AWS - Global Accelerator not working with Network Load Balancer

0
votes

I have a Network Load Balancer and an Application Load Balancer, they work just fine, but as I need fixed IPs/hostnames I decided to create a Global Accelerator for each one.

Global Accelerator with Application Load Balancer works but with Network Load Balancer it doesn't respond...
Example:

ALB:

$ nc -zv <application-load-balancer>.awsglobalaccelerator.com 80
Connection to <application-load-balancer>.awsglobalaccelerator.com 80 port [tcp/*] succeeded!

NLB:

$ nc -zv <network-load-balancer>.awsglobalaccelerator.com 1883
nc: connect to <network-load-balancer>.awsglobalaccelerator.com port 1883 (tcp) failed: Connection timed out

I have changed Health Check port configuration for the NLB to 1883, and the Global Accelerator is shown as " All healthy". And as I said, the Network Load Balancer itself works:

$ nc -zv <network-load-balancer>.elb.sa-east-1.amazonaws.com 1883
Connection to <network-load-balancer>.elb.sa-east-1.amazonaws.com 1883 port [tcp/*] succeeded!

Both load balancers are very similar (similar instances, same VPC, subnets, etc).
AWS docs say I can use Global Accelerator with both types of Load Balancers.

I don't know why the NLB Global Accelerator doesn't respond.
What am I missing?

More info:
- I'm testing in sa-east-1 region (South America)
- I need Global Accelerator because the LBs are part of terraform for deployment, so for every build the LBs hostname changes
- I could use Elastic IP's for NLB, but to do that I'd need to change my existing subnets (and as far as I know I can't use Elastic IPs for ALBs)...

1
amazon-web-servicesterraform-provider-aws
"I could use Elastic IP's for NLB, but to do that I'd need to change my existing subnets." This suggests that your NLB isn't on public subnets, and that seems to be a requirement. "You can target an Application Load Balancer or an EC2 instance in a private subnet using AWS Global Accelerator..." but NLB isn't mentioned. The way GA connects to the different balancer types isn't the same, so the requirements differ. Is this a possible explanation? – Michael - sqlbot
@Michael-sqlbot I'm using public subnets (I've set an internet gateway to 0.0.0.0/0), all instances are accessible from the internet :/ – Eric Sant'Anna

1 Answers

1
votes

If static ip is the only thing you need to achieve then I am not getting the point using Global accelerator and NLB together. Because both provides the features of static ip.

For static ip facility there are 2 options

  1. Use Global accelerator on top of ALB(easy configuration and high cost)

  2. Use NLB and forward your request to ALB(complex configuration and cost effective)

For 2nd option you can get reference from below link. https://www.bluematador.com/blog/static-ips-for-aws-application-load-balancer