[torqueusers] What is the difference between PBS and Torque?

Omer Khalid Omer.Khalid at cern.ch
Mon Feb 25 09:05:08 MST 2008


Hi Chris,

Thanks for the responses. I am wondering if there is any visual diagram
available for Torque/Maui depicting from the start when a job is submitted
to the batch system till it get completed/exited. I am searching some thing
similar to what is available for SLURM on slide 46 from this link:

https://computing.llnl.gov/linux/slurm/lci.7.tutorial.pdf

And also does Torque supports a plug-in architecture to add new modules and
is there a developer's guide some where?

Many thanks,
Omer

On Sun, Feb 24, 2008 at 4:56 AM, Chris Samuel <csamuel at vpac.org> wrote:

>
> ----- "Omer Khalid" <Omer.Khalid at cern.ch> wrote:
>
> > Hi James,
>
> Hello Omer,
>
> > Thanks for your prompt response and explaining me the history of
> > Torque and its architecture. So pbs_server is resource manager that
> > takes care of the resource provisioning, pbs_sched/Maui for scheduling
> > the job and resource allocation and pbs_mom for monitoring the
> > computer node.. right??
>
> That sounds about right.
>
> > So I guess that qmgr is part of pbs_server to set up the queues.
>
> It's actually a command that talks to the pbs_server process.
>
> > Now let's say for a cluster of 20 nodes, 5 different queues are
> specified
> > based on different type of execution priorities. Can all nodes consume
> > jobs from any of the queues
>
> Yes - though you can put rules into Maui/Moab to govern
> what you want to happen here.
>
> > or they are configured for specific queues?
>
> Not really.
>
> > And what happens when a job arrives at the pbs_server? Does
> > pbs_sched/Maui get the status info from the pbs_server about the
> > available resources, and based on a job's requirements put that into a
> > specific queue?
>
> No, queues are more of a Torque concept and Maui maps
> them to classes. pbs_sched might be more constrained though.
>
> > Does it also allocate a specific execution node for
> > that particular job
>
> Yes, it does.
>
> > or the compute node reads the job from a certain queue?
>
> No, it's a push system, rather than a pull one.
>
> > In this case, adding dynamic nodes is possible?
>
> Not easily.
>
> > My thoughts pretty are disjointed but I would much appreciate the
> > explanations..
>
> So are mine! :-)
>
> --
> Christopher Samuel - (03) 9925 4751 - Systems Manager
>  The Victorian Partnership for Advanced Computing
>  P.O. Box 201, Carlton South, VIC 3053, Australia
> VPAC is a not-for-profit Registered Research Agency
> _______________________________________________
> torqueusers mailing list
> torqueusers at supercluster.org
> http://www.supercluster.org/mailman/listinfo/torqueusers
>



-- 
Omer

-------------------------------------------------------
CERN -- European Organization for Nuclear
Research, IT Department, CH-1211,
Geneva 23, Switzerland

Phone: +41 (0) 22 767 2224
Fax:     +41 (0) 22 766 8683
E-mail : Omer.Khalid at cern.ch
Homepage: http://cern.ch/Omer.Khalid
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.supercluster.org/pipermail/torqueusers/attachments/20080225/b4d77363/attachment.html


More information about the torqueusers mailing list