Linux SNMP: Process name table No response from remote host

[msbensonstk]

Ok, I see what you mean.

So, no I have to use the full path as the application that gets run is a java app with about 12 lines of arguments. So the only way I can see if the java process is running is to look for the presence of the applications path in the object table.

That said, I went ahead and ran the modified command with the same result:
Thu Oct 17 08:34:10 CDT 2013
ERROR: Process name table : No response from remote host '<host>'.
Thu Oct 17 08:34:21 CDT 2013

[abrist]

Alright, this script is odd. Try changing:

Code: Select all

my $TIMEOUT = 15;

To:

Code: Select all

my $TIMEOUT = 60;

And then pass in your desired timeout with:

Code: Select all

-t 20

It looks like either the script is doubling the timeout (I looked and could not see where it would be doing it - so the above timeout will be approx 40 seconds), or there is a bug with Net::SNMP.
Let us know how it goes.

[msbensonstk]

Well, that definitely did something:

date; /usr/local/nagios/libexec/check_snmp_process_wizard.pl -t 20 -H <host> --login=nagiosxi --passwd=<pass> --privpass=<pass> --protocols=sha,aes -n 'ContegoSPOP' -f -w '0,1' -c '0,1'; date
Fri Oct 18 08:45:49 CDT 2013
ERROR: Process name table : No response from remote host '<host>'.
Fri Oct 18 08:46:29 CDT 2013

It took the longer timeout (20) and doubled it. So, that fixed the timeout issue, however it doesn't seem to have helped the issue with the host not sending an SNMP response.

[abrist]

This behavior is rather odd as I can test the plugin against an snmpv3 (aes,sha) box here just fine. You did declare a full path (-f) but then did not use the full path in your command (-n), though I don't think that would effect the device responding.

[msbensonstk]

That's why the check is coded as
-n 'ProcessName' -f

-n causes the name to be used as a regex against the -f table. It works great on 95% of our systems. It's just a couple that are having problems.

[abrist]

Is there any differences worth mention between the functioning checks and the non functioning checks? I tried full path checks this and full regex matches against different processes this morning on my test box without issue.

[msbensonstk]

No. They are the same check (I have it defined in 1 place and assigned to multiple hosts through group inheritance). Some systems process the check correctly, some do not.

[abrist]

Are both servers in the scenario using the same shell, environment, etc?