nix/hive-gateway: switch to JSON port table per mara on #740 (#15 v0, option B)

mara on PR #740 comment 9295: "we decided to go with the json" (issue #15 comment 9270:
"nginx container lives in system config, so it cannot be just rebuilt
from meta flake. go for the json file the c0re writes").

Drops:
- `cfg.agents` listOf str option
- Replicated FNV-1a hash + char-code table + manager-port special case
- Drift-hazard comment (no more rust↔nix constant sync)

Adds:
- `cfg.agentPortsFile = "/var/lib/hyperhive/agent-ports.json"` (default,
  nullable to disable) — path to a JSON map of `{ "<name>": <port> }`
  written by hive-c0re on every topology change.
- `agentPortsTable` reads the file at eval time via
  `builtins.fromJSON (builtins.readFile path)`, guarded by
  `builtins.pathExists` so a missing file gracefully defaults to `{}`.
- Per-agent locations generated via `lib.mapAttrs'` over the table —
  one location block per entry; empty table → empty attrset → no
  per-agent blocks, pre-#15 shape.

Rust-side dependency: hive-c0re needs to emit the JSON file on every
topology change. Coordinating with damocles via a separate ping — the
nix side ships now with safe defaults (missing file = no routes, no
behavior change vs main).

Verified:
- nix eval with `/tmp/test-agent-ports.json` → 4 per-agent blocks at
  correct ports (8178 iris, 8267 argus, 8304 atlas, 8549 damocles)
- nix eval with nonexistent file → only `/` location (graceful default)
- full container toplevel builds clean with matrix on

Empty file case mirrors the previous empty-list default — purely
additive, old `<host>:<port>/` direct reach untouched, no per-agent
blocks until c0re writes the JSON. Operator can also `null` the
option to disable entirely.
This commit is contained in:
atlas 2026-05-31 12:47:11 +02:00 committed by Mara
commit 09c1abe5b4

View file

@ -9,66 +9,30 @@ let
hyperhiveDomain = config.services.hyperhive.domain; hyperhiveDomain = config.services.hyperhive.domain;
matrixCfg = config.services.hyperhive.matrix; matrixCfg = config.services.hyperhive.matrix;
# Per-agent web port, mirroring `hive-c0re::lifecycle::agent_web_port` # Per-agent port table for `/agent/<name>/` routing (#15 v0). Single-
# exactly so the gateway and the harness agree on what port to talk # sourced from `cfg.agentPortsFile` (default
# to without a runtime contract. The manager is fixed at # `/var/lib/hyperhive/agent-ports.json`), written by hive-c0re on every
# `MANAGER_PORT`; every sub-agent is `WEB_PORT_BASE + FNV-1a(name) % # topology change in shape `{ "<name>": <port>, ... }`.
# WEB_PORT_RANGE`. Pure + reproducible from just the name, matches
# the rust constants line-for-line (lifecycle.rs:60-79).
# #
# **Drift hazard**: if the rust constants change (MANAGER_PORT, # Read at deploy time via `builtins.fromJSON (builtins.readFile ...)`
# WEB_PORT_BASE, WEB_PORT_RANGE, MANAGER_NAME, or the FNV-1a # — pure eval (the file lives outside the nix store; nix copies the
# parameters), this nix copy must change in lockstep — otherwise # content into the store as a fixed-output dep). When the file is
# `/agent/<name>/` requests get proxied to the wrong port. There's # missing (fresh install before c0re has had a chance to write it),
# no automated cross-check today (#15 v0 follow-up: have c0re emit # default to an empty map → no per-agent routes generated → gateway
# the port mapping as `/var/lib/hyperhive/agent-ports.json` and # falls back to its pre-#15 shape. The container rebuilds on every
# have this module read it via `lib.importJSON`, single-sourcing # `hivectl gateway-sync` (operator-initiated) or on the next
# the table). # `nixos-rebuild switch`, picking up whatever c0re has written
agentWebPortLib = # since the last build.
let #
# ASCII char → byte code lookup, limited to chars valid in agent # mara on #740 (comment 9295) + #15 (comment 9270): the gateway
# names (lowercase alpha, digits, dash, underscore). Names with # nginx container lives in system config (not meta), so it can't
# other chars are an eval-time error rather than a silent wrong # auto-rebuild from meta-flake events — the JSON file is what
# hash. Add entries here if hyperhive ever loosens the naming # bridges the host's nix eval to the agent-lifecycle data c0re owns.
# constraint. agentPortsTable =
charCode = { if cfg.agentPortsFile == null || !builtins.pathExists cfg.agentPortsFile then
"0" = 48; "1" = 49; "2" = 50; "3" = 51; "4" = 52; "5" = 53; { }
"6" = 54; "7" = 55; "8" = 56; "9" = 57; else
"-" = 45; "_" = 95; builtins.fromJSON (builtins.readFile cfg.agentPortsFile);
"a" = 97; "b" = 98; "c" = 99; "d" = 100; "e" = 101;
"f" = 102; "g" = 103; "h" = 104; "i" = 105; "j" = 106;
"k" = 107; "l" = 108; "m" = 109; "n" = 110; "o" = 111;
"p" = 112; "q" = 113; "r" = 114; "s" = 115; "t" = 116;
"u" = 117; "v" = 118; "w" = 119; "x" = 120; "y" = 121;
"z" = 122;
};
webPortBase = 8100;
webPortRange = 900;
managerPort = 8000;
managerName = "manager";
# nix uses signed 64-bit ints; mask each step to u32 to mirror
# rust's `u32::wrapping_mul`. FNV-1a constants identical to the
# rust version (offset basis 2_166_136_261, prime 16_777_619).
fnv1aU32 =
name:
let
chars = lib.stringToCharacters name;
mask32 = h: lib.bitAnd h 4294967295;
step =
acc: c:
mask32 (
(lib.bitXor acc (
charCode.${c}
or (throw "agent name '${name}' contains char '${c}' outside the [a-z0-9_-] alphabet hyperhive.user.name constraint mismatch?")
))
* 16777619
);
in
lib.foldl' step 2166136261 chars;
in
name: if name == managerName then managerPort else webPortBase + lib.mod (fnv1aU32 name) webPortRange;
in in
{ {
# Single nginx in front of every hyperhive surface (#609 / #15 v0). # Single nginx in front of every hyperhive surface (#609 / #15 v0).
@ -179,46 +143,39 @@ in
''; '';
}; };
agents = lib.mkOption { agentPortsFile = lib.mkOption {
type = lib.types.listOf lib.types.str; type = lib.types.nullOr lib.types.path;
default = [ ]; default = "/var/lib/hyperhive/agent-ports.json";
example = [ example = "/var/lib/hyperhive/agent-ports.json";
"iris"
"atlas"
"argus"
"damocles"
];
description = '' description = ''
Sub-agent names to expose at `http(s)://<gateway>/agent/<name>/` Path to a JSON file mapping sub-agent names to their web ports
through the gateway (#15 v0). For each name in the list, the for `/agent/<name>/` routing through the gateway (#15 v0).
gateway adds a `location /agent/<name>/` block that Shape: `{ "<name>": <port>, ... }`. Written by hive-c0re on
`proxy_pass`es to `http://127.0.0.1:<port>/`, where `<port>` is every topology change (the rust side knows the canonical port
derived from the same `agent_web_port(name)` hash hive-c0re allocation via `lifecycle::agent_web_port`; the gateway just
uses internally (`lifecycle.rs` constants + FNV-1a, replicated reads what it's told).
in this module's `let` block).
**Purely additive** the old `http://<host>:<port>/` direct For each `<name>: <port>` entry, the gateway adds a
reach still works in parallel; this just gives the operator a `location /agent/<name>/` block that `proxy_pass`es to
single-origin route. Manager isn't included (no per-agent `http://127.0.0.1:<port>/`. Empty / missing file no
prefix needed; manager already gets the `/` route via the per-agent routes generated gateway falls back to its pre-#15
c0re upstream block). shape (just `/` + matrix surfaces).
Empty list (default) leaves the gateway in its pre-#15 shape: **Purely additive**: the old `http://<host>:<port>/` direct
no per-agent routes, only `/` (c0re) + `/matrix/` (fluffychat reach keeps working in parallel; this just gives the operator
when matrix is on) + `.well-known/matrix/*` (matrix a single-origin route. Manager isn't included in the map (no
autodiscovery). per-agent prefix needed; manager already gets the `/` route
via the c0re upstream block).
Maintenance: this list is currently operator-managed (the Set to `null` to disable per-agent routing entirely without
gateway runs at the host level, the agent list lives in the creating the file. Set to a custom path if the operator's c0re
meta-flake at runtime, and the host's nix eval doesn't see writes the table elsewhere.
the meta-flake's agent dirs). Follow-up tracked at #15:
either auto-derive from **Rebuild trigger**: the gateway container picks up new entries
`/var/lib/hyperhive/meta/topology.json` via `lib.importJSON` on the next `nixos-rebuild switch` (or `hivectl gateway-sync`
in the gateway module (gives a single source of truth at the if that helper lands). c0re writes are not auto-applied to a
cost of an eval-time impurity), or have c0re write an nginx running gateway see the follow-up in #15 for runtime nginx
snippet that the container `include`s + reloads on topology include + reload + eventual per-agent unix sockets.
change (decouples from rebuilds entirely; also handles the
eventual move to per-agent unix sockets).
''; '';
}; };
}; };
@ -354,11 +311,10 @@ in
} }
// //
# Per-agent UIs (#15 v0). One `/agent/<name>/` # Per-agent UIs (#15 v0). One `/agent/<name>/`
# block per name in `services.hyperhive.gateway.agents`, # block per `<name>: <port>` entry in
# proxying to the agent's harness web server on # `agentPortsTable` (loaded from `cfg.agentPortsFile`
# `127.0.0.1:<port>` where `<port>` comes from the # — `/var/lib/hyperhive/agent-ports.json` by default,
# `agentWebPortLib` FNV-1a hash (matches # written by hive-c0re on every topology change).
# `lifecycle::agent_web_port` line-for-line).
# #
# Trailing-slash pair (`/agent/<name>/` + `proxy_pass # Trailing-slash pair (`/agent/<name>/` + `proxy_pass
# http://...:<port>/`) strips the `/agent/<name>` # http://...:<port>/`) strips the `/agent/<name>`
@ -378,23 +334,21 @@ in
# SSE / websocket support via `proxyWebsockets = true` # SSE / websocket support via `proxyWebsockets = true`
# (same as the c0re `/` block below). # (same as the c0re `/` block below).
# #
# Empty `cfg.agents` list → empty attrset → no # Empty / missing `cfg.agentPortsFile` → empty table
# per-agent blocks; old `<host>:<port>/` direct reach # → no per-agent blocks; old `<host>:<port>/` direct
# still works. # reach still works.
builtins.listToAttrs ( lib.mapAttrs' (name: port: {
builtins.map (name: { name = "/agent/${name}/";
name = "/agent/${name}/"; value = {
value = { proxyPass = "http://127.0.0.1:${toString port}/";
proxyPass = "http://127.0.0.1:${toString (agentWebPortLib name)}/"; proxyWebsockets = true;
proxyWebsockets = true; extraConfig = ''
extraConfig = '' proxy_set_header X-Forwarded-Prefix /agent/${name};
proxy_set_header X-Forwarded-Prefix /agent/${name}; proxy_buffering off;
proxy_buffering off; proxy_read_timeout 1d;
proxy_read_timeout 1d; '';
''; };
}; }) agentPortsTable
}) cfg.agents
)
// { // {
# Everything else proxies to hive-c0re. Upgrade # Everything else proxies to hive-c0re. Upgrade
# headers stay set so SSE (`/dashboard/stream`, # headers stay set so SSE (`/dashboard/stream`,