Provision a customized cluster

Learn how to provision a customized Konvoy cluster

In the Quick start, you provisioned a Konvoy cluster with default configuration settings. Customizing the cluster is an optional task, but one of the most common tasks you perform when deploying in a production environment.

If you are configuring the Konvoy cluster for a production deployment, you can use this tutorial to learn the basic steps involved in provisioning a custom cluster. However, provisioning a production cluster typically involves more planning and custom configuration than covered in this brief example.

This tutorial demonstrates how you can customize a few simple configuration settings for your Konvoy cluster, then provision the cluster using a public cloud infrastructure. The steps are similar if you are provisioning the cluster using an internal network (on-prem) infrastructure.

To provision a customized cluster on Amazon Web Services (AWS):

  1. Generate the configuration files by running the following command:

    konvoy init --provisioner=aws
    

    To provision the cluster with FIPS enabled, use the command:

    konvoy init --mode fips
    
  2. Edit the provisioner configuration settings in the cluster.yaml cluster configuration file.

    You can edit the cluster configuration settings to suit your needs. For example, you can change the node count or add custom tags to all resources created by the installer by modifying the corresponding settings in the cluster.yaml file under the ClusterProvisioner section.

    The following example illustrates the ClusterProvisioner settings defined in the cluster.yaml cluster configuration file:

    kind: ClusterProvisioner
    apiVersion: konvoy.mesosphere.io/v1beta2
    metadata:
      name: konvoy
      creationTimestamp: "2019-05-31T18:00:01.482791-04:00"
    spec:
      provider: aws
      providerOptions:
        region: us-west-2
        availabilityZones:
        - us-west-2c
        tags:
          owner: hector
      nodePools:
      - name: node
        count: 4
        machine:
          rootVolumeSize: 80
          rootVolumeType: gp2
          imagefsVolumeEnabled: true
          imagefsVolumeType: gp2
          imagefsVolumeSize: 160
          imagefsVolumeDevice: xvdb
          type: m5.2xlarge
      - name: control-plane
        controlPlane: true
        count: 3
        machine:
          rootVolumeSize: 80
          rootVolumeType: gp2
          imagefsVolumeEnabled: true
          imagefsVolumeType: gp2
          imagefsVolumeSize: 160
          imagefsVolumeDevice: xvdb
          type: m5.xlarge
      sshCredentials:
        user: centos
        publicKeyFile: konvoy-ssh.pub
        privateKeyFile: konvoy-ssh.pem
      version: v0.0.15-10-g57dff48
    

    As illustrated in this example, you can modify the nodePools section to configure the nodes of your cluster by changing the nodePools.count from 4 to 5 or the node type by changing the nodePools.machine.type from m5.2xlarge to m5.xlarge.

    You can also modify the tags section to extend the lifetime of your cluster. This change might be useful, for example, if your AWS administrator has created a job to remove cloud resources based on AWS resource tags. For example:

    ### needs both tags
    tags:
      owner: luxi
      expiration: 24h
    

    For more information about customizing provisioning settings, see Cluster configuration.

  3. Edit the ClusterConfiguration section of cluster.yaml configuration file to change which addons you want to enable or disable.

    The following example illustrates the ClusterConfiguration settings defined in the cluster.yaml cluster configuration file:

    kind: ClusterConfiguration
    apiVersion: konvoy.mesosphere.io/v1beta2
    metadata:
      name: konvoy
      creationTimestamp: "2019-05-31T18:00:00.844964-04:00"
    spec:
      kubernetes:
        version: 1.20.13
        networking:
          podSubnet: 192.168.0.0/16
          serviceSubnet: 10.0.0.0/18
        cloudProvider:
          provider: aws
        podSecurityPolicy:
          enabled: false
      containerRuntime:
        containerd:
          version: 1.2.5
      addons:
        configVersion: v0.0.11
        addonList:
        - name: velero
          enabled: true
        - name: helm
          enabled: true
        - name: awsebsprovisioner
          enabled: false
        - name: awsebscsiprovisioner
          enabled: true
        - name: opsportal
          enabled: true
        - name: elasticsearch
          enabled: true
        - name: fluentbit
          enabled: true
        - name: kibana
          enabled: true
        - name: prometheus
          enabled: true
        - name: traefik
          enabled: true
        - name: dashboard
          enabled: true
      version: v0.0.15-10-g57dff48
    

    In this example, you can disable the fluentbit addon by changing the enabled field from true to false.

    NOTE: The addons helm and opsportal are required and a cluster cannot be provisioned until they are enabled.

    For more information about customizing cluster configuration settings, see Cluster configuration.

  4. Provision the cluster with your customized settings by running the following command:

    konvoy up
    

The konvoy up command provisions the cluster similar to how it is provisioned using the default settings as described in the Quick start.

However, customized provisioning creates a cluster.tmp.yaml file that contains the default values merged with your user-provided overrides. The cluster.tmp.yaml file is the file that Ansible uses during its execution. You can delete this file after the cluster is created because it is regenerated every time you execute the konvoy up command.