Draft: CI Optimization - Increase Playwright workers and allow_failure for mirrornet-haf-node-replay
Summary
CI optimization changes to improve pipeline performance:
- Increase Playwright workers from 1 to 2 - Improves E2E test parallelism within each shard
- Add allow_failure to mirrornet-haf-node-replay - Prevents flaky job from blocking pipelines
Results
- Blog E2E tests: ~11.8% faster (3691s vs 4185s baseline)
- Pipeline 140545 passed successfully
Test Plan
-
Pipeline passes with all E2E tests green -
allow_failure correctly handles mirrornet-haf-node-replay failures
See REPORT.md for detailed analysis.
Edited by Dan Notestein