[ https://issues.apache.org/jira/browse/HADOOP-11538?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14303833#comment-14303833 ]
Allen Wittenauer commented on HADOOP-11538: ------------------------------------------- There are two ways to deal with this particular issue: 1. We pass JAVA_HOME on the command line instantiation 2. User configures SendEnv in the appropriate .ssh/config #1 isn't really viable because it assumes that nodes are homogeneous in a cluster. This leaves #2, which, at most, should be a documentation note rather than a code fix. > [Startup script ]Start-dfs.sh and start-yarn.sh does not work when we export > JAVA_HOME Manually > ----------------------------------------------------------------------------------------------- > > Key: HADOOP-11538 > URL: https://issues.apache.org/jira/browse/HADOOP-11538 > Project: Hadoop Common > Issue Type: Bug > Reporter: Jagadesh Kiran N > > Scenario: > ======= > Followed the document for installation of standalone cluster.. > http://hadoop.apache.org/docs/stable/hadoop-project-dist/hadoop-common/SingleCluster.html > export JAVA_HOME=/path/to/Java > When we execute start-dfs.sh , it will execute slaves.sh where it is list of > hostnamess are mentioned. > As slaves.sh will do ssh to machine and start the server, where JAVA_HOME > will not present. > *Actual:* > We will get error like JAVA_HOME is not present > *Expected:* > JAVA_HOME should to passed to slaves.sh as a argument. > -- This message was sent by Atlassian JIRA (v6.3.4#6332)