Kubernetes CI Metrics
Interactive views for CI job health data from
gs://k8s-metrics
Dashboards
Failures
Jobs that have been consistently failing, sorted by number of consecutive failing days. Helps identify long-standing broken jobs that need attention.
Weekly Flakes
Jobs with flaky test results over the past week. Shows consistency percentage and individual flaky tests within each job.
Daily Flakes
Jobs with flaky test results from the past day. Useful for catching newly introduced flakiness before it becomes chronic.
Job Health (Daily)
Daily snapshot of all CI jobs showing run counts, failure rates, test counts, and duration percentiles (p50, p75, p99).
Job Health (7 Days)
Rolling 7-day view of all CI jobs showing run counts, failure rates, test counts, and duration percentiles (p50, p75, p99).
Presubmit Health
Presubmit job health metrics showing PR failure rates, run counts, and average run times to help identify problematic presubmits.
Raw Data (JSON)
failures-latest.json
— Jobs failing for consecutive days
flakes-latest.json
— Weekly flake data with test-level details
flakes-daily-latest.json
— Daily flake data with test-level details
job-health-latest.json
— Daily job health with failure rates and durations
job-health-weekly-latest.json
— 7-day job health with failure rates and durations
presubmit-health-latest.json
— Presubmit job health with PR failure rates