I'm aware of the docs. An option to migrate data to a new VM cluster via the operator without manual work would be nice though.

ECK operator handles it like this for example. Therefore users are less annoyed ;)

Oh... Good to know...

I've heard the community wants some solution for rebalancing :D

In the end this recent change in the config was the problem:

resources:
requests:
cpu: 10m
memory: 64m
limits:
memory: 64m

Setting memory to "64Mi" fixed it.

So even "64m" is valid it seems just not enough and produces this weird error message....

#wtf

Warning FailedCreatePodSandBox 13s kubelet Failed to create pod sandbox: rpc error: code = Unknown desc = failed to create containerd task: failed to create shim task: OCI runtime create failed: runc create failed: unable to start container process:...

Turns out, it's not "no space left on device".

Plenty of space and inodes available.

I've set "automountServiceAccountToken: false" in the default #serviceaccount to see if the #container starts without mounting the token.

Now the errormessage changed to:

I mean... at least they won't go above 1MB 🤣

I hoped github.com/kubernetes-s... would take over at some point but the repo seems kinda dead :(

Quelle?