You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hive.apache.org by "Sergey Shelukhin (JIRA)" <ji...@apache.org> on 2015/09/25 19:58:04 UTC
[jira] [Comment Edited] (HIVE-11887) spark tests break the build on
a shared machine
[ https://issues.apache.org/jira/browse/HIVE-11887?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14908414#comment-14908414 ]
Sergey Shelukhin edited comment on HIVE-11887 at 9/25/15 5:57 PM:
------------------------------------------------------------------
Sorry, java, not jar
It's part of itests build:
{noformat}
$ grep -B 30 -A 2 UDFExa itests/pom.xml
<executions>
<execution>
<id>download-spark</id>
<phase>generate-sources</phase>
<goals>
<goal>run</goal>
</goals>
<configuration>
<target>
<echo file="target/download.sh">
set -x
/bin/pwd
BASE_DIR=./target
HIVE_ROOT=$BASE_DIR/../../../
DOWNLOAD_DIR=./../thirdparty
download() {
url=$1;
finalName=$2
tarName=$(basename $url)
rm -rf $BASE_DIR/$finalName
if [[ ! -f $DOWNLOAD_DIR/$tarName ]]
then
curl -Sso $DOWNLOAD_DIR/$tarName $url
fi
tar -zxf $DOWNLOAD_DIR/$tarName -C $BASE_DIR
mv $BASE_DIR/spark-${spark.version}-bin-hadoop2-without-hive $BASE_DIR/$finalName
}
mkdir -p $DOWNLOAD_DIR
download "http://d3jw87u4immizc.cloudfront.net/spark-tarball/spark-${spark.version}-bin-hadoop2-without-hive.tgz" "spark"
cp -f $HIVE_ROOT/data/conf/spark/log4j2.xml $BASE_DIR/spark/conf/
sed '/package /d' ${basedir}/${hive.path.to.root}/contrib/src/java/org/apache/hadoop/hive/contrib/udf/example/UDFExampleAdd.java > /tmp/UDFExampleAdd.java
javac -cp ${settings.localRepository}/org/apache/hive/hive-exec/${project.version}/hive-exec-${project.version}.jar /tmp/UDFExampleAdd.java -d /tmp
jar -cf /tmp/udfexampleadd-1.0.jar -C /tmp UDFExampleAdd.class
</echo>
</target>
{noformat}
See the last two lines
was (Author: sershe):
It's part of itests build:
{noformat}
$ grep -B 30 -A 2 UDFExa itests/pom.xml
<executions>
<execution>
<id>download-spark</id>
<phase>generate-sources</phase>
<goals>
<goal>run</goal>
</goals>
<configuration>
<target>
<echo file="target/download.sh">
set -x
/bin/pwd
BASE_DIR=./target
HIVE_ROOT=$BASE_DIR/../../../
DOWNLOAD_DIR=./../thirdparty
download() {
url=$1;
finalName=$2
tarName=$(basename $url)
rm -rf $BASE_DIR/$finalName
if [[ ! -f $DOWNLOAD_DIR/$tarName ]]
then
curl -Sso $DOWNLOAD_DIR/$tarName $url
fi
tar -zxf $DOWNLOAD_DIR/$tarName -C $BASE_DIR
mv $BASE_DIR/spark-${spark.version}-bin-hadoop2-without-hive $BASE_DIR/$finalName
}
mkdir -p $DOWNLOAD_DIR
download "http://d3jw87u4immizc.cloudfront.net/spark-tarball/spark-${spark.version}-bin-hadoop2-without-hive.tgz" "spark"
cp -f $HIVE_ROOT/data/conf/spark/log4j2.xml $BASE_DIR/spark/conf/
sed '/package /d' ${basedir}/${hive.path.to.root}/contrib/src/java/org/apache/hadoop/hive/contrib/udf/example/UDFExampleAdd.java > /tmp/UDFExampleAdd.java
javac -cp ${settings.localRepository}/org/apache/hive/hive-exec/${project.version}/hive-exec-${project.version}.jar /tmp/UDFExampleAdd.java -d /tmp
jar -cf /tmp/udfexampleadd-1.0.jar -C /tmp UDFExampleAdd.class
</echo>
</target>
{noformat}
See the last two lines
> spark tests break the build on a shared machine
> -----------------------------------------------
>
> Key: HIVE-11887
> URL: https://issues.apache.org/jira/browse/HIVE-11887
> Project: Hive
> Issue Type: Bug
> Reporter: Sergey Shelukhin
>
> Spark download creates UDFExampleAdd jar in /tmp; when building on a shared machine, someone else's jar from a build prevents this jar from being created (I have no permissions to this file because it was created by a different user) and the build fails.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)