SECURITY UPDATE: Prevent unauthorised users from setting state events in a room

when there is no `m.room.power_levels` event in force in the room. (PR #3397)
 
 Discussion around the Matrix Spec change proposal for this change can be
 followed at https://github.com/matrix-org/matrix-doc/issues/1304.
 -----BEGIN PGP SIGNATURE-----
 Version: GnuPG v1
 
 iQEcBAABAgAGBQJbIop9AAoJEIofk9V1tejV9lsIAJVH0l5dXROmy1KH/zt16AUA
 CXa6Vv4Vyo6hKad/fZ81OZVRr5ChK/TvbIJVn/SA/muCfdoIFdxhT8eo/pXzO2UW
 zReuLsDhAg+gSvpNus37oWj2FVsAE1HYDZ60lfaapAdZnkFit68d5DQZjO6nZHHA
 YUXcU3GUwj0ZYuUzFzYKMLu6uNNasNkN8h6SS2lF7Bm4JaKDW+mFMfCyJwdIVSEh
 BGhHoVpXdxFysD9s6Mwxqrz3KKg1Jtp7idDkk0x2S2Eh+gxyiDQQokv0oQ3+0+HG
 sgy5Iz2t2CkpS02/j+LOvAZljTmnD0bXu3srGR+25StsoDFP038Am3bfQwtD190=
 =9jsT
 -----END PGP SIGNATURE-----

Merge tag 'v0.31.2' into dinsic

.dockerignore

@@ -0,0 +1,5 @@
+Dockerfile
+.travis.yml
+.gitignore
+demo/etc
+tox.ini

.gitignore

@@ -1,5 +1,6 @@
 *.pyc
 .*.swp
+*~
 
 .DS_Store
 _trial_temp/
@@ -13,6 +14,7 @@ docs/build/
 cmdclient_config.json
 homeserver*.db
 homeserver*.log
+homeserver*.log.*
 homeserver*.pid
 homeserver*.yaml
 
@@ -32,6 +34,7 @@ demo/media_store.*
 demo/etc
 
 uploads
+cache
 
 .idea/
 media_store/
@@ -39,6 +42,7 @@ media_store/
 *.tac
 
 build/
+venv/
 
 localhost-800*/
 static/client/register/register_config.js
@@ -48,3 +52,4 @@ env/
 *.config
 
 .vscode/
+.ropeproject/

.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

AUTHORS.rst

@@ -60,3 +60,6 @@ Niklas Riekenbrauck <nikriek at gmail dot.com>
 
 Christoph Witzany <christoph at web.crofting.com>
  * Add LDAP support for authentication
+
+Pierre Jaury <pierre at jaury.eu>
+* Docker packaging

CHANGES.rst

@@ -1,3 +1,226 @@
+Changes in synapse v0.31.2 (2018-06-14)
+=======================================
+
+SECURITY UPDATE: Prevent unauthorised users from setting state events in a room
+when there is no ``m.room.power_levels`` event in force in the room. (PR #3397)
+
+Discussion around the Matrix Spec change proposal for this change can be
+followed at https://github.com/matrix-org/matrix-doc/issues/1304.
+
+Changes in synapse v0.31.1 (2018-06-08)
+=======================================
+
+v0.31.1 fixes a security bug in the ``get_missing_events`` federation API
+where event visibility rules were not applied correctly.
+
+We are not aware of it being actively exploited but please upgrade asap.
+
+Bug Fixes:
+
+* Fix event filtering in get_missing_events handler (PR #3371)
+
+Changes in synapse v0.31.0 (2018-06-06)
+=======================================
+
+Most notable change from v0.30.0 is to switch to the python prometheus library to improve system
+stats reporting. WARNING: this changes a number of prometheus metrics in a
+backwards-incompatible manner. For more details, see
+`docs/metrics-howto.rst <docs/metrics-howto.rst#removal-of-deprecated-metrics--time-based-counters-becoming-histograms-in-0310>`_.
+
+Bug Fixes:
+
+* Fix metric documentation tables (PR #3341)
+* Fix LaterGauge error handling (694968f)
+* Fix replication metrics (b7e7fd2)
+
+Changes in synapse v0.31.0-rc1 (2018-06-04)
+==========================================
+
+Features:
+
+* Switch to the Python Prometheus library (PR #3256, #3274)
+* Let users leave the server notice room after joining (PR #3287)
+
+
+Changes:
+
+* daily user type phone home stats (PR #3264)
+* Use iter* methods for _filter_events_for_server (PR #3267)
+* Docs on consent bits (PR #3268)
+* Remove users from user directory on deactivate (PR #3277)
+* Avoid sending consent notice to guest users (PR #3288)
+* disable CPUMetrics if no /proc/self/stat (PR #3299)
+* Consistently use six's iteritems and wrap lazy keys/values in list() if they're not meant to be lazy (PR #3307)
+* Add private IPv6 addresses to example config for url preview blacklist (PR #3317) Thanks to @thegcat!
+* Reduce stuck read-receipts: ignore depth when updating (PR #3318)
+* Put python's logs into Trial when running unit tests (PR #3319)
+
+Changes, python 3 migration:
+
+* Replace some more comparisons with six (PR #3243) Thanks to @NotAFile!
+* replace some iteritems with six (PR #3244) Thanks to @NotAFile!
+* Add batch_iter to utils (PR #3245) Thanks to @NotAFile!
+* use repr, not str (PR #3246) Thanks to @NotAFile!
+* Misc Python3 fixes (PR #3247) Thanks to @NotAFile!
+* Py3 storage/_base.py (PR #3278) Thanks to @NotAFile!
+* more six iteritems (PR #3279) Thanks to @NotAFile!
+* More Misc. py3 fixes (PR #3280) Thanks to @NotAFile!
+* remaining isintance fixes (PR #3281) Thanks to @NotAFile!
+* py3-ize state.py (PR #3283) Thanks to @NotAFile!
+* extend tox testing for py3 to avoid regressions (PR #3302) Thanks to @krombel!
+* use memoryview in py3 (PR #3303) Thanks to @NotAFile!
+
+Bugs:
+
+* Fix federation backfill bugs (PR #3261)
+* federation: fix LaterGauge usage (PR #3328) Thanks to @intelfx!
+
+
+Changes in synapse v0.30.0 (2018-05-24)
+==========================================
+
+'Server Notices' are a new feature introduced in Synapse 0.30. They provide a
+channel whereby server administrators can send messages to users on the server.
+
+They are used as part of communication of the server policies (see ``docs/consent_tracking.md``),
+however the intention is that they may also find a use for features such
+as "Message of the day".
+
+This feature is specific to Synapse, but uses standard Matrix communication mechanisms,
+so should work with any Matrix client. For more details see ``docs/server_notices.md``
+
+Further Server Notices/Consent Tracking Support:
+
+* Allow overriding the server_notices user's avatar (PR #3273)
+* Use the localpart in the consent uri (PR #3272)
+* Support for putting %(consent_uri)s in messages (PR #3271)
+* Block attempts to send server notices to remote users (PR #3270)
+* Docs on consent bits (PR #3268)
+
+
+
+Changes in synapse v0.30.0-rc1 (2018-05-23)
+==========================================
+
+Server Notices/Consent Tracking Support:
+
+* ConsentResource to gather policy consent from users (PR #3213)
+* Move RoomCreationHandler out of synapse.handlers.Handlers (PR #3225)
+* Infrastructure for a server notices room (PR #3232)
+* Send users a server notice about consent (PR #3236)
+* Reject attempts to send event before privacy consent is given (PR #3257)
+* Add a 'has_consented' template var to consent forms (PR #3262)
+* Fix dependency on jinja2 (PR #3263)
+
+Features:
+
+* Cohort analytics (PR #3163, #3241, #3251)
+* Add lxml to docker image for web previews (PR #3239) Thanks to @ptman!
+* Add in flight request metrics (PR #3252)
+
+Changes:
+
+* Remove unused `update_external_syncs` (PR #3233)
+* Use stream rather depth ordering for push actions (PR #3212)
+* Make purge_history operate on tokens (PR #3221)
+* Don't support limitless pagination (PR #3265)
+
+Bug Fixes:
+
+* Fix logcontext resource usage tracking (PR #3258)
+* Fix error in handling receipts (PR #3235)
+* Stop the transaction cache caching failures (PR #3255)
+
+
+Changes in synapse v0.29.1 (2018-05-17)
+==========================================
+Changes:
+
+* Update docker documentation (PR #3222)
+
+Changes in synapse v0.29.0 (2018-05-16)
+===========================================
+Not changes since v0.29.0-rc1
+
+Changes in synapse v0.29.0-rc1 (2018-05-14)
+===========================================
+
+Notable changes, a docker file for running Synapse (Thanks to @kaiyou!) and a
+closed spec bug in the Client Server API. Additionally further prep for Python 3
+migration.
+
+Potentially breaking change:
+
+* 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.
+
+Features:
+
+* Add a Dockerfile for synapse (PR #2846) Thanks to @kaiyou!
+
+Changes - General:
+
+* nuke-room-from-db.sh: added postgresql option and help (PR #2337) Thanks to @rubo77!
+* Part user from rooms on account deactivate (PR #3201)
+* Make 'unexpected logging context' into warnings (PR #3007)
+* Set Server header in SynapseRequest (PR #3208)
+* remove duplicates from groups tables (PR #3129)
+* Improve exception handling for background processes (PR #3138)
+* Add missing consumeErrors to improve exception handling (PR #3139)
+* reraise exceptions more carefully (PR #3142)
+* Remove redundant call to preserve_fn (PR #3143)
+* Trap exceptions thrown within run_in_background (PR #3144)
+
+Changes - Refactors:
+
+* Refactor /context to reuse pagination storage functions (PR #3193)
+* Refactor recent events func to use pagination func (PR #3195)
+* Refactor pagination DB API to return concrete type (PR #3196)
+* Refactor get_recent_events_for_room return type (PR #3198)
+* Refactor sync APIs to reuse pagination API (PR #3199)
+* Remove unused code path from member change DB func (PR #3200)
+* Refactor request handling wrappers (PR #3203)
+* transaction_id, destination defined twice (PR #3209) Thanks to @damir-manapov!
+* Refactor event storage to prepare for changes in state calculations (PR #3141)
+* Set Server header in SynapseRequest (PR #3208)
+* Use deferred.addTimeout instead of time_bound_deferred (PR #3127, #3178)
+* Use run_in_background in preference to preserve_fn (PR #3140)
+
+Changes - Python 3 migration:
+
+* Construct HMAC as bytes on py3 (PR #3156) Thanks to @NotAFile!
+* run config tests on py3 (PR #3159) Thanks to @NotAFile!
+* Open certificate files as bytes (PR #3084) Thanks to @NotAFile!
+* Open config file in non-bytes mode (PR #3085) Thanks to @NotAFile!
+* Make event properties raise AttributeError instead (PR #3102) Thanks to @NotAFile!
+* Use six.moves.urlparse (PR #3108) Thanks to @NotAFile!
+* Add py3 tests to tox with folders that work (PR #3145) Thanks to @NotAFile!
+* Don't yield in list comprehensions (PR #3150) Thanks to @NotAFile!
+* Move more xrange to six (PR #3151) Thanks to @NotAFile!
+* make imports local (PR #3152) Thanks to @NotAFile!
+* move httplib import to six (PR #3153) Thanks to @NotAFile!
+* Replace stringIO imports with six (PR #3154, #3168) Thanks to @NotAFile!
+* more bytes strings (PR #3155) Thanks to @NotAFile!
+
+Bug Fixes:
+
+* synapse fails to start under Twisted >= 18.4 (PR #3157)
+* Fix a class of logcontext leaks (PR #3170)
+* Fix a couple of logcontext leaks in unit tests (PR #3172)
+* Fix logcontext leak in media repo (PR #3174)
+* Escape label values in prometheus metrics (PR #3175, #3186)
+* Fix 'Unhandled Error' logs with Twisted 18.4 (PR #3182) Thanks to @Half-Shot!
+* Fix logcontext leaks in rate limiter (PR #3183)
+* notifications: Convert next_token to string according to the spec (PR #3190) Thanks to @mujx!
+* nuke-room-from-db.sh: fix deletion from search table (PR #3194) Thanks to @rubo77!
+* add guard for None on purge_history api (PR #3160) Thanks to @krombel!
+
 Changes in synapse v0.28.1 (2018-05-01)
 =======================================
 

Dockerfile

@@ -0,0 +1,19 @@
+FROM docker.io/python:2-alpine3.7
+
+RUN apk add --no-cache --virtual .nacl_deps su-exec build-base libffi-dev zlib-dev libressl-dev libjpeg-turbo-dev linux-headers postgresql-dev libxslt-dev
+
+COPY . /synapse
+
+# A wheel cache may be provided in ./cache for faster build
+RUN cd /synapse \
+ && pip install --upgrade pip setuptools psycopg2 lxml \
+ && mkdir -p /synapse/cache \
+ && pip install -f /synapse/cache --upgrade --process-dependency-links . \
+ && mv /synapse/contrib/docker/start.py /synapse/contrib/docker/conf / \
+ && rm -rf setup.py setup.cfg synapse
+
+VOLUME ["/data"]
+
+EXPOSE 8008/tcp 8448/tcp
+
+ENTRYPOINT ["/start.py"]

MANIFEST.in

@@ -25,6 +25,8 @@ recursive-include synapse/static *.js
 exclude jenkins.sh
 exclude jenkins*.sh
 exclude jenkins*
+exclude Dockerfile
+exclude .dockerignore
 recursive-exclude jenkins *.sh
 
 prune .github

README.rst

@@ -157,8 +157,9 @@ if you prefer.
 
 In case of problems, please see the _`Troubleshooting` section below.
 
-Alternatively, Andreas Peters (previously Silvio Fricke) has contributed a Dockerfile to automate the
+There is an offical synapse image available at https://hub.docker.com/r/matrixdotorg/synapse/tags/ which can be used with the docker-compose file available at `contrib/docker`. Further information on this including configuration options is available in `contrib/docker/README.md`.
-above in Docker at https://hub.docker.com/r/avhost/docker-matrix/tags/
+
+Alternatively, Andreas Peters (previously Silvio Fricke) has contributed a Dockerfile to automate a synapse server in a single Docker image, 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

contrib/docker/README.md

@@ -0,0 +1,153 @@
+# Synapse Docker
+
+The `matrixdotorg/synapse` Docker image will run Synapse as a single process. It does not provide a
+database server or a TURN server, you should run these separately.
+
+If you run a Postgres server, you should simply include it in the same Compose
+project or set the proper environment variables and the image will automatically
+use that server.
+
+## Build
+
+Build the docker image with the `docker build` command from the root of the synapse repository.
+
+```
+docker build -t docker.io/matrixdotorg/synapse .
+```
+
+The `-t` option sets the image tag. Official images are tagged `matrixdotorg/synapse:<version>` where `<version>` is the same as the release tag in the synapse git repository.
+
+You may have a local Python wheel cache available, in which case copy the relevant packages in the ``cache/`` directory at the root of the project.
+
+## Run
+
+This image is designed to run either with an automatically generated configuration
+file or with a custom configuration that requires manual edition.
+
+### Automated configuration
+
+It is recommended that you use Docker Compose to run your containers, including
+this image and a Postgres server. A sample ``docker-compose.yml`` is provided,
+including example labels for reverse proxying and other artifacts.
+
+Read the section about environment variables and set at least mandatory variables,
+then run the server:
+
+```
+docker-compose up -d
+```
+
+If secrets are not specified in the environment variables, they will be generated
+as part of the startup. Please ensure these secrets are kept between launches of the
+Docker container, as their loss may require users to log in again.
+
+### Manual configuration
+
+A sample ``docker-compose.yml`` is provided, including example labels for
+reverse proxying and other artifacts. The docker-compose file is an example,
+please comment/uncomment sections that are not suitable for your usecase.
+
+Specify a ``SYNAPSE_CONFIG_PATH``, preferably to a persistent path,
+to use manual configuration. To generate a fresh ``homeserver.yaml``, simply run:
+
+```
+docker-compose run --rm -e SYNAPSE_SERVER_NAME=my.matrix.host synapse generate
+```
+
+Then, customize your configuration and run the server:
+
+```
+docker-compose up -d
+```
+
+### Without Compose
+
+If you do not wish to use Compose, you may still run this image using plain
+Docker commands. Note that the following is just a guideline and you may need
+to add parameters to the docker run command to account for the network situation
+with your postgres database.
+
+```
+docker run \
+    -d \
+    --name synapse \
+    -v ${DATA_PATH}:/data \
+    -e SYNAPSE_SERVER_NAME=my.matrix.host \
+    -e SYNAPSE_REPORT_STATS=yes \
+    docker.io/matrixdotorg/synapse:latest
+```
+
+## Volumes
+
+The image expects a single volume, located at ``/data``, that will hold:
+
+* temporary files during uploads;
+* uploaded media and thumbnails;
+* the SQLite database if you do not configure postgres;
+* the appservices configuration.
+
+You are free to use separate volumes depending on storage endpoints at your
+disposal. For instance, ``/data/media`` coud be stored on a large but low
+performance hdd storage while other files could be stored on high performance
+endpoints.
+
+In order to setup an application service, simply create an ``appservices``
+directory in the data volume and write the application service Yaml
+configuration file there. Multiple application services are supported.
+
+## Environment
+
+Unless you specify a custom path for the configuration file, a very generic
+file will be generated, based on the following environment settings.
+These are a good starting point for setting up your own deployment.
+
+Global settings:
+
+* ``UID``, the user id Synapse will run as [default 991]
+* ``GID``, the group id Synapse will run as [default 991]
+* ``SYNAPSE_CONFIG_PATH``, path to a custom config file
+
+If ``SYNAPSE_CONFIG_PATH`` is set, you should generate a configuration file
+then customize it manually. No other environment variable is required.
+
+Otherwise, a dynamic configuration file will be used. The following environment
+variables are available for configuration:
+
+* ``SYNAPSE_SERVER_NAME`` (mandatory), the current server public hostname.
+* ``SYNAPSE_REPORT_STATS``, (mandatory, ``yes`` or ``no``), enable anonymous
+  statistics reporting back to the Matrix project which helps us to get funding.
+* ``SYNAPSE_NO_TLS``, set this variable to disable TLS in Synapse (use this if
+  you run your own TLS-capable reverse proxy).
+* ``SYNAPSE_ENABLE_REGISTRATION``, set this variable to enable registration on
+  the Synapse instance.
+* ``SYNAPSE_ALLOW_GUEST``, set this variable to allow guest joining this server.
+* ``SYNAPSE_EVENT_CACHE_SIZE``, the event cache size [default `10K`].
+* ``SYNAPSE_CACHE_FACTOR``, the cache factor [default `0.5`].
+* ``SYNAPSE_RECAPTCHA_PUBLIC_KEY``, set this variable to the recaptcha public
+  key in order to enable recaptcha upon registration.
+* ``SYNAPSE_RECAPTCHA_PRIVATE_KEY``, set this variable to the recaptcha private
+  key in order to enable recaptcha upon registration.
+* ``SYNAPSE_TURN_URIS``, set this variable to the coma-separated list of TURN
+  uris to enable TURN for this homeserver.
+* ``SYNAPSE_TURN_SECRET``, set this to the TURN shared secret if required.
+
+Shared secrets, that will be initialized to random values if not set:
+
+* ``SYNAPSE_REGISTRATION_SHARED_SECRET``, secret for registrering users if
+  registration is disable.
+* ``SYNAPSE_MACAROON_SECRET_KEY`` secret for signing access tokens
+  to the server.
+
+Database specific values (will use SQLite if not set):
+
+* `POSTGRES_DB` - The database name for the synapse postgres database. [default: `synapse`]
+* `POSTGRES_HOST` - The host of the postgres database if you wish to use postgresql instead of sqlite3. [default: `db` which is useful when using a container on the same docker network in a compose file where the postgres service is called `db`]
+* `POSTGRES_PASSWORD` - The password for the synapse postgres database. **If this is set then postgres will be used instead of sqlite3.** [default: none] **NOTE**: You are highly encouraged to use postgresql! Please use the compose file to make it easier to deploy.
+* `POSTGRES_USER` - The user for the synapse postgres database. [default: `matrix`]
+
+Mail server specific values (will not send emails if not set):
+
+* ``SYNAPSE_SMTP_HOST``, hostname to the mail server.
+* ``SYNAPSE_SMTP_PORT``, TCP port for accessing the mail server [default ``25``].
+* ``SYNAPSE_SMTP_USER``, username for authenticating against the mail server if any.
+* ``SYNAPSE_SMTP_PASSWORD``, password for authenticating against the mail server if any.

contrib/docker/conf/homeserver.yaml

@@ -0,0 +1,219 @@
+# vim:ft=yaml
+
+## TLS ##
+
+tls_certificate_path: "/data/{{ SYNAPSE_SERVER_NAME }}.tls.crt"
+tls_private_key_path: "/data/{{ SYNAPSE_SERVER_NAME }}.tls.key"
+tls_dh_params_path: "/data/{{ SYNAPSE_SERVER_NAME }}.tls.dh"
+no_tls: {{ "True" if SYNAPSE_NO_TLS else "False" }}
+tls_fingerprints: []
+
+## Server ##
+
+server_name: "{{ SYNAPSE_SERVER_NAME }}"
+pid_file: /homeserver.pid
+web_client: False
+soft_file_limit: 0
+
+## Ports ##
+
+listeners:
+  {% if not SYNAPSE_NO_TLS %}
+  -
+    port: 8448
+    bind_addresses: ['0.0.0.0']
+    type: http
+    tls: true
+    x_forwarded: false
+    resources:
+      - names: [client]
+        compress: true
+      - names: [federation]  # Federation APIs
+        compress: false
+  {% endif %}
+
+  - port: 8008
+    tls: false
+    bind_addresses: ['0.0.0.0']
+    type: http
+    x_forwarded: false
+
+    resources:
+      - names: [client]
+        compress: true
+      - names: [federation]
+        compress: false
+
+## Database ##
+
+{% if POSTGRES_PASSWORD %}
+database:
+  name: "psycopg2"
+  args:
+    user: "{{ POSTGRES_USER or "synapse" }}"
+    password: "{{ POSTGRES_PASSWORD }}"
+    database: "{{ POSTGRES_DB or "synapse" }}"
+    host: "{{ POSTGRES_HOST or "db" }}"
+    port: "{{ POSTGRES_PORT or "5432" }}"
+    cp_min: 5
+    cp_max: 10
+{% else %}
+database:
+  name: "sqlite3"
+  args:
+    database: "/data/homeserver.db"
+{% endif %}
+
+## Performance ##
+
+event_cache_size: "{{ SYNAPSE_EVENT_CACHE_SIZE or "10K" }}"
+verbose: 0
+log_file: "/data/homeserver.log"
+log_config: "/compiled/log.config"
+
+## Ratelimiting ##
+
+rc_messages_per_second: 0.2
+rc_message_burst_count: 10.0
+federation_rc_window_size: 1000
+federation_rc_sleep_limit: 10
+federation_rc_sleep_delay: 500
+federation_rc_reject_limit: 50
+federation_rc_concurrent: 3
+
+## Files ##
+
+media_store_path: "/data/media"
+uploads_path: "/data/uploads"
+max_upload_size: "10M"
+max_image_pixels: "32M"
+dynamic_thumbnails: false
+
+# List of thumbnail to precalculate when an image is uploaded.
+thumbnail_sizes:
+- width: 32
+  height: 32
+  method: crop
+- width: 96
+  height: 96
+  method: crop
+- width: 320
+  height: 240
+  method: scale
+- width: 640
+  height: 480
+  method: scale
+- width: 800
+  height: 600
+  method: scale
+
+url_preview_enabled: False
+max_spider_size: "10M"
+
+## Captcha ##
+
+{% if SYNAPSE_RECAPTCHA_PUBLIC_KEY %}
+recaptcha_public_key: "{{ SYNAPSE_RECAPTCHA_PUBLIC_KEY }}"
+recaptcha_private_key: "{{ SYNAPSE_RECAPTCHA_PRIVATE_KEY }}"
+enable_registration_captcha: True
+recaptcha_siteverify_api: "https://www.google.com/recaptcha/api/siteverify"
+{% else %}
+recaptcha_public_key: "YOUR_PUBLIC_KEY"
+recaptcha_private_key: "YOUR_PRIVATE_KEY"
+enable_registration_captcha: False
+recaptcha_siteverify_api: "https://www.google.com/recaptcha/api/siteverify"
+{% endif %}
+
+## Turn ##
+
+{% if SYNAPSE_TURN_URIS %}
+turn_uris:
+{% for uri in SYNAPSE_TURN_URIS.split(',') %}    - "{{ uri }}"
+{% endfor %}
+turn_shared_secret: "{{ SYNAPSE_TURN_SECRET }}"
+turn_user_lifetime: "1h"
+turn_allow_guests: True
+{% else %}
+turn_uris: []
+turn_shared_secret: "YOUR_SHARED_SECRET"
+turn_user_lifetime: "1h"
+turn_allow_guests: True
+{% endif %}
+
+## Registration ##
+
+enable_registration: {{ "True" if SYNAPSE_ENABLE_REGISTRATION else "False" }}
+registration_shared_secret: "{{ SYNAPSE_REGISTRATION_SHARED_SECRET }}"
+bcrypt_rounds: 12
+allow_guest_access: {{ "True" if SYNAPSE_ALLOW_GUEST else "False" }}
+enable_group_creation: true
+
+# The list of identity servers trusted to verify third party
+# identifiers by this server.
+trusted_third_party_id_servers:
+    - matrix.org
+    - vector.im
+    - riot.im
+
+## Metrics ###
+
+{% if SYNAPSE_REPORT_STATS.lower() == "yes" %}
+enable_metrics: True
+report_stats: True
+{% else %}
+enable_metrics: False
+report_stats: False
+{% endif %}
+
+## API Configuration ##
+
+room_invite_state_types:
+    - "m.room.join_rules"
+    - "m.room.canonical_alias"
+    - "m.room.avatar"
+    - "m.room.name"
+
+{% if SYNAPSE_APPSERVICES %}
+app_service_config_files:
+{% for appservice in SYNAPSE_APPSERVICES %}    - "{{ appservice }}"
+{% endfor %}
+{% else %}
+app_service_config_files: []
+{% endif %}
+
+macaroon_secret_key: "{{ SYNAPSE_MACAROON_SECRET_KEY }}"
+expire_access_token: False
+
+## Signing Keys ##
+
+signing_key_path: "/data/{{ SYNAPSE_SERVER_NAME }}.signing.key"
+old_signing_keys: {}
+key_refresh_interval: "1d" # 1 Day.
+
+# The trusted servers to download signing keys from.
+perspectives:
+  servers:
+    "matrix.org":
+      verify_keys:
+        "ed25519:auto":
+          key: "Noi6WqcDj0QmPxCNQqgezwTlBKrfqehY1u2FyWP9uYw"
+
+password_config:
+   enabled: true
+
+{% if SYNAPSE_SMTP_HOST %}
+email:
+   enable_notifs: false
+   smtp_host: "{{ SYNAPSE_SMTP_HOST }}"
+   smtp_port: {{ SYNAPSE_SMTP_PORT or "25" }}
+   smtp_user: "{{ SYNAPSE_SMTP_USER }}"
+   smtp_pass: "{{ SYNAPSE_SMTP_PASSWORD }}"
+   require_transport_security: False
+   notif_from: "{{ SYNAPSE_SMTP_FROM or "hostmaster@" + SYNAPSE_SERVER_NAME }}"
+   app_name: Matrix
+   template_dir: res/templates
+   notif_template_html: notif_mail.html
+   notif_template_text: notif_mail.txt
+   notif_for_new_users: True
+   riot_base_url: "https://{{ SYNAPSE_SERVER_NAME }}"
+{% endif %}

contrib/docker/conf/log.config

@@ -0,0 +1,29 @@
+version: 1
+
+formatters:
+  precise:
+   format: '%(asctime)s - %(name)s - %(lineno)d - %(levelname)s - %(request)s- %(message)s'
+
+filters:
+  context:
+    (): synapse.util.logcontext.LoggingContextFilter
+    request: ""
+
+handlers:
+  console:
+    class: logging.StreamHandler
+    formatter: precise
+    filters: [context]
+
+loggers:
+    synapse:
+        level: {{ SYNAPSE_LOG_LEVEL or "WARNING" }}
+
+    synapse.storage.SQL:
+        # beware: increasing this to DEBUG will make synapse log sensitive
+        # information such as access tokens.
+        level: {{ SYNAPSE_LOG_LEVEL or "WARNING" }}
+
+root:
+    level: {{ SYNAPSE_LOG_LEVEL or "WARNING" }}
+    handlers: [console]

contrib/docker/docker-compose.yml

@@ -0,0 +1,49 @@
+# This compose file is compatible with Compose itself, it might need some
+# adjustments to run properly with stack.
+
+version: '3'
+
+services:
+
+  synapse:
+    image: docker.io/matrixdotorg/synapse:latest
+    # Since snyapse does not retry to connect to the database, restart upon
+    # failure
+    restart: unless-stopped
+    # See the readme for a full documentation of the environment settings
+    environment:
+      - SYNAPSE_SERVER_NAME=my.matrix.host
+      - SYNAPSE_REPORT_STATS=no
+      - SYNAPSE_ENABLE_REGISTRATION=yes
+      - SYNAPSE_LOG_LEVEL=INFO
+      - POSTGRES_PASSWORD=changeme
+    volumes:
+      # You may either store all the files in a local folder
+      - ./files:/data
+      # .. or you may split this between different storage points
+      # - ./files:/data
+      # - /path/to/ssd:/data/uploads
+      # - /path/to/large_hdd:/data/media
+    depends_on:
+      - db
+    # In order to expose Synapse, remove one of the following, you might for
+    # instance expose the TLS port directly:
+    ports:
+      - 8448:8448/tcp
+    # ... or use a reverse proxy, here is an example for traefik:
+    labels:
+      - traefik.enable=true
+      - traefik.frontend.rule=Host:my.matrix.Host
+      - traefik.port=8448
+
+  db:
+    image: docker.io/postgres:10-alpine
+    # Change that password, of course!
+    environment:
+      - POSTGRES_USER=synapse
+      - POSTGRES_PASSWORD=changeme
+    volumes:
+      # You may store the database tables in a local folder..
+      - ./schemas:/var/lib/postgresql/data
+      # .. or store them on some high performance storage for better results
+      # - /path/to/ssd/storage:/var/lib/postfesql/data

contrib/docker/start.py

@@ -0,0 +1,66 @@
+#!/usr/local/bin/python
+
+import jinja2
+import os
+import sys
+import subprocess
+import glob
+
+# Utility functions
+convert = lambda src, dst, environ: open(dst, "w").write(jinja2.Template(open(src).read()).render(**environ))
+
+def check_arguments(environ, args):
+    for argument in args:
+        if argument not in environ:
+            print("Environment variable %s is mandatory, exiting." % argument)
+            sys.exit(2)
+
+def generate_secrets(environ, secrets):
+    for name, secret in secrets.items():
+        if secret not in environ:
+            filename = "/data/%s.%s.key" % (environ["SYNAPSE_SERVER_NAME"], name)
+            if os.path.exists(filename):
+                with open(filename) as handle: value = handle.read()
+            else:
+                print("Generating a random secret for {}".format(name))
+                value = os.urandom(32).encode("hex")
+                with open(filename, "w") as handle: handle.write(value)
+            environ[secret] = value
+
+# Prepare the configuration
+mode = sys.argv[1] if len(sys.argv) > 1 else None
+environ = os.environ.copy()
+ownership = "{}:{}".format(environ.get("UID", 991), environ.get("GID", 991))
+args = ["python", "-m", "synapse.app.homeserver"]
+
+# In generate mode, generate a configuration, missing keys, then exit
+if mode == "generate":
+    check_arguments(environ, ("SYNAPSE_SERVER_NAME", "SYNAPSE_REPORT_STATS", "SYNAPSE_CONFIG_PATH"))
+    args += [
+        "--server-name", environ["SYNAPSE_SERVER_NAME"],
+        "--report-stats", environ["SYNAPSE_REPORT_STATS"],
+        "--config-path", environ["SYNAPSE_CONFIG_PATH"],
+        "--generate-config"
+    ]
+    os.execv("/usr/local/bin/python", args)
+
+# In normal mode, generate missing keys if any, then run synapse
+else:
+    # Parse the configuration file
+    if "SYNAPSE_CONFIG_PATH" in environ:
+        args += ["--config-path", environ["SYNAPSE_CONFIG_PATH"]]
+    else:
+        check_arguments(environ, ("SYNAPSE_SERVER_NAME", "SYNAPSE_REPORT_STATS"))
+        generate_secrets(environ, {
+            "registration": "SYNAPSE_REGISTRATION_SHARED_SECRET",
+            "macaroon": "SYNAPSE_MACAROON_SECRET_KEY"
+        })
+        environ["SYNAPSE_APPSERVICES"] = glob.glob("/data/appservices/*.yaml")
+        if not os.path.exists("/compiled"): os.mkdir("/compiled")
+        convert("/conf/homeserver.yaml", "/compiled/homeserver.yaml", environ)
+        convert("/conf/log.config", "/compiled/log.config", environ)
+        subprocess.check_output(["chown", "-R", ownership, "/data"])
+        args += ["--config-path", "/compiled/homeserver.yaml"]
+    # Generate missing keys and start synapse
+    subprocess.check_output(args + ["--generate-keys"])
+    os.execv("/sbin/su-exec", ["su-exec", ownership] + args)

docs/consent_tracking.md

@@ -0,0 +1,160 @@
+Support in Synapse for tracking agreement to server terms and conditions
+========================================================================
+
+Synapse 0.30 introduces support for tracking whether users have agreed to the
+terms and conditions set by the administrator of a server - and blocking access
+to the server until they have.
+
+There are several parts to this functionality; each requires some specific
+configuration in `homeserver.yaml` to be enabled.
+
+Note that various parts of the configuation and this document refer to the
+"privacy policy": agreement with a privacy policy is one particular use of this
+feature, but of course adminstrators can specify other terms and conditions
+unrelated to "privacy" per se.
+
+Collecting policy agreement from a user
+---------------------------------------
+
+Synapse can be configured to serve the user a simple policy form with an
+"accept" button. Clicking "Accept" records the user's acceptance in the
+database and shows a success page.
+
+To enable this, first create templates for the policy and success pages.
+These should be stored on the local filesystem.
+
+These templates use the [Jinja2](http://jinja.pocoo.org) templating language,
+and [docs/privacy_policy_templates](privacy_policy_templates) gives
+examples of the sort of thing that can be done.
+
+Note that the templates must be stored under a name giving the language of the
+template - currently this must always be `en` (for "English");
+internationalisation support is intended for the future.
+
+The template for the policy itself should be versioned and named according to 
+the version: for example `1.0.html`. The version of the policy which the user
+has agreed to is stored in the database.
+
+Once the templates are in place, make the following changes to `homeserver.yaml`:
+
+ 1. Add a `user_consent` section, which should look like:
+
+    ```yaml
+    user_consent:
+      template_dir: privacy_policy_templates
+      version: 1.0
+    ```
+
+    `template_dir` points to the directory containing the policy
+    templates. `version` defines the version of the policy which will be served
+    to the user. In the example above, Synapse will serve
+    `privacy_policy_templates/en/1.0.html`.
+
+
+ 2. Add a `form_secret` setting at the top level:
+
+
+    ```yaml
+    form_secret: "<unique secret>"
+    ```
+
+    This should be set to an arbitrary secret string (try `pwgen -y 30` to
+    generate suitable secrets).
+
+    More on what this is used for below.
+
+ 3. Add `consent` wherever the `client` resource is currently enabled in the
+    `listeners` configuration. For example:
+
+    ```yaml
+    listeners:
+      - port: 8008
+        resources:
+          - names:
+            - client
+            - consent
+    ```
+
+
+Finally, ensure that `jinja2` is installed. If you are using a virtualenv, this
+should be a matter of `pip install Jinja2`. On debian, try `apt-get install
+python-jinja2`.
+
+Once this is complete, and the server has been restarted, try visiting
+`https://<server>/_matrix/consent`. If correctly configured, this should give
+an error "Missing string query parameter 'u'". It is now possible to manually
+construct URIs where users can give their consent.
+
+### Constructing the consent URI
+
+It may be useful to manually construct the "consent URI" for a given user - for
+instance, in order to send them an email asking them to consent. To do this,
+take the base `https://<server>/_matrix/consent` URL and add the following
+query parameters:
+
+ * `u`: the user id of the user. This can either be a full MXID
+   (`@user:server.com`) or just the localpart (`user`).
+
+ * `h`: hex-encoded HMAC-SHA256 of `u` using the `form_secret` as a key. It is
+   possible to calculate this on the commandline with something like:
+
+   ```bash
+   echo -n '<user>' | openssl sha256 -hmac '<form_secret>'
+   ```
+
+   This should result in a URI which looks something like:
+   `https://<server>/_matrix/consent?u=<user>&h=68a152465a4d...`.
+
+
+Sending users a server notice asking them to agree to the policy
+----------------------------------------------------------------
+
+It is possible to configure Synapse to send a [server
+notice](server_notices.md) to anybody who has not yet agreed to the current
+version of the policy. To do so:
+
+ * ensure that the consent resource is configured, as in the previous section
+
+ * ensure that server notices are configured, as in [server_notices.md](server_notices.md).
+
+ * Add `server_notice_content` under `user_consent` in `homeserver.yaml`. For
+   example:
+
+   ```yaml
+   user_consent:
+     server_notice_content:
+       msgtype: m.text
+       body: >-
+         Please give your consent to the privacy policy at %(consent_uri)s.
+   ```
+
+   Synapse automatically replaces the placeholder `%(consent_uri)s` with the
+   consent uri for that user.
+
+ * ensure that `public_baseurl` is set in `homeserver.yaml`, and gives the base
+   URI that clients use to connect to the server. (It is used to construct
+   `consent_uri` in the server notice.)
+
+
+Blocking users from using the server until they agree to the policy
+-------------------------------------------------------------------
+
+Synapse can be configured to block any attempts to join rooms or send messages
+until the user has given their agreement to the policy. (Joining the server
+notices room is exempted from this).
+
+To enable this, add `block_events_error` under `user_consent`. For example:
+
+```yaml
+user_consent:
+  block_events_error: >-
+    You can't send any messages until you consent to the privacy policy at
+    %(consent_uri)s.
+```
+
+Synapse automatically replaces the placeholder `%(consent_uri)s` with the
+consent uri for that user.
+
+ensure that `public_baseurl` is set in `homeserver.yaml`, and gives the base
+URI that clients use to connect to the server. (It is used to construct
+`consent_uri` in the error.)

docs/manhole.md

@@ -0,0 +1,43 @@
+Using the synapse manhole
+=========================
+
+The "manhole" allows server administrators to access a Python shell on a running
+Synapse installation. This is a very powerful mechanism for administration and
+debugging.
+
+To enable it, first uncomment the `manhole` listener configuration in
+`homeserver.yaml`:
+
+```yaml
+listeners:
+  - port: 9000
+    bind_addresses: ['::1', '127.0.0.1']
+    type: manhole
+```
+
+(`bind_addresses` in the above is important: it ensures that access to the
+manhole is only possible for local users).
+
+Note that this will give administrative access to synapse to **all users** with
+shell access to the server. It should therefore **not** be enabled in
+environments where untrusted users have shell access.
+
+Then restart synapse, and point an ssh client at port 9000 on localhost, using
+the username `matrix`:
+
+```bash
+ssh -p9000 matrix@localhost
+```
+
+The password is `rabbithole`.
+
+This gives a Python REPL in which `hs` gives access to the
+`synapse.server.HomeServer` object - which in turn gives access to many other
+parts of the process.
+
+As a simple example, retrieving an event from the database:
+
+```
+>>> hs.get_datastore().get_event('$1416420717069yeQaw:matrix.org')
+<Deferred at 0x7ff253fc6998 current result: <FrozenEvent event_id='$1416420717069yeQaw:matrix.org', type='m.room.create', state_key=''>>
+```

docs/metrics-howto.rst

@@ -1,25 +1,47 @@
 How to monitor Synapse metrics using Prometheus
 ===============================================
 
-1. Install prometheus:
+1. Install Prometheus:
 
    Follow instructions at http://prometheus.io/docs/introduction/install/
 
-2. Enable synapse metrics:
+2. Enable Synapse metrics:
 
-   Simply setting a (local) port number will enable it. Pick a port.
+   There are two methods of enabling metrics in Synapse.
-   prometheus itself defaults to 9090, so starting just above that for
-   locally monitored services seems reasonable. E.g. 9092:
 
-   Add to homeserver.yaml::
+   The first serves the metrics as a part of the usual web server and can be
+   enabled by adding the "metrics" resource to the existing listener as such::
 
-     metrics_port: 9092
+     resources:
+       - names:
+         - client
+         - metrics
 
-   Also ensure that ``enable_metrics`` is set to ``True``.
+   This provides a simple way of adding metrics to your Synapse installation,
+   and serves under ``/_synapse/metrics``. If you do not wish your metrics be
+   publicly exposed, you will need to either filter it out at your load
+   balancer, or use the second method.
 
-   Restart synapse.
+   The second method runs the metrics server on a different port, in a
+   different thread to Synapse. This can make it more resilient to heavy load
+   meaning metrics cannot be retrieved, and can be exposed to just internal
+   networks easier. The served metrics are available over HTTP only, and will
+   be available at ``/``.
 
-3. Add a prometheus target for synapse.
+   Add a new listener to homeserver.yaml::
+
+     listeners:
+       - type: metrics
+         port: 9000
+         bind_addresses:
+           - '0.0.0.0'
+
+   For both options, you will need to ensure that ``enable_metrics`` is set to
+   ``True``.
+
+   Restart Synapse.
+
+3. Add a Prometheus target for Synapse.
 
    It needs to set the ``metrics_path`` to a non-default value (under ``scrape_configs``)::
 
@@ -31,7 +53,50 @@ How to monitor Synapse metrics using Prometheus
    If your prometheus is older than 1.5.2, you will need to replace
    ``static_configs`` in the above with ``target_groups``.
 
-   Restart prometheus.
+   Restart Prometheus.
+
+
+Removal of deprecated metrics & time based counters becoming histograms in 0.31.0
+---------------------------------------------------------------------------------
+
+The duplicated metrics deprecated in Synapse 0.27.0 have been removed.
+
+All time duration-based metrics have been changed to be seconds. This affects:
+
++----------------------------------+
+| msec -> sec metrics              |
++==================================+
+| python_gc_time                   |
++----------------------------------+
+| python_twisted_reactor_tick_time |
++----------------------------------+
+| synapse_storage_query_time       |
++----------------------------------+
+| synapse_storage_schedule_time    |
++----------------------------------+
+| synapse_storage_transaction_time |
++----------------------------------+
+
+Several metrics have been changed to be histograms, which sort entries into
+buckets and allow better analysis. The following metrics are now histograms:
+
++-------------------------------------------+
+| Altered metrics                           |
++===========================================+
+| python_gc_time                            |
++-------------------------------------------+
+| python_twisted_reactor_pending_calls      |
++-------------------------------------------+
+| python_twisted_reactor_tick_time          |
++-------------------------------------------+
+| synapse_http_server_response_time_seconds |
++-------------------------------------------+
+| synapse_storage_query_time                |
++-------------------------------------------+
+| synapse_storage_schedule_time             |
++-------------------------------------------+
+| synapse_storage_transaction_time          |
++-------------------------------------------+
 
 
 Block and response metrics renamed for 0.27.0

docs/postgres.rst

@@ -6,7 +6,13 @@ Postgres version 9.4 or later is known to work.
 Set up database
 ===============
 
-The PostgreSQL database used *must* have the correct encoding set, otherwise
+Assuming your PostgreSQL database user is called ``postgres``, create a user
+``synapse_user`` with::
+
+ su - postgres
+ createuser --pwprompt synapse_user
+
+The PostgreSQL database used *must* have the correct encoding set, otherwise it
 would not be able to store UTF8 strings. To create a database with the correct
 encoding use, e.g.::
 
@@ -46,8 +52,8 @@ As with Debian/Ubuntu, postgres support depends on the postgres python connector
 Synapse config
 ==============
 
-When you are ready to start using PostgreSQL, add the following line to your
+When you are ready to start using PostgreSQL, edit the ``database`` section in
-config file::
+your config file to match the following lines::
 
     database:
         name: psycopg2
@@ -96,9 +102,12 @@ complete, restart synapse.  For instance::
     cp homeserver.db homeserver.db.snapshot
     ./synctl start
 
-Assuming your new config file (as described in the section *Synapse config*)
+Copy the old config file into a new config file::
-is named ``homeserver-postgres.yaml`` and the SQLite snapshot is at
+
-``homeserver.db.snapshot`` then simply run::
+    cp homeserver.yaml homeserver-postgres.yaml
+
+Edit the database section as described in the section *Synapse config* above
+and with the SQLite snapshot located at ``homeserver.db.snapshot`` simply run::
 
     synapse_port_db --sqlite-database homeserver.db.snapshot \
         --postgres-config homeserver-postgres.yaml
@@ -117,6 +126,11 @@ run::
         --postgres-config homeserver-postgres.yaml
 
 Once that has completed, change the synapse config to point at the PostgreSQL
-database configuration file ``homeserver-postgres.yaml`` (i.e. rename it to 
+database configuration file ``homeserver-postgres.yaml``:
-``homeserver.yaml``) and restart synapse. Synapse should now be running against
+
-PostgreSQL.
+    ./synctl stop
+    mv homeserver.yaml homeserver-old-sqlite.yaml 
+    mv homeserver-postgres.yaml homeserver.yaml 
+    ./synctl start
+
+Synapse should now be running against PostgreSQL.

docs/privacy_policy_templates/en/1.0.html

@@ -0,0 +1,23 @@
+<!doctype html>
+<html lang="en">
+  <head>
+    <title>Matrix.org Privacy policy</title>
+  </head>
+  <body>
+  {% if has_consented %}
+    <p>
+      Your base already belong to us.
+    </p>
+  {% else %}
+    <p>
+      All your base are belong to us.
+    </p>
+    <form method="post" action="consent">
+      <input type="hidden" name="v" value="{{version}}"/>
+      <input type="hidden" name="u" value="{{user}}"/>
+      <input type="hidden" name="h" value="{{userhmac}}"/>
+      <input type="submit" value="Sure thing!"/>
+    </form>
+  {% endif %}
+  </body>
+</html>

docs/privacy_policy_templates/en/success.html

@@ -0,0 +1,11 @@
+<!doctype html>
+<html lang="en">
+  <head>
+    <title>Matrix.org Privacy policy</title>
+  </head>
+  <body>
+    <p>
+      Sweet.
+    </p>
+  </body>
+</html>

docs/server_notices.md

@@ -0,0 +1,74 @@
+Server Notices
+==============
+
+'Server Notices' are a new feature introduced in Synapse 0.30. They provide a
+channel whereby server administrators can send messages to users on the server.
+
+They are used as part of communication of the server polices(see
+[consent_tracking.md](consent_tracking.md)), however the intention is that
+they may also find a use for features such as "Message of the day".
+
+This is a feature specific to Synapse, but it uses standard Matrix
+communication mechanisms, so should work with any Matrix client.
+
+User experience
+---------------
+
+When the user is first sent a server notice, they will get an invitation to a
+room (typically called 'Server Notices', though this is configurable in
+`homeserver.yaml`). They will be **unable to reject** this invitation -
+attempts to do so will receive an error.
+
+Once they accept the invitation, they will see the notice message in the room
+history; it will appear to have come from the 'server notices user' (see
+below).
+
+The user is prevented from sending any messages in this room by the power
+levels.
+
+Having joined the room, the user can leave the room if they want. Subsequent
+server notices will then cause a new room to be created.
+
+Synapse configuration
+---------------------
+
+Server notices come from a specific user id on the server. Server
+administrators are free to choose the user id - something like `server` is
+suggested, meaning the notices will come from
+`@server:<your_server_name>`. Once the Server Notices user is configured, that
+user id becomes a special, privileged user, so administrators should ensure
+that **it is not already allocated**.
+
+In order to support server notices, it is necessary to add some configuration
+to the `homeserver.yaml` file. In particular, you should add a `server_notices`
+section, which should look like this:
+
+```yaml
+server_notices:
+   system_mxid_localpart: server
+   system_mxid_display_name: "Server Notices"
+   system_mxid_avatar_url: "mxc://server.com/oumMVlgDnLYFaPVkExemNVVZ"
+   room_name: "Server Notices"
+```
+
+The only compulsory setting is `system_mxid_localpart`, which defines the user
+id of the Server Notices user, as above. `room_name` defines the name of the
+room which will be created.
+
+`system_mxid_display_name` and `system_mxid_avatar_url` can be used to set the
+displayname and avatar of the Server Notices user.
+
+Sending notices
+---------------
+
+As of the current version of synapse, there is no convenient interface for
+sending notices (other than the automated ones sent as part of consent
+tracking).
+
+In the meantime, it is possible to test this feature using the manhole. Having
+gone into the manhole as described in [manhole.md](manhole.md), a notice can be
+sent with something like:
+
+```
+>>> hs.get_server_notices_manager().send_notice('@user:server.com', {'msgtype':'m.text', 'body':'foo'})
+```

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-dev/nuke-room-from-db.sh

@@ -6,9 +6,19 @@
 
 ## Do not run it lightly.
 
+set -e
+
+if [ "$1" == "-h" ] || [ "$1" == "" ]; then
+  echo "Call with ROOM_ID as first option and then pipe it into the database. So for instance you might run"
+  echo " nuke-room-from-db.sh <room_id> | sqlite3 homeserver.db"
+  echo "or"
+  echo " nuke-room-from-db.sh <room_id> | psql --dbname=synapse"
+  exit
+fi
+
 ROOMID="$1"
 
-sqlite3 homeserver.db <<EOF
+cat <<EOF
 DELETE FROM event_forward_extremities WHERE room_id = '$ROOMID';
 DELETE FROM event_backward_extremities WHERE room_id = '$ROOMID';
 DELETE FROM event_edges WHERE room_id = '$ROOMID';
@@ -29,7 +39,7 @@ DELETE FROM state_groups WHERE room_id = '$ROOMID';
 DELETE FROM state_groups_state WHERE room_id = '$ROOMID';
 DELETE FROM receipts_graph WHERE room_id = '$ROOMID';
 DELETE FROM receipts_linearized WHERE room_id = '$ROOMID';
-DELETE FROM event_search_content WHERE c1room_id = '$ROOMID';
+DELETE FROM event_search WHERE room_id = '$ROOMID'; 
 DELETE FROM guest_access WHERE room_id = '$ROOMID';
 DELETE FROM history_visibility WHERE room_id = '$ROOMID';
 DELETE FROM room_tags WHERE room_id = '$ROOMID';

synapse/__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.
@@ -16,4 +17,4 @@
 """ This is a reference implementation of a Matrix home server.
 """
 
-__version__ = "0.28.1"
+__version__ = "0.31.2"

synapse/api/auth.py

@@ -15,6 +15,8 @@
 
 import logging
 
+from six import itervalues
+
 import pymacaroons
 from twisted.internet import defer
 
@@ -57,7 +59,7 @@ class Auth(object):
         self.TOKEN_NOT_FOUND_HTTP_STATUS = 401
 
         self.token_cache = LruCache(CACHE_SIZE_FACTOR * 10000)
-        register_cache("token_cache", self.token_cache)
+        register_cache("cache", "token_cache", self.token_cache)
 
     @defer.inlineCallbacks
     def check_from_context(self, event, context, do_sig_check=True):
@@ -66,7 +68,7 @@ class Auth(object):
         )
         auth_events = yield self.store.get_events(auth_events_ids)
         auth_events = {
-            (e.type, e.state_key): e for e in auth_events.values()
+            (e.type, e.state_key): e for e in itervalues(auth_events)
         }
         self.check(event, auth_events=auth_events, do_sig_check=do_sig_check)
 
@@ -653,7 +655,7 @@ class Auth(object):
             auth_events[(EventTypes.PowerLevels, "")] = power_level_event
 
         send_level = event_auth.get_send_level(
-            EventTypes.Aliases, "", auth_events
+            EventTypes.Aliases, "", power_level_event,
         )
         user_level = event_auth.get_user_power_level(user_id, auth_events)
 

synapse/api/errors.py

@@ -19,6 +19,7 @@ import logging
 
 import simplejson as json
 from six import iteritems
+from six.moves import http_client
 
 logger = logging.getLogger(__name__)
 
@@ -51,6 +52,8 @@ class Codes(object):
     THREEPID_DENIED = "M_THREEPID_DENIED"
     INVALID_USERNAME = "M_INVALID_USERNAME"
     SERVER_NOT_TRUSTED = "M_SERVER_NOT_TRUSTED"
+    CONSENT_NOT_GIVEN = "M_CONSENT_NOT_GIVEN"
+    CANNOT_LEAVE_SERVER_NOTICE_ROOM = "M_CANNOT_LEAVE_SERVER_NOTICE_ROOM"
 
 
 class CodeMessageException(RuntimeError):
@@ -138,6 +141,32 @@ class SynapseError(CodeMessageException):
         return res
 
 
+class ConsentNotGivenError(SynapseError):
+    """The error returned to the client when the user has not consented to the
+    privacy policy.
+    """
+    def __init__(self, msg, consent_uri):
+        """Constructs a ConsentNotGivenError
+
+        Args:
+            msg (str): The human-readable error message
+            consent_url (str): The URL where the user can give their consent
+        """
+        super(ConsentNotGivenError, self).__init__(
+            code=http_client.FORBIDDEN,
+            msg=msg,
+            errcode=Codes.CONSENT_NOT_GIVEN
+        )
+        self._consent_uri = consent_uri
+
+    def error_dict(self):
+        return cs_error(
+            self.msg,
+            self.errcode,
+            consent_uri=self._consent_uri
+        )
+
+
 class RegistrationError(SynapseError):
     """An error raised when a registration event fails."""
     pass
@@ -292,7 +321,7 @@ def cs_error(msg, code=Codes.UNKNOWN, **kwargs):
 
     Args:
         msg (str): The error message.
-        code (int): The error code.
+        code (str): The error code.
         kwargs : Additional keys to add to the response.
     Returns:
         A dict representing the error response JSON.

synapse/api/filtering.py

@@ -411,7 +411,7 @@ class Filter(object):
         return room_ids
 
     def filter(self, events):
-        return filter(self.check, events)
+        return list(filter(self.check, events))
 
     def limit(self):
         return self.filter_json.get("limit", 10)

synapse/api/urls.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.
@@ -14,6 +15,12 @@
 # limitations under the License.
 
 """Contains the URL paths to prefix various aspects of the server with. """
+from hashlib import sha256
+import hmac
+
+from six.moves.urllib.parse import urlencode
+
+from synapse.config import ConfigError
 
 CLIENT_PREFIX = "/_matrix/client/api/v1"
 CLIENT_V2_ALPHA_PREFIX = "/_matrix/client/v2_alpha"
@@ -25,3 +32,46 @@ SERVER_KEY_PREFIX = "/_matrix/key/v1"
 SERVER_KEY_V2_PREFIX = "/_matrix/key/v2"
 MEDIA_PREFIX = "/_matrix/media/r0"
 LEGACY_MEDIA_PREFIX = "/_matrix/media/v1"
+
+
+class ConsentURIBuilder(object):
+    def __init__(self, hs_config):
+        """
+        Args:
+            hs_config (synapse.config.homeserver.HomeServerConfig):
+        """
+        if hs_config.form_secret is None:
+            raise ConfigError(
+                "form_secret not set in config",
+            )
+        if hs_config.public_baseurl is None:
+            raise ConfigError(
+                "public_baseurl not set in config",
+            )
+
+        self._hmac_secret = hs_config.form_secret.encode("utf-8")
+        self._public_baseurl = hs_config.public_baseurl
+
+    def build_user_consent_uri(self, user_id):
+        """Build a URI which we can give to the user to do their privacy
+        policy consent
+
+        Args:
+            user_id (str): mxid or username of user
+
+        Returns
+            (str) the URI where the user can do consent
+        """
+        mac = hmac.new(
+            key=self._hmac_secret,
+            msg=user_id,
+            digestmod=sha256,
+        ).hexdigest()
+        consent_uri = "%s_matrix/consent?%s" % (
+            self._public_baseurl,
+            urlencode({
+                "u": user_id,
+                "h": mac
+            }),
+        )
+        return consent_uri

synapse/app/_base.py

@@ -124,6 +124,19 @@ def quit_with_error(error_string):
     sys.exit(1)
 
 
+def listen_metrics(bind_addresses, port):
+    """
+    Start Prometheus metrics server.
+    """
+    from synapse.metrics import RegistryProxy
+    from prometheus_client import start_http_server
+
+    for host in bind_addresses:
+        reactor.callInThread(start_http_server, int(port),
+                             addr=host, registry=RegistryProxy)
+        logger.info("Metrics now reporting on %s:%d", host, port)
+
+
 def listen_tcp(bind_addresses, port, factory, backlog=50):
     """
     Create a TCP socket for a port and several addresses

synapse/app/appservice.py

@@ -32,10 +32,10 @@ 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 NoResource
 
 logger = logging.getLogger("synapse.app.appservice")
@@ -74,6 +74,7 @@ class AppserviceServer(HomeServer):
                 site_tag,
                 listener_config,
                 root_resource,
+                self.version_string,
             )
         )
 
@@ -93,6 +94,13 @@ class AppserviceServer(HomeServer):
                         globals={"hs": self},
                     )
                 )
+            elif listener["type"] == "metrics":
+                if not self.get_config().enable_metrics:
+                    logger.warn(("Metrics listener configured, but "
+                                 "collect_metrics is not enabled!"))
+                else:
+                    _base.listen_metrics(listener["bind_addresses"],
+                                         listener["port"])
             else:
                 logger.warn("Unrecognized listener type: %s", listener["type"])
 
@@ -112,9 +120,14 @@ 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

@@ -25,6 +25,7 @@ from synapse.config.logger import setup_logging
 from synapse.crypto import context_factory
 from synapse.http.server import JsonResource
 from synapse.http.site import SynapseSite
+from synapse.metrics import RegistryProxy
 from synapse.metrics.resource import METRICS_PREFIX, MetricsResource
 from synapse.replication.slave.storage._base import BaseSlavedStore
 from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore
@@ -77,7 +78,7 @@ class ClientReaderServer(HomeServer):
         for res in listener_config["resources"]:
             for name in res["names"]:
                 if name == "metrics":
-                    resources[METRICS_PREFIX] = MetricsResource(self)
+                    resources[METRICS_PREFIX] = MetricsResource(RegistryProxy)
                 elif name == "client":
                     resource = JsonResource(self, canonical_json=False)
                     PublicRoomListRestServlet(self).register(resource)
@@ -98,6 +99,7 @@ class ClientReaderServer(HomeServer):
                 site_tag,
                 listener_config,
                 root_resource,
+                self.version_string,
             )
         )
 
@@ -117,7 +119,13 @@ class ClientReaderServer(HomeServer):
                         globals={"hs": self},
                     )
                 )
-
+            elif listener["type"] == "metrics":
+                if not self.get_config().enable_metrics:
+                    logger.warn(("Metrics listener configured, but "
+                                 "collect_metrics is not enabled!"))
+                else:
+                    _base.listen_metrics(listener["bind_addresses"],
+                                         listener["port"])
             else:
                 logger.warn("Unrecognized listener type: %s", listener["type"])
 

synapse/app/event_creator.py

@@ -25,6 +25,7 @@ from synapse.config.logger import setup_logging
 from synapse.crypto import context_factory
 from synapse.http.server import JsonResource
 from synapse.http.site import SynapseSite
+from synapse.metrics import RegistryProxy
 from synapse.metrics.resource import METRICS_PREFIX, MetricsResource
 from synapse.replication.slave.storage._base import BaseSlavedStore
 from synapse.replication.slave.storage.account_data import SlavedAccountDataStore
@@ -90,7 +91,7 @@ class EventCreatorServer(HomeServer):
         for res in listener_config["resources"]:
             for name in res["names"]:
                 if name == "metrics":
-                    resources[METRICS_PREFIX] = MetricsResource(self)
+                    resources[METRICS_PREFIX] = MetricsResource(RegistryProxy)
                 elif name == "client":
                     resource = JsonResource(self, canonical_json=False)
                     RoomSendEventRestServlet(self).register(resource)
@@ -114,6 +115,7 @@ class EventCreatorServer(HomeServer):
                 site_tag,
                 listener_config,
                 root_resource,
+                self.version_string,
             )
         )
 
@@ -133,6 +135,13 @@ class EventCreatorServer(HomeServer):
                         globals={"hs": self},
                     )
                 )
+            elif listener["type"] == "metrics":
+                if not self.get_config().enable_metrics:
+                    logger.warn(("Metrics listener configured, but "
+                                 "collect_metrics is not enabled!"))
+                else:
+                    _base.listen_metrics(listener["bind_addresses"],
+                                         listener["port"])
             else:
                 logger.warn("Unrecognized listener type: %s", listener["type"])
 

synapse/app/federation_reader.py

@@ -26,6 +26,7 @@ from synapse.config.logger import setup_logging
 from synapse.crypto import context_factory
 from synapse.federation.transport.server import TransportLayerServer
 from synapse.http.site import SynapseSite
+from synapse.metrics import RegistryProxy
 from synapse.metrics.resource import METRICS_PREFIX, MetricsResource
 from synapse.replication.slave.storage._base import BaseSlavedStore
 from synapse.replication.slave.storage.directory import DirectoryStore
@@ -71,7 +72,7 @@ class FederationReaderServer(HomeServer):
         for res in listener_config["resources"]:
             for name in res["names"]:
                 if name == "metrics":
-                    resources[METRICS_PREFIX] = MetricsResource(self)
+                    resources[METRICS_PREFIX] = MetricsResource(RegistryProxy)
                 elif name == "federation":
                     resources.update({
                         FEDERATION_PREFIX: TransportLayerServer(self),
@@ -87,6 +88,7 @@ class FederationReaderServer(HomeServer):
                 site_tag,
                 listener_config,
                 root_resource,
+                self.version_string,
             )
         )
 
@@ -106,6 +108,13 @@ class FederationReaderServer(HomeServer):
                         globals={"hs": self},
                     )
                 )
+            elif listener["type"] == "metrics":
+                if not self.get_config().enable_metrics:
+                    logger.warn(("Metrics listener configured, but "
+                                 "collect_metrics is not enabled!"))
+                else:
+                    _base.listen_metrics(listener["bind_addresses"],
+                                         listener["port"])
             else:
                 logger.warn("Unrecognized listener type: %s", listener["type"])
 

synapse/app/federation_sender.py

@@ -25,6 +25,7 @@ from synapse.config.logger import setup_logging
 from synapse.crypto import context_factory
 from synapse.federation import send_queue
 from synapse.http.site import SynapseSite
+from synapse.metrics import RegistryProxy
 from synapse.metrics.resource import METRICS_PREFIX, MetricsResource
 from synapse.replication.slave.storage.deviceinbox import SlavedDeviceInboxStore
 from synapse.replication.slave.storage.devices import SlavedDeviceStore
@@ -38,7 +39,7 @@ 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
@@ -89,7 +90,7 @@ class FederationSenderServer(HomeServer):
         for res in listener_config["resources"]:
             for name in res["names"]:
                 if name == "metrics":
-                    resources[METRICS_PREFIX] = MetricsResource(self)
+                    resources[METRICS_PREFIX] = MetricsResource(RegistryProxy)
 
         root_resource = create_resource_tree(resources, NoResource())
 
@@ -101,6 +102,7 @@ class FederationSenderServer(HomeServer):
                 site_tag,
                 listener_config,
                 root_resource,
+                self.version_string,
             )
         )
 
@@ -120,6 +122,13 @@ class FederationSenderServer(HomeServer):
                         globals={"hs": self},
                     )
                 )
+            elif listener["type"] == "metrics":
+                if not self.get_config().enable_metrics:
+                    logger.warn(("Metrics listener configured, but "
+                                 "collect_metrics is not enabled!"))
+                else:
+                    _base.listen_metrics(listener["bind_addresses"],
+                                         listener["port"])
             else:
                 logger.warn("Unrecognized listener type: %s", listener["type"])
 
@@ -229,7 +238,7 @@ 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 +246,7 @@ 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 +260,8 @@ 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

@@ -29,6 +29,7 @@ from synapse.http.servlet import (
     RestServlet, parse_json_object_from_request,
 )
 from synapse.http.site import SynapseSite
+from synapse.metrics import RegistryProxy
 from synapse.metrics.resource import METRICS_PREFIX, MetricsResource
 from synapse.replication.slave.storage._base import BaseSlavedStore
 from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore
@@ -131,7 +132,7 @@ class FrontendProxyServer(HomeServer):
         for res in listener_config["resources"]:
             for name in res["names"]:
                 if name == "metrics":
-                    resources[METRICS_PREFIX] = MetricsResource(self)
+                    resources[METRICS_PREFIX] = MetricsResource(RegistryProxy)
                 elif name == "client":
                     resource = JsonResource(self, canonical_json=False)
                     KeyUploadServlet(self).register(resource)
@@ -152,6 +153,7 @@ class FrontendProxyServer(HomeServer):
                 site_tag,
                 listener_config,
                 root_resource,
+                self.version_string,
             )
         )
 
@@ -171,6 +173,13 @@ class FrontendProxyServer(HomeServer):
                         globals={"hs": self},
                     )
                 )
+            elif listener["type"] == "metrics":
+                if not self.get_config().enable_metrics:
+                    logger.warn(("Metrics listener configured, but "
+                                 "collect_metrics is not enabled!"))
+                else:
+                    _base.listen_metrics(listener["bind_addresses"],
+                                         listener["port"])
             else:
                 logger.warn("Unrecognized listener type: %s", listener["type"])
 

synapse/app/homeserver.py

@@ -34,7 +34,7 @@ from synapse.module_api import ModuleApi
 from synapse.http.additional_resource import AdditionalResource
 from synapse.http.server import RootRedirect
 from synapse.http.site import SynapseSite
-from synapse.metrics import register_memory_metrics
+from synapse.metrics import RegistryProxy
 from synapse.metrics.resource import METRICS_PREFIX, MetricsResource
 from synapse.python_dependencies import CONDITIONAL_REQUIREMENTS, \
     check_requirements
@@ -140,6 +140,7 @@ class SynapseHomeServer(HomeServer):
                     site_tag,
                     listener_config,
                     root_resource,
+                    self.version_string,
                 ),
                 self.tls_server_context_factory,
             )
@@ -153,6 +154,7 @@ class SynapseHomeServer(HomeServer):
                     site_tag,
                     listener_config,
                     root_resource,
+                    self.version_string,
                 )
             )
         logger.info("Synapse now listening on port %d", port)
@@ -182,6 +184,15 @@ class SynapseHomeServer(HomeServer):
                 "/_matrix/client/versions": client_resource,
             })
 
+        if name == "consent":
+            from synapse.rest.consent.consent_resource import ConsentResource
+            consent_resource = ConsentResource(self)
+            if compress:
+                consent_resource = gz_wrap(consent_resource)
+            resources.update({
+                "/_matrix/consent": consent_resource,
+            })
+
         if name == "federation":
             resources.update({
                 FEDERATION_PREFIX: TransportLayerServer(self),
@@ -219,7 +230,7 @@ class SynapseHomeServer(HomeServer):
             resources[WEB_CLIENT_PREFIX] = build_resource_for_web_client(self)
 
         if name == "metrics" and self.get_config().enable_metrics:
-            resources[METRICS_PREFIX] = MetricsResource(self)
+            resources[METRICS_PREFIX] = MetricsResource(RegistryProxy)
 
         if name == "replication":
             resources[REPLICATION_PREFIX] = ReplicationRestResource(self)
@@ -252,6 +263,13 @@ class SynapseHomeServer(HomeServer):
                     reactor.addSystemEventTrigger(
                         "before", "shutdown", server_listener.stopListening,
                     )
+            elif listener["type"] == "metrics":
+                if not self.get_config().enable_metrics:
+                    logger.warn(("Metrics listener configured, but "
+                                 "collect_metrics is not enabled!"))
+                else:
+                    _base.listen_metrics(listener["bind_addresses"],
+                                         listener["port"])
             else:
                 logger.warn("Unrecognized listener type: %s", listener["type"])
 
@@ -351,8 +369,6 @@ def setup(config_options):
         hs.get_datastore().start_doing_background_updates()
         hs.get_federation_client().start_get_pdu_cache()
 
-        register_memory_metrics(hs)
-
     reactor.callWhenRunning(start)
 
     return hs
@@ -423,6 +439,10 @@ def run(hs):
         total_nonbridged_users = yield hs.get_datastore().count_nonbridged_users()
         stats["total_nonbridged_users"] = total_nonbridged_users
 
+        daily_user_type_results = yield hs.get_datastore().count_daily_user_type()
+        for name, count in daily_user_type_results.iteritems():
+            stats["daily_user_type_" + name] = count
+
         room_count = yield hs.get_datastore().get_room_count()
         stats["total_room_count"] = room_count
 
@@ -473,6 +493,14 @@ def run(hs):
                 " changes across releases."
             )
 
+    def generate_user_daily_visit_stats():
+        hs.get_datastore().generate_user_daily_visits()
+
+    # Rather than update on per session basis, batch up the requests.
+    # If you increase the loop period, the accuracy of user_daily_visits
+    # table will decrease
+    clock.looping_call(generate_user_daily_visit_stats, 5 * 60 * 1000)
+
     if hs.config.report_stats:
         logger.info("Scheduling stats reporting for 3 hour intervals")
         clock.looping_call(phone_stats_home, 3 * 60 * 60 * 1000)

synapse/app/media_repository.py

@@ -27,6 +27,7 @@ from synapse.config.homeserver import HomeServerConfig
 from synapse.config.logger import setup_logging
 from synapse.crypto import context_factory
 from synapse.http.site import SynapseSite
+from synapse.metrics import RegistryProxy
 from synapse.metrics.resource import METRICS_PREFIX, MetricsResource
 from synapse.replication.slave.storage._base import BaseSlavedStore
 from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore
@@ -73,7 +74,7 @@ class MediaRepositoryServer(HomeServer):
         for res in listener_config["resources"]:
             for name in res["names"]:
                 if name == "metrics":
-                    resources[METRICS_PREFIX] = MetricsResource(self)
+                    resources[METRICS_PREFIX] = MetricsResource(RegistryProxy)
                 elif name == "media":
                     media_repo = self.get_media_repository_resource()
                     resources.update({
@@ -94,6 +95,7 @@ class MediaRepositoryServer(HomeServer):
                 site_tag,
                 listener_config,
                 root_resource,
+                self.version_string,
             )
         )
 
@@ -113,6 +115,13 @@ class MediaRepositoryServer(HomeServer):
                         globals={"hs": self},
                     )
                 )
+            elif listener["type"] == "metrics":
+                if not self.get_config().enable_metrics:
+                    logger.warn(("Metrics listener configured, but "
+                                 "collect_metrics is not enabled!"))
+                else:
+                    _base.listen_metrics(listener["bind_addresses"],
+                                         listener["port"])
             else:
                 logger.warn("Unrecognized listener type: %s", listener["type"])
 

synapse/app/pusher.py

@@ -23,6 +23,7 @@ from synapse.config._base import ConfigError
 from synapse.config.homeserver import HomeServerConfig
 from synapse.config.logger import setup_logging
 from synapse.http.site import SynapseSite
+from synapse.metrics import RegistryProxy
 from synapse.metrics.resource import METRICS_PREFIX, MetricsResource
 from synapse.replication.slave.storage.account_data import SlavedAccountDataStore
 from synapse.replication.slave.storage.events import SlavedEventStore
@@ -33,7 +34,7 @@ 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
@@ -92,7 +93,7 @@ class PusherServer(HomeServer):
         for res in listener_config["resources"]:
             for name in res["names"]:
                 if name == "metrics":
-                    resources[METRICS_PREFIX] = MetricsResource(self)
+                    resources[METRICS_PREFIX] = MetricsResource(RegistryProxy)
 
         root_resource = create_resource_tree(resources, NoResource())
 
@@ -104,6 +105,7 @@ class PusherServer(HomeServer):
                 site_tag,
                 listener_config,
                 root_resource,
+                self.version_string,
             )
         )
 
@@ -123,6 +125,13 @@ class PusherServer(HomeServer):
                         globals={"hs": self},
                     )
                 )
+            elif listener["type"] == "metrics":
+                if not self.get_config().enable_metrics:
+                    logger.warn(("Metrics listener configured, but "
+                                 "collect_metrics is not enabled!"))
+                else:
+                    _base.listen_metrics(listener["bind_addresses"],
+                                         listener["port"])
             else:
                 logger.warn("Unrecognized listener type: %s", listener["type"])
 
@@ -140,10 +149,11 @@ 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 +168,8 @@ 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

@@ -26,6 +26,7 @@ from synapse.config.logger import setup_logging
 from synapse.handlers.presence import PresenceHandler, get_interested_parties
 from synapse.http.server import JsonResource
 from synapse.http.site import SynapseSite
+from synapse.metrics import RegistryProxy
 from synapse.metrics.resource import METRICS_PREFIX, MetricsResource
 from synapse.replication.slave.storage._base import BaseSlavedStore
 from synapse.replication.slave.storage.account_data import SlavedAccountDataStore
@@ -51,7 +52,7 @@ 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
@@ -257,7 +258,7 @@ class SynchrotronServer(HomeServer):
         for res in listener_config["resources"]:
             for name in res["names"]:
                 if name == "metrics":
-                    resources[METRICS_PREFIX] = MetricsResource(self)
+                    resources[METRICS_PREFIX] = MetricsResource(RegistryProxy)
                 elif name == "client":
                     resource = JsonResource(self, canonical_json=False)
                     sync.register_servlets(self, resource)
@@ -281,6 +282,7 @@ class SynchrotronServer(HomeServer):
                 site_tag,
                 listener_config,
                 root_resource,
+                self.version_string,
             )
         )
 
@@ -300,6 +302,13 @@ class SynchrotronServer(HomeServer):
                         globals={"hs": self},
                     )
                 )
+            elif listener["type"] == "metrics":
+                if not self.get_config().enable_metrics:
+                    logger.warn(("Metrics listener configured, but "
+                                 "collect_metrics is not enabled!"))
+                else:
+                    _base.listen_metrics(listener["bind_addresses"],
+                                         listener["port"])
             else:
                 logger.warn("Unrecognized listener type: %s", listener["type"])
 
@@ -327,8 +336,7 @@ 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()
@@ -340,6 +348,7 @@ 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.
@@ -389,6 +398,8 @@ 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/user_dir.py

@@ -26,6 +26,7 @@ from synapse.config.logger import setup_logging
 from synapse.crypto import context_factory
 from synapse.http.server import JsonResource
 from synapse.http.site import SynapseSite
+from synapse.metrics import RegistryProxy
 from synapse.metrics.resource import METRICS_PREFIX, MetricsResource
 from synapse.replication.slave.storage._base import BaseSlavedStore
 from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore
@@ -39,10 +40,10 @@ 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 NoResource
 
 logger = logging.getLogger("synapse.app.user_dir")
@@ -105,7 +106,7 @@ class UserDirectoryServer(HomeServer):
         for res in listener_config["resources"]:
             for name in res["names"]:
                 if name == "metrics":
-                    resources[METRICS_PREFIX] = MetricsResource(self)
+                    resources[METRICS_PREFIX] = MetricsResource(RegistryProxy)
                 elif name == "client":
                     resource = JsonResource(self, canonical_json=False)
                     user_directory.register_servlets(self, resource)
@@ -126,6 +127,7 @@ class UserDirectoryServer(HomeServer):
                 site_tag,
                 listener_config,
                 root_resource,
+                self.version_string,
             )
         )
 
@@ -145,6 +147,13 @@ class UserDirectoryServer(HomeServer):
                         globals={"hs": self},
                     )
                 )
+            elif listener["type"] == "metrics":
+                if not self.get_config().enable_metrics:
+                    logger.warn(("Metrics listener configured, but "
+                                 "collect_metrics is not enabled!"))
+                else:
+                    _base.listen_metrics(listener["bind_addresses"],
+                                         listener["port"])
             else:
                 logger.warn("Unrecognized listener type: %s", listener["type"])
 
@@ -164,7 +173,14 @@ 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/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/__init__.py

@@ -12,3 +12,9 @@
 # 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 ._base import ConfigError
+
+# export ConfigError if somebody does import *
+# this is largely a fudge to stop PEP8 moaning about the import
+__all__ = ["ConfigError"]

synapse/config/_base.py

@@ -281,15 +281,15 @@ 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((
                     "A config file has been generated in %r for server name"
                     " %r with corresponding SSL keys and self-signed"

synapse/config/appservice.py

@@ -17,11 +17,11 @@ 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__)
 
@@ -105,7 +105,7 @@ def _load_appservice(hostname, as_info, 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."
         )

synapse/config/consent_config.py

@@ -0,0 +1,85 @@
+# -*- coding: utf-8 -*-
+# 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.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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 ._base import Config
+
+DEFAULT_CONFIG = """\
+# User Consent configuration
+#
+# Parts of this section are required if enabling the 'consent' resource under
+# 'listeners', in particular 'template_dir' and 'version'.
+#
+# 'template_dir' gives the location of the templates for the HTML forms.
+# This directory should contain one subdirectory per language (eg, 'en', 'fr'),
+# and each language directory should contain the policy document (named as
+# '<version>.html') and a success page (success.html).
+#
+# 'version' specifies the 'current' version of the policy document. It defines
+# the version to be served by the consent resource if there is no 'v'
+# parameter.
+#
+# 'server_notice_content', if enabled, will send a user a "Server Notice"
+# asking them to consent to the privacy policy. The 'server_notices' section
+# must also be configured for this to work. Notices will *not* be sent to
+# guest users unless 'send_server_notice_to_guests' is set to true.
+#
+# 'block_events_error', if set, will block any attempts to send events
+# until the user consents to the privacy policy. The value of the setting is
+# used as the text of the error.
+#
+# user_consent:
+#   template_dir: res/templates/privacy
+#   version: 1.0
+#   server_notice_content:
+#     msgtype: m.text
+#     body: >-
+#       To continue using this homeserver you must review and agree to the
+#       terms and conditions at %(consent_uri)s
+#   send_server_notice_to_guests: True
+#   block_events_error: >-
+#     To continue using this homeserver you must review and agree to the
+#     terms and conditions at %(consent_uri)s
+#
+"""
+
+
+class ConsentConfig(Config):
+    def __init__(self):
+        super(ConsentConfig, self).__init__()
+
+        self.user_consent_version = None
+        self.user_consent_template_dir = None
+        self.user_consent_server_notice_content = None
+        self.user_consent_server_notice_to_guests = False
+        self.block_events_without_consent_error = None
+
+    def read_config(self, config):
+        consent_config = config.get("user_consent")
+        if consent_config is None:
+            return
+        self.user_consent_version = str(consent_config["version"])
+        self.user_consent_template_dir = consent_config["template_dir"]
+        self.user_consent_server_notice_content = consent_config.get(
+            "server_notice_content",
+        )
+        self.block_events_without_consent_error = consent_config.get(
+            "block_events_error",
+        )
+        self.user_consent_server_notice_to_guests = bool(consent_config.get(
+            "send_server_notice_to_guests", False,
+        ))
+
+    def default_config(self, **kwargs):
+        return DEFAULT_CONFIG

synapse/config/homeserver.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.
@@ -12,7 +13,6 @@
 # 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.
-
 from .tls import TlsConfig
 from .server import ServerConfig
 from .logger import LoggingConfig
@@ -37,6 +37,8 @@ from .push import PushConfig
 from .spam_checker import SpamCheckerConfig
 from .groups import GroupsConfig
 from .user_directory import UserDirectoryConfig
+from .consent_config import ConsentConfig
+from .server_notices_config import ServerNoticesConfig
 
 
 class HomeServerConfig(TlsConfig, ServerConfig, DatabaseConfig, LoggingConfig,
@@ -45,12 +47,15 @@ class HomeServerConfig(TlsConfig, ServerConfig, DatabaseConfig, LoggingConfig,
                        AppServiceConfig, KeyConfig, SAML2Config, CasConfig,
                        JWTConfig, PasswordConfig, EmailConfig,
                        WorkerConfig, PasswordAuthProviderConfig, PushConfig,
-                       SpamCheckerConfig, GroupsConfig, UserDirectoryConfig,):
+                       SpamCheckerConfig, GroupsConfig, UserDirectoryConfig,
+                       ConsentConfig,
+                       ServerNoticesConfig,
+                       ):
     pass
 
 
 if __name__ == '__main__':
     import sys
     sys.stdout.write(
-        HomeServerConfig().generate_config(sys.argv[1], sys.argv[2])[0]
+        HomeServerConfig().generate_config(sys.argv[1], sys.argv[2], True)[0]
     )

synapse/config/key.py

@@ -59,14 +59,20 @@ class KeyConfig(Config):
 
         self.expire_access_token = config.get("expire_access_token", False)
 
+        # a secret which is used to calculate HMACs for form values, to stop
+        # falsification of values
+        self.form_secret = config.get("form_secret", None)
+
     def default_config(self, config_dir_path, server_name, is_generating_file=False,
                        **kwargs):
         base_key_name = os.path.join(config_dir_path, server_name)
 
         if is_generating_file:
             macaroon_secret_key = random_string_with_symbols(50)
+            form_secret = '"%s"' % random_string_with_symbols(50)
         else:
             macaroon_secret_key = None
+            form_secret = 'null'
 
         return """\
         macaroon_secret_key: "%(macaroon_secret_key)s"
@@ -74,6 +80,10 @@ class KeyConfig(Config):
         # Used to enable access token expiration.
         expire_access_token: False
 
+        # a secret which is used to calculate HMACs for form values, to stop
+        # falsification of values
+        form_secret: %(form_secret)s
+
         ## Signing Keys ##
 
         # Path to the signing key to sign messages with

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/repository.py

@@ -250,6 +250,9 @@ class ContentRepositoryConfig(Config):
         # - '192.168.0.0/16'
         # - '100.64.0.0/10'
         # - '169.254.0.0/16'
+        # - '::1/128'
+        # - 'fe80::/64'
+        # - 'fc00::/7'
         #
         # List of IP address CIDR ranges that the URL preview spider is allowed
         # to access even if they are specified in url_preview_ip_range_blacklist.

synapse/config/server.py

@@ -14,8 +14,12 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+import logging
+
 from ._base import Config, ConfigError
 
+logger = logging.Logger(__name__)
+
 
 class ServerConfig(Config):
 
@@ -138,6 +142,12 @@ class ServerConfig(Config):
 
         metrics_port = config.get("metrics_port")
         if metrics_port:
+            logger.warn(
+                ("The metrics_port configuration option is deprecated in Synapse 0.31 "
+                 "in favour of a listener. Please see "
+                 "http://github.com/matrix-org/synapse/blob/master/docs/metrics-howto.rst"
+                 " on how to configure the new listener."))
+
             self.listeners.append({
                 "port": metrics_port,
                 "bind_addresses": [config.get("metrics_bind_host", "127.0.0.1")],

synapse/config/server_notices_config.py

@@ -0,0 +1,86 @@
+# -*- coding: utf-8 -*-
+# 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.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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 ._base import Config
+from synapse.types import UserID
+
+DEFAULT_CONFIG = """\
+# Server Notices room configuration
+#
+# Uncomment this section to enable a room which can be used to send notices
+# from the server to users. It is a special room which cannot be left; notices
+# come from a special "notices" user id.
+#
+# If you uncomment this section, you *must* define the system_mxid_localpart
+# setting, which defines the id of the user which will be used to send the
+# notices.
+#
+# It's also possible to override the room name, the display name of the
+# "notices" user, and the avatar for the user.
+#
+# server_notices:
+#   system_mxid_localpart: notices
+#   system_mxid_display_name: "Server Notices"
+#   system_mxid_avatar_url: "mxc://server.com/oumMVlgDnLYFaPVkExemNVVZ"
+#   room_name: "Server Notices"
+"""
+
+
+class ServerNoticesConfig(Config):
+    """Configuration for the server notices room.
+
+    Attributes:
+        server_notices_mxid (str|None):
+            The MXID to use for server notices.
+            None if server notices are not enabled.
+
+        server_notices_mxid_display_name (str|None):
+            The display name to use for the server notices user.
+            None if server notices are not enabled.
+
+        server_notices_mxid_avatar_url (str|None):
+            The display name to use for the server notices user.
+            None if server notices are not enabled.
+
+        server_notices_room_name (str|None):
+            The name to use for the server notices room.
+            None if server notices are not enabled.
+    """
+    def __init__(self):
+        super(ServerNoticesConfig, self).__init__()
+        self.server_notices_mxid = None
+        self.server_notices_mxid_display_name = None
+        self.server_notices_mxid_avatar_url = None
+        self.server_notices_room_name = None
+
+    def read_config(self, config):
+        c = config.get("server_notices")
+        if c is None:
+            return
+
+        mxid_localpart = c['system_mxid_localpart']
+        self.server_notices_mxid = UserID(
+            mxid_localpart, self.server_name,
+        ).to_string()
+        self.server_notices_mxid_display_name = c.get(
+            'system_mxid_display_name', None,
+        )
+        self.server_notices_mxid_avatar_url = c.get(
+            'system_mxid_avatar_url', None,
+        )
+        # todo: i18n
+        self.server_notices_room_name = c.get('room_name', "Server Notices")
+
+    def default_config(self, **kwargs):
+        return DEFAULT_CONFIG

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
             ],
@@ -358,7 +363,7 @@ class Keyring(object):
 
         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,
@@ -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()

synapse/event_auth.py

@@ -34,9 +34,11 @@ def check(event, auth_events, do_sig_check=True, do_size_check=True):
         event: the event being checked.
         auth_events (dict: event-key -> event): the existing room state.
 
+    Raises:
+        AuthError if the checks fail
 
     Returns:
-        True if the auth checks pass.
+         if the auth checks pass.
     """
     if do_size_check:
         _check_size_limits(event)
@@ -71,7 +73,7 @@ def check(event, auth_events, do_sig_check=True, do_size_check=True):
         # Oh, we don't know what the state of the room was, so we
         # are trusting that this is allowed (at least for now)
         logger.warn("Trusting event: %s", event.event_id)
-        return True
+        return
 
     if event.type == EventTypes.Create:
         room_id_domain = get_domain_from_id(event.room_id)
@@ -81,7 +83,8 @@ def check(event, auth_events, do_sig_check=True, do_size_check=True):
                 "Creation event's room_id domain does not match sender's"
             )
         # FIXME
-        return True
+        logger.debug("Allowing! %s", event)
+        return
 
     creation_event = auth_events.get((EventTypes.Create, ""), None)
 
@@ -118,7 +121,8 @@ def check(event, auth_events, do_sig_check=True, do_size_check=True):
                 403,
                 "Alias event's state_key does not match sender's domain"
             )
-        return True
+        logger.debug("Allowing! %s", event)
+        return
 
     if logger.isEnabledFor(logging.DEBUG):
         logger.debug(
@@ -127,14 +131,9 @@ def check(event, auth_events, do_sig_check=True, do_size_check=True):
         )
 
     if event.type == EventTypes.Member:
-        allowed = _is_membership_change_allowed(
+        _is_membership_change_allowed(event, auth_events)
-            event, auth_events
-        )
-        if allowed:
         logger.debug("Allowing! %s", event)
-        else:
+        return
-            logger.debug("Denying! %s", event)
-        return allowed
 
     _check_event_sender_in_room(event, auth_events)
 
@@ -153,7 +152,8 @@ def check(event, auth_events, do_sig_check=True, do_size_check=True):
                 )
             )
         else:
-            return True
+            logger.debug("Allowing! %s", event)
+            return
 
     _can_send_event(event, auth_events)
 
@@ -200,7 +200,7 @@ def _is_membership_change_allowed(event, auth_events):
         create = auth_events.get(key)
         if create and event.prev_events[0][0] == create.event_id:
             if create.content["creator"] == event.state_key:
-                return True
+                return
 
     target_user_id = event.state_key
 
@@ -265,13 +265,13 @@ def _is_membership_change_allowed(event, auth_events):
             raise AuthError(
                 403, "%s is banned from the room" % (target_user_id,)
             )
-        return True
+        return
 
     if Membership.JOIN != membership:
         if (caller_invited
                 and Membership.LEAVE == membership
                 and target_user_id == event.user_id):
-            return True
+            return
 
         if not caller_in_room:  # caller isn't joined
             raise AuthError(
@@ -334,8 +334,6 @@ def _is_membership_change_allowed(event, auth_events):
     else:
         raise AuthError(500, "Unknown membership %s" % membership)
 
-    return True
-
 
 def _check_event_sender_in_room(event, auth_events):
     key = (EventTypes.Member, event.user_id, )
@@ -355,35 +353,46 @@ def _check_joined_room(member, user_id, room_id):
         ))
 
 
-def get_send_level(etype, state_key, auth_events):
+def get_send_level(etype, state_key, power_levels_event):
-    key = (EventTypes.PowerLevels, "", )
+    """Get the power level required to send an event of a given type
-    send_level_event = auth_events.get(key)
+
-    send_level = None
+    The federation spec [1] refers to this as "Required Power Level".
-    if send_level_event:
+
-        send_level = send_level_event.content.get("events", {}).get(
+    https://matrix.org/docs/spec/server_server/unstable.html#definitions
-            etype
+
-        )
+    Args:
+        etype (str): type of event
+        state_key (str|None): state_key of state event, or None if it is not
+            a state event.
+        power_levels_event (synapse.events.EventBase|None): power levels event
+            in force at this point in the room
+    Returns:
+        int: power level required to send this event.
+    """
+
+    if power_levels_event:
+        power_levels_content = power_levels_event.content
+    else:
+        power_levels_content = {}
+
+    # see if we have a custom level for this event type
+    send_level = power_levels_content.get("events", {}).get(etype)
+
+    # otherwise, fall back to the state_default/events_default.
     if send_level is None:
         if state_key is not None:
-                send_level = send_level_event.content.get(
+            send_level = power_levels_content.get("state_default", 50)
-                    "state_default", 50
-                )
         else:
-                send_level = send_level_event.content.get(
+            send_level = power_levels_content.get("events_default", 0)
-                    "events_default", 0
-                )
 
-    if send_level:
+    return int(send_level)
-        send_level = int(send_level)
-    else:
-        send_level = 0
-
-    return send_level
 
 
 def _can_send_event(event, auth_events):
+    power_levels_event = _get_power_level_event(auth_events)
+
     send_level = get_send_level(
-        event.type, event.get("state_key", None), auth_events
+        event.type, event.get("state_key"), power_levels_event,
     )
     user_level = get_user_power_level(event.user_id, auth_events)
 
@@ -471,14 +480,14 @@ def _check_power_levels(event, auth_events):
     ]
 
     old_list = current_state.content.get("users", {})
-    for user in set(old_list.keys() + user_list.keys()):
+    for user in set(list(old_list) + list(user_list)):
         levels_to_check.append(
             (user, "users")
         )
 
     old_list = current_state.content.get("events", {})
     new_list = event.content.get("events", {})
-    for ev_id in set(old_list.keys() + new_list.keys()):
+    for ev_id in set(list(old_list) + list(new_list)):
         levels_to_check.append(
             (ev_id, "events")
         )
@@ -524,13 +533,22 @@ def _check_power_levels(event, auth_events):
 
 
 def _get_power_level_event(auth_events):
-    key = (EventTypes.PowerLevels, "", )
+    return auth_events.get((EventTypes.PowerLevels, ""))
-    return auth_events.get(key)
 
 
 def get_user_power_level(user_id, auth_events):
-    power_level_event = _get_power_level_event(auth_events)
+    """Get a user's power level
 
+    Args:
+        user_id (str): user's id to look up in power_levels
+        auth_events (dict[(str, str), synapse.events.EventBase]):
+            state in force at this point in the room (or rather, a subset of
+            it including at least the create event and power levels event.
+
+    Returns:
+        int: the user's power level in this room.
+    """
+    power_level_event = _get_power_level_event(auth_events)
     if power_level_event:
         level = power_level_event.content.get("users", {}).get(user_id)
         if not level:
@@ -541,6 +559,11 @@ def get_user_power_level(user_id, auth_events):
         else:
             return int(level)
     else:
+        # if there is no power levels event, the creator gets 100 and everyone
+        # else gets 0.
+
+        # some things which call this don't pass the create event: hack around
+        # that.
         key = (EventTypes.Create, "", )
         create_event = auth_events.get(key)
         if (create_event is not None and

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,
@@ -134,7 +146,7 @@ class EventBase(object):
         return field in self._event_dict
 
     def items(self):
-        return self._event_dict.items()
+        return list(self._event_dict.items())
 
 
 class FrozenEvent(EventBase):

synapse/events/utils.py

@@ -20,6 +20,8 @@ from frozendict import frozendict
 
 import re
 
+from six import string_types
+
 # Split strings on "." but not "\." This uses a negative lookbehind assertion for '\'
 # (?<!stuff) matches if the current position in the string is not preceded
 # by a match for 'stuff'.
@@ -277,7 +279,7 @@ def serialize_event(e, time_now_ms, as_client_event=True,
 
     if only_event_fields:
         if (not isinstance(only_event_fields, list) or
-                not all(isinstance(f, basestring) for f in only_event_fields)):
+                not all(isinstance(f, string_types) for f in only_event_fields)):
             raise TypeError("only_event_fields must be a list of strings")
         d = only_fields(d, only_event_fields)
 

synapse/events/validator.py

@@ -17,6 +17,8 @@ from synapse.types import EventID, RoomID, UserID
 from synapse.api.errors import SynapseError
 from synapse.api.constants import EventTypes, Membership
 
+from six import string_types
+
 
 class EventValidator(object):
 
@@ -49,7 +51,7 @@ class EventValidator(object):
             strings.append("state_key")
 
         for s in strings:
-            if not isinstance(getattr(event, s), basestring):
+            if not isinstance(getattr(event, s), string_types):
                 raise SynapseError(400, "Not '%s' a string type" % (s,))
 
         if event.type == EventTypes.Member:
@@ -88,5 +90,5 @@ class EventValidator(object):
         for s in keys:
             if s not in d:
                 raise SynapseError(400, "'%s' not in content" % (s,))
-            if not isinstance(d[s], basestring):
+            if not isinstance(d[s], string_types):
                 raise SynapseError(400, "Not '%s' a string type" % (s,))

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
@@ -30,20 +32,17 @@ from synapse.federation.federation_base import (
     FederationBase,
     event_from_pdu_json,
 )
-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
 
+from prometheus_client import Counter
+
 logger = logging.getLogger(__name__)
 
-
+sent_queries_counter = Counter("synapse_federation_client_sent_queries", "", ["type"])
-# synapse.federation.federation_client is a silly name
-metrics = synapse.metrics.get_metrics_for("synapse.federation.client")
-
-sent_queries_counter = metrics.register_counter("sent_queries", labels=["type"])
 
 
 PDU_RETRY_TIME_MS = 1 * 60 * 1000
@@ -106,7 +105,7 @@ class FederationClient(FederationBase):
             a Deferred which will eventually yield a JSON object from the
             response
         """
-        sent_queries_counter.inc(query_type)
+        sent_queries_counter.labels(query_type).inc()
 
         return self.transport_layer.make_query(
             destination, query_type, args, retry_on_dns_fail=retry_on_dns_fail,
@@ -125,7 +124,7 @@ class FederationClient(FederationBase):
             a Deferred which will eventually yield a JSON object from the
             response
         """
-        sent_queries_counter.inc("client_device_keys")
+        sent_queries_counter.labels("client_device_keys").inc()
         return self.transport_layer.query_client_keys(
             destination, content, timeout
         )
@@ -135,7 +134,7 @@ class FederationClient(FederationBase):
         """Query the device keys for a list of user ids hosted on a remote
         server.
         """
-        sent_queries_counter.inc("user_devices")
+        sent_queries_counter.labels("user_devices").inc()
         return self.transport_layer.query_user_devices(
             destination, user_id, timeout
         )
@@ -152,7 +151,7 @@ class FederationClient(FederationBase):
             a Deferred which will eventually yield a JSON object from the
             response
         """
-        sent_queries_counter.inc("client_one_time_keys")
+        sent_queries_counter.labels("client_one_time_keys").inc()
         return self.transport_layer.claim_client_keys(
             destination, content, timeout
         )
@@ -392,7 +391,7 @@ class FederationClient(FederationBase):
         """
         if return_local:
             seen_events = yield self.store.get_events(event_ids, allow_rejected=True)
-            signed_events = seen_events.values()
+            signed_events = list(seen_events.values())
         else:
             seen_events = yield self.store.have_seen_events(event_ids)
             signed_events = []
@@ -413,11 +412,12 @@ 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,
                 )
@@ -589,7 +589,7 @@ class FederationClient(FederationBase):
                 }
 
                 valid_pdus = yield self._check_sigs_and_hash_and_fetch(
-                    destination, pdus.values(),
+                    destination, list(pdus.values()),
                     outlier=True,
                 )
 

synapse/federation/federation_server.py

@@ -27,12 +27,13 @@ from synapse.federation.federation_base import (
 
 from synapse.federation.persistence import TransactionActions
 from synapse.federation.units import Edu, Transaction
-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.logutils import log_function
 
+from prometheus_client import Counter
+
 from six import iteritems
 
 # when processing incoming transactions, we try to handle multiple rooms in
@@ -41,17 +42,17 @@ TRANSACTION_CONCURRENCY_LIMIT = 10
 
 logger = logging.getLogger(__name__)
 
-# synapse.federation.federation_server is a silly name
+received_pdus_counter = Counter("synapse_federation_server_received_pdus", "")
-metrics = synapse.metrics.get_metrics_for("synapse.federation.server")
 
-received_pdus_counter = metrics.register_counter("received_pdus")
+received_edus_counter = Counter("synapse_federation_server_received_edus", "")
 
-received_edus_counter = metrics.register_counter("received_edus")
+received_queries_counter = Counter(
-
+    "synapse_federation_server_received_queries", "", ["type"]
-received_queries_counter = metrics.register_counter("received_queries", labels=["type"])
+)
 
 
 class FederationServer(FederationBase):
+
     def __init__(self, hs):
         super(FederationServer, self).__init__(hs)
 
@@ -131,7 +132,7 @@ class FederationServer(FederationBase):
 
         logger.debug("[%s] Transaction is new", transaction.transaction_id)
 
-        received_pdus_counter.inc_by(len(transaction.pdus))
+        received_pdus_counter.inc(len(transaction.pdus))
 
         pdus_by_room = {}
 
@@ -292,7 +293,7 @@ class FederationServer(FederationBase):
 
     @defer.inlineCallbacks
     def on_query_request(self, query_type, args):
-        received_queries_counter.inc(query_type)
+        received_queries_counter.labels(query_type).inc()
         resp = yield self.registry.on_query(query_type, args)
         defer.returnValue((200, resp))
 

synapse/federation/send_queue.py

@@ -33,7 +33,7 @@ from .units import Edu
 
 from synapse.storage.presence import UserPresenceState
 from synapse.util.metrics import Measure
-import synapse.metrics
+from synapse.metrics import LaterGauge
 
 from blist import sorteddict
 from collections import namedtuple
@@ -45,9 +45,6 @@ from six import itervalues, iteritems
 logger = logging.getLogger(__name__)
 
 
-metrics = synapse.metrics.get_metrics_for(__name__)
-
-
 class FederationRemoteSendQueue(object):
     """A drop in replacement for TransactionQueue"""
 
@@ -77,10 +74,8 @@ class FederationRemoteSendQueue(object):
         # lambda binds to the queue rather than to the name of the queue which
         # changes. ARGH.
         def register(name, queue):
-            metrics.register_callback(
+            LaterGauge("synapse_federation_send_queue_%s_size" % (queue_name,),
-                queue_name + "_size",
+                       "", [], lambda: len(queue))
-                lambda: len(queue),
-            )
 
         for queue_name in [
             "presence_map", "presence_changed", "keyed_edu", "keyed_edu_changed",
@@ -202,7 +197,7 @@ class FederationRemoteSendQueue(object):
 
         # We only want to send presence for our own users, so lets always just
         # filter here just in case.
-        local_states = filter(lambda s: self.is_mine_id(s.user_id), states)
+        local_states = list(filter(lambda s: self.is_mine_id(s.user_id), states))
 
         self.presence_map.update({state.user_id: state for state in local_states})
         self.presence_changed[pos] = [state.user_id for state in local_states]

synapse/federation/transaction_queue.py

@@ -26,23 +26,25 @@ from synapse.util.retryutils import NotRetryingDestination, get_retry_limiter
 from synapse.util.metrics import measure_func
 from synapse.handlers.presence import format_user_presence_state, get_interested_remotes
 import synapse.metrics
+from synapse.metrics import LaterGauge
+from synapse.metrics import (
+    sent_edus_counter,
+    sent_transactions_counter,
+    events_processed_counter,
+)
+
+from prometheus_client import Counter
+
+from six import itervalues
 
 import logging
 
 
 logger = logging.getLogger(__name__)
 
-metrics = synapse.metrics.get_metrics_for(__name__)
+sent_pdus_destination_dist = Counter(
-
+    "synapse_federation_transaction_queue_sent_pdu_destinations", ""
-client_metrics = synapse.metrics.get_metrics_for("synapse.federation.client")
-sent_pdus_destination_dist = client_metrics.register_distribution(
-    "sent_pdu_destinations"
 )
-sent_edus_counter = client_metrics.register_counter("sent_edus")
-
-sent_transactions_counter = client_metrics.register_counter("sent_transactions")
-
-events_processed_counter = client_metrics.register_counter("events_processed")
 
 
 class TransactionQueue(object):
@@ -69,8 +71,10 @@ class TransactionQueue(object):
         # done
         self.pending_transactions = {}
 
-        metrics.register_callback(
+        LaterGauge(
-            "pending_destinations",
+            "synapse_federation_transaction_queue_pending_destinations",
+            "",
+            [],
             lambda: len(self.pending_transactions),
         )
 
@@ -94,12 +98,16 @@ class TransactionQueue(object):
         # Map of destination -> (edu_type, key) -> Edu
         self.pending_edus_keyed_by_dest = edus_keyed = {}
 
-        metrics.register_callback(
+        LaterGauge(
-            "pending_pdus",
+            "synapse_federation_transaction_queue_pending_pdus",
+            "",
+            [],
             lambda: sum(map(len, pdus.values())),
         )
-        metrics.register_callback(
+        LaterGauge(
-            "pending_edus",
+            "synapse_federation_transaction_queue_pending_edus",
+            "",
+            [],
             lambda: (
                 sum(map(len, edus.values()))
                 + sum(map(len, presence.values()))
@@ -228,7 +236,7 @@ class TransactionQueue(object):
                 yield logcontext.make_deferred_yieldable(defer.gatherResults(
                     [
                         logcontext.run_in_background(handle_room_events, evs)
-                        for evs in events_by_room.itervalues()
+                        for evs in itervalues(events_by_room)
                     ],
                     consumeErrors=True
                 ))
@@ -241,18 +249,15 @@ class TransactionQueue(object):
                     now = self.clock.time_msec()
                     ts = yield self.store.get_received_ts(events[-1].event_id)
 
-                    synapse.metrics.event_processing_lag.set(
+                    synapse.metrics.event_processing_lag.labels(
-                        now - ts, "federation_sender",
+                        "federation_sender").set(now - ts)
-                    )
+                    synapse.metrics.event_processing_last_ts.labels(
-                    synapse.metrics.event_processing_last_ts.set(
+                        "federation_sender").set(ts)
-                        ts, "federation_sender",
-                    )
 
-                events_processed_counter.inc_by(len(events))
+                events_processed_counter.inc(len(events))
 
-                synapse.metrics.event_processing_positions.set(
+                synapse.metrics.event_processing_positions.labels(
-                    next_token, "federation_sender",
+                    "federation_sender").set(next_token)
-                )
 
         finally:
             self._is_processing = False
@@ -275,7 +280,7 @@ class TransactionQueue(object):
         if not destinations:
             return
 
-        sent_pdus_destination_dist.inc_by(len(destinations))
+        sent_pdus_destination_dist.inc(len(destinations))
 
         for destination in destinations:
             self.pending_pdus_by_dest.setdefault(destination, []).append(
@@ -322,7 +327,9 @@ class TransactionQueue(object):
                 if not states_map:
                     break
 
-                yield self._process_presence_inner(states_map.values())
+                yield self._process_presence_inner(list(states_map.values()))
+        except Exception:
+            logger.exception("Error sending presence states to servers")
         finally:
             self._processing_pending_presence = False
 

synapse/federation/transport/server.py

@@ -25,7 +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
@@ -152,11 +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

synapse/federation/units.py

@@ -74,8 +74,6 @@ class Transaction(JsonEncodedObject):
         "previous_ids",
         "pdus",
         "edus",
-        "transaction_id",
-        "destination",
         "pdu_failures",
     ]
 

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

@@ -20,6 +20,8 @@ from synapse.api.errors import SynapseError
 from synapse.types import GroupID, RoomID, UserID, get_domain_from_id
 from twisted.internet import defer
 
+from six import string_types
+
 logger = logging.getLogger(__name__)
 
 
@@ -431,7 +433,7 @@ class GroupsServerHandler(object):
                         "long_description"):
             if keyname in content:
                 value = content[keyname]
-                if not isinstance(value, basestring):
+                if not isinstance(value, string_types):
                     raise SynapseError(400, "%r value is not a string" % (keyname,))
                 profile[keyname] = value
 

synapse/handlers/__init__.py

@@ -14,9 +14,7 @@
 # limitations under the License.
 
 from .register import RegistrationHandler
-from .room import (
+from .room import RoomContextHandler
-    RoomCreationHandler, RoomContextHandler,
-)
 from .message import MessageHandler
 from .federation import FederationHandler
 from .directory import DirectoryHandler
@@ -47,7 +45,6 @@ class Handlers(object):
     def __init__(self, hs):
         self.registration_handler = RegistrationHandler(hs)
         self.message_handler = MessageHandler(hs)
-        self.room_creation_handler = RoomCreationHandler(hs)
         self.federation_handler = FederationHandler(hs)
         self.directory_handler = DirectoryHandler(hs)
         self.admin_handler = AdminHandler(hs)

synapse/handlers/_base.py

@@ -114,14 +114,14 @@ class BaseHandler(object):
             if guest_access != "can_join":
                 if context:
                     current_state = yield self.store.get_events(
-                        context.current_state_ids.values()
+                        list(context.current_state_ids.values())
                     )
                 else:
                     current_state = yield self.state_handler.get_current_state(
                         event.room_id
                     )
 
-                current_state = current_state.values()
+                current_state = list(current_state.values())
 
                 logger.info("maybe_kick_guest_users %r", current_state)
                 yield self.kick_guest_users(current_state)

synapse/handlers/appservice.py

@@ -15,20 +15,21 @@
 
 from twisted.internet import defer
 
+from six import itervalues
+
 import synapse
 from synapse.api.constants import EventTypes
 from synapse.util.metrics import Measure
 from synapse.util.logcontext import (
-    make_deferred_yieldable, preserve_fn, run_in_background,
+    make_deferred_yieldable, run_in_background,
 )
+from prometheus_client import Counter
 
 import logging
 
 logger = logging.getLogger(__name__)
 
-metrics = synapse.metrics.get_metrics_for(__name__)
+events_processed_counter = Counter("synapse_handlers_appservice_events_processed", "")
-
-events_processed_counter = metrics.register_counter("events_processed")
 
 
 def log_failure(failure):
@@ -111,9 +112,7 @@ 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):
@@ -122,7 +121,7 @@ class ApplicationServicesHandler(object):
 
                     yield make_deferred_yieldable(defer.gatherResults([
                         run_in_background(handle_room_events, evs)
-                        for evs in events_by_room.itervalues()
+                        for evs in itervalues(events_by_room)
                     ], consumeErrors=True))
 
                     yield self.store.set_appservice_last_pos(upper_bound)
@@ -130,18 +129,15 @@ class ApplicationServicesHandler(object):
                     now = self.clock.time_msec()
                     ts = yield self.store.get_received_ts(events[-1].event_id)
 
-                    synapse.metrics.event_processing_positions.set(
+                    synapse.metrics.event_processing_positions.labels(
-                        upper_bound, "appservice_sender",
+                        "appservice_sender").set(upper_bound)
-                    )
 
-                    events_processed_counter.inc_by(len(events))
+                    events_processed_counter.inc(len(events))
 
-                    synapse.metrics.event_processing_lag.set(
+                    synapse.metrics.event_processing_lag.labels(
-                        now - ts, "appservice_sender",
+                        "appservice_sender").set(now - ts)
-                    )
+                    synapse.metrics.event_processing_last_ts.labels(
-                    synapse.metrics.event_processing_last_ts.set(
+                        "appservice_sender").set(ts)
-                        ts, "appservice_sender",
-                    )
             finally:
                 self.is_processing = False
 
@@ -198,7 +194,10 @@ 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))
 
@@ -259,11 +258,15 @@ 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/auth.py

@@ -249,7 +249,7 @@ class AuthHandler(BaseHandler):
                 errordict = e.error_dict()
 
         for f in flows:
-            if len(set(f) - set(creds.keys())) == 0:
+            if len(set(f) - set(creds)) == 0:
                 # it's very useful to know what args are stored, but this can
                 # include the password in the case of registering, so only log
                 # the keys (confusingly, clientdict may contain a password
@@ -257,12 +257,12 @@ class AuthHandler(BaseHandler):
                 # and is not sensitive).
                 logger.info(
                     "Auth completed with creds: %r. Client dict has keys: %r",
-                    creds, clientdict.keys()
+                    creds, list(clientdict)
                 )
                 defer.returnValue((creds, clientdict, session['id']))
 
         ret = self._auth_dict_for_flows(flows, session)
-        ret['completed'] = creds.keys()
+        ret['completed'] = list(creds)
         ret.update(errordict)
         raise InteractiveAuthIncompleteError(
             ret,

synapse/handlers/deactivate_account.py

@@ -1,5 +1,5 @@
 # -*- coding: utf-8 -*-
-# Copyright 2017 New Vector Ltd
+# Copyright 2017, 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.
@@ -12,9 +12,11 @@
 # 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 import defer
+from twisted.internet import defer, reactor
 
 from ._base import BaseHandler
+from synapse.types import UserID, create_requester
+from synapse.util.logcontext import run_in_background
 
 import logging
 
@@ -27,6 +29,15 @@ class DeactivateAccountHandler(BaseHandler):
         super(DeactivateAccountHandler, self).__init__(hs)
         self._auth_handler = hs.get_auth_handler()
         self._device_handler = hs.get_device_handler()
+        self._room_member_handler = hs.get_room_member_handler()
+        self.user_directory_handler = hs.get_user_directory_handler()
+
+        # Flag that indicates whether the process to part users from rooms is running
+        self._user_parter_running = False
+
+        # Start the user parter loop so it can resume parting users from rooms where
+        # it left off (if it has work left to do).
+        reactor.callWhenRunning(self._start_user_parting)
 
     @defer.inlineCallbacks
     def deactivate_account(self, user_id):
@@ -50,3 +61,73 @@ class DeactivateAccountHandler(BaseHandler):
 
         yield self.store.user_delete_threepids(user_id)
         yield self.store.user_set_password_hash(user_id, None)
+
+        # Add the user to a table of users pending deactivation (ie.
+        # removal from all the rooms they're a member of)
+        yield self.store.add_user_pending_deactivation(user_id)
+
+        # delete from user directory
+        yield self.user_directory_handler.handle_user_deactivated(user_id)
+
+        # Now start the process that goes through that list and
+        # parts users from rooms (if it isn't already running)
+        self._start_user_parting()
+
+    def _start_user_parting(self):
+        """
+        Start the process that goes through the table of users
+        pending deactivation, if it isn't already running.
+
+        Returns:
+            None
+        """
+        if not self._user_parter_running:
+            run_in_background(self._user_parter_loop)
+
+    @defer.inlineCallbacks
+    def _user_parter_loop(self):
+        """Loop that parts deactivated users from rooms
+
+        Returns:
+            None
+        """
+        self._user_parter_running = True
+        logger.info("Starting user parter")
+        try:
+            while True:
+                user_id = yield self.store.get_user_pending_deactivation()
+                if user_id is None:
+                    break
+                logger.info("User parter parting %r", user_id)
+                yield self._part_user(user_id)
+                yield self.store.del_user_pending_deactivation(user_id)
+                logger.info("User parter finished parting %r", user_id)
+            logger.info("User parter finished: stopping")
+        finally:
+            self._user_parter_running = False
+
+    @defer.inlineCallbacks
+    def _part_user(self, user_id):
+        """Causes the given user_id to leave all the rooms they're joined to
+
+        Returns:
+            None
+        """
+        user = UserID.from_string(user_id)
+
+        rooms_for_user = yield self.store.get_rooms_for_user(user_id)
+        for room_id in rooms_for_user:
+            logger.info("User parter parting %r from %r", user_id, room_id)
+            try:
+                yield self._room_member_handler.update_membership(
+                    create_requester(user),
+                    user,
+                    room_id,
+                    "leave",
+                    ratelimit=False,
+                )
+            except Exception:
+                logger.exception(
+                    "Failed to part user %r from room %r: ignoring and continuing",
+                    user_id, room_id,
+                )

synapse/handlers/device.py

@@ -26,6 +26,8 @@ from ._base import BaseHandler
 
 import logging
 
+from six import itervalues, iteritems
+
 logger = logging.getLogger(__name__)
 
 
@@ -112,7 +114,7 @@ class DeviceHandler(BaseHandler):
             user_id, device_id=None
         )
 
-        devices = device_map.values()
+        devices = list(device_map.values())
         for device in devices:
             _update_device_from_client_ips(device, ips)
 
@@ -185,7 +187,7 @@ class DeviceHandler(BaseHandler):
             defer.Deferred:
         """
         device_map = yield self.store.get_devices_by_user(user_id)
-        device_ids = device_map.keys()
+        device_ids = list(device_map)
         if except_device_id is not None:
             device_ids = [d for d in device_ids if d != except_device_id]
         yield self.delete_devices(user_id, device_ids)
@@ -318,7 +320,7 @@ class DeviceHandler(BaseHandler):
             # The user may have left the room
             # TODO: Check if they actually did or if we were just invited.
             if room_id not in room_ids:
-                for key, event_id in current_state_ids.iteritems():
+                for key, event_id in iteritems(current_state_ids):
                     etype, state_key = key
                     if etype != EventTypes.Member:
                         continue
@@ -338,7 +340,7 @@ class DeviceHandler(BaseHandler):
             # special-case for an empty prev state: include all members
             # in the changed list
             if not event_ids:
-                for key, event_id in current_state_ids.iteritems():
+                for key, event_id in iteritems(current_state_ids):
                     etype, state_key = key
                     if etype != EventTypes.Member:
                         continue
@@ -354,10 +356,10 @@ class DeviceHandler(BaseHandler):
 
             # Check if we've joined the room? If so we just blindly add all the users to
             # the "possibly changed" users.
-            for state_dict in prev_state_ids.itervalues():
+            for state_dict in itervalues(prev_state_ids):
                 member_event = state_dict.get((EventTypes.Member, user_id), None)
                 if not member_event or member_event != current_member_id:
-                    for key, event_id in current_state_ids.iteritems():
+                    for key, event_id in iteritems(current_state_ids):
                         etype, state_key = key
                         if etype != EventTypes.Member:
                             continue
@@ -367,14 +369,14 @@ class DeviceHandler(BaseHandler):
             # If there has been any change in membership, include them in the
             # possibly changed list. We'll check if they are joined below,
             # and we're not toooo worried about spuriously adding users.
-            for key, event_id in current_state_ids.iteritems():
+            for key, event_id in iteritems(current_state_ids):
                 etype, state_key = key
                 if etype != EventTypes.Member:
                     continue
 
                 # check if this member has changed since any of the extremities
                 # at the stream_ordering, and add them to the list if so.
-                for state_dict in prev_state_ids.itervalues():
+                for state_dict in itervalues(prev_state_ids):
                     prev_event_id = state_dict.get(key, None)
                     if not prev_event_id or prev_event_id != event_id:
                         if state_key != user_id:

synapse/handlers/e2e_keys.py

@@ -19,12 +19,13 @@ import logging
 
 from canonicaljson import encode_canonical_json
 from twisted.internet import defer
+from six import iteritems
 
 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__)
@@ -92,7 +93,7 @@ class E2eKeysHandler(object):
         remote_queries_not_in_cache = {}
         if remote_queries:
             query_list = []
-            for user_id, device_ids in remote_queries.iteritems():
+            for user_id, device_ids in iteritems(remote_queries):
                 if device_ids:
                     query_list.extend((user_id, device_id) for device_id in device_ids)
                 else:
@@ -103,9 +104,9 @@ class E2eKeysHandler(object):
                     query_list
                 )
             )
-            for user_id, devices in remote_results.iteritems():
+            for user_id, devices in iteritems(remote_results):
                 user_devices = results.setdefault(user_id, {})
-                for device_id, device in devices.iteritems():
+                for device_id, device in iteritems(devices):
                     keys = device.get("keys", None)
                     device_display_name = device.get("device_display_name", None)
                     if keys:
@@ -139,9 +140,9 @@ class E2eKeysHandler(object):
                 failures[destination] = _exception_to_failure(e)
 
         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,
@@ -242,17 +243,17 @@ class E2eKeysHandler(object):
                 failures[destination] = _exception_to_failure(e)
 
         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",
             ",".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)
             )),
         )
 

synapse/handlers/events.py

@@ -48,6 +48,7 @@ class EventStreamHandler(BaseHandler):
 
         self.notifier = hs.get_notifier()
         self.state = hs.get_state_handler()
+        self._server_notices_sender = hs.get_server_notices_sender()
 
     @defer.inlineCallbacks
     @log_function
@@ -58,6 +59,10 @@ class EventStreamHandler(BaseHandler):
 
         If `only_keys` is not None, events from keys will be sent down.
         """
+
+        # send any outstanding server notices to the user.
+        yield self._server_notices_sender.on_user_syncing(auth_user_id)
+
         auth_user = UserID.from_string(auth_user_id)
         presence_handler = self.hs.get_presence_handler()
 

synapse/handlers/federation.py

@@ -16,12 +16,15 @@
 
 """Contains handlers for federation events."""
 
-import httplib
 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 six import iteritems
 from twisted.internet import defer
 from unpaddedbase64 import decode_base64
 
@@ -49,7 +52,6 @@ from synapse.util.retryutils import NotRetryingDestination
 
 from synapse.util.distributor import user_joined_room
 
-
 logger = logging.getLogger(__name__)
 
 
@@ -79,6 +81,7 @@ class FederationHandler(BaseHandler):
         self.pusher_pool = hs.get_pusherpool()
         self.spam_checker = hs.get_spam_checker()
         self.event_creation_handler = hs.get_event_creation_handler()
+        self._server_notices_mxid = hs.config.server_notices_mxid
 
         # When joining a room we need to queue any events for that room up
         self.room_queues = {}
@@ -476,7 +479,7 @@ class FederationHandler(BaseHandler):
         # to get all state ids that we're interested in.
         event_map = yield self.store.get_events([
             e_id
-            for key_to_eid in event_to_state_ids.values()
+            for key_to_eid in list(event_to_state_ids.values())
             for key, e_id in key_to_eid.items()
             if key[0] != EventTypes.Member or check_match(key[1])
         ])
@@ -484,10 +487,10 @@ class FederationHandler(BaseHandler):
         event_to_state = {
             e_id: {
                 key: event_map[inner_e_id]
-                for key, inner_e_id in key_to_eid.items()
+                for key, inner_e_id in key_to_eid.iteritems()
                 if inner_e_id in event_map
             }
-            for e_id, key_to_eid in event_to_state_ids.items()
+            for e_id, key_to_eid in event_to_state_ids.iteritems()
         }
 
         def redact_disallowed(event, state):
@@ -502,7 +505,7 @@ class FederationHandler(BaseHandler):
                     # membership states for the requesting server to determine
                     # if the server is either in the room or has been invited
                     # into the room.
-                    for ev in state.values():
+                    for ev in state.itervalues():
                         if ev.type != EventTypes.Member:
                             continue
                         try:
@@ -637,7 +640,8 @@ 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,
@@ -747,9 +751,19 @@ class FederationHandler(BaseHandler):
         curr_state = yield self.state_handler.get_current_state(room_id)
 
         def get_domains_from_state(state):
+            """Get joined domains from state
+
+            Args:
+                state (dict[tuple, FrozenEvent]): State map from type/state
+                    key to event.
+
+            Returns:
+                list[tuple[str, int]]: Returns a list of servers with the
+                lowest depth of their joins. Sorted by lowest depth first.
+            """
             joined_users = [
                 (state_key, int(event.depth))
-                for (e_type, state_key), event in state.items()
+                for (e_type, state_key), event in state.iteritems()
                 if e_type == EventTypes.Member
                 and event.membership == Membership.JOIN
             ]
@@ -766,7 +780,7 @@ class FederationHandler(BaseHandler):
                 except Exception:
                     pass
 
-            return sorted(joined_domains.items(), key=lambda d: d[1])
+            return sorted(joined_domains.iteritems(), key=lambda d: d[1])
 
         curr_domains = get_domains_from_state(curr_state)
 
@@ -783,7 +797,7 @@ class FederationHandler(BaseHandler):
                     yield self.backfill(
                         dom, room_id,
                         limit=100,
-                        extremities=[e for e in extremities.keys()]
+                        extremities=extremities,
                     )
                     # If this succeeded then we probably already have the
                     # appropriate stuff.
@@ -829,7 +843,7 @@ class FederationHandler(BaseHandler):
         tried_domains = set(likely_domains)
         tried_domains.add(self.server_name)
 
-        event_ids = list(extremities.keys())
+        event_ids = list(extremities.iterkeys())
 
         logger.debug("calling resolve_state_groups in _maybe_backfill")
         resolve = logcontext.preserve_fn(
@@ -839,31 +853,34 @@ class FederationHandler(BaseHandler):
             [resolve(room_id, [e]) for e in event_ids],
             consumeErrors=True,
         ))
+
+        # dict[str, dict[tuple, str]], a map from event_id to state map of
+        # event_ids.
         states = dict(zip(event_ids, [s.state for s in states]))
 
         state_map = yield self.store.get_events(
-            [e_id for ids in states.values() for e_id in ids],
+            [e_id for ids in states.itervalues() for e_id in ids.itervalues()],
             get_prev_content=False
         )
         states = {
             key: {
                 k: state_map[e_id]
-                for k, e_id in state_dict.items()
+                for k, e_id in state_dict.iteritems()
                 if e_id in state_map
-            } for key, state_dict in states.items()
+            } for key, state_dict in states.iteritems()
         }
 
         for e_id, _ in sorted_extremeties_tuple:
             likely_domains = get_domains_from_state(states[e_id])
 
             success = yield try_backfill([
-                dom for dom in likely_domains
+                dom for dom, _ in likely_domains
                 if dom not in tried_domains
             ])
             if success:
                 defer.returnValue(True)
 
-            tried_domains.update(likely_domains)
+            tried_domains.update(dom for dom, _ in likely_domains)
 
         defer.returnValue(False)
 
@@ -887,7 +904,7 @@ class FederationHandler(BaseHandler):
             logger.warn("Rejecting event %s which has %i prev_events",
                         ev.event_id, len(ev.prev_events))
             raise SynapseError(
-                httplib.BAD_REQUEST,
+                http_client.BAD_REQUEST,
                 "Too many prev_events",
             )
 
@@ -895,7 +912,7 @@ class FederationHandler(BaseHandler):
             logger.warn("Rejecting event %s which has %i auth_events",
                         ev.event_id, len(ev.auth_events))
             raise SynapseError(
-                httplib.BAD_REQUEST,
+                http_client.BAD_REQUEST,
                 "Too many auth_events",
             )
 
@@ -1023,7 +1040,7 @@ 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)
 
@@ -1131,13 +1148,13 @@ class FederationHandler(BaseHandler):
                 user = UserID.from_string(event.state_key)
                 yield user_joined_room(self.distributor, user, event.room_id)
 
-        state_ids = context.prev_state_ids.values()
+        state_ids = list(context.prev_state_ids.values())
         auth_chain = yield self.store.get_auth_chain(state_ids)
 
-        state = yield self.store.get_events(context.prev_state_ids.values())
+        state = yield self.store.get_events(list(context.prev_state_ids.values()))
 
         defer.returnValue({
-            "state": state.values(),
+            "state": list(state.values()),
             "auth_chain": auth_chain,
         })
 
@@ -1177,6 +1194,13 @@ class FederationHandler(BaseHandler):
         if not self.is_mine_id(event.state_key):
             raise SynapseError(400, "The invite event must be for this server")
 
+        # block any attempts to invite the server notices mxid
+        if event.state_key == self._server_notices_mxid:
+            raise SynapseError(
+                http_client.FORBIDDEN,
+                "Cannot invite this user",
+            )
+
         event.internal_metadata.outlier = True
         event.internal_metadata.invite_from_remote = True
 
@@ -1364,7 +1388,7 @@ class FederationHandler(BaseHandler):
         )
 
         if state_groups:
-            _, state = state_groups.items().pop()
+            _, state = list(iteritems(state_groups)).pop()
             results = {
                 (e.type, e.state_key): e for e in state
             }
@@ -1380,7 +1404,7 @@ class FederationHandler(BaseHandler):
                 else:
                     del results[(event.type, event.state_key)]
 
-            res = results.values()
+            res = list(results.values())
             for event in res:
                 # We sign these again because there was a bug where we
                 # incorrectly signed things the first time round
@@ -1421,7 +1445,7 @@ class FederationHandler(BaseHandler):
                 else:
                     results.pop((event.type, event.state_key), None)
 
-            defer.returnValue(results.values())
+            defer.returnValue(list(results.values()))
         else:
             defer.returnValue([])
 
@@ -1513,18 +1537,21 @@ 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))
@@ -1538,7 +1565,8 @@ 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"),
@@ -1766,6 +1794,10 @@ class FederationHandler(BaseHandler):
             min_depth=min_depth,
         )
 
+        missing_events = yield self._filter_events_for_server(
+            origin, room_id, missing_events,
+        )
+
         defer.returnValue(missing_events)
 
     @defer.inlineCallbacks
@@ -1867,7 +1899,8 @@ 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,
@@ -1885,7 +1918,7 @@ class FederationHandler(BaseHandler):
                 })
 
                 new_state = self.state_handler.resolve_events(
-                    [local_view.values(), remote_view.values()],
+                    [list(local_view.values()), list(remote_view.values())],
                     event
                 )
 
@@ -2005,7 +2038,7 @@ class FederationHandler(BaseHandler):
                 this will not be included in the current_state in the context.
         """
         state_updates = {
-            k: a.event_id for k, a in auth_events.iteritems()
+            k: a.event_id for k, a in iteritems(auth_events)
             if k != event_key
         }
         context.current_state_ids = dict(context.current_state_ids)
@@ -2015,7 +2048,7 @@ class FederationHandler(BaseHandler):
             context.delta_ids.update(state_updates)
         context.prev_state_ids = dict(context.prev_state_ids)
         context.prev_state_ids.update({
-            k: a.event_id for k, a in auth_events.iteritems()
+            k: a.event_id for k, a in iteritems(auth_events)
         })
         context.state_group = yield self.store.store_state_group(
             event.event_id,
@@ -2067,7 +2100,7 @@ class FederationHandler(BaseHandler):
 
         def get_next(it, opt=None):
             try:
-                return it.next()
+                return next(it)
             except Exception:
                 return opt
 

synapse/handlers/groups_local.py

@@ -15,6 +15,7 @@
 # limitations under the License.
 
 from twisted.internet import defer
+from six import iteritems
 
 from synapse.api.errors import SynapseError
 from synapse.types import get_domain_from_id
@@ -449,7 +450,7 @@ class GroupsLocalHandler(object):
 
         results = {}
         failed_results = []
-        for destination, dest_user_ids in destinations.iteritems():
+        for destination, dest_user_ids in iteritems(destinations):
             try:
                 r = yield self.transport_client.bulk_get_publicised_groups(
                     destination, list(dest_user_ids),

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,
@@ -180,8 +181,8 @@ class InitialSyncHandler(BaseHandler):
                     self.store, user_id, messages
                 )
 
-                start_token = now_token.copy_and_replace("room_key", token[0])
+                start_token = now_token.copy_and_replace("room_key", token)
-                end_token = now_token.copy_and_replace("room_key", token[1])
+                end_token = now_token.copy_and_replace("room_key", room_end_token)
                 time_now = self.clock.time_msec()
 
                 d["messages"] = {
@@ -324,8 +325,8 @@ class InitialSyncHandler(BaseHandler):
             self.store, user_id, messages, is_peeking=is_peeking
         )
 
-        start_token = StreamToken.START.copy_and_replace("room_key", token[0])
+        start_token = StreamToken.START.copy_and_replace("room_key", token)
-        end_token = StreamToken.START.copy_and_replace("room_key", token[1])
+        end_token = StreamToken.START.copy_and_replace("room_key", stream_token)
 
         time_now = self.clock.time_msec()
 
@@ -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,
@@ -406,8 +408,8 @@ class InitialSyncHandler(BaseHandler):
             self.store, user_id, messages, is_peeking=is_peeking,
         )
 
-        start_token = now_token.copy_and_replace("room_key", token[0])
+        start_token = now_token.copy_and_replace("room_key", token)
-        end_token = now_token.copy_and_replace("room_key", token[1])
+        end_token = now_token
 
         time_now = self.clock.time_msec()
 

synapse/handlers/message.py

@@ -13,11 +13,23 @@
 # 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 six import string_types, itervalues, iteritems
 from twisted.internet import defer, reactor
+from twisted.internet.defer import succeed
 from twisted.python.failure import Failure
 
 from synapse.api.constants import EventTypes, Membership, MAX_DEPTH
-from synapse.api.errors import AuthError, Codes, SynapseError
+from synapse.api.errors import (
+    AuthError, Codes, SynapseError,
+    ConsentNotGivenError,
+)
+from synapse.api.urls import ConsentURIBuilder
 from synapse.crypto.event_signing import add_hashes_and_signatures
 from synapse.events.utils import serialize_event
 from synapse.events.validator import EventValidator
@@ -25,7 +37,7 @@ 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 frozendict_json_encoder
 from synapse.util.stringutils import random_string
@@ -34,11 +46,6 @@ from synapse.replication.http.send_event import send_event_to_master
 
 from ._base import BaseHandler
 
-from canonicaljson import encode_canonical_json
-
-import logging
-import simplejson
-
 logger = logging.getLogger(__name__)
 
 
@@ -85,14 +92,14 @@ class MessageHandler(BaseHandler):
         # map from purge id to PurgeStatus
         self._purges_by_id = {}
 
-    def start_purge_history(self, room_id, topological_ordering,
+    def start_purge_history(self, room_id, token,
                             delete_local_events=False):
         """Start off a history purge on a room.
 
         Args:
             room_id (str): The room to purge from
 
-            topological_ordering (int): minimum topo ordering to preserve
+            token (str): topological token to delete events before
             delete_local_events (bool): True to delete local events as well as
                 remote ones
 
@@ -114,19 +121,19 @@ class MessageHandler(BaseHandler):
         self._purges_by_id[purge_id] = PurgeStatus()
         run_in_background(
             self._purge_history,
-            purge_id, room_id, topological_ordering, delete_local_events,
+            purge_id, room_id, token, delete_local_events,
         )
         return purge_id
 
     @defer.inlineCallbacks
-    def _purge_history(self, purge_id, room_id, topological_ordering,
+    def _purge_history(self, purge_id, room_id, token,
                        delete_local_events):
         """Carry out a history purge on a room.
 
         Args:
             purge_id (str): The id for this purge
             room_id (str): The room to purge from
-            topological_ordering (int): minimum topo ordering to preserve
+            token (str): topological token to delete events before
             delete_local_events (bool): True to delete local events as well as
                 remote ones
 
@@ -137,7 +144,7 @@ class MessageHandler(BaseHandler):
         try:
             with (yield self.pagination_lock.write(room_id)):
                 yield self.store.purge_history(
-                    room_id, topological_ordering, delete_local_events,
+                    room_id, token, delete_local_events,
                 )
             logger.info("[purge] complete")
             self._purges_by_id[purge_id].status = PurgeStatus.STATUS_COMPLETE
@@ -396,7 +403,7 @@ class MessageHandler(BaseHandler):
                 "avatar_url": profile.avatar_url,
                 "display_name": profile.display_name,
             }
-            for user_id, profile in users_with_profile.iteritems()
+            for user_id, profile in iteritems(users_with_profile)
         })
 
 
@@ -430,6 +437,9 @@ class EventCreationHandler(object):
 
         self.spam_checker = hs.get_spam_checker()
 
+        if self.config.block_events_without_consent_error is not None:
+            self._consent_uri_builder = ConsentURIBuilder(self.config)
+
     @defer.inlineCallbacks
     def create_event(self, requester, event_dict, token_id=None, txn_id=None,
                      prev_events_and_hashes=None):
@@ -481,6 +491,10 @@ class EventCreationHandler(object):
                         target, e
                     )
 
+        is_exempt = yield self._is_exempt_from_privacy_policy(builder)
+        if not is_exempt:
+            yield self.assert_accepted_privacy_policy(requester)
+
         if token_id is not None:
             builder.internal_metadata.token_id = token_id
 
@@ -495,6 +509,86 @@ class EventCreationHandler(object):
 
         defer.returnValue((event, context))
 
+    def _is_exempt_from_privacy_policy(self, builder):
+        """"Determine if an event to be sent is exempt from having to consent
+        to the privacy policy
+
+        Args:
+            builder (synapse.events.builder.EventBuilder): event being created
+
+        Returns:
+            Deferred[bool]: true if the event can be sent without the user
+                consenting
+        """
+        # the only thing the user can do is join the server notices room.
+        if builder.type == EventTypes.Member:
+            membership = builder.content.get("membership", None)
+            if membership == Membership.JOIN:
+                return self._is_server_notices_room(builder.room_id)
+        return succeed(False)
+
+    @defer.inlineCallbacks
+    def _is_server_notices_room(self, room_id):
+        if self.config.server_notices_mxid is None:
+            defer.returnValue(False)
+        user_ids = yield self.store.get_users_in_room(room_id)
+        defer.returnValue(self.config.server_notices_mxid in user_ids)
+
+    @defer.inlineCallbacks
+    def assert_accepted_privacy_policy(self, requester):
+        """Check if a user has accepted the privacy policy
+
+        Called when the given user is about to do something that requires
+        privacy consent. We see if the user is exempt and otherwise check that
+        they have given consent. If they have not, a ConsentNotGiven error is
+        raised.
+
+        Args:
+            requester (synapse.types.Requester):
+                The user making the request
+
+        Returns:
+            Deferred[None]: returns normally if the user has consented or is
+                exempt
+
+        Raises:
+            ConsentNotGivenError: if the user has not given consent yet
+        """
+        if self.config.block_events_without_consent_error is None:
+            return
+
+        # exempt AS users from needing consent
+        if requester.app_service is not None:
+            return
+
+        user_id = requester.user.to_string()
+
+        # exempt the system notices user
+        if (
+            self.config.server_notices_mxid is not None and
+            user_id == self.config.server_notices_mxid
+        ):
+            return
+
+        u = yield self.store.get_user_by_id(user_id)
+        assert u is not None
+        if u["appservice_id"] is not None:
+            # users registered by an appservice are exempt
+            return
+        if u["consent_version"] == self.config.user_consent_version:
+            return
+
+        consent_uri = self._consent_uri_builder.build_user_consent_uri(
+            requester.user.localpart,
+        )
+        msg = self.config.block_events_without_consent_error % {
+            'consent_uri': consent_uri,
+        }
+        raise ConsentNotGivenError(
+            msg=msg,
+            consent_uri=consent_uri,
+        )
+
     @defer.inlineCallbacks
     def send_nonmember_event(self, requester, event, context, ratelimit=True):
         """
@@ -577,7 +671,7 @@ class EventCreationHandler(object):
 
             spam_error = self.spam_checker.check_event_for_spam(event)
             if spam_error:
-                if not isinstance(spam_error, basestring):
+                if not isinstance(spam_error, string_types):
                     spam_error = "Spam is not permitted here"
                 raise SynapseError(
                     403, spam_error, Codes.FORBIDDEN
@@ -733,8 +827,14 @@ 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(
@@ -785,7 +885,7 @@ class EventCreationHandler(object):
 
                 state_to_include_ids = [
                     e_id
-                    for k, e_id in context.current_state_ids.iteritems()
+                    for k, e_id in iteritems(context.current_state_ids)
                     if k[0] in self.hs.config.room_invite_state_types
                     or k == (EventTypes.Member, event.sender)
                 ]
@@ -799,7 +899,7 @@ class EventCreationHandler(object):
                         "content": e.content,
                         "sender": e.sender,
                     }
-                    for e in state_to_include.itervalues()
+                    for e in itervalues(state_to_include)
                 ]
 
                 invitee = UserID.from_string(event.state_key)
@@ -854,22 +954,33 @@ 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

@@ -25,38 +25,42 @@ The methods that define policy are:
 from twisted.internet import defer, reactor
 from contextlib import contextmanager
 
+from six import itervalues, iteritems
+
 from synapse.api.errors import SynapseError
 from synapse.api.constants import PresenceState
 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
 from synapse.types import UserID, get_domain_from_id
-import synapse.metrics
+from synapse.metrics import LaterGauge
 
 import logging
 
+from prometheus_client import Counter
 
 logger = logging.getLogger(__name__)
 
-metrics = synapse.metrics.get_metrics_for(__name__)
 
-notified_presence_counter = metrics.register_counter("notified_presence")
+notified_presence_counter = Counter("synapse_handler_presence_notified_presence", "")
-federation_presence_out_counter = metrics.register_counter("federation_presence_out")
+federation_presence_out_counter = Counter(
-presence_updates_counter = metrics.register_counter("presence_updates")
+    "synapse_handler_presence_federation_presence_out", "")
-timers_fired_counter = metrics.register_counter("timers_fired")
+presence_updates_counter = Counter("synapse_handler_presence_presence_updates", "")
-federation_presence_counter = metrics.register_counter("federation_presence")
+timers_fired_counter = Counter("synapse_handler_presence_timers_fired", "")
-bump_active_time_counter = metrics.register_counter("bump_active_time")
+federation_presence_counter = Counter("synapse_handler_presence_federation_presence", "")
+bump_active_time_counter = Counter("synapse_handler_presence_bump_active_time", "")
 
-get_updates_counter = metrics.register_counter("get_updates", labels=["type"])
+get_updates_counter = Counter("synapse_handler_presence_get_updates", "", ["type"])
 
-notify_reason_counter = metrics.register_counter("notify_reason", labels=["reason"])
+notify_reason_counter = Counter(
-state_transition_counter = metrics.register_counter(
+    "synapse_handler_presence_notify_reason", "", ["reason"])
-    "state_transition", labels=["from", "to"]
+state_transition_counter = Counter(
+    "synapse_handler_presence_state_transition", "", ["from", "to"]
 )
 
 
@@ -87,6 +91,11 @@ assert LAST_ACTIVE_GRANULARITY < IDLE_TIMER
 class PresenceHandler(object):
 
     def __init__(self, hs):
+        """
+
+        Args:
+            hs (synapse.server.HomeServer):
+        """
         self.is_mine = hs.is_mine
         self.is_mine_id = hs.is_mine_id
         self.clock = hs.get_clock()
@@ -94,7 +103,6 @@ class PresenceHandler(object):
         self.wheel_timer = WheelTimer()
         self.notifier = hs.get_notifier()
         self.federation = hs.get_federation_sender()
-
         self.state = hs.get_state_handler()
 
         federation_registry = hs.get_federation_registry()
@@ -137,8 +145,9 @@ class PresenceHandler(object):
             for state in active_presence
         }
 
-        metrics.register_callback(
+        LaterGauge(
-            "user_to_current_state_size", lambda: len(self.user_to_current_state)
+            "synapse_handlers_presence_user_to_current_state_size", "", [],
+            lambda: len(self.user_to_current_state)
         )
 
         now = self.clock.time_msec()
@@ -208,7 +217,8 @@ class PresenceHandler(object):
             60 * 1000,
         )
 
-        metrics.register_callback("wheel_timer_size", lambda: len(self.wheel_timer))
+        LaterGauge("synapse_handlers_presence_wheel_timer_size", "", [],
+                   lambda: len(self.wheel_timer))
 
     @defer.inlineCallbacks
     def _on_shutdown(self):
@@ -254,6 +264,14 @@ 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
@@ -303,11 +321,11 @@ class PresenceHandler(object):
 
             # TODO: We should probably ensure there are no races hereafter
 
-            presence_updates_counter.inc_by(len(new_states))
+            presence_updates_counter.inc(len(new_states))
 
             if to_notify:
-                notified_presence_counter.inc_by(len(to_notify))
+                notified_presence_counter.inc(len(to_notify))
-                yield self._persist_and_notify(to_notify.values())
+                yield self._persist_and_notify(list(to_notify.values()))
 
             self.unpersisted_users_changes |= set(s.user_id for s in new_states)
             self.unpersisted_users_changes -= set(to_notify.keys())
@@ -317,7 +335,7 @@ class PresenceHandler(object):
                 if user_id not in to_notify
             }
             if to_federation_ping:
-                federation_presence_out_counter.inc_by(len(to_federation_ping))
+                federation_presence_out_counter.inc(len(to_federation_ping))
 
                 self._push_to_remotes(to_federation_ping.values())
 
@@ -355,7 +373,7 @@ class PresenceHandler(object):
                     for user_id in users_to_check
                 ]
 
-                timers_fired_counter.inc_by(len(states))
+                timers_fired_counter.inc(len(states))
 
                 changes = handle_timeouts(
                     states,
@@ -364,7 +382,7 @@ 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 +440,23 @@ 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())
 
@@ -452,61 +473,6 @@ class PresenceHandler(object):
             syncing_user_ids.update(user_ids)
         return syncing_user_ids
 
-    @defer.inlineCallbacks
-    def update_external_syncs(self, process_id, syncing_user_ids):
-        """Update the syncing users for an external process
-
-        Args:
-            process_id(str): An identifier for the process the users are
-                syncing against. This allows synapse to process updates
-                as user start and stop syncing against a given process.
-            syncing_user_ids(set(str)): The set of user_ids that are
-                currently syncing on that server.
-        """
-
-        # Grab the previous list of user_ids that were syncing on that process
-        prev_syncing_user_ids = (
-            self.external_process_to_current_syncs.get(process_id, set())
-        )
-        # Grab the current presence state for both the users that are syncing
-        # now and the users that were syncing before this update.
-        prev_states = yield self.current_state_for_users(
-            syncing_user_ids | prev_syncing_user_ids
-        )
-        updates = []
-        time_now_ms = self.clock.time_msec()
-
-        # For each new user that is syncing check if we need to mark them as
-        # being online.
-        for new_user_id in syncing_user_ids - prev_syncing_user_ids:
-            prev_state = prev_states[new_user_id]
-            if prev_state.state == PresenceState.OFFLINE:
-                updates.append(prev_state.copy_and_replace(
-                    state=PresenceState.ONLINE,
-                    last_active_ts=time_now_ms,
-                    last_user_sync_ts=time_now_ms,
-                ))
-            else:
-                updates.append(prev_state.copy_and_replace(
-                    last_user_sync_ts=time_now_ms,
-                ))
-
-        # For each user that is still syncing or stopped syncing update the
-        # last sync time so that we will correctly apply the grace period when
-        # they stop syncing.
-        for old_user_id in prev_syncing_user_ids:
-            prev_state = prev_states[old_user_id]
-            updates.append(prev_state.copy_and_replace(
-                last_user_sync_ts=time_now_ms,
-            ))
-
-        yield self._update_states(updates)
-
-        # Update the last updated time for the process. We expire the entries
-        # if we don't receive an update in the given timeframe.
-        self.external_process_last_updated_ms[process_id] = self.clock.time_msec()
-        self.external_process_to_current_syncs[process_id] = syncing_user_ids
-
     @defer.inlineCallbacks
     def update_external_syncs_row(self, process_id, user_id, is_syncing, sync_time_msec):
         """Update the syncing users for an external process as a delta.
@@ -570,7 +536,7 @@ class PresenceHandler(object):
                 prev_state.copy_and_replace(
                     last_user_sync_ts=time_now_ms,
                 )
-                for prev_state in prev_states.itervalues()
+                for prev_state in itervalues(prev_states)
             ])
             self.external_process_last_updated_ms.pop(process_id, None)
 
@@ -593,14 +559,14 @@ class PresenceHandler(object):
             for user_id in user_ids
         }
 
-        missing = [user_id for user_id, state in states.iteritems() if not state]
+        missing = [user_id for user_id, state in iteritems(states) if not state]
         if missing:
             # There are things not in our in memory cache. Lets pull them out of
             # the database.
             res = yield self.store.get_presence_for_users(missing)
             states.update(res)
 
-            missing = [user_id for user_id, state in states.iteritems() if not state]
+            missing = [user_id for user_id, state in iteritems(states) if not state]
             if missing:
                 new = {
                     user_id: UserPresenceState.default(user_id)
@@ -696,7 +662,7 @@ class PresenceHandler(object):
             updates.append(prev_state.copy_and_replace(**new_fields))
 
         if updates:
-            federation_presence_counter.inc_by(len(updates))
+            federation_presence_counter.inc(len(updates))
             yield self._update_states(updates)
 
     @defer.inlineCallbacks
@@ -721,7 +687,7 @@ class PresenceHandler(object):
         """
 
         updates = yield self.current_state_for_users(target_user_ids)
-        updates = updates.values()
+        updates = list(updates.values())
 
         for user_id in set(target_user_ids) - set(u.user_id for u in updates):
             updates.append(UserPresenceState.default(user_id))
@@ -787,11 +753,11 @@ class PresenceHandler(object):
             self._push_to_remotes([state])
         else:
             user_ids = yield self.store.get_users_in_room(room_id)
-            user_ids = filter(self.is_mine_id, user_ids)
+            user_ids = list(filter(self.is_mine_id, user_ids))
 
             states = yield self.current_state_for_users(user_ids)
 
-            self._push_to_remotes(states.values())
+            self._push_to_remotes(list(states.values()))
 
     @defer.inlineCallbacks
     def get_presence_list(self, observer_user, accepted=None):
@@ -971,28 +937,28 @@ def should_notify(old_state, new_state):
         return False
 
     if old_state.status_msg != new_state.status_msg:
-        notify_reason_counter.inc("status_msg_change")
+        notify_reason_counter.labels("status_msg_change").inc()
         return True
 
     if old_state.state != new_state.state:
-        notify_reason_counter.inc("state_change")
+        notify_reason_counter.labels("state_change").inc()
-        state_transition_counter.inc(old_state.state, new_state.state)
+        state_transition_counter.labels(old_state.state, new_state.state).inc()
         return True
 
     if old_state.state == PresenceState.ONLINE:
         if new_state.currently_active != old_state.currently_active:
-            notify_reason_counter.inc("current_active_change")
+            notify_reason_counter.labels("current_active_change").inc()
             return True
 
         if new_state.last_active_ts - old_state.last_active_ts > LAST_ACTIVE_GRANULARITY:
             # Only notify about last active bumps if we're not currently acive
             if not new_state.currently_active:
-                notify_reason_counter.inc("last_active_change_online")
+                notify_reason_counter.labels("last_active_change_online").inc()
                 return True
 
     elif new_state.last_active_ts - old_state.last_active_ts > LAST_ACTIVE_GRANULARITY:
         # Always notify for a transition where last active gets bumped.
-        notify_reason_counter.inc("last_active_change_not_online")
+        notify_reason_counter.labels("last_active_change_not_online").inc()
         return True
 
     return False
@@ -1066,14 +1032,14 @@ class PresenceEventSource(object):
             if changed is not None and len(changed) < 500:
                 # For small deltas, its quicker to get all changes and then
                 # work out if we share a room or they're in our presence list
-                get_updates_counter.inc("stream")
+                get_updates_counter.labels("stream").inc()
                 for other_user_id in changed:
                     if other_user_id in users_interested_in:
                         user_ids_changed.add(other_user_id)
             else:
                 # Too many possible updates. Find all users we can see and check
                 # if any of them have changed.
-                get_updates_counter.inc("full")
+                get_updates_counter.labels("full").inc()
 
                 if from_key:
                     user_ids_changed = stream_change_cache.get_entities_changed(
@@ -1085,10 +1051,10 @@ class PresenceEventSource(object):
             updates = yield presence.current_state_for_users(user_ids_changed)
 
         if include_offline:
-            defer.returnValue((updates.values(), max_token))
+            defer.returnValue((list(updates.values()), max_token))
         else:
             defer.returnValue(([
-                s for s in updates.itervalues()
+                s for s in itervalues(updates)
                 if s.state != PresenceState.OFFLINE
             ], max_token))
 
@@ -1146,7 +1112,7 @@ def handle_timeouts(user_states, is_mine_fn, syncing_user_ids, now):
         if new_state:
             changes[state.user_id] = new_state
 
-    return changes.values()
+    return list(changes.values())
 
 
 def handle_timeout(state, is_mine, syncing_user_ids, now):
@@ -1345,11 +1311,11 @@ def get_interested_remotes(store, states, state_handler):
     # hosts in those rooms.
     room_ids_to_states, users_to_states = yield get_interested_parties(store, states)
 
-    for room_id, states in room_ids_to_states.iteritems():
+    for room_id, states in iteritems(room_ids_to_states):
         hosts = yield state_handler.get_current_hosts_in_room(room_id)
         hosts_and_states.append((hosts, states))
 
-    for user_id, states in users_to_states.iteritems():
+    for user_id, states in iteritems(users_to_states):
         host = get_domain_from_id(user_id)
         hosts_and_states.append(([host], states))
 

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

@@ -34,6 +34,11 @@ logger = logging.getLogger(__name__)
 class RegistrationHandler(BaseHandler):
 
     def __init__(self, hs):
+        """
+
+        Args:
+            hs (synapse.server.HomeServer):
+        """
         super(RegistrationHandler, self).__init__(hs)
 
         self.auth = hs.get_auth()
@@ -49,6 +54,7 @@ class RegistrationHandler(BaseHandler):
         self._generate_user_id_linearizer = Linearizer(
             name="_generate_user_id_linearizer",
         )
+        self._server_notices_mxid = hs.config.server_notices_mxid
 
     @defer.inlineCallbacks
     def check_username(self, localpart, guest_access_token=None,
@@ -360,6 +366,14 @@ class RegistrationHandler(BaseHandler):
             yield identity_handler.bind_threepid(c, user_id)
 
     def check_user_id_not_appservice_exclusive(self, user_id, allowed_appservice=None):
+        # don't allow people to register the server notices mxid
+        if self._server_notices_mxid is not None:
+            if user_id == self._server_notices_mxid:
+                raise SynapseError(
+                    400, "This user ID is reserved.",
+                    errcode=Codes.EXCLUSIVE
+                )
+
         # valid user IDs must not clash with any user ID namespaces claimed by
         # application services.
         services = self.store.get_app_services()

synapse/handlers/room.py

@@ -70,14 +70,27 @@ class RoomCreationHandler(BaseHandler):
         self.event_creation_handler = hs.get_event_creation_handler()
 
     @defer.inlineCallbacks
-    def create_room(self, requester, config, ratelimit=True):
+    def create_room(self, requester, config, ratelimit=True,
+                    creator_join_profile=None):
         """ Creates a new room.
 
         Args:
-            requester (Requester): The user who requested the room creation.
+            requester (synapse.types.Requester):
+                The user who requested the room creation.
             config (dict) : A dict of configuration options.
+            ratelimit (bool): set to False to disable the rate limiter
+
+            creator_join_profile (dict|None):
+                Set to override the displayname and avatar for the creating
+                user in this room. If unset, displayname and avatar will be
+                derived from the user's profile. If set, should contain the
+                values to go in the body of the 'join' event (typically
+                `avatar_url` and/or `displayname`.
+
         Returns:
-            The new room ID.
+            Deferred[dict]:
+                a dict containing the keys `room_id` and, if an alias was
+                requested, `room_alias`.
         Raises:
             SynapseError if the room ID couldn't be stored, or something went
             horribly wrong.
@@ -115,6 +128,10 @@ class RoomCreationHandler(BaseHandler):
             except Exception:
                 raise SynapseError(400, "Invalid user_id: %s" % (i,))
 
+        yield self.event_creation_handler.assert_accepted_privacy_policy(
+            requester,
+        )
+
         invite_3pid_list = config.get("invite_3pid", [])
 
         visibility = config.get("visibility", None)
@@ -178,7 +195,8 @@ class RoomCreationHandler(BaseHandler):
             initial_state=initial_state,
             creation_content=creation_content,
             room_alias=room_alias,
-            power_level_content_override=config.get("power_level_content_override", {})
+            power_level_content_override=config.get("power_level_content_override", {}),
+            creator_join_profile=creator_join_profile,
         )
 
         if "name" in config:
@@ -258,6 +276,7 @@ class RoomCreationHandler(BaseHandler):
             creation_content,
             room_alias,
             power_level_content_override,
+            creator_join_profile,
     ):
         def create(etype, content, **kwargs):
             e = {
@@ -301,6 +320,7 @@ class RoomCreationHandler(BaseHandler):
             room_id,
             "join",
             ratelimit=False,
+            content=creator_join_profile,
         )
 
         # We treat the power levels override specially as this needs to be one
@@ -446,7 +466,7 @@ class RoomContextHandler(BaseHandler):
         state = yield self.store.get_state_for_events(
             [last_event_id], None
         )
-        results["state"] = state[last_event_id].values()
+        results["state"] = list(state[last_event_id].values())
 
         results["start"] = now_token.copy_and_replace(
             "room_key", results["start"]

synapse/handlers/room_list.py

@@ -15,6 +15,9 @@
 
 from twisted.internet import defer
 
+from six import iteritems
+from six.moves import range
+
 from ._base import BaseHandler
 
 from synapse.api.constants import (
@@ -200,7 +203,7 @@ 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(
@@ -305,7 +308,7 @@ class RoomListHandler(BaseHandler):
         )
 
         event_map = yield self.store.get_events([
-            event_id for key, event_id in current_state_ids.iteritems()
+            event_id for key, event_id in iteritems(current_state_ids)
             if key[0] in (
                 EventTypes.JoinRules,
                 EventTypes.Name,

synapse/handlers/room_member.py

@@ -17,11 +17,14 @@
 import abc
 import logging
 
+from six.moves import http_client
+
 from signedjson.key import decode_verify_key_bytes
 from signedjson.sign import verify_signed_json
 from twisted.internet import defer
 from unpaddedbase64 import decode_base64
 
+import synapse.server
 import synapse.types
 from synapse.api.constants import (
     EventTypes, Membership,
@@ -46,6 +49,11 @@ class RoomMemberHandler(object):
     __metaclass__ = abc.ABCMeta
 
     def __init__(self, hs):
+        """
+
+        Args:
+            hs (synapse.server.HomeServer):
+        """
         self.hs = hs
         self.store = hs.get_datastore()
         self.auth = hs.get_auth()
@@ -63,6 +71,7 @@ class RoomMemberHandler(object):
 
         self.clock = hs.get_clock()
         self.spam_checker = hs.get_spam_checker()
+        self._server_notices_mxid = self.config.server_notices_mxid
 
     @abc.abstractmethod
     def _remote_join(self, requester, remote_room_hosts, room_id, user, content):
@@ -290,11 +299,26 @@ class RoomMemberHandler(object):
             if is_blocked:
                 raise SynapseError(403, "This room has been blocked on this server")
 
-        if effective_membership_state == "invite":
+        if effective_membership_state == Membership.INVITE:
+            # block any attempts to invite the server notices mxid
+            if target.to_string() == self._server_notices_mxid:
+                raise SynapseError(
+                    http_client.FORBIDDEN,
+                    "Cannot invite this user",
+                )
+
             block_invite = False
+
+            if (self._server_notices_mxid is not None and
+                    requester.user.to_string() == self._server_notices_mxid):
+                # allow the server notices mxid to send invites
+                is_requester_admin = True
+
+            else:
                 is_requester_admin = yield self.auth.is_server_admin(
                     requester.user,
                 )
+
             if not is_requester_admin:
                 if self.config.block_non_admin_invites:
                     logger.info(
@@ -349,6 +373,20 @@ class RoomMemberHandler(object):
                 if same_sender and same_membership and same_content:
                     defer.returnValue(old_state)
 
+            # we don't allow people to reject invites to the server notice
+            # room, but they can leave it once they are joined.
+            if (
+                old_membership == Membership.INVITE and
+                effective_membership_state == Membership.LEAVE
+            ):
+                is_blocked = yield self._is_server_notice_room(room_id)
+                if is_blocked:
+                    raise SynapseError(
+                        http_client.FORBIDDEN,
+                        "You cannot reject this invite",
+                        errcode=Codes.CANNOT_LEAVE_SERVER_NOTICE_ROOM,
+                    )
+
         is_host_in_room = yield self._is_host_in_room(current_state_ids)
 
         if effective_membership_state == Membership.JOIN:
@@ -844,6 +882,13 @@ class RoomMemberHandler(object):
 
         defer.returnValue(False)
 
+    @defer.inlineCallbacks
+    def _is_server_notice_room(self, room_id):
+        if self._server_notices_mxid is None:
+            defer.returnValue(False)
+        user_ids = yield self.store.get_users_in_room(room_id)
+        defer.returnValue(self._server_notices_mxid in user_ids)
+
 
 class RoomMemberMasterHandler(RoomMemberHandler):
     def __init__(self, hs):

synapse/handlers/search.py

@@ -348,7 +348,7 @@ class SearchHandler(BaseHandler):
             rooms = set(e.room_id for e in allowed_events)
             for room_id in rooms:
                 state = yield self.state_handler.get_current_state(room_id)
-                state_results[room_id] = state.values()
+                state_results[room_id] = list(state.values())
 
             state_results.values()
 

synapse/handlers/sync.py

@@ -28,6 +28,8 @@ import collections
 import logging
 import itertools
 
+from six import itervalues, iteritems
+
 logger = logging.getLogger(__name__)
 
 
@@ -275,7 +277,7 @@ class SyncHandler(object):
                 # result returned by the event source is poor form (it might cache
                 # the object)
                 room_id = event["room_id"]
-                event_copy = {k: v for (k, v) in event.iteritems()
+                event_copy = {k: v for (k, v) in iteritems(event)
                               if k != "room_id"}
                 ephemeral_by_room.setdefault(room_id, []).append(event_copy)
 
@@ -294,7 +296,7 @@ class SyncHandler(object):
             for event in receipts:
                 room_id = event["room_id"]
                 # exclude room id, as above
-                event_copy = {k: v for (k, v) in event.iteritems()
+                event_copy = {k: v for (k, v) in iteritems(event)
                               if k != "room_id"}
                 ephemeral_by_room.setdefault(room_id, []).append(event_copy)
 
@@ -325,7 +327,7 @@ class SyncHandler(object):
                 current_state_ids = frozenset()
                 if any(e.is_state() for e in recents):
                     current_state_ids = yield self.state.get_current_state_ids(room_id)
-                    current_state_ids = frozenset(current_state_ids.itervalues())
+                    current_state_ids = frozenset(itervalues(current_state_ids))
 
                 recents = yield filter_events_for_client(
                     self.store,
@@ -354,12 +356,24 @@ class SyncHandler(object):
                 since_key = since_token.room_key
 
             while limited and len(recents) < timeline_limit and max_repeat:
+                # If we have a since_key then we are trying to get any events
+                # that have happened since `since_key` up to `end_key`, so we
+                # can just use `get_room_events_stream_for_room`.
+                # Otherwise, we want to return the last N events in the room
+                # in toplogical ordering.
+                if since_key:
                     events, end_key = yield self.store.get_room_events_stream_for_room(
                         room_id,
                         limit=load_limit + 1,
                         from_key=since_key,
                         to_key=end_key,
                     )
+                else:
+                    events, end_key = yield self.store.get_recent_events_for_room(
+                        room_id,
+                        limit=load_limit + 1,
+                        end_token=end_key,
+                    )
                 loaded_recents = sync_config.filter_collection.filter_room_timeline(
                     events
                 )
@@ -370,7 +384,7 @@ class SyncHandler(object):
                 current_state_ids = frozenset()
                 if any(e.is_state() for e in loaded_recents):
                     current_state_ids = yield self.state.get_current_state_ids(room_id)
-                    current_state_ids = frozenset(current_state_ids.itervalues())
+                    current_state_ids = frozenset(itervalues(current_state_ids))
 
                 loaded_recents = yield filter_events_for_client(
                     self.store,
@@ -429,7 +443,11 @@ class SyncHandler(object):
         Returns:
             A Deferred map from ((type, state_key)->Event)
         """
-        last_events, token = yield self.store.get_recent_events_for_room(
+        # FIXME this claims to get the state at a stream position, but
+        # get_recent_events_for_room operates by topo ordering. This therefore
+        # does not reliably give you the state at the given stream position.
+        # (https://github.com/matrix-org/synapse/issues/3305)
+        last_events, _ = yield self.store.get_recent_events_for_room(
             room_id, end_token=stream_position.room_key, limit=1,
         )
 
@@ -523,11 +541,11 @@ class SyncHandler(object):
 
         state = {}
         if state_ids:
-            state = yield self.store.get_events(state_ids.values())
+            state = yield self.store.get_events(list(state_ids.values()))
 
         defer.returnValue({
             (e.type, e.state_key): e
-            for e in sync_config.filter_collection.filter_room_state(state.values())
+            for e in sync_config.filter_collection.filter_room_state(list(state.values()))
         })
 
     @defer.inlineCallbacks
@@ -876,7 +894,7 @@ class SyncHandler(object):
             presence.extend(states)
 
             # Deduplicate the presence entries so that there's at most one per user
-            presence = {p.user_id: p for p in presence}.values()
+            presence = list({p.user_id: p for p in presence}.values())
 
         presence = sync_config.filter_collection.filter_presence(
             presence
@@ -972,7 +990,7 @@ class SyncHandler(object):
         if since_token:
             for joined_sync in sync_result_builder.joined:
                 it = itertools.chain(
-                    joined_sync.timeline.events, joined_sync.state.itervalues()
+                    joined_sync.timeline.events, itervalues(joined_sync.state)
                 )
                 for event in it:
                     if event.type == EventTypes.Member:
@@ -1028,7 +1046,13 @@ class SyncHandler(object):
 
         Returns:
             Deferred(tuple): Returns a tuple of the form:
-            `([RoomSyncResultBuilder], [InvitedSyncResult], newly_joined_rooms)`
+            `(room_entries, invited_rooms, newly_joined_rooms, newly_left_rooms)`
+
+            where:
+                room_entries is a list [RoomSyncResultBuilder]
+                invited_rooms is a list [InvitedSyncResult]
+                newly_joined rooms is a list[str] of room ids
+                newly_left_rooms is a list[str] of room ids
         """
         user_id = sync_result_builder.sync_config.user.to_string()
         since_token = sync_result_builder.since_token
@@ -1050,7 +1074,7 @@ class SyncHandler(object):
         newly_left_rooms = []
         room_entries = []
         invited = []
-        for room_id, events in mem_change_events_by_room_id.iteritems():
+        for room_id, events in iteritems(mem_change_events_by_room_id):
             non_joins = [e for e in events if e.membership != Membership.JOIN]
             has_join = len(non_joins) != len(events)
 

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/handlers/user_directory.py

@@ -22,6 +22,7 @@ from synapse.util.metrics import Measure
 from synapse.util.async import sleep
 from synapse.types import get_localpart_from_id
 
+from six import iteritems
 
 logger = logging.getLogger(__name__)
 
@@ -122,6 +123,13 @@ class UserDirectoryHandler(object):
             user_id, profile.display_name, profile.avatar_url, None,
         )
 
+    @defer.inlineCallbacks
+    def handle_user_deactivated(self, user_id):
+        """Called when a user ID is deactivated
+        """
+        yield self.store.remove_from_user_dir(user_id)
+        yield self.store.remove_from_user_in_public_room(user_id)
+
     @defer.inlineCallbacks
     def _unsafe_process(self):
         # If self.pos is None then means we haven't fetched it from DB
@@ -403,7 +411,7 @@ class UserDirectoryHandler(object):
 
         if change:
             users_with_profile = yield self.state.get_current_user_in_room(room_id)
-            for user_id, profile in users_with_profile.iteritems():
+            for user_id, profile in iteritems(users_with_profile):
                 yield self._handle_new_user(room_id, user_id, profile)
         else:
             users = yield self.store.get_users_in_public_due_to_room(room_id)

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/additional_resource.py

@@ -13,7 +13,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from synapse.http.server import wrap_request_handler
+from synapse.http.server import wrap_json_request_handler
 from twisted.web.resource import Resource
 from twisted.web.server import NOT_DONE_YET
 
@@ -42,14 +42,13 @@ class AdditionalResource(Resource):
         Resource.__init__(self)
         self._handler = handler
 
-        # these are required by the request_handler wrapper
+        # required by the request_handler wrapper
-        self.version_string = hs.version_string
         self.clock = hs.get_clock()
 
     def render(self, request):
         self._async_render(request)
         return NOT_DONE_YET
 
-    @wrap_request_handler
+    @wrap_json_request_handler
     def _async_render(self, request):
         return self._handler(request)

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.
@@ -18,10 +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
 
 from canonicaljson import encode_canonical_json
@@ -38,8 +39,9 @@ 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
 
+from prometheus_client import Counter
 import simplejson as json
 import logging
 import urllib
@@ -47,16 +49,9 @@ import urllib
 
 logger = logging.getLogger(__name__)
 
-metrics = synapse.metrics.get_metrics_for(__name__)
+outgoing_requests_counter = Counter("synapse_http_client_requests", "", ["method"])
-
+incoming_responses_counter = Counter("synapse_http_client_responses", "",
-outgoing_requests_counter = metrics.register_counter(
+                                     ["method", "code"])
-    "requests",
-    labels=["method"],
-)
-incoming_responses_counter = metrics.register_counter(
-    "responses",
-    labels=["method", "code"],
-)
 
 
 class SimpleHttpClient(object):
@@ -93,32 +88,28 @@ class SimpleHttpClient(object):
     def request(self, method, uri, *args, **kwargs):
         # A small wrapper around self.agent.request() so we can easily attach
         # counters to it
-        outgoing_requests_counter.inc(method)
+        outgoing_requests_counter.labels(method).inc()
-
-        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)
+            incoming_responses_counter.labels(method, response.code).inc()
             logger.info(
                 "Received response to  %s %s: %s",
                 method, uri, response.code
             )
             defer.returnValue(response)
         except Exception as e:
-            incoming_responses_counter.inc(method, "ERR")
+            incoming_responses_counter.labels(method, "ERR").inc()
             logger.info(
                 "Error sending request to  %s %s: %s %s",
                 method, uri, type(e).__name__, e.message
@@ -509,7 +500,7 @@ 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

@@ -115,9 +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):
@@ -286,7 +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:

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
 
@@ -38,22 +41,19 @@ import logging
 import random
 import sys
 import urllib
-import urlparse
+from six.moves.urllib import parse as urlparse
+from six import string_types
 
 
+from prometheus_client import Counter
+
 logger = logging.getLogger(__name__)
 outbound_logger = logging.getLogger("synapse.http.outbound")
 
-metrics = synapse.metrics.get_metrics_for(__name__)
+outgoing_requests_counter = Counter("synapse_http_matrixfederationclient_requests",
-
+                                    "", ["method"])
-outgoing_requests_counter = metrics.register_counter(
+incoming_responses_counter = Counter("synapse_http_matrixfederationclient_responses",
-    "requests",
+                                     "", ["method", "code"])
-    labels=["method"],
-)
-incoming_responses_counter = metrics.register_counter(
-    "responses",
-    labels=["method", "code"],
-)
 
 
 MAX_LONG_RETRIES = 10
@@ -184,21 +184,20 @@ 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
@@ -552,7 +551,7 @@ class MatrixFederationHttpClient(object):
 
         encoded_args = {}
         for k, vs in args.items():
-            if isinstance(vs, basestring):
+            if isinstance(vs, string_types):
                 vs = [vs]
             encoded_args[k] = [v.encode("UTF-8") for v in vs]
 
@@ -667,7 +666,7 @@ def check_content_type_is_json(headers):
         RuntimeError if the
 
     """
-    c_type = headers.getRawHeaders("Content-Type")
+    c_type = headers.getRawHeaders(b"Content-Type")
     if c_type is None:
         raise RuntimeError(
             "No Content-Type header"
@@ -684,7 +683,7 @@ def check_content_type_is_json(headers):
 def encode_query_args(args):
     encoded_args = {}
     for k, vs in args.items():
-        if isinstance(vs, basestring):
+        if isinstance(vs, string_types):
             vs = [vs]
         encoded_args[k] = [v.encode("UTF-8") for v in vs]
 

synapse/http/request_metrics.py

@@ -0,0 +1,273 @@
+# -*- 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.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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
+
+from prometheus_client.core import Counter, Histogram
+from synapse.metrics import LaterGauge
+
+from synapse.util.logcontext import LoggingContext
+
+logger = logging.getLogger(__name__)
+
+
+# total number of responses served, split by method/servlet/tag
+response_count = Counter(
+    "synapse_http_server_response_count", "", ["method", "servlet", "tag"]
+)
+
+requests_counter = Counter(
+    "synapse_http_server_requests_received", "", ["method", "servlet"]
+)
+
+outgoing_responses_counter = Counter(
+    "synapse_http_server_responses", "", ["method", "code"]
+)
+
+response_timer = Histogram(
+    "synapse_http_server_response_time_seconds", "sec", ["method", "servlet", "tag"]
+)
+
+response_ru_utime = Counter(
+    "synapse_http_server_response_ru_utime_seconds", "sec", ["method", "servlet", "tag"]
+)
+
+response_ru_stime = Counter(
+    "synapse_http_server_response_ru_stime_seconds", "sec", ["method", "servlet", "tag"]
+)
+
+response_db_txn_count = Counter(
+    "synapse_http_server_response_db_txn_count", "", ["method", "servlet", "tag"]
+)
+
+# seconds spent waiting for db txns, excluding scheduling time, when processing
+# this request
+response_db_txn_duration = Counter(
+    "synapse_http_server_response_db_txn_duration_seconds",
+    "",
+    ["method", "servlet", "tag"],
+)
+
+# seconds spent waiting for a db connection, when processing this request
+response_db_sched_duration = Counter(
+    "synapse_http_server_response_db_sched_duration_seconds",
+    "",
+    ["method", "servlet", "tag"],
+)
+
+# size in bytes of the response written
+response_size = Counter(
+    "synapse_http_server_response_size", "", ["method", "servlet", "tag"]
+)
+
+# In flight metrics are incremented while the requests are in flight, rather
+# than when the response was written.
+
+in_flight_requests_ru_utime = Counter(
+    "synapse_http_server_in_flight_requests_ru_utime_seconds",
+    "",
+    ["method", "servlet"],
+)
+
+in_flight_requests_ru_stime = Counter(
+    "synapse_http_server_in_flight_requests_ru_stime_seconds",
+    "",
+    ["method", "servlet"],
+)
+
+in_flight_requests_db_txn_count = Counter(
+    "synapse_http_server_in_flight_requests_db_txn_count", "", ["method", "servlet"]
+)
+
+# seconds spent waiting for db txns, excluding scheduling time, when processing
+# this request
+in_flight_requests_db_txn_duration = Counter(
+    "synapse_http_server_in_flight_requests_db_txn_duration_seconds",
+    "",
+    ["method", "servlet"],
+)
+
+# seconds spent waiting for a db connection, when processing this request
+in_flight_requests_db_sched_duration = Counter(
+    "synapse_http_server_in_flight_requests_db_sched_duration_seconds",
+    "",
+    ["method", "servlet"],
+)
+
+# The set of all in flight requests, set[RequestMetrics]
+_in_flight_requests = set()
+
+
+def _get_in_flight_counts():
+    """Returns a count of all in flight requests by (method, server_name)
+
+    Returns:
+        dict[tuple[str, str], int]
+    """
+    for rm in _in_flight_requests:
+        rm.update_metrics()
+
+    # Map from (method, name) -> int, the number of in flight requests of that
+    # type
+    counts = {}
+    for rm in _in_flight_requests:
+        key = (rm.method, rm.name,)
+        counts[key] = counts.get(key, 0) + 1
+
+    return counts
+
+
+LaterGauge(
+    "synapse_http_request_metrics_in_flight_requests_count",
+    "",
+    ["method", "servlet"],
+    _get_in_flight_counts,
+)
+
+
+class RequestMetrics(object):
+    def start(self, time_sec, name, method):
+        self.start = time_sec
+        self.start_context = LoggingContext.current_context()
+        self.name = name
+        self.method = method
+
+        self._request_stats = _RequestStats.from_context(self.start_context)
+
+        _in_flight_requests.add(self)
+
+    def stop(self, time_sec, request):
+        _in_flight_requests.discard(self)
+
+        context = LoggingContext.current_context()
+
+        tag = ""
+        if context:
+            tag = context.tag
+
+            if context != self.start_context:
+                logger.warn(
+                    "Context have unexpectedly changed %r, %r",
+                    context, self.start_context
+                )
+                return
+
+        outgoing_responses_counter.labels(request.method, str(request.code)).inc()
+
+        response_count.labels(request.method, self.name, tag).inc()
+
+        response_timer.labels(request.method, self.name, tag).observe(
+            time_sec - self.start
+        )
+
+        ru_utime, ru_stime = context.get_resource_usage()
+
+        response_ru_utime.labels(request.method, self.name, tag).inc(ru_utime)
+        response_ru_stime.labels(request.method, self.name, tag).inc(ru_stime)
+        response_db_txn_count.labels(request.method, self.name, tag).inc(
+            context.db_txn_count
+        )
+        response_db_txn_duration.labels(request.method, self.name, tag).inc(
+            context.db_txn_duration_sec
+        )
+        response_db_sched_duration.labels(request.method, self.name, tag).inc(
+            context.db_sched_duration_sec
+        )
+
+        response_size.labels(request.method, self.name, tag).inc(request.sentLength)
+
+        # We always call this at the end to ensure that we update the metrics
+        # regardless of whether a call to /metrics while the request was in
+        # flight.
+        self.update_metrics()
+
+    def update_metrics(self):
+        """Updates the in flight metrics with values from this request.
+        """
+        diff = self._request_stats.update(self.start_context)
+
+        in_flight_requests_ru_utime.labels(self.method, self.name).inc(diff.ru_utime)
+        in_flight_requests_ru_stime.labels(self.method, self.name).inc(diff.ru_stime)
+
+        in_flight_requests_db_txn_count.labels(self.method, self.name).inc(
+            diff.db_txn_count
+        )
+
+        in_flight_requests_db_txn_duration.labels(self.method, self.name).inc(
+            diff.db_txn_duration_sec
+        )
+
+        in_flight_requests_db_sched_duration.labels(self.method, self.name).inc(
+            diff.db_sched_duration_sec
+        )
+
+
+class _RequestStats(object):
+    """Keeps tracks of various metrics for an in flight request.
+    """
+
+    __slots__ = [
+        "ru_utime",
+        "ru_stime",
+        "db_txn_count",
+        "db_txn_duration_sec",
+        "db_sched_duration_sec",
+    ]
+
+    def __init__(
+        self, ru_utime, ru_stime, db_txn_count, db_txn_duration_sec, db_sched_duration_sec
+    ):
+        self.ru_utime = ru_utime
+        self.ru_stime = ru_stime
+        self.db_txn_count = db_txn_count
+        self.db_txn_duration_sec = db_txn_duration_sec
+        self.db_sched_duration_sec = db_sched_duration_sec
+
+    @staticmethod
+    def from_context(context):
+        ru_utime, ru_stime = context.get_resource_usage()
+
+        return _RequestStats(
+            ru_utime, ru_stime,
+            context.db_txn_count,
+            context.db_txn_duration_sec,
+            context.db_sched_duration_sec,
+        )
+
+    def update(self, context):
+        """Updates the current values and returns the difference between the
+        old and new values.
+
+        Returns:
+            _RequestStats: The difference between the old and new values
+        """
+        new = _RequestStats.from_context(context)
+
+        diff = _RequestStats(
+            new.ru_utime - self.ru_utime,
+            new.ru_stime - self.ru_stime,
+            new.db_txn_count - self.db_txn_count,
+            new.db_txn_duration_sec - self.db_txn_duration_sec,
+            new.db_sched_duration_sec - self.db_sched_duration_sec,
+        )
+
+        self.ru_utime = new.ru_utime
+        self.ru_stime = new.ru_stime
+        self.db_txn_count = new.db_txn_count
+        self.db_txn_duration_sec = new.db_txn_duration_sec
+        self.db_sched_duration_sec = new.db_sched_duration_sec
+
+        return diff

synapse/http/server.py

@@ -13,11 +13,15 @@
 # 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 cgi
+from six.moves import http_client
 
 from synapse.api.errors import (
     cs_exception, SynapseError, CodeMessageException, UnrecognizedRequestError, Codes
 )
+from synapse.http.request_metrics import (
+    requests_counter,
+)
 from synapse.util.logcontext import LoggingContext, PreserveLoggingContext
 from synapse.util.caches import intern_dict
 from synapse.util.metrics import Measure
@@ -41,133 +45,38 @@ import simplejson
 
 logger = logging.getLogger(__name__)
 
-metrics = synapse.metrics.get_metrics_for(__name__)
+HTML_ERROR_TEMPLATE = """<!DOCTYPE html>
-
+<html lang=en>
-# total number of responses served, split by method/servlet/tag
+  <head>
-response_count = metrics.register_counter(
+    <meta charset="utf-8">
-    "response_count",
+    <title>Error {code}</title>
-    labels=["method", "servlet", "tag"],
+  </head>
-    alternative_names=(
+  <body>
-        # the following are all deprecated aliases for the same metric
+     <p>{msg}</p>
-        metrics.name_prefix + x for x in (
+  </body>
-            "_requests",
+</html>
-            "_response_time:count",
+"""
-            "_response_ru_utime:count",
-            "_response_ru_stime:count",
-            "_response_db_txn_count:count",
-            "_response_db_txn_duration:count",
-        )
-    )
-)
-
-requests_counter = metrics.register_counter(
-    "requests_received",
-    labels=["method", "servlet", ],
-)
-
-outgoing_responses_counter = metrics.register_counter(
-    "responses",
-    labels=["method", "code"],
-)
-
-response_timer = metrics.register_counter(
-    "response_time_seconds",
-    labels=["method", "servlet", "tag"],
-    alternative_names=(
-        metrics.name_prefix + "_response_time:total",
-    ),
-)
-
-response_ru_utime = metrics.register_counter(
-    "response_ru_utime_seconds", labels=["method", "servlet", "tag"],
-    alternative_names=(
-        metrics.name_prefix + "_response_ru_utime:total",
-    ),
-)
-
-response_ru_stime = metrics.register_counter(
-    "response_ru_stime_seconds", labels=["method", "servlet", "tag"],
-    alternative_names=(
-        metrics.name_prefix + "_response_ru_stime:total",
-    ),
-)
-
-response_db_txn_count = metrics.register_counter(
-    "response_db_txn_count", labels=["method", "servlet", "tag"],
-    alternative_names=(
-        metrics.name_prefix + "_response_db_txn_count:total",
-    ),
-)
-
-# seconds spent waiting for db txns, excluding scheduling time, when processing
-# this request
-response_db_txn_duration = metrics.register_counter(
-    "response_db_txn_duration_seconds", labels=["method", "servlet", "tag"],
-    alternative_names=(
-        metrics.name_prefix + "_response_db_txn_duration:total",
-    ),
-)
-
-# seconds spent waiting for a db connection, when processing this request
-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
 
 
-def request_handler(include_metrics=False):
+def wrap_json_request_handler(h):
-    """Decorator for ``wrap_request_handler``"""
+    """Wraps a request handler method with exception handling.
-    return lambda request_handler: wrap_request_handler(request_handler, include_metrics)
 
+    Also adds logging as per wrap_request_handler_with_logging.
 
-def wrap_request_handler(request_handler, include_metrics=False):
+    The handler method must have a signature of "handle_foo(self, request)",
-    """Wraps a method that acts as a request handler with the necessary logging
+    where "self" must have a "clock" attribute (and "request" must be a
-    and exception handling.
+    SynapseRequest).
 
-    The method must have a signature of "handle_foo(self, request)". The
+    The handler must return a deferred. If the deferred succeeds we assume that
-    argument "self" must have "version_string" and "clock" attributes. The
-    argument "request" must be a twisted HTTP request.
-
-    The method must return a deferred. If the deferred succeeds we assume that
     a response has been sent. If the deferred fails with a SynapseError we use
     it to send a JSON response with the appropriate HTTP reponse code. If the
     deferred fails with any other type of error we send a 500 reponse.
-
-    We insert a unique request-id into the logging context for this request and
-    log the response and duration for this request.
     """
 
     @defer.inlineCallbacks
     def wrapped_request_handler(self, request):
-        global _next_request_id
-        request_id = "%s-%s" % (request.method, _next_request_id)
-        _next_request_id += 1
-
-        with LoggingContext(request_id) as request_context:
-            with Measure(self.clock, "wrapped_request_handler"):
-                request_metrics = RequestMetrics()
-                # we start the request metrics timer here with an initial stab
-                # at the servlet name. For most requests that name will be
-                # JsonResource (or a subclass), and JsonResource._async_render
-                # will update it once it picks a servlet.
-                servlet_name = self.__class__.__name__
-                request_metrics.start(self.clock, name=servlet_name)
-
-                request_context.request = request_id
-                with request.processing():
         try:
-                        with PreserveLoggingContext(request_context):
+            yield h(self, request)
-                            if include_metrics:
-                                yield request_handler(self, request, request_metrics)
-                            else:
-                                requests_counter.inc(request.method, servlet_name)
-                                yield request_handler(self, request)
         except CodeMessageException as e:
             code = e.code
             if isinstance(e, SynapseError):
@@ -176,22 +85,19 @@ def wrap_request_handler(request_handler, include_metrics=False):
                 )
             else:
                 logger.exception(e)
-                        outgoing_responses_counter.inc(request.method, str(code))
             respond_with_json(
                 request, code, cs_exception(e), send_cors=True,
                 pretty_print=_request_user_agent_is_curl(request),
-                            version_string=self.version_string,
             )
+
         except Exception:
             # failure.Failure() fishes the original Failure out
             # of our stack, and thus gives us a sensible stack
             # trace.
             f = failure.Failure()
             logger.error(
-                            "Failed handle request %s.%s on %r: %r: %s",
+                "Failed handle request via %r: %r: %s",
-                            request_handler.__module__,
+                h,
-                            request_handler.__name__,
-                            self,
                 request,
                 f.getTraceback().rstrip(),
             )
@@ -204,15 +110,109 @@ def wrap_request_handler(request_handler, include_metrics=False):
                 },
                 send_cors=True,
                 pretty_print=_request_user_agent_is_curl(request),
-                            version_string=self.version_string,
             )
-                    finally:
+
-                        try:
+    return wrap_request_handler_with_logging(wrapped_request_handler)
-                            request_metrics.stop(
+
-                                self.clock, request
+
+def wrap_html_request_handler(h):
+    """Wraps a request handler method with exception handling.
+
+    Also adds logging as per wrap_request_handler_with_logging.
+
+    The handler method must have a signature of "handle_foo(self, request)",
+    where "self" must have a "clock" attribute (and "request" must be a
+    SynapseRequest).
+    """
+    def wrapped_request_handler(self, request):
+        d = defer.maybeDeferred(h, self, request)
+        d.addErrback(_return_html_error, request)
+        return d
+
+    return wrap_request_handler_with_logging(wrapped_request_handler)
+
+
+def _return_html_error(f, request):
+    """Sends an HTML error page corresponding to the given failure
+
+    Args:
+        f (twisted.python.failure.Failure):
+        request (twisted.web.iweb.IRequest):
+    """
+    if f.check(CodeMessageException):
+        cme = f.value
+        code = cme.code
+        msg = cme.msg
+
+        if isinstance(cme, SynapseError):
+            logger.info(
+                "%s SynapseError: %s - %s", request, code, msg
             )
-                        except Exception as e:
+        else:
-                            logger.warn("Failed to stop metrics: %r", e)
+            logger.error(
+                "Failed handle request %r: %s",
+                request,
+                f.getTraceback().rstrip(),
+            )
+    else:
+        code = http_client.INTERNAL_SERVER_ERROR
+        msg = "Internal server error"
+
+        logger.error(
+            "Failed handle request %r: %s",
+            request,
+            f.getTraceback().rstrip(),
+        )
+
+    body = HTML_ERROR_TEMPLATE.format(
+        code=code, msg=cgi.escape(msg),
+    ).encode("utf-8")
+    request.setResponseCode(code)
+    request.setHeader(b"Content-Type", b"text/html; charset=utf-8")
+    request.setHeader(b"Content-Length", b"%i" % (len(body),))
+    request.write(body)
+    finish_request(request)
+
+
+def wrap_request_handler_with_logging(h):
+    """Wraps a request handler to provide logging and metrics
+
+    The handler method must have a signature of "handle_foo(self, request)",
+    where "self" must have a "clock" attribute (and "request" must be a
+    SynapseRequest).
+
+    As well as calling `request.processing` (which will log the response and
+    duration for this request), the wrapped request handler will insert the
+    request id into the logging context.
+    """
+    @defer.inlineCallbacks
+    def wrapped_request_handler(self, request):
+        """
+        Args:
+            self:
+            request (synapse.http.site.SynapseRequest):
+        """
+
+        request_id = request.get_request_id()
+        with LoggingContext(request_id) as request_context:
+            request_context.request = request_id
+            with Measure(self.clock, "wrapped_request_handler"):
+                # we start the request metrics timer here with an initial stab
+                # at the servlet name. For most requests that name will be
+                # JsonResource (or a subclass), and JsonResource._async_render
+                # will update it once it picks a servlet.
+                servlet_name = self.__class__.__name__
+                with request.processing(servlet_name):
+                    with PreserveLoggingContext(request_context):
+                        d = defer.maybeDeferred(h, self, request)
+
+                        # record the arrival of the request *after*
+                        # dispatching to the handler, so that the handler
+                        # can update the servlet name in the request
+                        # metrics
+                        requests_counter.labels(request.method,
+                                                request.request_metrics.name).inc()
+                        yield d
     return wrapped_request_handler
 
 
@@ -262,7 +262,6 @@ class JsonResource(HttpServer, resource.Resource):
         self.canonical_json = canonical_json
         self.clock = hs.get_clock()
         self.path_regexs = {}
-        self.version_string = hs.version_string
         self.hs = hs
 
     def register_paths(self, method, path_patterns, callback):
@@ -278,13 +277,9 @@ class JsonResource(HttpServer, resource.Resource):
         self._async_render(request)
         return server.NOT_DONE_YET
 
-    # Disable metric reporting because _async_render does its own metrics.
+    @wrap_json_request_handler
-    # It does its own metric reporting because _async_render dispatches to
-    # a callback and it's the class name of that callback we want to report
-    # against rather than the JsonResource itself.
-    @request_handler(include_metrics=True)
     @defer.inlineCallbacks
-    def _async_render(self, request, request_metrics):
+    def _async_render(self, request):
         """ This gets called from render() every time someone sends us a request.
             This checks if anyone has registered a callback for that method and
             path.
@@ -296,9 +291,7 @@ class JsonResource(HttpServer, resource.Resource):
             servlet_classname = servlet_instance.__class__.__name__
         else:
             servlet_classname = "%r" % callback
-
+        request.request_metrics.name = servlet_classname
-        request_metrics.name = servlet_classname
-        requests_counter.inc(request.method, servlet_classname)
 
         # Now trigger the callback. If it returns a response, we send it
         # here. If it throws an exception, that is handled by the wrapper
@@ -345,15 +338,12 @@ class JsonResource(HttpServer, resource.Resource):
 
     def _send_response(self, request, code, response_json_object,
                        response_code_message=None):
-        outgoing_responses_counter.inc(request.method, str(code))
-
         # TODO: Only enable CORS for the requests that need it.
         respond_with_json(
             request, code, response_json_object,
             send_cors=True,
             response_code_message=response_code_message,
             pretty_print=_request_user_agent_is_curl(request),
-            version_string=self.version_string,
             canonical_json=self.canonical_json,
         )
 
@@ -386,54 +376,6 @@ def _unrecognised_request_handler(request):
     raise UnrecognizedRequestError()
 
 
-class RequestMetrics(object):
-    def start(self, clock, name):
-        self.start = clock.time_msec()
-        self.start_context = LoggingContext.current_context()
-        self.name = name
-
-    def stop(self, clock, request):
-        context = LoggingContext.current_context()
-
-        tag = ""
-        if context:
-            tag = context.tag
-
-            if context != self.start_context:
-                logger.warn(
-                    "Context have unexpectedly changed %r, %r",
-                    context, self.start_context
-                )
-                return
-
-        response_count.inc(request.method, self.name, tag)
-
-        response_timer.inc_by(
-            clock.time_msec() - self.start, request.method,
-            self.name, tag
-        )
-
-        ru_utime, ru_stime = context.get_resource_usage()
-
-        response_ru_utime.inc_by(
-            ru_utime, request.method, self.name, tag
-        )
-        response_ru_stime.inc_by(
-            ru_stime, request.method, self.name, tag
-        )
-        response_db_txn_count.inc_by(
-            context.db_txn_count, request.method, self.name, tag
-        )
-        response_db_txn_duration.inc_by(
-            context.db_txn_duration_ms / 1000., request.method, self.name, tag
-        )
-        response_db_sched_duration.inc_by(
-            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."""
 
@@ -452,7 +394,7 @@ class RootRedirect(resource.Resource):
 
 def respond_with_json(request, code, json_object, send_cors=False,
                       response_code_message=None, pretty_print=False,
-                      version_string="", canonical_json=True):
+                      canonical_json=True):
     # could alternatively use request.notifyFinish() and flip a flag when
     # the Deferred fires, but since the flag is RIGHT THERE it seems like
     # a waste.
@@ -474,12 +416,11 @@ def respond_with_json(request, code, json_object, send_cors=False,
         request, code, json_bytes,
         send_cors=send_cors,
         response_code_message=response_code_message,
-        version_string=version_string
     )
 
 
 def respond_with_json_bytes(request, code, json_bytes, send_cors=False,
-                            version_string="", response_code_message=None):
+                            response_code_message=None):
     """Sends encoded JSON in response to the given request.
 
     Args:
@@ -493,7 +434,6 @@ def respond_with_json_bytes(request, code, json_bytes, send_cors=False,
 
     request.setResponseCode(code, message=response_code_message)
     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")
 
@@ -546,6 +486,6 @@ def _request_user_agent_is_curl(request):
         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

@@ -12,27 +12,51 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from synapse.util.logcontext import LoggingContext
-from twisted.web.server import Site, Request
-
 import contextlib
 import logging
 import re
 import time
 
+from twisted.web.server import Site, Request
+
+from synapse.http.request_metrics import RequestMetrics
+from synapse.util.logcontext import LoggingContext
+
+logger = logging.getLogger(__name__)
+
 ACCESS_TOKEN_RE = re.compile(br'(\?.*access(_|%5[Ff])token=)[^&]*(.*)$')
 
+_next_request_seq = 0
+
 
 class SynapseRequest(Request):
+    """Class which encapsulates an HTTP request to synapse.
+
+    All of the requests processed in synapse are of this type.
+
+    It extends twisted's twisted.web.server.Request, and adds:
+     * Unique request ID
+     * Redaction of access_token query-params in __repr__
+     * Logging at start and end
+     * Metrics to record CPU, wallclock and DB time by endpoint.
+
+    It provides a method `processing` which should be called by the Resource
+    which is handling the request, and returns a context manager.
+
+    """
     def __init__(self, site, *args, **kw):
         Request.__init__(self, *args, **kw)
         self.site = site
         self.authenticated_entity = None
         self.start_time = 0
 
+        global _next_request_seq
+        self.request_seq = _next_request_seq
+        _next_request_seq += 1
+
     def __repr__(self):
         # We overwrite this so that we don't log ``access_token``
-        return '<%s at 0x%x method=%s uri=%s clientproto=%s site=%s>' % (
+        return '<%s at 0x%x method=%r uri=%r clientproto=%r site=%r>' % (
             self.__class__.__name__,
             id(self),
             self.method,
@@ -41,6 +65,9 @@ class SynapseRequest(Request):
             self.site.site_tag,
         )
 
+    def get_request_id(self):
+        return "%s-%i" % (self.method, self.request_seq)
+
     def get_redacted_uri(self):
         return ACCESS_TOKEN_RE.sub(
             br'\1<redacted>\3',
@@ -50,7 +77,18 @@ class SynapseRequest(Request):
     def get_user_agent(self):
         return self.requestHeaders.getRawHeaders(b"User-Agent", [None])[-1]
 
-    def started_processing(self):
+    def render(self, resrc):
+        # override the Server header which is set by twisted
+        self.setHeader("Server", self.site.server_version_string)
+        return Request.render(self, resrc)
+
+    def _started_processing(self, servlet_name):
+        self.start_time = time.time()
+        self.request_metrics = RequestMetrics()
+        self.request_metrics.start(
+            self.start_time, name=servlet_name, method=self.method,
+        )
+
         self.site.access_logger.info(
             "%s - %s - Received request: %s %s",
             self.getClientIP(),
@@ -58,32 +96,32 @@ class SynapseRequest(Request):
             self.method,
             self.get_redacted_uri()
         )
-        self.start_time = int(time.time() * 1000)
-
-    def finished_processing(self):
 
+    def _finished_processing(self):
         try:
             context = LoggingContext.current_context()
             ru_utime, ru_stime = context.get_resource_usage()
             db_txn_count = context.db_txn_count
-            db_txn_duration_ms = context.db_txn_duration_ms
+            db_txn_duration_sec = context.db_txn_duration_sec
-            db_sched_duration_ms = context.db_sched_duration_ms
+            db_sched_duration_sec = context.db_sched_duration_sec
         except Exception:
             ru_utime, ru_stime = (0, 0)
-            db_txn_count, db_txn_duration_ms = (0, 0)
+            db_txn_count, db_txn_duration_sec = (0, 0)
+
+        end_time = time.time()
 
         self.site.access_logger.info(
             "%s - %s - {%s}"
-            " Processed request: %dms (%dms, %dms) (%dms/%dms/%d)"
+            " Processed request: %.3fsec (%.3fsec, %.3fsec) (%.3fsec/%.3fsec/%d)"
             " %sB %s \"%s %s %s\" \"%s\"",
             self.getClientIP(),
             self.site.site_tag,
             self.authenticated_entity,
-            int(time.time() * 1000) - self.start_time,
+            end_time - self.start_time,
-            int(ru_utime * 1000),
+            ru_utime,
-            int(ru_stime * 1000),
+            ru_stime,
-            db_sched_duration_ms,
+            db_sched_duration_sec,
-            db_txn_duration_ms,
+            db_txn_duration_sec,
             int(db_txn_count),
             self.sentLength,
             self.code,
@@ -93,11 +131,38 @@ class SynapseRequest(Request):
             self.get_user_agent(),
         )
 
+        try:
+            self.request_metrics.stop(end_time, self)
+        except Exception as e:
+            logger.warn("Failed to stop metrics: %r", e)
+
     @contextlib.contextmanager
-    def processing(self):
+    def processing(self, servlet_name):
-        self.started_processing()
+        """Record the fact that we are processing this request.
+
+        Returns a context manager; the correct way to use this is:
+
+        @defer.inlineCallbacks
+        def handle_request(request):
+            with request.processing("FooServlet"):
+                yield really_handle_the_request()
+
+        This will log the request's arrival. Once the context manager is
+        closed, the completion of the request will be logged, and the various
+        metrics will be updated.
+
+        Args:
+            servlet_name (str): the name of the servlet which will be
+                processing this request. This is used in the metrics.
+
+                It is possible to update this afterwards by updating
+                self.request_metrics.servlet_name.
+        """
+        # TODO: we should probably just move this into render() and finish(),
+        # to save having to call a separate method.
+        self._started_processing(servlet_name)
         yield
-        self.finished_processing()
+        self._finished_processing()
 
 
 class XForwardedForRequest(SynapseRequest):
@@ -135,7 +200,8 @@ class SynapseSite(Site):
     Subclass of a twisted http Site that does access logging with python's
     standard logging
     """
-    def __init__(self, logger_name, site_tag, config, resource, *args, **kwargs):
+    def __init__(self, logger_name, site_tag, config, resource,
+                 server_version_string, *args, **kwargs):
         Site.__init__(self, resource, *args, **kwargs)
 
         self.site_tag = site_tag
@@ -143,6 +209,7 @@ class SynapseSite(Site):
         proxied = config.get("x_forwarded", False)
         self.requestFactory = SynapseRequestFactory(self, proxied)
         self.access_logger = logging.getLogger(logger_name)
+        self.server_version_string = server_version_string
 
     def log(self, request):
         pass

synapse/metrics/__init__.py

@@ -17,165 +17,178 @@ import logging
 import functools
 import time
 import gc
+import os
 import platform
+import attr
+
+from prometheus_client import Gauge, Histogram, Counter
+from prometheus_client.core import GaugeMetricFamily, REGISTRY
 
 from twisted.internet import reactor
 
-from .metric import (
-    CounterMetric, CallbackMetric, DistributionMetric, CacheMetric,
-    MemoryUsageMetric, GaugeMetric,
-)
-from .process_collector import register_process_collector
-
 
 logger = logging.getLogger(__name__)
 
-
+running_on_pypy = platform.python_implementation() == "PyPy"
-running_on_pypy = platform.python_implementation() == 'PyPy'
 all_metrics = []
 all_collectors = []
+all_gauges = {}
+
+HAVE_PROC_SELF_STAT = os.path.exists("/proc/self/stat")
 
 
-class Metrics(object):
+class RegistryProxy(object):
-    """ A single Metrics object gives a (mutable) slice view of the all_metrics
-    dict, allowing callers to easily register new metrics that are namespaced
-    nicely."""
 
-    def __init__(self, name):
+    @staticmethod
-        self.name_prefix = name
+    def collect():
-
+        for metric in REGISTRY.collect():
-    def make_subspace(self, name):
+            if not metric.name.startswith("__"):
-        return Metrics("%s_%s" % (self.name_prefix, name))
+                yield metric
-
-    def register_collector(self, func):
-        all_collectors.append(func)
-
-    def _register(self, metric_class, name, *args, **kwargs):
-        full_name = "%s_%s" % (self.name_prefix, name)
-
-        metric = metric_class(full_name, *args, **kwargs)
-
-        all_metrics.append(metric)
-        return metric
-
-    def register_counter(self, *args, **kwargs):
-        """
-        Returns:
-            CounterMetric
-        """
-        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:
-            CallbackMetric
-        """
-        return self._register(CallbackMetric, *args, **kwargs)
-
-    def register_distribution(self, *args, **kwargs):
-        """
-        Returns:
-            DistributionMetric
-        """
-        return self._register(DistributionMetric, *args, **kwargs)
-
-    def register_cache(self, *args, **kwargs):
-        """
-        Returns:
-            CacheMetric
-        """
-        return self._register(CacheMetric, *args, **kwargs)
 
 
-def register_memory_metrics(hs):
+@attr.s(hash=True)
+class LaterGauge(object):
+
+    name = attr.ib()
+    desc = attr.ib()
+    labels = attr.ib(hash=False)
+    caller = attr.ib()
+
+    def collect(self):
+
+        g = GaugeMetricFamily(self.name, self.desc, labels=self.labels)
+
         try:
-        import psutil
+            calls = self.caller()
-        process = psutil.Process()
-        process.memory_info().rss
-    except (ImportError, AttributeError):
-        logger.warn(
-            "psutil is not installed or incorrect version."
-            " Disabling memory metrics."
-        )
-        return
-    metric = MemoryUsageMetric(hs, psutil)
-    all_metrics.append(metric)
-
-
-def get_metrics_for(pkg_name):
-    """ Returns a Metrics instance for conveniently creating metrics
-    namespaced with the given name prefix. """
-
-    # Convert a "package.name" to "package_name" because Prometheus doesn't
-    # let us use . in metric names
-    return Metrics(pkg_name.replace(".", "_"))
-
-
-def render_all():
-    strs = []
-
-    for collector in all_collectors:
-        collector()
-
-    for metric in all_metrics:
-        try:
-            strs += metric.render()
         except Exception:
-            strs += ["# FAILED to render"]
+            logger.exception(
-            logger.exception("Failed to render metric")
+                "Exception running callback for LaterGuage(%s)",
+                self.name,
+            )
+            yield g
+            return
 
-    strs.append("")  # to generate a final CRLF
+        if isinstance(calls, dict):
+            for k, v in calls.items():
+                g.add_metric(k, v)
+        else:
+            g.add_metric([], calls)
 
-    return "\n".join(strs)
+        yield g
+
+    def __attrs_post_init__(self):
+        self._register()
+
+    def _register(self):
+        if self.name in all_gauges.keys():
+            logger.warning("%s already registered, reregistering" % (self.name,))
+            REGISTRY.unregister(all_gauges.pop(self.name))
+
+        REGISTRY.register(self)
+        all_gauges[self.name] = self
 
 
-register_process_collector(get_metrics_for("process"))
+#
+# Detailed CPU metrics
+#
+
+class CPUMetrics(object):
+
+    def __init__(self):
+        ticks_per_sec = 100
+        try:
+            # Try and get the system config
+            ticks_per_sec = os.sysconf('SC_CLK_TCK')
+        except (ValueError, TypeError, AttributeError):
+            pass
+
+        self.ticks_per_sec = ticks_per_sec
+
+    def collect(self):
+        if not HAVE_PROC_SELF_STAT:
+            return
+
+        with open("/proc/self/stat") as s:
+            line = s.read()
+            raw_stats = line.split(") ", 1)[1].split(" ")
+
+            user = GaugeMetricFamily("process_cpu_user_seconds_total", "")
+            user.add_metric([], float(raw_stats[11]) / self.ticks_per_sec)
+            yield user
+
+            sys = GaugeMetricFamily("process_cpu_system_seconds_total", "")
+            sys.add_metric([], float(raw_stats[12]) / self.ticks_per_sec)
+            yield sys
 
 
-python_metrics = get_metrics_for("python")
+REGISTRY.register(CPUMetrics())
 
-gc_time = python_metrics.register_distribution("gc_time", labels=["gen"])
+#
-gc_unreachable = python_metrics.register_counter("gc_unreachable_total", labels=["gen"])
+# Python GC metrics
-python_metrics.register_callback(
+#
-    "gc_counts", lambda: {(i,): v for i, v in enumerate(gc.get_count())}, labels=["gen"]
+
+gc_unreachable = Gauge("python_gc_unreachable_total", "Unreachable GC objects", ["gen"])
+gc_time = Histogram(
+    "python_gc_time",
+    "Time taken to GC (sec)",
+    ["gen"],
+    buckets=[0.0025, 0.005, 0.01, 0.025, 0.05, 0.10, 0.25, 0.50, 1.00, 2.50,
+             5.00, 7.50, 15.00, 30.00, 45.00, 60.00],
 )
 
-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")
+class GCCounts(object):
+
+    def collect(self):
+        cm = GaugeMetricFamily("python_gc_counts", "GC cycle counts", labels=["gen"])
+        for n, m in enumerate(gc.get_count()):
+            cm.add_metric([str(n)], m)
+
+        yield cm
+
+
+REGISTRY.register(GCCounts())
+
+#
+# Twisted reactor metrics
+#
+
+tick_time = Histogram(
+    "python_twisted_reactor_tick_time",
+    "Tick time of the Twisted reactor (sec)",
+    buckets=[0.001, 0.002, 0.005, 0.01, 0.025, 0.05, 0.1, 0.2, 0.5, 1, 2, 5],
+)
+pending_calls_metric = Histogram(
+    "python_twisted_reactor_pending_calls",
+    "Pending calls",
+    buckets=[1, 2, 5, 10, 25, 50, 100, 250, 500, 1000],
+)
+
+#
+# Federation Metrics
+#
+
+sent_edus_counter = Counter("synapse_federation_client_sent_edus", "")
+
+sent_transactions_counter = Counter("synapse_federation_client_sent_transactions", "")
+
+events_processed_counter = Counter("synapse_federation_client_events_processed", "")
 
 # 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 = Gauge("synapse_event_processing_positions", "", ["name"])
-    "event_processing_positions", labels=["name"],
-)
 
 # Used to track the current max events stream position
-event_persisted_position = synapse_metrics.register_gauge(
+event_persisted_position = Gauge("synapse_event_persisted_position", "")
-    "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 = Gauge("synapse_event_processing_last_ts", "", ["name"])
-    "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 = Gauge("synapse_event_processing_lag", "", ["name"])
-    "event_processing_lag", labels=["name"],
-)
 
 
 def runUntilCurrentTimer(func):
@@ -197,17 +210,17 @@ def runUntilCurrentTimer(func):
             num_pending += 1
 
         num_pending += len(reactor.threadCallQueue)
-        start = time.time() * 1000
+        start = time.time()
         ret = func(*args, **kwargs)
-        end = time.time() * 1000
+        end = time.time()
 
         # record the amount of wallclock time spent running pending calls.
         # This is a proxy for the actual amount of time between reactor polls,
         # since about 25% of time is actually spent running things triggered by
         # I/O events, but that is harder to capture without rewriting half the
         # reactor.
-        tick_time.inc_by(end - start)
+        tick_time.observe(end - start)
-        pending_calls_metric.inc_by(num_pending)
+        pending_calls_metric.observe(num_pending)
 
         if running_on_pypy:
             return ret
@@ -220,12 +233,12 @@ def runUntilCurrentTimer(func):
             if threshold[i] < counts[i]:
                 logger.info("Collecting gc %d", i)
 
-                start = time.time() * 1000
+                start = time.time()
                 unreachable = gc.collect(i)
-                end = time.time() * 1000
+                end = time.time()
 
-                gc_time.inc_by(end - start, i)
+                gc_time.labels(i).observe(end - start)
-                gc_unreachable.inc_by(unreachable, i)
+                gc_unreachable.labels(i).set(unreachable)
 
         return ret
 

synapse/metrics/metric.py

@@ -1,301 +0,0 @@
-# -*- coding: utf-8 -*-
-# Copyright 2015, 2016 OpenMarket Ltd
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# 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 itertools import chain
-import logging
-
-logger = logging.getLogger(__name__)
-
-
-def flatten(items):
-    """Flatten a list of lists
-
-    Args:
-        items: iterable[iterable[X]]
-
-    Returns:
-        list[X]: flattened list
-    """
-    return list(chain.from_iterable(items))
-
-
-class BaseMetric(object):
-    """Base class for metrics which report a single value per label set
-    """
-
-    def __init__(self, name, labels=[], alternative_names=[]):
-        """
-        Args:
-            name (str): principal name for this metric
-            labels (list(str)): names of the labels which will be reported
-                for this metric
-            alternative_names (iterable(str)): list of alternative names for
-                 this metric. This can be useful to provide a migration path
-                when renaming metrics.
-        """
-        self._names = [name] + list(alternative_names)
-        self.labels = labels  # OK not to clone as we never write it
-
-    def dimension(self):
-        return len(self.labels)
-
-    def is_scalar(self):
-        return not len(self.labels)
-
-    def _render_labelvalue(self, value):
-        # TODO: escape backslashes, quotes and newlines
-        return '"%s"' % (value)
-
-    def _render_key(self, values):
-        if self.is_scalar():
-            return ""
-        return "{%s}" % (
-            ",".join(["%s=%s" % (k, self._render_labelvalue(v))
-                      for k, v in zip(self.labels, values)])
-        )
-
-    def _render_for_labels(self, label_values, value):
-        """Render this metric for a single set of labels
-
-        Args:
-            label_values (list[str]): values for each of the labels
-            value: value of the metric at with these labels
-
-        Returns:
-            iterable[str]: rendered metric
-        """
-        rendered_labels = self._render_key(label_values)
-        return (
-            "%s%s %.12g" % (name, rendered_labels, value)
-            for name in self._names
-        )
-
-    def render(self):
-        """Render this metric
-
-        Each metric is rendered as:
-
-            name{label1="val1",label2="val2"} value
-
-        https://prometheus.io/docs/instrumenting/exposition_formats/#text-format-details
-
-        Returns:
-            iterable[str]: rendered metrics
-        """
-        raise NotImplementedError()
-
-
-class CounterMetric(BaseMetric):
-    """The simplest kind of metric; one that stores a monotonically-increasing
-    value that counts events or running totals.
-
-    Example use cases for Counters:
-    - Number of requests processed
-    - Number of items that were inserted into a queue
-    - Total amount of data that a system has processed
-    Counters can only go up (and be reset when the process restarts).
-    """
-
-    def __init__(self, *args, **kwargs):
-        super(CounterMetric, 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.counts = {}
-
-        # Scalar metrics are never empty
-        if self.is_scalar():
-            self.counts[()] = 0.
-
-    def inc_by(self, incr, *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
-
-        if values not in self.counts:
-            self.counts[values] = incr
-        else:
-            self.counts[values] += incr
-
-    def inc(self, *values):
-        self.inc_by(1, *values)
-
-    def render(self):
-        return flatten(
-            self._render_for_labels(k, self.counts[k])
-            for k in sorted(self.counts.keys())
-        )
-
-
-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
-    size or other state of some in-memory object by actively querying it."""
-
-    def __init__(self, name, callback, labels=[]):
-        super(CallbackMetric, self).__init__(name, labels=labels)
-
-        self.callback = callback
-
-    def render(self):
-        try:
-            value = self.callback()
-        except Exception:
-            logger.exception("Failed to render %s", self.name)
-            return ["# FAILED to render " + self.name]
-
-        if self.is_scalar():
-            return list(self._render_for_labels([], value))
-
-        return flatten(
-            self._render_for_labels(k, value[k])
-            for k in sorted(value.keys())
-        )
-
-
-class DistributionMetric(object):
-    """A combination of an event counter and an accumulator, which counts
-    both the number of events and accumulates the total value. Typically this
-    could be used to keep track of method-running times, or other distributions
-    of values that occur in discrete occurances.
-
-    TODO(paul): Try to export some heatmap-style stats?
-    """
-
-    def __init__(self, name, *args, **kwargs):
-        self.counts = CounterMetric(name + ":count", **kwargs)
-        self.totals = CounterMetric(name + ":total", **kwargs)
-
-    def inc_by(self, inc, *values):
-        self.counts.inc(*values)
-        self.totals.inc_by(inc, *values)
-
-    def render(self):
-        return self.counts.render() + self.totals.render()
-
-
-class CacheMetric(object):
-    __slots__ = (
-        "name", "cache_name", "hits", "misses", "evicted_size", "size_callback",
-    )
-
-    def __init__(self, name, size_callback, cache_name):
-        self.name = name
-        self.cache_name = cache_name
-
-        self.hits = 0
-        self.misses = 0
-        self.evicted_size = 0
-
-        self.size_callback = size_callback
-
-    def inc_hits(self):
-        self.hits += 1
-
-    def inc_misses(self):
-        self.misses += 1
-
-    def inc_evictions(self, size=1):
-        self.evicted_size += size
-
-    def render(self):
-        size = self.size_callback()
-        hits = self.hits
-        total = self.misses + self.hits
-
-        return [
-            """%s:hits{name="%s"} %d""" % (self.name, self.cache_name, hits),
-            """%s:total{name="%s"} %d""" % (self.name, self.cache_name, total),
-            """%s:size{name="%s"} %d""" % (self.name, self.cache_name, size),
-            """%s:evicted_size{name="%s"} %d""" % (
-                self.name, self.cache_name, self.evicted_size
-            ),
-        ]
-
-
-class MemoryUsageMetric(object):
-    """Keeps track of the current memory usage, using psutil.
-
-    The class will keep the current min/max/sum/counts of rss over the last
-    WINDOW_SIZE_SEC, by polling UPDATE_HZ times per second
-    """
-
-    UPDATE_HZ = 2  # number of times to get memory per second
-    WINDOW_SIZE_SEC = 30  # the size of the window in seconds
-
-    def __init__(self, hs, psutil):
-        clock = hs.get_clock()
-        self.memory_snapshots = []
-
-        self.process = psutil.Process()
-
-        clock.looping_call(self._update_curr_values, 1000 / self.UPDATE_HZ)
-
-    def _update_curr_values(self):
-        max_size = self.UPDATE_HZ * self.WINDOW_SIZE_SEC
-        self.memory_snapshots.append(self.process.memory_info().rss)
-        self.memory_snapshots[:] = self.memory_snapshots[-max_size:]
-
-    def render(self):
-        if not self.memory_snapshots:
-            return []
-
-        max_rss = max(self.memory_snapshots)
-        min_rss = min(self.memory_snapshots)
-        sum_rss = sum(self.memory_snapshots)
-        len_rss = len(self.memory_snapshots)
-
-        return [
-            "process_psutil_rss:max %d" % max_rss,
-            "process_psutil_rss:min %d" % min_rss,
-            "process_psutil_rss:total %d" % sum_rss,
-            "process_psutil_rss:count %d" % len_rss,
-        ]

synapse/metrics/process_collector.py

@@ -1,122 +0,0 @@
-# -*- coding: utf-8 -*-
-# Copyright 2015, 2016 OpenMarket Ltd
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# 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 os
-
-
-TICKS_PER_SEC = 100
-BYTES_PER_PAGE = 4096
-
-HAVE_PROC_STAT = os.path.exists("/proc/stat")
-HAVE_PROC_SELF_STAT = os.path.exists("/proc/self/stat")
-HAVE_PROC_SELF_LIMITS = os.path.exists("/proc/self/limits")
-HAVE_PROC_SELF_FD = os.path.exists("/proc/self/fd")
-
-# Field indexes from /proc/self/stat, taken from the proc(5) manpage
-STAT_FIELDS = {
-    "utime": 14,
-    "stime": 15,
-    "starttime": 22,
-    "vsize": 23,
-    "rss": 24,
-}
-
-
-stats = {}
-
-# In order to report process_start_time_seconds we need to know the
-# machine's boot time, because the value in /proc/self/stat is relative to
-# this
-boot_time = None
-if HAVE_PROC_STAT:
-    with open("/proc/stat") as _procstat:
-        for line in _procstat:
-            if line.startswith("btime "):
-                boot_time = int(line.split()[1])
-
-
-def update_resource_metrics():
-    if HAVE_PROC_SELF_STAT:
-        global stats
-        with open("/proc/self/stat") as s:
-            line = s.read()
-            # line is PID (command) more stats go here ...
-            raw_stats = line.split(") ", 1)[1].split(" ")
-
-            for (name, index) in STAT_FIELDS.iteritems():
-                # subtract 3 from the index, because proc(5) is 1-based, and
-                # we've lost the first two fields in PID and COMMAND above
-                stats[name] = int(raw_stats[index - 3])
-
-
-def _count_fds():
-    # Not every OS will have a /proc/self/fd directory
-    if not HAVE_PROC_SELF_FD:
-        return 0
-
-    return len(os.listdir("/proc/self/fd"))
-
-
-def register_process_collector(process_metrics):
-    process_metrics.register_collector(update_resource_metrics)
-
-    if HAVE_PROC_SELF_STAT:
-        process_metrics.register_callback(
-            "cpu_user_seconds_total",
-            lambda: float(stats["utime"]) / TICKS_PER_SEC
-        )
-        process_metrics.register_callback(
-            "cpu_system_seconds_total",
-            lambda: float(stats["stime"]) / TICKS_PER_SEC
-        )
-        process_metrics.register_callback(
-            "cpu_seconds_total",
-            lambda: (float(stats["utime"] + stats["stime"])) / TICKS_PER_SEC
-        )
-
-        process_metrics.register_callback(
-            "virtual_memory_bytes",
-            lambda: int(stats["vsize"])
-        )
-        process_metrics.register_callback(
-            "resident_memory_bytes",
-            lambda: int(stats["rss"]) * BYTES_PER_PAGE
-        )
-
-        process_metrics.register_callback(
-            "start_time_seconds",
-            lambda: boot_time + int(stats["starttime"]) / TICKS_PER_SEC
-        )
-
-    if HAVE_PROC_SELF_FD:
-        process_metrics.register_callback(
-            "open_fds",
-            lambda: _count_fds()
-        )
-
-    if HAVE_PROC_SELF_LIMITS:
-        def _get_max_fds():
-            with open("/proc/self/limits") as limits:
-                for line in limits:
-                    if not line.startswith("Max open files "):
-                        continue
-                    # Line is  Max open files  $SOFT  $HARD
-                    return int(line.split()[3])
-            return None
-
-        process_metrics.register_callback(
-            "max_fds",
-            lambda: _get_max_fds()
-        )

synapse/metrics/resource.py

@@ -13,27 +13,8 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from twisted.web.resource import Resource
+from prometheus_client.twisted import MetricsResource
-
-import synapse.metrics
-
 
 METRICS_PREFIX = "/_synapse/metrics"
 
-
+__all__ = ["MetricsResource", "METRICS_PREFIX"]
-class MetricsResource(Resource):
-    isLeaf = True
-
-    def __init__(self, hs):
-        Resource.__init__(self)  # Resource is old-style, so no super()
-
-        self.hs = hs
-
-    def render_GET(self, request):
-        response = synapse.metrics.render_all()
-
-        request.setHeader("Content-Type", "text/plain")
-        request.setHeader("Content-Length", str(len(response)))
-
-        # Encode as UTF-8 (default)
-        return response.encode()

synapse/notifier.py

@@ -14,33 +14,34 @@
 # 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
-import synapse.metrics
+from synapse.metrics import LaterGauge
 
 from collections import namedtuple
+from prometheus_client import Counter
 
 import logging
 
 
 logger = logging.getLogger(__name__)
 
-metrics = synapse.metrics.get_metrics_for(__name__)
+notified_events_counter = Counter("synapse_notifier_notified_events", "")
 
-notified_events_counter = metrics.register_counter("notified_events")
+users_woken_by_stream_counter = Counter(
-
+    "synapse_notifier_users_woken_by_stream", "", ["stream"])
-users_woken_by_stream_counter = metrics.register_counter(
-    "users_woken_by_stream", labels=["stream"]
-)
 
 
 # TODO(paul): Should be shared somewhere
@@ -105,7 +106,7 @@ class _NotifierUserStream(object):
         self.last_notified_ms = time_now_ms
         noify_deferred = self.notify_deferred
 
-        users_woken_by_stream_counter.inc(stream_key)
+        users_woken_by_stream_counter.labels(stream_key).inc()
 
         with PreserveLoggingContext():
             self.notify_deferred = ObservableDeferred(defer.Deferred())
@@ -194,14 +195,14 @@ class Notifier(object):
                 all_user_streams.add(x)
 
             return sum(stream.count_listeners() for stream in all_user_streams)
-        metrics.register_callback("listeners", count_listeners)
+        LaterGauge("synapse_notifier_listeners", "", [], count_listeners)
 
-        metrics.register_callback(
+        LaterGauge(
-            "rooms",
+            "synapse_notifier_rooms", "", [],
             lambda: count(bool, self.room_to_user_streams.values()),
         )
-        metrics.register_callback(
+        LaterGauge(
-            "users",
+            "synapse_notifier_users", "", [],
             lambda: len(self.user_to_user_stream),
         )
 
@@ -251,9 +252,7 @@ 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)
@@ -267,6 +266,13 @@ 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.
 
@@ -331,11 +337,12 @@ 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
 
@@ -346,7 +353,7 @@ 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
@@ -551,13 +558,14 @@ 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