How to Deploy a New OS Image

Concept

When a virtual machine is created, some actions need to be done within the guest like resizing of the filesystem, creation of users,... To accomplish this, cloud-init is used. For most Linux distributions OS cloud images with cloud-init preinstalled are readily available for download. The rest of this page assumes you have such an OS image with cloud-init available. There is a guide on how to create a windows base image with cloudinit.

Clone the repository of your environment

From a well prepared computer, as documented here, your first step will be to clone the repository of your environment from GitHub:

git clone [email protected]:gig-projects/$name-of-your-env-repository$

Create a new image directory and the AYS service recipe for your new image

For each service you want to add to your environment you need to create a new subdirectory under $name-of-your-env-repository$/servicetemplates, so also for adding a new image you need to create a subdirectory there:

cd $name-of-your-env-repository$/servicetemplates

Let's create an subdirectory for the image available from https://cloud-images.ubuntu.com/xenial/current/xenial-server-cloudimg-amd64-uefi1.img:

mkdir image_xenial-server

Each directory contains 3 files:

services.hrd contains the all required information about ftp server that holds your image
- url: address of your FTP server from where the image is available, e.g. ftp://pub:[email protected]
- source: exact location on the FTP server from where the image can be downloaded, e.g. /images/image_windows2012
- checkmd5: whether the MD5 checksum needs to be checked, typically 'true'
- dest: directory where the image needs to be downloaded, e.g. /opt/jumpscale7/var/tmp/templates/image_windows2012.qcow2
actions.py defines the configure method that will be called to register the image once it got downloaded successfully
- name sets the image name, e.g. 'image_windows2012'
- imagename sets the disk image name as physically saved, e.g. 'image_windows2012.qcow2'
- registerImage(serviceObj, name, imagename, category, minimum-image-size) the actual registration method, where you have two additional parameters:
  - category specifies under which category the image will be available to the end user, typically 'Linux' or 'Windows'
  - minimum-image-size sets the minimum image size, in case of a windows image you have to set it to at least 20, otherwise 10 is fine for linux images
instance.hrd (leave it empty)

We actually only need to create 2 files, representing the AYS service recipe for this image:

service.hrd
actions.py

Starting with service.hrd:

vi service.hrd

Provide following service description:

platform.supported             = 'generic'

web.export.1                   =
    checkmd5:'false',
    dest:'/opt/jumpscale7/var/tmp/templates/xenial-server-cloudimg-amd64-uefi1.qcow2',
    source:'/xenial/current/xenial-server-cloudimg-amd64-uefi1.img',
    url:'https://cloud-images.ubuntu.com',

Save and close the new file by first pressing esc, typing :wq and hitting enter.

And now let's create the actions.py file:

vi actions.py

We only need to implement the configure() method:

from JumpScale import j

ActionsBase=j.atyourservice.getActionsBaseClass()

class Actions(ActionsBase):
    def configure(self, serviceObj):
        from CloudscalerLibcloud.imageutil import registerImage
        name = 'Xenial Server 16.04 amd64'
        imagename = 'xenial-server-cloudimg-amd64-uefi1.qcow2'
        registerImage(serviceObj, name, imagename, 'Linux', 10)

Save, commit and push your changes to the repo

Commit and push the newly created subdirectory and files to GitHub:

cd $name-of-your-env-repository$
git config --global push.default simple
git add servicetemplates/image_xenial-server
git commit -m "new image"
git push

Install the image

Connect to ovc_git as documented here.

Update the repository:

cd /opt/code/github/gig-projects/$name-of-your-env-repository$
git pull

Make the updated servicetemplates directory current:

cd /opt/code/github/gig-projects/$name-of-your-env-repository$/servicetemplates

And finally install the image, make sure to specify the name of one/any of the physical nodes (last option argument):

ays install -n image_xenial-server --targettype node.ssh --targetname $name-of-a-node-in-your-env$

In case you're updating an already previously installed image, use the -r option:

ays install -r -n image_xenial-server --targettype node.ssh --targetname $name-of-a-node-in-your-env$

Set image availability

In the Cloud Broker Portal go to the Image Details page for your newly added image and select Image Availability from the Actions menu:

Confirm on which nodes you want to make this image available:

Deploy a New Image