Spike: Upgrade containerd for an existing Kubernetes cluster

[eak13]

Problem description
As part of ongoing operations, we need the ability to perform various component upgrades & understand the behavior of the clusters when these upgrades occur. When it comes to upgrading containerd for an existing cluster, the plan is to use Cluster API & Metal 3 to deliver a new image with the upgraded containerd version.

Proposed change

• Generate a new image via image-builder with a new version of containerd
• Test the ability to deliver the image to an existing cluster via Cluster API/Metal3.
• Document & provide findings of the upgrade (e.g. What was the process? What issues were encountered, if any?, etc.)
• Are there any other components that need to be upgraded at the same time?

[pallavgupta]

pls assign it to me.

[pallavgupta]

It has been discussed in design call that this issue has below dependencies for actual implementation of the solution:

1. Kube API Server is deployed in HA mode (Kube API Server VIP)
2. Upgrade CAPI to v1alpha4 & CAPM3 to v1alpha5 https://review.opendev.org/c/airship/airshipctl/+/804834
3. Ironic deployed in HA mode (Ironic Provision IP VIP)
   Rolling upgrade of control plan node successfully tested and above dependencies are for upgrading 1 Target node. Pls find details in https://hackmd.io/@Pallav/BkU2FuWZY

[lb4368]

Closing this issue as rolling upgrade process is addressed in #621.