search

gocrane / crane-scheduler

Crane scheduler is a Kubernetes scheduler which can schedule pod based on actual node load.
Apache License 2.0
225 stars 64 forks source link

crane-scheduler-controller 健康检查失败，readness和liveness都失败 #41

Closed chinaboy007 closed 1 year ago

chinaboy007 commented 1 year ago

Events: Type Reason Age From Message

Normal Scheduled 5m27s default-scheduler Successfully assigned crane-system/crane-scheduler-controller-5c85f47c45-trmzp to 192.168.227.164 Normal Pulled 5m37s kubelet Container image "docker.io/gocrane/crane-scheduler-controller:0.0.23" already present on machine Normal Created 5m37s kubelet Created container crane-scheduler-controller Normal Started 5m36s kubelet Started container crane-scheduler-controller Warning Unhealthy 32s (x31 over 5m32s) kubelet Readiness probe failed: Get "http://10.244.27.203:8090/healthz": dial tcp 10.244.27.203:8090: connect: connection refused

readness和liveness都失败，后面直接注释掉才正常启动。请修复此问题

xieydd commented 1 year ago

maybe initialDelaySeconds is too small , we will check this and fix the issue.

chinaboy007 commented 1 year ago

maybe initialDelaySeconds is too small , we will check this and fix the issue.

I have change initialDelaySeconds is 120s or 180s,but Readiness probe failed

xieydd commented 1 year ago

@chinaboy007 We found the problem, because this commit https://github.com/gocrane/crane-scheduler/commit/ccd57b806e1c7109b440e8873ca75db91337b461 was merged in after image version 0.0.23 https://github.com/gocrane/helm-charts/blob/main/charts/scheduler/values.yaml#L18 , so the healthz healthz check will failure, you can update it to 0.0.24 image version. Thank you report the issue.

You can update your crane-scheduler via:

helm repo update
helm upgrade scheduler -n crane-system --create-namespace crane/scheduler
chinaboy007 commented 1 year ago

经验证，已解决，谢谢！