This message was deleted.

I went ahead and did it in production, because what's life without a little risk. It definitely causes some issues when you have attached volumes. I had about 150 attached volumes, and 50 Longhorn nodes. I'm seeing about 15 volumes in degraded status. From what I can tell when you upgrade Rancher it then goes through each node and runs a plan with the SUC, which starts up a new kube-proxy pod. My guess is that most of the time it's quick enough to not cause any problems, but sometimes it takes longer and causes a replica to stop. Just a hunch