Errors between different load balancer configurations

[MedLexAI]

Testing out this CLI interface for Fargate, and from following along with the video there are a couple of errors I don't understand.

For starters, I am specifying the exact command line from the video but get this error:

$ fargate service create app --port HTTP:8080 --lb test-app --num 2
[!] Invalid load balancer and protocol
network load balancer web-app only supports TCP

This is a Docker instance launched from the private ECR registry:

$ fargate task run --image 123456.dkr.ecr.us-east-1.amazonaws.com/test-app/test-app --subnet-id subnet-123456a --subnet-id subnet-123456b -v

I am also getting different errors based upon the ports I am specifying, for example:

$  fargate lb create test-app --port 8080 --certificate example.com
[i] Created load balancer test-app
$ fargate lb destroy test-app
[i] Destroyed load balancer test-app
$  fargate lb create test-app --port 443 --certificate example.com
[!] Could not create ELB load balancer
ValidationError: At least two subnets in two different Availability Zones must be specified
        status code: 400, request id: 1234-567-8912345

The test-app was launched with two subnets attached to different availability zones...?

Also, how does one go about launching two separate tasks within separate availability zones? I would assume that would be a requirement for the application load balancer? My goal is being able to use an alias so that https://example.com -> application load balancer -> Fargate tasks running web application

Thanks in advance, your project will be a big help if I can get it working properly!

[jpignata]

Hi there! Sorry my project is giving you trouble. A few things:

• By default, lb create will create an Application Load Balancer (HTTP/HTTPS) for ports 443 or 80 and a Network Load Balancer (TCP) for anty other port. What's happening above is that it's creating a Network Load Balancer for port 8080 which can only be used for TCP services so you're getting that error trying to use it with an HTTP service. You can manually specify the type by using --port HTTP:8080 which will create an Application Load Balancer.
• Application Load Balancers require two subnets. By default, the CLI will grab your default subnets and use them. I suspect you may only have one default subnet in your account. You can verify this in the console by checking the list of subnets and see which have the default flag. If I'm right, you can fix this in one of two ways:
  • Create a new default subnet in a different availability zone (https://docs.aws.amazon.com/AmazonVPC/latest/UserGuide/default-vpc.html#create-default-subnet)
  • Explicitly pass subnets to the tool via --subnet-id to place your ALB on a different VPC.

In terms of launching across AZs, the CLI will launch across your default subnets by default. You can control which subnets it uses by passing --subnet-id to either task run or servcie create.

I hope that helps!

[MedLexAI]

Thanks for the prompt response, I will test out your suggestions tomorrow and look forward to contributing feedback!