Merge remote-tracking branch 'upstream/release-v1.69'

.github/workflows/docker.yml

@@ -27,7 +27,7 @@ jobs:
 
       - name: Inspect builder
         run: docker buildx inspect
-          
+
       - name: Log in to DockerHub
         uses: docker/login-action@v2
         with:
@@ -55,3 +55,8 @@ jobs:
           tags: "${{ steps.set-tag.outputs.tags }}"
           file: "docker/Dockerfile"
           platforms: linux/amd64,linux/arm64
+
+          # arm64 builds OOM without the git fetch setting. c.f.
+          # https://github.com/rust-lang/cargo/issues/10583
+          build-args: |
+            CARGO_NET_GIT_FETCH_WITH_CLI=true

CHANGES.md

@@ -1,4 +1,4 @@
-Synapse 1.69.0rc3 (2022-10-12)
+Synapse 1.69.0rc4 (2022-10-14)
 ==============================
 
 Please note that legacy Prometheus metric names are now deprecated and will be removed in Synapse 1.73.0.
@@ -6,6 +6,21 @@ Server administrators should update their dashboards and alerting rules to avoid
 See the [upgrade notes](https://matrix-org.github.io/synapse/v1.69/upgrade.html#upgrading-to-v1690) for more details.
 
 
+Bugfixes
+--------
+
+- Fix poor performance of the `event_push_backfill_thread_id` background update, which was introduced in Synapse 1.68.0rc1. ([\#14172](https://github.com/matrix-org/synapse/issues/14172), [\#14181](https://github.com/matrix-org/synapse/issues/14181))
+
+
+Updates to the Docker image
+---------------------------
+
+- Fix docker build OOMing in CI for arm64 builds. ([\#14173](https://github.com/matrix-org/synapse/issues/14173))
+
+
+Synapse 1.69.0rc3 (2022-10-12)
+==============================
+
 Bugfixes
 --------
 

debian/changelog

@@ -1,3 +1,9 @@
+matrix-synapse-py3 (1.69.0~rc4) stable; urgency=medium
+
+  * New Synapse release 1.69.0rc4.
+
+ -- Synapse Packaging team <packages@matrix.org>  Fri, 14 Oct 2022 15:04:47 +0100
+
 matrix-synapse-py3 (1.69.0~rc3) stable; urgency=medium
 
   * New Synapse release 1.69.0rc3.

docker/Dockerfile

@@ -108,6 +108,12 @@ RUN mkdir /rust /cargo
 
 RUN curl -sSf https://sh.rustup.rs | sh -s -- -y --no-modify-path --default-toolchain stable
 
+
+# arm64 builds consume a lot of memory if `CARGO_NET_GIT_FETCH_WITH_CLI` is not
+# set to true, so we expose it as a build-arg.
+ARG CARGO_NET_GIT_FETCH_WITH_CLI=false
+ENV CARGO_NET_GIT_FETCH_WITH_CLI=$CARGO_NET_GIT_FETCH_WITH_CLI
+
 # To speed up rebuilds, install all of the dependencies before we copy over
 # the whole synapse project, so that this layer in the Docker cache can be
 # used while you develop on the source

pyproject.toml

@@ -57,7 +57,7 @@ manifest-path = "rust/Cargo.toml"
 
 [tool.poetry]
 name = "matrix-synapse"
-version = "1.69.0rc3"
+version = "1.69.0rc4"
 description = "Homeserver for the Matrix decentralised comms protocol"
 authors = ["Matrix.org Team and Contributors <packages@matrix.org>"]
 license = "Apache-2.0"

synapse/storage/databases/main/event_push_actions.py

@@ -269,11 +269,11 @@ class EventPushActionsWorkerStore(ReceiptsWorkerStore, StreamWorkerStore, SQLBas
         event_push_actions_done = progress.get("event_push_actions_done", False)
 
         def add_thread_id_txn(
-            txn: LoggingTransaction, table_name: str, start_stream_ordering: int
+            txn: LoggingTransaction, start_stream_ordering: int
         ) -> int:
-            sql = f"""
+            sql = """
             SELECT stream_ordering
-            FROM {table_name}
+            FROM event_push_actions
             WHERE
                 thread_id IS NULL
                 AND stream_ordering > ?
@@ -285,7 +285,7 @@ class EventPushActionsWorkerStore(ReceiptsWorkerStore, StreamWorkerStore, SQLBas
             # No more rows to process.
             rows = txn.fetchall()
             if not rows:
-                progress[f"{table_name}_done"] = True
+                progress["event_push_actions_done"] = True
                 self.db_pool.updates._background_update_progress_txn(
                     txn, "event_push_backfill_thread_id", progress
                 )
@@ -294,16 +294,65 @@ class EventPushActionsWorkerStore(ReceiptsWorkerStore, StreamWorkerStore, SQLBas
             # Update the thread ID for any of those rows.
             max_stream_ordering = rows[-1][0]
 
-            sql = f"""
-            UPDATE {table_name}
+            sql = """
+            UPDATE event_push_actions
             SET thread_id = 'main'
-            WHERE stream_ordering <= ? AND thread_id IS NULL
+            WHERE ? < stream_ordering AND stream_ordering <= ? AND thread_id IS NULL
             """
-            txn.execute(sql, (max_stream_ordering,))
+            txn.execute(
+                sql,
+                (
+                    start_stream_ordering,
+                    max_stream_ordering,
+                ),
+            )
 
             # Update progress.
             processed_rows = txn.rowcount
-            progress[f"max_{table_name}_stream_ordering"] = max_stream_ordering
+            progress["max_event_push_actions_stream_ordering"] = max_stream_ordering
+            self.db_pool.updates._background_update_progress_txn(
+                txn, "event_push_backfill_thread_id", progress
+            )
+
+            return processed_rows
+
+        def add_thread_id_summary_txn(txn: LoggingTransaction) -> int:
+            min_user_id = progress.get("max_summary_user_id", "")
+            min_room_id = progress.get("max_summary_room_id", "")
+
+            # Slightly overcomplicated query for getting the Nth user ID / room
+            # ID tuple, or the last if there are less than N remaining.
+            sql = """
+            SELECT user_id, room_id FROM (
+                SELECT user_id, room_id FROM event_push_summary
+                WHERE (user_id, room_id) > (?, ?)
+                    AND thread_id IS NULL
+                ORDER BY user_id, room_id
+                LIMIT ?
+            ) AS e
+            ORDER BY user_id DESC, room_id DESC
+            LIMIT 1
+            """
+
+            txn.execute(sql, (min_user_id, min_room_id, batch_size))
+            row = txn.fetchone()
+            if not row:
+                return 0
+
+            max_user_id, max_room_id = row
+
+            sql = """
+            UPDATE event_push_summary
+            SET thread_id = 'main'
+            WHERE
+                (?, ?) < (user_id, room_id) AND (user_id, room_id) <= (?, ?)
+                AND thread_id IS NULL
+            """
+            txn.execute(sql, (min_user_id, min_room_id, max_user_id, max_room_id))
+            processed_rows = txn.rowcount
+
+            progress["max_summary_user_id"] = max_user_id
+            progress["max_summary_room_id"] = max_room_id
             self.db_pool.updates._background_update_progress_txn(
                 txn, "event_push_backfill_thread_id", progress
             )
@@ -319,15 +368,12 @@ class EventPushActionsWorkerStore(ReceiptsWorkerStore, StreamWorkerStore, SQLBas
             result = await self.db_pool.runInteraction(
                 "event_push_backfill_thread_id",
                 add_thread_id_txn,
-                "event_push_actions",
                 progress.get("max_event_push_actions_stream_ordering", 0),
             )
         else:
             result = await self.db_pool.runInteraction(
                 "event_push_backfill_thread_id",
-                add_thread_id_txn,
-                "event_push_summary",
-                progress.get("max_event_push_summary_stream_ordering", 0),
+                add_thread_id_summary_txn,
             )
 
             # Only done after the event_push_summary table is done.