# Tooling

## Control
- lxc _(and sub-tools)_

## Building container images
- debootstrap
_Note: it allows importing docker rootfs aswell, however it isn't recommended_

### Base images
- Alpine - edge

### Build setup
- Overlay directories to reuse the base rootfs and add configs on top

### Build location
- Servers _(prefferably)_
- Local _(fallback)_

# Setup

## Initial clustering
- Manual
- Preseed _(needs reading and some dynamic variables, could be done via Ansible)_

## Deployment
- Manual _(automated with the help of the shell on need)_
- Likely small use of volumes _(live migration and snapshots allow easy migration)_
_TODO: Check if volume snapshots are a thing_
- Limited base image modification _(especially if it doesn't need any on-disk data)_
- Port exposure needs to be defined manually on a per container basiss using proxy devices, _see: [Proxy Protocol for Proxy device in LXD](https://discuss.linuxcontainers.org/t/performance-of-the-proxy-device-compared-to-traditional-iptables-rules/2052/6)_
- Ports can be forwarded in the firewall with DNAT aswell
	- Containers need IPv6 endpoint in DNAT mode

## Backup
- Container/Volume snapshots
- Eventual backup locations:
	- Amazon S3
	- Hetzner storage boxes
	- Local _(occasional)_

# Documenting/Testing list
- DNS (mostly unknown)
	- Endpoint from containers? _(/etc/resolv.conf? DHCP? Fixed?)_
	- TLD .lxd (default)
- Tiered networks
	- Most likely possible with special interfaces
	- Needs host configuration
- Replicas
	- Looks like it is not possible