## Why OEM?

Creating an "OEM Install" allows a user to create an image that can be installed

on many systems, which requests more information on the next boot up, or to use a

VM base image that can be cloned any number of times. Windows admins may be familiar

with this process using the "sysprep" tool. The same can be done on linux by combining

a few tools.

While Ubuntu desktop has a built-in OEM mode (guide available [here](https://help.ubuntu.com/community/Ubuntu_OEM_Installer_Overview), Ubuntu Server does not, requiring you to

build your own.

## Base Install

Start by doing a standard ubuntu server install (we're using 18.04).

Use "ubuntu" as the username and password.

When the install is done, make any configuration changes you want added to the base

image. For example:

- package updates

- package installs

- modifications to configuration files

- cronjobs

- modifications to /etc/skel

## Suggested scripts

TODO :

- Add partition expand script

something like this (NOTE: UNTESTED):

start=$(cat /sys/block/mmcblk0/mmcblk0p3/start)

end=$(($start+$(cat /sys/block/mmcblk0/mmcblk0p3/size)))

newend=$(($(cat /sys/block/mmcblk0/size)-8))

if [ "$newend" -gt "$end" ]

then

parted -s /dev/mmcblk0 unit s resize 3 $start $newend

fi

## Prepare for OEM Configuration

When these changes are complete, and you're ready to creae your clean image,

install package `oem-config`:

```bash

$ sudo apt update && sudo apt install -y oem-config ;

```

Once installed, run the following command. This will setup the system to refresh

itself on the next boot.

```bash

$ sudo oem-config-prepare

```

Finally, once complete, shut the system down.

Your system is now ready to clone and boot clones.

## TODO BEFORE COMPLETE

TODO: Look at this alternative option:

https://docs.openstack.org/image-guide/ubuntu-image.html

https://askubuntu.com/questions/736484/oem-install-ubuntu-server-14-04

https://askubuntu.com/questions/717842/oem-install-of-ubuntu-server-14-04-3

https://wiki.ubuntu.com/OemConfigServer

## ALTERNATIVE METHOD

https://docs.openstack.org/image-guide/ubuntu-image.html

``` bash

$ sudo apt-get install cloud-init -y

```

```bash

$ sudo dpkg-reconfigure cloud-init

```

Uncheck "ec2".

Run the sysprep command on the vmname in virsh

```bash

virt-sysprep -d vmname

```

## Method 2

Run a normal install. Reboot, make whatever changes are required.

Then, shut down, make a clone to sysprep.

Then, find the VM name, and run `virt-sysprep` on it. You can find VM

names with `virsh list --all` if you're not sure.

Run a dry run to see what commands will run. This will run the full process

but throw away the changes at the end.

```bash

$ virt-sysprep --dry-run --domain vmname

```

If you are satisfied, run without `--dry-run`:

```bash

$ virt-sysprep --domain vmname

```

Additioally, options like `--firstboot` or `--firstboot-command` can be used

to insert startup scripts.

These can be found here:

http://libguestfs.org/virt-sysprep.1.html

TODO : get a `--firstboot` option working

Untested example:

```bash

$ sudo virt-sysprep --domain ubuntu-apache-base --firstboot-command 'apt update && apt install -y apache2 && reboot' --hostname 'baseos1'

```

However, if you're just installing packages, you can do that at buildtime instad of boot

time, with a command like this:

The `--install` command installs the package into the vm image, and `--network` enables the

networking to do so at build time.

```bash

sudo virt-sysprep --domain ubuntu-apache-base --install apache2 --network --firstboot-command 'apt update && apt dist-upgrade -y && apt autoremove -y && reboot' --hostname 'ubuntu-apache-base'

```

For all the available options, see the [official documentation page](http://libguestfs.org/virt-sysprep.1.html).

# Oneliner example to create a cluster

for f in {00..05} ; do sudo virsh destroy rancher$f ; sudo virsh undefine rancher$f; sudo virt-clone --original=ubuntu-base --name=rancher$f --auto-clone ; sudo virt-sysprep --domain rancher$f --firstboot-command 'while true ; do apt update && break ; ps aux | grep dpk; sleep 1; done ; dpkg --configure -a && while true ; do apt update && break ; sleep 1; done ; apt install curl -y; while true ; do curl https://get.docker.com | sudo bash && break ; ps aux | dpk ; done ; dpkg --configure -a && while true ; do apt update && break ; sleep 1; done ; apt dist-upgrade -y && apt autoremove -y && reboot ; echo test' --hostname "rancher$f"; virt-manager --show-domain-console rancher$f --connect 'qemu:///system'; sudo virsh start rancher$f; done