actually reset retry schedule if we can successfuly talk to it

synapse/federation/replication.py

@@ -864,6 +864,9 @@ class _TransactionQueue(object):
 
             for deferred in deferreds:
                 if code == 200:
+                    if retry_last_ts:
+                        # this host is alive! reset retry schedule
+                        self.store.set_destination_retry_timings(destination, 0, 0)
                     deferred.callback(None)
                 else:
                     self.start_retrying(destination, retry_interval)

synapse/handlers/federation.py

@@ -208,6 +208,12 @@ class FederationHandler(BaseHandler):
                 affected=event.event_id,
             )
             
+        # if we're receiving valid events from an origin,
+        # it's probably a good idea to mark it as not in retry-state
+        # for sending (although this is a bit of a leap)
+        if ((self.store.get_destination_retry_timings(origin))[0]):
+            self.store.set_destination_retry_timings(origin, 0, 0)
+
         room = yield self.store.get_room(event.room_id)
 
         if not room: