Download TCP/IP (Parallel Library) Configuration and Management Manual

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