[
https://issues.apache.org/jira/browse/BEAM-4292?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16519908#comment-16519908
]
Henning Rohde commented on BEAM-4292:
-------------------------------------
I tried to run a simplified version of your example and it fails with:
```
java.lang.RuntimeException: java.lang.ClassCastException:
com.google.cloud.dataflow.worker.BatchModeExecutionContext cannot be cast to
com.google.cloud.dataflow.worker.StreamingModeExecutionContext
at
com.google.cloud.dataflow.worker.BeamFnMapTaskExecutorFactory$4.typedApply(BeamFnMapTaskExecutorFactory.java:442)
at
com.google.cloud.dataflow.worker.BeamFnMapTaskExecutorFactory$4.typedApply(BeamFnMapTaskExecutorFactory.java:413)
at
com.google.cloud.dataflow.worker.graph.Networks$TypeSafeNodeFunction.apply(Networks.java:63)
at
com.google.cloud.dataflow.worker.graph.Networks$TypeSafeNodeFunction.apply(Networks.java:50)
at
com.google.cloud.dataflow.worker.graph.Networks.replaceDirectedNetworkNodes(Networks.java:87)
at
com.google.cloud.dataflow.worker.BeamFnMapTaskExecutorFactory.create(BeamFnMapTaskExecutorFactory.java:175)
at
com.google.cloud.dataflow.worker.BatchDataflowWorker.doWork(BatchDataflowWorker.java:336)
at
com.google.cloud.dataflow.worker.BatchDataflowWorker.getAndPerformWork(BatchDataflowWorker.java:290)
at
com.google.cloud.dataflow.worker.DataflowRunnerHarness.start(DataflowRunnerHarness.java:179)
at
com.google.cloud.dataflow.worker.DataflowRunnerHarness.main(DataflowRunnerHarness.java:107)
Caused by: java.lang.ClassCastException:
com.google.cloud.dataflow.worker.BatchModeExecutionContext cannot be cast to
com.google.cloud.dataflow.worker.StreamingModeExecutionContext
at
com.google.cloud.dataflow.worker.PubsubSink$Factory.create(PubsubSink.java:111)
at
com.google.cloud.dataflow.worker.PubsubSink$Factory.create(PubsubSink.java:84)
at
com.google.cloud.dataflow.worker.SinkRegistry.create(SinkRegistry.java:103)
at
com.google.cloud.dataflow.worker.SinkRegistry.create(SinkRegistry.java:37)
at
com.google.cloud.dataflow.worker.BeamFnMapTaskExecutorFactory.createWriteOperation(BeamFnMapTaskExecutorFactory.java:483)
at
com.google.cloud.dataflow.worker.BeamFnMapTaskExecutorFactory$4.typedApply(BeamFnMapTaskExecutorFactory.java:429)
... 9 more
```
It looks like Dataflow supports pubsub only in streaming mode. I would say that
the simplest path forward is just not write #1 as a pipeline, but just read
using the Beam filesystem and publish to pubsub using the API similarly to how
streaming_wordcap works.
> Add streaming wordcount example
> -------------------------------
>
> Key: BEAM-4292
> URL: https://issues.apache.org/jira/browse/BEAM-4292
> Project: Beam
> Issue Type: Improvement
> Components: sdk-go
> Reporter: Henning Rohde
> Assignee: James Wilson
> Priority: Major
> Attachments: publish_wordline.go
>
>
> It is referenced on the Beam website as part of the Wordcount progression:
> https://beam.apache.org/get-started/wordcount-example/#streamingwordcount-example
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
