[Xymon] DURATION not working as expected

[J.C. Cleaver]

On Tue, March 22, 2016 7:39 am, Johan Sjöberg wrote:
> Hi,
> We are having problems with the DURATION setting in alerts.cfg, that it
> doesn't work as expected. The problem is that DURATION counts the time
> since the test went from OKCOLOR to ALERTCOLOR, and not the time since the
> alert-triggering color started. This means that if you have the default
> setting, with yellow as an ALERTCOLOR, and configure SMS alerts for red
> color with a certain DURATION, SMS will be sent to escalation teams
> immediately if a test is yellow for more than DURATION, and then goes red.
> If I configure an alert for red color, I would expect it to start counting
> the DURATION from when the red color was triggered.
> Is there some easy hack I can do to change this behavior? The current
> workaround is to have yellow as an OKCOLOR, but that means we cannot set
> up any alerts for yellow, and we cannot ack yellow tests.
>
> Regards,
> Johan

Hi Johan,

I can confirm that this is the case for DURATION. It's a reflection of the
fact that the timing is compared against the alert record as a whole
instead of the per-recipient record (where REPEAT values are stored, for
example).

I *think* I've got a fix for this, but I'm hesitant to put it into 4.3 as
this may be behavior that's being relied on by folks (in the same way that
REPEAT values *are* cleared when an alert escalates from yellow->red).
This should be fixed at the revision release, however.

I'm a bit surprised it hasn't been noticed before.


Regards,
-jc