Thanks. I had tried running the command as you suggested but got a ClassNotFoundException exception(for the mapper class) in the task trackers. I did a
bin/hadoop jar /root/myjob.jar (Already had the main class defined in manifest) There is probably something else that I am doing wrong. Appreciate your help. Thanks, Deepika -----Original Message----- From: Chandraprakash Bhagtani [mailto:[email protected]] Sent: Saturday, August 28, 2010 10:40 PM To: [email protected] Subject: Re: Deploying my job jar on hadoop cluster Deepika, You just have to run the following command on any of the cluster node HADOOP_HOME/bin/hadoop jar <job_jar_name> <class_name> <other_args> this command will automatically copy the jar on all the tasktrackers. On Sun, Aug 29, 2010 at 6:07 AM, Deepika Khera <[email protected]>wrote: > Hi, > > I want to deploy my map reduce job jar on the Hadoop cluster. I've always > done that by doing the following - > > 1. Copying the job jar to all datanodes > 2. Having the job jar on the hadoop classpath on all machines. > > Isn't hadoop capable of copying over the job jar to all machines in the > cluster? This is what I read (that job tracker copies the job jar, etc) , > but if I don't do the above the task trackers cannot find the job. I know I > am missing something. > > Could someone please let me know how I can run my job without having to > copy it over all machines in the cluster? > > Thanks, > Deepika > -- Thanks & Regards, Chandra Prakash Bhagtani, Nokia India Pvt. Ltd.
