Hello Anand,
Set your Java home in hadoop-env.sh - /usr/local/hadoop/etc/hadoop/hadoop-env.sh export JAVA_HOME='/usr/lib/jvm/java-7-openjdk-amd64' It would resolve your error. Thanks, S.RagavendraGanesh ViSolve Hadoop Support Team ViSolve Inc. | San Jose, California Website: www.visolve.com <http://www.visolve.com> email: servi...@visolve.com <mailto:servi...@visolve.com> | Phone: 408-850-2243 From: Alexandru Pacurar [mailto:alexandru.pacu...@propertyshark.com] Sent: Wednesday, March 25, 2015 3:17 PM To: user@hadoop.apache.org Subject: RE: Hadoop 2.6.0 Error Hello, I had a similar problem and my solution to this was setting JAVA_HOME in /etc/environment. The problem is, from what I remember, that the start-dfs.sh script calls hadoop-daemons.sh with the necessary options to start the Hadoop daemons. hadoop-daemons.sh in turn calls hadoop-daemon.sh with the necessary options via ssh, in an non-interactive fashion. When you are executing a command via ssh in a non-interactive manner (ex. ssh host1 ‘ls -la’ ) you have a minimal environment and you do not source the .profile file, and other environment related files. But the /etc/environment is sourced so you could set JAVA_HOME there. Technically you should set BASH_ENV there which should point to a file containing the environment variables you need. For more info see <http://stackoverflow.com/questions/216202/why-does-an-ssh-remote-command-get-fewer-environment-variables-then-when-run-man> http://stackoverflow.com/questions/216202/why-does-an-ssh-remote-command-get-fewer-environment-variables-then-when-run-man, or man bash Thank you, Alex From: Olivier Renault [ <mailto:orena...@hortonworks.com> mailto:orena...@hortonworks.com] Sent: Wednesday, March 25, 2015 10:44 AM To: <mailto:user@hadoop.apache.org> user@hadoop.apache.org; Anand Murali Subject: Re: Hadoop 2.6.0 Error It should be : export JAVA_HOME=… Olivier From: Brahma Reddy Battula Reply-To: " <mailto:user@hadoop.apache.org> user@hadoop.apache.org" Date: Wednesday, 25 March 2015 08:28 To: " <mailto:user@hadoop.apache.org> user@hadoop.apache.org", Anand Murali Subject: RE: Hadoop 2.6.0 Error HI Ideally it should take effect , if you configure in .profile or hadoop-env.sh.. As you told that you set in .profile ( hope you did source ~/.profile ),,, did you verify that take effect..? ( by checking echo $JAVA_HOME,, or jps..etc )... Thanks & Regards Brahma Reddy Battula _____ From: Anand Murali [ <mailto:anand_vi...@yahoo.com> anand_vi...@yahoo.com] Sent: Wednesday, March 25, 2015 1:30 PM To: <mailto:user@hadoop.apache.org> user@hadoop.apache.org; Anand Murali Subject: Re: Hadoop 2.6.0 Error Dear All: Even after setting JAVA_HOME in .profile I get JAVA_HOME is not set and could not be found -error. If anyone of you know of a more stable version please do let me know. Thanks, Anand Murali 11/7, 'Anand Vihar', Kandasamy St, Mylapore Chennai - 600 004, India Ph: (044)- 28474593/ 43526162 (voicemail) On Wednesday, March 25, 2015 12:57 PM, Anand Murali < <mailto:anand_vi...@yahoo.com> anand_vi...@yahoo.com> wrote: Dear Mr. Bhrama Reddy: Should I type SET JAVA_HOME=/usr/lib/jvm/java-1.7.0-openjdk-amd64 in root (profile) or at user level (.profile). Reply most welcome Thanks Regards Anand Murali 11/7, 'Anand Vihar', Kandasamy St, Mylapore Chennai - 600 004, India Ph: (044)- 28474593/ 43526162 (voicemail) On Wednesday, March 25, 2015 12:37 PM, Anand Murali < <mailto:anand_vi...@yahoo.com> anand_vi...@yahoo.com> wrote: Dear All: I get this error shall try setting JAVA_HOME in .profile Starting namenodes on [localhost] localhost: Error: JAVA_HOME is not set and could not be found. cat: /home/anand_vihar/hadoop-2.6.0/conf/slaves: No such file or directory Starting secondary namenodes [0.0.0.0] 0.0.0.0: Error: JAVA_HOME is not set and could not be found. <mailto:anand_vihar@Latitude-E5540:~/hadoop-2.6.0/sbin$> anand_vihar@Latitude-E5540:~/hadoop-2.6.0/sbin$ Thanks Anand Murali 11/7, 'Anand Vihar', Kandasamy St, Mylapore Chennai - 600 004, India Ph: (044)- 28474593/ 43526162 (voicemail) On Wednesday, March 25, 2015 12:22 PM, Brahma Reddy Battula < <mailto:brahmareddy.batt...@huawei.com> brahmareddy.batt...@huawei.com> wrote: Instead of exporting the JAVA_HOME, Please set JAVA_HOME in system level ( like putting in /etc/profile...) For more details please check the following jira. <https://urldefense.proofpoint.com/v2/url?u=https-3A__issues.apache.org_jira_browse_HADOOP-2D11538&d=AwMGaQ&c=011sAuWmEGuAVOWaydfJseI4cXS0FNE1rSze05FbvBE&r=ZAivIGqQeG1xHpYmaYPBcgf03sOn2v2J3r9hTy7sK4jmLRosJPeq02_uC8oa5Pa7&m=C3HqXnaRYKppI_bCvbqjXzXa6SZIULL08MPZDtxtGNM&s=7JTHoL-82HcBMB5Gdgtjowf5Dv96ZX8GY9TEEzMoIsU&e=> https://issues.apache.org/jira/browse/HADOOP-11538 Thanks & Regards Brahma Reddy Battula _____ From: Anand Murali [ <mailto:anand_vi...@yahoo.com> anand_vi...@yahoo.com] Sent: Wednesday, March 25, 2015 11:23 AM To: User Hadoop Subject: Hadoop 2.6.0 Error Dear All: Request help/advise as I am unable to start Hadoop. Performed follow steps in Ubuntu 14.10 1. ssh localhost 2. Did following exports in user defined hadoop.sh and ran it succesfully 1. EXPORT JAVA_HOME=/usr/lib/jvm/java-7-openjdk-amd64 2. EXPORT HADOOP_INSTALL=/home/anand_vihar/hadoop-2.6.0 3. EXPORT PATH=:$PATH:$HADOOP_INSTALL/sbin:$HADOOP_INSTALL/bin 3. Tested hadoop version succesfully 4. Ran $hadoop namenode -format successfully 5. Modified core-site.xml, hdfs-site.xml and mapred-site.xml to pseudo-distributed mode in /home/anand_vihar/conf directory 6. Ran $start-dfs.sh --config /home/anand_vihar/conf Got error JAVA_HOME not set and slaves not found in /conf. If I echo $JAVA_HOME it is pointing to /usr/lib/jvm/java-7-openjdk-amd6, correctly as set. Help appreciated. Thanks Regards, Anand Murali 11/7, 'Anand Vihar', Kandasamy St, Mylapore Chennai - 600 004, India Ph: (044)- 28474593/ 43526162 (voicemail)