Recover from daemon hello timeout (#1179)

src/main/daemon/client.test.ts

@@ -51,6 +51,7 @@ describe('DaemonClient', () => {
     onControlMessage?: (msg: unknown) => string | null
     onStreamHello?: (msg: HelloMessage) => void
     rejectVersion?: boolean
+    ignoreHello?: boolean
   }): Promise<void> {
     return new Promise((resolve) => {
       server = createServer((socket) => {
@@ -69,6 +70,9 @@ describe('DaemonClient', () => {
 
             if (msg.type === 'hello') {
               const hello = msg as HelloMessage
+              if (opts?.ignoreHello) {
+                return
+              }
               if (opts?.rejectVersion) {
                 socket.write(encodeNdjson({ type: 'hello', ok: false, error: 'Version mismatch' }))
                 return
@@ -112,6 +116,13 @@ describe('DaemonClient', () => {
       client = new DaemonClient({ socketPath, tokenPath })
       await expect(client.ensureConnected()).rejects.toThrow()
     })
+
+    it('rejects when the daemon accepts a socket but never answers hello', async () => {
+      await startMockDaemon({ ignoreHello: true })
+
+      client = new DaemonClient({ socketPath, tokenPath, handshakeTimeoutMs: 10 })
+      await expect(client.ensureConnected()).rejects.toThrow('Hello timed out')
+    })
   })
 
   describe('RPC', () => {

src/main/daemon/client.ts

@@ -7,12 +7,14 @@ import type { HelloMessage, HelloResponse, RpcResponse, DaemonEvent } from './ty
 import { addNodePtyRecoveryHint } from './node-pty-error-hints'
 
 const CONNECT_TIMEOUT_MS = 5000
+const HELLO_TIMEOUT_MS = 5000
 const REQUEST_TIMEOUT_MS = 30000
 
 export type DaemonClientOptions = {
   socketPath: string
   tokenPath: string
   protocolVersion?: number
+  handshakeTimeoutMs?: number
 }
 
 type PendingRequest = {
@@ -25,6 +27,7 @@ export class DaemonClient {
   private socketPath: string
   private tokenPath: string
   private protocolVersion: number
+  private handshakeTimeoutMs: number
   private clientId = randomUUID()
 
   private controlSocket: Socket | null = null
@@ -50,6 +53,7 @@ export class DaemonClient {
     this.socketPath = opts.socketPath
     this.tokenPath = opts.tokenPath
     this.protocolVersion = opts.protocolVersion ?? PROTOCOL_VERSION
+    this.handshakeTimeoutMs = opts.handshakeTimeoutMs ?? HELLO_TIMEOUT_MS
   }
 
   isConnected(): boolean {
@@ -215,6 +219,7 @@ export class DaemonClient {
         }
 
         socket.removeListener('data', onData)
+        clearTimeout(timer)
         const line = buffer.slice(0, newlineIdx)
         try {
           const response = JSON.parse(line) as HelloResponse
@@ -230,6 +235,13 @@ export class DaemonClient {
         }
       }
 
+      // Why: a wedged daemon can accept the socket but never answer hello,
+      // leaving terminal spawn permanently pending with a blank renderer pane.
+      const timer = setTimeout(() => {
+        socket.removeListener('data', onData)
+        reject(new DaemonProtocolError('Hello timed out'))
+      }, this.handshakeTimeoutMs)
+
       socket.on('data', onData)
       socket.write(encodeNdjson(hello))
     })

src/main/daemon/daemon-pty-adapter.ts

@@ -660,8 +660,9 @@ export class DaemonPtyAdapter implements IPtyProvider {
 // unreachable (daemon died). Checking syscall avoids false positives from
 // token-file ENOENT (readFileSync), which has no syscall or syscall='open'.
 // "Connection lost" / "Not connected" mean the daemon died while we had an
-// active or stale connection. All indicate the daemon is gone and a respawn
-// should be attempted.
+// active or stale connection. "Hello timed out" means the daemon accepted
+// the socket but did not complete the protocol handshake, which is also a
+// wedged-daemon state. All indicate a respawn should be attempted.
 function isDaemonGoneError(err: unknown): boolean {
   if (!(err instanceof Error)) {
     return false
@@ -671,5 +672,5 @@ function isDaemonGoneError(err: unknown): boolean {
     return true
   }
   const msg = err.message
-  return msg === 'Connection lost' || msg === 'Not connected'
+  return msg === 'Connection lost' || msg === 'Not connected' || msg === 'Hello timed out'
 }