Control Plane node unable to initialize

1

I am using Rancher 2.5 with hetzner nodes as a backend.

We have pools that normally we can just add a node and it will work. But I noticed that the control plan node was under strain and I wanted to increase the count.

However when I did that I now just keep on getting this error message: Driver "hetzner" not found. Do you have the plugin binary "docker-machine-driver-hetzner" accessible in your PATH?; Timeout waiting for ssh key

I am not sure on which machine it cannot find that binary as it exists on all the machines that is trying to start up.

The driver has been working, and it actually managed to spin up 1 node, before it all started breaking.

Now I have a cluster that doesn’t want to recover, has anyone here got any advise on how I can fix it, without destroying the current cluster, and without just upgrading the rancher instance?

Also I seem to be unable to remove a node. Clicking on “Delete” just stays in a loop of going from “Saving” to “Provisioning” state

2

If anyone else gets this issue. It seems that in Rancher 2.5 the driver isn’t always on all the rancher pods. So the way to fix this is to scale the rancher deployment down to a single pod, then deploy.

And once up then you can scale rancher back to what you had before.

1 Like