Harden room discovery and health metrics

This commit is contained in:
Koala
2026-06-03 10:48:49 +02:00
parent bf0fb5741f
commit 81c50eff16
16 changed files with 379 additions and 46 deletions
+3 -1
View File
@@ -1,4 +1,6 @@
PORT=3000
MIN_VERSION=1.0.0
MAX_ROOMS=1000
MAX_PEERS_PER_ROOM=50
MAX_PEERS_PER_ROOM=25
# Optional: enables aggregate-only admin metrics on /health with Authorization: Bearer <token>
ADMIN_METRICS_TOKEN=
+9 -1
View File
@@ -14,10 +14,17 @@ Copy `.env.example` to `.env` and configure your settings.
```bash
PORT=3000
MAX_ROOMS=1000
MAX_PEERS_PER_ROOM=50
MAX_PEERS_PER_ROOM=25
MIN_VERSION=1.0.0
# Optional: enables aggregate-only admin metrics on /health with Authorization: Bearer <token>
ADMIN_METRICS_TOKEN=
```
### Health & Metrics
`GET /` and `GET /health` are IP-rate-limited. By default `/health` returns only basic service status, uptime, room count, connection count, and a timestamp.
If `ADMIN_METRICS_TOKEN` is set, requests with `Authorization: Bearer <token>` receive additional aggregate metrics such as total peers, average peers per room, max room size, active lobby count, rate-limit map sizes, and process memory usage. The metrics response does not include room IDs, peer IDs, usernames, IP addresses, media titles, or other user-level data.
### Docker (Recommended)
The server is available as a pre-built image on GHCR.
```bash
@@ -38,6 +45,7 @@ npm start
## Security
- **Rate Limiting**: IP-based connection limits and socket-based event limits.
- **Room Discovery Throttle**: Room-list refreshes are rate-limited server-side to one request every 10 seconds per socket.
- **Token Handshake**: Requires a valid token defined in the root `shared/constants.js`.
- **Single Source of Truth**: The server imports constants directly from the root `shared/` directory.
- **In-Memory**: Rooms are automatically pruned after 2 hours of inactivity.
+28 -7
View File
@@ -4,6 +4,7 @@ import { Server } from 'socket.io';
import crypto from 'crypto';
import dotenv from 'dotenv';
import { EVENTS, OFFICIAL_SERVER_TOKEN, PROTOCOL_VERSION } from '../shared/constants.js';
import { buildHealthPayload, checkCooldown, isAdminMetricsAuthorized } from './ops.js';
dotenv.config();
@@ -15,8 +16,10 @@ function hashPassword(password) {
const PORT = process.env.PORT || 3000;
const MAX_ROOMS = parseInt(process.env.MAX_ROOMS) || 1000;
const MAX_PEERS_PER_ROOM = parseInt(process.env.MAX_PEERS_PER_ROOM) || 50;
const MAX_PEERS_PER_ROOM = parseInt(process.env.MAX_PEERS_PER_ROOM) || 25;
const MIN_VERSION = process.env.MIN_VERSION || '1.0.0';
const ADMIN_METRICS_TOKEN = process.env.ADMIN_METRICS_TOKEN || '';
const ROOM_LIST_COOLDOWN_MS = 10000;
const app = express();
app.set('trust proxy', 1); // For real client IP through reverse proxy
@@ -35,13 +38,20 @@ app.get('/health', (req, res) => {
if (!checkHealthRate(clientIp)) {
return res.status(429).json({ error: 'Rate limited' });
}
res.json({
status: 'ok',
uptime: process.uptime(),
rooms: rooms.size,
const includeMetrics = isAdminMetricsAuthorized(req.get('authorization'), ADMIN_METRICS_TOKEN);
res.json(buildHealthPayload({
rooms,
connections: io.engine?.clientsCount ?? 0,
timestamp: Date.now()
});
includeMetrics,
uptime: process.uptime(),
rateLimitSizes: {
connections: connectionCounts.size,
events: eventCounts.size,
health: healthCounts.size,
authFailures: failedAuthAttempts.size,
roomList: roomListCooldowns.size
}
}));
});
const httpServer = createServer(app);
@@ -139,6 +149,7 @@ setInterval(() => {
const eventCounts = new Map(); // socketId -> { count, resetTime }
const healthCounts = new Map(); // ip -> { count, resetTime }
const roomListCooldowns = new Map(); // socketId -> last allowed timestamp
// Clean up connection counts and event counts to prevent memory leak
setInterval(() => {
@@ -158,6 +169,11 @@ setInterval(() => {
healthCounts.delete(ip);
}
}
for (const [socketId] of roomListCooldowns.entries()) {
if (!io.sockets.sockets.has(socketId)) {
roomListCooldowns.delete(socketId);
}
}
}, 60000);
function checkConnectionRate(ip) {
@@ -541,6 +557,10 @@ io.on('connection', (socket) => {
socket.disconnect(true);
return;
}
if (!checkCooldown(roomListCooldowns, socket.id, ROOM_LIST_COOLDOWN_MS)) {
socket.emit(EVENTS.ERROR, { message: 'Room list refresh is rate limited. Try again in a few seconds.' });
return;
}
const list = Array.from(rooms.entries()).map(([id, r]) => ({
id,
peerCount: r.peers.size,
@@ -584,6 +604,7 @@ io.on('connection', (socket) => {
socket.on('disconnect', () => {
eventCounts.delete(socket.id);
roomListCooldowns.delete(socket.id);
const mapping = socketToRoom.get(socket.id);
if (mapping) {
// Socket is already disconnected — no need to call socket.leave().
+76
View File
@@ -0,0 +1,76 @@
import crypto from 'crypto';
export function checkCooldown(cooldowns, key, windowMs, now = Date.now()) {
const lastAllowedAt = cooldowns.get(key) || 0;
if (now - lastAllowedAt < windowMs) {
return false;
}
cooldowns.set(key, now);
return true;
}
export function isAdminMetricsAuthorized(authHeader, adminToken) {
if (!adminToken || typeof adminToken !== 'string') return false;
if (!authHeader || typeof authHeader !== 'string') return false;
const prefix = 'Bearer ';
if (!authHeader.startsWith(prefix)) return false;
const provided = authHeader.slice(prefix.length);
if (!provided) return false;
const expectedBuffer = Buffer.from(adminToken);
const providedBuffer = Buffer.from(provided);
if (expectedBuffer.length !== providedBuffer.length) return false;
return crypto.timingSafeEqual(expectedBuffer, providedBuffer);
}
export function buildHealthPayload({
rooms,
connections,
includeMetrics = false,
now = Date.now(),
uptime = 0,
memoryUsage = () => process.memoryUsage(),
rateLimitSizes = {}
}) {
const payload = {
status: 'ok',
uptime,
rooms: rooms.size,
connections,
timestamp: now
};
if (!includeMetrics) return payload;
const roomValues = Array.from(rooms.values());
const roomSizes = roomValues.map(room => room.peers?.size || 0);
const peers = roomSizes.reduce((sum, size) => sum + size, 0);
const maxPeersInRoom = roomSizes.length > 0 ? Math.max(...roomSizes) : 0;
const avgPeersPerRoom = roomSizes.length > 0
? Math.round((peers / roomSizes.length) * 100) / 100
: 0;
const mem = memoryUsage();
return {
...payload,
peers,
roomsWithLobby: roomValues.filter(room => !!room.activeLobby).length,
avgPeersPerRoom,
maxPeersInRoom,
rateLimitEntries: {
connections: rateLimitSizes.connections || 0,
events: rateLimitSizes.events || 0,
health: rateLimitSizes.health || 0,
authFailures: rateLimitSizes.authFailures || 0,
roomList: rateLimitSizes.roomList || 0
},
memory: {
rss: mem.rss,
heapUsed: mem.heapUsed,
heapTotal: mem.heapTotal
}
};
}
-10
View File
@@ -9,7 +9,6 @@
"version": "1.0.0",
"license": "ISC",
"dependencies": {
"bcryptjs": "^3.0.3",
"dotenv": "^17.4.2",
"express": "^5.2.1",
"socket.io": "^4.8.3"
@@ -70,15 +69,6 @@
"node": "^4.5.0 || >= 5.9"
}
},
"node_modules/bcryptjs": {
"version": "3.0.3",
"resolved": "https://registry.npmjs.org/bcryptjs/-/bcryptjs-3.0.3.tgz",
"integrity": "sha512-GlF5wPWnSa/X5LKM1o0wz0suXIINz1iHRLvTS+sLyi7XPbe5ycmYI3DlZqVGZZtDgl4DmasFg7gOB3JYbphV5g==",
"license": "BSD-3-Clause",
"bin": {
"bcrypt": "bin/bcrypt"
}
},
"node_modules/body-parser": {
"version": "2.2.2",
"resolved": "https://registry.npmjs.org/body-parser/-/body-parser-2.2.2.tgz",
-1
View File
@@ -11,7 +11,6 @@
"license": "ISC",
"type": "module",
"dependencies": {
"bcryptjs": "^3.0.3",
"dotenv": "^17.4.2",
"express": "^5.2.1",
"socket.io": "^4.8.3"