https://rancher.com/ logo
Join Slack
Powered by
This message was deleted.
# rke2
a
This message was deleted.
m
Did you also update the fleet-controller after the upgrade?
If you go to Apps->Installed in your Rancher cluster, in cattle-fleet-system, there should be "fleet". Edit/Upgrade it and pick the newer version (fleet:102.2.0+up0.8.0). This should also update the fleet agent on the downstream clusters. That might fix this. Otherwise, make sure the "rancher-webhook" is on the same version for rancher and downstream.
w
i did not. thank you, i'll give it a shot.
It looks like it did automatically upgrade (I upgraded to 2.7.6 via helm3) and the newest version I have the option to apply is 100.0.3+up0.3.9
the other issue that is probably suboptimal is that my local cluster is running k8s 1.20.15+rke2r2 yet. I'd like to update it to 1.23 or 1.24, but i suspect it'll fail to update
m
Yeah, looks like 1.23 is the minimally supported for it. By local, I assume you mean the one you're running Rancher on and not a downstream? If so, I think you should be good to upgrade it. Definitely take Rancher and etcd backups first, since it's always a good idea. But I imagine you'll have less issues upgrading than staying on an unsupported version.
I'm a little surprised you were able to upgrade to 2.7.6 on rke-1.20. Usually Rancher complains about an unsupported k8s and doesn't go through with the helm deployment.
w
nah, it just went. i should have updated local cluster to 1.23 first. I think that was allowed for the local cluster on 2.6.6 as i'd updated the other clusters to 1.23 after updating to 2.6.6.
the k8s upgrade to the local cluster doesn't appear to be doing anything. I don't see anything in the logs for it either. But the config is set up to 1.23. I do see this error, but i think we've had it for a while (perhaps even prior to the upgrade):
Copy code
2023/10/04 17:59:20 [ERROR] error syncing 'rancher-partner-charts': handler helm-clusterrepo-ensure: git -C /var/lib/rancher-data/local-catalogs/v2/rancher-partner-charts/8f17acdce9bffd6e05a58a3798840e408c4ea71783381ecd2e9af30baad65974 fetch origin -- 8c5817d641ed79e5cb6146685f688afe394d014b error: exit status 128, detail: error: Server does not allow request for unadvertised object 8c5817d641ed79e5cb6146685f688afe394d014b
, requeuing
I'm going to redeploy some worker nodes that seem to be in an inconsistent state (they're running k8s 1.23 and their control/etcd nodes are running 1.26)
m
Sounds like you got your CP and etcd nodes upgraded, though! Redeploying the workers one at a time should work from there.
w
I did clear up and reregister those nodes (they're also running ceph, outside of containers so i can't reprovision them fully) and they came back up with 1.23 again. I think the rancher install is kind of messed up.
m
If the config is set to 1.23 that makes sense
The weird thing there is how your CP and etcd nodes got to 1.26.
If I were in your shoes I'd go the full way to 1.26 so everything is on the same k8s.
And I'd do that by bumping the config to 1.26
w
okay, i'll give that a shot. thanks.
👍 1
Open in Slack
PreviousNext
Powered by