linux os process monitoring via snmp

[nbritton]

We have a set of appliances where i am unable to install an agent and need to monitor process cpu and memory that can be found in a mib table but referenced as the current running pid. I am trying to see how i can monitor for a process by its name. I know that by following this set of commands i am able to pull that information. But i can not see how i would do that in a template.

I am attempting to lookup the PID or could be pids for a process name to return the name and pid with the values of cpu and memory consumed by that resource.


C:\Program Files\netsnmp\bin>snmpwalk -v 1 -c "public" acw00040 1.3.6.1.2.1.25.4.2.1.2|findstr spoold
HOST-RESOURCES-MIB::hrSWRunName.51126 = STRING: "spoold"

C:\Program Files\netsnmp\bin>snmpwalk -v 1 -c "public" acw00040 1.3.6.1.2.1.25.4.2.1|findstr 51126
HOST-RESOURCES-MIB::hrSWRunIndex.51126 = INTEGER: 51126
HOST-RESOURCES-MIB::hrSWRunName.51126 = STRING: "spoold"
HOST-RESOURCES-MIB::hrSWRunID.51126 = OID: SNMPv2-SMI::zeroDotZero
HOST-RESOURCES-MIB::hrSWRunPath.51126 = STRING: "/usr/openv/pdde/pdcr/bin/spoold"
HOST-RESOURCES-MIB::hrSWRunParameters.51126 = ""
HOST-RESOURCES-MIB::hrSWRunType.51126 = INTEGER: application(4)
HOST-RESOURCES-MIB::hrSWRunStatus.51126 = INTEGER: runnable(2)

C:\Program Files\netsnmp\bin>snmpwalk -v 1 -c "public" acw00040 1.3.6.1.2.1.25.5.1.1|findstr 51126
HOST-RESOURCES-MIB::hrSWRunPerfCPU.51126 = INTEGER: 14069188
HOST-RESOURCES-MIB::hrSWRunPerfMem.51126 = INTEGER: 252857944 KBytes

Thoughts?

[LenR]

I think all you can do is create external scripts passed the process name and follow the steps above to return the values you want. Note that external scripts are expensive, so if you have to do a lot of them, it's not a good idea.

Something like this for item keys:

ext.hrSWRunPerfCPU(spoold)
ext.hrSWRunPerfMem(spoold)