Menu
Elastic Load Balancing
Developer Guide

Troubleshooting Elastic Load Balancing: Health Checks

Your load balancer checks the health of its registered instances using either the default health check configuration provided by Elastic Load Balancing or a custom health check configuration that you specify. The health check configuration contains information such as the protocol, ping port, ping path, response timeout, and health check interval. For more information health checks and how to check the health status of your registered EC2 instances, see Configure Health Checks.

If the current state of some or all your instances is OutOfService and the description field displays the message that the Instance has failed at least the Unhealthy Threshold number of health checks consecutively, the instances have failed the load balancer health check. The following are the issues to look for, the potential causes, and the steps you can take to resolve the issues.

Connection to the instances has timed out

Problem: Health check requests from your load balancer to your EC2 instances are timing out or failing intermittently.

First, verify the issue by connecting directly with the instance. We recommend that you connect to your instance from within the network using the private IP address of the instance.

Use the following command for a TCP connection:

telnet private-IP-address-of-the-instance port

Use the following command for an HTTP or HTTPS connection:

curl –I private-IP-address-of-the-instance:port/health-check-target-page

If you are using an HTTP/HTTPS connection and getting a non-200 response, see Health check target page error. If you are able to connect directly to the instance, check for the following:

Cause 1: The instance is failing to respond within the configured response timeout period.

Solution 1: Adjust the response timeout settings in your load balancer health check configuration.

Cause 2: The instance is under significant load and is taking longer than your configured response timeout period to respond.

Solution 2:

  • Check the monitoring graph for over-utilization of CPU. For information, see Get Statistics for a Specific EC2 Instance in the Amazon EC2 User Guide for Linux Instances.

  • Check the utilization of other application resources, such as memory or limits, by connecting to your EC2 instances.

  • If necessary, add more instances or enable Auto Scaling. For more information, see What is Auto Scaling? in the Auto Scaling User Guide.

Cause 3: If you are using an HTTP or an HTTPS connection and the health check is being performed on a target page specified in the ping path field (for example, HTTP:80/index.html), the target page might be taking longer to respond than your configured timeout.

Solution 3: Use a simpler health check target page or adjust the health check interval settings.

Connection to your Internet-facing load balancer launched in a VPC has timed out

Problem: Health check requests are not reaching your instances launched in a VPC because the front-end connection (client to load balancer) has timed out.

Cause: Your Internet-facing load balancer is attached to a private subnet.

Solution: Verify that the VPC has an Internet gateway and that the route table has a route to the Internet Gateway.

Health check target page error

Problem: An HTTP GET request issued to the instance on the specified ping port and the ping path (for example, HTTP:80/index.html) receives a non-200 response code. Or, some instances are failing the health check and some instances are healthy.

Cause 1: No target page is configured on some or all the instances.

Solution 1: Create a target page (for example, index.html) on all the registered instances.

Cause 2: The value of the Content-Length header in the response is not set.

Solution 2: If the response includes a body, then either set the Content-Length header to a value greater than or equal to zero, or set the Transfer-Encoding value to 'chunked'.

Cause 3: The application on the instances is not configured to receive request from the load balancer.

Solution 3: Check the application on your instance to investigate the cause for non-200 response.

Public key authentication is failing

Problem: A load balancer configured to use the HTTPS or SSL protocol with back-end authentication enabled fails public key authentication.

Cause: The public key on the SSL certificate does not match the public key configured on the load balancer. Use the s_client command to see the list of server certificates in the certificate chain. For more information, see https://www.openssl.org/docs/apps/s_client.html.

Solution: Your might need to update your SSL certificate. If your SSL certificate is current, try re-installing it on your load balancer. For more information, see Replace the SSL Certificate for Your Load Balancer.

Instance is not receiving traffic from the load balancer

Problem: The security group for the instance is blocking the traffic from the load balancer.

Do a packet capture on the instance to verify the issue. Use the following command:

# tcpdump port health-check-port

Cause 1: The security group associated with the instance does not allow traffic from the load balancer.

Solution 1: Edit the instance security group to allow traffic from the load balancer. Add a rule to allow all traffic from the load balancer security group.

Cause 2: The security group of your load balancer in a VPC does not allow traffic to the EC2 instances.

Solution 2: Edit the security group of your load balancer to allow traffic to the subnets and the EC2 instances.

For information about managing security groups for EC2-Classic, see Security Groups for Back-end Instances in EC2-Classic.

For information about managing security groups for a VPC, see Security Groups for Load Balancers in a VPC.

Ports on instance are not open

Problem: The health check sent to the EC2 instance by the load balancer is blocked by the port or a firewall.

Verify the issue by using the following command:

netstat –ant

Cause: The specified health port or the listener port (if configured differently) is not open. Both the port specified for the health check and the listener port must be open and listening.

Solution: Open up the listener port and the port specified in your health check configuration (if configured differently) on your instances to receive load balancer traffic.

Instances in an Auto Scaling group are failing the ELB health check

Problem: Instances in your Auto Scaling group pass the default Auto Scaling health check but fail the ELB health check.

Cause: Auto Scaling uses EC2 status checks to detect hardware and software issues with the instances, but the load balancer performs health checks by sending a request to the instance and waiting for a 200 response code, or by establishing a TCP connection (for a TCP-based health check) with the instance.

An instance might fail the ELB health check because an application running on the instance has issues that cause the load balancer to consider the instance out of service. This instance might pass the Auto Scaling health check; it would not be replaced by the Auto Scaling policy because it is considered healthy based on the EC2 status check.

Solution: Use the ELB health check for your Auto Scaling group. When you use the ELB health check, Auto Scaling determines the health status of your instances by checking the results of both the instance status check and the ELB health check. For more information, see Add an Elastic Load Balancing Health Check to your Auto Scaling Group in the Auto Scaling User Guide.