Add ASPIRE 1 example config to user guide

[annawoodard]

No description provided.

[annawoodard]

This depends on #908 and #909

[macintoshpie]

My notes on things to configure
pyzmq compilation:

# NOTE: on ASPIRE 1, it is necessary to force compilation of the zmq bindings.
# First, make sure that all versions of pyzmq are removed from the path.
python3 -m pip install --no-binary pyzmq pyzmq

Edit Parsl's torque template b/c it will always run with 1 ppn #669 . Then install your modified Parsl
#PBS -l nodes=${nodes_per_block}:ppn=24

Running parsl in job:
They will kill your parsl process when running on the login node

Message from root@nscc04 on <no tty> at 10:13 ...
This is an automated script.
The python process with PID 2366 has been killed because it exceeded the CPU time limit on the login nodes.
All long running or computationally intensive processes should be submitted to the PBS batch scheduler.
EOF

So you have to run parsl in a job. It can be helpful to make a bash file for running a parsl script - wall time will determine your queue fro what I can tell, and this will put you in medium which is pretty fast if I'm remembering correctly.
qsub runParsl.sh

runParsl.sh

#!/bin/bash

#PBS -q normal
#PBS -P Personal
#PBS -l select=1:ncpus=1:mem=2G
#PBS -l walltime=24:00:00
#PBS -N parsl_head
#PBS -o /home/users/industry/uchicago/tsummer2/outputfiles/parsl_head.o
#PBS -e /home/users/industry/uchicago/tsummer2/errorfiles/parsl_head.e

module load anaconda/3
source activate parsl_py36
python3 myParslScript.py

Another thought, that someone else will have to confirm, is that they limit you to only 2 active jobs (ie parsl and the job it's running).

[annawoodard]

Thanks for summarizing all this for us, @macintoshpie!!! This is super helpful!

Edit Parsl's torque template b/c

Only minor comment is that this shouldn't be necessary after we implement #908.

[benclifford]

@ravihansa3000 provided PR #1582 which I think resolves this