MILADY Docker

User guide

Milady users can run the code without compilation thanks to the pre-built image we provide. Depending on your environment (workstation, cluster, supercomputer), you will have to rely on different tools for running the image:

• Docker Engine (on workstations)
• Singularity (on shared resources, for rootless execution)
• pcocc (CEA alternative to Singularity, available on TGCC resources)

Locally, using Docker

You can pull and run the pre-built image using the following commands:

1. Pull the image from Docker Hub:

docker pull aallera/milady:latest

2. Run the container with the current directory mounted as a volume (of course without compiling the code):

docker run -it --rm -v "$(pwd):/workspace" aallera/milady

The option -it enables the interactive mode with a pseudo-TTY session, allowing to kill the process before it ends (ctrl-C).

Pass extra parameters to mpirun

Optionally, additional options to mpirun such as the number of processors (default is -np 4) can be passed as:

docker run -it --rm -v "$(pwd):/workspace" aallera/milady -np 8

Run an interactive shell in the container

docker run -it --rm -v "$(pwd):/workspace" aallera/milady shell

On Irene Joliot-Curie (CEA-TGCC) using pcocc

Locally (takes a long time):

docker save milady:<tag> | gzip > milady_<tag>.tar.gz

On Irene (after transfering the created file):

gzip -d milady_latest.tar.gz
pcocc image import docker-archive:milady_latest.tar milady

Interactive use:

# (optional) ccc_mprun  -p skylake -m scratch,work -s
pcocc run --mount "$(pwd):/workspace" -s -I milady -- -np 16

Submit as a job:

#!/bin/bash
#MSUB -n 4
#MSUB -c 8
#MSUB -q skylake
#MSUB -T 600
#MSUB -A <your submission group>
pcocc run -n ${BRIDGE_MSUB_NPROC}  -I my_docker_image [arg1, ...]

On Jean-Zay (IDRIS) using Singularity

1. Start a pre/post session to access a large RAM node (singularity build takes more ram than the standard 5 Go).

srun --pty --nodes=1 -A yxs@cpu --ntasks-per-node=1 --cpus-per-task=10 --partition=prepost --hint=nomultithread --time=01:00:00 bash

2. Build singularity image (.sif)

singularity build milady.sif docker://aallera/milady:1.0.1

3. Add to allowed directory

idrcontmgr cp milady.sif

4. Start a shell, with access to $WORK. Other directories are read-only.

singularity shell $SINGULARITY_ALLOWED_DIR/milady.sif --bind $WORK/:/MLD/work

Developer: Build the Docker Image

Build the image

1. Clone the current repository
2. Clone the milady repository inside it

./clone_milady.sh

3. Build the Docker image using the following command:

docker build -t milady .

This command will compile the MILADY code during the image build process.

Push the Docker Image to Docker Hub

1. Create a free account on Docker Hub (https://hub.docker.com/) if you don’t have one already.
2. Log in to your Docker Hub account on your machine using the following command:

docker login

3. Tag your local image with your Docker Hub username and a custom name for the image:

docker tag milady YOUR_DOCKERHUB_USERNAME/milady:<tag>

Replace YOUR_DOCKERHUB_USERNAME with your actual Docker Hub username and <tag> with a version.

4. Push the tagged image to Docker Hub:

docker push YOUR_DOCKERHUB_USERNAME/milady:<tag>