Any idea on this? https://rancher-users.slack.com/archives/C01PHNP149L/p1744728416903859

I think you’re putting your values in the wrong place. chartvalues should only be for the CNI and other things where the config is managed by rancher. If you want to provide a HelmChartConfig for rke2-coredns that should go in the additional manifest section. That’s why it’s getting removed when you edit in the UI - it doesn’t belong there.

We also currently store our

rke2-coredns

custom config in

chart_values

. We ran into an issue once where adding a node to the cluster through the UI erased those values and redeployed CoreDNS with the defaults. @creamy-pencil-82913 • Would you recommend moving these customizations to additional manifests instead? • Do you have any insight into why adding a node via the GUI would wipe that config? • Also, would it be safer in the meantime to add nodes directly by editing the cluster resource YAML?

additional manifest is the correct place for anything not managed by rancher itself

managed by rancher itself

Does that include cilium?

it kinda depends on how the cluster is configured. Out of the box, the only thing that goes in chartValues (“Add-on” tabs in the UI) is the CNI config for whatever CNI you have selected. If you pick a cloud provider that has chart options, then additional sections will be added in the UI to configure that as well. Anything that doesn’t have an “Add-on: ” section in the UI should go in Additional Manifest instead of Chart Values.

Even if we provision clusters via Terraform, never through the UI, your advice would stay the same? The only cluster manipulation we do in the UI is adding nodes

If you ever want to use the UI to do anything, then you need to play by the UI’s rules.

Okay, that's good to know, thank you