Upgrade CSE 2.0.0 to CSE 2.5.1

[ChandraRatra]

• Need step by step procedure for upgrade CSE 2.0.0 to CSE 2.5.1
• List of actions which can/can’t be performed on existing cluster after upgrade from CSE 2.0.0 to CSE 2.5.1

[sahithi]

Have you checked this? - https://vmware.github.io/container-service-extension/RELEASE_NOTES.html

[rocknes]

The release notes for CSE 2.5.0 describes the working of cse convert-cluster and why it is necessary. Without running the command you can't perform any operation (from CSE 2.5.0) on the clusters deployed via CSE 2.0.0. If you plan to upgrade to CSE 2.5.1, it's mandatory to run cse convert-cluster on pre-existing clusters.

You can find the step by step upgrade guide at https://vmware.github.io/container-service-extension/CSE_SERVER_MANAGEMENT.html#server_upgrade

[ChandraRatra]

Thanks for your reply.

On 1 cluster, "convert-cluster" command was completed after upgrade. but when tried to add node after convert-cluster. Got below error:- task: b5b99b66-05dc-48f0-9826-615b9eaa5a58, result: error, message: Join cluster script execution failed on worker node ['node-w3lq', 'node-ivaz']:['\t[WARNING IsDockerSystemdCheck]: detected "cgroupfs" as the Docker cgroup driver. The recommended driver is "systemd". Please follow the guide at https://kubernetes.io/docs/setup/cri/\nerror execution phase preflight: unable to fetch the kubeadm-config ConfigMap: this version of kubeadm only supports deploying clusters with the control plane version >= 1.13.0. Current version: v1.12.7\n', '\t[WARNING IsDockerSystemdCheck]: detected "cgroupfs" as the Docker cgroup driver. The recommended driver is "systemd". Please follow the guide at https://kubernetes.io/docs/setup/cri/\nerror execution phase preflight: unable to fetch the kubeadm-config ConfigMap: this version of kubeadm only supports deploying clusters with the control plane version >= 1.13.0. Current version: v1.12.7\n']

Where as on 1 cluster "convert-cluster" command didn't complete it keep on running forever :-(

[rocknes]

cse convert-cluster getting stuck on "Waiting for Guest Customization" is a known issue and a vCD bug. The operation on vCD finishes within a minute or two, but vCD never updates the status on the VM guest customization status page. As a result CSE keeps waiting indefinitely. You can safely terminate the command after few minutes and try interacting with the VM.

The other error you are seeing is most likely being caused by the fact that you are trying to add a node using the new template which has a higher Kubernetes version than the one installed on your cluster. If I am not mistaken your original cluster is a photon based one which is running K8s 1.12 and you are trying to add a node using ubuntu based template with K8s 1.15. Please try to use the 1.12 photon template.