Piik/docs/deployment.md

Minimal Deployment

Last verified against upstream documentation: 2026-08-19.

This section documents the currently deployable release contract: one Node.js process provides the built Web client, room API, and WebSocket signaling behind Caddy or nginx; coturn is the separate STUN/TURN service. Normal media remains browser-to-browser. Only an ICE pair that cannot connect directly consumes TURN bandwidth.

A deployment may additionally provide one single-node LiveKit process as the current controller's automatic final media fallback. This capacity is dormant unless the complete LIVEKIT_URL/key/secret tuple is configured. It does not replace the P2P path, the peer-assisted experiment, or the required application STUN/TURN settings.

The public staging infrastructure has already passed the relay checks recorded in docs/status.md, but every new deployment must run the direct, relay, and mixed-network procedures below. Configuration validation alone is not evidence that its DNS, firewall, NAT, or relay path works.

Topology and prerequisites

Use two DNS names, for example share.example.com and turn.example.com. They may resolve to the same public IP in the minimal deployment: the Web ingress owns TCP 443 while coturn owns UDP and TCP 3478. TURN/TLS is optional and uses its standard TCP port 5349 by default, which can also share that public IP.

browser -- HTTPS/WSS --> Caddy or nginx :443 --> Node.js :8787
browser <------------ DTLS-SRTP P2P ------------> browser
browser -- TURN only for failed ICE pairs --> coturn

The minimum runtime is Node.js 24 LTS and coturn 4.17.2 or a newer patched release. Provision a valid TLS certificate for the Web name and, only when enabling TURN/TLS, for the TURN name. Enable operating system time synchronization and keep the Node application port reachable only from its reverse proxy. A single Node process is intentional. Active participants and signaling remain in memory; an optional protected SQLite file can restore room identity after a restart, but it does not restore live WebRTC connections.

Build and validate the exact revision before starting it:

npm ci
npm run check
npm start

Run npm start under a service supervisor that injects the environment, restarts on failure, and applies bounded logs. For a simple untracked environment file, the equivalent direct launch is node --env-file=.env.production dist/server/server/index.js. Do not expose port 8787 to the Internet. The application default is 0.0.0.0 for LAN development and container compatibility, while the bare-metal production baseline below explicitly sets LISTEN_HOST=127.0.0.1 because only the same-host reverse proxy should connect. A container can instead use 0.0.0.0 and enforce the intended boundary with port publishing rules or a host firewall.

The optional same-host LiveKit layout needs no additional public hostname: LIVEKIT_URL=wss://share.example.com and nginx forwards only the exact /rtc endpoint and paths below /rtc/ to LiveKit. Its HTTP/WebSocket listener on TCP 7880 is private to nginx, while WebRTC media reaches LiveKit directly on TCP 7881 or UDP 7882.

Accepted flagship target after transport migration

The accepted target for share.bonfire.icu remains distributed and automatic: direct/peer UDP first, then an SFU virtual parent feeding normally one or two roots, whose peer descendants continue carrying media. TURN becomes an optional extreme-network transport for a controller-selected edge rather than a room-wide ICE default. Multiple exceptional viewers that cannot attach behind a healthy root may consume additional server egress only under a separate explicit cap.

This is not deployable from the current configuration contract. Release 769de201f7cc still refuses production startup without STUN, coturn credentials, and explicit TURN/UDP plus TURN/TCP URLs; it sends that ICE config to ordinary P2P peers. The tracked LiveKit example enables ICE/TCP 7881 and ICE/UDP mux 7882, omits embedded TURN, and pinned LiveKit 1.13.5 permits TCP fallback when the corresponding transport is configured. Keep the current environment, firewall, and services until a code PR changes validation/wire behavior and an exact-room canary passes.

The migration must make an entirely absent ordinary-peer coturn tuple a normal no-warning state, reject partial configuration, and add short-lived assignment/connection-generation grants only for an authorized exceptional ordinary edge. LiveKit participant transport is separate: the pinned integration issues/advertises embedded or external TURN only to LiveKit publishers/subscribers and does so participant-wide, not through Screener's per-edge grant. It remains a canary candidate and must be isolated/extended to the accepted exceptional-edge scope or rejected. The target default media path is UDP. Which optional TURN or media-TCP compatibility transport and port, if any, remains enabled is decided by the canary; HTTPS/WSS always remains TLS/TCP.

Do not collapse the public listener plan to 443/tcp plus one UDP port yet. LiveKit documents a single ICE/UDP mux port but its pinned sample recommends a multi-port mux range at least as wide as the CPU count for performance. Embedded TURN/UDP defaults to 3478 and recommends 443 only when HTTP/3/QUIC does not use that UDP port. The current nginx template has no HTTP/3 listener, but port choice, privileged binding, firewall, and one-port performance still require the bounded network/load gate in the low-server route research.

Production application environment

Keep the real values in the process secret store or an untracked, access-restricted environment file. Node.js 24 can load such a file with --env-file; the tracked .env.example is only a schema. Generate independent secrets with a password manager or a cross-platform Node command such as:

node -e "console.log(require('node:crypto').randomBytes(32).toString('base64url'))"

Use this production baseline:

NODE_ENV=production
LISTEN_HOST=127.0.0.1
PORT=8787
PUBLIC_BASE_URL=https://share.example.com
ALLOWED_ORIGINS=https://share.example.com
ACCESS_PASSWORD=<WHOLE_SITE_PASSWORD>
ROOM_DATABASE_PATH=/var/lib/screener/rooms.sqlite
ROOM_TTL_SECONDS=14400
MAX_ROOMS=1000
MAX_VIEWERS_PER_ROOM=8

STUN_URLS=stun:turn.example.com:3478
TURN_URLS=turn:turn.example.com:3478?transport=udp,turn:turn.example.com:3478?transport=tcp
TURN_SHARED_SECRET=<SAME_VALUE_AS_COTURN_STATIC_AUTH_SECRET>
TURN_CREDENTIAL_TTL_SECONDS=3600

To make automatic SFU fallback capacity available, add the complete tuple:

PEER_ASSISTED_MEDIA=true
PEER_ASSISTED_ROOM_IDS=1
LIVEKIT_URL=wss://share.example.com
LIVEKIT_API_KEY=<GENERATED_LIVEKIT_API_KEY>
LIVEKIT_API_SECRET=<INDEPENDENT_SECRET_OF_AT_LEAST_32_BYTES>
MAX_SFU_ROOTS_PER_ROOM=2

PEER_ASSISTED_ROOM_IDS is the deployment canary boundary. A non-empty value is a comma-separated set of exact positive numeric room IDs, using the same 1-to-12-digit syntax as room authentication. Duplicate IDs, empty entries, leading zeroes, and malformed IDs fail startup. A non-empty allowlist requires PEER_ASSISTED_MEDIA=true. Only listed rooms receive peer-assisted authentication, routing, room quality state, or optional LiveKit fallback; every other room keeps the legacy P2P wire and signaling behavior. Omitting the variable or leaving it empty deliberately preserves the earlier all-room behavior when peer assistance is enabled, so a canary deployment must set at least one exact ID. There is no browser control or percentage rollout.

ALLOWED_ORIGINS must list exact http or https origins, never *. ACCESS_PASSWORD is optional: omit it or leave it empty for a public site. A non-empty value must contain 1 through 128 visible ASCII characters (0x21 through 0x7e). For an Internet deployment intended to stay private, use an independent value that is not reused elsewhere. TURN_SHARED_SECRET must contain at least 32 bytes and must not reuse the access password. ROOM_DATABASE_PATH is optional but requires ACCESS_PASSWORD; startup fails if a database path is supplied without the whole-site gate. The baseline above enables persistent protected rooms. Omit ROOM_DATABASE_PATH to keep random temporary rooms, and omit both values for public mode. ROOM_TTL_SECONDS applies only to temporary rooms. MAX_VIEWERS_PER_ROOM defaults to 8 and accepts 1 through 16. It is an admission limit, not evidence that the publisher can sustain that many streams. TURN_SHARED_SECRET stays only in the Node environment and coturn configuration; browsers receive HMAC-SHA1-derived, time-limited credentials after room authentication. Keep host clocks synchronized because the credential username contains its Unix expiry time.

The three LIVEKIT_* values must either all be absent or all be present, and a complete tuple requires PEER_ASSISTED_MEDIA=true. An empty tuple keeps the optional SDK and server path dormant. LIVEKIT_URL must be a plain ws: or wss: origin with no /rtc suffix; production requires wss:. LIVEKIT_API_SECRET must contain at least 32 bytes and must not reuse ACCESS_PASSWORD or TURN_SHARED_SECRET. MAX_SFU_ROOTS_PER_ROOM defaults to 2 and accepts only 1 or 2; it is ignored when LiveKit is not configured. These credentials authorize short-lived LiveKit room tokens and do not provide E2EE: the LiveKit operator can access ordinary SFU media.

For the first canary, prefer an existing protected persistent room whose ID is stable across restarts. Set only that ID, restart the application, and verify that its authenticated message contains mediaMode: "peer-assisted" while a second non-allowlisted room contains none of mediaMode, qualitySettings, routeRevision, routeAssignment, or sfuStandbyUrl. Exercise join, offer and answer, stop, reconnect, and room deletion in both rooms. Roll back by setting PEER_ASSISTED_MEDIA=false and removing the allowlist and LiveKit tuple, then restart; no room database migration is involved.

When ACCESS_PASSWORD is configured, both host and viewer routes first show the same login gate. Only POST /api/session accepts the password in an Authorization: Bearer header, then returns a 12-hour stateless HMAC-SHA256 cookie with HttpOnly, SameSite=Strict, Path=/, a bounded Max-Age, and, under production HTTPS, Secure plus an __Host- name. The cookie contains no account or server-side session identifier. There is no account database, JWT, session map, or logout endpoint; expiry or clearing site cookies ends access. The optional room database is unrelated to access sessions.

POST /api/rooms and the /signal WebSocket upgrade use that cookie when the gate is enabled. The room API does not accept a direct Bearer credential as an alternate creation path. Viewers open /r/{code} or enter only the numeric code at /join; there is no viewer token or URL fragment, while the host token remains internal to the host page.

Room allocation has two deliberately small policies:

• With ROOM_DATABASE_PATH, SQLite allocates positive decimal IDs starting at 1. These protected rooms and links do not expire. An explicit stop or capture track ending stops the current publication and leaves viewers waiting; it does not delete the room. A brief signaling disconnect does not stop otherwise healthy P2P media.
• Without ROOM_DATABASE_PATH, rooms use random numeric IDs and expire according to ROOM_TTL_SECONDS. Stopping sharing does not immediately delete them. Public mode always uses this policy, so anonymous clients cannot simply walk a sequential namespace.

The built-in node:sqlite database stores only each room ID (the table rowid) and SHA-256 host-token digest. It must not contain plaintext tokens, passwords, access cookies, SDP, ICE candidates, IP addresses, viewer state, TURN credentials, or media. Protect and back up the file as service state. The tracked systemd unit creates /var/lib/screener with StateDirectory=screener and mode 0700; the production path above is writable despite ProtectSystem=strict. The database is designed for one application process, not shared storage across multiple instances.

For a simple consistent backup, stop screener.service before copying the SQLite file, then start it again. Restore while the service is stopped, restore ownership to the service account and mode 0600, run SQLite PRAGMA integrity_check, and only then start the service. An online backup must use SQLite's backup API or VACUUM INTO; do not copy only the live main file while it may have an active journal.

Enabling persistence does not migrate rooms that existed only in memory. The deployment restart invalidates those temporary links; the first subsequently created persistent room receives ID 1.

With no ACCESS_PASSWORD, the random room code is the sole viewing capability and does not provide strong privacy, so public mode is suitable only when public access is acceptable or another trusted access layer exists.

Production startup validates the configured ICE transport mix before the server listens. The baseline must contain STUN and separate TURN_URLS entries for turn: with explicit transport=udp and turn: with explicit transport=tcp. TURN/TLS is optional; when enabled, append turns:turn.example.com:5349?transport=tcp. This startup check only validates configuration shape; it does not prove that DNS, certificates, firewall rules, NAT mappings, or the coturn listeners work from an external network. Every STUN and TURN entry is also syntax-checked; use the exact lowercase TURN query forms shown above because one malformed ICE URL can make a browser reject the whole peer-connection configuration.

HTTPS and WSS ingress

Terminate public TLS at Caddy or nginx and proxy every application path to the single Node process. /signal must preserve the WebSocket upgrade, Host, and Origin, and its idle/read timeout must be longer than a normal sharing session. Do not cache /api/* responses. Caddy's minimal same-host form is:

share.example.com {
    reverse_proxy 127.0.0.1:8787
}

An equivalent nginx location is:

location / {
    proxy_pass http://127.0.0.1:8787;
    proxy_http_version 1.1;
    proxy_set_header Host $host;
    proxy_set_header Upgrade $http_upgrade;
    proxy_set_header Connection "upgrade";
    proxy_read_timeout 1h;
}

The tracked share.bonfire.icu.conf.example follows this simple model and listens directly on public TCP 443. It does not depend on the optional SNI router. Its optional exact /rtc and /rtc/ prefix locations preserve the LiveKit request URI and disable both access logging and request-line error logging. Legacy /rtc?token=... and current versioned requests carry a short-lived LiveKit JWT in the query string, so neither request target may fall through to the site's ordinary logs.

When Certbot manages the Web certificate, install the tracked reload-nginx.sh as an executable under /etc/letsencrypt/renewal-hooks/deploy/. Run certbot renew --dry-run after the first certificate is installed. This hook only reloads the Web ingress; if optional TURN/TLS later uses a renewed certificate, configure and verify a separate coturn reload or restart action.

Keep the proxy's access-log retention bounded and access controlled. Requests to /r/{code} put the room code in the path, so access logs can contain room codes as well as network metadata. They must not be treated as public artifacts; in public mode a current code is itself the only viewing capability.

For a process-level liveness probe, send GET /healthz. A running process returns HTTP 200 with {"status":"ok"} and Cache-Control: no-store; other methods return 405 with Allow: GET. The endpoint does not inspect room state, TURN reachability, or any external dependency, so use it only to decide whether the Node process can accept HTTP requests. It is not a deployment-readiness or end-to-end media check.

Optional single-node LiveKit fallback

Use the tracked livekit.yaml.example and livekit.service.example as the minimal single-node baseline. Install a reviewed, pinned LiveKit Server release from its official release assets, generate an independent key pair with livekit-server generate-keys, and place only the YAML key mapping in /etc/livekit/keys.yaml. Keep that file owned by the livekit service account with mode 0600; put the same values in Screener's untracked process secrets.

The example deliberately omits Redis and every recording, ingress, egress, webhook, and embedded-TURN service. Do not add Redis for this one-node workload. The service journal is the diagnostic log; keep its retention finite and access restricted. Never enable debug/Pion packet logging continuously or persist JWTs, SDP, ICE candidates, API secrets, or full /rtc and /rtc/* request targets.

Configure both the host firewall and the provider firewall independently:

Destination	Protocol	Exposure	Purpose
127.0.0.1:7880	TCP	Private	nginx to LiveKit signaling/API
public host :7881	TCP	Public	WebRTC ICE/TCP fallback
public host :7882	UDP	Public	WebRTC ICE/UDP mux

LiveKit may listen beyond loopback while it discovers and advertises public media addresses, so the public-firewall denial of TCP 7880 is mandatory; nginx must be its only ingress. TCP 7881 and UDP 7882 are direct media listeners and must not be placed behind the HTTP reverse proxy. If the host is behind NAT, forward them without port translation and verify the advertised candidate from an external network. LiveKit ICE/TCP is not equivalent to authenticated TURN; strict-network SFU reachability remains unproven until separately tested.

coturn

Copy deploy/coturn/turnserver.conf.example to an untracked service-owned location. Replace the FQDN and secret. The static-auth-secret value must exactly match the application's TURN_SHARED_SECRET; do not configure no-auth or permanent browser users. Certificate paths are needed only if the optional TURN/TLS block is enabled.

When the application and coturn configs share a parent directory, both service accounts must be able to traverse that directory. One minimal layout is a root:root directory with mode 0751, an application environment file owned by root:screener with mode 0640, and a coturn config owned by root:turnserver with mode 0640. Verify readability as the actual service user before starting coturn. The tracked systemd unit deliberately fails its pre-start check when the coturn config is missing or unreadable; coturn itself may otherwise warn and continue with unsafe defaults.

On a host with a public address directly on its interface, coturn can select the single listen/relay address automatically. On a multi-homed host, set listening-ip and relay-ip explicitly. Behind 1:1 NAT, also set external-ip=<PUBLIC_IP>/<LOCAL_RELAY_IP> and forward the complete relay range without port translation. A normal consumer NAT that changes relay port numbers is not a supported TURN-host topology.

The minimum deployment exposes TURN/UDP and TURN/TCP on 3478. turn: over TCP does not add TLS between the client and coturn, but relayed WebRTC media remains protected end to end by DTLS-SRTP. Optional TURN/TLS should listen directly on standard TCP 5349 and use turns:turn.example.com:5349?transport=tcp.

TCP 443 is an optional enhancement for networks that block uncommon ports. On the same public IP, coturn cannot bind it while the Web ingress already owns it. Use a second public IP or an explicitly validated layer-4/SNI router, such as the optional tls-sni-router.conf.example. That router requires moving the site's TLS virtual host to loopback TCP 8443; the default nginx site template intentionally does not do this. A normal Cloudflare HTTP proxy does not carry TURN even on port 443. Keep the TURN DNS record DNS-only, or use a compatible layer-4 service such as Cloudflare Spectrum; the Web hostname may remain HTTP-proxied independently.

Open only these public listeners:

Destination	Protocol	Purpose
share.example.com:443	TCP	HTTPS and WSS
turn.example.com:3478	UDP and TCP	STUN, TURN/UDP, and TURN/TCP
turn.example.com:49152-49251	UDP, bidirectional	Relayed media endpoints
turn.example.com:5349	TCP, optional	Standard TURN/TLS
turn.example.com:443	TCP, optional	TURN/TLS on a dedicated IP or validated L4/SNI route
host public address :7881	TCP, optional SFU	LiveKit WebRTC ICE/TCP
host public address :7882	UDP, optional SFU	LiveKit WebRTC ICE/UDP mux

The 100-port relay range is an initial small-room limit, not a universal sizing rule. Monitor 508/allocation failures and concurrent allocations before widening it. The example also sets user-quota=32, total-quota=100, and a per-allocation max-bps of 2,000,000 bytes/s. Tune these from measured 1080p60 traffic and the purchased egress capacity; never remove all quotas as a shortcut. The per-user value leaves allocation headroom for the host's independent connections and ICE restarts in an eight-viewer room; it is not a claim that the application or relay can sustain 1:8 media.

coturn 4.17.2 denies loopback peers unless allow-loopback-peers is enabled; the example deliberately does not enable it and additionally denies common private, carrier-grade NAT, link-local, and benchmark ranges. Keep no-multicast-peers and no-cli. If the TURN host must intentionally reach an internal peer, narrow an exception to that exact address instead of removing the deny list.

The example writes informational logs to stdout. Configure the supervisor to rotate by size and time with a finite retention, such as seven days initially, then reduce or extend it only for a documented operational need. coturn and the reverse proxy necessarily see client IP addresses; restrict log access, do not enable verbose or log-binding continuously, and never log credentials, SDP, ICE candidates, or the shared secret. When TURN/TLS is enabled, reload or restart coturn after certificate renewal. Confirm time synchronization remains healthy in every deployment.

Verification

Run these checks from real external networks before calling the deployment usable:

1. Open the official Trickle ICE sample and test each configured TURN URL separately with a short-lived credential. Each required UDP and TCP test must produce a relay candidate; test TLS too when it is configured. A successful STUN-only srflx candidate does not prove TURN works.
2. Create a normal Screener room on two different networks. Confirm media flows and the selected-pair stats report a non-relay path when direct ICE succeeds.
3. Open the host at https://share.example.com/?relay=1, create a new room, and connect a viewer. This diagnostic flag sets iceTransportPolicy=relay; media must still flow and the stats panel must report relay.
4. Block UDP on a test client while leaving TURN/TCP reachable. Confirm a new session succeeds through turn:turn.example.com:3478?transport=tcp.
5. If TURN/TLS is enabled, test its turns: URL separately on 5349 or the explicitly configured 443 route. Failure of an optional endpoint must be diagnosed, but an intentionally omitted endpoint is not a deployment failure.
6. Repeat at 1, 3, 5, and 8 viewers with a mixed direct/restrictive-network cohort. Record the selected path, RTT, bitrate, frame rate, packet loss, publisher upload/encode load, and coturn egress for each viewer. Verify a direct viewer does not start consuming relay bandwidth merely because another viewer needs TURN. A real 1:8 session is currently unverified and must not be inferred from the configured admission limit.

References: coturn 4.17.2 release, turnserver documentation, the official example configuration, TURN URI scheme RFC 7065, Cloudflare network-port guidance, RFC6265bis, Node.js Crypto, LiveKit firewall guidance, and the official LiveKit configuration sample.