summary refs log tree commit diff
path: root/synapse/storage
diff options
context:
space:
mode:
Diffstat (limited to 'synapse/storage')
-rw-r--r--synapse/storage/__init__.py2
-rw-r--r--synapse/storage/_base.py34
-rw-r--r--synapse/storage/client_ips.py34
-rw-r--r--synapse/storage/deviceinbox.py2
-rw-r--r--synapse/storage/devices.py10
-rw-r--r--synapse/storage/directory.py3
-rw-r--r--synapse/storage/e2e_room_keys.py320
-rw-r--r--synapse/storage/end_to_end_keys.py6
-rw-r--r--synapse/storage/engines/postgres.py9
-rw-r--r--synapse/storage/event_federation.py38
-rw-r--r--synapse/storage/events.py209
-rw-r--r--synapse/storage/events_worker.py9
-rw-r--r--synapse/storage/filtering.py6
-rw-r--r--synapse/storage/keys.py3
-rw-r--r--synapse/storage/monthly_active_users.py104
-rw-r--r--synapse/storage/profile.py4
-rw-r--r--synapse/storage/pusher.py14
-rw-r--r--synapse/storage/registration.py65
-rw-r--r--synapse/storage/room.py58
-rw-r--r--synapse/storage/roommember.py65
-rw-r--r--synapse/storage/schema/delta/51/e2e_room_keys.sql39
-rw-r--r--synapse/storage/signatures.py2
-rw-r--r--synapse/storage/state.py903
-rw-r--r--synapse/storage/stream.py16
-rw-r--r--synapse/storage/transactions.py40
25 files changed, 1496 insertions, 499 deletions
diff --git a/synapse/storage/__init__.py b/synapse/storage/__init__.py
index 23b4a8d76d..53c685c173 100644
--- a/synapse/storage/__init__.py
+++ b/synapse/storage/__init__.py
@@ -30,6 +30,7 @@ from .appservice import ApplicationServiceStore, ApplicationServiceTransactionSt
 from .client_ips import ClientIpStore
 from .deviceinbox import DeviceInboxStore
 from .directory import DirectoryStore
+from .e2e_room_keys import EndToEndRoomKeyStore
 from .end_to_end_keys import EndToEndKeyStore
 from .engines import PostgresEngine
 from .event_federation import EventFederationStore
@@ -77,6 +78,7 @@ class DataStore(RoomMemberStore, RoomStore,
                 ApplicationServiceTransactionStore,
                 ReceiptsStore,
                 EndToEndKeyStore,
+                EndToEndRoomKeyStore,
                 SearchStore,
                 TagsStore,
                 AccountDataStore,
diff --git a/synapse/storage/_base.py b/synapse/storage/_base.py
index 08dffd774f..d9d0255d0b 100644
--- a/synapse/storage/_base.py
+++ b/synapse/storage/_base.py
@@ -17,9 +17,10 @@ import sys
 import threading
 import time
 
-from six import iteritems, iterkeys, itervalues
-from six.moves import intern, range
+from six import PY2, iteritems, iterkeys, itervalues
+from six.moves import builtins, intern, range
 
+from canonicaljson import json
 from prometheus_client import Histogram
 
 from twisted.internet import defer
@@ -1216,3 +1217,32 @@ class _RollbackButIsFineException(Exception):
     something went wrong.
     """
     pass
+
+
+def db_to_json(db_content):
+    """
+    Take some data from a database row and return a JSON-decoded object.
+
+    Args:
+        db_content (memoryview|buffer|bytes|bytearray|unicode)
+    """
+    # psycopg2 on Python 3 returns memoryview objects, which we need to
+    # cast to bytes to decode
+    if isinstance(db_content, memoryview):
+        db_content = db_content.tobytes()
+
+    # psycopg2 on Python 2 returns buffer objects, which we need to cast to
+    # bytes to decode
+    if PY2 and isinstance(db_content, builtins.buffer):
+        db_content = bytes(db_content)
+
+    # Decode it to a Unicode string before feeding it to json.loads, so we
+    # consistenty get a Unicode-containing object out.
+    if isinstance(db_content, (bytes, bytearray)):
+        db_content = db_content.decode('utf8')
+
+    try:
+        return json.loads(db_content)
+    except Exception:
+        logging.warning("Tried to decode '%r' as JSON and failed", db_content)
+        raise
diff --git a/synapse/storage/client_ips.py b/synapse/storage/client_ips.py
index 8fc678fa67..9ad17b7c25 100644
--- a/synapse/storage/client_ips.py
+++ b/synapse/storage/client_ips.py
@@ -119,21 +119,25 @@ class ClientIpStore(background_updates.BackgroundUpdateStore):
         for entry in iteritems(to_update):
             (user_id, access_token, ip), (user_agent, device_id, last_seen) = entry
 
-            self._simple_upsert_txn(
-                txn,
-                table="user_ips",
-                keyvalues={
-                    "user_id": user_id,
-                    "access_token": access_token,
-                    "ip": ip,
-                    "user_agent": user_agent,
-                    "device_id": device_id,
-                },
-                values={
-                    "last_seen": last_seen,
-                },
-                lock=False,
-            )
+            try:
+                self._simple_upsert_txn(
+                    txn,
+                    table="user_ips",
+                    keyvalues={
+                        "user_id": user_id,
+                        "access_token": access_token,
+                        "ip": ip,
+                        "user_agent": user_agent,
+                        "device_id": device_id,
+                    },
+                    values={
+                        "last_seen": last_seen,
+                    },
+                    lock=False,
+                )
+            except Exception as e:
+                # Failed to upsert, log and continue
+                logger.error("Failed to insert client IP %r: %r", entry, e)
 
     @defer.inlineCallbacks
     def get_last_client_ip_by_device(self, user_id, device_id):
diff --git a/synapse/storage/deviceinbox.py b/synapse/storage/deviceinbox.py
index 73646da025..e06b0bc56d 100644
--- a/synapse/storage/deviceinbox.py
+++ b/synapse/storage/deviceinbox.py
@@ -169,7 +169,7 @@ class DeviceInboxStore(BackgroundUpdateStore):
         local_by_user_then_device = {}
         for user_id, messages_by_device in messages_by_user_then_device.items():
             messages_json_for_user = {}
-            devices = messages_by_device.keys()
+            devices = list(messages_by_device.keys())
             if len(devices) == 1 and devices[0] == "*":
                 # Handle wildcard device_ids.
                 sql = (
diff --git a/synapse/storage/devices.py b/synapse/storage/devices.py
index c0943ecf91..d10ff9e4b9 100644
--- a/synapse/storage/devices.py
+++ b/synapse/storage/devices.py
@@ -24,7 +24,7 @@ from synapse.api.errors import StoreError
 from synapse.metrics.background_process_metrics import run_as_background_process
 from synapse.util.caches.descriptors import cached, cachedInlineCallbacks, cachedList
 
-from ._base import Cache, SQLBaseStore
+from ._base import Cache, SQLBaseStore, db_to_json
 
 logger = logging.getLogger(__name__)
 
@@ -411,7 +411,7 @@ class DeviceStore(SQLBaseStore):
                 if device is not None:
                     key_json = device.get("key_json", None)
                     if key_json:
-                        result["keys"] = json.loads(key_json)
+                        result["keys"] = db_to_json(key_json)
                     device_display_name = device.get("device_display_name", None)
                     if device_display_name:
                         result["device_display_name"] = device_display_name
@@ -466,7 +466,7 @@ class DeviceStore(SQLBaseStore):
             retcol="content",
             desc="_get_cached_user_device",
         )
-        defer.returnValue(json.loads(content))
+        defer.returnValue(db_to_json(content))
 
     @cachedInlineCallbacks()
     def _get_cached_devices_for_user(self, user_id):
@@ -479,7 +479,7 @@ class DeviceStore(SQLBaseStore):
             desc="_get_cached_devices_for_user",
         )
         defer.returnValue({
-            device["device_id"]: json.loads(device["content"])
+            device["device_id"]: db_to_json(device["content"])
             for device in devices
         })
 
@@ -511,7 +511,7 @@ class DeviceStore(SQLBaseStore):
 
                 key_json = device.get("key_json", None)
                 if key_json:
-                    result["keys"] = json.loads(key_json)
+                    result["keys"] = db_to_json(key_json)
                 device_display_name = device.get("device_display_name", None)
                 if device_display_name:
                     result["device_display_name"] = device_display_name
diff --git a/synapse/storage/directory.py b/synapse/storage/directory.py
index 808194236a..61a029a53c 100644
--- a/synapse/storage/directory.py
+++ b/synapse/storage/directory.py
@@ -75,7 +75,6 @@ class DirectoryWorkerStore(SQLBaseStore):
             },
             retcol="creator",
             desc="get_room_alias_creator",
-            allow_none=True
         )
 
     @cached(max_entries=5000)
@@ -91,7 +90,7 @@ class DirectoryWorkerStore(SQLBaseStore):
 class DirectoryStore(DirectoryWorkerStore):
     @defer.inlineCallbacks
     def create_room_alias_association(self, room_alias, room_id, servers, creator=None):
-        """ Creates an associatin between  a room alias and room_id/servers
+        """ Creates an association between a room alias and room_id/servers
 
         Args:
             room_alias (RoomAlias)
diff --git a/synapse/storage/e2e_room_keys.py b/synapse/storage/e2e_room_keys.py
new file mode 100644
index 0000000000..f25ded2295
--- /dev/null
+++ b/synapse/storage/e2e_room_keys.py
@@ -0,0 +1,320 @@
+# -*- coding: utf-8 -*-
+# Copyright 2017 New Vector Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import json
+
+from twisted.internet import defer
+
+from synapse.api.errors import StoreError
+
+from ._base import SQLBaseStore
+
+
+class EndToEndRoomKeyStore(SQLBaseStore):
+
+    @defer.inlineCallbacks
+    def get_e2e_room_key(self, user_id, version, room_id, session_id):
+        """Get the encrypted E2E room key for a given session from a given
+        backup version of room_keys.  We only store the 'best' room key for a given
+        session at a given time, as determined by the handler.
+
+        Args:
+            user_id(str): the user whose backup we're querying
+            version(str): the version ID of the backup for the set of keys we're querying
+            room_id(str): the ID of the room whose keys we're querying.
+                This is a bit redundant as it's implied by the session_id, but
+                we include for consistency with the rest of the API.
+            session_id(str): the session whose room_key we're querying.
+
+        Returns:
+            A deferred dict giving the session_data and message metadata for
+            this room key.
+        """
+
+        row = yield self._simple_select_one(
+            table="e2e_room_keys",
+            keyvalues={
+                "user_id": user_id,
+                "version": version,
+                "room_id": room_id,
+                "session_id": session_id,
+            },
+            retcols=(
+                "first_message_index",
+                "forwarded_count",
+                "is_verified",
+                "session_data",
+            ),
+            desc="get_e2e_room_key",
+        )
+
+        row["session_data"] = json.loads(row["session_data"])
+
+        defer.returnValue(row)
+
+    @defer.inlineCallbacks
+    def set_e2e_room_key(self, user_id, version, room_id, session_id, room_key):
+        """Replaces or inserts the encrypted E2E room key for a given session in
+        a given backup
+
+        Args:
+            user_id(str): the user whose backup we're setting
+            version(str): the version ID of the backup we're updating
+            room_id(str): the ID of the room whose keys we're setting
+            session_id(str): the session whose room_key we're setting
+            room_key(dict): the room_key being set
+        Raises:
+            StoreError
+        """
+
+        yield self._simple_upsert(
+            table="e2e_room_keys",
+            keyvalues={
+                "user_id": user_id,
+                "room_id": room_id,
+                "session_id": session_id,
+            },
+            values={
+                "version": version,
+                "first_message_index": room_key['first_message_index'],
+                "forwarded_count": room_key['forwarded_count'],
+                "is_verified": room_key['is_verified'],
+                "session_data": json.dumps(room_key['session_data']),
+            },
+            lock=False,
+        )
+
+    @defer.inlineCallbacks
+    def get_e2e_room_keys(
+        self, user_id, version, room_id=None, session_id=None
+    ):
+        """Bulk get the E2E room keys for a given backup, optionally filtered to a given
+        room, or a given session.
+
+        Args:
+            user_id(str): the user whose backup we're querying
+            version(str): the version ID of the backup for the set of keys we're querying
+            room_id(str): Optional. the ID of the room whose keys we're querying, if any.
+                If not specified, we return the keys for all the rooms in the backup.
+            session_id(str): Optional. the session whose room_key we're querying, if any.
+                If specified, we also require the room_id to be specified.
+                If not specified, we return all the keys in this version of
+                the backup (or for the specified room)
+
+        Returns:
+            A deferred list of dicts giving the session_data and message metadata for
+            these room keys.
+        """
+
+        keyvalues = {
+            "user_id": user_id,
+            "version": version,
+        }
+        if room_id:
+            keyvalues['room_id'] = room_id
+            if session_id:
+                keyvalues['session_id'] = session_id
+
+        rows = yield self._simple_select_list(
+            table="e2e_room_keys",
+            keyvalues=keyvalues,
+            retcols=(
+                "user_id",
+                "room_id",
+                "session_id",
+                "first_message_index",
+                "forwarded_count",
+                "is_verified",
+                "session_data",
+            ),
+            desc="get_e2e_room_keys",
+        )
+
+        sessions = {'rooms': {}}
+        for row in rows:
+            room_entry = sessions['rooms'].setdefault(row['room_id'], {"sessions": {}})
+            room_entry['sessions'][row['session_id']] = {
+                "first_message_index": row["first_message_index"],
+                "forwarded_count": row["forwarded_count"],
+                "is_verified": row["is_verified"],
+                "session_data": json.loads(row["session_data"]),
+            }
+
+        defer.returnValue(sessions)
+
+    @defer.inlineCallbacks
+    def delete_e2e_room_keys(
+        self, user_id, version, room_id=None, session_id=None
+    ):
+        """Bulk delete the E2E room keys for a given backup, optionally filtered to a given
+        room or a given session.
+
+        Args:
+            user_id(str): the user whose backup we're deleting from
+            version(str): the version ID of the backup for the set of keys we're deleting
+            room_id(str): Optional. the ID of the room whose keys we're deleting, if any.
+                If not specified, we delete the keys for all the rooms in the backup.
+            session_id(str): Optional. the session whose room_key we're querying, if any.
+                If specified, we also require the room_id to be specified.
+                If not specified, we delete all the keys in this version of
+                the backup (or for the specified room)
+
+        Returns:
+            A deferred of the deletion transaction
+        """
+
+        keyvalues = {
+            "user_id": user_id,
+            "version": version,
+        }
+        if room_id:
+            keyvalues['room_id'] = room_id
+            if session_id:
+                keyvalues['session_id'] = session_id
+
+        yield self._simple_delete(
+            table="e2e_room_keys",
+            keyvalues=keyvalues,
+            desc="delete_e2e_room_keys",
+        )
+
+    @staticmethod
+    def _get_current_version(txn, user_id):
+        txn.execute(
+            "SELECT MAX(version) FROM e2e_room_keys_versions "
+            "WHERE user_id=? AND deleted=0",
+            (user_id,)
+        )
+        row = txn.fetchone()
+        if not row:
+            raise StoreError(404, 'No current backup version')
+        return row[0]
+
+    def get_e2e_room_keys_version_info(self, user_id, version=None):
+        """Get info metadata about a version of our room_keys backup.
+
+        Args:
+            user_id(str): the user whose backup we're querying
+            version(str): Optional. the version ID of the backup we're querying about
+                If missing, we return the information about the current version.
+        Raises:
+            StoreError: with code 404 if there are no e2e_room_keys_versions present
+        Returns:
+            A deferred dict giving the info metadata for this backup version
+        """
+
+        def _get_e2e_room_keys_version_info_txn(txn):
+            if version is None:
+                this_version = self._get_current_version(txn, user_id)
+            else:
+                this_version = version
+
+            result = self._simple_select_one_txn(
+                txn,
+                table="e2e_room_keys_versions",
+                keyvalues={
+                    "user_id": user_id,
+                    "version": this_version,
+                    "deleted": 0,
+                },
+                retcols=(
+                    "version",
+                    "algorithm",
+                    "auth_data",
+                ),
+            )
+            result["auth_data"] = json.loads(result["auth_data"])
+            return result
+
+        return self.runInteraction(
+            "get_e2e_room_keys_version_info",
+            _get_e2e_room_keys_version_info_txn
+        )
+
+    def create_e2e_room_keys_version(self, user_id, info):
+        """Atomically creates a new version of this user's e2e_room_keys store
+        with the given version info.
+
+        Args:
+            user_id(str): the user whose backup we're creating a version
+            info(dict): the info about the backup version to be created
+
+        Returns:
+            A deferred string for the newly created version ID
+        """
+
+        def _create_e2e_room_keys_version_txn(txn):
+            txn.execute(
+                "SELECT MAX(version) FROM e2e_room_keys_versions WHERE user_id=?",
+                (user_id,)
+            )
+            current_version = txn.fetchone()[0]
+            if current_version is None:
+                current_version = '0'
+
+            new_version = str(int(current_version) + 1)
+
+            self._simple_insert_txn(
+                txn,
+                table="e2e_room_keys_versions",
+                values={
+                    "user_id": user_id,
+                    "version": new_version,
+                    "algorithm": info["algorithm"],
+                    "auth_data": json.dumps(info["auth_data"]),
+                },
+            )
+
+            return new_version
+
+        return self.runInteraction(
+            "create_e2e_room_keys_version_txn", _create_e2e_room_keys_version_txn
+        )
+
+    def delete_e2e_room_keys_version(self, user_id, version=None):
+        """Delete a given backup version of the user's room keys.
+        Doesn't delete their actual key data.
+
+        Args:
+            user_id(str): the user whose backup version we're deleting
+            version(str): Optional. the version ID of the backup version we're deleting
+                If missing, we delete the current backup version info.
+        Raises:
+            StoreError: with code 404 if there are no e2e_room_keys_versions present,
+                or if the version requested doesn't exist.
+        """
+
+        def _delete_e2e_room_keys_version_txn(txn):
+            if version is None:
+                this_version = self._get_current_version(txn, user_id)
+            else:
+                this_version = version
+
+            return self._simple_update_one_txn(
+                txn,
+                table="e2e_room_keys_versions",
+                keyvalues={
+                    "user_id": user_id,
+                    "version": this_version,
+                },
+                updatevalues={
+                    "deleted": 1,
+                }
+            )
+
+        return self.runInteraction(
+            "delete_e2e_room_keys_version",
+            _delete_e2e_room_keys_version_txn
+        )
diff --git a/synapse/storage/end_to_end_keys.py b/synapse/storage/end_to_end_keys.py
index 523b4360c3..1f1721e820 100644
--- a/synapse/storage/end_to_end_keys.py
+++ b/synapse/storage/end_to_end_keys.py
@@ -14,13 +14,13 @@
 # limitations under the License.
 from six import iteritems
 
-from canonicaljson import encode_canonical_json, json
+from canonicaljson import encode_canonical_json
 
 from twisted.internet import defer
 
 from synapse.util.caches.descriptors import cached
 
-from ._base import SQLBaseStore
+from ._base import SQLBaseStore, db_to_json
 
 
 class EndToEndKeyStore(SQLBaseStore):
@@ -90,7 +90,7 @@ class EndToEndKeyStore(SQLBaseStore):
 
         for user_id, device_keys in iteritems(results):
             for device_id, device_info in iteritems(device_keys):
-                device_info["keys"] = json.loads(device_info.pop("key_json"))
+                device_info["keys"] = db_to_json(device_info.pop("key_json"))
 
         defer.returnValue(results)
 
diff --git a/synapse/storage/engines/postgres.py b/synapse/storage/engines/postgres.py
index 8a0386c1a4..42225f8a2a 100644
--- a/synapse/storage/engines/postgres.py
+++ b/synapse/storage/engines/postgres.py
@@ -41,13 +41,18 @@ class PostgresEngine(object):
         db_conn.set_isolation_level(
             self.module.extensions.ISOLATION_LEVEL_REPEATABLE_READ
         )
+
+        # Set the bytea output to escape, vs the default of hex
+        cursor = db_conn.cursor()
+        cursor.execute("SET bytea_output TO escape")
+
         # Asynchronous commit, don't wait for the server to call fsync before
         # ending the transaction.
         # https://www.postgresql.org/docs/current/static/wal-async-commit.html
         if not self.synchronous_commit:
-            cursor = db_conn.cursor()
             cursor.execute("SET synchronous_commit TO OFF")
-            cursor.close()
+
+        cursor.close()
 
     def is_deadlock(self, error):
         if isinstance(error, self.module.DatabaseError):
diff --git a/synapse/storage/event_federation.py b/synapse/storage/event_federation.py
index 24345b20a6..3faca2a042 100644
--- a/synapse/storage/event_federation.py
+++ b/synapse/storage/event_federation.py
@@ -376,33 +376,25 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore,
 
     @defer.inlineCallbacks
     def get_missing_events(self, room_id, earliest_events, latest_events,
-                           limit, min_depth):
+                           limit):
         ids = yield self.runInteraction(
             "get_missing_events",
             self._get_missing_events,
-            room_id, earliest_events, latest_events, limit, min_depth
+            room_id, earliest_events, latest_events, limit,
         )
-
         events = yield self._get_events(ids)
-
-        events = sorted(
-            [ev for ev in events if ev.depth >= min_depth],
-            key=lambda e: e.depth,
-        )
-
-        defer.returnValue(events[:limit])
+        defer.returnValue(events)
 
     def _get_missing_events(self, txn, room_id, earliest_events, latest_events,
-                            limit, min_depth):
-
-        earliest_events = set(earliest_events)
-        front = set(latest_events) - earliest_events
+                            limit):
 
-        event_results = set()
+        seen_events = set(earliest_events)
+        front = set(latest_events) - seen_events
+        event_results = []
 
         query = (
             "SELECT prev_event_id FROM event_edges "
-            "WHERE event_id = ? AND is_state = ? "
+            "WHERE room_id = ? AND event_id = ? AND is_state = ? "
             "LIMIT ?"
         )
 
@@ -411,18 +403,20 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore,
             for event_id in front:
                 txn.execute(
                     query,
-                    (event_id, False, limit - len(event_results))
+                    (room_id, event_id, False, limit - len(event_results))
                 )
 
-                for e_id, in txn:
-                    new_front.add(e_id)
+                new_results = set(t[0] for t in txn) - seen_events
 
-            new_front -= earliest_events
-            new_front -= event_results
+                new_front |= new_results
+                seen_events |= new_results
+                event_results.extend(new_results)
 
             front = new_front
-            event_results |= new_front
 
+        # we built the list working backwards from latest_events; we now need to
+        # reverse it so that the events are approximately chronological.
+        event_results.reverse()
         return event_results
 
 
diff --git a/synapse/storage/events.py b/synapse/storage/events.py
index 025a7fb6d9..8881b009df 100644
--- a/synapse/storage/events.py
+++ b/synapse/storage/events.py
@@ -19,7 +19,7 @@ import logging
 from collections import OrderedDict, deque, namedtuple
 from functools import wraps
 
-from six import iteritems
+from six import iteritems, text_type
 from six.moves import range
 
 from canonicaljson import json
@@ -34,10 +34,12 @@ from synapse.api.errors import SynapseError
 from synapse.events import EventBase  # noqa: F401
 from synapse.events.snapshot import EventContext  # noqa: F401
 from synapse.metrics.background_process_metrics import run_as_background_process
+from synapse.state import StateResolutionStore
 from synapse.storage.background_updates import BackgroundUpdateStore
 from synapse.storage.event_federation import EventFederationStore
 from synapse.storage.events_worker import EventsWorkerStore
 from synapse.types import RoomStreamToken, get_domain_from_id
+from synapse.util import batch_iter
 from synapse.util.async_helpers import ObservableDeferred
 from synapse.util.caches.descriptors import cached, cachedInlineCallbacks
 from synapse.util.frozenutils import frozendict_json_encoder
@@ -386,12 +388,10 @@ class EventsStore(EventFederationStore, EventsWorkerStore, BackgroundUpdateStore
                             )
 
                         for room_id, ev_ctx_rm in iteritems(events_by_room):
-                            # Work out new extremities by recursively adding and removing
-                            # the new events.
                             latest_event_ids = yield self.get_latest_event_ids_in_room(
                                 room_id
                             )
-                            new_latest_event_ids = yield self._calculate_new_extremeties(
+                            new_latest_event_ids = yield self._calculate_new_extremities(
                                 room_id, ev_ctx_rm, latest_event_ids
                             )
 
@@ -400,6 +400,12 @@ class EventsStore(EventFederationStore, EventsWorkerStore, BackgroundUpdateStore
                                 # No change in extremities, so no change in state
                                 continue
 
+                            # there should always be at least one forward extremity.
+                            # (except during the initial persistence of the send_join
+                            # results, in which case there will be no existing
+                            # extremities, so we'll `continue` above and skip this bit.)
+                            assert new_latest_event_ids, "No forward extremities left!"
+
                             new_forward_extremeties[room_id] = new_latest_event_ids
 
                             len_1 = (
@@ -517,44 +523,79 @@ class EventsStore(EventFederationStore, EventsWorkerStore, BackgroundUpdateStore
                     )
 
     @defer.inlineCallbacks
-    def _calculate_new_extremeties(self, room_id, event_contexts, latest_event_ids):
-        """Calculates the new forward extremeties for a room given events to
+    def _calculate_new_extremities(self, room_id, event_contexts, latest_event_ids):
+        """Calculates the new forward extremities for a room given events to
         persist.
 
         Assumes that we are only persisting events for one room at a time.
         """
-        new_latest_event_ids = set(latest_event_ids)
-        # First, add all the new events to the list
-        new_latest_event_ids.update(
-            event.event_id for event, ctx in event_contexts
+
+        # we're only interested in new events which aren't outliers and which aren't
+        # being rejected.
+        new_events = [
+            event for event, ctx in event_contexts
             if not event.internal_metadata.is_outlier() and not ctx.rejected
+        ]
+
+        # start with the existing forward extremities
+        result = set(latest_event_ids)
+
+        # add all the new events to the list
+        result.update(
+            event.event_id for event in new_events
         )
-        # Now remove all events that are referenced by the to-be-added events
-        new_latest_event_ids.difference_update(
+
+        # Now remove all events which are prev_events of any of the new events
+        result.difference_update(
             e_id
-            for event, ctx in event_contexts
+            for event in new_events
             for e_id, _ in event.prev_events
-            if not event.internal_metadata.is_outlier() and not ctx.rejected
         )
 
-        # And finally remove any events that are referenced by previously added
-        # events.
-        rows = yield self._simple_select_many_batch(
-            table="event_edges",
-            column="prev_event_id",
-            iterable=list(new_latest_event_ids),
-            retcols=["prev_event_id"],
-            keyvalues={
-                "is_state": False,
-            },
-            desc="_calculate_new_extremeties",
-        )
+        # Finally, remove any events which are prev_events of any existing events.
+        existing_prevs = yield self._get_events_which_are_prevs(result)
+        result.difference_update(existing_prevs)
 
-        new_latest_event_ids.difference_update(
-            row["prev_event_id"] for row in rows
-        )
+        defer.returnValue(result)
+
+    @defer.inlineCallbacks
+    def _get_events_which_are_prevs(self, event_ids):
+        """Filter the supplied list of event_ids to get those which are prev_events of
+        existing (non-outlier/rejected) events.
+
+        Args:
+            event_ids (Iterable[str]): event ids to filter
+
+        Returns:
+            Deferred[List[str]]: filtered event ids
+        """
+        results = []
+
+        def _get_events(txn, batch):
+            sql = """
+            SELECT prev_event_id
+            FROM event_edges
+                INNER JOIN events USING (event_id)
+                LEFT JOIN rejections USING (event_id)
+            WHERE
+                prev_event_id IN (%s)
+                AND NOT events.outlier
+                AND rejections.event_id IS NULL
+            """ % (
+                ",".join("?" for _ in batch),
+            )
+
+            txn.execute(sql, batch)
+            results.extend(r[0] for r in txn)
+
+        for chunk in batch_iter(event_ids, 100):
+            yield self.runInteraction(
+                "_get_events_which_are_prevs",
+                _get_events,
+                chunk,
+            )
 
-        defer.returnValue(new_latest_event_ids)
+        defer.returnValue(results)
 
     @defer.inlineCallbacks
     def _get_new_state_after_events(self, room_id, events_context, old_latest_event_ids,
@@ -586,10 +627,6 @@ class EventsStore(EventFederationStore, EventsWorkerStore, BackgroundUpdateStore
             the new current state is only returned if we've already calculated
             it.
         """
-
-        if not new_latest_event_ids:
-            return
-
         # map from state_group to ((type, key) -> event_id) state map
         state_groups_map = {}
 
@@ -695,19 +732,17 @@ class EventsStore(EventFederationStore, EventsWorkerStore, BackgroundUpdateStore
 
         # Ok, we need to defer to the state handler to resolve our state sets.
 
-        def get_events(ev_ids):
-            return self.get_events(
-                ev_ids, get_prev_content=False, check_redacted=False,
-            )
-
         state_groups = {
             sg: state_groups_map[sg] for sg in new_state_groups
         }
 
         events_map = {ev.event_id: ev for ev, _ in events_context}
+        room_version = yield self.get_room_version(room_id)
+
         logger.debug("calling resolve_state_groups from preserve_events")
         res = yield self._state_resolution_handler.resolve_state_groups(
-            room_id, state_groups, events_map, get_events
+            room_id, room_version, state_groups, events_map,
+            state_res_store=StateResolutionStore(self)
         )
 
         defer.returnValue((res.state, None))
@@ -816,6 +851,27 @@ class EventsStore(EventFederationStore, EventsWorkerStore, BackgroundUpdateStore
         # Insert into event_to_state_groups.
         self._store_event_state_mappings_txn(txn, events_and_contexts)
 
+        # We want to store event_auth mappings for rejected events, as they're
+        # used in state res v2.
+        # This is only necessary if the rejected event appears in an accepted
+        # event's auth chain, but its easier for now just to store them (and
+        # it doesn't take much storage compared to storing the entire event
+        # anyway).
+        self._simple_insert_many_txn(
+            txn,
+            table="event_auth",
+            values=[
+                {
+                    "event_id": event.event_id,
+                    "room_id": event.room_id,
+                    "auth_id": auth_id,
+                }
+                for event, _ in events_and_contexts
+                for auth_id, _ in event.auth_events
+                if event.is_state()
+            ],
+        )
+
         # _store_rejected_events_txn filters out any events which were
         # rejected, and returns the filtered list.
         events_and_contexts = self._store_rejected_events_txn(
@@ -928,6 +984,10 @@ class EventsStore(EventFederationStore, EventsWorkerStore, BackgroundUpdateStore
                 )
 
                 self._invalidate_cache_and_stream(
+                    txn, self.get_room_summary, (room_id,)
+                )
+
+                self._invalidate_cache_and_stream(
                     txn, self.get_current_state_ids, (room_id,)
                 )
 
@@ -1218,7 +1278,7 @@ class EventsStore(EventFederationStore, EventsWorkerStore, BackgroundUpdateStore
                     "sender": event.sender,
                     "contains_url": (
                         "url" in event.content
-                        and isinstance(event.content["url"], basestring)
+                        and isinstance(event.content["url"], text_type)
                     ),
                 }
                 for event, _ in events_and_contexts
@@ -1287,21 +1347,6 @@ class EventsStore(EventFederationStore, EventsWorkerStore, BackgroundUpdateStore
                     txn, event.room_id, event.redacts
                 )
 
-        self._simple_insert_many_txn(
-            txn,
-            table="event_auth",
-            values=[
-                {
-                    "event_id": event.event_id,
-                    "room_id": event.room_id,
-                    "auth_id": auth_id,
-                }
-                for event, _ in events_and_contexts
-                for auth_id, _ in event.auth_events
-                if event.is_state()
-            ],
-        )
-
         # Update the event_forward_extremities, event_backward_extremities and
         # event_edges tables.
         self._handle_mult_prev_events(
@@ -1527,7 +1572,7 @@ class EventsStore(EventFederationStore, EventsWorkerStore, BackgroundUpdateStore
 
                     contains_url = "url" in content
                     if contains_url:
-                        contains_url &= isinstance(content["url"], basestring)
+                        contains_url &= isinstance(content["url"], text_type)
                 except (KeyError, AttributeError):
                     # If the event is missing a necessary field then
                     # skip over it.
@@ -1884,20 +1929,6 @@ class EventsStore(EventFederationStore, EventsWorkerStore, BackgroundUpdateStore
             ")"
         )
 
-        # create an index on should_delete because later we'll be looking for
-        # the should_delete / shouldn't_delete subsets
-        txn.execute(
-            "CREATE INDEX events_to_purge_should_delete"
-            " ON events_to_purge(should_delete)",
-        )
-
-        # We do joins against events_to_purge for e.g. calculating state
-        # groups to purge, etc., so lets make an index.
-        txn.execute(
-            "CREATE INDEX events_to_purge_id"
-            " ON events_to_purge(event_id)",
-        )
-
         # First ensure that we're not about to delete all the forward extremeties
         txn.execute(
             "SELECT e.event_id, e.depth FROM events as e "
@@ -1908,9 +1939,9 @@ class EventsStore(EventFederationStore, EventsWorkerStore, BackgroundUpdateStore
             (room_id,)
         )
         rows = txn.fetchall()
-        max_depth = max(row[0] for row in rows)
+        max_depth = max(row[1] for row in rows)
 
-        if max_depth <= token.topological:
+        if max_depth < token.topological:
             # We need to ensure we don't delete all the events from the database
             # otherwise we wouldn't be able to send any events (due to not
             # having any backwards extremeties)
@@ -1924,19 +1955,45 @@ class EventsStore(EventFederationStore, EventsWorkerStore, BackgroundUpdateStore
         should_delete_params = ()
         if not delete_local_events:
             should_delete_expr += " AND event_id NOT LIKE ?"
-            should_delete_params += ("%:" + self.hs.hostname, )
+
+            # We include the parameter twice since we use the expression twice
+            should_delete_params += (
+                "%:" + self.hs.hostname,
+                "%:" + self.hs.hostname,
+            )
 
         should_delete_params += (room_id, token.topological)
 
+        # Note that we insert events that are outliers and aren't going to be
+        # deleted, as nothing will happen to them.
         txn.execute(
             "INSERT INTO events_to_purge"
             " SELECT event_id, %s"
             " FROM events AS e LEFT JOIN state_events USING (event_id)"
-            " WHERE e.room_id = ? AND topological_ordering < ?" % (
+            " WHERE (NOT outlier OR (%s)) AND e.room_id = ? AND topological_ordering < ?"
+            % (
+                should_delete_expr,
                 should_delete_expr,
             ),
             should_delete_params,
         )
+
+        # We create the indices *after* insertion as that's a lot faster.
+
+        # create an index on should_delete because later we'll be looking for
+        # the should_delete / shouldn't_delete subsets
+        txn.execute(
+            "CREATE INDEX events_to_purge_should_delete"
+            " ON events_to_purge(should_delete)",
+        )
+
+        # We do joins against events_to_purge for e.g. calculating state
+        # groups to purge, etc., so lets make an index.
+        txn.execute(
+            "CREATE INDEX events_to_purge_id"
+            " ON events_to_purge(event_id)",
+        )
+
         txn.execute(
             "SELECT event_id, should_delete FROM events_to_purge"
         )
@@ -2032,7 +2089,7 @@ class EventsStore(EventFederationStore, EventsWorkerStore, BackgroundUpdateStore
         for sg in remaining_state_groups:
             logger.info("[purge] de-delta-ing remaining state group %s", sg)
             curr_state = self._get_state_groups_from_groups_txn(
-                txn, [sg], types=None
+                txn, [sg],
             )
             curr_state = curr_state[sg]
 
diff --git a/synapse/storage/events_worker.py b/synapse/storage/events_worker.py
index 59822178ff..a8326f5296 100644
--- a/synapse/storage/events_worker.py
+++ b/synapse/storage/events_worker.py
@@ -12,6 +12,7 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+
 import itertools
 import logging
 from collections import namedtuple
@@ -265,7 +266,7 @@ class EventsWorkerStore(SQLBaseStore):
         """
         with Measure(self._clock, "_fetch_event_list"):
             try:
-                event_id_lists = zip(*event_list)[0]
+                event_id_lists = list(zip(*event_list))[0]
                 event_ids = [
                     item for sublist in event_id_lists for item in sublist
                 ]
@@ -299,14 +300,14 @@ class EventsWorkerStore(SQLBaseStore):
                 logger.exception("do_fetch")
 
                 # We only want to resolve deferreds from the main thread
-                def fire(evs):
+                def fire(evs, exc):
                     for _, d in evs:
                         if not d.called:
                             with PreserveLoggingContext():
-                                d.errback(e)
+                                d.errback(exc)
 
                 with PreserveLoggingContext():
-                    self.hs.get_reactor().callFromThread(fire, event_list)
+                    self.hs.get_reactor().callFromThread(fire, event_list, e)
 
     @defer.inlineCallbacks
     def _enqueue_events(self, events, check_redacted=True, allow_rejected=False):
diff --git a/synapse/storage/filtering.py b/synapse/storage/filtering.py
index 2d5896c5b4..6ddcc909bf 100644
--- a/synapse/storage/filtering.py
+++ b/synapse/storage/filtering.py
@@ -13,14 +13,14 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from canonicaljson import encode_canonical_json, json
+from canonicaljson import encode_canonical_json
 
 from twisted.internet import defer
 
 from synapse.api.errors import Codes, SynapseError
 from synapse.util.caches.descriptors import cachedInlineCallbacks
 
-from ._base import SQLBaseStore
+from ._base import SQLBaseStore, db_to_json
 
 
 class FilteringStore(SQLBaseStore):
@@ -44,7 +44,7 @@ class FilteringStore(SQLBaseStore):
             desc="get_user_filter",
         )
 
-        defer.returnValue(json.loads(bytes(def_json).decode("utf-8")))
+        defer.returnValue(db_to_json(def_json))
 
     def add_user_filter(self, user_localpart, user_filter):
         def_json = encode_canonical_json(user_filter)
diff --git a/synapse/storage/keys.py b/synapse/storage/keys.py
index f547977600..8af17921e3 100644
--- a/synapse/storage/keys.py
+++ b/synapse/storage/keys.py
@@ -32,7 +32,7 @@ logger = logging.getLogger(__name__)
 # py2 sqlite has buffer hardcoded as only binary type, so we must use it,
 # despite being deprecated and removed in favor of memoryview
 if six.PY2:
-    db_binary_type = buffer
+    db_binary_type = six.moves.builtins.buffer
 else:
     db_binary_type = memoryview
 
@@ -134,6 +134,7 @@ class KeyStore(SQLBaseStore):
         """
         key_id = "%s:%s" % (verify_key.alg, verify_key.version)
 
+        # XXX fix this to not need a lock (#3819)
         def _txn(txn):
             self._simple_upsert_txn(
                 txn,
diff --git a/synapse/storage/monthly_active_users.py b/synapse/storage/monthly_active_users.py
index 06f9a75a97..cf4104dc2e 100644
--- a/synapse/storage/monthly_active_users.py
+++ b/synapse/storage/monthly_active_users.py
@@ -33,20 +33,29 @@ class MonthlyActiveUsersStore(SQLBaseStore):
         self._clock = hs.get_clock()
         self.hs = hs
         self.reserved_users = ()
+        # Do not add more reserved users than the total allowable number
+        self._initialise_reserved_users(
+            dbconn.cursor(),
+            hs.config.mau_limits_reserved_threepids[:self.hs.config.max_mau_value],
+        )
 
-    @defer.inlineCallbacks
-    def initialise_reserved_users(self, threepids):
-        # TODO Why can't I do this in init?
-        store = self.hs.get_datastore()
+    def _initialise_reserved_users(self, txn, threepids):
+        """Ensures that reserved threepids are accounted for in the MAU table, should
+        be called on start up.
+
+        Args:
+            txn (cursor):
+            threepids (list[dict]): List of threepid dicts to reserve
+        """
         reserved_user_list = []
 
-        # Do not add more reserved users than the total allowable number
-        for tp in threepids[:self.hs.config.max_mau_value]:
-            user_id = yield store.get_user_id_by_threepid(
+        for tp in threepids:
+            user_id = self.get_user_id_by_threepid_txn(
+                txn,
                 tp["medium"], tp["address"]
             )
             if user_id:
-                yield self.upsert_monthly_active_user(user_id)
+                self.upsert_monthly_active_user_txn(txn, user_id)
                 reserved_user_list.append(user_id)
             else:
                 logger.warning(
@@ -56,8 +65,7 @@ class MonthlyActiveUsersStore(SQLBaseStore):
 
     @defer.inlineCallbacks
     def reap_monthly_active_users(self):
-        """
-        Cleans out monthly active user table to ensure that no stale
+        """Cleans out monthly active user table to ensure that no stale
         entries exist.
 
         Returns:
@@ -147,16 +155,63 @@ class MonthlyActiveUsersStore(SQLBaseStore):
             return count
         return self.runInteraction("count_users", _count_users)
 
+    @defer.inlineCallbacks
+    def get_registered_reserved_users_count(self):
+        """Of the reserved threepids defined in config, how many are associated
+        with registered users?
+
+        Returns:
+            Defered[int]: Number of real reserved users
+        """
+        count = 0
+        for tp in self.hs.config.mau_limits_reserved_threepids:
+            user_id = yield self.hs.get_datastore().get_user_id_by_threepid(
+                tp["medium"], tp["address"]
+            )
+            if user_id:
+                count = count + 1
+        defer.returnValue(count)
+
+    @defer.inlineCallbacks
     def upsert_monthly_active_user(self, user_id):
+        """Updates or inserts the user into the monthly active user table, which
+        is used to track the current MAU usage of the server
+
+        Args:
+            user_id (str): user to add/update
         """
-            Updates or inserts monthly active user member
-            Arguments:
-                user_id (str): user to add/update
-            Deferred[bool]: True if a new entry was created, False if an
-                existing one was updated.
+        is_insert = yield self.runInteraction(
+            "upsert_monthly_active_user", self.upsert_monthly_active_user_txn,
+            user_id
+        )
+
+        if is_insert:
+            self.user_last_seen_monthly_active.invalidate((user_id,))
+            self.get_monthly_active_count.invalidate(())
+
+    def upsert_monthly_active_user_txn(self, txn, user_id):
+        """Updates or inserts monthly active user member
+
+        Note that, after calling this method, it will generally be necessary
+        to invalidate the caches on user_last_seen_monthly_active and
+        get_monthly_active_count. We can't do that here, because we are running
+        in a database thread rather than the main thread, and we can't call
+        txn.call_after because txn may not be a LoggingTransaction.
+
+        Args:
+            txn (cursor):
+            user_id (str): user to add/update
+
+        Returns:
+            bool: True if a new entry was created, False if an
+            existing one was updated.
         """
-        is_insert = self._simple_upsert(
-            desc="upsert_monthly_active_user",
+        # Am consciously deciding to lock the table on the basis that is ought
+        # never be a big table and alternative approaches (batching multiple
+        # upserts into a single txn) introduced a lot of extra complexity.
+        # See https://github.com/matrix-org/synapse/issues/3854 for more
+        is_insert = self._simple_upsert_txn(
+            txn,
             table="monthly_active_users",
             keyvalues={
                 "user_id": user_id,
@@ -164,11 +219,9 @@ class MonthlyActiveUsersStore(SQLBaseStore):
             values={
                 "timestamp": int(self._clock.time_msec()),
             },
-            lock=False,
         )
-        if is_insert:
-            self.user_last_seen_monthly_active.invalidate((user_id,))
-            self.get_monthly_active_count.invalidate(())
+
+        return is_insert
 
     @cached(num_args=1)
     def user_last_seen_monthly_active(self, user_id):
@@ -199,7 +252,16 @@ class MonthlyActiveUsersStore(SQLBaseStore):
         Args:
             user_id(str): the user_id to query
         """
+
         if self.hs.config.limit_usage_by_mau:
+            # Trial users and guests should not be included as part of MAU group
+            is_guest = yield self.is_guest(user_id)
+            if is_guest:
+                return
+            is_trial = yield self.is_trial_user(user_id)
+            if is_trial:
+                return
+
             last_seen_timestamp = yield self.user_last_seen_monthly_active(user_id)
             now = self.hs.get_clock().time_msec()
 
diff --git a/synapse/storage/profile.py b/synapse/storage/profile.py
index e6848c70a0..10133f0a4a 100644
--- a/synapse/storage/profile.py
+++ b/synapse/storage/profile.py
@@ -123,8 +123,6 @@ class ProfileWorkerStore(SQLBaseStore):
             desc="get_from_remote_profile_cache",
         )
 
-
-class ProfileStore(ProfileWorkerStore):
     def set_profile_displayname(self, user_localpart, new_displayname, batchnum):
         return self._simple_upsert(
             table="profiles",
@@ -165,6 +163,8 @@ class ProfileStore(ProfileWorkerStore):
             lock=False  # we can do this because user_id has a unique index
         )
 
+
+class ProfileStore(ProfileWorkerStore):
     def add_remote_profile_cache(self, user_id, displayname, avatar_url):
         """Ensure we are caching the remote user's profiles.
 
diff --git a/synapse/storage/pusher.py b/synapse/storage/pusher.py
index 8443bd4c1b..2743b52bad 100644
--- a/synapse/storage/pusher.py
+++ b/synapse/storage/pusher.py
@@ -15,7 +15,8 @@
 # limitations under the License.
 
 import logging
-import types
+
+import six
 
 from canonicaljson import encode_canonical_json, json
 
@@ -27,6 +28,11 @@ from ._base import SQLBaseStore
 
 logger = logging.getLogger(__name__)
 
+if six.PY2:
+    db_binary_type = six.moves.builtins.buffer
+else:
+    db_binary_type = memoryview
+
 
 class PusherWorkerStore(SQLBaseStore):
     def _decode_pushers_rows(self, rows):
@@ -34,18 +40,18 @@ class PusherWorkerStore(SQLBaseStore):
             dataJson = r['data']
             r['data'] = None
             try:
-                if isinstance(dataJson, types.BufferType):
+                if isinstance(dataJson, db_binary_type):
                     dataJson = str(dataJson).decode("UTF8")
 
                 r['data'] = json.loads(dataJson)
             except Exception as e:
                 logger.warn(
                     "Invalid JSON in data for pusher %d: %s, %s",
-                    r['id'], dataJson, e.message,
+                    r['id'], dataJson, e.args[0],
                 )
                 pass
 
-            if isinstance(r['pushkey'], types.BufferType):
+            if isinstance(r['pushkey'], db_binary_type):
                 r['pushkey'] = str(r['pushkey']).decode("UTF8")
 
         return rows
diff --git a/synapse/storage/registration.py b/synapse/storage/registration.py
index 50706519aa..65061f4c61 100644
--- a/synapse/storage/registration.py
+++ b/synapse/storage/registration.py
@@ -26,6 +26,11 @@ from synapse.util.caches.descriptors import cached, cachedInlineCallbacks
 
 
 class RegistrationWorkerStore(SQLBaseStore):
+    def __init__(self, db_conn, hs):
+        super(RegistrationWorkerStore, self).__init__(db_conn, hs)
+
+        self.config = hs.config
+
     @cached()
     def get_user_by_id(self, user_id):
         return self._simple_select_one(
@@ -36,12 +41,33 @@ class RegistrationWorkerStore(SQLBaseStore):
             retcols=[
                 "name", "password_hash", "is_guest",
                 "consent_version", "consent_server_notice_sent",
-                "appservice_id",
+                "appservice_id", "creation_ts",
             ],
             allow_none=True,
             desc="get_user_by_id",
         )
 
+    @defer.inlineCallbacks
+    def is_trial_user(self, user_id):
+        """Checks if user is in the "trial" period, i.e. within the first
+        N days of registration defined by `mau_trial_days` config
+
+        Args:
+            user_id (str)
+
+        Returns:
+            Deferred[bool]
+        """
+
+        info = yield self.get_user_by_id(user_id)
+        if not info:
+            defer.returnValue(False)
+
+        now = self.clock.time_msec()
+        trial_duration_ms = self.config.mau_trial_days * 24 * 60 * 60 * 1000
+        is_trial = (now - info["creation_ts"] * 1000) < trial_duration_ms
+        defer.returnValue(is_trial)
+
     @cached()
     def get_user_by_access_token(self, token):
         """Get a user from the given access token.
@@ -436,17 +462,44 @@ class RegistrationStore(RegistrationWorkerStore,
 
     @defer.inlineCallbacks
     def get_user_id_by_threepid(self, medium, address):
-        ret = yield self._simple_select_one(
+        """Returns user id from threepid
+
+        Args:
+            medium (str): threepid medium e.g. email
+            address (str): threepid address e.g. me@example.com
+
+        Returns:
+            Deferred[str|None]: user id or None if no user id/threepid mapping exists
+        """
+        user_id = yield self.runInteraction(
+            "get_user_id_by_threepid", self.get_user_id_by_threepid_txn,
+            medium, address
+        )
+        defer.returnValue(user_id)
+
+    def get_user_id_by_threepid_txn(self, txn, medium, address):
+        """Returns user id from threepid
+
+        Args:
+            txn (cursor):
+            medium (str): threepid medium e.g. email
+            address (str): threepid address e.g. me@example.com
+
+        Returns:
+            str|None: user id or None if no user id/threepid mapping exists
+        """
+        ret = self._simple_select_one_txn(
+            txn,
             "user_threepids",
             {
                 "medium": medium,
                 "address": address
             },
-            ['user_id'], True, 'get_user_id_by_threepid'
+            ['user_id'], True
         )
         if ret:
-            defer.returnValue(ret['user_id'])
-        defer.returnValue(None)
+            return ret['user_id']
+        return None
 
     def user_delete_threepid(self, user_id, medium, address):
         return self._simple_delete(
@@ -529,7 +582,7 @@ class RegistrationStore(RegistrationWorkerStore,
         def _find_next_generated_user_id(txn):
             txn.execute("SELECT name FROM users")
 
-            regex = re.compile("^@(\d+):")
+            regex = re.compile(r"^@(\d+):")
 
             found = set()
 
diff --git a/synapse/storage/room.py b/synapse/storage/room.py
index 3378fc77d1..61013b8919 100644
--- a/synapse/storage/room.py
+++ b/synapse/storage/room.py
@@ -186,6 +186,35 @@ class RoomWorkerStore(SQLBaseStore):
             desc="is_room_blocked",
         )
 
+    @cachedInlineCallbacks(max_entries=10000)
+    def get_ratelimit_for_user(self, user_id):
+        """Check if there are any overrides for ratelimiting for the given
+        user
+
+        Args:
+            user_id (str)
+
+        Returns:
+            RatelimitOverride if there is an override, else None. If the contents
+            of RatelimitOverride are None or 0 then ratelimitng has been
+            disabled for that user entirely.
+        """
+        row = yield self._simple_select_one(
+            table="ratelimit_override",
+            keyvalues={"user_id": user_id},
+            retcols=("messages_per_second", "burst_count"),
+            allow_none=True,
+            desc="get_ratelimit_for_user",
+        )
+
+        if row:
+            defer.returnValue(RatelimitOverride(
+                messages_per_second=row["messages_per_second"],
+                burst_count=row["burst_count"],
+            ))
+        else:
+            defer.returnValue(None)
+
 
 class RoomStore(RoomWorkerStore, SearchStore):
 
@@ -469,35 +498,6 @@ class RoomStore(RoomWorkerStore, SearchStore):
             "get_all_new_public_rooms", get_all_new_public_rooms
         )
 
-    @cachedInlineCallbacks(max_entries=10000)
-    def get_ratelimit_for_user(self, user_id):
-        """Check if there are any overrides for ratelimiting for the given
-        user
-
-        Args:
-            user_id (str)
-
-        Returns:
-            RatelimitOverride if there is an override, else None. If the contents
-            of RatelimitOverride are None or 0 then ratelimitng has been
-            disabled for that user entirely.
-        """
-        row = yield self._simple_select_one(
-            table="ratelimit_override",
-            keyvalues={"user_id": user_id},
-            retcols=("messages_per_second", "burst_count"),
-            allow_none=True,
-            desc="get_ratelimit_for_user",
-        )
-
-        if row:
-            defer.returnValue(RatelimitOverride(
-                messages_per_second=row["messages_per_second"],
-                burst_count=row["burst_count"],
-            ))
-        else:
-            defer.returnValue(None)
-
     @defer.inlineCallbacks
     def block_room(self, room_id, user_id):
         yield self._simple_insert(
diff --git a/synapse/storage/roommember.py b/synapse/storage/roommember.py
index 9b4e6d6aa8..0707f9a86a 100644
--- a/synapse/storage/roommember.py
+++ b/synapse/storage/roommember.py
@@ -51,6 +51,12 @@ ProfileInfo = namedtuple(
     "ProfileInfo", ("avatar_url", "display_name")
 )
 
+# "members" points to a truncated list of (user_id, event_id) tuples for users of
+# a given membership type, suitable for use in calculating heroes for a room.
+# "count" points to the total numberr of users of a given membership type.
+MemberSummary = namedtuple(
+    "MemberSummary", ("members", "count")
+)
 
 _MEMBERSHIP_PROFILE_UPDATE_NAME = "room_membership_profile_update"
 
@@ -82,6 +88,65 @@ class RoomMemberWorkerStore(EventsWorkerStore):
             return [to_ascii(r[0]) for r in txn]
         return self.runInteraction("get_users_in_room", f)
 
+    @cached(max_entries=100000)
+    def get_room_summary(self, room_id):
+        """ Get the details of a room roughly suitable for use by the room
+        summary extension to /sync. Useful when lazy loading room members.
+        Args:
+            room_id (str): The room ID to query
+        Returns:
+            Deferred[dict[str, MemberSummary]:
+                dict of membership states, pointing to a MemberSummary named tuple.
+        """
+
+        def _get_room_summary_txn(txn):
+            # first get counts.
+            # We do this all in one transaction to keep the cache small.
+            # FIXME: get rid of this when we have room_stats
+            sql = """
+                SELECT count(*), m.membership FROM room_memberships as m
+                 INNER JOIN current_state_events as c
+                 ON m.event_id = c.event_id
+                 AND m.room_id = c.room_id
+                 AND m.user_id = c.state_key
+                 WHERE c.type = 'm.room.member' AND c.room_id = ?
+                 GROUP BY m.membership
+            """
+
+            txn.execute(sql, (room_id,))
+            res = {}
+            for count, membership in txn:
+                summary = res.setdefault(to_ascii(membership), MemberSummary([], count))
+
+            # we order by membership and then fairly arbitrarily by event_id so
+            # heroes are consistent
+            sql = """
+                SELECT m.user_id, m.membership, m.event_id
+                FROM room_memberships as m
+                 INNER JOIN current_state_events as c
+                 ON m.event_id = c.event_id
+                 AND m.room_id = c.room_id
+                 AND m.user_id = c.state_key
+                 WHERE c.type = 'm.room.member' AND c.room_id = ?
+                 ORDER BY
+                    CASE m.membership WHEN ? THEN 1 WHEN ? THEN 2 ELSE 3 END ASC,
+                    m.event_id ASC
+                 LIMIT ?
+            """
+
+            # 6 is 5 (number of heroes) plus 1, in case one of them is the calling user.
+            txn.execute(sql, (room_id, Membership.JOIN, Membership.INVITE, 6))
+            for user_id, membership, event_id in txn:
+                summary = res[to_ascii(membership)]
+                # we will always have a summary for this membership type at this
+                # point given the summary currently contains the counts.
+                members = summary.members
+                members.append((to_ascii(user_id), to_ascii(event_id)))
+
+            return res
+
+        return self.runInteraction("get_room_summary", _get_room_summary_txn)
+
     @cached()
     def get_invited_rooms_for_user(self, user_id):
         """ Get all the rooms the user is invited to
diff --git a/synapse/storage/schema/delta/51/e2e_room_keys.sql b/synapse/storage/schema/delta/51/e2e_room_keys.sql
new file mode 100644
index 0000000000..c0e66a697d
--- /dev/null
+++ b/synapse/storage/schema/delta/51/e2e_room_keys.sql
@@ -0,0 +1,39 @@
+/* Copyright 2017 New Vector Ltd
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *    http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+-- users' optionally backed up encrypted e2e sessions
+CREATE TABLE e2e_room_keys (
+    user_id TEXT NOT NULL,
+    room_id TEXT NOT NULL,
+    session_id TEXT NOT NULL,
+    version TEXT NOT NULL,
+    first_message_index INT,
+    forwarded_count INT,
+    is_verified BOOLEAN,
+    session_data TEXT NOT NULL
+);
+
+CREATE UNIQUE INDEX e2e_room_keys_idx ON e2e_room_keys(user_id, room_id, session_id);
+
+-- the metadata for each generation of encrypted e2e session backups
+CREATE TABLE e2e_room_keys_versions (
+    user_id TEXT NOT NULL,
+    version TEXT NOT NULL,
+    algorithm TEXT NOT NULL,
+    auth_data TEXT NOT NULL,
+    deleted SMALLINT DEFAULT 0 NOT NULL
+);
+
+CREATE UNIQUE INDEX e2e_room_keys_versions_idx ON e2e_room_keys_versions(user_id, version);
diff --git a/synapse/storage/signatures.py b/synapse/storage/signatures.py
index 5623391f6e..158e9dbe7b 100644
--- a/synapse/storage/signatures.py
+++ b/synapse/storage/signatures.py
@@ -27,7 +27,7 @@ from ._base import SQLBaseStore
 # py2 sqlite has buffer hardcoded as only binary type, so we must use it,
 # despite being deprecated and removed in favor of memoryview
 if six.PY2:
-    db_binary_type = buffer
+    db_binary_type = six.moves.builtins.buffer
 else:
     db_binary_type = memoryview
 
diff --git a/synapse/storage/state.py b/synapse/storage/state.py
index dd03c4168b..ef65929bb2 100644
--- a/synapse/storage/state.py
+++ b/synapse/storage/state.py
@@ -19,6 +19,8 @@ from collections import namedtuple
 from six import iteritems, itervalues
 from six.moves import range
 
+import attr
+
 from twisted.internet import defer
 
 from synapse.api.constants import EventTypes
@@ -48,6 +50,318 @@ class _GetStateGroupDelta(namedtuple("_GetStateGroupDelta", ("prev_group", "delt
         return len(self.delta_ids) if self.delta_ids else 0
 
 
+@attr.s(slots=True)
+class StateFilter(object):
+    """A filter used when querying for state.
+
+    Attributes:
+        types (dict[str, set[str]|None]): Map from type to set of state keys (or
+            None). This specifies which state_keys for the given type to fetch
+            from the DB. If None then all events with that type are fetched. If
+            the set is empty then no events with that type are fetched.
+        include_others (bool): Whether to fetch events with types that do not
+            appear in `types`.
+    """
+
+    types = attr.ib()
+    include_others = attr.ib(default=False)
+
+    def __attrs_post_init__(self):
+        # If `include_others` is set we canonicalise the filter by removing
+        # wildcards from the types dictionary
+        if self.include_others:
+            self.types = {
+                k: v for k, v in iteritems(self.types)
+                if v is not None
+            }
+
+    @staticmethod
+    def all():
+        """Creates a filter that fetches everything.
+
+        Returns:
+            StateFilter
+        """
+        return StateFilter(types={}, include_others=True)
+
+    @staticmethod
+    def none():
+        """Creates a filter that fetches nothing.
+
+        Returns:
+            StateFilter
+        """
+        return StateFilter(types={}, include_others=False)
+
+    @staticmethod
+    def from_types(types):
+        """Creates a filter that only fetches the given types
+
+        Args:
+            types (Iterable[tuple[str, str|None]]): A list of type and state
+                keys to fetch. A state_key of None fetches everything for
+                that type
+
+        Returns:
+            StateFilter
+        """
+        type_dict = {}
+        for typ, s in types:
+            if typ in type_dict:
+                if type_dict[typ] is None:
+                    continue
+
+            if s is None:
+                type_dict[typ] = None
+                continue
+
+            type_dict.setdefault(typ, set()).add(s)
+
+        return StateFilter(types=type_dict)
+
+    @staticmethod
+    def from_lazy_load_member_list(members):
+        """Creates a filter that returns all non-member events, plus the member
+        events for the given users
+
+        Args:
+            members (iterable[str]): Set of user IDs
+
+        Returns:
+            StateFilter
+        """
+        return StateFilter(
+            types={EventTypes.Member: set(members)},
+            include_others=True,
+        )
+
+    def return_expanded(self):
+        """Creates a new StateFilter where type wild cards have been removed
+        (except for memberships). The returned filter is a superset of the
+        current one, i.e. anything that passes the current filter will pass
+        the returned filter.
+
+        This helps the caching as the DictionaryCache knows if it has *all* the
+        state, but does not know if it has all of the keys of a particular type,
+        which makes wildcard lookups expensive unless we have a complete cache.
+        Hence, if we are doing a wildcard lookup, populate the cache fully so
+        that we can do an efficient lookup next time.
+
+        Note that since we have two caches, one for membership events and one for
+        other events, we can be a bit more clever than simply returning
+        `StateFilter.all()` if `has_wildcards()` is True.
+
+        We return a StateFilter where:
+            1. the list of membership events to return is the same
+            2. if there is a wildcard that matches non-member events we
+               return all non-member events
+
+        Returns:
+            StateFilter
+        """
+
+        if self.is_full():
+            # If we're going to return everything then there's nothing to do
+            return self
+
+        if not self.has_wildcards():
+            # If there are no wild cards, there's nothing to do
+            return self
+
+        if EventTypes.Member in self.types:
+            get_all_members = self.types[EventTypes.Member] is None
+        else:
+            get_all_members = self.include_others
+
+        has_non_member_wildcard = self.include_others or any(
+            state_keys is None
+            for t, state_keys in iteritems(self.types)
+            if t != EventTypes.Member
+        )
+
+        if not has_non_member_wildcard:
+            # If there are no non-member wild cards we can just return ourselves
+            return self
+
+        if get_all_members:
+            # We want to return everything.
+            return StateFilter.all()
+        else:
+            # We want to return all non-members, but only particular
+            # memberships
+            return StateFilter(
+                types={EventTypes.Member: self.types[EventTypes.Member]},
+                include_others=True,
+            )
+
+    def make_sql_filter_clause(self):
+        """Converts the filter to an SQL clause.
+
+        For example:
+
+            f = StateFilter.from_types([("m.room.create", "")])
+            clause, args = f.make_sql_filter_clause()
+            clause == "(type = ? AND state_key = ?)"
+            args == ['m.room.create', '']
+
+
+        Returns:
+            tuple[str, list]: The SQL string (may be empty) and arguments. An
+            empty SQL string is returned when the filter matches everything
+            (i.e. is "full").
+        """
+
+        where_clause = ""
+        where_args = []
+
+        if self.is_full():
+            return where_clause, where_args
+
+        if not self.include_others and not self.types:
+            # i.e. this is an empty filter, so we need to return a clause that
+            # will match nothing
+            return "1 = 2", []
+
+        # First we build up a lost of clauses for each type/state_key combo
+        clauses = []
+        for etype, state_keys in iteritems(self.types):
+            if state_keys is None:
+                clauses.append("(type = ?)")
+                where_args.append(etype)
+                continue
+
+            for state_key in state_keys:
+                clauses.append("(type = ? AND state_key = ?)")
+                where_args.extend((etype, state_key))
+
+        # This will match anything that appears in `self.types`
+        where_clause = " OR ".join(clauses)
+
+        # If we want to include stuff that's not in the types dict then we add
+        # a `OR type NOT IN (...)` clause to the end.
+        if self.include_others:
+            if where_clause:
+                where_clause += " OR "
+
+            where_clause += "type NOT IN (%s)" % (
+                ",".join(["?"] * len(self.types)),
+            )
+            where_args.extend(self.types)
+
+        return where_clause, where_args
+
+    def max_entries_returned(self):
+        """Returns the maximum number of entries this filter will return if
+        known, otherwise returns None.
+
+        For example a simple state filter asking for `("m.room.create", "")`
+        will return 1, whereas the default state filter will return None.
+
+        This is used to bail out early if the right number of entries have been
+        fetched.
+        """
+        if self.has_wildcards():
+            return None
+
+        return len(self.concrete_types())
+
+    def filter_state(self, state_dict):
+        """Returns the state filtered with by this StateFilter
+
+        Args:
+            state (dict[tuple[str, str], Any]): The state map to filter
+
+        Returns:
+            dict[tuple[str, str], Any]: The filtered state map
+        """
+        if self.is_full():
+            return dict(state_dict)
+
+        filtered_state = {}
+        for k, v in iteritems(state_dict):
+            typ, state_key = k
+            if typ in self.types:
+                state_keys = self.types[typ]
+                if state_keys is None or state_key in state_keys:
+                    filtered_state[k] = v
+            elif self.include_others:
+                filtered_state[k] = v
+
+        return filtered_state
+
+    def is_full(self):
+        """Whether this filter fetches everything or not
+
+        Returns:
+            bool
+        """
+        return self.include_others and not self.types
+
+    def has_wildcards(self):
+        """Whether the filter includes wildcards or is attempting to fetch
+        specific state.
+
+        Returns:
+            bool
+        """
+
+        return (
+            self.include_others
+            or any(
+                state_keys is None
+                for state_keys in itervalues(self.types)
+            )
+        )
+
+    def concrete_types(self):
+        """Returns a list of concrete type/state_keys (i.e. not None) that
+        will be fetched. This will be a complete list if `has_wildcards`
+        returns False, but otherwise will be a subset (or even empty).
+
+        Returns:
+            list[tuple[str,str]]
+        """
+        return [
+            (t, s)
+            for t, state_keys in iteritems(self.types)
+            if state_keys is not None
+            for s in state_keys
+        ]
+
+    def get_member_split(self):
+        """Return the filter split into two: one which assumes it's exclusively
+        matching against member state, and one which assumes it's matching
+        against non member state.
+
+        This is useful due to the returned filters giving correct results for
+        `is_full()`, `has_wildcards()`, etc, when operating against maps that
+        either exclusively contain member events or only contain non-member
+        events. (Which is the case when dealing with the member vs non-member
+        state caches).
+
+        Returns:
+            tuple[StateFilter, StateFilter]: The member and non member filters
+        """
+
+        if EventTypes.Member in self.types:
+            state_keys = self.types[EventTypes.Member]
+            if state_keys is None:
+                member_filter = StateFilter.all()
+            else:
+                member_filter = StateFilter({EventTypes.Member: state_keys})
+        elif self.include_others:
+            member_filter = StateFilter.all()
+        else:
+            member_filter = StateFilter.none()
+
+        non_member_filter = StateFilter(
+            types={k: v for k, v in iteritems(self.types) if k != EventTypes.Member},
+            include_others=self.include_others,
+        )
+
+        return member_filter, non_member_filter
+
+
 # this inherits from EventsWorkerStore because it calls self.get_events
 class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore):
     """The parts of StateGroupStore that can be called from workers.
@@ -60,8 +374,43 @@ class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore):
     def __init__(self, db_conn, hs):
         super(StateGroupWorkerStore, self).__init__(db_conn, hs)
 
+        # Originally the state store used a single DictionaryCache to cache the
+        # event IDs for the state types in a given state group to avoid hammering
+        # on the state_group* tables.
+        #
+        # The point of using a DictionaryCache is that it can cache a subset
+        # of the state events for a given state group (i.e. a subset of the keys for a
+        # given dict which is an entry in the cache for a given state group ID).
+        #
+        # However, this poses problems when performing complicated queries
+        # on the store - for instance: "give me all the state for this group, but
+        # limit members to this subset of users", as DictionaryCache's API isn't
+        # rich enough to say "please cache any of these fields, apart from this subset".
+        # This is problematic when lazy loading members, which requires this behaviour,
+        # as without it the cache has no choice but to speculatively load all
+        # state events for the group, which negates the efficiency being sought.
+        #
+        # Rather than overcomplicating DictionaryCache's API, we instead split the
+        # state_group_cache into two halves - one for tracking non-member events,
+        # and the other for tracking member_events.  This means that lazy loading
+        # queries can be made in a cache-friendly manner by querying both caches
+        # separately and then merging the result.  So for the example above, you
+        # would query the members cache for a specific subset of state keys
+        # (which DictionaryCache will handle efficiently and fine) and the non-members
+        # cache for all state (which DictionaryCache will similarly handle fine)
+        # and then just merge the results together.
+        #
+        # We size the non-members cache to be smaller than the members cache as the
+        # vast majority of state in Matrix (today) is member events.
+
         self._state_group_cache = DictionaryCache(
-            "*stateGroupCache*", 500000 * get_cache_factor_for("stateGroupCache")
+            "*stateGroupCache*",
+            # TODO: this hasn't been tuned yet
+            50000 * get_cache_factor_for("stateGroupCache")
+        )
+        self._state_group_members_cache = DictionaryCache(
+            "*stateGroupMembersCache*",
+            500000 * get_cache_factor_for("stateGroupMembersCache")
         )
 
     @defer.inlineCallbacks
@@ -117,61 +466,41 @@ class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore):
         )
 
     # FIXME: how should this be cached?
-    def get_filtered_current_state_ids(self, room_id, types, filtered_types=None):
+    def get_filtered_current_state_ids(self, room_id, state_filter=StateFilter.all()):
         """Get the current state event of a given type for a room based on the
         current_state_events table.  This may not be as up-to-date as the result
         of doing a fresh state resolution as per state_handler.get_current_state
+
         Args:
             room_id (str)
-            types (list[(Str, (Str|None))]): List of (type, state_key) tuples
-                which are used to filter the state fetched. `state_key` may be
-                None, which matches any `state_key`
-            filtered_types (list[Str]|None): List of types to apply the above filter to.
+            state_filter (StateFilter): The state filter used to fetch state
+                from the database.
+
         Returns:
-            deferred: dict of (type, state_key) -> event
+            Deferred[dict[tuple[str, str], str]]: Map from type/state_key to
+            event ID.
         """
 
-        include_other_types = False if filtered_types is None else True
-
         def _get_filtered_current_state_ids_txn(txn):
             results = {}
-            sql = """SELECT type, state_key, event_id FROM current_state_events
-                     WHERE room_id = ? %s"""
-            # Turns out that postgres doesn't like doing a list of OR's and
-            # is about 1000x slower, so we just issue a query for each specific
-            # type seperately.
-            if types:
-                clause_to_args = [
-                    (
-                        "AND type = ? AND state_key = ?",
-                        (etype, state_key)
-                    ) if state_key is not None else (
-                        "AND type = ?",
-                        (etype,)
-                    )
-                    for etype, state_key in types
-                ]
-
-                if include_other_types:
-                    unique_types = set(filtered_types)
-                    clause_to_args.append(
-                        (
-                            "AND type <> ? " * len(unique_types),
-                            list(unique_types)
-                        )
-                    )
-            else:
-                # If types is None we fetch all the state, and so just use an
-                # empty where clause with no extra args.
-                clause_to_args = [("", [])]
-            for where_clause, where_args in clause_to_args:
-                args = [room_id]
-                args.extend(where_args)
-                txn.execute(sql % (where_clause,), args)
-                for row in txn:
-                    typ, state_key, event_id = row
-                    key = (intern_string(typ), intern_string(state_key))
-                    results[key] = event_id
+            sql = """
+                SELECT type, state_key, event_id FROM current_state_events
+                WHERE room_id = ?
+            """
+
+            where_clause, where_args = state_filter.make_sql_filter_clause()
+
+            if where_clause:
+                sql += " AND (%s)" % (where_clause,)
+
+            args = [room_id]
+            args.extend(where_args)
+            txn.execute(sql, args)
+            for row in txn:
+                typ, state_key, event_id = row
+                key = (intern_string(typ), intern_string(state_key))
+                results[key] = event_id
+
             return results
 
         return self.runInteraction(
@@ -220,7 +549,17 @@ class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore):
         )
 
     @defer.inlineCallbacks
-    def get_state_groups_ids(self, room_id, event_ids):
+    def get_state_groups_ids(self, _room_id, event_ids):
+        """Get the event IDs of all the state for the state groups for the given events
+
+        Args:
+            _room_id (str): id of the room for these events
+            event_ids (iterable[str]): ids of the events
+
+        Returns:
+            Deferred[dict[int, dict[tuple[str, str], str]]]:
+                dict of state_group_id -> (dict of (type, state_key) -> event id)
+        """
         if not event_ids:
             defer.returnValue({})
 
@@ -235,7 +574,7 @@ class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore):
 
     @defer.inlineCallbacks
     def get_state_ids_for_group(self, state_group):
-        """Get the state IDs for the given state group
+        """Get the event IDs of all the state in the given state group
 
         Args:
             state_group (int)
@@ -251,7 +590,9 @@ class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore):
     def get_state_groups(self, room_id, event_ids):
         """ Get the state groups for the given list of event_ids
 
-        The return value is a dict mapping group names to lists of events.
+        Returns:
+            Deferred[dict[int, list[EventBase]]]:
+                dict of state_group_id -> list of state events.
         """
         if not event_ids:
             defer.returnValue({})
@@ -275,18 +616,17 @@ class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore):
         })
 
     @defer.inlineCallbacks
-    def _get_state_groups_from_groups(self, groups, types):
+    def _get_state_groups_from_groups(self, groups, state_filter):
         """Returns the state groups for a given set of groups, filtering on
         types of state events.
 
         Args:
             groups(list[int]): list of state group IDs to query
-            types (Iterable[str, str|None]|None): list of 2-tuples of the form
-                (`type`, `state_key`), where a `state_key` of `None` matches all
-                state_keys for the `type`. If None, all types are returned.
-
+            state_filter (StateFilter): The state filter used to fetch state
+                from the database.
         Returns:
-            dictionary state_group -> (dict of (type, state_key) -> event id)
+            Deferred[dict[int, dict[tuple[str, str], str]]]:
+                dict of state_group_id -> (dict of (type, state_key) -> event id)
         """
         results = {}
 
@@ -294,19 +634,23 @@ class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore):
         for chunk in chunks:
             res = yield self.runInteraction(
                 "_get_state_groups_from_groups",
-                self._get_state_groups_from_groups_txn, chunk, types,
+                self._get_state_groups_from_groups_txn, chunk, state_filter,
             )
             results.update(res)
 
         defer.returnValue(results)
 
     def _get_state_groups_from_groups_txn(
-        self, txn, groups, types=None,
+        self, txn, groups, state_filter=StateFilter.all(),
     ):
         results = {group: {} for group in groups}
 
-        if types is not None:
-            types = list(set(types))  # deduplicate types list
+        where_clause, where_args = state_filter.make_sql_filter_clause()
+
+        # Unless the filter clause is empty, we're going to append it after an
+        # existing where clause
+        if where_clause:
+            where_clause = " AND (%s)" % (where_clause,)
 
         if isinstance(self.database_engine, PostgresEngine):
             # Temporarily disable sequential scans in this transaction. This is
@@ -322,69 +666,33 @@ class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore):
             # group for the given type, state_key.
             # This may return multiple rows per (type, state_key), but last_value
             # should be the same.
-            sql = ("""
+            sql = """
                 WITH RECURSIVE state(state_group) AS (
                     VALUES(?::bigint)
                     UNION ALL
                     SELECT prev_state_group FROM state_group_edges e, state s
                     WHERE s.state_group = e.state_group
                 )
-                SELECT type, state_key, last_value(event_id) OVER (
+                SELECT DISTINCT type, state_key, last_value(event_id) OVER (
                     PARTITION BY type, state_key ORDER BY state_group ASC
                     ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING
                 ) AS event_id FROM state_groups_state
                 WHERE state_group IN (
                     SELECT state_group FROM state
                 )
-                %s
-            """)
-
-            # Turns out that postgres doesn't like doing a list of OR's and
-            # is about 1000x slower, so we just issue a query for each specific
-            # type seperately.
-            if types is not None:
-                clause_to_args = [
-                    (
-                        "AND type = ? AND state_key = ?",
-                        (etype, state_key)
-                    ) if state_key is not None else (
-                        "AND type = ?",
-                        (etype,)
-                    )
-                    for etype, state_key in types
-                ]
-            else:
-                # If types is None we fetch all the state, and so just use an
-                # empty where clause with no extra args.
-                clause_to_args = [("", [])]
+            """
 
-            for where_clause, where_args in clause_to_args:
-                for group in groups:
-                    args = [group]
-                    args.extend(where_args)
+            for group in groups:
+                args = [group]
+                args.extend(where_args)
 
-                    txn.execute(sql % (where_clause,), args)
-                    for row in txn:
-                        typ, state_key, event_id = row
-                        key = (typ, state_key)
-                        results[group][key] = event_id
+                txn.execute(sql + where_clause, args)
+                for row in txn:
+                    typ, state_key, event_id = row
+                    key = (typ, state_key)
+                    results[group][key] = event_id
         else:
-            where_args = []
-            where_clauses = []
-            wildcard_types = False
-            if types is not None:
-                for typ in types:
-                    if typ[1] is None:
-                        where_clauses.append("(type = ?)")
-                        where_args.append(typ[0])
-                        wildcard_types = True
-                    else:
-                        where_clauses.append("(type = ? AND state_key = ?)")
-                        where_args.extend([typ[0], typ[1]])
-
-                where_clause = "AND (%s)" % (" OR ".join(where_clauses))
-            else:
-                where_clause = ""
+            max_entries_returned = state_filter.max_entries_returned()
 
             # We don't use WITH RECURSIVE on sqlite3 as there are distributions
             # that ship with an sqlite3 version that doesn't support it (e.g. wheezy)
@@ -398,12 +706,11 @@ class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore):
                     # without the right indices (which we can't add until
                     # after we finish deduping state, which requires this func)
                     args = [next_group]
-                    if types:
-                        args.extend(where_args)
+                    args.extend(where_args)
 
                     txn.execute(
                         "SELECT type, state_key, event_id FROM state_groups_state"
-                        " WHERE state_group = ? %s" % (where_clause,),
+                        " WHERE state_group = ? " + where_clause,
                         args
                     )
                     results[group].update(
@@ -419,9 +726,8 @@ class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore):
                     # wildcards (i.e. Nones) in which case we have to do an exhaustive
                     # search
                     if (
-                        types is not None and
-                        not wildcard_types and
-                        len(results[group]) == len(types)
+                        max_entries_returned is not None and
+                        len(results[group]) == max_entries_returned
                     ):
                         break
 
@@ -436,20 +742,14 @@ class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore):
         return results
 
     @defer.inlineCallbacks
-    def get_state_for_events(self, event_ids, types, filtered_types=None):
+    def get_state_for_events(self, event_ids, state_filter=StateFilter.all()):
         """Given a list of event_ids and type tuples, return a list of state
-        dicts for each event. The state dicts will only have the type/state_keys
-        that are in the `types` list.
+        dicts for each event.
 
         Args:
             event_ids (list[string])
-            types (list[(str, str|None)]|None): List of (type, state_key) tuples
-                which are used to filter the state fetched. If `state_key` is None,
-                all events are returned of the given type.
-                May be None, which matches any key.
-            filtered_types(list[str]|None): Only apply filtering via `types` to this
-                list of event types.  Other types of events are returned unfiltered.
-                If None, `types` filtering is applied to all events.
+            state_filter (StateFilter): The state filter used to fetch state
+                from the database.
 
         Returns:
             deferred: A dict of (event_id) -> (type, state_key) -> [state_events]
@@ -459,7 +759,7 @@ class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore):
         )
 
         groups = set(itervalues(event_to_groups))
-        group_to_state = yield self._get_state_for_groups(groups, types, filtered_types)
+        group_to_state = yield self._get_state_for_groups(groups, state_filter)
 
         state_event_map = yield self.get_events(
             [ev_id for sd in itervalues(group_to_state) for ev_id in itervalues(sd)],
@@ -478,20 +778,15 @@ class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore):
         defer.returnValue({event: event_to_state[event] for event in event_ids})
 
     @defer.inlineCallbacks
-    def get_state_ids_for_events(self, event_ids, types=None, filtered_types=None):
+    def get_state_ids_for_events(self, event_ids, state_filter=StateFilter.all()):
         """
         Get the state dicts corresponding to a list of events, containing the event_ids
         of the state events (as opposed to the events themselves)
 
         Args:
             event_ids(list(str)): events whose state should be returned
-            types(list[(str, str|None)]|None): List of (type, state_key) tuples
-                which are used to filter the state fetched. If `state_key` is None,
-                all events are returned of the given type.
-                May be None, which matches any key.
-            filtered_types(list[str]|None): Only apply filtering via `types` to this
-                list of event types.  Other types of events are returned unfiltered.
-                If None, `types` filtering is applied to all events.
+            state_filter (StateFilter): The state filter used to fetch state
+                from the database.
 
         Returns:
             A deferred dict from event_id -> (type, state_key) -> event_id
@@ -501,7 +796,7 @@ class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore):
         )
 
         groups = set(itervalues(event_to_groups))
-        group_to_state = yield self._get_state_for_groups(groups, types, filtered_types)
+        group_to_state = yield self._get_state_for_groups(groups, state_filter)
 
         event_to_state = {
             event_id: group_to_state[group]
@@ -511,45 +806,35 @@ class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore):
         defer.returnValue({event: event_to_state[event] for event in event_ids})
 
     @defer.inlineCallbacks
-    def get_state_for_event(self, event_id, types=None, filtered_types=None):
+    def get_state_for_event(self, event_id, state_filter=StateFilter.all()):
         """
         Get the state dict corresponding to a particular event
 
         Args:
             event_id(str): event whose state should be returned
-            types(list[(str, str|None)]|None): List of (type, state_key) tuples
-                which are used to filter the state fetched. If `state_key` is None,
-                all events are returned of the given type.
-                May be None, which matches any key.
-            filtered_types(list[str]|None): Only apply filtering via `types` to this
-                list of event types.  Other types of events are returned unfiltered.
-                If None, `types` filtering is applied to all events.
+            state_filter (StateFilter): The state filter used to fetch state
+                from the database.
 
         Returns:
             A deferred dict from (type, state_key) -> state_event
         """
-        state_map = yield self.get_state_for_events([event_id], types, filtered_types)
+        state_map = yield self.get_state_for_events([event_id], state_filter)
         defer.returnValue(state_map[event_id])
 
     @defer.inlineCallbacks
-    def get_state_ids_for_event(self, event_id, types=None, filtered_types=None):
+    def get_state_ids_for_event(self, event_id, state_filter=StateFilter.all()):
         """
         Get the state dict corresponding to a particular event
 
         Args:
             event_id(str): event whose state should be returned
-            types(list[(str, str|None)]|None): List of (type, state_key) tuples
-                which are used to filter the state fetched. If `state_key` is None,
-                all events are returned of the given type.
-                May be None, which matches any key.
-            filtered_types(list[str]|None): Only apply filtering via `types` to this
-                list of event types.  Other types of events are returned unfiltered.
-                If None, `types` filtering is applied to all events.
+            state_filter (StateFilter): The state filter used to fetch state
+                from the database.
 
         Returns:
             A deferred dict from (type, state_key) -> state_event
         """
-        state_map = yield self.get_state_ids_for_events([event_id], types, filtered_types)
+        state_map = yield self.get_state_ids_for_events([event_id], state_filter)
         defer.returnValue(state_map[event_id])
 
     @cached(max_entries=50000)
@@ -580,179 +865,207 @@ class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore):
 
         defer.returnValue({row["event_id"]: row["state_group"] for row in rows})
 
-    def _get_some_state_from_cache(self, group, types, filtered_types=None):
+    def _get_state_for_group_using_cache(self, cache, group, state_filter):
         """Checks if group is in cache. See `_get_state_for_groups`
 
         Args:
+            cache(DictionaryCache): the state group cache to use
             group(int): The state group to lookup
-            types(list[str, str|None]): List of 2-tuples of the form
-                (`type`, `state_key`), where a `state_key` of `None` matches all
-                state_keys for the `type`.
-            filtered_types(list[str]|None): Only apply filtering via `types` to this
-                list of event types.  Other types of events are returned unfiltered.
-                If None, `types` filtering is applied to all events.
+            state_filter (StateFilter): The state filter used to fetch state
+                from the database.
 
         Returns 2-tuple (`state_dict`, `got_all`).
         `got_all` is a bool indicating if we successfully retrieved all
         requests state from the cache, if False we need to query the DB for the
         missing state.
         """
-        is_all, known_absent, state_dict_ids = self._state_group_cache.get(group)
+        is_all, known_absent, state_dict_ids = cache.get(group)
 
-        type_to_key = {}
+        if is_all or state_filter.is_full():
+            # Either we have everything or want everything, either way
+            # `is_all` tells us whether we've gotten everything.
+            return state_filter.filter_state(state_dict_ids), is_all
 
-        # tracks whether any of ourrequested types are missing from the cache
+        # tracks whether any of our requested types are missing from the cache
         missing_types = False
 
-        for typ, state_key in types:
-            key = (typ, state_key)
-
-            if (
-                state_key is None or
-                (filtered_types is not None and typ not in filtered_types)
-            ):
-                type_to_key[typ] = None
-                # we mark the type as missing from the cache because
-                # when the cache was populated it might have been done with a
-                # restricted set of state_keys, so the wildcard will not work
-                # and the cache may be incomplete.
-                missing_types = True
-            else:
-                if type_to_key.get(typ, object()) is not None:
-                    type_to_key.setdefault(typ, set()).add(state_key)
-
+        if state_filter.has_wildcards():
+            # We don't know if we fetched all the state keys for the types in
+            # the filter that are wildcards, so we have to assume that we may
+            # have missed some.
+            missing_types = True
+        else:
+            # There aren't any wild cards, so `concrete_types()` returns the
+            # complete list of event types we're wanting.
+            for key in state_filter.concrete_types():
                 if key not in state_dict_ids and key not in known_absent:
                     missing_types = True
+                    break
 
-        sentinel = object()
-
-        def include(typ, state_key):
-            valid_state_keys = type_to_key.get(typ, sentinel)
-            if valid_state_keys is sentinel:
-                return filtered_types is not None and typ not in filtered_types
-            if valid_state_keys is None:
-                return True
-            if state_key in valid_state_keys:
-                return True
-            return False
-
-        got_all = is_all
-        if not got_all:
-            # the cache is incomplete. We may still have got all the results we need, if
-            # we don't have any wildcards in the match list.
-            if not missing_types and filtered_types is None:
-                got_all = True
-
-        return {
-            k: v for k, v in iteritems(state_dict_ids)
-            if include(k[0], k[1])
-        }, got_all
-
-    def _get_all_state_from_cache(self, group):
-        """Checks if group is in cache. See `_get_state_for_groups`
+        return state_filter.filter_state(state_dict_ids), not missing_types
 
-        Returns 2-tuple (`state_dict`, `got_all`). `got_all` is a bool
-        indicating if we successfully retrieved all requests state from the
-        cache, if False we need to query the DB for the missing state.
+    @defer.inlineCallbacks
+    def _get_state_for_groups(self, groups, state_filter=StateFilter.all()):
+        """Gets the state at each of a list of state groups, optionally
+        filtering by type/state_key
 
         Args:
-            group: The state group to lookup
+            groups (iterable[int]): list of state groups for which we want
+                to get the state.
+            state_filter (StateFilter): The state filter used to fetch state
+                from the database.
+        Returns:
+            Deferred[dict[int, dict[tuple[str, str], str]]]:
+                dict of state_group_id -> (dict of (type, state_key) -> event id)
         """
-        is_all, _, state_dict_ids = self._state_group_cache.get(group)
 
-        return state_dict_ids, is_all
+        member_filter, non_member_filter = state_filter.get_member_split()
 
-    @defer.inlineCallbacks
-    def _get_state_for_groups(self, groups, types=None, filtered_types=None):
+        # Now we look them up in the member and non-member caches
+        non_member_state, incomplete_groups_nm, = (
+            yield self._get_state_for_groups_using_cache(
+                groups, self._state_group_cache,
+                state_filter=non_member_filter,
+            )
+        )
+
+        member_state, incomplete_groups_m, = (
+            yield self._get_state_for_groups_using_cache(
+                groups, self._state_group_members_cache,
+                state_filter=member_filter,
+            )
+        )
+
+        state = dict(non_member_state)
+        for group in groups:
+            state[group].update(member_state[group])
+
+        # Now fetch any missing groups from the database
+
+        incomplete_groups = incomplete_groups_m | incomplete_groups_nm
+
+        if not incomplete_groups:
+            defer.returnValue(state)
+
+        cache_sequence_nm = self._state_group_cache.sequence
+        cache_sequence_m = self._state_group_members_cache.sequence
+
+        # Help the cache hit ratio by expanding the filter a bit
+        db_state_filter = state_filter.return_expanded()
+
+        group_to_state_dict = yield self._get_state_groups_from_groups(
+            list(incomplete_groups),
+            state_filter=db_state_filter,
+        )
+
+        # Now lets update the caches
+        self._insert_into_cache(
+            group_to_state_dict,
+            db_state_filter,
+            cache_seq_num_members=cache_sequence_m,
+            cache_seq_num_non_members=cache_sequence_nm,
+        )
+
+        # And finally update the result dict, by filtering out any extra
+        # stuff we pulled out of the database.
+        for group, group_state_dict in iteritems(group_to_state_dict):
+            # We just replace any existing entries, as we will have loaded
+            # everything we need from the database anyway.
+            state[group] = state_filter.filter_state(group_state_dict)
+
+        defer.returnValue(state)
+
+    def _get_state_for_groups_using_cache(
+        self, groups, cache, state_filter,
+    ):
         """Gets the state at each of a list of state groups, optionally
-        filtering by type/state_key
+        filtering by type/state_key, querying from a specific cache.
 
         Args:
             groups (iterable[int]): list of state groups for which we want
                 to get the state.
-            types (None|iterable[(str, None|str)]):
-                indicates the state type/keys required. If None, the whole
-                state is fetched and returned.
-
-                Otherwise, each entry should be a `(type, state_key)` tuple to
-                include in the response. A `state_key` of None is a wildcard
-                meaning that we require all state with that type.
-            filtered_types(list[str]|None): Only apply filtering via `types` to this
-                list of event types.  Other types of events are returned unfiltered.
-                If None, `types` filtering is applied to all events.
+            cache (DictionaryCache): the cache of group ids to state dicts which
+                we will pass through - either the normal state cache or the specific
+                members state cache.
+            state_filter (StateFilter): The state filter used to fetch state
+                from the database.
 
         Returns:
-            Deferred[dict[int, dict[(type, state_key), EventBase]]]
-                a dictionary mapping from state group to state dictionary.
+            tuple[dict[int, dict[tuple[str, str], str]], set[int]]: Tuple of
+            dict of state_group_id -> (dict of (type, state_key) -> event id)
+            of entries in the cache, and the state group ids either missing
+            from the cache or incomplete.
         """
-        if types:
-            types = frozenset(types)
         results = {}
-        missing_groups = []
-        if types is not None:
-            for group in set(groups):
-                state_dict_ids, got_all = self._get_some_state_from_cache(
-                    group, types, filtered_types
-                )
-                results[group] = state_dict_ids
+        incomplete_groups = set()
+        for group in set(groups):
+            state_dict_ids, got_all = self._get_state_for_group_using_cache(
+                cache, group, state_filter
+            )
+            results[group] = state_dict_ids
 
-                if not got_all:
-                    missing_groups.append(group)
-        else:
-            for group in set(groups):
-                state_dict_ids, got_all = self._get_all_state_from_cache(
-                    group
-                )
+            if not got_all:
+                incomplete_groups.add(group)
+
+        return results, incomplete_groups
 
-                results[group] = state_dict_ids
+    def _insert_into_cache(self, group_to_state_dict, state_filter,
+                           cache_seq_num_members, cache_seq_num_non_members):
+        """Inserts results from querying the database into the relevant cache.
 
-                if not got_all:
-                    missing_groups.append(group)
+        Args:
+            group_to_state_dict (dict): The new entries pulled from database.
+                Map from state group to state dict
+            state_filter (StateFilter): The state filter used to fetch state
+                from the database.
+            cache_seq_num_members (int): Sequence number of member cache since
+                last lookup in cache
+            cache_seq_num_non_members (int): Sequence number of member cache since
+                last lookup in cache
+        """
 
-        if missing_groups:
-            # Okay, so we have some missing_types, lets fetch them.
-            cache_seq_num = self._state_group_cache.sequence
+        # We need to work out which types we've fetched from the DB for the
+        # member vs non-member caches. This should be as accurate as possible,
+        # but can be an underestimate (e.g. when we have wild cards)
 
-            # the DictionaryCache knows if it has *all* the state, but
-            # does not know if it has all of the keys of a particular type,
-            # which makes wildcard lookups expensive unless we have a complete
-            # cache. Hence, if we are doing a wildcard lookup, populate the
-            # cache fully so that we can do an efficient lookup next time.
+        member_filter, non_member_filter = state_filter.get_member_split()
+        if member_filter.is_full():
+            # We fetched all member events
+            member_types = None
+        else:
+            # `concrete_types()` will only return a subset when there are wild
+            # cards in the filter, but that's fine.
+            member_types = member_filter.concrete_types()
 
-            if filtered_types or (types and any(k is None for (t, k) in types)):
-                types_to_fetch = None
-            else:
-                types_to_fetch = types
+        if non_member_filter.is_full():
+            # We fetched all non member events
+            non_member_types = None
+        else:
+            non_member_types = non_member_filter.concrete_types()
 
-            group_to_state_dict = yield self._get_state_groups_from_groups(
-                missing_groups, types_to_fetch
-            )
+        for group, group_state_dict in iteritems(group_to_state_dict):
+            state_dict_members = {}
+            state_dict_non_members = {}
 
-            for group, group_state_dict in iteritems(group_to_state_dict):
-                state_dict = results[group]
-
-                # update the result, filtering by `types`.
-                if types:
-                    for k, v in iteritems(group_state_dict):
-                        (typ, _) = k
-                        if (
-                            (k in types or (typ, None) in types) or
-                            (filtered_types and typ not in filtered_types)
-                        ):
-                            state_dict[k] = v
+            for k, v in iteritems(group_state_dict):
+                if k[0] == EventTypes.Member:
+                    state_dict_members[k] = v
                 else:
-                    state_dict.update(group_state_dict)
-
-                # update the cache with all the things we fetched from the
-                # database.
-                self._state_group_cache.update(
-                    cache_seq_num,
-                    key=group,
-                    value=group_state_dict,
-                    fetched_keys=types_to_fetch,
-                )
+                    state_dict_non_members[k] = v
 
-        defer.returnValue(results)
+            self._state_group_members_cache.update(
+                cache_seq_num_members,
+                key=group,
+                value=state_dict_members,
+                fetched_keys=member_types,
+            )
+
+            self._state_group_cache.update(
+                cache_seq_num_non_members,
+                key=group,
+                value=state_dict_non_members,
+                fetched_keys=non_member_types,
+            )
 
     def store_state_group(self, event_id, room_id, prev_group, delta_ids,
                           current_state_ids):
@@ -847,15 +1160,33 @@ class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore):
                     ],
                 )
 
-            # Prefill the state group cache with this group.
+            # Prefill the state group caches with this group.
             # It's fine to use the sequence like this as the state group map
             # is immutable. (If the map wasn't immutable then this prefill could
             # race with another update)
+
+            current_member_state_ids = {
+                s: ev
+                for (s, ev) in iteritems(current_state_ids)
+                if s[0] == EventTypes.Member
+            }
+            txn.call_after(
+                self._state_group_members_cache.update,
+                self._state_group_members_cache.sequence,
+                key=state_group,
+                value=dict(current_member_state_ids),
+            )
+
+            current_non_member_state_ids = {
+                s: ev
+                for (s, ev) in iteritems(current_state_ids)
+                if s[0] != EventTypes.Member
+            }
             txn.call_after(
                 self._state_group_cache.update,
                 self._state_group_cache.sequence,
                 key=state_group,
-                value=dict(current_state_ids),
+                value=dict(current_non_member_state_ids),
             )
 
             return state_group
@@ -1043,12 +1374,12 @@ class StateStore(StateGroupWorkerStore, BackgroundUpdateStore):
                         continue
 
                     prev_state = self._get_state_groups_from_groups_txn(
-                        txn, [prev_group], types=None
+                        txn, [prev_group],
                     )
                     prev_state = prev_state[prev_group]
 
                     curr_state = self._get_state_groups_from_groups_txn(
-                        txn, [state_group], types=None
+                        txn, [state_group],
                     )
                     curr_state = curr_state[state_group]
 
diff --git a/synapse/storage/stream.py b/synapse/storage/stream.py
index 4c296d72c0..d6cfdba519 100644
--- a/synapse/storage/stream.py
+++ b/synapse/storage/stream.py
@@ -630,7 +630,21 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore):
 
     @defer.inlineCallbacks
     def get_all_new_events_stream(self, from_id, current_id, limit):
-        """Get all new events"""
+        """Get all new events
+
+         Returns all events with from_id < stream_ordering <= current_id.
+
+         Args:
+             from_id (int):  the stream_ordering of the last event we processed
+             current_id (int):  the stream_ordering of the most recently processed event
+             limit (int): the maximum number of events to return
+
+         Returns:
+             Deferred[Tuple[int, list[FrozenEvent]]]: A tuple of (next_id, events), where
+             `next_id` is the next value to pass as `from_id` (it will either be the
+             stream_ordering of the last returned event, or, if fewer than `limit` events
+             were found, `current_id`.
+         """
 
         def get_all_new_events_stream_txn(txn):
             sql = (
diff --git a/synapse/storage/transactions.py b/synapse/storage/transactions.py
index 428e7fa36e..d8bf953ec0 100644
--- a/synapse/storage/transactions.py
+++ b/synapse/storage/transactions.py
@@ -18,19 +18,19 @@ from collections import namedtuple
 
 import six
 
-from canonicaljson import encode_canonical_json, json
+from canonicaljson import encode_canonical_json
 
 from twisted.internet import defer
 
 from synapse.metrics.background_process_metrics import run_as_background_process
-from synapse.util.caches.descriptors import cached
+from synapse.util.caches.expiringcache import ExpiringCache
 
-from ._base import SQLBaseStore
+from ._base import SQLBaseStore, db_to_json
 
 # py2 sqlite has buffer hardcoded as only binary type, so we must use it,
 # despite being deprecated and removed in favor of memoryview
 if six.PY2:
-    db_binary_type = buffer
+    db_binary_type = six.moves.builtins.buffer
 else:
     db_binary_type = memoryview
 
@@ -50,6 +50,8 @@ _UpdateTransactionRow = namedtuple(
     )
 )
 
+SENTINEL = object()
+
 
 class TransactionStore(SQLBaseStore):
     """A collection of queries for handling PDUs.
@@ -60,6 +62,12 @@ class TransactionStore(SQLBaseStore):
 
         self._clock.looping_call(self._start_cleanup_transactions, 30 * 60 * 1000)
 
+        self._destination_retry_cache = ExpiringCache(
+            cache_name="get_destination_retry_timings",
+            clock=self._clock,
+            expiry_ms=5 * 60 * 1000,
+        )
+
     def get_received_txn_response(self, transaction_id, origin):
         """For an incoming transaction from a given origin, check if we have
         already responded to it. If so, return the response code and response
@@ -95,7 +103,8 @@ class TransactionStore(SQLBaseStore):
         )
 
         if result and result["response_code"]:
-            return result["response_code"], json.loads(str(result["response_json"]))
+            return result["response_code"], db_to_json(result["response_json"])
+
         else:
             return None
 
@@ -155,7 +164,7 @@ class TransactionStore(SQLBaseStore):
         """
         pass
 
-    @cached(max_entries=10000)
+    @defer.inlineCallbacks
     def get_destination_retry_timings(self, destination):
         """Gets the current retry timings (if any) for a given destination.
 
@@ -166,10 +175,20 @@ class TransactionStore(SQLBaseStore):
             None if not retrying
             Otherwise a dict for the retry scheme
         """
-        return self.runInteraction(
+
+        result = self._destination_retry_cache.get(destination, SENTINEL)
+        if result is not SENTINEL:
+            defer.returnValue(result)
+
+        result = yield self.runInteraction(
             "get_destination_retry_timings",
             self._get_destination_retry_timings, destination)
 
+        # We don't hugely care about race conditions between getting and
+        # invalidating the cache, since we time out fairly quickly anyway.
+        self._destination_retry_cache[destination] = result
+        defer.returnValue(result)
+
     def _get_destination_retry_timings(self, txn, destination):
         result = self._simple_select_one_txn(
             txn,
@@ -197,8 +216,7 @@ class TransactionStore(SQLBaseStore):
             retry_interval (int) - how long until next retry in ms
         """
 
-        # XXX: we could chose to not bother persisting this if our cache thinks
-        # this is a NOOP
+        self._destination_retry_cache.pop(destination, None)
         return self.runInteraction(
             "set_destination_retry_timings",
             self._set_destination_retry_timings,
@@ -211,10 +229,6 @@ class TransactionStore(SQLBaseStore):
                                        retry_last_ts, retry_interval):
         self.database_engine.lock_table(txn, "destinations")
 
-        self._invalidate_cache_and_stream(
-            txn, self.get_destination_retry_timings, (destination,)
-        )
-
         # We need to be careful here as the data may have changed from under us
         # due to a worker setting the timings.