since upgrading to 5.5.1 snmp checks timeout

[rzietlow]

still trying to get this to work. when I try to recreate it, it goes quickly, but when I look at service problems I see it still pop up from time to time.

[scottwilkerson]

so was it improved on the command line adding the -t 30 ?

[rzietlow]

the -t 30 doesn't seem to make a difference right now(from the CLI), it's all around 1-2 seconds response, but I still see it flapping in the services problem pane.

[tgriep]

Another thing to try is to increase the octet length as it may not be a timeout issue but that the buffer is to small and that is causing the error.
Try adding this to the command

Code: Select all

-o 32768

and see if the General Time-out stops.

-o, --octetlength=INTEGER

[rzietlow]

ok thank you.

Other than having to add this on to every check having issues, is there a universal place i could put this in that wouldn't get clobbered in upgrades should I want it to be a part of every check?

[tgriep]

Yes, you can edit the command in the CCM which will add the octet length option to all of the checks that use that command.

To do this, go to the Core Config Manager > Commands menu and search for the command name called check_xi_service_snmp_win_storage.
Edit it and change the command line from

Code: Select all

$USER1$/check_snmp_storage.pl -H $HOSTADDRESS$ $ARG1$

to

Code: Select all

$USER1$/check_snmp_storage.pl -H $HOSTADDRESS$ -o 32768 $ARG1$

Save it and Apply the Config. That will globally change it.

[rzietlow]

ok so I'll have to do this for each check (not the end of the world) we have CPU memory and other stuff that shows up. The storage was just my example. It's not that big of a deal. The -o should work regardless of snmp_check?

[tgriep]

Most of the Perl SNMP plugins have the -o option but you can run the plugin from a shell with the --help option and it will tell you if the option is available.