Version: main

Installation

How to: Install MCCP using a Helm chart#

Installing MCCP requires we:

Create a namespace
Choose a database
Create a secret for docker repository
Create a secret for creating pull requests on your git provider
Determine the public IP address of the worker nodes
Install the MCCP helm chart
Check that MCCP has been installed (optional)

1. Creating a namespace#

Create a new namespace that will be used to run the MCCP components:

$ kubectl create namespace mccp

2. Choosing a database#

MCCP stores incoming data from the connected clusters to a relational database. It supports SQLite and PostgreSQL. When using SQLite the database file may be stored on a host path volume or on a persistent volume. By default, MCCP will use SQLite on a host path volume.

info

Installing MCCP in its default configuration is ideal for trying it out but it is not recommended for production use. Using a SQLite database on a persistent volume or a cloud-hosted PostgreSQL database allows for increased reliability and scalability.

SQLite on a host path volume
SQLite on a persistent volume
PostgreSQL

In this configuration, the database file is stored on one of the cluster worker nodes. In order for all the MCCP pods to be able to access the host volume, they must be placed on the same node. We can enforce this by applying a label to one of the cluster worker nodes:

# list all worker nodes
$ kubectl get node --selector='!node-role.kubernetes.io/master' -o name
node/ip-192-168-40-197.eu-north-1.compute.internal
node/ip-192-168-68-212.eu-north-1.compute.internal

# choose the first one and label it for mccp database hosting
$ kubectl label node/ip-192-168-40-197.eu-north-1.compute.internal wkp-database-volume-node=true

This label will ensure that the MCCP pods will be deployed and run on the selected node.

If your cluster supports persistent volume storage and has a storage class defined, you can use it to request a persistent volume for the SQLite database file. The exact details of the persistent volume request vary by cluster type and the supported volume types. The following manifest is an example of a configuration of a PersistentVolumeClaim:

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: mccp-volume
  namespace: mccp
spec:
  storageClassName: default
  resources:
    requests:
      storage: 100G
  volumeMode: Filesystem
  accessModes:
    - ReadWriteOnce

3. Creating a secret for docker repository#

Create a secret that contains your docker repository credentials. This secret will be used by Kubernetes during deployment in order to pull down the MCCP images. You can find instructions on how to generate this secret here.

Add this secret to the target namespace. This needs to be the same namespace that the Helm chart will be installed.

$ kubectl create secret docker-registry \
  --namespace mccp docker-io-pull-secret \
  --docker-username=<your-docker-username> \
  --docker-password=<your-docker-password>

If you use a secrets management solution such as Sealed Secrets follow their instructions on how to create a new secret.

Take note of the secret name as you will need to supply it later when installing the chart.

4. Create a secret for creating pull requests on your git provider#

A Personal Access Token is required for the MCCP to create pull requests for new clusters. These tokens need certain permissions (scopes) set. For

GitHub: repo
GitLab: api

Save the token in secret called git-provider-credentials in the mccp's deployment namespace:

kubectl create secret generic git-provider-credentials \
  --namespace=<mccp-namespace>
  --from-literal="GIT_PROVIDER_TOKEN=$GITHUB_TOKEN"

5. Determining the public IP address of the worker nodes#

You need to determine the public IP address of the worker nodes of your cluster. This IP address is necessary in order to establish connectivity between agents and your MCCP instance. The way to determine this depends on your cluster type and provisioning method. Take note of that IP address as you will need to supply it later when installing the chart.

info

You may also use a domain name instead of an IP address.

6. Installing the Helm chart#

Before installing the chart, you need to add the Helm chart repository and then update its local cache. Run the following commands:

$ helm repo add wkpv3 https://s3.us-east-1.amazonaws.com/weaveworks-wkp/charts-v3
  "wkpv3" has been added to your repositories
$ helm repo update
  Hang tight while we grab the latest from your chart repositories...
  ...Successfully got an update from the "wkpv3" chart repository
  Update Complete. ⎈Happy Helming!⎈

Finally install the Helm chart to the target namespace by executing the following command using the Helm CLI (>= v3.5.4).

SQLite on a host path volume
SQLite on a persistent volume
PostgreSQL

$ helm install mccp wkpv3/mccp \
    --version <chart-version> \
    --namespace mccp \
    --set "imagePullSecrets[0].name=<secret-containing-docker-config>" \
    --set "agentTemplate.natsURL=<nats-address>:<exposed-port-for-nats>" \
    --set "nats.client.service.nodePort=<exposed-port-for-nats>" \
    --set "wkp-ui.image.pullSecrets[0]=<secret-containing-docker-config>" \
    --set "config.capi.repositoryURL=https://github.com/example/management-repo" \
    --set "config.capi.baseBranch=main"

7. Checking that MCCP is installed (optional)#

You should now be able to load the MCCP UI by running the following command:

$ kubectl port-forward --namespace mccp deployments.apps/mccp-nginx-ingress-controller 8000:80

The MCCP UI should now be accessible at http://localhost:8000.

How to: Install the MCCP CLI#

The MCCP CLI allows you to interact with MCCP clusters. You can use mccp to manage the lifecycle of your infrastructure declaratively using GitOps. The latest version of the MCCP CLI is available in the following links: