Remove usage of "conn_id" for presence. (#7128)

* Remove `conn_id` usage for UserSyncCommand.

Each tcp replication connection is assigned a "conn_id", which is used
to give an ID to a remotely connected worker. In a redis world, there
will no longer be a one to one mapping between connection and instance,
so instead we need to replace such usages with an ID generated by the
remote instances and included in the replicaiton commands.

This really only effects UserSyncCommand.

* Add CLEAR_USER_SYNCS command that is sent on shutdown.

This should help with the case where a synchrotron gets restarted
gracefully, rather than rely on 5 minute timeout.

1 files changed, 16 insertions, 4 deletions

diff --git a/synapse/app/generic_worker.py b/synapse/app/generic_worker.py
index fba7ad9551..1ee266f7c5 100644
--- a/synapse/app/generic_worker.py
+++ b/synapse/app/generic_worker.py
@@ -65,6 +65,7 @@ from synapse.replication.slave.storage.registration import SlavedRegistrationSto
 from synapse.replication.slave.storage.room import RoomStore
 from synapse.replication.slave.storage.transactions import SlavedTransactionStore
 from synapse.replication.tcp.client import ReplicationClientHandler
+from synapse.replication.tcp.commands import ClearUserSyncsCommand
 from synapse.replication.tcp.streams import (
     AccountDataStream,
     DeviceListsStream,
@@ -124,7 +125,6 @@ from synapse.types import ReadReceipt
 from synapse.util.async_helpers import Linearizer
 from synapse.util.httpresourcetree import create_resource_tree
 from synapse.util.manhole import manhole
-from synapse.util.stringutils import random_string
 from synapse.util.versionstring import get_version_string
 
 logger = logging.getLogger("synapse.app.generic_worker")
@@ -233,6 +233,7 @@ class GenericWorkerPresence(object):
         self.user_to_num_current_syncs = {}
         self.clock = hs.get_clock()
         self.notifier = hs.get_notifier()
+        self.instance_id = hs.get_instance_id()
 
         active_presence = self.store.take_presence_startup_info()
         self.user_to_current_state = {state.user_id: state for state in active_presence}
@@ -245,13 +246,24 @@ class GenericWorkerPresence(object):
             self.send_stop_syncing, UPDATE_SYNCING_USERS_MS
         )
 
-        self.process_id = random_string(16)
-        logger.info("Presence process_id is %r", self.process_id)
+        hs.get_reactor().addSystemEventTrigger(
+            "before",
+            "shutdown",
+            run_as_background_process,
+            "generic_presence.on_shutdown",
+            self._on_shutdown,
+        )
+
+    def _on_shutdown(self):
+        if self.hs.config.use_presence:
+            self.hs.get_tcp_replication().send_command(
+                ClearUserSyncsCommand(self.instance_id)
+            )
 
     def send_user_sync(self, user_id, is_syncing, last_sync_ms):
         if self.hs.config.use_presence:
             self.hs.get_tcp_replication().send_user_sync(
-                user_id, is_syncing, last_sync_ms
+                self.instance_id, user_id, is_syncing, last_sync_ms
             )
 
     def mark_as_coming_online(self, user_id):