Usually better to start your own thread for such purposes. However, my
On Sat, Dec 1, 2012 at 9:41 AM, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote:
> Hi JM,
> If you migrate 1.0.3 to 2.0.x, could you mind to share your migration steps? it is because I also have a 1.0.4 cluster (Ubuntu 12.04, Hadoop 1.0.4, Hbase 0.94.2 and ZooKeeper 3.4.4 ) and want to migrate it to 2.0.x in order to avoid the hardware failure of the NameNode.
HDFS upgrade should be simple enough to do, by invoking the namenode's
first startup with a -upgrade command.
For MR, since Apache Hadoop 2.0.x carries MR2 (which runs atop YARN
), you would have to recompile any custom jobs you have, against an
Apache Hadoop 2.x dependency, and use these new builds.
For HBase, following the HBase's hadoop-2 build instructions should
give you a usable tarball to deploy with.
In case you are open to using packages, you can also look at Apache
Bigtop , as they have RPM/DEB scripts for Hadoop 2.x and other
ecosystem components that you can leverage.
> I have a testing cluster ready for the migration test.
 - http://www.cloudera.com/blog/2012/10/mr2-and-yarn-briefly-explained/
 - http://bigtop.apache.org