bvolpato opened a new pull request, #40838: URL: https://github.com/apache/beam/pull/40838
The inference benchmark workflow submits with Python 3.11, but its vLLM image still installs Python 3.10 and copies that version's SDK harness. The SDK rejects the mismatch, so later benchmarks are skipped ([failure](https://github.com/apache/beam/actions/runs/40266709215/job/121390594847)). Use Python 3.11 and its matching Beam harness throughout the image, following the existing deadsnakes installation recipe for the Ubuntu base. Validated a full image build from a fresh SDK source distribution, the SDK compatibility check, and the worker launcher. The compatibility check accepts a Python 3.11 descriptor and rejects a Python 3.10 control. `pip check` reports the same existing PyGObject/pycairo dependency issue on the original Python 3.10 Dockerfile prefix and the updated image. GPU inference and Dataflow were not run. <!-- Thank you for your contribution! Follow this checklist to help us incorporate your contribution quickly and easily: - [ ] If this contribution is large, please file an Apache [Individual Contributor License Agreement](https://www.apache.org/licenses/icla.pdf). - [x] Replace these instructions with a description of your change and why you have made it. - [ ] Mention the appropriate issue in your description (for example: `addresses #123` or `fixes #456`). - [ ] Update `CHANGES.md` with noteworthy changes. See [CONTRIBUTING.md]() for more details. --> -- 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]
