brandon/oam

Fork 0

mirror of https://gitea.com/mcereda/oam.git synced 2026-02-08 21:34:25 +00:00

Files

Michele Cereda cba9d9193e chore(aws/ecs): improve section about capacity providers

2026-01-14 13:19:07 +01:00

82 KiB

Raw Blame History

Elastic Container Service

TL;DR
How it works
Execution and task roles
Standalone tasks
Services
Launch type
1. EC2 launch type
2. Fargate launch type
3. External launch type
Capacity providers
1. Capacity provider strategies
2. EC2 capacity providers
3. Fargate for ECS
Resource constraints
Environment variables
Storage
1. EBS volumes
2. EFS volumes
3. Docker volumes
4. Bind mounts
Networking
Execute commands in tasks' containers
Scale the number of tasks automatically
1. Target tracking
Allow tasks to communicate with each other
1. ECS Service Connect
2. ECS service discovery
3. VPC Lattice
Scrape metrics using Prometheus
Send logs to a central location
1. FireLens
2. Fluent Bit or Fluentd
Secrets
Best practices
Troubleshooting
1. Invalid 'cpu' setting for task
Cost-saving measures
Further readings
1. Sources

TL;DR

Tasks are the basic unit of deployment.
They are instances of the set of containers specified in their own task definition.

Tasks model and run one or more containers, much like Pods in Kubernetes.
Containers cannot run on ECS unless encapsulated in a task.

Standalone tasks start a single task, which is meant to perform some work to completion and then stop (much like batch processes would).
Services run and maintain a defined number of instances of the same task simultaneously, which are meant to stay active and act as replicas of some service (much like web servers would).

Tasks are executed depending on their launch type and capacity providers:

On EC2 instances that one owns, manages, and pays for.
On Fargate (an AWS-managed serverless environment for containers execution).

Unless explicitly restricted or capped, containers in tasks get access to all the CPU and memory capacity available on the host running it.

By default, containers behave like other Linux processes with respect to access to resources like CPU and memory.
Unless explicitly protected and guaranteed, all containers running on the same host share CPU, memory, and other resources much like normal processes running on that host share those very same resources.

Specify the execution role to allow ECS components to call AWS services when starting tasks.
Specify the task role to allow a task's containers to call AWS services.

Usage

# List services.
aws ecs list-services --cluster 'clusterName'

# Scale services.
aws ecs update-service --cluster 'clusterName' --service 'serviceName' --desired-count '0'
aws ecs update-service --cluster 'clusterName' --service 'serviceName' --desired-count '10' --no-cli-pager

# Wait for services to be running.
aws ecs wait services-stable --cluster 'clusterName' --services 'serviceName' …

# Delete services.
# Cannot really be deleted if scaled above 0.
aws ecs delete-service --cluster 'clusterName' --service 'serviceName'
aws ecs delete-service --cluster 'clusterName' --service 'serviceName' --force

# List task definitions.
aws ecs list-task-definitions --family-prefix 'familyPrefix'

# Deregister task definitions.
aws ecs deregister-task-definition --task-definition 'taskDefinitionArn'

# Delete task definitions.
# The task definition must be deregistered.
aws ecs delete-task-definitions --task-definitions 'taskDefinitionArn' …

# List tasks.
aws ecs list-tasks --cluster 'clusterName'
aws ecs list-tasks --cluster 'clusterName' --service-name 'serviceName'

# Get information about tasks.
aws ecs describe-tasks --cluster 'clusterName' --tasks 'taskIdOrArn' …

# Wait for tasks to be running.
aws ecs wait tasks-running --cluster 'clusterName' --tasks 'taskIdOrArn' …

# Access shells on containers in ECS.
aws ecs execute-command \
  --cluster 'clusterName' --task 'taskId' --container 'containerName' \
  --interactive --command '/bin/bash'

Real world use cases

# Get the ARNs of tasks for specific services.
aws ecs list-tasks --cluster 'testCluster' --service-name 'testService' --query 'taskArns' --output 'text'

# Get the private IP Address of containers.
aws ecs describe-tasks --output 'text' \
  --cluster 'testCluster' --tasks 'testTask' \
  --query "tasks[].attachments[].details[?(name=='privateDnsName')].value"

# Connect to the private DNS name of containers in ECS.
curl -fs "http://$( \
  aws ecs describe-tasks --cluster 'testCluster' --tasks "$( \
    aws ecs list-tasks --cluster 'testCluster' --service-name 'testService' --query 'taskArns' --output 'text' \
  )" --query "tasks[].attachments[].details[?(name=='privateDnsName')].value" --output 'text' \
):8080"

# Get the image of specific containers.
aws ecs list-tasks --cluster 'someCluster' --service-name 'someService' --query 'taskArns[0]' --output 'text' \
| xargs -oI '%%' \
    aws ecs describe-tasks --cluster 'someCluster' --task '%%' \
      --query 'tasks[].containers[?name==`someContainer`].image' --output 'text'

# Delete services.
aws ecs delete-service --cluster 'testCluster' --service 'testService' --force

# Delete task definitions.
aws ecs list-task-definitions --family-prefix 'testService' --output 'text' --query 'taskDefinitionArns' \
| xargs -n '1' aws ecs deregister-task-definition --task-definition

# Wait for tasks to be running.
aws ecs list-tasks --cluster 'testCluster' --family 'testService' --output 'text' --query 'taskArns' \
| xargs -p aws ecs wait tasks-running --cluster 'testCluster' --tasks
while [[ $(aws ecs list-tasks --query 'taskArns' --output 'text' --cluster 'testCluster' --service-name 'testService') == "" ]]; do sleep 1; done

# Restart tasks.
# No real way to do that, just stop the tasks and new ones will be eventually started in their place.
# To mimic a blue-green deployment, scale the service up by doubling its tasks, then down again to the normal amount.
aws ecs update-service --cluster 'someCluster' --service 'someService' --desired-count '0' --no-cli-pager \
&& aws ecs update-service --cluster 'someCluster' --service 'someService' --desired-count '1' --no-cli-pager

How it works

Tasks must be registered in task definitions before they can be launched.

Tasks can be executed as Standalone tasks or services.
Whatever the launch type or capacity provider:

On launch, a task is created and moved to the PROVISIONING state.
While in this state, ECS needs to find compute capacity for the task and neither the task nor its containers exist.
ECS selects the appropriate compute capacity for the task based on its launch type or capacity provider configuration.

Tasks will fail immediately should there be not enough compute capacity for the task in the launch type or capacity provider.

When using a capacity provider with managed scaling enabled, tasks that can't be started due to a lack of compute capacity are kept in the PROVISIONING state while ECS provisions the necessary attachments.
ECS uses the container agent to pull the task's container images.
ECS starts the task's containers.
ECS moves the task into the RUNNING state.

Important

Task definition's parameters differ depending on the launch type.

Execution and task roles

Specifying the Execution Role in a task definition grants that IAM Role's permissions to the ECS container agent, allowing it to make calls to other AWS services when starting tasks.
This is required when ECS itself (and not the app in the task's container) needs to make calls to, i.e., pull images from ECRs, write logs to CloudWatch, or retrieve secrets from Secrets Manager.

The Execution Role must allow ecs.amazonaws.com to assume it.

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Sid": "AllowECSToAssumeThisVeryRole",
            "Effect": "Allow",
            "Principal": {
                "Service": "ecs.amazonaws.com"
            },
            "Action": "sts:AssumeRole"
        }
    ]
}

It is common practice to attach the Execution Role the AmazonECSTaskExecutionRolePolicy IAM Policy (or equivalent permissions) to grant it the minimum permissions required to run Tasks.

Warning

For ECS to be able to start a task (OR):

[easier] The execution role itself must trust ecs-tasks.amazonaws.com in addition to ecs.amazonaws.com.

 {
     "Version": "2012-10-17",
     "Statement": [
         {
             "Sid": "AllowECSToAssumeThisVeryRole",
             "Effect": "Allow",
             "Principal": {
                 "Service": [
                     "ecs.amazonaws.com",
+                    "ecs-tasks.amazonaws.com",
                 ]
             },
             "Action": "sts:AssumeRole"
         }
     ]
 }

The IAM User or Role that creates the ECS service must have iam:PassRole permission for both the execution role and the task role.

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Sid": "AllowPassExecutionAndTaskRoles",
            "Effect": "Allow",
            "Action": "iam:PassRole",
            "Resource": [
                "arn:aws:iam::012345678901:role/SomeServiceECSExecutionRole",
                "arn:aws:iam::012345678901:role/SomeServiceECSTaskRole"
            ]
        }
    ]
}

Specifying the Task Role in a task definition grants that IAM Role's permissions to the task's container.
This is required when the apps in the task's containers (and not ECS) needs to make calls to, i.e., recover a file from S3 or read values from SQS.
This IAM Role must allow ecs-tasks.amazonaws.com to assume it.

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Sid": "AllowECSTasksToAssumeThisVeryRole",
            "Effect": "Allow",
            "Principal": {
                "Service": "ecs-tasks.amazonaws.com"
            },
            "Action": "sts:AssumeRole"
        }
    ]
}

Standalone tasks

Refer Amazon ECS standalone tasks.

Meant to perform some work, then stop similarly to batch processes.

Can be executed on schedules using the EventBridge Scheduler.

Services

Refer Amazon ECS services.

Services execute and maintain a defined number of instances of the same task simultaneously in a cluster.

Tasks executed in services are meant to stay active until decommissioned, much like web servers.
Should any of such tasks fail or stops, the service scheduler will launch another instance of the same task to replace the one that failed.

One can optionally expose services behind a load balancer to distribute traffic across the tasks that the service manages.

The service scheduler will replace unhealthy tasks should a container health check or a load balancer target group health check fail.
This depends on the maximumPercent and desiredCount parameters in the service's definition.

If a task is marked unhealthy, the service scheduler will first start a replacement task. Then:

If the replacement task is HEALTHY, the service scheduler stops the unhealthy task.
If the replacement task is also UNHEALTHY, the scheduler will stop either the unhealthy replacement task or the existing unhealthy task to get the total task count equal to the desiredCount value.

Should the maximumPercent parameter limit the scheduler from starting a replacement task first, the scheduler will:

Stop unhealthy tasks one at a time at random in order to free up capacity.
Start a replacement task.

The start and stop process continues until all unhealthy tasks are replaced with healthy tasks.
Should the total task count still exceed desiredCount once all unhealthy tasks have been replaced and only healthy tasks are running, healthy tasks are stopped at random until the total task count equals desiredCount.

The service scheduler includes logic that throttles how often tasks are restarted if they repeatedly fail to start.
If a task is stopped without having entered the RUNNING state, the service scheduler starts to slow down the launch attempts and sends out a service event message.
This prevents unnecessary resources from being used for failed tasks before one can resolve the issue.
On service update, the service scheduler resumes normal scheduling behavior.

Available service scheduler strategies:

REPLICA: places and maintains the desired number of tasks across one's cluster.
By default, tasks are spread across Availability Zones. Use task placement strategies and constraints to customize task placement decisions.
DAEMON: deploys exactly one task on each active container instance meeting all of the task placement constraints for the task.
There is no need to specify a desired number of tasks, a task placement strategy, or use Service Auto Scaling policies when using this strategy.

Important

Fargate does not support the DAEMON scheduling strategy.

Launch type

Defines the underlying infrastructure effectively running containers within ECS.

{
    "serviceName": "some-ecs-service",
    … ,
    "launchType": "FARGATE"
}

The setting is currently mutually exclusive with capacity provider strategies.
Prefer using those to leverage enhanced flexibility and advanced features for capacity management.

EC2 launch type

Starts tasks onto registered EC2 instances.

Instances can be registered:

Manually.
Automatically, by using the cluster auto scaling feature to dynamically scale the cluster's compute capacity.

Fargate launch type

Starts tasks on dedicated, managed EC2 instances that are not reachable by the users.

Instances are automatically provisioned, configured, and registered to scale one's cluster capacity.
The service takes care itself of all the infrastructure management for the tasks.

External launch type

Manages containers running outside the ECS ecosystem, e.g., on-premises servers, other cloud providers, or hybrid deployments.

Capacity providers

Refer Capacity providers.

Clusters can contain a mix of tasks that are hosted on Fargate, Amazon EC2 instances, or external instances.
Tasks can run on Fargate or EC2 infrastructure, as defined by their launch type or a capacity provider strategy.
Capacity providers offer enhanced flexibility and advanced features for capacity management compared to launch types.

Each cluster can have one or more capacity providers, and one optional capacity provider strategy.

To run tasks on Fargate, one only needs to associate one or more of the pre-defined Fargate-specific capacity providers (FARGATE, FARGATE_SPOT) with the cluster.
Leveraging the Fargate providers lifts the need to create or manage that cluster's capacity.

Warning

One cannot update a service that is using only an Auto Scaling Group capacity provider to use a Fargate-specific one, and vice versa.

Clusters can contain a mix of Fargate and Auto Scaling group capacity providers.

{
    "clusterName": "some-cluster",
    "capacityProviders": [
        "FARGATE",
        "FARGATE_SPOT",
        "some-custom-ec2-capacity-provider"
    ]
}

Capacity provider strategies

Capacity provider strategies determine how tasks are spread across a cluster's capacity providers.

One must associate at least one capacity provider with a cluster before specifying a capacity provider strategy for it.
Strategies allow to specify a maximum of 20 capacity providers.

Warning

Even if clusters can contain a mix of Fargate and Auto Scaling group capacity providers, capacity provider strategies can currently only contain either Fargate or Auto Scaling group capacity providers, but not both.

One can assign a default capacity provider strategy to a cluster.
Clusters that do not have a default capacity provider strategy will spread their tasks onto whatever configured provider will have enough capacity at the moment of deployment.

{
    "clusterName": "some-cluster",
    "capacityProviders": [
        "FARGATE",
        "FARGATE_SPOT"
    ],
    "defaultCapacityProviderStrategy": [
        {
            "capacityProvider": "FARGATE_SPOT",
            "weight": 100
        },
        {
            "capacityProvider": "FARGATE",
            "weight": 0
        }
    ]
}

When running a standalone task or creating a service, one can specify a capacity provider strategy to override the cluster's default one.

Important

The cluster's default capacity provider strategy only applies when a task or service specifies neither a launch type nor its own capacity provider strategy.
Should a task or service be configured with either of these parameters, it will ignore the cluster's default strategy.

Override the cluster's default strategy

{
    "serviceName": "some-ecs-service",
    … ,
    "capacityProviderStrategy": [
        {
            "capacityProvider": "FARGATE",
            "weight": 1,
            "base": 1
        },
        {
            "capacityProvider": "FARGATE_SPOT",
            "weight": 2
        },
        {
            "capacityProvider": "some-custom-ec2-capacity-provider",
            "weight": 0
        }
    ]
}

Strategies' weight value defaults to 1 when creating it from the Console, and to 0 if using the API or CLI.

A strategy's capacity provider can have a defined base value. This determines how many guaranteed tasks that provider will be given as minimum when enough replicas are requested.
Setting the base value higher than the service or standalone task's desiredCount only results in desiredCount tasks being placed on that provider. If no base value is specified for a provider, it defaults to 0.

Warning

Only one capacity provider can have a base value other than 0 in a strategy.

The weight value determines the relative ratio of tasks to execute over the long run.
This value is taken into account only after the base values are satisfied.
When multiple capacity providers are specified within a strategy, at least one of the providers must have a weight value greater than zero (0).

Aside from their base value (if not 0), capacity providers with a weight value of 0 are not considered when the scheduler decides where to place tasks. Should all providers in a strategy have a weight of 0, any RunTask or CreateService actions using that strategy will fail.

The weight ratio is computed by:

Summing up all providers' weights.
Determining the percentage per provider.

Simple example

Provider 1 is FARGATE, with weight of 1.
Provider 2 is FARGATE_SPOT, with weight of 3.

{
    …
    "capacityProviderStrategy": [
        {
            "capacityProvider": "FARGATE",
            "weight": 1
        },
        {
            "capacityProvider": "FARGATE_SPOT",
            "weight": 3
        }
    ]
}

Sum of weights: 1 + 3 = 4.
Percentage per provider:

FARGATE: 1 / 4 = 0.25.
FARGATE_SPOT: 3 / 4 = 0.75.

FARGATE will receive 25% of the tasks, while FARGATE_SPOT will receive the remaining 75%.

More advanced example

Provider 1 is FARGATE, with a weight of 1.
Provider 2 is FARGATE_SPOT, with a weight of 19.
Provider 3 is some-custom-ec2-capacity-provider, with a weight of 0 and base of 2.

{
    …
    "capacityProviderStrategy": [
        {
            "capacityProvider": "FARGATE",
            "weight": 1
        },
        {
            "capacityProvider": "FARGATE_SPOT",
            "weight": 19
        },
        {
            "capacityProvider": "some-custom-ec2-capacity-provider",
            "base": 2,
            "weight": 0
        }
    ]
}

some-custom-ec2-capacity-provider will run tasks just for being the provider with the base value defined.
After assigning it the 2 base tasks as per configuration, the scheduler will just ignore some-custom-ec2-capacity-provider due to its weight being 0.

Sum of the remaining weights: 1 + 19 = 20.
Percentage per provider:

FARGATE: 1 / 20 = 0.05.
FARGATE_SPOT: 19 / 20 = 0.95.

FARGATE will receive 5% of the tasks over the base, while FARGATE_SPOT will receive 95% of them.

A cluster can contain a mix of services and standalone tasks that use both capacity providers and launch types.
Services can be updated to use a capacity provider strategy instead of a launch type, but one will need to force a new deployment to do so.

EC2 capacity providers

Refer Amazon ECS capacity providers for the EC2 launch type.

When using EC2 instances for capacity, one really uses Auto Scaling groups to manage the EC2 instances.
Auto Scaling helps ensure that one has the correct number of EC2 instances available to handle the application's load.

Fargate for ECS

Refer AWS Fargate Spot Now Generally Available and Amazon ECS clusters for Fargate.

ECS can run tasks on the Fargate and Fargate Spot capacity when they are associated with a cluster.

The Fargate provider runs tasks on on-demand compute capacity.

Fargate Spot is intended for interruption tolerant tasks.
It runs tasks on spare compute capacity. This makes it cost less than Fargate's normal price, but allows AWS to interrupt those tasks when it needs capacity back.

During periods of extremely high demand, Fargate Spot capacity might be unavailable.
When this happens, ECS services retry launching tasks until the required capacity becomes available.

ECS sends a two-minute warning before Spot tasks are stopped due to a Spot interruption.
This warning is sent as a task state change event to EventBridge and as a SIGTERM signal to the running task.

EventBridge event example

{
    "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",
        …
    }
}

When Spot tasks are terminated, the service scheduler receives the interruption signal and attempts to launch additional tasks on Fargate Spot, possibly from a different Availability Zone, provided such capacity is available.

Fargate will not replace Spot capacity with on-demand capacity.

Ensure containers exit gracefully before the task stops by configuring the following:

Specify a stopTimeout value of 120 seconds or less in the container definition that the task is using.
The default value is 30 seconds. A higher value will provide more time between the moment that the task's state change event is received and the point in time when the container is forcefully stopped.
Make sure the SIGTERM signal is caught from within the container, and that it triggers any needed cleanup.
Not processing this signal results in the task receiving a SIGKILL signal after the configured stopTimeout value, which may result in data loss or corruption.

Resource constraints

ECS uses the CPU period and the CPU quota to control the task's CPU hard limits as a whole.
When specifying CPU values in task definitions, ECS translates that value to the CPU period and CPU quota settings that apply to the cgroup running all the containers in the task.

The CPU quota controls the amount of CPU time granted to a cgroup during a given CPU period. Both settings are expressed in terms of microseconds.
When the CPU quota equals the CPU period, a cgroup can execute up to 100% on one vCPU (or any other fraction that totals to 100% for multiple vCPUs). The CPU quota has a maximum of 1000000µs, and the CPU period has a minimum of 1ms.
Use these values to set the limits for the tasks' CPU count.

When changing the CPU period without changing the CPU quota, the task will have different effective limits than what is specified in the task definition.

The 100ms period allows for vCPUs ranging from 0.125 to 10.

Task-level CPU and memory parameters are ignored for Windows containers.

The cpu value must be expressed in CPU units or vCPUs. A CPU unit is 1/1024 of a full vCPU.
vCPUs values are converted to CPU units when task definitions are registered.

The memory value can be expressed in MiB or GB.
GB values are converted to MiB when tasks definitions are registered.

These fields are optional for tasks hosted on EC2.
Such tasks support CPU values between 0.25 and 10 vCPUs. these fields are optional

Task definitions specifying FARGATE as value for the requiresCompatibilities attribute, even if they also specify the EC2 value, are required to set both settings and to set them to one of the couples specified in the next table.
Fargate task definitions support only those specific values for tasks' CPU and memory.

CPU units	vCPUs	Memory values	Supported OSes	Notes
256	.25	512 MiB, 1 GB, or 2 GB	Linux
512	.5	Between 1 GB and 4 GB in 1 GB increments	Linux
1024	1	Between 2 GB and 8 GB in 1 GB increments	Linux, Windows
2048	2	Between 4 GB and 16 GB in 1 GB increments	Linux, Windows
4096	4	Between 8 GB and 30 GB in 1 GB increments	Linux, Windows
8192	8	Between 16 GB and 60 GB in 4 GB increments	Linux	Requires Linux platform >= 1.4.0
16384	16	Between 32 GB and 120 GB in 8 GB increments	Linux	Requires Linux platform >= 1.4.0

The task's settings are separate from the CPU and memory values that can be defined at the container definition level.

Reservations configure the minimum amount of resources that containers or tasks receive.
Using more than the reservation's amount is known as bursting.
ECS guarantees reservations. It doesn't place a task on an instance that cannot fulfill the task's reservation.

Limits are the maximum amount of resources that containers or tasks can use.
Attempts to use more CPU more than the limit results in throttling. Attempt to use more memory then the limit results in the container being stopped for OOM reasons.

Should both a container-level memory and memoryReservation value be set, the memory value must be higher than the memoryReservation value.
If specifying memoryReservation, that value is guaranteed to the container and subtracted from the available memory resources for the container instance that the container is placed on. Otherwise, the value of memory is used.

Swap usage is controlled at container-level.
Swap space must be enabled and allocated on the EC2 instance hosting the task, for the containers to use it. By default, ECS optimized AMIs do not have swap enabled. Also, Fargate does not support it.

maxSwap determines the total amount of swap memory in MiB a container can use.
It must be 0, or any positive integer number. Setting it to 0 disables swapping.
If omitted, the container uses the swap configuration for the container instance it is running on.

swappiness tunes a container's memory swappiness behavior.
It requires the maxSwap value to be set. If a value isn't specified for maxSwap, swappiness is ignored.
It accepts whole numbers between 0 and 100. 0 causes swapping to not occur unless required. 100 causes pages to be swapped aggressively.
If omitted, it defaults to 60.

{
    "containerDefinitions": [
        {
            "linuxParameters": {
                "maxSwap": 512,
                "swappiness": 10
            },
            …
        }
    ],
    …
}

Environment variables

Refer Amazon ECS environment variables.

ECS sets default environment variables for any task it runs.

$ aws ecs list-tasks --cluster 'devel' --service-name 'prometheus' --query 'taskArns' --output 'text' \
  | xargs -I '%%' aws ecs execute-command --cluster 'devel' --task '%%' --container 'prometheus' \
      --interactive --command 'printenv'

The Session Manager plugin was installed successfully. Use the AWS CLI to start a session.


Starting session with SessionId: ecs-execute-command-abcdefghijklmnopqrstuvwxyz
PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
HOSTNAME=ip-172-31-10-103.eu-west-1.compute.internal
AWS_CONTAINER_CREDENTIALS_RELATIVE_URI=/v2/credentials/abcdefgh-1234-abcd-9876-abcdefgh0123
AWS_DEFAULT_REGION=eu-west-1
AWS_EXECUTION_ENV=AWS_ECS_FARGATE
AWS_REGION=eu-west-1
ECS_AGENT_URI=http://169.254.170.2/api/abcdef0123456789abcdef0123456789-1111111111
ECS_CONTAINER_METADATA_URI=http://169.254.170.2/v3/abcdef0123456789abcdef0123456789-1111111111
ECS_CONTAINER_METADATA_URI_V4=http://169.254.170.2/v4/abcdef0123456789abcdef0123456789-1111111111
HOME=/root
TERM=xterm-256color
LANG=C.UTF-8


Exiting session with sessionId: ecs-execute-command-abcdefghijklmnopqrstuvwxyz.

Storage

Refer Storage options for Amazon ECS tasks.

Volume type	Launch type support	OS support	Persistence	Use cases
EBS volumes	EC2 Fargate	Linux	Can be persisted when used by a standalone task Ephemeral when attached to tasks maintained by a service	Transactional workloads
EFS volumes	EC2 Fargate	Linux	Persistent	Data analytics Media processing Content management Web serving
Docker volumes	EC2	Linux, Windows	Persistent	Provide a location for data persistence Sharing data between containers
Bind mounts	EC2 Fargate	Linux, Windows	Ephemeral	Data analytics Media processing Content management Web serving

EBS volumes

Refer Use Amazon EBS volumes with Amazon ECS.

One can attach at most one EBS volume to each ECS task, and it must be a new volume.
One cannot attach existing EBS volume to tasks. However, one can configure a new EBS volume at deployment to use the snapshot of an existing volume as starting point.

Provisioning volumes from snapshots of EBS volumes that contains partitions is not supported.

EBS volumes can be configured at deployment only for services that use the rolling update deployment type and the Replica scheduling strategy.

Containers in a task will be able to write to the mounted EBS volume only if the container runs as the root user.

ECS automatically adds the AmazonECSCreated and AmazonECSManaged reserved tags to attached volumes.
Should one remove these tags from the volumes, ECS won't be able to manage it anymore.

Volumes attached to tasks which are managed by a service are not preserved, and are always deleted upon task's termination.

One cannot configure EBS volumes for attachment to ECS tasks running on AWS Outposts.

EFS volumes

Refer Use Amazon EFS volumes with Amazon ECS.

Allows tasks with access to the same EFS volumes to share persistent storage.

Tasks must:

Reference the EFS volumes in the volumes attribute of their definition.
Reference the defined volumes in the mountPoints attribute in the containers' specifications.

{
    "volumes": [{
        "name": "myEfsVolume",
        "efsVolumeConfiguration": {
            "fileSystemId": "fs-1234",
            "rootDirectory": "/path/to/my/data",
            "transitEncryption": "ENABLED",
            "transitEncryptionPort": 9076,
            "authorizationConfig": {
                "accessPointId": "fsap-1234",
                "iam": "ENABLED"
            }
        }
    }],
    "containerDefinitions": [{
        "name": "container-using-efs",
        "image": "amazonlinux:2",
        "entryPoint": [
            "sh",
            "-c"
        ],
        "command": [ "ls -la /mount/efs" ],
        "mountPoints": [{
            "sourceVolume": "myEfsVolume",
            "containerPath": "/mount/efs",
            "readOnly": true
        }]
    }]
}

EFS file systems are supported on:

EC2 nodes using ECS-optimized AMI version 20200319 with container agent version 1.38.0.
Fargate since platform version 1.4.0 or later (Linux).

Not supported on external instances.

Docker volumes

Refer Use Docker volumes with Amazon ECS.

TODO

Only supported by EC2 or external instances.

Bind mounts

Refer Use bind mounts with Amazon ECS.

TODO

Mount files or directories from a host into a container.

Supported for tasks on both Fargate and EC2 instances.

Bind mounts are tied to the lifecycle of the container that uses them.
After all the containers using a specific bind mount stop, that data is removed.
The data can be tied to the lifecycle of an EC2 instance by specifying a host value in the task's definition.

Tasks running on Fargate receive a minimum of 20 GiB of ephemeral storage for bind mounts.
This can be increased up to a maximum of 200 GiB by specifying the ephemeralStorage parameter in the task's definition.

Networking

The networking behavior of tasks that are hosted on EC2 instances is dependent on the network mode that one defined in the task's definition.

In awsvpc network mode, each task is allocated its own Elastic Network Interface (ENI) and a primary private IPv4 address. This gives the task the same networking properties as EC2 instances.
AWS recommends using the awsvpc network mode, unless one has the specific need to use a different network mode.

In host network mode, the networking of the container is tied directly to the underlying host executing it.
Only supported for tasks hosted on EC2 instances, not supported when using Fargate.

With bridge mode, a virtual network bridge creates a layer between the host and the container's networking.
It allows to create port mappings to remap host ports to container ports. Mappings can be static or dynamic.
Only supported for tasks hosted on EC2 instances, not supported when using Fargate.

Tasks on Fargate are each provided an ENI with a primary private IP address, which allows them to use networking features such as VPC Flow Logs or PrivateLink.
When using a public subnet, one can optionally assign a public IP address to the task's ENI.
If the VPC is configured for dual-stack mode, and tasks are using a subnet with an IPv6 CIDR block, the tasks' ENI also receive an IPv6 address.

Fargate fully manages the ENIs it creates.
One cannot manually detach nor modify those ENIs. To release the ENIs for a task, stop the task.

A task can only have one ENI associated with it at a time.

Containers within the same task are placed on the same virtual network interface.
However, differently from Docker or Kubernetes, they must use localhost should they wish to communicate with each other. Container name-based DNS resolution (e.g. `postgresql://postgres:5432) will not work by default, and ECS will not create DNS records for container names inside a task.

Tasks on Fargate that need to pull a container image must have a route to the container registry.

An ECS service-linked role is required to provide ECS with the permissions to make calls to other AWS services on one's behalf.
Such role is automatically created when creating a cluster, or when creating or updating a service in the AWS Management Console.

Execute commands in tasks' containers

Refer Using Amazon ECS Exec to access your containers on AWS Fargate and Amazon EC2, A Step-by-Step Guide to Enabling Amazon ECS Exec, aws ecs execute-command results in TargetNotConnectedException The execute command failed due to an internal error and Amazon ECS Exec Checker.

Leverage ECS Exec, which in turn leverages SSM to create a secure channel between one's device and the target container.
It does so by bind-mounting the necessary SSM agent binaries into the container while the ECS (or Fargate) agent starts the SSM core agent inside the container.
The agent, when invoked, calls SSM to create the secure channel. In order to do so, the container's ECS task must have the proper IAM privileges for the SSM core agent to call the SSM service.

The SSM agent does not run as a separate container sidecar, but as an additional process inside the application container.
Refer ECS Execute-Command proposal for details.

The whole procedure is transparent and does not compel requirements changes in the container's content.

Requirements:

The required SSM components must be available on the EC2 instances hosting the container. Amazon's ECS optimized AMI and Fargate 1.4.0+ include their latest version already.
The container's image must have script and cat installed.
Required in order to have command logs uploaded correctly to S3 and/or CloudWatch.

The task's role (not the Task's execution role) must have specific permissions assigned.

Policy example

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Sid": "RequiredSSMPermissions",
            "Effect": "Allow",
            "Action": [
                "ssmmessages:CreateControlChannel",
                "ssmmessages:CreateDataChannel",
                "ssmmessages:OpenControlChannel",
                "ssmmessages:OpenDataChannel"
            ],
            "Resource": "*"
        },
        {
            "Sid": "RequiredGlobalCloudWatchPermissions",
            "Effect": "Allow",
            "Action": "logs:DescribeLogGroups",
            "Resource": "*"
        },
        {
            "Sid": "RequiredSpecificCloudWatchPermissions",
            "Effect": "Allow",
            "Action": [
                "logs:CreateLogStream",
                "logs:DescribeLogStreams",
                "logs:PutLogEvents"
            ],
            "Resource": [
                "arn:aws:logs:eu-west-1:012345678901:log-group:log-group-name",
                "arn:aws:logs:eu-west-1:012345678901:log-group:log-group-name:log-stream:log-stream-name"
            ]
        },
        {
            "Sid": "OptionalS3PermissionsIfSSMRecordsLogsInBuckets",
            "Effect": "Allow",
            "Action": [
                "s3:GetEncryptionConfiguration",
                "s3:PutObject"
            ],
            "Resource": [
                "arn:aws:s3:::ecs-exec-bucket",
                "arn:aws:s3:::ecs-exec-bucket/session-logs/*"
            ]
        },
        {
            "Sid": "OptionalKMSPermissionsIfSSMRecordsLogsInEncryptedBuckets",
            "Effect": "Allow",
            "Action": [
                "kms:Decrypt",
                "kms:GenerateDataKey"
            ],
            "Resource": "arn:aws:kms:eu-west-1:012345678901:key/abcdef01-2345-6789-abcd-ef0123456789"
        }
    ]
}

The service or the run-task command that start the task must have the enable-execute-command set to true.

Examples

aws ecs run-task … --enable-execute-command
aws ecs update-service --cluster 'stg' --service 'grafana' --enable-execute-command --force-new-deployment

new aws.ecs.Service(
    'whatever',
    { enableExecuteCommand: true,  …, },
);

Users initiating the execution:

Must install the Session Manager plugin for the AWS CLI.

Must be allowed the ecs:ExecuteCommand action on the ECS cluster.

Policy example

{
    "Version": "2012-10-17",
    "Statement": [{
        "Effect": "Allow",
        "Action": "ecs:ExecuteCommand",
        "Resource": "arn:aws:ecs:eu-west-1:012345678901:cluster/devel",
        "Condition": {
            "StringEquals": {
                "aws:ResourceTag/application": "someApp",
                "aws:ResourceTag/component": [
                    "someComponent",
                    "someOtherComponent"
                ],
                "ecs:container-name": "nginx"
            }
        }
    }]
}

Procedure:

Confirm that the task's ExecuteCommandAgent status is RUNNING and the enableExecuteCommand attribute is set to true.

Example

aws ecs describe-tasks --cluster 'devel' --tasks 'ef6260ed8aab49cf926667ab0c52c313' --output 'yaml' \
  --query 'tasks[0] | {
    "managedAgents": containers[].managedAgents[?@.name==`ExecuteCommandAgent`][],
    "enableExecuteCommand": enableExecuteCommand
  }'

aws ecs list-tasks --cluster 'devel' --service-name 'mimir' --query 'taskArns' --output 'text' \
| xargs \
    aws ecs describe-tasks --cluster 'devel' \
      --output 'yaml' --query 'tasks[0] | {
        "managedAgents": containers[].managedAgents[?@.name==`ExecuteCommandAgent`][],
        "enableExecuteCommand": enableExecuteCommand
      }' \
      --tasks

enableExecuteCommand: true
managedAgents:
- lastStartedAt: '2025-01-28T22:16:59.370000+01:00'
  lastStatus: RUNNING
  name: ExecuteCommandAgent

Execute the command.

Example

aws ecs execute-command --interactive --command 'df -h' \
  --cluster 'devel' --task 'ef6260ed8aab49cf926667ab0c52c313' --container 'nginx'

The Session Manager plugin was installed successfully. Use the AWS CLI to start a session.

Starting session with SessionId: ecs-execute-command-zobkrf3qrif9j962h9pecgnae8
Filesystem      Size  Used Avail Use% Mounted on
overlay          31G   12G   18G  40% /
tmpfs            64M     0   64M   0% /dev
shm             464M     0  464M   0% /dev/shm
tmpfs           464M     0  464M   0% /sys/fs/cgroup
/dev/nvme1n1     31G   12G   18G  40% /etc/hosts
/dev/nvme0n1p1  4.9G  2.1G  2.8G  43% /managed-agents/execute-command
tmpfs           464M     0  464M   0% /proc/acpi
tmpfs           464M     0  464M   0% /sys/firmware

Exiting session with sessionId: ecs-execute-command-zobkrf3qrif9j962h9pecgnae8.

Should one's command invoke a shell, one will gain interactive access to the container.
In this case, all commands and their outputs inside the shell session will be logged to S3 and/or CloudWatch. The shell invocation command and the user that invoked it will be logged in CloudTrail for auditing purposes as part of the ECS ExecuteCommand API call.

Should one's command invoke a single command, only the output of the command will be logged to S3 and/or CloudWatch. The command itself will still be logged in CloudTrail as part of the ECS ExecuteCommand API call.

Logging options are configured at the ECS cluster level.
The task's role will need to have IAM permissions to log the output to S3 and/or CloudWatch should the cluster be configured for the above options. If the options are not configured, then the permissions are not required.

Scale the number of tasks automatically

Refer Automatically scale your Amazon ECS service.

Scaling**out** increases the number of tasks, scaling-in decreases it.

ECS sends metrics in 1-minute intervals to CloudWatch.
Keep this in mind when tweaking the values for scaling.

Target tracking

Refer Target tracking scaling policies for Application Auto Scaling and How target tracking scaling for Application Auto Scaling works.

The only available metrics for the integrated checks are currently:

The service's average CPU utilization (ECSServiceCPUUtilization) for the last minute.
The service's average memory utilization (ECSServiceMemoryUtilization) for the last minute.
The service's Application Load Balancer's average requests count (ALBRequestCountPerTarget) for the last minute.

Allow tasks to communicate with each other

Refer How can I allow the tasks in my Amazon ECS services to communicate with each other? and Interconnect Amazon ECS services.

Tasks in a cluster are not normally able to communicate with each other.
Use ECS Service Connect, ECS service discovery or VPC Lattice to allow that.

ECS Service Connect

Refer Use Service Connect to connect Amazon ECS services with short names.

ECS Service Connect provides ECS clusters with the configuration they need for service-to-service discovery, connectivity, and traffic monitoring by building both service discovery and a service mesh in the clusters.

It provides:

The complete configuration services need to join the mesh.
A unified way to refer to services within namespaces that does not depend on the VPC's DNS configuration.
Standardized metrics and logs to monitor all the applications.

The feature creates a virtual network of related services.
The same service configuration can be used across different namespaces to run independent yet identical sets of applications.

When using Service Connect, ECS dynamically manages Service Connect endpoints for each task as they start and stop. It does so by injecting the definition of a sidecar proxy container in services. This does not change their task definition.
Each task created for each registered service will end up running the sidecar proxy container in order, so that the task is added to the mesh.

Injecting the proxy in the services and not in the task definitions allows for the same task definition to be reused to run identical applications in different namespaces with different Service Connect configurations.
It also means that, since the proxy is not in the task definition, it cannot be configured by users.

Service Connect only interconnects services within the same namespace.

One can add one Service Connect configuration to new or existing services.
When that happens, ECS creates:

A Service Connect endpoint in the namespace.
A new deployment in the service that replaces the tasks that are currently running with ones equipped with the proxy.

Existing tasks and other applications can continue to connect to existing endpoints and external applications.
If a service using Service Connect adds tasks by scaling out, new connections from clients will be load balanced between all of the running tasks. If the service is updated, new connections from clients will be load balanced only between the new version of the tasks.

The list of endpoints in the namespace changes every time any service in that namespace is deployed.
Existing tasks, and replacement tasks, continue to behave the same as they did after the most recent deployment.
Existing tasks cannot resolve and connect to new endpoints. Only tasks with a Service Connect configuration in the same namespace and that start running after this deployment can.

Applications can use short names and standard ports to connect to services in the same or other clusters.
This includes connecting across VPCs in the same AWS Region.

By default, the Service Connect proxy listens on the containerPort specified in the task definition's port mapping.
The service's Security Group rules must allow incoming traffic to this port from the subnets where clients will run.

The proxy will consume some of the resources allocated to their task.
It is recommended:

Adding at least 256 CPU units and 64 MiB of memory to the task's resources.
[If expecting tasks to receive more than 500 requests per second at their peak load] Increasing the sidecar's resources addition to at least 512 CPU units.
[If expecting to create more than 100 Service Connect services in the namespace, or 2000 tasks in total across all ECS services within the namespace], Adding 128 MiB extra of memory for the Service Connect proxy container.
One must do this in every task definition that is used by any of the ECS services in the namespace.

It is recommended one sets the log configuration in the Service Connect configuration.

Proxy configuration:

Tasks in a Service Connect endpoint are load balanced in a round-robin strategy.
The proxy uses data about prior failed connections to avoid sending new connections to the tasks that had the failed connections for some time.
At the time of writing, failing 5 or more connections in the last 30 seconds makes the proxy avoid that task for 30 to 300 seconds.
Connection that pass through the proxy and fail are retried, but avoid the host that failed the previous connection.
This ensures that each connection through Service Connect doesn't fail for one-off reasons.
Wait a maximum time for applications to respond.
The default timeout value is 15 seconds, but it can be updated.

Limitations

Service Connect does not support:

ECS' host network mode.
Windows containers.
HTTP 1.0.
Standalone tasks and any task created by other resources than services.
Services using the blue/green or external deployment types.
External container instance for ECS Anywhere.
PPv2.
Task definitions that set container memory limits.
It is required to set the task memory limit, though.

Tasks using the bridge network mode and Service Connect will not support the hostname container definition parameter.

Each service can belong to only one namespace.

Service Connect can use any AWS Cloud Map namespace, as long as they are in the same Region and AWS account.

Service Connect does not delete namespaces when clusters are deleted.
One must delete namespaces in AWS Cloud Map themselves.

Requirements

Tasks running in Fargate must use the Fargate Linux platform version 1.4.0 or higher.
The ECS agent on container instances must be version 1.67.2 or higher.
Container instances must run the ECS-optimized Amazon Linux 2023 AMI version 20230428 or later, or the ECS-optimized Amazon Linux 2 AMI version 2.0.20221115 or later.
These versions equip the Service Connect agent in addition to the ECS container agent.
Container instances must have the ecs:Poll permission assigned to them for resource arn:aws:ecs:{{region}}:{{accountId}}:task-set/cluster/*.
If using the ecsInstanceRole or AmazonEC2ContainerServiceforEC2Role IAM roles, there is no need for additional permissions.
Services must use the rolling deployment strategy, as it is the only one supported.
Task definitions must set their task's memory limit.
The task memory limit must be set to a number greater than the sum of the container memory limits.
The CPU and memory in the task limits that aren't allocated in the container limits will be used by the Service Connect's proxy container and other containers that don't set container limits.
All endpoints must be unique within their namespace.
All discovery names must be unique within their namespace.
One must redeploy existing services before applications can resolve the new endpoints.
New endpoints that are added to the namespace after the service's most recent deployment will not be added to the proxy configuration.
Application Load Balancer traffic defaults to routing through the Service Connect agent in awsvpc network mode.
If one wants non-service traffic to bypass the Service Connect agent, one will need to use the ingressPortOverride parameter in their Service Connect service configuration.

Procedure:

Configure the ECS cluster to use the desired AWS Cloud Map namespace.

Simplified process

Create the cluster with the desired name for the AWS Cloud Map namespace, and specify that name for the namespace when asked.
ECS will create a new HTTP namespace with the necessary configuration.
As reminder, Service Connect doesn't use or create DNS hosted zones in Amazon Route 53. FIXME: check this
Configure port names in the server services' task definitions for all the port mappings that the services will expose in Service Connect.
```
containerDefinitions: [{
    "name": "postgres",
    "protocol": "tcp",
    "containerPort": 5432
}]
```

Configure the server services to create Service Connect endpoints within the namespace.

"serviceConnectConfiguration": {
    "enabled": true,
    "namespace": "ecs-dev-cluster",
    "services": [{
        "portName": "postgres",
        "discoveryName": "postgres",
        "clientAliases": [{
            "port": 5432,
            "dnsName": "pgsql"
        }]
    }]
}

Deploy the services.
This will create the endpoints AWS Cloud Map namespace used by the cluster.
ECS also injects the Service Connect proxy container in each task.
Deploy the client applications as ECS services.
ECS connects them to the Service Connect endpoints through the Service Connect proxy in each task.
Applications only use the proxy to connect to Service Connect endpoints.
No additional configuration is required to use the proxy.
[optionally] Monitor traffic through the Service Connect proxy in Amazon CloudWatch.

ECS service discovery

Service discovery helps manage HTTP and DNS namespaces for ECS services.

ECS automatically registers and de-registers the list of launched tasks to AWS Cloud Map.
Cloud Map maintains DNS records that resolve to the internal IP addresses of one or more tasks from registered services.
Other services in the same VPC can use such DNS records to send traffic directly to containers using their internal IP addresses.

This approach provides low latency since traffic travels directly between the containers.

ECS service discovery is a good fit when using the awsvpc network mode, where:

Each task is assigned its own, unique IP address.
That IP address is an A record.
Each service can have a unique security group assigned.

When using bridged network mode, A records are no longer enough for service discovery and one must also use a SRV DNS record. This is due to containers sharing the same IP address and having ports mapped randomly.
SRV records can keep track of both IP addresses and port numbers, but requires applications to be appropriately configured.

Service discovery supports only the A and SRV DNS record types.
DNS records are automatically added or removed as tasks start or stop for ECS services.

Task registration in CloudMap might take some seconds to finish.
Until ECS registers the tasks, Containers in them might complain about being unable to resolve the services they are using.

DNS records have a TTL and it might happen that tasks died before this ended.
One must implement extra logic in one's applications, so that they can handle retries and deal with connection failures when the records are not yet updated.

Procedure:

Create the desired AWS Cloud Map namespace.
Create the desired Cloud Map service in the namespace.

Configure the ECS service offering acting as server to use the Cloud Map service.

"serviceRegistries": [{
    "registryArn": "arn:aws:servicediscovery:eu-west-1:012345678901:service/srv-uuf33b226vw93biy"
}]

NS lookup commands from within containers might fail, but they might still be able to resolve services registered in CloudMap namespaces.

$ aws ecs execute-command --cluster 'dev' \
    --task 'arn:aws:ecs:eu-west-1:012345678901:task/dev/abcdef0123456789abcdef0123456789' --container 'prometheus' \
    --interactive --command 'nslookup mimir.dev.ecs.internal'

The Session Manager plugin was installed successfully. Use the AWS CLI to start a session.

Starting session with SessionId: ecs-execute-command-p3pkkrysjdptxa8iu3cz3kxnke
Server:   172.16.0.2
Address:  172.16.0.2:53

Non-authoritative answer:

$ aws ecs execute-command --cluster 'dev' \
    --task 'arn:aws:ecs:eu-west-1:012345678901:task/dev/abcdef0123456789abcdef0123456789' --container 'prometheus' \
    --interactive --command 'wget -SO- mimir.dev.ecs.local:8080/ready'

The Session Manager plugin was installed successfully. Use the AWS CLI to start a session.

Starting session with SessionId: ecs-execute-command-hjgyio7n6nf2o9h4qn6ht7lzri
Connecting to mimir.dev.ecs.local:8080 (172.16.88.99:8080)
  HTTP/1.1 200 OK
  Date: Thu, 08 May 2025 09:35:02 GMT
  Content-Type: text/plain
  Content-Length: 5
  Connection: close

saving to '/dev/stdout'
stdout               100% |********************************|     5  0:00:00 ETA
'/dev/stdout' saved

Exiting session with sessionId: ecs-execute-command-hjgyio7n6nf2o9h4qn6ht7lzri.

VPC Lattice

Managed application networking service that customers can use to observe, secure, and monitor applications built across AWS compute services, VPCs, and accounts without having to modify their code.

VPC Lattice technically replaces the need for Application Load Balancers by leveraging target groups themselves.
Target groups which are a collection of compute resources, and can refer EC2 instances, IP addresses, Lambda functions, and Application Load Balancers.
Listeners are used to forward traffic to specified target groups when the conditions are met.
ECS also automatically replaces unhealthy tasks.

ECS tasks can be enabled as IP targets in VPC Lattice by associating their services with a VPC Lattice target group.
ECS automatically registers tasks to the VPC Lattice target group when they are launched for registered services.

Deployments might take longer when using VPC Lattice due to the extent of changes required.

Scrape metrics using Prometheus

Refer Prometheus service discovery for AWS ECS and Scraping Prometheus metrics from applications running in AWS ECS.

Important

Prometheus is not currently capable to automatically discover ECS components like services or tasks.

Solutions:

Push the metrics instead by instrumenting the applications, or using tools like AWS Distro for OpenTelemetry.
Use a load balancer to access the service, and point Prometheus to the load balancer.

Useful if needing to just monitor the availability of services.
The load balancer will take care of monitoring the tasks in the target group.

The scrape request needs to go through the load balancer.
This will cost money.
Target a lambda that returns a 308 Permanent Redirect code with the current IP addresses of the requested tasks.
Use dynamic service discovery mechanisms like AWS Cloud Map.
Refer Metrics collection from Amazon ECS using Amazon Managed Service for Prometheus and aws-cloudmap-prometheus-sd.

Send logs to a central location

FireLens

Refer Example Amazon ECS task definition: Route logs to FireLens, Under the hood: FireLens for Amazon ECS Tasks and Amazon ECS FireLens Examples.

Allows containers in ECS tasks to send logs to multiple destinations. Those can be AWS services (E.G. CloudWatch Logs and OpenSearch), AWS partners (E.G. Splunk and Datadog), or any service supporting Fluent* output.

It uses Fluent Bit or Fluentd under the hood.
One can tweak their behaviour using according custom Fluent Bit or Fluentd configuration files from S3 or the container image.

Requires a FireLens sidecar container to run alongside the main application's containers in order to process and forward logs from them.
This log router sidecar container should be marked as essential in order to prevent silent log loss should it crash.

The log router's container image can be amazon/aws-for-fluent-bit if one wants to send data to an AWS service or Partner.
It must be a custom image equipped with the required output plugins if not.

Example: send logs to OpenSearch

{
    "family": "nginx-to-opensearch",
    "networkMode": "awsvpc",
    "requiresCompatibilities": [ "FARGATE" ],
    "cpu": "256",
    "memory": "512",
    "executionRoleArn": "arn:aws:iam::012345678901:role/ecsTaskExecutionRole",
    "containerDefinitions": [
        {
            "name": "nginx",
            "essential": true,
            "image": "012345678901.dkr.ecr.eu-west-1.amazonaws.com/docker-hub-cache/nginx:latest",
            "portMappings": [{
                "protocol": "tcp",
                "containerPort": 80
            }],
            "logConfiguration": {
                "logDriver": "awsfirelens",
                "options": {
                    "Name": "ElasticSearch",
                    "Host": "sweet-os-domain-of-mine.eu-west-1.es.amazonaws.com",
                    "Port": "443",
                    "AWS_Auth": "On",
                    "AWS_Region": "eu-west-1",
                    "Index": "nginx-logs",
                    "Type": "_doc",
                    "tls": "On"
                }
            }
        },
        {
            "name": "log_router",
            "essential": true,
            "image": "amazon/aws-for-fluent-bit:latest",
            "memoryReservation": 128,
            "firelensConfiguration": {
                "type": "fluentbit",
                "options": {
                    "enable-ecs-log-metadata": "true"
                }
            }
        }
    ]
}

Example: send logs to Grafana Loki

{
    "family": "nginx-to-loki",
    "networkMode": "awsvpc",
    "requiresCompatibilities": [ "FARGATE" ],
    "cpu": "256",
    "memory": "512",
    "executionRoleArn": "arn:aws:iam::012345678901:role/ecsTaskExecutionRole",
    "containerDefinitions": [
        {
            "name": "nginx",
            "essential": true,
            "image": "012345678901.dkr.ecr.eu-west-1.amazonaws.com/docker-hub-cache/nginx:latest",
            "portMappings": [{
                "protocol": "tcp",
                "containerPort": 80
            }],
            "logConfiguration": {
                "logDriver": "awsfirelens",
                "options": {
                    "Name": "loki",
                    "Host": "loki.example.org",
                    "Port": "3100",
                    "LogLevel": "info",
                    "Labels": "{job=\"nginx\", container=\"nginx\"}",
                    "tls": "off",
                    "remove_keys": "ecs_task_arn,ecs_cluster"
                }
            }
        },
        {
            "name": "log_router",
            "essential": true,
            "image": "012345678901.dkr.ecr.eu-west-1.amazonaws.com/custom/fluent-bit-with-loki-output-plugin:latest",
            "memoryReservation": 128,
            "firelensConfiguration": {
                "type": "fluentbit",
                "options": {
                    "enable-ecs-log-metadata": "true",
                    "config-file-type": "s3",
                    "config-file-value": "s3://custom-configs-bucket/fluent-bit/nginx-log-router.conf"
                }
            }
        }
    ]
}

Fluent Bit or Fluentd

Refer Centralized Container Logging with Fluent Bit.

Use the fluentd log driver in task definitions.
The fluentd-address value is specified as a secret option as it may be treated as sensitive data.

"containerDefinitions": [{
    "logConfiguration": {
        "logDriver": "fluentd",
        "options": {
            "tag": "fluentd demo"
        },
        "secretOptions": [{
            "name": "fluentd-address",
            "valueFrom": "arn:aws:secretsmanager:region:aws_account_id:secret:fluentd-address-KnrBkD"
        }]
    },
    "entryPoint": [],
    "portMappings": [
        {
            "hostPort": 80,
            "protocol": "tcp",
            "containerPort": 80
        },
        {
            "hostPort": 24224,
            "protocol": "tcp",
            "containerPort": 24224
        }
    ]
}]

Secrets

Options:

Pass Secrets Manager secrets through Amazon ECS environment variables.

Use Secrets Manager in environment variables

When setting environment variables to secrets from Secrets Manager, it is the execution role (and not the task role) that must have the permissions required to access them.

Best practices

Consider configuring resource constraints.
Consider making sure the SIGTERM signal is caught from within the container, and that it triggers any cleanup action that might be needed.
When using spot compute capacity, consider ensuring containers exit gracefully before the task stops.
Refer Capacity providers.

Cost-saving measures:

Prefer using ARM-based compute capacity over the default X86_64, where feasible.
Specify the CPU architecture in the task's definition.

 {
     "family": "bb-arm64",
     "networkMode": "awsvpc",
     …,
+    "runtimePlatform": {
+        "cpuArchitecture": "ARM64"
+    }
 }

When configuring resource constraints:
- Consider granting tasks a reasonable amount of resources to work with.
- Keep an eye on the task's effective resource usage and adjust the constraints accordingly.
When deploying stateless or otherwise interruption tolerant tasks, consider only using spot compute capacity (e.g., FARGATE_SPOT).
Refer Capacity providers.
If deploying stateful or otherwise interruption sensitive tasks, consider using on-demand compute capacity (e.g., FARGATE) only for the minimum amount of required tasks.
Refer Capacity providers.
Ensure only a set number of tasks execute on on-demand capacity by specifying the base value and a zero weight value for the on-demand capacity provider.
```
{
    "capacityProvider": "FARGATE",
    "base": 2,
    "weight": 0
}
```
Ensure a percentage or ratio of all the desired tasks execute on on-demand capacity by specifying a low weight value for the on-demand capacity provider, and a higher weight value for a second, spot capacity provider.
Percentage-like
```
{
    "capacityProvider": "FARGATE",
    "weight": 5
}
{
    "capacityProvider": "FARGATE_SPOT",
    "weight": 95
}
```
Ratio-like
```
{
    "capacityProvider": "FARGATE",
    "weight": 1
}
{
    "capacityProvider": "FARGATE_SPOT",
    "weight": 19
}
```
Consider configuring Service auto scaling for the application to reduce the number of tasks to a minimum during schedules (e.g., at night) or when otherwise unused.

Caution

Mind the limitations that come with the auto scaling settings.
If only used internally (e.g., via a VPN), consider not using a load balancer, but configuring intra-network communication capabilities for the application in its place.
Refer Allow tasks to communicate with each other.

Troubleshooting

Invalid 'cpu' setting for task

Refer Troubleshoot Amazon ECS task definition invalid CPU or memory errors and Resource constraints.

Cause

One specified an invalid cpu or memory value for the task when registering a task definition using ECS's API or the AWS CLI.

Should the task definition specify FARGATE as value for the requiresCompatibilities attribute, the resource values must be one of the specific pairs supported by Fargate.

Solution

Specify a supported value for the task CPU and memory in your task definition.

Cost-saving measures

Prefer using spot capacity for non-critical services and tasks.
Consider stopping (scaling to 0) non-production services after working hours.
Consider applying for EC2 Instance and/or Compute Savings Plans if using EC2 capacity.
Consider applying for Compute Savings Plans if using Fargate capacity.

82 KiB Raw Blame History

Elastic Container Service

TL;DR

How it works

Execution and task roles

Standalone tasks

Services

Launch type

EC2 launch type

Fargate launch type

External launch type

Capacity providers

Capacity provider strategies

EC2 capacity providers

Fargate for ECS

Resource constraints

Environment variables

Storage

EBS volumes

EFS volumes

Docker volumes

Bind mounts

Networking

Execute commands in tasks' containers

Scale the number of tasks automatically

Target tracking

Allow tasks to communicate with each other

ECS Service Connect

ECS service discovery

VPC Lattice

Scrape metrics using Prometheus

Send logs to a central location

FireLens

Fluent Bit or Fluentd

Secrets

Best practices

Troubleshooting

Invalid 'cpu' setting for task

Cost-saving measures

Further readings

Sources

82 KiB

Raw Blame History