[SPARK-4056] Upgrade snappy-java to 1.1.1.5 #2911

JoshRosen · 2014-10-23T18:47:52Z

This upgrades snappy-java to 1.1.1.5, which improves error messages when attempting to deserialize empty inputs using SnappyInputStream (see xerial/snappy-java#89).

SparkQA · 2014-10-23T18:54:46Z

QA tests have started for PR 2911 at commit cc953d6.

This patch merges cleanly.

rxin · 2014-10-23T19:00:34Z

LGTM.

SparkQA · 2014-10-23T20:54:46Z

Tests timed out for PR 2911 at commit cc953d6 after a configured wait of 120m.

AmplabJenkins · 2014-10-23T20:54:49Z

Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/22077/
Test FAILed.

JoshRosen · 2014-10-23T21:00:59Z

Jenkins, retest this please.

SparkQA · 2014-10-23T21:04:59Z

QA tests have started for PR 2911 at commit cc953d6.

This patch merges cleanly.

JoshRosen · 2014-10-23T22:44:18Z

It looks like these test failures might be due missing classes in the snappy-java 1.1.1.4 JAR: xerial/snappy-java#90

SparkQA · 2014-10-23T23:05:00Z

Tests timed out for PR 2911 at commit cc953d6 after a configured wait of 120m.

AmplabJenkins · 2014-10-23T23:05:02Z

Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/22084/
Test FAILed.

xerial · 2014-10-24T00:53:55Z

Please use snappy-java-1.1.1.5, which fixes the broken build.

SparkQA · 2014-10-24T01:00:19Z

QA tests have started for PR 2911 at commit adec96c.

This patch merges cleanly.

JoshRosen · 2014-10-24T01:05:55Z

@xerial Thanks for fixing that so quickly!

SparkQA · 2014-10-24T03:00:19Z

Tests timed out for PR 2911 at commit adec96c after a configured wait of 120m.

AmplabJenkins · 2014-10-24T03:00:22Z

Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/22100/
Test FAILed.

JoshRosen · 2014-10-24T03:10:45Z

Jenkins, retest this please.

SparkQA · 2014-10-24T03:14:46Z

QA tests have started for PR 2911 at commit adec96c.

This patch merges cleanly.

SparkQA · 2014-10-24T04:09:10Z

QA tests have finished for PR 2911 at commit adec96c.

This patch fails Spark unit tests.
This patch merges cleanly.
This patch adds no public classes.

AmplabJenkins · 2014-10-24T04:09:14Z

Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/22106/
Test FAILed.

xerial · 2014-10-24T04:28:35Z

@JoshRosen If you have the stack trace of this error, please let me know. I would like to check it.

SparkQA · 2014-10-24T05:00:09Z

QA tests have started for PR 2911 at commit adec96c.

This patch merges cleanly.

JoshRosen · 2014-10-24T05:01:02Z

@xerial Here's a link to the exception from that most recent test failure:

https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/22106/testReport/junit/org.apache.spark.util.collection/ExternalAppendOnlyMapSuite/spilling_with_compression/

In case that link breaks, here's the driver stacktrace:

sbt.ForkMain$ForkError: Test failed with compression using codec org.apache.spark.io.SnappyCompressionCodec:

Job aborted due to stage failure: Task 0 in stage 4.0 failed 4 times, most recent failure: Lost task 0.3 in stage 4.0 (TID 14, localhost): java.io.IOException: unexpected exception type
        java.io.ObjectStreamClass.throwMiscException(ObjectStreamClass.java:1538)
        java.io.ObjectStreamClass.invokeReadObject(ObjectStreamClass.java:1025)
        java.io.ObjectInputStream.readSerialData(ObjectInputStream.java:1893)
        java.io.ObjectInputStream.readOrdinaryObject(ObjectInputStream.java:1798)
        java.io.ObjectInputStream.readObject0(ObjectInputStream.java:1350)
        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$TaskRunner.run(Executor.scala:163)
        java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
        java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
        java.lang.Thread.run(Thread.java:745)
Driver stacktrace:
    at org.apache.spark.scheduler.DAGScheduler.org$apache$spark$scheduler$DAGScheduler$$failJobAndIndependentStages(DAGScheduler.scala:1191)
    at org.apache.spark.scheduler.DAGScheduler$$anonfun$abortStage$1.apply(DAGScheduler.scala:1180)
    at org.apache.spark.scheduler.DAGScheduler$$anonfun$abortStage$1.apply(DAGScheduler.scala:1179)
    at scala.collection.mutable.ResizableArray$class.foreach(ResizableArray.scala:59)
    at scala.collection.mutable.ArrayBuffer.foreach(ArrayBuffer.scala:47)
    at org.apache.spark.scheduler.DAGScheduler.abortStage(DAGScheduler.scala:1179)
    at org.apache.spark.scheduler.DAGScheduler$$anonfun$handleTaskSetFailed$1.apply(DAGScheduler.scala:694)
    at org.apache.spark.scheduler.DAGScheduler$$anonfun$handleTaskSetFailed$1.apply(DAGScheduler.scala:694)
    at scala.Option.foreach(Option.scala:236)
    at org.apache.spark.scheduler.DAGScheduler.handleTaskSetFailed(DAGScheduler.scala:694)
    at org.apache.spark.scheduler.DAGSchedulerEventProcessActor$$anonfun$receive$2.applyOrElse(DAGScheduler.scala:1397)
    at akka.actor.Actor$class.aroundReceive(Actor.scala:465)
    at org.apache.spark.scheduler.DAGSchedulerEventProcessActor.aroundReceive(DAGScheduler.scala:1352)
    at akka.actor.ActorCell.receiveMessage(ActorCell.scala:516)
    at akka.actor.ActorCell.invoke(ActorCell.scala:487)
    at akka.dispatch.Mailbox.processMailbox(Mailbox.scala:238)
    at akka.dispatch.Mailbox.run(Mailbox.scala:220)
    at akka.dispatch.ForkJoinExecutorConfigurator$AkkaForkJoinTask.exec(AbstractDispatcher.scala:393)
    at scala.concurrent.forkjoin.ForkJoinTask.doExec(ForkJoinTask.java:260)
    at scala.concurrent.forkjoin.ForkJoinPool$WorkQueue.runTask(ForkJoinPool.java:1339)
    at scala.concurrent.forkjoin.ForkJoinPool.runWorker(ForkJoinPool.java:1979)
    at scala.concurrent.forkjoin.ForkJoinWorkerThread.run(ForkJoinWorkerThread.java:107)

This isn't the full stacktrace of the actual error, which took place in an executor. Until we merge #2845, I don't think that I'll have an easy way to grab the full executor logs from Jenkins.

I wasn't able to reproduce this failure locally. I'm going to try our new experimental "Deflake build" Jenkins button, which reruns only the failing tests, in order to see if I can reproduce this. If so, I'll SSH in and grab the full logs.

JoshRosen · 2014-10-24T05:02:30Z

Actually, I don't think that "deflake build" plugin will necessarily work as expected given all of the customization in our build; I guess it was added for another project that shares the Jenkins server with us.

SparkQA · 2014-10-24T06:12:18Z

QA tests have finished for PR 2911 at commit adec96c.

This patch passes all tests.
This patch merges cleanly.
This patch adds no public classes.

JoshRosen · 2014-10-24T06:33:17Z

Hmm, looks like that might have been a transient failure. Just to be sure, though, I'm going to run this one more time to make sure that it still passes, then merge it (since I don't think that any issues we'll observe will be caused by this small snappy-java version bump).

Jenkins, retest this please.

JoshRosen · 2014-10-24T16:21:12Z

Jenkins, retest this please.

SparkQA · 2014-10-24T16:24:54Z

Test build #22143 has started for PR 2911 at commit adec96c.

This patch merges cleanly.

AmplabJenkins · 2014-10-24T16:50:19Z

Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/22143/
Test FAILed.

pwendell · 2014-10-24T16:51:20Z

Jenkins, retest this please (sorry I had to abort this to clean the workspace).

SparkQA · 2014-10-24T16:55:01Z

Test build #22144 has started for PR 2911 at commit adec96c.

This patch merges cleanly.

SparkQA · 2014-10-24T17:49:13Z

Test build #22144 has finished for PR 2911 at commit adec96c.

This patch fails Spark unit tests.
This patch merges cleanly.
This patch adds no public classes.

AmplabJenkins · 2014-10-24T17:49:17Z

Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/22144/
Test FAILed.

JoshRosen · 2014-10-24T18:35:50Z

The "IOException: unexpected exception type" error here is actually masking the real error; this is due to us throwing something other than IOException from read/writeExternal; I've opened https://issues.apache.org/jira/browse/SPARK-4080 to fix this (working on a patch now).

I'm pretty sure that this isn't caused by Snappy, but instead is an instance of some longer-standing non-deterministic serialization issue in our code.

JoshRosen · 2014-10-24T22:23:14Z

Now that I've fixed the IOException issue via #2932, let's retest this to see if I get a more informative error message. I'm almost positive that this isn't a Snappy issue; I'm just curious to see what happens.

Jenkins, retest this please.

JoshRosen · 2014-10-24T23:04:34Z

Jenkins, retest this please.

SparkQA · 2014-10-24T23:07:32Z

Test build #22178 has started for PR 2911 at commit adec96c.

This patch merges cleanly.

SparkQA · 2014-10-25T00:00:36Z

Test build #22178 has finished for PR 2911 at commit adec96c.

This patch fails Spark unit tests.
This patch merges cleanly.
This patch adds no public classes.

AmplabJenkins · 2014-10-25T00:00:40Z

Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/22178/
Test FAILed.

JoshRosen · 2014-10-25T00:20:47Z

Those failures are related to known issues from the Hive 13 PR, not this, so I'm going to merge this across all of our backport branches in order to improve our error reporting.

This upgrades snappy-java to 1.1.1.5, which improves error messages when attempting to deserialize empty inputs using SnappyInputStream (see xerial/snappy-java#89). Author: Josh Rosen <[email protected]> Author: Josh Rosen <[email protected]> Closes #2911 from JoshRosen/upgrade-snappy-java and squashes the following commits: adec96c [Josh Rosen] Use snappy-java 1.1.1.5 cc953d6 [Josh Rosen] [SPARK-4056] Upgrade snappy-java to 1.1.1.4 (cherry picked from commit 898b22a) Signed-off-by: Josh Rosen <[email protected]> Conflicts: pom.xml

JoshRosen · 2014-10-26T00:04:03Z

Since merging this PR, I've started noticing some OOM failures in a test that uses SnappyOutputStream:

- partial aggregation without spill
*** RUN ABORTED ***
  java.lang.OutOfMemoryError: Java heap space
  at org.xerial.snappy.buffer.CachedBufferAllocator.allocate(CachedBufferAllocator.java:48)
  at org.xerial.snappy.SnappyOutputStream.<init>(SnappyOutputStream.java:96)
  at org.xerial.snappy.SnappyOutputStream.<init>(SnappyOutputStream.java:83)
  at org.apache.spark.io.SnappyCompressionCodec.compressedOutputStream(CompressionCodec.scala:125)
  at org.apache.spark.storage.BlockManager.wrapForCompression(BlockManager.scala:1084)
  at org.apache.spark.storage.BlockManager$$anonfun$7.apply(BlockManager.scala:579)
  at org.apache.spark.storage.BlockManager$$anonfun$7.apply(BlockManager.scala:579)
  at org.apache.spark.storage.DiskBlockObjectWriter.open(BlockObjectWriter.scala:126)
  at org.apache.spark.storage.DiskBlockObjectWriter.write(BlockObjectWriter.scala:192)
  at org.apache.spark.util.collection.ExternalSorter.spillToMergeableFile(ExternalSorter.scala:300)

@xerial do you think there could be a memory leak in the new CachedBufferAllocator? xerial/snappy-java@1.1.1.3...develop

JoshRosen · 2014-10-26T00:08:41Z

I've reverted this in commit 898b22a until we figure out what's going on.

We previously tried up upgrade to 1.1.1.5 in apache#2911 but reverted that patch after discovering a memory leak in snappy-java. This should leak have been fixed in 1.1.1.6, though.

This upgrades snappy-java to 1.1.1.6, which includes a patch that improves error messages when attempting to deserialize empty inputs using SnappyInputStream (see xerial/snappy-java#89). We previously tried up upgrade to 1.1.1.5 in #2911 but reverted that patch after discovering a memory leak in snappy-java. This should leak have been fixed in 1.1.1.6, though (see xerial/snappy-java#92). Author: Josh Rosen <[email protected]> Closes #3287 from JoshRosen/SPARK-4419 and squashes the following commits: 5d6f4cc [Josh Rosen] [SPARK-4419] Upgrade snappy-java to 1.1.1.6.

This upgrades snappy-java to 1.1.1.6, which includes a patch that improves error messages when attempting to deserialize empty inputs using SnappyInputStream (see xerial/snappy-java#89). We previously tried up upgrade to 1.1.1.5 in #2911 but reverted that patch after discovering a memory leak in snappy-java. This should leak have been fixed in 1.1.1.6, though (see xerial/snappy-java#92). Author: Josh Rosen <[email protected]> Closes #3287 from JoshRosen/SPARK-4419 and squashes the following commits: 5d6f4cc [Josh Rosen] [SPARK-4419] Upgrade snappy-java to 1.1.1.6. (cherry picked from commit 7d8e152) Signed-off-by: Reynold Xin <[email protected]>

[SPARK-4056] Upgrade snappy-java to 1.1.1.4

cc953d6

Use snappy-java 1.1.1.5

adec96c

JoshRosen changed the title ~~[SPARK-4056] Upgrade snappy-java to 1.1.1.4~~ [SPARK-4056] Upgrade snappy-java to 1.1.1.5 Oct 24, 2014

asfgit closed this in 898b22a Oct 25, 2014

JoshRosen mentioned this pull request Nov 15, 2014

[SPARK-4419] Upgrade snappy-java to 1.1.1.6 #3287

Closed

[SPARK-4056] Upgrade snappy-java to 1.1.1.5 #2911

[SPARK-4056] Upgrade snappy-java to 1.1.1.5 #2911

Conversation

JoshRosen commented Oct 23, 2014

SparkQA commented Oct 23, 2014

rxin commented Oct 23, 2014

SparkQA commented Oct 23, 2014

AmplabJenkins commented Oct 23, 2014

JoshRosen commented Oct 23, 2014

SparkQA commented Oct 23, 2014

JoshRosen commented Oct 23, 2014

SparkQA commented Oct 23, 2014

AmplabJenkins commented Oct 23, 2014

xerial commented Oct 24, 2014

SparkQA commented Oct 24, 2014

JoshRosen commented Oct 24, 2014

SparkQA commented Oct 24, 2014

AmplabJenkins commented Oct 24, 2014

JoshRosen commented Oct 24, 2014

SparkQA commented Oct 24, 2014

SparkQA commented Oct 24, 2014

AmplabJenkins commented Oct 24, 2014

xerial commented Oct 24, 2014

SparkQA commented Oct 24, 2014

JoshRosen commented Oct 24, 2014

JoshRosen commented Oct 24, 2014

SparkQA commented Oct 24, 2014

JoshRosen commented Oct 24, 2014

JoshRosen commented Oct 24, 2014

SparkQA commented Oct 24, 2014

AmplabJenkins commented Oct 24, 2014

pwendell commented Oct 24, 2014

SparkQA commented Oct 24, 2014

SparkQA commented Oct 24, 2014

AmplabJenkins commented Oct 24, 2014

JoshRosen commented Oct 24, 2014

JoshRosen commented Oct 24, 2014

JoshRosen commented Oct 24, 2014

SparkQA commented Oct 24, 2014

SparkQA commented Oct 25, 2014

AmplabJenkins commented Oct 25, 2014

JoshRosen commented Oct 25, 2014

JoshRosen commented Oct 26, 2014

JoshRosen commented Oct 26, 2014