[ https://issues.apache.org/jira/browse/HADOOP-9211?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13553778#comment-13553778 ]
Sarah Weissman commented on HADOOP-9211: ---------------------------------------- Maybe this can be better documented inside hadoop-env.sh? Out of the box changing HADOOP_HEAPSIZE adds a -Xmx argument to the beginning of the bin/hadoop java command, but the hard-coded 128M heap size in HADOOP_CLIENT_OPTS overrides that option with a second -Xmx. As a new user this was very unintuitive, especially since most advice given around the internet for earlier versions will refer to adjusting HADOOP_HEAPSIZE. Also, at least in my case, 128M was not enough to run various examples from the hadoop examples jar, so the first thing many new users might encounter is trying to figure out how to increase heap size. Maybe 128M is a bad value given the default settings of other variables? > HADOOP_CLIENT_OPTS default setting fixes max heap size at 128m, disregards > HADOOP_HEAPSIZE > ------------------------------------------------------------------------------------------ > > Key: HADOOP-9211 > URL: https://issues.apache.org/jira/browse/HADOOP-9211 > Project: Hadoop Common > Issue Type: Bug > Components: conf > Affects Versions: 2.0.2-alpha > Reporter: Sarah Weissman > Original Estimate: 1m > Remaining Estimate: 1m > > hadoop-env.sh as included in the 2.0.2alpha release tarball contains: > export HADOOP_CLIENT_OPTS="-Xmx128m $HADOOP_CLIENT_OPTS" > This overrides any heap settings in HADOOP_HEAPSIZE. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira