Go to file
Robert Collins 012116cad6 Move everything 'common' into base, making it avoidable for tests.
Change-Id: I87edd00f6e0edcd609856192c2c98604ef3274f0
2012-12-14 13:41:57 +13:00
bin Make it possible to not recompress the qcow2 image at the end. 2012-12-14 13:25:18 +13:00
docs Use base in all examples. 2012-11-22 15:07:15 +13:00
elements Move everything 'common' into base, making it avoidable for tests. 2012-12-14 13:41:57 +13:00
lib Move everything 'common' into base, making it avoidable for tests. 2012-12-14 13:41:57 +13:00
scripts Fix copyrights for HP work. 2012-11-15 16:20:32 +13:00
sudoers.d Jenkins image fixed 2012-11-27 18:37:57 +01:00
.gitignore Ignore temporary files. 2012-10-26 12:20:15 +13:00
.gitreview Add .gitreview file. 2012-11-28 08:57:46 -08:00
LICENSE Fix copyrights for HP work. 2012-11-15 16:20:32 +13:00
README.md Rename flavour to element. 2012-12-05 14:04:58 -08:00

Image building tools for Openstack

These tools are the components of tripleo (https://github.com/tripleo/demo) that do the plumbing involved in building disk images. Specific configs live in the demo repository, while the reusable tools live here.

What tools are there?

  • disk-image-create -o filename {element} [{element} ...] : Create an image of element {element}, optionally mixing in other elements.

  • ramdisk-image-create -o filename {element} [{element} ...] : Create a kernel+ ramdisk pair for running maintenance on bare metal machines (deployment, inventory, burnin etc).

    ramdisk-image-create -o deploy.ramdisk deploy

  • disk-image-get-kernel filename : Extract the appropriate kernel and ramdisk to use when doing PXE boot using filename as the image for a machine.

  • elements can be found in the top level elements directory.

Why?

Automation: While users and operators can manually script or put together ram disks and disk images, mature automation makes customisation and testing easier.

Design

Images are built using a chroot and bind mounted /proc /sys and /dev. The goal of the image building process is to produce blank slate machines that have all the necessary bits to fulfill a specific purpose in the running of an Openstack cloud: e.g. a nova-compute node.

An element is a particular set of code that alters how the image is built, or runs within the chroot to prepare the image. E.g. the local-config element copies in the http proxy and ssh keys of the user running the image build process into the image, whereas the vm element makes the image build a regular VM image with partition table and installed grub boot sector. The mellanox element adds support for mellanox infiniband hardware to both the deploy ramdisk and the built images.

Images start as a base ubuntu cloud image. Other distributions may be added in future, the infrastructure deliberately makes few assumptions about the exact operating system is use. The base image has opensshd running (a new key generated on first boot) and accepts use keys via the cloud metadata service, loading them into the 'ubuntu' user.

The goal of a built image is to have any global configuration ready to roll, but nothing that ties it to a specific cloud instance: images should be able to be dropped into a test cloud and validated, and then deployed into a production cloud (usually via bare metal nova) for production use. As such, the image contents can be modelled as three distinct portions:

  • global content: the actual code, kernel, always-applicable config (like disabling password authentication to sshd).
  • metadata / config management provided configuration: user ssh keys, network address and routes, configuration management server location and public key, credentials to access other servers in the cloud. These are typically refreshed on every boot.
  • persistent state: sshd server key, database contents, swift storage areas, nova instance disk images, disk image cache. These would typically be stored on a dedicated partition and not overwritten when re-deploying the image.

The goal of the image building tools is to create machine images that content the correct global content and are ready for 'last-mile' configuration by the nova metadata API, after which a configuration management system can take over (until the next deploy, when it all starts over from scratch).

Existing elements

Elements are found in the subdirectory elements. Each element is in a directory named after the element itself. Elements should have a README.md in the root of the element directory describing what it is for.

Writing an element

Make as many of the following subdirectories as you need, depending on what part of the process you need to customise:

  • block-device-size.d: Alter the size (in GB) of the disk image. This is useful when a particular element will require a certain minimum (or maximum) size. You can either error and stop the build, or adjust the size to match. NB: Due to the current simple implementation, the last output value wins so this should be used rarely - only one element in a mix can reliably set a size.

  • outputs: $IMAGE_SIZE={size_in_GB}

  • inputs: $IMAGE_SIZE={size_in_GB}

  • block-device.d: customise the block device that the image will be made on (e.g. to make partitions).

  • outputs: $IMAGE_BLOCK_DEVICE={path}

  • inputs: $IMAGE_BLOCK_DEVICE={path}

  • extra-data.d: pull in extra data from the host environment that hooks may need during image creation. This should copy any data (such as SSH keys, http proxy settings and the like) somewhere under $TMP_HOOKS_PATH.

  • outputs: None

  • inputs: $TMP_HOOKS_PATH

  • pre-install.d: Run code in the chroot before customisation or packages are installed. A good place to add apt repositories.

  • install.d: Runs after pre-install.d in the chroot. This is a good place to install packages, chain into configuration management tools or do other image specific operations.

  • first-boot.d: Runs inside the image before rc.local. Scripts from here are good for doing per-instance configuration based on cloud metadata.

Ramdisk elements support the following files in their element directories:

  • binary-deps : executables required to be fed into the ramdisk. These need to be present in your $PATH.

  • init : a POSIX shell script fragment that will be appended to the default script executed as the ramdisk is booted (/init)

Third party elements

Pending implementation. The idea is to have a search path for elements.

Installation

  • Clone the repository locally, then add bin to your path.

  • Copy sudoers.d/* into your /etc/sudoers.d/. (Warning, use visudo -c -f {filename} to check that each one parses successfully on your machine, so you don't break your machine).

Invocation

The scripts can generally just be run. Options can be set on the command line or by exporting variables to override those present in lib/img-defaults. -h to get help.

Copyright

Copyright 2012 Hewlett-Packard Development Company, L.P. Copyright (c) 2012 NTT DOCOMO, INC.

All Rights Reserved.

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at

http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.