Maximilian Michels commented on BEAM-593:

Until this is probably addressed on the Flink side, let us return 
{{State.DONE}} for all pipeline results which are created in blocking execution 
mode. For detached executions, we should always return {{State.UNKNOWN}} 
because we don't have a means yet to access the Job state.

This will change once Flink has introduced a client for job control in the API.

> Support unblocking run() in FlinkRunner and cancel() and waitUntilFinish() in 
> FlinkRunnerResult
> -----------------------------------------------------------------------------------------------
>                 Key: BEAM-593
>                 URL: https://issues.apache.org/jira/browse/BEAM-593
>             Project: Beam
>          Issue Type: New Feature
>          Components: runner-flink
>            Reporter: Pei He
> We introduced both functions to PipelineResult.
> Currently, both of them throw UnsupportedOperationException in Flink runner.

This message was sent by Atlassian JIRA

Reply via email to