Component pods Are Not in the Running State

Symptom

After cluster scheduling components are deployed, the kubectl get pods --all-namespaces -o wide command is executed to check the status of each component. However, the pod status of each component is not in the Running state.

Cause Analysis

• Cause 1: If the pod status is ImagePullBackOff ErrImagePull, the image version may be incorrect.
• Cause 2: If the pod status is CrashLoopBackOff Error, the image architecture may be incorrect or you do not have the permission to access the log path.
• Cause 3: If the pod status is Pending, the selector may not be configured on the management node.

Solution

For cause 1, check whether the image version of the corresponding component is correct. If not, prepare a new image by referring to Preparing an Image.

For cause 2, run the kubectl logs -f command to view logs. If the image architecture is incorrect, prepare a new image by referring to Preparing an Image. If the log path is not granted permission, set the permission and owner by referring to Creating a Log Directory.

For cause 3, configure the selector for the corresponding management node.

For more information about the pod status causes and solutions, see other sections in this document. If no solution is found, contact Huawei technical support.