Configuring Polling Intervals

You can configure the HP Insight Management Agents polling intervals to accommodate your data collection needs. However, HP recommends that you use the default polling intervals unless you are performance tuning your system.

You can list the process status table for all of the agents delivered with HP Insight Management Agent using the following command:

# /etc/init.d/hpima status
hpimaX (pid 29032) is running...
hpimahostd (pid 29547) is running...
hpimathreshd (pid 29557) is running...
hpimahealthd (pid 29570) is running...
hpimastdeqd (pid 29580) is running...
hpimaeventd (pid 29590) is running...
hpimaidad (pid 29608) is running...
hpimasasd (pid 29619) is running...
hpimascsid (pid 29631) is running...
hpimafcad (pid 29643) is running...
hpimanicd (pid 29661) is running...
hpimaE0traps (pid 29675) is running...

The list contains the process name, the ProcessID (PID) allocated for each process, and the process status. Depending on what hardware is present on a system, agents start and stop to monitor the health and inventory of that hardware then report and alert as to the condition of the system through SNMP traps. Many of these agents are running unique periodic polling loops that typically default to 15 or 30 seconds. These polling intervals are optimized based on the best means for error tracking and reporting of components monitored by the agent.

Changing a polling interval to an infrequent interval affects how frequently each agent attempts to query the system. In other words, a slower polling interval equates to obtaining less frequent information about the system. It is possible that critical issues (such as temperature and voltage spikes, fan failures, CPU failures, etc.) may be overlooked and unreported if the polling intervals are set too long.

Table 2-1 lists the available HP Insight Management Agents that can be queried or altered by name, process name, and the Object Identifier (OID):

Each OID listed in Table 2-1 is used with the snmpset or snmpget commands to view or modify the polling time for the corresponding agent. These SNMP commands also require the identification of your system's rwcommunity string. This string is a password that allows you access to read or write SNMP OIDs.

Obtaining the Community String

You can obtain the rwcommunity string by viewing the /etc/snmp/snmpd.conf on your system and finding the line similar to the following:

rwcommunity lYABuwPNFgmVCSkH   127.0.0.1

In this example, the rwcommunity string is set to lYABuwPNFgmVCSkH.

Viewing a Polling Interval

You can view the current polling interval for an agent with the snmpget command as follows:

# snmpget -mALL -v1 -c<community string> localhost <OID>.0

For example, to view the NIC agent polling interval and using the community string in the previous example, the command is as follows:

# snmpget -mALL -v1 -c lYABuwPNFgmVCSkH localhost cpqNicOsCommonPollFreq.0 CPQNIC-MIB::cpqSeOsCommonPollFreq.0 = INTEGER: 52

The response indicates that a 30 second polling interval is set for the NIC agent.

Changing a Polling Interval

The valid range that you can set polling intervals to is 0 - 65535 seconds. Any value within this range is valid and values outside the range are rejected. You can change the polling interval using either of the following methods:

Using the snmpset Command:

You can change the current polling interval for each agent with the snmpget command as follows:

snmpset -mALL -v1 -c<community string> localhost cpq<OID>CommonPollFreq.0 = <new interval>

For example, to change the NIC agent polling interval to 52 seconds and using the community string in the previous example, the command is as follows:

# snmpset -mALL -v1 -c lYABuwPNFgmVCSkH localhost cpqNicOsCommonPollFreq.0 = 52 CPQNIC-MIB::cpqSeOsCommonPollFreq.0 = INTEGER: 52

The response indicates that your set request to change the NIC agent polling interval to 52 seconds was successful.

Using the Environment Variable:

The environment variable, HPIMA_POLLFREQ, can be set to a global polling interval to operate all HP Insight Management Agents with the same polling frequency. You can set this environment variable using the following command format:

# HPIMA_POLLFREQ=<new interval> /etc/init.d/hpima/restart

For example, to set all agent polling intervals to 60 seconds the command is as follows:

# HPIMA_POLLFREQ=60 /etc/init.d/hpima/restart

All HP Insight Management Agent polling intervals are now set to 60 seconds and each agent process is restarted by the /etc/init.d/hpima/restart command.

Additionally, you can use the HPIMA_POLLFREQ variable to set the polling frequency for an individual agent using the following command format:

# HPIMA_POLLFREQ=<new interval> /etc/init.d/hpima/restart <agent>

where:

agent

Is an agent string with the 'hpima' prefix removed from the process names as described in Table 2-1. For example, the agent for hpimanicd is nicd.

For example, to set all the NIC agent polling interval to 180 seconds the command is as follows:

# HPIMA_POLLFREQ=180 /etc/init.d/hpima/restart nicd

	NOTE: The HPIMA_POLLFREQ is only available on HP Integrity systems running Linux.

You can validate any changed agent polling intervals by using the snmpget command as described in “Viewing a Polling Interval”.

	NOTE: The snmpget/snmpset commands fail if the corresponding hardware is not present on the system. For example, if the system has no SCSI devices the hpimascsid process can not run so attempting to get or set the cpqScsiOsCommonPollFreq.0 agent fails.

	TIP: If you want to manually trigger a single polling loop for all agents, without effecting the actual OID polling interval settings, to refresh data when the server is in a known quiescent state, you can use the following command: # /etc/init.d/hpima sample [agent] The use of the agent string is optional and when used only the agent specified is polled rather than all agents.

At this time, you must use SNMP commands to change the polling interval values as this facility is not available within the HP System Management Homepage.