Docker workspace and aliases

Docker workspace and aliases#

If Docker as a non-root user is configured:#

To set an interactive working directory (in an ubuntu environment), respectively using Docker and FEniCSx, the following commands can be used:

docker run -ti -v $(pwd):/home/fenics/shared -w /home/fenics/shared th0maslavigne/dolfinx:v0.9.0

Note: If you have several WSL Distros, enable them all in Docker Desktop (if used, in Windows). See Install.md

Remark: On a cluster, singularity is often installed instead of docker. In such cases, based on the image.sif file, the command is similar:

singularity exec /modules/containers/images/dolfinx/dolfinx-0.9.0.sif python3 file.py

To create a jupyter container, compute:

docker run --init -p 8888:8888 -v "$(pwd)":/root/shared --name=jupyter_dolfinx dolfinx/lab:v0.9.0

Then to use it, consider using:

docker container start -i jupyter_dolfinx

Remark: Docker can store the images and therefore fill a huge amount of space which you can purge with:

docker stop `docker ps -qa` > /dev/null 2>&1; docker system prune --volumes --all;

Sometimes a docker image is missing some python library one want to use. A new image based on an existing image can be created including additionnal packages. For instance, the image used for this workshop have been created using the following Dockerfile:

FROM dolfinx/dolfinx:v0.9.0
RUN apt update && \
    apt upgrade -y && \
    apt update && \
    apt install xvfb libgl1-mesa-dri libglx-mesa0 libglu1-mesa mesa-utils -y && \
    python3 -m pip install --upgrade pip && \
    apt update 
   
RUN pip3 install pandas \
         imageio \
         pyvista

Then to build the image, run in the folder where the ‘Dockerfile’ is present:

docker build -f Dockerfile -t FEniCSx:v0.9.0 .

Remark: Be careful, it is sensitive to the case so ensure your file is named ‘Dockerfile’.

You can list your local images using :

docker images

You can tag the images based on their ID:

docker tag ImageID meaningful_name

To save an image:

docker save ImageTag > name.tar

or

docker save -o name.tar ImageTag

Further commands are described in the cheat sheets available in the resources section.

Else, Docker as a non-root user is not configured:#

All the above commands are working but one need to put sudo before.

To set an interactive working directory (in an ubuntu environment), respectively using Docker and FEniCSx, the following commands can be used:

sudo docker run -ti -v $(pwd):/home/fenicsx/shared -w /home/fenicsx/shared th0maslavigne/dolfinx:v0.9.0

To create a jupyter container, compute:

sudo docker run --init -p 8888:8888 -v "$(pwd)":/root/shared --name=jupyter_dolfinx dolfinx/lab:v0.9.0

Then to use it, consider using:

sudo docker container start -i jupyter_dolfinx

And so on. Remark: if Docker as a non-root user is not configured, it is required to put sudo as part of the commands for the aliases defined here-after too.

Create an Alias#

The repeated use of a command can be reduced by the use of aliases (see create an alias fot linux). The idea of the alias is to execute commands from a meaningful name by introducing them in a sh ~/.bash_aliases.

To access this file, run:

gedit ~/.bash_aliases

It will open a window in which you can write your aliases on the following basis:

alias <Meaningful_name>='<the command>'

For example, in the present workshop, one could create:

alias fenicsx_v0_9_0='docker run -ti -v $(pwd):/home/fenicsx/shared -w /home/fenicsx/shared th0maslavigne/dolfinx:v0.9.0'

This is of interest and will allow you to have different versions of a same environment without conflicting package:

alias fenicsx_v0_6_0='docker run -ti -v $(pwd):/home/fenicsx/shared -w /home/fenicsx/shared dolfinx/dolfinx:v0.6.0'
alias fenicsx_v0_7_3='docker run -ti -v $(pwd):/home/fenicsx/shared -w /home/fenicsx/shared dolfinx/dolfinx:v0.7.3'
alias fenicsx_v0_8_0='docker run -ti -v $(pwd):/home/fenicsx/shared -w /home/fenicsx/shared dolfinx/dolfinx:v0.8.0'

If you prefer using the jupyter lab notebooks, the following commands can be created: Run once:

alias create_fenicsx_v0_8_0_jupyter='docker run --init -p 8888:8888 -v "$(pwd)":/root/shared --name=jupyter_dolfinx08 dolfinx/lab:v0.8.0'

or

alias create_fenicsx_v0_9_0_jupyter='docker run --init -p 8888:8888 -v "$(pwd)":/root/shared --name=jupyter_dolfinx09 dolfinx/lab:v0.9.0'

Then use:

alias run_fenicsx_v0_8_0_jupyter='docker container start -i jupyter_dolfinx08'
alias stop_fenicsx_v0_8_0_jupyter='docker stop jupyter_dolfinx08'
alias logs_fenicsx_v0_8_0_jupyter='docker logs jupyter_dolfinx08'

or

alias run_fenicsx_v0_9_0_jupyter='docker container start -i jupyter_dolfinx09'
alias stop_fenicsx_v0_9_0_jupyter='docker stop jupyter_dolfinx09'
alias logs_fenicsx_v0_9_0_jupyter='docker logs jupyter_dolfinx09'

Remark: Including the bash term at the end allows to exit the python environnment to the linux command. Here are two examples respectively for fenics legacy and pymesh:

alias fenics2019='docker run -ti -v $(pwd):/home/fenics/shared -w /home/fenics/shared pymor/fenics_py3.9 bash'
alias pymesh='docker run -ti -v $(pwd):/home/pymesh/shared -w /home/pymesh/shared pymesh/pymesh bash'
Contents