Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Bug]: Go SDK uses PTransform UniqueName as key instead of the global PID for a transform. #29848

Closed
1 of 16 tasks
lostluck opened this issue Dec 20, 2023 · 1 comment · Fixed by #29842
Closed
1 of 16 tasks
Assignees

Comments

@lostluck
Copy link
Contributor

lostluck commented Dec 20, 2023

What happened?

While adding a timer integration test, it was determined that

https://github.com/apache/beam/blob/master/sdks/go/pkg/beam/core/runtime/exec/translate.go#L530

is incorrect and the Go SDK is inconsistently referring to transforms when talking to the runner. It should be using the global map key for the transform here instead of the user readable UniqueName. (also used for combines later in the file)

A spot check shows this prevents timers from firing correctly on the Flink runner, but it's also lead to oddities in how Prism displays metrics to it's web UI needing to be worked around by having two indexes.

While this should not break update compatibility, it would change some things in user visible ways, so it needs to be made with care to fix them where possible.


The minimal fix is to simply update how Timers are looked up, as they're a relatively new feature, and would fix execution on Flink (and also, eventually Prism).

Issue Priority

Priority: 2 (default / most bugs should be filed as P2)

Issue Components

  • Component: Python SDK
  • Component: Java SDK
  • Component: Go SDK
  • Component: Typescript SDK
  • Component: IO connector
  • Component: Beam YAML
  • Component: Beam examples
  • Component: Beam playground
  • Component: Beam katas
  • Component: Website
  • Component: Spark Runner
  • Component: Flink Runner
  • Component: Samza Runner
  • Component: Twister2 Runner
  • Component: Hazelcast Jet Runner
  • Component: Google Cloud Dataflow Runner
@lostluck lostluck added the bug label Dec 20, 2023
@lostluck lostluck self-assigned this Dec 20, 2023
@lostluck
Copy link
Contributor Author

Internal Google testing show at least against FlumeGo and Dataflow this is a no-op. It does the UniqueName lookups as a fallback, and then proceeds to use the normal global ID.

#29842 shows that it allows the included "bounded" timer test to pass on Spark and Flink, and that some additional tests could be run by the py portable runner.

Once #29829 has been merged in, I'll produce a clean version of #29842 to resolve this issue.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging a pull request may close this issue.

1 participant