Add experimental support for sharding event persister. (#8170)

This is *not* ready for production yet. Caveats:

1. We should write some tests...
2. The stream token that we use for events can get stalled at the minimum position of all writers. This means that new events may not be processed and e.g. sent down sync streams if a writer isn't writing or is slow.

3 files changed, 12 insertions, 6 deletions

diff --git a/synapse/replication/http/federation.py b/synapse/replication/http/federation.py
index 6b56315148..5c8be747e1 100644
--- a/synapse/replication/http/federation.py
+++ b/synapse/replication/http/federation.py
@@ -65,10 +65,11 @@ class ReplicationFederationSendEventsRestServlet(ReplicationEndpoint):
         self.federation_handler = hs.get_handlers().federation_handler
 
     @staticmethod
-    async def _serialize_payload(store, event_and_contexts, backfilled):
+    async def _serialize_payload(store, room_id, event_and_contexts, backfilled):
         """
         Args:
             store
+            room_id (str)
             event_and_contexts (list[tuple[FrozenEvent, EventContext]])
             backfilled (bool): Whether or not the events are the result of
                 backfilling
@@ -88,7 +89,11 @@ class ReplicationFederationSendEventsRestServlet(ReplicationEndpoint):
                 }
             )
 
-        payload = {"events": event_payloads, "backfilled": backfilled}
+        payload = {
+            "events": event_payloads,
+            "backfilled": backfilled,
+            "room_id": room_id,
+        }
 
         return payload
 
@@ -96,6 +101,7 @@ class ReplicationFederationSendEventsRestServlet(ReplicationEndpoint):
         with Measure(self.clock, "repl_fed_send_events_parse"):
             content = parse_json_object_from_request(request)
 
+            room_id = content["room_id"]
             backfilled = content["backfilled"]
 
             event_payloads = content["events"]
@@ -120,7 +126,7 @@ class ReplicationFederationSendEventsRestServlet(ReplicationEndpoint):
         logger.info("Got %d events from federation", len(event_and_contexts))
 
         max_stream_id = await self.federation_handler.persist_events_and_notify(
-            event_and_contexts, backfilled
+            room_id, event_and_contexts, backfilled
         )
 
         return 200, {"max_stream_id": max_stream_id}
diff --git a/synapse/replication/tcp/handler.py b/synapse/replication/tcp/handler.py
index 1c303f3a46..b323841f73 100644
--- a/synapse/replication/tcp/handler.py
+++ b/synapse/replication/tcp/handler.py
@@ -109,7 +109,7 @@ class ReplicationCommandHandler:
             if isinstance(stream, (EventsStream, BackfillStream)):
                 # Only add EventStream and BackfillStream as a source on the
                 # instance in charge of event persistence.
-                if hs.config.worker.writers.events == hs.get_instance_name():
+                if hs.get_instance_name() in hs.config.worker.writers.events:
                     self._streams_to_replicate.append(stream)
 
                 continue
diff --git a/synapse/replication/tcp/streams/events.py b/synapse/replication/tcp/streams/events.py
index 16c63ff4ec..3705618b4f 100644
--- a/synapse/replication/tcp/streams/events.py
+++ b/synapse/replication/tcp/streams/events.py
@@ -19,7 +19,7 @@ from typing import List, Tuple, Type
 
 import attr
 
-from ._base import Stream, StreamUpdateResult, Token, current_token_without_instance
+from ._base import Stream, StreamUpdateResult, Token
 
 """Handling of the 'events' replication stream
 
@@ -117,7 +117,7 @@ class EventsStream(Stream):
         self._store = hs.get_datastore()
         super().__init__(
             hs.get_instance_name(),
-            current_token_without_instance(self._store.get_current_events_token),
+            self._store._stream_id_gen.get_current_token_for_writer,
             self._update_function,
         )