Last active
April 23, 2020 09:55
-
-
Save brynpickering/78ed2b32792cfd73997ff148b941a265 to your computer and use it in GitHub Desktop.
Load Jupyter lab on a ETH euler compute node
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
#!/bin/bash | |
#BSUB -J jupyter_lab[8889] | |
#BSUB -n 1 | |
#BSUB -W 480 | |
#BSUB -R "rusage[mem=80G]" | |
#BSUB -o jupyter-%I.log | |
# get tunnelling info | |
XDG_RUNTIME_DIR="" | |
node=$(hostname -s) | |
port=${LSB_JOBINDEX} | |
module load eth_proxy | |
# Run Jupyter | |
jupyter lab --no-browser --port=${port} --ip=${node} |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
name: jupyter_lab # Although I'm putting this in a new conda env, I generally install these in the base environment. | |
channels: | |
- conda-forge | |
- anaconda | |
dependencies: | |
- python | |
- jupyterlab | |
- nb_conda_kernels # Access all conda environments from one Jupyter lab session. On other conda envs, ensure you have ipython installed | |
- nodejs # to install jupyterlab extensions | |
- widgetsnbextension # On other conda envs, ensure you have ipywidgets installed | |
## Once the environment is installed, run any of the following extension loaders | |
## (also see https://jupyterlab.readthedocs.io/en/stable/user/extensions.html): | |
# jupyter labextension install @jupyter-widgets/jupyterlab-manager | |
# optional: jupyter labextension install @jupyterlab/plotly-extension |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
#!/bin/bash | |
usage="$(basename "$0") [-h] [-p W R] -- program to run jupyter lab on a compute node, for remote access | |
where: | |
-h show this help text | |
-p set the tunnelling port (default = 8889) | |
-W set the runtime, in minutes (default = 960) | |
-R set the memory allocation (default = 'rusage[mem=80G]'). Use 'light' to use a 5 day (1GB) node" | |
set -e | |
# Default values | |
port="8889" | |
time="960" | |
mem="rusage[mem=80G]" | |
while getopts ':hp:W:R:' option | |
do | |
case "$option" in | |
h) echo "$usage" | |
exit | |
;; | |
p) | |
port=${OPTARG} | |
;; | |
W) | |
time=${OPTARG} | |
;; | |
R) | |
mem=${OPTARG} | |
;; | |
esac | |
done | |
jobs=`bjobs -o "JOB_NAME" | sed -n 2p` | |
if [[ $jobs == *"jupyter_lab[${port}]"* ]]; then | |
echo "Jupyter lab instance already running or pending on this port! Exiting..." | |
exit | |
elif [[ $jobs == *"jupyter_lab"* ]]; then | |
echo "Warning: Jupyter Lab already running (${jobs}), be careful of not overwriting notebooks between your two lab instances." | |
fi | |
bsub -J jupyter_lab[${port}] -W ${time} -R ${mem} < jupyter_job.sh | |
until [ "$(bjobs -J jupyter_lab[${port}] -o 'STAT' | tail -1)" = "RUN" ]; | |
do | |
echo "Waiting for job to run..." | |
sleep 5 | |
done | |
user=$(whoami) | |
cluster="euler" | |
echo "Job 'jupyter_lab[${port}]' is running!" | |
node=`bjobs -J jupyter_lab[${port}] -o "EXEC_HOST" | tail -1` | |
# print tunnelling instructions jupyter-log | |
echo -e " | |
Job is running on ${node} | |
Command to create ssh tunnel: | |
> ssh -N -f -L ${port}:${node}:${port} ${user}@${cluster}.ethz.ch | |
Or, since it can be a pain to kill the tunnel: | |
> ssh -N -f -M -S /path/to/temp/file -L ${port}:${node}:${port} ${user}@${cluster}.ethz.ch | |
Then to kill: | |
> ssh -S /path/to/temp/file -O exit ${user}@${cluster}.ethz.ch | |
Where '/path/to/temp/file' is e.g. '/tmp/jupyter_lab_euler' | |
Use a Browser on your local machine to go to: | |
localhost:${port} (prefix w/ https:// if using password) | |
Once the job is complete (timeout, killed, etc.), you can find the log at 'jupyter-${port}.log' | |
" |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment