You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@hbase.apache.org by satish <sa...@gmail.com> on 2013/10/22 16:53:19 UTC

Need to update dfs.data.dir in Data Nodes. Best approach ?

Hi All, 

We have hbase 0.94 running with 6 slave nodes. I need to change the
dfs.data.dir (Data Node Storage Path) on two of the data nodes. What is the
cleanest  way to update the path. I googled and found two options
1. Remove the node from cluster (stop hbase region server, datanode,
tasktracker) --> update the path in xml -->  add the node to the cluster
2. Remove the node (stop hbase region server, datanode, tasktracker) -->Copy
the data manually from old dfs.data.dir to the new dfs.data.dir -->  Update
the  path in xml  ---> add the node back to the cluster.

which is cleanest option of above two. Is there any disadvantage of any of
the approach.



--
View this message in context: http://apache-hbase.679495.n3.nabble.com/Need-to-update-dfs-data-dir-in-Data-Nodes-Best-approach-tp4051992.html
Sent from the HBase User mailing list archive at Nabble.com.

RE: Need to update dfs.data.dir in Data Nodes. Best approach ?

Posted by Sandeep L <sa...@outlook.com>.
Both will work but 2nd one is better.

Thanks,Sandeep.

> Date: Tue, 22 Oct 2013 08:43:10 -0700
> Subject: Re: Need to update dfs.data.dir in Data Nodes. Best approach ?
> From: pradeepg26@gmail.com
> To: user@hbase.apache.org
> 
> I think you want to use option 2. It preserves the data thats on those data
> nodes.
> 
> P.S: The Hadoop mailing list might be a better list to post this (type) of
> question on. This is not really HBase specific.
> 
> 
> On Tue, Oct 22, 2013 at 7:53 AM, satish <sa...@gmail.com> wrote:
> 
> > Hi All,
> >
> > We have hbase 0.94 running with 6 slave nodes. I need to change the
> > dfs.data.dir (Data Node Storage Path) on two of the data nodes. What is the
> > cleanest  way to update the path. I googled and found two options
> > 1. Remove the node from cluster (stop hbase region server, datanode,
> > tasktracker) --> update the path in xml -->  add the node to the cluster
> > 2. Remove the node (stop hbase region server, datanode, tasktracker)
> > -->Copy
> > the data manually from old dfs.data.dir to the new dfs.data.dir -->  Update
> > the  path in xml  ---> add the node back to the cluster.
> >
> > which is cleanest option of above two. Is there any disadvantage of any of
> > the approach.
> >
> >
> >
> > --
> > View this message in context:
> > http://apache-hbase.679495.n3.nabble.com/Need-to-update-dfs-data-dir-in-Data-Nodes-Best-approach-tp4051992.html
> > Sent from the HBase User mailing list archive at Nabble.com.
> >
 		 	   		  

Re: Need to update dfs.data.dir in Data Nodes. Best approach ?

Posted by Pradeep Gollakota <pr...@gmail.com>.
I think you want to use option 2. It preserves the data thats on those data
nodes.

P.S: The Hadoop mailing list might be a better list to post this (type) of
question on. This is not really HBase specific.


On Tue, Oct 22, 2013 at 7:53 AM, satish <sa...@gmail.com> wrote:

> Hi All,
>
> We have hbase 0.94 running with 6 slave nodes. I need to change the
> dfs.data.dir (Data Node Storage Path) on two of the data nodes. What is the
> cleanest  way to update the path. I googled and found two options
> 1. Remove the node from cluster (stop hbase region server, datanode,
> tasktracker) --> update the path in xml -->  add the node to the cluster
> 2. Remove the node (stop hbase region server, datanode, tasktracker)
> -->Copy
> the data manually from old dfs.data.dir to the new dfs.data.dir -->  Update
> the  path in xml  ---> add the node back to the cluster.
>
> which is cleanest option of above two. Is there any disadvantage of any of
> the approach.
>
>
>
> --
> View this message in context:
> http://apache-hbase.679495.n3.nabble.com/Need-to-update-dfs-data-dir-in-Data-Nodes-Best-approach-tp4051992.html
> Sent from the HBase User mailing list archive at Nabble.com.
>