Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual Abstract This manual describes how to configure and manage the Parallel Library TCP/IP subsystem on an HP NonStop™ S-series server. Product Version Parallel Library TCP/IP G06 Supported Release Version Updates (RVUs) This manual supports G06.24 and all subsequent G-series RVUs until otherwise indicated by its replacement publication. Part Number Published 522271-006 March 2005 Document History Part Number Product Version Published 522271-002 Parallel Library TCP/IP G06 August 2002 522271-003 Parallel Library TCP/IP G06 September 2003 522271-004 Parallel Library TCP/IP G06 February 2004 522271-005 Parallel Library TCP/IP G06 September 2004 522271-006 Parallel Library TCP/IP G06 February 2005 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual Glossary Index What’s New in This Manual ix Manual Information ix New and Changed Information Examples Figures Tables x About This Manual xi Who Should Use This Manual xi How to Use This Manual xi Required Background xii Parallel Library TCP/IP Core Manuals xii Background Manuals and Prerequisite Materials Notation Conventions xvii Abbreviations xx xv 1. Configuration Quick Start Key Differences Between Parallel Library TCP/IP and Conventional (HP NonStop) TCP/IP 1-2 Starting Parallel Library TCP/IP With TELNET and LISTNER Using a HOSTS File 1-3 Task Summary 1-3 Tasks: Starting Parallel Library TCP/IP With HOSTS 1-3 Starting Parallel Library TCP/IP With TELNET and LISTNER Using DNS 1-10 Task Summary 1-10 Tasks: Starting Parallel Library TCP/IP With DNS 1-10 Starting TCPMAN Using the RUN Command 1-15 Task Summary 1-15 Tasks: Starting Parallel Library TCP/IP Using RUN Command 1-16 Starting Parallel Library TCP/IP Using the Persistence Manager 1-18 Task Summary 1-18 Tasks: Starting Parallel Library TCP/IP Using Persistence Manager 1-18 Stopping Parallel Library TCP/IP and Preserving the Current Configuration 1-19 Task Summary 1-19 Tasks: Stopping Parallel Library TCP/IP and Preserving the Database 1-19 Stopping Parallel Library TCP/IP and Clearing the Database 1-24 Task Summary 1-24 Hewlett-Packard Company—522271-006 i 1. Configuration Quick Start (continued) Contents 1. Configuration Quick Start (continued) Tasks: Stopping Parallel Library TCP/IP and Clearing the Database 1-24 Stopping Parallel Library TCP/IP as a Generic Process 1-29 Task Summary 1-29 Tasks: Stopping Parallel Library TCP/IP as a Generic Process 1-29 2. Introduction Background 2-1 Single IP Host 2-2 Round-Robin Filtering 2-4 Scalability 2-6 Transparency 2-6 Ethernet Failover 2-7 Architectural Overview 2-9 TCPMAN 2-10 TCPMON 2-11 TCPSAM 2-11 SRL 2-11 PTrace 2-12 SCF 2-12 QIO 2-13 Parallel Library TCP/IP and Other Products 2-15 NonStop Kernel Subsystem and the System Configuration Database Programming With the New Socket Provider (TCPSAM) 2-16 Restrictions of Parallel Library TCP/IP 2-16 RFC Compliance 2-16 How to Access Online Help 2-16 2-15 3. Configuring Parallel Library TCP/IP for Complex and HeavyUse Environments Introduction and Definitions 3-1 Four Listening Methods 3-2 Standard Listening Model 3-2 Monolithic Listening Model 3-4 Distributor Listening Model 3-6 Hybrid Listener Model 3-9 Configuration Example for the Standard Listening Model 3-11 Configuration Example for the Monolithic Listening Model 3-15 Configuration Example for the Distributor Listening Model 3-18 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 ii Contents 3. Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments (continued) 3. Configuring Parallel Library TCP/IP for Complex and HeavyUse Environments (continued) Configuration Example for the Hybrid Listening Model 3-21 Example for Two Gateways — Standard Listening Model 3-23 Parallel Library TCP/IP for Complex, Heavy-Use WAN Environments 3-29 4. Managing the Parallel Library TCP/IP Subsystem Running Applications in Both Environments 4-1 Managing the System Configuration Database 4-2 Configuration Database Management 4-2 Managing Persistence 4-3 Managing the TCPSAM Process 4-3 How to Manage TCPSAM-Dependent Applications 4-4 How to Add TCPMAN as a Generic Process to the System Configuration Database 4-5 Managing Performance 4-7 Strategy for Coexistence with Conventional TCP/IP 4-7 Falling Back to Conventional TCP/IP 4-7 Dynamically Loading SPRs 4-8 5. SCF Reference for Parallel Library TCP/IP SCF for Parallel Library TCP/IP 5-1 SCF Commands for TCPMAN Compared to SCF Commands for TCPSAM Object Types 5-2 ENTRY Object Type 5-3 MONITOR Object Type 5-4 null Object Type 5-4 PROCESS Object Type 5-4 ROUTE Object Type 5-5 SUBNET Object Type 5-5 Naming Convention Summary 5-7 Wild-Card Support 5-7 Summary States 5-8 Parallel Library TCP/IP SCF Commands 5-9 Supported Commands and Object Types 5-9 Entering SCF Commands 5-11 ABORT Command 5-12 ABORT MON Command for TCPMAN 5-12 ABORT PROCESS Command for TCPMAN 5-13 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 iii 5-1 Contents 5. SCF Reference for Parallel Library TCP/IP (continued) 5. SCF Reference for Parallel Library TCP/IP (continued) ABORT PROCESS Command for TCPSAM 5-14 ABORT ROUTE Command for TCPMAN 5-15 ABORT SUBNET Command for TCPMAN 5-16 ADD Command 5-17 ADD ENTRY Command for TCPMAN 5-17 ADD ROUTE Command for TCPMAN 5-18 ADD SUBNET Command for TCPMAN 5-21 ALTER Command 5-25 ALTER MON Command for TCPMAN 5-25 ALTER SUBNET Command for TCPMAN 5-30 DELETE Command 5-33 DELETE ENTRY Command for TCPMAN 5-33 DELETE ROUTE Command for TCPMAN 5-34 DELETE SUBNET Command for TCPMAN 5-35 INFO Command 5-36 INFO ENTRY Command for TCPMAN 5-36 INFO MON Command for TCPMAN 5-38 INFO PROCESS Command for TCPMAN 5-43 INFO PROCESS Command for TCPSAM 5-44 INFO ROUTE Command for TCPMAN 5-49 INFO ROUTE Command for TCPSAM 5-52 INFO SUBNET Command for TCPMAN 5-55 INFO SUBNET Command for TCPSAM 5-58 LISTOPENS Command 5-60 LISTOPENS MON Command for TCPMAN 5-60 LISTOPENS PROCESS Command for TCPSAM 5-63 NAMES Command 5-66 NAMES ENTRY Command for TCPMAN 5-66 NAMES ROUTE Command for TCPMAN 5-67 NAMES ROUTE Command for TCPSAM 5-68 NAMES SUBNET Command for TCPMAN 5-69 NAMES SUBNET Command for TCPSAM 5-70 PRIMARY Command 5-70 PRIMARY PROCESS Command for TCPMAN 5-70 PRIMARY PROCESS Command for TCPSAM 5-71 START Command 5-72 START MON Command for TCPMAN 5-72 START ROUTE Command for TCPMAN 5-73 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 iv Contents 5. SCF Reference for Parallel Library TCP/IP (continued) 5. SCF Reference for Parallel Library TCP/IP (continued) START SUBNET Command for TCPMAN 5-74 STATS Command 5-75 STATS MON Command for TCPMAN 5-76 STATS PROCESS Command for TCPSAM 5-96 STATS ROUTE Command for TCPMAN 5-115 STATS ROUTE Command for TCPSAM 5-117 STATS SUBNET Command for TCPMAN 5-118 STATS SUBNET Command for TCPSAM 5-121 STATUS Command 5-123 STATUS ENTRY Command for TCPMAN 5-124 STATUS MON Command for TCPMAN 5-125 STATUS PROCESS Command for TCPMAN 5-129 STATUS PROCESS Command for TCPSAM 5-130 STATUS ROUTE Command for TCPMAN 5-133 STATUS ROUTE Command for TCPSAM 5-135 STATUS SUBNET Command for TCPMAN 5-136 STATUS SUBNET Command for TCPSAM 5-137 STOP Command 5-139 STOP MON Command for TCPMAN 5-139 STOP PROCESS Command for TCPMAN 5-140 STOP PROCESS Command for TCPSAM 5-140 STOP ROUTE Command for TCPMAN 5-141 STOP SUBNET Command for TCPMAN 5-142 TRACE Command 5-143 TRACE MON Command for TCPMAN 5-143 TRACE PROCESS Command for TCPMAN 5-145 TRACE PROCESS Command for TCPSAM 5-148 TRACE SUBNET Command for TCPMAN 5-150 VERSION Command 5-152 VERSION MON Command for TCPMAN 5-152 VERSION PROCESS Command for TCPMAN 5-153 VERSION PROCESS Command for TCPSAM 5-154 Parallel Library TCP/IP Trace Facility 5-155 Introduction to PTrace 5-155 PTrace Commands 5-157 DETAIL Command 5-159 HEX Command 5-159 LABEL Command 5-160 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 v 5. SCF Reference for Parallel Library TCP/IP (continued) Contents 5. SCF Reference for Parallel Library TCP/IP (continued) OCTAL Command 5-160 SELECT Command 5-161 TEXT Command 5-163 Trace Record Formats 5-164 Socket Creation Records 5-165 Memory Buffer Allocation Records 5-168 Interprocess Communication Records 5-168 TCP Records 5-169 UDP Input Records 5-176 Detailed UDP Input Records 5-177 UDP Output Records 5-178 IP Input Records 5-179 IP Output Records 5-181 Route Records 5-182 Socket Command Records 5-183 UDP User Request Records 5-187 6. Troubleshooting Tips A. SCF Command Summary B. SCF Error Messages PTCPIP 00001 PTCPIP 00002 PTCPIP 00003 PTCPIP 00004 PTCPIP 00005 PTCPIP 00007 PTCPIP 00008 PTCPIP 00009 PTCPIP 00010 PTCPIP 00011 PTCPIP 00012 PTCPIP 00013 PTCPIP 00014 PTCPIP 00016 PTCPIP 00017 PTCPIP 00018 PTCPIP 00019 B-1 B-1 B-1 B-1 B-2 B-2 B-2 B-2 B-3 B-3 B-3 B-3 B-3 B-4 B-4 B-4 B-4 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 vi B. SCF Error Messages (continued) Contents B. SCF Error Messages (continued) PTCPIP 00020 PTCPIP 00022 PTCPIP 00027 PTCPIP 00035 PTCPIP 00036 PTCPIP 00037 PTCPIP 00038 PTCPIP 00039 PTCPIP 00040 B-5 B-5 B-5 B-5 B-6 B-6 B-6 B-6 B-7 C. Tracer Utility Running the Tracer Utility from a Terminal C-1 Glossary Index Examples Example 1-1. Example 1-2. Example 1-3. Example 1-4. Example 3-1. Example 3-2. Example 3-3. Example 3-4. Example 3-5. Example 3-6. Example 3-7. Example 3-8. Example 3-9. Example 4-1. Example 4-2. TCPIPUP Command File Using HOSTS file 1-8 TCPIPUP Command File Using DNS 1-14 TCPIPDN Command File 1-23 TCPIPDN Command File 1-28 TCPIPUP1 Command File 3-13 TCPIPUP2 for the LISTNER Process 3-14 SCFSBNT File for TCPIPUP2 3-15 TCPIPUP3 for the TELSERV Process 3-17 TCPIPUP4 for the Distrib Process 3-20 TCPIPUP5 for Hybrid Listening Model 3-22 TCPIPUP6 for LISTNER Environment and Two Gateways 3-25 SCFSBNT2 File for TCPIPUP6 3-26 HOSTS File for TCPIPUP6 3-28 SAMUP 4-4 Command File for Adding TCPMAN as a Generic Process 4-5 Figures Figure i. Figure ii. Figure 1-1. Figure 2-1. Parallel Library TCP/IP Core Manuals xiii Programming Manuals for Parallel Library TCP/IP xiv TACL STATUS Display 1-9 Multiple IP Appearance, Conventional TCP/IP 2-2 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 vii Figures (continued) Contents Figures (continued) Figure 2-2. Figure 2-3. Figure 2-4. Figure 3-1. Figure 3-2. Figure 3-3. Figure 3-4. Figure 3-5. Figure 3-6. Figure 3-7. Figure 3-8. Figure 3-9. Figure 3-10. Figure 3-11. Figure 5-1. Figure 5-2. Figure 5-3. Single IP Appearance, Parallel Library TCP/IP 2-3 Data Path Comparison: Conventional vs. Parallel Library TCP/IP 2-10 Parallel Library TCP/IP Subsystem Within the System 2-14 Standard Listening Model 3-3 Monolithic: Listening Model 3-5 Distributor Listening Model in Conventional TCP/IP 3-8 Distributor Listening Model in Parallel Library TCP/IP 3-9 Hybrid Listening Model in Conventional TCP/IP 3-10 Hybrid Listening Model in Parallel Library TCP/IP 3-11 Standard Listening Model Configuration Example: LISTNER 3-12 Configuration Example for Monolithic Listening Model: TELSERV 3-16 Configuration Example for Distributor Listening Model: Distrib 3-19 Configuration Example for Hybrid Listening Model: iTP WebServer 3-21 Two Gateways With LISTNER 3-24 TCPMAN Process Object Hierarchy 5-3 TCPSAM Process Object Hierarchy 5-3 Recording and Displaying Trace Data 5-156 Tables Table i. Table 1-1. Table 1-2. Table 1-3. Table 1-4. Table 5-1. Table 5-2. Table 5-3. Table 5-4. Table 5-5. Table 5-6. Table 5-7. Summary of Contents xi Starting the Parallel Library TCP/IP Environment 1-1 Stopping the Parallel Library TCP/IP Environment 1-1 Configuration Form 1 for HOSTS File Startup 1-5 Configuration Form 2 for HOSTS File Startup 1-12 Route Object Naming Conventions 5-5 Object Naming Convention Summary and Reserved Names 5-7 Object Summary States 5-8 Commands and Object Types for TCPMAN 5-9 Commands and Object Types for TCPSAM 5-10 Sensitive and Nonsensitive SCF Commands 5-11 Summary of Parallel Library TCP/IP PTrace Commands 5-158 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 viii What’s New in This Manual Manual Information HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual Abstract This manual describes how to configure and manage the Parallel Library TCP/IP subsystem on an HP NonStop™ S-series server. Product Version Parallel Library TCP/IP G06 Supported Release Version Updates (RVUs) This manual supports G06.24 and all subsequent G-series RVUs until otherwise indicated by its replacement publication. Part Number Published 522271-006 March 2005 Document History Part Number Product Version Published 522271-002 Parallel Library TCP/IP G06 August 2002 522271-003 Parallel Library TCP/IP G06 September 2003 522271-004 Parallel Library TCP/IP G06 February 2004 522271-005 Parallel Library TCP/IP G06 September 2004 522271-006 Parallel Library TCP/IP G06 February 2005 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 ix What’s New in This Manual New and Changed Information New and Changed Information • • In subsection ALTER MON Command for TCPMAN: ° The default value for the attribute DELAYACKSTIME int on page 5-26 has been changed from 5 to 20. ° The default (128) for the TCP-LISTEN-QUE-MIN attribute has been added under TCP-LISTEN-QUE-MIN int on page 5-28. The format of the display for STATUS MON, DETAIL under STATUS MON Display Format on page 5-126 has been updated to include the subnet name as part of the connection details. The descriptions of PPID and BPID have been removed and those of PID and OutSubNet have been added. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 x About This Manual This manual describes how to configure and manage the Parallel Library TCP/IP subsystem. Who Should Use This Manual System and network managers, operators, and others who configure and manage the Parallel Library TCP/IP subsystem should use this manual. How to Use This Manual Use this manual to configure Parallel Library TCP/IP on your system in conjunction with the TCP/IP (Parallel Library) Migration Guide. The TCP/IP (Parallel Library) Migration Guide lists migration considerations that could affect your configuration. Table i. Summary of Contents (page 1 of 2) Section Title This section... 1 Configuration Quick Start provides configuration procedures for setting up and shutting down a basic Parallel Library TCP/IP environment. Use this section if you want to get the subsystem started without first gaining a full understanding of Parallel Library TCP/IP, or if you are already familiar with the subsystem and just want basic subsystem configuration procedures. 2 Introduction explains the subsystem architecture and explains the new features of Parallel Library TCP/IP such as round-robin filtering and single IP. In addition, this section provides term definitions, and discusses the current restrictions of Parallel Library TCP/IP and its relationship to other products as well as RFC compliance and online help. 3 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments goes into more complex configuration examples that tell you how to maximize the benefits of Parallel Library TCP/IP. 4 Managing the Parallel Library TCP/IP Subsystem provides information about managing the subsystem. 5 SCF Reference for Parallel Library TCP/IP provides a reference for the SCF commands used for managing the subsystem. 6 Troubleshooting Tips provides tips for solving common problems encountered when configuring Parallel Library TCP/IP. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 xi Required Background About This Manual Table i. Summary of Contents (page 2 of 2) Section Title This section... A SCF Command Summary provides a quick reference of the syntax for all the SCF commands documented in Section 5. B SCF Error Messages describes the SCF errors returned by the Parallel Library TCP/IP subsystem including recovery procedures. C Tracer Utility describes a utility that displays the path taken by IP packets enroute to a network host. This manual also contains a glossary of technical terms and abbreviations used throughout the text. Required Background This manual assumes familiarity with the standard TCP/IP family of protocols described in the RFCs and IENs and familiarity with configuring IP networks. You should be familiar with HP NonStop S-series server architecture, the HP networking product ServerNet LAN Systems Access (SLSA), Ethernet 4 ServerNet adapters (E4SAs), Fast Ethernet ServerNet adapters (FESAs), Gigabit Ethernet 4-port ServerNet adapters (G4SAs), and Gigabit Ethernet ServerNet adapters (GESAs). This manual also assumes that you are familiar with the HP NonStop operating system. For a list of reference materials that can help provide this background, see Background Manuals and Prerequisite Materials on page xv. Parallel Library TCP/IP Core Manuals You should use this manual in conjunction with the TCP/IP (Parallel Library) Migration Guide. These are the two core manuals for Parallel Library TCP/IP. Figure i on page -xiii shows the core manuals for configuring and managing Parallel Library TCP/IP. The line between the TCP/IP (Parallel Library) Configuration and Management Manual and the TCP/IP (Parallel Library) Migration Guide indicates that you must use these two manuals together. The dashed lines connecting the TCP/IP (Parallel Library) Configuration and Management Manual to the LAN Configuration and Management Manual and Operator Messages Manual show that you might need to refer to these manuals when configuring and managing Parallel Library TCP/IP, but they are not always required. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 xii Adapter Manuals About This Manual Figure i. Parallel Library TCP/IP Core Manuals TCP/IP (Parallel Library) Configuration and Management Manual LAN Configuration and Management Manual TCP/IP (Parallel Library) Migration Guide Operator Messages Manual VST0001.vsd • • • • This manual, the TCP/IP (Parallel Library) Configuration and Management Manual, provides an introduction to the architecture, procedures for configuring and managing the subsystem, as well as reference information for the SCF commands used to manage the subsystem. The TCP/IP (Parallel Library) Migration Guide describes the differences between the conventional HP NonStop TCP/IP and Parallel Library TCP/IP subsystems and documents considerations for migrating to the Parallel Library TCP/IP subsystem. The Operator Messages Manual contains the operator messages distributed by the Event Management Service (EMS) for Parallel Library TCP/IP. The LAN Configuration and Management Manual describes the SLSA subsystem which provides parallel LAN I/O for NonStop S-series servers. In particular, read about logical interfaces (LIFs) and physical interfaces (PIFs) in that manual. Adapter Manuals Parallel Library TCP/IP supports the Fast Ethernet, Ethernet 4, Gigabit Ethernet, and Gigabit Ethernet 4-port ServerNet adapters (FESAs, E4SAs, GESAs, and G4SAs). For information about installing these adapters, see the Ethernet Adapter Installation and Support Guide, the Fast Ethernet Adapter Installation and Support Guide, the Gigabit Ethernet Adapter Installation and Support Guide and the Gigabit Ethernet 4-Port Adapter Installation and Support Guide. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 xiii Parallel Library TCP/IP Application Programming Manuals About This Manual Parallel Library TCP/IP Application Programming Manuals In addition to the core manuals for Parallel Library TCP/IP shown above, there are several programming manuals that affect applications that interface to TCP/IP. These manuals are shown in Figure ii. Figure ii. Programming Manuals for Parallel Library TCP/IP TCP/IP Programming Manual Open System Services Programmer's Guide Open System Services Porting Guide Open System Services System Calls Reference Manual Open System Services Library Calls Reference Manual VST0002.vsd • • • • • The IPX/SPX Programming Manual describes application development for the HP NonStop IPX/SPX subsystems using HP Guardian socket library routines. TCP/IP and TCP/IPv6 Programming Manual describes application development for the HP NonStop TCP/IP and HP NonStop TCP/IPv6 subsystem using the HP Guardian socket library routines. The Open System Services Programmer’s Guide describes how to write applications in C for the Open System Services (OSS) environment. The Open System Services System Calls Reference Manual contains reference information for OSS system functions, files and miscellaneous topics. The Open System Services Library Calls Reference Manual contains reference information for OSS library function calls. Parallel Library TCP/IP Application and Client Manuals The applications that are directly related to Parallel Library TCP/IP are described in the TCP/IP Applications and Utilities User Guide. In addition, the Expand communications product as well as the ServerNet wide area network (SWAN) subsystem are clients of the TCP/IP subsystem. Support of SWAN over Parallel Library TCP/IP is limited at this time. The procedures for configuring SWAN over Parallel Library TCP/IP are provided in the TCP/IP (Parallel Library) Migration Guide. The Expand subsystem and the application user’s manuals are described below: • The TCP/IP Applications and Utilities User Guide describes the interactive interfaces to the following NonStop TCP/IP applications: ECHO, FINGER, FTP, TFTP, TELNET, and TN6530. Server information is included for FTP, TFTP, and TELNET. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 xiv About This Manual • Background Manuals and Prerequisite Materials The Expand Configuration and Management Manual describes how to plan, configure, and manage the Expand subsystem on a NonStop S-series server. Use this manual for configuring Expand-over-IP lines. Background Manuals and Prerequisite Materials This subsection lists reference material that you can use to acquire the background required for managing TCP/IP on NonStop servers. For an overview of TCP/IP, see the books TCP/IP Illustrated Volume I and TCP/IP Illustrated Volume II by W. Richard Stevens, Prentice Hall, 1994. For an in-depth explanation of the Domain Name Server, see the book DNS and BIND by Paul Albitz and Cricket Liu, O’Reilly and Associates, Inc. Request for Comments (RFC) is a series of documents published by the Internet Engineering Task Force (IETF). The following RFCs related to TCP/IP can be located on the Internet. (As of this printing, the IETF home site is: http://www.ietf.org/ however, since URLs change frequently, if you cannot locate this site using that URL, try using your preferred search engine and the keyword IETF and, once on the home page for the IETF, navigate to “RFCs.”) • • • • • • • • • • • • • RFC 768 “User Datagram Protocol” RFC 791 “Internet Protocol” RFC 792 “Internet Control Message Protocol” RFC 793 “Transmission Control Protocol” RFC 819 “Domain Naming Convention for Internet User Applications” RFC 821 “Simple Mail Transfer Protocol” RFC 826 “Ethernet Address Resolution Protocol” RFC 894 “Standard for the Transmission of IP Datagrams Over Ethernet Networks” RFC 973 “Domain System Changes and Observations” RFC 974 “Mail Routing and Domain System” RFC 1034 “Domain Names — Concepts and Facilities” RFC 1042 “Standard for the Transmission of IP Datagrams Over IEEE 802 Networks” RFC 1323 “TCP Extensions for High Performance” HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 xv About This Manual NonStop S-Series System Configuration Manuals The following manuals provide background material that is helpful for fully using Parallel Library TCP/IP: • • • • • • • The Telserv Manual describes the TELSERV SCF interface. This guide is intended for configuration and support planners who are responsible for the operation of the TELSERV subsystem. This guide also provides information about the TN6530 terminal emulation utility. The TCP/IP TELNET Management Programming Manual describes the command/response interface and the Event Management Service (EMS) interface available to an application program for communication with the TCP/IP TELNET process. The QIO Configuration and Management Manual describes how to install and manage a QIO data communications subsystem. This manual also describes the SCF commands used to configure, control, and inquire about the QIO subsystem. The Open System Services Shell and Utilities Reference Manual documents the contents of the inetd configuration file. The TCP/IP Configuration and Management Manual provides some background information about TCP/IP fundamentals and complete information on the NonStop TCP/IP product. Introduction to Networking for HP NonStop S-Series Servers provides an overview of HP networking and data communications concepts, tasks, products, and manuals. It discusses ways to connect NonStop subsystems to various devices and networks and it introduces the tools and interfaces you can use. The Guardian User’s Guide provides introductory information and task-oriented instructions for using the HP Tandem Advanced Command Language (TACL) and various Guardian environment utilities. The utilities and procedures described in this guide include many of the more common operations/tasks that users will need to perform on a system running the NonStop operating system. NonStop S-Series System Configuration Manuals You might also need to refer to the following manuals for some configuration and management tasks: The SCF Reference Manual for G-Series RVUs describes the operation of SCF and the commands used to configure, control, and inquire about supported data communications subsystems. Of particular interest to the Parallel Library TCP/IP subsystem manager is the information about saving the system configuration database. SCF Reference Manual for the Kernel Subsystem describes the SCF interface to the NonStop Kernel subsystem and useful information about the persistence manager and generic processes. The NonStop S-Series FastPath Guide explains how to install and configure a two-processor or four-processor NonStop S-series server and provides a basic set of HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 xvi About This Manual Notation Conventions system operations procedures. This guide is intended for experienced HP customers who want to quickly install, configure, and operate their initial NonStop S-series server as a stand-alone server or as a development node within an existing Expand/IP network. Notation Conventions Hypertext Links Blue underline is used to indicate a hypertext link within text. By clicking a passage of text with a blue underline, you are taken to the location described. For example: This requirement is described under Backup DAM Volumes and Physical Disk Drives on page 3-2. General Syntax Notation The following list summarizes the notation conventions for syntax presentation in this manual. UPPERCASE LETTERS. Uppercase letters indicate keywords and reserved words; enter these items exactly as shown. Items not enclosed in brackets are required. For example: MAXATTACH lowercase italic letters. Lowercase italic letters indicate variable items that you supply. Items not enclosed in brackets are required. For example: file-name computer type. Computer type letters within text indicate C and Open System Services (OSS) keywords and reserved words; enter these items exactly as shown. Items not enclosed in brackets are required. For example: myfile.c italic computer type. Italic computer type letters within text indicate C and Open System Services (OSS) variable items that you supply. Items not enclosed in brackets are required. For example: pathname [ ] Brackets. Brackets enclose optional syntax items. For example: TERM [\system-name.]$terminal-name INT[ERRUPTS] A group of items enclosed in brackets is a list from which you can choose one item or none. The items in the list may be arranged either vertically, with aligned brackets on HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 xvii About This Manual General Syntax Notation each side of the list, or horizontally, enclosed in a pair of brackets and separated by vertical lines. For example: FC [ num ] [ -num] [ text] K [ X | D ] address-1 { } Braces. A group of items enclosed in braces is a list from which you are required to choose one item. The items in the list may be arranged either vertically, with aligned braces on each side of the list, or horizontally, enclosed in a pair of braces and separated by vertical lines. For example: LISTOPENS PROCESS { $appl-mgr-name } { $process-name } ALLOWSU { ON | OFF } | Vertical Line. A vertical line separates alternatives in a horizontal list that is enclosed in brackets or braces. For example: INSPECT { OFF | ON | SAVEABEND } … Ellipsis. An ellipsis immediately following a pair of brackets or braces indicates that you can repeat the enclosed sequence of syntax items any number of times. For example: M address-1 [ , new-value ]... [ - ] {0|1|2|3|4|5|6|7|8|9}... An ellipsis immediately following a single syntax item indicates that you can repeat that syntax item any number of times. For example: "s-char..." Punctuation. Parentheses, commas, semicolons, and other symbols not previously described must be entered as shown. For example: error := NEXTFILENAME ( file-name ) ; LISTOPENS SU $process-name.#su-name Quotation marks around a symbol such as a bracket or brace indicate the symbol is a required character that you must enter as shown. For example: "[" repetition-constant-list "]" Item Spacing. Spaces shown between items are required unless one of the items is a punctuation symbol such as a parenthesis or a comma. For example: CALL STEPMOM ( process-id ) ; If there is no space between two items, spaces are not permitted. In the following example, there are no spaces permitted between the period and any other items: $process-name.#su-name HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 xviii Notation for Messages About This Manual Line Spacing. If the syntax of a command is too long to fit on a single line, each continuation line is indented three spaces and is separated from the preceding line by a blank line. This spacing distinguishes items in a continuation line from items in a vertical list of selections. For example: ALTER [ / OUT file-spec / ] LINE [ , attribute-spec ]... Notation for Messages The following list summarizes the notation conventions for the presentation of displayed messages in this manual. Bold Text. Bold text in an example indicates user input entered at the terminal. For example: ENTER RUN CODE ?123 CODE RECEIVED: 123.00 The user must press the Return key after typing the input. Nonitalic text. Nonitalic letters, numbers, and punctuation indicate text that is displayed or returned exactly as shown. For example: Backup Up. lowercase italic letters. Lowercase italic letters indicate variable items whose values are displayed or returned. For example: p-register process-name [ ] Brackets. Brackets enclose items that are sometimes, but not always, displayed. For example: Event number = number [ Subject = first-subject-value ] A group of items enclosed in brackets is a list of all possible items that can be displayed, of which one or none might actually be displayed. The items in the list might be arranged either vertically, with aligned brackets on each side of the list, or horizontally, enclosed in a pair of brackets and separated by vertical lines. For example: proc-name trapped [ in SQL | in SQL file system ] { } Braces. A group of items enclosed in braces is a list of all possible items that can be displayed, of which one is actually displayed. The items in the list might be arranged HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 xix About This Manual Abbreviations either vertically, with aligned braces on each side of the list, or horizontally, enclosed in a pair of braces and separated by vertical lines. For example: obj-type obj-name state changed to state, caused by { Object | Operator | Service } process-name State changed from old-objstate to objstate { Operator Request. } { Unknown. } | Vertical Line. A vertical line separates alternatives in a horizontal list that is enclosed in brackets or braces. For example: Transfer status: { OK | Failed } % Percent Sign. A percent sign precedes a number that is not in decimal notation. The % notation precedes an octal number. The %B notation precedes a binary number. The %H notation precedes a hexadecimal number. For example: %005400 P=%p-register E=%e-register Abbreviations IPC. Inter-process communication TCPMAN. TCP/IP manager object TCPMON. TCP/IP monitor object TCPSAM. TCP/IP socket access method (transport-service provider) HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 xx 1 Configuration Quick Start This section provides concise examples of setting up the Parallel Library TCP/IP environment. If you prefer to read introductory information before configuring this subsystem, see Section 2, Introduction. In addition, before starting Parallel Library TCP/IP for the first time, see the TCP/IP (Parallel Library) Migration Guide for any considerations that affect your configuration. Table 1-1 and Table 1-2 describe the procedures you can follow to set up and stop the Parallel Library TCP/IP environment. Table 1-1. Starting the Parallel Library TCP/IP Environment Section This procedure... Page Starting Parallel Library TCP/IP With TELNET and LISTNER Using a HOSTS File Establishes a Parallel Library TCP/IP host environment that includes a TELNET service (TACL) and a LISTNER service (FTP server, ECHO server and FINGER server). Host-name resolution is through a HOSTS file. 1-3 Starting Parallel Library TCP/IP With TELNET and LISTNER Using DNS Establishes a Parallel Library TCP/IP host environment that uses an external Domain Name Server for host-name resolution. Many IP networks, particularly those connected to the Internet, include Domain Name Servers for host-name resolution. 1-10 Starting TCPMAN Using the RUN Command Tells you how to start Parallel Library TCP/IP by using the RUN command after the system-configuration database has been populated with the TCPMON, route, entry, and subnet objects. 1-15 Starting Parallel Library TCP/IP Using the Persistence Manager Tells you how to bring up the Parallel Library TCP/IP environment by using the persistence manager. 1-18 Table 1-2. Stopping the Parallel Library TCP/IP Environment Section This procedure... Page Stopping Parallel Library TCP/IP and Preserving the Current Configuration Tells you how to stop the Parallel Library TCP/IP without clearing the system-configuration database. Use this shut-down procedure when you want to restart Parallel Library TCP/IP using the same configuration. 1-19 Stopping Parallel Library TCP/IP and Clearing the Database Tells you how to stop your Parallel Library TCP/IP environment and clear the system-configuration database. Use this shut-down procedure when you want to restart Parallel Library TCP/IP using a new configuration. 1-24 Stopping Parallel Library TCP/IP as a Generic Process Tells you how to stop the Parallel Library TCP/IP environment when TCPMAN has been added to the system-configuration database as a generic process. 1-29 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1 -1 Configuration Quick Start Key Differences Between Parallel Library TCP/IP and Conventional (HP NonStop) TCP/IP Key Differences Between Parallel Library TCP/IP and Conventional (HP NonStop) TCP/IP Parallel Library TCP/IP presents a new architectural paradigm that requires some relearning for users who are familiar with the NonStop TCP/IP product. The later sections of this manual as well as the TCP/IP (Parallel Library) Migration Guide explain these differences in detail. This subsection highlights some of the major, conceptual differences between the conventional TCP/IP and Parallel Library TCP/IP products. Parallel Library TCP/IP doesn’t rely on processes in the same way that conventional TCP/IP does. However, Parallel Library TCP/IP provides a process for backward compatibility called the TCPSAM process. For applications that expect a TCP/IP process, you create a TCPSAM process and use that name when defining a TCPIP^PROCESS^NAME for application use. You can create as many TCPSAM processes as you want, but only one TCPSAM process is required for all the TCP/IP client applications in the system. So, for example, to run LISTNER and TELSERV as shown in Example 1-1 on page 1-8, just add a DEFINE/PARAM for a TCPSAM process name exactly as you would add a DEFINE for a TCP/IP process name in conventional TCP/IP. You no longer have to use different TCP/IP processes for different applications because you don’t use separate TCP/IP processes to balance the loads created by the different applications. In Parallel Library TCP/IP, all the applications can share one TCPSAM process without incurring an interprocessor-hop cost and without associating themselves with specific, underlying adapters. The TCPSAM process is not in the processing path for data transfer, it is a dummy process for backwards compatibility with applications. Parallel Library TCP/IP can distribute the load from all the TCP/IP-client applications across all the adapters available in the system. Another difference is that where conventional TCP/IP could have multiple TCP/IP processes, each having one or more LIFs uniquely associated with it, Parallel Library TCP/IP has one manager process ($ZZTCP) and all LIFs are associated with that process. Since, in Parallel Library TCP/IP, the LIF (and IP address) is no longer associated with the TCP/IP process used by the applications (TCPSAM), applications using Parallel Library TCP/IP do not know which LIF or IP address they will get. If you want to associate an application with a specific LIF (and IP address), you can still do this by using subnet-level binding. For information about this technique, see SubnetLevel Binding: How to Isolate Subnets in a Single-IP Environment on page 2-4. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1 -2 Configuration Quick Start Starting Parallel Library TCP/IP With TELNET and LISTNER Using a HOSTS File Starting Parallel Library TCP/IP With TELNET and LISTNER Using a HOSTS File Use this procedure to start Parallel Library TCP/IP if you are starting it for the first time (or have cleared the system configuration database, see Stopping Parallel Library TCP/IP and Clearing the Database on page 1-24) and want to configure a HOSTS file. Note. If you have installed Parallel Library TCP/IP before, you only need to execute the RUN command for the TCPMAN ($ZZTCP). (See Starting TCPMAN Using the RUN Command on page 1-15). Task Summary Task 1: Check for any considerations that affect your configuration. Task 2: Check all assumptions. Task 3: Fill in the configuration form. Task 4: Save the system configuration database. Task 5: Edit the HOSTS file. Task 6: Create a command file. Task 7: OBEY the command file. Task 8: Specify a DEFINE to set up the HOSTS file. Task 9: Test the new environment. Tasks: Starting Parallel Library TCP/IP With HOSTS 1. Check for any issues that affect your configuration. See the TCP/IP (Parallel Library) Migration Guide. 2. Check that the following assumptions are met. a. Check that the SCF environment is operational (that is, $ZNET is STARTED). Enter the following command at the TACL prompt. >STATUS $ZNET $ZNET should be running. If $ZNET is not running, see the SCF Reference Manual for G-Series RVUs. b. Check that QIOMON is running. Enter the following command at the SCF prompt: ->STATUS MON $ZM* You should see a $ZMnn process running in every processor in which you plan to run Parallel Library TCP/IP. If you do not, refer to the QIO Configuration and Management Manual. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1 -3 Configuration Quick Start Tasks: Starting Parallel Library TCP/IP With HOSTS c. Check that TCPMAN is not already running on the system. Enter the following command at the SCF prompt: ->LISTDEV PTCPIP $ZZTCP should not appear in the list of processes. If $ZZTCP does appear, you cannot start Parallel Library TCP/IP because Parallel Library TCP/IP is already running. If you want to make changes to the configuration, see Section 5, SCF Reference for Parallel Library TCP/IP. d. Check that the SLSA subsystem is configured. Enter the following command at the SCF prompt: ->STATUS PROCESS $ZZLAN You should see $ZZLAN in the STARTED state. If you do not, refer to the LAN Configuration and Management Manual. e. Select a LIF of TYPE ETHERNET. a. Obtain a list of all LIFs. Enter the following command at the SCF prompt: ->STATUS LIF $ZZLAN.* b. Using one of the LIF names from Step a, determine if the LIF is of TYPE ETHERNET by issuing the following command at the SCF prompt: ->INFO LIF $ZZLAN.LAN01 f. Ensure that the LIF you are configuring for Parallel Library TCP/IP is accessible to all processors on which you plan to run Parallel Library TCP/IP by entering the following command at the SCF prompt: ->STATUS LIF $ZZLAN.LAN01, DETAIL The CPUs with Data Path field lists the processors to which the LIF is accessible. g. Enter the name of the LIF you are configuring on Line # 5, SLSA LIF Name, of the Configuration Form 1 on page 1-5. h. Check that the TCPMAN process has not been added as a generic process to the system configuration database by entering the following command at the SCF prompt: ->STATUS PROCESS $ZZKRN.#ZZTCP 3. Fill in the Configuration Form 2 on page 1-12. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1 -4 Tasks: Starting Parallel Library TCP/IP With HOSTS Configuration Quick Start Configuration Form 1 Use the form in Table 1-3 to help collect the variables needed to start Parallel Library TCP/IP. This form lists the source of the variable including instructions, where appropriate, for obtaining it. The Line # column allows you to easily crossreference the variables in Example 1-1 on page 1-8 to the form in Table 1-3. Table 1-3. Configuration Form 1 for HOSTS File Startup (page 1 of 2) Step or Example 1-1 Uses... Line # Variable Name/ Note 1. Config Dbase # CONFIGxx.yy where xx and yy follow your system’s numbering scheme. See the SCF Reference Manual for G-Series RVUs. Step 4 uses 01.04 2. Host Name Arbitrary; you assign. ptcpip 3. Host ID You can assign a new IP address (obtained from your network administrator) or convert one of your existing IP addresses. If you convert an existing IP address, follow the procedures in Tasks for Migrating Your Environment of the TCP/IP (Parallel Library) Migration Guide. 172.17.215.27 4. Subnet Name Arbitrary; you assign. SN1 5. SLSA LIF Name Get this variable from Step f on page 1-4. LAN01 6. Subnet Mask Depends on the IP address class and your network setup (obtain from your network administrator or see the TCP/IP Configuration and Management Manual for an explanation of IP addresses and subnet masks). %hffffff00 (255.255.255.0) 7. Route Name Arbitrary; you assign. ROUTE1 8. Gateway/router address Must be an IP address associated with a router on your network subnet (obtain from your network administrator). 172.17.215.1 9. Location of TCPIPUP Get the location of your TACL command file from Step 6 on page 1-6. (Note: you can fill this field in later when you get to Step 6.) $GUEST.USER Source HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1 -5 Tasks: Starting Parallel Library TCP/IP With HOSTS Configuration Quick Start Table 1-3. Configuration Form 1 for HOSTS File Startup (page 2 of 2) Line # Variable Name/ Note Source Step or Example 1-1 Uses... 10. TCPSAM Name Arbitrary; you assign. $ZSAM1 11. LISTNER Name Arbitrary; you assign. $LSN1 12. TELSERV Name Arbitrary; you assign. $ZTN1 4. Save the system configuration database by entering the following SCF command. Substitute the numbering scheme you identified on Configuration Form 1 on page 1-5, line # 1, for the variable (01.04). ->SAVE CONFIGURATI0N 01.04 Note. See the TCP/IP (Parallel Library) Migration Guide for important information about saving your configuration database and preserving all your system configuration parameters. Also, see Configuration Database Management on page 4-2. 5. Edit the $SYSTEM.ZTCPIP.HOSTS file. a. Exit SCF, then enter the following TACL command: >TEDIT $SYSTEM.ZTCPIP.HOSTS b. Add the following entry to the file. (Use the name you identified on the Configuration Form 1 on page 1-5, line # 2, for the variable.) 172.17.215.27 ptcpip 6. Create a TACL command file called TCPIPUP as in Example 1-1 on page 1-8. Place the file in whichever volume you choose, but note the location on Line 9 of the Configuration Form 1 on page 1-5 for future reference. You can copy this command file from this manual available in the NonStop Technical Library (NTL). Replace the variables (in italics) with the values you recorded on Configuration Form 1 on page 1-5. The primary and backup processors for processes are also variable but were not included on Configuration Form 1. Select the primary and backup processors according to your system needs or use the ones in the example. Enter the following command at the TACL prompt: >TEDIT $GUEST.USER.TCPIPUP HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1 -6 Configuration Quick Start Tasks: Starting Parallel Library TCP/IP With HOSTS Note. If you receive an error saying that one or more of the LISTNER or TELSERV process file names is already in use, use the procedures for Stopping Parallel Library TCP/IP and Clearing the Database on page 1-24 to clear out Parallel Library TCP/IP environment, then use the TACL STOP command to stop the LISTNER and TELSERV processes that caused the error. (Socket applications can be bound to the SRL without having an open socket. Hence, their file names would be in use but they would not show up in the LISTOPENS MON command.) HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1 -7 Configuration Quick Start Tasks: Starting Parallel Library TCP/IP With HOSTS Example 1-1. TCPIPUP Command File Using HOSTS file ==Clear the system of any DEFINEs and PARAMs DELETE DEFINE =_SRL_01 DELETE DEFINE =TCPIP^PROCESS^NAME CLEAR ALL ==Start the TCPMAN process TCPMAN/NAME $ZZTCP, TERM $ZHOME, OUT $ZHOME, CPU 0,NOWAIT/1 ==SCF commands SCF/INLINE/ INLPREFIX + ==Start TCPMON objects in all processors, establish the host ==name and host ID, and set up loopback. + ASSUME PROCESS $ZZTCP + START MON * ==Give TCPMONs time to start + DELAY 21 ==Use host name from Configuration Form 1 on page 1-5, Line # 2 + ALTER MON *,HOSTNAME "ptcpip" ==Use host ID from Configuration Form 1 on page 1-5, Line # 3 + ALTER MON *,HOSTID 172.17.215.27 + STOP SUBNET LOOP0 ==Note in the above command that even though the MON is not ==assumed, you do not have to specify the MON. See ==STOP SUBNET Command for TCPMAN on page 5-142 for the STOP ==SUBNET command syntax. + ALTER SUBNET LOOP0,IPADDRESS 127.1 + START SUBNET LOOP0 ==Add and start a subnet and route for the TCPMONs. ==Use subnet name from Line # 4, DEVICENMAME from Line # 5, IP ==address from Line # 3, and subnet mask from Line # 6 of the ==Configuration Form 1 on page 1-5 + ADD SUBNET SN1,TYPE ETHERNET,DEVICENAME LAN01,IPADDRESS & 172.17.215.27,SUBNETMASK %HFFFFFF00 ==Use route name from Line # 7, and GATEWAY from Line # 8 of ==the Configuration Form 1 on page 1-5 + ADD ROUTE ROUTE1,DESTINATION 0.0.0.0,GATEWAY 172.17.215.1 + START SUBNET * + START ROUTE * INLEOF ==Add a DEFINE for the private SRL for TCPSAM. ADD DEFINE =_SRL_01,CLASS MAP,FILE ZTCPSRL ==Start a TCPSAM process. Use TCPSAM name from Line # 10 of ==the Configuration Form 1 on page 1-5 TCPSAM/NAME $ZSAM0,TERM $ZHOME, OUT $ZHOME, NOWAIT,CPU 0/1 ==Add a DEFINE to establish the TCPSAM process name ==as the transport-service provider for LISTNER. ==Use the TCPSAM name from Line # 10 of the Configuration Form 1 on page 1-5 ADD DEFINE =TCPIP^PROCESS^NAME, CLASS MAP, FILE $ZSAM0 ==Add a PARAM to establish the TCPSAM process name as ==the transport-service provider for TELSERV. PARAM TCPIP^PROCESS^NAME $ZSAM0 ==Add a PARAM to cause TELSERV to use the same process going out ==as coming in. Use the TCPSAM name from Line # 10 of the ==Configuration Form 1 on page 1-5 PARAM ZTNT^TRANSPORT^PROCESS^NAME, $ZSAM0 ==Start LISTNER and TELSERV. (See TELSERV in the ==TCP/IP (Parallel Library) Migration Guide for considerations ==about starting TELSERV.) Use LISTNER name from Line # 11 and ==TELSERV name from Line # 12 of the Configuration Form 1 on page 1-5. LISTNER/TERM $ZHOME, OUT $ZHOME, NAME $LSN1,CPU 2,NOWAIT, & PRI 160/1 $SYSTEM.ZTCPIP.PORTCONF TELSERV/TERM $ZHOME, OUT $ZHOME, NAME $ZTN1, CPU 1, & NOWAIT/-BACKUPCPU 3 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1 -8 Configuration Quick Start Tasks: Starting Parallel Library TCP/IP With HOSTS 7. Ensure that you are in the volume/subvolume that contains the SRL. a. First find out what the current SYSnn is: >STATUS 0,0 The STATUS command displays your subvolume information similarly to the sample display shown in Figure 1-1: Figure 1-1. TACL STATUS Display VST123.vsd In the display shown in Figure 1-1, the Program file column shows $SYSTEM.SYS00.OSIMAGE. The SYS00 portion of the Program file tells you that SYSnn is SYS00 in this example. b. Change your session location to the volume/subvolume location of the SYSnn as in the following example (substitute real values for 00): >VOLUME $SYSTEM.SYS00 8. Get the location of your TACL command file from Line 9 of Configuration Form 1 on page 1-5 and substitute this value for the variable location GUEST.USER in the following example command. Issue the following TACL OBEY command on the TCPIPUP command file while running as user SUPER.SUPER: >OBEY $GUEST.USER.TCPIPUP 9. For Parallel Library TCP/IP applications such as FTP and TELNET to use the HOSTS file for name resolution, you must specify the following DEFINE. Typically, this DEFINE is placed in the $SYSTEM.SYSTEM.TACLLOCL file so that all TACL users inherit the DEFINE. Enter the following command at the TACL prompt: >ADD DEFINE =TCPIP^HOST^FILE,CLASS MAP,FILE & $SYSTEM.ZTCPIP.HOSTS Once the Parallel Library TCP/IP environment is running, you can use FTP to make transfers to or from this host. The TELSERV configuration, by default, gives users access to a TACL service. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1 -9 Configuration Quick Start Starting Parallel Library TCP/IP With TELNET and LISTNER Using DNS 10. Test that the Parallel Library TCP/IP subsystem is running on the desired IP address by issuing the following commands at the TACL prompt. You can use the INFO SUBNET Command for TCPMAN to get the IP address for use in the following command sequence: >TELNET ip-address >TACL >LOGON SUPER.SUPER Starting Parallel Library TCP/IP With TELNET and LISTNER Using DNS Use this procedure to start Parallel Library TCP/IP if you are starting it for the first time (or have cleared the system configuration database, see Stopping Parallel Library TCP/IP and Clearing the Database on page 1-24) and want to use the Domain Name Server (DNS). Note. If you have installed Parallel Library TCP/IP before, you only need to execute the RUN command for the TCPMAN ($ZZTCP). (See Starting TCPMAN Using the RUN Command on page 1-15.) Task Summary Task 1: Check the TCP/IP (Parallel Library) Migration Guide for any considerations that affect your configuration. Task 2: Check that all assumptions are met. Task 3: Fill in the configuration form. Task 4: Save your system configuration database (CONFIG). Task 5: Create a startup command file to start Parallel Library TCP/IP. Task 6: OBEY the startup file to start the Parallel Library TCP/IP environment. Task 7: Test the Parallel Library TCP/IP environment. Tasks: Starting Parallel Library TCP/IP With DNS 1. Check for considerations that affect your configuration. See the TCP/IP (Parallel Library) Migration Guide. 2. Check that the following assumptions are met. a. Check that the SCF environment is operational (that is, $ZNET is STARTED). Enter the following command at the TACL prompt. >STATUS $ZNET If $ZNET is not running, see the SCF Reference Manual for G-Series RVUs. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1-10 Configuration Quick Start Tasks: Starting Parallel Library TCP/IP With DNS b. Check that QIOMON is running. Enter the following command at the SCF prompt: ->STATUS MON $ZM* You should see a $ZMnn process running in every processor in which you plan to run Parallel Library TCP/IP. If you do not, refer to the QIO Configuration and Management Manual. c. Check that TCPMAN is not already running on the system. Enter the following command at the SCF prompt: ->LISTDEV PTCPIP $ZZTCP should not appear in the list of processes. If $ZZTCP does appear, you cannot start Parallel Library TCP/IP because Parallel Library TCP/IP is already running. If you want to make changes to the configuration, see Section 5, SCF Reference for Parallel Library TCP/IP. d. Check that the SLSA subsystem is configured. Enter the following command at the SCF prompt: ->STATUS PROCESS $ZZLAN You should see $ZZLAN in the STARTED state. If you do not, refer to the LAN Configuration and Management Manual. e. Select a LIF of TYPE ETHERNET. a. Obtain a list of all LIFs. Enter the following command at the SCF prompt: ->STATUS LIF $ZZLAN.* b. Using one of the LIF names from Step a, determine if the LIF is of TYPE ETHERNET by issuing the following command at the SCF prompt: ->INFO LIF $ZZLAN.LAN02 f. Ensure that the LIF you are configuring for Parallel Library TCP/IP is accessible to all processors on which you plan to run Parallel Library TCP/IP by entering the following command at the SCF prompt: ->STATUS LIF $ZZLAN.LAN02, DETAIL The CPUs with Data Path field lists the processors to which the LIF is accessible. g. Enter the name of this LIF on Line # 5, SLSA LIF Name, of Configuration Form 2 on page 1-12. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1-11 Tasks: Starting Parallel Library TCP/IP With DNS Configuration Quick Start h. Ensure that the TCPMAN process has not been added as a generic process to the system configuration database by entering the following command at the SCF prompt: ->STATUS PROCESS $ZZKRN.#ZZTCP i. Ensure that DNS is configured properly and running on your system by checking with your DNS administrator. 3. Fill in Configuration Form 2. Configuration Form 2 Use the form provided in Table 1-4 to help collect the variables needed to start Parallel Library TCP/IP. This form lists the source of the variable including instructions, where appropriate, for obtaining it. The Line # column allows you to easily cross-reference the variable in Example 1-2 on page 1-14 to the form in Table 1-4. Table 1-4. Configuration Form 2 for HOSTS File Startup (page 1 of 2) Step or Example 1-2 Uses... Line # Variable Name/ Note 1. Config Dbase # CONFIGxx.yy where xx and yy follow your system’s numbering scheme. See the SCF Reference Manual for G-Series RVUs. Step 4 uses 01.05 2. Host Name Arbitrary; you assign. ptcpip 3. Host ID You can assign a new IP address (obtained from your network administrator) or convert one of your existing IP addresses. If you convert an existing IP address, follow the procedures in Tasks for Migrating Your Environment of the TCP/IP (Parallel Library) Migration Guide. 150.20.30.1 4. Subnet Name Arbitrary; you assign. SN2 5. SLSA LIF Name Get this variable from Step f on page 1-11. LAN02 6. Subnet Mask Depends on the IP address class and your network setup (obtain from your network administrator or see the TCP/IP Configuration and Management Manual for an explanation of IP addresses and subnet masks.) %hffffff00 (255.255.255.0) Source HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1-12 Tasks: Starting Parallel Library TCP/IP With DNS Configuration Quick Start Table 1-4. Configuration Form 2 for HOSTS File Startup (page 2 of 2) Line # Variable Name/ Note Source Step or Example 1-2 Uses... 7. Route Name Arbitrary; you assign. ROUTE2 8. Gateway/router address Must be an IP address associated with a router on your network subnet (obtain from your network administrator.) 150.20.30.2 9. Location of TCPIPUP Get the location of your TACL command file from Step 5 on page 1-13. (Note: you can fill this field in later when you get to Step 5.) $GUEST.USER 10. TCPSAM Name Arbitrary; you assign. $ZSAM2 11. LISTNER Name Arbitrary; you assign. $LSN2 12. TELSERV Name Arbitrary; you assign. $ZTN2 4. Save the system configuration database by entering the following SCF command. Substitute the numbering scheme you identified on Configuration Form 2, Line # 1, for the variable (01.05): ->SAVE CONFIGURATI0N 01.05 Note. See the TCP/IP (Parallel Library) Migration Guide for important information about saving your configuration database and preserving all your system configuration parameters. Also, see Configuration Database Management on page 4-2. 5. Create a TACL command file called TCPIPUP as in Example 1-2 on page 1-14. Place the file in your whichever volume you choose, but note the location on Line 9 of the Configuration Form 2 on page 1-12 for future reference. You can copy this command file from this manual available in the NonStop Technical Library (NTL). Replace the variables (in italics) with the values you recorded on Configuration Form 2 on page 1-12. The primary and backup processors for processes are also variable but were not included on Configuration Form 2 on page 1-12. Select these processors according to your system needs or use the ones in the example. 6. Enter the following command at the TACL prompt: >TEDIT TCPIPUP HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1-13 Configuration Quick Start Tasks: Starting Parallel Library TCP/IP With DNS Example 1-2. TCPIPUP Command File Using DNS ==Clear the system of any DEFINEs and PARAMs DELETE DEFINE =_SRL_01 DELETE DEFINE =TCPIP^PROCESS^NAME ==Set up environment to use DNS by deleting host file DEFINE DELETE DEFINE =TCPIP^HOST^FILE CLEAR ALL ==Start the TCPMAN process TCPMAN /NAME $ZZTCP, TERM $ZHOME, OUT $ZHOME, CPU 0,NOWAIT/1 ==SCF commands SCF/INLINE/ INLPREFIX + ==Start TCPMON objects in all processors, establish the host ==name and host ID, and set up loopback. + ASSUME PROCESS $ZZTCP + START MON * ==Give TCPMONs time to start + DELAY 21 ==This example ASSUMEs the TCPMON name + ASSUME MON $ZZTCP.* ==Use host name from Line #2 of the Configuration Form 2 + ALTER,HOSTNAME "ptcpip" ==Use host ID from Line # 3 of the Configuration Form 2 + ALTER,HOSTID 150.20.30.1 + ABORT SUBNET LOOP0 + ALTER SUBNET LOOP0, IPADDRESS 127.1 ==Add and start a subnet and route for the TCPMONs. ==Use subnet name from Line # 4, DEVICENMAME from Line # 5, IP ==address from Line # 3, and subnet mask from Line # 6 of the ==Configuration Form 1 + ADD SUBNET SN2,TYPE ETHERNET,DEVICENAME LAN02, & IPADDRESS 150.20.30.1, SUBNETMASK %HFFFFFF00 ==Use route name from Line # 7 and gateway from Line # 8 of the ==Configuration Form 2 + ADD ROUTE ROUTE2, DESTINATION 0.0.0.0, GATEWAY 150.20.30.2 + START SUBNET * + START ROUTE * INLEOF ==Add a DEFINE for the private SRL for TCPSAM. ADD DEFINE =_SRL_01, CLASS MAP, FILE ZTCPSRL ==Start a TCPSAM process. Use TCPSAM name from Line # 10 of ==the Configuration Form 2 TCPSAM/NAME $ZSAM2, TERM $ZHOME, OUT $ZHOME, NOWAIT,CPU 0/1 ==Add a DEFINE to establish the TCPSAM process name ==as the transport-service provider for LISTNER. ==Use the TCPSAM name from Line # 10 of the Configuration Form 2. ADD DEFINE =TCPIP^PROCESS^NAME, CLASS MAP, FILE $ZSAM2 ==Start the LISTNER. Use the LISTNER name from Line # 11 ==of the Configuration Form 2. LISTNER/TERM $ZHOME, OUT $ZHOME, NAME $LSN2,& CPU 0, NOWAIT, PRI 160/1 $SYSTEM.ZTCPIP.PORTCONF ==Add a PARAM to establish the TCPSAM process name ==as the transport-service provider for TELSERV. PARAM TCPIP^PROCESS^NAME $ZSAM2 ==Add a PARAM to cause TELSERV to use the same process going out ==as coming in. PARAM ZTNT^TRANSPORT^PROCESS^NAME, $ZSAM2 ==Start TELSERV. Use the TELSERV name from Line # 12 of the ==Configuration Form 2. See TELSERV of the ==TCP/IP (Parallel Library) Migration Guide for considerations. TELSERV/TERM $ZHOME, OUT $ZHOME, NAME $ZTN2, CPU 0, NOWAIT, & PRI 170/ -BACKUPCPU 1 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1-14 Configuration Quick Start Starting TCPMAN Using the RUN Command 7. Get the location of your TACL command file from Line 9 of Configuration Form 2 on page 1-12 and substitute this value for the variable location GUEST.USER in the following example command. Issue the following TACL OBEY command on the TCPIPUP command file while running as user SUPER.SUPER: >OBEY $GUEST.USER.TCPIPUP Note. If you receive an error saying that one or more of the LISTNER or TELSERV process file names is already in use, use the procedures for Stopping Parallel Library TCP/IP and Clearing the Database on page 1-24 to clear out Parallel Library TCP/IP environment and use the TACL STOP command to stop the LISTNER and TELSERV processes that caused the error. (It’s possible for socket applications to be bound to the SRL without having an open socket. Hence, their file names would be in use but they would not show up in the LISTOPENS MON command.) 8. Test that the Parallel Library TCP/IP subsystem is running on the desired IP address. You can use the INFO SUBNET Command for TCPMAN to get the IP address for use in the following command sequence. Enter the following commands at the TACL prompt: >TELNET ip-address >TACL >LOGON SUPER.SUPER Starting TCPMAN Using the RUN Command Use this procedure if you have stopped Parallel Library TCP/IP without clearing the system configuration database (see Stopping Parallel Library TCP/IP and Preserving the Current Configuration on page 1-19). This procedure assumes that you have started Parallel Library TCP/IP at least once before. SCF objects have been configured before and not deleted when Parallel Library TCP/IP was stopped, are available upon restart with this procedure. This procedure uses some variables, such as primary and backup processors and the name of the TCPSAM process. These variables are indicated in italics. For information on how to obtain real values for all variables except the backup and primary processors, see Configuration Form 1 on page 1-5. Select backup and primary processors according to your system needs. Task Summary Task 1: Check the TCP/IP (Parallel Library) Migration Guide for any considerations that affect your configuration. Task 2: Check that all assumptions are met. Task 3: Clear all DEFINEs and PARAMs. Task 4: Set up the environment to use a HOSTS file or DNS HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1-15 Configuration Quick Start Tasks: Starting Parallel Library TCP/IP Using RUN Command Task 5: Issue the TACL RUN command to start Parallel Library TCP/IP. Task 6: Start TCPSAM for socket programs. Task 7: Test the Parallel Library TCP/IP environment. Tasks: Starting Parallel Library TCP/IP Using RUN Command 1. Check for considerations that affect your configuration. See the TCP/IP (Parallel Library) Migration Guide. 2. Check that the following assumptions are met. a. Check that the SCF environment is operational (that is, $ZNET is STARTED). Enter the following command at the TACL prompt: >STATUS $ZNET $ZNET should be running. If $ZNET is not running, see the SCF Reference Manual for G-Series RVUs. b. Check that QIOMON is running. Enter the following command at the SCF prompt: ->STATUS MON $ZM* You should see a $ZMnn process running in every processor in which you plan to run Parallel Library TCP/IP. If you do not, refer to the QIO Configuration and Management Manual. c. Check that TCPMAN is not already running on the system. Enter the following command at the SCF prompt: ->LISTDEV PTCPIP $ZZTCP should not appear in the list of processes. If it does appear, you cannot start Parallel Library TCP/IP because Parallel Library TCP/IP is already running. If you want to make changes to the configuration, see Section 5, SCF Reference for Parallel Library TCP/IP. d. Check that the SLSA subsystem is configured. Enter the following command at the SCF prompt: ->STATUS PROCESS $ZZLAN You should see $ZZLAN in the STARTED state. If you do not, refer to the LAN Configuration and Management Manual. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1-16 Configuration Quick Start Tasks: Starting Parallel Library TCP/IP Using RUN Command 3. Clear the system of DEFINEs and PARAMs by exiting SCF and entering the following commands at the TACL prompt: > > > > DELETE DEFINE =_SRL_01 DELETE DEFINE =TCPIP^PROCESS^NAME DELETE DEFINE =TCPIP^HOST^FILE CLEAR ALL 4. (Skip this step if you are using DNS.) If you are not using DNS for address resolution, set up the environment to use a HOSTS file by adding the following DEFINE to $SYSTEM.SYSTEM.TACLLOCL: >ADD DEFINE =TCPIP^HOST^FILE,CLASS MAP,FILE & $SYSTEM.ZTCPIP.HOSTS 5. Issue the following TACL RUN command to start Parallel Library TCP/IP: >TCPMAN /NAME $ZZTCP, TERM $ZHOME, OUT $ZHOME, CPU 0, NOWAIT/1 Note. If you have stopped $ZZTCP by using the STOP PROCESS $ZZTCP, SUB ALL command, the TCPMONs are deleted from the configuration database and you still need to start the TCPMONs manually. You can work around this problem by stopping the TCPMAN process in two steps: STOP MON $ZZTCP.* STOP PROCESS $ZZTCP 6. Start TCPSAM for socket programs. a. ADD a DEFINE for the SRL for the TCPSAM process. (See Locating the SRL on page 2-12.) Enter the following command at the TACL prompt: >ADD DEFINE =_SRL_01, CLASS MAP, FILE ZTCPSRL b. Issue the following TACL RUN command to start TCPSAM: >TCPSAM /NAME $ZSAM3, TERM $ZHOME, OUT $ZHOME, NOWAIT, CPU 0/1 c. ADD a DEFINE and PARAM to establish the TCPSAM process name as the transport-service provider for socket programs by entering the following commands at the TACL prompt: >ADD DEFINE =TCPIP^PROCESS^NAME, CLASS MAP, FILE $ZSAM3 >PARAM TCPIP^PROCESS^NAME $ZSAM3 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1-17 Configuration Quick Start Starting Parallel Library TCP/IP Using the Persistence Manager 7. Test the Parallel Library TCP/IP environment. You can use the INFO SUBNET Command for TCPMAN to get the IP address for use in the following command sequence: >TELNET ip-address >TACL >LOGON SUPER.SUPER Starting Parallel Library TCP/IP Using the Persistence Manager Use this procedure to start Parallel Library TCP/IP if TCPMAN has been added to the system configuration database as a persistent generic process. (See Stopping Parallel Library TCP/IP as a Generic Process on page 1-29 and Managing the System Configuration Database on page 4-1.) Task Summary Task 1: Check all assumptions. Task 2: Start the generic process. Task 2: Add TCPSAM processes. Tasks: Starting Parallel Library TCP/IP Using Persistence Manager 1. Check that the following assumptions are met. a. Check that TCPMAN has been added as a generic process to the system configuration database. (See Managing the System Configuration Database on page 4-1 for more information.) Enter the following command at the SCF prompt: ->STATUS PROCESS $ZZKRN.#ZZTCP b. Check that the STARTMODE parameter of the generic TCPMAN process is MANUAL. (If the parameter is SYSTEM, you need not start the process since the persistence manager automatically starts it.) Enter the following command at the SCF prompt: ->INFO PROCESS $ZZKRN.#ZZTCP, DETAIL 2. Start the generic process for TCPMAN by entering the following SCF command: -> START PROCESS $ZZKRN.#ZZTCP HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1-18 Configuration Quick Start Stopping Parallel Library TCP/IP and Preserving the Current Configuration 3. Add a TCPSAM process by entering the following TACL command (replace the variables, indicated in italics, with real values): >ADD DEFINE =_SRL_01, CLASS MAP, FILE $SYSTEM.SYS03.ZTCPSRL > TCPSAM/TERM $ZHOME, 0/1 OUT $ZHOME, NAME $SIP02, NOWAIT, CPU Stopping Parallel Library TCP/IP and Preserving the Current Configuration Follow this shut-down procedure when you want to stop the Parallel Library TCP/IP environment and later restart it using the same configuration. Note that you must stop the Parallel Library TCP/IP from a conventional TCP/IP environment. Task Summary Use TSM to perform these tasks. Because TSM uses conventional TCP/IP, you will not have to worry about shutting down your operating environment. Task 1: Check that there is a Parallel Library TCP/IP environment on your system. Task 2: Ensure that you do not stop the TCP/IP process running your terminal. Task 3: Check for applications using Parallel Library TCP/IP. Task 4: Check for TCPSAM processes running in the Parallel Library TCP/IP environment. Task 5: Create a TACL command file to shut down the subsystem. Task 6: OBEY the TACL command file. Tasks: Stopping Parallel Library TCP/IP and Preserving the Database 1. Check that there is a Parallel Library TCP/IP environment running on your system by entering the following SCF command: ->LISTDEV PTCPIP The following sample display results from the LISTDEV PTCPIP and shows that Parallel Library TCP/IP is running on the system: -> listdev PTCPIP LDev Name PPID BPID Type RSize Pri Program 242 $ZPTM1 1,319 0,0 (68,0 ) 57344 201 \HOME.$JER01.THJAGUAR.TCPMON 254 $ZZTCP 3,276 0,0 (68,0 ) 132 200 \HOME.$JER01.THJAGUAR.TCPMAN 279 $ZPTM3 3,271 0,0 (68,0 ) 57344 201 \HOME.$JER01.THJAGUAR.TCPMON 287 $ZPTM2 2,282 0,0 (68,0 ) 57344 201 \HOME.$JER01.THJAGUAR.TCPMON 333 $ZPTM0 0,327 0,0 (68,0 ) 57344 201 \HOME.$SYSTEM.THJAGUAR.TCPMON HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1-19 Configuration Quick Start Tasks: Stopping Parallel Library TCP/IP and Preserving the Database The above sample display shows that there are four TCPMONs running (shown in the Program column) named $ZPTM1, $ZPTM3, $ZPTM2, and $ZPTM0. Also, one TCPMAN process ($ZZTCP) is running. 2. Perform this step if you are not using TSM. Ensure that you do not stop the TCP/IP process that is running your home terminal. a. Enter WHO at the TACL prompt: >WHO The following sample display results from the TACL WHO command: \HOME.$SYSTEM.SYSTEM 2> WHO Home terminal: $ZTNP1.#PTYPRAB TACL process: \HOME.$Z34A Primary CPU: 2 (NSR-G) Default Segment File: $SYSTEM.#0000382 Pages allocated: 24 Pages Maximum: 1024 Bytes Used: 32820 (1%) Bytes Maximum: 2097152 Current volume: $SYSTEM.SYSTEM Saved volume: $SYSTEM.SYSTEM Userid: 255,255 Username: SUPER.SUPER Security: "AAAA" Logon name: SUPER.SUPER b. The TELSERV process, $ZTNP1, is listed next to the HOME TERMINAL field. Make note of the TELSERV process (just the portion following the dollar sign ($)). a. Check all TCP/IP processes to find the one that has your TELSERV process listed as an opener. This is the process that you do not want to shut down. >SCF ->LISTDEV TCPIP The following sample display results from the LISTDEV TCPIP command and shows all the TCP/IP processes in the system: -> listdev tcpip LDev Name PPID BPID Type 204 $ZTC0 1,302 0,322 (48,0 ) 298 $TCPS3 3,278 0,0 (48,0 ) 305 $TCPS1 1,341 0,0 (48,0 ) 332 $ZTC01 0,301 1,389 (48,0 \HOME.$SYSTEM.THJAGUAR.TCPSAM RSize Pri Program 32000 200 \HOME.$SYSTEM.SYS07.TCPIP 57344 201 \HOME.$JER01.THJAGUAR.TCPSAM 57344 201 \HOME.$JER01.THJAGUAR.TCPSAM ) 57344 201 b. Issue a LISTOPENS PROCESS $process-name on each process listed in the display for LISTDEV TCPIP until you find the process that is running the TACL prompt of your home terminal (identified in Step b.) ->LISTOPENS PROCESS $ZTC0 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1- 20 Tasks: Stopping Parallel Library TCP/IP and Preserving the Database Configuration Quick Start The following sample display results from the LISTOPENS PROCESS command and shows all the processes depending on $ZTCO: TCPIP Listopens PROCESS \HOME.$ZTC0 Openers $ZPRP1 $ZPRP1 $ZPRP1 $ZTNP1 $ZTSM $ZCVP1 $ZPMP1 $ZPMP1 $ZTNP1 $ZNET PPID 1,304 1,304 1,304 1,305 0,307 0,324 1,266 1,266 1,305 0,21 BPID PLFN 4 5 6 3 22 1 2 3 4 2 BLFN 0 0 0 0 0 0 0 0 0 0 Protocol TCP TCP TCP TCP TCP UDP UDP TCP TCP #ZSPI Lport echo finger ftp telnet 980 548 111 111 telnet * In the Openers column, $ZTNP1 (the name of the TELSERV process identified in Step 2b on page 1-20) is an opener of the $ZTC0 process. Be sure that you do not stop this process while executing this shutdown procedure. This is the process that is running your home terminal. Comparing the process that you have just identified as running your home terminal to the output from the LISTDEV commands in Steps 1 on page 1-19 and 2ba on page 1-20, ensure that this process is not a TCPSAM process. If the process running your home terminal is a TCPSAM process, you must use TELNET to connect to a conventional TCP/IP process and subnet on your system, then execute these procedures from that process. 3. Determine if any applications are using the TCPMONs and make a note of the application names for Step 6. Enter the following command at the SCF prompt (a sample display follows this command): ->LISTOPENS MON $ZZTCP.* Note. Socket applications can be bound to the SRL without having an open socket. Hence, their file names would be in use but they would not show up in the LISTOPENS MON command. These applications could cause an error when you start the Parallel Library TCP/IP subsystem. If you receive an error (“NLD fatal error, cannot open ZTCPSRL” or “file already exists”) when running your startup files, repeat the procedure for shutting down and clearing out the database, then add the names of the processes listed in the error message to the STOP PROCESS commands in the command file. (See ==Stop the Opener Processes== in Example 1-3 on page 1-23.) HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1- 21 Tasks: Stopping Parallel Library TCP/IP and Preserving the Database Configuration Quick Start The following sample display results from the LISTOPENS MON command and shows all the processes depending on the Parallel Library TCP/IP subsystem: -> listopens mon $zztcp.* PTCPIP Listopens MON \HOME.$ZZTCP.#ZPTM0 Openers $ZPT0 $ZPT0 $ZPT0 $ZTN0 $ZTF0 $Z0KW $Z0KX $ZTN0 $ZTN0 $ZTN0 $ZTN0 $ZTN0 $ZTN0 $ZTN0 $ZTN0 $Z07S PPID 0,295 0,295 0,295 0,277 0,300 0,314 0,319 0,277 0,277 0,277 0,277 0,277 0,277 0,277 0,277 0,331 BPID PLFN 5 6 7 3 4 1 1 5 10 7 4 8 9 6 12 1 BLFN 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Protocol TCP TCP TCP TCP UDP TCP TCP TCP TCP TCP TCP TCP TCP TCP TCP TCP Lport echo finger ftp telnet 69 ftp ftp telnet telnet telnet telnet telnet telnet telnet telnet ftp In the above sample display, you would record all the opener processes: $ZPT0, $ZTN0, $ZTF0, $Z0KW, $Z0KX, $ZTN0, and $Z07S to be stopped in Example 1-3 on page 1-23. 4. List the names of the TCPSAM processes running in the Parallel Library TCP/IP environment. Enter the following command at the SCF prompt: ->LISTDEV TCPIP Make a note of the names of the running TCPSAM processes. (The process type is indicated in the last field of the Program column.) The following sample display results from the LISTDEV TCPIP command and shows all the TCP/IP processes running on the system: -> listdev tcpip LDev 170 186 215 240 Name PPID BPID $ZTC1 0,291 $ZTC0 1,293 $ZTC01 0,285 $TCPS1 1,267 Type RSize Pri Program 1,289 (48,0 ) 32000 200 \HOME.$SYSTEM.SYS07.TCPIP 0,302 (48,0 ) 32000 200 \HOME.$SYSTEM.SYS07.TCPIP 1,313 (48,0 ) 57344 201 \HOME.$SYSTEM.SYS07.TCPSAM 0,0 (48,0 ) 57344 201 \HOME.$JER01.THJAGUAR.TCPSAM In the above display, the TCPSAM processes are $TCPS1, and $ZTC01. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1- 22 Configuration Quick Start Tasks: Stopping Parallel Library TCP/IP and Preserving the Database 5. Create a TACL command file as shown in Example 1-3. Replace the italicized variables with the names of the applications and the TCPSAM processes you noted in the above steps. Enter the following command at the TACL prompt: >TEDIT TCPIPDN Example 1-3. TCPIPDN Command File ==Stop the opener processes. See Note on page 1-26. STOP $ZPT0 STOP $ZTN0 STOP $ZTF0 STOP $ZOKW STOP $ZOKX STOP $Z07S ==SCF Commands SCF/INLINE/ INLPREFIX + ==Stop TCPSAM processes + ABORT PROCESS $ZSAM0 ==Stop TCPMAN and all TCPMONs + STOP PROCESS $ZZTCP,SUB ALL ==Pause while all TCPMONs stop + DELAY 21 INLEOF 6. Issue the following TACL OBEY command on the TCPIPDN command file while running as user SUPER.SUPER: >OBEY TCPIPDN HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1- 23 Configuration Quick Start Stopping Parallel Library TCP/IP and Clearing the Database Stopping Parallel Library TCP/IP and Clearing the Database Follow this shut-down procedure when you want to stop the Parallel Library TCP/IP environment and later restart it using a new configuration. Note that you must stop the Parallel Library TCP/IP from a conventional TCP/IP environment in your system. Task Summary Use TSM to perform these tasks. Because TSM uses conventional TCP/IP, you will not have to worry about shutting down your operating environment. Task 1: Check that there is a Parallel Library TCP/IP environment on your system. Task 2: Ensure that you do not stop the TCP/IP environment that is running your home terminal. Task 3: Check for applications using Parallel Library TCP/IP. Task 4: Check TCPSAM processes running in the Parallel Library TCP/IP environment. Task 5: Obtain names of the subnets running in the Parallel Library TCP/IP environment. Task 6: Create a TACL command file to bring down the environment. Task 7: Issue the OBEY command to stop the Parallel Library TCP/IP environment. You must substitute real values for the TCPSAM process name, the LISTNER name, and the TELSERV name; however, these names are arbitrary. The variables are indicated in italics in the following procedures. Tasks: Stopping Parallel Library TCP/IP and Clearing the Database 1. Check that a Parallel Library TCP/IP environment is running on your system by entering the following SCF command: ->LISTDEV PTCPIP The following sample display results from a LISTDEV PTCPIP command and shows all the Parallel Library TCP/IP processes running on the system: -> listdev PTCPIP LDev Name PPID BPID Type RSize Pri Program 242 $ZPTM1 1,319 0,0 (68,0 ) 57344 201 \HOME.$JER01.THJAGUAR.TCPMON 254 $ZZTCP 3,276 0,0 (68,0 ) 132 200 \HOME.$JER01.THJAGUAR.TCPMAN 279 $ZPTM3 3,271 0,0 (68,0 ) 57344 201 \HOME.$JER01.THJAGUAR.TCPMON 287 $ZPTM2 2,282 0,0 (68,0 ) 57344 201 \HOME.$JER01.THJAGUAR.TCPMON 333 $ZPTM0 0,327 0,0 (68,0 ) 57344 201 \HOME.$SYSTEM.THJAGUAR.TCPMON HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1- 24 Configuration Quick Start Tasks: Stopping Parallel Library TCP/IP and Clearing the Database The above sample display shows that there are four TCPMONs running (shown in the Program column) named $ZPTM1, $ZPTM3, $ZPTM2, and $ZPTM0. Also, one TCPMAN process ($ZZTCP) is running. 2. Perform this step if you are not using TSM. Ensure that you do not stop the TCP/IP environment that is running your home terminal. a. Enter WHO at the TACL prompt: >WHO The following sample display results from the TACL WHO command: \HOME.$SYSTEM.SYSTEM 2> who Home terminal: $ZTNP1.#PTYPRAB TACL process: \HOME.$Z34A Primary CPU: 2 (NSR-G) Default Segment File: $SYSTEM.#0000382 Pages allocated: 24 Pages Maximum: 1024 Bytes Used: 32820 (1%) Bytes Maximum: 2097152 Current volume: $SYSTEM.SYSTEM Saved volume: $SYSTEM.SYSTEM Userid: 255,255 Username: SUPER.SUPER Security: "AAAA" Logon name: SUPER.SUPER b. The TELSERV process, $ZTNP1, is listed next to the HOME TERMINAL field. Make note of the TELSERV process name (just the portion following the dollar sign ($)). a. Check all TCP/IP processes and find the one that has your TELSERV process listed as an opener. This is the process that you do not want to shut down. >SCF ->LISTDEV TCPIP The following sample displays results from the LISTDEV TCPIP command and shows all the TCP/IP processes in the system: -> listdev tcpip LDev Name PPID BPID Type RSize Pri Program 204 $ZTC0 1,302 0,322 (48,0 ) 32000 200 \HOME.$SYSTEM.SYS07.TCPIP 298 $TCPS3 3,278 0,0 (48,0 ) 57344 201 \HOME.$JER01.THJAGUAR.TCPSAM 305 $TCPS1 1,341 0,0 (48,0 ) 57344 201 \HOME.$JER01.THJAGUAR.TCPSAM 332 $ZTC01 0,301 1,389 (48,0 ) 57344 201 \HOME.$SYSTEM.SYS07.TCPSAM b. Issue a LISTOPENS PROCESS $process-name on each of the processes listed in the display for LISTDEV TCPIP until you find the process that is running the TACL prompt of your home terminal (identified in Step 2b on page 1-25.) ->LISTOPENS PROCESS $ZTC0 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1- 25 Tasks: Stopping Parallel Library TCP/IP and Clearing the Database Configuration Quick Start The following sample display results from the LISTOPENS PROCESS command and shows all the processes depending on $ZTC0: TCPIP Listopens PROCESS \HOME.$ZTC0 Openers $ZPRP1 $ZPRP1 $ZPRP1 $ZTNP1 $ZTSM $ZCVP1 $ZPMP1 $ZPMP1 $ZTNP1 $ZNET PPID 1,304 1,304 1,304 1,305 0,307 0,324 1,266 1,266 1,305 0,21 BPID PLFN 4 5 6 3 22 1 2 3 4 2 BLFN 0 0 0 0 0 0 0 0 0 0 Protocol TCP TCP TCP TCP TCP UDP UDP TCP TCP #ZSPI Lport echo finger ftp telnet 980 548 111 111 telnet * In the Openers column, we see that $ZTNP1, the name of the TELSERV process identified in Step2b on page 1-25, is an opener of the $ZTC0 process. Note. Be sure that you do not stop this process while executing this shutdown procedure. This is the process that is running your home terminal. Comparing the process that you have just identified as running your home terminal to the output from the LISTDEV commands in Steps 1 on page 1-24 and 2ba on page 1-25, ensure that this process is not a TCPSAM process. If the process running your home terminal is a TCPSAM process, you must TELNET to a conventional TCP/IP process and subnet on your system and execute these procedures from that process. 3. Determine if any applications are using the TCPMONs and make a note of the application names for Step 6. Enter the following command at the SCF prompt (a sample display follows this command): ->LISTOPENS MON $ZZTCP.* Note. It is possible for socket applications to be bound to the SRL without having an open socket. Hence, their file names would be in use but they would not show up in the LISTOPENS MON command. These applications could cause an error when you start the Parallel Library TCP/IP subsystem. If you receive an error (“NLD fatal error, cannot open ZTCPSRL” or “file already exists”) when running those startup files, repeat the procedure for shutting down and clearing out the database, then add the names of the processes listed in the error message to the STOP PROCESS commands in the command file. (See ==Stop the Opener Processes== in Example 1-3 on page 1-23.) HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1- 26 Tasks: Stopping Parallel Library TCP/IP and Clearing the Database Configuration Quick Start The following sample display results from the LISTOPENS MON command and shows all the processes depending on the Parallel Library TCP/IP subsystem: -> listopens mon $zztcp.* PTCPIP Listopens MON \HOME.$ZZTCP.#ZPTM0 Openers $ZPT0 $ZPT0 $ZPT0 $ZTN0 $ZTF0 $Z0KW $Z0KX $ZTN0 $ZTN0 $ZTN0 $ZTN0 $ZTN0 $ZTN0 $ZTN0 $ZTN0 $Z07S PPID 0,295 0,295 0,295 0,277 0,300 0,314 0,319 0,277 0,277 0,277 0,277 0,277 0,277 0,277 0,277 0,331 BPID PLFN 5 6 7 3 4 1 1 5 10 7 4 8 9 6 12 1 BLFN 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Protocol TCP TCP TCP TCP UDP TCP TCP TCP TCP TCP TCP TCP TCP TCP TCP TCP Lport echo finger ftp telnet 69 ftp ftp telnet telnet telnet telnet telnet telnet telnet telnet ftp In the above sample display, you would record all the opener processes: $ZPT0, $ZTN0, $ZTF0, $Z0KW, $Z0KX, $ZTN0, and $Z07S to be stopped in Example 1-4 on page 1-28. 4. List the names of the TCPSAM processes running in the Parallel Library TCP/IP environment. Enter the following command at the SCF prompt: ->LISTDEV TCPIP Make a note of the names of the running TCPSAM processes (indicated in the last field of the Program column). The following sample display results from the LISTDEV TCPIP command and shows all the TCP/IP processes running the system: -> listdev tcpip LDev 170 186 215 240 Name $ZTC1 $ZTC0 $ZTC01 $TCPS1 PPID BPID Type RSize Pri Program 0,291 1,289 (48,0 ) 32000 200 \HOME.$SYSTEM.SYS07.TCPIP 1,293 0,302 (48,0 ) 32000 200 \HOME.$SYSTEM.SYS07.TCPIP 0,285 1,313 (48,0 ) 57344 201 \HOME.$SYSTEM.SYS07.TCPSAM 1,267 0,0 (48,0 ) 57344 201 \HOME.$SYSTEM.SYS07.TCPSAM In the above display, the TCPSAM processes are $TCPS1, and $ZTC01. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1- 27 Configuration Quick Start Tasks: Stopping Parallel Library TCP/IP and Clearing the Database 5. Obtain the names of the subnets running in the Parallel Library TCP/IP environment by entering the following SCF command (ignore LOOP0): ->INFO SUBNET $ZZTCP.* Since you cannot delete the LOOPBACK subnet, Example 1-4 uses the DELETE SUBNET SN* command instead of DELETE SUBNET * to avoid receiving an error. If you have followed the recommended naming convention (your subnets start with SN) you do not need to change this variable in Example 1-4. If you have used a different naming convention, substitute the appropriate leading characters for SN. 6. Create the TACL command file shown in Example 1-4. Replace the italicized variables in Example 1-4 with information you obtained in Steps 3, 4, and 5. Enter the following command at the TACL prompt: >TEDIT TCPIPDN Example 1-4. TCPIPDN Command File ==Stop the opener processes (identified in Step 3). STOP $ZPT0 STOP $ZTN0 STOP $ZTF0 STOP $ZOKW STOP $ZOKX STOP $Z07S ==SCF Commands SCF/INLINE/ INLPREFIX + ==Stop the $TCPSAM processes (identified in Step 4). + ABORT PROCESS $ZSAM0 ==Clear the system configuration database. (Use leading ==characters for the subnet names identified in Step 5.) Note that ==deleting the subnets also deletes the routes associated with ==the subnets so you don’t have to delete the routes explicitly. + ASSUME PROCESS $ZZTCP + STOP SUBNET* + DELETE SUBNET SN* + DELETE ENTRY * ==Stop the TCPMON objects + ABORT MON * ==Pause for TCPMONs to stop + DELAY 21 ==Stop the TCPMAN process + ABORT PROCESS $ZZTCP INLEOF 7. Issue the following TACL OBEY command on the TCPIPDN command file while running as user SUPER.SUPER: >OBEY TCPIPDN HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1- 28 Stopping Parallel Library TCP/IP as a Generic Process Configuration Quick Start Stopping Parallel Library TCP/IP as a Generic Process Use this procedure to stop Parallel Library TCP/IP when TCPMAN has been added as a generic process to the system configuration database. Note that you must stop the Parallel Library TCP/IP from a conventional TCP/IP environment in your system. Task Summary Task 1: Check that there is a Parallel Library TCP/IP environment on your system. Task 2: Check that $ZZTCP has been added as a generic process to the system configuration database. Task 3: Identify the TCP/IP process that is running your home terminal. Task 4: Check for applications using Parallel Library TCP/IP. Task 5: Check TCPSAM processes running in the Parallel Library TCP/IP environment. Task 6: Stop all openers of the TCPMONs. Task 7: Stop all TCPSAM processes. Task 8: Stop the generic TCPMAN process. Tasks: Stopping Parallel Library TCP/IP as a Generic Process 1. Check that a Parallel Library TCP/IP environment is running on your system by entering the following SCF command: ->LISTDEV PTCPIP The following sample display results from a LISTDEV PTCPIP command and shows that Parallel Library TCP/IP is running on the system: -> listdev PTCPIP LDev Name 242 $ZPTM1 254 $ZZTCP 279 $ZPTM3 287 $ZPTM2 333 $ZPTM0 PPID 1,319 3,276 3,271 2,282 0,327 BPID 0,0 0,0 0,0 0,0 0,0 Type (68,0 (68,0 (68,0 (68,0 (68,0 ) ) ) ) ) RSize Pri Program 57344 201 \HOME.$JER01.THJAGUAR.TCPMON 132 200 \HOME.$JER01.THJAGUAR.TCPMAN 57344 201 \HOME.$JER01.THJAGUAR.TCPMON 57344 201 \HOME.$JER01.PSLIB.TCPMON 57344 201 \HOME.$JER01.PSLIB.TCPMON The above sample display shows that there are four TCPMONs running (you can tell this by looking in the Program column) named $ZPTM1, $ZPTM3, $ZPTM2, and $ZPTM0. A TCPMAN process called $ZZTCP is also running. 2. Check that $ZZTCP has been added as a generic process to the system configuration database by issuing the following SCF command: ->STATUS PROCESS $ZZKRN.#ZZTCP HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1- 29 Configuration Quick Start Tasks: Stopping Parallel Library TCP/IP as a Generic Process 3. Ensure that you do not stop the TCP/IP environment that is running your home terminal. a. Enter WHO at the TACL prompt: >WHO The following sample display results from the TACL WHO command: \HOME.$SYSTEM.SYSTEM 2> who Home terminal: $ZTNP1.#PTYPRAB TACL process: \HOME.$Z34A Primary CPU: 2 (NSR-G) Default Segment File: $SYSTEM.#0000382 Pages allocated: 24 Pages Maximum: 1024 Bytes Used: 32820 (1%) Bytes Maximum: 2097152 Current volume: $SYSTEM.SYSTEM Saved volume: $SYSTEM.SYSTEM Userid: 255,255 Username: SUPER.SUPER Security: "AAAA" Logon name: SUPER.SUPER b. The TELSERV process, $ZTNP1, is listed next to the HOME TERMINAL field. Make note of the TELSERV process name (just the portion following the dollar sign ($)). a. Check all TCP/IP processes and find the one that has your TELSERV process listed as an opener. This is the process that you do not want to shutdown. >SCF ->LISTDEV TCPIP The following sample display results from the LISTDEV TCPIP command and shows all the TCP/IP processes in the system: -> listdev tcpip LDev Name PPID BPID Type RSize Pri Program 204 $ZTC0 1,302 0,322 (48,0 ) 32000 200 \HOME.$SYSTEM.SYS07.TCPIP 298 $TCPS3 3,278 0,0 (48,0 ) 57344 201 \HOME.$JER01.THJAGUAR.TCPSAM 305 $TCPS1 1,341 0,0 (48,0 ) 57344 201 \HOME.$JER01.THJAGUAR.TCPSAM 332 $ZTC01 0,301 1,389 (48,0 ) 57344 201 \HOME.$SYSTEM.SYS07.TCPSAM b. Issue a LISTOPENS PROCESS $process-name on each of the processes listed in the display for LISTDEV TCPIP until you find the process that is running the TACL prompt of your home terminal (identified in Step 3a on page 1-30.) ->LISTOPENS PROCESS $ZTC0 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1- 30 Tasks: Stopping Parallel Library TCP/IP as a Generic Process Configuration Quick Start The following sample display results from the LISTOPENS PROCESS command and shows all the processes depending on $ZTC0: TCPIP Listopens PROCESS \HOME.$ZTC0 Openers $ZPRP1 $ZPRP1 $ZPRP1 $ZTNP1 $ZTSM $ZCVP1 $ZPMP1 $ZPMP1 $ZTNP1 $ZNET PPID 1,304 1,304 1,304 1,305 0,307 0,324 1,266 1,266 1,305 0,21 BPID PLFN 4 5 6 3 22 1 2 3 4 2 BLFN 0 0 0 0 0 0 0 0 0 0 Protocol TCP TCP TCP TCP TCP UDP UDP TCP TCP #ZSPI Lport echo finger ftp telnet 980 548 111 111 telnet * In the Openers column, we see that $ZTNP1, the name of the TELSERV process identified in Step b on page 1-30b on page 1-30, is an opener of the $ZTC0 process. Note. Be sure that you do not stop this process while executing this shutdown procedure. This is the process that is running your home terminal. Comparing the process that you have just identified as running your home terminal to the output from the LISTDEV commands in Steps a on page 1-30 and 2ba on page 1-30, ensure that this process is not a TCPSAM process. If the process running your home terminal is a TCPSAM process, you must TELNET to a conventional TCP/IP process and subnet on your system and execute these procedures from that process. 4. Determine if any applications are using the TCPMONs and make a note of the application names for Step 6. Enter the following command at the SCF prompt (a sample display follows this command): ->LISTOPENS MON $ZZTCP.* Note. Socket applications can be bound to the SRL without having an open socket. Hence, their file names would be in use but they would not show up in the LISTOPENS MON command. These applications could cause an error when you start the Parallel Library TCP/IP subsystem. If you receive an error (“NLD fatal error, cannot open ZTCPSRL” or “file already exists”) when running your startup files, repeat the procedure for shutting down and clearing out the database, then add the names of those processes to the STOP PROCESS commands in the command file. (See ==Stop the Opener Processes== in Example 1-3 on page 1-23.) HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1- 31 Tasks: Stopping Parallel Library TCP/IP as a Generic Process Configuration Quick Start -> listopens mon $zztcp.* PTCPIP Listopens MON \HOME.$ZZTCP.#ZPTM0 Openers $ZLIS3 $ZLIS3 $ZLIS3 $ZTEL3 PPID 0,295 0,295 0,295 0,277 BPID PLFN 5 6 7 3 BLFN 0 0 0 0 Protocol TCP TCP TCP TCP Lport echo finger ftp telnet In the above sample display, you would record all the opener processes: $ZPT0, $ZTN0, $ZTF0, $Z0KW, $Z0KX, $ZTN0, and $Z07S to be stopped in Step 6. 5. List the names of the TCPSAM processes running in the Parallel Library TCP/IP environment. Enter the following command at the SCF prompt: ->LISTDEV TCPIP Make a note of the names of the running TCPSAM processes (indicated in the last field of the Program column). The following sample display results from the LISTDEV TCPIP command and shows all the TCP/IP processes running in the system: -> listdev tcpip LDev 170 186 215 240 Name $ZTC1 $ZTC0 $ZTC01 $TCPS1 PPID BPID 0,291 1,289 1,293 0,302 0,285 1,313 1,267 0,0 Type RSize Pri Program (48,0 ) 32000 200 \HOME.$SYSTEM.SYS07.TCPIP (48,0 ) 32000 200 \HOME.$SYSTEM.SYS07.TCPIP (48,0 ) 57344 201 \HOME.$SYSTEM.SYS07.TCPSAM (48,0 ) 57344 201 \HOME.$JER01.THJAGUAR.TCPSAM In the above display, the TCPSAM processes are $TCPS1, and $ZTC01. 6. Stop all openers of the TCPMONs (from Step 4). (Note that LISTNER and TELSERV do not support the SCF ABORT command so you must use the TACL STOP command to stop those processes.) Enter the following commands at the TACL prompt: >STOP PROCESS $ZLIS3 >STOP PROCESS $ZTEL3 7. Abort all TCPSAM processes (from Step 5). Enter the following commands at the SCF prompt: ->ABORT PROCESS $ZTC01 ->ABORT PROCESS $TCPS1 8. Issue the following SCF command to the NonStop Kernel subsystem: -> ABORT PROCESS $ZZKRN.#ZZTCP HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 1- 32 2 Introduction This section describes the purpose of Parallel Library TCP/IP, its architecture, benefits, components, features, and relationship to other HP products, and gives some how-to information such as locating the name of a socket access method (TCPSAM) and configuring the subsystem for round-robin distribution of incoming connection requests. Parallel Library TCP/IP is a new HP product that provides increased performance and scalability. For comparison, the previous product, NonStop TCP/IP, is referred to in this manual as conventional TCP/IP. The subsystem name for conventional TCP/IP is TCPIP. The subsystem name for Parallel Library TCP/IP is PTCPIP. Parallel Library TCP/IP coexists with conventional TCP/IP on NonStop S-series servers (see Strategy for Coexistence with Conventional TCP/IP on page 4-7). Parallel Library TCP/IP supports Ethernet, Fast Ethernet, and Gigabit Ethernet adapters (E4SAs, FESAs, GESAs and G4SAs) only (see the Ethernet Adapter Installation and Support Guide, the Fast Ethernet Adapter Installation and Support Guide, the Gigabit Ethernet Adapter Installation and Support Guide and the Gigabit Ethernet 4-Port Adapter Installation and Support Guide for information on installing these adapters). Use conventional TCP/IP for ATM, X.25 and token-ring support. Section 4, Managing the Parallel Library TCP/IP Subsystem contains important information about managing the system configuration database and coexistence with conventional TCP/IP. Be sure to read that section as well as this one. In addition, read the TCP/IP (Parallel Library) Migration Guide for migration considerations and a summary of the differences between conventional TCP/IP and Parallel Library TCP/IP. Background In the following discussions, the terms “physical port” and “PIF” are used extensively and interchangeably. A PIF is part of the ServerNet LAN Systems Access (SLSA) subsystem and represents the physical port on the adapter. When you configure Parallel Library TCP/IP, you actually use the “LIF,” which is another SLSA object that represents the logical interface to the port. The LIF is associated with the PIF which, in turn, represents the physical port. In Parallel Library TCP/IP as in conventional TCP/IP, you configure a SUBNET and assign it an IP address and the name of a LIF. In this discussion, the SUBNET is configured with an IP address and LIF which the SLSA subsystem then associates to the PIF/physical port. The architecture introduced by the NonStop S-series servers allows all processors in a system to access an adapter. Parallel Library TCP/IP takes advantage of this architecture by using the communications adapter and the ServerNet™ cloud to route packets directly to the processor containing the application. By directly routing packets to the correct processor from the adapter, Parallel Library TCP/IP eliminates the message-system hop that occurred between processes in the conventional TCP/IP architecture. By eliminating message-system hops, Parallel Library TCP/IP reduces the total path length from the application to the wire. This path-length reduction reduces individual HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 2 -1 Single IP Host Introduction request latency. In addition, more requests per second can be serviced using the same processor cost, resulting in higher throughput. Single IP Host In conventional TCP/IP, if you ran multiple process instances of a listening application in multiple processors (to increase computing power), you needed a different TCP/IP process (one per listening application process instance) in each processor. Each TCP/IP processes required a unique physical port (PIF) and presented a unique IP host to the outside world. Parallel Library TCP/IP allows multiple application process instances running in different processors to be presented to the outside world as a single IP host, because by using Parallel Library TCP/IP, you can run multiple process instances of a listening application in multiple processors, all sharing the same PIF. ServerNet™ allows all processors in a clustered system to access the same PIF; Parallel Library TCP/IP allows applications in different processors to access the same PIF and share a common listening TCP port number. Figure 2-1 shows the multiple IP hosts in the conventional TCP/IP environment discussed above, and Figure 2-2 on page 2-3 shows the single IP host possible in Parallel Library TCP/IP. Figure 2-1. Multiple IP Appearance, Conventional TCP/IP Message System Inter-Process Transfer Processor 0 Web 1 Processor 1 Processor 15 Web 2 Web 16 Message System Inter-Process Transfer TCP/IP 1 TCP/IP 2 TCP/IP 16 LAN Adapters LAN Adapters LAN Adapters VST0201.vsd HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 2 -2 Single IP Host Introduction Figure 2-1 on page 2-2 shows multiple instances of a Web application scaled to handle heavy traffic by running in all 16 processors. This configuration requires a different IP host for each of those Web applications. Figure 2-2 also shows multiple instances of a Web application scaled to handle heavy traffic by running in all 16 processors. However, in the Parallel Library TCP/IP environment shown in Figure 2-2, only one IP host is required for all those application instances. Figure 2-2. Single IP Appearance, Parallel Library TCP/IP Processor 0 MON 1 Web 1 TCP/IP Library LAN Driver DIH Processor 1 MON 2 Web 2 TCP/IP Library LAN Driver DIH ServerNet Processor 15 MON16 Web 16 TCP/IP Library LAN Driver DIH TCP/IP library is loaded into the application process space so no outbound messagesystem inter-process hop is needed for data transfer. LAN Adapter 1.2.3.4 VST0202.vsd Figure 2-1 on page 2-2 also shows the message system inter-process transfer that occurs for data transfer in conventional TCP/IP. Figure 2-2 shows that the data transfer in the Parallel Library TCP/IP environment happens within the library, a much faster way to transfer data. The differences between the data flows in the two environments is explained in more detail in Architectural Overview on page 2-9. Remote clients trying to connect to the Parallel Library TCP/IP network only need to know a single IP address to receive the processing power of up to 16 processors within the system. For multiple processors to share a physical port, a new, Parallel Library TCP/IP feature called round-robin filtering must be enabled. Round-robin filtering allows an adapter to distribute incoming requests among different listening processes. (See Round-Robin Filtering on page 2-4.) For example, a Web server residing in each processor could be configured to receive inbound requests in a round-robin manner. To use the roundHP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 2 -3 Introduction Round-Robin Filtering robin feature, however, you must explicitly configure it; the default configuration is for non-round-robin. (See Round-Robin Filtering.) Subnet-Level Binding: How to Isolate Subnets in a Single-IP Environment When you configure the SUBNET object in both conventional TCP/IP and Parallel Library TCP/IP, you specify a subnet IP address which associates the subnet with a particular physical interface (PIF). (See ADD SUBNET Command for TCPMAN on page 5-21.) In conventional TCP/IP, you could easily isolate an application on a particular subnet by associating the application with a separate TCP/IP process. By contrast, in Parallel Library TCP/IP, all applications share a single subnet and physical interface. In this configuration, you can still achieve application isolation on particular subnets/PIFs by using subnet-level binding. In Parallel Library TCP/IP, if you want to force traffic from a particular subnet to go to a particular application, bind the application to that subnet’s IP address rather than binding the application to INADDR_ANY. When the application is bound to a particular subnet’s IP address, TCP/IP directs traffic coming in from that subnet to that socket only. Traffic coming in from other subnets would not be directed to that socket. With subnet-level binding, you have a one-to-one correspondence between a PIF and a socket and correspondingly between an application and that socket. So if you have multiple application instances, each having performed a subnet-level socket bind, then traffic coming in on one subnet would go to application one, traffic from the next subnet would go to application two, and so on. Alternatively, you can have multiple application instances, each doing INADDR_ANY socket binds, and causing traffic from any of the subnets that are configured to be distributed to any of those sockets.. Round-Robin Filtering Parallel Library TCP/IP uses filters in the adapter to provide a new functionality called round-robin filtering. Round-robin filtering allows the adapter to distribute incoming connections to multiple listening processes in different processors sharing the same port (PIF). Round-robin filtering refers to the distribution of incoming connections to the first listening process in line, then the second, then the third, and so on, until the last listening process is reached, at which point the distribution returns again to the first listening process in line. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 2 -4 Introduction Round-Robin Filtering Configuring the System to use Round-Robin Filtering You must set the following DEFINE to enable round-robin filtering on your server processes: ADD DEFINE =PTCPIP^FILTER^KEY, class map, file file-name file-name is arbitrary but serves as the key or password. This key provides a measure of security: only users who know the key can access round-robin on the port. It can be up to eight alphanumeric characters; the first character must be a letter. The default setting is non-round-robin. If you don’t specify the file-name, round-robin filtering does not take effect. In this case, whoever registers the first application has exclusive use of the port. Each application that uses round-robin filtering on the same port must use the above ADD DEFINE with the same key. If you specify only the PTCPIP^FILTER^KEY, all applications that share that DEFINE also share all ports. To limit the shared ports, add one or both of the following DEFINEs: ADD DEFINE =PTCPIP^FILTER^TCP^PORTS, FILE Pstartport.Pendport ADD DEFINE =PTCPIP^FILTER^UDP^PORTS, FILE Pstartport.Pendport The startport and endport variables are integers specifying the allowable port range. The =PTCPIP^FILTER^TCP^PORTS key limits the shared TCP ports to the range defined in startport and endport. The =PTCPIP^FILTER^UDP^PORTS key limits the shared UDP ports to the range defined in startport and endport. Ports outside those ranges are not shared. You must always specify the =PTCPIP^FILTER^KEY DEFINE to enable round-robin filtering. If you want to limit shared TCP and UDP ports, add the appropriate DEFINE after the =PTCPIP^FILTER^KEY DEFINE. Port Collision Considerations for Listening Processes When you configure a set of listening processes for round robin, do not allow their primary and backup processors to overlap. That is, if you configure primary and backup listening processes, do so in distinct pairs. For example, if you have four processors, 0 through 3, and you want to configure primary and backup TELSERV processes for round-robin distribution, configure a primary and backup TELSERV pair in processors 0 and 1 and another primary and backup TELSERV pair in processors 2 and 3. Only one listening process per processor per port is allowed. If the processor running the primary listening process fails, the backup process in the other processor takes over listening on that port but if another listening process in the backup processor is already listening on the same port, the backup process receives an error and cannot listen on that port. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 2 -5 Introduction Scalability UDP Port Considerations If a process maintains a context for its messages, the process should not use roundrobin filtering on shared UDP ports. The processes sharing the UDP port should not maintain a context for previous messages because a sequence of messages might not be delivered to the same socket if the port is shared. In fact, with round-robin enabled, a sequence of messages is distributed to each of the port-sharing sockets, in turn. If, for example, an application assumes that all packets from a given source will be directed to it (the application process), the application must assume that it is the only instance of itself on that UDP port. If another instance of itself is sharing the UDP port, packets from the same source could go to two different instances of the application process resulting in one of the application processes missing some of the packets destined for it. To run multiple instances in parallel for applications which must, in the course of normal operation, maintain a context across multiple received messages, you can circumvent the problem introduced by round-robin filtering by changing the application to use subnet-level binding. (See Subnet-Level Binding: How to Isolate Subnets in a Single-IP Environment on page 2-4). Changing the application to use subnet-level binding allows one instance of the application for each subnet to be supported by Parallel Library TCP/IP, while still sharing the same port number. Parallel Library TCP/IP distributes incoming packets that came in from one subnet only to the application bound to that subnet. Thus, with subnet-level binding, the packets received by the application retain their contexts. Subnet-level binding circumvents the problem introduced by round-robin distribution of incoming packets among sockets sharing the same port. Scalability Having a TCP/IP stack in each processor tied together by a single IP address provides parallelism and thus, scalability; incoming connection requests directed to a specific LAN adapter can be distributed to one or more processors within a cluster. (See Introduction and Definitions on page 3-1 for a definition of scalable.) Parallel Library TCP/IP performs direct distribution of data flows to the processors containing the sockets that applications use. By doing so, Parallel Library TCP/IP allows applications to scale in parallel yet preserve the external image of a single IP host. (Section 3, Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments, provides examples that demonstrate this scalability.) Transparency Existing TCP/IP applications using the D30 and above socket library run transparently on Parallel Library TCP/IP without the need to be recompiled. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 2 -6 Introduction Ethernet Failover Ethernet Failover Ethernet failover, available with the G06.10 RVU of Parallel Library TCP/IP, provides fault tolerance at the adapter level. With Ethernet failover, you can configure your network to continue running if an adapter fails or during maintenance and replacement of an adapter. Ethernet failover allows TCP and UDP sessions to continue operating if there are cabling or adapter failures. With Ethernet failover, network traffic automatically migrates from the faulty LIF to the working LIF. Gains in Scalability Unlike other failover implementations, Parallel Library TCP/IP Ethernet failover does not require one of the LIFs to act as a “hot standby” in anticipation of a failure. Both LIFs are active, allowing inbound and outbound network traffic to be distributed between them. Therefore, you gain scalability when all your adapters are functioning correctly. In Parallel Library TCP/IP, a total of 64 SUBNETs can be configured. To achieve a failover configuration, two SUBNETs are associated as a failover pair, so a maximum of 32 failover pairs can be configured. Shared and Non-Shared IP Addresses With Ethernet failover, a single IP address can be shared between both LIFs (referred to as shared IP), or each LIF can be configured to have its own IP address (referred to as non-shared IP). HP recommends the shared IP configuration for most cases because shared IP provides a bandwidth advantage for outbound traffic. With shared IP, outbound traffic can flow over either SUBNET, and this effectively doubles the throughput capacity for outbound traffic. New connections are distributed across both adapters, but won't necessarily be distributed in a balanced manner. The benefits of shared IP are increased bandwidth for outbound traffic and flexibility for inbound traffic. Non-shared IP can provide the extra bandwidth for outbound traffic that shared IP provides if you add two routes to each subnet and if the application has selected INADDR_ANY as a source IP address (allowing Parallel Library TCP/IP to choose the interface to assign to it). If the application binds to a source IP address, Parallel Library TCP/IP assigns the outbound traffic to the subnet assigned to that IP address. Non-shared IP allows you to control the inbound traffic load, forcing the connections to be distributed over the two interfaces presented by the different IP addresses. This is handy when you have limited hardware resources or you want to maximize the use of LIFs. Both LIFs of a failover pair must be cabled to the same network segment. If different IP addresses are used, the IP addresses must be on the same network subnet. When a shared IP failover pair is configured, Parallel Library TCP/IP distributes new sessions over the two LIFs of the pair. In a similar fashion, each session is assigned one of the LIFs from the pair for its outbound traffic. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 2 -7 Introduction Ethernet Failover A shared IP failover pair also requires that a second IP address, referred to as the reserved IP address, be configured on each SUBNET in the pair. This address must be unique in the network and must reside on the same network subnet as the shared IP address. Parallel Library TCP/IP uses the reserved IP address when issuing Address Resolution Protocol (ARP) requests to resolve the media access control (MAC) address to an IP address. By using this reserved IP address, the ARP request does not affect ARP table entries in other network devices that may currently have sessions to the failover pair's shared IP address. When an adapter failure or cabling problem is detected on one of the LIFs of a failover pair, a gratuitous ARP packet containing the IP address of the failed LIF is sent using the operational LIF. This causes all devices on the network segment that had ARP table entries for this address to update their entries with the new MAC address of the operational LIF. Traffic for all the sessions on the failed LIF migrates to the operational LIF. When the failure is resolved and the failover pair is configured for non-shared IP, Parallel Library TCP/IP issues another gratuitous ARP packet on this LIF using its IP address. This causes all the devices that have sessions to this IP address to change their ARP table entries again and move the traffic to the new operational LIF. In the case of shared IP, the gratuitous ARP is not sent but new sessions are distributed over both LIFs of the failover pair. If configured for failover, during an adapter upgrade or removal, Parallel Library TCP/IP detects the LIF unavailability and automatically moves traffic over to the associate LIF. This feature eliminates the need for manual intervention for migrating traffic. Reinsertion of a replacement adapter (of the same type and in the same slot) automatically is detected and the replacement is initialized in the desired failover configuration. Similarly, if a LIF fails, Parallel Library TCP/IP detects it and moves traffic to the associate LIF. Note. If you have configured Ethernet failover as non-shared IP, you cannot have a failover pair consisting of addresses on different subnets. Configuration Guidelines The following are guidelines to use when configuring Parallel Library TCP/IP with Ethernet failover: • • • When selecting the LIF pair for the failover SUBNET pair, you should select LIFs on different adapters. When using Fast Ethernet adapters and Gigabit Ethernet adapters connected directly to Ethernet switches, failover recovery time may be impacted by the spanning tree feature used in a switch. When using multiple failover pairs on the same network subnet and adding static routes, it is best to add a copy of each route to one SUBNET in each failover pair. This increases the availability of the routes should both SUBNETs comprising a failover pair become unavailable. It also allows Parallel Library TCP/IP to distribute outbound connections over the failover pairs when the source IP address is not HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 2 -8 Introduction Architectural Overview selected by the application. In this case, static routes need to be added to both subnets of the pair if both LIFs are to participate. • • • • Ethernet failover may not function when directly connected to a firewall that uses Ethernet address (MAC) to IP address filtering. This problem can be overcome by adding a router between the LIFs and the firewall. When configuring multiple shared-IP failover pairs, the reserved IP address cannot be shared between pairs. If you use IP alias addresses, they must be added to both SUBNETs of a failover pair to preserve their availability during a failure. Also, alias IP addresses can only be shared between SUBNETs configured as a failover pair. When a pair of LIFs has been configured for Ethernet failover, separating them for use on distinct subnets requires a manual deletion of the SUBNET objects to disassociate them. The LIFs can then be used in the non-failover configuration (that is, one LIF per subnet). To configure your Parallel Library TCP/IP environment for Ethernet failover, you must use both the ADD SUBNET and ALTER SUBNET commands. See the ADD SUBNET Command for TCPMAN on page 5-21 and ALTER SUBNET Command for TCPMAN on page 5-30. Architectural Overview These are the product modules for Parallel Library TCP/IP: • • • • • • TCP/IP Manager Process (TCPMAN) TCP/IP Monitor Process (TCPMON) TCP/IP Socket Access Method (TCPSAM) TCP/IP Shared Runtime Library (SRL) (TCPLIB) TCP/IP Ptrace Product Module TCP/IP SCF Product Module For product numbers, see the TCP/IP (Parallel Library) Migration Guide. Some components of the Parallel Library TCP/IP subsystem are not involved in the data path; they exist for management purposes only. These components include the TCPMAN, TCPSAM, SCF, and PTrace product modules. Figure 2-3 on page 2-10 shows the data paths in Parallel Library TCP/IP and compares them to the data paths in conventional TCP/IP. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 2 -9 TCPMAN Introduction Figure 2-3. Data Path Comparison: Conventional vs. Parallel Library TCP/IP Conventional TCP/IP Parallel Library TCP/IP Application Inbound/Outbound Packets G06 TCP/IP Process Inbound/Outbound Packets TCPMON Inbound Packets Application TCPLIB Outbound Packets LAN Drivers/Interrupt Handlers SLSA DIH Legend Message system inter-process transfer Data transfer within the library VST0203.vsd Note that in Parallel Library TCP/IP, the TCPMON, TCPLIB and application components are in the data path. Figure 2-3 also shows that the TCP/IP library is pulled into the application context. Data transfer in the Parallel Library TCP/IP occurs within the library. The conventional TCP/IP environment shown in Figure 2-3 requires two message-system, inter-process, communication transfers. TCPMAN The manager process (TCPMAN) runs as a process pair and is the management point for the Parallel Library TCP/IP subsystem. Only one manager process pair exists for each system. TCPMAN is always named $ZZTCP. Once the monitors (TCPMON, see below) have been configured at least once in the system, TCPMAN starts the monitors along with any other subordinate objects from the system configuration database whenever TCPMAN is started. TCPMAN automatically configures one TCPMON for each system as the MASTER TCPMON. The MASTER TCPMON is usually the first TCPMON started. If the MASTER TCPMON processor fails, TCPMAN picks the TCPMON in the next configured processor as the MASTER. TCPMAN ensures one MASTER TCPMON always exists in the system. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 2- 10 Introduction TCPMON TCPMON The monitor object (TCPMON) provides the Parallel Library TCP/IP environment in each processor; one TCPMON exists in each configured processor. TCPMONs are controlled by the TCPMAN process. TCPMONs are named automatically. The naming convention for TCPMON is $ZZTCP.#ZPTMn where n is the processor number in which the TCPMON resides. The format for this processor number is hexadecimal (0-F). The TCPMON object has a MASTER attribute. The MASTER TCPMON receives and processes inbound frames that do not match any filter. The MASTER TCPMON also replies to all ICMP echo requests. TCPSAM The socket access method (TCPSAM) is a process pair provided for backward-compatibility for socket applications. TCPSAM provides applications with a name for a socket transport-service provider. By specifying the TCPSAM process as the name of the socket transport-service provider, the application programmer can access Parallel Library TCP/IP and gain the Parallel Library TCP/IP performance improvement without having to reconfigure the application. TCPSAM is provided for backward-compatibility only; no data passes through TCPSAM. See Figure 2-3 on page 2-10 for an illustration of the data path for Parallel Library TCP/IP. Any number of instances of TCPSAM can run in a system. The recommended naming convention for TCPSAM is $ZSAMx. The standard naming convention, $ZBnnn where nnn represents the LIF associated with the process, does not work in Parallel Library TCP/IP because the TCPSAM process, unlike the conventional TCP/IP process, is not associated with a specific LIF. Note that the Expand application expects a TCP/IP process name to start with Z, so you may want to ensure that at least one TCPSAM process starts with a Z. SRL Parallel Library TCP/IP places most of the protocol stack in a private shared runtime library (SRL) rather than in a process. Users of ZTCPSRL transparently use QIO functions without having to explicitly issue calls to initialize the QIO segment. This architecture allows TCP/IP to retain its context during processing and shortens the path-length. This library is dynamically loaded into the application’s process space as soon as the application issues a TCP/IP socket request. Figure 2-3 on page 2-10 shows the application laid over the TCP/IP library. Since the application invokes the HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 2- 11 Introduction PTrace TCP/IP library, TCP/IP retains the context of the application while processing the request. Caution. Parallel Library TCP/IP uses a private shared runtime library (SRL). If you run programs that use private SRLs and you are running Parallel Library TCP/IP, do not create private SRLs that overlap the following memory segments: 0x75800000 through 0X759FFFFF and 0x7FE00000 through 0x7FE1FFFF. In addition, QIO reserves 0x20000000 through 0x41FFFFFF. Locating the SRL In general, the Parallel Library TCP/IP SRL is located in the current SYSnn subvolume or in the same system subvolume as the system image file (OSIMAGE). You can ask your system administrator for the current SYSnn subvolume or you can find the OSIMAGE file by entering one of the following TACL commands: >FILEINFO $SYSTEM.SYS*.OSIMAGE or >STATUS 0,0 Once you know the location of the SYSnn, you also know the location of the SRL. The procedures in this manual have you use the TACL VOLUME command to ensure that you are located in the SYSnn where the SRL is located before issuing the OBEY command on the TACL command file. The TACL command file only specifies the file name (ZTCPSRL) for the SRL location and the system fills in your current location. By only specifying the file name of the SRL (ZTCPSRL) in the command file and following the procedure to ensure you are in the correct volume, you are assured of properly adding the SRL DEFINE. Programmatic Interfaces to the SRL Socket-program environments should be set up to use the correct SRL. Socket programs started in the same TACL command environment as the DEFINE inherit the DEFINE for the correct SRL (see Example 1-1 on page 1-8 and Example 1-2 on page 1-14). However, socket programs started in different TACL command environments need to either use a programmatic process-create or specify the SRL in the RUN command used to start the socket program. Alternatively, you can place the DEFINE for the SRL in the $SYSTEM.SYSTEM.TACLLOCL file so that all TACL users inherit the DEFINE. PTrace The PTrace product module formats the trace data records. SCF The SCF product module provides the command-line interface for managing the Parallel Library TCP/IP subsystem. Figure 2-4 shows a high-level view of the Parallel HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 2- 12 Introduction QIO Library TCP/IP subsystem and its relationship to the QIO and SLSA subsystems. The LAN drivers/interrupt handlers and the LAN adapters (for example, G4SA) are part of the SLSA subsystem. The Parallel Library TCP/IP subsystem components and part of the application run in the QIO shared memory segment. QIO The QIO subsystem has been enhanced as of G06.17 to allow you to have more control over certain aspects of memory management. You can now configure QIO to run in the Kseg2 memory segment and you can also control where QIO runs in the flat memory segment. Configuring QIO to run in Kseg2 can improve performance for Parallel Library TCP/IP but also imposes constraints that affect all QIO clients (including Parallel Library TCP/IP). As discussed in the QIO Configuration and Management Manual, you must consider these constraints as well as a variety of other factors before changing the default QIO configuration. Some of the constraints affecting Parallel Library TCP/IP (as well as other QIO clients) include the reduction of QIO memory space to 128 MB when QIO is moved to Kseg2. This impacts the number of LIFs that you can configure on your system because LIFs use QIO memory. It also impacts the number of sockets that can be opened because open sockets use QIO memory as well. 128 MB may not be sufficient for your Parallel Library TCP/IP or other QIO client needs. For details about planning for and using the new QIO features, see the QIO Configuration and Management Manual. Note. The default configuration for the QIO subsystem has not changed. Note. Whether you use the default QIO configuration or one of the newly supported custom configurations, you do not need to change anything in Parallel Library TCP/IP; all changes are made in the QIO subsystem. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 2- 13 QIO Introduction Figure 2-4. Parallel Library TCP/IP Subsystem Within the System SCF Commands and Responses DSM Management Applications SCF SPI-formatted messages SCP Open and Close Requests TCPMON QIO Shared Memory Segment TCPLIB TCPMAN Application TCPSAM Outbound Packets Inbound Packets LAN Drivers/Interrupt Handlers SLSA/DIH ServerNet Fabrics LAN LAN LAN Adapter Adapter Adapter SWAN Concentrator X.25 Network Public Data Network (PDN) Defense Data Network (DDN) Legend Message system inter-process transfer Library transfer VST0204.vsd Figure 2-4 also shows the management interfaces involved in running TCP/IP. The lines between the terminal, SCF, SCP, DSM and TCPMAN indicate management flow through the message system. The solid lines between the application and TCPSAM is also a message system transfer. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 2- 14 Introduction Parallel Library TCP/IP and Other Products The application, TCPMON, and TCPMAN communicate with each other through the TCP/IP library and transfer data within the library without message-system hops. See TCPMAN, TCPMON, and TCPSAM on page 2-11 for descriptions of these components of the Parallel Library TCP/IP subsystem. Parallel Library TCP/IP and Other Products Parallel Library TCP/IP allows transparent interactions with existing products. Products that use sockets to interface with TCP/IP need to select TCPSAM as the TCP/IP transport-service provider if they want Parallel Library TCP/IP instead of conventional TCP/IP. (Programming With the New Socket Provider (TCPSAM) on page 2-16 explains how to determine the TCPSAM name.) Note. Expand needs to have its TCP/IP transport provider name start with a Z. See the Expand Configuration and Management Manual for information about configuring Expand over Parallel Library TCP/IP. Caution. Parallel Library TCP/IP uses a private shared runtime library (SRL). If you run programs that use private SRLs and you are running Parallel Library TCP/IP, do not create private SRLs that overlap the following memory segments: 0x75800000 through 0X759FFFFF and 0x7FE00000 through 0x7FE1FFFF. In addition, QIO reserves 0x20000000 through 0x41FFFFFF. NonStop Kernel Subsystem and the System Configuration Database Parallel Library TCP/IP participates in the system configuration database (CONFIG) used by the persistence manager ($ZPM) in the NonStop Kernel subsystem. When you configure the MON, ROUTE, SUBNET, and ENTRY objects, those objects are stored in the system configuration database. If you also add the TCPMAN process to the system configuration database as a persistent, generic process, whenever you stop the TCPMAN or restart the system, the persistence manager restarts TCPMAN automatically and TCPMAN then retrieves and starts the MON, ROUTE, SUBNET, and ENTRY objects that are stored in the system configuration database. However, the persistence manager does not start TCPSAM processes so you must start those manually. See Managing the System Configuration Database on page 4-1 for information on how to add the TCPMAN process as a generic process and other details about managing the configuration database. The ABORT MON Command for TCPMAN on page 5-12 not only stops the monitors in the subsystem but also deletes them from the system configuration database. The DELETE ENTRY Command for TCPMAN on page 5-33, DELETE ROUTE Command for TCPMAN on page 5-34, and DELETE SUBNET Command for TCPMAN on page 5-35, deletes these objects from the system configuration database. For an example of these commands in a subsystem shutdown procedure, see Stopping Parallel Library TCP/IP and Clearing the Database on page 1-24. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 2- 15 Programming With the New Socket Provider (TCPSAM) Introduction Programming With the New Socket Provider (TCPSAM) Applications rely on the transport provider for making socket requests. With the choice of two different environments, the application programmer can now specify either the conventional TCP/IP or the Parallel Library TCP/IP environment just by choosing the appropriate transport-service provider. To select the Parallel Library TCP/IP environment, first determine the name of the TCPSAM process, then use that name in your socket calls. To determine the name of the TCPSAM process, issue the following SCF command: -> LISTDEV TCPIP This SCF LISTDEV command lists all the TCP/IP processes. A program name in the SCF LISTDEV display of TCPIP means that the process is a conventional TCP/IP process whereas a program name of TCPSAM means that process is a Parallel Library TCP/IP process. The following display shows a sample result of the SCF LISTDEV TCPIP command. 1 2 3 4 5 6 7 8 9 SCF - T9082G02 - (05AUT99) (26JUL99) - 12/22/1999 14:52:00 System \TIGGER Copyright Hewlett-Packard Company LDev Name 107 $ZTCP0 141 $ZTC03 154 $ZTC0 158 $ZTCP1 190 $ZTC02 PPID 0,285 3,269 0,299 1,293 1,310 BPID 1,287 0,0 1,286 0,302 0.0 Type (48,0) (48,0) (48,0) (48,0) (48,0) RSize 32000 57344 32000 32000 57344 Pri 200 201 200 200 201 Program \TIGGER.$SYSTEM.SYS03.TCPIP \TIGGER.$SYSTEM.SYS01.TCPSAM \TIGGER.$SYSTEM.SYS03.TCPIP \TIGGER.$SYSTEM.SYS03.TCPIP \TIGGER.$SYSTEM.SYS01.TCPSAM In this example, the processes $ZTC02 and $ZTC03 are TCPSAM processes. These are the processes you would select when specifying the transport-service provider name for your applications to access Parallel Library TCP/IP. Restrictions of Parallel Library TCP/IP The following features are not supported in Parallel Library TCP/IP: • • • ATM, X.25, and SNAP type subnets PMF and IOMF CRUs Network File System (NFS) For a complete list of migration issues for Parallel Library TCP/IP, see Summary of Differences Between Conventional TCP/IP and Parallel Library TCP/IP of the TCP/IP (Parallel Library) Migration Guide. RFC Compliance Parallel Library TCP/IP is based on the 4.4 BSD TCP/IP stack from Berkeley Software Design, Incorporated. How to Access Online Help To access online help for the Parallel Library TCP/IP subsystem, enter HELP at the command prompt, then enter PTCPIP at the SCF HELP prompt. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 2- 16 3 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments This section shows you how to configure your listening applications to take advantage of the architectural features of Parallel Library TCP/IP. The first part of this section describes different listening-application models and how those models can be configured to take advantage of Parallel Library TCP/IP. The second part provides configuration examples for each of the listener-application models. Finally, a configuration example is provided that emphasizes the networking aspect of configuring Parallel Library TCP/IP. As of the G06.14 RVU, complex, heavy-use SWAN configurations can benefit from using Parallel Library TCP/IP. The advantages of Parallel Library TCP/IP for SWAN are documented in this section (see Parallel Library TCP/IP for Complex, Heavy-Use WAN Environments on page 3-29). Introduction and Definitions In this discussion, scalable, parallel, and load-balancing mean: • • • Scalable — refers to the architectural capacity to grow to accommodate growing computing demands. A scalable architecture allows you to add processing power as your computing needs grow. Parallel — refers to the division of work among different processes and/or processors. Load-balancing — refers to algorithms that balance work-load between processes and/or processors. Scalable and parallel are closely related. An architecture is scalable if you can add parallel processing to it. By dividing the work-load among multiple processes and/or processors, you can scale your applications to meet increasing demand. However, parallel processing does not in and of itself provide scalability; you need load-balancing algorithms and/or architecture to avoid bottlenecks when your computing needs grow. With conventional TCP/IP, only one socket can be bound exclusively to a given incoming port number. With Parallel Library TCP/IP, multiple server-process instances in a system can all share the same incoming TCP (or UDP) port number if round-robin filtering is enabled. Round-robin filtering allows you to scale your system by multiplying the number of listening processes and by taking better advantage of the load-balancing applications available on NonStop S-series systems. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3 -1 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Four Listening Methods Four Listening Methods Networking methods fall into four types: • • • • Standard Listening Model Monolithic Listening Model Distributor Listening Model Hybrid Listener Model Standard Listening Model In this method, you have a single process which listens for incoming connections. When an incoming connection occurs, the listening process spawns a new server process (which can run in another processor) then passes that server process the caller's IP address and port number. That spawned process assumes ownership of the port and does its own socket I/O. This method allows for parallel instances of the server process. For example, the common LISTNER spawns instances of FTPSERV. The LISTNER process is the common point where all incoming connections are handled, but control and data flow is handed off to separate instances of server processes which can be distributed across processors. Because it passes connections off to other processors and can use all processors in the system, the standard listener model provides scalability. The standard listener model can benefit from the Parallel Library TCP/IP architecture because Parallel Library TCP/IP eliminates the hop that has to occur between the spawned server process and the processor where the TCP/IP process resides. That hop occurs on the server process’ first socket call and every subsequent send/receive call. In Parallel Library TCP/IP, those calls are handled locally, within the server’s processor. Figure 3-1 on page 3-3 shows a standard listener distributing a connection to a different processor. Step 3 of the conventional TCP/IP environment (of Figure 3-1 on page 3-3) shows that FTPSRV must hop to Processor 0 where the TCP/IP process resides to access the adapter. By contrast, in the Parallel Library TCP/IP environment, in Step 3, FTPSRV has direct access to the adapter. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3 -2 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Standard Listening Model Figure 3-1. Standard Listening Model Parallel Library TCP/IP Conventional TCP/IP 1. 1. Connection Req 1 LAN Adapter Processor 0 Connection Req 1 LAN Adapter LISTNER TCP/IP Processor 0 LISTNER TCP/IP Library 2. 2. Processor 0 Processor 0 TCP/IP LISTNER LISTNER Processor 1 TCP/IP Library LISTNER spawns FTPSRV, providing IP address and port # of connection 1 Processor 11 Processor FTPSRV LISTNER spawns FTPSRV, providing IP address and port number of connection FTPSRV TCP/IP Library 3. Processor 0 Processor 0 LISTNER Connection 1 LAN Adapter TCP/IP Processor 1 FTPSRV FTPSRV assumes ownership of connection 1 and communicates through the TCP/IP process in processor 0 LISTNER TCP/IP Library 3. Connection 1 LAN Adapter Processor 1 FTPSRV TCP/IP Library FTPSRV assumes ownership of connection 1 and communicates directly to adapter VST0301.vsd Configuration Example for the Standard Listening Model on page 3-11 tells you how to configure a similar configuration in Parallel Library TCP/IP. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3 -3 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Monolithic Listening Model Monolithic Listening Model In this method, the listener binds to a well-known port, (for example port 23 for TELSERV). Next, the listening process issues a standard accept call. Finally, the monolithic listening model then uses multi-threading to handle all connections within the same process creating sockets for the connections. The listening process is the common point where all incoming connections are handled as well as control and data flow. To achieve parallelism with the monolithic model in conventional TCP/IP, you run multiple instances of the listening process, each with a different well-known port number, and then inform the workstation clients to use a different destination port number. The monolithic listener model can benefit from the Parallel Library TCP/IP architecture because, with round-robin filtering enabled, all processors have access to the same port. Hence, you can run multiple copies of the listening process in different processors and bind them all to the same well-known port. By sharing the same port number among the processes, you no longer need to set up workstation clients with multiple port numbers to call. Figure 3-2 on page 3-5 compares the monolithic server model in the conventional TCP/IP and Parallel Library TCP/IP environments. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3 -4 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Monolithic Listening Model Figure 3-2. Monolithic: Listening Model Conventional TCP/IP IPC hop Processor 0 TCP/IP Process Processor 1 Monolithic Server Sockets Port xxxx (Exclusive) LAN Adapter Parallel Library TCP/IP Sockets Monolithic Server Monolithic Server Monolithic Server TCP/IP Library Port xxxx (Shared) TCP/IP Library Port xxxx (Shared) TCP/IP Library Port xxxx (Shared) TCP/IP Library LAN Adapter VST0302.vsd In the conventional TCP/IP environment shown in Figure 3-2, you have the typical monolithic listener configuration with just one instance of the listening process. It is shown in a different processor than the TCP/IP process to illustrate the remote IPC HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3 -5 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Distributor Listening Model hop required in the conventional TCP/IP environment and also because TCP/IP often runs in a different processor than the listening process for load distribution. In the Parallel Library TCP/IP environment, with round-robin filtering enabled, you can have multiple copies of the monolithic listener running in different processors, all sharing the same port (shown in Figure 3-2 on page 3-5). Because round-robin filtering is enabled, the adapter distributes the incoming connections to the different listening processes, and the listening processes, in a sense, distribute those connections within its own process by creating sockets for each connection. An example of how this kind of dual-layer distribution could work is described in the following steps: 1. The first incoming connection request goes to the listening process in the first processor and the listening process creates a socket for the connection. 2. The second connection request goes to the listening process in the second processor and the listening process creates a socket for the connection. 3. The third connection request goes to the listening process in the third processor and the listening process creates a socket for the connection. 4. The fourth connection request goes to the listening process in the first processor and the listening process creates a second socket for that connection. 5. The fifth connection request goes to the listening process in the second processor and the listening process creates a second socket for that connection. In summary, you can use the Parallel Library TCP/IP feature of all processors having access to one adapter to scale your monolithic listener server process (by adding more processes in different processors and distributing the work load among them), and eliminate an IPC hop between the server process and the TCP/IP process (which increases transmission speed). Configuration Example for the Monolithic Listening Model on page 3-15 tells you how to configure this listener model. Distributor Listening Model This method uses an interface process which serves as a distributor by using NonStop inter-process communication (IPC) to talk to multiple back-end server instances. In this method, the distributor binds to a well-known port and then accepts multiple connections by creating a socket for each connection. The distributor listener then performs the accepts, sends, receives, and so on, on those sockets, on behalf of the back-end servers. The distributor handles all data flow and control, and forwards the received data using NSK inter-process communication to back-end server processes. A good example of this method is a TCP/IP application developed for the Pathway environment. The Pathway environment provides a rich set of server process-management facilities and load-balancing facilities. A front-end distributor process using a set of verbs collectively referred to as PATHSEND, communicates with the back-end servers. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3 -6 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Distributor Listening Model PATHSEND allows load-balancing algorithms to distribute the data to the back-end server instances. The distributor model achieves some parallelism and load-balancing because of the use of the multiple, back-end server instances. However, the distributor model is limited by the fact that all data must flow through the distributor to the back-end server processes through PATHSEND. This situation creates a potential bottle-neck in the distributor. Figure 3-3 on page 3-8 shows the distributor listener model in conventional TCP/IP. To scale to accommodate growing computing needs in this conventional TCP/IP example, three instances of the distributor are running in three processors (3, 4, and 5) distributing connections to three sets of server instances in three different processors (0, 1, and 2). While you can achieve scalability in the conventional TCP/IP model by running multiple distributors in this manner, there are two hops involved in data flow (one inter-process hop between the distributor and the server and one between the distributor and the TCP/IP process). In addition, each processor must have its own physical interface (PIF) on the adapter. Finally, having run a TCP/IP process for each distributor, each distributor appears to be on a different IP host to the outside world. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3 -7 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Distributor Listening Model Figure 3-3. Distributor Listening Model in Conventional TCP/IP Processor 2 Processor 1 Processor 2 Processor 1 Processor 0 Processor 0 Processor 2 Processor 1 Processor 0 Processor 0 Processor 0 Processor 0 Server Classes Server A, D, G Classes Server A, D,Classes G A, D, G Server Classes Server A, D, G Classes Server A, D,Classes G A, D, G Server Classes Server A, D, G Classes Server A, D,Classes G A, D, G Distributor Distributor Server Class Send or Other IPC, Hop 1 Distributor Sockets Hop 2 TCP/IP TCP/IP Processor 3 Processor 4 Processor 5 LAN Adapter LAN Adapter LAN Adapter TCP/IP Three IP hosts VST0303.vsd The distributor listener model can benefit from the Parallel Library TCP/IP architecture in two ways: • • You can now run multiple distributor processes in multiple processors bound to the same port with round-robin filtering enabled. This arrangement allows you to spread the distributor’s work-load over as many processors as required achieving unlimited scalability while presenting a single IP host to the outside world. You shorten the path-length for data flow by eliminating the hop between the distributor and the TCP/IP process, thereby improving performance. Figure 3-4 on page 3-9 shows the distributor listener model in Parallel Library TCP/IP. The distributor can now be duplicated across processors and only one hop is required for outbound traffic because each processor housing the distributors has direct access to the adapter. Figure 3-4 also shows the round-robin distribution HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3 -8 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Hybrid Listener Model of connection requests performed by the adapter and the presentation of one IP host. Figure 3-4. Distributor Listening Model in Parallel Library TCP/IP Processor 2 Processor 1 Processor 2 Processor 1 Processor 0 Processor 0 Processor 2 Processor 1 Processor 0 Processor 0 Processor 0 Processor 0 Server Classes Server A, D, G Classes A, D,Classes G Server A, D, G Server Classes Server A, D, G Classes A, G ServerD, Classes A, D, G Server Classes Server A, D, G Classes A, D, G Server Classes A, D, G TCP/IP Library TCP/IP Library Distributor Distributor Server Class Send or Other IPC, Hop 1 TCP/IP Library Distributor Sockets No hop TCP/IP Library Processor 3 TCP/IP Library Processor 4 TCP/IP Library Processor 5 Round-robin distribution of incoming connection requests LAN Adapter Single IP host VST0304.vsd Configuration Example for the Distributor Listening Model on page 3-18 shows you how to configure this listener model. Hybrid Listener Model This method still has a distributor process using Pathway for managing and loadbalancing server process instances, but it uses the standard listener approach, handing off connections, so that data does not have to flow through the distributor process. This method de-couples the data and control flows. This de-coupling facilitates low-overhead scalability when combined with Parallel Library TCP/IP. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3 -9 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Hybrid Listener Model Figure 3-5 shows the hybrid listener model in conventional TCP/IP. The figure shows two distributors in processors 0 and 2 distributing connections and data and control flow to servers in their own and in one other processor. The servers inherit the connections and send data flow to the TCP/IP process. Note that each TCP/IP process represents a different IP host. Also note the remote IPC hop between the servers in the remote processors (1 and 3) and the processors with the TCP/IP stack (0 and 2). In addition, note the local IPC hop between the servers and the TCP/IP processes in the processors containing the TCP/IP stacks. Figure 3-5. Hybrid Listening Model in Conventional TCP/IP Processor 0 Processor 1 Server Server Server 1 Local IPC hop Server Server Server 2 1 Connection Distribution 1 Server Server Server 2 2 IPC Hop Distributor TCP/IP Process LAN Adapter Processor 3 Server Server Server 2 Distributor TCP/IP Process Processor 2 1 LAN Adapter Two IP Hosts Legend 1 Data flow 2 Connection hand offs VST0305.vsd The hybrid listener model can benefit from Parallel Library TCP/IP because you can enable round-robin filtering and run multiple instances of the hybrid distributor in different processors bound to the same port. Round-robin filtering enables the adapter to distribute incoming connections to the multiple distributor instances, for connection parallelism. Then the distributors hand off the connections to available servers, allowing reactive distribution. The servers have direct access to the TCP/IP library in their own processors, eliminating both the remote IPC hops and local IPC hops. All the distributors share the same port and present a single IP host to the outside world. Figure 3-6 shows a configuration (simplified for comparison) with two distributors: one in processor 0 and one in processor 2. The distributors hand off connections to servers in their own processor and in one remote processor. All connections go through the same G4SA LIF and the G4SA round-robin distributes connections among the distributors. A single IP host is presented to the outside world and the inter process HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3- 10 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Configuration Example for the Standard Listening Model hop between the remote processors (1 and 3) is eliminated as data flows directly to the local TCP/IP library and then out to the adapter. Figure 3-6. Hybrid Listening Model in Parallel Library TCP/IP Processor 0 Server Server Server Data flow Processor 1 Server Server Server Server Server Server Processor 3 Server Server Server No IPC hop Distributor Distributor TCP/IP Library Processor 2 TCP/IP Library TCP/IP Library TCP/IP Library G4SA One IP Host VST0306.vsd A similar configuration example is shown in Configuration Example for the Hybrid Listening Model on page 3-21. Configuration Example for the Standard Listening Model This example demonstrates the standard listener model discussed above (see Standard Listening Model on page 3-2) configured with round-robin filtering enabled. The startup files for establishing the standard listener model include commands to start several processes that are essential in the Parallel Library TCP/IP environment, commands to set different parameters, and SCF commands for adding and starting the SLSA DEVICE and Parallel Library TCP/IP PROCESS, SUBNET, and ROUTE objects. The files used in this example for starting and configuring the Parallel Library TCP/IP environment include: • • • TCPIPUP1 starts the Parallel Library TCP/IP environment. TCPIPUP2 starts the LISTNER processes. TCPIPUP2 also accesses an SCF file that adds, configures, and starts Parallel Library TCP/IP objects. SCFSBNT adds, configures, and starts the subnets and routes. The first configuration example of a Parallel Library TCP/IP environment is shown in Figure 3-7 on page 3-12. One LISTNER is configured in Processor 0 and its backup process is in. As it receives connection requests on port 21, it spawns FTPSERV HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3- 11 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Configuration Example for the Standard Listening Model processes and hands off the connection to those processes. The FTPSERV processes then assume direct control of the connections. Figure 3-7. Standard Listening Model Configuration Example: LISTNER NonStop S-Series System With 4 Processors 1. Processor 0 LAN Adapter LISTNER A connection request comes in to the LISTNER 4. 150.50.130. 2 Four processors handling four connections. LAN Adapter TCP/IP Library 150.50.130.2 Processor 0 LISTNER FTPSERV TCP/IP Library Backup LISTNER 2. LISTNER spawns an FTPSERV process and hands off the connection to it Processor 0 LAN Adapter LISTNER Processor 1 FTPSERV TCP/IP Library Processor 2 Processor 1 FTPSERV FTP SERV TCP/IP Library TCP/IP Library 3. FTPSERV in Processor 1 has direct access to the adapter. 150.50.130.2 Processor 3 Processor 0 LAN Adapter LISTNER FTP SERV Processor 1 FTPSERV TCP/IP Library VST0307.vsd HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3- 12 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Configuration Example for the Standard Listening Model Startup Files You must start the Parallel Library TCP/IP environment and create a TCPSAM process before running TCPIPUP2. The command file TCPIPUP1 performs these tasks. The TCPIPUP1 File The following TACL command file starts the Parallel Library TCP/IP environment including the TCPMAN process, the monitors, and the TCPSAM process. Issue the OBEY command on this file first, unless the Parallel Library TCP/IP environment is already running. Substitute real values for variables (indicated in italics). (See the Configuration Form 1 on page 1-5 and page 1-12 for procedures for determining these values.) Before running this file, change directories to the volume/subvolume SYSnn. For procedures for locating the SYSnn, see Step 7 on page 1-9. After you have changed to the volume containing the SRL (SYSnn), remember to fully qualify the name of the TCPIPUP1 command file since you probably did not store it in $SYSTEM.SYSnn. Example 3-1. TCPIPUP1 Command File DELETE DEFINE =_SRL_01 DELETE DEFINE =TCPIP^PROCESS^NAME CLEAR ALL TCPMAN/NAME $ZZTCP,TERM $ZHOME, OUT $ZHOME,& CPU 1,NOWAIT/3 SCF/INLINE/ INLPREFIX + + ASSUME PROCESS $ZZTCP + START MON * ==Give TCPMONs time to start + DELAY 21 + ALTER MON *,HOSTNAME "BOBAFET7" + ALTER MON *,HOSTID 150.50.130.2 INLEOF ADD DEFINE =_SRL_01,CLASS MAP,FILE ZTCPSRL & TCPSAM /NAME $ZSAM3, TERM $ZHOME, OUT $ZHOME, NOWAIT, CPU 0/1 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3- 13 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Configuration Example for the Standard Listening Model The TCPIPUP2 File The following TACL command file starts the processes, adds and starts subsystem objects through SCF, and sets appropriate parameters. To add comments, use the word “comment” or a double equal sign (==). Some lines are discussed separately after the example. Substitute real values for variables (indicated in italics). (See the Configuration Form 1 on page 1-5 and page 1-12 for procedures for determining these values.) Example 3-2. TCPIPUP2 for the LISTNER Process comment comment comment comment comment comment comment comment comment comment ==== TCPIPUP2 =========TCPIPUP2 ======== TACL command file to bring up Parallel Library TCP/IP subsystem Use DNS for name resolution; (no host file DEFINE) DELETE DEFINE =TCPIP^HOST^FILE ADD and START SUBNETS SCF/IN $SYSTEM.TCPIP.SCFSBNT/ Initialize LISTNER for FTPSERV, ECHOSERV, and FINGSERV Define the TCPSAM process for the LISTNER to use DELETE DEFINE =TCPIP^PROCESS^NAME ADD DEFINE =TCPIP^PROCESS^NAME, class map, file $ZSAM0 Start the LISTNER LISTNER/NAME $LSN0, NOWAIT, PRI 170, CPU 0/1, & OUT $ZHOME, TERM $ZHOME,& $SYSTEM.ZTCPIP.PORTCONF ====== END OF TCPIPUP2 ==== END OF TCPIPUP2 == The RUN command: LISTNER/NAME $LSN0, NOWAIT, PRI 170, CPU 0/1, & OUT $ZHOME, TERM $ZHOME, $SYSTEM.ZTCPIP.PORTCONF LOG_GOTCONN starts the LISTNER processes responsible for starting the ECHO, FINGER, and FTP servers when the LISTNER process receives a client request. Run these processes at a high priority. This command also specifies the location of the PORTCONF file used to designate which ports the LISTNER is to listen to. Since the LISTNER requires privileged access to some Parallel Library TCP/IP ports, always log on with a super group ID. Note. The LOG_GOTCONN option enables the logging of "got connection" messages. You can use these messages to monitor FTP requests. If you do not specify the LOG_GOTCONN option, the "got connection" messages are not logged. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3- 14 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Configuration Example for the Monolithic Listening Model The SCFSBNT File The SCFSBNT file adds and starts subnets and routes. Substitute real values for variables (indicated in italics). (See the Configuration Form 1 on page 1-5 and 1-12 for procedures for determining these values.) Example 3-3. SCFSBNT File for TCPIPUP2 === SCFSBNT ===== SCFSBNT ==== SCFSBNT ======== == SCF command file to ADD and START SUBNETs == This file is created for use by The TCPIPUP2 File, The TCPIPUP3 File, == The TCPIPUP4 File, and The TCPIPUP5 File ALLOW ALL ERRORS ALLOW ALL WARNINGS == ADD AND START SUBNET $ZZTCP.*.SN0 ASSUME PROCESS $ZZTCP ADD SUBNET SN0,TYPE ETHERNET,DEVICENAME LAN01,IPADDRESS 150.50.130.2, & SUBNETMASK %HFFFFFF00 == ALTER SUBNET for LOOPBACK STOP SUBNET LOOP0 ALTER SUBNET LOOP0, IPADDRESS 127.1 == START SUBNET on all monitors START SUBNET * =========== END OF SCFSBNT ============= END OF SCFSBNT =========== For the ADD SUBNET command, the subnet name can be anything under seven alphanumeric characters long beginning with a letter. The DEVICENAME attribute is required. The DEVICENAME attribute specifies the logical interface (LIF) name associated with the adapter accessed by the Parallel Library TCP/IP processes. (See Step e on page 1-4 for procedures for determining an appropriate LIF.) LOOPBACK When the monitors are started, a subnet named LOOP0 is added automatically. This subnet provides loopback capability without requiring the use of the TCP/IP network. When this LOOP0 subnet is created, it has an address of 0.0.0.0 in dotted decimal form. You must change this address; use the command: ALTER SUBNET LOOP0, IPADDRESS 127.1 The address 127.1 (or 127.0.0.1) is the standard for loopback operation. Configuration Example for the Monolithic Listening Model This example demonstrates the monolithic listener model discussed above (see Monolithic Listening Model on page 3-4) configured with round-robin filtering enabled. The startup files and subnet initiation files for establishing the Parallel Library TCP/IP environment and subnets are the same as Configuration Example for the Standard Listening Model on page 3-11. In this example, TCPIPUP3, the main command file, has been changed to use TELSERV instead of LISTNER to demonstrate the monolithic listener model. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3- 15 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Configuration Example for the Monolithic Listening Model Figure 3-8 shows the round-robin feature in this configuration. A TELSERV process is configured in each processor, the processes bound to the same port, and round-robin filtering enabled on those processes. (See Round-Robin Filtering on page 2-4.) Figure 3-8. Configuration Example for Monolithic Listening Model: TELSERV TCP/IP Serv ices Processor 0 TCP/IP Services Processor 1 TCP/IP Services LAN Adapter Processor 2 TCP/IP Services Host Processor 3 TELSERV TELSERV TELSERV TELSERV VST0308.vsd The first connection request establishes a connection to one of the TELSERV process’ sockets. The adapter then routes the next connection request to the next TELSERV process in the next processor, and so on. (See Monolithic Listening Model on page 3-4 for more information about the dual-layer distribution of a round-robin configured monolithic server.) HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3- 16 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Configuration Example for the Monolithic Listening Model The TCPIPUP3 File The following TACL command file starts the processes, adds and starts subsystem objects through SCF, and sets appropriate parameters. To add comments, use the word “comment” or a double equal sign (==). Some lines are discussed separately after the example. Example 3-4. TCPIPUP3 for the TELSERV Process comment comment comment comment comment comment comment comment comment comment comment ==== TCPIPUP3 =========TCPIPUP3 ======== TACL command file to bring up Parallel Library TCP/IP subsystem Use DNS for name resolution; (no host file DEFINE) DELETE DEFINE =TCPIP^HOST^FILE ADD and START SUBNETS SCF/IN $SYSTEM.TCPIP.SCFSBNT/ Define round-robin filtering for the TELSERV process (See Round-Robin Filtering on page 2-4) DELETE DEFINE =PTCPIP^FILTER^KEY ADD DEFINE =PTCPIP^FILTER^KEY, class map, file A234567 Initialize TELSERV processes Define the TCPSAM process for TELSERV to use PARAM TCPIP^PROCESS^NAME $ZSAM0 Start the TELSERV processes TELSERV/TERM $ZHOME, OUT $ZHOME, & NAME $ZTN0, CPU 0, NOWAIT, PRI 170/1 TELSERV/TERM $ZHOME, OUT $ZHOME, & NAME $ZTN1, CPU 2, NOWAIT, PRI 170/3 TELSERV/TERM $ZHOME, OUT $ZHOME, & NAME $ZTN2, CPU 4, NOWAIT, PRI 170/5 TELSERV/TERM $ZHOME, OUT $ZHOME, & NAME $ZTN3, CPU 6, NOWAIT, PRI 170/7 ====== END OF TCPIPUP3 ==== END OF TCPIPUP3 == Deleting the DEFINE before setting the new DEFINE for round-robin filtering: DELETE DEFINE =PTCPIP^FILTER^KEY ensures the new define does not conflict with any existing defines in this TACL session for the filter key. The configuration for round-robin filtering: ADD DEFINE =PTCPIP^FILTER^KEY, class map, file A234567 sets up all subsequent processes configured in this TACL session to use round-robin filtering. A234567 is an arbitrary file name that you select; it is equivalent to setting a password for use of the port. The lines starting the TELSERV processes, beginning with TELSERV/TERM $ZHOME, OUT $ZHOME, & NAME $ZTN0, CPU 0, NOWAIT, PRI 170/1 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3- 17 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Configuration Example for the Distributor Listening Model start TELSERV primary and backup processes in every other processor. Note that the backup TELSERV processes do not share any processors with other TELSERV processes. Running TELSERV processes in distinct processor pairs avoids potential port sharing conflicts in a failure situation. (See Port Collision Considerations for Listening Processes on page 2-5.) Configuration Example for the Distributor Listening Model This example demonstrates the distributor listener model discussed above (see Distributor Listening Model on page 3-6) using a hypothetical distributor called “Distrib” configured with round-robin enabled. This configuration has five processors running three Distrib servers and controlling the socket I/O for the server classes they control. All the connections share the same IP address on the same G4SA. The G4SA round-robin distributes incoming connections to the different Distrib instances. The startup files and subnet initiation files for establishing the Parallel Library TCP/IP environment and subnets are the same as Configuration Example for the Standard Listening Model. In this example, TCPIPUP4, the main command file, does not include the complete configuration commands for Distrib Server. (You must substitute real RUN commands for your distributor listener applications.) Figure 3-9 shows the round-robin feature in this configuration. We have configured a Distrib in processors 3 through 5, bound them to the same port, and enabled round-robin filtering on those processes. (See Round-Robin Filtering on page 2-4.) HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3- 18 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Configuration Example for the Distributor Listening Model Figure 3-9. Configuration Example for Distributor Listening Model: Distrib Processor 2 Processor 2 Server SQL/MP Processor 1 Server SQL/MPProcessor 0 Class A Pathway Server Processor 2 Server SQL/MP Processor 1 Class A SQL/MP Processor 0Server Class A Pathway Server SQL/MP Processor 1 Server Class A Server SQL/MP Processor 0 Class A Pathway Server Pathsend Distrib Server 1 Distrib Server 3 Distrib Server 2 TCP/IP Library Processor 5 TCP/IP Library Processor 4 Processor 3 Sockets TCP/IP Library Round-robin distribution of incoming connection requests G4SA VST0309.vsd The first connection request establishes a connection to one of the distributor’s sockets. The adapter then routes the next connection request to the next distributor in the next processor, and so on. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3- 19 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Configuration Example for the Distributor Listening Model The TCPIPUP4 File The following TACL command file starts the processes, adds and starts subsystem objects through SCF, and sets appropriate parameters. To add comments, use the word “comment” or a double equal sign (==). Some lines are discussed separately after the example. Example 3-5. TCPIPUP4 for the Distrib Process comment comment comment comment comment comment comment comment comment comment ==== TCPIPUP4 =========TCPIPUP4 ======== TACL command file to bring up Parallel Library TCP/IP subsystem Use DNS for name resolution; (no host file DEFINE) DELETE DEFINE =TCPIP^HOST^FILE ADD and START SUBNETS SCF/IN $SYSTEM.TCPIP.SCFSBNT/ Define round-robin filtering for the Distrib listener.(See Round-Robin Filtering on page 2-4.) DELETE DEFINE =PTCPIP^FILTER^KEY ADD DEFINE =PTCPIP^FILTER^KEY, class map, file A234567 Define the TCPSAM process for the distributor to use DELETE DEFINE =TCPIP^PROCESS^NAME ADD DEFINE =TCPIP^PROCESS^NAME, class map, file $ZSAM0 Start a Distrib in each processor RUN DISTRIB /NAME $DIST1, NOWAIT, PRI 160, CPU 3/0 RUN DISTRIB /NAME $DIST2, NOWAIT, PRI 160, CPU 4/1 RUN DISTRIB /NAME $DIST3, NOWAIT, PRI 160, CPU 5/2 ====== END OF TCPIPUP4 ==== END OF TCPIPUP4 == Deleting the DEFINE before setting the new DEFINE for round-robin filtering: DELETE DEFINE =PTCPIP^FILTER^KEY ensures the new define won’t conflict with any existing defines in this TACL session for the filter key. The line configuring round-robin filtering: ADD DEFINE =PTCPIP^FILTER^KEY, class map, file A234567 sets up all subsequent processes configured in this TACL session to use round-robin filtering. A234567 is an arbitrary file name that you select. Setting the filter key file is equivalent to setting a password for use of the port. The lines starting the Distrib processes, starting with: RUN DISTRIB /NAME $DIST1, NOWAIT, PRI 160, CPU 3/0 start a Distrib process in each of the processors. Note that the backup Distrib processes do not share any processors with other Distrib processes. Running Distrib processes in distinct processor pairs avoids potential port sharing conflicts in a failure situation. (See Port Collision Considerations for Listening Processes on page 2-5.) HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3- 20 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Configuration Example for the Hybrid Listening Model Configuration Example for the Hybrid Listening Model This example demonstrates the hybrid listener model discussed above (see Hybrid Listener Model on page 3-9) using iTP WebServer configured with round-robin filtering enabled. This configuration has four processors each running their own distributors. Each distributor listens for connections and hands over the connections to the web servers that it controls. In this configuration, each distributor has servers only in its own processor. All the connections share the same IP address on the same G4SA, which in turn uses round-robin distribution for incoming connections to the different processors. The startup files and subnet initiation files for establishing the Parallel Library TCP/IP environment and subnets are the same as Configuration Example for the Standard Listening Model. Figure 3-8 shows the round-robin feature in this configuration. An iTP WebServer process is configured in each processor, bound to the same port, and is round-robin filtering enabled. (See Round-Robin Filtering on page 2-4.) Figure 3-10. Configuration Example for Hybrid Listening Model: iTP WebServer Processor 0 Server Web Server Servers Distributor TCP/IP Library Processor 1 Server Server Web Servers Processor 2 Serve r Server Web Servers Distributor TCP/IP Library Distributor TCP/IP Library Processor 3 Server Server Web Servers Distributor TCP/IP Library Round-Robin Distribution G4SA Single IP Host VST0310.vsd The first connection request establishes a connection to one of the Distrib processes. The adapter then routes the next connection request to the next Distrib process in the next processor, and so on. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3- 21 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Configuration Example for the Hybrid Listening Model The TCPIPUP5 File The following TACL command file starts the processes, adds and starts subsystem objects through SCF, and sets appropriate parameters. To add comments, use the word “comment” or a double equal sign (==). Some lines are discussed separately after the example. Example 3-6. TCPIPUP5 for Hybrid Listening Model comment comment comment comment ==== TCPIPUP5 =========TCPIPUP5 ======== TACL command file to bring up Parallel Library TCP/IP subsystem Use DNS for name resolution; (no host file DEFINE) DELETE DEFINE =TCPIP^HOST^FILE comment ADD and START SUBNETS SCF/IN $SYSTEM.TCPIP.SCFSBNT/ comment Define round-robin filtering for the iTP WebServer comment (See Round-Robin Filtering on page 2-4) DELETE DEFINE =PTCPIP^FILTER^KEY ADD DEFINE =PTCPIP^FILTER^KEY, class map, file A234567 comment Start iTP WebServer. See the comment iTP Secure WebServer System Administrator’s Guide. Run comment the httpd.config file documented in that manual once comment for each processor, changing the name of HTPD1 comment to HTPD0, HTPD1, HTPD2, and HTPD3, to match the comment processor names. Change the TCP/IP Transport comment Provider name to $ZSAM0. ccomment ====== END OF TCPIPUP5 ==== END OF TCPIPUP5 == Deleting the DEFINE before setting the new DEFINE for round-robin filtering: DELETE DEFINE =PTCPIP^FILTER^KEY ensures that the new define won’t conflict with any existing defines in this TACL session for the filter key. Use the following command. ADD DEFINE =PTCPIP^FILTER^KEY, class map, file A234567 sets up all subsequent processes configured in this TACL session to use round-robin filtering. A234567 is an arbitrary file name that you select. It is equivalent to setting a password for use of the port. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3- 22 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Example for Two Gateways — Standard Listening Model Example for Two Gateways — Standard Listening Model This example demonstrates the standard listener model discussed above (see Standard Listening Model on page 3-2). This example shows an environment that has two gateways leading to two subnets on a NonStop S-series server (host). The startup files for establishing two gateways and subnets are the same as those for Configuration Example for the Standard Listening Model on page 3-11. The files used in this example for starting and configuring the Parallel Library TCP/IP environment include: • • • The TCPIPUP1 File on page 3-13 which starts the Parallel Library TCP/IP environment. The TCPIPUP6 File on page 3-25, the main command file (a TACL command file), which calls the other files, sets up the HOSTS file, calls SCFSBNT, and starts the LISTNER. The SCFSBNT2 File on page 3-26, which adds, configures, and starts the subnets and routes. This configuration example is shown in Figure 3-10. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3- 23 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Example for Two Gateways — Standard Listening Model Figure 3-11. Two Gateways With LISTNER Routers GTWY1 Host 150.50.130.1 150.50.130.2 LAN Adapter 150.50.130.4 GTWY2 LISTNER Processor 1 150.60.64.1 TCPLIB 128.30.128.2 Processor 0 TCPLIB 128.30.128.1 150.60.64.2 LAN Adapter 150.60.64.3 FTPSERV Backup LISTNER TCPLIB Processor 2 150.60.64.4 FTPSERV 150.60.64.5 TCPLIB Processor 3 FTPSERV VST0311.vsd HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3- 24 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Example for Two Gateways — Standard Listening Model The TCPIPUP6 File The following TACL command file starts the processes, adds and starts subsystem objects through SCF, and sets appropriate parameters. To add comments, use the word “comment” or a double equal sign (==). Some lines are discussed separately after the example. Example 3-7. TCPIPUP6 for LISTNER Environment and Two Gateways comment comment comment comment comment comment comment comment comment comment ==== TCPIPUP6 =========TCPIPUP6 ======== TACL command file to bring up Parallel Library TCP/IP subsystem Use HOSTS file for name resolution; not DNS DELETE DEFINE TCPIP^HOST^FILE ADD DEFINE =TCPIP^HOST^FILE, FILE & $SYSTEM.ZTCPIP.HOSTS ADD and START SUBNETS SCF/IN $SYSTEM.TCPIP.SCFSBNT2/ Initialize LISTNERs for FTPSERV, ECHOSERV, and FINGSERV Define the TCPSAM process for the LISTNER to use DELETE DEFINE =TCPIP^PROCESS^NAME ADD DEFINE =TCPIP^PROCESS^NAME, class map, file $ZSAM0 Start the LISTNER LISTNER/NAME $LSN0, NOWAIT, PRI 170, CPU 0/1, & OUT $ZHOME, TERM $ZHOME, & $SYSTEM.ZTCPIP.PORTCONF ====== END OF TCPIPUP6 ==== END OF TCPIPUP6 == The line: ADD DEFINE =TCPIP^HOST^FILE, FILE $SYSTEM.ZTCPIP.HOSTS sets the =TCPIP^HOST^FILE define to point to the desired HOSTS file. Having this define set informs the DNS to use the HOSTS file to translate host names to IP addresses. For information about the RESOLVER, see the TCP/IP and TCP/IPv6 Programming Manual. The RUN command: LISTNER/NAME $LSN0, NOWAIT, PRI 170, CPU 0/1, & OUT $ZHOME, TERM $ZHOME, & $SYSTEM.ZTCPIP.PORTCONF starts the LISTNER process responsible for starting the ECHO, FINGER, and FTP servers when a client request is received by the LISTNER process. You should run this process at a high priority. This command also specifies the location of the PORTCONF file used to designate which ports this process is to listen to. This process requires privileged access to some Parallel Library TCP/IP ports: therefore, always log on with a super group ID. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3- 25 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Example for Two Gateways — Standard Listening Model The SCFSBNT2 File The SCFSBNT2 file adds and starts subnets and routes. Example 3-8. SCFSBNT2 File for TCPIPUP6 === SCFSBNT2 ===== SCFSBNT2 ==== SCFSBNT2 ======== == SCF command file to ADD and START SUBNETs Example 3-7 ALLOW ALL ERRORS ALLOW ALL WARNINGS == ASSUME PROCESS $ZZTCP == Add subnets ADD SUBNET SN0,TYPE ETHERNET,DEVICENAME LAN01,IPADDRESS 150.50.130.2, & SUBNETMASK %HFFFFFF00 STOP SUBNET LOOP0 ALTER SUBNET LOOP0, IPADDRESS 127.1 ADD SUBNET SN1,TYPE ETHERNET,DEVICENAME LAN02,IPADDRESS 150.60.64.2, & SUBNETMASK %HFFFFFF00 ADD SUBNET SN2,TYPE ETHERNET,DEVICENAME LAN03,IPADDRESS 150.60.64.3, & SUBNETMASK %HFFFFFF00 ADD SUBNET SN3,TYPE ETHERNET,DEVICENAME LAN04,IPADDRESS 150.50.130.4, & SUBNETMASK %HFFFFFF00 == Add route ROU0 to direct traffic destined for network 128 to GTWY1 ADD ROUTE ROU0, DESTINATION 128.30.0.0, GATEWAY 150.50.130.1 == == Add route ROU2 to direct all other traffic to GTWY2 ADD ROUTE ROU2, DESTINATION 0.0.0.0, GATEWAY 150.60.64.1 == Start subnets & routes START SUBNET * START ROUTE * == =========== END OF SCFSBNT2 ============= END OF SCFSBNT2 =========== For the ADD SUBNET command, the subnet name can be anything under seven alphanumeric characters long beginning with an alpha character. The DEVICENAME attribute, which specifies the logical interface (LIF) name associated with the adapter that the Parallel Library TCP/IP process accesses, is required. (See step e on page 1-4 for determining an appropriate LIF.) In SCFSBNT2, four subnets are started (SN0, SN1, SN2, and SN3). Because LISTNER binds to the socket with INADDR_ANY, it listens for incoming connections on all configured subnets. So in this configuration, incoming connection requests are accepted from both subnets. Note. If the LISTNER had bound to the IP address specified for the subnet (in this case 150.60.64.3), instead of INADDR_ANY, the LISTNER would have been limited to accepting incoming connections only on SN1. While LISTNER would not do this, other applications following this model might. Hence, you should configure the listener with INADDR_ANY to accept connections on all subnets. (See Subnet-Level Binding: How to Isolate Subnets in a Single-IP Environment on page 2-4 for a related discussion of INADDR_ANY.) HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3- 26 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Example for Two Gateways — Standard Listening Model LOOPBACK When the monitors are started, a subnet named LOOP0 is added automatically. This subnet provides loopback capability without requiring the use of the TCP/IP network. When this LOOP0 subnet is created, it has an address of 0.0.0.0 in dotted decimal form. You must change this address; use the command: ALTER SUBNET LOOP0, IPADDRESS 127.1 The address 127.1 (or 127.0.0.1) is the standard for loopback operation. Routes A route is added to direct traffic destined for the 128 subnet through Gateway 1. Another route is added to direct all other traffic through Gateway 2. The ROUTE objects must be added to all monitors. (See ADD ROUTE Command for TCPMAN on page 5-18.) Subnet Mask The subnet mask causes the first three octets to be used for determining the correct network. Gateway In this example, there are potentially two destination networks to which HOST could communicate and thus, two routes, one for each destination network. Notice that in this case, the gateway address is the same for each route. As shown in Figure 3-11, HOST must route a datagram destined to the other network through GTWY1, which has the IP address of 150.50.130.1. This method is especially useful when you have multiple gateways to multiple networks. When all the routing is through a single gateway, however, there is a simpler way to set up your routing. Default routing establishes a single route as the default route. This action is particularly useful when you know that most of your TCP/IP traffic is going through a single gateway. The second route added in Example 3-8 implements default routing. The use of 0.0.0.0 to designate the destination network IP address is what indicates that this is a default route. You can add more routes for those networks which cannot be reached by using the default route. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3- 27 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Example for Two Gateways — Standard Listening Model The HOSTS File The HOSTS file is the file used in the absence of a Domain Name Server for resolving the common names of hosts into their corresponding IP addresses. (The HOSTS file shown is customized for this example.) Example 3-9. HOSTS File for TCPIPUP6 ########## HOSTS FOR HOST ########## HOSTS FOR HOST ############ # Filename = \CB1.$SYSTEM.ZTCPIP.HOSTS # Date = January 31/93 150.50.130.1 GTWY1 gtwy1 gw1 127.0.0.1 me loop 150.50.130.2 LAN01 lan01 con1 150.60.64.2 LAN02 lan02 con2 150.60.64.1 GTWY2 gtwy2 gw2 150.60.64.3 LAN03 lan03 con 3 150.50.130.4 LAN04 lan04 corp4 ###########END OF HOSTS ##################END OF HOSTS ########### All text following a pound sign (#) is comment text. Use comment text to note revisions made to the file. Begin the IP addresses of the hosts in column one of the HOSTS file. Separate the host name from the address by at least one space. You may have as many aliases as can fit on a single entry line. The lines in the HOSTS file: 127.0.0.1 me loop 150.50.130.2 LAN01 lan01 doc1 150.60.64.2 LAN02 lan02 doc2 provide flexibility in testing the environment. When you use the ECHO service to send an echo datagram to me or loop, you are testing the client and server capabilities of your own ECHO service. If you send an ECHO datagram to lan01 or lan02, you also are testing the actual physical-network connection for your HOST1 Parallel Library TCP/IP environment. The example assumes that all the hosts and gateways on this intranet are NonStop hosts. This HOSTS file easily accommodates the IP addresses and names of any host connected to the TCP/IP network. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3- 28 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Parallel Library TCP/IP for Complex, Heavy-Use WAN Environments Parallel Library TCP/IP for Complex, HeavyUse WAN Environments Parallel Library TCP/IP provides improved scalability for the SWAN subsystem. Whereas with conventional TCP/IP, traffic for a given path had to go through the processor which contained the conventional TCP/IP process, with Parallel Library TCP/IP, you can configure the WAN subsystem so that not only the workload of a WAN I/O process but also the workload of TCP/IP done on its behalf is spread across all the processors in which you have configured the WAN I/O processes. In conventional TCP/IP, there was a one-to-one correspondence between the TCP/IP process and a SLSA LIF/PIF. Traffic destined for that TCP/IP process would all flow through a single processor (potentially creating a bottle-neck). With Parallel Library TCP/IP, there is a one-to-any correspondence between the TCP/IP stack and all the LIFs/PIFs. So now you can place your WAN I/O processes in any processor and you can associate them with any LIFs/PIFs. Traffic now goes directly through the processors in which the WAN I/O processes have been configured, eliminating the interprocessor hop formerly required to get to the TCP/IP process servicing the LIF/PIF. In addition to increased scalability, Parallel Library TCP/IP also offers a feature called Ethernet failover. This provides another layer of fault-tolerance at the Ethernet adapter level, from which the SWAN subsystem and its client-I/O processes can benefit. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3- 29 Configuring Parallel Library TCP/IP for Complex and Heavy-Use Environments Parallel Library TCP/IP for Complex, Heavy-Use WAN Environments HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 3- 30 4 Managing the Parallel Library TCP/IP Subsystem This section describes the following aspects of managing the Parallel Library TCP/IP subsystem: • • • • • • Running Applications in Both Environments on page 4-1 Managing the System Configuration Database on page 4-1 Managing Performance on page 4-7 Strategy for Coexistence with Conventional TCP/IP on page 4-7 Falling Back to Conventional TCP/IP on page 4-7 Dynamically Loading SPRs on page 4-8 There are some new system management tasks for Parallel Library TCP/IP as well as some considerations for running applications in the Parallel Library TCP/IP and conventional TCP/IP environments. Running Applications in Both Environments You will probably run applications in both the conventional TCP/IP and the Parallel Library TCP/IP environments. If you are using ATM or token-ring adapters your applications must use conventional TCP/IP. If you are using Ethernet adapters, you can run your application in either environment. The two environments cannot share the same LIF but they can share an E4SA or G4SA because those adapters have four LIFs. However, a FESA and a GESA have only one LIF, so they can only support one environment. You may choose to dedicate some combination of these adapters to either the conventional TCP/IP or Parallel Library TCP/IP environments, but you don’t have to do anything differently to configure the FESA, E4SA, GESA, and G4SA adapters for use with Parallel Library TCP/IP. Managing the System Configuration Database The system configuration database (CONFIG) is part of the NonStop Kernel subsystem on NonStop S-series servers. The conventional TCP/IP subsystem (NonStop TCP/IP) does not participate in the system configuration database but Parallel Library TCP/IP does. As soon as you configure Parallel Library TCP/IP for the first time, the MON, ROUTE, ENTRY, and SUBNET objects are added to the system configuration database and any alterations to those objects also update the configuration of those objects in the system configuration database. The system configuration database HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 4 -1 Managing the Parallel Library TCP/IP Subsystem Configuration Database Management stores your subsystem configuration and can be accessed at any time to restore the subsystem to its last configuration. However, the TCPSAM process is not stored in the system configuration database so you must always start TCPSAM processes by using the TACL RUN command. Note. No dynamically created entries or routes are recorded in the system configuration database. The TCPMAN, when started, either by the persistence manager or by a TACL RUN command, starts any subordinate objects that are stored in the system configuration database. Configuration Database Management Save your configuration database prior to configuring Parallel Library TCP/IP for the first time and record the name and date of the saved database. This saved configuration database can be used if future RVUs of Parallel Library TCP/IP are incompatible with the Parallel Library TCP/IP records residing in the system configuration database. If a new RVU of Parallel Library TCP/IP is incompatible with the data stored in the configuration database, you can restore the saved configuration database and reconfigure Parallel Library TCP/IP. The following SCF command saves the current configuration database file in a new file located at $SYSTEM.ZYSCONF.CONF0104: ->SAVE CONFIGURATI0N 01.04 The full explanation of the SCF SAVE command is documented in SCF Reference Manual for G-Series RVUs. Caution. The configuration database stores all SCF commands that you issue to modify your Parallel Library TCP/IP environment. If you use startup scripts to start your Parallel Library TCP/IP subsystem, you should compare your configuration database to those startup files to ensure that the startup files reflect these additional modifications to the environment. For more detailed procedures and specific migration considerations, see the TCP/IP (Parallel Library) Migration Guide. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 4 -2 Managing the Parallel Library TCP/IP Subsystem Managing Persistence Managing Persistence You can add a generic process to the system configuration database and define that generic process in such a way that the persistence manager ($ZPM) will restart the generic process whenever the generic process abends, is stopped through TACL, or the system is reloaded. To define the generic process, set the STARTMODE to SYSTEM. If you add the TCPMAN process as a generic process configured in this way, TCPMAN starts automatically upon system reload and subsequently restores its stored, subordinate objects. Alternatively, when you add the TCPMAN process as a generic process to the system configuration database, you can choose to configure it using STARTMODE MANUAL; this method requires that you start $ZZKRN.#ZZTCP manually by using an SCF START command to the NonStop Kernel subsystem. Note that the persistence manager restarts persistent generic processes whenever they are stopped (if the generic process is configured with STARTMODE SYSTEM) in addition to starting those generic processes when the system is reloaded. Hence, if TCPMAN is a generic, persistent process (AUTORESTART > 0), any time you try to stop TCPMAN, it gets restarted by the persistence manager. To avoid this behavior and stop a persistent, generic, TCPMAN process, issue the ABORT command to the NonStop Kernel subsystem as in the following example: ->ABORT PROCESS $ZZKRN.#ZZTCP See also How to Stop the Generic Process for TCPMAN on page 4-5. For more information about generic processes and the persistence manager, see the SCF Reference Manual for the Kernel Subsystem. Managing the TCPSAM Process Remember that the TCPSAM process cannot be added as a generic process because processes that require PARAMs or DEFINEs cannot be configured as generic processes or added to the system configuration database. Therefore, even if you add $ZZTCP as a generic process, your Parallel Library TCP/IP environment is not completely persistent because you still must create any required TCPSAM processes as well as any applications that depend on TCPSAM. (See How to Manage TCPSAMDependent Applications on page 4-4.) The following procedures show how to create TCPSAM processes. TACL Commands for Starting a TCPSAM Process To create a TCPSAM process, perform the following steps: 1. Determine the location of the system image file (OSIMAGE) by entering the following TACL command: >FILEINFO $SYSTEM.SYS*.OSIMAGE Select one of the system subvolumes returned by this FILEINFO command for the next step. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 4 -3 Managing the Parallel Library TCP/IP Subsystem How to Manage TCPSAM-Dependent Applications 2. Change your session location to the subvolume of $SYSTEM.SYSnn by entering the following command (see also Locating the SRL on page 2-12): >VOLUME $SYSTEM.SYSnn 3. Enter the following command to tell the TCPSAM process the location of the Parallel Library TCP/IP private SRL: >ADD DEFINE =_SRL_01, CLASS MAP, FILE ZTCPSRL 4. Start a TCPSAM process by entering the following TACL command (substitute a name of your choice for the process name -- shown here as ZTC2): >TCPSAM /NAME $ZSAM3, TERM $ZHOME, OUT $ZHOME, NOWAIT, CPU 0/1 Command File for Starting a TCPSAM Process Because you have to start TCPSAM processes yourself without the aid of the persistence manager, HP recommends that you create a command file for this purpose. However, before issuing the OBEY command on the file, ensure that your session is still in the subvolume of the SRL file by using the FILEINFO command shown in TACL Commands for Starting a TCPSAM Process on page 4-3. 1. Create the SAMUP file. (Substitute real values for the variables indicated in italics.) Example 4-1. SAMUP ADD DEFINE =_SRL_01, CLASS MAP, FILE ZTCPSRL TCPSAM /NAME $ZSAM3, TERM $ZHOME, OUT $ZHOME, NOWAIT, CPU 0/1 2. Issue the following command: >OBEY SAMUP How to Manage TCPSAM-Dependent Applications Even though you can have Parallel Library TCP/IP started automatically by the persistence manager, the TCPSAM processes cannot be managed by the persistence manager. Since applications that depend on Parallel Library TCP/IP use TCPSAM, you should not configure those applications as generic processes to be started automatically by the persistence manager. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 4 -4 Managing the Parallel Library TCP/IP Subsystem How to Add TCPMAN as a Generic Process to the System Configuration Database How to Add TCPMAN as a Generic Process to the System Configuration Database Note. You can start the TCPMAN process from a RUN command rather than through the persistence manager and TCPMAN will restore the configuration of the subordinate objects from the system configuration database. Remember, that TCPMAN does not start any TCPSAM processes; you must start TCPSAM processes manually whether you are using the persistence manager or the RUN TCPMAN command. Although you should not need to add $ZZTCP as a generic process more than once, HP recommends that you create a command file for the procedure as a safeguard to allow you to resume the exact configuration of $ZZTCP as a generic process. To add a generic process to the system configuration database for $ZZTCP, create a command file containing the following commands like the one shown in Example 4-2 (substitute your own values for those parameters indicated in italics). See the SCF Reference Manual for the Kernel Subsystem for complete information about adding a generic process. Example 4-2. Command File for Adding TCPMAN as a Generic Process SCF/INLINE/ INLPREFIX + +ADD PROCESS $ZZKRN.#ZZTCP, AUTORESTART 10, BACKUPCPU 1, & DEFAULTVOL $SYSTEM.SYSTEM, HOMETERM $ZHOME, & NAME $ZZTCP, OUTFILE $ZHOME, PRIMARYCPU 0, PRIORITY 180, & PROGRAM $SYSTEM.SYSTEM.TCPMAN, STARTMODE SYSTEM,& STARTUPMSG “<BCKP-CPU>”, STOPMODE SYSMSG +START PROCESS $ZZKRN.#ZZTCP INLEOF Because you have set STARTMODE to SYSTEM, whenever the system is loaded or whenever the TCPMAN is stopped, the persistence manager restarts the generic TCPMAN process (#ZZTCP) and all the most recently configured SUBNET, ROUTE, ENTRY, and MON objects. See How to Stop the Generic Process for TCPMAN if you don’t want the persistence manager to restart #ZZTCP every time you stop the $ZZTCP using a TACL STOP command. How to Stop the Generic Process for TCPMAN If the TCPMAN process has been added as a generic process, you must use the SCF ABORT command to the NonStop Kernel subsystem to stop it (ABORT PROCESS $ZZKRN.#ZZTCP). If you issue an SCF STOP or ABORT command under the Parallel Library TCP/IP subsystem, you will receive an error. If you have also set AUTORESTART greater than zero, the process is persistent and the persistence manager will restart the process if the process stops due to a processor failure or abend. To stop $ZZTCP if it is a persistent, generic process, issue the following SCF command: ->ABORT $ZZKRN.#ZZTCP HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 4 -5 Managing the Parallel Library TCP/IP Subsystem How to Add TCPMAN as a Generic Process to the System Configuration Database The ABORT PROCESS command to $ZZKRN.#ZZTCP stops $ZZTCP and causes the persistence manager not to restart it until a system reload. To restart TCPMAN, issue the following SCF command: ->START PROCESS $ZZKRN.#ZZTCP If you don’t want the persistence manager to restart $ZZTCP automatically during a system reload, you can make one of the three following changes to the generic process (shown in Example 4-2, Command File for Adding TCPMAN as a Generic Process, on page 4-5): • • • Change STARTMODE to MANUAL. In this case, you must always issue an SCF START $ZZKRN.#ZZTCP to restart TCPMAN. Delete the generic process ($ZZKRN.#ZZTCP). In this case, TCPMAN is no longer a generic process and the persistence manager no longer starts it automatically upon system reload. Change STARTMODE to DISABLED. In this case, the generic process ($ZZKRN.#ZZTCP) remains in the system configuration database but is not started by the persistence manager. In addition, you cannot manually start the generic process ($ZZKRN.#ZZTCP) until the STARTMODE is changed back to SYSTEM or MANUAL. If you ABORT the process $ZZTCP by using the SCF command to the Parallel Library TCP/IP subsystem (ABORT PROCESS $ZZTCP instead of ABORT PROCESS $ZZKRN.#ZZTCP) and specify the SUB ALL in the ABORT command, only the TCPMAN process is restarted when you issue a RUN command or when the persistence manager restarts the process. The SUB ALL specification in the ABORT command deletes the MON objects from the configuration database. If you have issued the ABORT PROCESS $ZZTCP, SUB ALL command and want to restart the subsystem, issue a START MON * command to the NonStop TCP/IP subsystem. This action starts all subordinate objects with the configuration attributes that are stored in the system-configuration database. For information about managing generic processes, see SCF Reference Manual for the Kernel Subsystem. The following example shows how to stop the generic process #ZZTCP: Caution. Before stopping the TCPMAN ($ZZKRN.#ZZTCP), stop all applications that are using the Parallel Library TCP/IP environment. (See Stopping Parallel Library TCP/IP as a Generic Process on page 1-29 for procedures for checking what applications are using Parallel Library TCP/IP.) -> ABORT PROCESS $ZZKRN.#ZZTCP To manually restart TCPMAN after aborting it under the NonStop Kernel subsystem, issue the following command: ->START PROCESS $ZZKRN.#ZZTCP HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 4 -6 Managing the Parallel Library TCP/IP Subsystem Managing Performance To manually restart all the TCPMAN subordinate objects, issue the following commands: ->START MON * ->DELAY 21 Remember to start your TCPSAM processes since they will not be started by TCPMON or TCPMAN. Managing Performance When using Parallel Library TCP/IP, processor utilization measurements of client applications may tend to exhibit higher numbers because the TCP/IP processing is now done in the context of the application. In the conventional TCP/IP environment, this work was attributed to the TCP/IP process. Therefore, MEASURE analysis might show an increase in processor utilization by the application process. From a systems perspective, the overall processor utilization should be less than in the conventional TCP/IP environment because the number of dispatches and context switches is minimized. Strategy for Coexistence with Conventional TCP/IP To provide the features that are unsupported by Parallel Library TCP/IP, the conventional TCP/IP environment is present with the restriction that you can’t share the same IP subnet address (LIF) between environments. Conventional TCP/IP allows a fallback position as well as providing the unsupported features. Falling Back to Conventional TCP/IP 1. Follow one of the shutdown procedures in Section 1, Configuration Quick Start. (See Stopping Parallel Library TCP/IP and Preserving the Current Configuration on page 1-19, Stopping Parallel Library TCP/IP and Clearing the Database on page 1-24, or Stopping Parallel Library TCP/IP as a Generic Process on page 1-29.) 2. Change your system configuration database back to the previous, non-Parallel Library TCP/IP configuration database. The Compaq TSM system-load online help provides information about how to select a specific configuration file at system load. 3. Switch over to the existing conventional TCP/IP environment. Reset the DEFINEs, PARAMs, and/or transport service provider name-set procedure calls for your applications back to the conventional TCP/IP process name. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 4 -7 Managing the Parallel Library TCP/IP Subsystem Dynamically Loading SPRs a. Determine the name of your preferred transport service provider name. Use either $ZTC0, (if that is the name of the default conventional TCP/IP process), or use the LISTDEV command to obtain a list of running TCP/IP processes: ->LISTDEV TCPIP b. Change the transport service provider name for Guardian and OSS applications to the conventional TCP/IP process by entering one of the following commands: ° ADD DEFINE =TCPIP^PROCESS^NAME, class map, file $tcpip-process-name ° PARAM TCPIP^PROCESS^NAME $tcpip-process-name c. Change the following procedure calls for Guardian and OSS socket applications: ° ° set_inet_name() (for Guardian applications) socket_transport_name_set() (for OSS applications) Dynamically Loading SPRs You can install new Parallel Library TCP/IP SPRs on a processor-by-processor basis without having to halt and reload each processor. The following steps describe how to install a new Parallel Library TCP/IP software product revision (SPR) without halting processors: 1. Stop any socket applications that are using Parallel Library TCP/IP. 2. Rename the current TCPMON, TCPMAN, TCPSAM, and ZTCPSRL files in the current SYSnn. 3. Install the replacement SPR objects in the current SYSnn. 4. Shut down the Parallel Library TCP/IP environment in the normal way. (See Stopping Parallel Library TCP/IP and Preserving the Current Configuration on page 1-19.) 5. Restart the socket applications that you stopped for this procedure. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 4 -8 5 SCF Reference for Parallel Library TCP/IP This section provides information about: • • • The Subsystem Control Facility (SCF) SCF commands available for PTCPIP (to find a command quickly, see Table 5-4 on page 5-9.) The PTrace facility SCF for Parallel Library TCP/IP SCF provides an operator interface to an intermediate process, the Subsystem Control Point (SCP), which in turn provides the interface to the I/O processes of the various subsystems. The Parallel Library TCP/IP subsystem runs on the NonStop operating system and supports subnets using Ethernet LANs. Ethernet subnets use the ServerNet LAN systems access (SLSA) subsystem to provide access to Ethernet local area networks (LANs). The Parallel Library TCP/IP subsystem is a client of the SLSA subsystem which includes the LAN manager (LANMAN) and LAN monitor (LANMON) processes. SCF Commands for TCPMAN Compared to SCF Commands for TCPSAM This section describes SCF command syntax for both the TCPMAN and the TCPSAM processes. TCPSAM SCF syntax differs from TCPMAN syntax because TCPSAM provides backward-compatibility for applications. Existing applications expect objects to have the format $process-name.#subordinate-object-name which is the format that TCPSAM uses. By contrast, TCPMAN objects use the format $processname.#TCPMON-name.subordinate-object-name. The TCPSAM and TCPMAN processes yield different command results. For example, STATUS PROCESS $ZZTCP yields information about the primary and backup processor and identification numbers, whereas, STATUS PROCESS $tcpsam-name yields more extensive information (comparable to the STATUS PROCESS command in conventional TCP/IP.) In addition, when TCPSAM nonsensitive SCF commands are applied to the PROCESS, ROUTE, and SUBNET objects, the information returned reflects only those objects in the processor where the TCPSAM process resides. To HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -1 SCF Reference for Parallel Library TCP/IP Object Types get information about these objects on all configured processors, use the SCF commands for the TCPMAN process instead. Object Types You can monitor and control the Parallel Library TCP/IP subsystem by issuing commands that act on one or more Parallel Library TCP/IP subsystem objects. Each object has an object type and an object name. The object type describes the type of object. The object name uniquely identifies the object within the system. The Parallel Library TCP/IP subsystem has two PROCESS object types: • • TCPMAN TCPSAM The TCPMAN and TCPSAM processes support different subordinate objects and have different SCF command syntax, attribute definitions, and displays. This section describes each command in alphabetical order. In this section, when the command applies to both the TCPMAN and TCPSAM processes, the TCPMAN command syntax is described first with the TCPSAM command syntax immediately following. There are six object types supported by TCPMAN: • • • • • • PROCESS MON SUBNET ROUTE ENTRY null There are four object types supported by TCPSAM: • • • • PROCESS SUBNET ROUTE null Figure 5-1 on page 5-3 shows the object hierarchy for TCPMAN and that the route, subnet, and entry object types are peers. The route, subnet, and entry object types are subordinate to the process and monitor (TCPMON) objects. This hierarchy is important when issuing commands to the Parallel Library TCP/IP subsystem for processing. For example, because the route, subnet, and entry object types are subordinate to the process and monitor (TCPMON) object types, any commands pertaining to a route, subnet, or entry object type can be issued only when the process and monitor (TCPMON) objects are in the STARTED summary state. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -2 ENTRY Object Type SCF Reference for Parallel Library TCP/IP Figure 5-1. TCPMAN Process Object Hierarchy $PROCESS #MONITOR SUBNET ROUTE ENTRY VST00501.vsd Figure 5-2 shows the object hierarchy for the TCPSAM process. Figure 5-2. TCPSAM Process Object Hierarchy $PROCESS #SUBNET #ROUTE VST0502.vsd Note that there is a pound sign (#) in front of the SUBNET and ROUTE objects for TCPSAM and there is no intermediate monitor (TCPMON) object. (See SCF Commands for TCPMAN Compared to SCF Commands for TCPSAM on page 5-1.) ENTRY Object Type The ENTRY object allows you to view and add to the Address Resolution Protocol (ARP) table which maps physical (MAC) addresses to IP addresses. The ENTRY object name can have at most seven alphanumeric characters. The first two characters must be EA. If the ENTRY object was added dynamically, the name is provided by Parallel Library TCP/IP. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -3 SCF Reference for Parallel Library TCP/IP MONITOR Object Type MONITOR Object Type The MONITOR object (TCPMON) provides the Parallel Library TCP/IP environment in a processor. Only one TCPMON can exist in each configured processor. TCPMON has the reserved name of $ZPTMn where n is the processor number (hexadecimal) where TCPMON resides. Note. In the Parallel Library TCP/IP subsystem, a TCPMON can have more than one IP address associated with it (one per subnet). However, each TCPMON must have a valid TCPMON name, and each IP address must be unique within the network. null Object Type The null object is not an actual object type. The term “null” represents the lack of a specified object. Any SCF command that supports the null object type is issued without the specification of an object type. Commands support the null object type if an object type is irrelevant (as for the VERSION command), or if they refer to a collection of objects (as for with the NAMES command). To issue an SCF command using the null object, specify the name of the SCF command followed by a process name. The process name must be a valid NonStop operating system process name. Do not use the term “null” when you issue the command. PROCESS Object Type Two possible PROCESS objects can exist in the Parallel Library TCP/IP subsystem: TCPSAM and TCPMAN. TCPSAM is the socket access method. TCPMAN is the manager process. Both TCPMAN and TCPSAM run as process pairs. TCPMAN provides management functions for the PTCPIP subsystem and communicates with the TCPMONs in each processor. Only one TCPMAN process pair can exist in a system; however any number of TCPSAM process pairs can run in a system. Both the TCPMAN and TCPSAM processes can be started by using RUN commands. In addition, the TCPMAN process can be started by using the persistence manager. Only one name is supported for the TCPMAN process: $ZZTCP. You can assign any name to the TCPSAM process. When you assign a name to a TCPSAM process, HP recommends that you should use a name that conforms to the conventions for process names. The recommended form for TCPSAM process names is $ZTCx or $ZTCxx, where x is a letter or a numeric digit; for example, $ZTC01. Most HP client and server programs expect the name of a PTCPIP process to take this form. This convention allows applications to use a simple screening algorithm to locate Parallel Library TCP/IP processes in a system. However, since the conventional TCP/IP and Parallel Library TCP/IP environments coexist on the system, an application programmer wishing to select one or the other of the two environments needs to either obtain the PTCPIP process name from the system administrator or use the SCF LISTDEV TCPIP command to determine the HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -4 ROUTE Object Type SCF Reference for Parallel Library TCP/IP names of the processes associated with each environment. The LISTDEV command displays all the TCP/IP processes running on the system. The last field in the display is Program. A program name of TCPSAM indicates a Parallel Library TCP/IP process while a program name of TCPIP indicates a conventional TCP/IP process. To obtain a list of all running PTCPIP processes, enter the SCF LISTDEV PTCPIP command. (This command also gives you a list of the running TCPMON objects.) Again, the process type (TCPMON or TCPMAN) is identified in the program field. ROUTE Object Type The ROUTE object is the path a data packet travels to reach its destination. Instead of specifying a full path, a route specifies the packet’s first host address and the packet’s destination. The first host then routes the packet to the next appropriate address in-route to the destination. This sequence repeats until the packet reaches the destination. Often, a NonStop S-series server routes all packets to a default host, which in turn maintains a more complete routing table. Each time you add a subnet, a route is created automatically. You can add more routes as necessary. Refer to TCP/IP Configuration and Management Manual, for a full explanation of routes and routing. You must assign a unique ROUTE object name to each route associated with a given process. The ROUTE object name can have at most seven alphanumeric characters. The first character must be a letter. Table 5-1 shows an example of ROUTE object naming conventions. Names starting with DD, DA, RT, DR and EA are reserved. Table 5-1. Route Object Naming Conventions PROCESS $ZZTCP ROUTE Object Name Route 1 ROU1 Route 2 ROU2 Route3 ROU3 To omit the process name and period and just specify the route name, set the default process name with the ASSUME command. For further information on the ASSUME command, including the required syntax, refer to the SCF Reference Manual for GSeries RVUs. SUBNET Object Type The SUBNET is the point of connection between the Parallel Library TCP/IP and an I/O device. All subnets are associated with the TCPMAN process. Each subnet name must be unique within the system. The name can have at most seven alphanumeric characters. The first character must be a letter. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -5 SCF Reference for Parallel Library TCP/IP SUBNET Object Type The SUBNET object is also subordinate to the TCPMON object. A subnet is accessible to all TCPMONs in the system. You can view a subnet from a system-wide view or from a TCPMON/processor view. The fully-qualified subnet name includes the process name, a period, the TCPMON name, a period, and the subnet name. A STATUS on the $ZZTCP.#TCPMONname.subnet-name shows you the subnet status in one processor (the one where the specified TCPMON is running). A STATUS on the $ZZTCP.*.subnet-name shows you the subnet status in all processors. To omit the process name, use the ASSUME command to set the default process. In subsequent commands, just specify the TCPMON and subnet names as in the following example: SCF> ASSUME PROCESS $ZZTCP SCF> INFO SUBNET *.SN1 To omit the TCPMON name, use the ASSUME command to set the default TCPMON name. If you have also assumed the process name, you can specify only the subnet name in subsequent commands. The following examples show the ASSUME command for both the process and TCPMON, and for the process only. SCF> ASSUME PROCESS $ZZTCP SCF> ASSUME MON $ZZTCP.#ZPTM0 SCF> INFO SUBNET SN1 SCF> ASSUME PROCESS $ZZTCP SCF> INFO SUBNET *.SN1 HP recommends that you use the letters SN followed by a subnet number to identify a subnet; for example, SN1. The SUBNET object is the point of connection between PTCPIP and the SLSA LIF. Data coming from or going to an Ethernet LAN goes through a subnet. A subnet type of Ethernet specifies both Fast Ethernet, Gigabit Ethernet, and Ethernet type devices. The name LOOP0 is reserved for the loopback subnet. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -6 Naming Convention Summary SCF Reference for Parallel Library TCP/IP Naming Convention Summary Table 5-2 summarizes the reserved names for each object type and the naming convention rules. Table 5-2. Object Naming Convention Summary and Reserved Names Starting Symbol (Required) First Character Requirement First Character Recommendation Character Limit Letter EA 7 Object Type Reserve d Names ENTRY None MON #ZPTMx # Letter MON names are assigned automatically. 5 null N/A N/A N/A N/A N/A PROCESS (TCPMAN) $ZZTCP $ N/A The name is always $ZZTCP. 7 PROCESS (TCPSAM) None $ Letter ZSAMx where x is a letter or numeric digit. 7 ROUTE Names starting with: Letter None 7 Letter SN followed by a subnet number 7 RT, DD, DA, DR, EA SUBNET LOOP0 Wild-Card Support Normally, an SCF command line must include an object specifier composed of the object type and an object name. For many commands, the Parallel Library TCP/IP subsystem accepts object-name templates. In an object-name template, one object name can be used to indicate that multiple objects of a given object type are to be affected by the command. Object-name templates allow you to specify multiple objects by entering either a single wild-card character, or text and one or more wild-card characters. In the Parallel Library TCP/IP subsystem, you can use the following wild-card characters: * Use an asterisk (*) to represent a character string of undefined length. The following example deletes all subnets subordinate to $ZZTCP: SCF> DELETE SUBNET $ZZTCP.*.* The following example deletes all subnets subordinate to $ZZTCP that have names that start with SN: HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -7 Summary States SCF Reference for Parallel Library TCP/IP SCF> DELETE SUBNET $ZZTCP.*.SN* The following example deletes all routes subordinate to $ZZTCP that start with R and end with 5: SCF> DELETE ROUTE $ZZTCP.*.R*5 ? Use the question mark to represent a single unknown character in a specific position. For example, $ZZTCP.*.S?1 selects all object names subordinate to $ZZTCP that begin with S, end with 1, and contain exactly one character between the S and the 1. You can use wild-card characters in any combination. If you have set a default process name by using the ASSUME command, you can omit the process name and use the asterisk (*) to specify all objects of the specified object type under the assumed process. For example, the next two commands set the default process to $ZZTCP and display information about all subnets under $ZZTCP: SCF> ASSUME PROCESS $ZZTCP SCF> INFO SUBNET *.* Summary States The Parallel Library TCP/IP subsystem objects have operational states, known as summary states. The summary state of an object at a given instant is important; certain commands have no effect on an object when it is in one state but can affect the object when it is in another state. The summary states supported by the Parallel Library TCP/IP subsystem are STARTED, STARTING, and STOPPED. Table 5-3 shows the states for each object. Table 5-3. Object Summary States Object STOPPED ENTRY STARTED STARTING X null PROCESS X MON X X ROUTE X X SUBNET X X • • X X In the STARTED summary state, the object is available for data transfer. In the STOPPED summary state, the object is defined (that is, the object exists) but it is not available for data transfer. The STOPPED summary state is not applicable to the PROCESS object. If the PROCESS object is not STARTED, it is undefined (that is, the process does not exist). HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -8 Parallel Library TCP/IP SCF Commands SCF Reference for Parallel Library TCP/IP • In the STARTING summary state, the object has been initialized and is attempting to start. Parallel Library TCP/IP SCF Commands This subsection contains the following information: • • • A table describing the Subsystem Control Facility (SCF) commands supported by the Parallel Library TCP/IP subsystem and the object types supported for each command. Detailed information on object specification syntax. The following detailed information about each SCF command: ° ° ° A description of the command function. ° ° ° Descriptions, by object type, of the attributes. The command syntax. The object specification, which shows the supported object types and object names. Considerations you should be aware of before using the command. Command examples. Supported Commands and Object Types This section describes the SCF commands that are interpreted specifically for the Parallel Library TCP/IP subsystem. The SCF Reference Manual for G-Series RVUs provides general information about SCF commands. You should be familiar with that information before reading the Parallel Library TCP/IP subsystem-specific information provided here. Table 5-4 lists the SCF commands and object types supported by the TCPMAN process. The page number of the command description follows the command name. Table 5-4. Commands and Object Types for TCPMAN (page 1 of 2) Object Types SCF Command ENTRY ABORT Command, 5-12 ADD Command, 5-17 MON PROCESS ROUTE SUBNET X X X X X X X ALTER Command, 5-25 X DELETE Command, 5-33 X INFO Command, 5-36 X X X X X X X X HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -9 Supported Commands and Object Types SCF Reference for Parallel Library TCP/IP Table 5-4. Commands and Object Types for TCPMAN (page 2 of 2) Object Types SCF Command ENTRY LISTOPENS Command, 5-60 NAMES Command, 5-66 MON PROCESS ROUTE SUBNET X X X X PRIMARY Command, 5-70 X START Command, 5-72 X X X STATS Command, 5-75 X X X STATUS Command, 5-123 X X X X X STOP Command, 5-139 X X X X TRACE Command, 5-143 X X VERSION Command, 5-152 X X X Table 5-5 lists the SCF commands and object types supported by the TCPSAM process. The page number of the command description follows the command name. Table 5-5. Commands and Object Types for TCPSAM Object Types SCF Command PROCESS ABORT Command, 5-12 X INFO Command, 5-36 X LISTOPENS Command, 5-60 X NAMES Command, 5-66 ROUTE SUBNET X X X X PRIMARY Command, 5-70 X STATS Command, 5-75 X X X STATUS Command, 5-123 X X X STOP Command, 5-139 X TRACE Command, 5-143 X VERSION Command, 5-152 X HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 10 Entering SCF Commands SCF Reference for Parallel Library TCP/IP Table 5-6 lists the sensitive and nonsensitive Parallel Library TCP/IP SCF commands. Table 5-6. Sensitive and Nonsensitive SCF Commands Sensitive Commands Nonsensitive Commands ABORT Command INFO Command ADD Command LISTOPENS Command ALTER Command NAMES Command DELETE Command STATS Command (without the RESET option) PRIMARY Command STATUS Command START Command VERSION Command STATS Command (with the RESET option) STOP Command TRACE Command Entering SCF Commands You start SCF interactively by issuing the following TACL command: 1>SCF You rarely need to specify SCF RUN parameters because the default values are appropriate for most situations. For a more detailed description of the TACL RUN command parameters that apply to SCF, refer to the SCF Reference Manual for GSeries RVUs. At the beginning of an SCF session, SCF displays its product banner, which includes the HP product name, product number, version number, RVU date, and copyright statement. SCF waits for a command, followed by a carriage return. After the command has been received and processed, SCF displays its prompt for the next command. An SCF command always begins with a keyword identifying the command (such as ADD, ABORT, or ALTER). The keyword is followed by the object specifier, consisting of the object type and the object name, as in the following example: SCF> ABORT SUBNET $ZZTCP.#ZPTM1.SN1 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 11 SCF Reference for Parallel Library TCP/IP ABORT Command If additional attribute specifiers are required to define characteristics of the object, the object name is followed by a comma and the attribute name and value, as in the following example: SCF> ALTER MON $ZZTCP.#ZPTM*, DELAYACKS OFF Note. The SEL and SUM options, which apply to several of the SCF commands when used with other communications subsystems, cannot be used with the Parallel Library TCP/IP subsystem. You can enter multiple SCF commands at a single prompt by separating the commands with semicolons, as in the following example: SCF> ASSUME MON $ZZTCP.*;ALTER MON #ZPTM1, HOSTNAME "slugo1" When processing a command line that contains more than one command, SCF executes the commands one at a time from left to right. If an error occurs, SCF displays the appropriate error message and ignores the rest of the line. You can also continue a command that starts on one line onto a second line by terminating the first line with an ampersand (&). SCF prompts for additional input before executing the command, as in the following example: SCF> ADD SUBNET $ZZTCP.#ZPTM*.SN1, TYPE ETHERNET, & SCF> DEVICENAME LAN01, IPADDRESS 120.0.0.1 You must not enter more than 2048 characters for any input command. Note. SCF accepts input from either a terminal or a disk (OBEY) file and directs output to either a terminal, disk file, or printer. However, in this manual, all examples assume that a terminal is being used for both input and output. The rest of this section describes each SCF command for the Parallel Library TCP/IP subsystem. ABORT Command The ABORT command terminates the operation of specified Parallel Library TCP/IP subsystem processes, subnets, or routes as quickly as possible. Only enough processing is done to ensure the integrity of the subsystem. The objects are left in the STOPPED summary state. If any outstanding socket requests remain from the application, use the ABORT command instead of the STOP command. All pending socket requests are completed with an error. This is a sensitive command. ABORT MON Command for TCPMAN The ABORT MON command terminates the operation of the MON object as quickly as possible, without regard to open sockets. ABORT can be used to stop the PTCPIP HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 12 SCF Reference for Parallel Library TCP/IP ABORT PROCESS Command for TCPMAN TCPMON objects when open sockets exist. This command also deletes the MON object from the system configuration database. Command Syntax ABORT [ / OUT file-spec / ] MON [$ZZTCP.#ZPTMn] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. MON $ZZTCP.#ZPTM{0-F} is a valid MON name indicating the desired TCPMON. The MON object is always named $ZZTCP.#ZPTMn where n is the processor number where the TCPMON resides. You may substitute the wild card (*) for #ZPTMn and abort all running TCPMONs. You may also ASSUME the process name, in which case, you only need to enter the TCPMON name starting with the pound (#) sign or the wild card (*). Examples The first command aborts the MON object named #ZPTMA and the second group of commands aborts all running TCPMONs: SCF> ABORT MON $ZZTCP.#ZPTMA SCF> ASSUME PROCESS $ZZTCP SCF> ABORT MON * Considerations The ABORT MON command deletes the MON from the system configuration database. Because the MON has been deleted from the configuration database after the ABORT MON command, if you ABORT one MON and then restart it, it no longer shares the same non-default attributes with the other MONs in the system. For example, the hostname and hostids on the restarted MON would be blank and, if defined on the other MONs, would result in different hostnames and hostids being passed to different instances of a program, depending on which processor the program instance resides in. If you want to stop the MON but leave it in the system configuration database, use the STOP MON Command for TCPMAN on page 5-139. ABORT PROCESS Command for TCPMAN The ABORT PROCESS command terminates the operation of the TCPMAN process immediately, without regard to open sockets. This command stops the process and deletes it from the Parallel Library TCP/IP environment. However, the ABORT process HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 13 SCF Reference for Parallel Library TCP/IP ABORT PROCESS Command for TCPSAM command does not stop or delete the process from the system configuration database if the process has been added as a generic process. See Considerations. Command Syntax ABORT [ / OUT file-spec / ] [ PROCESS $ZZTCP ] [ , SUB ALL ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. PROCESS $ZZTCP is the name of the TCPMAN process. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. SUB ALL aborts all subordinate MON objects. If you use this option, the MON object is deleted from the configuration database. Note that when you use the RUN command to restart the $ZZTCP process or if the persistence manager restarts the $ZZTCP process, you must issue the START MON * command to restart the monitors and subordinate objects. Examples The following command aborts and deletes the TCPMAN process (named $ZZTCP) and all subordinate MON objects: SCF> ABORT PROCESS $ZZTCP, SUB ALL Considerations If the TCPMAN process has been added as a generic process, you must use the ABORT command under the Kernel subsystem (ABORT PROCESS $ZZKRN.#ZZTCP) to stop it. See How to Stop the Generic Process for TCPMAN on page 4-5 for more information about managing generic processes. The SUB ALL option deletes the TCPMON objects from the system configuration database. ABORT PROCESS Command for TCPSAM The ABORT PROCESS command terminates the operation of the TCPSAM process immediately, without regard to open sockets. This command stops and deletes the process. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 14 SCF Reference for Parallel Library TCP/IP ABORT ROUTE Command for TCPMAN Command Syntax ABORT [ / OUT file-spec / ] [ PROCESS $tcpsam-process-name ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. PROCESS $tcpsam-process-name is a valid process name indicating the desired TCPSAM process. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. Examples The following command aborts and deletes the TCPSAM process named $ZSAM2: SCF> ABORT PROCESS $ZSAM2 Considerations If there are any outstanding socket requests from the application, the ABORT command must be used instead of the STOP command. All pending socket requests with this TCPSAM as the transport provider are completed with an error. ABORT ROUTE Command for TCPMAN The ABORT ROUTE command terminates the activity of the specified route. Only enough processing is done to ensure the integrity of the subsystem. The object is left in the STOPPED summary state. The ABORT ROUTE command does not delete the ROUTE from the system configuration database. (See DELETE SUBNET Command for TCPMAN on page 5-35.) Command Syntax ABORT [ / OUT file-spec / ] [ROUTE $ZZTCP.*.route-name ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. ROUTE $ZZTCP.*.route-name is the name of the route. The fully-qualified route name is $ZZTCP.*.route-name. When you abort a route, you must do so on all configured TCPMONs. You can use the wild-card (*) notation for the TCPMON name, but if you do not, it is assumed. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 15 SCF Reference for Parallel Library TCP/IP ABORT SUBNET Command for TCPMAN For example, ABORT ROUTE *.RT1 is equivalent to ABORT ROUTE RT1. If you omit the process name, SCF uses the assumed process name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. Examples The following commands abort the specified routes in all TCPMONs: SCF> ABORT ROUTE $ZZTCP.*.RT1 SCF> ASSUME PROCESS $ZZTCP SCF> ABORT ROUTE *.RT1 ABORT SUBNET Command for TCPMAN The ABORT SUBNET command terminates the operation of a subnet as quickly as possible; only enough processing is done to ensure the integrity of the subsystem. The object is left in the STOPPED summary state. Since subnets are accessible to every processor with a configured TCPMON, the ABORT SUBNET command must be applied to all processors. Command Syntax ABORT [ / OUT file-spec / ] [SUBNET $ZZTCP.*.subnet-name] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. SUBNET $ZZTCP.*.subnet-name names the point of connection between the Parallel Library TCP/IP process and an I/O device. The fully-qualified subnet name is $ZZTCP.*.subnet-name (you must abort subnets on all TCPMONs). If you omit the process name, SCF uses the assumed process. If you omit the subnet name, SCF uses the assumed SUBNET object. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. Examples The following commands abort a subnet named $ZZTCP.*.SN1: SCF> ABORT SUBNET *.SN1 The following command aborts all subnets for the TCPMAN process. Note that you can omit the wild card (*) for the TCPMON. SCF> ASSUME PROCESS $ZZTCP SCF> ABORT SUBNET * HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 16 SCF Reference for Parallel Library TCP/IP ADD Command The following commands abort a subnet named SN0: SCF> ASSUME PROCESS $ZZTCP SCF> ABORT SUBNET *.SN0 Considerations • • • • The object-name template (wild-card notation) is supported. All activities being performed by the specified objects are halted. Use the STOP command if you want to stop the operation of objects in a more controlled manner. The STOP command does not abruptly terminate activities in progress. When a subnet is aborted, all the routes dependent on this subnet switch to use another interface/subnet if there are multiple interface cards in the same subnet range. ADD Command The ADD command adds a subnet, route or entry to the Parallel Library TCP/IP subsystem. You must enter an ADD SUBNET command for each subnet with which the Parallel Library TCP/IP subsystem is to communicate. Each subnet defines a point of attachment through which data is sent or received. This is a sensitive command. ADD ENTRY Command for TCPMAN The ADD ENTRY command creates an entry in an ARP table. Entries in the ARP table provide a static mapping between an IP address and a MAC address. The ENTRY object name must start with the letters “EA.” Command Syntax ADD [ /OUT file-spec/ ] [ ENTRY $ZZTCP.*.entry-name ] , TYPE ARP , IPADDRESS ip-addr , MACADDR mac-address OUT file-spec causes any SCF output generated for this command to be directed to the specified file. ENTRY $ZZTCP.*.entry-name specifies the name of the ENTRY object. The fully-qualified entry name is $ZZTCP.*.entry-name (you must add entries on all configured TCPMONs.) If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 17 SCF Reference for Parallel Library TCP/IP ADD ROUTE Command for TCPMAN The naming convention for entries is seven characters. The first two characters must be EA. TYPE ARP specifies the ENTRY type. The only supported type is ARP. A TYPE of ARP maps an IP address with an Ethernet MAC address. The ARP type requires an IP address and the MACADDR attribute. IPADDRESS ip addr specifies the internet address for the entry and is specified in dotted decimal notation. MACADDR mac address specifies the Ethernet address for the ENTRY. It is entered as a string of twelve hexadecimal digits preceded by a “%h”. Examples The following command creates an ENTRY in the ARP entry table: -> ADD ENTRY $ZZTCP.*.EA1, TYPE ARP, IPADDRESS & 1.2.3.4, MACADDR %H08008E003578 ADD ROUTE Command for TCPMAN The ADD ROUTE command creates a route. Command Syntax ADD [ / OUT file-spec / ] [ ROUTE $ZZTCP.*.route-name ] [ [ [ [ [ [ , , , , , , , , DESTINATION GATEWAY DESTTYPE NETMASK METRIC CLONING GENMASK SUBNET destination-ip-address gateway-ip-address { HOST | BROADCAST } ] mask-val ] metric-val ] { ON | OFF } ] mask-val ] subnet-name ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. ROUTE $ZZTCP.*.route-name is the name of the route. The route specifies the path on which data is sent in order to reach a destination. When you add a route, you must do so on all configured HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 18 SCF Reference for Parallel Library TCP/IP ADD ROUTE Command for TCPMAN TCPMONs. You can use the wild-card (*) notation for the TCPMON name, but if you do not, it is assumed anyway. For example, ADD ROUTE *.RT1 is equivalent to ADD ROUTE RT1. Names starting with DA, DD, DR, EA, and RT are reserved. The naming convention for routes is seven characters. The first character must be a letter. DESTINATION destination-ip-address specifies the Internet address of a single host or an entire network which can be reached through the system specified in GATEWAY. A zero in the local address portion of the destination Internet address acts as a wild card, representing all hosts on the network specified in the network portion of the Internet address. If the destination Internet address is 0 (0.0.0.0), this route specifies the default gateway. In this case, all packets with addresses for which routes cannot be determined are sent to the host specified in GATEWAY. This parameter is required. Default: If the route is added automatically when a subnet is added, the default address is the IP address of the subnet's host (the value for the subnet's IPADDRESS attribute) converted to broadcast form. GATEWAY gateway-ip-address specifies the Internet address of the gateway host through which the network or host addressed in DESTINATION can be reached. This is a required attribute. Default: The subsystem supplies no default value for Gateway unless the route is being added automatically as a part of the ADD SUBNET operation. The default value is the IP address of the subnet's host (the value for the subnet's IPADDRESS attribute). DESTTYPE { HOST | BROADCAST } specifies whether the route is a connection to a specific host (HOST) or to a network (BROADCAST). This is an optional attribute. Default: If you do not specify DESTTYPE in an ADD ROUTE command, the default value is BROADCAST. If a route is added automatically as the result of an ADD SUBNET command, the default value is BROADCAST. NETMASK mask-val specifies a subnet mask value to be associated with the route entry. This maskval is specified as dotted-decimal or hexadecimal notation. If it is not specified, it defaults to the default network mask of the specified IP address. The netmask value is set to be 0.0.0.0 for the default routes. The netmask value is set to 255.255.255.255 for the host route. Since non-contiguous mask values are not supported, netmask values such as 0.255.255.0, 255.0.255.0, 0.0.255.0 are considered invalid. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 19 SCF Reference for Parallel Library TCP/IP ADD ROUTE Command for TCPMAN METRIC metric-val indicates the number of hops to the destination. The metric is optional for add commands; it defaults to zero if the destination is on a directly-attached network and the metric is not specified. It is non-zero if the route uses one or more gateways. The default is 1 if the destination is not on a directly-attached network and the metric is not specified. CLONING { ON | OFF } enables (ON) or disables (OFF) the cloning capability of a route. A network route with cloning capability clones/generates an additional route with a subnet mask specified by GENMASK when it is referenced. This additional/cloned route is flagged with “c” in the “INFO ROUTE” display to indicate it’s a cloned route. GENMASK mask-val specifies a subnet mask value to be associated with the cloned route. This mask-val is specified as dotted-decimal or hexadecimal notation. If it is not specified, the cloning capability is turned off. Since non-contiguous mask values are not supported, genmask values such as 0.255.255.0, 255.0.255.0, 0.0.255.0 are considered invalid. The GENMASK value differs from the NETMASK value. In the example below, the route MR3 is created by the system administrator with a destination 155.186.0.0 and a network mask of 0xffff0000. Since the cloning capability is turned on (CLONING ON) and the genmask is set to 0xffffff00 which differs from the network mask. If the routing table is searched for 155.186.72.123 and the entry does not exist for 155.186.72.0 subnet, the entry for 155.186.0.0 with the mask of 0xffff0000 as the best match. A new route entry (since CLONING is ON) with a destination of 155.186.72.0 and a network mask of 0xffffff00 (the genmask value) would be created. The next time any host on this subnet is referenced, say 155.186.72.88, it matches this newly created entry. SUBNET subnet-name allows a route to be associated with a particular LIF. This feature allows up to n routes to be configured for n LIFs in the same subnet range; all these routes would go to the same destination but through a different LIF. Examples The following commands show how to add routes: -> ADD ROUTE $ZZTCP.*.MR1, DESTINATION 192.17.72.0, & NETMASK 255.255.255.0, GATEWAY 172.17.214.1, METRIC 1 -> ADD ROUTE $ZZTCP.*.MR2, DESTINATION 192.17.73.0,& NETMASK %HFFFFFF00, GATEWAY 172.17.214.2, METRIC 2 -> ADD ROUTE $ZZTCP.*.MR3, DESTINATION 155.186.0.0, & NETMASK 255.255.0.0, CLONING ON, GENMASK 255.255.255.0,& GATEWAY 172.17.214.1, METRIC 1 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 20 SCF Reference for Parallel Library TCP/IP ADD SUBNET Command for TCPMAN Considerations Routes can also be created dynamically through internal routing logic. • • • • • • • Routes created by internal route-redirect logic start with DDcpu where cpu is the CPU number in hexadecimal format where the route is generated. Routes created by internal ARP link-level logic start with DAcpu where cpu is the processor number in hexadecimal format where the route is generated. Routes generated implicitly because of an ADD SUBNET or ALTER SUBNET, subnetmask command, start with RT. The name specified in the ADD ENTRY command should start with the name EA. This ADD ENTRY name generates a link-level route with name name. All of the above route names are reserved; that is, names starting with DDcpu, DAcpu, RT, DRcpu and EA are reserved. When a subnet is added, a corresponding route to this subnet is added automatically. Both the subnet and the route are placed in the STOPPED state. To initiate the operation of the object, you must start it with the START command. The optional parameters CLONING and GENMASK are not allowed in the ADD ROUTE command when the SUBNET parameter is present. ADD SUBNET Command for TCPMAN The ADD SUBNET command creates a subnet in the Parallel Library TCP/IP environment. Command Syntax ADD [ /OUT file-spec/ ] [ SUBNET $ZZTCP.*.subnet-name ] , TYPE ETHERNET , DEVICENAME lif-name , IPADDRESS ip-addr [ , IRDP { ON | OFF } ] [ , SUBNETMASK mask-val ] [ , FAILOVER {SHAREDIP | NONSHAREDIP} ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. SUBNET $ZZTCP.*.subnet-name is the name of the subnet. If you omit this attribute, SCF uses the assumed SUBNET name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. The fully-qualified subnet name is $ZZTCP.*.subnet-name (you must add subnets to all configured TCPMONs.) The HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 21 SCF Reference for Parallel Library TCP/IP ADD SUBNET Command for TCPMAN naming convention for subnets is seven characters. The first character must be a letter. HP recommends making the first two characters SN. TYPE ETHERNET specifies the type of subnet to be added. The only valid type is Ethernet. This parameter is required. Default: None. DEVICENAME lif-name is the name of the device to be opened to connect to the network. This corresponds to the SLSA logical interface (LIF). The LIF provides access to the Ethernet LAN. For information on how to choose a SLSA device name, see step e on page 1-4. When adding a subnet, the DEVICENAME for the SLSA subnet does not begin with a dollar ($) character. Default: None. IPADDRESS ip-addr is the Internet address associated with this subnet interface. This parameter is required. Default: None. IRDP {ON | OFF} enables (ON) or disables (OFF) the ICMP Router Discovery Protocol on the subnet interface. IRDP is a mechanism for locating default routers and is specified in RFC 1256. IRDP also must be enabled on any local LAN routers. If redundant routers are configured with route hold-down times and advertisement intervals of approximately 30 seconds, IRDP can be used to provide a black hole, or dead gateway, detection mechanism. The Parallel Library TCP/IP subsystem implements IRDP using IP broadcasts rather than IP multicasts. Default: OFF. SUBNETMASK mask-val specifies that part of the IP address that has to be masked in order to make the host part of the IP address as a subnet. This is done normally to generate further subnets from CLASS A, CLASS B and CLASS C networks. This mask-val is specified as dotted-decimal or hexadecimal notation. If it is not specified, it defaults to the corresponding network mask of the specified IP address. A non-contiguous mask value is not supported. Since a non-contiguous mask value is not supported, netmask values such as 0.255.255.0, 255.0.255.0, 0.0.255.0 are considered invalid. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 22 SCF Reference for Parallel Library TCP/IP ADD SUBNET Command for TCPMAN FAILOVER {SHAREDIP | NONSHAREDIP} enables the SUBNET to be failover-capable using the following configuration: • • SHAREDIP has the same IP address as the associated SUBNET in the failover configuration. NONSHAREDIP has a different IP address than the associated SUBNET in the failover configuration. Examples The first example adds a subnet named SN1 of type Ethernet to all the TCPMONs in the system. The second example adds a subnet named SN4 to all the TCPMONs in the system. Note that in the second example, a subnet mask value has been specified, which was not possible in conventional TCP/IP. (In conventional TCP/IP, you had to alter the added subnet to add the subnet mask value.) -> ADD SUBNET $ZZTCP.*.SN1, TYPE ETHERNET, DEVICENAME LAN02, & IPADDRESS 50.0.0.3 -> ADD SUBNET $ZZTCP.*.SN4, TYPE ETHERNET, DEVICENAME LAN04, & IPADDRESS 50.0.0.1, SUBNETMASK %HFFFF0000 The first of the following examples adds a SUBNET and associated SUBNET with failover enabled for non-shared IP addresses and the second example adds them for shared IP addresses: -> ASSUME PROCESS $ZZTCP -> ADD SUBNET SN1, TYPE ETHERNET, DEVICENAME LANLIF2, IPADDRESS 172.17.217.232, SUBNETMASK 255.255.255.0, FAILOVER NONSHAREDIP -> ADD SUBNET SN2, TYPE ETHERNET, DEVICENAME LANLIF3, IPADDRESS 172.17.217.234, SUBNETMASK 255.255.255.0, FAILOVER NONSHAREDIP -> ADD SUBNET SN3, TYPE ETHERNET, DEVICENAME LANLIF4, IPADDRESS 172.17.217.44, SUBNETMASK 255.255.255.0, FAILOVER SHAREDIP -> ADD SUBNET SN4, TYPE ETHERNET, DEVICENAME LANLIF5, IPADDRESS 172.17.217.44, SUBNETMASK 255.255.255.0, FAILOVER SHAREDIP Considerations • • • • You can add up to 64 subnets in the Parallel Library TCP/IP environment. When you add a subnet, you must do so to all configured TCPMONs. Hence, only the wild card (*) is supported for the TCPMON name. The wild card, however, is optional; if you do not specify it, the wild card is assumed. Unlike conventional TCP/IP, you can now specify a subnet mask value in the ADD SUBNET command. When you specify the name of the route or subnet you are adding, be sure to specify the process name in the ASSUME command or in the ADD command, as shown in the examples. Verify that the name is unique for that process. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 23 SCF Reference for Parallel Library TCP/IP • • • • • ADD SUBNET Command for TCPMAN The SLSA subsystem must be operational for the ADD command to complete successfully. Refer to the LAN Configuration and Management Manual for more information. When a subnet is added, a corresponding route to this subnet is added automatically. Both the subnet and the route are placed in the STOPPED state. To initiate the operation of the object, you must start it with the START command. When adding a SLSA subnet type, the DEVICENAME does not begin with a dollar sign ($) character. Subnets from the conventional TCP/IP and Parallel Library TCP/IP environments cannot share a LIF. See Subnet-Level Binding: How to Isolate Subnets in a Single-IP Environment on page 2-4 for information about using the subnet IP address as opposed to INADDR_ANY when binding applications. The following are some guidelines to use when configuring Ethernet failover: • • • • • • The ALTER SUBNET command is required with the ADD SUBNET command to link the two LIFs for failover. When selecting the LIF pair for the failover SUBNET pair, you should select LIFs on different adapters. When using Fast Ethernet adapters and Gigabit Ethernet adapters connected directly to Ethernet switches, failover recovery time might be impacted by the spanning tree feature used in a switch. When using multiple failover pairs on the same network subnet and adding static routes, its best to add a copy of each route to one SUBNET in each failover pair. This increases the availability of the routes should both SUBNETs comprising a failover pair become unavailable. It also allows Parallel TCPIP Library TCP/IP to distribute outbound connections over the failover pairs when the source IP address is not selected by the application. Ethernet failover might not function when directly connected to a firewall that uses Ethernet-address (MAC)-to-IP address filtering. This problem can be overcome by adding a router between the LIFs and the firewall. When configuring multiple, shared IP fail over pairs, the reserved IP address cannot be shared between pairs. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 24 SCF Reference for Parallel Library TCP/IP ALTER Command ALTER Command The ALTER command changes attribute values associated with the specified PTCPIP object. This is a sensitive command. ALTER MON Command for TCPMAN The ALTER MON command is used to change the attribute values of the Parallel Library TCP/IP subsystem. When you alter attributes of a TCPMON process you must do so on all configured PTCPIP TCPMONs. Hence, only the wild card is supported for the TCPMON object. Command Syntax ALTER [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ /OUT file-spec/ ] MON $ZZTCP.* ,TCPSENDSPACE int ] ,TCPRECVSPACE int ] ,UDPSENDSPACE int ] ,UDPRECVSPACE int ] ,DELAYACKS { ON | OFF } ] ,DELAYACKSTIME int ] ,HOSTNAME string ] ,HOSTID int ] ,TCPKEEPIDLE int ] ,TCPKEEPINTVL int ] ,TCPKEEPCNT int ] ,DEBUG { ON | OFF } ] ,FULLDUMP { ON | OFF } ] ,ALLNETSARELOCAL { ON | OFF } ] ,TCPCOMPAT42 { ON | OFF } ] ,EXPANDSECURITY { ON | OFF } ] ,TCPPATHMTU { ON | OFF } ] ,TCPTIMEWAIT int ] ,RFC1323-ENABLE { ON | OFF } ] ,TCP-INIT-REXMIT-TIMEOUT int ] ,TCP-MIN-REXMIT-TIMEOUT int ] ,TCP-LISTEN-QUE-MIN int ] ,INITIAL-TTL int ] ,MIN-EPHEMERAL-PORT int ] ,MAX-EPHEMERAL-PORT int ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. MON $ZZTCP.* specifies all configured TCPMONs. When you alter the MON, you must do so on all configured MONs. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 25 SCF Reference for Parallel Library TCP/IP ALTER MON Command for TCPMAN TCPSENDSPACE int specifies the size of the window used for sending data for the TCP protocol. The recommended range for is 512 bytes to 12k bytes. The default value is 8K. TCPRECVSPACE int specifies the size of the window used for receiving data for the TCP protocol. This value affects the performance, as it is mapped to the advertised window. The recommended range is 512 bytes to 12k bytes. The default value is 8K. It is recommended that this value not be set too low (below 2K). UDPSENDSPACE int specifies the size of the window used for sending data for the UDP protocol. The recommended range is 512 bytes to 12k bytes. The default value is 9216. UDPRECVSPACE int specifies the size of the window that is used for receiving data for the UDP protocol. The recommended range is 512 bytes to 12k bytes. The default value is 41600. It is recommended that these values not be set too low (below 2K). DELAYACKS { ON | OFF } specifies whether acknowledgments for TCP packets be sent immediately (as soon as a packet is received). This mechanism allows more that one packet to be acknowledged with a single ACK. This helps reduce the network traffic. It also allows the TCP window to be filled up before an ACK is generated. The default value for DELAYACKS is ON. DELAYACKSTIME int specifies how much the delay time is before an ACK (acknowledgment) is sent for a packet. This is useful only if the DELAYACKS parameter is ON. It is specified in intervals of 0.01 seconds. The default value is 20 (200 milliseconds). The range is 1 through 50. It is recommended that the value of this variable not be greater than 20 (200 milliseconds). HOSTNAME string is the official name by which the host upon which the TCPMON is running is known to the Internet. This is a character string no longer than 50 characters. The default values is null. HOSTID int is the identification number (usually the host number part of the Internet address that is assigned to this host). It is a 32-bit number. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 26 SCF Reference for Parallel Library TCP/IP ALTER MON Command for TCPMAN TCPKEEPIDLE int is the amount of time in seconds before TCP issues a keep-alive packet on sockets that have enabled this option. The default is 45 seconds. The range is 1 to 7200. TCPKEEPINTVL int is the time interval in seconds between retransmissions of unacknowledged keepalive packets. The default is 45 seconds. The range is 1 to 1260. TCPKEEPCNT int is the number of times a keep-alive packet is sent without receiving an acknowledgment. After reaching int, the TCP connection is dropped. The default is 8. The range is 1 to 20. DEBUG { ON | OFF } is used by support personnel and development to enable the display of more TCP internal information for debugging purposes. FULLDUMP { ON | OFF } specifies whether the QIO segment is also saved when the TCPMON abends. The default is ON. When set to OFF, the TCPMON only saves its stack when abending which conserves disk space over a full dump. The preferred setting for this parameter is ON. ALLNETSARELOCAL { ON | OFF } causes TCP (when ON) to use the interface MTU as a base for determining the TCP Maximum Segment Size (MSS) for each non-local TCP connection. A nonlocal TCP connection is one that goes to another network (not just another subnetwork). The default is ON. If this switch is OFF, TCP conforms to RFC1323specified behavior and uses 512 bytes as the default MSS for non-local segments. When ON, for example for Ethernet, the non-local MSS is 1460. Setting this parameter to ON can benefit performance. TCPCOMPAT42 { ON | OFF } is the flag used to set the TCPMON compatible with BSD4.2 versions in the following regards: • • The default value of this flag is ON. If the flag is ON then the original ACK - 1 is sent in the keepalive packet; otherwise the original ACK is sent in the keepalive packet. EXPANDSECURITY { ON | OFF } is ON to cause TCP to check if a SOCKET request from another Expand node has passed the Expand security check. This means the user is valid on this system and HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 27 SCF Reference for Parallel Library TCP/IP ALTER MON Command for TCPMAN has correct remote passwords. If the check fails, the SOCKET request is rejected with file error 48. The default for this option is OFF. TCPPATHMTU { ON | OFF } is ON to cause TCP to use PATH MTU discovery on all TCP-type sockets (SOCK_STREAM), unless disabled by the SETSOCKOPT for SO_PMTU. The default is ON. TCPTIMEWAIT int is the amount of time in seconds that a TCP connection remains in the TIME_WAIT state. The default is 60 seconds. The range is 1 to 120. RFC1323-ENABLE { ON | OFF } is ON to cause TCP to support TCP Large Windows as documented in RFC 1323. When this option is enabled, Parallel Library TCP/IP uses the TCP Window Scale and Timestamp options as described in RFC 1323. The largest TCP window supported is 262144 bytes when this option is enabled, and 65535 when the option is disabled. The default is ON. TCP-INIT-REXMIT-TIMEOUT int is the initial retransmit timer-value in milliseconds to use on a TCP connection. When the first round-trip timer measurement is made on a TCP connection and the retransmission-timeout calculation for use on the next sent packet is done, this value is used (unless the calculated value is larger). This variable can be used to help reduce the number of premature retransmission timeouts. The default is 1000 milliseconds, or 1 second. The range is 200 to 30000 milliseconds. TCP-MIN-REXMIT-TIMEOUT int is the minimum value allowed for the TCP retransmission timeout. If this value is too low the TCPMON might generate premature retransmissions. If this value is set too high, real retransmissions are delayed, increasing the time for error recovery. The default is 1000 milliseconds. The range is 50 to 30000 milliseconds. TCP-LISTEN-QUE-MIN int is the minimum queue length that sits on a TCP socket when the TCPMON handles a socket LISTEN or ACCEPT_NW1 function call. This value is used if the queue length specified in the socket request is lower this minimum value; otherwise the queue length in the socket request is used. The default value is 128. INITIAL-TTL int specifies the initial value for UDP and TCP TTL. The default is 64, but may be altered to 30. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 28 SCF Reference for Parallel Library TCP/IP ALTER MON Command for TCPMAN MIN-EPHEMERAL-PORT int is the starting port number to allocate for TCP and UDP ephemeral ports. Ephemeral ports are those assigned by Parallel Library TCP/IP when an application has not bound to a specific port. The default is 1024. The allowable range is 1024 to (MAX-EPHEMERAL-PORT - 16). See Considerations and Examples. Everything below min-ephemeral-port requires super-group privileges. If you alter min-ephemeral-port to be greater than 1024, be aware that all ports between 1024 and min-ephemeral-port can only be opened by privileged users, that is, supergroup users. MAX-EPHEMERAL-PORT int is the largest port number to allocate for TCP and UDP ephemeral ports. The default is 65024. The allowable range is (MIN-EPHEMERAL-PORT + 16) to 65535. Each TCPMON is allocated one sixteenth of the range between min-ephemeralport and max-ephemeral-port. For example, using the defaults, #ZPTM0 is allocated 1024-5023, #ZPTM1 is allocated 5024-9023 and so on. See Considerations and Examples. Examples The following command alters the DELAYACKS and DELAYACKSTIME attributes on all configured TCPMONs. -> ALTER MON $ZZTCP.*, DELAYACKS ON, DELAYACKSTIME 20 The following command alters the TCPSENDSPACE to 4096 on all configured TCPMONs. -> ALTER MON $ZZTCP.*, TCPSENDSPACE 4096 The following command changes the TCP and UDP port range to 32768 to 65535. -> ALTER MON $ZZTCP.*, MIN-EPHEMERAL-PORT 32768 Considerations The MIN-EPHEMERAL-PORT and MAX-EPHEMERAL-PORT attributes have been added to allow you to modify the ephemeral port-range used by the TCPMONs to be more usable in your environment. The ephemeral port range is between 1024 and 65024. Each processor is allocated one sixteenth of this range. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 29 SCF Reference for Parallel Library TCP/IP ALTER SUBNET Command for TCPMAN ALTER SUBNET Command for TCPMAN The ALTER SUBNET command is used to change the attribute values of a subnet. Command Syntax ALTER [ /OUT file-spec/ ] [SUBNET $ZZTCP.*.subnet-name ] { [ ,IPADDRESS ip-addr ] [ ,SUBNETMASK %H0..FFFFFFFF ] [ ,IRDP { ON | OFF } ] [ ,ADDALIAS ip-addr,SUBNETMASK %H0..FFFFFFFF ] [ ,DELETEALIAS ip-addr ] } | { [ ,ASSOCIATESUB "subnet-name" ] [, RESERVEDIP ip-addr] } OUT file-spec causes any SCF output generated for this command to be directed to the specified file. SUBNET $ZZTCP.*.subnet-name is the name of the subnet. The fully-qualified subnet name is $ZZTCP.*.subnetname (you must alter the subnet on all configured TCPMONs.) You can substitute the wild card (*) for the subnet-name; doing so alters all subnets on all TCPMONs. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for GSeries RVUs. IPADDRESS ip_address is the 32-bit integer that identifies the subnet. This is the IP address assigned to the subnet by the network administrator. SUBNETMASK subnet-mask is a 32-bit integer in hexadecimal format that specifies the subnet mask for this subnet. A subnet mask identifies which portion of the IP local address represents the subnet number and which part represents the host ID. If bits in the subnet mask are set to 1, the corresponding bits in the IP address are part of the network (and subnet) address. If bits in the subnet mask are set to 0, the corresponding bits in the IP address are part of the host ID. That is, that portion of the local address masked with 1s identifies the subnet, and the remainder of the local address uniquely identifies a host connected to the subnet. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 30 SCF Reference for Parallel Library TCP/IP ALTER SUBNET Command for TCPMAN IRDP enables (ON) or disables (OFF) the ICMP Router Discovery Protocol on the subnet interface. IRDP is a mechanism for locating default routers and is specified in RFC 1256. IRDP must also be enabled on any local LAN routers. If redundant routers are configured with route hold-down times and advertisement intervals of approximately 30 seconds, IRDP can be used to provide a black hole, or dead gateway, detection mechanism. The Parallel Library TCP/IP subsystem implements IRDP using IP broadcasts rather than IP multicasts. ADDALIAS ip-addr allows the addition of the alias IP address to the subnet specified in the ALTER SUBNET command. The IP alias feature allows a process to be known to the Internet by different IP addresses. DELETEALIAS ip-addr allows the deletion of alias IP addresses that have been added by the ADDALIAS attribute. ASSOCIATESUB “subnet-name" links two adapters together to be a failover-pair configuration. As the result of this command, an IP alias address is configured for each subnet internally. For a SHAREDIP configuration, the RESERVEDIP address is configured as an alias address for both subnets. For a NONSHAREDIP configuration, the subnet IP address of the first subnet is configured as an alias address of the second subnet and vice-versa. RESERVEDIP ip-addr is a required parameter for the two subnets configured to be failover-enabled and also share the same subnet IP address. This parameter is not valid for failover configurations that have two different subnet IP addresses. Examples The following examples alter the subnets in all TCPMONs. -> ALTER SUBNET $ZZTCP.*.SN1, SUBNETMASK 255.255.0.0 -> ALTER SUBNET $ZZTCP.*.SN2, IPADDRESS 172.17.217.234 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 31 SCF Reference for Parallel Library TCP/IP ALTER SUBNET Command for TCPMAN The following example links the two failover-enabled subnets, SN1 and SN2, together as an adapter failover pair. The subnets SN1 and SN2 are configured to have different subnet IP address. -> ASSUME PROCESS $ZZTCP -> ADD SUBNET SN1,TYPE ETHERNET, DEVICENAME LANLIF2, & IPADDRESS 172.17.217.232, SUBNETMASK 255.255.255.0, & FAILOVER NONSHAREDIP -> ADD SUBNET SN2,TYPE ETHERNET, DEVICENAME LANLIF3, & IPADDRESS 172.17.217.234, SUBNETMASK 255.255.255.0, & FAILOVER NONSHAREDIP -> ALTER SUBNET SN1, ASSOCIATESUB "SN2" The following example links the two failover-enabled subnets, SN3 and SN4, together as an adapter failover pair. The subnets SN3 and SN4 are configured to have the same subnet IP address. -> ASSUME PROCESS $ZZTCP -> ADD SUBNET SN3,TYPE ETHERNET,DEVICENAME LANLIF4, & IPADDRESS 172.17.217.44, SUBNETMASK 255.255.255.0, FAILOVER & SHAREDIP -> ADD SUBNET SN4,TYPE ETHERNET,DEVICENAME LANLIF5,& IPADDRESS 172.17.217.44, SUBNETMASK 255.255.255.0, FAILOVER & SHAREDIP -> ALTER SUBNET SN3, ASSOCIATESUB "SN4", RESERVEDIP & 172.17.217.45 Considerations • • • You cannot ALTER a subnet attribute on only one TCPMON. ALTER SUBNET only accepts the wild-card (*) notation for the process name on the TCPMON object; hence, alterations to a subnet change all TCPMONs. The wild card, however, is optional; if you do not specify it, the wild card is assumed. See Table 5-2 on page 5-7 for naming conventions and reserved object names. The object must be in the STOPPED summary state when the ALTER command is issued. Note. The ADDALIAS and DELETEALIAS attributes are exceptions. Both these attributes can also be altered when the subnet is in the STARTED state. • • • When the ALTER command is completed, the object remains in the same summary state that existed before you issued the command. Use the INFO command to view the current attribute values. Do not use the RESERVED IP address for data traffic because the RESERVED IP address is not guaranteed to survive an adapter failure. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 32 SCF Reference for Parallel Library TCP/IP • • • • • DELETE Command No other option parameters are allowed in the ALTER SUBNET command when the ASSOCIATESUB parameter is present. For the two subnets configured as a failover pair, you cannot assign an alias address before the ASSOCIATESUB is done. For the two subnets configured as a failover pair, you cannot alter the subnet IP address or subnetmask once they are configured. In a failover-enabled SUBNET pair with SHAREDIP, one SUBNET is designated as primary after completion of the ALTER SUBNET, ASSOCIATESUB command. Alias IP addresses are added by using the ALTER SUBNET , ADDALIAS command. This consideration documents the failover behavior of alias IP addresses. If the SUBNET is configured for failover, all IP aliases are also configured for failover as long as the IP alias address is added to both SUBNETs in the failover pair. This arrangement is true for both SHARED and NONSHARED failover configurations. If the alias IP address is only added to one SUBNET in a failover pair, the alias IP address does not switch to another SUBNET when its SUBNET fails. DELETE Command The DELETE command removes entry names, subnets, and routes from the Parallel Library TCP/IP subsystem. You cannot delete a process or a TCPMON. (To delete TCPMONs from the subsystem, see the ABORT MON Command for TCPMAN on page 5-12.) This is a sensitive command. DELETE ENTRY Command for TCPMAN The DELETE ENTRY command removes entries from the ARP table. Entries can be deleted by specifying the entry name or by specifying the IP address. Specifying the IP address is the only way dynamically added entries can be deleted. Command Syntax DELETE [ /OUT file-spec/ ] [ ENTRY $ZZTCP.*.entry-name ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 33 SCF Reference for Parallel Library TCP/IP DELETE ROUTE Command for TCPMAN ENTRY $ZZTCP.*.entry-name is the name of the ENTRY object to be deleted. The fully-qualified entry name is $ZZTCP.*.entry-name (you must alter the subnet on all configured TCPMONs). You can delete all entries by substituting the wild card (*) for the entry-name. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. Examples The following example deletes the entry on all TCPMONs: -> DELETE ENTRY EA1 Considerations • • • When the DELETE operation is completed, the definition of the rout you specified for deletion is removed from the subsystem. Before you can delete a route, it must be in the STOPPED summary state. The DELETE SUBNET operation also deletes all the routes dependent on this interface/subnet and removes the static routes dependent on this interface/subnet from the configuration database. See DELETE SUBNET Command for TCPMAN on page 5-35. DELETE ROUTE Command for TCPMAN The DELETE ROUTE command removes a ROUTE from the Parallel Library TCP/IP subsystem. Only ROUTEs in the STOPPED state may be deleted. Command Syntax DELETE [ / OUT file-spec / ] [ROUTE $ZZTCP.#ZPTMn.route-name] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. ROUTE $ZZTCP.#ZPTMn.route-name is the name of the route. When you delete a route, you must do so on all configured TCPMONs (except dynamic routes, see the second example below. You can use the wild-card (*) notation for the TCPMON name, but if you do not, it is assumed. For example, DELETE ROUTE *.RT1 is equivalent to DELETE ROUTE RT1. You can substitute the wild card (*) for the route-name to delete all routes. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 34 SCF Reference for Parallel Library TCP/IP DELETE SUBNET Command for TCPMAN Examples The following example deletes the specified route from all TCPMONs. Note that the wild card (#) is assumed for the TCPMON. -> DELETE ROUTE $ZZTCP.*.RT0 The following command is valid because the dynamic route DR1_1 was created in processor 1. -> DELETE ROUTE $ZZTCP.#ZPTM1.DR1_1 Considerations • • • Only link-level routes, generated internally by the ARP logic, can be deleted without being brought to a STOPPED state. The DELETE SUBNET Command for TCPMAN also deletes all the routes dependent on the SUBNET (including static routes) from the system configuration database. Deleting a dynamic route not created in that processor is not allowed. DELETE SUBNET Command for TCPMAN The DELETE SUBNET command removes a subnet from the Parallel Library TCP/IP subsystem. When you delete a SUBNET, you must do so on all configured TCPMONs. Only subnets in the STOPPED summary state may be deleted. Command Syntax DELETE [/ OUT file-spec / ] [ SUBNET $ZZTCP.*.subnet-name] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. SUBNET $ZZTCP.*.subnet-name is the name of the subnet. Since you must delete subnets on all configured TCPMONs, the wild card (*) is assumed for the TCPMON name. You can also substitute the wild card (*) for the subnet-name to delete all subnets. If you omit the process name, or the subnet name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for GSeries RVUs. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 35 SCF Reference for Parallel Library TCP/IP INFO Command Examples The following command deletes all subnets in the Parallel Library TCP/IP: SCF> DELETE SUBNET $ZZTCP.*.* Considerations • • • • • • • The object-name template (wild-card notation) is supported. When the DELETE operation is completed, the definition of the subnet you specified for deletion is removed from the system configuration database. Before you can delete a subnet, it must be in the STOPPED summary state. The DELETE SUBNET operation also deletes all the routes dependent on this interface/subnet and removes the static routes dependent on this interface/subnet from the system configuration database. A subnet configured with failover enabled cannot be deleted unless its associated brother is also brought to the STOPPED state. When one of the subnet in the failover pair is deleted, the associated subnet of this subnet is also deleted. When one of the subnet in the failover pair is deleted, the Failover Alias IP address and all the additional aliases in its brother are also removed. INFO Command The INFO command displays the current attribute values for the specified PTCPIP object. Alterable attributes are indicated with an asterisk (*). This is a nonsensitive command. INFO ENTRY Command for TCPMAN The INFO ENTRY command displays the ARP table for the given entry. INFO [ /OUT file-spec/ ] [ ENTRY $ZZTCP.#ZPTMn.entry-name ] [ , IPADDRESS ip-addr | , OBEYFORM] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 36 SCF Reference for Parallel Library TCP/IP INFO ENTRY Command for TCPMAN Entry $ZZTCP.#ZPTMn.entry-name is the name of the entry. The fully-qualified entry name is $ZZTCP.#ZPTMn.entry-name. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. IPADDRESS ip-addr is the IP address of the entry. OBEYFORM causes the static ARP table configuration to be displayed in ADD ENTRY format, so that this configuration can be re-created. Examples The first example returns information about all entries in the ARP table. The second example returns information about the static ARP table in ADD ROUTE format. -> INFO ENTRY $ZZTCP.*.* -> INFO ENTRY *, OBEYFORM INFO ENTRY Display Format The format of the display for the INFO ENTRY command table is: TCPIP Info ENTRY \SAMCAT.$ZZTCP.*.* Name: (ARP) IPADDRESS.... 172.16.119.1 MacAddress... %H00 000C 3920CE Name is the name of the entry. The entry type is indicated in parentheses to the right. The only possible type is ARP. IPADDRESS is the IP address for the entry in dotted decimal format. MacAddress is the MAC (physical) address of the entry in hexadecimal format. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 37 INFO MON Command for TCPMAN SCF Reference for Parallel Library TCP/IP INFO ENTRY With OBEYFORM Display Format The format of the display for the INFO ENTRY command with the OBEYFORM attribute specified is: ADD ENTRY EA1 , TYPE ARP,& IPADDRESS 172.17.220.10 ,& MACADDR %H08008E003578 INFO MON Command for TCPMAN The INFO MON command displays the current attribute settings for the PTCPIP subsystem in a given TCPMON or in all configured TCPMONs. Command Syntax INFO[ /OUT file-spec/] [ MON [, DETAIL | , OBEYFORM] $ZZTCP.#ZPTMn ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. MON $ZZTCP.#ZPTMn is the name of the TCPMON. The TCPMON is always named #ZPTMn where n is the hexadecimal number of the processor in which the TCPMON is running. If you want to get info on a specific TCPMON, you must specify the TCPMON number. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for GSeries RVUs. OBEYFORM causes the altered TCPMON attributes to be displayed in ALTER MON format, so that you can re-create these TCPMON attributes. DETAIL specifies that the display is to include additional detailed information on the object. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 38 SCF Reference for Parallel Library TCP/IP INFO MON Command for TCPMAN Examples The first command displays the current attribute settings for the TCPMON in processor 0. The second command displays the altered TCPMON attributes in ALTER MON format. -> INFO MON $ZZTCP.#ZPTM0, DETAIL -> INFO MON *, OBEYFORM INFO MON Display Format The format of the display for the INFO MON $ZZTCP.#ZPTM0 , DETAIL command is: TCPMAN Detailed Info MON \OSCAR.$ZPTM0 *TCP Send Space......... *UDP Send Space......... *Delay Ack Time......... *Keep Alive Idle........ *Keep Alive Interval.... *Host ID ............... *Host Name ............. Program Filename ...... *Debug.................. *Full Dump.............. *All Nets Are Local..... *TCP Compat 42.......... *EXPAND Security........ *TCP Path MTU........... *TCP Time Wait.......... Trace Status........... Trace Filename ........ *RFC1323 Enable ........ *TCP Init Rexmit Timeout *TCP Min Rexmit Timeout. *TCP Listen Queue Min... *Initial TTL............ *Min-Ephemeral-Port..... *Max-Ephemeral-Port..... 8192 *TCP Receive Space...... 9216 *UDP Receive Space...... 5 *Delay Ack.............. 45 *Keep Alive Retry Cnt... 45 QIO Limit.............. 0D tcp0 \OSCAR.$SYSTEM.SYS02.TCPMON OFF ON ON ON OFF ON 60 OFF 8192 41600 ON 8 100% ON 1000 ms 1000 ms 5 64 1024 65024 TCP Send Space is the space reserved for send operations for the TCP protocol. TCP Receive Space is the space reserved for receive operations for the TCP protocol. UDP Send Space is the space reserved for send operations for the UDP protocol. UDP Receive Space is the space reserved for receive operations for the UDP protocol. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 39 SCF Reference for Parallel Library TCP/IP INFO MON Command for TCPMAN Delay Ack Time is the amount of time in 10 ms intervals that the acknowledgments are delayed. Delay Ack is a switch indicating if TCP is delaying acknowledgments. Keep Alive Idle is the amount of time in seconds before TCP issues a keep alive packet on sockets that have enabled this option. Keep Alive Retry Cnt is the number of times a keep alive packet is sent without receiving an acknowledgment, after which the TCP connection is dropped. Keep Alive Interval is the time interval in seconds between retransmissions of unacknowledged keep-alive packets. QIO Limit is a percentage between 0 and 100, representing the amount of queued I/O or shared memory allowed to this process. This attribute is not used by Parallel Library TCP/IP. The default is 100 percent. Host Id is the ID (usually the host number part of the internet address that is assigned to this host). It is a 32-bit number. Host Name is the official name of the host upon which the PTCPIP process is running is known in the Internet. This is a character string no longer than 50 characters. The default is the EXPAND node name with the leading “\” stripped off. Program File Name is the name of the file that is being executed for this process. Debug is the current setting (ON or OFF) of the DEBUG attribute. Debug is used by HP support and development personnel. Full Dump is a switch that allows the PTCPIP process to either save the QIO segment when abending if set to ON or just the stack if set to OFF. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 40 SCF Reference for Parallel Library TCP/IP INFO MON Command for TCPMAN ALLNETSARELOCAL is ON to cause TCP to use the interface MTU as a base for the determination of the TCP Maximum Segment Size (MSS) for each non-local TCP connection. A non-local TCP connection is one that goes to another network (not just another subnetwork). The default is ON. If this switch is OFF, TCP conforms to RFCspecified behavior and use 512 bytes as the default MSS for non-local segments. When ON, for example for Ethernet, the non-local MSS is 1460. Having this parameter set to ON can benefit performance. TCPCOMPAT42 is the flag used to set the PTCPIP process compatible with BSD4.2 versions. The default value of this flag is ON. If the flag is ON then original ACK - 1 is sent in the keepalive packet, otherwise the original ACK is sent in the keepalive packet. EXPAND Security is ON to cause TCP to check if a SOCKET request from another HP Expand node has passed the Expand security check. This means the user is valid on this system and has correct remote passwords. If the check fails then the SOCKET request is rejected with file error 48. The default for this option is OFF. TCP Path MTU is ON to cause TCP to use PATH MTU discovery on all TCP type sockets (SOCK_STREAM), unless disabled by the SETSOCKOPT for SO_PMTU. The default for this option is OFF. TCP Time Wait is the amount of time in seconds that a TCP connection remains in the TIME_WAIT state. The default is 60 seconds. The range is 1 to 120. Trace Status is ON when the process is being traced using SCF. Trace Filename is the name of the current trace file. RFC1323 ENABLE is ON to cause TCP to support TCP Large Windows as documented in RFC 1323. When this option is enabled, Parallel Library TCP/IP uses the TCP Window Scale and Timestamp options as described in RFC 1323. The largest TCP window supported is 262144 bytes when this option is enabled, and 65535 when the option is disabled. The default for this option is ON. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 41 SCF Reference for Parallel Library TCP/IP INFO MON Command for TCPMAN TCP-INIT-REXMIT-TIMEOUT is the initial retransmit timer value in milliseconds to use on a TCP connection. When the first round-trip timer measurement is made on a TCP connection and the calculation is done to arrive at the retransmission timeout to use on the next packet sent, this value is used unless the calculated value is larger. This variable can be used to help reduce the number of premature retransmission timeouts. The default is 1000 milliseconds, or 1 second. The range is 200 to 30000 milliseconds. TCP-MIN-REXMIT-TIMEOUT is the minimum value allowed for the TCP retransmission timeout. If this value is too low, the PTCPIP process might generate premature retransmissions. If this value is set too high, real retransmissions are delayed, increasing the time for error recovery. The default is 1000 milliseconds. The range is 50 to 30000 milliseconds. TCP-LISTEN-QUE-MIN is the minimum queue length that is set on a TCP socket when the PTCPIP process handles a socket LISTEN or ACCEPT_NW1 function call. This value is used if the queue length specified in the socket request is lower, otherwise the queue length in the socket request is used. The default value is 5. The range is 1 to 1024. INITIAL TTL specifies the initial value for UDP and TCP TTL (Time To Live). The default is 64, but may be altered to 30. MIN-EPHEMERAL-PORT is the starting port number to allocate for TCP and UDP ephemeral ports. Ephemeral ports are those assigned by Parallel Library TCP/IP when an application has not bound to a specific port. The default is 1024. The allowable range is 1024 to (MAX-EPHEMERAL-PORT - 16). See Considerations on page 5-29. Everything below min-ephemeral-port requires super-group privileges. If you alter min-ephemeral-port to be greater than 1024, be aware that all ports between 1024 and min-ephemeral-port can only be opened by privileged users, that is, supergroup users. MAX-EPHEMERAL-PORT is the largest port number to allocate for TCP and UDP ephemeral ports. The default is 65024. The allowable range is (MIN-EPHEMERAL-PORT + 16) to 65535. Each TCPMON is allocated one sixteenth of the range between min-ephemeralport and max-ephemeral-port. For example, using the defaults, #ZPTM0 is allocated 1024-5023, #ZPTM1 is allocated 5024-9023 and so on. See Considerations on page 5-29. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 42 SCF Reference for Parallel Library TCP/IP INFO PROCESS Command for TCPMAN INFO MON With OBEYFORM Display Format The format of the display for INFO MON * , OBEYFORM is: ALTER MON *, & TCPSENDSPACE 9120 ,& UDPSENDSPACE 41600,& RFC1323-ENABLE OFF,& TCP-LISTEN-QUE-MIN 64,& MIN-EPHEMERAL-PORT 1024,& MAX-EPHEMERAL-PORT 65024 INFO PROCESS Command for TCPMAN The INFO PROCESS command displays the current attribute values for the TCPMAN process. The INFO PROCESS display is the same as INFO PROCESS, DETAIL. Command Syntax INFO [ / OUT file-spec / ] [ PROCESS $ZZTCP ] [ , DETAIL ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. PROCESS $ZZTCP is the name of the manager process (TCPMAN). If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. DETAIL specifies that the display is to include additional detailed information on the object. Examples The following commands request non-detailed and detailed information about the TCPMAN process. SCF> INFO PROCESS $ZZTCP SCF> INFO PROCESS $ZZTCP , DETAIL HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 43 SCF Reference for Parallel Library TCP/IP INFO PROCESS Command for TCPSAM INFO PROCESS Display Format The format of the display for the INFO PROCESS command both with and without the DETAIL option is the same: TCPMAN Info Process \SYSTEM.$ZZTCP PPID............ ( 2,289) BPID................... ( 3,271) PROCESS $ZZTCP is the name of the manager process (TCPMAN). If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. PPID is the primary processor and PIN of the process. BPID is the backup processor and PIN of the process. INFO PROCESS Command for TCPSAM This command displays the current attribute settings for the TCPSAM process. Command Syntax INFO [ / OUT file-spec / ] [ PROCESS tcpsam-name ] [, DETAIL] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. DETAIL specifies that the display is to include additional detailed information on the object. PROCESS tcpman-name is the name of the socket access method process (TCPSAM). If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 44 SCF Reference for Parallel Library TCP/IP INFO PROCESS Command for TCPSAM Examples The following commands request the non-detailed and the detailed information for the TCPSAM process named $SAM1: -> INFO PROCESS $SAM1 -> INFO PROCESS $SAM1 , DETAIL INFO PROCESS Display Format The format of the display for the INFO PROCESS command without the DETAIL option is (an asterisk (*) an alterable attribute; however, see Considerations on page 5-48): TCPIP Info PROCESS \BOBAFET.$SAM1 *TCPSendSpace 8192 *TCPReceiveSpace *UDPSendSpace *UDPReceiveSpace 8192 9216 41600 The format of the display for the INFO PROCESS command with the DETAIL option is: TCPIP Detailed Info PROCESS \BEAR.$SAM1 *TCP Send Space ...... 8192 *TCP Receive Space .. *UDP Send Space ...... 2048 *UDP Receive Space .. *Delay Ack Time....... 5 *Delay Ack........... *Keep Alive Idle...... 7200 *Keep Alive Retry Cnt *Keep Alive Interval.. 75 QIO Limit........... *Host Id.............. 0D *Host Name ........... tcp0 Program Filename..... \BEAR.$SYSTEM.SYS02.TCPSAM *Debug ............... OFF *Full Dump............ ON *All Nets Are Local... ON *TCP Compat 42........ ON *EXPAND Security...... OFF *TCP Path MTU......... OFF *TCP Time Wait........ 60 Trace Status.......... OFF Trace Filename ....... *RFC1323 Enable ...... ON *TCP Init Rexmit Timeout 1000 ms *TCP Min Rexmit Timeout. 400 ms *TCP Listen Queue Min... 5 *Initial TTL............ 6 8192 4128 ON 8 100% TCP Send Space is the amount of data (in bytes) that can be buffered in the TCP layer when sending data to a remote site. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 45 SCF Reference for Parallel Library TCP/IP INFO PROCESS Command for TCPSAM TCP Receive Space is the amount of data (in bytes) that can be buffered in the TCP layer when receiving data from a remote site. UDP Send Space is the amount of data (in bytes) that can be buffered in the UDP layer when sending data to a remote site. UDP Receive Space is the amount of data (in bytes) that can be buffered in the UDP layer when receiving data from a remote site. Delay Ack Time is the amount of time (in .01-second units) that acknowledgments are delayed. Delay Ack indicates whether the acknowledgment (ACK) should be delayed when a TCP packet is received from a remote site. Keep Alive Idle is the amount of time, in seconds, before TCP issues a keep-alive packet on sockets that have enabled this option. Keep Alive Retry Cnt is the number of times a keep-alive packet is sent without receiving an acknowledgment. When this value is exceeded, the TCP connection is dropped. Keep Alive Interval is the time interval, in seconds, between retransmissions of unacknowledged keepalive packets. QIO Limit is a percentage between 0 and 100, representing the amount of queued I/O or shared memory allowed to this process. This attribute is not used by Parallel Library TCP/IP. The default is 100 percent. Host ID identifies the host by number. Host Name is the host on which the PTCPIP process is running. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 46 SCF Reference for Parallel Library TCP/IP INFO PROCESS Command for TCPSAM Program Filename is the name of the file being executed for this process. Debug is the current setting (ON or OFF) of the DEBUG attribute. Debug is used by HP support and development personnel. Full Dump is the current setting (ON or OFF) of the FULLDUMP attribute. All Nets Are Local The default is ON. ON causes TCP to use the interface MTU as a base for determining the TCP Maximum Segment Size (MSS) for each non-local TCP connection. A non-local TCP connection is one that goes to another network (not just another subnetwork). If ALLNETSARELOCAL is OFF, TCP conforms to RFC-specified behavior and use 512 bytes as the default MSS for non-local segments. For example, for Ethernet, when ALLNETSARELOCAL is ON, the non-local MSS is 1460; setting ALLNETSARELOCAL to ON can improve performance. TCPCOMPAT42 is the flag that sets the PTCPIP process compatible with BSD4.2 versions as follows: • • The default value of this flag is ON. If the flag is ON, then the original ACK minus 1 is sent in the keepalive packet; if the flag is OFF, the original ACK is sent in the keepalive packet. EXPAND Security EXPANDSECURITY is ON to cause TCP to check if a SOCKET request from another Expand node has passed the Expand security check. This means the user is valid on this system and has correct remote passwords. If the check fails then the SOCKET request is rejected with file error 48. The default for this option is OFF. TCPPATHMTU is ON to cause TCP to use PATH MTU discovery on all TCP-type sockets (SOCK_STREAM) unless disabled by the SETSOCKOPT for SO_PMTU. The default for this option is OFF. TCPTIMEWAIT is the amount of time in seconds that a TCP connection remains in the TIME_WAIT state. The default is 60 seconds. The range is 1 to 120. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 47 SCF Reference for Parallel Library TCP/IP INFO PROCESS Command for TCPSAM Trace Status is ON when the process is being traced using SCF. Trace Filename is the name of the current trace file. RFC1323 Enable is ON to cause TCP to support TCP Large Windows as documented in RFC 1323. When this option is enabled, Parallel Library TCP/IP uses the TCP Window Scale and Timestamp options as described in RFC 1323. The largest TCP window supported is 262144 bytes when this option is enabled, and 65535 when the option is disabled. The default for this option is ON. TCP-INIT-REXMIT-TIMEOUT is the initial retransmit timer value in milliseconds to use on a TCP connection. When the first round trip timer measurement is made on a TCP connection and the calculation is done to arrive at the retransmission timeout to use on the next packet sent, this value is used unless the calculated value is larger. This variable can be used to help reduce the number of premature retransmission timeouts. The default is 1000 milliseconds, or 1 second. The range is 200 to 30000 milliseconds. TCP-MIN-REXMIT-TIMEOUT is the minimum value allowed for the TCP retransmission timeout. If this value is too low the PTCPIP process might generate premature retransmissions. If this value is set too high, real retransmissions is delayed, increasing the time for error recovery. The default is 1000 milliseconds. The range is 50 to 30000 milliseconds. TCP Listen Queue Min is the minimum queue length that is set on a TCP socket when the PTCPIP process handles a socket LISTEN or ACCEPT_NW1 function call. This value is used if the queue length specified in the socket request is lower, otherwise the queue length in the socket request is used. The default value is 5. The range is 1 to 1024. INITIAL-TTL specifies the initial value for UDP and TCP TTL. The default is 64, but may be altered to 30. The only valid values are 30 and 64. Considerations Even though the detailed display option for the PROCESS object has an asterisk (*) in front of some fields, they are not alterable. TCPSAM does not support the ALTER command. In order to alter those parameters which have an asterisk (*) in front of them in the display, alter the TCPMON object instead. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 48 SCF Reference for Parallel Library TCP/IP INFO ROUTE Command for TCPMAN INFO ROUTE Command for TCPMAN The INFO ROUTE command for TCPMAN displays attribute values for the specified route(s). Command Syntax INFO [ / OUT file-spec / ] [ ROUTE $ZZTCP.#ZPTMn.route-name ] [, OBEYFORM ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. ROUTE route-name is the name of the route. To obtain info about a route on all configured TCPMONs, use the wild-card (*) notation for the TCPMON name. For example, INFO ROUTE *.RT1. To obtain info about a ROUTE on one TCPMON, qualify the TCPMON name. For example, INFO ROUTE #ZPTM1.RT1. Note that both of these examples had ASSUME(d) the process $ZZTCP. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. OBEYFORM causes the static route configuration to be displayed in ADD ROUTE format, so that this configuration can be re-created. Examples The first command returns the attributes of routes configured on the TCPMON in processor 2. The second command returns the route information in ADD ROUTE format. -> INFO ROUTE $ZZTCP.#ZPTM2.* -> INFO ROUTE *, OBEYFORM HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 49 INFO ROUTE Command for TCPMAN SCF Reference for Parallel Library TCP/IP INFO ROUTE Display Format The following display shows the output of the first example: PTCPIP Info ROUTE \BOBAFET.$ZZTCP.#ZPTM2.* Name Subnet Destination name RT2 EN1 172.17.215.0 RT3 EN2 172.17.215.0 RT4 EN3 172.17.195.0 RT5 DA2_2 DA2_3 MR3 MR4 MR5 DR2_1 DEF RT6 LOOP0 EN1 EN1 EN1 EN1 EN1 EN3 EN1 EN1 127.0.0.1 172.17.215.1 172.17.215.2 155.186.70.0 155.186.70.0 130.186.0.0 0.0.0.0 0.0.0.0 130.186.72.0 Netmask Gateway 255.255.255.0 255.255.255.255 255.255.255.0 255.255.255.255 255.255.255.0 255.255.255.255 255.255.255.255 255.255.255.255 255.255.255.255 255.255.255.0 255.255.255.0 255.255.0.0 0.0.0.0 0.0.0.0 255.255.255.0 Type Metric 172.17.215.32 C 0 172.17.215.34 C 0 172.17.195.34 C 0 127.0.0.1 172.17.215.32 172.17.215.32 172.17.215.1 172.17.215.2 172.17.215.2 172.17.195.2 172.17.215.1 172.17.215.2 H HLc HLc GS GS GSC GR GS Gc 0 0 0 1 1 1 1 1 1 Name • • • • • Routes created by internal route-redirect logic have the name of format DDcpu_n where cpu is the processor number in hexadecimal format where the route is generated, and n is a decimal number. Routes created by ARP link-level logic have the name of format DAcpu_n where cpu is the processor number in hexadecimal format where the route is generated, and n is a decimal number. Routes created by internal IRDP logic have the name of format DRcpu_n where cpu is the processor number in hexadecimal format where the route is generated, and n is a decimal number. Routes generated implicitly because of an ADD SUBNET or ALTER SUBNET, subnetmask command, start with “RT”. These kinds of routes are called implicit routes. The link-level route generated from the ADD ENTRY command has the same name as the entry name. The entry name must start with “EA”. Subnetname specifies the name of the subnetwork interface that is used by a specific route. Destination is the remote machine or network that can be reached via the machine specified in the GATEWAY. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 50 INFO ROUTE Command for TCPMAN SCF Reference for Parallel Library TCP/IP Netmask is the subnetmask associated with the route entry. Gateway is the machine through which the remote machine or network specified in DESTINATION is to be reached. Type indicates one of the following: blank routes to a network. H host Route G gateway Route C route with cloning capability c route cloned from a Cloning Route. S manually generated route. L route generated by ARP logic (Link level route). R route generated by IRDP logic. If ICMP Router Discovery Protocol (IRDP) is enabled on a subnet, default routes discovered by IRDP is indicated as Type Gateway/Router (G, R). D route generated from rtredirect (route redirect) logic. Metric indicates the number of hops to the destination. INFO ROUTE With OBEYFORM Display Format The following display shows the output of the second command: ADD ROUTE MR3 , DESTINATION GATEWAY 172.17.215.1 , NETMASK METRIC 1 ADD ROUTE MR4 , DESTINATION GATEWAY 172.17.215.2 , NETMASK METRIC 1 ADD ROUTE MR5 , DESTINATION GATEWAY 172.17.215.2 , NETMASK METRIC 1 ADD ROUTE DEF , DESTINATION GATEWAY 172.17.215.2 , NETMASK METRIC 1 155.186.70.0 %FFFFFF00 ,& ,& 155.186.70.0 %FFFFFF00 ,& ,& 130.186.0.0 %FFFF0000 ,& ,& 0 %00000000 ,& ,& HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 51 SCF Reference for Parallel Library TCP/IP INFO ROUTE Command for TCPSAM Considerations • • The implicit route generated internally from the ADD SUBNET command has the cloning flag set. See the ADD ROUTE help text for a detailed description of the cloning capability of a route. Link level routes, generated internally by the ARP logic, cannot be stopped externally through the SCF ABORT or STOP ROUTE commands but can be deleted externally through the SCF DELETE ROUTE command. INFO ROUTE Command for TCPSAM The INFO ROUTE command displays attribute settings for the specified route(s) configured on the TCPMON object in the TCPSAM primary processor. This is a nonsensitive command. Command Syntax INFO [ /OUT file-spec/ ] [ ROUTE $tcpsam-name.route-name ] [ , OBEYFORM] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. ROUTE $tcpsam-name.route-name is the name of the route. If you omit the object name, SCF uses the assumed object name. You do not need to ASSUME the MON object; if you omit it, the wild card (*) is assumed. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. OBEYFORM where OBEYFORM causes the static route configuration to be displayed in ADD ROUTE format, so that this configuration can be re-created. Examples -> INFO ROUTE $ZSAM2.* -> INFO ROUTE *, OBEYFORM HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 52 INFO ROUTE Command for TCPSAM SCF Reference for Parallel Library TCP/IP INFO ROUTE Display Format The display format for INFO ROUTE for TCPSAM is: TCPIP Info ROUTE \BOBAFET.$ZSAM2.* Name Subnetname Destination Gateway Type #RT2 #RT3 #RT4 #RT5 #DA2_2 #DA2_3 #MR3 #MR4 #MR5 #DR2_1 #DEF #RT6 #EN1 #EN2 #EN3 #LOOP0 #EN1 #EN1 #EN1 #EN1 #EN1 #EN3 #EN1 #EN1 172.17.215.0 172.17.215.0 172.17.195.0 127.0.0.1 172.17.215.1 172.17.215.2 155.186.70.0 155.186.70.0 130.186.0.0 0.0.0.0 0.0.0.0 130.186.72.0 172.17.215.32 172.17.215.34 172.17.195.34 127.0.0.1 172.17.215.32 172.17.215.32 172.17.215.1 172.17.215.2 172.17.215.2 172.17.195.2 172.17.215.1 172.17.215.2 H H H G G G GR G G Name • • • • • Routes created by internal route-redirect logic have the name of format DDcpu_n where cpu is the CPU number in hexadecimal format where the route is generated, and n is a decimal number. Routes created by ARP link-level logic have the name of format DAcpu_n where cpu is the CPU number in hexadecimal format where the route is generated, and n is a decimal number. Routes created by internal IRDP logic have the name of format DRcpu_n where cpu is the CPU number in hexadecimal format where the route is generated, and n is a decimal number. Routes generated implicitly because of an ADD SUBNET or ALTER SUBNET, subnetmask command, start with “RT”. These kinds of routes are called implicit routes. The link-level route generated from the ADD ENTRY command has the same name as the entry name. The entry name must start with “EA”. Subnetname specifies the name of the subnetwork interface that is used by a specific route. Destination is the remote machine or network that can be reached via the machine specified in the GATEWAY. Netmask is the subnetmask associated with the route entry. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 53 SCF Reference for Parallel Library TCP/IP INFO ROUTE Command for TCPSAM Gateway is the machine through which the remote machine or network specified in DESTINATION is to be reached. Type indicates one of the following: blank routes to a network. H host Route G gateway Route R route generated by IRDP logic. If ICMP Router Discovery Protocol (IRDP) is enabled on a subnet, default routes discovered by IRDP is indicated as Type Gateway/Router (G, R). D route generated from rtredirect (route redirect) logic. The format of the display for INFO ROUTE, OBEYFORM is: -> INFO ROUTE *, OBEYFORM ADD ROUTE MR3 , DESTINATION GATEWAY 172.17.215.1 , NETMASK METRIC 1 ADD ROUTE MR4 , DESTINATION GATEWAY 172.17.215.2 , NETMASK METRIC 1 ADD ROUTE MR5 , DESTINATION GATEWAY 172.17.215.2 , NETMASK METRIC 1 ADD ROUTE DEF , DESTINATION GATEWAY 172.17.215.2 , NETMASK METRIC 1 155.186.70.0 %FFFFFF00 ,& ,& 155.186.70.0 %FFFFFF00 ,& ,& 130.186.0.0 %FFFF0000 ,& ,& 0 %00000000 ,& ,& HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 54 SCF Reference for Parallel Library TCP/IP INFO SUBNET Command for TCPMAN INFO SUBNET Command for TCPMAN The INFO SUBNET command displays the current attribute values for the specified subnets. Command Syntax INFO [ / OUT file-spec / ] [SUBNET $ZZTCP.#ZPTMn.subnet-name] [, DETAIL | , OBEYFORM] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. SUBNET $ZZTCP.#ZPTMn.subnet-name is the name of the subnet. To obtain info about the subnets on all configured TCPMONs, use the wild-card (*) notation for the TCPMON name. For example, INFO SUBNET $ZZTCP.*.SN1. To obtain info about the subnet on one TCPMON, qualify the TCPMON name. For example, INFO SUBNET $ZZTCP.#ZPTM1.SN1. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. DETAIL specifies that the display is to include additional detailed information about the object. OBEYFORM causes the subnet configuration to be displayed in ADD SUBNET and ALTER SUBNET formats, so that this configuration can be re-created. Examples The first example returns information about a specific subnet, the second example returns information about all running subnets on the system, the third example returns detailed information about a specific subnet, and the fourth example displays the subnet configuration in ADD SUBNET and ALTER SUBNET formats: -> INFO SUBNET $ZZTCP.#ZPTM1.SN1 -> INFO SUBNET $ZZTCP.#ZPTM1.* -> INFO SUBNET $ZZTCP.#ZPTM2.SN2, DETAIL -> INFO SUBNET *, OBEYFORM HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 55 INFO SUBNET Command for TCPMAN SCF Reference for Parallel Library TCP/IP INFO SUBNET for TCPMAN Display Format The format of the display for the first example is the following (an asterisk (*) indicates an alterable attribute): PTCPIP Info SUBNET \BEAR.$ZZTCP.#ZPTM1.SN1 Name SN1 Devicename \BEAR.LANLIF2 *IPADDRESS 172.17.217.234 TYPE *SUBNETMASK ETHERNET %HFFFFFF00 QIO *R ON N The format of the display for the second example is (an asterisk (*) indicates an alterable attribute): PTCPIP Info SUBNET \BEAR.$ZZTCP.#ZPTM1.* Name LOOP0 SN1 SN2 Devicename \NOSYS.$NOIOP \BEAR.LANLIF2 \BEAR.LANLIF3 *IPAddRESS 127.0.0.1 172.17.217.234 172.17.217.232 TYPE *SUBNETMASK QIO *R LOOP-BACK %HFF000000 ETHERNET %HFFFFFF00 ETHERNET %HFFFFFF00 OFF N ON N ON N The format of the display for the DETAIL example is (an asterisk (*) indicates an alterable attribute): PTCPIP Detailed Info SUBNET \BEAR.$ZZTCP.#ZPTM2.SN2 Name Devicename *IPADDRESS TYPE SN2 \BEAR.LAN03 172.17.208.22 ETHERNET Trace Status .......... OFF Trace Filename ........ Interface MTU ......... 1500 ---Multicast Groups-----State--239.246.67.20 STARTED 239.31.50.19 STARTED 238.72.33.18 STARTED 237.113.16.17 STARTED 224.0.0.1 STARTED *SUBNETMASK %HFFFFFF00 QIO *R ON N Name is the name of the subnet. Devicename is the name of the SLSA LIF that provides access to the Ethernet LAN. Note that with loopback subnets, the value \NOSYS.NOIOP, meaning no system, no device name, is displayed. This value is displayed because loopback subnets are routed internally so that there is no device name to display. IPADDRESS is the Internet address of this subnet and all the IP addresses of the aliases associated with the subnet. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 56 SCF Reference for Parallel Library TCP/IP INFO SUBNET Command for TCPMAN TYPE is the subnet type. Possible values are Ethernet and loopback. SUBNETMASK is a 32-bit integer that specifies which portion of the network number and the IP host address is to be masked to define a subnet. QIO shows whether or not the subnet is currently using the QIO interface. QIO is always on for Ethernet type subnets. ON indicates that the interface is currently using QIO mode. OFF indicates that the interface is not currently using QIO mode. Trace Status shows whether the subnet is being traced. ON indicates that it is being traced. Trace Filename is the name of the current trace file. Interface MTU is the maximum transmission unit that can be used on the subnet. R shows whether or not the ICMP Router Discovery Protocol (IRDP) has been enabled on the subnet. The displayed value can be Y (IRDP is ON), or N (IRDP is OFF). Multicast Groups is the list of internet addresses joined by an application. State is the filter registration state. The possible state values are: STARTED, STARTING, and STOPPED. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 57 SCF Reference for Parallel Library TCP/IP INFO SUBNET Command for TCPSAM The format of the display for the fourth example, OBEYFORM, is: ALTER SUBNET LOOP0 , IPADDRESS 127.0.0.1 ADD SUBNET EN1 , TYPE ETHERNET,& IPADDRESS 172.17.222.15 , SUBNETMASK %HFFFFFF00 ,& DEVICENAME \SAMCAT.LANLIF1, FAILOVER NONSHAREDIP ALTER SUBNET EN1 , ASSOCIATESUB "EN2" ALTER SUBNET EN1 , ADDALIAS 172.17.222.120 ALTER SUBNET EN1 , ADDALIAS 172.17.222.121 ALTER SUBNET EN1 , ADDALIAS 172.17.222.122 ALTER SUBNET EN1 , ADDALIAS 172.17.222.123 ALTER SUBNET EN1 , ADDALIAS 172.17.222.124 ALTER SUBNET EN1 , ADDALIAS 172.17.222.125 ALTER SUBNET EN1 , ADDALIAS 172.17.222.126 ALTER SUBNET EN1 , ADDALIAS 172.17.222.127 ADD SUBNET EN2 , TYPE ETHERNET,& IPADDRESS 172.17.222.16 , SUBNETMASK %HFFFFFF00 ,& DEVICENAME \SAMCAT.LANLIF2, FAILOVER NONSHAREDIP ALTER SUBNET EN2 , ASSOCIATESUB "EN1" INFO SUBNET Command for TCPSAM The INFO SUBNET command displays attribute settings for the specified subnet(s) configured in the TCPSAM primary processor. Command Syntax INFO [ / OUT file-spec / ] [ SUBNET $tcpsam-name.subnet-name ] [, DETAIL ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. SUBNET $tcpsam-name.subnet-name is the name of the subnet. To obtain info about all subnets configured for the TCPSAM process, use the wild-card (*) notation. For example, INFO SUBNET $ZTC1.*. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for GSeries RVUs. Examples The following command returns information about all running subnets on the TCPMON object running in TCPSAM’s primary processor. -> INFO SUBNET $ZTC1.* HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 58 INFO SUBNET Command for TCPSAM SCF Reference for Parallel Library TCP/IP INFO SUBNET for TCPSAM Display Format The format of the INFO SUBNET display for TCPSAM is (an asterisk (*) indicates an alterable attribute): TCPIP Info SUBNET \OSCAR.$ZTC1.* Name Devicename *IPADDRESS #LOOP0 \NOSYS.$NOIOP 127.0.0.1 TYPE *SUBNETMASK SuName LOOP-BACK %HFF000000 QIO *R OFF N TCPIP Info SUBNET \OSCAR.$ZTC1.* Name Devicename *IPADDRESS #SN2 \OSCAR.LAN04 172.17.221.74 TYPE *SUBNETMASK ETHERNET SuName %HFFFFFF00 QIO *R ON N Name is the name of the subnet. Devicename is the name of the SLSA LIF that provides access to the Ethernet LAN. Note that with loopback subnets, the value \NOSYS.NOIOP, meaning no system, no device name, is displayed. This value is displayed because loopback subnets are routed internally so that there is no device name to display. IPADDRESS is the Internet address of this subnet and all the IP addresses of the aliases associated with the subnet. TYPE is the subnet type. Possible values are Ethernet and loopback. SUBNETMASK is a 32-bit integer that specifies which portion of the network number and the IP host address is to be masked to define a subnet. SuName is not supported for Parallel Library TCP/IP. QIO shows whether or not the subnet is currently using the QIO interface. QIO is always on for Ethernet type subnets. ON indicates that the interface is currently using QIO mode. OFF indicates that the interface is not currently using QIO mode. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 59 SCF Reference for Parallel Library TCP/IP LISTOPENS Command R shows whether or not the ICMP Router Discovery Protocol (IRDP) has been enabled on the subnet. The displayed value can be Y (IRDP is ON), or N (IRDP is OFF). LISTOPENS Command The LISTOPENS command returns information on openers of the TCPMONs. This is a nonsensitive command. LISTOPENS MON Command for TCPMAN The LISTOPENS MON command displays information identifying the origins of the connections in a given TCPMON or in all TCPMONs. Command Syntax LISTOPENS[ /OUT file-spec/ ] [ MON $ZZTCP.#ZPTM{0-F } ] [,DETAIL ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. MON $ZZTCP.#ZPTM{0-F } is the name of the TCPMON object. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. DETAIL specifies that the display is to include additional detailed information on the object. Examples The following commands request non-detailed and detailed information about the openers of the specified process: -> LISTOPENS MON $ZZTCP.#ZPTM1 -> LISTOPENS MON $ZZTCP.#ZPTM1, DETAIL HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 60 LISTOPENS MON Command for TCPMAN SCF Reference for Parallel Library TCP/IP LISTOPENS MON Display Format The format of the display for the LISTOPENS command without the DETAIL option is (an asterisk (*) indicates an alterable attribute): PTCPIP LISTOPENS MON \BEAR.$ZZTCP.#ZPTM1 OPENERS $ZNET $ZPORT PPID 13,234 3,52 BPID PLFN 2 5 BLFN 0 0 PROTOCOL #ZSPI TCP LPORT * FTP OPENERS is the process name of the opener of the TCPMON. PPID is the primary processor and PIN of the opener. BPID is the backup processor and PIN of the opener. PLFN is the logical file number of the primary opener process. BLFN is the logical file number of the backup opener process. Protocol is the protocol accessed by the opener. Lport is the local port number for either TCP or UDP, depending on the value of Protocol. The more common port values are displayed in text form; others are displayed as four-decimal octets. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 61 LISTOPENS MON Command for TCPMAN SCF Reference for Parallel Library TCP/IP LISTOPENS MON Display Format With DETAIL The format of the display for the LISTOPENS MON command with the DETAIL option is: PTCPIP LISTOPENS MON \BEAR.$ZZTCP.#ZPTM3 <DETAIL-DISPLAY> OPENER OPENER $ZNET PROTO LADDR FADDR #ZSPI 0.0.0.0 0.0.0.0 $ZPORT PROTO TCP LADDR 0.0.0.0 FADDR 0.0.0.0 PPID 0,47 STATE LPORT FPORT BPID SENDQ 0 PLFN 3 RECVQ BLFN 0 0 0 PLFN 3 RECVQ BLFN 0 0 * * PPID 0,47 BPID STATE LISTEN SENDQ LPORT FTP FPORT * OPENER is the system name and process name of a opener of the TCPSAM Process. PPID is the primary processor and process ID of the opener. BPID is the backup processor and process ID of the opener. PLFN is the logical file number of the primary opener process. BLFN is the logical file number of the backup opener process. PROTO is the protocol of the opener. STATE is the state a particular socket is in. Only sockets with TCP protocol have states associated with them. The possible state values are: CLOSED, LISTEN, SYNSENT, SYN-RCVD, ESTAB, CLOSING, FIN-WAIT-1, FIN-WAIT-2, TIME-WAIT, CLOSE-WAIT, LAST-ACK, and FAIL-WAIT. SENDQ specifies the number of bytes of data in the send queue and receive queue of the socket HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 62 SCF Reference for Parallel Library TCP/IP LISTOPENS PROCESS Command for TCPSAM RECVQ specifies the number of bytes of data in the send queue and receive queue of the socket. LADDR specifies the local internet address associated with the socket (IP addresses). LPORT is the local port number for either TCP or UDP depending on the protocol listed in the PROTO field. LPORT is displayed in text form for the more common port values. It is displayed in decimal if they are not recognized as a common port. FADDR specifies the foreign Internet address associated with the socket (IP addresses). FPORT is the foreign port number for either TCP or UDP depending on the protocol listed in the PROTO field. FPORT is displayed in text form for the more common port values. It is displayed in decimal if they are not recognized as a common port. LISTOPENS PROCESS Command for TCPSAM The SCF LISTOPENS command returns the list of sockets, the source IP address, source port, destination IP address and destination port for connected sockets and the local port for listening sockets. TCPSAM routes this command to all the TCPMONs which are running on that system. However, it displays only information that is relevant to that TCPSAM process. Thus, you can display all the socket (OSS and Guardian) opens that use that particular TCPSAM as the transport provider on the entire system. Command Syntax LISTOPENS[ /OUT file-spec/ ] [ PROCESS $tcpsam-name ] [,DETAIL ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. PROCESS $tcpsam-name is the name of the TCPSAM process. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 63 LISTOPENS PROCESS Command for TCPSAM SCF Reference for Parallel Library TCP/IP DETAIL specifies that the display is to include additional detailed information on the object. Examples The following commands request non-detailed and detailed information about the openers of the specified process: -> LISTOPENS PROCESS $ZTC1 -> LISTOPENS PROCESS $ZTC1, DETAIL LISTOPENS PROCESS Display Format The format of the display for the LISTOPENS command without the DETAIL option is: TCPIP Listopens PROCESS \BOBAFET.$ZTC1 Openers $ZPT0 $ZPT0 $ZPT0 $ZTN0 PPID 0,314 0,314 0,314 0,305 BPID PLFN 4 5 6 3 BLFN 0 0 0 0 Protocol TCP TCP TCP TCP Lport echo finger ftp telnet The format of the display for the LISTOPENS command with the DETAIL option is: TCPIP Detailed Listopens PROCESS \BOBAFET.$ZTCP1 Opener $ZPT0 Proto Laddr Faddr TCP 0.0.0.0 0.0.0.0 Opener $ZPT0 Proto Laddr Faddr TCP 0.0.0.0 0.0.0.0 Opener $ZPT0 Proto Laddr Faddr TCP 0.0.0.0 0.0.0.0 Opener $ZTN0 Proto Laddr Faddr TCP 0.0.0.0 0.0.0.0 Ppid State 0,314 Bpid LISTEN SendQ Lport echo Fport * Plfn 4 0 RecvQ Blfn 0 0 Ppid State 0,314 Bpid LISTEN SendQ Lport finger Fport * Plfn 5 0 RecvQ Blfn 0 0 Ppid State 0,314 Bpid LISTEN SendQ Lport ftp Fport * Plfn 6 0 RecvQ Blfn 0 0 Ppid State 0,305 Bpid LISTEN SendQ Lport telnet Fport * Plfn 3 0 RecvQ Blfn 0 0 Opener is the system name and process name of a opener of the TCPSAM Process. Ppid is the primary processor and process ID of the opener. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 64 SCF Reference for Parallel Library TCP/IP LISTOPENS PROCESS Command for TCPSAM Bpid is the backup processor and process ID of the opener. Plfn is the logical file number of the primary opener process. Blfn is the logical file number of the backup opener process. Proto is the protocol of the opener. State is the state a particular socket is in. Only sockets with TCP protocol have states associated with them. The possible state values are: CLOSED, LISTEN, SYNSENT, SYN-RCVD, ESTAB, CLOSING, FIN-WAIT-1, FIN-WAIT-2, TIME-WAIT, CLOSE-WAIT, LAST-ACK, and FAIL-WAIT. SendQ specifies the number of bytes of data in the send queue and receive queue of the socket RecvQ specifies the number of bytes of data in the send queue and receive queue of the socket. Laddr specifies the local internet address associated with the socket (IP addresses). Lport is the local port number for either TCP or UDP depending on the protocol listed in the PROTO field. LPORT is displayed in text form for the more common port values. It is displayed in decimal if they are not recognized as a common port. Faddr specifies the foreign internet address associated with the socket (IP addresses). Fport is the foreign port number for either TCP or UDP depending on the protocol listed in the PROTO field. FPORT is displayed in text form for the more common port values. It is displayed in decimal if they are not recognized as a common port. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 65 SCF Reference for Parallel Library TCP/IP NAMES Command NAMES Command The NAMES command displays the names of the specified PTCPIP objects. This is a nonsensitive command. NAMES ENTRY Command for TCPMAN The NAMES ENTRY command displays the names of the ENTRY objects for the Parallel Library TCP/IP subsystem in a configured TCPMON or in all configured TCPMONs. Command Syntax NAMES [ /OUT file-spec/ ] [ ENTRY $ZZTCP.#ZPTMn.* ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. ENTRY $ZZTCP.#ZPTMn.* is the name of the entry. The wild card (*) is used in place of the entry-name because the purpose of the NAMES ENTRY command is to obtain a list of all entries. If you omit the process name, SCF uses the process name established in a previous ASSUME command. See the SCF Reference Manual for G-Series RVUs for more information about the ASSUME command. You may assume the process and TCPMON and you may use the wild card (*) for the TCPMON and entry. The wild card used in place of the TCPMON yields a list of entries on all configured TCPMONs. Examples The following command provides a list of entries for all entries on the #ZPTM2 TCPMON. -> NAMES ENTRY $ZZTCP.#ZPTM2.* NAMES ENTRY Display Format The display format of the NAMES ENTRY command is: TCPMAN Names ENTRY \BEAR.$ZZTCP.#ZPTM2.* ENTRY DA2_1 EA01 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 66 SCF Reference for Parallel Library TCP/IP NAMES ROUTE Command for TCPMAN NAMES ROUTE Command for TCPMAN The NAMES ROUTE command displays the names of the routes for the Parallel Library TCP/IP subsystem. Command Syntax NAMES [ / OUT file-spec / ] [ROUTE $ZZTCP.#ZPTMn.* ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. ROUTE ROUTE $ZZTCP.#ZPTMn.* is the name of the route. The route name is $ZZTCP.#ZPTMn.*. You can substitute the wild card (*) for the TCPMON name; doing yields NAMES information for routes on all TCPMONs. The wild card is used in place of the route-name because the purpose of the NAMES ROUTE command is to obtain a list of all routes. If you omit the process name, SCF uses the process name established in a previous ASSUME command. See the SCF Reference Manual for G-Series RVUs for more information about the ASSUME command. Examples The following commands request a list of the routes associated with $ZZTCP.#ZPTM1 and a list of all the routes in the Parallel Library TCP/IP subsystem: SCF> NAMES ROUTE $ZZTCP.#ZPTM1.* SCF> NAMES ROUTE $ZZTCP.*.* NAMES ROUTE Display Format The format of the display for the NAMES ROUTE command is: TCPMAN Names ROUTE \OSCAR.$ZZTCP.#zptm1.* ROUTE RT1 RT3 DEF The format of the display for the NAMES ROUTE command for all routes in the Parallel Library TCP/IP is: TCPMAN Names ROUTE \OSCAR.$ZZTCP.*.* ROUTE RT1 RT3 DEF HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 67 NAMES ROUTE Command for TCPSAM SCF Reference for Parallel Library TCP/IP NAMES ROUTE Command for TCPSAM The NAMES ROUTE command for TCPSAM displays the names of the routes configured in the same processor as the TCPSAM process. Command Syntax NAMES [ / OUT file-spec / ] [ROUTE $tcpsam-name.*] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. ROUTE $tcpsam-name.* is the name of the route. The wild card (*) is used in place of the route-name because the purpose of the NAMES ROUTE command is to obtain a list of all routes. If you omit the process name, SCF uses the process name established in a previous ASSUME command. See the SCF Reference Manual for G-Series RVUs for more information about the ASSUME command. Examples The following command requests a list of the routes running in the same processor as TCPSAM: SCF> NAMES ROUTE $ZTC1.* NAMES ROUTE Display Format The format of the display for the NAMES ROUTE command is: TCPIP Names ROUTE \BOBAFET.$ZTC1.* ROUTE #RT7 #RT8 #RT10 #DA2_2 #DEF HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 68 SCF Reference for Parallel Library TCP/IP NAMES SUBNET Command for TCPMAN NAMES SUBNET Command for TCPMAN The NAMES SUBNET command for TCPMAN displays the names of the SUBNETs for the Parallel Library TCP/IP subsystem in a configured TCPMON or in all configured TCPMONs. Command Syntax NAMES [ / OUT file-spec / ] [ SUBNET $ZZTCP.#ZPTMn.* ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. SUBNET $ZZTCP.#ZPTMn.* is the name of the subnet. The wild card (*) is used in place of the subnet-name because the purpose of the NAMES SUBNET command is to obtain a list of all subnets. If you omit the process or TCPMON name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. You may substitute the wild card (*) for the subnet name which yields the subnet names on all configured subnets. Examples The following command requests the names of the subnets associated with #ZPTM2: SCF> NAMES SUBNET $ZZTCP.#ZPTM2.* NAMES SUBNET Display Format The format of the display for the NAMES SUBNET command is: TCPMAN Names SUBNET \BEAR.$ZZTCP.#ZPTM2.* SUBNET LOOP0 EN1 EN2 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 69 SCF Reference for Parallel Library TCP/IP NAMES SUBNET Command for TCPSAM NAMES SUBNET Command for TCPSAM NAMES SUBNET for TCPSAM displays the names of the subnets configured on the TCPMON process in the same processor. Command Syntax NAMES [ / OUT file-spec / ] [ SUBNET $tcpsam-name.* ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. SUBNET $tcpsam-name.subnet-name is the name of the subnet. The fully-qualified name of the subnet for TCPSAM is $tcpsam-name.subnet-name. If you omit the process name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. You may substitute the wild card (*) for the subnet name which yields the subnet names on all configured subnets. Examples The following command requests the names of the subnets associated with $ZTC1: SCF> NAMES SUBNET $ZTC1.* NAMES SUBNET Display Format The format of the display for the NAMES SUBNET command is: TCPMAN Names SUBNET \BEAR.$ZTC1.* SUBNET #LOOP0 #EN1 #EN2 PRIMARY Command The PRIMARY command can be used when the Parallel Library TCP/IP subsystem is running as a fault-tolerant process pair. This command causes the backup processor to become the primary processor and the primary processor to become the backup processor. This is a sensitive command. PRIMARY PROCESS Command for TCPMAN The PRIMARY command for TCPMAN causes the backup processor for the TCPMAN process ($ZZTCP) to become the primary and the primary processor to become the backup. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 70 SCF Reference for Parallel Library TCP/IP PRIMARY PROCESS Command for TCPSAM Command Syntax PRIMARY [ / OUT file-spec / ] [ PROCESS $ZZTCP ] , CPU cpu-number OUT file-spec causes any SCF output generated for this command to be directed to the specified file. PROCESS $ZZTCP is the name of the TCPMAN process. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. CPU cpu-number is the number of the processor of the backup process. This attribute is required. Examples The following command causes the $ZTC1 primary process to become the backup process (the attribute CPU 2 identifies the processor where the former backup process resided): SCF> PRIMARY PROCESS $ZZTCP, CPU 2 Considerations If the specified processor is not the processor of the backup process, the command is rejected. PRIMARY PROCESS Command for TCPSAM The SCF PRIMARY PROCESS in TCPSAM lets you switch connections to a backup processor. The information maintained about the Guardian sockets is checkpointed to the backup when the backup is started. This lets the TCPSAM process switch the roles of the primary and backup processes. Command Syntax PRIMARY [ / OUT file-spec / ] [ PROCESS $tcpsam-name ] , CPU cpu-number OUT file-spec causes any SCF output generated for this command to be directed to the specified file. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 71 SCF Reference for Parallel Library TCP/IP START Command PROCESS tcpsam-name is the name of the TCPSAM process. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. CPU cpu-number is the number of the processor of the backup process. This attribute is required. Examples The following command causes the TCPSAM primary process to become the backup process (the attribute processor 2 identifies the processor where the former backup process resided): SCF> PRIMARY PROCESS $ZTC1, CPU 2 Considerations If the specified processor is not the processor of the backup process, the command is rejected. START Command The START command initiates the operation of TCPMONs, subnets and routes for the Parallel Library TCP/IP subsystem. When the subsystem has successfully completed processing this command, the specified object is placed in the STARTED summary state. You can start a subnet or a route, but not a process. If a process is not started, it is undefined. This is a sensitive command. START MON Command for TCPMAN The START MON command is used to start individual TCPMON objects on each processor. MONs are automatically started by TCPMAN if they were previously started by using SCF and were not aborted. Command Syntax START [ / OUT file-spec / ] MON $ZZTCP.#ZPTM{0-F } OUT file-spec causes any SCF output generated for this command to be directed to the specified file. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 72 SCF Reference for Parallel Library TCP/IP START ROUTE Command for TCPMAN $ZZTCP.#ZPTM{0-F } is the name of the TCPMON. The wild card (*) is supported. If you substitute the wild card for the TCPMON names, a TCPMON is started in every running processor. You can start TCPMONs in specific processors by listing a string of TCPMONs in parentheses, such as: ->ASSUME PROCESS $ZZTCP ->START MON (#ZPTM0, #ZPTM1, #ZPTM2, #ZPTM3) Caution. Starting TCPMONs in only some processors has repercussions for certain applications. If your application can be spawned in any processor, and you do not configure a TCPMON in every available processor, you need to change your application so that it does not spawn to a processor without a TCPMON object. Examples This command starts a TCPMON on processor 15: -> START MON $ZZTCP.#ZPTMF This command starts TCPMONs in all processors: -> START MON $ZZTCP.* Considerations The START MON command adds the MON to the system configuration database. You must follow the START MON command with a DELAY command to ensure that all the MONs start before you start using them. DELAY 21 will suffice. When you start a single MON after stopping it with the ABORT MON command, be sure to use the ALTER MON * command to re-configure the non-default attributes on the restarted MON. The preferred way to stop the MON to preserve the non-default attributes is to use the STOP MON command rather than the ABORT MON command. START ROUTE Command for TCPMAN The START ROUTE command creates implicit connections to and from a route. The successful completion of the START command leaves the route in the STARTED summary state. Command Syntax START [ / OUT file-spec / ] [ROUTE $ZZTCP.*.route-name ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 73 SCF Reference for Parallel Library TCP/IP START SUBNET Command for TCPMAN ROUTE $ZZTCP.*.route-name is the name of the route. The fully-qualified route name is $ZZTCP.*. route-name (you must start the route on all configured TCPMONs). If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. Examples The following command starts all routes under the assumed process and TCPMON: SCF> START ROUTE * The following command starts the specified route in all TCPMONs: SCF> START ROUTE $ZZTCP.*.RT1 START SUBNET Command for TCPMAN The START SUBNET command creates implicit connections to and from a subnet. The subnet transitions through the STARTING state and, upon successful completion, ends in the STARTED summary state. You must start the subnet in all configured TCPMONs. Command Syntax START [ / OUT file-spec / ] [SUBNET $ZZTCP.#ZPTMn.subnet-name] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. SUBNET $ZZTCP.#ZPTMn.subnet-name is the name of the subnet. The fully-qualified subnet name is $ZZTCP.*.subnetname (you must alter the subnet on all configured TCPMONs). If you do not substitute the wild card (*) for the TCPMON name, it is assumed. If you omit the process or subnet name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. Examples The following command starts subnet SN1 under the specified TCPMON: -> START SUBNET $ZZTCP.#ZPTM1.SN1 The following command starts subnet SN1 under all TCPMONs: -> START SUBNET $ZZTCP.*.SN1 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 74 SCF Reference for Parallel Library TCP/IP STATS Command The following command is valid even if you have not assumed the TCPMON name: -> START SUBNET $ZZTCP.SN1 Considerations • • • • • The object-name template (wild-card notation) is supported. When you use the START command, the object must be in the STOPPED summary state. The SLSA subsystem must be operational before subnets can be started successfully. To terminate the operation of subnets, use the STOP or ABORT command. For the two subnets configured as a failover pair, you cannot start either of the subnets before the ASSOCIATESUB command is done. STATS Command The STATS command returns statistics for a specified PTCPIP object. The STATS MON command returns statistics for the TCPMONs running in every processor while the STATS PROCESS command for TCPSAM returns statistics only for the processor in which the TCPSAM process resides. All statistics are 32-bit numbers. The letter D in the display values indicates that the value is a doubleword. Whenever a RESET option is included, the counters associated with the specified objects are displayed and reset to 0, and the timestamp for the reset is recorded. Any STATS command returns the time at which the current statistics were sampled and the time at which the counters were last reset. This is a nonsensitive command except when used with the reset option. When used with the reset option, it is a sensitive command. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 75 SCF Reference for Parallel Library TCP/IP STATS MON Command for TCPMAN STATS MON Command for TCPMAN The STATS MON command displays the PTCPIP subsystems statistics for each of the protocol layers in a given TCPMON or in all configured TCPMONs. Command Syntax STATS [ / OUT file-spec / ] [MON $ZZTCP.#ZPTMn.mon-name] [ , RESET ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. MON $ZZTCP.#ZPTMn.mon-name is the name of the TCPMON. The fully-qualified TCPMON name is $ZZTCP.#ZPTM{0-F}.mon-name. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. RESET resets the statistical counters to zero. Examples The following command requests statistics about the specified TCPMON: -> STATS MON $ZZTCP.#ZPTM2 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 76 SCF Reference for Parallel Library TCP/IP STATS MON Command for TCPMAN STATS MON Display Format The format of the display for the STATS MON command is: TCPMAN Stats MON \BEAR.$ZZTCP.#ZPTM2 Sample Time ... 16 Jun 1999, 6:46:37.534 Reset Time .... 16 Jun 1999, 6:28:26.279 TCP LAYER STATS Bad Checksum..........0 Bad Offset..........0 Too Short.............0 Bad Sequence........0 Retransmitted PKTs....6 Connection Timeouts.0 Total PKTs Input......310 Total PKTs Output...218 Incoming Connections..15 Outgoing Connects...9 No Ports For PKTs.....2 Urgent PKTs Recv....0 PKTs Unacknowledged...0 Established Connect.22 Connections Dropped...2 Embryonic Conn Drop.2 Connections Closed....54 Segments RTT........51 RTT Updated...........44 Delayed ACKs Sent...8 Conn Dropped Timeouts.0 Retransmit Timeouts.18 Persist Timeouts......0 Keep-Alive Timeouts.40 KeepAlive Probes Sent.0 Keep-Alive Dropped..0 Data Packets Sent.....12 Data Bytes Sent....34105 Retransmitted Bytes...36000 ACK PKTs Sent.......56 Window Probes Sent....0 Urgent PKTs Sent....0 Win Update PKT Sent...104 Control PKTs Sent...40 Data Packets Rcvd.....226 Data Bytes Rcv....796384 Duplicate PKTs Recv...4 Duplicate Bytes Rcv.0 Partial Dup PKTs......0 Partial Dup Byte....0 Out Of Order PKTs Rcv.7 Out Of Order B Rcv.28384 PKTs Rcv After Window.0 Bytes Rcv After Win.0 PKT Rcv After Close...2 Win Probe PKTs Rcv..0 Dup ACK PKTs Recv.....2 TooMuch ACK PKT Rcv.0 ACK PKTs Recv.........37 ACK Bytes Received.32864 Win Update PKTs Rcv...3 ACK Predictions OK..2 Data Predictions OK..211 PawsDrop.............0 PCB Cache Missed......57 Persist State Drop..0 Premature ACKs........0 SYN Dropped.........0 Fast Retransmits......0 TCP LAYER / SYN ATTACK STATISTICS SYN Cache Added.......0 SYN Cache Completed.0 SYN Cache timed Out...0 SYN Cache Drop, OvF.0 SYN Cache Drop, RST...0 SYN Cache Drop, UnR.0 SYN Cache Drop, BOvF..0 SYN Cache Aborted...0 SYN Cache Duplicated..0 SYN Cache Dropped...0 UDP LAYER STATS Bad Checksum..........0 PKTS with no Chksum.0 Invalid Header Size...0 Bad Packet Size.....0 Total PKTS Input......0 Total PKTS Output...0 Input PKTs Dropped....0 Output PKTs Dropped.0 Nosock on port,Bcast..0 No sock on port.....0 Pkts, Miss pcb Cach...0 NotDeliver,Sockfull.0 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 77 STATS MON Command for TCPMAN SCF Reference for Parallel Library TCP/IP The STATS MON command display (continued) is: IP Bad Checksum..........0 Invalid Header Size...0 Fragments Input.......0 Packets Cant Forward..0 Short Packets.........0 Fragments Timed Out...0 Total Packets Input...310 Deliverd to Upper.....310 PKTs Lost, No Buffer..0 Packets Fragmented....0 Packets, Dont Fragment.0 Discarded, No Route...0 PKTs, Raw IP Generate.0 Frags, Exceed Limit...0 Bad Route Redirects..0 New Gateway Redirect.0 Unreachable..........2 Bad Checksum.........0 Invalid Header Size..0 Reflect Packets......0 Bad ICMP Code........0 In Echo Reply........0 In Dest Unreachable..0 In Source Quench.....0 In Redirect..........0 In Echo..............0 In Time Exceeded.....0 In Param Problem.....0 In Timestamp.........0 In Timestamp Reply...0 In Info Request......0 In Info Reply........0 Bad Router Adv Sub...0 Bad Router Wrds/Addr.0 Router Advertisement.0 LAYER STATS UnKnown Protocol....0 Bad Packet Size.....0 Fragments Dropped...0 ICMP Redirects Sent.0 Packets Too Small...0 Packets Forwarded...0 Total PKTS Output...222 PKTS,Generated Here.222 Packets,Reassembled.0 Out Frags Created...0 Packets, Bad Option.0 Pkts, IP Ver != 4...0 Bad Source Interface0 IP ROUTING STATS Dynamic Redirects...0 wild-Card Matches....0 ICMP LAYER STATS Errors...............0 Short IP Packets.....0 Bad ICMP Packets.....0 Packets Too Short....0 Out Echo Reply.......0 Out Dest Unreach.....0 Out Source Quench....0 Out Redirect.........0 Out Echo.............0 Out Time Exceeded....0 Out Param Problem....0 Out Timestamp........0 Out Timestamp Reply..0 Out Info Request.....0 Out Info Reply.......0 Bad Router Addr List.0 Good Routes Recorded.0 Router Solicitation..0 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 78 STATS MON Command for TCPMAN SCF Reference for Parallel Library TCP/IP The STATS MON command display (continued) is: Data MDs In Use........ Dup MDs In Use......... Dup Driver MDs In Use.. No Data MDs Avail...... MD Queue Limits........ QIO Driver Errors...... Current Pool Allocation Pool Allocation Fails.. Size Size Size Size Size Size QIO STATS 1D 0D 0D 0D 0D 0D 497992D 0D SOCKET SEND SIZE HISTOGRAM 1-128............. 2D Size 129-256........... 0D 257-512........... 0D Size 513-1024.......... 0D 1025-2048......... 0D Size 2049-4096........ 12D 4097-8192......... 0D Size 8193-12288........ 0D 12289-16384....... 0D Size 16385-32768....... 0D 32769 and larger.. 0D ARP In In In In In Maximum Data MDs Used.. 2D Maximum Dup MDs Used... 8D Max Dup Driv MDs Used.. 0D No Dup MDs Avail....... 0D QIO Limit Warnings..... 0D No Dup Driv MDs Avail.. 0D Maximum Pool Alloc 498296D ARP Requests........ ARP Replys.......... InARP Requests...... InARP Replys........ ARP Naks............ 7 0 0 0 0 Total Packets Input.... Short Packets.......... Total Queries Input.... Total Reports Input.... Reports For Groups..... 0 0 0 0 0 IGMP STATS Out Out Out Out Out ARP Requests....... ARP Replys......... InARP Requests..... InARP Replys....... ARP Naks........... 4 0 0 0 0 STATS Total Reports Sent..... Bad Checksum........... Bad Queries............ Bad Reports............ 0 0 0 0 Statistics Definitions (in Alphabetical Order) Reset Time is the time at which the counters were last initialized (set to zero). Sample Time is the time at which the statistics were sampled. Description of Statistics for the TCP Layer ACK Packets Sent is the number of ACK packets sent. Bad Checksum is the number of packets received with invalid checksum values. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 79 SCF Reference for Parallel Library TCP/IP STATS MON Command for TCPMAN Bad Offset is the number of packets received with invalid data offsets in their TCP headers. An invalid data offset usually indicates that either the sender of the packet made an internal error in generating the packet, or the receiver of the packet had a byteswapping problem. This error is rare and is usually seen only during the development of the protocol. Bad Sequence is the number of packets with bad sequence numbers. Bytes Recv After Win is the number of bytes received exceeding the window boundary. Conn Dropped Timeouts is the number of connections dropped in a transmit timeout. Connections Closed is the number of connections closed (this value includes the number of connections dropped). Connections Dropped is the number of connections dropped. Control Packets Sent is the number of SYN, FIN, and RST control packets sent. Data Bytes Received is the number of bytes received in sequence. Data Bytes Sent is the total number of data bytes sent. Data Packets Received is the number of packets received in sequence. Data Packets Sent is the total number of data packets sent. Delayed ACKs Sent is the number of delayed ACKs sent. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 80 SCF Reference for Parallel Library TCP/IP STATS MON Command for TCPMAN Duplicate Bytes Recv is the number of duplicate bytes received. Duplicate PKTs Recv is the number of duplicate packets received. Embryonic Conn Dropped is the number of embryonic connections dropped. Established Connects is the number of connections established. Incoming Connections is the number of incoming connection requests. Keep-Alive Dropped is the number of connections dropped because of keep-alive timeouts. Keep-Alive Probes Sent is the number of keep-alive probes sent. Keep-Alive Timeouts is the number of keep-alive timeouts. No Ports For Packets is the number of packets received for a connection that has been closed or does not exist. This event can be a normal occurrence or it can be caused by a faulty PTCPIP implementation that does not conform to the PTCPIP state table. Outgoing Connect is the number of connection requests sent to remote hosts. Out Of Order PKTs Rcv is the number of out-of-order packets received. Out Of Order B Rcv is the number of out-of-order bytes received. Partial Duplicate Byte is the number of duplicate bytes received in partially duplicate packets. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 81 SCF Reference for Parallel Library TCP/IP STATS MON Command for TCPMAN Partial Dup PKTs is the number of packets received with some duplicate data. PCB Cache Missed is the number of input packets missing PCB cache. Persist Timeouts is the number of persistent timeouts. PKTs Rcv After Close is the number of packets received after close. PKTs Rcv After Window is the number of packets received exceeding the window boundary. PKTs Unacknowledged is the number of unacknowledged packets. Retransmitted Bytes is the number of bytes retransmitted. Retransmitted Packets is the number of packets retransmitted. Packets are retransmitted when a packet is not acknowledged within a certain time period. Packets can be retransmitted for any of the following reasons: the network is overloaded; the other end of the connection is overloaded (so that appropriate acknowledgments cannot be received or sent); or a corrupted packet (that is, a packet with an invalid checksum) has been received. Retransmit Timeouts is the number of retransmit timeouts. RTT Updated is the number of round-trip times updated. Segments RTT is the number of segments where round-trip time was attempted. Too Short is the number of packets received that were too short. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 82 SCF Reference for Parallel Library TCP/IP STATS MON Command for TCPMAN Total PKTs Input is the number of packets received. Total PKTs Output is the number of packets sent down to the IP layer. Urgent PKTs Recv is the number of packets received with the URG bit set. Urgent PKTs Sent is the number of packets sent with the URG bit set. Window Probes Sent is the number of window probes sent. Window Update PKT Sent is the number of window update packets sent. Window Probe PKTs Recv is the number of window-probes packets received. Description of TCP Layer / SYN Attack Statistics (in Displayed Order) SYN Cache Added is the number of SYN cache entries added. SYN CACHE COMPLETED is the number of SYN cache connections completed. SYN CACHE TIMED OUT is the number of SYN cache entries timed out. SYN CACHE DROP, OvF is the number of SYN cache entries dropped due to overflow. SYN CACHE DROP, RST is the number of SYN cache entries dropped due to RST. SYN CACHE DROP, UnR is the number of SYN cache entries dropped due to ICMP unreachable. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 83 SCF Reference for Parallel Library TCP/IP STATS MON Command for TCPMAN SYN CACHE DROP, BOvF is the number of SYN cache entries dropped due to bucket overflow. SYN CACHE ABORTED is the number of SYN cache aborted (no memory). SYN CACHE DUPLICATED is the number of duplicated SYNs received. SYN CACHE DROPPED is the number of SYNs dropped (no route/mem). Description of Statistics for the UDP Layer (in Alphabetical Order) Bad Checksum is the number of packets received with invalid checksum values. An invalid checksum is usually caused by a noisy link. Bad Packet Size is the number of packets received that contain either more or less data than has been specified in their headers. This error indicates the sender has a protocol error or that the receiver has a byte-ordering problem. Input PKTs Dropped is the number of packets not forwarded to socket applications because of receive socket space being full. Invalid Header Size is the number of packets received with invalid header size. This error indicates a problem between IP and UDP. No sock on port is the number of sockets with no port. Nosock on port, Bdcst is the number of sockets with no port which arrived as broadcast. NotDeliver,Sockfull is the number of packets not delivered, input socket full. Output PKTs Dropped is the number of packets not sent because of interface problems. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 84 SCF Reference for Parallel Library TCP/IP STATS MON Command for TCPMAN Pkts, Miss pcb Cach is the number of input packets missing pcb cache. PKTS with no Chksum the number of packets received with no checksums. Total PKTs Input is the number of packets received. Total PKTs Output is the number of packets sent to the IP layer. Description of Statistics for the IP Layer (in Alphabetical Order) Bad Checksum is the number of packets received with invalid checksum values. An invalid checksum is usually caused by a noisy link. Bad Packet Size is the number of packets received with a packet length shorter than expected. This error is very similar to the Invalid Header Size and is usually caused by similar conditions. Bad Src Interface is the number of packets with incorrect source interface or no route. Delivered to Upper is the number of IP packets delivered to upper level. Discarded, No Route is the number of packets discarded due to no route. Fragments Dropped is the number of packet fragments dropped. A fragment is dropped either when memory cannot be allocated for the fragment or when the fragment is a duplicate of a fragment that has already been received. Fragments Input is the number of packet fragments received. Usually, a packet is fragmented when it is too large for a particular gateway or network. This statistic might indicate that the sender's maximum segment size is too large for the connection. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 85 SCF Reference for Parallel Library TCP/IP STATS MON Command for TCPMAN Fragments Timed Out is the number of packet fragments received that timed out before the whole packet was received. This is usually caused by congestion, noisy links, or some event that prevents one of the fragments from being received with the rest. Frags, Exceed Limit is the number of fragments that exceeded the limit. ICMP Redirects Sent is the number of ICMP Redirect messages sent. Redirect messages are sent to the source host to indicate that there is a shorter path to the destination. The source host should send the packet directly to the destination host or to another gateway. Invalid Header Size is the number of packets received with a header size that is larger than the header length provided in the packet. This error indicates a problem with the sender of the packet or a problem in reading the data from the link controller to IP. Out Frags Created is the number of output fragments created. Packets, Bad Option the number of error in IP option processing. Packets Cant Forward is the number of packets destined for another host that were received but could not be forwarded. The packets could not be forwarded because either the local host is not configured as a gateway or no route is available to the specified destination. Packets, Dont Fragment is the number of packets with don't fragment flag set. Packets Forwarded is the number of packets destined for another host that were forwarded. Packets Fragmented is the number of datagrams successfully fragmented. Packets,reassembled shows the total number of IP packets successfully reassembled. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 86 SCF Reference for Parallel Library TCP/IP STATS MON Command for TCPMAN Packets Too Small is the number of packets that contained less data than was expected when the packet was read into the local buffers. This error usually indicates a problem with the local machine's buffering scheme. Pkts, Ip Ver != 4 is the number of packets with IP version not equal to 4. PKTs, Generated Here is the total number of IP packets generated here. PKTs Lost, No Buffer is the number of IP packets lost here due to no buffer. PKTs, Raw IP Generate is the total number of raw IP packets generated. Short Packets is the number of packets that contained less data than specified in their header. This can be caused by noisy links, a protocol error by the sender of the packet, or a byte-swapping problem on the receiver. Total Packets Input is the number of packets received. Total Packets Output is the number of packets sent to the IP layer. Un Known/Supp Proto is the number of packets with either an unknown or unsupported protocol specified. Description of Statistics for IP Routing (in Alphabetical Order) Bad Route Redirects is the number of Redirect messages received. Dynamic Redirects is the number of dynamic route messages received. These messages indicate where the Parallel Library TCP/IP subsystem should route messages for a specific destination. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 87 SCF Reference for Parallel Library TCP/IP STATS MON Command for TCPMAN New Gateway Redirect is the number of messages received that established a route for a new or an unknown gateway. Unreachable is the number of messages received that indicated that the specified destination was unreachable. wild-Card Matches is the number of wild-card matches found when zeros were given in the destination Internet address for a route. Description of Statistics for the ICMP Layer (in Alphabetical Order) Bad Checksum is the number of packets received with invalid checksum values. An invalid checksum is usually caused by a noisy link. Bad ICMP Code is the number of packets received that contain invalid ICMP packet-type codes in the header. The Parallel Library TCP/IP subsystem supports the following ICMP packet types and packet-type code: Echo Reply (0) Destination Unreachable (3) Source Quench (4) Redirect (5) Echo (8) Time Exceeded (11) Parameter Problem (12) Timestamp (13) Timestamp Reply (14) Information Request (15) Information Reply (1 For more detailed descriptions of these packet types, refer to the descriptions of the individual packet types below. Bad ICMP Packets is the number of invalid ICMP packets received. Bad Router ADDR List is the number of IRDP messages with a bad address list. Bad Router ADV Subcode is the number of IRDP messages with a bad ICMP subcode. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 88 SCF Reference for Parallel Library TCP/IP STATS MON Command for TCPMAN Bad Router Words/ADDR is the number of IRDP messages with an incorrect address length. Errors is the number of times an ICMP error was generated. Note that Redirect messages are not included in the total. ICMP errors can be caused by any of the following reasons: invalid IP options, problems in IP packet forwarding, or a UDP server crash. Good Routes Recorded is the number of valid routes discovered by IRDP messages that have been entered in the PTCPIP route table. In Dest Unreachable is the number of Destination Unreachable (type 3) messages received. A Destination Unreachable message is sent to the Parallel Library TCP/IP subsystem when another host or gateway determines that a destination host or port is unreachable. This message can be caused by the following reasons: either there is no route to the destination or the route to the destination has gone down; a nonexistent address has been specified; the process listening on the port has gone down; the destination host has crashed; or fragmentation is needed but the Don't Fragment flag is set. In Echo is the number of Echo (type 8) messages received. The Echo message is sent from the source address to the destination address. An Echo Reply message containing the same data is expected from the destination address. In Echo Reply is the number of Echo Reply (type 0) messages received. This ICMP message is the reply to the Echo (type 8) message. Essentially, an Echo Reply message is just the original Echo message with the type changed from 8 to 0 and the destination and source addresses reversed; the data returned in the Echo Reply message is the same as that sent in the Echo message. The receipt of an Echo Reply message informs the local host that the remote host is still alive. The data returned also gives the local host a means of testing the integrity of the link. In Info Reply is the number of Information Reply (type 16) messages received. A host or gateway sends this message—with the source and destination addresses fully specified—in reply to an Information Request message. Note that the Information Request/Reply facility, although supported, is rarely used. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 89 SCF Reference for Parallel Library TCP/IP STATS MON Command for TCPMAN In Info Request is the number of Information Request (type 15) messages received. A host or gateway can send this message—with the network portion of the source address and the destination address set to 0—to determine the number of the network on which it is running. Any host on the network can respond to this request with an Information Reply message. In Param Problem is the number of Parameter Problem (type 12) messages received. A host or gateway sends this message to notify the Parallel Library TCP/IP subsystem (functioning as a source host) that one of its datagrams has been discarded because the header parameters are incorrect. In Redirect is the number of Redirect (type 5) messages received. A gateway sends this message to the Parallel Library TCP/IP subsystem (functioning as a source host) to indicate that there is a shorter path to the destination through another gateway. When the Parallel Library TCP/IP subsystem receives a Redirect message, it corrects its routing table to reflect the new route. If a host receives many Redirect messages in a short period of time, it is usually an indication that the host is not correcting its routing table. When the Parallel Library TCP/IP subsystem services the In Redirect messages, it adds a dynamic route entry of the name #DYRTn. This dynamic route is used in lieu of the previous route which has been redirected. In Source Quench is the number of Source Quench (type 4) messages received. A gateway sends this message to the Parallel Library TCP/IP subsystem to indicate that the gateway is receiving datagrams more quickly than it can process them. When the Parallel Library TCP/IP subsystem receives this message, it reduces the rate at which it is sending datagrams by implementing a slow start. To implement a slow start, the Parallel Library TCP/IP subsystem first stops sending datagrams, then restarts sending them, and gradually increases the number of datagrams sent. If the Parallel Library TCP/IP subsystem is doing a lot of retransmissions, you should check to see if Source Quench messages are being received. If they are, you should reduce the number of packets being transmitted by your applications. In Time Exceeded is the number of Time Exceeded (type 11) messages received. A gateway sends this message to notify the Parallel Library TCP/IP subsystem (functioning as a source host) that the time-to-live field is 0 and that the gateway discarded the HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 90 SCF Reference for Parallel Library TCP/IP STATS MON Command for TCPMAN datagram. A destination host sends this message if the host cannot reassemble a fragmented datagram within the time limit because fragments are missing. The destination host then discards the datagram. When a Time Exceeded message is received, you should check for routing loops. In Timestamp is the number of Timestamp (type 13) messages received. A host or gateway sends this message to indicate the last time it handled the message before sending it. In Timestamp Reply is the number of Timestamp Reply (type 14) messages received. A host or gateway sends this message in reply to a Timestamp message. This message indicates the time in the original Timestamp message and the time at which the Timestamp message was received by the destination. The Timestamp facility is used to obtain the network time. Special applications can be written to use this facility. Invalid Header Size is the number of packets received with a length that is shorter than the length specified in the header. This error, usually caused by a noisy link, is rarely reported because the checksum routine also detects this problem. Packets Too Short is the number of packets received that were shorter than the minimum length allowed for an ICMP packet. Short packets are usually caused by a noisy link. Reflect Packets is the number of ICMP packets received that have been sent a response. Note that not all ICMP packets require a response. Short IP Packets is the number of packets received that were too short. Out Dest Unreachable is the number of Destination Unreachable messages sent. Out Echo is the number of Echo messages sent. Out Echo Reply is the number of Echo Reply messages sent. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 91 SCF Reference for Parallel Library TCP/IP STATS MON Command for TCPMAN Out Info Reply is the number of Information Reply messages sent. Out Info Request is the number of Information Request messages sent. Out Param Problem is the number of Parameter Problem messages sent. Out Redirect is the number of Redirect messages sent. Out Source Quench is the number of Source Quench messages sent. Out Time Exceeded is the number of Time Exceeded messages sent. Out Timestamp is the number of Timestamp messages sent. Out Timestamp Reply is the number of Timestamp Reply messages sent. Router Advertisement is the number of IRDP discovery messages detected by the Parallel Library TCP/IP subsystem. The Parallel Library TCP/IP subsystem either records these routes or ignore them, depending on how IRDP is configured and according to route preference. Router Solicitation is the number of IRDP solicitation messages sent by the Parallel Library TCP/IP subsystem. Description of Statistics for QIO (in Alphabetical Order) Current MBUFs Used is the current number of MBUFs in use. Current Pool Allocation is the current number of bytes of pool space in use. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 92 SCF Reference for Parallel Library TCP/IP STATS MON Command for TCPMAN Data MDs In Use is the current number of data MDs in use by the process. Dup Driver MDs In Use is the current number of duplicate MDs assigned to inbound driver MDs in use by the process. Dup MDs in Use is the current number of duplicate MDs not assigned to inbound driver MDs in use by the process. Maximum Data MDs Used is the maximum number of data MDs that have been in use. Maximum Dup MDs Used is the maximum number of duplicate MDs not assigned to inbound driver MDs that have been in use by the process. Max Dup Driv MDs Used is the maximum number of duplicate MDs assigned to inbound driver MDs in use by the process. Maximum MBUFs Used is the maximum number of MBUFs to be used. Maximum Pool Allocation is the maximum pool space used. MBUF Allocation Fails is the number of times an MBUF was not available. MD Queue Limits is the number of times the send or receive queue on a TCP session exceeded a predefined limit of MDs queued. The process attempts to decrease the number queued by collapsing the data into a smaller number of MDs. No Data MDs Avail is the number of times the process failed to obtain a data MD. No Dup Driv MDs Avail is the number of times the process failed to obtain a duplicate MD for a driver inbound MD. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 93 SCF Reference for Parallel Library TCP/IP STATS MON Command for TCPMAN No Dup MDs Avail is the number of times the process failed to obtain a duplicate MD. Pool Allocation Fails is the number of times a pool space request failed. QIO Driver Errors is the number of times the QIO driver returned an error. QIO Limit Warnings is the number of times the process received an event signifying a pool or an MD shortage from the QIO monitor. Total MBUFs Allocated is the current number of MBUFs allocated. Description of Statistics for Socket Send Size Histogram (in Displayed Order) Size 1-128 is the count of socket sends between 1 and 128 bytes. Size 129-256 is the count of socket sends between 129 and 256 bytes. Size 257-512 is the count of socket sends between 257 and 512 bytes. Size 513-1024 is the count of socket sends between 513 and 1024 bytes. Size 1025-2048 is the count of socket sends between 1025 and 2048 bytes. Size 2049-4096 is the count of socket sends between 2049 and 4096 bytes. Size 4097-8192 is the count of socket sends between 4097 and 8192 bytes. Size 8193-12288 is the count of socket sends between 8193 and 12288 bytes. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 94 SCF Reference for Parallel Library TCP/IP STATS MON Command for TCPMAN Size 12289-16384 is the count of socket sends between 12289 and 16384 bytes. Size 16385-32768 is the count of socket sends between 16385 and 32768 bytes. Description of Statistics for the ARP STATS (in Displayed Order) In ARP Requests is the number of ARP requests received. Out ARP Requests is the number of ARP requests sent. In ARP Replys is the number of ARP replies received. Out ARP Replys is the number of ARP replies sent. In InARP Requests is the number of inverse ARP requests received. Out InARP Requests is the number of inverse ARP requests sent. In InARP Replys is the number of inverse ARP replies received. Out InARP Replys is the number of inverse ARP replies sent. In ARP Naks is the number of ARP Naks received. Out ARP Naks is the number of ARP Naks sent. Description of Statistics for IGMP Statistics (in Displayed Order) Total Packets Input is the total number of IGMP packets received. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 95 SCF Reference for Parallel Library TCP/IP STATS PROCESS Command for TCPSAM Total Reports Sent is the total number of IGMP report packets sent by this process. Short Packets is the total number of IGMP packets received that were too short. Bad Checksum is the total number of IGMP packets received that had an incorrect checksum. Total Queries Input is the total number of IGMP query packets received. Bad Queries is the total number of IGMP query packets received with the IP destination address not equal to the all hosts group. Total Reports Input is the total number of IGMP membership reports received. Bad Reports is the total number of bad IGMP membership reports received. Reports For Our Groups is the total number of IGMP membership reports received for groups we belong to. STATS PROCESS Command for TCPSAM The STATS PROCESS command for TCPSAM displays the statistics for the TCPSAM process running in processor that contains the TCPSAM process. Command Syntax STATS [ / OUT file-spec / ] [PROCESS $tcpsam-name] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 96 SCF Reference for Parallel Library TCP/IP STATS PROCESS Command for TCPSAM PROCESS $tcpsam-name is the name of the TCPSAM process. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. Examples The following command requests statistics about the specified TCPMON: -> STATS MON $ZSAM1 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 97 STATS PROCESS Command for TCPSAM SCF Reference for Parallel Library TCP/IP STATS PROCESS Display Format The format of the display for the STATS PROCESS command is: TCPIP Stats PROCESS \SYSA.$ZSAM1 Sample Time ... 17 Oct 1999, 17:17:41.169 Reset Time .... Invalid date/time Bad Checksum.......... Invalid Header Size... Retransmitted Packets. Total Packets Input... Incoming Connections.. No Ports For Packets.. Packets Unacknowledged Connections Dropped... Connections Closed.... RTT Updated........... Conn Dropped Timeouts. Persist Timeouts...... Keep-Alive Probes Sent Data Packets Sent..... Retransmitted Bytes... Window Probes Sent.... Window Update PKT Sent Data Packets Received. Duplicate PKTs Recv... Partial Duplicate PKTs Out Of Order PKTs Recv PKTs Recv After Window PKTs Recv After Close. Duplicate ACKs Recv... ACK Packets Received.. Window Update PKTs.... Data Predictions OK... 0D 0D 608D 103579D 804D 9D 0D 0D 0D 0D 0D 0D 0D 788D 17890D 0D 0D 344D 0D 0D 0D 0D 0D 0D 0D 0D 501D Bad Checksum.......... Invalid Header Size... Total Packets Input... Input Packets Dropped 0D 0D 0D OD Bad Checksum.......... Invalid Header Size... Fragments Input....... Packets Cant Forward.. Short Packets......... Fragments Timed Out... Total Packets Input... 0D 0D 0D 3D 0D 0D 12327D TCP LAYER STATS Bad Offset............ Bad Segment Size...... Connection Timeouts... Total Packets Output.. Outgoing Connections.. Urgent Packets Recv... Established Connects.. Embryonic Conn Dropped Segments RTT.......... Delayed ACKs Sent..... Retransmit Timeouts... Keep-Alive Timeouts... Keep-Alive Dropped.... Data Bytes Sent....... ACK Packets Sent...... Urgent Packets Sent... Control Packets Sent.. Data Bytes Received... Duplicate Bytes Recv.. Partial Duplicate Byte Out Of Order Byte Recv Bytes Recv After Win.. Window Probe PKTs Recv Too Much ACK Received. ACK Bytes Received.... ACK Predictions OK.... 3D 0D 129D 9847D 1096D 3D 1D 0D 0D 0D 0D 0D 0D 67897D 1298D 0D 0D 34489D 0D 0D 0D 0D 0D 0D 0D 110D UDP LAYER STATS Bad Route Redirects... 0D New Gateway Redirects. 0D Unreachable........... 10D Bad Packet Size....... 0D Total Packets Output.. 0D Output Packets Dropped OD IP LAYER STATS Bad Packet Size....... 0D Fragments Dropped..... 0D ICMP Redirect Sent.... 0D Packets Too Small..... 0D Packets Forwarded..... 0D Total Packets Output.. 8129D IP ROUTING STATS Dynamic Redirects..... 0D wild-Card Matches...... 0D HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 98 STATS PROCESS Command for TCPSAM SCF Reference for Parallel Library TCP/IP The STATS PROCESS command display (continued): Bad Checksum.......... Invalid Header Size... Reflect Packets....... Bad ICMP Code......... In Echo Reply......... In Dest Unreachable... In Source Quench...... In Redirect........... In Echo............... In Time Exceeded...... In Parameter Problem.. In Timestamp.......... In Timestamp Reply.... In Info Request....... In Info Reply......... Bad Router Adv Subcode Bad Router Words/Addr. Router Advertisement.. 0D 0D 4D 1D 0D 643D 3D 0D 0D 1D 0D 0D 0D 0D 0D 0D 0D 0D ICMP LAYER STATS Errors................ Short IP Packets...... Bad ICMP Packets...... Packets Too Short..... Out Echo Reply........ Out Dest Unreachable.. Out Source Quench..... Out Redirect.......... Out Echo.............. Out Time Exceeded..... Out Parameter Problem. Out Timestamp......... Out Timestamp Reply... Out Info Request...... Out Info Reply........ Bad Router Addr List.. Good Routes Recorded.. Router Solicitation... Data MDs In Use........ Dup MDs In Use......... Dup Driver MDs In Use.. No Data MDs Avail...... MD Queue Limits........ QIO Driver Errors...... Current Pool Allocation Pool Allocation Fails.. Total MBUFs Allocated.. Maximum MBUFs Used..... QIO STATS 0D Maximum Data MDs Used. 0D Maximum Dup MDs Used.. 0D Max Dup Driv MDs Used. 0D No Dup MDs Avail...... 0D QIO Limit Warnings.... 0D No Dup Driv MDs Avail. 452156D Maximum Pool Allocation 0D 1008D Current MBUFs Used.... 14D MBUF Allocation Fails. Size Size Size Size Size 1-128............. 257-512........... 1025-2048......... 4097-8192......... 12289-16384....... 7D 0D 0D 0D 0D ARP Requests........ ARP Replys.......... InARP Requests...... InARP Replys........ ARP Naks............ 0D 4D 2D 1D 1D Total Packets Input.... Short Packets.......... Total Queries Input.... Total Reports Input.... Reports For Our Groups. 0D 0D 0D 0D 0D SOCKET SEND SIZE HISTOGRAM Size 129-256........... Size 513-1024.......... Size 2049-4096......... Size 8193-12288........ Size 16385-32768....... 4248D 0D 0D 0D 4D 4248D 0D 0D 0D 0D 0D 0D 0D 0D 0D 0D 0D 0D 1D 0D 0D 0D 0D 0D 452156D 8D 0D 0D 0D 0D 0D 0D ARP STATS In In In In In Out Out Out Out Out ARP Requests....... ARP Replys......... InARP Requests..... InARP Replys....... ARP Naks........... 5D 0D 2D 2D 0D Total Reports Sent.... Bad Checksum.......... Bad Queries........... Bad Reports........... 0D 0D 0D 0D IGMP STATS Statistics Definitions (in Alphabetical Order) Reset Time shows invalid date and time because the RESET option is not supported for the TCPSAM process. Sample Time is the time at which the statistics were sampled. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5- 99 SCF Reference for Parallel Library TCP/IP STATS PROCESS Command for TCPSAM Description of Statistics for the TCP Layer (in Alphabetical Order) ACK Bytes Received is the number of ACK bytes acknowledged by received ACKs. ACK Packets Received is the number of ACK packets received. ACK Packets Sent is the number of ACK packets sent. ACK Predictions OK is the number of times the header predictions were correct for ACKs. Bad Checksum is the number of packets received with invalid checksum values. Bad Offset is the number of packets received with invalid data offsets in their TCP headers. An invalid data offset usually indicates that either the sender of the packet made an internal error in generating the packet, or the receiver of the packet had a byteswapping problem. This error is rare and is usually seen only during the development of the protocol. Bad Segment Size is the number of packets received with invalid segment sizes. Bytes Recv After Win is the number of bytes received exceeding the window boundary. Conn Dropped Timeouts is the number of connections dropped in a transmit timeout. Connection Timeouts is the number of connections (including partial connections) that timed out. A connection timeout is recorded each time the keep-alive timer or retransmission timer expires. The keep-alive timer expires when the connection is inactive for a certain period of time. The inactivity can be caused by a lost connection or by network congestion. The retransmission timer expires when a packet is not acknowledged within a certain time. Packet retransmission can be caused by any of the following conditions: the network is overloaded; the other end of the connection is overloaded (so that HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -100 SCF Reference for Parallel Library TCP/IP STATS PROCESS Command for TCPSAM appropriate acknowledgments cannot be received and/or sent); or a corrupted packet (that is, a packet with an invalid checksum) is received. Connections Closed is the number of connections closed (this value includes the number of connections dropped). Connections Dropped is the number of connections dropped. Control Packets Sent is the number of SYN, FIN, and RST control packets sent. Data Bytes Received is the number of bytes received in sequence. Data Bytes Sent is the total number of data bytes sent. Data Packets Received is the number of packets received in sequence. Data Packets Sent is the total number of data packets sent. Data Predictions OK is the number of times the header predictions were correct for data packets. Delayed ACKs Sent is the number of delayed ACKs sent. Duplicate ACKs Recv is the number of duplicate ACK packets received. Duplicate Bytes Recv is the number of duplicate bytes received. Duplicate PKTs Recv is the number of duplicate packets received. Embryonic Conn Dropped is the number of embryonic connections dropped. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -101 SCF Reference for Parallel Library TCP/IP STATS PROCESS Command for TCPSAM Established Connects is the number of connections established. Incoming Connections is the number of incoming connection requests. Invalid Header Size is the number of packets received with an invalid header size. This error usually indicates a problem between IP and TCP. Keep-Alive Dropped is the number of connections dropped because of keep-alive timeouts. Keep-Alive Probes Sent is the number of keep-alive probes sent. Keep-Alive Timeouts is the number of keep-alive timeouts. No Ports For Packets is the number of packets received for a connection that has been closed or does not exist. This event can be a normal occurrence or it can be caused by a faulty TCP/IP implementation that does not conform to the TCP/IP state table. Outgoing Connections is the number of connection requests sent to remote hosts. Out Of Order PKTs Recv is the number of out-of-order packets received. Out Of Order Byte Recv is the number of out-of-order bytes received. Packets Unacknowledged is the number of unacknowledged packets. Partial Duplicate Byte is the number of duplicate bytes received in partially duplicate packets. Partial Duplicate PKTs is the number of packets received with some duplicate data. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -102 SCF Reference for Parallel Library TCP/IP STATS PROCESS Command for TCPSAM Persist Timeouts is the number of persistent timeouts. PKTs Recv After Close is the number of packets received after close. PKTs Recv After Window is the number of packets received exceeding the window boundary. Retransmitted Bytes is the number of bytes retransmitted. Retransmitted Packets is the number of packets retransmitted. Packets are retransmitted when a packet is not acknowledged within a certain time period. Packets can be retransmitted for any of the following reasons: the network is overloaded; the other end of the connection is overloaded (so that appropriate acknowledgments cannot be received or sent); or a corrupted packet (that is, a packet with an invalid checksum) has been received. Retransmit Timeouts is the number of retransmit timeouts. RTT Updated is the number of round-trip times updated. Segments RTT is the number of segments where round-trip time was attempted. Too Much ACK Received is the number of ACK packets received for unsent data. Total Packets Input is the number of packets received. Total Packets Output is the number of packets sent down to the IP layer. Urgent Packets Recv is the number of packets received with the URG bit set. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -103 SCF Reference for Parallel Library TCP/IP STATS PROCESS Command for TCPSAM Urgent Packets Sent is the number of packets sent with the URG bit set. Window Probes Sent is the number of window probes sent. Window Update PKT Sent is the number of window update packets sent. Window Probe PKTs Recv is the number of window-probes packets received. Window Update Pkts is the number of window update packets received. Description of Statistics for the UDP Layer (in Alphabetical Order) Bad Checksum is the number of packets received with invalid checksum values. An invalid checksum is usually caused by a noisy link. Bad Packet Size is the number of packets received that contain either more or less data than has been specified in their headers. This error indicates the sender has a protocol error or that the receiver has a byte-ordering problem. Input Packets Dropped is the number of packets not forwarded to socket applications because of receive socket space being full. Invalid Header Size is the number of packets received with invalid header size. This error indicates a problem between IP and UDP. Output Packets Dropped is the number of packets not sent because of interface problems. Total Packets Input is the number of packets received. Total Packets Output is the number of packets sent to the IP layer. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -104 SCF Reference for Parallel Library TCP/IP STATS PROCESS Command for TCPSAM Description of Statistics for the IP Layer (in Alphabetical Order) Bad Checksum is the number of packets received with invalid checksum values. An invalid checksum is usually caused by a noisy link. Bad Packet Size is the number of packets received with a packet length shorter than expected. This error is very similar to the Invalid Header Size and is usually caused by similar conditions. Fragments Dropped is the number of packet fragments dropped. A fragment is dropped either when memory cannot be allocated for the fragment or when the fragment is a duplicate of a fragment that has already been received. Fragments Input is the number of packet fragments received. Usually, a packet is fragmented when it is too large for a particular gateway or network. This statistic might indicate that the sender's maximum segment size is too large for the connection. Fragments Timed Out is the number of packet fragments received that timed out before the whole packet was received. This is usually caused by congestion, noisy links, or some event that prevents one of the fragments from being received with the rest. ICMP Redirects Sent is the number of ICMP Redirect messages sent. Redirect messages are sent to the source host to indicate that there is a shorter path to the destination. The source host should send the packet directly to the destination host or to another gateway. Invalid Header Size is the number of packets received with a header size that is larger than the header length provided in the packet. This error indicates a problem with the sender of the packet or a problem in reading the data from the link controller to IP. Packets Cant Forward is the number of packets destined for another host that were received but could not be forwarded. The packets could not be forwarded because either the local host is not configured as a gateway or no route is available to the specified destination. Packets Forwarded is the number of packets destined for another host that were forwarded. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -105 SCF Reference for Parallel Library TCP/IP STATS PROCESS Command for TCPSAM Packets Too Small is the number of packets that contained less data than was expected when the packet was read into the local buffers. This error usually indicates a problem with the local machine's buffering scheme. Short Packets is the number of packets that contained less data than specified in their header. This can be caused by noisy links, a protocol error by the sender of the packet, or a byte-swapping problem on the receiver. Total Packets Input is the number of packets received. Total Packets Output is the number of packets sent to the IP layer. Description of Statistics for IP Routing Bad Route Redirects is the number of Redirect messages received. Dynamic Redirects is the number of dynamic route messages received. These messages indicate where the NonStop TCP/IP subsystem should route messages for a specific destination. New Gateway Redirects is the number of messages received that established a route for a new or an unknown gateway. Unreachable is the number of messages received that indicated that the specified destination was unreachable. wild-Card Matches is the number of wild-card matches found when zeros were given in the destination Internet address for a route. Description of Statistics for the ICMP Layer (in Alphabetical Order) Bad Checksum is the number of packets received with invalid checksum values. An invalid checksum is usually caused by a noisy link. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -106 SCF Reference for Parallel Library TCP/IP STATS PROCESS Command for TCPSAM Bad ICMP Code is the number of packets received that contain invalid ICMP packet-type codes in the header. The NonStop TCP/IP subsystem supports the following ICMP packet types and packet-type code: Echo Reply (0) Destination Unreachable (3) Source Quench (4) Redirect (5) Echo (8) Time Exceeded (11) Parameter Problem (12) Timestamp (13) Timestamp Reply (14) Information Request (15) Information Reply (1 For more detailed descriptions of these packet types, refer to the descriptions of the individual packet types below. Bad ICMP Packets is the number of invalid ICMP packets received. Bad Router Addr List is the number of IRDP messages with a bad address list. Bad Router ADV Subcode is the number of IRDP messages with a bad ICMP subcode. Bad Router Words/Addr is the number of IRDP messages with an incorrect address length. Errors is the number of times an ICMP error was generated. Note that Redirect messages are not included in the total. ICMP errors can be caused by any of the following reasons: invalid IP options, problems in IP packet forwarding, or a UDP server crash. Good Routes Recorded is the number of valid routes discovered by IRDP messages that have been entered in the TCP/IP route table. In Dest Unreachable is the number of Destination Unreachable (type 3) messages received. A Destination Unreachable message is sent to the NonStop TCP/IP subsystem when another host or gateway determines that a destination host or port is HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -107 SCF Reference for Parallel Library TCP/IP STATS PROCESS Command for TCPSAM unreachable. This message can be caused by the following reasons: either there is no route to the destination or the route to the destination has gone down; a nonexistent address has been specified; the process listening on the port has gone down; the destination host has crashed; or fragmentation is needed but the Don't Fragment flag is set. In Echo is the number of Echo (type 8) messages received. The Echo message is sent from the source address to the destination address. An Echo Reply message containing the same data is expected from the destination address. In Echo Reply is the number of Echo Reply (type 0) messages received. This ICMP message is the reply to the Echo (type 8) message. Essentially, an Echo Reply message is just the original Echo message with the type changed from 8 to 0 and the destination and source addresses reversed; the data returned in the Echo Reply message is the same as that sent in the Echo message. The receipt of an Echo Reply message informs the local host that the remote host is still alive. The data returned also gives the local host a means of testing the integrity of the link. In Info Reply is the number of Information Reply (type 16) messages received. A host or gateway sends this message—with the source and destination addresses fully specified—in reply to an Information Request message. Note that the Information Request/Reply facility, although supported, is rarely used. In Info Request is the number of Information Request (type 15) messages received. A host or gateway can send this message—with the network portion of the source address and the destination address set to 0—to determine the number of the network on which it is running. Any host on the network can respond to this request with an Information Reply message. In Parameter Problem is the number of Parameter Problem (type 12) messages received. A host or gateway sends this message to notify the NonStop TCP/IP subsystem (functioning as a source host) that one of its datagrams has been discarded because the header parameters are incorrect. In Redirect is the number of Redirect (type 5) messages received. A gateway sends this message to the NonStop TCP/IP subsystem (functioning as a source host) to indicate that there is a shorter path to the destination through another gateway. When the NonStop TCP/IP subsystem receives a Redirect message, it corrects its HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -108 SCF Reference for Parallel Library TCP/IP STATS PROCESS Command for TCPSAM routing table to reflect the new route. If a host receives many Redirect messages in a short period of time, it is usually an indication that the host is not correcting its routing table. When the NonStop TCP/IP subsystem services the In Redirect messages, it adds a dynamic route entry of the name #DYRTn. This dynamic route is used in lieu of the previous route which has been redirected. In Source Quench is the number of Source Quench (type 4) messages received. A gateway sends this message to the NonStop TCP/IP subsystem to indicate that the gateway is receiving datagrams more quickly than it can process them. When the NonStop TCP/IP subsystem receives this message, it reduces the rate at which it is sending datagrams by implementing a slow start. To implement a slow start, the NonStop TCP/IP subsystem first stops sending datagrams, then restarts sending them, and gradually increases the number of datagrams sent. If the NonStop TCP/IP subsystem is doing a lot of retransmissions, you should check to see if Source Quench messages are being received. If they are, you should reduce the number of packets being transmitted by your applications. In Time Exceeded is the number of Time Exceeded (type 11) messages received. A gateway sends this message to notify the NonStop TCP/IP subsystem (functioning as a source host) that the time-to-live field is 0 and that the gateway discarded the datagram. A destination host sends this message if the host cannot reassemble a fragmented datagram within the time limit because fragments are missing. The destination host then discards the datagram. When a Time Exceeded message is received, you should check for routing loops. In Timestamp is the number of Timestamp (type 13) messages received. A host or gateway sends this message to indicate the last time it handled the message before sending it. In Timestamp Reply is the number of Timestamp Reply (type 14) messages received. A host or gateway sends this message in reply to a Timestamp message. This message indicates the time in the original Timestamp message and the time at which the Timestamp message was received by the destination. The Timestamp facility is used to obtain the network time. Special applications can be written to use this facility. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -109 SCF Reference for Parallel Library TCP/IP STATS PROCESS Command for TCPSAM Invalid Header Size is the number of packets received with a length that is shorter than the length specified in the header. This error, usually caused by a noisy link, is rarely reported because the checksum routine also detects this problem. Packets Too Short is the number of packets received that were shorter than the minimum length allowed for an ICMP packet. Short packets are usually caused by a noisy link. Reflect Packets is the number of ICMP packets received that have been sent a response. Note that not all ICMP packets require a response. Short IP Packets is the number of packets received that were too short. Out Dest Unreachable is the number of Destination Unreachable messages sent. Out Echo is the number of Echo messages sent. Out Echo Reply is the number of Echo Reply messages sent. Out Info Reply is the number of Information Reply messages sent. Out Info Request is the number of Information Request messages sent. Out Parameter Problem is the number of Parameter Problem messages sent. Out Redirect is the number of Redirect messages sent. Out Source Quench is the number of Source Quench messages sent. Out Time Exceeded is the number of Time Exceeded messages sent. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -110 SCF Reference for Parallel Library TCP/IP STATS PROCESS Command for TCPSAM Out Timestamp is the number of Timestamp messages sent. Out Timestamp Reply is the number of Timestamp Reply messages sent. Router Advertisement is the number of IRDP discovery messages detected by the NonStop TCP/IP subsystem. The NonStop TCP/IP subsystem either records these routes or ignore them, depending on how IRDP is configured and according to route preference. Router Solicitation is the number of IRDP solicitation messages sent by the NonStop TCP/IP subsystem. Description of Statistics for QIO Current MBUFs Used is the current number of MBUFs in use. Current Pool Allocation is the current number of bytes of pool space in use. Data MDs In Use is the current number of data MDs in use by the process. Dup Driver MDs In Use is the current number of duplicate MDs assigned to inbound driver MDs in use by the process. Dup MDs in Use is the current number of duplicate MDs not assigned to inbound driver MDs in use by the process. Maximum Data MDs Used is the maximum number of data MDs that have been in use. Maximum Dup MDs Used is the maximum number of duplicate MDs not assigned to inbound driver MDs that have been in use by the process. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -111 SCF Reference for Parallel Library TCP/IP STATS PROCESS Command for TCPSAM Max Dup Driv MDs Used is the maximum number of duplicate MDs assigned to inbound driver MDs in use by the process. Maximum MBUFs Used is the maximum number of MBUFs to be used. Maximum Pool Allocation is the maximum pool space used. MBUF Allocation Fails is the number of times an MBUF was not available. MD Queue Limits is the number of times the send or receive queue on a TCP session exceeded a predefined limit of MDs queued. The process attempts to decrease the number queued by collapsing the data into a smaller number of MDs. No Data MDs Avail is the number of times the process failed to obtain a data MD. No Dup Driv MDs Avail is the number of times the process failed to obtain a duplicate MD for a driver inbound MD. No Dup MDs Avail is the number of times the process failed to obtain a duplicate MD. Pool Allocation Fails is the number of times a pool space request failed. QIO Driver Errors is the number of times the QIO driver returned an error. QIO Limit Warnings is the number of times the process received an event signifying a pool or an MD shortage from the QIO monitor. Total MBUFs Allocated is the current number of MBUFs allocated. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -112 SCF Reference for Parallel Library TCP/IP STATS PROCESS Command for TCPSAM Description of Statistics for Socket Send Size Histogram Size 1-128 is the count of socket sends between 1 and 128 bytes. Size 129-256 is the count of socket sends between 129 and 256 bytes. Size 257-512 is the count of socket sends between 257 and 512 bytes. Size 513-1024 is the count of socket sends between 513 and 1024 bytes. Size 1025-2048 is the count of socket sends between 1025 and 2048 bytes. Size 2049-4096 is the count of socket sends between 2049 and 4096 bytes. Size 4097-8192 is the count of socket sends between 4097 and 8192 bytes. Size 8193-12288 is the count of socket sends between 8193 and 12288 bytes. Size 12289-16384 is the count of socket sends between 12289 and 16384 bytes. Size 16385-32768 is the count of socket sends between 16385 and 32768 bytes. Description of Statistics for the ARP STATS In ARP Requests is the number of ARP requests received. Out ARP Requests is the number of ARP requests sent. In ARP Replys is the number of ARP replies received. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -113 SCF Reference for Parallel Library TCP/IP STATS PROCESS Command for TCPSAM Out ARP Replys is the number of ARP replies sent. In InARP Requests is the number of inverse ARP requests received. Out InARP Requests is the number of inverse ARP requests sent. In InARP Replys is the number of inverse ARP replies received. Out InARP Replys is the number of inverse ARP replies sent. In ARP Naks is the number of ARP Naks received. Out ARP Naks is the number of ARP Naks sent. Description of Statistics for IGMP Statistics Total Packets Input is the total number of IGMP packets received. Total Reports Sent is the total number of IGMP report packets sent by this process. Short Packets is the total number of IGMP packets received that were too short. Bad Checksum is the total number of IGMP packets received that had an incorrect checksum. Total Queries Input is the total number of IGMP query packets received. Bad Queries is the total number of IGMP query packets received with the IP destination address not equal to the all hosts group. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -114 SCF Reference for Parallel Library TCP/IP STATS ROUTE Command for TCPMAN Total Reports Input is the total number of IGMP membership reports received. Bad Reports is the total number of bad IGMP membership reports received. Reports For Our Groups is the total number of IGMP membership reports received for groups we belong to. STATS ROUTE Command for TCPMAN The STATS ROUTE command displays the Parallel Library TCP/IP subsystem statistics for the specified routes. Note. STATS ROUTE with the RESET option is sensitive. Command Syntax STATS [ / OUT file-spec / ] [ROUTE $ZZTCP.#ZPTMn.route-name] [, RESET ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. ROUTE $ZZTCP.#ZPTMn.route-name is the name of the route. The fully-qualified name for the ROUTE is $ZZTCP.#ZPTM{0-F}.route-name. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. You may use the wild card (*) in place of the TCPMON name; this yields statistics for the route on all TCPMONs. You also may use the wild card in place of the route name; this yields statistics for all routes either on the specified TCPMON or, if the wild card is also used for the TCPMON, on all TCPMONs. RESET resets the statistical counters to zero. (This option is sensitive.) Examples The following command requests statistics about all running routes: SCF> STATS ROUTE $ZZTCP.*.* HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -115 SCF Reference for Parallel Library TCP/IP STATS ROUTE Command for TCPMAN STATS ROUTE Display Format The format of the display for a ROUTE object is: TCPMAN Stats ROUTE \OSCAR.$ZPTM0.* Sample Time ... 04 Jan 2000, 16:53:45.447 Reset Time .... 03 Jan 2000, 10:55:16.071 Name RT1 Route Usage 0D Sample Time ... 04 Jan 2000, 16:53:45.447 Reset Time .... 03 Jan 2000, 10:55:16.091 Name RT3 Route Usage 0D Sample Time ... 04 Jan 2000, 16:53:45.447 Reset Time .... 04 Jan 2000, 15:56:52.068 Name RT4 Route Usage 0D Sample Time ... 04 Jan 2000, 16:53:45.447 Reset Time .... 03 Jan 2000, 10:55:16.098 Name DEF Route Usage 0D TCPMAN Stats ROUTE \OSCAR.$ZPTM1.* Sample Time ... 04 Jan 2000, 16:53:51.612 Reset Time .... 03 Jan 2000, 10:55:16.609 Name RT1 Route Usage 0D Sample Time ... 04 Jan 2000, 16:53:51.612 Reset Time .... 03 Jan 2000, 10:55:16.615 Name RT3 Route Usage 0D Sample Time ... 04 Jan 2000, 16:53:51.612 Reset Time .... 04 Jan 2000, 15:56:52.155 Name Route Usage RT4 0D Sample Time ... 04 Jan 2000, 16:53:51.612 Reset Time .... 03 Jan 2000, 10:55:16.619 Name DEF Route Usage 0D Sample Time is the time when the statistics were sampled (displayed or written to a file). Reset Time is the time when the counters were last reset to zero. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -116 SCF Reference for Parallel Library TCP/IP STATS ROUTE Command for TCPSAM Name is the name of the route. Route Usage is the number of times this route was used to send IP datagrams. STATS ROUTE Command for TCPSAM The STATS ROUTE command displays the Parallel Library TCP/IP subsystem statistics for the specified routes in the processor containing the TCPSAM process. Note. STATS ROUTE with the RESET option is sensitive. Command Syntax STATS [ / OUT file-spec / ] [ROUTE $tcpsam-name.route-name] [, RESET ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. ROUTE $tcpsam-name.route-name is the name of the route. The fully-qualified name for the ROUTE is $tcpsamname.route-name. You may substitute the wild card (*) for the route name; doing so retrieves STATS for all routes in the TCPSAM primary processor. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. RESET resets the statistical counters to zero. (This option is sensitive.) Example The following command requests statistics for the specified TCPSAM process. ->STATS ROUTE $ZSAM1.* HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -117 SCF Reference for Parallel Library TCP/IP STATS SUBNET Command for TCPMAN STATS ROUTE Display Format The format of the display for a ROUTE object is: TCPIP Stats ROUTE \SYSA.$ZSAM1.* Sample Time ... 23 March 1996, 17:18:25.334 Reset Time .... 23 March 1996, 11:47:47.166 Name #ROU1 Route Usage 10709D Sample Time is the time when the statistics were sampled (displayed or written to a file). Reset Time is the time when the counters were last reset to zero. Name is the name of the route. Route Usage is the number of times this route was used to send IP datagrams. STATS SUBNET Command for TCPMAN The STATS SUBNET command displays the statistical information for the specified subnets in a given TCPMON or in all configured TCPMONs. Command Syntax STATS [ / OUT file-spec / ] [SUBNET $ZZTCP.#ZPTMn.subnet-name] [ , RESET ] [ , DETAIL ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. SUBNET $ZZTCP.#ZPTMn.subnet-name is the name of the subnet. The fully-qualified name of the subnet is $ZZTCP.#ZPTM{0-f}.subnet-name. You may substitute the wild card (*) for the TCPMON name; doing so yields statistical information for subnets configured on all TCPMONs. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for GSeries RVUs. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -118 STATS SUBNET Command for TCPMAN SCF Reference for Parallel Library TCP/IP RESET resets the statistical counters to zero. DETAIL requests the detailed status information for the subnet. Examples The following example requests statistics for subnets for starting with SN: -> STATS SUBNET $ZZTCP.#ZPTM3.SN* The following example requests statistics for all running subnets: ->STAT SUBNET $ZZTCP.*.* STATS SUBNET Display Format The format of the display for the first example of the STATS SUBNET command is: TCPMAN Stats SUBNET \BEAR.$ZZTCP.#ZPTM3.SN* Sample Time ... 28 Jan 2000, 13:49:48.912 Reset Time .... 28 Jan 2000, 13:40:30.682 Name EN1 Filter Errors........0 Output Packets.......0 Output Errors........0 TCP filters Reg......0 TCP filters Dereg....0 UDP filters Error....0 Port filters Drop....0 Name Filter Timeout.....0 Input Packets......54 Input Errors.......0 TCP filters Error..0 UDP filters Reg....0 UDP filters Dereg..0 EN2 Sample Time ... 28 Jan 2000, 13:49:48.912 Reset Time .... 28 Jan 2000, 13:40:30.682 Filter Errors........0 Output Packets.......0 Output Errors........0 TCP filters Reg......0 TCP filters Dereg....0 UDP filters Error....0 Port filters Drop...0 Filter Timeout.....0 Input Packets......54 Input Errors.......0 TCP filters Error..0 UDP filters Reg....0 UDP filters Dereg..0 Media State Down...1 Sample Time is the time when the statistics were sampled (displayed or written to a file). HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -119 SCF Reference for Parallel Library TCP/IP STATS SUBNET Command for TCPMAN Reset Time is the time when the counters were last initialized (set to zero). Name is the name of the subnet. Filter Errors indicates the number of errors received from SLSA for filter registrations. Filter Timeouts indicates that the filter registration is not receiving a reply from SLSA in the allowed time. Output Packets is the number of packets sent by the subnet. Input Packets is the number of packets received by the subnet. Output Errors is the number of errors that occurred when packets were sent by the subnet. Each output error also generates one of the following operator messages: DEVICE READ ERROR error ON IOP iopname DEVICE WRITE ERROR error ON IOP iopname ERROR error ON IOP iopname Input Errors is the number of errors detected when packets were received by the subnet. Each input error also generates one of the following operator messages: DEVICE READ ERROR error ON IOP iopname DEVICE WRITE ERROR error ON IOP iopname ERROR error ON IOP iopname TCP filters Reg is the number of TCP filters registered. TCP filters Error is the number of TCP filter registration errors. TCP filters Dereg is the number of TCP filters de-registered. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -120 SCF Reference for Parallel Library TCP/IP STATS SUBNET Command for TCPSAM UDP filters Reg is the number of UDP filters registered. UDP filters Error is the number of UDP filter registration errors. UDP filters Dereg is the number of UDP filters de-registered. Port filters Drop is the number of port filters dropped. Media State Down shows the total media down events received from the adapter. Considerations • • • • The object-name template (wild-card notation) is supported. STATS is a nonsensitive command without the RESET option; it is a sensitive command with the RESET option. To initialize (set to zero) the statistical counters, use the RESET option. STATS, RESET is sensitive. The STATS command returns the time at which the current statistics were sampled and the time at which the counters were last reset. STATS SUBNET Command for TCPSAM The STATS SUBNET command displays the statistical information for the specified subnets in the processor containing the TCPSAM process. Command Syntax STATS [ / OUT file-spec / ] [SUBNET $tcpsam-process.subnet-name] [ , RESET ] [ , DETAIL ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -121 STATS SUBNET Command for TCPSAM SCF Reference for Parallel Library TCP/IP SUBNET $tcpsam-process.subnet-name is the name of the subnet. The fully-qualified name of the subnet is $tcpsamprocess.subnet-name. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. RESET resets the statistical counters to zero. DETAIL requests the detailed status information for the subnet. Example The following example requests statistics for all running subnets in the processor containing the TCPSAM process: ->STATS SUBNET $ZSAM1.* STATS SUBNET Display Format The format of the display for the STATS SUBNET command is: TCPIP Stats SUBNET \SYSTEM.$ZSAM1.* Sample Time ... 19 Feb 1998, 9:00:56:.054 Reset time ... 18 Feb 1998, 21:09:10.986 Name #LOOP0 Output Packets 0D Input Output Packets Errors 0D 0D Input Errors 0D Filter Errors 0D Filter Timeouts OD Filter Errors Filter Timeouts SAMPLE TIME ... 19 FEB 1998, 9:00:56:055 RESET TIME .... 19 FEB 1998, 7:36:15.674 NAME Output Packets #SN1 1033D Input Output Packets Errors 4496D 0D Input Errors 0D 0D 0D Sample Time is the time when the statistics were sampled (displayed or written to a file). Reset Time is the time when the counters were last initialized (set to zero). Name is the name of the subnet. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -122 SCF Reference for Parallel Library TCP/IP STATUS Command Output Packets is the number of packets sent by the subnet. Input Packets is the number of packets received by the subnet. Filter Errors indicates the number of errors received from SLSA for filter registrations. Filter Timeouts indicates that the filter registration is not receiving a reply from SLSA in the allowed time. Output Errors is the number of errors that occurred when packets were sent by the subnet. Each output error also generates one of the following operator messages: DEVICE READ ERROR error ON IOP iopname DEVICE WRITE ERROR error ON IOP iopname ERROR error ON IOP iopname Input Errors is the number of errors detected when packets were received by the subnet. Each input error also generates one of the following operator messages: DEVICE READ ERROR error ON IOP iopname DEVICE WRITE ERROR error ON IOP iopname ERROR error ON IOP iopname STATUS Command The STATUS command reports the status of the specified PTCPIP object. Use the status command with the wild-card (*) option whenever you want to find out the names of TCPMON, route, entry and subnet objects in the Parallel Library TCP/IP subsystem. For example, if you have ASSUMED PROCESS $ZZTCP, the STATUS MON * command lists all the running TCPMON objects in the system and the STATUS ROUTE *.* command lists all the running routes in the system. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -123 SCF Reference for Parallel Library TCP/IP STATUS ENTRY Command for TCPMAN STATUS ENTRY Command for TCPMAN The STATUS ENTRY command displays the dynamic status of the specified entry in a given TCPMON or in all configured TCPMONs. Command Syntax STATUS [ / OUT file-spec / ] [ ENTRY $ZZTCP.#ZPTMn.entry-name] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. ENTRY $ZZTCP.#ZPTMn.entry-name is the name of the entry. The fully-qualified name of the entry object is $ZZTCP.#ZPTMn.entry-name. You may substitute the wild card (*) for the TCPMON name; doing so yields the status information for the specified entry on all TCPMONs (in all processors). You may substitute the wild card (*) for the entry name; doing so yields the status information for the all entries either on all TCPMONs, if you used the wild card for the TCPMON or on the specified TCPMON. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. Examples The following commands return status information about all entries contained in the ARP entry table: -> ASSUME PROCESS $ZZTCP -> STATUS ENTRY *.* STATUS ENTRY Response Display The format of the STATUS ENTRY display is: Name:(ARP) IPADDRESS........ 172.16.119.1 Arp Timer........... 19 (Min) Arp Flags........ (INUSE,COM) MacAddress.......... %H00 000C 3920CE Name is the name of the entry. The entry type is indicated in parentheses to the right of the name. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -124 SCF Reference for Parallel Library TCP/IP STATUS MON Command for TCPMAN IPADDRESS is the IP address of the entry. Arp Timer is the time in minutes left to expire. Arp Flags is the state of the ARP table entry; possible values are: COMPLETED indicates a resolved entry. That is, the reply to an ARP request has been received. PERMANENT indicates a static entry that is never cleared from the ARP cache. INUSE indicates an entry that is currently in use. INCOMPL indicates an unresolved entry. That is, an ARP request has been sent but a reply has not been received yet. Timer (2 minutes) specifies how long it will wait for a response before this entry is flushed. No further action is necessary. IDLE indicates an entry which was not used at all for 10 minutes. The timer (10 minutes) specifies the time after which this entry will be flushed. No further action is necessary. MacAddress is the MAC address of the entry in hexadecimal format. STATUS MON Command for TCPMAN The STATUS MON command displays the dynamic state of a TCPMON process or of all configured TCPMON processes and any in-use ports. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -125 STATUS MON Command for TCPMAN SCF Reference for Parallel Library TCP/IP Command Syntax STATUS [ / OUT file spec / ] [ MON $ZZTCP.#ZPTMn ] [ , DETAIL ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. MON $ZZTCP.#ZPTMn is the name of the TCPMON. The fully-qualified name of the TCPMON object is $ZZTCP.#ZPTMn. You may substitute the wild card (*) for the TCPMON name; doing so yields the status information on all TCPMONs (in all processors). If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. DETAIL requests the detailed status information for the TCPMON. Examples This command displays the dynamic state of all TCPMON objects, and any in-use ports. -> STATUS MON $ZZTCP.* This command displays the dynamic state of TCPMON object #ZPTM1, and any in-use ports. -> STATUS MON $ZZTCP.#ZPTM1, DETAIL STATUS MON Display Format The format of the display for STATUS MON, DETAIL is: PTCPIP Detailed Status MON \IDC26.$ZZTCP.#ZPTM1 Status: STARTED PID............ ( 1,342) Proto State TCP SYN-SENT TCP ESTAB TCP ESTAB TCP LISTEN UDP Laddr/OutSubNet 172.31.45.90 SN1 172.31.45.90 LOOP0 172.31.45.90 LOOP0 0.0.0.0 0.0.0.0 Lport 5027 Faddr 172.31.45.142 Fport SendQ telnet 0 RecvQ 0 telnet 172.31.45.90 5025 0 0 5025 172.31.45.90 telnet 0 0 telnet 0.0.0.0 * 0 0 4444 0.0.0.0 * 0 0 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -126 SCF Reference for Parallel Library TCP/IP STATUS MON Command for TCPMAN Status indicates the state of the TCPMON. PID is the process ID of the TCPMON in the PTCPIP subsystem. Proto is the protocol associated with the socket, which can be UDP (for a UDP socket), TCP (for a TCP socket), or a protocol number (for a raw IP socket). State is the current state of the socket; it applies only to sockets whose Proto value is TCP. The possible values are: CLOSING if waiting for a terminate connection request acknowledgment from the remote site. CLOSE-WAIT if waiting for a terminate connection request from the local user. ESTAB if the connection is open and the user can send and receive data. This is the normal state for data transfer. FIN-WAIT-1 if waiting for a terminate connection request from the remote TCP site or if waiting for acknowledgment of the terminate connection request that the process has sent previously. FIN-WAIT-2 if waiting for a termination of data to be received after having sent a FIN (termination of data being sent). LISTEN if waiting for a connection request from any remote TCP site. LAST-ACK if waiting for acknowledgment of the terminate connection request previously sent to the remote site (which includes an acknowledgment of its terminate connection request). HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -127 SCF Reference for Parallel Library TCP/IP STATUS MON Command for TCPMAN SYN-RCVD if waiting for an acknowledgment of a SYN-ACK sent in response to a SYN. SYN-SENT if waiting for a SYN-ACK after having sent a SYN. TIME-WAIT if waiting for sufficient time to pass (about two round trips) to be sure that stray packets are flushed from the network. UNKNOWN the socket was in the closing state when the command was issued. Laddr is the local Internet address associated with the socket, displayed as four-decimal octets. OutSubNet is the subnet associated with the socket used for outbound traffic. Lport is the local port number for either TCP or UDP, depending on the value of Proto. The more common port values are displayed in text form; others are displayed as four-decimal octets. Faddr is the foreign (remote) Internet address associated with the socket, displayed in four-decimal octets. Fport is the foreign port number for either TCP or UDP, depending on the value of Proto. The more common port values are displayed in text form; others are displayed as four-decimal octets. SendQ is the number of bytes of data in the send queue of the socket. RecvQ is the number of bytes of data in the receive queue of the socket. Multicast Groups is the list of internet addresses joined by an application. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -128 STATUS PROCESS Command for TCPMAN SCF Reference for Parallel Library TCP/IP Multicast Group States is the state of the multicast groups. Possible values are: STARTED indicates that multicast is operational for the group. STARTING indicates that the multicast group is transitioning to the STARTED (and operational) state but is not yet fully operational. STOPPED indicates that multicast is not operational for the group. STATUS PROCESS Command for TCPMAN The STATUS PROCESS command displays the TCPMAN's primary and backup information. Command Syntax STATUS [ / OUT file spec / ] [ PROCESS $ZZTCP ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. PROCESS $ZZTCP is the name of the TCPMAN process. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. Examples This command displays the primary and backup information for the TCPMAN process, $ZZTCP. -> STATUS PROCESS $ZZTCP STATUS PROCESS Display Format The format of the display for STATUS PROCESS is: TCPMAN Status PROCESS \BEAR.$ZZTCP PPID............ ( 2,269) BPID................... ( 3, 272) HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -129 SCF Reference for Parallel Library TCP/IP STATUS PROCESS Command for TCPSAM Status always indicates that the process is STARTED. PPID is the processor and process ID of the TCPMAN primary process. BPID is the processor and process ID of the TCPMAN backup process. If TCPMAN is running without a backup process, this field shows ( 0, 0). STATUS PROCESS Command for TCPSAM The STATUS PROCESS command displays the dynamic state of the TCPSAM process and any in-use ports. Command Syntax STATUS [ / OUT file spec / ] [ PROCESS $tcpsam-name ] [, DETAIL] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. PROCESS $tcpsam-name is the name of the TCPSAM process. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. Examples The following command displays the dynamic state of TCPSAM process, $SAM2, and any in-use ports: -> STATUS PROCESS $SAM2 , detail HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -130 STATUS PROCESS Command for TCPSAM SCF Reference for Parallel Library TCP/IP STATUS PROCESS Display Format The format of the display for STATUS PROCESS both with the detail option is: TCPIP Detailed Status PROCESS \BEAR.$SAM2 Status: STARTED PPID............( 1,54) BPID..................( 2, 32) Proto State Laddr Lport TCP ESTAB 50.0.0.3 ftp TCP LISTEN 0.0.0.0 echo UDP 0.0.0.0 8000 ---Multicast Groups--224.0.0.1 230.17.123.55 239.1.2.3 UDP 0.0.0.0 7000 Faddr 50.0.0.1 0.0.0.0 0.0.0.0 0.0.0.0 Fport SendQ RecvQ 1953 0 0 0 0 0 * 0 0 ---State--STARTED STARTING STOPPED * 0 0 Status always indicates that the process is STARTED. PPID is the processor and process ID of the TCPSAM primary process. BPID is the processor and process ID of the TCPSAM backup process. If TCPSAM is running without a backup process, this field shows ( 0, 0). Proto is the protocol associated with the socket, which can be UDP (for a UDP socket), TCP (for a TCP socket), or a protocol number (for a raw IP socket). State is the current state of the socket; it applies only to sockets whose Proto value is TCP. The possible values are: CLOSING if waiting for a terminate connection request acknowledgment from the remote site. CLOSE-WAIT if waiting for a terminate connection request from the local user. ESTAB if the connection is open and the user can send and receive data. This is the normal state for data transfer. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -131 SCF Reference for Parallel Library TCP/IP STATUS PROCESS Command for TCPSAM FIN-WAIT-1 if waiting for a terminate connection request from the remote TCP site or if waiting for acknowledgment of the terminate connection request that the process has sent previously. FIN-WAIT-2 if waiting for a termination of data to be received after having sent a FIN (termination of data being sent). LISTEN if waiting for a connection request from any remote TCP site. LAST-ACK if waiting for acknowledgment of the terminate connection request previously sent to the remote site (which includes an acknowledgment of its terminate connection request). SYN-RCVD if waiting for an acknowledgment of a SYN-ACK sent in response to a SYN. SYN-SENT if waiting for a SYN-ACK after having sent a SYN. TIME-WAIT if waiting for sufficient time to pass (about two round trips) to be sure that stray packets are flushed from the network. UNKNOWN the socket was in the closing state when the command was issued. Laddr is the local Internet address associated with the socket, displayed as four-decimal octets. Lport is the local port number for either TCP or UDP, depending on the value of Proto. The more common port values are displayed in text form; others are displayed as four-decimal octets. Faddr is the foreign (remote) Internet address associated with the socket, displayed in four-decimal octets. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -132 SCF Reference for Parallel Library TCP/IP STATUS ROUTE Command for TCPMAN Fport is the foreign port number for either TCP or UDP, depending on the value of Proto. The more common port values are displayed in text form; others are displayed as four-decimal octets. SendQ is the number of bytes of data in the send queue of the socket. RecvQ is the number of bytes of data in the receive queue of the socket. Multicast Groups indicates the IP multicast group addresses that the PTCPIP connection is listening to. Multicast Group States is the state of the multicast groups. Possible values are: STARTED indicates that multicast is operational for the group. STARTING indicates that the multicast group is transitioning to the STARTED (and operational) state but is not yet fully operational. STOPPED indicates that multicast is not operational for the group. STATUS ROUTE Command for TCPMAN The STATUS ROUTE command displays the current status of the specified routes on a given TCPMON or on all configured TCPMONs. Command Syntax STATUS [ / OUT file spec / ] [ROUTE $ZZTCP.#ZPTMn.route-name] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -133 STATUS ROUTE Command for TCPMAN SCF Reference for Parallel Library TCP/IP ROUTE $ZZTCP.#ZPTMn.route-name is the specification of the route name. The fully-qualified route name for TCPMAN is $ZZTCP.#ZPTMn.route-name. To obtain status information about a route on all configured TCPMONs, use the wild-card (*) notation for the TCPMON name. For example, STATUS ROUTE *.RT1. To obtain status information about a ROUTE on one TCPMON, qualify the TCPMON name. For example, STATUS ROUTE #ZPTM1.RT1. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. Examples The following command requests status information about all routes configured on the ZPTM1 TCPMON: -> STATUS ROUTE $ZZTCP.#ZPTM2.* STATUS ROUTE Display Format The format of the display for the STATUS ROUTE command is: TCPMAN Status ROUTE \BEAR.$ZZTCP.#ZPTM2.* Name RT2 RT3 DA2_2 DEF1 DEF2 Status STARTED STARTED STARTED STARTED STARTED RefCnt 1D 0D 0D 3D 1D SecondaryRoutes 1D 1D 0D 1D 1D Name is the name of the route. Status is the summary state of the route. RefCnt specifies the number of users currently using the specific route. If the value is greater than zero, an application is currently using the specified route. SECONDARYROUTES indicates the number of shadow routes associated with this route. For multiple routes to the same destination, all the routes in addition to the primary route (which is the route visible to Radix Routing topology) are called shadow/secondary routes. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -134 STATUS ROUTE Command for TCPSAM SCF Reference for Parallel Library TCP/IP STATUS ROUTE Command for TCPSAM This command displays the status of the routes configured in the TCPMON on the TCPSAM primary processor. This is a nonsensitive command. Command Syntax STATUS [ / OUT file spec / ] [ROUTE $tcpsam-name.#route-name ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. ROUTE $tcpsam-name.#route-name is the specification of the route name. The fully-qualified route name for TCPSAM is $tcpsam-name.#route-name. To obtain status information about all routes configured on the TCPSAM primary processor, use the wild-card (*) notation for the route name. For example, STATUS ROUTE $ZTC1.*. To obtain status information about one route, qualify the route name. For example, STATUS ROUTE $ZTC1.#RT1. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. Examples -> STATUS ROUTE $ZTC1.* STATUS ROUTE Display Format The format of the display for the STATUS ROUTE command is: TCPIP Status ROUTE \BOBAFET.$ZTC1.* Name #RT2 #RT3 #DA2_2 Status STARTED STARTED STARTED RefCnt 1D 0D 0D Name is the name of the route. Status is the summary state of the route. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -135 STATUS SUBNET Command for TCPMAN SCF Reference for Parallel Library TCP/IP Considerations A pound sign (#) precedes the ROUTE name for backward compatibility with applications that expect this naming convention for ROUTEs. See Supported Commands and Object Types on page 5-9. STATUS SUBNET Command for TCPMAN The STATUS SUBNET command displays the current status of the specified subnets. Command Syntax STATUS [ / OUT file spec / ] [SUBNET $ZZTCP.#ZPTMn.subnet-name] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. SUBNET $ZZTCP.#ZPTMn.subnet-name is the specification of the subnet. The fully-qualified subnet name for TCPMAN is $ZZTCP.#ZPTMn.subnet-name. To obtain information about a subnet’s status in all TCPMONs, substitute the wild card (*) for the TCPMON name. To obtain information about all running subnets, substitute the wild card (*) for both the TCPMON name and the subnet name. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. Examples The following example shows the command and display for STATUS SUBNET. The wild card (*) is used to obtain status information on all subnets. -> STATUS SUBNET $ZZTCP.#ZPTM2.SN* STATUS SUBNET Display Format The format of the display for the STATUS SUBNET command is: PTCPIP Status SUBNET \BEAR.$ZZTCP.#ZPTM2.SN* Name Status SN1 SN2 SN3 SN4 STARTED STARTED STARTED STARTED FailOver YES YES YES YES SharedIP NO NO YES-PRIMARY YES-BACKUP HasAlias YES YES YES YES AssociateSub SN2 SN1 SN4 SN3 Media State UP UP UP DOWN HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -136 SCF Reference for Parallel Library TCP/IP STATUS SUBNET Command for TCPSAM Name is the name of the subnet. Status is the summary state of the subnet. FailOver indicates whether the subnet is failover enabled or not. SharedIP indicates whether the subnet has the same subnet IP address as its brother. YES-PRIMARY indicates that this subnet is failover-capable and has the same subnet IP address as another subnet. YES-BACKUP indicates that this subnet is failover-capable and has the same subnet IP address as another subnet. NO indicates that this subnet does not have the same subnet IP address as another subnet. HasAlias indicates that this subnet has been configured for IP aliasing addresses. AssociateSub is the associated subnet in a failover pair. Media State is in the UP state when the subnet’s adapter has detected a link pulse from the network connection. A DOWN state indicates problems with the cabling or the associated network equipment. STATUS SUBNET Command for TCPSAM This command displays the status of the subnets configured in the TCPMON process on the TCPSAM primary processor. This is a nonsensitive command. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -137 STATUS SUBNET Command for TCPSAM SCF Reference for Parallel Library TCP/IP Command Syntax STATUS [ / OUT file spec / ] [SUBNET $tcpsam-name.#subnet-name] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. SUBNET $tcpsam-name.#subnet-name is the specification of the subnet. The fully-qualified subnet name for TCPSAM is $tcpsam-name.#subnet-name. To obtain information about all running subnets in the TCPSAM primary processor, substitute the wild card (*) for the subnet name. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. Example The following example shows the command and display for STATUS SUBNET. The wild card (*) is used to obtain status information on all subnets. -> STATUS SUBNET $ZTC1.* STATUS SUBNET Display Format The format of the display for the STATUS SUBNET command is: TCPIP Status SUBNET \BOBAFET.$ZTC1.* Name #LOOP0 #SN1 #SN2 Status STARTED STARTED STOPPED Name is the name of the subnet. Status is the summary state of the subnet. Considerations A pound sign (#) precedes the SUBNET name for backward compatibility with applications that expect this naming convention for SUBNETs. See Supported Commands and Object Types on page 5-9. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -138 SCF Reference for Parallel Library TCP/IP STOP Command STOP Command The STOP command terminates the operation of the specified PTCPIP object. You can stop processes, subnets, and routes. When the operation is complete, the object(s) is in the STOPPED summary state. If the specified objects are in use, the STOP command is not completed. If you attempt to stop an object that is in use or is already in the STOPPED summary state, the Parallel Library TCP/IP subsystem returns a warning. This is a sensitive command. STOP MON Command for TCPMAN The STOP MON command terminates the activity on a given TCPMON or all configured TCPMONs in a normal, orderly manner. The STOP command can't be used if open sockets exist; (use ABORT MON instead). This command does not delete the MON object from the system configuration database. Command Syntax STOP [ /OUT file-spec/ ] MON $ZZTCP.#ZPTMn OUT file-spec causes any SCF output generated for this command to be directed to the specified file. MON $ZZTCP.#ZPTMn is the name of the TCPMON object. If you omit the object name, SCF uses the assumed object name. You can substitute the wild card (*) for the TCPMON name; doing so stops all TCPMONs. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. Examples The following command terminates the operation of all TCPMONs: SCF> STOP MON $ZZTCP.* Considerations By using the STOP MON command instead of the ABORT MON, you ensure that when you restart the MON, it comes up with the non-default attributes restored from the configuration database. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -139 SCF Reference for Parallel Library TCP/IP STOP PROCESS Command for TCPMAN STOP PROCESS Command for TCPMAN The STOP PROCESS command terminates the activity of the specified TCPMAN process in a normal, orderly manner. This is a sensitive command. Command Syntax STOP [ / OUT file-spec / ] [ PROCESS $ZZTCP ] [, SUB ALL ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. PROCESS $ZZTCP is the TCPMAN process. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. SUB ALL The SUB ALL modifier stops all configured TCPMON objects. Examples The following command terminates the operation of the TCPMAN process: SCF> STOP PROCESS $ZZTCP Considerations • • • To stop a process immediately, use the ABORT command. If the TCPMAN process has been added as a generic process, you must use the ABORT command under the Kernel subsystem (ABORT PROCESS $ZZKRN.#ZZTCP) to stop it. See the SCF Reference Manual for G-Series RVUs for more information about managing generic processes. The SUB ALL option stops the TCPMON objects but leaves them in the system configuration database. STOP PROCESS Command for TCPSAM The STOP PROCESS command terminates the activity of the specified TCPSAM process without stopping the Parallel Library TCP/IP subsystem. This is a sensitive command. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -140 SCF Reference for Parallel Library TCP/IP STOP ROUTE Command for TCPMAN Command Syntax STOP [ / OUT file-spec / ] [ PROCESS $tcpsam-name ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. PROCESS $tcpsam-name is the name of the TCPSAM process. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. Examples The following command terminates the operation of the TCPSAM process: SCF> STOP PROCESS $ZTC1 Considerations To stop a process immediately, use the ABORT command. STOP ROUTE Command for TCPMAN The STOP ROUTE command terminates the activity of the specified route in a normal, orderly manner. This is a sensitive command. Command Syntax STOP [ / OUT file-spec / ] [ROUTE $ZZTCP.#ZPTMn.route-name ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. ROUTE $ZZTCP.#ZPTMn.route-name is the specification for the route. The fully-qualified route name is $ZZTCP.#ZPTMn.route-name. However, you must stop the routes on all configured monitors (by substituting the wild card (*) for the monitor name) unless you are stopping a dynamic route. Dynamic routes run only in one monitor and you must delete the dynamic route on the monitor in the processor where it was HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -141 SCF Reference for Parallel Library TCP/IP STOP SUBNET Command for TCPMAN created. You can also substitute the wild card for the route name; doing so stops all routes on all TCPMONs. Examples The following command terminates the operation of the specified route: SCF> STOP ROUTE $ZZTCP.#ZPTM1.RT1 Considerations • • • • • • Link-level routes, generated internally by the ARP logic, cannot be stopped externally by using the SCF ABORT/STOP ROUTE commands but can be deleted externally by using the SCF DELETE ROUTE command. When you stop a static or implicit route, you must do so on all configured TCPMONs. You can use the wild-card (*) notation for the TCPMON name, but if you do not, it is assumed. When you stop a dynamic route on a configured TCPMON object, the dynamic route must be the one created in that processor. To stop a route immediately, use the ABORT command. To remove a route from the subsystem configuration database, use DELETE ROUTE. STOP SUBNET Command for TCPMAN The STOP SUBNET command terminates the activity of the specified subnets in a normal manner. Command Syntax STOP [ / OUT file-spec / ] [SUBNET $ZZTCP.#ZPTMn.subnet-name] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. SUBNET $ZZTCP.*.subnet-name is the specification of the subnet. The fully-qualified name of the subnet is $ZZTCP.*.subnet-name (you must stop subnets on all TCPMONs). However, specifying the MON object in the command is optional. (See the example below.) You can also substitute the wild card for the subnet name; doing so stops all subnets on all TCPMONs. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -142 SCF Reference for Parallel Library TCP/IP TRACE Command Examples The following command terminates the operation of the subnet SN2: SCF> STOP SUBNET $ZZTCP.*.SN2 Considerations • • • To stop a subnet immediately, use the ABORT command. To remove a subnet from the system configuration database, use the DELETE SUBNET command. You can use the wild-card (*) notation for the TCPMON name, but if you do not, it is assumed. TRACE Command The TRACE command allows you to capture and store records that you can then display using the PTrace utility. The TRACE command can request the capture of data items, alter trace attributes that were set by a previous use of the command, or stop a previously requested trace operation. An SCF trace produces a trace file that can be displayed using the commands available in the PTrace program. The trace file is created by SCF. The PTrace program is described in the PTrace Reference Manual. This is a sensitive command. Caution. The trace operation can significantly increase processor use by the PTCPIP process. TRACE MON Command for TCPMAN The TRACE MON command creates trace records corresponding to TCPMON process activities. In the case of multiple TCPMON processes, separate TRACE commands are required for each process. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -143 TRACE MON Command for TCPMAN SCF Reference for Parallel Library TCP/IP Command Syntax TRACE [ /OUT file-spec/ ] {, STOP } | {, TO file-spec MON [ $ZZTCP.#ZPTMn ] [ [ [ [ [ [ , , , , , , BULKIO / NOBULKIO COUNT count NOCOLL RECSIZE size SELECT select-spec PAGES pages ] ] ] ] ] ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. MON $ZZTCP.#ZPTMn, is the name of the TCPMON you want to trace. The wild card (*) is not supported. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. STOP ends the trace operation. A TRACE command must include either the STOP option or the TO option. TO file-spec specifies the name of the file into which the results of the trace operation are to be placed. It is a required option if STOP is not used. BULKIO / NOBULKIO specifies whether TRACE should use bulk I/O for tracing. BULKIO, the default parameter, specifies that the TRACE collector use bulk I/O to write data to the disk file, thus reducing the number of missing frame errors reported by PTRACE. NOBULKIO and BULKIO are optional parameters. SELECT selects the operations to be traced. For the SUBNET object, you can specify the following for select-spec: ALL All records SOCKCMD Socket requests (bind, listen, accept, connect send) MSGSYS Message system interface MALLOC Resource allocation and deallocation events ROUTING Requests for route changes HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -144 SCF Reference for Parallel Library TCP/IP TRACE PROCESS Command for TCPMAN UDP UDP interface layer TCP Transmission Control Protocol message layer IP IP layer LOGIC Several of the above selections including socket requests (SOCKCMD) and message system interface (MSGSYS) COUNT count specifies the number of trace records to be captured. count is an integer in the range -1 through 32767. If this option is omitted or if count equals -1, records are accumulated until you use the STOP option. NOCOLL indicates that the trace collector process should not be initiated. PAGES pages designates how much space, in units of pages, is allocated in the extended data segment used for tracing. PAGES can be specified only when a trace is being initiated, not when its attributes are being modified. pages is an integer in the range 4 through 64, or it is equal to 0. If you omit this option or specify 0, the default value of 64 is applied to the trace. RECSIZE specifies the length, in bytes, of the data in the trace data records. size is an integer in the range 300 through 4050. The length of the trace header, which is 8 bytes, is not included in size. If you omit this option or specify 0, an error occurs. Example ->TRACE MON $ZZTCP.#ZPTM0 TO TRACE1,RECSIZE 300,NOBULKIO ->TRACE,STOP ->PTRACE FROM TRACE1 TRACE PROCESS Command for TCPMAN TRACE can request the capture of target-defined data items, alter trace parameters, and end tracing of the TCPMAN PROCESS. This is a sensitive command. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -145 SCF Reference for Parallel Library TCP/IP TRACE PROCESS Command for TCPMAN Command Syntax TRACE [ /OUT file-spec/ ] PROCESS $ZZTCP { , STOP [ , BACKUP ] | [ , TO file-spec [ , BACKUP ] [ , COUNT count ] [ , NOCOLL ] [ , RECSIZE size ] [ , SELECT select-spec ] [ , PAGES pages ] ] } OUT file-spec causes any SCF output generated for this command to be directed to the specified file. PROCESS $ZZTCP is the name of the TCPMAN process. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. STOP Discontinues the trace currently in progress. TO file-spec specifies the name of the file into which the results of the trace operation are to be placed. It is a required option if STOP is not used. BACKUP If BACKUP is specified, then the command applies to the back up TCPMAN process (i.e. the trace is stopped or started on the backup). If omitted the primary is assumed. The TCPMAN must be running as a NonStop process pair if this syntax is used. If the primary TCPMAN is being traced when a takeover by the backup TCPMAN occurs, the trace of the same TCPMAN continues, but most events that were being traced prior to the TCPMAN switch are no longer traced. This is because the TCPMAN being traced is no longer the primary. If neither PRIMARY nor BACKUP is designated, the primary TCPMAN is traced. COUNT count count is an integer in the range -1 to (32k-1). It specifies the number of trace records to be captured. If COUNT is not specified (or is specified as -1), records are accumulated until the trace is stopped. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -146 SCF Reference for Parallel Library TCP/IP TRACE PROCESS Command for TCPMAN NOCOLL Indicates that the trace collector process should not be initiated. The disk file is to be written to by Guardian. The attributes WRAP and NOCOLL might not be specified together. PAGES pages pages specifies how many extended data segment pages are allocated when tracing. An integer value in the range 4 to 64 is expected. The default is 64 pages. RECSIZE size size is an integer in the range 1024 to 4050. It controls the length of the data in the trace data records. The trace header not included in the RECSIZE. The default is 120 bytes. Eight bytes are used for the header, and 120 bytes are trace data. SELECT select-spec selects the operations to be traced. For the PROCESS object, you can specify the following for select-spec: ALL All records. COMMON Common module events. MEMORY Management of resources for TCPMAN internal memory. MSGSYS Message system interface. The SELECT option is used to select a subset of the possible record types for capture. Default is select ALL. TO file-spec file-spec specifies the file to which tracing is to be initiated. The file may be an unstructured file created by the user previously. Examples The following command traces the TCPMAN process, writes results into the file named $DATA1.TRC.TRCFILE allows the trace data to be overwritten when the EOF is reached, and selects tracing of all PTCPIP process activity: -> TRACE PROCESS $ZZTCP, SELECT (MEMORY, COMMON), TO $DAT11.TRC.TRCFILE The following command stops the tracing on the TCPMAN process: -> TRACE PROCESS $ZZTCP, STOP HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -147 TRACE PROCESS Command for TCPSAM SCF Reference for Parallel Library TCP/IP Considerations The TCPMAN trace facilities for the primary and the backup TCPMAN processes are independent. The primary and the backup traces can be active at the same time. In the event of a primary/backup process switch, such as when the PRIMARY command is issued, the original primary TCPMAN process becomes the new backup process and the backup process becomes the primary. Any activated trace is maintained, however, the BACKUP modifier must be used on all further TRACE commands to the original primary process. TRACE commands to the original backup process should no longer have the BACKUP modifier. TRACE PROCESS Command for TCPSAM TRACE can request the capture of target-defined data items, alter trace parameters, and end tracing of the TCPSAM PROCESS. This is a sensitive command. Command Syntax TRACE [ /OUT file-spec/ ] PROCESS $tcpsam-name { , STOP [ , BACKUP ] } | { [ , TO file-spec [ , BACKUP ] [ , COUNT count ] [ , NOCOLL ] [ , RECSIZE size ] [ , PAGES pages ] ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. PROCESS $tcpsam-name is the name of the TCPSAM process. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. STOP Discontinues the trace currently in progress. TO file-spec specifies the name of the file into which the results of the trace operation are to be placed. It is a required option if STOP is not used. BACKUP If BACKUP is specified, then the command applies to the back up TCPSAM process (i.e. the trace is stopped or started on the backup). If omitted the primary HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -148 SCF Reference for Parallel Library TCP/IP TRACE PROCESS Command for TCPSAM is assumed. The TCPSAM process must be running as a NonStop process pair if this syntax is used. If the primary TCPSAM is being traced when a takeover by the backup TCPSAM occurs, the trace of the same TCPSAM continues, but most events that were being traced prior to the TCPSAM switch is no longer traced. This is because the TCPSAM being traced is no longer the primary. If neither PRIMARY nor BACKUP is designated, the primary TCPSAM is traced. COUNT count count is an integer in the range -1 to (32k-1). It specifies the number of trace records to be captured. If COUNT is not specified (or is specified as -1), records are accumulated until the trace is stopped. NOCOLL Indicates that the trace collector process should not be initiated. The disk file is to be written to by Guardian. WRAP and NOCOLL cannot be specified at the same time. PAGES pages pages specifies how many extended data segment pages are allocated when tracing. An integer value in the range 4 to 64 is expected. The default is 64 pages. RECSIZE size size is an integer in the range 1024 to 4050. It controls the length of the data in the trace data records. The trace header not included in the RECSIZE. The default is 120 bytes. Eight bytes are used for the header, and 120 bytes are trace data. TO file-spec file-spec specifies the file to which tracing is to be initiated. The file may be an unstructured file created by the user previously. Examples The following command traces the TCPSAM process, writes results into the file named $DATA1.TRC.TRCFILE allows the trace data to be overwritten when the EOF is reached, and selects tracing of all PTCPIP process activity: -> TRACE PROCESS $ZTC1, SELECT (MEMORY, COMMON), TO $DAT11.TRC.TRCFILE The following command stops the tracing on the TCPSAM process: -> TRACE PROCESS $ZTC1, STOP HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -149 TRACE SUBNET Command for TCPMAN SCF Reference for Parallel Library TCP/IP Considerations The TCPSAM trace facilities for the primary and the backup TCPSAM processes are independent. The primary and the backup traces can be active at the same time. In the event of a primary/backup process switch, such as when the PRIMARY command is issued, the original primary TCPSAM process becomes the new backup process and the backup process becomes the primary. Any activated trace is maintained, however, the BACKUP modifier must be used on all further TRACE commands to the original primary process. TRACE commands to the original backup process should no longer have the BACKUP modifier. TRACE SUBNET Command for TCPMAN The TRACE SUBNET command creates trace records corresponding to data traffic being exchanged over a defined IP subnet. Command Syntax TRACE [/OUT file-spec/] [SUBNET $ZZTCP.#ZPTMn.subnet-name] {, STOP } | {, TO file-spec [ [ [ [ [ [ , , , , , , BULKIO / NOBULKIO COUNT count NOCOLL RECSIZE size SELECT select-spec PAGES pages ] ] ] ] ] ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. SUBNET $ZZTCP.#ZPTMn.subnet-name is the name of the subnet. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. STOP ends the trace operation. A TRACE command must include either the STOP option or the TO option. TO file-spec specifies the name of the file into which the results of the trace operation are to be placed. It is a required option if STOP is not used. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -150 SCF Reference for Parallel Library TCP/IP TRACE SUBNET Command for TCPMAN BULKIO / NOBULKIO specifies whether TRACE should use bulk I/O for tracing. BULKIO, the default parameter, specifies that the TRACE collector use bulk I/O to write data to the disk file, thus reducing the number of missing frame errors reported by PTRACE. BULKIO and NOBULKIO are optional parameters. SELECT selects the operations to be traced. For the SUBNET object, you can specify the following for select-spec: ALL All records IPI IP Input records IPO IP Output records ARPI ARP Input records ARPO ARP Output records LOGIC A combination of all the above records USERDATA Used with IPI and IPO to display user data COUNT count specifies the number of trace records to be captured. count is an integer in the range -1 through 32767. If this option is omitted or if count equals -1, records are accumulated until you use the STOP option. NOCOLL indicates that the trace collector process should not be initiated. WRAP and NOCOLL cannot be specified at the same time. PAGES pages designates how much space, in units of pages, is allocated in the extended data segment used for tracing. PAGES can be specified only when a trace is being initiated, not when its attributes are being modified. pages is an integer in the range 4 through 64, or it is equal to 0. If you omit this option or specify 0, the default value of 64 is applied to the trace. RECSIZE specifies the length, in bytes, of the data in the trace data records. size is an integer in the range 300 through 4050. The length of the trace header, which is 8 bytes, is not included in size. If you omit this option or specify 0, an error occurs. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -151 SCF Reference for Parallel Library TCP/IP VERSION Command Examples The following command traces the $ZZTCP.#ZPTM1.SN2 subnet, writes the results into the file $SYSA.TRACES.TCPSUB and traces all PTCPIP process activity on the subnet: SCF> TRACE SUBNET $ZZTCP.#ZPTM1.SN2, TO $SYSA.TRACES.TCPSUB,& RECSIZE 300, NOBULKIO VERSION Command The VERSION command displays the Parallel Library TCP/IP subsystem version number, product name, product number, and RVU date. Use the DETAIL option to display this information about the NonStop operating system, the SCF Kernel, and the Parallel Library TCP/IP product module. This is a nonsensitive command. VERSION MON Command for TCPMAN The VERSION MON command displays the version level of the TCPMON object. VERSION [ /OUT file-spec/ ] [ MON $ZZTCP.#ZPTMn.mon-name ] [, DETAIL ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. MON $ZZTCP.#ZPTMn.mon-name is the MON object. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. DETAIL designates that complete version information is to be returned for the specified object. If DETAIL is omitted, a single line of version information is returned for the object. Examples The following examples show the VERSION command for the MON object with and without the DETAIL option. ->VERSION MON $ZPTM1 ->VERSION MON $ZPTM1, DETAIL HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -152 SCF Reference for Parallel Library TCP/IP VERSION PROCESS Command for TCPMAN Version Command Display Format VERSION MON \BEAR.$ZZTCP.#ZPTM1:T0470G06_15APR2000_G06_MO_A0614 Detailed VERSION MON \BEAR.$ZZTCP.#PTM2 SYSTEM \BEAR T0470G06_15APR2000_G06_MO_A0614 GUARDIAN - T9050 - (Q06) SCF KERNEL - T9082G02 - (24SEP99) (26JUL99) PTCPIP PM - T0473G40 - (01JAN2000) - (A0602) VERSION PROCESS Command for TCPMAN The VERSION PROCESS command displays the version level of the PTCPIP subsystem. Command Syntax VERSION [ / OUT file-spec / ] [ PROCESS $ZZTCP ] [ , DETAIL ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. PROCESS $ZZTCP is the TCPMAN process. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. DETAIL designates that complete version information is to be returned for the specified object. If DETAIL is omitted, a single line of version information is returned for the object. Examples The second example shows the null command. ->VERSION PROCESS $ZZTCP HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -153 SCF Reference for Parallel Library TCP/IP VERSION PROCESS Command for TCPSAM VERSION Command Display Format The format of the display of the VERSION PROCESS command without the DETAIL option is: VERSION PROCESS \BOBAFET.$ZZTCP: T0468G40_01JAN2000_G40_MA_A0603 The format of the display of the VERSION PROCESS command with the DETAIL option is: Detailed VERSION PROCESS \BEAR.$ZZTCP SYSTEM \BEAR T0468G40_01JAN2000_G40_MA_A0603 GUARDIAN - T9050 - (Q06) SCF KERNEL - T9082G02 - (05AUG99) (26JUL99) TCPMAN PM - T0468G06 - (10AUG99) VERSION PROCESS Command for TCPSAM The VERSION PROCESS for TCPSAM displays the version level of the TCPSAM process. Command Syntax VERSION [ /OUT file-spec/ ] [ PROCESS $tcpsam-name ] [, DETAIL ] OUT file-spec causes any SCF output generated for this command to be directed to the specified file. PROCESS $tcpsam-name is the name of the TCPSAM process. If you omit the object name, SCF uses the assumed object name. For information about the ASSUME command, see the SCF Reference Manual for G-Series RVUs. DETAIL designates that complete version information is to be returned for the specified object. If DETAIL is omitted, a single line of version information is returned for the object. Examples This command requests the version level of the TCPSAM process named $ZTC1: ->VERSION PROCESS $ZTC1 ->VERSION PROCESS $ZTC1, DETAIL HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -154 SCF Reference for Parallel Library TCP/IP Parallel Library TCP/IP Trace Facility VERSION Command Display Format The format of the display of the VERSION command without the DETAIL option is: VERSION PROCESS \BOBAFET.$ZTC1: T0471G40_01JAN2000_G40_SA_A0603 The format of the display of the VERSION command with the DETAIL option is: Detailed VERSION PROCESS \BEAR.$ZTC1 SYSTEM \BEAR T0471G06_15APR2000_G06_SA_A0605 GUARDIAN - T9050 - (Q06) SCF KERNEL - T9082G02 - (05AUG99) (26JUL99) TCPIP PM - T6243G06 - (10AUG98) - (ABH00A) Parallel Library TCP/IP Trace Facility This section contains the following information: • • • An introduction to the Parallel Library TCP/IPtrace facility A description of the subsystem-specific PTrace commands and any special considerations for using these commands with the Parallel Library TCP/IP An example of each type of trace record display Introduction to PTrace Trace files contain a record of the communications between processes. Each subsystem determines what information is recorded in its trace files. This information varies as to the type of events that are recorded, the amount of detail that is included, and other subsystem-specific attributes. You can generate a Parallel Library TCP/IP trace file interactively or programmatically. To start a trace and capture data interactively, you use the SCF TRACE command. To start a trace and capture data programmatically, you use the Subsystem Programmatic Interface (SPI). The trace files created with either SCF or SPI are unstructured and cannot be printed or displayed directly. You use PTrace to display and examine the trace files. The PTrace program formats the data stored in these unstructured trace files for output to terminals, printers, or disk files. Figure 5-3 on page 5-156 shows the four general steps involved in recording and formatting trace data. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -155 SCF Reference for Parallel Library TCP/IP Introduction to PTrace Figure 5-3. Recording and Displaying Trace Data Start the trace interactively with the SCF TRACE command or programmatically through SPI. Collect trace data. Stop the trace with the SCF TRACE command or through SPI. Display the trace file with PTrace. VST0503.vsd 1. Start the trace interactively with the SCF TRACE command or programmatically with SPI. 2. The TRACE command allows you to specify attributes, such as the size of the trace records and the name and maximum size of the trace file. 3. Collect trace data. Send and receive data or perform other operations related to the problem you are analyzing. 4. Stop the trace with another SCF TRACE command or with SPI. 5. Display the trace file with PTrace. For additional information on using PTrace, refer to the PTrace Reference Manual. Device Type and Subtype When a trace file is created, the type and subtype of the device being traced are recorded in that file. When PTrace opens the trace file, it uses this information to determine for which subsystem PTrace is formatting records. The device type and subtype for the TCPMAN are 68 and 0, respectively. The device type and subtype for TCPSAM are 48 and 0 respectively. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -156 SCF Reference for Parallel Library TCP/IP PTrace Commands PTrace Commands The PTrace commands provide options for selecting trace records for display, so you can suppress those records that do not relate to the problem you are investigating. The PTrace commands also provide options for specifying the way in which the trace records are formatted. Although PTrace provides a common set of commands for displaying trace records, not all of the PTrace commands are supported by each subsystem. This is because of the structure of the PTrace code. The PTrace code actually consists of two modules. The first module contains the code shared by all subsystems; the second contains the additional, subsystem-specific code that actually displays the PTrace records. Thus, those commands implemented by the first PTrace module are supported by all subsystems: ALLOW, COUNT, ENV, EXIT, FC, FIND, FROM, HELP, LIMIT, LOG, NEXT, OBEY, OUT, PAGESIZE, RECORD, and RESET. Those additional commands implemented by the subsystem-specific PTrace modules vary from subsystem to subsystem. Of the commands that fall into the subsystem-dependent category, the Parallel Library TCP/IP subsystem supports the following: DETAIL HEX LABEL OCTAL SELECT TEXT The HEX, OCTAL, and TEXT commands are implemented in the standard manner. The LABEL and SELECT commands vary slightly from the standard, as described later in this section. The following subsystem-dependent commands are not supported: EBCDIC FILTER SETTRANSLATE TEST TRANSLATE Table 5-7 lists and describes all the PTrace commands supported by the Parallel Library TCP/IP subsystem. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -157 SCF Reference for Parallel Library TCP/IP PTrace Commands Table 5-7. Summary of Parallel Library TCP/IP PTrace Commands Command Description ALLOW Specifies the number of errors or warnings permitted during the execution of a command COUNT Counts the records in the trace file DETAIL Turns on the detailed display formatting of records ENV Displays the settings of the PTrace session attributes EXIT Terminates a PTrace session FC Allows correction of the last PTrace command line entered FIND Searches the formatted output for a specified string FROM Specifies the trace file to be displayed HELP Displays information on TRACE commands HEX Sets the hexadecimal display option LABEL Turns on subsystem-controlled formatting and display of trace data LIMIT Limits the number of records displayed by a single command LOG Directs a copy of PTrace input and output to a file NEXT Displays the next trace data record(s) in the file OBEY Causes commands to be read from a different input file OCTAL Sets the octal display option OUT Redirects PTrace output PAGESIZE Sets the terminal screen size for interactive mode RECORD Displays record(s) selected by record number RESET Resets session attributes to their default values SELECT Establishes selection criteria for displaying records TEXT Sets the text display option The remainder of this subsection describes in detail the subsystem-dependent commands supported by the Parallel Library TCP/IP subsystem (DETAIL, HEX, LABEL, OCTAL, SELECT, and TEXT). Each command description includes a brief description of the command, the command's syntax, and any special considerations applicable to the command. The commands are presented in alphabetical order. See the Notation Conventions on page xvii for a description of the notation scheme used here. For information on starting PTrace and entering PTrace commands, and for more detailed descriptions of the standard PTrace commands available to all subsystems, refer to the PTrace Reference Manual. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -158 SCF Reference for Parallel Library TCP/IP DETAIL Command DETAIL Command The DETAIL command controls the detailed display option. When DETAIL is set to ON, PTrace displays extended formatted versions of some records (for example, ARP traffic). Command Syntax DETAIL [ ON | OFF ] ON turns on detailed display mode. OFF turns off detailed display mode. Considerations • The Parallel Library TCP/IP DETAIL command is implemented in the standards defined in the PTrace Reference Manual. • • If the DETAIL command is not used, the OFF attribute is assumed. If DETAIL is specified without the ON or OFF attribute, the ON attribute is assumed. • The RESET and FROM commands set the DETAIL command to OFF. HEX Command The HEX command controls the hexadecimal display option. When HEX is set to ON, PTrace displays a hexadecimal dump of trace-file records (excluding the record header), with character equivalents printed to the right of the dump. Command Syntax HEX [ ON | OFF ] ON turns on hexadecimal display mode. OFF turns off hexadecimal display mode. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -159 LABEL Command SCF Reference for Parallel Library TCP/IP Considerations • The Parallel Library TCP/IP HEX command is implemented in the standards defined in the PTrace Reference Manual. • • • If the HEX command is not used, the OFF attribute is assumed. If HEX is specified without the ON or OFF attribute, the ON attribute is assumed. The RESET and FROM commands set the HEX command to OFF. LABEL Command The LABEL command controls the formatted display of trace records. Command Syntax LABEL [ ON | OFF ] ON turns on the formatted display of trace records. The default value is ON. OFF turns off the formatted display of trace records, but the record header for the trace record is displayed. Note. The LABEL ON command is the only way to display Parallel records. Library TCP/IP trace Considerations • • • If the LABEL command is not used, the ON attribute is assumed. If LABEL is specified without the ON or OFF attribute, the ON attribute is assumed. The RESET and FROM commands set the LABEL command to ON. OCTAL Command The OCTAL command controls the octal display option. When OCTAL is set to ON, PTrace displays an octal dump of trace-file records (excluding the record header), with character equivalents printed to the right of the dump. Command Syntax OCTAL [ ON | OFF ] HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -160 SCF Reference for Parallel Library TCP/IP SELECT Command ON turns on octal display mode. OFF turns off octal display mode. Considerations • • • If the OCTAL command is not used, the OFF attribute is assumed. If OCTAL is specified without the ON or OFF attribute, the ON attribute is assumed. The RESET and FROM commands set the OCTAL command to OFF. SELECT Command The SELECT command establishes the selection criteria that control which trace records are to be displayed. Command Syntax SELECT [ [ [ mask ] keyword ] ( keyword [ , keyword ] ... ) ] mask is a decimal integer that specifies a selection mask. The number is converted into a 32-bit mask and saved as an enumerated value. The acceptable range is 0 through 65535. keyword is a keyword either for the PROCESS object or the SUBNET object. The following keywords apply to the PROCESS object: ALL All records SOCKCMD Socket requests (bind, listen, accept, connect, send) MSGSYS Message system interface MALLOC Resource allocation and deallocation events ROUTING Requests for route changes UDP IDP interface layer HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -161 SCF Reference for Parallel Library TCP/IP SELECT Command TCP Transmission Control Protocol message layer IP IP layer LOGIC Several of the above selections including socket requests (SOCKCMD) and message system interface (MSGSYS) The following keywords apply to the SUBNET object: ALL All records IPI IP input records IPO IP output records ARPI ARP input records ARPO ARP output records LOGIC A combination of all the above records USERDATA Used with IPI and IPO to display user data Considerations • • • • • • • If the SELECT command is not entered, the default mask and keyword is ALL. If the SELECT command is specified with no mask or keywords, the ALL keyword is assumed. The ENV command allows you to see which SELECT keywords are currently being used. PTrace prints the Parallel Library TCP/IP VPROC version of the Parallel Library TCP/IP process that is creating a trace file. Parallel Library TCP/IP collects detailed location debugging information with each trace point. The ENV command allows you to see which SELECT keywords are currently being used. Parallel Library TCP/IP and PTrace collect and decode detailed SOCKET and internal TCP control block information when SOCKCMD or TCP are selected for the PROCESS object. Examples 1. The following command sequence can be used to trace and decode SOCKET command requests and responses during a PROCESS object trace: TRACE PROCESS $ZTC0,TO TRACEFL,RECSIZE 750,SELECT SOCKCMD HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -162 TEXT Command SCF Reference for Parallel Library TCP/IP 2. The following command sequence can be used to trace and decode ARP protocol traffic during a SUBNET object trace: TRACE SUBNET #E1,TO TRACEFL,RECSIZE 500,SELECT (ARPI,ARPO) 3. The following command sequence can be used to trace and decode IP application data (by providing the keyword USERDATA) during a SUBNET object trace: TRACE SUBNET #EN1,TO TRACEFL,RECSIZE 750, SELECT (IPI,IPO,USERDATA) & TEXT Command The TEXT command controls the text display option. When TEXT is set to ON, PTrace displays an interpreted text of trace-file records (excluding the record header). The textual display appears below labeled data, the HEX display, and the OCTAL display, if they are present. The textual display consists of ASCII characters, with control codes represented by two- or three-character mnemonics. Command Syntax TEXT [ ON | OFF ] ON turns on text display mode. OFF turns off text display mode. Considerations • • • • The Parallel Library TCP/IP TEXT command is implemented in the standards defined in the PTrace Reference Manual. If the TEXT command is not used, the OFF attribute is assumed. If TEXT is specified without the ON or OFF attribute, the ON attribute is assumed. The RESET and FROM commands set the TEXT command to OFF. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -163 SCF Reference for Parallel Library TCP/IP Trace Record Formats Trace Record Formats This subsection describes the formatted Parallel Library TCP/IP trace records. The records are presented in alphabetical order under the SELECT keyword used to display them. The SELECT keyword categories are presented in numeric order, based on their record-type code, as follows: Type Record 1 SOCKCR 2 MBUF 3 IPC 4 TCP 5 and 6 UDPI and UDPDI 7 UDPO and UDPDO 9 IPI 10 IPO 11 ROUTE 12 SOCKCMD 13 UDPUREQ Each description includes the time when the record is generated, the record-type code, the text of the record, and the definitions of any values contained in the record. Header Format Each trace record displayed is preceded by a header line having the following format: Date Time timestamp >Delta Time time #Record rec-type Line line-num of file-name (time on date) line-num of file-name (time on date) is the line number that caused the event, the fully-qualified file name, and the last time the file was compiled. rec-no indicates the record number. Records are numbered sequentially based on age. The oldest record in the file (the trace file header record) is record 0. The oldest data record is record 1. The newest record in the file is record number n - 1, where n is the number of records in the file. The letter D following the record number indicates that it is in double-integer format. rec-type indicates the type of trace record. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -164 SCF Reference for Parallel Library TCP/IP Socket Creation Records seq-no indicates the sequence number. The sequence number is included to keep track of records that are lost when the trace file is written to disk. The sequence number counts from 0 to 255 and then begins again. time indicates the time since the last trace run on this line. timestamp indicates the timestamp of the record. The timestamp reports the time at which the record was captured. The resolution is to one hundredth of a second. type indicates the record-type code. The record-type code identifies the type of information contained in the record. It is subsystem-dependent. Socket Creation Records This subsection describes the formatted trace records displayed when the SOCKCR keyword is specified for the PTrace SELECT command. Note that all of the socket creation records are preceded by a header containing the record-type code 1. The records are presented in alphabetical order, based on the text format. Attach Socket Protocol Record The attach socket protocol record is generated when a socket is attached to a protocol. header attach socket_handle nnnnaaaa proto #n nnnnaaaa indicates the internal socket ID of the socket being attached to a protocol. n indicates the IP number of the protocol being attached to the socket. For a list of commonly used IP numbers, refer to the TCP/IP and TCP/IPv6 Programming Manual. For a complete list of the IP numbers, refer to Request for Comments document 1010, Assigned Numbers. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -165 SCF Reference for Parallel Library TCP/IP Socket Creation Records Soclose Record The soclose record is generated each time the SOCLOSE procedure is called. The SOCLOSE procedure completes the close of a socket. header procedure:soclose socket_handle nnnnaaaa line-num of file-name (time on date) is the edit-line number that caused the event, the fully-qualified edit-file name, and the last time the edit file was compiled. nnnnaaaa indicates the internal socket ID of the socket being closed. Sofree Record The sofree record is generated each time the SOFREE procedure is called. The SOFREE procedure frees up a socket data structure. header procedure:sofree freeing socket_handle nnnaaa nnnnaaaa indicates the internal ID of the socket being freed. Socket Closing Record The socket closing record is generated when the process initiates the close of a socket. header socket closing socket_handle nnnnaaaa nnnnaaaa indicates the internal ID of the socket being closed. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -166 SCF Reference for Parallel Library TCP/IP Socket Creation Records Allocating PCB Record The allocating PCB record is generated each time a protocol control block (PCB) is allocated for a TCP socket. header socket_handle nnnnaaaa allocating PCB for TCP socket nnnnaaaa indicates the internal ID of the socket for which the PCB is being allocated. Can’t Create New TCPCB Record The can't create new TCPCB record is generated each time the Parallel Library TCP/IP process can't create a new control block for a TCP socket. header socket_handle nnnnaaaa can't create new tcpcb for TCP socket nnnnaaaa indicates the internal ID of the socket for which the new control block could not be created. Creating New TCPCB Record The creating new TCPCB record is generated each time a new control block is created for a TCP socket. header socket_handle nnnnaaaa creating new tcpcb for TCP socket nnnnaaaa indicates the internal ID of the socket for which the new control block is being created. Reserve Space Record The reserve space record is generated each time a socket structure needs to be created for an incoming TCP connection. header socket_handle nnnnaaaa reserve space for incoming connection HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -167 SCF Reference for Parallel Library TCP/IP Memory Buffer Allocation Records nnnnaaaa indicates the internal ID of the socket being reserved. Memory Buffer Allocation Records This subsection describes the formatted trace records displayed when the MBUF keyword is specified for the PTrace SELECT command. Note that there is only one memory buffer allocation record and that each memory buffer allocation record in the trace file is preceded by a header containing the record-type code 2. Memory Buffer Allocation Record The memory buffer allocation record is generated each time the Parallel Library TCP/IP process attempts to allocate memory buffers (MBUFs). Note that this record is generated even when the allocation attempt fails. header m_mbufalloc nnnnnnnnnn bytes result: rrrrrrr nnnnnnnnnn indicates the number of bytes allocated. rrrrrrr indicates whether the memory allocation attempt succeeded or not. The value can be succeed or failed. Interprocess Communication Records This subsection describes the formatted trace records displayed when the IPC keyword is specified for the PTrace SELECT command. Note that there is only one interprocess communication record and that each socket system call record in the trace file is preceded by a header containing the record-type code 3. Socket System Call Record The socket system call record is generated each time a socket call is made by an application-level program. header socket sys call #c socket_handle nnnnaaaa received bbb bytes c indicates the socket call number being invoked. The socket call number is described in the SYSCALH INCLUDE file. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -168 SCF Reference for Parallel Library TCP/IP TCP Records nnnnaaaa indicates the internal ID of the socket to which the system call applies. bbb indicates the number of bytes of data received in the socket call. TCP Records This subsection describes the formatted trace records displayed when the TCP keyword is specified for the PTrace SELECT command. Note that TCP records are preceded by a header containing the record-type code 4. The records are presented in alphabetical order, based on their text format. Data Acked Record The data acked record is generated each time an ACK is received for the local socket. header socket_handle nnnnaaaa: acked ack-bytes, sb_cc unack_bytes nnnnaaaa indicates the internal socket ID. ack-bytes indicates the number of bytes of data acknowledged. unack-bytes indicates the number of bytes of data in the queue waiting to be acknowledged. All Data Acked Record The all data acked record is generated each time all of the data in the queue has been acknowledged. header socket_handle nnnnaaaa: acked ack-bytes > sb_cc unack-bytes, all data acked nnnnaaaa indicates the internal socket ID. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -169 SCF Reference for Parallel Library TCP/IP TCP Records ack-bytes indicates the number of bytes of data acknowledged. unack-bytes indicates the number of bytes of data in the queue waiting to be acknowledged. Because the number of bytes acknowledged is greater than this value (>), all of the data in the queue has been acknowledged. After Changes Record The after changes record is generated each time data or an ACK is received for a TCP socket. Note that the values reported indicate the values of these variables after they have been updated by the packet. The preliminary values are reported in the send next record. header socket_handle nnnnaaaa: After Changes: snd_nxt snd-nxt, snd_una snd-una, snd_max snd-max nnnnaaaa indicates the internal socket ID. snd-nxt indicates the next sequence number to be sent. snd-una indicates the oldest unacknowledged sequence number. snd-max indicates the maximum sequence number that can be sent. Send Next Record The send next record is generated each time data or an ACK is received for a TCP socket. Note that the values reported indicate the values of these variables before they have been updated by the packet. The updated values are reported in the after changes record. header socket_handle nnnnaaaa: snd_nxt snd-nxt, snd_una snd-una ti_ack ti-ack HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -170 SCF Reference for Parallel Library TCP/IP TCP Records nnnnaaaa indicates the internal socket ID. snd-nxt indicates the next sequence number to be sent. snd-una indicates the oldest unacknowledged sequence number. ti-ack indicates the sequence number of the data currently being acknowledged. Receive State Change Record The receive send state change record is generated when data is received. header socket_handle nnnnaaaa tcp_handle nnnnn init-state: input (start-no..end-no) @ ack-no, urp=urp [f1,f2,f3,f4,f5,f6] -> fin-state... rcv_(nxt,wnd,up) (rcv-nxt, rcv-wnd, rcv-up) snd_(una,nxt,max) (snd-una, snd-nxt, snd-max) snd_(wl1,wl2,wnd) (snd-wl1, snd-wl2, snd-wnd) nnnnaaaa indicates the internal socket ID. nnnnn indicates the internal ID of the TCP packet. init-state indicates the initial state before the data was received. The possible states are: CLOSE-WAIT LAST-ACK CLOSED LISTEN CLOSING SYN-RECVD ESTABLISHED SYN-SENT FIN-WAIT-1 TIME-WAIT FIN-WAIT-2 start-no indicates the starting sequence number of the data received. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -171 SCF Reference for Parallel Library TCP/IP TCP Records end-no indicates the ending sequence number of the data received. ack-no indicates the acknowledgment number. urp indicates the urgent pointer. [f1,f2,f3,f4,f5,f6] indicates the control flags set. The possible flags that can be set are SYN, ACK, FIN, RST, PUSH, and URG. fin-state indicates the final state after the data was received. The possible states are: CLOSE-WAIT LAST-ACK CLOSED LISTEN CLOSING SYN-RECVD ESTABLISHED SYN-SENT FIN-WAIT-1 TIME-WAIT FIN-WAIT-2 rcv-nxt indicates the next sequence number expected to be received. rcv-wnd indicates the receive window. rcv-up indicates the receive urgent pointer. snd-una indicates the oldest unacknowledged sequence number. snd-nxt indicates the next sequence number to be sent. snd-max indicates the maximum sequence number that can be sent. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -172 SCF Reference for Parallel Library TCP/IP TCP Records snd-wl1 indicates the sequence number used for the last window update. snd-wl2 indicates the acknowledgment number used for the last window update. snd-wnd indicates the send window. Send State Change Record The send state change record is generated when a user sends data. header socket_handle nnnnaaaa tcp_handle nnnnn init-state: user req-type -> fin-state... rcv_(nxt,wnd,up) (rcv-nxt, rcv-wnd, rcv-up) snd_(una,nxt,max) (snd-una, snd-nxt, snd-max) snd_(wl1,wl2,wnd) (snd-wl1, snd-wl2, snd-wnd) nnnnaaaa indicates the internal socket ID. nnnnn indicates the internal ID of the TCP packet. init-state indicates the initial state before the data was sent. The possible states are: CLOSE-WAIT LAST-ACK CLOSED LISTEN CLOSING SYN-RECVD ESTABLISHED SYN-SENT FIN-WAIT-1 TIME-WAIT FIN-WAIT-2 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -173 SCF Reference for Parallel Library TCP/IP TCP Records req-type indicates the request type. The possible request types are: ABORT PEERADDR ACCEPT PROTORCV ATTACH PROTOSEND BIND RCVD CONNECT RCVOOB CONNECT2 SEND CONTROL SENDOOB DETACH SENSE DISCONNECT SHUTDOWN FASTIMO SLOWTIMO LISTEN SOCKADDR fin-state indicates the final state after the data was sent. The possible states are: CLOSE-WAIT LAST-ACK CLOSED LISTEN CLOSING SYN-RECVD ESTABLISHED SYN-SENT FIN-WAIT-1 TIME-WAIT FIN-WAIT-2 rcv-nxt indicates the next sequence number expected to be received. rcv-wnd indicates the receive window. rcv-up indicates the receive urgent pointer. snd-una indicates the oldest unacknowledged sequence number. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -174 SCF Reference for Parallel Library TCP/IP TCP Records snd-nxt indicates the next sequence number to be sent. snd-max indicates the maximum sequence number that can be sent. snd-wl1 indicates the sequence number used for the last window update. snd-wl2 indicates the acknowledgment number used for the last window update. snd-wnd indicates the send window. Accepting Connection Record The accepting connection record is generated each time an incoming connection is accepted on a local socket. header socket_handle nnnnaaaa: tcp_usrreq: PRU_ACCEPT faddr forgn-addr fport forgn-port nnnnaaaa indicates the internal socket ID. forgn-addr indicates the remote Internet address associated with the incoming connection. forgn-port indicates the remote port number associated with the incoming connection. Incoming Connection Record The incoming connection record is generated each time an incoming connection request is received on a local socket. header socket_handle nnnnaaaa: tcp_usrreq: PRU_CONNIND faddr forgn-addr fport forgn-port HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -175 SCF Reference for Parallel Library TCP/IP UDP Input Records nnnnaaaa indicates the internal socket ID. forgn-addr indicates the remote Internet address associated with the incoming connection. forgn-port indicates the remote port number associated with the incoming connection. TCP Socket Request Record The TCP socket request record is generated each time a TCP socket request is made. header socket_handle nnnnaaaa: tcp_usrreq: socket request #nnnnn nnnnaaaa indicates the internal socket ID. nnnnn indicates the internal request number used to manipulate the TCP socket. The possible values that can appear and their meanings are discussed in the PROTOSWH INCLUDE file. UDP Input Records This subsection describes the formatted trace records displayed when the UDPI keyword is specified for the PTrace SELECT command. Note that UDP input records are preceded by a header containing the record-type code 5 or 6. The records are presented in alphabetical order, based on their text format. Received UDP Packet Record The received UDP packet record is generated each time the UDP input routine is executed. This record is preceded by a header containing the record-type code 6. header Received UDP packet for udp_header_handle nnnnaaaa nnnnaaaa indicates the internal ID of the UDP packet. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -176 SCF Reference for Parallel Library TCP/IP Detailed UDP Input Records Sent UDP Packet to User Record The sent UDP packet to user record is generated each time a valid user is identified for an incoming UDP packet and the packet is delivered to the user. This record is preceded by a header containing the record-type code 5. header udp_input: Sent UDP packet to user --> udp_header_handle nnnnaaaa nnnnaaaa indicates the internal ID of the UDP packet. Detailed UDP Input Records This subsection describes the formatted trace records displayed when the UDPDI keyword is specified for the PTrace SELECT command. Note that detailed UDP input records are preceded by a header containing the record-type code 5 or 6. The records are presented in alphabetical order, based on their text format. Destination Address and Port Record The destination address and port record is generated each time a UDP packet is received. This record is preceded by a header containing the record-type code 5. header udp_input: dst dst-addr, dport port-no udp_header_handle nnnnaaaa dest-addr indicates the packet's destination Internet address. port-no indicates the packet's destination UDP port number. nnnnaaaa indicates the internal ID of the UDP packet. Packet Length Record The packet length record is generated each time a UDP packet is received. This record is preceded by a header containing the record-type code 6. header udp_input: packetlen lllll udp_header_handle nnnnaaaa HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -177 UDP Output Records SCF Reference for Parallel Library TCP/IP lllll indicates the packet's length. nnnnaaaa indicates the internal ID of the UDP packet. Source Address and Port Record The source address and port record is generated each time a UDP packet is received. This record is preceded by a header containing the record-type code 5. header udp_input: src ip-addr, sport portno udp_header_handle nnnnaaaa ip-addr indicates the packet's source Internet address. portno indicates the packet's source UDP port number. nnnnaaaa indicates the internal ID of the UDP packet. UDP Output Records This subsection describes the formatted trace records displayed when either the UDPO or UDPDO keyword is specified for the PTrace SELECT command. Note that UDP output records are preceded by a header containing the record-type code 7. The records are presented in alphabetical order, based on their text format. UDP Sending From Record The UDP sending from record is generated each time the Parallel Library TCP/IP process sends a packet. header udp_output: sending from ip-addr.udp-port ip-addr indicates the source Internet address. udp-port indicates the source UDP port number. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -178 SCF Reference for Parallel Library TCP/IP IP Input Records UDP Sending to Record The UDP sending to record is generated each time the Parallel Library TCP/IP process sends a packet. header udp_output: sending to ip-addr.udp-port ip-addr indicates the destination IP address. udp-port indicates the destination UDP port number. IP Input Records This subsection describes the formatted trace records displayed when the IPI keyword is specified for the PTrace SELECT command. Note that IP input records are preceded by a header containing the record-type code 9. The records are presented in alphabetical order, based on their text format. Sending ICMP Error Record The sending ICMP error record is generated each time the IP detects an error and requests the generation of an ICMP error packet. header ip_forward: ip_handle nnnnaaaa sending icmp error dst 1234cccc, code ptype nnnnaaaa indicates the internal ID of the IP packet. 1234cccc indicates the destination Internet address in the packet containing the error. ptype indicates the type of ICMP packet requested. The Parallel Library TCP/IP subsystem supports the following packet types and packet-type codes: Echo Reply (0) Destination Unreachable (3) Source Quench (4) Redirect (5) Echo (8) Time Exceeded (11) HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -179 SCF Reference for Parallel Library TCP/IP IP Input Records w Problem (12) Timestamp (13) Timestamp Reply (14) Information Request (15) Information Reply (16) Forwarding to IP Address Record The forwarding to IP address record is generated each time the IP input routines receive a packet destined for another destination. header ipintr: ip_handle nnnnaaaa forwarding to ip address ip-addr nnnnaaaa indicates the internal ID of the IP packet. ip-addr indicates the address to which the packet is forwarded. Got Fragment Record The got fragment record is generated each time the IP input routines receive a packet fragment. header ipintr: ip_handle nnnnaaaa got fragment offset bbbbb nnnnaaaa indicates the internal ID of the IP packet. bbbbb indicates the IP offset (in bytes). Packet for Us Record The packet for us record is generated each time the IP input routines receive a packet destined for this address. header ipintr: ip_handle nnnnaaaa packet for us, proto #nnnnn nnnnaaaa indicates the internal ID of the IP packet. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -180 SCF Reference for Parallel Library TCP/IP IP Output Records nnnnn indicates the IP protocol number (either 6 for TCP or 17 for UDP). Rebuilt Fragment Record The rebuilt fragment record is generated each time the IP input routines rebuild a packet from packet fragments. header ipintr: ip_handle nnnnaaaa rebuilt fragment len lllll nnnnaaaa indicates the internal ID of the IP packet. lllll indicates the rebuilt packet's total length. Message Buffer Length Record The message buffer length record is generated each time the IP input routines are executed. header ipintr: mbuflen lllll lllll indicates the length of the packet received. IP Output Records This subsection describes the formatted trace records displayed when the IPO keyword is specified for the PTrace SELECT command. Note that IP output records are preceded by a header containing the record-type code 10. The records are presented in alphabetical order, based on their text format. Destination IP Address Record The destination IP address record is generated each time the IP sends a packet with a standard destination address. header ip_output: dest ip address ip-addr, proto ppppp ip-addr indicates the destination IP address. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -181 SCF Reference for Parallel Library TCP/IP Route Records ppppp indicates the IP number associated with the packet sent. For a list of the commonly used IP numbers, refer to the TCP/IP and TCP/IPv6 Programming Manual For a complete list of the IP numbers, refer to Request for Comments document 1010, Assigned Numbers. Fragmenting Record The fragmenting record is generated each time the IP must fragment a packet. header ip_output: fragmenting offset bbbbb bbbbb indicates the IP offset of the fragments (in bytes). Sending Broadcast Record The sending broadcast record is generated each time the IP sends a packet with a broadcast address. header ip_output: sending broadcast len lllll lllll indicates the length of the broadcast packet sent. Route Records This subsection describes the formatted trace records displayed when the ROUTE keyword is specified for the PTrace SELECT command. Note that route records are preceded by a header containing the record-type code 11. The records are presented in alphabetical order, based on their text format. Flags Record The flags record is generated each time a route change request is received. header flags ffff ffff indicates the internal flags set during the routing change. The value displayed represents a bit pattern in which bit 0 is the low-order bit and bits 1 through 5 correspond to the following flags: bit 1 indicates whether the route is UP, bit 2 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -182 SCF Reference for Parallel Library TCP/IP Socket Command Records indicates whether the route is to a gateway, bit 3 indicates whether the route is to a point-to-point connection, bit 4 indicates whether the route is marked down, and bit 5 indicates whether the route is a dynamic route. Route Addition Record The route addition record is generated each time a route is added. Note that this record does not return any values. header req SIOCADDRT Route Deletion Record The route deletion record is generated each time a route is deleted. Note that this record does not return any values. header req SIOCDELRT Route Request Record The route request record is generated each time a route change request is received. header rtreq: dst dst-addr, gateway gw-addr, flags 0 dst-addr indicates the destination address associated with the route change request. gw-addr indicates the gateway address associated with the route change request. Socket Command Records This subsection describes the formatted trace records displayed when the SOCKCMD keyword is specified for the PTrace SELECT command. Note that socket command records are preceded by a header containing the record-type code 12. The records are presented in alphabetical order, based on their text format. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -183 SCF Reference for Parallel Library TCP/IP Socket Command Records Accept Record The accept record is generated each time a connection is accepted on the local socket. header accept: socket_handle nnnnaaaa connection on 1234abcd.12345 nnnnaaaa indicates the internal socket ID. 1234abcd.12345 indicates the remote IP address and port number. Address Family Record The address family record is generated each time a connection request is received on the local socket. header AF fffff fffff indicates the address family for the new connection. Bind Record The bind record is generated each time a name (consisting of a local Internet address and port number) is bound to a socket. header bind: socket_handle nnnnaaaa, port ppppp, local_addr loc-addr nnnnaaaa indicates the internal socket ID. ppppp indicates the local port number to be associated with the socket. loc-addr indicates the local Internet address to be associated with the socket. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -184 SCF Reference for Parallel Library TCP/IP Socket Command Records Connection Request Record The connection request record is generated each time a connection request is received on the local socket. header connect: socket_handle nnnnaaaa, to address 1234abcd.12345 nnnnaaaa indicates the internal socket ID. 1234abcd.12345 indicates the remote IP address and port number. Connection Waiting Record The connection waiting record is generated each time the socket has to wait for a connection to complete. header connect: waiting socket_handle nnnnaaaa nnnnaaaa indicates the internal socket ID. Queue Length Record The queue length record is generated when a listen call is made. header listen: socket_handle nnnnaaaa qlen lllll nnnnaaaa indicates the internal socket ID. lllll indicates the maximum queue length of pending TCP connections on the socket. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -185 SCF Reference for Parallel Library TCP/IP Socket Command Records Waiting for Reply Record The waiting for reply record is generated each time an accept call is not completed immediately (that is, if the socket has to wait for an incoming connection). header listen: socket_handle nnnnaaaa waiting for reply nnnnaaaa indicates the internal socket ID. Send Record The send record is generated each time a send call is made. header send: socket_handle nnnnaaaa bbbbb nnnnaaaa indicates the internal socket ID. bbbbb is number of bytes transferred. Send to Record The send to record is generated each time a sendto call is made. header sendto: socket_handle nnnnaaaa bbbbb, to address 1234abcd.1234 nnnnaaaa indicates the internal socket ID. bbbbb is number of bytes transferred. 1234abcd,12345 indicates the remote IP address and port number. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -186 SCF Reference for Parallel Library TCP/IP UDP User Request Records Socket Family Record The socket family record is generated each time a socket is created. header sock_reply: family fffff, type ttttt, proto proto fffff indicates the address family specified by the programmer in the socket call. ttttt indicates the socket type specified by the programmer in the socket call. proto indicates the IP number specified by the programmer in the socket call (either 0 for IP, 6 for TCP, or 17 for UDP). Socket Reply Record The socket reply record is generated each time a socket request is completed. header sock_reply: socket_handle nnnnaaaa nnnnaaaa indicates the internal socket ID. UDP User Request Records This subsection describes the formatted trace records displayed when the UDPUREQ keyword is specified for the PTrace SELECT command. Note that UDP user request records are preceded by a header containing the record-type code 13. The records are presented in alphabetical order, based on their text format. Socket Request Record The socket request record is generated each time a UDP socket request is made. header udp_usrreq: socket_handle nnnnaaaa, socket request #nnnnn nnnnaaaa indicates the internal socket ID. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -187 SCF Reference for Parallel Library TCP/IP UDP User Request Records nnnnn indicates the internal request number used to manipulate the UDP socket. The possible values that can appear and their meanings are explained in the PROTOSWH INCLUDE file. UDP Socket Request Completed Record The UDP socket request completed record is generated each time a UDP socket request is completed with an error. header udp_usrreq: socket request nnnnn completed with error err-no nnnnn indicates the internal request number associated with the UDP socket request. The possible values that can appear and their meanings are explained in the PROTOSWH INCLUDE file. err-no indicates the error code returned as the result of the socket request. For descriptions of the error codes returned, refer to the TCP/IP and TCP/IPv6 Programming Manual. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 5 -188 6 Troubleshooting Tips This section provides some conditions to check if you encounter some problems with your Parallel Library TCP/IP configuration. Review the following list for suggestions that may pertain to your configuration: • • • Check the adapter configuration and ensure that the SACs are configured with the correct Access List. Make sure that all processors running a TCPMON process are listed in the Access List. Ensure that you have cleared the system of all DEFINEs and PARAMs before adding new ones. Ensure that you add a DEFINE and PARAM to specify a TCPSAM name before starting your LISTNER, TELSERV, or other client application. The following examples show the DEFINE and PARAM TACL commands: ADD DEFINE =TCPIP^PROCESS^NAME, FILE $ZSAM0 PARAM TCPIP^PROCESS^NAME $ZSAM0 • • • Ensure that you add the DEFINE to specify the location of the SRL before starting the TCPSAM process. See the ADD DEFINE =_SRL command in Example 1-1 on page 1-8. Ensure that you set your home terminal for the TCPMAN process to $ZHOME. Check your TCPMAN process creation script to ensure that TERM $ZHOME (and OUT $ZHOME) are included. You should also add these attributes to the TCPSAM, LISTNER, and TELSERV process startup commands. If you have a large configuration that has many LISTNERs and if you want to use different TCPSAM processes for each of them, make sure you delete and add DEFINEs and PARAMs for the TCPSAM process that will be associated with each LISTNER. For an example of deleting the DEFINEs and PARAMs, see the Considerations on page 5-73 in the START MON Command for TCPMAN on page 5-72. Note. Only one TCPSAM process is needed for all the applications in the system; creating more TCPSAM processes does not provide more bandwidth. If you use only one TCPSAM process for all the LISTNERs in the system, you only need to delete and add the DEFINE for TCPIP^PROCESS^NAME once. • • You must wait for all the TCPMONs to start before starting TCPSAM. Check your configuration scripts to ensure that a DELAY command exists after starting the TCPMONs. See the DELAY command in Example 1-1 on page 1-8. When you configure a set of listening processes for round robin, do not allow their primary and backup processors to overlap. That is, if you configure primary and backup listening processes, do so in distinct pairs. For example, if you have four processors, 0 through 3, and you want to configure primary and backup TELSERV processes for round-robin distribution, configure a primary and backup TELSERV HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 6 -1 Troubleshooting Tips pair in processors 0 and 1 and another primary and backup TELSERV pair in processors 2 and 3. • • • • • • The TCPSAM process must be in a processor that has a TCPMON running. Check to ensure that all primary and backup TCPSAM processes are configured in processors which contain a running TCPMON. If you replace the ZTCPSRL file while you have the TCPMAN -- or any TCPMONs or TCPSAMs -- running and a processor stops, you get Error 48: “PTCPIP: TCPMON can’t be started after CPU failure.” Before replacing the ZTCPSRL file you must stop the Parallel Library TCP/IP environment. If you receive Error 48 in this circumstance, stop the Parallel Library TCP/IP subsystem then restart it. (See Section 1, Configuration Quick Start for shut-down procedures.) If you have TCPMAN ($ZZTCP) configured as a generic process, make sure that its STOPMODE attribute is set to SYSMSG. For more information, see Managing Persistence on page 4-3 and Example 4-2 on page 4-5. If you are migrating from a NonStop K-series system where you are using multiple Telserv processes on each 3615 communication controller with ports 23, 2 and 3, you cannot create the same configuration in Parallel Library TCP/IP because Parallel Library TCP/IP does not allow sharing of ports between processes. However, you can create a similar configuration in the Parallel Library TCP/IP environment by enabling round-robin processing. In this case, on the NonStop Sseries system, you create multiple Telserv processes in multiple processors and enable round-robin filtering in the Parallel Library TCP/IP environment, (see Round-Robin Filtering on page 2-4 for round-robin filter-enabling procedures); then the Telserv processes can share all configured ports. If you configure Telserv in this manner, remember to run the process pairs in alternative processor pairs (see Port Collision Considerations for Listening Processes on page 2-5). When Telserv is running as a fault-tolerant process pair in the Parallel Library TCP/IP environment, all Telserv STATIC service and STATIC window information is captured and retained by the backup Telserv process when the primary Telserv fails. However, when you run multiple Telserv processes not fault-tolerant process pairs with round-robin filtering enabled, you can configure the persistence manager to restart Telserv. In this case, when a failed processor is restarted, the persistence manager can execute an automatic script to re-launch the Telserv process. However, the primary Telserv process launched by the persistence manager does not have the knowledge of the STATIC service and STATIC window information. To avoid or fix this problem, modify the Telserv startup script that the persistence manager launches upon processor restart so that the script re-configures the Telserv STATIC service and STATIC window. TCPMAN start up changes the fingerprint of the ZTCPSRL file. When DSM/SCM validates the fingerprints of all the SRLs (during a BUILD/APPLY for any SPR update), it notices that the fingerprint of ZTCPSRL differs from the version in the archive. Hence, TCPMAN renames ZTCPSRL to a fictitious name, and replaces HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 6 -2 Troubleshooting Tips ZTCPSRL with the original version from the archive. The DSM/SCM ZPHIRNM log file shows the fictitious file name. This replacement does not affect the subsystem. However, if after the replacement of ZTCPSRL by DSM/SCM, either TCPSAM alone or TCPMON alone is restarted, a mismatch occurs in the version of ZTCPSRL that the TCPSAM or TCPMON is bound to. This situation could lead to a malfunction in the subsystem. Hence, you must rename the fictitious file as ZTCPSRL after a DSC/SCM BUILD/APPLY step. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 6 -3 Troubleshooting Tips HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 6 -4 A SCF Command Summary ABORT [ / OUT file-spec / ] MON [$ZZTCP.#ZPTMn] ABORT [ / OUT file-spec / ] [ PROCESS $ZZTCP ] [ , SUB ALL ] ABORT [ / OUT file-spec / ] [ PROCESS $tcpsam-process-name ] ABORT [ / OUT file-spec / ] [ROUTE $ZZTCP.*.route-name ] ABORT [ / OUT file-spec / ] [SUBNET $ZZTCP.*.subnet-name] ADD [ /OUT file-spec/ ] [ ENTRY $ZZTCP.*.entry-name ] , TYPE ARP , IPADDRESS ip-addr , MACADDR mac-address ADD [ / OUT file-spec / ] [ ROUTE $ZZTCP.*.route-name ] , DESTINATION destination-ip-address , GATEWAY gateway-ip-address [ , DESTTYPE { HOST | BROADCAST } ] [ , NETMASK mask-val ] [ , METRIC metric-val ] [ , CLONING { ON | OFF } [ , GENMASK mask-val ] [ , SUBNET subnet-name ] HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 A- 1 SCF Command Summary ADD [ /OUT file-spec/ ] [ SUBNET $ZZTCP.*.subnet-name ] , TYPE ETHERNET , DEVICENAME lif-name , IPADDRESS ip-addr [ , IRDP { ON | OFF } ] [ , SUBNETMASK mask-val ] [ , FAILOVER {SHAREDIP | NON-SHAREDIP} ] ALTER [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ [ /OUT file-spec/ ] MON $ZZTCP.* ,TCPSENDSPACE int ] ,TCPRECVSPACE int ] ,UDPSENDSPACE int ] ,UDPRECVSPACE int ] ,DELAYACKS { ON | OFF } ] ,DELAYACKSTIME int ] ,HOSTNAME string ] ,HOSTID int ] ,TCPKEEPIDLE int ] ,TCPKEEPINTVL int ] ,TCPKEEPCNT int ] ,DEBUG { ON | OFF } ] ,FULLDUMP { ON | OFF } ] ,ALLNETSARELOCAL { ON | OFF } ] ,TCPCOMPAT42 { ON | OFF } ] ,EXPANDSECURITY { ON | OFF } ] ,TCPPATHMTU { ON | OFF } ] ,TCPTIMEWAIT int ] ,RFC1323-ENABLE { ON | OFF } ] ,TCP-INIT-REXMIT-TIMEOUT int ] ,TCP-MIN-REXMIT-TIMEOUT int ] ,TCP-LISTEN-QUE-MIN int ] ,INITIAL-TTL int ] ,MIN-EPHEMERAL-PORT int ] ,MAX-EPHEMERAL-PORT int ] HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 A- 2 SCF Command Summary ALTER [ /OUT file-spec/ ] [SUBNET $ZZTCP.*.subnet-name ] { [ ,IPADDRESS ip-addr ] [ ,SUBNETMASK %H0..FFFFFFFF ] [ ,IRDP { ON | OFF } ] [ ,ADDALIAS ip-addr,SUBNETMASK %H0..FFFFFFFF ] [ ,DELETEALIAS ip-addr ] } | { [ ,ASSOCIATESUB "subnet-name" ] [, RESERVEDIP ip-addr] } DELETE [ /OUT file-spec/ ] [ ENTRY $ZZTCP.*.entry-name ] DELETE [ / OUT file-spec / ] [ROUTE $ZZTCP.#ZPTMn.route-name] DELETE [/ OUT file-spec / ] [ SUBNET $ZZTCP.*.subnet-name] INFO [ /OUT file-spec/ ] [ ENTRY $ZZTCP.#ZPTMn.entry-name ] [ , IPADDRESS ip-addr | , OBEYFORM] INFO[ /OUT file-spec/] [ MON [, DETAIL | OBEYFORM] $ZZTCP.#ZPTMn ] INFO [ / OUT file-spec / ] [ PROCESS $ZZTCP ] [ , DETAIL ] INFO [ / OUT file-spec / ] [ PROCESS tcpsam-name ] [, DETAIL] HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 A- 3 SCF Command Summary INFO [ / OUT file-spec / ] [ ROUTE $ZZTCP.#ZPTMn.route-name | OBEYFORM ] INFO [ /OUT file-spec/ ] [ ROUTE $tcpsam-name.route-name ] INFO [ / OUT file-spec / ] [SUBNET $ZZTCP.#ZPTMn.subnet-name] [, DETAIL | , OBEYFORM] INFO [ / OUT file-spec / ] [ SUBNET $tcpsam-name.subnet-name ] [, DETAIL ] LISTOPENS[ /OUT file-spec/ ] [ MON $ZZTCP.#ZPTM{0-F } ] [,DETAIL ] LISTOPENS[ /OUT file-spec/ ] [ PROCESS $tcpsam-name ] [,DETAIL ] NAMES [ /OUT file-spec/ ] [ ENTRY $ZZTCP.#ZPTMn.* ] NAMES [ / OUT file-spec / ] [ROUTE $ZZTCP.#ZPTMn.* ] NAMES [ / OUT file-spec / ] [ROUTE $tcpsam-name.*] NAMES [ / OUT file-spec / ] [ SUBNET $ZZTCP.#ZPTMn.* ] HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 A- 4 SCF Command Summary NAMES [ / OUT file-spec / ] [SUBNET $tcpsam-name.*] PRIMARY [ / OUT file-spec / ] [ PROCESS $ZZTCP ] , CPU cpu-number PRIMARY [ / OUT file-spec / ] [ PROCESS $tcpsam-name ] , CPU cpu-number START [ / OUT file-spec / ] MON $ZZTCP.#ZPTM{0-F } START [ / OUT file-spec / ] [ROUTE $ZZTCP.*.route-name ] START [ / OUT file-spec / ] [SUBNET $ZZTCP.#ZPTMn.subnet-name] STATS [ / OUT file-spec / ] [MON $ZZTCP.#ZPTMn.mon-name] [ , RESET ] STATS [ / OUT file-spec / ] [PROCESS $tcpsam-name] [ , RESET ] STATS [ / OUT file-spec / ] [ROUTE $ZZTCP.#ZPTMn.route-name] [, RESET ] STATS [ / OUT file-spec / ] [ROUTE $tcpsam-name.route-name] [, RESET ] HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 A- 5 SCF Command Summary STATS [ / OUT file-spec / ] [SUBNET $ZZTCP.#ZPTMn.subnet-name] [ , RESET ] [ , DETAIL ] STATS [ / OUT file-spec / ] [SUBNET $tcpsam-process.subnet-name] [ , RESET ] [ , DETAIL ] STATUS [ / OUT file-spec / ] [ ENTRY $ZZTCP.#ZPTMn.entry-name] STATUS [ / OUT file spec / ] [ MON $ZZTCP.#ZPTMn ] [ , DETAIL ] STATUS [ / OUT file spec / ] [ PROCESS $ZZTCP ] STATUS [ / OUT file spec / ] [ PROCESS $tcpsam-name ] [, DETAIL] STATUS [ / OUT file spec / ] [ROUTE $ZZTCP.#ZPTMn.route-name] STATUS [ / OUT file spec / ] [ROUTE $tcpsam-name.#route-name ] STATUS [ / OUT file spec / ] [SUBNET $ZZTCP.#ZPTMn.subnet-name] HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 A- 6 SCF Command Summary STATUS [ / OUT file spec / ] [SUBNET $tcpsam-name.#subnet-name] STOP [ /OUT file-spec/ ] MON $ZZTCP.#ZPTMn STOP [ / OUT file-spec / ] [ PROCESS $ZZTCP ] [, SUB ALL ] STOP [ / OUT file-spec / ] [ PROCESS $tcpsam-name ] STOP [ / OUT file-spec / ] [ROUTE $ZZTCP.#ZPTMn.route-name ] STOP [ / OUT file-spec / ] [SUBNET $ZZTCP.#ZPTMn.subnet-name] TRACE [ /OUT file-spec/ ] MON [ $ZZTCP.#ZPTMn ] {, STOP } | , TO file-spec , NOBULKIO} [ [ [ [ [ , , , , , COUNT count NOCOLL RECSIZE size SELECT select-spec PAGES pages TRACE [ /OUT file-spec/ ] PROCESS $ZZTCP { , STOP [ , BACKUP ] } | { [ , TO file-spec [ , BACKUP ] [ , COUNT count ] [ , NOCOLL ] [ , RECSIZE size ] [ , SELECT select-spec ] [ , PAGES pages ] ] ] ] ] ] ] HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 A- 7 SCF Command Summary TRACE [ /OUT file-spec/ ] PROCESS $tcpsam-name { , STOP [ , BACKUP ] } | { [ , TO file-spec [ , BACKUP ] [ , COUNT count ] [ , NOCOLL ] [ , RECSIZE size ] [ , PAGES pages ] ] TRACE [/OUT file-spec/] [SUBNET $ZZTCP.#ZPTMn.subnet-name] {, STOP } | {, TO file-spec [ [ [ [ [ , NOBULKIO , COUNT count , NOCOLL , RECSIZE size , SELECT select-spec , PAGES pages ] ] ] ] ] VERSION [ /OUT file-spec/ ] [ MON $ZZTCP.#ZPTMn.mon-name ] [, DETAIL ] VERSION [ / OUT file-spec / ] [ PROCESS $ZZTCP ] [ , DETAIL ] VERSION [ /OUT file-spec/ ] [ PROCESS $tcpsam-name ] [, DETAIL ] HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 A- 8 B SCF Error Messages This appendix contains a description of the PTCPIP subsystem SCF error messages. For the operator display of event messages, see the Operator Messages Manual. PTCPIP 00001 PTCPIP 00001 Invalid file name. Cause. You specified a file with an invalid format. Effect. The command is not executed. Recovery. Verify the file-name format and retry the command. PTCPIP 00002 PTCPIP 00002 INTERNAL ERROR: Case value out of range. Cause. An invalid case value was generated, with no associated case label. Effect. The SCF command you entered is not executed. Recovery. Send complete error information to your Global Customer Support Center analyst for analysis. PTCPIP 00003 PTCPIP 00003 Missing Attribute. Cause. You omitted a required attribute from the command. Effect. The command is not executed. Recovery. Verify that the required attribute has been included and retry the command. PTCPIP 00004 PTCPIP 00004 Duplicate Attribute. Cause. You specified the same attribute twice in a command. Effect. The command is not executed. Recovery. Omit the duplicate attribute and retry the command. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 B- 1 PTCPIP 00005 SCF Error Messages PTCPIP 00005 PTCPIP 00005 Attribute value out of range attribute-name. attribute-name is the name of the attribute you specified in an ALTER PROCESS command. Cause. You specified a value for the ALTER PROCESS command that is outside the valid range. Effect. The command is not executed. Recovery. Enter a valid range for the command and retry it. Refer to the ALTER command in Section 5, SCF Reference for Parallel Library TCP/IP, for more information on valid ranges. PTCPIP 00007 PTCPIP 00007 Duplicate address. Cause. The IP address you specified in the ADD SUBNET or ALTER SUBNET command is already being used by another interface. Effect. The command is not executed. Recovery. Specify a different IP address and retry the command. PTCPIP 00008 PTCPIP 00008 Gateway Network Unreachable. Cause. The gateway you specified in the ADD ROUTE command is unavailable. Effect. The command is not executed. Recovery. Specify an available gateway and retry the command. PTCPIP 00009 PTCPIP 00009 Filesystem error. Cause. A file-system error occurred. Effect. The command is not executed. Recovery. Take an action based on the file system error received. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 B- 2 PTCPIP 00010 SCF Error Messages PTCPIP 00010 PTCPIP 00010 SNAP MTU not available. Cause. TCP/IP cannot communicate with the manager process to obtain the MTU size. Effect. The command is not executed. Recovery. Check or start the manager process. PTCPIP 00011 PTCPIP 00011 Invalid IP address. Cause. The IP address is invalid. Effect. The command is not executed. Recovery. Use a correct IP address. PTCPIP 00012 PTCPIP 00012 Invalid CPU number. Cause. The processor number is invalid. Effect. The command is not executed. Recovery. Use a correct processor number. PTCPIP 00013 PTCPIP 00013 CPU is already a primary CPU. Cause. The processor number is the primary processor number. Effect. The command is not executed. Recovery. Use the backup CPU number. PTCPIP 00014 PTCPIP 00014 RECSIZE must be at least 300 bytes. Cause. The record size given, or implied, in a TRACE command is too small. It must be at least 300 bytes. The minimum trace record size for PTCPIP has expanded to accommodate additional trace information kept by the PTCPIP subsystem. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 B- 3 PTCPIP 00016 SCF Error Messages Effect. The command is not executed. Recovery. Use the RECSIZE parameter while starting a trace. When a larger trace record size is used, there is less chance of trace records being truncated. PTCPIP 00016 PTCPIP 00016 Primary not allowed, some subnets still in STARTED state. Cause. This occurs when a Primary command is rejected because at least one subnet is still in the started state and switching to another CPU. This takes any started subnets out of service. Effect. The command is not executed. Recovery. Try moving the LAN access interfaces to TCP/IP subsystem backup CPU first. PTCPIP 00017 PTCPIP 00017 TCPMAN. Device access not available from same CPU as Cause. Device selected for subnet not available between the same CPU pair as the TCP/IP process. Effect. The command is not executed. Recovery. Select a device that is available between the same CPU pair as the TCP/IP process or stop the TCP/IP process and restart it specifying a CPU pair that has access to the device. PTCPIP 00018 PTCPIP 00018 Device access not available from any CPU. Cause. Physical access to the device does not exist. Either the device is not installed or a failure has occurred. Effect. The command is not executed. Recovery. Select a device that is available or correct the problem. PTCPIP 00019 PTCPIP 00019 Unknown LIF device name. Cause. Incorrect device name specified. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 B- 4 SCF Error Messages PTCPIP 00020 Effect. The command is not executed. Recovery. Select a device that is available or correct the problem. PTCPIP 00020 PTCPIP 00020 The Device selected for the subnet returned a NULL MAC address. Cause. The device selected for the subnet returned a NULL MAC address. Effect. The command is not executed. Recovery. Try the operation again. PTCPIP 00022 PTCPIP 00022 Invalid MAC address. Cause. MAC address is invalid. Effect. The command is not executed. Recovery. Use a correct MAC address. PTCPIP 00027 PTCPIP 00027 Subnet does not exist. Cause. The subnet selected does not exist. Effect. The command is not executed. Recovery. Use a correct subnet name or add the subnet and retry the request. PTCPIP 00035 PTCPIP 00035 The subnets configured in the FAILOVER are invalid. Cause. The two subnets configured for failover are not in the same LAN or the subnet is not failover-enabled. Effect. The command is rejected with the reason. Recovery. Configure two subnets that have two IP addresses in the same subnet range with failover enabled. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 B- 5 SCF Error Messages PTCPIP 00036 PTCPIP 00036 PTCPIP 00036 The subnet intended to be used for FAILOVER is not configured. Cause. The subnet intended to be used for failover is not configured. Effect. The command is rejected with the reason. Recovery. Configure the subnet associated with the error and re-issue the command. PTCPIP 00037 PTCPIP 00037 The command issued to configure FAILOVER is not valid. Cause. The command issued to configure failover is not valid. Two subnets intended to be linked as a failover pair should have the same failover-enabled type, either SHAREDIP or NONSHAREDIP. Effect. The command is rejected with the reason. Recovery. Re-issue the command with the two subnets having the same failover-enabled type, either SHAREDIP or NONSHAREDIP. If both subnets are SHAREDIP FAILOVER type, the RESERVEDIP parameter is also required. PTCPIP 00038 PTCPIP 00038 The FAILOVER brother needs to be in STOPPED state. Cause. The failover brother needs to be in STOPPED state before either subnet in the failover pair can be deleted. Effect. The command is rejected with the reason. Recovery. Abort the associated subnet/brother and re-issue the command. PTCPIP 00039 PTCPIP 00039 The FAILOVER brother was already associated with other subnet. Cause. The failover brother was already associated with another subnet. Effect. The command is rejected with the reason. Recovery. Re-issue the command using another subnet not linked in a failover pair. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 B- 6 SCF Error Messages PTCPIP 00040 PTCPIP 00040 PTCPIP 00040 The command is not valid for FAILOVER enabled subnet. Cause. The command is invalid in the failover-enabled subnet. Causes include: • • • For a failover-enabled subnet, the ADDALIAS is not allowed until the subnet is associated with another subnet as a failover pair. For a failover-enabled subnet, the DELETEALIAS is not allowed. For a failover-enabled subnet, the ALTER subnet, IPADDDRESS or ALTER subnet, SUBNETMASK command is not allowed. Effect. The command is rejected with the reason. Recovery. Re-issue the command with correct parameters. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 B- 7 SCF Error Messages PTCPIP 00040 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 B- 8 C Tracer Utility The Tracer Utility displays the path taken by IP packets enroute to a network host. Use the Tracer Utility to determine any problems that these packets might encounter. From each gateway system along the path, the Tracer Utility attempts to elicit an ICMP TIME_EXCEEDED message. From the destination remote host, it attempts to elicit a ICMP_PORT_UNREACHABLE message. Running the Tracer Utility from a Terminal You can use Tracer Utility only if your user ID is SUPER.SUPER. Output from the Tracer Utility appears on the screen of the terminal from which the utility was launched. You can also choose to have the output logged to a file. TRACER [ / run-option ] [ , run-option... / ][ -d ] [ -m max-ttl ][-n ][ -p port-num ] [ q nqueries ] [ -r ][ -s src-addr ] [ -v ] [ -w wait-time ] remote-host-name [ data-size ] [run-option ] specifies an operating system RUN command option. For a complete description of all RUN options, see the TACL Reference Manual. Note that the OUT option allows you to send the output of a trace to a log file. Examples: The following command directs the output of a trace to be sent to a remote system named \IDEV to a disk file named $fiti.trace. traceout on the local system. >TACL TRACER/OUT $fiti.trace.traceout/IDEV The following command directs the output of a trace to be sent to a remote system named \IDEV to a disk file named $wpo.trace. traceout on the system named \igate. >TACL TRACER/OUT \igate.$wpo.trace.traceout/IDEV [ -d ] sets the SO_DEBUG option on the socket being used. [ -m max-ttl ] specifies the maximum time-to-live value (or number of hops) used in out-going probe packets. If you do not specify this option, the Tracer Utility uses the default value of 30 hops. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 C- 1 Running the Tracer Utility from a Terminal Tracer Utility [ -n ] specifies that the address of each gateway be printed numerically rather than both symbolically and numerically. Thus, only the IP address, rather than the address and gateway name, appears in the output. Specifying -n avoids having the Tracer Utility do a time-consuming address-to-name lookup. HP recommends use of this option. [ -p port-num ] specifies the base UDP port number used in probes. If you do not specify this option, the Tracer utility uses the default value 33434 for port-num. The UDP port number, whether it is the default number or a number you specify through this option, should not be an actual port range on the remote host to which the probe is destined. The remote host should not process the probe packet. Instead, the remote host should send back a ICMP_UNREACH_PORT message to conclude route tracing. The Tracer Utility informs you of this occurrence by printing an exclamation point either on your screen or in the disk file you specified. Specifying the -p option is useful when the default value (33434) does specify an actual port range on the destination host. In such cases, -p option allows you to specify an unused port range. [ -q nqueries ] specifies the number of probes, or queries, for each TTL. If you do not specify this option, the Tracer Utility uses the default value 3 for nqueries. [-r ] specifies that the routing tables be bypassed and that probes should be sent directly to a host on an attached network. If the host is not on a directly attached network, an error message is returned. You can use this option to send an ICMP echo request to a local host through an interface that does not involve routing. [ -s src-addr ] specifies that the IP address in src-add should be used as the source address in outgoing probe packets. The address specified in src-add must be an IP number rather than a host name. On NonStop systems that have more than one IP address, use the -s option to change the source address to an address that differs from the IP address of the interface on which the probe packet is sent. The IP address you specify for src-addr must be one of the IP addresses of the NonStop system on which you launch the trace. Otherwise, an error message is returned, and the Tracer Utility does not send any probes. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 C- 2 Tracer Utility Running the Tracer Utility from a Terminal [ -v ] specifies verbose output. If you specify this option, all received ICMP packets are listed. If you do not specify this option, only the ICMP packets TIME_EXCEEDED and UNREACHABLE are listed. [ -w wait-time ] specifies the time, in seconds, that the Tracer Utility waits for a response to a probe. If you do not specify this option, the Tracer Utility uses the default value of 5 seconds for wait-time. remote-host-name specifies the name or IP address of the remote host system to which the Tracer Utility is to trace the path. This parameter is required. You must specify it following any Tracer options (-d, -m, -n, -p, -q, -r, -s, -v, or -w). [ data-size ] specifies the packet-size in bytes. If you do not specify data-size, the Tracer Utility uses the default probe-datagram length. The default probe-datagram length is 38 bytes. You must specify data-size immediately following the specification of host-name. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 C- 3 Tracer Utility Running the Tracer Utility from a Terminal HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 C- 4 Glossary This glossary defines terms used both in this manual and in other NonStop TCP/IP manuals. Both industry-standard terms and HP terms are included. address mask. A bit mask used to select bits from an Internet address for subnet addressing. The mask is 32 bits long and selects the network portion of the Internet address and one or more bits from the local portion. address resolution. Conversion of an Internet address into a corresponding physical address. Depending on the underlying network, resolution may require broadcasting on a local network. See also address resolution. Address Resolution Protocol (ARP). The Internet protocol used to dynamically bind a high-level Internet Address to a low-level physical hardware address. ARP applies only across a single physical network and is limited to networks that support hardware broadcast. Advanced Projects Research Agency (ARPA). Former name of DARPA, the government agency that funded the ARPANET and DARPA Internet. ARP. See Address Resolution Protocol (ARP). ARPA (Advanced Projects Research Agency). See Advanced Projects Research Agency (ARPA). ARPANET. A pioneering long-haul network funded by ARPA (later DARPA) and built by Bolt, Baranek, and Newman (BBN). It served as the basis for early networking research as well as a central backbone during the development of the Internet. asynchronous. A mode of serial-data transmission in which characters are sent at random; there is no timing relationship between the end of one character and the start of the next, that is, the transmission is not synchronized with a separate clock signal. The data contains extra bits: a start bit to signal the beginning of a byte and one or more stop bits to signal the end of the byte. These start and stop bits allow the receiver to determine the correct synchronization. attribute. In DSM, a characteristic of an entity. For example, two attributes of a communications line might be its baud rate and its retry count. In a token-oriented interface based on SPI, an attribute of an object is usually expressed as either a simple token or as a field within an extensible structured token. See also simple token or extensible structured token. autonomous system. A collection of gateways and networks that fall under one administrative entity and cooperate closely to propagate network reachability (and routing) information among themselves using an interior gateway protocol of their choice. Gateways within an autonomous system have a high degree of trust. At least HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Glossary- 1 baseband Glossary one gateway in an autonomous system must advertise networks in that system to a core gateway using EGP. baseband. Characteristic of any network technology (like Ethernet) that uses a single carrier frequency and requires all stations attached to the network to participate in every transmission. See broadband. bridge. A router that connects two or more networks and forwards packets among them. Usually, bridges operate at the physical network level. For example, an Ethernet bridge connects two physical Ethernet cables and forwards from one cable to the other exactly those packets that are not local. Bridges differ from repeaters; bridges store and forward complete packets, while repeaters forward electrical signals. broadband. Characteristic of any network technology that multiplexes multiple, independent network carriers onto a single cable (usually using frequency division multiplexing). For example, a single 100 mbps broadband cable can be divided into ten 10 mbps carriers, with each treated as an independent Ethernet. The advantage of broadband is less cable; the disadvantage is higher cost for equipment. See baseband. broadcast. A packet delivery system that delivers a copy of a given packet to all hosts that attach to it is said to broadcast the packet. Broadcast may be implemented with hardware or software. brother. See failover brother. BSD. Berkeley Software Distribution. Carrier Sense Multiple Access (CSMA). A characteristic of network hardware that operates by allowing multiple stations to contend for access to a transmission medium by listening to see if it is idle. Carrier Sense Multiple Access with Collision Detection (CSMA/CD). A characteristic of network hardware that uses CSMA access combined with a mechanism that allows the hardware to detect when two stations simultaneously attempt transmission. Ethernet is an example of a well-known network based on CSMA/CD technology. Class A. The network number is 1 through 127 (1 octet); that is, the first octet is in the range 1-127. The remaining three octets in the address are used for the subnet number and host number. Class B. The network number is 128 through 191.255 (2 octets); that is, the first octet is in the range 128-191, the second octet is in the range 0-255. The remaining two octets are used for the subnet number and host number. Class C. The network number is 192.0.0 through 255.255.255 (3 octets); that is, the first octet is in the range 192-255, the second octet is in the range 0-255, and the third octet is in the range 0-255. The remaining octet is used for the subnet number and host number. The subnet number varies in length. The subnet number's width is typically represented by a bit mask. The rest of the available bits uniquely identify the HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Glossary- 2 Class D Glossary host connected to the subnetwork. LANs connected by way of a gateway to the INTERNET get their subnet class from the DCA's NIC (Network Information Center). The address classes of standalone, or entirely private, LANs are administered by the LAN administrator. Typical usage calls for all CLASS A addresses to have private LANs. Class D. A Class D address is a 4-octet multicast group address. The four high-order bits of the address are always 1110; therefore, the first octet is a number in the range 224 through 239 (%HE0 through %HEF). This means that an Internet can have a total of 268,435,456 multicast groups. collector. An EMS process that accepts event messages from subsystems and logs them in the event log. See also Event Management Service (EMS). Compare distributor. command message. A SPI message, containing a command, sent from an application program to a subsystem. See also SPI message. Compare response message or event message. common definition. In DSM programmatic interfaces, a definition (data declaration) used in several commands, responses, or event messages in an SPI interface to a subsystem. See also definition. compatibility distributor. An EMS distributor process that filters event messages according to fixed (rather than user-specified) criteria, obtains text for these messages that is compatible with the operator console of Guardian operating system versions earlier than C00, and writes the text to the standard Guardian console-message destinations. See also distributor. conditional token. In DSM event management, a token that is sometimes, but not always, present in a particular event message. connection. The path between two protocol modules that provides reliable stream delivery service. In the Internet, a connection extends from a TCP module on one machine to a TCP module on another machine. connectionless service. Characteristic of the packet delivery service offered by most hardware and by the Internet Protocol (IP). The connectionless service treats each packet or datagram as a separate entity that contains the source and destination address. Usually, connectionless services can drop packets or deliver them out of sequence. consumer distributor. An EMS distributor process that returns selected event messages to management applications upon request. See also distributor. context token. In DSM programmatic interfaces, a token in an SPI response message that indicates (by its presence or absence) whether or not the response is continued in the following message. If this token is present, the response is continued. To obtain the next message, the application program reissues the original command with one modification: the context token is included in the new command message. When the HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Glossary- 3 control and inquiry Glossary subsystem sends a response message that does not contain a context token, the series of response messages is complete. control and inquiry. In DSM, those aspects of object management that affect the state or configuration of an object, such as inquiries about the object and commands pertaining to the environment (for example, commands that set default values for the session). Compare event management. critical event. A DSM event that is considered to be crucial to the operation of the system or network. Each subsystem determines which of its events are critical, designating them as such by setting the value of the emphasis token to TRUE. Compare noncritical event. CSMA. See Carrier Sense Multiple Access (CSMA). CSMA/CD. See Carrier Sense Multiple Access with Collision Detection (CSMA/CD). DARPA. See Defense Advanced Projects Research Agency (DARPA). data communications standard definitions. In DSM, the set of declarations provided by HP for use in all management programs that manage or retrieve event messages from NonStop data communications subsystems. The names of these definitions start with either ZCOM or ZCMK. See also definition or definition files. Compare SPI standard definitions or EMS standard definitions. data list. In DSM programmatic interfaces, a group of tokens used to separate response records within an SPI message for a response, or used to enclose a single response record, if the program so requests. A data list consists of a list token that denotes a data list (different from the token that starts an error list or a generic list), followed by a response record and an end-list token. See also response record. DDN. See Defense Data Network (DDN). Defense Advanced Projects Research Agency (DARPA). Formerly called ARPA. The government agency that funded research and experimentation with the ARPANET and DARPA Internet. Defense Data Network (DDN). Used to loosely refer to the MILNET, ARPANET, and the TCP/IP protocols they use. More literally, it is the MILNET and associated parts of the Internet that connect military installations. definition. One of the declarations provided by HP for use in applications that call APS or SPI procedures. These definitions are provided in definition files. See also definition files. definition files. A set of files containing declarations for use in applications that call SPI procedures. SPI has a standard definition file for the Data Definition Language (DDL) and one for each of the programming languages supporting SPI; the latter files are derived from the DDL definition file. Likewise, each subsystem that has a tokenHP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Glossary- 4 Distributed Systems Management Glossary oriented programmatic interface has one definition file for DDL and one for each programming language. Some subsystems—for instance, data communications subsystems—have additional, shared definition files. See also SPI standard definitions, data communications standard definitions, or EMS standard definitions. Distributed Systems Management. A set of tools used to manage NonStop systems and EXPAND networks. These tools include the VIEWPOINT console application, the Subsystem Control Facility (SCF) for data communications subsystems, the Subsystem Programmatic Interface (SPI), the Event Management Service (EMS), the Distributed Name Service (DNS), and token-oriented programmatic interfaces to the management processes for various NonStop subsystems. distributor. An EMS process that distributes event messages from event logs to requesting management applications, to Guardian console message destinations, or to a collector on another node. See also consumer distributor. and compatibility distributor. Contrast collector. DNS. See Domain Name Server (DNS). Domain Name Server (DNS). A method for naming resources. The basic function of the domain name server is to provide information about network objects by answering queries. Domain. In the Internet, a part of the naming hierarchy. Syntactically, a domain name consists of a sequence of names (labels) separated by periods (dots). DSM. See Distributed Systems Management. E4SA. See Ethernet 4 ServerNet adapter (E4SA). ECHO. The name of a program used in the Internet to test the reachability of destinations by sending them an ICMP echo request and waiting for a reply. EGP (Exterior Gateway Protocol). The protocol used by a gateway in one autonomous system to advise the Internet addresses of networks in that autonomous system to a gateway in another autonomous system. Every autonomous system must use EGP to advertise network reachability to the core gateway system. empty response record. In DSM programmatic interfaces, a response record containing only a return token with a value that means “no more response records.” See also return token. EMS. See Event Management Service (EMS). EMS standard definitions. The set of declarations provided by EMS for use in event management, regardless of the subsystem. Any application that retrieves tokens from event messages needs the EMS standard definitions. See also definition or definition files. Comparedata communications standard definitions or SPI standard definitions. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Glossary- 5 error Glossary error. In DSM interfaces, a condition that causes a command or other operation to fail. Contrast Warning. error list. In DSM programmatic interfaces, a group of tokens used within a response record to provide error and warning information. An error list consists of a list token that denotes an error list (different from the token that starts a data list or a generic list), followed by an error token, other tokens explaining the error (optional), and an end-list token. Error lists can be nested within other error lists. The return token cannot be included in an error list. See also return token. error number. In DSM programmatic interfaces, a value that can be assigned to a return token, or to the last field of an error token, to identify an error that occurred. Some error numbers are defined in the data communications (ZCOM) definitions; others are defined by individual subsystems. error token. In DSM programmatic interfaces, a token in a response message that indicates the reason an error occurred during a programmatic command. NonStop subsystems enclose each error token in an error list, which can also contain additional information about the error. A response record must contain a return token and can also contain error lists to explain the error further. The token code for the error token is ZSPI-TKNERROR. Its value is a structure consisting of the subsystem ID and an error number identifying the error. See also error list, error number, or return token. Ethernet. A popular local area network technology invented at the Xerox Corporation Palo Alto Research Center. An Ethernet itself is a passive coaxial cable; the interconnections all contain active components. Ethernet is a best-effort delivery system that uses CSMA/CD technology. Xerox Corporation, Digital Equipment Corporation, and Intel Corporation developed and published the standard for 10 Mbps Ethernet. Ethernet 4 ServerNet adapter (E4SA). A ServerNet adapter for Ethernet local area networks (LANs) that contains four Ethernet ports. Ethernet meltdown. An event that causes saturation or near saturation on an Ethernet. It usually results from illegal or misrouted packets and typically lasts only a short time. As an example, consider an IP datagram directed to a nonexistent host and delivered by way of hardware broadcast to all machines on the network. Gateways receiving the broadcast will send out ARP packets in an attempt to find the host and deliver the datagram. event. In DSM terms, a significant change in some condition in the system or network. Events can be operational errors, notifications of limits exceeded, requests for action needed, and so on. event log. A file or set of files maintained by EMS to store event messages generated by subsystems. Event Management Service (EMS). A part of DSM used to provide event collection, event logging, and event distribution facilities. It provides for different event descriptions for HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Glossary- 6 event management Glossary interactive and programmatic interfaces, lets an operator or application select specific event-message data, and allows for the flexible distribution of event messages within a system or network. EMS has an SPI-based programmatic interface for both reporting and retrieving events. See also DSM or event message. event management. The reporting and logging of events, the distribution and retrieval of information concerning those events, and the actions taken by operations personnel or software in response to the events. Compare control and inquiry. event message. A special kind of SPI message that describes an event occurring in the system or network. Compare command message. Expand. The Expand subsystem enables you to link together as many as 255 geographically dispersed NonStop systems to create a network with the same reliability, capacity to preserve data integrity, and potential expansion as a single NonStop system. extensible structure. In DSM programmatic interfaces, a structure declared for the value of an extensible structured token. See also extensible structured token. Compare fixed structure. extensible structured token. In DSM programmatic interfaces, a token consisting of a token code and a value that is an extensible structure. Extensible structures can be extended by adding new fields at the end in later RVUs. Such structures are typically used to indicate the attributes of an object being operated on and to return status and statistics information in responses; they can also be used for other purposes. The token is referenced by a token map that describes the structure to SPI so that SPI can provide compatibility between different versions of the structure. Compare simple token. fabric. A simplified way of representing a complex set of interconnections through which there can be multiple and (to the user) unknown paths from point to point. The term fabric is used to refer to the X or Y portion of the ServerNet system area network (ServerNet SAN), for example the X fabric. failover. A feature that you can enable in your Parallel Library TCP/IP environment that provides continuous access to the network during a LIF failure. failover brother. The other subnet associated with the first subnet in a failover-enabled configuration. Fast Ethernet ServerNet adapter (FESA). The FESA is a CRU that supports one Ethernet 10 Base-T or 100 Base-TX connection and communicates with multiple processors through its dual ServerNet interfaces to the ServerNet fabrics. FDDI. See Fiber Distribution Data Interface (FDDI). FESA. See Fast Ethernet ServerNet adapter (FESA). HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Glossary- 7 Fiber Distribution Data Interface (FDDI) Glossary Fiber Distribution Data Interface (FDDI). An emerging standard for a network technology based on fiber optics. FDDI specifies a 100-mbps data rate using 1300-nanometer light wavelength, and limits networks to approximately 200 km in length, with repeaters every 2 km or less. The access control mechanism uses token-ring technology. File Transfer Protocol (FTP). The Internet standard, high-level protocol for transferring files from one machine to another. Usually implemented as application level programs, FTP uses the TELNET and TCP protocols. The server side requires a client to supply a login identifier and password before it will honor requests. filter. In EMS, a file containing a list of criteria against which incoming event messages can be compared. The messages are allowed to pass (all criteria met) or not pass (one or more criteria failed). In the ServerNet LAN Systems Access (SLSA) subsystem (for NonStop S-series systems), filters are logical entities which allow frames received from the LAN to be sorted and delivered to a client. In the SLSA subsystem, filters replace the PORT objects used in K-series systems in the sense that filters are the final destination for data received from the LAN. FINGER. A protocol providing a method for retrieving status information about one or all of the users on a particular system. fixed structure. In DSM programmatic interfaces, a multifield structure declared for the value of a simple token. Fields cannot be added to fixed structures in later RVUs. Compare extensible structure. forwarding distributor. An EMS distributor process that sends selected event messages to an EMS collector on another network node. See also distributor. FTP. See File Transfer Protocol (FTP). full-duplex mode. The communication mode in which data can be transferred in both directions simultaneously. In the Session Layer, no data token is needed. gateway. A special-purpose, dedicated computer that attaches to two or more networks and routes packets from one to the other. In particular, an Internet gateway routes IP datagrams among the networks to which is connected. Gateways route packets to other gateways until they can be delivered to the final destination directly across one physical network. The term is loosely applied to any machine that transfers information from one network to another, as in mail gateway. Gateway to Gateway Protocol. The protocol core gateways used to exchange routing information, GGP implements a distributed shortest path routing computation. Under normal circumstances, all GGP participants reach a steady state in which the routing information at all gateways agrees. G4SA. See Gigabit Ethernet 4-port ServerNet Adapter (G4SA). GESA. See Gigabit Ethernet ServerNet Adapter (GESA). HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Glossary- 8 GGP Glossary GGP. See Gateway to Gateway Protocol. Gigabit Ethernet 4-port ServerNet Adapter (G4SA). A multiport ServerNet adapter that provides 1000 megabits/second (Mbps) data transfer rates between HP NonStop S-series systems and Ethernet LANs. The G4SA is the only LAN adapter supported for the I/O Adapter Module (IOAM) enclosure, and it is installed in slots 1, 2, 3, 4, and 5 of an IOAM. Although the G4SA supersedes the Ethernet 4 ServerNet adapter (E4SA), Fast Ethernet ServerNet adapter (FESA), and the Gigabit Ethernet ServerNet adapter (GESA), it cannot be installed in an HP NonStop S-series enclosure. Gigabit Ethernet ServerNet Adapter (GESA). A single-port ServerNet adapter that provides Gigabit connectivity on a NonStop S-series server. The GESA installs directly into an existing Ethernet port, and multiple GESAs are supported in a system enclosure. half-duplex mode. The communications mode in which data can be transferred in both directions, but only in one direction at a time, and in which the direction of data flow alternates. In the Session Layer, the data token indicates which side can send data. header. The initial part of an SPI message. The first word of this header always contains the value -28; the remainder of the header contains descriptive information about the SPI message, most of which is accessible as header tokens. The tokens in an SPI message header differ according to the type of message: the header of a message that contains a command or response differs somewhat from the header of an event message. An application can use SSGET or EMSGET calls to retrieve the values of header tokens, and can use SSPUT calls to change the values of some tokens. However, there are certain basic differences between header tokens and other tokens. See also header token. header token. In an SPI message, a token that provides information pertaining to the message as a whole. Header tokens differ from other tokens in several ways: they exist in the buffer at initialization and their values are usually set by SSINIT, they can occur only once in a buffer, they are never enclosed in a list, they cannot be moved to another buffer with SSMOVE, and programs cannot position to them or retrieve their values using the NEXTCODE or NEXTTOKEN operation. Programs retrieve the values of header tokens by passing appropriate token codes to SSGET and can change the values of some header tokens by passing their token codes to SSPUT. Examples of header tokens for commands are the command number, the object type, the maximum-response token, the server-version token, the maximum-field-version token, and the checksum token. Command and response messages contain a specified set of header tokens; event messages, a different set with some overlap. See also SPI message. hierarchical routing. Routing based on a hierarchical addressing scheme. Most Internet routing is based on a two-level hierarchy in which an Internet address is divided into a network portion and a host portion. Gateways use only the network portion until the HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Glossary- 9 hop count Glossary datagram reaches a gateway that can deliver it directly. Subnetting introduces additional levels of hierarchical routing. hop count. A measure of distance between two points in the Internet. A hop count of n means that n gateways separate the source destination. ICMP. See Interior Gateway Protocol (IGP). IEEE. See Institute of Electrical and Electronics Engineers (IEEE). IEEE 802.3. A local area network protocol suite commonly known as Ethernet. Ethernet has either a 10Mbps or 100Mbps throughput and uses Carrier Sense Multiple Access bus with Collision Detection (CSMA/CD. This method allows users to share the network cable. However, only one station can use the cable at a time. A variety of physical medium dependent protocols are supported. IEE 802.5. A local area network protocol suite commonly known as token ring. A standard originated by IBM for a token-passing ring network that can be configured in a star topology. Versions supported are 4Mbps and 16 Mbps. IEN. See Internet Engineering Note (IEN). IGP. See Interior Gateway Protocol (IGP). Interior Gateway Protocol (IGP). The generic term applied to any protocol used to propagate network reachability and routing information within an autonomous system. Although no standard Internet IGP exists, RIP is among the most popular. Institute of Electrical and Electronics Engineers (IEEE). An international industry group that develops standards for many areas of electrical engineering and computers. interactive command. In DSM, a command entered by a human operator rather than by a program. See also programmatic command. International Organization for Standardization (ISO). A United Nations organization, established to promote the development of standards to facilitate the international exchange of goods and services and to develop mutual cooperation in areas of intellectual, scientific, technological, and economic activity. International Telecommunications Union Telecommunications (ITU-T). An international body of member countries whose task is to define recommendations and standards relating to the international telecommunications industry. The fundamental standards for ATM have been defined and published by the ITU-T (previously CCITT). Internet. Physically, a collection of packet switching networks interconnected by gateways, along with protocols that allow them to function logically as a single, large, virtual network. When written in uppercase, INTERNET refers specifically to the DARPA Internet and the TCP/IP protocols it uses. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Glossary -10 Internet address Glossary Internet address. The 32-bit address assigned to hosts that want to participate in the Internet using TCP/IP. Internet addresses are the abstraction of physical hardware addresses, just as the Internet is an abstraction of physical networks. Actually assigned to the interconnection of a host to a physical network, an Internet address consists of a network portion and a host portion. The partition makes routing efficient. Internet Control Message Protocol (ICMP). An integral part of the Internet Protocol (IP) that handles error and control messages. Specifically, gateways and hosts use ICMP to send reports of problems about datagrams back to the original source that sent the datagram. ICMP also includes an echo request/reply used to test whether a destination is reachable and responding. Internet Engineering Note (IEN). A series of notes developed in parallel to RFCs and available across the Internet from the INIC. IENs contain many of the early theories on the Internet. Internet Protocol (IP). The Internet standard protocol that defines the Internet datagram as the unit of information passed across the Internet, and that provides the basis for the Internet, connectionless, best-effort, packet-delivery service. interoperability. The ability of software and hardware on multiple machines from multiple vendors to communicate meaningfully. IOP. Input/output process. An input/output process (IOP) is a privileged process, residing in a fault-tolerant system processor, which provides an application access to a communications line. IP. See Internet Protocol (IP). IP datagram. The basic unit of information passed across the Internet. An IP datagram is to the Internet as a hardware packet is to a physical network. It contains source and destination addresses, along with data. ISO. See International Organization for Standardization (ISO). ITU-T. See International Telecommunications Union Telecommunications (ITU-T). LANMAN. See LAN manager (LANMAN) process. LAN. See local area network (LAN). LAN manager (LANMAN) process. The process provided as part of the ServerNet local area network (LAN) systems access (SLSA) subsystem that starts and manages the SLSA subsystem objects and the LAN monitor (LANMON) process and assigns ownership of Ethernet adapters to the LANMON processes in the system. Subsystem Control Facility (SCF) commands are directed to the LANMON processes for configuring and managing the SLSA subsystem and the Ethernet adapters. LANMON. See LAN monitor (LANMON) process. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Glossary -11 LAN monitor (LANMON) process. Glossary LAN monitor (LANMON) process. The process provided as part of the ServerNet local area network (LAN) systems access (SLSA) subsystem that has ownership of the Ethernet adapters controlled by the SLSA subsystem. LAPB (Link Access Protocol —Balanced). ITU-T standards that define in the Data Link Layer the requirements for X.25 connections over wide area networks (WANs). Level 2. A reference to LINK LEVEL communication (for example, frame formats) or linklevel connections derived from the ISO 7-layer reference model. For long-haul networks, level 2 refers to the communication between a host computer and a network packet switch (for example, HDLC/LAPB). For local area networks, level 2 refers to physical packet transmission. Thus, a level 2 address is a physical hardware address. Level 3. A reference to NETWORK-level communication derived from the ISO 7-layer reference model. For the Internet, level 3 refers to the IP and IP datagram formats. Thus, a level 3 address is an Internet address. LIF. See logical interface (LIF). LLC (Logical Link Control). See Logical Link Control (LLC). local area network (LAN). A network that is located in a small geographical area and whose communications technology provides a high-bandwidth, low-cost medium to which low-cost nodes can be connected. One or more LANs can be connected to the system such that the LAN users can access the system as if their workstations were connected directly to it. logical interface (LIF). The interface that allows an application or another process to communicate with data communications hardware. Logical Link Control (LLC). An IEEE 802.2 standard for the Data Link Layer of the OSI Reference Model that defines both connection-oriented and connectionless standards over LAN networks. MAC address. See Media Access Control (MAC) Address. management applications. In DSM, an application process that opens a management or subsystem process to control a subsystem. This process can issue SPI commands to subsystems and retrieve EMS event messages to assist in the management of a computer system or a network of systems. A management application is a requester to the subsystems to which it sends commands; the subsystems are servers to the management application. management process. In DSM, an HP process through which an application issues commands to a subsystem. A management process can be part of a subsystem, or it can be associated with more than one subsystem; in the latter case, the management process is logically part of each of the subsystems. SCP is the management process for all NonStop data communications subsystems that support DSM. See also subsystem. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Glossary -12 manager process Glossary manager process. In DSM, an HP subsystem process with which the SCP management process communicates to control a particular data communications subsystem. Media Access Control (MAC) Address. A MAC address is a value in the Medium Access Control sublayer of the IEEE/ISO/ANSI LAN architecture, that uniquely identifies an individual station that implements a single point of physical attachment to a LAN. MFIOB. See multifunction I/O board (MFIOB). MILNET (Military Network). Originally part of the ARPANET, MILNET was partitioned in 1984 to make it possible for military installations to have reliable network service, while the ARPANET continues to be used for research. MILNET uses exactly the same hardware and protocol technology as ARPANET, and there are several interconnection points between the two. Thus, under normal circumstances, MILNET sites are part of the Internet. multicast. A technique that allows copies of a single packet to be passed to a selected subset of all possible destinations. Some hardware (for example, Ethernet) supports multicast by allowing a network interface to belong to one or more multicast groups. Broadcast is a special form of multicast in which the subset of machines selected to receive a copy of a packet consists of the entire set. multifunction I/O board (MFIOB). A ServerNet adapter that contains ServerNet addressable controllers (SACs) for SCSI and Ethernet; a service processor; ServerNet links to the processor, to the two ServerNet adapter slots, and to one of the ServerNet expansion board (SEB) slots; and provides connections to the serial maintenance bus (SMB), which connects components within an enclosure to the service processor. Network File System (NFS). A protocol developed by SUN Microsystems that uses IP to allow a set of cooperating computers to access each other's file systems as if they were local. The key advantage of NFS over conventional file transfer protocols is that NFS hides the differences between local and remote files by placing them in the same name space. NFS is used primarily on UNIX systems, but has been implemented for many systems, including personal computers like an IBM PC and Apple Macintosh. NFS. See Network File System (NFS). noncritical event. A DSM event not too crucial to system or network operations. Each subsystem determines which of its events are noncritical by setting the value of the emphasis token to FALSE. Compare critical event. nonsensitive command. A DSM command that can be issued by any user or program allowed access to the target subsystem—that is, a command on which the subsystem imposes no further security restrictions. For NonStop data communications subsystems, the nonsensitive commands are all those that cannot change the state or configuration of objects (usually information commands). Compare sensitive command. nowait mode. In Guardian file-system operations and in some APS operations, the mode in which the called procedure initiates an I/O operation but does not wait for it to complete HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Glossary -13 object Glossary before returning control to the caller. In order to make the called procedure wait for the completion of the operation, the application calls a separate procedure. Compare wait mode. object. (1) In general HP use, one or more of the devices, lines, processes, and files in a NonStop subsystem; any entity subject to independent reference or control by one or more subsystems. (2) In DSM use, an entity subject to independent reference and control by a subsystem: for example, the disk volume $DATA or the data communications line $X2502. An object typically has a name and a type known to the controlling subsystem. object-name template. In DSM, a name that stands for more than one object. Such a name includes one or more wild-card characters, such as * and ?. See also wild-card character. object type. In DSM, the category of objects to which a specific object belongs: for example, a specific disk file might have the object type FILE, and a specific terminal might have the object type SU (subdevice). A subsystem identifies a set of object types by the objects it manages. The SCF interfaces to NonStop data communications subsystems use standard keywords to identify the types. The corresponding programmatic interfaces have object-type numbers (represented by symbolic names such as ZCOM-OBJ-SU) suitable for passing to the SPI SSINIT procedure. open system. Any computer system that adheres to the OSI standards. Open Systems Interconnection. A set of standards used for the interconnection of heterogeneous computer systems, thus providing universal connectivity. OSI. See Open Systems Interconnection. OSI Reference Model. A communications architecture, adopted by the ISO in 1984, that includes seven layers that define the functions involved in communications between two systems, the services required to perform these functions, and the protocols associated with these services. packet. The unit of data sent across a packet switching network. While some Internet literature uses it to refer specifically to data sent across a physical network, other literature views the Internet as a packet switching network and describes IP datagrams as packets. Packet Internet Groper (PING). The name of a program used in the Internet to test the reachability of destinations by sending them an ICMP echo request and waiting for a reply. The term has survived the original program and is now used as a verb, as in “please ping host A to see if it is alive.” packet switching. A technique in which messages are broken into smaller units, called packets, that can be individually addressed and routed through the network. The receiving-end node ascertains whether all the packets are received and in the proper sequence before forwarding the complete message to the addressee. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Glossary -14 PDN Glossary PDN. See Public Data Network (PDN). physical interface (PIF). The hardware components that connect a system node to a network. physical layer. Layer 1 in the OSI Reference Model. This layer establishes the actual physical connection between the network and the computer equipment. Protocols at the Physical Layer include rules for the transmission of bits across the physical medium and rules for connectors and wiring. PIF. See physical interface (PIF). PING. See PING. predefined value. A commonly used value—for instance, a value for a token or a field in a token—that is given a name in a set of definition files. process. A running entity that is managed by the operating system, as opposed to a program, which is a collection of code and data. When a program is taken from a file on a disk and run in a processor, the running entity is called a process. programmatic command. In DSM, a command issued by a program rather than by a human operator. Compare interactive command. protocol. A formal description of message formats and the rules two or more machines must follow to exchange those messages. Protocols can describe low level details of machine-to-machine interfaces (for example, the order in which the bits from a byte are sent across a wire), or high-level exchanges between application programs (for example, the way in which two programs transfer a file across the Internet). Most protocols include both intuitive descriptions of the expected interactions as well as more formal specifications using finite state-machine models. Public Data Network (PDN). A network with data communications services available to any subscriber. Request for Comments (RFC). he name of a series of notes that contain surveys, measurements, ideas, techniques, and observations, as well as proposed and accepted Internet protocol standards. RFCs are edited but not referenced. They are available across the Internet. response. In DSM use, the information or confirmation supplied (as part of a response message) to an application by a subsystem in response to a DSM command. response message. An SPI message sent from a subsystem to an application program in reaction to a command message. Compare command message or event message. response record. In DSM programmatic interfaces, a set of response tokens usually describing the result when a command is performed on one object. Every response record in a response from a NonStop subsystem contains a return token; a response HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Glossary -15 return token Glossary record can also contain error lists that include error tokens. A response can consist of multiple response records, spread across one or more response messages. A response record cannot be split between two response messages. If multiple response records are in a response message, each response record is enclosed in a data list. See also data list. Each response record is required to contain a return token. See also return token. return token. In DSM programmatic interfaces, the token that indicates whether a command was successful and, if not, why it failed. The token code for the return token is ZSPI-TKN-RETCODE. Its value consists of a single integer field. Compare error token. RFC. See Request for Comments (RFC). SAC. See ServerNet addressable controller (SAC). SCF. See Subsystem Control Facility (SCF). SCP. See Subsystem Control Point (SCP). secondary route. For multiple routes to the same destination, all the routes in addition to the primary route (the route visible to Radix Routing topology) are called shadow/secondary routes. Also called shadow route on page -17. sensitive command. In DSM, a command that can be issued only by a restricted set of Guardian users, such as the owner of a subsystem process. For NonStop data communications subsystems, the sensitive commands are those that can change the state or configuration of objects, start or stop tracing, or change the values of statistics counters. Compare nonsensitive command. ServerNet adapter. A customer-replaceable unit (CRU) that connects peripheral devices to the rest of the system through a ServerNet bus interface (SBI). A ServerNet adapter is similar in function to an I/O controller logic board (LB) and backplane interconnect card (BIC) in NonStop K-series servers. ServerNet addressable controller (SAC). A controller that is uniquely addressable within one or more ServerNet address domains (SADs) through the node ID and address fields in a request packet. A SAC typically is implemented on some portion of a processor multifunction (PMF) customer-replaceable unit (CRU), an I/O multifunction (IOMF) CRU, or a ServerNet adapter. ServerNet LAN Systems Access (SLSA) subsystem. A subsystem of the NonStop operating system. The SLSA subsystem enables the protocol I/O processes (IOPs) and drivers to access the ServerNet adapters. ServerNet wide area network (SWAN) concentrator. An HP data communications peripheral that provides connectivity to a NonStop S-series server. The SWAN concentrator supports both synchronous and asynchronous data over RS-232, RS-449, X.21, and V.35 electrical and physical interfaces. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Glossary -16 service Glossary service. A set of primitives (operations) that a layer provides to the layer above it. The service defines what operations the layer can perform on behalf of its users, but not how these operations are implemented. A service relates to an interface between two layers: the lower layer is the service provider, and the upper layer is the service user. Compare protocol. session. For a management application, the period during which an application can issue commands to a subsystem. shadow route. For multiple routes to the same destination (all the routes in addition to the primary route) the route visible to Radix Routing topology, are called shadow/secondary routes. Also called secondary route on page -16. Simple Mail Transfer Protocol (SMTP). The Internet standard protocol for transferring electronic mail messages from one machine to another. SMTP specifies how two mail systems interact, and specifies the format of control messages the two mail systems exchange to transfer mail. simple token. In DSM programmatic interfaces, a token consisting of a token code and a value that is either a single elementary field, such as an integer or a character string, or a fixed (nonextensible) structure. Compare extensible structured token. SLSA Subsystem. See ServerNet wide area network (SWAN) concentrator on page -16 SMTP. See Simple Mail Transfer Protocol (SMTP). SNAP. See Subnetwork Access Protocol (SNAP). SPI. See Subsystem Programmatic Interface (SPI). SPI buffer. The buffer that contains an SPI message. See also SPI message. SPI message. In DSM programmatic interfaces, a message specially formatted by the SPI procedures for communication between a management application and a subsystem or between one subsystem and another. An SPI message consists of a collection of tokens. Note that an SPI message is a single block of information sent at one time, as one interprocess message. There are two types of SPI messages, distinguished by different sets of tokens in the header: command and response messages, and event messages. SPI procedures. In DSM, the set of Guardian procedures used to build and decode buffers for use in system and network management and in certain other applications. SPI standard definitions. In DSM programmatic interfaces, the set of declarations available for use with the SPI procedures, regardless of the subsystem. There is also a set of subsystem-specific declarations for each subsystem, and some sets of declarations that apply to multiple subsystems. An application using SPI needs the SPI standard definitions and also the subsystem definitions for all subsystems with which it HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Glossary -17 Glossary subject token communicates. See also definition. Compare data communications standard definitions or EMS standard definitions. subject token. In event management, a device, process, or other named entity about which a given event message has information. subnet address. An extension of the Internet addressing scheme that allows a site to use a single Internet address for multiple physical networks. Outside of the site using subnet addressing, routing continues as usual by dividing the destination address into an Internet portion and local portion. Gateways and hosts inside a site using subnet addressing interpret the local portion of the address by dividing it into a physical network portion and host portion. subnetwork. One or more intermediate systems that provide relaying and through which end open systems may establish network connections. Subnetwork Access Protocol (SNAP). In order to run the TCP/IP protocol suite over IEEE networks, the Subnetwork Access Protocol (SNAP) defines the interface between the IP layer and the LLC layer. The interface is accomplished through the use of an extension of the LLC header that contains a predefined Service Access Point (SAP) for use in the Source SAP (SSAP) and the Destination SAP (DSAP) fields of the LLC header. subsystem. (1) The software and/or hardware facilities that provide users with access to a set of communications service. (2) For DSM, a program or set of processes that manages a cohesive set of objects. Each subsystem has a process through which applications can request services by issuing commands defined by that subsystem; in some cases, this process is the entire subsystem. Many subsystems also have interactive interfaces. Subsystem Control Facility (SCF). A part of DSM, used to provide a common, interactive management interface for configuring, controlling, and collecting information from HP data communications products. Subsystem Control Point (SCP). In DSM, the management process for all NonStop data communications subsystems. There can be several instances of this process. Applications using SPI send all commands for data communications subsystems to an instance of this process, which in turn sends the commands on to the manager processes of the target subsystems. SCP also processes a few commands itself. It provides security features, version compatibility, support for tracing, and support for applications implemented as fault-tolerant process pairs. See also management process or manager process. Subsystem ID (SSID). In DSM programmatic interfaces, a data structure that uniquely identifies a subsystem to SPI. It consists of the name of the owner of the subsystem (such as HP), a subsystem number that identifies that particular subsystem, and a subsystem version number. The subsystem ID is an argument to most of the SPI procedures. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Glossary -18 Subsystem Programmatic Interface (SPI) Glossary Subsystem Programmatic Interface (SPI). In DSM, a set of procedures and associated definition files used to define common message-based programmatic interfaces for communication between requesters and servers—for instance, in a management application. SPI includes procedures to build and decode specially formatted messages; definition files in Pascal, TAL, C, COBOL85, and TACL for inclusion in programs, macros, and routines using the SPI procedures; and definition files in DDL for programmers writing their own subsystems. summary state. In DSM interfaces to NonStop data communications subsystems, one of the generally defined possible conditions of an object, with respect to the management of that object. A summary state differs from a state in two ways. First, a summary state pertains to the management of an object, whereas a state may convey other kinds of information about the object. Second, summary states are defined in the same way for all NonStop data communications subsystems, whereas the set of possible states differs from subsystem to subsystem. The management programming interfaces to NonStop data communications subsystems refer to summary states rather than to states. Examples of summary states are STARTED, STOPPED, SUSPENDED, and ABORTING. SWAN concentrator. See ServerNet wide area network (SWAN) concentrator. symbolic name. In DSM programmatic interfaces, a name used in programs to refer to commonly used values, token codes, token maps, extensible structures, and other related variables for use in management programs. TCP. See Transmission Control Protocol (TCP). TELNET. The Internet standard protocol for remote terminal connection service. TELNET allows a user at one site to interact with remote timesharing systems at another site just as if the user's terminal is connected directly to the remote machine. That is, the user invokes a TELNET application program that connects to a remote machine, prompts for a login id and password, then passes keystrokes from the user's terminal to the remote machine and displays output from the remote machine on the user's terminal. TFTP. See Trivial File Transfer Protocol (TFTP). token. In DSM use, a distinguishable unit in a SPI message. Programs place tokens in an SPI buffer using the SSPUT or SSINIT procedures and retrieve them from the buffer with the SSGET procedure. A token has two parts: an identifying code, or token code, and a token value. In command and response messages, a token normally represents a parameter to a command, an item of information in a response, or control information for the subsystem. In event messages, a token normally represents an item of information about an event or about the event message itself. See also header token. token number. In DSM programmatic interfaces, the number used by a subsystem to identify each DSM token that it defines. The token type and the token number together form the token code. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Glossary -19 token ring Glossary token ring. 1)þþThe token access procedure used on a network with a sequential or ring topology. (2) A data link level protocol designed to transfer data over ring-oriented LANs. The token ring technique is based on the use of a particular bit pattern called a token that circulates around the ring when all stations are idle. token type. In DSM programmatic interfaces, the part of a DSM token code that identifies the data type and length of the token value. The token type and the token number together form the token code. token value. In DSM programmatic interfaces, the value assigned to a DSM token. Transmission Control Protocol (TCP). The Internet standard transport-level protocol that provides the reliable, full-duplex stream service on which many application protocols depend. TCP allows a process on one machine to send a stream of data to a process on another. It is connection-oriented, in the sense that before transmitting data participants must establish a connection. Software implementing TCP usually resides on the operating system and uses the IP protocol to transmit information across the Internet. It is possible to terminate (shut down) one direction of flow across a TCP connection, leaving a one-way (simplex) connection. The Internet protocol suite is often referred to as TCP/IP because TCP is one of the two most fundamental protocols. Trivial File Transfer Protocol (TFTP). The Internet standard protocol for file transfer with minimal capability and minimal overhead. TFTP depends only on the unreliable, connectionless datagram delivery service (UDP), so it can be used on machines like diskless workstations that keep such software in ROM and use it to bootstrap themselves. UDP. See User Datagram Protocol (UDP). User Datagram Protocol (UDP). The Internet standard protocol that allows an application program on one machine to send a datagram to an application program on another machine. UDP uses the Internet Protocol to deliver datagrams. Conceptually, the important difference between UDP and IP is that UDP messages include a protocol port number, allowing the sender to distinguish among multiple destinations (application programs) on the remote machine. In practice, UDP also includes a checksum over the data being sent. wait mode. In the Guardian operating system, the mode in which the called procedure waits for the completion of an I/O operation before returning a condition code to the caller. Compare nowait mode. WAN. See wide area network (WAN). WAN manager process. The WAN manager process starts and manages the WAN subsystem objects including the ConMgr and WANBoot processes. WAN subsystem. See wide area network (WAN) subsystem. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Glossary -20 Warning Glossary Warning. In DSM interfaces, a condition encountered in performing a command or other operation, that can be significant but does not cause the command or operation to fail. A warning is less serious than an error. Compare error. well-known port. Any of a set of protocol ports preassigned for specific uses by transport level protocols (that is, TCP and UDP). Servers follow the well-known port assignments so clients can locate them. Examples of well-known port numbers include ports assigned to echo servers, time servers, remote login (TELNET) servers, and file transfer (FTP) servers. wide area network (WAN). A network that operates over a larger geographical area than a local area network (LAN)—typically, an area with a radius greater than one kilometer. The elements of a WAN may be separated by distances great enough to require telephone communications. Contrast with local area network (LAN). wide area network (WAN) subsystem. The Subsystem Control Facility (SCF) subsystem for configuration and management of WAN objects in G-series RVUs. wild-card character. A character that stands for any possible character(s) in a search string or in a name applying to multiple objects. In DSM object-name templates, two wild-card characters can appear: question mark (?) for a single character and asterisk (*) for zero, one, or more consecutive characters. See also object-name template. X.25. The CCITT standard protocol for transport-level network service. Originally designed to connect terminals to computers, X.25 provides a reliable stream transmission service that can support remote login. X.25AM. See X.25 Access Method (X.25AM). X.25 Access Method (X.25AM). An HP product that implements, for WANs, the services of the Network Layer and layers below. X.25 network. Any network or subnetwork linked using X.25 standards. X.25 standards are CCITT standards that define packet switching carrier communication in the Network Layer over wide area networks (WANs). See also International Telecommunications Union Telecommunications (ITU-T) and packet switching. $ZZLAN. See LAN manager (LANMAN) process. $ZZWAN. See WAN manager process. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Glossary -21 Glossary See WAN manager process. HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Glossary -22 Index A ABORT command LISTNER, not supported 1-32 MON, example 1-28 MON, specification 5-12 MON, summary A-1 PROCESS, example 1-23, 1-28 ROUTE, specification 5-15 ROUTE, summary A-1 SUBNET, example 1-14 SUBNET, specification 5-16 SUBNET, summary A-1 TCPMAN process, specification 5-13 TCPMAN process, summary A-1 TCPSAM process 1-32 TCPSAM process, specification 5-14 TCPSAM process, summary A-1 TELSERV, not supported 1-32 Access list 6-1 ACK Predictions OK 5-100 ADD command DEFINE, HOSTS file 1-9, 1-17 DEFINE, hosts file 3-25 DEFINE, process 1-14 DEFINE, round-robin filtering 2-5 DEFINE, SRL 1-14, 1-19, 3-13 ENTRY, specification 5-17 ENTRY, summary A-1 ROUTE, example 1-14, 3-26 ROUTE, specification 5-20 SUBNET, example 1-14, 3-26 SUBNET, specification 5-21 ADD DEFINE, limiting port sharing 2-5 ADDALIAS attribute 5-31 Address Resolution Protocol See ARP table Alias 5-31, 5-33 ALLNETSARELOCAL 5-41 ALTER command HOSTID 1-14 HOSTNAME 1-14 MON, example 1-8, 1-14 MON, specification 5-25 SUBNET, example 1-8, 1-14, 3-26 SUBNET, specification 5-30 SUBNET, summary A-3 Applications and port sharing 2-5 and single IP host 2-2 binding 2-6 configuring subnet-level binding for 2-6 context maintaining, caution 2-6 in conventional TCP/IP 2-2 multiple instances of 2-3 path length from 2-1 routing to 2-1, 2-6 running in both environments 4-1 scaling 2-6 sharing ports among 2-2 socket-transport-service provider for 2-11 transparency for 2-6 using the SRL 2-11 Architecture, conventional TCP/IP 2-1 Arp Flags 5-125 ARP table adding to 5-17 deleting from 5-33 entry type 5-18 viewing 5-3, 5-36 Arp Timer 5-125 ASSOCIATESUB attribute 5-31 ASSUME command 1-8 ATM, restriction 2-16 Attribute 3-26 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Index -1 B Index B Backup CPU configuring 2-5, 3-14, 3-18, 3-20 distributor 3-20 LISTNER 3-14 TELSERV 3-18 Backward-compatibility 2-11 Bad checksum attribute 5-88, 5-106 Bad ICMP code attribute 5-88, 5-107 Bad ICMP packets attribute 5-88, 5-107 Bad router ADDR list attribute 5-88, 5-107 Bad router ADV subcode attribute 5-88, 5-107 Bad router words/ADDR attribute 5-89, 5-107 Balancing, load 3-1 Berkeley software design 2-16 Binding, subnet-level 2-6 BSD 4.4 2-16 Bytes Maximum 1-20 Bytes Used 1-20 C Caution applications bound to SRL without an open 1-21 binding the listener with INADDR_ANY 3-26 not stopping home terminal 1-20, 1-26 port collisions, among listening processes 2-5 SRL 2-12 UDP port sharing 2-6 Caution, SRL 2-15 Class map 2-5 CLEAR command ALL 1-8, 3-13 PARAM 1-17 Clearing system current configuration 1-24/1-28, 2-15 of DEFINEs and PARAMs 1-8, 3-13 CLOSE-WAIT socket state 5-127 CLOSING socket state 5-127 Coexistence Parallel Library and conventional TCP/IP 2-1 Coexistence, Parallel Library and conventional TCP/IP 4-1 Command file HOSTS 3-28 SCFSBNT 3-15 SCFSBNT2 3-26 TCPIPDN 1-23, 1-28 TCPIPUP (quick start) 1-8, 1-14 TCPIPUP2 (for LISTNER) 3-14 TCPIPUP3 (for TELSERV) 3-17 TCPIPUP4 (for ODBC) 3-20 TCPIPUP5 (for iTP WebServer) 3-22 TCPIPUP6 (for 2 gateways) 3-25 Commands See also individual command names list of 5-9 Comment, adding to file 3-25 CONFIG 2-15 Configuration database 2-15 default 2-4 distributor listening model 3-18/3-20 hybrid listening model 3-21/3-22 quick start 1-1/1-32 round-robin 2-4 round-robin filtering 2-5 standard listening model 3-11/3-15 standard listening model, two gateways 3-23/3-28 Configuration example 3-16, 3-18, 3-21, 3-23 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Index -2 D Index Configuring backup processors 2-5 Ethernet failover 2-8/2-9 master TCPMON 2-10 primary processors 2-5 round-robin, caution 2-5 Connections, incoming distribution 2-4 Considerations for listening processes 2-5 UDP port 2-6 Context, maintaining, caution 2-6 Conventional TCP/IP coexistence 2-1 coexistence with 4-1 data path 2-9 definition 2-1 differences, where to find 2-1 distributor listening model definition 3-6 distributor listening model figure 3-8 hybrid listening model definition 3-9 message-system hop 2-1, 2-10 monolithic listening model 3-4 monolithic listening model figure 3-5 monolithic listening model hybrid 3-10 multiple IP hosts 2-2 port sharing 3-1 CPU parameter 5-71 Current MBUFs used attribute 5-92, 5-111 Current pool allocation attribute 5-92, 5-111 D Data flow distributor listening model 3-6 hops 3-7 in hybrid listening model 3-9 monolithic listening model 3-4 shortening path-length for 3-8 standard listening model 3-2 Data MDs in use attribute 5-93, 5-111 Data path 2-9, 2-11 Data Predictions OK 5-101 Dead gateway detection 5-22 Debug attribute 5-40 Default configuration 2-4 routers, locating 5-22 segment file 1-20, 1-25 DEFINE adding 1-14 clearing 1-8, 1-17 HOSTS file 1-9, 1-17 limiting port sharing 2-5 SRL 1-14 DEFINE command in TACLLOCL, SRL 2-12 inheriting 2-12 Delay ack attribute 5-40 Delay ack time attribute 5-40 DELAY command 1-8 DELETE command DEFINE 1-17 DEFINE, example 1-8 ENTRY, specification 5-33 ENTRY, summary A-3 ROUTE, specification 5-34 ROUTE, summary A-3 SUBNET, specification 5-35 SUBNET, summary A-3 DELETE DEFINE 1-8, 3-13, 3-25 DELETEALIAS attribute 5-31 DETAIL command, PTrace 5-159 DETAIL parameter 5-38 Detailed UDP input records 5-177 Determining home terminal environment 1-20 Device type 5-156 Device type and subtype 5-156 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Index -3 E Index DEVICENAME attribute 5-22 description of 3-26 G-series considerations 5-24 Display example INFO PROCESS 5-36, 5-39, 5-44 INFO SUBNET 5-56, 5-59 LISTOPENS MON 5-61 LISTOPENS PROCESS 5-64 NAMES ROUTE 5-67 NAMES SUBNET 5-69 STATS PROCESS 5-77 STATS ROUTE 5-116, 5-118 STATS SUBNET 5-119, 5-122 STATUS PROCESS 5-129 STATUS ROUTE 5-134 STATUS SUBNET 5-136, 5-138 VERSION 5-154, 5-155 Distributed Systems Management 2-12 Distribution caution for UDP ports 2-6 incoming requests 2-3 of connections 3-6 of connections, hybrid listening model 3-10 round-robin 2-4 Distributor listening model definition 3-6 figure 3-8 Distributor, backup CPU configuration 3-20 DNS 1-10, 3-25 Domain Name Server See DNS DSM See also Distributed Systems Management management flow 2-14 Dup driver MDs in use attribute 5-93, 5-111 Dup MDs in use attribute 5-93, 5-111 Dynamic loading of SPRs 4-8 E Echo 2-11 ECHOSERV 3-25 Echo, determining name of opener 1-21 ENTRY and system configuration database 2-15 names 5-3 object 5-3 object hierarchy 5-2 object type 5-3 object type definition 5-3 specifying in ADD command 5-17 Environment, determining your home terminal’s 1-20 Error file already exists 1-21 NLD fatal 1-21 port collision 2-5 Errors attribute 5-89, 5-107 ESTAB socket state 5-127 Ethernet 4 ServerNet adapter sharing 4-1 support 2-1 Ethernet failover 2-7/2-9 Event messages B-1 Example TCPIPDN command file 1-23, 1-28 TCPIPUP command file 1-8 Expand application requirements for naming 2-15 transport-provider naming convention requirements 2-11 F Faddr attribute 5-128 FAILOVER attribute 5-23 Failover, Ethernet 2-7 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Index -4 G Index Fast Ethernet ServerNet adapter (FESA) sharing 4-1 support 2-1 FESA See Fast Ethernet ServerNet adapter (FESA) File already exists, error 1-21 Filter inbound frames and 2-11 key, round-robin 2-5 Filter Errors statistic 5-120, 5-123 Filter Timeout statistic 5-120, 5-123 Finger, determining name of opener 1-21 FINGSERV 3-25 FIN-WAIT-1 socket state 5-127 FIN-WAIT-2 socket state 5-127 Firewall workaround 5-24 Formats trace record header 5-164 trace records detailed UDP input records 5-177 interprocess communication 5-168 IP input records 5-179 IP output records 5-181 memory buffer allocation 5-168 route records 5-182 socket command records 5-183 socket creation 5-165 TCP records 5-169 UDP input records 5-176 UDP output records 5-178 UDP user request records 5-187 Fport attribute 5-128 FTP determining name of opener 1-21 using 1-9 FTPSERV 3-2, 3-12, 3-25 Full dump attribute 5-40 G G4SA See Gigabit Ethernet 4-port ServerNet adapter (GESA) Gateways description of 3-27 example of 3-23 Generic process checking if TCPMAN is 1-18 stopping 1-29/1-32 GESA See Gigabit Ethernet ServerNet adapter (GESA) Gigabit Ethernet 4-port ServerNet adapter (G4SA) 2-1, 3-18, 4-1 Gigabit Ethernet ServerNet adapter (GESA) 2-1, 4-1 Good routes recorded attribute 5-89, 5-107 H Header formats 5-164 HEX command 5-159 Hexadecimal format 2-11 Hierarchy, SCF objects 5-2 Home terminal 1-20, 1-25 Hop distributor listening model 3-7 elimination 3-8, 3-11 elimination of 3-2 in hybrid listening model 3-10 message-system 2-1, 2-10 message-system inter-process transfer 2-3 monolithic listening model 3-6 HOSTID altering 1-8, 1-14 definition 5-40 HOSTNAME altering 1-8, 1-14 definition 5-40 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Index -5 I Index HOSTS command file 3-28 HOSTS file altering 3-28 defining 1-9, 1-17 Hybrid listening model definition 3-9 figure 3-10 I ICMP 2-11 ICMP Router Discovery Protocol 5-22 identifying name of 2-16 In ARP Requests 5-113 In ARP requests attribute 5-95 In dest unreachable attribute 5-89, 5-107 In echo attribute 5-89, 5-108 In echo reply attribute 5-89, 5-108 In info reply attribute 5-89, 5-108 In info request attribute 5-90, 5-108 In parameter problem attribute 5-90, 5-108 In redirect attribute 5-90, 5-108 In source quench attribute 5-90, 5-109 In time exceeded attribute 5-90, 5-109 In timestamp attribute 5-91, 5-109 In timestamp reply attribute 5-91, 5-109 INADDR_ANY 2-4, 2-7, 3-26 INFO command ENTRY, specification 5-36 ENTRY, summary A-3 MON, specification 5-38 MON, summary A-3 PROCESS, example 1-18 ROUTE, specification 5-49, 5-52 ROUTE, summary A-4 SUBNET, specification 5-55, 5-58 SUBNET, summary A-4 TCPMAN process, specification 5-43 TCPMAN process, summary A-3 TCPSAM process, specification 5-44 TCSAM process, summary A-3 INITIAL TTL attribute 5-42 Input Errors statistic 5-120, 5-123 Input packets dropped attribute 5-104 Input Packets Dropped statistic 5-84 Input Packets statistic 5-120, 5-123 Interprocess communication records 5-168 Inter-process communication (IPC) 3-6 Introduction, Parallel Library TCP/IP 2-1/2-16 Invalid header size attribute 5-91, 5-110 IOMF, restriction 2-16 IP address 5-4 IP alias 5-31, 5-33 IP hosts conventional TCP/IP 2-2 in hybrid listening model 3-10 Parallel Library TCP/IP 2-2 IP input records 5-179 IP output records 5-181 IPADDRESS 5-125 IPADDRESS attribute 5-18, 5-37 IPC See Inter-process communication IRDP attribute 5-22 K Keep Alive Idle 5-40 Keep Alive Retry Cnt 5-40 Key, filter 2-5 Kseg2 memory segment 2-13 L LABEL command 5-160 Laddr attribute 5-128 LAN drivers/interrupt handlers 2-13 LAST-ACK socket state 5-127 Library invocation of, by applications 2-12 message-system transfer 2-3 role 2-15 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Index -6 M Index LIF definition 2-1 determining name of 1-4, 1-11 in hybrid listening model 3-10 sharing 4-1 LISTDEV 1-4, 1-16, 1-19, 1-22, 1-27, 1-32, 2-16 LISTEN socket state 5-127 Listening models 3-2 distributor, definition 3-6 distributor, figure 3-8 hybrid, definition 3-9 hybrid, figure 3-10 monolithic 3-4 monolithic, figure 3-5 standard, configuration example 3-11 standard, figure 3-12 standard, startup file 3-13 Listening processes distribution among 2-4 distributor 3-6 hybrid 3-9 monolithic 3-4 port collisions, caution 2-5 standard listening model 3-2 UDP port considerations 2-6 LISTNER adding 3-25 backup CPU configuration 3-14 example with two gateways 3-24 standard listening model, example of 3-2 starting 1-14, 3-14 LISTOPENS command MON, identifying applications using TCP/IP 1-21 MON, specification 5-60 MON, summary A-4 PROCESS 1-21 PROCESS, example 1-25 LISTOPENS command (continued) PROCESS, specification 5-63 TCPSAM process, specification 5-63 TCPSAM process, summary A-4 Loading SPRs 4-8 Load-balancing, definition 3-1 Locating default routers 5-22 Logical interface, definition 2-1 LOGON, SUPER.SUPER 1-15, 1-18 LOOP0 altering 1-8 reserved name 5-7 Loopback altering IP address for 1-8 altering to correct address 3-27 default name 5-7 establishing 1-8 LOOP0 1-8 stopping 1-8 Lport attribute 5-128 M MACADDR attribute 5-18 MacAddress attribute 5-37 Manager process, definition 2-10 Managing Parallel Library TCP/IP 4-1/4-8 Master TCPMON, assignment 2-10 Max dup driv MDs used attribute 5-93, 5-112 Maximum data MDs used attribute 5-93, 5-111 Maximum dup MDs used attribute 5-93, 5-111 Maximum MBUFs used attribute 5-93, 5-112 Maximum pool allocation attribute 5-93, 5-112 MAX-EPHEMERAL-PORT attribute 5-29, 5-42 MBUF allocation fails attribute 5-93, 5-112 MD queue limits attribute 5-93, 5-112 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Index -7 N Index Media state attribute 5-137 Media state down attribute 5-121 Memory buffer allocation records 5-168 Memory management 2-13 Message system 2-14 Message system hop 2-10 Messages, context maintenance for, caution 2-6 Message-system hop 2-1 Migration, transparency 2-6 MIN-EPHEMERAL-PORT attribute 5-29, 5-42 MON definition 5-4 names 5-4 object 5-4 object type 5-4 Monolithic listening model definition 3-4 figure 3-5 Multicast groups 5-57, 5-128 Multicast groups attribute 5-133 N Name attribute 5-37 NAMES command ENTRY, specification 5-66 ENTRY, summary A-4 ROUTE, specification 5-67 ROUTE, summary A-4 SUBNET, specification 5-69 SUBNET, summary A-4 Names, suggested 5-7 Naming conventions SCF 5-7 TCPSAM 2-11, 5-3 Network file system, restriction 2-16 NFS See Network file system NLD fatal error 1-21 No data MDs avail attribute 5-93, 5-112 No dup driv MDs avail attribute 5-93, 5-112 No dup MDs avail attribute 5-94, 5-112 NOBULKIO 5-144 Nonsensitive commands INFO 5-36 listed 5-11 LISTOPENS 5-60 NAMES 5-66 STATS without RESET option 5-75 STATUS 5-123 VERSION 5-152 Null object 5-2, 5-4 O OBEY file HOSTS 3-28 SCFSBNT 3-15 SCFSBNT2 3-26 TCPIPDN 1-23, 1-28 TCPIPUP (quick start) 1-8, 1-15 TCPIPUP2 (for LISTNER) 3-14 TCPIPUP3 (for TELSERV) 3-17 TCPIPUP4 (for ODBC) 3-20 TCPIPUP5 (for iTP WebServer) 3-22 TCPIPUP6 (for 2 gateways) 3-25 Object specifiers 5-7 Object types descriptions 5-2/5-4 ENTRY 5-3 MON 5-4 null 5-4 PROCESS 5-4 ROUTE 5-5 SUBNET 5-6 Object-name templates, definition 5-7 OCTAL command 5-160 Online help 2-16 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Index -8 P Index Openers of TCPMONs 1-21, 1-26 of TCPMONs, stopping 1-28 of TCP/IP process 1-21, 1-25 Operator messages B-1 OSIMAGE file 2-12 Out dest unreachable attribute 5-91, 5-110 Out echo attribute 5-91, 5-110 Out echo reply attribute 5-91, 5-110 Out info reply attribute 5-92, 5-110 Out info request attribute 5-92, 5-110 OUT parameter in RUN command 1-8, 3-13 to file 5-13 Out parameter problem attribute 5-92, 5-110 Out redirect attribute 5-92, 5-110 Out source quench attribute 5-92, 5-110 Out time exceeded attribute 5-92, 5-110 Out timestamp attribute 5-92, 5-111 Out timestamp reply attribute 5-92, 5-111 Output Errors statistic 5-120, 5-123 Output packets dropped attribute 5-84, 5-104 Overview Parallel Library TCP/IP 2-1/2-16 PTrace 5-155 SCF for Parallel Library TCP/IP 5-1/5-12 P Packets data path 2-9 inbound 2-11 incoming request distribution 2-3 inter-process transfer, conventional TCP/IP 2-3 library transfer 2-3 missed, avoiding 2-6 request processing speed 2-2 request-latency reduction 2-2 Packets (continued) routing 2-1 Packets too short attribute 5-91, 5-110 Pages allocated 1-20, 1-25 Pages Maximum 1-20, 1-25 Parallel Library TCP/IP advantages 2-2, 2-6, 2-11, 3-1, 3-4, 3-8 and other products 2-15 application instances, single IP 2-2 caution, private SRL 2-12 circumventing missed packets in context-maintaining listening applications 2-6 coexistence with conventional 4-1 components 2-13 current restrictions 2-16 data path 2-9 differences, where to find 2-1 distributor listening model definition 3-6 hybrid listening model definition 3-9 introduction 2-1/2-16 library functionality 2-3 managing 4-1/4-8 monolithic listener model 3-4 monolithic listening model figure 3-5 path-length reduction 2-1 place in system 2-14 port sharing 3-1 product modules 2-9 request latency 2-2 round-robin filtering 2-4 selecting as environment 2-15 shared listening port 2-2 single IP host 2-2 Standard listening model definition 3-2 figure 3-3 standard listening model configuration example 3-11 standard listening model figure 3-12 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Index -9 Q Index Parallel Library TCP/IP (continued) standard listening model startup file 3-13 startup files for standard listening model 3-11 throughput 2-2 transparency for existing applications 2-6 Parallel, definition 3-1 PARAM adding 1-14 clearing 1-8, 1-17 clearing all 1-17 Password, filter key 2-5 PATHSEND 3-6 Pathway 3-6, 3-9 Path-length reduction 2-1, 2-11 Persistence manager and system configuration database 2-15 function 4-3 starting using 1-18 Physical port, definition 2-1 PIF definition 2-1 distributor listening model 3-7 sharing 2-2 PMF, restriction 2-16 Pool allocation fails attribute 5-94, 5-112 Port binding to 3-8 definition 2-1 limiting sharing of 2-5 ownership 3-2 sharing, caution 2-5 sharing, round-robin filtering 2-3 UDP sharing considerations 2-6 well-known, binding to 3-6 Port filters drop statistic 5-121 Port sharing, limiting 2-5 PORTCONF 3-25 PPID attribute 5-127 PRIMARY command definition 5-70 TCPMAN process, specification 5-70 TCPMAN process, summary A-5 TCPSAM process, specification 5-71 TCPSAM process, summary A-5 Primary CPU configuring 2-5 home terminal 1-20, 1-25 Private SRL, caution 2-12 PROCESS names 5-4 object 5-4 object type 5-4 parameter 5-44 starting 3-11 Process defining as transport provider 1-14 defining for TELSERV 1-14 Process-create command, for SRL 2-12 Program column, LISTDEV display 1-22 Program file name attribute 5-40 Programmatic interface to SRL 2-12 Protocol attribute 5-127 PTCPIP online help for 2-16 subsystem name 2-1 PTCPIP^FILTER^TCP^PORTS 2-5 PTCPIP^FILTER^UDP^PORTS 2-5 PTrace commands 5-157/5-163 overview 5-155 product module 2-9, 2-12 Q QIO 2-13 QIO configuration 2-13 QIO driver errors attribute 5-94, 5-112 QIO limit attribute 5-40 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Index -10 R Index QIO limit warnings attribute 5-94, 5-112 QIO memory space 2-13 QIOMON, determining if running 1-3, 1-11, 1-16 QIO, determining where it runs in flat memory segment 2-13 QIO, reserved segment 2-12 QIO, use of 2-11 Quick start, configuration 1-1/1-32 R Recording and displaying trace data 5-155 RecvQ attribute 5-128 Reflect packets attribute 5-91, 5-110 Requests incoming 2-3 latency reduction 2-2 processing speed 2-2 Reserved names for ROUTEs 5-7 LOOP0 5-7 #ZPTM 5-7 $ZZTCP 5-7 RESERVEDIP attribute 5-31 RFC BSD 4.4 2-16 compliance 2-16 RFC1323-ENABLE 5-41 Round-robin filtering and shared ports 2-3 configuration 2-5 configuring, caution 2-5 default configuration 2-4 definition 2-3, 2-4 for listening models 3-1 in hybrid listening model 3-10 in monolithic listening model 3-4 monolithic listening model 3-6 port-sharing considerations 2-6 ROUTE and system configuration database 2-15 attribute 5-34 definition 5-5 names 5-5 object 5-2 object type 5-5 Route records 5-182 Router advertisement attribute 5-92, 5-111 Router Discovery Protocol (ICMP) 5-22, 5-31 Router solicitation attribute 5-92, 5-111 Router, adding for firewalls 5-24 Route, example explanation 3-27 ROUTE, starting 3-11 Routing packets, in Parallel Library TCP/IP 2-1 RUN command, LISTNER 3-14 S Saving, current configuration 1-19/1-23 Scalable, definition 2-6, 3-1 SCF command summary A-1/A-8 determining if running 1-3, 1-10, 1-16 help 2-16 INLINE 1-8 LISTDEV command 1-4 object hierarchy 5-2 overview 2-12 product module 2-12 SCF commands ABORT 5-12 ADD 5-17 ALTER 5-25 DELETE 5-33 INFO 5-36 LISTOPENS 5-60 NAMES 5-34, 5-66 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Index- 11 S Index SCF commands (continued) PRIMARY 5-70 START 5-72 STATS 5-75, 5-115 STATUS 5-123 STOP 5-139 TCPSAM 5-130 TRACE 5-143 VERSION 5-152 SCFSBNT command file 3-15 SCFSBNT2 command file 3-26 SCP 2-14 SECONDARYROUTES attribute 5-134 SEL and SUM options, not supported by TCP/IP 5-11 SELECT command 5-161 SendQ attribute 5-128 Sensitive commands 5-11 ABORT 5-12 ADD 5-17 ALTER 5-25 DELETE 5-33 PRIMARY 5-70 START 5-72 STATS command with RESET option 5-75 STOP 5-139 TRACE 5-143 ServerNet 2-1 shared PIFs 2-2 use of 2-1 Shared runtime library See SRL Sharing ports among listening processes, caution 2-5 UDP considerations 2-6 Short IP packets attribute 5-91, 5-110 Single IP host definition 2-2 in hybrid listening model 3-10 Size 1025-2048 socket statistic 5-94, 5-113 Size 12289-16384 socket statistic 5-95, 5-113 Size 129-256 socket statistic 5-94, 5-113 Size 16385-32768 socket statistic 5-95, 5-113 Size 1-128 socket statistic 5-94, 5-113 Size 2049-4096 socket statistic 5-94, 5-113 Size 257-512 socket statistic 5-94, 5-113 Size 4097-8192 socket statistic 5-94, 5-113 Size 513-1024 socket statistic 5-94, 5-113 Size 8193-12288 socket statistic 5-94, 5-113 SLSA description of subsystem 2-13 determining if running 1-4, 1-16 determining LIF name 1-4 objects 2-1 relationship to 2-13 SNAP, restriction 2-16 Socket access method See TCPSAM Socket command records 5-183 Socket creation records 5-165 Socket-transport-service provider 2-16 for selecting Parallel Library TCP/IP 2-15 TCPSAM, definition Spawning processes 3-2 Spawning, FTPSERV 3-11 SPRs, loading 4-8 SRL caution 2-12, 2-15 defining 1-14, 1-17, 1-19 defining in TACLLOCL 2-12 deleting 1-8, 1-17 locating 2-12 process-create 2-12 product module 2-9 programmatic interfaces to 2-12 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Index -12 S Index Standard listening model configuration example 3-11 definition 3-2 figure 3-3, 3-12 startup file 3-13 START command MON, example 1-8 MON, specification 5-72 MON, summary A-5 PROCESS, example 1-18 ROUTE, example 1-14, 3-26 ROUTE, specification 5-73 ROUTE, summary A-5 SUBNET, example 1-14, 3-26 SUBNET, specification 5-74 STARTED, multicast state 5-129, 5-133 Starting LISTNER 1-14 loopback 1-8 TCPMAN 1-8 TCPMON 1-8 TELSERV 1-14 using DNS 1-10/1-15 using HOSTS file 1-3/1-10 using persistence manager 1-18 using RUN command 1-15/1-18 STARTING, multicast state 5-129, 5-133 Startup files standard listener 3-13 Startup files for LAN-based connections 3-11 State attribute 5-57 socket status 5-127 summary, defined 5-8 STATS command MON, specification 5-76 MON, summary A-5 ROUTE, display example 5-118 ROUTE, specification 5-115, 5-117 STATS command (continued) ROUTE, summary A-5 SUBNET, display example 5-122 SUBNET, specification 5-118, 5-121 SUBNET, summary A-6 TCPSAM process 5-96/5-115 TCPSAM process, summary A-5 Status attribute 5-127 STATUS command ENTRY, specification 5-124 ENTRY, status A-6 MON, specification 5-125 MON, summary A-6 PROCESS example 1-4 PROCESS, example 1-16, 1-18 ROUTE, specification 5-133 ROUTE, summary A-6 SUBNET, specification 5-136, 5-137 SUBNET, summary A-6, A-7 TCPMAN process, specification 5-129 TCPMAN process, summary A-6 TCPSAM process 5-130/5-133 TCPSAM process, summary A-6 STOP command 5-139 LISTNER process 1-32 MON, specification 5-139 MON, summary A-7 ROUTE, specification 5-141 ROUTE, summary A-7 SUBNET, example 1-8, 3-26 SUBNET, specification 5-142 SUBNET, summary A-7 TCPMAN process, specification 5-140 TCPMAN process, summary A-7 TCPSAM process, specification 5-140 TCPSAM process, summary A-7 TELSERV process 1-32 STOPPED, multicast state 5-129, 5-133 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Index -13 T Index Stopping generic process 1-29/1-32 openers of TCPMON 1-28 preserving configuration 1-19/1-23 without preserving configuration 1-24/1-28 SUBNET and system configuration database 2-15 attribute 5-35 definition 5-5 maximum configurable 2-7, 5-23 names 5-5 object type 5-2, 5-6 starting 3-11 Subnet mask 3-27 SUBNETMASK attribute 5-30 Subnet-level binding 2-4 Subnet-level binding, and applications 2-6 Subsystem name PTCPIP 2-1 TCPIP 2-1 Summary states 5-8 SUPER.SUPER logon 1-15, 1-18 SWAN configuration 3-29 SYN-RCVD socket state 5-128 SYN-SENT socket state 5-128 System configuration database clearing 2-15 description of 2-15 managing 4-2 T TACL process 1-20, 1-25 RUN command 1-17 starting 1-9, 1-18 WHO command 1-20 Task summary starting using DNS 1-10 starting using HOSTS file 1-3 starting using persistence manager 1-18 starting using RUN 1-15 stopping generic process 1-29 stopping, clearing configuration 1-24 stopping, preserving configuration 1-19 Tasks starting Parallel Library TCP/IP using persistence manager 1-18 starting Parallel Library TCP/IP using RUN command 1-16/1-18 starting Parallel Library TCP/IP with DNS 1-10/1-15 starting Parallel Library TCP/IP with HOSTS 1-3/1-10 stopping Parallel Library TCP/IP as a generic process 1-29/1-32 stopping Parallel Library TCP/IP, clearing the configuration 1-24/1-28 stopping Parallel Library TCP/IP, preserving configuration 1-19/1-23 TCP filters dereg statistic 5-120 TCP filters error statistic 5-120 TCP filters reg statistic 5-120 TCP records 5-169 TCP send space attribute 5-39 TCPCOMPAT42 attribute 5-41 TCPIP 2-1 TCPIPUP command file 1-8 TCPIPUP1 3-13 TCPIPUP2 command file 3-14 TCPIPUP3 command file 3-17 TCPIPUP4 command file 3-20 TCPIPUP5 command file 3-22 TCPIPUP6 command file 3-25 TCPIP, subsystem name 2-1 TCPIP^HOST^FILE 3-25 TCPIP^HOST^FILE, deleting 1-17 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Index -14 T Index TCPIP^PROCESS^NAME adding 1-14, 1-17 deleting 1-8, 1-17 TCPLIB 2-9, 2-10 TCPMAN ABORT command 5-13 checking if generic process 1-18 definition 2-10 determining if running 1-4, 1-16, 1-19 INFO command 5-43 master, assignment 2-10 object types supported by 5-2 PRIMARY command 5-70 product module 2-9 SCF object hierarchy 5-3 starting 1-8 STATUS command 5-129 STOP command 5-140 TRACE command 5-145 VERSION command 5-153 TCPMON definition 2-11 determining if running 1-19 determining openers of 1-21, 1-26 identifying applications using 1-21 in data path, explanation 2-10 product module 2-9 stopping openers of 1-28 TCPPATHMTU attribute 5-41 TCPSAM ABORT command 5-14 defining 1-17, 1-19 definition 2-11 determining name of 1-20, 1-22, 2-16 determining the name of 2-16 INFO command 5-44 LISTOPENS command 5-63 naming conventions 2-11 object types supported by 5-2 TCPSAM (continued) PRIMARY command 5-71 product module 2-9 programming with 2-16 restrictions 2-11 SCF object hierarchy 5-3 starting 1-14, 2-15 STATS command 5-96/5-115 STATUS command 5-130/5-133 STOP command 5-140 TRACE command 5-148 VERSION command 5-154 TCPTIMEWAIT attribute 5-41 TCP-INIT-REXMIT-TIMEOUT 5-42 TCP-LISTEN-QUE-MIN 5-42 TCP-MIN-REXMIT-TIMEOUT 5-42 TCP/IP conventional See Conventional TCP/IP interactive management interface to SCF 2-12 programmatic management interface to SCF 2-12 TCP/IP process defining for applications 1-14 defining for TELSERV 1-14 determining openers of 1-21, 1-25 PARAM for TELSERV 1-14 TEDIT 1-23 TELNET determining name of opener 1-21 entering system with 1-15, 1-18 TELSERV adding a PARAM for 1-14 backup CPU configuration 3-18 example of monolithic listener 3-4 primary and backup configuration 2-5 starting 1-14 Templates for object names 5-7 TERM $ZHOME 1-8, 3-13 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Index -15 U Index Testing, running Parallel Library TCP/IP 1-15 TEXT command 5-163 TIME-WAIT socket state 5-128 Token ring support 2-1 Token-ring, restriction 2-16 Total MBUFs allocated attribute 5-94, 5-112 TRACE command MON, specification 5-143 MON, summary A-7 SUBNET, specification 5-150 SUBNET, summary A-8 TCPMAN process, specification 5-145 TCPMAN process, summary A-7 TCPSAM process, specification 5-148 TCPSAM process, summary A-8 Trace filename attribute 5-41 Trace record formats detailed UDP input records 5-177 header 5-164 interprocess communication 5-168 IP input records 5-179 IP output records 5-181 memory buffer allocation 5-168 route records 5-182 socket command records 5-183 socket creation 5-165 TCP records 5-169 UDP input records 5-176 UDP output records 5-178 UDP user request records 5-187 Trace status attribute 5-41 Tracing data records 2-12 Transparency for existing applications 2-6 TCPSAM provided for 2-11 Transport provider See Socket-transport-service provider TYPE ARP attribute 5-18 TYPE attribute 5-22 U UDP filters dereg statistic 5-121 filters error statistic 5-121 filters reg statistic 5-121 input records 5-176 output records 5-178 port-sharing considerations 2-6 receive space attribute 5-39 send space attribute 5-39 user request records 5-187 UNKNOWN socket state 5-128 V VERSION command MON, specification 5-152 MON, summary A-8 TCPMAN process, specification 5-153 TCPMAN process, summary A-8 TCPSAM process, specification 5-154 TCPSAM process, summary A-8 W WAN configuration 3-29 WHO command 1-25 display 1-20 entering 1-20 Wild-card characters 5-7 Wild-Card support 5-7 X X.25, restriction 2-16 Z ZTCPSRL 2-11 ZTCPSRL, warning re replacing 6-2 ZTNT^TRANSPORT^PROCESS^NAME 114 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Index -16 Special Characters Index Special Characters #LOOP0 5-6 #ZPTMx 5-7 $SYSTEM.SYSTEM.TACLLOCL 1-9 $SYSTEM.ZTCPIP.HOSTS 3-25 $SYSTEM.ZTCPIP.PORTCONF 3-25 $ZHOME 1-8, 3-13 $ZM 1-3, 1-11, 1-16 $ZNET 1-3, 1-10, 1-16 $ZPM 2-15 $ZTCx 5-4 $ZZKRN.#ZZTCP 1-18 $ZZLAN 1-4, 1-11, 1-16 $ZZTCP 5-7 $ZZTCP.#ZPTMn 2-11 HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Index -17 Index Special Characters HP NonStop TCP/IP (Parallel Library) Configuration and Management Manual— 522271-006 Index -18