[
https://issues.apache.org/jira/browse/BEAM-6349?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16736648#comment-16736648
]
Boyuan Zhang commented on BEAM-6349:
------------------------------------
Hey Lukasz,
The root cause of this problem is because Craig's PR changed the sdk and
dataflow runner harness at the same time but your test is running against the
pre-build dataflow runner image(which does not include Craig's changes) with
the latest sdk(which includes Craig's sdk part changes). This results in a
mismatch issue. For a temporary workaround, you can launch your job with
customized dataflow worker jar.
> Exceptions (IllegalArgumentException or NoClassDefFoundError) when running
> tests on Dataflow runner
> ---------------------------------------------------------------------------------------------------
>
> Key: BEAM-6349
> URL: https://issues.apache.org/jira/browse/BEAM-6349
> Project: Beam
> Issue Type: Improvement
> Components: testing
> Reporter: Lukasz Gajowy
> Assignee: Craig Chambers
> Priority: Major
> Time Spent: 50m
> Remaining Estimate: 0h
>
> Running GroupByKeyLoadTest results in the following error on Dataflow runner:
>
> {code:java}
> java.lang.ExceptionInInitializerError
> at
> org.apache.beam.runners.dataflow.worker.IntrinsicMapTaskExecutorFactory$2.typedApply(IntrinsicMapTaskExecutorFactory.java:344)
> at
> org.apache.beam.runners.dataflow.worker.IntrinsicMapTaskExecutorFactory$2.typedApply(IntrinsicMapTaskExecutorFactory.java:338)
> at
> org.apache.beam.runners.dataflow.worker.graph.Networks$TypeSafeNodeFunction.apply(Networks.java:63)
> at
> org.apache.beam.runners.dataflow.worker.graph.Networks$TypeSafeNodeFunction.apply(Networks.java:50)
> at
> org.apache.beam.runners.dataflow.worker.graph.Networks.replaceDirectedNetworkNodes(Networks.java:87)
> at
> org.apache.beam.runners.dataflow.worker.IntrinsicMapTaskExecutorFactory.create(IntrinsicMapTaskExecutorFactory.java:120)
> at
> org.apache.beam.runners.dataflow.worker.BatchDataflowWorker.doWork(BatchDataflowWorker.java:337)
> at
> org.apache.beam.runners.dataflow.worker.BatchDataflowWorker.getAndPerformWork(BatchDataflowWorker.java:291)
> at
> org.apache.beam.runners.dataflow.worker.DataflowBatchWorkerHarness$WorkerThread.doWork(DataflowBatchWorkerHarness.java:135)
> at
> org.apache.beam.runners.dataflow.worker.DataflowBatchWorkerHarness$WorkerThread.call(DataflowBatchWorkerHarness.java:115)
> at
> org.apache.beam.runners.dataflow.worker.DataflowBatchWorkerHarness$WorkerThread.call(DataflowBatchWorkerHarness.java:102)
> at java.util.concurrent.FutureTask.run(FutureTask.java:266)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: java.lang.IllegalArgumentException: Multiple entries with same
> key:
> kind:varint=org.apache.beam.runners.dataflow.util.CloudObjectTranslators$8@39b69c48
> and
> kind:varint=org.apache.beam.runners.dataflow.worker.RunnerHarnessCoderCloudObjectTranslatorRegistrar$1@7966f294
> at
> org.apache.beam.repackaged.beam_runners_google_cloud_dataflow_java.com.google.common.collect.ImmutableMap.checkNoConflict(ImmutableMap.java:136)
> at
> org.apache.beam.repackaged.beam_runners_google_cloud_dataflow_java.com.google.common.collect.RegularImmutableMap.checkNoConflictInKeyBucket(RegularImmutableMap.java:100)
> at
> org.apache.beam.repackaged.beam_runners_google_cloud_dataflow_java.com.google.common.collect.RegularImmutableMap.fromEntryArray(RegularImmutableMap.java:86)
> at
> org.apache.beam.repackaged.beam_runners_google_cloud_dataflow_java.com.google.common.collect.ImmutableMap$Builder.build(ImmutableMap.java:300)
> at
> org.apache.beam.runners.dataflow.util.CloudObjects.populateCloudObjectTranslators(CloudObjects.java:60)
> at
> org.apache.beam.runners.dataflow.util.CloudObjects.<clinit>(CloudObjects.java:39)
> ... 15 more
> {code}
>
> Example command to run the tests (FWIW, it also runs the "clean" task
> although I don't know if it's necessary):
> {code:java}
> ./gradlew clean :beam-sdks-java-load-tests:run --info
> -PloadTest.mainClass=org.apache.beam.sdk.loadtests.GroupByKeyLoadTest
> -Prunner=:beam-runners-google-cloud-dataflow-java
> '-PloadTest.args=--sourceOptions={"numRecords":1000,"splitPointFrequencyRecords":1,"keySizeBytes":1,"valueSizeBytes":9,"numHotKeys":0,"hotKeyFraction":0,"seed":123456,"bundleSizeDistribution":{"type":"const","const":42},"forceNumInitialBundles":100,"progressShape":"LINEAR","initializeDelayDistribution":{"type":"const","const":42}}
>
> --stepOptions={"outputRecordsPerInputRecord":1,"preservesInputKeyDistribution":true,"perBundleDelay":1,"perBundleDelayType":"MIXED","cpuUtilizationInMixedDelay":0.5}
> --fanout=1 --iterations=1 --runner=DataflowRunner'{code}
>
> After reverting commit bac909b8e237ef8a2ab7e17ac986e5cc90143e5b ([PR:
> 7351|https://github.com/apache/beam/pull/7351]) I can no longer reproduce
> this issue.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
