Make event persisters periodically announce position over replication. (#8499)

Currently background proccesses stream the events stream use the "minimum persisted position" (i.e. `get_current_token()`) rather than the vector clock style tokens. This is broadly fine as it doesn't matter if the background processes lag a small amount. However, in extreme cases (i.e. SyTests) where we only write to one event persister the background processes will never make progress.

This PR changes it so that the `MultiWriterIDGenerator` keeps the current position of a given instance as up to date as possible (i.e using the latest token it sees if its not in the process of persisting anything), and then periodically announces that over replication. This then allows the "minimum persisted position" to advance, albeit with a small lag.

1 files changed, 16 insertions, 9 deletions

diff --git a/tests/storage/test_id_generators.py b/tests/storage/test_id_generators.py
index 392b08832b..cc0612cf65 100644
--- a/tests/storage/test_id_generators.py
+++ b/tests/storage/test_id_generators.py
@@ -199,10 +199,17 @@ class MultiWriterIdGeneratorTestCase(HomeserverTestCase):
         first_id_gen = self._create_id_generator("first", writers=["first", "second"])
         second_id_gen = self._create_id_generator("second", writers=["first", "second"])
 
-        self.assertEqual(first_id_gen.get_positions(), {"first": 3, "second": 7})
-        self.assertEqual(first_id_gen.get_current_token_for_writer("first"), 3)
+        # The first ID gen will notice that it can advance its token to 7 as it
+        # has no in progress writes...
+        self.assertEqual(first_id_gen.get_positions(), {"first": 7, "second": 7})
+        self.assertEqual(first_id_gen.get_current_token_for_writer("first"), 7)
         self.assertEqual(first_id_gen.get_current_token_for_writer("second"), 7)
 
+        # ... but the second ID gen doesn't know that.
+        self.assertEqual(second_id_gen.get_positions(), {"first": 3, "second": 7})
+        self.assertEqual(second_id_gen.get_current_token_for_writer("first"), 3)
+        self.assertEqual(second_id_gen.get_current_token_for_writer("second"), 7)
+
         # Try allocating a new ID gen and check that we only see position
         # advanced after we leave the context manager.
 
@@ -211,7 +218,7 @@ class MultiWriterIdGeneratorTestCase(HomeserverTestCase):
                 self.assertEqual(stream_id, 8)
 
                 self.assertEqual(
-                    first_id_gen.get_positions(), {"first": 3, "second": 7}
+                    first_id_gen.get_positions(), {"first": 7, "second": 7}
                 )
 
         self.get_success(_get_next_async())
@@ -279,7 +286,7 @@ class MultiWriterIdGeneratorTestCase(HomeserverTestCase):
         self._insert_row_with_id("first", 3)
         self._insert_row_with_id("second", 5)
 
-        id_gen = self._create_id_generator("first", writers=["first", "second"])
+        id_gen = self._create_id_generator("worker", writers=["first", "second"])
 
         self.assertEqual(id_gen.get_positions(), {"first": 3, "second": 5})
 
@@ -319,14 +326,14 @@ class MultiWriterIdGeneratorTestCase(HomeserverTestCase):
 
         id_gen = self._create_id_generator("first", writers=["first", "second"])
 
-        self.assertEqual(id_gen.get_positions(), {"first": 3, "second": 5})
+        self.assertEqual(id_gen.get_positions(), {"first": 5, "second": 5})
 
-        self.assertEqual(id_gen.get_persisted_upto_position(), 3)
+        self.assertEqual(id_gen.get_persisted_upto_position(), 5)
 
         async def _get_next_async():
             async with id_gen.get_next() as stream_id:
                 self.assertEqual(stream_id, 6)
-                self.assertEqual(id_gen.get_persisted_upto_position(), 3)
+                self.assertEqual(id_gen.get_persisted_upto_position(), 5)
 
         self.get_success(_get_next_async())
 
@@ -388,7 +395,7 @@ class MultiWriterIdGeneratorTestCase(HomeserverTestCase):
         self._insert_row_with_id("second", 5)
 
         # Initial config has two writers
-        id_gen = self._create_id_generator("first", writers=["first", "second"])
+        id_gen = self._create_id_generator("worker", writers=["first", "second"])
         self.assertEqual(id_gen.get_persisted_upto_position(), 3)
         self.assertEqual(id_gen.get_current_token_for_writer("first"), 3)
         self.assertEqual(id_gen.get_current_token_for_writer("second"), 5)
@@ -568,7 +575,7 @@ class BackwardsMultiWriterIdGeneratorTestCase(HomeserverTestCase):
 
         self.get_success(_get_next_async2())
 
-        self.assertEqual(id_gen_1.get_positions(), {"first": -1, "second": -2})
+        self.assertEqual(id_gen_1.get_positions(), {"first": -2, "second": -2})
         self.assertEqual(id_gen_2.get_positions(), {"first": -1, "second": -2})
         self.assertEqual(id_gen_1.get_persisted_upto_position(), -2)
         self.assertEqual(id_gen_2.get_persisted_upto_position(), -2)