Am monitoring ~100 remote hosts via a VPN using check_snmp_process.pl. For many months this has worked just fine. Over the weekend I started seeing ERROR: Alarm signal (Nagios time-out) errors from just about every host/process. I can use the command on the command line and get a successful response so I'm not clear why it would timeout under normal usage.

This morning I tried upping the 'timeout' param on the plugin to 20 seconds. For about an hour this appeared to work then in a matter of minutes the failure rate returned to its previous level.

The VPN server doesn't appear to be under any abnormal load. Nor does the nagios machine.

Suggestions on where else to look for the source of this?

Nagios machine: CentOS 6.5
Nagios version: 3.5.1
Plugin version: 1.10

EDIT: When the 'mass timeout' happens it's all within a few seconds. Every host shows the same time (+- 5 seconds) on the report. This may be due to nagios forcing rechecks on 'orphaned processes' from a restart of the service. Not sure yet. Just seems ominous when 40-50 timeouts hit the log all at once.

  • What's your average check latency and exec time? What's the load on the nagios box? – Keith Feb 17 '15 at 17:21
  • Also may want to check your routing tables between your VPN and remote systems. And try a few pings between some of them. If nothing has changed on this server, then I'd start looking at the network and/or VPN host. – Jim Black Dec 28 '15 at 21:45

I had the same issue but after editing the script check_snmp_process.pl time out from 15 to 40 it worked. my $TIMEOUT = 40;

  • 1
    It's just a workaround,doesn't sounds like an optimal solution? – Sachin Singh Aug 25 '16 at 11:03
  • I don't have control over the network, and this is a network latency issue so i have no choice but to increase the time out buffer on the script. – user372318 Aug 26 '16 at 8:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.