cmpnyinfo: deploying through aws III(fin)

5 minute read

Subtitle: uploading image to ecr; deploying with ecs

Introduction

In this post, I am going to talk about

uploading the docker image to ECR
creating cluster in ECS
writing task definition for my cmpnyinfo application
creating service & deploying with ECS

Uploading to ECR

I am going to explain how I setup awscli in my Ubuntu WSL. This should work just fine even if you are just using Ubuntu that’s not WSL.

First, you have to create an user in the IAM console. Create an access key and set the policies for Public&Private registry.

Secondly, type the command sudo apt install awscli to install awscli

If this give you an error, try typing:
pip install 'urllib3<2'

Thirdly, type the command aws configure; and configure your settings accordingly.

You can type aws configure list to see all the configurations of the profiles you created

Lastly, copy the commands in provided by ECR when you push the View push commands button. It should look something like this

If you facen and error, configure the ~/docker/config.json file and delete this line
```
"credStore": "desktop"
```

The full code would look sth like this

{
"auths": {
  "https://index.docker.io/v1/": {},
  "public.ecr.aws": {
    "auth": "some-stuff
  }
}
}

Mind the comma! Make the code compiable!

Creating cluster

Now you have to create the cluster; which, as I explained, is the cluster of underlying infrastructures, such as EC2 instances, required to run your application.

There are two options for your infrastructure for your cluster

Fargate
EC2

For my application, I am going to use EC2. However, I believe most people prefer fargate, due to its simplicity and extra features. So if you want to launch your application with fargate, you can go ahead and watch this tutorial.

https://www.youtube.com/watch?v=esISkPlnxL0&list=WL&index=13&t=916s

It is so far the best tutorial I have found, so I strongly recommend you watching it!

Now you can just go to the ecs page.

https://us-east-1.console.aws.amazon.com/ecs

Infrastructure

And click the “Create cluster” button, and choose the infrastructure type as EC2. Like this:

Configure you OS and instance type accordingly. I used Amazon Linux 2 and t2.micro(free tier) for my application

Then, modify the rest of the setting in the “infrastructure tab”

set the minimum capacity as 1, or else no EC2 instance will be created and your tasks will be stuck in provisioning state.

create a new key pair if you don’t have one! choose .pem type unless you are going to use PuTTY for SSH.

Network settings

If you don’t have a default VPC and/or subnets, you have to create one manually. If you have no idea what they are, or how to create them, please refer to this video

https://www.youtube.com/watch?v=2doSoMN2xvI&list=PL0yqfPf4QbKGmyi-8DG1GVoESHs9tiY3z&index=73&t=1365s

You also have to create your security group. My security group allows

http request into port 80 from anywhere around the world
SSH into port 22 from anywhere
- This is not recommended
- I set it as anywhere sepcifically for this post
- You can configure the setting so that only your IP address can access your containers.

I am not going to configure the “monitoring” and the “tags” tab, so go ahead and click that “create” button to create your cluster.

Create Task Definition

Now let’s create our task definition

Infrastructure requirements

set the launch type as EC2, and EC2 only (if you are going to launch it with EC2)
set the os accordingly
set the network mode to bridge!! This will enable the containers to refer to eachother by names, when we link them together.
- originally, if you launch with fargate, “awsvpc” network mode will automatically allow you to refer the containers by names.
- but if you use EC2 launch type, you have to use the “bridge” mode and manually link the containers together
set the CPU and memory accordingly
- since we are using t2 micro, which provides 1 vCPU and 1GB of memory, we are going to set the CPU and Memory as 0.8 vCPU and 0.8 GB. Don’t mind the photo!

set the name and the URI to the image in your ECR accordingly
this container is the django container, so it will listen on port 8000
set the CPU and Memory accordingly
- CPU should be 0.35 vCPU! NOT 0.45!
you should also set your environmental variables. Variables such as SECRET_KEY or DEBUG should be set as environmental variable

Now let’s define the nginx container
This container is listening on port 80
set the CPU and Memory accordingly
- CPU should be 0.35 vCPU! NOT 0.45!

startup dependency
- just like we did in the docker-compose file, we are going to set the nginx container dependent to the dj-gunicorn container
Container network
- since we are using the “bridge” network mode, we have to link the containers.

Create service

Since we have created our task definition, now we have to deploy it to the cluster through ECS!

Go to the “Cluster” tab and select the cluster you’ve created, then go to the “Service” and click “Create”

select Launch type and EC2

set the family as the task definition you’ve created (for me is cmpnyinfo-td )
set the desired tasks to 1

Finishing

Now that’s basically it! Now click “create” and pray that your service deploys just fine; which, barely happens. Here are some recommendations for you when you face error

When you’re stuck in provisioning state

make sure that in the “infrastructure” tab in the cluster, there’s one or more EC2 instances running
If you don’t see any instances, make sure that you set the minimum capacity as 1 when you created the cluster
If you still couldn’t get out of the provisioning state, GOOD LUCK. Don’t give up. If you think you’ve got enough of it, just use fargate to launch

When the containers fail, or is in “stopped” state

enable log checking when you create your task definition
chekc the log and, GOOD LUCK!

Written by

Roger Kim

Share on

Twitter Facebook LinkedIn

Roger Kim