Merge branch 'rei/rss_target' into rei/rss_inc2

changelog.d/5771.feature

@@ -0,0 +1 @@
+Make Opentracing work in worker mode.

changelog.d/5776.misc

@@ -0,0 +1 @@
+Update opentracing docs to use the unified `trace` method.

changelog.d/5845.feature

@@ -0,0 +1 @@
+Add an admin API to purge old rooms from the database.

changelog.d/5850.feature

@@ -0,0 +1 @@
+Add retry to well-known lookups if we have recently seen a valid well-known record for the server.

changelog.d/5852.feature

@@ -0,0 +1 @@
+Pass opentracing contexts between servers when transmitting EDUs.

changelog.d/5855.misc

@@ -0,0 +1 @@
+Opentracing for room and e2e keys.

changelog.d/5860.misc

@@ -0,0 +1 @@
+Remove log line for debugging issue #5407.

changelog.d/5877.removal

@@ -0,0 +1 @@
+Remove shared secret registration from client/r0/register endpoint. Contributed by Awesome Technologies Innovationslabor GmbH.

changelog.d/5878.feature

@@ -0,0 +1 @@
+Add admin API endpoint for setting whether or not a user is a server administrator.

changelog.d/5885.bugfix

@@ -0,0 +1 @@
+Fix stack overflow when recovering an appservice which had an outage.

changelog.d/5886.misc

@@ -0,0 +1 @@
+Refactor the Appservice scheduler code.

changelog.d/5893.misc

@@ -0,0 +1 @@
+Drop some unused tables.

changelog.d/5894.misc

@@ -0,0 +1 @@
+Add missing index on users_in_public_rooms to improve the performance of directory queries.

changelog.d/5895.feature

@@ -0,0 +1 @@
+Add config option to sign remote key query responses with a separate key.

changelog.d/5896.misc

@@ -0,0 +1 @@
+Improve the logging when we have an error when fetching signing keys.

changelog.d/5906.feature

@@ -0,0 +1 @@
+Increase max display name size to 256.

changelog.d/5909.misc

@@ -0,0 +1 @@
+Fix error message which referred to public_base_url instead of public_baseurl. Thanks to @aaronraimist for the fix!

changelog.d/5911.misc

@@ -0,0 +1 @@
+Add support for database engine-specific schema deltas, based on file extension.

docs/admin_api/purge_room.md

@@ -0,0 +1,18 @@
+Purge room API
+==============
+
+This API will remove all trace of a room from your database.
+
+All local users must have left the room before it can be removed.
+
+The API is:
+
+```
+POST /_synapse/admin/v1/purge_room
+
+{
+    "room_id": "!room:id"
+}
+```
+
+You must authenticate using the access token of an admin user.

docs/admin_api/user_admin_api.rst

@@ -84,3 +84,23 @@ with a body of:
    }
 
 including an ``access_token`` of a server admin.
+
+
+Change whether a user is a server administrator or not
+======================================================
+
+Note that you cannot demote yourself.
+
+The api is::
+
+    PUT /_synapse/admin/v1/users/<user_id>/admin
+
+with a body of:
+
+.. code:: json
+
+    {
+        "admin": true
+    }
+
+including an ``access_token`` of a server admin.

docs/opentracing.rst

@@ -32,7 +32,7 @@ It is up to the remote server to decide what it does with the spans
 it creates. This is called the sampling policy and it can be configured
 through Jaeger's settings.
 
-For OpenTracing concepts see 
+For OpenTracing concepts see
 https://opentracing.io/docs/overview/what-is-tracing/.
 
 For more information about Jaeger's implementation see
@@ -79,7 +79,7 @@ Homeserver whitelisting
 
 The homeserver whitelist is configured using regular expressions. A list of regular
 expressions can be given and their union will be compared when propagating any
-spans contexts to another homeserver. 
+spans contexts to another homeserver.
 
 Though it's mostly safe to send and receive span contexts to and from
 untrusted users since span contexts are usually opaque ids it can lead to
@@ -92,6 +92,29 @@ two problems, namely:
   but that doesn't prevent another server sending you baggage which will be logged
   to OpenTracing's logs.
 
+==========
+EDU FORMAT
+==========
+
+EDUs can contain tracing data in their content. This is not specced but
+it could be of interest for other homeservers.
+
+EDU format (if you're using jaeger):
+
+.. code-block:: json
+
+   {
+     "edu_type": "type",
+     "content": {
+       "org.matrix.opentracing_context": {
+         "uber-trace-id": "fe57cf3e65083289"
+       }
+     }
+   }
+
+Though you don't have to use jaeger you must inject the span context into
+`org.matrix.opentracing_context` using the opentracing `Format.TEXT_MAP` inject method.
+
 ==================
 Configuring Jaeger
 ==================

docs/sample_config.yaml

@@ -1027,6 +1027,14 @@ signing_key_path: "CONFDIR/SERVERNAME.signing.key"
 #
 #trusted_key_servers:
 #  - server_name: "matrix.org"
+#
+
+# The signing keys to use when acting as a trusted key server. If not specified
+# defaults to the server signing key.
+#
+# Can contain multiple keys, one per line.
+#
+#key_server_signing_keys_path: "key_server_signing_keys.key"
 
 
 # Enable SAML2 for registration and login. Uses pysaml2.

synapse/appservice/scheduler.py

@@ -70,35 +70,37 @@ def __init__(self, hs):
         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 @@ def _send_request(self, service):
 
 
 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):
@@ -154,61 +172,63 @@ def send(self, service, events):
                 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)
         return state == ApplicationServiceState.UP or state is None
 
 
 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
@@ -224,7 +244,9 @@ def _retry():
                 "as-recoverer-%s" % (self.service.id,), self.retry
             )
 
-        self.clock.call_later((2 ** self.backoff_counter), _retry)
+        delay = 2 ** self.backoff_counter
+        logger.info("Scheduling retries on %s in %fs", self.service.id, delay)
+        self.clock.call_later(delay, _retry)
 
     def _backoff(self):
         # cap the backoff to be around 8.5min => (2^9) = 512 secs
@@ -234,25 +256,30 @@ def _backoff(self):
 
     @defer.inlineCallbacks
     def retry(self):
+        logger.info("Starting retries on %s", self.service.id)
         try:
-            txn = yield self.store.get_oldest_unsent_txn(self.service)
-            if txn:
+            while True:
+                txn = yield self.store.get_oldest_unsent_txn(self.service)
+                if not txn:
+                    # nothing left: we're done!
+                    self.callback(self)
+                    return
+
                 logger.info(
                     "Retrying transaction %s for AS ID %s", txn.id, txn.service.id
                 )
                 sent = yield txn.send(self.as_api)
-                if sent:
-                    yield txn.complete(self.store)
-                    # reset the backoff counter and retry immediately
-                    self.backoff_counter = 1
-                    yield self.retry()
-                else:
-                    self._backoff()
-            else:
-                self._set_service_recovered()
-        except Exception as e:
-            logger.exception(e)
-            self._backoff()
-
-    def _set_service_recovered(self):
-        self.callback(self)
+                if not sent:
+                    break
+
+                yield txn.complete(self.store)
+
+                # reset the backoff counter and then process the next transaction
+                self.backoff_counter = 1
+
+        except Exception:
+            logger.exception("Unexpected error running retries")
+
+        # we didn't manage to send all of the transactions before we got an error of
+        # some flavour: reschedule the next retry.
+        self._backoff()

synapse/config/emailconfig.py

@@ -115,7 +115,7 @@ def read_config(self, config, **kwargs):
                     missing.append("email." + k)
 
             if config.get("public_baseurl") is None:
-                missing.append("public_base_url")
+                missing.append("public_baseurl")
 
             if len(missing) > 0:
                 raise RuntimeError(