Power-cycling check on OpenShift and Cloud Pak for Integration

If you cycle the power on an API Connect cluster that is deployed on OpenShift or Cloud Pak for Integration, you might need to reset the IP address for the kubernetes.io/hostname.

About this task

You will know there is a problem if you cycle the power for the cluster and receive an error, or find that the deployment remains in the pending state indefinitely because some of the pods failed to start. Complete the following steps to resolve the problem:


  1. Run the following command to check the label on the worker nodes:
    oc get nodes -l kubernetes.io/hostname=localhost

    If the response shows the following message, then the power cycle did not cause a problem and you can skip the rest of this task:

    No resources found

    If the response looks like the following example (the IP address will be different), proceed to the next step:

    NAME                                         STATUS   ROLES    AGE   VERSION
    ip-    Ready    worker   27h   v1.19.0+8d12420
  2. Run the following command to correct the problem:
    for i in $(oc get nodes -l kubernetes.io/hostname=localhost|grep worker|awk '{print $1}'); do echo $i;  oc label nodes $i kubernetes.io/hostname=`echo $i|awk -F'.' '{print $1}'` --overwrite=true; done