What Can I Do If proxy-agent Fails to Be Deployed?

Symptom

The deployed proxy-agent is not in the Running state.

Procedure

1. Log in to the master node of the cluster.
2. Check the running status of proxy-agent.

   kubectl -n kube-system get pod | grep proxy-agent

   The following command output shows the pods are in the ImagePullBackOff and Pending states:

   proxy-agent-59ddf7597b-rq4j6            0/1     ImagePullBackOff   0          2d16h
   proxy-agent-59ddf7597b-sjf55            0/1     Pending            0          2d16h

3. Query the details about the pods:

   kubectl describe pod proxy-agent-*** -nkube-system

   The following errors may occur:

   • The Kubernetes event indicates that the cluster cannot pull the proxy-agent image. In this case, check whether the cluster can access the public network to pull the SWR image.

     

   • The Kubernetes event indicates that the CPU or memory resources of the node are insufficient. In this case, scale up the node.

     

   • The Kubernetes event shows that the scheduling failed. To achieve high availability, proxy-agent is deployed with two pods and they are scheduled to different nodes by default. Ensure that your cluster has at least two nodes with sufficient resources.

     

   • If gatekeeper is displayed in the Kubernetes event, the created policy may have performed interception. To solve this problem, run the following command in the cluster to delete the corresponding policy:

     kubectl delete constraint --all

4. After the preceding problems are resolved, check the running status of proxy-agent again. All pods should now be in the Running state.