GreenDoc - SUSEAI Demo Lab

Author: Alessandro Festa

Rev. Number: v1.0

SUSE AI Demo enviroment is a simple document that is intended to describe how to implement a local lab to demonstrate capabilites of SUSE AI in the context of specific usecase If you are interested in understanding more what is SUSE AI please refer to http://suse.com/products/ai

The applications in this repo are pulled from the SUSE Application Collection https://apps.rancher.io/ check the documentation at https://docs.apps.rancher.io/ to learn more about AppCo

This GreenDoc is structured in the following way:

• Cluster setup
• Rancher Fleet setup (for automation of the GreenDoc)
• GreenDoc Configuration
• GreenDoc deployment
• Demo

Cluster Setup

For the purpose of this GreenDoc we'll use K3s.

Create namespaces used in the environment

kubectl create namespace cert-manager

kubectl create namespace suseai

Create secrets for namespaces

kubectl create secret docker-registry application-collection \
--docker-server=dp.apps.rancher.io \
--docker-username=<your registered account on appco> \
--docker-password=<INSERT YOUR TOKEN FROM APPCO> \
-n cert-manager

kubectl create secret docker-registry application-collection \
--docker-server=dp.apps.rancher.io \
--docker-username=<your registered account on appco>\
--docker-password=<INSERT YOUR TOKEN FROM APPCO> \
-n suseai

kubectl create secret docker-registry application-collection \
--docker-server=dp.apps.rancher.io \
--docker-username=<your registered account on appco> \
--docker-password=<INSERT YOUR TOKEN FROM APPCO> \
-n fleet-local

Use Rancher Application Collection for components

Deploy cert-manager helm chart

helm upgrade --install cert-manager \
  oci://dp.apps.rancher.io/charts/cert-manager \
  --version 1.17.1 \
  -n cert-manager \
  --create-namespace \
  --set crds.enabled=true \
  --set "global.imagePullSecrets[0].name"="application-collection"

Deploy Rancher Fleet

helm -n cattle-fleet-system install --create-namespace --wait \
    fleet-crd https://github.com/rancher/fleet/releases/download/v0.12.0/fleet-crd-0.12.0.tgz

helm -n cattle-fleet-system install --create-namespace --wait \
    fleet https://github.com/rancher/fleet/releases/download/v0.12.0/fleet-0.12.0.tgz

At this point we are ready to pickup the greendoc recipe

Deploy a minimal SUSEAI Stack with Rancher Fleet

Let's create a minimal github configuration and save it as suseai.yaml

apiVersion: fleet.cattle.io/v1alpha1
kind: GitRepo
metadata:
  name: suseai
  # This namespace is special and auto-wired to deploy to the local cluster
  namespace: fleet-local
spec:
  repo: https://github.com/alessandro-festa/bundles
  helmSecretName: basic-auth-secret
  ociRegistry:
    authSecretName: application-collection
  branch: main
  paths:
  - k8s-assistant

and add a generic secret for basic authentication

kubectl create secret generic basic-auth-secret \--from-literal=username=<your registered account on appco> \
--from-literal=password=<INSERT YOUR TOKEN FROM APPCO> \
-n fleet-local

Now let's execute the config with kubectl

kubectl apply -f rancher-config.yaml

You should have the SUSE AI stack plus an mcpo server installed. We used as example the kubernetes mcp server from https://github.com/manusa/kubernetes-mcp-server

Let's check that!

Checkin MCPO Server

Open a browser and go to http://localhost:8000 you should see an image similar to the one below

Click on the kubernetes you should see a similar page

Let's authorize and try a simple query Click on the Authorize button and type top-secretas token

Now let's move to a test like get node informations. Use the try out option and check the results.

It works!

Configure Open-WebUI

We now need to configure OWUI to use the MCP server. Last release of OWUI has a new feature names "Tools Servers".

Let's open the admin settings

Move to Tools

Add the mcp server connection `http://mcp-service.suseai.svc.cluster.local:8000/kubernetes``

if everything goes right you shouild see a succesfull message

Now let's configure the model to use the tool

First move to models and select the model you want to use with through the pen icon. Second check two parameters as per images below

Switch Function Calling to `native``

And finally select the `tool``

Click save and try it ... you should see a tool in the chat (see the number 1 in the picture?)

Ask something like "list the pods in the namespace suseai"

without and with the tool check the difference!