Amazon ECS clusters for the Fargate launch type

Amazon ECS capacity providers manage the scaling of infrastructure for tasks in your clusters. Each cluster can have one or more capacity providers and an optional capacity provider strategy. The capacity provider strategy determines how the tasks are spread across the cluster's capacity providers. When you run a standalone task or create a service, you either use the cluster's default capacity provider strategy or a capacity provider strategy that overrides the default one.

When you run your tasks on AWS Fargate, you do not need to create or manage the capacity. You just need to associate any of the following pre-defined capacity providers with the cluster:

• Fargate

• Fargate Spot

With Amazon ECS on AWS Fargate capacity providers, you can use both Fargate and Fargate Spot capacity with your Amazon ECS tasks.

With Fargate Spot, you can run interruption tolerant Amazon ECS tasks at a rate that's discounted compared to the Fargate price. Fargate Spot runs tasks on spare compute capacity. When AWS needs the capacity back, your tasks are interrupted with a two-minute warning. Fargate Spot only supports Linux tasks with the X86_64 architecture on platform version 1.3.0 or later .

When tasks that use the Fargate and Fargate Spot capacity providers are stopped, the task state change event is sent to Amazon EventBridge. The stopped reason describes the cause. For more information, see Amazon ECS task state change events.

A cluster can contain a mix of Fargate and Auto Scaling group capacity providers. However, a capacity provider strategy can only contain either Fargate or Auto Scaling group capacity providers, but not both. For more information, see Auto Scaling Group Capacity Providers.

Consider the following when using capacity providers:

• You must associate a capacity provider with a cluster before you associate it with the capacity provider strategy.

• You can specify a maximum of 20 capacity providers for a capacity provider strategy.

• You can't update a service using an Auto Scaling group capacity provider to use a Fargate capacity provider. The opposite is also the case.

• In a capacity provider strategy, if no weight value is specified for a capacity provider in the console, then the default value of 1 is used. If using the API or AWS CLI, the default value of 0 is used.

• When multiple capacity providers are specified within a capacity provider strategy, at least one of the capacity providers must have a weight value that's greater than zero. Any capacity providers with a weight of zero aren't used to place tasks. If you specify multiple capacity providers in a strategy with all the same weight of zero, then any RunTask or CreateService actions using the capacity provider strategy fail.

• In a capacity provider strategy, only one capacity provider can have a defined base value. If no base value is specified, the default value of zero is used.

• A cluster can contain a mix of both Auto Scaling group capacity providers and Fargate capacity providers. However, a capacity provider strategy can only contain Auto Scaling group or Fargate capacity providers, but not both.

• A cluster can contain a mix of services and standalone tasks that use both capacity providers and launch types. A service can be updated to use a capacity provider strategy rather than a launch type. However, you must force a new deployment when doing so.

Fargate Spot termination notices

During periods of extremely high demand, Fargate Spot capacity might be unavailable. This can cause Fargate Spot tasks to be delayed. When this happens, Amazon ECS services retry launching tasks until the required capacity becomes available. Fargate doesn't replace Spot capacity with on-demand capacity.

When tasks using Fargate Spot capacity are stopped due to a Spot interruption, a two-minute warning is sent before a task is stopped. The warning is sent as a task state change event to Amazon EventBridge and as a SIGTERM signal to the running task. If you use Fargate Spot as part of a service, then in this scenario the service scheduler receives the interruption signal and attempts to launch additional tasks on Fargate Spot if there's capacity available. A service with only one task is interrupted until capacity is available. For more information about a graceful shutdown, see Graceful shutdowns with ECS .

To ensure that your containers exit gracefully before the task stops, you can configure the following:

• A stopTimeout value of 120 seconds or less can be specified in the container definition that the task is using. The default stopTimeout value is 30 seconds. You can specify a longer stopTimeout value to give yourself more time between the moment that the task state change event is received and the point in time when the container is forcefully stopped. For more information, see Container timeouts.

• The SIGTERM signal must be received from within the container to perform any cleanup actions. Failure to process this signal results in the task receiving a SIGKILL signal after the configured stopTimeout and may result in data loss or corruption.

The following is a snippet of a task state change event. This snippet displays the stopped reason and stop code for a Fargate Spot interruption.

{
  "version": "0",
  "id": "9bcdac79-b31f-4d3d-9410-fbd727c29fab",
  "detail-type": "ECS Task State Change",
  "source": "aws.ecs",
  "account": "111122223333",
  "resources": [
    "arn:aws:ecs:us-east-1:111122223333:task/b99d40b3-5176-4f71-9a52-9dbd6f1cebef"
  ],
  "detail": {
    "clusterArn": "arn:aws:ecs:us-east-1:111122223333:cluster/default",
    "createdAt": "2016-12-06T16:41:05.702Z",
    "desiredStatus": "STOPPED",
    "lastStatus": "RUNNING",
    "stoppedReason": "Your Spot Task was interrupted.",
    "stopCode": "SpotInterruption",
    "taskArn": "arn:aws:ecs:us-east-1:111122223333:task/b99d40b3-5176-4f71-9a52-9dbd6fEXAMPLE",
    ...
  }
}

The following is an event pattern that's used to create an EventBridge rule for Amazon ECS task state change events. You can optionally specify a cluster in the detail field. Doing so means that you will receive task state change events for that cluster. For more information, see Creating an EventBridge Rule in the Amazon EventBridge User Guide.

{
    "source": [
        "aws.ecs"
    ],
    "detail-type": [
        "ECS Task State Change"
    ],
    "detail": {
        "clusterArn": [
            "arn:aws:ecs:us-west-2:111122223333:cluster/default"
        ]
    }
}