You are viewing a plain text version of this content. The canonical link for it is here.
Posted to mapreduce-dev@hadoop.apache.org by Ted Yu <yu...@gmail.com> on 2010/11/25 07:19:07 UTC
using pipe in the command executed by slaves.sh
Hi,
We use cdh3b2
I want to get per-tasktracker statictics, such as the count of map/reduce
tasks on each node.
The following returns total:
slaves.sh /usr/java/default/bin/jps | /bin/grep Child | /usr/bin/wc
How do I get per-node count ?
Thanks
Re: using pipe in the command executed by slaves.sh
Posted by Ted Yu <yu...@gmail.com>.
[hadoop@us01-ciqps1-name01 ~]$ slaves.sh '/usr/java/default/bin/jps |
/bin/grep Child | /usr/bin/wc -l'
us01-ciqps1-grid01.carrieriq.com: bash: /usr/java/default/bin/jps : No such
file or directory
us01-ciqps1-grid01.carrieriq.com: bash: /bin/grep Child : No such file or
directory
us01-ciqps1-grid01.carrieriq.com: bash: /usr/bin/wc -l: No such file or
directory
But I verified that /usr/java/default/bin/jps exists on each node.
On Thu, Nov 25, 2010 at 12:40 AM, Harsh J <qw...@gmail.com> wrote:
> Hi,
>
> On Thu, Nov 25, 2010 at 11:49 AM, Ted Yu <yu...@gmail.com> wrote:
> > Hi,
> > We use cdh3b2
> > I want to get per-tasktracker statictics, such as the count of map/reduce
> > tasks on each node.
> > The following returns total:
> > slaves.sh /usr/java/default/bin/jps | /bin/grep Child | /usr/bin/wc
> >
> > How do I get per-node count ?
>
> You could get it by passing that entire command to the slaves?
> Something like slaves.sh '/usr/java/default/bin/jps | /bin/grep Child
> | /usr/bin/wc -l' should work I think, instead of aggregating in your
> shell..
>
>
>
> --
> Harsh J
> www.harshj.com
>
Re: using pipe in the command executed by slaves.sh
Posted by Harsh J <qw...@gmail.com>.
Hi,
On Thu, Nov 25, 2010 at 11:49 AM, Ted Yu <yu...@gmail.com> wrote:
> Hi,
> We use cdh3b2
> I want to get per-tasktracker statictics, such as the count of map/reduce
> tasks on each node.
> The following returns total:
> slaves.sh /usr/java/default/bin/jps | /bin/grep Child | /usr/bin/wc
>
> How do I get per-node count ?
You could get it by passing that entire command to the slaves?
Something like slaves.sh '/usr/java/default/bin/jps | /bin/grep Child
| /usr/bin/wc -l' should work I think, instead of aggregating in your
shell..
--
Harsh J
www.harshj.com