Robert Burke
Robert Burke
It looks like Python does have some drain tests already authored, so the big question is can we make them work in loopback mode. They're currently skipped in the portable_runner_test.py...
Technically yes, but marked as skipped by the hierarchy, since portable runners couldn't trigger drain previously. The "pre work" would be making it so the python sdk could call drain...
@dependabot recreate
R: @byronellis @damondouglas This is a smaller change than it looks, it only touches so many files because of the swap to the Standard Library slog package. Otherwise most files...
* PostCommit XVR GoUsingJava Dataflow - Permared ( low priority to investigate and fix. (#30519, and earlier with #28339), likely a docker push issue) * XVR_Direct PostCommit 3.8 - (Other...
> https://github.com/apache/beam/actions/workflows/beam_PostCommit_XVR_Direct.yml?query=branch%3Amaster looks permared Yup. That one is #30517
> Which of the errors above are a regression in 2.59.0 (if any) ? None of them, given the age of the root issues that I linked.
It looks like it wasn't able to find the Go SDK container? ``` 2024-08-23T13:25:39.8797158Z Error response from daemon: manifest for apache/beam_go_sdk:dev not found: manifest unknown: manifest unknown ``` Very weird....
> Will try to downgrade GRPC IO, looks like it had a release around Aug 22 > > https://pypi.org/project/grpcio/#history Agreed. The failures seem largely GRPC related.
With various discussions, this is moved to the 2.60.0 release scheduled to be cut on October 2nd.