Metaflow UI is a tool to monitor Metaflow workflows in real-time.
Using Metaflow UI requires Metaflow Service for now.
To setup a local development environment, see docs/README.md.
Deploying Metaflow UI requires setting up a UI service (which is different from the Metaflow service but uses the same backing database). To deploy the UI service, follow instructions at Metaflow UI Service.
Dockerfile provides support for an nginx
container hosting the production build of the application.
# Build Docker image
$ docker build --tag metaflow-ui:latest .
# Run Docker container on port 3000
$ docker run -p 3000:3000 metaflow-ui:latest
# Run Docker container using custom API endpoint
$ docker run -p 3000:3000 -e METAFLOW_SERVICE=http://custom-ui-backend/api metaflow-ui:latest
For example, when used with a locally deployed Metaflow UI Service, the UI can be launched with
docker run -p 3000:3000 -e METAFLOW_SERVICE=http://localhost:8083/ metaflow-ui:latest
Dockerfile also supports the following environment variables to inject content into the UI's index.html:
METAFLOW_HEAD
- Inject content tohead
elementMETAFLOW_BODY_BEFORE
- Inject content at the beginning ofbody
elementMETAFLOW_BODY_AFTER
- Inject content at the end ofbody
element
Use cases for these variables range from additional meta tags to analytics script injection.
Example on how to add a keyword meta tag to Metaflow UI:
METAFLOW_HEAD='<meta name="keywords" content="metaflow" />'
See docs/plugin-system.md to get started with plugins development.
See docs/README.md to learn more.
General Metaflow documentation available here:
We welcome contributions to Metaflow. Please see our contribution guide for more details.
There are several ways to get in touch with us:
- Open an issue at: https://github.com/Netflix/metaflow-ui
- Email us at: [email protected]
- Chat with us on: http://chat.metaflow.org