Avoid resetting state on rejected events

When we get a rejected event, give it the same state_group as its prev_event,
rather than no state_group at all.

This should fix https://github.com/matrix-org/synapse/issues/1935.

1 files changed, 10 insertions, 0 deletions

diff --git a/synapse/storage/state.py b/synapse/storage/state.py

index 27f1ec89ec..1b42bea07a 100644
--- a/synapse/storage/state.py
+++ b/synapse/storage/state.py
@@ -136,6 +136,16 @@ class StateStore(SQLBaseStore):
                 continue
 
             if context.current_state_ids is None:
+                # AFAIK, this can never happen
+                logger.error(
+                    "Non-outlier event %s had current_state_ids==None",
+                    event.event_id)
+                continue
+
+            # if the event was rejected, just give it the same state as its
+            # predecessor.
+            if context.rejected:
+                state_groups[event.event_id] = context.prev_group
                 continue
 
             state_groups[event.event_id] = context.state_group