Hi everyone,
I package a bunch of applications as RPM file. Part of the post-install process
is to issue "monit reload" followed by a "monit monitor all". I do this to make
monit read the (possibly updated) config files under /etc/monit.d. However I
noticed, that this kicks of a whole new cycle in monit (even though the
previous cycle is just over).
My rpm post-script reads like
------------------
%post
#! /bin/bash
if [ -x /sbin/service ]; then
/sbin/service "%{name}" start 2>&1 | logger -t rpm
fi
/usr/bin/monit reload
sleep 2
/usr/bin/monit -g "%{name}" monitor all >/dev/null 2>&1
exit 0
------------------
My monit file for the services read like
------------------
check process @project.name@[email protected]@.process with pidfile
/var/run/@project.name@[email protected]@.pid
start program = "/etc/init.d/@project.name@[email protected]@ start"
stop program = "/etc/init.d/@project.name@[email protected]@ stop"
if not exist for 5 cycles then alert
if cpu > 75% for 5 cycles then alert
if failed host 127.0.0.1 port @service.port@ for 5 cycles then alert
if not exist for 2 cycles then restart
------------------
The main config file has
------------------
...
set daemon 60
....
------------------
The intention was, to get the service restarted after 2minutes and get a alert
after 5minutes. However when I install a couple of those RPMs, where each comes
with its own "monit reload", I now have 5cycles in 15 seconds. This leads to
false alarms, that I like to prevent.
I there a best practice in the monit world I could follow? Do you guys have
recommendations how to solve this?
Thanks a lot in advance,
Jens
--
To unsubscribe:
https://lists.nongnu.org/mailman/listinfo/monit-general
