[BEAM-1751] Singleton ByteKeyRange with BigtableIO and Dataflow runner

• Type: Bug
• Status: Resolved
• Priority: P2
• Resolution: Fixed
• Affects Version/s: 0.5.0
• Fix Version/s: 2.0.0
• Component/s: io-java-gcp, runner-dataflow
• Assignee: Dan Halperin
• Reporter: peay
• Created: 19/Mar/17 12:50
• Updated: 24/Jul/20 20:19
• Resolved: 24/Mar/17 22:58

Description

I am getting this exception on a smallish table of a couple hundreds of rows from Bigtable, when running on Dataflow with a single worker.

This doesn't occur with the direct runner on my laptop, only when running on Dataflow. Backtrace is from Beam 0.5.

java.lang.IllegalArgumentException: Start [xxxxxxxxxx] must be less than end [xxxxxxxxxx]
	at org.apache.beam.sdk.repackaged.com.google.common.base.Preconditions.checkArgument(Preconditions.java:146)
	at org.apache.beam.sdk.io.range.ByteKeyRange.<init>(ByteKeyRange.java:288)
	at org.apache.beam.sdk.io.range.ByteKeyRange.withEndKey(ByteKeyRange.java:278)
	at org.apache.beam.sdk.io.gcp.bigtable.BigtableIO$BigtableSource.withEndKey(BigtableIO.java:728)
	at org.apache.beam.sdk.io.gcp.bigtable.BigtableIO$BigtableReader.splitAtFraction(BigtableIO.java:1034)
	at org.apache.beam.sdk.io.gcp.bigtable.BigtableIO$BigtableReader.splitAtFraction(BigtableIO.java:953)
	at org.apache.beam.runners.dataflow.DataflowUnboundedReadFromBoundedSource$BoundedToUnboundedSourceAdapter$ResidualSource.getCheckpointMark(DataflowUnboundedReadFromBoundedSource.java:530)
	at org.apache.beam.runners.dataflow.DataflowUnboundedReadFromBoundedSource$BoundedToUnboundedSourceAdapter$Reader.getCheckpointMark(DataflowUnboundedReadFromBoundedSource.java:386)
	at org.apache.beam.runners.dataflow.DataflowUnboundedReadFromBoundedSource$BoundedToUnboundedSourceAdapter$Reader.getCheckpointMark(DataflowUnboundedReadFromBoundedSource.java:283)
	at com.google.cloud.dataflow.worker.runners.worker.StreamingModeExecutionContext.flushState(StreamingModeExecutionContext.java:278)
	at com.google.cloud.dataflow.worker.runners.worker.StreamingDataflowWorker.process(StreamingDataflowWorker.java:778)
	at com.google.cloud.dataflow.worker.runners.worker.StreamingDataflowWorker.access$700(StreamingDataflowWorker.java:105)
	at com.google.cloud.dataflow.worker.runners.worker.StreamingDataflowWorker$9.run(StreamingDataflowWorker.java:858)
	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
	at java.lang.Thread.run(Thread.java:745)

This is in the log right before:

"Proposing to split ByteKeyRangeTracker{range=ByteKeyRange{startKey=[xxxxxxxxxx], endKey=[]}, position=null} at fraction 0.0 (key [xxxxxxxxxx])"   

I have replaced the actual key with xxxxxxxxxx, but it is always the same everywhere. In https://github.com/apache/beam/blob/e68a70e08c9fe00df9ec163d1532da130f69588a/sdks/java/core/src/main/java/org/apache/beam/sdk/io/range/ByteKeyRange.java#L260, the end position is obtained by truncating the fractional part of size * fraction, such that the resulting offset can just be zero if fraction is too small. `ByteKeyRange` does not allow a singleton range, however. Since fraction is zero here, the call to splitAtFraction fails.

Issue Links

• links to: GitHub Pull Request #2324