Endless notification after edit of trigger in active escalation

[Slash]

Hello everyone,

I have a very annoying issue with Zabbix... I first configured several triggers to warn me when some value are higher than a specific threshold.

I then noticed that my threshold was too low because I had alert all the time, so I changed the threshold when the trigger was in the "PROBLEM" state.

The alert immediately disappeared from the dashboard, as expected, but the notification is still being sent, and since I have an action configured with escalation to send emails every 12 hours when an issue is still in the problem state, my team receive every 12 hours an e-mail like this :

IO-drive average temperature last 15m > 60° C on srv-157: PROBLEM
ioDrive internal temperature: 50.2 °C

Additional information:
SYSADMIN ALERT
Host: srv-157
DateTimeNOW: 2012.12.26-01:57:40
EventDateTime: 2012.10.15-14:32:14
EventAge: 71d 12h 25m
ItemName: ioDrive internal temperature
ItemLastvalue: 50.2 °C
Severity: Average

Escalation history:
Problem started: 2012.10.15 14:32:14 Age: 71d 12h 25m

I truncated the end, it has the list of all the e-mails previously sent.

About the item monitored, its maximum value ever is 56.1 °C, so it is impossible that the threshold is exceeded. Data is collected normally (item is named "io.internal_temperature", collected via a UserParameter).

Also the trigger expression is : {srv-157:io.internal_temperature.avg(15m)}>60.

I talked about it on irc and it seems like "apparently it is not a good idea to edit expressions for triggers involved in active escalations". But here I had no choice...

Is there any way that I can solve this issue ? I have 3 notifications stuck like that. I looked in the database but I can't find what to update/change/delete to kill the notification.

EDIT : forgot to tell the Zabbix version : 2.0.4, but this issue is here since 2.0.2, I upgraded 2 times zabbix from source since it began.

[Slash]

Just tried this, hopefully it will solve the issue :

zabbix=> select * from escalations;
 escalationid | actionid | triggerid | eventid | r_eventid | nextcheck  | esc_step | status 
--------------+----------+-----------+---------+-----------+------------+----------+--------
          280 |        8 |     13731 |   20052 |           | 1350036302 |        1 |      2
          283 |        8 |     13730 |   20053 |           | 1350036326 |        1 |      2
          288 |        5 |     13732 |   20054 |           | 1356556250 |      151 |      0
          287 |        7 |     13732 |   20054 |           | 1356556254 |      151 |      0
          309 |        5 |     13723 |   34081 |           | 1356569864 |      145 |      0
          308 |        7 |     13723 |   34081 |           | 1356569868 |      145 |      0
(6 rows)

zabbix=> delete from escalations where esc_step = '151';
DELETE 2
zabbix=> delete from escalations where esc_step = '145';
DELETE 2
zabbix=> select * from escalations;
 escalationid | actionid | triggerid | eventid | r_eventid | nextcheck  | esc_step | status 
--------------+----------+-----------+---------+-----------+------------+----------+--------
          280 |        8 |     13731 |   20052 |           | 1350036302 |        1 |      2
          283 |        8 |     13730 |   20053 |           | 1350036326 |        1 |      2
(2 rows)

[Rawlings]

What's the update frequency on your item ?

Use average function with caution

[Slash]

The update frequency is 60 secs on the items in question.

[Slash]

After 2 days without alert, I can confirm that removing the rows from the "escalations" table solved the issue.