Re: Optionally monitoring, depending on local config?

[Martin Pala]

I think it could be possible to use 'manual' monitoring mode and monit 
SYSV init script to start service monitoring depending on installed 
server environment.

For example:

1.) monitrc is common for all servers - services which are not common 
are defined as 'manual' monitoring mode:

--8<--
  check process myproc with pidfile /var/run/myproc.pid
    ... (service check definitions) ...
    mode manual
--8<--


2.) monit SYSV init script is extended by tests which can enable service 
monitoring in the case that it is installed on the server:

--8<--
  if test -x /usr/bin/myproc
  then
    echo "Starting myproc monitoring"
    /usr/bin/monit monitor myproc > /dev/null
  fi
--8<--

Note that monit daemon must be started already - the command 'monit 
monitor myproc' instructs the running monit daemon to enable monitoring 
of myproc service.

In manual monitoring mode will Monit not start monitoring automaticaly 
=> you can use simple test for existence of monitored daemons binaries 
and enable monitoring explicitly in the case that it is found.

This way you can have common monitrc configuration file and common SYSV 
init script for all machines.


Another option could be to use monit service dependencies and existence 
file test. However, existence test has hardcoded default action 
currently => to support this directly, monit configuration file parser 
needs to be extended little bit (the modification is quite simple).


Cheers,
Martin



Bryan K. Wright wrote:
> Hi folks,
>
> address@hidden said:
>
> > Why not simply do something like this at the end of the monit conf
> > file:
>
>
> > include /etc/monitrc.d/* 
>
>
>         That's an option I considered, but I manage around 150 machines
> and I'd like to just push out monit (with a standard config file) onto
> them without having to configure each machine individually.  I do nightly
> automated updates on the machines, so it'd be trivial to turn on monit
> on all of them at once if I could come up with a general-purpose
> configuration file.
>
> 	Another option would be to write a script that creates a 
> tailored monitrc for each node.  That's probably the route I'll go
> if I decide to pursue monit (and if there's no way to write a 
> general-purpose monitrc file).
>
>                                                 Thanks,
>                                                 Bryan