mirror of
https://github.com/djdevin/recflare.git
synced 2026-10-25 15:09:18 -07:00
[rooms] fix room instance index
This commit is contained in:
@@ -0,0 +1,26 @@
|
||||
-- Restore the two UNIQUE indexes on `room_instance`.
|
||||
--
|
||||
-- Generated from packages/domain/src/room-instance-db.ts (ROOM_INSTANCE_SCHEMA_DDL) —
|
||||
-- keep in sync. The DDL and migration 0004 have always declared them, but the deployed
|
||||
-- table carried only the non-unique `idx_room_instance_room_id`: the two `CREATE UNIQUE
|
||||
-- INDEX` statements never took, so nothing has enforced uniqueness on `id` or
|
||||
-- `photon_room_id` in production. Two rows sharing one `photon_room_id` were found —
|
||||
-- the signature of a replayed INSERT (same blob written twice), which the index would
|
||||
-- have refused. Without the `id` index the `MAX(id) + 1` allocation in
|
||||
-- createRoomInstance can also hand two concurrent matchmakes the same id, which the
|
||||
-- index would have failed loudly instead of silently minting two instances.
|
||||
--
|
||||
-- A UNIQUE index refuses to build over duplicates, so they're removed first. Instances
|
||||
-- are live sessions, not records: of each duplicate set the earliest row (lowest rowid)
|
||||
-- is kept and the rest dropped. A replayed insert's twins are identical, so nothing is
|
||||
-- lost; presence rows point at instances by `id`, which the survivor still carries.
|
||||
-- Dedupe by `id` first so a row that is a duplicate on both counts is only counted once.
|
||||
|
||||
DELETE FROM room_instance
|
||||
WHERE rowid NOT IN (SELECT MIN(rowid) FROM room_instance GROUP BY id);
|
||||
|
||||
DELETE FROM room_instance
|
||||
WHERE rowid NOT IN (SELECT MIN(rowid) FROM room_instance GROUP BY photon_room_id);
|
||||
|
||||
CREATE UNIQUE INDEX IF NOT EXISTS idx_room_instance_id ON room_instance (id);
|
||||
CREATE UNIQUE INDEX IF NOT EXISTS idx_room_instance_photon_room_id ON room_instance (photon_room_id);
|
||||
Reference in New Issue
Block a user