Skip to content
Snippets Groups Projects
Commit 6230249d authored by Jens Henrik Göbbert's avatar Jens Henrik Göbbert
Browse files

update build instruction for kernels

parent 18bab2db
Branches
No related tags found
No related merge requests found
%% Cell type:markdown id: tags:
![header.png](attachment:9f53dcb1-00d6-4245-955a-b527f1540865.png)
<h5 style="text-align: right">Author: <a href="mailto:j.goebbert@fz-juelich.de?subject=Jupyter-JSC%20documentation">Jens Henrik Göbbert</a></h5>
<h5><a href="../index.ipynb">Index</a></h5>
<h1 style="text-align: center">Create your own Jupyter Kernel</h1>
%% Cell type:markdown id: tags:
Often the standard kernel do not provide all features you need for your work. This might be that certain modules are not loaded or packages are not installed.
With your own kernel you can overcome that problem easily and define your own environment, in which you work.
This notebook shows you how you can build your own kernel for a **python environment**.
<div class="alert alert-block alert-info">
<b>Attention:</b>
This notebook is meant to run out of a JupyterLab on JSC's HPC systems.</br>
</div>
-------------------------
%% Cell type:markdown id: tags:
## Building your own Jupyter kernel is a three step process
1. Create/Pimp new virtual Python environment
* venv
2. Create/Edit launch script for the Jupyter kernel
* kernel.sh
3. Create/Edit Jupyter kernel configuration
* kernel.json
%% Cell type:markdown id: tags:
### Settings
%% Cell type:markdown id: tags:
* Set kernel name
#### Set the kernel name
- must be lower case
- change if you like
%% Cell type:code id: tags:
``` bash
# INPUT NEEDED:
KERNEL_NAME=${USER}_kernel
export KERNEL_NAME=$(echo "${KERNEL_NAME}" | awk '{print tolower($0)}')
echo ${KERNEL_NAME} # double check
```
%% Cell type:markdown id: tags:
* List directories where JupyterLab will search for kernels
%% Cell type:code id: tags:
``` bash
# JUPYTER SEARCH PATH (for kernels-directory)
echo "jupyter search paths for kernels-directories"
if [ -z $JUPYTER_PATH ]; then
echo "$HOME/.local/share/jupyter"
else
tr ':' '\n' <<< "$JUPYTER_PATH"
fi
```
%% Cell type:markdown id: tags:
<div class="alert alert-block alert-info">
<b>Attention:</b>
Please choose 'private kernel' if you are unsure.</br>
Using 'project kernel's need to be enabled for your project first by our Jupyter-JSC admins.
</div>
* Set kernel type
- private kernel = "\${HOME}/.local/"
- project kernel = "\${PROJECT}/.local/"
- other kernel = "\<your-path\>" (ensure it is part of $JUPYTER_PATH or your kernel will not be found by JuypterLab)
#### Set the kernel directory
- check that the kernel name is unique
- print the location of the new kernel
%% Cell type:code id: tags:
``` bash
# INPUT NEEDED:
export KERNEL_TYPE=private # private, project or other
# define KERNEL_SPECS_DIR
export KERNEL_SPECS_PREFIX=${HOME}/.local
###################
# project kernel
if [ "${KERNEL_TYPE}" == "project" ]; then
export KERNEL_SPECS_PREFIX=${PROJECT}/.local
echo "project kernel"
# private kernel
elif [ "${KERNEL_TYPE}" == "private" ]; then
export KERNEL_SPECS_PREFIX=${HOME}/.local
echo "private kernel"
else
if [ ! -d "$KERNEL_SPECS_PREFIX" ]; then
echo "ERROR: please create directory $KERNEL_SPECS_PREFIX"
fi
echo "other kernel"
if [ ! -d "$KERNEL_SPECS_PREFIX" ]; then
echo "ERROR: please create directory $KERNEL_SPECS_PREFIX"
fi
export KERNEL_SPECS_DIR=${KERNEL_SPECS_PREFIX}/share/jupyter/kernels
# check if kernel name is unique
if [ -d "${KERNEL_SPECS_DIR}/${KERNEL_NAME}" ]; then
echo "ERROR: Kernel already exists in ${KERNEL_SPECS_DIR}/${KERNEL_NAME}"
echo " Rename kernel name or remove directory."
fi
echo ${KERNEL_SPECS_DIR}/${KERNEL_NAME} # double check
# print the location of the new kernel
echo ${KERNEL_SPECS_DIR}/${KERNEL_NAME}
```
%% Cell type:markdown id: tags:
* Set directory for kernels virtual environment
- change if you like
#### Set the kernel's virtual environment
- by default it is located at $PROJECT
- print the location of the new kernels virtual environment
%% Cell type:code id: tags:
``` bash
# INPUT NEEDED:
# define KERNEL_VENVS_DIR
export KERNEL_VENVS_DIR=${PROJECT}/${USER}/jupyter/kernels
###################
mkdir -p ${KERNEL_VENVS_DIR}
if [ "${KERNEL_TYPE}" != "private" ] && [ "${KERNEL_TYPE}" != "other" ]; then
echo "Please check the permissions and ensure your project partners have read/execute permissions:"
namei -l ${KERNEL_VENVS_DIR}
fi
echo ${KERNEL_VENVS_DIR} # double check
ls -lt ${KERNEL_VENVS_DIR}
# print the location of the new kernels virtual environment
echo ${KERNEL_VENVS_DIR}
```
%% Cell type:markdown id: tags:
---
%% Cell type:markdown id: tags:
## 1. Create/Pimp new virtual Python environment
%% Cell type:markdown id: tags:
* 1.1 - Load basic Python module
#### 1.1 - Load basic Python module
%% Cell type:code id: tags:
``` bash
module -q purge
module -q use $OTHERSTAGES
module -q load Stages/2020 2> /dev/null # any stage can be used
module -q load GCCcore/.9.3.0 2> /dev/null
module -q load Python/3.8.5 # only Python is required
module list # double check
module -q load Stages/2023 # any stage can be used
module -q load GCC
module -q load Python # only Python is required
```
%% Cell type:code id: tags:
``` bash
# get Python version
export PYV=$(python -c 'import sys; print(".".join(map(str, sys.version_info[:2])))')
echo $PYV
```
%% Cell type:markdown id: tags:
* 1.2 - Load extra modules you need for your kernel
#### 1.2 - Load extra modules you need for your kernel
%% Cell type:code id: tags:
``` bash
# module load <module you need>
```
%% Cell type:markdown id: tags:
* 1.3 - Create and activate a virtual environment for the kernel
#### 1.3 - Create and activate a virtual environment for the kernel
and ensure python packages installed in the virtual environment are always prefered
%% Cell type:code id: tags:
``` bash
if [ -d "${KERNEL_VENVS_DIR}/${KERNEL_NAME}" ]; then
echo "ERROR: Directory for virtual environment already ${KERNEL_VENVS_DIR}/${KERNEL_NAME}"
export VIRTUAL_ENV=${KERNEL_VENVS_DIR}/${KERNEL_NAME}
if [ -d "${VIRTUAL_ENV}" ]; then
echo "ERROR: Directory for virtual environment already ${VIRTUAL_ENV}"
echo " Rename kernel name or remove directory."
else
python -m venv --system-site-packages ${KERNEL_VENVS_DIR}/${KERNEL_NAME}
source ${KERNEL_VENVS_DIR}/${KERNEL_NAME}/bin/activate
export PYTHONPATH=${VIRTUAL_ENV}/lib/python3.8/site-packages:${PYTHONPATH}
python -m venv --system-site-packages ${VIRTUAL_ENV}
source ${VIRTUAL_ENV}/bin/activate
export PYTHONPATH=${VIRTUAL_ENV}/lib/python${PYV}/site-packages:${PYTHONPATH}
echo ${VIRTUAL_ENV} # double check
fi
```
%% Cell type:markdown id: tags:
* 1.4 - Install Python libraries required for communication with Jupyter
#### 1.4 - Install Python libraries required for communication with Jupyter
%% Cell type:code id: tags:
``` bash
which pip
if [ -z "${VIRTUAL_ENV}" ]; then
echo "ERROR: Virtual environment not successfully initialized."
else
pip install --ignore-installed ipykernel
ls ${VIRTUAL_ENV}/lib/python3.8/site-packages/ # double check
pip install ipykernel
fi
```
%% Cell type:markdown id: tags:
* 1.5 - Install whatever else you need in your Python virtual environment (using pip)
#### 1.5 - Install whatever else you need in your Python virtual environment (using pip)
%% Cell type:code id: tags:
``` bash
#pip install <python-package you need>
```
%% Cell type:markdown id: tags:
---
%% Cell type:markdown id: tags:
## 2. Create/Edit launch script for the Jupyter kernel
%% Cell type:markdown id: tags:
* 2.1 - Create launch script, which loads your Python virtual environment and starts the ipykernel process inside:
#### 2.1 - Create launch script, which loads your Python virtual environment and starts the ipykernel process inside:
<div class="alert alert-block alert-info">
<b>Attention:</b>
You MUST load the exactly the same modules as you did above for your virtual Python environment.
</div>
%% Cell type:code id: tags:
``` bash
echo '#!/bin/bash'"
# Load basic Python module
module purge
module use "'$OTHERSTAGES'"
module load Stages/2020
module load GCCcore/.9.3.0
module load Python/3.8.5
module load Stages/2023
module load GCC
module load Python
# Load extra modules you need for your kernel (as you did in step 1.2)
#module load <module you need>
# Activate your Python virtual environment
source ${KERNEL_VENVS_DIR}/${KERNEL_NAME}/bin/activate
source ${VIRTUAL_ENV}/bin/activate
# Ensure python packages installed in the virtual environment are always prefered
export PYTHONPATH=${VIRTUAL_ENV}/lib/python3.8/site-packages:"'${PYTHONPATH}'"
export PYTHONPATH=${VIRTUAL_ENV}/lib/python${PYV}/site-packages:"'${PYTHONPATH}'"
exec python -m ipykernel "'$@' > ${VIRTUAL_ENV}/kernel.sh
chmod +x ${VIRTUAL_ENV}/kernel.sh
cat ${VIRTUAL_ENV}/kernel.sh # double check
```
%% Cell type:markdown id: tags:
---
%% Cell type:markdown id: tags:
## 3. Create/Edit Jupyter kernel configuration
%% Cell type:markdown id: tags:
* 3.1 - Create Jupyter kernel configuration directory and files
#### 3.1 - Create Jupyter kernel configuration directory and files
%% Cell type:code id: tags:
``` bash
python -m ipykernel install --name=${KERNEL_NAME} --prefix ${VIRTUAL_ENV}
export VIRTUAL_ENV_KERNELS=${VIRTUAL_ENV}/share/jupyter/kernels
```
%% Cell type:markdown id: tags:
* 3.2 - Adjust kernel.json file
#### 3.2 - Adjust kernel.json file
%% Cell type:code id: tags:
``` bash
mv ${VIRTUAL_ENV_KERNELS}/${KERNEL_NAME}/kernel.json ${VIRTUAL_ENV_KERNELS}/${KERNEL_NAME}/kernel.json.orig
echo '{
"argv": [
"'${KERNEL_VENVS_DIR}/${KERNEL_NAME}/kernel.sh'",
"-m",
"ipykernel_launcher",
"-f",
"{connection_file}"
],
"display_name": "'${KERNEL_NAME}'",
"language": "python"
"language": "python",
"metadata": {
"debugger": true
}
}' > ${VIRTUAL_ENV_KERNELS}/${KERNEL_NAME}/kernel.json
cat ${VIRTUAL_ENV_KERNELS}/${KERNEL_NAME}/kernel.json # double check
```
%% Cell type:markdown id: tags:
* 3.3 - Create link to kernel specs
#### 3.3 - Create link to kernel specs
%% Cell type:code id: tags:
``` bash
mkdir -p ${KERNEL_SPECS_DIR}
cd ${KERNEL_SPECS_DIR}
ln -s ${VIRTUAL_ENV_KERNELS}/${KERNEL_NAME} .
echo -e "\n\nThe new kernel '${KERNEL_NAME}' was added to your kernels in '${KERNEL_SPECS_DIR}/'\n"
ls ${KERNEL_SPECS_DIR} # double check
```
%% Cell type:markdown id: tags:
* 3.4 - You can select the new kernel in the top right corner of your notebook. The kernel icon will be added to your launcher, once you've restarted the JupyterLab
#### 3.4 - Use the kernel
- You can select the new kernel in the top right corner of your notebook or from JupyterLab's Launchpad
- The kernel icon will be added to your launcher, after a while by JupyterLab automatically or once you've restarted the JupyterLab
%% Cell type:markdown id: tags:
---
%% Cell type:markdown id: tags:
## 4. Cleanup
%% Cell type:code id: tags:
``` bash
deactivate
```
......
0% Loading or .
You are about to add 0 people to the discussion. Proceed with caution.
Please register or to comment