Agent Checks fail on servers with 100% utilization

[thestraycat]

Has anyone else experienced this? When build servers are running a 100% load the agent times out and generates email alerts creating a load of noise. Any fixes that anyones aware of?

Is it possible to add some funky logic like "If CPU >95% then ignore agent uncontactable check" I don't see a way round it other than guessing how low long my server is likely to run at 100%!

[LenR]

It might be interesting to see if increasing the zabbix agent task priorities would help, either renice all agent processes (temporary) or "Nice=-15" in the zabbix-agent.service file and restart the agent.

[thestraycat]

It;s only when boxes (Windows / Linux) servers are hit hard that they miss there heartbeat... What are my options? Longer timeouts for the agent?? I've set realtime priority via windows... Dosn't seem to have done anything tbh.