Windows Agent becomes unresponsive (5.0.18)

[jhboricua]

We recently updated our agents (Zabbix Agent 2) from 5.0.7 to 5.0.18. Since then, we have observed that on our Windows fleet, agents will become unresponsive at random times. The service/process is running, but there is no communication flowing. Zabbix reports the agent as unreachable due to this. Doing netstat shows a bunch of connections in the 'closing' state. When we try to restart the Zabbix Agent service, it causes the Agent to finally terminate, then we are able to start the service again and all is good. Until the next Agent crash event. So it appears the Agent is getting hung up and not terminating properly unless we force it to.

I've seen this bug report: https://support.zabbix.com/browse/ZBX-20378

So it appears this hasn't been resolved in the 5.0.19 release either. Anyone else experiencing this on their Windows agents?

[tim.mooney]

You mention the Windows agent, but the bug report you link to is for "Agent 2".

We've only upgraded our Linux clients to the 5.0.18 agent currently, we don't yet have any Windows clients running it yet. And now, seeing your post, I think we'll delay until whatever this issue is is fixed.

[jhboricua]

My mistake, we are running the Agent 2 on Windows and Linux. Been running Agent 2 for so long I keep forgetting there's two agents.

[rthonpm]

I've seen this happen on Windows servers and clients with multiple versions of the Zabbix Agent. It generally seems to happen after a marked increase in resource usage on the systems, generally from some other service on the system. Over the past six years of using Zabbix, I haven't really found a logical reason for the service to stop.

Sent from my BBE100-5 using Tapatalk