scottslowe-learning-tools/traefik/aws-tf-traefik

Using Traefik with Docker on CoreOS on AWS

These files help establish a simple Docker Swarm cluster on which you can deploy Traefik (https://traefik.io), a dynamic reverse proxy, to help direct traffic for services deployed on the Swarm cluster.

Prerequisites

This demo environment assumes that you have the following software already installed on your system:

• Terraform 0.9.x (tested with 0.9.11, should work with any 0.9.x version)

Before trying to use this demo environment, please ensure that you've installed the necessary prerequisites.

This demo environment was tested on macOS 10.12 "Sierra", but it should work on any recent macOS release or recent Linux distribution. I don't know what would be required to make it work on Windows, sorry.

Contents

• compute.tf: This Terraform configuration launches the CoreOS-based instances used for the Docker Swarm setup.

• data.tf: This Terraform configuration file provides information to Terraform on which AWS AMIs to use.

• networking.tf: This Terraform configuration file creates all the networking constructs needed for the environment (VPC, Internet gateway, subnet, gateway attachment, route table, and route table associations).

• output.tf: This Terraform configuration outputs information about the resources created (the public IP addresses of the instances, specifically).

• provider.tf: This Terraform configuration files configures the AWS provider.

• README.md: The file you're currently reading.

• security.tf: This Terraform configuration creates the security groups used by the instances.

• variables.tf: This Terraform configuration file specifies variables that Terraform will need in order to create the desired AWS infrastructure.

Instructions

As mentioned in "Prerequisites" above, these instructions assume that you have a working Terraform installation capable of working with AWS.

1. Place the files from the traefik/aws-tf-traefik directory of the "learning-tools" GitHub repository into a direcotry on your local system. You can clone the entire repository (using git clone) or just download the specific files from the traefik/aws-tf-traefik directory.

2. Create a file named terraform.tfvars and populate it with the name of the AWS keypair you'd like to use for the instances, the AWS region to use, the ID of the security group you'd like to use, the flavor of the instances (such as "t2.micro"), and the number of worker nodes you'd like created. Refer to this URL for specific details on the syntax of this file. You can refer to the contents of variables.tf for the names of the variables that need to be defined.

3. Run terraform validate to ensure there are no errors in the Terraform configuration. (If using Terraform 0.10.x, you will also need to run terraform init first; however, I haven't tested this environment with 0.10.x yet).

4. Run terraform plan to have Terraform examine the current infrastructure and determine what changes are necessary to realize the desired configuration.

5. Run terraform apply to have Terraform make the changes necessary to realize the desired configuration. When this command has completed, it will output the public IP addresses assigned to the created instances.

6. Use SSH to log into the "manager" instance. While logged into the "manager" instance, establish this node as a Docker Swarm manager using this command:

    docker swarm init
   

   Copy the output of this command; you'll need it for the next step.

7. Use SSH to log into the first "worker" node. While logged into the first "worker" node, use the output from the previous step to have this node join the Swarm cluster.

8. Repeat step 7 with all remaining worker nodes.

9. You now have a working Docker Swarm cluster running on CoreOS on AWS. To deploy Traefik into this cluster, first create an overlay network:

    docker network create --driver=overlay demo-net
   

10. Next, create a service (constrained to the manager node) to run the Traefik reverse proxy:

    docker service create --name traefik \
    --constraint 'node.role==manager' \
    --publish 80:80 --publish 8080:8080 \
    --mount type=bind,source=/var/run/docker.sock,target=/var/run/docker.sock \
    --network demo-net
    traefik --web --docker --docker.watch \
    --docker.swarmmode --docker.domain=docker.local
    

    If you'd like additional logging, add --logLevel=DEBUG to the above command.

11. Deploy a web service to be used behind Traefik:

    docker service create --name web \
    --label 'traefik.port=5000' \
    --network demo-net slowe/flask-demo-app:1.0
    

12. Run this command against the IP address of the manager and note the output:

    curl -H "Host:web.docker.local" http://<manager_ip_address>
    

    If you used a name other than "web" in step 11, replace that name in the "Host" portion of the above command.

13. Run docker service scale web=3 to scale up the "web" service (replace "web" with whatever name you used in step 11). Repeat step 12 and note that Traefik will load balance across the different containers hosting the service.

14. Deploy a new web service:

    docker service create --name api \
    --label 'traefik.port=5000' \
    --network demo-net slowe/flask-demo-app:1.0
    

15. Run this command to access the "api" service behind Traefik:

    curl -H "Host:api.docker.local" http://<manager_ip_address>
    

    If you used a name other than "api", use that name in the above command.

16. Note that Traefik is taking inbound traffic to the manager and correctly routing it to the appropriate backend container based on the "Host" header. Feel free to deploy additional services with different names to see Traefik in action.

License

This content is licensed under the MIT License.