Multiple pods in Terminating state after node state changes to Not Ready

Last Updated : Oct 25, 2021 |
Prolog information
Metadata
Keywords:

Condition

Multiple pods in a node display in Terminating or Crashed state after the node state changes to Not Ready.

Cause

This is a known issue. When Kubelet loses contact with the API server, the state of the node changes to Not Ready. When this happens, multiple pods go into Terminating or Crashed state.

Solution

  1. Log in to the node that is in Not Ready state using the cluster_ssh console.
  2. Type sudo su at the prompt and press Enter.
  3. Enter the following command to restart the Kubelet service and restore all the pods:
    ADDITIONAL INFORMATION:
    monit restart kubelet
    STEP RESULT:
    The terminated or crashed pods take 20 to 30 minutes to be restored to the normal state.
  4. Enter the following command to verify whether all the components are working:
    ADDITIONAL INFORMATION:
    ccm smoke-test