[torqueusers] torque configuration question

Jacqueline Scoggins jscoggins at lbl.gov
Wed Aug 24 17:48:27 MDT 2005


That would not work as well.  One way I did get it to work was the
following way:

adding dualcore and shared as two separate properties allow for the
users to specify what type of node group they wish to run on.  All they
have to do is the 

qsub -l nodes=x:dualcore   or qsub -l nodes-x:shared.

Without any property tagged they can use any nodes.  At this time we
want to not allow that and the only way to do this is to have the users
specify.  I could also wrap qsub if I wanted to to check if they did
specify the default would be shared.  I would rewrite their qsub command
for them in the script and make shared be enforced.   I did not want to
do this at this time.

Thanks for your input.

Jackie

On Wed, 2005-08-24 at 16:44, Dave Jackson wrote:
> Jacqueline,
>   
> > qsub -l nodes:4  job       
> 
> > Sends the job to all free nodes even dual core ones.
> 
>   Can you try submitting the job as 'qsub -l nodes=1:ppn=4'?  This
> should force the job to only run quad proc nodes.  If you want to do
> anything much more advanced than this, you may need to bail on the
> default scheduler.
> 
> Dave
>  
> On Wed, 2005-08-24 at 13:09 -0700, Jacqueline Scoggins wrote:
> > Does anyone know how to allow for 2 class of hosts to be included in the
> > nodes file to allow users to have to specify via a property tag and the
> > jobs will only go to those nodes?
> > 
> > i.e. 
> > 
> > nodes 
> > 
> >   node0001  np=2 shared
> >   node0002  np=2 shared
> >    ...
> > 
> >   node0030 np=4 dualcore
> >   node0031 np=4 dualcore
> > 
> > qsub -l nodes=4:dualcore    <jobs>     Sends the jobs correctly to the 
> >                                                    dual core queues even
> > if the 
> > 					shared ones are free
> > 
> > but
> > 
> > qsub -l nodes:4  job                    Sends the job to all free nodes
> > even
> >                                                dual core ones.
> > 
> > I have the following queue set up:
> > 
> > Queue            Max Tot Ena Str Que Run Hld Wat Trn Ext Type
> > ---------------- --- --- --- --- --- --- --- --- --- --- ----------
> > default            0   0 yes yes   0   0   0   0   0   0 Route     
> > long               0   7 yes yes   0   7   0   0   0   0 Execution 
> > 
> > 
> > qmgr -c 'l q default'
> > Queue default
> >         queue_type = Route
> >         total_jobs = 0
> >         state_count = Transit:0 Queued:0 Held:0 Waiting:0 Running:0
> > Exiting:0 
> >         route_destinations = long
> >         enabled = True
> >         started = True
> > 
> >  qmgr -c 'l q long'
> > Queue long
> >         queue_type = Execution
> >         total_jobs = 7
> >         state_count = Transit:0 Queued:0 Held:0 Waiting:0 Running:7
> > Exiting:0 
> >         from_route_only = True
> >         resources_default.nodect = 1
> >         resources_default.nodes = 1
> >         resources_assigned.nodect = 32
> >         enabled = True
> >         started = True
> > 
> > 
> > 
> > What can I do?  I am only running torque version 1.1.0p2-aspen3.
> > 
> > Thanks
> > 
> > 
> > 
> > 
> > _______________________________________________
> > torqueusers mailing list
> > torqueusers at supercluster.org
> > http://www.supercluster.org/mailman/listinfo/torqueusers
> 



More information about the torqueusers mailing list