From 078ea74ba9b0a676db35b919437a6aad81c911d4 Mon Sep 17 00:00:00 2001
From: sock <sock@hyperhive>
Date: Fri, 10 Jul 2026 17:30:25 +0200
Subject: [PATCH] fix(hive-ci): static bridge IP instead of DHCP (no DHCP
 server on the bridge)

hive-ci is the one container on the hive bridge that used DHCP
(networking.interfaces.eth0.useDHCP = true, from #2310 / #2336). But the
bridge has no DHCP server: dnsmasq on the bridge is DNS-only and agent
containers get deterministic static IPs (lifecycle::agent_network_ip),
so the DHCP client never gets a lease and the container hangs at boot:

  A start job is running for DHCP Client (5s / 1min 31s)

Assign a static address on eth0 (top host address of the subnet) plus a
default route via the bridge gateway, and drop useDHCP. Preserves the
#2310 netns isolation (no host-loopback reach) while letting the
container boot and reach the forge through the gateway.

Refs #2310, #2336.
---
 nix/modules/hive-ci.nix | 36 ++++++++++++++++++++++++++++--------
 1 file changed, 28 insertions(+), 8 deletions(-)

diff --git a/nix/modules/hive-ci.nix b/nix/modules/hive-ci.nix
index 96ddebdb..986d468b 100644
--- a/nix/modules/hive-ci.nix
+++ b/nix/modules/hive-ci.nix
@@ -11,6 +11,19 @@ let
   networkCfg = config.services.hyperhive.network;
   tlsCfg = config.services.hyperhive.tls;
 
+  # Static bridge address for the hive-ci container. The hive bridge has
+  # NO DHCP server: agent containers get deterministic static IPs
+  # (`lifecycle::agent_network_ip` hashes each name across .2..broadcast-1)
+  # and dnsmasq on the bridge is DNS-only. hive-ci is the one service
+  # container on the bridge, so it needs a static address too — `useDHCP`
+  # here only hangs the boot waiting for a lease nothing serves. Reserve the
+  # top host address of the (default /24) subnet; a clash with an agent that
+  # happens to hash here is the same rename-to-resolve case as any
+  # agent/agent IP collision. Operators on a non-/24 bridge (or with
+  # `bridgeIp` set to the top address) should pick a free host address.
+  ciBridgeOctets = lib.splitString "." networkCfg.bridgeIp;
+  ciBridgeIp = "${lib.elemAt ciBridgeOctets 0}.${lib.elemAt ciBridgeOctets 1}.${lib.elemAt ciBridgeOctets 2}.254";
+
   # Self-signed TLS is the gateway default (no operator cert / ACME). When
   # active, forgejo's ROOT_URL is `https://forge.<domain>` and the leaf is
   # signed by the host hive CA — so the runner's Node-based actions (e.g.
@@ -410,14 +423,21 @@ in
           # affect traffic destined for the bridge IP itself.
           networking.nameservers = [ networkCfg.bridgeIp ];
           # With privateNetwork=true + hostBridge the container's veth
-          # is bridge-attached. Enable DHCP on eth0 so the container gets
-          # an IP from dnsmasq on the bridge (hive-gateway serves the
-          # bridge subnet). Per-interface rather than global: nixos-containers
-          # sets networking.useDHCP = false for all containers (to silence
-          # the deprecated global DHCP warning), so setting useDHCP = true
-          # globally would conflict. eth0 is the inner veth name assigned
-          # by systemd-nspawn when hostBridge is used.
-          networking.interfaces.eth0.useDHCP = true;
+          # (eth0) is bridge-attached. There is no DHCP server on the hive
+          # bridge (dnsmasq is DNS-only; agents use static IPs), so assign a
+          # static address + default route via the bridge gateway rather than
+          # DHCP — `useDHCP` here just hangs boot on a lease that never
+          # arrives. See `ciBridgeIp` above.
+          networking.interfaces.eth0.ipv4.addresses = [
+            {
+              address = ciBridgeIp;
+              prefixLength = networkCfg.bridgePrefixLength;
+            }
+          ];
+          networking.defaultGateway = {
+            address = networkCfg.bridgeIp;
+            interface = "eth0";
+          };
 
           # nspawn containers can't create user-namespaces, so nix
           # sandboxing always fails. Fall back to unsandboxed builds.