Merge remote-tracking branch 'origin/develop' into rav/warn_on_logcontext_fail

.travis.yml

@@ -1,14 +1,22 @@
 sudo: false
 language: python
-python: 2.7
 
 # tell travis to cache ~/.cache/pip
 cache: pip
 
-env:
+matrix:
-  - TOX_ENV=packaging
+  include:
-  - TOX_ENV=pep8
+  - python: 2.7
-  - TOX_ENV=py27
+    env: TOX_ENV=packaging
+
+  - python: 2.7
+    env: TOX_ENV=pep8
+
+  - python: 2.7
+    env: TOX_ENV=py27
+    
+  - python: 3.6
+    env: TOX_ENV=py36
 
 install:
   - pip install tox

CHANGES.rst

@@ -1,11 +1,245 @@
-Unreleased
+Changes in synapse <unreleased>
-==========
+===============================
 
-synctl no longer starts the main synapse when using ``-a`` option with workers.
+Potentially breaking change:
-A new worker file should be added with ``worker_app: synapse.app.homeserver``.
+
+* Make Client-Server API return 401 for invalid token (PR #3161).
+
+  This changes the Client-server spec to return a 401 error code instead of 403
+  when the access token is unrecognised. This is the behaviour required by the
+  specification, but some clients may be relying on the old, incorrect
+  behaviour.
+
+  Thanks to @NotAFile for fixing this.
+
+
+Changes in synapse v0.28.1 (2018-05-01)
+=======================================
+
+SECURITY UPDATE
+
+* Clamp the allowed values of event depth received over federation to be
+  [0, 2^63 - 1].  This mitigates an attack where malicious events
+  injected with depth = 2^63 - 1 render rooms unusable.  Depth is used to
+  determine the cosmetic ordering of events within a room, and so the ordering
+  of events in such a room will default to using stream_ordering rather than depth
+  (topological_ordering).
+
+  This is a temporary solution to mitigate abuse in the wild, whilst a long term solution
+  is being implemented to improve how the depth parameter is used.
+
+  Full details at
+  https://docs.google.com/document/d/1I3fi2S-XnpO45qrpCsowZv8P8dHcNZ4fsBsbOW7KABI
+
+* Pin Twisted to <18.4 until we stop using the private _OpenSSLECCurve API.
+
+
+Changes in synapse v0.28.0 (2018-04-26)
+=======================================
+
+Bug Fixes:
+
+* Fix quarantine media admin API and search reindex (PR #3130)
+* Fix media admin APIs (PR #3134)
+
+
+Changes in synapse v0.28.0-rc1 (2018-04-24)
+===========================================
+
+Minor performance improvement to federation sending and bug fixes.
+
+(Note: This release does not include the delta state resolution implementation discussed in matrix live)
+
+
+Features:
+
+* Add metrics for event processing lag (PR #3090)
+* Add metrics for ResponseCache (PR #3092)
+
+Changes:
+
+* Synapse on PyPy (PR #2760) Thanks to @Valodim!
+* move handling of auto_join_rooms to RegisterHandler (PR #2996) Thanks to @krombel!
+* Improve handling of SRV records for federation connections (PR #3016) Thanks to @silkeh!
+* Document the behaviour of ResponseCache (PR #3059)
+* Preparation for py3 (PR #3061, #3073, #3074, #3075, #3103, #3104, #3106, #3107, #3109, #3110) Thanks to @NotAFile!
+* update prometheus dashboard to use new metric names (PR #3069) Thanks to @krombel!
+* use python3-compatible prints (PR #3074) Thanks to @NotAFile!
+* Send federation events concurrently (PR #3078)
+* Limit concurrent event sends for a room (PR #3079)
+* Improve R30 stat definition (PR #3086)
+* Send events to ASes concurrently (PR #3088)
+* Refactor ResponseCache usage (PR #3093)
+* Clarify that SRV may not point to a CNAME (PR #3100) Thanks to @silkeh!
+* Use str(e) instead of e.message (PR #3103) Thanks to @NotAFile!
+* Use six.itervalues in some places (PR #3106) Thanks to @NotAFile!
+* Refactor store.have_events (PR #3117)
+
+Bug Fixes:
+
+* Return 401 for invalid access_token on logout (PR #2938) Thanks to @dklug!
+* Return a 404 rather than a 500 on rejoining empty rooms (PR #3080)
+* fix federation_domain_whitelist (PR #3099)
+* Avoid creating events with huge numbers of prev_events (PR #3113)
+* Reject events which have lots of prev_events (PR #3118)
+
+
+Changes in synapse v0.27.4 (2018-04-13)
+======================================
+
+Changes:
+
+* Update canonicaljson dependency (#3095)
+
+
+Changes in synapse v0.27.3 (2018-04-11)
+======================================
+
+Bug fixes:
+
+* URL quote path segments over federation (#3082)
+
+Changes in synapse v0.27.3-rc2 (2018-04-09)
+==========================================
+
+v0.27.3-rc1 used a stale version of the develop branch so the changelog overstates
+the functionality. v0.27.3-rc2 is up to date, rc1 should be ignored.
+
+Changes in synapse v0.27.3-rc1 (2018-04-09)
+=======================================
+
+Notable changes include API support for joinability of groups. Also new metrics
+and phone home stats. Phone home stats include better visibility of system usage
+so we can tweak synpase to work better for all users rather than our own experience
+with matrix.org. Also, recording 'r30' stat which is the measure we use to track
+overal growth of the Matrix ecosystem. It is defined as:-
+
+Counts the number of native 30 day retained users, defined as:-
+         * Users who have created their accounts more than 30 days
+         * Where last seen at most 30 days ago
+         * Where account creation and last_seen are > 30 days"
+
+
+Features:
+
+* Add joinability for groups (PR #3045)
+* Implement group join API (PR #3046)
+* Add counter metrics for calculating state delta (PR #3033)
+* R30 stats (PR #3041)
+* Measure time it takes to calculate state group ID (PR #3043)
+* Add basic performance statistics to phone home (PR #3044)
+* Add response size metrics (PR #3071)
+* phone home cache size configurations (PR #3063)
+
+Changes:
+
+* Add a blurb explaining the main synapse worker (PR #2886) Thanks to @turt2live!
+* Replace old style error catching with 'as' keyword (PR #3000) Thanks to @NotAFile!
+* Use .iter* to avoid copies in StateHandler (PR #3006)
+* Linearize calls to _generate_user_id (PR #3029)
+* Remove last usage of ujson (PR #3030)
+* Use simplejson throughout (PR #3048)
+* Use static JSONEncoders (PR #3049)
+* Remove uses of events.content (PR #3060)
+* Improve database cache performance (PR #3068)
+
+Bug fixes:
+
+* Add room_id to the response of `rooms/{roomId}/join` (PR #2986) Thanks to @jplatte!
+* Fix replication after switch to simplejson (PR #3015)
+* 404 correctly on missing paths via NoResource (PR #3022)
+* Fix error when claiming e2e keys from offline servers (PR #3034)
+* fix tests/storage/test_user_directory.py (PR #3042)
+* use PUT instead of POST for federating groups/m.join_policy (PR #3070) Thanks to @krombel!
+* postgres port script: fix state_groups_pkey error (PR #3072)
+
+
+Changes in synapse v0.27.2 (2018-03-26)
+=======================================
+
+Bug fixes:
+
+* Fix bug which broke TCP replication between workers (PR #3015)
+
+
+Changes in synapse v0.27.1 (2018-03-26)
+=======================================
+
+Meta release as v0.27.0 temporarily pointed to the wrong commit
+
+
+Changes in synapse v0.27.0 (2018-03-26)
+=======================================
+
+No changes since v0.27.0-rc2
+
+
+Changes in synapse v0.27.0-rc2 (2018-03-19)
+===========================================
+
+Pulls in v0.26.1
+
+Bug fixes:
+
+* Fix bug introduced in v0.27.0-rc1 that causes much increased memory usage in state cache (PR #3005)
+
+
+Changes in synapse v0.26.1 (2018-03-15)
+=======================================
+
+Bug fixes:
+
+* Fix bug where an invalid event caused server to stop functioning correctly,
+  due to parsing and serializing bugs in ujson library (PR #3008)
+
+
+Changes in synapse v0.27.0-rc1 (2018-03-14)
+===========================================
+
+The common case for running Synapse is not to run separate workers, but for those that do, be aware that synctl no longer starts the main synapse when using ``-a`` option with workers. A new worker file should be added with ``worker_app: synapse.app.homeserver``.
 
 This release also begins the process of renaming a number of the metrics
 reported to prometheus. See `docs/metrics-howto.rst <docs/metrics-howto.rst#block-and-response-metrics-renamed-for-0-27-0>`_.
+Note that the v0.28.0 release will remove the deprecated metric names.
+
+Features:
+
+* Add ability for ASes to override message send time (PR #2754)
+* Add support for custom storage providers for media repository (PR #2867, #2777, #2783, #2789, #2791, #2804, #2812, #2814, #2857, #2868, #2767)
+* Add purge API features, see `docs/admin_api/purge_history_api.rst <docs/admin_api/purge_history_api.rst>`_ for full details (PR #2858, #2867, #2882, #2946, #2962, #2943)
+* Add support for whitelisting 3PIDs that users can register. (PR #2813)
+* Add ``/room/{id}/event/{id}`` API (PR #2766)
+* Add an admin API to get all the media in a room (PR #2818) Thanks to @turt2live!
+* Add ``federation_domain_whitelist`` option (PR #2820, #2821)
+
+
+Changes:
+
+* Continue to factor out processing from main process and into worker processes. See updated `docs/workers.rst <docs/workers.rst>`_ (PR #2892 - #2904, #2913, #2920 - #2926, #2947, #2847, #2854, #2872, #2873, #2874, #2928, #2929, #2934, #2856, #2976 - #2984, #2987 - #2989, #2991 - #2993, #2995, #2784)
+* Ensure state cache is used when persisting events (PR #2864, #2871, #2802, #2835, #2836, #2841, #2842, #2849)
+* Change the default config to bind on both IPv4 and IPv6 on all platforms (PR #2435) Thanks to @silkeh!
+* No longer require a specific version of saml2 (PR #2695) Thanks to @okurz!
+* Remove ``verbosity``/``log_file`` from generated config (PR #2755)
+* Add and improve metrics and logging (PR #2770, #2778, #2785, #2786, #2787, #2793, #2794, #2795, #2809, #2810, #2833, #2834, #2844, #2965, #2927, #2975, #2790, #2796, #2838)
+* When using synctl with workers, don't start the main synapse automatically (PR #2774)
+* Minor performance improvements (PR #2773, #2792)
+* Use a connection pool for non-federation outbound connections (PR #2817)
+* Make it possible to run unit tests against postgres (PR #2829)
+* Update pynacl dependency to 1.2.1 or higher (PR #2888) Thanks to @bachp!
+* Remove ability for AS users to call /events and /sync (PR #2948)
+* Use bcrypt.checkpw (PR #2949) Thanks to @krombel!
+
+Bug fixes:
+
+* Fix broken ``ldap_config`` config option (PR #2683) Thanks to @seckrv!
+* Fix error message when user is not allowed to unban (PR #2761) Thanks to @turt2live!
+* Fix publicised groups GET API (singular) over federation (PR #2772)
+* Fix user directory when using ``user_directory_search_all_users`` config option (PR #2803, #2831)
+* Fix error on ``/publicRooms`` when no rooms exist (PR #2827)
+* Fix bug in quarantine_media (PR #2837)
+* Fix url_previews when no Content-Type is returned from URL (PR #2845)
+* Fix rare race in sync API when joining room (PR #2944)
+* Fix slow event search, switch back from GIST to GIN indexes (PR #2769, #2848)
 
 
 Changes in synapse v0.26.0 (2018-01-05)

CONTRIBUTING.rst

@@ -30,8 +30,12 @@ use github's pull request workflow to review the contribution, and either ask
 you to make any refinements needed or merge it and make them ourselves. The
 changes will then land on master when we next do a release.
 
-We use Jenkins for continuous integration (http://matrix.org/jenkins), and
+We use `Jenkins <http://matrix.org/jenkins>`_ and 
-typically all pull requests get automatically tested Jenkins: if your change breaks the build, Jenkins will yell about it in #matrix-dev:matrix.org so please lurk there and keep an eye open.
+`Travis <https://travis-ci.org/matrix-org/synapse>`_ for continuous
+integration. All pull requests to synapse get automatically tested by Travis; 
+the Jenkins builds require an adminstrator to start them. If your change 
+breaks the build, this will be shown in github, so please keep an eye on the 
+pull request for feedback.
 
 Code style
 ~~~~~~~~~~

README.rst

@@ -157,8 +157,8 @@ if you prefer.
 
 In case of problems, please see the _`Troubleshooting` section below.
 
-Alternatively, Silvio Fricke has contributed a Dockerfile to automate the
+Alternatively, Andreas Peters (previously Silvio Fricke) has contributed a Dockerfile to automate the
-above in Docker at https://registry.hub.docker.com/u/silviof/docker-matrix/.
+above in Docker at https://hub.docker.com/r/avhost/docker-matrix/tags/
 
 Also, Martin Giess has created an auto-deployment process with vagrant/ansible,
 tested with VirtualBox/AWS/DigitalOcean - see https://github.com/EMnify/matrix-synapse-auto-deploy
@@ -354,6 +354,10 @@ https://matrix.org/docs/projects/try-matrix-now.html (or build your own with one
 Fedora
 ------
 
+Synapse is in the Fedora repositories as ``matrix-synapse``::
+
+    sudo dnf install matrix-synapse
+
 Oleg Girko provides Fedora RPMs at
 https://obs.infoserver.lv/project/monitor/matrix-synapse
 
@@ -610,6 +614,9 @@ should have the format ``_matrix._tcp.<yourdomain.com> <ttl> IN SRV 10 0 <port>
     $ dig -t srv _matrix._tcp.example.com
     _matrix._tcp.example.com. 3600    IN      SRV     10 0 8448 synapse.example.com.
 
+Note that the server hostname cannot be an alias (CNAME record): it has to point
+directly to the server hosting the synapse instance.
+
 You can then configure your homeserver to use ``<yourdomain.com>`` as the domain in
 its user-ids, by setting ``server_name``::
 
@@ -890,6 +897,17 @@ This should end with a 'PASSED' result::
 
     PASSED (successes=143)
 
+Running the Integration Tests
+=============================
+
+Synapse is accompanied by `SyTest <https://github.com/matrix-org/sytest>`_,
+a Matrix homeserver integration testing suite, which uses HTTP requests to
+access the API as a Matrix client would. It is able to run Synapse directly from
+the source tree, so installation of the server is not required.
+
+Testing with SyTest is recommended for verifying that changes related to the
+Client-Server API are functioning correctly. See the `installation instructions
+<https://github.com/matrix-org/sytest#installing>`_ for details.
 
 Building Internal API Documentation
 ===================================

UPGRADE.rst

@@ -48,6 +48,18 @@ returned by the Client-Server API:
     # configured on port 443.
     curl -kv https://<host.name>/_matrix/client/versions 2>&1 | grep "Server:"
 
+Upgrading to $NEXT_VERSION
+====================
+
+This release expands the anonymous usage stats sent if the opt-in
+``report_stats`` configuration is set to ``true``. We now capture RSS memory 
+and cpu use at a very coarse level. This requires administrators to install
+the optional ``psutil`` python module.
+
+We would appreciate it if you could assist by ensuring this module is available
+and ``report_stats`` is enabled. This will let us see if performance changes to
+synapse are having an impact to the general community.
+
 Upgrading to v0.15.0
 ====================
 

contrib/README.rst

@@ -0,0 +1,10 @@
+Community Contributions
+=======================
+
+Everything in this directory are projects submitted by the community that may be useful
+to others. As such, the project maintainers cannot guarantee support, stability
+or backwards compatibility of these projects. 
+
+Files in this directory should *not* be relied on directly, as they may not
+continue to work or exist in future. If you wish to use any of these files then
+they should be copied to avoid them breaking from underneath you.

contrib/graph/graph3.py

@@ -22,6 +22,8 @@ import argparse
 from synapse.events import FrozenEvent
 from synapse.util.frozenutils import unfreeze
 
+from six import string_types
+
 
 def make_graph(file_name, room_id, file_prefix, limit):
     print "Reading lines"
@@ -58,7 +60,7 @@ def make_graph(file_name, room_id, file_prefix, limit):
         for key, value in unfreeze(event.get_dict()["content"]).items():
             if value is None:
                 value = "<null>"
-            elif isinstance(value, basestring):
+            elif isinstance(value, string_types):
                 pass
             else:
                 value = json.dumps(value)

contrib/prometheus/consoles/synapse.html

@@ -202,11 +202,11 @@ new PromConsole.Graph({
 <h1>Requests</h1>
 
 <h3>Requests by Servlet</h3>
-<div id="synapse_http_server_requests_servlet"></div>
+<div id="synapse_http_server_request_count_servlet"></div>
 <script>
 new PromConsole.Graph({
-  node: document.querySelector("#synapse_http_server_requests_servlet"),
+  node: document.querySelector("#synapse_http_server_request_count_servlet"),
-  expr: "rate(synapse_http_server_requests:servlet[2m])",
+  expr: "rate(synapse_http_server_request_count:servlet[2m])",
   name: "[[servlet]]",
   yAxisFormatter: PromConsole.NumberFormatter.humanize,
   yHoverFormatter: PromConsole.NumberFormatter.humanize,
@@ -215,11 +215,11 @@ new PromConsole.Graph({
 })
 </script>
 <h4>&nbsp;(without <tt>EventStreamRestServlet</tt> or <tt>SyncRestServlet</tt>)</h4>
-<div id="synapse_http_server_requests_servlet_minus_events"></div>
+<div id="synapse_http_server_request_count_servlet_minus_events"></div>
 <script>
 new PromConsole.Graph({
-  node: document.querySelector("#synapse_http_server_requests_servlet_minus_events"),
+  node: document.querySelector("#synapse_http_server_request_count_servlet_minus_events"),
-  expr: "rate(synapse_http_server_requests:servlet{servlet!=\"EventStreamRestServlet\", servlet!=\"SyncRestServlet\"}[2m])",
+  expr: "rate(synapse_http_server_request_count:servlet{servlet!=\"EventStreamRestServlet\", servlet!=\"SyncRestServlet\"}[2m])",
   name: "[[servlet]]",
   yAxisFormatter: PromConsole.NumberFormatter.humanize,
   yHoverFormatter: PromConsole.NumberFormatter.humanize,
@@ -233,7 +233,7 @@ new PromConsole.Graph({
 <script>
 new PromConsole.Graph({
   node: document.querySelector("#synapse_http_server_response_time_avg"),
-  expr: "rate(synapse_http_server_response_time:total[2m]) / rate(synapse_http_server_response_time:count[2m]) / 1000",
+  expr: "rate(synapse_http_server_response_time_seconds[2m]) / rate(synapse_http_server_response_count[2m]) / 1000",
   name: "[[servlet]]",
   yAxisFormatter: PromConsole.NumberFormatter.humanize,
   yHoverFormatter: PromConsole.NumberFormatter.humanize,
@@ -276,7 +276,7 @@ new PromConsole.Graph({
 <script>
 new PromConsole.Graph({
   node: document.querySelector("#synapse_http_server_response_ru_utime"),
-  expr: "rate(synapse_http_server_response_ru_utime:total[2m])",
+  expr: "rate(synapse_http_server_response_ru_utime_seconds[2m])",
   name: "[[servlet]]",
   yAxisFormatter: PromConsole.NumberFormatter.humanize,
   yHoverFormatter: PromConsole.NumberFormatter.humanize,
@@ -291,7 +291,7 @@ new PromConsole.Graph({
 <script>
 new PromConsole.Graph({
   node: document.querySelector("#synapse_http_server_response_db_txn_duration"),
-  expr: "rate(synapse_http_server_response_db_txn_duration:total[2m])",
+  expr: "rate(synapse_http_server_response_db_txn_duration_seconds[2m])",
   name: "[[servlet]]",
   yAxisFormatter: PromConsole.NumberFormatter.humanize,
   yHoverFormatter: PromConsole.NumberFormatter.humanize,
@@ -306,7 +306,7 @@ new PromConsole.Graph({
 <script>
 new PromConsole.Graph({
   node: document.querySelector("#synapse_http_server_send_time_avg"),
-  expr: "rate(synapse_http_server_response_time:total{servlet='RoomSendEventRestServlet'}[2m]) / rate(synapse_http_server_response_time:count{servlet='RoomSendEventRestServlet'}[2m]) / 1000",
+  expr: "rate(synapse_http_server_response_time_second{servlet='RoomSendEventRestServlet'}[2m]) / rate(synapse_http_server_response_count{servlet='RoomSendEventRestServlet'}[2m]) / 1000",
   name: "[[servlet]]",
   yAxisFormatter: PromConsole.NumberFormatter.humanize,
   yHoverFormatter: PromConsole.NumberFormatter.humanize,

contrib/prometheus/synapse-v1.rules

@@ -1,10 +1,10 @@
 synapse_federation_transaction_queue_pendingEdus:total = sum(synapse_federation_transaction_queue_pendingEdus or absent(synapse_federation_transaction_queue_pendingEdus)*0)
 synapse_federation_transaction_queue_pendingPdus:total = sum(synapse_federation_transaction_queue_pendingPdus or absent(synapse_federation_transaction_queue_pendingPdus)*0)
 
-synapse_http_server_requests:method{servlet=""} = sum(synapse_http_server_requests) by (method)
+synapse_http_server_request_count:method{servlet=""} = sum(synapse_http_server_request_count) by (method)
-synapse_http_server_requests:servlet{method=""} = sum(synapse_http_server_requests) by (servlet)
+synapse_http_server_request_count:servlet{method=""} = sum(synapse_http_server_request_count) by (servlet)
 
-synapse_http_server_requests:total{servlet=""} = sum(synapse_http_server_requests:by_method) by (servlet)
+synapse_http_server_request_count:total{servlet=""} = sum(synapse_http_server_request_count:by_method) by (servlet)
 
 synapse_cache:hit_ratio_5m = rate(synapse_util_caches_cache:hits[5m]) / rate(synapse_util_caches_cache:total[5m])
 synapse_cache:hit_ratio_30s = rate(synapse_util_caches_cache:hits[30s]) / rate(synapse_util_caches_cache:total[30s])

contrib/prometheus/synapse-v2.rules

@@ -5,19 +5,19 @@ groups:
     expr: "sum(synapse_federation_transaction_queue_pendingEdus or absent(synapse_federation_transaction_queue_pendingEdus)*0)"
   - record: "synapse_federation_transaction_queue_pendingPdus:total"
     expr:   "sum(synapse_federation_transaction_queue_pendingPdus or absent(synapse_federation_transaction_queue_pendingPdus)*0)"
-  - record: 'synapse_http_server_requests:method'
+  - record: 'synapse_http_server_request_count:method'
     labels:
       servlet: ""
-    expr: "sum(synapse_http_server_requests) by (method)"
+    expr: "sum(synapse_http_server_request_count) by (method)"
-  - record: 'synapse_http_server_requests:servlet'
+  - record: 'synapse_http_server_request_count:servlet'
     labels:
       method: ""
-    expr: 'sum(synapse_http_server_requests) by (servlet)'
+    expr: 'sum(synapse_http_server_request_count) by (servlet)'
 
-  - record: 'synapse_http_server_requests:total'
+  - record: 'synapse_http_server_request_count:total'
     labels:
       servlet: ""
-    expr: 'sum(synapse_http_server_requests:by_method) by (servlet)'
+    expr: 'sum(synapse_http_server_request_count:by_method) by (servlet)'
 
   - record: 'synapse_cache:hit_ratio_5m'
     expr: 'rate(synapse_util_caches_cache:hits[5m]) / rate(synapse_util_caches_cache:total[5m])'

contrib/systemd/synapse.service

@@ -2,6 +2,9 @@
 # (e.g. https://www.archlinux.org/packages/community/any/matrix-synapse/ for ArchLinux)
 # rather than in a user home directory or similar under virtualenv.
 
+# **NOTE:** This is an example service file that may change in the future. If you
+# wish to use this please copy rather than symlink it.
+
 [Unit]
 Description=Synapse Matrix homeserver
 
@@ -12,6 +15,7 @@ Group=synapse
 WorkingDirectory=/var/lib/synapse
 ExecStart=/usr/bin/python2.7 -m synapse.app.homeserver --config-path=/etc/synapse/homeserver.yaml
 ExecStop=/usr/bin/synctl stop /etc/synapse/homeserver.yaml
+# EnvironmentFile=-/etc/sysconfig/synapse  # Can be used to e.g. set SYNAPSE_CACHE_FACTOR 
 
 [Install]
 WantedBy=multi-user.target

docs/admin_api/purge_history_api.rst

@@ -16,9 +16,11 @@ including an ``access_token`` of a server admin.
 
 By default, events sent by local users are not deleted, as they may represent
 the only copies of this content in existence. (Events sent by remote users are
-deleted, and room state data before the cutoff is always removed).
+deleted.)
 
-To delete local events as well, set ``delete_local_events`` in the body:
+Room state data (such as joins, leaves, topic) is always preserved.
+
+To delete local message events as well, set ``delete_local_events`` in the body:
 
 .. code:: json
 

docs/workers.rst

@@ -55,7 +55,12 @@ synapse process.)
 
 You then create a set of configs for the various worker processes.  These
 should be worker configuration files, and should be stored in a dedicated
-subdirectory, to allow synctl to manipulate them.
+subdirectory, to allow synctl to manipulate them. An additional configuration
+for the master synapse process will need to be created because the process will
+not be started automatically. That configuration should look like this::
+
+    worker_app: synapse.app.homeserver
+    daemonize: true
 
 Each worker configuration file inherits the configuration of the main homeserver
 configuration file.  You can then override configuration specific to that worker,
@@ -230,9 +235,11 @@ file. For example::
 ``synapse.app.event_creator``
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
-Handles non-state event creation. It can handle REST endpoints matching:
+Handles some event creation. It can handle REST endpoints matching::
 
     ^/_matrix/client/(api/v1|r0|unstable)/rooms/.*/send
+    ^/_matrix/client/(api/v1|r0|unstable)/rooms/.*/(join|invite|leave|ban|unban|kick)$
+    ^/_matrix/client/(api/v1|r0|unstable)/join/
 
 It will create events locally and then send them on to the main synapse
 instance to be persisted and handled.

jenkins/prepare_synapse.sh

@@ -1,5 +1,7 @@
 #! /bin/bash
 
+set -eux
+
 cd "`dirname $0`/.."
 
 TOX_DIR=$WORKSPACE/.tox
@@ -14,7 +16,20 @@ fi
 tox -e py27 --notest -v
 
 TOX_BIN=$TOX_DIR/py27/bin
-$TOX_BIN/pip install setuptools
+
+# cryptography 2.2 requires setuptools >= 18.5.
+#
+# older versions of virtualenv (?) give us a virtualenv with the same version
+# of setuptools as is installed on the system python (and tox runs virtualenv
+# under python3, so we get the version of setuptools that is installed on that).
+#
+# anyway, make sure that we have a recent enough setuptools.
+$TOX_BIN/pip install 'setuptools>=18.5'
+
+# we also need a semi-recent version of pip, because old ones fail to install
+# the "enum34" dependency of cryptography.
+$TOX_BIN/pip install 'pip>=10'
+
 { python synapse/python_dependencies.py
   echo lxml psycopg2
 } | xargs $TOX_BIN/pip install

scripts/synapse_port_db

@@ -1,6 +1,7 @@
 #!/usr/bin/env python
 # -*- coding: utf-8 -*-
 # Copyright 2015, 2016 OpenMarket Ltd
+# Copyright 2018 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.
@@ -29,6 +30,8 @@ import time
 import traceback
 import yaml
 
+from six import string_types
+
 
 logger = logging.getLogger("synapse_port_db")
 
@@ -250,6 +253,12 @@ class Porter(object):
     @defer.inlineCallbacks
     def handle_table(self, table, postgres_size, table_size, forward_chunk,
                      backward_chunk):
+        logger.info(
+            "Table %s: %i/%i (rows %i-%i) already ported",
+            table, postgres_size, table_size,
+            backward_chunk+1, forward_chunk-1,
+        )
+
         if not table_size:
             return
 
@@ -467,31 +476,10 @@ class Porter(object):
             self.progress.set_state("Preparing PostgreSQL")
             self.setup_db(postgres_config, postgres_engine)
 
-            # Step 2. Get tables.
+            self.progress.set_state("Creating port tables")
-            self.progress.set_state("Fetching tables")
-            sqlite_tables = yield self.sqlite_store._simple_select_onecol(
-                table="sqlite_master",
-                keyvalues={
-                    "type": "table",
-                },
-                retcol="name",
-            )
-
-            postgres_tables = yield self.postgres_store._simple_select_onecol(
-                table="information_schema.tables",
-                keyvalues={},
-                retcol="distinct table_name",
-            )
-
-            tables = set(sqlite_tables) & set(postgres_tables)
-
-            self.progress.set_state("Creating tables")
-
-            logger.info("Found %d tables", len(tables))
-
             def create_port_table(txn):
                 txn.execute(
-                    "CREATE TABLE port_from_sqlite3 ("
+                    "CREATE TABLE IF NOT EXISTS port_from_sqlite3 ("
                     " table_name varchar(100) NOT NULL UNIQUE,"
                     " forward_rowid bigint NOT NULL,"
                     " backward_rowid bigint NOT NULL"
@@ -517,18 +505,33 @@ class Porter(object):
                     "alter_table", alter_table
                 )
             except Exception as e:
-                logger.info("Failed to create port table: %s", e)
+                pass
 
-            try:
             yield self.postgres_store.runInteraction(
                 "create_port_table", create_port_table
             )
-            except Exception as e:
-                logger.info("Failed to create port table: %s", e)
 
-            self.progress.set_state("Setting up")
+            # Step 2. Get tables.
+            self.progress.set_state("Fetching tables")
+            sqlite_tables = yield self.sqlite_store._simple_select_onecol(
+                table="sqlite_master",
+                keyvalues={
+                    "type": "table",
+                },
+                retcol="name",
+            )
 
-            # Set up tables.
+            postgres_tables = yield self.postgres_store._simple_select_onecol(
+                table="information_schema.tables",
+                keyvalues={},
+                retcol="distinct table_name",
+            )
+
+            tables = set(sqlite_tables) & set(postgres_tables)
+            logger.info("Found %d tables", len(tables))
+
+            # Step 3. Figure out what still needs copying
+            self.progress.set_state("Checking on port progress")
             setup_res = yield defer.gatherResults(
                 [
                     self.setup_table(table)
@@ -539,7 +542,8 @@ class Porter(object):
                 consumeErrors=True,
             )
 
-            # Process tables.
+            # Step 4. Do the copying.
+            self.progress.set_state("Copying to postgres")
             yield defer.gatherResults(
                 [
                     self.handle_table(*res)
@@ -548,6 +552,9 @@ class Porter(object):
                 consumeErrors=True,
             )
 
+            # Step 5. Do final post-processing
+            yield self._setup_state_group_id_seq()
+
             self.progress.done()
         except:
             global end_error_exec_info
@@ -569,7 +576,7 @@ class Porter(object):
         def conv(j, col):
             if j in bool_cols:
                 return bool(col)
-            elif isinstance(col, basestring) and "\0" in col:
+            elif isinstance(col, string_types) and "\0" in col:
                 logger.warn("DROPPING ROW: NUL value in table %s col %s: %r", table, headers[j], col)
                 raise BadValueException();
             return col
@@ -707,6 +714,16 @@ class Porter(object):
 
         defer.returnValue((done, remaining + done))
 
+    def _setup_state_group_id_seq(self):
+        def r(txn):
+            txn.execute("SELECT MAX(id) FROM state_groups")
+            next_id = txn.fetchone()[0]+1
+            txn.execute(
+                "ALTER SEQUENCE state_group_id_seq RESTART WITH %s",
+                (next_id,),
+            )
+        return self.postgres_store.runInteraction("setup_state_group_id_seq", r)
+
 
 ##############################################
 ###### The following is simply UI stuff ######

synapse/__init__.py

@@ -16,4 +16,4 @@
 """ This is a reference implementation of a Matrix home server.
 """
 
-__version__ = "0.26.0"
+__version__ = "0.28.1"

synapse/api/auth.py

@@ -204,8 +204,8 @@ class Auth(object):
 
             ip_addr = self.hs.get_ip_from_request(request)
             user_agent = request.requestHeaders.getRawHeaders(
-                "User-Agent",
+                b"User-Agent",
-                default=[""]
+                default=[b""]
             )[0]
             if user and access_token and ip_addr:
                 self.store.insert_client_ip(
@@ -672,7 +672,7 @@ def has_access_token(request):
         bool: False if no access_token was given, True otherwise.
     """
     query_params = request.args.get("access_token")
-    auth_headers = request.requestHeaders.getRawHeaders("Authorization")
+    auth_headers = request.requestHeaders.getRawHeaders(b"Authorization")
     return bool(query_params) or bool(auth_headers)
 
 
@@ -692,8 +692,8 @@ def get_access_token_from_request(request, token_not_found_http_status=401):
         AuthError: If there isn't an access_token in the request.
     """
 
-    auth_headers = request.requestHeaders.getRawHeaders("Authorization")
+    auth_headers = request.requestHeaders.getRawHeaders(b"Authorization")
-    query_params = request.args.get("access_token")
+    query_params = request.args.get(b"access_token")
     if auth_headers:
         # Try the get the access_token from a "Authorization: Bearer"
         # header

synapse/api/constants.py

@@ -16,6 +16,9 @@
 
 """Contains constants from the specification."""
 
+# the "depth" field on events is limited to 2**63 - 1
+MAX_DEPTH = 2**63 - 1
+
 
 class Membership(object):
 

synapse/api/errors.py

@@ -15,9 +15,11 @@
 
 """Contains exceptions and error codes."""
 
-import json
 import logging
 
+import simplejson as json
+from six import iteritems
+
 logger = logging.getLogger(__name__)
 
 
@@ -296,7 +298,7 @@ def cs_error(msg, code=Codes.UNKNOWN, **kwargs):
         A dict representing the error response JSON.
     """
     err = {"error": msg, "errcode": code}
-    for key, value in kwargs.iteritems():
+    for key, value in iteritems(kwargs):
         err[key] = value
     return err
 

synapse/api/filtering.py

@@ -17,7 +17,7 @@ from synapse.storage.presence import UserPresenceState
 from synapse.types import UserID, RoomID
 from twisted.internet import defer
 
-import ujson as json
+import simplejson as json
 import jsonschema
 from jsonschema import FormatChecker
 

synapse/app/appservice.py

@@ -32,11 +32,11 @@ from synapse.replication.tcp.client import ReplicationClientHandler
 from synapse.server import HomeServer
 from synapse.storage.engines import create_engine
 from synapse.util.httpresourcetree import create_resource_tree
-from synapse.util.logcontext import LoggingContext, preserve_fn
+from synapse.util.logcontext import LoggingContext, run_in_background
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
-from twisted.internet import reactor
+from twisted.internet import reactor, defer
-from twisted.web.resource import Resource
+from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.appservice")
 
@@ -64,7 +64,7 @@ class AppserviceServer(HomeServer):
                 if name == "metrics":
                     resources[METRICS_PREFIX] = MetricsResource(self)
 
-        root_resource = create_resource_tree(resources, Resource())
+        root_resource = create_resource_tree(resources, NoResource())
 
         _base.listen_tcp(
             bind_addresses,
@@ -112,9 +112,14 @@ class ASReplicationHandler(ReplicationClientHandler):
 
         if stream_name == "events":
             max_stream_id = self.store.get_room_max_stream_ordering()
-            preserve_fn(
+            run_in_background(self._notify_app_services, max_stream_id)
-                self.appservice_handler.notify_interested_services
+
-            )(max_stream_id)
+    @defer.inlineCallbacks
+    def _notify_app_services(self, room_stream_id):
+        try:
+            yield self.appservice_handler.notify_interested_services(room_stream_id)
+        except Exception:
+            logger.exception("Error notifying application services of event")
 
 
 def start(config_options):

synapse/app/client_reader.py

@@ -44,7 +44,7 @@ from synapse.util.logcontext import LoggingContext
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 from twisted.internet import reactor
-from twisted.web.resource import Resource
+from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.client_reader")
 
@@ -88,7 +88,7 @@ class ClientReaderServer(HomeServer):
                         "/_matrix/client/api/v1": resource,
                     })
 
-        root_resource = create_resource_tree(resources, Resource())
+        root_resource = create_resource_tree(resources, NoResource())
 
         _base.listen_tcp(
             bind_addresses,

synapse/app/event_creator.py

@@ -52,7 +52,7 @@ from synapse.util.logcontext import LoggingContext
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 from twisted.internet import reactor
-from twisted.web.resource import Resource
+from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.event_creator")
 
@@ -104,7 +104,7 @@ class EventCreatorServer(HomeServer):
                         "/_matrix/client/api/v1": resource,
                     })
 
-        root_resource = create_resource_tree(resources, Resource())
+        root_resource = create_resource_tree(resources, NoResource())
 
         _base.listen_tcp(
             bind_addresses,

synapse/app/federation_reader.py

@@ -41,7 +41,7 @@ from synapse.util.logcontext import LoggingContext
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 from twisted.internet import reactor
-from twisted.web.resource import Resource
+from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.federation_reader")
 
@@ -77,7 +77,7 @@ class FederationReaderServer(HomeServer):
                         FEDERATION_PREFIX: TransportLayerServer(self),
                     })
 
-        root_resource = create_resource_tree(resources, Resource())
+        root_resource = create_resource_tree(resources, NoResource())
 
         _base.listen_tcp(
             bind_addresses,

synapse/app/federation_sender.py

@@ -38,11 +38,11 @@ from synapse.server import HomeServer
 from synapse.storage.engines import create_engine
 from synapse.util.async import Linearizer
 from synapse.util.httpresourcetree import create_resource_tree
-from synapse.util.logcontext import LoggingContext, preserve_fn
+from synapse.util.logcontext import LoggingContext, run_in_background
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 from twisted.internet import defer, reactor
-from twisted.web.resource import Resource
+from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.federation_sender")
 
@@ -91,7 +91,7 @@ class FederationSenderServer(HomeServer):
                 if name == "metrics":
                     resources[METRICS_PREFIX] = MetricsResource(self)
 
-        root_resource = create_resource_tree(resources, Resource())
+        root_resource = create_resource_tree(resources, NoResource())
 
         _base.listen_tcp(
             bind_addresses,
@@ -229,7 +229,7 @@ class FederationSenderHandler(object):
         # presence, typing, etc.
         if stream_name == "federation":
             send_queue.process_rows_for_federation(self.federation_sender, rows)
-            preserve_fn(self.update_token)(token)
+            run_in_background(self.update_token, token)
 
         # We also need to poke the federation sender when new events happen
         elif stream_name == "events":
@@ -237,6 +237,7 @@ class FederationSenderHandler(object):
 
     @defer.inlineCallbacks
     def update_token(self, token):
+        try:
             self.federation_position = token
 
             # We linearize here to ensure we don't have races updating the token
@@ -250,6 +251,8 @@ class FederationSenderHandler(object):
                     # its in memory queues
                     self.replication_client.send_federation_ack(self.federation_position)
                     self._last_ack = self.federation_position
+        except Exception:
+            logger.exception("Error updating federation stream position")
 
 
 if __name__ == '__main__':

synapse/app/frontend_proxy.py

@@ -44,7 +44,7 @@ from synapse.util.logcontext import LoggingContext
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 from twisted.internet import defer, reactor
-from twisted.web.resource import Resource
+from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.frontend_proxy")
 
@@ -90,7 +90,7 @@ class KeyUploadServlet(RestServlet):
             # They're actually trying to upload something, proxy to main synapse.
             # Pass through the auth headers, if any, in case the access token
             # is there.
-            auth_headers = request.requestHeaders.getRawHeaders("Authorization", [])
+            auth_headers = request.requestHeaders.getRawHeaders(b"Authorization", [])
             headers = {
                 "Authorization": auth_headers,
             }
@@ -142,7 +142,7 @@ class FrontendProxyServer(HomeServer):
                         "/_matrix/client/api/v1": resource,
                     })
 
-        root_resource = create_resource_tree(resources, Resource())
+        root_resource = create_resource_tree(resources, NoResource())
 
         _base.listen_tcp(
             bind_addresses,

synapse/app/homeserver.py

@@ -48,6 +48,7 @@ from synapse.server import HomeServer
 from synapse.storage import are_all_users_on_domain
 from synapse.storage.engines import IncorrectDatabaseSetup, create_engine
 from synapse.storage.prepare_database import UpgradeDatabaseException, prepare_database
+from synapse.util.caches import CACHE_SIZE_FACTOR
 from synapse.util.httpresourcetree import create_resource_tree
 from synapse.util.logcontext import LoggingContext
 from synapse.util.manhole import manhole
@@ -56,7 +57,7 @@ from synapse.util.rlimit import change_resource_limit
 from synapse.util.versionstring import get_version_string
 from twisted.application import service
 from twisted.internet import defer, reactor
-from twisted.web.resource import EncodingResourceWrapper, Resource
+from twisted.web.resource import EncodingResourceWrapper, NoResource
 from twisted.web.server import GzipEncoderFactory
 from twisted.web.static import File
 
@@ -126,7 +127,7 @@ class SynapseHomeServer(HomeServer):
         if WEB_CLIENT_PREFIX in resources:
             root_resource = RootRedirect(WEB_CLIENT_PREFIX)
         else:
-            root_resource = Resource()
+            root_resource = NoResource()
 
         root_resource = create_resource_tree(resources, root_resource)
 
@@ -402,6 +403,10 @@ def run(hs):
 
     stats = {}
 
+    # Contains the list of processes we will be monitoring
+    # currently either 0 or 1
+    stats_process = []
+
     @defer.inlineCallbacks
     def phone_stats_home():
         logger.info("Gathering stats for reporting")
@@ -425,8 +430,21 @@ def run(hs):
         stats["daily_active_rooms"] = yield hs.get_datastore().count_daily_active_rooms()
         stats["daily_messages"] = yield hs.get_datastore().count_daily_messages()
 
+        r30_results = yield hs.get_datastore().count_r30_users()
+        for name, count in r30_results.iteritems():
+            stats["r30_users_" + name] = count
+
         daily_sent_messages = yield hs.get_datastore().count_daily_sent_messages()
         stats["daily_sent_messages"] = daily_sent_messages
+        stats["cache_factor"] = CACHE_SIZE_FACTOR
+        stats["event_cache_size"] = hs.config.event_cache_size
+
+        if len(stats_process) > 0:
+            stats["memory_rss"] = 0
+            stats["cpu_average"] = 0
+            for process in stats_process:
+                stats["memory_rss"] += process.memory_info().rss
+                stats["cpu_average"] += int(process.cpu_percent(interval=None))
 
         logger.info("Reporting stats to matrix.org: %s" % (stats,))
         try:
@@ -437,10 +455,32 @@ def run(hs):
         except Exception as e:
             logger.warn("Error reporting stats: %s", e)
 
+    def performance_stats_init():
+        try:
+            import psutil
+            process = psutil.Process()
+            # Ensure we can fetch both, and make the initial request for cpu_percent
+            # so the next request will use this as the initial point.
+            process.memory_info().rss
+            process.cpu_percent(interval=None)
+            logger.info("report_stats can use psutil")
+            stats_process.append(process)
+        except (ImportError, AttributeError):
+            logger.warn(
+                "report_stats enabled but psutil is not installed or incorrect version."
+                " Disabling reporting of memory/cpu stats."
+                " Ensuring psutil is available will help matrix.org track performance"
+                " changes across releases."
+            )
+
     if hs.config.report_stats:
         logger.info("Scheduling stats reporting for 3 hour intervals")
         clock.looping_call(phone_stats_home, 3 * 60 * 60 * 1000)
 
+        # We need to defer this init for the cases that we daemonize
+        # otherwise the process ID we get is that of the non-daemon process
+        clock.call_later(0, performance_stats_init)
+
         # We wait 5 minutes to send the first set of stats as the server can
         # be quite busy the first few minutes
         clock.call_later(5 * 60, phone_stats_home)

synapse/app/media_repository.py

@@ -43,7 +43,7 @@ from synapse.util.logcontext import LoggingContext
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 from twisted.internet import reactor
-from twisted.web.resource import Resource
+from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.media_repository")
 
@@ -84,7 +84,7 @@ class MediaRepositoryServer(HomeServer):
                         ),
                     })
 
-        root_resource = create_resource_tree(resources, Resource())
+        root_resource = create_resource_tree(resources, NoResource())
 
         _base.listen_tcp(
             bind_addresses,

synapse/app/pusher.py

@@ -33,11 +33,11 @@ from synapse.server import HomeServer
 from synapse.storage import DataStore
 from synapse.storage.engines import create_engine
 from synapse.util.httpresourcetree import create_resource_tree
-from synapse.util.logcontext import LoggingContext, preserve_fn
+from synapse.util.logcontext import LoggingContext, run_in_background
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 from twisted.internet import defer, reactor
-from twisted.web.resource import Resource
+from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.pusher")
 
@@ -94,7 +94,7 @@ class PusherServer(HomeServer):
                 if name == "metrics":
                     resources[METRICS_PREFIX] = MetricsResource(self)
 
-        root_resource = create_resource_tree(resources, Resource())
+        root_resource = create_resource_tree(resources, NoResource())
 
         _base.listen_tcp(
             bind_addresses,
@@ -140,10 +140,11 @@ class PusherReplicationHandler(ReplicationClientHandler):
 
     def on_rdata(self, stream_name, token, rows):
         super(PusherReplicationHandler, self).on_rdata(stream_name, token, rows)
-        preserve_fn(self.poke_pushers)(stream_name, token, rows)
+        run_in_background(self.poke_pushers, stream_name, token, rows)
 
     @defer.inlineCallbacks
     def poke_pushers(self, stream_name, token, rows):
+        try:
             if stream_name == "pushers":
                 for row in rows:
                     if row.deleted:
@@ -158,6 +159,8 @@ class PusherReplicationHandler(ReplicationClientHandler):
                 yield self.pusher_pool.on_new_receipts(
                     token, token, set(row.room_id for row in rows)
                 )
+        except Exception:
+            logger.exception("Error poking pushers")
 
     def stop_pusher(self, user_id, app_id, pushkey):
         key = "%s:%s" % (app_id, pushkey)

synapse/app/synchrotron.py

@@ -51,12 +51,14 @@ from synapse.storage.engines import create_engine
 from synapse.storage.presence import UserPresenceState
 from synapse.storage.roommember import RoomMemberStore
 from synapse.util.httpresourcetree import create_resource_tree
-from synapse.util.logcontext import LoggingContext, preserve_fn
+from synapse.util.logcontext import LoggingContext, run_in_background
 from synapse.util.manhole import manhole
 from synapse.util.stringutils import random_string
 from synapse.util.versionstring import get_version_string
 from twisted.internet import defer, reactor
-from twisted.web.resource import Resource
+from twisted.web.resource import NoResource
+
+from six import iteritems
 
 logger = logging.getLogger("synapse.app.synchrotron")
 
@@ -211,7 +213,7 @@ class SynchrotronPresence(object):
 
     def get_currently_syncing_users(self):
         return [
-            user_id for user_id, count in self.user_to_num_current_syncs.iteritems()
+            user_id for user_id, count in iteritems(self.user_to_num_current_syncs)
             if count > 0
         ]
 
@@ -269,7 +271,7 @@ class SynchrotronServer(HomeServer):
                         "/_matrix/client/api/v1": resource,
                     })
 
-        root_resource = create_resource_tree(resources, Resource())
+        root_resource = create_resource_tree(resources, NoResource())
 
         _base.listen_tcp(
             bind_addresses,
@@ -325,8 +327,7 @@ class SyncReplicationHandler(ReplicationClientHandler):
 
     def on_rdata(self, stream_name, token, rows):
         super(SyncReplicationHandler, self).on_rdata(stream_name, token, rows)
-
+        run_in_background(self.process_and_notify, stream_name, token, rows)
-        preserve_fn(self.process_and_notify)(stream_name, token, rows)
 
     def get_streams_to_replicate(self):
         args = super(SyncReplicationHandler, self).get_streams_to_replicate()
@@ -338,6 +339,7 @@ class SyncReplicationHandler(ReplicationClientHandler):
 
     @defer.inlineCallbacks
     def process_and_notify(self, stream_name, token, rows):
+        try:
             if stream_name == "events":
                 # We shouldn't get multiple rows per token for events stream, so
                 # we don't need to optimise this for multiple rows.
@@ -387,6 +389,8 @@ class SyncReplicationHandler(ReplicationClientHandler):
                 self.notifier.on_new_event(
                     "groups_key", token, users=[row.user_id for row in rows],
                 )
+        except Exception:
+            logger.exception("Error processing replication")
 
 
 def start(config_options):

synapse/app/synctl.py

@@ -38,7 +38,7 @@ def pid_running(pid):
     try:
         os.kill(pid, 0)
         return True
-    except OSError, err:
+    except OSError as err:
         if err.errno == errno.EPERM:
             return True
         return False
@@ -98,7 +98,7 @@ def stop(pidfile, app):
         try:
             os.kill(pid, signal.SIGTERM)
             write("stopped %s" % (app,), colour=GREEN)
-        except OSError, err:
+        except OSError as err:
             if err.errno == errno.ESRCH:
                 write("%s not running" % (app,), colour=YELLOW)
             elif err.errno == errno.EPERM:
@@ -252,6 +252,7 @@ def main():
             for running_pid in running_pids:
                 while pid_running(running_pid):
                     time.sleep(0.2)
+            write("All processes exited; now restarting...")
 
     if action == "start" or action == "restart":
         if start_stop_synapse:

synapse/app/user_dir.py

@@ -39,11 +39,11 @@ from synapse.storage.engines import create_engine
 from synapse.storage.user_directory import UserDirectoryStore
 from synapse.util.caches.stream_change_cache import StreamChangeCache
 from synapse.util.httpresourcetree import create_resource_tree
-from synapse.util.logcontext import LoggingContext, preserve_fn
+from synapse.util.logcontext import LoggingContext, run_in_background
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
-from twisted.internet import reactor
+from twisted.internet import reactor, defer
-from twisted.web.resource import Resource
+from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.user_dir")
 
@@ -116,7 +116,7 @@ class UserDirectoryServer(HomeServer):
                         "/_matrix/client/api/v1": resource,
                     })
 
-        root_resource = create_resource_tree(resources, Resource())
+        root_resource = create_resource_tree(resources, NoResource())
 
         _base.listen_tcp(
             bind_addresses,
@@ -164,7 +164,14 @@ class UserDirectoryReplicationHandler(ReplicationClientHandler):
             stream_name, token, rows
         )
         if stream_name == "current_state_deltas":
-            preserve_fn(self.user_directory.notify_new_event)()
+            run_in_background(self._notify_directory)
+
+    @defer.inlineCallbacks
+    def _notify_directory(self):
+        try:
+            yield self.user_directory.notify_new_event()
+        except Exception:
+            logger.exception("Error notifiying user directory of state update")
 
 
 def start(config_options):

synapse/appservice/__init__.py

@@ -21,6 +21,8 @@ from twisted.internet import defer
 import logging
 import re
 
+from six import string_types
+
 logger = logging.getLogger(__name__)
 
 
@@ -146,7 +148,7 @@ class ApplicationService(object):
                         )
 
                 regex = regex_obj.get("regex")
-                if isinstance(regex, basestring):
+                if isinstance(regex, string_types):
                     regex_obj["regex"] = re.compile(regex)  # Pre-compile regex
                 else:
                     raise ValueError(

synapse/appservice/api.py

@@ -18,7 +18,6 @@ from synapse.api.constants import ThirdPartyEntityKind
 from synapse.api.errors import CodeMessageException
 from synapse.http.client import SimpleHttpClient
 from synapse.events.utils import serialize_event
-from synapse.util.logcontext import preserve_fn, make_deferred_yieldable
 from synapse.util.caches.response_cache import ResponseCache
 from synapse.types import ThirdPartyInstanceID
 
@@ -73,7 +72,8 @@ class ApplicationServiceApi(SimpleHttpClient):
         super(ApplicationServiceApi, self).__init__(hs)
         self.clock = hs.get_clock()
 
-        self.protocol_meta_cache = ResponseCache(hs, timeout_ms=HOUR_IN_MS)
+        self.protocol_meta_cache = ResponseCache(hs, "as_protocol_meta",
+                                                 timeout_ms=HOUR_IN_MS)
 
     @defer.inlineCallbacks
     def query_user(self, service, user_id):
@@ -193,12 +193,7 @@ class ApplicationServiceApi(SimpleHttpClient):
                 defer.returnValue(None)
 
         key = (service.id, protocol)
-        result = self.protocol_meta_cache.get(key)
+        return self.protocol_meta_cache.wrap(key, _get)
-        if not result:
-            result = self.protocol_meta_cache.set(
-                key, preserve_fn(_get)()
-            )
-        return make_deferred_yieldable(result)
 
     @defer.inlineCallbacks
     def push_bulk(self, service, events, txn_id=None):

synapse/appservice/scheduler.py

@@ -51,7 +51,7 @@ components.
 from twisted.internet import defer
 
 from synapse.appservice import ApplicationServiceState
-from synapse.util.logcontext import preserve_fn
+from synapse.util.logcontext import run_in_background
 from synapse.util.metrics import Measure
 
 import logging
@@ -106,7 +106,7 @@ class _ServiceQueuer(object):
     def enqueue(self, service, event):
         # if this service isn't being sent something
         self.queued_events.setdefault(service.id, []).append(event)
-        preserve_fn(self._send_request)(service)
+        run_in_background(self._send_request, service)
 
     @defer.inlineCallbacks
     def _send_request(self, service):
@@ -152,10 +152,10 @@ class _TransactionController(object):
                 if sent:
                     yield txn.complete(self.store)
                 else:
-                    preserve_fn(self._start_recoverer)(service)
+                    run_in_background(self._start_recoverer, service)
-        except Exception as e:
+        except Exception:
-            logger.exception(e)
+            logger.exception("Error creating appservice transaction")
-            preserve_fn(self._start_recoverer)(service)
+            run_in_background(self._start_recoverer, service)
 
     @defer.inlineCallbacks
     def on_recovered(self, recoverer):
@@ -176,6 +176,7 @@ class _TransactionController(object):
 
     @defer.inlineCallbacks
     def _start_recoverer(self, service):
+        try:
             yield self.store.set_appservice_state(
                 service,
                 ApplicationServiceState.DOWN
@@ -187,6 +188,8 @@ class _TransactionController(object):
             recoverer = self.recoverer_fn(service, self.on_recovered)
             self.add_recoverers([recoverer])
             recoverer.recover()
+        except Exception:
+            logger.exception("Error starting AS recoverer")
 
     @defer.inlineCallbacks
     def _is_service_up(self, service):

synapse/config/_base.py

@@ -19,6 +19,8 @@ import os
 import yaml
 from textwrap import dedent
 
+from six import integer_types
+
 
 class ConfigError(Exception):
     pass
@@ -49,7 +51,7 @@ Missing mandatory `server_name` config option.
 class Config(object):
     @staticmethod
     def parse_size(value):
-        if isinstance(value, int) or isinstance(value, long):
+        if isinstance(value, integer_types):
             return value
         sizes = {"K": 1024, "M": 1024 * 1024}
         size = 1
@@ -61,7 +63,7 @@ class Config(object):
 
     @staticmethod
     def parse_duration(value):
-        if isinstance(value, int) or isinstance(value, long):
+        if isinstance(value, integer_types):
             return value
         second = 1000
         minute = 60 * second
@@ -279,31 +281,31 @@ class Config(object):
                     )
                 if not cls.path_exists(config_dir_path):
                     os.makedirs(config_dir_path)
-                with open(config_path, "wb") as config_file:
+                with open(config_path, "w") as config_file:
-                    config_bytes, config = obj.generate_config(
+                    config_str, config = obj.generate_config(
                         config_dir_path=config_dir_path,
                         server_name=server_name,
                         report_stats=(config_args.report_stats == "yes"),
                         is_generating_file=True
                     )
                     obj.invoke_all("generate_files", config)
-                    config_file.write(config_bytes)
+                    config_file.write(config_str)
-                print (
+                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"
                     " to your needs."
-                ) % (config_path, server_name)
+                ) % (config_path, server_name))
-                print (
+                print(
                     "If this server name is incorrect, you will need to"
                     " regenerate the SSL certificates"
                 )
                 return
             else:
-                print (
+                print((
                     "Config file %r already exists. Generating any missing key"
                     " files."
-                ) % (config_path,)
+                ) % (config_path,))
                 generate_keys = True
 
         parser = argparse.ArgumentParser(

synapse/config/appservice.py

@@ -17,10 +17,12 @@ from ._base import Config, ConfigError
 from synapse.appservice import ApplicationService
 from synapse.types import UserID
 
-import urllib
 import yaml
 import logging
 
+from six import string_types
+from six.moves.urllib import parse as urlparse
+
 logger = logging.getLogger(__name__)
 
 
@@ -89,21 +91,21 @@ def _load_appservice(hostname, as_info, config_filename):
         "id", "as_token", "hs_token", "sender_localpart"
     ]
     for field in required_string_fields:
-        if not isinstance(as_info.get(field), basestring):
+        if not isinstance(as_info.get(field), string_types):
             raise KeyError("Required string field: '%s' (%s)" % (
                 field, config_filename,
             ))
 
     # 'url' must either be a string or explicitly null, not missing
     # to avoid accidentally turning off push for ASes.
-    if (not isinstance(as_info.get("url"), basestring) and
+    if (not isinstance(as_info.get("url"), string_types) and
             as_info.get("url", "") is not None):
         raise KeyError(
             "Required string field or explicit null: 'url' (%s)" % (config_filename,)
         )
 
     localpart = as_info["sender_localpart"]
-    if urllib.quote(localpart) != localpart:
+    if urlparse.quote(localpart) != localpart:
         raise ValueError(
             "sender_localpart needs characters which are not URL encoded."
         )
@@ -128,7 +130,7 @@ def _load_appservice(hostname, as_info, config_filename):
                         "Expected namespace entry in %s to be an object,"
                         " but got %s", ns, regex_obj
                     )
-                if not isinstance(regex_obj.get("regex"), basestring):
+                if not isinstance(regex_obj.get("regex"), string_types):
                     raise ValueError(
                         "Missing/bad type 'regex' key in %s", regex_obj
                     )

synapse/config/logger.py

@@ -117,7 +117,7 @@ class LoggingConfig(Config):
         log_config = config.get("log_config")
         if log_config and not os.path.exists(log_config):
             log_file = self.abspath("homeserver.log")
-            with open(log_config, "wb") as log_config_file:
+            with open(log_config, "w") as log_config_file:
                 log_config_file.write(
                     DEFAULT_LOG_CONFIG.substitute(log_file=log_file)
                 )

synapse/config/registration.py

@@ -77,7 +77,9 @@ class RegistrationConfig(Config):
 
         # Set the number of bcrypt rounds used to generate password hash.
         # Larger numbers increase the work factor needed to generate the hash.
-        # The default number of rounds is 12.
+        # The default number is 12 (which equates to 2^12 rounds).
+        # N.B. that increasing this will exponentially increase the time required
+        # to register or login - e.g. 24 => 2^24 rounds which will take >20 mins.
         bcrypt_rounds: 12
 
         # Allows users to register as guests without a password/email/etc, and

synapse/config/tls.py

@@ -133,7 +133,7 @@ class TlsConfig(Config):
         tls_dh_params_path = config["tls_dh_params_path"]
 
         if not self.path_exists(tls_private_key_path):
-            with open(tls_private_key_path, "w") as private_key_file:
+            with open(tls_private_key_path, "wb") as private_key_file:
                 tls_private_key = crypto.PKey()
                 tls_private_key.generate_key(crypto.TYPE_RSA, 2048)
                 private_key_pem = crypto.dump_privatekey(
@@ -148,7 +148,7 @@ class TlsConfig(Config):
                 )
 
         if not self.path_exists(tls_certificate_path):
-            with open(tls_certificate_path, "w") as certificate_file:
+            with open(tls_certificate_path, "wb") as certificate_file:
                 cert = crypto.X509()
                 subject = cert.get_subject()
                 subject.CN = config["server_name"]

synapse/crypto/context_factory.py

@@ -13,8 +13,8 @@
 # limitations under the License.
 
 from twisted.internet import ssl
-from OpenSSL import SSL
+from OpenSSL import SSL, crypto
-from twisted.internet._sslverify import _OpenSSLECCurve, _defaultCurveName
+from twisted.internet._sslverify import _defaultCurveName
 
 import logging
 
@@ -32,8 +32,9 @@ class ServerContextFactory(ssl.ContextFactory):
     @staticmethod
     def configure_context(context, config):
         try:
-            _ecCurve = _OpenSSLECCurve(_defaultCurveName)
+            _ecCurve = crypto.get_elliptic_curve(_defaultCurveName)
-            _ecCurve.addECKeyToContext(context)
+            context.set_tmp_ecdh(_ecCurve)
+
         except Exception:
             logger.exception("Failed to enable elliptic curve for TLS")
         context.set_options(SSL.OP_NO_SSLv2 | SSL.OP_NO_SSLv3)

synapse/crypto/keyring.py

@@ -19,7 +19,8 @@ from synapse.api.errors import SynapseError, Codes
 from synapse.util import unwrapFirstError, logcontext
 from synapse.util.logcontext import (
     PreserveLoggingContext,
-    preserve_fn
+    preserve_fn,
+    run_in_background,
 )
 from synapse.util.metrics import Measure
 
@@ -127,7 +128,7 @@ class Keyring(object):
 
             verify_requests.append(verify_request)
 
-        preserve_fn(self._start_key_lookups)(verify_requests)
+        run_in_background(self._start_key_lookups, verify_requests)
 
         # Pass those keys to handle_key_deferred so that the json object
         # signatures can be verified
@@ -146,6 +147,7 @@ class Keyring(object):
             verify_requests (List[VerifyKeyRequest]):
         """
 
+        try:
             # create a deferred for each server we're going to look up the keys
             # for; we'll resolve them once we have completed our lookups.
             # These will be passed into wait_for_previous_lookups to block
@@ -192,6 +194,8 @@ class Keyring(object):
                 verify_request.deferred.addBoth(
                     remove_deferreds, verify_request,
                 )
+        except Exception:
+            logger.exception("Error starting key lookups")
 
     @defer.inlineCallbacks
     def wait_for_previous_lookups(self, server_names, server_to_deferred):
@@ -313,7 +317,7 @@ class Keyring(object):
                     if not verify_request.deferred.called:
                         verify_request.deferred.errback(err)
 
-        preserve_fn(do_iterations)().addErrback(on_err)
+        run_in_background(do_iterations).addErrback(on_err)
 
     @defer.inlineCallbacks
     def get_keys_from_store(self, server_name_and_key_ids):
@@ -329,8 +333,9 @@ class Keyring(object):
         """
         res = yield logcontext.make_deferred_yieldable(defer.gatherResults(
             [
-                preserve_fn(self.store.get_server_verify_keys)(
+                run_in_background(
-                    server_name, key_ids
+                    self.store.get_server_verify_keys,
+                    server_name, key_ids,
                 ).addCallback(lambda ks, server: (server, ks), server_name)
                 for server_name, key_ids in server_name_and_key_ids
             ],
@@ -352,13 +357,13 @@ class Keyring(object):
                 logger.exception(
                     "Unable to get key from %r: %s %s",
                     perspective_name,
-                    type(e).__name__, str(e.message),
+                    type(e).__name__, str(e),
                 )
                 defer.returnValue({})
 
         results = yield logcontext.make_deferred_yieldable(defer.gatherResults(
             [
-                preserve_fn(get_key)(p_name, p_keys)
+                run_in_background(get_key, p_name, p_keys)
                 for p_name, p_keys in self.perspective_servers.items()
             ],
             consumeErrors=True,
@@ -384,7 +389,7 @@ class Keyring(object):
                 logger.info(
                     "Unable to get key %r for %r directly: %s %s",
                     key_ids, server_name,
-                    type(e).__name__, str(e.message),
+                    type(e).__name__, str(e),
                 )
 
             if not keys:
@@ -398,7 +403,7 @@ class Keyring(object):
 
         results = yield logcontext.make_deferred_yieldable(defer.gatherResults(
             [
-                preserve_fn(get_key)(server_name, key_ids)
+                run_in_background(get_key, server_name, key_ids)
                 for server_name, key_ids in server_name_and_key_ids
             ],
             consumeErrors=True,
@@ -481,7 +486,8 @@ class Keyring(object):
 
         yield logcontext.make_deferred_yieldable(defer.gatherResults(
             [
-                preserve_fn(self.store_keys)(
+                run_in_background(
+                    self.store_keys,
                     server_name=server_name,
                     from_server=perspective_name,
                     verify_keys=response_keys,
@@ -539,7 +545,8 @@ class Keyring(object):
 
         yield logcontext.make_deferred_yieldable(defer.gatherResults(
             [
-                preserve_fn(self.store_keys)(
+                run_in_background(
+                    self.store_keys,
                     server_name=key_server_name,
                     from_server=server_name,
                     verify_keys=verify_keys,
@@ -615,7 +622,8 @@ class Keyring(object):
 
         yield logcontext.make_deferred_yieldable(defer.gatherResults(
             [
-                preserve_fn(self.store.store_server_keys_json)(
+                run_in_background(
+                    self.store.store_server_keys_json,
                     server_name=server_name,
                     key_id=key_id,
                     from_server=server_name,
@@ -716,7 +724,8 @@ class Keyring(object):
         # TODO(markjh): Store whether the keys have expired.
         return logcontext.make_deferred_yieldable(defer.gatherResults(
             [
-                preserve_fn(self.store.store_server_verify_key)(
+                run_in_background(
+                    self.store.store_server_verify_key,
                     server_name, server_name, key.time_added, key
                 )
                 for key_id, key in verify_keys.items()
@@ -734,7 +743,7 @@ def _handle_key_deferred(verify_request):
     except IOError as e:
         logger.warn(
             "Got IOError when downloading keys for %s: %s %s",
-            server_name, type(e).__name__, str(e.message),
+            server_name, type(e).__name__, str(e),
         )
         raise SynapseError(
             502,
@@ -744,7 +753,7 @@ def _handle_key_deferred(verify_request):
     except Exception as e:
         logger.exception(
             "Got Exception when downloading keys for %s: %s %s",
-            server_name, type(e).__name__, str(e.message),
+            server_name, type(e).__name__, str(e),
         )
         raise SynapseError(
             401,

synapse/events/__init__.py

@@ -47,14 +47,26 @@ class _EventInternalMetadata(object):
 
 
 def _event_dict_property(key):
+    # We want to be able to use hasattr with the event dict properties.
+    # However, (on python3) hasattr expects AttributeError to be raised. Hence,
+    # we need to transform the KeyError into an AttributeError
     def getter(self):
+        try:
             return self._event_dict[key]
+        except KeyError:
+            raise AttributeError(key)
 
     def setter(self, v):
+        try:
             self._event_dict[key] = v
+        except KeyError:
+            raise AttributeError(key)
 
     def delete(self):
+        try:
             del self._event_dict[key]
+        except KeyError:
+            raise AttributeError(key)
 
     return property(
         getter,

synapse/federation/federation_base.py

@@ -14,7 +14,10 @@
 # limitations under the License.
 import logging
 
-from synapse.api.errors import SynapseError
+import six
+
+from synapse.api.constants import MAX_DEPTH
+from synapse.api.errors import SynapseError, Codes
 from synapse.crypto.event_signing import check_event_content_hash
 from synapse.events import FrozenEvent
 from synapse.events.utils import prune_event
@@ -190,11 +193,23 @@ def event_from_pdu_json(pdu_json, outlier=False):
         FrozenEvent
 
     Raises:
-        SynapseError: if the pdu is missing required fields
+        SynapseError: if the pdu is missing required fields or is otherwise
+            not a valid matrix event
     """
     # we could probably enforce a bunch of other fields here (room_id, sender,
     # origin, etc etc)
-    assert_params_in_request(pdu_json, ('event_id', 'type'))
+    assert_params_in_request(pdu_json, ('event_id', 'type', 'depth'))
+
+    depth = pdu_json['depth']
+    if not isinstance(depth, six.integer_types):
+        raise SynapseError(400, "Depth %r not an intger" % (depth, ),
+                           Codes.BAD_JSON)
+
+    if depth < 0:
+        raise SynapseError(400, "Depth too small", Codes.BAD_JSON)
+    elif depth > MAX_DEPTH:
+        raise SynapseError(400, "Depth too large", Codes.BAD_JSON)
+
     event = FrozenEvent(
         pdu_json
     )

synapse/federation/federation_client.py

@@ -19,6 +19,8 @@ import itertools
 import logging
 import random
 
+from six.moves import range
+
 from twisted.internet import defer
 
 from synapse.api.constants import Membership
@@ -33,7 +35,7 @@ from synapse.federation.federation_base import (
 import synapse.metrics
 from synapse.util import logcontext, unwrapFirstError
 from synapse.util.caches.expiringcache import ExpiringCache
-from synapse.util.logcontext import make_deferred_yieldable, preserve_fn
+from synapse.util.logcontext import make_deferred_yieldable, run_in_background
 from synapse.util.logutils import log_function
 from synapse.util.retryutils import NotRetryingDestination
 
@@ -394,7 +396,7 @@ class FederationClient(FederationBase):
             seen_events = yield self.store.get_events(event_ids, allow_rejected=True)
             signed_events = seen_events.values()
         else:
-            seen_events = yield self.store.have_events(event_ids)
+            seen_events = yield self.store.have_seen_events(event_ids)
             signed_events = []
 
         failed_to_fetch = set()
@@ -413,11 +415,12 @@ class FederationClient(FederationBase):
 
         batch_size = 20
         missing_events = list(missing_events)
-        for i in xrange(0, len(missing_events), batch_size):
+        for i in range(0, len(missing_events), batch_size):
             batch = set(missing_events[i:i + batch_size])
 
             deferreds = [
-                preserve_fn(self.get_pdu)(
+                run_in_background(
+                    self.get_pdu,
                     destinations=random_server_list(),
                     event_id=e_id,
                 )

synapse/federation/federation_server.py

@@ -1,5 +1,6 @@
 # -*- coding: utf-8 -*-
 # Copyright 2015, 2016 OpenMarket Ltd
+# Copyright 2018 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.
@@ -30,9 +31,10 @@ import synapse.metrics
 from synapse.types import get_domain_from_id
 from synapse.util import async
 from synapse.util.caches.response_cache import ResponseCache
-from synapse.util.logcontext import make_deferred_yieldable, preserve_fn
 from synapse.util.logutils import log_function
 
+from six import iteritems
+
 # when processing incoming transactions, we try to handle multiple rooms in
 # parallel, up to this limit.
 TRANSACTION_CONCURRENCY_LIMIT = 10
@@ -65,7 +67,7 @@ class FederationServer(FederationBase):
 
         # We cache responses to state queries, as they take a while and often
         # come in waves.
-        self._state_resp_cache = ResponseCache(hs, timeout_ms=30000)
+        self._state_resp_cache = ResponseCache(hs, "state_resp", timeout_ms=30000)
 
     @defer.inlineCallbacks
     @log_function
@@ -212,16 +214,17 @@ class FederationServer(FederationBase):
         if not in_room:
             raise AuthError(403, "Host not in room.")
 
-        result = self._state_resp_cache.get((room_id, event_id))
+        # we grab the linearizer to protect ourselves from servers which hammer
-        if not result:
+        # us. In theory we might already have the response to this query
+        # in the cache so we could return it without waiting for the linearizer
+        # - but that's non-trivial to get right, and anyway somewhat defeats
+        # the point of the linearizer.
         with (yield self._server_linearizer.queue((origin, room_id))):
-                d = self._state_resp_cache.set(
+            resp = yield self._state_resp_cache.wrap(
                 (room_id, event_id),
-                    preserve_fn(self._on_context_state_request_compute)(room_id, event_id)
+                self._on_context_state_request_compute,
+                room_id, event_id,
             )
-                resp = yield make_deferred_yieldable(d)
-        else:
-            resp = yield make_deferred_yieldable(result)
 
         defer.returnValue((200, resp))
 
@@ -425,9 +428,9 @@ class FederationServer(FederationBase):
             "Claimed one-time-keys: %s",
             ",".join((
                 "%s for %s:%s" % (key_id, user_id, device_id)
-                for user_id, user_keys in json_result.iteritems()
+                for user_id, user_keys in iteritems(json_result)
-                for device_id, device_keys in user_keys.iteritems()
+                for device_id, device_keys in iteritems(user_keys)
-                for key_id, _ in device_keys.iteritems()
+                for key_id, _ in iteritems(device_keys)
             )),
         )
 
@@ -494,12 +497,32 @@ class FederationServer(FederationBase):
     def _handle_received_pdu(self, origin, pdu):
         """ Process a PDU received in a federation /send/ transaction.
 
+        If the event is invalid, then this method throws a FederationError.
+        (The error will then be logged and sent back to the sender (which
+        probably won't do anything with it), and other events in the
+        transaction will be processed as normal).
+
+        It is likely that we'll then receive other events which refer to
+        this rejected_event in their prev_events, etc.  When that happens,
+        we'll attempt to fetch the rejected event again, which will presumably
+        fail, so those second-generation events will also get rejected.
+
+        Eventually, we get to the point where there are more than 10 events
+        between any new events and the original rejected event. Since we
+        only try to backfill 10 events deep on received pdu, we then accept the
+        new event, possibly introducing a discontinuity in the DAG, with new
+        forward extremities, so normal service is approximately returned,
+        until we try to backfill across the discontinuity.
+
         Args:
             origin (str): server which sent the pdu
             pdu (FrozenEvent): received pdu
 
         Returns (Deferred): completes with None
-        Raises: FederationError if the signatures / hash do not match
+
+        Raises: FederationError if the signatures / hash do not match, or
+            if the event was unacceptable for any other reason (eg, too large,
+            too many prev_events, couldn't find the prev_events)
         """
         # check that it's actually being sent from a valid destination to
         # workaround bug #1753 in 0.18.5 and 0.18.6

synapse/federation/send_queue.py

@@ -40,6 +40,8 @@ from collections import namedtuple
 
 import logging
 
+from six import itervalues, iteritems
+
 logger = logging.getLogger(__name__)
 
 
@@ -122,7 +124,7 @@ class FederationRemoteSendQueue(object):
 
             user_ids = set(
                 user_id
-                for uids in self.presence_changed.itervalues()
+                for uids in itervalues(self.presence_changed)
                 for user_id in uids
             )
 
@@ -276,7 +278,7 @@ class FederationRemoteSendQueue(object):
         # stream position.
         keyed_edus = {self.keyed_edu_changed[k]: k for k in keys[i:j]}
 
-        for ((destination, edu_key), pos) in keyed_edus.iteritems():
+        for ((destination, edu_key), pos) in iteritems(keyed_edus):
             rows.append((pos, KeyedEduRow(
                 key=edu_key,
                 edu=self.keyed_edu[(destination, edu_key)],
@@ -309,7 +311,7 @@ class FederationRemoteSendQueue(object):
         j = keys.bisect_right(to_token) + 1
         device_messages = {self.device_messages[k]: k for k in keys[i:j]}
 
-        for (destination, pos) in device_messages.iteritems():
+        for (destination, pos) in iteritems(device_messages):
             rows.append((pos, DeviceRow(
                 destination=destination,
             )))
@@ -528,19 +530,19 @@ def process_rows_for_federation(transaction_queue, rows):
     if buff.presence:
         transaction_queue.send_presence(buff.presence)
 
-    for destination, edu_map in buff.keyed_edus.iteritems():
+    for destination, edu_map in iteritems(buff.keyed_edus):
         for key, edu in edu_map.items():
             transaction_queue.send_edu(
                 edu.destination, edu.edu_type, edu.content, key=key,
             )
 
-    for destination, edu_list in buff.edus.iteritems():
+    for destination, edu_list in iteritems(buff.edus):
         for edu in edu_list:
             transaction_queue.send_edu(
                 edu.destination, edu.edu_type, edu.content, key=None,
             )
 
-    for destination, failure_list in buff.failures.iteritems():
+    for destination, failure_list in iteritems(buff.failures):
         for failure in failure_list:
             transaction_queue.send_failure(destination, failure)
 

synapse/federation/transaction_queue.py

@@ -169,7 +169,7 @@ class TransactionQueue(object):
             while True:
                 last_token = yield self.store.get_federation_out_pos("events")
                 next_token, events = yield self.store.get_all_new_events_stream(
-                    last_token, self._last_poked_id, limit=20,
+                    last_token, self._last_poked_id, limit=100,
                 )
 
                 logger.debug("Handling %s -> %s", last_token, next_token)
@@ -177,13 +177,15 @@ class TransactionQueue(object):
                 if not events and next_token >= self._last_poked_id:
                     break
 
-                for event in events:
+                @defer.inlineCallbacks
+                def handle_event(event):
                     # Only send events for this server.
                     send_on_behalf_of = event.internal_metadata.get_send_on_behalf_of()
                     is_mine = self.is_mine_id(event.event_id)
                     if not is_mine and send_on_behalf_of is None:
-                        continue
+                        return
 
+                    try:
                         # Get the state from before the event.
                         # We need to make sure that this is the state from before
                         # the event and not from after it.
@@ -195,6 +197,13 @@ class TransactionQueue(object):
                                 prev_id for prev_id, _ in event.prev_events
                             ],
                         )
+                    except Exception:
+                        logger.exception(
+                            "Failed to calculate hosts in room for event: %s",
+                            event.event_id,
+                        )
+                        return
+
                     destinations = set(destinations)
 
                     if send_on_behalf_of is not None:
@@ -207,12 +216,44 @@ class TransactionQueue(object):
 
                     self._send_pdu(event, destinations)
 
-                events_processed_counter.inc_by(len(events))
+                @defer.inlineCallbacks
+                def handle_room_events(events):
+                    for event in events:
+                        yield handle_event(event)
+
+                events_by_room = {}
+                for event in events:
+                    events_by_room.setdefault(event.room_id, []).append(event)
+
+                yield logcontext.make_deferred_yieldable(defer.gatherResults(
+                    [
+                        logcontext.run_in_background(handle_room_events, evs)
+                        for evs in events_by_room.itervalues()
+                    ],
+                    consumeErrors=True
+                ))
 
                 yield self.store.update_federation_out_pos(
                     "events", next_token
                 )
 
+                if events:
+                    now = self.clock.time_msec()
+                    ts = yield self.store.get_received_ts(events[-1].event_id)
+
+                    synapse.metrics.event_processing_lag.set(
+                        now - ts, "federation_sender",
+                    )
+                    synapse.metrics.event_processing_last_ts.set(
+                        ts, "federation_sender",
+                    )
+
+                events_processed_counter.inc_by(len(events))
+
+                synapse.metrics.event_processing_positions.set(
+                    next_token, "federation_sender",
+                )
+
         finally:
             self._is_processing = False
 
@@ -282,6 +323,8 @@ class TransactionQueue(object):
                     break
 
                 yield self._process_presence_inner(states_map.values())
+        except Exception:
+            logger.exception("Error sending presence states to servers")
         finally:
             self._processing_pending_presence = False
 

synapse/federation/transport/client.py

@@ -1,5 +1,6 @@
 # -*- coding: utf-8 -*-
 # Copyright 2014-2016 OpenMarket Ltd
+# Copyright 2018 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.
@@ -20,6 +21,7 @@ from synapse.api.urls import FEDERATION_PREFIX as PREFIX
 from synapse.util.logutils import log_function
 
 import logging
+import urllib
 
 
 logger = logging.getLogger(__name__)
@@ -49,7 +51,7 @@ class TransportLayerClient(object):
         logger.debug("get_room_state dest=%s, room=%s",
                      destination, room_id)
 
-        path = PREFIX + "/state/%s/" % room_id
+        path = _create_path(PREFIX, "/state/%s/", room_id)
         return self.client.get_json(
             destination, path=path, args={"event_id": event_id},
         )
@@ -71,7 +73,7 @@ class TransportLayerClient(object):
         logger.debug("get_room_state_ids dest=%s, room=%s",
                      destination, room_id)
 
-        path = PREFIX + "/state_ids/%s/" % room_id
+        path = _create_path(PREFIX, "/state_ids/%s/", room_id)
         return self.client.get_json(
             destination, path=path, args={"event_id": event_id},
         )
@@ -93,7 +95,7 @@ class TransportLayerClient(object):
         logger.debug("get_pdu dest=%s, event_id=%s",
                      destination, event_id)
 
-        path = PREFIX + "/event/%s/" % (event_id, )
+        path = _create_path(PREFIX, "/event/%s/", event_id)
         return self.client.get_json(destination, path=path, timeout=timeout)
 
     @log_function
@@ -119,7 +121,7 @@ class TransportLayerClient(object):
             # TODO: raise?
             return
 
-        path = PREFIX + "/backfill/%s/" % (room_id,)
+        path = _create_path(PREFIX, "/backfill/%s/", room_id)
 
         args = {
             "v": event_tuples,
@@ -157,9 +159,11 @@ class TransportLayerClient(object):
         # generated by the json_data_callback.
         json_data = transaction.get_dict()
 
+        path = _create_path(PREFIX, "/send/%s/", transaction.transaction_id)
+
         response = yield self.client.put_json(
             transaction.destination,
-            path=PREFIX + "/send/%s/" % transaction.transaction_id,
+            path=path,
             data=json_data,
             json_data_callback=json_data_callback,
             long_retries=True,
@@ -177,7 +181,7 @@ class TransportLayerClient(object):
     @log_function
     def make_query(self, destination, query_type, args, retry_on_dns_fail,
                    ignore_backoff=False):
-        path = PREFIX + "/query/%s" % query_type
+        path = _create_path(PREFIX, "/query/%s", query_type)
 
         content = yield self.client.get_json(
             destination=destination,
@@ -222,7 +226,7 @@ class TransportLayerClient(object):
                 "make_membership_event called with membership='%s', must be one of %s" %
                 (membership, ",".join(valid_memberships))
             )
-        path = PREFIX + "/make_%s/%s/%s" % (membership, room_id, user_id)
+        path = _create_path(PREFIX, "/make_%s/%s/%s", membership, room_id, user_id)
 
         ignore_backoff = False
         retry_on_dns_fail = False
@@ -248,7 +252,7 @@ class TransportLayerClient(object):
     @defer.inlineCallbacks
     @log_function
     def send_join(self, destination, room_id, event_id, content):
-        path = PREFIX + "/send_join/%s/%s" % (room_id, event_id)
+        path = _create_path(PREFIX, "/send_join/%s/%s", room_id, event_id)
 
         response = yield self.client.put_json(
             destination=destination,
@@ -261,7 +265,7 @@ class TransportLayerClient(object):
     @defer.inlineCallbacks
     @log_function
     def send_leave(self, destination, room_id, event_id, content):
-        path = PREFIX + "/send_leave/%s/%s" % (room_id, event_id)
+        path = _create_path(PREFIX, "/send_leave/%s/%s", room_id, event_id)
 
         response = yield self.client.put_json(
             destination=destination,
@@ -280,7 +284,7 @@ class TransportLayerClient(object):
     @defer.inlineCallbacks
     @log_function
     def send_invite(self, destination, room_id, event_id, content):
-        path = PREFIX + "/invite/%s/%s" % (room_id, event_id)
+        path = _create_path(PREFIX, "/invite/%s/%s", room_id, event_id)
 
         response = yield self.client.put_json(
             destination=destination,
@@ -322,7 +326,7 @@ class TransportLayerClient(object):
     @defer.inlineCallbacks
     @log_function
     def exchange_third_party_invite(self, destination, room_id, event_dict):
-        path = PREFIX + "/exchange_third_party_invite/%s" % (room_id,)
+        path = _create_path(PREFIX, "/exchange_third_party_invite/%s", room_id,)
 
         response = yield self.client.put_json(
             destination=destination,
@@ -335,7 +339,7 @@ class TransportLayerClient(object):
     @defer.inlineCallbacks
     @log_function
     def get_event_auth(self, destination, room_id, event_id):
-        path = PREFIX + "/event_auth/%s/%s" % (room_id, event_id)
+        path = _create_path(PREFIX, "/event_auth/%s/%s", room_id, event_id)
 
         content = yield self.client.get_json(
             destination=destination,
@@ -347,7 +351,7 @@ class TransportLayerClient(object):
     @defer.inlineCallbacks
     @log_function
     def send_query_auth(self, destination, room_id, event_id, content):
-        path = PREFIX + "/query_auth/%s/%s" % (room_id, event_id)
+        path = _create_path(PREFIX, "/query_auth/%s/%s", room_id, event_id)
 
         content = yield self.client.post_json(
             destination=destination,
@@ -409,7 +413,7 @@ class TransportLayerClient(object):
         Returns:
             A dict containg the device keys.
         """
-        path = PREFIX + "/user/devices/" + user_id
+        path = _create_path(PREFIX, "/user/devices/%s", user_id)
 
         content = yield self.client.get_json(
             destination=destination,
@@ -459,7 +463,7 @@ class TransportLayerClient(object):
     @log_function
     def get_missing_events(self, destination, room_id, earliest_events,
                            latest_events, limit, min_depth, timeout):
-        path = PREFIX + "/get_missing_events/%s" % (room_id,)
+        path = _create_path(PREFIX, "/get_missing_events/%s", room_id,)
 
         content = yield self.client.post_json(
             destination=destination,
@@ -479,7 +483,7 @@ class TransportLayerClient(object):
     def get_group_profile(self, destination, group_id, requester_user_id):
         """Get a group profile
         """
-        path = PREFIX + "/groups/%s/profile" % (group_id,)
+        path = _create_path(PREFIX, "/groups/%s/profile", group_id,)
 
         return self.client.get_json(
             destination=destination,
@@ -498,7 +502,7 @@ class TransportLayerClient(object):
             requester_user_id (str)
             content (dict): The new profile of the group
         """
-        path = PREFIX + "/groups/%s/profile" % (group_id,)
+        path = _create_path(PREFIX, "/groups/%s/profile", group_id,)
 
         return self.client.post_json(
             destination=destination,
@@ -512,7 +516,7 @@ class TransportLayerClient(object):
     def get_group_summary(self, destination, group_id, requester_user_id):
         """Get a group summary
         """
-        path = PREFIX + "/groups/%s/summary" % (group_id,)
+        path = _create_path(PREFIX, "/groups/%s/summary", group_id,)
 
         return self.client.get_json(
             destination=destination,
@@ -525,7 +529,7 @@ class TransportLayerClient(object):
     def get_rooms_in_group(self, destination, group_id, requester_user_id):
         """Get all rooms in a group
         """
-        path = PREFIX + "/groups/%s/rooms" % (group_id,)
+        path = _create_path(PREFIX, "/groups/%s/rooms", group_id,)
 
         return self.client.get_json(
             destination=destination,
@@ -538,7 +542,7 @@ class TransportLayerClient(object):
                           content):
         """Add a room to a group
         """
-        path = PREFIX + "/groups/%s/room/%s" % (group_id, room_id,)
+        path = _create_path(PREFIX, "/groups/%s/room/%s", group_id, room_id,)
 
         return self.client.post_json(
             destination=destination,
@@ -552,7 +556,10 @@ class TransportLayerClient(object):
                              config_key, content):
         """Update room in group
         """
-        path = PREFIX + "/groups/%s/room/%s/config/%s" % (group_id, room_id, config_key,)
+        path = _create_path(
+            PREFIX, "/groups/%s/room/%s/config/%s",
+            group_id, room_id, config_key,
+        )
 
         return self.client.post_json(
             destination=destination,
@@ -565,7 +572,7 @@ class TransportLayerClient(object):
     def remove_room_from_group(self, destination, group_id, requester_user_id, room_id):
         """Remove a room from a group
         """
-        path = PREFIX + "/groups/%s/room/%s" % (group_id, room_id,)
+        path = _create_path(PREFIX, "/groups/%s/room/%s", group_id, room_id,)
 
         return self.client.delete_json(
             destination=destination,
@@ -578,7 +585,7 @@ class TransportLayerClient(object):
     def get_users_in_group(self, destination, group_id, requester_user_id):
         """Get users in a group
         """
-        path = PREFIX + "/groups/%s/users" % (group_id,)
+        path = _create_path(PREFIX, "/groups/%s/users", group_id,)
 
         return self.client.get_json(
             destination=destination,
@@ -591,7 +598,7 @@ class TransportLayerClient(object):
     def get_invited_users_in_group(self, destination, group_id, requester_user_id):
         """Get users that have been invited to a group
         """
-        path = PREFIX + "/groups/%s/invited_users" % (group_id,)
+        path = _create_path(PREFIX, "/groups/%s/invited_users", group_id,)
 
         return self.client.get_json(
             destination=destination,
@@ -604,7 +611,23 @@ class TransportLayerClient(object):
     def accept_group_invite(self, destination, group_id, user_id, content):
         """Accept a group invite
         """
-        path = PREFIX + "/groups/%s/users/%s/accept_invite" % (group_id, user_id)
+        path = _create_path(
+            PREFIX, "/groups/%s/users/%s/accept_invite",
+            group_id, user_id,
+        )
+
+        return self.client.post_json(
+            destination=destination,
+            path=path,
+            data=content,
+            ignore_backoff=True,
+        )
+
+    @log_function
+    def join_group(self, destination, group_id, user_id, content):
+        """Attempts to join a group
+        """
+        path = _create_path(PREFIX, "/groups/%s/users/%s/join", group_id, user_id)
 
         return self.client.post_json(
             destination=destination,
@@ -617,7 +640,7 @@ class TransportLayerClient(object):
     def invite_to_group(self, destination, group_id, user_id, requester_user_id, content):
         """Invite a user to a group
         """
-        path = PREFIX + "/groups/%s/users/%s/invite" % (group_id, user_id)
+        path = _create_path(PREFIX, "/groups/%s/users/%s/invite", group_id, user_id)
 
         return self.client.post_json(
             destination=destination,
@@ -633,7 +656,7 @@ class TransportLayerClient(object):
         invited.
         """
 
-        path = PREFIX + "/groups/local/%s/users/%s/invite" % (group_id, user_id)
+        path = _create_path(PREFIX, "/groups/local/%s/users/%s/invite", group_id, user_id)
 
         return self.client.post_json(
             destination=destination,
@@ -647,7 +670,7 @@ class TransportLayerClient(object):
                                user_id, content):
         """Remove a user fron a group
         """
-        path = PREFIX + "/groups/%s/users/%s/remove" % (group_id, user_id)
+        path = _create_path(PREFIX, "/groups/%s/users/%s/remove", group_id, user_id)
 
         return self.client.post_json(
             destination=destination,
@@ -664,7 +687,7 @@ class TransportLayerClient(object):
         kicked from the group.
         """
 
-        path = PREFIX + "/groups/local/%s/users/%s/remove" % (group_id, user_id)
+        path = _create_path(PREFIX, "/groups/local/%s/users/%s/remove", group_id, user_id)
 
         return self.client.post_json(
             destination=destination,
@@ -679,7 +702,7 @@ class TransportLayerClient(object):
         the attestations
         """
 
-        path = PREFIX + "/groups/%s/renew_attestation/%s" % (group_id, user_id)
+        path = _create_path(PREFIX, "/groups/%s/renew_attestation/%s", group_id, user_id)
 
         return self.client.post_json(
             destination=destination,
@@ -694,11 +717,12 @@ class TransportLayerClient(object):
         """Update a room entry in a group summary
         """
         if category_id:
-            path = PREFIX + "/groups/%s/summary/categories/%s/rooms/%s" % (
+            path = _create_path(
+                PREFIX, "/groups/%s/summary/categories/%s/rooms/%s",
                 group_id, category_id, room_id,
             )
         else:
-            path = PREFIX + "/groups/%s/summary/rooms/%s" % (group_id, room_id,)
+            path = _create_path(PREFIX, "/groups/%s/summary/rooms/%s", group_id, room_id,)
 
         return self.client.post_json(
             destination=destination,
@@ -714,11 +738,12 @@ class TransportLayerClient(object):
         """Delete a room entry in a group summary
         """
         if category_id:
-            path = PREFIX + "/groups/%s/summary/categories/%s/rooms/%s" % (
+            path = _create_path(
+                PREFIX + "/groups/%s/summary/categories/%s/rooms/%s",
                 group_id, category_id, room_id,
             )
         else:
-            path = PREFIX + "/groups/%s/summary/rooms/%s" % (group_id, room_id,)
+            path = _create_path(PREFIX, "/groups/%s/summary/rooms/%s", group_id, room_id,)
 
         return self.client.delete_json(
             destination=destination,
@@ -731,7 +756,7 @@ class TransportLayerClient(object):
     def get_group_categories(self, destination, group_id, requester_user_id):
         """Get all categories in a group
         """
-        path = PREFIX + "/groups/%s/categories" % (group_id,)
+        path = _create_path(PREFIX, "/groups/%s/categories", group_id,)
 
         return self.client.get_json(
             destination=destination,
@@ -744,7 +769,7 @@ class TransportLayerClient(object):
     def get_group_category(self, destination, group_id, requester_user_id, category_id):
         """Get category info in a group
         """
-        path = PREFIX + "/groups/%s/categories/%s" % (group_id, category_id,)
+        path = _create_path(PREFIX, "/groups/%s/categories/%s", group_id, category_id,)
 
         return self.client.get_json(
             destination=destination,
@@ -758,7 +783,7 @@ class TransportLayerClient(object):
                               content):
         """Update a category in a group
         """
-        path = PREFIX + "/groups/%s/categories/%s" % (group_id, category_id,)
+        path = _create_path(PREFIX, "/groups/%s/categories/%s", group_id, category_id,)
 
         return self.client.post_json(
             destination=destination,
@@ -773,7 +798,7 @@ class TransportLayerClient(object):
                               category_id):
         """Delete a category in a group
         """
-        path = PREFIX + "/groups/%s/categories/%s" % (group_id, category_id,)
+        path = _create_path(PREFIX, "/groups/%s/categories/%s", group_id, category_id,)
 
         return self.client.delete_json(
             destination=destination,
@@ -786,7 +811,7 @@ class TransportLayerClient(object):
     def get_group_roles(self, destination, group_id, requester_user_id):
         """Get all roles in a group
         """
-        path = PREFIX + "/groups/%s/roles" % (group_id,)
+        path = _create_path(PREFIX, "/groups/%s/roles", group_id,)
 
         return self.client.get_json(
             destination=destination,
@@ -799,7 +824,7 @@ class TransportLayerClient(object):
     def get_group_role(self, destination, group_id, requester_user_id, role_id):
         """Get a roles info
         """
-        path = PREFIX + "/groups/%s/roles/%s" % (group_id, role_id,)
+        path = _create_path(PREFIX, "/groups/%s/roles/%s", group_id, role_id,)
 
         return self.client.get_json(
             destination=destination,
@@ -813,7 +838,7 @@ class TransportLayerClient(object):
                           content):
         """Update a role in a group
         """
-        path = PREFIX + "/groups/%s/roles/%s" % (group_id, role_id,)
+        path = _create_path(PREFIX, "/groups/%s/roles/%s", group_id, role_id,)
 
         return self.client.post_json(
             destination=destination,
@@ -827,7 +852,7 @@ class TransportLayerClient(object):
     def delete_group_role(self, destination, group_id, requester_user_id, role_id):
         """Delete a role in a group
         """
-        path = PREFIX + "/groups/%s/roles/%s" % (group_id, role_id,)
+        path = _create_path(PREFIX, "/groups/%s/roles/%s", group_id, role_id,)
 
         return self.client.delete_json(
             destination=destination,
@@ -842,11 +867,12 @@ class TransportLayerClient(object):
         """Update a users entry in a group
         """
         if role_id:
-            path = PREFIX + "/groups/%s/summary/roles/%s/users/%s" % (
+            path = _create_path(
+                PREFIX, "/groups/%s/summary/roles/%s/users/%s",
                 group_id, role_id, user_id,
             )
         else:
-            path = PREFIX + "/groups/%s/summary/users/%s" % (group_id, user_id,)
+            path = _create_path(PREFIX, "/groups/%s/summary/users/%s", group_id, user_id,)
 
         return self.client.post_json(
             destination=destination,
@@ -856,17 +882,33 @@ class TransportLayerClient(object):
             ignore_backoff=True,
         )
 
+    @log_function
+    def set_group_join_policy(self, destination, group_id, requester_user_id,
+                              content):
+        """Sets the join policy for a group
+        """
+        path = _create_path(PREFIX, "/groups/%s/settings/m.join_policy", group_id,)
+
+        return self.client.put_json(
+            destination=destination,
+            path=path,
+            args={"requester_user_id": requester_user_id},
+            data=content,
+            ignore_backoff=True,
+        )
+
     @log_function
     def delete_group_summary_user(self, destination, group_id, requester_user_id,
                                   user_id, role_id):
         """Delete a users entry in a group
         """
         if role_id:
-            path = PREFIX + "/groups/%s/summary/roles/%s/users/%s" % (
+            path = _create_path(
+                PREFIX, "/groups/%s/summary/roles/%s/users/%s",
                 group_id, role_id, user_id,
             )
         else:
-            path = PREFIX + "/groups/%s/summary/users/%s" % (group_id, user_id,)
+            path = _create_path(PREFIX, "/groups/%s/summary/users/%s", group_id, user_id,)
 
         return self.client.delete_json(
             destination=destination,
@@ -889,3 +931,22 @@ class TransportLayerClient(object):
             data=content,
             ignore_backoff=True,
         )
+
+
+def _create_path(prefix, path, *args):
+    """Creates a path from the prefix, path template and args. Ensures that
+    all args are url encoded.
+
+    Example:
+
+        _create_path(PREFIX, "/event/%s/", event_id)
+
+    Args:
+        prefix (str)
+        path (str): String template for the path
+        args: ([str]): Args to insert into path. Each arg will be url encoded
+
+    Returns:
+        str
+    """
+    return prefix + path % tuple(urllib.quote(arg, "") for arg in args)

synapse/federation/transport/server.py

@@ -1,5 +1,6 @@
 # -*- coding: utf-8 -*-
 # Copyright 2014-2016 OpenMarket Ltd
+# Copyright 2018 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.
@@ -24,7 +25,7 @@ from synapse.http.servlet import (
 )
 from synapse.util.ratelimitutils import FederationRateLimiter
 from synapse.util.versionstring import get_version_string
-from synapse.util.logcontext import preserve_fn
+from synapse.util.logcontext import run_in_background
 from synapse.types import ThirdPartyInstanceID, get_domain_from_id
 
 import functools
@@ -93,12 +94,6 @@ class Authenticator(object):
             "signatures": {},
         }
 
-        if (
-            self.federation_domain_whitelist is not None and
-            self.server_name not in self.federation_domain_whitelist
-        ):
-            raise FederationDeniedError(self.server_name)
-
         if content is not None:
             json_request["content"] = content
 
@@ -137,6 +132,12 @@ class Authenticator(object):
                 json_request["origin"] = origin
                 json_request["signatures"].setdefault(origin, {})[key] = sig
 
+        if (
+            self.federation_domain_whitelist is not None and
+            origin not in self.federation_domain_whitelist
+        ):
+            raise FederationDeniedError(origin)
+
         if not json_request["signatures"]:
             raise NoAuthenticationError(
                 401, "Missing Authorization headers", Codes.UNAUTHORIZED,
@@ -151,11 +152,18 @@ class Authenticator(object):
         # alive
         retry_timings = yield self.store.get_destination_retry_timings(origin)
         if retry_timings and retry_timings["retry_last_ts"]:
-            logger.info("Marking origin %r as up", origin)
+            run_in_background(self._reset_retry_timings, origin)
-            preserve_fn(self.store.set_destination_retry_timings)(origin, 0, 0)
 
         defer.returnValue(origin)
 
+    @defer.inlineCallbacks
+    def _reset_retry_timings(self, origin):
+        try:
+            logger.info("Marking origin %r as up", origin)
+            yield self.store.set_destination_retry_timings(origin, 0, 0)
+        except Exception:
+            logger.exception("Error resetting retry timings on %s", origin)
+
 
 class BaseFederationServlet(object):
     REQUIRE_AUTH = True
@@ -802,6 +810,23 @@ class FederationGroupsAcceptInviteServlet(BaseFederationServlet):
         defer.returnValue((200, new_content))
 
 
+class FederationGroupsJoinServlet(BaseFederationServlet):
+    """Attempt to join a group
+    """
+    PATH = "/groups/(?P<group_id>[^/]*)/users/(?P<user_id>[^/]*)/join$"
+
+    @defer.inlineCallbacks
+    def on_POST(self, origin, content, query, group_id, user_id):
+        if get_domain_from_id(user_id) != origin:
+            raise SynapseError(403, "user_id doesn't match origin")
+
+        new_content = yield self.handler.join_group(
+            group_id, user_id, content,
+        )
+
+        defer.returnValue((200, new_content))
+
+
 class FederationGroupsRemoveUserServlet(BaseFederationServlet):
     """Leave or kick a user from the group
     """
@@ -1124,6 +1149,24 @@ class FederationGroupsBulkPublicisedServlet(BaseFederationServlet):
         defer.returnValue((200, resp))
 
 
+class FederationGroupsSettingJoinPolicyServlet(BaseFederationServlet):
+    """Sets whether a group is joinable without an invite or knock
+    """
+    PATH = "/groups/(?P<group_id>[^/]*)/settings/m.join_policy$"
+
+    @defer.inlineCallbacks
+    def on_PUT(self, origin, content, query, group_id):
+        requester_user_id = parse_string_from_args(query, "requester_user_id")
+        if get_domain_from_id(requester_user_id) != origin:
+            raise SynapseError(403, "requester_user_id doesn't match origin")
+
+        new_content = yield self.handler.set_group_join_policy(
+            group_id, requester_user_id, content
+        )
+
+        defer.returnValue((200, new_content))
+
+
 FEDERATION_SERVLET_CLASSES = (
     FederationSendServlet,
     FederationPullServlet,
@@ -1163,6 +1206,7 @@ GROUP_SERVER_SERVLET_CLASSES = (
     FederationGroupsInvitedUsersServlet,
     FederationGroupsInviteServlet,
     FederationGroupsAcceptInviteServlet,
+    FederationGroupsJoinServlet,
     FederationGroupsRemoveUserServlet,
     FederationGroupsSummaryRoomsServlet,
     FederationGroupsCategoriesServlet,
@@ -1172,6 +1216,7 @@ GROUP_SERVER_SERVLET_CLASSES = (
     FederationGroupsSummaryUsersServlet,
     FederationGroupsAddRoomsServlet,
     FederationGroupsAddRoomsConfigServlet,
+    FederationGroupsSettingJoinPolicyServlet,
 )
 
 

synapse/groups/attestations.py

@@ -42,7 +42,7 @@ from twisted.internet import defer
 
 from synapse.api.errors import SynapseError
 from synapse.types import get_domain_from_id
-from synapse.util.logcontext import preserve_fn
+from synapse.util.logcontext import run_in_background
 
 from signedjson.sign import sign_json
 
@@ -165,6 +165,7 @@ class GroupAttestionRenewer(object):
 
         @defer.inlineCallbacks
         def _renew_attestation(group_id, user_id):
+            try:
                 if not self.is_mine_id(group_id):
                     destination = get_domain_from_id(group_id)
                 elif not self.is_mine_id(user_id):
@@ -187,9 +188,12 @@ class GroupAttestionRenewer(object):
                 yield self.store.update_attestation_renewal(
                     group_id, user_id, attestation
                 )
+            except Exception:
+                logger.exception("Error renewing attestation of %r in %r",
+                                 user_id, group_id)
 
         for row in rows:
             group_id = row["group_id"]
             user_id = row["user_id"]
 
-            preserve_fn(_renew_attestation)(group_id, user_id)
+            run_in_background(_renew_attestation, group_id, user_id)

synapse/groups/groups_server.py

@@ -1,5 +1,6 @@
 # -*- coding: utf-8 -*-
 # Copyright 2017 Vector Creations Ltd
+# Copyright 2018 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.
@@ -205,6 +206,28 @@ class GroupsServerHandler(object):
 
         defer.returnValue({})
 
+    @defer.inlineCallbacks
+    def set_group_join_policy(self, group_id, requester_user_id, content):
+        """Sets the group join policy.
+
+        Currently supported policies are:
+         - "invite": an invite must be received and accepted in order to join.
+         - "open": anyone can join.
+        """
+        yield self.check_group_is_ours(
+            group_id, requester_user_id, and_exists=True, and_is_admin=requester_user_id
+        )
+
+        join_policy = _parse_join_policy_from_contents(content)
+        if join_policy is None:
+            raise SynapseError(
+                400, "No value specified for 'm.join_policy'"
+            )
+
+        yield self.store.set_group_join_policy(group_id, join_policy=join_policy)
+
+        defer.returnValue({})
+
     @defer.inlineCallbacks
     def get_group_categories(self, group_id, requester_user_id):
         """Get all categories in a group (as seen by user)
@@ -381,9 +404,16 @@ class GroupsServerHandler(object):
 
         yield self.check_group_is_ours(group_id, requester_user_id)
 
-        group_description = yield self.store.get_group(group_id)
+        group = yield self.store.get_group(group_id)
+
+        if group:
+            cols = [
+                "name", "short_description", "long_description",
+                "avatar_url", "is_public",
+            ]
+            group_description = {key: group[key] for key in cols}
+            group_description["is_openly_joinable"] = group["join_policy"] == "open"
 
-        if group_description:
             defer.returnValue(group_description)
         else:
             raise SynapseError(404, "Unknown group")
@@ -654,6 +684,40 @@ class GroupsServerHandler(object):
         else:
             raise SynapseError(502, "Unknown state returned by HS")
 
+    @defer.inlineCallbacks
+    def _add_user(self, group_id, user_id, content):
+        """Add a user to a group based on a content dict.
+
+        See accept_invite, join_group.
+        """
+        if not self.hs.is_mine_id(user_id):
+            local_attestation = self.attestations.create_attestation(
+                group_id, user_id,
+            )
+
+            remote_attestation = content["attestation"]
+
+            yield self.attestations.verify_attestation(
+                remote_attestation,
+                user_id=user_id,
+                group_id=group_id,
+            )
+        else:
+            local_attestation = None
+            remote_attestation = None
+
+        is_public = _parse_visibility_from_contents(content)
+
+        yield self.store.add_user_to_group(
+            group_id, user_id,
+            is_admin=False,
+            is_public=is_public,
+            local_attestation=local_attestation,
+            remote_attestation=remote_attestation,
+        )
+
+        defer.returnValue(local_attestation)
+
     @defer.inlineCallbacks
     def accept_invite(self, group_id, requester_user_id, content):
         """User tries to accept an invite to the group.
@@ -670,30 +734,27 @@ class GroupsServerHandler(object):
         if not is_invited:
             raise SynapseError(403, "User not invited to group")
 
-        if not self.hs.is_mine_id(requester_user_id):
+        local_attestation = yield self._add_user(group_id, requester_user_id, content)
-            local_attestation = self.attestations.create_attestation(
-                group_id, requester_user_id,
-            )
-            remote_attestation = content["attestation"]
 
-            yield self.attestations.verify_attestation(
+        defer.returnValue({
-                remote_attestation,
+            "state": "join",
-                user_id=requester_user_id,
+            "attestation": local_attestation,
-                group_id=group_id,
+        })
-            )
-        else:
-            local_attestation = None
-            remote_attestation = None
 
-        is_public = _parse_visibility_from_contents(content)
+    @defer.inlineCallbacks
+    def join_group(self, group_id, requester_user_id, content):
+        """User tries to join the group.
 
-        yield self.store.add_user_to_group(
+        This will error if the group requires an invite/knock to join
-            group_id, requester_user_id,
+        """
-            is_admin=False,
+
-            is_public=is_public,
+        group_info = yield self.check_group_is_ours(
-            local_attestation=local_attestation,
+            group_id, requester_user_id, and_exists=True
-            remote_attestation=remote_attestation,
         )
+        if group_info['join_policy'] != "open":
+            raise SynapseError(403, "Group is not publicly joinable")
+
+        local_attestation = yield self._add_user(group_id, requester_user_id, content)
 
         defer.returnValue({
             "state": "join",
@@ -835,6 +896,31 @@ class GroupsServerHandler(object):
         })
 
 
+def _parse_join_policy_from_contents(content):
+    """Given a content for a request, return the specified join policy or None
+    """
+
+    join_policy_dict = content.get("m.join_policy")
+    if join_policy_dict:
+        return _parse_join_policy_dict(join_policy_dict)
+    else:
+        return None
+
+
+def _parse_join_policy_dict(join_policy_dict):
+    """Given a dict for the "m.join_policy" config return the join policy specified
+    """
+    join_policy_type = join_policy_dict.get("type")
+    if not join_policy_type:
+        return "invite"
+
+    if join_policy_type not in ("invite", "open"):
+        raise SynapseError(
+            400, "Synapse only supports 'invite'/'open' join rule"
+        )
+    return join_policy_type
+
+
 def _parse_visibility_from_contents(content):
     """Given a content for a request parse out whether the entity should be
     public or not

synapse/handlers/appservice.py

@@ -18,7 +18,9 @@ from twisted.internet import defer
 import synapse
 from synapse.api.constants import EventTypes
 from synapse.util.metrics import Measure
-from synapse.util.logcontext import make_deferred_yieldable, preserve_fn
+from synapse.util.logcontext import (
+    make_deferred_yieldable, run_in_background,
+)
 
 import logging
 
@@ -84,11 +86,16 @@ class ApplicationServicesHandler(object):
                     if not events:
                         break
 
+                    events_by_room = {}
                     for event in events:
+                        events_by_room.setdefault(event.room_id, []).append(event)
+
+                    @defer.inlineCallbacks
+                    def handle_event(event):
                         # Gather interested services
                         services = yield self._get_services_for_event(event)
                         if len(services) == 0:
-                            continue  # no services need notifying
+                            return  # no services need notifying
 
                         # Do we know this user exists? If not, poke the user
                         # query API for all services which match that user regex.
@@ -104,13 +111,35 @@ class ApplicationServicesHandler(object):
 
                         # Fork off pushes to these services
                         for service in services:
-                            preserve_fn(self.scheduler.submit_event_for_as)(
+                            self.scheduler.submit_event_for_as(service, event)
-                                service, event
+
+                    @defer.inlineCallbacks
+                    def handle_room_events(events):
+                        for event in events:
+                            yield handle_event(event)
+
+                    yield make_deferred_yieldable(defer.gatherResults([
+                        run_in_background(handle_room_events, evs)
+                        for evs in events_by_room.itervalues()
+                    ], consumeErrors=True))
+
+                    yield self.store.set_appservice_last_pos(upper_bound)
+
+                    now = self.clock.time_msec()
+                    ts = yield self.store.get_received_ts(events[-1].event_id)
+
+                    synapse.metrics.event_processing_positions.set(
+                        upper_bound, "appservice_sender",
                     )
 
                     events_processed_counter.inc_by(len(events))
 
-                    yield self.store.set_appservice_last_pos(upper_bound)
+                    synapse.metrics.event_processing_lag.set(
+                        now - ts, "appservice_sender",
+                    )
+                    synapse.metrics.event_processing_last_ts.set(
+                        ts, "appservice_sender",
+                    )
             finally:
                 self.is_processing = False
 
@@ -167,7 +196,10 @@ class ApplicationServicesHandler(object):
         services = yield self._get_services_for_3pn(protocol)
 
         results = yield make_deferred_yieldable(defer.DeferredList([
-            preserve_fn(self.appservice_api.query_3pe)(service, kind, protocol, fields)
+            run_in_background(
+                self.appservice_api.query_3pe,
+                service, kind, protocol, fields,
+            )
             for service in services
         ], consumeErrors=True))
 
@@ -228,11 +260,15 @@ class ApplicationServicesHandler(object):
             event based on the service regex.
         """
         services = self.store.get_app_services()
-        interested_list = [
+
-            s for s in services if (
+        # we can't use a list comprehension here. Since python 3, list
-                yield s.is_interested(event, self.store)
+        # comprehensions use a generator internally. This means you can't yield
-            )
+        # inside of a list comprehension anymore.
-        ]
+        interested_list = []
+        for s in services:
+            if (yield s.is_interested(event, self.store)):
+                interested_list.append(s)
+
         defer.returnValue(interested_list)
 
     def _get_services_for_user(self, user_id):

synapse/handlers/device.py

@@ -155,7 +155,7 @@ class DeviceHandler(BaseHandler):
 
         try:
             yield self.store.delete_device(user_id, device_id)
-        except errors.StoreError, e:
+        except errors.StoreError as e:
             if e.code == 404:
                 # no match
                 pass
@@ -204,7 +204,7 @@ class DeviceHandler(BaseHandler):
 
         try:
             yield self.store.delete_devices(user_id, device_ids)
-        except errors.StoreError, e:
+        except errors.StoreError as e:
             if e.code == 404:
                 # no match
                 pass
@@ -243,7 +243,7 @@ class DeviceHandler(BaseHandler):
                 new_display_name=content.get("display_name")
             )
             yield self.notify_device_update(user_id, [device_id])
-        except errors.StoreError, e:
+        except errors.StoreError as e:
             if e.code == 404:
                 raise errors.NotFoundError()
             else:

synapse/handlers/e2e_keys.py

@@ -1,5 +1,6 @@
 # -*- coding: utf-8 -*-
 # Copyright 2016 OpenMarket Ltd
+# Copyright 2018 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.
@@ -13,7 +14,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import ujson as json
+import simplejson as json
 import logging
 
 from canonicaljson import encode_canonical_json
@@ -23,7 +24,7 @@ from synapse.api.errors import (
     SynapseError, CodeMessageException, FederationDeniedError,
 )
 from synapse.types import get_domain_from_id, UserID
-from synapse.util.logcontext import preserve_fn, make_deferred_yieldable
+from synapse.util.logcontext import make_deferred_yieldable, run_in_background
 from synapse.util.retryutils import NotRetryingDestination
 
 logger = logging.getLogger(__name__)
@@ -134,28 +135,13 @@ class E2eKeysHandler(object):
                     if user_id in destination_query:
                         results[user_id] = keys
 
-            except CodeMessageException as e:
-                failures[destination] = {
-                    "status": e.code, "message": e.message
-                }
-            except NotRetryingDestination as e:
-                failures[destination] = {
-                    "status": 503, "message": "Not ready for retry",
-                }
-            except FederationDeniedError as e:
-                failures[destination] = {
-                    "status": 403, "message": "Federation Denied",
-                }
             except Exception as e:
-                # include ConnectionRefused and other errors
+                failures[destination] = _exception_to_failure(e)
-                failures[destination] = {
-                    "status": 503, "message": e.message
-                }
 
         yield make_deferred_yieldable(defer.gatherResults([
-            preserve_fn(do_remote_query)(destination)
+            run_in_background(do_remote_query, destination)
             for destination in remote_queries_not_in_cache
-        ]))
+        ], consumeErrors=True))
 
         defer.returnValue({
             "device_keys": results, "failures": failures,
@@ -252,24 +238,13 @@ class E2eKeysHandler(object):
                 for user_id, keys in remote_result["one_time_keys"].items():
                     if user_id in device_keys:
                         json_result[user_id] = keys
-            except CodeMessageException as e:
-                failures[destination] = {
-                    "status": e.code, "message": e.message
-                }
-            except NotRetryingDestination as e:
-                failures[destination] = {
-                    "status": 503, "message": "Not ready for retry",
-                }
             except Exception as e:
-                # include ConnectionRefused and other errors
+                failures[destination] = _exception_to_failure(e)
-                failures[destination] = {
-                    "status": 503, "message": e.message
-                }
 
         yield make_deferred_yieldable(defer.gatherResults([
-            preserve_fn(claim_client_keys)(destination)
+            run_in_background(claim_client_keys, destination)
             for destination in remote_queries
-        ]))
+        ], consumeErrors=True))
 
         logger.info(
             "Claimed one-time-keys: %s",
@@ -362,6 +337,31 @@ class E2eKeysHandler(object):
         )
 
 
+def _exception_to_failure(e):
+    if isinstance(e, CodeMessageException):
+        return {
+            "status": e.code, "message": e.message,
+        }
+
+    if isinstance(e, NotRetryingDestination):
+        return {
+            "status": 503, "message": "Not ready for retry",
+        }
+
+    if isinstance(e, FederationDeniedError):
+        return {
+            "status": 403, "message": "Federation Denied",
+        }
+
+    # include ConnectionRefused and other errors
+    #
+    # Note that some Exceptions (notably twisted's ResponseFailed etc) don't
+    # give a string for e.message, which simplejson then fails to serialize.
+    return {
+        "status": 503, "message": str(e.message),
+    }
+
+
 def _one_time_keys_match(old_key_json, new_key):
     old_key = json.loads(old_key_json)
 

synapse/handlers/federation.py

@@ -15,8 +15,16 @@
 # limitations under the License.
 
 """Contains handlers for federation events."""
+
+import itertools
+import logging
+import sys
+
 from signedjson.key import decode_verify_key_bytes
 from signedjson.sign import verify_signed_json
+import six
+from six.moves import http_client
+from twisted.internet import defer
 from unpaddedbase64 import decode_base64
 
 from ._base import BaseHandler
@@ -43,10 +51,6 @@ from synapse.util.retryutils import NotRetryingDestination
 
 from synapse.util.distributor import user_joined_room
 
-from twisted.internet import defer
-
-import itertools
-import logging
 
 logger = logging.getLogger(__name__)
 
@@ -115,6 +119,19 @@ class FederationHandler(BaseHandler):
             logger.debug("Already seen pdu %s", pdu.event_id)
             return
 
+        # do some initial sanity-checking of the event. In particular, make
+        # sure it doesn't have hundreds of prev_events or auth_events, which
+        # could cause a huge state resolution or cascade of event fetches.
+        try:
+            self._sanity_check_event(pdu)
+        except SynapseError as err:
+            raise FederationError(
+                "ERROR",
+                err.code,
+                err.msg,
+                affected=pdu.event_id,
+            )
+
         # If we are currently in the process of joining this room, then we
         # queue up events for later processing.
         if pdu.room_id in self.room_queues:
@@ -149,10 +166,6 @@ class FederationHandler(BaseHandler):
 
         auth_chain = []
 
-        have_seen = yield self.store.have_events(
-            [ev for ev, _ in pdu.prev_events]
-        )
-
         fetch_state = False
 
         # Get missing pdus if necessary.
@@ -168,7 +181,7 @@ class FederationHandler(BaseHandler):
             )
 
             prevs = {e_id for e_id, _ in pdu.prev_events}
-            seen = set(have_seen.keys())
+            seen = yield self.store.have_seen_events(prevs)
 
             if min_depth and pdu.depth < min_depth:
                 # This is so that we don't notify the user about this
@@ -196,8 +209,7 @@ class FederationHandler(BaseHandler):
 
                         # Update the set of things we've seen after trying to
                         # fetch the missing stuff
-                        have_seen = yield self.store.have_events(prevs)
+                        seen = yield self.store.have_seen_events(prevs)
-                        seen = set(have_seen.iterkeys())
 
                         if not prevs - seen:
                             logger.info(
@@ -248,8 +260,7 @@ class FederationHandler(BaseHandler):
             min_depth (int): Minimum depth of events to return.
         """
         # We recalculate seen, since it may have changed.
-        have_seen = yield self.store.have_events(prevs)
+        seen = yield self.store.have_seen_events(prevs)
-        seen = set(have_seen.keys())
 
         if not prevs - seen:
             return
@@ -361,9 +372,7 @@ class FederationHandler(BaseHandler):
             if auth_chain:
                 event_ids |= {e.event_id for e in auth_chain}
 
-            seen_ids = set(
+            seen_ids = yield self.store.have_seen_events(event_ids)
-                (yield self.store.have_events(event_ids)).keys()
-            )
 
             if state and auth_chain is not None:
                 # If we have any state or auth_chain given to us by the replication
@@ -527,9 +536,16 @@ class FederationHandler(BaseHandler):
     def backfill(self, dest, room_id, limit, extremities):
         """ Trigger a backfill request to `dest` for the given `room_id`
 
-        This will attempt to get more events from the remote. This may return
+        This will attempt to get more events from the remote. If the other side
-        be successfull and still return no events if the other side has no new
+        has no new events to offer, this will return an empty list.
-        events to offer.
+
+        As the events are received, we check their signatures, and also do some
+        sanity-checking on them. If any of the backfilled events are invalid,
+        this method throws a SynapseError.
+
+        TODO: make this more useful to distinguish failures of the remote
+        server from invalid events (there is probably no point in trying to
+        re-fetch invalid events from every other HS in the room.)
         """
         if dest == self.server_name:
             raise SynapseError(400, "Can't backfill from self.")
@@ -541,6 +557,16 @@ class FederationHandler(BaseHandler):
             extremities=extremities,
         )
 
+        # ideally we'd sanity check the events here for excess prev_events etc,
+        # but it's hard to reject events at this point without completely
+        # breaking backfill in the same way that it is currently broken by
+        # events whose signature we cannot verify (#3121).
+        #
+        # So for now we accept the events anyway. #3124 tracks this.
+        #
+        # for ev in events:
+        #     self._sanity_check_event(ev)
+
         # Don't bother processing events we already have.
         seen_events = yield self.store.have_events_in_timeline(
             set(e.event_id for e in events)
@@ -613,7 +639,8 @@ class FederationHandler(BaseHandler):
 
                 results = yield logcontext.make_deferred_yieldable(defer.gatherResults(
                     [
-                        logcontext.preserve_fn(self.replication_layer.get_pdu)(
+                        logcontext.run_in_background(
+                            self.replication_layer.get_pdu,
                             [dest],
                             event_id,
                             outlier=True,
@@ -633,7 +660,7 @@ class FederationHandler(BaseHandler):
 
                 failed_to_fetch = missing_auth - set(auth_events)
 
-        seen_events = yield self.store.have_events(
+        seen_events = yield self.store.have_seen_events(
             set(auth_events.keys()) | set(state_events.keys())
         )
 
@@ -843,6 +870,38 @@ class FederationHandler(BaseHandler):
 
         defer.returnValue(False)
 
+    def _sanity_check_event(self, ev):
+        """
+        Do some early sanity checks of a received event
+
+        In particular, checks it doesn't have an excessive number of
+        prev_events or auth_events, which could cause a huge state resolution
+        or cascade of event fetches.
+
+        Args:
+            ev (synapse.events.EventBase): event to be checked
+
+        Returns: None
+
+        Raises:
+            SynapseError if the event does not pass muster
+        """
+        if len(ev.prev_events) > 20:
+            logger.warn("Rejecting event %s which has %i prev_events",
+                        ev.event_id, len(ev.prev_events))
+            raise SynapseError(
+                http_client.BAD_REQUEST,
+                "Too many prev_events",
+            )
+
+        if len(ev.auth_events) > 10:
+            logger.warn("Rejecting event %s which has %i auth_events",
+                        ev.event_id, len(ev.auth_events))
+            raise SynapseError(
+                http_client.BAD_REQUEST,
+                "Too many auth_events",
+            )
+
     @defer.inlineCallbacks
     def send_invite(self, target_host, event):
         """ Sends the invite to the remote server for signing.
@@ -967,7 +1026,7 @@ class FederationHandler(BaseHandler):
             # lots of requests for missing prev_events which we do actually
             # have. Hence we fire off the deferred, but don't wait for it.
 
-            logcontext.preserve_fn(self._handle_queued_pdus)(room_queue)
+            logcontext.run_in_background(self._handle_queued_pdus, room_queue)
 
         defer.returnValue(True)
 
@@ -1457,18 +1516,21 @@ class FederationHandler(BaseHandler):
                 backfilled=backfilled,
             )
         except:  # noqa: E722, as we reraise the exception this is fine.
-            # Ensure that we actually remove the entries in the push actions
+            tp, value, tb = sys.exc_info()
-            # staging area
+
-            logcontext.preserve_fn(
+            logcontext.run_in_background(
-                self.store.remove_push_actions_from_staging
+                self.store.remove_push_actions_from_staging,
-            )(event.event_id)
+                event.event_id,
-            raise
+            )
+
+            six.reraise(tp, value, tb)
 
         if not backfilled:
             # this intentionally does not yield: we don't care about the result
             # and don't need to wait for it.
-            logcontext.preserve_fn(self.pusher_pool.on_new_notifications)(
+            logcontext.run_in_background(
-                event_stream_id, max_stream_id
+                self.pusher_pool.on_new_notifications,
+                event_stream_id, max_stream_id,
             )
 
         defer.returnValue((context, event_stream_id, max_stream_id))
@@ -1482,7 +1544,8 @@ class FederationHandler(BaseHandler):
         """
         contexts = yield logcontext.make_deferred_yieldable(defer.gatherResults(
             [
-                logcontext.preserve_fn(self._prep_event)(
+                logcontext.run_in_background(
+                    self._prep_event,
                     origin,
                     ev_info["event"],
                     state=ev_info.get("state"),
@@ -1736,7 +1799,8 @@ class FederationHandler(BaseHandler):
             event_key = None
 
         if event_auth_events - current_state:
-            have_events = yield self.store.have_events(
+            # TODO: can we use store.have_seen_events here instead?
+            have_events = yield self.store.get_seen_events_with_rejections(
                 event_auth_events - current_state
             )
         else:
@@ -1759,12 +1823,12 @@ class FederationHandler(BaseHandler):
                     origin, event.room_id, event.event_id
                 )
 
-                seen_remotes = yield self.store.have_events(
+                seen_remotes = yield self.store.have_seen_events(
                     [e.event_id for e in remote_auth_chain]
                 )
 
                 for e in remote_auth_chain:
-                    if e.event_id in seen_remotes.keys():
+                    if e.event_id in seen_remotes:
                         continue
 
                     if e.event_id == event.event_id:
@@ -1791,7 +1855,7 @@ class FederationHandler(BaseHandler):
                     except AuthError:
                         pass
 
-                have_events = yield self.store.have_events(
+                have_events = yield self.store.get_seen_events_with_rejections(
                     [e_id for e_id, _ in event.auth_events]
                 )
                 seen_events = set(have_events.keys())
@@ -1810,7 +1874,8 @@ class FederationHandler(BaseHandler):
 
             different_events = yield logcontext.make_deferred_yieldable(
                 defer.gatherResults([
-                    logcontext.preserve_fn(self.store.get_event)(
+                    logcontext.run_in_background(
+                        self.store.get_event,
                         d,
                         allow_none=True,
                         allow_rejected=False,
@@ -1876,13 +1941,13 @@ class FederationHandler(BaseHandler):
                         local_auth_chain,
                     )
 
-                    seen_remotes = yield self.store.have_events(
+                    seen_remotes = yield self.store.have_seen_events(
                         [e.event_id for e in result["auth_chain"]]
                     )
 
                     # 3. Process any remote auth chain events we haven't seen.
                     for ev in result["auth_chain"]:
-                        if ev.event_id in seen_remotes.keys():
+                        if ev.event_id in seen_remotes:
                             continue
 
                         if ev.event_id == event.event_id:

synapse/handlers/groups_local.py

@@ -1,5 +1,6 @@
 # -*- coding: utf-8 -*-
 # Copyright 2017 Vector Creations Ltd
+# Copyright 2018 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.
@@ -90,6 +91,8 @@ class GroupsLocalHandler(object):
     get_group_role = _create_rerouter("get_group_role")
     get_group_roles = _create_rerouter("get_group_roles")
 
+    set_group_join_policy = _create_rerouter("set_group_join_policy")
+
     @defer.inlineCallbacks
     def get_group_summary(self, group_id, requester_user_id):
         """Get the group summary for a group.
@@ -226,7 +229,45 @@ class GroupsLocalHandler(object):
     def join_group(self, group_id, user_id, content):
         """Request to join a group
         """
-        raise NotImplementedError()  # TODO
+        if self.is_mine_id(group_id):
+            yield self.groups_server_handler.join_group(
+                group_id, user_id, content
+            )
+            local_attestation = None
+            remote_attestation = None
+        else:
+            local_attestation = self.attestations.create_attestation(group_id, user_id)
+            content["attestation"] = local_attestation
+
+            res = yield self.transport_client.join_group(
+                get_domain_from_id(group_id), group_id, user_id, content,
+            )
+
+            remote_attestation = res["attestation"]
+
+            yield self.attestations.verify_attestation(
+                remote_attestation,
+                group_id=group_id,
+                user_id=user_id,
+                server_name=get_domain_from_id(group_id),
+            )
+
+        # TODO: Check that the group is public and we're being added publically
+        is_publicised = content.get("publicise", False)
+
+        token = yield self.store.register_user_group_membership(
+            group_id, user_id,
+            membership="join",
+            is_admin=False,
+            local_attestation=local_attestation,
+            remote_attestation=remote_attestation,
+            is_publicised=is_publicised,
+        )
+        self.notifier.on_new_event(
+            "groups_key", token, users=[user_id],
+        )
+
+        defer.returnValue({})
 
     @defer.inlineCallbacks
     def accept_invite(self, group_id, user_id, content):

synapse/handlers/identity.py

@@ -15,6 +15,11 @@
 # limitations under the License.
 
 """Utilities for interacting with Identity Servers"""
+
+import logging
+
+import simplejson as json
+
 from twisted.internet import defer
 
 from synapse.api.errors import (
@@ -24,9 +29,6 @@ from ._base import BaseHandler
 from synapse.util.async import run_on_reactor
 from synapse.api.errors import SynapseError, Codes
 
-import json
-import logging
-
 logger = logging.getLogger(__name__)
 
 

synapse/handlers/initial_sync.py

@@ -27,7 +27,7 @@ from synapse.types import (
 from synapse.util import unwrapFirstError
 from synapse.util.async import concurrently_execute
 from synapse.util.caches.snapshot_cache import SnapshotCache
-from synapse.util.logcontext import make_deferred_yieldable, preserve_fn
+from synapse.util.logcontext import make_deferred_yieldable, run_in_background
 from synapse.visibility import filter_events_for_client
 
 from ._base import BaseHandler
@@ -166,7 +166,8 @@ class InitialSyncHandler(BaseHandler):
                 (messages, token), current_state = yield make_deferred_yieldable(
                     defer.gatherResults(
                         [
-                            preserve_fn(self.store.get_recent_events_for_room)(
+                            run_in_background(
+                                self.store.get_recent_events_for_room,
                                 event.room_id,
                                 limit=limit,
                                 end_token=room_end_token,
@@ -391,9 +392,10 @@ class InitialSyncHandler(BaseHandler):
 
         presence, receipts, (messages, token) = yield defer.gatherResults(
             [
-                preserve_fn(get_presence)(),
+                run_in_background(get_presence),
-                preserve_fn(get_receipts)(),
+                run_in_background(get_receipts),
-                preserve_fn(self.store.get_recent_events_for_room)(
+                run_in_background(
+                    self.store.get_recent_events_for_room,
                     room_id,
                     limit=limit,
                     end_token=now_token.room_key,

synapse/handlers/message.py

@@ -13,10 +13,16 @@
 # 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 simplejson
+import sys
+
+from canonicaljson import encode_canonical_json
+import six
 from twisted.internet import defer, reactor
 from twisted.python.failure import Failure
 
-from synapse.api.constants import EventTypes, Membership
+from synapse.api.constants import EventTypes, Membership, MAX_DEPTH
 from synapse.api.errors import AuthError, Codes, SynapseError
 from synapse.crypto.event_signing import add_hashes_and_signatures
 from synapse.events.utils import serialize_event
@@ -25,21 +31,15 @@ from synapse.types import (
     UserID, RoomAlias, RoomStreamToken,
 )
 from synapse.util.async import run_on_reactor, ReadWriteLock, Limiter
-from synapse.util.logcontext import preserve_fn, run_in_background
+from synapse.util.logcontext import run_in_background
 from synapse.util.metrics import measure_func
-from synapse.util.frozenutils import unfreeze
+from synapse.util.frozenutils import frozendict_json_encoder
 from synapse.util.stringutils import random_string
 from synapse.visibility import filter_events_for_client
 from synapse.replication.http.send_event import send_event_to_master
 
 from ._base import BaseHandler
 
-from canonicaljson import encode_canonical_json
-
-import logging
-import random
-import ujson
-
 logger = logging.getLogger(__name__)
 
 
@@ -433,7 +433,7 @@ class EventCreationHandler(object):
 
     @defer.inlineCallbacks
     def create_event(self, requester, event_dict, token_id=None, txn_id=None,
-                     prev_event_ids=None):
+                     prev_events_and_hashes=None):
         """
         Given a dict from a client, create a new event.
 
@@ -447,14 +447,19 @@ class EventCreationHandler(object):
             event_dict (dict): An entire event
             token_id (str)
             txn_id (str)
-            prev_event_ids (list): The prev event ids to use when creating the event
+
+            prev_events_and_hashes (list[(str, dict[str, str], int)]|None):
+                the forward extremities to use as the prev_events for the
+                new event. For each event, a tuple of (event_id, hashes, depth)
+                where *hashes* is a map from algorithm to hash.
+
+                If None, they will be requested from the database.
 
         Returns:
             Tuple of created event (FrozenEvent), Context
         """
         builder = self.event_builder_factory.new(event_dict)
 
-        with (yield self.limiter.queue(builder.room_id)):
         self.validator.validate_new(builder)
 
         if builder.type == EventTypes.Member:
@@ -486,7 +491,7 @@ class EventCreationHandler(object):
         event, context = yield self.create_new_client_event(
             builder=builder,
             requester=requester,
-                prev_event_ids=prev_event_ids,
+            prev_events_and_hashes=prev_events_and_hashes,
         )
 
         defer.returnValue((event, context))
@@ -557,6 +562,13 @@ class EventCreationHandler(object):
 
         See self.create_event and self.send_nonmember_event.
         """
+
+        # We limit the number of concurrent event sends in a room so that we
+        # don't fork the DAG too much. If we don't limit then we can end up in
+        # a situation where event persistence can't keep up, causing
+        # extremities to pile up, which in turn leads to state resolution
+        # taking longer.
+        with (yield self.limiter.queue(event_dict["room_id"])):
             event, context = yield self.create_event(
                 requester,
                 event_dict,
@@ -582,38 +594,47 @@ class EventCreationHandler(object):
 
     @measure_func("create_new_client_event")
     @defer.inlineCallbacks
-    def create_new_client_event(self, builder, requester=None, prev_event_ids=None):
+    def create_new_client_event(self, builder, requester=None,
-        if prev_event_ids:
+                                prev_events_and_hashes=None):
-            prev_events = yield self.store.add_event_hashes(prev_event_ids)
+        """Create a new event for a local client
-            prev_max_depth = yield self.store.get_max_depth_of_events(prev_event_ids)
+
-            depth = prev_max_depth + 1
+        Args:
-        else:
+            builder (EventBuilder):
-            latest_ret = yield self.store.get_latest_event_ids_and_hashes_in_room(
+
-                builder.room_id,
+            requester (synapse.types.Requester|None):
+
+            prev_events_and_hashes (list[(str, dict[str, str], int)]|None):
+                the forward extremities to use as the prev_events for the
+                new event. For each event, a tuple of (event_id, hashes, depth)
+                where *hashes* is a map from algorithm to hash.
+
+                If None, they will be requested from the database.
+
+        Returns:
+            Deferred[(synapse.events.EventBase, synapse.events.snapshot.EventContext)]
+        """
+
+        if prev_events_and_hashes is not None:
+            assert len(prev_events_and_hashes) <= 10, \
+                "Attempting to create an event with %i prev_events" % (
+                    len(prev_events_and_hashes),
             )
+        else:
+            prev_events_and_hashes = \
+                yield self.store.get_prev_events_for_room(builder.room_id)
 
-            # We want to limit the max number of prev events we point to in our
+        if prev_events_and_hashes:
-            # new event
+            depth = max([d for _, _, d in prev_events_and_hashes]) + 1
-            if len(latest_ret) > 10:
+            # we cap depth of generated events, to ensure that they are not
-                # Sort by reverse depth, so we point to the most recent.
+            # rejected by other servers (and so that they can be persisted in
-                latest_ret.sort(key=lambda a: -a[2])
+            # the db)
-                new_latest_ret = latest_ret[:5]
+            depth = min(depth, MAX_DEPTH)
-
-                # We also randomly point to some of the older events, to make
-                # sure that we don't completely ignore the older events.
-                if latest_ret[5:]:
-                    sample_size = min(5, len(latest_ret[5:]))
-                    new_latest_ret.extend(random.sample(latest_ret[5:], sample_size))
-                latest_ret = new_latest_ret
-
-            if latest_ret:
-                depth = max([d for _, _, d in latest_ret]) + 1
         else:
             depth = 1
 
         prev_events = [
             (event_id, prev_hashes)
-                for event_id, prev_hashes, _ in latest_ret
+            for event_id, prev_hashes, _ in prev_events_and_hashes
         ]
 
         builder.prev_events = prev_events
@@ -678,8 +699,8 @@ class EventCreationHandler(object):
 
         # Ensure that we can round trip before trying to persist in db
         try:
-            dump = ujson.dumps(unfreeze(event.content))
+            dump = frozendict_json_encoder.encode(event.content)
-            ujson.loads(dump)
+            simplejson.loads(dump)
         except Exception:
             logger.exception("Failed to encode content: %r", event.content)
             raise
@@ -713,8 +734,14 @@ class EventCreationHandler(object):
         except:  # noqa: E722, as we reraise the exception this is fine.
             # Ensure that we actually remove the entries in the push actions
             # staging area, if we calculated them.
-            preserve_fn(self.store.remove_push_actions_from_staging)(event.event_id)
+            tp, value, tb = sys.exc_info()
-            raise
+
+            run_in_background(
+                self.store.remove_push_actions_from_staging,
+                event.event_id,
+            )
+
+            six.reraise(tp, value, tb)
 
     @defer.inlineCallbacks
     def persist_and_notify_client_event(
@@ -834,22 +861,33 @@ class EventCreationHandler(object):
 
         # this intentionally does not yield: we don't care about the result
         # and don't need to wait for it.
-        preserve_fn(self.pusher_pool.on_new_notifications)(
+        run_in_background(
+            self.pusher_pool.on_new_notifications,
             event_stream_id, max_stream_id
         )
 
         @defer.inlineCallbacks
         def _notify():
             yield run_on_reactor()
+            try:
                 self.notifier.on_new_room_event(
                     event, event_stream_id, max_stream_id,
                     extra_users=extra_users
                 )
+            except Exception:
+                logger.exception("Error notifying about new room event")
 
-        preserve_fn(_notify)()
+        run_in_background(_notify)
 
         if event.type == EventTypes.Message:
-            presence = self.hs.get_presence_handler()
             # We don't want to block sending messages on any presence code. This
             # matters as sometimes presence code can take a while.
-            preserve_fn(presence.bump_presence_active_time)(requester.user)
+            run_in_background(self._bump_active_time, requester.user)
+
+    @defer.inlineCallbacks
+    def _bump_active_time(self, user):
+        try:
+            presence = self.hs.get_presence_handler()
+            yield presence.bump_presence_active_time(user)
+        except Exception:
+            logger.exception("Error bumping presence active time")

synapse/handlers/presence.py

@@ -31,7 +31,7 @@ from synapse.storage.presence import UserPresenceState
 
 from synapse.util.caches.descriptors import cachedInlineCallbacks
 from synapse.util.async import Linearizer
-from synapse.util.logcontext import preserve_fn
+from synapse.util.logcontext import run_in_background
 from synapse.util.logutils import log_function
 from synapse.util.metrics import Measure
 from synapse.util.wheel_timer import WheelTimer
@@ -254,6 +254,14 @@ class PresenceHandler(object):
 
         logger.info("Finished _persist_unpersisted_changes")
 
+    @defer.inlineCallbacks
+    def _update_states_and_catch_exception(self, new_states):
+        try:
+            res = yield self._update_states(new_states)
+            defer.returnValue(res)
+        except Exception:
+            logger.exception("Error updating presence")
+
     @defer.inlineCallbacks
     def _update_states(self, new_states):
         """Updates presence of users. Sets the appropriate timeouts. Pokes
@@ -364,7 +372,7 @@ class PresenceHandler(object):
                     now=now,
                 )
 
-            preserve_fn(self._update_states)(changes)
+            run_in_background(self._update_states_and_catch_exception, changes)
         except Exception:
             logger.exception("Exception in _handle_timeouts loop")
 
@@ -422,20 +430,23 @@ class PresenceHandler(object):
 
         @defer.inlineCallbacks
         def _end():
-            if affect_presence:
+            try:
                 self.user_to_num_current_syncs[user_id] -= 1
 
                 prev_state = yield self.current_state_for_user(user_id)
                 yield self._update_states([prev_state.copy_and_replace(
                     last_user_sync_ts=self.clock.time_msec(),
                 )])
+            except Exception:
+                logger.exception("Error updating presence after sync")
 
         @contextmanager
         def _user_syncing():
             try:
                 yield
             finally:
-                preserve_fn(_end)()
+                if affect_presence:
+                    run_in_background(_end)
 
         defer.returnValue(_user_syncing())
 

synapse/handlers/receipts.py

@@ -135,6 +135,7 @@ class ReceiptsHandler(BaseHandler):
         """Given a list of receipts, works out which remote servers should be
         poked and pokes them.
         """
+        try:
             # TODO: Some of this stuff should be coallesced.
             for receipt in receipts:
                 room_id = receipt["room_id"]
@@ -166,6 +167,8 @@ class ReceiptsHandler(BaseHandler):
                         },
                         key=(room_id, receipt_type, user_id),
                     )
+        except Exception:
+            logger.exception("Error pushing receipts to remote servers")
 
     @defer.inlineCallbacks
     def get_receipts_for_room(self, room_id, to_key):

synapse/handlers/register.py

@@ -23,8 +23,8 @@ from synapse.api.errors import (
 )
 from synapse.http.client import CaptchaServerHttpClient
 from synapse import types
-from synapse.types import UserID
+from synapse.types import UserID, create_requester, RoomID, RoomAlias
-from synapse.util.async import run_on_reactor
+from synapse.util.async import run_on_reactor, Linearizer
 from synapse.util.threepids import check_3pid_allowed
 from ._base import BaseHandler
 
@@ -46,6 +46,10 @@ class RegistrationHandler(BaseHandler):
 
         self.macaroon_gen = hs.get_macaroon_generator()
 
+        self._generate_user_id_linearizer = Linearizer(
+            name="_generate_user_id_linearizer",
+        )
+
     @defer.inlineCallbacks
     def check_username(self, localpart, guest_access_token=None,
                        assigned_user_id=None):
@@ -201,10 +205,17 @@ class RegistrationHandler(BaseHandler):
                     token = None
                     attempts += 1
 
+        # auto-join the user to any rooms we're supposed to dump them into
+        fake_requester = create_requester(user_id)
+        for r in self.hs.config.auto_join_rooms:
+            try:
+                yield self._join_user_to_room(fake_requester, r)
+            except Exception as e:
+                logger.error("Failed to join new user to %r: %r", r, e)
+
         # We used to generate default identicons here, but nowadays
         # we want clients to generate their own as part of their branding
         # rather than there being consistent matrix-wide ones, so we don't.
-
         defer.returnValue((user_id, token))
 
     @defer.inlineCallbacks
@@ -344,6 +355,8 @@ class RegistrationHandler(BaseHandler):
 
     @defer.inlineCallbacks
     def _generate_user_id(self, reseed=False):
+        if reseed or self._next_generated_user_id is None:
+            with (yield self._generate_user_id_linearizer.queue(())):
                 if reseed or self._next_generated_user_id is None:
                     self._next_generated_user_id = (
                         yield self.store.find_next_generated_user_id_localpart()
@@ -477,3 +490,28 @@ class RegistrationHandler(BaseHandler):
         )
 
         defer.returnValue((user_id, access_token))
+
+    @defer.inlineCallbacks
+    def _join_user_to_room(self, requester, room_identifier):
+        room_id = None
+        room_member_handler = self.hs.get_room_member_handler()
+        if RoomID.is_valid(room_identifier):
+            room_id = room_identifier
+        elif RoomAlias.is_valid(room_identifier):
+            room_alias = RoomAlias.from_string(room_identifier)
+            room_id, remote_room_hosts = (
+                yield room_member_handler.lookup_room_alias(room_alias)
+            )
+            room_id = room_id.to_string()
+        else:
+            raise SynapseError(400, "%s was not legal room ID or room alias" % (
+                room_identifier,
+            ))
+
+        yield room_member_handler.update_membership(
+            requester=requester,
+            target=requester.user,
+            room_id=room_id,
+            remote_room_hosts=remote_room_hosts,
+            action="join",
+        )

synapse/handlers/room_list.py

@@ -15,12 +15,13 @@
 
 from twisted.internet import defer
 
+from six.moves import range
+
 from ._base import BaseHandler
 
 from synapse.api.constants import (
     EventTypes, JoinRules,
 )
-from synapse.util.logcontext import make_deferred_yieldable, preserve_fn
 from synapse.util.async import concurrently_execute
 from synapse.util.caches.descriptors import cachedInlineCallbacks
 from synapse.util.caches.response_cache import ResponseCache
@@ -44,8 +45,9 @@ EMTPY_THIRD_PARTY_ID = ThirdPartyInstanceID(None, None)
 class RoomListHandler(BaseHandler):
     def __init__(self, hs):
         super(RoomListHandler, self).__init__(hs)
-        self.response_cache = ResponseCache(hs)
+        self.response_cache = ResponseCache(hs, "room_list")
-        self.remote_response_cache = ResponseCache(hs, timeout_ms=30 * 1000)
+        self.remote_response_cache = ResponseCache(hs, "remote_room_list",
+                                                   timeout_ms=30 * 1000)
 
     def get_local_public_room_list(self, limit=None, since_token=None,
                                    search_filter=None,
@@ -77,18 +79,11 @@ class RoomListHandler(BaseHandler):
             )
 
         key = (limit, since_token, network_tuple)
-        result = self.response_cache.get(key)
+        return self.response_cache.wrap(
-        if not result:
-            logger.info("No cached result, calculating one.")
-            result = self.response_cache.set(
             key,
-                preserve_fn(self._get_public_room_list)(
+            self._get_public_room_list,
-                    limit, since_token, network_tuple=network_tuple
+            limit, since_token, network_tuple=network_tuple,
         )
-            )
-        else:
-            logger.info("Using cached deferred result.")
-        return make_deferred_yieldable(result)
 
     @defer.inlineCallbacks
     def _get_public_room_list(self, limit=None, since_token=None,
@@ -207,7 +202,7 @@ class RoomListHandler(BaseHandler):
             step = len(rooms_to_scan) if len(rooms_to_scan) != 0 else 1
 
         chunk = []
-        for i in xrange(0, len(rooms_to_scan), step):
+        for i in range(0, len(rooms_to_scan), step):
             batch = rooms_to_scan[i:i + step]
             logger.info("Processing %i rooms for result", len(batch))
             yield concurrently_execute(
@@ -422,18 +417,14 @@ class RoomListHandler(BaseHandler):
             server_name, limit, since_token, include_all_networks,
             third_party_instance_id,
         )
-        result = self.remote_response_cache.get(key)
+        return self.remote_response_cache.wrap(
-        if not result:
-            result = self.remote_response_cache.set(
             key,
-                repl_layer.get_public_rooms(
+            repl_layer.get_public_rooms,
             server_name, limit=limit, since_token=since_token,
             search_filter=search_filter,
             include_all_networks=include_all_networks,
             third_party_instance_id=third_party_instance_id,
         )
-            )
-        return result
 
 
 class RoomListNextBatch(namedtuple("RoomListNextBatch", (

synapse/handlers/room_member.py

@@ -149,7 +149,7 @@ class RoomMemberHandler(object):
     @defer.inlineCallbacks
     def _local_membership_update(
         self, requester, target, room_id, membership,
-        prev_event_ids,
+        prev_events_and_hashes,
         txn_id=None,
         ratelimit=True,
         content=None,
@@ -175,7 +175,7 @@ class RoomMemberHandler(object):
             },
             token_id=requester.access_token_id,
             txn_id=txn_id,
-            prev_event_ids=prev_event_ids,
+            prev_events_and_hashes=prev_events_and_hashes,
         )
 
         # Check if this event matches the previous membership event for the user.
@@ -314,7 +314,12 @@ class RoomMemberHandler(object):
                     403, "Invites have been disabled on this server",
                 )
 
-        latest_event_ids = yield self.store.get_latest_event_ids_in_room(room_id)
+        prev_events_and_hashes = yield self.store.get_prev_events_for_room(
+            room_id,
+        )
+        latest_event_ids = (
+            event_id for (event_id, _, _) in prev_events_and_hashes
+        )
         current_state_ids = yield self.state_handler.get_current_state_ids(
             room_id, latest_event_ids=latest_event_ids,
         )
@@ -403,7 +408,7 @@ class RoomMemberHandler(object):
             membership=effective_membership_state,
             txn_id=txn_id,
             ratelimit=ratelimit,
-            prev_event_ids=latest_event_ids,
+            prev_events_and_hashes=prev_events_and_hashes,
             content=content,
         )
         defer.returnValue(res)
@@ -852,6 +857,14 @@ class RoomMemberMasterHandler(RoomMemberHandler):
     def _remote_join(self, requester, remote_room_hosts, room_id, user, content):
         """Implements RoomMemberHandler._remote_join
         """
+        # filter ourselves out of remote_room_hosts: do_invite_join ignores it
+        # and if it is the only entry we'd like to return a 404 rather than a
+        # 500.
+
+        remote_room_hosts = [
+            host for host in remote_room_hosts if host != self.hs.hostname
+        ]
+
         if len(remote_room_hosts) == 0:
             raise SynapseError(404, "No known servers")
 

synapse/handlers/sync.py

@@ -15,7 +15,7 @@
 
 from synapse.api.constants import Membership, EventTypes
 from synapse.util.async import concurrently_execute
-from synapse.util.logcontext import LoggingContext, make_deferred_yieldable, preserve_fn
+from synapse.util.logcontext import LoggingContext
 from synapse.util.metrics import Measure, measure_func
 from synapse.util.caches.response_cache import ResponseCache
 from synapse.push.clientformat import format_push_rules_for_user
@@ -52,6 +52,7 @@ class TimelineBatch(collections.namedtuple("TimelineBatch", [
         to tell if room needs to be part of the sync result.
         """
         return bool(self.events)
+    __bool__ = __nonzero__  # python3
 
 
 class JoinedSyncResult(collections.namedtuple("JoinedSyncResult", [
@@ -76,6 +77,7 @@ class JoinedSyncResult(collections.namedtuple("JoinedSyncResult", [
             # nb the notification count does not, er, count: if there's nothing
             # else in the result, we don't need to send it.
         )
+    __bool__ = __nonzero__  # python3
 
 
 class ArchivedSyncResult(collections.namedtuple("ArchivedSyncResult", [
@@ -95,6 +97,7 @@ class ArchivedSyncResult(collections.namedtuple("ArchivedSyncResult", [
             or self.state
             or self.account_data
         )
+    __bool__ = __nonzero__  # python3
 
 
 class InvitedSyncResult(collections.namedtuple("InvitedSyncResult", [
@@ -106,6 +109,7 @@ class InvitedSyncResult(collections.namedtuple("InvitedSyncResult", [
     def __nonzero__(self):
         """Invited rooms should always be reported to the client"""
         return True
+    __bool__ = __nonzero__  # python3
 
 
 class GroupsSyncResult(collections.namedtuple("GroupsSyncResult", [
@@ -117,6 +121,7 @@ class GroupsSyncResult(collections.namedtuple("GroupsSyncResult", [
 
     def __nonzero__(self):
         return bool(self.join or self.invite or self.leave)
+    __bool__ = __nonzero__  # python3
 
 
 class DeviceLists(collections.namedtuple("DeviceLists", [
@@ -127,6 +132,7 @@ class DeviceLists(collections.namedtuple("DeviceLists", [
 
     def __nonzero__(self):
         return bool(self.changed or self.left)
+    __bool__ = __nonzero__  # python3
 
 
 class SyncResult(collections.namedtuple("SyncResult", [
@@ -159,6 +165,7 @@ class SyncResult(collections.namedtuple("SyncResult", [
             self.device_lists or
             self.groups
         )
+    __bool__ = __nonzero__  # python3
 
 
 class SyncHandler(object):
@@ -169,7 +176,7 @@ class SyncHandler(object):
         self.presence_handler = hs.get_presence_handler()
         self.event_sources = hs.get_event_sources()
         self.clock = hs.get_clock()
-        self.response_cache = ResponseCache(hs)
+        self.response_cache = ResponseCache(hs, "sync")
         self.state = hs.get_state_handler()
 
     def wait_for_sync_for_user(self, sync_config, since_token=None, timeout=0,
@@ -180,15 +187,11 @@ class SyncHandler(object):
         Returns:
             A Deferred SyncResult.
         """
-        result = self.response_cache.get(sync_config.request_key)
+        return self.response_cache.wrap(
-        if not result:
-            result = self.response_cache.set(
             sync_config.request_key,
-                preserve_fn(self._wait_for_sync_for_user)(
+            self._wait_for_sync_for_user,
-                    sync_config, since_token, timeout, full_state
+            sync_config, since_token, timeout, full_state,
         )
-            )
-        return make_deferred_yieldable(result)
 
     @defer.inlineCallbacks
     def _wait_for_sync_for_user(self, sync_config, since_token, timeout,

synapse/handlers/typing.py

@@ -16,7 +16,7 @@
 from twisted.internet import defer
 
 from synapse.api.errors import SynapseError, AuthError
-from synapse.util.logcontext import preserve_fn
+from synapse.util.logcontext import run_in_background
 from synapse.util.metrics import Measure
 from synapse.util.wheel_timer import WheelTimer
 from synapse.types import UserID, get_domain_from_id
@@ -97,7 +97,8 @@ class TypingHandler(object):
             if self.hs.is_mine_id(member.user_id):
                 last_fed_poke = self._member_last_federation_poke.get(member, None)
                 if not last_fed_poke or last_fed_poke + FEDERATION_PING_INTERVAL <= now:
-                    preserve_fn(self._push_remote)(
+                    run_in_background(
+                        self._push_remote,
                         member=member,
                         typing=True
                     )
@@ -196,7 +197,7 @@ class TypingHandler(object):
     def _push_update(self, member, typing):
         if self.hs.is_mine_id(member.user_id):
             # Only send updates for changes to our own users.
-            preserve_fn(self._push_remote)(member, typing)
+            run_in_background(self._push_remote, member, typing)
 
         self._push_update_local(
             member=member,
@@ -205,6 +206,7 @@ class TypingHandler(object):
 
     @defer.inlineCallbacks
     def _push_remote(self, member, typing):
+        try:
             users = yield self.state.get_current_user_in_room(member.room_id)
             self._member_last_federation_poke[member] = self.clock.time_msec()
 
@@ -227,6 +229,8 @@ class TypingHandler(object):
                         },
                         key=member,
                     )
+        except Exception:
+            logger.exception("Error pushing typing notif to remotes")
 
     @defer.inlineCallbacks
     def _recv_edu(self, origin, content):

synapse/http/__init__.py

@@ -1,5 +1,6 @@
 # -*- coding: utf-8 -*-
 # Copyright 2014-2016 OpenMarket Ltd
+# Copyright 2018 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.
@@ -12,3 +13,24 @@
 # 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 twisted.internet.defer import CancelledError
+from twisted.python import failure
+
+from synapse.api.errors import SynapseError
+
+
+class RequestTimedOutError(SynapseError):
+    """Exception representing timeout of an outbound request"""
+    def __init__(self):
+        super(RequestTimedOutError, self).__init__(504, "Timed out")
+
+
+def cancelled_to_request_timed_out_error(value, timeout):
+    """Turns CancelledErrors into RequestTimedOutErrors.
+
+    For use with async.add_timeout_to_deferred
+    """
+    if isinstance(value, failure.Failure):
+        value.trap(CancelledError)
+        raise RequestTimedOutError()
+    return value

synapse/http/client.py

@@ -1,5 +1,6 @@
 # -*- coding: utf-8 -*-
 # Copyright 2014-2016 OpenMarket Ltd
+# Copyright 2018 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.
@@ -18,9 +19,10 @@ from OpenSSL.SSL import VERIFY_NONE
 from synapse.api.errors import (
     CodeMessageException, MatrixCodeMessageException, SynapseError, Codes,
 )
+from synapse.http import cancelled_to_request_timed_out_error
+from synapse.util.async import add_timeout_to_deferred
 from synapse.util.caches import CACHE_SIZE_FACTOR
 from synapse.util.logcontext import make_deferred_yieldable
-from synapse.util import logcontext
 import synapse.metrics
 from synapse.http.endpoint import SpiderEndpoint
 
@@ -38,7 +40,7 @@ from twisted.web.http import PotentialDataLoss
 from twisted.web.http_headers import Headers
 from twisted.web._newclient import ResponseDone
 
-from StringIO import StringIO
+from six import StringIO
 
 import simplejson as json
 import logging
@@ -95,21 +97,17 @@ class SimpleHttpClient(object):
         # counters to it
         outgoing_requests_counter.inc(method)
 
-        def send_request():
-            request_deferred = self.agent.request(
-                method, uri, *args, **kwargs
-            )
-
-            return self.clock.time_bound_deferred(
-                request_deferred,
-                time_out=60,
-            )
-
         logger.info("Sending request %s %s", method, uri)
 
         try:
-            with logcontext.PreserveLoggingContext():
+            request_deferred = self.agent.request(
-                response = yield send_request()
+                method, uri, *args, **kwargs
+            )
+            add_timeout_to_deferred(
+                request_deferred,
+                60, cancelled_to_request_timed_out_error,
+            )
+            response = yield make_deferred_yieldable(request_deferred)
 
             incoming_responses_counter.inc(method, response.code)
             logger.info(
@@ -509,7 +507,7 @@ class SpiderHttpClient(SimpleHttpClient):
                     reactor,
                     SpiderEndpointFactory(hs)
                 )
-            ), [('gzip', GzipDecoder)]
+            ), [(b'gzip', GzipDecoder)]
         )
         # We could look like Chrome:
         # self.user_agent = ("Mozilla/5.0 (%s) (KHTML, like Gecko)

synapse/http/endpoint.py

@@ -12,8 +12,6 @@
 # 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 socket
-
 from twisted.internet.endpoints import HostnameEndpoint, wrapClientTLS
 from twisted.internet import defer, reactor
 from twisted.internet.error import ConnectError
@@ -33,7 +31,7 @@ SERVER_CACHE = {}
 
 # our record of an individual server which can be tried to reach a destination.
 #
-# "host" is actually a dotted-quad or ipv6 address string. Except when there's
+# "host" is the hostname acquired from the SRV record. Except when there's
 # no SRV record, in which case it is the original hostname.
 _Server = collections.namedtuple(
     "_Server", "priority weight host port expires"
@@ -117,9 +115,14 @@ class _WrappedConnection(object):
         if time.time() - self.last_request >= 2.5 * 60:
             self.abort()
             # Abort the underlying TLS connection. The abort() method calls
-            # loseConnection() on the underlying TLS connection which tries to
+            # loseConnection() on the TLS connection which tries to
             # shutdown the connection cleanly. We call abortConnection()
-            # since that will promptly close the underlying TCP connection.
+            # since that will promptly close the TLS connection.
+            #
+            # In Twisted >18.4; the TLS connection will be None if it has closed
+            # which will make abortConnection() throw. Check that the TLS connection
+            # is not None before trying to close it.
+            if self.transport.getHandle() is not None:
                 self.transport.abortConnection()
 
     def request(self, request):
@@ -288,7 +291,7 @@ def resolve_service(service_name, dns_client=client, cache=SERVER_CACHE, clock=t
         if (len(answers) == 1
                 and answers[0].type == dns.SRV
                 and answers[0].payload
-                and answers[0].payload.target == dns.Name('.')):
+                and answers[0].payload.target == dns.Name(b'.')):
             raise ConnectError("Service %s unavailable" % service_name)
 
         for answer in answers:
@@ -297,19 +300,12 @@ def resolve_service(service_name, dns_client=client, cache=SERVER_CACHE, clock=t
 
             payload = answer.payload
 
-            hosts = yield _get_hosts_for_srv_record(
-                dns_client, str(payload.target)
-            )
-
-            for (ip, ttl) in hosts:
-                host_ttl = min(answer.ttl, ttl)
-
             servers.append(_Server(
-                    host=ip,
+                host=str(payload.target),
                 port=int(payload.port),
                 priority=int(payload.priority),
                 weight=int(payload.weight),
-                    expires=int(clock.time()) + host_ttl,
+                expires=int(clock.time()) + answer.ttl,
             ))
 
         servers.sort()
@@ -328,81 +324,3 @@ def resolve_service(service_name, dns_client=client, cache=SERVER_CACHE, clock=t
             raise e
 
     defer.returnValue(servers)
-
-
-@defer.inlineCallbacks
-def _get_hosts_for_srv_record(dns_client, host):
-    """Look up each of the hosts in a SRV record
-
-    Args:
-        dns_client (twisted.names.dns.IResolver):
-        host (basestring): host to look up
-
-    Returns:
-        Deferred[list[(str, int)]]: a list of (host, ttl) pairs
-
-    """
-    ip4_servers = []
-    ip6_servers = []
-
-    def cb(res):
-        # lookupAddress and lookupIP6Address return a three-tuple
-        # giving the answer, authority, and additional sections of the
-        # response.
-        #
-        # we only care about the answers.
-
-        return res[0]
-
-    def eb(res, record_type):
-        if res.check(DNSNameError):
-            return []
-        logger.warn("Error looking up %s for %s: %s", record_type, host, res)
-        return res
-
-    # no logcontexts here, so we can safely fire these off and gatherResults
-    d1 = dns_client.lookupAddress(host).addCallbacks(
-        cb, eb, errbackArgs=("A", ))
-    d2 = dns_client.lookupIPV6Address(host).addCallbacks(
-        cb, eb, errbackArgs=("AAAA", ))
-    results = yield defer.DeferredList(
-        [d1, d2], consumeErrors=True)
-
-    # if all of the lookups failed, raise an exception rather than blowing out
-    # the cache with an empty result.
-    if results and all(s == defer.FAILURE for (s, _) in results):
-        defer.returnValue(results[0][1])
-
-    for (success, result) in results:
-        if success == defer.FAILURE:
-            continue
-
-        for answer in result:
-            if not answer.payload:
-                continue
-
-            try:
-                if answer.type == dns.A:
-                    ip = answer.payload.dottedQuad()
-                    ip4_servers.append((ip, answer.ttl))
-                elif answer.type == dns.AAAA:
-                    ip = socket.inet_ntop(
-                        socket.AF_INET6, answer.payload.address,
-                    )
-                    ip6_servers.append((ip, answer.ttl))
-                else:
-                    # the most likely candidate here is a CNAME record.
-                    # rfc2782 says srvs may not point to aliases.
-                    logger.warn(
-                        "Ignoring unexpected DNS record type %s for %s",
-                        answer.type, host,
-                    )
-                    continue
-            except Exception as e:
-                logger.warn("Ignoring invalid DNS response for %s: %s",
-                            host, e)
-                continue
-
-    # keep the ipv4 results before the ipv6 results, mostly to match historical
-    # behaviour.
-    defer.returnValue(ip4_servers + ip6_servers)

synapse/http/matrixfederationclient.py

@@ -1,5 +1,6 @@
 # -*- coding: utf-8 -*-
 # Copyright 2014-2016 OpenMarket Ltd
+# Copyright 2018 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.
@@ -12,17 +13,19 @@
 # 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
 from twisted.web.http_headers import Headers
 from twisted.web._newclient import ResponseDone
 
+from synapse.http import cancelled_to_request_timed_out_error
 from synapse.http.endpoint import matrix_federation_endpoint
-from synapse.util.async import sleep
-from synapse.util import logcontext
 import synapse.metrics
+from synapse.util.async import sleep, add_timeout_to_deferred
+from synapse.util import logcontext
+from synapse.util.logcontext import make_deferred_yieldable
+import synapse.util.retryutils
 
 from canonicaljson import encode_canonical_json
 
@@ -38,8 +41,7 @@ import logging
 import random
 import sys
 import urllib
-import urlparse
+from six.moves.urllib import parse as urlparse
-
 
 logger = logging.getLogger(__name__)
 outbound_logger = logging.getLogger("synapse.http.outbound")
@@ -184,21 +186,20 @@ class MatrixFederationHttpClient(object):
                         producer = body_callback(method, http_url_bytes, headers_dict)
 
                     try:
-                        def send_request():
                         request_deferred = self.agent.request(
                             method,
                             url_bytes,
                             Headers(headers_dict),
                             producer
                         )
-
+                        add_timeout_to_deferred(
-                            return self.clock.time_bound_deferred(
+                            request_deferred,
+                            timeout / 1000. if timeout else 60,
+                            cancelled_to_request_timed_out_error,
+                        )
+                        response = yield make_deferred_yieldable(
                             request_deferred,
-                                time_out=timeout / 1000. if timeout else 60,
                         )
-
-                        with logcontext.PreserveLoggingContext():
-                            response = yield send_request()
 
                         log_result = "%d %s" % (response.code, response.phrase,)
                         break
@@ -286,7 +287,8 @@ class MatrixFederationHttpClient(object):
         headers_dict[b"Authorization"] = auth_headers
 
     @defer.inlineCallbacks
-    def put_json(self, destination, path, data={}, json_data_callback=None,
+    def put_json(self, destination, path, args={}, data={},
+                 json_data_callback=None,
                  long_retries=False, timeout=None,
                  ignore_backoff=False,
                  backoff_on_404=False):
@@ -296,6 +298,7 @@ class MatrixFederationHttpClient(object):
             destination (str): The remote server to send the HTTP request
                 to.
             path (str): The HTTP path.
+            args (dict): query params
             data (dict): A dict containing the data that will be used as
                 the request body. This will be encoded as JSON.
             json_data_callback (callable): A callable returning the dict to
@@ -342,6 +345,7 @@ class MatrixFederationHttpClient(object):
             path,
             body_callback=body_callback,
             headers_dict={"Content-Type": ["application/json"]},
+            query_bytes=encode_query_args(args),
             long_retries=long_retries,
             timeout=timeout,
             ignore_backoff=ignore_backoff,
@@ -373,6 +377,7 @@ class MatrixFederationHttpClient(object):
                 giving up. None indicates no timeout.
             ignore_backoff (bool): true to ignore the historical backoff data and
                 try the request anyway.
+            args (dict): query params
         Returns:
             Deferred: Succeeds when we get a 2xx HTTP response. The result
             will be the decoded JSON body.

synapse/http/server.py

@@ -37,7 +37,7 @@ from twisted.web.util import redirectTo
 import collections
 import logging
 import urllib
-import ujson
+import simplejson
 
 logger = logging.getLogger(__name__)
 
@@ -113,6 +113,11 @@ response_db_sched_duration = metrics.register_counter(
     "response_db_sched_duration_seconds", labels=["method", "servlet", "tag"]
 )
 
+# size in bytes of the response written
+response_size = metrics.register_counter(
+    "response_size", labels=["method", "servlet", "tag"]
+)
+
 _next_request_id = 0
 
 
@@ -324,7 +329,7 @@ class JsonResource(HttpServer, resource.Resource):
                 register_paths, so will return (possibly via Deferred) either
                 None, or a tuple of (http code, response body).
         """
-        if request.method == "OPTIONS":
+        if request.method == b"OPTIONS":
             return _options_handler, {}
 
         # Loop through all the registered callbacks to check if the method
@@ -426,6 +431,8 @@ class RequestMetrics(object):
             context.db_sched_duration_ms / 1000., request.method, self.name, tag
         )
 
+        response_size.inc_by(request.sentLength, request.method, self.name, tag)
+
 
 class RootRedirect(resource.Resource):
     """Redirects the root '/' path to another path."""
@@ -461,8 +468,7 @@ def respond_with_json(request, code, json_object, send_cors=False,
         if canonical_json or synapse.events.USE_FROZEN_DICTS:
             json_bytes = encode_canonical_json(json_object)
         else:
-            # ujson doesn't like frozen_dicts.
+            json_bytes = simplejson.dumps(json_object)
-            json_bytes = ujson.dumps(json_object, ensure_ascii=False)
 
     return respond_with_json_bytes(
         request, code, json_bytes,
@@ -489,6 +495,7 @@ def respond_with_json_bytes(request, code, json_bytes, send_cors=False,
     request.setHeader(b"Content-Type", b"application/json")
     request.setHeader(b"Server", version_string)
     request.setHeader(b"Content-Length", b"%d" % (len(json_bytes),))
+    request.setHeader(b"Cache-Control", b"no-cache, no-store, must-revalidate")
 
     if send_cors:
         set_cors_headers(request)
@@ -536,9 +543,9 @@ def finish_request(request):
 
 def _request_user_agent_is_curl(request):
     user_agents = request.requestHeaders.getRawHeaders(
-        "User-Agent", default=[]
+        b"User-Agent", default=[]
     )
     for user_agent in user_agents:
-        if "curl" in user_agent:
+        if b"curl" in user_agent:
             return True
     return False

synapse/http/site.py

@@ -20,7 +20,7 @@ import logging
 import re
 import time
 
-ACCESS_TOKEN_RE = re.compile(r'(\?.*access(_|%5[Ff])token=)[^&]*(.*)$')
+ACCESS_TOKEN_RE = re.compile(br'(\?.*access(_|%5[Ff])token=)[^&]*(.*)$')
 
 
 class SynapseRequest(Request):
@@ -43,12 +43,12 @@ class SynapseRequest(Request):
 
     def get_redacted_uri(self):
         return ACCESS_TOKEN_RE.sub(
-            r'\1<redacted>\3',
+            br'\1<redacted>\3',
             self.uri
         )
 
     def get_user_agent(self):
-        return self.requestHeaders.getRawHeaders("User-Agent", [None])[-1]
+        return self.requestHeaders.getRawHeaders(b"User-Agent", [None])[-1]
 
     def started_processing(self):
         self.site.access_logger.info(

synapse/metrics/__init__.py

@@ -17,12 +17,13 @@ import logging
 import functools
 import time
 import gc
+import platform
 
 from twisted.internet import reactor
 
 from .metric import (
     CounterMetric, CallbackMetric, DistributionMetric, CacheMetric,
-    MemoryUsageMetric,
+    MemoryUsageMetric, GaugeMetric,
 )
 from .process_collector import register_process_collector
 
@@ -30,6 +31,7 @@ from .process_collector import register_process_collector
 logger = logging.getLogger(__name__)
 
 
+running_on_pypy = platform.python_implementation() == 'PyPy'
 all_metrics = []
 all_collectors = []
 
@@ -63,6 +65,13 @@ class Metrics(object):
         """
         return self._register(CounterMetric, *args, **kwargs)
 
+    def register_gauge(self, *args, **kwargs):
+        """
+        Returns:
+            GaugeMetric
+        """
+        return self._register(GaugeMetric, *args, **kwargs)
+
     def register_callback(self, *args, **kwargs):
         """
         Returns:
@@ -142,6 +151,32 @@ reactor_metrics = get_metrics_for("python.twisted.reactor")
 tick_time = reactor_metrics.register_distribution("tick_time")
 pending_calls_metric = reactor_metrics.register_distribution("pending_calls")
 
+synapse_metrics = get_metrics_for("synapse")
+
+# Used to track where various components have processed in the event stream,
+# e.g. federation sending, appservice sending, etc.
+event_processing_positions = synapse_metrics.register_gauge(
+    "event_processing_positions", labels=["name"],
+)
+
+# Used to track the current max events stream position
+event_persisted_position = synapse_metrics.register_gauge(
+    "event_persisted_position",
+)
+
+# Used to track the received_ts of the last event processed by various
+# components
+event_processing_last_ts = synapse_metrics.register_gauge(
+    "event_processing_last_ts", labels=["name"],
+)
+
+# Used to track the lag processing events. This is the time difference
+# between the last processed event's received_ts and the time it was
+# finished being processed.
+event_processing_lag = synapse_metrics.register_gauge(
+    "event_processing_lag", labels=["name"],
+)
+
 
 def runUntilCurrentTimer(func):
 
@@ -174,6 +209,9 @@ def runUntilCurrentTimer(func):
         tick_time.inc_by(end - start)
         pending_calls_metric.inc_by(num_pending)
 
+        if running_on_pypy:
+            return ret
+
         # Check if we need to do a manual GC (since its been disabled), and do
         # one if necessary.
         threshold = gc.get_threshold()
@@ -206,6 +244,7 @@ try:
 
     # We manually run the GC each reactor tick so that we can get some metrics
     # about time spent doing GC,
+    if not running_on_pypy:
         gc.disable()
 except AttributeError:
     pass

synapse/metrics/metric.py

@@ -16,6 +16,7 @@
 
 from itertools import chain
 import logging
+import re
 
 logger = logging.getLogger(__name__)
 
@@ -56,8 +57,7 @@ class BaseMetric(object):
         return not len(self.labels)
 
     def _render_labelvalue(self, value):
-        # TODO: escape backslashes, quotes and newlines
+        return '"%s"' % (_escape_label_value(value),)
-        return '"%s"' % (value)
 
     def _render_key(self, values):
         if self.is_scalar():
@@ -115,7 +115,7 @@ class CounterMetric(BaseMetric):
         # dict[list[str]]: value for each set of label values. the keys are the
         # label values, in the same order as the labels in self.labels.
         #
-        # (if the metric is a scalar, the (single) key is the empty list).
+        # (if the metric is a scalar, the (single) key is the empty tuple).
         self.counts = {}
 
         # Scalar metrics are never empty
@@ -145,6 +145,36 @@ class CounterMetric(BaseMetric):
         )
 
 
+class GaugeMetric(BaseMetric):
+    """A metric that can go up or down
+    """
+
+    def __init__(self, *args, **kwargs):
+        super(GaugeMetric, self).__init__(*args, **kwargs)
+
+        # dict[list[str]]: value for each set of label values. the keys are the
+        # label values, in the same order as the labels in self.labels.
+        #
+        # (if the metric is a scalar, the (single) key is the empty tuple).
+        self.guages = {}
+
+    def set(self, v, *values):
+        if len(values) != self.dimension():
+            raise ValueError(
+                "Expected as many values to inc() as labels (%d)" % (self.dimension())
+            )
+
+        # TODO: should assert that the tag values are all strings
+
+        self.guages[values] = v
+
+    def render(self):
+        return flatten(
+            self._render_for_labels(k, self.guages[k])
+            for k in sorted(self.guages.keys())
+        )
+
+
 class CallbackMetric(BaseMetric):
     """A metric that returns the numeric value returned by a callback whenever
     it is rendered. Typically this is used to implement gauges that yield the
@@ -269,3 +299,29 @@ class MemoryUsageMetric(object):
             "process_psutil_rss:total %d" % sum_rss,
             "process_psutil_rss:count %d" % len_rss,
         ]
+
+
+def _escape_character(m):
+    """Replaces a single character with its escape sequence.
+
+    Args:
+        m (re.MatchObject): A match object whose first group is the single
+            character to replace
+
+    Returns:
+        str
+    """
+    c = m.group(1)
+    if c == "\\":
+        return "\\\\"
+    elif c == "\"":
+        return "\\\""
+    elif c == "\n":
+        return "\\n"
+    return c
+
+
+def _escape_label_value(value):
+    """Takes a label value and escapes quotes, newlines and backslashes
+    """
+    return re.sub(r"([\n\"\\])", _escape_character, value)

synapse/notifier.py

@@ -14,14 +14,17 @@
 # limitations under the License.
 
 from twisted.internet import defer
+
 from synapse.api.constants import EventTypes, Membership
 from synapse.api.errors import AuthError
 from synapse.handlers.presence import format_user_presence_state
 
-from synapse.util import DeferredTimedOutError
 from synapse.util.logutils import log_function
-from synapse.util.async import ObservableDeferred
+from synapse.util.async import (
-from synapse.util.logcontext import PreserveLoggingContext, preserve_fn
+    ObservableDeferred, add_timeout_to_deferred,
+    DeferredTimeoutError,
+)
+from synapse.util.logcontext import PreserveLoggingContext, run_in_background
 from synapse.util.metrics import Measure
 from synapse.types import StreamToken
 from synapse.visibility import filter_events_for_client
@@ -144,6 +147,7 @@ class _NotifierUserStream(object):
 class EventStreamResult(namedtuple("EventStreamResult", ("events", "tokens"))):
     def __nonzero__(self):
         return bool(self.events)
+    __bool__ = __nonzero__  # python3
 
 
 class Notifier(object):
@@ -250,9 +254,7 @@ class Notifier(object):
     def _on_new_room_event(self, event, room_stream_id, extra_users=[]):
         """Notify any user streams that are interested in this room event"""
         # poke any interested application service.
-        preserve_fn(self.appservice_handler.notify_interested_services)(
+        run_in_background(self._notify_app_services, room_stream_id)
-            room_stream_id
-        )
 
         if self.federation_sender:
             self.federation_sender.notify_new_events(room_stream_id)
@@ -266,6 +268,13 @@ class Notifier(object):
             rooms=[event.room_id],
         )
 
+    @defer.inlineCallbacks
+    def _notify_app_services(self, room_stream_id):
+        try:
+            yield self.appservice_handler.notify_interested_services(room_stream_id)
+        except Exception:
+            logger.exception("Error notifying application services of event")
+
     def on_new_event(self, stream_key, new_token, users=[], rooms=[]):
         """ Used to inform listeners that something has happend event wise.
 
@@ -330,11 +339,12 @@ class Notifier(object):
                     # Now we wait for the _NotifierUserStream to be told there
                     # is a new token.
                     listener = user_stream.new_listener(prev_token)
-                    with PreserveLoggingContext():
+                    add_timeout_to_deferred(
-                        yield self.clock.time_bound_deferred(
                         listener.deferred,
-                            time_out=(end_time - now) / 1000.
+                        (end_time - now) / 1000.,
                     )
+                    with PreserveLoggingContext():
+                        yield listener.deferred
 
                     current_token = user_stream.current_token
 
@@ -345,7 +355,7 @@ class Notifier(object):
                     # Update the prev_token to the current_token since nothing
                     # has happened between the old prev_token and the current_token
                     prev_token = current_token
-                except DeferredTimedOutError:
+                except DeferredTimeoutError:
                     break
                 except defer.CancelledError:
                     break
@@ -550,13 +560,14 @@ class Notifier(object):
             if end_time <= now:
                 break
 
+            add_timeout_to_deferred(
+                listener.deferred.addTimeout,
+                (end_time - now) / 1000.,
+            )
             try:
                 with PreserveLoggingContext():
-                    yield self.clock.time_bound_deferred(
+                    yield listener.deferred
-                        listener.deferred,
+            except DeferredTimeoutError:
-                        time_out=(end_time - now) / 1000.
-                    )
-            except DeferredTimedOutError:
                 break
             except defer.CancelledError:
                 break

synapse/push/emailpusher.py

@@ -77,10 +77,13 @@ class EmailPusher(object):
     @defer.inlineCallbacks
     def on_started(self):
         if self.mailer is not None:
+            try:
                 self.throttle_params = yield self.store.get_throttle_params_by_room(
                     self.pusher_id
                 )
                 yield self._process()
+            except Exception:
+                logger.exception("Error starting email pusher")
 
     def on_stop(self):
         if self.timed_call:

synapse/push/httppusher.py

@@ -18,8 +18,8 @@ import logging
 from twisted.internet import defer, reactor
 from twisted.internet.error import AlreadyCalled, AlreadyCancelled
 
-import push_rule_evaluator
+from . import push_rule_evaluator
-import push_tools
+from . import push_tools
 import synapse
 from synapse.push import PusherConfigException
 from synapse.util.logcontext import LoggingContext
@@ -94,7 +94,10 @@ class HttpPusher(object):
 
     @defer.inlineCallbacks
     def on_started(self):
+        try:
             yield self._process()
+        except Exception:
+            logger.exception("Error starting http pusher")
 
     @defer.inlineCallbacks
     def on_new_notifications(self, min_stream_ordering, max_stream_ordering):

synapse/push/pusher.py

@@ -13,7 +13,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from httppusher import HttpPusher
+from .httppusher import HttpPusher
 
 import logging
 logger = logging.getLogger(__name__)

synapse/push/pusherpool.py

@@ -14,13 +14,13 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+import logging
+
 from twisted.internet import defer
 
-from .pusher import PusherFactory
+from synapse.push.pusher import PusherFactory
-from synapse.util.logcontext import make_deferred_yieldable, preserve_fn
 from synapse.util.async import run_on_reactor
-
+from synapse.util.logcontext import make_deferred_yieldable, run_in_background
-import logging
 
 logger = logging.getLogger(__name__)
 
@@ -137,12 +137,15 @@ class PusherPool:
                 if u in self.pushers:
                     for p in self.pushers[u].values():
                         deferreds.append(
-                            preserve_fn(p.on_new_notifications)(
+                            run_in_background(
-                                min_stream_id, max_stream_id
+                                p.on_new_notifications,
+                                min_stream_id, max_stream_id,
                             )
                         )
 
-            yield make_deferred_yieldable(defer.gatherResults(deferreds))
+            yield make_deferred_yieldable(
+                defer.gatherResults(deferreds, consumeErrors=True),
+            )
         except Exception:
             logger.exception("Exception in pusher on_new_notifications")
 
@@ -164,10 +167,15 @@ class PusherPool:
                 if u in self.pushers:
                     for p in self.pushers[u].values():
                         deferreds.append(
-                            preserve_fn(p.on_new_receipts)(min_stream_id, max_stream_id)
+                            run_in_background(
+                                p.on_new_receipts,
+                                min_stream_id, max_stream_id,
+                            )
                         )
 
-            yield make_deferred_yieldable(defer.gatherResults(deferreds))
+            yield make_deferred_yieldable(
+                defer.gatherResults(deferreds, consumeErrors=True),
+            )
         except Exception:
             logger.exception("Exception in pusher on_new_receipts")
 
@@ -207,7 +215,7 @@ class PusherPool:
                 if appid_pushkey in byuser:
                     byuser[appid_pushkey].on_stop()
                 byuser[appid_pushkey] = p
-                preserve_fn(p.on_started)()
+                run_in_background(p.on_started)
 
         logger.info("Started pushers")
 

synapse/python_dependencies.py

@@ -1,5 +1,6 @@
 # Copyright 2015, 2016 OpenMarket Ltd
 # Copyright 2017 Vector Creations Ltd
+# Copyright 2018 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.
@@ -18,28 +19,43 @@ from distutils.version import LooseVersion
 
 logger = logging.getLogger(__name__)
 
+# this dict maps from python package name to a list of modules we expect it to
+# provide.
+#
+# the key is a "requirement specifier", as used as a parameter to `pip
+# install`[1], or an `install_requires` argument to `setuptools.setup` [2].
+#
+# the value is a sequence of strings; each entry should be the name of the
+# python module, optionally followed by a version assertion which can be either
+# ">=<ver>" or "==<ver>".
+#
+# [1] https://pip.pypa.io/en/stable/reference/pip_install/#requirement-specifiers.
+# [2] https://setuptools.readthedocs.io/en/latest/setuptools.html#declaring-dependencies
 REQUIREMENTS = {
     "jsonschema>=2.5.1": ["jsonschema>=2.5.1"],
     "frozendict>=0.4": ["frozendict"],
     "unpaddedbase64>=1.1.0": ["unpaddedbase64>=1.1.0"],
-    "canonicaljson>=1.0.0": ["canonicaljson>=1.0.0"],
+    "canonicaljson>=1.1.3": ["canonicaljson>=1.1.3"],
     "signedjson>=1.0.0": ["signedjson>=1.0.0"],
     "pynacl>=1.2.1": ["nacl>=1.2.1", "nacl.bindings"],
     "service_identity>=1.0.0": ["service_identity>=1.0.0"],
     "Twisted>=16.0.0": ["twisted>=16.0.0"],
-    "pyopenssl>=0.14": ["OpenSSL>=0.14"],
+
+    # We use crypto.get_elliptic_curve which is only supported in >=0.15
+    "pyopenssl>=0.15": ["OpenSSL>=0.15"],
+
     "pyyaml": ["yaml"],
     "pyasn1": ["pyasn1"],
     "daemonize": ["daemonize"],
     "bcrypt": ["bcrypt>=3.1.0"],
     "pillow": ["PIL"],
     "pydenticon": ["pydenticon"],
-    "ujson": ["ujson"],
     "blist": ["blist"],
     "pysaml2>=3.0.0": ["saml2>=3.0.0"],
     "pymacaroons-pynacl": ["pymacaroons"],
     "msgpack-python>=0.3.0": ["msgpack"],
     "phonenumbers>=8.2.0": ["phonenumbers"],
+    "six": ["six"],
 }
 CONDITIONAL_REQUIREMENTS = {
     "web_client": {

synapse/replication/http/send_event.py

@@ -23,7 +23,6 @@ from synapse.events.snapshot import EventContext
 from synapse.http.servlet import RestServlet, parse_json_object_from_request
 from synapse.util.async import sleep
 from synapse.util.caches.response_cache import ResponseCache
-from synapse.util.logcontext import make_deferred_yieldable, preserve_fn
 from synapse.util.metrics import Measure
 from synapse.types import Requester, UserID
 
@@ -115,20 +114,15 @@ class ReplicationSendEventRestServlet(RestServlet):
         self.clock = hs.get_clock()
 
         # The responses are tiny, so we may as well cache them for a while
-        self.response_cache = ResponseCache(hs, timeout_ms=30 * 60 * 1000)
+        self.response_cache = ResponseCache(hs, "send_event", timeout_ms=30 * 60 * 1000)
 
     def on_PUT(self, request, event_id):
-        result = self.response_cache.get(event_id)
+        return self.response_cache.wrap(
-        if not result:
-            result = self.response_cache.set(
             event_id,
-                self._handle_request(request)
+            self._handle_request,
+            request
         )
-        else:
-            logger.warn("Returning cached response")
-        return make_deferred_yieldable(result)
 
-    @preserve_fn
     @defer.inlineCallbacks
     def _handle_request(self, request):
         with Measure(self.clock, "repl_send_event_parse"):

synapse/replication/tcp/commands.py

@@ -19,11 +19,13 @@ allowed to be sent by which side.
 """
 
 import logging
-import ujson as json
+import simplejson
 
 
 logger = logging.getLogger(__name__)
 
+_json_encoder = simplejson.JSONEncoder(namedtuple_as_object=False)
+
 
 class Command(object):
     """The base command class.
@@ -100,14 +102,14 @@ class RdataCommand(Command):
         return cls(
             stream_name,
             None if token == "batch" else int(token),
-            json.loads(row_json)
+            simplejson.loads(row_json)
         )
 
     def to_line(self):
         return " ".join((
             self.stream_name,
             str(self.token) if self.token is not None else "batch",
-            json.dumps(self.row),
+            _json_encoder.encode(self.row),
         ))
 
 
@@ -298,10 +300,12 @@ class InvalidateCacheCommand(Command):
     def from_line(cls, line):
         cache_func, keys_json = line.split(" ", 1)
 
-        return cls(cache_func, json.loads(keys_json))
+        return cls(cache_func, simplejson.loads(keys_json))
 
     def to_line(self):
-        return " ".join((self.cache_func, json.dumps(self.keys)))
+        return " ".join((
+            self.cache_func, _json_encoder.encode(self.keys),
+        ))
 
 
 class UserIpCommand(Command):
@@ -325,14 +329,14 @@ class UserIpCommand(Command):
     def from_line(cls, line):
         user_id, jsn = line.split(" ", 1)
 
-        access_token, ip, user_agent, device_id, last_seen = json.loads(jsn)
+        access_token, ip, user_agent, device_id, last_seen = simplejson.loads(jsn)
 
         return cls(
             user_id, access_token, ip, user_agent, device_id, last_seen
         )
 
     def to_line(self):
-        return self.user_id + " " + json.dumps((
+        return self.user_id + " " + _json_encoder.encode((
             self.access_token, self.ip, self.user_agent, self.device_id,
             self.last_seen,
         ))

synapse/replication/tcp/protocol.py

@@ -53,12 +53,12 @@ from twisted.internet import defer
 from twisted.protocols.basic import LineOnlyReceiver
 from twisted.python.failure import Failure
 
-from commands import (
+from .commands import (
     COMMAND_MAP, VALID_CLIENT_COMMANDS, VALID_SERVER_COMMANDS,
     ErrorCommand, ServerCommand, RdataCommand, PositionCommand, PingCommand,
     NameCommand, ReplicateCommand, UserSyncCommand, SyncCommand,
 )
-from streams import STREAMS_MAP
+from .streams import STREAMS_MAP
 
 from synapse.util.stringutils import random_string
 from synapse.metrics.metric import CounterMetric

synapse/replication/tcp/resource.py

@@ -18,8 +18,8 @@
 from twisted.internet import defer, reactor
 from twisted.internet.protocol import Factory
 
-from streams import STREAMS_MAP, FederationStream
+from .streams import STREAMS_MAP, FederationStream
-from protocol import ServerReplicationStreamProtocol
+from .protocol import ServerReplicationStreamProtocol
 
 from synapse.util.metrics import Measure, measure_func
 

synapse/rest/client/v1/admin.py

@@ -168,11 +168,24 @@ class PurgeHistoryRestServlet(ClientV1RestServlet):
                 yield self.store.find_first_stream_ordering_after_ts(ts)
             )
 
-            (_, depth, _) = (
+            room_event_after_stream_ordering = (
                 yield self.store.get_room_event_after_stream_ordering(
                     room_id, stream_ordering,
                 )
             )
+            if room_event_after_stream_ordering:
+                (_, depth, _) = room_event_after_stream_ordering
+            else:
+                logger.warn(
+                    "[purge] purging events not possible: No event found "
+                    "(received_ts %i => stream_ordering %i)",
+                    ts, stream_ordering,
+                )
+                raise SynapseError(
+                    404,
+                    "there is no event to be purged",
+                    errcode=Codes.NOT_FOUND,
+                )
             logger.info(
                 "[purge] purging up to depth %i (received_ts %i => "
                 "stream_ordering %i)",

synapse/rest/client/v1/base.py

@@ -52,6 +52,10 @@ class ClientV1RestServlet(RestServlet):
     """A base Synapse REST Servlet for the client version 1 API.
     """
 
+    # This subclass was presumably created to allow the auth for the v1
+    # protocol version to be different, however this behaviour was removed.
+    # it may no longer be necessary
+
     def __init__(self, hs):
         """
         Args:
@@ -59,5 +63,5 @@ class ClientV1RestServlet(RestServlet):
         """
         self.hs = hs
         self.builder_factory = hs.get_event_builder_factory()
-        self.auth = hs.get_v1auth()
+        self.auth = hs.get_auth()
         self.txns = HttpTransactionCache(hs.get_clock())

synapse/rest/client/v1/login.py

@@ -25,7 +25,7 @@ from .base import ClientV1RestServlet, client_path_patterns
 
 import simplejson as json
 import urllib
-import urlparse
+from six.moves.urllib import parse as urlparse
 
 import logging
 from saml2 import BINDING_HTTP_POST

synapse/rest/client/v1/logout.py

@@ -44,7 +44,10 @@ class LogoutRestServlet(ClientV1RestServlet):
             requester = yield self.auth.get_user_by_req(request)
         except AuthError:
             # this implies the access token has already been deleted.
-            pass
+            defer.returnValue((401, {
+                "errcode": "M_UNKNOWN_TOKEN",
+                "error": "Access Token unknown or expired"
+            }))
         else:
             if requester.device_id is None:
                 # the acccess token wasn't associated with a device.

synapse/rest/client/v1/pusher.py

@@ -150,7 +150,7 @@ class PushersRemoveRestServlet(RestServlet):
         super(RestServlet, self).__init__()
         self.hs = hs
         self.notifier = hs.get_notifier()
-        self.auth = hs.get_v1auth()
+        self.auth = hs.get_auth()
         self.pusher_pool = self.hs.get_pusherpool()
 
     @defer.inlineCallbacks

synapse/rest/client/v1/register.py

@@ -30,6 +30,8 @@ from hashlib import sha1
 import hmac
 import logging
 
+from six import string_types
+
 logger = logging.getLogger(__name__)
 
 
@@ -333,11 +335,11 @@ class RegisterRestServlet(ClientV1RestServlet):
     def _do_shared_secret(self, request, register_json, session):
         yield run_on_reactor()
 
-        if not isinstance(register_json.get("mac", None), basestring):
+        if not isinstance(register_json.get("mac", None), string_types):
             raise SynapseError(400, "Expected mac.")
-        if not isinstance(register_json.get("user", None), basestring):
+        if not isinstance(register_json.get("user", None), string_types):
             raise SynapseError(400, "Expected 'user' key.")
-        if not isinstance(register_json.get("password", None), basestring):
+        if not isinstance(register_json.get("password", None), string_types):
             raise SynapseError(400, "Expected 'password' key.")
 
         if not self.hs.config.registration_shared_secret:
@@ -348,9 +350,9 @@ class RegisterRestServlet(ClientV1RestServlet):
         admin = register_json.get("admin", None)
 
         # Its important to check as we use null bytes as HMAC field separators
-        if "\x00" in user:
+        if b"\x00" in user:
             raise SynapseError(400, "Invalid user")
-        if "\x00" in password:
+        if b"\x00" in password:
             raise SynapseError(400, "Invalid password")
 
         # str() because otherwise hmac complains that 'unicode' does not
@@ -358,14 +360,14 @@ class RegisterRestServlet(ClientV1RestServlet):
         got_mac = str(register_json["mac"])
 
         want_mac = hmac.new(
-            key=self.hs.config.registration_shared_secret,
+            key=self.hs.config.registration_shared_secret.encode(),
             digestmod=sha1,
         )
         want_mac.update(user)
-        want_mac.update("\x00")
+        want_mac.update(b"\x00")
         want_mac.update(password)
-        want_mac.update("\x00")
+        want_mac.update(b"\x00")
-        want_mac.update("admin" if admin else "notadmin")
+        want_mac.update(b"admin" if admin else b"notadmin")
         want_mac = want_mac.hexdigest()
 
         if compare_digest(want_mac, got_mac):

synapse/rest/client/v1/room.py

@@ -28,9 +28,10 @@ from synapse.http.servlet import (
     parse_json_object_from_request, parse_string, parse_integer
 )
 
+from six.moves.urllib import parse as urlparse
+
 import logging
-import urllib
+import simplejson as json
-import ujson as json
 
 logger = logging.getLogger(__name__)
 
@@ -165,17 +166,12 @@ class RoomStateEventRestServlet(ClientV1RestServlet):
                 content=content,
             )
         else:
-            event, context = yield self.event_creation_hander.create_event(
+            event = yield self.event_creation_hander.create_and_send_nonmember_event(
                 requester,
                 event_dict,
-                token_id=requester.access_token_id,
                 txn_id=txn_id,
             )
 
-            yield self.event_creation_hander.send_nonmember_event(
-                requester, event, context,
-            )
-
         ret = {}
         if event:
             ret = {"event_id": event.event_id}
@@ -438,7 +434,7 @@ class RoomMessageListRestServlet(ClientV1RestServlet):
         as_client_event = "raw" not in request.args
         filter_bytes = request.args.get("filter", None)
         if filter_bytes:
-            filter_json = urllib.unquote(filter_bytes[-1]).decode("UTF-8")
+            filter_json = urlparse.unquote(filter_bytes[-1]).decode("UTF-8")
             event_filter = Filter(json.loads(filter_json))
         else:
             event_filter = None
@@ -655,7 +651,12 @@ class RoomMembershipRestServlet(ClientV1RestServlet):
             content=event_content,
         )
 
-        defer.returnValue((200, {}))
+        return_value = {}
+
+        if membership_action == "join":
+            return_value["room_id"] = room_id
+
+        defer.returnValue((200, return_value))
 
     def _has_3pid_invite_keys(self, content):
         for key in {"id_server", "medium", "address"}:
@@ -718,8 +719,8 @@ class RoomTypingRestServlet(ClientV1RestServlet):
     def on_PUT(self, request, room_id, user_id):
         requester = yield self.auth.get_user_by_req(request)
 
-        room_id = urllib.unquote(room_id)
+        room_id = urlparse.unquote(room_id)
-        target_user = UserID.from_string(urllib.unquote(user_id))
+        target_user = UserID.from_string(urlparse.unquote(user_id))
 
         content = parse_json_object_from_request(request)
 

synapse/rest/client/v2_alpha/groups.py

@@ -1,5 +1,6 @@
 # -*- coding: utf-8 -*-
 # Copyright 2017 Vector Creations Ltd
+# Copyright 2018 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.
@@ -401,6 +402,32 @@ class GroupInvitedUsersServlet(RestServlet):
         defer.returnValue((200, result))
 
 
+class GroupSettingJoinPolicyServlet(RestServlet):
+    """Set group join policy
+    """
+    PATTERNS = client_v2_patterns("/groups/(?P<group_id>[^/]*)/settings/m.join_policy$")
+
+    def __init__(self, hs):
+        super(GroupSettingJoinPolicyServlet, self).__init__()
+        self.auth = hs.get_auth()
+        self.groups_handler = hs.get_groups_local_handler()
+
+    @defer.inlineCallbacks
+    def on_PUT(self, request, group_id):
+        requester = yield self.auth.get_user_by_req(request)
+        requester_user_id = requester.user.to_string()
+
+        content = parse_json_object_from_request(request)
+
+        result = yield self.groups_handler.set_group_join_policy(
+            group_id,
+            requester_user_id,
+            content,
+        )
+
+        defer.returnValue((200, result))
+
+
 class GroupCreateServlet(RestServlet):
     """Create a group
     """
@@ -738,6 +765,7 @@ def register_servlets(hs, http_server):
     GroupInvitedUsersServlet(hs).register(http_server)
     GroupUsersServlet(hs).register(http_server)
     GroupRoomServlet(hs).register(http_server)
+    GroupSettingJoinPolicyServlet(hs).register(http_server)
     GroupCreateServlet(hs).register(http_server)
     GroupAdminRoomsServlet(hs).register(http_server)
     GroupAdminRoomsConfigServlet(hs).register(http_server)

synapse/rest/client/v2_alpha/register.py

@@ -20,7 +20,6 @@ import synapse
 import synapse.types
 from synapse.api.auth import get_access_token_from_request, has_access_token
 from synapse.api.constants import LoginType
-from synapse.types import RoomID, RoomAlias
 from synapse.api.errors import SynapseError, Codes, UnrecognizedRequestError
 from synapse.http.servlet import (
     RestServlet, parse_json_object_from_request, assert_params_in_request, parse_string
@@ -36,6 +35,8 @@ from hashlib import sha1
 from synapse.util.async import run_on_reactor
 from synapse.util.ratelimitutils import FederationRateLimiter
 
+from six import string_types
+
 
 # We ought to be using hmac.compare_digest() but on older pythons it doesn't
 # exist. It's a _really minor_ security flaw to use plain string comparison
@@ -211,14 +212,14 @@ class RegisterRestServlet(RestServlet):
         # in sessions. Pull out the username/password provided to us.
         desired_password = None
         if 'password' in body:
-            if (not isinstance(body['password'], basestring) or
+            if (not isinstance(body['password'], string_types) or
                     len(body['password']) > 512):
                 raise SynapseError(400, "Invalid password")
             desired_password = body["password"]
 
         desired_username = None
         if 'username' in body:
-            if (not isinstance(body['username'], basestring) or
+            if (not isinstance(body['username'], string_types) or
                     len(body['username']) > 512):
                 raise SynapseError(400, "Invalid username")
             desired_username = body['username']
@@ -244,7 +245,7 @@ class RegisterRestServlet(RestServlet):
 
             access_token = get_access_token_from_request(request)
 
-            if isinstance(desired_username, basestring):
+            if isinstance(desired_username, string_types):
                 result = yield self._do_appservice_registration(
                     desired_username, access_token, body
                 )
@@ -405,14 +406,6 @@ class RegisterRestServlet(RestServlet):
                 generate_token=False,
             )
 
-            # auto-join the user to any rooms we're supposed to dump them into
-            fake_requester = synapse.types.create_requester(registered_user_id)
-            for r in self.hs.config.auto_join_rooms:
-                try:
-                    yield self._join_user_to_room(fake_requester, r)
-                except Exception as e:
-                    logger.error("Failed to join new user to %r: %r", r, e)
-
             # remember that we've now registered that user account, and with
             #  what user ID (since the user may not have specified)
             self.auth_handler.set_session_data(
@@ -445,29 +438,6 @@ class RegisterRestServlet(RestServlet):
     def on_OPTIONS(self, _):
         return 200, {}
 
-    @defer.inlineCallbacks
-    def _join_user_to_room(self, requester, room_identifier):
-        room_id = None
-        if RoomID.is_valid(room_identifier):
-            room_id = room_identifier
-        elif RoomAlias.is_valid(room_identifier):
-            room_alias = RoomAlias.from_string(room_identifier)
-            room_id, remote_room_hosts = (
-                yield self.room_member_handler.lookup_room_alias(room_alias)
-            )
-            room_id = room_id.to_string()
-        else:
-            raise SynapseError(400, "%s was not legal room ID or room alias" % (
-                room_identifier,
-            ))
-
-        yield self.room_member_handler.update_membership(
-            requester=requester,
-            target=requester.user,
-            room_id=room_id,
-            action="join",
-        )
-
     @defer.inlineCallbacks
     def _do_appservice_registration(self, username, as_token, body):
         user_id = yield self.registration_handler.appservice_register(
@@ -496,7 +466,7 @@ class RegisterRestServlet(RestServlet):
         # includes the password and admin flag in the hashed text. Why are
         # these different?
         want_mac = hmac.new(
-            key=self.hs.config.registration_shared_secret,
+            key=self.hs.config.registration_shared_secret.encode(),
             msg=user,
             digestmod=sha1,
         ).hexdigest()

synapse/rest/client/v2_alpha/sync.py

@@ -33,7 +33,7 @@ from ._base import set_timeline_upper_limit
 import itertools
 import logging
 
-import ujson as json
+import simplejson as json
 
 logger = logging.getLogger(__name__)
 

synapse/rest/media/v1/_base.py

@@ -28,7 +28,7 @@ import os
 
 import logging
 import urllib
-import urlparse
+from six.moves.urllib import parse as urlparse
 
 logger = logging.getLogger(__name__)
 
@@ -143,6 +143,7 @@ def respond_with_responder(request, responder, media_type, file_size, upload_nam
         respond_404(request)
         return
 
+    logger.debug("Responding to media request with responder %s")
     add_file_headers(request, media_type, file_size, upload_name)
     with responder:
         yield responder.write_to_consumer(request)

synapse/rest/media/v1/media_repository.py

@@ -47,7 +47,7 @@ import shutil
 
 import cgi
 import logging
-import urlparse
+from six.moves.urllib import parse as urlparse
 
 logger = logging.getLogger(__name__)
 

synapse/rest/media/v1/media_storage.py

@@ -16,6 +16,8 @@
 from twisted.internet import defer, threads
 from twisted.protocols.basic import FileSender
 
+import six
+
 from ._base import Responder
 
 from synapse.util.file_consumer import BackgroundFileConsumer
@@ -119,7 +121,7 @@ class MediaStorage(object):
                 os.remove(fname)
             except Exception:
                 pass
-            raise t, v, tb
+            six.reraise(t, v, tb)
 
         if not finished_called:
             raise Exception("Finished callback not called")
@@ -253,7 +255,9 @@ class FileResponder(Responder):
         self.open_file = open_file
 
     def write_to_consumer(self, consumer):
-        return FileSender().beginFileTransfer(self.open_file, consumer)
+        return make_deferred_yieldable(
+            FileSender().beginFileTransfer(self.open_file, consumer)
+        )
 
     def __exit__(self, exc_type, exc_val, exc_tb):
         self.open_file.close()

synapse/rest/media/v1/preview_url_resource.py

@@ -23,7 +23,7 @@ import re
 import shutil
 import sys
 import traceback
-import ujson as json
+import simplejson as json
 import urlparse
 
 from twisted.web.server import NOT_DONE_YET
@@ -35,7 +35,7 @@ from ._base import FileInfo
 from synapse.api.errors import (
     SynapseError, Codes,
 )
-from synapse.util.logcontext import preserve_fn, make_deferred_yieldable
+from synapse.util.logcontext import make_deferred_yieldable, run_in_background
 from synapse.util.stringutils import random_string
 from synapse.util.caches.expiringcache import ExpiringCache
 from synapse.http.client import SpiderHttpClient
@@ -144,7 +144,8 @@ class PreviewUrlResource(Resource):
         observable = self._cache.get(url)
 
         if not observable:
-            download = preserve_fn(self._do_preview)(
+            download = run_in_background(
+                self._do_preview,
                 url, requester.user, ts,
             )
             observable = ObservableDeferred(

synapse/rest/media/v1/storage_provider.py

@@ -18,7 +18,7 @@ from twisted.internet import defer, threads
 from .media_storage import FileResponder
 
 from synapse.config._base import Config
-from synapse.util.logcontext import preserve_fn
+from synapse.util.logcontext import run_in_background
 
 import logging
 import os
@@ -87,7 +87,12 @@ class StorageProviderWrapper(StorageProvider):
             return self.backend.store_file(path, file_info)
         else:
             # TODO: Handle errors.
-            preserve_fn(self.backend.store_file)(path, file_info)
+            def store():
+                try:
+                    return self.backend.store_file(path, file_info)
+                except Exception:
+                    logger.exception("Error storing file")
+            run_in_background(store)
             return defer.succeed(None)
 
     def fetch(self, path, file_info):