Since this hasn't appeared in the MON releases yet, I'd like to note that
the patch included below solved some serious scheduler failures. I just
brought mon up in a badly designed network, and the scheduler was dying
every 3-4 hours. Applying the patch included here solved the problem, and
the scheduler has been stable for nearly 2 weeks without a single failure.
Thanks to Theo, and Jim could you please make sure this or something
similar gets into mon 1.0 ?
From: Theo Van Dinter
Subject: Re: Occasional scheduler death
Date: Thu, 30 Aug 2001 12:11:30 -0700
On Thu, Aug 30, 2001 at 08:11:25AM -0400, Bates, C Thomas wrote:
> I think I've finally found what has been causing the MON daemon to die. It
> finally died with debug running, with this in stderr:
> /^**** Time Out/: nested *?+ in regexp at /usr/local/mon/prod/mon line 587.
Line 593 has another bug (for which I submitted a patch to Jim) where alerts
will be skipped if a monitor fails with summary "alpha bravo", and then
fails again with a summary "alpha". The summarys are different, so there
should be an alert but mon won't alert until the "alertevery" period has
passed.
I've attached my patch which I believe will fix both your problem and
the one in the above paragraph. :)
--
Randomly Generated Tagline:
"You must lash out with every limb, like the octopus who plays drums."
- The Sphinx in Mystery Men
586a587
> my($prevsumm) = split("\n", $sref->{"_failure_output"});
593c594
< ($sref->{"_failure_output"} =~ /^$summary/)
---
> (!$pref->{"_observe_detail"} && $prevsumm eq
$summary)
--
Joe Rhett Chief Geek
[EMAIL PROTECTED] ISite Services, Inc.
