I currently have all my clusters updating simultaneously on Rancher 2.12.0, all seemingly stuck with the message "KDM release data is empty for v1.33.3+rke2r1" (or slight variation in release), all with one controller stating "Reconciling. Waiting for plan to be applied". Should I be worried?

did you hardcode a custom KDM URL in your settings, and forget to update it when upgrading rancher to 2.12?

I have changed the rke-metadata-config entry from 2.11 to the one that you've shown. There were a number of other 2.11 urls too - ui-index and ui-dashboard-index. I definitely didn't deliberately hardcode anything, just tried following instructions to install, then to update. I might well have missed something. Is there a doc showing what I should be doing in these circumstances? Do I just wait until the next update to see if it clears up?

those would normally be handled automatically as part of the upgrade. if they were not, that indicates someone went in and changed them at some point so that they are not using the default that is built into the product. If that is done, you need to update them manually after upgrading

They definitely weren't changed manually. I must have missed something when I did the upgrade, not entered something I should have. Should I just wait for the daily update to make the warnings go away?

Are you running Rancher as a standalone docker container? Or deployed via helm chart?

If that's just adjusting the global variables via Settings, that was done after you first mentioned it. If there's more to do, if you could just point me towards the right part of the docs I'll run through that. ui-index https://releases.rancher.com/ui/release-2.12/index.html ui-dashboard-index https://releases.rancher.com/dashboard/release-2.12/index.html rke-metadata-config

{
  "refresh-interval-minutes": "144",
  "url": "<https://releases.rancher.com/kontainer-driver-metadata/release-v2.12/data.json>"
}
{
  "refresh-interval-minutes": "144",
  "url": "<https://releases.rancher.com/kontainer-driver-metadata/release-v2.12/data.json>"
}

I cut down 1440 to 144. It's a helm installation, three Rancher servers, three MariaDB as a Galera cluster in place for etcd (though only the first was in the initial Rancher configuration).

Hi guys! I ran into the same issue after upgrading to Rancher v2.12.1 and fixed it like this: 1. First I verified the

rke-metadata-config

was set to v2.12 — it already was, so no change needed. 2. In Rancher: ☰ → Cluster Management → Drivers → Refresh Kubernetes Metadata. 3. After it refreshed, I went to the local cluster (where Rancher runs) and redeployed the Fleet deployments — both the agent and the controller. 4. Finally, I redeployed the

rancher

deployment itself. After that, a quick refresh showed all my RKE2 clusters finishing their updates and returning to Active with no messages. Hope this helps! @best-appointment-66450

Thank you very much! I'll take a poke at it.

Awesome! Glad that did the trick. Good luck with the new cluster! 🍻