Hi-

I'd like to catch processes that are logging more than an expected
amount. I can write an event handler on the "TICK_60" (or similar)
event, and then scan the log directory to do this. A somewhat more
efficient solution would be to receive an event when the process logs
are rotated (and then take some secondary action to see if the process
is logging too much).

The latter approach would require something like a PROCESS_LOG_ROTATE
event (which doesn't yet exist). I've looked at the supervisrod code and
prototyped this event -- but I'd like to know if this is overkill in
general. Should I just stick with the TICK_60 event and then scan the
logs directory? Would a "logs rotated" event be generally useful?

Thanks,
-- David
_______________________________________________
Supervisor-users mailing list
[email protected]
http://lists.supervisord.org/mailman/listinfo/supervisor-users

Reply via email to