Escalator crashing

[Brismtedt]

Hello!

Yesteday I activated escalation, and this morning i noticed zabbix server was no longer running and is unable to restart.

By enabling debugging output, ive come to the conclusion that the escalator thread crashes (it is started, server notice one child crashing, all other threads than escalators exits..)

I've tried to remove the escalator action in the web frontend, but the escalator thread still crashes...

Some extract from my log:

 
 12338:20090726:231510 server #15 started [Escalator]
 12338:20090726:231510 In main_escalator_loop()
 ...
 12338:20090726:231510 Query [select i.itemid,i.value_type,i.valuemapid,i.units,i.lastvalue from items i,functions f where i.itemid=f.itemid and f.functionid=12342]

+--------+------------+------------+-------+-----------+
| itemid | value_type | valuemapid | units | lastvalue |
+--------+------------+------------+-------+-----------+
|  22551 |          3 |          2 |       | NULL      |
+--------+------------+------------+-------+-----------+

The query shown is the last thing done by the thread before the shutdown.
The query result is what this query gives me if i run it manually in the db.

I cannot tell if this result is ok (for example: is NULL a valid possibility here, or could it be related to the crash?)


First of all im of course eager to clean my db so i can get the server to run again, but Im also willing to do what I can to help you find the bug.


any one got a clue what to do ?

cheers
/linus

[Brismtedt]

Can i for example just try and empty the escalations table? Or is there relations I need to clean up first/too ?

Or will this have unwanted sideeffects?

/L

[Brismtedt]

Updateing the escalation status (one row had 0, rest 2) to 2 made the server run again... for now at least :-)

Update: When new status = 0 rows appears, it crashes again.
/L

[Brismtedt]

Hi!

Noone has any ideas? Maybe at least what else I can clean up in db so that new scallation rows are not generated?

Problem is, that now everything works fine as long as there are no high level problems.. When a high level problem appear, escalation row is generated and server crashes :-)

/L

[richlv]

did i miss it or didn't you mention which zabbix version is that ?
if it's latest stable, you could try compiling zabbix server with debugging information, obtaining a full backtrace and submitting a bugreport

[Brismtedt]

It seems like this was solved with the 1.6.6 version (i experienced it in 1.6.5)

br

/B