Running CoreOS Container Linux on EC2

The current AMIs for all Container Linux channels and EC2 regions are listed below and updated frequently. Using CloudFormation is the easiest way to launch a cluster, but it is also possible to follow the manual steps at the end of the article. Questions can be directed to the CoreOS IRC channel or user mailing list.

Choosing a channel

Container Linux is designed to be updated automatically with different schedules per channel. You can disable this feature, although we don't recommend it. Read the release notes for specific features and bug fixes.

The Alpha channel closely tracks master and is released frequently. The newest versions of system libraries and utilities will be available for testing. The current version is Container Linux 1590.0.0.

View as json feed
EC2 Region AMI Type AMI ID CloudFormation
ap-northeast-1 PV ami-3b17bf5d Launch Stack
HVM ami-4a11b92c Launch Stack
ap-northeast-2 PV ami-1545e17b Launch Stack
HVM ami-ba44e0d4 Launch Stack
ap-south-1 PV ami-af571bc0 Launch Stack
HVM ami-de501cb1 Launch Stack
ap-southeast-1 PV ami-8bda94e8 Launch Stack
HVM ami-71df9112 Launch Stack
ap-southeast-2 PV ami-ac2fc6ce Launch Stack
HVM ami-e32cc581 Launch Stack
ca-central-1 PV ami-e61fa782 Launch Stack
HVM ami-cb1ea6af Launch Stack
cn-north-1 PV ami-4a63b027 Launch Stack
HVM ami-6563b008 Launch Stack
eu-central-1 PV ami-bb7ef9d4 Launch Stack
HVM ami-ad7ff8c2 Launch Stack
eu-west-1 PV ami-1d0fa264 Launch Stack
HVM ami-b20da0cb Launch Stack
eu-west-2 PV ami-b4fee2d0 Launch Stack
HVM ami-8ffce0eb Launch Stack
sa-east-1 PV ami-29e29945 Launch Stack
HVM ami-dce19ab0 Launch Stack
us-east-1 PV ami-8f289af5 Launch Stack
HVM ami-4a1daf30 Launch Stack
us-east-2 PV ami-32daf557 Launch Stack
HVM ami-76dbf413 Launch Stack
us-gov-west-1 PV ami-7cd65a1d Launch Stack
HVM ami-b4d75bd5 Launch Stack
us-west-1 PV ami-d64876b6 Launch Stack
HVM ami-a44977c4 Launch Stack
us-west-2 PV ami-90eb3fe8 Launch Stack
HVM ami-7beb3f03 Launch Stack

The Beta channel consists of promoted Alpha releases. The current version is Container Linux 1576.2.0.

View as json feed
EC2 Region AMI Type AMI ID CloudFormation
ap-northeast-1 PV ami-9216bef4 Launch Stack
HVM ami-8616bee0 Launch Stack
ap-northeast-2 PV ami-3545e15b Launch Stack
HVM ami-3845e156 Launch Stack
ap-south-1 PV ami-7c541813 Launch Stack
HVM ami-08571b67 Launch Stack
ap-southeast-1 PV ami-ebd49a88 Launch Stack
HVM ami-f7d99794 Launch Stack
ap-southeast-2 PV ami-212ec743 Launch Stack
HVM ami-922dc4f0 Launch Stack
ca-central-1 PV ami-2411a940 Launch Stack
HVM ami-971ca4f3 Launch Stack
cn-north-1 PV ami-d87cafb5 Launch Stack
HVM ami-4363b02e Launch Stack
eu-central-1 PV ami-6e7afd01 Launch Stack
HVM ami-107cfb7f Launch Stack
eu-west-1 PV ami-a60ba6df Launch Stack
HVM ami-da0fa2a3 Launch Stack
eu-west-2 PV ami-eefce08a Launch Stack
HVM ami-ecfce088 Launch Stack
sa-east-1 PV ami-c1e79cad Launch Stack
HVM ami-b4e299d8 Launch Stack
us-east-1 PV ami-bc09bbc6 Launch Stack
HVM ami-aff944d5 Launch Stack
us-east-2 PV ami-d7dcf3b2 Launch Stack
HVM ami-4bdbf42e Launch Stack
us-gov-west-1 PV ami-f6d55997 Launch Stack
HVM ami-d8d75bb9 Launch Stack
us-west-1 PV ami-b2447ad2 Launch Stack
HVM ami-904678f0 Launch Stack
us-west-2 PV ami-2e10c456 Launch Stack
HVM ami-dc12c6a4 Launch Stack

The Stable channel should be used by production clusters. Versions of Container Linux are battle-tested within the Beta and Alpha channels before being promoted. The current version is Container Linux 1520.8.0.

View as json feed
EC2 Region AMI Type AMI ID CloudFormation
ap-northeast-1 PV ami-7d69c81b Launch Stack
HVM ami-8f65c4e9 Launch Stack
ap-northeast-2 PV ami-6b02a705 Launch Stack
HVM ami-5901a437 Launch Stack
ap-south-1 PV ami-1fd89a70 Launch Stack
HVM ami-8ad89ae5 Launch Stack
ap-southeast-1 PV ami-c4f2b3a7 Launch Stack
HVM ami-64f1b007 Launch Stack
ap-southeast-2 PV ami-27896645 Launch Stack
HVM ami-6e89660c Launch Stack
ca-central-1 PV ami-fd853d99 Launch Stack
HVM ami-91853df5 Launch Stack
cn-north-1 PV ami-d627f4bb Launch Stack
HVM ami-d727f4ba Launch Stack
eu-central-1 PV ami-7350eb1c Launch Stack
HVM ami-ea53e885 Launch Stack
eu-west-1 PV ami-a5ae0bdc Launch Stack
HVM ami-bbaf0ac2 Launch Stack
eu-west-2 PV ami-fa908d9e Launch Stack
HVM ami-c3978aa7 Launch Stack
sa-east-1 PV ami-051b6369 Launch Stack
HVM ami-181c6474 Launch Stack
us-east-1 PV ami-eb9b3c91 Launch Stack
HVM ami-a89d3ad2 Launch Stack
us-east-2 PV ami-2280ac47 Launch Stack
HVM ami-1c81ad79 Launch Stack
us-gov-west-1 PV ami-674dc006 Launch Stack
HVM ami-644dc005 Launch Stack
us-west-1 PV ami-cf566aaf Launch Stack
HVM ami-23566a43 Launch Stack
us-west-2 PV ami-af4d82d7 Launch Stack
HVM ami-7c488704 Launch Stack

CloudFormation will launch a cluster of Container Linux machines with a security and autoscaling group.

Container Linux Configs

Container Linux allows you to configure machine parameters, configure networking, launch systemd units on startup, and more via Container Linux Configs. These configs are then transpiled into Ignition configs and given to booting machines. Head over to the docs to learn about the supported features.

You can provide a raw Ignition config to Container Linux via the Amazon web console or via the EC2 API.

As an example, this Container Linux Config will configure and start etcd:

This is the human-readable config file. This should not be immediately passed to Container Linux. Learn more.
# This config is meant to be consumed by the config transpiler, which will
# generate the corresponding Ignition config. Do not pass this config directly
# to instances of Container Linux.

etcd:
  # All options get passed as command line flags to etcd.
  # Any information inside curly braces comes from the machine at boot time.

  # multi_region and multi_cloud deployments need to use {PUBLIC_IPV4}
  advertise_client_urls:       "http://{PRIVATE_IPV4}:2379"
  initial_advertise_peer_urls: "http://{PRIVATE_IPV4}:2380"
  # listen on both the official ports and the legacy ports
  # legacy ports can be omitted if your application doesn't depend on them
  listen_client_urls:          "http://0.0.0.0:2379"
  listen_peer_urls:            "http://{PRIVATE_IPV4}:2380"
  # generate a new token for each unique cluster from https://discovery.etcd.io/new?size=3
  # specify the initial size of your cluster with ?size=X
  discovery:                   "https://discovery.etcd.io/<token>"
This is the raw machine configuration, which is not intended for editing. Learn more. Validate the config here.
{
  "ignition": {
    "version": "2.0.0",
    "config": {}
  },
  "storage": {},
  "systemd": {
    "units": [
      {
        "name": "etcd-member.service",
        "enable": true,
        "dropins": [
          {
            "name": "20-clct-etcd-member.conf",
            "contents": "[Unit]\nRequires=coreos-metadata.service\nAfter=coreos-metadata.service\n\n[Service]\nEnvironmentFile=/run/metadata/coreos\nExecStart=\nExecStart=/usr/lib/coreos/etcd-wrapper $ETCD_OPTS \\\n  --listen-peer-urls=\"http://${COREOS_EC2_IPV4_LOCAL}:2380\" \\\n  --listen-client-urls=\"http://0.0.0.0:2379\" \\\n  --initial-advertise-peer-urls=\"http://${COREOS_EC2_IPV4_LOCAL}:2380\" \\\n  --advertise-client-urls=\"http://${COREOS_EC2_IPV4_LOCAL}:2379\" \\\n  --discovery=\"https://discovery.etcd.io/\u003ctoken\u003e\""
          }
        ]
      }
    ]
  },
  "networkd": {},
  "passwd": {}
}

Instance storage

Ephemeral disks and additional EBS volumes attached to instances can be mounted with a .mount unit. Amazon's block storage devices are attached differently depending on the instance type. Here's the Container Linux Config to format and mount the first ephemeral disk, xvdb, on most instance types:

This is the human-readable config file. This should not be immediately passed to Container Linux. Learn more.
# This config is meant to be consumed by the config transpiler, which will
# generate the corresponding Ignition config. Do not pass this config directly
# to instances of Container Linux.

storage:
  filesystems:
    - mount:
        device: /dev/xvdb
        format: ext4
        create:
          force: false

systemd:
  units:
    - name: media-ephemeral.mount
      enable: true
      contents: |
        [Mount]
        What=/dev/xvdb
        Where=/media/ephemeral
        Type=ext4

        [Install]
        RequiredBy=local-fs.target
This is the raw machine configuration, which is not intended for editing. Learn more. Validate the config here.
{
  "ignition": {
    "version": "2.0.0",
    "config": {}
  },
  "storage": {
    "filesystems": [
      {
        "mount": {
          "device": "/dev/xvdb",
          "format": "ext4",
          "create": {}
        }
      }
    ]
  },
  "systemd": {
    "units": [
      {
        "name": "media-ephemeral.mount",
        "enable": true,
        "contents": "[Mount]\nWhat=/dev/xvdb\nWhere=/media/ephemeral\nType=ext4\n\n[Install]\nRequiredBy=local-fs.target"
      }
    ]
  },
  "networkd": {},
  "passwd": {}
}

For more information about mounting storage, Amazon's own documentation is the best source. You can also read about mounting storage on Container Linux.

Adding more machines

To add more instances to the cluster, just launch more with the same Container Linux Config, the appropriate security group and the AMI for that region. New instances will join the cluster regardless of region if the security groups are configured correctly.

SSH to your instances

Container Linux is set up to be a little more secure than other cloud images. By default, it uses the core user instead of root and doesn't use a password for authentication. You'll need to add an SSH key(s) via the AWS console or add keys/passwords via your Container Linux Config in order to log in.

To connect to an instance after it's created, run:

ssh core@<ip address>

Optionally, you may want to configure your ssh-agent to more easily run fleet commands.

Multiple clusters

If you would like to create multiple clusters you will need to change the "Stack Name". You can find the direct template file on S3.

Manual setup

TL;DR: launch three instances of ami-8f289af5 in us-east-1 with a security group that has open port 22, 2379, 2380, 4001, and 7001 and the same "User Data" of each host. SSH uses the core user and you have etcd and Docker to play with.

Creating the security group

You need open port 2379, 2380, 7001 and 4001 between servers in the etcd cluster. Step by step instructions below.

This step is only needed once

First we need to create a security group to allow Container Linux instances to communicate with one another.

  1. Go to the security group page in the EC2 console.
  2. Click "Create Security Group"
    • Name: coreos-testing
    • Description: Container Linux instances
    • VPC: No VPC
    • Click: "Yes, Create"
  3. In the details of the security group, click the Inbound tab
  4. First, create a security group rule for SSH
    • Create a new rule: SSH
    • Source: 0.0.0.0/0
    • Click: "Add Rule"
  5. Add two security group rules for etcd communication
    • Create a new rule: Custom TCP rule
    • Port range: 2379
    • Source: type "coreos-testing" until your security group auto-completes. Should be something like "sg-8d4feabc"
    • Click: "Add Rule"
    • Repeat this process for port range 2380, 4001 and 7001 as well
  6. Click "Apply Rule Changes"

Launching a test cluster

We will be launching three instances, with a few parameters in the User Data, and selecting our security group.

  1. Open the quick launch wizard to boot ami-8f289af5.
  2. On the second page of the wizard, launch 3 servers to test our clustering
    • Number of instances: 3
    • Click "Continue"
  3. Next, we need to specify a discovery URL, which contains a unique token that allows us to find other hosts in our cluster. If you're launching your first machine, generate one at https://discovery.etcd.io/new?size=3, configure the `?size=` to your initial cluster size and add it to the metadata. You should re-use this key for each machine in the cluster.
  4. Use ct to convert the following configuration into an Ignition config, and back in the EC2 dashboard, paste it into the "User Data" field.
    This is the human-readable config file. This should not be immediately passed to Container Linux. Learn more.
    # This config is meant to be consumed by the config transpiler, which will
    # generate the corresponding Ignition config. Do not pass this config directly
    # to instances of Container Linux.
    
              etcd:
                # All options get passed as command line flags to etcd.
                # Any information inside curly braces comes from the machine at boot time.
              
                # multi_region and multi_cloud deployments need to use {PUBLIC_IPV4}
                advertise_client_urls:       "http://{PRIVATE_IPV4}:2379"
                initial_advertise_peer_urls: "http://{PRIVATE_IPV4}:2380"
                # listen on both the official ports and the legacy ports
                # legacy ports can be omitted if your application doesn't depend on them
                listen_client_urls:          "http://0.0.0.0:2379"
                listen_peer_urls:            "http://{PRIVATE_IPV4}:2380"
                # generate a new token for each unique cluster from https://discovery.etcd.io/new?size=3
                # specify the initial size of your cluster with ?size=X
                discovery:                   "https://discovery.etcd.io/<token>"
    
    This is the raw machine configuration, which is not intended for editing. Learn more. Validate the config here.
    {
      "ignition": {
        "version": "2.0.0",
        "config": {}
      },
      "storage": {},
      "systemd": {
        "units": [
          {
            "name": "etcd-member.service",
            "enable": true,
            "dropins": [
              {
                "name": "20-clct-etcd-member.conf",
                "contents": "[Unit]\nRequires=coreos-metadata.service\nAfter=coreos-metadata.service\n\n[Service]\nEnvironmentFile=/run/metadata/coreos\nExecStart=\nExecStart=/usr/lib/coreos/etcd-wrapper $ETCD_OPTS \\\n  --listen-peer-urls=\"http://${COREOS_EC2_IPV4_LOCAL}:2380\" \\\n  --listen-client-urls=\"http://0.0.0.0:2379\" \\\n  --initial-advertise-peer-urls=\"http://${COREOS_EC2_IPV4_LOCAL}:2380\" \\\n  --advertise-client-urls=\"http://${COREOS_EC2_IPV4_LOCAL}:2379\" \\\n  --discovery=\"https://discovery.etcd.io/\u003ctoken\u003e\""
              }
            ]
          }
        ]
      },
      "networkd": {},
      "passwd": {}
    }
    
    • Paste configuration into "User Data"
    • "Continue"
  5. Storage Configuration
    • "Continue"
  6. Tags
    • "Continue"
  7. Create Key Pair
    • Choose a key of your choice, it will be added in addition to the one in the gist.
    • "Continue"
  8. Choose one or more of your existing Security Groups
    • "coreos-testing" as above.
    • "Continue"
  9. Launch!

We will be launching three instances, with a few parameters in the User Data, and selecting our security group.

  1. Open the quick launch wizard to boot ami-bc09bbc6.
  2. On the second page of the wizard, launch 3 servers to test our clustering
    • Number of instances: 3
    • Click "Continue"
  3. Next, we need to specify a discovery URL, which contains a unique token that allows us to find other hosts in our cluster. If you're launching your first machine, generate one at https://discovery.etcd.io/new?size=3, configure the `?size=` to your initial cluster size and add it to the metadata. You should re-use this key for each machine in the cluster.
  4. Use ct to convert the following configuration into an Ignition config, and back in the EC2 dashboard, paste it into the "User Data" field.
    This is the human-readable config file. This should not be immediately passed to Container Linux. Learn more.
    # This config is meant to be consumed by the config transpiler, which will
    # generate the corresponding Ignition config. Do not pass this config directly
    # to instances of Container Linux.
    
              etcd:
                # All options get passed as command line flags to etcd.
                # Any information inside curly braces comes from the machine at boot time.
              
                # multi_region and multi_cloud deployments need to use {PUBLIC_IPV4}
                advertise_client_urls:       "http://{PRIVATE_IPV4}:2379"
                initial_advertise_peer_urls: "http://{PRIVATE_IPV4}:2380"
                # listen on both the official ports and the legacy ports
                # legacy ports can be omitted if your application doesn't depend on them
                listen_client_urls:          "http://0.0.0.0:2379"
                listen_peer_urls:            "http://{PRIVATE_IPV4}:2380"
                # generate a new token for each unique cluster from https://discovery.etcd.io/new?size=3
                # specify the initial size of your cluster with ?size=X
                discovery:                   "https://discovery.etcd.io/<token>"
    
    This is the raw machine configuration, which is not intended for editing. Learn more. Validate the config here.
    {
      "ignition": {
        "version": "2.0.0",
        "config": {}
      },
      "storage": {},
      "systemd": {
        "units": [
          {
            "name": "etcd-member.service",
            "enable": true,
            "dropins": [
              {
                "name": "20-clct-etcd-member.conf",
                "contents": "[Unit]\nRequires=coreos-metadata.service\nAfter=coreos-metadata.service\n\n[Service]\nEnvironmentFile=/run/metadata/coreos\nExecStart=\nExecStart=/usr/lib/coreos/etcd-wrapper $ETCD_OPTS \\\n  --listen-peer-urls=\"http://${COREOS_EC2_IPV4_LOCAL}:2380\" \\\n  --listen-client-urls=\"http://0.0.0.0:2379\" \\\n  --initial-advertise-peer-urls=\"http://${COREOS_EC2_IPV4_LOCAL}:2380\" \\\n  --advertise-client-urls=\"http://${COREOS_EC2_IPV4_LOCAL}:2379\" \\\n  --discovery=\"https://discovery.etcd.io/\u003ctoken\u003e\""
              }
            ]
          }
        ]
      },
      "networkd": {},
      "passwd": {}
    }
    
    • Paste configuration into "User Data"
    • "Continue"
  5. Storage Configuration
    • "Continue"
  6. Tags
    • "Continue"
  7. Create Key Pair
    • Choose a key of your choice, it will be added in addition to the one in the gist.
    • "Continue"
  8. Choose one or more of your existing Security Groups
    • "coreos-testing" as above.
    • "Continue"
  9. Launch!

We will be launching three instances, with a few parameters in the User Data, and selecting our security group.

  1. Open the quick launch wizard to boot ami-eb9b3c91.
  2. On the second page of the wizard, launch 3 servers to test our clustering
    • Number of instances: 3
    • Click "Continue"
  3. Next, we need to specify a discovery URL, which contains a unique token that allows us to find other hosts in our cluster. If you're launching your first machine, generate one at https://discovery.etcd.io/new?size=3, configure the `?size=` to your initial cluster size and add it to the metadata. You should re-use this key for each machine in the cluster.
  4. Use ct to convert the following configuration into an Ignition config, and back in the EC2 dashboard, paste it into the "User Data" field.
    This is the human-readable config file. This should not be immediately passed to Container Linux. Learn more.
    # This config is meant to be consumed by the config transpiler, which will
    # generate the corresponding Ignition config. Do not pass this config directly
    # to instances of Container Linux.
    
              etcd:
                # All options get passed as command line flags to etcd.
                # Any information inside curly braces comes from the machine at boot time.
              
                # multi_region and multi_cloud deployments need to use {PUBLIC_IPV4}
                advertise_client_urls:       "http://{PRIVATE_IPV4}:2379"
                initial_advertise_peer_urls: "http://{PRIVATE_IPV4}:2380"
                # listen on both the official ports and the legacy ports
                # legacy ports can be omitted if your application doesn't depend on them
                listen_client_urls:          "http://0.0.0.0:2379"
                listen_peer_urls:            "http://{PRIVATE_IPV4}:2380"
                # generate a new token for each unique cluster from https://discovery.etcd.io/new?size=3
                # specify the initial size of your cluster with ?size=X
                discovery:                   "https://discovery.etcd.io/<token>"
    
    This is the raw machine configuration, which is not intended for editing. Learn more. Validate the config here.
    {
      "ignition": {
        "version": "2.0.0",
        "config": {}
      },
      "storage": {},
      "systemd": {
        "units": [
          {
            "name": "etcd-member.service",
            "enable": true,
            "dropins": [
              {
                "name": "20-clct-etcd-member.conf",
                "contents": "[Unit]\nRequires=coreos-metadata.service\nAfter=coreos-metadata.service\n\n[Service]\nEnvironmentFile=/run/metadata/coreos\nExecStart=\nExecStart=/usr/lib/coreos/etcd-wrapper $ETCD_OPTS \\\n  --listen-peer-urls=\"http://${COREOS_EC2_IPV4_LOCAL}:2380\" \\\n  --listen-client-urls=\"http://0.0.0.0:2379\" \\\n  --initial-advertise-peer-urls=\"http://${COREOS_EC2_IPV4_LOCAL}:2380\" \\\n  --advertise-client-urls=\"http://${COREOS_EC2_IPV4_LOCAL}:2379\" \\\n  --discovery=\"https://discovery.etcd.io/\u003ctoken\u003e\""
              }
            ]
          }
        ]
      },
      "networkd": {},
      "passwd": {}
    }
    
    • Paste configuration into "User Data"
    • "Continue"
  5. Storage Configuration
    • "Continue"
  6. Tags
    • "Continue"
  7. Create Key Pair
    • Choose a key of your choice, it will be added in addition to the one in the gist.
    • "Continue"
  8. Choose one or more of your existing Security Groups
    • "coreos-testing" as above.
    • "Continue"
  9. Launch!

Using CoreOS Container Linux

Now that you have a machine booted it is time to play around. Check out the Container Linux Quickstart guide or dig into more specific topics.