hyperhive/nix/modules/hive-matrix.nix
atlas 85790b0cca nix: add breaking-change note to each openFirewall description (argus #653)
argus picked option (a) on #653: put the upgrade note in each option's
`description` so it shows up in `nix flake show` + the rendered
options docs, right next to the option itself. cheapest option, no
eval-time noise (a `warnings` block would fire on every new
deployment that wants false — the normal case now).

Appended a `**Breaking change as of #651**` paragraph to each of the
three `openFirewall` descriptions, naming the exact option string the
operator needs to set to restore the old behaviour.

Gateway's note specifically calls out that external reach is the
common case (operator's primary entry point), so the upgrade hint
is most likely needed there.
2026-05-30 19:30:11 +02:00

365 lines
16 KiB
Nix

{
pkgs,
lib,
config,
...
}:
let
cfg = config.services.hyperhive.matrix;
hyperhiveDomain = config.services.hyperhive.domain;
effectiveServerName =
if cfg.serverName != null then cfg.serverName else "matrix.${hyperhiveDomain}";
in
{
# Private Matrix homeserver (matrix-tuwunel — the official conduwuit
# successor) for hyperhive agents, wrapped in a nixos-container so it
# doesn't fight any existing `services.matrix-*` the operator may
# already run on the host. Same shape as `nix/modules/hive-forge.nix`:
# shared host netns (`privateNetwork = false`) so agents reach it at
# `http://localhost:<httpPort>` (or via the configured server_name
# for federation), nixos-container only here for state + systemd-unit
# isolation.
#
# Container name `hive-matrix` (not `h-*`) so the lifecycle scanner
# ignores it; operator manages via the standard `nixos-container` CLI.
#
# Persistent state at `/var/lib/nixos-containers/hive-matrix/var/lib/
# matrix-tuwunel/` (survives container restart / host reboot). To
# wipe, destroy the container.
#
# Initial rollout (#548): federation enabled (needed for multi-hive
# swarms; trusted_servers starts empty so no actual federation traffic
# leaves until peers are explicitly listed), registration enabled via
# a `registration_token_file` known only to hive-c0re (so agents can't
# self-register without going through the coordinator), e2ee disabled
# per operator call (tracked for follow-up at #551).
#
# Provisioning model (matches `nix/modules/hive-forge.nix` shape):
# hive-c0re generates a 32-byte random `registration_token` on first
# boot, writes it to `/var/lib/hyperhive/matrix-register-token` (mode
# 0600, root-only), and bind-mounts that file read-only into the
# tuwunel container at the same path so tuwunel can read it via
# `registration_token_file`. hive-c0re then uses the token to register
# each agent account via the matrix-spec UIAA registration flow, and
# persists the returned `access_token` to `<agent-state>/matrix-token`
# so the agent's matrix MCP client can authenticate without ever
# seeing the shared registration token.
options.services.hyperhive.matrix = {
enable = lib.mkOption {
type = lib.types.bool;
default = false;
description = ''
Run hive-matrix a private matrix-tuwunel homeserver (in a
nixos-container) for hyperhive agents. Off by default while
the integration phases in; flip to `true` once the operator
has set `services.hyperhive.domain` and is ready to onboard agents.
'';
};
package = lib.mkOption {
type = lib.types.package;
default = pkgs.matrix-tuwunel;
defaultText = lib.literalExpression "pkgs.matrix-tuwunel";
description = ''
matrix-tuwunel package to run inside the container. Defaults
to nixpkgs's `pkgs.matrix-tuwunel`. Override to pin a
specific upstream if you need an unreleased feature.
'';
};
serverName = lib.mkOption {
type = lib.types.nullOr lib.types.str;
default = null;
example = "chat.example.org";
description = ''
Matrix `server_name` the host part of every user ID
(`@argus:<server_name>`) and room ID minted on this
homeserver. CRITICAL: must be stable from day one because
it's embedded irrevocably in the identifiers. Defaults to
`matrix.''${services.hyperhive.domain}` (always a subdomain keeps
the root domain free for the dashboard or forge). Override
here only if you need a name that doesn't follow the
`matrix.<domain>` shape.
'';
};
httpPort = lib.mkOption {
type = lib.types.port;
default = 8008;
description = ''
TCP port tuwunel serves the matrix client-server API on.
Default 8008 is the matrix-spec well-known port. Sits
outside hyperhive's claimed ranges (dashboard 7000, manager
8000, sub-agents 8100..8999). Federation listens on
`federationPort` separately.
'';
};
openFirewall = lib.mkOption {
type = lib.types.bool;
default = false;
example = true;
description = ''
Open `httpPort` in the host firewall. Off by default (#651,
secure-by-default): the homeserver is reachable from the
host + every agent container via `localhost` either way
(shared netns), so the firewall open only matters for
access from outside the host. Flip to `true` when announcing
the homeserver to other hives or when an external matrix
client needs to reach the client-server API directly.
**Breaking change as of #651**: this used to default to
`true`. If you relied on the old default for external reach,
add `services.hyperhive.matrix.openFirewall = true;` to
your host config before rebuilding.
Note: federation (the matrix-spec well-known port 8448) is
intentionally not opened here. tuwunel serves the federation
API on the same `httpPort` as the client-server API by
default; reaching it on 8448 requires either binding tuwunel
to that port explicitly OR a reverse-proxy + `.well-known/
matrix/server` delegation, neither of which lives in this
module. Add that proxy config alongside whatever serves your
dashboard or forge on 443.
'';
};
trustedServers = lib.mkOption {
type = lib.types.listOf lib.types.str;
default = [ ];
example = [ "matrix.org" ];
description = ''
List of trusted matrix servers (homeservers whose signing
keys this server will fetch identity-server-style). Empty
by default federation is enabled at the protocol level
but no peer is trusted until listed here, so the homeserver
is effectively closed until the operator declares hive
peers explicitly.
'';
};
maxRequestSize = lib.mkOption {
type = lib.types.ints.positive;
default = 20000000;
description = ''
Maximum size in bytes of a single matrix client request body.
Default 20 MB matches the matrix-spec recommendation for
media uploads + the upstream tuwunel default.
'';
};
registrationTokenFile = lib.mkOption {
type = lib.types.path;
default = "/var/lib/hyperhive/matrix-register-token";
description = ''
Host path to a file containing the matrix registration token
tuwunel reads to authorise new-account creation. The token is
generated automatically by `hive-c0re` on first boot (32-byte
random hex, mode 0600) and is bind-mounted read-only into the
tuwunel container at the same path. Agents never see this
token hive-c0re uses it to provision per-agent accounts
and the agent only receives the resulting `access_token`.
Override only when integrating with externally-managed
registration tokens.
'';
};
gui = {
enable = lib.mkOption {
type = lib.types.bool;
default = cfg.enable;
defaultText = lib.literalExpression "config.services.hyperhive.matrix.enable";
description = ''
Serve a matrix web client (default `pkgs.fluffychat-web`) as
a static dist at `/matrix/` via the hive-gateway nginx
(#607 / #634). Defaults to whatever
`services.hyperhive.matrix.enable` is turning on the
homeserver gives you the web client by default; set to
`false` explicitly to opt out of the GUI while keeping
the homeserver running for agents. Requires
`services.hyperhive.gateway.enable` (default on); when
gateway is off no one hosts the GUI and the
`M4TR1X ` dashboard tab is hidden.
fluffychat-web supports per-login server pick point it at
the in-host tuwunel URL (`http://localhost:8008` by
default) the first time. The post-#15 nginx-front re-root
(`https://matrix.''${services.hyperhive.domain}`) is tracked
separately in #609.
'';
};
package = lib.mkOption {
type = lib.types.package;
default = pkgs.fluffychat-web.overrideAttrs (old: {
# fluffychat-web ships with `<base href="/">` baked into its
# index.html — flutter's `--base-href` build flag replaces
# that placeholder. Default upstream build is `--base-href "/"`
# which is wrong for hyperhive's `/matrix/` sub-path mount:
# the browser resolves relative asset paths (`Imaging.js`,
# `flutter.js`, `splash/*`) against the document ROOT,
# producing 404s for every asset (#634). Inject the right
# base-href into the actual `flutter build web` invocation
# via the upstream derivation's `flutterBuildFlags` string
# (the buildPhase is literally
# `flutter build web -v $flutterBuildFlags`). When subdomain
# routing lands (#609 — `matrix.${hyperhiveDomain}`), drop
# this override; upstream's `/` base-href is correct at the
# root of a dedicated subdomain.
flutterBuildFlags = (old.flutterBuildFlags or [ ]) ++ [
"--base-href"
"/matrix/"
];
});
defaultText = lib.literalMD "`pkgs.fluffychat-web` rebuilt with `--base-href /matrix/` via `flutterBuildFlags`.";
description = ''
Static web client dist to serve at `/matrix/`. Defaults to
`pkgs.fluffychat-web` rebuilt with `--base-href "/matrix/"`
so relative asset paths resolve under the sub-path mount
(#634). Override to swap for `hydrogen-web` (lightest),
`cinny` (no threads), `element-web` (heaviest, full
features), or an out-of-tree client dist any replacement
also needs its `<base href>` aligned with the mount path.
'';
};
};
};
config = lib.mkIf cfg.enable {
# mara on #548: "there is no default, but it is required. add
# assertion." — fail eval with a helpful message rather than
# spawning a homeserver with a bogus server_name we can never
# change later. `services.hyperhive.domain` is host-wide; matrix derives
# the server_name from it (or from `cfg.serverName` if the
# operator wants to override).
assertions = [
{
assertion = hyperhiveDomain != null || cfg.serverName != null;
message = ''
services.hyperhive.matrix.enable = true requires either:
- services.hyperhive.domain set to your host's canonical domain
(recommended; shared with forge / dashboard), or
- services.hyperhive.matrix.serverName set explicitly.
The matrix server_name is embedded into every user ID and
room ID on this homeserver it cannot be changed later
without losing every account and chat history. Pick a
stable hostname before enabling.
'';
}
];
# Pin the `tuwunel` group at a fixed GID on BOTH the host and the
# hive-matrix container. The registration token file lives on the
# host bind-mounted into the container; for tuwunel's non-root
# user inside the container to read it, the file gets `chown
# root:tuwunel` + mode `0640` in the activation script below. That
# ownership only works if the numeric GID resolves to the same
# name on both sides of the bind — without an explicit pin, the
# host's auto-allocated GID for `tuwunel` (if any) almost
# certainly wouldn't match the container's. 10042 sits well
# outside nixos's auto-allocated system-user range (200..399).
users.groups.tuwunel.gid = 10042;
# Generate the registration token at system activation time, BEFORE
# the hive-matrix container would otherwise start with an empty
# bind-mount target (argus nit on #565: nspawn creates an empty
# file when the host path is missing, tuwunel reads it as
# `registration_token_file=""` and rejects every registration
# until the next restart). Idempotent: only writes when the file
# doesn't exist. 32-byte hex = 64 chars, same shape hive-c0re's
# `matrix::ensure_register_token` would produce.
#
# Ownership: tuwunel inside the hive-matrix container runs as its
# own non-root user (nixpkgs's `services.matrix-tuwunel`), so a
# 0600 root-owned file denies it open(2) and tuwunel boots loop-
# fails with `Permission denied (os error 13)` (#644). Fix:
# `chown root:tuwunel` + `chmod 0640` so only the tuwunel group
# gains read access (no world-readable footgun, per mara). The
# `tuwunel` group GID is pinned to 10042 above so the host's name
# → number lookup matches what the container sees.
system.activationScripts.hive-matrix-register-token = lib.stringAfter [ "var" "users" ] ''
tokenFile=${lib.escapeShellArg (toString cfg.registrationTokenFile)}
if [ ! -s "$tokenFile" ]; then
mkdir -p "$(dirname "$tokenFile")"
head -c 32 /dev/urandom | od -An -tx1 | tr -d ' \n' > "$tokenFile"
echo >> "$tokenFile"
echo "hive-matrix: generated registration token at $tokenFile"
fi
# Always re-apply ownership + mode (covers existing 0600 root-
# owned files from pre-#644 deployments; `chown` + `chmod` are
# both idempotent).
chown root:tuwunel "$tokenFile"
chmod 0640 "$tokenFile"
'';
containers.hive-matrix = {
autoStart = true;
ephemeral = false;
# Share host netns — tuwunel's listeners look exactly like
# host-side services, no port-forward plumbing, and agent
# containers (also host netns) reach it via plain `localhost`.
privateNetwork = false;
# Read-only bind of the host-managed registration token so
# tuwunel can resolve `registration_token_file` to a real
# file inside the container. The activation script above
# ensures the host path exists with a valid 64-char hex token
# before any container starts, so the bind always finds real
# content (no first-boot empty-file race; argus #565 nit).
bindMounts.${cfg.registrationTokenFile} = {
hostPath = cfg.registrationTokenFile;
isReadOnly = true;
};
config =
{ ... }:
{
system.stateVersion = "26.05";
# Mirror the host's pinned `tuwunel` GID so the bind-mounted
# registration token (chowned `root:tuwunel` on the host)
# resolves to the same group inside the container. Without
# this pin nixos auto-allocates whatever's free, the two
# sides diverge, and tuwunel's user falls back to the
# "other" mode bits (= no read) on the file. See the host-
# side `users.groups.tuwunel.gid` above.
users.groups.tuwunel.gid = 10042;
services.matrix-tuwunel = {
enable = true;
package = cfg.package;
settings.global = {
server_name = effectiveServerName;
# `address` is `listOf nonEmptyStr` upstream (multi-bind
# support). Single-host bind goes through as a one-element list.
address = [ "0.0.0.0" ];
# `port` is `listOf port` upstream. Same shape.
port = [ cfg.httpPort ];
max_request_size = cfg.maxRequestSize;
# Federation enabled at the protocol level so swarms
# can be wired up later by extending `trustedServers`
# without a homeserver restart. Empty trusted_servers
# keeps it effectively closed until peers are listed.
allow_federation = true;
trusted_servers = cfg.trustedServers;
# Token-gated registration: hive-c0re holds the token,
# agents never see it. allow_registration must be true
# for the token flow to engage; the absent
# `yes_i_am_very_very_sure_…_open_registration_…` flag
# keeps the server closed to anyone without the token.
allow_registration = true;
registration_token_file = toString cfg.registrationTokenFile;
# E2EE disabled in initial rollout per operator call
# (#548) — re-enabling tracked at #551.
allow_encryption = false;
};
};
environment.systemPackages = [ cfg.package ];
};
};
networking.firewall = lib.mkIf cfg.openFirewall {
allowedTCPPorts = [
cfg.httpPort
];
};
};
}