support federation queries through http connect proxy

changelog.d/10475.feature

@@ -0,0 +1 @@
+Add support for sending federation requests through a proxy. Contributed by @Bubu and @dklimpel.

docs/upgrade.md

@@ -86,6 +86,30 @@ process, for example:
     ```
 
 
+# Upgrading to v1.xx.0
+
+## Add support for routing outbound HTTP requests via a proxy for federation
+
+Since Synapse 1.6.0 (2019-11-26) you can set a proxy for outbound HTTP requests via
+http_proxy/https_proxy environment variables. This proxy was set for:
+- push
+- url previews
+- phone-home stats
+- recaptcha validation
+- CAS auth validation
+- OpenID Connect
+
+In this version we have added support for outbound requests for:
+- federation
+- download remote media
+- fetch keys from other servers
+- Identity servers
+
+These requests use the same proxy configuration. If you have a proxy configuration we
+recommend to verify the configuration. It may be necessary to adjust the `no_proxy`
+environment variable.
+
+
 # Upgrading to v1.39.0
 
 ## Deprecation of the current third-party rules module interface

synapse/http/federation/matrix_federation_agent.py

@@ -14,6 +14,10 @@
 import logging
 import urllib.parse
 from typing import Any, Generator, List, Optional
+from urllib.request import (  # type: ignore[attr-defined]
+    getproxies_environment,
+    proxy_bypass_environment,
+)
 
 from netaddr import AddrFormatError, IPAddress, IPSet
 from zope.interface import implementer
@@ -25,14 +29,17 @@
     IReactorCore,
     IStreamClientEndpoint,
 )
-from twisted.web.client import URI, Agent, HTTPConnectionPool
+from twisted.web.client import URI, Agent, BrowserLikePolicyForHTTPS, HTTPConnectionPool
 from twisted.web.http_headers import Headers
 from twisted.web.iweb import IAgent, IAgentEndpointFactory, IBodyProducer, IResponse
 
 from synapse.crypto.context_factory import FederationPolicyForHTTPS
+from synapse.http import proxyagent
 from synapse.http.client import BlacklistingAgentWrapper
+from synapse.http.connectproxyclient import HTTPConnectProxyEndpoint
 from synapse.http.federation.srv_resolver import Server, SrvResolver
 from synapse.http.federation.well_known_resolver import WellKnownResolver
+from synapse.http.proxyagent import ProxyAgent
 from synapse.logging.context import make_deferred_yieldable, run_in_background
 from synapse.types import ISynapseReactor
 from synapse.util import Clock
@@ -57,6 +64,12 @@ class MatrixFederationAgent:
         user_agent:
             The user agent header to use for federation requests.
 
+        ip_blacklist: Disallowed IP addresses.
+
+        proxy_reactor: twisted reactor to use for connections to the proxy server
+           reactor might have some blacklisting applied (i.e. for DNS queries),
+           but we need unblocked access to the proxy.
+
         _srv_resolver:
             SrvResolver implementation to use for looking up SRV records. None
             to use a default implementation.
@@ -72,6 +85,7 @@ def __init__(
         tls_client_options_factory: Optional[FederationPolicyForHTTPS],
         user_agent: bytes,
         ip_blacklist: IPSet,
+        proxy_reactor: Optional[ISynapseReactor] = None,
         _srv_resolver: Optional[SrvResolver] = None,
         _well_known_resolver: Optional[WellKnownResolver] = None,
     ):
@@ -82,10 +96,18 @@ def __init__(
         self._pool.maxPersistentPerHost = 5
         self._pool.cachedConnectionTimeout = 2 * 60
 
+        if proxy_reactor is None:
+            self.proxy_reactor = reactor
+        else:
+            self.proxy_reactor = proxy_reactor
+
         self._agent = Agent.usingEndpointFactory(
             self._reactor,
             MatrixHostnameEndpointFactory(
-                reactor, tls_client_options_factory, _srv_resolver
+                reactor,
+                self.proxy_reactor,
+                tls_client_options_factory,
+                _srv_resolver,
             ),
             pool=self._pool,
         )
@@ -97,10 +119,12 @@ def __init__(
             _well_known_resolver = WellKnownResolver(
                 self._reactor,
                 agent=BlacklistingAgentWrapper(
-                    Agent(
+                    ProxyAgent(
                         self._reactor,
+                        self.proxy_reactor,
                         pool=self._pool,
                         contextFactory=tls_client_options_factory,
+                        use_proxy=True,
                     ),
                     ip_blacklist=ip_blacklist,
                 ),
@@ -200,20 +224,23 @@ class MatrixHostnameEndpointFactory:
     def __init__(
         self,
         reactor: IReactorCore,
+        proxy_reactor: IReactorCore,
         tls_client_options_factory: Optional[FederationPolicyForHTTPS],
         srv_resolver: Optional[SrvResolver],
     ):
         self._reactor = reactor
+        self._proxy_reactor = proxy_reactor
         self._tls_client_options_factory = tls_client_options_factory
 
         if srv_resolver is None:
             srv_resolver = SrvResolver()
 
         self._srv_resolver = srv_resolver
 
-    def endpointForURI(self, parsed_uri):
+    def endpointForURI(self, parsed_uri: URI):
         return MatrixHostnameEndpoint(
             self._reactor,
+            self._proxy_reactor,
             self._tls_client_options_factory,
             self._srv_resolver,
             parsed_uri,
@@ -227,6 +254,9 @@ class MatrixHostnameEndpoint:
 
     Args:
         reactor: twisted reactor to use for underlying requests
+        proxy_reactor: twisted reactor to use for connections to the proxy server
+           reactor might have some blacklisting applied (i.e. for DNS queries),
+           but we need unblocked access to the proxy.
         tls_client_options_factory:
             factory to use for fetching client tls options, or none to disable TLS.
         srv_resolver: The SRV resolver to use
@@ -236,14 +266,28 @@ class MatrixHostnameEndpoint:
     def __init__(
         self,
         reactor: IReactorCore,
+        proxy_reactor: IReactorCore,
         tls_client_options_factory: Optional[FederationPolicyForHTTPS],
         srv_resolver: SrvResolver,
         parsed_uri: URI,
     ):
         self._reactor = reactor
-
         self._parsed_uri = parsed_uri
 
+        # http_proxy is not needed because federation is always over TLS
+        proxies = getproxies_environment()
+        https_proxy = proxies["https"].encode() if "https" in proxies else None
+        self.no_proxy = proxies["no"] if "no" in proxies else None
+
+        (
+            self.https_proxy_endpoint,
+            self.https_proxy_creds,
+        ) = proxyagent.http_proxy_endpoint(
+            https_proxy,
+            proxy_reactor,
+            tls_client_options_factory or BrowserLikePolicyForHTTPS(),
+        )
+
         # set up the TLS connection params
         #
         # XXX disabling TLS is really only supported here for the benefit of the
@@ -273,9 +317,42 @@ async def _do_connect(self, protocol_factory: IProtocolFactory) -> None:
             host = server.host
             port = server.port
 
+            should_skip_proxy = False
+            if self.no_proxy is not None:
+                should_skip_proxy = proxy_bypass_environment(
+                    host.decode(),
+                    proxies={"no": self.no_proxy},
+                )
+
+            endpoint: IStreamClientEndpoint
             try:
-                logger.debug("Connecting to %s:%i", host.decode("ascii"), port)
-                endpoint = HostnameEndpoint(self._reactor, host, port)
+                if self.https_proxy_endpoint and not should_skip_proxy:
+                    logger.debug(
+                        "Connecting to %s:%i via %s",
+                        host.decode("ascii"),
+                        port,
+                        self.https_proxy_endpoint,
+                    )
+                    connect_headers = Headers()
+                    # Determine whether we need to set Proxy-Authorization headers
+                    if self.https_proxy_creds:
+                        # Set a Proxy-Authorization header
+                        connect_headers.addRawHeader(
+                            b"Proxy-Authorization",
+                            self.https_proxy_creds.as_proxy_authorization_value(),
+                        )
+
+                    endpoint = HTTPConnectProxyEndpoint(
+                        self._reactor,
+                        self.https_proxy_endpoint,
+                        host,
+                        port,
+                        headers=connect_headers,
+                    )
+                else:
+                    logger.debug("Connecting to %s:%i", host.decode("ascii"), port)
+                    # not using a proxy
+                    endpoint = HostnameEndpoint(self._reactor, host, port)
                 if self._tls_options:
                     endpoint = wrapClientTLS(self._tls_options, endpoint)
                 result = await make_deferred_yieldable(

synapse/http/matrixfederationclient.py

@@ -343,6 +343,7 @@ def __init__(self, hs, tls_client_options_factory):
             tls_client_options_factory,
             user_agent,
             hs.config.federation_ip_range_blacklist,
+            proxy_reactor=hs.get_reactor(),
         )
 
         # Use a BlacklistingAgentWrapper to prevent circumventing the IP

synapse/http/proxyagent.py

@@ -131,11 +131,11 @@ def __init__(
             https_proxy = proxies["https"].encode() if "https" in proxies else None
             no_proxy = proxies["no"] if "no" in proxies else None
 
-        self.http_proxy_endpoint, self.http_proxy_creds = _http_proxy_endpoint(
+        self.http_proxy_endpoint, self.http_proxy_creds = http_proxy_endpoint(
             http_proxy, self.proxy_reactor, contextFactory, **self._endpoint_kwargs
         )
 
-        self.https_proxy_endpoint, self.https_proxy_creds = _http_proxy_endpoint(
+        self.https_proxy_endpoint, self.https_proxy_creds = http_proxy_endpoint(
             https_proxy, self.proxy_reactor, contextFactory, **self._endpoint_kwargs
         )
 
@@ -268,7 +268,7 @@ def request(
         )
 
 
-def _http_proxy_endpoint(
+def http_proxy_endpoint(
     proxy: Optional[bytes],
     reactor: IReactorCore,
     tls_options_factory: IPolicyForHTTPS,