Contribute to MLflow


This document outlines the processes and practices recommended for contributing enhancements to this operator.

Talk to us First

Before developing enhancements to this charm, you should open an issue explaining your use case. If you would like to chat with us about your use-cases or proposed implementation, you can reach us at MLOps Mattermost public channel or on Discourse.

Pull Requests

Please help us out in ensuring easy to review branches by rebasing your pull request branch onto the main branch. This also avoids merge commits and creates a linear Git commit history.

All pull requests require review before being merged. Code review typically examines:

  • code quality

  • test coverage

  • user experience for Juju administrators of this charm.


You can use the environments created by tox for development:

tox --notest -e unit
source .tox/unit/bin/activate


tox -e lint          # code style
tox -e unit          # unit tests
tox -e integration   # integration tests
tox                  # runs 'lint' and 'unit' environments

Build Charm

Build the charm in this git repository using:

charmcraft pack


# Create a model
juju add-model dev
# Enable DEBUG logging
juju model-config logging-config="<root>=INFO;unit=DEBUG"
# Deploy the charm
juju deploy ./mlflow-server_ubuntu-20.04-amd64.charm \
    --resource oci-image=$(yq '.resources."oci-image"."upstream-source"' metadata.yaml)

Updating the charm for new versions of the workload

To upgrade the source and resources of this charm, you must:

  1. Bump the oci-image in metadata.yaml

  2. Update the charm source for any changes, such as:

    • YAML manifests in src/ and/or any Kubernetes resource in pod_spec

    • New or changed configurations passed to pebble workloads or through pod.set_spec

  3. Ensure integration and unit tests are passing; fix/adapt them otherwise

Canonical Contributor Agreement

Canonical welcomes contributions to this charm. Please check out our contributor agreement if you’re interested in contributing.