Nagios checks fail

[Tom Haddon]

Using cs:squid-reverseproxy-7, and adding a relation with nrpe, the outcome is a nagios check of:

# check squidpeers
command[check_squidpeers]=/usr/lib/nagios/plugins/check_squidpeers -p 3128 -w 50

Running this commands gives:

ubuntu@juju-f66069-charm-testing-squid-reverseproxy-trusty-20-1:~$ /usr/lib/nagios/plugins/check_squidpeers -p 3128 -w 50
Traceback (most recent call last):
  File "/usr/lib/nagios/plugins/check_squidpeers", line 87, in <module>
    sys.exit(main())
  File "/usr/lib/nagios/plugins/check_squidpeers", line 81, in main
    retcode, message = get_status(peers, args.weight)
  File "/usr/lib/nagios/plugins/check_squidpeers", line 51, in get_status
    perc = calculate_perc_up(peers)
  File "/usr/lib/nagios/plugins/check_squidpeers", line 46, in calculate_perc_up
    return ((reduce(iadd, [1. for i in peers if i['status'] == STATUS_UP]))
TypeError: reduce() of empty sequence with no initial value

[William Grant (wgrant)]

The linked branch fixes the crash but still makes the check fail when there are no peers. This is problematic in some use cases, e.g. where mojo is deploying base infrastructure like squid first, verifying it's all working, then relating app services, meaning that during the first verify there are legitimately no peers.