8000 Precompile the fixture in the suite-timeout integration test by onsi · Pull Request #1689 · onsi/ginkgo · GitHub
[go: up one dir, main page]

Skip to content

Precompile the fixture in the suite-timeout integration test - #1689

Merged
onsi merged 1 commit into
masterfrom
fix-timeout-test-flake
Oct 9, 2026
Merged

onsi merged 1 commit into
masterfrom
fix-timeout-test-flake

Conversation

@onsi
@onsi onsi commented Oct 9, 2026
Copy link
Copy Markdown
Owner

Interrupt and Timeout when the suite times out has failed on master 4 times since August, most recently 38002110196. Each time it failed at gbytes.Say("Sleeping..."): the hanging fixture never got far enough to print anything.

The ginkgo --timeout clock starts before compilation (ginkgo/run/run_command.go). On a loaded runner, compiling the fixture could use up most of the test's 5s budget.

The fix builds the fixture with ginkgo build first, the same way the neighbouring interrupt test does. It then runs ./hanging.test with the timeout, so the 5s covers only the suite run.

The spec passes locally with -race.

🤖 Generated with Claude Code

https://claude.ai/code/session_01PHpCwq4zwCazRetYBgay4u

ginkgo's --timeout clock starts before it compiles the suite.  On a
loaded CI runner, compiling the hanging fixture could eat most of the 5s
timeout, so the spec never printed "Sleeping..." and the test failed.  It
has flaked this way on master four times since August.

Build the fixture first, as the neighbouring interrupt test already does,
then run the precompiled binary with the timeout.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PHpCwq4zwCazRetYBgay4u
@onsi
onsi marked this pull request as ready for review October 9, 2026 23:36
@onsi
onsi merged commit 699632a into master Oct 9, 2026
9 checks passed
@onsi
onsi deleted the fix-timeout-test-flake branch October 9, 2026 23:49
onsi added a commit that referenced this pull request Nov 2, 2026
This is the same flake #1689 fixed for its neighbour. ginkgo's --timeout
clock starts before it compiles, and on a loaded CI runner (with -race)
compiling the four timeout fixtures used up the whole 10s budget. Ginkgo
then reported that all four suites "did not run because the timeout
elapsed" before TimeoutA ever started. This failed the master build for
db1c4c8.

The test now builds the fixtures first and runs the precompiled
binaries. They keep the same suite names and paths in the output.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FockBur8KAmXY5teeBBkn7
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant

0