custom nagios script with nrpe resulting in non-zero exit status 1

Question

I am trying to run a python script using nrpe to monitor rabbitmq. Inside the script is a command 'sudo rabbiqmqctl list_queues' which gives me a message count on each queue. However this is resulting in nagios giving htis message:

CRITICAL - Command '['sudo', 'rabbitmqctl', 'list_queues']' returned non-zero exit status 1 

I thought this might be a permissions issue so proceeded in the following manner

/etc/group:

ec2-user:x:500:
rabbitmq:x:498:nrpe,nagios,ec2-user
nagios:x:497:
nrpe:x:496:
rpc:x:32:

/etc/sudoers:

%rabbitmq ALL=NOPASSWD: /usr/sbin/rabbitmqctl

nagios configuration:

command[check_rabbitmq_queuecount_prod]=/usr/bin/python27 /etc/nagios/check_rabbitmq_prod -a queues_count -C 3000 -W 1500

check_rabbitmq_prod:

#!/usr/bin/env python
from optparse import OptionParser
import shlex
import subprocess
import sys

class RabbitCmdWrapper(object):
    """So basically this just runs rabbitmqctl commands and returns parsed output.
       Typically this means you need root privs for this to work.

       Made this it's own class so it could be used in other monitoring tools
       if desired."""

    @classmethod
    def list_queues(cls):
        args = shlex.split('sudo rabbitmqctl list_queues')
        cmd_result = subprocess.check_output(args).strip()
        results = cls._parse_list_results(cmd_result)
        return results

    @classmethod
    def _parse_list_results(cls, result_string):
        results = result_string.strip().split('\n')
        #remove text fluff
        results.remove(results[-1])
        results.remove(results[0])
        return_data = []
        for row in results:
            return_data.append(row.split('\t'))
        return return_data

def check_queues_count(critical=1000, warning=1000):
    """
    A blanket check to make sure all queues are within count parameters.
    TODO: Possibly break this out so test can be done on individual queues.
    """
    try:
        critical_q = []
        warning_q = []
        ok_q = []
        results = RabbitCmdWrapper.list_queues()

        for queue in results:
            if queue[0] == 'SFS_Production_Queue':
                count = int(queue[1])
                if count >= critical:
                        critical_q.append("%s: %s" % (queue[0], count))
                elif count >= warning:
                        warning_q.append("%s: %s" % (queue[0], count))
                else:
                        ok_q.append("%s: %s" % (queue[0], count))
        if critical_q:
            print "CRITICAL - %s" % ", ".join(critical_q)
            sys.exit(2)
        elif warning_q:
            print "WARNING - %s" % ", ".join(warning_q)
            sys.exit(1)
        else:
            print "OK - %s" % ", ".join(ok_q)
            sys.exit(0)
    except Exception, err:
        print "CRITICAL - %s" % err
        sys.exit(2)

USAGE = """Usage: ./check_rabbitmq -a [action] -C [critical] -W [warning]
           Actions:
           - queues_count
             checks the count in each of the queues in rabbitmq's list_queues"""

if __name__ == "__main__":
    parser = OptionParser(USAGE)
    parser.add_option("-a", "--action", dest="action",
                      help="Action to Check")
    parser.add_option("-C", "--critical", dest="critical",
                      type="int", help="Critical Threshold")
    parser.add_option("-W", "--warning", dest="warning",
                      type="int", help="Warning Threshold")
    (options, args) = parser.parse_args()

    if options.action == "queues_count":
        check_queues_count(options.critical, options.warning)
    else:
        print "Invalid action: %s" % options.action
        print USAGE

At this point I'm not sure what is preventing the script from running. It runs fine via the command-line. Any help is appreciated.

Answer 1

The "non-zero exit code" error is often associated with requiretty being applied to all users by default in your sudoers file.

Disabling "requiretty" in your sudoers file for the user that runs the check is safe, and may potentially fix the issue.

E.g. (assuming nagios/nrpe are the users)

@ /etc/sudoers

Defaults:nagios !requiretty
Defaults:nrpe !requiretty