damccorm opened a new pull request, #41057: URL: https://github.com/apache/beam/pull/41057
### Summary Hardens `.github/workflows/build_wheels.yml` (`Build wheel` and `Build RC wheels` steps) by retrying `cibuildwheel --output-dir wheelhouse` up to 3 times (cleaning `wheelhouse/` before retrying), matching the existing retry workaround in `:sdks:python:bdistPy*` (`sdks/python/build.gradle`). ### Root Cause In `Build python source distribution and wheels` ([run 42285495566](https://github.com/apache/beam/actions/runs/42285495566), job `128125638525`, as well as 5 other `ubuntu-24.04` wheel jobs in recent days), `cibuildwheel` finished building, repairing, and copying the wheel to the host, and then failed with: ``` File ".../cibuildwheel/oci_container.py", line 349, in __exit__ self.process.wait(timeout=30) subprocess.TimeoutExpired: Command '['docker', 'start', '--attach', '--interactive', 'cibuildwheel-...']' timed out after 30 seconds ``` This is the same sporadic `OCIContainer.__exit__` timeout tracked in https://github.com/apache/beam/issues/28703 and https://github.com/pypa/cibuildwheel/issues/1692, which was previously worked around with a 3-attempt retry loop in `sdks/python/build.gradle` but not in `.github/workflows/build_wheels.yml`. ### Verification - Triggered `Build python source distribution and wheels` on branch `fix/build-wheels-cibuildwheel-retry`: https://github.com/apache/beam/actions/runs/42297136697 -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
