Watch
0
0
Fork
You've already forked hyperhive
0

nix: move the remaining service containers onto the swarm-container module

hive-ci, hive-forge, hive-matrix, swarm-authelia, swarm-bao, swarm-grafana,
swarm-nats, swarm-otel and swarm-victorialogs now import
./swarm-container.nix and drop their own copies of the stateVersion,
firewall and resolvconf lines. Each binds privateNetwork once in its
top-level let and passes it to both the host attr and the in-container
option, as swarm-victoriametrics already does.

hive-forge (25.11) and swarm-otel (the host's value) keep their own
stateVersion over the module's mkDefault. hive-ci sets privateNetwork =
true and writesOwnResolvConf = false, which leaves its firewall and
resolvconf on, as before. hive-matrix keeps its useHostResolvConf
override and static resolv.conf; its resolvconf mkForce now comes from
the module default.

Every container's system.build.toplevel drvPath and host-side attrs
evaluate identical to the parent commit.

module-eval-swarm-services-switch gains a fixture with all ten service
containers and checks that each one's in-container privateNetwork equals
its host-side value, that the nine on the host netns run no firewall or
resolvconf, that hive-ci keeps both, and that hive-forge keeps its pinned
stateVersion.

Refs #3773
This commit is contained in:
atlas 2026-09-29 19:10:54 +02:00
commit 10d579ecaf
10 changed files with 133 additions and 135 deletions

View file

@ -242,6 +242,19 @@ let
# grants it (./swarm-bao.nix), so the daily timer below has two weeks of
# retries before it lapses.
leafRenewSeconds = 360 * 3600;
# Shared host netns, like every sibling swarm container.
#
# ⚠️ Which is exactly why the server below must refuse everyone
# until the callout responder exists: the queue is on the host's
# own loopback, in reach of every process there and every sibling
# on this netns, and each remote hive in a multi-host swarm dials
# it directly. An unauthenticated interim state would be a hole
# rather than a rough edge.
#
# Not agent containers, though: they have a netns of their own and
# the bridge firewall does not open this port.
privateNetwork = false;
in
{
# The swarm's message queue: one NATS server, reached by every hive at
@ -723,18 +736,7 @@ in
# Journal files on the host, not inside the container: nixpkgs hardcodes
# --link-journal=try-guest, and EXTRA_NSPAWN_FLAGS expands after it.
extraFlags = [ "--link-journal=host" ];
# Shared host netns, like every sibling swarm container.
#
# ⚠️ Which is exactly why the server below must refuse everyone
# until the callout responder exists: the queue is on the host's
# own loopback, in reach of every process there and every sibling
# on this netns, and each remote hive in a multi-host swarm dials
# it directly. An unauthenticated interim state would be a hole
# rather than a rough edge.
#
# Not agent containers, though: they have a netns of their own and
# the bridge firewall does not open this port.
privateNetwork = false;
inherit privateNetwork;
# The public trust bundle (empty when the gateway is not self-signed)
# and the queue's own TLS leaf, both read-only.
bindMounts = caTrust.bindMount // {
@ -747,6 +749,7 @@ in
{ ... }:
{
imports = [
../container-modules/swarm-container.nix
(import ../container-modules/swarm-container-resolver.nix {
inherit (networkCfg) bridgeIp;
# The responder introspects authelia BY NAME on every auth
@ -758,17 +761,7 @@ in
caBundleModule
];
system.stateVersion = "26.05";
# Shared host netns: this container's own firewall.service
# would rewrite the HOST ruleset at every boot. The host
# firewall owns all filtering.
networking.firewall.enable = false;
# resolvconf stays off because the resolver unit imported above
# owns /etc/resolv.conf. Leaving it on would let host-tracking
# regenerate the file empty, since the host's copy does not
# cross the boundary after start.
networking.resolvconf.enable = lib.mkForce false;
services.hyperhive.swarmContainer = { inherit privateNetwork; };
services.nats = {
enable = true;