Tom Hubrecht
31e3aabc8f
All checks were successful
Build all the nodes / ap01 (push) Successful in 1m4s
Build all the nodes / bridge01 (push) Successful in 1m48s
Build all the nodes / geo01 (push) Successful in 1m48s
Build all the nodes / compute01 (push) Successful in 2m14s
Build all the nodes / geo02 (push) Successful in 1m25s
Build all the nodes / netcore02 (push) Successful in 36s
Build all the nodes / hypervisor02 (push) Successful in 1m46s
Build all the nodes / hypervisor01 (push) Successful in 1m47s
Build all the nodes / hypervisor03 (push) Successful in 1m31s
Build all the nodes / rescue01 (push) Successful in 1m54s
Build all the nodes / vault01 (push) Successful in 1m45s
Build all the nodes / storage01 (push) Successful in 1m50s
Run pre-commit on all files / pre-commit (push) Successful in 36s
Build all the nodes / web01 (push) Successful in 2m21s
Build all the nodes / web02 (push) Successful in 1m53s
Build all the nodes / web03 (push) Successful in 1m41s
This version has exponential backoff for trying to connect to the upstream server |
||
---|---|---|
.forgejo/workflows | ||
iso | ||
keys | ||
lib | ||
LICENSES | ||
machines | ||
meta | ||
modules | ||
npins | ||
patches | ||
scripts | ||
workflows | ||
.envrc | ||
.gitignore | ||
CONTRIBUTE.md | ||
default.nix | ||
hive.nix | ||
README.md | ||
REUSE.toml | ||
shell.nix |
❄️ infrastructure
The dgnum infrastructure.
Contributing
Some instruction on how to contribute are available (in french) in /CONTRIBUTE.md. You're expected to read this document before commiting to the repo.
Some documentation for the development tools are provided in the aforementioned file.
Using the binary cache
Add the following module to your configuration (and pin this repo using your favorite tool: npins, lon, etc...):
{ lib, ... }:
let
dgnum-infra = PINNED_PATH_TO_INFRA;
in {
nix.settings = (import dgnum-infra { }).mkCacheSettings {
caches = [ "infra" ];
};
}
Adding a new machine
The first step is to create a minimal viable NixOS host, using tha means necessary. The second step is to find a name for this host, it must be unique from the other hosts.
Tip
For the rest of this part, we assume that the host is named
host02
Download the keys
The public SSH keys of host02
have to be saved to keys
, preferably only the ssh-ed25519
one.
It can be retreived with :
ssh-keyscan address.of.host02 2>/dev/null | awk '/ssh-ed25519/ {print $2,$3}'
Initialize the machine folder and configuration
- Create a folder
host02
undermachines/
- Copy the hardware configuration file generated by
nixos-generate-config
tomachines/host02/_hardware-configuration.nix
- Create a
machines/host02/_configuration.nix
file, it will contain the main configuration options, the basic content of this file should be the following
{ lib, ... }:
lib.extra.mkConfig {
enabledModules = [
# List of modules to enable
];
enabledServices = [
# List of services to enable
];
extraConfig = {
services.netbird.enable = true;
};
root = ./.;
}
Fill in the metadata
Network configuration
The network is declared in meta/network.nix
, the necessary hostId
value can be generated with :
head -c4 /dev/urandom | od -A none -t x4 | sed 's/ //'
Other details
The general metadata is declared in meta/nodes.nix
, the main values to declare are :
site
, where the node is physically locatedstateVersion
nixpkgs
, the nixpkgs version to use
Initialize secrets
Create the directory secrets
in the configuration folder, and add a secrets.nix
file containing :
(import ../../../keys).mkSecrets [ "host02" ] [
# List of secrets for host02
]
This will be used for future secret management.
Update encrypted files
Both the Arkheon, Netbox and notification modules have secrets that are deployed on all machines. To make those services work correctly, run in modules/dgn-records
, modules/dgn-netbox-agent
and modules/dgn-notify
:
agenix -r
Commit and create a PR
Once all of this is done, check that the configuration builds correctly :
colmena build --on host02
Apply it, and create a Pull Request.