Downgrade socket problems to warnings

Previously, these were logged at exception - which logs at ERROR and
prints a traceback.  This led to two problems: first, that they're not
kdcproxy errors (rather problems with the other end); and second, that
the traceback is quite noisy.  Log a simplified version of the
exception instead of the traceback.

In the process, correct the sendall() error message to refer to
sendall(), not recv().

Author: frozencemetery

kdcproxy/__init__.py

@@ -99,17 +99,17 @@ def __await_reply(self, pr, rsocks, wsocks, timeout):
                     else:
                         sock.sendall(pr.request)
                         extra = 10  # New connections get 10 extra seconds
-                except Exception:
-                    logging.exception('Error in recv() of %s', sock)
+                except Exception as e:
+                    logging.warning("Conection broken while writing (%s)", e)
                     continue
                 rsocks.append(sock)
                 wsocks.remove(sock)
 
             for sock in r:
                 try:
                     reply = self.__handle_recv(sock, read_buffers)
-                except Exception:
-                    logging.exception('Error in recv() of %s', sock)
+                except Exception as e:
+                    logging.warning("Connection broken while reading (%s)", e)
                     if self.sock_type(sock) == socket.SOCK_STREAM:
                         # Remove broken TCP socket from readers
                         rsocks.remove(sock)