Best practice for dealing with missing pidfile while service is running

[Bill Durant]

Greetings:

Is there a best practice for dealing with a situation when a service's
PID file is deleted by something other than the service itself?

For example, given the following monit rule:

check process ntpd with pidfile /var/run/ntpd.pid
start program "/etc/init.d/ntpd start"
stop program "/etc/init.d/ntpd stop"

If /var/run/ntpd.pid is deleted by the root user from the command line,
then monit will start it again resulting in two instances of ntpd.

A workaround that I discovered is to tell monit to 'restart' ntpd
instead of 'starting' it as follows:

check process ntpd with pidfile /var/run/ntpd.pid
start program "/etc/init.d/ntpd restart"
stop program "/etc/init.d/ntpd stop"

Is this a common practice or is there a better way?

Thanks!

Bill