kubecare cluster manager

kubecare cluster manager is a tool for generating manifests used by ArgoCD. By using kubecare cluster manager you'll get the following benefits:

simplified syntax for defining cluster applications
you don't repeat yourself with managing multiple clusters
auto creation of namespaces for helm charts
addon support to extract common cod
support for Helm and Kustomize (as well as plain manifests)
support for multiple variants of addons

Usage

After installing and configuring ArgoCD to work with kubecare cluster manager you need to:

Create new git repository

You can use any provider that ArgoCD can connect to.

Configure ArgoCD to be able to read from that repo

In order to allow ArgoCD to connect to the repo you need to add private ssh key for that repo in ArgoCD configuration as well as add public key in your git repo configuration. See Installation part below to see how to configure ssh private key in ArgoCD.

Create basic structure inside the repo

CLUSTER_NAME=my-cluster
mkdir clusters
mkdir addons
mkdir clusters/$CLUSTER_NAME
mkdir clusters/$CLUSTER_NAME/addons
touch clusters/$CLUSTER_NAME/cluster.yaml

Edit cluster definition file

Now you need to edit clusters/$CLUSTER_NAME/cluster.yaml file and provide some basic information:

cluster:
  name: my-cluster
  server: https://url-to-kube-api-server
  # optional: repoURL: url-to-external-repo, by default it's the same url as the repo with cluster.yaml file

helmApplications:
# see below
kustomizeApplications:
# see below

Define kustomize application

Define helm application

Create helm addon

Splitting cluster definition file into multiple files

Installation on ArgoCD

When using a chart from https://github.com/argoproj/argo-helm/ (charts/argo-cd) alter your values.yaml file and set the following:

repoServer:
  volumes:
  - name: custom-tools
    emptyDir: {}

  initContainers:
  - name: download-tools
    image: luktom/ws
    command: [sh, -c]
    args:
    - |
      rm -fr /custom-tools/kubecare-cluster-manager \
      && git clone https://github.com/kubecare/cluster-manager /custom-tools/kubecare-cluster-manager \
      && bash /custom-tools/kubecare-cluster-manager/scripts/init.sh
    volumeMounts:
    - mountPath: /custom-tools
      name: custom-tools

  volumeMounts:
  - mountPath: /opt/kubecare-cluster-manager
    name: custom-tools
    subPath: kubecare-cluster-manager

server:
  config:
    configManagementPlugins: |
      - name: kubecare-cluster-manager
        generate:
          command: ["sh", "-c"]
          args:
          - |
            p=$(pwd);
            cd /opt/kubecare-cluster-manager/addons;
            git pull > /dev/null;
            cd $p;
            /opt/kubecare-cluster-manager/kubecare-cluster-manager

Init container for repoServer does the following:

clones the repo with cluster manager
launches the script that downloads the newest version of cluster manager
and also checks out addons repository (https://github.com/kubecare/cluster-manager-addons)

For server we configure custom config management plugin that:

updates addons repo for each generation of manifests (to provide newest version of configuration)
runs cluster manager to generate manifests

krecik / cluster-manager