Update all stream IDs after processing replication rows (#14723)

This creates a new store method, `process_replication_position` that
is called after `process_replication_rows`. By moving stream ID advances
here this guarantees any relevant cache invalidations will have been
applied before the stream is advanced.

This avoids race conditions where Python switches between threads mid
way through processing the `process_replication_rows` method where stream
IDs may be advanced before caches are invalidated due to class resolution
ordering.

See this comment/issue for further discussion:
	https://github.com/matrix-org/synapse/issues/14158#issuecomment-1344048703

1 files changed, 7 insertions, 0 deletions

diff --git a/synapse/storage/databases/main/deviceinbox.py b/synapse/storage/databases/main/deviceinbox.py
index 48a54d9cb8..713be91c5d 100644
--- a/synapse/storage/databases/main/deviceinbox.py
+++ b/synapse/storage/databases/main/deviceinbox.py
@@ -157,6 +157,13 @@ class DeviceInboxWorkerStore(SQLBaseStore):
                     )
         return super().process_replication_rows(stream_name, instance_name, token, rows)
 
+    def process_replication_position(
+        self, stream_name: str, instance_name: str, token: int
+    ) -> None:
+        if stream_name == ToDeviceStream.NAME:
+            self._device_inbox_id_gen.advance(instance_name, token)
+        super().process_replication_position(stream_name, instance_name, token)
+
     def get_to_device_stream_token(self) -> int:
         return self._device_inbox_id_gen.get_current_token()