From 8c4937b216309eec62f9262f23162878fa3b772c Mon Sep 17 00:00:00 2001
From: Erik Johnston <erikj@element.io>
Date: Mon, 10 Jun 2024 15:56:57 +0100
Subject: Fix bug where device lists would break sync (#17292)

If the stream ID in the unconverted table is ahead of the device lists
ID gen, then it can break all /sync requests that had an ID from ahead
of the table.

The fix is to make sure we add the unconverted table to the list of
tables we check at start up.

Broke in https://github.com/element-hq/synapse/pull/17229
---
 changelog.d/17292.bugfix                           |  1 +
 synapse/storage/databases/main/devices.py          | 28 +++++++++++++++-------
 .../85/05_add_instance_names_converted_pos.sql     | 16 +++++++++++++
 3 files changed, 36 insertions(+), 9 deletions(-)
 create mode 100644 changelog.d/17292.bugfix
 create mode 100644 synapse/storage/schema/main/delta/85/05_add_instance_names_converted_pos.sql

diff --git a/changelog.d/17292.bugfix b/changelog.d/17292.bugfix
new file mode 100644
index 0000000000..c067a98ce8
--- /dev/null
+++ b/changelog.d/17292.bugfix
@@ -0,0 +1 @@
+Fix bug where `/sync` could get stuck due to edge case in device lists handling. Introduced in v1.109.0rc1.
diff --git a/synapse/storage/databases/main/devices.py b/synapse/storage/databases/main/devices.py
index 1c771e48f7..40187496e2 100644
--- a/synapse/storage/databases/main/devices.py
+++ b/synapse/storage/databases/main/devices.py
@@ -108,6 +108,11 @@ class DeviceWorkerStore(RoomMemberWorkerStore, EndToEndKeyWorkerStore):
                 ("device_lists_outbound_pokes", "instance_name", "stream_id"),
                 ("device_lists_changes_in_room", "instance_name", "stream_id"),
                 ("device_lists_remote_pending", "instance_name", "stream_id"),
+                (
+                    "device_lists_changes_converted_stream_position",
+                    "instance_name",
+                    "stream_id",
+                ),
             ],
             sequence_name="device_lists_sequence",
             writers=["master"],
@@ -2394,15 +2399,16 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore):
         `FALSE` have not been converted.
         """
 
-        return cast(
-            Tuple[int, str],
-            await self.db_pool.simple_select_one(
-                table="device_lists_changes_converted_stream_position",
-                keyvalues={},
-                retcols=["stream_id", "room_id"],
-                desc="get_device_change_last_converted_pos",
-            ),
+        # There should be only one row in this table, though we want to
+        # future-proof ourselves for when we have multiple rows (one for each
+        # instance). So to handle that case we take the minimum of all rows.
+        rows = await self.db_pool.simple_select_list(
+            table="device_lists_changes_converted_stream_position",
+            keyvalues={},
+            retcols=["stream_id", "room_id"],
+            desc="get_device_change_last_converted_pos",
         )
+        return cast(Tuple[int, str], min(rows))
 
     async def set_device_change_last_converted_pos(
         self,
@@ -2417,6 +2423,10 @@ class DeviceStore(DeviceWorkerStore, DeviceBackgroundUpdateStore):
         await self.db_pool.simple_update_one(
             table="device_lists_changes_converted_stream_position",
             keyvalues={},
-            updatevalues={"stream_id": stream_id, "room_id": room_id},
+            updatevalues={
+                "stream_id": stream_id,
+                "instance_name": self._instance_name,
+                "room_id": room_id,
+            },
             desc="set_device_change_last_converted_pos",
         )
diff --git a/synapse/storage/schema/main/delta/85/05_add_instance_names_converted_pos.sql b/synapse/storage/schema/main/delta/85/05_add_instance_names_converted_pos.sql
new file mode 100644
index 0000000000..c3f2b6a1dd
--- /dev/null
+++ b/synapse/storage/schema/main/delta/85/05_add_instance_names_converted_pos.sql
@@ -0,0 +1,16 @@
+--
+-- This file is licensed under the Affero General Public License (AGPL) version 3.
+--
+-- Copyright (C) 2024 New Vector, Ltd
+--
+-- This program is free software: you can redistribute it and/or modify
+-- it under the terms of the GNU Affero General Public License as
+-- published by the Free Software Foundation, either version 3 of the
+-- License, or (at your option) any later version.
+--
+-- See the GNU Affero General Public License for more details:
+-- <https://www.gnu.org/licenses/agpl-3.0.html>.
+
+-- Add `instance_name` columns to stream tables to allow them to be used with
+-- `MultiWriterIdGenerator`
+ALTER TABLE device_lists_changes_converted_stream_position ADD COLUMN instance_name TEXT;
-- 
cgit 1.4.1