Determine Amazon ECS task health using container health checks

When you create a task definition, you can configure a health check for you containers. Health checks are commands that run locally on a container and validate application health and availability.

The Amazon ECS container agent only monitors and reports on the health checks that are specified in the task definition. Amazon ECS doesn't monitor Docker health checks that are embedded in a container image but aren't specified in the container definition. Health check parameters that are specified in a container definition override any Docker health checks that exist in the container image.

When a health check is defined in a task definition, the container runs the health check process inside the container, and then evaluate the exit code to determine the application health.

The health check consists the following parameters:

• Command – The command that the container runs to determine if it's healthy. The string array can start with CMD to run the command arguments directly, or CMD-SHELL to run the command with the container's default shell.

• Interval – The period of time (in seconds) between each health check.

• Timeout – The period of time (in seconds) to wait for a health check to succeed before it's considered a failure.

• Retries – The number of times to retry a failed health check before the container is considered unhealthy.

• Start period – The optional grace period to provide containers time to bootstrap in before failed health checks count towards the maximum number of retries.

For information about how to specify a health check in a task definition, see Health check.

The following describes the possible health status values for a container:

• HEALTHY–The container health check has passed successfully.

• UNHEALTHY–The container health check has failed.

• UNKNOWN–The container health check is being evaluated, there's no container health check defined, or Amazon ECS doesn't have the health status of the container.

The health check commands run on the container. Therefore you must include the commands in the container image.

The health check connects to the application through the container's loopback interface at localhost or 127.0.0.1. An exit code of 0 indicates success, and non-zero exit code indicates failure.

Consider the following when using container health checks:

• Container health checks require version 1.17.0 or greater of the Amazon ECS container agent.

• Container health checks are supported for Fargate tasks if you're using Linux platform version 1.1.0 or greater or Windows platform version 1.1.0 or greater

How Amazon ECS determines task health

Containers that are essential and have health check command in the task definition are the only ones considered to determine the task health.

The following rules are evaluated in order:

1. If the status of one essential container is UNHEALTHY, then the task status is UNHEALTHY.

2. If he status of one essential container is UNKNOWN, then the task status is UNKNOWN.

3. If the status of all essential containers are HEALTHY, then the task status is HEALTHY.

Consider the following task health example with 2 essential containers.

Container 1 health	Container 2 health	Task health
UNHEALTHY	UNKNOWN	UNHEALTHY
UNHEALTHY	HEALTHY	UNHEALTHY
HEALTHY	UNKNOWN	UNKNOWN
HEALTHY	HEALTHY	HEALTHY

Consider the following task health example with 3 containers.

Container 1 health	Container 2 health	Container 3 health	Task health
UNHEALTHY	UNKNOWN	UNKNOWN	UNHEALTHY
UNHEALTHY	UNKNOWN	HEALTHY	UNHEALTHY
UNHEALTHY	HEALTHY	HEALTHY	UNHEALTHY
HEALTHY	UNKNOWN	HEALTHY	UNKNOWN
HEALTHY	UNKNOWN	UNKNOWN	UNKNOWN
HEALTHY	HEALTHY	HEALTHY	HEALTHY

How health checks are affected by agent disconnects

If the Amazon ECS container agent becomes disconnected from the Amazon ECS service, this won't cause a container to transition to an UNHEALTHY status. This is by design, to ensure that containers remain running during agent restarts or temporary unavailability. The health check status is the "last heard from" response from the Amazon ECS agent, so if the container was considered HEALTHY prior to the disconnect, that status will remain until the agent reconnects and another health check occurs. There are no assumptions made about the status of the container health checks.