From 042eedfa2b56caef0d4873583e768ca3664a881f Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 21 Feb 2018 20:49:55 +0000 Subject: Add hacky cache factor override system --- synapse/util/caches/__init__.py | 10 ++++++++++ synapse/util/caches/descriptors.py | 4 ++-- 2 files changed, 12 insertions(+), 2 deletions(-) (limited to 'synapse/util') diff --git a/synapse/util/caches/__init__.py b/synapse/util/caches/__init__.py index 183faf75a1..900575eb3c 100644 --- a/synapse/util/caches/__init__.py +++ b/synapse/util/caches/__init__.py @@ -22,6 +22,16 @@ import six CACHE_SIZE_FACTOR = float(os.environ.get("SYNAPSE_CACHE_FACTOR", 0.5)) + +def get_cache_factor_for(cache_name): + env_var = "SYNAPSE_CACHE_FACTOR_" + cache_name.upper() + factor = os.environ.get(env_var) + if factor: + return float(factor) + + return CACHE_SIZE_FACTOR + + caches_by_name = {} collectors_by_name = {} diff --git a/synapse/util/caches/descriptors.py b/synapse/util/caches/descriptors.py index fc1874b65b..65a1042de1 100644 --- a/synapse/util/caches/descriptors.py +++ b/synapse/util/caches/descriptors.py @@ -17,7 +17,7 @@ import logging from synapse.util.async import ObservableDeferred from synapse.util import unwrapFirstError, logcontext -from synapse.util.caches import CACHE_SIZE_FACTOR +from synapse.util.caches import get_cache_factor_for from synapse.util.caches.lrucache import LruCache from synapse.util.caches.treecache import TreeCache, iterate_tree_cache_entry from synapse.util.stringutils import to_ascii @@ -313,7 +313,7 @@ class CacheDescriptor(_CacheDescriptorBase): orig, num_args=num_args, inlineCallbacks=inlineCallbacks, cache_context=cache_context) - max_entries = int(max_entries * CACHE_SIZE_FACTOR) + max_entries = int(max_entries * get_cache_factor_for(orig.__name__)) self.max_entries = max_entries self.tree = tree -- cgit 1.5.1 From f7869f8f8b5a9817fde250c8c9dd44869e50a796 Mon Sep 17 00:00:00 2001 From: Amber Brown Date: Wed, 6 Jun 2018 00:13:57 +1000 Subject: Port to sortedcontainers (with tests!) (#3332) --- .gitignore | 1 + setup.cfg | 3 +- synapse/federation/send_queue.py | 14 +- synapse/python_dependencies.py | 3 +- synapse/util/caches/stream_change_cache.py | 57 +++++---- tests/util/test_stream_change_cache.py | 198 +++++++++++++++++++++++++++++ 6 files changed, 241 insertions(+), 35 deletions(-) create mode 100644 tests/util/test_stream_change_cache.py (limited to 'synapse/util') diff --git a/.gitignore b/.gitignore index 7940158c5b..9f42a7568f 100644 --- a/.gitignore +++ b/.gitignore @@ -43,6 +43,7 @@ media_store/ build/ venv/ +venv*/ localhost-800*/ static/client/register/register_config.js diff --git a/setup.cfg b/setup.cfg index da8eafbb39..fa6f2d1ce4 100644 --- a/setup.cfg +++ b/setup.cfg @@ -17,4 +17,5 @@ ignore = [flake8] max-line-length = 90 # W503 requires that binary operators be at the end, not start, of lines. Erik doesn't like it. -ignore = W503 +# E203 is contrary to PEP8. +ignore = W503,E203 diff --git a/synapse/federation/send_queue.py b/synapse/federation/send_queue.py index 9f1142b5a9..0f4aea95a3 100644 --- a/synapse/federation/send_queue.py +++ b/synapse/federation/send_queue.py @@ -35,7 +35,7 @@ from synapse.storage.presence import UserPresenceState from synapse.util.metrics import Measure from synapse.metrics import LaterGauge -from blist import sorteddict +from sortedcontainers import SortedDict from collections import namedtuple import logging @@ -55,19 +55,19 @@ class FederationRemoteSendQueue(object): self.is_mine_id = hs.is_mine_id self.presence_map = {} # Pending presence map user_id -> UserPresenceState - self.presence_changed = sorteddict() # Stream position -> user_id + self.presence_changed = SortedDict() # Stream position -> user_id self.keyed_edu = {} # (destination, key) -> EDU - self.keyed_edu_changed = sorteddict() # stream position -> (destination, key) + self.keyed_edu_changed = SortedDict() # stream position -> (destination, key) - self.edus = sorteddict() # stream position -> Edu + self.edus = SortedDict() # stream position -> Edu - self.failures = sorteddict() # stream position -> (destination, Failure) + self.failures = SortedDict() # stream position -> (destination, Failure) - self.device_messages = sorteddict() # stream position -> destination + self.device_messages = SortedDict() # stream position -> destination self.pos = 1 - self.pos_time = sorteddict() + self.pos_time = SortedDict() # EVERYTHING IS SAD. In particular, python only makes new scopes when # we make a new function, so we need to make a new function so the inner diff --git a/synapse/python_dependencies.py b/synapse/python_dependencies.py index 478c497722..001c798fe3 100644 --- a/synapse/python_dependencies.py +++ b/synapse/python_dependencies.py @@ -50,7 +50,7 @@ REQUIREMENTS = { "bcrypt": ["bcrypt>=3.1.0"], "pillow": ["PIL"], "pydenticon": ["pydenticon"], - "blist": ["blist"], + "sortedcontainers": ["sortedcontainers"], "pysaml2>=3.0.0": ["saml2>=3.0.0"], "pymacaroons-pynacl": ["pymacaroons"], "msgpack-python>=0.3.0": ["msgpack"], @@ -58,6 +58,7 @@ REQUIREMENTS = { "six": ["six"], "prometheus_client": ["prometheus_client"], } + CONDITIONAL_REQUIREMENTS = { "web_client": { "matrix_angular_sdk>=0.6.8": ["syweb>=0.6.8"], diff --git a/synapse/util/caches/stream_change_cache.py b/synapse/util/caches/stream_change_cache.py index a7fe0397fa..817118e30f 100644 --- a/synapse/util/caches/stream_change_cache.py +++ b/synapse/util/caches/stream_change_cache.py @@ -13,10 +13,10 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.util.caches import register_cache, CACHE_SIZE_FACTOR +from synapse.util import caches -from blist import sorteddict +from sortedcontainers import SortedDict import logging @@ -32,16 +32,18 @@ class StreamChangeCache(object): entities that may have changed since that position. If position key is too old then the cache will simply return all given entities. """ - def __init__(self, name, current_stream_pos, max_size=10000, prefilled_cache={}): - self._max_size = int(max_size * CACHE_SIZE_FACTOR) + + def __init__(self, name, current_stream_pos, max_size=10000, prefilled_cache=None): + self._max_size = int(max_size * caches.CACHE_SIZE_FACTOR) self._entity_to_key = {} - self._cache = sorteddict() + self._cache = SortedDict() self._earliest_known_stream_pos = current_stream_pos self.name = name - self.metrics = register_cache("cache", self.name, self._cache) + self.metrics = caches.register_cache("cache", self.name, self._cache) - for entity, stream_pos in prefilled_cache.items(): - self.entity_has_changed(entity, stream_pos) + if prefilled_cache: + for entity, stream_pos in prefilled_cache.items(): + self.entity_has_changed(entity, stream_pos) def has_entity_changed(self, entity, stream_pos): """Returns True if the entity may have been updated since stream_pos @@ -65,22 +67,25 @@ class StreamChangeCache(object): return False def get_entities_changed(self, entities, stream_pos): - """Returns subset of entities that have had new things since the - given position. If the position is too old it will just return the given list. + """ + Returns subset of entities that have had new things since the given + position. Entities unknown to the cache will be returned. If the + position is too old it will just return the given list. """ assert type(stream_pos) is int if stream_pos >= self._earliest_known_stream_pos: - keys = self._cache.keys() - i = keys.bisect_right(stream_pos) + not_known_entities = set(entities) - set(self._entity_to_key) - result = set( - self._cache[k] for k in keys[i:] - ).intersection(entities) + result = ( + set(self._cache.values()[self._cache.bisect_right(stream_pos) :]) + .intersection(entities) + .union(not_known_entities) + ) self.metrics.inc_hits() else: - result = entities + result = set(entities) self.metrics.inc_misses() return result @@ -90,12 +95,13 @@ class StreamChangeCache(object): """ assert type(stream_pos) is int + if not self._cache: + # If we have no cache, nothing can have changed. + return False + if stream_pos >= self._earliest_known_stream_pos: self.metrics.inc_hits() - keys = self._cache.keys() - i = keys.bisect_right(stream_pos) - - return i < len(keys) + return self._cache.bisect_right(stream_pos) < len(self._cache) else: self.metrics.inc_misses() return True @@ -107,10 +113,7 @@ class StreamChangeCache(object): assert type(stream_pos) is int if stream_pos >= self._earliest_known_stream_pos: - keys = self._cache.keys() - i = keys.bisect_right(stream_pos) - - return [self._cache[k] for k in keys[i:]] + return self._cache.values()[self._cache.bisect_right(stream_pos) :] else: return None @@ -129,8 +132,10 @@ class StreamChangeCache(object): self._entity_to_key[entity] = stream_pos while len(self._cache) > self._max_size: - k, r = self._cache.popitem() - self._earliest_known_stream_pos = max(k, self._earliest_known_stream_pos) + k, r = self._cache.popitem(0) + self._earliest_known_stream_pos = max( + k, self._earliest_known_stream_pos, + ) self._entity_to_key.pop(r, None) def get_max_pos_of_last_change(self, entity): diff --git a/tests/util/test_stream_change_cache.py b/tests/util/test_stream_change_cache.py new file mode 100644 index 0000000000..67ece166c7 --- /dev/null +++ b/tests/util/test_stream_change_cache.py @@ -0,0 +1,198 @@ +from tests import unittest +from mock import patch + +from synapse.util.caches.stream_change_cache import StreamChangeCache + + +class StreamChangeCacheTests(unittest.TestCase): + """ + Tests for StreamChangeCache. + """ + + def test_prefilled_cache(self): + """ + Providing a prefilled cache to StreamChangeCache will result in a cache + with the prefilled-cache entered in. + """ + cache = StreamChangeCache("#test", 1, prefilled_cache={"user@foo.com": 2}) + self.assertTrue(cache.has_entity_changed("user@foo.com", 1)) + + def test_has_entity_changed(self): + """ + StreamChangeCache.entity_has_changed will mark entities as changed, and + has_entity_changed will observe the changed entities. + """ + cache = StreamChangeCache("#test", 3) + + cache.entity_has_changed("user@foo.com", 6) + cache.entity_has_changed("bar@baz.net", 7) + + # If it's been changed after that stream position, return True + self.assertTrue(cache.has_entity_changed("user@foo.com", 4)) + self.assertTrue(cache.has_entity_changed("bar@baz.net", 4)) + + # If it's been changed at that stream position, return False + self.assertFalse(cache.has_entity_changed("user@foo.com", 6)) + + # If there's no changes after that stream position, return False + self.assertFalse(cache.has_entity_changed("user@foo.com", 7)) + + # If the entity does not exist, return False. + self.assertFalse(cache.has_entity_changed("not@here.website", 7)) + + # If we request before the stream cache's earliest known position, + # return True, whether it's a known entity or not. + self.assertTrue(cache.has_entity_changed("user@foo.com", 0)) + self.assertTrue(cache.has_entity_changed("not@here.website", 0)) + + @patch("synapse.util.caches.CACHE_SIZE_FACTOR", 1.0) + def test_has_entity_changed_pops_off_start(self): + """ + StreamChangeCache.entity_has_changed will respect the max size and + purge the oldest items upon reaching that max size. + """ + cache = StreamChangeCache("#test", 1, max_size=2) + + cache.entity_has_changed("user@foo.com", 2) + cache.entity_has_changed("bar@baz.net", 3) + cache.entity_has_changed("user@elsewhere.org", 4) + + # The cache is at the max size, 2 + self.assertEqual(len(cache._cache), 2) + + # The oldest item has been popped off + self.assertTrue("user@foo.com" not in cache._entity_to_key) + + # If we update an existing entity, it keeps the two existing entities + cache.entity_has_changed("bar@baz.net", 5) + self.assertEqual( + set(["bar@baz.net", "user@elsewhere.org"]), set(cache._entity_to_key) + ) + + def test_get_all_entities_changed(self): + """ + StreamChangeCache.get_all_entities_changed will return all changed + entities since the given position. If the position is before the start + of the known stream, it returns None instead. + """ + cache = StreamChangeCache("#test", 1) + + cache.entity_has_changed("user@foo.com", 2) + cache.entity_has_changed("bar@baz.net", 3) + cache.entity_has_changed("user@elsewhere.org", 4) + + self.assertEqual( + cache.get_all_entities_changed(1), + ["user@foo.com", "bar@baz.net", "user@elsewhere.org"], + ) + self.assertEqual( + cache.get_all_entities_changed(2), ["bar@baz.net", "user@elsewhere.org"] + ) + self.assertEqual(cache.get_all_entities_changed(3), ["user@elsewhere.org"]) + self.assertEqual(cache.get_all_entities_changed(0), None) + + def test_has_any_entity_changed(self): + """ + StreamChangeCache.has_any_entity_changed will return True if any + entities have been changed since the provided stream position, and + False if they have not. If the cache has entries and the provided + stream position is before it, it will return True, otherwise False if + the cache has no entries. + """ + cache = StreamChangeCache("#test", 1) + + # With no entities, it returns False for the past, present, and future. + self.assertFalse(cache.has_any_entity_changed(0)) + self.assertFalse(cache.has_any_entity_changed(1)) + self.assertFalse(cache.has_any_entity_changed(2)) + + # We add an entity + cache.entity_has_changed("user@foo.com", 2) + + # With an entity, it returns True for the past, the stream start + # position, and False for the stream position the entity was changed + # on and ones after it. + self.assertTrue(cache.has_any_entity_changed(0)) + self.assertTrue(cache.has_any_entity_changed(1)) + self.assertFalse(cache.has_any_entity_changed(2)) + self.assertFalse(cache.has_any_entity_changed(3)) + + def test_get_entities_changed(self): + """ + StreamChangeCache.get_entities_changed will return the entities in the + given list that have changed since the provided stream ID. If the + stream position is earlier than the earliest known position, it will + return all of the entities queried for. + """ + cache = StreamChangeCache("#test", 1) + + cache.entity_has_changed("user@foo.com", 2) + cache.entity_has_changed("bar@baz.net", 3) + cache.entity_has_changed("user@elsewhere.org", 4) + + # Query all the entries, but mid-way through the stream. We should only + # get the ones after that point. + self.assertEqual( + cache.get_entities_changed( + ["user@foo.com", "bar@baz.net", "user@elsewhere.org"], stream_pos=2 + ), + set(["bar@baz.net", "user@elsewhere.org"]), + ) + + # Query all the entries mid-way through the stream, but include one + # that doesn't exist in it. We should get back the one that doesn't + # exist, too. + self.assertEqual( + cache.get_entities_changed( + [ + "user@foo.com", + "bar@baz.net", + "user@elsewhere.org", + "not@here.website", + ], + stream_pos=2, + ), + set(["bar@baz.net", "user@elsewhere.org", "not@here.website"]), + ) + + # Query all the entries, but before the first known point. We will get + # all the entries we queried for, including ones that don't exist. + self.assertEqual( + cache.get_entities_changed( + [ + "user@foo.com", + "bar@baz.net", + "user@elsewhere.org", + "not@here.website", + ], + stream_pos=0, + ), + set( + [ + "user@foo.com", + "bar@baz.net", + "user@elsewhere.org", + "not@here.website", + ] + ), + ) + + def test_max_pos(self): + """ + StreamChangeCache.get_max_pos_of_last_change will return the most + recent point where the entity could have changed. If the entity is not + known, the stream start is provided instead. + """ + cache = StreamChangeCache("#test", 1) + + cache.entity_has_changed("user@foo.com", 2) + cache.entity_has_changed("bar@baz.net", 3) + cache.entity_has_changed("user@elsewhere.org", 4) + + # Known entities will return the point where they were changed. + self.assertEqual(cache.get_max_pos_of_last_change("user@foo.com"), 2) + self.assertEqual(cache.get_max_pos_of_last_change("bar@baz.net"), 3) + self.assertEqual(cache.get_max_pos_of_last_change("user@elsewhere.org"), 4) + + # Unknown entities will return the stream start position. + self.assertEqual(cache.get_max_pos_of_last_change("not@here.website"), 1) -- cgit 1.5.1 From a61738b316db70a4184d5c355696e0a039e7867f Mon Sep 17 00:00:00 2001 From: Amber Brown Date: Thu, 14 Jun 2018 18:27:37 +1000 Subject: Remove run_on_reactor (#3395) --- synapse/federation/transaction_queue.py | 4 ---- synapse/handlers/auth.py | 8 ++------ synapse/handlers/federation.py | 6 +----- synapse/handlers/identity.py | 8 -------- synapse/handlers/message.py | 4 +--- synapse/handlers/register.py | 5 +---- synapse/push/pusherpool.py | 3 --- synapse/rest/client/v1/register.py | 7 ------- synapse/rest/client/v2_alpha/account.py | 7 ------- synapse/rest/client/v2_alpha/register.py | 3 --- synapse/util/async.py | 10 +--------- tests/test_distributor.py | 2 -- tests/util/caches/test_descriptors.py | 2 -- 13 files changed, 6 insertions(+), 63 deletions(-) (limited to 'synapse/util') diff --git a/synapse/federation/transaction_queue.py b/synapse/federation/transaction_queue.py index f0aeb5a0d3..bcbce7f6eb 100644 --- a/synapse/federation/transaction_queue.py +++ b/synapse/federation/transaction_queue.py @@ -21,7 +21,6 @@ from .units import Transaction, Edu from synapse.api.errors import HttpResponseException, FederationDeniedError from synapse.util import logcontext, PreserveLoggingContext -from synapse.util.async import run_on_reactor from synapse.util.retryutils import NotRetryingDestination, get_retry_limiter from synapse.util.metrics import measure_func from synapse.handlers.presence import format_user_presence_state, get_interested_remotes @@ -451,9 +450,6 @@ class TransactionQueue(object): # hence why we throw the result away. yield get_retry_limiter(destination, self.clock, self.store) - # XXX: what's this for? - yield run_on_reactor() - pending_pdus = [] while True: device_message_edus, device_stream_id, dev_list_id = ( diff --git a/synapse/handlers/auth.py b/synapse/handlers/auth.py index 912136534d..dabc744890 100644 --- a/synapse/handlers/auth.py +++ b/synapse/handlers/auth.py @@ -13,6 +13,7 @@ # 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. + from twisted.internet import defer, threads from ._base import BaseHandler @@ -23,7 +24,6 @@ from synapse.api.errors import ( ) from synapse.module_api import ModuleApi from synapse.types import UserID -from synapse.util.async import run_on_reactor from synapse.util.caches.expiringcache import ExpiringCache from synapse.util.logcontext import make_deferred_yieldable @@ -423,15 +423,11 @@ class AuthHandler(BaseHandler): def _check_msisdn(self, authdict, _): return self._check_threepid('msisdn', authdict) - @defer.inlineCallbacks def _check_dummy_auth(self, authdict, _): - yield run_on_reactor() - defer.returnValue(True) + return defer.succeed(True) @defer.inlineCallbacks def _check_threepid(self, medium, authdict): - yield run_on_reactor() - if 'threepid_creds' not in authdict: raise LoginError(400, "Missing threepid_creds", Codes.MISSING_PARAM) diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py index 495ac4c648..af94bf33bc 100644 --- a/synapse/handlers/federation.py +++ b/synapse/handlers/federation.py @@ -39,7 +39,7 @@ from synapse.events.validator import EventValidator from synapse.util import unwrapFirstError, logcontext from synapse.util.metrics import measure_func from synapse.util.logutils import log_function -from synapse.util.async import run_on_reactor, Linearizer +from synapse.util.async import Linearizer from synapse.util.frozenutils import unfreeze from synapse.crypto.event_signing import ( compute_event_signature, add_hashes_and_signatures, @@ -1381,8 +1381,6 @@ class FederationHandler(BaseHandler): def get_state_for_pdu(self, room_id, event_id): """Returns the state at the event. i.e. not including said event. """ - yield run_on_reactor() - state_groups = yield self.store.get_state_groups( room_id, [event_id] ) @@ -1425,8 +1423,6 @@ class FederationHandler(BaseHandler): def get_state_ids_for_pdu(self, room_id, event_id): """Returns the state at the event. i.e. not including said event. """ - yield run_on_reactor() - state_groups = yield self.store.get_state_groups_ids( room_id, [event_id] ) diff --git a/synapse/handlers/identity.py b/synapse/handlers/identity.py index 529400955d..f00dfe1d3e 100644 --- a/synapse/handlers/identity.py +++ b/synapse/handlers/identity.py @@ -27,7 +27,6 @@ from synapse.api.errors import ( MatrixCodeMessageException, CodeMessageException ) from ._base import BaseHandler -from synapse.util.async import run_on_reactor from synapse.api.errors import SynapseError, Codes logger = logging.getLogger(__name__) @@ -62,8 +61,6 @@ class IdentityHandler(BaseHandler): @defer.inlineCallbacks def threepid_from_creds(self, creds): - yield run_on_reactor() - if 'id_server' in creds: id_server = creds['id_server'] elif 'idServer' in creds: @@ -106,7 +103,6 @@ class IdentityHandler(BaseHandler): @defer.inlineCallbacks def bind_threepid(self, creds, mxid): - yield run_on_reactor() logger.debug("binding threepid %r to %s", creds, mxid) data = None @@ -188,8 +184,6 @@ class IdentityHandler(BaseHandler): @defer.inlineCallbacks def requestEmailToken(self, id_server, email, client_secret, send_attempt, **kwargs): - yield run_on_reactor() - if not self._should_trust_id_server(id_server): raise SynapseError( 400, "Untrusted ID server '%s'" % id_server, @@ -224,8 +218,6 @@ class IdentityHandler(BaseHandler): self, id_server, country, phone_number, client_secret, send_attempt, **kwargs ): - yield run_on_reactor() - if not self._should_trust_id_server(id_server): raise SynapseError( 400, "Untrusted ID server '%s'" % id_server, diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py index 1cb81b6cf8..18dcc6d196 100644 --- a/synapse/handlers/message.py +++ b/synapse/handlers/message.py @@ -36,7 +36,7 @@ from synapse.events.validator import EventValidator from synapse.types import ( UserID, RoomAlias, RoomStreamToken, ) -from synapse.util.async import run_on_reactor, ReadWriteLock, Limiter +from synapse.util.async import ReadWriteLock, Limiter from synapse.util.logcontext import run_in_background from synapse.util.metrics import measure_func from synapse.util.frozenutils import frozendict_json_encoder @@ -959,9 +959,7 @@ class EventCreationHandler(object): event_stream_id, max_stream_id ) - @defer.inlineCallbacks def _notify(): - yield run_on_reactor() try: self.notifier.on_new_room_event( event, event_stream_id, max_stream_id, diff --git a/synapse/handlers/register.py b/synapse/handlers/register.py index 7e52adda3c..e76ef5426d 100644 --- a/synapse/handlers/register.py +++ b/synapse/handlers/register.py @@ -24,7 +24,7 @@ from synapse.api.errors import ( from synapse.http.client import CaptchaServerHttpClient from synapse import types from synapse.types import UserID, create_requester, RoomID, RoomAlias -from synapse.util.async import run_on_reactor, Linearizer +from synapse.util.async import Linearizer from synapse.util.threepids import check_3pid_allowed from ._base import BaseHandler @@ -139,7 +139,6 @@ class RegistrationHandler(BaseHandler): Raises: RegistrationError if there was a problem registering. """ - yield run_on_reactor() password_hash = None if password: password_hash = yield self.auth_handler().hash(password) @@ -431,8 +430,6 @@ class RegistrationHandler(BaseHandler): Raises: RegistrationError if there was a problem registering. """ - yield run_on_reactor() - if localpart is None: raise SynapseError(400, "Request must include user id") diff --git a/synapse/push/pusherpool.py b/synapse/push/pusherpool.py index 750d11ca38..36bb5bbc65 100644 --- a/synapse/push/pusherpool.py +++ b/synapse/push/pusherpool.py @@ -19,7 +19,6 @@ import logging from twisted.internet import defer from synapse.push.pusher import PusherFactory -from synapse.util.async import run_on_reactor from synapse.util.logcontext import make_deferred_yieldable, run_in_background logger = logging.getLogger(__name__) @@ -125,7 +124,6 @@ class PusherPool: @defer.inlineCallbacks def on_new_notifications(self, min_stream_id, max_stream_id): - yield run_on_reactor() try: users_affected = yield self.store.get_push_action_users_in_range( min_stream_id, max_stream_id @@ -151,7 +149,6 @@ class PusherPool: @defer.inlineCallbacks def on_new_receipts(self, min_stream_id, max_stream_id, affected_room_ids): - yield run_on_reactor() try: # Need to subtract 1 from the minimum because the lower bound here # is not inclusive diff --git a/synapse/rest/client/v1/register.py b/synapse/rest/client/v1/register.py index 9b3022e0b0..c10320dedf 100644 --- a/synapse/rest/client/v1/register.py +++ b/synapse/rest/client/v1/register.py @@ -24,8 +24,6 @@ import synapse.util.stringutils as stringutils from synapse.http.servlet import parse_json_object_from_request from synapse.types import create_requester -from synapse.util.async import run_on_reactor - from hashlib import sha1 import hmac import logging @@ -272,7 +270,6 @@ class RegisterRestServlet(ClientV1RestServlet): @defer.inlineCallbacks def _do_password(self, request, register_json, session): - yield run_on_reactor() if (self.hs.config.enable_registration_captcha and not session[LoginType.RECAPTCHA]): # captcha should've been done by this stage! @@ -333,8 +330,6 @@ class RegisterRestServlet(ClientV1RestServlet): @defer.inlineCallbacks def _do_shared_secret(self, request, register_json, session): - yield run_on_reactor() - if not isinstance(register_json.get("mac", None), string_types): raise SynapseError(400, "Expected mac.") if not isinstance(register_json.get("user", None), string_types): @@ -423,8 +418,6 @@ class CreateUserRestServlet(ClientV1RestServlet): @defer.inlineCallbacks def _do_create(self, requester, user_json): - yield run_on_reactor() - if "localpart" not in user_json: raise SynapseError(400, "Expected 'localpart' key.") diff --git a/synapse/rest/client/v2_alpha/account.py b/synapse/rest/client/v2_alpha/account.py index 0291fba9e7..e1281cfbb6 100644 --- a/synapse/rest/client/v2_alpha/account.py +++ b/synapse/rest/client/v2_alpha/account.py @@ -24,7 +24,6 @@ from synapse.http.servlet import ( RestServlet, assert_params_in_request, parse_json_object_from_request, ) -from synapse.util.async import run_on_reactor from synapse.util.msisdn import phone_number_to_msisdn from synapse.util.threepids import check_3pid_allowed from ._base import client_v2_patterns, interactive_auth_handler @@ -300,8 +299,6 @@ class ThreepidRestServlet(RestServlet): @defer.inlineCallbacks def on_GET(self, request): - yield run_on_reactor() - requester = yield self.auth.get_user_by_req(request) threepids = yield self.datastore.user_get_threepids( @@ -312,8 +309,6 @@ class ThreepidRestServlet(RestServlet): @defer.inlineCallbacks def on_POST(self, request): - yield run_on_reactor() - body = parse_json_object_from_request(request) threePidCreds = body.get('threePidCreds') @@ -365,8 +360,6 @@ class ThreepidDeleteRestServlet(RestServlet): @defer.inlineCallbacks def on_POST(self, request): - yield run_on_reactor() - body = parse_json_object_from_request(request) required = ['medium', 'address'] diff --git a/synapse/rest/client/v2_alpha/register.py b/synapse/rest/client/v2_alpha/register.py index 5cab00aea9..97e7c0f7c6 100644 --- a/synapse/rest/client/v2_alpha/register.py +++ b/synapse/rest/client/v2_alpha/register.py @@ -32,7 +32,6 @@ from ._base import client_v2_patterns, interactive_auth_handler import logging import hmac from hashlib import sha1 -from synapse.util.async import run_on_reactor from synapse.util.ratelimitutils import FederationRateLimiter from six import string_types @@ -191,8 +190,6 @@ class RegisterRestServlet(RestServlet): @interactive_auth_handler @defer.inlineCallbacks def on_POST(self, request): - yield run_on_reactor() - body = parse_json_object_from_request(request) kind = "user" diff --git a/synapse/util/async.py b/synapse/util/async.py index 9dd4e6b5bc..b8e57efc54 100644 --- a/synapse/util/async.py +++ b/synapse/util/async.py @@ -13,7 +13,6 @@ # See the License for the specific language governing permissions and # limitations under the License. - from twisted.internet import defer, reactor from twisted.internet.defer import CancelledError from twisted.python import failure @@ -41,13 +40,6 @@ def sleep(seconds): defer.returnValue(res) -def run_on_reactor(): - """ This will cause the rest of the function to be invoked upon the next - iteration of the main loop - """ - return sleep(0) - - class ObservableDeferred(object): """Wraps a deferred object so that we can add observer deferreds. These observer deferreds do not affect the callback chain of the original @@ -227,7 +219,7 @@ class Linearizer(object): # the context manager, but it needs to happen while we hold the # lock, and the context manager's exit code must be synchronous, # so actually this is the only sensible place. - yield run_on_reactor() + yield sleep(0) else: logger.info("Acquired uncontended linearizer lock %r for key %r", diff --git a/tests/test_distributor.py b/tests/test_distributor.py index 010aeaee7e..c066381698 100644 --- a/tests/test_distributor.py +++ b/tests/test_distributor.py @@ -19,7 +19,6 @@ from twisted.internet import defer from mock import Mock, patch from synapse.util.distributor import Distributor -from synapse.util.async import run_on_reactor class DistributorTestCase(unittest.TestCase): @@ -95,7 +94,6 @@ class DistributorTestCase(unittest.TestCase): @defer.inlineCallbacks def observer(): - yield run_on_reactor() raise MyException("Oopsie") self.dist.observe("whail", observer) diff --git a/tests/util/caches/test_descriptors.py b/tests/util/caches/test_descriptors.py index 2516fe40f4..24754591df 100644 --- a/tests/util/caches/test_descriptors.py +++ b/tests/util/caches/test_descriptors.py @@ -18,7 +18,6 @@ import logging import mock from synapse.api.errors import SynapseError -from synapse.util import async from synapse.util import logcontext from twisted.internet import defer from synapse.util.caches import descriptors @@ -195,7 +194,6 @@ class DescriptorTestCase(unittest.TestCase): def fn(self, arg1): @defer.inlineCallbacks def inner_fn(): - yield async.run_on_reactor() raise SynapseError(400, "blah") return inner_fn() -- cgit 1.5.1 From b088aafcae71fa61433f92ce0b139802a9cbd304 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Thu, 21 Jun 2018 06:15:03 +0100 Subject: Log number of events fetched from DB When we finish processing a request, log the number of events we fetched from the database to handle it. [I'm trying to figure out which requests are responsible for large amounts of event cache churn. It may turn out to be more helpful to add counts to the prometheus per-request/block metrics, but that is an extension to this code anyway.] --- synapse/http/site.py | 5 ++++- synapse/storage/events_worker.py | 4 ++++ synapse/util/logcontext.py | 4 ++++ 3 files changed, 12 insertions(+), 1 deletion(-) (limited to 'synapse/util') diff --git a/synapse/http/site.py b/synapse/http/site.py index 2664006f8c..74a752d6cf 100644 --- a/synapse/http/site.py +++ b/synapse/http/site.py @@ -99,16 +99,18 @@ class SynapseRequest(Request): db_txn_count = context.db_txn_count db_txn_duration_sec = context.db_txn_duration_sec db_sched_duration_sec = context.db_sched_duration_sec + evt_db_fetch_count = context.evt_db_fetch_count except Exception: ru_utime, ru_stime = (0, 0) db_txn_count, db_txn_duration_sec = (0, 0) + evt_db_fetch_count = 0 end_time = time.time() self.site.access_logger.info( "%s - %s - {%s}" " Processed request: %.3fsec (%.3fsec, %.3fsec) (%.3fsec/%.3fsec/%d)" - " %sB %s \"%s %s %s\" \"%s\"", + " %sB %s \"%s %s %s\" \"%s\" [%d dbevts]", self.getClientIP(), self.site.site_tag, self.authenticated_entity, @@ -124,6 +126,7 @@ class SynapseRequest(Request): self.get_redacted_uri(), self.clientproto, self.get_user_agent(), + evt_db_fetch_count, ) try: diff --git a/synapse/storage/events_worker.py b/synapse/storage/events_worker.py index 32d9d00ffb..cf4efa9d12 100644 --- a/synapse/storage/events_worker.py +++ b/synapse/storage/events_worker.py @@ -21,6 +21,7 @@ from synapse.events.utils import prune_event from synapse.util.logcontext import ( PreserveLoggingContext, make_deferred_yieldable, run_in_background, + LoggingContext, ) from synapse.util.metrics import Measure from synapse.api.errors import SynapseError @@ -145,6 +146,9 @@ class EventsWorkerStore(SQLBaseStore): missing_events_ids = [e for e in event_ids if e not in event_entry_map] if missing_events_ids: + log_ctx = LoggingContext.current_context() + log_ctx.evt_db_fetch_count += len(missing_events_ids) + missing_events = yield self._enqueue_events( missing_events_ids, check_redacted=check_redacted, diff --git a/synapse/util/logcontext.py b/synapse/util/logcontext.py index a58c723403..2bfd4bee96 100644 --- a/synapse/util/logcontext.py +++ b/synapse/util/logcontext.py @@ -60,6 +60,7 @@ class LoggingContext(object): __slots__ = [ "previous_context", "name", "ru_stime", "ru_utime", "db_txn_count", "db_txn_duration_sec", "db_sched_duration_sec", + "evt_db_fetch_count", "usage_start", "main_thread", "alive", "request", "tag", @@ -109,6 +110,9 @@ class LoggingContext(object): # sec spent waiting for db txns to be scheduled self.db_sched_duration_sec = 0 + # number of events this thread has fetched from the db + self.evt_db_fetch_count = 0 + # If alive has the thread resource usage when the logcontext last # became active. self.usage_start = None -- cgit 1.5.1 From 77ac14b960cb8daef76062ce85fc0427749b48af Mon Sep 17 00:00:00 2001 From: Amber Brown Date: Fri, 22 Jun 2018 09:37:10 +0100 Subject: Pass around the reactor explicitly (#3385) --- synapse/handlers/auth.py | 30 +++++++++++++++++++---------- synapse/handlers/message.py | 1 + synapse/handlers/user_directory.py | 9 ++++----- synapse/http/client.py | 6 +++--- synapse/http/matrixfederationclient.py | 5 +++-- synapse/notifier.py | 3 +++ synapse/replication/http/send_event.py | 6 +++--- synapse/rest/media/v1/media_repository.py | 3 ++- synapse/rest/media/v1/media_storage.py | 7 +++++-- synapse/server.py | 19 ++++++++++++++---- synapse/storage/background_updates.py | 3 +-- synapse/storage/client_ips.py | 6 ++++-- synapse/storage/event_push_actions.py | 3 +-- synapse/storage/events_worker.py | 6 +++--- synapse/util/__init__.py | 32 +++++++++++++++++++------------ synapse/util/async.py | 25 +++++++++++------------- synapse/util/file_consumer.py | 16 +++++++++++----- synapse/util/ratelimitutils.py | 3 +-- tests/crypto/test_keyring.py | 9 +++++---- tests/rest/client/test_transactions.py | 6 +++--- tests/rest/media/v1/test_media_storage.py | 5 +++-- tests/util/test_file_consumer.py | 6 +++--- tests/util/test_linearizer.py | 7 ++++--- tests/util/test_logcontext.py | 11 ++++++----- tests/utils.py | 7 ++++++- 25 files changed, 141 insertions(+), 93 deletions(-) (limited to 'synapse/util') diff --git a/synapse/handlers/auth.py b/synapse/handlers/auth.py index dabc744890..a131b7f73f 100644 --- a/synapse/handlers/auth.py +++ b/synapse/handlers/auth.py @@ -33,6 +33,7 @@ import logging import bcrypt import pymacaroons import simplejson +import attr import synapse.util.stringutils as stringutils @@ -854,7 +855,11 @@ class AuthHandler(BaseHandler): return bcrypt.hashpw(password.encode('utf8') + self.hs.config.password_pepper, bcrypt.gensalt(self.bcrypt_rounds)) - return make_deferred_yieldable(threads.deferToThread(_do_hash)) + return make_deferred_yieldable( + threads.deferToThreadPool( + self.hs.get_reactor(), self.hs.get_reactor().getThreadPool(), _do_hash + ), + ) def validate_hash(self, password, stored_hash): """Validates that self.hash(password) == stored_hash. @@ -874,16 +879,21 @@ class AuthHandler(BaseHandler): ) if stored_hash: - return make_deferred_yieldable(threads.deferToThread(_do_validate_hash)) + return make_deferred_yieldable( + threads.deferToThreadPool( + self.hs.get_reactor(), + self.hs.get_reactor().getThreadPool(), + _do_validate_hash, + ), + ) else: return defer.succeed(False) -class MacaroonGeneartor(object): - def __init__(self, hs): - self.clock = hs.get_clock() - self.server_name = hs.config.server_name - self.macaroon_secret_key = hs.config.macaroon_secret_key +@attr.s +class MacaroonGenerator(object): + + hs = attr.ib() def generate_access_token(self, user_id, extra_caveats=None): extra_caveats = extra_caveats or [] @@ -901,7 +911,7 @@ class MacaroonGeneartor(object): def generate_short_term_login_token(self, user_id, duration_in_ms=(2 * 60 * 1000)): macaroon = self._generate_base_macaroon(user_id) macaroon.add_first_party_caveat("type = login") - now = self.clock.time_msec() + now = self.hs.get_clock().time_msec() expiry = now + duration_in_ms macaroon.add_first_party_caveat("time < %d" % (expiry,)) return macaroon.serialize() @@ -913,9 +923,9 @@ class MacaroonGeneartor(object): def _generate_base_macaroon(self, user_id): macaroon = pymacaroons.Macaroon( - location=self.server_name, + location=self.hs.config.server_name, identifier="key", - key=self.macaroon_secret_key) + key=self.hs.config.macaroon_secret_key) macaroon.add_first_party_caveat("gen = 1") macaroon.add_first_party_caveat("user_id = %s" % (user_id,)) return macaroon diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py index 18dcc6d196..7b9946ab91 100644 --- a/synapse/handlers/message.py +++ b/synapse/handlers/message.py @@ -806,6 +806,7 @@ class EventCreationHandler(object): # If we're a worker we need to hit out to the master. if self.config.worker_app: yield send_event_to_master( + self.hs.get_clock(), self.http_client, host=self.config.worker_replication_host, port=self.config.worker_replication_http_port, diff --git a/synapse/handlers/user_directory.py b/synapse/handlers/user_directory.py index a39f0f7343..7e4a114d4f 100644 --- a/synapse/handlers/user_directory.py +++ b/synapse/handlers/user_directory.py @@ -19,7 +19,6 @@ from twisted.internet import defer from synapse.api.constants import EventTypes, JoinRules, Membership from synapse.storage.roommember import ProfileInfo from synapse.util.metrics import Measure -from synapse.util.async import sleep from synapse.types import get_localpart_from_id from six import iteritems @@ -174,7 +173,7 @@ class UserDirectoryHandler(object): logger.info("Handling room %d/%d", num_processed_rooms + 1, len(room_ids)) yield self._handle_initial_room(room_id) num_processed_rooms += 1 - yield sleep(self.INITIAL_ROOM_SLEEP_MS / 1000.) + yield self.clock.sleep(self.INITIAL_ROOM_SLEEP_MS / 1000.) logger.info("Processed all rooms.") @@ -188,7 +187,7 @@ class UserDirectoryHandler(object): logger.info("Handling user %d/%d", num_processed_users + 1, len(user_ids)) yield self._handle_local_user(user_id) num_processed_users += 1 - yield sleep(self.INITIAL_USER_SLEEP_MS / 1000.) + yield self.clock.sleep(self.INITIAL_USER_SLEEP_MS / 1000.) logger.info("Processed all users") @@ -236,7 +235,7 @@ class UserDirectoryHandler(object): count = 0 for user_id in user_ids: if count % self.INITIAL_ROOM_SLEEP_COUNT == 0: - yield sleep(self.INITIAL_ROOM_SLEEP_MS / 1000.) + yield self.clock.sleep(self.INITIAL_ROOM_SLEEP_MS / 1000.) if not self.is_mine_id(user_id): count += 1 @@ -251,7 +250,7 @@ class UserDirectoryHandler(object): continue if count % self.INITIAL_ROOM_SLEEP_COUNT == 0: - yield sleep(self.INITIAL_ROOM_SLEEP_MS / 1000.) + yield self.clock.sleep(self.INITIAL_ROOM_SLEEP_MS / 1000.) count += 1 user_set = (user_id, other_user_id) diff --git a/synapse/http/client.py b/synapse/http/client.py index 8064a84c5c..46ffb41de1 100644 --- a/synapse/http/client.py +++ b/synapse/http/client.py @@ -98,8 +98,8 @@ class SimpleHttpClient(object): method, uri, *args, **kwargs ) add_timeout_to_deferred( - request_deferred, - 60, cancelled_to_request_timed_out_error, + request_deferred, 60, self.hs.get_reactor(), + cancelled_to_request_timed_out_error, ) response = yield make_deferred_yieldable(request_deferred) @@ -115,7 +115,7 @@ class SimpleHttpClient(object): "Error sending request to %s %s: %s %s", method, redact_uri(uri), type(e).__name__, e.message ) - raise e + raise @defer.inlineCallbacks def post_urlencoded_get_json(self, uri, args={}, headers=None): diff --git a/synapse/http/matrixfederationclient.py b/synapse/http/matrixfederationclient.py index 993dc06e02..4e0399e762 100644 --- a/synapse/http/matrixfederationclient.py +++ b/synapse/http/matrixfederationclient.py @@ -22,7 +22,7 @@ from twisted.web._newclient import ResponseDone from synapse.http import cancelled_to_request_timed_out_error from synapse.http.endpoint import matrix_federation_endpoint import synapse.metrics -from synapse.util.async import sleep, add_timeout_to_deferred +from synapse.util.async import add_timeout_to_deferred from synapse.util import logcontext from synapse.util.logcontext import make_deferred_yieldable import synapse.util.retryutils @@ -193,6 +193,7 @@ class MatrixFederationHttpClient(object): add_timeout_to_deferred( request_deferred, timeout / 1000. if timeout else 60, + self.hs.get_reactor(), cancelled_to_request_timed_out_error, ) response = yield make_deferred_yieldable( @@ -234,7 +235,7 @@ class MatrixFederationHttpClient(object): delay = min(delay, 2) delay *= random.uniform(0.8, 1.4) - yield sleep(delay) + yield self.clock.sleep(delay) retries_left -= 1 else: raise diff --git a/synapse/notifier.py b/synapse/notifier.py index 6dce20a284..3c0622a294 100644 --- a/synapse/notifier.py +++ b/synapse/notifier.py @@ -161,6 +161,7 @@ class Notifier(object): self.user_to_user_stream = {} self.room_to_user_streams = {} + self.hs = hs self.event_sources = hs.get_event_sources() self.store = hs.get_datastore() self.pending_new_room_events = [] @@ -340,6 +341,7 @@ class Notifier(object): add_timeout_to_deferred( listener.deferred, (end_time - now) / 1000., + self.hs.get_reactor(), ) with PreserveLoggingContext(): yield listener.deferred @@ -561,6 +563,7 @@ class Notifier(object): add_timeout_to_deferred( listener.deferred.addTimeout, (end_time - now) / 1000., + self.hs.get_reactor(), ) try: with PreserveLoggingContext(): diff --git a/synapse/replication/http/send_event.py b/synapse/replication/http/send_event.py index a9baa2c1c3..f080f96cc1 100644 --- a/synapse/replication/http/send_event.py +++ b/synapse/replication/http/send_event.py @@ -21,7 +21,6 @@ from synapse.api.errors import ( from synapse.events import FrozenEvent from synapse.events.snapshot import EventContext from synapse.http.servlet import RestServlet, parse_json_object_from_request -from synapse.util.async import sleep from synapse.util.caches.response_cache import ResponseCache from synapse.util.metrics import Measure from synapse.types import Requester, UserID @@ -33,11 +32,12 @@ logger = logging.getLogger(__name__) @defer.inlineCallbacks -def send_event_to_master(client, host, port, requester, event, context, +def send_event_to_master(clock, client, host, port, requester, event, context, ratelimit, extra_users): """Send event to be handled on the master Args: + clock (synapse.util.Clock) client (SimpleHttpClient) host (str): host of master port (int): port on master listening for HTTP replication @@ -77,7 +77,7 @@ def send_event_to_master(client, host, port, requester, event, context, # If we timed out we probably don't need to worry about backing # off too much, but lets just wait a little anyway. - yield sleep(1) + yield clock.sleep(1) except MatrixCodeMessageException as e: # We convert to SynapseError as we know that it was a SynapseError # on the master process that we should send to the client. (And diff --git a/synapse/rest/media/v1/media_repository.py b/synapse/rest/media/v1/media_repository.py index 2ac767d2dc..218ba7a083 100644 --- a/synapse/rest/media/v1/media_repository.py +++ b/synapse/rest/media/v1/media_repository.py @@ -58,6 +58,7 @@ UPDATE_RECENTLY_ACCESSED_TS = 60 * 1000 class MediaRepository(object): def __init__(self, hs): + self.hs = hs self.auth = hs.get_auth() self.client = MatrixFederationHttpClient(hs) self.clock = hs.get_clock() @@ -94,7 +95,7 @@ class MediaRepository(object): storage_providers.append(provider) self.media_storage = MediaStorage( - self.primary_base_path, self.filepaths, storage_providers, + self.hs, self.primary_base_path, self.filepaths, storage_providers, ) self.clock.looping_call( diff --git a/synapse/rest/media/v1/media_storage.py b/synapse/rest/media/v1/media_storage.py index d23fe10b07..d6b8ebbedb 100644 --- a/synapse/rest/media/v1/media_storage.py +++ b/synapse/rest/media/v1/media_storage.py @@ -37,13 +37,15 @@ class MediaStorage(object): """Responsible for storing/fetching files from local sources. Args: + hs (synapse.server.Homeserver) local_media_directory (str): Base path where we store media on disk filepaths (MediaFilePaths) storage_providers ([StorageProvider]): List of StorageProvider that are used to fetch and store files. """ - def __init__(self, local_media_directory, filepaths, storage_providers): + def __init__(self, hs, local_media_directory, filepaths, storage_providers): + self.hs = hs self.local_media_directory = local_media_directory self.filepaths = filepaths self.storage_providers = storage_providers @@ -175,7 +177,8 @@ class MediaStorage(object): res = yield provider.fetch(path, file_info) if res: with res: - consumer = BackgroundFileConsumer(open(local_path, "w")) + consumer = BackgroundFileConsumer( + open(local_path, "w"), self.hs.get_reactor()) yield res.write_to_consumer(consumer) yield consumer.wait() defer.returnValue(local_path) diff --git a/synapse/server.py b/synapse/server.py index 58dbf78437..c29c19289a 100644 --- a/synapse/server.py +++ b/synapse/server.py @@ -40,7 +40,7 @@ from synapse.federation.transport.client import TransportLayerClient from synapse.federation.transaction_queue import TransactionQueue from synapse.handlers import Handlers from synapse.handlers.appservice import ApplicationServicesHandler -from synapse.handlers.auth import AuthHandler, MacaroonGeneartor +from synapse.handlers.auth import AuthHandler, MacaroonGenerator from synapse.handlers.deactivate_account import DeactivateAccountHandler from synapse.handlers.devicemessage import DeviceMessageHandler from synapse.handlers.device import DeviceHandler @@ -165,15 +165,19 @@ class HomeServer(object): 'server_notices_sender', ] - def __init__(self, hostname, **kwargs): + def __init__(self, hostname, reactor=None, **kwargs): """ Args: hostname : The hostname for the server. """ + if not reactor: + from twisted.internet import reactor + + self._reactor = reactor self.hostname = hostname self._building = {} - self.clock = Clock() + self.clock = Clock(reactor) self.distributor = Distributor() self.ratelimiter = Ratelimiter() @@ -186,6 +190,12 @@ class HomeServer(object): self.datastore = DataStore(self.get_db_conn(), self) logger.info("Finished setting up.") + def get_reactor(self): + """ + Fetch the Twisted reactor in use by this HomeServer. + """ + return self._reactor + def get_ip_from_request(self, request): # X-Forwarded-For is handled by our custom request type. return request.getClientIP() @@ -261,7 +271,7 @@ class HomeServer(object): return AuthHandler(self) def build_macaroon_generator(self): - return MacaroonGeneartor(self) + return MacaroonGenerator(self) def build_device_handler(self): return DeviceHandler(self) @@ -328,6 +338,7 @@ class HomeServer(object): return adbapi.ConnectionPool( name, + cp_reactor=self.get_reactor(), **self.db_config.get("args", {}) ) diff --git a/synapse/storage/background_updates.py b/synapse/storage/background_updates.py index 8af325a9f5..b7e9c716c8 100644 --- a/synapse/storage/background_updates.py +++ b/synapse/storage/background_updates.py @@ -12,7 +12,6 @@ # 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 synapse.util.async from ._base import SQLBaseStore from . import engines @@ -92,7 +91,7 @@ class BackgroundUpdateStore(SQLBaseStore): logger.info("Starting background schema updates") while True: - yield synapse.util.async.sleep( + yield self.hs.get_clock().sleep( self.BACKGROUND_UPDATE_INTERVAL_MS / 1000.) try: diff --git a/synapse/storage/client_ips.py b/synapse/storage/client_ips.py index ce338514e8..968d2fed22 100644 --- a/synapse/storage/client_ips.py +++ b/synapse/storage/client_ips.py @@ -15,7 +15,7 @@ import logging -from twisted.internet import defer, reactor +from twisted.internet import defer from ._base import Cache from . import background_updates @@ -70,7 +70,9 @@ class ClientIpStore(background_updates.BackgroundUpdateStore): self._client_ip_looper = self._clock.looping_call( self._update_client_ips_batch, 5 * 1000 ) - reactor.addSystemEventTrigger("before", "shutdown", self._update_client_ips_batch) + self.hs.get_reactor().addSystemEventTrigger( + "before", "shutdown", self._update_client_ips_batch + ) def insert_client_ip(self, user_id, access_token, ip, user_agent, device_id, now=None): diff --git a/synapse/storage/event_push_actions.py b/synapse/storage/event_push_actions.py index d0350ee5fe..c4a0208ce4 100644 --- a/synapse/storage/event_push_actions.py +++ b/synapse/storage/event_push_actions.py @@ -16,7 +16,6 @@ from synapse.storage._base import SQLBaseStore, LoggingTransaction from twisted.internet import defer -from synapse.util.async import sleep from synapse.util.caches.descriptors import cachedInlineCallbacks import logging @@ -800,7 +799,7 @@ class EventPushActionsStore(EventPushActionsWorkerStore): ) if caught_up: break - yield sleep(5) + yield self.hs.get_clock().sleep(5) finally: self._doing_notif_rotation = False diff --git a/synapse/storage/events_worker.py b/synapse/storage/events_worker.py index 32d9d00ffb..38fcf7d444 100644 --- a/synapse/storage/events_worker.py +++ b/synapse/storage/events_worker.py @@ -14,7 +14,7 @@ # limitations under the License. from ._base import SQLBaseStore -from twisted.internet import defer, reactor +from twisted.internet import defer from synapse.events import FrozenEvent from synapse.events.utils import prune_event @@ -265,7 +265,7 @@ class EventsWorkerStore(SQLBaseStore): except Exception: logger.exception("Failed to callback") with PreserveLoggingContext(): - reactor.callFromThread(fire, event_list, row_dict) + self.hs.get_reactor().callFromThread(fire, event_list, row_dict) except Exception as e: logger.exception("do_fetch") @@ -278,7 +278,7 @@ class EventsWorkerStore(SQLBaseStore): if event_list: with PreserveLoggingContext(): - reactor.callFromThread(fire, event_list) + self.hs.get_reactor().callFromThread(fire, event_list) @defer.inlineCallbacks def _enqueue_events(self, events, check_redacted=True, allow_rejected=False): diff --git a/synapse/util/__init__.py b/synapse/util/__init__.py index fc11e26623..2a3df7c71d 100644 --- a/synapse/util/__init__.py +++ b/synapse/util/__init__.py @@ -13,15 +13,14 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.util.logcontext import PreserveLoggingContext - -from twisted.internet import defer, reactor, task - -import time import logging - from itertools import islice +import attr +from twisted.internet import defer, task + +from synapse.util.logcontext import PreserveLoggingContext + logger = logging.getLogger(__name__) @@ -31,16 +30,24 @@ def unwrapFirstError(failure): return failure.value.subFailure +@attr.s class Clock(object): - """A small utility that obtains current time-of-day so that time may be - mocked during unit-tests. - - TODO(paul): Also move the sleep() functionality into it """ + A Clock wraps a Twisted reactor and provides utilities on top of it. + """ + _reactor = attr.ib() + + @defer.inlineCallbacks + def sleep(self, seconds): + d = defer.Deferred() + with PreserveLoggingContext(): + self._reactor.callLater(seconds, d.callback, seconds) + res = yield d + defer.returnValue(res) def time(self): """Returns the current system time in seconds since epoch.""" - return time.time() + return self._reactor.seconds() def time_msec(self): """Returns the current system time in miliseconds since epoch.""" @@ -56,6 +63,7 @@ class Clock(object): msec(float): How long to wait between calls in milliseconds. """ call = task.LoopingCall(f) + call.clock = self._reactor call.start(msec / 1000.0, now=False) return call @@ -73,7 +81,7 @@ class Clock(object): callback(*args, **kwargs) with PreserveLoggingContext(): - return reactor.callLater(delay, wrapped_callback, *args, **kwargs) + return self._reactor.callLater(delay, wrapped_callback, *args, **kwargs) def cancel_call_later(self, timer, ignore_errs=False): try: diff --git a/synapse/util/async.py b/synapse/util/async.py index b8e57efc54..1668df4ce6 100644 --- a/synapse/util/async.py +++ b/synapse/util/async.py @@ -13,14 +13,14 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer, reactor +from twisted.internet import defer from twisted.internet.defer import CancelledError from twisted.python import failure from .logcontext import ( PreserveLoggingContext, make_deferred_yieldable, run_in_background ) -from synapse.util import logcontext, unwrapFirstError +from synapse.util import logcontext, unwrapFirstError, Clock from contextlib import contextmanager @@ -31,15 +31,6 @@ from six.moves import range logger = logging.getLogger(__name__) -@defer.inlineCallbacks -def sleep(seconds): - d = defer.Deferred() - with PreserveLoggingContext(): - reactor.callLater(seconds, d.callback, seconds) - res = yield d - defer.returnValue(res) - - class ObservableDeferred(object): """Wraps a deferred object so that we can add observer deferreds. These observer deferreds do not affect the callback chain of the original @@ -172,13 +163,18 @@ class Linearizer(object): # do some work. """ - def __init__(self, name=None): + def __init__(self, name=None, clock=None): if name is None: self.name = id(self) else: self.name = name self.key_to_defer = {} + if not clock: + from twisted.internet import reactor + clock = Clock(reactor) + self._clock = clock + @defer.inlineCallbacks def queue(self, key): # If there is already a deferred in the queue, we pull it out so that @@ -219,7 +215,7 @@ class Linearizer(object): # the context manager, but it needs to happen while we hold the # lock, and the context manager's exit code must be synchronous, # so actually this is the only sensible place. - yield sleep(0) + yield self._clock.sleep(0) else: logger.info("Acquired uncontended linearizer lock %r for key %r", @@ -396,7 +392,7 @@ class DeferredTimeoutError(Exception): """ -def add_timeout_to_deferred(deferred, timeout, on_timeout_cancel=None): +def add_timeout_to_deferred(deferred, timeout, reactor, on_timeout_cancel=None): """ Add a timeout to a deferred by scheduling it to be cancelled after timeout seconds. @@ -411,6 +407,7 @@ def add_timeout_to_deferred(deferred, timeout, on_timeout_cancel=None): Args: deferred (defer.Deferred): deferred to be timed out timeout (Number): seconds to time out after + reactor (twisted.internet.reactor): the Twisted reactor to use on_timeout_cancel (callable): A callable which is called immediately after the deferred times out, and not if this deferred is diff --git a/synapse/util/file_consumer.py b/synapse/util/file_consumer.py index 3380970e4e..c78801015b 100644 --- a/synapse/util/file_consumer.py +++ b/synapse/util/file_consumer.py @@ -13,7 +13,7 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import threads, reactor +from twisted.internet import threads from synapse.util.logcontext import make_deferred_yieldable, run_in_background @@ -27,6 +27,7 @@ class BackgroundFileConsumer(object): Args: file_obj (file): The file like object to write to. Closed when finished. + reactor (twisted.internet.reactor): the Twisted reactor to use """ # For PushProducers pause if we have this many unwritten slices @@ -34,9 +35,11 @@ class BackgroundFileConsumer(object): # And resume once the size of the queue is less than this _RESUME_ON_QUEUE_SIZE = 2 - def __init__(self, file_obj): + def __init__(self, file_obj, reactor): self._file_obj = file_obj + self._reactor = reactor + # Producer we're registered with self._producer = None @@ -71,7 +74,10 @@ class BackgroundFileConsumer(object): self._producer = producer self.streaming = streaming self._finished_deferred = run_in_background( - threads.deferToThread, self._writer + threads.deferToThreadPool, + self._reactor, + self._reactor.getThreadPool(), + self._writer, ) if not streaming: self._producer.resumeProducing() @@ -109,7 +115,7 @@ class BackgroundFileConsumer(object): # producer. if self._producer and self._paused_producer: if self._bytes_queue.qsize() <= self._RESUME_ON_QUEUE_SIZE: - reactor.callFromThread(self._resume_paused_producer) + self._reactor.callFromThread(self._resume_paused_producer) bytes = self._bytes_queue.get() @@ -121,7 +127,7 @@ class BackgroundFileConsumer(object): # If its a pull producer then we need to explicitly ask for # more stuff. if not self.streaming and self._producer: - reactor.callFromThread(self._producer.resumeProducing) + self._reactor.callFromThread(self._producer.resumeProducing) except Exception as e: self._write_exception = e raise diff --git a/synapse/util/ratelimitutils.py b/synapse/util/ratelimitutils.py index 0ab63c3d7d..c5a45cef7c 100644 --- a/synapse/util/ratelimitutils.py +++ b/synapse/util/ratelimitutils.py @@ -17,7 +17,6 @@ from twisted.internet import defer from synapse.api.errors import LimitExceededError -from synapse.util.async import sleep from synapse.util.logcontext import ( run_in_background, make_deferred_yieldable, PreserveLoggingContext, @@ -153,7 +152,7 @@ class _PerHostRatelimiter(object): "Ratelimit [%s]: sleeping req", id(request_id), ) - ret_defer = run_in_background(sleep, self.sleep_msec / 1000.0) + ret_defer = run_in_background(self.clock.sleep, self.sleep_msec / 1000.0) self.sleeping_requests.add(request_id) diff --git a/tests/crypto/test_keyring.py b/tests/crypto/test_keyring.py index 149e443022..cc1c862ba4 100644 --- a/tests/crypto/test_keyring.py +++ b/tests/crypto/test_keyring.py @@ -19,10 +19,10 @@ import signedjson.sign from mock import Mock from synapse.api.errors import SynapseError from synapse.crypto import keyring -from synapse.util import async, logcontext +from synapse.util import logcontext, Clock from synapse.util.logcontext import LoggingContext from tests import unittest, utils -from twisted.internet import defer +from twisted.internet import defer, reactor class MockPerspectiveServer(object): @@ -118,6 +118,7 @@ class KeyringTestCase(unittest.TestCase): @defer.inlineCallbacks def test_verify_json_objects_for_server_awaits_previous_requests(self): + clock = Clock(reactor) key1 = signedjson.key.generate_signing_key(1) kr = keyring.Keyring(self.hs) @@ -167,7 +168,7 @@ class KeyringTestCase(unittest.TestCase): # wait a tick for it to send the request to the perspectives server # (it first tries the datastore) - yield async.sleep(1) # XXX find out why this takes so long! + yield clock.sleep(1) # XXX find out why this takes so long! self.http_client.post_json.assert_called_once() self.assertIs(LoggingContext.current_context(), context_11) @@ -183,7 +184,7 @@ class KeyringTestCase(unittest.TestCase): res_deferreds_2 = kr.verify_json_objects_for_server( [("server10", json1)], ) - yield async.sleep(1) + yield clock.sleep(1) self.http_client.post_json.assert_not_called() res_deferreds_2[0].addBoth(self.check_context, None) diff --git a/tests/rest/client/test_transactions.py b/tests/rest/client/test_transactions.py index b5bc2fa255..6a757289db 100644 --- a/tests/rest/client/test_transactions.py +++ b/tests/rest/client/test_transactions.py @@ -1,9 +1,9 @@ from synapse.rest.client.transactions import HttpTransactionCache from synapse.rest.client.transactions import CLEANUP_PERIOD_MS -from twisted.internet import defer +from twisted.internet import defer, reactor from mock import Mock, call -from synapse.util import async +from synapse.util import Clock from synapse.util.logcontext import LoggingContext from tests import unittest from tests.utils import MockClock @@ -46,7 +46,7 @@ class HttpTransactionCacheTestCase(unittest.TestCase): def test_logcontexts_with_async_result(self): @defer.inlineCallbacks def cb(): - yield async.sleep(0) + yield Clock(reactor).sleep(0) defer.returnValue("yay") @defer.inlineCallbacks diff --git a/tests/rest/media/v1/test_media_storage.py b/tests/rest/media/v1/test_media_storage.py index eef38b6781..c5e2f5549a 100644 --- a/tests/rest/media/v1/test_media_storage.py +++ b/tests/rest/media/v1/test_media_storage.py @@ -14,7 +14,7 @@ # limitations under the License. -from twisted.internet import defer +from twisted.internet import defer, reactor from synapse.rest.media.v1._base import FileInfo from synapse.rest.media.v1.media_storage import MediaStorage @@ -38,6 +38,7 @@ class MediaStorageTests(unittest.TestCase): self.secondary_base_path = os.path.join(self.test_dir, "secondary") hs = Mock() + hs.get_reactor = Mock(return_value=reactor) hs.config.media_store_path = self.primary_base_path storage_providers = [FileStorageProviderBackend( @@ -46,7 +47,7 @@ class MediaStorageTests(unittest.TestCase): self.filepaths = MediaFilePaths(self.primary_base_path) self.media_storage = MediaStorage( - self.primary_base_path, self.filepaths, storage_providers, + hs, self.primary_base_path, self.filepaths, storage_providers, ) def tearDown(self): diff --git a/tests/util/test_file_consumer.py b/tests/util/test_file_consumer.py index d6e1082779..c2aae8f54c 100644 --- a/tests/util/test_file_consumer.py +++ b/tests/util/test_file_consumer.py @@ -30,7 +30,7 @@ class FileConsumerTests(unittest.TestCase): @defer.inlineCallbacks def test_pull_consumer(self): string_file = StringIO() - consumer = BackgroundFileConsumer(string_file) + consumer = BackgroundFileConsumer(string_file, reactor=reactor) try: producer = DummyPullProducer() @@ -54,7 +54,7 @@ class FileConsumerTests(unittest.TestCase): @defer.inlineCallbacks def test_push_consumer(self): string_file = BlockingStringWrite() - consumer = BackgroundFileConsumer(string_file) + consumer = BackgroundFileConsumer(string_file, reactor=reactor) try: producer = NonCallableMock(spec_set=[]) @@ -80,7 +80,7 @@ class FileConsumerTests(unittest.TestCase): @defer.inlineCallbacks def test_push_producer_feedback(self): string_file = BlockingStringWrite() - consumer = BackgroundFileConsumer(string_file) + consumer = BackgroundFileConsumer(string_file, reactor=reactor) try: producer = NonCallableMock(spec_set=["pauseProducing", "resumeProducing"]) diff --git a/tests/util/test_linearizer.py b/tests/util/test_linearizer.py index 4865eb4bc6..bf7e3aa885 100644 --- a/tests/util/test_linearizer.py +++ b/tests/util/test_linearizer.py @@ -12,10 +12,11 @@ # 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 synapse.util import async, logcontext + +from synapse.util import logcontext, Clock from tests import unittest -from twisted.internet import defer +from twisted.internet import defer, reactor from synapse.util.async import Linearizer from six.moves import range @@ -53,7 +54,7 @@ class LinearizerTestCase(unittest.TestCase): self.assertEqual( logcontext.LoggingContext.current_context(), lc) if sleep: - yield async.sleep(0) + yield Clock(reactor).sleep(0) self.assertEqual( logcontext.LoggingContext.current_context(), lc) diff --git a/tests/util/test_logcontext.py b/tests/util/test_logcontext.py index ad78d884e0..9cf90fcfc4 100644 --- a/tests/util/test_logcontext.py +++ b/tests/util/test_logcontext.py @@ -3,8 +3,7 @@ from twisted.internet import defer from twisted.internet import reactor from .. import unittest -from synapse.util.async import sleep -from synapse.util import logcontext +from synapse.util import logcontext, Clock from synapse.util.logcontext import LoggingContext @@ -22,18 +21,20 @@ class LoggingContextTestCase(unittest.TestCase): @defer.inlineCallbacks def test_sleep(self): + clock = Clock(reactor) + @defer.inlineCallbacks def competing_callback(): with LoggingContext() as competing_context: competing_context.request = "competing" - yield sleep(0) + yield clock.sleep(0) self._check_test_key("competing") reactor.callLater(0, competing_callback) with LoggingContext() as context_one: context_one.request = "one" - yield sleep(0) + yield clock.sleep(0) self._check_test_key("one") def _test_run_in_background(self, function): @@ -87,7 +88,7 @@ class LoggingContextTestCase(unittest.TestCase): def test_run_in_background_with_blocking_fn(self): @defer.inlineCallbacks def blocking_function(): - yield sleep(0) + yield Clock(reactor).sleep(0) return self._test_run_in_background(blocking_function) diff --git a/tests/utils.py b/tests/utils.py index 262c4a5714..189fd2711c 100644 --- a/tests/utils.py +++ b/tests/utils.py @@ -37,11 +37,15 @@ USE_POSTGRES_FOR_TESTS = False @defer.inlineCallbacks -def setup_test_homeserver(name="test", datastore=None, config=None, **kargs): +def setup_test_homeserver(name="test", datastore=None, config=None, reactor=None, + **kargs): """Setup a homeserver suitable for running tests against. Keyword arguments are passed to the Homeserver constructor. If no datastore is supplied a datastore backed by an in-memory sqlite db will be given to the HS. """ + if reactor is None: + from twisted.internet import reactor + if config is None: config = Mock() config.signing_key = [MockKey()] @@ -110,6 +114,7 @@ def setup_test_homeserver(name="test", datastore=None, config=None, **kargs): database_engine=db_engine, room_list_handler=object(), tls_server_context_factory=Mock(), + reactor=reactor, **kargs ) db_conn = hs.get_db_conn() -- cgit 1.5.1 From 0495fe00350a3610cbfd9b0097ff1a8a1c31f5f4 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Fri, 22 Jun 2018 10:42:28 +0100 Subject: Indirect evt_count updates via method call so that we can stub it for the sentinel and not have a billion failing UTs --- synapse/storage/events_worker.py | 2 +- synapse/util/logcontext.py | 11 +++++++++++ 2 files changed, 12 insertions(+), 1 deletion(-) (limited to 'synapse/util') diff --git a/synapse/storage/events_worker.py b/synapse/storage/events_worker.py index cf4efa9d12..3a634842ca 100644 --- a/synapse/storage/events_worker.py +++ b/synapse/storage/events_worker.py @@ -147,7 +147,7 @@ class EventsWorkerStore(SQLBaseStore): if missing_events_ids: log_ctx = LoggingContext.current_context() - log_ctx.evt_db_fetch_count += len(missing_events_ids) + log_ctx.record_event_fetch(len(missing_events_ids)) missing_events = yield self._enqueue_events( missing_events_ids, diff --git a/synapse/util/logcontext.py b/synapse/util/logcontext.py index 2bfd4bee96..df2b71b791 100644 --- a/synapse/util/logcontext.py +++ b/synapse/util/logcontext.py @@ -91,6 +91,9 @@ class LoggingContext(object): def add_database_scheduled(self, sched_sec): pass + def record_event_fetch(self, event_count): + pass + def __nonzero__(self): return False __bool__ = __nonzero__ # python3 @@ -247,6 +250,14 @@ class LoggingContext(object): """ self.db_sched_duration_sec += sched_sec + def record_event_fetch(self, event_count): + """Record a number of events being fetched from the db + + Args: + event_count (int): number of events being fetched + """ + self.evt_db_fetch_count += event_count + class LoggingContextFilter(logging.Filter): """Logging filter that adds values from the current logging context to each -- cgit 1.5.1 From 43e02c409d163700a293ae67015584699d557c3c Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Mon, 11 Jun 2018 23:13:06 +0100 Subject: Disable partial state group caching for wildcard lookups When _get_state_for_groups is given a wildcard filter, just do a complete lookup. Hopefully this will give us the best of both worlds by not filling up the ram if we only need one or two keys, but also making the cache still work for the federation reader usecase. --- synapse/storage/state.py | 56 +++++++++++++++++++++++++-------- synapse/util/caches/dictionary_cache.py | 25 +++++++-------- tests/util/test_dict_cache.py | 12 +++---- 3 files changed, 61 insertions(+), 32 deletions(-) (limited to 'synapse/util') diff --git a/synapse/storage/state.py b/synapse/storage/state.py index 986a20400c..cd9821c270 100644 --- a/synapse/storage/state.py +++ b/synapse/storage/state.py @@ -526,10 +526,23 @@ class StateGroupWorkerStore(SQLBaseStore): @defer.inlineCallbacks def _get_state_for_groups(self, groups, types=None): - """Given list of groups returns dict of group -> list of state events - with matching types. `types` is a list of `(type, state_key)`, where - a `state_key` of None matches all state_keys. If `types` is None then - all events are returned. + """Gets the state at each of a list of state groups, optionally + filtering by type/state_key + + Args: + groups (iterable[int]): list of state groups for which we want + to get the state. + types (None|iterable[(str, None|str)]): + indicates the state type/keys required. If None, the whole + state is fetched and returned. + + Otherwise, each entry should be a `(type, state_key)` tuple to + include in the response. A `state_key` of None is a wildcard + meaning that we require all state with that type. + + Returns: + Deferred[dict[int, dict[(type, state_key), EventBase]]] + a dictionary mapping from state group to state dictionary. """ if types: types = frozenset(types) @@ -538,7 +551,7 @@ class StateGroupWorkerStore(SQLBaseStore): if types is not None: for group in set(groups): state_dict_ids, _, got_all = self._get_some_state_from_cache( - group, types + group, types, ) results[group] = state_dict_ids @@ -559,22 +572,40 @@ class StateGroupWorkerStore(SQLBaseStore): # Okay, so we have some missing_types, lets fetch them. cache_seq_num = self._state_group_cache.sequence + # the DictionaryCache knows if it has *all* the state, but + # does not know if it has all of the keys of a particular type, + # which makes wildcard lookups expensive unless we have a complete + # cache. Hence, if we are doing a wildcard lookup, populate the + # cache fully so that we can do an efficient lookup next time. + + if types and any(k is None for (t, k) in types): + types_to_fetch = None + else: + types_to_fetch = types + group_to_state_dict = yield self._get_state_groups_from_groups( - missing_groups, types + missing_groups, types_to_fetch, ) - # Now we want to update the cache with all the things we fetched - # from the database. for group, group_state_dict in iteritems(group_to_state_dict): state_dict = results[group] - state_dict.update(group_state_dict) + # update the result, filtering by `types`. + if types: + for k, v in iteritems(group_state_dict): + (typ, _) = k + if k in types or (typ, None) in types: + state_dict[k] = v + else: + state_dict.update(group_state_dict) + + # update the cache with all the things we fetched from the + # database. self._state_group_cache.update( cache_seq_num, key=group, - value=state_dict, - full=(types is None), - known_absent=types, + value=group_state_dict, + fetched_keys=types_to_fetch, ) defer.returnValue(results) @@ -681,7 +712,6 @@ class StateGroupWorkerStore(SQLBaseStore): self._state_group_cache.sequence, key=state_group, value=dict(current_state_ids), - full=True, ) return state_group diff --git a/synapse/util/caches/dictionary_cache.py b/synapse/util/caches/dictionary_cache.py index bdc21e348f..95793d466d 100644 --- a/synapse/util/caches/dictionary_cache.py +++ b/synapse/util/caches/dictionary_cache.py @@ -107,29 +107,28 @@ class DictionaryCache(object): self.sequence += 1 self.cache.clear() - def update(self, sequence, key, value, full=False, known_absent=None): + def update(self, sequence, key, value, fetched_keys=None): """Updates the entry in the cache Args: sequence - key - value (dict): The value to update the cache with. - full (bool): Whether the given value is the full dict, or just a - partial subset there of. If not full then any existing entries - for the key will be updated. - known_absent (set): Set of keys that we know don't exist in the full - dict. + key (K) + value (dict[X,Y]): The value to update the cache with. + fetched_keys (None|set[X]): All of the dictionary keys which were + fetched from the database. + + If None, this is the complete value for key K. Otherwise, it + is used to infer a list of keys which we know don't exist in + the full dict. """ self.check_thread() if self.sequence == sequence: # Only update the cache if the caches sequence number matches the # number that the cache had before the SELECT was started (SYN-369) - if known_absent is None: - known_absent = set() - if full: - self._insert(key, value, known_absent) + if fetched_keys is None: + self._insert(key, value, set()) else: - self._update_or_insert(key, value, known_absent) + self._update_or_insert(key, value, fetched_keys) def _update_or_insert(self, key, value, known_absent): # We pop and reinsert as we need to tell the cache the size may have diff --git a/tests/util/test_dict_cache.py b/tests/util/test_dict_cache.py index bc92f85fa6..543ac5bed9 100644 --- a/tests/util/test_dict_cache.py +++ b/tests/util/test_dict_cache.py @@ -32,7 +32,7 @@ class DictCacheTestCase(unittest.TestCase): seq = self.cache.sequence test_value = {"test": "test_simple_cache_hit_full"} - self.cache.update(seq, key, test_value, full=True) + self.cache.update(seq, key, test_value) c = self.cache.get(key) self.assertEqual(test_value, c.value) @@ -44,7 +44,7 @@ class DictCacheTestCase(unittest.TestCase): test_value = { "test": "test_simple_cache_hit_partial" } - self.cache.update(seq, key, test_value, full=True) + self.cache.update(seq, key, test_value) c = self.cache.get(key, ["test"]) self.assertEqual(test_value, c.value) @@ -56,7 +56,7 @@ class DictCacheTestCase(unittest.TestCase): test_value = { "test": "test_simple_cache_miss_partial" } - self.cache.update(seq, key, test_value, full=True) + self.cache.update(seq, key, test_value) c = self.cache.get(key, ["test2"]) self.assertEqual({}, c.value) @@ -70,7 +70,7 @@ class DictCacheTestCase(unittest.TestCase): "test2": "test_simple_cache_hit_miss_partial2", "test3": "test_simple_cache_hit_miss_partial3", } - self.cache.update(seq, key, test_value, full=True) + self.cache.update(seq, key, test_value) c = self.cache.get(key, ["test2"]) self.assertEqual({"test2": "test_simple_cache_hit_miss_partial2"}, c.value) @@ -82,13 +82,13 @@ class DictCacheTestCase(unittest.TestCase): test_value_1 = { "test": "test_simple_cache_hit_miss_partial", } - self.cache.update(seq, key, test_value_1, full=False) + self.cache.update(seq, key, test_value_1, fetched_keys=set("test")) seq = self.cache.sequence test_value_2 = { "test2": "test_simple_cache_hit_miss_partial2", } - self.cache.update(seq, key, test_value_2, full=False) + self.cache.update(seq, key, test_value_2, fetched_keys=set("test2")) c = self.cache.get(key) self.assertEqual( -- cgit 1.5.1 From 07cad26d65dcf237643a43ff2bdfac73ca7a7257 Mon Sep 17 00:00:00 2001 From: Amber Brown Date: Mon, 25 Jun 2018 14:08:28 +0100 Subject: Remove all global reactor imports & pass it around explicitly (#3424) --- synapse/handlers/deactivate_account.py | 4 ++-- synapse/handlers/message.py | 4 ++-- synapse/handlers/presence.py | 4 ++-- synapse/http/endpoint.py | 16 +++++++++------- synapse/push/emailpusher.py | 4 ++-- synapse/push/httppusher.py | 6 ++++-- synapse/replication/tcp/client.py | 6 +++--- synapse/replication/tcp/resource.py | 4 ++-- synapse/util/__init__.py | 3 +++ 9 files changed, 29 insertions(+), 22 deletions(-) (limited to 'synapse/util') diff --git a/synapse/handlers/deactivate_account.py b/synapse/handlers/deactivate_account.py index 8ec5ba2012..6116842764 100644 --- a/synapse/handlers/deactivate_account.py +++ b/synapse/handlers/deactivate_account.py @@ -12,7 +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. -from twisted.internet import defer, reactor +from twisted.internet import defer from ._base import BaseHandler from synapse.types import UserID, create_requester @@ -39,7 +39,7 @@ class DeactivateAccountHandler(BaseHandler): # Start the user parter loop so it can resume parting users from rooms where # it left off (if it has work left to do). - reactor.callWhenRunning(self._start_user_parting) + hs.get_reactor().callWhenRunning(self._start_user_parting) @defer.inlineCallbacks def deactivate_account(self, user_id): diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py index 7b9946ab91..a812117dea 100644 --- a/synapse/handlers/message.py +++ b/synapse/handlers/message.py @@ -20,7 +20,7 @@ import sys from canonicaljson import encode_canonical_json import six from six import string_types, itervalues, iteritems -from twisted.internet import defer, reactor +from twisted.internet import defer from twisted.internet.defer import succeed from twisted.python.failure import Failure @@ -157,7 +157,7 @@ class MessageHandler(BaseHandler): # remove the purge from the list 24 hours after it completes def clear_purge(): del self._purges_by_id[purge_id] - reactor.callLater(24 * 3600, clear_purge) + self.hs.get_reactor().callLater(24 * 3600, clear_purge) def get_purge_status(self, purge_id): """Get the current status of an active purge diff --git a/synapse/handlers/presence.py b/synapse/handlers/presence.py index 7fe568132f..7db59fba00 100644 --- a/synapse/handlers/presence.py +++ b/synapse/handlers/presence.py @@ -22,7 +22,7 @@ The methods that define policy are: - should_notify """ -from twisted.internet import defer, reactor +from twisted.internet import defer from contextlib import contextmanager from six import itervalues, iteritems @@ -179,7 +179,7 @@ class PresenceHandler(object): # have not yet been persisted self.unpersisted_users_changes = set() - reactor.addSystemEventTrigger("before", "shutdown", self._on_shutdown) + hs.get_reactor().addSystemEventTrigger("before", "shutdown", self._on_shutdown) self.serial_to_user = {} self._next_serial = 1 diff --git a/synapse/http/endpoint.py b/synapse/http/endpoint.py index 87a482650d..928c1c7407 100644 --- a/synapse/http/endpoint.py +++ b/synapse/http/endpoint.py @@ -13,7 +13,7 @@ # See the License for the specific language governing permissions and # limitations under the License. from twisted.internet.endpoints import HostnameEndpoint, wrapClientTLS -from twisted.internet import defer, reactor +from twisted.internet import defer from twisted.internet.error import ConnectError from twisted.names import client, dns from twisted.names.error import DNSNameError, DomainError @@ -78,17 +78,18 @@ def matrix_federation_endpoint(reactor, destination, ssl_context_factory=None, else: return _WrappingEndpointFac(transport_endpoint( reactor, domain, port, **endpoint_kw_args - )) + ), reactor) class _WrappingEndpointFac(object): - def __init__(self, endpoint_fac): + def __init__(self, endpoint_fac, reactor): self.endpoint_fac = endpoint_fac + self.reactor = reactor @defer.inlineCallbacks def connect(self, protocolFactory): conn = yield self.endpoint_fac.connect(protocolFactory) - conn = _WrappedConnection(conn) + conn = _WrappedConnection(conn, self.reactor) defer.returnValue(conn) @@ -98,9 +99,10 @@ class _WrappedConnection(object): """ __slots__ = ["conn", "last_request"] - def __init__(self, conn): + def __init__(self, conn, reactor): object.__setattr__(self, "conn", conn) object.__setattr__(self, "last_request", time.time()) + self._reactor = reactor def __getattr__(self, name): return getattr(self.conn, name) @@ -131,14 +133,14 @@ class _WrappedConnection(object): # Time this connection out if we haven't send a request in the last # N minutes # TODO: Cancel the previous callLater? - reactor.callLater(3 * 60, self._time_things_out_maybe) + self._reactor.callLater(3 * 60, self._time_things_out_maybe) d = self.conn.request(request) def update_request_time(res): self.last_request = time.time() # TODO: Cancel the previous callLater? - reactor.callLater(3 * 60, self._time_things_out_maybe) + self._reactor.callLater(3 * 60, self._time_things_out_maybe) return res d.addCallback(update_request_time) diff --git a/synapse/push/emailpusher.py b/synapse/push/emailpusher.py index ba7286cb72..52d4f087ee 100644 --- a/synapse/push/emailpusher.py +++ b/synapse/push/emailpusher.py @@ -13,7 +13,7 @@ # See the License for the specific language governing permissions and # limitations under the License. -from twisted.internet import defer, reactor +from twisted.internet import defer from twisted.internet.error import AlreadyCalled, AlreadyCancelled import logging @@ -199,7 +199,7 @@ class EmailPusher(object): self.timed_call = None if soonest_due_at is not None: - self.timed_call = reactor.callLater( + self.timed_call = self.hs.get_reactor().callLater( self.seconds_until(soonest_due_at), self.on_timer ) diff --git a/synapse/push/httppusher.py b/synapse/push/httppusher.py index bf7ff74a1a..7a481b5a1e 100644 --- a/synapse/push/httppusher.py +++ b/synapse/push/httppusher.py @@ -15,7 +15,7 @@ # limitations under the License. import logging -from twisted.internet import defer, reactor +from twisted.internet import defer from twisted.internet.error import AlreadyCalled, AlreadyCancelled from . import push_rule_evaluator @@ -220,7 +220,9 @@ class HttpPusher(object): ) else: logger.info("Push failed: delaying for %ds", self.backoff_delay) - self.timed_call = reactor.callLater(self.backoff_delay, self.on_timer) + self.timed_call = self.hs.get_reactor().callLater( + self.backoff_delay, self.on_timer + ) self.backoff_delay = min(self.backoff_delay * 2, self.MAX_BACKOFF_SEC) break diff --git a/synapse/replication/tcp/client.py b/synapse/replication/tcp/client.py index 6d2513c4e2..bb852b00af 100644 --- a/synapse/replication/tcp/client.py +++ b/synapse/replication/tcp/client.py @@ -15,7 +15,7 @@ """A replication client for use by synapse workers. """ -from twisted.internet import reactor, defer +from twisted.internet import defer from twisted.internet.protocol import ReconnectingClientFactory from .commands import ( @@ -44,7 +44,7 @@ class ReplicationClientFactory(ReconnectingClientFactory): self.server_name = hs.config.server_name self._clock = hs.get_clock() # As self.clock is defined in super class - reactor.addSystemEventTrigger("before", "shutdown", self.stopTrying) + hs.get_reactor().addSystemEventTrigger("before", "shutdown", self.stopTrying) def startedConnecting(self, connector): logger.info("Connecting to replication: %r", connector.getDestination()) @@ -95,7 +95,7 @@ class ReplicationClientHandler(object): factory = ReplicationClientFactory(hs, client_name, self) host = hs.config.worker_replication_host port = hs.config.worker_replication_port - reactor.connectTCP(host, port, factory) + hs.get_reactor().connectTCP(host, port, factory) def on_rdata(self, stream_name, token, rows): """Called when we get new replication data. By default this just pokes diff --git a/synapse/replication/tcp/resource.py b/synapse/replication/tcp/resource.py index 63bd6d2652..95ad8c1b4c 100644 --- a/synapse/replication/tcp/resource.py +++ b/synapse/replication/tcp/resource.py @@ -15,7 +15,7 @@ """The server side of the replication stream. """ -from twisted.internet import defer, reactor +from twisted.internet import defer from twisted.internet.protocol import Factory from .streams import STREAMS_MAP, FederationStream @@ -109,7 +109,7 @@ class ReplicationStreamer(object): self.is_looping = False self.pending_updates = False - reactor.addSystemEventTrigger("before", "shutdown", self.on_shutdown) + hs.get_reactor().addSystemEventTrigger("before", "shutdown", self.on_shutdown) def on_shutdown(self): # close all connections on shutdown diff --git a/synapse/util/__init__.py b/synapse/util/__init__.py index 2a3df7c71d..e9886ef299 100644 --- a/synapse/util/__init__.py +++ b/synapse/util/__init__.py @@ -34,6 +34,9 @@ def unwrapFirstError(failure): class Clock(object): """ A Clock wraps a Twisted reactor and provides utilities on top of it. + + Args: + reactor: The Twisted reactor to use. """ _reactor = attr.ib() -- cgit 1.5.1 From abfe4b2957645194c59da359779a4d77dcc6f493 Mon Sep 17 00:00:00 2001 From: Amber Brown Date: Tue, 26 Jun 2018 17:25:34 +0100 Subject: try and make loading items from the cache faster --- synapse/util/caches/stream_change_cache.py | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'synapse/util') diff --git a/synapse/util/caches/stream_change_cache.py b/synapse/util/caches/stream_change_cache.py index 817118e30f..c81518c6bc 100644 --- a/synapse/util/caches/stream_change_cache.py +++ b/synapse/util/caches/stream_change_cache.py @@ -78,7 +78,8 @@ class StreamChangeCache(object): not_known_entities = set(entities) - set(self._entity_to_key) result = ( - set(self._cache.values()[self._cache.bisect_right(stream_pos) :]) + {self._cache[k] for k in in self._cache.islice( + start=self._cache.bisect_right(stream_pos))} .intersection(entities) .union(not_known_entities) ) @@ -113,7 +114,8 @@ class StreamChangeCache(object): assert type(stream_pos) is int if stream_pos >= self._earliest_known_stream_pos: - return self._cache.values()[self._cache.bisect_right(stream_pos) :] + return {self._cache[k] for k in in self._cache.islice( + start=self._cache.bisect_right(stream_pos))} else: return None -- cgit 1.5.1 From bd3d329c88a59fa1adff8354d7e7105e721e2e9e Mon Sep 17 00:00:00 2001 From: Amber Brown Date: Tue, 26 Jun 2018 17:28:12 +0100 Subject: fixes --- changelog.d/3447.misc | 0 synapse/util/caches/stream_change_cache.py | 4 ++-- 2 files changed, 2 insertions(+), 2 deletions(-) create mode 100644 changelog.d/3447.misc (limited to 'synapse/util') diff --git a/changelog.d/3447.misc b/changelog.d/3447.misc new file mode 100644 index 0000000000..e69de29bb2 diff --git a/synapse/util/caches/stream_change_cache.py b/synapse/util/caches/stream_change_cache.py index c81518c6bc..b19feb3b11 100644 --- a/synapse/util/caches/stream_change_cache.py +++ b/synapse/util/caches/stream_change_cache.py @@ -78,7 +78,7 @@ class StreamChangeCache(object): not_known_entities = set(entities) - set(self._entity_to_key) result = ( - {self._cache[k] for k in in self._cache.islice( + {self._cache[k] for k in self._cache.islice( start=self._cache.bisect_right(stream_pos))} .intersection(entities) .union(not_known_entities) @@ -114,7 +114,7 @@ class StreamChangeCache(object): assert type(stream_pos) is int if stream_pos >= self._earliest_known_stream_pos: - return {self._cache[k] for k in in self._cache.islice( + return {self._cache[k] for k in self._cache.islice( start=self._cache.bisect_right(stream_pos))} else: return None -- cgit 1.5.1 From 1202508067d392cbc71e1c9e0b4d9eca3494143a Mon Sep 17 00:00:00 2001 From: Amber Brown Date: Tue, 26 Jun 2018 17:29:01 +0100 Subject: fixes --- synapse/util/caches/stream_change_cache.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'synapse/util') diff --git a/synapse/util/caches/stream_change_cache.py b/synapse/util/caches/stream_change_cache.py index b19feb3b11..0fb8620001 100644 --- a/synapse/util/caches/stream_change_cache.py +++ b/synapse/util/caches/stream_change_cache.py @@ -114,8 +114,8 @@ class StreamChangeCache(object): assert type(stream_pos) is int if stream_pos >= self._earliest_known_stream_pos: - return {self._cache[k] for k in self._cache.islice( - start=self._cache.bisect_right(stream_pos))} + return [self._cache[k] for k in self._cache.islice( + start=self._cache.bisect_right(stream_pos))] else: return None -- cgit 1.5.1 From 8057489b260a19cdd36b2f4ac5b587565ca7aac0 Mon Sep 17 00:00:00 2001 From: Matthew Hodgson Date: Tue, 26 Jun 2018 18:09:01 +0100 Subject: Revert "Try to not use as much CPU in the StreamChangeCache" --- changelog.d/3447.misc | 0 synapse/util/caches/stream_change_cache.py | 6 ++---- 2 files changed, 2 insertions(+), 4 deletions(-) delete mode 100644 changelog.d/3447.misc (limited to 'synapse/util') diff --git a/changelog.d/3447.misc b/changelog.d/3447.misc deleted file mode 100644 index e69de29bb2..0000000000 diff --git a/synapse/util/caches/stream_change_cache.py b/synapse/util/caches/stream_change_cache.py index 0fb8620001..817118e30f 100644 --- a/synapse/util/caches/stream_change_cache.py +++ b/synapse/util/caches/stream_change_cache.py @@ -78,8 +78,7 @@ class StreamChangeCache(object): not_known_entities = set(entities) - set(self._entity_to_key) result = ( - {self._cache[k] for k in self._cache.islice( - start=self._cache.bisect_right(stream_pos))} + set(self._cache.values()[self._cache.bisect_right(stream_pos) :]) .intersection(entities) .union(not_known_entities) ) @@ -114,8 +113,7 @@ class StreamChangeCache(object): assert type(stream_pos) is int if stream_pos >= self._earliest_known_stream_pos: - return [self._cache[k] for k in self._cache.islice( - start=self._cache.bisect_right(stream_pos))] + return self._cache.values()[self._cache.bisect_right(stream_pos) :] else: return None -- cgit 1.5.1 From 72d2143ea8fe43464b1db59936acef53293b1b2a Mon Sep 17 00:00:00 2001 From: Amber Brown Date: Thu, 28 Jun 2018 11:04:18 +0100 Subject: Revert "Revert "Try to not use as much CPU in the StreamChangeCache"" (#3454) --- changelog.d/3447.misc | 0 synapse/util/caches/stream_change_cache.py | 6 ++++-- 2 files changed, 4 insertions(+), 2 deletions(-) create mode 100644 changelog.d/3447.misc (limited to 'synapse/util') diff --git a/changelog.d/3447.misc b/changelog.d/3447.misc new file mode 100644 index 0000000000..e69de29bb2 diff --git a/synapse/util/caches/stream_change_cache.py b/synapse/util/caches/stream_change_cache.py index 817118e30f..0fb8620001 100644 --- a/synapse/util/caches/stream_change_cache.py +++ b/synapse/util/caches/stream_change_cache.py @@ -78,7 +78,8 @@ class StreamChangeCache(object): not_known_entities = set(entities) - set(self._entity_to_key) result = ( - set(self._cache.values()[self._cache.bisect_right(stream_pos) :]) + {self._cache[k] for k in self._cache.islice( + start=self._cache.bisect_right(stream_pos))} .intersection(entities) .union(not_known_entities) ) @@ -113,7 +114,8 @@ class StreamChangeCache(object): assert type(stream_pos) is int if stream_pos >= self._earliest_known_stream_pos: - return self._cache.values()[self._cache.bisect_right(stream_pos) :] + return [self._cache[k] for k in self._cache.islice( + start=self._cache.bisect_right(stream_pos))] else: return None -- cgit 1.5.1 From 6350bf925e8651f2fae70a1e7eb7182e9161c34a Mon Sep 17 00:00:00 2001 From: Amber Brown Date: Thu, 28 Jun 2018 14:49:57 +0100 Subject: Attempt to be more performant on PyPy (#3462) --- changelog.d/3462.feature | 1 + synapse/api/errors.py | 3 ++- synapse/api/filtering.py | 3 ++- synapse/crypto/keyclient.py | 2 +- synapse/federation/federation_server.py | 2 +- synapse/handlers/auth.py | 5 +++-- synapse/handlers/e2e_keys.py | 5 ++--- synapse/handlers/identity.py | 2 +- synapse/handlers/message.py | 5 ++--- synapse/http/client.py | 2 +- synapse/http/matrixfederationclient.py | 3 +-- synapse/http/server.py | 5 ++--- synapse/http/servlet.py | 6 ++++-- synapse/metrics/__init__.py | 3 ++- synapse/replication/tcp/commands.py | 16 ++++++++++------ synapse/rest/client/v1/login.py | 3 ++- synapse/rest/client/v1/room.py | 2 +- synapse/rest/client/v2_alpha/sync.py | 2 +- synapse/rest/media/v0/content_repository.py | 3 ++- synapse/rest/media/v1/preview_url_resource.py | 3 ++- synapse/storage/account_data.py | 3 ++- synapse/storage/appservice.py | 2 +- synapse/storage/background_updates.py | 3 ++- synapse/storage/deviceinbox.py | 13 +++++++------ synapse/storage/devices.py | 3 ++- synapse/storage/end_to_end_keys.py | 3 +-- synapse/storage/event_push_actions.py | 3 ++- synapse/storage/events.py | 3 ++- synapse/storage/events_worker.py | 3 ++- synapse/storage/filtering.py | 3 +-- synapse/storage/group_server.py | 2 +- synapse/storage/push_rule.py | 3 ++- synapse/storage/pusher.py | 3 +-- synapse/storage/receipts.py | 3 ++- synapse/storage/room.py | 3 ++- synapse/storage/roommember.py | 2 +- synapse/storage/search.py | 2 +- synapse/storage/tags.py | 3 ++- synapse/storage/transactions.py | 3 +-- synapse/util/frozenutils.py | 2 +- 40 files changed, 79 insertions(+), 62 deletions(-) create mode 100644 changelog.d/3462.feature (limited to 'synapse/util') diff --git a/changelog.d/3462.feature b/changelog.d/3462.feature new file mode 100644 index 0000000000..305dbbeddd --- /dev/null +++ b/changelog.d/3462.feature @@ -0,0 +1 @@ +Synapse now uses the best performing JSON encoder/decoder according to your runtime (simplejson on CPython, stdlib json on PyPy). \ No newline at end of file diff --git a/synapse/api/errors.py b/synapse/api/errors.py index e6ad3768f0..227a0713b2 100644 --- a/synapse/api/errors.py +++ b/synapse/api/errors.py @@ -17,7 +17,8 @@ import logging -import simplejson as json +from canonicaljson import json + from six import iteritems from six.moves import http_client diff --git a/synapse/api/filtering.py b/synapse/api/filtering.py index dbc0e7e445..aae25e7a47 100644 --- a/synapse/api/filtering.py +++ b/synapse/api/filtering.py @@ -17,7 +17,8 @@ from synapse.storage.presence import UserPresenceState from synapse.types import UserID, RoomID from twisted.internet import defer -import simplejson as json +from canonicaljson import json + import jsonschema from jsonschema import FormatChecker diff --git a/synapse/crypto/keyclient.py b/synapse/crypto/keyclient.py index f1fd488b90..2a0eddbea1 100644 --- a/synapse/crypto/keyclient.py +++ b/synapse/crypto/keyclient.py @@ -18,7 +18,7 @@ from twisted.web.http import HTTPClient from twisted.internet.protocol import Factory from twisted.internet import defer, reactor from synapse.http.endpoint import matrix_federation_endpoint -import simplejson as json +from canonicaljson import json import logging diff --git a/synapse/federation/federation_server.py b/synapse/federation/federation_server.py index d4dd967c60..a00420a24b 100644 --- a/synapse/federation/federation_server.py +++ b/synapse/federation/federation_server.py @@ -15,7 +15,7 @@ # limitations under the License. import logging -import simplejson as json +from canonicaljson import json from twisted.internet import defer from synapse.api.errors import AuthError, FederationError, SynapseError, NotFoundError diff --git a/synapse/handlers/auth.py b/synapse/handlers/auth.py index a131b7f73f..cbef1f2770 100644 --- a/synapse/handlers/auth.py +++ b/synapse/handlers/auth.py @@ -16,6 +16,8 @@ from twisted.internet import defer, threads +from canonicaljson import json + from ._base import BaseHandler from synapse.api.constants import LoginType from synapse.api.errors import ( @@ -32,7 +34,6 @@ from twisted.web.client import PartialDownloadError import logging import bcrypt import pymacaroons -import simplejson import attr import synapse.util.stringutils as stringutils @@ -403,7 +404,7 @@ class AuthHandler(BaseHandler): except PartialDownloadError as pde: # Twisted is silly data = pde.response - resp_body = simplejson.loads(data) + resp_body = json.loads(data) if 'success' in resp_body: # Note that we do NOT check the hostname here: we explicitly diff --git a/synapse/handlers/e2e_keys.py b/synapse/handlers/e2e_keys.py index 4d7bf5defd..62b4892a4e 100644 --- a/synapse/handlers/e2e_keys.py +++ b/synapse/handlers/e2e_keys.py @@ -14,10 +14,9 @@ # See the License for the specific language governing permissions and # limitations under the License. -import simplejson as json import logging -from canonicaljson import encode_canonical_json +from canonicaljson import encode_canonical_json, json from twisted.internet import defer from six import iteritems @@ -357,7 +356,7 @@ def _exception_to_failure(e): # include ConnectionRefused and other errors # # Note that some Exceptions (notably twisted's ResponseFailed etc) don't - # give a string for e.message, which simplejson then fails to serialize. + # give a string for e.message, which json then fails to serialize. return { "status": 503, "message": str(e.message), } diff --git a/synapse/handlers/identity.py b/synapse/handlers/identity.py index f00dfe1d3e..277c2b7760 100644 --- a/synapse/handlers/identity.py +++ b/synapse/handlers/identity.py @@ -19,7 +19,7 @@ import logging -import simplejson as json +from canonicaljson import json from twisted.internet import defer diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py index 8467284758..cbadf3c88e 100644 --- a/synapse/handlers/message.py +++ b/synapse/handlers/message.py @@ -14,10 +14,9 @@ # See the License for the specific language governing permissions and # limitations under the License. import logging -import simplejson import sys -from canonicaljson import encode_canonical_json +from canonicaljson import encode_canonical_json, json import six from six import string_types, itervalues, iteritems from twisted.internet import defer @@ -797,7 +796,7 @@ class EventCreationHandler(object): # Ensure that we can round trip before trying to persist in db try: dump = frozendict_json_encoder.encode(event.content) - simplejson.loads(dump) + json.loads(dump) except Exception: logger.exception("Failed to encode content: %r", event.content) raise diff --git a/synapse/http/client.py b/synapse/http/client.py index 46ffb41de1..5bdc484c15 100644 --- a/synapse/http/client.py +++ b/synapse/http/client.py @@ -42,7 +42,7 @@ from twisted.web._newclient import ResponseDone from six import StringIO from prometheus_client import Counter -import simplejson as json +from canonicaljson import json import logging import urllib diff --git a/synapse/http/matrixfederationclient.py b/synapse/http/matrixfederationclient.py index 4e0399e762..2cb9e3e231 100644 --- a/synapse/http/matrixfederationclient.py +++ b/synapse/http/matrixfederationclient.py @@ -27,7 +27,7 @@ from synapse.util import logcontext from synapse.util.logcontext import make_deferred_yieldable import synapse.util.retryutils -from canonicaljson import encode_canonical_json +from canonicaljson import encode_canonical_json, json from synapse.api.errors import ( SynapseError, Codes, HttpResponseException, FederationDeniedError, @@ -36,7 +36,6 @@ from synapse.api.errors import ( from signedjson.sign import sign_json import cgi -import simplejson as json import logging import random import sys diff --git a/synapse/http/server.py b/synapse/http/server.py index bc09b8b2be..517aaf7b5a 100644 --- a/synapse/http/server.py +++ b/synapse/http/server.py @@ -29,7 +29,7 @@ import synapse.metrics import synapse.events from canonicaljson import ( - encode_canonical_json, encode_pretty_printed_json + encode_canonical_json, encode_pretty_printed_json, json ) from twisted.internet import defer @@ -41,7 +41,6 @@ from twisted.web.util import redirectTo import collections import logging import urllib -import simplejson logger = logging.getLogger(__name__) @@ -410,7 +409,7 @@ def respond_with_json(request, code, json_object, send_cors=False, if canonical_json or synapse.events.USE_FROZEN_DICTS: json_bytes = encode_canonical_json(json_object) else: - json_bytes = simplejson.dumps(json_object) + json_bytes = json.dumps(json_object) return respond_with_json_bytes( request, code, json_bytes, diff --git a/synapse/http/servlet.py b/synapse/http/servlet.py index ef8e62901b..ef3a01ddc7 100644 --- a/synapse/http/servlet.py +++ b/synapse/http/servlet.py @@ -18,7 +18,9 @@ from synapse.api.errors import SynapseError, Codes import logging -import simplejson + +from canonicaljson import json + logger = logging.getLogger(__name__) @@ -171,7 +173,7 @@ def parse_json_value_from_request(request, allow_empty_body=False): return None try: - content = simplejson.loads(content_bytes) + content = json.loads(content_bytes) except Exception as e: logger.warn("Unable to parse JSON: %s", e) raise SynapseError(400, "Content not JSON.", errcode=Codes.NOT_JSON) diff --git a/synapse/metrics/__init__.py b/synapse/metrics/__init__.py index 7d6e0232ed..2d2397caae 100644 --- a/synapse/metrics/__init__.py +++ b/synapse/metrics/__init__.py @@ -147,7 +147,8 @@ class GCCounts(object): yield cm -REGISTRY.register(GCCounts()) +if not running_on_pypy: + REGISTRY.register(GCCounts()) # # Twisted reactor metrics diff --git a/synapse/replication/tcp/commands.py b/synapse/replication/tcp/commands.py index 12aac3cc6b..f3908df642 100644 --- a/synapse/replication/tcp/commands.py +++ b/synapse/replication/tcp/commands.py @@ -19,13 +19,17 @@ allowed to be sent by which side. """ import logging -import simplejson +import platform +if platform.python_implementation() == "PyPy": + import json + _json_encoder = json.JSONEncoder() +else: + import simplejson as json + _json_encoder = json.JSONEncoder(namedtuple_as_object=False) logger = logging.getLogger(__name__) -_json_encoder = simplejson.JSONEncoder(namedtuple_as_object=False) - class Command(object): """The base command class. @@ -102,7 +106,7 @@ class RdataCommand(Command): return cls( stream_name, None if token == "batch" else int(token), - simplejson.loads(row_json) + json.loads(row_json) ) def to_line(self): @@ -300,7 +304,7 @@ class InvalidateCacheCommand(Command): def from_line(cls, line): cache_func, keys_json = line.split(" ", 1) - return cls(cache_func, simplejson.loads(keys_json)) + return cls(cache_func, json.loads(keys_json)) def to_line(self): return " ".join(( @@ -329,7 +333,7 @@ class UserIpCommand(Command): def from_line(cls, line): user_id, jsn = line.split(" ", 1) - access_token, ip, user_agent, device_id, last_seen = simplejson.loads(jsn) + access_token, ip, user_agent, device_id, last_seen = json.loads(jsn) return cls( user_id, access_token, ip, user_agent, device_id, last_seen diff --git a/synapse/rest/client/v1/login.py b/synapse/rest/client/v1/login.py index 34df5be4e9..88ca5184cd 100644 --- a/synapse/rest/client/v1/login.py +++ b/synapse/rest/client/v1/login.py @@ -23,7 +23,8 @@ from synapse.util.msisdn import phone_number_to_msisdn from .base import ClientV1RestServlet, client_path_patterns -import simplejson as json +from canonicaljson import json + import urllib from six.moves.urllib import parse as urlparse diff --git a/synapse/rest/client/v1/room.py b/synapse/rest/client/v1/room.py index 0b984987ed..e6ae5db79b 100644 --- a/synapse/rest/client/v1/room.py +++ b/synapse/rest/client/v1/room.py @@ -31,7 +31,7 @@ from synapse.http.servlet import ( from six.moves.urllib import parse as urlparse import logging -import simplejson as json +from canonicaljson import json logger = logging.getLogger(__name__) diff --git a/synapse/rest/client/v2_alpha/sync.py b/synapse/rest/client/v2_alpha/sync.py index a291cffbf1..d2aa47b326 100644 --- a/synapse/rest/client/v2_alpha/sync.py +++ b/synapse/rest/client/v2_alpha/sync.py @@ -33,7 +33,7 @@ from ._base import set_timeline_upper_limit import itertools import logging -import simplejson as json +from canonicaljson import json logger = logging.getLogger(__name__) diff --git a/synapse/rest/media/v0/content_repository.py b/synapse/rest/media/v0/content_repository.py index 956bd5da75..e44d4276d2 100644 --- a/synapse/rest/media/v0/content_repository.py +++ b/synapse/rest/media/v0/content_repository.py @@ -22,8 +22,9 @@ from synapse.api.errors import ( from twisted.protocols.basic import FileSender from twisted.web import server, resource +from canonicaljson import json + import base64 -import simplejson as json import logging import os import re diff --git a/synapse/rest/media/v1/preview_url_resource.py b/synapse/rest/media/v1/preview_url_resource.py index 565cef2b8d..adca490640 100644 --- a/synapse/rest/media/v1/preview_url_resource.py +++ b/synapse/rest/media/v1/preview_url_resource.py @@ -23,7 +23,8 @@ import re import shutil import sys import traceback -import simplejson as json + +from canonicaljson import json from six.moves import urllib_parse as urlparse from six import string_types diff --git a/synapse/storage/account_data.py b/synapse/storage/account_data.py index 284ec3c970..7034a61399 100644 --- a/synapse/storage/account_data.py +++ b/synapse/storage/account_data.py @@ -22,8 +22,9 @@ from synapse.storage.util.id_generators import StreamIdGenerator from synapse.util.caches.stream_change_cache import StreamChangeCache from synapse.util.caches.descriptors import cached, cachedInlineCallbacks +from canonicaljson import json + import abc -import simplejson as json import logging logger = logging.getLogger(__name__) diff --git a/synapse/storage/appservice.py b/synapse/storage/appservice.py index 12ea8a158c..4d32d0bdf6 100644 --- a/synapse/storage/appservice.py +++ b/synapse/storage/appservice.py @@ -15,8 +15,8 @@ # limitations under the License. import logging import re -import simplejson as json from twisted.internet import defer +from canonicaljson import json from synapse.appservice import AppServiceTransaction from synapse.config.appservice import load_appservices diff --git a/synapse/storage/background_updates.py b/synapse/storage/background_updates.py index b7e9c716c8..af18964510 100644 --- a/synapse/storage/background_updates.py +++ b/synapse/storage/background_updates.py @@ -18,7 +18,8 @@ from . import engines from twisted.internet import defer -import simplejson as json +from canonicaljson import json + import logging logger = logging.getLogger(__name__) diff --git a/synapse/storage/deviceinbox.py b/synapse/storage/deviceinbox.py index a879e5bfc1..38addbf9c0 100644 --- a/synapse/storage/deviceinbox.py +++ b/synapse/storage/deviceinbox.py @@ -14,7 +14,8 @@ # limitations under the License. import logging -import simplejson + +from canonicaljson import json from twisted.internet import defer @@ -85,7 +86,7 @@ class DeviceInboxStore(BackgroundUpdateStore): ) rows = [] for destination, edu in remote_messages_by_destination.items(): - edu_json = simplejson.dumps(edu) + edu_json = json.dumps(edu) rows.append((destination, stream_id, now_ms, edu_json)) txn.executemany(sql, rows) @@ -177,7 +178,7 @@ class DeviceInboxStore(BackgroundUpdateStore): " WHERE user_id = ?" ) txn.execute(sql, (user_id,)) - message_json = simplejson.dumps(messages_by_device["*"]) + message_json = json.dumps(messages_by_device["*"]) for row in txn: # Add the message for all devices for this user on this # server. @@ -199,7 +200,7 @@ class DeviceInboxStore(BackgroundUpdateStore): # Only insert into the local inbox if the device exists on # this server device = row[0] - message_json = simplejson.dumps(messages_by_device[device]) + message_json = json.dumps(messages_by_device[device]) messages_json_for_user[device] = message_json if messages_json_for_user: @@ -253,7 +254,7 @@ class DeviceInboxStore(BackgroundUpdateStore): messages = [] for row in txn: stream_pos = row[0] - messages.append(simplejson.loads(row[1])) + messages.append(json.loads(row[1])) if len(messages) < limit: stream_pos = current_stream_id return (messages, stream_pos) @@ -389,7 +390,7 @@ class DeviceInboxStore(BackgroundUpdateStore): messages = [] for row in txn: stream_pos = row[0] - messages.append(simplejson.loads(row[1])) + messages.append(json.loads(row[1])) if len(messages) < limit: stream_pos = current_stream_id return (messages, stream_pos) diff --git a/synapse/storage/devices.py b/synapse/storage/devices.py index d149d8392e..2ed9ada783 100644 --- a/synapse/storage/devices.py +++ b/synapse/storage/devices.py @@ -13,7 +13,6 @@ # See the License for the specific language governing permissions and # limitations under the License. import logging -import simplejson as json from twisted.internet import defer @@ -21,6 +20,8 @@ from synapse.api.errors import StoreError from ._base import SQLBaseStore, Cache from synapse.util.caches.descriptors import cached, cachedList, cachedInlineCallbacks +from canonicaljson import json + from six import itervalues, iteritems logger = logging.getLogger(__name__) diff --git a/synapse/storage/end_to_end_keys.py b/synapse/storage/end_to_end_keys.py index b146487943..181047c8b7 100644 --- a/synapse/storage/end_to_end_keys.py +++ b/synapse/storage/end_to_end_keys.py @@ -16,8 +16,7 @@ from twisted.internet import defer from synapse.util.caches.descriptors import cached -from canonicaljson import encode_canonical_json -import simplejson as json +from canonicaljson import encode_canonical_json, json from ._base import SQLBaseStore diff --git a/synapse/storage/event_push_actions.py b/synapse/storage/event_push_actions.py index 8cb24b7d59..05cb3f61ce 100644 --- a/synapse/storage/event_push_actions.py +++ b/synapse/storage/event_push_actions.py @@ -19,7 +19,8 @@ from twisted.internet import defer from synapse.util.caches.descriptors import cachedInlineCallbacks import logging -import simplejson as json + +from canonicaljson import json from six import iteritems diff --git a/synapse/storage/events.py b/synapse/storage/events.py index 7d0e59538a..d816d4883c 100644 --- a/synapse/storage/events.py +++ b/synapse/storage/events.py @@ -19,7 +19,8 @@ from functools import wraps import itertools import logging -import simplejson as json +from canonicaljson import json + from twisted.internet import defer from synapse.storage.events_worker import EventsWorkerStore diff --git a/synapse/storage/events_worker.py b/synapse/storage/events_worker.py index f6a6e46b43..896225aab9 100644 --- a/synapse/storage/events_worker.py +++ b/synapse/storage/events_worker.py @@ -29,7 +29,8 @@ from synapse.api.errors import SynapseError from collections import namedtuple import logging -import simplejson as json + +from canonicaljson import json # these are only included to make the type annotations work from synapse.events import EventBase # noqa: F401 diff --git a/synapse/storage/filtering.py b/synapse/storage/filtering.py index 2e2763126d..eae6027cee 100644 --- a/synapse/storage/filtering.py +++ b/synapse/storage/filtering.py @@ -19,8 +19,7 @@ from ._base import SQLBaseStore from synapse.api.errors import SynapseError, Codes from synapse.util.caches.descriptors import cachedInlineCallbacks -from canonicaljson import encode_canonical_json -import simplejson as json +from canonicaljson import encode_canonical_json, json class FilteringStore(SQLBaseStore): diff --git a/synapse/storage/group_server.py b/synapse/storage/group_server.py index da05ccb027..b77402d295 100644 --- a/synapse/storage/group_server.py +++ b/synapse/storage/group_server.py @@ -20,7 +20,7 @@ from synapse.api.errors import SynapseError from ._base import SQLBaseStore -import simplejson as json +from canonicaljson import json # The category ID for the "default" category. We don't store as null in the diff --git a/synapse/storage/push_rule.py b/synapse/storage/push_rule.py index 04a0b59a39..9e52e992b3 100644 --- a/synapse/storage/push_rule.py +++ b/synapse/storage/push_rule.py @@ -25,9 +25,10 @@ from synapse.push.baserules import list_with_base_rules from synapse.api.constants import EventTypes from twisted.internet import defer +from canonicaljson import json + import abc import logging -import simplejson as json logger = logging.getLogger(__name__) diff --git a/synapse/storage/pusher.py b/synapse/storage/pusher.py index 307660b99a..c6def861cf 100644 --- a/synapse/storage/pusher.py +++ b/synapse/storage/pusher.py @@ -17,12 +17,11 @@ from ._base import SQLBaseStore from twisted.internet import defer -from canonicaljson import encode_canonical_json +from canonicaljson import encode_canonical_json, json from synapse.util.caches.descriptors import cachedInlineCallbacks, cachedList import logging -import simplejson as json import types logger = logging.getLogger(__name__) diff --git a/synapse/storage/receipts.py b/synapse/storage/receipts.py index c93c228f6e..f230a3bab7 100644 --- a/synapse/storage/receipts.py +++ b/synapse/storage/receipts.py @@ -21,9 +21,10 @@ from synapse.util.caches.stream_change_cache import StreamChangeCache from twisted.internet import defer +from canonicaljson import json + import abc import logging -import simplejson as json logger = logging.getLogger(__name__) diff --git a/synapse/storage/room.py b/synapse/storage/room.py index ea6a189185..ca0eb187e5 100644 --- a/synapse/storage/room.py +++ b/synapse/storage/room.py @@ -20,9 +20,10 @@ from synapse.storage._base import SQLBaseStore from synapse.storage.search import SearchStore from synapse.util.caches.descriptors import cached, cachedInlineCallbacks +from canonicaljson import json + import collections import logging -import simplejson as json import re logger = logging.getLogger(__name__) diff --git a/synapse/storage/roommember.py b/synapse/storage/roommember.py index 829cc4a207..8fc9549a75 100644 --- a/synapse/storage/roommember.py +++ b/synapse/storage/roommember.py @@ -28,7 +28,7 @@ from synapse.api.constants import Membership, EventTypes from synapse.types import get_domain_from_id import logging -import simplejson as json +from canonicaljson import json from six import itervalues, iteritems diff --git a/synapse/storage/search.py b/synapse/storage/search.py index f0fa5d7631..9b77c45318 100644 --- a/synapse/storage/search.py +++ b/synapse/storage/search.py @@ -16,7 +16,7 @@ from collections import namedtuple import logging import re -import simplejson as json +from canonicaljson import json from six import string_types diff --git a/synapse/storage/tags.py b/synapse/storage/tags.py index 6671d3cfca..04d123ed95 100644 --- a/synapse/storage/tags.py +++ b/synapse/storage/tags.py @@ -19,7 +19,8 @@ from synapse.storage.account_data import AccountDataWorkerStore from synapse.util.caches.descriptors import cached from twisted.internet import defer -import simplejson as json +from canonicaljson import json + import logging from six.moves import range diff --git a/synapse/storage/transactions.py b/synapse/storage/transactions.py index e485d19b84..acbc03446e 100644 --- a/synapse/storage/transactions.py +++ b/synapse/storage/transactions.py @@ -19,12 +19,11 @@ from synapse.util.caches.descriptors import cached from twisted.internet import defer import six -from canonicaljson import encode_canonical_json +from canonicaljson import encode_canonical_json, json from collections import namedtuple import logging -import simplejson as json # py2 sqlite has buffer hardcoded as only binary type, so we must use it, # despite being deprecated and removed in favor of memoryview diff --git a/synapse/util/frozenutils.py b/synapse/util/frozenutils.py index 15f0a7ba9e..535e7d0e7a 100644 --- a/synapse/util/frozenutils.py +++ b/synapse/util/frozenutils.py @@ -14,7 +14,7 @@ # limitations under the License. from frozendict import frozendict -import simplejson as json +from canonicaljson import json from six import string_types -- cgit 1.5.1