enforcing ppn=X limits on per queue level

Dear All,

in our cluster, we have a mixture of nodes with 2 and 4 CPUs which
are served by different queues (but the same tqoeue server). Is
there a way that e.g. the "max2cpus-per-node" queue directly
rejects any jobs which specify "ppn=4" as they cannot be started
there at all? Right now they are accepted but get deferred (and
users sometimes do not realize why).

A submit filter might be an option in some cases _BUT_ submit
filters can only check the script but not the command line
arguments (as far as I understand) and thus are generally only of
very limited use - at least in my view.


Regards,

thomas

-- 
Thomas Zeiser, HPC Services
Friedrich-Alexander-Universitaet Erlangen-Nuernberg
Regionales Rechenzentrum Erlangen (RRZE)
Martensstrasse 1, 91058 Erlangen, Germany