Re: [Condor-users] MPI and Flocked Pools

[Andrew Howard <ahoward@xxxxxxxxxx>]

I don't see why this wouldn't work from a technical aspect, but the  
biggest question is whether the jobs would ever be evicted from the  
flocked-to pools. If you're running a job and one node has to evict  
because the user came back, your entire MPI job is hosed.

--
Andrew Howard
Assistant Research Programmer
Rosen Center for Advanced Computing
Purdue University


On Feb 29, 2008, at 2:19 PM, Philip Papadopoulos wrote:

> I have a setup where I have a number of clusters around campus that  
> can satisfy
> job requests.  I have a "submit" pool (a cluster head node with no  
> compute nodes)
> where users submit condor jobs. Since it has no local resources, all  
> requests are satisfied
> by the flocked-to clusters.  It is simple and works well.
>
> I'm trying to figure out if I can submit MPI jobs on my submit pool  
> and have them run on
> one of the flocked-to clusters? From the online manual, it seems  
> that there is a dedicated
> scheduler node that needs definition and configuration.  The  
> question is, can I have  a dedicated scheduler
> node per cluster and then use flocking to satisfy parallel (MPI) job  
> requests? That is submit an MPI
> job on my "submit pool" and have it's resource request satisfied by  
> one of my flocked-to clusters?
>
>  Or do I need
> one scheduler node ((e.g. my "submit" pool node)).
>
> Thanks,
> Phil
>
>
> --
> Philip Papadopoulos, PhD
> University of California, San Diego
> 858-822-3628 _______________________________________________
> Condor-users mailing list
> To unsubscribe, send a message to condor-users-request@xxxxxxxxxxx  
> with a
> subject: Unsubscribe
> You can also unsubscribe by visiting
> https://lists.cs.wisc.edu/mailman/listinfo/condor-users
>
> The archives can be found at:
> https://lists.cs.wisc.edu/archive/condor-users/