From 14e4d4f4bf894f4e70118e03f4f4575e1eb6dab6 Mon Sep 17 00:00:00 2001 From: Amber Brown Date: Fri, 31 Aug 2018 00:19:58 +1000 Subject: Port storage/ to Python 3 (#3725) --- synapse/storage/_base.py | 32 +++++++++++++++++++++++++++++++- synapse/storage/deviceinbox.py | 2 +- synapse/storage/devices.py | 10 +++++----- synapse/storage/end_to_end_keys.py | 6 +++--- synapse/storage/engines/postgres.py | 9 +++++++-- synapse/storage/events.py | 10 +++++----- synapse/storage/events_worker.py | 9 +++++---- synapse/storage/filtering.py | 6 +++--- synapse/storage/pusher.py | 14 ++++++++++---- synapse/storage/transactions.py | 7 ++++--- 10 files changed, 74 insertions(+), 31 deletions(-) (limited to 'synapse/storage') diff --git a/synapse/storage/_base.py b/synapse/storage/_base.py index 08dffd774f..be61147b9b 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 import PY2, iteritems, iterkeys, itervalues from six.moves import 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, 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/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/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/events.py b/synapse/storage/events.py index f39c8c8461..8bf87f38f7 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 @@ -1220,7 +1220,7 @@ 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 @@ -1529,7 +1529,7 @@ 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. @@ -1910,9 +1910,9 @@ 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) 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/pusher.py b/synapse/storage/pusher.py index 8443bd4c1b..c7987bfcdd 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 = 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/transactions.py b/synapse/storage/transactions.py index 428e7fa36e..0c42bd3322 100644 --- a/synapse/storage/transactions.py +++ b/synapse/storage/transactions.py @@ -18,14 +18,14 @@ 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 ._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 @@ -95,7 +95,8 @@ 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 -- cgit 1.5.1 From 09f3cf1a7ef0c533d052a5c87257503b710093c6 Mon Sep 17 00:00:00 2001 From: Neil Johnson Date: Fri, 31 Aug 2018 15:42:51 +0100 Subject: ensure post registration auth checks do not fail erroneously --- synapse/api/auth.py | 7 ++----- synapse/rest/client/v1_only/register.py | 4 ++++ synapse/rest/client/v2_alpha/register.py | 4 ++++ synapse/storage/monthly_active_users.py | 15 ++++++++++++++- 4 files changed, 24 insertions(+), 6 deletions(-) (limited to 'synapse/storage') diff --git a/synapse/api/auth.py b/synapse/api/auth.py index 9c207b9537..6a97c06110 100644 --- a/synapse/api/auth.py +++ b/synapse/api/auth.py @@ -809,11 +809,8 @@ class Auth(object): elif threepid: # If the user does not exist yet, but is signing up with a # reserved threepid then pass auth check - for tp in self.hs.config.mau_limits_reserved_threepids: - if (threepid['medium'] == tp['medium'] - and threepid['address'] == tp['address']): - return - + if is_threepid_reserved(threepid): + return # Else if there is no room in the MAU bucket, bail current_mau = yield self.store.get_monthly_active_count() if current_mau >= self.hs.config.max_mau_value: diff --git a/synapse/rest/client/v1_only/register.py b/synapse/rest/client/v1_only/register.py index 2c7bbcb171..95873e03d5 100644 --- a/synapse/rest/client/v1_only/register.py +++ b/synapse/rest/client/v1_only/register.py @@ -291,6 +291,10 @@ class RegisterRestServlet(ClientV1RestServlet): password=password, threepid=threepid, ) + # Necessary due to auth checks prior to the threepid being + # written to the db + if self.store.is_threepid_reserved(threepid): + self.store.upsert_monthly_active_user(registered_user_id) if session[LoginType.EMAIL_IDENTITY]: logger.debug("Binding emails %s to %s" % ( diff --git a/synapse/rest/client/v2_alpha/register.py b/synapse/rest/client/v2_alpha/register.py index 45113e5386..f22b7577ea 100644 --- a/synapse/rest/client/v2_alpha/register.py +++ b/synapse/rest/client/v2_alpha/register.py @@ -406,6 +406,10 @@ class RegisterRestServlet(RestServlet): generate_token=False, threepid=threepid, ) + # Necessary due to auth checks prior to the threepid being + # written to the db + if self.store.is_threepid_reserved(threepid): + self.store.upsert_monthly_active_user(registered_user_id) # remember that we've now registered that user account, and with # what user ID (since the user may not have specified) diff --git a/synapse/storage/monthly_active_users.py b/synapse/storage/monthly_active_users.py index d178f5c5ba..173867c4b1 100644 --- a/synapse/storage/monthly_active_users.py +++ b/synapse/storage/monthly_active_users.py @@ -36,7 +36,6 @@ class MonthlyActiveUsersStore(SQLBaseStore): @defer.inlineCallbacks def initialise_reserved_users(self, threepids): - # TODO Why can't I do this in init? store = self.hs.get_datastore() reserved_user_list = [] @@ -220,3 +219,17 @@ class MonthlyActiveUsersStore(SQLBaseStore): yield self.upsert_monthly_active_user(user_id) elif now - last_seen_timestamp > LAST_SEEN_GRANULARITY: yield self.upsert_monthly_active_user(user_id) + + def is_threepid_reserved(self, threepid): + """Check the threepid against the reserved threepid config + Args: + threepid(dict) - The threepid to test for + Returns: + boolean Is the threepid undertest reserved_user + """ + for tp in self.hs.config.mau_limits_reserved_threepids: + if (threepid['medium'] == tp['medium'] + and threepid['address'] == tp['address']): + return True + else: + return False -- cgit 1.5.1 From 0b01281e77aee7e69925f36dbb6a798772a98a45 Mon Sep 17 00:00:00 2001 From: Neil Johnson Date: Fri, 31 Aug 2018 17:11:11 +0100 Subject: move threepid checker to config, add missing yields --- synapse/api/auth.py | 13 +++++++++++-- synapse/config/server.py | 17 +++++++++++++++++ synapse/rest/client/v1_only/register.py | 7 ++++--- synapse/rest/client/v2_alpha/register.py | 5 +++-- synapse/storage/monthly_active_users.py | 14 -------------- tests/utils.py | 6 ++++++ 6 files changed, 41 insertions(+), 21 deletions(-) (limited to 'synapse/storage') diff --git a/synapse/api/auth.py b/synapse/api/auth.py index a36fb6b3bd..a89687f420 100644 --- a/synapse/api/auth.py +++ b/synapse/api/auth.py @@ -26,6 +26,7 @@ import synapse.types from synapse import event_auth from synapse.api.constants import EventTypes, JoinRules, Membership from synapse.api.errors import AuthError, Codes, ResourceLimitError +from synapse.config.server import is_threepid_reserved from synapse.types import UserID from synapse.util.caches import CACHE_SIZE_FACTOR, register_cache from synapse.util.caches.lrucache import LruCache @@ -782,6 +783,11 @@ class Auth(object): Args: user_id(str|None): If present, checks for presence against existing MAU cohort + threepid(dict|None): If present, checks for presence against configured + reserved threepid. Used in cases where the user is trying register + with a MAU blocked server, normally they would be rejected but their + threepid is on the reserved list. user_id and + threepid should never be set at the same time. """ # Never fail an auth check for the server notices users @@ -797,6 +803,10 @@ class Auth(object): limit_type=self.hs.config.hs_disabled_limit_type ) if self.hs.config.limit_usage_by_mau is True: + + if user_id and threepid: + logger.warn("Called with both user_id and threepid, this shoudn't happen") + # If the user is already part of the MAU cohort or a trial user if user_id: timestamp = yield self.store.user_last_seen_monthly_active(user_id) @@ -809,14 +819,13 @@ class Auth(object): elif threepid: # If the user does not exist yet, but is signing up with a # reserved threepid then pass auth check - if self.store.is_threepid_reserved(threepid): + if is_threepid_reserved(self.hs.config, threepid): return # Else if there is no room in the MAU bucket, bail current_mau = yield self.store.get_monthly_active_count() if current_mau >= self.hs.config.max_mau_value: raise ResourceLimitError( 403, "Monthly Active User Limit Exceeded", - admin_contact=self.hs.config.admin_contact, errcode=Codes.RESOURCE_LIMIT_EXCEEDED, limit_type="monthly_active_user" diff --git a/synapse/config/server.py b/synapse/config/server.py index 2faf472189..c1c7c0105e 100644 --- a/synapse/config/server.py +++ b/synapse/config/server.py @@ -404,6 +404,23 @@ class ServerConfig(Config): " service on the given port.") +def is_threepid_reserved(config, threepid): + """Check the threepid against the reserved threepid config + Args: + config(ServerConfig) - to access server config attributes + threepid(dict) - The threepid to test for + + Returns: + boolean Is the threepid undertest reserved_user + """ + + for tp in config.mau_limits_reserved_threepids: + if (threepid['medium'] == tp['medium'] + and threepid['address'] == tp['address']): + return True + return False + + def read_gc_thresholds(thresholds): """Reads the three integer thresholds for garbage collection. Ensures that the thresholds are integers if thresholds are supplied. diff --git a/synapse/rest/client/v1_only/register.py b/synapse/rest/client/v1_only/register.py index 95873e03d5..dadb376b02 100644 --- a/synapse/rest/client/v1_only/register.py +++ b/synapse/rest/client/v1_only/register.py @@ -23,6 +23,7 @@ from twisted.internet import defer import synapse.util.stringutils as stringutils from synapse.api.constants import LoginType from synapse.api.errors import Codes, SynapseError +from synapse.config.server import is_threepid_reserved from synapse.http.servlet import assert_params_in_dict, parse_json_object_from_request from synapse.rest.client.v1.base import ClientV1RestServlet from synapse.types import create_requester @@ -282,7 +283,7 @@ class RegisterRestServlet(ClientV1RestServlet): if "user" in register_json else None ) threepid = None - if session[LoginType.EMAIL_IDENTITY]: + if session.get(LoginType.EMAIL_IDENTITY): threepid = session["threepidCreds"] handler = self.handlers.registration_handler @@ -293,8 +294,8 @@ class RegisterRestServlet(ClientV1RestServlet): ) # Necessary due to auth checks prior to the threepid being # written to the db - if self.store.is_threepid_reserved(threepid): - self.store.upsert_monthly_active_user(registered_user_id) + if is_threepid_reserved(self.hs.config, threepid): + yield self.store.upsert_monthly_active_user(user_id) if session[LoginType.EMAIL_IDENTITY]: logger.debug("Binding emails %s to %s" % ( diff --git a/synapse/rest/client/v2_alpha/register.py b/synapse/rest/client/v2_alpha/register.py index f22b7577ea..2fb4d43ccb 100644 --- a/synapse/rest/client/v2_alpha/register.py +++ b/synapse/rest/client/v2_alpha/register.py @@ -26,6 +26,7 @@ import synapse import synapse.types from synapse.api.constants import LoginType from synapse.api.errors import Codes, SynapseError, UnrecognizedRequestError +from synapse.config.server import is_threepid_reserved from synapse.http.servlet import ( RestServlet, assert_params_in_dict, @@ -408,8 +409,8 @@ class RegisterRestServlet(RestServlet): ) # Necessary due to auth checks prior to the threepid being # written to the db - if self.store.is_threepid_reserved(threepid): - self.store.upsert_monthly_active_user(registered_user_id) + if is_threepid_reserved(self.hs.config, threepid): + yield self.store.upsert_monthly_active_user(registered_user_id) # remember that we've now registered that user account, and with # what user ID (since the user may not have specified) diff --git a/synapse/storage/monthly_active_users.py b/synapse/storage/monthly_active_users.py index 173867c4b1..c7899d7fd2 100644 --- a/synapse/storage/monthly_active_users.py +++ b/synapse/storage/monthly_active_users.py @@ -219,17 +219,3 @@ class MonthlyActiveUsersStore(SQLBaseStore): yield self.upsert_monthly_active_user(user_id) elif now - last_seen_timestamp > LAST_SEEN_GRANULARITY: yield self.upsert_monthly_active_user(user_id) - - def is_threepid_reserved(self, threepid): - """Check the threepid against the reserved threepid config - Args: - threepid(dict) - The threepid to test for - Returns: - boolean Is the threepid undertest reserved_user - """ - for tp in self.hs.config.mau_limits_reserved_threepids: - if (threepid['medium'] == tp['medium'] - and threepid['address'] == tp['address']): - return True - else: - return False diff --git a/tests/utils.py b/tests/utils.py index 179b592501..63e30dc6c0 100644 --- a/tests/utils.py +++ b/tests/utils.py @@ -26,6 +26,7 @@ from twisted.internet import defer, reactor from synapse.api.constants import EventTypes from synapse.api.errors import CodeMessageException, cs_error +from synapse.config.server import ServerConfig from synapse.federation.transport import server from synapse.http.server import HttpServer from synapse.server import HomeServer @@ -158,6 +159,11 @@ def setup_test_homeserver( # background, which upsets the test runner. config.update_user_directory = False + def is_threepid_reserved(threepid): + return ServerConfig.is_threepid_reserved(config, threepid) + + config.is_threepid_reserved.side_effect = is_threepid_reserved + config.use_frozen_dicts = True config.ldap_enabled = False -- cgit 1.5.1 From 61b05727fa6ba879f507d13ff40c0197916ef795 Mon Sep 17 00:00:00 2001 From: Neil Johnson Date: Wed, 5 Sep 2018 22:30:36 +0100 Subject: guest users should not be part of mau total --- synapse/storage/monthly_active_users.py | 7 ++++- tests/storage/test_monthly_active_users.py | 44 ++++++++++++++++++++++++++++-- 2 files changed, 48 insertions(+), 3 deletions(-) (limited to 'synapse/storage') diff --git a/synapse/storage/monthly_active_users.py b/synapse/storage/monthly_active_users.py index c7899d7fd2..e34e16c079 100644 --- a/synapse/storage/monthly_active_users.py +++ b/synapse/storage/monthly_active_users.py @@ -192,14 +192,19 @@ class MonthlyActiveUsersStore(SQLBaseStore): )) @defer.inlineCallbacks - def populate_monthly_active_users(self, user_id): + def populate_monthly_active_users(self, user_id, is_guest=False): """Checks on the state of monthly active user limits and optionally add the user to the monthly active tables Args: user_id(str): the user_id to query """ + if self.hs.config.limit_usage_by_mau: + # Guests should not be included as part of MAU group + if is_guest: + return + is_trial = yield self.is_trial_user(user_id) if is_trial: # we don't track trial users in the MAU table. diff --git a/tests/storage/test_monthly_active_users.py b/tests/storage/test_monthly_active_users.py index 2036287288..cfe0a7c77d 100644 --- a/tests/storage/test_monthly_active_users.py +++ b/tests/storage/test_monthly_active_users.py @@ -12,6 +12,9 @@ # 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. +from mock import Mock + +from twisted.internet import defer from tests.unittest import HomeserverTestCase @@ -23,7 +26,8 @@ class MonthlyActiveUsersTestCase(HomeserverTestCase): hs = self.setup_test_homeserver() self.store = hs.get_datastore() - + hs.config.limit_usage_by_mau = True + hs.config.max_mau_value = 50 # Advance the clock a bit reactor.advance(FORTY_DAYS) @@ -73,7 +77,7 @@ class MonthlyActiveUsersTestCase(HomeserverTestCase): active_count = self.store.get_monthly_active_count() self.assertEquals(self.get_success(active_count), user_num) - # Test that regalar users are removed from the db + # Test that regular users are removed from the db ru_count = 2 self.store.upsert_monthly_active_user("@ru1:server") self.store.upsert_monthly_active_user("@ru2:server") @@ -139,3 +143,39 @@ class MonthlyActiveUsersTestCase(HomeserverTestCase): count = self.store.get_monthly_active_count() self.assertEquals(self.get_success(count), 0) + + def test_populate_monthly_users_is_guest(self): + # Test that guest users are not added to mau list + self.store.upsert_monthly_active_user = Mock() + self.store.populate_monthly_active_users('user_id', True) + self.pump() + self.store.upsert_monthly_active_user.assert_not_called() + + def test_populate_monthly_users_should_update(self): + self.store.upsert_monthly_active_user = Mock() + + self.store.is_trial_user = Mock( + return_value=defer.succeed(False) + ) + + self.store.user_last_seen_monthly_active = Mock( + return_value=defer.succeed(None) + ) + self.store.populate_monthly_active_users('user_id') + self.pump() + self.store.upsert_monthly_active_user.assert_called_once() + + def test_populate_monthly_users_should_not_update(self): + self.store.upsert_monthly_active_user = Mock() + + self.store.is_trial_user = Mock( + return_value=defer.succeed(False) + ) + self.store.user_last_seen_monthly_active = Mock( + return_value=defer.succeed( + self.hs.get_clock().time_msec() + ) + ) + self.store.populate_monthly_active_users('user_id') + self.pump() + self.store.upsert_monthly_active_user.assert_not_called() -- cgit 1.5.1 From 84a750e0c36f653b71a06c564154f257c1b7dee3 Mon Sep 17 00:00:00 2001 From: Neil Johnson Date: Thu, 6 Sep 2018 17:22:53 +0100 Subject: ensure guests never enter mau list --- synapse/storage/monthly_active_users.py | 7 +++---- tests/storage/test_client_ips.py | 4 +--- tests/storage/test_monthly_active_users.py | 6 +++++- tests/utils.py | 1 + 4 files changed, 10 insertions(+), 8 deletions(-) (limited to 'synapse/storage') diff --git a/synapse/storage/monthly_active_users.py b/synapse/storage/monthly_active_users.py index e34e16c079..b890c152db 100644 --- a/synapse/storage/monthly_active_users.py +++ b/synapse/storage/monthly_active_users.py @@ -192,7 +192,7 @@ class MonthlyActiveUsersStore(SQLBaseStore): )) @defer.inlineCallbacks - def populate_monthly_active_users(self, user_id, is_guest=False): + def populate_monthly_active_users(self, user_id): """Checks on the state of monthly active user limits and optionally add the user to the monthly active tables @@ -201,13 +201,12 @@ class MonthlyActiveUsersStore(SQLBaseStore): """ if self.hs.config.limit_usage_by_mau: - # Guests should not be included as part of MAU group + # 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: - # we don't track trial users in the MAU table. return last_seen_timestamp = yield self.user_last_seen_monthly_active(user_id) diff --git a/tests/storage/test_client_ips.py b/tests/storage/test_client_ips.py index c2e88bdbaf..c9b02a062b 100644 --- a/tests/storage/test_client_ips.py +++ b/tests/storage/test_client_ips.py @@ -101,13 +101,11 @@ class ClientIpStoreTestCase(tests.unittest.TestCase): self.hs.config.limit_usage_by_mau = True self.hs.config.max_mau_value = 50 user_id = "@user:server" + yield self.store.register(user_id=user_id, token="123", password_hash=None) active = yield self.store.user_last_seen_monthly_active(user_id) self.assertFalse(active) - yield self.store.insert_client_ip( - user_id, "access_token", "ip", "user_agent", "device_id" - ) yield self.store.insert_client_ip( user_id, "access_token", "ip", "user_agent", "device_id" ) diff --git a/tests/storage/test_monthly_active_users.py b/tests/storage/test_monthly_active_users.py index cfe0a7c77d..ccfc21b7fc 100644 --- a/tests/storage/test_monthly_active_users.py +++ b/tests/storage/test_monthly_active_users.py @@ -146,8 +146,12 @@ class MonthlyActiveUsersTestCase(HomeserverTestCase): def test_populate_monthly_users_is_guest(self): # Test that guest users are not added to mau list + user_id = "user_id" + self.store.register( + user_id=user_id, token="123", password_hash=None, make_guest=True + ) self.store.upsert_monthly_active_user = Mock() - self.store.populate_monthly_active_users('user_id', True) + self.store.populate_monthly_active_users(user_id) self.pump() self.store.upsert_monthly_active_user.assert_not_called() diff --git a/tests/utils.py b/tests/utils.py index 114470d641..f24bd81bf6 100644 --- a/tests/utils.py +++ b/tests/utils.py @@ -147,6 +147,7 @@ def setup_test_homeserver( config.hs_disabled_message = "" config.hs_disabled_limit_type = "" config.max_mau_value = 50 + config.mau_trial_days = 0 config.mau_limits_reserved_threepids = [] config.admin_contact = None config.rc_messages_per_second = 10000 -- cgit 1.5.1 From cd7ef4387205eaeb64b77860a82f782f91e41e93 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Thu, 6 Sep 2018 23:56:47 +0100 Subject: clearer logging when things fail, too --- synapse/federation/federation_base.py | 34 +++++++++++++++++++++++++++------- synapse/storage/keys.py | 1 + 2 files changed, 28 insertions(+), 7 deletions(-) (limited to 'synapse/storage') diff --git a/synapse/federation/federation_base.py b/synapse/federation/federation_base.py index 5be8e66fb8..61782ae1c0 100644 --- a/synapse/federation/federation_base.py +++ b/synapse/federation/federation_base.py @@ -143,11 +143,31 @@ class FederationBase(object): def callback(_, pdu): with logcontext.PreserveLoggingContext(ctx): if not check_event_content_hash(pdu): - logger.warn( - "Event content has been tampered, redacting %s: %s", - pdu.event_id, pdu.get_pdu_json() - ) - return prune_event(pdu) + # let's try to distinguish between failures because the event was + # redacted (which are somewhat expected) vs actual ball-tampering + # incidents. + # + # This is just a heuristic, so we just assume that if the keys are + # about the same between the redacted and received events, then the + # received event was probably a redacted copy (but we then use our + # *actual* redacted copy to be on the safe side.) + redacted_event = prune_event(pdu) + if ( + set(six.iterkeys(redacted_event)) == set(six.iterkeys(pdu)) and + set(six.iterkeys(redacted_event.content)) + == set(six.iterkeys(pdu.content)) + ): + logger.info( + "Event %s seems to have been redacted; using our redacted " + "copy", + pdu.event_id, + ) + else: + logger.warning( + "Event %s content has been tampered, redacting", + pdu.event_id, pdu.get_pdu_json(), + ) + return redacted_event if self.spam_checker.check_event_for_spam(pdu): logger.warn( @@ -162,8 +182,8 @@ class FederationBase(object): failure.trap(SynapseError) with logcontext.PreserveLoggingContext(ctx): logger.warn( - "Signature check failed for %s", - pdu.event_id, + "Signature check failed for %s: %s", + pdu.event_id, failure.getErrorMessage(), ) return failure diff --git a/synapse/storage/keys.py b/synapse/storage/keys.py index f547977600..a1331c1a61 100644 --- a/synapse/storage/keys.py +++ b/synapse/storage/keys.py @@ -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, -- cgit 1.5.1 From b041115415191f8353177f7e0c1beec81be0c921 Mon Sep 17 00:00:00 2001 From: Matthew Hodgson Date: Wed, 12 Sep 2018 00:50:39 +0100 Subject: Speed up lazy loading (#3827) * speed up room summaries by pulling their data from room_memberships rather than room state * disable LL for incr syncs, and log incr sync stats (#3840) --- changelog.d/3827.misc | 1 + changelog.d/3840.misc | 1 + synapse/handlers/sync.py | 118 +++++++++++++++++++++++++++++++----------- synapse/storage/events.py | 4 ++ synapse/storage/roommember.py | 65 +++++++++++++++++++++++ 5 files changed, 159 insertions(+), 30 deletions(-) create mode 100644 changelog.d/3827.misc create mode 100644 changelog.d/3840.misc (limited to 'synapse/storage') diff --git a/changelog.d/3827.misc b/changelog.d/3827.misc new file mode 100644 index 0000000000..bc294706cf --- /dev/null +++ b/changelog.d/3827.misc @@ -0,0 +1 @@ +speed up lazy loading by 2-3x diff --git a/changelog.d/3840.misc b/changelog.d/3840.misc new file mode 100644 index 0000000000..b9585ae9be --- /dev/null +++ b/changelog.d/3840.misc @@ -0,0 +1 @@ +Disable lazy loading for incremental syncs for now diff --git a/synapse/handlers/sync.py b/synapse/handlers/sync.py index 7eed2fcc9b..23983a51ab 100644 --- a/synapse/handlers/sync.py +++ b/synapse/handlers/sync.py @@ -24,6 +24,7 @@ from twisted.internet import defer from synapse.api.constants import EventTypes, Membership from synapse.push.clientformat import format_push_rules_for_user +from synapse.storage.roommember import MemberSummary from synapse.types import RoomStreamToken from synapse.util.async_helpers import concurrently_execute from synapse.util.caches.expiringcache import ExpiringCache @@ -525,6 +526,8 @@ class SyncHandler(object): A deferred dict describing the room summary """ + # FIXME: we could/should get this from room_stats when matthew/stats lands + # FIXME: this promulgates https://github.com/matrix-org/synapse/issues/3305 last_events, _ = yield self.store.get_recent_event_ids_for_room( room_id, end_token=now_token.room_key, limit=1, @@ -537,44 +540,54 @@ class SyncHandler(object): last_event = last_events[-1] state_ids = yield self.store.get_state_ids_for_event( last_event.event_id, [ - (EventTypes.Member, None), (EventTypes.Name, ''), (EventTypes.CanonicalAlias, ''), ] ) - member_ids = { - state_key: event_id - for (t, state_key), event_id in iteritems(state_ids) - if t == EventTypes.Member - } + # this is heavily cached, thus: fast. + details = yield self.store.get_room_summary(room_id) + name_id = state_ids.get((EventTypes.Name, '')) canonical_alias_id = state_ids.get((EventTypes.CanonicalAlias, '')) summary = {} - - # FIXME: it feels very heavy to load up every single membership event - # just to calculate the counts. - member_events = yield self.store.get_events(member_ids.values()) - - joined_user_ids = [] - invited_user_ids = [] - - for ev in member_events.values(): - if ev.content.get("membership") == Membership.JOIN: - joined_user_ids.append(ev.state_key) - elif ev.content.get("membership") == Membership.INVITE: - invited_user_ids.append(ev.state_key) + empty_ms = MemberSummary([], 0) # TODO: only send these when they change. - summary["m.joined_member_count"] = len(joined_user_ids) - summary["m.invited_member_count"] = len(invited_user_ids) + summary["m.joined_member_count"] = ( + details.get(Membership.JOIN, empty_ms).count + ) + summary["m.invited_member_count"] = ( + details.get(Membership.INVITE, empty_ms).count + ) if name_id or canonical_alias_id: defer.returnValue(summary) - # FIXME: order by stream ordering, not alphabetic + joined_user_ids = [ + r[0] for r in details.get(Membership.JOIN, empty_ms).members + ] + invited_user_ids = [ + r[0] for r in details.get(Membership.INVITE, empty_ms).members + ] + gone_user_ids = ( + [r[0] for r in details.get(Membership.LEAVE, empty_ms).members] + + [r[0] for r in details.get(Membership.BAN, empty_ms).members] + ) + + # FIXME: only build up a member_ids list for our heroes + member_ids = {} + for membership in ( + Membership.JOIN, + Membership.INVITE, + Membership.LEAVE, + Membership.BAN + ): + for user_id, event_id in details.get(membership, empty_ms).members: + member_ids[user_id] = event_id + # FIXME: order by stream ordering rather than as returned by SQL me = sync_config.user.to_string() if (joined_user_ids or invited_user_ids): summary['m.heroes'] = sorted( @@ -586,7 +599,11 @@ class SyncHandler(object): )[0:5] else: summary['m.heroes'] = sorted( - [user_id for user_id in member_ids.keys() if user_id != me] + [ + user_id + for user_id in gone_user_ids + if user_id != me + ] )[0:5] if not sync_config.filter_collection.lazy_load_members(): @@ -719,6 +736,26 @@ class SyncHandler(object): lazy_load_members=lazy_load_members, ) elif batch.limited: + state_at_timeline_start = yield self.store.get_state_ids_for_event( + batch.events[0].event_id, types=types, + filtered_types=filtered_types, + ) + + # for now, we disable LL for gappy syncs - see + # https://github.com/vector-im/riot-web/issues/7211#issuecomment-419976346 + # N.B. this slows down incr syncs as we are now processing way + # more state in the server than if we were LLing. + # + # We still have to filter timeline_start to LL entries (above) in order + # for _calculate_state's LL logic to work, as we have to include LL + # members for timeline senders in case they weren't loaded in the initial + # sync. We do this by (counterintuitively) by filtering timeline_start + # members to just be ones which were timeline senders, which then ensures + # all of the rest get included in the state block (if we need to know + # about them). + types = None + filtered_types = None + state_at_previous_sync = yield self.get_state_at( room_id, stream_position=since_token, types=types, filtered_types=filtered_types, @@ -729,24 +766,21 @@ class SyncHandler(object): filtered_types=filtered_types, ) - state_at_timeline_start = yield self.store.get_state_ids_for_event( - batch.events[0].event_id, types=types, - filtered_types=filtered_types, - ) - state_ids = _calculate_state( timeline_contains=timeline_state, timeline_start=state_at_timeline_start, previous=state_at_previous_sync, current=current_state_ids, + # we have to include LL members in case LL initial sync missed them lazy_load_members=lazy_load_members, ) else: state_ids = {} if lazy_load_members: if types: - # We're returning an incremental sync, with no "gap" since - # the previous sync, so normally there would be no state to return + # We're returning an incremental sync, with no + # "gap" since the previous sync, so normally there would be + # no state to return. # But we're lazy-loading, so the client might need some more # member events to understand the events in this timeline. # So we fish out all the member events corresponding to the @@ -1616,10 +1650,24 @@ class SyncHandler(object): ) summary = {} + + # we include a summary in room responses when we're lazy loading + # members (as the client otherwise doesn't have enough info to form + # the name itself). if ( sync_config.filter_collection.lazy_load_members() and ( + # we recalulate the summary: + # if there are membership changes in the timeline, or + # if membership has changed during a gappy sync, or + # if this is an initial sync. any(ev.type == EventTypes.Member for ev in batch.events) or + ( + # XXX: this may include false positives in the form of LL + # members which have snuck into state + batch.limited and + any(t == EventTypes.Member for (t, k) in state) + ) or since_token is None ) ): @@ -1649,6 +1697,16 @@ class SyncHandler(object): unread_notifications["highlight_count"] = notifs["highlight_count"] sync_result_builder.joined.append(room_sync) + + if batch.limited: + user_id = sync_result_builder.sync_config.user.to_string() + logger.info( + "Incremental syncing room %s for user %s with %d state events" % ( + room_id, + user_id, + len(state), + ) + ) elif room_builder.rtype == "archived": room_sync = ArchivedSyncResult( room_id=room_id, diff --git a/synapse/storage/events.py b/synapse/storage/events.py index 8bf87f38f7..30ff87a4c4 100644 --- a/synapse/storage/events.py +++ b/synapse/storage/events.py @@ -929,6 +929,10 @@ class EventsStore(EventFederationStore, EventsWorkerStore, BackgroundUpdateStore txn, self.get_users_in_room, (room_id,) ) + 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,) ) 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 -- cgit 1.5.1 From 0ddf48672421aa64920342b959dc77432c869889 Mon Sep 17 00:00:00 2001 From: Neil Johnson Date: Wed, 12 Sep 2018 11:58:52 +0100 Subject: expose number of real reserved users --- synapse/app/homeserver.py | 10 +++++++--- synapse/storage/monthly_active_users.py | 17 ++++++++++++++++ tests/storage/test_monthly_active_users.py | 31 ++++++++++++++++++++++++++++++ 3 files changed, 55 insertions(+), 3 deletions(-) (limited to 'synapse/storage') diff --git a/synapse/app/homeserver.py b/synapse/app/homeserver.py index 3eb5b663de..e6372bcec2 100755 --- a/synapse/app/homeserver.py +++ b/synapse/app/homeserver.py @@ -307,6 +307,7 @@ class SynapseHomeServer(HomeServer): # Gauges to expose monthly active user control metrics current_mau_gauge = Gauge("synapse_admin_mau:current", "Current MAU") max_mau_gauge = Gauge("synapse_admin_mau:max", "MAU Limit") +reserved_mau_gauge = Gauge("synapse_admin_mau:reserved", "Reserved real MAU users") def setup(config_options): @@ -531,10 +532,13 @@ def run(hs): @defer.inlineCallbacks def generate_monthly_active_users(): - count = 0 + current_mau_count = 0 + reserved_mau_count = 0 if hs.config.limit_usage_by_mau: - count = yield hs.get_datastore().get_monthly_active_count() - current_mau_gauge.set(float(count)) + current_mau_count = yield hs.get_datastore().get_monthly_active_count() + reserved_mau_count = yield hs.get_datastore().get_reserved_real_user_account() + current_mau_gauge.set(float(current_mau_count)) + reserved_mau_gauge.set(float(reserved_mau_count)) max_mau_gauge.set(float(hs.config.max_mau_value)) hs.get_datastore().initialise_reserved_users( diff --git a/synapse/storage/monthly_active_users.py b/synapse/storage/monthly_active_users.py index b890c152db..53d125d305 100644 --- a/synapse/storage/monthly_active_users.py +++ b/synapse/storage/monthly_active_users.py @@ -146,6 +146,23 @@ class MonthlyActiveUsersStore(SQLBaseStore): return count return self.runInteraction("count_users", _count_users) + @defer.inlineCallbacks + def get_reserved_real_user_account(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): """ diff --git a/tests/storage/test_monthly_active_users.py b/tests/storage/test_monthly_active_users.py index ccfc21b7fc..662f2ed845 100644 --- a/tests/storage/test_monthly_active_users.py +++ b/tests/storage/test_monthly_active_users.py @@ -183,3 +183,34 @@ class MonthlyActiveUsersTestCase(HomeserverTestCase): self.store.populate_monthly_active_users('user_id') self.pump() self.store.upsert_monthly_active_user.assert_not_called() + + def test_get_reserved_real_user_account(self): + # Test no reserved users, or reserved threepids + count = self.store.get_reserved_real_user_account() + self.assertEquals(self.get_success(count), 0) + # Test reserved users but no registered users + + user1 = '@user1:example.com' + user2 = '@user2:example.com' + user1_email = 'user1@example.com' + user2_email = 'user2@example.com' + threepids = [ + {'medium': 'email', 'address': user1_email}, + {'medium': 'email', 'address': user2_email}, + ] + self.hs.config.mau_limits_reserved_threepids = threepids + self.store.initialise_reserved_users(threepids) + self.pump() + count = self.store.get_reserved_real_user_account() + self.assertEquals(self.get_success(count), 0) + + # Test reserved registed users + self.store.register(user_id=user1, token="123", password_hash=None) + self.store.register(user_id=user2, token="456", password_hash=None) + self.pump() + + now = int(self.hs.get_clock().time_msec()) + self.store.user_add_threepid(user1, "email", user1_email, now, now) + self.store.user_add_threepid(user2, "email", user2_email, now, now) + count = self.store.get_reserved_real_user_account() + self.assertEquals(self.get_success(count), len(threepids)) -- cgit 1.5.1 From 8decd6233dabd87160794949ffd95282e60ab01e Mon Sep 17 00:00:00 2001 From: Neil Johnson Date: Wed, 12 Sep 2018 16:22:15 +0100 Subject: improve naming --- changelog.d/3846.feature | 2 +- synapse/app/homeserver.py | 14 +++++++++----- synapse/storage/monthly_active_users.py | 2 +- tests/storage/test_monthly_active_users.py | 6 +++--- 4 files changed, 14 insertions(+), 10 deletions(-) (limited to 'synapse/storage') diff --git a/changelog.d/3846.feature b/changelog.d/3846.feature index dc9e9d86cc..453c11d3f8 100644 --- a/changelog.d/3846.feature +++ b/changelog.d/3846.feature @@ -1 +1 @@ -create synapse_admin_mau:reserved metric to expose number of real reaserved users +Add synapse_admin_mau:registered_reserved_users metric to expose number of real reaserved users diff --git a/synapse/app/homeserver.py b/synapse/app/homeserver.py index e6372bcec2..ac97e19649 100755 --- a/synapse/app/homeserver.py +++ b/synapse/app/homeserver.py @@ -307,7 +307,10 @@ class SynapseHomeServer(HomeServer): # Gauges to expose monthly active user control metrics current_mau_gauge = Gauge("synapse_admin_mau:current", "Current MAU") max_mau_gauge = Gauge("synapse_admin_mau:max", "MAU Limit") -reserved_mau_gauge = Gauge("synapse_admin_mau:reserved", "Reserved real MAU users") +registered_reserved_users_mau_gauge = Gauge( + "synapse_admin_mau:registered_reserved_users", + "Registered users with reserved threepids" +) def setup(config_options): @@ -533,12 +536,13 @@ def run(hs): @defer.inlineCallbacks def generate_monthly_active_users(): current_mau_count = 0 - reserved_mau_count = 0 + reserved_count = 0 + store = hs.get_datastore() if hs.config.limit_usage_by_mau: - current_mau_count = yield hs.get_datastore().get_monthly_active_count() - reserved_mau_count = yield hs.get_datastore().get_reserved_real_user_account() + current_mau_count = yield store.get_monthly_active_count() + reserved_count = yield store.get_registered_reserved_users_count() current_mau_gauge.set(float(current_mau_count)) - reserved_mau_gauge.set(float(reserved_mau_count)) + registered_reserved_users_mau_gauge.set(float(reserved_count)) max_mau_gauge.set(float(hs.config.max_mau_value)) hs.get_datastore().initialise_reserved_users( diff --git a/synapse/storage/monthly_active_users.py b/synapse/storage/monthly_active_users.py index 53d125d305..59580949f1 100644 --- a/synapse/storage/monthly_active_users.py +++ b/synapse/storage/monthly_active_users.py @@ -147,7 +147,7 @@ class MonthlyActiveUsersStore(SQLBaseStore): return self.runInteraction("count_users", _count_users) @defer.inlineCallbacks - def get_reserved_real_user_account(self): + def get_registered_reserved_users_count(self): """Of the reserved threepids defined in config, how many are associated with registered users? diff --git a/tests/storage/test_monthly_active_users.py b/tests/storage/test_monthly_active_users.py index 662f2ed845..686f12a0dc 100644 --- a/tests/storage/test_monthly_active_users.py +++ b/tests/storage/test_monthly_active_users.py @@ -186,7 +186,7 @@ class MonthlyActiveUsersTestCase(HomeserverTestCase): def test_get_reserved_real_user_account(self): # Test no reserved users, or reserved threepids - count = self.store.get_reserved_real_user_account() + count = self.store.get_registered_reserved_users_count() self.assertEquals(self.get_success(count), 0) # Test reserved users but no registered users @@ -201,7 +201,7 @@ class MonthlyActiveUsersTestCase(HomeserverTestCase): self.hs.config.mau_limits_reserved_threepids = threepids self.store.initialise_reserved_users(threepids) self.pump() - count = self.store.get_reserved_real_user_account() + count = self.store.get_registered_reserved_users_count() self.assertEquals(self.get_success(count), 0) # Test reserved registed users @@ -212,5 +212,5 @@ class MonthlyActiveUsersTestCase(HomeserverTestCase): now = int(self.hs.get_clock().time_msec()) self.store.user_add_threepid(user1, "email", user1_email, now, now) self.store.user_add_threepid(user2, "email", user2_email, now, now) - count = self.store.get_reserved_real_user_account() + count = self.store.get_registered_reserved_users_count() self.assertEquals(self.get_success(count), len(threepids)) -- cgit 1.5.1 From c857f5ef9b77b07179555a63973f11b8aa8c1752 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Thu, 13 Sep 2018 12:48:10 +0100 Subject: Make purge history slightly faster Don't pull out events that are outliers and won't be deleted, as nothing should happen to them. --- synapse/storage/events.py | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) (limited to 'synapse/storage') diff --git a/synapse/storage/events.py b/synapse/storage/events.py index 30ff87a4c4..1c8e01a47e 100644 --- a/synapse/storage/events.py +++ b/synapse/storage/events.py @@ -1930,15 +1930,22 @@ 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, ) + 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, -- cgit 1.5.1 From 9dbe38ea7de66f6e5dfab1570dd3a5217a4883eb Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Thu, 13 Sep 2018 15:05:52 +0100 Subject: Create indices after insertion --- synapse/storage/events.py | 32 ++++++++++++++++++-------------- 1 file changed, 18 insertions(+), 14 deletions(-) (limited to 'synapse/storage') diff --git a/synapse/storage/events.py b/synapse/storage/events.py index 1c8e01a47e..622f2ababf 100644 --- a/synapse/storage/events.py +++ b/synapse/storage/events.py @@ -1890,20 +1890,6 @@ 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 " @@ -1950,6 +1936,24 @@ class EventsStore(EventFederationStore, EventsWorkerStore, BackgroundUpdateStore ), 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" ) -- cgit 1.5.1 From 9cbd0094f0dcf03676b3234896e73af80cb75021 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Thu, 13 Sep 2018 15:15:35 +0100 Subject: pep8 --- synapse/storage/events.py | 1 - 1 file changed, 1 deletion(-) (limited to 'synapse/storage') diff --git a/synapse/storage/events.py b/synapse/storage/events.py index 622f2ababf..bc1c18023b 100644 --- a/synapse/storage/events.py +++ b/synapse/storage/events.py @@ -1953,7 +1953,6 @@ class EventsStore(EventFederationStore, EventsWorkerStore, BackgroundUpdateStore " ON events_to_purge(event_id)", ) - txn.execute( "SELECT event_id, should_delete FROM events_to_purge" ) -- cgit 1.5.1 From ed5331a6272280efd4fc98ae425711acec8c6be5 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Thu, 13 Sep 2018 16:10:56 +0100 Subject: comment --- synapse/storage/events.py | 2 ++ 1 file changed, 2 insertions(+) (limited to 'synapse/storage') diff --git a/synapse/storage/events.py b/synapse/storage/events.py index bc1c18023b..e7487311ce 100644 --- a/synapse/storage/events.py +++ b/synapse/storage/events.py @@ -1916,6 +1916,8 @@ class EventsStore(EventFederationStore, EventsWorkerStore, BackgroundUpdateStore should_delete_params = () if not delete_local_events: should_delete_expr += " AND event_id NOT LIKE ?" + + # We include the parameter twice since we use the expression twice should_delete_params += ( "%:" + self.hs.hostname, "%:" + self.hs.hostname, -- cgit 1.5.1 From 85a43f4167b5e775f7c3afe96d71137818561272 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Mon, 17 Sep 2018 13:19:00 +0100 Subject: Return a 404 when deleting unknown room alias As per https://github.com/matrix-org/matrix-doc/issues/1675 Fixes https://github.com/matrix-org/synapse/issues/2782 --- synapse/handlers/directory.py | 19 ++++++++++++++++--- synapse/storage/directory.py | 1 - 2 files changed, 16 insertions(+), 4 deletions(-) (limited to 'synapse/storage') diff --git a/synapse/handlers/directory.py b/synapse/handlers/directory.py index ef866da1b6..c745e6740b 100644 --- a/synapse/handlers/directory.py +++ b/synapse/handlers/directory.py @@ -20,7 +20,14 @@ import string from twisted.internet import defer from synapse.api.constants import EventTypes -from synapse.api.errors import AuthError, CodeMessageException, Codes, SynapseError +from synapse.api.errors import ( + AuthError, + CodeMessageException, + Codes, + NotFoundError, + StoreError, + SynapseError, +) from synapse.types import RoomAlias, UserID, get_domain_from_id from ._base import BaseHandler @@ -109,7 +116,13 @@ class DirectoryHandler(BaseHandler): def delete_association(self, requester, user_id, room_alias): # association deletion for human users - can_delete = yield self._user_can_delete_alias(room_alias, user_id) + try: + can_delete = yield self._user_can_delete_alias(room_alias, user_id) + except StoreError as e: + if e.code == 404: + raise NotFoundError("Unknown room alias") + raise + if not can_delete: raise AuthError( 403, "You don't have permission to delete the alias.", @@ -320,7 +333,7 @@ class DirectoryHandler(BaseHandler): def _user_can_delete_alias(self, alias, user_id): creator = yield self.store.get_room_alias_creator(alias.to_string()) - if creator and creator == user_id: + if creator == user_id: defer.returnValue(True) is_admin = yield self.auth.is_server_admin(UserID.from_string(user_id)) diff --git a/synapse/storage/directory.py b/synapse/storage/directory.py index 808194236a..cfb687cb53 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) -- cgit 1.5.1 From b9158ac2bf46e15107162be136964c17bd8a948c Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 19 Sep 2018 14:16:34 +0100 Subject: Remove get_destination_retry_timings cache Currently we rely on the master to invalidate this cache promptly. However, after having moved most federation endpoints off of master this no longer happens, causing outbound fedeariont to get blackholed. Fixes #3798 --- synapse/storage/transactions.py | 5 ----- 1 file changed, 5 deletions(-) (limited to 'synapse/storage') diff --git a/synapse/storage/transactions.py b/synapse/storage/transactions.py index 0c42bd3322..7d51f9612a 100644 --- a/synapse/storage/transactions.py +++ b/synapse/storage/transactions.py @@ -156,7 +156,6 @@ class TransactionStore(SQLBaseStore): """ pass - @cached(max_entries=10000) def get_destination_retry_timings(self, destination): """Gets the current retry timings (if any) for a given destination. @@ -212,10 +211,6 @@ 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. -- cgit 1.5.1 From 392a54128c03128255ee538e17ed9bd87b95d4ca Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 19 Sep 2018 14:37:49 +0100 Subject: pep8 --- synapse/storage/transactions.py | 1 - 1 file changed, 1 deletion(-) (limited to 'synapse/storage') diff --git a/synapse/storage/transactions.py b/synapse/storage/transactions.py index 7d51f9612a..449bdc1e49 100644 --- a/synapse/storage/transactions.py +++ b/synapse/storage/transactions.py @@ -23,7 +23,6 @@ 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 ._base import SQLBaseStore, db_to_json -- cgit 1.5.1 From bbab6ebfd9257ba9ad5e576c4b9b8bdea1c166cf Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 19 Sep 2018 14:45:14 +0100 Subject: Fix up changelog and remove spurious comment --- changelog.d/3914.bugfix | 2 +- synapse/storage/transactions.py | 2 -- 2 files changed, 1 insertion(+), 3 deletions(-) (limited to 'synapse/storage') diff --git a/changelog.d/3914.bugfix b/changelog.d/3914.bugfix index 60fb8ec1ae..27e6bad590 100644 --- a/changelog.d/3914.bugfix +++ b/changelog.d/3914.bugfix @@ -1 +1 @@ -Fix bug where outboud federation would stop talking to some servers when using workers +Fix bug where outbound federation would stop talking to some servers when using workers diff --git a/synapse/storage/transactions.py b/synapse/storage/transactions.py index 449bdc1e49..baf0379a68 100644 --- a/synapse/storage/transactions.py +++ b/synapse/storage/transactions.py @@ -196,8 +196,6 @@ 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 return self.runInteraction( "set_destination_retry_timings", self._set_destination_retry_timings, -- cgit 1.5.1 From 1f3f5fcf52efdf5215dc6c7cd3805b1357bf8236 Mon Sep 17 00:00:00 2001 From: Amber Brown Date: Thu, 20 Sep 2018 20:14:34 +1000 Subject: Fix client IPs being broken on Python 3 (#3908) --- .travis.yml | 5 + changelog.d/3908.bugfix | 1 + synapse/http/site.py | 2 +- synapse/storage/client_ips.py | 34 ++++--- tests/server.py | 8 +- tests/storage/test_client_ips.py | 202 ++++++++++++++++++++++++++++++++------- tests/unittest.py | 7 +- tests/utils.py | 27 +++++- tox.ini | 10 ++ 9 files changed, 238 insertions(+), 58 deletions(-) create mode 100644 changelog.d/3908.bugfix (limited to 'synapse/storage') diff --git a/.travis.yml b/.travis.yml index b3ee66da8f..b6faca4b92 100644 --- a/.travis.yml +++ b/.travis.yml @@ -31,6 +31,11 @@ matrix: - python: 3.6 env: TOX_ENV=py36 + - python: 3.6 + env: TOX_ENV=py36-postgres TRIAL_FLAGS="-j 4" + services: + - postgresql + - python: 3.6 env: TOX_ENV=check_isort diff --git a/changelog.d/3908.bugfix b/changelog.d/3908.bugfix new file mode 100644 index 0000000000..518aee6c4d --- /dev/null +++ b/changelog.d/3908.bugfix @@ -0,0 +1 @@ +Fix adding client IPs to the database failing on Python 3. \ No newline at end of file diff --git a/synapse/http/site.py b/synapse/http/site.py index d41d672934..50be2de3bb 100644 --- a/synapse/http/site.py +++ b/synapse/http/site.py @@ -308,7 +308,7 @@ class XForwardedForRequest(SynapseRequest): C{b"-"}. """ return self.requestHeaders.getRawHeaders( - b"x-forwarded-for", [b"-"])[0].split(b",")[0].strip() + b"x-forwarded-for", [b"-"])[0].split(b",")[0].strip().decode('ascii') class SynapseRequestFactory(object): 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/tests/server.py b/tests/server.py index ccea3baa55..7bee58dff1 100644 --- a/tests/server.py +++ b/tests/server.py @@ -98,7 +98,7 @@ class FakeSite: return FakeLogger() -def make_request(method, path, content=b"", access_token=None): +def make_request(method, path, content=b"", access_token=None, request=SynapseRequest): """ Make a web request using the given method and path, feed it the content, and return the Request and the Channel underneath. @@ -120,14 +120,16 @@ def make_request(method, path, content=b"", access_token=None): site = FakeSite() channel = FakeChannel() - req = SynapseRequest(site, channel) + req = request(site, channel) req.process = lambda: b"" req.content = BytesIO(content) if access_token: req.requestHeaders.addRawHeader(b"Authorization", b"Bearer " + access_token) - req.requestHeaders.addRawHeader(b"X-Forwarded-For", b"127.0.0.1") + if content: + req.requestHeaders.addRawHeader(b"Content-Type", b"application/json") + req.requestReceived(method, path, b"1.1") return req, channel diff --git a/tests/storage/test_client_ips.py b/tests/storage/test_client_ips.py index c9b02a062b..2ffbb9f14f 100644 --- a/tests/storage/test_client_ips.py +++ b/tests/storage/test_client_ips.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2016 OpenMarket Ltd +# Copyright 2018 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. @@ -12,35 +13,45 @@ # 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 hashlib +import hmac +import json + from mock import Mock from twisted.internet import defer -import tests.unittest -import tests.utils +from synapse.http.site import XForwardedForRequest +from synapse.rest.client.v1 import admin, login + +from tests import unittest -class ClientIpStoreTestCase(tests.unittest.TestCase): - def __init__(self, *args, **kwargs): - super(ClientIpStoreTestCase, self).__init__(*args, **kwargs) - self.store = None # type: synapse.storage.DataStore - self.clock = None # type: tests.utils.MockClock +class ClientIpStoreTestCase(unittest.HomeserverTestCase): + def make_homeserver(self, reactor, clock): + hs = self.setup_test_homeserver() + return hs - @defer.inlineCallbacks - def setUp(self): - self.hs = yield tests.utils.setup_test_homeserver(self.addCleanup) + def prepare(self, hs, reactor, clock): self.store = self.hs.get_datastore() - self.clock = self.hs.get_clock() - @defer.inlineCallbacks def test_insert_new_client_ip(self): - self.clock.now = 12345678 + self.reactor.advance(12345678) + user_id = "@user:id" - yield self.store.insert_client_ip( - user_id, "access_token", "ip", "user_agent", "device_id" + self.get_success( + self.store.insert_client_ip( + user_id, "access_token", "ip", "user_agent", "device_id" + ) ) - result = yield self.store.get_last_client_ip_by_device(user_id, "device_id") + # Trigger the storage loop + self.reactor.advance(10) + + result = self.get_success( + self.store.get_last_client_ip_by_device(user_id, "device_id") + ) r = result[(user_id, "device_id")] self.assertDictContainsSubset( @@ -55,18 +66,18 @@ class ClientIpStoreTestCase(tests.unittest.TestCase): r, ) - @defer.inlineCallbacks def test_disabled_monthly_active_user(self): self.hs.config.limit_usage_by_mau = False self.hs.config.max_mau_value = 50 user_id = "@user:server" - yield self.store.insert_client_ip( - user_id, "access_token", "ip", "user_agent", "device_id" + self.get_success( + self.store.insert_client_ip( + user_id, "access_token", "ip", "user_agent", "device_id" + ) ) - active = yield self.store.user_last_seen_monthly_active(user_id) + active = self.get_success(self.store.user_last_seen_monthly_active(user_id)) self.assertFalse(active) - @defer.inlineCallbacks def test_adding_monthly_active_user_when_full(self): self.hs.config.limit_usage_by_mau = True self.hs.config.max_mau_value = 50 @@ -76,38 +87,159 @@ class ClientIpStoreTestCase(tests.unittest.TestCase): self.store.get_monthly_active_count = Mock( return_value=defer.succeed(lots_of_users) ) - yield self.store.insert_client_ip( - user_id, "access_token", "ip", "user_agent", "device_id" + self.get_success( + self.store.insert_client_ip( + user_id, "access_token", "ip", "user_agent", "device_id" + ) ) - active = yield self.store.user_last_seen_monthly_active(user_id) + active = self.get_success(self.store.user_last_seen_monthly_active(user_id)) self.assertFalse(active) - @defer.inlineCallbacks def test_adding_monthly_active_user_when_space(self): self.hs.config.limit_usage_by_mau = True self.hs.config.max_mau_value = 50 user_id = "@user:server" - active = yield self.store.user_last_seen_monthly_active(user_id) + active = self.get_success(self.store.user_last_seen_monthly_active(user_id)) self.assertFalse(active) - yield self.store.insert_client_ip( - user_id, "access_token", "ip", "user_agent", "device_id" + # Trigger the saving loop + self.reactor.advance(10) + + self.get_success( + self.store.insert_client_ip( + user_id, "access_token", "ip", "user_agent", "device_id" + ) ) - active = yield self.store.user_last_seen_monthly_active(user_id) + active = self.get_success(self.store.user_last_seen_monthly_active(user_id)) self.assertTrue(active) - @defer.inlineCallbacks def test_updating_monthly_active_user_when_space(self): self.hs.config.limit_usage_by_mau = True self.hs.config.max_mau_value = 50 user_id = "@user:server" - yield self.store.register(user_id=user_id, token="123", password_hash=None) + self.get_success( + self.store.register(user_id=user_id, token="123", password_hash=None) + ) - active = yield self.store.user_last_seen_monthly_active(user_id) + active = self.get_success(self.store.user_last_seen_monthly_active(user_id)) self.assertFalse(active) - yield self.store.insert_client_ip( - user_id, "access_token", "ip", "user_agent", "device_id" + # Trigger the saving loop + self.reactor.advance(10) + + self.get_success( + self.store.insert_client_ip( + user_id, "access_token", "ip", "user_agent", "device_id" + ) ) - active = yield self.store.user_last_seen_monthly_active(user_id) + active = self.get_success(self.store.user_last_seen_monthly_active(user_id)) self.assertTrue(active) + + +class ClientIpAuthTestCase(unittest.HomeserverTestCase): + + servlets = [admin.register_servlets, login.register_servlets] + + def make_homeserver(self, reactor, clock): + hs = self.setup_test_homeserver() + return hs + + def prepare(self, hs, reactor, clock): + self.hs.config.registration_shared_secret = u"shared" + self.store = self.hs.get_datastore() + + # Create the user + request, channel = self.make_request("GET", "/_matrix/client/r0/admin/register") + self.render(request) + nonce = channel.json_body["nonce"] + + want_mac = hmac.new(key=b"shared", digestmod=hashlib.sha1) + want_mac.update(nonce.encode('ascii') + b"\x00bob\x00abc123\x00admin") + want_mac = want_mac.hexdigest() + + body = json.dumps( + { + "nonce": nonce, + "username": "bob", + "password": "abc123", + "admin": True, + "mac": want_mac, + } + ) + request, channel = self.make_request( + "POST", "/_matrix/client/r0/admin/register", body.encode('utf8') + ) + self.render(request) + + self.assertEqual(channel.code, 200) + self.user_id = channel.json_body["user_id"] + + def test_request_with_xforwarded(self): + """ + The IP in X-Forwarded-For is entered into the client IPs table. + """ + self._runtest( + {b"X-Forwarded-For": b"127.9.0.1"}, + "127.9.0.1", + {"request": XForwardedForRequest}, + ) + + def test_request_from_getPeer(self): + """ + The IP returned by getPeer is entered into the client IPs table, if + there's no X-Forwarded-For header. + """ + self._runtest({}, "127.0.0.1", {}) + + def _runtest(self, headers, expected_ip, make_request_args): + device_id = "bleb" + + body = json.dumps( + { + "type": "m.login.password", + "user": "bob", + "password": "abc123", + "device_id": device_id, + } + ) + request, channel = self.make_request( + "POST", "/_matrix/client/r0/login", body.encode('utf8'), **make_request_args + ) + self.render(request) + self.assertEqual(channel.code, 200) + access_token = channel.json_body["access_token"].encode('ascii') + + # Advance to a known time + self.reactor.advance(123456 - self.reactor.seconds()) + + request, channel = self.make_request( + "GET", + "/_matrix/client/r0/admin/users/" + self.user_id, + body.encode('utf8'), + access_token=access_token, + **make_request_args + ) + request.requestHeaders.addRawHeader(b"User-Agent", b"Mozzila pizza") + + # Add the optional headers + for h, v in headers.items(): + request.requestHeaders.addRawHeader(h, v) + self.render(request) + + # Advance so the save loop occurs + self.reactor.advance(100) + + result = self.get_success( + self.store.get_last_client_ip_by_device(self.user_id, device_id) + ) + r = result[(self.user_id, device_id)] + self.assertDictContainsSubset( + { + "user_id": self.user_id, + "device_id": device_id, + "ip": expected_ip, + "user_agent": "Mozzila pizza", + "last_seen": 123456100, + }, + r, + ) diff --git a/tests/unittest.py b/tests/unittest.py index 56f3dca394..ef905e6389 100644 --- a/tests/unittest.py +++ b/tests/unittest.py @@ -26,6 +26,7 @@ from twisted.internet.defer import Deferred from twisted.trial import unittest from synapse.http.server import JsonResource +from synapse.http.site import SynapseRequest from synapse.server import HomeServer from synapse.types import UserID, create_requester from synapse.util.logcontext import LoggingContextFilter @@ -237,7 +238,9 @@ class HomeserverTestCase(TestCase): Function to optionally be overridden in subclasses. """ - def make_request(self, method, path, content=b""): + def make_request( + self, method, path, content=b"", access_token=None, request=SynapseRequest + ): """ Create a SynapseRequest at the path using the method and containing the given content. @@ -255,7 +258,7 @@ class HomeserverTestCase(TestCase): if isinstance(content, dict): content = json.dumps(content).encode('utf8') - return make_request(method, path, content) + return make_request(method, path, content, access_token, request) def render(self, request): """ diff --git a/tests/utils.py b/tests/utils.py index 215226debf..aaed1149c3 100644 --- a/tests/utils.py +++ b/tests/utils.py @@ -16,7 +16,9 @@ import atexit import hashlib import os +import time import uuid +import warnings from inspect import getcallargs from mock import Mock, patch @@ -237,20 +239,41 @@ def setup_test_homeserver( else: # We need to do cleanup on PostgreSQL def cleanup(): + import psycopg2 + # Close all the db pools hs.get_db_pool().close() + dropped = False + # Drop the test database db_conn = db_engine.module.connect( database=POSTGRES_BASE_DB, user=POSTGRES_USER ) db_conn.autocommit = True cur = db_conn.cursor() - cur.execute("DROP DATABASE IF EXISTS %s;" % (test_db,)) - db_conn.commit() + + # Try a few times to drop the DB. Some things may hold on to the + # database for a few more seconds due to flakiness, preventing + # us from dropping it when the test is over. If we can't drop + # it, warn and move on. + for x in range(5): + try: + cur.execute("DROP DATABASE IF EXISTS %s;" % (test_db,)) + db_conn.commit() + dropped = True + except psycopg2.OperationalError as e: + warnings.warn( + "Couldn't drop old db: " + str(e), category=UserWarning + ) + time.sleep(0.5) + cur.close() db_conn.close() + if not dropped: + warnings.warn("Failed to drop old DB.", category=UserWarning) + if not LEAVE_DB: # Register the cleanup hook cleanup_func(cleanup) diff --git a/tox.ini b/tox.ini index 88875c3318..e4db563b4b 100644 --- a/tox.ini +++ b/tox.ini @@ -70,6 +70,16 @@ usedevelop=true [testenv:py36] usedevelop=true +[testenv:py36-postgres] +usedevelop=true +deps = + {[base]deps} + psycopg2 +setenv = + {[base]setenv} + SYNAPSE_POSTGRES = 1 + + [testenv:packaging] deps = check-manifest -- cgit 1.5.1 From fdd1a62e8d09ddccbe685fe7d7840990a9c06241 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Fri, 21 Sep 2018 14:55:47 +0100 Subject: Add a five minute cache to get_destination_retry_timings Hopefully helps with #3931 --- synapse/storage/transactions.py | 23 ++++++++++++++++++++++- synapse/util/caches/expiringcache.py | 13 +++++++++++++ 2 files changed, 35 insertions(+), 1 deletion(-) (limited to 'synapse/storage') diff --git a/synapse/storage/transactions.py b/synapse/storage/transactions.py index baf0379a68..ab54977a75 100644 --- a/synapse/storage/transactions.py +++ b/synapse/storage/transactions.py @@ -23,6 +23,7 @@ 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.expiringcache import ExpiringCache from ._base import SQLBaseStore, db_to_json @@ -49,6 +50,8 @@ _UpdateTransactionRow = namedtuple( ) ) +SENTINEL = object() + class TransactionStore(SQLBaseStore): """A collection of queries for handling PDUs. @@ -59,6 +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 @@ -155,6 +164,7 @@ class TransactionStore(SQLBaseStore): """ pass + @defer.inlineCallbacks def get_destination_retry_timings(self, destination): """Gets the current retry timings (if any) for a given destination. @@ -165,10 +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, @@ -196,6 +216,7 @@ class TransactionStore(SQLBaseStore): retry_interval (int) - how long until next retry in ms """ + self._destination_retry_cache.pop(destination) return self.runInteraction( "set_destination_retry_timings", self._set_destination_retry_timings, diff --git a/synapse/util/caches/expiringcache.py b/synapse/util/caches/expiringcache.py index 48ca2d634d..346669e4ce 100644 --- a/synapse/util/caches/expiringcache.py +++ b/synapse/util/caches/expiringcache.py @@ -24,6 +24,9 @@ from synapse.util.caches import register_cache logger = logging.getLogger(__name__) +SENTINEL = object() + + class ExpiringCache(object): def __init__(self, cache_name, clock, max_len=0, expiry_ms=0, reset_expiry_on_get=False, iterable=False): @@ -102,6 +105,16 @@ class ExpiringCache(object): return entry.value + def pop(self, key, default=None): + value = self._cache.pop(key, SENTINEL) + if value is SENTINEL: + return default + + if self.iterable: + self._size_estimate -= len(value.value) + + return value + def __contains__(self, key): return key in self._cache -- cgit 1.5.1 From 28781b65e7cb8b0d1fc409d0d78253662b49fe06 Mon Sep 17 00:00:00 2001 From: Neil Johnson Date: Wed, 26 Sep 2018 16:16:41 +0100 Subject: fix #3854 --- synapse/storage/monthly_active_users.py | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'synapse/storage') diff --git a/synapse/storage/monthly_active_users.py b/synapse/storage/monthly_active_users.py index 59580949f1..0fe8c8e24c 100644 --- a/synapse/storage/monthly_active_users.py +++ b/synapse/storage/monthly_active_users.py @@ -172,6 +172,10 @@ class MonthlyActiveUsersStore(SQLBaseStore): Deferred[bool]: True if a new entry was created, False if an existing one was updated. """ + # 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 = yield self._simple_upsert( desc="upsert_monthly_active_user", table="monthly_active_users", @@ -181,7 +185,6 @@ class MonthlyActiveUsersStore(SQLBaseStore): values={ "timestamp": int(self._clock.time_msec()), }, - lock=False, ) if is_insert: self.user_last_seen_monthly_active.invalidate((user_id,)) -- cgit 1.5.1 From ae6ad4cf4190dbd2dadd72fc8131e4c139f8eb4a Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Thu, 27 Sep 2018 11:22:25 +0100 Subject: docstrings and unittests for storage.state (#3958) I spent ages trying to figure out how I was going mad... --- changelog.d/3958.misc | 1 + synapse/storage/state.py | 30 ++++++++++++++++++++++-------- tests/storage/test_state.py | 39 +++++++++++++++++++++++++++++++++++++++ 3 files changed, 62 insertions(+), 8 deletions(-) create mode 100644 changelog.d/3958.misc (limited to 'synapse/storage') diff --git a/changelog.d/3958.misc b/changelog.d/3958.misc new file mode 100644 index 0000000000..5931d06dcf --- /dev/null +++ b/changelog.d/3958.misc @@ -0,0 +1 @@ +Fix docstrings and add tests for state store methods diff --git a/synapse/storage/state.py b/synapse/storage/state.py index 4b971efdba..3f4cbd61c4 100644 --- a/synapse/storage/state.py +++ b/synapse/storage/state.py @@ -255,7 +255,17 @@ 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({}) @@ -270,7 +280,7 @@ 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) @@ -286,7 +296,9 @@ 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({}) @@ -324,7 +336,9 @@ class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore): member events (if True), or to exclude member events (if False) Returns: - dictionary state_group -> (dict of (type, state_key) -> event id) + Returns: + Deferred[dict[int, dict[tuple[str, str], str]]]: + dict of state_group_id -> (dict of (type, state_key) -> event id) """ results = {} @@ -732,8 +746,8 @@ class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore): If None, `types` filtering is applied to all events. Returns: - Deferred[dict[int, dict[(type, state_key), EventBase]]] - a dictionary mapping from state group to state dictionary. + Deferred[dict[int, dict[tuple[str, str], str]]]: + dict of state_group_id -> (dict of (type, state_key) -> event id) """ if types is not None: non_member_types = [t for t in types if t[0] != EventTypes.Member] @@ -788,8 +802,8 @@ class StateGroupWorkerStore(EventsWorkerStore, SQLBaseStore): If None, `types` filtering is applied to all events. Returns: - Deferred[dict[int, dict[(type, state_key), EventBase]]] - a dictionary mapping from state group to state dictionary. + Deferred[dict[int, dict[tuple[str, str], str]]]: + dict of state_group_id -> (dict of (type, state_key) -> event id) """ if types: types = frozenset(types) diff --git a/tests/storage/test_state.py b/tests/storage/test_state.py index b910965932..b9c5b39d59 100644 --- a/tests/storage/test_state.py +++ b/tests/storage/test_state.py @@ -74,6 +74,45 @@ class StateStoreTestCase(tests.unittest.TestCase): self.assertEqual(s1[t].event_id, s2[t].event_id) self.assertEqual(len(s1), len(s2)) + @defer.inlineCallbacks + def test_get_state_groups_ids(self): + e1 = yield self.inject_state_event( + self.room, self.u_alice, EventTypes.Create, '', {} + ) + e2 = yield self.inject_state_event( + self.room, self.u_alice, EventTypes.Name, '', {"name": "test room"} + ) + + state_group_map = yield self.store.get_state_groups_ids(self.room, [e2.event_id]) + self.assertEqual(len(state_group_map), 1) + state_map = list(state_group_map.values())[0] + self.assertDictEqual( + state_map, + { + (EventTypes.Create, ''): e1.event_id, + (EventTypes.Name, ''): e2.event_id, + }, + ) + + @defer.inlineCallbacks + def test_get_state_groups(self): + e1 = yield self.inject_state_event( + self.room, self.u_alice, EventTypes.Create, '', {} + ) + e2 = yield self.inject_state_event( + self.room, self.u_alice, EventTypes.Name, '', {"name": "test room"} + ) + + state_group_map = yield self.store.get_state_groups( + self.room, [e2.event_id]) + self.assertEqual(len(state_group_map), 1) + state_list = list(state_group_map.values())[0] + + self.assertEqual( + {ev.event_id for ev in state_list}, + {e1.event_id, e2.event_id}, + ) + @defer.inlineCallbacks def test_get_state_for_event(self): -- cgit 1.5.1 From 7258d081a5523040dddf5a99ff745acf4fe238bf Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Tue, 2 Oct 2018 15:47:57 +0100 Subject: Fix bug when invalidating destination retry timings --- synapse/storage/transactions.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'synapse/storage') diff --git a/synapse/storage/transactions.py b/synapse/storage/transactions.py index ab54977a75..a3032cdce9 100644 --- a/synapse/storage/transactions.py +++ b/synapse/storage/transactions.py @@ -216,7 +216,7 @@ class TransactionStore(SQLBaseStore): retry_interval (int) - how long until next retry in ms """ - self._destination_retry_cache.pop(destination) + self._destination_retry_cache.pop(destination, None) return self.runInteraction( "set_destination_retry_timings", self._set_destination_retry_timings, -- cgit 1.5.1 From ae61ade8919085ad0c90e0b54c4e8338998ee64a Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Tue, 2 Oct 2018 23:33:29 +0100 Subject: Fix bug in forward_extremity update logic An event does not stop being a forward_extremity just because an outlier or rejected event refers to it. --- changelog.d/3995.bug | 1 + synapse/storage/events.py | 111 ++++++++++++++++++++++++++++++++-------------- 2 files changed, 79 insertions(+), 33 deletions(-) create mode 100644 changelog.d/3995.bug (limited to 'synapse/storage') diff --git a/changelog.d/3995.bug b/changelog.d/3995.bug new file mode 100644 index 0000000000..2adc36756b --- /dev/null +++ b/changelog.d/3995.bug @@ -0,0 +1 @@ +Fix bug in event persistence logic which caused 'NoneType is not iterable' \ No newline at end of file diff --git a/synapse/storage/events.py b/synapse/storage/events.py index e7487311ce..046174edb3 100644 --- a/synapse/storage/events.py +++ b/synapse/storage/events.py @@ -38,6 +38,7 @@ 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 +387,10 @@ 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 +399,12 @@ 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 +522,88 @@ 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 - ) + if not result: + logger.warn( + "Forward extremity list A+B-C-D is now empty in %s. " + "Old extremities (A): %s, new events (B): %s, " + "existing events which are reffed by new events (C): %s, " + "new events which are reffed by existing events (D): %s", + room_id, latest_event_ids, new_events, + [e_id for event in new_events for e_id, _ in event.prev_events], + existing_prevs, + ) + defer.returnValue(result) - defer.returnValue(new_latest_event_ids) + @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) 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 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(results) @defer.inlineCallbacks def _get_new_state_after_events(self, room_id, events_context, old_latest_event_ids, @@ -586,10 +635,6 @@ 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 = {} -- cgit 1.5.1 From 3e39783d5d7ca5da7b3619d0328f6aeec48854de Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Tue, 2 Oct 2018 23:44:14 +0100 Subject: remove debugging --- synapse/storage/events.py | 10 ---------- 1 file changed, 10 deletions(-) (limited to 'synapse/storage') diff --git a/synapse/storage/events.py b/synapse/storage/events.py index 046174edb3..8822dc7bcb 100644 --- a/synapse/storage/events.py +++ b/synapse/storage/events.py @@ -555,16 +555,6 @@ class EventsStore(EventFederationStore, EventsWorkerStore, BackgroundUpdateStore existing_prevs = yield self._get_events_which_are_prevs(result) result.difference_update(existing_prevs) - if not result: - logger.warn( - "Forward extremity list A+B-C-D is now empty in %s. " - "Old extremities (A): %s, new events (B): %s, " - "existing events which are reffed by new events (C): %s, " - "new events which are reffed by existing events (D): %s", - room_id, latest_event_ids, new_events, - [e_id for event in new_events for e_id, _ in event.prev_events], - existing_prevs, - ) defer.returnValue(result) @defer.inlineCallbacks -- cgit 1.5.1 From 9693625e556df1af66ba376d49411064c2d0f47e Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Wed, 3 Oct 2018 10:19:41 +0100 Subject: actually exclude outliers --- synapse/storage/events.py | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'synapse/storage') diff --git a/synapse/storage/events.py b/synapse/storage/events.py index 8822dc7bcb..03cedf3a75 100644 --- a/synapse/storage/events.py +++ b/synapse/storage/events.py @@ -560,7 +560,7 @@ class EventsStore(EventFederationStore, EventsWorkerStore, BackgroundUpdateStore @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) events. + existing (non-outlier/rejected) events. Args: event_ids (Iterable[str]): event ids to filter @@ -578,6 +578,7 @@ class EventsStore(EventFederationStore, EventsWorkerStore, BackgroundUpdateStore 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), -- cgit 1.5.1