What Can I Do If a Large Number of Pods in a Cluster Are in the UnexpectedAdmissionError State?

Symptom

NAME                              READY      STATUS                      RESTARTS       AGE
aos-apiserver-5f8f5b5585-s9l92     0/1       UnexpectedAdmissionError        0          3d1h
aos-cmdbserver-789bf5b497-6rwrg    0/1       UnexpectedAdmissionError        0          3d1h
aos-controller-545d78bs8d-vm6j9    0/1       UnexpectedAdmissionError        3          3d1h
...

Possible Cause

After a scheduler assigns a pod to a node but the node lacks sufficient resources, such as CPUs, memory, or heterogeneous resources, to meet the pod requirements, kubelet will reject the pod and mark it as failed.

Solution

Pods in the UnexpectedAdmissionError state are not cleared immediately. When the total number of pods in the Completed or UnexpectedAdmissionError state exceeds 1,000, CCE performs a centralized cleanup. At this stage, resources previously occupied by these pods have already been released, and only their status records remain for troubleshooting and issue diagnosis. If these pods are not needed, you can manually delete them.