Micromamba

Micromamba#

Download and Install micromamba#

cd $WORKDIR
curl -Ls https://micro.mamba.pm/api/micromamba/linux-64/latest | tar -xvj bin/micromamba
./bin/micromamba shell init -s bash  ./micromamba
source ~/.bashrc

Caution

WORKDIR should be a large storage, do not use the HOME directory for installation

On gricad , it could be /bettik/PROJECTS/PROJECT_NAME/USER_NAME

On ige-clusters, it could be /workdir/TEAM_NAME/USER_NAME or /workdir2/TEAM_NAME/USER_NAME

Create an environment#

micromamba create -n myenv python=3.10 -c conda-forge

Activate the environment and install a package#

micromamba activate myenv
micromamba install package_name -c conda-forge

Warning

In a submission job, you will probably add the following to activate your environment

. $WORKDIR/micromamba/etc/profile.d/micromamba.sh
OR
. $WORKDIR/micromamba/etc/profile.d/mamba.sh
and then 
micromamba activate myenv

Warning

On some clusters (for example Dahu), your micromamba process will likely be killed by the system when you try to install environments or packages, because it uses too much memory. In this case, log in to a computation node to perform such tasks.

Example 1: Create R environment and install R packages

# Create an environment with python=3.10
micromamba create -n Renv python=3.10 -c conda-forge

# Activate the environment
micromamba activate Renv

# Install R+ netcdf package
micromamba install r r-base r-essentials -c conda-forge
micromamba install r-ncdf4 -c conda-forge

In order to install the corresponding kernel and use it from a jupyter notebook

Get R version and run R command

install.packages('IRkernel')
# if R version is  R 4.4.2
IRkernel::installspec(name = 'ir442', displayname = 'R 4.4.2')

Once you start a jupyterlab server, an icon will shows up on the right with R 4.4.2 The difference between a simple R file and using this kernel, is that with the kernel, you will be able to run your code cell by cell in a notebook style

Example 2: Create a ferret environment

# Create an environment with python=3.8 (needed for Gricad clusters to get the display)
micromamba create -n pyferret python=3.8  -c conda-forge pyferret ferret_datasets --yes

# Activate the environment
micromamba activate pyferret

Caution

Always create an environment for a single tool/application. This way , if you break an environment for a tool, you will still be able to use the others

Clone an environment#

Create a yaml file to copy an existing environment:

micromamba env export -n myenv > myenv.yaml

Create an environment from the yaml file:

micromamba env create -n myenv-copy  -f  myenv.yaml

Install a kernel on a jupyter notebook#

Once your environment is created it is easy to make it available on a jupyter notebook.

Install the ipykernel package

micromamba activate myenv
micromamba install ipykernel  -c conda-forge

Then make the environment “myenv” available under the name “My First Env”

python -m ipykernel install --name myenv --user --display-name "My First Env"

Restart your jupyter notebook and you will be able to choose your environment My First Env among other available environments

Contents