From 17e1e807264ce13774d9b343c96406795ea24c27 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Mon, 12 Aug 2019 15:39:14 +0100 Subject: Retry well-known lookup before expiry. This gives a bit of a grace period where we can attempt to refetch a remote `well-known`, while still using the cached result if that fails. Hopefully this will make the well-known resolution a bit more torelant of failures, rather than it immediately treating failures as "no result" and caching that for an hour. --- .../federation/test_matrix_federation_agent.py | 69 ++++++++++++++++++++++ tests/util/caches/test_ttlcache.py | 4 +- 2 files changed, 71 insertions(+), 2 deletions(-) (limited to 'tests') diff --git a/tests/http/federation/test_matrix_federation_agent.py b/tests/http/federation/test_matrix_federation_agent.py index 1435baede2..2c568788b3 100644 --- a/tests/http/federation/test_matrix_federation_agent.py +++ b/tests/http/federation/test_matrix_federation_agent.py @@ -987,6 +987,75 @@ class MatrixFederationAgentTests(TestCase): r = self.successResultOf(fetch_d) self.assertEqual(r.delegated_server, b"other-server") + def test_well_known_cache_with_temp_failure(self): + """Test that we refetch well-known before the cache expires, and that + it ignores transient errors. + """ + + well_known_resolver = WellKnownResolver( + self.reactor, + Agent(self.reactor, contextFactory=self.tls_factory), + well_known_cache=self.well_known_cache, + ) + + self.reactor.lookups["testserv"] = "1.2.3.4" + + fetch_d = well_known_resolver.get_well_known(b"testserv") + + # there should be an attempt to connect on port 443 for the .well-known + clients = self.reactor.tcpClients + self.assertEqual(len(clients), 1) + (host, port, client_factory, _timeout, _bindAddress) = clients.pop(0) + self.assertEqual(host, "1.2.3.4") + self.assertEqual(port, 443) + + well_known_server = self._handle_well_known_connection( + client_factory, + expected_sni=b"testserv", + response_headers={b"Cache-Control": b"max-age=1000"}, + content=b'{ "m.server": "target-server" }', + ) + + r = self.successResultOf(fetch_d) + self.assertEqual(r.delegated_server, b"target-server") + + # close the tcp connection + well_known_server.loseConnection() + + # Get close to the cache expiry, this will cause the resolver to do + # another lookup. + self.reactor.pump((900.0,)) + + fetch_d = well_known_resolver.get_well_known(b"testserv") + clients = self.reactor.tcpClients + (host, port, client_factory, _timeout, _bindAddress) = clients.pop(0) + + # fonx the connection attempt, this will be treated as a temporary + # failure. + client_factory.clientConnectionFailed(None, Exception("nope")) + + # attemptdelay on the hostnameendpoint is 0.3, so takes that long before the + # .well-known request fails. + self.reactor.pump((0.4,)) + + # Resolver should return cached value, despite the lookup failing. + r = self.successResultOf(fetch_d) + self.assertEqual(r.delegated_server, b"target-server") + + # Expire the cache and repeat the request + self.reactor.pump((100.0,)) + + # Repated the request, this time it should fail if the lookup fails. + fetch_d = well_known_resolver.get_well_known(b"testserv") + + clients = self.reactor.tcpClients + (host, port, client_factory, _timeout, _bindAddress) = clients.pop(0) + client_factory.clientConnectionFailed(None, Exception("nope")) + self.reactor.pump((0.4,)) + + r = self.successResultOf(fetch_d) + self.assertEqual(r.delegated_server, None) + class TestCachePeriodFromHeaders(TestCase): def test_cache_control(self): diff --git a/tests/util/caches/test_ttlcache.py b/tests/util/caches/test_ttlcache.py index c94cbb662b..816795c136 100644 --- a/tests/util/caches/test_ttlcache.py +++ b/tests/util/caches/test_ttlcache.py @@ -36,7 +36,7 @@ class CacheTestCase(unittest.TestCase): self.assertTrue("one" in self.cache) self.assertEqual(self.cache.get("one"), "1") self.assertEqual(self.cache["one"], "1") - self.assertEqual(self.cache.get_with_expiry("one"), ("1", 110)) + self.assertEqual(self.cache.get_with_expiry("one"), ("1", 110, 10)) self.assertEqual(self.cache._metrics.hits, 3) self.assertEqual(self.cache._metrics.misses, 0) @@ -77,7 +77,7 @@ class CacheTestCase(unittest.TestCase): self.assertEqual(self.cache["two"], "2") self.assertEqual(self.cache["three"], "3") - self.assertEqual(self.cache.get_with_expiry("two"), ("2", 120)) + self.assertEqual(self.cache.get_with_expiry("two"), ("2", 120, 20)) self.assertEqual(self.cache._metrics.hits, 5) self.assertEqual(self.cache._metrics.misses, 0) -- cgit 1.5.1 From e6e136decca12648933f974e4151fb936ad9e1fa Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Tue, 13 Aug 2019 18:04:46 +0100 Subject: Retry well known on fail. If we have recently seen a valid well-known for a domain we want to retry on (non-final) errors a few times, to handle temporary blips in networking/etc. --- synapse/http/federation/matrix_federation_agent.py | 26 +++-- synapse/http/federation/well_known_resolver.py | 122 +++++++++++++++++---- .../federation/test_matrix_federation_agent.py | 79 +++++++------ 3 files changed, 160 insertions(+), 67 deletions(-) (limited to 'tests') diff --git a/synapse/http/federation/matrix_federation_agent.py b/synapse/http/federation/matrix_federation_agent.py index 71a15f434d..64f62aaeec 100644 --- a/synapse/http/federation/matrix_federation_agent.py +++ b/synapse/http/federation/matrix_federation_agent.py @@ -51,9 +51,9 @@ class MatrixFederationAgent(object): SRVResolver impl to use for looking up SRV records. None to use a default implementation. - _well_known_cache (TTLCache|None): - TTLCache impl for storing cached well-known lookups. None to use a default - implementation. + _well_known_resolver (WellKnownResolver|None): + WellKnownResolver to use to perform well-known lookups. None to use a + default implementation. """ def __init__( @@ -61,7 +61,7 @@ class MatrixFederationAgent(object): reactor, tls_client_options_factory, _srv_resolver=None, - _well_known_cache=None, + _well_known_resolver=None, ): self._reactor = reactor self._clock = Clock(reactor) @@ -76,15 +76,17 @@ class MatrixFederationAgent(object): self._pool.maxPersistentPerHost = 5 self._pool.cachedConnectionTimeout = 2 * 60 - self._well_known_resolver = WellKnownResolver( - self._reactor, - agent=Agent( + if _well_known_resolver is None: + _well_known_resolver = WellKnownResolver( self._reactor, - pool=self._pool, - contextFactory=tls_client_options_factory, - ), - well_known_cache=_well_known_cache, - ) + agent=Agent( + self._reactor, + pool=self._pool, + contextFactory=tls_client_options_factory, + ), + ) + + self._well_known_resolver = _well_known_resolver @defer.inlineCallbacks def request(self, method, uri, headers=None, bodyProducer=None): diff --git a/synapse/http/federation/well_known_resolver.py b/synapse/http/federation/well_known_resolver.py index bb250c6922..d59864e298 100644 --- a/synapse/http/federation/well_known_resolver.py +++ b/synapse/http/federation/well_known_resolver.py @@ -38,6 +38,13 @@ WELL_KNOWN_DEFAULT_CACHE_PERIOD_JITTER = 10 * 60 # period to cache failure to fetch .well-known for WELL_KNOWN_INVALID_CACHE_PERIOD = 1 * 3600 +# period to cache failure to fetch .well-known if there has recently been a +# valid well-known for that domain. +WELL_KNOWN_DOWN_CACHE_PERIOD = 2 * 60 + +# period to remember there was a valid well-known after valid record expires +WELL_KNOWN_REMEMBER_DOMAIN_HAD_VALID = 2 * 3600 + # cap for .well-known cache period WELL_KNOWN_MAX_CACHE_PERIOD = 48 * 3600 @@ -49,11 +56,16 @@ WELL_KNOWN_MIN_CACHE_PERIOD = 5 * 60 # we'll start trying to refetch 1 minute before it expires. WELL_KNOWN_GRACE_PERIOD_FACTOR = 0.2 +# Number of times we retry fetching a well-known for a domain we know recently +# had a valid entry. +WELL_KNOWN_RETRY_ATTEMPTS = 3 + logger = logging.getLogger(__name__) _well_known_cache = TTLCache("well-known") +_had_valid_well_known_cache = TTLCache("had-valid-well-known") @attr.s(slots=True, frozen=True) @@ -65,14 +77,20 @@ class WellKnownResolver(object): """Handles well-known lookups for matrix servers. """ - def __init__(self, reactor, agent, well_known_cache=None): + def __init__( + self, reactor, agent, well_known_cache=None, had_well_known_cache=None + ): self._reactor = reactor self._clock = Clock(reactor) if well_known_cache is None: well_known_cache = _well_known_cache + if had_well_known_cache is None: + had_well_known_cache = _had_valid_well_known_cache + self._well_known_cache = well_known_cache + self._had_valid_well_known_cache = had_well_known_cache self._well_known_agent = RedirectAgent(agent) @defer.inlineCallbacks @@ -100,7 +118,7 @@ class WellKnownResolver(object): # requests for the same server in parallel? try: with Measure(self._clock, "get_well_known"): - result, cache_period = yield self._do_get_well_known(server_name) + result, cache_period = yield self._fetch_well_known(server_name) except _FetchWellKnownFailure as e: if prev_result and e.temporary: @@ -111,10 +129,20 @@ class WellKnownResolver(object): result = None - # add some randomness to the TTL to avoid a stampeding herd every hour - # after startup - cache_period = WELL_KNOWN_INVALID_CACHE_PERIOD - cache_period += random.uniform(0, WELL_KNOWN_DEFAULT_CACHE_PERIOD_JITTER) + if self._had_valid_well_known_cache.get(server_name, False): + # We have recently seen a valid well-known record for this + # server, so we cache the lack of well-known for a shorter time. + cache_period = WELL_KNOWN_DOWN_CACHE_PERIOD + cache_period += random.uniform( + 0, WELL_KNOWN_DEFAULT_CACHE_PERIOD_JITTER + ) + else: + # add some randomness to the TTL to avoid a stampeding herd every hour + # after startup + cache_period = WELL_KNOWN_INVALID_CACHE_PERIOD + cache_period += random.uniform( + 0, WELL_KNOWN_DEFAULT_CACHE_PERIOD_JITTER + ) if cache_period > 0: self._well_known_cache.set(server_name, result, cache_period) @@ -122,7 +150,7 @@ class WellKnownResolver(object): return WellKnownLookupResult(delegated_server=result) @defer.inlineCallbacks - def _do_get_well_known(self, server_name): + def _fetch_well_known(self, server_name): """Actually fetch and parse a .well-known, without checking the cache Args: @@ -134,24 +162,17 @@ class WellKnownResolver(object): Returns: Deferred[Tuple[bytes,int]]: The lookup result and cache period. """ - uri = b"https://%s/.well-known/matrix/server" % (server_name,) - uri_str = uri.decode("ascii") - logger.info("Fetching %s", uri_str) + + had_valid_well_known = bool( + self._had_valid_well_known_cache.get(server_name, False) + ) # We do this in two steps to differentiate between possibly transient # errors (e.g. can't connect to host, 503 response) and more permenant # errors (such as getting a 404 response). - try: - response = yield make_deferred_yieldable( - self._well_known_agent.request(b"GET", uri) - ) - body = yield make_deferred_yieldable(readBody(response)) - - if 500 <= response.code < 600: - raise Exception("Non-200 response %s" % (response.code,)) - except Exception as e: - logger.info("Error fetching %s: %s", uri_str, e) - raise _FetchWellKnownFailure(temporary=True) + response, body = yield self._make_well_known_request( + server_name, retry=had_valid_well_known + ) try: if response.code != 200: @@ -161,8 +182,11 @@ class WellKnownResolver(object): logger.info("Response from .well-known: %s", parsed_body) result = parsed_body["m.server"].encode("ascii") + except defer.CancelledError: + # Bail if we've been cancelled + raise except Exception as e: - logger.info("Error fetching %s: %s", uri_str, e) + logger.info("Error parsing well-known for %s: %s", server_name, e) raise _FetchWellKnownFailure(temporary=False) cache_period = _cache_period_from_headers( @@ -177,8 +201,62 @@ class WellKnownResolver(object): cache_period = min(cache_period, WELL_KNOWN_MAX_CACHE_PERIOD) cache_period = max(cache_period, WELL_KNOWN_MIN_CACHE_PERIOD) + # We got a success, mark as such in the cache + self._had_valid_well_known_cache.set( + server_name, + bool(result), + cache_period + WELL_KNOWN_REMEMBER_DOMAIN_HAD_VALID, + ) + return (result, cache_period) + @defer.inlineCallbacks + def _make_well_known_request(self, server_name, retry): + """Make the well known request. + + This will retry the request if requested and it fails (with unable + to connect or receives a 5xx error). + + Args: + server_name (bytes) + retry (bool): Whether to retry the request if it fails. + + Returns: + Deferred[tuple[IResponse, bytes]] Returns the response object and + body. Response may be a non-200 response. + """ + uri = b"https://%s/.well-known/matrix/server" % (server_name,) + uri_str = uri.decode("ascii") + + i = 0 + while True: + i += 1 + + logger.info("Fetching %s", uri_str) + try: + response = yield make_deferred_yieldable( + self._well_known_agent.request(b"GET", uri) + ) + body = yield make_deferred_yieldable(readBody(response)) + + if 500 <= response.code < 600: + raise Exception("Non-200 response %s" % (response.code,)) + + return response, body + except defer.CancelledError: + # Bail if we've been cancelled + raise + except Exception as e: + logger.info("Retry: %s", retry) + if not retry or i >= WELL_KNOWN_RETRY_ATTEMPTS: + logger.info("Error fetching %s: %s", uri_str, e) + raise _FetchWellKnownFailure(temporary=True) + + logger.info("Error fetching %s: %s. Retrying", uri_str, e) + + # Sleep briefly in the hopes that they come back up + yield self._clock.sleep(0.5) + def _cache_period_from_headers(headers, time_now=time.time): cache_controls = _parse_cache_control(headers) diff --git a/tests/http/federation/test_matrix_federation_agent.py b/tests/http/federation/test_matrix_federation_agent.py index 2c568788b3..4d3f31d18c 100644 --- a/tests/http/federation/test_matrix_federation_agent.py +++ b/tests/http/federation/test_matrix_federation_agent.py @@ -73,8 +73,6 @@ class MatrixFederationAgentTests(TestCase): self.mock_resolver = Mock() - self.well_known_cache = TTLCache("test_cache", timer=self.reactor.seconds) - config_dict = default_config("test", parse=False) config_dict["federation_custom_ca_list"] = [get_test_ca_cert_file()] @@ -82,11 +80,21 @@ class MatrixFederationAgentTests(TestCase): config.parse_config_dict(config_dict, "", "") self.tls_factory = ClientTLSOptionsFactory(config) + + self.well_known_cache = TTLCache("test_cache", timer=self.reactor.seconds) + self.had_well_known_cache = TTLCache("test_cache", timer=self.reactor.seconds) + self.well_known_resolver = WellKnownResolver( + self.reactor, + Agent(self.reactor, contextFactory=self.tls_factory), + well_known_cache=self.well_known_cache, + had_well_known_cache=self.had_well_known_cache, + ) + self.agent = MatrixFederationAgent( reactor=self.reactor, tls_client_options_factory=self.tls_factory, _srv_resolver=self.mock_resolver, - _well_known_cache=self.well_known_cache, + _well_known_resolver=self.well_known_resolver, ) def _make_connection(self, client_factory, expected_sni): @@ -701,11 +709,18 @@ class MatrixFederationAgentTests(TestCase): config = default_config("test", parse=True) + # Build a new agent and WellKnownResolver with a different tls factory + tls_factory = ClientTLSOptionsFactory(config) agent = MatrixFederationAgent( reactor=self.reactor, - tls_client_options_factory=ClientTLSOptionsFactory(config), + tls_client_options_factory=tls_factory, _srv_resolver=self.mock_resolver, - _well_known_cache=self.well_known_cache, + _well_known_resolver=WellKnownResolver( + self.reactor, + Agent(self.reactor, contextFactory=tls_factory), + well_known_cache=self.well_known_cache, + had_well_known_cache=self.had_well_known_cache, + ), ) test_d = agent.request(b"GET", b"matrix://testserv/foo/bar") @@ -932,15 +947,9 @@ class MatrixFederationAgentTests(TestCase): self.successResultOf(test_d) def test_well_known_cache(self): - well_known_resolver = WellKnownResolver( - self.reactor, - Agent(self.reactor, contextFactory=self.tls_factory), - well_known_cache=self.well_known_cache, - ) - self.reactor.lookups["testserv"] = "1.2.3.4" - fetch_d = well_known_resolver.get_well_known(b"testserv") + fetch_d = self.well_known_resolver.get_well_known(b"testserv") # there should be an attempt to connect on port 443 for the .well-known clients = self.reactor.tcpClients @@ -963,7 +972,7 @@ class MatrixFederationAgentTests(TestCase): well_known_server.loseConnection() # repeat the request: it should hit the cache - fetch_d = well_known_resolver.get_well_known(b"testserv") + fetch_d = self.well_known_resolver.get_well_known(b"testserv") r = self.successResultOf(fetch_d) self.assertEqual(r.delegated_server, b"target-server") @@ -971,7 +980,7 @@ class MatrixFederationAgentTests(TestCase): self.reactor.pump((1000.0,)) # now it should connect again - fetch_d = well_known_resolver.get_well_known(b"testserv") + fetch_d = self.well_known_resolver.get_well_known(b"testserv") self.assertEqual(len(clients), 1) (host, port, client_factory, _timeout, _bindAddress) = clients.pop(0) @@ -992,15 +1001,9 @@ class MatrixFederationAgentTests(TestCase): it ignores transient errors. """ - well_known_resolver = WellKnownResolver( - self.reactor, - Agent(self.reactor, contextFactory=self.tls_factory), - well_known_cache=self.well_known_cache, - ) - self.reactor.lookups["testserv"] = "1.2.3.4" - fetch_d = well_known_resolver.get_well_known(b"testserv") + fetch_d = self.well_known_resolver.get_well_known(b"testserv") # there should be an attempt to connect on port 443 for the .well-known clients = self.reactor.tcpClients @@ -1026,27 +1029,37 @@ class MatrixFederationAgentTests(TestCase): # another lookup. self.reactor.pump((900.0,)) - fetch_d = well_known_resolver.get_well_known(b"testserv") - clients = self.reactor.tcpClients - (host, port, client_factory, _timeout, _bindAddress) = clients.pop(0) + fetch_d = self.well_known_resolver.get_well_known(b"testserv") - # fonx the connection attempt, this will be treated as a temporary - # failure. - client_factory.clientConnectionFailed(None, Exception("nope")) + # The resolver may retry a few times, so fonx all requests that come along + attempts = 0 + while self.reactor.tcpClients: + clients = self.reactor.tcpClients + (host, port, client_factory, _timeout, _bindAddress) = clients.pop(0) - # attemptdelay on the hostnameendpoint is 0.3, so takes that long before the - # .well-known request fails. - self.reactor.pump((0.4,)) + attempts += 1 + + # fonx the connection attempt, this will be treated as a temporary + # failure. + client_factory.clientConnectionFailed(None, Exception("nope")) + + # There's a few sleeps involved, so we have to pump the reactor a + # bit. + self.reactor.pump((1.0, 1.0)) + + # We expect to see more than one attempt as there was previously a valid + # well known. + self.assertGreater(attempts, 1) # Resolver should return cached value, despite the lookup failing. r = self.successResultOf(fetch_d) self.assertEqual(r.delegated_server, b"target-server") - # Expire the cache and repeat the request - self.reactor.pump((100.0,)) + # Expire both caches and repeat the request + self.reactor.pump((10000.0,)) # Repated the request, this time it should fail if the lookup fails. - fetch_d = well_known_resolver.get_well_known(b"testserv") + fetch_d = self.well_known_resolver.get_well_known(b"testserv") clients = self.reactor.tcpClients (host, port, client_factory, _timeout, _bindAddress) = clients.pop(0) -- cgit 1.5.1 From f299c5414c2dd300103b0e11e7114123d8eb58a1 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Thu, 8 Aug 2019 15:30:04 +0100 Subject: Refactor MatrixFederationAgent to retry SRV. This refactors MatrixFederationAgent to move the SRV lookup into the endpoint code, this has two benefits: 1. Its easier to retry different host/ports in the same way as HostnameEndpoint. 2. We avoid SRV lookups if we have a free connection in the pool --- synapse/http/federation/matrix_federation_agent.py | 356 ++++++++++----------- synapse/http/federation/srv_resolver.py | 35 +- .../federation/test_matrix_federation_agent.py | 63 +++- tests/http/federation/test_srv_resolver.py | 8 +- 4 files changed, 268 insertions(+), 194 deletions(-) (limited to 'tests') diff --git a/synapse/http/federation/matrix_federation_agent.py b/synapse/http/federation/matrix_federation_agent.py index 71a15f434d..c208185791 100644 --- a/synapse/http/federation/matrix_federation_agent.py +++ b/synapse/http/federation/matrix_federation_agent.py @@ -14,21 +14,21 @@ # limitations under the License. import logging +import urllib -import attr -from netaddr import IPAddress +from netaddr import AddrFormatError, IPAddress from zope.interface import implementer from twisted.internet import defer from twisted.internet.endpoints import HostnameEndpoint, wrapClientTLS from twisted.internet.interfaces import IStreamClientEndpoint -from twisted.web.client import URI, Agent, HTTPConnectionPool +from twisted.web.client import Agent, HTTPConnectionPool from twisted.web.http_headers import Headers -from twisted.web.iweb import IAgent +from twisted.web.iweb import IAgent, IAgentEndpointFactory -from synapse.http.federation.srv_resolver import SrvResolver, pick_server_from_list +from synapse.http.federation.srv_resolver import Server, SrvResolver from synapse.http.federation.well_known_resolver import WellKnownResolver -from synapse.logging.context import make_deferred_yieldable +from synapse.logging.context import PreserveLoggingContext, make_deferred_yieldable from synapse.util import Clock logger = logging.getLogger(__name__) @@ -36,8 +36,9 @@ logger = logging.getLogger(__name__) @implementer(IAgent) class MatrixFederationAgent(object): - """An Agent-like thing which provides a `request` method which will look up a matrix - server and send an HTTP request to it. + """An Agent-like thing which provides a `request` method which correctly + handles resolving matrix server names when using matrix://. Handles standard + https URIs as normal. Doesn't implement any retries. (Those are done in MatrixFederationHttpClient.) @@ -65,23 +66,25 @@ class MatrixFederationAgent(object): ): self._reactor = reactor self._clock = Clock(reactor) - - self._tls_client_options_factory = tls_client_options_factory - if _srv_resolver is None: - _srv_resolver = SrvResolver() - self._srv_resolver = _srv_resolver - self._pool = HTTPConnectionPool(reactor) self._pool.retryAutomatically = False self._pool.maxPersistentPerHost = 5 self._pool.cachedConnectionTimeout = 2 * 60 + self._agent = Agent.usingEndpointFactory( + self._reactor, + MatrixHostnameEndpointFactory( + reactor, tls_client_options_factory, _srv_resolver + ), + pool=self._pool, + ) + self._well_known_resolver = WellKnownResolver( self._reactor, agent=Agent( self._reactor, - pool=self._pool, contextFactory=tls_client_options_factory, + pool=self._pool, ), well_known_cache=_well_known_cache, ) @@ -91,19 +94,15 @@ class MatrixFederationAgent(object): """ Args: method (bytes): HTTP method: GET/POST/etc - uri (bytes): Absolute URI to be retrieved - headers (twisted.web.http_headers.Headers|None): HTTP headers to send with the request, or None to send no extra headers. - bodyProducer (twisted.web.iweb.IBodyProducer|None): An object which can generate bytes to make up the body of this request (for example, the properly encoded contents of a file for a file upload). Or None if the request is to have no body. - Returns: Deferred[twisted.web.iweb.IResponse]: fires when the header of the response has been received (regardless of the @@ -111,210 +110,195 @@ class MatrixFederationAgent(object): response from being received (including problems that prevent the request from being sent). """ - parsed_uri = URI.fromBytes(uri, defaultPort=-1) - res = yield self._route_matrix_uri(parsed_uri) + # We use urlparse as that will set `port` to None if there is no + # explicit port. + parsed_uri = urllib.parse.urlparse(uri) - # set up the TLS connection params + # If this is a matrix:// URI check if the server has delegated matrix + # traffic using well-known delegation. # - # XXX disabling TLS is really only supported here for the benefit of the - # unit tests. We should make the UTs cope with TLS rather than having to make - # the code support the unit tests. - if self._tls_client_options_factory is None: - tls_options = None - else: - tls_options = self._tls_client_options_factory.get_options( - res.tls_server_name.decode("ascii") + # We have to do this here and not in the endpoint as we need to rewrite + # the host header with the delegated server name. + delegated_server = None + if ( + parsed_uri.scheme == b"matrix" + and not _is_ip_literal(parsed_uri.hostname) + and not parsed_uri.port + ): + well_known_result = yield self._well_known_resolver.get_well_known( + parsed_uri.hostname + ) + delegated_server = well_known_result.delegated_server + + if delegated_server: + # Ok, the server has delegated matrix traffic to somewhere else, so + # lets rewrite the URL to replace the server with the delegated + # server name. + uri = urllib.parse.urlunparse( + ( + parsed_uri.scheme, + delegated_server, + parsed_uri.path, + parsed_uri.params, + parsed_uri.query, + parsed_uri.fragment, + ) ) + parsed_uri = urllib.parse.urlparse(uri) - # make sure that the Host header is set correctly + # We need to make sure the host header is set to the netloc of the + # server. if headers is None: headers = Headers() else: headers = headers.copy() if not headers.hasHeader(b"host"): - headers.addRawHeader(b"host", res.host_header) + headers.addRawHeader(b"host", parsed_uri.netloc) - class EndpointFactory(object): - @staticmethod - def endpointForURI(_uri): - ep = LoggingHostnameEndpoint( - self._reactor, res.target_host, res.target_port - ) - if tls_options is not None: - ep = wrapClientTLS(tls_options, ep) - return ep + with PreserveLoggingContext(): + res = yield self._agent.request(method, uri, headers, bodyProducer) - agent = Agent.usingEndpointFactory(self._reactor, EndpointFactory(), self._pool) - res = yield make_deferred_yieldable( - agent.request(method, uri, headers, bodyProducer) - ) return res - @defer.inlineCallbacks - def _route_matrix_uri(self, parsed_uri, lookup_well_known=True): - """Helper for `request`: determine the routing for a Matrix URI - Args: - parsed_uri (twisted.web.client.URI): uri to route. Note that it should be - parsed with URI.fromBytes(uri, defaultPort=-1) to set the `port` to -1 - if there is no explicit port given. +@implementer(IAgentEndpointFactory) +class MatrixHostnameEndpointFactory(object): + """Factory for MatrixHostnameEndpoint for parsing to an Agent. + """ - lookup_well_known (bool): True if we should look up the .well-known file if - there is no SRV record. + def __init__(self, reactor, tls_client_options_factory, srv_resolver): + self._reactor = reactor + self._tls_client_options_factory = tls_client_options_factory - Returns: - Deferred[_RoutingResult] - """ - # check for an IP literal - try: - ip_address = IPAddress(parsed_uri.host.decode("ascii")) - except Exception: - # not an IP address - ip_address = None - - if ip_address: - port = parsed_uri.port - if port == -1: - port = 8448 - return _RoutingResult( - host_header=parsed_uri.netloc, - tls_server_name=parsed_uri.host, - target_host=parsed_uri.host, - target_port=port, - ) + if srv_resolver is None: + srv_resolver = SrvResolver() - if parsed_uri.port != -1: - # there is an explicit port - return _RoutingResult( - host_header=parsed_uri.netloc, - tls_server_name=parsed_uri.host, - target_host=parsed_uri.host, - target_port=parsed_uri.port, - ) + self._srv_resolver = srv_resolver - if lookup_well_known: - # try a .well-known lookup - well_known_result = yield self._well_known_resolver.get_well_known( - parsed_uri.host - ) - well_known_server = well_known_result.delegated_server - - if well_known_server: - # if we found a .well-known, start again, but don't do another - # .well-known lookup. - - # parse the server name in the .well-known response into host/port. - # (This code is lifted from twisted.web.client.URI.fromBytes). - if b":" in well_known_server: - well_known_host, well_known_port = well_known_server.rsplit(b":", 1) - try: - well_known_port = int(well_known_port) - except ValueError: - # the part after the colon could not be parsed as an int - # - we assume it is an IPv6 literal with no port (the closing - # ']' stops it being parsed as an int) - well_known_host, well_known_port = well_known_server, -1 - else: - well_known_host, well_known_port = well_known_server, -1 - - new_uri = URI( - scheme=parsed_uri.scheme, - netloc=well_known_server, - host=well_known_host, - port=well_known_port, - path=parsed_uri.path, - params=parsed_uri.params, - query=parsed_uri.query, - fragment=parsed_uri.fragment, - ) + def endpointForURI(self, parsed_uri): + return MatrixHostnameEndpoint( + self._reactor, + self._tls_client_options_factory, + self._srv_resolver, + parsed_uri, + ) - res = yield self._route_matrix_uri(new_uri, lookup_well_known=False) - return res - - # try a SRV lookup - service_name = b"_matrix._tcp.%s" % (parsed_uri.host,) - server_list = yield self._srv_resolver.resolve_service(service_name) - - if not server_list: - target_host = parsed_uri.host - port = 8448 - logger.debug( - "No SRV record for %s, using %s:%i", - parsed_uri.host.decode("ascii"), - target_host.decode("ascii"), - port, - ) + +@implementer(IStreamClientEndpoint) +class MatrixHostnameEndpoint(object): + """An endpoint that resolves matrix:// URLs using Matrix server name + resolution (i.e. via SRV). Does not check for well-known delegation. + """ + + def __init__(self, reactor, tls_client_options_factory, srv_resolver, parsed_uri): + self._reactor = reactor + + # We reparse the URI so that defaultPort is -1 rather than 80 + self._parsed_uri = parsed_uri + + # set up the TLS connection params + # + # XXX disabling TLS is really only supported here for the benefit of the + # unit tests. We should make the UTs cope with TLS rather than having to make + # the code support the unit tests. + + if tls_client_options_factory is None: + self._tls_options = None else: - target_host, port = pick_server_from_list(server_list) - logger.debug( - "Picked %s:%i from SRV records for %s", - target_host.decode("ascii"), - port, - parsed_uri.host.decode("ascii"), + self._tls_options = tls_client_options_factory.get_options( + self._parsed_uri.host.decode("ascii") ) - return _RoutingResult( - host_header=parsed_uri.netloc, - tls_server_name=parsed_uri.host, - target_host=target_host, - target_port=port, - ) + self._srv_resolver = srv_resolver + @defer.inlineCallbacks + def connect(self, protocol_factory): + """Implements IStreamClientEndpoint interface + """ -@implementer(IStreamClientEndpoint) -class LoggingHostnameEndpoint(object): - """A wrapper for HostnameEndpint which logs when it connects""" + first_exception = None - def __init__(self, reactor, host, port, *args, **kwargs): - self.host = host - self.port = port - self.ep = HostnameEndpoint(reactor, host, port, *args, **kwargs) + server_list = yield self._resolve_server() - def connect(self, protocol_factory): - logger.info("Connecting to %s:%i", self.host.decode("ascii"), self.port) - return self.ep.connect(protocol_factory) + for server in server_list: + host = server.host + port = server.port + try: + logger.info("Connecting to %s:%i", host.decode("ascii"), port) + endpoint = HostnameEndpoint(self._reactor, host, port) + if self._tls_options: + endpoint = wrapClientTLS(self._tls_options, endpoint) + result = yield make_deferred_yieldable( + endpoint.connect(protocol_factory) + ) -@attr.s -class _RoutingResult(object): - """The result returned by `_route_matrix_uri`. + return result + except Exception as e: + logger.info( + "Failed to connect to %s:%i: %s", host.decode("ascii"), port, e + ) + if not first_exception: + first_exception = e - Contains the parameters needed to direct a federation connection to a particular - server. + # We return the first failure because that's probably the most interesting. + if first_exception: + raise first_exception - Where a SRV record points to several servers, this object contains a single server - chosen from the list. - """ + # This shouldn't happen as we should always have at least one host/port + # to try and if that doesn't work then we'll have an exception. + raise Exception("Failed to resolve server %r" % (self._parsed_uri.netloc,)) - host_header = attr.ib() - """ - The value we should assign to the Host header (host:port from the matrix - URI, or .well-known). + @defer.inlineCallbacks + def _resolve_server(self): + """Resolves the server name to a list of hosts and ports to attempt to + connect to. - :type: bytes - """ + Returns: + Deferred[list[Server]] + """ - tls_server_name = attr.ib() - """ - The server name we should set in the SNI (typically host, without port, from the - matrix URI or .well-known) + if self._parsed_uri.scheme != b"matrix": + return [Server(host=self._parsed_uri.host, port=self._parsed_uri.port)] - :type: bytes - """ + # Note: We don't do well-known lookup as that needs to have happened + # before now, due to needing to rewrite the Host header of the HTTP + # request. - target_host = attr.ib() - """ - The hostname (or IP literal) we should route the TCP connection to (the target of the - SRV record, or the hostname from the URL/.well-known) + parsed_uri = urllib.parse.urlparse(self._parsed_uri.toBytes()) - :type: bytes - """ + host = parsed_uri.hostname + port = parsed_uri.port - target_port = attr.ib() - """ - The port we should route the TCP connection to (the target of the SRV record, or - the port from the URL/.well-known, or 8448) + # If there is an explicit port or the host is an IP address we bypass + # SRV lookups and just use the given host/port. + if port or _is_ip_literal(host): + return [Server(host, port or 8448)] - :type: int + server_list = yield self._srv_resolver.resolve_service(b"_matrix._tcp." + host) + + if server_list: + return server_list + + # No SRV records, so we fallback to host and 8448 + return [Server(host, 8448)] + + +def _is_ip_literal(host): + """Test if the given host name is either an IPv4 or IPv6 literal. + + Args: + host (bytes) + + Returns: + bool """ + + host = host.decode("ascii") + + try: + IPAddress(host) + return True + except AddrFormatError: + return False diff --git a/synapse/http/federation/srv_resolver.py b/synapse/http/federation/srv_resolver.py index b32188766d..bbda0a23f4 100644 --- a/synapse/http/federation/srv_resolver.py +++ b/synapse/http/federation/srv_resolver.py @@ -32,7 +32,7 @@ logger = logging.getLogger(__name__) SERVER_CACHE = {} -@attr.s +@attr.s(slots=True, frozen=True) class Server(object): """ Our record of an individual server which can be tried to reach a destination. @@ -83,6 +83,35 @@ def pick_server_from_list(server_list): raise RuntimeError("pick_server_from_list got to end of eligible server list.") +def _sort_server_list(server_list): + """Given a list of SRV records sort them into priority order and shuffle + each priority with the given weight. + """ + priority_map = {} + + for server in server_list: + priority_map.setdefault(server.priority, []).append(server) + + results = [] + for priority in sorted(priority_map): + servers = priority_map.pop(priority) + + while servers: + total_weight = sum(s.weight for s in servers) + target_weight = random.randint(0, total_weight) + + for s in servers: + target_weight -= s.weight + + if target_weight <= 0: + break + + results.append(s) + servers.remove(s) + + return results + + class SrvResolver(object): """Interface to the dns client to do SRV lookups, with result caching. @@ -120,7 +149,7 @@ class SrvResolver(object): if cache_entry: if all(s.expires > now for s in cache_entry): servers = list(cache_entry) - return servers + return _sort_server_list(servers) try: answers, _, _ = yield make_deferred_yieldable( @@ -169,4 +198,4 @@ class SrvResolver(object): ) self._cache[service_name] = list(servers) - return servers + return _sort_server_list(servers) diff --git a/tests/http/federation/test_matrix_federation_agent.py b/tests/http/federation/test_matrix_federation_agent.py index 2c568788b3..f97c8a59f6 100644 --- a/tests/http/federation/test_matrix_federation_agent.py +++ b/tests/http/federation/test_matrix_federation_agent.py @@ -41,9 +41,9 @@ from synapse.http.federation.well_known_resolver import ( from synapse.logging.context import LoggingContext from synapse.util.caches.ttlcache import TTLCache +from tests import unittest from tests.http import TestServerTLSConnectionFactory, get_test_ca_cert_file from tests.server import FakeTransport, ThreadedMemoryReactorClock -from tests.unittest import TestCase from tests.utils import default_config logger = logging.getLogger(__name__) @@ -67,7 +67,8 @@ def get_connection_factory(): return test_server_connection_factory -class MatrixFederationAgentTests(TestCase): +@unittest.DEBUG +class MatrixFederationAgentTests(unittest.TestCase): def setUp(self): self.reactor = ThreadedMemoryReactorClock() @@ -1056,8 +1057,64 @@ class MatrixFederationAgentTests(TestCase): r = self.successResultOf(fetch_d) self.assertEqual(r.delegated_server, None) + def test_srv_fallbacks(self): + """Test that other SRV results are tried if the first one fails. + """ + + self.mock_resolver.resolve_service.side_effect = lambda _: [ + Server(host=b"target.com", port=8443), + Server(host=b"target.com", port=8444), + ] + self.reactor.lookups["target.com"] = "1.2.3.4" + + test_d = self._make_get_request(b"matrix://testserv/foo/bar") + + # Nothing happened yet + self.assertNoResult(test_d) + + self.mock_resolver.resolve_service.assert_called_once_with( + b"_matrix._tcp.testserv" + ) + + # We should see an attempt to connect to the first server + clients = self.reactor.tcpClients + self.assertEqual(len(clients), 1) + (host, port, client_factory, _timeout, _bindAddress) = clients.pop(0) + self.assertEqual(host, "1.2.3.4") + self.assertEqual(port, 8443) + + # Fonx the connection + client_factory.clientConnectionFailed(None, Exception("nope")) + + # There's a 300ms delay in HostnameEndpoint + self.reactor.pump((0.4,)) + + # Hasn't failed yet + self.assertNoResult(test_d) + + # We shouldnow see an attempt to connect to the second server + clients = self.reactor.tcpClients + self.assertEqual(len(clients), 1) + (host, port, client_factory, _timeout, _bindAddress) = clients.pop(0) + self.assertEqual(host, "1.2.3.4") + self.assertEqual(port, 8444) + + # make a test server, and wire up the client + http_server = self._make_connection(client_factory, expected_sni=b"testserv") + + self.assertEqual(len(http_server.requests), 1) + request = http_server.requests[0] + self.assertEqual(request.method, b"GET") + self.assertEqual(request.path, b"/foo/bar") + self.assertEqual(request.requestHeaders.getRawHeaders(b"host"), [b"testserv"]) + + # finish the request + request.finish() + self.reactor.pump((0.1,)) + self.successResultOf(test_d) + -class TestCachePeriodFromHeaders(TestCase): +class TestCachePeriodFromHeaders(unittest.TestCase): def test_cache_control(self): # uppercase self.assertEqual( diff --git a/tests/http/federation/test_srv_resolver.py b/tests/http/federation/test_srv_resolver.py index 3b885ef64b..df034ab237 100644 --- a/tests/http/federation/test_srv_resolver.py +++ b/tests/http/federation/test_srv_resolver.py @@ -83,8 +83,10 @@ class SrvResolverTestCase(unittest.TestCase): service_name = b"test_service.example.com" - entry = Mock(spec_set=["expires"]) + entry = Mock(spec_set=["expires", "priority", "weight"]) entry.expires = 0 + entry.priority = 0 + entry.weight = 0 cache = {service_name: [entry]} resolver = SrvResolver(dns_client=dns_client_mock, cache=cache) @@ -105,8 +107,10 @@ class SrvResolverTestCase(unittest.TestCase): service_name = b"test_service.example.com" - entry = Mock(spec_set=["expires"]) + entry = Mock(spec_set=["expires", "priority", "weight"]) entry.expires = 999999999 + entry.priority = 0 + entry.weight = 0 cache = {service_name: [entry]} resolver = SrvResolver( -- cgit 1.5.1 From 1dec31560e5712306e368a0adc6d9f84f924bdc9 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Tue, 20 Aug 2019 11:46:00 +0100 Subject: Change jitter to be a factor rather than absolute value --- synapse/http/federation/well_known_resolver.py | 23 +++++++++++----------- .../federation/test_matrix_federation_agent.py | 4 ++-- 2 files changed, 14 insertions(+), 13 deletions(-) (limited to 'tests') diff --git a/synapse/http/federation/well_known_resolver.py b/synapse/http/federation/well_known_resolver.py index c846003886..5e9b0befb0 100644 --- a/synapse/http/federation/well_known_resolver.py +++ b/synapse/http/federation/well_known_resolver.py @@ -32,8 +32,8 @@ from synapse.util.metrics import Measure # period to cache .well-known results for by default WELL_KNOWN_DEFAULT_CACHE_PERIOD = 24 * 3600 -# jitter to add to the .well-known default cache ttl -WELL_KNOWN_DEFAULT_CACHE_PERIOD_JITTER = 10 * 60 +# jitter factor to add to the .well-known default cache ttls +WELL_KNOWN_DEFAULT_CACHE_PERIOD_JITTER = 0.1 # period to cache failure to fetch .well-known for WELL_KNOWN_INVALID_CACHE_PERIOD = 1 * 3600 @@ -133,16 +133,14 @@ class WellKnownResolver(object): # We have recently seen a valid well-known record for this # server, so we cache the lack of well-known for a shorter time. cache_period = WELL_KNOWN_DOWN_CACHE_PERIOD - cache_period += random.uniform( - 0, WELL_KNOWN_DEFAULT_CACHE_PERIOD_JITTER - ) else: - # add some randomness to the TTL to avoid a stampeding herd every hour - # after startup cache_period = WELL_KNOWN_INVALID_CACHE_PERIOD - cache_period += random.uniform( - 0, WELL_KNOWN_DEFAULT_CACHE_PERIOD_JITTER - ) + + # add some randomness to the TTL to avoid a stampeding herd + cache_period *= random.uniform( + 1 - WELL_KNOWN_DEFAULT_CACHE_PERIOD_JITTER, + 1 + WELL_KNOWN_DEFAULT_CACHE_PERIOD_JITTER, + ) if cache_period > 0: self._well_known_cache.set(server_name, result, cache_period) @@ -194,7 +192,10 @@ class WellKnownResolver(object): cache_period = WELL_KNOWN_DEFAULT_CACHE_PERIOD # add some randomness to the TTL to avoid a stampeding herd every 24 hours # after startup - cache_period += random.uniform(0, WELL_KNOWN_DEFAULT_CACHE_PERIOD_JITTER) + cache_period *= random.uniform( + 1 - WELL_KNOWN_DEFAULT_CACHE_PERIOD_JITTER, + 1 + WELL_KNOWN_DEFAULT_CACHE_PERIOD_JITTER, + ) else: cache_period = min(cache_period, WELL_KNOWN_MAX_CACHE_PERIOD) cache_period = max(cache_period, WELL_KNOWN_MIN_CACHE_PERIOD) diff --git a/tests/http/federation/test_matrix_federation_agent.py b/tests/http/federation/test_matrix_federation_agent.py index 4d3f31d18c..c55aad8e11 100644 --- a/tests/http/federation/test_matrix_federation_agent.py +++ b/tests/http/federation/test_matrix_federation_agent.py @@ -551,7 +551,7 @@ class MatrixFederationAgentTests(TestCase): self.assertEqual(self.well_known_cache[b"testserv"], b"target-server") # check the cache expires - self.reactor.pump((25 * 3600,)) + self.reactor.pump((48 * 3600,)) self.well_known_cache.expire() self.assertNotIn(b"testserv", self.well_known_cache) @@ -639,7 +639,7 @@ class MatrixFederationAgentTests(TestCase): self.assertEqual(self.well_known_cache[b"testserv"], b"target-server") # check the cache expires - self.reactor.pump((25 * 3600,)) + self.reactor.pump((48 * 3600,)) self.well_known_cache.expire() self.assertNotIn(b"testserv", self.well_known_cache) -- cgit 1.5.1 From 7777d353bfffc840b79391da107e593338a1a2fe Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Tue, 20 Aug 2019 11:46:54 +0100 Subject: Remove test debugs --- tests/federation/test_federation_server.py | 1 - tests/http/federation/test_matrix_federation_agent.py | 1 - tests/test_visibility.py | 1 - 3 files changed, 3 deletions(-) (limited to 'tests') diff --git a/tests/federation/test_federation_server.py b/tests/federation/test_federation_server.py index af15f4cc5a..b08be451aa 100644 --- a/tests/federation/test_federation_server.py +++ b/tests/federation/test_federation_server.py @@ -20,7 +20,6 @@ from synapse.federation.federation_server import server_matches_acl_event from tests import unittest -@unittest.DEBUG class ServerACLsTestCase(unittest.TestCase): def test_blacklisted_server(self): e = _create_acl_event({"allow": ["*"], "deny": ["evil.com"]}) diff --git a/tests/http/federation/test_matrix_federation_agent.py b/tests/http/federation/test_matrix_federation_agent.py index f97c8a59f6..445a0e76ab 100644 --- a/tests/http/federation/test_matrix_federation_agent.py +++ b/tests/http/federation/test_matrix_federation_agent.py @@ -67,7 +67,6 @@ def get_connection_factory(): return test_server_connection_factory -@unittest.DEBUG class MatrixFederationAgentTests(unittest.TestCase): def setUp(self): self.reactor = ThreadedMemoryReactorClock() diff --git a/tests/test_visibility.py b/tests/test_visibility.py index e0605dac2f..18f1a0035d 100644 --- a/tests/test_visibility.py +++ b/tests/test_visibility.py @@ -74,7 +74,6 @@ class FilterEventsForServerTestCase(tests.unittest.TestCase): self.assertEqual(events_to_filter[i].event_id, filtered[i].event_id) self.assertEqual(filtered[i].content["a"], "b") - @tests.unittest.DEBUG @defer.inlineCallbacks def test_erased_user(self): # 4 message events, from erased and unerased users, with a membership -- cgit 1.5.1 From 501994582899ad9d790029b3d7c48ba32f5720a9 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff Date: Tue, 20 Aug 2019 11:20:10 +0100 Subject: Refactor the Appservice scheduler code Get rid of the labyrinthine `recoverer_fn` code, and clean up the startup code (it seemed to be previously inexplicably split between `ApplicationServiceScheduler.start` and `_Recoverer.start`). Add some docstrings too. --- changelog.d/5886.misc | 1 + synapse/appservice/scheduler.py | 110 ++++++++++++++++++++++--------------- tests/appservice/test_scheduler.py | 6 +- 3 files changed, 68 insertions(+), 49 deletions(-) create mode 100644 changelog.d/5886.misc (limited to 'tests') diff --git a/changelog.d/5886.misc b/changelog.d/5886.misc new file mode 100644 index 0000000000..22adba3d85 --- /dev/null +++ b/changelog.d/5886.misc @@ -0,0 +1 @@ +Refactor the Appservice scheduler code. diff --git a/synapse/appservice/scheduler.py b/synapse/appservice/scheduler.py index 42a350bff8..03a14402c5 100644 --- a/synapse/appservice/scheduler.py +++ b/synapse/appservice/scheduler.py @@ -70,35 +70,37 @@ class ApplicationServiceScheduler(object): self.store = hs.get_datastore() self.as_api = hs.get_application_service_api() - def create_recoverer(service, callback): - return _Recoverer(self.clock, self.store, self.as_api, service, callback) - - self.txn_ctrl = _TransactionController( - self.clock, self.store, self.as_api, create_recoverer - ) + self.txn_ctrl = _TransactionController(self.clock, self.store, self.as_api) self.queuer = _ServiceQueuer(self.txn_ctrl, self.clock) @defer.inlineCallbacks def start(self): logger.info("Starting appservice scheduler") + # check for any DOWN ASes and start recoverers for them. - recoverers = yield _Recoverer.start( - self.clock, self.store, self.as_api, self.txn_ctrl.on_recovered + services = yield self.store.get_appservices_by_state( + ApplicationServiceState.DOWN ) - self.txn_ctrl.add_recoverers(recoverers) + + for service in services: + self.txn_ctrl.start_recoverer(service) def submit_event_for_as(self, service, event): self.queuer.enqueue(service, event) class _ServiceQueuer(object): - """Queues events for the same application service together, sending - transactions as soon as possible. Once a transaction is sent successfully, - this schedules any other events in the queue to run. + """Queue of events waiting to be sent to appservices. + + Groups events into transactions per-appservice, and sends them on to the + TransactionController. Makes sure that we only have one transaction in flight per + appservice at a given time. """ def __init__(self, txn_ctrl, clock): self.queued_events = {} # dict of {service_id: [events]} + + # the appservices which currently have a transaction in flight self.requests_in_flight = set() self.txn_ctrl = txn_ctrl self.clock = clock @@ -136,13 +138,29 @@ class _ServiceQueuer(object): class _TransactionController(object): - def __init__(self, clock, store, as_api, recoverer_fn): + """Transaction manager. + + Builds AppServiceTransactions and runs their lifecycle. Also starts a Recoverer + if a transaction fails. + + (Note we have only have one of these in the homeserver.) + + Args: + clock (synapse.util.Clock): + store (synapse.storage.DataStore): + as_api (synapse.appservice.api.ApplicationServiceApi): + """ + + def __init__(self, clock, store, as_api): self.clock = clock self.store = store self.as_api = as_api - self.recoverer_fn = recoverer_fn - # keep track of how many recoverers there are - self.recoverers = [] + + # map from service id to recoverer instance + self.recoverers = {} + + # for UTs + self.RECOVERER_CLASS = _Recoverer @defer.inlineCallbacks def send(self, service, events): @@ -154,42 +172,45 @@ class _TransactionController(object): if sent: yield txn.complete(self.store) else: - run_in_background(self._start_recoverer, service) + run_in_background(self._on_txn_fail, service) except Exception: logger.exception("Error creating appservice transaction") - run_in_background(self._start_recoverer, service) + run_in_background(self._on_txn_fail, service) @defer.inlineCallbacks def on_recovered(self, recoverer): - self.recoverers.remove(recoverer) logger.info( "Successfully recovered application service AS ID %s", recoverer.service.id ) + self.recoverers.pop(recoverer.service.id) logger.info("Remaining active recoverers: %s", len(self.recoverers)) yield self.store.set_appservice_state( recoverer.service, ApplicationServiceState.UP ) - def add_recoverers(self, recoverers): - for r in recoverers: - self.recoverers.append(r) - if len(recoverers) > 0: - logger.info("New active recoverers: %s", len(self.recoverers)) - @defer.inlineCallbacks - def _start_recoverer(self, service): + def _on_txn_fail(self, service): try: yield self.store.set_appservice_state(service, ApplicationServiceState.DOWN) - logger.info( - "Application service falling behind. Starting recoverer. AS ID %s", - service.id, - ) - recoverer = self.recoverer_fn(service, self.on_recovered) - self.add_recoverers([recoverer]) - recoverer.recover() + self.start_recoverer(service) except Exception: logger.exception("Error starting AS recoverer") + def start_recoverer(self, service): + """Start a Recoverer for the given service + + Args: + service (synapse.appservice.ApplicationService): + """ + logger.info("Starting recoverer for AS ID %s", service.id) + assert service.id not in self.recoverers + recoverer = self.RECOVERER_CLASS( + self.clock, self.store, self.as_api, service, self.on_recovered + ) + self.recoverers[service.id] = recoverer + recoverer.recover() + logger.info("Now %i active recoverers", len(self.recoverers)) + @defer.inlineCallbacks def _is_service_up(self, service): state = yield self.store.get_appservice_state(service) @@ -197,18 +218,17 @@ class _TransactionController(object): class _Recoverer(object): - @staticmethod - @defer.inlineCallbacks - def start(clock, store, as_api, callback): - services = yield store.get_appservices_by_state(ApplicationServiceState.DOWN) - recoverers = [_Recoverer(clock, store, as_api, s, callback) for s in services] - for r in recoverers: - logger.info( - "Starting recoverer for AS ID %s which was marked as " "DOWN", - r.service.id, - ) - r.recover() - return recoverers + """Manages retries and backoff for a DOWN appservice. + + We have one of these for each appservice which is currently considered DOWN. + + Args: + clock (synapse.util.Clock): + store (synapse.storage.DataStore): + as_api (synapse.appservice.api.ApplicationServiceApi): + service (synapse.appservice.ApplicationService): the service we are managing + callback (callable[_Recoverer]): called once the service recovers. + """ def __init__(self, clock, store, as_api, service, callback): self.clock = clock diff --git a/tests/appservice/test_scheduler.py b/tests/appservice/test_scheduler.py index 04b8c2c07c..52f89d3f83 100644 --- a/tests/appservice/test_scheduler.py +++ b/tests/appservice/test_scheduler.py @@ -37,11 +37,9 @@ class ApplicationServiceSchedulerTransactionCtrlTestCase(unittest.TestCase): self.recoverer = Mock() self.recoverer_fn = Mock(return_value=self.recoverer) self.txnctrl = _TransactionController( - clock=self.clock, - store=self.store, - as_api=self.as_api, - recoverer_fn=self.recoverer_fn, + clock=self.clock, store=self.store, as_api=self.as_api ) + self.txnctrl.RECOVERER_CLASS = self.recoverer_fn def test_single_service_up_txn_sent(self): # Test: The AS is up and the txn is successfully sent. -- cgit 1.5.1 From 0fb5189072f61416e7f616aa9d90d8c981e6b8b3 Mon Sep 17 00:00:00 2001 From: Half-Shot Date: Fri, 23 Aug 2019 09:25:35 +0100 Subject: Fix registration test --- tests/storage/test_registration.py | 1 + 1 file changed, 1 insertion(+) (limited to 'tests') diff --git a/tests/storage/test_registration.py b/tests/storage/test_registration.py index 0253c4ac05..4578cc3b60 100644 --- a/tests/storage/test_registration.py +++ b/tests/storage/test_registration.py @@ -49,6 +49,7 @@ class RegistrationStoreTestCase(unittest.TestCase): "consent_server_notice_sent": None, "appservice_id": None, "creation_ts": 1000, + "user_type": None, }, (yield self.store.get_user_by_id(self.user_id)), ) -- cgit 1.5.1 From 7dc398586c2156a456d9526ac0e42c1fec9f8143 Mon Sep 17 00:00:00 2001 From: Amber Brown Date: Wed, 28 Aug 2019 21:18:53 +1000 Subject: Implement a structured logging output system. (#5680) --- .buildkite/docker-compose.py35.pg95.yaml | 1 + .buildkite/docker-compose.py37.pg11.yaml | 1 + .buildkite/docker-compose.py37.pg95.yaml | 1 + .buildkite/pipeline.yml | 10 +- .gitignore | 5 +- changelog.d/5680.misc | 1 + docs/structured_logging.md | 83 +++++++ synapse/app/_base.py | 12 +- synapse/app/admin_cmd.py | 4 +- synapse/app/appservice.py | 4 +- synapse/app/client_reader.py | 4 +- synapse/app/event_creator.py | 4 +- synapse/app/federation_reader.py | 4 +- synapse/app/federation_sender.py | 4 +- synapse/app/frontend_proxy.py | 4 +- synapse/app/homeserver.py | 4 +- synapse/app/media_repository.py | 4 +- synapse/app/pusher.py | 4 +- synapse/app/synchrotron.py | 4 +- synapse/app/user_dir.py | 4 +- synapse/config/logger.py | 103 +++++---- synapse/handlers/federation.py | 5 +- synapse/logging/_structured.py | 374 +++++++++++++++++++++++++++++++ synapse/logging/_terse_json.py | 278 +++++++++++++++++++++++ synapse/logging/context.py | 14 +- synapse/python_dependencies.py | 6 +- tests/logging/__init__.py | 0 tests/logging/test_structured.py | 197 ++++++++++++++++ tests/logging/test_terse_json.py | 234 +++++++++++++++++++ tests/server.py | 27 ++- tox.ini | 10 + 31 files changed, 1328 insertions(+), 82 deletions(-) create mode 100644 changelog.d/5680.misc create mode 100644 docs/structured_logging.md create mode 100644 synapse/logging/_structured.py create mode 100644 synapse/logging/_terse_json.py create mode 100644 tests/logging/__init__.py create mode 100644 tests/logging/test_structured.py create mode 100644 tests/logging/test_terse_json.py (limited to 'tests') diff --git a/.buildkite/docker-compose.py35.pg95.yaml b/.buildkite/docker-compose.py35.pg95.yaml index 2f14387fbc..aaea33006b 100644 --- a/.buildkite/docker-compose.py35.pg95.yaml +++ b/.buildkite/docker-compose.py35.pg95.yaml @@ -6,6 +6,7 @@ services: image: postgres:9.5 environment: POSTGRES_PASSWORD: postgres + command: -c fsync=off testenv: image: python:3.5 diff --git a/.buildkite/docker-compose.py37.pg11.yaml b/.buildkite/docker-compose.py37.pg11.yaml index f3eec05ceb..1b32675e78 100644 --- a/.buildkite/docker-compose.py37.pg11.yaml +++ b/.buildkite/docker-compose.py37.pg11.yaml @@ -6,6 +6,7 @@ services: image: postgres:11 environment: POSTGRES_PASSWORD: postgres + command: -c fsync=off testenv: image: python:3.7 diff --git a/.buildkite/docker-compose.py37.pg95.yaml b/.buildkite/docker-compose.py37.pg95.yaml index 2a41db8eba..7679f6508d 100644 --- a/.buildkite/docker-compose.py37.pg95.yaml +++ b/.buildkite/docker-compose.py37.pg95.yaml @@ -6,6 +6,7 @@ services: image: postgres:9.5 environment: POSTGRES_PASSWORD: postgres + command: -c fsync=off testenv: image: python:3.7 diff --git a/.buildkite/pipeline.yml b/.buildkite/pipeline.yml index b75269a155..d9327227ed 100644 --- a/.buildkite/pipeline.yml +++ b/.buildkite/pipeline.yml @@ -45,8 +45,15 @@ steps: - docker#v3.0.1: image: "python:3.6" - - wait + - command: + - "python -m pip install tox" + - "tox -e mypy" + label: ":mypy: mypy" + plugins: + - docker#v3.0.1: + image: "python:3.5" + - wait - command: - "apt-get update && apt-get install -y python3.5 python3.5-dev python3-pip libxml2-dev libxslt-dev zlib1g-dev" @@ -55,6 +62,7 @@ steps: label: ":python: 3.5 / SQLite / Old Deps" env: TRIAL_FLAGS: "-j 2" + LANG: "C.UTF-8" plugins: - docker#v3.0.1: image: "ubuntu:xenial" # We use xenail to get an old sqlite and python diff --git a/.gitignore b/.gitignore index f6168a8819..e53d4908d5 100644 --- a/.gitignore +++ b/.gitignore @@ -20,6 +20,7 @@ _trial_temp*/ /*.signing.key /env/ /homeserver*.yaml +/logs /media_store/ /uploads @@ -29,8 +30,9 @@ _trial_temp*/ /.vscode/ # build products -/.coverage* !/.coveragerc +/.coverage* +/.mypy_cache/ /.tox /build/ /coverage.* @@ -38,4 +40,3 @@ _trial_temp*/ /docs/build/ /htmlcov /pip-wheel-metadata/ - diff --git a/changelog.d/5680.misc b/changelog.d/5680.misc new file mode 100644 index 0000000000..46a403a188 --- /dev/null +++ b/changelog.d/5680.misc @@ -0,0 +1 @@ +Lay the groundwork for structured logging output. diff --git a/docs/structured_logging.md b/docs/structured_logging.md new file mode 100644 index 0000000000..decec9b8fa --- /dev/null +++ b/docs/structured_logging.md @@ -0,0 +1,83 @@ +# Structured Logging + +A structured logging system can be useful when your logs are destined for a machine to parse and process. By maintaining its machine-readable characteristics, it enables more efficient searching and aggregations when consumed by software such as the "ELK stack". + +Synapse's structured logging system is configured via the file that Synapse's `log_config` config option points to. The file must be YAML and contain `structured: true`. It must contain a list of "drains" (places where logs go to). + +A structured logging configuration looks similar to the following: + +```yaml +structured: true + +loggers: + synapse: + level: INFO + synapse.storage.SQL: + level: WARNING + +drains: + console: + type: console + location: stdout + file: + type: file_json + location: homeserver.log +``` + +The above logging config will set Synapse as 'INFO' logging level by default, with the SQL layer at 'WARNING', and will have two logging drains (to the console and to a file, stored as JSON). + +## Drain Types + +Drain types can be specified by the `type` key. + +### `console` + +Outputs human-readable logs to the console. + +Arguments: + +- `location`: Either `stdout` or `stderr`. + +### `console_json` + +Outputs machine-readable JSON logs to the console. + +Arguments: + +- `location`: Either `stdout` or `stderr`. + +### `console_json_terse` + +Outputs machine-readable JSON logs to the console, separated by newlines. This +format is not designed to be read and re-formatted into human-readable text, but +is optimal for a logging aggregation system. + +Arguments: + +- `location`: Either `stdout` or `stderr`. + +### `file` + +Outputs human-readable logs to a file. + +Arguments: + +- `location`: An absolute path to the file to log to. + +### `file_json` + +Outputs machine-readable logs to a file. + +Arguments: + +- `location`: An absolute path to the file to log to. + +### `network_json_terse` + +Delivers machine-readable JSON logs to a log aggregator over TCP. This is +compatible with LogStash's TCP input with the codec set to `json_lines`. + +Arguments: + +- `host`: Hostname or IP address of the log aggregator. +- `port`: Numerical port to contact on the host. \ No newline at end of file diff --git a/synapse/app/_base.py b/synapse/app/_base.py index 69dcf3523f..c30fdeee9a 100644 --- a/synapse/app/_base.py +++ b/synapse/app/_base.py @@ -36,18 +36,20 @@ from synapse.util.versionstring import get_version_string logger = logging.getLogger(__name__) +# list of tuples of function, args list, kwargs dict _sighup_callbacks = [] -def register_sighup(func): +def register_sighup(func, *args, **kwargs): """ Register a function to be called when a SIGHUP occurs. Args: func (function): Function to be called when sent a SIGHUP signal. - Will be called with a single argument, the homeserver. + Will be called with a single default argument, the homeserver. + *args, **kwargs: args and kwargs to be passed to the target function. """ - _sighup_callbacks.append(func) + _sighup_callbacks.append((func, args, kwargs)) def start_worker_reactor(appname, config, run_command=reactor.run): @@ -248,8 +250,8 @@ def start(hs, listeners=None): # we're not using systemd. sdnotify(b"RELOADING=1") - for i in _sighup_callbacks: - i(hs) + for i, args, kwargs in _sighup_callbacks: + i(hs, *args, **kwargs) sdnotify(b"READY=1") diff --git a/synapse/app/admin_cmd.py b/synapse/app/admin_cmd.py index 1fd52a5526..04751a6a5e 100644 --- a/synapse/app/admin_cmd.py +++ b/synapse/app/admin_cmd.py @@ -227,8 +227,6 @@ def start(config_options): config.start_pushers = False config.send_federation = False - setup_logging(config, use_worker_options=True) - synapse.events.USE_FROZEN_DICTS = config.use_frozen_dicts database_engine = create_engine(config.database_config) @@ -241,6 +239,8 @@ def start(config_options): database_engine=database_engine, ) + setup_logging(ss, config, use_worker_options=True) + ss.setup() # We use task.react as the basic run command as it correctly handles tearing diff --git a/synapse/app/appservice.py b/synapse/app/appservice.py index 54bb114dec..767b87d2db 100644 --- a/synapse/app/appservice.py +++ b/synapse/app/appservice.py @@ -141,8 +141,6 @@ def start(config_options): assert config.worker_app == "synapse.app.appservice" - setup_logging(config, use_worker_options=True) - events.USE_FROZEN_DICTS = config.use_frozen_dicts database_engine = create_engine(config.database_config) @@ -167,6 +165,8 @@ def start(config_options): database_engine=database_engine, ) + setup_logging(ps, config, use_worker_options=True) + ps.setup() reactor.addSystemEventTrigger( "before", "startup", _base.start, ps, config.worker_listeners diff --git a/synapse/app/client_reader.py b/synapse/app/client_reader.py index 721bb5b119..86193d35a8 100644 --- a/synapse/app/client_reader.py +++ b/synapse/app/client_reader.py @@ -179,8 +179,6 @@ def start(config_options): assert config.worker_app == "synapse.app.client_reader" - setup_logging(config, use_worker_options=True) - events.USE_FROZEN_DICTS = config.use_frozen_dicts database_engine = create_engine(config.database_config) @@ -193,6 +191,8 @@ def start(config_options): database_engine=database_engine, ) + setup_logging(ss, config, use_worker_options=True) + ss.setup() reactor.addSystemEventTrigger( "before", "startup", _base.start, ss, config.worker_listeners diff --git a/synapse/app/event_creator.py b/synapse/app/event_creator.py index 473c8895d0..c67fe69a50 100644 --- a/synapse/app/event_creator.py +++ b/synapse/app/event_creator.py @@ -175,8 +175,6 @@ def start(config_options): assert config.worker_replication_http_port is not None - setup_logging(config, use_worker_options=True) - # This should only be done on the user directory worker or the master config.update_user_directory = False @@ -192,6 +190,8 @@ def start(config_options): database_engine=database_engine, ) + setup_logging(ss, config, use_worker_options=True) + ss.setup() reactor.addSystemEventTrigger( "before", "startup", _base.start, ss, config.worker_listeners diff --git a/synapse/app/federation_reader.py b/synapse/app/federation_reader.py index 5255d9e8cc..1ef027a88c 100644 --- a/synapse/app/federation_reader.py +++ b/synapse/app/federation_reader.py @@ -160,8 +160,6 @@ def start(config_options): assert config.worker_app == "synapse.app.federation_reader" - setup_logging(config, use_worker_options=True) - events.USE_FROZEN_DICTS = config.use_frozen_dicts database_engine = create_engine(config.database_config) @@ -174,6 +172,8 @@ def start(config_options): database_engine=database_engine, ) + setup_logging(ss, config, use_worker_options=True) + ss.setup() reactor.addSystemEventTrigger( "before", "startup", _base.start, ss, config.worker_listeners diff --git a/synapse/app/federation_sender.py b/synapse/app/federation_sender.py index c5a2880e69..04fbb407af 100644 --- a/synapse/app/federation_sender.py +++ b/synapse/app/federation_sender.py @@ -171,8 +171,6 @@ def start(config_options): assert config.worker_app == "synapse.app.federation_sender" - setup_logging(config, use_worker_options=True) - events.USE_FROZEN_DICTS = config.use_frozen_dicts database_engine = create_engine(config.database_config) @@ -197,6 +195,8 @@ def start(config_options): database_engine=database_engine, ) + setup_logging(ss, config, use_worker_options=True) + ss.setup() reactor.addSystemEventTrigger( "before", "startup", _base.start, ss, config.worker_listeners diff --git a/synapse/app/frontend_proxy.py b/synapse/app/frontend_proxy.py index e2822ca848..611d285421 100644 --- a/synapse/app/frontend_proxy.py +++ b/synapse/app/frontend_proxy.py @@ -232,8 +232,6 @@ def start(config_options): assert config.worker_main_http_uri is not None - setup_logging(config, use_worker_options=True) - events.USE_FROZEN_DICTS = config.use_frozen_dicts database_engine = create_engine(config.database_config) @@ -246,6 +244,8 @@ def start(config_options): database_engine=database_engine, ) + setup_logging(ss, config, use_worker_options=True) + ss.setup() reactor.addSystemEventTrigger( "before", "startup", _base.start, ss, config.worker_listeners diff --git a/synapse/app/homeserver.py b/synapse/app/homeserver.py index 8233905844..04f1ed14f3 100644 --- a/synapse/app/homeserver.py +++ b/synapse/app/homeserver.py @@ -341,8 +341,6 @@ def setup(config_options): # generating config files and shouldn't try to continue. sys.exit(0) - synapse.config.logger.setup_logging(config, use_worker_options=False) - events.USE_FROZEN_DICTS = config.use_frozen_dicts database_engine = create_engine(config.database_config) @@ -356,6 +354,8 @@ def setup(config_options): database_engine=database_engine, ) + synapse.config.logger.setup_logging(hs, config, use_worker_options=False) + logger.info("Preparing database: %s...", config.database_config["name"]) try: diff --git a/synapse/app/media_repository.py b/synapse/app/media_repository.py index 3a168577c7..2ac783ffa3 100644 --- a/synapse/app/media_repository.py +++ b/synapse/app/media_repository.py @@ -155,8 +155,6 @@ def start(config_options): "Please add ``enable_media_repo: false`` to the main config\n" ) - setup_logging(config, use_worker_options=True) - events.USE_FROZEN_DICTS = config.use_frozen_dicts database_engine = create_engine(config.database_config) @@ -169,6 +167,8 @@ def start(config_options): database_engine=database_engine, ) + setup_logging(ss, config, use_worker_options=True) + ss.setup() reactor.addSystemEventTrigger( "before", "startup", _base.start, ss, config.worker_listeners diff --git a/synapse/app/pusher.py b/synapse/app/pusher.py index 692ffa2f04..d84732ee3c 100644 --- a/synapse/app/pusher.py +++ b/synapse/app/pusher.py @@ -184,8 +184,6 @@ def start(config_options): assert config.worker_app == "synapse.app.pusher" - setup_logging(config, use_worker_options=True) - events.USE_FROZEN_DICTS = config.use_frozen_dicts if config.start_pushers: @@ -210,6 +208,8 @@ def start(config_options): database_engine=database_engine, ) + setup_logging(ps, config, use_worker_options=True) + ps.setup() def start(): diff --git a/synapse/app/synchrotron.py b/synapse/app/synchrotron.py index a1c3b162f7..473026fce5 100644 --- a/synapse/app/synchrotron.py +++ b/synapse/app/synchrotron.py @@ -435,8 +435,6 @@ def start(config_options): assert config.worker_app == "synapse.app.synchrotron" - setup_logging(config, use_worker_options=True) - synapse.events.USE_FROZEN_DICTS = config.use_frozen_dicts database_engine = create_engine(config.database_config) @@ -450,6 +448,8 @@ def start(config_options): application_service_handler=SynchrotronApplicationService(), ) + setup_logging(ss, config, use_worker_options=True) + ss.setup() reactor.addSystemEventTrigger( "before", "startup", _base.start, ss, config.worker_listeners diff --git a/synapse/app/user_dir.py b/synapse/app/user_dir.py index cb29a1afab..e01afb39f2 100644 --- a/synapse/app/user_dir.py +++ b/synapse/app/user_dir.py @@ -197,8 +197,6 @@ def start(config_options): assert config.worker_app == "synapse.app.user_dir" - setup_logging(config, use_worker_options=True) - events.USE_FROZEN_DICTS = config.use_frozen_dicts database_engine = create_engine(config.database_config) @@ -223,6 +221,8 @@ def start(config_options): database_engine=database_engine, ) + setup_logging(ss, config, use_worker_options=True) + ss.setup() reactor.addSystemEventTrigger( "before", "startup", _base.start, ss, config.worker_listeners diff --git a/synapse/config/logger.py b/synapse/config/logger.py index d321d00b80..981df5a10c 100644 --- a/synapse/config/logger.py +++ b/synapse/config/logger.py @@ -25,6 +25,10 @@ from twisted.logger import STDLibLogObserver, globalLogBeginner import synapse from synapse.app import _base as appbase +from synapse.logging._structured import ( + reload_structured_logging, + setup_structured_logging, +) from synapse.logging.context import LoggingContextFilter from synapse.util.versionstring import get_version_string @@ -119,21 +123,10 @@ class LoggingConfig(Config): log_config_file.write(DEFAULT_LOG_CONFIG.substitute(log_file=log_file)) -def setup_logging(config, use_worker_options=False): - """ Set up python logging - - Args: - config (LoggingConfig | synapse.config.workers.WorkerConfig): - configuration data - - use_worker_options (bool): True to use the 'worker_log_config' option - instead of 'log_config'. - - register_sighup (func | None): Function to call to register a - sighup handler. +def _setup_stdlib_logging(config, log_config): + """ + Set up Python stdlib logging. """ - log_config = config.worker_log_config if use_worker_options else config.log_config - if log_config is None: log_format = ( "%(asctime)s - %(name)s - %(lineno)d - %(levelname)s - %(request)s" @@ -151,35 +144,10 @@ def setup_logging(config, use_worker_options=False): handler.addFilter(LoggingContextFilter(request="")) logger.addHandler(handler) else: + logging.config.dictConfig(log_config) - def load_log_config(): - with open(log_config, "r") as f: - logging.config.dictConfig(yaml.safe_load(f)) - - def sighup(*args): - # it might be better to use a file watcher or something for this. - load_log_config() - logging.info("Reloaded log config from %s due to SIGHUP", log_config) - - load_log_config() - appbase.register_sighup(sighup) - - # make sure that the first thing we log is a thing we can grep backwards - # for - logging.warn("***** STARTING SERVER *****") - logging.warn("Server %s version %s", sys.argv[0], get_version_string(synapse)) - logging.info("Server hostname: %s", config.server_name) - - # It's critical to point twisted's internal logging somewhere, otherwise it - # stacks up and leaks kup to 64K object; - # see: https://twistedmatrix.com/trac/ticket/8164 - # - # Routing to the python logging framework could be a performance problem if - # the handlers blocked for a long time as python.logging is a blocking API - # see https://twistedmatrix.com/documents/current/core/howto/logger.html - # filed as https://github.com/matrix-org/synapse/issues/1727 - # - # However this may not be too much of a problem if we are just writing to a file. + # Route Twisted's native logging through to the standard library logging + # system. observer = STDLibLogObserver() def _log(event): @@ -201,3 +169,54 @@ def setup_logging(config, use_worker_options=False): ) if not config.no_redirect_stdio: print("Redirected stdout/stderr to logs") + + +def _reload_stdlib_logging(*args, log_config=None): + logger = logging.getLogger("") + + if not log_config: + logger.warn("Reloaded a blank config?") + + logging.config.dictConfig(log_config) + + +def setup_logging(hs, config, use_worker_options=False): + """ + Set up the logging subsystem. + + Args: + config (LoggingConfig | synapse.config.workers.WorkerConfig): + configuration data + + use_worker_options (bool): True to use the 'worker_log_config' option + instead of 'log_config'. + """ + log_config = config.worker_log_config if use_worker_options else config.log_config + + def read_config(*args, callback=None): + if log_config is None: + return None + + with open(log_config, "rb") as f: + log_config_body = yaml.safe_load(f.read()) + + if callback: + callback(log_config=log_config_body) + logging.info("Reloaded log config from %s due to SIGHUP", log_config) + + return log_config_body + + log_config_body = read_config() + + if log_config_body and log_config_body.get("structured") is True: + setup_structured_logging(hs, config, log_config_body) + appbase.register_sighup(read_config, callback=reload_structured_logging) + else: + _setup_stdlib_logging(config, log_config_body) + appbase.register_sighup(read_config, callback=_reload_stdlib_logging) + + # make sure that the first thing we log is a thing we can grep backwards + # for + logging.warn("***** STARTING SERVER *****") + logging.warn("Server %s version %s", sys.argv[0], get_version_string(synapse)) + logging.info("Server hostname: %s", config.server_name) diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py index c86903b98b..94306c94a9 100644 --- a/synapse/handlers/federation.py +++ b/synapse/handlers/federation.py @@ -326,8 +326,9 @@ class FederationHandler(BaseHandler): ours = yield self.store.get_state_groups_ids(room_id, seen) # state_maps is a list of mappings from (type, state_key) to event_id - # type: list[dict[tuple[str, str], str]] - state_maps = list(ours.values()) + state_maps = list( + ours.values() + ) # type: list[dict[tuple[str, str], str]] # we don't need this any more, let's delete it. del ours diff --git a/synapse/logging/_structured.py b/synapse/logging/_structured.py new file mode 100644 index 0000000000..0367d6dfc4 --- /dev/null +++ b/synapse/logging/_structured.py @@ -0,0 +1,374 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging +import os.path +import sys +import typing +import warnings + +import attr +from constantly import NamedConstant, Names, ValueConstant, Values +from zope.interface import implementer + +from twisted.logger import ( + FileLogObserver, + FilteringLogObserver, + ILogObserver, + LogBeginner, + Logger, + LogLevel, + LogLevelFilterPredicate, + LogPublisher, + eventAsText, + globalLogBeginner, + jsonFileLogObserver, +) + +from synapse.config._base import ConfigError +from synapse.logging._terse_json import ( + TerseJSONToConsoleLogObserver, + TerseJSONToTCPLogObserver, +) +from synapse.logging.context import LoggingContext + + +def stdlib_log_level_to_twisted(level: str) -> LogLevel: + """ + Convert a stdlib log level to Twisted's log level. + """ + lvl = level.lower().replace("warning", "warn") + return LogLevel.levelWithName(lvl) + + +@attr.s +@implementer(ILogObserver) +class LogContextObserver(object): + """ + An ILogObserver which adds Synapse-specific log context information. + + Attributes: + observer (ILogObserver): The target parent observer. + """ + + observer = attr.ib() + + def __call__(self, event: dict) -> None: + """ + Consume a log event and emit it to the parent observer after filtering + and adding log context information. + + Args: + event (dict) + """ + # Filter out some useless events that Twisted outputs + if "log_text" in event: + if event["log_text"].startswith("DNSDatagramProtocol starting on "): + return + + if event["log_text"].startswith("(UDP Port "): + return + + if event["log_text"].startswith("Timing out client") or event[ + "log_format" + ].startswith("Timing out client"): + return + + context = LoggingContext.current_context() + + # Copy the context information to the log event. + if context is not None: + context.copy_to_twisted_log_entry(event) + else: + # If there's no logging context, not even the root one, we might be + # starting up or it might be from non-Synapse code. Log it as if it + # came from the root logger. + event["request"] = None + event["scope"] = None + + self.observer(event) + + +class PythonStdlibToTwistedLogger(logging.Handler): + """ + Transform a Python stdlib log message into a Twisted one. + """ + + def __init__(self, observer, *args, **kwargs): + """ + Args: + observer (ILogObserver): A Twisted logging observer. + *args, **kwargs: Args/kwargs to be passed to logging.Handler. + """ + self.observer = observer + super().__init__(*args, **kwargs) + + def emit(self, record: logging.LogRecord) -> None: + """ + Emit a record to Twisted's observer. + + Args: + record (logging.LogRecord) + """ + + self.observer( + { + "log_time": record.created, + "log_text": record.getMessage(), + "log_format": "{log_text}", + "log_namespace": record.name, + "log_level": stdlib_log_level_to_twisted(record.levelname), + } + ) + + +def SynapseFileLogObserver(outFile: typing.io.TextIO) -> FileLogObserver: + """ + A log observer that formats events like the traditional log formatter and + sends them to `outFile`. + + Args: + outFile (file object): The file object to write to. + """ + + def formatEvent(_event: dict) -> str: + event = dict(_event) + event["log_level"] = event["log_level"].name.upper() + event["log_format"] = "- {log_namespace} - {log_level} - {request} - " + ( + event.get("log_format", "{log_text}") or "{log_text}" + ) + return eventAsText(event, includeSystem=False) + "\n" + + return FileLogObserver(outFile, formatEvent) + + +class DrainType(Names): + CONSOLE = NamedConstant() + CONSOLE_JSON = NamedConstant() + CONSOLE_JSON_TERSE = NamedConstant() + FILE = NamedConstant() + FILE_JSON = NamedConstant() + NETWORK_JSON_TERSE = NamedConstant() + + +class OutputPipeType(Values): + stdout = ValueConstant(sys.__stdout__) + stderr = ValueConstant(sys.__stderr__) + + +@attr.s +class DrainConfiguration(object): + name = attr.ib() + type = attr.ib() + location = attr.ib() + options = attr.ib(default=None) + + +@attr.s +class NetworkJSONTerseOptions(object): + maximum_buffer = attr.ib(type=int) + + +DEFAULT_LOGGERS = {"synapse": {"level": "INFO"}} + + +def parse_drain_configs( + drains: dict +) -> typing.Generator[DrainConfiguration, None, None]: + """ + Parse the drain configurations. + + Args: + drains (dict): A list of drain configurations. + + Yields: + DrainConfiguration instances. + + Raises: + ConfigError: If any of the drain configuration items are invalid. + """ + for name, config in drains.items(): + if "type" not in config: + raise ConfigError("Logging drains require a 'type' key.") + + try: + logging_type = DrainType.lookupByName(config["type"].upper()) + except ValueError: + raise ConfigError( + "%s is not a known logging drain type." % (config["type"],) + ) + + if logging_type in [ + DrainType.CONSOLE, + DrainType.CONSOLE_JSON, + DrainType.CONSOLE_JSON_TERSE, + ]: + location = config.get("location") + if location is None or location not in ["stdout", "stderr"]: + raise ConfigError( + ( + "The %s drain needs the 'location' key set to " + "either 'stdout' or 'stderr'." + ) + % (logging_type,) + ) + + pipe = OutputPipeType.lookupByName(location).value + + yield DrainConfiguration(name=name, type=logging_type, location=pipe) + + elif logging_type in [DrainType.FILE, DrainType.FILE_JSON]: + if "location" not in config: + raise ConfigError( + "The %s drain needs the 'location' key set." % (logging_type,) + ) + + location = config.get("location") + if os.path.abspath(location) != location: + raise ConfigError( + "File paths need to be absolute, '%s' is a relative path" + % (location,) + ) + yield DrainConfiguration(name=name, type=logging_type, location=location) + + elif logging_type in [DrainType.NETWORK_JSON_TERSE]: + host = config.get("host") + port = config.get("port") + maximum_buffer = config.get("maximum_buffer", 1000) + yield DrainConfiguration( + name=name, + type=logging_type, + location=(host, port), + options=NetworkJSONTerseOptions(maximum_buffer=maximum_buffer), + ) + + else: + raise ConfigError( + "The %s drain type is currently not implemented." + % (config["type"].upper(),) + ) + + +def setup_structured_logging( + hs, + config, + log_config: dict, + logBeginner: LogBeginner = globalLogBeginner, + redirect_stdlib_logging: bool = True, +) -> LogPublisher: + """ + Set up Twisted's structured logging system. + + Args: + hs: The homeserver to use. + config (HomeserverConfig): The configuration of the Synapse homeserver. + log_config (dict): The log configuration to use. + """ + if config.no_redirect_stdio: + raise ConfigError( + "no_redirect_stdio cannot be defined using structured logging." + ) + + logger = Logger() + + if "drains" not in log_config: + raise ConfigError("The logging configuration requires a list of drains.") + + observers = [] + + for observer in parse_drain_configs(log_config["drains"]): + # Pipe drains + if observer.type == DrainType.CONSOLE: + logger.debug( + "Starting up the {name} console logger drain", name=observer.name + ) + observers.append(SynapseFileLogObserver(observer.location)) + elif observer.type == DrainType.CONSOLE_JSON: + logger.debug( + "Starting up the {name} JSON console logger drain", name=observer.name + ) + observers.append(jsonFileLogObserver(observer.location)) + elif observer.type == DrainType.CONSOLE_JSON_TERSE: + logger.debug( + "Starting up the {name} terse JSON console logger drain", + name=observer.name, + ) + observers.append( + TerseJSONToConsoleLogObserver(observer.location, metadata={}) + ) + + # File drains + elif observer.type == DrainType.FILE: + logger.debug("Starting up the {name} file logger drain", name=observer.name) + log_file = open(observer.location, "at", buffering=1, encoding="utf8") + observers.append(SynapseFileLogObserver(log_file)) + elif observer.type == DrainType.FILE_JSON: + logger.debug( + "Starting up the {name} JSON file logger drain", name=observer.name + ) + log_file = open(observer.location, "at", buffering=1, encoding="utf8") + observers.append(jsonFileLogObserver(log_file)) + + elif observer.type == DrainType.NETWORK_JSON_TERSE: + metadata = {"server_name": hs.config.server_name} + log_observer = TerseJSONToTCPLogObserver( + hs=hs, + host=observer.location[0], + port=observer.location[1], + metadata=metadata, + maximum_buffer=observer.options.maximum_buffer, + ) + log_observer.start() + observers.append(log_observer) + else: + # We should never get here, but, just in case, throw an error. + raise ConfigError("%s drain type cannot be configured" % (observer.type,)) + + publisher = LogPublisher(*observers) + log_filter = LogLevelFilterPredicate() + + for namespace, namespace_config in log_config.get( + "loggers", DEFAULT_LOGGERS + ).items(): + # Set the log level for twisted.logger.Logger namespaces + log_filter.setLogLevelForNamespace( + namespace, + stdlib_log_level_to_twisted(namespace_config.get("level", "INFO")), + ) + + # Also set the log levels for the stdlib logger namespaces, to prevent + # them getting to PythonStdlibToTwistedLogger and having to be formatted + if "level" in namespace_config: + logging.getLogger(namespace).setLevel(namespace_config.get("level")) + + f = FilteringLogObserver(publisher, [log_filter]) + lco = LogContextObserver(f) + + if redirect_stdlib_logging: + stuff_into_twisted = PythonStdlibToTwistedLogger(lco) + stdliblogger = logging.getLogger() + stdliblogger.addHandler(stuff_into_twisted) + + # Always redirect standard I/O, otherwise other logging outputs might miss + # it. + logBeginner.beginLoggingTo([lco], redirectStandardIO=True) + + return publisher + + +def reload_structured_logging(*args, log_config=None) -> None: + warnings.warn( + "Currently the structured logging system can not be reloaded, doing nothing" + ) diff --git a/synapse/logging/_terse_json.py b/synapse/logging/_terse_json.py new file mode 100644 index 0000000000..7f1e8f23fe --- /dev/null +++ b/synapse/logging/_terse_json.py @@ -0,0 +1,278 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +""" +Log formatters that output terse JSON. +""" + +import sys +from collections import deque +from ipaddress import IPv4Address, IPv6Address, ip_address +from math import floor +from typing.io import TextIO + +import attr +from simplejson import dumps + +from twisted.application.internet import ClientService +from twisted.internet.endpoints import ( + HostnameEndpoint, + TCP4ClientEndpoint, + TCP6ClientEndpoint, +) +from twisted.internet.protocol import Factory, Protocol +from twisted.logger import FileLogObserver, Logger +from twisted.python.failure import Failure + + +def flatten_event(event: dict, metadata: dict, include_time: bool = False): + """ + Flatten a Twisted logging event to an dictionary capable of being sent + as a log event to a logging aggregation system. + + The format is vastly simplified and is not designed to be a "human readable + string" in the sense that traditional logs are. Instead, the structure is + optimised for searchability and filtering, with human-understandable log + keys. + + Args: + event (dict): The Twisted logging event we are flattening. + metadata (dict): Additional data to include with each log message. This + can be information like the server name. Since the target log + consumer does not know who we are other than by host IP, this + allows us to forward through static information. + include_time (bool): Should we include the `time` key? If False, the + event time is stripped from the event. + """ + new_event = {} + + # If it's a failure, make the new event's log_failure be the traceback text. + if "log_failure" in event: + new_event["log_failure"] = event["log_failure"].getTraceback() + + # If it's a warning, copy over a string representation of the warning. + if "warning" in event: + new_event["warning"] = str(event["warning"]) + + # Stdlib logging events have "log_text" as their human-readable portion, + # Twisted ones have "log_format". For now, include the log_format, so that + # context only given in the log format (e.g. what is being logged) is + # available. + if "log_text" in event: + new_event["log"] = event["log_text"] + else: + new_event["log"] = event["log_format"] + + # We want to include the timestamp when forwarding over the network, but + # exclude it when we are writing to stdout. This is because the log ingester + # (e.g. logstash, fluentd) can add its own timestamp. + if include_time: + new_event["time"] = round(event["log_time"], 2) + + # Convert the log level to a textual representation. + new_event["level"] = event["log_level"].name.upper() + + # Ignore these keys, and do not transfer them over to the new log object. + # They are either useless (isError), transferred manually above (log_time, + # log_level, etc), or contain Python objects which are not useful for output + # (log_logger, log_source). + keys_to_delete = [ + "isError", + "log_failure", + "log_format", + "log_level", + "log_logger", + "log_source", + "log_system", + "log_time", + "log_text", + "observer", + "warning", + ] + + # If it's from the Twisted legacy logger (twisted.python.log), it adds some + # more keys we want to purge. + if event.get("log_namespace") == "log_legacy": + keys_to_delete.extend(["message", "system", "time"]) + + # Rather than modify the dictionary in place, construct a new one with only + # the content we want. The original event should be considered 'frozen'. + for key in event.keys(): + + if key in keys_to_delete: + continue + + if isinstance(event[key], (str, int, bool, float)) or event[key] is None: + # If it's a plain type, include it as is. + new_event[key] = event[key] + else: + # If it's not one of those basic types, write out a string + # representation. This should probably be a warning in development, + # so that we are sure we are only outputting useful data. + new_event[key] = str(event[key]) + + # Add the metadata information to the event (e.g. the server_name). + new_event.update(metadata) + + return new_event + + +def TerseJSONToConsoleLogObserver(outFile: TextIO, metadata: dict) -> FileLogObserver: + """ + A log observer that formats events to a flattened JSON representation. + + Args: + outFile: The file object to write to. + metadata: Metadata to be added to each log object. + """ + + def formatEvent(_event: dict) -> str: + flattened = flatten_event(_event, metadata) + return dumps(flattened, ensure_ascii=False, separators=(",", ":")) + "\n" + + return FileLogObserver(outFile, formatEvent) + + +@attr.s +class TerseJSONToTCPLogObserver(object): + """ + An IObserver that writes JSON logs to a TCP target. + + Args: + hs (HomeServer): The Homeserver that is being logged for. + host: The host of the logging target. + port: The logging target's port. + metadata: Metadata to be added to each log entry. + """ + + hs = attr.ib() + host = attr.ib(type=str) + port = attr.ib(type=int) + metadata = attr.ib(type=dict) + maximum_buffer = attr.ib(type=int) + _buffer = attr.ib(default=attr.Factory(deque), type=deque) + _writer = attr.ib(default=None) + _logger = attr.ib(default=attr.Factory(Logger)) + + def start(self) -> None: + + # Connect without DNS lookups if it's a direct IP. + try: + ip = ip_address(self.host) + if isinstance(ip, IPv4Address): + endpoint = TCP4ClientEndpoint( + self.hs.get_reactor(), self.host, self.port + ) + elif isinstance(ip, IPv6Address): + endpoint = TCP6ClientEndpoint( + self.hs.get_reactor(), self.host, self.port + ) + except ValueError: + endpoint = HostnameEndpoint(self.hs.get_reactor(), self.host, self.port) + + factory = Factory.forProtocol(Protocol) + self._service = ClientService(endpoint, factory, clock=self.hs.get_reactor()) + self._service.startService() + + def _write_loop(self) -> None: + """ + Implement the write loop. + """ + if self._writer: + return + + self._writer = self._service.whenConnected() + + @self._writer.addBoth + def writer(r): + if isinstance(r, Failure): + r.printTraceback(file=sys.__stderr__) + self._writer = None + self.hs.get_reactor().callLater(1, self._write_loop) + return + + try: + for event in self._buffer: + r.transport.write( + dumps(event, ensure_ascii=False, separators=(",", ":")).encode( + "utf8" + ) + ) + r.transport.write(b"\n") + self._buffer.clear() + except Exception as e: + sys.__stderr__.write("Failed writing out logs with %s\n" % (str(e),)) + + self._writer = False + self.hs.get_reactor().callLater(1, self._write_loop) + + def _handle_pressure(self) -> None: + """ + Handle backpressure by shedding events. + + The buffer will, in this order, until the buffer is below the maximum: + - Shed DEBUG events + - Shed INFO events + - Shed the middle 50% of the events. + """ + if len(self._buffer) <= self.maximum_buffer: + return + + # Strip out DEBUGs + self._buffer = deque( + filter(lambda event: event["level"] != "DEBUG", self._buffer) + ) + + if len(self._buffer) <= self.maximum_buffer: + return + + # Strip out INFOs + self._buffer = deque( + filter(lambda event: event["level"] != "INFO", self._buffer) + ) + + if len(self._buffer) <= self.maximum_buffer: + return + + # Cut the middle entries out + buffer_split = floor(self.maximum_buffer / 2) + + old_buffer = self._buffer + self._buffer = deque() + + for i in range(buffer_split): + self._buffer.append(old_buffer.popleft()) + + end_buffer = [] + for i in range(buffer_split): + end_buffer.append(old_buffer.pop()) + + self._buffer.extend(reversed(end_buffer)) + + def __call__(self, event: dict) -> None: + flattened = flatten_event(event, self.metadata, include_time=True) + self._buffer.append(flattened) + + # Handle backpressure, if it exists. + try: + self._handle_pressure() + except Exception: + # If handling backpressure fails,clear the buffer and log the + # exception. + self._buffer.clear() + self._logger.failure("Failed clearing backpressure") + + # Try and write immediately. + self._write_loop() diff --git a/synapse/logging/context.py b/synapse/logging/context.py index b456c31f70..63379bfb93 100644 --- a/synapse/logging/context.py +++ b/synapse/logging/context.py @@ -25,6 +25,7 @@ See doc/log_contexts.rst for details on how this works. import logging import threading import types +from typing import Any, List from twisted.internet import defer, threads @@ -194,7 +195,7 @@ class LoggingContext(object): class Sentinel(object): """Sentinel to represent the root context""" - __slots__ = [] + __slots__ = [] # type: List[Any] def __str__(self): return "sentinel" @@ -202,6 +203,10 @@ class LoggingContext(object): def copy_to(self, record): pass + def copy_to_twisted_log_entry(self, record): + record["request"] = None + record["scope"] = None + def start(self): pass @@ -330,6 +335,13 @@ class LoggingContext(object): # we also track the current scope: record.scope = self.scope + def copy_to_twisted_log_entry(self, record): + """ + Copy logging fields from this context to a Twisted log record. + """ + record["request"] = self.request + record["scope"] = self.scope + def start(self): if get_thread_id() != self.main_thread: logger.warning("Started logcontext %s on different thread", self) diff --git a/synapse/python_dependencies.py b/synapse/python_dependencies.py index c6465c0386..ec0ac547c1 100644 --- a/synapse/python_dependencies.py +++ b/synapse/python_dependencies.py @@ -47,9 +47,9 @@ REQUIREMENTS = [ "idna>=2.5", # validating SSL certs for IP addresses requires service_identity 18.1. "service_identity>=18.1.0", - # our logcontext handling relies on the ability to cancel inlineCallbacks - # (https://twistedmatrix.com/trac/ticket/4632) which landed in Twisted 18.7. - "Twisted>=18.7.0", + # Twisted 18.9 introduces some logger improvements that the structured + # logger utilises + "Twisted>=18.9.0", "treq>=15.1", # Twisted has required pyopenssl 16.0 since about Twisted 16.6. "pyopenssl>=16.0.0", diff --git a/tests/logging/__init__.py b/tests/logging/__init__.py new file mode 100644 index 0000000000..e69de29bb2 diff --git a/tests/logging/test_structured.py b/tests/logging/test_structured.py new file mode 100644 index 0000000000..a786de0233 --- /dev/null +++ b/tests/logging/test_structured.py @@ -0,0 +1,197 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import os +import os.path +import shutil +import sys +import textwrap + +from twisted.logger import Logger, eventAsText, eventsFromJSONLogFile + +from synapse.config.logger import setup_logging +from synapse.logging._structured import setup_structured_logging +from synapse.logging.context import LoggingContext + +from tests.unittest import DEBUG, HomeserverTestCase + + +class FakeBeginner(object): + def beginLoggingTo(self, observers, **kwargs): + self.observers = observers + + +class StructuredLoggingTestCase(HomeserverTestCase): + """ + Tests for Synapse's structured logging support. + """ + + def test_output_to_json_round_trip(self): + """ + Synapse logs can be outputted to JSON and then read back again. + """ + temp_dir = self.mktemp() + os.mkdir(temp_dir) + self.addCleanup(shutil.rmtree, temp_dir) + + json_log_file = os.path.abspath(os.path.join(temp_dir, "out.json")) + + log_config = { + "drains": {"jsonfile": {"type": "file_json", "location": json_log_file}} + } + + # Begin the logger with our config + beginner = FakeBeginner() + setup_structured_logging( + self.hs, self.hs.config, log_config, logBeginner=beginner + ) + + # Make a logger and send an event + logger = Logger( + namespace="tests.logging.test_structured", observer=beginner.observers[0] + ) + logger.info("Hello there, {name}!", name="wally") + + # Read the log file and check it has the event we sent + with open(json_log_file, "r") as f: + logged_events = list(eventsFromJSONLogFile(f)) + self.assertEqual(len(logged_events), 1) + + # The event pulled from the file should render fine + self.assertEqual( + eventAsText(logged_events[0], includeTimestamp=False), + "[tests.logging.test_structured#info] Hello there, wally!", + ) + + def test_output_to_text(self): + """ + Synapse logs can be outputted to text. + """ + temp_dir = self.mktemp() + os.mkdir(temp_dir) + self.addCleanup(shutil.rmtree, temp_dir) + + log_file = os.path.abspath(os.path.join(temp_dir, "out.log")) + + log_config = {"drains": {"file": {"type": "file", "location": log_file}}} + + # Begin the logger with our config + beginner = FakeBeginner() + setup_structured_logging( + self.hs, self.hs.config, log_config, logBeginner=beginner + ) + + # Make a logger and send an event + logger = Logger( + namespace="tests.logging.test_structured", observer=beginner.observers[0] + ) + logger.info("Hello there, {name}!", name="wally") + + # Read the log file and check it has the event we sent + with open(log_file, "r") as f: + logged_events = f.read().strip().split("\n") + self.assertEqual(len(logged_events), 1) + + # The event pulled from the file should render fine + self.assertTrue( + logged_events[0].endswith( + " - tests.logging.test_structured - INFO - None - Hello there, wally!" + ) + ) + + def test_collects_logcontext(self): + """ + Test that log outputs have the attached logging context. + """ + log_config = {"drains": {}} + + # Begin the logger with our config + beginner = FakeBeginner() + publisher = setup_structured_logging( + self.hs, self.hs.config, log_config, logBeginner=beginner + ) + + logs = [] + + publisher.addObserver(logs.append) + + # Make a logger and send an event + logger = Logger( + namespace="tests.logging.test_structured", observer=beginner.observers[0] + ) + + with LoggingContext("testcontext", request="somereq"): + logger.info("Hello there, {name}!", name="steve") + + self.assertEqual(len(logs), 1) + self.assertEqual(logs[0]["request"], "somereq") + + +class StructuredLoggingConfigurationFileTestCase(HomeserverTestCase): + def make_homeserver(self, reactor, clock): + + tempdir = self.mktemp() + os.mkdir(tempdir) + log_config_file = os.path.abspath(os.path.join(tempdir, "log.config.yaml")) + self.homeserver_log = os.path.abspath(os.path.join(tempdir, "homeserver.log")) + + config = self.default_config() + config["log_config"] = log_config_file + + with open(log_config_file, "w") as f: + f.write( + textwrap.dedent( + """\ + structured: true + + drains: + file: + type: file_json + location: %s + """ + % (self.homeserver_log,) + ) + ) + + self.addCleanup(self._sys_cleanup) + + return self.setup_test_homeserver(config=config) + + def _sys_cleanup(self): + sys.stdout = sys.__stdout__ + sys.stderr = sys.__stderr__ + + # Do not remove! We need the logging system to be set other than WARNING. + @DEBUG + def test_log_output(self): + """ + When a structured logging config is given, Synapse will use it. + """ + setup_logging(self.hs, self.hs.config) + + # Make a logger and send an event + logger = Logger(namespace="tests.logging.test_structured") + + with LoggingContext("testcontext", request="somereq"): + logger.info("Hello there, {name}!", name="steve") + + with open(self.homeserver_log, "r") as f: + logged_events = [ + eventAsText(x, includeTimestamp=False) for x in eventsFromJSONLogFile(f) + ] + + logs = "\n".join(logged_events) + self.assertTrue("***** STARTING SERVER *****" in logs) + self.assertTrue("Hello there, steve!" in logs) diff --git a/tests/logging/test_terse_json.py b/tests/logging/test_terse_json.py new file mode 100644 index 0000000000..514282591d --- /dev/null +++ b/tests/logging/test_terse_json.py @@ -0,0 +1,234 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import json +from collections import Counter + +from twisted.logger import Logger + +from synapse.logging._structured import setup_structured_logging + +from tests.server import connect_client +from tests.unittest import HomeserverTestCase + +from .test_structured import FakeBeginner + + +class TerseJSONTCPTestCase(HomeserverTestCase): + def test_log_output(self): + """ + The Terse JSON outputter delivers simplified structured logs over TCP. + """ + log_config = { + "drains": { + "tersejson": { + "type": "network_json_terse", + "host": "127.0.0.1", + "port": 8000, + } + } + } + + # Begin the logger with our config + beginner = FakeBeginner() + setup_structured_logging( + self.hs, self.hs.config, log_config, logBeginner=beginner + ) + + logger = Logger( + namespace="tests.logging.test_terse_json", observer=beginner.observers[0] + ) + logger.info("Hello there, {name}!", name="wally") + + # Trigger the connection + self.pump() + + _, server = connect_client(self.reactor, 0) + + # Trigger data being sent + self.pump() + + # One log message, with a single trailing newline + logs = server.data.decode("utf8").splitlines() + self.assertEqual(len(logs), 1) + self.assertEqual(server.data.count(b"\n"), 1) + + log = json.loads(logs[0]) + + # The terse logger should give us these keys. + expected_log_keys = [ + "log", + "time", + "level", + "log_namespace", + "request", + "scope", + "server_name", + "name", + ] + self.assertEqual(set(log.keys()), set(expected_log_keys)) + + # It contains the data we expect. + self.assertEqual(log["name"], "wally") + + def test_log_backpressure_debug(self): + """ + When backpressure is hit, DEBUG logs will be shed. + """ + log_config = { + "loggers": {"synapse": {"level": "DEBUG"}}, + "drains": { + "tersejson": { + "type": "network_json_terse", + "host": "127.0.0.1", + "port": 8000, + "maximum_buffer": 10, + } + }, + } + + # Begin the logger with our config + beginner = FakeBeginner() + setup_structured_logging( + self.hs, + self.hs.config, + log_config, + logBeginner=beginner, + redirect_stdlib_logging=False, + ) + + logger = Logger( + namespace="synapse.logging.test_terse_json", observer=beginner.observers[0] + ) + + # Send some debug messages + for i in range(0, 3): + logger.debug("debug %s" % (i,)) + + # Send a bunch of useful messages + for i in range(0, 7): + logger.info("test message %s" % (i,)) + + # The last debug message pushes it past the maximum buffer + logger.debug("too much debug") + + # Allow the reconnection + _, server = connect_client(self.reactor, 0) + self.pump() + + # Only the 7 infos made it through, the debugs were elided + logs = server.data.splitlines() + self.assertEqual(len(logs), 7) + + def test_log_backpressure_info(self): + """ + When backpressure is hit, DEBUG and INFO logs will be shed. + """ + log_config = { + "loggers": {"synapse": {"level": "DEBUG"}}, + "drains": { + "tersejson": { + "type": "network_json_terse", + "host": "127.0.0.1", + "port": 8000, + "maximum_buffer": 10, + } + }, + } + + # Begin the logger with our config + beginner = FakeBeginner() + setup_structured_logging( + self.hs, + self.hs.config, + log_config, + logBeginner=beginner, + redirect_stdlib_logging=False, + ) + + logger = Logger( + namespace="synapse.logging.test_terse_json", observer=beginner.observers[0] + ) + + # Send some debug messages + for i in range(0, 3): + logger.debug("debug %s" % (i,)) + + # Send a bunch of useful messages + for i in range(0, 10): + logger.warn("test warn %s" % (i,)) + + # Send a bunch of info messages + for i in range(0, 3): + logger.info("test message %s" % (i,)) + + # The last debug message pushes it past the maximum buffer + logger.debug("too much debug") + + # Allow the reconnection + client, server = connect_client(self.reactor, 0) + self.pump() + + # The 10 warnings made it through, the debugs and infos were elided + logs = list(map(json.loads, server.data.decode("utf8").splitlines())) + self.assertEqual(len(logs), 10) + + self.assertEqual(Counter([x["level"] for x in logs]), {"WARN": 10}) + + def test_log_backpressure_cut_middle(self): + """ + When backpressure is hit, and no more DEBUG and INFOs cannot be culled, + it will cut the middle messages out. + """ + log_config = { + "loggers": {"synapse": {"level": "DEBUG"}}, + "drains": { + "tersejson": { + "type": "network_json_terse", + "host": "127.0.0.1", + "port": 8000, + "maximum_buffer": 10, + } + }, + } + + # Begin the logger with our config + beginner = FakeBeginner() + setup_structured_logging( + self.hs, + self.hs.config, + log_config, + logBeginner=beginner, + redirect_stdlib_logging=False, + ) + + logger = Logger( + namespace="synapse.logging.test_terse_json", observer=beginner.observers[0] + ) + + # Send a bunch of useful messages + for i in range(0, 20): + logger.warn("test warn", num=i) + + # Allow the reconnection + client, server = connect_client(self.reactor, 0) + self.pump() + + # The first five and last five warnings made it through, the debugs and + # infos were elided + logs = list(map(json.loads, server.data.decode("utf8").splitlines())) + self.assertEqual(len(logs), 10) + self.assertEqual(Counter([x["level"] for x in logs]), {"WARN": 10}) + self.assertEqual([0, 1, 2, 3, 4, 15, 16, 17, 18, 19], [x["num"] for x in logs]) diff --git a/tests/server.py b/tests/server.py index e573c4e4c5..c8269619b1 100644 --- a/tests/server.py +++ b/tests/server.py @@ -11,9 +11,13 @@ from twisted.internet import address, threads, udp from twisted.internet._resolver import SimpleResolverComplexifier from twisted.internet.defer import Deferred, fail, succeed from twisted.internet.error import DNSLookupError -from twisted.internet.interfaces import IReactorPluggableNameResolver, IResolverSimple +from twisted.internet.interfaces import ( + IReactorPluggableNameResolver, + IReactorTCP, + IResolverSimple, +) from twisted.python.failure import Failure -from twisted.test.proto_helpers import MemoryReactorClock +from twisted.test.proto_helpers import AccumulatingProtocol, MemoryReactorClock from twisted.web.http import unquote from twisted.web.http_headers import Headers @@ -465,3 +469,22 @@ class FakeTransport(object): self.buffer = self.buffer[len(to_write) :] if self.buffer and self.autoflush: self._reactor.callLater(0.0, self.flush) + + +def connect_client(reactor: IReactorTCP, client_id: int) -> AccumulatingProtocol: + """ + Connect a client to a fake TCP transport. + + Args: + reactor + factory: The connecting factory to build. + """ + factory = reactor.tcpClients[client_id][2] + client = factory.buildProtocol(None) + server = AccumulatingProtocol() + server.makeConnection(FakeTransport(client, reactor)) + client.makeConnection(FakeTransport(server, reactor)) + + reactor.tcpClients.pop(client_id) + + return client, server diff --git a/tox.ini b/tox.ini index 09b4b8fc3c..f9a3b7e49a 100644 --- a/tox.ini +++ b/tox.ini @@ -146,3 +146,13 @@ commands = coverage combine coverage xml codecov -X gcov + +[testenv:mypy] +basepython = python3.5 +deps = + {[base]deps} + mypy +extras = all +commands = mypy --ignore-missing-imports \ + synapse/logging/_structured.py \ + synapse/logging/_terse_json.py \ No newline at end of file -- cgit 1.5.1 From 6d97843793d59bc5d307475a6a6185ff107e116b Mon Sep 17 00:00:00 2001 From: Jorik Schellekens Date: Wed, 28 Aug 2019 13:12:22 +0100 Subject: Config templating (#5900) Template config files * Imagine a system composed entirely of x, y, z etc and the basic operations.. Wait George, why XOR? Why not just neq? George: Eh, I didn't think of that.. Co-Authored-By: Erik Johnston --- changelog.d/5900.feature | 1 + docs/sample_config.yaml | 16 +++---- synapse/config/_base.py | 37 ++++++++++++++++ synapse/config/database.py | 27 +++++++---- synapse/config/server.py | 84 ++++++++++++++++++++++++++++------- synapse/config/tls.py | 50 ++++++++++++++++----- tests/config/test_database.py | 52 ++++++++++++++++++++++ tests/config/test_server.py | 101 +++++++++++++++++++++++++++++++++++++++++- tests/config/test_tls.py | 44 ++++++++++++++++++ 9 files changed, 366 insertions(+), 46 deletions(-) create mode 100644 changelog.d/5900.feature create mode 100644 tests/config/test_database.py (limited to 'tests') diff --git a/changelog.d/5900.feature b/changelog.d/5900.feature new file mode 100644 index 0000000000..b62d88a76b --- /dev/null +++ b/changelog.d/5900.feature @@ -0,0 +1 @@ +Add support for config templating. diff --git a/docs/sample_config.yaml b/docs/sample_config.yaml index ae1cafc5f3..6da1167632 100644 --- a/docs/sample_config.yaml +++ b/docs/sample_config.yaml @@ -205,9 +205,9 @@ listeners: # - port: 8008 tls: false - bind_addresses: ['::1', '127.0.0.1'] type: http x_forwarded: true + bind_addresses: ['::1', '127.0.0.1'] resources: - names: [client, federation] @@ -392,10 +392,10 @@ listeners: # permission to listen on port 80. # acme: - # ACME support is disabled by default. Uncomment the following line - # (and tls_certificate_path and tls_private_key_path above) to enable it. + # ACME support is disabled by default. Set this to `true` and uncomment + # tls_certificate_path and tls_private_key_path above to enable it. # - #enabled: true + enabled: False # Endpoint to use to request certificates. If you only want to test, # use Let's Encrypt's staging url: @@ -406,17 +406,17 @@ acme: # Port number to listen on for the HTTP-01 challenge. Change this if # you are forwarding connections through Apache/Nginx/etc. # - #port: 80 + port: 80 # Local addresses to listen on for incoming connections. # Again, you may want to change this if you are forwarding connections # through Apache/Nginx/etc. # - #bind_addresses: ['::', '0.0.0.0'] + bind_addresses: ['::', '0.0.0.0'] # How many days remaining on a certificate before it is renewed. # - #reprovision_threshold: 30 + reprovision_threshold: 30 # The domain that the certificate should be for. Normally this # should be the same as your Matrix domain (i.e., 'server_name'), but, @@ -430,7 +430,7 @@ acme: # # If not set, defaults to your 'server_name'. # - #domain: matrix.example.com + domain: matrix.example.com # file to use for the account key. This will be generated if it doesn't # exist. diff --git a/synapse/config/_base.py b/synapse/config/_base.py index 6ce5cd07fb..31f6530978 100644 --- a/synapse/config/_base.py +++ b/synapse/config/_base.py @@ -181,6 +181,11 @@ class Config(object): generate_secrets=False, report_stats=None, open_private_ports=False, + listeners=None, + database_conf=None, + tls_certificate_path=None, + tls_private_key_path=None, + acme_domain=None, ): """Build a default configuration file @@ -207,6 +212,33 @@ class Config(object): open_private_ports (bool): True to leave private ports (such as the non-TLS HTTP listener) open to the internet. + listeners (list(dict)|None): A list of descriptions of the listeners + synapse should start with each of which specifies a port (str), a list of + resources (list(str)), tls (bool) and type (str). For example: + [{ + "port": 8448, + "resources": [{"names": ["federation"]}], + "tls": True, + "type": "http", + }, + { + "port": 443, + "resources": [{"names": ["client"]}], + "tls": False, + "type": "http", + }], + + + database (str|None): The database type to configure, either `psycog2` + or `sqlite3`. + + tls_certificate_path (str|None): The path to the tls certificate. + + tls_private_key_path (str|None): The path to the tls private key. + + acme_domain (str|None): The domain acme will try to validate. If + specified acme will be enabled. + Returns: str: the yaml config file """ @@ -220,6 +252,11 @@ class Config(object): generate_secrets=generate_secrets, report_stats=report_stats, open_private_ports=open_private_ports, + listeners=listeners, + database_conf=database_conf, + tls_certificate_path=tls_certificate_path, + tls_private_key_path=tls_private_key_path, + acme_domain=acme_domain, ) ) diff --git a/synapse/config/database.py b/synapse/config/database.py index 746a6cd1f4..118aafbd4a 100644 --- a/synapse/config/database.py +++ b/synapse/config/database.py @@ -13,6 +13,9 @@ # See the License for the specific language governing permissions and # limitations under the License. import os +from textwrap import indent + +import yaml from ._base import Config @@ -38,20 +41,28 @@ class DatabaseConfig(Config): self.set_databasepath(config.get("database_path")) - def generate_config_section(self, data_dir_path, **kwargs): - database_path = os.path.join(data_dir_path, "homeserver.db") - return ( - """\ - ## Database ## - - database: - # The database engine name + def generate_config_section(self, data_dir_path, database_conf, **kwargs): + if not database_conf: + database_path = os.path.join(data_dir_path, "homeserver.db") + database_conf = ( + """# The database engine name name: "sqlite3" # Arguments to pass to the engine args: # Path to the database database: "%(database_path)s" + """ + % locals() + ) + else: + database_conf = indent(yaml.dump(database_conf), " " * 10).lstrip() + + return ( + """\ + ## Database ## + database: + %(database_conf)s # Number of events to cache in memory. # #event_cache_size: 10K diff --git a/synapse/config/server.py b/synapse/config/server.py index 15449695d1..2abdef0971 100644 --- a/synapse/config/server.py +++ b/synapse/config/server.py @@ -17,8 +17,11 @@ import logging import os.path +import re +from textwrap import indent import attr +import yaml from netaddr import IPSet from synapse.api.room_versions import KNOWN_ROOM_VERSIONS @@ -352,7 +355,7 @@ class ServerConfig(Config): return any(l["tls"] for l in self.listeners) def generate_config_section( - self, server_name, data_dir_path, open_private_ports, **kwargs + self, server_name, data_dir_path, open_private_ports, listeners, **kwargs ): _, bind_port = parse_and_validate_server_name(server_name) if bind_port is not None: @@ -366,11 +369,68 @@ class ServerConfig(Config): # Bring DEFAULT_ROOM_VERSION into the local-scope for use in the # default config string default_room_version = DEFAULT_ROOM_VERSION + secure_listeners = [] + unsecure_listeners = [] + private_addresses = ["::1", "127.0.0.1"] + if listeners: + for listener in listeners: + if listener["tls"]: + secure_listeners.append(listener) + else: + # If we don't want open ports we need to bind the listeners + # to some address other than 0.0.0.0. Here we chose to use + # localhost. + # If the addresses are already bound we won't overwrite them + # however. + if not open_private_ports: + listener.setdefault("bind_addresses", private_addresses) + + unsecure_listeners.append(listener) + + secure_http_bindings = indent( + yaml.dump(secure_listeners), " " * 10 + ).lstrip() + + unsecure_http_bindings = indent( + yaml.dump(unsecure_listeners), " " * 10 + ).lstrip() + + if not unsecure_listeners: + unsecure_http_bindings = ( + """- port: %(unsecure_port)s + tls: false + type: http + x_forwarded: true""" + % locals() + ) + + if not open_private_ports: + unsecure_http_bindings += ( + "\n bind_addresses: ['::1', '127.0.0.1']" + ) + + unsecure_http_bindings += """ + + resources: + - names: [client, federation] + compress: false""" + + if listeners: + # comment out this block + unsecure_http_bindings = "#" + re.sub( + "\n {10}", + lambda match: match.group(0) + "#", + unsecure_http_bindings, + ) - unsecure_http_binding = "port: %i\n tls: false" % (unsecure_port,) - if not open_private_ports: - unsecure_http_binding += ( - "\n bind_addresses: ['::1', '127.0.0.1']" + if not secure_listeners: + secure_http_bindings = ( + """#- port: %(bind_port)s + # type: http + # tls: true + # resources: + # - names: [client, federation]""" + % locals() ) return ( @@ -556,11 +616,7 @@ class ServerConfig(Config): # will also need to give Synapse a TLS key and certificate: see the TLS section # below.) # - #- port: %(bind_port)s - # type: http - # tls: true - # resources: - # - names: [client, federation] + %(secure_http_bindings)s # Unsecure HTTP listener: for when matrix traffic passes through a reverse proxy # that unwraps TLS. @@ -568,13 +624,7 @@ class ServerConfig(Config): # If you plan to use a reverse proxy, please see # https://github.com/matrix-org/synapse/blob/master/docs/reverse_proxy.rst. # - - %(unsecure_http_binding)s - type: http - x_forwarded: true - - resources: - - names: [client, federation] - compress: false + %(unsecure_http_bindings)s # example additional_resources: # diff --git a/synapse/config/tls.py b/synapse/config/tls.py index ca508a224f..c0148aa95c 100644 --- a/synapse/config/tls.py +++ b/synapse/config/tls.py @@ -239,12 +239,38 @@ class TlsConfig(Config): self.tls_fingerprints.append({"sha256": sha256_fingerprint}) def generate_config_section( - self, config_dir_path, server_name, data_dir_path, **kwargs + self, + config_dir_path, + server_name, + data_dir_path, + tls_certificate_path, + tls_private_key_path, + acme_domain, + **kwargs ): + """If the acme_domain is specified acme will be enabled. + If the TLS paths are not specified the default will be certs in the + config directory""" + base_key_name = os.path.join(config_dir_path, server_name) - tls_certificate_path = base_key_name + ".tls.crt" - tls_private_key_path = base_key_name + ".tls.key" + if bool(tls_certificate_path) != bool(tls_private_key_path): + raise ConfigError( + "Please specify both a cert path and a key path or neither." + ) + + tls_enabled = ( + "" if tls_certificate_path and tls_private_key_path or acme_domain else "#" + ) + + if not tls_certificate_path: + tls_certificate_path = base_key_name + ".tls.crt" + if not tls_private_key_path: + tls_private_key_path = base_key_name + ".tls.key" + + acme_enabled = bool(acme_domain) + acme_domain = "matrix.example.com" + default_acme_account_file = os.path.join(data_dir_path, "acme_account.key") # this is to avoid the max line length. Sorrynotsorry @@ -269,11 +295,11 @@ class TlsConfig(Config): # instance, if using certbot, use `fullchain.pem` as your certificate, # not `cert.pem`). # - #tls_certificate_path: "%(tls_certificate_path)s" + %(tls_enabled)stls_certificate_path: "%(tls_certificate_path)s" # PEM-encoded private key for TLS # - #tls_private_key_path: "%(tls_private_key_path)s" + %(tls_enabled)stls_private_key_path: "%(tls_private_key_path)s" # Whether to verify TLS server certificates for outbound federation requests. # @@ -340,10 +366,10 @@ class TlsConfig(Config): # permission to listen on port 80. # acme: - # ACME support is disabled by default. Uncomment the following line - # (and tls_certificate_path and tls_private_key_path above) to enable it. + # ACME support is disabled by default. Set this to `true` and uncomment + # tls_certificate_path and tls_private_key_path above to enable it. # - #enabled: true + enabled: %(acme_enabled)s # Endpoint to use to request certificates. If you only want to test, # use Let's Encrypt's staging url: @@ -354,17 +380,17 @@ class TlsConfig(Config): # Port number to listen on for the HTTP-01 challenge. Change this if # you are forwarding connections through Apache/Nginx/etc. # - #port: 80 + port: 80 # Local addresses to listen on for incoming connections. # Again, you may want to change this if you are forwarding connections # through Apache/Nginx/etc. # - #bind_addresses: ['::', '0.0.0.0'] + bind_addresses: ['::', '0.0.0.0'] # How many days remaining on a certificate before it is renewed. # - #reprovision_threshold: 30 + reprovision_threshold: 30 # The domain that the certificate should be for. Normally this # should be the same as your Matrix domain (i.e., 'server_name'), but, @@ -378,7 +404,7 @@ class TlsConfig(Config): # # If not set, defaults to your 'server_name'. # - #domain: matrix.example.com + domain: %(acme_domain)s # file to use for the account key. This will be generated if it doesn't # exist. diff --git a/tests/config/test_database.py b/tests/config/test_database.py new file mode 100644 index 0000000000..151d3006ac --- /dev/null +++ b/tests/config/test_database.py @@ -0,0 +1,52 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 New Vector Ltd +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import yaml + +from synapse.config.database import DatabaseConfig + +from tests import unittest + + +class DatabaseConfigTestCase(unittest.TestCase): + def test_database_configured_correctly_no_database_conf_param(self): + conf = yaml.safe_load( + DatabaseConfig().generate_config_section("/data_dir_path", None) + ) + + expected_database_conf = { + "name": "sqlite3", + "args": {"database": "/data_dir_path/homeserver.db"}, + } + + self.assertEqual(conf["database"], expected_database_conf) + + def test_database_configured_correctly_database_conf_param(self): + + database_conf = { + "name": "my super fast datastore", + "args": { + "user": "matrix", + "password": "synapse_database_password", + "host": "synapse_database_host", + "database": "matrix", + }, + } + + conf = yaml.safe_load( + DatabaseConfig().generate_config_section("/data_dir_path", database_conf) + ) + + self.assertEqual(conf["database"], database_conf) diff --git a/tests/config/test_server.py b/tests/config/test_server.py index 1ca5ea54ca..a10d017120 100644 --- a/tests/config/test_server.py +++ b/tests/config/test_server.py @@ -13,7 +13,9 @@ # See the License for the specific language governing permissions and # limitations under the License. -from synapse.config.server import is_threepid_reserved +import yaml + +from synapse.config.server import ServerConfig, is_threepid_reserved from tests import unittest @@ -29,3 +31,100 @@ class ServerConfigTestCase(unittest.TestCase): self.assertTrue(is_threepid_reserved(config, user1)) self.assertFalse(is_threepid_reserved(config, user3)) self.assertFalse(is_threepid_reserved(config, user1_msisdn)) + + def test_unsecure_listener_no_listeners_open_private_ports_false(self): + conf = yaml.safe_load( + ServerConfig().generate_config_section( + "che.org", "/data_dir_path", False, None + ) + ) + + expected_listeners = [ + { + "port": 8008, + "tls": False, + "type": "http", + "x_forwarded": True, + "bind_addresses": ["::1", "127.0.0.1"], + "resources": [{"names": ["client", "federation"], "compress": False}], + } + ] + + self.assertEqual(conf["listeners"], expected_listeners) + + def test_unsecure_listener_no_listeners_open_private_ports_true(self): + conf = yaml.safe_load( + ServerConfig().generate_config_section( + "che.org", "/data_dir_path", True, None + ) + ) + + expected_listeners = [ + { + "port": 8008, + "tls": False, + "type": "http", + "x_forwarded": True, + "resources": [{"names": ["client", "federation"], "compress": False}], + } + ] + + self.assertEqual(conf["listeners"], expected_listeners) + + def test_listeners_set_correctly_open_private_ports_false(self): + listeners = [ + { + "port": 8448, + "resources": [{"names": ["federation"]}], + "tls": True, + "type": "http", + }, + { + "port": 443, + "resources": [{"names": ["client"]}], + "tls": False, + "type": "http", + }, + ] + + conf = yaml.safe_load( + ServerConfig().generate_config_section( + "this.one.listens", "/data_dir_path", True, listeners + ) + ) + + self.assertEqual(conf["listeners"], listeners) + + def test_listeners_set_correctly_open_private_ports_true(self): + listeners = [ + { + "port": 8448, + "resources": [{"names": ["federation"]}], + "tls": True, + "type": "http", + }, + { + "port": 443, + "resources": [{"names": ["client"]}], + "tls": False, + "type": "http", + }, + { + "port": 1243, + "resources": [{"names": ["client"]}], + "tls": False, + "type": "http", + "bind_addresses": ["this_one_is_bound"], + }, + ] + + expected_listeners = listeners.copy() + expected_listeners[1]["bind_addresses"] = ["::1", "127.0.0.1"] + + conf = yaml.safe_load( + ServerConfig().generate_config_section( + "this.one.listens", "/data_dir_path", True, listeners + ) + ) + + self.assertEqual(conf["listeners"], expected_listeners) diff --git a/tests/config/test_tls.py b/tests/config/test_tls.py index 4f8a87a3df..8e0c4b9533 100644 --- a/tests/config/test_tls.py +++ b/tests/config/test_tls.py @@ -16,6 +16,8 @@ import os +import yaml + from OpenSSL import SSL from synapse.config.tls import ConfigError, TlsConfig @@ -191,3 +193,45 @@ s4niecZKPBizL6aucT59CsunNmmb5Glq8rlAcU+1ZTZZzGYqVYhF6axB9Qg= self.assertEqual(cf._verify_ssl._options & SSL.OP_NO_TLSv1, 0) self.assertEqual(cf._verify_ssl._options & SSL.OP_NO_TLSv1_1, 0) self.assertEqual(cf._verify_ssl._options & SSL.OP_NO_TLSv1_2, 0) + + def test_acme_disabled_in_generated_config_no_acme_domain_provied(self): + """ + Checks acme is disabled by default. + """ + conf = TestConfig() + conf.read_config( + yaml.safe_load( + TestConfig().generate_config_section( + "/config_dir_path", + "my_super_secure_server", + "/data_dir_path", + "/tls_cert_path", + "tls_private_key", + None, # This is the acme_domain + ) + ), + "/config_dir_path", + ) + + self.assertFalse(conf.acme_enabled) + + def test_acme_enabled_in_generated_config_domain_provided(self): + """ + Checks acme is enabled if the acme_domain arg is set to some string. + """ + conf = TestConfig() + conf.read_config( + yaml.safe_load( + TestConfig().generate_config_section( + "/config_dir_path", + "my_super_secure_server", + "/data_dir_path", + "/tls_cert_path", + "tls_private_key", + "my_supe_secure_server", # This is the acme_domain + ) + ), + "/config_dir_path", + ) + + self.assertTrue(conf.acme_enabled) -- cgit 1.5.1 From a4bf72c30c5953b721a64eae89db186fa8735bb3 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Thu, 29 Aug 2019 17:38:51 +0100 Subject: Censor redactions in DB after a month --- synapse/storage/events.py | 88 +++++++++++++++++++++- .../storage/schema/delta/56/redaction_censor.sql | 17 +++++ tests/storage/test_redaction.py | 71 +++++++++++++++++ 3 files changed, 175 insertions(+), 1 deletion(-) create mode 100644 synapse/storage/schema/delta/56/redaction_censor.sql (limited to 'tests') diff --git a/synapse/storage/events.py b/synapse/storage/events.py index 5a95c36a8b..2970da6829 100644 --- a/synapse/storage/events.py +++ b/synapse/storage/events.py @@ -23,7 +23,7 @@ from functools import wraps from six import iteritems, text_type from six.moves import range -from canonicaljson import json +from canonicaljson import encode_canonical_json, json from prometheus_client import Counter, Histogram from twisted.internet import defer @@ -33,6 +33,7 @@ from synapse.api.constants import EventTypes from synapse.api.errors import SynapseError from synapse.events import EventBase # noqa: F401 from synapse.events.snapshot import EventContext # noqa: F401 +from synapse.events.utils import prune_event_dict from synapse.logging.context import PreserveLoggingContext, make_deferred_yieldable from synapse.logging.utils import log_function from synapse.metrics import BucketCollector @@ -262,6 +263,13 @@ class EventsStore( hs.get_clock().looping_call(read_forward_extremities, 60 * 60 * 1000) + def _censor_redactions(): + return run_as_background_process( + "_censor_redactions", self._censor_redactions + ) + + hs.get_clock().looping_call(_censor_redactions, 10 * 60 * 1000) + @defer.inlineCallbacks def _read_forward_extremities(self): def fetch(txn): @@ -1548,6 +1556,84 @@ class EventsStore( (event.event_id, event.redacts), ) + @defer.inlineCallbacks + def _censor_redactions(self): + """Censors all redactions older than a month that haven't been censored. + + By censor we mean update the event_json table with the redacted event. + + Returns: + Deferred + """ + + if self.stream_ordering_month_ago is None: + return + + max_pos = self.stream_ordering_month_ago + + # We fetch all redactions that point to an event that we have that has + # a stream ordering from over a month ago, that we haven't yet censored + # in the DB. + sql = """ + SELECT er.event_id, redacts FROM redactions + INNER JOIN events AS er USING (event_id) + INNER JOIN events AS eb ON (er.room_id = eb.room_id AND redacts = eb.event_id) + WHERE NOT have_censored + AND ? <= er.stream_ordering AND er.stream_ordering <= ? + ORDER BY er.stream_ordering ASC + LIMIT ? + """ + + rows = yield self._execute( + "_censor_redactions_fetch", None, sql, -max_pos, max_pos, 100 + ) + + updates = [] + + for redaction_id, event_id in rows: + redaction_event = yield self.get_event(redaction_id, allow_none=True) + original_event = yield self.get_event( + event_id, allow_rejected=True, allow_none=True + ) + + # The SQL above ensures that we have both the redaction and + # original event, so if the `get_event` calls return None it + # means that the redaction wasn't allowed. Either way we know that + # the result won't change so we mark the fact that we've checked. + if ( + redaction_event + and original_event + and original_event.internal_metadata.is_redacted() + ): + # Redaction was allowed + pruned_json = encode_canonical_json( + prune_event_dict(original_event.get_dict()) + ) + else: + # Redaction wasn't allowed + pruned_json = None + + updates.append((redaction_id, event_id, pruned_json)) + + def _update_censor_txn(txn): + for redaction_id, event_id, pruned_json in updates: + if pruned_json: + self._simple_update_one_txn( + txn, + table="event_json", + keyvalues={"event_id": event_id}, + updatevalues={"json": pruned_json}, + ) + + self._simple_update_one_txn( + txn, + table="redactions", + keyvalues={"event_id": redaction_id}, + updatevalues={"have_censored": True}, + ) + + yield self.runInteraction("_update_censor_txn", _update_censor_txn) + @defer.inlineCallbacks def count_daily_messages(self): """ diff --git a/synapse/storage/schema/delta/56/redaction_censor.sql b/synapse/storage/schema/delta/56/redaction_censor.sql new file mode 100644 index 0000000000..fe51b02309 --- /dev/null +++ b/synapse/storage/schema/delta/56/redaction_censor.sql @@ -0,0 +1,17 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C. + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +ALTER TABLE redactions ADD COLUMN have_censored BOOL NOT NULL DEFAULT false; +CREATE INDEX redactions_have_censored ON redactions(event_id) WHERE not have_censored; diff --git a/tests/storage/test_redaction.py b/tests/storage/test_redaction.py index d961b81d48..0c9f3c7071 100644 --- a/tests/storage/test_redaction.py +++ b/tests/storage/test_redaction.py @@ -17,6 +17,8 @@ from mock import Mock +from canonicaljson import json + from twisted.internet import defer from synapse.api.constants import EventTypes, Membership @@ -286,3 +288,72 @@ class RedactionTestCase(unittest.HomeserverTestCase): self.assertEqual( fetched.unsigned["redacted_because"].event_id, redaction_event_id2 ) + + def test_redact_censor(self): + """Test that a redacted event gets censored in the DB after a month + """ + + self.get_success( + self.inject_room_member(self.room1, self.u_alice, Membership.JOIN) + ) + + msg_event = self.get_success(self.inject_message(self.room1, self.u_alice, "t")) + + # Check event has not been redacted: + event = self.get_success(self.store.get_event(msg_event.event_id)) + + self.assertObjectHasAttributes( + { + "type": EventTypes.Message, + "user_id": self.u_alice.to_string(), + "content": {"body": "t", "msgtype": "message"}, + }, + event, + ) + + self.assertFalse("redacted_because" in event.unsigned) + + # Redact event + reason = "Because I said so" + self.get_success( + self.inject_redaction(self.room1, msg_event.event_id, self.u_alice, reason) + ) + + event = self.get_success(self.store.get_event(msg_event.event_id)) + + self.assertTrue("redacted_because" in event.unsigned) + + self.assertObjectHasAttributes( + { + "type": EventTypes.Message, + "user_id": self.u_alice.to_string(), + "content": {}, + }, + event, + ) + + event_json = self.get_success( + self.store._simple_select_one_onecol( + table="event_json", + keyvalues={"event_id": msg_event.event_id}, + retcol="json", + ) + ) + + self.assert_dict( + {"content": {"body": "t", "msgtype": "message"}}, json.loads(event_json) + ) + + # Advance by 30 days + self.reactor.advance(60 * 60 * 24 * 31) + self.reactor.advance(60 * 60 * 2) + + event_json = self.get_success( + self.store._simple_select_one_onecol( + table="event_json", + keyvalues={"event_id": msg_event.event_id}, + retcol="json", + ) + ) + + self.assert_dict({"content": {}}, json.loads(event_json)) -- cgit 1.5.1 From 4548d1f87e3ff3dc24b0af8f944276137d3228e3 Mon Sep 17 00:00:00 2001 From: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com> Date: Fri, 30 Aug 2019 16:28:26 +0100 Subject: Remove unnecessary parentheses around return statements (#5931) Python will return a tuple whether there are parentheses around the returned values or not. I'm just sick of my editor complaining about this all over the place :) --- changelog.d/5931.misc | 1 + synapse/api/auth.py | 14 ++--- synapse/app/frontend_proxy.py | 8 +-- synapse/crypto/event_signing.py | 4 +- synapse/federation/federation_client.py | 8 +-- synapse/federation/federation_server.py | 22 ++++---- synapse/handlers/account_data.py | 4 +- synapse/handlers/auth.py | 8 +-- synapse/handlers/federation.py | 2 +- synapse/handlers/initial_sync.py | 4 +- synapse/handlers/presence.py | 4 +- synapse/handlers/receipts.py | 2 +- synapse/handlers/register.py | 2 +- synapse/handlers/room_member.py | 2 +- synapse/handlers/sync.py | 8 +-- synapse/handlers/typing.py | 2 +- synapse/http/federation/well_known_resolver.py | 2 +- synapse/module_api/__init__.py | 2 +- synapse/notifier.py | 6 +- synapse/push/bulk_push_rule_evaluator.py | 2 +- synapse/replication/http/federation.py | 8 +-- synapse/replication/http/login.py | 2 +- synapse/replication/http/membership.py | 6 +- synapse/replication/http/register.py | 4 +- synapse/replication/http/send_event.py | 2 +- synapse/replication/tcp/streams/_base.py | 8 +-- synapse/rest/admin/__init__.py | 26 ++++----- synapse/rest/admin/media.py | 6 +- synapse/rest/admin/purge_room_servlet.py | 2 +- synapse/rest/admin/server_notice_servlet.py | 2 +- synapse/rest/admin/users.py | 4 +- synapse/rest/client/v1/directory.py | 16 +++--- synapse/rest/client/v1/events.py | 8 +-- synapse/rest/client/v1/initial_sync.py | 2 +- synapse/rest/client/v1/login.py | 6 +- synapse/rest/client/v1/logout.py | 8 +-- synapse/rest/client/v1/presence.py | 6 +- synapse/rest/client/v1/profile.py | 18 +++--- synapse/rest/client/v1/push_rule.py | 10 ++-- synapse/rest/client/v1/pusher.py | 6 +- synapse/rest/client/v1/room.py | 48 ++++++++-------- synapse/rest/client/v1/voip.py | 4 +- synapse/rest/client/v2_alpha/account.py | 24 ++++---- synapse/rest/client/v2_alpha/account_data.py | 8 +-- synapse/rest/client/v2_alpha/capabilities.py | 2 +- synapse/rest/client/v2_alpha/devices.py | 10 ++-- synapse/rest/client/v2_alpha/filter.py | 4 +- synapse/rest/client/v2_alpha/groups.py | 64 +++++++++++----------- synapse/rest/client/v2_alpha/keys.py | 8 +-- synapse/rest/client/v2_alpha/notifications.py | 2 +- synapse/rest/client/v2_alpha/read_marker.py | 2 +- synapse/rest/client/v2_alpha/receipts.py | 2 +- synapse/rest/client/v2_alpha/register.py | 10 ++-- synapse/rest/client/v2_alpha/relations.py | 8 +-- synapse/rest/client/v2_alpha/report_event.py | 2 +- synapse/rest/client/v2_alpha/room_keys.py | 14 ++--- .../client/v2_alpha/room_upgrade_rest_servlet.py | 2 +- synapse/rest/client/v2_alpha/sync.py | 2 +- synapse/rest/client/v2_alpha/tags.py | 6 +- synapse/rest/client/v2_alpha/thirdparty.py | 10 ++-- synapse/rest/client/v2_alpha/user_directory.py | 4 +- synapse/rest/media/v1/media_repository.py | 4 +- synapse/rest/media/v1/thumbnailer.py | 4 +- .../resource_limits_server_notices.py | 2 +- synapse/storage/account_data.py | 8 +-- synapse/storage/appservice.py | 2 +- synapse/storage/deviceinbox.py | 4 +- synapse/storage/devices.py | 10 ++-- synapse/storage/events.py | 10 ++-- synapse/storage/presence.py | 2 +- synapse/storage/pusher.py | 2 +- synapse/storage/receipts.py | 2 +- synapse/storage/stream.py | 12 ++-- synapse/storage/util/id_generators.py | 4 +- synapse/streams/config.py | 2 +- tests/handlers/test_register.py | 2 +- tests/rest/client/v2_alpha/test_register.py | 2 +- tests/server.py | 2 +- tests/test_server.py | 2 +- tests/test_state.py | 2 +- tests/utils.py | 2 +- 81 files changed, 287 insertions(+), 286 deletions(-) create mode 100644 changelog.d/5931.misc (limited to 'tests') diff --git a/changelog.d/5931.misc b/changelog.d/5931.misc new file mode 100644 index 0000000000..ac8e74f5b9 --- /dev/null +++ b/changelog.d/5931.misc @@ -0,0 +1 @@ +Remove unnecessary parentheses in return statements. \ No newline at end of file diff --git a/synapse/api/auth.py b/synapse/api/auth.py index 7b3a5a8221..fd3cdf50b0 100644 --- a/synapse/api/auth.py +++ b/synapse/api/auth.py @@ -276,25 +276,25 @@ class Auth(object): self.get_access_token_from_request(request) ) if app_service is None: - return (None, None) + return None, None if app_service.ip_range_whitelist: ip_address = IPAddress(self.hs.get_ip_from_request(request)) if ip_address not in app_service.ip_range_whitelist: - return (None, None) + return None, None if b"user_id" not in request.args: - return (app_service.sender, app_service) + return app_service.sender, app_service user_id = request.args[b"user_id"][0].decode("utf8") if app_service.sender == user_id: - return (app_service.sender, app_service) + return app_service.sender, app_service if not app_service.is_interested_in_user(user_id): raise AuthError(403, "Application service cannot masquerade as this user.") if not (yield self.store.get_user_by_id(user_id)): raise AuthError(403, "Application service has not registered this user") - return (user_id, app_service) + return user_id, app_service @defer.inlineCallbacks def get_user_by_access_token(self, token, rights="access"): @@ -694,7 +694,7 @@ class Auth(object): # * The user is a guest user, and has joined the room # else it will throw. member_event = yield self.check_user_was_in_room(room_id, user_id) - return (member_event.membership, member_event.event_id) + return member_event.membership, member_event.event_id except AuthError: visibility = yield self.state.get_current_state( room_id, EventTypes.RoomHistoryVisibility, "" @@ -703,7 +703,7 @@ class Auth(object): visibility and visibility.content["history_visibility"] == "world_readable" ): - return (Membership.JOIN, None) + return Membership.JOIN, None return raise AuthError( 403, "Guest access not allowed", errcode=Codes.GUEST_ACCESS_FORBIDDEN diff --git a/synapse/app/frontend_proxy.py b/synapse/app/frontend_proxy.py index 611d285421..9504bfbc70 100644 --- a/synapse/app/frontend_proxy.py +++ b/synapse/app/frontend_proxy.py @@ -70,12 +70,12 @@ class PresenceStatusStubServlet(RestServlet): except HttpResponseException as e: raise e.to_synapse_error() - return (200, result) + return 200, result @defer.inlineCallbacks def on_PUT(self, request, user_id): yield self.auth.get_user_by_req(request) - return (200, {}) + return 200, {} class KeyUploadServlet(RestServlet): @@ -126,11 +126,11 @@ class KeyUploadServlet(RestServlet): self.main_uri + request.uri.decode("ascii"), body, headers=headers ) - return (200, result) + return 200, result else: # Just interested in counts. result = yield self.store.count_e2e_one_time_keys(user_id, device_id) - return (200, {"one_time_key_counts": result}) + return 200, {"one_time_key_counts": result} class FrontendProxySlavedStore( diff --git a/synapse/crypto/event_signing.py b/synapse/crypto/event_signing.py index 41eabbe717..694fb2c816 100644 --- a/synapse/crypto/event_signing.py +++ b/synapse/crypto/event_signing.py @@ -83,7 +83,7 @@ def compute_content_hash(event_dict, hash_algorithm): event_json_bytes = encode_canonical_json(event_dict) hashed = hash_algorithm(event_json_bytes) - return (hashed.name, hashed.digest()) + return hashed.name, hashed.digest() def compute_event_reference_hash(event, hash_algorithm=hashlib.sha256): @@ -106,7 +106,7 @@ def compute_event_reference_hash(event, hash_algorithm=hashlib.sha256): event_dict.pop("unsigned", None) event_json_bytes = encode_canonical_json(event_dict) hashed = hash_algorithm(event_json_bytes) - return (hashed.name, hashed.digest()) + return hashed.name, hashed.digest() def compute_event_signature(event_dict, signature_name, signing_key): diff --git a/synapse/federation/federation_client.py b/synapse/federation/federation_client.py index bec3080895..6ee6216660 100644 --- a/synapse/federation/federation_client.py +++ b/synapse/federation/federation_client.py @@ -355,7 +355,7 @@ class FederationClient(FederationBase): auth_chain.sort(key=lambda e: e.depth) - return (pdus, auth_chain) + return pdus, auth_chain except HttpResponseException as e: if e.code == 400 or e.code == 404: logger.info("Failed to use get_room_state_ids API, falling back") @@ -404,7 +404,7 @@ class FederationClient(FederationBase): signed_auth.sort(key=lambda e: e.depth) - return (signed_pdus, signed_auth) + return signed_pdus, signed_auth @defer.inlineCallbacks def get_events_from_store_or_dest(self, destination, room_id, event_ids): @@ -429,7 +429,7 @@ class FederationClient(FederationBase): missing_events.discard(k) if not missing_events: - return (signed_events, failed_to_fetch) + return signed_events, failed_to_fetch logger.debug( "Fetching unknown state/auth events %s for room %s", @@ -465,7 +465,7 @@ class FederationClient(FederationBase): # We removed all events we successfully fetched from `batch` failed_to_fetch.update(batch) - return (signed_events, failed_to_fetch) + return signed_events, failed_to_fetch @defer.inlineCallbacks @log_function diff --git a/synapse/federation/federation_server.py b/synapse/federation/federation_server.py index 05fd49f3c1..e5f0b90aec 100644 --- a/synapse/federation/federation_server.py +++ b/synapse/federation/federation_server.py @@ -100,7 +100,7 @@ class FederationServer(FederationBase): res = self._transaction_from_pdus(pdus).get_dict() - return (200, res) + return 200, res @defer.inlineCallbacks @log_function @@ -163,7 +163,7 @@ class FederationServer(FederationBase): yield self.transaction_actions.set_response( origin, transaction, 400, response ) - return (400, response) + return 400, response received_pdus_counter.inc(len(transaction.pdus)) @@ -265,7 +265,7 @@ class FederationServer(FederationBase): logger.debug("Returning: %s", str(response)) yield self.transaction_actions.set_response(origin, transaction, 200, response) - return (200, response) + return 200, response @defer.inlineCallbacks def received_edu(self, origin, edu_type, content): @@ -298,7 +298,7 @@ class FederationServer(FederationBase): event_id, ) - return (200, resp) + return 200, resp @defer.inlineCallbacks def on_state_ids_request(self, origin, room_id, event_id): @@ -315,7 +315,7 @@ class FederationServer(FederationBase): state_ids = yield self.handler.get_state_ids_for_pdu(room_id, event_id) auth_chain_ids = yield self.store.get_auth_chain_ids(state_ids) - return (200, {"pdu_ids": state_ids, "auth_chain_ids": auth_chain_ids}) + return 200, {"pdu_ids": state_ids, "auth_chain_ids": auth_chain_ids} @defer.inlineCallbacks def _on_context_state_request_compute(self, room_id, event_id): @@ -345,15 +345,15 @@ class FederationServer(FederationBase): pdu = yield self.handler.get_persisted_pdu(origin, event_id) if pdu: - return (200, self._transaction_from_pdus([pdu]).get_dict()) + return 200, self._transaction_from_pdus([pdu]).get_dict() else: - return (404, "") + return 404, "" @defer.inlineCallbacks def on_query_request(self, query_type, args): received_queries_counter.labels(query_type).inc() resp = yield self.registry.on_query(query_type, args) - return (200, resp) + return 200, resp @defer.inlineCallbacks def on_make_join_request(self, origin, room_id, user_id, supported_versions): @@ -435,7 +435,7 @@ class FederationServer(FederationBase): logger.debug("on_send_leave_request: pdu sigs: %s", pdu.signatures) yield self.handler.on_send_leave_request(origin, pdu) - return (200, {}) + return 200, {} @defer.inlineCallbacks def on_event_auth(self, origin, room_id, event_id): @@ -446,7 +446,7 @@ class FederationServer(FederationBase): time_now = self._clock.time_msec() auth_pdus = yield self.handler.on_event_auth(event_id) res = {"auth_chain": [a.get_pdu_json(time_now) for a in auth_pdus]} - return (200, res) + return 200, res @defer.inlineCallbacks def on_query_auth_request(self, origin, content, room_id, event_id): @@ -499,7 +499,7 @@ class FederationServer(FederationBase): "missing": ret.get("missing", []), } - return (200, send_content) + return 200, send_content @log_function def on_query_client_keys(self, origin, content): diff --git a/synapse/handlers/account_data.py b/synapse/handlers/account_data.py index 8acd9f9a83..38bc67191c 100644 --- a/synapse/handlers/account_data.py +++ b/synapse/handlers/account_data.py @@ -51,8 +51,8 @@ class AccountDataEventSource(object): {"type": account_data_type, "content": content, "room_id": room_id} ) - return (results, current_stream_id) + return results, current_stream_id @defer.inlineCallbacks def get_pagination_rows(self, user, config, key): - return ([], config.to_id) + return [], config.to_id diff --git a/synapse/handlers/auth.py b/synapse/handlers/auth.py index 0f3ebf7ef8..f844409d21 100644 --- a/synapse/handlers/auth.py +++ b/synapse/handlers/auth.py @@ -280,7 +280,7 @@ class AuthHandler(BaseHandler): creds, list(clientdict), ) - return (creds, clientdict, session["id"]) + return creds, clientdict, session["id"] ret = self._auth_dict_for_flows(flows, session) ret["completed"] = list(creds) @@ -722,7 +722,7 @@ class AuthHandler(BaseHandler): known_login_type = True is_valid = yield provider.check_password(qualified_user_id, password) if is_valid: - return (qualified_user_id, None) + return qualified_user_id, None if not hasattr(provider, "get_supported_login_types") or not hasattr( provider, "check_auth" @@ -766,7 +766,7 @@ class AuthHandler(BaseHandler): ) if canonical_user_id: - return (canonical_user_id, None) + return canonical_user_id, None if not known_login_type: raise SynapseError(400, "Unknown login type %s" % login_type) @@ -816,7 +816,7 @@ class AuthHandler(BaseHandler): result = (result, None) return result - return (None, None) + return None, None @defer.inlineCallbacks def _check_local_password(self, user_id, password): diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py index 94306c94a9..538b16efd6 100644 --- a/synapse/handlers/federation.py +++ b/synapse/handlers/federation.py @@ -1428,7 +1428,7 @@ class FederationHandler(BaseHandler): assert event.user_id == user_id assert event.state_key == user_id assert event.room_id == room_id - return (origin, event, format_ver) + return origin, event, format_ver @defer.inlineCallbacks @log_function diff --git a/synapse/handlers/initial_sync.py b/synapse/handlers/initial_sync.py index 42d6650ed9..595f75400b 100644 --- a/synapse/handlers/initial_sync.py +++ b/synapse/handlers/initial_sync.py @@ -449,7 +449,7 @@ class InitialSyncHandler(BaseHandler): # * The user is a guest user, and has joined the room # else it will throw. member_event = yield self.auth.check_user_was_in_room(room_id, user_id) - return (member_event.membership, member_event.event_id) + return member_event.membership, member_event.event_id return except AuthError: visibility = yield self.state_handler.get_current_state( @@ -459,7 +459,7 @@ class InitialSyncHandler(BaseHandler): visibility and visibility.content["history_visibility"] == "world_readable" ): - return (Membership.JOIN, None) + return Membership.JOIN, None return raise AuthError( 403, "Guest access not allowed", errcode=Codes.GUEST_ACCESS_FORBIDDEN diff --git a/synapse/handlers/presence.py b/synapse/handlers/presence.py index 94a9ca0357..8377a0ddc2 100644 --- a/synapse/handlers/presence.py +++ b/synapse/handlers/presence.py @@ -1032,7 +1032,7 @@ class PresenceEventSource(object): # # Hence this guard where we just return nothing so that the sync # doesn't return. C.f. #5503. - return ([], max_token) + return [], max_token presence = self.get_presence_handler() stream_change_cache = self.store.presence_stream_cache @@ -1279,7 +1279,7 @@ def get_interested_parties(store, states): # Always notify self users_to_states.setdefault(state.user_id, []).append(state) - return (room_ids_to_states, users_to_states) + return room_ids_to_states, users_to_states @defer.inlineCallbacks diff --git a/synapse/handlers/receipts.py b/synapse/handlers/receipts.py index 73973502a4..6854c751a6 100644 --- a/synapse/handlers/receipts.py +++ b/synapse/handlers/receipts.py @@ -148,7 +148,7 @@ class ReceiptEventSource(object): to_key = yield self.get_current_key() if from_key == to_key: - return ([], to_key) + return [], to_key events = yield self.store.get_linearized_receipts_for_rooms( room_ids, from_key=from_key, to_key=to_key diff --git a/synapse/handlers/register.py b/synapse/handlers/register.py index 4631fab94e..be0425a33b 100644 --- a/synapse/handlers/register.py +++ b/synapse/handlers/register.py @@ -622,7 +622,7 @@ class RegistrationHandler(BaseHandler): initial_display_name=initial_display_name, is_guest=is_guest, ) - return (r["device_id"], r["access_token"]) + return r["device_id"], r["access_token"] valid_until_ms = None if self.session_lifetime is not None: diff --git a/synapse/handlers/room_member.py b/synapse/handlers/room_member.py index 249a6d9c5d..f03a2bd540 100644 --- a/synapse/handlers/room_member.py +++ b/synapse/handlers/room_member.py @@ -903,7 +903,7 @@ class RoomMemberHandler(object): if not public_keys: public_keys.append(fallback_public_key) display_name = data["display_name"] - return (token, public_keys, fallback_public_key, display_name) + return token, public_keys, fallback_public_key, display_name @defer.inlineCallbacks def _is_host_in_room(self, current_state_ids): diff --git a/synapse/handlers/sync.py b/synapse/handlers/sync.py index ef7f2ca980..d582f8e494 100644 --- a/synapse/handlers/sync.py +++ b/synapse/handlers/sync.py @@ -378,7 +378,7 @@ class SyncHandler(object): event_copy = {k: v for (k, v) in iteritems(event) if k != "room_id"} ephemeral_by_room.setdefault(room_id, []).append(event_copy) - return (now_token, ephemeral_by_room) + return now_token, ephemeral_by_room @defer.inlineCallbacks def _load_filtered_recents( @@ -1332,7 +1332,7 @@ class SyncHandler(object): ) if not tags_by_room: logger.debug("no-oping sync") - return ([], [], [], []) + return [], [], [], [] ignored_account_data = yield self.store.get_global_account_data_by_type_for_user( "m.ignored_user_list", user_id=user_id @@ -1642,7 +1642,7 @@ class SyncHandler(object): ) room_entries.append(entry) - return (room_entries, invited, newly_joined_rooms, newly_left_rooms) + return room_entries, invited, newly_joined_rooms, newly_left_rooms @defer.inlineCallbacks def _get_all_rooms(self, sync_result_builder, ignored_users): @@ -1716,7 +1716,7 @@ class SyncHandler(object): ) ) - return (room_entries, invited, []) + return room_entries, invited, [] @defer.inlineCallbacks def _generate_room_entry( diff --git a/synapse/handlers/typing.py b/synapse/handlers/typing.py index f882330293..ca8ae9fb5b 100644 --- a/synapse/handlers/typing.py +++ b/synapse/handlers/typing.py @@ -319,4 +319,4 @@ class TypingNotificationEventSource(object): return self.get_typing_handler()._latest_room_serial def get_pagination_rows(self, user, pagination_config, key): - return ([], pagination_config.from_key) + return [], pagination_config.from_key diff --git a/synapse/http/federation/well_known_resolver.py b/synapse/http/federation/well_known_resolver.py index 5e9b0befb0..7ddfad286d 100644 --- a/synapse/http/federation/well_known_resolver.py +++ b/synapse/http/federation/well_known_resolver.py @@ -207,7 +207,7 @@ class WellKnownResolver(object): cache_period + WELL_KNOWN_REMEMBER_DOMAIN_HAD_VALID, ) - return (result, cache_period) + return result, cache_period @defer.inlineCallbacks def _make_well_known_request(self, server_name, retry): diff --git a/synapse/module_api/__init__.py b/synapse/module_api/__init__.py index 41147d4292..735b882363 100644 --- a/synapse/module_api/__init__.py +++ b/synapse/module_api/__init__.py @@ -101,7 +101,7 @@ class ModuleApi(object): ) user_id = yield self.register_user(localpart, displayname, emails) _, access_token = yield self.register_device(user_id) - return (user_id, access_token) + return user_id, access_token def register_user(self, localpart, displayname=None, emails=[]): """Registers a new user with given localpart and optional displayname, emails. diff --git a/synapse/notifier.py b/synapse/notifier.py index bd80c801b6..4e091314e6 100644 --- a/synapse/notifier.py +++ b/synapse/notifier.py @@ -472,11 +472,11 @@ class Notifier(object): joined_room_ids = yield self.store.get_rooms_for_user(user.to_string()) if explicit_room_id: if explicit_room_id in joined_room_ids: - return ([explicit_room_id], True) + return [explicit_room_id], True if (yield self._is_world_readable(explicit_room_id)): - return ([explicit_room_id], False) + return [explicit_room_id], False raise AuthError(403, "Non-joined access not allowed") - return (joined_room_ids, True) + return joined_room_ids, True @defer.inlineCallbacks def _is_world_readable(self, room_id): diff --git a/synapse/push/bulk_push_rule_evaluator.py b/synapse/push/bulk_push_rule_evaluator.py index c831975635..22491f3700 100644 --- a/synapse/push/bulk_push_rule_evaluator.py +++ b/synapse/push/bulk_push_rule_evaluator.py @@ -134,7 +134,7 @@ class BulkPushRuleEvaluator(object): pl_event = auth_events.get(POWER_KEY) - return (pl_event.content if pl_event else {}, sender_level) + return pl_event.content if pl_event else {}, sender_level @defer.inlineCallbacks def action_for_event_by_user(self, event, context): diff --git a/synapse/replication/http/federation.py b/synapse/replication/http/federation.py index fed4f08820..2f16955954 100644 --- a/synapse/replication/http/federation.py +++ b/synapse/replication/http/federation.py @@ -113,7 +113,7 @@ class ReplicationFederationSendEventsRestServlet(ReplicationEndpoint): event_and_contexts, backfilled ) - return (200, {}) + return 200, {} class ReplicationFederationSendEduRestServlet(ReplicationEndpoint): @@ -156,7 +156,7 @@ class ReplicationFederationSendEduRestServlet(ReplicationEndpoint): result = yield self.registry.on_edu(edu_type, origin, edu_content) - return (200, result) + return 200, result class ReplicationGetQueryRestServlet(ReplicationEndpoint): @@ -204,7 +204,7 @@ class ReplicationGetQueryRestServlet(ReplicationEndpoint): result = yield self.registry.on_query(query_type, args) - return (200, result) + return 200, result class ReplicationCleanRoomRestServlet(ReplicationEndpoint): @@ -238,7 +238,7 @@ class ReplicationCleanRoomRestServlet(ReplicationEndpoint): def _handle_request(self, request, room_id): yield self.store.clean_room_for_join(room_id) - return (200, {}) + return 200, {} def register_servlets(hs, http_server): diff --git a/synapse/replication/http/login.py b/synapse/replication/http/login.py index f17d3a2da4..786f5232b2 100644 --- a/synapse/replication/http/login.py +++ b/synapse/replication/http/login.py @@ -64,7 +64,7 @@ class RegisterDeviceReplicationServlet(ReplicationEndpoint): user_id, device_id, initial_display_name, is_guest ) - return (200, {"device_id": device_id, "access_token": access_token}) + return 200, {"device_id": device_id, "access_token": access_token} def register_servlets(hs, http_server): diff --git a/synapse/replication/http/membership.py b/synapse/replication/http/membership.py index 4217335d88..b9ce3477ad 100644 --- a/synapse/replication/http/membership.py +++ b/synapse/replication/http/membership.py @@ -83,7 +83,7 @@ class ReplicationRemoteJoinRestServlet(ReplicationEndpoint): remote_room_hosts, room_id, user_id, event_content ) - return (200, {}) + return 200, {} class ReplicationRemoteRejectInviteRestServlet(ReplicationEndpoint): @@ -153,7 +153,7 @@ class ReplicationRemoteRejectInviteRestServlet(ReplicationEndpoint): yield self.store.locally_reject_invite(user_id, room_id) ret = {} - return (200, ret) + return 200, ret class ReplicationUserJoinedLeftRoomRestServlet(ReplicationEndpoint): @@ -202,7 +202,7 @@ class ReplicationUserJoinedLeftRoomRestServlet(ReplicationEndpoint): else: raise Exception("Unrecognized change: %r", change) - return (200, {}) + return 200, {} def register_servlets(hs, http_server): diff --git a/synapse/replication/http/register.py b/synapse/replication/http/register.py index 3341320a87..87fe2dd9b0 100644 --- a/synapse/replication/http/register.py +++ b/synapse/replication/http/register.py @@ -90,7 +90,7 @@ class ReplicationRegisterServlet(ReplicationEndpoint): address=content["address"], ) - return (200, {}) + return 200, {} class ReplicationPostRegisterActionsServlet(ReplicationEndpoint): @@ -143,7 +143,7 @@ class ReplicationPostRegisterActionsServlet(ReplicationEndpoint): bind_msisdn=bind_msisdn, ) - return (200, {}) + return 200, {} def register_servlets(hs, http_server): diff --git a/synapse/replication/http/send_event.py b/synapse/replication/http/send_event.py index eff7bd7305..adb9b2f7f4 100644 --- a/synapse/replication/http/send_event.py +++ b/synapse/replication/http/send_event.py @@ -117,7 +117,7 @@ class ReplicationSendEventRestServlet(ReplicationEndpoint): requester, event, context, ratelimit=ratelimit, extra_users=extra_users ) - return (200, {}) + return 200, {} def register_servlets(hs, http_server): diff --git a/synapse/replication/tcp/streams/_base.py b/synapse/replication/tcp/streams/_base.py index c10b85d2ff..f03111c259 100644 --- a/synapse/replication/tcp/streams/_base.py +++ b/synapse/replication/tcp/streams/_base.py @@ -158,7 +158,7 @@ class Stream(object): updates, current_token = yield self.get_updates_since(self.last_token) self.last_token = current_token - return (updates, current_token) + return updates, current_token @defer.inlineCallbacks def get_updates_since(self, from_token): @@ -172,14 +172,14 @@ class Stream(object): sent over the replication steam. """ if from_token in ("NOW", "now"): - return ([], self.upto_token) + return [], self.upto_token current_token = self.upto_token from_token = int(from_token) if from_token == current_token: - return ([], current_token) + return [], current_token if self._LIMITED: rows = yield self.update_function( @@ -198,7 +198,7 @@ class Stream(object): if self._LIMITED and len(updates) >= MAX_EVENTS_BEHIND: raise Exception("stream %s has fallen behind" % (self.NAME)) - return (updates, current_token) + return updates, current_token def current_token(self): """Gets the current token of the underlying streams. Should be provided diff --git a/synapse/rest/admin/__init__.py b/synapse/rest/admin/__init__.py index fa91cc8dee..b4761adaed 100644 --- a/synapse/rest/admin/__init__.py +++ b/synapse/rest/admin/__init__.py @@ -69,7 +69,7 @@ class UsersRestServlet(RestServlet): ret = yield self.handlers.admin_handler.get_users() - return (200, ret) + return 200, ret class VersionServlet(RestServlet): @@ -120,7 +120,7 @@ class UserRegisterServlet(RestServlet): nonce = self.hs.get_secrets().token_hex(64) self.nonces[nonce] = int(self.reactor.seconds()) - return (200, {"nonce": nonce}) + return 200, {"nonce": nonce} @defer.inlineCallbacks def on_POST(self, request): @@ -212,7 +212,7 @@ class UserRegisterServlet(RestServlet): ) result = yield register._create_registration_details(user_id, body) - return (200, result) + return 200, result class WhoisRestServlet(RestServlet): @@ -237,7 +237,7 @@ class WhoisRestServlet(RestServlet): ret = yield self.handlers.admin_handler.get_whois(target_user) - return (200, ret) + return 200, ret class PurgeHistoryRestServlet(RestServlet): @@ -322,7 +322,7 @@ class PurgeHistoryRestServlet(RestServlet): room_id, token, delete_local_events=delete_local_events ) - return (200, {"purge_id": purge_id}) + return 200, {"purge_id": purge_id} class PurgeHistoryStatusRestServlet(RestServlet): @@ -347,7 +347,7 @@ class PurgeHistoryStatusRestServlet(RestServlet): if purge_status is None: raise NotFoundError("purge id '%s' not found" % purge_id) - return (200, purge_status.asdict()) + return 200, purge_status.asdict() class DeactivateAccountRestServlet(RestServlet): @@ -379,7 +379,7 @@ class DeactivateAccountRestServlet(RestServlet): else: id_server_unbind_result = "no-support" - return (200, {"id_server_unbind_result": id_server_unbind_result}) + return 200, {"id_server_unbind_result": id_server_unbind_result} class ShutdownRoomRestServlet(RestServlet): @@ -549,7 +549,7 @@ class ResetPasswordRestServlet(RestServlet): yield self._set_password_handler.set_password( target_user_id, new_password, requester ) - return (200, {}) + return 200, {} class GetUsersPaginatedRestServlet(RestServlet): @@ -591,7 +591,7 @@ class GetUsersPaginatedRestServlet(RestServlet): logger.info("limit: %s, start: %s", limit, start) ret = yield self.handlers.admin_handler.get_users_paginate(order, start, limit) - return (200, ret) + return 200, ret @defer.inlineCallbacks def on_POST(self, request, target_user_id): @@ -619,7 +619,7 @@ class GetUsersPaginatedRestServlet(RestServlet): logger.info("limit: %s, start: %s", limit, start) ret = yield self.handlers.admin_handler.get_users_paginate(order, start, limit) - return (200, ret) + return 200, ret class SearchUsersRestServlet(RestServlet): @@ -662,7 +662,7 @@ class SearchUsersRestServlet(RestServlet): logger.info("term: %s ", term) ret = yield self.handlers.admin_handler.search_users(term) - return (200, ret) + return 200, ret class DeleteGroupAdminRestServlet(RestServlet): @@ -685,7 +685,7 @@ class DeleteGroupAdminRestServlet(RestServlet): raise SynapseError(400, "Can only delete local groups") yield self.group_server.delete_group(group_id, requester.user.to_string()) - return (200, {}) + return 200, {} class AccountValidityRenewServlet(RestServlet): @@ -716,7 +716,7 @@ class AccountValidityRenewServlet(RestServlet): ) res = {"expiration_ts": expiration_ts} - return (200, res) + return 200, res ######################################################################################## diff --git a/synapse/rest/admin/media.py b/synapse/rest/admin/media.py index 824df919f2..f3f63f0be7 100644 --- a/synapse/rest/admin/media.py +++ b/synapse/rest/admin/media.py @@ -49,7 +49,7 @@ class QuarantineMediaInRoom(RestServlet): room_id, requester.user.to_string() ) - return (200, {"num_quarantined": num_quarantined}) + return 200, {"num_quarantined": num_quarantined} class ListMediaInRoom(RestServlet): @@ -70,7 +70,7 @@ class ListMediaInRoom(RestServlet): local_mxcs, remote_mxcs = yield self.store.get_media_mxcs_in_room(room_id) - return (200, {"local": local_mxcs, "remote": remote_mxcs}) + return 200, {"local": local_mxcs, "remote": remote_mxcs} class PurgeMediaCacheRestServlet(RestServlet): @@ -89,7 +89,7 @@ class PurgeMediaCacheRestServlet(RestServlet): ret = yield self.media_repository.delete_old_remote_media(before_ts) - return (200, ret) + return 200, ret def register_servlets_for_media_repo(hs, http_server): diff --git a/synapse/rest/admin/purge_room_servlet.py b/synapse/rest/admin/purge_room_servlet.py index 2922eb543e..f474066542 100644 --- a/synapse/rest/admin/purge_room_servlet.py +++ b/synapse/rest/admin/purge_room_servlet.py @@ -54,4 +54,4 @@ class PurgeRoomServlet(RestServlet): await self.pagination_handler.purge_room(body["room_id"]) - return (200, {}) + return 200, {} diff --git a/synapse/rest/admin/server_notice_servlet.py b/synapse/rest/admin/server_notice_servlet.py index 656526fea5..ae2cbe2e0a 100644 --- a/synapse/rest/admin/server_notice_servlet.py +++ b/synapse/rest/admin/server_notice_servlet.py @@ -92,7 +92,7 @@ class SendServerNoticeServlet(RestServlet): event_content=body["content"], ) - return (200, {"event_id": event.event_id}) + return 200, {"event_id": event.event_id} def on_PUT(self, request, txn_id): return self.txns.fetch_or_execute_request( diff --git a/synapse/rest/admin/users.py b/synapse/rest/admin/users.py index 5364117420..9720a3bab0 100644 --- a/synapse/rest/admin/users.py +++ b/synapse/rest/admin/users.py @@ -71,7 +71,7 @@ class UserAdminServlet(RestServlet): is_admin = yield self.handlers.admin_handler.get_user_server_admin(target_user) is_admin = bool(is_admin) - return (200, {"admin": is_admin}) + return 200, {"admin": is_admin} @defer.inlineCallbacks def on_PUT(self, request, user_id): @@ -97,4 +97,4 @@ class UserAdminServlet(RestServlet): target_user, set_admin_to ) - return (200, {}) + return 200, {} diff --git a/synapse/rest/client/v1/directory.py b/synapse/rest/client/v1/directory.py index 4284738021..4ea3666874 100644 --- a/synapse/rest/client/v1/directory.py +++ b/synapse/rest/client/v1/directory.py @@ -54,7 +54,7 @@ class ClientDirectoryServer(RestServlet): dir_handler = self.handlers.directory_handler res = yield dir_handler.get_association(room_alias) - return (200, res) + return 200, res @defer.inlineCallbacks def on_PUT(self, request, room_alias): @@ -87,7 +87,7 @@ class ClientDirectoryServer(RestServlet): requester, room_alias, room_id, servers ) - return (200, {}) + return 200, {} @defer.inlineCallbacks def on_DELETE(self, request, room_alias): @@ -102,7 +102,7 @@ class ClientDirectoryServer(RestServlet): service.url, room_alias.to_string(), ) - return (200, {}) + return 200, {} except InvalidClientCredentialsError: # fallback to default user behaviour if they aren't an AS pass @@ -118,7 +118,7 @@ class ClientDirectoryServer(RestServlet): "User %s deleted alias %s", user.to_string(), room_alias.to_string() ) - return (200, {}) + return 200, {} class ClientDirectoryListServer(RestServlet): @@ -136,7 +136,7 @@ class ClientDirectoryListServer(RestServlet): if room is None: raise NotFoundError("Unknown room") - return (200, {"visibility": "public" if room["is_public"] else "private"}) + return 200, {"visibility": "public" if room["is_public"] else "private"} @defer.inlineCallbacks def on_PUT(self, request, room_id): @@ -149,7 +149,7 @@ class ClientDirectoryListServer(RestServlet): requester, room_id, visibility ) - return (200, {}) + return 200, {} @defer.inlineCallbacks def on_DELETE(self, request, room_id): @@ -159,7 +159,7 @@ class ClientDirectoryListServer(RestServlet): requester, room_id, "private" ) - return (200, {}) + return 200, {} class ClientAppserviceDirectoryListServer(RestServlet): @@ -193,4 +193,4 @@ class ClientAppserviceDirectoryListServer(RestServlet): requester.app_service.id, network_id, room_id, visibility ) - return (200, {}) + return 200, {} diff --git a/synapse/rest/client/v1/events.py b/synapse/rest/client/v1/events.py index 53ebed2203..6651b4cf07 100644 --- a/synapse/rest/client/v1/events.py +++ b/synapse/rest/client/v1/events.py @@ -67,10 +67,10 @@ class EventStreamRestServlet(RestServlet): is_guest=is_guest, ) - return (200, chunk) + return 200, chunk def on_OPTIONS(self, request): - return (200, {}) + return 200, {} # TODO: Unit test gets, with and without auth, with different kinds of events. @@ -91,9 +91,9 @@ class EventRestServlet(RestServlet): time_now = self.clock.time_msec() if event: event = yield self._event_serializer.serialize_event(event, time_now) - return (200, event) + return 200, event else: - return (404, "Event not found.") + return 404, "Event not found." def register_servlets(hs, http_server): diff --git a/synapse/rest/client/v1/initial_sync.py b/synapse/rest/client/v1/initial_sync.py index 70b8478e90..2da3cd7511 100644 --- a/synapse/rest/client/v1/initial_sync.py +++ b/synapse/rest/client/v1/initial_sync.py @@ -42,7 +42,7 @@ class InitialSyncRestServlet(RestServlet): include_archived=include_archived, ) - return (200, content) + return 200, content def register_servlets(hs, http_server): diff --git a/synapse/rest/client/v1/login.py b/synapse/rest/client/v1/login.py index 5762b9fd06..25a1b67092 100644 --- a/synapse/rest/client/v1/login.py +++ b/synapse/rest/client/v1/login.py @@ -121,10 +121,10 @@ class LoginRestServlet(RestServlet): ({"type": t} for t in self.auth_handler.get_supported_login_types()) ) - return (200, {"flows": flows}) + return 200, {"flows": flows} def on_OPTIONS(self, request): - return (200, {}) + return 200, {} @defer.inlineCallbacks def on_POST(self, request): @@ -152,7 +152,7 @@ class LoginRestServlet(RestServlet): well_known_data = self._well_known_builder.get_well_known() if well_known_data: result["well_known"] = well_known_data - return (200, result) + return 200, result @defer.inlineCallbacks def _do_other_login(self, login_submission): diff --git a/synapse/rest/client/v1/logout.py b/synapse/rest/client/v1/logout.py index 2769f3a189..4785a34d75 100644 --- a/synapse/rest/client/v1/logout.py +++ b/synapse/rest/client/v1/logout.py @@ -33,7 +33,7 @@ class LogoutRestServlet(RestServlet): self._device_handler = hs.get_device_handler() def on_OPTIONS(self, request): - return (200, {}) + return 200, {} @defer.inlineCallbacks def on_POST(self, request): @@ -49,7 +49,7 @@ class LogoutRestServlet(RestServlet): requester.user.to_string(), requester.device_id ) - return (200, {}) + return 200, {} class LogoutAllRestServlet(RestServlet): @@ -62,7 +62,7 @@ class LogoutAllRestServlet(RestServlet): self._device_handler = hs.get_device_handler() def on_OPTIONS(self, request): - return (200, {}) + return 200, {} @defer.inlineCallbacks def on_POST(self, request): @@ -75,7 +75,7 @@ class LogoutAllRestServlet(RestServlet): # .. and then delete any access tokens which weren't associated with # devices. yield self._auth_handler.delete_access_tokens_for_user(user_id) - return (200, {}) + return 200, {} def register_servlets(hs, http_server): diff --git a/synapse/rest/client/v1/presence.py b/synapse/rest/client/v1/presence.py index 1eb1068c98..0153525cef 100644 --- a/synapse/rest/client/v1/presence.py +++ b/synapse/rest/client/v1/presence.py @@ -56,7 +56,7 @@ class PresenceStatusRestServlet(RestServlet): state = yield self.presence_handler.get_state(target_user=user) state = format_user_presence_state(state, self.clock.time_msec()) - return (200, state) + return 200, state @defer.inlineCallbacks def on_PUT(self, request, user_id): @@ -88,10 +88,10 @@ class PresenceStatusRestServlet(RestServlet): if self.hs.config.use_presence: yield self.presence_handler.set_state(user, state) - return (200, {}) + return 200, {} def on_OPTIONS(self, request): - return (200, {}) + return 200, {} def register_servlets(hs, http_server): diff --git a/synapse/rest/client/v1/profile.py b/synapse/rest/client/v1/profile.py index 2657ae45bb..bbce2e2b71 100644 --- a/synapse/rest/client/v1/profile.py +++ b/synapse/rest/client/v1/profile.py @@ -48,7 +48,7 @@ class ProfileDisplaynameRestServlet(RestServlet): if displayname is not None: ret["displayname"] = displayname - return (200, ret) + return 200, ret @defer.inlineCallbacks def on_PUT(self, request, user_id): @@ -61,14 +61,14 @@ class ProfileDisplaynameRestServlet(RestServlet): try: new_name = content["displayname"] except Exception: - return (400, "Unable to parse name") + return 400, "Unable to parse name" yield self.profile_handler.set_displayname(user, requester, new_name, is_admin) - return (200, {}) + return 200, {} def on_OPTIONS(self, request, user_id): - return (200, {}) + return 200, {} class ProfileAvatarURLRestServlet(RestServlet): @@ -98,7 +98,7 @@ class ProfileAvatarURLRestServlet(RestServlet): if avatar_url is not None: ret["avatar_url"] = avatar_url - return (200, ret) + return 200, ret @defer.inlineCallbacks def on_PUT(self, request, user_id): @@ -110,14 +110,14 @@ class ProfileAvatarURLRestServlet(RestServlet): try: new_name = content["avatar_url"] except Exception: - return (400, "Unable to parse name") + return 400, "Unable to parse name" yield self.profile_handler.set_avatar_url(user, requester, new_name, is_admin) - return (200, {}) + return 200, {} def on_OPTIONS(self, request, user_id): - return (200, {}) + return 200, {} class ProfileRestServlet(RestServlet): @@ -150,7 +150,7 @@ class ProfileRestServlet(RestServlet): if avatar_url is not None: ret["avatar_url"] = avatar_url - return (200, ret) + return 200, ret def register_servlets(hs, http_server): diff --git a/synapse/rest/client/v1/push_rule.py b/synapse/rest/client/v1/push_rule.py index c3ae8b98a8..9f8c3d09e3 100644 --- a/synapse/rest/client/v1/push_rule.py +++ b/synapse/rest/client/v1/push_rule.py @@ -69,7 +69,7 @@ class PushRuleRestServlet(RestServlet): if "attr" in spec: yield self.set_rule_attr(user_id, spec, content) self.notify_user(user_id) - return (200, {}) + return 200, {} if spec["rule_id"].startswith("."): # Rule ids starting with '.' are reserved for server default rules. @@ -106,7 +106,7 @@ class PushRuleRestServlet(RestServlet): except RuleNotFoundException as e: raise SynapseError(400, str(e)) - return (200, {}) + return 200, {} @defer.inlineCallbacks def on_DELETE(self, request, path): @@ -123,7 +123,7 @@ class PushRuleRestServlet(RestServlet): try: yield self.store.delete_push_rule(user_id, namespaced_rule_id) self.notify_user(user_id) - return (200, {}) + return 200, {} except StoreError as e: if e.code == 404: raise NotFoundError() @@ -151,10 +151,10 @@ class PushRuleRestServlet(RestServlet): ) if path[0] == "": - return (200, rules) + return 200, rules elif path[0] == "global": result = _filter_ruleset_with_path(rules["global"], path[1:]) - return (200, result) + return 200, result else: raise UnrecognizedRequestError() diff --git a/synapse/rest/client/v1/pusher.py b/synapse/rest/client/v1/pusher.py index ebc3dec516..41660682d9 100644 --- a/synapse/rest/client/v1/pusher.py +++ b/synapse/rest/client/v1/pusher.py @@ -62,7 +62,7 @@ class PushersRestServlet(RestServlet): if k not in allowed_keys: del p[k] - return (200, {"pushers": pushers}) + return 200, {"pushers": pushers} def on_OPTIONS(self, _): return 200, {} @@ -94,7 +94,7 @@ class PushersSetRestServlet(RestServlet): yield self.pusher_pool.remove_pusher( content["app_id"], content["pushkey"], user_id=user.to_string() ) - return (200, {}) + return 200, {} assert_params_in_dict( content, @@ -143,7 +143,7 @@ class PushersSetRestServlet(RestServlet): self.notifier.on_new_replication_data() - return (200, {}) + return 200, {} def on_OPTIONS(self, _): return 200, {} diff --git a/synapse/rest/client/v1/room.py b/synapse/rest/client/v1/room.py index 4b2344e696..f244e8f469 100644 --- a/synapse/rest/client/v1/room.py +++ b/synapse/rest/client/v1/room.py @@ -91,14 +91,14 @@ class RoomCreateRestServlet(TransactionRestServlet): requester, self.get_room_config(request) ) - return (200, info) + return 200, info def get_room_config(self, request): user_supplied_config = parse_json_object_from_request(request) return user_supplied_config def on_OPTIONS(self, request): - return (200, {}) + return 200, {} # TODO: Needs unit testing for generic events @@ -173,9 +173,9 @@ class RoomStateEventRestServlet(TransactionRestServlet): if format == "event": event = format_event_for_client_v2(data.get_dict()) - return (200, event) + return 200, event elif format == "content": - return (200, data.get_dict()["content"]) + return 200, data.get_dict()["content"] @defer.inlineCallbacks def on_PUT(self, request, room_id, event_type, state_key, txn_id=None): @@ -210,7 +210,7 @@ class RoomStateEventRestServlet(TransactionRestServlet): ret = {} if event: ret = {"event_id": event.event_id} - return (200, ret) + return 200, ret # TODO: Needs unit testing for generic events + feedback @@ -244,10 +244,10 @@ class RoomSendEventRestServlet(TransactionRestServlet): requester, event_dict, txn_id=txn_id ) - return (200, {"event_id": event.event_id}) + return 200, {"event_id": event.event_id} def on_GET(self, request, room_id, event_type, txn_id): - return (200, "Not implemented") + return 200, "Not implemented" def on_PUT(self, request, room_id, event_type, txn_id): return self.txns.fetch_or_execute_request( @@ -307,7 +307,7 @@ class JoinRoomAliasServlet(TransactionRestServlet): third_party_signed=content.get("third_party_signed", None), ) - return (200, {"room_id": room_id}) + return 200, {"room_id": room_id} def on_PUT(self, request, room_identifier, txn_id): return self.txns.fetch_or_execute_request( @@ -360,7 +360,7 @@ class PublicRoomListRestServlet(TransactionRestServlet): limit=limit, since_token=since_token ) - return (200, data) + return 200, data @defer.inlineCallbacks def on_POST(self, request): @@ -405,7 +405,7 @@ class PublicRoomListRestServlet(TransactionRestServlet): network_tuple=network_tuple, ) - return (200, data) + return 200, data # TODO: Needs unit testing @@ -456,7 +456,7 @@ class RoomMemberListRestServlet(RestServlet): continue chunk.append(event) - return (200, {"chunk": chunk}) + return 200, {"chunk": chunk} # deprecated in favour of /members?membership=join? @@ -477,7 +477,7 @@ class JoinedRoomMemberListRestServlet(RestServlet): requester, room_id ) - return (200, {"joined": users_with_profile}) + return 200, {"joined": users_with_profile} # TODO: Needs better unit testing @@ -510,7 +510,7 @@ class RoomMessageListRestServlet(RestServlet): event_filter=event_filter, ) - return (200, msgs) + return 200, msgs # TODO: Needs unit testing @@ -531,7 +531,7 @@ class RoomStateRestServlet(RestServlet): user_id=requester.user.to_string(), is_guest=requester.is_guest, ) - return (200, events) + return 200, events # TODO: Needs unit testing @@ -550,7 +550,7 @@ class RoomInitialSyncRestServlet(RestServlet): content = yield self.initial_sync_handler.room_initial_sync( room_id=room_id, requester=requester, pagin_config=pagination_config ) - return (200, content) + return 200, content class RoomEventServlet(RestServlet): @@ -581,7 +581,7 @@ class RoomEventServlet(RestServlet): time_now = self.clock.time_msec() if event: event = yield self._event_serializer.serialize_event(event, time_now) - return (200, event) + return 200, event return SynapseError(404, "Event not found.", errcode=Codes.NOT_FOUND) @@ -633,7 +633,7 @@ class RoomEventContextServlet(RestServlet): results["state"], time_now ) - return (200, results) + return 200, results class RoomForgetRestServlet(TransactionRestServlet): @@ -652,7 +652,7 @@ class RoomForgetRestServlet(TransactionRestServlet): yield self.room_member_handler.forget(user=requester.user, room_id=room_id) - return (200, {}) + return 200, {} def on_PUT(self, request, room_id, txn_id): return self.txns.fetch_or_execute_request( @@ -702,7 +702,7 @@ class RoomMembershipRestServlet(TransactionRestServlet): requester, txn_id, ) - return (200, {}) + return 200, {} return target = requester.user @@ -729,7 +729,7 @@ class RoomMembershipRestServlet(TransactionRestServlet): if membership_action == "join": return_value["room_id"] = room_id - return (200, return_value) + return 200, return_value def _has_3pid_invite_keys(self, content): for key in {"id_server", "medium", "address"}: @@ -771,7 +771,7 @@ class RoomRedactEventRestServlet(TransactionRestServlet): txn_id=txn_id, ) - return (200, {"event_id": event.event_id}) + return 200, {"event_id": event.event_id} def on_PUT(self, request, room_id, event_id, txn_id): return self.txns.fetch_or_execute_request( @@ -816,7 +816,7 @@ class RoomTypingRestServlet(RestServlet): target_user=target_user, auth_user=requester.user, room_id=room_id ) - return (200, {}) + return 200, {} class SearchRestServlet(RestServlet): @@ -838,7 +838,7 @@ class SearchRestServlet(RestServlet): requester.user, content, batch ) - return (200, results) + return 200, results class JoinedRoomsRestServlet(RestServlet): @@ -854,7 +854,7 @@ class JoinedRoomsRestServlet(RestServlet): requester = yield self.auth.get_user_by_req(request, allow_guest=True) room_ids = yield self.store.get_rooms_for_user(requester.user.to_string()) - return (200, {"joined_rooms": list(room_ids)}) + return 200, {"joined_rooms": list(room_ids)} def register_txn_path(servlet, regex_string, http_server, with_get=False): diff --git a/synapse/rest/client/v1/voip.py b/synapse/rest/client/v1/voip.py index 497cddf8b8..2afdbb89e5 100644 --- a/synapse/rest/client/v1/voip.py +++ b/synapse/rest/client/v1/voip.py @@ -60,7 +60,7 @@ class VoipRestServlet(RestServlet): password = turnPassword else: - return (200, {}) + return 200, {} return ( 200, @@ -73,7 +73,7 @@ class VoipRestServlet(RestServlet): ) def on_OPTIONS(self, request): - return (200, {}) + return 200, {} def register_servlets(hs, http_server): diff --git a/synapse/rest/client/v2_alpha/account.py b/synapse/rest/client/v2_alpha/account.py index 934ed5d16d..0620a4d0cf 100644 --- a/synapse/rest/client/v2_alpha/account.py +++ b/synapse/rest/client/v2_alpha/account.py @@ -117,7 +117,7 @@ class EmailPasswordRequestTokenRestServlet(RestServlet): # Wrap the session id in a JSON object ret = {"sid": sid} - return (200, ret) + return 200, ret @defer.inlineCallbacks def send_password_reset(self, email, client_secret, send_attempt, next_link=None): @@ -221,7 +221,7 @@ class MsisdnPasswordRequestTokenRestServlet(RestServlet): raise SynapseError(400, "MSISDN not found", Codes.THREEPID_NOT_FOUND) ret = yield self.identity_handler.requestMsisdnToken(**body) - return (200, ret) + return 200, ret class PasswordResetSubmitTokenServlet(RestServlet): @@ -330,7 +330,7 @@ class PasswordResetSubmitTokenServlet(RestServlet): ) response_code = 200 if valid else 400 - return (response_code, {"success": valid}) + return response_code, {"success": valid} class PasswordRestServlet(RestServlet): @@ -399,7 +399,7 @@ class PasswordRestServlet(RestServlet): yield self._set_password_handler.set_password(user_id, new_password, requester) - return (200, {}) + return 200, {} def on_OPTIONS(self, _): return 200, {} @@ -434,7 +434,7 @@ class DeactivateAccountRestServlet(RestServlet): yield self._deactivate_account_handler.deactivate_account( requester.user.to_string(), erase ) - return (200, {}) + return 200, {} yield self.auth_handler.validate_user_via_ui_auth( requester, body, self.hs.get_ip_from_request(request) @@ -447,7 +447,7 @@ class DeactivateAccountRestServlet(RestServlet): else: id_server_unbind_result = "no-support" - return (200, {"id_server_unbind_result": id_server_unbind_result}) + return 200, {"id_server_unbind_result": id_server_unbind_result} class EmailThreepidRequestTokenRestServlet(RestServlet): @@ -481,7 +481,7 @@ class EmailThreepidRequestTokenRestServlet(RestServlet): raise SynapseError(400, "Email is already in use", Codes.THREEPID_IN_USE) ret = yield self.identity_handler.requestEmailToken(**body) - return (200, ret) + return 200, ret class MsisdnThreepidRequestTokenRestServlet(RestServlet): @@ -516,7 +516,7 @@ class MsisdnThreepidRequestTokenRestServlet(RestServlet): raise SynapseError(400, "MSISDN is already in use", Codes.THREEPID_IN_USE) ret = yield self.identity_handler.requestMsisdnToken(**body) - return (200, ret) + return 200, ret class ThreepidRestServlet(RestServlet): @@ -536,7 +536,7 @@ class ThreepidRestServlet(RestServlet): threepids = yield self.datastore.user_get_threepids(requester.user.to_string()) - return (200, {"threepids": threepids}) + return 200, {"threepids": threepids} @defer.inlineCallbacks def on_POST(self, request): @@ -568,7 +568,7 @@ class ThreepidRestServlet(RestServlet): logger.debug("Binding threepid %s to %s", threepid, user_id) yield self.identity_handler.bind_threepid(threePidCreds, user_id) - return (200, {}) + return 200, {} class ThreepidDeleteRestServlet(RestServlet): @@ -603,7 +603,7 @@ class ThreepidDeleteRestServlet(RestServlet): else: id_server_unbind_result = "no-support" - return (200, {"id_server_unbind_result": id_server_unbind_result}) + return 200, {"id_server_unbind_result": id_server_unbind_result} class WhoamiRestServlet(RestServlet): @@ -617,7 +617,7 @@ class WhoamiRestServlet(RestServlet): def on_GET(self, request): requester = yield self.auth.get_user_by_req(request) - return (200, {"user_id": requester.user.to_string()}) + return 200, {"user_id": requester.user.to_string()} def register_servlets(hs, http_server): diff --git a/synapse/rest/client/v2_alpha/account_data.py b/synapse/rest/client/v2_alpha/account_data.py index 98f2f6f4b5..f0db204ffa 100644 --- a/synapse/rest/client/v2_alpha/account_data.py +++ b/synapse/rest/client/v2_alpha/account_data.py @@ -55,7 +55,7 @@ class AccountDataServlet(RestServlet): self.notifier.on_new_event("account_data_key", max_id, users=[user_id]) - return (200, {}) + return 200, {} @defer.inlineCallbacks def on_GET(self, request, user_id, account_data_type): @@ -70,7 +70,7 @@ class AccountDataServlet(RestServlet): if event is None: raise NotFoundError("Account data not found") - return (200, event) + return 200, event class RoomAccountDataServlet(RestServlet): @@ -112,7 +112,7 @@ class RoomAccountDataServlet(RestServlet): self.notifier.on_new_event("account_data_key", max_id, users=[user_id]) - return (200, {}) + return 200, {} @defer.inlineCallbacks def on_GET(self, request, user_id, room_id, account_data_type): @@ -127,7 +127,7 @@ class RoomAccountDataServlet(RestServlet): if event is None: raise NotFoundError("Room account data not found") - return (200, event) + return 200, event def register_servlets(hs, http_server): diff --git a/synapse/rest/client/v2_alpha/capabilities.py b/synapse/rest/client/v2_alpha/capabilities.py index a4fa45fe11..acd58af193 100644 --- a/synapse/rest/client/v2_alpha/capabilities.py +++ b/synapse/rest/client/v2_alpha/capabilities.py @@ -58,7 +58,7 @@ class CapabilitiesRestServlet(RestServlet): "m.change_password": {"enabled": change_password}, } } - return (200, response) + return 200, response def register_servlets(hs, http_server): diff --git a/synapse/rest/client/v2_alpha/devices.py b/synapse/rest/client/v2_alpha/devices.py index 9adf76cc0c..26d0235208 100644 --- a/synapse/rest/client/v2_alpha/devices.py +++ b/synapse/rest/client/v2_alpha/devices.py @@ -48,7 +48,7 @@ class DevicesRestServlet(RestServlet): devices = yield self.device_handler.get_devices_by_user( requester.user.to_string() ) - return (200, {"devices": devices}) + return 200, {"devices": devices} class DeleteDevicesRestServlet(RestServlet): @@ -91,7 +91,7 @@ class DeleteDevicesRestServlet(RestServlet): yield self.device_handler.delete_devices( requester.user.to_string(), body["devices"] ) - return (200, {}) + return 200, {} class DeviceRestServlet(RestServlet): @@ -114,7 +114,7 @@ class DeviceRestServlet(RestServlet): device = yield self.device_handler.get_device( requester.user.to_string(), device_id ) - return (200, device) + return 200, device @interactive_auth_handler @defer.inlineCallbacks @@ -137,7 +137,7 @@ class DeviceRestServlet(RestServlet): ) yield self.device_handler.delete_device(requester.user.to_string(), device_id) - return (200, {}) + return 200, {} @defer.inlineCallbacks def on_PUT(self, request, device_id): @@ -147,7 +147,7 @@ class DeviceRestServlet(RestServlet): yield self.device_handler.update_device( requester.user.to_string(), device_id, body ) - return (200, {}) + return 200, {} def register_servlets(hs, http_server): diff --git a/synapse/rest/client/v2_alpha/filter.py b/synapse/rest/client/v2_alpha/filter.py index 22be0ee3c5..c6ddf24c8d 100644 --- a/synapse/rest/client/v2_alpha/filter.py +++ b/synapse/rest/client/v2_alpha/filter.py @@ -56,7 +56,7 @@ class GetFilterRestServlet(RestServlet): user_localpart=target_user.localpart, filter_id=filter_id ) - return (200, filter.get_filter_json()) + return 200, filter.get_filter_json() except (KeyError, StoreError): raise SynapseError(400, "No such filter", errcode=Codes.NOT_FOUND) @@ -89,7 +89,7 @@ class CreateFilterRestServlet(RestServlet): user_localpart=target_user.localpart, user_filter=content ) - return (200, {"filter_id": str(filter_id)}) + return 200, {"filter_id": str(filter_id)} def register_servlets(hs, http_server): diff --git a/synapse/rest/client/v2_alpha/groups.py b/synapse/rest/client/v2_alpha/groups.py index e629c4256d..999a0fa80c 100644 --- a/synapse/rest/client/v2_alpha/groups.py +++ b/synapse/rest/client/v2_alpha/groups.py @@ -47,7 +47,7 @@ class GroupServlet(RestServlet): group_id, requester_user_id ) - return (200, group_description) + return 200, group_description @defer.inlineCallbacks def on_POST(self, request, group_id): @@ -59,7 +59,7 @@ class GroupServlet(RestServlet): group_id, requester_user_id, content ) - return (200, {}) + return 200, {} class GroupSummaryServlet(RestServlet): @@ -83,7 +83,7 @@ class GroupSummaryServlet(RestServlet): group_id, requester_user_id ) - return (200, get_group_summary) + return 200, get_group_summary class GroupSummaryRoomsCatServlet(RestServlet): @@ -120,7 +120,7 @@ class GroupSummaryRoomsCatServlet(RestServlet): content=content, ) - return (200, resp) + return 200, resp @defer.inlineCallbacks def on_DELETE(self, request, group_id, category_id, room_id): @@ -131,7 +131,7 @@ class GroupSummaryRoomsCatServlet(RestServlet): group_id, requester_user_id, room_id=room_id, category_id=category_id ) - return (200, resp) + return 200, resp class GroupCategoryServlet(RestServlet): @@ -157,7 +157,7 @@ class GroupCategoryServlet(RestServlet): group_id, requester_user_id, category_id=category_id ) - return (200, category) + return 200, category @defer.inlineCallbacks def on_PUT(self, request, group_id, category_id): @@ -169,7 +169,7 @@ class GroupCategoryServlet(RestServlet): group_id, requester_user_id, category_id=category_id, content=content ) - return (200, resp) + return 200, resp @defer.inlineCallbacks def on_DELETE(self, request, group_id, category_id): @@ -180,7 +180,7 @@ class GroupCategoryServlet(RestServlet): group_id, requester_user_id, category_id=category_id ) - return (200, resp) + return 200, resp class GroupCategoriesServlet(RestServlet): @@ -204,7 +204,7 @@ class GroupCategoriesServlet(RestServlet): group_id, requester_user_id ) - return (200, category) + return 200, category class GroupRoleServlet(RestServlet): @@ -228,7 +228,7 @@ class GroupRoleServlet(RestServlet): group_id, requester_user_id, role_id=role_id ) - return (200, category) + return 200, category @defer.inlineCallbacks def on_PUT(self, request, group_id, role_id): @@ -240,7 +240,7 @@ class GroupRoleServlet(RestServlet): group_id, requester_user_id, role_id=role_id, content=content ) - return (200, resp) + return 200, resp @defer.inlineCallbacks def on_DELETE(self, request, group_id, role_id): @@ -251,7 +251,7 @@ class GroupRoleServlet(RestServlet): group_id, requester_user_id, role_id=role_id ) - return (200, resp) + return 200, resp class GroupRolesServlet(RestServlet): @@ -275,7 +275,7 @@ class GroupRolesServlet(RestServlet): group_id, requester_user_id ) - return (200, category) + return 200, category class GroupSummaryUsersRoleServlet(RestServlet): @@ -312,7 +312,7 @@ class GroupSummaryUsersRoleServlet(RestServlet): content=content, ) - return (200, resp) + return 200, resp @defer.inlineCallbacks def on_DELETE(self, request, group_id, role_id, user_id): @@ -323,7 +323,7 @@ class GroupSummaryUsersRoleServlet(RestServlet): group_id, requester_user_id, user_id=user_id, role_id=role_id ) - return (200, resp) + return 200, resp class GroupRoomServlet(RestServlet): @@ -347,7 +347,7 @@ class GroupRoomServlet(RestServlet): group_id, requester_user_id ) - return (200, result) + return 200, result class GroupUsersServlet(RestServlet): @@ -371,7 +371,7 @@ class GroupUsersServlet(RestServlet): group_id, requester_user_id ) - return (200, result) + return 200, result class GroupInvitedUsersServlet(RestServlet): @@ -395,7 +395,7 @@ class GroupInvitedUsersServlet(RestServlet): group_id, requester_user_id ) - return (200, result) + return 200, result class GroupSettingJoinPolicyServlet(RestServlet): @@ -420,7 +420,7 @@ class GroupSettingJoinPolicyServlet(RestServlet): group_id, requester_user_id, content ) - return (200, result) + return 200, result class GroupCreateServlet(RestServlet): @@ -450,7 +450,7 @@ class GroupCreateServlet(RestServlet): group_id, requester_user_id, content ) - return (200, result) + return 200, result class GroupAdminRoomsServlet(RestServlet): @@ -477,7 +477,7 @@ class GroupAdminRoomsServlet(RestServlet): group_id, requester_user_id, room_id, content ) - return (200, result) + return 200, result @defer.inlineCallbacks def on_DELETE(self, request, group_id, room_id): @@ -488,7 +488,7 @@ class GroupAdminRoomsServlet(RestServlet): group_id, requester_user_id, room_id ) - return (200, result) + return 200, result class GroupAdminRoomsConfigServlet(RestServlet): @@ -516,7 +516,7 @@ class GroupAdminRoomsConfigServlet(RestServlet): group_id, requester_user_id, room_id, config_key, content ) - return (200, result) + return 200, result class GroupAdminUsersInviteServlet(RestServlet): @@ -546,7 +546,7 @@ class GroupAdminUsersInviteServlet(RestServlet): group_id, user_id, requester_user_id, config ) - return (200, result) + return 200, result class GroupAdminUsersKickServlet(RestServlet): @@ -573,7 +573,7 @@ class GroupAdminUsersKickServlet(RestServlet): group_id, user_id, requester_user_id, content ) - return (200, result) + return 200, result class GroupSelfLeaveServlet(RestServlet): @@ -598,7 +598,7 @@ class GroupSelfLeaveServlet(RestServlet): group_id, requester_user_id, requester_user_id, content ) - return (200, result) + return 200, result class GroupSelfJoinServlet(RestServlet): @@ -623,7 +623,7 @@ class GroupSelfJoinServlet(RestServlet): group_id, requester_user_id, content ) - return (200, result) + return 200, result class GroupSelfAcceptInviteServlet(RestServlet): @@ -648,7 +648,7 @@ class GroupSelfAcceptInviteServlet(RestServlet): group_id, requester_user_id, content ) - return (200, result) + return 200, result class GroupSelfUpdatePublicityServlet(RestServlet): @@ -672,7 +672,7 @@ class GroupSelfUpdatePublicityServlet(RestServlet): publicise = content["publicise"] yield self.store.update_group_publicity(group_id, requester_user_id, publicise) - return (200, {}) + return 200, {} class PublicisedGroupsForUserServlet(RestServlet): @@ -694,7 +694,7 @@ class PublicisedGroupsForUserServlet(RestServlet): result = yield self.groups_handler.get_publicised_groups_for_user(user_id) - return (200, result) + return 200, result class PublicisedGroupsForUsersServlet(RestServlet): @@ -719,7 +719,7 @@ class PublicisedGroupsForUsersServlet(RestServlet): result = yield self.groups_handler.bulk_get_publicised_groups(user_ids) - return (200, result) + return 200, result class GroupsForUserServlet(RestServlet): @@ -741,7 +741,7 @@ class GroupsForUserServlet(RestServlet): result = yield self.groups_handler.get_joined_groups(requester_user_id) - return (200, result) + return 200, result def register_servlets(hs, http_server): diff --git a/synapse/rest/client/v2_alpha/keys.py b/synapse/rest/client/v2_alpha/keys.py index b218a3f334..64b6898eb8 100644 --- a/synapse/rest/client/v2_alpha/keys.py +++ b/synapse/rest/client/v2_alpha/keys.py @@ -105,7 +105,7 @@ class KeyUploadServlet(RestServlet): result = yield self.e2e_keys_handler.upload_keys_for_user( user_id, device_id, body ) - return (200, result) + return 200, result class KeyQueryServlet(RestServlet): @@ -159,7 +159,7 @@ class KeyQueryServlet(RestServlet): timeout = parse_integer(request, "timeout", 10 * 1000) body = parse_json_object_from_request(request) result = yield self.e2e_keys_handler.query_devices(body, timeout) - return (200, result) + return 200, result class KeyChangesServlet(RestServlet): @@ -200,7 +200,7 @@ class KeyChangesServlet(RestServlet): results = yield self.device_handler.get_user_ids_changed(user_id, from_token) - return (200, results) + return 200, results class OneTimeKeyServlet(RestServlet): @@ -235,7 +235,7 @@ class OneTimeKeyServlet(RestServlet): timeout = parse_integer(request, "timeout", 10 * 1000) body = parse_json_object_from_request(request) result = yield self.e2e_keys_handler.claim_one_time_keys(body, timeout) - return (200, result) + return 200, result def register_servlets(hs, http_server): diff --git a/synapse/rest/client/v2_alpha/notifications.py b/synapse/rest/client/v2_alpha/notifications.py index d034863a3c..10c1ad5b07 100644 --- a/synapse/rest/client/v2_alpha/notifications.py +++ b/synapse/rest/client/v2_alpha/notifications.py @@ -88,7 +88,7 @@ class NotificationsServlet(RestServlet): returned_push_actions.append(returned_pa) next_token = str(pa["stream_ordering"]) - return (200, {"notifications": returned_push_actions, "next_token": next_token}) + return 200, {"notifications": returned_push_actions, "next_token": next_token} def register_servlets(hs, http_server): diff --git a/synapse/rest/client/v2_alpha/read_marker.py b/synapse/rest/client/v2_alpha/read_marker.py index d93d6a9f24..b3bf8567e1 100644 --- a/synapse/rest/client/v2_alpha/read_marker.py +++ b/synapse/rest/client/v2_alpha/read_marker.py @@ -59,7 +59,7 @@ class ReadMarkerRestServlet(RestServlet): event_id=read_marker_event_id, ) - return (200, {}) + return 200, {} def register_servlets(hs, http_server): diff --git a/synapse/rest/client/v2_alpha/receipts.py b/synapse/rest/client/v2_alpha/receipts.py index 98a97b7059..0dab03d227 100644 --- a/synapse/rest/client/v2_alpha/receipts.py +++ b/synapse/rest/client/v2_alpha/receipts.py @@ -52,7 +52,7 @@ class ReceiptRestServlet(RestServlet): room_id, receipt_type, user_id=requester.user.to_string(), event_id=event_id ) - return (200, {}) + return 200, {} def register_servlets(hs, http_server): diff --git a/synapse/rest/client/v2_alpha/register.py b/synapse/rest/client/v2_alpha/register.py index 9510a1e2b0..65f9fce2ff 100644 --- a/synapse/rest/client/v2_alpha/register.py +++ b/synapse/rest/client/v2_alpha/register.py @@ -94,7 +94,7 @@ class EmailRegisterRequestTokenRestServlet(RestServlet): raise SynapseError(400, "Email is already in use", Codes.THREEPID_IN_USE) ret = yield self.identity_handler.requestEmailToken(**body) - return (200, ret) + return 200, ret class MsisdnRegisterRequestTokenRestServlet(RestServlet): @@ -137,7 +137,7 @@ class MsisdnRegisterRequestTokenRestServlet(RestServlet): ) ret = yield self.identity_handler.requestMsisdnToken(**body) - return (200, ret) + return 200, ret class UsernameAvailabilityRestServlet(RestServlet): @@ -177,7 +177,7 @@ class UsernameAvailabilityRestServlet(RestServlet): yield self.registration_handler.check_username(username) - return (200, {"available": True}) + return 200, {"available": True} class RegisterRestServlet(RestServlet): @@ -279,7 +279,7 @@ class RegisterRestServlet(RestServlet): result = yield self._do_appservice_registration( desired_username, access_token, body ) - return (200, result) # we throw for non 200 responses + return 200, result # we throw for non 200 responses return # for regular registration, downcase the provided username before @@ -487,7 +487,7 @@ class RegisterRestServlet(RestServlet): bind_msisdn=params.get("bind_msisdn"), ) - return (200, return_dict) + return 200, return_dict def on_OPTIONS(self, _): return 200, {} diff --git a/synapse/rest/client/v2_alpha/relations.py b/synapse/rest/client/v2_alpha/relations.py index 1538b247e5..040b37c504 100644 --- a/synapse/rest/client/v2_alpha/relations.py +++ b/synapse/rest/client/v2_alpha/relations.py @@ -118,7 +118,7 @@ class RelationSendServlet(RestServlet): requester, event_dict=event_dict, txn_id=txn_id ) - return (200, {"event_id": event.event_id}) + return 200, {"event_id": event.event_id} class RelationPaginationServlet(RestServlet): @@ -198,7 +198,7 @@ class RelationPaginationServlet(RestServlet): return_value["chunk"] = events return_value["original_event"] = original_event - return (200, return_value) + return 200, return_value class RelationAggregationPaginationServlet(RestServlet): @@ -270,7 +270,7 @@ class RelationAggregationPaginationServlet(RestServlet): to_token=to_token, ) - return (200, pagination_chunk.to_dict()) + return 200, pagination_chunk.to_dict() class RelationAggregationGroupPaginationServlet(RestServlet): @@ -356,7 +356,7 @@ class RelationAggregationGroupPaginationServlet(RestServlet): return_value = result.to_dict() return_value["chunk"] = events - return (200, return_value) + return 200, return_value def register_servlets(hs, http_server): diff --git a/synapse/rest/client/v2_alpha/report_event.py b/synapse/rest/client/v2_alpha/report_event.py index 3fdd4584a3..e7449864cd 100644 --- a/synapse/rest/client/v2_alpha/report_event.py +++ b/synapse/rest/client/v2_alpha/report_event.py @@ -72,7 +72,7 @@ class ReportEventRestServlet(RestServlet): received_ts=self.clock.time_msec(), ) - return (200, {}) + return 200, {} def register_servlets(hs, http_server): diff --git a/synapse/rest/client/v2_alpha/room_keys.py b/synapse/rest/client/v2_alpha/room_keys.py index 10dec96208..df4f44cd36 100644 --- a/synapse/rest/client/v2_alpha/room_keys.py +++ b/synapse/rest/client/v2_alpha/room_keys.py @@ -135,7 +135,7 @@ class RoomKeysServlet(RestServlet): body = {"rooms": {room_id: body}} yield self.e2e_room_keys_handler.upload_room_keys(user_id, version, body) - return (200, {}) + return 200, {} @defer.inlineCallbacks def on_GET(self, request, room_id, session_id): @@ -218,7 +218,7 @@ class RoomKeysServlet(RestServlet): else: room_keys = room_keys["rooms"][room_id] - return (200, room_keys) + return 200, room_keys @defer.inlineCallbacks def on_DELETE(self, request, room_id, session_id): @@ -242,7 +242,7 @@ class RoomKeysServlet(RestServlet): yield self.e2e_room_keys_handler.delete_room_keys( user_id, version, room_id, session_id ) - return (200, {}) + return 200, {} class RoomKeysNewVersionServlet(RestServlet): @@ -293,7 +293,7 @@ class RoomKeysNewVersionServlet(RestServlet): info = parse_json_object_from_request(request) new_version = yield self.e2e_room_keys_handler.create_version(user_id, info) - return (200, {"version": new_version}) + return 200, {"version": new_version} # we deliberately don't have a PUT /version, as these things really should # be immutable to avoid people footgunning @@ -338,7 +338,7 @@ class RoomKeysVersionServlet(RestServlet): except SynapseError as e: if e.code == 404: raise SynapseError(404, "No backup found", Codes.NOT_FOUND) - return (200, info) + return 200, info @defer.inlineCallbacks def on_DELETE(self, request, version): @@ -358,7 +358,7 @@ class RoomKeysVersionServlet(RestServlet): user_id = requester.user.to_string() yield self.e2e_room_keys_handler.delete_version(user_id, version) - return (200, {}) + return 200, {} @defer.inlineCallbacks def on_PUT(self, request, version): @@ -392,7 +392,7 @@ class RoomKeysVersionServlet(RestServlet): ) yield self.e2e_room_keys_handler.update_version(user_id, version, info) - return (200, {}) + return 200, {} def register_servlets(hs, http_server): diff --git a/synapse/rest/client/v2_alpha/room_upgrade_rest_servlet.py b/synapse/rest/client/v2_alpha/room_upgrade_rest_servlet.py index 14ba61a63e..d2c3316eb7 100644 --- a/synapse/rest/client/v2_alpha/room_upgrade_rest_servlet.py +++ b/synapse/rest/client/v2_alpha/room_upgrade_rest_servlet.py @@ -80,7 +80,7 @@ class RoomUpgradeRestServlet(RestServlet): ret = {"replacement_room": new_room_id} - return (200, ret) + return 200, ret def register_servlets(hs, http_server): diff --git a/synapse/rest/client/v2_alpha/sync.py b/synapse/rest/client/v2_alpha/sync.py index 7b32dd2212..c98c5a3802 100644 --- a/synapse/rest/client/v2_alpha/sync.py +++ b/synapse/rest/client/v2_alpha/sync.py @@ -174,7 +174,7 @@ class SyncRestServlet(RestServlet): time_now, sync_result, requester.access_token_id, filter ) - return (200, response_content) + return 200, response_content @defer.inlineCallbacks def encode_response(self, time_now, sync_result, access_token_id, filter): diff --git a/synapse/rest/client/v2_alpha/tags.py b/synapse/rest/client/v2_alpha/tags.py index d173544355..3b555669a0 100644 --- a/synapse/rest/client/v2_alpha/tags.py +++ b/synapse/rest/client/v2_alpha/tags.py @@ -45,7 +45,7 @@ class TagListServlet(RestServlet): tags = yield self.store.get_tags_for_room(user_id, room_id) - return (200, {"tags": tags}) + return 200, {"tags": tags} class TagServlet(RestServlet): @@ -76,7 +76,7 @@ class TagServlet(RestServlet): self.notifier.on_new_event("account_data_key", max_id, users=[user_id]) - return (200, {}) + return 200, {} @defer.inlineCallbacks def on_DELETE(self, request, user_id, room_id, tag): @@ -88,7 +88,7 @@ class TagServlet(RestServlet): self.notifier.on_new_event("account_data_key", max_id, users=[user_id]) - return (200, {}) + return 200, {} def register_servlets(hs, http_server): diff --git a/synapse/rest/client/v2_alpha/thirdparty.py b/synapse/rest/client/v2_alpha/thirdparty.py index 158e686b01..2e8d672471 100644 --- a/synapse/rest/client/v2_alpha/thirdparty.py +++ b/synapse/rest/client/v2_alpha/thirdparty.py @@ -40,7 +40,7 @@ class ThirdPartyProtocolsServlet(RestServlet): yield self.auth.get_user_by_req(request, allow_guest=True) protocols = yield self.appservice_handler.get_3pe_protocols() - return (200, protocols) + return 200, protocols class ThirdPartyProtocolServlet(RestServlet): @@ -60,9 +60,9 @@ class ThirdPartyProtocolServlet(RestServlet): only_protocol=protocol ) if protocol in protocols: - return (200, protocols[protocol]) + return 200, protocols[protocol] else: - return (404, {"error": "Unknown protocol"}) + return 404, {"error": "Unknown protocol"} class ThirdPartyUserServlet(RestServlet): @@ -85,7 +85,7 @@ class ThirdPartyUserServlet(RestServlet): ThirdPartyEntityKind.USER, protocol, fields ) - return (200, results) + return 200, results class ThirdPartyLocationServlet(RestServlet): @@ -108,7 +108,7 @@ class ThirdPartyLocationServlet(RestServlet): ThirdPartyEntityKind.LOCATION, protocol, fields ) - return (200, results) + return 200, results def register_servlets(hs, http_server): diff --git a/synapse/rest/client/v2_alpha/user_directory.py b/synapse/rest/client/v2_alpha/user_directory.py index 7ab2b80e46..2863affbab 100644 --- a/synapse/rest/client/v2_alpha/user_directory.py +++ b/synapse/rest/client/v2_alpha/user_directory.py @@ -60,7 +60,7 @@ class UserDirectorySearchRestServlet(RestServlet): user_id = requester.user.to_string() if not self.hs.config.user_directory_search_enabled: - return (200, {"limited": False, "results": []}) + return 200, {"limited": False, "results": []} body = parse_json_object_from_request(request) @@ -76,7 +76,7 @@ class UserDirectorySearchRestServlet(RestServlet): user_id, search_term, limit ) - return (200, results) + return 200, results def register_servlets(hs, http_server): diff --git a/synapse/rest/media/v1/media_repository.py b/synapse/rest/media/v1/media_repository.py index cf5759e9a6..d4ea09260c 100644 --- a/synapse/rest/media/v1/media_repository.py +++ b/synapse/rest/media/v1/media_repository.py @@ -318,14 +318,14 @@ class MediaRepository(object): responder = yield self.media_storage.fetch_media(file_info) if responder: - return (responder, media_info) + return responder, media_info # Failed to find the file anywhere, lets download it. media_info = yield self._download_remote_file(server_name, media_id, file_id) responder = yield self.media_storage.fetch_media(file_info) - return (responder, media_info) + return responder, media_info @defer.inlineCallbacks def _download_remote_file(self, server_name, media_id, file_id): diff --git a/synapse/rest/media/v1/thumbnailer.py b/synapse/rest/media/v1/thumbnailer.py index 90d8e6bffe..c995d7e043 100644 --- a/synapse/rest/media/v1/thumbnailer.py +++ b/synapse/rest/media/v1/thumbnailer.py @@ -78,9 +78,9 @@ class Thumbnailer(object): """ if max_width * self.height < max_height * self.width: - return (max_width, (max_width * self.height) // self.width) + return max_width, (max_width * self.height) // self.width else: - return ((max_height * self.width) // self.height, max_height) + return (max_height * self.width) // self.height, max_height def scale(self, width, height, output_type): """Rescales the image to the given dimensions. diff --git a/synapse/server_notices/resource_limits_server_notices.py b/synapse/server_notices/resource_limits_server_notices.py index 729c097e6d..81c4aff496 100644 --- a/synapse/server_notices/resource_limits_server_notices.py +++ b/synapse/server_notices/resource_limits_server_notices.py @@ -193,4 +193,4 @@ class ResourceLimitsServerNotices(object): if event_id in referenced_events: referenced_events.remove(event.event_id) - return (currently_blocked, referenced_events) + return currently_blocked, referenced_events diff --git a/synapse/storage/account_data.py b/synapse/storage/account_data.py index 9fa5b4f3d6..6afbfc0d74 100644 --- a/synapse/storage/account_data.py +++ b/synapse/storage/account_data.py @@ -90,7 +90,7 @@ class AccountDataWorkerStore(SQLBaseStore): room_data = by_room.setdefault(row["room_id"], {}) room_data[row["account_data_type"]] = json.loads(row["content"]) - return (global_account_data, by_room) + return global_account_data, by_room return self.runInteraction( "get_account_data_for_user", get_account_data_for_user_txn @@ -205,7 +205,7 @@ class AccountDataWorkerStore(SQLBaseStore): ) txn.execute(sql, (last_room_id, current_id, limit)) room_results = txn.fetchall() - return (global_results, room_results) + return global_results, room_results return self.runInteraction( "get_all_updated_account_data_txn", get_updated_account_data_txn @@ -244,13 +244,13 @@ class AccountDataWorkerStore(SQLBaseStore): room_account_data = account_data_by_room.setdefault(row[0], {}) room_account_data[row[1]] = json.loads(row[2]) - return (global_account_data, account_data_by_room) + return global_account_data, account_data_by_room changed = self._account_data_stream_cache.has_entity_changed( user_id, int(stream_id) ) if not changed: - return ({}, {}) + return {}, {} return self.runInteraction( "get_updated_account_data_for_user", get_updated_account_data_for_user_txn diff --git a/synapse/storage/appservice.py b/synapse/storage/appservice.py index 05d9c05c3f..36657753cd 100644 --- a/synapse/storage/appservice.py +++ b/synapse/storage/appservice.py @@ -358,7 +358,7 @@ class ApplicationServiceTransactionWorkerStore( events = yield self.get_events_as_list(event_ids) - return (upper_bound, events) + return upper_bound, events class ApplicationServiceTransactionStore(ApplicationServiceTransactionWorkerStore): diff --git a/synapse/storage/deviceinbox.py b/synapse/storage/deviceinbox.py index 79bb0ea46d..4dca9de617 100644 --- a/synapse/storage/deviceinbox.py +++ b/synapse/storage/deviceinbox.py @@ -66,7 +66,7 @@ class DeviceInboxWorkerStore(SQLBaseStore): messages.append(json.loads(row[1])) if len(messages) < limit: stream_pos = current_stream_id - return (messages, stream_pos) + return messages, stream_pos return self.runInteraction( "get_new_messages_for_device", get_new_messages_for_device_txn @@ -157,7 +157,7 @@ class DeviceInboxWorkerStore(SQLBaseStore): messages.append(json.loads(row[1])) if len(messages) < limit: stream_pos = current_stream_id - return (messages, stream_pos) + return messages, stream_pos return self.runInteraction( "get_new_device_msgs_for_remote", diff --git a/synapse/storage/devices.py b/synapse/storage/devices.py index e11881161d..76542c512d 100644 --- a/synapse/storage/devices.py +++ b/synapse/storage/devices.py @@ -94,7 +94,7 @@ class DeviceWorkerStore(SQLBaseStore): destination, int(from_stream_id) ) if not has_changed: - return (now_stream_id, []) + return now_stream_id, [] # We retrieve n+1 devices from the list of outbound pokes where n is # our outbound device update limit. We then check if the very last @@ -117,7 +117,7 @@ class DeviceWorkerStore(SQLBaseStore): # Return an empty list if there are no updates if not updates: - return (now_stream_id, []) + return now_stream_id, [] # if we have exceeded the limit, we need to exclude any results with the # same stream_id as the last row. @@ -167,13 +167,13 @@ class DeviceWorkerStore(SQLBaseStore): # skip that stream_id and return an empty list, and continue with the next # stream_id next time. if not query_map: - return (stream_id_cutoff, []) + return stream_id_cutoff, [] results = yield self._get_device_update_edus_by_remote( destination, from_stream_id, query_map ) - return (now_stream_id, results) + return now_stream_id, results def _get_devices_by_remote_txn( self, txn, destination, from_stream_id, now_stream_id, limit @@ -352,7 +352,7 @@ class DeviceWorkerStore(SQLBaseStore): else: results[user_id] = yield self._get_cached_devices_for_user(user_id) - return (user_ids_not_in_cache, results) + return user_ids_not_in_cache, results @cachedInlineCallbacks(num_args=2, tree=True) def _get_cached_user_device(self, user_id, device_id): diff --git a/synapse/storage/events.py b/synapse/storage/events.py index 5a95c36a8b..32050868ff 100644 --- a/synapse/storage/events.py +++ b/synapse/storage/events.py @@ -810,7 +810,7 @@ class EventsStore( # If they old and new groups are the same then we don't need to do # anything. if old_state_groups == new_state_groups: - return (None, None) + return None, None if len(new_state_groups) == 1 and len(old_state_groups) == 1: # If we're going from one state group to another, lets check if @@ -827,7 +827,7 @@ class EventsStore( # the current state in memory then lets also return that, # but it doesn't matter if we don't. new_state = state_groups_map.get(new_state_group) - return (new_state, delta_ids) + return new_state, delta_ids # Now that we have calculated new_state_groups we need to get # their state IDs so we can resolve to a single state set. @@ -839,7 +839,7 @@ class EventsStore( if len(new_state_groups) == 1: # If there is only one state group, then we know what the current # state is. - return (state_groups_map[new_state_groups.pop()], None) + return state_groups_map[new_state_groups.pop()], None # Ok, we need to defer to the state handler to resolve our state sets. @@ -868,7 +868,7 @@ class EventsStore( state_res_store=StateResolutionStore(self), ) - return (res.state, None) + return res.state, None @defer.inlineCallbacks def _calculate_state_delta(self, room_id, current_state): @@ -891,7 +891,7 @@ class EventsStore( if ev_id != existing_state.get(key) } - return (to_delete, to_insert) + return to_delete, to_insert @log_function def _persist_events_txn( diff --git a/synapse/storage/presence.py b/synapse/storage/presence.py index 1a0f2d5768..5db6f2d84a 100644 --- a/synapse/storage/presence.py +++ b/synapse/storage/presence.py @@ -90,7 +90,7 @@ class PresenceStore(SQLBaseStore): presence_states, ) - return (stream_orderings[-1], self._presence_id_gen.get_current_token()) + return stream_orderings[-1], self._presence_id_gen.get_current_token() def _update_presence_txn(self, txn, stream_orderings, presence_states): for stream_id, state in zip(stream_orderings, presence_states): diff --git a/synapse/storage/pusher.py b/synapse/storage/pusher.py index b431d24b8a..3e0e834a62 100644 --- a/synapse/storage/pusher.py +++ b/synapse/storage/pusher.py @@ -133,7 +133,7 @@ class PusherWorkerStore(SQLBaseStore): txn.execute(sql, (last_id, current_id, limit)) deleted = txn.fetchall() - return (updated, deleted) + return updated, deleted return self.runInteraction( "get_all_updated_pushers", get_all_updated_pushers_txn diff --git a/synapse/storage/receipts.py b/synapse/storage/receipts.py index 6aa6d98ebb..290ddb30e8 100644 --- a/synapse/storage/receipts.py +++ b/synapse/storage/receipts.py @@ -478,7 +478,7 @@ class ReceiptsStore(ReceiptsWorkerStore): max_persisted_id = self._receipts_id_gen.get_current_token() - return (stream_id, max_persisted_id) + return stream_id, max_persisted_id def insert_graph_receipt(self, room_id, receipt_type, user_id, event_ids, data): return self.runInteraction( diff --git a/synapse/storage/stream.py b/synapse/storage/stream.py index 856c2ee8d8..490454f19a 100644 --- a/synapse/storage/stream.py +++ b/synapse/storage/stream.py @@ -364,7 +364,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): the chunk of events returned. """ if from_key == to_key: - return ([], from_key) + return [], from_key from_id = RoomStreamToken.parse_stream_token(from_key).stream to_id = RoomStreamToken.parse_stream_token(to_key).stream @@ -374,7 +374,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): ) if not has_changed: - return ([], from_key) + return [], from_key def f(txn): sql = ( @@ -407,7 +407,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): # get. key = from_key - return (ret, key) + return ret, key @defer.inlineCallbacks def get_membership_changes_for_user(self, user_id, from_key, to_key): @@ -496,7 +496,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): """ # Allow a zero limit here, and no-op. if limit == 0: - return ([], end_token) + return [], end_token end_token = RoomStreamToken.parse(end_token) @@ -511,7 +511,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): # We want to return the results in ascending order. rows.reverse() - return (rows, token) + return rows, token def get_room_event_after_stream_ordering(self, room_id, stream_ordering): """Gets details of the first event in a room at or after a stream ordering @@ -783,7 +783,7 @@ class StreamWorkerStore(EventsWorkerStore, SQLBaseStore): events = yield self.get_events_as_list(event_ids) - return (upper_bound, events) + return upper_bound, events def get_federation_out_pos(self, typ): return self._simple_select_one_onecol( diff --git a/synapse/storage/util/id_generators.py b/synapse/storage/util/id_generators.py index f1c8d99419..cbb0a4810a 100644 --- a/synapse/storage/util/id_generators.py +++ b/synapse/storage/util/id_generators.py @@ -195,6 +195,6 @@ class ChainedIdGenerator(object): with self._lock: if self._unfinished_ids: stream_id, chained_id = self._unfinished_ids[0] - return (stream_id - 1, chained_id) + return stream_id - 1, chained_id - return (self._current_max, self.chained_generator.get_current_token()) + return self._current_max, self.chained_generator.get_current_token() diff --git a/synapse/streams/config.py b/synapse/streams/config.py index f7f5906a99..02994ab2a5 100644 --- a/synapse/streams/config.py +++ b/synapse/streams/config.py @@ -37,7 +37,7 @@ class SourcePaginationConfig(object): self.limit = min(int(limit), MAX_LIMIT) if limit is not None else None def __repr__(self): - return ("StreamConfig(from_key=%r, to_key=%r, direction=%r, limit=%r)") % ( + return "StreamConfig(from_key=%r, to_key=%r, direction=%r, limit=%r)" % ( self.from_key, self.to_key, self.direction, diff --git a/tests/handlers/test_register.py b/tests/handlers/test_register.py index 0ad0a88165..e10296a5e4 100644 --- a/tests/handlers/test_register.py +++ b/tests/handlers/test_register.py @@ -283,4 +283,4 @@ class RegistrationTestCase(unittest.HomeserverTestCase): user, requester, displayname, by_admin=True ) - return (user_id, token) + return user_id, token diff --git a/tests/rest/client/v2_alpha/test_register.py b/tests/rest/client/v2_alpha/test_register.py index bb867150f4..ab4d7d70d0 100644 --- a/tests/rest/client/v2_alpha/test_register.py +++ b/tests/rest/client/v2_alpha/test_register.py @@ -472,7 +472,7 @@ class AccountValidityRenewalByEmailTestCase(unittest.HomeserverTestCase): added_at=now, ) ) - return (user_id, tok) + return user_id, tok def test_manual_email_send_expired_account(self): user_id = self.register_user("kermit", "monkey") diff --git a/tests/server.py b/tests/server.py index c8269619b1..e397ebe8fa 100644 --- a/tests/server.py +++ b/tests/server.py @@ -338,7 +338,7 @@ def setup_test_homeserver(cleanup_func, *args, **kwargs): def get_clock(): clock = ThreadedMemoryReactorClock() hs_clock = Clock(clock) - return (clock, hs_clock) + return clock, hs_clock @attr.s(cmp=False) diff --git a/tests/test_server.py b/tests/test_server.py index 2a7d407c98..98fef21d55 100644 --- a/tests/test_server.py +++ b/tests/test_server.py @@ -57,7 +57,7 @@ class JsonResourceTests(unittest.TestCase): def _callback(request, **kwargs): got_kwargs.update(kwargs) - return (200, kwargs) + return 200, kwargs res = JsonResource(self.homeserver) res.register_paths( diff --git a/tests/test_state.py b/tests/test_state.py index 6d33566f47..610ec9fb46 100644 --- a/tests/test_state.py +++ b/tests/test_state.py @@ -106,7 +106,7 @@ class StateGroupStore(object): } def get_state_group_delta(self, name): - return (None, None) + return None, None def register_events(self, events): for e in events: diff --git a/tests/utils.py b/tests/utils.py index f1eb9a545c..46ef2959f2 100644 --- a/tests/utils.py +++ b/tests/utils.py @@ -464,7 +464,7 @@ class MockHttpResource(HttpServer): args = [urlparse.unquote(u) for u in matcher.groups()] (code, response) = yield func(mock_request, *args) - return (code, response) + return code, response except CodeMessageException as e: return (e.code, cs_error(e.msg, code=e.errcode)) -- cgit 1.5.1 From 6e834e94fcc97811e4cc8185e86c6b9da06eb28e Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 4 Sep 2019 13:04:27 +0100 Subject: Fix and refactor room and user stats (#5971) Previously the stats were not being correctly populated. --- changelog.d/5971.bugfix | 1 + docs/room_and_user_statistics.md | 62 ++ synapse/config/stats.py | 13 +- synapse/handlers/stats.py | 307 +++--- synapse/storage/events.py | 5 +- synapse/storage/registration.py | 12 + synapse/storage/roommember.py | 44 +- .../storage/schema/delta/56/stats_separated.sql | 152 +++ synapse/storage/stats.py | 1036 ++++++++++++++------ tests/handlers/test_stats.py | 643 +++++++++--- tests/rest/client/v1/utils.py | 8 +- 11 files changed, 1642 insertions(+), 641 deletions(-) create mode 100644 changelog.d/5971.bugfix create mode 100644 docs/room_and_user_statistics.md create mode 100644 synapse/storage/schema/delta/56/stats_separated.sql (limited to 'tests') diff --git a/changelog.d/5971.bugfix b/changelog.d/5971.bugfix new file mode 100644 index 0000000000..9ea095103b --- /dev/null +++ b/changelog.d/5971.bugfix @@ -0,0 +1 @@ +Fix room and user stats tracking. diff --git a/docs/room_and_user_statistics.md b/docs/room_and_user_statistics.md new file mode 100644 index 0000000000..e1facb38d4 --- /dev/null +++ b/docs/room_and_user_statistics.md @@ -0,0 +1,62 @@ +Room and User Statistics +======================== + +Synapse maintains room and user statistics (as well as a cache of room state), +in various tables. These can be used for administrative purposes but are also +used when generating the public room directory. + + +# Synapse Developer Documentation + +## High-Level Concepts + +### Definitions + +* **subject**: Something we are tracking stats about – currently a room or user. +* **current row**: An entry for a subject in the appropriate current statistics + table. Each subject can have only one. +* **historical row**: An entry for a subject in the appropriate historical + statistics table. Each subject can have any number of these. + +### Overview + +Stats are maintained as time series. There are two kinds of column: + +* absolute columns – where the value is correct for the time given by `end_ts` + in the stats row. (Imagine a line graph for these values) + * They can also be thought of as 'gauges' in Prometheus, if you are familiar. +* per-slice columns – where the value corresponds to how many of the occurrences + occurred within the time slice given by `(end_ts − bucket_size)…end_ts` + or `start_ts…end_ts`. (Imagine a histogram for these values) + +Stats are maintained in two tables (for each type): current and historical. + +Current stats correspond to the present values. Each subject can only have one +entry. + +Historical stats correspond to values in the past. Subjects may have multiple +entries. + +## Concepts around the management of stats + +### Current rows + +Current rows contain the most up-to-date statistics for a room. +They only contain absolute columns + +### Historical rows + +Historical rows can always be considered to be valid for the time slice and +end time specified. + +* historical rows will not exist for every time slice – they will be omitted + if there were no changes. In this case, the following assumptions can be + made to interpolate/recreate missing rows: + - absolute fields have the same values as in the preceding row + - per-slice fields are zero (`0`) +* historical rows will not be retained forever – rows older than a configurable + time will be purged. + +#### Purge + +The purging of historical rows is not yet implemented. diff --git a/synapse/config/stats.py b/synapse/config/stats.py index b518a3ed9c..b18ddbd1fa 100644 --- a/synapse/config/stats.py +++ b/synapse/config/stats.py @@ -27,19 +27,16 @@ class StatsConfig(Config): def read_config(self, config, **kwargs): self.stats_enabled = True - self.stats_bucket_size = 86400 + self.stats_bucket_size = 86400 * 1000 self.stats_retention = sys.maxsize stats_config = config.get("stats", None) if stats_config: self.stats_enabled = stats_config.get("enabled", self.stats_enabled) - self.stats_bucket_size = ( - self.parse_duration(stats_config.get("bucket_size", "1d")) / 1000 + self.stats_bucket_size = self.parse_duration( + stats_config.get("bucket_size", "1d") ) - self.stats_retention = ( - self.parse_duration( - stats_config.get("retention", "%ds" % (sys.maxsize,)) - ) - / 1000 + self.stats_retention = self.parse_duration( + stats_config.get("retention", "%ds" % (sys.maxsize,)) ) def generate_config_section(self, config_dir_path, server_name, **kwargs): diff --git a/synapse/handlers/stats.py b/synapse/handlers/stats.py index 4449da6669..921735edb3 100644 --- a/synapse/handlers/stats.py +++ b/synapse/handlers/stats.py @@ -14,15 +14,14 @@ # limitations under the License. import logging +from collections import Counter from twisted.internet import defer -from synapse.api.constants import EventTypes, JoinRules, Membership +from synapse.api.constants import EventTypes, Membership from synapse.handlers.state_deltas import StateDeltasHandler from synapse.metrics import event_processing_positions from synapse.metrics.background_process_metrics import run_as_background_process -from synapse.types import UserID -from synapse.util.metrics import Measure logger = logging.getLogger(__name__) @@ -62,11 +61,10 @@ class StatsHandler(StateDeltasHandler): def notify_new_event(self): """Called when there may be more deltas to process """ - if not self.hs.config.stats_enabled: + if not self.hs.config.stats_enabled or self._is_processing: return - if self._is_processing: - return + self._is_processing = True @defer.inlineCallbacks def process(): @@ -75,39 +73,72 @@ class StatsHandler(StateDeltasHandler): finally: self._is_processing = False - self._is_processing = True run_as_background_process("stats.notify_new_event", process) @defer.inlineCallbacks def _unsafe_process(self): # If self.pos is None then means we haven't fetched it from DB if self.pos is None: - self.pos = yield self.store.get_stats_stream_pos() - - # If still None then the initial background update hasn't happened yet - if self.pos is None: - return None + self.pos = yield self.store.get_stats_positions() # Loop round handling deltas until we're up to date + while True: - with Measure(self.clock, "stats_delta"): - deltas = yield self.store.get_current_state_deltas(self.pos) - if not deltas: - return + deltas = yield self.store.get_current_state_deltas(self.pos) + + if deltas: + logger.debug("Handling %d state deltas", len(deltas)) + room_deltas, user_deltas = yield self._handle_deltas(deltas) + + max_pos = deltas[-1]["stream_id"] + else: + room_deltas = {} + user_deltas = {} + max_pos = yield self.store.get_room_max_stream_ordering() - logger.info("Handling %d state deltas", len(deltas)) - yield self._handle_deltas(deltas) + # Then count deltas for total_events and total_event_bytes. + room_count, user_count = yield self.store.get_changes_room_total_events_and_bytes( + self.pos, max_pos + ) + + for room_id, fields in room_count.items(): + room_deltas.setdefault(room_id, {}).update(fields) + + for user_id, fields in user_count.items(): + user_deltas.setdefault(user_id, {}).update(fields) + + logger.debug("room_deltas: %s", room_deltas) + logger.debug("user_deltas: %s", user_deltas) - self.pos = deltas[-1]["stream_id"] - yield self.store.update_stats_stream_pos(self.pos) + # Always call this so that we update the stats position. + yield self.store.bulk_update_stats_delta( + self.clock.time_msec(), + updates={"room": room_deltas, "user": user_deltas}, + stream_id=max_pos, + ) + + event_processing_positions.labels("stats").set(max_pos) - event_processing_positions.labels("stats").set(self.pos) + if self.pos == max_pos: + break + + self.pos = max_pos @defer.inlineCallbacks def _handle_deltas(self, deltas): + """Called with the state deltas to process + + Returns: + Deferred[tuple[dict[str, Counter], dict[str, counter]]] + Resovles to two dicts, the room deltas and the user deltas, + mapping from room/user ID to changes in the various fields. """ - Called with the state deltas to process - """ + + room_to_stats_deltas = {} + user_to_stats_deltas = {} + + room_to_state_updates = {} + for delta in deltas: typ = delta["type"] state_key = delta["state_key"] @@ -115,11 +146,10 @@ class StatsHandler(StateDeltasHandler): event_id = delta["event_id"] stream_id = delta["stream_id"] prev_event_id = delta["prev_event_id"] - stream_pos = delta["stream_id"] - logger.debug("Handling: %r %r, %s", typ, state_key, event_id) + logger.debug("Handling: %r, %r %r, %s", room_id, typ, state_key, event_id) - token = yield self.store.get_earliest_token_for_room_stats(room_id) + token = yield self.store.get_earliest_token_for_stats("room", room_id) # If the earliest token to begin from is larger than our current # stream ID, skip processing this delta. @@ -131,203 +161,130 @@ class StatsHandler(StateDeltasHandler): continue if event_id is None and prev_event_id is None: - # Errr... + logger.error( + "event ID is None and so is the previous event ID. stream_id: %s", + stream_id, + ) continue event_content = {} + sender = None if event_id is not None: event = yield self.store.get_event(event_id, allow_none=True) if event: event_content = event.content or {} + sender = event.sender + + # All the values in this dict are deltas (RELATIVE changes) + room_stats_delta = room_to_stats_deltas.setdefault(room_id, Counter()) - # We use stream_pos here rather than fetch by event_id as event_id - # may be None - now = yield self.store.get_received_ts_by_stream_pos(stream_pos) + room_state = room_to_state_updates.setdefault(room_id, {}) - # quantise time to the nearest bucket - now = (now // 1000 // self.stats_bucket_size) * self.stats_bucket_size + if prev_event_id is None: + # this state event doesn't overwrite another, + # so it is a new effective/current state event + room_stats_delta["current_state_events"] += 1 if typ == EventTypes.Member: # we could use _get_key_change here but it's a bit inefficient # given we're not testing for a specific result; might as well # just grab the prev_membership and membership strings and # compare them. - prev_event_content = {} + # We take None rather than leave as a previous membership + # in the absence of a previous event because we do not want to + # reduce the leave count when a new-to-the-room user joins. + prev_membership = None if prev_event_id is not None: prev_event = yield self.store.get_event( prev_event_id, allow_none=True ) if prev_event: prev_event_content = prev_event.content + prev_membership = prev_event_content.get( + "membership", Membership.LEAVE + ) membership = event_content.get("membership", Membership.LEAVE) - prev_membership = prev_event_content.get("membership", Membership.LEAVE) - - if prev_membership == membership: - continue - if prev_membership == Membership.JOIN: - yield self.store.update_stats_delta( - now, "room", room_id, "joined_members", -1 - ) + if prev_membership is None: + logger.debug("No previous membership for this user.") + elif membership == prev_membership: + pass # noop + elif prev_membership == Membership.JOIN: + room_stats_delta["joined_members"] -= 1 elif prev_membership == Membership.INVITE: - yield self.store.update_stats_delta( - now, "room", room_id, "invited_members", -1 - ) + room_stats_delta["invited_members"] -= 1 elif prev_membership == Membership.LEAVE: - yield self.store.update_stats_delta( - now, "room", room_id, "left_members", -1 - ) + room_stats_delta["left_members"] -= 1 elif prev_membership == Membership.BAN: - yield self.store.update_stats_delta( - now, "room", room_id, "banned_members", -1 - ) + room_stats_delta["banned_members"] -= 1 else: - err = "%s is not a valid prev_membership" % (repr(prev_membership),) - logger.error(err) - raise ValueError(err) + raise ValueError( + "%r is not a valid prev_membership" % (prev_membership,) + ) + if membership == prev_membership: + pass # noop if membership == Membership.JOIN: - yield self.store.update_stats_delta( - now, "room", room_id, "joined_members", +1 - ) + room_stats_delta["joined_members"] += 1 elif membership == Membership.INVITE: - yield self.store.update_stats_delta( - now, "room", room_id, "invited_members", +1 - ) + room_stats_delta["invited_members"] += 1 + + if sender and self.is_mine_id(sender): + user_to_stats_deltas.setdefault(sender, Counter())[ + "invites_sent" + ] += 1 + elif membership == Membership.LEAVE: - yield self.store.update_stats_delta( - now, "room", room_id, "left_members", +1 - ) + room_stats_delta["left_members"] += 1 elif membership == Membership.BAN: - yield self.store.update_stats_delta( - now, "room", room_id, "banned_members", +1 - ) + room_stats_delta["banned_members"] += 1 else: - err = "%s is not a valid membership" % (repr(membership),) - logger.error(err) - raise ValueError(err) + raise ValueError("%r is not a valid membership" % (membership,)) user_id = state_key if self.is_mine_id(user_id): - # update user_stats as it's one of our users - public = yield self._is_public_room(room_id) - - if membership == Membership.LEAVE: - yield self.store.update_stats_delta( - now, - "user", - user_id, - "public_rooms" if public else "private_rooms", - -1, - ) - elif membership == Membership.JOIN: - yield self.store.update_stats_delta( - now, - "user", - user_id, - "public_rooms" if public else "private_rooms", - +1, - ) + # this accounts for transitions like leave → ban and so on. + has_changed_joinedness = (prev_membership == Membership.JOIN) != ( + membership == Membership.JOIN + ) - elif typ == EventTypes.Create: - # Newly created room. Add it with all blank portions. - yield self.store.update_room_state( - room_id, - { - "join_rules": None, - "history_visibility": None, - "encryption": None, - "name": None, - "topic": None, - "avatar": None, - "canonical_alias": None, - }, - ) + if has_changed_joinedness: + delta = +1 if membership == Membership.JOIN else -1 - elif typ == EventTypes.JoinRules: - yield self.store.update_room_state( - room_id, {"join_rules": event_content.get("join_rule")} - ) + user_to_stats_deltas.setdefault(user_id, Counter())[ + "joined_rooms" + ] += delta - is_public = yield self._get_key_change( - prev_event_id, event_id, "join_rule", JoinRules.PUBLIC - ) - if is_public is not None: - yield self.update_public_room_stats(now, room_id, is_public) + room_stats_delta["local_users_in_room"] += delta + elif typ == EventTypes.Create: + room_state["is_federatable"] = event_content.get("m.federate", True) + if sender and self.is_mine_id(sender): + user_to_stats_deltas.setdefault(sender, Counter())[ + "rooms_created" + ] += 1 + elif typ == EventTypes.JoinRules: + room_state["join_rules"] = event_content.get("join_rule") elif typ == EventTypes.RoomHistoryVisibility: - yield self.store.update_room_state( - room_id, - {"history_visibility": event_content.get("history_visibility")}, - ) - - is_public = yield self._get_key_change( - prev_event_id, event_id, "history_visibility", "world_readable" + room_state["history_visibility"] = event_content.get( + "history_visibility" ) - if is_public is not None: - yield self.update_public_room_stats(now, room_id, is_public) - elif typ == EventTypes.Encryption: - yield self.store.update_room_state( - room_id, {"encryption": event_content.get("algorithm")} - ) + room_state["encryption"] = event_content.get("algorithm") elif typ == EventTypes.Name: - yield self.store.update_room_state( - room_id, {"name": event_content.get("name")} - ) + room_state["name"] = event_content.get("name") elif typ == EventTypes.Topic: - yield self.store.update_room_state( - room_id, {"topic": event_content.get("topic")} - ) + room_state["topic"] = event_content.get("topic") elif typ == EventTypes.RoomAvatar: - yield self.store.update_room_state( - room_id, {"avatar": event_content.get("url")} - ) + room_state["avatar"] = event_content.get("url") elif typ == EventTypes.CanonicalAlias: - yield self.store.update_room_state( - room_id, {"canonical_alias": event_content.get("alias")} - ) + room_state["canonical_alias"] = event_content.get("alias") + elif typ == EventTypes.GuestAccess: + room_state["guest_access"] = event_content.get("guest_access") - @defer.inlineCallbacks - def update_public_room_stats(self, ts, room_id, is_public): - """ - Increment/decrement a user's number of public rooms when a room they are - in changes to/from public visibility. + for room_id, state in room_to_state_updates.items(): + yield self.store.update_room_state(room_id, state) - Args: - ts (int): Timestamp in seconds - room_id (str) - is_public (bool) - """ - # For now, blindly iterate over all local users in the room so that - # we can handle the whole problem of copying buckets over as needed - user_ids = yield self.store.get_users_in_room(room_id) - - for user_id in user_ids: - if self.hs.is_mine(UserID.from_string(user_id)): - yield self.store.update_stats_delta( - ts, "user", user_id, "public_rooms", +1 if is_public else -1 - ) - yield self.store.update_stats_delta( - ts, "user", user_id, "private_rooms", -1 if is_public else +1 - ) - - @defer.inlineCallbacks - def _is_public_room(self, room_id): - join_rules = yield self.state.get_current_state(room_id, EventTypes.JoinRules) - history_visibility = yield self.state.get_current_state( - room_id, EventTypes.RoomHistoryVisibility - ) - - if (join_rules and join_rules.content.get("join_rule") == JoinRules.PUBLIC) or ( - ( - history_visibility - and history_visibility.content.get("history_visibility") - == "world_readable" - ) - ): - return True - else: - return False + return room_to_stats_deltas, user_to_stats_deltas diff --git a/synapse/storage/events.py b/synapse/storage/events.py index 32050868ff..1958afe1d7 100644 --- a/synapse/storage/events.py +++ b/synapse/storage/events.py @@ -2270,8 +2270,9 @@ class EventsStore( "room_aliases", "room_depth", "room_memberships", - "room_state", - "room_stats", + "room_stats_state", + "room_stats_current", + "room_stats_historical", "room_stats_earliest_token", "rooms", "stream_ordering_to_exterm", diff --git a/synapse/storage/registration.py b/synapse/storage/registration.py index 3f50324253..2d3c7e2dc9 100644 --- a/synapse/storage/registration.py +++ b/synapse/storage/registration.py @@ -869,6 +869,17 @@ class RegistrationStore( (user_id_obj.localpart, create_profile_with_displayname), ) + if self.hs.config.stats_enabled: + # we create a new completed user statistics row + + # we don't strictly need current_token since this user really can't + # have any state deltas before now (as it is a new user), but still, + # we include it for completeness. + current_token = self._get_max_stream_id_in_current_state_deltas_txn(txn) + self._update_stats_delta_txn( + txn, now, "user", user_id, {}, complete_with_stream_id=current_token + ) + self._invalidate_cache_and_stream(txn, self.get_user_by_id, (user_id,)) txn.call_after(self.is_guest.invalidate, (user_id,)) @@ -1140,6 +1151,7 @@ class RegistrationStore( deferred str|None: A str representing a link to redirect the user to if there is one. """ + # Insert everything into a transaction in order to run atomically def validate_threepid_session_txn(txn): row = self._simple_select_one_txn( diff --git a/synapse/storage/roommember.py b/synapse/storage/roommember.py index eecb276465..f8b682ebd9 100644 --- a/synapse/storage/roommember.py +++ b/synapse/storage/roommember.py @@ -112,29 +112,31 @@ class RoomMemberWorkerStore(EventsWorkerStore): @cached(max_entries=100000, iterable=True) def get_users_in_room(self, room_id): - def f(txn): - # If we can assume current_state_events.membership is up to date - # then we can avoid a join, which is a Very Good Thing given how - # frequently this function gets called. - if self._current_state_events_membership_up_to_date: - sql = """ - SELECT state_key FROM current_state_events - WHERE type = 'm.room.member' AND room_id = ? AND membership = ? - """ - else: - sql = """ - SELECT state_key FROM room_memberships as m - INNER JOIN current_state_events as c - ON m.event_id = c.event_id - AND m.room_id = c.room_id - AND m.user_id = c.state_key - WHERE c.type = 'm.room.member' AND c.room_id = ? AND m.membership = ? - """ + return self.runInteraction( + "get_users_in_room", self.get_users_in_room_txn, room_id + ) - txn.execute(sql, (room_id, Membership.JOIN)) - return [to_ascii(r[0]) for r in txn] + def get_users_in_room_txn(self, txn, room_id): + # If we can assume current_state_events.membership is up to date + # then we can avoid a join, which is a Very Good Thing given how + # frequently this function gets called. + if self._current_state_events_membership_up_to_date: + sql = """ + SELECT state_key FROM current_state_events + WHERE type = 'm.room.member' AND room_id = ? AND membership = ? + """ + else: + sql = """ + SELECT state_key FROM room_memberships as m + INNER JOIN current_state_events as c + ON m.event_id = c.event_id + AND m.room_id = c.room_id + AND m.user_id = c.state_key + WHERE c.type = 'm.room.member' AND c.room_id = ? AND m.membership = ? + """ - return self.runInteraction("get_users_in_room", f) + txn.execute(sql, (room_id, Membership.JOIN)) + return [to_ascii(r[0]) for r in txn] @cached(max_entries=100000) def get_room_summary(self, room_id): diff --git a/synapse/storage/schema/delta/56/stats_separated.sql b/synapse/storage/schema/delta/56/stats_separated.sql new file mode 100644 index 0000000000..163529c071 --- /dev/null +++ b/synapse/storage/schema/delta/56/stats_separated.sql @@ -0,0 +1,152 @@ +/* Copyright 2018 New Vector Ltd + * Copyright 2019 The Matrix.org Foundation C.I.C. + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +----- First clean up from previous versions of room stats. + +-- First remove old stats stuff +DROP TABLE IF EXISTS room_stats; +DROP TABLE IF EXISTS room_state; +DROP TABLE IF EXISTS room_stats_state; +DROP TABLE IF EXISTS user_stats; +DROP TABLE IF EXISTS room_stats_earliest_tokens; +DROP TABLE IF EXISTS _temp_populate_stats_position; +DROP TABLE IF EXISTS _temp_populate_stats_rooms; +DROP TABLE IF EXISTS stats_stream_pos; + +-- Unschedule old background updates if they're still scheduled +DELETE FROM background_updates WHERE update_name IN ( + 'populate_stats_createtables', + 'populate_stats_process_rooms', + 'populate_stats_process_users', + 'populate_stats_cleanup' +); + +INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES + ('populate_stats_process_rooms', '{}', ''); + +INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES + ('populate_stats_process_users', '{}', 'populate_stats_process_rooms'); + +----- Create tables for our version of room stats. + +-- single-row table to track position of incremental updates +DROP TABLE IF EXISTS stats_incremental_position; +CREATE TABLE stats_incremental_position ( + Lock CHAR(1) NOT NULL DEFAULT 'X' UNIQUE, -- Makes sure this table only has one row. + stream_id BIGINT NOT NULL, + CHECK (Lock='X') +); + +-- insert a null row and make sure it is the only one. +INSERT INTO stats_incremental_position ( + stream_id +) SELECT COALESCE(MAX(stream_ordering), 0) from events; + +-- represents PRESENT room statistics for a room +-- only holds absolute fields +DROP TABLE IF EXISTS room_stats_current; +CREATE TABLE room_stats_current ( + room_id TEXT NOT NULL PRIMARY KEY, + + -- These are absolute counts + current_state_events INT NOT NULL, + joined_members INT NOT NULL, + invited_members INT NOT NULL, + left_members INT NOT NULL, + banned_members INT NOT NULL, + + local_users_in_room INT NOT NULL, + + -- The maximum delta stream position that this row takes into account. + completed_delta_stream_id BIGINT NOT NULL +); + + +-- represents HISTORICAL room statistics for a room +DROP TABLE IF EXISTS room_stats_historical; +CREATE TABLE room_stats_historical ( + room_id TEXT NOT NULL, + -- These stats cover the time from (end_ts - bucket_size)...end_ts (in ms). + -- Note that end_ts is quantised. + end_ts BIGINT NOT NULL, + bucket_size BIGINT NOT NULL, + + -- These stats are absolute counts + current_state_events BIGINT NOT NULL, + joined_members BIGINT NOT NULL, + invited_members BIGINT NOT NULL, + left_members BIGINT NOT NULL, + banned_members BIGINT NOT NULL, + local_users_in_room BIGINT NOT NULL, + + -- These stats are per time slice + total_events BIGINT NOT NULL, + total_event_bytes BIGINT NOT NULL, + + PRIMARY KEY (room_id, end_ts) +); + +-- We use this index to speed up deletion of ancient room stats. +CREATE INDEX room_stats_historical_end_ts ON room_stats_historical (end_ts); + +-- represents PRESENT statistics for a user +-- only holds absolute fields +DROP TABLE IF EXISTS user_stats_current; +CREATE TABLE user_stats_current ( + user_id TEXT NOT NULL PRIMARY KEY, + + joined_rooms BIGINT NOT NULL, + + -- The maximum delta stream position that this row takes into account. + completed_delta_stream_id BIGINT NOT NULL +); + +-- represents HISTORICAL statistics for a user +DROP TABLE IF EXISTS user_stats_historical; +CREATE TABLE user_stats_historical ( + user_id TEXT NOT NULL, + end_ts BIGINT NOT NULL, + bucket_size BIGINT NOT NULL, + + joined_rooms BIGINT NOT NULL, + + invites_sent BIGINT NOT NULL, + rooms_created BIGINT NOT NULL, + total_events BIGINT NOT NULL, + total_event_bytes BIGINT NOT NULL, + + PRIMARY KEY (user_id, end_ts) +); + +-- We use this index to speed up deletion of ancient user stats. +CREATE INDEX user_stats_historical_end_ts ON user_stats_historical (end_ts); + + +CREATE TABLE room_stats_state ( + room_id TEXT NOT NULL, + name TEXT, + canonical_alias TEXT, + join_rules TEXT, + history_visibility TEXT, + encryption TEXT, + avatar TEXT, + guest_access TEXT, + is_federatable BOOLEAN, + topic TEXT +); + +CREATE UNIQUE INDEX room_stats_state_room ON room_stats_state(room_id); diff --git a/synapse/storage/stats.py b/synapse/storage/stats.py index e13efed417..6560173c08 100644 --- a/synapse/storage/stats.py +++ b/synapse/storage/stats.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2018, 2019 New Vector Ltd +# Copyright 2019 The Matrix.org Foundation C.I.C. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -14,17 +15,22 @@ # limitations under the License. import logging +from itertools import chain from twisted.internet import defer +from twisted.internet.defer import DeferredLock from synapse.api.constants import EventTypes, Membership -from synapse.storage.prepare_database import get_statements +from synapse.storage import PostgresEngine from synapse.storage.state_deltas import StateDeltasStore from synapse.util.caches.descriptors import cached logger = logging.getLogger(__name__) # these fields track absolutes (e.g. total number of rooms on the server) +# You can think of these as Prometheus Gauges. +# You can draw these stats on a line graph. +# Example: number of users in a room ABSOLUTE_STATS_FIELDS = { "room": ( "current_state_events", @@ -32,14 +38,23 @@ ABSOLUTE_STATS_FIELDS = { "invited_members", "left_members", "banned_members", - "state_events", + "local_users_in_room", ), - "user": ("public_rooms", "private_rooms"), + "user": ("joined_rooms",), } -TYPE_TO_ROOM = {"room": ("room_stats", "room_id"), "user": ("user_stats", "user_id")} +# these fields are per-timeslice and so should be reset to 0 upon a new slice +# You can draw these stats on a histogram. +# Example: number of events sent locally during a time slice +PER_SLICE_FIELDS = { + "room": ("total_events", "total_event_bytes"), + "user": ("invites_sent", "rooms_created", "total_events", "total_event_bytes"), +} + +TYPE_TO_TABLE = {"room": ("room_stats", "room_id"), "user": ("user_stats", "user_id")} -TEMP_TABLE = "_temp_populate_stats" +# these are the tables (& ID columns) which contain our actual subjects +TYPE_TO_ORIGIN_TABLE = {"room": ("rooms", "room_id"), "user": ("users", "name")} class StatsStore(StateDeltasStore): @@ -51,136 +66,102 @@ class StatsStore(StateDeltasStore): self.stats_enabled = hs.config.stats_enabled self.stats_bucket_size = hs.config.stats_bucket_size - self.register_background_update_handler( - "populate_stats_createtables", self._populate_stats_createtables - ) + self.stats_delta_processing_lock = DeferredLock() + self.register_background_update_handler( "populate_stats_process_rooms", self._populate_stats_process_rooms ) self.register_background_update_handler( - "populate_stats_cleanup", self._populate_stats_cleanup + "populate_stats_process_users", self._populate_stats_process_users ) + # we no longer need to perform clean-up, but we will give ourselves + # the potential to reintroduce it in the future – so documentation + # will still encourage the use of this no-op handler. + self.register_noop_background_update("populate_stats_cleanup") + self.register_noop_background_update("populate_stats_prepare") - @defer.inlineCallbacks - def _populate_stats_createtables(self, progress, batch_size): - - if not self.stats_enabled: - yield self._end_background_update("populate_stats_createtables") - return 1 - - # Get all the rooms that we want to process. - def _make_staging_area(txn): - # Create the temporary tables - stmts = get_statements( - """ - -- We just recreate the table, we'll be reinserting the - -- correct entries again later anyway. - DROP TABLE IF EXISTS {temp}_rooms; - - CREATE TABLE IF NOT EXISTS {temp}_rooms( - room_id TEXT NOT NULL, - events BIGINT NOT NULL - ); - - CREATE INDEX {temp}_rooms_events - ON {temp}_rooms(events); - CREATE INDEX {temp}_rooms_id - ON {temp}_rooms(room_id); - """.format( - temp=TEMP_TABLE - ).splitlines() - ) - - for statement in stmts: - txn.execute(statement) - - sql = ( - "CREATE TABLE IF NOT EXISTS " - + TEMP_TABLE - + "_position(position TEXT NOT NULL)" - ) - txn.execute(sql) - - # Get rooms we want to process from the database, only adding - # those that we haven't (i.e. those not in room_stats_earliest_token) - sql = """ - INSERT INTO %s_rooms (room_id, events) - SELECT c.room_id, count(*) FROM current_state_events AS c - LEFT JOIN room_stats_earliest_token AS t USING (room_id) - WHERE t.room_id IS NULL - GROUP BY c.room_id - """ % ( - TEMP_TABLE, - ) - txn.execute(sql) + def quantise_stats_time(self, ts): + """ + Quantises a timestamp to be a multiple of the bucket size. - new_pos = yield self.get_max_stream_id_in_current_state_deltas() - yield self.runInteraction("populate_stats_temp_build", _make_staging_area) - yield self._simple_insert(TEMP_TABLE + "_position", {"position": new_pos}) - self.get_earliest_token_for_room_stats.invalidate_all() + Args: + ts (int): the timestamp to quantise, in milliseconds since the Unix + Epoch - yield self._end_background_update("populate_stats_createtables") - return 1 + Returns: + int: a timestamp which + - is divisible by the bucket size; + - is no later than `ts`; and + - is the largest such timestamp. + """ + return (ts // self.stats_bucket_size) * self.stats_bucket_size @defer.inlineCallbacks - def _populate_stats_cleanup(self, progress, batch_size): + def _populate_stats_process_users(self, progress, batch_size): """ - Update the user directory stream position, then clean up the old tables. + This is a background update which regenerates statistics for users. """ if not self.stats_enabled: - yield self._end_background_update("populate_stats_cleanup") + yield self._end_background_update("populate_stats_process_users") return 1 - position = yield self._simple_select_one_onecol( - TEMP_TABLE + "_position", None, "position" + last_user_id = progress.get("last_user_id", "") + + def _get_next_batch(txn): + sql = """ + SELECT DISTINCT name FROM users + WHERE name > ? + ORDER BY name ASC + LIMIT ? + """ + txn.execute(sql, (last_user_id, batch_size)) + return [r for r, in txn] + + users_to_work_on = yield self.runInteraction( + "_populate_stats_process_users", _get_next_batch ) - yield self.update_stats_stream_pos(position) - def _delete_staging_area(txn): - txn.execute("DROP TABLE IF EXISTS " + TEMP_TABLE + "_rooms") - txn.execute("DROP TABLE IF EXISTS " + TEMP_TABLE + "_position") + # No more rooms -- complete the transaction. + if not users_to_work_on: + yield self._end_background_update("populate_stats_process_users") + return 1 - yield self.runInteraction("populate_stats_cleanup", _delete_staging_area) + for user_id in users_to_work_on: + yield self._calculate_and_set_initial_state_for_user(user_id) + progress["last_user_id"] = user_id - yield self._end_background_update("populate_stats_cleanup") - return 1 + yield self.runInteraction( + "populate_stats_process_users", + self._background_update_progress_txn, + "populate_stats_process_users", + progress, + ) + + return len(users_to_work_on) @defer.inlineCallbacks def _populate_stats_process_rooms(self, progress, batch_size): - + """ + This is a background update which regenerates statistics for rooms. + """ if not self.stats_enabled: yield self._end_background_update("populate_stats_process_rooms") return 1 - # If we don't have progress filed, delete everything. - if not progress: - yield self.delete_all_stats() + last_room_id = progress.get("last_room_id", "") def _get_next_batch(txn): - # Only fetch 250 rooms, so we don't fetch too many at once, even - # if those 250 rooms have less than batch_size state events. sql = """ - SELECT room_id, events FROM %s_rooms - ORDER BY events DESC - LIMIT 250 - """ % ( - TEMP_TABLE, - ) - txn.execute(sql) - rooms_to_work_on = txn.fetchall() - - if not rooms_to_work_on: - return None - - # Get how many are left to process, so we can give status on how - # far we are in processing - txn.execute("SELECT COUNT(*) FROM " + TEMP_TABLE + "_rooms") - progress["remaining"] = txn.fetchone()[0] - - return rooms_to_work_on + SELECT DISTINCT room_id FROM current_state_events + WHERE room_id > ? + ORDER BY room_id ASC + LIMIT ? + """ + txn.execute(sql, (last_room_id, batch_size)) + return [r for r, in txn] rooms_to_work_on = yield self.runInteraction( - "populate_stats_temp_read", _get_next_batch + "populate_stats_rooms_get_batch", _get_next_batch ) # No more rooms -- complete the transaction. @@ -188,154 +169,28 @@ class StatsStore(StateDeltasStore): yield self._end_background_update("populate_stats_process_rooms") return 1 - logger.info( - "Processing the next %d rooms of %d remaining", - len(rooms_to_work_on), - progress["remaining"], - ) - - # Number of state events we've processed by going through each room - processed_event_count = 0 - - for room_id, event_count in rooms_to_work_on: - - current_state_ids = yield self.get_current_state_ids(room_id) - - join_rules_id = current_state_ids.get((EventTypes.JoinRules, "")) - history_visibility_id = current_state_ids.get( - (EventTypes.RoomHistoryVisibility, "") - ) - encryption_id = current_state_ids.get((EventTypes.RoomEncryption, "")) - name_id = current_state_ids.get((EventTypes.Name, "")) - topic_id = current_state_ids.get((EventTypes.Topic, "")) - avatar_id = current_state_ids.get((EventTypes.RoomAvatar, "")) - canonical_alias_id = current_state_ids.get((EventTypes.CanonicalAlias, "")) - - event_ids = [ - join_rules_id, - history_visibility_id, - encryption_id, - name_id, - topic_id, - avatar_id, - canonical_alias_id, - ] - - state_events = yield self.get_events( - [ev for ev in event_ids if ev is not None] - ) - - def _get_or_none(event_id, arg): - event = state_events.get(event_id) - if event: - return event.content.get(arg) - return None - - yield self.update_room_state( - room_id, - { - "join_rules": _get_or_none(join_rules_id, "join_rule"), - "history_visibility": _get_or_none( - history_visibility_id, "history_visibility" - ), - "encryption": _get_or_none(encryption_id, "algorithm"), - "name": _get_or_none(name_id, "name"), - "topic": _get_or_none(topic_id, "topic"), - "avatar": _get_or_none(avatar_id, "url"), - "canonical_alias": _get_or_none(canonical_alias_id, "alias"), - }, - ) + for room_id in rooms_to_work_on: + yield self._calculate_and_set_initial_state_for_room(room_id) + progress["last_room_id"] = room_id - now = self.hs.get_reactor().seconds() - - # quantise time to the nearest bucket - now = (now // self.stats_bucket_size) * self.stats_bucket_size - - def _fetch_data(txn): - - # Get the current token of the room - current_token = self._get_max_stream_id_in_current_state_deltas_txn(txn) - - current_state_events = len(current_state_ids) - - membership_counts = self._get_user_counts_in_room_txn(txn, room_id) - - total_state_events = self._get_total_state_event_counts_txn( - txn, room_id - ) - - self._update_stats_txn( - txn, - "room", - room_id, - now, - { - "bucket_size": self.stats_bucket_size, - "current_state_events": current_state_events, - "joined_members": membership_counts.get(Membership.JOIN, 0), - "invited_members": membership_counts.get(Membership.INVITE, 0), - "left_members": membership_counts.get(Membership.LEAVE, 0), - "banned_members": membership_counts.get(Membership.BAN, 0), - "state_events": total_state_events, - }, - ) - self._simple_insert_txn( - txn, - "room_stats_earliest_token", - {"room_id": room_id, "token": current_token}, - ) - - # We've finished a room. Delete it from the table. - self._simple_delete_one_txn( - txn, TEMP_TABLE + "_rooms", {"room_id": room_id} - ) - - yield self.runInteraction("update_room_stats", _fetch_data) - - # Update the remaining counter. - progress["remaining"] -= 1 - yield self.runInteraction( - "populate_stats", - self._background_update_progress_txn, - "populate_stats_process_rooms", - progress, - ) - - processed_event_count += event_count - - if processed_event_count > batch_size: - # Don't process any more rooms, we've hit our batch size. - return processed_event_count + yield self.runInteraction( + "_populate_stats_process_rooms", + self._background_update_progress_txn, + "populate_stats_process_rooms", + progress, + ) - return processed_event_count + return len(rooms_to_work_on) - def delete_all_stats(self): + def get_stats_positions(self): """ - Delete all statistics records. + Returns the stats processor positions. """ - - def _delete_all_stats_txn(txn): - txn.execute("DELETE FROM room_state") - txn.execute("DELETE FROM room_stats") - txn.execute("DELETE FROM room_stats_earliest_token") - txn.execute("DELETE FROM user_stats") - - return self.runInteraction("delete_all_stats", _delete_all_stats_txn) - - def get_stats_stream_pos(self): return self._simple_select_one_onecol( - table="stats_stream_pos", + table="stats_incremental_position", keyvalues={}, retcol="stream_id", - desc="stats_stream_pos", - ) - - def update_stats_stream_pos(self, stream_id): - return self._simple_update_one( - table="stats_stream_pos", - keyvalues={}, - updatevalues={"stream_id": stream_id}, - desc="update_stats_stream_pos", + desc="stats_incremental_position", ) def update_room_state(self, room_id, fields): @@ -361,42 +216,87 @@ class StatsStore(StateDeltasStore): fields[col] = None return self._simple_upsert( - table="room_state", + table="room_stats_state", keyvalues={"room_id": room_id}, values=fields, desc="update_room_state", ) - def get_deltas_for_room(self, room_id, start, size=100): + def get_statistics_for_subject(self, stats_type, stats_id, start, size=100): """ - Get statistics deltas for a given room. + Get statistics for a given subject. Args: - room_id (str) + stats_type (str): The type of subject + stats_id (str): The ID of the subject (e.g. room_id or user_id) start (int): Pagination start. Number of entries, not timestamp. size (int): How many entries to return. Returns: Deferred[list[dict]], where the dict has the keys of - ABSOLUTE_STATS_FIELDS["room"] and "ts". + ABSOLUTE_STATS_FIELDS[stats_type], and "bucket_size" and "end_ts". """ - return self._simple_select_list_paginate( - "room_stats", - {"room_id": room_id}, - "ts", + return self.runInteraction( + "get_statistics_for_subject", + self._get_statistics_for_subject_txn, + stats_type, + stats_id, + start, + size, + ) + + def _get_statistics_for_subject_txn( + self, txn, stats_type, stats_id, start, size=100 + ): + """ + Transaction-bound version of L{get_statistics_for_subject}. + """ + + table, id_col = TYPE_TO_TABLE[stats_type] + selected_columns = list( + ABSOLUTE_STATS_FIELDS[stats_type] + PER_SLICE_FIELDS[stats_type] + ) + + slice_list = self._simple_select_list_paginate_txn( + txn, + table + "_historical", + {id_col: stats_id}, + "end_ts", start, size, - retcols=(list(ABSOLUTE_STATS_FIELDS["room"]) + ["ts"]), + retcols=selected_columns + ["bucket_size", "end_ts"], order_direction="DESC", ) - def get_all_room_state(self): - return self._simple_select_list( - "room_state", None, retcols=("name", "topic", "canonical_alias") + return slice_list + + def get_room_stats_state(self, room_id): + """ + Returns the current room_stats_state for a room. + + Args: + room_id (str): The ID of the room to return state for. + + Returns (dict): + Dictionary containing these keys: + "name", "topic", "canonical_alias", "avatar", "join_rules", + "history_visibility" + """ + return self._simple_select_one( + "room_stats_state", + {"room_id": room_id}, + retcols=( + "name", + "topic", + "canonical_alias", + "avatar", + "join_rules", + "history_visibility", + ), ) @cached() - def get_earliest_token_for_room_stats(self, room_id): + def get_earliest_token_for_stats(self, stats_type, id): """ Fetch the "earliest token". This is used by the room stats delta processor to ignore deltas that have been processed between the @@ -406,79 +306,571 @@ class StatsStore(StateDeltasStore): Returns: Deferred[int] """ + table, id_col = TYPE_TO_TABLE[stats_type] + return self._simple_select_one_onecol( - "room_stats_earliest_token", - {"room_id": room_id}, - retcol="token", + "%s_current" % (table,), + keyvalues={id_col: id}, + retcol="completed_delta_stream_id", allow_none=True, ) - def update_stats(self, stats_type, stats_id, ts, fields): - table, id_col = TYPE_TO_ROOM[stats_type] - return self._simple_upsert( - table=table, - keyvalues={id_col: stats_id, "ts": ts}, - values=fields, - desc="update_stats", + def bulk_update_stats_delta(self, ts, updates, stream_id): + """Bulk update stats tables for a given stream_id and updates the stats + incremental position. + + Args: + ts (int): Current timestamp in ms + updates(dict[str, dict[str, dict[str, Counter]]]): The updates to + commit as a mapping stats_type -> stats_id -> field -> delta. + stream_id (int): Current position. + + Returns: + Deferred + """ + + def _bulk_update_stats_delta_txn(txn): + for stats_type, stats_updates in updates.items(): + for stats_id, fields in stats_updates.items(): + self._update_stats_delta_txn( + txn, + ts=ts, + stats_type=stats_type, + stats_id=stats_id, + fields=fields, + complete_with_stream_id=stream_id, + ) + + self._simple_update_one_txn( + txn, + table="stats_incremental_position", + keyvalues={}, + updatevalues={"stream_id": stream_id}, + ) + + return self.runInteraction( + "bulk_update_stats_delta", _bulk_update_stats_delta_txn ) - def _update_stats_txn(self, txn, stats_type, stats_id, ts, fields): - table, id_col = TYPE_TO_ROOM[stats_type] - return self._simple_upsert_txn( - txn, table=table, keyvalues={id_col: stats_id, "ts": ts}, values=fields + def update_stats_delta( + self, + ts, + stats_type, + stats_id, + fields, + complete_with_stream_id, + absolute_field_overrides=None, + ): + """ + Updates the statistics for a subject, with a delta (difference/relative + change). + + Args: + ts (int): timestamp of the change + stats_type (str): "room" or "user" – the kind of subject + stats_id (str): the subject's ID (room ID or user ID) + fields (dict[str, int]): Deltas of stats values. + complete_with_stream_id (int, optional): + If supplied, converts an incomplete row into a complete row, + with the supplied stream_id marked as the stream_id where the + row was completed. + absolute_field_overrides (dict[str, int]): Current stats values + (i.e. not deltas) of absolute fields. + Does not work with per-slice fields. + """ + + return self.runInteraction( + "update_stats_delta", + self._update_stats_delta_txn, + ts, + stats_type, + stats_id, + fields, + complete_with_stream_id=complete_with_stream_id, + absolute_field_overrides=absolute_field_overrides, ) - def update_stats_delta(self, ts, stats_type, stats_id, field, value): - def _update_stats_delta(txn): - table, id_col = TYPE_TO_ROOM[stats_type] - - sql = ( - "SELECT * FROM %s" - " WHERE %s=? and ts=(" - " SELECT MAX(ts) FROM %s" - " WHERE %s=?" - ")" - ) % (table, id_col, table, id_col) - txn.execute(sql, (stats_id, stats_id)) - rows = self.cursor_to_dict(txn) - if len(rows) == 0: - # silently skip as we don't have anything to apply a delta to yet. - # this tries to minimise any race between the initial sync and - # subsequent deltas arriving. - return - - current_ts = ts - latest_ts = rows[0]["ts"] - if current_ts < latest_ts: - # This one is in the past, but we're just encountering it now. - # Mark it as part of the current bucket. - current_ts = latest_ts - elif ts != latest_ts: - # we have to copy our absolute counters over to the new entry. - values = { - key: rows[0][key] for key in ABSOLUTE_STATS_FIELDS[stats_type] - } - values[id_col] = stats_id - values["ts"] = ts - values["bucket_size"] = self.stats_bucket_size - - self._simple_insert_txn(txn, table=table, values=values) - - # actually update the new value - if stats_type in ABSOLUTE_STATS_FIELDS[stats_type]: - self._simple_update_txn( - txn, - table=table, - keyvalues={id_col: stats_id, "ts": current_ts}, - updatevalues={field: value}, + def _update_stats_delta_txn( + self, + txn, + ts, + stats_type, + stats_id, + fields, + complete_with_stream_id, + absolute_field_overrides=None, + ): + if absolute_field_overrides is None: + absolute_field_overrides = {} + + table, id_col = TYPE_TO_TABLE[stats_type] + + quantised_ts = self.quantise_stats_time(int(ts)) + end_ts = quantised_ts + self.stats_bucket_size + + # Lets be paranoid and check that all the given field names are known + abs_field_names = ABSOLUTE_STATS_FIELDS[stats_type] + slice_field_names = PER_SLICE_FIELDS[stats_type] + for field in chain(fields.keys(), absolute_field_overrides.keys()): + if field not in abs_field_names and field not in slice_field_names: + # guard against potential SQL injection dodginess + raise ValueError( + "%s is not a recognised field" + " for stats type %s" % (field, stats_type) ) + + # Per slice fields do not get added to the _current table + + # This calculates the deltas (`field = field + ?` values) + # for absolute fields, + # * defaulting to 0 if not specified + # (required for the INSERT part of upserting to work) + # * omitting overrides specified in `absolute_field_overrides` + deltas_of_absolute_fields = { + key: fields.get(key, 0) + for key in abs_field_names + if key not in absolute_field_overrides + } + + # Keep the delta stream ID field up to date + absolute_field_overrides = absolute_field_overrides.copy() + absolute_field_overrides["completed_delta_stream_id"] = complete_with_stream_id + + # first upsert the `_current` table + self._upsert_with_additive_relatives_txn( + txn=txn, + table=table + "_current", + keyvalues={id_col: stats_id}, + absolutes=absolute_field_overrides, + additive_relatives=deltas_of_absolute_fields, + ) + + per_slice_additive_relatives = { + key: fields.get(key, 0) for key in slice_field_names + } + self._upsert_copy_from_table_with_additive_relatives_txn( + txn=txn, + into_table=table + "_historical", + keyvalues={id_col: stats_id}, + extra_dst_insvalues={"bucket_size": self.stats_bucket_size}, + extra_dst_keyvalues={"end_ts": end_ts}, + additive_relatives=per_slice_additive_relatives, + src_table=table + "_current", + copy_columns=abs_field_names, + ) + + def _upsert_with_additive_relatives_txn( + self, txn, table, keyvalues, absolutes, additive_relatives + ): + """Used to update values in the stats tables. + + This is basically a slightly convoluted upsert that *adds* to any + existing rows. + + Args: + txn + table (str): Table name + keyvalues (dict[str, any]): Row-identifying key values + absolutes (dict[str, any]): Absolute (set) fields + additive_relatives (dict[str, int]): Fields that will be added onto + if existing row present. + """ + if self.database_engine.can_native_upsert: + absolute_updates = [ + "%(field)s = EXCLUDED.%(field)s" % {"field": field} + for field in absolutes.keys() + ] + + relative_updates = [ + "%(field)s = EXCLUDED.%(field)s + %(table)s.%(field)s" + % {"table": table, "field": field} + for field in additive_relatives.keys() + ] + + insert_cols = [] + qargs = [] + + for (key, val) in chain( + keyvalues.items(), absolutes.items(), additive_relatives.items() + ): + insert_cols.append(key) + qargs.append(val) + + sql = """ + INSERT INTO %(table)s (%(insert_cols_cs)s) + VALUES (%(insert_vals_qs)s) + ON CONFLICT (%(key_columns)s) DO UPDATE SET %(updates)s + """ % { + "table": table, + "insert_cols_cs": ", ".join(insert_cols), + "insert_vals_qs": ", ".join( + ["?"] * (len(keyvalues) + len(absolutes) + len(additive_relatives)) + ), + "key_columns": ", ".join(keyvalues), + "updates": ", ".join(chain(absolute_updates, relative_updates)), + } + + txn.execute(sql, qargs) + else: + self.database_engine.lock_table(txn, table) + retcols = list(chain(absolutes.keys(), additive_relatives.keys())) + current_row = self._simple_select_one_txn( + txn, table, keyvalues, retcols, allow_none=True + ) + if current_row is None: + merged_dict = {**keyvalues, **absolutes, **additive_relatives} + self._simple_insert_txn(txn, table, merged_dict) + else: + for (key, val) in additive_relatives.items(): + current_row[key] += val + current_row.update(absolutes) + self._simple_update_one_txn(txn, table, keyvalues, current_row) + + def _upsert_copy_from_table_with_additive_relatives_txn( + self, + txn, + into_table, + keyvalues, + extra_dst_keyvalues, + extra_dst_insvalues, + additive_relatives, + src_table, + copy_columns, + ): + """Updates the historic stats table with latest updates. + + This involves copying "absolute" fields from the `_current` table, and + adding relative fields to any existing values. + + Args: + txn: Transaction + into_table (str): The destination table to UPSERT the row into + keyvalues (dict[str, any]): Row-identifying key values + extra_dst_keyvalues (dict[str, any]): Additional keyvalues + for `into_table`. + extra_dst_insvalues (dict[str, any]): Additional values to insert + on new row creation for `into_table`. + additive_relatives (dict[str, any]): Fields that will be added onto + if existing row present. (Must be disjoint from copy_columns.) + src_table (str): The source table to copy from + copy_columns (iterable[str]): The list of columns to copy + """ + if self.database_engine.can_native_upsert: + ins_columns = chain( + keyvalues, + copy_columns, + additive_relatives, + extra_dst_keyvalues, + extra_dst_insvalues, + ) + sel_exprs = chain( + keyvalues, + copy_columns, + ( + "?" + for _ in chain( + additive_relatives, extra_dst_keyvalues, extra_dst_insvalues + ) + ), + ) + keyvalues_where = ("%s = ?" % f for f in keyvalues) + + sets_cc = ("%s = EXCLUDED.%s" % (f, f) for f in copy_columns) + sets_ar = ( + "%s = EXCLUDED.%s + %s.%s" % (f, f, into_table, f) + for f in additive_relatives + ) + + sql = """ + INSERT INTO %(into_table)s (%(ins_columns)s) + SELECT %(sel_exprs)s + FROM %(src_table)s + WHERE %(keyvalues_where)s + ON CONFLICT (%(keyvalues)s) + DO UPDATE SET %(sets)s + """ % { + "into_table": into_table, + "ins_columns": ", ".join(ins_columns), + "sel_exprs": ", ".join(sel_exprs), + "keyvalues_where": " AND ".join(keyvalues_where), + "src_table": src_table, + "keyvalues": ", ".join( + chain(keyvalues.keys(), extra_dst_keyvalues.keys()) + ), + "sets": ", ".join(chain(sets_cc, sets_ar)), + } + + qargs = list( + chain( + additive_relatives.values(), + extra_dst_keyvalues.values(), + extra_dst_insvalues.values(), + keyvalues.values(), + ) + ) + txn.execute(sql, qargs) + else: + self.database_engine.lock_table(txn, into_table) + src_row = self._simple_select_one_txn( + txn, src_table, keyvalues, copy_columns + ) + all_dest_keyvalues = {**keyvalues, **extra_dst_keyvalues} + dest_current_row = self._simple_select_one_txn( + txn, + into_table, + keyvalues=all_dest_keyvalues, + retcols=list(chain(additive_relatives.keys(), copy_columns)), + allow_none=True, + ) + + if dest_current_row is None: + merged_dict = { + **keyvalues, + **extra_dst_keyvalues, + **extra_dst_insvalues, + **src_row, + **additive_relatives, + } + self._simple_insert_txn(txn, into_table, merged_dict) else: - sql = ("UPDATE %s SET %s=%s+? WHERE %s=? AND ts=?") % ( - table, - field, - field, - id_col, + for (key, val) in additive_relatives.items(): + src_row[key] = dest_current_row[key] + val + self._simple_update_txn(txn, into_table, all_dest_keyvalues, src_row) + + def get_changes_room_total_events_and_bytes(self, min_pos, max_pos): + """Fetches the counts of events in the given range of stream IDs. + + Args: + min_pos (int) + max_pos (int) + + Returns: + Deferred[dict[str, dict[str, int]]]: Mapping of room ID to field + changes. + """ + + return self.runInteraction( + "stats_incremental_total_events_and_bytes", + self.get_changes_room_total_events_and_bytes_txn, + min_pos, + max_pos, + ) + + def get_changes_room_total_events_and_bytes_txn(self, txn, low_pos, high_pos): + """Gets the total_events and total_event_bytes counts for rooms and + senders, in a range of stream_orderings (including backfilled events). + + Args: + txn + low_pos (int): Low stream ordering + high_pos (int): High stream ordering + + Returns: + tuple[dict[str, dict[str, int]], dict[str, dict[str, int]]]: The + room and user deltas for total_events/total_event_bytes in the + format of `stats_id` -> fields + """ + + if low_pos >= high_pos: + # nothing to do here. + return {}, {} + + if isinstance(self.database_engine, PostgresEngine): + new_bytes_expression = "OCTET_LENGTH(json)" + else: + new_bytes_expression = "LENGTH(CAST(json AS BLOB))" + + sql = """ + SELECT events.room_id, COUNT(*) AS new_events, SUM(%s) AS new_bytes + FROM events INNER JOIN event_json USING (event_id) + WHERE (? < stream_ordering AND stream_ordering <= ?) + OR (? <= stream_ordering AND stream_ordering <= ?) + GROUP BY events.room_id + """ % ( + new_bytes_expression, + ) + + txn.execute(sql, (low_pos, high_pos, -high_pos, -low_pos)) + + room_deltas = { + room_id: {"total_events": new_events, "total_event_bytes": new_bytes} + for room_id, new_events, new_bytes in txn + } + + sql = """ + SELECT events.sender, COUNT(*) AS new_events, SUM(%s) AS new_bytes + FROM events INNER JOIN event_json USING (event_id) + WHERE (? < stream_ordering AND stream_ordering <= ?) + OR (? <= stream_ordering AND stream_ordering <= ?) + GROUP BY events.sender + """ % ( + new_bytes_expression, + ) + + txn.execute(sql, (low_pos, high_pos, -high_pos, -low_pos)) + + user_deltas = { + user_id: {"total_events": new_events, "total_event_bytes": new_bytes} + for user_id, new_events, new_bytes in txn + if self.hs.is_mine_id(user_id) + } + + return room_deltas, user_deltas + + @defer.inlineCallbacks + def _calculate_and_set_initial_state_for_room(self, room_id): + """Calculate and insert an entry into room_stats_current. + + Args: + room_id (str) + + Returns: + Deferred[tuple[dict, dict, int]]: A tuple of room state, membership + counts and stream position. + """ + + def _fetch_current_state_stats(txn): + pos = self.get_room_max_stream_ordering() + + rows = self._simple_select_many_txn( + txn, + table="current_state_events", + column="type", + iterable=[ + EventTypes.Create, + EventTypes.JoinRules, + EventTypes.RoomHistoryVisibility, + EventTypes.Encryption, + EventTypes.Name, + EventTypes.Topic, + EventTypes.RoomAvatar, + EventTypes.CanonicalAlias, + ], + keyvalues={"room_id": room_id, "state_key": ""}, + retcols=["event_id"], + ) + + event_ids = [row["event_id"] for row in rows] + + txn.execute( + """ + SELECT membership, count(*) FROM current_state_events + WHERE room_id = ? AND type = 'm.room.member' + GROUP BY membership + """, + (room_id,), + ) + membership_counts = {membership: cnt for membership, cnt in txn} + + txn.execute( + """ + SELECT COALESCE(count(*), 0) FROM current_state_events + WHERE room_id = ? + """, + (room_id,), + ) + + current_state_events_count, = txn.fetchone() + + users_in_room = self.get_users_in_room_txn(txn, room_id) + + return ( + event_ids, + membership_counts, + current_state_events_count, + users_in_room, + pos, + ) + + ( + event_ids, + membership_counts, + current_state_events_count, + users_in_room, + pos, + ) = yield self.runInteraction( + "get_initial_state_for_room", _fetch_current_state_stats + ) + + state_event_map = yield self.get_events(event_ids, get_prev_content=False) + + room_state = { + "join_rules": None, + "history_visibility": None, + "encryption": None, + "name": None, + "topic": None, + "avatar": None, + "canonical_alias": None, + "is_federatable": True, + } + + for event in state_event_map.values(): + if event.type == EventTypes.JoinRules: + room_state["join_rules"] = event.content.get("join_rule") + elif event.type == EventTypes.RoomHistoryVisibility: + room_state["history_visibility"] = event.content.get( + "history_visibility" ) - txn.execute(sql, (value, stats_id, current_ts)) + elif event.type == EventTypes.Encryption: + room_state["encryption"] = event.content.get("algorithm") + elif event.type == EventTypes.Name: + room_state["name"] = event.content.get("name") + elif event.type == EventTypes.Topic: + room_state["topic"] = event.content.get("topic") + elif event.type == EventTypes.RoomAvatar: + room_state["avatar"] = event.content.get("url") + elif event.type == EventTypes.CanonicalAlias: + room_state["canonical_alias"] = event.content.get("alias") + elif event.type == EventTypes.Create: + room_state["is_federatable"] = event.content.get("m.federate", True) + + yield self.update_room_state(room_id, room_state) + + local_users_in_room = [u for u in users_in_room if self.hs.is_mine_id(u)] + + yield self.update_stats_delta( + ts=self.clock.time_msec(), + stats_type="room", + stats_id=room_id, + fields={}, + complete_with_stream_id=pos, + absolute_field_overrides={ + "current_state_events": current_state_events_count, + "joined_members": membership_counts.get(Membership.JOIN, 0), + "invited_members": membership_counts.get(Membership.INVITE, 0), + "left_members": membership_counts.get(Membership.LEAVE, 0), + "banned_members": membership_counts.get(Membership.BAN, 0), + "local_users_in_room": len(local_users_in_room), + }, + ) + + @defer.inlineCallbacks + def _calculate_and_set_initial_state_for_user(self, user_id): + def _calculate_and_set_initial_state_for_user_txn(txn): + pos = self._get_max_stream_id_in_current_state_deltas_txn(txn) - return self.runInteraction("update_stats_delta", _update_stats_delta) + txn.execute( + """ + SELECT COUNT(distinct room_id) FROM current_state_events + WHERE type = 'm.room.member' AND state_key = ? + AND membership = 'join' + """, + (user_id,), + ) + count, = txn.fetchone() + return count, pos + + joined_rooms, pos = yield self.runInteraction( + "calculate_and_set_initial_state_for_user", + _calculate_and_set_initial_state_for_user_txn, + ) + + yield self.update_stats_delta( + ts=self.clock.time_msec(), + stats_type="user", + stats_id=user_id, + fields={}, + complete_with_stream_id=pos, + absolute_field_overrides={"joined_rooms": joined_rooms}, + ) diff --git a/tests/handlers/test_stats.py b/tests/handlers/test_stats.py index a8b858eb4f..7569b6fab5 100644 --- a/tests/handlers/test_stats.py +++ b/tests/handlers/test_stats.py @@ -13,16 +13,17 @@ # See the License for the specific language governing permissions and # limitations under the License. -from mock import Mock - -from twisted.internet import defer - -from synapse.api.constants import EventTypes, Membership +from synapse import storage from synapse.rest import admin from synapse.rest.client.v1 import login, room from tests import unittest +# The expected number of state events in a fresh public room. +EXPT_NUM_STATE_EVTS_IN_FRESH_PUBLIC_ROOM = 5 +# The expected number of state events in a fresh private room. +EXPT_NUM_STATE_EVTS_IN_FRESH_PRIVATE_ROOM = 6 + class StatsRoomTests(unittest.HomeserverTestCase): @@ -33,7 +34,6 @@ class StatsRoomTests(unittest.HomeserverTestCase): ] def prepare(self, reactor, clock, hs): - self.store = hs.get_datastore() self.handler = self.hs.get_stats_handler() @@ -47,7 +47,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): self.get_success( self.store._simple_insert( "background_updates", - {"update_name": "populate_stats_createtables", "progress_json": "{}"}, + {"update_name": "populate_stats_prepare", "progress_json": "{}"}, ) ) self.get_success( @@ -56,7 +56,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): { "update_name": "populate_stats_process_rooms", "progress_json": "{}", - "depends_on": "populate_stats_createtables", + "depends_on": "populate_stats_prepare", }, ) ) @@ -64,18 +64,58 @@ class StatsRoomTests(unittest.HomeserverTestCase): self.store._simple_insert( "background_updates", { - "update_name": "populate_stats_cleanup", + "update_name": "populate_stats_process_users", "progress_json": "{}", "depends_on": "populate_stats_process_rooms", }, ) ) + self.get_success( + self.store._simple_insert( + "background_updates", + { + "update_name": "populate_stats_cleanup", + "progress_json": "{}", + "depends_on": "populate_stats_process_users", + }, + ) + ) + + def get_all_room_state(self): + return self.store._simple_select_list( + "room_stats_state", None, retcols=("name", "topic", "canonical_alias") + ) + + def _get_current_stats(self, stats_type, stat_id): + table, id_col = storage.stats.TYPE_TO_TABLE[stats_type] + + cols = list(storage.stats.ABSOLUTE_STATS_FIELDS[stats_type]) + list( + storage.stats.PER_SLICE_FIELDS[stats_type] + ) + + end_ts = self.store.quantise_stats_time(self.reactor.seconds() * 1000) + + return self.get_success( + self.store._simple_select_one( + table + "_historical", + {id_col: stat_id, end_ts: end_ts}, + cols, + allow_none=True, + ) + ) + + def _perform_background_initial_update(self): + # Do the initial population of the stats via the background update + self._add_background_updates() + + while not self.get_success(self.store.has_completed_background_updates()): + self.get_success(self.store.do_next_background_update(100), by=0.1) def test_initial_room(self): """ The background updates will build the table from scratch. """ - r = self.get_success(self.store.get_all_room_state()) + r = self.get_success(self.get_all_room_state()) self.assertEqual(len(r), 0) # Disable stats @@ -91,7 +131,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): ) # Stats disabled, shouldn't have done anything - r = self.get_success(self.store.get_all_room_state()) + r = self.get_success(self.get_all_room_state()) self.assertEqual(len(r), 0) # Enable stats @@ -104,7 +144,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): while not self.get_success(self.store.has_completed_background_updates()): self.get_success(self.store.do_next_background_update(100), by=0.1) - r = self.get_success(self.store.get_all_room_state()) + r = self.get_success(self.get_all_room_state()) self.assertEqual(len(r), 1) self.assertEqual(r[0]["topic"], "foo") @@ -114,6 +154,7 @@ class StatsRoomTests(unittest.HomeserverTestCase): Ingestion via notify_new_event will ignore tokens that the background update have already processed. """ + self.reactor.advance(86401) self.hs.config.stats_enabled = False @@ -138,12 +179,18 @@ class StatsRoomTests(unittest.HomeserverTestCase): self.hs.config.stats_enabled = True self.handler.stats_enabled = True self.store._all_done = False - self.get_success(self.store.update_stats_stream_pos(None)) + self.get_success( + self.store._simple_update_one( + table="stats_incremental_position", + keyvalues={}, + updatevalues={"stream_id": 0}, + ) + ) self.get_success( self.store._simple_insert( "background_updates", - {"update_name": "populate_stats_createtables", "progress_json": "{}"}, + {"update_name": "populate_stats_prepare", "progress_json": "{}"}, ) ) @@ -154,6 +201,8 @@ class StatsRoomTests(unittest.HomeserverTestCase): self.helper.invite(room=room_1, src=u1, targ=u2, tok=u1_token) self.helper.join(room=room_1, user=u2, tok=u2_token) + # orig_delta_processor = self.store. + # Now do the initial ingestion. self.get_success( self.store._simple_insert( @@ -185,8 +234,15 @@ class StatsRoomTests(unittest.HomeserverTestCase): self.helper.invite(room=room_1, src=u1, targ=u3, tok=u1_token) self.helper.join(room=room_1, user=u3, tok=u3_token) - # Get the deltas! There should be two -- day 1, and day 2. - r = self.get_success(self.store.get_deltas_for_room(room_1, 0)) + # self.handler.notify_new_event() + + # We need to let the delta processor advance… + self.pump(10 * 60) + + # Get the slices! There should be two -- day 1, and day 2. + r = self.get_success(self.store.get_statistics_for_subject("room", room_1, 0)) + + self.assertEqual(len(r), 2) # The oldest has 2 joined members self.assertEqual(r[-1]["joined_members"], 2) @@ -194,111 +250,476 @@ class StatsRoomTests(unittest.HomeserverTestCase): # The newest has 3 self.assertEqual(r[0]["joined_members"], 3) - def test_incorrect_state_transition(self): - """ - If the state transition is not one of (JOIN, INVITE, LEAVE, BAN) to - (JOIN, INVITE, LEAVE, BAN), an error is raised. - """ - events = { - "a1": {"membership": Membership.LEAVE}, - "a2": {"membership": "not a real thing"}, - } - - def get_event(event_id, allow_none=True): - m = Mock() - m.content = events[event_id] - d = defer.Deferred() - self.reactor.callLater(0.0, d.callback, m) - return d - - def get_received_ts(event_id): - return defer.succeed(1) - - self.store.get_received_ts = get_received_ts - self.store.get_event = get_event - - deltas = [ - { - "type": EventTypes.Member, - "state_key": "some_user", - "room_id": "room", - "event_id": "a1", - "prev_event_id": "a2", - "stream_id": 60, - } - ] - - f = self.get_failure(self.handler._handle_deltas(deltas), ValueError) + def test_create_user(self): + """ + When we create a user, it should have statistics already ready. + """ + + u1 = self.register_user("u1", "pass") + + u1stats = self._get_current_stats("user", u1) + + self.assertIsNotNone(u1stats) + + # not in any rooms by default + self.assertEqual(u1stats["joined_rooms"], 0) + + def test_create_room(self): + """ + When we create a room, it should have statistics already ready. + """ + + self._perform_background_initial_update() + + u1 = self.register_user("u1", "pass") + u1token = self.login("u1", "pass") + r1 = self.helper.create_room_as(u1, tok=u1token) + r1stats = self._get_current_stats("room", r1) + r2 = self.helper.create_room_as(u1, tok=u1token, is_public=False) + r2stats = self._get_current_stats("room", r2) + + self.assertIsNotNone(r1stats) + self.assertIsNotNone(r2stats) + + # contains the default things you'd expect in a fresh room self.assertEqual( - f.value.args[0], "'not a real thing' is not a valid prev_membership" - ) - - # And the other way... - deltas = [ - { - "type": EventTypes.Member, - "state_key": "some_user", - "room_id": "room", - "event_id": "a2", - "prev_event_id": "a1", - "stream_id": 100, - } - ] - - f = self.get_failure(self.handler._handle_deltas(deltas), ValueError) + r1stats["total_events"], + EXPT_NUM_STATE_EVTS_IN_FRESH_PUBLIC_ROOM, + "Wrong number of total_events in new room's stats!" + " You may need to update this if more state events are added to" + " the room creation process.", + ) self.assertEqual( - f.value.args[0], "'not a real thing' is not a valid membership" + r2stats["total_events"], + EXPT_NUM_STATE_EVTS_IN_FRESH_PRIVATE_ROOM, + "Wrong number of total_events in new room's stats!" + " You may need to update this if more state events are added to" + " the room creation process.", ) - def test_redacted_prev_event(self): + self.assertEqual( + r1stats["current_state_events"], EXPT_NUM_STATE_EVTS_IN_FRESH_PUBLIC_ROOM + ) + self.assertEqual( + r2stats["current_state_events"], EXPT_NUM_STATE_EVTS_IN_FRESH_PRIVATE_ROOM + ) + + self.assertEqual(r1stats["joined_members"], 1) + self.assertEqual(r1stats["invited_members"], 0) + self.assertEqual(r1stats["banned_members"], 0) + + self.assertEqual(r2stats["joined_members"], 1) + self.assertEqual(r2stats["invited_members"], 0) + self.assertEqual(r2stats["banned_members"], 0) + + def test_send_message_increments_total_events(self): """ - If the prev_event does not exist, then it is assumed to be a LEAVE. + When we send a message, it increments total_events. """ + + self._perform_background_initial_update() + u1 = self.register_user("u1", "pass") - u1_token = self.login("u1", "pass") + u1token = self.login("u1", "pass") + r1 = self.helper.create_room_as(u1, tok=u1token) + r1stats_ante = self._get_current_stats("room", r1) - room_1 = self.helper.create_room_as(u1, tok=u1_token) + self.helper.send(r1, "hiss", tok=u1token) - # Do the initial population of the user directory via the background update - self._add_background_updates() + r1stats_post = self._get_current_stats("room", r1) + + self.assertEqual(r1stats_post["total_events"] - r1stats_ante["total_events"], 1) + + def test_send_state_event_nonoverwriting(self): + """ + When we send a non-overwriting state event, it increments total_events AND current_state_events + """ + + self._perform_background_initial_update() + + u1 = self.register_user("u1", "pass") + u1token = self.login("u1", "pass") + r1 = self.helper.create_room_as(u1, tok=u1token) + + self.helper.send_state( + r1, "cat.hissing", {"value": True}, tok=u1token, state_key="tabby" + ) + + r1stats_ante = self._get_current_stats("room", r1) + + self.helper.send_state( + r1, "cat.hissing", {"value": False}, tok=u1token, state_key="moggy" + ) + + r1stats_post = self._get_current_stats("room", r1) + + self.assertEqual(r1stats_post["total_events"] - r1stats_ante["total_events"], 1) + self.assertEqual( + r1stats_post["current_state_events"] - r1stats_ante["current_state_events"], + 1, + ) + + def test_send_state_event_overwriting(self): + """ + When we send an overwriting state event, it increments total_events ONLY + """ + + self._perform_background_initial_update() + + u1 = self.register_user("u1", "pass") + u1token = self.login("u1", "pass") + r1 = self.helper.create_room_as(u1, tok=u1token) + + self.helper.send_state( + r1, "cat.hissing", {"value": True}, tok=u1token, state_key="tabby" + ) + + r1stats_ante = self._get_current_stats("room", r1) + + self.helper.send_state( + r1, "cat.hissing", {"value": False}, tok=u1token, state_key="tabby" + ) + + r1stats_post = self._get_current_stats("room", r1) + + self.assertEqual(r1stats_post["total_events"] - r1stats_ante["total_events"], 1) + self.assertEqual( + r1stats_post["current_state_events"] - r1stats_ante["current_state_events"], + 0, + ) + + def test_join_first_time(self): + """ + When a user joins a room for the first time, total_events, current_state_events and + joined_members should increase by exactly 1. + """ + + self._perform_background_initial_update() + + u1 = self.register_user("u1", "pass") + u1token = self.login("u1", "pass") + r1 = self.helper.create_room_as(u1, tok=u1token) + + u2 = self.register_user("u2", "pass") + u2token = self.login("u2", "pass") + + r1stats_ante = self._get_current_stats("room", r1) + + self.helper.join(r1, u2, tok=u2token) + + r1stats_post = self._get_current_stats("room", r1) + + self.assertEqual(r1stats_post["total_events"] - r1stats_ante["total_events"], 1) + self.assertEqual( + r1stats_post["current_state_events"] - r1stats_ante["current_state_events"], + 1, + ) + self.assertEqual( + r1stats_post["joined_members"] - r1stats_ante["joined_members"], 1 + ) + + def test_join_after_leave(self): + """ + When a user joins a room after being previously left, total_events and + joined_members should increase by exactly 1. + current_state_events should not increase. + left_members should decrease by exactly 1. + """ + + self._perform_background_initial_update() + + u1 = self.register_user("u1", "pass") + u1token = self.login("u1", "pass") + r1 = self.helper.create_room_as(u1, tok=u1token) + + u2 = self.register_user("u2", "pass") + u2token = self.login("u2", "pass") + + self.helper.join(r1, u2, tok=u2token) + self.helper.leave(r1, u2, tok=u2token) + + r1stats_ante = self._get_current_stats("room", r1) + + self.helper.join(r1, u2, tok=u2token) + + r1stats_post = self._get_current_stats("room", r1) + + self.assertEqual(r1stats_post["total_events"] - r1stats_ante["total_events"], 1) + self.assertEqual( + r1stats_post["current_state_events"] - r1stats_ante["current_state_events"], + 0, + ) + self.assertEqual( + r1stats_post["joined_members"] - r1stats_ante["joined_members"], +1 + ) + self.assertEqual( + r1stats_post["left_members"] - r1stats_ante["left_members"], -1 + ) + + def test_invited(self): + """ + When a user invites another user, current_state_events, total_events and + invited_members should increase by exactly 1. + """ + + self._perform_background_initial_update() + + u1 = self.register_user("u1", "pass") + u1token = self.login("u1", "pass") + r1 = self.helper.create_room_as(u1, tok=u1token) + + u2 = self.register_user("u2", "pass") + + r1stats_ante = self._get_current_stats("room", r1) + + self.helper.invite(r1, u1, u2, tok=u1token) + + r1stats_post = self._get_current_stats("room", r1) + + self.assertEqual(r1stats_post["total_events"] - r1stats_ante["total_events"], 1) + self.assertEqual( + r1stats_post["current_state_events"] - r1stats_ante["current_state_events"], + 1, + ) + self.assertEqual( + r1stats_post["invited_members"] - r1stats_ante["invited_members"], +1 + ) + + def test_join_after_invite(self): + """ + When a user joins a room after being invited, total_events and + joined_members should increase by exactly 1. + current_state_events should not increase. + invited_members should decrease by exactly 1. + """ + + self._perform_background_initial_update() + + u1 = self.register_user("u1", "pass") + u1token = self.login("u1", "pass") + r1 = self.helper.create_room_as(u1, tok=u1token) + + u2 = self.register_user("u2", "pass") + u2token = self.login("u2", "pass") + + self.helper.invite(r1, u1, u2, tok=u1token) + + r1stats_ante = self._get_current_stats("room", r1) + + self.helper.join(r1, u2, tok=u2token) + + r1stats_post = self._get_current_stats("room", r1) + + self.assertEqual(r1stats_post["total_events"] - r1stats_ante["total_events"], 1) + self.assertEqual( + r1stats_post["current_state_events"] - r1stats_ante["current_state_events"], + 0, + ) + self.assertEqual( + r1stats_post["joined_members"] - r1stats_ante["joined_members"], +1 + ) + self.assertEqual( + r1stats_post["invited_members"] - r1stats_ante["invited_members"], -1 + ) + + def test_left(self): + """ + When a user leaves a room after joining, total_events and + left_members should increase by exactly 1. + current_state_events should not increase. + joined_members should decrease by exactly 1. + """ + + self._perform_background_initial_update() + + u1 = self.register_user("u1", "pass") + u1token = self.login("u1", "pass") + r1 = self.helper.create_room_as(u1, tok=u1token) + + u2 = self.register_user("u2", "pass") + u2token = self.login("u2", "pass") + + self.helper.join(r1, u2, tok=u2token) + + r1stats_ante = self._get_current_stats("room", r1) + + self.helper.leave(r1, u2, tok=u2token) + + r1stats_post = self._get_current_stats("room", r1) + + self.assertEqual(r1stats_post["total_events"] - r1stats_ante["total_events"], 1) + self.assertEqual( + r1stats_post["current_state_events"] - r1stats_ante["current_state_events"], + 0, + ) + self.assertEqual( + r1stats_post["left_members"] - r1stats_ante["left_members"], +1 + ) + self.assertEqual( + r1stats_post["joined_members"] - r1stats_ante["joined_members"], -1 + ) + + def test_banned(self): + """ + When a user is banned from a room after joining, total_events and + left_members should increase by exactly 1. + current_state_events should not increase. + banned_members should decrease by exactly 1. + """ + + self._perform_background_initial_update() + + u1 = self.register_user("u1", "pass") + u1token = self.login("u1", "pass") + r1 = self.helper.create_room_as(u1, tok=u1token) + + u2 = self.register_user("u2", "pass") + u2token = self.login("u2", "pass") + + self.helper.join(r1, u2, tok=u2token) + + r1stats_ante = self._get_current_stats("room", r1) + + self.helper.change_membership(r1, u1, u2, "ban", tok=u1token) + + r1stats_post = self._get_current_stats("room", r1) + + self.assertEqual(r1stats_post["total_events"] - r1stats_ante["total_events"], 1) + self.assertEqual( + r1stats_post["current_state_events"] - r1stats_ante["current_state_events"], + 0, + ) + self.assertEqual( + r1stats_post["banned_members"] - r1stats_ante["banned_members"], +1 + ) + self.assertEqual( + r1stats_post["joined_members"] - r1stats_ante["joined_members"], -1 + ) + + def test_initial_background_update(self): + """ + Test that statistics can be generated by the initial background update + handler. + + This test also tests that stats rows are not created for new subjects + when stats are disabled. However, it may be desirable to change this + behaviour eventually to still keep current rows. + """ + + self.hs.config.stats_enabled = False + + u1 = self.register_user("u1", "pass") + u1token = self.login("u1", "pass") + r1 = self.helper.create_room_as(u1, tok=u1token) + + # test that these subjects, which were created during a time of disabled + # stats, do not have stats. + self.assertIsNone(self._get_current_stats("room", r1)) + self.assertIsNone(self._get_current_stats("user", u1)) + + self.hs.config.stats_enabled = True + + self._perform_background_initial_update() + + r1stats = self._get_current_stats("room", r1) + u1stats = self._get_current_stats("user", u1) + + self.assertEqual(r1stats["joined_members"], 1) + self.assertEqual( + r1stats["current_state_events"], EXPT_NUM_STATE_EVTS_IN_FRESH_PUBLIC_ROOM + ) + + self.assertEqual(u1stats["joined_rooms"], 1) + + def test_incomplete_stats(self): + """ + This tests that we track incomplete statistics. + + We first test that incomplete stats are incrementally generated, + following the preparation of a background regen. + + We then test that these incomplete rows are completed by the background + regen. + """ + + u1 = self.register_user("u1", "pass") + u1token = self.login("u1", "pass") + u2 = self.register_user("u2", "pass") + u2token = self.login("u2", "pass") + u3 = self.register_user("u3", "pass") + r1 = self.helper.create_room_as(u1, tok=u1token, is_public=False) + + # preparation stage of the initial background update + # Ugh, have to reset this flag + self.store._all_done = False + + self.get_success( + self.store._simple_delete( + "room_stats_current", {"1": 1}, "test_delete_stats" + ) + ) + self.get_success( + self.store._simple_delete( + "user_stats_current", {"1": 1}, "test_delete_stats" + ) + ) + + self.helper.invite(r1, u1, u2, tok=u1token) + self.helper.join(r1, u2, tok=u2token) + self.helper.invite(r1, u1, u3, tok=u1token) + self.helper.send(r1, "thou shalt yield", tok=u1token) + + # now do the background updates + + self.store._all_done = False + self.get_success( + self.store._simple_insert( + "background_updates", + { + "update_name": "populate_stats_process_rooms", + "progress_json": "{}", + "depends_on": "populate_stats_prepare", + }, + ) + ) + self.get_success( + self.store._simple_insert( + "background_updates", + { + "update_name": "populate_stats_process_users", + "progress_json": "{}", + "depends_on": "populate_stats_process_rooms", + }, + ) + ) + self.get_success( + self.store._simple_insert( + "background_updates", + { + "update_name": "populate_stats_cleanup", + "progress_json": "{}", + "depends_on": "populate_stats_process_users", + }, + ) + ) while not self.get_success(self.store.has_completed_background_updates()): self.get_success(self.store.do_next_background_update(100), by=0.1) - events = {"a1": None, "a2": {"membership": Membership.JOIN}} - - def get_event(event_id, allow_none=True): - if events.get(event_id): - m = Mock() - m.content = events[event_id] - else: - m = None - d = defer.Deferred() - self.reactor.callLater(0.0, d.callback, m) - return d - - def get_received_ts(event_id): - return defer.succeed(1) - - self.store.get_received_ts = get_received_ts - self.store.get_event = get_event - - deltas = [ - { - "type": EventTypes.Member, - "state_key": "some_user:test", - "room_id": room_1, - "event_id": "a2", - "prev_event_id": "a1", - "stream_id": 100, - } - ] - - # Handle our fake deltas, which has a user going from LEAVE -> JOIN. - self.get_success(self.handler._handle_deltas(deltas)) - - # One delta, with two joined members -- the room creator, and our fake - # user. - r = self.get_success(self.store.get_deltas_for_room(room_1, 0)) - self.assertEqual(len(r), 1) - self.assertEqual(r[0]["joined_members"], 2) + r1stats_complete = self._get_current_stats("room", r1) + u1stats_complete = self._get_current_stats("user", u1) + u2stats_complete = self._get_current_stats("user", u2) + + # now we make our assertions + + # check that _complete rows are complete and correct + self.assertEqual(r1stats_complete["joined_members"], 2) + self.assertEqual(r1stats_complete["invited_members"], 1) + + self.assertEqual( + r1stats_complete["current_state_events"], + 2 + EXPT_NUM_STATE_EVTS_IN_FRESH_PRIVATE_ROOM, + ) + + self.assertEqual(u1stats_complete["joined_rooms"], 1) + self.assertEqual(u2stats_complete["joined_rooms"], 1) diff --git a/tests/rest/client/v1/utils.py b/tests/rest/client/v1/utils.py index 9915367144..cdded88b7f 100644 --- a/tests/rest/client/v1/utils.py +++ b/tests/rest/client/v1/utils.py @@ -128,8 +128,12 @@ class RestHelper(object): return channel.json_body - def send_state(self, room_id, event_type, body, tok, expect_code=200): - path = "/_matrix/client/r0/rooms/%s/state/%s" % (room_id, event_type) + def send_state(self, room_id, event_type, body, tok, expect_code=200, state_key=""): + path = "/_matrix/client/r0/rooms/%s/state/%s/%s" % ( + room_id, + event_type, + state_key, + ) if tok: path = path + "?access_token=%s" % tok -- cgit 1.5.1 From 3ff0422d2dbfa668df365da99a4b7caeea85528d Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Thu, 5 Sep 2019 17:16:03 +0100 Subject: Make redaction retention period configurable --- docs/sample_config.yaml | 5 +++++ synapse/config/server.py | 15 +++++++++++++++ synapse/storage/events.py | 6 ++++-- tests/storage/test_redaction.py | 4 +++- 4 files changed, 27 insertions(+), 3 deletions(-) (limited to 'tests') diff --git a/docs/sample_config.yaml b/docs/sample_config.yaml index 43969bbb70..e23b80d2b8 100644 --- a/docs/sample_config.yaml +++ b/docs/sample_config.yaml @@ -306,6 +306,11 @@ listeners: # #allow_per_room_profiles: false +# How long to keep redacted events in unredacted form in the database. +# By default redactions are kept indefinitely. +# +#redaction_retention_period: 30d + ## TLS ## diff --git a/synapse/config/server.py b/synapse/config/server.py index 2abdef0971..8efab924d4 100644 --- a/synapse/config/server.py +++ b/synapse/config/server.py @@ -162,6 +162,16 @@ class ServerConfig(Config): self.mau_trial_days = config.get("mau_trial_days", 0) + # How long to keep redacted events in the database in unredacted form + # before redacting them. + redaction_retention_period = config.get("redaction_retention_period") + if redaction_retention_period: + self.redaction_retention_period = self.parse_duration( + redaction_retention_period + ) + else: + self.redaction_retention_period = None + # Options to disable HS self.hs_disabled = config.get("hs_disabled", False) self.hs_disabled_message = config.get("hs_disabled_message", "") @@ -718,6 +728,11 @@ class ServerConfig(Config): # Defaults to 'true'. # #allow_per_room_profiles: false + + # How long to keep redacted events in unredacted form in the database. + # By default redactions are kept indefinitely. + # + #redaction_retention_period: 30d """ % locals() ) diff --git a/synapse/storage/events.py b/synapse/storage/events.py index 2970da6829..d0d1781c90 100644 --- a/synapse/storage/events.py +++ b/synapse/storage/events.py @@ -1566,10 +1566,12 @@ class EventsStore( Deferred """ - if self.stream_ordering_month_ago is None: + if not self.hs.config.redaction_retention_period: return - max_pos = self.stream_ordering_month_ago + max_pos = yield self.find_first_stream_ordering_after_ts( + self._clock.time_msec() - self.hs.config.redaction_retention_period + ) # We fetch all redactions that point to an event that we have that has # a stream ordering from over a month ago, that we haven't yet censored diff --git a/tests/storage/test_redaction.py b/tests/storage/test_redaction.py index 0c9f3c7071..f0e86d41a8 100644 --- a/tests/storage/test_redaction.py +++ b/tests/storage/test_redaction.py @@ -344,7 +344,9 @@ class RedactionTestCase(unittest.HomeserverTestCase): {"content": {"body": "t", "msgtype": "message"}}, json.loads(event_json) ) - # Advance by 30 days + # Advance by 30 days, then advance again to ensure that the looping call + # for updating the stream position gets called and then the looping call + # for the censoring gets called. self.reactor.advance(60 * 60 * 24 * 31) self.reactor.advance(60 * 60 * 2) -- cgit 1.5.1 From ad9b64b4969537ac339469152eaa437bcf4b6609 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Thu, 5 Sep 2019 17:17:47 +0100 Subject: Fix test --- tests/storage/test_redaction.py | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'tests') diff --git a/tests/storage/test_redaction.py b/tests/storage/test_redaction.py index f0e86d41a8..deecfad9fb 100644 --- a/tests/storage/test_redaction.py +++ b/tests/storage/test_redaction.py @@ -31,8 +31,10 @@ from tests.utils import create_room class RedactionTestCase(unittest.HomeserverTestCase): def make_homeserver(self, reactor, clock): + config = self.default_config() + config["redaction_retention_period"] = "30d" return self.setup_test_homeserver( - resource_for_federation=Mock(), http_client=None + resource_for_federation=Mock(), http_client=None, config=config ) def prepare(self, reactor, clock, hs): -- cgit 1.5.1 From 55d5b3af8863167432017f23cd8a04a0c14c9d23 Mon Sep 17 00:00:00 2001 From: Amber Brown Date: Sat, 7 Sep 2019 01:45:51 +1000 Subject: Servers-known-about statistic (#5981) --- changelog.d/5981.feature | 1 + docs/sample_config.yaml | 10 ++++ synapse/config/metrics.py | 31 ++++++++++ synapse/storage/roommember.py | 59 ++++++++++++++++++ tests/config/test_generate.py | 25 ++++---- tests/config/test_load.py | 34 +++++++---- tests/storage/test_roommember.py | 126 +++++++++++++++++++++++++++------------ 7 files changed, 226 insertions(+), 60 deletions(-) create mode 100644 changelog.d/5981.feature (limited to 'tests') diff --git a/changelog.d/5981.feature b/changelog.d/5981.feature new file mode 100644 index 0000000000..e39514273d --- /dev/null +++ b/changelog.d/5981.feature @@ -0,0 +1 @@ +Setting metrics_flags.known_servers to True in the configuration will publish the synapse_federation_known_servers metric over Prometheus. This represents the total number of servers your server knows about (i.e. is in rooms with), including itself. diff --git a/docs/sample_config.yaml b/docs/sample_config.yaml index 186cdbedd2..93c0edd8ce 100644 --- a/docs/sample_config.yaml +++ b/docs/sample_config.yaml @@ -958,6 +958,16 @@ account_threepid_delegates: #sentry: # dsn: "..." +# Flags to enable Prometheus metrics which are not suitable to be +# enabled by default, either for performance reasons or limited use. +# +metrics_flags: + # Publish synapse_federation_known_servers, a g auge of the number of + # servers this homeserver knows about, including itself. May cause + # performance problems on large homeservers. + # + #known_servers: true + # Whether or not to report anonymized homeserver usage statistics. # report_stats: true|false diff --git a/synapse/config/metrics.py b/synapse/config/metrics.py index 3698441963..653b990e67 100644 --- a/synapse/config/metrics.py +++ b/synapse/config/metrics.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2015, 2016 OpenMarket Ltd +# Copyright 2019 The Matrix.org Foundation C.I.C. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,6 +14,8 @@ # See the License for the specific language governing permissions and # limitations under the License. +import attr + from ._base import Config, ConfigError MISSING_SENTRY = """Missing sentry-sdk library. This is required to enable sentry @@ -20,6 +23,18 @@ MISSING_SENTRY = """Missing sentry-sdk library. This is required to enable sentr """ +@attr.s +class MetricsFlags(object): + known_servers = attr.ib(default=False, validator=attr.validators.instance_of(bool)) + + @classmethod + def all_off(cls): + """ + Instantiate the flags with all options set to off. + """ + return cls(**{x.name: False for x in attr.fields(cls)}) + + class MetricsConfig(Config): def read_config(self, config, **kwargs): self.enable_metrics = config.get("enable_metrics", False) @@ -27,6 +42,12 @@ class MetricsConfig(Config): self.metrics_port = config.get("metrics_port") self.metrics_bind_host = config.get("metrics_bind_host", "127.0.0.1") + if self.enable_metrics: + _metrics_config = config.get("metrics_flags") or {} + self.metrics_flags = MetricsFlags(**_metrics_config) + else: + self.metrics_flags = MetricsFlags.all_off() + self.sentry_enabled = "sentry" in config if self.sentry_enabled: try: @@ -58,6 +79,16 @@ class MetricsConfig(Config): #sentry: # dsn: "..." + # Flags to enable Prometheus metrics which are not suitable to be + # enabled by default, either for performance reasons or limited use. + # + metrics_flags: + # Publish synapse_federation_known_servers, a g auge of the number of + # servers this homeserver knows about, including itself. May cause + # performance problems on large homeservers. + # + #known_servers: true + # Whether or not to report anonymized homeserver usage statistics. """ diff --git a/synapse/storage/roommember.py b/synapse/storage/roommember.py index f8b682ebd9..4df8ebdacd 100644 --- a/synapse/storage/roommember.py +++ b/synapse/storage/roommember.py @@ -24,8 +24,10 @@ from canonicaljson import json from twisted.internet import defer from synapse.api.constants import EventTypes, Membership +from synapse.metrics import LaterGauge from synapse.metrics.background_process_metrics import run_as_background_process from synapse.storage._base import LoggingTransaction +from synapse.storage.engines import Sqlite3Engine from synapse.storage.events_worker import EventsWorkerStore from synapse.types import get_domain_from_id from synapse.util.async_helpers import Linearizer @@ -74,6 +76,63 @@ class RoomMemberWorkerStore(EventsWorkerStore): self._check_safe_current_state_events_membership_updated_txn(txn) txn.close() + if self.hs.config.metrics_flags.known_servers: + self._known_servers_count = 1 + self.hs.get_clock().looping_call( + run_as_background_process, + 60 * 1000, + "_count_known_servers", + self._count_known_servers, + ) + self.hs.get_clock().call_later( + 1000, + run_as_background_process, + "_count_known_servers", + self._count_known_servers, + ) + LaterGauge( + "synapse_federation_known_servers", + "", + [], + lambda: self._known_servers_count, + ) + + @defer.inlineCallbacks + def _count_known_servers(self): + """ + Count the servers that this server knows about. + + The statistic is stored on the class for the + `synapse_federation_known_servers` LaterGauge to collect. + """ + + def _transact(txn): + if isinstance(self.database_engine, Sqlite3Engine): + query = """ + SELECT COUNT(DISTINCT substr(out.user_id, pos+1)) + FROM ( + SELECT rm.user_id as user_id, instr(rm.user_id, ':') + AS pos FROM room_memberships as rm + INNER JOIN current_state_events as c ON rm.event_id = c.event_id + WHERE c.type = 'm.room.member' + ) as out + """ + else: + query = """ + SELECT COUNT(DISTINCT split_part(state_key, ':', 2)) + FROM current_state_events + WHERE type = 'm.room.member' AND membership = 'join'; + """ + txn.execute(query) + return list(txn)[0][0] + + count = yield self.runInteraction("get_known_servers", _transact) + + # We always know about ourselves, even if we have nothing in + # room_memberships (for example, the server is new). + self._known_servers_count = max([count, 1]) + return self._known_servers_count + def _check_safe_current_state_events_membership_updated_txn(self, txn): """Checks if it is safe to assume the new current_state_events membership column is up to date diff --git a/tests/config/test_generate.py b/tests/config/test_generate.py index 5017cbce85..2684e662de 100644 --- a/tests/config/test_generate.py +++ b/tests/config/test_generate.py @@ -17,6 +17,8 @@ import os.path import re import shutil import tempfile +from contextlib import redirect_stdout +from io import StringIO from synapse.config.homeserver import HomeServerConfig @@ -32,17 +34,18 @@ class ConfigGenerationTestCase(unittest.TestCase): shutil.rmtree(self.dir) def test_generate_config_generates_files(self): - HomeServerConfig.load_or_generate_config( - "", - [ - "--generate-config", - "-c", - self.file, - "--report-stats=yes", - "-H", - "lemurs.win", - ], - ) + with redirect_stdout(StringIO()): + HomeServerConfig.load_or_generate_config( + "", + [ + "--generate-config", + "-c", + self.file, + "--report-stats=yes", + "-H", + "lemurs.win", + ], + ) self.assertSetEqual( set(["homeserver.yaml", "lemurs.win.log.config", "lemurs.win.signing.key"]), diff --git a/tests/config/test_load.py b/tests/config/test_load.py index 6bfc1970ad..b3e557bd6a 100644 --- a/tests/config/test_load.py +++ b/tests/config/test_load.py @@ -15,6 +15,8 @@ import os.path import shutil import tempfile +from contextlib import redirect_stdout +from io import StringIO import yaml @@ -26,7 +28,6 @@ from tests import unittest class ConfigLoadingTestCase(unittest.TestCase): def setUp(self): self.dir = tempfile.mkdtemp() - print(self.dir) self.file = os.path.join(self.dir, "homeserver.yaml") def tearDown(self): @@ -94,18 +95,27 @@ class ConfigLoadingTestCase(unittest.TestCase): ) self.assertTrue(config.enable_registration) + def test_stats_enabled(self): + self.generate_config_and_remove_lines_containing("enable_metrics") + self.add_lines_to_config(["enable_metrics: true"]) + + # The default Metrics Flags are off by default. + config = HomeServerConfig.load_config("", ["-c", self.file]) + self.assertFalse(config.metrics_flags.known_servers) + def generate_config(self): - HomeServerConfig.load_or_generate_config( - "", - [ - "--generate-config", - "-c", - self.file, - "--report-stats=yes", - "-H", - "lemurs.win", - ], - ) + with redirect_stdout(StringIO()): + HomeServerConfig.load_or_generate_config( + "", + [ + "--generate-config", + "-c", + self.file, + "--report-stats=yes", + "-H", + "lemurs.win", + ], + ) def generate_config_and_remove_lines_containing(self, needle): self.generate_config() diff --git a/tests/storage/test_roommember.py b/tests/storage/test_roommember.py index 64cb294c37..447a3c6ffb 100644 --- a/tests/storage/test_roommember.py +++ b/tests/storage/test_roommember.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2014-2016 OpenMarket Ltd +# Copyright 2019 The Matrix.org Foundation C.I.C. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,78 +14,129 @@ # See the License for the specific language governing permissions and # limitations under the License. - -from mock import Mock - -from twisted.internet import defer +from unittest.mock import Mock from synapse.api.constants import EventTypes, Membership from synapse.api.room_versions import RoomVersions -from synapse.types import Requester, RoomID, UserID +from synapse.rest.admin import register_servlets_for_client_rest_resource +from synapse.rest.client.v1 import login, room +from synapse.types import Requester, UserID from tests import unittest -from tests.utils import create_room, setup_test_homeserver -class RoomMemberStoreTestCase(unittest.TestCase): - @defer.inlineCallbacks - def setUp(self): - hs = yield setup_test_homeserver( - self.addCleanup, resource_for_federation=Mock(), http_client=None +class RoomMemberStoreTestCase(unittest.HomeserverTestCase): + + servlets = [ + login.register_servlets, + register_servlets_for_client_rest_resource, + room.register_servlets, + ] + + def make_homeserver(self, reactor, clock): + hs = self.setup_test_homeserver( + resource_for_federation=Mock(), http_client=None ) + return hs + + def prepare(self, reactor, clock, hs): + # We can't test the RoomMemberStore on its own without the other event # storage logic self.store = hs.get_datastore() self.event_builder_factory = hs.get_event_builder_factory() self.event_creation_handler = hs.get_event_creation_handler() - self.u_alice = UserID.from_string("@alice:test") - self.u_bob = UserID.from_string("@bob:test") + self.u_alice = self.register_user("alice", "pass") + self.t_alice = self.login("alice", "pass") + self.u_bob = self.register_user("bob", "pass") # User elsewhere on another host self.u_charlie = UserID.from_string("@charlie:elsewhere") - self.room = RoomID.from_string("!abc123:test") - - yield create_room(hs, self.room.to_string(), self.u_alice.to_string()) - - @defer.inlineCallbacks def inject_room_member(self, room, user, membership, replaces_state=None): builder = self.event_builder_factory.for_room_version( RoomVersions.V1, { "type": EventTypes.Member, - "sender": user.to_string(), - "state_key": user.to_string(), - "room_id": room.to_string(), + "sender": user, + "state_key": user, + "room_id": room, "content": {"membership": membership}, }, ) - event, context = yield self.event_creation_handler.create_new_client_event( - builder + event, context = self.get_success( + self.event_creation_handler.create_new_client_event(builder) ) - yield self.store.persist_event(event, context) + self.get_success(self.store.persist_event(event, context)) return event - @defer.inlineCallbacks def test_one_member(self): - yield self.inject_room_member(self.room, self.u_alice, Membership.JOIN) - - self.assertEquals( - [self.room.to_string()], - [ - m.room_id - for m in ( - yield self.store.get_rooms_for_user_where_membership_is( - self.u_alice.to_string(), [Membership.JOIN] - ) - ) - ], + + # Alice creates the room, and is automatically joined + self.room = self.helper.create_room_as(self.u_alice, tok=self.t_alice) + + rooms_for_user = self.get_success( + self.store.get_rooms_for_user_where_membership_is( + self.u_alice, [Membership.JOIN] + ) ) + self.assertEquals([self.room], [m.room_id for m in rooms_for_user]) + + def test_count_known_servers(self): + """ + _count_known_servers will calculate how many servers are in a room. + """ + self.room = self.helper.create_room_as(self.u_alice, tok=self.t_alice) + self.inject_room_member(self.room, self.u_bob, Membership.JOIN) + self.inject_room_member(self.room, self.u_charlie.to_string(), Membership.JOIN) + + servers = self.get_success(self.store._count_known_servers()) + self.assertEqual(servers, 2) + + def test_count_known_servers_stat_counter_disabled(self): + """ + If enabled, the metrics for how many servers are known will be counted. + """ + self.assertTrue("_known_servers_count" not in self.store.__dict__.keys()) + + self.room = self.helper.create_room_as(self.u_alice, tok=self.t_alice) + self.inject_room_member(self.room, self.u_bob, Membership.JOIN) + self.inject_room_member(self.room, self.u_charlie.to_string(), Membership.JOIN) + + self.pump(20) + + self.assertTrue("_known_servers_count" not in self.store.__dict__.keys()) + + @unittest.override_config( + {"enable_metrics": True, "metrics_flags": {"known_servers": True}} + ) + def test_count_known_servers_stat_counter_enabled(self): + """ + If enabled, the metrics for how many servers are known will be counted. + """ + # Initialises to 1 -- itself + self.assertEqual(self.store._known_servers_count, 1) + + self.pump(20) + + # No rooms have been joined, so technically the SQL returns 0, but it + # will still say it knows about itself. + self.assertEqual(self.store._known_servers_count, 1) + + self.room = self.helper.create_room_as(self.u_alice, tok=self.t_alice) + self.inject_room_member(self.room, self.u_bob, Membership.JOIN) + self.inject_room_member(self.room, self.u_charlie.to_string(), Membership.JOIN) + + self.pump(20) + + # It now knows about Charlie's server. + self.assertEqual(self.store._known_servers_count, 2) + class CurrentStateMembershipUpdateTestCase(unittest.HomeserverTestCase): def prepare(self, reactor, clock, homeserver): -- cgit 1.5.1 From be618e055178f4aa9865ab426182218312bed07f Mon Sep 17 00:00:00 2001 From: Jason Robinson Date: Mon, 9 Sep 2019 14:43:51 +0300 Subject: Only count real users when checking for auto-creation of auto-join room Previously if the first registered user was a "support" or "bot" user, when the first real user registers, the auto-join rooms were not created. Fix to exclude non-real (ie users with a special user type) users when counting how many users there are to determine whether we should auto-create a room. Signed-off-by: Jason Robinson --- changelog.d/6004.bugfix | 1 + synapse/handlers/register.py | 12 ++++-------- synapse/storage/registration.py | 39 +++++++++++++++++++++++++++++++++++++++ tests/handlers/test_register.py | 29 +++++++++++++++++++++++++++-- 4 files changed, 71 insertions(+), 10 deletions(-) create mode 100644 changelog.d/6004.bugfix (limited to 'tests') diff --git a/changelog.d/6004.bugfix b/changelog.d/6004.bugfix new file mode 100644 index 0000000000..45c179c8fd --- /dev/null +++ b/changelog.d/6004.bugfix @@ -0,0 +1 @@ +Only count real users when checking for auto-creation of auto-join room. diff --git a/synapse/handlers/register.py b/synapse/handlers/register.py index 975da57ffd..06bd03b77c 100644 --- a/synapse/handlers/register.py +++ b/synapse/handlers/register.py @@ -275,16 +275,12 @@ class RegistrationHandler(BaseHandler): fake_requester = create_requester(user_id) # try to create the room if we're the first real user on the server. Note - # that an auto-generated support user is not a real user and will never be + # that an auto-generated support or bot user is not a real user and will never be # the user to create the room should_auto_create_rooms = False - is_support = yield self.store.is_support_user(user_id) - # There is an edge case where the first user is the support user, then - # the room is never created, though this seems unlikely and - # recoverable from given the support user being involved in the first - # place. - if self.hs.config.autocreate_auto_join_rooms and not is_support: - count = yield self.store.count_all_users() + is_real_user = yield self.store.is_real_user(user_id) + if self.hs.config.autocreate_auto_join_rooms and is_real_user: + count = yield self.store.count_real_users() should_auto_create_rooms = count == 1 for r in self.hs.config.auto_join_rooms: logger.info("Auto-joining %s to %s", user_id, r) diff --git a/synapse/storage/registration.py b/synapse/storage/registration.py index 5138792a5f..b054d86ae0 100644 --- a/synapse/storage/registration.py +++ b/synapse/storage/registration.py @@ -322,6 +322,21 @@ class RegistrationWorkerStore(SQLBaseStore): return None + @cachedInlineCallbacks() + def is_real_user(self, user_id): + """Determines if the user is a real user, ie does not have a 'user_type'. + + Args: + user_id (str): user id to test + + Returns: + Deferred[bool]: True if user 'user_type' is null or empty string + """ + res = yield self.runInteraction( + "is_real_user", self.is_real_user_txn, user_id + ) + return res + @cachedInlineCallbacks() def is_support_user(self, user_id): """Determines if the user is of type UserTypes.SUPPORT @@ -337,6 +352,16 @@ class RegistrationWorkerStore(SQLBaseStore): ) return res + def is_real_user_txn(self, txn, user_id): + res = self._simple_select_one_onecol_txn( + txn=txn, + table="users", + keyvalues={"name": user_id}, + retcol="user_type", + allow_none=True, + ) + return True if res is None or res == "" else False + def is_support_user_txn(self, txn, user_id): res = self._simple_select_one_onecol_txn( txn=txn, @@ -421,6 +446,20 @@ class RegistrationWorkerStore(SQLBaseStore): ret = yield self.runInteraction("count_users", _count_users) return ret + @defer.inlineCallbacks + def count_real_users(self): + """Counts all users without a special user_type registered on the homeserver.""" + + def _count_users(txn): + txn.execute("SELECT COUNT(*) AS users FROM users where user_type is null or user_type = ''") + rows = self.cursor_to_dict(txn) + if rows: + return rows[0]["users"] + return 0 + + ret = yield self.runInteraction("count_real_users", _count_users) + return ret + @defer.inlineCallbacks def find_next_generated_user_id_localpart(self): """ diff --git a/tests/handlers/test_register.py b/tests/handlers/test_register.py index e10296a5e4..1e9ba3a201 100644 --- a/tests/handlers/test_register.py +++ b/tests/handlers/test_register.py @@ -171,11 +171,11 @@ class RegistrationTestCase(unittest.HomeserverTestCase): rooms = self.get_success(self.store.get_rooms_for_user(user_id)) self.assertEqual(len(rooms), 0) - def test_auto_create_auto_join_rooms_when_support_user_exists(self): + def test_auto_create_auto_join_rooms_when_user_is_not_a_real_user(self): room_alias_str = "#room:test" self.hs.config.auto_join_rooms = [room_alias_str] - self.store.is_support_user = Mock(return_value=True) + self.store.is_real_user = Mock(return_value=False) user_id = self.get_success(self.handler.register_user(localpart="support")) rooms = self.get_success(self.store.get_rooms_for_user(user_id)) self.assertEqual(len(rooms), 0) @@ -183,6 +183,31 @@ class RegistrationTestCase(unittest.HomeserverTestCase): room_alias = RoomAlias.from_string(room_alias_str) self.get_failure(directory_handler.get_association(room_alias), SynapseError) + def test_auto_create_auto_join_rooms_when_user_is_the_first_real_user(self): + room_alias_str = "#room:test" + self.hs.config.auto_join_rooms = [room_alias_str] + + self.store.count_real_users = Mock(return_value=1) + self.store.is_real_user = Mock(return_value=True) + user_id = self.get_success(self.handler.register_user(localpart="real")) + rooms = self.get_success(self.store.get_rooms_for_user(user_id)) + directory_handler = self.hs.get_handlers().directory_handler + room_alias = RoomAlias.from_string(room_alias_str) + room_id = self.get_success(directory_handler.get_association(room_alias)) + + self.assertTrue(room_id["room_id"] in rooms) + self.assertEqual(len(rooms), 1) + + def test_auto_create_auto_join_rooms_when_user_is_not_the_first_real_user(self): + room_alias_str = "#room:test" + self.hs.config.auto_join_rooms = [room_alias_str] + + self.store.count_real_users = Mock(return_value=2) + self.store.is_real_user = Mock(return_value=True) + user_id = self.get_success(self.handler.register_user(localpart="real")) + rooms = self.get_success(self.store.get_rooms_for_user(user_id)) + self.assertEqual(len(rooms), 0) + def test_auto_create_auto_join_where_no_consent(self): """Test to ensure that the first user is not auto-joined to a room if they have not given general consent. -- cgit 1.5.1 From aeb9b2179eaa4b468bec937570d3ac7de7ccaaea Mon Sep 17 00:00:00 2001 From: Amber Brown Date: Tue, 10 Sep 2019 00:14:58 +1000 Subject: Add a build info metric to Prometheus (#6005) --- changelog.d/6005.feature | 1 + synapse/metrics/__init__.py | 12 ++++++++++++ tests/test_metrics.py | 22 ++++++++++++++++++++-- 3 files changed, 33 insertions(+), 2 deletions(-) create mode 100644 changelog.d/6005.feature (limited to 'tests') diff --git a/changelog.d/6005.feature b/changelog.d/6005.feature new file mode 100644 index 0000000000..ed6491d3e4 --- /dev/null +++ b/changelog.d/6005.feature @@ -0,0 +1 @@ +The new Prometheus metric `synapse_build_info` exposes the Python version, OS version, and Synapse version of the running server. diff --git a/synapse/metrics/__init__.py b/synapse/metrics/__init__.py index 488280b4a6..b5c9595cb9 100644 --- a/synapse/metrics/__init__.py +++ b/synapse/metrics/__init__.py @@ -29,11 +29,13 @@ from prometheus_client.core import REGISTRY, GaugeMetricFamily, HistogramMetricF from twisted.internet import reactor +import synapse from synapse.metrics._exposition import ( MetricsResource, generate_latest, start_http_server, ) +from synapse.util.versionstring import get_version_string logger = logging.getLogger(__name__) @@ -385,6 +387,16 @@ event_processing_last_ts = Gauge("synapse_event_processing_last_ts", "", ["name" # finished being processed. event_processing_lag = Gauge("synapse_event_processing_lag", "", ["name"]) +# Build info of the running server. +build_info = Gauge( + "synapse_build_info", "Build information", ["pythonversion", "version", "osversion"] +) +build_info.labels( + " ".join([platform.python_implementation(), platform.python_version()]), + get_version_string(synapse), + " ".join([platform.system(), platform.release()]), +).set(1) + last_ticked = time.time() diff --git a/tests/test_metrics.py b/tests/test_metrics.py index 2edbae5c6d..270f853d60 100644 --- a/tests/test_metrics.py +++ b/tests/test_metrics.py @@ -1,5 +1,6 @@ # -*- coding: utf-8 -*- # Copyright 2018 New Vector Ltd +# Copyright 2019 Matrix.org Foundation C.I.C. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -13,8 +14,7 @@ # See the License for the specific language governing permissions and # limitations under the License. - -from synapse.metrics import InFlightGauge +from synapse.metrics import REGISTRY, InFlightGauge, generate_latest from tests import unittest @@ -111,3 +111,21 @@ class TestMauLimit(unittest.TestCase): } return results + + +class BuildInfoTests(unittest.TestCase): + def test_get_build(self): + """ + The synapse_build_info metric reports the OS version, Python version, + and Synapse version. + """ + items = list( + filter( + lambda x: b"synapse_build_info{" in x, + generate_latest(REGISTRY).split(b"\n"), + ) + ) + self.assertEqual(len(items), 1) + self.assertTrue(b"osversion=" in items[0]) + self.assertTrue(b"pythonversion=" in items[0]) + self.assertTrue(b"version=" in items[0]) -- cgit 1.5.1 From caa9d6fed719a8a80eb4a998d32f09577d04f927 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 11 Sep 2019 11:16:23 +0100 Subject: Add test for admin redaction ratelimiting. --- tests/rest/client/test_redactions.py | 25 +++++++++++++++++++++++++ 1 file changed, 25 insertions(+) (limited to 'tests') diff --git a/tests/rest/client/test_redactions.py b/tests/rest/client/test_redactions.py index fe66e397c4..1b1e991c42 100644 --- a/tests/rest/client/test_redactions.py +++ b/tests/rest/client/test_redactions.py @@ -30,6 +30,14 @@ class RedactionsTestCase(HomeserverTestCase): sync.register_servlets, ] + def make_homeserver(self, reactor, clock): + config = self.default_config() + + config["rc_message"] = {"per_second": 0.2, "burst_count": 10} + config["rc_admin_redaction"] = {"per_second": 1, "burst_count": 100} + + return self.setup_test_homeserver(config=config) + def prepare(self, reactor, clock, hs): # register a couple of users self.mod_user_id = self.register_user("user1", "pass") @@ -177,3 +185,20 @@ class RedactionsTestCase(HomeserverTestCase): self._redact_event( self.other_access_token, self.room_id, create_event_id, expect_code=403 ) + + def test_redact_event_as_moderator_ratelimit(self): + """Tests that the correct ratelimiting is applied to redactions + """ + + message_ids = [] + # as a regular user, send messages to redact + for _ in range(20): + b = self.helper.send(room_id=self.room_id, tok=self.other_access_token) + message_ids.append(b["event_id"]) + self.reactor.advance(10) # To get around ratelimits + + # as the moderator, send a bunch of redactions redaction + for msg_id in message_ids: + # These should all succeed, even though this would be denied by + # standard message ratelimiter + self._redact_event(self.mod_access_token, self.room_id, msg_id) -- cgit 1.5.1 From 57dd41a45b4df5d736e2f30d40926b60f367b500 Mon Sep 17 00:00:00 2001 From: Erik Johnston Date: Wed, 11 Sep 2019 13:54:50 +0100 Subject: Fix comments Co-Authored-By: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com> --- synapse/config/ratelimiting.py | 2 +- synapse/handlers/_base.py | 2 +- synapse/handlers/message.py | 2 +- tests/rest/client/test_redactions.py | 4 ++-- 4 files changed, 5 insertions(+), 5 deletions(-) (limited to 'tests') diff --git a/synapse/config/ratelimiting.py b/synapse/config/ratelimiting.py index b4df6612d6..587e2862b7 100644 --- a/synapse/config/ratelimiting.py +++ b/synapse/config/ratelimiting.py @@ -110,7 +110,7 @@ class RatelimitConfig(Config): # attempts for this account. # - one for ratelimiting redactions by room admins. If this is not explicitly # set then it uses the same ratelimiting as per rc_message. This is useful - # to allow room admins to quickly deal with abuse quickly. + # to allow room admins to deal with abuse quickly. # # The defaults are as shown below. # diff --git a/synapse/handlers/_base.py b/synapse/handlers/_base.py index 853b72d8e7..d15c6282fb 100644 --- a/synapse/handlers/_base.py +++ b/synapse/handlers/_base.py @@ -105,7 +105,7 @@ class BaseHandler(object): if is_admin_redaction and self.hs.config.rc_admin_redaction: # If we have separate config for admin redactions we use a separate - # ratelimiter. + # ratelimiter allowed, time_allowed = self.admin_redaction_ratelimiter.can_do_action( user_id, time_now, diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py index f975909416..1f8272784e 100644 --- a/synapse/handlers/message.py +++ b/synapse/handlers/message.py @@ -731,7 +731,7 @@ class EventCreationHandler(object): if ratelimit: # We check if this is a room admin redacting an event so that we # can apply different ratelimiting. We do this by simply checking - # its not a self-redaction (to avoid having to look up whether the + # it's not a self-redaction (to avoid having to look up whether the # user is actually admin or not). is_admin_redaction = False if event.type == EventTypes.Redaction: diff --git a/tests/rest/client/test_redactions.py b/tests/rest/client/test_redactions.py index 1b1e991c42..d2bcf256fa 100644 --- a/tests/rest/client/test_redactions.py +++ b/tests/rest/client/test_redactions.py @@ -197,8 +197,8 @@ class RedactionsTestCase(HomeserverTestCase): message_ids.append(b["event_id"]) self.reactor.advance(10) # To get around ratelimits - # as the moderator, send a bunch of redactions redaction + # as the moderator, send a bunch of redactions for msg_id in message_ids: # These should all succeed, even though this would be denied by - # standard message ratelimiter + # the standard message ratelimiter self._redact_event(self.mod_access_token, self.room_id, msg_id) -- cgit 1.5.1 From 6d847d8ce69f2cb849633265aaeb4a9df4ff713d Mon Sep 17 00:00:00 2001 From: Jason Robinson Date: Wed, 11 Sep 2019 20:22:18 +0300 Subject: Ensure support users can be registered even if MAU limit is reached This allows support users to be created even on MAU limits via the admin API. Support users are excluded from MAU after creation, so it makes sense to exclude them in creation - except if the whole host is in disabled state. Signed-off-by: Jason Robinson --- changelog.d/6020.bugfix | 1 + synapse/api/auth.py | 11 +++++++++-- tests/api/test_auth.py | 18 ++++++++++++++++++ 3 files changed, 28 insertions(+), 2 deletions(-) create mode 100644 changelog.d/6020.bugfix (limited to 'tests') diff --git a/changelog.d/6020.bugfix b/changelog.d/6020.bugfix new file mode 100644 index 0000000000..58a7deba9d --- /dev/null +++ b/changelog.d/6020.bugfix @@ -0,0 +1 @@ +Ensure support users can be registered even if MAU limit is reached. diff --git a/synapse/api/auth.py b/synapse/api/auth.py index ddc195bc32..9e445cd808 100644 --- a/synapse/api/auth.py +++ b/synapse/api/auth.py @@ -25,7 +25,7 @@ from twisted.internet import defer import synapse.logging.opentracing as opentracing import synapse.types from synapse import event_auth -from synapse.api.constants import EventTypes, JoinRules, Membership +from synapse.api.constants import EventTypes, JoinRules, Membership, UserTypes from synapse.api.errors import ( AuthError, Codes, @@ -709,7 +709,7 @@ class Auth(object): ) @defer.inlineCallbacks - def check_auth_blocking(self, user_id=None, threepid=None): + def check_auth_blocking(self, user_id=None, threepid=None, user_type=None): """Checks if the user should be rejected for some external reason, such as monthly active user limiting or global disable flag @@ -722,6 +722,9 @@ class Auth(object): with a MAU blocked server, normally they would be rejected but their threepid is on the reserved list. user_id and threepid should never be set at the same time. + + user_type(str|None): If present, is used to decide whether to check against + certain blocking reasons like MAU. """ # Never fail an auth check for the server notices users or support user @@ -759,6 +762,10 @@ class Auth(object): self.hs.config.mau_limits_reserved_threepids, threepid ): return + elif user_type == UserTypes.SUPPORT: + # If the user does not exist yet and is of type "support", + # allow registration. Support users are excluded from MAU checks. + return # Else if there is no room in the MAU bucket, bail current_mau = yield self.store.get_monthly_active_count() if current_mau >= self.hs.config.max_mau_value: diff --git a/tests/api/test_auth.py b/tests/api/test_auth.py index c0cb8ef296..6121efcfa9 100644 --- a/tests/api/test_auth.py +++ b/tests/api/test_auth.py @@ -21,6 +21,7 @@ from twisted.internet import defer import synapse.handlers.auth from synapse.api.auth import Auth +from synapse.api.constants import UserTypes from synapse.api.errors import ( AuthError, Codes, @@ -335,6 +336,23 @@ class AuthTestCase(unittest.TestCase): ) yield self.auth.check_auth_blocking() + @defer.inlineCallbacks + def test_blocking_mau__depending_on_user_type(self): + self.hs.config.max_mau_value = 50 + self.hs.config.limit_usage_by_mau = True + + self.store.get_monthly_active_count = Mock(return_value=defer.succeed(100)) + # Support users allowed + yield self.auth.check_auth_blocking(user_type=UserTypes.SUPPORT) + self.store.get_monthly_active_count = Mock(return_value=defer.succeed(100)) + # Bots not allowed + with self.assertRaises(ResourceLimitError): + yield self.auth.check_auth_blocking(user_type=UserTypes.BOT) + self.store.get_monthly_active_count = Mock(return_value=defer.succeed(100)) + # Real users not allowed + with self.assertRaises(ResourceLimitError): + yield self.auth.check_auth_blocking() + @defer.inlineCallbacks def test_reserved_threepid(self): self.hs.config.limit_usage_by_mau = True -- cgit 1.5.1 From b617864cd9f81109e818bc5ae95bee317d917b72 Mon Sep 17 00:00:00 2001 From: Amber Brown Date: Fri, 13 Sep 2019 02:29:55 +1000 Subject: Fix for structured logging tests stomping on logs (#6023) --- MANIFEST.in | 12 +++++---- changelog.d/6023.misc | 1 + mypy.ini | 54 ++++++++++++++++++++++++++++++++++++++++ synapse/config/logger.py | 33 ++++++++++++++++++------ synapse/logging/_structured.py | 8 +++--- synapse/logging/_terse_json.py | 8 +++--- synapse/logging/opentracing.py | 4 +-- synapse/metrics/__init__.py | 5 ++-- synapse/metrics/_exposition.py | 4 ++- synapse/python_dependencies.py | 7 +++--- tests/logging/test_structured.py | 25 ++++++++++++++++--- tests/logging/test_terse_json.py | 4 +-- tox.ini | 30 +++++++++++++++++----- 13 files changed, 154 insertions(+), 41 deletions(-) create mode 100644 changelog.d/6023.misc create mode 100644 mypy.ini (limited to 'tests') diff --git a/MANIFEST.in b/MANIFEST.in index 919cd8a1cd..9c2902b8d2 100644 --- a/MANIFEST.in +++ b/MANIFEST.in @@ -38,14 +38,16 @@ exclude sytest-blacklist include pyproject.toml recursive-include changelog.d * -prune .github -prune demo/etc -prune docker +prune .buildkite prune .circleci +prune .codecov.yml prune .coveragerc +prune .github prune debian -prune .codecov.yml -prune .buildkite +prune demo/etc +prune docker +prune mypy.ini +prune stubs exclude jenkins* recursive-exclude jenkins *.sh diff --git a/changelog.d/6023.misc b/changelog.d/6023.misc new file mode 100644 index 0000000000..d80410c22c --- /dev/null +++ b/changelog.d/6023.misc @@ -0,0 +1 @@ +Fix the structured logging tests stomping on the global log configuration for subsequent tests. diff --git a/mypy.ini b/mypy.ini new file mode 100644 index 0000000000..8788574ee3 --- /dev/null +++ b/mypy.ini @@ -0,0 +1,54 @@ +[mypy] +namespace_packages=True +plugins=mypy_zope:plugin +follow_imports=skip +mypy_path=stubs + +[mypy-synapse.config.homeserver] +# this is a mess because of the metaclass shenanigans +ignore_errors = True + +[mypy-zope] +ignore_missing_imports = True + +[mypy-constantly] +ignore_missing_imports = True + +[mypy-twisted.*] +ignore_missing_imports = True + +[mypy-treq.*] +ignore_missing_imports = True + +[mypy-hyperlink] +ignore_missing_imports = True + +[mypy-h11] +ignore_missing_imports = True + +[mypy-opentracing] +ignore_missing_imports = True + +[mypy-OpenSSL] +ignore_missing_imports = True + +[mypy-netaddr] +ignore_missing_imports = True + +[mypy-saml2.*] +ignore_missing_imports = True + +[mypy-unpaddedbase64] +ignore_missing_imports = True + +[mypy-canonicaljson] +ignore_missing_imports = True + +[mypy-jaeger_client] +ignore_missing_imports = True + +[mypy-jsonschema] +ignore_missing_imports = True + +[mypy-signedjson.*] +ignore_missing_imports = True diff --git a/synapse/config/logger.py b/synapse/config/logger.py index 2704c18720..767ecfdf09 100644 --- a/synapse/config/logger.py +++ b/synapse/config/logger.py @@ -21,7 +21,12 @@ from string import Template import yaml -from twisted.logger import STDLibLogObserver, globalLogBeginner +from twisted.logger import ( + ILogObserver, + LogBeginner, + STDLibLogObserver, + globalLogBeginner, +) import synapse from synapse.app import _base as appbase @@ -124,7 +129,7 @@ class LoggingConfig(Config): log_config_file.write(DEFAULT_LOG_CONFIG.substitute(log_file=log_file)) -def _setup_stdlib_logging(config, log_config): +def _setup_stdlib_logging(config, log_config, logBeginner: LogBeginner): """ Set up Python stdlib logging. """ @@ -165,12 +170,12 @@ def _setup_stdlib_logging(config, log_config): return observer(event) - globalLogBeginner.beginLoggingTo( - [_log], redirectStandardIO=not config.no_redirect_stdio - ) + logBeginner.beginLoggingTo([_log], redirectStandardIO=not config.no_redirect_stdio) if not config.no_redirect_stdio: print("Redirected stdout/stderr to logs") + return observer + def _reload_stdlib_logging(*args, log_config=None): logger = logging.getLogger("") @@ -181,7 +186,9 @@ def _reload_stdlib_logging(*args, log_config=None): logging.config.dictConfig(log_config) -def setup_logging(hs, config, use_worker_options=False): +def setup_logging( + hs, config, use_worker_options=False, logBeginner: LogBeginner = globalLogBeginner +) -> ILogObserver: """ Set up the logging subsystem. @@ -191,6 +198,12 @@ def setup_logging(hs, config, use_worker_options=False): use_worker_options (bool): True to use the 'worker_log_config' option instead of 'log_config'. + + logBeginner: The Twisted logBeginner to use. + + Returns: + The "root" Twisted Logger observer, suitable for sending logs to from a + Logger instance. """ log_config = config.worker_log_config if use_worker_options else config.log_config @@ -210,10 +223,12 @@ def setup_logging(hs, config, use_worker_options=False): log_config_body = read_config() if log_config_body and log_config_body.get("structured") is True: - setup_structured_logging(hs, config, log_config_body) + logger = setup_structured_logging( + hs, config, log_config_body, logBeginner=logBeginner + ) appbase.register_sighup(read_config, callback=reload_structured_logging) else: - _setup_stdlib_logging(config, log_config_body) + logger = _setup_stdlib_logging(config, log_config_body, logBeginner=logBeginner) appbase.register_sighup(read_config, callback=_reload_stdlib_logging) # make sure that the first thing we log is a thing we can grep backwards @@ -221,3 +236,5 @@ def setup_logging(hs, config, use_worker_options=False): logging.warn("***** STARTING SERVER *****") logging.warn("Server %s version %s", sys.argv[0], get_version_string(synapse)) logging.info("Server hostname: %s", config.server_name) + + return logger diff --git a/synapse/logging/_structured.py b/synapse/logging/_structured.py index 0367d6dfc4..3220e985a9 100644 --- a/synapse/logging/_structured.py +++ b/synapse/logging/_structured.py @@ -18,6 +18,7 @@ import os.path import sys import typing import warnings +from typing import List import attr from constantly import NamedConstant, Names, ValueConstant, Values @@ -33,7 +34,6 @@ from twisted.logger import ( LogLevelFilterPredicate, LogPublisher, eventAsText, - globalLogBeginner, jsonFileLogObserver, ) @@ -134,7 +134,7 @@ class PythonStdlibToTwistedLogger(logging.Handler): ) -def SynapseFileLogObserver(outFile: typing.io.TextIO) -> FileLogObserver: +def SynapseFileLogObserver(outFile: typing.IO[str]) -> FileLogObserver: """ A log observer that formats events like the traditional log formatter and sends them to `outFile`. @@ -265,7 +265,7 @@ def setup_structured_logging( hs, config, log_config: dict, - logBeginner: LogBeginner = globalLogBeginner, + logBeginner: LogBeginner, redirect_stdlib_logging: bool = True, ) -> LogPublisher: """ @@ -286,7 +286,7 @@ def setup_structured_logging( if "drains" not in log_config: raise ConfigError("The logging configuration requires a list of drains.") - observers = [] + observers = [] # type: List[ILogObserver] for observer in parse_drain_configs(log_config["drains"]): # Pipe drains diff --git a/synapse/logging/_terse_json.py b/synapse/logging/_terse_json.py index 7f1e8f23fe..0ebbde06f2 100644 --- a/synapse/logging/_terse_json.py +++ b/synapse/logging/_terse_json.py @@ -21,10 +21,11 @@ import sys from collections import deque from ipaddress import IPv4Address, IPv6Address, ip_address from math import floor -from typing.io import TextIO +from typing import IO import attr from simplejson import dumps +from zope.interface import implementer from twisted.application.internet import ClientService from twisted.internet.endpoints import ( @@ -33,7 +34,7 @@ from twisted.internet.endpoints import ( TCP6ClientEndpoint, ) from twisted.internet.protocol import Factory, Protocol -from twisted.logger import FileLogObserver, Logger +from twisted.logger import FileLogObserver, ILogObserver, Logger from twisted.python.failure import Failure @@ -129,7 +130,7 @@ def flatten_event(event: dict, metadata: dict, include_time: bool = False): return new_event -def TerseJSONToConsoleLogObserver(outFile: TextIO, metadata: dict) -> FileLogObserver: +def TerseJSONToConsoleLogObserver(outFile: IO[str], metadata: dict) -> FileLogObserver: """ A log observer that formats events to a flattened JSON representation. @@ -146,6 +147,7 @@ def TerseJSONToConsoleLogObserver(outFile: TextIO, metadata: dict) -> FileLogObs @attr.s +@implementer(ILogObserver) class TerseJSONToTCPLogObserver(object): """ An IObserver that writes JSON logs to a TCP target. diff --git a/synapse/logging/opentracing.py b/synapse/logging/opentracing.py index 7246253018..308a27213b 100644 --- a/synapse/logging/opentracing.py +++ b/synapse/logging/opentracing.py @@ -223,8 +223,8 @@ try: from jaeger_client import Config as JaegerConfig from synapse.logging.scopecontextmanager import LogContextScopeManager except ImportError: - JaegerConfig = None - LogContextScopeManager = None + JaegerConfig = None # type: ignore + LogContextScopeManager = None # type: ignore logger = logging.getLogger(__name__) diff --git a/synapse/metrics/__init__.py b/synapse/metrics/__init__.py index b5c9595cb9..bec3b13397 100644 --- a/synapse/metrics/__init__.py +++ b/synapse/metrics/__init__.py @@ -20,6 +20,7 @@ import os import platform import threading import time +from typing import Dict, Union import six @@ -42,9 +43,7 @@ logger = logging.getLogger(__name__) METRICS_PREFIX = "/_synapse/metrics" running_on_pypy = platform.python_implementation() == "PyPy" -all_metrics = [] -all_collectors = [] -all_gauges = {} +all_gauges = {} # type: Dict[str, Union[LaterGauge, InFlightGauge, BucketCollector]] HAVE_PROC_SELF_STAT = os.path.exists("/proc/self/stat") diff --git a/synapse/metrics/_exposition.py b/synapse/metrics/_exposition.py index 1933ecd3e3..74d9c3ecd3 100644 --- a/synapse/metrics/_exposition.py +++ b/synapse/metrics/_exposition.py @@ -36,7 +36,9 @@ from twisted.web.resource import Resource try: from prometheus_client.samples import Sample except ImportError: - Sample = namedtuple("Sample", ["name", "labels", "value", "timestamp", "exemplar"]) + Sample = namedtuple( + "Sample", ["name", "labels", "value", "timestamp", "exemplar"] + ) # type: ignore CONTENT_TYPE_LATEST = str("text/plain; version=0.0.4; charset=utf-8") diff --git a/synapse/python_dependencies.py b/synapse/python_dependencies.py index 07345e916a..0bd563edc7 100644 --- a/synapse/python_dependencies.py +++ b/synapse/python_dependencies.py @@ -15,6 +15,7 @@ # limitations under the License. import logging +from typing import Set from pkg_resources import ( DistributionNotFound, @@ -97,7 +98,7 @@ CONDITIONAL_REQUIREMENTS = { "jwt": ["pyjwt>=1.6.4"], } -ALL_OPTIONAL_REQUIREMENTS = set() +ALL_OPTIONAL_REQUIREMENTS = set() # type: Set[str] for name, optional_deps in CONDITIONAL_REQUIREMENTS.items(): # Exclude systemd as it's a system-based requirement. @@ -174,8 +175,8 @@ def check_requirements(for_feature=None): pass if deps_needed: - for e in errors: - logging.error(e) + for err in errors: + logging.error(err) raise DependencyException(deps_needed) diff --git a/tests/logging/test_structured.py b/tests/logging/test_structured.py index a786de0233..451d05c0f0 100644 --- a/tests/logging/test_structured.py +++ b/tests/logging/test_structured.py @@ -13,6 +13,7 @@ # See the License for the specific language governing permissions and # limitations under the License. +import logging import os import os.path import shutil @@ -33,7 +34,20 @@ class FakeBeginner(object): self.observers = observers -class StructuredLoggingTestCase(HomeserverTestCase): +class StructuredLoggingTestBase(object): + """ + Test base that registers a cleanup handler to reset the stdlib log handler + to 'unset'. + """ + + def prepare(self, reactor, clock, hs): + def _cleanup(): + logging.getLogger("synapse").setLevel(logging.NOTSET) + + self.addCleanup(_cleanup) + + +class StructuredLoggingTestCase(StructuredLoggingTestBase, HomeserverTestCase): """ Tests for Synapse's structured logging support. """ @@ -139,7 +153,9 @@ class StructuredLoggingTestCase(HomeserverTestCase): self.assertEqual(logs[0]["request"], "somereq") -class StructuredLoggingConfigurationFileTestCase(HomeserverTestCase): +class StructuredLoggingConfigurationFileTestCase( + StructuredLoggingTestBase, HomeserverTestCase +): def make_homeserver(self, reactor, clock): tempdir = self.mktemp() @@ -179,10 +195,11 @@ class StructuredLoggingConfigurationFileTestCase(HomeserverTestCase): """ When a structured logging config is given, Synapse will use it. """ - setup_logging(self.hs, self.hs.config) + beginner = FakeBeginner() + publisher = setup_logging(self.hs, self.hs.config, logBeginner=beginner) # Make a logger and send an event - logger = Logger(namespace="tests.logging.test_structured") + logger = Logger(namespace="tests.logging.test_structured", observer=publisher) with LoggingContext("testcontext", request="somereq"): logger.info("Hello there, {name}!", name="steve") diff --git a/tests/logging/test_terse_json.py b/tests/logging/test_terse_json.py index 514282591d..4cf81f7128 100644 --- a/tests/logging/test_terse_json.py +++ b/tests/logging/test_terse_json.py @@ -23,10 +23,10 @@ from synapse.logging._structured import setup_structured_logging from tests.server import connect_client from tests.unittest import HomeserverTestCase -from .test_structured import FakeBeginner +from .test_structured import FakeBeginner, StructuredLoggingTestBase -class TerseJSONTCPTestCase(HomeserverTestCase): +class TerseJSONTCPTestCase(StructuredLoggingTestBase, HomeserverTestCase): def test_log_output(self): """ The Terse JSON outputter delivers simplified structured logs over TCP. diff --git a/tox.ini b/tox.ini index 7cb40847b5..1bce10a4ce 100644 --- a/tox.ini +++ b/tox.ini @@ -2,6 +2,7 @@ envlist = packaging, py35, py36, py37, check_codestyle, check_isort [base] +basepython = python3.7 deps = mock python-subunit @@ -137,18 +138,35 @@ commands = {toxinidir}/scripts-dev/generate_sample_config --check skip_install = True deps = coverage -whitelist_externals = - bash commands= coverage combine coverage report +[testenv:cov-erase] +skip_install = True +deps = + coverage +commands= + coverage erase + +[testenv:cov-html] +skip_install = True +deps = + coverage +commands= + coverage html + [testenv:mypy] -basepython = python3.5 +basepython = python3.7 +skip_install = True deps = {[base]deps} mypy + mypy-zope + typeshed +env = + MYPYPATH = stubs/ extras = all -commands = mypy --ignore-missing-imports \ - synapse/logging/_structured.py \ - synapse/logging/_terse_json.py +commands = mypy --show-traceback \ + synapse/logging/ \ + synapse/config/ -- cgit 1.5.1 From 850dcfd2d3a1d689042fb38c8a16b652244068c2 Mon Sep 17 00:00:00 2001 From: Amber Brown Date: Sat, 14 Sep 2019 04:58:38 +1000 Subject: Fix well-known lookups with the federation certificate whitelist (#5997) --- changelog.d/5996.bugfix | 1 + synapse/config/tls.py | 9 ++++- synapse/crypto/context_factory.py | 26 +++++++------- synapse/http/federation/matrix_federation_agent.py | 2 +- tests/config/test_tls.py | 40 ++++++++++++++++++++++ 5 files changed, 63 insertions(+), 15 deletions(-) create mode 100644 changelog.d/5996.bugfix (limited to 'tests') diff --git a/changelog.d/5996.bugfix b/changelog.d/5996.bugfix new file mode 100644 index 0000000000..05e31faaa2 --- /dev/null +++ b/changelog.d/5996.bugfix @@ -0,0 +1 @@ +federation_certificate_verification_whitelist now will not cause TypeErrors to be raised (a regression in 1.3). Additionally, it now supports internationalised domain names in their non-canonical representation. diff --git a/synapse/config/tls.py b/synapse/config/tls.py index c0148aa95c..fc47ba3e9a 100644 --- a/synapse/config/tls.py +++ b/synapse/config/tls.py @@ -110,8 +110,15 @@ class TlsConfig(Config): # Support globs (*) in whitelist values self.federation_certificate_verification_whitelist = [] for entry in fed_whitelist_entries: + try: + entry_regex = glob_to_regex(entry.encode("ascii").decode("ascii")) + except UnicodeEncodeError: + raise ConfigError( + "IDNA domain names are not allowed in the " + "federation_certificate_verification_whitelist: %s" % (entry,) + ) + # Convert globs to regex - entry_regex = glob_to_regex(entry) self.federation_certificate_verification_whitelist.append(entry_regex) # List of custom certificate authorities for federation traffic validation diff --git a/synapse/crypto/context_factory.py b/synapse/crypto/context_factory.py index 06e63a96b5..e93f0b3705 100644 --- a/synapse/crypto/context_factory.py +++ b/synapse/crypto/context_factory.py @@ -15,7 +15,6 @@ import logging -import idna from service_identity import VerificationError from service_identity.pyopenssl import verify_hostname, verify_ip_address from zope.interface import implementer @@ -114,14 +113,20 @@ class ClientTLSOptionsFactory(object): self._no_verify_ssl_context = self._no_verify_ssl.getContext() self._no_verify_ssl_context.set_info_callback(self._context_info_cb) - def get_options(self, host): + def get_options(self, host: bytes): + + # IPolicyForHTTPS.get_options takes bytes, but we want to compare + # against the str whitelist. The hostnames in the whitelist are already + # IDNA-encoded like the hosts will be here. + ascii_host = host.decode("ascii") + # Check if certificate verification has been enabled should_verify = self._config.federation_verify_certificates # Check if we've disabled certificate verification for this host if should_verify: for regex in self._config.federation_certificate_verification_whitelist: - if regex.match(host): + if regex.match(ascii_host): should_verify = False break @@ -162,7 +167,7 @@ class SSLClientConnectionCreator(object): Replaces twisted.internet.ssl.ClientTLSOptions """ - def __init__(self, hostname, ctx, verify_certs): + def __init__(self, hostname: bytes, ctx, verify_certs: bool): self._ctx = ctx self._verifier = ConnectionVerifier(hostname, verify_certs) @@ -190,21 +195,16 @@ class ConnectionVerifier(object): # This code is based on twisted.internet.ssl.ClientTLSOptions. - def __init__(self, hostname, verify_certs): + def __init__(self, hostname: bytes, verify_certs): self._verify_certs = verify_certs - if isIPAddress(hostname) or isIPv6Address(hostname): - self._hostnameBytes = hostname.encode("ascii") + _decoded = hostname.decode("ascii") + if isIPAddress(_decoded) or isIPv6Address(_decoded): self._is_ip_address = True else: - # twisted's ClientTLSOptions falls back to the stdlib impl here if - # idna is not installed, but points out that lacks support for - # IDNA2008 (http://bugs.python.org/issue17305). - # - # We can rely on having idna. - self._hostnameBytes = idna.encode(hostname) self._is_ip_address = False + self._hostnameBytes = hostname self._hostnameASCII = self._hostnameBytes.decode("ascii") def verify_context_info_cb(self, ssl_connection, where): diff --git a/synapse/http/federation/matrix_federation_agent.py b/synapse/http/federation/matrix_federation_agent.py index feae7de5be..647d26dc56 100644 --- a/synapse/http/federation/matrix_federation_agent.py +++ b/synapse/http/federation/matrix_federation_agent.py @@ -217,7 +217,7 @@ class MatrixHostnameEndpoint(object): self._tls_options = None else: self._tls_options = tls_client_options_factory.get_options( - self._parsed_uri.host.decode("ascii") + self._parsed_uri.host ) self._srv_resolver = srv_resolver diff --git a/tests/config/test_tls.py b/tests/config/test_tls.py index 8e0c4b9533..b02780772a 100644 --- a/tests/config/test_tls.py +++ b/tests/config/test_tls.py @@ -16,6 +16,7 @@ import os +import idna import yaml from OpenSSL import SSL @@ -235,3 +236,42 @@ s4niecZKPBizL6aucT59CsunNmmb5Glq8rlAcU+1ZTZZzGYqVYhF6axB9Qg= ) self.assertTrue(conf.acme_enabled) + + def test_whitelist_idna_failure(self): + """ + The federation certificate whitelist will not allow IDNA domain names. + """ + config = { + "federation_certificate_verification_whitelist": [ + "example.com", + "*.ドメイン.テスト", + ] + } + t = TestConfig() + e = self.assertRaises( + ConfigError, t.read_config, config, config_dir_path="", data_dir_path="" + ) + self.assertIn("IDNA domain names", str(e)) + + def test_whitelist_idna_result(self): + """ + The federation certificate whitelist will match on IDNA encoded names. + """ + config = { + "federation_certificate_verification_whitelist": [ + "example.com", + "*.xn--eckwd4c7c.xn--zckzah", + ] + } + t = TestConfig() + t.read_config(config, config_dir_path="", data_dir_path="") + + cf = ClientTLSOptionsFactory(t) + + # Not in the whitelist + opts = cf.get_options(b"notexample.com") + self.assertTrue(opts._verifier._verify_certs) + + # Caught by the wildcard + opts = cf.get_options(idna.encode("テスト.ドメイン.テスト")) + self.assertFalse(opts._verifier._verify_certs) -- cgit 1.5.1 From 1e19ce00bff8d67168d39201cdf9424f7b2f22f6 Mon Sep 17 00:00:00 2001 From: Richard van der Hoff <1389908+richvdh@users.noreply.github.com> Date: Tue, 17 Sep 2019 11:41:54 +0100 Subject: Add 'failure_ts' column to 'destinations' table (#6016) Track the time that a server started failing at, for general analysis purposes. --- changelog.d/6016.misc | 1 + .../schema/delta/56/destinations_failure_ts.sql | 25 ++++ synapse/storage/transactions.py | 23 ++-- synapse/util/retryutils.py | 16 ++- tests/handlers/test_typing.py | 7 +- tests/storage/test_transactions.py | 8 +- tests/util/test_retryutils.py | 127 +++++++++++++++++++++ 7 files changed, 195 insertions(+), 12 deletions(-) create mode 100644 changelog.d/6016.misc create mode 100644 synapse/storage/schema/delta/56/destinations_failure_ts.sql create mode 100644 tests/util/test_retryutils.py (limited to 'tests') diff --git a/changelog.d/6016.misc b/changelog.d/6016.misc new file mode 100644 index 0000000000..91cf164714 --- /dev/null +++ b/changelog.d/6016.misc @@ -0,0 +1 @@ +Add a 'failure_ts' column to the 'destinations' database table. diff --git a/synapse/storage/schema/delta/56/destinations_failure_ts.sql b/synapse/storage/schema/delta/56/destinations_failure_ts.sql new file mode 100644 index 0000000000..f00889290b --- /dev/null +++ b/synapse/storage/schema/delta/56/destinations_failure_ts.sql @@ -0,0 +1,25 @@ +/* Copyright 2019 The Matrix.org Foundation C.I.C + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +/* + * Record the timestamp when a given server started failing + */ +ALTER TABLE destinations ADD failure_ts BIGINT; + +/* as a rough approximation, we assume that the server started failing at + * retry_interval before the last retry + */ +UPDATE destinations SET failure_ts = retry_last_ts - retry_interval + WHERE retry_last_ts > 0; diff --git a/synapse/storage/transactions.py b/synapse/storage/transactions.py index d81ace0ece..289c117396 100644 --- a/synapse/storage/transactions.py +++ b/synapse/storage/transactions.py @@ -165,7 +165,7 @@ class TransactionStore(SQLBaseStore): txn, table="destinations", keyvalues={"destination": destination}, - retcols=("destination", "retry_last_ts", "retry_interval"), + retcols=("destination", "failure_ts", "retry_last_ts", "retry_interval"), allow_none=True, ) @@ -174,12 +174,15 @@ class TransactionStore(SQLBaseStore): else: return None - def set_destination_retry_timings(self, destination, retry_last_ts, retry_interval): + def set_destination_retry_timings( + self, destination, failure_ts, retry_last_ts, retry_interval + ): """Sets the current retry timings for a given destination. Both timings should be zero if retrying is no longer occuring. Args: destination (str) + failure_ts (int|None) - when the server started failing (ms since epoch) retry_last_ts (int) - time of last retry attempt in unix epoch ms retry_interval (int) - how long until next retry in ms """ @@ -189,12 +192,13 @@ class TransactionStore(SQLBaseStore): "set_destination_retry_timings", self._set_destination_retry_timings, destination, + failure_ts, retry_last_ts, retry_interval, ) def _set_destination_retry_timings( - self, txn, destination, retry_last_ts, retry_interval + self, txn, destination, failure_ts, retry_last_ts, retry_interval ): if self.database_engine.can_native_upsert: @@ -202,9 +206,12 @@ class TransactionStore(SQLBaseStore): # resetting it) or greater than the existing retry interval. sql = """ - INSERT INTO destinations (destination, retry_last_ts, retry_interval) - VALUES (?, ?, ?) + INSERT INTO destinations ( + destination, failure_ts, retry_last_ts, retry_interval + ) + VALUES (?, ?, ?, ?) ON CONFLICT (destination) DO UPDATE SET + failure_ts = EXCLUDED.failure_ts, retry_last_ts = EXCLUDED.retry_last_ts, retry_interval = EXCLUDED.retry_interval WHERE @@ -212,7 +219,7 @@ class TransactionStore(SQLBaseStore): OR destinations.retry_interval < EXCLUDED.retry_interval """ - txn.execute(sql, (destination, retry_last_ts, retry_interval)) + txn.execute(sql, (destination, failure_ts, retry_last_ts, retry_interval)) return @@ -225,7 +232,7 @@ class TransactionStore(SQLBaseStore): txn, table="destinations", keyvalues={"destination": destination}, - retcols=("retry_last_ts", "retry_interval"), + retcols=("failure_ts", "retry_last_ts", "retry_interval"), allow_none=True, ) @@ -235,6 +242,7 @@ class TransactionStore(SQLBaseStore): table="destinations", values={ "destination": destination, + "failure_ts": failure_ts, "retry_last_ts": retry_last_ts, "retry_interval": retry_interval, }, @@ -245,6 +253,7 @@ class TransactionStore(SQLBaseStore): "destinations", keyvalues={"destination": destination}, updatevalues={ + "failure_ts": failure_ts, "retry_last_ts": retry_last_ts, "retry_interval": retry_interval, }, diff --git a/synapse/util/retryutils.py b/synapse/util/retryutils.py index b740913b58..a5f2fbef5c 100644 --- a/synapse/util/retryutils.py +++ b/synapse/util/retryutils.py @@ -80,11 +80,13 @@ def get_retry_limiter(destination, clock, store, ignore_backoff=False, **kwargs) # We aren't ready to retry that destination. raise """ + failure_ts = None retry_last_ts, retry_interval = (0, 0) retry_timings = yield store.get_destination_retry_timings(destination) if retry_timings: + failure_ts = retry_timings["failure_ts"] retry_last_ts, retry_interval = ( retry_timings["retry_last_ts"], retry_timings["retry_interval"], @@ -108,6 +110,7 @@ def get_retry_limiter(destination, clock, store, ignore_backoff=False, **kwargs) destination, clock, store, + failure_ts, retry_interval, backoff_on_failure=backoff_on_failure, **kwargs @@ -120,6 +123,7 @@ class RetryDestinationLimiter(object): destination, clock, store, + failure_ts, retry_interval, backoff_on_404=False, backoff_on_failure=True, @@ -133,6 +137,8 @@ class RetryDestinationLimiter(object): destination (str) clock (Clock) store (DataStore) + failure_ts (int|None): when this destination started failing (in ms since + the epoch), or zero if the last request was successful retry_interval (int): The next retry interval taken from the database in milliseconds, or zero if the last request was successful. @@ -145,6 +151,7 @@ class RetryDestinationLimiter(object): self.store = store self.destination = destination + self.failure_ts = failure_ts self.retry_interval = retry_interval self.backoff_on_404 = backoff_on_404 self.backoff_on_failure = backoff_on_failure @@ -186,6 +193,7 @@ class RetryDestinationLimiter(object): logger.debug( "Connection to %s was successful; clearing backoff", self.destination ) + self.failure_ts = None retry_last_ts = 0 self.retry_interval = 0 elif not self.backoff_on_failure: @@ -211,11 +219,17 @@ class RetryDestinationLimiter(object): ) retry_last_ts = int(self.clock.time_msec()) + if self.failure_ts is None: + self.failure_ts = retry_last_ts + @defer.inlineCallbacks def store_retry_timings(): try: yield self.store.set_destination_retry_timings( - self.destination, retry_last_ts, self.retry_interval + self.destination, + self.failure_ts, + retry_last_ts, + self.retry_interval, ) except Exception: logger.exception("Failed to store destination_retry_timings") diff --git a/tests/handlers/test_typing.py b/tests/handlers/test_typing.py index 5d5e324df2..1f2ef5d01f 100644 --- a/tests/handlers/test_typing.py +++ b/tests/handlers/test_typing.py @@ -99,7 +99,12 @@ class TypingNotificationsTestCase(unittest.HomeserverTestCase): self.event_source = hs.get_event_sources().sources["typing"] self.datastore = hs.get_datastore() - retry_timings_res = {"destination": "", "retry_last_ts": 0, "retry_interval": 0} + retry_timings_res = { + "destination": "", + "retry_last_ts": 0, + "retry_interval": 0, + "failure_ts": None, + } self.datastore.get_destination_retry_timings.return_value = defer.succeed( retry_timings_res ) diff --git a/tests/storage/test_transactions.py b/tests/storage/test_transactions.py index 14169afa96..a771d5af29 100644 --- a/tests/storage/test_transactions.py +++ b/tests/storage/test_transactions.py @@ -29,17 +29,19 @@ class TransactionStoreTestCase(HomeserverTestCase): r = self.get_success(d) self.assertIsNone(r) - d = self.store.set_destination_retry_timings("example.com", 50, 100) + d = self.store.set_destination_retry_timings("example.com", 1000, 50, 100) self.get_success(d) d = self.store.get_destination_retry_timings("example.com") r = self.get_success(d) - self.assert_dict({"retry_last_ts": 50, "retry_interval": 100}, r) + self.assert_dict( + {"retry_last_ts": 50, "retry_interval": 100, "failure_ts": 1000}, r + ) def test_initial_set_transactions(self): """Tests that we can successfully set the destination retries (there was a bug around invalidating the cache that broke this) """ - d = self.store.set_destination_retry_timings("example.com", 50, 100) + d = self.store.set_destination_retry_timings("example.com", 1000, 50, 100) self.get_success(d) diff --git a/tests/util/test_retryutils.py b/tests/util/test_retryutils.py new file mode 100644 index 0000000000..9e348694ad --- /dev/null +++ b/tests/util/test_retryutils.py @@ -0,0 +1,127 @@ +# -*- coding: utf-8 -*- +# Copyright 2019 The Matrix.org Foundation C.I.C. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +from synapse.util.retryutils import ( + MIN_RETRY_INTERVAL, + RETRY_MULTIPLIER, + NotRetryingDestination, + get_retry_limiter, +) + +from tests.unittest import HomeserverTestCase + + +class RetryLimiterTestCase(HomeserverTestCase): + def test_new_destination(self): + """A happy-path case with a new destination and a successful operation""" + store = self.hs.get_datastore() + d = get_retry_limiter("test_dest", self.clock, store) + self.pump() + limiter = self.successResultOf(d) + + # advance the clock a bit before making the request + self.pump(1) + + with limiter: + pass + + d = store.get_destination_retry_timings("test_dest") + self.pump() + new_timings = self.successResultOf(d) + self.assertIsNone(new_timings) + + def test_limiter(self): + """General test case which walks through the process of a failing request""" + store = self.hs.get_datastore() + + d = get_retry_limiter("test_dest", self.clock, store) + self.pump() + limiter = self.successResultOf(d) + + self.pump(1) + try: + with limiter: + self.pump(1) + failure_ts = self.clock.time_msec() + raise AssertionError("argh") + except AssertionError: + pass + + # wait for the update to land + self.pump() + + d = store.get_destination_retry_timings("test_dest") + self.pump() + new_timings = self.successResultOf(d) + self.assertEqual(new_timings["failure_ts"], failure_ts) + self.assertEqual(new_timings["retry_last_ts"], failure_ts) + self.assertEqual(new_timings["retry_interval"], MIN_RETRY_INTERVAL) + + # now if we try again we should get a failure + d = get_retry_limiter("test_dest", self.clock, store) + self.pump() + self.failureResultOf(d, NotRetryingDestination) + + # + # advance the clock and try again + # + + self.pump(MIN_RETRY_INTERVAL) + d = get_retry_limiter("test_dest", self.clock, store) + self.pump() + limiter = self.successResultOf(d) + + self.pump(1) + try: + with limiter: + self.pump(1) + retry_ts = self.clock.time_msec() + raise AssertionError("argh") + except AssertionError: + pass + + # wait for the update to land + self.pump() + + d = store.get_destination_retry_timings("test_dest") + self.pump() + new_timings = self.successResultOf(d) + self.assertEqual(new_timings["failure_ts"], failure_ts) + self.assertEqual(new_timings["retry_last_ts"], retry_ts) + self.assertGreaterEqual( + new_timings["retry_interval"], MIN_RETRY_INTERVAL * RETRY_MULTIPLIER * 0.5 + ) + self.assertLessEqual( + new_timings["retry_interval"], MIN_RETRY_INTERVAL * RETRY_MULTIPLIER * 2.0 + ) + + # + # one more go, with success + # + self.pump(MIN_RETRY_INTERVAL * RETRY_MULTIPLIER * 2.0) + d = get_retry_limiter("test_dest", self.clock, store) + self.pump() + limiter = self.successResultOf(d) + + self.pump(1) + with limiter: + self.pump(1) + + # wait for the update to land + self.pump() + + d = store.get_destination_retry_timings("test_dest") + self.pump() + new_timings = self.successResultOf(d) + self.assertIsNone(new_timings) -- cgit 1.5.1