engine: More work to fix cascading error dialogs

If a remote network connection stalls, the tick queue becomes backed
up while we wait for the hung connection to continue. While this
happens, the queue is filled up with other requests to poll the hung
connection.

When the connection finally times out, the tick thread closes the
connection via an idle callback. However before that callback gets
a chance to run, all the other poll requests for the dead connection
are processed, all launching their own error dialog.

Mark the connection as 'closing' before conn.close is scheduled, and
use it to short circuit the tick() routine.

virtManager/connection.py

@@ -87,6 +87,7 @@ class vmmConnection(vmmGObject):
         self.connectThread = None
         self.connectError = None
         self._backend = virtinst.VirtualConnection(self._uri)
+        self._closing = False
 
         self._caps = None
         self._caps_xml = None
@@ -957,6 +958,10 @@ class vmmConnection(vmmGObject):
         self.config.set_conn_autoconnect(self.get_uri(), val)
 
     def close(self):
+        if self.state != self.STATE_DISCONNECTED:
+            logging.debug("conn.close() uri=%s", self.get_uri())
+        self._closing = True
+
         def cleanup(devs):
             for dev in devs.values():
                 try:
@@ -1002,6 +1007,7 @@ class vmmConnection(vmmGObject):
         self.vms = {}
 
         self._change_state(self.STATE_DISCONNECTED)
+        self._closing = False
 
     def _cleanup(self):
         self.close()
@@ -1164,7 +1170,43 @@ class vmmConnection(vmmGObject):
             kwargs["stats_update"] = False
         self.idle_emit("priority-tick", kwargs)
 
-    def tick(self, stats_update,
+    def tick(self, *args, **kwargs):
+        e = None
+        try:
+            self._tick(*args, **kwargs)
+        except KeyboardInterrupt:
+            raise
+        except Exception, e:
+            pass
+
+        if e is None:
+            return
+
+        from_remote = getattr(libvirt, "VIR_FROM_REMOTE", None)
+        from_rpc = getattr(libvirt, "VIR_FROM_RPC", None)
+        sys_error = getattr(libvirt, "VIR_ERR_SYSTEM_ERROR", None)
+
+        dom = -1
+        code = -1
+        if isinstance(e, libvirt.libvirtError):
+            dom = e.get_error_domain()
+            code = e.get_error_code()
+
+        logging.debug("Error polling connection %s",
+            self.get_uri(), exc_info=True)
+
+        if (dom in [from_remote, from_rpc] and
+            code in [sys_error]):
+            e = None
+            logging.debug("Not showing user error since libvirtd "
+                "appears to have stopped.")
+
+        self._closing = True
+        self.idle_add(self.close)
+        if e:
+            raise e  # pylint: disable=raising-bad-type
+
+    def _tick(self, stats_update,
              pollvm=False, pollnet=False,
              pollpool=False, polliface=False,
              pollnodedev=False, pollmedia=False,
@@ -1173,7 +1215,7 @@ class vmmConnection(vmmGObject):
         main update function: polls for new objects, updates stats, ...
         @force: Perform the requested polling even if async events are in use
         """
-        if self.state != self.STATE_ACTIVE:
+        if self.state != self.STATE_ACTIVE or self._closing:
             return
 
         if not pollvm:

virtManager/engine.py

@@ -26,8 +26,8 @@ import logging
 import re
 import Queue
 import threading
+import traceback
 
-import libvirt
 from virtinst import util
 
 from virtManager import packageutils
@@ -336,45 +336,18 @@ class vmmEngine(vmmGObject):
 
     def _handle_tick_queue(self):
         while True:
-            ignore1, ignore2, obj, kwargs = self._tick_queue.get()
+            ignore1, ignore2, conn, kwargs = self._tick_queue.get()
-            self._tick_single_conn(obj, kwargs)
-            self._tick_queue.task_done()
-        return 1
-
-    def _tick_single_conn(self, conn, kwargs):
-        e = None
             try:
                 conn.tick(**kwargs)
-        except KeyboardInterrupt:
-            raise
             except Exception, e:
-            pass
+                tb = "".join(traceback.format_exc())
+                error_msg = (_("Error polling connection '%s': %s")
+                    % (conn.get_uri(), e))
+                self.idle_add(lambda: self.err.show_err(error_msg,
+                    details=tb))
 
-        if e is None:
+            self._tick_queue.task_done()
-            return
+        return 1
-
-        from_remote = getattr(libvirt, "VIR_FROM_REMOTE", None)
-        from_rpc = getattr(libvirt, "VIR_FROM_RPC", None)
-        sys_error = getattr(libvirt, "VIR_ERR_SYSTEM_ERROR", None)
-
-        dom = -1
-        code = -1
-        if isinstance(e, libvirt.libvirtError):
-            dom = e.get_error_domain()
-            code = e.get_error_code()
-
-        if (dom in [from_remote, from_rpc] and
-            code in [sys_error]):
-            logging.exception("Could not refresh connection %s",
-                              conn.get_uri())
-            logging.debug("Closing connection since libvirtd "
-                          "appears to have stopped")
-        else:
-            error_msg = _("Error polling connection '%s': %s") \
-                % (conn.get_uri(), e)
-            self.idle_add(lambda: self.err.show_err(error_msg))
-
-        self.idle_add(conn.close)
 
 
     def increment_window_counter(self, src):