Add `event.internal_metadata.instance_name` (#17300)

Add `event.internal_metadata.instance_name` (the worker instance that persisted the event) to go alongside the existing `event.internal_metadata.stream_ordering`.

`instance_name` is useful to properly compare and query for events with a token since you need to compare both the `stream_ordering` and `instance_name` against the vector clock/`instance_map` in the `RoomStreamToken`.

This is pre-requisite work and may be used in https://github.com/element-hq/synapse/pull/17293

Adding `event.internal_metadata.instance_name` was first mentioned in the initial Sliding Sync PR while pairing with @erikjohnston, see https://github.com/element-hq/synapse/pull/17187/commits/09609cb0dbca3a4cfd9fbf90cc962e765ec469c0#diff-5cd773fb307aa754bd3948871ba118b1ef0303f4d72d42a2d21e38242bf4e096R405-R410

5 files changed, 16 insertions, 6 deletions

diff --git a/synapse/events/utils.py b/synapse/events/utils.py
index 0772472312..b997d82d71 100644
--- a/synapse/events/utils.py
+++ b/synapse/events/utils.py
@@ -90,6 +90,7 @@ def prune_event(event: EventBase) -> EventBase:
     pruned_event.internal_metadata.stream_ordering = (
         event.internal_metadata.stream_ordering
     )
+    pruned_event.internal_metadata.instance_name = event.internal_metadata.instance_name
     pruned_event.internal_metadata.outlier = event.internal_metadata.outlier
 
     # Mark the event as redacted
@@ -116,6 +117,7 @@ def clone_event(event: EventBase) -> EventBase:
     new_event.internal_metadata.stream_ordering = (
         event.internal_metadata.stream_ordering
     )
+    new_event.internal_metadata.instance_name = event.internal_metadata.instance_name
     new_event.internal_metadata.outlier = event.internal_metadata.outlier
 
     return new_event
diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py
index de5bd44a5f..721ef04f41 100644
--- a/synapse/handlers/message.py
+++ b/synapse/handlers/message.py
@@ -1551,6 +1551,7 @@ class EventCreationHandler:
                     # stream_ordering entry manually (as it was persisted on
                     # another worker).
                     event.internal_metadata.stream_ordering = stream_id
+                    event.internal_metadata.instance_name = writer_instance
 
                 return event
 
diff --git a/synapse/storage/databases/main/events.py b/synapse/storage/databases/main/events.py
index f1bd85aa27..66428e6c8e 100644
--- a/synapse/storage/databases/main/events.py
+++ b/synapse/storage/databases/main/events.py
@@ -207,6 +207,7 @@ class PersistEventsStore:
         async with stream_ordering_manager as stream_orderings:
             for (event, _), stream in zip(events_and_contexts, stream_orderings):
                 event.internal_metadata.stream_ordering = stream
+                event.internal_metadata.instance_name = self._instance_name
 
             await self.db_pool.runInteraction(
                 "persist_events",
diff --git a/synapse/storage/databases/main/events_worker.py b/synapse/storage/databases/main/events_worker.py
index c06c44deb1..e264d36f02 100644
--- a/synapse/storage/databases/main/events_worker.py
+++ b/synapse/storage/databases/main/events_worker.py
@@ -156,6 +156,7 @@ class _EventRow:
 
     event_id: str
     stream_ordering: int
+    instance_name: str
     json: str
     internal_metadata: str
     format_version: Optional[int]
@@ -1354,6 +1355,7 @@ class EventsWorkerStore(SQLBaseStore):
                 rejected_reason=rejected_reason,
             )
             original_ev.internal_metadata.stream_ordering = row.stream_ordering
+            original_ev.internal_metadata.instance_name = row.instance_name
             original_ev.internal_metadata.outlier = row.outlier
 
             # Consistency check: if the content of the event has been modified in the
@@ -1439,6 +1441,7 @@ class EventsWorkerStore(SQLBaseStore):
                 SELECT
                   e.event_id,
                   e.stream_ordering,
+                  e.instance_name,
                   ej.internal_metadata,
                   ej.json,
                   ej.format_version,
@@ -1462,13 +1465,14 @@ class EventsWorkerStore(SQLBaseStore):
                 event_dict[event_id] = _EventRow(
                     event_id=event_id,
                     stream_ordering=row[1],
-                    internal_metadata=row[2],
-                    json=row[3],
-                    format_version=row[4],
-                    room_version_id=row[5],
-                    rejected_reason=row[6],
+                    instance_name=row[2],
+                    internal_metadata=row[3],
+                    json=row[4],
+                    format_version=row[5],
+                    room_version_id=row[6],
+                    rejected_reason=row[7],
                     redactions=[],
-                    outlier=bool(row[7]),  # This is an int in SQLite3
+                    outlier=bool(row[8]),  # This is an int in SQLite3
                 )
 
             # check for redactions
diff --git a/synapse/synapse_rust/events.pyi b/synapse/synapse_rust/events.pyi
index 69837617f5..1682d0d151 100644
--- a/synapse/synapse_rust/events.pyi
+++ b/synapse/synapse_rust/events.pyi
@@ -19,6 +19,8 @@ class EventInternalMetadata:
 
     stream_ordering: Optional[int]
     """the stream ordering of this event. None, until it has been persisted."""
+    instance_name: Optional[str]
+    """the instance name of the server that persisted this event. None, until it has been persisted."""
 
     outlier: bool
     """whether this event is an outlier (ie, whether we have the state at that