Flaky E2E tests – tracking intermittent failures

[zhzhuang-zju]

Our E2E tests have been failing randomly lately. These flaky failures make it difficult to determine if a genuine issue was introduced or if it’s just a test problem.

This issue is to collect and track intermittent E2E failures in one place. If you see a test fail and it passes on retry, please add a comment with:

• The test name
• A short error snippet or symptom
• Link to the CI run

Identified E2E test failures:

• metrics presence testing
  failure e2e link: https://github.com/karmada-io/karmada/actions/runs/18395251404/job/52413766044?pr=4574#step:6:742

  • skip calculate available replicas when no need #4574, @zhzhuang-zju

• Application failover testing with purgeMode gracefully
  failure e2e link: https://github.com/karmada-io/karmada/actions/runs/18553955121/job/52887389171#step:6:444

  • e2e: enhance application failover testing #6845, @zhzhuang-zju

• Karmada Operator E2E test: PriorityClass configuration testing

  • e2e: Should ignore conflict when updating resources #6843, @zhzhuang-zju

• Karmada instance status testing

  • operator: avoid unnecessary reconcile #6908 (@zhzhuang-zju )

• Deploy workloads should be rejected in case of no enough quota left

  • flaky test: deploy workloads should be rejected in case of no enough quota left #6907 (@zhzhuang-zju )

• Multi-Components: FederatedResourceQuota enforcement testing

  • flake: Multi-Components: FederatedResourceQuota enforcement testing #6941 (@zhzhuang-zju )