Reprocess rejected input events (#2647)

* Reprocess outliers that were previously rejected

* Might as well do all events this way

* More useful errors

* Fix queries

* Tweak condition

* Don't wrap errors

* Report more useful error

* Flatten error on `r.Queryer.QueryStateAfterEvents`

* Some more debug logging

* Flatten error in `QueryRestrictedJoinAllowed`

* Revert "Flatten error in `QueryRestrictedJoinAllowed`"

This reverts commit 1238b4184c30e0c31ffb0f364806fa1275aba483.

* Tweak `QueryStateAfterEvents`

* Handle MissingStateError too

* Scope to room

* Clean up

* Fix the error

* Only apply rejection check to outliers

7 files changed, 65 insertions, 28 deletions

diff --git a/roomserver/internal/input/input_events.go b/roomserver/internal/input/input_events.go
index 81541260..53ccd597 100644
--- a/roomserver/internal/input/input_events.go
+++ b/roomserver/internal/input/input_events.go
@@ -17,8 +17,8 @@
 package input
 
 import (
-	"bytes"
 	"context"
+	"database/sql"
 	"fmt"
 	"time"
 
@@ -107,28 +107,6 @@ func (r *Inputer) processRoomEvent(
 		})
 	}
 
-	// if we have already got this event then do not process it again, if the input kind is an outlier.
-	// Outliers contain no extra information which may warrant a re-processing.
-	if input.Kind == api.KindOutlier {
-		evs, err2 := r.DB.EventsFromIDs(ctx, []string{event.EventID()})
-		if err2 == nil && len(evs) == 1 {
-			// check hash matches if we're on early room versions where the event ID was a random string
-			idFormat, err2 := headered.RoomVersion.EventIDFormat()
-			if err2 == nil {
-				switch idFormat {
-				case gomatrixserverlib.EventIDFormatV1:
-					if bytes.Equal(event.EventReference().EventSHA256, evs[0].EventReference().EventSHA256) {
-						logger.Debugf("Already processed event; ignoring")
-						return nil
-					}
-				default:
-					logger.Debugf("Already processed event; ignoring")
-					return nil
-				}
-			}
-		}
-	}
-
 	// Don't waste time processing the event if the room doesn't exist.
 	// A room entry locally will only be created in response to a create
 	// event.
@@ -141,6 +119,29 @@ func (r *Inputer) processRoomEvent(
 		return fmt.Errorf("room %s does not exist for event %s", event.RoomID(), event.EventID())
 	}
 
+	// If we already know about this outlier and it hasn't been rejected
+	// then we won't attempt to reprocess it. If it was rejected or has now
+	// arrived as a different kind of event, then we can attempt to reprocess,
+	// in case we have learned something new or need to weave the event into
+	// the DAG now.
+	if input.Kind == api.KindOutlier && roomInfo != nil {
+		wasRejected, werr := r.DB.IsEventRejected(ctx, roomInfo.RoomNID, event.EventID())
+		switch {
+		case werr == sql.ErrNoRows:
+			// We haven't seen this event before so continue.
+		case werr != nil:
+			// Something has gone wrong trying to find out if we rejected
+			// this event already.
+			logger.WithError(werr).Errorf("Failed to check if event %q is already seen", event.EventID())
+			return werr
+		case !wasRejected:
+			// We've seen this event before and it wasn't rejected so we
+			// should ignore it.
+			logger.Debugf("Already processed event %q, ignoring", event.EventID())
+			return nil
+		}
+	}
+
 	var missingAuth, missingPrev bool
 	serverRes := &fedapi.QueryJoinedHostServerNamesInRoomResponse{}
 	if !isCreateEvent {
diff --git a/roomserver/internal/query/query.go b/roomserver/internal/query/query.go
index f3061c22..f5d8c2d4 100644
--- a/roomserver/internal/query/query.go
+++ b/roomserver/internal/query/query.go
@@ -72,13 +72,10 @@ func (r *Queryer) QueryStateAfterEvents(
 
 	prevStates, err := r.DB.StateAtEventIDs(ctx, request.PrevEventIDs)
 	if err != nil {
-		switch err.(type) {
-		case types.MissingEventError:
-			util.GetLogger(ctx).Errorf("QueryStateAfterEvents: MissingEventError: %s", err)
+		if _, ok := err.(types.MissingEventError); ok {
 			return nil
-		default:
-			return err
 		}
+		return err
 	}
 	response.PrevEventsExist = true
 
@@ -95,6 +92,12 @@ func (r *Queryer) QueryStateAfterEvents(
 		)
 	}
 	if err != nil {
+		if _, ok := err.(types.MissingEventError); ok {
+			return nil
+		}
+		if _, ok := err.(types.MissingStateError); ok {
+			return nil
+		}
 		return err
 	}
 
diff --git a/roomserver/storage/interface.go b/roomserver/storage/interface.go
index b12025c4..5c068873 100644
--- a/roomserver/storage/interface.go
+++ b/roomserver/storage/interface.go
@@ -94,6 +94,8 @@ type Database interface {
 	// Opens and returns a room updater, which locks the room and opens a transaction.
 	// The GetRoomUpdater must have Commit or Rollback called on it if this doesn't return an error.
 	// If this returns an error then no further action is required.
+	// IsEventRejected returns true if the event is known and rejected.
+	IsEventRejected(ctx context.Context, roomNID types.RoomNID, eventID string) (rejected bool, err error)
 	GetRoomUpdater(ctx context.Context, roomInfo *types.RoomInfo) (*shared.RoomUpdater, error)
 	// Look up event references for the latest events in the room and the current state snapshot.
 	// Returns the latest events, the current state and the maximum depth of the latest events plus 1.
diff --git a/roomserver/storage/postgres/events_table.go b/roomserver/storage/postgres/events_table.go
index a4d05756..c7748d2b 100644
--- a/roomserver/storage/postgres/events_table.go
+++ b/roomserver/storage/postgres/events_table.go
@@ -136,6 +136,9 @@ const selectMaxEventDepthSQL = "" +
 const selectRoomNIDsForEventNIDsSQL = "" +
 	"SELECT event_nid, room_nid FROM roomserver_events WHERE event_nid = ANY($1)"
 
+const selectEventRejectedSQL = "" +
+	"SELECT is_rejected FROM roomserver_events WHERE room_nid = $1 AND event_id = $2"
+
 type eventStatements struct {
 	insertEventStmt                        *sql.Stmt
 	selectEventStmt                        *sql.Stmt
@@ -153,6 +156,7 @@ type eventStatements struct {
 	bulkSelectUnsentEventNIDStmt           *sql.Stmt
 	selectMaxEventDepthStmt                *sql.Stmt
 	selectRoomNIDsForEventNIDsStmt         *sql.Stmt
+	selectEventRejectedStmt                *sql.Stmt
 }
 
 func CreateEventsTable(db *sql.DB) error {
@@ -180,6 +184,7 @@ func PrepareEventsTable(db *sql.DB) (tables.Events, error) {
 		{&s.bulkSelectUnsentEventNIDStmt, bulkSelectUnsentEventNIDSQL},
 		{&s.selectMaxEventDepthStmt, selectMaxEventDepthSQL},
 		{&s.selectRoomNIDsForEventNIDsStmt, selectRoomNIDsForEventNIDsSQL},
+		{&s.selectEventRejectedStmt, selectEventRejectedSQL},
 	}.Prepare(db)
 }
 
@@ -540,3 +545,11 @@ func eventNIDsAsArray(eventNIDs []types.EventNID) pq.Int64Array {
 	}
 	return nids
 }
+
+func (s *eventStatements) SelectEventRejected(
+	ctx context.Context, txn *sql.Tx, roomNID types.RoomNID, eventID string,
+) (rejected bool, err error) {
+	stmt := sqlutil.TxStmt(txn, s.selectEventRejectedStmt)
+	err = stmt.QueryRowContext(ctx, roomNID, eventID).Scan(&rejected)
+	return
+}
diff --git a/roomserver/storage/shared/storage.go b/roomserver/storage/shared/storage.go
index cbf9c8b2..4f92adf1 100644
--- a/roomserver/storage/shared/storage.go
+++ b/roomserver/storage/shared/storage.go
@@ -567,6 +567,10 @@ func (d *Database) GetRoomUpdater(
 	return updater, err
 }
 
+func (d *Database) IsEventRejected(ctx context.Context, roomNID types.RoomNID, eventID string) (bool, error) {
+	return d.EventsTable.SelectEventRejected(ctx, nil, roomNID, eventID)
+}
+
 func (d *Database) StoreEvent(
 	ctx context.Context, event *gomatrixserverlib.Event,
 	authEventNIDs []types.EventNID, isRejected bool,
diff --git a/roomserver/storage/sqlite3/events_table.go b/roomserver/storage/sqlite3/events_table.go
index 1dda34c3..174e3a9a 100644
--- a/roomserver/storage/sqlite3/events_table.go
+++ b/roomserver/storage/sqlite3/events_table.go
@@ -109,6 +109,9 @@ const selectMaxEventDepthSQL = "" +
 const selectRoomNIDsForEventNIDsSQL = "" +
 	"SELECT event_nid, room_nid FROM roomserver_events WHERE event_nid IN ($1)"
 
+const selectEventRejectedSQL = "" +
+	"SELECT is_rejected FROM roomserver_events WHERE room_nid = $1 AND event_id = $2"
+
 type eventStatements struct {
 	db                                     *sql.DB
 	insertEventStmt                        *sql.Stmt
@@ -122,6 +125,7 @@ type eventStatements struct {
 	bulkSelectStateAtEventAndReferenceStmt *sql.Stmt
 	bulkSelectEventReferenceStmt           *sql.Stmt
 	bulkSelectEventIDStmt                  *sql.Stmt
+	selectEventRejectedStmt                *sql.Stmt
 	//bulkSelectEventNIDStmt               *sql.Stmt
 	//bulkSelectUnsentEventNIDStmt         *sql.Stmt
 	//selectRoomNIDsForEventNIDsStmt       *sql.Stmt
@@ -152,6 +156,7 @@ func PrepareEventsTable(db *sql.DB) (tables.Events, error) {
 		//{&s.bulkSelectEventNIDStmt, bulkSelectEventNIDSQL},
 		//{&s.bulkSelectUnsentEventNIDStmt, bulkSelectUnsentEventNIDSQL},
 		//{&s.selectRoomNIDForEventNIDStmt, selectRoomNIDForEventNIDSQL},
+		{&s.selectEventRejectedStmt, selectEventRejectedSQL},
 	}.Prepare(db)
 }
 
@@ -614,3 +619,11 @@ func eventNIDsAsArray(eventNIDs []types.EventNID) string {
 	b, _ := json.Marshal(eventNIDs)
 	return string(b)
 }
+
+func (s *eventStatements) SelectEventRejected(
+	ctx context.Context, txn *sql.Tx, roomNID types.RoomNID, eventID string,
+) (rejected bool, err error) {
+	stmt := sqlutil.TxStmt(txn, s.selectEventRejectedStmt)
+	err = stmt.QueryRowContext(ctx, roomNID, eventID).Scan(&rejected)
+	return
+}
diff --git a/roomserver/storage/tables/interface.go b/roomserver/storage/tables/interface.go
index 0bc389b8..ed67c43d 100644
--- a/roomserver/storage/tables/interface.go
+++ b/roomserver/storage/tables/interface.go
@@ -66,6 +66,7 @@ type Events interface {
 	BulkSelectUnsentEventNID(ctx context.Context, txn *sql.Tx, eventIDs []string) (map[string]types.EventNID, error)
 	SelectMaxEventDepth(ctx context.Context, txn *sql.Tx, eventNIDs []types.EventNID) (int64, error)
 	SelectRoomNIDsForEventNIDs(ctx context.Context, txn *sql.Tx, eventNIDs []types.EventNID) (roomNIDs map[types.EventNID]types.RoomNID, err error)
+	SelectEventRejected(ctx context.Context, txn *sql.Tx, roomNID types.RoomNID, eventID string) (rejected bool, err error)
 }
 
 type Rooms interface {