ES agent doesn't feed "host.cpu.usage" metric when CPU loading 100%

[rickywong]

Hi teams,

Once my VM CPU loading is up to 100%, ES agent can't feed log to ES cloud. So the detection rules can't work as well. How to solve this problem?

Thanks
Ricky

[Christian_Dahlqvist]

Looks like you are overloading the cluster, so either reduce the load or assign more resources.

[rickywong]

The ES cloud cluster was not overloading.

Actually it's my monitoring target VM (installed with ES agent) was overloading, and I found during the timeframe of 100% CPU overload, there is no any metrics were ingested. So no detection can be triggered in this timeframe.

Thanks
Ricky

[rickywong]

Maybe I elaborate not good enough before. When my target monitoring host (with ES agent installed) CPU is up to 100% load, I found the metric "host.cpu.usage" is no longer to be found in the data feeding in. So basically the CPU load detection rule can't be triggered.

I would like to see any experience for you guys to handle such situation? Thanks!

[Agi_K_Thomas]

In situations like that, no-data alerts can generate alerts.

Additional resource for your reference is here

[Agi_K_Thomas]

You may configure index threshold alert if the number of documents in your index is lower than some specific value for a specific time interval.