CentOS - Jul 2009 - Simple resource manager?

I need to serialize computing job requests for two different multicore
machines, and in some near future, for a cluster. I have worked with
SGE but it requires NFS and other administrative steps, plus it seems
a bit overkill for my needs. I guess some simpler queue managing
engine may have been developed, possibly over SSH. Any pointers? TIA.

-- 
Eduardo Grosclaude
Universidad Nacional del Comahue
Neuquen, Argentina

Hi,

have a look at Torque
(http://www.clusterresources.com/products/torque-resource-manager.php)

It might be a bit overkill, but it might also suit your needs :)

	Regards,

	Michel


On Mon, 2009-07-27 at 15:22 -0300, Eduardo Grosclaude
wrote:
> I need to serialize computing job requests for two different multicore
> machines, and in some near future, for a cluster. I have worked with
> SGE but it requires NFS and other administrative steps, plus it seems
> a bit overkill for my needs. I guess some simpler queue managing
> engine may have been developed, possibly over SSH. Any pointers? TIA.

Hi,

On Mon, Jul 27, 2009 at 14:22, Eduardo
Grosclaude<eduardo.grosclaude at gmail.com> wrote:
> I have worked with SGE but it requires NFS
I believe it is possible to set up SGE without any NFS at all. On the
other hand, you will have to handle the distribution of the scripts
and programs you want to run on your grid anyway, so you will probably
need NFS set up for that...

Have a look at GXP, it seems to be exactly what you are looking for:
http://www.logos.t.u-tokyo.ac.jp/gxp/index.php?FrontPage

You can also look at SQS, it seems to be very simple:
http://sqs.sourceforge.net/

GXP looks more powerful, though.

As another post suggested, there is Torque, which is similar to SGE
and probably has similar requirements.

And there is also Condor, which does not need NFS but is more complex
(does checkpoints, etc.):
http://www.cs.wisc.edu/condor/

Then there are the commercial ones, of which Platform LSF seems to be
the more complete one, but as you are looking for something simple, I
recommend you look at the above first.

HTH,
Filipe