Xymon Mailing List Archive search

Unwanted delay in hobbit alerts

2 messages in this thread

list Brian Bernstein · Wed, 12 Jul 2006 16:26:50 -0400 ·
It seems that hobbitd_alert for whatever reason, is sending alerts
only once every half hour.

In other words, if I was to test an alert as such:
bbcmd hobbitd_alert --test [hostname] [test]

It will immediately output that it found a matching rule, and states
the action being taken
(mail -s etc. . .), however, the notification will not come up in the
notifications.log or in the mail server logs until either :06 or :36
past the hour.

Also, the output of the resulting messages in the alerts are time
stamped just before the message is sent/received as opposed to when
the command was first called to run the test.

I have not tested to see if this happens when a machine actually goes
down (the test in question for this particular alert), just with the
test option in hobbitd_alert (actually, the host I've chosen for the
test was already down).


The alert rule in question is rather simple:
HOST=*  SERVICE=conn
MAIL [user]@[server].[tld] RECOVERED

Any help in this would be greatly appreciated.

-Brian
list Henrik Størner · Wed, 12 Jul 2006 22:59:19 +0200 ·
quoted from Brian Bernstein
On Wed, Jul 12, 2006 at 04:26:50PM -0400, Brian Bernstein wrote:
It seems that hobbitd_alert for whatever reason, is sending alerts
only once every half hour.

In other words, if I was to test an alert as such:
bbcmd hobbitd_alert --test [hostname] [test]

It will immediately output that it found a matching rule, and states
the action being taken
(mail -s etc. . .), however, the notification will not come up in the
notifications.log or in the mail server logs until either :06 or :36
past the hour.
I don't understand. The "--test" option is a simulation; it doesn't
generate any alert messages. 

I think what you're seeing is a *real* alert being sent out because
of a system that is down. The 30 minute interval between alerts
would match the default REPEAT setting.
quoted from Brian Bernstein
I have not tested to see if this happens when a machine actually goes
down (the test in question for this particular alert), just with the
test option in hobbitd_alert (actually, the host I've chosen for the
test was already down).
Precisely. So you have a host that is down, and alert rules configured.
Then you get the alerts - not from the testing you're doing, but from 
the Hobbit system.


Regards,
Henrik