I have an issue that i try to catch where pods are failing readiness checks (that are just simple http endpoints returning 200 OK).
I expect this to be related to their CPU Load, but when monitoring metrics from cAdvisor (with interval of 15s) i can only see averages, so if a pod is utilized at 100% during 2s of those 15s, the averages are rather low.
Are there any tools that i could use to monitor microbursting? I thought about adding some script/binaries to container images that would report avg/min/max, but i am not sure if that is the correct approach.
Cluster information:
Kubernetes version:
Cloud being used: AWS
Installation method: EKS
Host OS: Ubuntu 22.04