Test Trends

Test Trend Graphs and Flaky Test Handling

Once a job has two builds with results, its page draws a Test Result Trend: passed, skipped and failed counts per build. The branch's two builds show the break and the fix:

The Test Result Trend of feature/id-range: one failure in build 1, none in build 2
The Test Result Trend of feature/id-range: one failure in build 1, none in build 2

The axis starts at 8, so one failure looks large. A failing test's Age column counts how many builds it has failed, separating a new break from an old one.

A flaky test passes and fails on the same code, through timing, ordering or shared state. Retrying the whole stage (retry(2) { ... }) hides it and doubles the time of real failures. Better options: