[Fwd: [Mauiusers] Patch for nodeaccespolicy SINGLEJOB and MAXPS for SMPmachines]

Dave Jackson jacksond at clusterresources.com
Sat Nov 26 14:26:37 MST 2005


Bas,

  The patch looks good but I do not know if the resulting behavior is
globally desired.  It is clear that if a user requests two procs per
node he should be 'charged' for them.  However, if the scheduler
allocates resources to the job based on other admin-specified policies,
the correct action is less clear.

  The correct solution may be to create a new policy, perhaps
'USAGELIMITTRACKINGPOLICY' with a value of either 'requested' or
'dedicated'.  In the first, the user will have his usage limits
'charged' based on what he requested, regardless of how the scheduler or
other admin policies delivered resources, the second would charge based
on what was delivered, regardless of what was requested.

  Would this approach cover your needs?

Dave
  

On Thu, 2005-11-24 at 15:14 +0100, Bas van der Vlies wrote:
> I have just installed maui 3.2.6p13 and i have still have the same 
> problem. I could apply the patch without any problems ;-)
> 
> My question is will the patch be applied or rejected? See explanation below.
> 
> Regards and Thanks
> 
> -------- Original Message --------
> Subject: 	[Mauiusers] Patch for nodeaccespolicy SINGLEJOB and MAXPS for
> SMPmachines
> Date: 	Mon, 14 Feb 2005 17:45:06 +0100
> From: 	Bas van der Vlies <basv at sara.nl>
> To: 	Maui Users <mauiusers at supercluster.org>
> 
> 
> 
> At our side we run one job per nodes and have an MAXPS setting of 600
> hours and max walltime 120 hours. Our nodes have 2 processors. When the
> user submits
> an job for eg:
>     1) qsub -I -lnodes=60:ppn=1 -lwalltime=10:00:00 ( will run )
>     2) qsub -I -lnodes=60:ppn=2 -lwalltime=10:00:00 ( wil not run MAXPS
>                                                       violation)
> 
> Now when job 1 runs is allocates the whole node and maui sees that it
> oocupies 4 task ( 2 nodes and each node two cpu's = 4 tasks). So the
> used tme will becalculated as 60 * 2 * 10 = 1200 hours. What is far more
> then allowed!
> 
> The next example will only run one job instead of 2:
>     qsub -I -lnodes=30:ppn=1 -lwalltime=10:00:00 ( will run )
>     qsub -I -lnodes=30:ppn=1 -lwalltime=10:00:00 ( will not  run MAXPS
>                                                    violation )
> 
> I have an patch that checks if NODEACCESSPOLICY SINGLEJOB is set. If so
> then it forgets the the cpu's per node.
> 
> --- PATCH attached --
> 
> --
> --
> ********************************************************************
> *                                                                  *
> *  Bas van der Vlies                     e-mail: basv at sara.nl      *
> *  SARA - Academic Computing Services    phone:  +31 20 592 8012   *
> *  Kruislaan 415                         fax:    +31 20 6683167    *
> *  1098 SJ Amsterdam                                               *
> *                                                                  *
> ********************************************************************
> 
> _______________________________________________
> mauiusers mailing list
> mauiusers at supercluster.org
> http://www.supercluster.org/mailman/listinfo/mauiusers



More information about the mauiusers mailing list