Support any process writing to cache invalidation stream. (#7436)

changelog.d/7436.misc

@@ -0,0 +1 @@
+Support any process writing to cache invalidation stream.

docs/tcp_replication.md

@@ -219,10 +219,6 @@ Asks the server for the current position of all streams.
 
    Inform the server a pusher should be removed
 
-#### INVALIDATE_CACHE (C)
-
-   Inform the server a cache should be invalidated
-
 ### REMOTE_SERVER_UP (S, C)
 
    Inform other processes that a remote server may have come back online.

scripts/synapse_port_db

@@ -122,7 +122,7 @@ APPEND_ONLY_TABLES = [
     "presence_stream",
     "push_rules_stream",
     "ex_outlier_stream",
-    "cache_invalidation_stream",
+    "cache_invalidation_stream_by_instance",
     "public_room_list_stream",
     "state_group_edges",
     "stream_ordering_to_exterm",
@@ -188,7 +188,7 @@ class MockHomeserver:
         self.clock = Clock(reactor)
         self.config = config
         self.hostname = config.server_name
-        self.version_string = "Synapse/"+get_version_string(synapse)
+        self.version_string = "Synapse/" + get_version_string(synapse)
 
     def get_clock(self):
         return self.clock

synapse/replication/slave/storage/_base.py

@@ -18,14 +18,10 @@
 
 import six
 
-from synapse.storage.data_stores.main.cache import (
-    CURRENT_STATE_CACHE_NAME,
-    CacheInvalidationWorkerStore,
-)
+from synapse.storage.data_stores.main.cache import CacheInvalidationWorkerStore
 from synapse.storage.database import Database
 from synapse.storage.engines import PostgresEngine
-
-from ._slaved_id_tracker import SlavedIdTracker
+from synapse.storage.util.id_generators import MultiWriterIdGenerator
 
 logger = logging.getLogger(__name__)
 
@@ -41,40 +37,16 @@ class BaseSlavedStore(CacheInvalidationWorkerStore):
     def __init__(self, database: Database, db_conn, hs):
         super(BaseSlavedStore, self).__init__(database, db_conn, hs)
         if isinstance(self.database_engine, PostgresEngine):
-            self._cache_id_gen = SlavedIdTracker(
-                db_conn, "cache_invalidation_stream", "stream_id"
-            )  # type: Optional[SlavedIdTracker]
+            self._cache_id_gen = MultiWriterIdGenerator(
+                db_conn,
+                database,
+                instance_name=hs.get_instance_name(),
+                table="cache_invalidation_stream_by_instance",
+                instance_column="instance_name",
+                id_column="stream_id",
+                sequence_name="cache_invalidation_stream_seq",
+            )  # type: Optional[MultiWriterIdGenerator]
         else:
             self._cache_id_gen = None
 
         self.hs = hs
-
-    def get_cache_stream_token(self):
-        if self._cache_id_gen:
-            return self._cache_id_gen.get_current_token()
-        else:
-            return 0
-
-    def process_replication_rows(self, stream_name, token, rows):
-        if stream_name == "caches":
-            if self._cache_id_gen:
-                self._cache_id_gen.advance(token)
-            for row in rows:
-                if row.cache_func == CURRENT_STATE_CACHE_NAME:
-                    if row.keys is None:
-                        raise Exception(
-                            "Can't send an 'invalidate all' for current state cache"
-                        )
-
-                    room_id = row.keys[0]
-                    members_changed = set(row.keys[1:])
-                    self._invalidate_state_caches(room_id, members_changed)
-                else:
-                    self._attempt_to_invalidate_cache(row.cache_func, row.keys)
-
-    def _invalidate_cache_and_stream(self, txn, cache_func, keys):
-        txn.call_after(cache_func.invalidate, keys)
-        txn.call_after(self._send_invalidation_poke, cache_func, keys)
-
-    def _send_invalidation_poke(self, cache_func, keys):
-        self.hs.get_tcp_replication().send_invalidate_cache(cache_func, keys)

synapse/replication/slave/storage/account_data.py

@@ -32,7 +32,7 @@ def __init__(self, database: Database, db_conn, hs):
     def get_max_account_data_stream_id(self):
         return self._account_data_id_gen.get_current_token()
 
-    def process_replication_rows(self, stream_name, token, rows):
+    def process_replication_rows(self, stream_name, instance_name, token, rows):
         if stream_name == "tag_account_data":
             self._account_data_id_gen.advance(token)
             for row in rows:
@@ -51,6 +51,4 @@ def process_replication_rows(self, stream_name, token, rows):
                     (row.user_id, row.room_id, row.data_type)
                 )
                 self._account_data_stream_cache.entity_has_changed(row.user_id, token)
-        return super(SlavedAccountDataStore, self).process_replication_rows(
-            stream_name, token, rows
-        )
+        return super().process_replication_rows(stream_name, instance_name, token, rows)

synapse/replication/slave/storage/deviceinbox.py

@@ -43,7 +43,7 @@ def __init__(self, database: Database, db_conn, hs):
             expiry_ms=30 * 60 * 1000,
         )
 
-    def process_replication_rows(self, stream_name, token, rows):
+    def process_replication_rows(self, stream_name, instance_name, token, rows):
         if stream_name == "to_device":
             self._device_inbox_id_gen.advance(token)
             for row in rows:
@@ -55,6 +55,4 @@ def process_replication_rows(self, stream_name, token, rows):
                     self._device_federation_outbox_stream_cache.entity_has_changed(
                         row.entity, token
                     )
-        return super(SlavedDeviceInboxStore, self).process_replication_rows(
-            stream_name, token, rows
-        )
+        return super().process_replication_rows(stream_name, instance_name, token, rows)

synapse/replication/slave/storage/devices.py

@@ -48,17 +48,15 @@ def __init__(self, database: Database, db_conn, hs):
             "DeviceListFederationStreamChangeCache", device_list_max
         )
 
-    def process_replication_rows(self, stream_name, token, rows):
+    def process_replication_rows(self, stream_name, instance_name, token, rows):
         if stream_name == DeviceListsStream.NAME:
             self._device_list_id_gen.advance(token)
             self._invalidate_caches_for_devices(token, rows)
         elif stream_name == UserSignatureStream.NAME:
             self._device_list_id_gen.advance(token)
             for row in rows:
                 self._user_signature_stream_cache.entity_has_changed(row.user_id, token)
-        return super(SlavedDeviceStore, self).process_replication_rows(
-            stream_name, token, rows
-        )
+        return super().process_replication_rows(stream_name, instance_name, token, rows)
 
     def _invalidate_caches_for_devices(self, token, rows):
         for row in rows:

synapse/replication/slave/storage/events.py

@@ -93,7 +93,7 @@ def get_room_max_stream_ordering(self):
     def get_room_min_stream_ordering(self):
         return self._backfill_id_gen.get_current_token()
 
-    def process_replication_rows(self, stream_name, token, rows):
+    def process_replication_rows(self, stream_name, instance_name, token, rows):
         if stream_name == "events":
             self._stream_id_gen.advance(token)
             for row in rows:
@@ -111,9 +111,7 @@ def process_replication_rows(self, stream_name, token, rows):
                     row.relates_to,
                     backfilled=True,
                 )
-        return super(SlavedEventStore, self).process_replication_rows(
-            stream_name, token, rows
-        )
+        return super().process_replication_rows(stream_name, instance_name, token, rows)
 
     def _process_event_stream_row(self, token, row):
         data = row.data

synapse/replication/slave/storage/groups.py

@@ -37,12 +37,10 @@ def __init__(self, database: Database, db_conn, hs):
     def get_group_stream_token(self):
         return self._group_updates_id_gen.get_current_token()
 
-    def process_replication_rows(self, stream_name, token, rows):
+    def process_replication_rows(self, stream_name, instance_name, token, rows):
         if stream_name == "groups":
             self._group_updates_id_gen.advance(token)
             for row in rows:
                 self._group_updates_stream_cache.entity_has_changed(row.user_id, token)
 
-        return super(SlavedGroupServerStore, self).process_replication_rows(
-            stream_name, token, rows
-        )
+        return super().process_replication_rows(stream_name, instance_name, token, rows)

synapse/replication/slave/storage/presence.py

@@ -41,12 +41,10 @@ def __init__(self, database: Database, db_conn, hs):
     def get_current_presence_token(self):
         return self._presence_id_gen.get_current_token()
 
-    def process_replication_rows(self, stream_name, token, rows):
+    def process_replication_rows(self, stream_name, instance_name, token, rows):
         if stream_name == "presence":
             self._presence_id_gen.advance(token)
             for row in rows:
                 self.presence_stream_cache.entity_has_changed(row.user_id, token)
                 self._get_presence_for_user.invalidate((row.user_id,))
-        return super(SlavedPresenceStore, self).process_replication_rows(
-            stream_name, token, rows
-        )
+        return super().process_replication_rows(stream_name, instance_name, token, rows)

synapse/replication/slave/storage/push_rule.py

@@ -37,13 +37,11 @@ def get_push_rules_stream_token(self):
     def get_max_push_rules_stream_id(self):
         return self._push_rules_stream_id_gen.get_current_token()
 
-    def process_replication_rows(self, stream_name, token, rows):
+    def process_replication_rows(self, stream_name, instance_name, token, rows):
         if stream_name == "push_rules":
             self._push_rules_stream_id_gen.advance(token)
             for row in rows:
                 self.get_push_rules_for_user.invalidate((row.user_id,))
                 self.get_push_rules_enabled_for_user.invalidate((row.user_id,))
                 self.push_rules_stream_cache.entity_has_changed(row.user_id, token)
-        return super(SlavedPushRuleStore, self).process_replication_rows(
-            stream_name, token, rows
-        )
+        return super().process_replication_rows(stream_name, instance_name, token, rows)

synapse/replication/slave/storage/pushers.py

@@ -31,9 +31,7 @@ def __init__(self, database: Database, db_conn, hs):
     def get_pushers_stream_token(self):
         return self._pushers_id_gen.get_current_token()
 
-    def process_replication_rows(self, stream_name, token, rows):
+    def process_replication_rows(self, stream_name, instance_name, token, rows):
         if stream_name == "pushers":
             self._pushers_id_gen.advance(token)
-        return super(SlavedPusherStore, self).process_replication_rows(
-            stream_name, token, rows
-        )
+        return super().process_replication_rows(stream_name, instance_name, token, rows)

synapse/replication/slave/storage/receipts.py

@@ -51,7 +51,7 @@ def invalidate_caches_for_receipt(self, room_id, receipt_type, user_id):
         self._invalidate_get_users_with_receipts_in_room(room_id, receipt_type, user_id)
         self.get_receipts_for_room.invalidate((room_id, receipt_type))
 
-    def process_replication_rows(self, stream_name, token, rows):
+    def process_replication_rows(self, stream_name, instance_name, token, rows):
         if stream_name == "receipts":
             self._receipts_id_gen.advance(token)
             for row in rows:
@@ -60,6 +60,4 @@ def process_replication_rows(self, stream_name, token, rows):
                 )
                 self._receipts_stream_cache.entity_has_changed(row.room_id, token)
 
-        return super(SlavedReceiptsStore, self).process_replication_rows(
-            stream_name, token, rows
-        )
+        return super().process_replication_rows(stream_name, instance_name, token, rows)

synapse/replication/slave/storage/room.py

@@ -30,8 +30,8 @@ def __init__(self, database: Database, db_conn, hs):
     def get_current_public_room_stream_id(self):
         return self._public_room_id_gen.get_current_token()
 
-    def process_replication_rows(self, stream_name, token, rows):
+    def process_replication_rows(self, stream_name, instance_name, token, rows):
         if stream_name == "public_rooms":
             self._public_room_id_gen.advance(token)
 
-        return super(RoomStore, self).process_replication_rows(stream_name, token, rows)
+        return super().process_replication_rows(stream_name, instance_name, token, rows)

synapse/replication/tcp/client.py

@@ -100,10 +100,10 @@ async def on_rdata(
             token: stream token for this batch of rows
             rows: a list of Stream.ROW_TYPE objects as returned by Stream.parse_row.
         """
-        self.store.process_replication_rows(stream_name, token, rows)
+        self.store.process_replication_rows(stream_name, instance_name, token, rows)
 
-    async def on_position(self, stream_name: str, token: int):
-        self.store.process_replication_rows(stream_name, token, [])
+    async def on_position(self, stream_name: str, instance_name: str, token: int):
+        self.store.process_replication_rows(stream_name, instance_name, token, [])
 
     def on_remote_server_up(self, server: str):
         """Called when get a new REMOTE_SERVER_UP command."""

synapse/replication/tcp/commands.py

@@ -341,37 +341,6 @@ def to_line(self):
         return " ".join((self.app_id, self.push_key, self.user_id))
 
 
-class InvalidateCacheCommand(Command):
-    """Sent by the client to invalidate an upstream cache.
-
-    THIS IS NOT RELIABLE, AND SHOULD *NOT* BE USED ACCEPT FOR THINGS THAT ARE
-    NOT DISASTROUS IF WE DROP ON THE FLOOR.
-
-    Mainly used to invalidate destination retry timing caches.
-
-    Format::
-
-        INVALIDATE_CACHE <cache_func> <keys_json>
-
-    Where <keys_json> is a json list.
-    """
-
-    NAME = "INVALIDATE_CACHE"
-
-    def __init__(self, cache_func, keys):
-        self.cache_func = cache_func
-        self.keys = keys
-
-    @classmethod
-    def from_line(cls, line):
-        cache_func, keys_json = line.split(" ", 1)
-
-        return cls(cache_func, json.loads(keys_json))
-
-    def to_line(self):
-        return " ".join((self.cache_func, _json_encoder.encode(self.keys)))
-
-
 class UserIpCommand(Command):
     """Sent periodically when a worker sees activity from a client.
 
@@ -439,7 +408,6 @@ class RemoteServerUpCommand(_SimpleCommand):
     UserSyncCommand,
     FederationAckCommand,
     RemovePusherCommand,
-    InvalidateCacheCommand,
     UserIpCommand,
     RemoteServerUpCommand,
     ClearUserSyncsCommand,
@@ -467,7 +435,6 @@ class RemoteServerUpCommand(_SimpleCommand):
     ClearUserSyncsCommand.NAME,
     FederationAckCommand.NAME,
     RemovePusherCommand.NAME,
-    InvalidateCacheCommand.NAME,
     UserIpCommand.NAME,
     ErrorCommand.NAME,
     RemoteServerUpCommand.NAME,