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, 11 insertions, 0 deletions

diff --git a/synapse/server.py b/synapse/server.py
index c7ca2bda0d..cd86475d6b 100644
--- a/synapse/server.py
+++ b/synapse/server.py
@@ -103,6 +103,7 @@ from synapse.storage import DataStores, Storage
 from synapse.streams.events import EventSources
 from synapse.util import Clock
 from synapse.util.distributor import Distributor
+from synapse.util.stringutils import random_string
 
 logger = logging.getLogger(__name__)
 
@@ -230,6 +231,8 @@ class HomeServer(object):
         self._listening_services = []
         self.start_time = None
 
+        self.instance_id = random_string(5)
+
         self.clock = Clock(reactor)
         self.distributor = Distributor()
         self.ratelimiter = Ratelimiter()
@@ -242,6 +245,14 @@ class HomeServer(object):
         for depname in kwargs:
             setattr(self, depname, kwargs[depname])
 
+    def get_instance_id(self):
+        """A unique ID for this synapse process instance.
+
+        This is used to distinguish running instances in worker-based
+        deployments.
+        """
+        return self.instance_id
+
     def setup(self):
         logger.info("Setting up.")
         self.start_time = int(self.get_clock().time())