I mean , I tried that already. In my .bash_profile , I set up HADOOP_HOME, HADOOP_MAPRED_HOME, HADOOP_COMMON_HOME, HADOOP_HDFS_HOME, HADOOP_YARN_HOME, HADOOP_CONF_DIR to point to hadoop2 directory. And similarly I have HADOOP_PREFIX and HADOOP_HOME for hadoop-1. So , i comment out hadoop-2 specific conf lines in .bash_profile while running hadoop-1 and vice-versa. And here the problem comes, that when trying to run hadoop-2 , it asks to format the hadoop.tmp.dir set up by hadoop-1.
Just for context- I have done fair amount of work on hadoop-1 and this is my first attempt at hadoop-2. -- Chandra On Mon, May 5, 2014 at 11:31 AM, chandra kant < chandralakshmikan...@gmail.com> wrote: > Thanks.. > I set it accordingly.. > > > On Monday, 5 May 2014, Shengjun Xin <s...@gopivotal.com> wrote: > >> I think you need to set two different environment variables for hadoop1 >> and hadoop2, such as HADOOP_HOME, HADOOP_CONF_DIR and before you run hadoop >> command, you need to make sure the correct environment variables are enable. >> >> >> On Mon, May 5, 2014 at 12:36 PM, chandra kant < >> chandralakshmikan...@gmail.com> wrote: >> >>> >>> Hi, >>> Is it possible to install both hadoop1 and hadoop2 side by side on >>> single node for development purpose so that I can choose any one of them by >>> just shutting down one and starting another? >>> I installed hadoop 1.2.1 and ran air successfully. >>> Next, when I try to do "hdfs namenode - format " from hadoop2.2.0 , it >>> tries to format the hadoop.tmp.dir set up by hadoop1.2.1 , which is clearly >>> not desirable. I set up "dfs.namenode.name.dir" and "dfs.datanode.data.dir" >>> in "hdfs-site.xml" to different >>> locations. >>> But again the same problem... >>> Any suggestions.... >>> >>> -- >>> Chandra >>> >>> >> >> >> -- >> Regards >> Shengjun >> >