[Xymon] alerting issue?

[Gavin Stone-Tolcher]

Thanks for that information Henrik. The criteria for the bug was triggered by alerting rules on our setup overnight and I am glad to report that the behaviour is now  correct. i.e. alerts were generated for second recipient at the start of their TIME window.

BTW, I am using RHEL 6 systems and am not seeing core dumps.

Many thanks for the rapid response to this issue.

Cheers,
Gavin....

-----Original Message-----
From: henrik at hswn.dk [mailto:henrik at hswn.dk] 
Sent: Wednesday, 22 January 2014 6:52 PM
To: Gavin Stone-Tolcher
Cc: xymon at xymon.com
Subject: RE: [Xymon] alerting issue?

Den 22.01.2014 00:55, Gavin Stone-Tolcher skrev:
> Thanks, I will patch our systems with this today.
>
> Is there a reliable way of reproducing the issue? My test case 
> appeared to work OK unlike the documented failure case.

I setup a test installation with the settings from your original report, just modified so the TIME restriction on the second alert was 15 minutes after I started Xymon. So the first alert fired while the TIME restriction was active on the second alert. The second alert then failed to trigger.

The heart of the problem is that when the first alert triggers, Xymon sets a timestamp for the next time it should send an alert - and since the alert that triggered has a "REPEAT=1w" setting, then xymond_alert ignores the event for a week. It simply fails to take into account that there is another recipient defined which triggers before that time. So the circumstances required to hit this bug are

1) you need two or more recipients for an alert
2) you must have a REPEAT setting of more than 24 hours
3) one of the recipients must have a TIME restriction which disables alerts when the event first triggers.


Regards,
Henrik