Amazon Elastic Kubernetes Service (EKS)

This tutorial demonstrates how to deploy the Open AMT Cloud Toolkit on a Kubernetes cluster using EKS. To perform a simpler test deployment, use a single-mode cluster locally. See Kubernetes (K8s).

Amazon EKS offers serverless Kubernetes, an integrated continuous integration and continuous delivery (CI/CD) experience, and enterprise-grade security and governance. Learn more about EKS here.

Prerequisites

• kubectl
• AWS CLI
• eksctl CLI
• Helm CLI (v3.5+)
• PSQL CLI (11.13)

Get the Toolkit

1. Clone the Open AMT Cloud Toolkit.

   git clone https://github.com/open-amt-cloud-toolkit/open-amt-cloud-toolkit --branch v2.13.0
   

Create a New EKS Cluster

1. Follow steps for aws configure to finish configuration of AWS CLI.

2. Follow steps to Create a key pair using Amazon EC2 to create a SSH key for accessing the cluster.

3. Create a new EKS cluster and supporting components.

   eksctl create cluster --name <cluster-name> --region <region> --with-oidc --ssh-access --ssh-public-key <ssh-keypair-name> --managed
   

   Where:

   • <cluster-name> is the name of the new EKS cluster.
   • <region> is the AWS region to deploy the stack (Ex: us-west-2).
   • <ssh-keypair-name> is the name of the SSH key from the previous step.

Configure EKS Instance

Ensure your kubectl is connected to the correct EKS cluster to manage.

1. Provide your region and cluster name.

   aws eks update-kubeconfig --region <region> --name <cluster-name>
   

   Where:

   • <cluster-name> is the name of your EKS cluster.
   • <region> is the AWS region where the cluster is (Ex: us-west-2).

Update Access Permissions

In order to be able to see cluster details like resources, networking, and more with the Amazon EKS console, we must configure permissions in the ConfigMap. More information can be found at How do I resolve the "Your current user or role does not have access to Kubernetes objects on this EKS cluster" error in Amazon EKS?

1. Get the configuration of your AWS CLI user or role.

   aws sts get-caller-identity
   

2. Edit aws-auth ConfigMap in a text editor.

   kubectl edit configmap aws-auth -n kube-system
   

3. Add the IAM user OR IAM role to the ConfigMap. To allow superuser access for performing any action on any resource, add system:masters instead of system:bootstrappers and system:nodes.

   Add a RoleAdd a User

   # Add under existing mapRoles section
   # Replace [ROLE-NAME] with your IAM Role
   
   mapRoles: |
     - rolearn: arn:aws:iam::XXXXXXXXXXXX:role/[ROLE-NAME]
     username: [ROLE-NAME]
     groups:
     - system:bootstrappers
     - system:nodes
   

   # Alternatively, you can create permissions for a single User rather than a Role
   # Create a new mapUsers section
   # Replace [USER-NAME] with your IAM User
   
   mapUsers: |
     - rolearn: arn:aws:iam::XXXXXXXXXXXX:role/[USER-NAME]
     username: [USER-NAME]
     groups:
     - system:bootstrappers
     - system:nodes
   

4. Save and close the text editor window. A success or error message will print to the console after closing the text editor window. If an error shows, verify the correct syntax was used. Additionally, a more detailed error message will be printed within the ConfigMap text file.

Add EBS CSI driver to Cluster

The Amazon EBS CSI plugin requires IAM permissions to make calls to Amazon APIs on your behalf. This is required for Vault. Without the driver, Vault will be stuck pending since its volume will be unable to be created. This is a new requirement starting in Kubernetes 1.23 and later.

Find additional information at Creating the Amazon EBS CSI driver IAM role for service accounts.

1. Create a new IAM role and attach the required Amazon managed policy. Replace <cluster-name> with the name of your cluster.

   eksctl create iamserviceaccount \
       --name ebs-csi-controller-sa \
       --namespace kube-system \
       --cluster <cluster-name> \
       --attach-policy-arn arn:aws-cn:iam::aws:policy/service-role/AmazonEBSCSIDriverPolicy \
       --approve \
       --role-only \
       --role-name AmazonEKS_EBS_CSI_DriverRole
   

2. Add the EBS CSI add-on to the cluster. Replace <cluster-name> with the name of your cluster and <account-ID> with your Account ID. Find more information at Managing the Amazon EBS CSI driver as an Amazon EKS add-on.

   eksctl create addon --name aws-ebs-csi-driver --cluster <cluster-name> --service-account-role-arn arn:aws:iam::<account-ID>:role/AmazonEKS_EBS_CSI_DriverRole --force
   

Create Postgres DB in RDS

1. Create a Postgres DB by following the steps for Creating an Amazon RDS DB instance.

   Make sure to set the following configuration settings:

   Field	Set to
   Virtual private cloud (VPC)	Choose the VPC created from your cluster. It should follow the format: 'eksctl-<cluster-name>-cluster/VPC'
   Public access	Yes. In the next steps, we will create Security rules to limit access.
   VPC security group	Choose existing
   Existing VPC security groups	default

Configure Virtual Private Cloud (VPC) for access

1. Go to RDS home.
2. Select 'Databases' from the left-hand side menu.
3. Select your created database (Ex: database-1).
4. Under Security in Connectivity & security, click on the VPC under VPC security groups (Ex: default (sg-01b4767ggdcb52825)).
5. Select Inbound rules.

6. Select Edit inbound rules.

   Add Two New Rules

   Rule One:

   1. Select Add rule.
   2. Under 'Type' select PostgresSQL.
   3. Under 'Source' select My IP.

   Rule Two:

   1. Select Add rule.
   2. Under 'Type' select PostgresSQL.
   3. Under 'Source' select Custom.
   4. In the search box, select the security group starting with the label 'eks-cluster-sg'.

7. Select Save rules.

Create Databases and Schema

1. Use the database schema files to initialize the hosted Postgres DB in the following steps.

   Note

   The following commands will prompt for the database password you chose here.

   Where:

   • <SERVERURL> is the location of the Postgres database (Ex: database-1.jotd7t2abapq.us-west-2.rds.amazonaws.com).
   • <USERNAME> is the username for the Postgres database.

2. Create the RPS database.

   psql -h <SERVERURL> -p 5432 -d postgres -U <USERNAME> -W -c "CREATE DATABASE rpsdb"
   

3. Create tables for the new 'rpsdb'.

   psql -h <SERVERURL> -p 5432 -d rpsdb -U <USERNAME> -W -f ./open-amt-cloud-toolkit/data/init.sql
   

4. Create the MPS database.

   psql -h <SERVERURL> -p 5432 -d postgres -U <USERNAME> -W -f ./open-amt-cloud-toolkit/data/initMPS.sql
   

Create Secrets

1. MPS/KONG JWT

This is the secret used for generating and verifying JWTs.

kubectl create secret generic open-amt-admin-jwt --from-literal=kongCredType=jwt --from-literal=key="admin-issuer" --from-literal=algorithm=HS256 --from-literal=secret="<your-secret>"

Where:

• <your-secret> is your chosen strong secret.

2. KONG ACL for JWT

This configures KONG with an Access Control List (ACL) to allow an admin user open-amt-admin to access endpoints using the JWT retrieved when logging in.

kubectl create secret generic open-amt-admin-acl --from-literal=kongCredType=acl --from-literal=group=open-amt-admin

3. MPS Web Username and Password

This is the username and password that is used for requesting a JWT. These credentials are also used for logging into the Sample Web UI.

kubectl create secret generic mpsweb --from-literal=user=<your-username> --from-literal=password=<your-password>

Where:

• <your-username> is a username of your choice.

• <your-password> is a strong password of your choice.

  Important - Using Strong Passwords

  The password must meet standard, strong password requirements:

  • 8 to 32 characters
  • One uppercase, one lowercase, one numerical digit, one special character

4. Database connection strings

Warning - Using SSL/TLS with AWS RDS

Postgres 14

By default, AWS pre-selects Postgres 14. This tutorial uses the connection string setting of no-verify for ease of setup. To fully configure SSL, follow the links below. For production, it is recommended to use a SSL connection.

Find more information at Using SSL with a PostgreSQL DB instance and Updating applications to connect to PostgreSQL DB instances using new SSL/TLS certificates.

Postgres 15

Alternatively, if Postgres 15 is preferred and selected, the sslmode in the connection strings must be updated from no-verify/disable to require for the services to be able to connect to the database. No other work is required for a test environment.

Note: For a fully secured, certificate-based SSL connection, the following steps must be taken in Using SSL with a PostgreSQL DB instance. It will also require updating sslmode to verify-full or verify-ca. For production, it is highly recommended.

1. Configure the database connection strings used by MPS, RPS, and MPS Router.

   Where:

   • <USERNAME> is the username for the Postgres database.
   • <PASSWORD> is the password for the Postgres database.
   • <SERVERURL> is the url for the AWS-hosted Postgres database (Ex: database-1.jotd7t2abapq.us-west-2.rds.amazonaws.com).

2. Create RPS connection string secret.

   Postgres 14Postgres 15

   kubectl create secret generic rps --from-literal=connectionString=postgresql://<USERNAME>:<PASSWORD>@<SERVERURL>:5432/rpsdb?sslmode=no-verify
   

   kubectl create secret generic rps --from-literal=connectionString=postgresql://<USERNAME>:<PASSWORD>@<SERVERURL>:5432/rpsdb?sslmode=require
   

3. Create MPS Router connection string secret.

   Postgres 14Postgres 15

   kubectl create secret generic mpsrouter --from-literal=connectionString=postgresql://<USERNAME>:<PASSWORD>@<SERVERURL>:5432/mpsdb?sslmode=disable
   

   kubectl create secret generic mpsrouter --from-literal=connectionString=postgresql://<USERNAME>:<PASSWORD>@<SERVERURL>:5432/mpsdb?sslmode=require
   

4. Create MPS connection string secret.

   Postgres 14Postgres 15

   kubectl create secret generic mps --from-literal=connectionString=postgresql://<USERNAME>:<PASSWORD>@<SERVERURL>:5432/mpsdb?sslmode=no-verify
   

   kubectl create secret generic mps --from-literal=connectionString=postgresql://<USERNAME>:<PASSWORD>@<SERVERURL>:5432/mpsdb?sslmode=require
   

Update Configuration

Edit values.yaml

1. Open the values.yaml file in the ./open-amt-cloud-toolkit/kubernetes/charts/ directory.

2. Remove the annotations section and service.beta.kubernetes.io/azure-dns-label-name key in the kong: section. These are Azure-specific implementations.

   kong:
     proxy:
       annotations: # Delete this line
         service.beta.kubernetes.io/azure-dns-label-name: "<your-domain-name>" # Delete this line
   

3. Save and close the file.

Deploy Open AMT Cloud Toolkit using Helm

1. Deploy using Helm.

   helm install openamtstack ./kubernetes/charts
   

   Success

   NAME: openamtstack
   LAST DEPLOYED: Thu Jul 15 11:17:38 2021
   NAMESPACE: default
   STATUS: deployed
   REVISION: 1
   TEST SUITE: None
   

Initialize and Unseal Vault

Danger - Download and Save Vault Keys

Make sure to download your Vault credentials and save them in a secure location when unsealing Vault. If the keys are lost, a new Vault will need to be started and any stored data will be lost.

Tip - Finding the Vault UI External IP Address

The external IP of your Vault UI service can be found by running:

kubectl get services openamtstack-vault-ui

1. Please refer to HashiCorp documentation on how to Initialize and unseal Vault. Stop and return here after signing in to Vault with the root_token.

2. After initializing and unsealing the vault, you need to enable the Key Value engine.

3. Click Enable New Engine +.

4. Choose KV.

5. Click Next.

6. Leave the default path and choose version 2 from the drop down.

7. Click Enable Engine.

Vault Token Secret

1. Add the root token as a secret to the EKS cluster so that the services can access Vault.

   kubectl create secret generic vault --from-literal=vaultKey=<your-root-token>
   

   Where:

   • <your-root-token> is your root_token generated by Vault.

Update commonName in values.yml

1. Get the External-IP for accessing the UI. Note and save the value under 'EXTERNAL-IP'.

   kubectl get service openamtstack-kong-proxy
   

2. Update the value for commonName in the mps section in the values.yml file with the External-IP from above. Recall that values.yml is located in ./kubernetes/charts/.

   mps:
       commonName: "" # update with External-IP from `kubectl get services`
       replicaCount: 1
       logLevel: "silly"
       jwtExpiration: 1440
   

3. Update the stack using helm.

   helm upgrade openamtstack ./kubernetes/charts 
   

Verify running pods

1. View the pods. All pods should now be Ready and Running.

   kubectl get pods
   

   Success

   NAME                                                 READY   STATUS      RESTARTS   AGE
   mps-69786bfb47-92mpc                                 1/1     Running     0          4m5s
   mpsrouter-9b9bc499b-2tkb2                            1/1     Running     0          4m5s
   openamtstack-kong-68d6c84bcc-fp8dl                   2/2     Running     0          4m5s
   openamtstack-vault-0                                 1/1     Running     0          4m5s
   openamtstack-vault-agent-injector-6b564845db-zss78   1/1     Running     0          4m5s
   rps-79877bf5c5-dsg5p                                 1/1     Running     0          4m5s
   webui-6cc48f4d68-6r8b5                               1/1     Running     0          4m5s
   

2. Check that the MPS Certificate is correct in your browser. Go to your FQDN at port 4433.

   https://<Your-AWS-FQDN>:4433
   

3. Verify the MPS Certificate in your browser has the correct Issuer information and is Issued to your FQDN.

   Troubleshoot - Issued to field showing NaN or blank

   If your certificate is incorrect, the AMT device will not connect to the MPS server. See Figure 1.

   Follow the steps below to correct the problem.

   Example - Incorrect MPS Certificate

   

   Figure 1: Incorrect Certificate

   1. Open and Login to Vault UI.

   2. Go to kv/data/MPSCerts/ directory.

   3. Delete the existing MPS Certificate.

   4. In a terminal, run the following command.

      kubectl rollout restart deployment mps 
      

   5. A new, correct MPS Cert should be generated.

   6. Go back to the webserver in your browser.

      https://<Your-AWS-FQDN>:4433
      

   7. Verify the Issued to: field is no longer NaN/blank and now shows the correct FQDN.

   8. Continue to Next Steps section.

Next Steps

Visit the Sample Web UI using the FQDN name and Continue from the Get Started steps.