/applications-platform

Fully-automated Kubernetes platform with focus on monitoring, security, accessability, and ease-of-use

Primary LanguageHCLMIT LicenseMIT

application-platform

This project aims to provide you with a fully-automated Kubernetes platform with focus on monitoring, security, accessability, and ease-of-use.

Setup

Preparation

Make sure you have:

  • A Github account with escalated permissions to set secrets.
  • A GCP service account with escalated permissions to create resources. For example, for only the duration of the pipeline, you can attach the service account with roles/owner role. Note: please remove the escalated permissions when done with pipeline steps day-0 and day-1 for security reasons.
  • A GCP storage bucket for storing Terraform state.
  • GCP APIs enabled:
    • compute.googleapis.com
    • container.googleapis.com
    • dns.googleapis.com
  • A new RSA keypair with no passphrase for ArgoCD to access the Github repo. The public key of this keypair has to be then uploaded to the deploy keys of the repository; the private key has to be set as a secret variable ARGOCD_CREDENTIALS_KEY. This key should not be used by any other service accounts / users; hence do not bother storing this key. You can generate the keypair by running:
ssh-keygen -b 2048 -t rsa -f /tmp/id_rsa -q -N "" -C rpch-alligator

Secret variables

We try to keep as little secret variables as possible by design. For the sake of convenience, define the following secrets in your Github secrets section:

Non-secret variables

For non-secret variables, simply edit/add them in the .env file, which gets sourced during pipeline runs, e.g.:

TF_VAR_name="rpch-alligator"
TF_VAR_domain="rpch.tech"
TF_VAR_argocd_repo_url="git@github.com:Rpc-h/infrastructure.git"
TF_VAR_argocd_credentials_url="git@github.com:Rpc-h"

Installation

Github

Run the day-0-apply workflow in Github to install day-0 resources such as:

  • GKE Kubernetes cluster and node pools.
  • IAM service accounts and bindings for the Kubernetes cluster.
  • VPC networks and firewall rules for the Kubernetes cluster.

After successful completion of day-0-apply, run the day-1-apply workflow in Github to install day-1 resources such as:

  • ArgoCD helm chart and the initial ArgoCD app-of-apps.
  • IAM service accounts and bindings for day-2 applications, e.g. cert-manager, external-dns, etc.

Local

To run Terraform locally, change into the day you want to run, generate new service account credentials or re-use existing ones, export some env variables and initialize the backend config with the specific bucket:

cd day-0 #TODO - change this to the day you want to run

gcloud iam service-accounts keys create credentials.json --iam-account rpch-staging-initial@rpch-375921.iam.gserviceaccount.com

export GOOGLE_APPLICATION_CREDENTIALS=${PWD}/credentials.json
export TF_VAR_google_project="rpch-375921"
export TF_VAR_google_region="europe-west6"

terraform init -backend-config="bucket=rpch-alligator-terraform"
terraform plan

Uninstallation

Run the day-1-destroy workflow in Github to destroy day-1. After successful completion of day-1-destroy, run the day-0-destroy workflow in Github to destroy day-0. Note that some of the day-2 cloud provider resources created by your apps, such as load balancers and DNS entries, might interfere with the current destruction of day-1 and day-0. Either make sure everything in day-2 is uninstalled cleanly, or you might have to do remove the stuck resources manually.

Usage

Access

  1. Ensure your administrator has given enough permissions, e.g. your IAM user is assigned with roles/container.developer role.
  2. Install Google Cloud CLI as described here: https://cloud.google.com/sdk/docs/install-sdk.
  3. Acquire the cluster related info by running gcloud container clusters list.
  4. Follow the instructions on how to access the Kubernetes cluster here: https://cloud.google.com/kubernetes-engine/docs/how-to/cluster-access-for-kubectl. Run gcloud auth login and follow the on-screen instructions and pick the correct Google project. Next run `gcloud get generate
  5. Verify you're able to access the cluster and see the nodes by running: kubectl get nodes.

Applications

These are just some convenience commands for developers that might come in handy. Please see the official upstream docs of the respective applications for more info.

Authelia

To generate authelia password, run the following command:

export AUTHELIA_PASSWORD="your-password"
docker run authelia/authelia:latest authelia crypto hash generate argon2 --password ${AUTHELIA_PASSWORD}

ArgoCD

To retrieve ArgoCD password run:

kubectl -n argocd get secrets argocd-initial-admin-secret -o yaml | yq -r '.data["password"]' | base64 -d

Grafana

To retrieve Grafana password run:

kubectl -n monitoring get secrets kube-prometheus-stack-grafana -o yaml | yq -r '.data["admin-password"]' | base64 -d

Sealed Secrets

Make sure you have the kubeseal binary installed: https://github.com/bitnami-labs/sealed-secrets. To generate a sealed secret run:

export NAMESPACE=your-namespace
export SECRET_NAME=your-secret

cat << EOF > /tmp/env
your-key=your-value
your-another-key=your-another-value
EOF

kubectl create secret generic -n $NAMESPACE $SECRET_NAME --from-env-file /tmp/env -oyaml --dry-run=client > /tmp/${SECRET_NAME}-secret.yaml
cat /tmp/${SECRET_NAME}.yaml | kubeseal --controller-namespace sealed-secrets --controller-name sealed-secrets -oyaml | tee /tmp/${SECRET_NAME}-sealed-secret.yaml

Now you can copy the sealed-secret file to where you need it.

Prometheus

To generate the sealed secret execute the following command:

export HOPRD_API_TOKEN=XXXXXX
sed 's#__SCRAPE_USERNAME__#'"$HOPRD_API_TOKEN"'#g' ./prometheus-scrape-config.yaml | kubeseal --controller-name=sealed-secrets --controller-namespace=sealed-secrets --format yaml > ./day-2/monitoring/kube-prometheus-stack/sealed-secrets/prometheus-scrape-config.yaml