[torqueusers] maui-3.2.6p13 - Configuring Generic Consumable Floating Resources Patch

[Dave Jackson]

Jeffrey,

  Thanks for the patch!  A new snapshot has been released.  Please let
us know if you find anything further.

Dave



On Mon, 2005-12-12 at 16:15 -0600, Jeffrey B. Reed wrote:
> I have also sent this to mauiusers group, but I thought it might be helpful to 
> others that are trying to use Generic Consumable Floating Resources.
> 
> Configuring maui-3.2.6p13 and torque-2.0.0p2 to enable Generic Consumable
> Floating Resources, I came across a bug that caused maui to segfault.
> 
> maui.cfg
> --------
> 
> ...
> DEFERTIME 0
> NODECFG[GLOBAL] GRES=verilog:1
> ...
> 
> This defines a psuedo-node named GLOBAL.  You can actually view the node using:
> 
> % checknode GLOBAL
> checking node GLOBAL
> 
> State:      Idle  (in current state for 00:56:15)
> Configured Resources: verilog: 1
> Utilized   Resources: [NONE]
> Dedicated  Resources: [NONE]
> Opsys:        [NONE]  Arch:      [NONE]
> Speed:      0.00  Load:       0.000
> Features:   [NONE]
> Classes:    [NONE]
> 
> Reservations:
> NOTE:  no reservations on node
> 
> Using the -l construct of torque/pbs qsub command you can submit a job
> requesting this floating resource:
> 
> % qsub -l software=verilog job.sh
> 
> However, this caused maui to core dump!  In locating the cause of the
> segmentation fault, I discovered that the psuedo-node GLOBAL is created without
> a valid DM value.  I corrected this by checking for a valid DM value prior to
> accessing the DM->Type field.
> 
> Patch is enclosed.
> 
> Jeff
> 
> plain text document attachment (maui-3.2.6p13-patch)
> --- maui-3.2.6p13.orig/src/moab/MJob.c	2005-05-05 17:03:06.000000000 -0500
> +++ maui-3.2.6p13/src/moab/MJob.c	2005-12-12 14:59:16.000000000 -0600
> @@ -9708,7 +9708,7 @@
>      }
>    else
>      {
> -    if (N->RM->Type == mrmtLL)
> +    if (N->RM && N->RM->Type == mrmtLL)
>        {
>        if (RQ->NodeCount > 0)
> 
> _______________________________________________
> torqueusers mailing list
> torqueusers at supercluster.org
> http://www.supercluster.org/mailman/listinfo/torqueusers