[1.4.2] Items stuck in queue
Hello,
I'm using Ubuntu 8.04 with Zabbix 1.4.2 on MySQL (all installed from repository) to monitor a dozen of machines (all of them on Linux, agents are mostly 1.1.6). Not sure if this has any meaning, but the database was converted from some 1.1 version.
The first problem I noticed was that 5 of zabbix_server processes seem to have memory leakage. From Google I learnt it apparently was fixed in 1.4.4, so as a temporary workaround I added a cron job that restarts zabbix server every night.
There's another problem, though. When I look at the queue, there are about 200 items in "more than 5 minutes old" category. For instance: CPU load 5 from server X from yesterday at 10:51 AM. When I look at latest data or graphs, there's no data for this item from this server since yesterday 10:51. About 80-90% of the 200 items got stuck between 10:51 and 10:53 AM. The time is just an example - the important point is that they all got stuck in a very narrow time period.
Do you have any ideas on what the reason can be? How to solve it? Will moving the DB to PGSQL help? One possibility is to restart the server every hour or so, but for obvious reasons it's not really appealing.
Thanks,
Konrad Garus
I'm using Ubuntu 8.04 with Zabbix 1.4.2 on MySQL (all installed from repository) to monitor a dozen of machines (all of them on Linux, agents are mostly 1.1.6). Not sure if this has any meaning, but the database was converted from some 1.1 version.
The first problem I noticed was that 5 of zabbix_server processes seem to have memory leakage. From Google I learnt it apparently was fixed in 1.4.4, so as a temporary workaround I added a cron job that restarts zabbix server every night.
There's another problem, though. When I look at the queue, there are about 200 items in "more than 5 minutes old" category. For instance: CPU load 5 from server X from yesterday at 10:51 AM. When I look at latest data or graphs, there's no data for this item from this server since yesterday 10:51. About 80-90% of the 200 items got stuck between 10:51 and 10:53 AM. The time is just an example - the important point is that they all got stuck in a very narrow time period.
Do you have any ideas on what the reason can be? How to solve it? Will moving the DB to PGSQL help? One possibility is to restart the server every hour or so, but for obvious reasons it's not really appealing.
Thanks,
Konrad Garus