Deploy Docker Container as serverless architecture to AWS Fargate
Serverless containerized deployment of a Python application using AWS Fargate.
This introductory guide walks through the process of deploying the docker container to AWS Fargate. The main focus is on creating a docker container and deployment process to AWS Fargate rather than automatically build a pipeline or continuous deployment (CD). The code repository accompanies the guide to get you started with this tutorial. Basic knowledge of AWS concepts by the reader is recommended.
Let us start with our tutorial.
Prepare the simple project
You can easily clone the sample project from our Github here. Or you can follow this tutorial to create the basic application which will be used in this tutorial. Please use your terminal to execute this command in your working directory.
git clone https://github.com/tribasuki74/mypython.gitCloning into 'mypython'...
remote: Enumerating objects: 1914, done.
remote: Counting objects: 100% (1914/1914), done.
remote: Compressing objects: 100% (1702/1702), done.
remote: Total 1914 (delta 125), reused 1902 (delta 121), pack-reused 0
Receiving objects: 100% (1914/1914), 7.99 MiB | 2.32 MiB/s, done.
Resolving deltas: 100% (125/125), done.
Next, lets us create Dockerfile for this application.
FROM python:3WORKDIR /appCOPY . .RUN python3 -m pip install --upgrade pipRUN python3 -m pip install --no-cache-dir -r requirements.txtENV FLASK_APP=mypython.pyENV FLASK_ENV=developmentENV FLASK_RUN_HOST=0.0.0.0EXPOSE 5000CMD ["flask", "run"]
To create the docker image, you can run this command
docker build -t mypython .After a certain time of the process, if all processes are successful, you will get the docker images. To check that image, you should check it by this command
docker images -aREPOSITORY TAG IMAGE ID CREATED SIZE
mypython latest 24b7006722b4 15 minutes ago 929MB
Lets us running our docker image
docker run -d --name mypython -p 5000:5000 mypython:latestOk, we will check it using our browser. Please type localhost:5000/into your browser. If everything is fine, you will get this display.
and you also can try another route localhost:5000/add?data=1,2,3,4,5.
Now that we have a working simple docker container, let’s see how it fits in the context of AWS Fargate.
Introduction
AWS, which stands for Amazon Web Services, provides many great services, including Elastic Container Registration (ECR), where we can register our docker images, same as Docker Hub functionality and Elastic Container Service (ECS), which is consist of two different services, Fargate and EC2. In this tutorial, we will deploy our docker image only to the Fargate service in ECS.
Create Free AWS Account
First, we need an AWS account. Please click this link to start. You will bring it to this page.
The AWS provides 12 months of free tier access, which you can use for this tutorial.
After you finished the registration process and login, you will be directed to the main page.
Register Docker Image to ECR
Please select ECR service by type ECR in the search area and then click on the ECR title. You will be directed to the ECR dashboard.
Lets us create a new repository by click [Create repository] button (which has orange color) at the right top of the dashboard.
Please fill the Repository name with mypython and set Enabled for Tag immutability and Scan on push properties, and then press [Create repository] button at the bottom of the form.
Select the mypython repository and then click [View push commands] button to see the instruction on how to push our local docker images into the ECR.
There are two types of commands: the macOS or Linux platform and the other one for the Windows platform. This tutorial assumes you are using the macOS or Linux platform; if not, please refer to the instruction for Windows.
There are 4 (four) steps. You can copy (by clicking the copy icon on the right of the command) and paste it to your terminal. If you have an error, it probably you not configure your security information yet.
Please set up your configuration by type command
aws configureand followed by input the AWS Access Key ID, AWS Secret Access Key from the information that you obtain when you create your account. Also, we need to input the Default region name (you can choose based on your region setting) and Default output format which you can type json as an input.
If everything is gone well, you will get a result similar to this.
Deploy the Docker Image to ECS
Similar to the previous step, now we need to access ECS service. Please search in the search area by typing ECS and then click on the ECS title. You will be directed to ECS dashboard.
First, please click the Cluster on the left side of the sidebar’s menu to create our cluster and then click [Create cluster] button.
Please select [Network only] tab under Select cluster template and then click [Next step] button on the bottom of the screen.
Please give a name for your cluster. In this tutorial, we use mypython-cluster as cluster’s name. Click [Create] button to create our cluster.
The cluster will be created, and then you can view the cluster by clicking the [View Cluster] button.
Creation of Task Definition
Next, we can start creating Task Definitions and Services to run our application. Please click the Task Definition sidebar menu at the left of the display and then click [Create new Task Definition] button to start to create a new one.
Since we need to deploy to AWS Fargate, we select Fargate tab and then click [Next step] button.
Please use mypython-tasks as task’s name and Task Role, we no need setting anything because of our application no need that for right now. For Network Mode, already set to awsvpc as default. Then, for the Task execution role, we can set Create new role to create a new role for us.
Next, we need to specify how much memory and CPU you need for this container. In this tutorial, please use 1GB for Task memory and 0.5vCPU for Task CPU. After this, we need to add our container, which previously already we push into our ECR. Please click [Add container] button.
For container definition, first, we need to define a name for the container; in this tutorial, we use mypython-container. For the Image setting, we need to supply a link which our docker images in ECR. Please copy from our images repository and paste it here. The last one is we need to define a port number exposed by the docker image that we already defined in our Dockerfile. In this tutorial, we use 5000. Next, we click [Add] button.
Our containers will be displayed in the list of containers. To create the Task Definition, please click [Create] button.
Then we will get results as display above.
Next, click [View task definition] button to see our task definition configuration just we created.
Creation of Load Balancer
The next step is creating the service, but for doing that, we need a load balancer, which will be used in the process of creating the service. To create the load balancer, we can search in the AWS services search area by typing EC2 load balancer.
Please click [Create Load Balancer] button to start to create a new load balancer.
Please select Application Load Balancer by clicking the [Create] button.
We need to provide the name of our load balancer. In this tutorial, we used mypython-alb, which alb is standing for application load balancer. By default, we will use HTTP with port 80 for our load balancer, so our application's access will be managed from this port.
We also need to set Availability Zones by checked all provided availability zones, and AWS will automatically be assigned the IPv4.
Next, we click [Next: Configure Security Setting] button.
We still continue. Please click [Next: Configure Security Group] button.
On this page, first, we need to check the Create a new security group and then give a name for that. In this tutorial, we usemypython-albsg. Then we select HTTP for the type of rule, which will have port 80 as default. To continue the process, we click [Next: Configure Routing] button.
Since, by default, we don’t yet have any Target group, we need to create a new one. So, we will define a name for this new target group, called mypython-albtg. Then we should select IP for the Target type. This selection is because the service which we will be created initiates a task from within our vpc.
The last one, we need to define the path where the health code was placed.
Click [Next: Regular Targets] button to continue.
We don’t need to change anything here. Just click [Next: Review] button and then click the [Create] button on the next page. We will see the process of load balancer will be created.
Just click [Close] button. Congratulations, we already successfully created our load balancer. Next, we will use our load balancer on the creation of services.
Creation of Services
To create the service, please click the Cluster sidebar menu and then click our cluster’s name.
Please click [Create] button in the Services tab to create new services for our cluster.
First, we set Launch Type as FARGATE. Next, we select mypython-tasks as Task Definition and select the latest’s tag for Revision. For cluster, default is our current cluster is mypython-cluster, and we give our service name as mypython-service. The last one, we need to define how many tasks will be managed by these services. We set it 2, which means the services will manage 2 of the tasks, which if the first task fails, the second task will be running.
Please click [Nest step] button to continue the process.
On this page, we need to select our cluster VPC, which based on our load balancer already created in the previous process is vpc-d77892b1 (see image below).
Then, we set the subnet; in this case, we select all available subnet for our region. Next, we click [Edit] button to set our security group, which we already created when we create our load balancer.
We select Select existing security groupfor Assigned security groups, and then from a list, we select our load balancer security group, which is mypython-albsg, and then click [Save] button to back to our Configuration network page.
Then, we set 0 to Health check grace period, that means we will ignore the checking it process. For Load balancer type, we select Application Load Balanacer.
Then, for the next set, we need to select our load balancer, mypython-alb. For Container to load balancer, we will add our container. mypython-container:5000:5000. Please click [Add to load balancer] button.
For our container, we select 80:HTTP from a dropdown component, and also we select our Target group name, which is mypython-albtg.
For Enable service discovery integration, we should not check because for right now, we no need any other microservices in our application. But, this feature can be handy when we used other microservices in our application.
Please click [Next step] to continue.
We don’t need to change anything on this page, and please continue by clicking [Next step] button.
And then, we click [Create Service] button.
We will get the result page. Congratulation again, we did it.
We can see our service detail by clicking the [View Service] button.
We can see here; there are two tasks to have the status PENDING. It should need more time to be PROVISIONING, and then the status is RUNNING.
Now, we will back to our load balancer. Please going back to EC2 dashboard.
Please click Load Balancer sidebar menu.
Check our load balancer, and then in Description tab, click the copy icon to copy our DNS name. Then please paste it into your browser. You will get a result similar to the image below.
Congratulations, we have finished our tutorial. Good job!
Summary
Setting load balancer with services that running two tasks simultaneously will give us high availability to our application. That means our load balancer will have two available services which always ready to serve our request. If one of the tasks fails, the other will be initiated right aways to serve.
Hopefully, this article will help you learn more about AWS Fargate as a serverless architecture and learn about Docker and Docker image technologies.
Thank you for your time!
References
[1] https://dzone.com/articles/serverless-architectures-on-aws
[2] https://levelup.gitconnected.com/a-practical-guide-to-deploying-microservices-on-aws-6a8ddd298f7
[3] https://aws.amazon.com/id/getting-started/hands-on/build-modern-app-fargate-lambda-dynamodb-python/
[4] https://www.learnaws.org/2018/02/06/Introduction-AWS-Fargate/
