Resolving Cluster Import Failure for Multicloud Manager v3.2.0

Introduction

I will show you how to resolve a failure while trying to import a cluster on Multicloud Manager version 3.2.0. This procedure only works for Multicloud Manager v3.2.0.

Problem

The problem may occur while you are trying to import a cluster. The import times out and generates the following error message:

# cloudctl mc cluster import -f mycluster -K kubeconfig | tee cluster-import.log
secret/klusterlet-bootstrap created
serviceaccount/klusterlet-sa created
clusterrolebinding.rbac.authorization.k8s.io/klusterlet-crb created
secret/klusterlet-config created
job.batch/install-klusterlet created
failed to start bootstrap job: rpc error: code = Unknown desc = Error response from daemon: Get https://registry-1.docker.io/v2/: net/http: request canceled while waiting for connection (Client.Timeout exceeded while awaiting headers)
...

Solution

First, you may need to clean-up the failed cluster. While logged into the hub cluster, run the following and note if the cluster exists in the list:

kubectl get clusters --all-namespaces

If the failed managed cluster is in the list, remove it while logged into the hub cluster. If not, then skip this step.

In this example, I am deleting a cluster called mycluster:

kubectl delete cluster mycluster -n mycluster

Next, while logged into the hub cluster, regenerate your cluster-config.yaml:

cloudctl mc cluster template mycluster -n mycluster > cluster-import.yaml

Edit the cluster-import.yaml and change the inception image value from latest:

inception_image: ibmcom/icp-inception:latest

…to 3.2.0:

inception_image: ibmcom/icp-inception:3.2.0

Save the file and retry the import. The cluster import will now complete without timing out.