You are viewing a plain text version of this content. The canonical link for it is here.
Posted to common-user@hadoop.apache.org by Usman Waheed <us...@opera.com> on 2009/04/02 23:05:19 UTC

Question about upgrading

Hello,

I have a 5 node cluster with one master node. I am upgrading from 16.4 to  
18.3 but am a little confused if i am doing it the right way. I read up on  
the documentatin and how to use the -upgrade switch but want to make sure  
i havent missed any step.

First i took down the cluster by issuing stop-all.sh on the master node.
I installed the new hadoop by untaring the tar ball and then copied the  
config files from the old setup 16.4/conf/* into 18.3/conf/*. Changed some  
symlinks to point to the new version. Performed this step on all the  
master and slave nodes.

Then i went on the master and started the master node only with the  
-upgrade switch using the command in the new hadoop version directory.  
Waited for everything to go smoothly. No error were reported i didnt  
change any setting in the config files and just copied from the old  
version conf directory.

Then i started the other data nodes, it should work right. Did i miss  
anything, i am the sysadmin for this setup so want to make sure i do this  
right and not have to reformat the file system and cant afford to loose  
the data for every upgrade. Want to keep the file system as is and upgrade  
 from 16.4 to 18.3.

If i missed any important detail or steps please advise.

Thanks,
Usman