You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
helm-release-enable-tests policy is never ready because its match covers every resource #6135
A read-only check of prod on 2026-11-12 found the helm-release-enable-tests MutatingPolicy at ready: false since it was created on 2026-10-26. Its RBACPermissionsGranted condition says Kyverno's reports controller lacks get, list and watch on every resource (*/*). The mutation itself works: 7 of 49 HelmReleases carry the opt-in label and all 7 have tests enabled.
The cause is the policy's deliberately broad match (apiGroups, resources and apiVersions all *, narrowed to HelmRelease only by a CEL condition). Kyverno wants read access to everything the match names before it will report on the policy.
Problem
A policy that is never ready produces no reports, so the posture tooling cannot show whether it applied. It also makes ready: false normal in the cluster, which hides a policy that is not ready for a real reason (as restrict-tenant-image-repositories was in #6080).
Hypothesis and success signal
Matching only helm.toolkit.fluxcd.iohelmreleases (all served versions) keeps the mutation identical and lets the policy become ready with the reports controller's existing grants. Success: the policy reports ready: true, the existing Kyverno CLI test still passes, and the 7 opted-in HelmReleases still have tests enabled after the next reconcile.
Acceptance criteria
The match names only HelmReleases, without widening the reports controller's permissions.
The existing chainsaw or Kyverno CLI tests for this policy pass, including the label-removed case.
Prod shows ready: true for the policy after deploy, read back read-only.
#5360 - refactor(policy): preserve Helm release defaults on CEL policies [merged]
📝 Issue Planner
Check the box below or use the @coderabbitai plan command to generate an implementation plan and prompts that you can use with your favorite coding assistant.
Create Plan
🧪 Issue enrichment is currently in open beta.
You can configure auto-planning by selecting labels in the issue_enrichment configuration.
To disable automatic issue enrichment, add the following to your .coderabbit.yaml:
issue_enrichment:
auto_enrich:
enabled: false
💬 Have feedback or questions? Drop into our discord!
Delivered and verified in production.#6140 merged as 3dbd0347000a01bb17cb67cc713c6d24ae71ce30 through the merge queue, whose run deployed it to prod (run 43299132983). Read back read-only at about 21:08 UTC:
Generation 2, same match, ready: true, RBACPermissionsGranted=True
helm-release-install-crds
Unchanged, still ready
Kyverno's mutating webhook for each policy
Now registered for HelmReleases only, instead of every resource
Test opt-ins outside kube-system and kyverno
7 of 7 have spec.test.enable: true
Remediation opt-ins outside those namespaces
38 of 38 have retries 3 and remediateLastFailure on install and upgrade
Eight more releases carry an opt-in label but not the defaults: cilium, descheduler, hcloud-cloud-controller-manager, hcloud-csi, metrics-server, snapshot-controller, tetragon and kyverno. All live in kube-system or kyverno. Kyverno's webhooks exclude both namespaces for all three Helm release defaults, the unchanged install-crds policy included, so those releases were never mutated, before this change or after it. Their own manifests set the values they run with (for example, unlimited install retries).
All three acceptance criteria are met:
The match names only HelmReleases, with no change to the reports controller's access.
The CLI tests, including the label-removed cases, pass.
Activity
coderabbitai commentedon Nov 12, 2026
🔗 Related PRs
#5360 - refactor(policy): preserve Helm release defaults on CEL policies [merged]
📝 Issue Planner
Check the box below or use the
@coderabbitai plancommand to generate an implementation plan and prompts that you can use with your favorite coding assistant.🧪 Issue enrichment is currently in open beta.
You can configure auto-planning by selecting labels in the issue_enrichment configuration.
To disable automatic issue enrichment, add the following to your
.coderabbit.yaml:💬 Have feedback or questions? Drop into our discord!
chore: agent-claim #6135 nonce=d7c8e21ff791214d9306110bb284f6e6
chore: agent-claim #6135 nonce=9dcc33d281b276293ef92e76aa0ae1f1
fix(kyverno): re-approve the narrowed Helm release default policies
devantler commentedon Nov 12, 2026
Delivered and verified in production. #6140 merged as
3dbd0347000a01bb17cb67cc713c6d24ae71ce30through the merge queue, whose run deployed it to prod (run 43299132983). Read back read-only at about 21:08 UTC:helm-release-enable-testshelm.toolkit.fluxcd.io/helmreleases,ready: true,RBACPermissionsGranted=Truehelm-release-remediation-retriesready: true,RBACPermissionsGranted=Truehelm-release-install-crdskube-systemandkyvernospec.test.enable: trueremediateLastFailureon install and upgradeEight more releases carry an opt-in label but not the defaults:
cilium,descheduler,hcloud-cloud-controller-manager,hcloud-csi,metrics-server,snapshot-controller,tetragonandkyverno. All live inkube-systemorkyverno. Kyverno's webhooks exclude both namespaces for all three Helm release defaults, the unchangedinstall-crdspolicy included, so those releases were never mutated, before this change or after it. Their own manifests set the values they run with (for example, unlimited install retries).All three acceptance criteria are met:
ready: truefor both policies.