What is the relation between CpuThreshold option and request/limit cpu resource k8s deployment config?

[SgtDaJim]

I am using go-zero in a service deployed on k8s, and I set request / limit cpu 4 cores (4000m). I found that the range of CpuThreshold is [0:1000), so I set 950. Under this config, I found my service cannot occupy all the cpu I request, and the rpc service will response service overloaded when cpu usage is near 1000m。 What can I do to make my service occupy all the resource I request？

[kevwan]

950 means 95% of your cpu limit. And 95% is a moving average value, you can consider it as the average value of last 5 seconds.

[SgtDaJim]

950 means 95% of your cpu limit. And 95% is a moving average value, you can consider it as the average value of last 5 seconds.

that is strange. I review my config, I set resource like this:

then i got an error before (time is UTC):

but at that time, cpu monitoring looked like this:

it seems not touching the threshold.

[SgtDaJim]

Hi, any update?

[github-actions[bot]]

This issue is stale because it has been open for 30 days with no activity.

[github-actions[bot]]

This issue was closed because it has been inactive for 14 days since being marked as stale.