Skip to content

Latest commit

 

History

History
787 lines (525 loc) · 27.4 KB

installation.rst

File metadata and controls

787 lines (525 loc) · 27.4 KB

Installation

Note

ESMValTool now uses mamba instead of conda for the recommended installation. For more information about the change, have a look at :ref:`Move to Mamba<move-to-mamba>`.

ESMValTool supports Python 3.10 and later and requires Linux or MacOS. Successful usage on Windows has been reported by following the Linux installation instructions with WSL.

ESMValTool can be installed in multiple ways.

Recommended installation method:

Install the :ref:`mamba package manager <install_with_mamba>` and then follow the instructions for

Further options for installation are:

The next sections will detail the procedure to install ESMValTool through each of these methods.

There is also a lesson available in the ESMValTool tutorial that describes the installation of the ESMValTool in more detail. It can be found here.

See common installation issues if you run into trouble.

Mamba/Conda installation

In order to install ESMValTool and its dependencies from conda-forge, you will first need to install the mamba package manager. We recommend using mamba as a package manager for your conda environments instead of conda because it is much faster, see move-to-mamba for more information.

For a minimal mamba installation (recommended) go to https://mamba.readthedocs.io/en/latest/installation.html.

Note

It is recommended that you always use the latest version of mamba, as problems have been reported when trying to use older versions.

Note

Some systems provide a pre-installed version of conda or mamba (e.g. via the module environment). However, several users reported problems when installing with such versions. It is therefore preferable to use a local, fully user-controlled mamba installation.

First download the installation file for Linux or MacOSX. After downloading the installation file from one of the links above, execute it by running (Linux example):

bash Miniforge3-Linux-x86_64.sh

and follow the instructions on your screen.

Note

Make sure to choose an installation location where you have at least 10 GB of disk space available.

During installation, mamba will ask you if you want mamba to be automatically loaded from your .bashrc or .bash-profile files. It is recommended that you answer yes. If you answered no, you can load the correct paths and environment variables later by running:

source <prefix>/etc/profile.d/conda.sh

where <prefix> is the installation location of mamba (e.g. /home/$USER/miniforge3 if you chose the default installation path).

If you use another shell than Bash, have a look at the available configurations in the <prefix>/etc/profile.d directory.

You can check that mamba installed correctly by running

which mamba

this should show the path to your mamba executable, e.g. ~/miniforge3/bin/mamba.

It is recommended to update both mamba and conda after installing:

mamba update --name base mamba conda

ESMValTool installation on Linux

Once you have installed the mamba package manager, you can install the entire ESMValTool package by running:

mamba create --name esmvaltool esmvaltool

It is also possible to install just a subset of the ESMValTool dependencies by installing one or more of the :ref:`subpackages <conda subpackages>` described in the next section.

The command above will create a new conda environment called esmvaltool, and install ESMValTool in it. Of course it is also possible to choose a different name than esmvaltool for the environment.

Note

Creating a new conda environment is often much faster and more reliable than trying to update an existing conda environment. Therefore it is recommended that you create a new environment when you want to upgrade to the latest version.

The next step is to check that the installation works properly.

First activate the environment with the command:

conda activate esmvaltool

and then run the tool with the command:

esmvaltool --help

If everything was installed properly, ESMValTool should have printed a help message to the console.

Installation of subpackages

The diagnostics bundled in ESMValTool are scripts in four different programming languages: Python, NCL, R, and Julia.

There are three language specific packages available:

  • esmvaltool-ncl
  • esmvaltool-python
  • esmvaltool-r

The main esmvaltool package contains all three subpackages listed above. For the Julia dependencies, there is no subpackage yet, but there are special :ref:`installation instructions <install_julia_dependencies>`. If you only need to run a recipe with diagnostics in some of these languages, it is possible to install only the dependencies needed to do just that. The diagnostic script(s) used in each recipe, are documented in :ref:`recipes`. The extension of the diagnostic script can be used to see in which language a diagnostic script is written (.py for Python, .ncl for NCL, .R for R, and .jl for Julia diagnostics).

To install support for diagnostics written in Python and NCL into an existing environment, run

mamba install esmvaltool-python esmvaltool-ncl

Some of the CMORization scripts are written in Python, while others are written in NCL. Therefore, both esmvaltool-python and esmvaltool-ncl need to be installed in order to be able to run all CMORization scripts.

Note that the ESMValTool source code is contained in the esmvaltool-python package, so this package will always be installed as a dependency if you install one or more of the packages for other languages.

Installation of Julia dependencies

If you want to use the ESMValTool Julia functionality, you will also need to install Julia. If you are just getting started, we suggest that you come back to this step later when, and if you need it. To perform the Julia installation, make sure that your conda environment is activated and then execute

mamba install julia
esmvaltool install Julia

ESMValTool installation on MacOS

The Python diagnostics of the ESMValTool are supported on MacOS, but Julia, NCL, and R are not. If any of these are needed, deployment through a :ref:`Docker<install_with_docker>` container is advised.

The esmvaltool-python diagnostics can be installed as follows:

First, ensure mamba is installed (see install_with_mamba for more details).

Create a new environment with the esmvaltool-python package:

mamba create --name esmvaltool esmvaltool-python

Activate the new environment:

conda activate esmvaltool

Confirm that the ESMValTool is working with:

esmvaltool --help

Note that some recipes may depend on the OpenMP library, which does not install via mamba on MacOS. To install this library, run:

brew install libomp

to install the library with Homebrew. In case you do not have Homebrew, follow installation instructions here.

Install from source

Installing the tool from source is recommended if you need the very latest features or if you would like to contribute to its development.

Obtaining the source code

The ESMValTool source code is available on a public GitHub repository: https://github.com/ESMValGroup/ESMValTool

The easiest way to obtain it is to clone the repository using git (see https://git-scm.com/). To clone the public repository:

git clone https://github.com/ESMValGroup/ESMValTool

or

git clone git@github.com:ESMValGroup/ESMValTool

if you prefer to connect to the repository over SSH.

The command above will create a folder called ESMValTool containing the source code of the tool in the current working directory.

Note

Using SSH is much more convenient if you push to the repository regularly (recommended to back up your work), because then you do not need to type your password over and over again. See this guide for information on how to set it up if you have not done so yet. If you are developing ESMValTool on a shared compute cluster, you can set up SSH agent forwarding to use your local SSH keys also from the remote machine.

It is also possible to work in one of the ESMValTool private repositories, e.g.:

git clone https://github.com/ESMValGroup/ESMValTool-private

GitHub also allows one to download the source code in as a tar.gz or zip file. If you choose to use this option, download the compressed file and extract its contents at the desired location.

Install dependencies

It is recommended to use mamba to manage ESMValTool dependencies. See the :ref:`mamba installation instructions <install_with_mamba>` at the top of this page for instructions on installing mamba. To simplify the installation process, an environment definition file is provided in the repository (environment.yml in the root folder).

The ESMValTool conda environment file can also be used as a requirements list for those cases in which a mamba installation is not possible or advisable. From now on, we will assume that the installation is going to be done through mamba.

Ideally, you should create a separate conda environment for ESMValTool, so it is independent from any other Python tools present in the system.

To create an environment, go to the directory containing the ESMValTool source code that you just downloaded. It is called ESMValTool if you did not choose a different name.

cd ESMValTool

and create a new environment called esmvaltool with the command (when on Linux):

mamba env create --name esmvaltool --file environment.yml

or (when on MacOS)

mamba env create --name esmvaltool --file environment_osx.yml

This will install all of the required development dependencies. Note that the MacOS environment file contains only Python dependencies, so you will not be able to run NCL, R, or Julia diagnostics with it.

Note

The environment is called esmvaltool in the example above, but it is possible to use the option --name some_environment_name to define a different name. This can be useful when you have an older ESMValTool installation that you would like to keep. It is recommended that you create a new environment when updating ESMValTool.

Next, activate the environment by using the command:

conda activate esmvaltool

Attention!

From now on, we assume that the conda environment containing the development dependencies for ESMValTool is activated.

Install ESMValTool

Once all dependencies have been installed, ESMValTool itself can be installed by running the following command in the directory containing the ESMValTool source code (called ESMValTool if you did not choose a different name):

pip install --editable '.[develop]'

Using the --editable flag will cause the installer to create a symbolic link from the installation location to your source code, so any changes you make to the source code will immediately be available in the installed version of the tool.

If you would like to run Julia diagnostic scripts, you will need to install the ESMValTool Julia dependencies:

esmvaltool install Julia

If you are planning to do any coding, install the :ref:`esmvaltool:pre-commit` hooks by running:

pre-commit install

these will make sure that when you commit your changes, they will be formatted correctly.

The next step is to check that the installation works properly. To do this, run the tool with:

esmvaltool --help

If everything was installed properly, ESMValTool should have printed a help message to the console.

Note

MacOS users: some recipes may depend on the OpenMP library, which does not install via mamba on MacOS. Instead run

brew install libomp

to install the library with Homebrew. In case you do not have Homebrew, follow installation instructions here.

For a more complete installation verification, run the automated tests and confirm that no errors are reported:

pytest -m "not installation"

or if you want to run the full test suite remove the -m "not installation" flag; also if you want to run the tests on multiple threads, making the run faster, use the -n N flag where N is the number of available threads e.g:

pytest -n 4

This concludes the installation from source guide. However, if you would like to do development work on ESMValCore, please read on.

Using the development version of the ESMValCore package

If you need the latest developments of the ESMValCore package, you can install it from source into the same conda environment.

Attention!

The recipes and diagnostics in the ESMValTool repository are compatible with the latest released version of the ESMValCore. Using the development version of the ESMValCore package is only recommended if you are planning to develop new features for the ESMValCore, e.g. you want to implement a new preprocessor function.

First follow the steps in the section above to :ref:`install ESMValTool from source <install_from_source>`. Next, go to the place where you would like to keep the source code and clone the ESMValCore github repository:

git clone https://github.com/ESMValGroup/ESMValCore

or

git clone git@github.com:ESMValGroup/ESMValCore

The command above will create a folder called ESMValCore containing the source code of the tool in the current working directory.

Go into the folder you just downloaded

cd ESMValCore

and then install ESMValCore in development mode

pip install --editable '.[develop]'

To check that the installation was successful, run

python -c 'import esmvalcore; print(esmvalcore.__path__[0])'

this should show the directory of the source code that you just downloaded.

If the command above shows a directory inside your conda environment instead, e.g. ~/miniforge3/envs/esmvaltool/lib/python3.11/site-packages/esmvalcore, you may need to manually remove that directory and run pip install --editable '.[develop]' again.

Finally, also install the :ref:`esmvaltool:pre-commit` hooks by running:

pre-commit install

these will make sure that when you commit your changes, they will be formatted correctly.

Pre-installed versions on HPC clusters / other servers

ESMValTool is available on the HPC clusters CEDA-JASMIN and DKRZ-Levante, and on the Met Office Linux estate, so there is no need to install ESMValTool if you are just running recipes:

  • CEDA-JASMIN: esmvaltool is available on the scientific compute nodes (sciX.jasmin.ac.uk where X = 1, 2, 3, 4, 5) after login and module loading via module load esmvaltool; see the helper page at CEDA .
  • DKRZ-Levante: esmvaltool is available on login nodes (levante.dkrz.de) after login and module loading via module load esmvaltool; the command module help esmvaltool provides some information about the module. A Jupyter kernel based on the latest module is available from DKRZ-JupyterHub.
  • Met Office: esmvaltool is available on the Linux estate after login and module loading via module load; see the ESMValTool Community of Practice SharePoint site for more details.

The ESMValTool Tutorial provides a quickstart guide that is particularly suited for new users that have an access to pre-installed version of ESMValTool.

Information on how to request an account at CEDA-JASMIN and DKRZ-Levante and to get started with these HPC clusters can be found on the setup page of the tutorial here.

Docker installation

ESMValTool is also provided through DockerHub in the form of docker containers. See https://docs.docker.com for more information about docker containers and how to run them.

You can get the latest release with

docker pull esmvalgroup/esmvaltool:stable

If you want to use the current main branch, use

docker pull esmvalgroup/esmvaltool:latest

To run a container using those images, use:

docker run esmvalgroup/esmvaltool:stable --help

Note that the container does not see the data or environmental variables available in the host by default. You can make data available with -v /path:/path/in/container and environmental variables with -e VARNAME.

For example, the following command would run a recipe

docker run -e HOME -v "$HOME":"$HOME" -v /data:/data esmvalgroup/esmvaltool:stable run examples/recipe_python.yml

with the environmental variable $HOME available inside the container and the data in the directories $HOME and /data, so these can be used to find the configuration file, recipe, and data.

It might be useful to define a bash alias or script to abbreviate the above command, for example

alias esmvaltool="docker run -e HOME -v $HOME:$HOME -v /data:/data esmvalgroup/esmvaltool:stable"

would allow using the esmvaltool command without even noticing that the tool is running inside a Docker container.

Singularity installation

Docker is usually forbidden in clusters due to security reasons. However, there is a more secure alternative to run containers that is usually available on them: Singularity.

Singularity can use docker containers directly from DockerHub with the following command

singularity run docker://esmvalgroup/esmvaltool:stable run examples/recipe_python.yml

Note that the container does not see the data available in the host by default. You can make host data available with -B /path:/path/in/container.

It might be useful to define a bash alias or script to abbreviate the above command, for example

alias esmvaltool="singularity run -B $HOME:$HOME -B /data:/data docker://esmvalgroup/esmvaltool:stable"

would allow using the esmvaltool command without even noticing that the tool is running inside a Singularity container.

Some clusters may not allow to connect to external services, in those cases you can first create a singularity image locally:

singularity build esmvaltool.sif docker://esmvalgroup/esmvaltool:stable

and then upload the image file esmvaltool.sif to the cluster. To run the container using the image file esmvaltool.sif use:

singularity run esmvaltool.sif run examples/recipe_python.yml

Pip installation

It is also possible to install ESMValTool from PyPI. However, this requires first installing dependencies that are not available on PyPI in some other way. The list of required dependencies can be found in :download:`environment.yml <../../../../environment.yml>`.

Warning

It is recommended to use the installation with mamba instead, as it may not be easy to install the correct versions of all dependencies.

After installing the dependencies that are not available from PyPI, install ESMValTool and any remaining Python dependencies with the command:

pip install esmvaltool

If you would like to run Julia diagnostic scripts, you will also need to install the Julia dependencies:

esmvaltool install Julia

Installation from the conda lock file

The conda lock file is an alternative to the environment.yml file used in the :ref:`installation from source instructions <install_from_source>`. All other steps in those installation instructions are the same.

The conda lock file can be used to install the dependencies of ESMValTool whenever the conda environment defined by environment.yml can not be solved for some reason. A conda lock file is a reproducible environment file that contains links to dependency packages as they are hosted on the Anaconda cloud; these have frozen version numbers, build hashes, and channel names. These parameters are established at the time of the conda lock file creation, so may be outdated after a while. Therefore, we regenerate these lock files every 10 days through automatic Pull Requests (or more frequently, since the automatic generator runs on merges on the main branch too), to minimize the risk of dependencies becoming outdated.

Conda environment creation from a lock file is done with the following command:

conda create --name esmvaltool --file conda-linux-64.lock

The latest, most up-to-date file can always be downloaded directly from the source code repository, a direct download link can be found here.

Note

For instructions on how to manually create the lock file, see :ref:`these instructions <esmvalcore:condalock-installation-creation>`.

Common installation problems and their solutions

Problems with proxies

If you are installing ESMValTool from source from behind a proxy that does not trust the usual PyPI URLs you can declare them with the option --trusted-host, e.g.

pip install --trusted-host=pypi.python.org --trusted-host=pypi.org --trusted-host=files.pythonhosted.org -e .[develop]

If R packages fail to download, you might be able to solve this by setting the environment variable http_proxy to the correct value, e.g. in bash:

export http_proxy=http://user:pass@proxy_server:port

the username and password can be omitted if they are not required. See e.g. here for more information.

Anaconda servers connection issues

HTTP connection errors (of e.g. type 404) to the Anaconda servers are rather common, and usually a retry will solve the problem.

Installation of R packages fails

Problems have been reported if the R interpreter was made available through the module load command in addition to installation from mamba. If your ESMValTool conda environment is called esmvaltool and you want to use the R interpreter installed from mamba, the path to the R interpreter should end with mamba/envs/esmvaltool/bin/R or conda/envs/esmvaltool/bin/R. When the conda environment for ESMValTool is activated, you can check which R interpreter is used by running

which R

The Modules package is often used by system administrators to make software available to users of scientific compute clusters. To list any currently loaded modules run module list, run module help or man module for more information about the Modules package.

Problems when using ssh

If you log in to a cluster or other device via SSH and your origin machine sends the locale environment via the SSH connection, make sure the environment is set correctly, specifically LANG and LC_ALL are set correctly (for GB English UTF-8 encoding these variables must be set to en_GB.UTF-8; you can set them by adding export LANG=en_GB.UTF-8 and export LC_ALL=en_GB.UTF-8) in your origin or login machines’ .profile.

Problems when updating the conda environment

Usually mamba is much better at solving new environments than updating older environments, so it is often a good idea to create a new environment if updating does not work.

Do not run mamba update --update-all in the esmvaltool environment since that will update some packages that are pinned to specific versions for the correct functionality of the tool.

Move to Mamba

Mamba is a much faster alternative to conda, and environment creation and updating benefits from the use of a much faster (C++ backend) dependency solver; tests have been performed to verify the integrity of the esmvaltool environment built with mamba, and we are now confident that the change will not affect the way ESMValTool is installed and run, whether it be on a Linux or OS platform. From the user's perspective, it is a straightforward use change: the CLI (command line interface) of mamba is identical to conda: any command that was run with conda before will now be run with mamba instead, keeping all the other command line arguments and flags as they were before. The only place where conda should not be replaced with mamba at command line level is at the environment activation point: conda activate will still have to be used.