Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
manifests/0000_90_kube-apiserver-operator_04_servicemonitor-apiserver…
…: Rename to kube-apiserver-performance-recording-rules Exactly like 7457da4 (manifests: rename API performance dashboard, 2023-10-10, #1565), this pivot: * Turns the original (group, namepspace, name) into a delete manifest, with no capability annotation (all clusters should delete this resource). * Creates a (group, namespace, newName) manifest with the original spec content and capability annotation (only clusters with the cap enabled should have this resource). which allows updates like: 1. Old release with the old (group, namepspace, name) not annotated for a capability. 2. Request update to new release. 3. Outgoing cluster-version operator compares the outgoing manifests with the incoming manifests to decide if any capabilities need to be implicitly enabled: a. It sees that the old (group, namepspace, name) isn't labeled with any capabilities, so no need to implicitly enable anything there. b. It sees that the new (group, namespace, newName) is annotated for the capability, but it doesn't see anything with that g,n,n getting reconciled in the outgoing manifest set, so it thinks "just some new manifest for a capability I do not care about", and does not enable the annotated capability. In a separate pull request than 7457da4, because this one doesn't need to go back to 4.14. The RecordingRule is from 04f36e7 (manifests: add new PrometheusRule for recording rules, 2023-06-28, #1521), which landed in 4.14, so 4.13 clusters do not have a matching (group, namespace, name) manifest to worry about matching.
- Loading branch information