Spin up the environment

First, clone the actual Kube Project repository here: https://github.com/kube-project/kube-project.

This repository contains all of the services as a monorepo.

Let's run the cluster now. This project uses a kind configuration to create a specific cluster.

Here is the configuration:

# three node (two workers) cluster config
kind: Cluster
name: kube-facerecog-test-cluster
apiVersion: kind.x-k8s.io/v1alpha4
nodes:
  - role: control-plane
  - role: worker
    extraMounts:
      - hostPath: /Users/skarlso/temp/known_people
        containerPath: /known_people
      - hostPath: /Users/skarlso/temp/unknown_people
        containerPath: /unknown_people
  - role: worker
    extraMounts:
      - hostPath: /Users/skarlso/temp/known_people
        containerPath: /known_people
      - hostPath: /Users/skarlso/temp/unknown_people
        containerPath: /unknown_people

The design here is that the face-recognition part has access to a volumn which contains the images it needs to work with. These images are either on an NFS volumn, or some other storage media. For testing purposes, these images will be located on your own machine.

For this to work, Kind needs access to these folders. Please edit the hostPath locations to a folder on your disk that you'll need to create.

One for known_people and the other for unknown_people. The unknown_people will contain new images that the face recognition service will need to compute.

Let's spin up the cluster. To use kind with a config, run the following command:

kind create cluster --config kind_test_cluster.yaml

Once executed, we should see the kind pods running and the nodes healthy.

We can now run tilt up to start all of the services.