[Paraview] pvserver submission thru qsub/PBS
Renato Elias
rnelias at gmail.com
Fri Apr 17 11:26:38 EDT 2009
I also use reverse connection with Torque (same commands used by PBS).
Here it goes the script that I use to start my PV session (reverse
connection and offscreen rendering):
#====================================================================
#PBS -l nodes=1:ppn=8
#PBS -l walltime=12:00:00
#PBS -N pv-omp
#PBS -j oe
#PBS -V
# PBS -m ae
# PBS -M user at domain.ufrj.br
PVCLIENT=ip.to.pvclient.machine
export
LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:/sw/shared/x86_64/paraview/openmpi/3.4.0/lib/paraview-3.4:/sw/shared/x86_64/Mesa/7.2/lib64
# change directory
cd ${PBS_O_WORKDIR}
# get the number of processors and nodes
NUM_PROC=`cat ${PBS_NODEFILE} | wc -l`
# mpi program launch
mpirun -np ${NUM_PROC} /sw/shared/x86_64/paraview/openmpi/3.4.0/bin/pvserver
--use-offscreen-rendering -rc -ch=${PVCLIENT}
#====================================================================
Of course, you'll need to leave your pvclient waiting the reverse
connection.
I hope it helps you...
Regards
Renato.
On Fri, Apr 17, 2009 at 11:12 AM, Moreland, Kenneth <kmorel at sandia.gov>wrote:
> For what it’s worth, I use reverse connect all the time to connect a
> server launched with pbs, and it works quite well (although you will have
> problems if your client is behind a firewall). Reverse connection also has
> the added advantage of having the connection happen as soon as the pbs job
> launches; you don’t have to guess how long the job is going to take to
> launch to connect to the server.
>
> -Ken
>
>
>
> On 4/16/09 2:01 PM, "John Patchett" <patchett at lanl.gov> wrote:
>
> I have not tried a reverse connection from a batch job to a waiting
> client, but it sounds like it should work. Have you tried using a
> normal batch script something like:
> #PBS -A myproject
> #PBS -N paraview
> #PBS -j oe
> #PBS -l walltime=00:01:00,size=512
> cd /myparaviewbindir
> mpirun -n 512 ./pvserver --reverse-connection --client-
> host=myheadnode
>
>
> as an aside,
> forwarding X is generally much slower than it has to be.
> My environment makes it very difficult to get the reverse connection
> all the way to my desktop but i can do some port forwarding to get
> port 11111 from my desktop to the back end node.
> ssh -L11111:process0:11111 myheadnode
> which would probably work a lot better/faster than forwarding X
> We have a some machines that have a gateway and then a front end node,
> which gets a lot more complicated, but can still be done with port
> forwarding.
>
> -John.
>
> On Apr 16, 2009, at 1:47 PM, Jack Gundrum wrote:
>
> > John,
> > If the client is running on the login/head node (with X-forwarding
> > to my laptop) and from the client I want to start the parallel
> > backend I would enter the commands in the connect gui to choose a
> > server. Have you ever tried using qsub this way?
> >
>
>
>
> >
> > Patchett wrote:
> >> Hi Jack,
> >> Usually you have to figure out what process 0 is going to be, so
> >> you get an interactive session using something like:
> >> qsub -I -A myproject -q debug -V -lsize=64,walltime=20:00
> >> which I use on one of the machines I use...
> >>
> >> I then do an
> >> mpirun -n 1 hostname
> >> to find out the first node from MPI's perspective so I can connect
> >> to it with a client.
> >>
> >> then:
> >> mpirun -n 64 pvserver
> >>
> >> If you are really trying to do batch I think there is still a
> >> pvbatch.
> >>
> >> -John.
> >> On Apr 16, 2009, at 12:32 PM, Jack Gundrum wrote:
> >>
> >>> Hi,
> >>> I would like to launch pvserver from paraview using a PBS script
> >>> for submitting thru qsub.
> >>> Are there docs or better yet examples that could be shared about
> >>> this? - Thanks
> >>>
> >>> --
> >>> Jack Gundrum
> >>> ITS - Research Computing & Cyberinfrastucture
> >>> Penn State University
> >>>
> >>> _______________________________________________
> >>> Powered by www.kitware.com
> >>>
> >>> Visit other Kitware open-source projects at
> http://www.kitware.com/opensource/opensource.html
> >>>
> >>> Please keep messages on-topic and check the ParaView Wiki at:
> http://paraview.org/Wiki/ParaView
> >>>
> >>> Follow this link to subscribe/unsubscribe:
> >>> http://www.paraview.org/mailman/listinfo/paraview
> >>
> >
> >
> > --
> > Jack Gundrum
> > ITS - Research Computing & Cyberinfrastucture
> > Penn State University
> >
>
> _______________________________________________
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at
> http://www.kitware.com/opensource/opensource.html
>
> Please keep messages on-topic and check the ParaView Wiki at:
> http://paraview.org/Wiki/ParaView
>
> Follow this link to subscribe/unsubscribe:
> http://www.paraview.org/mailman/listinfo/paraview
>
>
>
>
> **** Kenneth Moreland
> *** Sandia National Laboratories
> ***********
> *** *** *** email: kmorel at sandia.gov
> ** *** ** phone: (505) 844-8919
> *** web: http://www.cs.unm.edu/~kmorel<http://www.cs.unm.edu/%7Ekmorel>
>
>
> _______________________________________________
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at
> http://www.kitware.com/opensource/opensource.html
>
> Please keep messages on-topic and check the ParaView Wiki at:
> http://paraview.org/Wiki/ParaView
>
> Follow this link to subscribe/unsubscribe:
> http://www.paraview.org/mailman/listinfo/paraview
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.paraview.org/pipermail/paraview/attachments/20090417/b8c2f716/attachment-0001.htm>
More information about the ParaView
mailing list