Re: monit restart bug/race condition (3.1 and current CVS version behavior)

[Jan-Henrik Haukeland]

Martin Pala <address@hidden> writes:

[..]
> You can replicate the problem easily when you let monit run in
> daemon mode and wakeup every 1s.

I think we will need to add a restart action to check_process() which
does the restart in one atomic operation (instead of the current
check_process(stop) -> wait -> check_process(start). That is, call
do_stop -> do_start in sequence in check_process(). And to avoid any
race condition (if the monit daemon thread has a low poll frequency)
use an action-mutex so only one thread can execute an action in
check_process() at the same time.

Something like

check_process(P, restart) {

  LOCK(action_mutex)
      do_stop(P);
      do_start(P);
  END_LOCK;

}

I can look into it, so you can concentrate on the device stuff :-)

-- 
Jan-Henrik Haukeland