Skip to content

Latest commit

 

History

History
95 lines (60 loc) · 4.46 KB

clusters.md

File metadata and controls

95 lines (60 loc) · 4.46 KB

Deploying a new environment

Note

The following documentation outlines how to deploy a new ArgoCD envrionment onto a management cluster. See Deploying a Child Cluster if you are deploying a cluster using an existing environment.

Prerequisites:

  1. Create an application credential for your new cluster on your project
  2. Ensure enough quota for the cluster (RAM, CPU, instances etc)
  3. Provision a floating ip on your project for kubernetes API Server access
  4. (Optional) Provision a second floating IP for nginx ingress controller
  5. Create a self-managed cluster called *-management-cluster - replace * with the environment - either prod dev or staging
❗ Make sure you already have a Kubernetes cluster deployed before proceeding with rest of the documentation

Motivation

You may want to make your own environment so you can test gitOps config or charts idependently of running clusters on test/feature branches

We don't want too many environments on main - especially those that aren't being used so there's less clutter

Steps

To deploy another environment - follow these steps:

  1. Create a new branch for your work

  2. Create a new folder charts/<your-environment>

  3. Copy the charts from another environment - from either prod, dev, or staging.

  4. Create a new folder clusters/<your-environment>

  5. Copy management subfolder from another environment into your newly created folder in clusters. You can optionally copy or create any other cluster subfolders you want to be in this environment here as well

  6. Edit apps.yaml:

  • Any entries that begin with path should use charts from the new environment

  • Any entries that begin with valuesFiles should use paths from the new environment

  • remember to change spec.template.metadata.name so the prefix matches your environment name

  • (Optional) Any entries with targetRevision or revision should point to your new branch if you are using this branch for testing or as a feature branch and not intending on keeping it long-term

  1. Modify the infra-values.yaml and any other cluster-specific values files as required
  1. Modify/add any cluster-specific values for any apps you want to manage.

  1. Create a new folder mkdir -p ./secrets/<your-environment>/management/infra You will also need to create another subfolder for each extra cluster subfolder you've copied/added

  2. Add secret files .sops.yaml, api-sever-fip.yaml, and app-creds.yaml as above See (Deploying a child cluster on an existing environment steps 5-7). Repeat for all other clusters you want to add

Caution

Make sure these files has been encrypted using SOPS in the steps above before committing and pushing changes to your branch.

  1. (Optional) If this is not a temporary environment - and you want to keep it around long term in main make a PR for it and get it merged

  2. Deploy age private key secret to management cluster

New to generating age keys and generating secrets? See secrets for more information

cd scripts; ./deploy-helm-secret.sh <path-to-age-private-key>
  1. Run deploy.sh on your self-managed cluster *-management-cluster like so:
cd scripts; ./deploy.sh management <your-environment>

when deploying argo onto child clusters - replace management with the cluster's folder name

e.g. to deploy worker cluster apps - run

cd scripts; ./deploy.sh worker <your-environment>

  1. Wait for argocd to deploy and it should spring to life and spin up any other clusters you've defined

  2. Perform any Post-Deployment steps - see infra-setup and app-setup for any apps you want to manage

  3. Repeat step 12-14 for each extra cluster that you have running that you want to manage apps with.

You can get the kubeconfig to access these clusters by accessing the management cluster and running clusterctl get kubeconfig $CLUSTER_NAME -n clusters > $CLUSTER_NAME.kubeconfig