Run the unit tests on two workers again, and fail at once when one is killed (#1184) #1185

Merged

tiagoagueda merged 1 commit from 1184-unit-two-workers into main

2026-10-31 19:29:56 +00:00

tiagoagueda commented

2026-10-31 17:59:08 +00:00

Owner

Runs 1778 and 1793 on 31 October each lost a unit worker in the measured step, with [gwN] node down: Not properly terminated under -n 3 in the job's 2 GB container. xdist replaced the worker and never finished, so the job hung until the runner's deadline. Every test it reached had passed, and the log named no test.

ci.yml and every-python.yml: back to -n 2, which was stable before #1085. This costs about 3.5 minutes of the unit step.
--max-worker-restart=0: a killed worker fails the job at once, with the test it was running in the report.
CONTRIBUTING.md: says -n 2.

Re-measuring three workers is for #1033 (0.6.0).

Release blocker: the v0.5.0 tag needs a green CI on main.

Locally: the tests that read the workflows pass (tests/test_release_tools.py, 40 tests).

Closes #1184

🤖 Generated with Claude Code

Runs 1778 and 1793 on 31 October each lost a unit worker in the measured step, with `[gwN] node down: Not properly terminated` under `-n 3` in the job's 2 GB container. xdist replaced the worker and never finished, so the job hung until the runner's deadline. Every test it reached had passed, and the log named no test. - `ci.yml` and `every-python.yml`: back to `-n 2`, which was stable before #1085. This costs about 3.5 minutes of the unit step. - `--max-worker-restart=0`: a killed worker fails the job at once, with the test it was running in the report. - `CONTRIBUTING.md`: says `-n 2`. Re-measuring three workers is for #1033 (0.6.0). **Release blocker:** the `v0.5.0` tag needs a green CI on `main`. Locally: the tests that read the workflows pass (`tests/test_release_tools.py`, 40 tests). Closes #1184 🤖 Generated with [Claude Code](https://claude.com/claude-code)

tiagoagueda added 1 commit

2026-10-31 17:59:08 +00:00

Run the unit tests on two workers again, and fail at once when one is killed (#1184 )

CI / Unit tests and coverage (Python 3.14) (pull_request) Successful in 13m43s

Details

CI / Checks: lint, migrations, catalogues, settings, built files (pull_request) Successful in 1m23s

Details

CI / Browser tests (Chromium) (pull_request) Successful in 32m34s

Details

CI / Security audit (pull_request) Successful in 48s

Details

CI / Database tests on PostgreSQL (pull_request) Successful in 3m56s

Details

0cdac558fb

Runs 1778 and 1793 (31 October) each lost a worker in the measured step,
`[gwN] node down: Not properly terminated` under -n 3 in the job's 2 GB,
the failure #1085 had measured three workers against; tiers 3-4 have added
to the suite since, and --cov adds to each worker. xdist replaced the
worker and never finished, so the job hung to the runner's deadline with
every test it reached passed, naming nothing.

Back to -n 2 in ci.yml and every-python.yml, stable before #1085, and
--max-worker-restart=0, so a killed worker fails the job at once with the
test it was running. Three workers are for #1033 to measure again.

Closes #1184

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>