[torqueusers] managing consumable resources

Brian O'Connor briano at sgi.com
Thu Aug 21 22:27:08 MDT 2008


Hi 

We have a 32p Altix running SLES10sp1. It has a 5 user nodelocked
licensed
application, lets call it "bogus" ;)

We want torque-2.3.0/maui to manage the allocation of the licenses as a
consumable
resource, in the same way that it manages cpu allocation. There is no
other
interactive use of bogus, and so it should just be a node locked generic
consumable resource. The jobs should wait in the queue until a license
becomes
available.

I am having a bit of trouble understanding the somewhat confusing
torque/maui
documentation.

I first tried without maui(ie using pbs_sched) and adding to
<torque>mom_priv/config

bogus 5

and then "qsub -l other=bogus script.sh" and
"qsub -l software=bogus script.sh"

but these jobs run regardless. The bogus resource was not decremented

This is discussed 

http://www.clusterresources.com/torquedocs21/a.cmomconfig.shtml

I then tried using maui. I have used maui before in simple
configuations.

The maui.cfg is

--------------------
# maui.cfg 3.2.6p19

SERVERHOST            mel-dem1.melbourne.sgi.com
# primary admin must be first in list
ADMIN1                root sgi
ADMIN3                wwwrun

# Resource Manager Definition

RMCFG[base] TYPE=PBS

# Allocation Manager Definition

AMCFG[bank]  TYPE=NONE

# full parameter docs at
http://supercluster.org/mauidocs/a.fparameters.html
# use the 'schedctl -l' command to display current configuration

RMPOLLINTERVAL        00:00:30

SERVERPORT            42559

# Admin: http://supercluster.org/mauidocs/a.esecurity.html


LOGFILE               maui.log
LOGFILEMAXSIZE        10000000
LOGLEVEL              3

# Job Priority:
http://supercluster.org/mauidocs/5.1jobprioritization.html

QUEUETIMEWEIGHT       1

# FairShare: http://supercluster.org/mauidocs/6.3fairshare.html

FSPOLICY              PSDEDICATED
FSDEPTH               7
FSINTERVAL            604800
FSDECAY               0.80

USERCFG[DEFAULT] MAXIJOB=20 MAXJOB=10,20

NODEMAXLOAD 40

# allow jobs to share node
NODEACCESSPOLICY  SHARED

# track background load
NODELOADPOLICY            ADJUSTPROCS
NODEUNTRACKEDLOADFACTOR   1.2

# favor short jobs, disfavor large jobs
QUEUETIMEWEIGHT   0
RESOURCEWEIGHT    -10
PROCWEIGHT        128
MEMWEIGHT         1
XFACTOR           1000


# kill resource hogs
RESOURCEUTILIZATIONPOLICY ALWAYS
RESOURCEUTILIZATIONACTION CANCEL
#I added 

NODECFG[GLOBAL]  GRES=bogus:5
---------------------------------

and used "qsub -W x=GRES:bogus+1 script.sh"

but these also ran regardless.

what am I missing?

Brian O'Connor
-----------------------------------------------------------------------
SGI Consulting
Email: briano at sgi.com, Mobile +61 417 746 452
Phone: +61 3 9963 1900, Fax:  +61 3 9963 1902
357 Camberwell Road, Camberwell, Victoria, 3124
AUSTRALIA
http://www.sgi.com/support/services
-----------------------------------------------------------------------


More information about the torqueusers mailing list