Amazon Elastic Kubernetes Service

This page describes how to install and configure Styra DAS On-premises on Amazon Elastic Kubernetes Service (EKS) with Relational Database Service (RDS) and managed Elasticsearch Service (ES).

Prerequisites

The following shows the list of software requirements to install and configure Styra DAS On-premises on EKS with RDS and ES.

Ensure you have a Kubernetes cluster with version 1.11 or later, with a minimum of six vCPUs and 32GB of memory.
Install the kubectl client.
Configure SMTP (Simple Mail Transfer Protocol): An SMTP server allows Styra to send emails. For example, Styra can send an email for user activation, and Styra can also send an email to assist the user for password recovery. During installation, the SMTP information must be specified as listed in the Install and Configure Styra DAS On-Premises on EKS with RDS and Managed ES section.
- You need an SMTP host, SMTP port, SMTP username and password, and SMTP From email address.
- For Gmail, you can create a username or password for your account by clicking App Passwords from Google's security page.
Styra DAS can connect to AWS PostgreSQL RDS and Elasticsearch.
- For RDS, use the database URL in the format: postgres://USERNAME:PASSWORD@ENDPOINT/postgres, where the endpoint is available on the RDS database dashboard in the AWS console, the username and password are valid PostgreSQL credentials.
- Similarly, you need the VPC endpoint for Elasticsearch, that is available on the Elasticsearch dashboard in the AWS console, for example: https://vpc-myelastic-7icwafgw6u3kyowfu2yhhcsh4u.us-west-2.es.amazonaws.com.
important
1. PostgreSQL should not use Identity and Access Management (IAM) based authentication. An example of a valid endpoint URL like mypostgres.cxfo17xhlaq4.us-west-2.rds.amazonaws.com is used.
2. AWS managed Elasticsearch will use IAM-based authentication. You should grant access to your Elasticsearch domain to an IAM user or role, and then configure Styra DAS appropriately for the user or role credentials (as described below).

Install and Configure Styra DAS On-Premises on EKS with RDS and Managed ES

To ensure kubectl is pointing to the correct cluster, run:
```
kubectl config get-contexts
```
Download the Styra DAS On-premises YAML configuration files from the Overview page.
If you are using the images directly from registry.styra.com, then configure the cluster nodes to use a private registry with the following command. In this case, skip step #4 and step #5.
```
export REPOSITORY_URL="registry.styra.com"
```
For more information, see the Overview page.

info
It is assumed that either the cluster nodes are configured to use a private registry (to use images directly from registry.styra.com) OR Docker images are already downloaded from the Overview page and are available in the Cloud Registries (GCP or Azure or AWS).

To setup the Project ID or authentication details, run:

export ACCOUNT=<account-id> && export REGION=<aws-region>

To set the REPOSITORY_URL environment variable, run:

export REPOSITORY_URL=$ACCOUNT.dkr.$REGION.amazonaws.com

Complete the following installation steps with either helm(recommended) or kubectl.

Using `kubectl`

Use the following instructions to install Styra DAS with kubectl and configure the Kubernetes resources:

Modify the YAML files to use the container registry.

TMP_FILE=`mktemp /tmp/on-premises.yaml.XXXXXXXXXX`; sed -e "s/REPOSITORY_URL/$REPOSITORY_URL/" on-premises.yaml > $TMP_FILE; mv $TMP_FILE on-premises.yaml

(Optional) Modify Elasticsearch and Postgres YAML files to use Standard Kubernetes container registry. You should skip this step if you are using AWS managed Elasticsearch and Postgres RDS.

TMP_FILE=`mktemp /tmp/postgres-deployment.yaml.XXXXXXXXXX`; sed -e "s/REPOSITORY_URL/$REPOSITORY_URL/" postgres/postgres-deployment.yaml > $TMP_FILE; mv $TMP_FILE postgres/postgres-deployment.yaml; \
TMP_FILE=`mktemp /tmp/es-client.yaml.XXXXXXXXXX`; sed -e "s/REPOSITORY_URL/$REPOSITORY_URL/" elastic/es-client.yaml > $TMP_FILE; mv $TMP_FILE elastic/es-client.yaml; \
TMP_FILE=`mktemp /tmp/es-data-stateful.yaml.XXXXXXXXXX`; sed -e "s/REPOSITORY_URL/$REPOSITORY_URL/" elastic/es-data-stateful.yaml > $TMP_FILE; mv $TMP_FILE elastic/es-data-stateful.yaml; \
TMP_FILE=`mktemp /tmp/es-master.yaml.XXXXXXXXXX`; sed -e "s/REPOSITORY_URL/$REPOSITORY_URL/" elastic/es-master.yaml > $TMP_FILE; mv $TMP_FILE elastic/es-master.yaml

When using AWS managed Elasticsearch and PostgreSQL RDS:

Styra DAS must connect to AWS PostgreSQL RDS and Elasticsearch.
- For RDS, use the database URL in the format postgres://USERNAME:PASSWORD@ENDPOINT/postgres. The endpoint is available on the RDS database dashboard in the AWS console, the username and password are valid PostgreSQL credentials.
- Similarly, you need the VPC endpoint for Elasticsearch that is available on the Elasticsearch dashboard in the AWS console.

Create LoadBalancer IP address.
- Create an ingress for the installation and determine the external URL assigned for it. You must use the ingress URL in settings.yaml.
Configure the LoadBalancer IP or NodePort in the DAS gateway-public service specification by updating one of the following files.
- (Optional) Add spec.loadBalancerIP in standard-external-http/load-balancer-svc.yaml
  
  OR
- (Optional) Add spec.ports.nodePort in standard-external-http/node-port-svc.yaml.
Configure settings by modifying settings.yaml.
- ingress_url: http://<Cluster Ingress>
- from_email_address: "YOUR EMAIL ADDRESS"
- smtp_address: "smtp.gmail.com:PORT NUMBER" [25 or 465 or 587]
Configure credentials by modifying credentials.yaml.
- smtp_username: "YOUR SMTP USERNAME"
- smtp_password: "YOUR SMTP PASSWORD"

(Optional) When the SMTP server is not configured, then a root user must be added, so that you can login after installation to add or update other users.

Set the root user and password in environment variables.

export ROOT_USER=<root user email> && export ROOT_USER_PWD=<root user password>

Update the settings.yaml with the root user and password.

TMP_FILE=`mktemp /tmp/settings.yaml.XXXXXXXXXX`; sed "s/.*config.json.*/  config.json: '{\"tenants\": {\"default\": {\"root_users\": {\"$ROOT_USER\": {\"encrypted_password\": \"\", \"password\": \"$ROOT_USER_PWD\"}}}}}'/" settings.yaml > $TMP_FILE; mv $TMP_FILE settings.yaml

(Optional) Configure TLS:
- Update gateway-tls/deployment.yaml.
```
TMP_FILE=`mktemp /tmp/gateway-tls-deploy.yaml.XXXXXXXXXX`; sed -e "s/REPOSITORY_URL/$REPOSITORY_URL/" gateway-tls/deployment.yaml > $TMP_FILE; mv $TMP_FILE gateway-tls/deployment.yaml
```
- Update credentials.yaml and settings.yaml by mounting the gateway_tls_private_key.pem and gateway_tls_cert.pem.
```
data:
  gateway_tls_cert.pem: |
    <TLS_CERT_CONTENT>
```
- Similarly, update the secret (credentials.yaml) with gateway_tls_private_key.pem content.
```
data:
  gateway_tls_private_key.pem: |
    <TLS_KEY_CONTENT>
```
- Update gateway-tls/gateway-public-tls-service.yaml.
  - Add spec.type: LoadBalancer
  - (Optional) Add spec.loadBalancerIP: <LoadBalancerIP>
    
    OR
  - Add spec.type: NodePort
  - (Optional) Add spec.ports.nodePort: <NodePort>
- Install Gateway TLS YAML manifest files with:
```
kubectl apply -f gateway-tls/
```
Configure Elasticsearch.
- Update settings.yaml
  - elasticsearch_url: "YOUR ES URL"
  - es_sniff: "true"
  - es_ec2_iam_role_auth: "true" or "false"
  - aws_region: "YOUR AWS REGION"
- Update credentials.yaml if using IAM user authentication. (If using IAM role authentication, then skip this step.)
  - aws_access_key_id: "YOUR ACCESS KEY ID"
  - aws_secret_access_key: "YOUR SECRET ACCESS KEY"
  info
  The es_username and es_password values in credentials.yaml should remain empty (""). They are ignored in an AWS-managed ES environment in favor of IAM authentication.
Configure PostgreSQL.
- Update settings.yaml
  - db_url: "YOUR POSTGRES URL"
- Update credentials.yaml
  - db_username: "YOUR POSTGRES USERNAME"
  - db_password: "YOUR POSTGRES PASSWORD"
(Optional) Install the Elasticsearch and Postgres YAML files (if using the bundled Elasticsearch and Postgres). If using AWS managed Elasticsearch and Postgres, then skip this step.
```
for f in elastic/*.yaml postgres/*.yaml; do kubectl apply -f $f; done
```

a. Install the Styra DAS YAML files.

for f in *.yaml; do kubectl apply -f $f; done

b. Expose Styra DAS:

Using LoadBalancer service type:

kubectl apply -f standard-external-http/load-balancer-svc.yaml

Using NodePort service type:

kubectl apply -f standard-external-http/node-port-svc.yaml

Wait until the status on all of the pods is running. To check the status of the pods, run the kubectl get pods command.
Point your browser to http://<YOUR LoadBalancer IP ADDRESS>.
Reset the password for your email address (from_email_address:) by using the Forgot Password flow.

tip

The UI can be accessed without a public IP by port-forwarding port 8080 to the Gateway pod. You can use the kubectl port-forward <YOUR-GATEWAY-POD> 8080:8080 command to access the UI from localhost.

Using Helm

note

The supported Helm versions are Helm v2.16.1 and Helm v3.

Use the following instructions to install Styra DAS with helm and configure the Kubernetes resources:

Navigate to the charts directory:
```
cd charts
```
Edit values.yaml by using your code editor of choice. The Helm chart provides Helm values and its description.
Update the following values.
- email.from_address: "Your-Email"
- global.repo: $REPOSITORY_URL
- service.loadbalancerip: "LoadBalancerIP"
- smtp.username: "username"
- smtp.password: "password"
- ingress.url: http://<YOUR LoadBalancer IP ADDRESS>
(Optional) Update the following values.
- (Optional) If you want to use managed services for Postgres and Elasticsearch, update the values as per the instructions in the values.yaml.
- (Optional) Similarly, you can configure s3_decision_streaming, root_user credentials, and the oidc configuration as per the instructions present under their respective sections.
(Optional) Configure TLS to enable secure HTTPS communication between the cluster ingress and the Styra gateway-secure service.
- Change the value of global.tls.enable to true in values.yaml.
- Create the tls directory under styra-das/ with:
```
mkdir -p styra-das/tls
```
- Copy your TLS key and certificate files (with the names gateway_tls_private_key.pem and gateway_tls_cert.pem, respectively) into the tls/ directory by using the following commands.
```
cp <path-to-ssl-cert> styra-das/tls/gateway_tls_cert.pem
cp <path-to-ssl-key> styra-das/tls/gateway_tls_private_key.pem
```
(Optional) Configure TLS to enable secure HTTPS communication between the Elasticsearch service and the Styra DAS ES clients (agentloader and analysis-api services).
- Change the value of elasticsearch.load_custom_ca to true in values.yaml.
- Create the tls-ca directory under styra-das/ with:
```
mkdir -p styra-das/tls-ca
```
- Copy your custom Certificate Authority (CA) file under the newly created tls-ca directory and rename as ca-cert-es.pem with the following command:
```
cp <path-to-tls-ca-file> styra-das/tls-ca/ca-cert-es.pem
```
Install the charts directory with:
```
helm install styra-das styra-das/
```
tip
Storing the hardcoded secrets in values.yaml can be avoided by passing the values as arguments while installing the Helm chart. For example, the email and password values for the root_user can be set dynamically by using environment variables:
export PASSWORD="top-secret" helm install styra-das styra-das/ --set root_user.email=admin@example.com --set root_user.password=$PASSWORD
Wait until the status on all of the pods is running. To check the status of the pods, run the kubectl get pods command.
Point your browser to http://<YOUR LoadBalancer IP ADDRESS>.
Reset the password for your email address (from_email_address:) by using the Forgot Password flow.

Reference

Helm Values

The following shows the description of various values that can be configured with values.yaml in the Helm chart for Styra DAS.

# smtp: Configures the SMTP server settings. If SMTP is not used, then
# you must leave "serveraddress" blank ("")
smtp:
  serveraddress: smtp.gmail.com:587
  username: ""
  password: ""

# email.from_address: The user for SMTP server and also the default
# admin user created on installation.
#
# If you have SMTP configured, then you can use the "Forgot Password"
# flow to reset the password and login.
# If SMTP is not configured, then you must set the "root_user" with a
# valid login created during installation.
email:
  from_address: support@styra.com

# cluster: For future use. Leave this as "gke" for now.
cluster: gke

# The 'gke.loadbalancerip' field has been deprecated - use 'service.loadBalancerIP' instead
gke:
  loadbalancerip: ""

service:
  type: LoadBalancer

  # (optional) For the LoadBalancer service type, define the public IP that is used to configure Styra public gateway.
  loadBalancerIP: ""

  # (optional) Specify the nodePort values for the NodePort or LoadBalancer service types.
  nodePorts:
    http: ""
    https: ""

# ingress.url: URL for the ingress.
ingress:
  url: http://styra.customer.com

#ddb.enabled: Set to true, if want DAS to use AWS DynamoDB service instead of Postgres. It Needs AWS creds or IRSA enabled.
ddb:
  enabled: false

postgres:
  # postgres.enabled: Set to false, if managed RDS is used.
  enabled: true
  db_url: "postgres://postgres/postgresdb?sslmode=disable"
  db_username: "postgresadmin"
  db_password: "admin123"
  # If using AWS managed Postgres, set to `true` to use IAM role authentication. If you are not using role
  # based auth, then set the `aws.access_key_id` and `aws.secret_access_key` fields below
  db_iam_auth: false

elasticsearch:
  # elasticsearch.enabled: Set to false, if managed Elasticsearch
  # is used.
  enabled: true
  # elasticsearch.url: Set the elasticsearch URL if you are using
  # managed ES; otherwise, leave it unchanged.
  # only effective if "enabled: false".
  url: ""
  # elasticsearch.username and password enables authentication.
  # enabled flag doesn't affect either of them.
  username: ""
  password: ""
  # sniff enables the elasticsearch client calls to the `_nodes/_all/http`
  # endpoint in order to keep the state of the ES cluster in sync with the client’s
  # connection pool. In a AWS managed ES environment, 'sniff' should be 'true'
  sniff: false
  # If your Elasticsearch server is secured with a custom CA certificate, then
  # copy your CA file under the newly created tls-ca directory using below commands(Rename as below).
  #   mkdir -p charts/styra-das/tls-ca
  #   cp <path-to-tls-ca-file> tls-ca/ca-cert-es.pem
  # and set `load_custom_ca` to true
  load_custom_ca: false
  # If using AWS managed ES, set to `true` to use IAM role authentication. If you are not using role
  # based auth, then set the `aws.access_key_id` and `aws.secret_access_key` fields below.
  aws_iam_role_auth: false

# The following values are used to configure the connection to an AWS managed Elasticsearch instance
aws:
  access_key_id: ""
  secret_access_key: ""
  region: ""
  # To Enable the AWS IRSA (fine-grained IAM roles for service accounts), follow the steps
  # 1. Update value of "aws_irsa.enabled" to "true"
  # 2. Update global.serviceAccountName with the serviceAccount name which is configured with
  # desired IAM role. (Note: You can use an existing ServiceAccount or create a new one. In both the cases
  # the SA name has to be updated at "global.serviceAccountName".)
  irsa:
    enabled: false # mandatory
    serviceAccount:
      # Create the new serviceAccount (Optional: Skip this step if the desired serviceAccount already exists)
      # Update the value of "create" to "true"
      # Update the values of "aws_account_id" and "aws_iam_role"
      create: false
      aws_account_id: AWS_ACCOUNT_ID
      aws_iam_role: AWS_IAM_ROLE


# s3_decision_streaming: decisions S3 settings for streaming decisions.
s3_decision_streaming:
  s3_endpoint: ""
  s3_url: ""
  s3_region: ""
  s3_access_key: ""
  s3_secret_access_key_id: ""
  s3_iam_role_auth: false

# root_user: Defines a root user which will be created by default as
# admin. Must be set if SMTP is not configured.
# If SMTP is configured, this can be left blank.
root_user:
  #root_user.email: Email address of the root user.
  email: ""
  password: ""

# oidc fields configure the OIDC provider provisioned and enabled.
# oidc.issuer_url is the OIDC issuer URL provided by the provider. For
# example, "https://oidc.customer.com".
# oidc.client_id and client_secret are the credentials provided by the
# provider.
# oidc.allowed_domains is the domains for allowed user accounts, e.g.,
# "customer.com".
# oidc.scopes is the OIDC scopes requirested from the provider. The
# value is an array of strings. If an empty array is provided, then the
# default ["openid", "profile", "email"] is used.
oidc:
  id: "OIDC"
  issuer_url: ""
  client_id: ""
  client_secret: ""
  allowed_domains: []
  scopes: []

# global.repo: Docker image repository URL
global:
  # global.repo: Docker image repository URL for Styra images
  repo: REPOSITORY_URL
  # global.elasticRepo: Docker image repository URL for Elasticsearch images
  elasticRepo: REPOSITORY_URL
  # global.postgresRepo: Docker image repository URL for Postgres images
  postgresRepo: REPOSITORY_URL

  # Image pull secrets
  imagePullSecrets:
    - name: styra-reg

  # TLS configurations to enable secure HTTPS communication between the cluster ingress and the Styra `gateway-secure` service.
  #    mark the "enable: true", and
  #    create the tls directory under charts/styra-das/
  #      mkdir -p charts/styra-das/tls
  #    Copy your tls key and certificate files under the newly created tls/ directory.
  #      cp <path-to-ssl-cert> tls/gateway_tls_cert.pem
  #      cp <path-to-ssl-key> tls/gateway_tls_private_key.pem
  tls:
    enable: false
  # Change the value of "serviceAccountName" if you want to deploy Styra DAS's microservices-pods with a custom serviceAccount. Otherwise keep it's value "default"
  serviceAccountName: default
  # This field should be updated if the cluster version is equal to or greated than 1.20 to avoid file permissions errors.
  # kubernetesClusterVersion should be in format <major>.<minor>, dataType: float64. For example
  # kubernetesClusterVersion: 1.20
  kubernetesClusterVersion:
  # If changed to 'true', then all the DAS service pods (including ElasticSearch and Postgres pods) will be compatible to run with istio-proxy sidercar container.
  istio:
    enabled: false
  # To configure DAS pods (Datasources and Systems) network traffic to the outside world via a custom proxy server,
  # uncomment and update the properties under `proxyServerEnv` section so that they will be mounted as env variables in the respective containers.
  proxyServerEnv: []
  # - name: http_proxy
  #   value: ''
  # - name: https_proxy
  #   value: ''
  # - name: no_proxy
  #   value: ''
  # Set true to enable secondary ingress controller
  enable_secondary_ingress: false

tenants:
  features:
    # By default, the decision logs are searchable for three days and retained
    # in storage for 14 days. These can be configured by updating below values (e.g., 72h for 3 days).
    # `LOG_VIEW_LIMIT_TIME` (for searchable) and `DECISIONS_GC_HISTORY` (for retension)
    LOG_VIEW_LIMIT_TIME: "72h"
    DECISIONS_GC_HISTORY: "336h"
    # Leave empty unless you want to use a second ingress for status and discovery.
    SECONDARY_INGRESS_URL:
    # Maximum connection lifetime.  Leave blank to use default
    DATABASE_CONNECTION_MAX_LIFETIME:
    # Maximum number of idle connections to the database.  Leave blank to use default
    DATABASE_MAX_IDLE_CONNECTIONS:
    # Maximum number of open connections to the database.  Leave blank to use default
    DATABASE_MAX_OPEN_CONNECTIONS:

Prerequisites​

Install and Configure Styra DAS On-Premises on EKS with RDS and Managed ES​

Using kubectl​

Using Helm​

Reference​

Helm Values​