monit-general
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Delaying tests for a number of cycles


From: Martin Pala
Subject: Re: Delaying tests for a number of cycles
Date: Tue, 9 Oct 2007 22:30:45 +0200

As stated - *hard* dependency can solve this problem, since monit will wait for the parent to start. Currently just soft dependency is supported, so this just provides correct sequence, without waiting for parent.

Martin

On Oct 9, 2007, at 10:04 PM, Claus Klein wrote:

Martin Pala schrieb:
Monit supports the dependency for such situations (the 'depends on' statement).
I can't understand how the dependency should help in such a case?

If I know right, monit will not wait for dependent files.
We had a discussion about this 2 years ago, see http:// lists.gnu.org/archive/html/monit-general/2005-11/msg00049.html.

Claus

Currently there is just soft dependency, hard dependency option is planned.

Martin

MosheC wrote:
Thanks.
The workaround is not enough in general. It is not always a case of starting a service. For example, I check files and directories that are created and initialized by processes yet to be invoked and they might depend on some other factors. I would like the checks to start only in , let's say, 10
cycles.

Moshe


Martin Pala wrote:
If monit is starting the process, then it expects that the next monitoring cycle will be the service running otherwise it reports error.

This behavior cannot be changed currently, but i added the todo item for
it:
http://www.tildeslash.com/monit/doc/next.php#38

As a workaround you can set your monitoring cycle so, that the service will have time enough to start.


Martin


MosheC wrote:
With "every N cycles" we can lower the test frequency.

Is it possible to only delay the first test for a number of cycles and
then
continue normally or possibly delay only the alerts.

I have a lot of alerts when I start monit that are a result of the new processes being invoked by monit. It is just noise. I would like to get alerts only after I have reached "steady state" and then the alerts will
really mean something "bad" happened.

Moshe


--
To unsubscribe:
http://lists.nongnu.org/mailman/listinfo/monit-general





--
To unsubscribe:
http://lists.nongnu.org/mailman/listinfo/monit-general



--
To unsubscribe:
http://lists.nongnu.org/mailman/listinfo/monit-general





reply via email to

[Prev in Thread] Current Thread [Next in Thread]