Kubernetes upgrade of the local cluster causes Rancher UI downtime

[mrolmedo]

Related Issues

[List any issues or tickets on other platforms that are associated with the request. For example, include a link]to the issue tracking that feature in the Rancher repo, or list the Jira ticket number for the request.

• https://jira.suse.com/browse/SURE-8107

• Upgrading RKE2 with system-upgrade-controller causes downtime in nginx-ingress rke2#2184

• Distinguish wait-shutdown command from standard k8s SIGTERM kubernetes/ingress-nginx#6287 (comment)

Summary

Hi team,

During the kubernetes upgrade of the local cluster, the ingress controllers will be recycled causing the Rancher UI not to be accessible, depending on the cluster, it might be inaccessible for 5-10 minutes. This is pointed it out in the Kubernetes Master Class: A Seamless Approach to Rancher and Kubernetes Upgrades(minute 11-12) and also in https://github.com/mattmattox/Kubernetes-Master-Class-Upgrades.

Effect / Impact during the change window

**Kubernetes upgrade of the local cluster

• The Rancher UI should disconnect and reconnect after a few mins due to the ingress-controllers being restarted.**

It might worth including a note, tip or warning in our Docs: https://ranchermanager.docs.rancher.com/getting-started/installation-and-upgrade/install-upgrade-on-a-kubernetes-cluster/upgrades#3-upgrade-rancher