Shen Yinjie created AMBARI-24190: ------------------------------------ Summary: HADOOP_CLASSPATH should always put HADOOP_CONFIG_DIR first Key: AMBARI-24190 URL: https://issues.apache.org/jira/browse/AMBARI-24190 Project: Ambari Issue Type: Bug Reporter: Shen Yinjie
Currently,in hadoop-env.xml template, {code}MAPREDUCE_LIBS=/usr/lib/hadoop-mapreduce/* export HADOOP_CLASSPATH=${HADOOP_CLASSPATH}${JAVA_JDBC_LIBS}:${MAPREDUCE_LIBS} if [ -d "/usr/lib/tez" ]; then export HADOOP_CLASSPATH=$HADOOP_CLASSPATH:/usr/lib/tez/*:/usr/lib/tez/lib/*:/etc/tez/conf fi {code} In a scenario when a hadoop client uses hadoop shell and sets HADOOP_USER_CLASSPATH_FIRST , since HADOOP_CLASSPATH contains MAPREDUCE_LIBS,which has a core-site.xml in hadoop-mapreduce-client-jobclient-xx-tests.jar. This could pollute job runtime environment. So, we should put HADOOP_CONFIG_DIR always in front of HADOOP_CLASSPATH. I found that in Hadoop 3.0,this has already been done in hadoop-functions.sh. -- This message was sent by Atlassian JIRA (v7.6.3#76005)