Skip to content

Deploy an MLflow tracking server in one sitting.

Choose a server name. Trackshund provisions the private MLflow instance, gives it an HTTPS address and waits until its core pieces answer before reporting it ready.

export MLFLOW_TRACKING_URI=https://your-team.mlflow.example
export MLFLOW_TRACKING_TOKEN=<your-token>

import mlflow
mlflow.set_experiment("forecasting")

Illustrative hostname. Your dashboard supplies the real address and token instructions.

Three steps on your side.

  1. Name the server

    The name becomes the server address. The dashboard shows progress while Trackshund provisions the required cloud resources.

  2. Create your token

    Sign in to the server and create an access token for scripts or CI. Each teammate signs in separately.

  3. Point MLflow at it

    Set the tracking URI and token. Your training code continues to use the normal MLflow SDK.

Checks completed before a server is ready

Before the dashboard marks a server ready, Trackshund checks the API, database, artifact path, and routing. A failed attempt rolls back so it does not leave a partially created server.

Update the tracking URI in your existing code.

Deploy the server, create a token, and get back to the experiment.

Deploy your MLflow server