Skip to content

Conversation

@calvinjia
Copy link
Contributor

This is a reopening of #9204 which failed hadoop1 sbt tests.

With the original PR, a classpath issue would occur due to the MIMA plugin pulling in hadoop-2.2 dependencies regardless of the hadoop version when building the oldDeps project. These affect the hadoop1 sbt build because they are placed in lib_managed and Tachyon 0.8.0's default hadoop version is 2.2.

@calvinjia
Copy link
Contributor Author

@yhuai
Could you take a look? Thanks!

@yhuai
Copy link
Contributor

yhuai commented Nov 1, 2015

cc @pwendell @srowen This is the new pr that upgrades Tachyon. I reverted the original one, which caused hadoop 1 test failures.

@yhuai
Copy link
Contributor

yhuai commented Nov 1, 2015

btw, the profiles that we used in hadoop 1 tests are -Phadoop-1 -Dhadoop.version=1.2.1 -Pkinesis-asl -Phive-thriftserver -Phive.

@SparkQA
Copy link

SparkQA commented Nov 1, 2015

Test build #44756 has finished for PR 9395 at commit 6802ecd.

  • This patch passes all tests.
  • This patch merges cleanly.
  • This patch adds the following public classes (experimental):\n * logInfo(s\"Hive class not found $e\")\n * logDebug(\"Hive class not found\", e)\n

@calvinjia
Copy link
Contributor Author

@yhuai
Thanks for the the response! The last PR would pass with those parameters, but and would only fail if the entire dev/runtests cycle was ran (specifically if MIMA ran before). I've tested locally that the issues do not occur when running dev/runtests with this change.

@pwendell
Copy link
Contributor

pwendell commented Nov 2, 2015

@calvinjia can you add "[test-hadoop1.0]" to the title of this PR and then retest it? That will run the tests with hadoop 1. See more info here:

https://cwiki.apache.org/confluence/display/SPARK/Useful+Developer+Tools

@calvinjia calvinjia changed the title [SPARK-11236][CORE] Update Tachyon dependency 0.7.1 -> 0.8.1 [test-hadoop1.0][SPARK-11236][CORE] Update Tachyon dependency 0.7.1 -> 0.8.1 Nov 2, 2015
@calvinjia
Copy link
Contributor Author

@pwendell could you trigger a build for me? Or should I just make an empty commit?

@pwendell
Copy link
Contributor

pwendell commented Nov 2, 2015

Jenkins, test this please.

@calvinjia
Copy link
Contributor Author

Thanks!

@SparkQA
Copy link

SparkQA commented Nov 2, 2015

Test build #44788 has finished for PR 9395 at commit 6802ecd.

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

@calvinjia
Copy link
Contributor Author

@pwendell

It seems like the previous issue is gone. The only error (which seems to be triggered by an oom / jvm crash?) that occurs is:

Exception in thread "Thread-58" Exception in thread "Thread-62" java.net.SocketException: Connection reset
    at java.net.SocketInputStream.read(SocketInputStream.java:196)
    at java.net.SocketInputStream.read(SocketInputStream.java:122)
    at java.net.SocketInputStream.read(SocketInputStream.java:210)
    at java.io.ObjectInputStream$PeekInputStream.peek(ObjectInputStream.java:2293)
    at java.io.ObjectInputStream$BlockDataInputStream.peek(ObjectInputStream.java:2586)
    at java.io.ObjectInputStream$BlockDataInputStream.peekByte(ObjectInputStream.java:2596)
    at java.io.ObjectInputStream.readObject0(ObjectInputStream.java:1318)
    at java.io.ObjectInputStream.readObject(ObjectInputStream.java:370)
    at sbt.React.react(ForkTests.scala:114)
    at sbt.ForkTests$$anonfun$mainTestTask$1$Acceptor$2$.run(ForkTests.scala:74)
    at java.lang.Thread.run(Thread.java:745)
java.io.EOFException
    at java.io.ObjectInputStream$BlockDataInputStream.peekByte(ObjectInputStream.java:2598)
    at java.io.ObjectInputStream.readObject0(ObjectInputStream.java:1318)
    at java.io.ObjectInputStream.readObject(ObjectInputStream.java:370)
    at org.scalatest.tools.Framework$ScalaTestRunner$Skeleton$1$React.react(Framework.scala:945)
    at org.scalatest.tools.Framework$ScalaTestRunner$Skeleton$1.run(Framework.scala:934)
    at java.lang.Thread.run(Thread.java:745)

It doesn't seem like its related to this change.

@yhuai
Copy link
Contributor

yhuai commented Nov 2, 2015

test this please

@calvinjia
Copy link
Contributor Author

@yhuai Thanks for the retest. I'm not sure if this will go away by re-running or if there is something up with Jenkins/Spark master branch. It seems like the current Spark-Master-SBT build is not happy on hadoop1.0 for the same reason. (See: https://amplab.cs.berkeley.edu/jenkins/job/Spark-Master-SBT/3911/AMPLAB_JENKINS_BUILD_PROFILE=hadoop1.0,label=spark-test/consoleFull)

@SparkQA
Copy link

SparkQA commented Nov 2, 2015

Test build #44790 has finished for PR 9395 at commit 6802ecd.

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

@pwendell
Copy link
Contributor

pwendell commented Nov 2, 2015

Hm - can you try also adding [test-maven] - might be better to test with
maven.

On Sun, Nov 1, 2015 at 9:40 PM, Calvin Jia [email protected] wrote:

@yhuai https://github.com/yhuai Thanks for the retest. I'm not sure if
this will go away by re-running or if there is something up with
Jenkins/Spark master branch. It seems like the current Spark-Master-SBT
build is not happy on hadoop1.0 for the same reason. (See:
https://amplab.cs.berkeley.edu/jenkins/job/Spark-Master-SBT/3911/AMPLAB_JENKINS_BUILD_PROFILE=hadoop1.0,label=spark-test/consoleFull
)


Reply to this email directly or view it on GitHub
#9395 (comment).

@calvinjia calvinjia changed the title [test-hadoop1.0][SPARK-11236][CORE] Update Tachyon dependency 0.7.1 -> 0.8.1 [test-maven][test-hadoop1.0][SPARK-11236][CORE] Update Tachyon dependency 0.7.1 -> 0.8.1 Nov 2, 2015
@calvinjia
Copy link
Contributor Author

@pwendell Sure, I've updated it.

@pwendell
Copy link
Contributor

pwendell commented Nov 2, 2015

Jenkins, test this please.

@SparkQA
Copy link

SparkQA commented Nov 2, 2015

Test build #44798 has finished for PR 9395 at commit 6802ecd.

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

@calvinjia
Copy link
Contributor Author

@pwendell
It seems like the maven tests failed with:

java.lang.IllegalArgumentException: Logging directory specified is not a directory: file:/home/jenkins/workspace/SparkPullRequestBuilder/core/target/tmp/spark-ab60a5c8-38dd-4340-a4e9-109cf8dcf18e/

This doesn't seem like something that would be introduced by this change?

@calvinjia
Copy link
Contributor Author

@pwendell

It seems like the Spark-Master-Maven build running for hadoop1.0 is also failing with the same problem (see: https://amplab.cs.berkeley.edu/jenkins/job/Spark-Master-Maven-pre-YARN/HADOOP_VERSION=1.2.1,label=spark-test/4631/).

@calvinjia
Copy link
Contributor Author

@pwendell @yhuai
Is the only blocker for this PR the stability of hadoop1.0 tests in Spark master? Thanks!

@pwendell
Copy link
Contributor

pwendell commented Nov 3, 2015

Hey @calvinjia seems okay to merge since this is just triggering other failures. Can you explain more though how this works around the mima issue - the patch seems the same as #9204. Did something change between Tachyon 0.8.0 and 0.8.1?

@calvinjia
Copy link
Contributor Author

@pwendell
We changed the default hadoop version from 2.2. MIMA pulls in hadoop2.2 jars into the lib_managed directory when it builds oldDeps. In spark.utils.ShutdownHookManager, we do some reflection to see if we are using Hadoop 2.x, and it will get a false positive leading to the failures. (See: https://github.com/apache/spark/blob/master/core/src/main/scala/org/apache/spark/util/ShutdownHookManager.scala#L221)

@pwendell
Copy link
Contributor

pwendell commented Nov 3, 2015

Got it- thanks I can merge it then.

@asfgit asfgit closed this in 476f434 Nov 3, 2015
@calvinjia
Copy link
Contributor Author

@pwendell @yhuai
Thanks for the pointers and reviewing this PR.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants