These sample exercises demonstrate exposing and accessing the NGINX™ service by using both Marathon-LB and Edge-LB. It demonstrates the differences in configuration in terms of both load balancers.

Before you begin

You must have:

The Edge-LB API server installed as described in the Edge-LB installation instructions.
The DC/OS command-line interface (CLI) installed and configured to communicate with the DC/OS™ cluster.
The edgelb command-line interface (CLI) installed.
An active and properly-configured DC/OS Enterprise cluster with at least one DC/OS private agent node to run the load-balanced service and at least one DC/OS public agent node for exposing the load-balanced service.
Marathon-LB installed as described in the Marathon-LB installation instructions.

Preview of what you’ll do

This tutorial illustrates the differences between configuring load balancing using Marathon-LB and an Edge-LB pool to provide public access to a simple Marathon app. In this tutorial, you will:

Create and deploy a sample Marathon app called nginx.
Configure and deploy Marathon-LB using the Marathon service nginx-mlb.
Configure and deploy load balancing using the Edge-LB pool instance called nginx-edgelb.
Access the sample nginx app through the public agent URL.

Configure the sample app for Marathon-LB

These steps illustrate how you can configure and deploy the nginx service through Marathon-LB.

Copy and paste the following sample app definition into your text editor to create the nginx-mlb.json file:

{
  "id": "/nginx-mlb",
  "cpus": 0.1,
  "instances": 1,
  "mem": 128,
  "cmd": "sed -i 's:Welcome to nginx!:Welcome to nginx! - through Marathon-LB:g' /usr/share/nginx/html/index.html; nginx -g 'daemon off;'",
  "container": {
    "portMappings": [
      {
        "containerPort": 80,
        "protocol": "tcp",
        "servicePort": 10020
      }
    ],
    "type": "MESOS",
    "docker": {
      "image": "nginx"
    }
  },
  "networks": [
    {
      "mode": "container/bridge"
    }
],
  "labels": {
    "HAPROXY_GROUP": "external",
    "HAPROXY_0_STICKY": "true",
    "HAPROXY_0_VHOST": "<Public agent IP address>"
  }
}

This sample app definition includes the public IP address used to expose and access the nginx service.

Deploy the nginx service by adding the nginx-mlb.json app definition to the cluster:
```
dcos marathon app add nginx-mlb.json
```
Verify access to the service by opening a web browser and navigating to the public agent IP address and service port defined in the nginx-mlb.json file.
```
http://<Public agent IP address>:10020
```

Configure the sample app for Edge-LB

These steps illustrate how you can configure and deploy the nginx service through Edge-LB.

Copy and paste the following sample app definition in your text editor to create the nginx.json file:

{
  "id": "/nginx",
  "cpus": 0.1,
  "instances": 1,
  "mem": 128,
  "cmd": "sed -i 's:Welcome to nginx!:Welcome to nginx! - through Edge-LB:g' /usr/share/nginx/html/index.html; nginx -g 'daemon off;'",
  "container": {
    "portMappings": [
      {
        "containerPort": 80,
        "protocol": "tcp",
        "name": "nginx-port"
      }
    ],
    "type": "MESOS",
    "docker": {
      "image": "nginx"
    }
  },
  "networks": [
    {
      "mode": "container/bridge"
    }
]
}

Deploy the nginx app:
```
dcos marathon app add nginx.json
```

Copy and paste the following in your text editor to create the nginx-edgelb.json pool configuration file to expose and access the nginx service:

{
  "apiVersion": "V2",
  "name": "nginx-edgelb",
  "count": 1,
  "haproxy": {
    "frontends": [
    {
      "bindPort": 15001,
      "protocol": "HTTP",
      "linkBackend": {
        "defaultBackend": "nginx-backend"
      }
    }
  ],
    "backends": [
    {
      "name": "nginx-backend",
      "protocol": "HTTP",
      "services": [
        {
        "marathon": {
          "serviceID": "/nginx"
        },
        "endpoint": {
          "portName": "nginx-port"
        }
      }
    ]
    }
  ],
    "stats": {
      "bindPort": 1025
    }
  }
}

Deploy the Edge-LB pool configuration file to expose and access the nginx service:
```
dcos edgelb create nginx-edgelb.json
```

NOTE: When you deploy a cluster with Terraform, the default firewall configuration allows you to access only ports 80 (http) and 443 (https). Everything else is blocked. You may want to add public agent ports to the main.tf file (the Terraform configuration file of the running cluster) in module "dcos." To add public agent ports, add a line like this one, public_agents_additional_ports = [ 15001, 10020, 9090, 1025 ], and then re-deploy the cluster.

Verify deployment status for sample apps

Verify that the Marathon-LB and Edge-LB API server packages deployed successfully:
```
dcos package list
```
Verify the services and pool instances has been deployed sucessfully:
```
dcos marathon app list
```
Verify the Apache® Mesos® task relevant to services and the pool instances:
```
dcos task list
```
Verify that the Edge-LB pool named nginx-edgelb deployed successfully:
```
dcos edgelb list
```
Verify that the Edge-LB pool instance deployed successfully with the configured frontend and backend ports:
```
dcos edgelb endpoints nginx-edgelb
```

Verify service access

Access the nginx-mlb service using the public IP address and port specified in the nginx-mlb.json file.
```
http://<public_agent_public_IP>:10020
```
You should see a page for Welcome to Nginx - through Marathon-LB. For example:
View the load balancing statistics for the nginx service deployed using Marathon-LB using the default HAProxy stats port 9090:
```
http://<public_agent_public_IP>:9090
```
For example:
Access the nginx service that was exposed through Edge-LB using the public agent IP and the frontend port number.
```
http://<public_agent_public_IP>:15001
```
You should see a page for Welcome to Nginx - through Edge-LB. For example:
View the load balancing statistics for the nginx service deployed using Edge-LB using the predefined HAProxy stats port 1025:
```
http://<public_agent_public_IP>:1025
```
For example: