docs: follow the swarm service names to the swarm domain

The three moves above falsify prose in five files, and none of it is
caught by a gate: clippy, cargo test and nix flake check read exactly
zero markdown.

Corrected where a doc stated a DEFAULT or enumerated the vhosts --
gateway (ACME requirements + the header table + the two Forgejo
reachability lines), matrix (gatewayHost, firewall rationale, agent
reachability), network (container table + the HTTP resolution bullet),
turn-loop config, dashboard.

Two mentions are deliberately left alone. `docs/tools/hivectl.md` says
the URL is read from the daemon "instead of assuming forge.<domain>" --
that sentence is about not assuming a shape and stays true. And
matrix.md's breaking-change note describes what `serverName` USED to
default to; it is history, correctly labelled, and rewriting it would
erase the record of an earlier migration.

The ACME line gained a requirement rather than a rename: with the
service names under the swarm domain, every one of those names has to
resolve to this host, not just the hive's own.
This commit is contained in:
atlas 2026-08-05 23:30:53 +02:00 committed by mara
commit a1243fe04a
5 changed files with 24 additions and 15 deletions

View file

@ -46,7 +46,7 @@ untouched by any of it.
| container | netns | IPv4 | listens / reached via |
| -------------- | ----------------------- | -------------------- | -------------------------------------------------------------------------------------------- |
| `hive-gateway` | host (shared) | host addresses | nginx `:80`/`:443` (every vhost); dnsmasq `bridgeIp:53` + DHCP `:67` on the bridge |
| `hive-forge` | host (shared) | host addresses | forgejo `:3000` http, `:2222` git-ssh; fronted by the `forge.<domain>` vhost |
| `hive-forge` | host (shared) | host addresses | forgejo `:3000` http, `:2222` git-ssh; fronted by the `forge.<swarm-domain>` vhost |
| `hive-matrix` | host (shared) | host addresses | tuwunel `:8008` (+ optional federation port); fronted by the matrix vhost |
| `hive-ci` | private, veth on bridge | DHCP pool | outbound only (runner → forge); no inbound surface |
| `h-<agent>` | private, veth on bridge | DHCP pool | web UI via UDS `/run/hive-agent/<name>` → nginx sub-path; in-container UI port hashed 81008999 |
@ -58,8 +58,9 @@ The flows, end to end:
- **DNS** — agents query `bridgeIp:53`; hive zones are answered
authoritatively with the bridge IP, everything else forwards to the
host's resolvers (see *Resolver behaviour* below).
- **HTTP**`forge.<domain>` / matrix / dashboard names all resolve
to the bridge IP, land on nginx `:80`/`:443`, and proxy to forgejo
- **HTTP**`forge.` and `chat.` (under `swarm.domain`) plus the hive's
own dashboard name resolve to the bridge IP, land on nginx
`:80`/`:443`, and proxy to forgejo
`:3000`, tuwunel `:8008`, hive-c0re `127.0.0.1:7000`, or a per-agent
UI unix socket.
- **Internet egress** — agent default route points at the bridge IP;