bug: run the e2e gunicorn with 2 workers, not 1 #40

Merged

Gilles merged 1 commit from fix/e2e-gunicorn-single-worker into master

2026-11-02 07:56:41 +00:00

Gilles commented

2026-11-02 07:08:00 +00:00

Owner

Summary

The e2e job's gunicorn ran with no --workers flag (single sync worker), serializing every request during the test run.
AttachmentList.vue's write actions (restore/replace) reload the whole list after writing (settle() -> load()), a second full round trip on top of the write itself; under a single worker this occasionally exceeded wdio's 10s waitforTimeout, more so as the shared CI runner got busier (matches runs #96's two flaky failures, both ~10.07s after the preceding log line).
Production (Dockerfile) already runs --workers 3; gave the e2e dev server a lighter --workers 2.

Closes #39

Test plan

CI e2e job green
Re-run e2e a few times (or check history) to confirm the flake is gone, not just hidden

🤖 Generated with Claude Code

https://claude.ai/code/session_014HiJrn6J6vhyFfC2qyY4uh

## Summary - The e2e job's gunicorn ran with no `--workers` flag (single sync worker), serializing every request during the test run. - `AttachmentList.vue`'s write actions (restore/replace) reload the whole list after writing (`settle()` -> `load()`), a second full round trip on top of the write itself; under a single worker this occasionally exceeded wdio's 10s `waitforTimeout`, more so as the shared CI runner got busier (matches runs #96's two flaky failures, both ~10.07s after the preceding log line). - Production (`Dockerfile`) already runs `--workers 3`; gave the e2e dev server a lighter `--workers 2`. Closes #39 ## Test plan - [ ] CI e2e job green - [ ] Re-run e2e a few times (or check history) to confirm the flake is gone, not just hidden 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_014HiJrn6J6vhyFfC2qyY4uh

Gilles added 1 commit

2026-11-02 07:08:00 +00:00

bug: run the e2e gunicorn with 2 workers, not 1

ci / test (pull_request) Successful in 3m28s

Details

ci / e2e (pull_request) Has been skipped

Details

ci / docker (pull_request) Has been skipped

Details

ci / deploy (pull_request) Has been skipped

Details

66c0f4ffbe

The e2e job's gunicorn had no --workers flag (single sync worker),
serializing every request during the test run -- including the
write+reload round trip that attachment restore/replace triggers
(AttachmentList.vue's settle() reloads the whole list after every
write). That round trip occasionally exceeded wdio.conf.ts's 10s
waitforTimeout, more often as the shared CI runner got busier.
Production already runs --workers 3; give e2e a lighter --workers 2.

Closes #39

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_014HiJrn6J6vhyFfC2qyY4uh

Gilles added the

bug

labels

2026-11-02 07:08:07 +00:00

Gilles commented

2026-11-02 07:35:23 +00:00

Author

Owner