Re: [monit] How to Reset Monit's Cycle Counters?

[Art Age Software]

Thank you very much - this is great! Can this change be patched into
4.10? Or will I need to update to the 5.0 beta?

Sam

On Thu, Jul 3, 2008 at 1:43 PM, Martin Pala <address@hidden> wrote:
> Hi,
>
> i have modified the stop and unmonitor actions to cleanup the event state.
> You can get the source code from CVS:
> http://savannah.nongnu.org/cvs/?group=monit
>
> Thanks,
>
> Cheers,
> Martin
>
>
> On Jul 2, 2008, at 3:32 AM, Art Age Software wrote:
>
>> I have a service that looks something like this:
>>
>> -------------
>> check file some-file with path /var/run/check.tmp
>>  start program = "/bin/touch /var/run/check.tmp"
>>  mode manual
>>  every 2 cycles
>>  if changed timestamp 5 times within 10 cycles then exec "/some/script"
>>  if 10 restarts within 20 cycles then timeout
>> -------------
>>
>> I would like to find some way of resetting monit's check on the
>> timestamp after the exec "/some/script" occurs. Currently, exec
>> "/some/script" (indirectly) invokes "monit unmonitor some-file"
>> (through heartbeat). However, when monitoring is re-enabled on
>> some-file in the future, monit will treat an immediate  occurrence of
>> a changed timestamp as being within the window of "5 times within 10
>> cycles" which causes it to immediately invoke "/some/script" again. In
>> my case, "/some/script" is a heartbeat failover command, so the
>> unfortunate effect is that the two nodes just ping-pong the services
>> back and forth between them every minute.
>>
>> I have tried restarting the some-file service (as opposed to
>> unmonitor/monitor), which also does not work.
>>
>> So far, the only thing that seems to work is to force a "monit reload".
>>
>> Is there any other way?
>>
>>
>> --
>> To unsubscribe:
>> http://lists.nongnu.org/mailman/listinfo/monit-general
>
>
>
> --
> To unsubscribe:
> http://lists.nongnu.org/mailman/listinfo/monit-general
>