Changing health check options

Each listener for a standard accelerator routes requests only to healthy, active endpoints. When you add an endpoint, it must pass a health check to be considered healthy. AWS Global Accelerator also regularly sends health check requests to all endpoints on standard accelerators, to test their status. Global Accelerator automatically runs these regular health checks. After each health check is completed, the listener closes the connection that was established for the health check.

Note that if there aren't any healthy endpoints to route traffic to, Global Accelerator routes incoming client requests to all endpoints in the endpoint group. For more information, see Failover for unhealthy endpoints.

Details about how health checks work, and guidance about using health checks, depends on the type of endpoint resource. This topic provides information about how to work with health checks for different endpoint types, including steps for updating health check options in Global Accelerator (applies to EC2 instance or Elastic IP address endpoints).

Security and access for health checks

Global Accelerator requires that your router and firewall rules allow inbound traffic from the IP addresses associated with Amazon Route 53 health checkers to complete health checks for EC2 instance or Elastic IP address endpoints. To see the list of IP address ranges associated with Route 53 health checkers, see IP address ranges of Route 53 servers in the Amazon Route 53 Developer Guide.

Global Accelerator health checks work by receiving traffic for Route 53 health checks, which is forwarded to the configured health check port for the endpoint group. Typically, the ports configured for health checks match the listener configuration. If you configure a different port for health checks instead, review your security group configuration to make sure that you don't allow public traffic on the port.

For example, if your listener is configured on port 80, then your health check port is also 80. If you choose to configure health ports on another port, for example, port 83, then make sure that you configure your security groups to allow traffic on port 83 only from IP addresses that are in the IP address range for Route 53 health checks.

Health check guidance for different endpoint types

Overall, make sure that the health checks that you choose for endpoints with HTTP workloads are representative of the overall health of your application, and that you follow the guidance in the Security and access for health checks section.

In addition, review the following information about health checks for each endpoint type.

• For Network Load Balancer or Application Load Balancer endpoints, be aware of the following:

  • The health check options that you choose in Global Accelerator do not affect Network Load Balancers or Application Load Balancers that you've added as endpoints. That is, health check options that you specify in Global Accelerator are used for Amazon EC2 and Elastic IP address health checks, but not for health checks on load balancer endpoints.

    For load balancer endpoints, configure health checks by using Elastic Load Balancing configuration options. For more information, see Health checks for your target groups.

  • Global Accelerator considers a Network Load Balancer or Application Load Balancer healthy if there is at least one healthy Availability Zone. An Availability Zone is healthy if all load balancer target groups in that Availability Zone are healthy. For more information, see Health checks for your target groups.

• For EC2 instance or Elastic IP address endpoints, be aware of the following:

  • When you add EC2 instance or Elastic IP address endpoints to a listener configured with TCP, you can specify the port to use for health checks. By default, if you don't specify a port for health checks, Global Accelerator uses the listener port that you specified for your accelerator.

  • When you add these endpoint types with a UDP listener, Global Accelerator uses the listener port and the TCP protocol for health checks, so you must have a TCP server on your endpoint.

    Make sure to check that the port that you've configured for the TCP server on each endpoint is the same as the port that you specify for the health check in Global Accelerator. If the port numbers aren't the same, or if you haven't set up a TCP server for the endpoint, Global Accelerator marks the endpoint as unhealthy, regardless of the endpoint's health.

  • Make sure to follow the guidance for security and access when you configure ports for health checks for your EC2 instance or Elastic IP address endpoints.

Setting health check options

You can add the following health check options for an endpoint group.

Health check port

The port to use when Global Accelerator performs health checks on endpoints that are part of this endpoint group.

Note that you can't set a port override for health check ports.

Health check protocol

The protocol to use when Global Accelerator performs health checks on endpoints that are part of this endpoint group.

Health check interval

The interval, in seconds, between each health check for an endpoint.

Threshold count

The number of consecutive health checks required before considering an unhealthy target healthy or a healthy target unhealthy.