[BEAM-112] Using non-IntervalWindows seems to fail

• Type: Bug
• Status: Resolved
• Priority: P3
• Resolution: Duplicate
• Fix Version/s: Not applicable
• Component/s: runner-spark
• Assignee: Amit Sela
• Reporter: Ben Chambers
• Created: 14/Mar/16 18:37
• Updated: 24/Jul/20 20:56
• Resolved: 14/Apr/16 18:59

Description

See here for more details: http://stackoverflow.com/questions/35993777/globalwindow-cannot-be-cast-to-intervalwindow

The linked stack trace indicates this is using the Spark Runner:

Exception

java.lang.ClassCastException: com.google.cloud.dataflow.sdk.transforms.windowing.GlobalWindow cannot be cast to com.google.cloud.dataflow.sdk.transforms.windowing.IntervalWindow
    at com.google.cloud.dataflow.sdk.transforms.windowing.IntervalWindow$IntervalWindowCoder.encode(IntervalWindow.java:171)
    at com.google.cloud.dataflow.sdk.coders.IterableLikeCoder.encode(IterableLikeCoder.java:113)
    at com.google.cloud.dataflow.sdk.coders.IterableLikeCoder.encode(IterableLikeCoder.java:59)
    at com.google.cloud.dataflow.sdk.util.WindowedValue$FullWindowedValueCoder.encode(WindowedValue.java:599)
    at com.google.cloud.dataflow.sdk.util.WindowedValue$FullWindowedValueCoder.encode(WindowedValue.java:540)
    at com.cloudera.dataflow.spark.CoderHelpers.toByteArray(CoderHelpers.java:48)
    at com.cloudera.dataflow.spark.CoderHelpers$3.call(CoderHelpers.java:134)
    at com.cloudera.dataflow.spark.CoderHelpers$3.call(CoderHelpers.java:131)
    at org.apache.spark.api.java.JavaPairRDD$$anonfun$pairFunToScalaFun$1.apply(JavaPairRDD.scala:1018)
    at org.apache.spark.api.java.JavaPairRDD$$anonfun$pairFunToScalaFun$1.apply(JavaPairRDD.scala:1018)
    at scala.collection.Iterator$$anon$11.next(Iterator.scala:328)
    at org.apache.spark.shuffle.sort.BypassMergeSortShuffleWriter.write(BypassMergeSortShuffleWriter.java:149)
    at org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:73)
    at org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:41)
    at org.apache.spark.scheduler.Task.run(Task.scala:89)
    at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:214)
    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)

It seems likely that at some point the Spark runner is assuming that all windows are IntervalWindows, which may not be true. Specifically the GlobalWindow+Triggers is valid, as is any custom implementation of BoundedWindow.

Issue Links

• is related to: BEAM-189 The Spark runner uses valueInEmptyWindow which causes values to be dropped