[torqueusers] Problems with using torque on a single workstation.

Bo Peng ben.bob at gmail.com
Mon Aug 21 15:34:48 MDT 2006


Dear list,

I have a decent linux workstation (two dual-core x64 CPUs). To make it
easier to submit jobs to this workstation, I installed torque today.
The services are running but the pbs scheduler seems not be able to
accept many jobs.. For example, if I submit 100 jobs, most of them
will fail with message like (in sched_logs)

08/21/2006 16:04:28;0040; pbs_sched;Job;3631.localhost.localdomain;Not
enough of the right type of nodes available

My server_priv/nodes has
localhost np=4

and  mom_priv/config has
$pbsserver bp8
$max_load 4.0
opsys RHEL4

(I do not really know what max_load is).

Also, I notice that the /proc/cpuinfo shows 8 CPUs so I guess the
system can also make use of hyper threading of each core. I try to set
nodes as 'localhost np=8', but the service crashes quickly. Can torque
make use of hyper-threading?

Many thanks in advance.
Bo


More information about the torqueusers mailing list