[torqueusers] Torque module for pdsh ?

Joshua Bernstein jbernstein at penguincomputing.com
Thu May 14 14:24:07 MDT 2009


<shameless-pitch>
With Scyld ClusterWare we provide server tools that work outside of TORQUE. One 
tool is called bpsh and allows for a command to be executed sequetially over a 
set of nodes:

$ bpsh 0,2-3 hostname
n2
n3
n0

There is also a called beorun, that lauches the given command in parallel (not 
MPI parallel mind you) across a given set of nodes.

Further, Scyld has a unified process space, so running a ps on the central head 
node shows the PIDs running on the compute nodes!

If you are interested in reading more about Scyld ClusterWare you can check here:

http://www.penguincomputing.com/cluster_computing/scyld_clusterware
</shameless-pitch>

Otherwise what would you be looking to add to your existing script?

-Joshua Bernstein
Software Engineer
Penguin Computing


Ole Holm Nielsen wrote:
> Si Hammond wrote:
>> Have you looked at the pbsdsh command?
> 
> Yes, pbsdsh could be used within a batch job.  I'm interested
> in cluster management tools where you can execute a command
> from the central administration server, for example to get
> the process status on the nodes that run a particular job.
> 
> I currently use a simple script that loops over the job's nodes,
> see ftp://ftp.fysik.dtu.dk/pub/Torque/sshjob but I would like
> to improve this by using pdsh.
> 
> Thanks,
> Ole
> _______________________________________________
> torqueusers mailing list
> torqueusers at supercluster.org
> http://www.supercluster.org/mailman/listinfo/torqueusers


More information about the torqueusers mailing list