errors with trigger stuck in loop?

[zeki893]

Using 1.6.2, I have a host with icmpping check on with trigger {test:icmpping. last( 0 ) }#1
It seemed to be working fine until I rebooted the server that was being checked and set off the trigger. It behaved as its supposed to sending me an alert that it failed. Then when it came back online it said OK, BUT it keeps sending me alerts with problem, then ok, problem then ok. I must disable the item then few minutes later enable it.

I can reproduce this problem as it happens whenever an alert goes off, it doesn't stop sending me alerts.

This is the notification I receive when an alert is triggerred.
NOTICE THAT LAST VALUE IS 1. The trigger should only go off when LAST VALUE is 0

NOTE: Escalation cancelled: Item 'test' disabled.
test: PROBLEM
Severity: Disaster
Date: 2009.01.27 - 14:10:34
Event - Date - Time: 2009.01.27 - 12:10:26
Event Age : 2h 0m
Item Name: test
Item Last Value: 1
Trigger Comment:
Trigger Status: PROBLEM
Hostname: test

30 seconds later after I receive this alert that status is OK

test: OK
Severity: Disaster
Date: 2009.01.27 - 13:47:03
Event - Date - Time: 2009.01.27 - 12:11:55
Event Age : 1h 35m
Item Name: test
Item Last Value: 1
Trigger Comment:
Trigger Status: OK
Hostname: test

anybody know whats wrong?

[pace]

This is happening to me as well, but only with web tests.


pace

[zeki893]

I'm glad I'm not the only one. I'm going crazy. I'm going to try recompile on a new system. BTW what distro are you using?

[pace]

CentOS 5.2 32 bit

The thing sent out 200 alerts to one of my customers last night at about midnight. They're really happy with me right now, but I guess they finally know a little bit about what happens to me all the time. heh


pace

[Neurotox]

I got probably the same issue. If I have escalation ON and a trigger goes ON, I will start receiving email in loop until I turn it off (both OK and PROBLEM notification).

http://www.zabbix.com/forum/showthread.php?t=11623

My setup:

Zabbix 1.6.2 compile from source
Debian Etch stable (x86)
MySQL 5.0.32
Postfix 2.3.8

Any one have an Idea where to look? Apparently a few people got it working correctly but not us

[Neurotox]

Any one have an update on this problem ? :\

[pace]

I've got an update. This just happened to me again. I paid more attention this time instead of being in a panic to stop it. I'm getting recovery messages after a web scenario fails. The problem is that I don't even have recovery messages enabled. So I get hundreds of recovery messages, but the dashboard never shows the scenario has having recovered (even tho it did recover).

I'm running 1.6.2 with the memory leak patch on Centos 5.2 (32 bit) with all the latest patches and with 1.6.2 compiled the day that the memory patch came out.


pace

[pace]

I opened a bug report on this yesterday. It happened again, twice, in the middle of last night. The system was sending recovery messages out like crazy when my web scenario failed and it needed to have me disable/re-enable the web scenario to convince it to stop sending messages. I had to restart zabbix_server to get it to recognize that the web scenario was working.

Since I did it twice in relatively close succession last night the routine was: disable web scenario that failed, wait a few minutes (I'm not sure this is necessary), re-enable, restart zabbix_server.


pace