SciPy is a powerful python library, but it has many dependencies including Fortran. So Running your Scipy code in a docker container makes absolute sense. We will use a private registry:
docker run -d -p 5000:5000 --name registry registry:2
I will use a Centos image. Centos is a very popular linux distribution based on RedHat which is a commercial Linux distribution. Oracle’s Linux and Amazon Linux is based on Red Hat Linux.
docker pull centos docker tag centos localhost:5000/centos docker push localhost:5000/centos
Then we start a container
docker run -i -t --name centoscontainer localhost:5000/centos /bin/bash
We install all binary dependencies
yum install -y epel-release yum -y update yum -y groupinstall "Development Tools" yum -y install python-devel yum -y install blas --enablerepo=epel yum -y install lapack --enablerepo=epel yum -y install Cython --enablerepo=epel yum -y install python-pip
Then we install the scipy stack
pip install boto3 pip install numpy pip install pandas pip install scipy
And we are ready. Now we should proceed on committing the image.
docker commit -m 'Added scipy stack' -a "Emmanouil Gkatziouras" 4954f603d93b localhost:5000/scipy docker push localhost:5000/scipy
Now we are ok to run our SciPy enabled container.
docker run -t -i localhost:5000/scipy /bin/bash
Last but not least we clear our registry.
docker stop registry && docker rm -v registry
| Reference: | Implement a SciPy Stack Docker Image from our JCG partner Emmanouil Gkatziouras at the gkatzioura blog. |
