[HIVE-8300] Missing guava lib causes IllegalStateException when deserializing a task [Spark Branch]

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.1.0
• Component/s: Spark

Environment

Spark-1.2.0-SNAPSHOT

• Assignee: Unassigned
• Reporter: Rui Li
• Created: 30/Sep/14 00:16
• Updated: 29/May/15 02:27
• Resolved: 30/Sep/14 14:09

Description

In spark-1.2, we have guava shaded in spark-assembly. And we only ship hive-exec to spark cluster. So spark executor won't have (original) guava in its class path.
This can cause some problem when TaskRunner deserializes a task, and throws something like this:

org.apache.spark.SparkException: Job aborted due to stage failure: Task 0 in stage 0.0 failed 4 times, most recent failure: Lost task 0.3 in stage 0.0 (TID 3, node13-1): java.lang.IllegalStateException: unread block data
        java.io.ObjectInputStream$BlockDataInputStream.setBlockDataMode(ObjectInputStream.java:2421)
        java.io.ObjectInputStream.readObject0(ObjectInputStream.java:1382)
        java.io.ObjectInputStream.defaultReadFields(ObjectInputStream.java:1990)
        java.io.ObjectInputStream.readSerialData(ObjectInputStream.java:1915)
        java.io.ObjectInputStream.readOrdinaryObject(ObjectInputStream.java:1798)
        java.io.ObjectInputStream.readObject0(ObjectInputStream.java:1350)
        java.io.ObjectInputStream.readObject(ObjectInputStream.java:370)
        org.apache.spark.serializer.JavaDeserializationStream.readObject(JavaSerializer.scala:62)
        org.apache.spark.serializer.JavaSerializerInstance.deserialize(JavaSerializer.scala:87)
        org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:164)
        java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
        java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
        java.lang.Thread.run(Thread.java:744)

We may have to verify this issue and ship guava to spark cluster.

Issue Links

• is part of: HIVE-7292 Hive on Spark