Merge pull request #2695 from makeittech/fix/gh-2638-chat-ui-freeze

fix(server): rebind message-stream upstreams after a managed OpenCode restart (#2638)

packages/web/server/lib/opencode/DOCUMENTATION.md

@@ -112,7 +112,7 @@ This module provides OpenCode server integration utilities for the web server ru
 The runtime maintains active-session count incrementally from idempotent activity phase transitions. Upstream stall-timeout and lifecycle health checks read it in O(1); the hourly cleanup removes activity phases older than 24 hours without broadcasting synthetic state transitions. Snapshot generation remains reserved for the session-activity API.
 
 ## Public exports (lifecycle.js)
-- `createOpenCodeLifecycleRuntime(dependencies)`: creates lifecycle runtime for managed/external OpenCode process orchestration.
+- `createOpenCodeLifecycleRuntime(dependencies)`: creates lifecycle runtime for managed/external OpenCode process orchestration. The optional `onOpenCodeRestarted` dependency (default `null`) is fired after a successful managed restart; `index.js` wires it to `messageStreamRuntime.rebindUpstream()` so event-stream readers rebind to the possibly-new port (a restart can land on a new port while an orphaned process keeps the old one, which would otherwise leave the chat UI silent — issue #2638).
 - Returned API:
   - `startOpenCode()`
   - `restartOpenCode()`

packages/web/server/lib/opencode/lifecycle.js

@@ -48,6 +48,7 @@ export const createOpenCodeLifecycleRuntime = (deps) => {
     getActiveSessionCount = () => 0,
     reapManagedOrphanedProcesses = reapOrphanedProcesses,
     getWarmupDirectories = async () => [],
+    onOpenCodeRestarted = null,
     now = Date.now,
   } = deps;
 
@@ -694,6 +695,17 @@ export const createOpenCodeLifecycleRuntime = (deps) => {
         setupProxy(state.expressApp);
         ensureOpenCodeApiPrefix();
       }
+
+      // The restart may have landed on a NEW port (the old one can remain
+      // occupied by an orphaned process, e.g. Windows killProcessOnPort is a
+      // no-op). Upstream event readers pinned to the old process would keep
+      // the UI silent forever, so rebind them to the current port. Best
+      // effort: a failure here must not fail the restart itself.
+      try {
+        onOpenCodeRestarted?.();
+      } catch (error) {
+        console.warn('Failed to rebind event stream after OpenCode restart:', error?.message ?? error);
+      }
     })();
 
     try {

packages/web/server/lib/opencode/lifecycle.test.js

@@ -287,6 +287,70 @@ describe('OpenCode lifecycle', () => {
     expect(spawnMock).toHaveBeenCalledTimes(1);
   });
 
+  it('calls onOpenCodeRestarted after a successful managed restart', async () => {
+    const close = vi.fn(async () => {});
+    const replacement = createMockChild();
+    const onOpenCodeRestarted = vi.fn();
+    globalThis.fetch = vi.fn(async () => ({
+      ok: false,
+      json: async () => null,
+    }));
+    spawnMock.mockImplementationOnce(() => {
+      queueMicrotask(() => {
+        replacement.stdout.emit('data', 'opencode server listening on http://127.0.0.1:45678\n');
+      });
+      return replacement;
+    });
+    const runtime = createRuntime({ onOpenCodeRestarted }, {
+      openCodePort: 45678,
+      openCodeProcess: {
+        pid: null,
+        exitCode: 1,
+        signalCode: null,
+        close,
+      },
+    });
+
+    await runtime.triggerHealthCheck();
+
+    expect(close).toHaveBeenCalledTimes(1);
+    expect(spawnMock).toHaveBeenCalledTimes(1);
+    // The restart completed on a (possibly new) port — the event-stream
+    // upstreams must rebind so the UI keeps receiving events (#2638).
+    expect(onOpenCodeRestarted).toHaveBeenCalledTimes(1);
+  });
+
+  it('does not call onOpenCodeRestarted when a managed restart fails', async () => {
+    const close = vi.fn(async () => {});
+    const onOpenCodeRestarted = vi.fn();
+    globalThis.fetch = vi.fn(async () => ({
+      ok: false,
+      json: async () => null,
+    }));
+    spawnMock.mockImplementation(() => {
+      const child = createMockChild();
+      queueMicrotask(() => {
+        child.emit('error', new Error('spawn failed'));
+      });
+      return child;
+    });
+    const runtime = createRuntime({ onOpenCodeRestarted }, {
+      openCodePort: 45678,
+      openCodeProcess: {
+        pid: null,
+        exitCode: 1,
+        signalCode: null,
+        close,
+      },
+    });
+
+    // triggerHealthCheck logs instead of rethrowing; call restartOpenCode
+    // directly to observe the failure result.
+    await expect(runtime.restartOpenCode()).rejects.toThrow();
+
+    expect(onOpenCodeRestarted).not.toHaveBeenCalled();
+  });
+
   it('launches managed OpenCode with the managed PATH', async () => {
     delete process.env.OPENCODE_BINARY;
     const child = createMockChild();