[2mind]

Summary

The internal proxy's server certificate was built with a hard-coded cluster domain:

dnsNames := []string{
	fpService,
	fmt.Sprintf("%s.%s", fpService, r.rootShard.Namespace),
	fmt.Sprintf("%s.%s.svc.cluster.local", fpService, r.rootShard.Namespace),
}

while everything that dials the proxy goes through resources.GetRootShardProxyBaseHost() → <service>.<namespace>.svc.<clusterDomain>. With spec.clusterDomain: cozy.local a certificate for the proxy looks like this, and the operator's own client cannot verify it:

["kcp-proxy","kcp-proxy.kcp-system","kcp-proxy.kcp-system.svc.cluster.local"]

The affected client is client.NewInternalKubeconfigClient() (via NewRootShardProxyClient), which the kubeconfig-rbac controller uses to create the RBAC for Kubeconfig.spec.authorization — so scoped permissions were silently not granted for every Kubeconfig whose target is a FrontProxy.

Change

• resources.GetRootShardClusterDomain() — new helper for the "cluster domain or default" lookup that was open-coded in several places; GetRootShardBaseHost() and GetRootShardProxyBaseHost() now use it too.
• frontproxy.serverDNSNames() — the DNS name list is factored out of serverCertificateReconciler() and its fully qualified Service name is derived from that cluster domain instead of the hard-coded cluster.local.
• Unit test covering the root-shard proxy and front-proxy cases, with the default and a custom cluster domain.

The SAN list keeps the short forms, and for front-proxies the external hostnames are added exactly as before; only the fully qualified name changes.

Verification

• go build ./..., go vet ./internal/..., go test ./internal/... ./pkg/... — all pass.
• With spec.clusterDomain: cozy.local the generated Certificate now carries kcp-proxy.kcp-system.svc.cozy.local, which is exactly the entry we had to add by hand through spec.proxy.certificateTemplates in our installation (kcp-operator v0.9.0, chart 0.7.10), where this was the reason the kubeconfig-rbac bindings never appeared — Kubeconfig.status.authorization.provisionedCluster stayed empty.

What Type of PR Is This?

/kind bug

Related Issue(s)

Fixes #303

Release Notes

Use the RootShard's cluster domain in the internal proxy server certificate instead of a hard-coded cluster.local

[kcp-ci-bot]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by:
Once this PR has been reviewed and has the lgtm label, please assign mjudeikis for approval. For more information see the Code Review Process.

The full list of commands accepted by this bot can be found here.

Details

Needs approval from an approver in each of these files:

• OWNERS

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[kcp-ci-bot]

Hi @2mind. Thanks for your PR.

I'm waiting for a kcp-dev member to verify that this patch is reasonable to test. If it is, they should reply with /ok-to-test on its own line. Until that is done, I will not automatically test new commits in this PR, but the usual testing commands by org members will still work. Regular contributors should join the org to skip this step.

Once the patch is verified, the new status will be reflected by the ok-to-test label.

I understand the commands that are listed here.

Details

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository.

[2mind]

/kind bug

[2mind]

Friendly ping for review (fixes #303). We currently work around the missing internal-proxy SAN by setting spec.proxy.certificateTemplates by hand; happy to drop that once this lands and is released.