After upgrading k3s Adding GroupVersion is being repeated in the k3s.log

[hickersonj]

Environmental Info:
K3s Version:

k3s version v1.27.3+k3s1 (fe9604ca)
go version go1.20.5

Node(s) CPU architecture, OS, and Version:
Linux build 5.14.18-100.fc33.x86_64 #1 SMP Fri Nov 12 17:38:44 UTC 2021 x86_64 x86_64 x86_64 GNU/Linux

Cluster Configuration:

1 node, 1 server and 1 agent

Describe the bug:

I upgraded k3s to v1.27.3+k3s1 and have noticed a log message regarding Adding GroupVersion is repeated in the k3s.log. I do not see this when upgrading using another k8s orchestration manager with kubernetes v1.27.3.

Steps To Reproduce:

• Upgrade k3s from version v1.26.2+k3s1 to v1.27.3+k3s1

Expected behavior:

The k3s.log should not printout repeated logs.

Actual behavior:

The following messages are repeated in the k3s.log:

I0723 00:08:17.208415   18058 handler.go:232] Adding GroupVersion helm.cattle.io v1 to ResourceManager
I0723 00:08:17.208638   18058 handler.go:232] Adding GroupVersion traefik.containo.us v1alpha1 to ResourceManager
I0723 00:08:17.208933   18058 handler.go:232] Adding GroupVersion migrations.kubevirt.io v1alpha1 to ResourceManager
I0723 00:08:17.209021   18058 handler.go:232] Adding GroupVersion kubevirt.io v1 to ResourceManager
I0723 00:08:17.209151   18058 handler.go:232] Adding GroupVersion kubevirt.io v1alpha3 to ResourceManager
I0723 00:08:17.209358   18058 handler.go:232] Adding GroupVersion instancetype.kubevirt.io v1alpha1 to ResourceManager
I0723 00:08:17.209439   18058 handler.go:232] Adding GroupVersion instancetype.kubevirt.io v1alpha2 to ResourceManager
I0723 00:08:17.209752   18058 handler.go:232] Adding GroupVersion crd.projectcalico.org v1 to ResourceManager
I0723 00:08:17.210052   18058 handler.go:232] Adding GroupVersion traefik.containo.us v1alpha1 to ResourceManager
...

Additional context / logs:

I tried minikube version: v1.31.1 which is using kubernetes v1.27.3 and it is not repeating the line Adding GroupVersion in the log.

[brandond]

This is coming from the core Kubernetes apiserver, not K3s:
https://github.com/kubernetes/kubernetes/blob/v1.27.3/staging/src/k8s.io/apiserver/pkg/endpoints/discovery/aggregated/handler.go#L232

I'm not aware of anything we're doing that would cause custom resource discovery to run or re-run more frequently on K3s than it would on any other Kubernetes distribution.

[brandond]

This appears to be coming from the ResourceDiscoveryManager controller, which is gated behind the AggregatedDiscoveryEndpoint feature-gate, which is on by default in Kubernetes 1.27. I am going to close this out, as it appears to be the expected behavior when using the default feature-gates and log levels.