BUG: upgrade from k8s v1.26.x to v1.27.y , kubelet restart failed

labring / sealos

Sealos is a production-ready Kubernetes distribution. You can run any Docker image on sealos, start high availability databases like mysql/pgsql/redis/mongo, develop applications using any Programming language.

https://cloud.sealos.io

Apache License 2.0

14.08k stars 2.08k forks source link

BUG: upgrade from k8s v1.26.x to v1.27.y , kubelet restart failed #4848

Closed yangxggo closed 4 months ago

yangxggo commented 4 months ago

Sealos Version

both v4.3.7 and v5.0.0_beta

How to reproduce the bug?

upgrade k8s from v1.26.x to v1.27.y

What is the expected behavior?

all pods restart and work fine

What do you see instead?

some pods (include kube-scheduler) are pending or terminating for a long time

Operating environment

No response

Additional information

No response

yangxggo commented 4 months ago

The reason why the pods do not work properly is that the kubelet fails to restart. The root cause is that since 1.27.0, the kubelet startup parameters do not support --container-runtime and --pod-infra-container-image.

yangxggo commented 4 months ago

/assign