You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@nutch.apache.org by Rod Taylor <rb...@sitesell.com> on 2005/09/29 21:01:45 UTC
nutch-daemon.sh patch (PID file and IDENT string)
On machines with more than 1 CPU it is useful to run more than a single
instance of Nutch under the same user.
The attached patch allows the admins to override the default storage
locations for pids and use something instead of $USER for instance
identification.
--
Rod Taylor <rb...@sitesell.com>
Re: nutch-daemon.sh patch (PID file and IDENT string)
Posted by Doug Cutting <cu...@nutch.org>.
Rod Taylor wrote:
> On Thu, 2005-09-29 at 14:02 -0700, Doug Cutting wrote:
>
>>With MapReduce on multi-CPU machines it probably makes more sense to set
>>mapred.tasktracker.tasks.maxiumum, 2 by default. This determines the
>>number of tasks a tasktracker will execute simultaneously. Each task is
>>run in a separate JVM.
>
> That does make more sense. I went through the config files a couple of
> times looking for something like that.
Sorry. In the heat of coding I frequently forget to add important
properties to nutch-default.xml. I've added this one now.
Doug
Re: nutch-daemon.sh patch (PID file and IDENT string)
Posted by Gal Nitzan <gn...@usa.net>.
Rod Taylor wrote:
> On Thu, 2005-09-29 at 14:02 -0700, Doug Cutting wrote:
>
>> Rod Taylor wrote:
>>
>>> On machines with more than 1 CPU it is useful to run more than a single
>>> instance of Nutch under the same user.
>>>
>> With MapReduce on multi-CPU machines it probably makes more sense to set
>> mapred.tasktracker.tasks.maxiumum, 2 by default. This determines the
>>
mapred.tasktracker.tasks.maxiumum
where is this property should be? couldn't find it anywhere.
>> number of tasks a tasktracker will execute simultaneously. Each task is
>> run in a separate JVM.
>>
>
> That does make more sense. I went through the config files a couple of
> times looking for something like that.
>
> Thanks.
>
>
>> That said, the patch is still good and I've committed it.
>>
>> Doug
>>
>>
Re: nutch-daemon.sh patch (PID file and IDENT string)
Posted by Rod Taylor <rb...@sitesell.com>.
On Thu, 2005-09-29 at 14:02 -0700, Doug Cutting wrote:
> Rod Taylor wrote:
> > On machines with more than 1 CPU it is useful to run more than a single
> > instance of Nutch under the same user.
>
> With MapReduce on multi-CPU machines it probably makes more sense to set
> mapred.tasktracker.tasks.maxiumum, 2 by default. This determines the
> number of tasks a tasktracker will execute simultaneously. Each task is
> run in a separate JVM.
That does make more sense. I went through the config files a couple of
times looking for something like that.
Thanks.
> That said, the patch is still good and I've committed it.
>
> Doug
>
--
Rod Taylor <rb...@sitesell.com>
Re: nutch-daemon.sh patch (PID file and IDENT string)
Posted by Doug Cutting <cu...@nutch.org>.
Rod Taylor wrote:
> On machines with more than 1 CPU it is useful to run more than a single
> instance of Nutch under the same user.
With MapReduce on multi-CPU machines it probably makes more sense to set
mapred.tasktracker.tasks.maxiumum, 2 by default. This determines the
number of tasks a tasktracker will execute simultaneously. Each task is
run in a separate JVM.
That said, the patch is still good and I've committed it.
Doug