Merge branch 'rav/fix_event_sig_checks' into matrix-org/fix_event_sig_checks

Use the actual origin for push transactions, rather than whatever the remote
server claimed.

.circleci/config.yml

@@ -0,0 +1,48 @@
+version: 2
+jobs:
+  sytestpy2:
+    machine: true
+    steps:
+      - checkout
+      - run: docker pull matrixdotorg/sytest-synapsepy2
+      - run: docker run --rm -it -v $(pwd)\:/src -v $(pwd)/logs\:/logs matrixdotorg/sytest-synapsepy2
+      - store_artifacts:
+          path: ~/project/logs
+          destination: logs
+  sytestpy2postgres:
+    machine: true
+    steps:
+      - checkout
+      - run: docker pull matrixdotorg/sytest-synapsepy2
+      - run: docker run --rm -it -v $(pwd)\:/src -v $(pwd)/logs\:/logs -e POSTGRES=1 matrixdotorg/sytest-synapsepy2
+      - store_artifacts:
+          path: ~/project/logs
+          destination: logs
+  sytestpy3:
+    machine: true
+    steps:
+      - checkout
+      - run: docker pull matrixdotorg/sytest-synapsepy3
+      - run: docker run --rm -it -v $(pwd)\:/src -v $(pwd)/logs\:/logs hawkowl/sytestpy3
+      - store_artifacts:
+          path: ~/project/logs
+          destination: logs
+  sytestpy3postgres:
+    machine: true
+    steps:
+      - checkout
+      - run: docker pull matrixdotorg/sytest-synapsepy3
+      - run: docker run --rm -it -v $(pwd)\:/src -v $(pwd)/logs\:/logs -e POSTGRES=1 matrixdotorg/sytest-synapsepy3
+      - store_artifacts:
+          path: ~/project/logs
+          destination: logs
+
+workflows:
+  version: 2
+  build:
+    jobs:
+      - sytestpy2
+      - sytestpy2postgres
+# Currently broken while the Python 3 port is incomplete
+#      - sytestpy3
+#      - sytestpy3postgres

.dockerignore

@@ -3,3 +3,6 @@ Dockerfile
 .gitignore
 demo/etc
 tox.ini
+synctl
+.git/*
+.tox/*

.github/ISSUE_TEMPLATE.md

@@ -27,8 +27,9 @@ Describe here the problem that you are experiencing, or the feature you are requ
 
 Describe how what happens differs from what you expected.
 
-If you can identify any relevant log snippets from _homeserver.log_, please include
-those here (please be careful to remove any personal or private data):
+<!-- If you can identify any relevant log snippets from _homeserver.log_, please include
+those (please be careful to remove any personal or private data). Please surround them with
+``` (three backticks, on a line on their own), so that they are formatted legibly. -->
 
 ### Version information
 

.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
 
@@ -40,6 +42,8 @@ media_store/
 *.tac
 
 build/
+venv/
+venv*/
 
 localhost-800*/
 static/client/register/register_config.js

.travis.yml

@@ -4,7 +4,15 @@ language: python
 # tell travis to cache ~/.cache/pip
 cache: pip
 
+before_script:
+  - git remote set-branches --add origin develop
+  - git fetch origin develop
+
+services:
+  - postgresql
+
 matrix:
+  fast_finish: true
   include:
   - python: 2.7
     env: TOX_ENV=packaging
@@ -14,10 +22,23 @@ matrix:
 
   - python: 2.7
     env: TOX_ENV=py27
-    
+
+  - python: 2.7
+    env: TOX_ENV=py27-postgres TRIAL_FLAGS="-j 4"
+
   - python: 3.6
     env: TOX_ENV=py36
 
+  - python: 3.6
+    env: TOX_ENV=check_isort
+
+  - python: 3.6
+    env: TOX_ENV=check-newsfragment
+
+  allow_failures:
+  - python: 2.7
+    env: TOX_ENV=py27-postgres TRIAL_FLAGS="-j 4"
+
 install:
   - pip install tox
 

AUTHORS.rst

@@ -62,4 +62,7 @@ Christoph Witzany <christoph at web.crofting.com>
  * Add LDAP support for authentication
 
 Pierre Jaury <pierre at jaury.eu>
-* Docker packaging
+* Docker packaging
+
+Serban Constantin <serban.constantin at gmail dot com>
+ * Small bug fix

CONTRIBUTING.rst

@@ -30,11 +30,11 @@ use github's pull request workflow to review the contribution, and either ask
 you to make any refinements needed or merge it and make them ourselves. The
 changes will then land on master when we next do a release.
 
-We use `Jenkins <http://matrix.org/jenkins>`_ and 
+We use `Jenkins <http://matrix.org/jenkins>`_ and
 `Travis <https://travis-ci.org/matrix-org/synapse>`_ for continuous
-integration. All pull requests to synapse get automatically tested by Travis; 
-the Jenkins builds require an adminstrator to start them. If your change 
-breaks the build, this will be shown in github, so please keep an eye on the 
+integration. All pull requests to synapse get automatically tested by Travis;
+the Jenkins builds require an adminstrator to start them. If your change
+breaks the build, this will be shown in github, so please keep an eye on the
 pull request for feedback.
 
 Code style
@@ -48,6 +48,26 @@ Please ensure your changes match the cosmetic style of the existing project,
 and **never** mix cosmetic and functional changes in the same commit, as it
 makes it horribly hard to review otherwise.
 
+Changelog
+~~~~~~~~~
+
+All changes, even minor ones, need a corresponding changelog / newsfragment
+entry. These are managed by Towncrier
+(https://github.com/hawkowl/towncrier).
+
+To create a changelog entry, make a new file in the ``changelog.d``
+file named in the format of ``PRnumber.type``. The type can be
+one of ``feature``, ``bugfix``, ``removal`` (also used for
+deprecations), or ``misc`` (for internal-only changes). The content of
+the file is your changelog entry, which can contain RestructuredText
+formatting. A note of contributors is welcomed in changelogs for
+non-misc changes (the content of misc changes is not displayed).
+
+For example, a fix in PR #1234 would have its changelog entry in
+``changelog.d/1234.bugfix``, and contain content like "The security levels of
+Florbs are now validated when recieved over federation. Contributed by Jane
+Matrix".
+
 Attribution
 ~~~~~~~~~~~
 
@@ -105,16 +125,20 @@ the contribution or otherwise have the right to contribute it to Matrix::
         personal information I submit with it, including my sign-off) is
         maintained indefinitely and may be redistributed consistent with
         this project or the open source license(s) involved.
-        
+
 If you agree to this for your contribution, then all that's needed is to
 include the line in your commit or pull request comment::
 
     Signed-off-by: Your Name <your@email.example.org>
-    
-...using your real name; unfortunately pseudonyms and anonymous contributions
-can't be accepted. Git makes this trivial - just use the -s flag when you do
-``git commit``, having first set ``user.name`` and ``user.email`` git configs
-(which you should have done anyway :)
+
+We accept contributions under a legally identifiable name, such as
+your name on government documentation or common-law names (names
+claimed by legitimate usage or repute). Unfortunately, we cannot
+accept anonymous contributions at this time.
+
+Git allows you to add this signoff automatically when using the ``-s``
+flag to ``git commit``, which uses the name and email set in your
+``user.name`` and ``user.email`` git configs.
 
 Conclusion
 ~~~~~~~~~~

Dockerfile

@@ -1,19 +0,0 @@
-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

@@ -2,6 +2,7 @@ include synctl
 include LICENSE
 include VERSION
 include *.rst
+include *.md
 include demo/README
 include demo/demo.tls.dh
 include demo/*.py
@@ -29,5 +30,10 @@ exclude Dockerfile
 exclude .dockerignore
 recursive-exclude jenkins *.sh
 
+include pyproject.toml
+recursive-include changelog.d *
+
 prune .github
 prune demo/etc
+prune docker
+prune .circleci

README.rst

@@ -71,7 +71,7 @@ We'd like to invite you to join #matrix:matrix.org (via
 https://matrix.org/docs/projects/try-matrix-now.html), run a homeserver, take a look
 at the `Matrix spec <https://matrix.org/docs/spec>`_, and experiment with the
 `APIs <https://matrix.org/docs/api>`_ and `Client SDKs
-<http://matrix.org/docs/projects/try-matrix-now.html#client-sdks>`_.
+<https://matrix.org/docs/projects/try-matrix-now.html#client-sdks>`_.
 
 Thanks for using Matrix!
 
@@ -157,12 +157,19 @@ if you prefer.
 
 In case of problems, please see the _`Troubleshooting` section below.
 
-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`.
+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 <contrib/docker>`_. Further information on
+this including configuration options is available in the README on
+hub.docker.com.
 
-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/
+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
+tested with VirtualBox/AWS/DigitalOcean - see 
+https://github.com/EMnify/matrix-synapse-auto-deploy
 for details.
 
 Configuring synapse
@@ -283,7 +290,7 @@ Connecting to Synapse from a client
 
 The easiest way to try out your new Synapse installation is by connecting to it
 from a web client. The easiest option is probably the one at
-http://riot.im/app. You will need to specify a "Custom server" when you log on
+https://riot.im/app. You will need to specify a "Custom server" when you log on
 or register: set this to ``https://domain.tld`` if you setup a reverse proxy
 following the recommended setup, or ``https://localhost:8448`` - remember to specify the
 port (``:8448``) if not ``:443`` unless you changed the configuration. (Leave the identity
@@ -329,7 +336,7 @@ Security Note
 =============
 
 Matrix serves raw user generated data in some APIs - specifically the `content
-repository endpoints <http://matrix.org/docs/spec/client_server/latest.html#get-matrix-media-r0-download-servername-mediaid>`_.
+repository endpoints <https://matrix.org/docs/spec/client_server/latest.html#get-matrix-media-r0-download-servername-mediaid>`_.
 
 Whilst we have tried to mitigate against possible XSS attacks (e.g.
 https://github.com/matrix-org/synapse/pull/1021) we recommend running
@@ -348,7 +355,7 @@ Platform-Specific Instructions
 Debian
 ------
 
-Matrix provides official Debian packages via apt from http://matrix.org/packages/debian/.
+Matrix provides official Debian packages via apt from https://matrix.org/packages/debian/.
 Note that these packages do not include a client - choose one from
 https://matrix.org/docs/projects/try-matrix-now.html (or build your own with one of our SDKs :)
 
@@ -362,6 +369,19 @@ Synapse is in the Fedora repositories as ``matrix-synapse``::
 Oleg Girko provides Fedora RPMs at
 https://obs.infoserver.lv/project/monitor/matrix-synapse
 
+OpenSUSE
+--------
+
+Synapse is in the OpenSUSE repositories as ``matrix-synapse``::
+
+    sudo zypper install matrix-synapse
+
+SUSE Linux Enterprise Server
+----------------------------
+
+Unofficial package are built for SLES 15 in the openSUSE:Backports:SLE-15 repository at
+https://download.opensuse.org/repositories/openSUSE:/Backports:/SLE-15/standard/
+
 ArchLinux
 ---------
 
@@ -524,7 +544,7 @@ Troubleshooting Running
 -----------------------
 
 If synapse fails with ``missing "sodium.h"`` crypto errors, you may need
-to manually upgrade PyNaCL, as synapse uses NaCl (http://nacl.cr.yp.to/) for
+to manually upgrade PyNaCL, as synapse uses NaCl (https://nacl.cr.yp.to/) for
 encryption and digital signatures.
 Unfortunately PyNACL currently has a few issues
 (https://github.com/pyca/pynacl/issues/53) and
@@ -672,8 +692,8 @@ useful just for development purposes. See `<demo/README>`_.
 Using PostgreSQL
 ================
 
-As of Synapse 0.9, `PostgreSQL <http://www.postgresql.org>`_ is supported as an
-alternative to the `SQLite <http://sqlite.org/>`_ database that Synapse has
+As of Synapse 0.9, `PostgreSQL <https://www.postgresql.org>`_ is supported as an
+alternative to the `SQLite <https://sqlite.org/>`_ database that Synapse has
 traditionally used for convenience and simplicity.
 
 The advantages of Postgres include:
@@ -697,7 +717,7 @@ Using a reverse proxy with Synapse
 It is recommended to put a reverse proxy such as
 `nginx <https://nginx.org/en/docs/http/ngx_http_proxy_module.html>`_,
 `Apache <https://httpd.apache.org/docs/current/mod/mod_proxy_http.html>`_ or
-`HAProxy <http://www.haproxy.org/>`_ in front of Synapse. One advantage of
+`HAProxy <https://www.haproxy.org/>`_ in front of Synapse. One advantage of
 doing so is that it means that you can expose the default https port (443) to
 Matrix clients without needing to run Synapse with root privileges.
 

changelog.d/.gitignore

@@ -0,0 +1 @@
+!.gitignore

changelog.d/3659.feature

@@ -0,0 +1 @@
+Support profile API endpoints on workers

changelog.d/3673.misc

@@ -0,0 +1 @@
+Refactor state module to support multiple room versions

changelog.d/3722.bugfix

@@ -0,0 +1 @@
+Fix error collecting prometheus metrics when run on dedicated thread due to threading concurrency issues

changelog.d/3726.misc

@@ -0,0 +1 @@
+Split the state_group_cache into member and non-member state events (and so speed up LL /sync)

changelog.d/3727.misc

@@ -0,0 +1 @@
+Log failure to authenticate remote servers as warnings (without stack traces)

changelog.d/3735.misc

@@ -0,0 +1 @@
+Fix minor spelling error in federation client documentation.

contrib/docker/README.md

@@ -1,29 +1,5 @@
 # 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
@@ -60,94 +36,6 @@ Then, customize your configuration and run the server:
 docker-compose up -d
 ```
 
-### Without Compose
+### More information
 
-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.
+For more information on required environment variables and mounts, see the main docker documentation at [/docker/README.md](../../docker/README.md)

contrib/docker/docker-compose.yml

@@ -6,6 +6,7 @@ version: '3'
 services:
 
   synapse:
+    build: ../..
     image: docker.io/matrixdotorg/synapse:latest
     # Since snyapse does not retry to connect to the database, restart upon
     # failure

contrib/grafana/README.md

@@ -0,0 +1,6 @@
+# Using the Synapse Grafana dashboard
+
+0. Set up Prometheus and Grafana. Out of scope for this readme. Useful documentation about using Grafana with Prometheus: http://docs.grafana.org/features/datasources/prometheus/
+1. Have your Prometheus scrape your Synapse. https://github.com/matrix-org/synapse/blob/master/docs/metrics-howto.rst
+2. Import dashboard into Grafana. Download `synapse.json`. Import it to Grafana and select the correct Prometheus datasource. http://docs.grafana.org/reference/export_import/
+3. Set up additional recording rules

docker/Dockerfile

@@ -0,0 +1,35 @@
+FROM docker.io/python:2-alpine3.8
+
+RUN apk add --no-cache --virtual .nacl_deps \
+        build-base \
+        libffi-dev \
+        libjpeg-turbo-dev \
+        libressl-dev \
+        libxslt-dev \
+        linux-headers \
+        postgresql-dev \
+        su-exec \
+        zlib-dev
+
+COPY . /synapse
+
+# A wheel cache may be provided in ./cache for faster build
+RUN cd /synapse \
+ && pip install --upgrade \
+        lxml \
+        pip \
+        psycopg2 \
+        setuptools \
+ && mkdir -p /synapse/cache \
+ && pip install -f /synapse/cache --upgrade --process-dependency-links . \
+ && mv /synapse/docker/start.py /synapse/docker/conf / \
+ && rm -rf \
+        setup.cfg \
+        setup.py \
+        synapse
+
+VOLUME ["/data"]
+
+EXPOSE 8008/tcp 8448/tcp
+
+ENTRYPOINT ["/start.py"]

docker/README.md

@@ -0,0 +1,124 @@
+# Synapse Docker
+
+This 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.
+
+## Run
+
+We do not currently offer a `latest` image, as this has somewhat undefined semantics.
+We instead release only tagged versions so upgrading between releases is entirely
+within your control.
+
+### Using docker-compose (easier)
+
+This image is designed to run either with an automatically generated configuration
+file or with a custom configuration that requires manual editing.
+
+An easy way to make use of this image is via docker-compose. See the
+[contrib/docker](../contrib/docker)
+section of the synapse project for examples.
+
+### Without Compose (harder)
+
+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.
+
+## Build
+
+Build the docker image with the `docker build` command from the root of the synapse repository.
+
+```
+docker build -t docker.io/matrixdotorg/synapse . -f docker/Dockerfile
+```
+
+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.

docs/admin_api/register_api.rst

@@ -0,0 +1,63 @@
+Shared-Secret Registration
+==========================
+
+This API allows for the creation of users in an administrative and
+non-interactive way. This is generally used for bootstrapping a Synapse
+instance with administrator accounts.
+
+To authenticate yourself to the server, you will need both the shared secret
+(``registration_shared_secret`` in the homeserver configuration), and a
+one-time nonce. If the registration shared secret is not configured, this API
+is not enabled.
+
+To fetch the nonce, you need to request one from the API::
+
+  > GET /_matrix/client/r0/admin/register
+
+  < {"nonce": "thisisanonce"}
+
+Once you have the nonce, you can make a ``POST`` to the same URL with a JSON
+body containing the nonce, username, password, whether they are an admin
+(optional, False by default), and a HMAC digest of the content.
+
+As an example::
+
+  > POST /_matrix/client/r0/admin/register
+  > {
+     "nonce": "thisisanonce",
+     "username": "pepper_roni",
+     "password": "pizza",
+     "admin": true,
+     "mac": "mac_digest_here"
+    }
+
+  < {
+     "access_token": "token_here",
+     "user_id": "@pepper_roni:localhost",
+     "home_server": "test",
+     "device_id": "device_id_here"
+    }
+
+The MAC is the hex digest output of the HMAC-SHA1 algorithm, with the key being
+the shared secret and the content being the nonce, user, password, and either
+the string "admin" or "notadmin", each separated by NULs. For an example of
+generation in Python::
+
+  import hmac, hashlib
+
+  def generate_mac(nonce, user, password, admin=False):
+
+      mac = hmac.new(
+        key=shared_secret,
+        digestmod=hashlib.sha1,
+      )
+
+      mac.update(nonce.encode('utf8'))
+      mac.update(b"\x00")
+      mac.update(user.encode('utf8'))
+      mac.update(b"\x00")
+      mac.update(password.encode('utf8'))
+      mac.update(b"\x00")
+      mac.update(b"admin" if admin else b"notadmin")
+
+      return mac.hexdigest()

docs/admin_api/user_admin_api.rst

@@ -44,13 +44,26 @@ Deactivate Account
 
 This API deactivates an account. It removes active access tokens, resets the
 password, and deletes third-party IDs (to prevent the user requesting a
-password reset).
+password reset). It can also mark the user as GDPR-erased (stopping their data
+from distributed further, and deleting it entirely if there are no other
+references to it).
 
 The api is::
 
     POST /_matrix/client/r0/admin/deactivate/<user_id>
 
-including an ``access_token`` of a server admin, and an empty request body.
+with a body of:
+
+.. code:: json
+
+    {
+        "erase": true
+    }
+
+including an ``access_token`` of a server admin.
+
+The erase parameter is optional and defaults to 'false'.
+An empty body may be passed for backwards compatibility.
 
 
 Reset password

docs/code_style.rst

@@ -16,7 +16,7 @@
       print("I am a fish %s" %
             "moo")
 
-      and this::
+    and this::
 
         print(
             "I am a fish %s" %

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.
-   prometheus itself defaults to 9090, so starting just above that for
-   locally monitored services seems reasonable. E.g. 9092:
+   There are two methods of enabling metrics in Synapse.
 
-   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

@@ -9,19 +9,19 @@ Set up database
 Assuming your PostgreSQL database user is called ``postgres``, create a user
 ``synapse_user`` with::
 
- su - postgres
- createuser --pwprompt synapse_user
+   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.::
 
- CREATE DATABASE synapse
-  ENCODING 'UTF8'
-  LC_COLLATE='C'
-  LC_CTYPE='C'
-  template=template0
-  OWNER synapse_user;
+   CREATE DATABASE synapse
+    ENCODING 'UTF8'
+    LC_COLLATE='C'
+    LC_CTYPE='C'
+    template=template0
+    OWNER synapse_user;
 
 This would create an appropriate database named ``synapse`` owned by the
 ``synapse_user`` user (which must already exist).
@@ -126,7 +126,7 @@ 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``:
+database configuration file ``homeserver-postgres.yaml``::
 
     ./synctl stop
     mv homeserver.yaml homeserver-old-sqlite.yaml 

docs/server_notices.md

@@ -5,7 +5,7 @@ Server Notices
 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 
+[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
@@ -24,7 +24,10 @@ 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. They also cannot leave it.
+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
 ---------------------

docs/workers.rst

@@ -173,10 +173,23 @@ endpoints matching the following regular expressions::
     ^/_matrix/federation/v1/backfill/
     ^/_matrix/federation/v1/get_missing_events/
     ^/_matrix/federation/v1/publicRooms
+    ^/_matrix/federation/v1/query/
+    ^/_matrix/federation/v1/make_join/
+    ^/_matrix/federation/v1/make_leave/
+    ^/_matrix/federation/v1/send_join/
+    ^/_matrix/federation/v1/send_leave/
+    ^/_matrix/federation/v1/invite/
+    ^/_matrix/federation/v1/query_auth/
+    ^/_matrix/federation/v1/event_auth/
+    ^/_matrix/federation/v1/exchange_third_party_invite/
+    ^/_matrix/federation/v1/send/
 
 The above endpoints should all be routed to the federation_reader worker by the
 reverse-proxy configuration.
 
+The `^/_matrix/federation/v1/send/` endpoint must only be handled by a single
+instance.
+
 ``synapse.app.federation_sender``
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
@@ -206,6 +219,10 @@ Handles client API endpoints. It can handle REST endpoints matching the
 following regular expressions::
 
     ^/_matrix/client/(api/v1|r0|unstable)/publicRooms$
+    ^/_matrix/client/(api/v1|r0|unstable)/rooms/.*/joined_members$
+    ^/_matrix/client/(api/v1|r0|unstable)/rooms/.*/context/.*$
+    ^/_matrix/client/(api/v1|r0|unstable)/rooms/.*/members$
+    ^/_matrix/client/(api/v1|r0|unstable)/rooms/.*/state$
 
 ``synapse.app.user_dir``
 ~~~~~~~~~~~~~~~~~~~~~~~~
@@ -224,6 +241,14 @@ regular expressions::
 
     ^/_matrix/client/(api/v1|r0|unstable)/keys/upload
 
+If ``use_presence`` is False in the homeserver config, it can also handle REST
+endpoints matching the following regular expressions::
+
+    ^/_matrix/client/(api/v1|r0|unstable)/presence/[^/]+/status
+
+This "stub" presence handler will pass through ``GET`` request but make the
+``PUT`` effectively a no-op.
+
 It will proxy any requests it cannot handle to the main synapse instance. It
 must therefore be configured with the location of the main instance, via
 the ``worker_main_http_uri`` setting in the frontend_proxy worker configuration
@@ -240,6 +265,7 @@ Handles some event creation. It can handle REST endpoints matching::
     ^/_matrix/client/(api/v1|r0|unstable)/rooms/.*/send
     ^/_matrix/client/(api/v1|r0|unstable)/rooms/.*/(join|invite|leave|ban|unban|kick)$
     ^/_matrix/client/(api/v1|r0|unstable)/join/
+    ^/_matrix/client/(api/v1|r0|unstable)/profile/
 
 It will create events locally and then send them on to the main synapse
 instance to be persisted and handled.

pyproject.toml

@@ -0,0 +1,30 @@
+[tool.towncrier]
+    package = "synapse"
+    filename = "CHANGES.md"
+    directory = "changelog.d"
+    issue_format = "[\\#{issue}](https://github.com/matrix-org/synapse/issues/{issue})"
+
+    [[tool.towncrier.type]]
+        directory = "feature"
+        name = "Features"
+        showcontent = true
+
+    [[tool.towncrier.type]]
+        directory = "bugfix"
+        name = "Bugfixes"
+        showcontent = true
+
+    [[tool.towncrier.type]]
+        directory = "doc"
+        name = "Improved Documentation"
+        showcontent = true
+
+    [[tool.towncrier.type]]
+        directory = "removal"
+        name = "Deprecations and Removals"
+        showcontent = true
+
+    [[tool.towncrier.type]]
+        directory = "misc"
+        name = "Internal Changes"
+        showcontent = true

scripts-dev/federation_client.py

@@ -18,14 +18,22 @@
 from __future__ import print_function
 
 import argparse
+from urlparse import urlparse, urlunparse
+
 import nacl.signing
 import json
 import base64
 import requests
 import sys
+
+from requests.adapters import HTTPAdapter
 import srvlookup
 import yaml
 
+# uncomment the following to enable debug logging of http requests
+#from httplib import HTTPConnection
+#HTTPConnection.debuglevel = 1
+
 def encode_base64(input_bytes):
     """Encode bytes as a base64 string without any padding."""
 
@@ -113,17 +121,6 @@ def read_signing_keys(stream):
     return keys
 
 
-def lookup(destination, path):
-    if ":" in destination:
-        return "https://%s%s" % (destination, path)
-    else:
-        try:
-            srv = srvlookup.lookup("matrix", "tcp", destination)[0]
-            return "https://%s:%d%s" % (srv.host, srv.port, path)
-        except:
-            return "https://%s:%d%s" % (destination, 8448, path)
-
-
 def request_json(method, origin_name, origin_key, destination, path, content):
     if method is None:
         if content is None:
@@ -152,13 +149,19 @@ def request_json(method, origin_name, origin_key, destination, path, content):
         authorization_headers.append(bytes(header))
         print ("Authorization: %s" % header, file=sys.stderr)
 
-    dest = lookup(destination, path)
+    dest = "matrix://%s%s" % (destination, path)
     print ("Requesting %s" % dest, file=sys.stderr)
 
-    result = requests.request(
+    s = requests.Session()
+    s.mount("matrix://", MatrixConnectionAdapter())
+
+    result = s.request(
         method=method,
         url=dest,
-        headers={"Authorization": authorization_headers[0]},
+        headers={
+            "Host": destination,
+            "Authorization": authorization_headers[0]
+        },
         verify=False,
         data=content,
     )
@@ -242,5 +245,39 @@ def read_args_from_config(args):
             args.signing_key_path = config['signing_key_path']
 
 
+class MatrixConnectionAdapter(HTTPAdapter):
+    @staticmethod
+    def lookup(s):
+        if s[-1] == ']':
+            # ipv6 literal (with no port)
+            return s, 8448
+
+        if ":" in s:
+            out = s.rsplit(":",1)
+            try:
+                port = int(out[1])
+            except ValueError:
+                raise ValueError("Invalid host:port '%s'" % s)
+            return out[0], port
+
+        try:
+            srv = srvlookup.lookup("matrix", "tcp", s)[0]
+            return srv.host, srv.port
+        except:
+            return s, 8448
+
+    def get_connection(self, url, proxies=None):
+        parsed = urlparse(url)
+
+        (host, port) = self.lookup(parsed.netloc)
+        netloc = "%s:%d" % (host, port)
+        print("Connecting to %s" % (netloc,), file=sys.stderr)
+        url = urlunparse((
+            "https", netloc, parsed.path, parsed.params, parsed.query,
+            parsed.fragment,
+        ))
+        return super(MatrixConnectionAdapter, self).get_connection(url, proxies)
+
+
 if __name__ == "__main__":
     main()

scripts/register_new_matrix_user

@@ -26,11 +26,37 @@ import yaml
 
 
 def request_registration(user, password, server_location, shared_secret, admin=False):
+    req = urllib2.Request(
+        "%s/_matrix/client/r0/admin/register" % (server_location,),
+        headers={'Content-Type': 'application/json'}
+    )
+
+    try:
+        if sys.version_info[:3] >= (2, 7, 9):
+            # As of version 2.7.9, urllib2 now checks SSL certs
+            import ssl
+            f = urllib2.urlopen(req, context=ssl.SSLContext(ssl.PROTOCOL_SSLv23))
+        else:
+            f = urllib2.urlopen(req)
+        body = f.read()
+        f.close()
+        nonce = json.loads(body)["nonce"]
+    except urllib2.HTTPError as e:
+        print "ERROR! Received %d %s" % (e.code, e.reason,)
+        if 400 <= e.code < 500:
+            if e.info().type == "application/json":
+                resp = json.load(e)
+                if "error" in resp:
+                    print resp["error"]
+        sys.exit(1)
+
     mac = hmac.new(
         key=shared_secret,
         digestmod=hashlib.sha1,
     )
 
+    mac.update(nonce)
+    mac.update("\x00")
     mac.update(user)
     mac.update("\x00")
     mac.update(password)
@@ -40,10 +66,10 @@ def request_registration(user, password, server_location, shared_secret, admin=F
     mac = mac.hexdigest()
 
     data = {
-        "user": user,
+        "nonce": nonce,
+        "username": user,
         "password": password,
         "mac": mac,
-        "type": "org.matrix.login.shared_secret",
         "admin": admin,
     }
 
@@ -52,7 +78,7 @@ def request_registration(user, password, server_location, shared_secret, admin=F
     print "Sending registration request..."
 
     req = urllib2.Request(
-        "%s/_matrix/client/api/v1/register" % (server_location,),
+        "%s/_matrix/client/r0/admin/register" % (server_location,),
         data=json.dumps(data),
         headers={'Content-Type': 'application/json'}
     )

setup.cfg

@@ -14,7 +14,26 @@ ignore =
     pylint.cfg
     tox.ini
 
-[flake8]
+[pep8]
 max-line-length = 90
-#  W503 requires that binary operators be at the end, not start, of lines. Erik doesn't like it.
-ignore = W503
+#  W503 requires that binary operators be at the end, not start, of lines. Erik
+#  doesn't like it.  E203 is contrary to PEP8.
+ignore = W503,E203
+
+[flake8]
+# note that flake8 inherits the "ignore" settings from "pep8" (because it uses
+# pep8 to do those checks), but not the "max-line-length" setting
+max-line-length = 90
+
+[isort]
+line_length = 89
+not_skip = __init__.py
+sections=FUTURE,STDLIB,COMPAT,THIRDPARTY,TWISTED,FIRSTPARTY,TESTS,LOCALFOLDER
+default_section=THIRDPARTY
+known_first_party = synapse
+known_tests=tests
+known_compat = mock,six
+known_twisted=twisted,OpenSSL
+multi_line_output=3
+include_trailing_comma=true
+combine_as_imports=true

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__ = "0.30.0"
+__version__ = "0.33.3"

synapse/api/auth.py

@@ -15,15 +15,19 @@
 
 import logging
 
+from six import itervalues
+
 import pymacaroons
+from netaddr import IPAddress
+
 from twisted.internet import defer
 
 import synapse.types
 from synapse import event_auth
-from synapse.api.constants import EventTypes, Membership, JoinRules
-from synapse.api.errors import AuthError, Codes
+from synapse.api.constants import EventTypes, JoinRules, Membership
+from synapse.api.errors import AuthError, Codes, ResourceLimitError
 from synapse.types import UserID
-from synapse.util.caches import register_cache, CACHE_SIZE_FACTOR
+from synapse.util.caches import CACHE_SIZE_FACTOR, register_cache
 from synapse.util.caches.lrucache import LruCache
 from synapse.util.metrics import Measure
 
@@ -57,16 +61,17 @@ 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):
+        prev_state_ids = yield context.get_prev_state_ids(self.store)
         auth_events_ids = yield self.compute_auth_events(
-            event, context.prev_state_ids, for_verification=True,
+            event, prev_state_ids, for_verification=True,
         )
         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)
 
@@ -189,7 +194,7 @@ class Auth(object):
                     synapse.types.create_requester(user_id, app_service=app_service)
                 )
 
-            access_token = get_access_token_from_request(
+            access_token = self.get_access_token_from_request(
                 request, self.TOKEN_NOT_FOUND_HTTP_STATUS
             )
 
@@ -206,9 +211,9 @@ class Auth(object):
             user_agent = request.requestHeaders.getRawHeaders(
                 b"User-Agent",
                 default=[b""]
-            )[0]
+            )[0].decode('ascii', 'surrogateescape')
             if user and access_token and ip_addr:
-                self.store.insert_client_ip(
+                yield self.store.insert_client_ip(
                     user_id=user.to_string(),
                     access_token=access_token,
                     ip=ip_addr,
@@ -235,17 +240,22 @@ class Auth(object):
     @defer.inlineCallbacks
     def _get_appservice_user_id(self, request):
         app_service = self.store.get_app_service_by_token(
-            get_access_token_from_request(
+            self.get_access_token_from_request(
                 request, self.TOKEN_NOT_FOUND_HTTP_STATUS
             )
         )
         if app_service is None:
             defer.returnValue((None, None))
 
-        if "user_id" not in request.args:
+        if app_service.ip_range_whitelist:
+            ip_address = IPAddress(self.hs.get_ip_from_request(request))
+            if ip_address not in app_service.ip_range_whitelist:
+                defer.returnValue((None, None))
+
+        if b"user_id" not in request.args:
             defer.returnValue((app_service.sender, app_service))
 
-        user_id = request.args["user_id"][0]
+        user_id = request.args[b"user_id"][0].decode('utf8')
         if app_service.sender == user_id:
             defer.returnValue((app_service.sender, app_service))
 
@@ -486,7 +496,7 @@ class Auth(object):
     def _look_up_user_by_access_token(self, token):
         ret = yield self.store.get_user_by_access_token(token)
         if not ret:
-            logger.warn("Unrecognised access token - not in store: %s" % (token,))
+            logger.warn("Unrecognised access token - not in store.")
             raise AuthError(
                 self.TOKEN_NOT_FOUND_HTTP_STATUS, "Unrecognised access token.",
                 errcode=Codes.UNKNOWN_TOKEN
@@ -504,12 +514,12 @@ class Auth(object):
 
     def get_appservice_by_req(self, request):
         try:
-            token = get_access_token_from_request(
+            token = self.get_access_token_from_request(
                 request, self.TOKEN_NOT_FOUND_HTTP_STATUS
             )
             service = self.store.get_app_service_by_token(token)
             if not service:
-                logger.warn("Unrecognised appservice access token: %s" % (token,))
+                logger.warn("Unrecognised appservice access token.")
                 raise AuthError(
                     self.TOKEN_NOT_FOUND_HTTP_STATUS,
                     "Unrecognised access token.",
@@ -535,7 +545,8 @@ class Auth(object):
 
     @defer.inlineCallbacks
     def add_auth_events(self, builder, context):
-        auth_ids = yield self.compute_auth_events(builder, context.prev_state_ids)
+        prev_state_ids = yield context.get_prev_state_ids(self.store)
+        auth_ids = yield self.compute_auth_events(builder, prev_state_ids)
 
         auth_events_entries = yield self.store.add_event_hashes(
             auth_ids
@@ -653,7 +664,7 @@ 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)
 
@@ -664,67 +675,134 @@ class Auth(object):
                 " edit its room list entry"
             )
 
+    @staticmethod
+    def has_access_token(request):
+        """Checks if the request has an access_token.
 
-def has_access_token(request):
-    """Checks if the request has an access_token.
+        Returns:
+            bool: False if no access_token was given, True otherwise.
+        """
+        query_params = request.args.get(b"access_token")
+        auth_headers = request.requestHeaders.getRawHeaders(b"Authorization")
+        return bool(query_params) or bool(auth_headers)
 
-    Returns:
-        bool: False if no access_token was given, True otherwise.
-    """
-    query_params = request.args.get("access_token")
-    auth_headers = request.requestHeaders.getRawHeaders(b"Authorization")
-    return bool(query_params) or bool(auth_headers)
+    @staticmethod
+    def get_access_token_from_request(request, token_not_found_http_status=401):
+        """Extracts the access_token from the request.
 
+        Args:
+            request: The http request.
+            token_not_found_http_status(int): The HTTP status code to set in the
+                AuthError if the token isn't found. This is used in some of the
+                legacy APIs to change the status code to 403 from the default of
+                401 since some of the old clients depended on auth errors returning
+                403.
+        Returns:
+            unicode: The access_token
+        Raises:
+            AuthError: If there isn't an access_token in the request.
+        """
 
-def get_access_token_from_request(request, token_not_found_http_status=401):
-    """Extracts the access_token from the request.
-
-    Args:
-        request: The http request.
-        token_not_found_http_status(int): The HTTP status code to set in the
-            AuthError if the token isn't found. This is used in some of the
-            legacy APIs to change the status code to 403 from the default of
-            401 since some of the old clients depended on auth errors returning
-            403.
-    Returns:
-        str: The access_token
-    Raises:
-        AuthError: If there isn't an access_token in the request.
-    """
-
-    auth_headers = request.requestHeaders.getRawHeaders(b"Authorization")
-    query_params = request.args.get(b"access_token")
-    if auth_headers:
-        # Try the get the access_token from a "Authorization: Bearer"
-        # header
-        if query_params is not None:
-            raise AuthError(
-                token_not_found_http_status,
-                "Mixing Authorization headers and access_token query parameters.",
-                errcode=Codes.MISSING_TOKEN,
-            )
-        if len(auth_headers) > 1:
-            raise AuthError(
-                token_not_found_http_status,
-                "Too many Authorization headers.",
-                errcode=Codes.MISSING_TOKEN,
-            )
-        parts = auth_headers[0].split(" ")
-        if parts[0] == "Bearer" and len(parts) == 2:
-            return parts[1]
+        auth_headers = request.requestHeaders.getRawHeaders(b"Authorization")
+        query_params = request.args.get(b"access_token")
+        if auth_headers:
+            # Try the get the access_token from a "Authorization: Bearer"
+            # header
+            if query_params is not None:
+                raise AuthError(
+                    token_not_found_http_status,
+                    "Mixing Authorization headers and access_token query parameters.",
+                    errcode=Codes.MISSING_TOKEN,
+                )
+            if len(auth_headers) > 1:
+                raise AuthError(
+                    token_not_found_http_status,
+                    "Too many Authorization headers.",
+                    errcode=Codes.MISSING_TOKEN,
+                )
+            parts = auth_headers[0].split(b" ")
+            if parts[0] == b"Bearer" and len(parts) == 2:
+                return parts[1].decode('ascii')
+            else:
+                raise AuthError(
+                    token_not_found_http_status,
+                    "Invalid Authorization header.",
+                    errcode=Codes.MISSING_TOKEN,
+                )
         else:
-            raise AuthError(
-                token_not_found_http_status,
-                "Invalid Authorization header.",
-                errcode=Codes.MISSING_TOKEN,
+            # Try to get the access_token from the query params.
+            if not query_params:
+                raise AuthError(
+                    token_not_found_http_status,
+                    "Missing access token.",
+                    errcode=Codes.MISSING_TOKEN
+                )
+
+            return query_params[0].decode('ascii')
+
+    @defer.inlineCallbacks
+    def check_in_room_or_world_readable(self, room_id, user_id):
+        """Checks that the user is or was in the room or the room is world
+        readable. If it isn't then an exception is raised.
+
+        Returns:
+            Deferred[tuple[str, str|None]]: Resolves to the current membership of
+            the user in the room and the membership event ID of the user. If
+            the user is not in the room and never has been, then
+            `(Membership.JOIN, None)` is returned.
+        """
+
+        try:
+            # check_user_was_in_room will return the most recent membership
+            # event for the user if:
+            #  * The user is a non-guest user, and was ever in the room
+            #  * The user is a guest user, and has joined the room
+            # else it will throw.
+            member_event = yield self.check_user_was_in_room(room_id, user_id)
+            defer.returnValue((member_event.membership, member_event.event_id))
+        except AuthError:
+            visibility = yield self.state.get_current_state(
+                room_id, EventTypes.RoomHistoryVisibility, ""
             )
-    else:
-        # Try to get the access_token from the query params.
-        if not query_params:
+            if (
+                visibility and
+                visibility.content["history_visibility"] == "world_readable"
+            ):
+                defer.returnValue((Membership.JOIN, None))
+                return
             raise AuthError(
-                token_not_found_http_status,
-                "Missing access token.",
-                errcode=Codes.MISSING_TOKEN
+                403, "Guest access not allowed", errcode=Codes.GUEST_ACCESS_FORBIDDEN
             )
 
-        return query_params[0]
+    @defer.inlineCallbacks
+    def check_auth_blocking(self, user_id=None):
+        """Checks if the user should be rejected for some external reason,
+        such as monthly active user limiting or global disable flag
+
+        Args:
+            user_id(str|None): If present, checks for presence against existing
+            MAU cohort
+        """
+        if self.hs.config.hs_disabled:
+            raise ResourceLimitError(
+                403, self.hs.config.hs_disabled_message,
+                errcode=Codes.RESOURCE_LIMIT_EXCEED,
+                admin_uri=self.hs.config.admin_uri,
+                limit_type=self.hs.config.hs_disabled_limit_type
+            )
+        if self.hs.config.limit_usage_by_mau is True:
+            # If the user is already part of the MAU cohort
+            if user_id:
+                timestamp = yield self.store.user_last_seen_monthly_active(user_id)
+                if timestamp:
+                    return
+            # Else if there is no room in the MAU bucket, bail
+            current_mau = yield self.store.get_monthly_active_count()
+            if current_mau >= self.hs.config.max_mau_value:
+                raise ResourceLimitError(
+                    403, "Monthly Active User Limit Exceeded",
+
+                    admin_uri=self.hs.config.admin_uri,
+                    errcode=Codes.RESOURCE_LIMIT_EXCEED,
+                    limit_type="monthly_active_user"
+                )

synapse/api/constants.py

@@ -1,6 +1,7 @@
 # -*- coding: utf-8 -*-
 # Copyright 2014-2016 OpenMarket Ltd
 # Copyright 2017 Vector Creations Ltd
+# Copyright 2018 New Vector Ltd.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -76,6 +77,8 @@ class EventTypes(object):
     Topic = "m.room.topic"
     Name = "m.room.name"
 
+    ServerACL = "m.room.server_acl"
+
 
 class RejectedReason(object):
     AUTH_ERROR = "auth_error"
@@ -92,3 +95,16 @@ class RoomCreationPreset(object):
 class ThirdPartyEntityKind(object):
     USER = "user"
     LOCATION = "location"
+
+
+class RoomVersions(object):
+    V1 = "1"
+    VDH_TEST = "vdh-test-version"
+
+
+# the version we will give rooms which are created on this server
+DEFAULT_ROOM_VERSION = RoomVersions.V1
+
+# vdh-test-version is a placeholder to get room versioning support working and tested
+# until we have a working v2.
+KNOWN_ROOM_VERSIONS = {RoomVersions.V1, RoomVersions.VDH_TEST}

synapse/api/errors.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.
@@ -17,10 +18,11 @@
 
 import logging
 
-import simplejson as json
 from six import iteritems
 from six.moves import http_client
 
+from canonicaljson import json
+
 logger = logging.getLogger(__name__)
 
 
@@ -54,6 +56,9 @@ class Codes(object):
     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"
+    RESOURCE_LIMIT_EXCEED = "M_RESOURCE_LIMIT_EXCEED"
+    UNSUPPORTED_ROOM_VERSION = "M_UNSUPPORTED_ROOM_VERSION"
+    INCOMPATIBLE_ROOM_VERSION = "M_INCOMPATIBLE_ROOM_VERSION"
 
 
 class CodeMessageException(RuntimeError):
@@ -68,20 +73,6 @@ class CodeMessageException(RuntimeError):
         self.code = code
         self.msg = msg
 
-    def error_dict(self):
-        return cs_error(self.msg)
-
-
-class MatrixCodeMessageException(CodeMessageException):
-    """An error from a general matrix endpoint, eg. from a proxied Matrix API call.
-
-    Attributes:
-        errcode (str): Matrix error code e.g 'M_FORBIDDEN'
-    """
-    def __init__(self, code, msg, errcode=Codes.UNKNOWN):
-        super(MatrixCodeMessageException, self).__init__(code, msg)
-        self.errcode = errcode
-
 
 class SynapseError(CodeMessageException):
     """A base exception type for matrix errors which have an errcode and error
@@ -107,38 +98,28 @@ class SynapseError(CodeMessageException):
             self.errcode,
         )
 
-    @classmethod
-    def from_http_response_exception(cls, err):
-        """Make a SynapseError based on an HTTPResponseException
 
-        This is useful when a proxied request has failed, and we need to
-        decide how to map the failure onto a matrix error to send back to the
-        client.
+class ProxiedRequestError(SynapseError):
+    """An error from a general matrix endpoint, eg. from a proxied Matrix API call.
 
-        An attempt is made to parse the body of the http response as a matrix
-        error. If that succeeds, the errcode and error message from the body
-        are used as the errcode and error message in the new synapse error.
+    Attributes:
+        errcode (str): Matrix error code e.g 'M_FORBIDDEN'
+    """
+    def __init__(self, code, msg, errcode=Codes.UNKNOWN, additional_fields=None):
+        super(ProxiedRequestError, self).__init__(
+            code, msg, errcode
+        )
+        if additional_fields is None:
+            self._additional_fields = {}
+        else:
+            self._additional_fields = dict(additional_fields)
 
-        Otherwise, the errcode is set to M_UNKNOWN, and the error message is
-        set to the reason code from the HTTP response.
-
-        Args:
-            err (HttpResponseException):
-
-        Returns:
-            SynapseError:
-        """
-        # try to parse the body as json, to get better errcode/msg, but
-        # default to M_UNKNOWN with the HTTP status as the error text
-        try:
-            j = json.loads(err.response)
-        except ValueError:
-            j = {}
-        errcode = j.get('errcode', Codes.UNKNOWN)
-        errmsg = j.get('error', err.msg)
-
-        res = SynapseError(err.code, errmsg, errcode)
-        return res
+    def error_dict(self):
+        return cs_error(
+            self.msg,
+            self.errcode,
+            **self._additional_fields
+        )
 
 
 class ConsentNotGivenError(SynapseError):
@@ -250,6 +231,30 @@ class AuthError(SynapseError):
         super(AuthError, self).__init__(*args, **kwargs)
 
 
+class ResourceLimitError(SynapseError):
+    """
+    Any error raised when there is a problem with resource usage.
+    For instance, the monthly active user limit for the server has been exceeded
+    """
+    def __init__(
+        self, code, msg,
+        errcode=Codes.RESOURCE_LIMIT_EXCEED,
+        admin_uri=None,
+        limit_type=None,
+    ):
+        self.admin_uri = admin_uri
+        self.limit_type = limit_type
+        super(ResourceLimitError, self).__init__(code, msg, errcode=errcode)
+
+    def error_dict(self):
+        return cs_error(
+            self.msg,
+            self.errcode,
+            admin_uri=self.admin_uri,
+            limit_type=self.limit_type
+        )
+
+
 class EventSizeError(SynapseError):
     """An error raised when an event is too big."""
 
@@ -307,12 +312,25 @@ class LimitExceededError(SynapseError):
         )
 
 
-def cs_exception(exception):
-    if isinstance(exception, CodeMessageException):
-        return exception.error_dict()
-    else:
-        logger.error("Unknown exception type: %s", type(exception))
-        return {}
+class IncompatibleRoomVersionError(SynapseError):
+    """A server is trying to join a room whose version it does not support."""
+
+    def __init__(self, room_version):
+        super(IncompatibleRoomVersionError, self).__init__(
+            code=400,
+            msg="Your homeserver does not support the features required to "
+                "join this room",
+            errcode=Codes.INCOMPATIBLE_ROOM_VERSION,
+        )
+
+        self._room_version = room_version
+
+    def error_dict(self):
+        return cs_error(
+            self.msg,
+            self.errcode,
+            room_version=self._room_version,
+        )
 
 
 def cs_error(msg, code=Codes.UNKNOWN, **kwargs):
@@ -371,7 +389,7 @@ class HttpResponseException(CodeMessageException):
     Represents an HTTP-level failure of an outbound request
 
     Attributes:
-        response (str): body of response
+        response (bytes): body of response
     """
     def __init__(self, code, msg, response):
         """
@@ -379,7 +397,39 @@ class HttpResponseException(CodeMessageException):
         Args:
             code (int): HTTP status code
             msg (str): reason phrase from HTTP response status line
-            response (str): body of response
+            response (bytes): body of response
         """
         super(HttpResponseException, self).__init__(code, msg)
         self.response = response
+
+    def to_synapse_error(self):
+        """Make a SynapseError based on an HTTPResponseException
+
+        This is useful when a proxied request has failed, and we need to
+        decide how to map the failure onto a matrix error to send back to the
+        client.
+
+        An attempt is made to parse the body of the http response as a matrix
+        error. If that succeeds, the errcode and error message from the body
+        are used as the errcode and error message in the new synapse error.
+
+        Otherwise, the errcode is set to M_UNKNOWN, and the error message is
+        set to the reason code from the HTTP response.
+
+        Returns:
+            SynapseError:
+        """
+        # try to parse the body as json, to get better errcode/msg, but
+        # default to M_UNKNOWN with the HTTP status as the error text
+        try:
+            j = json.loads(self.response)
+        except ValueError:
+            j = {}
+
+        if not isinstance(j, dict):
+            j = {}
+
+        errcode = j.pop('errcode', Codes.UNKNOWN)
+        errmsg = j.pop('error', self.msg)
+
+        return ProxiedRequestError(self.code, errmsg, errcode, j)

synapse/api/filtering.py

@@ -12,14 +12,15 @@
 # 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.
-from synapse.api.errors import SynapseError
-from synapse.storage.presence import UserPresenceState
-from synapse.types import UserID, RoomID
+import jsonschema
+from canonicaljson import json
+from jsonschema import FormatChecker
+
 from twisted.internet import defer
 
-import simplejson as json
-import jsonschema
-from jsonschema import FormatChecker
+from synapse.api.errors import SynapseError
+from synapse.storage.presence import UserPresenceState
+from synapse.types import RoomID, UserID
 
 FILTER_SCHEMA = {
     "additionalProperties": False,
@@ -112,7 +113,13 @@ ROOM_EVENT_FILTER_SCHEMA = {
         },
         "contains_url": {
             "type": "boolean"
-        }
+        },
+        "lazy_load_members": {
+            "type": "boolean"
+        },
+        "include_redundant_members": {
+            "type": "boolean"
+        },
     }
 }
 
@@ -260,6 +267,12 @@ class FilterCollection(object):
     def ephemeral_limit(self):
         return self._room_ephemeral_filter.limit()
 
+    def lazy_load_members(self):
+        return self._room_state_filter.lazy_load_members()
+
+    def include_redundant_members(self):
+        return self._room_state_filter.include_redundant_members()
+
     def filter_presence(self, events):
         return self._presence_filter.filter(events)
 
@@ -411,11 +424,17 @@ 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)
 
+    def lazy_load_members(self):
+        return self.filter_json.get("lazy_load_members", False)
+
+    def include_redundant_members(self):
+        return self.filter_json.get("include_redundant_members", False)
+
 
 def _matches_wildcard(actual_value, filter_value):
     if filter_value.endswith("*"):

synapse/api/ratelimiting.py

@@ -72,7 +72,7 @@ class Ratelimiter(object):
         return allowed, time_allowed
 
     def prune_message_counts(self, time_now_s):
-        for user_id in self.message_counts.keys():
+        for user_id in list(self.message_counts.keys()):
             message_count, time_start, msg_rate_hz = (
                 self.message_counts[user_id]
             )

synapse/api/urls.py

@@ -15,8 +15,8 @@
 # limitations under the License.
 
 """Contains the URL paths to prefix various aspects of the server with. """
-from hashlib import sha256
 import hmac
+from hashlib import sha256
 
 from six.moves.urllib.parse import urlencode
 

synapse/app/__init__.py

@@ -14,9 +14,11 @@
 # limitations under the License.
 
 import sys
+
+from synapse import python_dependencies  # noqa: E402
+
 sys.dont_write_bytecode = True
 
-from synapse import python_dependencies   # noqa: E402
 
 try:
     python_dependencies.check_requirements()

synapse/app/_base.py

@@ -17,15 +17,18 @@ import gc
 import logging
 import sys
 
+from daemonize import Daemonize
+
+from twisted.internet import error, reactor
+
+from synapse.util import PreserveLoggingContext
+from synapse.util.rlimit import change_resource_limit
+
 try:
     import affinity
 except Exception:
     affinity = None
 
-from daemonize import Daemonize
-from synapse.util import PreserveLoggingContext
-from synapse.util.rlimit import change_resource_limit
-from twisted.internet import error, reactor
 
 logger = logging.getLogger(__name__)
 
@@ -124,7 +127,20 @@ def quit_with_error(error_string):
     sys.exit(1)
 
 
-def listen_tcp(bind_addresses, port, factory, backlog=50):
+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, reactor=reactor, backlog=50):
     """
     Create a TCP socket for a port and several addresses
     """
@@ -140,7 +156,9 @@ def listen_tcp(bind_addresses, port, factory, backlog=50):
             check_bind_error(e, address, bind_addresses)
 
 
-def listen_ssl(bind_addresses, port, factory, context_factory, backlog=50):
+def listen_ssl(
+    bind_addresses, port, factory, context_factory, reactor=reactor, backlog=50
+):
     """
     Create an SSL socket for a port and several addresses
     """

synapse/app/appservice.py

@@ -16,6 +16,9 @@
 import logging
 import sys
 
+from twisted.internet import defer, reactor
+from twisted.web.resource import NoResource
+
 import synapse
 from synapse import events
 from synapse.app import _base
@@ -23,6 +26,7 @@ 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.appservice import SlavedApplicationServiceStore
 from synapse.replication.slave.storage.directory import DirectoryStore
@@ -35,8 +39,6 @@ from synapse.util.httpresourcetree import create_resource_tree
 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, defer
-from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.appservice")
 
@@ -62,7 +64,7 @@ class AppserviceServer(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())
 
@@ -94,6 +96,13 @@ class AppserviceServer(HomeServer):
                         globals={"hs": self},
                     )
                 )
+            elif listener["type"] == "metrics":
+                if not self.get_config().enable_metrics:
+                    logger.warn(("Metrics listener configured, but "
+                                 "enable_metrics is not True!"))
+                else:
+                    _base.listen_metrics(listener["bind_addresses"],
+                                         listener["port"])
             else:
                 logger.warn("Unrecognized listener type: %s", listener["type"])
 
@@ -108,8 +117,9 @@ class ASReplicationHandler(ReplicationClientHandler):
         super(ASReplicationHandler, self).__init__(hs.get_datastore())
         self.appservice_handler = hs.get_application_service_handler()
 
+    @defer.inlineCallbacks
     def on_rdata(self, stream_name, token, rows):
-        super(ASReplicationHandler, self).on_rdata(stream_name, token, rows)
+        yield super(ASReplicationHandler, self).on_rdata(stream_name, token, rows)
 
         if stream_name == "events":
             max_stream_id = self.store.get_room_max_stream_ordering()

synapse/app/client_reader.py

@@ -16,6 +16,9 @@
 import logging
 import sys
 
+from twisted.internet import reactor
+from twisted.web.resource import NoResource
+
 import synapse
 from synapse import events
 from synapse.app import _base
@@ -25,8 +28,10 @@ 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
 from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore
 from synapse.replication.slave.storage.client_ips import SlavedClientIpStore
 from synapse.replication.slave.storage.directory import DirectoryStore
@@ -34,29 +39,34 @@ from synapse.replication.slave.storage.events import SlavedEventStore
 from synapse.replication.slave.storage.keys import SlavedKeyStore
 from synapse.replication.slave.storage.registration import SlavedRegistrationStore
 from synapse.replication.slave.storage.room import RoomStore
-from synapse.replication.slave.storage.transactions import TransactionStore
+from synapse.replication.slave.storage.transactions import SlavedTransactionStore
 from synapse.replication.tcp.client import ReplicationClientHandler
-from synapse.rest.client.v1.room import PublicRoomListRestServlet
+from synapse.rest.client.v1.room import (
+    JoinedRoomMemberListRestServlet,
+    PublicRoomListRestServlet,
+    RoomEventContextServlet,
+    RoomMemberListRestServlet,
+    RoomStateRestServlet,
+)
 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
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
-from twisted.internet import reactor
-from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.client_reader")
 
 
 class ClientReaderSlavedStore(
+    SlavedAccountDataStore,
     SlavedEventStore,
     SlavedKeyStore,
     RoomStore,
     DirectoryStore,
     SlavedApplicationServiceStore,
     SlavedRegistrationStore,
-    TransactionStore,
+    SlavedTransactionStore,
     SlavedClientIpStore,
     BaseSlavedStore,
 ):
@@ -77,10 +87,16 @@ 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)
+                    RoomMemberListRestServlet(self).register(resource)
+                    JoinedRoomMemberListRestServlet(self).register(resource)
+                    RoomStateRestServlet(self).register(resource)
+                    RoomEventContextServlet(self).register(resource)
+
                     resources.update({
                         "/_matrix/client/r0": resource,
                         "/_matrix/client/unstable": resource,
@@ -118,7 +134,13 @@ class ClientReaderServer(HomeServer):
                         globals={"hs": self},
                     )
                 )
-
+            elif listener["type"] == "metrics":
+                if not self.get_config().enable_metrics:
+                    logger.warn(("Metrics listener configured, but "
+                                 "enable_metrics is not True!"))
+                else:
+                    _base.listen_metrics(listener["bind_addresses"],
+                                         listener["port"])
             else:
                 logger.warn("Unrecognized listener type: %s", listener["type"])
 
@@ -146,11 +168,13 @@ def start(config_options):
     database_engine = create_engine(config.database_config)
 
     tls_server_context_factory = context_factory.ServerContextFactory(config)
+    tls_client_options_factory = context_factory.ClientTLSOptionsFactory(config)
 
     ss = ClientReaderServer(
         config.server_name,
         db_config=config.database_config,
         tls_server_context_factory=tls_server_context_factory,
+        tls_client_options_factory=tls_client_options_factory,
         config=config,
         version_string="Synapse/" + get_version_string(synapse),
         database_engine=database_engine,

synapse/app/event_creator.py

@@ -16,6 +16,9 @@
 import logging
 import sys
 
+from twisted.internet import reactor
+from twisted.web.resource import NoResource
+
 import synapse
 from synapse import events
 from synapse.app import _base
@@ -25,6 +28,7 @@ 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
@@ -39,27 +43,36 @@ from synapse.replication.slave.storage.pushers import SlavedPusherStore
 from synapse.replication.slave.storage.receipts import SlavedReceiptsStore
 from synapse.replication.slave.storage.registration import SlavedRegistrationStore
 from synapse.replication.slave.storage.room import RoomStore
-from synapse.replication.slave.storage.transactions import TransactionStore
+from synapse.replication.slave.storage.transactions import SlavedTransactionStore
 from synapse.replication.tcp.client import ReplicationClientHandler
+from synapse.rest.client.v1.profile import (
+    ProfileAvatarURLRestServlet,
+    ProfileDisplaynameRestServlet,
+    ProfileRestServlet,
+)
 from synapse.rest.client.v1.room import (
-    RoomSendEventRestServlet, RoomMembershipRestServlet, RoomStateEventRestServlet,
     JoinRoomAliasServlet,
+    RoomMembershipRestServlet,
+    RoomSendEventRestServlet,
+    RoomStateEventRestServlet,
 )
 from synapse.server import HomeServer
 from synapse.storage.engines import create_engine
+from synapse.storage.user_directory import UserDirectoryStore
 from synapse.util.httpresourcetree import create_resource_tree
 from synapse.util.logcontext import LoggingContext
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
-from twisted.internet import reactor
-from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.event_creator")
 
 
 class EventCreatorSlavedStore(
+    # FIXME(#3714): We need to add UserDirectoryStore as we write directly
+    # rather than going via the correct worker.
+    UserDirectoryStore,
     DirectoryStore,
-    TransactionStore,
+    SlavedTransactionStore,
     SlavedProfileStore,
     SlavedAccountDataStore,
     SlavedPusherStore,
@@ -90,13 +103,16 @@ 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)
                     RoomMembershipRestServlet(self).register(resource)
                     RoomStateEventRestServlet(self).register(resource)
                     JoinRoomAliasServlet(self).register(resource)
+                    ProfileAvatarURLRestServlet(self).register(resource)
+                    ProfileDisplaynameRestServlet(self).register(resource)
+                    ProfileRestServlet(self).register(resource)
                     resources.update({
                         "/_matrix/client/r0": resource,
                         "/_matrix/client/unstable": resource,
@@ -134,6 +150,13 @@ class EventCreatorServer(HomeServer):
                         globals={"hs": self},
                     )
                 )
+            elif listener["type"] == "metrics":
+                if not self.get_config().enable_metrics:
+                    logger.warn(("Metrics listener configured, but "
+                                 "enable_metrics is not True!"))
+                else:
+                    _base.listen_metrics(listener["bind_addresses"],
+                                         listener["port"])
             else:
                 logger.warn("Unrecognized listener type: %s", listener["type"])
 
@@ -163,11 +186,13 @@ def start(config_options):
     database_engine = create_engine(config.database_config)
 
     tls_server_context_factory = context_factory.ServerContextFactory(config)
+    tls_client_options_factory = context_factory.ClientTLSOptionsFactory(config)
 
     ss = EventCreatorServer(
         config.server_name,
         db_config=config.database_config,
         tls_server_context_factory=tls_server_context_factory,
+        tls_client_options_factory=tls_client_options_factory,
         config=config,
         version_string="Synapse/" + get_version_string(synapse),
         database_engine=database_engine,

synapse/app/federation_reader.py

@@ -16,6 +16,9 @@
 import logging
 import sys
 
+from twisted.internet import reactor
+from twisted.web.resource import NoResource
+
 import synapse
 from synapse import events
 from synapse.api.urls import FEDERATION_PREFIX
@@ -26,13 +29,20 @@ 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.account_data import SlavedAccountDataStore
+from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore
 from synapse.replication.slave.storage.directory import DirectoryStore
 from synapse.replication.slave.storage.events import SlavedEventStore
 from synapse.replication.slave.storage.keys import SlavedKeyStore
+from synapse.replication.slave.storage.profile import SlavedProfileStore
+from synapse.replication.slave.storage.push_rule import SlavedPushRuleStore
+from synapse.replication.slave.storage.pushers import SlavedPusherStore
+from synapse.replication.slave.storage.receipts import SlavedReceiptsStore
 from synapse.replication.slave.storage.room import RoomStore
-from synapse.replication.slave.storage.transactions import TransactionStore
+from synapse.replication.slave.storage.transactions import SlavedTransactionStore
 from synapse.replication.tcp.client import ReplicationClientHandler
 from synapse.server import HomeServer
 from synapse.storage.engines import create_engine
@@ -40,18 +50,22 @@ from synapse.util.httpresourcetree import create_resource_tree
 from synapse.util.logcontext import LoggingContext
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
-from twisted.internet import reactor
-from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.federation_reader")
 
 
 class FederationReaderSlavedStore(
+    SlavedAccountDataStore,
+    SlavedProfileStore,
+    SlavedApplicationServiceStore,
+    SlavedPusherStore,
+    SlavedPushRuleStore,
+    SlavedReceiptsStore,
     SlavedEventStore,
     SlavedKeyStore,
     RoomStore,
     DirectoryStore,
-    TransactionStore,
+    SlavedTransactionStore,
     BaseSlavedStore,
 ):
     pass
@@ -71,7 +85,7 @@ 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),
@@ -107,6 +121,13 @@ class FederationReaderServer(HomeServer):
                         globals={"hs": self},
                     )
                 )
+            elif listener["type"] == "metrics":
+                if not self.get_config().enable_metrics:
+                    logger.warn(("Metrics listener configured, but "
+                                 "enable_metrics is not True!"))
+                else:
+                    _base.listen_metrics(listener["bind_addresses"],
+                                         listener["port"])
             else:
                 logger.warn("Unrecognized listener type: %s", listener["type"])
 
@@ -134,11 +155,13 @@ def start(config_options):
     database_engine = create_engine(config.database_config)
 
     tls_server_context_factory = context_factory.ServerContextFactory(config)
+    tls_client_options_factory = context_factory.ClientTLSOptionsFactory(config)
 
     ss = FederationReaderServer(
         config.server_name,
         db_config=config.database_config,
         tls_server_context_factory=tls_server_context_factory,
+        tls_client_options_factory=tls_client_options_factory,
         config=config,
         version_string="Synapse/" + get_version_string(synapse),
         database_engine=database_engine,

synapse/app/federation_sender.py

@@ -16,6 +16,9 @@
 import logging
 import sys
 
+from twisted.internet import defer, reactor
+from twisted.web.resource import NoResource
+
 import synapse
 from synapse import events
 from synapse.app import _base
@@ -25,6 +28,7 @@ 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
@@ -32,23 +36,21 @@ from synapse.replication.slave.storage.events import SlavedEventStore
 from synapse.replication.slave.storage.presence import SlavedPresenceStore
 from synapse.replication.slave.storage.receipts import SlavedReceiptsStore
 from synapse.replication.slave.storage.registration import SlavedRegistrationStore
-from synapse.replication.slave.storage.transactions import TransactionStore
+from synapse.replication.slave.storage.transactions import SlavedTransactionStore
 from synapse.replication.tcp.client import ReplicationClientHandler
 from synapse.server import HomeServer
 from synapse.storage.engines import create_engine
-from synapse.util.async import Linearizer
+from synapse.util.async_helpers import Linearizer
 from synapse.util.httpresourcetree import create_resource_tree
 from synapse.util.logcontext import LoggingContext, run_in_background
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
-from twisted.internet import defer, reactor
-from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.federation_sender")
 
 
 class FederationSenderSlaveStore(
-    SlavedDeviceInboxStore, TransactionStore, SlavedReceiptsStore, SlavedEventStore,
+    SlavedDeviceInboxStore, SlavedTransactionStore, SlavedReceiptsStore, SlavedEventStore,
     SlavedRegistrationStore, SlavedDeviceStore, SlavedPresenceStore,
 ):
     def __init__(self, db_conn, hs):
@@ -89,7 +91,7 @@ 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())
 
@@ -121,6 +123,13 @@ class FederationSenderServer(HomeServer):
                         globals={"hs": self},
                     )
                 )
+            elif listener["type"] == "metrics":
+                if not self.get_config().enable_metrics:
+                    logger.warn(("Metrics listener configured, but "
+                                 "enable_metrics is not True!"))
+                else:
+                    _base.listen_metrics(listener["bind_addresses"],
+                                         listener["port"])
             else:
                 logger.warn("Unrecognized listener type: %s", listener["type"])
 
@@ -135,8 +144,9 @@ class FederationSenderReplicationHandler(ReplicationClientHandler):
         super(FederationSenderReplicationHandler, self).__init__(hs.get_datastore())
         self.send_handler = FederationSenderHandler(hs, self)
 
+    @defer.inlineCallbacks
     def on_rdata(self, stream_name, token, rows):
-        super(FederationSenderReplicationHandler, self).on_rdata(
+        yield super(FederationSenderReplicationHandler, self).on_rdata(
             stream_name, token, rows
         )
         self.send_handler.process_replication_rows(stream_name, token, rows)
@@ -177,11 +187,13 @@ def start(config_options):
     config.send_federation = True
 
     tls_server_context_factory = context_factory.ServerContextFactory(config)
+    tls_client_options_factory = context_factory.ClientTLSOptionsFactory(config)
 
     ps = FederationSenderServer(
         config.server_name,
         db_config=config.database_config,
         tls_server_context_factory=tls_server_context_factory,
+        tls_client_options_factory=tls_client_options_factory,
         config=config,
         version_string="Synapse/" + get_version_string(synapse),
         database_engine=database_engine,

synapse/app/frontend_proxy.py

@@ -16,6 +16,9 @@
 import logging
 import sys
 
+from twisted.internet import defer, reactor
+from twisted.web.resource import NoResource
+
 import synapse
 from synapse import events
 from synapse.api.errors import SynapseError
@@ -25,10 +28,9 @@ from synapse.config.homeserver import HomeServerConfig
 from synapse.config.logger import setup_logging
 from synapse.crypto import context_factory
 from synapse.http.server import JsonResource
-from synapse.http.servlet import (
-    RestServlet, parse_json_object_from_request,
-)
+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
@@ -36,6 +38,7 @@ from synapse.replication.slave.storage.client_ips import SlavedClientIpStore
 from synapse.replication.slave.storage.devices import SlavedDeviceStore
 from synapse.replication.slave.storage.registration import SlavedRegistrationStore
 from synapse.replication.tcp.client import ReplicationClientHandler
+from synapse.rest.client.v1.base import ClientV1RestServlet, client_path_patterns
 from synapse.rest.client.v2_alpha._base import client_v2_patterns
 from synapse.server import HomeServer
 from synapse.storage.engines import create_engine
@@ -43,12 +46,39 @@ from synapse.util.httpresourcetree import create_resource_tree
 from synapse.util.logcontext import LoggingContext
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
-from twisted.internet import defer, reactor
-from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.frontend_proxy")
 
 
+class PresenceStatusStubServlet(ClientV1RestServlet):
+    PATTERNS = client_path_patterns("/presence/(?P<user_id>[^/]*)/status")
+
+    def __init__(self, hs):
+        super(PresenceStatusStubServlet, self).__init__(hs)
+        self.http_client = hs.get_simple_http_client()
+        self.auth = hs.get_auth()
+        self.main_uri = hs.config.worker_main_http_uri
+
+    @defer.inlineCallbacks
+    def on_GET(self, request, user_id):
+        # Pass through the auth headers, if any, in case the access token
+        # is there.
+        auth_headers = request.requestHeaders.getRawHeaders("Authorization", [])
+        headers = {
+            "Authorization": auth_headers,
+        }
+        result = yield self.http_client.get_json(
+            self.main_uri + request.uri,
+            headers=headers,
+        )
+        defer.returnValue((200, result))
+
+    @defer.inlineCallbacks
+    def on_PUT(self, request, user_id):
+        yield self.auth.get_user_by_req(request)
+        defer.returnValue((200, {}))
+
+
 class KeyUploadServlet(RestServlet):
     PATTERNS = client_v2_patterns("/keys/upload(/(?P<device_id>[^/]+))?$")
 
@@ -131,10 +161,16 @@ 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)
+
+                    # If presence is disabled, use the stub servlet that does
+                    # not allow sending presence
+                    if not self.config.use_presence:
+                        PresenceStatusStubServlet(self).register(resource)
+
                     resources.update({
                         "/_matrix/client/r0": resource,
                         "/_matrix/client/unstable": resource,
@@ -153,7 +189,8 @@ class FrontendProxyServer(HomeServer):
                 listener_config,
                 root_resource,
                 self.version_string,
-            )
+            ),
+            reactor=self.get_reactor()
         )
 
         logger.info("Synapse client reader now listening on port %d", port)
@@ -172,6 +209,13 @@ class FrontendProxyServer(HomeServer):
                         globals={"hs": self},
                     )
                 )
+            elif listener["type"] == "metrics":
+                if not self.get_config().enable_metrics:
+                    logger.warn(("Metrics listener configured, but "
+                                 "enable_metrics is not True!"))
+                else:
+                    _base.listen_metrics(listener["bind_addresses"],
+                                         listener["port"])
             else:
                 logger.warn("Unrecognized listener type: %s", listener["type"])
 
@@ -201,11 +245,13 @@ def start(config_options):
     database_engine = create_engine(config.database_config)
 
     tls_server_context_factory = context_factory.ServerContextFactory(config)
+    tls_client_options_factory = context_factory.ClientTLSOptionsFactory(config)
 
     ss = FrontendProxyServer(
         config.server_name,
         db_config=config.database_config,
         tls_server_context_factory=tls_server_context_factory,
+        tls_client_options_factory=tls_client_options_factory,
         config=config,
         version_string="Synapse/" + get_version_string(synapse),
         database_engine=database_engine,

synapse/app/homeserver.py

@@ -18,27 +18,44 @@ import logging
 import os
 import sys
 
+from six import iteritems
+
+from prometheus_client import Gauge
+
+from twisted.application import service
+from twisted.internet import defer, reactor
+from twisted.web.resource import EncodingResourceWrapper, NoResource
+from twisted.web.server import GzipEncoderFactory
+from twisted.web.static import File
+
 import synapse
 import synapse.config.logger
 from synapse import events
-from synapse.api.urls import CONTENT_REPO_PREFIX, FEDERATION_PREFIX, \
-    LEGACY_MEDIA_PREFIX, MEDIA_PREFIX, SERVER_KEY_PREFIX, SERVER_KEY_V2_PREFIX, \
-    STATIC_PREFIX, WEB_CLIENT_PREFIX
+from synapse.api.urls import (
+    CONTENT_REPO_PREFIX,
+    FEDERATION_PREFIX,
+    LEGACY_MEDIA_PREFIX,
+    MEDIA_PREFIX,
+    SERVER_KEY_PREFIX,
+    SERVER_KEY_V2_PREFIX,
+    STATIC_PREFIX,
+    WEB_CLIENT_PREFIX,
+)
 from synapse.app import _base
-from synapse.app._base import quit_with_error, listen_ssl, listen_tcp
+from synapse.app._base import listen_ssl, listen_tcp, quit_with_error
 from synapse.config._base import ConfigError
 from synapse.config.homeserver import HomeServerConfig
 from synapse.crypto import context_factory
 from synapse.federation.transport.server import TransportLayerServer
-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.background_process_metrics import run_as_background_process
 from synapse.metrics.resource import METRICS_PREFIX, MetricsResource
-from synapse.python_dependencies import CONDITIONAL_REQUIREMENTS, \
-    check_requirements
-from synapse.replication.http import ReplicationRestResource, REPLICATION_PREFIX
+from synapse.module_api import ModuleApi
+from synapse.python_dependencies import CONDITIONAL_REQUIREMENTS, check_requirements
+from synapse.replication.http import REPLICATION_PREFIX, ReplicationRestResource
 from synapse.replication.tcp.resource import ReplicationStreamProtocolFactory
 from synapse.rest import ClientRestResource
 from synapse.rest.key.v1.server_key_resource import LocalKey
@@ -55,11 +72,6 @@ from synapse.util.manhole import manhole
 from synapse.util.module_loader import load_module
 from synapse.util.rlimit import change_resource_limit
 from synapse.util.versionstring import get_version_string
-from twisted.application import service
-from twisted.internet import defer, reactor
-from twisted.web.resource import EncodingResourceWrapper, NoResource
-from twisted.web.server import GzipEncoderFactory
-from twisted.web.static import File
 
 logger = logging.getLogger("synapse.app.homeserver")
 
@@ -230,7 +242,7 @@ 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)
@@ -263,6 +275,13 @@ 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 "
+                                 "enable_metrics is not True!"))
+                else:
+                    _base.listen_metrics(listener["bind_addresses"],
+                                         listener["port"])
             else:
                 logger.warn("Unrecognized listener type: %s", listener["type"])
 
@@ -283,6 +302,11 @@ class SynapseHomeServer(HomeServer):
             quit_with_error(e.message)
 
 
+# Gauges to expose monthly active user control metrics
+current_mau_gauge = Gauge("synapse_admin_mau:current", "Current MAU")
+max_mau_gauge = Gauge("synapse_admin_mau:max", "MAU Limit")
+
+
 def setup(config_options):
     """
     Args:
@@ -311,14 +335,10 @@ def setup(config_options):
     # check any extra requirements we have now we have a config
     check_requirements(config)
 
-    version_string = "Synapse/" + get_version_string(synapse)
-
-    logger.info("Server hostname: %s", config.server_name)
-    logger.info("Server version: %s", version_string)
-
     events.USE_FROZEN_DICTS = config.use_frozen_dicts
 
     tls_server_context_factory = context_factory.ServerContextFactory(config)
+    tls_client_options_factory = context_factory.ClientTLSOptionsFactory(config)
 
     database_engine = create_engine(config.database_config)
     config.database_config["args"]["cp_openfun"] = database_engine.on_new_connection
@@ -327,8 +347,9 @@ def setup(config_options):
         config.server_name,
         db_config=config.database_config,
         tls_server_context_factory=tls_server_context_factory,
+        tls_client_options_factory=tls_client_options_factory,
         config=config,
-        version_string=version_string,
+        version_string="Synapse/" + get_version_string(synapse),
         database_engine=database_engine,
     )
 
@@ -362,8 +383,6 @@ 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
@@ -418,6 +437,9 @@ def run(hs):
     # currently either 0 or 1
     stats_process = []
 
+    def start_phone_stats_home():
+        return run_as_background_process("phone_stats_home", phone_stats_home)
+
     @defer.inlineCallbacks
     def phone_stats_home():
         logger.info("Gathering stats for reporting")
@@ -434,6 +456,10 @@ 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 iteritems(daily_user_type_results):
+            stats["daily_user_type_" + name] = count
+
         room_count = yield hs.get_datastore().get_room_count()
         stats["total_room_count"] = room_count
 
@@ -442,7 +468,7 @@ def run(hs):
         stats["daily_messages"] = yield hs.get_datastore().count_daily_messages()
 
         r30_results = yield hs.get_datastore().count_r30_users()
-        for name, count in r30_results.iteritems():
+        for name, count in iteritems(r30_results):
             stats["r30_users_" + name] = count
 
         daily_sent_messages = yield hs.get_datastore().count_daily_sent_messages()
@@ -485,16 +511,41 @@ def run(hs):
             )
 
     def generate_user_daily_visit_stats():
-        hs.get_datastore().generate_user_daily_visits()
+        return run_as_background_process(
+            "generate_user_daily_visits",
+            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)
 
+    # monthly active user limiting functionality
+    clock.looping_call(
+        hs.get_datastore().reap_monthly_active_users, 1000 * 60 * 60
+    )
+    hs.get_datastore().reap_monthly_active_users()
+
+    @defer.inlineCallbacks
+    def generate_monthly_active_users():
+        count = 0
+        if hs.config.limit_usage_by_mau:
+            count = yield hs.get_datastore().get_monthly_active_count()
+        current_mau_gauge.set(float(count))
+        max_mau_gauge.set(float(hs.config.max_mau_value))
+
+    hs.get_datastore().initialise_reserved_users(
+        hs.config.mau_limits_reserved_threepids
+    )
+    generate_monthly_active_users()
+    if hs.config.limit_usage_by_mau:
+        clock.looping_call(generate_monthly_active_users, 5 * 60 * 1000)
+    # End of monthly active user settings
+
     if hs.config.report_stats:
         logger.info("Scheduling stats reporting for 3 hour intervals")
-        clock.looping_call(phone_stats_home, 3 * 60 * 60 * 1000)
+        clock.looping_call(start_phone_stats_home, 3 * 60 * 60 * 1000)
 
         # We need to defer this init for the cases that we daemonize
         # otherwise the process ID we get is that of the non-daemon process
@@ -502,7 +553,7 @@ def run(hs):
 
         # We wait 5 minutes to send the first set of stats as the server can
         # be quite busy the first few minutes
-        clock.call_later(5 * 60, phone_stats_home)
+        clock.call_later(5 * 60, start_phone_stats_home)
 
     if hs.config.daemonize and hs.config.print_pidfile:
         print (hs.config.pid_file)

synapse/app/media_repository.py

@@ -16,23 +16,25 @@
 import logging
 import sys
 
+from twisted.internet import reactor
+from twisted.web.resource import NoResource
+
 import synapse
 from synapse import events
-from synapse.api.urls import (
-    CONTENT_REPO_PREFIX, LEGACY_MEDIA_PREFIX, MEDIA_PREFIX
-)
+from synapse.api.urls import CONTENT_REPO_PREFIX, LEGACY_MEDIA_PREFIX, MEDIA_PREFIX
 from synapse.app import _base
 from synapse.config._base import ConfigError
 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
 from synapse.replication.slave.storage.client_ips import SlavedClientIpStore
 from synapse.replication.slave.storage.registration import SlavedRegistrationStore
-from synapse.replication.slave.storage.transactions import TransactionStore
+from synapse.replication.slave.storage.transactions import SlavedTransactionStore
 from synapse.replication.tcp.client import ReplicationClientHandler
 from synapse.rest.media.v0.content_repository import ContentRepoResource
 from synapse.server import HomeServer
@@ -42,8 +44,6 @@ from synapse.util.httpresourcetree import create_resource_tree
 from synapse.util.logcontext import LoggingContext
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
-from twisted.internet import reactor
-from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.media_repository")
 
@@ -52,7 +52,7 @@ class MediaRepositorySlavedStore(
     SlavedApplicationServiceStore,
     SlavedRegistrationStore,
     SlavedClientIpStore,
-    TransactionStore,
+    SlavedTransactionStore,
     BaseSlavedStore,
     MediaRepositoryStore,
 ):
@@ -73,7 +73,7 @@ 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({
@@ -114,6 +114,13 @@ class MediaRepositoryServer(HomeServer):
                         globals={"hs": self},
                     )
                 )
+            elif listener["type"] == "metrics":
+                if not self.get_config().enable_metrics:
+                    logger.warn(("Metrics listener configured, but "
+                                 "enable_metrics is not True!"))
+                else:
+                    _base.listen_metrics(listener["bind_addresses"],
+                                         listener["port"])
             else:
                 logger.warn("Unrecognized listener type: %s", listener["type"])
 
@@ -148,11 +155,13 @@ def start(config_options):
     database_engine = create_engine(config.database_config)
 
     tls_server_context_factory = context_factory.ServerContextFactory(config)
+    tls_client_options_factory = context_factory.ClientTLSOptionsFactory(config)
 
     ss = MediaRepositoryServer(
         config.server_name,
         db_config=config.database_config,
         tls_server_context_factory=tls_server_context_factory,
+        tls_client_options_factory=tls_client_options_factory,
         config=config,
         version_string="Synapse/" + get_version_string(synapse),
         database_engine=database_engine,

synapse/app/pusher.py

@@ -16,6 +16,9 @@
 import logging
 import sys
 
+from twisted.internet import defer, reactor
+from twisted.web.resource import NoResource
+
 import synapse
 from synapse import events
 from synapse.app import _base
@@ -23,6 +26,7 @@ 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
@@ -36,8 +40,6 @@ from synapse.util.httpresourcetree import create_resource_tree
 from synapse.util.logcontext import LoggingContext, run_in_background
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
-from twisted.internet import defer, reactor
-from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.pusher")
 
@@ -92,7 +94,7 @@ 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())
 
@@ -124,6 +126,13 @@ class PusherServer(HomeServer):
                         globals={"hs": self},
                     )
                 )
+            elif listener["type"] == "metrics":
+                if not self.get_config().enable_metrics:
+                    logger.warn(("Metrics listener configured, but "
+                                 "enable_metrics is not True!"))
+                else:
+                    _base.listen_metrics(listener["bind_addresses"],
+                                         listener["port"])
             else:
                 logger.warn("Unrecognized listener type: %s", listener["type"])
 
@@ -139,8 +148,9 @@ class PusherReplicationHandler(ReplicationClientHandler):
 
         self.pusher_pool = hs.get_pusherpool()
 
+    @defer.inlineCallbacks
     def on_rdata(self, stream_name, token, rows):
-        super(PusherReplicationHandler, self).on_rdata(stream_name, token, rows)
+        yield super(PusherReplicationHandler, self).on_rdata(stream_name, token, rows)
         run_in_background(self.poke_pushers, stream_name, token, rows)
 
     @defer.inlineCallbacks
@@ -153,11 +163,11 @@ class PusherReplicationHandler(ReplicationClientHandler):
                     else:
                         yield self.start_pusher(row.user_id, row.app_id, row.pushkey)
             elif stream_name == "events":
-                yield self.pusher_pool.on_new_notifications(
+                self.pusher_pool.on_new_notifications(
                     token, token,
                 )
             elif stream_name == "receipts":
-                yield self.pusher_pool.on_new_receipts(
+                self.pusher_pool.on_new_receipts(
                     token, token, set(row.room_id for row in rows)
                 )
         except Exception:

synapse/app/synchrotron.py

@@ -17,6 +17,11 @@ import contextlib
 import logging
 import sys
 
+from six import iteritems
+
+from twisted.internet import defer, reactor
+from twisted.web.resource import NoResource
+
 import synapse
 from synapse.api.constants import EventTypes
 from synapse.app import _base
@@ -26,6 +31,7 @@ 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
@@ -35,12 +41,12 @@ from synapse.replication.slave.storage.deviceinbox import SlavedDeviceInboxStore
 from synapse.replication.slave.storage.devices import SlavedDeviceStore
 from synapse.replication.slave.storage.events import SlavedEventStore
 from synapse.replication.slave.storage.filtering import SlavedFilteringStore
+from synapse.replication.slave.storage.groups import SlavedGroupServerStore
 from synapse.replication.slave.storage.presence import SlavedPresenceStore
 from synapse.replication.slave.storage.push_rule import SlavedPushRuleStore
 from synapse.replication.slave.storage.receipts import SlavedReceiptsStore
 from synapse.replication.slave.storage.registration import SlavedRegistrationStore
 from synapse.replication.slave.storage.room import RoomStore
-from synapse.replication.slave.storage.groups import SlavedGroupServerStore
 from synapse.replication.tcp.client import ReplicationClientHandler
 from synapse.rest.client.v1 import events
 from synapse.rest.client.v1.initial_sync import InitialSyncRestServlet
@@ -49,16 +55,11 @@ from synapse.rest.client.v2_alpha import sync
 from synapse.server import HomeServer
 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, run_in_background
 from synapse.util.manhole import manhole
 from synapse.util.stringutils import random_string
 from synapse.util.versionstring import get_version_string
-from twisted.internet import defer, reactor
-from twisted.web.resource import NoResource
-
-from six import iteritems
 
 logger = logging.getLogger("synapse.app.synchrotron")
 
@@ -79,9 +80,7 @@ class SynchrotronSlavedStore(
     RoomStore,
     BaseSlavedStore,
 ):
-    did_forget = (
-        RoomMemberStore.__dict__["did_forget"]
-    )
+    pass
 
 
 UPDATE_SYNCING_USERS_MS = 10 * 1000
@@ -115,7 +114,10 @@ class SynchrotronPresence(object):
         logger.info("Presence process_id is %r", self.process_id)
 
     def send_user_sync(self, user_id, is_syncing, last_sync_ms):
-        self.hs.get_tcp_replication().send_user_sync(user_id, is_syncing, last_sync_ms)
+        if self.hs.config.use_presence:
+            self.hs.get_tcp_replication().send_user_sync(
+                user_id, is_syncing, last_sync_ms
+            )
 
     def mark_as_coming_online(self, user_id):
         """A user has started syncing. Send a UserSync to the master, unless they
@@ -212,10 +214,13 @@ class SynchrotronPresence(object):
         yield self.notify_from_replication(states, stream_id)
 
     def get_currently_syncing_users(self):
-        return [
-            user_id for user_id, count in iteritems(self.user_to_num_current_syncs)
-            if count > 0
-        ]
+        if self.hs.config.use_presence:
+            return [
+                user_id for user_id, count in iteritems(self.user_to_num_current_syncs)
+                if count > 0
+            ]
+        else:
+            return set()
 
 
 class SynchrotronTyping(object):
@@ -257,7 +262,7 @@ 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)
@@ -301,6 +306,13 @@ class SynchrotronServer(HomeServer):
                         globals={"hs": self},
                     )
                 )
+            elif listener["type"] == "metrics":
+                if not self.get_config().enable_metrics:
+                    logger.warn(("Metrics listener configured, but "
+                                 "enable_metrics is not True!"))
+                else:
+                    _base.listen_metrics(listener["bind_addresses"],
+                                         listener["port"])
             else:
                 logger.warn("Unrecognized listener type: %s", listener["type"])
 
@@ -326,8 +338,9 @@ class SyncReplicationHandler(ReplicationClientHandler):
         self.presence_handler = hs.get_presence_handler()
         self.notifier = hs.get_notifier()
 
+    @defer.inlineCallbacks
     def on_rdata(self, stream_name, token, rows):
-        super(SyncReplicationHandler, self).on_rdata(stream_name, token, rows)
+        yield super(SyncReplicationHandler, self).on_rdata(stream_name, token, rows)
         run_in_background(self.process_and_notify, stream_name, token, rows)
 
     def get_streams_to_replicate(self):

synapse/app/synctl.py

@@ -16,16 +16,19 @@
 
 import argparse
 import collections
+import errno
 import glob
 import os
 import os.path
 import signal
 import subprocess
 import sys
-import yaml
-import errno
 import time
 
+from six import iteritems
+
+import yaml
+
 SYNAPSE = [sys.executable, "-B", "-m", "synapse.app.homeserver"]
 
 GREEN = "\x1b[1;32m"
@@ -108,7 +111,7 @@ def stop(pidfile, app):
 
 
 Worker = collections.namedtuple("Worker", [
-    "app", "configfile", "pidfile", "cache_factor"
+    "app", "configfile", "pidfile", "cache_factor", "cache_factors",
 ])
 
 
@@ -171,6 +174,10 @@ def main():
     if cache_factor:
         os.environ["SYNAPSE_CACHE_FACTOR"] = str(cache_factor)
 
+    cache_factors = config.get("synctl_cache_factors", {})
+    for cache_name, factor in iteritems(cache_factors):
+        os.environ["SYNAPSE_CACHE_FACTOR_" + cache_name.upper()] = str(factor)
+
     worker_configfiles = []
     if options.worker:
         start_stop_synapse = False
@@ -211,6 +218,10 @@ def main():
                 or pidfile
             )
             worker_cache_factor = worker_config.get("synctl_cache_factor") or cache_factor
+            worker_cache_factors = (
+                worker_config.get("synctl_cache_factors")
+                or cache_factors
+            )
             daemonize = worker_config.get("daemonize") or config.get("daemonize")
             assert daemonize, "Main process must have daemonize set to true"
 
@@ -226,8 +237,10 @@ def main():
             assert worker_daemonize, "In config %r: expected '%s' to be True" % (
                 worker_configfile, "worker_daemonize")
             worker_cache_factor = worker_config.get("synctl_cache_factor")
+            worker_cache_factors = worker_config.get("synctl_cache_factors", {})
         workers.append(Worker(
             worker_app, worker_configfile, worker_pidfile, worker_cache_factor,
+            worker_cache_factors,
         ))
 
     action = options.action
@@ -262,15 +275,19 @@ def main():
             start(configfile)
 
         for worker in workers:
+            env = os.environ.copy()
+
             if worker.cache_factor:
                 os.environ["SYNAPSE_CACHE_FACTOR"] = str(worker.cache_factor)
 
+            for cache_name, factor in worker.cache_factors.iteritems():
+                os.environ["SYNAPSE_CACHE_FACTOR_" + cache_name.upper()] = str(factor)
+
             start_worker(worker.app, configfile, worker.configfile)
 
-            if cache_factor:
-                os.environ["SYNAPSE_CACHE_FACTOR"] = str(cache_factor)
-            else:
-                os.environ.pop("SYNAPSE_CACHE_FACTOR", None)
+            # Reset env back to the original
+            os.environ.clear()
+            os.environ.update(env)
 
 
 if __name__ == "__main__":

synapse/app/user_dir.py

@@ -17,6 +17,9 @@
 import logging
 import sys
 
+from twisted.internet import defer, reactor
+from twisted.web.resource import NoResource
+
 import synapse
 from synapse import events
 from synapse.app import _base
@@ -26,6 +29,7 @@ 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
@@ -42,8 +46,6 @@ from synapse.util.httpresourcetree import create_resource_tree
 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, defer
-from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.user_dir")
 
@@ -105,7 +107,7 @@ 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)
@@ -146,6 +148,13 @@ class UserDirectoryServer(HomeServer):
                         globals={"hs": self},
                     )
                 )
+            elif listener["type"] == "metrics":
+                if not self.get_config().enable_metrics:
+                    logger.warn(("Metrics listener configured, but "
+                                 "enable_metrics is not True!"))
+                else:
+                    _base.listen_metrics(listener["bind_addresses"],
+                                         listener["port"])
             else:
                 logger.warn("Unrecognized listener type: %s", listener["type"])
 
@@ -160,8 +169,9 @@ class UserDirectoryReplicationHandler(ReplicationClientHandler):
         super(UserDirectoryReplicationHandler, self).__init__(hs.get_datastore())
         self.user_directory = hs.get_user_directory_handler()
 
+    @defer.inlineCallbacks
     def on_rdata(self, stream_name, token, rows):
-        super(UserDirectoryReplicationHandler, self).on_rdata(
+        yield super(UserDirectoryReplicationHandler, self).on_rdata(
             stream_name, token, rows
         )
         if stream_name == "current_state_deltas":
@@ -205,11 +215,13 @@ def start(config_options):
     config.update_user_directory = True
 
     tls_server_context_factory = context_factory.ServerContextFactory(config)
+    tls_client_options_factory = context_factory.ClientTLSOptionsFactory(config)
 
     ps = UserDirectoryServer(
         config.server_name,
         db_config=config.database_config,
         tls_server_context_factory=tls_server_context_factory,
+        tls_client_options_factory=tls_client_options_factory,
         config=config,
         version_string="Synapse/" + get_version_string(synapse),
         database_engine=database_engine,

synapse/appservice/__init__.py

@@ -12,17 +12,17 @@
 # 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 synapse.api.constants import EventTypes
-from synapse.util.caches.descriptors import cachedInlineCallbacks
-from synapse.types import GroupID, get_domain_from_id
-
-from twisted.internet import defer
-
 import logging
 import re
 
 from six import string_types
 
+from twisted.internet import defer
+
+from synapse.api.constants import EventTypes
+from synapse.types import GroupID, get_domain_from_id
+from synapse.util.caches.descriptors import cachedInlineCallbacks
+
 logger = logging.getLogger(__name__)
 
 
@@ -85,7 +85,8 @@ class ApplicationService(object):
     NS_LIST = [NS_USERS, NS_ALIASES, NS_ROOMS]
 
     def __init__(self, token, hostname, url=None, namespaces=None, hs_token=None,
-                 sender=None, id=None, protocols=None, rate_limited=True):
+                 sender=None, id=None, protocols=None, rate_limited=True,
+                 ip_range_whitelist=None):
         self.token = token
         self.url = url
         self.hs_token = hs_token
@@ -93,6 +94,7 @@ class ApplicationService(object):
         self.server_name = hostname
         self.namespaces = self._check_namespaces(namespaces)
         self.id = id
+        self.ip_range_whitelist = ip_range_whitelist
 
         if "|" in self.id:
             raise Exception("application service ID cannot contain '|' character")
@@ -292,4 +294,8 @@ class ApplicationService(object):
         return self.rate_limited
 
     def __str__(self):
-        return "ApplicationService: %s" % (self.__dict__,)
+        # copy dictionary and redact token fields so they don't get logged
+        dict_copy = self.__dict__.copy()
+        dict_copy["token"] = "<redacted>"
+        dict_copy["hs_token"] = "<redacted>"
+        return "ApplicationService: %s" % (dict_copy,)

synapse/appservice/api.py

@@ -12,20 +12,39 @@
 # 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 urllib
+
+from prometheus_client import Counter
+
 from twisted.internet import defer
 
 from synapse.api.constants import ThirdPartyEntityKind
 from synapse.api.errors import CodeMessageException
-from synapse.http.client import SimpleHttpClient
 from synapse.events.utils import serialize_event
-from synapse.util.caches.response_cache import ResponseCache
+from synapse.http.client import SimpleHttpClient
 from synapse.types import ThirdPartyInstanceID
-
-import logging
-import urllib
+from synapse.util.caches.response_cache import ResponseCache
 
 logger = logging.getLogger(__name__)
 
+sent_transactions_counter = Counter(
+    "synapse_appservice_api_sent_transactions",
+    "Number of /transactions/ requests sent",
+    ["service"]
+)
+
+failed_transactions_counter = Counter(
+    "synapse_appservice_api_failed_transactions",
+    "Number of /transactions/ requests that failed to send",
+    ["service"]
+)
+
+sent_events_counter = Counter(
+    "synapse_appservice_api_sent_events",
+    "Number of events sent to the AS",
+    ["service"]
+)
 
 HOUR_IN_MS = 60 * 60 * 1000
 
@@ -219,12 +238,15 @@ class ApplicationServiceApi(SimpleHttpClient):
                 args={
                     "access_token": service.hs_token
                 })
+            sent_transactions_counter.labels(service.id).inc()
+            sent_events_counter.labels(service.id).inc(len(events))
             defer.returnValue(True)
             return
         except CodeMessageException as e:
             logger.warning("push_bulk to %s received %s", uri, e.code)
         except Exception as ex:
             logger.warning("push_bulk to %s threw exception %s", uri, ex)
+        failed_transactions_counter.labels(service.id).inc()
         defer.returnValue(False)
 
     def _serialize(self, events):

synapse/appservice/scheduler.py

@@ -48,14 +48,14 @@ UP & quit           +---------- YES                       SUCCESS
 This is all tied together by the AppServiceScheduler which DIs the required
 components.
 """
+import logging
+
 from twisted.internet import defer
 
 from synapse.appservice import ApplicationServiceState
 from synapse.util.logcontext import run_in_background
 from synapse.util.metrics import Measure
 
-import logging
-
 logger = logging.getLogger(__name__)
 
 

synapse/config/_base.py

@@ -16,11 +16,12 @@
 import argparse
 import errno
 import os
-import yaml
 from textwrap import dedent
 
 from six import integer_types
 
+import yaml
+
 
 class ConfigError(Exception):
     pass

synapse/config/api.py

@@ -12,10 +12,10 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from ._base import Config
-
 from synapse.api.constants import EventTypes
 
+from ._base import Config
+
 
 class ApiConfig(Config):
 

synapse/config/appservice.py

@@ -12,17 +12,19 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from ._base import Config, ConfigError
-
-from synapse.appservice import ApplicationService
-from synapse.types import UserID
-
-import yaml
 import logging
 
 from six import string_types
 from six.moves.urllib import parse as urlparse
 
+import yaml
+from netaddr import IPSet
+
+from synapse.appservice import ApplicationService
+from synapse.types import UserID
+
+from ._base import Config, ConfigError
+
 logger = logging.getLogger(__name__)
 
 
@@ -154,6 +156,13 @@ def _load_appservice(hostname, as_info, config_filename):
             " will not receive events or queries.",
             config_filename,
         )
+
+    ip_range_whitelist = None
+    if as_info.get('ip_range_whitelist'):
+        ip_range_whitelist = IPSet(
+            as_info.get('ip_range_whitelist')
+        )
+
     return ApplicationService(
         token=as_info["as_token"],
         hostname=hostname,
@@ -163,5 +172,6 @@ def _load_appservice(hostname, as_info, config_filename):
         sender=user_id,
         id=as_info["id"],
         protocols=protocols,
-        rate_limited=rate_limited
+        rate_limited=rate_limited,
+        ip_range_whitelist=ip_range_whitelist,
     )

synapse/config/consent_config.py

@@ -18,6 +18,9 @@ from ._base import Config
 DEFAULT_CONFIG = """\
 # User Consent configuration
 #
+# for detailed instructions, see
+# https://github.com/matrix-org/synapse/blob/master/docs/consent_tracking.md
+#
 # Parts of this section are required if enabling the 'consent' resource under
 # 'listeners', in particular 'template_dir' and 'version'.
 #
@@ -32,7 +35,8 @@ DEFAULT_CONFIG = """\
 #
 # '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.
+# 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
@@ -46,6 +50,7 @@ DEFAULT_CONFIG = """\
 #     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
@@ -60,6 +65,7 @@ class ConsentConfig(Config):
         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):
@@ -74,6 +80,9 @@ class ConsentConfig(Config):
         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

@@ -13,32 +13,32 @@
 # 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
-from .database import DatabaseConfig
-from .ratelimiting import RatelimitConfig
-from .repository import ContentRepositoryConfig
-from .captcha import CaptchaConfig
-from .voip import VoipConfig
-from .registration import RegistrationConfig
-from .metrics import MetricsConfig
 from .api import ApiConfig
 from .appservice import AppServiceConfig
-from .key import KeyConfig
-from .saml2 import SAML2Config
+from .captcha import CaptchaConfig
 from .cas import CasConfig
-from .password import PasswordConfig
-from .jwt import JWTConfig
-from .password_auth_providers import PasswordAuthProviderConfig
-from .emailconfig import EmailConfig
-from .workers import WorkerConfig
-from .push import PushConfig
-from .spam_checker import SpamCheckerConfig
-from .groups import GroupsConfig
-from .user_directory import UserDirectoryConfig
 from .consent_config import ConsentConfig
+from .database import DatabaseConfig
+from .emailconfig import EmailConfig
+from .groups import GroupsConfig
+from .jwt import JWTConfig
+from .key import KeyConfig
+from .logger import LoggingConfig
+from .metrics import MetricsConfig
+from .password import PasswordConfig
+from .password_auth_providers import PasswordAuthProviderConfig
+from .push import PushConfig
+from .ratelimiting import RatelimitConfig
+from .registration import RegistrationConfig
+from .repository import ContentRepositoryConfig
+from .saml2 import SAML2Config
+from .server import ServerConfig
 from .server_notices_config import ServerNoticesConfig
+from .spam_checker import SpamCheckerConfig
+from .tls import TlsConfig
+from .user_directory import UserDirectoryConfig
+from .voip import VoipConfig
+from .workers import WorkerConfig
 
 
 class HomeServerConfig(TlsConfig, ServerConfig, DatabaseConfig, LoggingConfig,

synapse/config/jwt.py

@@ -15,7 +15,6 @@
 
 from ._base import Config, ConfigError
 
-
 MISSING_JWT = (
     """Missing jwt library. This is required for jwt login.
 

synapse/config/key.py

@@ -13,21 +13,24 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from ._base import Config, ConfigError
-
-from synapse.util.stringutils import random_string
-from signedjson.key import (
-    generate_signing_key, is_signing_algorithm_supported,
-    decode_signing_key_base64, decode_verify_key_bytes,
-    read_signing_keys, write_signing_keys, NACL_ED25519
-)
-from unpaddedbase64 import decode_base64
-from synapse.util.stringutils import random_string_with_symbols
-
-import os
 import hashlib
 import logging
+import os
 
+from signedjson.key import (
+    NACL_ED25519,
+    decode_signing_key_base64,
+    decode_verify_key_bytes,
+    generate_signing_key,
+    is_signing_algorithm_supported,
+    read_signing_keys,
+    write_signing_keys,
+)
+from unpaddedbase64 import decode_base64
+
+from synapse.util.stringutils import random_string, random_string_with_symbols
+
+from ._base import Config, ConfigError
 
 logger = logging.getLogger(__name__)
 

synapse/config/logger.py

@@ -12,17 +12,22 @@
 # 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.util.logcontext import LoggingContextFilter
-from twisted.logger import globalLogBeginner, STDLibLogObserver
 import logging
 import logging.config
-import yaml
-from string import Template
 import os
 import signal
+import sys
+from string import Template
 
+import yaml
+
+from twisted.logger import STDLibLogObserver, globalLogBeginner
+
+import synapse
+from synapse.util.logcontext import LoggingContextFilter
+from synapse.util.versionstring import get_version_string
+
+from ._base import Config
 
 DEFAULT_LOG_CONFIG = Template("""
 version: 1
@@ -163,7 +168,8 @@ def setup_logging(config, use_worker_options=False):
         if log_file:
             # TODO: Customisable file size / backup count
             handler = logging.handlers.RotatingFileHandler(
-                log_file, maxBytes=(1000 * 1000 * 100), backupCount=3
+                log_file, maxBytes=(1000 * 1000 * 100), backupCount=3,
+                encoding='utf8'
             )
 
             def sighup(signum, stack):
@@ -188,9 +194,8 @@ def setup_logging(config, use_worker_options=False):
 
         def sighup(signum, stack):
             # it might be better to use a file watcher or something for this.
-            logging.info("Reloading log config from %s due to SIGHUP",
-                         log_config)
             load_log_config()
+            logging.info("Reloaded log config from %s due to SIGHUP", log_config)
 
         load_log_config()
 
@@ -202,6 +207,15 @@ def setup_logging(config, use_worker_options=False):
     if getattr(signal, "SIGHUP"):
         signal.signal(signal.SIGHUP, sighup)
 
+    # make sure that the first thing we log is a thing we can grep backwards
+    # for
+    logging.warn("***** STARTING SERVER *****")
+    logging.warn(
+        "Server %s version %s",
+        sys.argv[0], get_version_string(synapse),
+    )
+    logging.info("Server hostname: %s", config.server_name)
+
     # It's critical to point twisted's internal logging somewhere, otherwise it
     # stacks up and leaks kup to 64K object;
     # see: https://twistedmatrix.com/trac/ticket/8164

synapse/config/password_auth_providers.py

@@ -13,10 +13,10 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from ._base import Config
-
 from synapse.util.module_loader import load_module
 
+from ._base import Config
+
 LDAP_PROVIDER = 'ldap_auth_provider.LdapAuthProvider'
 
 

synapse/config/registration.py

@@ -13,11 +13,11 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from ._base import Config
+from distutils.util import strtobool
 
 from synapse.util.stringutils import random_string_with_symbols
 
-from distutils.util import strtobool
+from ._base import Config
 
 
 class RegistrationConfig(Config):

synapse/config/repository.py

@@ -13,11 +13,11 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from ._base import Config, ConfigError
 from collections import namedtuple
 
 from synapse.util.module_loader import load_module
 
+from ._base import Config, ConfigError
 
 MISSING_NETADDR = (
     "Missing netaddr library. This is required for URL preview API."
@@ -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

@@ -14,13 +14,25 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+import logging
+
+from synapse.http.endpoint import parse_and_validate_server_name
+
 from ._base import Config, ConfigError
 
+logger = logging.Logger(__name__)
+
 
 class ServerConfig(Config):
 
     def read_config(self, config):
         self.server_name = config["server_name"]
+
+        try:
+            parse_and_validate_server_name(self.server_name)
+        except ValueError as e:
+            raise ConfigError(str(e))
+
         self.pid_file = self.abspath(config.get("pid_file"))
         self.web_client = config["web_client"]
         self.web_client_location = config.get("web_client_location", None)
@@ -37,6 +49,9 @@ class ServerConfig(Config):
         # "disable" federation
         self.send_federation = config.get("send_federation", True)
 
+        # Whether to enable user presence.
+        self.use_presence = config.get("use_presence", True)
+
         # Whether to update the user directory or not. This should be set to
         # false only if we are updating the user directory in a worker
         self.update_user_directory = config.get("update_user_directory", True)
@@ -55,6 +70,26 @@ class ServerConfig(Config):
             "block_non_admin_invites", False,
         )
 
+        # Options to control access by tracking MAU
+        self.limit_usage_by_mau = config.get("limit_usage_by_mau", False)
+        self.max_mau_value = 0
+        if self.limit_usage_by_mau:
+            self.max_mau_value = config.get(
+                "max_mau_value", 0,
+            )
+        self.mau_limits_reserved_threepids = config.get(
+            "mau_limit_reserved_threepids", []
+        )
+
+        # Options to disable HS
+        self.hs_disabled = config.get("hs_disabled", False)
+        self.hs_disabled_message = config.get("hs_disabled_message", "")
+        self.hs_disabled_limit_type = config.get("hs_disabled_limit_type", "")
+
+        # Admin uri to direct users at should their instance become blocked
+        # due to resource constraints
+        self.admin_uri = config.get("admin_uri", None)
+
         # FIXME: federation_domain_whitelist needs sytests
         self.federation_domain_whitelist = None
         federation_domain_whitelist = config.get(
@@ -138,6 +173,12 @@ 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")],
@@ -152,8 +193,8 @@ class ServerConfig(Config):
             })
 
     def default_config(self, server_name, **kwargs):
-        if ":" in server_name:
-            bind_port = int(server_name.split(":")[1])
+        _, bind_port = parse_and_validate_server_name(server_name)
+        if bind_port is not None:
             unsecure_port = bind_port - 400
         else:
             bind_port = 8448
@@ -191,6 +232,8 @@ class ServerConfig(Config):
         # different cores. See
         # https://www.mirantis.com/blog/improve-performance-python-programs-restricting-single-cpu/.
         #
+        # This setting requires the affinity package to be installed!
+        #
         # cpu_affinity: 0xFFFFFFFF
 
         # Whether to serve a web client from the HTTP/HTTPS root resource.
@@ -210,6 +253,9 @@ class ServerConfig(Config):
         # hard limit.
         soft_file_limit: 0
 
+        # Set to false to disable presence tracking on this homeserver.
+        use_presence: true
+
         # The GC threshold parameters to pass to `gc.set_threshold`, if defined
         # gc_thresholds: [700, 10, 10]
 
@@ -301,6 +347,32 @@ class ServerConfig(Config):
           # - port: 9000
           #   bind_addresses: ['::1', '127.0.0.1']
           #   type: manhole
+
+
+          # Homeserver blocking
+          #
+          # How to reach the server admin, used in ResourceLimitError
+          # admin_uri: 'mailto:admin@server.com'
+          #
+          # Global block config
+          #
+          # hs_disabled: False
+          # hs_disabled_message: 'Human readable reason for why the HS is blocked'
+          # hs_disabled_limit_type: 'error code(str), to help clients decode reason'
+          #
+          # Monthly Active User Blocking
+          #
+          # Enables monthly active user checking
+          # limit_usage_by_mau: False
+          # max_mau_value: 50
+          #
+          # Sometimes the server admin will want to ensure certain accounts are
+          # never blocked by mau checking. These accounts are specified here.
+          #
+          # mau_limit_reserved_threepids:
+          # - medium: 'email'
+          #   address: 'reserved_user@example.com'
+
         """ % locals()
 
     def read_arguments(self, args):

synapse/config/server_notices_config.py

@@ -12,9 +12,10 @@
 # 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
 
+from ._base import Config
+
 DEFAULT_CONFIG = """\
 # Server Notices room configuration
 #

synapse/config/tls.py

@@ -13,14 +13,15 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from ._base import Config
+import os
+import subprocess
+from hashlib import sha256
+
+from unpaddedbase64 import encode_base64
 
 from OpenSSL import crypto
-import subprocess
-import os
 
-from hashlib import sha256
-from unpaddedbase64 import encode_base64
+from ._base import Config
 
 GENERATE_DH_PARAMS = False
 

synapse/config/voip.py

@@ -30,10 +30,10 @@ class VoipConfig(Config):
         ## Turn ##
 
         # The public URIs of the TURN server to give to clients
-        turn_uris: []
+        #turn_uris: []
 
         # The shared secret used to compute passwords for the TURN server
-        turn_shared_secret: "YOUR_SHARED_SECRET"
+        #turn_shared_secret: "YOUR_SHARED_SECRET"
 
         # The Username and password if the TURN server needs them and
         # does not use a token

synapse/crypto/context_factory.py

@@ -11,19 +11,22 @@
 # 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 zope.interface import implementer
 
-from twisted.internet import ssl
 from OpenSSL import SSL, crypto
 from twisted.internet._sslverify import _defaultCurveName
-
-import logging
+from twisted.internet.interfaces import IOpenSSLClientConnectionCreator
+from twisted.internet.ssl import CertificateOptions, ContextFactory
+from twisted.python.failure import Failure
 
 logger = logging.getLogger(__name__)
 
 
-class ServerContextFactory(ssl.ContextFactory):
+class ServerContextFactory(ContextFactory):
     """Factory for PyOpenSSL SSL contexts that are used to handle incoming
-    connections and to make connections to remote servers."""
+    connections."""
 
     def __init__(self, config):
         self._context = SSL.Context(SSL.SSLv23_METHOD)
@@ -48,3 +51,78 @@ class ServerContextFactory(ssl.ContextFactory):
 
     def getContext(self):
         return self._context
+
+
+def _idnaBytes(text):
+    """
+    Convert some text typed by a human into some ASCII bytes. This is a
+    copy of twisted.internet._idna._idnaBytes. For documentation, see the
+    twisted documentation.
+    """
+    try:
+        import idna
+    except ImportError:
+        return text.encode("idna")
+    else:
+        return idna.encode(text)
+
+
+def _tolerateErrors(wrapped):
+    """
+    Wrap up an info_callback for pyOpenSSL so that if something goes wrong
+    the error is immediately logged and the connection is dropped if possible.
+    This is a copy of twisted.internet._sslverify._tolerateErrors. For
+    documentation, see the twisted documentation.
+    """
+
+    def infoCallback(connection, where, ret):
+        try:
+            return wrapped(connection, where, ret)
+        except:  # noqa: E722, taken from the twisted implementation
+            f = Failure()
+            logger.exception("Error during info_callback")
+            connection.get_app_data().failVerification(f)
+
+    return infoCallback
+
+
+@implementer(IOpenSSLClientConnectionCreator)
+class ClientTLSOptions(object):
+    """
+    Client creator for TLS without certificate identity verification. This is a
+    copy of twisted.internet._sslverify.ClientTLSOptions with the identity
+    verification left out. For documentation, see the twisted documentation.
+    """
+
+    def __init__(self, hostname, ctx):
+        self._ctx = ctx
+        self._hostname = hostname
+        self._hostnameBytes = _idnaBytes(hostname)
+        ctx.set_info_callback(
+            _tolerateErrors(self._identityVerifyingInfoCallback)
+        )
+
+    def clientConnectionForTLS(self, tlsProtocol):
+        context = self._ctx
+        connection = SSL.Connection(context, None)
+        connection.set_app_data(tlsProtocol)
+        return connection
+
+    def _identityVerifyingInfoCallback(self, connection, where, ret):
+        if where & SSL.SSL_CB_HANDSHAKE_START:
+            connection.set_tlsext_host_name(self._hostnameBytes)
+
+
+class ClientTLSOptionsFactory(object):
+    """Factory for Twisted ClientTLSOptions that are used to make connections
+    to remote servers for federation."""
+
+    def __init__(self, config):
+        # We don't use config options yet
+        pass
+
+    def get_options(self, host):
+        return ClientTLSOptions(
+            host.decode('utf-8'),
+            CertificateOptions(verify=False).getContext()
+        )

synapse/crypto/event_signing.py

@@ -15,16 +15,16 @@
 # limitations under the License.
 
 
-from synapse.api.errors import SynapseError, Codes
-from synapse.events.utils import prune_event
-
-from canonicaljson import encode_canonical_json
-from unpaddedbase64 import encode_base64, decode_base64
-from signedjson.sign import sign_json
-
 import hashlib
 import logging
 
+from canonicaljson import encode_canonical_json
+from signedjson.sign import sign_json
+from unpaddedbase64 import decode_base64, encode_base64
+
+from synapse.api.errors import Codes, SynapseError
+from synapse.events.utils import prune_event
+
 logger = logging.getLogger(__name__)
 
 

synapse/crypto/keyclient.py

@@ -13,14 +13,18 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from synapse.util import logcontext
-from twisted.web.http import HTTPClient
-from twisted.internet.protocol import Factory
-from twisted.internet import defer, reactor
-from synapse.http.endpoint import matrix_federation_endpoint
-import simplejson as json
 import logging
 
+from canonicaljson import json
+
+from twisted.internet import defer, reactor
+from twisted.internet.error import ConnectError
+from twisted.internet.protocol import Factory
+from twisted.names.error import DomainError
+from twisted.web.http import HTTPClient
+
+from synapse.http.endpoint import matrix_federation_endpoint
+from synapse.util import logcontext
 
 logger = logging.getLogger(__name__)
 
@@ -28,14 +32,14 @@ KEY_API_V1 = b"/_matrix/key/v1/"
 
 
 @defer.inlineCallbacks
-def fetch_server_key(server_name, ssl_context_factory, path=KEY_API_V1):
+def fetch_server_key(server_name, tls_client_options_factory, path=KEY_API_V1):
     """Fetch the keys for a remote server."""
 
     factory = SynapseKeyClientFactory()
     factory.path = path
     factory.host = server_name
     endpoint = matrix_federation_endpoint(
-        reactor, server_name, ssl_context_factory, timeout=30
+        reactor, server_name, tls_client_options_factory, timeout=30
     )
 
     for i in range(5):
@@ -45,12 +49,14 @@ def fetch_server_key(server_name, ssl_context_factory, path=KEY_API_V1):
                 server_response, server_certificate = yield protocol.remote_key
                 defer.returnValue((server_response, server_certificate))
         except SynapseKeyClientError as e:
-            logger.exception("Error getting key for %r" % (server_name,))
+            logger.warn("Error getting key for %r: %s", server_name, e)
             if e.status.startswith("4"):
                 # Don't retry for 4xx responses.
                 raise IOError("Cannot get key for %r" % server_name)
+        except (ConnectError, DomainError) as e:
+            logger.warn("Error getting key for %r: %s", server_name, e)
         except Exception as e:
-            logger.exception(e)
+            logger.exception("Error getting key for %r", server_name)
     raise IOError("Cannot get key for %r" % server_name)
 
 

synapse/crypto/keyring.py

@@ -14,9 +14,31 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+import hashlib
+import logging
+import urllib
+from collections import namedtuple
+
+from signedjson.key import (
+    decode_verify_key_bytes,
+    encode_verify_key_base64,
+    is_signing_algorithm_supported,
+)
+from signedjson.sign import (
+    SignatureVerifyException,
+    encode_canonical_json,
+    sign_json,
+    signature_ids,
+    verify_signed_json,
+)
+from unpaddedbase64 import decode_base64, encode_base64
+
+from OpenSSL import crypto
+from twisted.internet import defer
+
+from synapse.api.errors import Codes, SynapseError
 from synapse.crypto.keyclient import fetch_server_key
-from synapse.api.errors import SynapseError, Codes
-from synapse.util import unwrapFirstError, logcontext
+from synapse.util import logcontext, unwrapFirstError
 from synapse.util.logcontext import (
     PreserveLoggingContext,
     preserve_fn,
@@ -24,24 +46,6 @@ from synapse.util.logcontext import (
 )
 from synapse.util.metrics import Measure
 
-from twisted.internet import defer
-
-from signedjson.sign import (
-    verify_signed_json, signature_ids, sign_json, encode_canonical_json
-)
-from signedjson.key import (
-    is_signing_algorithm_supported, decode_verify_key_bytes
-)
-from unpaddedbase64 import decode_base64, encode_base64
-
-from OpenSSL import crypto
-
-from collections import namedtuple
-import urllib
-import hashlib
-import logging
-
-
 logger = logging.getLogger(__name__)
 
 
@@ -56,7 +60,7 @@ Attributes:
     key_ids(set(str)): The set of key_ids to that could be used to verify the
         JSON object
     json_object(dict): The JSON object to verify.
-    deferred(twisted.internet.defer.Deferred):
+    deferred(Deferred[str, str, nacl.signing.VerifyKey]):
         A deferred (server_name, key_id, verify_key) tuple that resolves when
         a verify key has been fetched. The deferreds' callbacks are run with no
         logcontext.
@@ -508,7 +512,7 @@ class Keyring(object):
                 continue
 
             (response, tls_certificate) = yield fetch_server_key(
-                server_name, self.hs.tls_server_context_factory,
+                server_name, self.hs.tls_client_options_factory,
                 path=(b"/_matrix/key/v2/server/%s" % (
                     urllib.quote(requested_key_id),
                 )).encode("ascii"),
@@ -651,7 +655,7 @@ class Keyring(object):
         # Try to fetch the key from the remote server.
 
         (response, tls_certificate) = yield fetch_server_key(
-            server_name, self.hs.tls_server_context_factory
+            server_name, self.hs.tls_client_options_factory
         )
 
         # Check the response.
@@ -736,6 +740,17 @@ class Keyring(object):
 
 @defer.inlineCallbacks
 def _handle_key_deferred(verify_request):
+    """Waits for the key to become available, and then performs a verification
+
+    Args:
+        verify_request (VerifyKeyRequest):
+
+    Returns:
+        Deferred[None]
+
+    Raises:
+        SynapseError if there was a problem performing the verification
+    """
     server_name = verify_request.server_name
     try:
         with PreserveLoggingContext():
@@ -768,11 +783,17 @@ def _handle_key_deferred(verify_request):
     ))
     try:
         verify_signed_json(json_object, server_name, verify_key)
-    except Exception:
+    except SignatureVerifyException as e:
+        logger.debug(
+            "Error verifying signature for %s:%s:%s with key %s: %s",
+            server_name, verify_key.alg, verify_key.version,
+            encode_verify_key_base64(verify_key),
+            str(e),
+        )
         raise SynapseError(
             401,
-            "Invalid signature for server %s with key %s:%s" % (
-                server_name, verify_key.alg, verify_key.version
+            "Invalid signature for server %s with key %s:%s: %s" % (
+                server_name, verify_key.alg, verify_key.version, str(e),
             ),
             Codes.UNAUTHORIZED,
         )

synapse/event_auth.py

@@ -17,11 +17,11 @@ import logging
 
 from canonicaljson import encode_canonical_json
 from signedjson.key import decode_verify_key_bytes
-from signedjson.sign import verify_signed_json, SignatureVerifyException
+from signedjson.sign import SignatureVerifyException, verify_signed_json
 from unpaddedbase64 import decode_base64
 
-from synapse.api.constants import EventTypes, Membership, JoinRules
-from synapse.api.errors import AuthError, SynapseError, EventSizeError
+from synapse.api.constants import KNOWN_ROOM_VERSIONS, EventTypes, JoinRules, Membership
+from synapse.api.errors import AuthError, EventSizeError, SynapseError
 from synapse.types import UserID, get_domain_from_id
 
 logger = logging.getLogger(__name__)
@@ -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)
@@ -71,17 +73,27 @@ 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:
+        sender_domain = get_domain_from_id(event.sender)
         room_id_domain = get_domain_from_id(event.room_id)
         if room_id_domain != sender_domain:
             raise AuthError(
                 403,
                 "Creation event's room_id domain does not match sender's"
             )
+
+        room_version = event.content.get("room_version", "1")
+        if room_version not in KNOWN_ROOM_VERSIONS:
+            raise AuthError(
+                403,
+                "room appears to have unsupported version %s" % (
+                    room_version,
+                ))
         # FIXME
-        return True
+        logger.debug("Allowing! %s", event)
+        return
 
     creation_event = auth_events.get((EventTypes.Create, ""), None)
 
@@ -118,7 +130,8 @@ 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 +140,9 @@ def check(event, auth_events, do_sig_check=True, do_size_check=True):
         )
 
     if event.type == EventTypes.Member:
-        allowed = _is_membership_change_allowed(
-            event, auth_events
-        )
-        if allowed:
-            logger.debug("Allowing! %s", event)
-        else:
-            logger.debug("Denying! %s", event)
-        return allowed
+        _is_membership_change_allowed(event, auth_events)
+        logger.debug("Allowing! %s", event)
+        return
 
     _check_event_sender_in_room(event, auth_events)
 
@@ -153,7 +161,8 @@ 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 +209,7 @@ 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 +274,13 @@ 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 +343,6 @@ 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 +362,46 @@ def _check_joined_room(member, user_id, room_id):
         ))
 
 
-def get_send_level(etype, state_key, auth_events):
-    key = (EventTypes.PowerLevels, "", )
-    send_level_event = auth_events.get(key)
-    send_level = None
-    if send_level_event:
-        send_level = send_level_event.content.get("events", {}).get(
-            etype
-        )
-        if send_level is None:
-            if state_key is not None:
-                send_level = send_level_event.content.get(
-                    "state_default", 50
-                )
-            else:
-                send_level = send_level_event.content.get(
-                    "events_default", 0
-                )
+def get_send_level(etype, state_key, power_levels_event):
+    """Get the power level required to send an event of a given type
 
-    if send_level:
-        send_level = int(send_level)
+    The federation spec [1] refers to this as "Required Power Level".
+
+    https://matrix.org/docs/spec/server_server/unstable.html#definitions
+
+    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:
-        send_level = 0
+        power_levels_content = {}
 
-    return send_level
+    # 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 = power_levels_content.get("state_default", 50)
+        else:
+            send_level = power_levels_content.get("events_default", 0)
+
+    return int(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 +489,14 @@ 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")
         )
@@ -515,7 +533,11 @@ def _check_power_levels(event, auth_events):
                     "to your own"
                 )
 
-        if old_level > user_level or new_level > user_level:
+        # Check if the old and new levels are greater than the user level
+        # (if defined)
+        old_level_too_big = old_level is not None and old_level > user_level
+        new_level_too_big = new_level is not None and new_level > user_level
+        if old_level_too_big or new_level_too_big:
             raise AuthError(
                 403,
                 "You don't have permission to add ops level greater "
@@ -524,13 +546,22 @@ def _check_power_levels(event, auth_events):
 
 
 def _get_power_level_event(auth_events):
-    key = (EventTypes.PowerLevels, "", )
-    return auth_events.get(key)
+    return auth_events.get((EventTypes.PowerLevels, ""))
 
 
 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 +572,11 @@ 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

@@ -13,9 +13,8 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from synapse.util.frozenutils import freeze
 from synapse.util.caches import intern_dict
-
+from synapse.util.frozenutils import freeze
 
 # Whether we should use frozen_dict in FrozenEvent. Using frozen_dicts prevents
 # bugs where we accidentally share e.g. signature dicts. However, converting
@@ -146,7 +145,7 @@ 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/builder.py

@@ -13,13 +13,12 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from . import EventBase, FrozenEvent, _event_dict_property
+import copy
 
 from synapse.types import EventID
-
 from synapse.util.stringutils import random_string
 
-import copy
+from . import EventBase, FrozenEvent, _event_dict_property
 
 
 class EventBuilder(EventBase):

synapse/events/snapshot.py

@@ -13,22 +13,18 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from twisted.internet import defer
+from six import iteritems
 
 from frozendict import frozendict
 
+from twisted.internet import defer
+
+from synapse.util.logcontext import make_deferred_yieldable, run_in_background
+
 
 class EventContext(object):
     """
     Attributes:
-        current_state_ids (dict[(str, str), str]):
-            The current state map including the current event.
-            (type, state_key) -> event_id
-
-        prev_state_ids (dict[(str, str), str]):
-            The current state map excluding the current event.
-            (type, state_key) -> event_id
-
         state_group (int|None): state group id, if the state has been stored
             as a state group. This is usually only None if e.g. the event is
             an outlier.
@@ -45,38 +41,77 @@ class EventContext(object):
 
         prev_state_events (?): XXX: is this ever set to anything other than
             the empty list?
+
+        _current_state_ids (dict[(str, str), str]|None):
+            The current state map including the current event. None if outlier
+            or we haven't fetched the state from DB yet.
+            (type, state_key) -> event_id
+
+        _prev_state_ids (dict[(str, str), str]|None):
+            The current state map excluding the current event. None if outlier
+            or we haven't fetched the state from DB yet.
+            (type, state_key) -> event_id
+
+        _fetching_state_deferred (Deferred|None): Resolves when *_state_ids have
+            been calculated. None if we haven't started calculating yet
+
+        _event_type (str): The type of the event the context is associated with.
+            Only set when state has not been fetched yet.
+
+        _event_state_key (str|None): The state_key of the event the context is
+            associated with. Only set when state has not been fetched yet.
+
+        _prev_state_id (str|None): If the event associated with the context is
+            a state event, then `_prev_state_id` is the event_id of the state
+            that was replaced.
+            Only set when state has not been fetched yet.
     """
 
     __slots__ = [
-        "current_state_ids",
-        "prev_state_ids",
         "state_group",
         "rejected",
         "prev_group",
         "delta_ids",
         "prev_state_events",
         "app_service",
+        "_current_state_ids",
+        "_prev_state_ids",
+        "_prev_state_id",
+        "_event_type",
+        "_event_state_key",
+        "_fetching_state_deferred",
     ]
 
     def __init__(self):
-        # The current state including the current event
-        self.current_state_ids = None
-        # The current state excluding the current event
-        self.prev_state_ids = None
-        self.state_group = None
-
+        self.prev_state_events = []
         self.rejected = False
+        self.app_service = None
+
+    @staticmethod
+    def with_state(state_group, current_state_ids, prev_state_ids,
+                   prev_group=None, delta_ids=None):
+        context = EventContext()
+
+        # The current state including the current event
+        context._current_state_ids = current_state_ids
+        # The current state excluding the current event
+        context._prev_state_ids = prev_state_ids
+        context.state_group = state_group
+
+        context._prev_state_id = None
+        context._event_type = None
+        context._event_state_key = None
+        context._fetching_state_deferred = defer.succeed(None)
 
         # A previously persisted state group and a delta between that
         # and this state.
-        self.prev_group = None
-        self.delta_ids = None
+        context.prev_group = prev_group
+        context.delta_ids = delta_ids
 
-        self.prev_state_events = None
+        return context
 
-        self.app_service = None
-
-    def serialize(self, event):
+    @defer.inlineCallbacks
+    def serialize(self, event, store):
         """Converts self to a type that can be serialized as JSON, and then
         deserialized by `deserialize`
 
@@ -92,11 +127,12 @@ class EventContext(object):
         # the prev_state_ids, so if we're a state event we include the event
         # id that we replaced in the state.
         if event.is_state():
-            prev_state_id = self.prev_state_ids.get((event.type, event.state_key))
+            prev_state_ids = yield self.get_prev_state_ids(store)
+            prev_state_id = prev_state_ids.get((event.type, event.state_key))
         else:
             prev_state_id = None
 
-        return {
+        defer.returnValue({
             "prev_state_id": prev_state_id,
             "event_type": event.type,
             "event_state_key": event.state_key if event.is_state() else None,
@@ -106,10 +142,9 @@ class EventContext(object):
             "delta_ids": _encode_state_dict(self.delta_ids),
             "prev_state_events": self.prev_state_events,
             "app_service_id": self.app_service.id if self.app_service else None
-        }
+        })
 
     @staticmethod
-    @defer.inlineCallbacks
     def deserialize(store, input):
         """Converts a dict that was produced by `serialize` back into a
         EventContext.
@@ -122,32 +157,115 @@ class EventContext(object):
             EventContext
         """
         context = EventContext()
-        context.state_group = input["state_group"]
-        context.rejected = input["rejected"]
-        context.prev_group = input["prev_group"]
-        context.delta_ids = _decode_state_dict(input["delta_ids"])
-        context.prev_state_events = input["prev_state_events"]
 
         # We use the state_group and prev_state_id stuff to pull the
         # current_state_ids out of the DB and construct prev_state_ids.
-        prev_state_id = input["prev_state_id"]
-        event_type = input["event_type"]
-        event_state_key = input["event_state_key"]
+        context._prev_state_id = input["prev_state_id"]
+        context._event_type = input["event_type"]
+        context._event_state_key = input["event_state_key"]
 
-        context.current_state_ids = yield store.get_state_ids_for_group(
-            context.state_group,
-        )
-        if prev_state_id and event_state_key:
-            context.prev_state_ids = dict(context.current_state_ids)
-            context.prev_state_ids[(event_type, event_state_key)] = prev_state_id
-        else:
-            context.prev_state_ids = context.current_state_ids
+        context._current_state_ids = None
+        context._prev_state_ids = None
+        context._fetching_state_deferred = None
+
+        context.state_group = input["state_group"]
+        context.prev_group = input["prev_group"]
+        context.delta_ids = _decode_state_dict(input["delta_ids"])
+
+        context.rejected = input["rejected"]
+        context.prev_state_events = input["prev_state_events"]
 
         app_service_id = input["app_service_id"]
         if app_service_id:
             context.app_service = store.get_app_service_by_id(app_service_id)
 
-        defer.returnValue(context)
+        return context
+
+    @defer.inlineCallbacks
+    def get_current_state_ids(self, store):
+        """Gets the current state IDs
+
+        Returns:
+            Deferred[dict[(str, str), str]|None]: Returns None if state_group
+            is None, which happens when the associated event is an outlier.
+        """
+
+        if not self._fetching_state_deferred:
+            self._fetching_state_deferred = run_in_background(
+                self._fill_out_state, store,
+            )
+
+        yield make_deferred_yieldable(self._fetching_state_deferred)
+
+        defer.returnValue(self._current_state_ids)
+
+    @defer.inlineCallbacks
+    def get_prev_state_ids(self, store):
+        """Gets the prev state IDs
+
+        Returns:
+            Deferred[dict[(str, str), str]|None]: Returns None if state_group
+            is None, which happens when the associated event is an outlier.
+        """
+
+        if not self._fetching_state_deferred:
+            self._fetching_state_deferred = run_in_background(
+                self._fill_out_state, store,
+            )
+
+        yield make_deferred_yieldable(self._fetching_state_deferred)
+
+        defer.returnValue(self._prev_state_ids)
+
+    def get_cached_current_state_ids(self):
+        """Gets the current state IDs if we have them already cached.
+
+        Returns:
+            dict[(str, str), str]|None: Returns None if we haven't cached the
+            state or if state_group is None, which happens when the associated
+            event is an outlier.
+        """
+
+        return self._current_state_ids
+
+    @defer.inlineCallbacks
+    def _fill_out_state(self, store):
+        """Called to populate the _current_state_ids and _prev_state_ids
+        attributes by loading from the database.
+        """
+        if self.state_group is None:
+            return
+
+        self._current_state_ids = yield store.get_state_ids_for_group(
+            self.state_group,
+        )
+        if self._prev_state_id and self._event_state_key is not None:
+            self._prev_state_ids = dict(self._current_state_ids)
+
+            key = (self._event_type, self._event_state_key)
+            self._prev_state_ids[key] = self._prev_state_id
+        else:
+            self._prev_state_ids = self._current_state_ids
+
+    @defer.inlineCallbacks
+    def update_state(self, state_group, prev_state_ids, current_state_ids,
+                     prev_group, delta_ids):
+        """Replace the state in the context
+        """
+
+        # We need to make sure we wait for any ongoing fetching of state
+        # to complete so that the updated state doesn't get clobbered
+        if self._fetching_state_deferred:
+            yield make_deferred_yieldable(self._fetching_state_deferred)
+
+        self.state_group = state_group
+        self._prev_state_ids = prev_state_ids
+        self.prev_group = prev_group
+        self._current_state_ids = current_state_ids
+        self.delta_ids = delta_ids
+
+        # We need to ensure that that we've marked as having fetched the state
+        self._fetching_state_deferred = defer.succeed(None)
 
 
 def _encode_state_dict(state_dict):
@@ -159,7 +277,7 @@ def _encode_state_dict(state_dict):
 
     return [
         (etype, state_key, v)
-        for (etype, state_key), v in state_dict.iteritems()
+        for (etype, state_key), v in iteritems(state_dict)
     ]
 
 

synapse/events/utils.py

@@ -13,12 +13,15 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from synapse.api.constants import EventTypes
-from . import EventBase
+import re
+
+from six import string_types
 
 from frozendict import frozendict
 
-import re
+from synapse.api.constants import EventTypes
+
+from . import EventBase
 
 # Split strings on "." but not "\." This uses a negative lookbehind assertion for '\'
 # (?<!stuff) matches if the current position in the string is not preceded
@@ -277,7 +280,7 @@ 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

@@ -13,9 +13,11 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from synapse.types import EventID, RoomID, UserID
-from synapse.api.errors import SynapseError
+from six import string_types
+
 from synapse.api.constants import EventTypes, Membership
+from synapse.api.errors import SynapseError
+from synapse.types import EventID, RoomID, UserID
 
 
 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_base.py

@@ -13,17 +13,21 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 import logging
+from collections import namedtuple
 
 import six
 
-from synapse.api.constants import MAX_DEPTH
-from synapse.api.errors import SynapseError, Codes
+from twisted.internet import defer
+from twisted.internet.defer import DeferredList
+
+from synapse.api.constants import MAX_DEPTH, EventTypes, Membership
+from synapse.api.errors import Codes, SynapseError
 from synapse.crypto.event_signing import check_event_content_hash
 from synapse.events import FrozenEvent
 from synapse.events.utils import prune_event
-from synapse.http.servlet import assert_params_in_request
-from synapse.util import unwrapFirstError, logcontext
-from twisted.internet import defer
+from synapse.http.servlet import assert_params_in_dict
+from synapse.types import get_domain_from_id
+from synapse.util import logcontext, unwrapFirstError
 
 logger = logging.getLogger(__name__)
 
@@ -132,34 +136,25 @@ class FederationBase(object):
               * throws a SynapseError if the signature check failed.
             The deferreds run their callbacks in the sentinel logcontext.
         """
-
-        redacted_pdus = [
-            prune_event(pdu)
-            for pdu in pdus
-        ]
-
-        deferreds = self.keyring.verify_json_objects_for_server([
-            (p.origin, p.get_pdu_json())
-            for p in redacted_pdus
-        ])
+        deferreds = _check_sigs_on_pdus(self.keyring, pdus)
 
         ctx = logcontext.LoggingContext.current_context()
 
-        def callback(_, pdu, redacted):
+        def callback(_, pdu):
             with logcontext.PreserveLoggingContext(ctx):
                 if not check_event_content_hash(pdu):
                     logger.warn(
                         "Event content has been tampered, redacting %s: %s",
                         pdu.event_id, pdu.get_pdu_json()
                     )
-                    return redacted
+                    return prune_event(pdu)
 
                 if self.spam_checker.check_event_for_spam(pdu):
                     logger.warn(
                         "Event contains spam, redacting %s: %s",
                         pdu.event_id, pdu.get_pdu_json()
                     )
-                    return redacted
+                    return prune_event(pdu)
 
                 return pdu
 
@@ -172,16 +167,116 @@ class FederationBase(object):
                 )
             return failure
 
-        for deferred, pdu, redacted in zip(deferreds, pdus, redacted_pdus):
+        for deferred, pdu in zip(deferreds, pdus):
             deferred.addCallbacks(
                 callback, errback,
-                callbackArgs=[pdu, redacted],
+                callbackArgs=[pdu],
                 errbackArgs=[pdu],
             )
 
         return deferreds
 
 
+class PduToCheckSig(namedtuple("PduToCheckSig", [
+    "pdu", "redacted_pdu_json", "event_id_domain", "sender_domain", "deferreds",
+])):
+    pass
+
+
+def _check_sigs_on_pdus(keyring, pdus):
+    """Check that the given events are correctly signed
+
+    Args:
+        keyring (synapse.crypto.Keyring): keyring object to do the checks
+        pdus (Collection[EventBase]): the events to be checked
+
+    Returns:
+        List[Deferred]: a Deferred for each event in pdus, which will either succeed if
+           the signatures are valid, or fail (with a SynapseError) if not.
+    """
+
+    # (currently this is written assuming the v1 room structure; we'll probably want a
+    # separate function for checking v2 rooms)
+
+    # we want to check that the event is signed by:
+    #
+    # (a) the server which created the event_id
+    #
+    # (b) the sender's server.
+    #
+    #     - except in the case of invites created from a 3pid invite, which are exempt
+    #     from this check, because the sender has to match that of the original 3pid
+    #     invite, but the event may come from a different HS, for reasons that I don't
+    #     entirely grok (why do the senders have to match? and if they do, why doesn't the
+    #     joining server ask the inviting server to do the switcheroo with
+    #     exchange_third_party_invite?).
+    #
+    #     That's pretty awful, since redacting such an invite will render it invalid
+    #     (because it will then look like a regular invite without a valid signature),
+    #     and signatures are *supposed* to be valid whether or not an event has been
+    #     redacted. But this isn't the worst of the ways that 3pid invites are broken.
+    #
+    # let's start by getting the domain for each pdu, and flattening the event back
+    # to JSON.
+    pdus_to_check = [
+        PduToCheckSig(
+            pdu=p,
+            redacted_pdu_json=prune_event(p).get_pdu_json(),
+            event_id_domain=get_domain_from_id(p.event_id),
+            sender_domain=get_domain_from_id(p.sender),
+            deferreds=[],
+        )
+        for p in pdus
+    ]
+
+    # first make sure that the event is signed by the event_id's domain
+    deferreds = keyring.verify_json_objects_for_server([
+        (p.event_id_domain, p.redacted_pdu_json)
+        for p in pdus_to_check
+    ])
+
+    for p, d in zip(pdus_to_check, deferreds):
+        p.deferreds.append(d)
+
+    # now let's look for events where the sender's domain is different to the
+    # event id's domain (normally only the case for joins/leaves), and add additional
+    # checks.
+    pdus_to_check_sender = [
+        p for p in pdus_to_check
+        if p.sender_domain != p.event_id_domain and not _is_invite_via_3pid(p.pdu)
+    ]
+
+    more_deferreds = keyring.verify_json_objects_for_server([
+        (p.sender_domain, p.redacted_pdu_json)
+        for p in pdus_to_check_sender
+    ])
+
+    for p, d in zip(pdus_to_check_sender, more_deferreds):
+        p.deferreds.append(d)
+
+    # replace lists of deferreds with single Deferreds
+    return [_flatten_deferred_list(p.deferreds) for p in pdus_to_check]
+
+
+def _flatten_deferred_list(deferreds):
+    """Given a list of one or more deferreds, either return the single deferred, or
+    combine into a DeferredList.
+    """
+    if len(deferreds) > 1:
+        return DeferredList(deferreds, fireOnOneErrback=True, consumeErrors=True)
+    else:
+        assert len(deferreds) == 1
+        return deferreds[0]
+
+
+def _is_invite_via_3pid(event):
+    return (
+        event.type == EventTypes.Member
+        and event.membership == Membership.INVITE
+        and "third_party_invite" in event.content
+    )
+
+
 def event_from_pdu_json(pdu_json, outlier=False):
     """Construct a FrozenEvent from an event json received over federation
 
@@ -198,7 +293,7 @@ def event_from_pdu_json(pdu_json, outlier=False):
     """
     # we could probably enforce a bunch of other fields here (room_id, sender,
     # origin, etc etc)
-    assert_params_in_request(pdu_json, ('event_id', 'type', 'depth'))
+    assert_params_in_dict(pdu_json, ('event_id', 'type', 'depth'))
 
     depth = pdu_json['depth']
     if not isinstance(depth, six.integer_types):

synapse/federation/federation_client.py

@@ -21,18 +21,19 @@ import random
 
 from six.moves import range
 
+from prometheus_client import Counter
+
 from twisted.internet import defer
 
-from synapse.api.constants import Membership
+from synapse.api.constants import KNOWN_ROOM_VERSIONS, EventTypes, Membership
 from synapse.api.errors import (
-    CodeMessageException, HttpResponseException, SynapseError, FederationDeniedError
+    CodeMessageException,
+    FederationDeniedError,
+    HttpResponseException,
+    SynapseError,
 )
 from synapse.events import builder
-from synapse.federation.federation_base import (
-    FederationBase,
-    event_from_pdu_json,
-)
-import synapse.metrics
+from synapse.federation.federation_base import FederationBase, event_from_pdu_json
 from synapse.util import logcontext, unwrapFirstError
 from synapse.util.caches.expiringcache import ExpiringCache
 from synapse.util.logcontext import make_deferred_yieldable, run_in_background
@@ -41,16 +42,19 @@ from synapse.util.retryutils import NotRetryingDestination
 
 logger = logging.getLogger(__name__)
 
-
-# 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"])
+sent_queries_counter = Counter("synapse_federation_client_sent_queries", "", ["type"])
 
 
 PDU_RETRY_TIME_MS = 1 * 60 * 1000
 
 
+class InvalidResponseError(RuntimeError):
+    """Helper for _try_destination_list: indicates that the server returned a response
+    we couldn't parse
+    """
+    pass
+
+
 class FederationClient(FederationBase):
     def __init__(self, hs):
         super(FederationClient, self).__init__(hs)
@@ -108,7 +112,7 @@ 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,
@@ -127,7 +131,7 @@ 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
         )
@@ -137,7 +141,7 @@ 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
         )
@@ -154,7 +158,7 @@ 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
         )
@@ -394,7 +398,7 @@ 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 = []
@@ -461,8 +465,63 @@ class FederationClient(FederationBase):
         defer.returnValue(signed_auth)
 
     @defer.inlineCallbacks
+    def _try_destination_list(self, description, destinations, callback):
+        """Try an operation on a series of servers, until it succeeds
+
+        Args:
+            description (unicode): description of the operation we're doing, for logging
+
+            destinations (Iterable[unicode]): list of server_names to try
+
+            callback (callable):  Function to run for each server. Passed a single
+                argument: the server_name to try. May return a deferred.
+
+                If the callback raises a CodeMessageException with a 300/400 code,
+                attempts to perform the operation stop immediately and the exception is
+                reraised.
+
+                Otherwise, if the callback raises an Exception the error is logged and the
+                next server tried. Normally the stacktrace is logged but this is
+                suppressed if the exception is an InvalidResponseError.
+
+        Returns:
+            The [Deferred] result of callback, if it succeeds
+
+        Raises:
+            SynapseError if the chosen remote server returns a 300/400 code.
+
+            RuntimeError if no servers were reachable.
+        """
+        for destination in destinations:
+            if destination == self.server_name:
+                continue
+
+            try:
+                res = yield callback(destination)
+                defer.returnValue(res)
+            except InvalidResponseError as e:
+                logger.warn(
+                    "Failed to %s via %s: %s",
+                    description, destination, e,
+                )
+            except HttpResponseException as e:
+                if not 500 <= e.code < 600:
+                    raise e.to_synapse_error()
+                else:
+                    logger.warn(
+                        "Failed to %s via %s: %i %s",
+                        description, destination, e.code, e.message,
+                    )
+            except Exception:
+                logger.warn(
+                    "Failed to %s via %s",
+                    description, destination, exc_info=1,
+                )
+
+        raise RuntimeError("Failed to %s via any server" % (description, ))
+
     def make_membership_event(self, destinations, room_id, user_id, membership,
-                              content={},):
+                              content, params):
         """
         Creates an m.room.member event, with context, without participating in the room.
 
@@ -478,13 +537,15 @@ class FederationClient(FederationBase):
             user_id (str): The user whose membership is being evented.
             membership (str): The "membership" property of the event. Must be
                 one of "join" or "leave".
-            content (object): Any additional data to put into the content field
+            content (dict): Any additional data to put into the content field
                 of the event.
+            params (dict[str, str|Iterable[str]]): Query parameters to include in the
+                request.
         Return:
             Deferred: resolves to a tuple of (origin (str), event (object))
             where origin is the remote homeserver which generated the event.
 
-            Fails with a ``CodeMessageException`` if the chosen remote server
+            Fails with a ``SynapseError`` if the chosen remote server
             returns a 300/400 code.
 
             Fails with a ``RuntimeError`` if no servers were reachable.
@@ -495,50 +556,37 @@ class FederationClient(FederationBase):
                 "make_membership_event called with membership='%s', must be one of %s" %
                 (membership, ",".join(valid_memberships))
             )
-        for destination in destinations:
-            if destination == self.server_name:
-                continue
 
-            try:
-                ret = yield self.transport_layer.make_membership_event(
-                    destination, room_id, user_id, membership
-                )
+        @defer.inlineCallbacks
+        def send_request(destination):
+            ret = yield self.transport_layer.make_membership_event(
+                destination, room_id, user_id, membership, params,
+            )
 
-                pdu_dict = ret["event"]
+            pdu_dict = ret.get("event", None)
+            if not isinstance(pdu_dict, dict):
+                raise InvalidResponseError("Bad 'event' field in response")
 
-                logger.debug("Got response to make_%s: %s", membership, pdu_dict)
+            logger.debug("Got response to make_%s: %s", membership, pdu_dict)
 
-                pdu_dict["content"].update(content)
+            pdu_dict["content"].update(content)
 
-                # The protoevent received over the JSON wire may not have all
-                # the required fields. Lets just gloss over that because
-                # there's some we never care about
-                if "prev_state" not in pdu_dict:
-                    pdu_dict["prev_state"] = []
+            # The protoevent received over the JSON wire may not have all
+            # the required fields. Lets just gloss over that because
+            # there's some we never care about
+            if "prev_state" not in pdu_dict:
+                pdu_dict["prev_state"] = []
 
-                ev = builder.EventBuilder(pdu_dict)
+            ev = builder.EventBuilder(pdu_dict)
 
-                defer.returnValue(
-                    (destination, ev)
-                )
-                break
-            except CodeMessageException as e:
-                if not 500 <= e.code < 600:
-                    raise
-                else:
-                    logger.warn(
-                        "Failed to make_%s via %s: %s",
-                        membership, destination, e.message
-                    )
-            except Exception as e:
-                logger.warn(
-                    "Failed to make_%s via %s: %s",
-                    membership, destination, e.message
-                )
+            defer.returnValue(
+                (destination, ev)
+            )
 
-        raise RuntimeError("Failed to send to any server.")
+        return self._try_destination_list(
+            "make_" + membership, destinations, send_request,
+        )
 
-    @defer.inlineCallbacks
     def send_join(self, destinations, pdu):
         """Sends a join event to one of a list of homeservers.
 
@@ -555,103 +603,111 @@ class FederationClient(FederationBase):
             giving the serer the event was sent to, ``state`` (?) and
             ``auth_chain``.
 
-            Fails with a ``CodeMessageException`` if the chosen remote server
+            Fails with a ``SynapseError`` if the chosen remote server
             returns a 300/400 code.
 
             Fails with a ``RuntimeError`` if no servers were reachable.
         """
 
-        for destination in destinations:
-            if destination == self.server_name:
-                continue
-
-            try:
-                time_now = self._clock.time_msec()
-                _, content = yield self.transport_layer.send_join(
-                    destination=destination,
-                    room_id=pdu.room_id,
-                    event_id=pdu.event_id,
-                    content=pdu.get_pdu_json(time_now),
+        def check_authchain_validity(signed_auth_chain):
+            for e in signed_auth_chain:
+                if e.type == EventTypes.Create:
+                    create_event = e
+                    break
+            else:
+                raise InvalidResponseError(
+                    "no %s in auth chain" % (EventTypes.Create,),
                 )
 
-                logger.debug("Got content: %s", content)
+            # the room version should be sane.
+            room_version = create_event.content.get("room_version", "1")
+            if room_version not in KNOWN_ROOM_VERSIONS:
+                # This shouldn't be possible, because the remote server should have
+                # rejected the join attempt during make_join.
+                raise InvalidResponseError(
+                    "room appears to have unsupported version %s" % (
+                        room_version,
+                    ))
 
-                state = [
-                    event_from_pdu_json(p, outlier=True)
-                    for p in content.get("state", [])
-                ]
+        @defer.inlineCallbacks
+        def send_request(destination):
+            time_now = self._clock.time_msec()
+            _, content = yield self.transport_layer.send_join(
+                destination=destination,
+                room_id=pdu.room_id,
+                event_id=pdu.event_id,
+                content=pdu.get_pdu_json(time_now),
+            )
 
-                auth_chain = [
-                    event_from_pdu_json(p, outlier=True)
-                    for p in content.get("auth_chain", [])
-                ]
+            logger.debug("Got content: %s", content)
 
-                pdus = {
-                    p.event_id: p
-                    for p in itertools.chain(state, auth_chain)
-                }
+            state = [
+                event_from_pdu_json(p, outlier=True)
+                for p in content.get("state", [])
+            ]
 
-                valid_pdus = yield self._check_sigs_and_hash_and_fetch(
-                    destination, pdus.values(),
-                    outlier=True,
-                )
+            auth_chain = [
+                event_from_pdu_json(p, outlier=True)
+                for p in content.get("auth_chain", [])
+            ]
 
-                valid_pdus_map = {
-                    p.event_id: p
-                    for p in valid_pdus
-                }
+            pdus = {
+                p.event_id: p
+                for p in itertools.chain(state, auth_chain)
+            }
 
-                # NB: We *need* to copy to ensure that we don't have multiple
-                # references being passed on, as that causes... issues.
-                signed_state = [
-                    copy.copy(valid_pdus_map[p.event_id])
-                    for p in state
-                    if p.event_id in valid_pdus_map
-                ]
+            valid_pdus = yield self._check_sigs_and_hash_and_fetch(
+                destination, list(pdus.values()),
+                outlier=True,
+            )
 
-                signed_auth = [
-                    valid_pdus_map[p.event_id]
-                    for p in auth_chain
-                    if p.event_id in valid_pdus_map
-                ]
+            valid_pdus_map = {
+                p.event_id: p
+                for p in valid_pdus
+            }
 
-                # NB: We *need* to copy to ensure that we don't have multiple
-                # references being passed on, as that causes... issues.
-                for s in signed_state:
-                    s.internal_metadata = copy.deepcopy(s.internal_metadata)
+            # NB: We *need* to copy to ensure that we don't have multiple
+            # references being passed on, as that causes... issues.
+            signed_state = [
+                copy.copy(valid_pdus_map[p.event_id])
+                for p in state
+                if p.event_id in valid_pdus_map
+            ]
 
-                auth_chain.sort(key=lambda e: e.depth)
+            signed_auth = [
+                valid_pdus_map[p.event_id]
+                for p in auth_chain
+                if p.event_id in valid_pdus_map
+            ]
 
-                defer.returnValue({
-                    "state": signed_state,
-                    "auth_chain": signed_auth,
-                    "origin": destination,
-                })
-            except CodeMessageException as e:
-                if not 500 <= e.code < 600:
-                    raise
-                else:
-                    logger.exception(
-                        "Failed to send_join via %s: %s",
-                        destination, e.message
-                    )
-            except Exception as e:
-                logger.exception(
-                    "Failed to send_join via %s: %s",
-                    destination, e.message
-                )
+            # NB: We *need* to copy to ensure that we don't have multiple
+            # references being passed on, as that causes... issues.
+            for s in signed_state:
+                s.internal_metadata = copy.deepcopy(s.internal_metadata)
 
-        raise RuntimeError("Failed to send to any server.")
+            check_authchain_validity(signed_auth)
+
+            defer.returnValue({
+                "state": signed_state,
+                "auth_chain": signed_auth,
+                "origin": destination,
+            })
+        return self._try_destination_list("send_join", destinations, send_request)
 
     @defer.inlineCallbacks
     def send_invite(self, destination, room_id, event_id, pdu):
         time_now = self._clock.time_msec()
-        code, content = yield self.transport_layer.send_invite(
-            destination=destination,
-            room_id=room_id,
-            event_id=event_id,
-            content=pdu.get_pdu_json(time_now),
-        )
+        try:
+            code, content = yield self.transport_layer.send_invite(
+                destination=destination,
+                room_id=room_id,
+                event_id=event_id,
+                content=pdu.get_pdu_json(time_now),
+            )
+        except HttpResponseException as e:
+            if e.code == 403:
+                raise e.to_synapse_error()
+            raise
 
         pdu_dict = content["event"]
 
@@ -666,7 +722,6 @@ class FederationClient(FederationBase):
 
         defer.returnValue(pdu)
 
-    @defer.inlineCallbacks
     def send_leave(self, destinations, pdu):
         """Sends a leave event to one of a list of homeservers.
 
@@ -683,35 +738,25 @@ class FederationClient(FederationBase):
         Return:
             Deferred: resolves to None.
 
-            Fails with a ``CodeMessageException`` if the chosen remote server
-            returns a non-200 code.
+            Fails with a ``SynapseError`` if the chosen remote server
+            returns a 300/400 code.
 
             Fails with a ``RuntimeError`` if no servers were reachable.
         """
-        for destination in destinations:
-            if destination == self.server_name:
-                continue
+        @defer.inlineCallbacks
+        def send_request(destination):
+            time_now = self._clock.time_msec()
+            _, content = yield self.transport_layer.send_leave(
+                destination=destination,
+                room_id=pdu.room_id,
+                event_id=pdu.event_id,
+                content=pdu.get_pdu_json(time_now),
+            )
 
-            try:
-                time_now = self._clock.time_msec()
-                _, content = yield self.transport_layer.send_leave(
-                    destination=destination,
-                    room_id=pdu.room_id,
-                    event_id=pdu.event_id,
-                    content=pdu.get_pdu_json(time_now),
-                )
+            logger.debug("Got content: %s", content)
+            defer.returnValue(None)
 
-                logger.debug("Got content: %s", content)
-                defer.returnValue(None)
-            except CodeMessageException:
-                raise
-            except Exception as e:
-                logger.exception(
-                    "Failed to send_leave via %s: %s",
-                    destination, e.message
-                )
-
-        raise RuntimeError("Failed to send to any server.")
+        return self._try_destination_list("send_leave", destinations, send_request)
 
     def get_public_rooms(self, destination, limit=None, since_token=None,
                          search_filter=None, include_all_networks=False,

synapse/federation/federation_server.py

@@ -14,52 +14,65 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 import logging
+import re
+
+import six
+from six import iteritems
+
+from canonicaljson import json
+from prometheus_client import Counter
 
-import simplejson as json
 from twisted.internet import defer
+from twisted.internet.abstract import isIPAddress
+from twisted.python import failure
 
-from synapse.api.errors import AuthError, FederationError, SynapseError, NotFoundError
-from synapse.crypto.event_signing import compute_event_signature
-from synapse.federation.federation_base import (
-    FederationBase,
-    event_from_pdu_json,
+from synapse.api.constants import EventTypes
+from synapse.api.errors import (
+    AuthError,
+    FederationError,
+    IncompatibleRoomVersionError,
+    NotFoundError,
+    SynapseError,
 )
-
+from synapse.crypto.event_signing import compute_event_signature
+from synapse.federation.federation_base import FederationBase, event_from_pdu_json
 from synapse.federation.persistence import TransactionActions
 from synapse.federation.units import Edu, Transaction
-import synapse.metrics
+from synapse.http.endpoint import parse_server_name
+from synapse.replication.http.federation import (
+    ReplicationFederationSendEduRestServlet,
+    ReplicationGetQueryRestServlet,
+)
 from synapse.types import get_domain_from_id
-from synapse.util import async
+from synapse.util.async_helpers import Linearizer, concurrently_execute
 from synapse.util.caches.response_cache import ResponseCache
 from synapse.util.logutils import log_function
 
-from six import iteritems
-
 # when processing incoming transactions, we try to handle multiple rooms in
 # parallel, up to this limit.
 TRANSACTION_CONCURRENCY_LIMIT = 10
 
 logger = logging.getLogger(__name__)
 
-# synapse.federation.federation_server is a silly name
-metrics = synapse.metrics.get_metrics_for("synapse.federation.server")
+received_pdus_counter = Counter("synapse_federation_server_received_pdus", "")
 
-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 = metrics.register_counter("received_queries", labels=["type"])
+received_queries_counter = Counter(
+    "synapse_federation_server_received_queries", "", ["type"]
+)
 
 
 class FederationServer(FederationBase):
+
     def __init__(self, hs):
         super(FederationServer, self).__init__(hs)
 
         self.auth = hs.get_auth()
         self.handler = hs.get_handlers().federation_handler
 
-        self._server_linearizer = async.Linearizer("fed_server")
-        self._transaction_linearizer = async.Linearizer("fed_txn_handler")
+        self._server_linearizer = Linearizer("fed_server")
+        self._transaction_linearizer = Linearizer("fed_txn_handler")
 
         self.transaction_actions = TransactionActions(self.store)
 
@@ -73,6 +86,9 @@ class FederationServer(FederationBase):
     @log_function
     def on_backfill_request(self, origin, room_id, versions, limit):
         with (yield self._server_linearizer.queue((origin, room_id))):
+            origin_host, _ = parse_server_name(origin)
+            yield self.check_server_matches_acl(origin_host, room_id)
+
             pdus = yield self.handler.on_backfill_request(
                 origin, room_id, versions, limit
             )
@@ -83,7 +99,7 @@ class FederationServer(FederationBase):
 
     @defer.inlineCallbacks
     @log_function
-    def on_incoming_transaction(self, transaction_data):
+    def on_incoming_transaction(self, origin, transaction_data):
         # keep this as early as possible to make the calculated origin ts as
         # accurate as possible.
         request_time = self._clock.time_msec()
@@ -92,34 +108,33 @@ class FederationServer(FederationBase):
 
         if not transaction.transaction_id:
             raise Exception("Transaction missing transaction_id")
-        if not transaction.origin:
-            raise Exception("Transaction missing origin")
 
         logger.debug("[%s] Got transaction", transaction.transaction_id)
 
         # use a linearizer to ensure that we don't process the same transaction
         # multiple times in parallel.
         with (yield self._transaction_linearizer.queue(
-                (transaction.origin, transaction.transaction_id),
+                (origin, transaction.transaction_id),
         )):
             result = yield self._handle_incoming_transaction(
-                transaction, request_time,
+                origin, transaction, request_time,
             )
 
         defer.returnValue(result)
 
     @defer.inlineCallbacks
-    def _handle_incoming_transaction(self, transaction, request_time):
+    def _handle_incoming_transaction(self, origin, transaction, request_time):
         """ Process an incoming transaction and return the HTTP response
 
         Args:
+            origin (unicode): the server making the request
             transaction (Transaction): incoming transaction
             request_time (int): timestamp that the HTTP request arrived at
 
         Returns:
             Deferred[(int, object)]: http response code and body
         """
-        response = yield self.transaction_actions.have_responded(transaction)
+        response = yield self.transaction_actions.have_responded(origin, transaction)
 
         if response:
             logger.debug(
@@ -131,7 +146,9 @@ 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))
+
+        origin_host, _ = parse_server_name(origin)
 
         pdus_by_room = {}
 
@@ -153,24 +170,40 @@ class FederationServer(FederationBase):
         # we can process different rooms in parallel (which is useful if they
         # require callouts to other servers to fetch missing events), but
         # impose a limit to avoid going too crazy with ram/cpu.
+
         @defer.inlineCallbacks
         def process_pdus_for_room(room_id):
             logger.debug("Processing PDUs for %s", room_id)
+            try:
+                yield self.check_server_matches_acl(origin_host, room_id)
+            except AuthError as e:
+                logger.warn(
+                    "Ignoring PDUs for room %s from banned server", room_id,
+                )
+                for pdu in pdus_by_room[room_id]:
+                    event_id = pdu.event_id
+                    pdu_results[event_id] = e.error_dict()
+                return
+
             for pdu in pdus_by_room[room_id]:
                 event_id = pdu.event_id
                 try:
                     yield self._handle_received_pdu(
-                        transaction.origin, pdu
+                        origin, pdu
                     )
                     pdu_results[event_id] = {}
                 except FederationError as e:
                     logger.warn("Error handling PDU %s: %s", event_id, e)
                     pdu_results[event_id] = {"error": str(e)}
                 except Exception as e:
+                    f = failure.Failure()
                     pdu_results[event_id] = {"error": str(e)}
-                    logger.exception("Failed to handle PDU %s", event_id)
+                    logger.error(
+                        "Failed to handle PDU %s: %s",
+                        event_id, f.getTraceback().rstrip(),
+                    )
 
-        yield async.concurrently_execute(
+        yield concurrently_execute(
             process_pdus_for_room, pdus_by_room.keys(),
             TRANSACTION_CONCURRENCY_LIMIT,
         )
@@ -178,15 +211,11 @@ class FederationServer(FederationBase):
         if hasattr(transaction, "edus"):
             for edu in (Edu(**x) for x in transaction.edus):
                 yield self.received_edu(
-                    transaction.origin,
+                    origin,
                     edu.edu_type,
                     edu.content
                 )
 
-        pdu_failures = getattr(transaction, "pdu_failures", [])
-        for failure in pdu_failures:
-            logger.info("Got failure %r", failure)
-
         response = {
             "pdus": pdu_results,
         }
@@ -194,6 +223,7 @@ class FederationServer(FederationBase):
         logger.debug("Returning: %s", str(response))
 
         yield self.transaction_actions.set_response(
+            origin,
             transaction,
             200, response
         )
@@ -210,6 +240,9 @@ class FederationServer(FederationBase):
         if not event_id:
             raise NotImplementedError("Specify an event")
 
+        origin_host, _ = parse_server_name(origin)
+        yield self.check_server_matches_acl(origin_host, room_id)
+
         in_room = yield self.auth.check_host_in_room(room_id, origin)
         if not in_room:
             raise AuthError(403, "Host not in room.")
@@ -233,6 +266,9 @@ class FederationServer(FederationBase):
         if not event_id:
             raise NotImplementedError("Specify an event")
 
+        origin_host, _ = parse_server_name(origin)
+        yield self.check_server_matches_acl(origin_host, room_id)
+
         in_room = yield self.auth.check_host_in_room(room_id, origin)
         if not in_room:
             raise AuthError(403, "Host not in room.")
@@ -276,7 +312,7 @@ class FederationServer(FederationBase):
     @defer.inlineCallbacks
     @log_function
     def on_pdu_request(self, origin, event_id):
-        pdu = yield self._get_persisted_pdu(origin, event_id)
+        pdu = yield self.handler.get_persisted_pdu(origin, event_id)
 
         if pdu:
             defer.returnValue(
@@ -292,19 +328,32 @@ 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))
 
     @defer.inlineCallbacks
-    def on_make_join_request(self, room_id, user_id):
+    def on_make_join_request(self, origin, room_id, user_id, supported_versions):
+        origin_host, _ = parse_server_name(origin)
+        yield self.check_server_matches_acl(origin_host, room_id)
+
+        room_version = yield self.store.get_room_version(room_id)
+        if room_version not in supported_versions:
+            logger.warn("Room version %s not in %s", room_version, supported_versions)
+            raise IncompatibleRoomVersionError(room_version=room_version)
+
         pdu = yield self.handler.on_make_join_request(room_id, user_id)
         time_now = self._clock.time_msec()
-        defer.returnValue({"event": pdu.get_pdu_json(time_now)})
+        defer.returnValue({
+            "event": pdu.get_pdu_json(time_now),
+            "room_version": room_version,
+        })
 
     @defer.inlineCallbacks
     def on_invite_request(self, origin, content):
         pdu = event_from_pdu_json(content)
+        origin_host, _ = parse_server_name(origin)
+        yield self.check_server_matches_acl(origin_host, pdu.room_id)
         ret_pdu = yield self.handler.on_invite_request(origin, pdu)
         time_now = self._clock.time_msec()
         defer.returnValue((200, {"event": ret_pdu.get_pdu_json(time_now)}))
@@ -313,6 +362,10 @@ class FederationServer(FederationBase):
     def on_send_join_request(self, origin, content):
         logger.debug("on_send_join_request: content: %s", content)
         pdu = event_from_pdu_json(content)
+
+        origin_host, _ = parse_server_name(origin)
+        yield self.check_server_matches_acl(origin_host, pdu.room_id)
+
         logger.debug("on_send_join_request: pdu sigs: %s", pdu.signatures)
         res_pdus = yield self.handler.on_send_join_request(origin, pdu)
         time_now = self._clock.time_msec()
@@ -324,7 +377,9 @@ class FederationServer(FederationBase):
         }))
 
     @defer.inlineCallbacks
-    def on_make_leave_request(self, room_id, user_id):
+    def on_make_leave_request(self, origin, room_id, user_id):
+        origin_host, _ = parse_server_name(origin)
+        yield self.check_server_matches_acl(origin_host, room_id)
         pdu = yield self.handler.on_make_leave_request(room_id, user_id)
         time_now = self._clock.time_msec()
         defer.returnValue({"event": pdu.get_pdu_json(time_now)})
@@ -333,6 +388,10 @@ class FederationServer(FederationBase):
     def on_send_leave_request(self, origin, content):
         logger.debug("on_send_leave_request: content: %s", content)
         pdu = event_from_pdu_json(content)
+
+        origin_host, _ = parse_server_name(origin)
+        yield self.check_server_matches_acl(origin_host, pdu.room_id)
+
         logger.debug("on_send_leave_request: pdu sigs: %s", pdu.signatures)
         yield self.handler.on_send_leave_request(origin, pdu)
         defer.returnValue((200, {}))
@@ -340,6 +399,9 @@ class FederationServer(FederationBase):
     @defer.inlineCallbacks
     def on_event_auth(self, origin, room_id, event_id):
         with (yield self._server_linearizer.queue((origin, room_id))):
+            origin_host, _ = parse_server_name(origin)
+            yield self.check_server_matches_acl(origin_host, room_id)
+
             time_now = self._clock.time_msec()
             auth_pdus = yield self.handler.on_event_auth(event_id)
             res = {
@@ -368,6 +430,9 @@ class FederationServer(FederationBase):
             Deferred: Results in `dict` with the same format as `content`
         """
         with (yield self._server_linearizer.queue((origin, room_id))):
+            origin_host, _ = parse_server_name(origin)
+            yield self.check_server_matches_acl(origin_host, room_id)
+
             auth_chain = [
                 event_from_pdu_json(e)
                 for e in content["auth_chain"]
@@ -380,6 +445,7 @@ class FederationServer(FederationBase):
             ret = yield self.handler.on_query_auth(
                 origin,
                 event_id,
+                room_id,
                 signed_auth,
                 content.get("rejects", []),
                 content.get("missing", []),
@@ -441,6 +507,9 @@ class FederationServer(FederationBase):
     def on_get_missing_events(self, origin, room_id, earliest_events,
                               latest_events, limit, min_depth):
         with (yield self._server_linearizer.queue((origin, room_id))):
+            origin_host, _ = parse_server_name(origin)
+            yield self.check_server_matches_acl(origin_host, room_id)
+
             logger.info(
                 "on_get_missing_events: earliest_events: %r, latest_events: %r,"
                 " limit: %d, min_depth: %d",
@@ -469,17 +538,6 @@ class FederationServer(FederationBase):
         ts_now_ms = self._clock.time_msec()
         return self.store.get_user_id_for_open_id_token(token, ts_now_ms)
 
-    @log_function
-    def _get_persisted_pdu(self, origin, event_id, do_auth=True):
-        """ Get a PDU from the database with given origin and id.
-
-        Returns:
-            Deferred: Results in a `Pdu`.
-        """
-        return self.handler.get_persisted_pdu(
-            origin, event_id, do_auth=do_auth
-        )
-
     def _transaction_from_pdus(self, pdu_list):
         """Returns a new Transaction containing the given PDUs suitable for
         transmission.
@@ -559,7 +617,9 @@ class FederationServer(FederationBase):
                 affected=pdu.event_id,
             )
 
-        yield self.handler.on_receive_pdu(origin, pdu, get_missing=True)
+        yield self.handler.on_receive_pdu(
+            origin, pdu, get_missing=True, sent_to_us_directly=True,
+        )
 
     def __str__(self):
         return "<ReplicationLayer(%s)>" % self.server_name
@@ -587,6 +647,101 @@ class FederationServer(FederationBase):
         )
         defer.returnValue(ret)
 
+    @defer.inlineCallbacks
+    def check_server_matches_acl(self, server_name, room_id):
+        """Check if the given server is allowed by the server ACLs in the room
+
+        Args:
+            server_name (str): name of server, *without any port part*
+            room_id (str): ID of the room to check
+
+        Raises:
+            AuthError if the server does not match the ACL
+        """
+        state_ids = yield self.store.get_current_state_ids(room_id)
+        acl_event_id = state_ids.get((EventTypes.ServerACL, ""))
+
+        if not acl_event_id:
+            return
+
+        acl_event = yield self.store.get_event(acl_event_id)
+        if server_matches_acl_event(server_name, acl_event):
+            return
+
+        raise AuthError(code=403, msg="Server is banned from room")
+
+
+def server_matches_acl_event(server_name, acl_event):
+    """Check if the given server is allowed by the ACL event
+
+    Args:
+        server_name (str): name of server, without any port part
+        acl_event (EventBase): m.room.server_acl event
+
+    Returns:
+        bool: True if this server is allowed by the ACLs
+    """
+    logger.debug("Checking %s against acl %s", server_name, acl_event.content)
+
+    # first of all, check if literal IPs are blocked, and if so, whether the
+    # server name is a literal IP
+    allow_ip_literals = acl_event.content.get("allow_ip_literals", True)
+    if not isinstance(allow_ip_literals, bool):
+        logger.warn("Ignorning non-bool allow_ip_literals flag")
+        allow_ip_literals = True
+    if not allow_ip_literals:
+        # check for ipv6 literals. These start with '['.
+        if server_name[0] == '[':
+            return False
+
+        # check for ipv4 literals. We can just lift the routine from twisted.
+        if isIPAddress(server_name):
+            return False
+
+    # next,  check the deny list
+    deny = acl_event.content.get("deny", [])
+    if not isinstance(deny, (list, tuple)):
+        logger.warn("Ignorning non-list deny ACL %s", deny)
+        deny = []
+    for e in deny:
+        if _acl_entry_matches(server_name, e):
+            # logger.info("%s matched deny rule %s", server_name, e)
+            return False
+
+    # then the allow list.
+    allow = acl_event.content.get("allow", [])
+    if not isinstance(allow, (list, tuple)):
+        logger.warn("Ignorning non-list allow ACL %s", allow)
+        allow = []
+    for e in allow:
+        if _acl_entry_matches(server_name, e):
+            # logger.info("%s matched allow rule %s", server_name, e)
+            return True
+
+    # everything else should be rejected.
+    # logger.info("%s fell through", server_name)
+    return False
+
+
+def _acl_entry_matches(server_name, acl_entry):
+    if not isinstance(acl_entry, six.string_types):
+        logger.warn("Ignoring non-str ACL entry '%s' (is %s)", acl_entry, type(acl_entry))
+        return False
+    regex = _glob_to_regex(acl_entry)
+    return regex.match(server_name)
+
+
+def _glob_to_regex(glob):
+    res = ''
+    for c in glob:
+        if c == '*':
+            res = res + '.*'
+        elif c == '?':
+            res = res + '.'
+        else:
+            res = res + re.escape(c)
+    return re.compile(res + "\\Z", re.IGNORECASE)
+
 
 class FederationHandlerRegistry(object):
     """Allows classes to register themselves as handlers for a given EDU or
@@ -609,6 +764,8 @@ class FederationHandlerRegistry(object):
         if edu_type in self.edu_handlers:
             raise KeyError("Already have an EDU handler for %s" % (edu_type,))
 
+        logger.info("Registering federation EDU handler for %r", edu_type)
+
         self.edu_handlers[edu_type] = handler
 
     def register_query_handler(self, query_type, handler):
@@ -627,6 +784,8 @@ class FederationHandlerRegistry(object):
                 "Already have a Query handler for %s" % (query_type,)
             )
 
+        logger.info("Registering federation query handler for %r", query_type)
+
         self.query_handlers[query_type] = handler
 
     @defer.inlineCallbacks
@@ -649,3 +808,49 @@ class FederationHandlerRegistry(object):
             raise NotFoundError("No handler for Query type '%s'" % (query_type,))
 
         return handler(args)
+
+
+class ReplicationFederationHandlerRegistry(FederationHandlerRegistry):
+    """A FederationHandlerRegistry for worker processes.
+
+    When receiving EDU or queries it will check if an appropriate handler has
+    been registered on the worker, if there isn't one then it calls off to the
+    master process.
+    """
+
+    def __init__(self, hs):
+        self.config = hs.config
+        self.http_client = hs.get_simple_http_client()
+        self.clock = hs.get_clock()
+
+        self._get_query_client = ReplicationGetQueryRestServlet.make_client(hs)
+        self._send_edu = ReplicationFederationSendEduRestServlet.make_client(hs)
+
+        super(ReplicationFederationHandlerRegistry, self).__init__()
+
+    def on_edu(self, edu_type, origin, content):
+        """Overrides FederationHandlerRegistry
+        """
+        handler = self.edu_handlers.get(edu_type)
+        if handler:
+            return super(ReplicationFederationHandlerRegistry, self).on_edu(
+                edu_type, origin, content,
+            )
+
+        return self._send_edu(
+                edu_type=edu_type,
+                origin=origin,
+                content=content,
+        )
+
+    def on_query(self, query_type, args):
+        """Overrides FederationHandlerRegistry
+        """
+        handler = self.query_handlers.get(query_type)
+        if handler:
+            return handler(args)
+
+        return self._get_query_client(
+                query_type=query_type,
+                args=args,
+        )

synapse/federation/persistence.py

@@ -19,13 +19,12 @@ package.
 These actions are mostly only used by the :py:mod:`.replication` module.
 """
 
+import logging
+
 from twisted.internet import defer
 
 from synapse.util.logutils import log_function
 
-import logging
-
-
 logger = logging.getLogger(__name__)
 
 
@@ -37,7 +36,7 @@ class TransactionActions(object):
         self.store = datastore
 
     @log_function
-    def have_responded(self, transaction):
+    def have_responded(self, origin, transaction):
         """ Have we already responded to a transaction with the same id and
         origin?
 
@@ -51,11 +50,11 @@ class TransactionActions(object):
                                "transaction_id")
 
         return self.store.get_received_txn_response(
-            transaction.transaction_id, transaction.origin
+            transaction.transaction_id, origin
         )
 
     @log_function
-    def set_response(self, transaction, code, response):
+    def set_response(self, origin, transaction, code, response):
         """ Persist how we responded to a transaction.
 
         Returns:
@@ -67,7 +66,7 @@ class TransactionActions(object):
 
         return self.store.set_received_txn_response(
             transaction.transaction_id,
-            transaction.origin,
+            origin,
             code,
             response,
         )

synapse/federation/send_queue.py

@@ -29,25 +29,22 @@ dead worker doesn't cause the queues to grow limitlessly.
 Events are replicated via a separate events stream.
 """
 
-from .units import Edu
-
-from synapse.storage.presence import UserPresenceState
-from synapse.util.metrics import Measure
-import synapse.metrics
-
-from blist import sorteddict
+import logging
 from collections import namedtuple
 
-import logging
+from six import iteritems, itervalues
 
-from six import itervalues, iteritems
+from sortedcontainers import SortedDict
+
+from synapse.metrics import LaterGauge
+from synapse.storage.presence import UserPresenceState
+from synapse.util.metrics import Measure
+
+from .units import Edu
 
 logger = logging.getLogger(__name__)
 
 
-metrics = synapse.metrics.get_metrics_for(__name__)
-
-
 class FederationRemoteSendQueue(object):
     """A drop in replacement for TransactionQueue"""
 
@@ -58,33 +55,29 @@ class FederationRemoteSendQueue(object):
         self.is_mine_id = hs.is_mine_id
 
         self.presence_map = {}  # Pending presence map user_id -> UserPresenceState
-        self.presence_changed = sorteddict()  # Stream position -> user_id
+        self.presence_changed = SortedDict()  # Stream position -> user_id
 
         self.keyed_edu = {}  # (destination, key) -> EDU
-        self.keyed_edu_changed = sorteddict()  # stream position -> (destination, key)
+        self.keyed_edu_changed = SortedDict()  # stream position -> (destination, key)
 
-        self.edus = sorteddict()  # stream position -> Edu
+        self.edus = SortedDict()  # stream position -> Edu
 
-        self.failures = sorteddict()  # stream position -> (destination, Failure)
-
-        self.device_messages = sorteddict()  # stream position -> destination
+        self.device_messages = SortedDict()  # stream position -> destination
 
         self.pos = 1
-        self.pos_time = sorteddict()
+        self.pos_time = SortedDict()
 
         # EVERYTHING IS SAD. In particular, python only makes new scopes when
         # we make a new function, so we need to make a new function so the inner
         # lambda binds to the queue rather than to the name of the queue which
         # changes. ARGH.
         def register(name, queue):
-            metrics.register_callback(
-                queue_name + "_size",
-                lambda: len(queue),
-            )
+            LaterGauge("synapse_federation_send_queue_%s_size" % (queue_name,),
+                       "", [], lambda: len(queue))
 
         for queue_name in [
             "presence_map", "presence_changed", "keyed_edu", "keyed_edu_changed",
-            "edus", "failures", "device_messages", "pos_time",
+            "edus", "device_messages", "pos_time",
         ]:
             register(queue_name, getattr(self, queue_name))
 
@@ -103,7 +96,7 @@ class FederationRemoteSendQueue(object):
         now = self.clock.time_msec()
 
         keys = self.pos_time.keys()
-        time = keys.bisect_left(now - FIVE_MINUTES_AGO)
+        time = self.pos_time.bisect_left(now - FIVE_MINUTES_AGO)
         if not keys[:time]:
             return
 
@@ -118,7 +111,7 @@ class FederationRemoteSendQueue(object):
         with Measure(self.clock, "send_queue._clear"):
             # Delete things out of presence maps
             keys = self.presence_changed.keys()
-            i = keys.bisect_left(position_to_delete)
+            i = self.presence_changed.bisect_left(position_to_delete)
             for key in keys[:i]:
                 del self.presence_changed[key]
 
@@ -136,7 +129,7 @@ class FederationRemoteSendQueue(object):
 
             # Delete things out of keyed edus
             keys = self.keyed_edu_changed.keys()
-            i = keys.bisect_left(position_to_delete)
+            i = self.keyed_edu_changed.bisect_left(position_to_delete)
             for key in keys[:i]:
                 del self.keyed_edu_changed[key]
 
@@ -150,19 +143,13 @@ class FederationRemoteSendQueue(object):
 
             # Delete things out of edu map
             keys = self.edus.keys()
-            i = keys.bisect_left(position_to_delete)
+            i = self.edus.bisect_left(position_to_delete)
             for key in keys[:i]:
                 del self.edus[key]
 
-            # Delete things out of failure map
-            keys = self.failures.keys()
-            i = keys.bisect_left(position_to_delete)
-            for key in keys[:i]:
-                del self.failures[key]
-
             # Delete things out of device map
             keys = self.device_messages.keys()
-            i = keys.bisect_left(position_to_delete)
+            i = self.device_messages.bisect_left(position_to_delete)
             for key in keys[:i]:
                 del self.device_messages[key]
 
@@ -202,20 +189,13 @@ 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]
 
         self.notifier.on_new_replication_data()
 
-    def send_failure(self, failure, destination):
-        """As per TransactionQueue"""
-        pos = self._next_pos()
-
-        self.failures[pos] = (destination, str(failure))
-        self.notifier.on_new_replication_data()
-
     def send_device_messages(self, destination):
         """As per TransactionQueue"""
         pos = self._next_pos()
@@ -255,13 +235,12 @@ class FederationRemoteSendQueue(object):
             self._clear_queue_before_pos(federation_ack)
 
         # Fetch changed presence
-        keys = self.presence_changed.keys()
-        i = keys.bisect_right(from_token)
-        j = keys.bisect_right(to_token) + 1
+        i = self.presence_changed.bisect_right(from_token)
+        j = self.presence_changed.bisect_right(to_token) + 1
         dest_user_ids = [
             (pos, user_id)
-            for pos in keys[i:j]
-            for user_id in self.presence_changed[pos]
+            for pos, user_id_list in self.presence_changed.items()[i:j]
+            for user_id in user_id_list
         ]
 
         for (key, user_id) in dest_user_ids:
@@ -270,13 +249,12 @@ class FederationRemoteSendQueue(object):
             )))
 
         # Fetch changes keyed edus
-        keys = self.keyed_edu_changed.keys()
-        i = keys.bisect_right(from_token)
-        j = keys.bisect_right(to_token) + 1
+        i = self.keyed_edu_changed.bisect_right(from_token)
+        j = self.keyed_edu_changed.bisect_right(to_token) + 1
         # We purposefully clobber based on the key here, python dict comprehensions
         # always use the last value, so this will correctly point to the last
         # stream position.
-        keyed_edus = {self.keyed_edu_changed[k]: k for k in keys[i:j]}
+        keyed_edus = {v: k for k, v in self.keyed_edu_changed.items()[i:j]}
 
         for ((destination, edu_key), pos) in iteritems(keyed_edus):
             rows.append((pos, KeyedEduRow(
@@ -285,31 +263,17 @@ class FederationRemoteSendQueue(object):
             )))
 
         # Fetch changed edus
-        keys = self.edus.keys()
-        i = keys.bisect_right(from_token)
-        j = keys.bisect_right(to_token) + 1
-        edus = ((k, self.edus[k]) for k in keys[i:j])
+        i = self.edus.bisect_right(from_token)
+        j = self.edus.bisect_right(to_token) + 1
+        edus = self.edus.items()[i:j]
 
         for (pos, edu) in edus:
             rows.append((pos, EduRow(edu)))
 
-        # Fetch changed failures
-        keys = self.failures.keys()
-        i = keys.bisect_right(from_token)
-        j = keys.bisect_right(to_token) + 1
-        failures = ((k, self.failures[k]) for k in keys[i:j])
-
-        for (pos, (destination, failure)) in failures:
-            rows.append((pos, FailureRow(
-                destination=destination,
-                failure=failure,
-            )))
-
         # Fetch changed device messages
-        keys = self.device_messages.keys()
-        i = keys.bisect_right(from_token)
-        j = keys.bisect_right(to_token) + 1
-        device_messages = {self.device_messages[k]: k for k in keys[i:j]}
+        i = self.device_messages.bisect_right(from_token)
+        j = self.device_messages.bisect_right(to_token) + 1
+        device_messages = {v: k for k, v in self.device_messages.items()[i:j]}
 
         for (destination, pos) in iteritems(device_messages):
             rows.append((pos, DeviceRow(
@@ -427,34 +391,6 @@ class EduRow(BaseFederationRow, namedtuple("EduRow", (
         buff.edus.setdefault(self.edu.destination, []).append(self.edu)
 
 
-class FailureRow(BaseFederationRow, namedtuple("FailureRow", (
-    "destination",  # str
-    "failure",
-))):
-    """Streams failures to a remote server. Failures are issued when there was
-    something wrong with a transaction the remote sent us, e.g. it included
-    an event that was invalid.
-    """
-
-    TypeId = "f"
-
-    @staticmethod
-    def from_data(data):
-        return FailureRow(
-            destination=data["destination"],
-            failure=data["failure"],
-        )
-
-    def to_data(self):
-        return {
-            "destination": self.destination,
-            "failure": self.failure,
-        }
-
-    def add_to_buffer(self, buff):
-        buff.failures.setdefault(self.destination, []).append(self.failure)
-
-
 class DeviceRow(BaseFederationRow, namedtuple("DeviceRow", (
     "destination",  # str
 ))):
@@ -481,7 +417,6 @@ TypeToRow = {
         PresenceRow,
         KeyedEduRow,
         EduRow,
-        FailureRow,
         DeviceRow,
     )
 }
@@ -491,7 +426,6 @@ ParsedFederationStreamData = namedtuple("ParsedFederationStreamData", (
     "presence",  # list(UserPresenceState)
     "keyed_edus",  # dict of destination -> { key -> Edu }
     "edus",  # dict of destination -> [Edu]
-    "failures",  # dict of destination -> [failures]
     "device_destinations",  # set of destinations
 ))
 
@@ -513,7 +447,6 @@ def process_rows_for_federation(transaction_queue, rows):
         presence=[],
         keyed_edus={},
         edus={},
-        failures={},
         device_destinations=set(),
     )
 
@@ -542,9 +475,5 @@ def process_rows_for_federation(transaction_queue, rows):
                 edu.destination, edu.edu_type, edu.content, key=None,
             )
 
-    for destination, failure_list in iteritems(buff.failures):
-        for failure in failure_list:
-            transaction_queue.send_failure(destination, failure)
-
     for destination in buff.device_destinations:
         transaction_queue.send_device_messages(destination)

synapse/federation/transaction_queue.py

@@ -13,36 +13,41 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 import datetime
+import logging
+
+from six import itervalues
+
+from prometheus_client import Counter
 
 from twisted.internet import defer
 
-from .persistence import TransactionActions
-from .units import Transaction, Edu
-
-from synapse.api.errors import HttpResponseException, FederationDeniedError
-from synapse.util import logcontext, PreserveLoggingContext
-from synapse.util.async import run_on_reactor
-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.api.errors import FederationDeniedError, HttpResponseException
+from synapse.handlers.presence import format_user_presence_state, get_interested_remotes
+from synapse.metrics import (
+    LaterGauge,
+    event_processing_loop_counter,
+    event_processing_loop_room_count,
+    events_processed_counter,
+    sent_edus_counter,
+    sent_transactions_counter,
+)
+from synapse.metrics.background_process_metrics import run_as_background_process
+from synapse.util import logcontext
+from synapse.util.metrics import measure_func
+from synapse.util.retryutils import NotRetryingDestination, get_retry_limiter
 
-import logging
-
+from .persistence import TransactionActions
+from .units import Edu, Transaction
 
 logger = logging.getLogger(__name__)
 
-metrics = synapse.metrics.get_metrics_for(__name__)
-
-client_metrics = synapse.metrics.get_metrics_for("synapse.federation.client")
-sent_pdus_destination_dist = client_metrics.register_distribution(
-    "sent_pdu_destinations"
+sent_pdus_destination_dist_count = Counter(
+    "synapse_federation_client_sent_pdu_destinations:count", ""
+)
+sent_pdus_destination_dist_total = Counter(
+    "synapse_federation_client_sent_pdu_destinations:total", ""
 )
-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):
@@ -53,6 +58,7 @@ class TransactionQueue(object):
     """
 
     def __init__(self, hs):
+        self.hs = hs
         self.server_name = hs.hostname
 
         self.store = hs.get_datastore()
@@ -69,8 +75,10 @@ class TransactionQueue(object):
         # done
         self.pending_transactions = {}
 
-        metrics.register_callback(
-            "pending_destinations",
+        LaterGauge(
+            "synapse_federation_transaction_queue_pending_destinations",
+            "",
+            [],
             lambda: len(self.pending_transactions),
         )
 
@@ -94,12 +102,16 @@ class TransactionQueue(object):
         # Map of destination -> (edu_type, key) -> Edu
         self.pending_edus_keyed_by_dest = edus_keyed = {}
 
-        metrics.register_callback(
-            "pending_pdus",
+        LaterGauge(
+            "synapse_federation_transaction_queue_pending_pdus",
+            "",
+            [],
             lambda: sum(map(len, pdus.values())),
         )
-        metrics.register_callback(
-            "pending_edus",
+        LaterGauge(
+            "synapse_federation_transaction_queue_pending_edus",
+            "",
+            [],
             lambda: (
                 sum(map(len, edus.values()))
                 + sum(map(len, presence.values()))
@@ -107,9 +119,6 @@ class TransactionQueue(object):
             ),
         )
 
-        # destination -> list of tuple(failure, deferred)
-        self.pending_failures_by_dest = {}
-
         # destination -> stream_id of last successfully sent to-device message.
         # NB: may be a long or an int.
         self.last_device_stream_id_by_dest = {}
@@ -157,10 +166,11 @@ class TransactionQueue(object):
         if self._is_processing:
             return
 
-        # fire off a processing loop in the background. It's likely it will
-        # outlast the current request, so run it in the sentinel logcontext.
-        with PreserveLoggingContext():
-            self._process_event_queue_loop()
+        # fire off a processing loop in the background
+        run_as_background_process(
+            "process_event_queue_for_federation",
+            self._process_event_queue_loop,
+        )
 
     @defer.inlineCallbacks
     def _process_event_queue_loop(self):
@@ -228,7 +238,7 @@ 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 +251,21 @@ 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(
-                        now - ts, "federation_sender",
-                    )
-                    synapse.metrics.event_processing_last_ts.set(
-                        ts, "federation_sender",
-                    )
+                    synapse.metrics.event_processing_lag.labels(
+                        "federation_sender").set(now - ts)
+                    synapse.metrics.event_processing_last_ts.labels(
+                        "federation_sender").set(ts)
 
-                events_processed_counter.inc_by(len(events))
+                    events_processed_counter.inc(len(events))
 
-                synapse.metrics.event_processing_positions.set(
-                    next_token, "federation_sender",
-                )
+                    event_processing_loop_room_count.labels(
+                        "federation_sender"
+                    ).inc(len(events_by_room))
+
+                event_processing_loop_counter.labels("federation_sender").inc()
+
+                synapse.metrics.event_processing_positions.labels(
+                    "federation_sender").set(next_token)
 
         finally:
             self._is_processing = False
@@ -275,7 +288,8 @@ class TransactionQueue(object):
         if not destinations:
             return
 
-        sent_pdus_destination_dist.inc_by(len(destinations))
+        sent_pdus_destination_dist_total.inc(len(destinations))
+        sent_pdus_destination_dist_count.inc()
 
         for destination in destinations:
             self.pending_pdus_by_dest.setdefault(destination, []).append(
@@ -295,6 +309,9 @@ class TransactionQueue(object):
         Args:
             states (list(UserPresenceState))
         """
+        if not self.hs.config.use_presence:
+            # No-op if presence is disabled.
+            return
 
         # First we queue up the new presence by user ID, so multiple presence
         # updates in quick successtion are correctly handled
@@ -322,7 +339,7 @@ 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:
@@ -374,19 +391,6 @@ class TransactionQueue(object):
 
         self._attempt_new_transaction(destination)
 
-    def send_failure(self, failure, destination):
-        if destination == self.server_name or destination == "localhost":
-            return
-
-        if not self.can_send_to(destination):
-            return
-
-        self.pending_failures_by_dest.setdefault(
-            destination, []
-        ).append(failure)
-
-        self._attempt_new_transaction(destination)
-
     def send_device_messages(self, destination):
         if destination == self.server_name or destination == "localhost":
             return
@@ -426,14 +430,11 @@ class TransactionQueue(object):
 
         logger.debug("TX [%s] Starting transaction loop", destination)
 
-        # Drop the logcontext before starting the transaction. It doesn't
-        # really make sense to log all the outbound transactions against
-        # whatever path led us to this point: that's pretty arbitrary really.
-        #
-        # (this also means we can fire off _perform_transaction without
-        # yielding)
-        with logcontext.PreserveLoggingContext():
-            self._transaction_transmission_loop(destination)
+        run_as_background_process(
+            "federation_transaction_transmission_loop",
+            self._transaction_transmission_loop,
+            destination,
+        )
 
     @defer.inlineCallbacks
     def _transaction_transmission_loop(self, destination):
@@ -446,9 +447,6 @@ class TransactionQueue(object):
             # hence why we throw the result away.
             yield get_retry_limiter(destination, self.clock, self.store)
 
-            # XXX: what's this for?
-            yield run_on_reactor()
-
             pending_pdus = []
             while True:
                 device_message_edus, device_stream_id, dev_list_id = (
@@ -467,7 +465,6 @@ class TransactionQueue(object):
                 pending_pdus = self.pending_pdus_by_dest.pop(destination, [])
                 pending_edus = self.pending_edus_by_dest.pop(destination, [])
                 pending_presence = self.pending_presence_by_dest.pop(destination, {})
-                pending_failures = self.pending_failures_by_dest.pop(destination, [])
 
                 pending_edus.extend(
                     self.pending_edus_keyed_by_dest.pop(destination, {}).values()
@@ -495,7 +492,7 @@ class TransactionQueue(object):
                     logger.debug("TX [%s] len(pending_pdus_by_dest[dest]) = %d",
                                  destination, len(pending_pdus))
 
-                if not pending_pdus and not pending_edus and not pending_failures:
+                if not pending_pdus and not pending_edus:
                     logger.debug("TX [%s] Nothing to send", destination)
                     self.last_device_stream_id_by_dest[destination] = (
                         device_stream_id
@@ -505,7 +502,7 @@ class TransactionQueue(object):
                 # END CRITICAL SECTION
 
                 success = yield self._send_new_transaction(
-                    destination, pending_pdus, pending_edus, pending_failures,
+                    destination, pending_pdus, pending_edus,
                 )
                 if success:
                     sent_transactions_counter.inc()
@@ -582,14 +579,12 @@ class TransactionQueue(object):
 
     @measure_func("_send_new_transaction")
     @defer.inlineCallbacks
-    def _send_new_transaction(self, destination, pending_pdus, pending_edus,
-                              pending_failures):
+    def _send_new_transaction(self, destination, pending_pdus, pending_edus):
 
         # Sort based on the order field
         pending_pdus.sort(key=lambda t: t[1])
         pdus = [x[0] for x in pending_pdus]
         edus = pending_edus
-        failures = [x.get_dict() for x in pending_failures]
 
         success = True
 
@@ -599,11 +594,10 @@ class TransactionQueue(object):
 
         logger.debug(
             "TX [%s] {%s} Attempting new transaction"
-            " (pdus: %d, edus: %d, failures: %d)",
+            " (pdus: %d, edus: %d)",
             destination, txn_id,
             len(pdus),
             len(edus),
-            len(failures)
         )
 
         logger.debug("TX [%s] Persisting transaction...", destination)
@@ -615,7 +609,6 @@ class TransactionQueue(object):
             destination=destination,
             pdus=pdus,
             edus=edus,
-            pdu_failures=failures,
         )
 
         self._next_txn_id += 1
@@ -625,12 +618,11 @@ class TransactionQueue(object):
         logger.debug("TX [%s] Persisted transaction", destination)
         logger.info(
             "TX [%s] {%s} Sending transaction [%s],"
-            " (PDUs: %d, EDUs: %d, failures: %d)",
+            " (PDUs: %d, EDUs: %d)",
             destination, txn_id,
             transaction.transaction_id,
             len(pdus),
             len(edus),
-            len(failures),
         )
 
         # Actually send the transaction

synapse/federation/transport/client.py

@@ -14,15 +14,14 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from twisted.internet import defer
-from synapse.api.constants import Membership
-
-from synapse.api.urls import FEDERATION_PREFIX as PREFIX
-from synapse.util.logutils import log_function
-
 import logging
 import urllib
 
+from twisted.internet import defer
+
+from synapse.api.constants import Membership
+from synapse.api.urls import FEDERATION_PREFIX as PREFIX
+from synapse.util.logutils import log_function
 
 logger = logging.getLogger(__name__)
 
@@ -107,7 +106,7 @@ class TransportLayerClient(object):
             dest (str)
             room_id (str)
             event_tuples (list)
-            limt (int)
+            limit (int)
 
         Returns:
             Deferred: Results in a dict received from the remote homeserver.
@@ -196,7 +195,7 @@ class TransportLayerClient(object):
 
     @defer.inlineCallbacks
     @log_function
-    def make_membership_event(self, destination, room_id, user_id, membership):
+    def make_membership_event(self, destination, room_id, user_id, membership, params):
         """Asks a remote server to build and sign us a membership event
 
         Note that this does not append any events to any graphs.
@@ -206,6 +205,8 @@ class TransportLayerClient(object):
             room_id (str): room to join/leave
             user_id (str): user to be joined/left
             membership (str): one of join/leave
+            params (dict[str, str|Iterable[str]]): Query parameters to include in the
+                request.
 
         Returns:
             Deferred: Succeeds when we get a 2xx HTTP response. The result
@@ -242,6 +243,7 @@ class TransportLayerClient(object):
         content = yield self.client.get_json(
             destination=destination,
             path=path,
+            args=params,
             retry_on_dns_fail=retry_on_dns_fail,
             timeout=20000,
             ignore_backoff=ignore_backoff,

synapse/federation/transport/server.py

@@ -14,25 +14,27 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from twisted.internet import defer
-
-from synapse.api.urls import FEDERATION_PREFIX as PREFIX
-from synapse.api.errors import Codes, SynapseError, FederationDeniedError
-from synapse.http.server import JsonResource
-from synapse.http.servlet import (
-    parse_json_object_from_request, parse_integer_from_args, parse_string_from_args,
-    parse_boolean_from_args,
-)
-from synapse.util.ratelimitutils import FederationRateLimiter
-from synapse.util.versionstring import get_version_string
-from synapse.util.logcontext import run_in_background
-from synapse.types import ThirdPartyInstanceID, get_domain_from_id
-
 import functools
 import logging
 import re
-import synapse
 
+from twisted.internet import defer
+
+import synapse
+from synapse.api.errors import Codes, FederationDeniedError, SynapseError
+from synapse.api.urls import FEDERATION_PREFIX as PREFIX
+from synapse.http.endpoint import parse_and_validate_server_name
+from synapse.http.server import JsonResource
+from synapse.http.servlet import (
+    parse_boolean_from_args,
+    parse_integer_from_args,
+    parse_json_object_from_request,
+    parse_string_from_args,
+)
+from synapse.types import ThirdPartyInstanceID, get_domain_from_id
+from synapse.util.logcontext import run_in_background
+from synapse.util.ratelimitutils import FederationRateLimiter
+from synapse.util.versionstring import get_version_string
 
 logger = logging.getLogger(__name__)
 
@@ -99,26 +101,6 @@ class Authenticator(object):
 
         origin = None
 
-        def parse_auth_header(header_str):
-            try:
-                params = auth.split(" ")[1].split(",")
-                param_dict = dict(kv.split("=") for kv in params)
-
-                def strip_quotes(value):
-                    if value.startswith("\""):
-                        return value[1:-1]
-                    else:
-                        return value
-
-                origin = strip_quotes(param_dict["origin"])
-                key = strip_quotes(param_dict["key"])
-                sig = strip_quotes(param_dict["sig"])
-                return (origin, key, sig)
-            except Exception:
-                raise AuthenticationError(
-                    400, "Malformed Authorization header", Codes.UNAUTHORIZED
-                )
-
         auth_headers = request.requestHeaders.getRawHeaders(b"Authorization")
 
         if not auth_headers:
@@ -127,8 +109,8 @@ class Authenticator(object):
             )
 
         for auth in auth_headers:
-            if auth.startswith("X-Matrix"):
-                (origin, key, sig) = parse_auth_header(auth)
+            if auth.startswith(b"X-Matrix"):
+                (origin, key, sig) = _parse_auth_header(auth)
                 json_request["origin"] = origin
                 json_request["signatures"].setdefault(origin, {})[key] = sig
 
@@ -165,7 +147,84 @@ class Authenticator(object):
             logger.exception("Error resetting retry timings on %s", origin)
 
 
+def _parse_auth_header(header_bytes):
+    """Parse an X-Matrix auth header
+
+    Args:
+        header_bytes (bytes): header value
+
+    Returns:
+        Tuple[str, str, str]: origin, key id, signature.
+
+    Raises:
+        AuthenticationError if the header could not be parsed
+    """
+    try:
+        header_str = header_bytes.decode('utf-8')
+        params = header_str.split(" ")[1].split(",")
+        param_dict = dict(kv.split("=") for kv in params)
+
+        def strip_quotes(value):
+            if value.startswith("\""):
+                return value[1:-1]
+            else:
+                return value
+
+        origin = strip_quotes(param_dict["origin"])
+
+        # ensure that the origin is a valid server name
+        parse_and_validate_server_name(origin)
+
+        key = strip_quotes(param_dict["key"])
+        sig = strip_quotes(param_dict["sig"])
+        return origin, key, sig
+    except Exception as e:
+        logger.warn(
+            "Error parsing auth header '%s': %s",
+            header_bytes.decode('ascii', 'replace'),
+            e,
+        )
+        raise AuthenticationError(
+            400, "Malformed Authorization header", Codes.UNAUTHORIZED,
+        )
+
+
 class BaseFederationServlet(object):
+    """Abstract base class for federation servlet classes.
+
+    The servlet object should have a PATH attribute which takes the form of a regexp to
+    match against the request path (excluding the /federation/v1 prefix).
+
+    The servlet should also implement one or more of on_GET, on_POST, on_PUT, to match
+    the appropriate HTTP method. These methods have the signature:
+
+        on_<METHOD>(self, origin, content, query, **kwargs)
+
+        With arguments:
+
+            origin (unicode|None): The authenticated server_name of the calling server,
+                unless REQUIRE_AUTH is set to False and authentication failed.
+
+            content (unicode|None): decoded json body of the request. None if the
+                request was a GET.
+
+            query (dict[bytes, list[bytes]]): Query params from the request. url-decoded
+                (ie, '+' and '%xx' are decoded) but note that it is *not* utf8-decoded
+                yet.
+
+            **kwargs (dict[unicode, unicode]): the dict mapping keys to path
+                components as specified in the path match regexp.
+
+        Returns:
+            Deferred[(int, object)|None]: either (response code, response object) to
+                 return a JSON response, or None if the request has already been handled.
+
+        Raises:
+            SynapseError: to return an error code
+
+            Exception: other exceptions will be caught, logged, and a 500 will be
+                returned.
+    """
     REQUIRE_AUTH = True
 
     def __init__(self, handler, authenticator, ratelimiter, server_name):
@@ -180,6 +239,18 @@ class BaseFederationServlet(object):
         @defer.inlineCallbacks
         @functools.wraps(func)
         def new_func(request, *args, **kwargs):
+            """ A callback which can be passed to HttpServer.RegisterPaths
+
+            Args:
+                request (twisted.web.http.Request):
+                *args: unused?
+                **kwargs (dict[unicode, unicode]): the dict mapping keys to path
+                    components as specified in the path match regexp.
+
+            Returns:
+                Deferred[(int, object)|None]: (response code, response object) as returned
+                    by the callback method. None if the request has already been handled.
+            """
             content = None
             if request.method in ["PUT", "POST"]:
                 # TODO: Handle other method types? other content types?
@@ -190,10 +261,10 @@ class BaseFederationServlet(object):
             except NoAuthenticationError:
                 origin = None
                 if self.REQUIRE_AUTH:
-                    logger.exception("authenticate_request failed")
+                    logger.warn("authenticate_request failed: missing authentication")
                     raise
-            except Exception:
-                logger.exception("authenticate_request failed")
+            except Exception as e:
+                logger.warn("authenticate_request failed: %s", e)
                 raise
 
             if origin:
@@ -259,11 +330,10 @@ class FederationSendServlet(BaseFederationServlet):
             )
 
             logger.info(
-                "Received txn %s from %s. (PDUs: %d, EDUs: %d, failures: %d)",
+                "Received txn %s from %s. (PDUs: %d, EDUs: %d)",
                 transaction_id, origin,
                 len(transaction_data.get("pdus", [])),
                 len(transaction_data.get("edus", [])),
-                len(transaction_data.get("failures", [])),
             )
 
             # We should ideally be getting this from the security layer.
@@ -283,7 +353,7 @@ class FederationSendServlet(BaseFederationServlet):
 
         try:
             code, response = yield self.handler.on_incoming_transaction(
-                transaction_data
+                origin, transaction_data,
             )
         except Exception:
             logger.exception("on_incoming_transaction failed")
@@ -361,8 +431,32 @@ class FederationMakeJoinServlet(BaseFederationServlet):
     PATH = "/make_join/(?P<context>[^/]*)/(?P<user_id>[^/]*)"
 
     @defer.inlineCallbacks
-    def on_GET(self, origin, content, query, context, user_id):
-        content = yield self.handler.on_make_join_request(context, user_id)
+    def on_GET(self, origin, _content, query, context, user_id):
+        """
+        Args:
+            origin (unicode): The authenticated server_name of the calling server
+
+            _content (None): (GETs don't have bodies)
+
+            query (dict[bytes, list[bytes]]): Query params from the request.
+
+            **kwargs (dict[unicode, unicode]): the dict mapping keys to path
+                components as specified in the path match regexp.
+
+        Returns:
+            Deferred[(int, object)|None]: either (response code, response object) to
+                 return a JSON response, or None if the request has already been handled.
+        """
+        versions = query.get(b'ver')
+        if versions is not None:
+            supported_versions = [v.decode("utf-8") for v in versions]
+        else:
+            supported_versions = ["1"]
+
+        content = yield self.handler.on_make_join_request(
+            origin, context, user_id,
+            supported_versions=supported_versions,
+        )
         defer.returnValue((200, content))
 
 
@@ -371,15 +465,17 @@ class FederationMakeLeaveServlet(BaseFederationServlet):
 
     @defer.inlineCallbacks
     def on_GET(self, origin, content, query, context, user_id):
-        content = yield self.handler.on_make_leave_request(context, user_id)
+        content = yield self.handler.on_make_leave_request(
+            origin, context, user_id,
+        )
         defer.returnValue((200, content))
 
 
 class FederationSendLeaveServlet(BaseFederationServlet):
-    PATH = "/send_leave/(?P<room_id>[^/]*)/(?P<txid>[^/]*)"
+    PATH = "/send_leave/(?P<room_id>[^/]*)/(?P<event_id>[^/]*)"
 
     @defer.inlineCallbacks
-    def on_PUT(self, origin, content, query, room_id, txid):
+    def on_PUT(self, origin, content, query, room_id, event_id):
         content = yield self.handler.on_send_leave_request(origin, content)
         defer.returnValue((200, content))
 

synapse/federation/units.py

@@ -17,10 +17,9 @@
 server protocol.
 """
 
-from synapse.util.jsonobject import JsonEncodedObject
-
 import logging
 
+from synapse.util.jsonobject import JsonEncodedObject
 
 logger = logging.getLogger(__name__)
 
@@ -74,7 +73,6 @@ class Transaction(JsonEncodedObject):
         "previous_ids",
         "pdus",
         "edus",
-        "pdu_failures",
     ]
 
     internal_keys = [

synapse/groups/attestations.py

@@ -23,9 +23,9 @@ If a user leaves (or gets kicked out of) a group, either side can still use
 their attestation to "prove" their membership, until the attestation expires.
 Therefore attestations shouldn't be relied on to prove membership in important
 cases, but can for less important situtations, e.g. showing a users membership
-of groups on their profile, showing flairs, etc.abs
+of groups on their profile, showing flairs, etc.
 
-An attestsation is a signed blob of json that looks like:
+An attestation is a signed blob of json that looks like:
 
     {
         "user_id": "@foo:a.example.com",
@@ -38,15 +38,15 @@ An attestsation is a signed blob of json that looks like:
 import logging
 import random
 
+from signedjson.sign import sign_json
+
 from twisted.internet import defer
 
 from synapse.api.errors import SynapseError
+from synapse.metrics.background_process_metrics import run_as_background_process
 from synapse.types import get_domain_from_id
 from synapse.util.logcontext import run_in_background
 
-from signedjson.sign import sign_json
-
-
 logger = logging.getLogger(__name__)
 
 
@@ -130,7 +130,7 @@ class GroupAttestionRenewer(object):
         self.attestations = hs.get_groups_attestation_signing()
 
         self._renew_attestations_loop = self.clock.looping_call(
-            self._renew_attestations, 30 * 60 * 1000,
+            self._start_renew_attestations, 30 * 60 * 1000,
         )
 
     @defer.inlineCallbacks
@@ -152,6 +152,9 @@ class GroupAttestionRenewer(object):
 
         defer.returnValue({})
 
+    def _start_renew_attestations(self):
+        return run_as_background_process("renew_attestations", self._renew_attestations)
+
     @defer.inlineCallbacks
     def _renew_attestations(self):
         """Called periodically to check if we need to update any of our attestations

synapse/groups/groups_server.py

@@ -16,9 +16,12 @@
 
 import logging
 
+from six import string_types
+
+from twisted.internet import defer
+
 from synapse.api.errors import SynapseError
 from synapse.types import GroupID, RoomID, UserID, get_domain_from_id
-from twisted.internet import defer
 
 logger = logging.getLogger(__name__)
 
@@ -431,7 +434,7 @@ 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

@@ -13,13 +13,11 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from .register import RegistrationHandler
-from .room import RoomContextHandler
-from .message import MessageHandler
-from .federation import FederationHandler
-from .directory import DirectoryHandler
 from .admin import AdminHandler
+from .directory import DirectoryHandler
+from .federation import FederationHandler
 from .identity import IdentityHandler
+from .register import RegistrationHandler
 from .search import SearchHandler
 
 
@@ -44,10 +42,8 @@ class Handlers(object):
 
     def __init__(self, hs):
         self.registration_handler = RegistrationHandler(hs)
-        self.message_handler = MessageHandler(hs)
         self.federation_handler = FederationHandler(hs)
         self.directory_handler = DirectoryHandler(hs)
         self.admin_handler = AdminHandler(hs)
         self.identity_handler = IdentityHandler(hs)
         self.search_handler = SearchHandler(hs)
-        self.room_context_handler = RoomContextHandler(hs)