I am always having issues upgrading nodes, and in some mission critical applications I am starting to think that I am better off being standalone.
So the question what happens to running containers if I do a incus cluster remove. Can I do it while the containers are running in node I want to remove? On the converse side can I add a node with running containers using incus cluster add. All this without rebooting and containers running none stop.
I would play with it myself but since this is a production cluster with real working containers I decided not to play with it unless I get ‘official’ approval from the top incus brains…
Thank you for your answer.
SO I guess if I have 4 machines A,B,C,D and I remove A,B,C, from cluster individually since they have nothing then D is left by itself. I can then rejoin A,B,C into a new Cluster and D stays by itself as a cluster of one. And this can be done without a reboot?
Also if I remember a two node cluster is really not a full cluster. Is that correct?
Yeah, a 2 machine cluster isn’t HA, it’s still a cluster but if whatever machine is running the database goes down, the other server will not respond until that machine comes back.