E. The LSF SNMP Agent

This appendix describes how to use the LSF SNMP (Simple Network Management Protocol) agent. It is assumed that you are already familiar with SNMP concepts.

About the Agent

To integrate with existing network and system management frameworks, LSF supports SNMP, an IETF (Internet Engineering Task Force) standard protocol used to monitor and manage devices and software on the network. Platform has also defined a Management Information Base (MIB) specific to LSF.

Any SNMP client, from command-line utilities to full network and system management frameworks, can monitor information provided by the LSF SNMP agent. It does this by reading or modifying the values of objects in the LSF MIB. The MIB defines these variables according to internet standards RFC 1155 and RFC 1212, so any fully compliant SNMP client can query the LSF MIB.

In addition, the LSF SNMP agent supports the Internet standard Management Information Base (MIB-II) for use in TCP/IP based internets (RFC 1213). This allows the network manager to run the LSF agent while still being able to retrieve basic MIB-II statistics.

Requirements

The agent can be installed on any LSF host running UNIX and it is compatible with all SNMP version 1 clients, including HP OpenView, CA Unicenter TNG, and Tivoli TME 10. For information about a particular network or system management framework, please refer to the documentation supplied by the vendor.

Distribution

The agent is available from Platform's FTP site. Installation instructions are included. The following files are provided by Platform:

• the SNMP agent (snmpd)
• the LSF MIB (LSF_CONFDIR/snmp/lsf-agent-mib.txt)
• a configuration file (LSF_CONFDIR/snmp/snmpd.conf)
• a script to start the SNMP agent (lsfsnmpd)

Starting the Agent

To simplify the startup process, the LSF SNMP binary file (snmpd)is accompanied by a script (lsfsnmpd). This script is customized to start the agent in a specific LSF environment. For example, it contains the location of the configuration file used by the agent, (LSF_CONFDIR/snmp/snmpd.conf), and the log file that is created when the agent runs (LSF_LOGDIR/snmpd.log).

The LSF administrator can modify the parameters in the lsfsnmpd script, or run the snmpd binary file without using the script.

Options which can be set in this file include:

-p port_number
Listen on port port_number (default: port 161).

-l log_file
Log all output from the agent (including stdout/err) to log_file.

-c conf_file
Read conf_file as a configuration file.

-f
Don't fork from the calling shell.

-L
Don't open a log file; use stdout/err instead.

Structure of the LSF MIB

The LSF MIB (LSF_CONFDIR/snmp/lsf-agent-mib.txt) consists of several tables of information, organized into three groups:

• LSF Base host information (lsfHosts group)
• LSF Base resource information (lsfResources group)
• LSF Batch information (lsfBatch group)

The lsfHosts MIB Group

lsfStaticTable
Consists of one row for each LSF host, indexed by host IP address. Each row contains static host information, corresponding to the lshosts command.

lsfDynamicTable
Consists of one row for each LSF server, indexed by host IP address. Each row contains dynamic host information, (corresponding to the lsload command) composed of the built-in load indices and the host status.

The lsfResources MIB Group

lsfNumericTable
Consists of several rows for each resource, one for each LSF host using that resource, indexed by resource number (generated by the agent) and host IP address. Each row contains the name of a numeric shared resource or external index, a location (a host using the resource), and the resource value.

For shared resources, the resource value is the same for all the hosts that share the resource instance.

The lsfBatch MIB Group

lsbHostsTable
Consists of one row for each LSF batch server, indexed by host IP address. Each row contains the host limits as well as the host counters.

lsbQueuesTable
Consists of one row for each LSF batch queue, indexed by a number generated by the agent (corresponding to alphabetical order of queue names). Each row contains queue limits and queue counters.

lsbJobsTable
Consists of one row for each running batch job, indexed by job ID (for performance reasons, only running jobs are displayed). Each row contains information such as the queue, user and execution hosts, and job resource information.

Optional Configuration of the Agent

The configuration file (LSF_CONFDIR/snmp/snmpd.conf) has the format of one directive per line. Lines preceded by the '#' character are treated as comments, and not parsed.

Directives which can be set in this file are:

syslocation string
This sets the system location for the agent in the system table of the MIB-II tree to string.

syscontact string
This sets the system contact for the agent in the system table of the MIB-II tree to string.

trapsink host
This sets the host host to receive traps (e.g., the agent sends a Cold Start trap when it starts up). To enable multiple hosts to receive traps, add a new line for each additional host. The default value is null (no hosts receive traps).

trapcommunity string
This sets the community string in the trap PDU (Protocol Data Unit) to string.

authtrapenable number
This enables the sending of authentication failure traps when number is set to 1 (enable). The default value is 2 (disable).

community number string
This sets the community string in slot number to string.

The agent has 5 slots available to keep community strings, so the acceptable values for number are from 1 to 5. SNMP PDUs sent to the agent should contain one of the communities in the 5 slots. The default values for each slot are:

1. public
2. private
3. regional
4. proxy
5. core

If the agent receives a PDU without a known community, it will discard the request, and if authtrapenable is set to 1, it will generate an authentication failure trap.