Thank you very much, Gustavo!<br><br>I forgot about that file. I just check it and I think I read it some time ago... Anyway, there won't be any parallel job running on my cluster, so "1 job 1 core" policy will be easier.<br>
<br>But I'm seeing that the jobs are sent to the nodes in the order they are listed on the nodes file. Maybe it's version dependent.<br><br>Thanks a lot for your answer and sorry for my late reply.<br><br>Fernando.<br>
<br><br><div class="gmail_quote">On Tue, Mar 30, 2010 at 19:55, Gus Correa <span dir="ltr"><<a href="mailto:gus@ldeo.columbia.edu" target="_blank">gus@ldeo.columbia.edu</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">Hi Fernando<div><br>
<br>
Fernando Campos wrote:<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Hi all!!<br>
<br>
I'm running a TORQUE cluster with "pbs_sched" scheduler. Which is the policy for node allocation?? <br>
</blockquote>
<br></div>
As far as I know, the standard pbs_sched job policy just first in first out (FIFO).<div><br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Can you define it anywhere on Torque?? <br>
</blockquote>
<br></div>
There are a few things that can be done in your Torque server<br>
on the ${TORQUE}/sched_priv/sched_config file.<br>
I found it useful to reduce the job starvation limit<br>
(default 24h, I think), to prevent large parallel jobs to<br>
be bypassed forever by serial jobs and small parallel ones.<br>
<br>
For more job control you need to install the Maui scheduler,<br>
and use it instead of pbs_sched.<div><br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Is this case depending on the nodes order in "nodes" file? Should I then put the better performance nodes at the beggining?<br>
<br>
</blockquote>
<br></div>
I think Torque picks the nodes in the reverse order that you list them<br>
in the nodes file.<br>
Hence, if you want the better performance nodes to be used first,<br>
put them on the bottom of the nodes file.<br>
<br>
I hope this helps.<br>
Gus Correa<br>
---------------------------------------------------------------------<br>
Gustavo Correa<br>
Lamont-Doherty Earth Observatory - Columbia University<br>
Palisades, NY, 10964-8000 - USA<br>
---------------------------------------------------------------------<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Thanks!<br>
<br>
Fer.<div><br>
<br>
<br>
<br>
On Tue, Mar 30, 2010 at 16:54, Charles Johnson <<a href="mailto:charles.johnson@accre.vanderbilt.edu" target="_blank">charles.johnson@accre.vanderbilt.edu</a> <mailto:<a href="mailto:charles.johnson@accre.vanderbilt.edu" target="_blank">charles.johnson@accre.vanderbilt.edu</a>>> wrote:<br>
<br>
On Mar 30, 2010, at 9:40 AM, Ken Nielson wrote:<br>
<br>
> Glen Beane wrote:<br>
>><br>
>><br>
>> On Tue, Mar 30, 2010 at 10:07 AM, Glen Beane<br>
<<a href="mailto:glen.beane@gmail.com" target="_blank">glen.beane@gmail.com</a> <mailto:<a href="mailto:glen.beane@gmail.com" target="_blank">glen.beane@gmail.com</a>><br></div>
<mailto:<a href="mailto:glen.beane@gmail.com" target="_blank">glen.beane@gmail.com</a> <mailto:<a href="mailto:glen.beane@gmail.com" target="_blank">glen.beane@gmail.com</a>><div><div></div><div><br>
>> >> wrote:<br>
>><br>
>><br>
>><br>
>> On Tue, Mar 30, 2010 at 9:55 AM, Charles Johnson<br>
>> <<a href="mailto:charles.johnson@accre.vanderbilt.edu" target="_blank">charles.johnson@accre.vanderbilt.edu</a><br>
<mailto:<a href="mailto:charles.johnson@accre.vanderbilt.edu" target="_blank">charles.johnson@accre.vanderbilt.edu</a>><br>
>> <mailto:<a href="mailto:charles.johnson@accre.vanderbilt.edu" target="_blank">charles.johnson@accre.vanderbilt.edu</a><br>
<mailto:<a href="mailto:charles.johnson@accre.vanderbilt.edu" target="_blank">charles.johnson@accre.vanderbilt.edu</a>>>> wrote:<br>
>><br>
>> About the nodes file ... I have always been under the<br>
>> impression that<br>
>> the nodes file gives an ordering to nodes selected for jobs,<br>
>> i.e.,<br>
>> nodes at the top of the list are considered before nodes at<br>
>> the<br>
>> bottom. We are currently in a down time for refreshing<br>
>> hardware, and<br>
>> the whole cluster is quiescent. As a test of hardware we<br>
>> submitted a<br>
>> single job suitable for any one of several hundred nodes at<br>
>> the top of<br>
>> the nodes file. The job ran on a node roughly halfway<br>
down the<br>
>> nodes<br>
>> file. Again, there were no other jobs on the cluster.<br>
>><br>
>> I am curious as to why? Any ideas?<br>
>><br>
>> We are using torque 2.4.5 and moab 5.3.6<br>
>><br>
>><br>
>><br>
>> This is really a Moab question since Moab selects the node that<br>
>> the job will run on, it has nothing to do with the order of the<br>
>> nodes in the TORQUE node file. I think the default for Moab might<br>
>> be "last fit", so as it scans the available nodes it will select<br>
>> the last one it finds that satisfies the requirements for the<br>
>> job. There is a "first fit" and "best fit" option if I remember<br>
>> correctly.<br>
>><br>
>> With the fifo scheduler, then yes, I think the job would run on<br>
>> the first available node in the node list.<br>
>><br>
>><br>
>> actually, I was thinking of LASTAVAILABLE, not "last fit", so the<br>
>> definition would be different than what I stated. This is the<br>
>> correct definition: "This algorithm is a best fit in time algorithm<br>
>> that minimizes the impact of reservation based node-time<br>
>> fragmentation." I think this might be the default, but I don't<br>
>> remember for sure.<br>
>><br>
------------------------------------------------------------------------<br>
>><br>
>> _______________________________________________<br>
>> torqueusers mailing list<br></div></div>
>> <a href="mailto:torqueusers@supercluster.org" target="_blank">torqueusers@supercluster.org</a> <mailto:<a href="mailto:torqueusers@supercluster.org" target="_blank">torqueusers@supercluster.org</a>><div>
<br>
>> <a href="http://www.supercluster.org/mailman/listinfo/torqueusers" target="_blank">http://www.supercluster.org/mailman/listinfo/torqueusers</a><br>
>><br>
> If you were to run TORQUE without a scheduler and started the job<br>
> manually TORQUE will choose the last node in the nodes file to<br>
run on.<br>
><br>
> Ken Nielson<br>
> Adaptive Computing<br>
<br>
<br>
Thanks to all who replied. Clarity reigns. :)<br>
<br>
~Cheers--<br>
<br>
Charles<br>
--<br>
Charles Johnson<br>
Advanced Computing Center for Research and Education<br>
Office: 615-343-2776<br>
Cell: 615-478-5743<br>
_______________________________________________<br>
torqueusers mailing list<br></div>
<a href="mailto:torqueusers@supercluster.org" target="_blank">torqueusers@supercluster.org</a> <mailto:<a href="mailto:torqueusers@supercluster.org" target="_blank">torqueusers@supercluster.org</a>><div>
<br>
<a href="http://www.supercluster.org/mailman/listinfo/torqueusers" target="_blank">http://www.supercluster.org/mailman/listinfo/torqueusers</a><br>
<br>
<br>
<br>
<br>
-- <br>
---------------------------------------------------------------------------------------------------------<br>
Fernando Campos Del Pozo<br>
Departamento de Física Teórica<br>
Facultad de Ciencias / Módulo 15 (C-XI) / Despacho 512<br>
Universidad Autónoma de Madrid<br>
Tlf.: +34-914974893<br></div>
e-mail: <a href="mailto:fernando.campos@uam.es" target="_blank">fernando.campos@uam.es</a> <mailto:<a href="mailto:fernando.campos@uam.es" target="_blank">fernando.campos@uam.es</a>><div><br>
<br>
<br>
------------------------------------------------------------------------<br>
<br>
_______________________________________________<br>
torqueusers mailing list<br>
<a href="mailto:torqueusers@supercluster.org" target="_blank">torqueusers@supercluster.org</a><br>
<a href="http://www.supercluster.org/mailman/listinfo/torqueusers" target="_blank">http://www.supercluster.org/mailman/listinfo/torqueusers</a><br>
</div></blockquote>
<br>
<br>
</blockquote></div><br><br clear="all"><br>-- <br>---------------------------------------------------------------------------------------------------------<br>Fernando Campos Del Pozo<br>Departamento de Física Teórica<br>
Facultad de Ciencias / Módulo 15 (C-XI) / Despacho 512<br>Universidad Autónoma de Madrid<br>Tlf.: +34-914974893<br>e-mail: <a href="mailto:fernando.campos@uam.es" target="_blank">fernando.campos@uam.es</a><br>