BinAC/Software/Jupyterlab: Difference between revisions

From bwHPC Wiki
Jump to navigation Jump to search
No edit summary
No edit summary
 
(9 intermediate revisions by the same user not shown)
Line 21: Line 21:


JupyterLab is a web-based interactive development environment for notebooks, code, and data.
JupyterLab is a web-based interactive development environment for notebooks, code, and data.

Currently BinAC provides the following [https://jupyter-docker-stacks.readthedocs.io/en/latest/using/selecting.html#jupyter-minimal-notebook JupyterLab Docker images] via Apptainer:

* minimal-notebook
* r-notebook


= Usage =
= Usage =


This guide is valid for for <code>minimal-notebook</code>. You can follow the guide also for <code>r-notebook</code>, but you have to use <code>r-notebook.pbs.template</code> as template for your jobscript.
== Start Jupyterlab ==

== Start JupyterLab ==


The module provides a job script for starting a JupyterLab instance on the BinAC <code>inter</code> queue.
The module provides a job script for starting a JupyterLab instance on the BinAC <code>inter</code> queue.
Line 44: Line 51:
Please note the restrictions of the inter queue:
Please note the restrictions of the inter queue:
* max. walltime: 12 hours
* max. walltime: 12 hours
* max. nodes: 1
* max. cores: 28
* max. cores: 28
* max jobs per user: 1
* max jobs per user: 1
Line 55: Line 63:
== Create SSH tunnel ==
== Create SSH tunnel ==


The compute node on which JupyterLab is running is not reachable from your workstation.
The compute node on which JupyterLab is running is not reachable directly from your workstation.
Hence you have to create an SSH tunnel from your workstation to the compute node through a BinAC login node.
Hence you have to create an SSH tunnel from your workstation to the compute node through a BinAC login node.


Line 65: Line 73:
</pre>
</pre>


[[File:Binac_jupyterlab_connection_details.png | 800px | center | Jupyterlab connection info]]
[[File:Binac_jupyterlab_connection_details.png | 800px | center | JupyterLab connection info]]


=== Linux Users ===
=== Linux Users ===
Line 78: Line 86:
=== Windows Users ===
=== Windows Users ===


If you are using Windows you will need to create the SSH tunnel in the SSH client of your choice (e.g. MobaXTerm, PuTTY, etc.).
TODO

==== MobaXTerm ====

Select <code>Tunneling</code> in the top ribbon. Then press <code>New SSH tunnel</code>.
Then configure the SSH tunnel with the correct values taken the SSH tunnel infos above.
For the example in this tutorial it looks as follows:

[[File:Binac_jupyterlab_mobaxterm.png | 800px | center ]]


== Access JupyterLab ==
== Access JupyterLab ==
Line 90: Line 106:


[[File:Binac_jupyterlab_browser_lab.png | 800px | center ]]
[[File:Binac_jupyterlab_browser_lab.png | 800px | center ]]

== Access <code>/beegfs/work/</code> in file browser ==

Jupyterlab's root directory will be your home directory. As your home directory is backuped daily you may want to store your notebooks there.

In order to access data in your workspace (e.g. somewhere under <code>/beegfs/work</code>) via the file browser you will need to create a symbolic link from your home directory to you workspace:

<pre>
ln -s /beegfs/work/<path to your project data> $HOME/<link name>
</pre>

Through that link in your home directory you can move around your research data in Jupyterlab's file explorer.

Here is an example how I linked to my directory:

[[ File:Binac_jupyterlab_link.png | 800px | center ]]



== Shut Down JupyterLab ==
== Shut Down JupyterLab ==
Line 96: Line 129:
Please note that this will also terminate your compute job on the BinAC!
Please note that this will also terminate your compute job on the BinAC!


[[File:Binac_jupyterlab_browser_shutdown.png | 800px | center ]]
[[ File:Binac_jupyterlab_browser_shutdown.png | 800px | center ]]


== Add new Kernels ==
= Managing Kernels =


The kernels are stored in your Home directory on BinAC: <code>/$HOME/.local/share/jupyter/kernels/</code>.
TODO
You can install new kernels from within the JupyterLab browser window, but you will have to install Miniconda beforehand.
With Miniconda available, open a new terminal window.

== Add a new Kernel ==

=== Python ===

There is only a Python 3 kernel installed when you first start JupyterLab.
Because there are nearly endless combinations of Python versions and packages we encourage you to install the software yourself via Conda.

This is an example how you create new kernels for Jupyterlab. It's so simple that three commmands suffice:

<pre>
conda create --name kernel_env python=3.8 pandas numpy matplotlib ipykernel # 1
conda activate kernel_env # 2
python -m ipykernel install --user --name pandas --display-name="Python 3.8 (pandas)" # 3

# Installed kernelspec pandas in /home/tu/tu_tu/tu_iioba01/.local/share/jupyter/kernels/pandas
</pre>

The first command creates a new Conda environment called <code>kernel_env</code> and installs a specific Python packages plus a few Python packages. It's important that you also install <code>ipykernel</code>. We need <code>ipykernel</code> later to create the JupyterLab kernel.

The second command activates the <code>kernel_env</code> Conda environment.
The third command creates the new JupyterLab kernel.

<pre>
$ ls -lh $HOME/.local/share/jupyter/kernels/
total 0
drwxr-xr-x 2 tu_iioba01 tu_tu 109 Jul 26 10:38 pandas
</pre>

[[ File:Binac_jupyterlab_new_kernel.png | 800px | center ]]

=== R ===

The instructions for new R-Kernels are a bit different.

<pre>
conda config --add channels r
conda create --name r_kernel_env r-base=4.4.1 r-irkernel
conda activate r_kernel_env
R
# In the R-Session:
install.packages(...)
IRkernel::installspec(name = 'ir44', displayname = 'R 4.4.1')
</pre>

The first command creates a new Conda environment called <code>r_kernel_env</code> and installs a specific R version. It's important that you also install <code>r-irkernel</code>. We need <code>r-irkernel</code> later to create the JupyterLab kernel.
The second command activates the <code>r_kernel_env</code> Conda environment and open an R session. In this session you can install whatever R-package you need in your kernel.
Last, create the new kernel with the <code>installspec</code> command.

== Remove a Kernel ==

In order to remove a kernel from Jupyterlab, simply remove the corresponding directory in <code>/$HOME/.local/share/jupyter/kernels/</code>:

<pre>
# Remove the JupyterLab kernel installed in the previous example
rm -rf /$HOME/.local/share/jupyter/kernels/pandas
</pre>

Also remove the corresponding Conda environment if you don't need it any more:

<pre>
conda env remove --name kernel_env
</pre>

Latest revision as of 16:15, 31 July 2024

The main documentation is available via module help devel/jupyterlab on the cluster. Most software modules for applications provide working example batch scripts.


Description Content
module load devel/jupyterlab
License JupyterLab License
Links Homepage
Graphical Interface Yes

Description

JupyterLab is a web-based interactive development environment for notebooks, code, and data.

Currently BinAC provides the following JupyterLab Docker images via Apptainer:

  • minimal-notebook
  • r-notebook

Usage

This guide is valid for for minimal-notebook. You can follow the guide also for r-notebook, but you have to use r-notebook.pbs.template as template for your jobscript.

Start JupyterLab

The module provides a job script for starting a JupyterLab instance on the BinAC inter queue. Load the module and copy the job script into your workspace:

module load devel/jupyterlab/7.2.1
cp $JUPYTERLAB_EXA_DIR/jupyterlab.pbs.template jupyterlab.pbs

You can adjust the following settings in the job script according to your needs.

#PBS -l nodes=1:ppn=1          # adjust the number of cpu cores (ppn)
#PBS -l mem=2gb
#PBS -l walltime=6:00:00

Please note the restrictions of the inter queue:

  • max. walltime: 12 hours
  • max. nodes: 1
  • max. cores: 28
  • max jobs per user: 1

Then submit the job.

jobid=$(qsub jupyterlab.pbs)

Create SSH tunnel

The compute node on which JupyterLab is running is not reachable directly from your workstation. Hence you have to create an SSH tunnel from your workstation to the compute node through a BinAC login node.

The job's standard output file (Jupyterlab.<jobid>) contains the SSH command for this tunnel. Please note that details like IP, port number, and access URL will vary.

cat JupyterLab.o${jobid}
JupyterLab connection info

Linux Users

Copy the ssh -N -L ... command and execute it in a shell on your workstation. After a successfull authentication the SSH tunnel is ready to use. The ssh command does not return a result. If there is no error message everything should be fine:

Creation of SSH tunnel on Linux

Windows Users

If you are using Windows you will need to create the SSH tunnel in the SSH client of your choice (e.g. MobaXTerm, PuTTY, etc.).

MobaXTerm

Select Tunneling in the top ribbon. Then press New SSH tunnel. Then configure the SSH tunnel with the correct values taken the SSH tunnel infos above. For the example in this tutorial it looks as follows:

Binac jupyterlab mobaxterm.png

Access JupyterLab

JupyterLab is now running on a BinAC compute node and you created an SSH tunnel from your workstation to that compute node. Open a browser and copy the URL with the access token into the address field:

Binac jupyterlab browser url.png

Your browser should now display the JupyterLab user interface:

Binac jupyterlab browser lab.png

Access /beegfs/work/ in file browser

Jupyterlab's root directory will be your home directory. As your home directory is backuped daily you may want to store your notebooks there.

In order to access data in your workspace (e.g. somewhere under /beegfs/work) via the file browser you will need to create a symbolic link from your home directory to you workspace:

ln -s /beegfs/work/<path to your project data> $HOME/<link name>

Through that link in your home directory you can move around your research data in Jupyterlab's file explorer.

Here is an example how I linked to my directory:

Binac jupyterlab link.png


Shut Down JupyterLab

You can shut down JupyterLab via File -> Shut Down. Please note that this will also terminate your compute job on the BinAC!

Binac jupyterlab browser shutdown.png

Managing Kernels

The kernels are stored in your Home directory on BinAC: /$HOME/.local/share/jupyter/kernels/. You can install new kernels from within the JupyterLab browser window, but you will have to install Miniconda beforehand. With Miniconda available, open a new terminal window.

Add a new Kernel

Python

There is only a Python 3 kernel installed when you first start JupyterLab. Because there are nearly endless combinations of Python versions and packages we encourage you to install the software yourself via Conda.

This is an example how you create new kernels for Jupyterlab. It's so simple that three commmands suffice:

conda create --name kernel_env python=3.8 pandas numpy matplotlib ipykernel             # 1
conda activate kernel_env                                                               # 2
python -m ipykernel install --user --name pandas --display-name="Python 3.8 (pandas)"   # 3

# Installed kernelspec pandas in /home/tu/tu_tu/tu_iioba01/.local/share/jupyter/kernels/pandas

The first command creates a new Conda environment called kernel_env and installs a specific Python packages plus a few Python packages. It's important that you also install ipykernel. We need ipykernel later to create the JupyterLab kernel.

The second command activates the kernel_env Conda environment. The third command creates the new JupyterLab kernel.

$ ls -lh $HOME/.local/share/jupyter/kernels/
total 0
drwxr-xr-x 2 tu_iioba01 tu_tu 109 Jul 26 10:38 pandas
Binac jupyterlab new kernel.png

R

The instructions for new R-Kernels are a bit different.

conda config --add channels r
conda create --name r_kernel_env r-base=4.4.1 r-irkernel
conda activate r_kernel_env 
R
# In the R-Session:
install.packages(...)
IRkernel::installspec(name = 'ir44', displayname = 'R 4.4.1')

The first command creates a new Conda environment called r_kernel_env and installs a specific R version. It's important that you also install r-irkernel. We need r-irkernel later to create the JupyterLab kernel. The second command activates the r_kernel_env Conda environment and open an R session. In this session you can install whatever R-package you need in your kernel. Last, create the new kernel with the installspec command.

Remove a Kernel

In order to remove a kernel from Jupyterlab, simply remove the corresponding directory in /$HOME/.local/share/jupyter/kernels/:

# Remove the JupyterLab kernel installed in the previous example
rm -rf /$HOME/.local/share/jupyter/kernels/pandas

Also remove the corresponding Conda environment if you don't need it any more:

conda env remove --name kernel_env