This message was deleted.

you should move one version at a time - you should also make sure you follow the support matrix to assure ALL of the software around Rancher (such as the OS, Docker and K8s). Is supported on the versions you are jumping too. you might have to play a game of leap frog to get to the latest version. (honestly - at a certain point its easier to just start over)

starting over is probably not really going to be feasible

I cant say for certain but probably. 2.6 nor 2.7 have been certified with anything that old

they're legacy clusters we're moving workloads off them

This is why we STRONGLY advise following the support matrix. It's what we test to work

in what way do you mean "won't work quite right"? we won't be making any changes to those clusters and they will be replaced with new ones within supported version range, we just need them to be connected to rancher in the meantime

Rancher talks to clusters via the kubernetes API if the kubernetes API is too old there are likely to be communication and control issues with rancher and it's agents on the downstream clusters. I have no idea what exact issues you will face or if you will even face any... But that's the point - it's not tested so no one does. You're breaking ground testing for us essentially.

i see