CentOS ECS Agent Ansible Role

This role helps configure the ECS container agent on standard CentOS 7 servers.

Requirements

• Ansible version 2.2 or newer
• Works on CentOS version 7.4.1708

Role Variables

For more details, check:

http://docs.aws.amazon.com/AmazonECS/latest/developerguide/ecs-agent-config.html

• ecs_agent_loglevel: ECS_LOGLEVEL (Default: info)
• ecs_agent_cluster_name: ECS_CLUSTER (Default: default)
• ecs_agent_enable_iam_role: ECS_ENABLE_TASK_IAM_ROLE (Default: true)
• ecs_agent_enable_task_iam_role_network_host: ECS_ENABLE_TASK_IAM_ROLE_NETWORK_HOST (Default: true)
• ecs_agent_reserved_ports: ECS_RESERVED_PORTS (Default: "[22, 2375, 2376, 51678]")
• ecs_agent_container_stop_timeout: ECS_CONTAINER_STOP_TIMEOUT (Default: 30s)
• ecs_agent_auth_type: ECS_ENGINE_AUTH_TYPE (Default: "")
• ecs_agent_auth_data: ECS_ENGINE_AUTH_DATA (Default: "")
• ecs_agent_data_dir: ECS_DATADIR (Default: "/data")
• ecs_agent_log_file: ECS_LOGFILE (Default: "/log/ecs_agent.log")

Dependencies

Docker must be installed and running.

See:

• https://github.com/bitintheskud/ansible-role-docker-ce-centos.git

Caveats

This role sets up the AWS ECS agent as suggested in the documentation and adds iptables rules. However, it does not save these iptables rules for you (using iptables-save or other methods). If you want to save iptables rules so that they remain after a reboot without needing to run Ansible again, you'll need to manage that separately.

Example Playbook

---
- name: AWS ECS Agent Playbook
  hosts: all
  become: yes
  vars:
    - ecs_agent_cluster_name: MyClusterName
  roles:
    - ansible-role-ecs-agent

Plans for Future Versions

• Add Travis CI build checks.
• Support for log drivers. See: https://github.com/open-guides/og-aws#ecs-tips
• Options for cleaning:

ECS_ENGINE_TASK_CLEANUP_WAIT_DURATION This variable sets how long to wait before removing containers from stopped tasks. The default is 3 hours, but it can be lowered to 1 minute.

ECS_DISABLE_IMAGE_CLEANUP Setting this to true disables automatic image cleanup on your container instance.

ECS_IMAGE_CLEANUP_INTERVAL This sets how often the image cleanup process checks for images to delete. The default is every 30 minutes, but it can be reduced to 10 minutes.

ECS_IMAGE_MINIMUM_CLEANUP_AGE This specifies the minimum time an image must have been pulled before it can be removed. The default is 1 hour.

ECS_NUM_IMAGES_DELETE_PER_CYCLE This specifies how many images can be removed in one cleanup cycle. The default is 5, with a minimum of 1.

License

Licensed under the MIT License. See the LICENSE file for more details.