fix(gateway): restore WebSocket heartbeat ping on Windows (#722)
This commit is contained in:
committed by
GitHub
Unverified
parent
ec8db0be75
commit
eb948d5820
@@ -787,14 +787,7 @@ export class GatewayManager extends EventEmitter {
|
||||
port,
|
||||
connectedAt: Date.now(),
|
||||
});
|
||||
// On Windows, skip WebSocket heartbeat ping to avoid cascading failures:
|
||||
// heartbeat timeout → terminate socket → reconnect → port conflict
|
||||
// (old process holds port due to TCP TIME_WAIT) → ~2 min downtime.
|
||||
// Gateway is a local child process; actual crashes are caught by the
|
||||
// process exit handler, and graceful restarts use code=1012 close frames.
|
||||
if (process.platform !== 'win32') {
|
||||
this.startPing();
|
||||
}
|
||||
this.startPing();
|
||||
},
|
||||
onMessage: (message) => {
|
||||
this.handleMessage(message);
|
||||
@@ -900,6 +893,14 @@ export class GatewayManager extends EventEmitter {
|
||||
} catch (error) {
|
||||
logger.warn('Failed to terminate stale Gateway socket after heartbeat timeout:', error);
|
||||
}
|
||||
|
||||
// On Windows, onCloseAfterHandshake intentionally skips scheduleReconnect()
|
||||
// to avoid double-reconnect races with the process exit handler. However,
|
||||
// a heartbeat timeout means the socket is stale while the process may still
|
||||
// be alive (no exit event), so we must explicitly trigger reconnect here.
|
||||
if (process.platform === 'win32') {
|
||||
this.scheduleReconnect();
|
||||
}
|
||||
},
|
||||
});
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user