diff options
Diffstat (limited to 'synapse')
| -rw-r--r-- | synapse/app/_base.py | 17 | ||||
| -rwxr-xr-x | synapse/app/homeserver.py | 5 | ||||
| -rw-r--r-- | synapse/config/_base.py | 9 | ||||
| -rw-r--r-- | synapse/config/homeserver.py | 2 | ||||
| -rw-r--r-- | synapse/config/server.py | 185 | ||||
| -rw-r--r-- | synapse/config/tls.py | 123 | ||||
| -rw-r--r-- | synapse/crypto/context_factory.py | 4 | ||||
| -rw-r--r-- | synapse/handlers/e2e_room_keys.py | 74 | ||||
| -rw-r--r-- | synapse/http/matrixfederationclient.py | 4 | ||||
| -rw-r--r-- | synapse/rest/client/v2_alpha/room_keys.py | 34 | ||||
| -rw-r--r-- | synapse/server.py | 6 | ||||
| -rw-r--r-- | synapse/storage/client_ips.py | 87 | ||||
| -rw-r--r-- | synapse/storage/e2e_room_keys.py | 21 | ||||
| -rw-r--r-- | synapse/storage/schema/delta/53/user_ips_index.sql | 10 |
14 files changed, 436 insertions, 145 deletions
diff --git a/synapse/app/_base.py b/synapse/app/_base.py index 62c633146f..5b0ca312e2 100644 --- a/synapse/app/_base.py +++ b/synapse/app/_base.py @@ -213,16 +213,17 @@ def refresh_certificate(hs): Refresh the TLS certificates that Synapse is using by re-reading them from disk and updating the TLS context factories to use them. """ - logging.info("Loading certificate from disk...") - hs.config.read_certificate_from_disk() + + if not hs.config.has_tls_listener(): + # attempt to reload the certs for the good of the tls_fingerprints + hs.config.read_certificate_from_disk(require_cert_and_key=False) + return + + hs.config.read_certificate_from_disk(require_cert_and_key=True) hs.tls_server_context_factory = context_factory.ServerContextFactory(hs.config) - hs.tls_client_options_factory = context_factory.ClientTLSOptionsFactory( - hs.config - ) - logging.info("Certificate loaded.") if hs._listening_services: - logging.info("Updating context factories...") + logger.info("Updating context factories...") for i in hs._listening_services: # When you listenSSL, it doesn't make an SSL port but a TCP one with # a TLS wrapping factory around the factory you actually want to get @@ -237,7 +238,7 @@ def refresh_certificate(hs): False, i.factory.wrappedFactory ) - logging.info("Context factories updated.") + logger.info("Context factories updated.") def start(hs, listeners=None): diff --git a/synapse/app/homeserver.py b/synapse/app/homeserver.py index b4476bf16e..dbd98d394f 100755 --- a/synapse/app/homeserver.py +++ b/synapse/app/homeserver.py @@ -90,11 +90,6 @@ class SynapseHomeServer(HomeServer): tls = listener_config.get("tls", False) site_tag = listener_config.get("tag", port) - if tls and config.no_tls: - raise ConfigError( - "Listener on port %i has TLS enabled, but no_tls is set" % (port,), - ) - resources = {} for res in listener_config["resources"]: for name in res["names"]: diff --git a/synapse/config/_base.py b/synapse/config/_base.py index 5858fb92b4..5aec43b702 100644 --- a/synapse/config/_base.py +++ b/synapse/config/_base.py @@ -257,7 +257,7 @@ class Config(object): "--keys-directory", metavar="DIRECTORY", help="Used with 'generate-*' options to specify where files such as" - " certs and signing keys should be stored in, unless explicitly" + " signing keys should be stored, unless explicitly" " specified in the config.", ) config_parser.add_argument( @@ -313,16 +313,11 @@ class Config(object): print( ( "A config file has been generated in %r for server name" - " %r with corresponding SSL keys and self-signed" - " certificates. Please review this file and customise it" + " %r. Please review this file and customise it" " to your needs." ) % (config_path, server_name) ) - print( - "If this server name is incorrect, you will need to" - " regenerate the SSL certificates" - ) return else: print( diff --git a/synapse/config/homeserver.py b/synapse/config/homeserver.py index 5aad062c36..727fdc54d8 100644 --- a/synapse/config/homeserver.py +++ b/synapse/config/homeserver.py @@ -42,7 +42,7 @@ from .voip import VoipConfig from .workers import WorkerConfig -class HomeServerConfig(TlsConfig, ServerConfig, DatabaseConfig, LoggingConfig, +class HomeServerConfig(ServerConfig, TlsConfig, DatabaseConfig, LoggingConfig, RatelimitConfig, ContentRepositoryConfig, CaptchaConfig, VoipConfig, RegistrationConfig, MetricsConfig, ApiConfig, AppServiceConfig, KeyConfig, SAML2Config, CasConfig, diff --git a/synapse/config/server.py b/synapse/config/server.py index f0a60cc712..c5c3aac8ed 100644 --- a/synapse/config/server.py +++ b/synapse/config/server.py @@ -24,6 +24,14 @@ from ._base import Config, ConfigError logger = logging.Logger(__name__) +# by default, we attempt to listen on both '::' *and* '0.0.0.0' because some OSes +# (Windows, macOS, other BSD/Linux where net.ipv6.bindv6only is set) will only listen +# on IPv6 when '::' is set. +# +# We later check for errors when binding to 0.0.0.0 and ignore them if :: is also in +# in the list. +DEFAULT_BIND_ADDRESSES = ['::', '0.0.0.0'] + class ServerConfig(Config): @@ -118,16 +126,34 @@ class ServerConfig(Config): self.public_baseurl += '/' self.start_pushers = config.get("start_pushers", True) - self.listeners = config.get("listeners", []) + self.listeners = [] + for listener in config.get("listeners", []): + if not isinstance(listener.get("port", None), int): + raise ConfigError( + "Listener configuration is lacking a valid 'port' option" + ) + + if listener.setdefault("tls", False): + # no_tls is not really supported any more, but let's grandfather it in + # here. + if config.get("no_tls", False): + logger.info( + "Ignoring TLS-enabled listener on port %i due to no_tls" + ) + continue - for listener in self.listeners: bind_address = listener.pop("bind_address", None) bind_addresses = listener.setdefault("bind_addresses", []) + # if bind_address was specified, add it to the list of addresses if bind_address: bind_addresses.append(bind_address) - elif not bind_addresses: - bind_addresses.append('') + + # if we still have an empty list of addresses, use the default list + if not bind_addresses: + bind_addresses.extend(DEFAULT_BIND_ADDRESSES) + + self.listeners.append(listener) if not self.web_client_location: _warn_if_webclient_configured(self.listeners) @@ -136,6 +162,9 @@ class ServerConfig(Config): bind_port = config.get("bind_port") if bind_port: + if config.get("no_tls", False): + raise ConfigError("no_tls is incompatible with bind_port") + self.listeners = [] bind_host = config.get("bind_host", "") gzip_responses = config.get("gzip_responses", True) @@ -182,6 +211,7 @@ class ServerConfig(Config): "port": manhole, "bind_addresses": ["127.0.0.1"], "type": "manhole", + "tls": False, }) metrics_port = config.get("metrics_port") @@ -207,6 +237,9 @@ class ServerConfig(Config): _check_resource_config(self.listeners) + def has_tls_listener(self): + return any(l["tls"] for l in self.listeners) + def default_config(self, server_name, data_dir_path, **kwargs): _, bind_port = parse_and_validate_server_name(server_name) if bind_port is not None: @@ -295,84 +328,106 @@ class ServerConfig(Config): # List of ports that Synapse should listen on, their purpose and their # configuration. + # + # Options for each listener include: + # + # port: the TCP port to bind to + # + # bind_addresses: a list of local addresses to listen on. The default is + # 'all local interfaces'. + # + # type: the type of listener. Normally 'http', but other valid options are: + # 'manhole' (see docs/manhole.md), + # 'metrics' (see docs/metrics-howto.rst), + # 'replication' (see docs/workers.rst). + # + # tls: set to true to enable TLS for this listener. Will use the TLS + # key/cert specified in tls_private_key_path / tls_certificate_path. + # + # x_forwarded: Only valid for an 'http' listener. Set to true to use the + # X-Forwarded-For header as the client IP. Useful when Synapse is + # behind a reverse-proxy. + # + # resources: Only valid for an 'http' listener. A list of resources to host + # on this port. Options for each resource are: + # + # names: a list of names of HTTP resources. See below for a list of + # valid resource names. + # + # compress: set to true to enable HTTP comression for this resource. + # + # additional_resources: Only valid for an 'http' listener. A map of + # additional endpoints which should be loaded via dynamic modules. + # + # Valid resource names are: + # + # client: the client-server API (/_matrix/client). Also implies 'media' and + # 'static'. + # + # consent: user consent forms (/_matrix/consent). See + # docs/consent_tracking.md. + # + # federation: the server-server API (/_matrix/federation). Also implies + # 'media', 'keys', 'openid' + # + # keys: the key discovery API (/_matrix/keys). + # + # media: the media API (/_matrix/media). + # + # metrics: the metrics interface. See docs/metrics-howto.rst. + # + # openid: OpenID authentication. + # + # replication: the HTTP replication API (/_synapse/replication). See + # docs/workers.rst. + # + # static: static resources under synapse/static (/_matrix/static). (Mostly + # useful for 'fallback authentication'.) + # + # webclient: A web client. Requires web_client_location to be set. + # listeners: - # Main HTTPS listener - # For when matrix traffic is sent directly to synapse. - - - # The port to listen for HTTPS requests on. - port: %(bind_port)s - - # Local addresses to listen on. - # On Linux and Mac OS, `::` will listen on all IPv4 and IPv6 - # addresses by default. For most other OSes, this will only listen - # on IPv6. - bind_addresses: - - '::' - - '0.0.0.0' - - # This is a 'http' listener, allows us to specify 'resources'. + # TLS-enabled listener: for when matrix traffic is sent directly to synapse. + # + # Disabled by default. To enable it, uncomment the following. (Note that you + # 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] + + # Unsecure HTTP listener: for when matrix traffic passes through a reverse proxy + # that unwraps TLS. + # + # If you plan to use a reverse proxy, please see + # https://github.com/matrix-org/synapse/blob/master/docs/reverse_proxy.rst. + # + - port: %(unsecure_port)s + tls: false + bind_addresses: ['::1', '127.0.0.1'] type: http + x_forwarded: true - tls: true - - # Use the X-Forwarded-For (XFF) header as the client IP and not the - # actual client IP. - x_forwarded: false - - # List of HTTP resources to serve on this listener. resources: - - - # List of resources to host on this listener. - names: - - client # The client-server APIs, both v1 and v2 - # - webclient # A web client. Requires web_client_location to be set. - - # Should synapse compress HTTP responses to clients that support it? - # This should be disabled if running synapse behind a load balancer - # that can do automatic compression. - compress: true - - - names: [federation] # Federation APIs + - names: [client, federation] compress: false - # # If federation is disabled synapse can still expose the open ID endpoint - # # to allow integrations to authenticate users - # - names: [openid] - # compress: false - - # optional list of additional endpoints which can be loaded via - # dynamic modules + # example additonal_resources: + # # additional_resources: # "/_matrix/my/custom/endpoint": # module: my_module.CustomRequestHandler # config: {} - # Unsecure HTTP listener, - # For when matrix traffic passes through loadbalancer that unwraps TLS. - - port: %(unsecure_port)s - tls: false - bind_addresses: ['::', '0.0.0.0'] - type: http - - x_forwarded: false - - resources: - - names: [client] - compress: true - - names: [federation] - compress: false - # # If federation is disabled synapse can still expose the open ID endpoint - # # to allow integrations to authenticate users - # - names: [openid] - # compress: false - # Turn on the twisted ssh manhole service on localhost on the given # port. # - port: 9000 # bind_addresses: ['::1', '127.0.0.1'] # type: manhole - # Homeserver blocking # # How to reach the server admin, used in ResourceLimitError diff --git a/synapse/config/tls.py b/synapse/config/tls.py index 9fcc79816d..5fb3486db1 100644 --- a/synapse/config/tls.py +++ b/synapse/config/tls.py @@ -23,9 +23,9 @@ from unpaddedbase64 import encode_base64 from OpenSSL import crypto -from synapse.config._base import Config +from synapse.config._base import Config, ConfigError -logger = logging.getLogger() +logger = logging.getLogger(__name__) class TlsConfig(Config): @@ -45,13 +45,25 @@ class TlsConfig(Config): self.tls_certificate_file = self.abspath(config.get("tls_certificate_path")) self.tls_private_key_file = self.abspath(config.get("tls_private_key_path")) + + if self.has_tls_listener(): + if not self.tls_certificate_file: + raise ConfigError( + "tls_certificate_path must be specified if TLS-enabled listeners are " + "configured." + ) + if not self.tls_private_key_file: + raise ConfigError( + "tls_certificate_path must be specified if TLS-enabled listeners are " + "configured." + ) + self._original_tls_fingerprints = config.get("tls_fingerprints", []) if self._original_tls_fingerprints is None: self._original_tls_fingerprints = [] self.tls_fingerprints = list(self._original_tls_fingerprints) - self.no_tls = config.get("no_tls", False) # This config option applies to non-federation HTTP clients # (e.g. for talking to recaptcha, identity servers, and such) @@ -106,36 +118,40 @@ class TlsConfig(Config): days_remaining = (expires_on - now).days return days_remaining - def read_certificate_from_disk(self): - """ - Read the certificates from disk. + def read_certificate_from_disk(self, require_cert_and_key): """ - self.tls_certificate = self.read_tls_certificate(self.tls_certificate_file) + Read the certificates and private key from disk. - # Check if it is self-signed, and issue a warning if so. - if self.tls_certificate.get_issuer() == self.tls_certificate.get_subject(): - warnings.warn( - ( - "Self-signed TLS certificates will not be accepted by Synapse 1.0. " - "Please either provide a valid certificate, or use Synapse's ACME " - "support to provision one." + Args: + require_cert_and_key (bool): set to True to throw an error if the certificate + and key file are not given + """ + if require_cert_and_key: + self.tls_private_key = self.read_tls_private_key() + self.tls_certificate = self.read_tls_certificate() + elif self.tls_certificate_file: + # we only need the certificate for the tls_fingerprints. Reload it if we + # can, but it's not a fatal error if we can't. + try: + self.tls_certificate = self.read_tls_certificate() + except Exception as e: + logger.info( + "Unable to read TLS certificate (%s). Ignoring as no " + "tls listeners enabled.", e, ) - ) - - if not self.no_tls: - self.tls_private_key = self.read_tls_private_key(self.tls_private_key_file) self.tls_fingerprints = list(self._original_tls_fingerprints) - # Check that our own certificate is included in the list of fingerprints - # and include it if it is not. - x509_certificate_bytes = crypto.dump_certificate( - crypto.FILETYPE_ASN1, self.tls_certificate - ) - sha256_fingerprint = encode_base64(sha256(x509_certificate_bytes).digest()) - sha256_fingerprints = set(f["sha256"] for f in self.tls_fingerprints) - if sha256_fingerprint not in sha256_fingerprints: - self.tls_fingerprints.append({u"sha256": sha256_fingerprint}) + if self.tls_certificate: + # Check that our own certificate is included in the list of fingerprints + # and include it if it is not. + x509_certificate_bytes = crypto.dump_certificate( + crypto.FILETYPE_ASN1, self.tls_certificate + ) + sha256_fingerprint = encode_base64(sha256(x509_certificate_bytes).digest()) + sha256_fingerprints = set(f["sha256"] for f in self.tls_fingerprints) + if sha256_fingerprint not in sha256_fingerprints: + self.tls_fingerprints.append({u"sha256": sha256_fingerprint}) def default_config(self, config_dir_path, server_name, **kwargs): base_key_name = os.path.join(config_dir_path, server_name) @@ -151,6 +167,8 @@ class TlsConfig(Config): return ( """\ + ## TLS ## + # PEM-encoded X509 certificate for TLS. # This certificate, as of Synapse 1.0, will need to be a valid and verifiable # certificate, signed by a recognised Certificate Authority. @@ -158,10 +176,10 @@ class TlsConfig(Config): # See 'ACME support' below to enable auto-provisioning this certificate via # Let's Encrypt. # - tls_certificate_path: "%(tls_certificate_path)s" + # tls_certificate_path: "%(tls_certificate_path)s" # PEM-encoded private key for TLS - tls_private_key_path: "%(tls_private_key_path)s" + # tls_private_key_path: "%(tls_private_key_path)s" # ACME support: This will configure Synapse to request a valid TLS certificate # for your configured `server_name` via Let's Encrypt. @@ -186,7 +204,7 @@ class TlsConfig(Config): # acme: # ACME support is disabled by default. Uncomment the following line - # to enable it. + # (and tls_certificate_path and tls_private_key_path above) to enable it. # # enabled: true @@ -211,13 +229,6 @@ class TlsConfig(Config): # # reprovision_threshold: 30 - # If your server runs behind a reverse-proxy which terminates TLS connections - # (for both client and federation connections), it may be useful to disable - # All TLS support for incoming connections. Setting no_tls to True will - # do so (and avoid the need to give synapse a TLS private key). - # - # no_tls: True - # List of allowed TLS fingerprints for this server to publish along # with the signing keys for this server. Other matrix servers that # make HTTPS requests to this server will check that the TLS @@ -250,10 +261,38 @@ class TlsConfig(Config): % locals() ) - def read_tls_certificate(self, cert_path): - cert_pem = self.read_file(cert_path, "tls_certificate") - return crypto.load_certificate(crypto.FILETYPE_PEM, cert_pem) + def read_tls_certificate(self): + """Reads the TLS certificate from the configured file, and returns it + + Also checks if it is self-signed, and warns if so + + Returns: + OpenSSL.crypto.X509: the certificate + """ + cert_path = self.tls_certificate_file + logger.info("Loading TLS certificate from %s", cert_path) + cert_pem = self.read_file(cert_path, "tls_certificate_path") + cert = crypto.load_certificate(crypto.FILETYPE_PEM, cert_pem) + + # Check if it is self-signed, and issue a warning if so. + if cert.get_issuer() == cert.get_subject(): + warnings.warn( + ( + "Self-signed TLS certificates will not be accepted by Synapse 1.0. " + "Please either provide a valid certificate, or use Synapse's ACME " + "support to provision one." + ) + ) + + return cert + + def read_tls_private_key(self): + """Reads the TLS private key from the configured file, and returns it - def read_tls_private_key(self, private_key_path): - private_key_pem = self.read_file(private_key_path, "tls_private_key") + Returns: + OpenSSL.crypto.PKey: the private key + """ + private_key_path = self.tls_private_key_file + logger.info("Loading TLS key from %s", private_key_path) + private_key_pem = self.read_file(private_key_path, "tls_private_key_path") return crypto.load_privatekey(crypto.FILETYPE_PEM, private_key_pem) diff --git a/synapse/crypto/context_factory.py b/synapse/crypto/context_factory.py index 286ad80100..85f2848fb1 100644 --- a/synapse/crypto/context_factory.py +++ b/synapse/crypto/context_factory.py @@ -43,9 +43,7 @@ class ServerContextFactory(ContextFactory): logger.exception("Failed to enable elliptic curve for TLS") context.set_options(SSL.OP_NO_SSLv2 | SSL.OP_NO_SSLv3) context.use_certificate_chain_file(config.tls_certificate_file) - - if not config.no_tls: - context.use_privatekey(config.tls_private_key) + context.use_privatekey(config.tls_private_key) # https://hynek.me/articles/hardening-your-web-servers-ssl-ciphers/ context.set_cipher_list( diff --git a/synapse/handlers/e2e_room_keys.py b/synapse/handlers/e2e_room_keys.py index 42b040375f..7bc174070e 100644 --- a/synapse/handlers/e2e_room_keys.py +++ b/synapse/handlers/e2e_room_keys.py @@ -19,7 +19,13 @@ from six import iteritems from twisted.internet import defer -from synapse.api.errors import NotFoundError, RoomKeysVersionError, StoreError +from synapse.api.errors import ( + Codes, + NotFoundError, + RoomKeysVersionError, + StoreError, + SynapseError, +) from synapse.util.async_helpers import Linearizer logger = logging.getLogger(__name__) @@ -267,7 +273,7 @@ class E2eRoomKeysHandler(object): version(str): Optional; if None gives the most recent version otherwise a historical one. Raises: - StoreError: code 404 if the requested backup version doesn't exist + NotFoundError: if the requested backup version doesn't exist Returns: A deferred of a info dict that gives the info about the new version. @@ -279,7 +285,13 @@ class E2eRoomKeysHandler(object): """ with (yield self._upload_linearizer.queue(user_id)): - res = yield self.store.get_e2e_room_keys_version_info(user_id, version) + try: + res = yield self.store.get_e2e_room_keys_version_info(user_id, version) + except StoreError as e: + if e.code == 404: + raise NotFoundError("Unknown backup version") + else: + raise defer.returnValue(res) @defer.inlineCallbacks @@ -290,8 +302,60 @@ class E2eRoomKeysHandler(object): user_id(str): the user whose current backup version we're deleting version(str): the version id of the backup being deleted Raises: - StoreError: code 404 if this backup version doesn't exist + NotFoundError: if this backup version doesn't exist """ with (yield self._upload_linearizer.queue(user_id)): - yield self.store.delete_e2e_room_keys_version(user_id, version) + try: + yield self.store.delete_e2e_room_keys_version(user_id, version) + except StoreError as e: + if e.code == 404: + raise NotFoundError("Unknown backup version") + else: + raise + + @defer.inlineCallbacks + def update_version(self, user_id, version, version_info): + """Update the info about a given version of the user's backup + + Args: + user_id(str): the user whose current backup version we're updating + version(str): the backup version we're updating + version_info(dict): the new information about the backup + Raises: + NotFoundError: if the requested backup version doesn't exist + Returns: + A deferred of an empty dict. + """ + if "version" not in version_info: + raise SynapseError( + 400, + "Missing version in body", + Codes.MISSING_PARAM + ) + if version_info["version"] != version: + raise SynapseError( + 400, + "Version in body does not match", + Codes.INVALID_PARAM + ) + with (yield self._upload_linearizer.queue(user_id)): + try: + old_info = yield self.store.get_e2e_room_keys_version_info( + user_id, version + ) + except StoreError as e: + if e.code == 404: + raise NotFoundError("Unknown backup version") + else: + raise + if old_info["algorithm"] != version_info["algorithm"]: + raise SynapseError( + 400, + "Algorithm does not match", + Codes.INVALID_PARAM + ) + + yield self.store.update_e2e_room_keys_version(user_id, version, version_info) + + defer.returnValue({}) diff --git a/synapse/http/matrixfederationclient.py b/synapse/http/matrixfederationclient.py index 5ee4d528d2..3c24bf3805 100644 --- a/synapse/http/matrixfederationclient.py +++ b/synapse/http/matrixfederationclient.py @@ -168,7 +168,7 @@ class MatrixFederationHttpClient(object): requests. """ - def __init__(self, hs): + def __init__(self, hs, tls_client_options_factory): self.hs = hs self.signing_key = hs.config.signing_key[0] self.server_name = hs.hostname @@ -176,7 +176,7 @@ class MatrixFederationHttpClient(object): self.agent = MatrixFederationAgent( hs.get_reactor(), - hs.tls_client_options_factory, + tls_client_options_factory, ) self.clock = hs.get_clock() self._store = hs.get_datastore() diff --git a/synapse/rest/client/v2_alpha/room_keys.py b/synapse/rest/client/v2_alpha/room_keys.py index ab3f1bd21a..220a0de30b 100644 --- a/synapse/rest/client/v2_alpha/room_keys.py +++ b/synapse/rest/client/v2_alpha/room_keys.py @@ -380,6 +380,40 @@ class RoomKeysVersionServlet(RestServlet): ) defer.returnValue((200, {})) + @defer.inlineCallbacks + def on_PUT(self, request, version): + """ + Update the information about a given version of the user's room_keys backup. + + POST /room_keys/version/12345 HTTP/1.1 + Content-Type: application/json + { + "algorithm": "m.megolm_backup.v1", + "auth_data": { + "public_key": "abcdefg", + "signatures": { + "ed25519:something": "hijklmnop" + } + }, + "version": "42" + } + + HTTP/1.1 200 OK + Content-Type: application/json + {} + """ + requester = yield self.auth.get_user_by_req(request, allow_guest=False) + user_id = requester.user.to_string() + info = parse_json_object_from_request(request) + + if version is None: + raise SynapseError(400, "No version specified to update", Codes.MISSING_PARAM) + + yield self.e2e_room_keys_handler.update_version( + user_id, version, info + ) + defer.returnValue((200, {})) + def register_servlets(hs, http_server): RoomKeysServlet(hs).register(http_server) diff --git a/synapse/server.py b/synapse/server.py index a2cf8a91cd..8615b67ad4 100644 --- a/synapse/server.py +++ b/synapse/server.py @@ -31,6 +31,7 @@ from synapse.api.filtering import Filtering from synapse.api.ratelimiting import Ratelimiter from synapse.appservice.api import ApplicationServiceApi from synapse.appservice.scheduler import ApplicationServiceScheduler +from synapse.crypto import context_factory from synapse.crypto.keyring import Keyring from synapse.events.builder import EventBuilderFactory from synapse.events.spamcheck import SpamChecker @@ -367,7 +368,10 @@ class HomeServer(object): return PusherPool(self) def build_http_client(self): - return MatrixFederationHttpClient(self) + tls_client_options_factory = context_factory.ClientTLSOptionsFactory( + self.config + ) + return MatrixFederationHttpClient(self, tls_client_options_factory) def build_db_pool(self): name = self.db_config["name"] diff --git a/synapse/storage/client_ips.py b/synapse/storage/client_ips.py index 091d7116c5..9c21362226 100644 --- a/synapse/storage/client_ips.py +++ b/synapse/storage/client_ips.py @@ -66,6 +66,11 @@ class ClientIpStore(background_updates.BackgroundUpdateStore): ) self.register_background_update_handler( + "user_ips_analyze", + self._analyze_user_ip, + ) + + self.register_background_update_handler( "user_ips_remove_dupes", self._remove_user_ip_dupes, ) @@ -109,6 +114,25 @@ class ClientIpStore(background_updates.BackgroundUpdateStore): defer.returnValue(1) @defer.inlineCallbacks + def _analyze_user_ip(self, progress, batch_size): + # Background update to analyze user_ips table before we run the + # deduplication background update. The table may not have been analyzed + # for ages due to the table locks. + # + # This will lock out the naive upserts to user_ips while it happens, but + # the analyze should be quick (28GB table takes ~10s) + def user_ips_analyze(txn): + txn.execute("ANALYZE user_ips") + + yield self.runInteraction( + "user_ips_analyze", user_ips_analyze + ) + + yield self._end_background_update("user_ips_analyze") + + defer.returnValue(1) + + @defer.inlineCallbacks def _remove_user_ip_dupes(self, progress, batch_size): # This works function works by scanning the user_ips table in batches # based on `last_seen`. For each row in a batch it searches the rest of @@ -167,12 +191,16 @@ class ClientIpStore(background_updates.BackgroundUpdateStore): clause = "? <= last_seen AND last_seen < ?" args = (begin_last_seen, end_last_seen) + # (Note: The DISTINCT in the inner query is important to ensure that + # the COUNT(*) is accurate, otherwise double counting may happen due + # to the join effectively being a cross product) txn.execute( """ SELECT user_id, access_token, ip, - MAX(device_id), MAX(user_agent), MAX(last_seen) + MAX(device_id), MAX(user_agent), MAX(last_seen), + COUNT(*) FROM ( - SELECT user_id, access_token, ip + SELECT DISTINCT user_id, access_token, ip FROM user_ips WHERE {} ) c @@ -186,7 +214,60 @@ class ClientIpStore(background_updates.BackgroundUpdateStore): # We've got some duplicates for i in res: - user_id, access_token, ip, device_id, user_agent, last_seen = i + user_id, access_token, ip, device_id, user_agent, last_seen, count = i + + # We want to delete the duplicates so we end up with only a + # single row. + # + # The naive way of doing this would be just to delete all rows + # and reinsert a constructed row. However, if there are a lot of + # duplicate rows this can cause the table to grow a lot, which + # can be problematic in two ways: + # 1. If user_ips is already large then this can cause the + # table to rapidly grow, potentially filling the disk. + # 2. Reinserting a lot of rows can confuse the table + # statistics for postgres, causing it to not use the + # correct indices for the query above, resulting in a full + # table scan. This is incredibly slow for large tables and + # can kill database performance. (This seems to mainly + # happen for the last query where the clause is simply `? < + # last_seen`) + # + # So instead we want to delete all but *one* of the duplicate + # rows. That is hard to do reliably, so we cheat and do a two + # step process: + # 1. Delete all rows with a last_seen strictly less than the + # max last_seen. This hopefully results in deleting all but + # one row the majority of the time, but there may be + # duplicate last_seen + # 2. If multiple rows remain, we fall back to the naive method + # and simply delete all rows and reinsert. + # + # Note that this relies on no new duplicate rows being inserted, + # but if that is happening then this entire process is futile + # anyway. + + # Do step 1: + + txn.execute( + """ + DELETE FROM user_ips + WHERE user_id = ? AND access_token = ? AND ip = ? AND last_seen < ? + """, + (user_id, access_token, ip, last_seen) + ) + if txn.rowcount == count - 1: + # We deleted all but one of the duplicate rows, i.e. there + # is exactly one remaining and so there is nothing left to + # do. + continue + elif txn.rowcount >= count: + raise Exception( + "We deleted more duplicate rows from 'user_ips' than expected", + ) + + # The previous step didn't delete enough rows, so we fallback to + # step 2: # Drop all the duplicates txn.execute( diff --git a/synapse/storage/e2e_room_keys.py b/synapse/storage/e2e_room_keys.py index 45cebe61d1..9a3aec759e 100644 --- a/synapse/storage/e2e_room_keys.py +++ b/synapse/storage/e2e_room_keys.py @@ -298,6 +298,27 @@ class EndToEndRoomKeyStore(SQLBaseStore): "create_e2e_room_keys_version_txn", _create_e2e_room_keys_version_txn ) + def update_e2e_room_keys_version(self, user_id, version, info): + """Update a given backup version + + Args: + user_id(str): the user whose backup version we're updating + version(str): the version ID of the backup version we're updating + info(dict): the new backup version info to store + """ + + return self._simple_update( + table="e2e_room_keys_versions", + keyvalues={ + "user_id": user_id, + "version": version, + }, + updatevalues={ + "auth_data": json.dumps(info["auth_data"]), + }, + desc="update_e2e_room_keys_version" + ) + def delete_e2e_room_keys_version(self, user_id, version=None): """Delete a given backup version of the user's room keys. Doesn't delete their actual key data. diff --git a/synapse/storage/schema/delta/53/user_ips_index.sql b/synapse/storage/schema/delta/53/user_ips_index.sql index 4ca346c111..b812c5794f 100644 --- a/synapse/storage/schema/delta/53/user_ips_index.sql +++ b/synapse/storage/schema/delta/53/user_ips_index.sql @@ -13,9 +13,13 @@ * limitations under the License. */ --- delete duplicates + -- analyze user_ips, to help ensure the correct indices are used INSERT INTO background_updates (update_name, progress_json) VALUES - ('user_ips_remove_dupes', '{}'); + ('user_ips_analyze', '{}'); + +-- delete duplicates +INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES + ('user_ips_remove_dupes', '{}', 'user_ips_analyze'); -- add a new unique index to user_ips table INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES @@ -23,4 +27,4 @@ INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES -- drop the old original index INSERT INTO background_updates (update_name, progress_json, depends_on) VALUES - ('user_ips_drop_nonunique_index', '{}', 'user_ips_device_unique_index'); \ No newline at end of file + ('user_ips_drop_nonunique_index', '{}', 'user_ips_device_unique_index'); |