Singularity Container Workflow Part 5: Running the Tensorflow container in WSL2
Now that we have readied the WSL2 environment with Singularity and the relevant CUDA libraries, it’s time to run the sample Keras workflow.
Jupyter Notebook containerized with Singularity
By Staff
Jupyter Notebook is an open source web application environment used to create and share documents containing code, equations, visualizations and narrative text. Use cases include data visualization, statistical modeling and machine learning.
Singularity provides a basis for repeatability with a container runtime that does not require root owned daemon or allow escalation of user privilege when ran. With it’s single file image format (SIF) Singularity is ideal for BYOE science (bring your own environment) and for the mobility of compute with deep learning applications.
We’ll use the following Singularity definition file when building Jupyter:
Bootstrap: library
From: debian:9
%help
Container with Anaconda (Conda 4.5.11) and Jupyter Notebook 5.6.0 for Debian 9.x (Stretch).
This installation is based on Python 2.7.15
%setup
#Create the .condarc file where the environments/channels from conda are specified,
#these are pulled with preference to root
cd /
touch .condarc
%post
#Installing all dependencies
apt-get update && apt-get -y upgrade
apt-get -y install \
build-essential \
wget \
bzip2 \
ca-certificates \
libglib2.0-0 \
libxext6 \
libsm6 \
libxrender1 \
git
rm -rf /var/lib/apt/lists/*
apt-get clean
#Installing Anaconda 2 and Conda 4.5.11
wget -c https://repo.continuum.io/archive/Anaconda2-5.3.0-Linux-x86_64.sh
/bin/bash Anaconda2-5.3.0-Linux-x86_64.sh -bfp /usr/local
#Conda configuration of channels from .condarc file
conda config --file /.condarc --add channels defaults
conda config --file /.condarc --add channels conda-forge
conda update conda
#List installed environments
conda list
%runscript
jupyter notebook --allow-root "$@"To build the image run the following command:
$ sudo singularity build jupyter.sif jupyter.defThe definition file above has a %post section in which all the dependencies are installed at build time. After that, you can start the container and, it will listen on localhost:8888 by default.
$ singularity run jupyter.sifIf you would like to run it on another port (e.g. 9000) instead you can do so by:
$ singularity run jupyter.sif --port=9000This will run Jupyter Notebook on localhost:9000.
And if you would like to set another hostname use ––ip when running the image:
$ singularity run jupyter.sif --ip=1.1.1.1 --port=9000This will run Jupyter Notebook on 1.1.1.1:9000
For the files used in this example, see our GitHub repository at:
https://github.com/sylabs/examples/tree/master/machinelearning/jupyter-notebook
Join Our Mailing List
Related Posts
read more
Singularity Container Workflow: Part 4 – Signing the Container
Signing the Container The Singularity 3.0 family introduced the ability to create (and manage) PGP keys to sign and verify containers. This provides a trusted method for Singularity users to share containers and ensures a bit-for-bit reproduction of the original...
Singularity Container Workflow: Part 3 – Create an Account & Authentication Token
Create an Account & Authentication Token Now that we have SingularityCE installed in WSL2, and NVIDIA GPU support is enabled, we will create a Singularity Container Services account and configure the local Singularity client, followed by building a remote...