We currently have an issue currently where if we enable more than 22 to 23 items in a hist across all templates a host has the items start to queue and we then see Zabbix agent unreachable.
The checks are a mixture of bash scripts, java files connecting to Oracle and Zabbix builtin commands to check processess, cpu and memory information as well as checking log files for certain phrases.
The items are all Active checks and it seem to be random on what item triggers the queuing. To get around the issue I have to disable all items and only enable the 22 that we deem the most important of all the checks we have, ideally we would like all check enabled without causing them to appear on the queue.
On one attempt we could find that its a check designed to check the log that causes the issue, but another time it could a database check and another time it might be a check to get the cpu details.
We are using Zabbix server V3.2 and our clients use a mixture of v2.2 and v3. We have no agents installed locally and they are on customer sites and connect over the internet to our locally hosted server.
When we have seen the items queuing we have gone onto the affected server and we run the same check with the followoing comand which always returns a value to be sent to the server
/usr/sbin/zabbix_agentd -c /etc/zabbix/zabbix_agentd.conf -t <check>
I have also at the same time use telnet to check connectivity from the agent to the server and each time it is always open
Any help is appreciated on what to check or what to enable to allow us to have the checks we would like to have.
The checks are a mixture of bash scripts, java files connecting to Oracle and Zabbix builtin commands to check processess, cpu and memory information as well as checking log files for certain phrases.
The items are all Active checks and it seem to be random on what item triggers the queuing. To get around the issue I have to disable all items and only enable the 22 that we deem the most important of all the checks we have, ideally we would like all check enabled without causing them to appear on the queue.
On one attempt we could find that its a check designed to check the log that causes the issue, but another time it could a database check and another time it might be a check to get the cpu details.
We are using Zabbix server V3.2 and our clients use a mixture of v2.2 and v3. We have no agents installed locally and they are on customer sites and connect over the internet to our locally hosted server.
When we have seen the items queuing we have gone onto the affected server and we run the same check with the followoing comand which always returns a value to be sent to the server
/usr/sbin/zabbix_agentd -c /etc/zabbix/zabbix_agentd.conf -t <check>
I have also at the same time use telnet to check connectivity from the agent to the server and each time it is always open
Any help is appreciated on what to check or what to enable to allow us to have the checks we would like to have.