[torquedev] pbsdsh and number of processors allocated

Bogdan Costescu Bogdan.Costescu at iwr.uni-heidelberg.de
Fri Jun 19 22:39:50 MDT 2009


On Thu, 18 Jun 2009, Garrick wrote:

> Since some of my users have started using sshfs religiously, I've 
> been thinking the same thing. Pbsfs could be pretty cool.

Where would it exist ? Submit node, server node, mom node, all nodes ?
At least right now the information about the job does not exist on all 
nodes.

>> People are already using awk, perl, sed, and cat to get some of 
>> this info from qstat, pbsnodes, the $PBS_NODEFILE, etc.  All of 
>> these things are hacks.

For monitoring of jobs, I use the scheduler's (Maui) view. Inside a 
job, I use the native mechanism of the MPI lib for launching (well, 
mpiexec for MPICH(2)) so no parsing of $PBS_NODEFILE. Then pbs_python 
for everything else. So I don't parse qstat/pbsnodes output and this 
means that I don't really miss a pbsfs...

I would understand better a need for it if this would become the 
standard way of providing information for Torque and even Torque tools 
like qstat and pbsnodes would take their information from it instead 
of querying the server. Or if it would become writable too as a way to 
interact with Torque...

-- 
Bogdan Costescu

IWR, University of Heidelberg, INF 368, D-69120 Heidelberg, Germany
Phone: +49 6221 54 8240, Fax: +49 6221 54 8850
E-mail: bogdan.costescu at iwr.uni-heidelberg.de


More information about the torquedev mailing list