Well, the pidfile needs to point to path that appropriate system initscript uses.

Also, in my system, i found that i needed separate dir, because daemon runs as nagios user and that user can not write to /var/run, so i created /var/run/nrpe dir in that system.

which os/distro are you using? i'm using ubuntu 12.04 and only root can write to /var/run. Also nrpe runs as nagios user, but i think it creates the pid file still as root and afterwards does the daemon fork magic.

i run pld linux, but no, it fails to reload if using /var/run. the change itself is very old, so i do not have log. but i'm very sure it was during shutdown it was unable to remove pidfile with nagios user

pld-linux/nagios-nrpe@ce845d3
pld-linux/nagios-nrpe@3c30292
pld-linux/nagios-nrpe@6d1e064

ok, maybe it would make sense if the pid_dir is created within the init script? this script at least should run as root, so it will have the privileges to do so?

not knowing your system, but seeing the diff, i would say if you stay with /var/run/nagios dir, it should work fine.

because:

1. /var/run/nagios dir seems to exist, and is likely writable for nagios user
2. recipe creates nrpe.conf which specifies where nrpe daemon writes the pid

also, recipe may add 'directory "/var/run/nagios"' resource (with proper nagios-writable permissions) if the dir does not exist.

looked the PR diff again. you shouldn't modify 'freebsd' case, if you don't know or can test that system!

ahh ok, i'm running this cookbook with the source install method, then the /var/run/nagios does not exist for me. also after any reboot the /var/run directory is cleaned again. so even if chef creates the directory , i must ensure it runs after a reboot, which is not the case for me.

ok with the freebsd a totally agree

and for the fix for #7, are you sure debian initscript has "reload" action? maybe you need :restart instead?

# service nrpe reload
Usage: /etc/rc.d/init.d/nrpe {start|stop|restart|try-restart|force-reload|status}
# echo $?
3

this is from my system and it lacks reload, because nrpe daemon itself does not handle any signals

hmm so this cookbook provides an init script ... and it also supports the reload command https://github.com/schubergphilis/nrpe/blob/master/templates/default/nagios-nrpe-server.erb#L54

i assume you run /var/run as tmpfs and systemd? so you need systemd-tmpfiles configuration so the dir is created at boot.

if initscript comes from recipe, you can then mkdir in initscript in 'start' action. more universal than systemd config, i guess.

install -o nagios -g nagios -d /var/run/nagios

oo, it has signals support indeed. but test whether it actually reloads configuration?

# kill -HUP `cat /var/run/nrpe/nrpe.pid`
Aug  5 18:28:04 wintersunset nrpe[6463]: Caught SIGHUP - restarting...
Aug  5 18:28:04 wintersunset nrpe[6463]: Could not open config directory '/etc/nagios/nrpe.d/CVS' for reading.
Aug  5 18:28:04 wintersunset nrpe[6463]: Continuing with errors...
Aug  5 18:28:04 wintersunset nrpe[6463]: Server listening on 192.168.2.23 port 5666.
Aug  5 18:28:04 wintersunset nrpe[6463]: Listening for connections on port 0
Aug  5 18:28:04 wintersunset nrpe[6463]: Allowing connections from: 127.0.0.1

Hi @5c077yP I am trying to fix the travis issues in this cookbook and going over all open PRs either failed in travis CI or did not even run Travis.

Do you think that this PR is still valid and required?

@snkutlu hmm, this PR is quite some time ago, let me think that through again

If you want to pick this up again, just raise a new PR. 👍

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.