Merge branch 'develop' into cross-signing_sig_upload

37 files changed, 2261 insertions, 2234 deletions

diff --git a/docs/CAPTCHA_SETUP.rst b/docs/CAPTCHA_SETUP.md
index 0c22ee4ff6..5f9057530b 100644
--- a/docs/CAPTCHA_SETUP.rst
+++ b/docs/CAPTCHA_SETUP.md
@@ -1,30 +1,31 @@
+# Overview
 Captcha can be enabled for this home server. This file explains how to do that.
 The captcha mechanism used is Google's ReCaptcha. This requires API keys from Google.
 
-Getting keys
-------------
+## Getting keys
+
 Requires a public/private key pair from:
 
-https://developers.google.com/recaptcha/
+<https://developers.google.com/recaptcha/>
 
 Must be a reCAPTCHA v2 key using the "I'm not a robot" Checkbox option
 
-Setting ReCaptcha Keys
-----------------------
+## Setting ReCaptcha Keys
+
 The keys are a config option on the home server config. If they are not
-visible, you can generate them via --generate-config. Set the following value::
+visible, you can generate them via `--generate-config`. Set the following value:
+
+    recaptcha_public_key: YOUR_PUBLIC_KEY
+    recaptcha_private_key: YOUR_PRIVATE_KEY
 
-  recaptcha_public_key: YOUR_PUBLIC_KEY
-  recaptcha_private_key: YOUR_PRIVATE_KEY
+In addition, you MUST enable captchas via:
 
-In addition, you MUST enable captchas via::
+    enable_registration_captcha: true
 
-  enable_registration_captcha: true
+## Configuring IP used for auth
 
-Configuring IP used for auth
-----------------------------
 The ReCaptcha API requires that the IP address of the user who solved the
 captcha is sent. If the client is connecting through a proxy or load balancer,
-it may be required to use the X-Forwarded-For (XFF) header instead of the origin
-IP address. This can be configured using the x_forwarded directive in the
+it may be required to use the `X-Forwarded-For` (XFF) header instead of the origin
+IP address. This can be configured using the `x_forwarded` directive in the
 listeners section of the homeserver.yaml configuration file.
diff --git a/docs/MSC1711_certificates_FAQ.md b/docs/MSC1711_certificates_FAQ.md
index 83497380df..80bd1294c7 100644
--- a/docs/MSC1711_certificates_FAQ.md
+++ b/docs/MSC1711_certificates_FAQ.md
@@ -147,7 +147,7 @@ your domain, you can simply route all traffic through the reverse proxy by
 updating the SRV record appropriately (or removing it, if the proxy listens on
 8448).
 
-See [reverse_proxy.rst](reverse_proxy.rst) for information on setting up a
+See [reverse_proxy.md](reverse_proxy.md) for information on setting up a
 reverse proxy.
 
 #### Option 3: add a .well-known file to delegate your matrix traffic
@@ -319,7 +319,7 @@ We no longer actively recommend against using a reverse proxy. Many admins will
 find it easier to direct federation traffic to a reverse proxy and manage their
 own TLS certificates, and this is a supported configuration.
 
-See [reverse_proxy.rst](reverse_proxy.rst) for information on setting up a
+See [reverse_proxy.md](reverse_proxy.md) for information on setting up a
 reverse proxy.
 
 ### Do I still need to give my TLS certificates to Synapse if I am using a reverse proxy?
diff --git a/docs/README.md b/docs/README.md
new file mode 100644
index 0000000000..3c6ea48c66
--- /dev/null
+++ b/docs/README.md
@@ -0,0 +1,7 @@
+# Synapse Documentation
+
+This directory contains documentation specific to the `synapse` homeserver.
+
+All matrix-generic documentation now lives in its own project, located at [matrix-org/matrix-doc](https://github.com/matrix-org/matrix-doc)
+
+(Note:  some items here may be moved to [matrix-org/matrix-doc](https://github.com/matrix-org/matrix-doc) at some point in the future.)
diff --git a/docs/README.rst b/docs/README.rst
deleted file mode 100644
index 3012da8b19..0000000000
--- a/docs/README.rst
+++ /dev/null
@@ -1,6 +0,0 @@
-All matrix-generic documentation now lives in its own project at
-
-github.com/matrix-org/matrix-doc.git
-
-Only Synapse implementation-specific documentation lives here now
-(together with some older stuff will be shortly migrated over to matrix-doc)
diff --git a/docs/admin_api/README.rst b/docs/admin_api/README.rst
index d4f564cfae..191806c5b4 100644
--- a/docs/admin_api/README.rst
+++ b/docs/admin_api/README.rst
@@ -10,3 +10,15 @@ server admin by updating the database directly, e.g.:
 ``UPDATE users SET admin = 1 WHERE name = '@foo:bar.com'``
 
 Restarting may be required for the changes to register.
+
+Using an admin access_token
+###########################
+
+Many of the API calls listed in the documentation here will require to include an admin `access_token`.
+Finding your user's `access_token` is client-dependent, but will usually be shown in the client's settings.
+
+Once you have your `access_token`, to include it in a request, the best option is to add the token to a request header:
+
+``curl --header "Authorization: Bearer <access_token>" <the_rest_of_your_API_request>``
+
+Fore more details, please refer to the complete `matrix spec documentation <https://matrix.org/docs/spec/client_server/r0.5.0#using-access-tokens>`_.
diff --git a/docs/ancient_architecture_notes.md b/docs/ancient_architecture_notes.md
new file mode 100644
index 0000000000..3ea8976cc7
--- /dev/null
+++ b/docs/ancient_architecture_notes.md
@@ -0,0 +1,81 @@
+> **Warning**
+>  These architecture notes are spectacularly old, and date back
+> to when Synapse was just federation code in isolation. This should be
+> merged into the main spec.
+
+# Server to Server
+
+## Server to Server Stack
+
+To use the server to server stack, home servers should only need to
+interact with the Messaging layer.
+
+The server to server side of things is designed into 4 distinct layers:
+
+1.  Messaging Layer
+2.  Pdu Layer
+3.  Transaction Layer
+4.  Transport Layer
+
+Where the bottom (the transport layer) is what talks to the internet via
+HTTP, and the top (the messaging layer) talks to the rest of the Home
+Server with a domain specific API.
+
+1. **Messaging Layer**
+
+    This is what the rest of the Home Server hits to send messages, join rooms,
+    etc. It also allows you to register callbacks for when it get's notified by
+    lower levels that e.g. a new message has been received.
+
+    It is responsible for serializing requests to send to the data
+    layer, and to parse requests received from the data layer.
+
+2. **PDU Layer**
+
+    This layer handles:
+
+		- duplicate `pdu_id`'s - i.e., it makes sure we ignore them.
+		- responding to requests for a given `pdu_id`
+		- responding to requests for all metadata for a given context (i.e. room)
+		- handling incoming backfill requests
+
+		So it has to parse incoming messages to discover which are metadata and
+    which aren't, and has to correctly clobber existing metadata where
+    appropriate.
+
+    For incoming PDUs, it has to check the PDUs it references to see
+    if we have missed any. If we have go and ask someone (another
+    home server) for it.
+
+3. **Transaction Layer**
+
+		This layer makes incoming requests idempotent. i.e., it stores
+		which transaction id's we have seen and what our response were.
+		If we have already seen a message with the given transaction id,
+		we do not notify higher levels but simply respond with the
+		previous response.
+
+		`transaction_id` is from "`GET /send/<tx_id>/`"
+
+		It's also responsible for batching PDUs into single transaction for
+		sending to remote destinations, so that we only ever have one
+		transaction in flight to a given destination at any one time.
+
+		This is also responsible for answering requests for things after a
+		given set of transactions, i.e., ask for everything after 'ver' X.
+
+4. **Transport Layer**
+
+		This is responsible for starting a HTTP server and hitting the
+		correct callbacks on the Transaction layer, as well as sending
+		both data and requests for data.
+
+## Persistence
+
+We persist things in a single sqlite3 database. All database queries get
+run on a separate, dedicated thread. This that we only ever have one
+query running at a time, making it a lot easier to do things in a safe
+manner.
+
+The queries are located in the `synapse.persistence.transactions` module,
+and the table information in the `synapse.persistence.tables` module.
diff --git a/docs/ancient_architecture_notes.rst b/docs/ancient_architecture_notes.rst
deleted file mode 100644
index 2a5a2613c4..0000000000
--- a/docs/ancient_architecture_notes.rst
+++ /dev/null
@@ -1,59 +0,0 @@
-.. WARNING::
-  These architecture notes are spectacularly old, and date back to when Synapse 
-  was just federation code in isolation.  This should be merged into the main
-  spec.
-  
-
-= Server to Server =
-
-== Server to Server Stack ==
-
-To use the server to server stack, home servers should only need to interact with the Messaging layer.
-
-The server to server side of things is designed into 4 distinct layers:
-
-    1. Messaging Layer
-    2. Pdu Layer
-    3. Transaction Layer
-    4. Transport Layer
-
-Where the bottom (the transport layer) is what talks to the internet via HTTP, and the top (the messaging layer) talks to the rest of the Home Server with a domain specific API.
-
-1. Messaging Layer
-    This is what the rest of the Home Server hits to send messages, join rooms, etc. It also allows you to register callbacks for when it get's notified by lower levels that e.g. a new message has been received.
-
-    It is responsible for serializing requests to send to the data layer, and to parse requests received from the data layer.
-
-
-2. PDU Layer
-    This layer handles: 
-        * duplicate pdu_id's - i.e., it makes sure we ignore them. 
-        * responding to requests for a given pdu_id
-        * responding to requests for all metadata for a given context (i.e. room)
-        * handling incoming backfill requests
-
-    So it has to parse incoming messages to discover which are metadata and which aren't, and has to correctly clobber existing metadata where appropriate.
-
-    For incoming PDUs, it has to check the PDUs it references to see if we have missed any. If we have go and ask someone (another home server) for it.    
-
-
-3. Transaction Layer
-    This layer makes incoming requests idempotent. I.e., it stores which transaction id's we have seen and what our response were. If we have already seen a message with the given transaction id, we do not notify higher levels but simply respond with the previous response.
-
-transaction_id is from "GET /send/<tx_id>/"
-
-    It's also responsible for batching PDUs into single transaction for sending to remote destinations, so that we only ever have one transaction in flight to a given destination at any one time.
-
-    This is also responsible for answering requests for things after a given set of transactions, i.e., ask for everything after 'ver' X.
-
-
-4. Transport Layer
-    This is responsible for starting a HTTP server and hitting the correct callbacks on the Transaction layer, as well as sending both data and requests for data.
-
-
-== Persistence ==
-
-We persist things in a single sqlite3 database. All database queries get run on a separate, dedicated thread. This that we only ever have one query running at a time, making it a lot easier to do things in a safe manner.
-
-The queries are located in the synapse.persistence.transactions module, and the table information in the synapse.persistence.tables module.
-
diff --git a/docs/application_services.md b/docs/application_services.md
new file mode 100644
index 0000000000..06cb79f1f9
--- /dev/null
+++ b/docs/application_services.md
@@ -0,0 +1,31 @@
+# Registering an Application Service
+
+The registration of new application services depends on the homeserver used. 
+In synapse, you need to create a new configuration file for your AS and add it
+to the list specified under the `app_service_config_files` config
+option in your synapse config.
+
+For example:
+
+```yaml
+app_service_config_files:
+- /home/matrix/.synapse/<your-AS>.yaml
+```
+
+The format of the AS configuration file is as follows:
+
+```yaml
+url: <base url of AS>
+as_token: <token AS will add to requests to HS>
+hs_token: <token HS will add to requests to AS>
+sender_localpart: <localpart of AS user>
+namespaces:
+  users:  # List of users we're interested in
+    - exclusive: <bool>
+      regex: <regex>
+    - ...
+  aliases: []  # List of aliases we're interested in
+  rooms: [] # List of room ids we're interested in
+```
+
+See the [spec](https://matrix.org/docs/spec/application_service/unstable.html) for further details on how application services work.
diff --git a/docs/application_services.rst b/docs/application_services.rst
deleted file mode 100644
index fbc0c7e960..0000000000
--- a/docs/application_services.rst
+++ /dev/null
@@ -1,35 +0,0 @@
-Registering an Application Service
-==================================
-
-The registration of new application services depends on the homeserver used. 
-In synapse, you need to create a new configuration file for your AS and add it
-to the list specified under the ``app_service_config_files`` config
-option in your synapse config.
-
-For example:
-
-.. code-block:: yaml
-
-  app_service_config_files:
-  - /home/matrix/.synapse/<your-AS>.yaml
-
-
-The format of the AS configuration file is as follows:
-
-..  code-block:: yaml
-
-    url: <base url of AS>
-    as_token: <token AS will add to requests to HS>
-    hs_token: <token HS will add to requests to AS>
-    sender_localpart: <localpart of AS user>
-    namespaces:
-      users:  # List of users we're interested in
-        - exclusive: <bool>
-          regex: <regex>
-        - ...
-      aliases: []  # List of aliases we're interested in
-      rooms: [] # List of room ids we're interested in
-
-See the spec_ for further details on how application services work.
-
-.. _spec: https://matrix.org/docs/spec/application_service/unstable.html
diff --git a/docs/architecture.md b/docs/architecture.md
new file mode 100644
index 0000000000..0c7f315f3f
--- /dev/null
+++ b/docs/architecture.md
@@ -0,0 +1,65 @@
+# Synapse Architecture
+
+As of the end of Oct 2014, Synapse's overall architecture looks like:
+
+        synapse
+        .-----------------------------------------------------.
+        |                          Notifier                   |
+        |                            ^  |                     |
+        |                            |  |                     |
+        |                  .------------|------.              |
+        |                  | handlers/  |      |              |
+        |                  |            v      |              |
+        |                  | Event*Handler <--------> rest/* <=> Client
+        |                  | Rooms*Handler     |              |
+    HS <=> federation/* <==> FederationHandler |              |
+        |      |           | PresenceHandler   |              |
+        |      |           | TypingHandler     |              |
+        |      |           '-------------------'              |
+        |      |                 |     |                      |
+        |      |              state/*  |                      |
+        |      |                 |     |                      |
+        |      |                 v     v                      |
+        |      `--------------> storage/*                     |
+        |                          |                          |
+        '--------------------------|--------------------------'
+                                   v
+                                .----.
+                                | DB |
+                                '----'
+
+-   Handlers: business logic of synapse itself. Follows a set contract of BaseHandler:
+    -   BaseHandler gives us onNewRoomEvent which: (TODO: flesh this out and make it less cryptic):
+        -   handle_state(event)
+        -   auth(event)
+        -   persist_event(event)
+        -   notify notifier or federation(event)
+    -   PresenceHandler: use distributor to get EDUs out of Federation.
+        Very lightweight logic built on the distributor
+    -   TypingHandler: use distributor to get EDUs out of Federation.
+        Very lightweight logic built on the distributor
+    -   EventsHandler: handles the events stream...
+    -   FederationHandler: - gets PDU from Federation Layer; turns into
+        an event; follows basehandler functionality.
+    -   RoomsHandler: does all the room logic, including members - lots
+        of classes in RoomsHandler.
+    -   ProfileHandler: talks to the storage to store/retrieve profile
+        info.
+-   EventFactory: generates events of particular event types.
+-   Notifier: Backs the events handler
+-   REST: Interfaces handlers and events to the outside world via
+    HTTP/JSON. Converts events back and forth from JSON.
+-   Federation: holds the HTTP client & server to talk to other servers.
+    Does replication to make sure there's nothing missing in the graph.
+    Handles reliability. Handles txns.
+-   Distributor: generic event bus. used for presence & typing only
+    currently. Notifier could be implemented using Distributor - so far
+    we are only using for things which actually /require/ dynamic
+    pluggability however as it can obfuscate the actual flow of control.
+-   Auth: helper singleton to say whether a given event is allowed to do
+    a given thing (TODO: put this on the diagram)
+-   State: helper singleton: does state conflict resolution. You give it
+    an event and it tells you if it actually updates the state or not,
+    and annotates the event up properly and handles merge conflict
+    resolution.
+-   Storage: abstracts the storage engine.
diff --git a/docs/architecture.rst b/docs/architecture.rst
deleted file mode 100644
index 98050428b9..0000000000
--- a/docs/architecture.rst
+++ /dev/null
@@ -1,68 +0,0 @@
-Synapse Architecture
-====================
-
-As of the end of Oct 2014, Synapse's overall architecture looks like::
-
-        synapse
-        .-----------------------------------------------------.
-        |                          Notifier                   |
-        |                            ^  |                     |
-        |                            |  |                     |
-        |                  .------------|------.              |
-        |                  | handlers/  |      |              |
-        |                  |            v      |              |
-        |                  | Event*Handler <--------> rest/* <=> Client
-        |                  | Rooms*Handler     |              |
-  HSes <=> federation/* <==> FederationHandler |              |
-        |      |           | PresenceHandler   |              |
-        |      |           | TypingHandler     |              |
-        |      |           '-------------------'              |
-        |      |                 |     |                      |
-        |      |              state/*  |                      |
-        |      |                 |     |                      |
-        |      |                 v     v                      |
-        |      `--------------> storage/*                     |
-        |                          |                          |
-        '--------------------------|--------------------------'
-                                   v
-                                .----.
-                                | DB |
-                                '----'
-
-* Handlers: business logic of synapse itself.  Follows a set contract of BaseHandler:
-
-  - BaseHandler gives us onNewRoomEvent which: (TODO: flesh this out and make it less cryptic):
- 
-    + handle_state(event)
-    + auth(event)
-    + persist_event(event)
-    + notify notifier or federation(event)
-   
-  - PresenceHandler: use distributor to get EDUs out of Federation.  Very
-    lightweight logic built on the distributor
-  - TypingHandler: use distributor to get EDUs out of Federation.  Very
-    lightweight logic built on the distributor
-  - EventsHandler: handles the events stream...
-  - FederationHandler: - gets PDU from Federation Layer; turns into an event;
-    follows basehandler functionality.
-  - RoomsHandler: does all the room logic, including members - lots of classes in
-    RoomsHandler.
-  - ProfileHandler: talks to the storage to store/retrieve profile info.
-
-* EventFactory: generates events of particular event types.
-* Notifier: Backs the events handler
-* REST: Interfaces handlers and events to the outside world via HTTP/JSON.
-  Converts events back and forth from JSON.
-* Federation: holds the HTTP client & server to talk to other servers.  Does
-  replication to make sure there's nothing missing in the graph.  Handles
-  reliability.  Handles txns.
-* Distributor: generic event bus. used for presence & typing only currently. 
-  Notifier could be implemented using Distributor - so far we are only using for
-  things which actually /require/ dynamic pluggability however as it can
-  obfuscate the actual flow of control.
-* Auth: helper singleton to say whether a given event is allowed to do a given
-  thing  (TODO: put this on the diagram)
-* State: helper singleton: does state conflict resolution. You give it an event
-  and it tells you if it actually updates the state or not, and annotates the
-  event up properly and handles merge conflict resolution.
-* Storage: abstracts the storage engine.
diff --git a/docs/code_style.md b/docs/code_style.md
new file mode 100644
index 0000000000..f983f72d6c
--- /dev/null
+++ b/docs/code_style.md
@@ -0,0 +1,169 @@
+# Code Style
+
+## Formatting tools
+
+The Synapse codebase uses a number of code formatting tools in order to
+quickly and automatically check for formatting (and sometimes logical)
+errors in code.
+
+The necessary tools are detailed below.
+
+-   **black**
+
+    The Synapse codebase uses [black](https://pypi.org/project/black/)
+    as an opinionated code formatter, ensuring all comitted code is
+    properly formatted.
+
+    First install `black` with:
+
+        pip install --upgrade black
+
+    Have `black` auto-format your code (it shouldn't change any
+    functionality) with:
+
+        black . --exclude="\.tox|build|env"
+
+-   **flake8**
+
+    `flake8` is a code checking tool. We require code to pass `flake8`
+    before being merged into the codebase.
+
+    Install `flake8` with:
+
+        pip install --upgrade flake8
+
+    Check all application and test code with:
+
+        flake8 synapse tests
+
+-   **isort**
+
+    `isort` ensures imports are nicely formatted, and can suggest and
+    auto-fix issues such as double-importing.
+
+    Install `isort` with:
+
+        pip install --upgrade isort
+
+    Auto-fix imports with:
+
+        isort -rc synapse tests
+
+    `-rc` means to recursively search the given directories.
+
+It's worth noting that modern IDEs and text editors can run these tools
+automatically on save. It may be worth looking into whether this
+functionality is supported in your editor for a more convenient
+development workflow. It is not, however, recommended to run `flake8` on
+save as it takes a while and is very resource intensive.
+
+## General rules
+
+-   **Naming**:
+    -   Use camel case for class and type names
+    -   Use underscores for functions and variables.
+-   **Docstrings**: should follow the [google code
+    style](https://google.github.io/styleguide/pyguide.html#38-comments-and-docstrings).
+    This is so that we can generate documentation with
+    [sphinx](http://sphinxcontrib-napoleon.readthedocs.org/en/latest/).
+    See the
+    [examples](http://sphinxcontrib-napoleon.readthedocs.io/en/latest/example_google.html)
+    in the sphinx documentation.
+-   **Imports**:
+    -   Imports should be sorted by `isort` as described above.
+    -   Prefer to import classes and functions rather than packages or
+        modules.
+
+        Example:
+
+            from synapse.types import UserID
+            ...
+            user_id = UserID(local, server)
+
+        is preferred over:
+
+            from synapse import types
+            ...
+            user_id = types.UserID(local, server)
+
+        (or any other variant).
+
+        This goes against the advice in the Google style guide, but it
+        means that errors in the name are caught early (at import time).
+
+    -   Avoid wildcard imports (`from synapse.types import *`) and
+        relative imports (`from .types import UserID`).
+
+## Configuration file format
+
+The [sample configuration file](./sample_config.yaml) acts as a
+reference to Synapse's configuration options for server administrators.
+Remember that many readers will be unfamiliar with YAML and server
+administration in general, so that it is important that the file be as
+easy to understand as possible, which includes following a consistent
+format.
+
+Some guidelines follow:
+
+-   Sections should be separated with a heading consisting of a single
+    line prefixed and suffixed with `##`. There should be **two** blank
+    lines before the section header, and **one** after.
+-   Each option should be listed in the file with the following format:
+    -   A comment describing the setting. Each line of this comment
+        should be prefixed with a hash (`#`) and a space.
+
+        The comment should describe the default behaviour (ie, what
+        happens if the setting is omitted), as well as what the effect
+        will be if the setting is changed.
+
+        Often, the comment end with something like "uncomment the
+        following to <do action>".
+
+    -   A line consisting of only `#`.
+    -   A commented-out example setting, prefixed with only `#`.
+
+        For boolean (on/off) options, convention is that this example
+        should be the *opposite* to the default (so the comment will end
+        with "Uncomment the following to enable [or disable]
+        <feature>." For other options, the example should give some
+        non-default value which is likely to be useful to the reader.
+
+-   There should be a blank line between each option.
+-   Where several settings are grouped into a single dict, *avoid* the
+    convention where the whole block is commented out, resulting in
+    comment lines starting `# #`, as this is hard to read and confusing
+    to edit. Instead, leave the top-level config option uncommented, and
+    follow the conventions above for sub-options. Ensure that your code
+    correctly handles the top-level option being set to `None` (as it
+    will be if no sub-options are enabled).
+-   Lines should be wrapped at 80 characters.
+
+Example:
+
+    ## Frobnication ##
+
+    # The frobnicator will ensure that all requests are fully frobnicated.
+    # To enable it, uncomment the following.
+    #
+    #frobnicator_enabled: true
+
+    # By default, the frobnicator will frobnicate with the default frobber.
+    # The following will make it use an alternative frobber.
+    #
+    #frobincator_frobber: special_frobber
+
+    # Settings for the frobber
+    #
+    frobber:
+       # frobbing speed. Defaults to 1.
+       #
+       #speed: 10
+
+       # frobbing distance. Defaults to 1000.
+       #
+       #distance: 100
+
+Note that the sample configuration is generated from the synapse code
+and is maintained by a script, `scripts-dev/generate_sample_config`.
+Making sure that the output from this script matches the desired format
+is left as an exercise for the reader!
diff --git a/docs/code_style.rst b/docs/code_style.rst
deleted file mode 100644
index 39ac4ebedc..0000000000
--- a/docs/code_style.rst
+++ /dev/null
@@ -1,180 +0,0 @@
-Code Style
-==========
-
-Formatting tools
-----------------
-
-The Synapse codebase uses a number of code formatting tools in order to
-quickly and automatically check for formatting (and sometimes logical) errors
-in code.
-
-The necessary tools are detailed below.
-
-- **black**
-
-  The Synapse codebase uses `black <https://pypi.org/project/black/>`_ as an
-  opinionated code formatter, ensuring all comitted code is properly
-  formatted.
-
-  First install ``black`` with::
-
-    pip install --upgrade black
-
-  Have ``black`` auto-format your code (it shouldn't change any functionality)
-  with::
-
-    black . --exclude="\.tox|build|env"
-
-- **flake8**
-
-  ``flake8`` is a code checking tool. We require code to pass ``flake8`` before being merged into the codebase.
-
-  Install ``flake8`` with::
-
-    pip install --upgrade flake8
-
-  Check all application and test code with::
-
-    flake8 synapse tests
-
-- **isort**
-
-  ``isort`` ensures imports are nicely formatted, and can suggest and
-  auto-fix issues such as double-importing.
-
-  Install ``isort`` with::
-
-    pip install --upgrade isort
-
-  Auto-fix imports with::
-
-    isort -rc synapse tests
-
-  ``-rc`` means to recursively search the given directories.
-
-It's worth noting that modern IDEs and text editors can run these tools
-automatically on save. It may be worth looking into whether this
-functionality is supported in your editor for a more convenient development
-workflow. It is not, however, recommended to run ``flake8`` on save as it
-takes a while and is very resource intensive.
-
-General rules
--------------
-
-- **Naming**:
-
-  - Use camel case for class and type names
-  - Use underscores for functions and variables.
-
-- **Docstrings**: should follow the `google code style
-  <https://google.github.io/styleguide/pyguide.html#38-comments-and-docstrings>`_.
-  This is so that we can generate documentation with `sphinx
-  <http://sphinxcontrib-napoleon.readthedocs.org/en/latest/>`_. See the
-  `examples
-  <http://sphinxcontrib-napoleon.readthedocs.io/en/latest/example_google.html>`_
-  in the sphinx documentation.
-
-- **Imports**:
-
-  - Imports should be sorted by ``isort`` as described above.
-
-  - Prefer to import classes and functions rather than packages or modules.
-
-    Example::
-
-      from synapse.types import UserID
-      ...
-      user_id = UserID(local, server)
-
-    is preferred over::
-
-      from synapse import types
-      ...
-      user_id = types.UserID(local, server)
-
-    (or any other variant).
-
-    This goes against the advice in the Google style guide, but it means that
-    errors in the name are caught early (at import time).
-
-  - Avoid wildcard imports (``from synapse.types import *``) and relative
-    imports (``from .types import UserID``).
-
-Configuration file format
--------------------------
-
-The `sample configuration file <./sample_config.yaml>`_ acts as a reference to
-Synapse's configuration options for server administrators. Remember that many
-readers will be unfamiliar with YAML and server administration in general, so
-that it is important that the file be as easy to understand as possible, which
-includes following a consistent format.
-
-Some guidelines follow:
-
-* Sections should be separated with a heading consisting of a single line
-  prefixed and suffixed with ``##``. There should be **two** blank lines
-  before the section header, and **one** after.
-
-* Each option should be listed in the file with the following format:
-
-  * A comment describing the setting. Each line of this comment should be
-    prefixed with a hash (``#``) and a space.
-
-    The comment should describe the default behaviour (ie, what happens if
-    the setting is omitted), as well as what the effect will be if the
-    setting is changed.
-
-    Often, the comment end with something like "uncomment the
-    following to \<do action>".
-
-  * A line consisting of only ``#``.
-
-  * A commented-out example setting, prefixed with only ``#``.
-
-    For boolean (on/off) options, convention is that this example should be
-    the *opposite* to the default (so the comment will end with "Uncomment
-    the following to enable [or disable] \<feature\>." For other options,
-    the example should give some non-default value which is likely to be
-    useful to the reader.
-
-* There should be a blank line between each option.
-
-* Where several settings are grouped into a single dict, *avoid* the
-  convention where the whole block is commented out, resulting in comment
-  lines starting ``# #``, as this is hard to read and confusing to
-  edit. Instead, leave the top-level config option uncommented, and follow
-  the conventions above for sub-options. Ensure that your code correctly
-  handles the top-level option being set to ``None`` (as it will be if no
-  sub-options are enabled).
-
-* Lines should be wrapped at 80 characters.
-
-Example::
-
-    ## Frobnication ##
-
-    # The frobnicator will ensure that all requests are fully frobnicated.
-    # To enable it, uncomment the following.
-    #
-    #frobnicator_enabled: true
-
-    # By default, the frobnicator will frobnicate with the default frobber.
-    # The following will make it use an alternative frobber.
-    #
-    #frobincator_frobber: special_frobber
-
-    # Settings for the frobber
-    #
-    frobber:
-       # frobbing speed. Defaults to 1.
-       #
-       #speed: 10
-
-       # frobbing distance. Defaults to 1000.
-       #
-       #distance: 100
-
-Note that the sample configuration is generated from the synapse code and is
-maintained by a script, ``scripts-dev/generate_sample_config``. Making sure
-that the output from this script matches the desired format is left as an
-exercise for the reader!
diff --git a/docs/dev/saml.md b/docs/dev/saml.md
new file mode 100644
index 0000000000..f41aadce47
--- /dev/null
+++ b/docs/dev/saml.md
@@ -0,0 +1,37 @@
+# How to test SAML as a developer without a server
+
+https://capriza.github.io/samling/samling.html (https://github.com/capriza/samling) is a great
+resource for being able to tinker with the SAML options within Synapse without needing to
+deploy and configure a complicated software stack.
+
+To make Synapse (and therefore Riot) use it:
+
+1. Use the samling.html URL above or deploy your own and visit the IdP Metadata tab.
+2. Copy the XML to your clipboard.
+3. On your Synapse server, create a new file `samling.xml` next to your `homeserver.yaml` with
+   the XML from step 2 as the contents.
+4. Edit your `homeserver.yaml` to include:
+   ```yaml
+   saml2_config:
+     sp_config:
+       allow_unknown_attributes: true  # Works around a bug with AVA Hashes: https://github.com/IdentityPython/pysaml2/issues/388
+       metadata:
+         local: ["samling.xml"]   
+   ```
+5. Run `apt-get install xmlsec1` and `pip install --upgrade --force 'pysaml2>=4.5.0'` to ensure
+   the dependencies are installed and ready to go.
+6. Restart Synapse.
+
+Then in Riot:
+
+1. Visit the login page with a Riot pointing at your homeserver.
+2. Click the Single Sign-On button.
+3. On the samling page, enter a Name Identifier and add a SAML Attribute for `uid=your_localpart`.
+   The response must also be signed.
+4. Click "Next".
+5. Click "Post Response" (change nothing).
+6. You should be logged in.
+
+If you try and repeat this process, you may be automatically logged in using the information you
+gave previously. To fix this, open your developer console (`F12` or `Ctrl+Shift+I`) while on the
+samling page and clear the site data. In Chrome, this will be a button on the Application tab.
diff --git a/docs/federate.md b/docs/federate.md
index 6d6bb85e15..193e2d2dfe 100644
--- a/docs/federate.md
+++ b/docs/federate.md
@@ -148,7 +148,7 @@ We no longer actively recommend against using a reverse proxy. Many admins will
 find it easier to direct federation traffic to a reverse proxy and manage their
 own TLS certificates, and this is a supported configuration.
 
-See [reverse_proxy.rst](reverse_proxy.rst) for information on setting up a
+See [reverse_proxy.md](reverse_proxy.md) for information on setting up a
 reverse proxy.
 
 #### Do I still need to give my TLS certificates to Synapse if I am using a reverse proxy?
@@ -184,7 +184,7 @@ a complicated dance which requires connections in both directions).
 
 Another common problem is that people on other servers can't join rooms that
 you invite them to. This can be caused by an incorrectly-configured reverse
-proxy: see [reverse_proxy.rst](<reverse_proxy.rst>) for instructions on how to correctly
+proxy: see [reverse_proxy.md](<reverse_proxy.md>) for instructions on how to correctly
 configure a reverse proxy.
 
 ## Running a Demo Federation of Synapses
diff --git a/docs/log_contexts.md b/docs/log_contexts.md
new file mode 100644
index 0000000000..5331e8c88b
--- /dev/null
+++ b/docs/log_contexts.md
@@ -0,0 +1,494 @@
+# Log Contexts
+
+To help track the processing of individual requests, synapse uses a
+'`log context`' to track which request it is handling at any given
+moment. This is done via a thread-local variable; a `logging.Filter` is
+then used to fish the information back out of the thread-local variable
+and add it to each log record.
+
+Logcontexts are also used for CPU and database accounting, so that we
+can track which requests were responsible for high CPU use or database
+activity.
+
+The `synapse.logging.context` module provides a facilities for managing
+the current log context (as well as providing the `LoggingContextFilter`
+class).
+
+Deferreds make the whole thing complicated, so this document describes
+how it all works, and how to write code which follows the rules.
+
+##Logcontexts without Deferreds
+
+In the absence of any Deferred voodoo, things are simple enough. As with
+any code of this nature, the rule is that our function should leave
+things as it found them:
+
+```python
+from synapse.logging import context         # omitted from future snippets
+
+def handle_request(request_id):
+    request_context = context.LoggingContext()
+
+    calling_context = context.LoggingContext.current_context()
+    context.LoggingContext.set_current_context(request_context)
+    try:
+        request_context.request = request_id
+        do_request_handling()
+        logger.debug("finished")
+    finally:
+        context.LoggingContext.set_current_context(calling_context)
+
+def do_request_handling():
+    logger.debug("phew")  # this will be logged against request_id
+```
+
+LoggingContext implements the context management methods, so the above
+can be written much more succinctly as:
+
+```python
+def handle_request(request_id):
+    with context.LoggingContext() as request_context:
+        request_context.request = request_id
+        do_request_handling()
+        logger.debug("finished")
+
+def do_request_handling():
+    logger.debug("phew")
+```
+
+## Using logcontexts with Deferreds
+
+Deferreds --- and in particular, `defer.inlineCallbacks` --- break the
+linear flow of code so that there is no longer a single entry point
+where we should set the logcontext and a single exit point where we
+should remove it.
+
+Consider the example above, where `do_request_handling` needs to do some
+blocking operation, and returns a deferred:
+
+```python
+@defer.inlineCallbacks
+def handle_request(request_id):
+    with context.LoggingContext() as request_context:
+        request_context.request = request_id
+        yield do_request_handling()
+        logger.debug("finished")
+```
+
+In the above flow:
+
+-   The logcontext is set
+-   `do_request_handling` is called, and returns a deferred
+-   `handle_request` yields the deferred
+-   The `inlineCallbacks` wrapper of `handle_request` returns a deferred
+
+So we have stopped processing the request (and will probably go on to
+start processing the next), without clearing the logcontext.
+
+To circumvent this problem, synapse code assumes that, wherever you have
+a deferred, you will want to yield on it. To that end, whereever
+functions return a deferred, we adopt the following conventions:
+
+**Rules for functions returning deferreds:**
+
+> -   If the deferred is already complete, the function returns with the
+>     same logcontext it started with.
+> -   If the deferred is incomplete, the function clears the logcontext
+>     before returning; when the deferred completes, it restores the
+>     logcontext before running any callbacks.
+
+That sounds complicated, but actually it means a lot of code (including
+the example above) "just works". There are two cases:
+
+-   If `do_request_handling` returns a completed deferred, then the
+    logcontext will still be in place. In this case, execution will
+    continue immediately after the `yield`; the "finished" line will
+    be logged against the right context, and the `with` block restores
+    the original context before we return to the caller.
+-   If the returned deferred is incomplete, `do_request_handling` clears
+    the logcontext before returning. The logcontext is therefore clear
+    when `handle_request` yields the deferred. At that point, the
+    `inlineCallbacks` wrapper adds a callback to the deferred, and
+    returns another (incomplete) deferred to the caller, and it is safe
+    to begin processing the next request.
+
+    Once `do_request_handling`'s deferred completes, it will reinstate
+    the logcontext, before running the callback added by the
+    `inlineCallbacks` wrapper. That callback runs the second half of
+    `handle_request`, so again the "finished" line will be logged
+    against the right context, and the `with` block restores the
+    original context.
+
+As an aside, it's worth noting that `handle_request` follows our rules
+-though that only matters if the caller has its own logcontext which it
+cares about.
+
+The following sections describe pitfalls and helpful patterns when
+implementing these rules.
+
+Always yield your deferreds
+---------------------------
+
+Whenever you get a deferred back from a function, you should `yield` on
+it as soon as possible. (Returning it directly to your caller is ok too,
+if you're not doing `inlineCallbacks`.) Do not pass go; do not do any
+logging; do not call any other functions.
+
+```python
+@defer.inlineCallbacks
+def fun():
+    logger.debug("starting")
+    yield do_some_stuff()       # just like this
+
+    d = more_stuff()
+    result = yield d            # also fine, of course
+
+    return result
+
+def nonInlineCallbacksFun():
+    logger.debug("just a wrapper really")
+    return do_some_stuff()      # this is ok too - the caller will yield on
+                                # it anyway.
+```
+
+Provided this pattern is followed all the way back up to the callchain
+to where the logcontext was set, this will make things work out ok:
+provided `do_some_stuff` and `more_stuff` follow the rules above, then
+so will `fun` (as wrapped by `inlineCallbacks`) and
+`nonInlineCallbacksFun`.
+
+It's all too easy to forget to `yield`: for instance if we forgot that
+`do_some_stuff` returned a deferred, we might plough on regardless. This
+leads to a mess; it will probably work itself out eventually, but not
+before a load of stuff has been logged against the wrong context.
+(Normally, other things will break, more obviously, if you forget to
+`yield`, so this tends not to be a major problem in practice.)
+
+Of course sometimes you need to do something a bit fancier with your
+Deferreds - not all code follows the linear A-then-B-then-C pattern.
+Notes on implementing more complex patterns are in later sections.
+
+## Where you create a new Deferred, make it follow the rules
+
+Most of the time, a Deferred comes from another synapse function.
+Sometimes, though, we need to make up a new Deferred, or we get a
+Deferred back from external code. We need to make it follow our rules.
+
+The easy way to do it is with a combination of `defer.inlineCallbacks`,
+and `context.PreserveLoggingContext`. Suppose we want to implement
+`sleep`, which returns a deferred which will run its callbacks after a
+given number of seconds. That might look like:
+
+```python
+# not a logcontext-rules-compliant function
+def get_sleep_deferred(seconds):
+    d = defer.Deferred()
+    reactor.callLater(seconds, d.callback, None)
+    return d
+```
+
+That doesn't follow the rules, but we can fix it by wrapping it with
+`PreserveLoggingContext` and `yield` ing on it:
+
+```python
+@defer.inlineCallbacks
+def sleep(seconds):
+    with PreserveLoggingContext():
+        yield get_sleep_deferred(seconds)
+```
+
+This technique works equally for external functions which return
+deferreds, or deferreds we have made ourselves.
+
+You can also use `context.make_deferred_yieldable`, which just does the
+boilerplate for you, so the above could be written:
+
+```python
+def sleep(seconds):
+    return context.make_deferred_yieldable(get_sleep_deferred(seconds))
+```
+
+## Fire-and-forget
+
+Sometimes you want to fire off a chain of execution, but not wait for
+its result. That might look a bit like this:
+
+```python
+@defer.inlineCallbacks
+def do_request_handling():
+    yield foreground_operation()
+
+    # *don't* do this
+    background_operation()
+
+    logger.debug("Request handling complete")
+
+@defer.inlineCallbacks
+def background_operation():
+    yield first_background_step()
+    logger.debug("Completed first step")
+    yield second_background_step()
+    logger.debug("Completed second step")
+```
+
+The above code does a couple of steps in the background after
+`do_request_handling` has finished. The log lines are still logged
+against the `request_context` logcontext, which may or may not be
+desirable. There are two big problems with the above, however. The first
+problem is that, if `background_operation` returns an incomplete
+Deferred, it will expect its caller to `yield` immediately, so will have
+cleared the logcontext. In this example, that means that 'Request
+handling complete' will be logged without any context.
+
+The second problem, which is potentially even worse, is that when the
+Deferred returned by `background_operation` completes, it will restore
+the original logcontext. There is nothing waiting on that Deferred, so
+the logcontext will leak into the reactor and possibly get attached to
+some arbitrary future operation.
+
+There are two potential solutions to this.
+
+One option is to surround the call to `background_operation` with a
+`PreserveLoggingContext` call. That will reset the logcontext before
+starting `background_operation` (so the context restored when the
+deferred completes will be the empty logcontext), and will restore the
+current logcontext before continuing the foreground process:
+
+```python
+@defer.inlineCallbacks
+def do_request_handling():
+    yield foreground_operation()
+
+    # start background_operation off in the empty logcontext, to
+    # avoid leaking the current context into the reactor.
+    with PreserveLoggingContext():
+        background_operation()
+
+    # this will now be logged against the request context
+    logger.debug("Request handling complete")
+```
+
+Obviously that option means that the operations done in
+`background_operation` would be not be logged against a logcontext
+(though that might be fixed by setting a different logcontext via a
+`with LoggingContext(...)` in `background_operation`).
+
+The second option is to use `context.run_in_background`, which wraps a
+function so that it doesn't reset the logcontext even when it returns
+an incomplete deferred, and adds a callback to the returned deferred to
+reset the logcontext. In other words, it turns a function that follows
+the Synapse rules about logcontexts and Deferreds into one which behaves
+more like an external function --- the opposite operation to that
+described in the previous section. It can be used like this:
+
+```python
+@defer.inlineCallbacks
+def do_request_handling():
+    yield foreground_operation()
+
+    context.run_in_background(background_operation)
+
+    # this will now be logged against the request context
+    logger.debug("Request handling complete")
+```
+
+## Passing synapse deferreds into third-party functions
+
+A typical example of this is where we want to collect together two or
+more deferred via `defer.gatherResults`:
+
+```python
+d1 = operation1()
+d2 = operation2()
+d3 = defer.gatherResults([d1, d2])
+```
+
+This is really a variation of the fire-and-forget problem above, in that
+we are firing off `d1` and `d2` without yielding on them. The difference
+is that we now have third-party code attached to their callbacks. Anyway
+either technique given in the [Fire-and-forget](#fire-and-forget)
+section will work.
+
+Of course, the new Deferred returned by `gatherResults` needs to be
+wrapped in order to make it follow the logcontext rules before we can
+yield it, as described in [Where you create a new Deferred, make it
+follow the
+rules](#where-you-create-a-new-deferred-make-it-follow-the-rules).
+
+So, option one: reset the logcontext before starting the operations to
+be gathered:
+
+```python
+@defer.inlineCallbacks
+def do_request_handling():
+    with PreserveLoggingContext():
+        d1 = operation1()
+        d2 = operation2()
+        result = yield defer.gatherResults([d1, d2])
+```
+
+In this case particularly, though, option two, of using
+`context.preserve_fn` almost certainly makes more sense, so that
+`operation1` and `operation2` are both logged against the original
+logcontext. This looks like:
+
+```python
+@defer.inlineCallbacks
+def do_request_handling():
+    d1 = context.preserve_fn(operation1)()
+    d2 = context.preserve_fn(operation2)()
+
+    with PreserveLoggingContext():
+        result = yield defer.gatherResults([d1, d2])
+```
+
+## Was all this really necessary?
+
+The conventions used work fine for a linear flow where everything
+happens in series via `defer.inlineCallbacks` and `yield`, but are
+certainly tricky to follow for any more exotic flows. It's hard not to
+wonder if we could have done something else.
+
+We're not going to rewrite Synapse now, so the following is entirely of
+academic interest, but I'd like to record some thoughts on an
+alternative approach.
+
+I briefly prototyped some code following an alternative set of rules. I
+think it would work, but I certainly didn't get as far as thinking how
+it would interact with concepts as complicated as the cache descriptors.
+
+My alternative rules were:
+
+-   functions always preserve the logcontext of their caller, whether or
+    not they are returning a Deferred.
+-   Deferreds returned by synapse functions run their callbacks in the
+    same context as the function was orignally called in.
+
+The main point of this scheme is that everywhere that sets the
+logcontext is responsible for clearing it before returning control to
+the reactor.
+
+So, for example, if you were the function which started a
+`with LoggingContext` block, you wouldn't `yield` within it --- instead
+you'd start off the background process, and then leave the `with` block
+to wait for it:
+
+```python
+def handle_request(request_id):
+    with context.LoggingContext() as request_context:
+        request_context.request = request_id
+        d = do_request_handling()
+
+    def cb(r):
+        logger.debug("finished")
+
+    d.addCallback(cb)
+    return d
+```
+
+(in general, mixing `with LoggingContext` blocks and
+`defer.inlineCallbacks` in the same function leads to slighly
+counter-intuitive code, under this scheme).
+
+Because we leave the original `with` block as soon as the Deferred is
+returned (as opposed to waiting for it to be resolved, as we do today),
+the logcontext is cleared before control passes back to the reactor; so
+if there is some code within `do_request_handling` which needs to wait
+for a Deferred to complete, there is no need for it to worry about
+clearing the logcontext before doing so:
+
+```python
+def handle_request():
+    r = do_some_stuff()
+    r.addCallback(do_some_more_stuff)
+    return r
+```
+
+--- and provided `do_some_stuff` follows the rules of returning a
+Deferred which runs its callbacks in the original logcontext, all is
+happy.
+
+The business of a Deferred which runs its callbacks in the original
+logcontext isn't hard to achieve --- we have it today, in the shape of
+`context._PreservingContextDeferred`:
+
+```python
+def do_some_stuff():
+    deferred = do_some_io()
+    pcd = _PreservingContextDeferred(LoggingContext.current_context())
+    deferred.chainDeferred(pcd)
+    return pcd
+```
+
+It turns out that, thanks to the way that Deferreds chain together, we
+automatically get the property of a context-preserving deferred with
+`defer.inlineCallbacks`, provided the final Defered the function
+`yields` on has that property. So we can just write:
+
+```python
+@defer.inlineCallbacks
+def handle_request():
+    yield do_some_stuff()
+    yield do_some_more_stuff()
+```
+
+To conclude: I think this scheme would have worked equally well, with
+less danger of messing it up, and probably made some more esoteric code
+easier to write. But again --- changing the conventions of the entire
+Synapse codebase is not a sensible option for the marginal improvement
+offered.
+
+## A note on garbage-collection of Deferred chains
+
+It turns out that our logcontext rules do not play nicely with Deferred
+chains which get orphaned and garbage-collected.
+
+Imagine we have some code that looks like this:
+
+```python
+listener_queue = []
+
+def on_something_interesting():
+    for d in listener_queue:
+        d.callback("foo")
+
+@defer.inlineCallbacks
+def await_something_interesting():
+    new_deferred = defer.Deferred()
+    listener_queue.append(new_deferred)
+
+    with PreserveLoggingContext():
+        yield new_deferred
+```
+
+Obviously, the idea here is that we have a bunch of things which are
+waiting for an event. (It's just an example of the problem here, but a
+relatively common one.)
+
+Now let's imagine two further things happen. First of all, whatever was
+waiting for the interesting thing goes away. (Perhaps the request times
+out, or something *even more* interesting happens.)
+
+Secondly, let's suppose that we decide that the interesting thing is
+never going to happen, and we reset the listener queue:
+
+```python
+def reset_listener_queue():
+    listener_queue.clear()
+```
+
+So, both ends of the deferred chain have now dropped their references,
+and the deferred chain is now orphaned, and will be garbage-collected at
+some point. Note that `await_something_interesting` is a generator
+function, and when Python garbage-collects generator functions, it gives
+them a chance to clean up by making the `yield` raise a `GeneratorExit`
+exception. In our case, that means that the `__exit__` handler of
+`PreserveLoggingContext` will carefully restore the request context, but
+there is now nothing waiting for its return, so the request context is
+never cleared.
+
+To reiterate, this problem only arises when *both* ends of a deferred
+chain are dropped. Dropping the the reference to a deferred you're
+supposed to be calling is probably bad practice, so this doesn't
+actually happen too much. Unfortunately, when it does happen, it will
+lead to leaked logcontexts which are incredibly hard to track down.
diff --git a/docs/log_contexts.rst b/docs/log_contexts.rst
deleted file mode 100644
index 4502cd9454..0000000000
--- a/docs/log_contexts.rst
+++ /dev/null
@@ -1,498 +0,0 @@
-Log Contexts
-============
-
-.. contents::
-
-To help track the processing of individual requests, synapse uses a
-'log context' to track which request it is handling at any given moment. This
-is done via a thread-local variable; a ``logging.Filter`` is then used to fish
-the information back out of the thread-local variable and add it to each log
-record.
-
-Logcontexts are also used for CPU and database accounting, so that we can track
-which requests were responsible for high CPU use or database activity.
-
-The ``synapse.logging.context`` module provides a facilities for managing the
-current log context (as well as providing the ``LoggingContextFilter`` class).
-
-Deferreds make the whole thing complicated, so this document describes how it
-all works, and how to write code which follows the rules.
-
-Logcontexts without Deferreds
------------------------------
-
-In the absence of any Deferred voodoo, things are simple enough. As with any
-code of this nature, the rule is that our function should leave things as it
-found them:
-
-.. code:: python
-
-    from synapse.logging import context         # omitted from future snippets
-
-    def handle_request(request_id):
-        request_context = context.LoggingContext()
-
-        calling_context = context.LoggingContext.current_context()
-        context.LoggingContext.set_current_context(request_context)
-        try:
-            request_context.request = request_id
-            do_request_handling()
-            logger.debug("finished")
-        finally:
-            context.LoggingContext.set_current_context(calling_context)
-
-    def do_request_handling():
-        logger.debug("phew")  # this will be logged against request_id
-
-
-LoggingContext implements the context management methods, so the above can be
-written much more succinctly as:
-
-.. code:: python
-
-    def handle_request(request_id):
-        with context.LoggingContext() as request_context:
-            request_context.request = request_id
-            do_request_handling()
-            logger.debug("finished")
-
-    def do_request_handling():
-        logger.debug("phew")
-
-
-Using logcontexts with Deferreds
---------------------------------
-
-Deferreds — and in particular, ``defer.inlineCallbacks`` — break
-the linear flow of code so that there is no longer a single entry point where
-we should set the logcontext and a single exit point where we should remove it.
-
-Consider the example above, where ``do_request_handling`` needs to do some
-blocking operation, and returns a deferred:
-
-.. code:: python
-
-    @defer.inlineCallbacks
-    def handle_request(request_id):
-        with context.LoggingContext() as request_context:
-            request_context.request = request_id
-            yield do_request_handling()
-            logger.debug("finished")
-
-
-In the above flow:
-
-* The logcontext is set
-* ``do_request_handling`` is called, and returns a deferred
-* ``handle_request`` yields the deferred
-* The ``inlineCallbacks`` wrapper of ``handle_request`` returns a deferred
-
-So we have stopped processing the request (and will probably go on to start
-processing the next), without clearing the logcontext.
-
-To circumvent this problem, synapse code assumes that, wherever you have a
-deferred, you will want to yield on it. To that end, whereever functions return
-a deferred, we adopt the following conventions:
-
-**Rules for functions returning deferreds:**
-
-  * If the deferred is already complete, the function returns with the same
-    logcontext it started with.
-  * If the deferred is incomplete, the function clears the logcontext before
-    returning; when the deferred completes, it restores the logcontext before
-    running any callbacks.
-
-That sounds complicated, but actually it means a lot of code (including the
-example above) "just works". There are two cases:
-
-* If ``do_request_handling`` returns a completed deferred, then the logcontext
-  will still be in place. In this case, execution will continue immediately
-  after the ``yield``; the "finished" line will be logged against the right
-  context, and the ``with`` block restores the original context before we
-  return to the caller.
-
-* If the returned deferred is incomplete, ``do_request_handling`` clears the
-  logcontext before returning. The logcontext is therefore clear when
-  ``handle_request`` yields the deferred. At that point, the ``inlineCallbacks``
-  wrapper adds a callback to the deferred, and returns another (incomplete)
-  deferred to the caller, and it is safe to begin processing the next request.
-
-  Once ``do_request_handling``'s deferred completes, it will reinstate the
-  logcontext, before running the callback added by the ``inlineCallbacks``
-  wrapper. That callback runs the second half of ``handle_request``, so again
-  the "finished" line will be logged against the right
-  context, and the ``with`` block restores the original context.
-
-As an aside, it's worth noting that ``handle_request`` follows our rules -
-though that only matters if the caller has its own logcontext which it cares
-about.
-
-The following sections describe pitfalls and helpful patterns when implementing
-these rules.
-
-Always yield your deferreds
----------------------------
-
-Whenever you get a deferred back from a function, you should ``yield`` on it
-as soon as possible. (Returning it directly to your caller is ok too, if you're
-not doing ``inlineCallbacks``.) Do not pass go; do not do any logging; do not
-call any other functions.
-
-.. code:: python
-
-    @defer.inlineCallbacks
-    def fun():
-        logger.debug("starting")
-        yield do_some_stuff()       # just like this
-
-        d = more_stuff()
-        result = yield d            # also fine, of course
-
-        return result
-
-    def nonInlineCallbacksFun():
-        logger.debug("just a wrapper really")
-        return do_some_stuff()      # this is ok too - the caller will yield on
-                                    # it anyway.
-
-Provided this pattern is followed all the way back up to the callchain to where
-the logcontext was set, this will make things work out ok: provided
-``do_some_stuff`` and ``more_stuff`` follow the rules above, then so will
-``fun`` (as wrapped by ``inlineCallbacks``) and ``nonInlineCallbacksFun``.
-
-It's all too easy to forget to ``yield``: for instance if we forgot that
-``do_some_stuff`` returned a deferred, we might plough on regardless. This
-leads to a mess; it will probably work itself out eventually, but not before
-a load of stuff has been logged against the wrong context. (Normally, other
-things will break, more obviously, if you forget to ``yield``, so this tends
-not to be a major problem in practice.)
-
-Of course sometimes you need to do something a bit fancier with your Deferreds
-- not all code follows the linear A-then-B-then-C pattern. Notes on
-implementing more complex patterns are in later sections.
-
-Where you create a new Deferred, make it follow the rules
----------------------------------------------------------
-
-Most of the time, a Deferred comes from another synapse function. Sometimes,
-though, we need to make up a new Deferred, or we get a Deferred back from
-external code. We need to make it follow our rules.
-
-The easy way to do it is with a combination of ``defer.inlineCallbacks``, and
-``context.PreserveLoggingContext``. Suppose we want to implement ``sleep``,
-which returns a deferred which will run its callbacks after a given number of
-seconds. That might look like:
-
-.. code:: python
-
-    # not a logcontext-rules-compliant function
-    def get_sleep_deferred(seconds):
-        d = defer.Deferred()
-        reactor.callLater(seconds, d.callback, None)
-        return d
-
-That doesn't follow the rules, but we can fix it by wrapping it with
-``PreserveLoggingContext`` and ``yield`` ing on it:
-
-.. code:: python
-
-    @defer.inlineCallbacks
-    def sleep(seconds):
-        with PreserveLoggingContext():
-            yield get_sleep_deferred(seconds)
-
-This technique works equally for external functions which return deferreds,
-or deferreds we have made ourselves.
-
-You can also use ``context.make_deferred_yieldable``, which just does the
-boilerplate for you, so the above could be written:
-
-.. code:: python
-
-    def sleep(seconds):
-        return context.make_deferred_yieldable(get_sleep_deferred(seconds))
-
-
-Fire-and-forget
----------------
-
-Sometimes you want to fire off a chain of execution, but not wait for its
-result. That might look a bit like this:
-
-.. code:: python
-
-    @defer.inlineCallbacks
-    def do_request_handling():
-        yield foreground_operation()
-
-        # *don't* do this
-        background_operation()
-
-        logger.debug("Request handling complete")
-
-    @defer.inlineCallbacks
-    def background_operation():
-        yield first_background_step()
-        logger.debug("Completed first step")
-        yield second_background_step()
-        logger.debug("Completed second step")
-
-The above code does a couple of steps in the background after
-``do_request_handling`` has finished. The log lines are still logged against
-the ``request_context`` logcontext, which may or may not be desirable. There
-are two big problems with the above, however. The first problem is that, if
-``background_operation`` returns an incomplete Deferred, it will expect its
-caller to ``yield`` immediately, so will have cleared the logcontext. In this
-example, that means that 'Request handling complete' will be logged without any
-context.
-
-The second problem, which is potentially even worse, is that when the Deferred
-returned by ``background_operation`` completes, it will restore the original
-logcontext. There is nothing waiting on that Deferred, so the logcontext will
-leak into the reactor and possibly get attached to some arbitrary future
-operation.
-
-There are two potential solutions to this.
-
-One option is to surround the call to ``background_operation`` with a
-``PreserveLoggingContext`` call. That will reset the logcontext before
-starting ``background_operation`` (so the context restored when the deferred
-completes will be the empty logcontext), and will restore the current
-logcontext before continuing the foreground process:
-
-.. code:: python
-
-    @defer.inlineCallbacks
-    def do_request_handling():
-        yield foreground_operation()
-
-        # start background_operation off in the empty logcontext, to
-        # avoid leaking the current context into the reactor.
-        with PreserveLoggingContext():
-            background_operation()
-
-        # this will now be logged against the request context
-        logger.debug("Request handling complete")
-
-Obviously that option means that the operations done in
-``background_operation`` would be not be logged against a logcontext (though
-that might be fixed by setting a different logcontext via a ``with
-LoggingContext(...)`` in ``background_operation``).
-
-The second option is to use ``context.run_in_background``, which wraps a
-function so that it doesn't reset the logcontext even when it returns an
-incomplete deferred, and adds a callback to the returned deferred to reset the
-logcontext. In other words, it turns a function that follows the Synapse rules
-about logcontexts and Deferreds into one which behaves more like an external
-function — the opposite operation to that described in the previous section.
-It can be used like this:
-
-.. code:: python
-
-    @defer.inlineCallbacks
-    def do_request_handling():
-        yield foreground_operation()
-
-        context.run_in_background(background_operation)
-
-        # this will now be logged against the request context
-        logger.debug("Request handling complete")
-
-Passing synapse deferreds into third-party functions
-----------------------------------------------------
-
-A typical example of this is where we want to collect together two or more
-deferred via ``defer.gatherResults``:
-
-.. code:: python
-
-    d1 = operation1()
-    d2 = operation2()
-    d3 = defer.gatherResults([d1, d2])
-
-This is really a variation of the fire-and-forget problem above, in that we are
-firing off ``d1`` and ``d2`` without yielding on them. The difference
-is that we now have third-party code attached to their callbacks. Anyway either
-technique given in the `Fire-and-forget`_ section will work.
-
-Of course, the new Deferred returned by ``gatherResults`` needs to be wrapped
-in order to make it follow the logcontext rules before we can yield it, as
-described in `Where you create a new Deferred, make it follow the rules`_.
-
-So, option one: reset the logcontext before starting the operations to be
-gathered:
-
-.. code:: python
-
-    @defer.inlineCallbacks
-    def do_request_handling():
-        with PreserveLoggingContext():
-            d1 = operation1()
-            d2 = operation2()
-            result = yield defer.gatherResults([d1, d2])
-
-In this case particularly, though, option two, of using
-``context.preserve_fn`` almost certainly makes more sense, so that
-``operation1`` and ``operation2`` are both logged against the original
-logcontext. This looks like:
-
-.. code:: python
-
-    @defer.inlineCallbacks
-    def do_request_handling():
-        d1 = context.preserve_fn(operation1)()
-        d2 = context.preserve_fn(operation2)()
-
-        with PreserveLoggingContext():
-            result = yield defer.gatherResults([d1, d2])
-
-
-Was all this really necessary?
-------------------------------
-
-The conventions used work fine for a linear flow where everything happens in
-series via ``defer.inlineCallbacks`` and ``yield``, but are certainly tricky to
-follow for any more exotic flows. It's hard not to wonder if we could have done
-something else.
-
-We're not going to rewrite Synapse now, so the following is entirely of
-academic interest, but I'd like to record some thoughts on an alternative
-approach.
-
-I briefly prototyped some code following an alternative set of rules. I think
-it would work, but I certainly didn't get as far as thinking how it would
-interact with concepts as complicated as the cache descriptors.
-
-My alternative rules were:
-
-* functions always preserve the logcontext of their caller, whether or not they
-  are returning a Deferred.
-
-* Deferreds returned by synapse functions run their callbacks in the same
-  context as the function was orignally called in.
-
-The main point of this scheme is that everywhere that sets the logcontext is
-responsible for clearing it before returning control to the reactor.
-
-So, for example, if you were the function which started a ``with
-LoggingContext`` block, you wouldn't ``yield`` within it — instead you'd start
-off the background process, and then leave the ``with`` block to wait for it:
-
-.. code:: python
-
-    def handle_request(request_id):
-        with context.LoggingContext() as request_context:
-            request_context.request = request_id
-            d = do_request_handling()
-
-        def cb(r):
-            logger.debug("finished")
-
-        d.addCallback(cb)
-        return d
-
-(in general, mixing ``with LoggingContext`` blocks and
-``defer.inlineCallbacks`` in the same function leads to slighly
-counter-intuitive code, under this scheme).
-
-Because we leave the original ``with`` block as soon as the Deferred is
-returned (as opposed to waiting for it to be resolved, as we do today), the
-logcontext is cleared before control passes back to the reactor; so if there is
-some code within ``do_request_handling`` which needs to wait for a Deferred to
-complete, there is no need for it to worry about clearing the logcontext before
-doing so:
-
-.. code:: python
-
-    def handle_request():
-        r = do_some_stuff()
-        r.addCallback(do_some_more_stuff)
-        return r
-
-— and provided ``do_some_stuff`` follows the rules of returning a Deferred which
-runs its callbacks in the original logcontext, all is happy.
-
-The business of a Deferred which runs its callbacks in the original logcontext
-isn't hard to achieve — we have it today, in the shape of
-``context._PreservingContextDeferred``:
-
-.. code:: python
-
-    def do_some_stuff():
-        deferred = do_some_io()
-        pcd = _PreservingContextDeferred(LoggingContext.current_context())
-        deferred.chainDeferred(pcd)
-        return pcd
-
-It turns out that, thanks to the way that Deferreds chain together, we
-automatically get the property of a context-preserving deferred with
-``defer.inlineCallbacks``, provided the final Defered the function ``yields``
-on has that property. So we can just write:
-
-.. code:: python
-
-    @defer.inlineCallbacks
-    def handle_request():
-        yield do_some_stuff()
-        yield do_some_more_stuff()
-
-To conclude: I think this scheme would have worked equally well, with less
-danger of messing it up, and probably made some more esoteric code easier to
-write. But again — changing the conventions of the entire Synapse codebase is
-not a sensible option for the marginal improvement offered.
-
-
-A note on garbage-collection of Deferred chains
------------------------------------------------
-
-It turns out that our logcontext rules do not play nicely with Deferred
-chains which get orphaned and garbage-collected.
-
-Imagine we have some code that looks like this:
-
-.. code:: python
-
-    listener_queue = []
-
-    def on_something_interesting():
-        for d in listener_queue:
-            d.callback("foo")
-
-    @defer.inlineCallbacks
-    def await_something_interesting():
-        new_deferred = defer.Deferred()
-        listener_queue.append(new_deferred)
-
-        with PreserveLoggingContext():
-            yield new_deferred
-
-Obviously, the idea here is that we have a bunch of things which are waiting
-for an event. (It's just an example of the problem here, but a relatively
-common one.)
-
-Now let's imagine two further things happen. First of all, whatever was
-waiting for the interesting thing goes away. (Perhaps the request times out,
-or something *even more* interesting happens.)
-
-Secondly, let's suppose that we decide that the interesting thing is never
-going to happen, and we reset the listener queue:
-
-.. code:: python
-
-    def reset_listener_queue():
-        listener_queue.clear()
-
-So, both ends of the deferred chain have now dropped their references, and the
-deferred chain is now orphaned, and will be garbage-collected at some point.
-Note that ``await_something_interesting`` is a generator function, and when
-Python garbage-collects generator functions, it gives them a chance to clean
-up by making the ``yield`` raise a ``GeneratorExit`` exception. In our case,
-that means that the ``__exit__`` handler of ``PreserveLoggingContext`` will
-carefully restore the request context, but there is now nothing waiting for
-its return, so the request context is never cleared.
-
-To reiterate, this problem only arises when *both* ends of a deferred chain
-are dropped. Dropping the the reference to a deferred you're supposed to be
-calling is probably bad practice, so this doesn't actually happen too much.
-Unfortunately, when it does happen, it will lead to leaked logcontexts which
-are incredibly hard to track down.
diff --git a/docs/media_repository.md b/docs/media_repository.md
new file mode 100644
index 0000000000..1bf8f16f55
--- /dev/null
+++ b/docs/media_repository.md
@@ -0,0 +1,30 @@
+# Media Repository 
+
+*Synapse implementation-specific details for the media repository*
+
+The media repository is where attachments and avatar photos are stored.
+It stores attachment content and thumbnails for media uploaded by local users.
+It caches attachment content and thumbnails for media uploaded by remote users.
+
+## Storage
+
+Each item of media is assigned a `media_id` when it is uploaded.
+The `media_id` is a randomly chosen, URL safe 24 character string.
+
+Metadata such as the MIME type, upload time and length are stored in the
+sqlite3 database indexed by `media_id`.
+
+Content is stored on the filesystem under a `"local_content"` directory.
+
+Thumbnails are stored under a `"local_thumbnails"` directory.
+
+The item with `media_id` `"aabbccccccccdddddddddddd"` is stored under
+`"local_content/aa/bb/ccccccccdddddddddddd"`. Its thumbnail with width
+`128` and height `96` and type `"image/jpeg"` is stored under
+`"local_thumbnails/aa/bb/ccccccccdddddddddddd/128-96-image-jpeg"`
+
+Remote content is cached under `"remote_content"` directory. Each item of
+remote content is assigned a local `"filesystem_id"` to ensure that the
+directory structure `"remote_content/server_name/aa/bb/ccccccccdddddddddddd"`
+is appropriate. Thumbnails for remote content are stored under
+`"remote_thumbnails/server_name/..."`
diff --git a/docs/media_repository.rst b/docs/media_repository.rst
deleted file mode 100644
index 1037b5be63..0000000000
--- a/docs/media_repository.rst
+++ /dev/null
@@ -1,27 +0,0 @@
-Media Repository 
-================
-
-*Synapse implementation-specific details for the media repository*
-
-The media repository is where attachments and avatar photos are stored.
-It stores attachment content and thumbnails for media uploaded by local users.
-It caches attachment content and thumbnails for media uploaded by remote users.
-
-Storage
--------
-
-Each item of media is assigned a ``media_id`` when it is uploaded.
-The ``media_id`` is a randomly chosen, URL safe 24 character string.
-Metadata such as the MIME type, upload time and length are stored in the
-sqlite3 database indexed by ``media_id``.
-Content is stored on the filesystem under a ``"local_content"`` directory.
-Thumbnails are stored under a ``"local_thumbnails"`` directory.
-The item with ``media_id`` ``"aabbccccccccdddddddddddd"`` is stored under
-``"local_content/aa/bb/ccccccccdddddddddddd"``. Its thumbnail with width
-``128`` and height ``96`` and type ``"image/jpeg"`` is stored under
-``"local_thumbnails/aa/bb/ccccccccdddddddddddd/128-96-image-jpeg"``
-Remote content is cached under ``"remote_content"`` directory. Each item of
-remote content is assigned a local "``filesystem_id``" to ensure that the
-directory structure ``"remote_content/server_name/aa/bb/ccccccccdddddddddddd"``
-is appropriate. Thumbnails for remote content are stored under
-``"remote_thumbnails/server_name/..."``
diff --git a/docs/metrics-howto.md b/docs/metrics-howto.md
new file mode 100644
index 0000000000..32abb9f44e
--- /dev/null
+++ b/docs/metrics-howto.md
@@ -0,0 +1,217 @@
+# How to monitor Synapse metrics using Prometheus
+
+1.  Install Prometheus:
+
+    Follow instructions at
+    <http://prometheus.io/docs/introduction/install/>
+
+1.  Enable Synapse metrics:
+
+    There are two methods of enabling metrics in Synapse.
+
+    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:
+
+        resources:
+          - names:
+            - client
+            - metrics
+
+    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.
+
+    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 `/`.
+
+    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`.
+
+1.  Restart Synapse.
+
+1.  Add a Prometheus target for Synapse.
+
+    It needs to set the `metrics_path` to a non-default value (under
+    `scrape_configs`):
+
+        - job_name: "synapse"
+          metrics_path: "/_synapse/metrics"
+          static_configs:
+            - targets: ["my.server.here:port"]
+
+    where `my.server.here` is the IP address of Synapse, and `port` is
+    the listener port configured with the `metrics` resource.
+
+    If your prometheus is older than 1.5.2, you will need to replace
+    `static_configs` in the above with `target_groups`.
+
+1.  Restart Prometheus.
+
+## Renaming of metrics & deprecation of old names in 1.2
+
+Synapse 1.2 updates the Prometheus metrics to match the naming
+convention of the upstream `prometheus_client`. The old names are
+considered deprecated and will be removed in a future version of
+Synapse.
+
+| New Name                                                                     | Old Name                                                               |
+| ---------------------------------------------------------------------------- | ---------------------------------------------------------------------- |
+| python_gc_objects_collected_total                                            | python_gc_objects_collected                                            |
+| python_gc_objects_uncollectable_total                                        | python_gc_objects_uncollectable                                        |
+| python_gc_collections_total                                                  | python_gc_collections                                                  |
+| process_cpu_seconds_total                                                    | process_cpu_seconds                                                    |
+| synapse_federation_client_sent_transactions_total                            | synapse_federation_client_sent_transactions                            |
+| synapse_federation_client_events_processed_total                             | synapse_federation_client_events_processed                             |
+| synapse_event_processing_loop_count_total                                    | synapse_event_processing_loop_count                                    |
+| synapse_event_processing_loop_room_count_total                               | synapse_event_processing_loop_room_count                               |
+| synapse_util_metrics_block_count_total                                       | synapse_util_metrics_block_count                                       |
+| synapse_util_metrics_block_time_seconds_total                                | synapse_util_metrics_block_time_seconds                                |
+| synapse_util_metrics_block_ru_utime_seconds_total                            | synapse_util_metrics_block_ru_utime_seconds                            |
+| synapse_util_metrics_block_ru_stime_seconds_total                            | synapse_util_metrics_block_ru_stime_seconds                            |
+| synapse_util_metrics_block_db_txn_count_total                                | synapse_util_metrics_block_db_txn_count                                |
+| synapse_util_metrics_block_db_txn_duration_seconds_total                     | synapse_util_metrics_block_db_txn_duration_seconds                     |
+| synapse_util_metrics_block_db_sched_duration_seconds_total                   | synapse_util_metrics_block_db_sched_duration_seconds                   |
+| synapse_background_process_start_count_total                                 | synapse_background_process_start_count                                 |
+| synapse_background_process_ru_utime_seconds_total                            | synapse_background_process_ru_utime_seconds                            |
+| synapse_background_process_ru_stime_seconds_total                            | synapse_background_process_ru_stime_seconds                            |
+| synapse_background_process_db_txn_count_total                                | synapse_background_process_db_txn_count                                |
+| synapse_background_process_db_txn_duration_seconds_total                     | synapse_background_process_db_txn_duration_seconds                     |
+| synapse_background_process_db_sched_duration_seconds_total                   | synapse_background_process_db_sched_duration_seconds                   |
+| synapse_storage_events_persisted_events_total                                | synapse_storage_events_persisted_events                                |
+| synapse_storage_events_persisted_events_sep_total                            | synapse_storage_events_persisted_events_sep                            |
+| synapse_storage_events_state_delta_total                                     | synapse_storage_events_state_delta                                     |
+| synapse_storage_events_state_delta_single_event_total                        | synapse_storage_events_state_delta_single_event                        |
+| synapse_storage_events_state_delta_reuse_delta_total                         | synapse_storage_events_state_delta_reuse_delta                         |
+| synapse_federation_server_received_pdus_total                                | synapse_federation_server_received_pdus                                |
+| synapse_federation_server_received_edus_total                                | synapse_federation_server_received_edus                                |
+| synapse_handler_presence_notified_presence_total                             | synapse_handler_presence_notified_presence                             |
+| synapse_handler_presence_federation_presence_out_total                       | synapse_handler_presence_federation_presence_out                       |
+| synapse_handler_presence_presence_updates_total                              | synapse_handler_presence_presence_updates                              |
+| synapse_handler_presence_timers_fired_total                                  | synapse_handler_presence_timers_fired                                  |
+| synapse_handler_presence_federation_presence_total                           | synapse_handler_presence_federation_presence                           |
+| synapse_handler_presence_bump_active_time_total                              | synapse_handler_presence_bump_active_time                              |
+| synapse_federation_client_sent_edus_total                                    | synapse_federation_client_sent_edus                                    |
+| synapse_federation_client_sent_pdu_destinations_count_total                  | synapse_federation_client_sent_pdu_destinations:count                  |
+| synapse_federation_client_sent_pdu_destinations_total                        | synapse_federation_client_sent_pdu_destinations:total                  |
+| synapse_handlers_appservice_events_processed_total                           | synapse_handlers_appservice_events_processed                           |
+| synapse_notifier_notified_events_total                                       | synapse_notifier_notified_events                                       |
+| synapse_push_bulk_push_rule_evaluator_push_rules_invalidation_counter_total  | synapse_push_bulk_push_rule_evaluator_push_rules_invalidation_counter  |
+| synapse_push_bulk_push_rule_evaluator_push_rules_state_size_counter_total    | synapse_push_bulk_push_rule_evaluator_push_rules_state_size_counter    |
+| synapse_http_httppusher_http_pushes_processed_total                          | synapse_http_httppusher_http_pushes_processed                          |
+| synapse_http_httppusher_http_pushes_failed_total                             | synapse_http_httppusher_http_pushes_failed                             |
+| synapse_http_httppusher_badge_updates_processed_total                        | synapse_http_httppusher_badge_updates_processed                        |
+| synapse_http_httppusher_badge_updates_failed_total                           | synapse_http_httppusher_badge_updates_failed                           |
+
+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
+---------------------------------------------
+
+Synapse 0.27.0 begins the process of rationalising the duplicate
+`*:count` metrics reported for the resource tracking for code blocks and
+HTTP requests.
+
+At the same time, the corresponding `*:total` metrics are being renamed,
+as the `:total` suffix no longer makes sense in the absence of a
+corresponding `:count` metric.
+
+To enable a graceful migration path, this release just adds new names
+for the metrics being renamed. A future release will remove the old
+ones.
+
+The following table shows the new metrics, and the old metrics which
+they are replacing.
+
+| New name                                                      | Old name                                                   |
+| ------------------------------------------------------------- | ---------------------------------------------------------- |
+| synapse_util_metrics_block_count                              | synapse_util_metrics_block_timer:count                     |
+| synapse_util_metrics_block_count                              | synapse_util_metrics_block_ru_utime:count                  |
+| synapse_util_metrics_block_count                              | synapse_util_metrics_block_ru_stime:count                  |
+| synapse_util_metrics_block_count                              | synapse_util_metrics_block_db_txn_count:count              |
+| synapse_util_metrics_block_count                              | synapse_util_metrics_block_db_txn_duration:count           |
+| synapse_util_metrics_block_time_seconds                       | synapse_util_metrics_block_timer:total                     |
+| synapse_util_metrics_block_ru_utime_seconds                   | synapse_util_metrics_block_ru_utime:total                  |
+| synapse_util_metrics_block_ru_stime_seconds                   | synapse_util_metrics_block_ru_stime:total                  |
+| synapse_util_metrics_block_db_txn_count                       | synapse_util_metrics_block_db_txn_count:total              |
+| synapse_util_metrics_block_db_txn_duration_seconds            | synapse_util_metrics_block_db_txn_duration:total           |
+| synapse_http_server_response_count                            | synapse_http_server_requests                               |
+| synapse_http_server_response_count                            | synapse_http_server_response_time:count                    |
+| synapse_http_server_response_count                            | synapse_http_server_response_ru_utime:count                |
+| synapse_http_server_response_count                            | synapse_http_server_response_ru_stime:count                |
+| synapse_http_server_response_count                            | synapse_http_server_response_db_txn_count:count            |
+| synapse_http_server_response_count                            | synapse_http_server_response_db_txn_duration:count         |
+| synapse_http_server_response_time_seconds                     | synapse_http_server_response_time:total                    |
+| synapse_http_server_response_ru_utime_seconds                 | synapse_http_server_response_ru_utime:total                |
+| synapse_http_server_response_ru_stime_seconds                 | synapse_http_server_response_ru_stime:total                |
+| synapse_http_server_response_db_txn_count                     | synapse_http_server_response_db_txn_count:total            |
+| synapse_http_server_response_db_txn_duration_seconds          | synapse_http_server_response_db_txn_duration:total         |
+
+Standard Metric Names
+---------------------
+
+As of synapse version 0.18.2, the format of the process-wide metrics has
+been changed to fit prometheus standard naming conventions. Additionally
+the units have been changed to seconds, from miliseconds.
+
+| New name                                 | Old name                          |
+| ---------------------------------------- | --------------------------------- |
+| process_cpu_user_seconds_total           | process_resource_utime / 1000     |
+| process_cpu_system_seconds_total         | process_resource_stime / 1000     |
+| process_open_fds (no \'type\' label)     | process_fds                       |
+
+The python-specific counts of garbage collector performance have been
+renamed.
+
+| New name                         | Old name                   |
+| -------------------------------- | -------------------------- |
+| python_gc_time                   | reactor_gc_time            |
+| python_gc_unreachable_total      | reactor_gc_unreachable     |
+| python_gc_counts                 | reactor_gc_counts          |
+
+The twisted-specific reactor metrics have been renamed.
+
+| New name                               | Old name                |
+| -------------------------------------- | ----------------------- |
+| python_twisted_reactor_pending_calls   | reactor_pending_calls   |
+| python_twisted_reactor_tick_time       | reactor_tick_time       |
diff --git a/docs/metrics-howto.rst b/docs/metrics-howto.rst
deleted file mode 100644
index 973641f3dc..0000000000
--- a/docs/metrics-howto.rst
+++ /dev/null
@@ -1,285 +0,0 @@
-How to monitor Synapse metrics using Prometheus
-===============================================
-
-1. Install Prometheus:
-
-   Follow instructions at http://prometheus.io/docs/introduction/install/
-
-2. Enable Synapse metrics:
-
-   There are two methods of enabling metrics in Synapse.
-
-   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::
-
-     resources:
-       - names:
-         - client
-         - metrics
-
-   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.
-
-   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 ``/``.
-
-   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``)::
-
-    - job_name: "synapse"
-      metrics_path: "/_synapse/metrics"
-      static_configs:
-        - targets: ["my.server.here:port"]
-
-   where ``my.server.here`` is the IP address of Synapse, and ``port`` is the listener port
-   configured with the ``metrics`` resource.
-
-   If your prometheus is older than 1.5.2, you will need to replace
-   ``static_configs`` in the above with ``target_groups``.
-
-   Restart Prometheus.
-
-
-Renaming of metrics & deprecation of old names in 1.2
------------------------------------------------------
-
-Synapse 1.2 updates the Prometheus metrics to match the naming convention of the
-upstream ``prometheus_client``. The old names are considered deprecated and will
-be removed in a future version of Synapse.
-
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-|                                  New Name                                   |                               Old Name                                |
-+=============================================================================+=======================================================================+
-| python_gc_objects_collected_total                                           | python_gc_objects_collected                                           |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| python_gc_objects_uncollectable_total                                       | python_gc_objects_uncollectable                                       |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| python_gc_collections_total                                                 | python_gc_collections                                                 |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| process_cpu_seconds_total                                                   | process_cpu_seconds                                                   |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_federation_client_sent_transactions_total                           | synapse_federation_client_sent_transactions                           |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_federation_client_events_processed_total                            | synapse_federation_client_events_processed                            |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_event_processing_loop_count_total                                   | synapse_event_processing_loop_count                                   |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_event_processing_loop_room_count_total                              | synapse_event_processing_loop_room_count                              |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_util_metrics_block_count_total                                      | synapse_util_metrics_block_count                                      |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_util_metrics_block_time_seconds_total                               | synapse_util_metrics_block_time_seconds                               |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_util_metrics_block_ru_utime_seconds_total                           | synapse_util_metrics_block_ru_utime_seconds                           |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_util_metrics_block_ru_stime_seconds_total                           | synapse_util_metrics_block_ru_stime_seconds                           |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_util_metrics_block_db_txn_count_total                               | synapse_util_metrics_block_db_txn_count                               |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_util_metrics_block_db_txn_duration_seconds_total                    | synapse_util_metrics_block_db_txn_duration_seconds                    |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_util_metrics_block_db_sched_duration_seconds_total                  | synapse_util_metrics_block_db_sched_duration_seconds                  |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_background_process_start_count_total                                | synapse_background_process_start_count                                |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_background_process_ru_utime_seconds_total                           | synapse_background_process_ru_utime_seconds                           |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_background_process_ru_stime_seconds_total                           | synapse_background_process_ru_stime_seconds                           |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_background_process_db_txn_count_total                               | synapse_background_process_db_txn_count                               |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_background_process_db_txn_duration_seconds_total                    | synapse_background_process_db_txn_duration_seconds                    |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_background_process_db_sched_duration_seconds_total                  | synapse_background_process_db_sched_duration_seconds                  |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_storage_events_persisted_events_total                               | synapse_storage_events_persisted_events                               |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_storage_events_persisted_events_sep_total                           | synapse_storage_events_persisted_events_sep                           |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_storage_events_state_delta_total                                    | synapse_storage_events_state_delta                                    |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_storage_events_state_delta_single_event_total                       | synapse_storage_events_state_delta_single_event                       |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_storage_events_state_delta_reuse_delta_total                        | synapse_storage_events_state_delta_reuse_delta                        |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_federation_server_received_pdus_total                               | synapse_federation_server_received_pdus                               |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_federation_server_received_edus_total                               | synapse_federation_server_received_edus                               |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_handler_presence_notified_presence_total                            | synapse_handler_presence_notified_presence                            |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_handler_presence_federation_presence_out_total                      | synapse_handler_presence_federation_presence_out                      |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_handler_presence_presence_updates_total                             | synapse_handler_presence_presence_updates                             |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_handler_presence_timers_fired_total                                 | synapse_handler_presence_timers_fired                                 |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_handler_presence_federation_presence_total                          | synapse_handler_presence_federation_presence                          |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_handler_presence_bump_active_time_total                             | synapse_handler_presence_bump_active_time                             |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_federation_client_sent_edus_total                                   | synapse_federation_client_sent_edus                                   |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_federation_client_sent_pdu_destinations_count_total                 | synapse_federation_client_sent_pdu_destinations:count                 |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_federation_client_sent_pdu_destinations_total                       | synapse_federation_client_sent_pdu_destinations:total                 |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_handlers_appservice_events_processed_total                          | synapse_handlers_appservice_events_processed                          |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_notifier_notified_events_total                                      | synapse_notifier_notified_events                                      |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_push_bulk_push_rule_evaluator_push_rules_invalidation_counter_total | synapse_push_bulk_push_rule_evaluator_push_rules_invalidation_counter |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_push_bulk_push_rule_evaluator_push_rules_state_size_counter_total   | synapse_push_bulk_push_rule_evaluator_push_rules_state_size_counter   |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_http_httppusher_http_pushes_processed_total                         | synapse_http_httppusher_http_pushes_processed                         |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_http_httppusher_http_pushes_failed_total                            | synapse_http_httppusher_http_pushes_failed                            |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_http_httppusher_badge_updates_processed_total                       | synapse_http_httppusher_badge_updates_processed                       |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-| synapse_http_httppusher_badge_updates_failed_total                          | synapse_http_httppusher_badge_updates_failed                          |
-+-----------------------------------------------------------------------------+-----------------------------------------------------------------------+
-
-
-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
----------------------------------------------
-
-Synapse 0.27.0 begins the process of rationalising the duplicate ``*:count``
-metrics reported for the resource tracking for code blocks and HTTP requests.
-
-At the same time, the corresponding ``*:total`` metrics are being renamed, as
-the ``:total`` suffix no longer makes sense in the absence of a corresponding
-``:count`` metric.
-
-To enable a graceful migration path, this release just adds new names for the
-metrics being renamed. A future release will remove the old ones.
-
-The following table shows the new metrics, and the old metrics which they are
-replacing.
-
-==================================================== ===================================================
-New name                                             Old name
-==================================================== ===================================================
-synapse_util_metrics_block_count                     synapse_util_metrics_block_timer:count
-synapse_util_metrics_block_count                     synapse_util_metrics_block_ru_utime:count
-synapse_util_metrics_block_count                     synapse_util_metrics_block_ru_stime:count
-synapse_util_metrics_block_count                     synapse_util_metrics_block_db_txn_count:count
-synapse_util_metrics_block_count                     synapse_util_metrics_block_db_txn_duration:count
-
-synapse_util_metrics_block_time_seconds              synapse_util_metrics_block_timer:total
-synapse_util_metrics_block_ru_utime_seconds          synapse_util_metrics_block_ru_utime:total
-synapse_util_metrics_block_ru_stime_seconds          synapse_util_metrics_block_ru_stime:total
-synapse_util_metrics_block_db_txn_count              synapse_util_metrics_block_db_txn_count:total
-synapse_util_metrics_block_db_txn_duration_seconds   synapse_util_metrics_block_db_txn_duration:total
-
-synapse_http_server_response_count                   synapse_http_server_requests
-synapse_http_server_response_count                   synapse_http_server_response_time:count
-synapse_http_server_response_count                   synapse_http_server_response_ru_utime:count
-synapse_http_server_response_count                   synapse_http_server_response_ru_stime:count
-synapse_http_server_response_count                   synapse_http_server_response_db_txn_count:count
-synapse_http_server_response_count                   synapse_http_server_response_db_txn_duration:count
-
-synapse_http_server_response_time_seconds            synapse_http_server_response_time:total
-synapse_http_server_response_ru_utime_seconds        synapse_http_server_response_ru_utime:total
-synapse_http_server_response_ru_stime_seconds        synapse_http_server_response_ru_stime:total
-synapse_http_server_response_db_txn_count            synapse_http_server_response_db_txn_count:total
-synapse_http_server_response_db_txn_duration_seconds synapse_http_server_response_db_txn_duration:total
-==================================================== ===================================================
-
-
-Standard Metric Names
----------------------
-
-As of synapse version 0.18.2, the format of the process-wide metrics has been
-changed to fit prometheus standard naming conventions. Additionally the units
-have been changed to seconds, from miliseconds.
-
-================================== =============================
-New name                           Old name
-================================== =============================
-process_cpu_user_seconds_total     process_resource_utime / 1000
-process_cpu_system_seconds_total   process_resource_stime / 1000
-process_open_fds (no 'type' label) process_fds
-================================== =============================
-
-The python-specific counts of garbage collector performance have been renamed.
-
-=========================== ======================
-New name                    Old name
-=========================== ======================
-python_gc_time              reactor_gc_time
-python_gc_unreachable_total reactor_gc_unreachable
-python_gc_counts            reactor_gc_counts
-=========================== ======================
-
-The twisted-specific reactor metrics have been renamed.
-
-==================================== =====================
-New name                             Old name
-==================================== =====================
-python_twisted_reactor_pending_calls reactor_pending_calls
-python_twisted_reactor_tick_time     reactor_tick_time
-==================================== =====================
diff --git a/docs/opentracing.md b/docs/opentracing.md
new file mode 100644
index 0000000000..4c7a56a5d7
--- /dev/null
+++ b/docs/opentracing.md
@@ -0,0 +1,93 @@
+# OpenTracing
+
+## Background
+
+OpenTracing is a semi-standard being adopted by a number of distributed
+tracing platforms. It is a common api for facilitating vendor-agnostic
+tracing instrumentation. That is, we can use the OpenTracing api and
+select one of a number of tracer implementations to do the heavy lifting
+in the background. Our current selected implementation is Jaeger.
+
+OpenTracing is a tool which gives an insight into the causal
+relationship of work done in and between servers. The servers each track
+events and report them to a centralised server - in Synapse's case:
+Jaeger. The basic unit used to represent events is the span. The span
+roughly represents a single piece of work that was done and the time at
+which it occurred. A span can have child spans, meaning that the work of
+the child had to be completed for the parent span to complete, or it can
+have follow-on spans which represent work that is undertaken as a result
+of the parent but is not depended on by the parent to in order to
+finish.
+
+Since this is undertaken in a distributed environment a request to
+another server, such as an RPC or a simple GET, can be considered a span
+(a unit or work) for the local server. This causal link is what
+OpenTracing aims to capture and visualise. In order to do this metadata
+about the local server's span, i.e the 'span context', needs to be
+included with the request to the remote.
+
+It is up to the remote server to decide what it does with the spans it
+creates. This is called the sampling policy and it can be configured
+through Jaeger's settings.
+
+For OpenTracing concepts see
+<https://opentracing.io/docs/overview/what-is-tracing/>.
+
+For more information about Jaeger's implementation see
+<https://www.jaegertracing.io/docs/>
+
+## Setting up OpenTracing
+
+To receive OpenTracing spans, start up a Jaeger server. This can be done
+using docker like so:
+
+```sh
+docker run -d --name jaeger
+  -p 6831:6831/udp \
+  -p 6832:6832/udp \
+  -p 5778:5778 \
+  -p 16686:16686 \
+  -p 14268:14268 \
+  jaegertracing/all-in-one:1.13
+```
+
+Latest documentation is probably at
+<https://www.jaegertracing.io/docs/1.13/getting-started/>
+
+## Enable OpenTracing in Synapse
+
+OpenTracing is not enabled by default. It must be enabled in the
+homeserver config by uncommenting the config options under `opentracing`
+as shown in the [sample config](./sample_config.yaml). For example:
+
+```yaml
+opentracing:
+  tracer_enabled: true
+  homeserver_whitelist:
+    - "mytrustedhomeserver.org"
+    - "*.myotherhomeservers.com"
+```
+
+## Homeserver whitelisting
+
+The homeserver whitelist is configured using regular expressions. A list
+of regular expressions can be given and their union will be compared
+when propagating any spans contexts to another homeserver.
+
+Though it's mostly safe to send and receive span contexts to and from
+untrusted users since span contexts are usually opaque ids it can lead
+to two problems, namely:
+
+-   If the span context is marked as sampled by the sending homeserver
+    the receiver will sample it. Therefore two homeservers with wildly
+    different sampling policies could incur higher sampling counts than
+    intended.
+-   Sending servers can attach arbitrary data to spans, known as
+    'baggage'. For safety this has been disabled in Synapse but that
+    doesn't prevent another server sending you baggage which will be
+    logged to OpenTracing's logs.
+
+## Configuring Jaeger
+
+Sampling strategies can be set as in this document:
+<https://www.jaegertracing.io/docs/1.13/sampling/>
diff --git a/docs/opentracing.rst b/docs/opentracing.rst
deleted file mode 100644
index 6e98ab56ba..0000000000
--- a/docs/opentracing.rst
+++ /dev/null
@@ -1,123 +0,0 @@
-===========
-OpenTracing
-===========
-
-Background
-----------
-
-OpenTracing is a semi-standard being adopted by a number of distributed tracing
-platforms. It is a common api for facilitating vendor-agnostic tracing
-instrumentation. That is, we can use the OpenTracing api and select one of a
-number of tracer implementations to do the heavy lifting in the background.
-Our current selected implementation is Jaeger.
-
-OpenTracing is a tool which gives an insight into the causal relationship of
-work done in and between servers. The servers each track events and report them
-to a centralised server - in Synapse's case: Jaeger. The basic unit used to
-represent events is the span. The span roughly represents a single piece of work
-that was done and the time at which it occurred. A span can have child spans,
-meaning that the work of the child had to be completed for the parent span to
-complete, or it can have follow-on spans which represent work that is undertaken
-as a result of the parent but is not depended on by the parent to in order to
-finish.
-
-Since this is undertaken in a distributed environment a request to another
-server, such as an RPC or a simple GET, can be considered a span (a unit or
-work) for the local server. This causal link is what OpenTracing aims to
-capture and visualise. In order to do this metadata about the local server's
-span, i.e the 'span context', needs to be included with the request to the
-remote.
-
-It is up to the remote server to decide what it does with the spans
-it creates. This is called the sampling policy and it can be configured
-through Jaeger's settings.
-
-For OpenTracing concepts see
-https://opentracing.io/docs/overview/what-is-tracing/.
-
-For more information about Jaeger's implementation see
-https://www.jaegertracing.io/docs/
-
-=====================
-Seting up OpenTracing
-=====================
-
-To receive OpenTracing spans, start up a Jaeger server. This can be done
-using docker like so:
-
-.. code-block:: bash
-
-   docker run -d --name jaeger
-     -p 6831:6831/udp \
-     -p 6832:6832/udp \
-     -p 5778:5778 \
-     -p 16686:16686 \
-     -p 14268:14268 \
-     jaegertracing/all-in-one:1.13
-
-Latest documentation is probably at
-https://www.jaegertracing.io/docs/1.13/getting-started/
-
-
-Enable OpenTracing in Synapse
------------------------------
-
-OpenTracing is not enabled by default. It must be enabled in the homeserver
-config by uncommenting the config options under ``opentracing`` as shown in
-the `sample config <./sample_config.yaml>`_. For example:
-
-.. code-block:: yaml
-
-  opentracing:
-    tracer_enabled: true
-    homeserver_whitelist:
-      - "mytrustedhomeserver.org"
-      - "*.myotherhomeservers.com"
-
-Homeserver whitelisting
------------------------
-
-The homeserver whitelist is configured using regular expressions. A list of regular
-expressions can be given and their union will be compared when propagating any
-spans contexts to another homeserver.
-
-Though it's mostly safe to send and receive span contexts to and from
-untrusted users since span contexts are usually opaque ids it can lead to
-two problems, namely:
-
-- If the span context is marked as sampled by the sending homeserver the receiver will
-  sample it. Therefore two homeservers with wildly different sampling policies
-  could incur higher sampling counts than intended.
-- Sending servers can attach arbitrary data to spans, known as 'baggage'. For safety this has been disabled in Synapse
-  but that doesn't prevent another server sending you baggage which will be logged
-  to OpenTracing's logs.
-
-==========
-EDU FORMAT
-==========
-
-EDUs can contain tracing data in their content. This is not specced but
-it could be of interest for other homeservers.
-
-EDU format (if you're using jaeger):
-
-.. code-block:: json
-
-   {
-     "edu_type": "type",
-     "content": {
-       "org.matrix.opentracing_context": {
-         "uber-trace-id": "fe57cf3e65083289"
-       }
-     }
-   }
-
-Though you don't have to use jaeger you must inject the span context into
-`org.matrix.opentracing_context` using the opentracing `Format.TEXT_MAP` inject method.
-
-==================
-Configuring Jaeger
-==================
-
-Sampling strategies can be set as in this document:
-https://www.jaegertracing.io/docs/1.13/sampling/
diff --git a/docs/password_auth_providers.md b/docs/password_auth_providers.md
new file mode 100644
index 0000000000..0db1a3804a
--- /dev/null
+++ b/docs/password_auth_providers.md
@@ -0,0 +1,116 @@
+# Password auth provider modules
+
+Password auth providers offer a way for server administrators to
+integrate their Synapse installation with an existing authentication
+system.
+
+A password auth provider is a Python class which is dynamically loaded
+into Synapse, and provides a number of methods by which it can integrate
+with the authentication system.
+
+This document serves as a reference for those looking to implement their
+own password auth providers.
+
+## Required methods
+
+Password auth provider classes must provide the following methods:
+
+*class* `SomeProvider.parse_config`(*config*)
+
+> This method is passed the `config` object for this module from the
+> homeserver configuration file.
+>
+> It should perform any appropriate sanity checks on the provided
+> configuration, and return an object which is then passed into
+> `__init__`.
+
+*class* `SomeProvider`(*config*, *account_handler*)
+
+> The constructor is passed the config object returned by
+> `parse_config`, and a `synapse.module_api.ModuleApi` object which
+> allows the password provider to check if accounts exist and/or create
+> new ones.
+
+## Optional methods
+
+Password auth provider classes may optionally provide the following
+methods.
+
+*class* `SomeProvider.get_db_schema_files`()
+
+> This method, if implemented, should return an Iterable of
+> `(name, stream)` pairs of database schema files. Each file is applied
+> in turn at initialisation, and a record is then made in the database
+> so that it is not re-applied on the next start.
+
+`someprovider.get_supported_login_types`()
+
+> This method, if implemented, should return a `dict` mapping from a
+> login type identifier (such as `m.login.password`) to an iterable
+> giving the fields which must be provided by the user in the submission
+> to the `/login` api. These fields are passed in the `login_dict`
+> dictionary to `check_auth`.
+>
+> For example, if a password auth provider wants to implement a custom
+> login type of `com.example.custom_login`, where the client is expected
+> to pass the fields `secret1` and `secret2`, the provider should
+> implement this method and return the following dict:
+>
+>     {"com.example.custom_login": ("secret1", "secret2")}
+
+`someprovider.check_auth`(*username*, *login_type*, *login_dict*)
+
+> This method is the one that does the real work. If implemented, it
+> will be called for each login attempt where the login type matches one
+> of the keys returned by `get_supported_login_types`.
+>
+> It is passed the (possibly UNqualified) `user` provided by the client,
+> the login type, and a dictionary of login secrets passed by the
+> client.
+>
+> The method should return a Twisted `Deferred` object, which resolves
+> to the canonical `@localpart:domain` user id if authentication is
+> successful, and `None` if not.
+>
+> Alternatively, the `Deferred` can resolve to a `(str, func)` tuple, in
+> which case the second field is a callback which will be called with
+> the result from the `/login` call (including `access_token`,
+> `device_id`, etc.)
+
+`someprovider.check_3pid_auth`(*medium*, *address*, *password*)
+
+> This method, if implemented, is called when a user attempts to
+> register or log in with a third party identifier, such as email. It is
+> passed the medium (ex. "email"), an address (ex.
+> "<jdoe@example.com>") and the user's password.
+>
+> The method should return a Twisted `Deferred` object, which resolves
+> to a `str` containing the user's (canonical) User ID if
+> authentication was successful, and `None` if not.
+>
+> As with `check_auth`, the `Deferred` may alternatively resolve to a
+> `(user_id, callback)` tuple.
+
+`someprovider.check_password`(*user_id*, *password*)
+
+> This method provides a simpler interface than
+> `get_supported_login_types` and `check_auth` for password auth
+> providers that just want to provide a mechanism for validating
+> `m.login.password` logins.
+>
+> Iif implemented, it will be called to check logins with an
+> `m.login.password` login type. It is passed a qualified
+> `@localpart:domain` user id, and the password provided by the user.
+>
+> The method should return a Twisted `Deferred` object, which resolves
+> to `True` if authentication is successful, and `False` if not.
+
+`someprovider.on_logged_out`(*user_id*, *device_id*, *access_token*)
+
+> This method, if implemented, is called when a user logs out. It is
+> passed the qualified user ID, the ID of the deactivated device (if
+> any: access tokens are occasionally created without an associated
+> device ID), and the (now deactivated) access token.
+>
+> It may return a Twisted `Deferred` object; the logout request will
+> wait for the deferred to complete but the result is ignored.
diff --git a/docs/password_auth_providers.rst b/docs/password_auth_providers.rst
deleted file mode 100644
index 6149ba7458..0000000000
--- a/docs/password_auth_providers.rst
+++ /dev/null
@@ -1,113 +0,0 @@
-Password auth provider modules
-==============================
-
-Password auth providers offer a way for server administrators to integrate
-their Synapse installation with an existing authentication system.
-
-A password auth provider is a Python class which is dynamically loaded into
-Synapse, and provides a number of methods by which it can integrate with the
-authentication system.
-
-This document serves as a reference for those looking to implement their own
-password auth providers.
-
-Required methods
-----------------
-
-Password auth provider classes must provide the following methods:
-
-*class* ``SomeProvider.parse_config``\(*config*)
-
-    This method is passed the ``config`` object for this module from the
-    homeserver configuration file.
-
-    It should perform any appropriate sanity checks on the provided
-    configuration, and return an object which is then passed into ``__init__``.
-
-*class* ``SomeProvider``\(*config*, *account_handler*)
-
-    The constructor is passed the config object returned by ``parse_config``,
-    and a ``synapse.module_api.ModuleApi`` object which allows the
-    password provider to check if accounts exist and/or create new ones.
-
-Optional methods
-----------------
-
-Password auth provider classes may optionally provide the following methods.
-
-*class* ``SomeProvider.get_db_schema_files``\()
-
-    This method, if implemented, should return an Iterable of ``(name,
-    stream)`` pairs of database schema files. Each file is applied in turn at
-    initialisation, and a record is then made in the database so that it is
-    not re-applied on the next start.
-
-``someprovider.get_supported_login_types``\()
-
-    This method, if implemented, should return a ``dict`` mapping from a login
-    type identifier (such as ``m.login.password``) to an iterable giving the
-    fields which must be provided by the user in the submission to the
-    ``/login`` api. These fields are passed in the ``login_dict`` dictionary
-    to ``check_auth``.
-
-    For example, if a password auth provider wants to implement a custom login
-    type of ``com.example.custom_login``, where the client is expected to pass
-    the fields ``secret1`` and ``secret2``, the provider should implement this
-    method and return the following dict::
-
-      {"com.example.custom_login": ("secret1", "secret2")}
-
-``someprovider.check_auth``\(*username*, *login_type*, *login_dict*)
-
-    This method is the one that does the real work. If implemented, it will be
-    called for each login attempt where the login type matches one of the keys
-    returned by ``get_supported_login_types``.
-
-    It is passed the (possibly UNqualified) ``user`` provided by the client,
-    the login type, and a dictionary of login secrets passed by the client.
-
-    The method should return a Twisted ``Deferred`` object, which resolves to
-    the canonical ``@localpart:domain`` user id if authentication is successful,
-    and ``None`` if not.
-
-    Alternatively, the ``Deferred`` can resolve to a ``(str, func)`` tuple, in
-    which case the second field is a callback which will be called with the
-    result from the ``/login`` call (including ``access_token``, ``device_id``,
-    etc.)
-
-``someprovider.check_3pid_auth``\(*medium*, *address*, *password*)
-
-    This method, if implemented, is called when a user attempts to register or
-    log in with a third party identifier, such as email. It is passed the
-    medium (ex. "email"), an address (ex. "jdoe@example.com") and the user's
-    password.
-
-    The method should return a Twisted ``Deferred`` object, which resolves to
-    a ``str`` containing the user's (canonical) User ID if authentication was
-    successful, and ``None`` if not.
-
-    As with ``check_auth``, the ``Deferred`` may alternatively resolve to a
-    ``(user_id, callback)`` tuple.
-
-``someprovider.check_password``\(*user_id*, *password*)
-
-    This method provides a simpler interface than ``get_supported_login_types``
-    and ``check_auth`` for password auth providers that just want to provide a
-    mechanism for validating ``m.login.password`` logins.
-
-    Iif implemented, it will be called to check logins with an
-    ``m.login.password`` login type. It is passed a qualified
-    ``@localpart:domain`` user id, and the password provided by the user.
-
-    The method should return a Twisted ``Deferred`` object, which resolves to
-    ``True`` if authentication is successful, and ``False`` if not.
-
-``someprovider.on_logged_out``\(*user_id*, *device_id*, *access_token*)
-
-    This method, if implemented, is called when a user logs out. It is passed
-    the qualified user ID, the ID of the deactivated device (if any: access
-    tokens are occasionally created without an associated device ID), and the
-    (now deactivated) access token.
-
-    It may return a Twisted ``Deferred`` object; the logout request will wait
-    for the deferred to complete but the result is ignored.
diff --git a/docs/postgres.md b/docs/postgres.md
new file mode 100644
index 0000000000..29cf762858
--- /dev/null
+++ b/docs/postgres.md
@@ -0,0 +1,164 @@
+# Using Postgres
+
+Postgres version 9.5 or later is known to work.
+
+## Install postgres client libraries
+
+Synapse will require the python postgres client library in order to
+connect to a postgres database.
+
+-   If you are using the [matrix.org debian/ubuntu
+    packages](../INSTALL.md#matrixorg-packages), the necessary python
+    library will already be installed, but you will need to ensure the
+    low-level postgres library is installed, which you can do with
+    `apt install libpq5`.
+-   For other pre-built packages, please consult the documentation from
+    the relevant package.
+-   If you installed synapse [in a
+    virtualenv](../INSTALL.md#installing-from-source), you can install
+    the library with:
+
+        ~/synapse/env/bin/pip install matrix-synapse[postgres]
+
+    (substituting the path to your virtualenv for `~/synapse/env`, if
+    you used a different path). You will require the postgres
+    development files. These are in the `libpq-dev` package on
+    Debian-derived distributions.
+
+## Set up database
+
+Assuming your PostgreSQL database user is called `postgres`, create a
+user `synapse_user` with:
+
+    su - postgres
+    createuser --pwprompt synapse_user
+
+Before you can authenticate with the `synapse_user`, you must create a
+database that it can access. To create a database, first connect to the
+database with your database user:
+
+    su - postgres
+    psql
+
+and then run:
+
+    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 have been created as above).
+
+Note that the PostgreSQL database *must* have the correct encoding set
+(as shown above), otherwise it will not be able to store UTF8 strings.
+
+You may need to enable password authentication so `synapse_user` can
+connect to the database. See
+<https://www.postgresql.org/docs/11/auth-pg-hba-conf.html>.
+
+## Tuning Postgres
+
+The default settings should be fine for most deployments. For larger
+scale deployments tuning some of the settings is recommended, details of
+which can be found at
+<https://wiki.postgresql.org/wiki/Tuning_Your_PostgreSQL_Server>.
+
+In particular, we've found tuning the following values helpful for
+performance:
+
+-   `shared_buffers`
+-   `effective_cache_size`
+-   `work_mem`
+-   `maintenance_work_mem`
+-   `autovacuum_work_mem`
+
+Note that the appropriate values for those fields depend on the amount
+of free memory the database host has available.
+
+## Synapse config
+
+When you are ready to start using PostgreSQL, edit the `database`
+section in your config file to match the following lines:
+
+    database:
+        name: psycopg2
+        args:
+            user: <user>
+            password: <pass>
+            database: <db>
+            host: <host>
+            cp_min: 5
+            cp_max: 10
+
+All key, values in `args` are passed to the `psycopg2.connect(..)`
+function, except keys beginning with `cp_`, which are consumed by the
+twisted adbapi connection pool.
+
+## Porting from SQLite
+
+### Overview
+
+The script `synapse_port_db` allows porting an existing synapse server
+backed by SQLite to using PostgreSQL. This is done in as a two phase
+process:
+
+1.  Copy the existing SQLite database to a separate location (while the
+    server is down) and running the port script against that offline
+    database.
+2.  Shut down the server. Rerun the port script to port any data that
+    has come in since taking the first snapshot. Restart server against
+    the PostgreSQL database.
+
+The port script is designed to be run repeatedly against newer snapshots
+of the SQLite database file. This makes it safe to repeat step 1 if
+there was a delay between taking the previous snapshot and being ready
+to do step 2.
+
+It is safe to at any time kill the port script and restart it.
+
+### Using the port script
+
+Firstly, shut down the currently running synapse server and copy its
+database file (typically `homeserver.db`) to another location. Once the
+copy is complete, restart synapse. For instance:
+
+    ./synctl stop
+    cp homeserver.db homeserver.db.snapshot
+    ./synctl start
+
+Copy the old config file into a new config file:
+
+    cp homeserver.yaml homeserver-postgres.yaml
+
+Edit the database section as described in the section *Synapse config*
+above and with the SQLite snapshot located at `homeserver.db.snapshot`
+simply run:
+
+    synapse_port_db --sqlite-database homeserver.db.snapshot \
+        --postgres-config homeserver-postgres.yaml
+
+The flag `--curses` displays a coloured curses progress UI.
+
+If the script took a long time to complete, or time has otherwise passed
+since the original snapshot was taken, repeat the previous steps with a
+newer snapshot.
+
+To complete the conversion shut down the synapse server and run the port
+script one last time, e.g. if the SQLite database is at `homeserver.db`
+run:
+
+    synapse_port_db --sqlite-database homeserver.db \
+        --postgres-config homeserver-postgres.yaml
+
+Once that has completed, change the synapse config to point at the
+PostgreSQL database configuration file `homeserver-postgres.yaml`:
+
+    ./synctl stop
+    mv homeserver.yaml homeserver-old-sqlite.yaml
+    mv homeserver-postgres.yaml homeserver.yaml
+    ./synctl start
+
+Synapse should now be running against PostgreSQL.
diff --git a/docs/postgres.rst b/docs/postgres.rst
deleted file mode 100644
index e08a5116b9..0000000000
--- a/docs/postgres.rst
+++ /dev/null
@@ -1,166 +0,0 @@
-Using Postgres
---------------
-
-Postgres version 9.5 or later is known to work.
-
-Install postgres client libraries
-=================================
-
-Synapse will require the python postgres client library in order to connect to
-a postgres database.
-
-* If you are using the `matrix.org debian/ubuntu
-  packages <../INSTALL.md#matrixorg-packages>`_,
-  the necessary python library will already be installed, but you will need to
-  ensure the low-level postgres library is installed, which you can do with
-  ``apt install libpq5``.
-
-* For other pre-built packages, please consult the documentation from the
-  relevant package.
-
-* If you installed synapse `in a virtualenv
-  <../INSTALL.md#installing-from-source>`_, you can install the library with::
-
-      ~/synapse/env/bin/pip install matrix-synapse[postgres]
-
-  (substituting the path to your virtualenv for ``~/synapse/env``, if you used a
-  different path). You will require the postgres development files. These are in
-  the ``libpq-dev`` package on Debian-derived distributions.
-
-Set up database
-===============
-
-Assuming your PostgreSQL database user is called ``postgres``, create a user
-``synapse_user`` with::
-
-   su - postgres
-   createuser --pwprompt synapse_user
-
-Before you can authenticate with the ``synapse_user``, you must create a
-database that it can access. To create a database, first connect to the database
-with your database user::
-
-   su - postgres
-   psql
-
-and then run::
-
-   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 have been created as above).
-
-Note that the PostgreSQL database *must* have the correct encoding set (as
-shown above), otherwise it will not be able to store UTF8 strings.
-
-You may need to enable password authentication so ``synapse_user`` can connect
-to the database. See https://www.postgresql.org/docs/11/auth-pg-hba-conf.html.
-
-Tuning Postgres
-===============
-
-The default settings should be fine for most deployments. For larger scale
-deployments tuning some of the settings is recommended, details of which can be
-found at https://wiki.postgresql.org/wiki/Tuning_Your_PostgreSQL_Server.
-
-In particular, we've found tuning the following values helpful for performance:
-
-- ``shared_buffers``
-- ``effective_cache_size``
-- ``work_mem``
-- ``maintenance_work_mem``
-- ``autovacuum_work_mem``
-
-Note that the appropriate values for those fields depend on the amount of free
-memory the database host has available.
-
-Synapse config
-==============
-
-When you are ready to start using PostgreSQL, edit the ``database`` section in
-your config file to match the following lines::
-
-    database:
-        name: psycopg2
-        args:
-            user: <user>
-            password: <pass>
-            database: <db>
-            host: <host>
-            cp_min: 5
-            cp_max: 10
-
-All key, values in ``args`` are passed to the ``psycopg2.connect(..)``
-function, except keys beginning with ``cp_``, which are consumed by the twisted
-adbapi connection pool.
-
-
-Porting from SQLite
-===================
-
-Overview
-~~~~~~~~
-
-The script ``synapse_port_db`` allows porting an existing synapse server
-backed by SQLite to using PostgreSQL. This is done in as a two phase process:
-
-1. Copy the existing SQLite database to a separate location (while the server
-   is down) and running the port script against that offline database.
-2. Shut down the server. Rerun the port script to port any data that has come
-   in since taking the first snapshot. Restart server against the PostgreSQL
-   database.
-
-The port script is designed to be run repeatedly against newer snapshots of the
-SQLite database file. This makes it safe to repeat step 1 if there was a delay
-between taking the previous snapshot and being ready to do step 2.
-
-It is safe to at any time kill the port script and restart it.
-
-Using the port script
-~~~~~~~~~~~~~~~~~~~~~
-
-Firstly, shut down the currently running synapse server and copy its database
-file (typically ``homeserver.db``) to another location. Once the copy is
-complete, restart synapse.  For instance::
-
-    ./synctl stop
-    cp homeserver.db homeserver.db.snapshot
-    ./synctl start
-
-Copy the old config file into a new config file::
-
-    cp homeserver.yaml homeserver-postgres.yaml
-
-Edit the database section as described in the section *Synapse config* above
-and with the SQLite snapshot located at ``homeserver.db.snapshot`` simply run::
-
-    synapse_port_db --sqlite-database homeserver.db.snapshot \
-        --postgres-config homeserver-postgres.yaml
-
-The flag ``--curses`` displays a coloured curses progress UI.
-
-If the script took a long time to complete, or time has otherwise passed since
-the original snapshot was taken, repeat the previous steps with a newer
-snapshot.
-
-To complete the conversion shut down the synapse server and run the port
-script one last time, e.g. if the SQLite database is at  ``homeserver.db``
-run::
-
-    synapse_port_db --sqlite-database homeserver.db \
-        --postgres-config homeserver-postgres.yaml
-
-Once that has completed, change the synapse config to point at the PostgreSQL
-database configuration file ``homeserver-postgres.yaml``::
-
-    ./synctl stop
-    mv homeserver.yaml homeserver-old-sqlite.yaml
-    mv homeserver-postgres.yaml homeserver.yaml
-    ./synctl start
-
-Synapse should now be running against PostgreSQL.
diff --git a/docs/replication.md b/docs/replication.md
new file mode 100644
index 0000000000..ed88233157
--- /dev/null
+++ b/docs/replication.md
@@ -0,0 +1,37 @@
+# Replication Architecture
+
+## Motivation
+
+We'd like to be able to split some of the work that synapse does into
+multiple python processes. In theory multiple synapse processes could
+share a single postgresql database and we\'d scale up by running more
+synapse processes. However much of synapse assumes that only one process
+is interacting with the database, both for assigning unique identifiers
+when inserting into tables, notifying components about new updates, and
+for invalidating its caches.
+
+So running multiple copies of the current code isn't an option. One way
+to run multiple processes would be to have a single writer process and
+multiple reader processes connected to the same database. In order to do
+this we'd need a way for the reader process to invalidate its in-memory
+caches when an update happens on the writer. One way to do this is for
+the writer to present an append-only log of updates which the readers
+can consume to invalidate their caches and to push updates to listening
+clients or pushers.
+
+Synapse already stores much of its data as an append-only log so that it
+can correctly respond to `/sync` requests so the amount of code changes
+needed to expose the append-only log to the readers should be fairly
+minimal.
+
+## Architecture
+
+### The Replication Protocol
+
+See [tcp_replication.md](tcp_replication.md)
+
+### The Slaved DataStore
+
+There are read-only version of the synapse storage layer in
+`synapse/replication/slave/storage` that use the response of the
+replication API to invalidate their caches.
diff --git a/docs/replication.rst b/docs/replication.rst
deleted file mode 100644
index 310abb3488..0000000000
--- a/docs/replication.rst
+++ /dev/null
@@ -1,40 +0,0 @@
-Replication Architecture
-========================
-
-Motivation
-----------
-
-We'd like to be able to split some of the work that synapse does into multiple
-python processes. In theory multiple synapse processes could share a single
-postgresql database and we'd scale up by running more synapse processes.
-However much of synapse assumes that only one process is interacting with the
-database, both for assigning unique identifiers when inserting into tables,
-notifying components about new updates, and for invalidating its caches.
-
-So running multiple copies of the current code isn't an option. One way to
-run multiple processes would be to have a single writer process and multiple
-reader processes connected to the same database. In order to do this we'd need
-a way for the reader process to invalidate its in-memory caches when an update
-happens on the writer. One way to do this is for the writer to present an
-append-only log of updates which the readers can consume to invalidate their
-caches and to push updates to listening clients or pushers.
-
-Synapse already stores much of its data as an append-only log so that it can
-correctly respond to /sync requests so the amount of code changes needed to
-expose the append-only log to the readers should be fairly minimal.
-
-Architecture
-------------
-
-The Replication Protocol
-~~~~~~~~~~~~~~~~~~~~~~~~
-
-See ``tcp_replication.rst``
-
-
-The Slaved DataStore
-~~~~~~~~~~~~~~~~~~~~
-
-There are read-only version of the synapse storage layer in
-``synapse/replication/slave/storage`` that use the response of the replication
-API to invalidate their caches.
diff --git a/docs/reverse_proxy.md b/docs/reverse_proxy.md
new file mode 100644
index 0000000000..dcfc5c64aa
--- /dev/null
+++ b/docs/reverse_proxy.md
@@ -0,0 +1,123 @@
+# 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),
+[Caddy](https://caddyserver.com/docs/proxy) or
+[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.
+
+> **NOTE**: Your reverse proxy must not `canonicalise` or `normalise`
+the requested URI in any way (for example, by decoding `%xx` escapes).
+Beware that Apache *will* canonicalise URIs unless you specifify
+`nocanon`.
+
+When setting up a reverse proxy, remember that Matrix clients and other
+Matrix servers do not necessarily need to connect to your server via the
+same server name or port. Indeed, clients will use port 443 by default,
+whereas servers default to port 8448. Where these are different, we
+refer to the 'client port' and the \'federation port\'. See [Setting
+up federation](federate.md) for more details of the algorithm used for
+federation connections.
+
+Let's assume that we expect clients to connect to our server at
+`https://matrix.example.com`, and other servers to connect at
+`https://example.com:8448`.  The following sections detail the configuration of
+the reverse proxy and the homeserver.
+
+## Webserver configuration examples
+
+> **NOTE**: You only need one of these.
+
+### nginx
+
+        server {
+            listen 443 ssl;
+            listen [::]:443 ssl;
+            server_name matrix.example.com;
+
+            location /_matrix {
+                proxy_pass http://localhost:8008;
+                proxy_set_header X-Forwarded-For $remote_addr;
+            }
+        }
+
+        server {
+            listen 8448 ssl default_server;
+            listen [::]:8448 ssl default_server;
+            server_name example.com;
+
+            location / {
+                proxy_pass http://localhost:8008;
+                proxy_set_header X-Forwarded-For $remote_addr;
+            }
+        }
+
+> **NOTE**: Do not add a `/` after the port in `proxy_pass`, otherwise nginx will
+canonicalise/normalise the URI.
+
+### Caddy
+
+        matrix.example.com {
+          proxy /_matrix http://localhost:8008 {
+            transparent
+          }
+        }
+
+        example.com:8448 {
+          proxy / http://localhost:8008 {
+            transparent
+          }
+        }
+
+### Apache
+
+        <VirtualHost *:443>
+            SSLEngine on
+            ServerName matrix.example.com;
+
+            AllowEncodedSlashes NoDecode
+            ProxyPass /_matrix http://127.0.0.1:8008/_matrix nocanon
+            ProxyPassReverse /_matrix http://127.0.0.1:8008/_matrix
+        </VirtualHost>
+
+        <VirtualHost *:8448>
+            SSLEngine on
+            ServerName example.com;
+
+            AllowEncodedSlashes NoDecode
+            ProxyPass /_matrix http://127.0.0.1:8008/_matrix nocanon
+            ProxyPassReverse /_matrix http://127.0.0.1:8008/_matrix
+        </VirtualHost>
+
+> **NOTE**: ensure the  `nocanon` options are included.
+
+### HAProxy
+
+        frontend https
+          bind :::443 v4v6 ssl crt /etc/ssl/haproxy/ strict-sni alpn h2,http/1.1
+
+          # Matrix client traffic
+          acl matrix-host hdr(host) -i matrix.example.com
+          acl matrix-path path_beg /_matrix
+
+          use_backend matrix if matrix-host matrix-path
+
+        frontend matrix-federation
+          bind :::8448 v4v6 ssl crt /etc/ssl/haproxy/synapse.pem alpn h2,http/1.1
+          default_backend matrix
+
+        backend matrix
+          server matrix 127.0.0.1:8008
+
+## Homeserver Configuration
+
+You will also want to set `bind_addresses: ['127.0.0.1']` and
+`x_forwarded: true` for port 8008 in `homeserver.yaml` to ensure that
+client IP addresses are recorded correctly.
+
+Having done so, you can then use `https://matrix.example.com` (instead
+of `https://matrix.example.com:8448`) as the "Custom server" when
+connecting to Synapse from a client.
diff --git a/docs/reverse_proxy.rst b/docs/reverse_proxy.rst
deleted file mode 100644
index 4b640ffc4f..0000000000
--- a/docs/reverse_proxy.rst
+++ /dev/null
@@ -1,112 +0,0 @@
-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>`_,
-`Caddy <https://caddyserver.com/docs/proxy>`_ or
-`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.
-
-**NOTE**: Your reverse proxy must not 'canonicalise' or 'normalise' the
-requested URI in any way (for example, by decoding ``%xx`` escapes). Beware
-that Apache *will* canonicalise URIs unless you specifify ``nocanon``.
-
-When setting up a reverse proxy, remember that Matrix clients and other Matrix
-servers do not necessarily need to connect to your server via the same server
-name or port. Indeed, clients will use port 443 by default, whereas servers
-default to port 8448. Where these are different, we refer to the 'client port'
-and the 'federation port'. See `Setting up federation
-<federate.md>`_ for more details of the algorithm used for
-federation connections.
-
-Let's assume that we expect clients to connect to our server at
-``https://matrix.example.com``, and other servers to connect at
-``https://example.com:8448``. Here are some example configurations:
-
-* nginx::
-
-      server {
-          listen 443 ssl;
-          listen [::]:443 ssl;
-          server_name matrix.example.com;
-
-          location /_matrix {
-              proxy_pass http://localhost:8008;
-              proxy_set_header X-Forwarded-For $remote_addr;
-          }
-      }
-
-      server {
-          listen 8448 ssl default_server;
-          listen [::]:8448 ssl default_server;
-          server_name example.com;
-
-          location / {
-              proxy_pass http://localhost:8008;
-              proxy_set_header X-Forwarded-For $remote_addr;
-          }
-      }
-      
-  Do not add a `/` after the port in `proxy_pass`, otherwise nginx will canonicalise/normalise the URI.
-
-* Caddy::
-
-      matrix.example.com {
-        proxy /_matrix http://localhost:8008 {
-          transparent
-        }
-      }
-
-      example.com:8448 {
-        proxy / http://localhost:8008 {
-          transparent
-        }
-      }
-
-* Apache (note the ``nocanon`` options here!)::
-
-      <VirtualHost *:443>
-          SSLEngine on
-          ServerName matrix.example.com;
-
-          AllowEncodedSlashes NoDecode
-          ProxyPass /_matrix http://127.0.0.1:8008/_matrix nocanon
-          ProxyPassReverse /_matrix http://127.0.0.1:8008/_matrix
-      </VirtualHost>
-
-      <VirtualHost *:8448>
-          SSLEngine on
-          ServerName example.com;
-          
-          AllowEncodedSlashes NoDecode
-          ProxyPass /_matrix http://127.0.0.1:8008/_matrix nocanon
-          ProxyPassReverse /_matrix http://127.0.0.1:8008/_matrix
-      </VirtualHost>
-
-* HAProxy::
-
-      frontend https
-        bind :::443 v4v6 ssl crt /etc/ssl/haproxy/ strict-sni alpn h2,http/1.1
-
-        # Matrix client traffic
-        acl matrix-host hdr(host) -i matrix.example.com
-        acl matrix-path path_beg /_matrix
-
-        use_backend matrix if matrix-host matrix-path
-
-      frontend matrix-federation
-        bind :::8448 v4v6 ssl crt /etc/ssl/haproxy/synapse.pem alpn h2,http/1.1
-        default_backend matrix
-
-      backend matrix
-        server matrix 127.0.0.1:8008
-
-You will also want to set ``bind_addresses: ['127.0.0.1']`` and ``x_forwarded: true``
-for port 8008 in ``homeserver.yaml`` to ensure that client IP addresses are
-recorded correctly.
-
-Having done so, you can then use ``https://matrix.example.com`` (instead of
-``https://matrix.example.com:8448``) as the "Custom server" when connecting to
-Synapse from a client.
diff --git a/docs/sample_config.yaml b/docs/sample_config.yaml
index 93c0edd8ce..8226978ba6 100644
--- a/docs/sample_config.yaml
+++ b/docs/sample_config.yaml
@@ -110,6 +110,9 @@ pid_file: DATADIR/homeserver.pid
 # blacklist IP address CIDR ranges. If this option is not specified, or
 # specified with an empty list, no ip range blacklist will be enforced.
 #
+# As of Synapse v1.4.0 this option also affects any outbound requests to identity
+# servers provided by user input.
+#
 # (0.0.0.0 and :: are always blacklisted, whether or not they are explicitly
 # listed here, since they correspond to unroutable addresses.)
 #
@@ -136,8 +139,8 @@ federation_ip_range_blacklist:
 #
 #   type: the type of listener. Normally 'http', but other valid options are:
 #       'manhole' (see docs/manhole.md),
-#       'metrics' (see docs/metrics-howto.rst),
-#       'replication' (see docs/workers.rst).
+#       'metrics' (see docs/metrics-howto.md),
+#       'replication' (see docs/workers.md).
 #
 #   tls: set to true to enable TLS for this listener. Will use the TLS
 #       key/cert specified in tls_private_key_path / tls_certificate_path.
@@ -172,12 +175,12 @@ federation_ip_range_blacklist:
 #
 #   media: the media API (/_matrix/media).
 #
-#   metrics: the metrics interface. See docs/metrics-howto.rst.
+#   metrics: the metrics interface. See docs/metrics-howto.md.
 #
 #   openid: OpenID authentication.
 #
 #   replication: the HTTP replication API (/_synapse/replication). See
-#       docs/workers.rst.
+#       docs/workers.md.
 #
 #   static: static resources under synapse/static (/_matrix/static). (Mostly
 #       useful for 'fallback authentication'.)
@@ -201,7 +204,7 @@ listeners:
   # that unwraps TLS.
   #
   # If you plan to use a reverse proxy, please see
-  # https://github.com/matrix-org/synapse/blob/master/docs/reverse_proxy.rst.
+  # https://github.com/matrix-org/synapse/blob/master/docs/reverse_proxy.md.
   #
   - port: 8008
     tls: false
@@ -306,6 +309,19 @@ listeners:
 #
 #allow_per_room_profiles: false
 
+# How long to keep redacted events in unredacted form in the database. After
+# this period redacted events get replaced with their redacted form in the DB.
+#
+# Defaults to `7d`. Set to `null` to disable.
+#
+#redaction_retention_period: 28d
+
+# How long to track users' last seen time and IPs in the database.
+#
+# Defaults to `28d`. Set to `null` to disable clearing out of old rows.
+#
+#user_ips_max_age: 14d
+
 
 ## TLS ##
 
@@ -511,6 +527,9 @@ log_config: "CONFDIR/SERVERNAME.log.config"
 #   - one for login that ratelimits login requests based on the account the
 #     client is attempting to log into, based on the amount of failed login
 #     attempts for this account.
+#   - one for ratelimiting redactions by room admins. If this is not explicitly
+#     set then it uses the same ratelimiting as per rc_message. This is useful
+#     to allow room admins to deal with abuse quickly.
 #
 # The defaults are as shown below.
 #
@@ -532,6 +551,10 @@ log_config: "CONFDIR/SERVERNAME.log.config"
 #  failed_attempts:
 #    per_second: 0.17
 #    burst_count: 3
+#
+#rc_admin_redaction:
+#  per_second: 1
+#  burst_count: 50
 
 
 # Ratelimiting settings for incoming federation
@@ -923,8 +946,10 @@ uploads_path: "DATADIR/uploads"
 # by the Matrix Identity Service API specification:
 # https://matrix.org/docs/spec/identity_service/latest
 #
+# If a delegate is specified, the config option public_baseurl must also be filled out.
+#
 account_threepid_delegates:
-    #email: https://example.com     # Delegate email sending to matrix.org
+    #email: https://example.com     # Delegate email sending to example.org
     #msisdn: http://localhost:8090  # Delegate SMS sending to this local process
 
 # Users who register on this homeserver will automatically be joined
@@ -971,6 +996,11 @@ metrics_flags:
 # Whether or not to report anonymized homeserver usage statistics.
 # report_stats: true|false
 
+# The endpoint to report the anonymized homeserver usage statistics to.
+# Defaults to https://matrix.org/report-usage-stats/push
+#
+#report_stats_endpoint: https://example.com/report-usage-stats/push
+
 
 ## API Configuration ##
 
@@ -1042,6 +1072,10 @@ signing_key_path: "CONFDIR/SERVERNAME.signing.key"
 # This setting supercedes an older setting named `perspectives`. The old format
 # is still supported for backwards-compatibility, but it is deprecated.
 #
+# 'trusted_key_servers' defaults to matrix.org, but using it will generate a
+# warning on start-up. To suppress this warning, set
+# 'suppress_key_server_warning' to true.
+#
 # Options for each entry in the list include:
 #
 #    server_name: the name of the server. required.
@@ -1066,11 +1100,13 @@ signing_key_path: "CONFDIR/SERVERNAME.signing.key"
 #      "ed25519:auto": "abcdefghijklmnopqrstuvwxyzabcdefghijklmopqr"
 #  - server_name: "my_other_trusted_server.example.com"
 #
-# The default configuration is:
-#
-#trusted_key_servers:
-#  - server_name: "matrix.org"
+trusted_key_servers:
+  - server_name: "matrix.org"
+
+# Uncomment the following to disable the warning that is emitted when the
+# trusted_key_servers include 'matrix.org'. See above.
 #
+#suppress_key_server_warning: true
 
 # The signing keys to use when acting as a trusted key server. If not specified
 # defaults to the server signing key.
@@ -1082,12 +1118,13 @@ signing_key_path: "CONFDIR/SERVERNAME.signing.key"
 
 # Enable SAML2 for registration and login. Uses pysaml2.
 #
-# `sp_config` is the configuration for the pysaml2 Service Provider.
-# See pysaml2 docs for format of config.
+# At least one of `sp_config` or `config_path` must be set in this section to
+# enable SAML login.
 #
-# Default values will be used for the 'entityid' and 'service' settings,
-# so it is not normally necessary to specify them unless you need to
-# override them.
+# (You will probably also want to set the following options to `false` to
+# disable the regular login/registration flows:
+#   * enable_registration
+#   * password_config.enabled
 #
 # Once SAML support is enabled, a metadata file will be exposed at
 # https://<server>:<port>/_matrix/saml2/metadata.xml, which you may be able to
@@ -1095,52 +1132,85 @@ signing_key_path: "CONFDIR/SERVERNAME.signing.key"
 # the IdP to use an ACS location of
 # https://<server>:<port>/_matrix/saml2/authn_response.
 #
-#saml2_config:
-#  sp_config:
-#    # point this to the IdP's metadata. You can use either a local file or
-#    # (preferably) a URL.
-#    metadata:
-#      #local: ["saml2/idp.xml"]
-#      remote:
-#        - url: https://our_idp/metadata.xml
-#
-#    # By default, the user has to go to our login page first. If you'd like to
-#    # allow IdP-initiated login, set 'allow_unsolicited: True' in a
-#    # 'service.sp' section:
-#    #
-#    #service:
-#    #  sp:
-#    #    allow_unsolicited: True
-#
-#    # The examples below are just used to generate our metadata xml, and you
-#    # may well not need it, depending on your setup. Alternatively you
-#    # may need a whole lot more detail - see the pysaml2 docs!
-#
-#    description: ["My awesome SP", "en"]
-#    name: ["Test SP", "en"]
-#
-#    organization:
-#      name: Example com
-#      display_name:
-#        - ["Example co", "en"]
-#      url: "http://example.com"
-#
-#    contact_person:
-#      - given_name: Bob
-#        sur_name: "the Sysadmin"
-#        email_address": ["admin@example.com"]
-#        contact_type": technical
-#
-#  # Instead of putting the config inline as above, you can specify a
-#  # separate pysaml2 configuration file:
-#  #
-#  config_path: "CONFDIR/sp_conf.py"
-#
-#  # the lifetime of a SAML session. This defines how long a user has to
-#  # complete the authentication process, if allow_unsolicited is unset.
-#  # The default is 5 minutes.
-#  #
-#  # saml_session_lifetime: 5m
+saml2_config:
+  # `sp_config` is the configuration for the pysaml2 Service Provider.
+  # See pysaml2 docs for format of config.
+  #
+  # Default values will be used for the 'entityid' and 'service' settings,
+  # so it is not normally necessary to specify them unless you need to
+  # override them.
+  #
+  #sp_config:
+  #  # point this to the IdP's metadata. You can use either a local file or
+  #  # (preferably) a URL.
+  #  metadata:
+  #    #local: ["saml2/idp.xml"]
+  #    remote:
+  #      - url: https://our_idp/metadata.xml
+  #
+  #    # By default, the user has to go to our login page first. If you'd like
+  #    # to allow IdP-initiated login, set 'allow_unsolicited: True' in a
+  #    # 'service.sp' section:
+  #    #
+  #    #service:
+  #    #  sp:
+  #    #    allow_unsolicited: true
+  #
+  #    # The examples below are just used to generate our metadata xml, and you
+  #    # may well not need them, depending on your setup. Alternatively you
+  #    # may need a whole lot more detail - see the pysaml2 docs!
+  #
+  #    description: ["My awesome SP", "en"]
+  #    name: ["Test SP", "en"]
+  #
+  #    organization:
+  #      name: Example com
+  #      display_name:
+  #        - ["Example co", "en"]
+  #      url: "http://example.com"
+  #
+  #    contact_person:
+  #      - given_name: Bob
+  #        sur_name: "the Sysadmin"
+  #        email_address": ["admin@example.com"]
+  #        contact_type": technical
+
+  # Instead of putting the config inline as above, you can specify a
+  # separate pysaml2 configuration file:
+  #
+  #config_path: "CONFDIR/sp_conf.py"
+
+  # the lifetime of a SAML session. This defines how long a user has to
+  # complete the authentication process, if allow_unsolicited is unset.
+  # The default is 5 minutes.
+  #
+  #saml_session_lifetime: 5m
+
+  # The SAML attribute (after mapping via the attribute maps) to use to derive
+  # the Matrix ID from. 'uid' by default.
+  #
+  #mxid_source_attribute: displayName
+
+  # The mapping system to use for mapping the saml attribute onto a matrix ID.
+  # Options include:
+  #  * 'hexencode' (which maps unpermitted characters to '=xx')
+  #  * 'dotreplace' (which replaces unpermitted characters with '.').
+  # The default is 'hexencode'.
+  #
+  #mxid_mapping: dotreplace
+
+  # In previous versions of synapse, the mapping from SAML attribute to MXID was
+  # always calculated dynamically rather than stored in a table. For backwards-
+  # compatibility, we will look for user_ids matching such a pattern before
+  # creating a new account.
+  #
+  # This setting controls the SAML attribute which will be used for this
+  # backwards-compatibility lookup. Typically it should be 'uid', but if the
+  # attribute maps are changed, it may be necessary to change it.
+  #
+  # The default is 'uid'.
+  #
+  #grandfathered_mxid_source_attribute: upn
 
 
 
@@ -1150,6 +1220,7 @@ signing_key_path: "CONFDIR/SERVERNAME.signing.key"
 #   enabled: true
 #   server_url: "https://cas-server.com"
 #   service_url: "https://homeserver.domain.com:8448"
+#   #displayname_attribute: name
 #   #required_attributes:
 #   #    name: value
 
@@ -1242,6 +1313,12 @@ password_config:
 #   #registration_template_html: registration.html
 #   #registration_template_text: registration.txt
 #
+#   # Templates for validation emails sent by the homeserver when adding an email to
+#   # your user account
+#   #
+#   #add_threepid_template_html: add_threepid.html
+#   #add_threepid_template_text: add_threepid.txt
+#
 #   # Templates for password reset success and failure pages that a user
 #   # will see after attempting to reset their password
 #   #
@@ -1253,6 +1330,12 @@ password_config:
 #   #
 #   #registration_template_success_html: registration_success.html
 #   #registration_template_failure_html: registration_failure.html
+#
+#   # Templates for success and failure pages that a user will see after attempting
+#   # to add an email or phone to their account
+#   #
+#   #add_threepid_success_html: add_threepid_success.html
+#   #add_threepid_failure_html: add_threepid_failure.html
 
 
 #password_providers:
diff --git a/docs/tcp_replication.md b/docs/tcp_replication.md
new file mode 100644
index 0000000000..e099d8a87b
--- /dev/null
+++ b/docs/tcp_replication.md
@@ -0,0 +1,249 @@
+# TCP Replication
+
+## Motivation
+
+Previously the workers used an HTTP long poll mechanism to get updates
+from the master, which had the problem of causing a lot of duplicate
+work on the server. This TCP protocol replaces those APIs with the aim
+of increased efficiency.
+
+## Overview
+
+The protocol is based on fire and forget, line based commands. An
+example flow would be (where '>' indicates master to worker and
+'<' worker to master flows):
+
+    > SERVER example.com
+    < REPLICATE events 53
+    > RDATA events 54 ["$foo1:bar.com", ...]
+    > RDATA events 55 ["$foo4:bar.com", ...]
+
+The example shows the server accepting a new connection and sending its
+identity with the `SERVER` command, followed by the client asking to
+subscribe to the `events` stream from the token `53`. The server then
+periodically sends `RDATA` commands which have the format
+`RDATA <stream_name> <token> <row>`, where the format of `<row>` is
+defined by the individual streams.
+
+Error reporting happens by either the client or server sending an ERROR
+command, and usually the connection will be closed.
+
+Since the protocol is a simple line based, its possible to manually
+connect to the server using a tool like netcat. A few things should be
+noted when manually using the protocol:
+
+-   When subscribing to a stream using `REPLICATE`, the special token
+    `NOW` can be used to get all future updates. The special stream name
+    `ALL` can be used with `NOW` to subscribe to all available streams.
+-   The federation stream is only available if federation sending has
+    been disabled on the main process.
+-   The server will only time connections out that have sent a `PING`
+    command. If a ping is sent then the connection will be closed if no
+    further commands are receieved within 15s. Both the client and
+    server protocol implementations will send an initial PING on
+    connection and ensure at least one command every 5s is sent (not
+    necessarily `PING`).
+-   `RDATA` commands *usually* include a numeric token, however if the
+    stream has multiple rows to replicate per token the server will send
+    multiple `RDATA` commands, with all but the last having a token of
+    `batch`. See the documentation on `commands.RdataCommand` for
+    further details.
+
+## Architecture
+
+The basic structure of the protocol is line based, where the initial
+word of each line specifies the command. The rest of the line is parsed
+based on the command. For example, the RDATA command is defined as:
+
+    RDATA <stream_name> <token> <row_json>
+
+(Note that <row_json> may contains spaces, but cannot contain
+newlines.)
+
+Blank lines are ignored.
+
+### Keep alives
+
+Both sides are expected to send at least one command every 5s or so, and
+should send a `PING` command if necessary. If either side do not receive
+a command within e.g. 15s then the connection should be closed.
+
+Because the server may be connected to manually using e.g. netcat, the
+timeouts aren't enabled until an initial `PING` command is seen. Both
+the client and server implementations below send a `PING` command
+immediately on connection to ensure the timeouts are enabled.
+
+This ensures that both sides can quickly realize if the tcp connection
+has gone and handle the situation appropriately.
+
+### Start up
+
+When a new connection is made, the server:
+
+-   Sends a `SERVER` command, which includes the identity of the server,
+    allowing the client to detect if its connected to the expected
+    server
+-   Sends a `PING` command as above, to enable the client to time out
+    connections promptly.
+
+The client:
+
+-   Sends a `NAME` command, allowing the server to associate a human
+    friendly name with the connection. This is optional.
+-   Sends a `PING` as above
+-   For each stream the client wishes to subscribe to it sends a
+    `REPLICATE` with the `stream_name` and token it wants to subscribe
+    from.
+-   On receipt of a `SERVER` command, checks that the server name
+    matches the expected server name.
+
+### Error handling
+
+If either side detects an error it can send an `ERROR` command and close
+the connection.
+
+If the client side loses the connection to the server it should
+reconnect, following the steps above.
+
+### Congestion
+
+If the server sends messages faster than the client can consume them the
+server will first buffer a (fairly large) number of commands and then
+disconnect the client. This ensures that we don't queue up an unbounded
+number of commands in memory and gives us a potential oppurtunity to
+squawk loudly. When/if the client recovers it can reconnect to the
+server and ask for missed messages.
+
+### Reliability
+
+In general the replication stream should be considered an unreliable
+transport since e.g. commands are not resent if the connection
+disappears.
+
+The exception to that are the replication streams, i.e. RDATA commands,
+since these include tokens which can be used to restart the stream on
+connection errors.
+
+The client should keep track of the token in the last RDATA command
+received for each stream so that on reconneciton it can start streaming
+from the correct place. Note: not all RDATA have valid tokens due to
+batching. See `RdataCommand` for more details.
+
+### Example
+
+An example iteraction is shown below. Each line is prefixed with '>'
+or '<' to indicate which side is sending, these are *not* included on
+the wire:
+
+    * connection established *
+    > SERVER localhost:8823
+    > PING 1490197665618
+    < NAME synapse.app.appservice
+    < PING 1490197665618
+    < REPLICATE events 1
+    < REPLICATE backfill 1
+    < REPLICATE caches 1
+    > POSITION events 1
+    > POSITION backfill 1
+    > POSITION caches 1
+    > RDATA caches 2 ["get_user_by_id",["@01register-user:localhost:8823"],1490197670513]
+    > RDATA events 14 ["$149019767112vOHxz:localhost:8823",
+        "!AFDCvgApUmpdfVjIXm:localhost:8823","m.room.guest_access","",null]
+    < PING 1490197675618
+    > ERROR server stopping
+    * connection closed by server *
+
+The `POSITION` command sent by the server is used to set the clients
+position without needing to send data with the `RDATA` command.
+
+An example of a batched set of `RDATA` is:
+
+    > RDATA caches batch ["get_user_by_id",["@test:localhost:8823"],1490197670513]
+    > RDATA caches batch ["get_user_by_id",["@test2:localhost:8823"],1490197670513]
+    > RDATA caches batch ["get_user_by_id",["@test3:localhost:8823"],1490197670513]
+    > RDATA caches 54 ["get_user_by_id",["@test4:localhost:8823"],1490197670513]
+
+In this case the client shouldn't advance their caches token until it
+sees the the last `RDATA`.
+
+### List of commands
+
+The list of valid commands, with which side can send it: server (S) or
+client (C):
+
+#### SERVER (S)
+
+   Sent at the start to identify which server the client is talking to
+
+#### RDATA (S)
+
+   A single update in a stream
+
+#### POSITION (S)
+
+   The position of the stream has been updated. Sent to the client
+    after all missing updates for a stream have been sent to the client
+    and they're now up to date.
+
+#### ERROR (S, C)
+
+   There was an error
+
+#### PING (S, C)
+
+   Sent periodically to ensure the connection is still alive
+
+#### NAME (C)
+
+   Sent at the start by client to inform the server who they are
+
+#### REPLICATE (C)
+
+   Asks the server to replicate a given stream
+
+#### USER_SYNC (C)
+
+   A user has started or stopped syncing
+
+#### FEDERATION_ACK (C)
+
+   Acknowledge receipt of some federation data
+
+#### REMOVE_PUSHER (C)
+
+   Inform the server a pusher should be removed
+
+#### INVALIDATE_CACHE (C)
+
+   Inform the server a cache should be invalidated
+
+#### SYNC (S, C)
+
+   Used exclusively in tests
+
+See `synapse/replication/tcp/commands.py` for a detailed description and
+the format of each command.
+
+### Cache Invalidation Stream
+
+The cache invalidation stream is used to inform workers when they need
+to invalidate any of their caches in the data store. This is done by
+streaming all cache invalidations done on master down to the workers,
+assuming that any caches on the workers also exist on the master.
+
+Each individual cache invalidation results in a row being sent down
+replication, which includes the cache name (the name of the function)
+and they key to invalidate. For example:
+
+    > RDATA caches 550953771 ["get_user_by_id", ["@bob:example.com"], 1550574873251]
+
+However, there are times when a number of caches need to be invalidated
+at the same time with the same key. To reduce traffic we batch those
+invalidations into a single poke by defining a special cache name that
+workers understand to mean to expand to invalidate the correct caches.
+
+Currently the special cache names are declared in
+`synapse/storage/_base.py` and are:
+
+1.  `cs_cache_fake` ─ invalidates caches that depend on the current
+    state
diff --git a/docs/tcp_replication.rst b/docs/tcp_replication.rst
deleted file mode 100644
index 75e723484c..0000000000
--- a/docs/tcp_replication.rst
+++ /dev/null
@@ -1,249 +0,0 @@
-TCP Replication
-===============
-
-Motivation
-----------
-
-Previously the workers used an HTTP long poll mechanism to get updates from the
-master, which had the problem of causing a lot of duplicate work on the server.
-This TCP protocol replaces those APIs with the aim of increased efficiency.
-
-
-
-Overview
---------
-
-The protocol is based on fire and forget, line based commands. An example flow
-would be (where '>' indicates master to worker and '<' worker to master flows)::
-
-    > SERVER example.com
-    < REPLICATE events 53
-    > RDATA events 54 ["$foo1:bar.com", ...]
-    > RDATA events 55 ["$foo4:bar.com", ...]
-
-The example shows the server accepting a new connection and sending its identity
-with the ``SERVER`` command, followed by the client asking to subscribe to the
-``events`` stream from the token ``53``. The server then periodically sends ``RDATA``
-commands which have the format ``RDATA <stream_name> <token> <row>``, where the
-format of ``<row>`` is defined by the individual streams.
-
-Error reporting happens by either the client or server sending an `ERROR`
-command, and usually the connection will be closed.
-
-
-Since the protocol is a simple line based, its possible to manually connect to
-the server using a tool like netcat. A few things should be noted when manually
-using the protocol:
-
-* When subscribing to a stream using ``REPLICATE``, the special token ``NOW`` can
-  be used to get all future updates. The special stream name ``ALL`` can be used
-  with ``NOW`` to subscribe to all available streams.
-* The federation stream is only available if federation sending has been
-  disabled on the main process.
-* The server will only time connections out that have sent a ``PING`` command.
-  If a ping is sent then the connection will be closed if no further commands
-  are receieved within 15s. Both the client and server protocol implementations
-  will send an initial PING on connection and ensure at least one command every
-  5s is sent (not necessarily ``PING``).
-* ``RDATA`` commands *usually* include a numeric token, however if the stream
-  has multiple rows to replicate per token the server will send multiple
-  ``RDATA`` commands, with all but the last having a token of ``batch``. See
-  the documentation on ``commands.RdataCommand`` for further details.
-
-
-Architecture
-------------
-
-The basic structure of the protocol is line based, where the initial word of
-each line specifies the command. The rest of the line is parsed based on the
-command. For example, the `RDATA` command is defined as::
-
-    RDATA <stream_name> <token> <row_json>
-
-(Note that `<row_json>` may contains spaces, but cannot contain newlines.)
-
-Blank lines are ignored.
-
-
-Keep alives
-~~~~~~~~~~~
-
-Both sides are expected to send at least one command every 5s or so, and
-should send a ``PING`` command if necessary. If either side do not receive a
-command within e.g. 15s then the connection should be closed.
-
-Because the server may be connected to manually using e.g. netcat, the timeouts
-aren't enabled until an initial ``PING`` command is seen. Both the client and
-server implementations below send a ``PING`` command immediately on connection to
-ensure the timeouts are enabled.
-
-This ensures that both sides can quickly realize if the tcp connection has gone
-and handle the situation appropriately.
-
-
-Start up
-~~~~~~~~
-
-When a new connection is made, the server:
-
-* Sends a ``SERVER`` command, which includes the identity of the server, allowing
-  the client to detect if its connected to the expected server
-* Sends a ``PING`` command as above, to enable the client to time out connections
-  promptly.
-
-The client:
-
-* Sends a ``NAME`` command, allowing the server to associate a human friendly
-  name with the connection. This is optional.
-* Sends a ``PING`` as above
-* For each stream the client wishes to subscribe to it sends a ``REPLICATE``
-  with the stream_name and token it wants to subscribe from.
-* On receipt of a ``SERVER`` command, checks that the server name matches the
-  expected server name.
-
-
-Error handling
-~~~~~~~~~~~~~~
-
-If either side detects an error it can send an ``ERROR`` command and close the
-connection.
-
-If the client side loses the connection to the server it should reconnect,
-following the steps above.
-
-
-Congestion
-~~~~~~~~~~
-
-If the server sends messages faster than the client can consume them the server
-will first buffer a (fairly large) number of commands and then disconnect the
-client. This ensures that we don't queue up an unbounded number of commands in
-memory and gives us a potential oppurtunity to squawk loudly. When/if the client
-recovers it can reconnect to the server and ask for missed messages.
-
-
-Reliability
-~~~~~~~~~~~
-
-In general the replication stream should be considered an unreliable transport
-since e.g. commands are not resent if the connection disappears.
-
-The exception to that are the replication streams, i.e. RDATA commands, since
-these include tokens which can be used to restart the stream on connection
-errors.
-
-The client should keep track of the token in the last RDATA command received
-for each stream so that on reconneciton it can start streaming from the correct
-place. Note: not all RDATA have valid tokens due to batching. See
-``RdataCommand`` for more details.
-
-Example
-~~~~~~~
-
-An example iteraction is shown below. Each line is prefixed with '>' or '<' to
-indicate which side is sending, these are *not* included on the wire::
-
-    * connection established *
-    > SERVER localhost:8823
-    > PING 1490197665618
-    < NAME synapse.app.appservice
-    < PING 1490197665618
-    < REPLICATE events 1
-    < REPLICATE backfill 1
-    < REPLICATE caches 1
-    > POSITION events 1
-    > POSITION backfill 1
-    > POSITION caches 1
-    > RDATA caches 2 ["get_user_by_id",["@01register-user:localhost:8823"],1490197670513]
-    > RDATA events 14 ["$149019767112vOHxz:localhost:8823",
-        "!AFDCvgApUmpdfVjIXm:localhost:8823","m.room.guest_access","",null]
-    < PING 1490197675618
-    > ERROR server stopping
-    * connection closed by server *
-
-The ``POSITION`` command sent by the server is used to set the clients position
-without needing to send data with the ``RDATA`` command.
-
-
-An example of a batched set of ``RDATA`` is::
-
-    > RDATA caches batch ["get_user_by_id",["@test:localhost:8823"],1490197670513]
-    > RDATA caches batch ["get_user_by_id",["@test2:localhost:8823"],1490197670513]
-    > RDATA caches batch ["get_user_by_id",["@test3:localhost:8823"],1490197670513]
-    > RDATA caches 54 ["get_user_by_id",["@test4:localhost:8823"],1490197670513]
-
-In this case the client shouldn't advance their caches token until it sees the
-the last ``RDATA``.
-
-
-List of commands
-~~~~~~~~~~~~~~~~
-
-The list of valid commands, with which side can send it: server (S) or client (C):
-
-SERVER (S)
-    Sent at the start to identify which server the client is talking to
-
-RDATA (S)
-    A single update in a stream
-
-POSITION (S)
-    The position of the stream has been updated. Sent to the client after all
-    missing updates for a stream have been sent to the client and they're now
-    up to date.
-
-ERROR (S, C)
-    There was an error
-
-PING (S, C)
-    Sent periodically to ensure the connection is still alive
-
-NAME (C)
-    Sent at the start by client to inform the server who they are
-
-REPLICATE (C)
-    Asks the server to replicate a given stream
-
-USER_SYNC (C)
-    A user has started or stopped syncing
-
-FEDERATION_ACK (C)
-    Acknowledge receipt of some federation data
-
-REMOVE_PUSHER (C)
-    Inform the server a pusher should be removed
-
-INVALIDATE_CACHE (C)
-    Inform the server a cache should be invalidated
-
-SYNC (S, C)
-    Used exclusively in tests
-
-
-See ``synapse/replication/tcp/commands.py`` for a detailed description and the
-format of each command.
-
-
-Cache Invalidation Stream
-~~~~~~~~~~~~~~~~~~~~~~~~~
-
-The cache invalidation stream is used to inform workers when they need to
-invalidate any of their caches in the data store. This is done by streaming all
-cache invalidations done on master down to the workers, assuming that any caches
-on the workers also exist on the master.
-
-Each individual cache invalidation results in a row being sent down replication,
-which includes the cache name (the name of the function) and they key to
-invalidate. For example::
-
-    > RDATA caches 550953771 ["get_user_by_id", ["@bob:example.com"], 1550574873251]
-
-However, there are times when a number of caches need to be invalidated at the
-same time with the same key. To reduce traffic we batch those invalidations into
-a single poke by defining a special cache name that workers understand to mean
-to expand to invalidate the correct caches.
-
-Currently the special cache names are declared in ``synapse/storage/_base.py``
-and are:
-
-1. ``cs_cache_fake`` ─ invalidates caches that depend on the current state
diff --git a/docs/turn-howto.md b/docs/turn-howto.md
new file mode 100644
index 0000000000..4a983621e5
--- /dev/null
+++ b/docs/turn-howto.md
@@ -0,0 +1,123 @@
+# Overview
+
+This document explains how to enable VoIP relaying on your Home Server with
+TURN.
+
+The synapse Matrix Home Server supports integration with TURN server via the
+[TURN server REST API](<http://tools.ietf.org/html/draft-uberti-behave-turn-rest-00>). This
+allows the Home Server to generate credentials that are valid for use on the
+TURN server through the use of a secret shared between the Home Server and the
+TURN server.
+
+The following sections describe how to install [coturn](<https://github.com/coturn/coturn>) (which implements the TURN REST API) and integrate it with synapse.
+
+## `coturn` Setup
+
+### Initial installation
+
+The TURN daemon `coturn` is available from a variety of sources such as native package managers, or installation from source.
+
+#### Debian installation
+
+    # apt install coturn
+
+#### Source installation
+
+1. Download the [latest release](https://github.com/coturn/coturn/releases/latest) from github.  Unpack it and `cd` into the directory.
+
+1.  Configure it:
+
+        ./configure
+
+    > You may need to install `libevent2`: if so, you should do so in
+    > the way recommended by your operating system. You can ignore
+    > warnings about lack of database support: a database is unnecessary
+    > for this purpose.
+
+1.  Build and install it:
+
+        make
+        make install
+
+1.  Create or edit the config file in `/etc/turnserver.conf`. The relevant
+    lines, with example values, are:
+
+        use-auth-secret
+        static-auth-secret=[your secret key here]
+        realm=turn.myserver.org
+
+    See `turnserver.conf` for explanations of the options. One way to generate
+    the `static-auth-secret` is with `pwgen`:
+
+        pwgen -s 64 1
+
+1.  Consider your security settings. TURN lets users request a relay which will
+    connect to arbitrary IP addresses and ports. The following configuration is
+    suggested as a minimum starting point:
+    
+        # VoIP traffic is all UDP. There is no reason to let users connect to arbitrary TCP endpoints via the relay.
+        no-tcp-relay
+        
+        # don't let the relay ever try to connect to private IP address ranges within your network (if any)
+        # given the turn server is likely behind your firewall, remember to include any privileged public IPs too.
+        denied-peer-ip=10.0.0.0-10.255.255.255
+        denied-peer-ip=192.168.0.0-192.168.255.255
+        denied-peer-ip=172.16.0.0-172.31.255.255
+        
+        # special case the turn server itself so that client->TURN->TURN->client flows work
+        allowed-peer-ip=10.0.0.1
+        
+        # consider whether you want to limit the quota of relayed streams per user (or total) to avoid risk of DoS.
+        user-quota=12 # 4 streams per video call, so 12 streams = 3 simultaneous relayed calls per user.
+        total-quota=1200
+
+    Ideally coturn should refuse to relay traffic which isn't SRTP; see
+    <https://github.com/matrix-org/synapse/issues/2009>
+
+1.  Ensure your firewall allows traffic into the TURN server on the ports
+    you've configured it to listen on (remember to allow both TCP and UDP TURN
+    traffic)
+
+1.  If you've configured coturn to support TLS/DTLS, generate or import your
+    private key and certificate.
+
+1.  Start the turn server:
+
+         bin/turnserver -o
+
+## synapse Setup
+
+Your home server configuration file needs the following extra keys:
+
+1.  "`turn_uris`": This needs to be a yaml list of public-facing URIs
+    for your TURN server to be given out to your clients. Add separate
+    entries for each transport your TURN server supports.
+2.  "`turn_shared_secret`": This is the secret shared between your
+    Home server and your TURN server, so you should set it to the same
+    string you used in turnserver.conf.
+3.  "`turn_user_lifetime`": This is the amount of time credentials
+    generated by your Home Server are valid for (in milliseconds).
+    Shorter times offer less potential for abuse at the expense of
+    increased traffic between web clients and your home server to
+    refresh credentials. The TURN REST API specification recommends
+    one day (86400000).
+4.  "`turn_allow_guests`": Whether to allow guest users to use the
+    TURN server. This is enabled by default, as otherwise VoIP will
+    not work reliably for guests. However, it does introduce a
+    security risk as it lets guests connect to arbitrary endpoints
+    without having gone through a CAPTCHA or similar to register a
+    real account.
+
+As an example, here is the relevant section of the config file for matrix.org:
+
+    turn_uris: [ "turn:turn.matrix.org:3478?transport=udp", "turn:turn.matrix.org:3478?transport=tcp" ]
+    turn_shared_secret: n0t4ctuAllymatr1Xd0TorgSshar3d5ecret4obvIousreAsons
+    turn_user_lifetime: 86400000
+    turn_allow_guests: True
+
+After updating the homeserver configuration, you must restart synapse:
+
+    cd /where/you/run/synapse
+    ./synctl restart
+
+..and your Home Server now supports VoIP relaying!
diff --git a/docs/turn-howto.rst b/docs/turn-howto.rst
deleted file mode 100644
index a2fc5c8820..0000000000
--- a/docs/turn-howto.rst
+++ /dev/null
@@ -1,127 +0,0 @@
-How to enable VoIP relaying on your Home Server with TURN
-
-Overview
---------
-The synapse Matrix Home Server supports integration with TURN server via the
-TURN server REST API
-(http://tools.ietf.org/html/draft-uberti-behave-turn-rest-00). This allows
-the Home Server to generate credentials that are valid for use on the TURN
-server through the use of a secret shared between the Home Server and the
-TURN server.
-
-This document describes how to install coturn
-(https://github.com/coturn/coturn) which also supports the TURN REST API,
-and integrate it with synapse.
-
-coturn Setup
-============
-
-You may be able to setup coturn via your package manager,  or set it up manually using the usual ``configure, make, make install`` process.  
-
- 1. Check out coturn::
- 
-      git clone https://github.com/coturn/coturn.git coturn
-      cd coturn
-
- 2. Configure it::
- 
-      ./configure
-
-    You may need to install ``libevent2``: if so, you should do so
-    in the way recommended by your operating system.
-    You can ignore warnings about lack of database support: a
-    database is unnecessary for this purpose.
-
- 3. Build and install it::
- 
-      make
-      make install
-
- 4. Create or edit the config file in ``/etc/turnserver.conf``. The relevant
-    lines, with example values, are::
-
-      use-auth-secret
-      static-auth-secret=[your secret key here]
-      realm=turn.myserver.org
-
-    See turnserver.conf for explanations of the options.
-    One way to generate the static-auth-secret is with pwgen::
-
-       pwgen -s 64 1
-
- 5. Consider your security settings.  TURN lets users request a relay
-    which will connect to arbitrary IP addresses and ports.  At the least
-    we recommend::
-
-       # VoIP traffic is all UDP. There is no reason to let users connect to arbitrary TCP endpoints via the relay.
-       no-tcp-relay
-
-       # don't let the relay ever try to connect to private IP address ranges within your network (if any)
-       # given the turn server is likely behind your firewall, remember to include any privileged public IPs too.
-       denied-peer-ip=10.0.0.0-10.255.255.255
-       denied-peer-ip=192.168.0.0-192.168.255.255
-       denied-peer-ip=172.16.0.0-172.31.255.255
-
-       # special case the turn server itself so that client->TURN->TURN->client flows work
-       allowed-peer-ip=10.0.0.1
-
-       # consider whether you want to limit the quota of relayed streams per user (or total) to avoid risk of DoS.
-       user-quota=12 # 4 streams per video call, so 12 streams = 3 simultaneous relayed calls per user.
-       total-quota=1200
-
-    Ideally coturn should refuse to relay traffic which isn't SRTP;
-    see https://github.com/matrix-org/synapse/issues/2009
-
- 6. Ensure your firewall allows traffic into the TURN server on
-    the ports you've configured it to listen on (remember to allow
-    both TCP and UDP TURN traffic)
-
- 7. If you've configured coturn to support TLS/DTLS, generate or
-    import your private key and certificate.
-
- 8. Start the turn server::
- 
-       bin/turnserver -o
-
-
-synapse Setup
-=============
-
-Your home server configuration file needs the following extra keys:
-
- 1. "turn_uris": This needs to be a yaml list
-    of public-facing URIs for your TURN server to be given out 
-    to your clients. Add separate entries for each transport your
-    TURN server supports.
-
- 2. "turn_shared_secret": This is the secret shared between your Home
-    server and your TURN server, so you should set it to the same
-    string you used in turnserver.conf.
-
- 3. "turn_user_lifetime": This is the amount of time credentials
-    generated by your Home Server are valid for (in milliseconds).
-    Shorter times offer less potential for abuse at the expense
-    of increased traffic between web clients and your home server
-    to refresh credentials. The TURN REST API specification recommends
-    one day (86400000).
-
- 4. "turn_allow_guests": Whether to allow guest users to use the TURN
-    server.  This is enabled by default, as otherwise VoIP will not
-    work reliably for guests.  However, it does introduce a security risk
-    as it lets guests connect to arbitrary endpoints without having gone
-    through a CAPTCHA or similar to register a real account.
-
-As an example, here is the relevant section of the config file for
-matrix.org::
-
-    turn_uris: [ "turn:turn.matrix.org:3478?transport=udp", "turn:turn.matrix.org:3478?transport=tcp" ]
-    turn_shared_secret: n0t4ctuAllymatr1Xd0TorgSshar3d5ecret4obvIousreAsons
-    turn_user_lifetime: 86400000
-    turn_allow_guests: True
-
-Now, restart synapse::
-
-    cd /where/you/run/synapse
-    ./synctl restart
-
-...and your Home Server now supports VoIP relaying!
diff --git a/docs/workers.rst b/docs/workers.md
index e11e117418..4bd60ba0a0 100644
--- a/docs/workers.rst
+++ b/docs/workers.md
@@ -1,5 +1,4 @@
-Scaling synapse via workers
-===========================
+# Scaling synapse via workers
 
 Synapse has experimental support for splitting out functionality into
 multiple separate python processes, helping greatly with scalability.  These
@@ -20,17 +19,16 @@ TCP protocol called 'replication' - analogous to MySQL or Postgres style
 database replication; feeding a stream of relevant data to the workers so they
 can be kept in sync with the main synapse process and database state.
 
-Configuration
--------------
+## Configuration
 
 To make effective use of the workers, you will need to configure an HTTP
 reverse-proxy such as nginx or haproxy, which will direct incoming requests to
 the correct worker, or to the main synapse instance. Note that this includes
-requests made to the federation port. See `<reverse_proxy.rst>`_ for
-information on setting up a reverse proxy.
+requests made to the federation port. See [reverse_proxy.md](reverse_proxy.md)
+for information on setting up a reverse proxy.
 
 To enable workers, you need to add two replication listeners to the master
-synapse, e.g.::
+synapse, e.g.:
 
     listeners:
       # The TCP replication port
@@ -56,7 +54,7 @@ You then create a set of configs for the various worker processes.  These
 should be worker configuration files, and should be stored in a dedicated
 subdirectory, to allow synctl to manipulate them. An additional configuration
 for the master synapse process will need to be created because the process will
-not be started automatically. That configuration should look like this::
+not be started automatically. That configuration should look like this:
 
     worker_app: synapse.app.homeserver
     daemonize: true
@@ -66,17 +64,17 @@ configuration file.  You can then override configuration specific to that worker
 e.g. the HTTP listener that it provides (if any); logging configuration; etc.
 You should minimise the number of overrides though to maintain a usable config.
 
-You must specify the type of worker application (``worker_app``). The currently
+You must specify the type of worker application (`worker_app`). The currently
 available worker applications are listed below. You must also specify the
 replication endpoints that it's talking to on the main synapse process.
-``worker_replication_host`` should specify the host of the main synapse,
-``worker_replication_port`` should point to the TCP replication listener port and
-``worker_replication_http_port`` should point to the HTTP replication port.
+`worker_replication_host` should specify the host of the main synapse,
+`worker_replication_port` should point to the TCP replication listener port and
+`worker_replication_http_port` should point to the HTTP replication port.
 
-Currently, the ``event_creator`` and ``federation_reader`` workers require specifying
-``worker_replication_http_port``.
+Currently, the `event_creator` and `federation_reader` workers require specifying
+`worker_replication_http_port`.
 
-For instance::
+For instance:
 
     worker_app: synapse.app.synchrotron
 
@@ -97,15 +95,15 @@ For instance::
     worker_log_config: /home/matrix/synapse/config/synchrotron_log_config.yaml
 
 ...is a full configuration for a synchrotron worker instance, which will expose a
-plain HTTP ``/sync`` endpoint on port 8083 separately from the ``/sync`` endpoint provided
+plain HTTP `/sync` endpoint on port 8083 separately from the `/sync` endpoint provided
 by the main synapse.
 
 Obviously you should configure your reverse-proxy to route the relevant
-endpoints to the worker (``localhost:8083`` in the above example).
+endpoints to the worker (`localhost:8083` in the above example).
 
 Finally, to actually run your worker-based synapse, you must pass synctl the -a
 commandline option to tell it to operate on all the worker configurations found
-in the given directory, e.g.::
+in the given directory, e.g.:
 
     synctl -a $CONFIG/workers start
 
@@ -114,28 +112,24 @@ synapse, unless you explicitly know it's safe not to.  For instance, restarting
 synapse without restarting all the synchrotrons may result in broken typing
 notifications.
 
-To manipulate a specific worker, you pass the -w option to synctl::
+To manipulate a specific worker, you pass the -w option to synctl:
 
     synctl -w $CONFIG/workers/synchrotron.yaml restart
 
+## Available worker applications
 
-Available worker applications
------------------------------
-
-``synapse.app.pusher``
-~~~~~~~~~~~~~~~~~~~~~~
+### `synapse.app.pusher`
 
 Handles sending push notifications to sygnal and email. Doesn't handle any
-REST endpoints itself, but you should set ``start_pushers: False`` in the
+REST endpoints itself, but you should set `start_pushers: False` in the
 shared configuration file to stop the main synapse sending these notifications.
 
 Note this worker cannot be load-balanced: only one instance should be active.
 
-``synapse.app.synchrotron``
-~~~~~~~~~~~~~~~~~~~~~~~~~~~
+### `synapse.app.synchrotron`
 
-The synchrotron handles ``sync`` requests from clients. In particular, it can
-handle REST endpoints matching the following regular expressions::
+The synchrotron handles `sync` requests from clients. In particular, it can
+handle REST endpoints matching the following regular expressions:
 
     ^/_matrix/client/(v2_alpha|r0)/sync$
     ^/_matrix/client/(api/v1|v2_alpha|r0)/events$
@@ -151,20 +145,18 @@ load-balance across the instances, though it will be more efficient if all
 requests from a particular user are routed to a single instance. Extracting
 a userid from the access token is currently left as an exercise for the reader.
 
-``synapse.app.appservice``
-~~~~~~~~~~~~~~~~~~~~~~~~~~
+### `synapse.app.appservice`
 
 Handles sending output traffic to Application Services. Doesn't handle any
-REST endpoints itself, but you should set ``notify_appservices: False`` in the
+REST endpoints itself, but you should set `notify_appservices: False` in the
 shared configuration file to stop the main synapse sending these notifications.
 
 Note this worker cannot be load-balanced: only one instance should be active.
 
-``synapse.app.federation_reader``
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+### `synapse.app.federation_reader`
 
 Handles a subset of federation endpoints. In particular, it can handle REST
-endpoints matching the following regular expressions::
+endpoints matching the following regular expressions:
 
     ^/_matrix/federation/v1/event/
     ^/_matrix/federation/v1/state/
@@ -190,40 +182,36 @@ reverse-proxy configuration.
 The `^/_matrix/federation/v1/send/` endpoint must only be handled by a single
 instance.
 
-``synapse.app.federation_sender``
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+### `synapse.app.federation_sender`
 
 Handles sending federation traffic to other servers. Doesn't handle any
-REST endpoints itself, but you should set ``send_federation: False`` in the
+REST endpoints itself, but you should set `send_federation: False` in the
 shared configuration file to stop the main synapse sending this traffic.
 
 Note this worker cannot be load-balanced: only one instance should be active.
 
-``synapse.app.media_repository``
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+### `synapse.app.media_repository`
 
-Handles the media repository. It can handle all endpoints starting with::
+Handles the media repository. It can handle all endpoints starting with:
 
     /_matrix/media/
 
-And the following regular expressions matching media-specific administration
-APIs::
+And the following regular expressions matching media-specific administration APIs:
 
     ^/_synapse/admin/v1/purge_media_cache$
     ^/_synapse/admin/v1/room/.*/media$
     ^/_synapse/admin/v1/quarantine_media/.*$
 
-You should also set ``enable_media_repo: False`` in the shared configuration
+You should also set `enable_media_repo: False` in the shared configuration
 file to stop the main synapse running background jobs related to managing the
 media repository.
 
 Note this worker cannot be load-balanced: only one instance should be active.
 
-``synapse.app.client_reader``
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+### `synapse.app.client_reader`
 
 Handles client API endpoints. It can handle REST endpoints matching the
-following regular expressions::
+following regular expressions:
 
     ^/_matrix/client/(api/v1|r0|unstable)/publicRooms$
     ^/_matrix/client/(api/v1|r0|unstable)/rooms/.*/joined_members$
@@ -237,60 +225,55 @@ following regular expressions::
     ^/_matrix/client/versions$
     ^/_matrix/client/(api/v1|r0|unstable)/voip/turnServer$
 
-Additionally, the following REST endpoints can be handled for GET requests::
+Additionally, the following REST endpoints can be handled for GET requests:
 
     ^/_matrix/client/(api/v1|r0|unstable)/pushrules/.*$
 
 Additionally, the following REST endpoints can be handled, but all requests must
-be routed to the same instance::
+be routed to the same instance:
 
     ^/_matrix/client/(r0|unstable)/register$
 
 Pagination requests can also be handled, but all requests with the same path
 room must be routed to the same instance. Additionally, care must be taken to
 ensure that the purge history admin API is not used while pagination requests
-for the room are in flight::
+for the room are in flight:
 
     ^/_matrix/client/(api/v1|r0|unstable)/rooms/.*/messages$
 
-
-``synapse.app.user_dir``
-~~~~~~~~~~~~~~~~~~~~~~~~
+### `synapse.app.user_dir`
 
 Handles searches in the user directory. It can handle REST endpoints matching
-the following regular expressions::
+the following regular expressions:
 
     ^/_matrix/client/(api/v1|r0|unstable)/user_directory/search$
 
-``synapse.app.frontend_proxy``
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+### `synapse.app.frontend_proxy`
 
 Proxies some frequently-requested client endpoints to add caching and remove
 load from the main synapse. It can handle REST endpoints matching the following
-regular expressions::
+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::
+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.
+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
-file. For example::
+the `worker_main_http_uri` setting in the `frontend_proxy` worker configuration
+file. For example:
 
     worker_main_http_uri: http://127.0.0.1:8008
 
+### `synapse.app.event_creator`
 
-``synapse.app.event_creator``
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
-Handles some event creation. It can handle REST endpoints matching::
+Handles some event creation. It can handle REST endpoints matching:
 
     ^/_matrix/client/(api/v1|r0|unstable)/rooms/.*/send
     ^/_matrix/client/(api/v1|r0|unstable)/rooms/.*/(join|invite|leave|ban|unban|kick)$