<property>
<name>mapred.tasktracker.map.tasks.maximum</name>
<value>*2*</value>
<description>The maximum number of map tasks that will be run
simultaneously by a task tracker.
</description>
</property>
<property>
<name>mapred.tasktracker.reduce.tasks.maximum</name>
<value>*2*</value>
<description>The maximum number of reduce tasks that will be run
simultaneously by a task tracker.
</description>
</property>
but the above in your hadoop-site.xml and change the *2* to the number
of your cpus, that will essentially do what you are asking
L. Mark Hall wrote:
Sorry, I sent this earlier from the wrong account.
Hello to the Hadoop Core development group,
I am interested as to weather or not I could use Hadoop in the
pseudo-distributed mode to launch multiple client processes on all
cores of a multi-core processor and then re-assemble the results.
Map-reduce is probably overkill for this kind of thing, but I have
very limited engineering resources and can't really start from scratch
with pipes or sockets or an MPI implementation. If Hadoop can be
configured to do this out of the box, that would be very useful. I am
also dumping my results into a Java database (H2), so it would be nice
to keep the entire application in Java.
I would greatly appreciate any information that could be provided.
Mark
LMH_medchemist
--
Jason Venner
Attributor - Publish with Confidence <http://www.attributor.com/>
Attributor is hiring Hadoop Wranglers and coding wizards, contact if
interested
