I would like to know if is possible to update a RANCHER SERVER - FULL ACTIVE/ACTIVE HA installation, from v1.6.14 to v1.6.26 without causing any downtime to the application served by the cluster.
The cluster serve a real production application, so no downtime can be accepted.
The upgrade guide is documented here: Upgrading Rancher
Anyway I have some questions:
- “On each node in the HA setup, stop the running Rancher server containers.” So if I well understood, during the update no rancher server node will be live but rancher agents should not face any issues for that. Is this right?
- So any rancher server node can be stopped / destroyed and then, one by one, we can start / create a new rancher server node with the new desired version. Is this right?
- Should also rancher agents be updated? If yes, how? Because “The
rancher/agentversion is correlated to the Rancher server version. You will need to check the custom command to get the appropriate tag for the version to use.” FYI Hosts in Rancher
- Also how exactly works “Upgrade” yellow button close to kubernetes infrastructure stack? Should I first update kubernetes stack and only then proceed to rancher server update? (see screenshot below)