monit – make broken things work

I set up monit, a utility for monitoring services on a Unix system. My concrete use case is a FreeBSD machine, that runs "N2N Edge", a Peer-to-peer (P2P) virtual private network (VPN) software. monit pretty much automates the "have you tried turning it off and on again?" process. Configurations like the following illustrate this:

check process myproc
        matching "myproc"
        start program = "/etc/init.d/myprocstart"
        stop program = "/usr/bin/killall myproc"
        if cpu usage > 95% for 10 cycles then restart

In order to be monitored, every process needes a file containing the process id, when it runs (a so called pidfile). I used the following script to wrap the start_edge script, which in turn holds the actual command to start the VPN software:

#!/usr/local/bin/bash
case $1 in
    start)
       echo $$ > /var/run/edge.pid;
       exec 2>&1 /root/start_edge 1>/tmp/edge.out
       ;;
     stop)
       killall edge;;
     *)
       echo "usage: edge {start|stop}" ;;
esac
exit 0

After pkg install monit an example config file can be found at /usr/local/etc/monitrc.sample. I removed the .sample part of the file name and appended the following to the very end:

check process edge                           
        pidfile /var/run/edge.pid            
        start program = "/usr/bin/edge start"
        stop program = "/usr/bin/edge stop"  

Make sure to check the config for obvious errors with monit -t, reload the configuration and start the process:

monit reload
monit start all