🎓 Intranet

Install CASM 0.3.x in a docker container

3 min read

Running CASM in Docker Container

  • The purpose of this note is to install CASM on some new Linux distributions where some old libraries are needed but don’t exist
  • Here we use docker to build the image because it is compatible with Vscode plugin for software development
  • Docker needs root privileges, we need to install a rootless version

Install Rootless version of Docker

  • Run below will install docker in ${HOME}/bin
curl -fsSL https://get.docker.com/rootless | sh
  • You also need to change ~/.bashrc as suggested by the output of the above script, shown below
  • YOUR_USER_ID can be obtained using id
  • YOUR_USER_NAME is your username
export DOCKER_HOST=unix:///run/user/YOUR_USER_ID/docker.sock
export PATH=/home/YOUR_USER_NAME/bin:$PATH
  • Then you need to source ~/.bashrc

Configure an image using a Dockerfile

  • You can import CASM 0.3.x directly from my Docker Hub channel dengzeyu/casm.0.3.x:latest
  • For the CASM with source code stored in /app/CASMcode, you can pull dengzeyu/casm.0.3.x:latest-dev
  • CASM source code is in /app/CASMcode and Miniconda is installed in /opt/conda
  • The Dockerfile below will install Miniconda 3, and CASM

Build a CASM image for development

  • You can do development of CASM, e.g. changing the source code, based on the version on my Docker Hub channel
  • Prepare a Dockerfile:casm_dev.dockerfile
# syntax=docker/dockerfile:1
FROM dengzeyu/casm.0.3.x:latest
 
RUN apt-get update && \
    apt-get install -y vim make gcc autoconf m4 bash-completion libtool pkg-config automake  && \
    apt-get clean && \
    rm -rf /var/lib/apt/lists/*
 
WORKDIR /app
 
# Make RUN commands use `bash --login`:
SHELL ["/bin/bash", "--login", "-c"]
 
# Install CASM dependencies
RUN  conda activate casm_dev && conda remove casm casm-python casm-cpp --force
RUN git clone https://github.com/caneparesearch/CASMcode.git /app/CASMcode
WORKDIR /app/CASMcode
ENV CASM_DIRTY_IS_OK=1
RUN conda activate casm_dev && bash build.sh
RUN make install
RUN conda activate casm_dev && pip install python/casm && bash clean.sh && conda clean -a -y
WORKDIR /
  • Build this image name as casm_dev
docker build ./ -f casm_dev.dockerfile -t casm_dev

Development using VS Code

  • Use vscode-remote and ssh to a machine, you need to configure the machine in ~/.ssh/config
  • Install docker in the remote machine
  • Start a container from an image
  • Then right-click the container and use Attach Visual Studio Code to enter the container

Create a container

  • More details can be found here: https://docs.docker.com/engine/reference/commandline/run/
  • Every time you execute docker run you will create a new container
  • -d run in detachable mode (background)
  • -i run in interactive mode
  • -t is related to pseudo-TTY
  • --name is the name of this container
  • -v mount /home/jerry outside the container to /home/jerry inside the container
  • dengzeyu/casm.0.3.x:latest is the image that we want to run
  • --rm automatically remove this container when it is stopped
docker run -itd -v /home/jerry:/home/jerry --name casm_dev_container dengzeyu/casm.0.3.x:latest

Run a command in container outside of container interactively

  • Here I’m running casm init command in a conda environment casm_dev inside a container casm_dev_container in the current path outside of container: $PWD
  • The command works only when you type this in the shell interactively
docker exec -it -w $PWD casm_dev_container conda run -n casm_dev casm init

Run a command in container outside of container in a script

#!/bin/bash
CASM="docker exec -w $PWD casm_dev_container conda run -n casm_dev casm"
PYTHON="docker exec -w $PWD casm_dev_container conda run -n casm_dev python"
$CASM monte -s monte.json >> stdout.txt
$PYTHON ../../analysis.py

Graph View