Merge branch 'develop' of github.com:matrix-org/synapse into use-simplejson

synapse/storage/__init__.py

@@ -163,6 +163,9 @@ class DataStore(RoomMemberStore, RoomStore,
                            stream_ordering=None, is_new_state=True,
                            current_state=None):
 
+        # Remove the any existing cache entries for the event_id
+        self._get_event_cache.pop(event.event_id)
+
         # We purposefully do this first since if we include a `current_state`
         # key, we *want* to update the `current_state_events` table
         if current_state:
@@ -426,6 +429,8 @@ class DataStore(RoomMemberStore, RoomStore,
         )
 
     def _store_redaction(self, txn, event):
+        # invalidate the cache for the redacted event
+        self._get_event_cache.pop(event.redacts)
         txn.execute(
             "INSERT OR IGNORE INTO redactions "
             "(event_id, redacts) VALUES (?,?)",

synapse/storage/_base.py

@@ -19,6 +19,7 @@ from synapse.events import FrozenEvent
 from synapse.events.utils import prune_event
 from synapse.util.logutils import log_function
 from synapse.util.logcontext import PreserveLoggingContext, LoggingContext
+from synapse.util.lrucache import LruCache
 
 from twisted.internet import defer
 
@@ -128,6 +129,8 @@ class SQLBaseStore(object):
         self._txn_perf_counters = PerformanceCounters()
         self._get_event_counters = PerformanceCounters()
 
+        self._get_event_cache = LruCache(hs.config.event_cache_size)
+
     def start_profiling(self):
         self._previous_loop_ts = self._clock.time_msec()
 
@@ -579,6 +582,19 @@ class SQLBaseStore(object):
 
     def _get_event_txn(self, txn, event_id, check_redacted=True,
                        get_prev_content=False, allow_rejected=False):
+
+        start_time = time.time() * 1000
+        update_counter = self._get_event_counters.update
+
+        try:
+            cache = self._get_event_cache.setdefault(event_id, {})
+            # Separate cache entries for each way to invoke _get_event_txn
+            return cache[(check_redacted, get_prev_content, allow_rejected)]
+        except KeyError:
+            pass
+        finally:
+            start_time = update_counter("event_cache", start_time)
+
         sql = (
             "SELECT e.internal_metadata, e.json, r.event_id, rej.reason "
             "FROM event_json as e "
@@ -588,8 +604,6 @@ class SQLBaseStore(object):
             "LIMIT 1 "
         )
 
-        start_time = time.time() * 1000
-
         txn.execute(sql, (event_id,))
 
         res = txn.fetchone()
@@ -599,14 +613,16 @@ class SQLBaseStore(object):
 
         internal_metadata, js, redacted, rejected_reason = res
 
-        self._get_event_counters.update("select_event", start_time)
+        start_time = update_counter("select_event", start_time)
 
         if allow_rejected or not rejected_reason:
-            return self._get_event_from_row_txn(
+            result = self._get_event_from_row_txn(
                 txn, internal_metadata, js, redacted,
                 check_redacted=check_redacted,
                 get_prev_content=get_prev_content,
             )
+            cache[(check_redacted, get_prev_content, allow_rejected)] = result
+            return result
         else:
             return None
 

synapse/storage/roommember.py

@@ -35,6 +35,11 @@ RoomsForUser = namedtuple(
 
 class RoomMemberStore(SQLBaseStore):
 
+    def __init__(self, *args, **kw):
+        super(RoomMemberStore, self).__init__(*args, **kw)
+
+        self._user_rooms_cache = {}
+
     def _store_room_member_txn(self, txn, event):
         """Store a room member in the database.
         """
@@ -98,6 +103,8 @@ class RoomMemberStore(SQLBaseStore):
 
                 txn.execute(sql, (event.room_id, domain))
 
+        self.invalidate_rooms_for_user(target_user_id)
+
     @defer.inlineCallbacks
     def get_room_member(self, user_id, room_id):
         """Retrieve the current state of a room member.
@@ -240,28 +247,53 @@ class RoomMemberStore(SQLBaseStore):
         results = self._parse_events_txn(txn, rows)
         return results
 
+    # TODO(paul): Create a nice @cached decorator to do this
+    #    @cached
+    #    def get_foo(...)
+    #        ...
+    #    invalidate_foo = get_foo.invalidator
+
+    @defer.inlineCallbacks
+    def get_rooms_for_user(self, user_id):
+        # TODO(paul): put some performance counters in here so we can easily
+        #   track what impact this cache is having
+        if user_id in self._user_rooms_cache:
+            defer.returnValue(self._user_rooms_cache[user_id])
+
+        rooms = yield self.get_rooms_for_user_where_membership_is(
+            user_id, membership_list=[Membership.JOIN],
+        )
+
+        # TODO(paul): Consider applying a maximum size; just evict things at
+        #   random, or consider LRU?
+
+        self._user_rooms_cache[user_id] = rooms
+        defer.returnValue(rooms)
+
+    def invalidate_rooms_for_user(self, user_id):
+        if user_id in self._user_rooms_cache:
+            del self._user_rooms_cache[user_id]
+
+    @defer.inlineCallbacks
     def user_rooms_intersect(self, user_id_list):
         """ Checks whether all the users whose IDs are given in a list share a
         room.
+
+        This is a "hot path" function that's called a lot, e.g. by presence for
+        generating the event stream. As such, it is implemented locally by
+        wrapping logic around heavily-cached database queries.
         """
-        def interaction(txn):
-            user_list_clause = " OR ".join(["m.user_id = ?"] * len(user_id_list))
-            sql = (
-                "SELECT m.room_id FROM room_memberships as m "
-                "INNER JOIN current_state_events as c "
-                "ON m.event_id = c.event_id "
-                "WHERE m.membership = 'join' "
-                "AND (%(clause)s) "
-                # TODO(paul): We've got duplicate rows in the database somewhere
-                #   so we have to DISTINCT m.user_id here
-                "GROUP BY m.room_id HAVING COUNT(DISTINCT m.user_id) = ?"
-            ) % {"clause": user_list_clause}
+        if len(user_id_list) < 2:
+            defer.returnValue(True)
 
-            args = list(user_id_list)
-            args.append(len(user_id_list))
+        deferreds = [self.get_rooms_for_user(u) for u in user_id_list]
 
-            txn.execute(sql, args)
+        results = yield defer.DeferredList(deferreds)
 
-            return len(txn.fetchall()) > 0
+        # A list of sets of strings giving room IDs for each user
+        room_id_lists = [set([r.room_id for r in result[1]]) for result in results]
 
-        return self.runInteraction("user_rooms_intersect", interaction)
+        # There isn't a setintersection(*list_of_sets)
+        ret = len(room_id_lists.pop(0).intersection(*room_id_lists)) > 0
+
+        defer.returnValue(ret)