Closed joelsmith closed 5 months ago
Changes to config/manager/manager.yaml and the resulting generated changes to bundle/manifests/keda.clusterserviceversion.yaml are to make them match https://github.com/kedacore/keda-olm-operator/blob/46bdc560770b75b2ed2b04813941695f877d57e2/config/manifests/bases/keda.clusterserviceversion.yaml#L632-L633 and https://github.com/kedacore/keda-olm-operator/blob/46bdc560770b75b2ed2b04813941695f877d57e2/config/manifests/bases/keda.clusterserviceversion.yaml#L623-L627
The change to add a label in config/manifests/bases/keda.clusterserviceversion.yaml is to make it match the bundle-generated CSV.
Thanks for the detail and background, very much appreciated.
Oh yuck, and you had to change the keda.yaml parsing too. Chores have been rough this round :smile:
/lgtm
Also:
Checklist
Notes for reviewers: While doing the upstream 2.12.1 release, I noticed a little weirdness in the KEDA OLM operator's
hack/relprep.sh
script. Namely, that it isn't keepingconfig/manifests/bases/keda.clusterserviceversion.yaml
andbundle/manifests/keda.clusterserviceversion.yaml
in sync. Apparently, the former doesn't get updated by theoperator-sdk generate kustomize manifests -q
command like I thought it did, and the latter does get updated, but is only used for our manual testing workflow (when we're developing or doing a release). While playing around with it, I also did some test runs on 2.13.0 and noticed that there is a new CRD coming our way which lives ineventing.keda.sh
instead ofkeda.sh
, so that also broke the script. So I've opened a PR to "fix" all of that. I'm using quotes because some of the things aren't actually fixed by the script, only detected, so that a human knows which things to manually update. I didn't figure it was worth the work to automate them at this point, but we can always change our mind later if I'm wrong.