Knife::Server

A Chef Knife plugin to manage Chef Servers. Bootstrap a new Chef Server on Amazon's EC2, Digital Ocean, Linode, OpenStack or a standalone server. Backup and restore your Chef Server or Hosted Chef's node, role, data bag, and environment JSON data.

Follow the installation instructions, then you are ready to create your very own Chef Server running Ubuntu on Amazon's EC2 service:

$ knife server bootstrap ec2 \
  --node-name chefapalooza.example.com \
  --aws-access-key-id $AWS_ACCESS_KEY_ID \
  --aws-secret-access-key $AWS_SECRET_ACCESS_KEY \
  --region us-east-1 \
  --availability-zone us-east-1b \
  --image ami-d017b2b8 \
  --ssh-user ubuntu \
  --flavor m1.small \
  --ssh-key id_rsa-aws \
  --identity-file ~/.ssh/id_rsa-aws

Be sure to substitute with your access key id, secret access key, ssh key id, and identity file (private SSH key). If you have most of these settings set up in a knife.rb file, it becomes much shorter:

$ knife server bootstrap ec2 \
  --node-name chefapalooza.example.com \
  --ssh-user ubuntu

To spin up your Chef Server on Digital Ocean:

knife server bootstrap digitalocean \
  --node-name chefapalooza.example.com \
  --digital_ocean_access_token $DIGITAL_OCEAN_ACCESS_TOKEN \
  --location sfo1 \
  --size 1gb \
  --image ubuntu-12-04-x64 \
  --ssh-keys $DIGITAL_OCEAN_SSH_KEY_ID \
  --identity-file ~/.ssh/id_rsa-do

To spin up your Chef Server on Linode:

knife server bootstrap linode \
  --node-name linny.example.com \
  --linode-node-name linny \
  --linode-api-key $LINODE_API_KEY \
  --linode-datacenter 3 \
  --linode-flavor 1 \
  --linode-image 126 \
  --ssh-password 'testing1234'

To spin up your Chef Server on OpenStack:

knife server bootstrap openstack \
  --node-name openstack.example.com \
  --openstack-node-name openstack \
  --openstack-username $OS_USERNAME \
  --openstack-password $OS_PASSWORD \
  --openstack-auth-url $OS_AUTH_URL \
  --ssh-password 'testing1234'

Or maybe you want to try out a Chef Server using Vagrant?

$ cat <<VAGRANTFILE > Vagrantfile
Vagrant.configure("2") do |config|
  config.vm.box = "precise64"
  config.vm.box_url = "http://files.vagrantup.com/precise64.box"
  config.vm.network :private_network, ip: "192.168.33.10"
  config.vm.provider :virtualbox do |vb|
    vb.customize ["modifyvm", :id, "--memory", 2048]
  end
end
VAGRANTFILE
$ vagrant up
$ knife server bootstrap standalone \
  --node-name chefapalooza.example.com \
  --host 192.168.33.10 \
  --ssh-user vagrant \
  --ssh-password vagrant

Taking a backup of all node, role, data bag, and environment data is also a snap:

$ knife server backup

Restoring all that data isn't too hard either:

$ knife server restore

Let's say you have all your roles, environments, and data bags in a Git repository (remember, Infrastructure as Code) and you want to build a new Chef Server. Try this:

$ knife server backup nodes --backup-dir /tmp/old_server
$ knife server bootstrap ec2 --ssh-user ubuntu \
  --node-name chef.example.com
$ knife server restore nodes --backup-dir /tmp/old_server

Now fire up your cookbooks, roles, environments, and data bags. Good to go! Well except for old client keys on your nodes, but this is being looked into.

See below for more details.

Add this line to your application's Gemfile:

gem 'knife-server'

Note If you want to use the bootstrap ec2, bootstrap digitalocean, bootstrap linode, or bootstrap openstack subcommands you will need to explicitly add this to your Gemfile with:

gem 'knife-ec2'
gem 'knife-digital_ocean'
gem 'knife-linode'
gem 'knife-openstack'

Finally execute:

$ bundle

Or install it yourself as:

$ gem install knife-server

(Don't forget a gem install knife-ec2, gem install knife-digital_ocean, gem install knife-linode, or gem install knife-openstack if using any cloud-specific subcommands).

Next, you must set up a knife.rb configuration so that the shipped Knife subcommands know where to place and modify key files, backup directories, etc.

When creating a Chef Server the validation key and admin client key will be installed on your workstation. Therefore, several knife configuration options are required (descriptions courtesy of the Chef docs site):

• node_name: User or client identity (i.e., name) to use for authenticating requests to the Chef Server.
• client_key: Private key file to authenticate to the Chef server. Corresponds to the -k or --key option
• validation_key: Specifies the private key file to use when bootstrapping new hosts. See knife-client(1) for more information about the validation client.

It is important to note that the knife server bootstrap subcommands will need write access to the client_key and validation_key files as well as write access the their directories in order to create backup files. This means that if you do not explicitly set client_key and validation_key, the knife server bootstrap subcommands will attempt to write to "/etc/chef/client.pem" and "/etc/chef/validation.pem." respectively.

For example, you could add lines similar to the following in your user's ~/.chef/knife.rb or your project's .chef/knife.rb file (assuming you have a directory "$HOME/.chef" created with mkdir -p $HOME/.chef):

node_name       "gramsay"
client_key      "#{ENV['HOME']}/.chef/gramsay.pem"
validation_key  "#{ENV['HOME']}/.chef/chef-validator.pem"

Most options can be passed to the knife subcommands explicitly but this quickly becomes tiring, repetitive, and error-prone. A better solution is to add some of the common configuration to your ~/.chef/knife.rb or your projects .chef/knife.rb file like so:

# for aws/ec2
knife[:aws_access_key_id] = "MY_KEY"
knife[:aws_secret_access_key] = "MY_SECRET"
knife[:region] = "us-west-2"
knife[:availability_zone] = "us-west-2a"
knife[:flavor] = "t1.micro"

# for linode
knife[:linode_api_key] = "MY_KEY"

# for digitalocean
knife[:digital_ocean_acceess_token] = "MY_TOKEN"

Better yet, why not try a more generic knife.rb file from the chef-bootstrap-repo project?

These subcommands will install and configure an Open Source Chef Server on several different clouds/environments. The high level step taken are as follows:

1. Provision or use a node and install the Chef Server software. If it is a Chef 10 server, configure an Apache2 instance to front the server instance handling SSL for the API port (TCP/443) and the WebUI web application (TCP/444, if enabled).
2. Fetch the validation key from the server and install it onto the workstation issuing the knife subcommand. The validation key will be installed at the path defined in the knife validation_key variable. If a key already exists at that path a backup copy will be made in the same directory.
3. Create an initial admin user key called root in the root user's account on the server which can be used for local administration of the Chef Server. If it is a Chef 10 installation, an admin client key will be generated instead.
4. Create an admin user key with the name defined in the knife node_name configuration variable and install it onto the workstation issuing the knife subcommand. The user key will be installed at the path defined in the knife client_key configuration variable. If a key already exists at that path a backup copy will be made in the same directory. If it is a Chef 10 installation, an admin client key will be generated instead.

Note knife server bootstrap can not be invoked directly; a subcommand must be selected which determines the provisioning strategy.

The name of your new Chef Server. The hostname of the system will be set to this value and the self signed SSL certificate will use this value as its CN. Ideally this will correspond to the A or CNAME DNS record.

This option is required.

The platform type that will be bootstrapped. By convention a bootstrap template of chef-server-#{platform}.erb will be searched for in the template lookup locations (gems, .chef directory, etc.).

The default value is "omnibus" which supports all platforms for which Omnibus packages have been created. For more details, visit the Install Chef page and click on "Chef Server".

The SSH username used when bootstrapping the Chef Server node. Note that the some Amazon Machine Images (AMIs) such as the official Canonical Ubuntu images use non-root SSH users ("ubuntu" for Ubuntu AMIs).

The default value is "root".

The SSH port used when bootstrapping the Chef Server node.

The default value is "22".

Verify and cache SSH host key when connecting.

The default value is true

The SSH identity file (private SSH key) used for authentication.

Installs a pre-release Chef gem rather than a stable release version.

The version of Chef to install. For Chef 11 installations (the default), you can provide the version of Omnibus package you want installed. To install a Chef 10 server, set this value to "10".

The full path to location of template to use.

Bootstraps the Chef Server using a particular bootstrap template.

The default is "chef11/#{platform}".

Whether or no the WebUI interface will be installed and enabled.

The default value is false.

The initial password for the WebUI admin account, root user account, and workstation user account.

The default value is "chefchef".

The initial password for AMQP.

The default value is "chefchef".

Note: You must install the knife-ec2 gem to use this subcommand. This was done to keep the dependencies of this library lighter and to make future cloud adapter support easier to add.

Provisions an EC2 instance on the Amazon Web Services (AWS) cloud and sets up an Open Source Chef Server as described above. In addition, the following steps are taken initially:

1. Create and configure an EC2 Security Group called "infrastructure" for the Chef Server instance. TCP ports 22, 443, and 444 are permitted inbound for SSH, the API endpoint, and the WebUI web application respectively.
2. An EC2 instance will be provisioned using configuration and/or defaults present using the knife-ec2 plugin.

This subcommand imports all relavent options from the knife-ec2 gem. For detailed documentation relating to these options, please visit the docs page.

The availability zone for the EC2 instance.

The default value is "us-east-1b".

Your AWS access key ID.

This option is required.

Your AWS API secret access key.

This option is required.

Enabled optimized EBS I/O.

The size of the EBS volume in GB, for EBS-backed instances.

Do not delete EBS volumn on instance termination.

The flavor of EC2 instance (m1.small, m1.medium, etc).

The default value is "m1.small".

The security groups for this EC2 instance.

The default value is "infrastructure".

The AMI for the EC2 instance.

The desired AWS region, such as "us-east-1" or "us-west-2".

The default value is "us-east-1" but is strongly encouraged to be set explicitly.

Create node in this Virtual Private Cloud Subnet ID (implies VPC mode).

The AWS SSH key id.

The tags for this EC2 instance.

The resulting set will include:

• "Node=#{config[:chef_node_name]}"
• "Role=chef_server"

Note: You must install the knife-digital_ocean gem to use this subcommand. This was done to keep the dependencies of this library lighter and to make future cloud adapter support easier to add.

Provisions a Digital Ocean droplet and sets up an Open Source Chef Server as described above.

This subcommand imports all relavent options from the knife-digital_ocean gem. For detailed documentation relating to these options, please visit the project page.

Note: You must install the knife-linode gem to use this subcommand. This was done to keep the dependencies of this library lighter and to make future cloud adapter support easier to add.

Provisions a Linode instance and sets up an Open Source Chef Server as described above.

This subcommand imports all relavent options from the knife-linode gem. For detailed documentation relating to these options, please visit the docs page.

Your Linode API Key.

This option is required.

The datacenter for the server.

The default value is 3 (Use knife linode datacenter list for a list of choices)

The flavor of server.

The default value is 1. (Use knife linode flavor list for a list of choices)

The image for the server.

The default value is 93. (Use knife linode image list for a list of choices)

The kernel for the server.

The default value is 138. (Use knife linode kernel list for a list of choices)

The Linode node name for your new node.

This option is required.

The ssh password. If a password is not provided, then a random password will be generated and echoed in logging output. It is recommended that you specify a password so that you know how to connect later.

The default value is a random password.

Note: You must install the knife-openstack gem to use this subcommand. This was done to keep the dependencies of this library lighter and to make future cloud adapter support easier to add.

Provisions a Openstack instance and sets up an Open Source Chef Server as described above.

Provisions a standalone server that is reachable on the network and sets up an Open Source Chef Server as described above. You are responsible for providing the server so it could be a physical machine, Vagrant VM with host-only or bridged networking, or a cloud server instance with a known IP address or host name.

Host name or IP address of the host to bootstrap.

This option is required.

The SSH password used when bootstrapping the Chef Server node. If no password is provided but an SSH key-based authentication fails, then you will be prompted interactively for a password. In other words, if your server requires password authentication you can skip this option and type it in after the plugin starts.

Pulls Chef data primitives from a Chef Server as JSON for backup. Backups can be taken of:

• nodes
• roles
• environments
• data bags

A big thanks to Steven Danna and Joshua Timberman for the BackupExport knife plugin which was the inspiration for this implementation.

The following component types are valid:

• nodes
• roles
• environments
• data_bags (note the underscore character)

When no component types are specified, all will be selected for backup. This is equivalent to invoking:

$ knife server backup nodes roles environments data_bags

The directory to host backup files. A sub-directory for each data primitive type will be created. For example if the backup-dir was /backups/chef then all all node JSON representations would be written to /backups/chef/nodes and data bag JSON representations would be written to /backups/chef/data_bags.

The default uses the :file_backup_path configuration option, the :chef_server_url and the current time to construct a unique directory (within a second). For example, if the time was "2012-04-01 08:47:11 UTC", and given the following configuration (in knife.rb):

file_backup_path  = "/var/chef/backups"
chef_server_url   = "https://api.opscode.com/organizations/coolinc"

then a backup directory of /var/chef/backups/api.opscode.com_20120401T084711-0000 would be created.

Restores Chef data primitives from JSON backups to a Chef Server. You can restore some or all of:

• nodes
• roles
• environments
• data bags

A big thanks to Steven Danna and Joshua Timberman for the BackupRestore knife plugin which was the inspiration for this implementation.

The following component types are valid:

• nodes
• roles
• environments
• data_bags (note the underscore character)

When no component types are specified, all will be selected for restore. This is equivalent to invoking:

$ knife server restore nodes roles environments data_bags \
  --backup-dir /my/backup/dir

The directory to containing backup JSON files. A sub-directory for each data primitive type is expected (the knife server backup subcommand provides this format).

This option is required.

• Support for other platforms (alternative bootstrap templates)
• Support for Rackspace provisioning (use knife-rackspace gem)
• knife server backup backed by s3 (fog api)
• knife server restore from s3 archive (fog api)

• 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

Created and maintained by Fletcher Nichol (fnichol@nichol.ca)

Apache License, Version 2.0 (see LICENSE)