2

I created a simple script for nagios to check a port on a remote server. Originally, I tried to use lsof. 

#!/bin/bash

RESULT=$(lsof -i :8080 | grep LISTEN)

if [ -n "$RESULT" ]
  then
  echo "$RESULT"
exit 0
  else
   echo "Check application port is down"
exit 2
fi

On the local machine, the nagios user could run and execute the script and get the expected output. But remotely, using check_nrpe, I always got the else echo back

Check application port is down

Even though as I said running the script locally it worked properly

The nrpe user had all rights needed to execute the script, SELinux was in permissive mode, but I could never get it to work. So I changed the RESULT line to use netstat. 

RESULT=$(netstat -anp | grep ":8080" | grep LISTEN)

And then I could get the expected response back using check_nrpe remotely. So I assume this is some sort of problem with check_nrpe and lsof. But I could never find anything in the logs.

Anyone have any ideas or suggestions to get this to work right with lsof?

chicks
  • 3,793
  • 10
  • 27
  • 36
user53029
  • 629
  • 3
  • 14
  • 36

1 Answers1

3

Non-root users don't see any output from lsof -i. (Confirmed on Debian 8 and Cent 7, without SELinux.)

If you wanted to use this in a nagios check, you'd have to run lsof through sudo.

If sudo is configured to require a tty (might be the default on CentOS), you can do something like this:

Defaults:nagios !requiretty
nagios ALL=NOPASSWD:/usr/bin/lsof

Replace nagios with nrpe if necessary, and modify path to lsof if necessary.

Keith
  • 4,637
  • 15
  • 25
  • I gave the nagios user sudo for lsof and called sudo in the nrpe.cfg file. I even tried RESULT=$(sudo lsof -i :8080 | grep LISTEN) in the script and it still did not work. I confirmed sudo lsof worked locally with the account, just remotely it does not. – user53029 Apr 25 '16 at 23:49
  • See updated answer – Keith Apr 26 '16 at 18:07