Skip to content

fix(operational-review): correct kyverno container label and memory limit#11

Open
drewr wants to merge 1 commit into
mainfrom
fix/ops-review-kyverno-query-labels
Open

fix(operational-review): correct kyverno container label and memory limit#11
drewr wants to merge 1 commit into
mainfrom
fix/ops-review-kyverno-query-labels

Conversation

@drewr

@drewr drewr commented Apr 27, 2026

Copy link
Copy Markdown
Contributor

Summary

Three fixes to skills/operational-review/queries.md, all diagnosed from live VictoriaMetrics prod data:

  • Container label mismatch (3 queries): Queries filtering container="kyverno-admission-controller" returned no data. The actual label value used by cAdvisor and kube-state-metrics is container="kyverno" (confirmed via label_values). All three affected queries (memory working set, memory limit, restart rate) are corrected to container="kyverno" and scoped with namespace="kyverno-system".

  • Stale memory limit comment: The comment referencing the post-infra#2220 memory limit said 1536Mi. The actual current limit confirmed across all 18 clusters is 2048Mi (2Gi). Updated accordingly.

  • kyverno_update_requests_total unavailable: This metric is not emitted by the Kyverno version running in prod (absent from VictoriaMetrics). A NOTE block is added to that query section directing operators to use the Kubernetes MCP UpdateRequest path (already documented above it) as the only viable approach.

Also bumps datum-platform plugin version 1.6.0 → 1.6.1 per release checklist.

Test plan

  • Verify container_memory_working_set_bytes{namespace="kyverno-system", container="kyverno"} returns data in prod VictoriaMetrics
  • Verify kube_pod_container_resource_limits{namespace="kyverno-system", resource="memory", container="kyverno"} returns data and limit value matches 2048Mi
  • Verify increase(kube_pod_container_status_restarts_total{namespace="kyverno-system", container="kyverno"}[1d]) returns data
  • Confirm kyverno_update_requests_total remains absent from VictoriaMetrics (NOTE block is accurate)

Generated with Claude Code

…imit

- Replace container="kyverno-admission-controller" with container="kyverno"
  in all three metric queries (memory usage, memory limit, restart rate);
  the former label returns no data in VictoriaMetrics — cAdvisor and
  kube-state-metrics use the container name "kyverno", not the pod name.
- Add namespace="kyverno-system" scoping to each corrected query.
- Update stale 1536Mi memory limit comment to 2048Mi (2Gi), confirmed
  across all 18 clusters from live prod metrics.
- Flag kyverno_update_requests_total as unavailable in the current Kyverno
  deployment; direct operators to the Kubernetes MCP UpdateRequest path.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant