Configuring instances at boot time

Users often want to do some configuration to their instances after booting. For example, you may want to install some packages, start services, or manage the instance using a Puppet or Chef server. When launching instances in an OpenStack cloud, there are two technologies that work together to support automated configuration of instances at boot time: user data and cloud-init.

User data is the mechanism by which a user can pass information contained in a local file to an instance at launch time. The typical use case is to pass something like a shell script or a configuration file as user data.

User data is sent using the --user-data /path/to/filename option when calling nova boot. The following example creates a text file and then send the contents of that file as user data to the instance.

$ echo "This is some text" > myfile.txt
$ nova boot --user-data ./myfile.txt --image myimage myinstance

The instance can retrieve user data by querying the metadata service at using either the OpenStack metadata API or the EC2 compatibility API:

$ curl http://169.254.169.254/2009-04-04/user-data
This is some text
$ curl http://169.254.169.254/openstack/2012-08-10/user_data
This is some text

Note that the Compute service treats user data as a blob. While the example above used a text file, user data can be in any format.

To do something useful with the user data, the virtual machine image must be configured to run a service on boot that retrieves the user data from the metadata service and take some action based on the contents of the data. The cloud-init package was designed to do exactly this. In particular, cloud-init is compatible with the Compute metadata service as well as the Compute config drive.

Note that cloud-init is not an OpenStack technology. Rather, it is a package that is designed to support multiple cloud providers, so that the same virtual machine image can be used in different clouds without modification. Cloud-init is an open source project, and the source code is available on Launchpad. It is maintained by Canonical, the company which runs the Ubuntu project. All Ubuntu cloud images come pre-installed with cloud-init. However, cloud-init is not designed to be Ubuntu-specific, and has been successfully ported to Fedora.

We recommend installing cloud-init on images that you create to simplify the task of configuring your instances on boot. Even if you do not wish to use user data to configure instance behavior at boot time, cloud-init provides useful functionality such as copying the public key to an account (the ubuntu account by default on Ubuntu instances, the ec2-user by default in Fedora instances).

If you do not have cloud-init installed, you will need to manually configure your image to retrieve the public key from the metadata service on boot and copy it to the appropriate account.

We recommend taking a look at the cloud-init doc/userdata.txt file the examples directory as well as the Ubuntu community documentation for details on how to use cloud-init. We provide some basic examples here.

Cloud-init supports several different input formats for user data. We briefly discuss two commonly used formats:

Assuming you have cloud-init installed, the simplest way to configure an instance on boot is to pass a shell script as user data. The shell file must begin with #! in order for cloud-init to recognize it as a shell script. Here's an example of a script that creates an account called clouduser.

#!/bin/bash
adduser --disabled-password --gecos "" clouduser

Sending a shell script as user data has a similar effect to writing an /etc/rc.local script: it will be executed very late in the boot sequence as root.

Cloud-init supports a YAML-based config format that allows the user to configure a large number of options on a system. User data that begins with #cloud-config will be interpreted by cloud-init as cloud-config format.

This cloud-init user data example sets the hostname and the FQDN, as well as updating /etc/hosts on the instance:

#cloud-config
hostname: mynode
fqdn: mynode.example.com
manage_etc_hosts: true

This cloud-init user data example, based on doc/examples/cloud-config-puppet.txt, would configure the instance to contact a Puppet server at puppetmaster.example.org and verify its identity using a certificate.

#cloud-config
puppet:
 conf:
   agent:
     server: "puppetmaster.example.org"
   ca_cert: |
     -----BEGIN CERTIFICATE-----
     MIICCTCCAXKgAwIBAgIBATANBgkqhkiG9w0BAQUFADANMQswCQYDVQQDDAJjYTAe
     Fw0xMDAyMTUxNzI5MjFaFw0xNTAyMTQxNzI5MjFaMA0xCzAJBgNVBAMMAmNhMIGf
     MA0GCSqGSIb3DQEBAQUAA4GNADCBiQKBgQCu7Q40sm47/E1Pf+r8AYb/V/FWGPgc
     b014OmNoX7dgCxTDvps/h8Vw555PdAFsW5+QhsGr31IJNI3kSYprFQcYf7A8tNWu
     1MASW2CfaEiOEi9F1R3R4Qlz4ix+iNoHiUDTjazw/tZwEdxaQXQVLwgTGRwVa+aA
     qbutJKi93MILLwIDAQABo3kwdzA4BglghkgBhvhCAQ0EKxYpUHVwcGV0IFJ1Ynkv
     T3BlblNTTCBHZW5lcmF0ZWQgQ2VydGlmaWNhdGUwDwYDVR0TAQH/BAUwAwEB/zAd
     BgNVHQ4EFgQUu4+jHB+GYE5Vxo+ol1OAhevspjAwCwYDVR0PBAQDAgEGMA0GCSqG
     SIb3DQEBBQUAA4GBAH/rxlUIjwNb3n7TXJcDJ6MMHUlwjr03BDJXKb34Ulndkpaf
     +GAlzPXWa7bO908M9I8RnPfvtKnteLbvgTK+h+zX1XCty+S2EQWk29i2AdoqOTxb
     hppiGMp0tT5Havu4aceCXiy2crVcudj3NFciy8X66SoECemW9UYDCb9T5D0d
     -----END CERTIFICATE-----
 

This cloud-init user data example, based on doc/examples/cloud-config/chef.txt, and intended for use in an Ubuntu image, would add the Chef apt repository, install Chef, connect to a Chef server at https://chefserver.example.com:4000 and install Apache.

#cloud-config
apt_sources:
 - source: "deb http://apt.opscode.com/ $RELEASE-0.10 main"
   key: |
     -----BEGIN PGP PUBLIC KEY BLOCK-----
     Version: GnuPG v1.4.9 (GNU/Linux)

     mQGiBEppC7QRBADfsOkZU6KZK+YmKw4wev5mjKJEkVGlus+NxW8wItX5sGa6kdUu
     twAyj7Yr92rF+ICFEP3gGU6+lGo0Nve7KxkN/1W7/m3G4zuk+ccIKmjp8KS3qn99
     dxy64vcji9jIllVa+XXOGIp0G8GEaj7mbkixL/bMeGfdMlv8Gf2XPpp9vwCgn/GC
     JKacfnw7MpLKUHOYSlb//JsEAJqao3ViNfav83jJKEkD8cf59Y8xKia5OpZqTK5W
     ShVnNWS3U5IVQk10ZDH97Qn/YrK387H4CyhLE9mxPXs/ul18ioiaars/q2MEKU2I
     XKfV21eMLO9LYd6Ny/Kqj8o5WQK2J6+NAhSwvthZcIEphcFignIuobP+B5wNFQpe
     DbKfA/0WvN2OwFeWRcmmd3Hz7nHTpcnSF+4QX6yHRF/5BgxkG6IqBIACQbzPn6Hm
     sMtm/SVf11izmDqSsQptCrOZILfLX/mE+YOl+CwWSHhl+YsFts1WOuh1EhQD26aO
     Z84HuHV5HFRWjDLw9LriltBVQcXbpfSrRP5bdr7Wh8vhqJTPjrQnT3BzY29kZSBQ
     YWNrYWdlcyA8cGFja2FnZXNAb3BzY29kZS5jb20+iGAEExECACAFAkppC7QCGwMG
     CwkIBwMCBBUCCAMEFgIDAQIeAQIXgAAKCRApQKupg++Caj8sAKCOXmdG36gWji/K
     +o+XtBfvdMnFYQCfTCEWxRy2BnzLoBBFCjDSK6sJqCu5Ag0ESmkLtBAIAIO2SwlR
     lU5i6gTOp42RHWW7/pmW78CwUqJnYqnXROrt3h9F9xrsGkH0Fh1FRtsnncgzIhvh
     DLQnRHnkXm0ws0jV0PF74ttoUT6BLAUsFi2SPP1zYNJ9H9fhhK/pjijtAcQwdgxu
     wwNJ5xCEscBZCjhSRXm0d30bK1o49Cow8ZIbHtnXVP41c9QWOzX/LaGZsKQZnaMx
     EzDk8dyyctR2f03vRSVyTFGgdpUcpbr9eTFVgikCa6ODEBv+0BnCH6yGTXwBid9g
     w0o1e/2DviKUWCC+AlAUOubLmOIGFBuI4UR+rux9affbHcLIOTiKQXv79lW3P7W8
     AAfniSQKfPWXrrcAAwUH/2XBqD4Uxhbs25HDUUiM/m6Gnlj6EsStg8n0nMggLhuN
     QmPfoNByMPUqvA7sULyfr6xCYzbzRNxABHSpf85FzGQ29RF4xsA4vOOU8RDIYQ9X
     Q8NqqR6pydprRFqWe47hsAN7BoYuhWqTtOLSBmnAnzTR5pURoqcquWYiiEavZixJ
     3ZRAq/HMGioJEtMFrvsZjGXuzef7f0ytfR1zYeLVWnL9Bd32CueBlI7dhYwkFe+V
     Ep5jWOCj02C1wHcwt+uIRDJV6TdtbIiBYAdOMPk15+VBdweBXwMuYXr76+A7VeDL
     zIhi7tKFo6WiwjKZq0dzctsJJjtIfr4K4vbiD9Ojg1iISQQYEQIACQUCSmkLtAIb
     DAAKCRApQKupg++CauISAJ9CxYPOKhOxalBnVTLeNUkAHGg2gACeIsbobtaD4ZHG
     0GLl8EkfA8uhluM=
     =zKAm
     -----END PGP PUBLIC KEY BLOCK-----

chef:
 install_type: "packages"
 server_url: "https://chefserver.example.com:4000"
 node_name: "your-node-name"
 environment: "production"
 validation_name: "yourorg-validator"
 validation_key: |
     -----BEGIN RSA PRIVATE KEY-----
     YOUR-ORGS-VALIDATION-KEY-HERE
     -----END RSA PRIVATE KEY-----
 run_list:
  - "recipe[apache2]"
  - "role[db]"
 initial_attributes:
    apache:
      prefork:
        maxclients: 100
      keepalive: "off"