Skip to main content

Performance Guidance

Scale targets​

The dashboard is designed for lazy loading:

  • 100 repositories
  • 500 branches
  • 100,000 workflow runs
  • Millions of tests (via per-run lazy load)

Recommendations​

  • Use history and retain-days action inputs to prune old runs
  • Each run is a separate JSON file — pruning removes entries from history.json but orphan run files may remain until manually cleaned
  • The run detail page virtualizes the test list for large suites
  • Index files stay small by storing summaries only

Git churn​

  • Only touched repositories and branches are updated per workflow run
  • Run files are append-only (never rewritten)
  • Keep commits scoped to the publishing repository's subtree