1. Running Jupyter notebook on TSCC

Create a script, say "jupter_notebook", with the following contents on TSCC.

#!/bin/bash
#PBS -q condo
#PBS -N notebook
#PBS -l nodes=1:ppn=1:ib
#PBS -l walltime=8:00:00
#PBS -V
#PBS -A YOUR-group-account

cd $PBS_O_WORKDIR

# choose your own unique port between 8000 and 9999
NOTEBOOKPORT=8181

# setup tunnel between computing and login node
ssh -N -f -R $NOTEBOOKPORT:localhost:$NOTEBOOKPORT $PBS_O_HOST

# launch the notebook
jupyter notebook --port=$NOTEBOOKPORT --no-browser > notebook.log

Create a script, say "tunnel_notebook", with the following contents at your laptop

ssh -f -N -L 8181:localhost:8181 username@tscc-login$1.sdsc.edu

Open a terminal, and run "bash tunnel_notebook login_node" , here login_node is a number. For instance, [username@tscc-login2 ~], login_node=2

Open ur favorite web browser, e.g Chrome. Type "localhost:8181" in the address. It will open the jupyter notebook that you do data analysis on TSCC.

2. Running Jupyter notebook on Comet

Create a script, say "jupter_notebook", with the following contents on Comet.

#!/bin/bash
#SBATCH --nodes 1

#### just 1 core (5.3GB RAM)
#SBATCH --ntasks-per-node 1
#SBATCH --partition shared  

#### If you need a large memory(with maximum of 1.5T) uncomment below two lines
####SBATCH -p large-shared
####SBATCH --mem=10G

#SBATCH --time  8:00:00
#SBATCH --job-name notebook
#SBATCH --output notebook.log

# choose your own unique port between 8000 and 9999
NOTEBOOKPORT=8181

# setup tunnel between computing and login node
ssh -N -f -R $NOTEBOOKPORT:localhost:$NOTEBOOKPORT $SLURM_SUBMIT_HOST

# launch the notebook
ipython notebook --port=$NOTEBOOKPORT --no-browser

Create a script, say "tunnel_notebook", with the following contents at your laptop

ssh -f -N -L 8181:localhost:8181 username@comet-ln$1.sdsc.edu

Open a terminal, and run "bash tunnel_notebook login_node" , here login_node is a number. For instance, [username@comet-ln3 ~], login_node=3

Open ur favorite web browser, e.g Chrome. Type "localhost:8181" in the address. It will open the jupyter notebook that you do data analysis on TSCC.

In [ ]:
 

3. Running Jupyter notebook on Cori

Go go "https://jupyter-dev.nersc.gov" to login and use the notebook