Merge pull request #114 from matrix-org/improve_get_event_cache

Improve get event cache

synapse/events/__init__.py

@@ -46,9 +46,10 @@ def _event_dict_property(key):
 
 class EventBase(object):
     def __init__(self, event_dict, signatures={}, unsigned={},
-                 internal_metadata_dict={}):
+                 internal_metadata_dict={}, rejected_reason=None):
         self.signatures = signatures
         self.unsigned = unsigned
+        self.rejected_reason = rejected_reason
 
         self._event_dict = event_dict
 
@@ -109,7 +110,7 @@ class EventBase(object):
 
 
 class FrozenEvent(EventBase):
-    def __init__(self, event_dict, internal_metadata_dict={}):
+    def __init__(self, event_dict, internal_metadata_dict={}, rejected_reason=None):
         event_dict = dict(event_dict)
 
         # Signatures is a dict of dicts, and this is faster than doing a
@@ -128,6 +129,7 @@ class FrozenEvent(EventBase):
             signatures=signatures,
             unsigned=unsigned,
             internal_metadata_dict=internal_metadata_dict,
+            rejected_reason=rejected_reason,
         )
 
     @staticmethod

synapse/storage/_base.py

@@ -242,10 +242,8 @@ class SQLBaseStore(object):
         self._txn_perf_counters = PerformanceCounters()
         self._get_event_counters = PerformanceCounters()
 
-        self._get_event_cache = LruCache(hs.config.event_cache_size)
-
-        # Pretend the getEventCache is just another named cache
-        caches_by_name["*getEvent*"] = self._get_event_cache
+        self._get_event_cache = Cache("*getEvent*", keylen=3, lru=True,
+                                      max_entries=hs.config.event_cache_size)
 
     def start_profiling(self):
         self._previous_loop_ts = self._clock.time_msec()
@@ -733,6 +731,12 @@ class SQLBaseStore(object):
 
         return [e for e in events if e]
 
+    def _invalidate_get_event_cache(self, event_id):
+        for check_redacted in (False, True):
+            for get_prev_content in (False, True):
+                self._get_event_cache.invalidate(event_id, check_redacted,
+                                                 get_prev_content)
+
     def _get_event_txn(self, txn, event_id, check_redacted=True,
                        get_prev_content=False, allow_rejected=False):
 
@@ -743,16 +747,14 @@ class SQLBaseStore(object):
             sql_getevents_timer.inc_by(curr_time - last_time, desc)
             return curr_time
 
-        cache = self._get_event_cache.setdefault(event_id, {})
-
         try:
-            # Separate cache entries for each way to invoke _get_event_txn
-            ret = cache[(check_redacted, get_prev_content, allow_rejected)]
+            ret = self._get_event_cache.get(event_id, check_redacted, get_prev_content)
 
-            cache_counter.inc_hits("*getEvent*")
-            return ret
+            if allow_rejected or not ret.rejected_reason:
+                return ret
+            else:
+                return None
         except KeyError:
-            cache_counter.inc_misses("*getEvent*")
             pass
         finally:
             start_time = update_counter("event_cache", start_time)
@@ -777,19 +779,22 @@ class SQLBaseStore(object):
 
         start_time = update_counter("select_event", start_time)
 
+        result = self._get_event_from_row_txn(
+            txn, internal_metadata, js, redacted,
+            check_redacted=check_redacted,
+            get_prev_content=get_prev_content,
+            rejected_reason=rejected_reason,
+        )
+        self._get_event_cache.prefill(event_id, check_redacted, get_prev_content, result)
+
         if allow_rejected or not rejected_reason:
-            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
 
     def _get_event_from_row_txn(self, txn, internal_metadata, js, redacted,
-                                check_redacted=True, get_prev_content=False):
+                                check_redacted=True, get_prev_content=False,
+                                rejected_reason=None):
 
         start_time = time.time() * 1000
 
@@ -804,7 +809,11 @@ class SQLBaseStore(object):
         internal_metadata = json.loads(internal_metadata)
         start_time = update_counter("decode_internal", start_time)
 
-        ev = FrozenEvent(d, internal_metadata_dict=internal_metadata)
+        ev = FrozenEvent(
+            d,
+            internal_metadata_dict=internal_metadata,
+            rejected_reason=rejected_reason,
+        )
         start_time = update_counter("build_frozen_event", start_time)
 
         if check_redacted and redacted:

synapse/storage/events.py

@@ -94,7 +94,7 @@ class EventsStore(SQLBaseStore):
                            current_state=None):
 
         # Remove the any existing cache entries for the event_id
-        self._get_event_cache.pop(event.event_id)
+        self._invalidate_get_event_cache(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
@@ -356,7 +356,7 @@ class EventsStore(SQLBaseStore):
 
     def _store_redaction(self, txn, event):
         # invalidate the cache for the redacted event
-        self._get_event_cache.pop(event.redacts)
+        self._invalidate_get_event_cache(event.redacts)
         txn.execute(
             "INSERT INTO redactions (event_id, redacts) VALUES (?,?)",
             (event.event_id, event.redacts)