bridge (dbus): Add time out on connection:

- Add a 5s timeout on connection to avoid infinite hang when socket is not working.
- Raise `BridgeInitError` on time-out (needed to make `auto` bridge fallback correctly).

libervia/backend/bridge/bridge_constructor/constructors/dbus/dbus_core_template.py

@@ -36,6 +36,7 @@ const_ERROR_PREFIX = const_INT_PREFIX + ".error"
 const_OBJ_PATH = "/org/libervia/Libervia/bridge"
 const_CORE_SUFFIX = ".core"
 const_PLUGIN_SUFFIX = ".plugin"
+const_CONNECT_TIMEOUT = 5
 
 
 class ParseError(Exception):
@@ -123,7 +124,23 @@ class bridge:
 
     async def post_init(self):
         try:
-            self.conn = await client.connect(reactor)
+            d = client.connect(reactor)
+            reactor.callLater(const_CONNECT_TIMEOUT, d.cancel)
+
+            def on_cancel(failure):
+                if failure.check(defer.CancelledError):
+                    log.error(
+                        f"D-Bus connection timed out after {const_CONNECT_TIMEOUT} s, "
+                        "please check the D-Bus service"
+                    )
+                return failure
+
+            d.addErrback(on_cancel)
+            self.conn = await d
+        except defer.CancelledError:
+            # The connect deferred was cancelled by the timeout, we raise BridgeInitError
+            # so that the "auto" bridge can fall back to the next one.
+            raise BridgeInitError from None
         except error.DBusException as e:
             if e.errName == "org.freedesktop.DBus.Error.NotSupported":
                 log.error(

libervia/backend/bridge/bridge_constructor/constructors/dbus/dbus_frontend_template.py

@@ -19,7 +19,7 @@
 import asyncio
 from functools import partial
 
-from twisted.internet import reactor
+from twisted.internet import defer, reactor
 from txdbus import client
 
 from .bridge_frontend import BridgeException
@@ -40,6 +40,27 @@ const_OBJ_PATH = "/org/libervia/Libervia/bridge"
 const_CORE_SUFFIX = ".core"
 const_PLUGIN_SUFFIX = ".plugin"
 const_TIMEOUT = 120
+const_CONNECT_TIMEOUT = 5
+
+
+def _connect_dbus():
+    """Connect to the D-Bus session bus with a timeout.
+
+    @return: Deferred to the D-Bus connection
+    """
+    d = client.connect(reactor)
+    reactor.callLater(const_CONNECT_TIMEOUT, d.cancel)
+
+    def on_cancel(failure):
+        if failure.check(defer.CancelledError):
+            log.error(
+                f"D-Bus connection timed out after {const_CONNECT_TIMEOUT} s, "
+                "please check the D-Bus service"
+            )
+        return failure
+
+    d.addErrback(on_cancel)
+    return d
 
 
 def dbus_to_bridge_exception(failure):
@@ -192,7 +213,7 @@ class bridge:
     async def bridge_connect(self) -> None:
         """Connect to the D-Bus bridge."""
         try:
-            self.conn = await client.connect(reactor)
+            self.conn = await _connect_dbus()
             self.db_object = await self.conn.getRemoteObject(
                 const_INT_PREFIX, const_OBJ_PATH
             )
@@ -218,6 +239,10 @@ class bridge:
             )
         except BridgeInitError as e:
             raise e
+        except defer.CancelledError:
+            # The connect deferred was cancelled by the timeout, we raise BridgeInitError
+            # so that the "auto" bridge can fall back to the next one.
+            raise BridgeInitError from None
         except Exception as e:
             msg = str(e)
             if (
@@ -281,7 +306,7 @@ class AIOBridge(bridge):
     async def bridge_connect(self):
         """Connect to the D-Bus bridge (async version)."""
         try:
-            self.conn = await client.connect(reactor).asFuture(asyncio.get_event_loop())
+            self.conn = await _connect_dbus().asFuture(asyncio.get_event_loop())
             self.db_object = await self.conn.getRemoteObject(
                 const_INT_PREFIX, const_OBJ_PATH
             ).asFuture(asyncio.get_event_loop())
@@ -308,6 +333,10 @@ class AIOBridge(bridge):
             )
         except BridgeInitError as e:
             raise e
+        except defer.CancelledError:
+            # The connect deferred was cancelled by the timeout, we raise BridgeInitError
+            # so that the "auto" bridge can fall back to the next one.
+            raise BridgeInitError from None
         except Exception as e:
             msg = str(e)
             if (

libervia/backend/bridge/dbus_bridge.py

@@ -36,6 +36,7 @@ const_ERROR_PREFIX = const_INT_PREFIX + ".error"
 const_OBJ_PATH = "/org/libervia/Libervia/bridge"
 const_CORE_SUFFIX = ".core"
 const_PLUGIN_SUFFIX = ".plugin"
+const_CONNECT_TIMEOUT = 5
 
 
 class ParseError(Exception):
@@ -571,7 +572,23 @@ class bridge:
 
     async def post_init(self):
         try:
-            self.conn = await client.connect(reactor)
+            d = client.connect(reactor)
+            reactor.callLater(const_CONNECT_TIMEOUT, d.cancel)
+
+            def on_cancel(failure):
+                if failure.check(defer.CancelledError):
+                    log.error(
+                        f"D-Bus connection timed out after {const_CONNECT_TIMEOUT} s, "
+                        "please check the D-Bus service"
+                    )
+                return failure
+
+            d.addErrback(on_cancel)
+            self.conn = await d
+        except defer.CancelledError:
+            # The connect deferred was cancelled by the timeout, we raise BridgeInitError
+            # so that the "auto" bridge can fall back to the next one.
+            raise BridgeInitError from None
         except error.DBusException as e:
             if e.errName == "org.freedesktop.DBus.Error.NotSupported":
                 log.error(

libervia/frontends/bridge/dbus_bridge.py

@@ -19,7 +19,7 @@
 import asyncio
 from functools import partial
 
-from twisted.internet import reactor
+from twisted.internet import defer, reactor
 from txdbus import client
 
 from .bridge_frontend import BridgeException
@@ -40,6 +40,27 @@ const_OBJ_PATH = "/org/libervia/Libervia/bridge"
 const_CORE_SUFFIX = ".core"
 const_PLUGIN_SUFFIX = ".plugin"
 const_TIMEOUT = 120
+const_CONNECT_TIMEOUT = 5
+
+
+def _connect_dbus():
+    """Connect to the D-Bus session bus with a timeout.
+
+    @return: Deferred to the D-Bus connection
+    """
+    d = client.connect(reactor)
+    reactor.callLater(const_CONNECT_TIMEOUT, d.cancel)
+
+    def on_cancel(failure):
+        if failure.check(defer.CancelledError):
+            log.error(
+                f"D-Bus connection timed out after {const_CONNECT_TIMEOUT} s, "
+                "please check the D-Bus service"
+            )
+        return failure
+
+    d.addErrback(on_cancel)
+    return d
 
 
 def dbus_to_bridge_exception(failure):
@@ -192,7 +213,7 @@ class bridge:
     async def bridge_connect(self) -> None:
         """Connect to the D-Bus bridge."""
         try:
-            self.conn = await client.connect(reactor)
+            self.conn = await _connect_dbus()
             self.db_object = await self.conn.getRemoteObject(
                 const_INT_PREFIX, const_OBJ_PATH
             )
@@ -218,6 +239,10 @@ class bridge:
             )
         except BridgeInitError as e:
             raise e
+        except defer.CancelledError:
+            # The connect deferred was cancelled by the timeout, we raise BridgeInitError
+            # so that the "auto" bridge can fall back to the next one.
+            raise BridgeInitError from None
         except Exception as e:
             msg = str(e)
             if (
@@ -963,7 +988,7 @@ class AIOBridge(bridge):
     async def bridge_connect(self):
         """Connect to the D-Bus bridge (async version)."""
         try:
-            self.conn = await client.connect(reactor).asFuture(asyncio.get_event_loop())
+            self.conn = await _connect_dbus().asFuture(asyncio.get_event_loop())
             self.db_object = await self.conn.getRemoteObject(
                 const_INT_PREFIX, const_OBJ_PATH
             ).asFuture(asyncio.get_event_loop())
@@ -990,6 +1015,10 @@ class AIOBridge(bridge):
             )
         except BridgeInitError as e:
             raise e
+        except defer.CancelledError:
+            # The connect deferred was cancelled by the timeout, we raise BridgeInitError
+            # so that the "auto" bridge can fall back to the next one.
+            raise BridgeInitError from None
         except Exception as e:
             msg = str(e)
             if (