There can be several possible reasons for a cluster to get block in the "Creating" state:

  • The Pods of the components are unable to become ready, with possible states like ImagePullBackOff,Pending,CrashBackLoopOff etc
  • The metadata of the metadb in GMS is not prepared successfully
  • Unable to retrieve the version from CN
  • ...

The troubleshooting approach mainly involves two steps:

  1. Check the status of the Pods in the cluster to see if there are any Pods in abnormal states
  2. Check the polardbx-operator logs,for any ERROR logs related to the cluster
kubectl get pods -l polardbx/name={name}
Pod Status Possible Causes Troubleshooting Steps
  • READY STATUS
  • 0/ 3 ImagePullBackOff
Image pull failure
  • Incorrectly specified image
  • Insufficient permissions for private registry
Use kubectl describe to further determine the issue
  • If the image is incorrect, update the spec of PolarDBXCluster
  • If using a private registry,[add necessary permissions](https://kubernetes.io/docs/tasks/configure-pod-container/pull-image-private-registry/)
  • READY STATUS
  • 0/ 3 Pending
Insufficient resources Use kubectl describe to further determine the issue
  • Add nodes
  • Reallocate resources
  • READY STATUS
  • 2/ 3 CrashBackLoopOff
  • Repeated container crashes
  • Failure of the cn process
Use kubectl describe to further determine the issue
  • Analyze the specific problem
  • If the describe output does not show the error information, you can bring up the Pod by temporarily [disabling the liveness probe ](../ops/component/cn/2-liveness.md) and check the relevant logs inside the Pod

results matching ""

    No results matching ""