[
https://issues.apache.org/jira/browse/SPARK-6971?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14522151#comment-14522151
]
Patrick Wendell commented on SPARK-6971:
----------------------------------------
I took care of this for the master and 1.3 maven builds. They now do this:
{code}
rm -rf ./work
git clean -fdx
# Generate random point for Zinc
export ZINC_PORT=$(python -S -c "import random; print
random.randrange(3030,4030)")
build/mvn -DskipTests -P${HADOOP_PROFILE} -Pyarn -Phive clean package
build/mvn test -P${HADOOP_PROFILE} -Pyarn -Phive --fail-at-end
# Clean-up Zinc nailgun process
/usr/sbin/lsof -P |grep $ZINC_PORT | grep LISTEN | awk '{ print $2; }' | xargs
kill
{code}
> Each Jenkins build should use a distinct Zinc port
> --------------------------------------------------
>
> Key: SPARK-6971
> URL: https://issues.apache.org/jira/browse/SPARK-6971
> Project: Spark
> Issue Type: Bug
> Components: Project Infra
> Reporter: Patrick Wendell
> Assignee: Patrick Wendell
> Priority: Critical
> Fix For: 1.4.0
>
>
> Contamination between the builds causes random compiler crashes. Our Spark
> packaging scripts now use unique zinc ports for each build:
> https://github.com/apache/spark/commit/3980ebdf188d77799b55b407b115cdc82f51d532#diff-c849de7463d08ba5f589d5732ef07061R229
> It should be too hard to fix this - we can just set ZINC_PORT differently in
> each build, to a specific number or a random number within a range.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
