AWS NLB to ECS Cluster failing health check

Question

I've configured a Network Load Balancer to route TCP traffic to an ECS cluster over port 80

The ECS cluster is running an ASP.NET Core 2.2 API task in Fargate configuration. The base route for the API is http://ip_address:80/api/v1/

At first, everything works as expected, I can hit the load balancer's DNS and route requests to the appropriate .NET API route

However, eventually the health check on the NLB fails, the containers are drained and new container replace them.

With an Application Load Balancer I would just configure the health check to hit a /healthcheck route on the API and everything would be fine.

However, a Network Balancer can't be configured this way.

My question is: What is the strategy for carrying out health checks on an ECS container?

Answer 1

OK so the answer was fairly simple...

You just set up a resource on the / path so that the NLB get's a response

My concrete solution for asp.net core:

ASP.NET Core 2.2 has a built in package for this. It was just a matter of adding the following in the Startup.cs class

Add the service to service collection

public void ConfigureServices(IServiceCollection services)
{
    services.AddHealthChecks();
}

Configure the health check

public void Configure(IApplicationBuilder app, IHostingEnvironment env, ILoggerFactory loggerFactory)
{
    // Network Load Balancer hits this path for the health check
    app.UseHealthChecks("/", 80);
    ...
}