zabbix[queue] exploded since upgrade to zabbix 1.6.5

[rrr]

Since upgrade from 1.6.4 to 1.6.5 the value of zabbix[queue] exploded, while displaying normal (low) queue-values at the queue-page.

The first picture shows the the zabbix[queue]-value, which is 250 in average before upgrade and 1100 in average after the upgrade.

The second picture shows the queue-page AFTER the upgrade, which looks quite normal, but the zabbix[queue] isn't normal...

Which is the right value? The queue-values on the queue-page, or the zabbix[queue]-value? And how do I fix it?

[Andreas Bollhalder]

I'm faced with the same problem. After upgrade from 1.6.4 to 1.6.5 I had a lot of outstanding items in the queue. I realized it, after the last Sunday night. I got a bunch of messages, that the agent ping did not return.

Currently, I have increased the "StartPollers" and "StartTrappers" to 8 and under "Administration" -> "Queue" in the frontend, the values are fine again. But I have to investigate more, because with some higher values (10/10), the ZABBIX server died after around of 24h uptime... MySQL connections aren't eated up and the MySQL server shouldn't be the problem (my CPU usage is <5%, RAM is OK).

I'm using the database cache of ZABBIX ("StartDBSyncers=1" in the server config).

Are there others experience similar problems with 1.6.5 ?

Andreas