Thanks for confirming, then I am fine to call it a known issue in 2.62.0, 2.61.0 and fix for 2.63.0 and I retract my -1 vote.
On Fri, Jan 17, 2025 at 5:49 PM Chamikara Jayalath via dev < dev@beam.apache.org> wrote: > Yeah, this seems to be happening for numpy 2.1.3. as well so likely the > issue existed before the current release. > > On Fri, Jan 17, 2025 at 6:50 AM XQ Hu via dev <dev@beam.apache.org> wrote: > >> I tested this with my mackbook and with a clean venv and pip install >> "apache-beam[gcp]==2.61.0". numpy==2.1.3 is installed. >> >> On Fri, Jan 17, 2025 at 5:25 AM Valentyn Tymofieiev via dev < >> dev@beam.apache.org> wrote: >> >>> @Chamikara Jayalath <chamik...@google.com> Does this bug also happen in >>> 2.61.0 ? >>> >>> I am seeing that the numpy 2.x upgrade was first introduced earlier in >>> 2.61.0: >>> >>> >>> https://github.com/apache/beam/commit/6129c9a56d52ebb060417cb397e0764cdd8791bc >>> >>> In this case the regression would be a preexisting known issue and >>> according to our release process it doesn't have to block the release, and >>> I would retract my -1 vote as per: >>> >>> >>> - Is this a regression from a previous release? (If no, fix could go >>> to a newer version.) >>> >>> >>> However we also had: >>> https://github.com/apache/beam/commit/fd17dcea9cae8a444fcd1fc9708dffdd2e33cda0 >>> , and it is possible that this commit has introduced the regression that >>> you observe, then it would be better to revert >>> fd17dcea9cae8a444fcd1fc9708dffdd2e33cda0 commit on the release branch and >>> make RC2. >>> >>> I also saw that we reverted a wrong commit on the release branch, >>> sending: https://github.com/apache/beam/pull/33638 >>> >>> >>> On Thu, Jan 16, 2025 at 8:07 PM Valentyn Tymofieiev <valen...@google.com> >>> wrote: >>> >>>> Thanks Cham for flagging this. It sounds like an inconvenience at >>>> minimum, I vote -1 and suggest to rollback the numpy upgrade on the release >>>> branch. Alternative: we upgrade numpy in the SDK containers but that leaves >>>> us with releasing a configuration that had less time to be tested. >>>> >>>> On Thu, Jan 16, 2025 at 6:55 PM Chamikara Jayalath via dev < >>>> dev@beam.apache.org> wrote: >>>> >>>>> BTW my vote is -0 since I'm unsure of the true impact of this :) >>>>> >>>>> On Thu, Jan 16, 2025 at 9:34 AM Chamikara Jayalath < >>>>> chamik...@google.com> wrote: >>>>> >>>>>> Seems like we updated the "numpy" version for this release [1] but we >>>>>> did not update SDK harness containers [2]. >>>>>> One of the jobs I ran (Java multi-lang that uses Python Dataframe) >>>>>> failed due to this with the following error. I suspect pure Python >>>>>> pipelines may also fail due to similar errors though unless users pin the >>>>>> numpy version during job submission. >>>>>> >>>>>> ``` >>>>>> ModuleNotFoundError: No module named 'numpy._core.numeric' >>>>>> ``` >>>>>> >>>>>> May be we can address this by updating the >>>>>> "base_image_requirements.txt" files and re-generating the Python SDK >>>>>> harness containers ? >>>>>> >>>>>> Thanks, >>>>>> Cham >>>>>> >>>>>> [1] https://github.com/apache/beam/pull/33325 >>>>>> [2] >>>>>> https://github.com/apache/beam/blob/77f38cd74735cc732c99d0752ee10c3585a3b0c7/sdks/python/container/py311/base_image_requirements.txt#L106 >>>>>> >>>>>> On Wed, Jan 15, 2025 at 7:43 PM Yi Hu via dev <dev@beam.apache.org> >>>>>> wrote: >>>>>> >>>>>>> +1 (non-binding) >>>>>>> >>>>>>> tested Dataflow Template integration tests with Java 11 and 17: >>>>>>> https://github.com/GoogleCloudPlatform/DataflowTemplates/pull/2133 >>>>>>> >>>>>>> On Tue, Jan 14, 2025 at 1:22 PM Danny McCormick via dev < >>>>>>> dev@beam.apache.org> wrote: >>>>>>> >>>>>>>> +1 (binding) - tested some example ML pipelines on the local >>>>>>>> (interactive) and Dataflow runners. >>>>>>>> >>>>>>>> Thanks, >>>>>>>> Danny >>>>>>>> >>>>>>>> On Mon, Jan 13, 2025 at 12:53 PM XQ Hu via dev <dev@beam.apache.org> >>>>>>>> wrote: >>>>>>>> >>>>>>>>> +1 (non-binding) - tested this with a simple Dataflow ML pipeline: >>>>>>>>> https://github.com/google/dataflow-ml-starter/actions/runs/12752033114/job/35540477801 >>>>>>>>> >>>>>>>>> On Mon, Jan 13, 2025 at 11:28 AM Jack McCluskey via dev < >>>>>>>>> dev@beam.apache.org> wrote: >>>>>>>>> >>>>>>>>>> +1 (non-binding) - downloaded python artifacts, ran direct and >>>>>>>>>> dataflow wordcount pipleines >>>>>>>>>> >>>>>>>>>> On Mon, Jan 13, 2025 at 10:54 AM Kenneth Knowles <k...@apache.org> >>>>>>>>>> wrote: >>>>>>>>>> >>>>>>>>>>> Hi everyone, >>>>>>>>>>> >>>>>>>>>>> Please review and vote on the release candidate #1 for the >>>>>>>>>>> version 2.62.0, as follows: >>>>>>>>>>> >>>>>>>>>>> [ ] +1, Approve the release >>>>>>>>>>> [ ] -1, Do not approve the release (please provide specific >>>>>>>>>>> comments) >>>>>>>>>>> >>>>>>>>>>> Reviewers are encouraged to test their own use cases with the >>>>>>>>>>> release candidate, and vote +1 if no issues are found. For >>>>>>>>>>> guidelines on how to try the release in your projects, check out >>>>>>>>>>> our RC >>>>>>>>>>> testing guide at >>>>>>>>>>> https://github.com/apache/beam/blob/master/contributor-docs/rc-testing-guide.md >>>>>>>>>>> >>>>>>>>>>> Only PMC member votes will count towards the final vote, but >>>>>>>>>>> votes from all community members are encouraged and helpful for >>>>>>>>>>> finding >>>>>>>>>>> regressions. >>>>>>>>>>> >>>>>>>>>>> The complete staging area is available for your review, which >>>>>>>>>>> includes: >>>>>>>>>>> >>>>>>>>>>> - GitHub Release notes: >>>>>>>>>>> https://github.com/apache/beam/milestone/26 >>>>>>>>>>> - The official Apache source release: >>>>>>>>>>> https://dist.apache.org/repos/dist/dev/beam/2.62.0/ which is >>>>>>>>>>> signed with the key with fingerprint D20316F712213422, found in >>>>>>>>>>> https://dist.apache.org/repos/dist/release/beam/KEYS >>>>>>>>>>> - All artifacts to be deployed to the Maven Central >>>>>>>>>>> Repository: >>>>>>>>>>> >>>>>>>>>>> https://repository.apache.org/content/repositories/orgapachebeam-1395/ >>>>>>>>>>> - source code tag "v2.62.0-RC1" ( >>>>>>>>>>> https://github.com/apache/beam/tree/v2.62.0-RC1) >>>>>>>>>>> - website pull request listing the release and blog post: >>>>>>>>>>> https://github.com/apache/beam/pull/33567 >>>>>>>>>>> - website pull request publishing the API reference manual: >>>>>>>>>>> https://github.com/apache/beam-site/pull/675 >>>>>>>>>>> - Python artifacts are deployed along with the source >>>>>>>>>>> release to >>>>>>>>>>> https://dist.apache.org/repos/dist/dev/beam/2.62.0/ and PyPI >>>>>>>>>>> https://pypi.org/project/apache-beam/2.62.0rc1/. >>>>>>>>>>> - Go artifacts and documentation are available at >>>>>>>>>>> >>>>>>>>>>> https://pkg.go.dev/github.com/apache/beam/sdks/v2@v2.62.0-RC1/go/pkg/beam >>>>>>>>>>> - Validation sheet with a tab for 2.62.0 release to help >>>>>>>>>>> with validation: >>>>>>>>>>> >>>>>>>>>>> https://docs.google.com/spreadsheets/d/1qk-N5vjXvbcEk68GjbkSZTR8AGqyNUM-oLFo_ZXBpJw/edit?gid=1493515573#gid=1493515573 >>>>>>>>>>> - Docker images published to Docker Hub: >>>>>>>>>>> https://hub.docker.com/search?q=apache%2Fbeam&type=image >>>>>>>>>>> - Pull request to run tests against release branch: >>>>>>>>>>> https://github.com/apache/beam/pull/33424 >>>>>>>>>>> - Github Release pre-release page: >>>>>>>>>>> https://github.com/apache/beam/releases/tag/v2.62.0-RC1 >>>>>>>>>>> >>>>>>>>>>> The vote will be open for at least 72 hours. It is adopted by >>>>>>>>>>> majority approval, with at least 3 PMC affirmative votes. >>>>>>>>>>> >>>>>>>>>>> Thanks, >>>>>>>>>>> Kenn >>>>>>>>>>> >>>>>>>>>>
