Release v0.51.234 — Release HB (stage-q4): duplicate-instance startup guard + remote-terminal workspace paths

CHANGELOG.md

@@ -3,6 +3,12 @@
 
 ## [Unreleased]
 
+## [v0.51.234] — 2026-06-03 — Release HB (stage-q4 — duplicate-instance startup guard + remote-terminal workspace paths)
+
+### Fixed
+- The server now refuses to start when a live instance is already responding on the configured port, instead of silently sharing it (a Windows/macOS hazard where `SO_REUSEADDR` semantics let two processes bind 8787 at once, #3289). Rather than globally disabling `SO_REUSEADDR` (which would brick legitimate fast restarts — `ctl.sh restart` and the `os.execv` self-update path rebind immediately and would hit the TIME_WAIT window), startup now runs a live-listener probe (`_abort_if_already_serving`): a TCP connect + `GET /health` with a 2s timeout. A live instance answers and startup aborts with a clear message; a dying instance whose socket still lingers in the kernel backlog accepts the connection but never responds, so the probe times out and startup proceeds — preserving fast restart. On Windows, `SO_EXCLUSIVEADDRUSE` is set in a `server_bind()` override to get true exclusive binding (POSIX keeps the inherited `allow_reuse_address = True`) (#3289, @rodboev).
+- Remote/SSH terminal profiles can now use target-side workspace paths that don't exist on the WebUI host. Workspace validation/resolution previously `stat()`-ed every path against the WebUI server's local filesystem, so a `terminal.cwd` (or session workspace) living on the remote target was rejected as nonexistent. For profiles whose terminal backend is non-local, paths **under the configured `terminal.cwd`** now pass validation without a server-local existence check, and stale server-local `last_workspace` values are ignored unless they fall under the remote cwd. Local profiles are unchanged — the bypass only fires for remote backends and only for paths contained within `terminal.cwd` (#3486, @dso2ng).
+
 ## [v0.51.233] — 2026-06-03 — Release HA (stage-q3 — session-truncate keep_count guard against silent transcript loss)
 
 ### Fixed

api/workspace.py

@@ -57,6 +57,46 @@ def _last_workspace_file() -> Path:
     return _profile_state_dir() / 'last_workspace.txt'
 
 
+def _is_remote_terminal_backend(terminal_cfg: dict | None) -> bool:
+    """Return True when the active terminal backend runs outside this WebUI host."""
+    if not isinstance(terminal_cfg, dict):
+        return False
+    backend = str(terminal_cfg.get('backend') or '').strip().lower()
+    return backend not in ('', 'local')
+
+
+def _remote_terminal_cwd() -> str | None:
+    """Return target-side terminal cwd for remote profiles, without local stat()."""
+    try:
+        from api.config import get_config
+
+        terminal_cfg = get_config().get('terminal', {})
+        if not _is_remote_terminal_backend(terminal_cfg):
+            return None
+        cwd = str(terminal_cfg.get('cwd') or '').strip()
+        if not cwd or cwd == '.':
+            return None
+        return cwd
+    except Exception:
+        logger.debug("Failed to read remote terminal cwd", exc_info=True)
+        return None
+
+
+def _remote_terminal_workspace_candidate(path: str | Path) -> Path | None:
+    """Return a non-stat'ed target-side Path when it is under terminal.cwd."""
+    cwd = _remote_terminal_cwd()
+    if not cwd:
+        return None
+    raw = _strip_surrounding_quotes(str(path)).strip()
+    if not raw:
+        return None
+    candidate = Path(raw).expanduser().resolve()
+    base = Path(cwd).expanduser().resolve()
+    if candidate == base or _is_within(candidate, base):
+        return candidate
+    return None
+
+
 def _profile_default_workspace() -> str:
     """Read the profile's default workspace from its config.yaml.
 
@@ -65,25 +105,31 @@ def _profile_default_workspace() -> str:
       2. 'default_workspace' — alternate explicit key
       3. 'terminal.cwd'      — hermes-agent terminal working dir (most common)
 
+    For remote/SSH terminal profiles, ``terminal.cwd`` lives on the target
+    machine, not on the WebUI server. In that case return it without a
+    server-local existence check so WebUI can send the correct workspace hint
+    to the agent/tool backend.
+
     Falls back to the live DEFAULT_WORKSPACE from api.config.
     """
     try:
         from api.config import get_config
         cfg = get_config()
+        terminal_cfg = cfg.get('terminal', {})
+        remote_terminal = _is_remote_terminal_backend(terminal_cfg)
         # Explicit webui workspace keys first
         for key in ('workspace', 'default_workspace'):
             ws = cfg.get(key)
             if ws:
                 p = Path(str(ws)).expanduser().resolve()
-                if p.is_dir():
+                if remote_terminal or p.is_dir():
                     return str(p)
         # Fall through to terminal.cwd — the agent's configured working directory
-        terminal_cfg = cfg.get('terminal', {})
         if isinstance(terminal_cfg, dict):
             cwd = terminal_cfg.get('cwd', '')
             if cwd and str(cwd) not in ('.', ''):
                 p = Path(str(cwd)).expanduser().resolve()
-                if p.is_dir():
+                if remote_terminal or p.is_dir():
                     return str(p)
     except (ImportError, Exception):
         logger.debug("Failed to load profile default workspace config")
@@ -225,19 +271,35 @@ def save_workspaces(workspaces: list) -> None:
 
 
 def get_last_workspace() -> str:
+    remote_cwd = _remote_terminal_cwd()
+
+    def valid_last_workspace(raw: str) -> str | None:
+        if not raw:
+            return None
+        if remote_cwd:
+            # For remote/SSH profiles, last_workspace is target-side state. Do
+            # not accept stale server-local paths merely because they exist on
+            # the WebUI host; require the value to stay under terminal.cwd.
+            if _remote_terminal_workspace_candidate(raw) is not None:
+                return raw
+            return None
+        if Path(raw).is_dir():
+            return raw
+        return None
+
     lw_file = _last_workspace_file()
     if lw_file.exists():
         try:
-            p = lw_file.read_text(encoding='utf-8').strip()
-            if p and Path(p).is_dir():
+            p = valid_last_workspace(lw_file.read_text(encoding='utf-8').strip())
+            if p:
                 return p
         except Exception:
             logger.debug("Failed to read last workspace from %s", lw_file)
     # Fallback: try global file
     if _GLOBAL_LW_FILE.exists():
         try:
-            p = _GLOBAL_LW_FILE.read_text(encoding='utf-8').strip()
-            if p and Path(p).is_dir():
+            p = valid_last_workspace(_GLOBAL_LW_FILE.read_text(encoding='utf-8').strip())
+            if p:
                 return p
         except Exception:
             logger.debug("Failed to read global last workspace")
@@ -574,8 +636,17 @@ def resolve_trusted_workspace(path: str | Path | None = None) -> Path:
     candidate = Path(path).expanduser().resolve()
 
     access_error = _workspace_access_error(candidate)
+    remote_candidate = _remote_terminal_workspace_candidate(path)
     if access_error:
-        raise ValueError(access_error)
+        # For remote terminal profiles, workspace paths belong to the target
+        # machine. Allow paths under terminal.cwd so session switching can
+        # update the workspace hint even though this WebUI host cannot stat
+        # the target-side path.
+        if remote_candidate is None:
+            raise ValueError(access_error)
+
+    if remote_candidate is not None:
+        return remote_candidate
 
     # (A) Trusted if under the user's home directory — cross-platform via Path.home()
     # Must be checked before system roots to allow symlinks like /var/home.
@@ -658,8 +729,16 @@ def validate_workspace_to_add(path: str) -> Path:
     candidate = Path(path).expanduser().resolve()
 
     access_error = _workspace_access_error(candidate)
+    remote_candidate = _remote_terminal_workspace_candidate(path)
     if access_error:
-        raise ValueError(access_error)
+        # Remote terminal profiles validate workspace existence on the target
+        # machine, not on the WebUI server. Permit target-side paths under
+        # terminal.cwd.
+        if remote_candidate is None:
+            raise ValueError(access_error)
+
+    if remote_candidate is not None:
+        return remote_candidate
 
     # Home directory is always trusted regardless of where it lives on disk
     # (e.g. /var/home/... on systemd-homed Fedora/RHEL).

server.py

@@ -184,6 +184,13 @@ def __init__(self, *args, **kwargs):
         self.accept_loop_requests_total = 0
         self.accept_loop_last_request_at = 0.0
 
+    def server_bind(self):
+        if sys.platform == 'win32':
+            self.allow_reuse_address = False
+            SO_EXCLUSIVEADDRUSE = getattr(socket, 'SO_EXCLUSIVEADDRUSE', -5)
+            self.socket.setsockopt(socket.SOL_SOCKET, SO_EXCLUSIVEADDRUSE, 1)
+        super().server_bind()
+
     def _handle_request_noblock(self):
         """Record accept-loop progress before dispatching a request handler.
 
@@ -477,6 +484,25 @@ def _log_shutdown_audit(reason: str = "serve_forever_exit") -> None:
     )
 
 
+def _abort_if_already_serving(host: str, port: int) -> None:
+    """Refuse to start if a live HTTP server is already responding on this port."""
+    probe_host = '127.0.0.1' if host in ('0.0.0.0', '', '::') else host
+    try:
+        with socket.create_connection((probe_host, port), timeout=2) as s:
+            s.sendall(b'GET /health HTTP/1.0\r\nHost: localhost\r\n\r\n')
+            s.settimeout(2)
+            data = s.recv(512)
+            if data:
+                print(
+                    f'[!!] FATAL: Another server is already responding on'
+                    f' {probe_host}:{port}. Stop the existing instance first.',
+                    flush=True,
+                )
+                sys.exit(1)
+    except (ConnectionRefusedError, ConnectionResetError, OSError, socket.timeout):
+        pass
+
+
 def main() -> None:
     from api.config import print_startup_config, verify_hermes_imports, _HERMES_FOUND
 
@@ -572,6 +598,7 @@ def main() -> None:
     except Exception as e:
         print(f'[!!] WARNING: Plugin loading failed: {e}', flush=True)
 
+    _abort_if_already_serving(HOST, PORT)
     httpd = QuietHTTPServer((HOST, PORT), Handler)
 
     # ── TLS/HTTPS setup (optional) ─────────────────────────────────────────
@@ -597,6 +624,7 @@ def main() -> None:
     try:
         httpd.serve_forever()
     finally:
+        httpd.server_close()
         _log_shutdown_audit()
         # Stop the gateway watcher on shutdown
         try:

tests/test_remote_terminal_workspace.py

@@ -0,0 +1,57 @@
+from pathlib import Path
+
+import pytest
+
+from api import config as api_config
+from api import workspace
+
+
+REMOTE_CWD = "/Users/joeyshiue"
+
+
+def _remote_config(**overrides):
+    cfg = {"terminal": {"backend": "ssh", "cwd": REMOTE_CWD}}
+    cfg.update(overrides)
+    return cfg
+
+
+def test_remote_terminal_cwd_is_profile_default_without_local_stat(monkeypatch, tmp_path):
+    fallback = tmp_path / "fallback"
+    fallback.mkdir()
+
+    monkeypatch.setattr(api_config, "DEFAULT_WORKSPACE", fallback)
+    monkeypatch.setattr(api_config, "get_config", lambda: _remote_config())
+
+    assert workspace._profile_default_workspace() == REMOTE_CWD
+
+
+def test_remote_terminal_last_workspace_ignores_stale_local_path(monkeypatch, tmp_path):
+    stale_local = tmp_path / "stale-local"
+    stale_local.mkdir()
+    last_workspace = tmp_path / "last_workspace.txt"
+    last_workspace.write_text(str(stale_local), encoding="utf-8")
+
+    monkeypatch.setattr(api_config, "get_config", lambda: _remote_config())
+    monkeypatch.setattr(workspace, "_last_workspace_file", lambda: last_workspace)
+    monkeypatch.setattr(workspace, "_GLOBAL_LW_FILE", tmp_path / "missing-global-last-workspace.txt")
+
+    assert workspace.get_last_workspace() == REMOTE_CWD
+
+
+def test_remote_terminal_workspace_paths_under_cwd_do_not_require_local_existence(monkeypatch):
+    monkeypatch.setattr(api_config, "get_config", lambda: _remote_config())
+
+    target_side_project = f"{REMOTE_CWD}/projects/demo"
+
+    assert workspace.validate_workspace_to_add(target_side_project) == Path(target_side_project).resolve()
+    assert workspace.resolve_trusted_workspace(target_side_project) == Path(target_side_project).resolve()
+
+
+def test_remote_terminal_workspace_paths_outside_cwd_still_reject(monkeypatch):
+    monkeypatch.setattr(api_config, "get_config", lambda: _remote_config())
+
+    with pytest.raises(ValueError, match="Path does not exist"):
+        workspace.validate_workspace_to_add("/Users/other/projects/demo")
+
+    with pytest.raises(ValueError, match="Path does not exist"):
+        workspace.resolve_trusted_workspace("/Users/other/projects/demo")

tests/test_server_port_exclusivity.py

@@ -0,0 +1,90 @@
+"""Duplicate-instance guard: a second server on the same port must be detected
+and refused before bind, not silently shared (#3289)."""
+
+from __future__ import annotations
+
+import socket
+import sys
+import threading
+from http.server import BaseHTTPRequestHandler, HTTPServer
+
+import pytest
+
+from tests._pytest_port import TEST_PORT
+
+
+# ── SO_EXCLUSIVEADDRUSE on Windows ──────────────────────────────────────────
+
+@pytest.mark.skipif(sys.platform != 'win32', reason='Windows-only socket option')
+def test_exclusive_addr_use_set_on_windows():
+    from server import QuietHTTPServer
+    port = TEST_PORT + 901
+    httpd = QuietHTTPServer(('127.0.0.1', port), BaseHTTPRequestHandler)
+    try:
+        val = httpd.socket.getsockopt(
+            socket.SOL_SOCKET,
+            getattr(socket, 'SO_EXCLUSIVEADDRUSE', -5),
+        )
+        assert val != 0, 'SO_EXCLUSIVEADDRUSE should be set on Windows'
+    finally:
+        httpd.server_close()
+
+
+# ── Live-listener probe ─────────────────────────────────────────────────────
+
+def test_probe_detects_live_server():
+    """_abort_if_already_serving must call sys.exit when a live server responds."""
+    from server import _abort_if_already_serving
+
+    port = TEST_PORT + 902
+
+    class Handler(BaseHTTPRequestHandler):
+        def do_GET(self):  # noqa: N802
+            self.send_response(200)
+            self.end_headers()
+            self.wfile.write(b'ok')
+        def log_message(self, *a):
+            pass
+
+    httpd = HTTPServer(('127.0.0.1', port), Handler)
+    t = threading.Thread(target=httpd.serve_forever, daemon=True)
+    t.start()
+    try:
+        with pytest.raises(SystemExit):
+            _abort_if_already_serving('127.0.0.1', port)
+    finally:
+        httpd.shutdown()
+        httpd.server_close()
+
+
+def test_probe_allows_startup_when_nothing_listening():
+    """_abort_if_already_serving must return normally on a free port."""
+    from server import _abort_if_already_serving
+
+    port = TEST_PORT + 903
+    _abort_if_already_serving('127.0.0.1', port)
+
+
+def test_probe_allows_startup_on_unresponsive_socket():
+    """A socket that accepts but never responds (e.g. dying instance still in
+    kernel backlog) should not block startup."""
+    from server import _abort_if_already_serving
+
+    port = TEST_PORT + 904
+    srv = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
+    srv.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
+    srv.bind(('127.0.0.1', port))
+    srv.listen(1)
+    try:
+        _abort_if_already_serving('127.0.0.1', port)
+    finally:
+        srv.close()
+
+
+def test_probe_normalizes_wildcard_host():
+    """0.0.0.0 and :: should probe 127.0.0.1, not the literal wildcard."""
+    from server import _abort_if_already_serving
+
+    port = TEST_PORT + 905
+    _abort_if_already_serving('0.0.0.0', port)
+    _abort_if_already_serving('::', port)