== Parallel !ParaView on JURECA
The aim of this instruction is to explain how the following scenario can be set up:
- !ParaView client (GUI) running inside a VNC session on one of JURECAs visualization login nodes
- !ParaView servers (pvserver) running on an couple of visualization batch nodes


=== 1. Establish a VNC connection to JURECA visualization login nodes
This step is described in detail [wiki:vnc3d/manual here]. On this page, please follow instruction branch 2.a., which is about connecting to a visualization login node.

=== 2. Open a !ParaView client in the VNC server
Once the VNC connection is established, open a terminal on the VNC desktop of JURECA and start the !ParaView client by the following script:

{{{ #!sh
module purge
module use /usr/local/software/jureca/OtherStages
module load Stages/Devel-2018b
module load GCC
module load ParaStationMPI
module load ParaView/5.6.0-OSPRay-1.8.2-Python-3.6.6
vglrun paraview
}}}

The !ParaView graphical user interface (GUI) should show up on the screen.

=== 3. Start !ParaView servers on visualization batch nodes
Here we start pvserver on 4 visualization batch nodes (partition 'vis'). To do this, first create (not execute!) the following slurm batch file, e.g. with name '''pvserver.slurm''':
{{{ #!sh
#!/bin/bash
srun -n 4 --cpu_bind=none vglrun pvserver --disable-xdisplay-test --force-offscreen-rendering --mpi
}}}

Now start the batch job via this script (please fill in the name of your contingent beforehand):
{{{ #!sh
module purge
module use /usr/local/software/jureca/OtherStages
module load Stages/Devel-2018b
module load GCC
module load ParaStationMPI
module load ParaView/5.6.0-OSPRay-1.8.2-Python-3.6.6

export DISPLAY=:0.0
sbatch --account=YOUR_CONTINGENT --partition=vis --nodes=4 --time=24:00:00 --gres=mem512,gpu:2 pvserver.slurm
}}}

After you submitted the job, the following output should be seen: "Submitted batch job ''1234567 (some number)''". Wait for the job to start. You can check the status of the job by
{{{ #!sh
squeue -u YOUR_ACCOUNT_NAME
}}}
When the job is running, a slurm output file is created, lets say the filename is '''slurm-1234567.out'''. Look at the last lines of this file and notice the name of the pvserver head node, e.g.:
{{{ #!sh
Waiting for client...
Connection URL: cs://jrc1390:11111
Accepting connection(s): jrc1390:11111
}}}
This tells you, that your !ParaView client should connect to node '''jrc1390''

=== 4. Connect ParaView client to server
Now enter your !ParaView client GUI and click on '''connect'''.
If not already done, add jrc1390 with port 11111 to the list of known servers. Then click on "Connect". If everything works, you should now have a session with 4 !ParaView servers running in parallel. You can check this e.g. by looking at the "Memory Inspector" in the GUI of the !ParaView client.

Once you disconnect the !ParaView client from the server, the batch job should be terminated. You can check this by "squeue -u YOUR_ACCOUNT_NAME".