stale server sockets, passive non-initiators, and race conditions during peer connection setup. Fix users unable to see or hear each other in voice channels due to stale server sockets, passive non-initiators, and race conditions during peer connection setup. Server: - Close stale WebSocket connections sharing the same oderId in handleIdentify instead of letting them linger up to 45s - Make user_joined/user_left broadcasts identity-aware so duplicate sockets don't produce phantom join/leave events - Include serverIds in user_left payload for multi-room presence - Simplify findUserByOderId now that stale sockets are cleaned up Client - signaling: - Add fallback offer system with 1s timer for missed user_joined races - Add non-initiator takeover after 5s when the initiator fails to send an offer (NON_INITIATOR_GIVE_UP_MS) - Scope peerServerMap per signaling URL to prevent cross-server collisions - Add socket identity guards on all signaling event handlers - Replace canReusePeerConnection with hasActivePeerConnection and isPeerConnectionNegotiating with extended grace periods Client - peer connections: - Extract replaceUnusablePeer helper to deduplicate stale peer replacement in offer and ICE handlers - Add stale connectionstatechange guard to ignore events from replaced RTCPeerConnection instances - Use deterministic initiator election in peer recovery reconnects - Track createdAt on PeerData for staleness detection Client - presence: - Add multi-room presence tracking via presenceServerIds on User - Replace clearUsers + individual userJoined with syncServerPresence for atomic server roster updates - Make userLeft handle partial server removal instead of full eviction Documentation: - Add server-side connection hygiene, non-initiator takeover, and stale peer replacement sections to the realtime README
290 lines
8.2 KiB
TypeScript
290 lines
8.2 KiB
TypeScript
import {
|
|
CONNECTION_STATE_CONNECTED,
|
|
DATA_CHANNEL_STATE_OPEN,
|
|
P2P_TYPE_VOICE_STATE_REQUEST,
|
|
PEER_DISCONNECT_GRACE_MS,
|
|
PEER_RECONNECT_INTERVAL_MS,
|
|
PEER_RECONNECT_MAX_ATTEMPTS
|
|
} from '../../realtime.constants';
|
|
import {
|
|
PeerConnectionManagerContext,
|
|
PeerConnectionManagerState,
|
|
RecoveryHandlers,
|
|
RemovePeerOptions
|
|
} from '../shared';
|
|
import { clearAllPingTimers, stopPingInterval } from '../messaging/ping';
|
|
|
|
/**
|
|
* Close and remove a peer connection, data channel, and emit a disconnect event.
|
|
*/
|
|
export function removePeer(
|
|
context: PeerConnectionManagerContext,
|
|
peerId: string,
|
|
options?: RemovePeerOptions
|
|
): void {
|
|
const { state } = context;
|
|
const peerData = state.activePeerConnections.get(peerId);
|
|
const preserveReconnectState = options?.preserveReconnectState === true;
|
|
|
|
clearPeerDisconnectGraceTimer(state, peerId);
|
|
|
|
if (!preserveReconnectState) {
|
|
clearPeerReconnectTimer(state, peerId);
|
|
state.disconnectedPeerTracker.delete(peerId);
|
|
}
|
|
|
|
state.remotePeerStreams.delete(peerId);
|
|
state.remotePeerVoiceStreams.delete(peerId);
|
|
state.remotePeerScreenShareStreams.delete(peerId);
|
|
|
|
if (peerData) {
|
|
if (peerData.dataChannel)
|
|
peerData.dataChannel.close();
|
|
|
|
peerData.connection.close();
|
|
state.activePeerConnections.delete(peerId);
|
|
state.peerNegotiationQueue.delete(peerId);
|
|
removeFromConnectedPeers(state, peerId);
|
|
stopPingInterval(state, peerId);
|
|
state.peerLatencies.delete(peerId);
|
|
state.pendingPings.delete(peerId);
|
|
state.peerDisconnected$.next(peerId);
|
|
}
|
|
}
|
|
|
|
/** Close every active peer connection and clear internal state. */
|
|
export function closeAllPeers(state: PeerConnectionManagerState): void {
|
|
clearAllPeerReconnectTimers(state);
|
|
clearAllPeerDisconnectGraceTimers(state);
|
|
clearAllPingTimers(state);
|
|
|
|
state.activePeerConnections.forEach((peerData) => {
|
|
if (peerData.dataChannel)
|
|
peerData.dataChannel.close();
|
|
|
|
peerData.connection.close();
|
|
});
|
|
|
|
state.activePeerConnections.clear();
|
|
state.remotePeerStreams.clear();
|
|
state.remotePeerVoiceStreams.clear();
|
|
state.remotePeerScreenShareStreams.clear();
|
|
state.peerNegotiationQueue.clear();
|
|
state.peerLatencies.clear();
|
|
state.pendingPings.clear();
|
|
state.connectedPeersChanged$.next([]);
|
|
}
|
|
|
|
export function trackDisconnectedPeer(state: PeerConnectionManagerState, peerId: string): void {
|
|
state.disconnectedPeerTracker.set(peerId, {
|
|
lastSeenTimestamp: Date.now(),
|
|
reconnectAttempts: 0
|
|
});
|
|
}
|
|
|
|
export function clearPeerReconnectTimer(
|
|
state: PeerConnectionManagerState,
|
|
peerId: string
|
|
): void {
|
|
const timer = state.peerReconnectTimers.get(peerId);
|
|
|
|
if (timer) {
|
|
clearInterval(timer);
|
|
state.peerReconnectTimers.delete(peerId);
|
|
}
|
|
}
|
|
|
|
export function clearPeerDisconnectGraceTimer(
|
|
state: PeerConnectionManagerState,
|
|
peerId: string
|
|
): void {
|
|
const timer = state.peerDisconnectGraceTimers.get(peerId);
|
|
|
|
if (timer) {
|
|
clearTimeout(timer);
|
|
state.peerDisconnectGraceTimers.delete(peerId);
|
|
}
|
|
}
|
|
|
|
/** Cancel all pending peer reconnect timers and clear the tracker. */
|
|
export function clearAllPeerReconnectTimers(state: PeerConnectionManagerState): void {
|
|
state.peerReconnectTimers.forEach((timer) => clearInterval(timer));
|
|
state.peerReconnectTimers.clear();
|
|
state.disconnectedPeerTracker.clear();
|
|
}
|
|
|
|
export function clearAllPeerDisconnectGraceTimers(state: PeerConnectionManagerState): void {
|
|
state.peerDisconnectGraceTimers.forEach((timer) => clearTimeout(timer));
|
|
state.peerDisconnectGraceTimers.clear();
|
|
}
|
|
|
|
export function schedulePeerDisconnectRecovery(
|
|
context: PeerConnectionManagerContext,
|
|
peerId: string,
|
|
handlers: RecoveryHandlers
|
|
): void {
|
|
const { logger, state } = context;
|
|
|
|
if (state.peerDisconnectGraceTimers.has(peerId))
|
|
return;
|
|
|
|
logger.warn('Peer temporarily disconnected; waiting before reconnect', { peerId });
|
|
|
|
const timer = setTimeout(() => {
|
|
state.peerDisconnectGraceTimers.delete(peerId);
|
|
|
|
const peerData = state.activePeerConnections.get(peerId);
|
|
|
|
if (!peerData)
|
|
return;
|
|
|
|
const connectionState = peerData.connection.connectionState;
|
|
|
|
if (connectionState === CONNECTION_STATE_CONNECTED || connectionState === 'connecting') {
|
|
logger.info('Peer recovered before disconnect grace expired', {
|
|
peerId,
|
|
state: connectionState
|
|
});
|
|
|
|
return;
|
|
}
|
|
|
|
logger.warn('Peer still disconnected after grace period; recreating connection', {
|
|
peerId,
|
|
state: connectionState
|
|
});
|
|
|
|
trackDisconnectedPeer(state, peerId);
|
|
handlers.removePeer(peerId, { preserveReconnectState: true });
|
|
schedulePeerReconnect(context, peerId, handlers);
|
|
}, PEER_DISCONNECT_GRACE_MS);
|
|
|
|
state.peerDisconnectGraceTimers.set(peerId, timer);
|
|
}
|
|
|
|
export function schedulePeerReconnect(
|
|
context: PeerConnectionManagerContext,
|
|
peerId: string,
|
|
handlers: RecoveryHandlers
|
|
): void {
|
|
const { callbacks, logger, state } = context;
|
|
|
|
if (state.peerReconnectTimers.has(peerId))
|
|
return;
|
|
|
|
logger.info('Scheduling P2P reconnect', { peerId });
|
|
|
|
const timer = setInterval(() => {
|
|
const info = state.disconnectedPeerTracker.get(peerId);
|
|
|
|
if (!info) {
|
|
clearPeerReconnectTimer(state, peerId);
|
|
return;
|
|
}
|
|
|
|
info.reconnectAttempts++;
|
|
logger.info('P2P reconnect attempt', {
|
|
peerId,
|
|
attempt: info.reconnectAttempts
|
|
});
|
|
|
|
if (info.reconnectAttempts >= PEER_RECONNECT_MAX_ATTEMPTS) {
|
|
logger.info('P2P reconnect max attempts reached', { peerId });
|
|
clearPeerReconnectTimer(state, peerId);
|
|
state.disconnectedPeerTracker.delete(peerId);
|
|
return;
|
|
}
|
|
|
|
if (!callbacks.isSignalingConnected()) {
|
|
logger.info('Skipping P2P reconnect - no signaling connection', { peerId });
|
|
return;
|
|
}
|
|
|
|
attemptPeerReconnect(context, peerId, handlers);
|
|
}, PEER_RECONNECT_INTERVAL_MS);
|
|
|
|
state.peerReconnectTimers.set(peerId, timer);
|
|
}
|
|
|
|
export function attemptPeerReconnect(
|
|
context: PeerConnectionManagerContext,
|
|
peerId: string,
|
|
handlers: RecoveryHandlers
|
|
): void {
|
|
const { callbacks, logger, state } = context;
|
|
|
|
if (state.activePeerConnections.has(peerId)) {
|
|
handlers.removePeer(peerId, { preserveReconnectState: true });
|
|
}
|
|
|
|
const localOderId = callbacks.getIdentifyCredentials()?.oderId ?? null;
|
|
|
|
if (!localOderId) {
|
|
logger.info('Skipping reconnect offer until logical identity is ready', { peerId });
|
|
handlers.createPeerConnection(peerId, false);
|
|
return;
|
|
}
|
|
|
|
const shouldInitiate = peerId !== localOderId && localOderId < peerId;
|
|
|
|
handlers.createPeerConnection(peerId, shouldInitiate);
|
|
|
|
if (shouldInitiate) {
|
|
void handlers.createAndSendOffer(peerId);
|
|
return;
|
|
}
|
|
|
|
logger.info('Waiting for remote reconnect offer based on deterministic initiator selection', {
|
|
localOderId,
|
|
peerId
|
|
});
|
|
}
|
|
|
|
export function requestVoiceStateFromPeer(
|
|
state: PeerConnectionManagerState,
|
|
logger: PeerConnectionManagerContext['logger'],
|
|
peerId: string
|
|
): void {
|
|
const peerData = state.activePeerConnections.get(peerId);
|
|
|
|
if (peerData?.dataChannel?.readyState === DATA_CHANNEL_STATE_OPEN) {
|
|
try {
|
|
peerData.dataChannel.send(JSON.stringify({ type: P2P_TYPE_VOICE_STATE_REQUEST }));
|
|
} catch (error) {
|
|
logger.warn('Failed to request voice state', error);
|
|
}
|
|
}
|
|
}
|
|
|
|
/** Return a snapshot copy of the currently-connected peer IDs. */
|
|
export function getConnectedPeerIds(state: PeerConnectionManagerState): string[] {
|
|
return [...state.connectedPeersList];
|
|
}
|
|
|
|
export function addToConnectedPeers(state: PeerConnectionManagerState, peerId: string): void {
|
|
if (!state.connectedPeersList.includes(peerId)) {
|
|
state.connectedPeersList = [...state.connectedPeersList, peerId];
|
|
state.connectedPeersChanged$.next(state.connectedPeersList);
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Remove a peer from the connected list and notify subscribers.
|
|
*/
|
|
export function removeFromConnectedPeers(
|
|
state: PeerConnectionManagerState,
|
|
peerId: string
|
|
): void {
|
|
state.connectedPeersList = state.connectedPeersList.filter(
|
|
(connectedId) => connectedId !== peerId
|
|
);
|
|
|
|
state.connectedPeersChanged$.next(state.connectedPeersList);
|
|
}
|
|
|
|
/** Reset the connected peers list to empty and notify subscribers. */
|
|
export function resetConnectedPeers(state: PeerConnectionManagerState): void {
|
|
state.connectedPeersList = [];
|
|
state.connectedPeersChanged$.next([]);
|
|
}
|