summary refs log tree commit diff
path: root/synapse/http/matrixfederationclient.py
diff options
context:
space:
mode:
authorRichard van der Hoff <github@rvanderhoff.org.uk>2017-03-23 22:27:01 +0000
committerGitHub <noreply@github.com>2017-03-23 22:27:01 +0000
commit9397edb28bb9aa091d71728f24b2765180bda6ba (patch)
tree3e9e44ed68ef1fbe1ecc2b39599cd458294c2125 /synapse/http/matrixfederationclient.py
parentMerge pull request #2052 from matrix-org/rav/time_bound_deferred (diff)
parentfix tests (diff)
downloadsynapse-9397edb28bb9aa091d71728f24b2765180bda6ba.tar.xz
Merge pull request #2050 from matrix-org/rav/federation_backoff
push federation retry limiter down to matrixfederationclient
Diffstat (limited to '')
-rw-r--r--synapse/http/matrixfederationclient.py294
1 files changed, 173 insertions, 121 deletions
diff --git a/synapse/http/matrixfederationclient.py b/synapse/http/matrixfederationclient.py
index 82586e3dea..f9e32ef03d 100644
--- a/synapse/http/matrixfederationclient.py
+++ b/synapse/http/matrixfederationclient.py
@@ -12,8 +12,7 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-
-
+import synapse.util.retryutils
 from twisted.internet import defer, reactor, protocol
 from twisted.internet.error import DNSLookupError
 from twisted.web.client import readBody, HTTPConnectionPool, Agent
@@ -94,6 +93,7 @@ class MatrixFederationHttpClient(object):
             reactor, MatrixFederationEndpointFactory(hs), pool=pool
         )
         self.clock = hs.get_clock()
+        self._store = hs.get_datastore()
         self.version_string = hs.version_string
         self._next_id = 1
 
@@ -103,129 +103,151 @@ class MatrixFederationHttpClient(object):
         )
 
     @defer.inlineCallbacks
-    def _create_request(self, destination, method, path_bytes,
-                        body_callback, headers_dict={}, param_bytes=b"",
-                        query_bytes=b"", retry_on_dns_fail=True,
-                        timeout=None, long_retries=False):
-        """ Creates and sends a request to the given url
+    def _request(self, destination, method, path,
+                 body_callback, headers_dict={}, param_bytes=b"",
+                 query_bytes=b"", retry_on_dns_fail=True,
+                 timeout=None, long_retries=False,
+                 ignore_backoff=False,
+                 backoff_on_404=False):
+        """ Creates and sends a request to the given server
+        Args:
+            destination (str): The remote server to send the HTTP request to.
+            method (str): HTTP method
+            path (str): The HTTP path
+            ignore_backoff (bool): true to ignore the historical backoff data
+                and try the request anyway.
+            backoff_on_404 (bool): Back off if we get a 404
 
         Returns:
             Deferred: resolves with the http response object on success.
 
             Fails with ``HTTPRequestException``: if we get an HTTP response
-            code >= 300.
+                code >= 300.
+            Fails with ``NotRetryingDestination`` if we are not yet ready
+                to retry this server.
         """
-        headers_dict[b"User-Agent"] = [self.version_string]
-        headers_dict[b"Host"] = [destination]
-
-        url_bytes = self._create_url(
-            destination, path_bytes, param_bytes, query_bytes
+        limiter = yield synapse.util.retryutils.get_retry_limiter(
+            destination,
+            self.clock,
+            self._store,
+            backoff_on_404=backoff_on_404,
+            ignore_backoff=ignore_backoff,
         )
 
-        txn_id = "%s-O-%s" % (method, self._next_id)
-        self._next_id = (self._next_id + 1) % (sys.maxint - 1)
+        destination = destination.encode("ascii")
+        path_bytes = path.encode("ascii")
+        with limiter:
+            headers_dict[b"User-Agent"] = [self.version_string]
+            headers_dict[b"Host"] = [destination]
 
-        outbound_logger.info(
-            "{%s} [%s] Sending request: %s %s",
-            txn_id, destination, method, url_bytes
-        )
+            url_bytes = self._create_url(
+                destination, path_bytes, param_bytes, query_bytes
+            )
 
-        # XXX: Would be much nicer to retry only at the transaction-layer
-        # (once we have reliable transactions in place)
-        if long_retries:
-            retries_left = MAX_LONG_RETRIES
-        else:
-            retries_left = MAX_SHORT_RETRIES
+            txn_id = "%s-O-%s" % (method, self._next_id)
+            self._next_id = (self._next_id + 1) % (sys.maxint - 1)
 
-        http_url_bytes = urlparse.urlunparse(
-            ("", "", path_bytes, param_bytes, query_bytes, "")
-        )
+            outbound_logger.info(
+                "{%s} [%s] Sending request: %s %s",
+                txn_id, destination, method, url_bytes
+            )
 
-        log_result = None
-        try:
-            while True:
-                producer = None
-                if body_callback:
-                    producer = body_callback(method, http_url_bytes, headers_dict)
-
-                try:
-                    def send_request():
-                        request_deferred = preserve_context_over_fn(
-                            self.agent.request,
-                            method,
-                            url_bytes,
-                            Headers(headers_dict),
-                            producer
-                        )
+            # XXX: Would be much nicer to retry only at the transaction-layer
+            # (once we have reliable transactions in place)
+            if long_retries:
+                retries_left = MAX_LONG_RETRIES
+            else:
+                retries_left = MAX_SHORT_RETRIES
 
-                        return self.clock.time_bound_deferred(
-                            request_deferred,
-                            time_out=timeout / 1000. if timeout else 60,
-                        )
+            http_url_bytes = urlparse.urlunparse(
+                ("", "", path_bytes, param_bytes, query_bytes, "")
+            )
 
-                    response = yield preserve_context_over_fn(send_request)
+            log_result = None
+            try:
+                while True:
+                    producer = None
+                    if body_callback:
+                        producer = body_callback(method, http_url_bytes, headers_dict)
+
+                    try:
+                        def send_request():
+                            request_deferred = preserve_context_over_fn(
+                                self.agent.request,
+                                method,
+                                url_bytes,
+                                Headers(headers_dict),
+                                producer
+                            )
+
+                            return self.clock.time_bound_deferred(
+                                request_deferred,
+                                time_out=timeout / 1000. if timeout else 60,
+                            )
+
+                        response = yield preserve_context_over_fn(send_request)
+
+                        log_result = "%d %s" % (response.code, response.phrase,)
+                        break
+                    except Exception as e:
+                        if not retry_on_dns_fail and isinstance(e, DNSLookupError):
+                            logger.warn(
+                                "DNS Lookup failed to %s with %s",
+                                destination,
+                                e
+                            )
+                            log_result = "DNS Lookup failed to %s with %s" % (
+                                destination, e
+                            )
+                            raise
 
-                    log_result = "%d %s" % (response.code, response.phrase,)
-                    break
-                except Exception as e:
-                    if not retry_on_dns_fail and isinstance(e, DNSLookupError):
                         logger.warn(
-                            "DNS Lookup failed to %s with %s",
+                            "{%s} Sending request failed to %s: %s %s: %s - %s",
+                            txn_id,
                             destination,
-                            e
+                            method,
+                            url_bytes,
+                            type(e).__name__,
+                            _flatten_response_never_received(e),
                         )
-                        log_result = "DNS Lookup failed to %s with %s" % (
-                            destination, e
+
+                        log_result = "%s - %s" % (
+                            type(e).__name__, _flatten_response_never_received(e),
                         )
-                        raise
-
-                    logger.warn(
-                        "{%s} Sending request failed to %s: %s %s: %s - %s",
-                        txn_id,
-                        destination,
-                        method,
-                        url_bytes,
-                        type(e).__name__,
-                        _flatten_response_never_received(e),
-                    )
-
-                    log_result = "%s - %s" % (
-                        type(e).__name__, _flatten_response_never_received(e),
-                    )
-
-                    if retries_left and not timeout:
-                        if long_retries:
-                            delay = 4 ** (MAX_LONG_RETRIES + 1 - retries_left)
-                            delay = min(delay, 60)
-                            delay *= random.uniform(0.8, 1.4)
+
+                        if retries_left and not timeout:
+                            if long_retries:
+                                delay = 4 ** (MAX_LONG_RETRIES + 1 - retries_left)
+                                delay = min(delay, 60)
+                                delay *= random.uniform(0.8, 1.4)
+                            else:
+                                delay = 0.5 * 2 ** (MAX_SHORT_RETRIES - retries_left)
+                                delay = min(delay, 2)
+                                delay *= random.uniform(0.8, 1.4)
+
+                            yield sleep(delay)
+                            retries_left -= 1
                         else:
-                            delay = 0.5 * 2 ** (MAX_SHORT_RETRIES - retries_left)
-                            delay = min(delay, 2)
-                            delay *= random.uniform(0.8, 1.4)
-
-                        yield sleep(delay)
-                        retries_left -= 1
-                    else:
-                        raise
-        finally:
-            outbound_logger.info(
-                "{%s} [%s] Result: %s",
-                txn_id,
-                destination,
-                log_result,
-            )
+                            raise
+            finally:
+                outbound_logger.info(
+                    "{%s} [%s] Result: %s",
+                    txn_id,
+                    destination,
+                    log_result,
+                )
 
-        if 200 <= response.code < 300:
-            pass
-        else:
-            # :'(
-            # Update transactions table?
-            body = yield preserve_context_over_fn(readBody, response)
-            raise HttpResponseException(
-                response.code, response.phrase, body
-            )
+            if 200 <= response.code < 300:
+                pass
+            else:
+                # :'(
+                # Update transactions table?
+                body = yield preserve_context_over_fn(readBody, response)
+                raise HttpResponseException(
+                    response.code, response.phrase, body
+                )
 
-        defer.returnValue(response)
+            defer.returnValue(response)
 
     def sign_request(self, destination, method, url_bytes, headers_dict,
                      content=None):
@@ -254,7 +276,9 @@ class MatrixFederationHttpClient(object):
 
     @defer.inlineCallbacks
     def put_json(self, destination, path, data={}, json_data_callback=None,
-                 long_retries=False, timeout=None):
+                 long_retries=False, timeout=None,
+                 ignore_backoff=False,
+                 backoff_on_404=False):
         """ Sends the specifed json data using PUT
 
         Args:
@@ -269,11 +293,19 @@ class MatrixFederationHttpClient(object):
                 retry for a short or long time.
             timeout(int): How long to try (in ms) the destination for before
                 giving up. None indicates no timeout.
+            ignore_backoff (bool): true to ignore the historical backoff data
+                and try the request anyway.
+            backoff_on_404 (bool): True if we should count a 404 response as
+                a failure of the server (and should therefore back off future
+                requests)
 
         Returns:
             Deferred: Succeeds when we get a 2xx HTTP response. The result
             will be the decoded JSON body. On a 4xx or 5xx error response a
             CodeMessageException is raised.
+
+            Fails with ``NotRetryingDestination`` if we are not yet ready
+            to retry this server.
         """
 
         if not json_data_callback:
@@ -288,14 +320,16 @@ class MatrixFederationHttpClient(object):
             producer = _JsonProducer(json_data)
             return producer
 
-        response = yield self._create_request(
-            destination.encode("ascii"),
+        response = yield self._request(
+            destination,
             "PUT",
-            path.encode("ascii"),
+            path,
             body_callback=body_callback,
             headers_dict={"Content-Type": ["application/json"]},
             long_retries=long_retries,
             timeout=timeout,
+            ignore_backoff=ignore_backoff,
+            backoff_on_404=backoff_on_404,
         )
 
         if 200 <= response.code < 300:
@@ -307,7 +341,7 @@ class MatrixFederationHttpClient(object):
 
     @defer.inlineCallbacks
     def post_json(self, destination, path, data={}, long_retries=False,
-                  timeout=None):
+                  timeout=None, ignore_backoff=False):
         """ Sends the specifed json data using POST
 
         Args:
@@ -320,11 +354,15 @@ class MatrixFederationHttpClient(object):
                 retry for a short or long time.
             timeout(int): How long to try (in ms) the destination for before
                 giving up. None indicates no timeout.
-
+            ignore_backoff (bool): true to ignore the historical backoff data and
+                try the request anyway.
         Returns:
             Deferred: Succeeds when we get a 2xx HTTP response. The result
             will be the decoded JSON body. On a 4xx or 5xx error response a
             CodeMessageException is raised.
+
+            Fails with ``NotRetryingDestination`` if we are not yet ready
+            to retry this server.
         """
 
         def body_callback(method, url_bytes, headers_dict):
@@ -333,14 +371,15 @@ class MatrixFederationHttpClient(object):
             )
             return _JsonProducer(data)
 
-        response = yield self._create_request(
-            destination.encode("ascii"),
+        response = yield self._request(
+            destination,
             "POST",
-            path.encode("ascii"),
+            path,
             body_callback=body_callback,
             headers_dict={"Content-Type": ["application/json"]},
             long_retries=long_retries,
             timeout=timeout,
+            ignore_backoff=ignore_backoff,
         )
 
         if 200 <= response.code < 300:
@@ -353,7 +392,7 @@ class MatrixFederationHttpClient(object):
 
     @defer.inlineCallbacks
     def get_json(self, destination, path, args={}, retry_on_dns_fail=True,
-                 timeout=None):
+                 timeout=None, ignore_backoff=False):
         """ GETs some json from the given host homeserver and path
 
         Args:
@@ -365,11 +404,16 @@ class MatrixFederationHttpClient(object):
             timeout (int): How long to try (in ms) the destination for before
                 giving up. None indicates no timeout and that the request will
                 be retried.
+            ignore_backoff (bool): true to ignore the historical backoff data
+                and try the request anyway.
         Returns:
             Deferred: Succeeds when we get *any* HTTP response.
 
             The result of the deferred is a tuple of `(code, response)`,
             where `response` is a dict representing the decoded JSON body.
+
+            Fails with ``NotRetryingDestination`` if we are not yet ready
+            to retry this server.
         """
         logger.debug("get_json args: %s", args)
 
@@ -386,14 +430,15 @@ class MatrixFederationHttpClient(object):
             self.sign_request(destination, method, url_bytes, headers_dict)
             return None
 
-        response = yield self._create_request(
-            destination.encode("ascii"),
+        response = yield self._request(
+            destination,
             "GET",
-            path.encode("ascii"),
+            path,
             query_bytes=query_bytes,
             body_callback=body_callback,
             retry_on_dns_fail=retry_on_dns_fail,
             timeout=timeout,
+            ignore_backoff=ignore_backoff,
         )
 
         if 200 <= response.code < 300:
@@ -406,19 +451,25 @@ class MatrixFederationHttpClient(object):
 
     @defer.inlineCallbacks
     def get_file(self, destination, path, output_stream, args={},
-                 retry_on_dns_fail=True, max_size=None):
+                 retry_on_dns_fail=True, max_size=None,
+                 ignore_backoff=False):
         """GETs a file from a given homeserver
         Args:
             destination (str): The remote server to send the HTTP request to.
             path (str): The HTTP path to GET.
             output_stream (file): File to write the response body to.
             args (dict): Optional dictionary used to create the query string.
+            ignore_backoff (bool): true to ignore the historical backoff data
+                and try the request anyway.
         Returns:
             Deferred: resolves with an (int,dict) tuple of the file length and
             a dict of the response headers.
 
             Fails with ``HTTPRequestException`` if we get an HTTP response code
             >= 300
+
+            Fails with ``NotRetryingDestination`` if we are not yet ready
+            to retry this server.
         """
 
         encoded_args = {}
@@ -434,13 +485,14 @@ class MatrixFederationHttpClient(object):
             self.sign_request(destination, method, url_bytes, headers_dict)
             return None
 
-        response = yield self._create_request(
-            destination.encode("ascii"),
+        response = yield self._request(
+            destination,
             "GET",
-            path.encode("ascii"),
+            path,
             query_bytes=query_bytes,
             body_callback=body_callback,
-            retry_on_dns_fail=retry_on_dns_fail
+            retry_on_dns_fail=retry_on_dns_fail,
+            ignore_backoff=ignore_backoff,
         )
 
         headers = dict(response.headers.getAllRawHeaders())