Purpose

This GitHub repository contains code for deploying a proof of concept of Vault as a credential broker for Boundary, with a PostgreSQL database as target.

This article explains the code: https://security.padok.fr/blog/vault-credential-broker

Installation

Tools

You should have all these tools installed:

• terraform: https://developer.hashicorp.com/terraform/tutorials/aws-get-started/install-cli

• AWS CLI: https://docs.aws.amazon.com/cli/latest/userguide/getting-started-install.html

• Helm: https://helm.sh/docs/intro/install/

• direnv: https://direnv.net/

Prerequisites

• Clone this repository

• Have a VPC with 2 subnets and an EKS cluster deployed on AWS

• Create AWS profile "padok_univ_3" in ~/.aws/config, or if you want to use a profile of yours, search each occurence of "padok_univ_3" and replace them with your profile name

  • then direnv allow

• Login to AWS: aws sso login

• Replace domain name with your domain name: search each occurence of "padok.school" and replace them with your domain name

1. Bootstrap

• Create terraform backend for remote state: in folder terraform/layers/0-bootstrap
  • terraform init
  • terraform apply

2. Backbone

• Create RDS, EIP and KMS keys: in folder terraform/layers/1-backbone
  • In file locals.tf:
    • Replace local.name with your EKS cluster name
    • Complete vpc_id and vpc_private_subnets_ids with your VPC information
    • Complete node_group_iam_role_name with the name of the node group's iam role of your cluster
  • terraform init
  • terraform apply
• Set up kubectl to use the newly deployed EKS cluster by running

  aws eks update-kubeconfig --name boundary-eks

3. Boundary installation

• Create POSTGRESQL database for Boundary

  helm repo add bitnami https://charts.bitnami.com/bitnami
  helm repo update
  helm install postgresql bitnami/postgresql -n boundary --create-namespace
  kubectl get secret --namespace boundary postgresql -o jsonpath="{.data.postgres-password}" | base64 -d

  • /!\ WARNING: If you have a single node cluster, you need to check in which zone your node is:

    kubectl get nodes
    kubectl describe node <NODE_ID>

    Look for zone in result, for example: topology.kubernetes.io/zone=eu-west-3a

• In helm/boundary-controller, edit values.yaml file:

  • service.beta.kubernetes.io/aws-load-balancer-eip-allocations: eipalloc-xxxxxxxx,eipalloc-yyyyyyyy

    • write eip ids of both boundary-controller eip

  • public_cluster_ip: "xx.xx.xx.xx"

    • change ip with a controller elastic ip created for controller in backbone/eip.tf

      /!\ WARNING: If you have a single node cluster, you need to use the first elastic ip you wrote in the controller load balancer config if your node is in the first zone, else the second elastic ip

  • postgres_url: "postgresql://postgres:[email protected]:5432/postgres"

    • write password of your POSTGRESQL database that you got at the end of the previous step when you created the database

  • kms: root: "xxxxxxxx" worker: "xxxxxxxx" recovery: "xxxxxxxx"

    • write kms key ids

• In helm/boundary-worker, edit values.yaml file:

  • service.beta.kubernetes.io/aws-load-balancer-eip-allocations: eipalloc-xxxxxxxx,eipalloc-yyyyyyyy

    • write eip ids of both boundary-worker eip

  • controller: ip: "xx.xx.xx.xx"

    • change ip with the same controller elastic ip you used in values.yaml of boundary controller

  • kms: worker: "xxxxxxxx"

    • write worker kms key id

  • config.hostname + ingress.hosts.host :

    • change both ips with the worker elastic ip in the zone where your node is (eip created for worker in backbone/eip.tf)

    /!\ WARNING: If you have a single node cluster, you need to use the first elastic ip you wrote in the worker load balancer config if your node is in the first zone, else the second elastic ip

• Install boundary

helm install boundary-controller -n boundary . #in helm/boundary-controller
helm install boundary-worker -n boundary . #in helm/boundary-worker

• Wait for network load balancers to be active, then restart boundary-controller and boundary-worker deployments

4. Populate RDS

• Open a shell in a boundary pod
• Install postgresql : apk add postgresql
• Copy both files of rds folder in pod : northwind-database.sql et northwind-roles.sql
• Populate RDS with these commands ($PG_URL is connexion url to RDS: postgresql://{user}:{password}@{rds_endpoint}:5432/postgres)

$ psql -d $PG_URL -f northwind-database.sql
$ psql -d $PG_URL -f northwind-roles.sql

5. DNS

• Apply the 2-dns layer

6. Vault installation and configuration

• Install vault

helm repo add hashicorp https://helm.releases.hashicorp.com
helm install vault hashicorp/vault -n vault --create-namespace

• Initialize vault

kubectl exec -it vault-0 -n vault -- /bin/sh
$ vault operator init # /!\ Note three unseal keys and initial root token
$ vault operator unseal <key1>
$ vault operator unseal <key2>
$ vault operator unseal <key3>

• Port-forward the Vault service

kubectl port-forward svc/vault -n vault 8200:8200

• In 3-vault/providers.tf, edit token with your initial root token

• Apply 3-vault layer

7. Boundary configuration

• /!\ WARNING: As stated in this documentation, you need to provide access_key, secret_key and session token from SSO and export is as env variables

export AWS_ACCESS_KEY_ID="..."
export AWS_SECRET_ACCESS_KEY="..."
export AWS_SESSION_TOKEN="..."

• Then, port-forward the Boundary service

kubectl port-forward svc/boundary-controller-api -n boundary 9200:80

• Apply 4-boundary layer

Usage

To log into Boundary (password: $uper$ecure):

boundary authenticate password -addr http://boundary-api.padok.school -auth-method-id <AUTH_METHOD_ID> -login-name jeff

To list organization:

boundary scopes list -addr http://boundary-api.padok.school

To list projects in organization:

boundary scopes list -addr http://boundary-api.padok.school -scope-id <ORG_ID>

To list targets in project:

boundary targets list -addr http://boundary-api.padok.school -scope-id <PROJECT_ID>

To connect to PostgreSQL target:

boundary connect postgres -target-id <TARGET_ID> -addr http://boundary-api.padok.school -dbname postgres