10-17, 13:45–13:50 (UTC), Main Stream
We will explore how to build and manage dozens of NixOS virtual machines in a Proxmox cluster using NixOps in a declarative fashion.
Managing a fleet of NixOS machines can be achieved using supported backends of NixOps or other tooling (Disnix, Krops, Morph, etc.), but most of them do not solve the "initialization" of your host, e.g. installing NixOS, starting the
node, except if you're using a major cloud provider like AWS for example.
If you're running your own mini-datacenter and want to manage the whole infrastructure end to end, using the Nix language, it would prove quite difficult.
nixops-proxmox is one step towards a NixOS-heavy datacenter, e.g. storage backends, PCI devices, network adapters can be considered as resources and ultimately all of them can be controlled from a "reproducible" and centralized Nix expression.
This talk will feature how to use IPv6 (!) inside a NixOps plugin, especially useful for IPv6-only networks.
FOSS developer for more than five years, (eternal?) student in mathematics and computer science, interested into formal methods applied to computer security, networks and more.
I enjoy playing around with infrastructure and run my own micro-datacenter in order to understand how can we develop an (economic viable?) model of an NixOS-only and IPv6-only datacenter as much as possible.
And then, someday, I will write a formal proof that the Nix core is sound. But that will be for another day.