Skip to content

Commit 1ff83f5

Browse files
committed
test(e2e): fix false claim that a job timeout preserves the report
The catch-timeout comment justified the raised ceiling by asserting that if: always() upload survives a job timeout. It does not: a 180-minute cap ends the job cancelled, and the post-cancellation grace window is consumed by the earlier report-collection step, so both cozyreport.tgz and the job log are lost. Correct the comment to describe the real failure mode. Assisted-By: Claude <[email protected]> Signed-off-by: Aleksei Sviridkin <[email protected]>
1 parent 07c67a8 commit 1ff83f5

1 file changed

Lines changed: 5 additions & 2 deletions

File tree

‎hack/e2e-chainsaw/.chainsaw.yaml‎

Lines changed: 5 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -56,8 +56,11 @@ spec:
5656
# bought its speed by collecting nothing at all — and it is bounded: both legs
5757
# only approach their ceilings on a genuinely degraded cluster, a systemic
5858
# failure usually reds the install gate and skips the E2E step entirely, and
59-
# the report-upload steps are `if: always()` so a job timeout still yields the
60-
# artifact. If this ever does start truncating runs, bound the number of
59+
# the report-upload steps are `if: always()`. That last point does NOT extend
60+
# to a job timeout: a 180-minute cap ends the job `cancelled`, and the ~4m48s
61+
# grace window is consumed by the earlier `Collect report` step, so both
62+
# cozyreport.tgz and the job log are lost (measured on runs 29557003429 and
63+
# 29568562337). If this ever does start truncating runs, bound the number of
6164
# snapshotted tests rather than shrinking these timeouts back to a size that
6265
# guarantees a kill.
6366
error:

0 commit comments

Comments
 (0)