Deployed as a K8s helm chart, pod is being throttled with no apparent reason

[luciano-buono]

Hi, I have deployed the app using the Helm chart: https://github.com/promhippie/charts/tree/master/stable/github-exporter and set the following resources block:

      resources:
        limits:
          cpu: 300m
          memory: 250Mi
        requests:
          cpu: 50m
          memory: 50Mi

Checking the CPU usage with ice:

➜ k ice cpu
PODNAME                                                 CONTAINER                           USED  REQUEST  LIMIT  %REQ  %LIMIT
prometheus-promhippie-github-exporter-7fdd9fbd9c-2sqkb  github-exporter                     9m    50m      300m   17.03 2.84

I see that the reported usage is very low, however I'm facing throttling issues

I suspect this may be related to the container trying to burst for very small time, not enough for the metrics to catch it

[tboerger]

CPU throttling is always a difficult part. I'm open to improvements, I have not set CPU limits where I run into this issue.