Monitor linux processes with parameters thru net-snmp

Question

I´m migrating from nagios and I´ve managed to move some checks to APM and the "Process Monitor - SNMP". This component monitor does only monitor the process name and can´t handle the process parameters, for example I have 5 java processes but with different parameter like:

/usr/bin/java --server=srv1 --someotherparam
/usr/bin/java --server=srv2 --someotherparam

This is easily done in nagios but I can´t find a good way to do this in Orion NPM.
(Note: I´m not allowing SSH to my servers, only snmp for security reasons)

I´ve been researching:

1) Process Monitor - SNMP: doesn´t handle process parameters

2) Create a universal device poller for each process to poll the HOST-RESOURCES-MIB::hrSWRunName hrSWRunParameters. Only one OID can be polled and + alerted however in Orion so this approach can´t be used since hrSWRunName and hrSWRunParameters has different OIDs.

3) Add the processes as PROC directives in snmpd.conf and send snmp traps if a process goes down.
Too much administrative work since different conf files on each server, hard to maintain.

4) Create a custom component monitor by copying the "Process Monitor - SNMP" and customizing it. Looks like this isn´t supported...(?). Could be an option if I knew how ...

Maybe I´m missing something since I´m quite new to Orion but how is this done best with SNMP?

Thanks for any help ...

pserwe · Answer

Yeah, I'd be interested in testing out the bash scripts, I'm running net-snmp on some linux boxen, and there's a few things I really care about.

APM, interestingly enough, isn't capable of telling me if the mail server is accepting messages for derivery..

Apparently capturing "200 OK Message accepted for delivery" is beyond the scope of what APM can currently handle.

I think it's funny that the stock answer when using fairly expensive software is "well, if you have the expertise to develop it yourself, you can display the results in our web page"..

Peter

FormerMember · Answer

See this thread:

It's something we're going to try to fix.

branfarm · Answer

This discussion is probably better suited for the APM forum. That being said,  what you are asking for is something I'm looking for too. I posted a long rant about this particular APM shortcoming over there.  What I've found is that you will either have to write a perl or bash script yourself, or find someone who can.   As long as you can find a way to get the individual pid, you should be able to work some magic using either ps or top to get the info you want. In fact, I'll go over the APM forum and post the bash script I'm testing right now so you can take a look. Maybe if we get enough people together we can get some scripts into the content exchange that will do what people need.