You are viewing a plain text version of this content. The canonical link for it is here.
Posted to general@hadoop.apache.org by Mohammed A Hassan <mh...@gmu.edu> on 2010/03/12 01:07:10 UTC
hadoop 0.20.2 and Ubuntu 9.04 and 9.10 problem.
Hi,
I am working on Ubuntu 9.04/9.10 and hadoop 0.20.2. Here the problem is that for Pseudo Distributed and Distributed clusters, it takes undefined time (1 minute to several minutes, sometimes more than 10 minutes without any result) to up the live nodes.
It shows that the number of live nodes is zero in http://localhost:50070/dfshealth.jsp for an uncertain amount of time.
Regards
M. Hassan
Re: hadoop 0.20.2 and Ubuntu 9.04 and 9.10 problem.
Posted by Steve Loughran <st...@apache.org>.
Todd Lipcon wrote:
> Hi,
>
> My guess is that the hosts in your cluster are low on entropy.
>
> If you jstack the daemons as they are starting up, do you see them stuck in
> a stace trace inside SecureRandom?
>
> -Todd
Or the machines don't know who they are and are stuck in DNS operations.
Again a stack trace will show you where things lie
>
> On Thu, Mar 11, 2010 at 4:07 PM, Mohammed A Hassan <mh...@gmu.edu> wrote:
>
>> Hi,
>>
>> I am working on Ubuntu 9.04/9.10 and hadoop 0.20.2. Here the problem is
>> that for Pseudo Distributed and Distributed clusters, it takes undefined
>> time (1 minute to several minutes, sometimes more than 10 minutes without
>> any result) to up the live nodes.
>> It shows that the number of live nodes is zero in
>> http://localhost:50070/dfshealth.jsp for an uncertain amount of time.
>>
>> Regards
>> M. Hassan
>>
>
>
>
Re: hadoop 0.20.2 and Ubuntu 9.04 and 9.10 problem.
Posted by Todd Lipcon <to...@cloudera.com>.
Hi,
My guess is that the hosts in your cluster are low on entropy.
If you jstack the daemons as they are starting up, do you see them stuck in
a stace trace inside SecureRandom?
-Todd
On Thu, Mar 11, 2010 at 4:07 PM, Mohammed A Hassan <mh...@gmu.edu> wrote:
> Hi,
>
> I am working on Ubuntu 9.04/9.10 and hadoop 0.20.2. Here the problem is
> that for Pseudo Distributed and Distributed clusters, it takes undefined
> time (1 minute to several minutes, sometimes more than 10 minutes without
> any result) to up the live nodes.
> It shows that the number of live nodes is zero in
> http://localhost:50070/dfshealth.jsp for an uncertain amount of time.
>
> Regards
> M. Hassan
>
--
Todd Lipcon
Software Engineer, Cloudera