[Mauiusers] job submitted to multiple nodes is blocked
rjacobi at email.arizona.edu
Tue Mar 29 19:30:23 MDT 2011
We're running Maui with Torque on our cluster (10 nodes, 8 cores each).
When I submit a job on multiple nodes specifying them in a certain why
my job gets blocked by Maui, otherwise it runs just fine.
Here's the line I usually use in submission script to submit jobs to
mutliple nodes (say if I need 16 cores): "#PBS -l nodes=2:ppn=8". That
works just fine.
But when I don't need a multiple of 8 processors and don't want to block
processors unnecessarily, I tried the following (say to get 20 cores):
"#PBS -l nodes=2:ppn=8+1:ppn=4"
According to the Torque user manual that should get me 16 cores on 2
nodes and 4 cores on the third. I can submit it without a problem, but
then it is blocked by Maui.
For testing I also tried "#PBS -l nodes=1:ppn=1+1:ppn=1" and "#PBS -l
nodes=1:ppn=8+1:ppn=8", which are blocked as well
Output from showq for the blocked job:
JOBNAME USERNAME STATE PROC WCLIMIT
1272 robert BatchHold 1 00:00:30 Tue Mar 29
The relevant policies (as far as I know), which are set in maui.cfg, are:
Please, can anybody point me to where the problem might be? If there are
queue specific settings that could cause this, I'd be more than happy to
provide more information.
PS: I am aware that I could just reserve multiple of 8 processors and
let some of them sit idle, but on our relatively small cluster that
amounts to s significant waste of resources.
University of Arizona
Department of Aerospace & Mechanical Engineering
1130 N. Mountain Ave.
Tucson, AZ, 85721-0119
tel: +1 (520) 621 4369
mail: rjacobi at email.arizona.edu
The less time you spent on algebra in life, the more time you have to be a happy person. (Kerschen)
Doubt is not a pleasant condition, but certainty is absurd. (Voltaire)
All great truths begin as blasphemies. (Shaw)
Denken ist etwas, das auf Schwierigkeiten folgt und dem das Handeln vorausgeht.(Brecht)
More information about the mauiusers