Remove fake prev events from historical state chain

Fix #11091

We have to allow creation of events with no prev_events
but do have auth_events.

And since the historical member events are outliers
with no prev_events to resolve them, we want to avoid
putting them as backward extremeties.

synapse/federation/federation_server.py

@@ -200,6 +200,8 @@ async def on_backfill_request(
 
             res = self._transaction_dict_from_pdus(pdus)
 
+            logger.info("on_backfill_request res=%s", res)
+
         return 200, res
 
     async def on_incoming_transaction(

synapse/handlers/federation.py

@@ -1061,6 +1061,7 @@ async def on_backfill_request(
         events = await self.store.get_backfill_events(room_id, pdu_list, limit)
 
         events = await filter_events_for_server(self.storage, origin, events)
+        logger.info("on_backfill_request resultant events(%d)=%s", len(events), events)
 
         return events
 

synapse/handlers/federation_event.py

@@ -417,7 +417,8 @@ async def backfill(
             dest, room_id, limit=limit, extremities=extremities
         )
         logger.info(
-            "from remote server: got backfill response events=%s",
+            "from remote server: got backfill response events(%d)=%s",
+            len(events),
             [
                 {
                     "event_id": ev.event_id,

synapse/handlers/message.py

@@ -955,8 +955,10 @@ async def create_new_client_event(
         # event and then try to auth it (which fails with a somewhat confusing "No
         # create event in auth events")
         assert (
-            builder.type == EventTypes.Create or len(prev_event_ids) > 0
-        ), "Attempting to create an event with no prev_events"
+            builder.type == EventTypes.Create
+            or len(prev_event_ids) > 0
+            or len(auth_event_ids) > 0
+        ), "Attempting to create an event with no prev_events or auth_event_ids"
 
         event = await builder.build(
             prev_event_ids=prev_event_ids,

synapse/handlers/room_batch.py

@@ -184,7 +184,7 @@ async def persist_state_events_at_start(
 
         # Make the state events float off on their own so we don't have a
         # bunch of `@mxid joined the room` noise between each batch
-        prev_event_ids_for_state_chain = [generate_fake_event_id()]
+        prev_event_ids_for_state_chain = []  # generate_fake_event_id()
 
         for state_event in state_events_at_start:
             assert_params_in_dict(
@@ -227,6 +227,15 @@ async def persist_state_events_at_start(
                     # reference and also update in the event when we append later.
                     auth_event_ids=auth_event_ids.copy(),
                 )
+
+                mem_event = await self.store.get_event(event_id)
+                logger.info(
+                    "room_batch mem_event_id=%s depth=%s stream_ordering=%s prev_event_ids=%s",
+                    mem_event.event_id,
+                    mem_event.depth,
+                    mem_event.internal_metadata.stream_ordering,
+                    mem_event.prev_event_ids(),
+                )
             else:
                 # TODO: Add some complement tests that adds state that is not member joins
                 # and will use this code path. Maybe we only want to support join state events

synapse/handlers/room_member.py

@@ -644,7 +644,7 @@ async def update_membership_locked(
             if block_invite:
                 raise SynapseError(403, "Invites have been disabled on this server")
 
-        if prev_event_ids:
+        if prev_event_ids is not None:
             return await self._local_membership_update(
                 requester=requester,
                 target=target,

synapse/storage/databases/main/event_federation.py

@@ -1127,7 +1127,7 @@ def _get_backfill_events(self, txn, room_id, event_list, limit):
                     ev["type"],
                     ev["depth"],
                     event_lookup_result["stream_ordering"],
-                    ev["content"].get("body", None),
+                    ev["content"].get("body", ev["content"]),
                 )
             else:
                 logger.info(

synapse/storage/databases/main/events.py

@@ -2139,6 +2139,38 @@ def _update_backward_extremeties(self, txn, events):
 
         Forward extremities are handled when we first start persisting the events.
         """
+        logger.info(
+            "_update_backward_extremeties events=%s",
+            [
+                {
+                    "event_id": ev.event_id,
+                    "prev_events": ev.prev_event_ids(),
+                    "outlier": ev.internal_metadata.is_outlier(),
+                }
+                for ev in events
+            ],
+        )
+
+        for ev in events:
+            for e_id in ev.prev_event_ids():
+                query = """
+                        SELECT 1 FROM event_edges
+                        INNER JOIN events AS e USING (event_id, room_id)
+                        WHERE event_id = ? AND room_id = ? AND e.outlier = TRUE
+                """
+
+                txn.execute(
+                    query,
+                    (e_id, ev.room_id),
+                )
+                result = txn.fetchall()
+                logger.info(
+                    "_update_backward_extremeties test ev=%s prev_event_id=%s result=%s",
+                    ev.event_id,
+                    e_id,
+                    result,
+                )
+
         # From the events passed in, add all of the prev events as backwards extremities.
         # Ignore any events that are already backwards extrems or outliers.
         query = (
@@ -2147,22 +2179,45 @@ def _update_backward_extremeties(self, txn, events):
             "   SELECT 1 FROM event_backward_extremities"
             "   WHERE event_id = ? AND room_id = ?"
             " )"
+            # 1. Don't add an event as a extremity again if we already persisted it
+            # as a non-outlier.
+            # 2. Don't add an outlier as an extremity if it has no prev_events
             " AND NOT EXISTS ("
-            "   SELECT 1 FROM events WHERE event_id = ? AND room_id = ? "
-            "   AND outlier = ?"
+            "   SELECT 1 FROM events"
+            "   LEFT JOIN event_edges edge"
+            "   ON edge.event_id = events.event_id"
+            "   WHERE events.event_id = ? AND events.room_id = ? AND (events.outlier = FALSE OR edge.event_id IS NULL)"
             " )"
         )
 
         txn.execute_batch(
             query,
             [
-                (e_id, ev.room_id, e_id, ev.room_id, e_id, ev.room_id, False)
+                (e_id, ev.room_id, e_id, ev.room_id, e_id, ev.room_id)
                 for ev in events
                 for e_id in ev.prev_event_ids()
                 if not ev.internal_metadata.is_outlier()
             ],
         )
 
+        for ev in events:
+            for e_id in ev.prev_event_ids():
+                query = """
+                        SELECT * FROM event_backward_extremities
+                        WHERE event_id = ? AND room_id = ?
+                """
+
+                txn.execute(
+                    query,
+                    (e_id, ev.room_id),
+                )
+                result = txn.fetchall()
+                logger.info(
+                    "_update_backward_extremeties ended up as prev_event_id=%s result=%s",
+                    e_id,
+                    result,
+                )
+
         # Delete all these events that we've already fetched and now know that their
         # prev events are the new backwards extremeties.
         query = (
@@ -2175,6 +2230,10 @@ def _update_backward_extremeties(self, txn, events):
                 (ev.event_id, ev.room_id)
                 for ev in events
                 if not ev.internal_metadata.is_outlier()
+                # If we encountered an event with no prev_events, then we might
+                # as well remove it now because it won't ever have anything else
+                # to backfill from.
+                or len(ev.prev_event_ids()) == 0
             ],
         )