Kitchen::Ec2: A Test Kitchen Driver for Amazon EC2

A Test Kitchen Driver for Amazon EC2.

This driver uses the aws sdk gem to provision and destroy EC2 instances. Use Amazon's cloud for your infrastructure testing!

Quick Start

1. Install ChefDK. If testing things other than Chef cookbooks, please consult your driver's documentation for information on what to install.

2. Install the AWS command line tools.

3. Run aws configure. This will set up your AWS credentials for both the AWS CLI tools and kitchen-ec2.

4. Add or exit the driver section of your .kitchen.yml:

   driver:
     name: ec2

5. Run kitchen test.

Requirements

There are no external system requirements for this driver. However you will need access to an AWS account. IAM users should have, at a minimum, permission to manage the lifecycle of an EC2 instance along with modifying components specified in kitchen driver configs. Consider using a permissive managed IAM policy like arn:aws:iam::aws:policy/AmazonEC2FullAccess or tailor one specific to your security requirements.

Configuration

By automatically applying reasonable defaults wherever possible, kitchen-ec2 does a lot of work to make your life easier. Here is a description of some of the configuration parameters and what we do to default them.

Specifying the Image

There are three ways to specify the image you use for the instance: the platform name, image_id, and image_search.

platform Name

The third way to specify the image is by leaving image_id and image_search blank, and specifying a standard platform name.

platforms:
  - name: ubuntu-14.04

If you use the platform name ubuntu, windows, rhel, debian, centos, freebsd or fedora, kitchen-ec2 will search for the latest matching official image of the given OS in your region. You may leave versions off, specify partial versions, and you may specify architecture to distinguish 32- and 64-bit. Some examples:

platforms:
  # The latest stable minor+patch release of rhel 6
  - name: rhel-6
  # The latest patch release of CentOS 6.3
  - name: centos-6.3
  # The latest patch release of Amazon Linux 2017.03
  - name: amazon-2017.03
  # 32-bit version of latest major+minor+patch release of Ubuntu
  - name: ubuntu-i386
  # 32-bit version of Debian 6
  - name: debian-6-i386
  # Latest 32-bit stable minor release of freebsd 10
  - name: freebsd-10-i386
  # The latest stable major+minor+patch release of Fedora
  - name: fedora
  # The most recent service-pack for Windows 2012 (not R2)
  - name: windows-2012
  # The most recent service-pack for Windows 2012R2
  - name: windows-2012r2
  # Windows 2008 RTM (not R2, no service pack)
  - name: windows-2008rtm
  # Windows 2008R2 SP1
  - name: windows-2008r2sp1

We always pick the highest released stable version that matches your regex, and follow the other image_search rules for preference.

image_id

image_id can be set explicitly. It must be an ami in the region you are working with!

platforms:
  - name: centos-7
    driver:
      image_id: ami-96a818fe

image_id's have a format like ami-748e2903. The image_id values appear next to the image names when you select 'Launch Instance' from the AWS EC2 console. You can also see the list from the AWS CLI aws ec2 describe-images.

image_search

image_search lets you specify a series of key/value pairs to search for the image. If a value is set to an array, then any of those values will match. You can learn more about the available filters in the AWS CLI doc under --filters here.

platforms:
  - name: ubuntu-14.04
    driver:
      image_search:
        owner-id: "099720109477"
        name: ubuntu/images/*/ubuntu-*-14.04*

In the event that there are multiple matches (as sometimes happens), we sort to get the best results. In order of priority from greatest to least, we prefer:

• HVM images over paravirtual
• SSD support over magnetic drives
• 64-bit over 32-bit
• The most recently created image (to pick up patch releases)

Note that the image_search method requires that the AMI image names be in a specific format. Some examples are:

• Windows-2012
• Windows-2012r2
• Windows-2012r2sp1
• RHEL-7.2

It is safest to use the same naming convention as used by the public images published by the OS vendors on the AWS marketplace.

AWS Authentication

In order to connect to AWS, you must specify AWS credentials. We rely on the SDK to find credentials in the standard way, documented here: https://github.com/aws/aws-sdk-ruby/#configuration

The SDK Chain will search environment variables, then config files, then IAM role data from the instance profile, in that order. In the case config files being present, the 'default' profile will be used unless shared_credentials_profile is defined to point to another profile.

Because the Test Kitchen test should be checked into source control and ran through CI we no longer support storing the AWS credentials in the .kitchen.yml file.

Instance Login Configuration

The instances you create use credentials you specify which are separate from the AWS credentials. Generally, SSH and WinRM use an AWS key pair which you specify.

SSH

The aws_ssh_key_id value is the name of the AWS key pair you want to use.

The value can be one of:

• nil (default)

By default the key ID is read from the AWS_SSH_KEY_ID environment variable. If the environment variable is not set, the value will be nil. In this case, a temporary key will be created for you (>= 2.1.0).

• name_of_some_existing_aws_key_pair

By setting the environment variable AWS_SSH_KEY_ID or by setting the aws_ssh_key_id driver option in kitchen.yml. This will be the key that becomes associated with the test EC2 instances. The key must already exist in AWS. To see a list of existing key pair IDs in a region, use the aws CLI tool. For example, in the US-East-1 region: aws ec2 describe-key-pairs --region us-east-1.

• _disable

This will not directly associate the EC2 instance with an AWS-managed key pair (pre-existing or auto-generated). This may be useful in environments that have disabled AWS-managed keys. Getting SSH keys onto the instance then becomes an exercise for the reader, though it can be done, for example, with scripting in user_data or if the credentials are already baked into the AMI.

When using an existing key, either an AWS-managed key pair or keys that exist on the instance through some other means, ensure that the private key is configured in your Test Kitchen transport section, either directly or made available via ssh-agent:

transport:
  ssh_key: ~/.ssh/mykey.pem

For standard platforms we automatically provide the SSH username, but when specifying your own AMI you may need to configure that as well.

WinRM

For Windows instances the generated Administrator password is fetched automatically from Amazon EC2 with the same private key as we use for SSH.

Unfortunately the RDP file format does not allow including login credentials, so kitchen login with WinRM cannot automatically log in for you.

Other Configuration

availability_zone

The AWS availability zone to use. Only request the letter designation - will attach this to the region used.

If not specified, your instances will be placed in an AZ of AWS's choice in your region.

instance_type

The EC2 instance type (also known as size) to use.

The default is t2.micro or t1.micro, depending on whether the image is hvm or paravirtual. (paravirtual images are incompatible with t2.micro.)

security_group_ids

An Array of EC2 security groups which will be applied to the instance. If no security group is specified, a temporary group will be created automatically which allows SSH and WinRM (>= 2.1.0).

security_group_filter

The EC2 security group which will be applied to the instance, specified by tag. Only one group can be specified this way.

The default is unset, or nil.

An example of usage:

security_group_filter:
  tag:   'Name'
  value: 'example-group-name'

region

Required The AWS region to use.

If the environment variable AWS_REGION is populated that will be used. Otherwise the default is "us-east-1".

subnet_id

The EC2 subnet to use.

The default is unset, or nil.

subnet_filter

The EC2 subnet to use, specified by tag.

The default is unset, or nil.

An example of usage:

subnet_filter:
  tag:   'Name'
  value: 'example-subnet-name'

tags

The Hash of EC tag name/value pairs which will be applied to the instance.

The default is { "created-by" => "test-kitchen" }.

user_data

The user_data script or the path to a script to feed the instance. Use bash to install dependencies or download artifacts before chef runs. This is just for some cases. If you can do the stuff with chef, then do it with chef!

On linux instances the default is unset, or nil.

On Windows instances we specify a default that enables winrm and adds a non-administrator user specified in the username transport options to the Administrator's User Group.

iam_profile_name

The EC2 IAM profile name to use. The default is nil.

Note: The user, whose AWS credentials you have defined, not only needs AmazonEC2FullAccess permissions, but also the ability to execute iam:PassRole. Hence, use a policy like below when using this option:

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Action": [
                "ec2:*"
            ],
            "Resource": "*"
        },
        {
            "Effect": "Allow",
            "Action": "iam:PassRole",
            "Resource": "arn:aws:iam::123456789:role/RoleName"
        }
    ]
}

See AWS documentation for more details.

spot_price

The price you bid in order to submit a spot request. An additional step will be required during the spot request process submission. If no price is set, it will use an on-demand instance.

The default is nil.

instance_initiated_shutdown_behavior

Control whether an instance should stop or terminate when shutdown is initiated from the instance using an operating system command for system shutdown.

The default is nil.

block_duration_minutes

The specified duration for a spot instance, in minutes. This value must be a multiple of 60 (60, 120, 180, 240, 300, or 360). If no duration is set, the spot instance will remain active until it is terminated.

The default is nil.

http_proxy

Specify a proxy to send AWS requests through. Should be of the format http://<host>:<port>.

The default is ENV["HTTPS_PROXY"] || ENV["HTTP_PROXY"]. If you have these environment variables set and do not want to use a proxy when contacting aws set http_proxy: nil.

Note - The AWS command line utility allow you to specify two proxies, one for HTTP and one for HTTPS. The AWS Ruby SDK only allows you to specify 1 proxy and because all requests are https:// this proxy needs to support HTTPS.

ssl_verify_peer

If you need to turn off ssl certificate verification for HTTP calls made to AWS, set ssl_verify_peer: false.

Disk Configuration

block_device_mappings

A list of block device mappings for the machine. An example of all available keys looks like:

block_device_mappings:
  - device_name: /dev/sda
    ebs:
      volume_size: 20
      delete_on_termination: true
  - device_name: /dev/sdb
    ebs:
      volume_type: gp2
      virtual_name: test
      volume_size: 15
      delete_on_termination: true
      snapshot_id: snap-0015d0bc
  - device_name: /dev/sdc
    ebs:
      volume_size: 100
      delete_on_termination: true
      volume_type: io1
      iops: 100

See Amazon EBS Volume Types to find out more about volume types.

If you have a block device mapping with a device_name equal to the root storage device name on your image then the provided mapping will replace the settings in the image.

If this is not provided it will use the default block_device_mappings from the AMI.

ebs_optimized

Option to launch EC2 instance with optimized EBS volume. See Amazon EC2 Instance Types to find out more about instance types that can be launched as EBS-optimized instances.

The default is false.

Network and Communication Configuration

associate_public_ip

AWS does not automatically allocate public IP addresses for instances created within non-default subnets. Set this option to true to force allocation of a public IP and associate it with the launched instance.

If you set this option to false when launching into a non-default subnet, Test Kitchen will be unable to communicate with the instance unless you have a VPN connection to your Virtual Private Cloud.

The default is true if you have configured a subnet_id, or false otherwise.

private_ip_address

The primary private IP address of your instance.

If you don't set this it will default to whatever DHCP address EC2 hands out.

interface

The place from which to derive the hostname for communicating with the instance. May be dns, public, private or private_dns. If this is unset, the driver will derive the hostname by failing back in the following order:

1. DNS Name
2. Public IP Address
3. Private IP Address
4. Private DNS Name

The default is unset. Under normal circumstances, the lookup will return the Private IP Address.

If the Private DNS Name is preferred over the private IP, it must be specified in the .kitchen.yml file

driver:
  interface: private_dns

Example

The following could be used in a .kitchen.yml or in a .kitchen.local.yml to override default configuration.

---
driver:
  name: ec2
  aws_ssh_key_id: id_rsa-aws
  security_group_ids: ["sg-1a2b3c4d"]
  region: us-west-2
  availability_zone: b
  require_chef_omnibus: true
  subnet_id: subnet-6e5d4c3b
  iam_profile_name: chef-client
  instance_type: m3.medium
  associate_public_ip: true
  interface: dns

transport:
  ssh_key: /path/to/id_rsa-aws
  connection_timeout: 10
  connection_retries: 5
  username: ubuntu

platforms:
  - name: ubuntu-16.04
  - name: centos-6.9
  - name: centos-7
    driver:
      image_id: ami-c7d092f7
      block_device_mappings:
        - device_name: /dev/sdb
          ebs:
            volume_type: gp2
            virtual_name: test
            volume_size: 8
            delete_on_termination: true
    transport:
      username: centos
  - name: windows-2012r2
  - name: windows-2016

suites:
# ...

Development

• Source hosted at GitHub
• Report issues/questions/feature requests on GitHub Issues

Pull requests are very welcome! Make sure your patches are well tested. Ideally create a topic branch for every separate change you make. For example:

1. Fork the repo
2. Create your feature branch (git checkout -b my-new-feature)
3. Commit your changes (git commit -am 'Added some feature')
4. Push to the branch (git push origin my-new-feature)
5. Create new Pull Request

License

Apache 2.0 (see LICENSE)