Network Loadbalancer and health checked

To configure your Network Load Balancer in Elastic Beanstalk to perform application-level health checks and replace instances that are returning 5xx errors, you need to make several changes to your configuration.

First, it's important to understand that Network Load Balancers (NLBs) perform health checks at the transport layer (TCP) by default, which only verifies that the port is open and accepting connections. They don't check HTTP status codes like 5xx errors.

To achieve what you want, you need to:

1. Configure the health check settings for your target group
2. Configure your Auto Scaling group to use the Elastic Load Balancing health check

Here's how to update your Terraform configuration:

/* Configure health check settings for the default process */
setting {
namespace = "aws:elasticbeanstalk:environment:process:default"
name      = "HealthCheckPath"
value     = "/your-health-check-path"  # Path to check on your application
}
setting {
namespace = "aws:elasticbeanstalk:environment:process:default"
name      = "Port"
value     = "5000"  # Your application port
}
setting {
namespace = "aws:elasticbeanstalk:environment:process:default"
name      = "HealthCheckInterval"
value     = "10"  # Seconds between health checks
}
setting {
namespace = "aws:elasticbeanstalk:environment:process:default"
name      = "HealthyThresholdCount"
value     = "3"  # Number of consecutive successful checks to consider healthy
}
setting {
namespace = "aws:elasticbeanstalk:environment:process:default"
name      = "UnhealthyThresholdCount"
value     = "3"  # Number of consecutive failed checks to consider unhealthy
}

/* Configure Auto Scaling to use ELB health checks */
setting {
namespace = "aws:autoscaling:launchconfiguration"
name      = "HealthCheckType"
value     = "ELB"
}
setting {
namespace = "aws:autoscaling:launchconfiguration"
name      = "HealthCheckGracePeriod"
value     = "300"  # Time in seconds to wait before checking health after instance launch
}

This configuration does two important things:

1. It sets up the health check to target your application on port 5000 with a specific path
2. It configures the Auto Scaling group to use the "ELB" health check type instead of just "EC2"

The "ELB" health check type is crucial because it tells Auto Scaling to replace instances that fail the load balancer's health check. Without this setting, an unhealthy instance would be removed from the load balancer but not automatically replaced.

Note that even with a Network Load Balancer, you can configure health checks to check specific paths in your application, which will help detect 5xx errors. When an instance consistently returns errors, it will fail the health check, be marked as unhealthy, removed from the load balancer, and then replaced by Auto Scaling.
Sources
Troubleshoot a Classic Load Balancer: Health checks - Elastic Load Balancing
Basic health reporting - AWS Elastic Beanstalk
Configuring a Network Load Balancer - AWS Elastic Beanstalk

You’re seeing this because a Network Load Balancer (NLB) with a TCP health check only verifies that the port is open; it does not evaluate HTTP status codes. So a service returning 5xx can stay “healthy” to the NLB.

If you need app-level health (e.g., fail an instance on sustained 5xx), switch your Elastic Beanstalk env to an Application Load Balancer (ALB) and configure an HTTP health check on your app (port 5000, path like /health). Also tell the Auto Scaling Group to use ELB health.