Hi, Sonal

       Thanks for your information.
       Because some users want to modify the source codes of hadoop, these
users want to install their own hadoop version in the same clusters. After
they modify their hadoop version, they may compile and install the modified
hadoop version and don't want to make any influence to other users.

       Does your method make effect on this?

Thanks a lot,

May

 

 

---------------------------------------------------

Hi,

 

Do your users want different versions of Hadoop? Or can they share the same
hadoop cluster and schedule their jobs? If the latter, Hadoop can be
configured to run for multiple users, and each user can submit their data
and jobs to the same cluster. Hence you can maintain a single cluster and
utilize your resources more efficiently. You can read more here:

 

http://www.ibm.com/developerworks/linux/library/os-hadoop-scheduling/index.h
tml

 

http://www.cloudera.com/blog/2008/11/job-scheduling-in-hadoop/

 

Best Regards,
Sonal
Crux: Reporting for HBase <https://github.com/sonalgoyal/crux>
Nube Technologies <http://www.nubetech.co>  
On Fri, Aug 24, 2012 at 9:13 AM, Li Shengmei <[EMAIL PROTECTED]> wrote:

Hi, all

         There are many users in hadoop platform. Can they install their own
hadoop version on the same clusters platform?

I tried to do this but failed. There exsited a user account and the user
install his hadoop. I create another account and install his hadoop. The
logs display "ERROR org.apache.hadoop.hdfs.server.namenode.NameNode:
java.net.BindException: Problem binding to hadoop01/10.3.1.91:9000 : Address
already in use". So I change the port no. to 8000, but still failed.

When I "start-all.sh", the namenode can't start, the logs display "ERROR
org.apache.hadoop.security.UserGroupInformation: PriviledgedActionException
as:lismhadoop cause:java.net.BindException: Address already in use"

Can anyone give some suggestions?

Thanks,

May

 

Hi Li,

The approach of everyone having their own version of hadoop on same
cluster is way too much complicated.

Good aproach would be all of them test the patches they are compiling
for hadoop in pseudo mode on their person laptops/desktops and then
merge the features or patches with a single version to cluster.

It is never recommended that you run multiple version of hadoops on
single hardware cluster.

Thanks,
nitin

On Fri, Aug 24, 2012 at 1:42 PM, Li Shengmei <[EMAIL PROTECTED]> wrote:
> Hi, Sonal
>
>        Thanks for your information.
>        Because some users want to modify the source codes of hadoop, these
> users want to install their own hadoop version in the same clusters. After
> they modify their hadoop version, they may compile and install the modified
> hadoop version and don’t want to make any influence to other users.
>
>        Does your method make effect on this?
>
> Thanks a lot,
>
> May
>
>
>
>
>
> ---------------------------------------------------
>
> Hi,
>
>
>
> Do your users want different versions of Hadoop? Or can they share the same
> hadoop cluster and schedule their jobs? If the latter, Hadoop can be
> configured to run for multiple users, and each user can submit their data
> and jobs to the same cluster. Hence you can maintain a single cluster and
> utilize your resources more efficiently. You can read more here:
>
>
>
> http://www.ibm.com/developerworks/linux/library/os-hadoop-scheduling/index.html
>
>
>
> http://www.cloudera.com/blog/2008/11/job-scheduling-in-hadoop/
>
>
>
> Best Regards,
> Sonal
> Crux: Reporting for HBase
> Nube Technologies
>
>
>
>
>
>
> On Fri, Aug 24, 2012 at 9:13 AM, Li Shengmei <[EMAIL PROTECTED]> wrote:
>
> Hi, all
>
>          There are many users in hadoop platform. Can they install their own
> hadoop version on the same clusters platform?
>
> I tried to do this but failed. There exsited a user account and the user
> install his hadoop. I create another account and install his hadoop. The
> logs display “ERROR org.apache.hadoop.hdfs.server.namenode.NameNode:
> java.net.BindException: Problem binding to hadoop01/10.3.1.91:9000 : Address
> already in use”. So I change the port no. to 8000, but still failed.
>
> When I “start-all.sh”, the namenode can’t start, the logs display “ERROR
> org.apache.hadoop.security.UserGroupInformation: PriviledgedActionException
> as:lismhadoop cause:java.net.BindException: Address already in use”
>
> Can anyone give some suggestions?
>
> Thanks,
>
> May
>
>

--
Nitin Pawar