Kube apiserver error logs caused by system.antrea.tanzu.vmware.com

[antoninbas]

Describe the bug
I observe the following in the apiserver logs since we merged the antctl changes (about every minute or so):

I0117 03:41:45.385678       1 controller.go:107] OpenAPI AggregationController: Processing item v1beta1.system.antrea.tanzu.vmware.com
E0117 03:41:45.388894       1 controller.go:114] loading OpenAPI spec for "v1beta1.system.antrea.tanzu.vmware.com" failed with: OpenAPI spec does not exist
I0117 03:41:45.388913       1 controller.go:127] OpenAPI AggregationController: action for item v1beta1.system.antrea.tanzu.vmware.com: Rate Limited Requeue.

To Reproduce
Deploy Antrea on a cluster, check the logs of the kube-apiserver Pod using kubectl.

Versions:
Antrea version:

===> Version information <===
VERSION: v0.3.0-dev
GIT_SHA: 223eb44
GIT_TREE_STATE: clean
RELEASE_STATUS: unreleased
DOCKER_IMG_VERSION: v0.3.0-dev-223eb44

[antoninbas]

@weiqiangt I think you have a fix ready for this (I asked you to remove it from the antctl framework refactor PR). Do you think you can open a new PR with the fix so we can close this for 0.4.0? Thanks.

[weiqiangt]

@weiqiangt I think you have a fix ready for this (I asked you to remove it from the antctl framework refactor PR). Do you think you can open a new PR with the fix so we can close this for 0.4.0? Thanks.

The OpenAPI changes depend on the refactoring, I think it is not able to have a separate PR for the fix.

[tnqn]

@weiqiangt I met this issue and did some investigation, I see the OpenAPI endpoint doesn't have to have specs of all its paths (even kube-apiserver doesn't have all OpenAPI specs of its paths).
I see your refactoring PR doesn't have it yet, I can push a separate PR to fix it. Please let me know if it's ok.

[weiqiangt]

Thanks @tnqn, the OpenAPI changes were moved to another CL which depends on the refactor. I appreciate if you could have a fix.