Setting up the Cluster

Once the Astarte Operator has been installed, and any prerequisite has been fulfilled, you can move forward and deploy an Astarte Cluster.

Using astartectl

You can use astartectl to deploy an instance through the astartectl cluster instances deploy command. This is an interactive command that will inspect your cluster and provide you with a set of profiles that can be deployed. When you choose a Profile, you will be prompted with a number of questions that will be needed to configure your instance correctly. Upon completion, astartectl will prepare and execute the deployment automatically.

astartectl Profiles

In astartectl, profiles allow for easy scaling, enhanced management, and automated upgrade upon release series without any action on behalf of the user. They're the way to go if you plan on having a standard, managed installation.

astartectl comes packed with a set of default profiles, but you can write your own ones. Profiles can be either written as Go resources, or (in a much easier fashion) as yaml resources. You can have a look at the Profiles schema here.

Writing your own profile

This guide will be extended in the future, as more recent versions of astartectl will support loading yaml profiles.

Using a standard Astarte CR

If you do not want to use astartectl or Profiles, you can create your own Astarte Custom Resource. This gives you a higher degree of customization, allowing you to tweak any single parameter in the Astarte setup. The main Astarte CRD contains extensive documentation on the available fields in OpenAPIv3 format. Just create your Astarte Custom Resource, which will look something like this:

apiVersion: api.astarte-platform.org/v1alpha1
kind: Astarte
metadata:
  name: example-minimal
  namespace: astarte
spec:
  # This is the most minimal set of reasonable configuration to spin up an Astarte
  # instance with reasonable defaults and enough control over the deployment.
  version: 0.11.4
  api:
    host: "api.astarte.yourdomain.com" # MANDATORY
  rabbitmq:
    resources:
      requests:
        cpu: 300m
        memory: 512M
      limits:
        cpu: 1
        memory: 1000M
  cassandra:
    maxHeapSize: 1024M
    heapNewSize: 256M
    storage:
      size: 30Gi
    resources:
      requests:
        cpu: 1
        memory: 1024M
      limits:
        cpu: 2
        memory: 2048M
  vernemq:
    host: "broker.astarte.yourdomain.com"
    resources:
      requests:
        cpu: 200m
        memory: 1024M
      limits:
        cpu: 1000m
        memory: 2048M
  cfssl:
    resources:
      requests:
        cpu: 100m
        memory: 128M
      limits:
        cpu: 200m
        memory: 256M
    storage:
      size: 2Gi
  components:
    # Global resource allocation. Automatically allocates resources to components weighted in a
    # reasonable way.
    resources:
      requests:
        cpu: 1200m
        memory: 3072M
      limits:
        cpu: 3000m
        memory: 6144M

You can simply apply this resource in your Kubernetes cluster with kubectl apply -f. The Operator will take over from there.