You can configure OpenStack to write metadata to a special configuration drive that attaches to the instance when it boots. The instance can mount this drive and read files from it to get information that is normally available through the metadata service. One use case for the configuration drive is to pass a networking configuration when you do not use DHCP to assign IP addresses to instances. For example, you might pass the IP configuration for the instance through the configuration drive, which the instance can mount and access before the you configure the network settings for the instance. Any modern guest operating system that is capable of mounting an ISO9660 or VFAT file system can use the configuration drive.

The following hypervisors support the configuration drive: libvirt, xenserver, hyper-v, and vmware. To use configuration drive with libvirt, xenserver, or vmware, you must first install the genisoimage package on each Compute host. Otherwise, instances do not boot properly. Use the mkisofs_cmd flag to set the path where you install the genisoimage program. If genisoimage is in same path as the nova-compute service, you do not need to set this flag. By default, Ubuntu packages do not install this package. See bug #1165174. To use configuration drive with hyper-v, you must set the mkisofs_cmd value to the full path to an mkisofs.exe installation. Additionally, you must set the qemu_img_cmd value in the hyperv configuration section to the full path to an qemu-img command installation. An image built with a recent version of the cloud-init package can automatically access metadata passed through the configuration drive. The cloud-init package version 0.7.1 works with Ubuntu and Fedora-based images, such as RHEL. If an image does not have the cloud-init package installed, you must customize the image to run a script that mounts the configuration drive on boot, reads the data from the drive, and takes appropriate action such as adding the public key to an account. See for details on how data is organized on the configuration drive. If you use Xen with a configuration drive, use the xenapi_disable_agent configuration parameter to disable the agent. Do not rely on the presence of the EC2 metadata present in the configuration drive, as this content might be removed in a future release. For example, do not rely on files in the ec2 directory. When you create images that access configuration drive data and multiple directories are under the openstack directory, always select the highest API version by date that your consumer supports. For example, if your guest image supports the 2012-03-05, 2012-08-05, 2013-04-13 versions, try 2013-04-13 first and fall back to a previous version if 2013-04-13 is not present.

To enable the configuration drive, pass the --config-drive=true parameter to the nova boot command. This example enables the configuration drive and passes user data, two files, and two key/value metadata pairs, all of which are accessible from the configuration drive: $ nova boot --config-drive=true --image my-image-name --key-name mykey --flavor 1 --user-data ./my-user-data.txt myinstance --file /etc/network/interfaces=/home/myuser/instance-interfaces --file known_hosts=/home/myuser/.ssh/known_hosts --meta role=webservers --meta essential=false You can also configure the Compute service to always create a configuration drive. Set this option in the /etc/nova/nova.conf file: force_config_drive=true If a user passes the --config-drive=true flag to the nova boot command, an administrator cannot disable the configuration drive. The configuration drive has the config-2 volume label. If your guest operating system supports accessing disk by label, you can mount the configuration drive as the /dev/disk/by-label/config-2 device. For example: # mkdir -p /mnt/config # mount /dev/disk/by-label/config-2 /mnt/config Make sure that you use at least version 0.3.1 of CirrOS for configuration drive support. If your guest operating system does not use udev, the /dev/disk/by-label directory is not present. You can use the blkid command to identify the block device that corresponds to the configuration drive. For example, when you boot the CirrOS image with the m1.tiny flavor, the device is /dev/vdb: # blkid -t LABEL="config-2" -odevice /dev/vdb Once identified, you can mount the device: # mkdir -p /mnt/config # mount /dev/vdb /mnt/config In this example, the contents of the configuration drive are: ec2/2009-04-04/meta-data.json ec2/2009-04-04/user-data ec2/latest/meta-data.json ec2/latest/user-data openstack/2012-08-10/meta_data.json openstack/2012-08-10/user_data openstack/content openstack/content/0000 openstack/content/0001 openstack/latest/meta_data.json openstack/latest/user_data The files that appear on the configuration drive depend on the arguments that you pass to the nova boot command. The following example shows the contents of the openstack/2012-08-10/meta_data.json and openstack/latest/meta_data.json files. These files are identical. The file contents are formatted for readability: Note the effect of the --file /etc/network/interfaces=/home/myuser/instance-interfaces argument that was passed to the nova boot command. The contents of this file are contained in the openstack/content/0000 file on the configuration drive, and the path is specified as /etc/network/interfaces in the meta_data.json file. The following example shows the contents of the ec2/2009-04-04/meta-data.json, latest/meta-data.json files. These files are identical. The file contents are formatted to improve readability: The openstack/2012-08-10/user_data, openstack/latest/user_data, ec2/2009-04-04/user-data, and ec2/latest/user-data file are present only if the --user-data flag and the contents of the user data file are passed to the nova boot command. The default format of the configuration drive as an ISO 9660 file system. To explicitly specify the ISO 9660 format, add the following line to the /etc/nova/nova.conf file: config_drive_format=iso9660 By default, you cannot attach the configuration drive image as a CD drive instead of as a disk drive. To attach a CD drive, add this line to the /etc/nova/nova.conf file: config_drive_cdrom=true For legacy reasons, you can configure the configuration drive to use VFAT format instead of ISO 9660. It is unlikely that you would require VFAT format because ISO 9660 is widely supported across operating systems. However, to use the VFAT format, add the following line to the /etc/nova/nova.conf file: config_drive_format=vfat If you choose VFAT, the configuration drive is 64 MBs.

The following table shows the configuration options for the configuration drive: