Comment 6 for bug 1563026

Frankly, whether or not I can run 100 gallion Ubuntu servers with LXD on my server with KVM/Xen/a host which needs no virtualization at all on it is irrevelant.

On a brand-new server install (which I prefer to keep to an absolute minimal) on bare-metal on Xenial, I now have:

- A bridge that does nothing for me
- By extension, a stateful firewall with iptables which
a) Could have a negative impact on high throughput hosts
b) Gets in the way of hosts which have their firewalls statically configured with Puppet/Ansible/etc
- dnsmasq which is running a DNS and DHCP service

Do we really need those things on a m1.tiny instance running on AWS?

Even purging LXD on a host causes stuff to be left behind:

Removing lxc-templates (2.0.0~rc12-0ubuntu1) ...
Removing ubuntu-server (1.350) ...
Removing lxd (2.0.0~rc5-0ubuntu1) ...
Warning: Stopping lxd.service, but it can still be activated by:
  lxd.socket
Purging configuration files for lxd (2.0.0~rc5-0ubuntu1) ...
Failed to stop lxd.service: Unit lxd.service not loaded.
dpkg: warning: while removing lxd, directory '/var/lib/lxd' not empty so not removed
Removing lxc (2.0.0~rc12-0ubuntu1) ...
Removing lxc1 (2.0.0~rc12-0ubuntu1) ...
Removing lxc dnsmasq configuration
Purging configuration files for lxc1 (2.0.0~rc12-0ubuntu1) ...
dpkg: warning: while removing lxc1, directory '/etc/lxc' not empty so not removed