Skip to content

RELEASE-2373: Drop noisy validation and pre-processing latency alerts - #1291

Draft
happybhati wants to merge 1 commit into
redhat-appstudio:mainfrom
happybhati:RELEASE-2373-drop-noisy-latency-alerts
Draft

RELEASE-2373: Drop noisy validation and pre-processing latency alerts#1291
happybhati wants to merge 1 commit into
redhat-appstudio:mainfrom
happybhati:RELEASE-2373-drop-noisy-latency-alerts

Conversation

@happybhati

@happybhati happybhati commented Aug 26, 2026

Copy link
Copy Markdown

Summary

Draft for review with Leandro / release team — please do not merge until agreed.

Proposal for RELEASE-2373: remove the two early-dev latency alerts that generate noise for SPRE without reflecting actionable release-service controller health, and keep controller availability alerts.

Removed

  • ReleaseServiceValidationDurationSeconds (90% under 5s)
  • ReleaseServicePreProcessingDurationSeconds (PromQL used 60s; text said 10s)
  • Related recording rules used only by those alerts

Kept

  • ReleaseServiceControllerManagerPodNotReady
  • ReleaseServiceControllerManagerDown (slo: true / critical)

Why (prior guidance + Slack history)

  • Leandro (SPRE validation flap thread): validation alert came from early SLO guesses before client onboarding; often heals by itself and has not impacted service quality; severity → high was an interim step while RELEASE-2373 redesigns SLOs.
  • Pre-processing was similarly called out as not a meaningful SLO for the service.
  • SPRE (Aurel) linked SPRE-4674 / SPRE-5192 and asked for a permanent solution beyond severity downgrade.
  • Metrics (release_validation_duration_seconds, release_pre_processing_duration_seconds) remain on the operator for Grafana — this PR only stops alerting on them.

Follow-ups (not in this PR)

  • Update formal SLOs in app-interface slo-document.yml (remove or revise validation / pre-processing / unused “finish <1h” entries).
  • Archive or update SOPs linked from the removed alerts.
  • Optional later: new SLIs only if we identify controller-health signals that are actionable (separate from catalog/pipeline failures).

Test plan

  • make selective-check-and-test RULE_FILES="rhobs/alerting/data_plane/prometheus.release_service_alerts.yaml" TEST_CASE_FILES="test/promql/tests/data_plane/release_service_test.yaml"
  • Review with Leandro before marking ready for review / merge
  • After merge: open app-interface + SOP follow-ups; notify SPRE that permanent cleanup landed

Made with Cursor

Remove early-dev latency alerts that flap and rarely indicate actionable
controller failures. Keep controller availability alerts as the paging
signal. Metrics remain available for dashboards.

Co-authored-by: Cursor <cursoragent@cursor.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