Server stuck not reported as unreachable

[gnollet]

Hi,
I'm implementing zabbix since few days on our network.
I setup agent on Linux server and this server is now "stuck". The server only answer to ping and doesn't answer to ssh, http or other open ports. But the server didn't reject connection.
The problem is zabbix is still trying to get informations from agent and doens't declare this server as unreachable. At this time, the server is in this state since yesterday evening and zabbix server is still trying to get information from yesterday evening, the queries are still on the queue and then impact other checks, other checks are deleyad for some minutes.

My question is to know if if there any timeout on the checks to declare it as unreachable and/or put some warning.

Thanks a lot

[James Wells]

Greetings,

What I do to get around this problem is use the version[agent] and then check to see if I get an answer. If I fail to get an answer after one test cycle, I trigger an alert that the system is hung.

EDIT: Hit return before I intended to.... Anyway, I depend this check on simple icmpping and and agent ping.

[gnollet]

Thanks for reply !
But I'm already usingthe agent version.
This morning I found 30 checks in pending for the stuck host on the queue since yesterday evening. There was no supervision during the night.
This morning we reboot the server and then the Zabbix server report an alarms about the server is unreachable and few minutes later, about server was just restarted.
After some investigation on the server side, it appears all Memory and swap were used and probably explain why the server was stuck, we fixed this by increasing the swap. I see this with the last values on zabbix.

Do you think if it's possible to setup a timeout on the zabbix server side ?
I know there is timeout paramter on agent on each hosts but I didn't see this on server parameter.