Skip to content

Quick Start

In order to use FuseML one may simply do the following:

1. Install Requirements

We assume that a Kubernetes cluster is already up and running. If not a good way to start with a local cluster is to use KinD, Rancher K3s or k3d.

It is also necessary that your cluster setup includes a storage class marked as default.

IMPORTANT: if you're installing a Rancher K3s cluster, it's recommended to skip installing the default Traefik ingress controller. The FuseML installer deploys Istio as a component and Istio is also required by some of the 3rd party tools that FuseML is integrated with (e.g. Knative, KServe).

Additional software requirements:

2. Get fuseml-installer

# On Linux
curl -sfL https://raw.githubusercontent.com/fuseml/fuseml/main/install.sh | sh -

You should get an output similar to this:

Verifying checksum... Done.
Preparing to install fuseml-installer into /usr/local/bin
fuseml-installer installed into /usr/local/bin/fuseml-installer
Run 'fuseml-installer --help' to see what you can do with it.

3. Install FuseML

fuseml-installer install

You should get an output similar to this:

> fuseml-installer install

🚢 FuseML installing...

Configuration...
  🧭  system_domain:
  🧭  extensions_repository: https://raw.githubusercontent.com/fuseml/extensions/main/installer/
  🧭  force_reinstall: false

🚢 Deploying Istio.....
✔️  Istio deployed
✔️  Created system_domain: 172.18.0.3.nip.io

🚢 Deploying Workloads...
✔️  Workloads deployed

🚢 Deploying Gitea......................................................
✔️  Gitea deployed (http://gitea.172.18.0.3.nip.io).

🚢 Deploying Registry...............
✔️  Registry deployed

🚢 Deploying Tekton................................................................................
✔️  Tekton deployed (http://tekton.172.18.0.3.nip.io).

🚢 Deploying Core..............
✔️  FuseML core component deployed (http://fuseml-core.172.18.0.3.nip.io).

🚢 Downloading command line client...
🚢 FuseML command line client saved as /Users/flaviosr/workspace/fuseml/fuseml.
Copy it to the location within your PATH (e.g. /usr/local/bin).

🚢 To use the FuseML CLI, you must point it to the FuseML server URL, e.g.:

    export FUSEML_SERVER_URL=http://fuseml-core.172.18.0.3.nip.io

✔️  FuseML installed.
System domain: 172.18.0.3.nip.io

The FuseML installer prints out URLs for gitea, tekton and fuseml-core components. You can access those sub-services directly at any given time. The fuseml-core URL is especially important, as you'll need that to use the FuseML CLI.

You can also see that the installer downloaded fuseml command line client which you will use for actual FuseML operations. You can use the latest version which was downloaded automatically or manually choose one of released versions from github.

Run the following command to see the list of available commands:

fuseml-installer install --help

4. Ensure FuseML components are up and running

# Check Istio
kubectl get pods -n istio-system -o wide
# Check Gitea
kubectl get pods -n gitea -o wide
# Check registry
kubectl get pods -n fuseml-registry -o wide
# Check tekton
kubectl get pods -n tekton-pipelines -o wide
# Check fuseml-core
kubectl get pods -n fuseml-core -o wide

If everything is in running or completed status, you are good to go. Continue with one of the available tutorials, such as the MLFlow and KServe basic example and start to have fun with FuseML.