@jolly-processor-88759 thanks for the advice on upgrading to 2.6.8, that got some things moving. Unfortunately now I'm running into another issue where nodes aren't provisioning because of a duplicate mount error for kubelet from an extra bind that was accidentally left in the cluster yaml while trying to get things moving before upgrading Rancher:

level=error msg="Remotedialer proxy error" error="Error response from daemon: Duplicate mount point: /opt/rke/var/lib/kubelet"

. I've removed the offending extra bind from the cluster yaml, but no matter what I do the plan the agent receives doesn't seem to change (still includes the new/unwanted bind:

level=info msg="For process kubelet, Binds has changed from

) and the error persists, preventing any progress. How can I force Rancher to send the new plan without this bind to the agents?

have you tried restarting

rancher-system-agent

Is that different from the

cattle-node-agent

? Where does it run?

oh wait you are running this bare-metal and not downstream provisioned, correct?

My Rancher install is 3 or 4 years old, drug through a bunch of upgrades, running as a docker container on a Digital Ocean droplet. The downstream cluster is bare metal, Rancher provisioned and started on something like k8s 1.16.

So its a RKE install vs K3s or RKE2

Correct it's RKE1

I haven't had to deal with RKE1 in a while. Maybe @creamy-pencil-82913 Might be able to weigh in on the issue.