We use the following trigger to make sure our agents are up: {Template App Zabbix Agent:agent.ping.nodata(5m)}=1. This works fine if an agent process is stopped. However, I recently noticed it does not work if the server does not have permission to the agent. We had an issue where the "Server" setting was mistakenly changed on a server, so our Zabbix server no longer had access to it. We never received any alerts for this, and the Zabbix server UI Hosts window still showed the agent availability as green. Even if the Zabbix server isn't allowed to the agent because of the Server setting, why wouldn't this trigger the agent.ping.nodata trigger? Is there some other trigger we should use in case this ever happens again?
-
-
Update
Apparently this only happens on Windows, not Linux. I made the same change on a Linux agent, modifying the Server setting so it wouldn't allow incoming connections from our Zabbix server. We immediately received an alert from the agent.ping trigger. Tried this on a Windows host and no alert. In the Zabbix UI->Hosts, the availability for the Windows host shows as green. The Items for the host show as "Unsupported - Get value from agent failed: ZBX_TCP_READ() [104]". The Triggers how as "Unknown" with the same error as Items.
So I guess my real question is, if the Server parameter is changed on a client so the Zabbix server is no longer allowed to connect, why does Zabbix alert for a Linux agent but not for a Windows agent? We were running Windows agent 2.2.1. I just upgraded to latest, 2.4.4, and it makes no difference.
Comment