Getting started

Activate e3sm_unified environment

If you have an account on one of the E3SM supported machines (NERSC, Compy, Acme1, LCRC, Cooley, Rhea), you can access zppy by activating e3sm_unified, which is a conda environment that pulls together Python and other E3SM tools such as e3sm_diags and zstash.

The paths to e3sm_unified activation scripts are machine dependent:

source /share/apps/E3SM/conda_envs/
source /global/common/software/e3sm/anaconda_envs/
source /lcrc/soft/climate/e3sm-unified/
source /lcrc/soft/climate/e3sm-unified/
source /lus/theta-fs0/projects/ccsm/acme/tools/e3sm-unified/
source /p/user_pub/e3sm_unified/envs/
source /gpfs/alpine/proj-shared/cli115/e3sm-unified/

Change .sh to .csh for csh shells.

Note that e3sm_unified’s development cycle is not in phase with zppy. Therefore the version of zppy included may not be the latest. To install the latest stable release, refer to the following:

Installation in a Conda Environment

If the E3SM Unified environment doesn’t serve your needs, you can alternatively install the latest version in your own custom conda environment.

First, activate conda or install it if it’s not available. Details vary amongst machines.


If the system doesn’t come with conda pre-installed, follow these instructions:

  1. Download Mambaforge

    MacOS x86_64 (note that zppy is not supported on MacOS, but it may be useful to contribute to the documentation on MacOS)
  2. Install Mambaforge

    bash ./
    MacOS x86_64
    bash ./

    When you see:

    by running conda init? [yes|no]
    [no] >>> yes

    respond with yes so conda and mamba commands are available on initializing a new bash terminal.

3. If you are working on a machine/network that intercepts SSL communications (such as acme1), you will get an SSL error unless you disable the SSL verification:

conda config --set ssl_verify false
binstar config --set ssl_verify False

4. Once conda and mamba are properly working, you can install the (a) Latest Stable Release or create a (b) Development Environment.

(a) Latest Stable Release

Installation using mamba

First, make sure that you’re using bash.


You must have a conda base enviornment installed as well. See “Installation in a Conda Environment” section above for installing conda.

These steps should not be necessary if you installed Mambaforge as suggested above but may be needed if you have previously installed Miniconda3 instead:

conda install -y -n base mamba
conda config --add channels conda-forge
conda config --set channel_priority strict

Create a new conda environment with zppy installed and activate it:

mamba create -n zppy_env zppy
conda activate zppy_env

Or (less recommended because of potential conflicts) you can install zppy in an existing environment.

mamba install zppy


If you installed into your own conda environment (e.g., not through the unified environment), you can update zppy by doing the following:

mamba update zppy

(b) Development Environment

Unlike the latest stable release (i.e., the user environment), the development environment does not include zppy. Instead, the developer will python -m pip install -e . to build zppy with changes (see step 7 below).

Furthermore, the dev environment includes quality assurance (QA) tools such as code formatters, linters, and pre-commit. NOTE: These QA tools are enforced using pre-commit checks in the continuous integration/continuous delivery (CI/CD) build, so you must use the dev environment for all contributions.

  1. Follow “Others/Local” section for installing conda.

  2. Clone your fork and keep it in sync with the main repo’s main

    # Go to
    # Click "Fork" in the upper right hand corner. This will fork the main repo.
    # Click the green "Code" button
    # Choose the HTTPS or SSH option.
    # (To use the SSH option, you need to have a SSH connection to GitHub set up).
    # Click the clipboard icon to copy the path.
    # On your command line:
    git clone <path>
    git remote -v
    # You should see your fork listed as `origin`

    or if you already have a clone of your fork, rebase your fork on the main repo’s main to keep it in sync:

    # Add the main repo as a remote.
    # You can call it anything but "upstream" is recommended.
    # We'll use `<upstream-origin>` here.
    git remote add <upstream-origin> <path from the green "Code" button mentioned above>
    # Fetch all the branches of that remote into remote-tracking branches
    git fetch <upstream-origin>
    # Make sure that you're on your ``main`` branch:
    git checkout main
    # Rewrite your `main` branch so that any of your commits that
    # aren't already in <upstream-origin>/main are replayed on top of that branch:
    git rebase <upstream-origin>/main
    # Push your main branch to your GitHub fork:
    # Note that <fork-origin> should be `origin` if you cloned your fork as above.
    git push -f <fork-origin> main

    Checkout a new branch from main:

    git checkout -b <branch-name> <remote-origin>/main
  3. Remove any cached conda packages. This will ensure that you always get the latest packages.

    mamba clean --all
  4. Enter the fork’s clone.

    cd zppy

5. Use conda to create a new dev environment. (zppy is not included in this environment).

  • Tip: Add the flag -n <name_of_env> to customize the name of the environment

mamba env create -f conda/dev.yml
conda activate zppy_dev
  1. Install pre-commit.

    pre-commit install
  2. install zppy into the conda environment in edit mode (so code changes are reflected in the environment as you make them) with:

    python -m pip install -e .
  3. Make the desired changes to zppy.

  4. Commit changes and make sure pre-commit checks pass:

    git commit -m "commit-message"
    pre-commit Output

    pre-commit Output

Configuration file

The configuration files consists of sections ([...]) and subsections ([[...]]). There is a default section at the top ([default]) to define some common settings, followed by a separate section for each available task. Within each task section, you can optionally include an arbitrary number of subsections for multiple renditions of a given task. The name of the subsections are arbitrary. They are used to name the batch jobs and resolve dependencies.

Please note that the configuration file follows an inheritance model: [[ subsections ]] inherit settings from their parent [section], which itself inherits settings from the [default] section. Settings can be defined at arbitrary levels, with the lower level definition taking precedence: [[ subsection ]] settings can overwrite [section] settings which can overwrite [default] settings. Many settings also take on sensible default values if they are not set.


To start the post-processing:

zppy -c <configuration file>

zppy will parse the configuration file and then generate and submit all batch jobs. zppy can be invoked safely multiple times – it will simply check the status of previously submitted tasks, only submitting new or previously failed tasks.