Linux – How to get monit to do multiple actions when a monitored process has failed

linuxmonit

I have a custom service that I want to monitor with monit. When the process fails I want to copy the log to a shared file system and restart the service. Something like the following but I am not not sure what. Any hints would be appreciated. 

check process pipeline with pidfile /var/run/pipeline.pid
   start program = "/sbin/start pipeline"
   stop  program = "/sbin/stop pipeline"
   if 10 restarts within 10 cycles then timeout
   # Not sure what to write next
   if <service has failed> 
      restart and 
      exec "/bin/bash -c 'cp /var/log/upstart/pipeline.log /nfs/logs/`hostname`.`date +'%m-%d-%Y_%H.%M.%S'`.log'"

Best Answer

I would write a small bash script containing the desired event actions. Call that script from Monit.

It's cleaner, more modular and will behave more predictably. The same idea applies to cron jobs.

For instance, from the Monit examples page, would you rather have to support this:

check directory httpd_core with path /var/crash/core if changed timestamp then exec "/bin/bash -c 'if [ /bin/cat /tmp/monit_httpd_core.tmp | head -1 != /bin/ls /var/crash/core/core.httpd* | tail -1 ]; then /usr/bin/gdb -x /etc/gdb.batch /usr/sbin/httpd /bin/ls /var/crash/core/core.httpd* | tail -1 | tee /tmp/monit_httpd_core.tmp | mail -s httpd_crash admin@foo.bar webmaster@foo.bar; fi'"

or this:

check directory httpd_core with path /var/crash/core   if changed
timestamp then exec script.sh

Where script.sh contains all of the ugliness.

Related Topic