Merge tag 'v1.4.0'

Synapse 1.4.0 (2019-10-03)
==========================

Bugfixes
--------

- Redact `client_secret` in server logs. ([\#6158](https://github.com/matrix-org/synapse/issues/6158))

Synapse 1.4.0rc2 (2019-10-02)
=============================

Bugfixes
--------

- Fix bug in background update that adds last seen information to the `devices` table, and improve its performance on Postgres. ([\#6135](https://github.com/matrix-org/synapse/issues/6135))
- Fix bad performance of censoring redactions background task. ([\#6141](https://github.com/matrix-org/synapse/issues/6141))
- Fix fetching censored redactions from DB, which caused APIs like initial sync to fail if it tried to include the censored redaction. ([\#6145](https://github.com/matrix-org/synapse/issues/6145))
- Fix exceptions when storing large retry intervals for down remote servers. ([\#6146](https://github.com/matrix-org/synapse/issues/6146))

Internal Changes
----------------

- Fix up sample config entry for `redaction_retention_period` option. ([\#6117](https://github.com/matrix-org/synapse/issues/6117))

Synapse 1.4.0rc1 (2019-09-26)
=============================

Note that this release includes significant changes around 3pid
verification. Administrators are reminded to review the [upgrade notes](UPGRADE.rst#upgrading-to-v140).

Features
--------

- Changes to 3pid verification:
  - Add the ability to send registration emails from the homeserver rather than delegating to an identity server. ([\#5835](https://github.com/matrix-org/synapse/issues/5835), [\#5940](https://github.com/matrix-org/synapse/issues/5940), [\#5993](https://github.com/matrix-org/synapse/issues/5993), [\#5994](https://github.com/matrix-org/synapse/issues/5994), [\#5868](https://github.com/matrix-org/synapse/issues/5868))
  - Replace `trust_identity_server_for_password_resets` config option with `account_threepid_delegates`, and make the `id_server` parameteter optional on `*/requestToken` endpoints, as per [MSC2263](https://github.com/matrix-org/matrix-doc/pull/2263). ([\#5876](https://github.com/matrix-org/synapse/issues/5876), [\#5969](https://github.com/matrix-org/synapse/issues/5969), [\#6028](https://github.com/matrix-org/synapse/issues/6028))
  - Switch to using the v2 Identity Service `/lookup` API where available, with fallback to v1. (Implements [MSC2134](https://github.com/matrix-org/matrix-doc/pull/2134) plus `id_access_token authentication` for v2 Identity Service APIs from [MSC2140](https://github.com/matrix-org/matrix-doc/pull/2140)). ([\#5897](https://github.com/matrix-org/synapse/issues/5897))
  - Remove `bind_email` and `bind_msisdn` parameters from `/register` ala [MSC2140](https://github.com/matrix-org/matrix-doc/pull/2140). ([\#5964](https://github.com/matrix-org/synapse/issues/5964))
  - Add `m.id_access_token` to `unstable_features` in `/versions` as per [MSC2264](https://github.com/matrix-org/matrix-doc/pull/2264). ([\#5974](https://github.com/matrix-org/synapse/issues/5974))
  - Use the v2 Identity Service API for 3PID invites. ([\#5979](https://github.com/matrix-org/synapse/issues/5979))
  - Add `POST /_matrix/client/unstable/account/3pid/unbind` endpoint from [MSC2140](https://github.com/matrix-org/matrix-doc/pull/2140) for unbinding a 3PID from an identity server without removing it from the homeserver user account. ([\#5980](https://github.com/matrix-org/synapse/issues/5980), [\#6062](https://github.com/matrix-org/synapse/issues/6062))
  - Use `account_threepid_delegate.email` and `account_threepid_delegate.msisdn` for validating threepid sessions. ([\#6011](https://github.com/matrix-org/synapse/issues/6011))
  - Allow homeserver to handle or delegate email validation when adding an email to a user's account. ([\#6042](https://github.com/matrix-org/synapse/issues/6042))
  - Implement new Client Server API endpoints `/account/3pid/add` and `/account/3pid/bind` as per [MSC2290](https://github.com/matrix-org/matrix-doc/pull/2290). ([\#6043](https://github.com/matrix-org/synapse/issues/6043))
  - Add an unstable feature flag for separate add/bind 3pid APIs. ([\#6044](https://github.com/matrix-org/synapse/issues/6044))
  - Remove `bind` parameter from Client Server POST `/account` endpoint as per [MSC2290](https://github.com/matrix-org/matrix-doc/pull/2290/). ([\#6067](https://github.com/matrix-org/synapse/issues/6067))
  - Add `POST /add_threepid/msisdn/submit_token` endpoint for proxying submitToken on an `account_threepid_handler`. ([\#6078](https://github.com/matrix-org/synapse/issues/6078))
  - Add `submit_url` response parameter to `*/msisdn/requestToken` endpoints. ([\#6079](https://github.com/matrix-org/synapse/issues/6079))
  - Add `m.require_identity_server` flag to /version's unstable_features. ([\#5972](https://github.com/matrix-org/synapse/issues/5972))
- Enhancements to OpenTracing support:
  - Make OpenTracing work in worker mode. ([\#5771](https://github.com/matrix-org/synapse/issues/5771))
  - Pass OpenTracing contexts between servers when transmitting EDUs. ([\#5852](https://github.com/matrix-org/synapse/issues/5852))
  - OpenTracing for device list updates. ([\#5853](https://github.com/matrix-org/synapse/issues/5853))
  - Add a tag recording a request's authenticated entity and corresponding servlet in OpenTracing. ([\#5856](https://github.com/matrix-org/synapse/issues/5856))
  - Add minimum OpenTracing for client servlets. ([\#5983](https://github.com/matrix-org/synapse/issues/5983))
  - Check at setup that OpenTracing is installed if it's enabled in the config. ([\#5985](https://github.com/matrix-org/synapse/issues/5985))
  - Trace replication send times. ([\#5986](https://github.com/matrix-org/synapse/issues/5986))
  - Include missing OpenTracing contexts in outbout replication requests. ([\#5982](https://github.com/matrix-org/synapse/issues/5982))
  - Fix sending of EDUs when OpenTracing is enabled with an empty whitelist. ([\#5984](https://github.com/matrix-org/synapse/issues/5984))
  - Fix invalid references to None while OpenTracing if the log context slips. ([\#5988](https://github.com/matrix-org/synapse/issues/5988), [\#5991](https://github.com/matrix-org/synapse/issues/5991))
  - OpenTracing for room and e2e keys. ([\#5855](https://github.com/matrix-org/synapse/issues/5855))
  - Add OpenTracing span over HTTP push processing. ([\#6003](https://github.com/matrix-org/synapse/issues/6003))
- Add an admin API to purge old rooms from the database. ([\#5845](https://github.com/matrix-org/synapse/issues/5845))
- Retry well-known lookups if we have recently seen a valid well-known record for the server. ([\#5850](https://github.com/matrix-org/synapse/issues/5850))
- Add support for filtered room-directory search requests over federation ([MSC2197](https://github.com/matrix-org/matrix-doc/pull/2197), in order to allow upcoming room directory query performance improvements. ([\#5859](https://github.com/matrix-org/synapse/issues/5859))
- Correctly retry all hosts returned from SRV when we fail to connect. ([\#5864](https://github.com/matrix-org/synapse/issues/5864))
- Add admin API endpoint for setting whether or not a user is a server administrator. ([\#5878](https://github.com/matrix-org/synapse/issues/5878))
- Enable cleaning up extremities with dummy events by default to prevent undue build up of forward extremities. ([\#5884](https://github.com/matrix-org/synapse/issues/5884))
- Add config option to sign remote key query responses with a separate key. ([\#5895](https://github.com/matrix-org/synapse/issues/5895))
- Add support for config templating. ([\#5900](https://github.com/matrix-org/synapse/issues/5900))
- Users with the type of "support" or "bot" are no longer required to consent. ([\#5902](https://github.com/matrix-org/synapse/issues/5902))
- Let synctl accept a directory of config files. ([\#5904](https://github.com/matrix-org/synapse/issues/5904))
- Increase max display name size to 256. ([\#5906](https://github.com/matrix-org/synapse/issues/5906))
- Add admin API endpoint for getting whether or not a user is a server administrator. ([\#5914](https://github.com/matrix-org/synapse/issues/5914))
- Redact events in the database that have been redacted for a week. ([\#5934](https://github.com/matrix-org/synapse/issues/5934))
- New prometheus metrics:
  - `synapse_federation_known_servers`: represents the total number of servers your server knows about (i.e. is in rooms with), including itself. Enable by setting `metrics_flags.known_servers` to True in the configuration.([\#5981](https://github.com/matrix-org/synapse/issues/5981))
  - `synapse_build_info`: exposes the Python version, OS version, and Synapse version of the running server. ([\#6005](https://github.com/matrix-org/synapse/issues/6005))
- Give appropriate exit codes when synctl fails. ([\#5992](https://github.com/matrix-org/synapse/issues/5992))
- Apply the federation blacklist to requests to identity servers. ([\#6000](https://github.com/matrix-org/synapse/issues/6000))
- Add `report_stats_endpoint` option to configure where stats are reported to, if enabled. Contributed by @Sorunome. ([\#6012](https://github.com/matrix-org/synapse/issues/6012))
- Add config option to increase ratelimits for room admins redacting messages. ([\#6015](https://github.com/matrix-org/synapse/issues/6015))
- Stop sending federation transactions to servers which have been down for a long time. ([\#6026](https://github.com/matrix-org/synapse/issues/6026))
- Make the process for mapping SAML2 users to matrix IDs more flexible. ([\#6037](https://github.com/matrix-org/synapse/issues/6037))
- Return a clearer error message when a timeout occurs when attempting to contact an identity server. ([\#6073](https://github.com/matrix-org/synapse/issues/6073))
- Prevent password reset's submit_token endpoint from accepting trailing slashes. ([\#6074](https://github.com/matrix-org/synapse/issues/6074))
- Return 403 on `/register/available` if registration has been disabled. ([\#6082](https://github.com/matrix-org/synapse/issues/6082))
- Explicitly log when a homeserver does not have the `trusted_key_servers` config field configured. ([\#6090](https://github.com/matrix-org/synapse/issues/6090))
- Add support for pruning old rows in `user_ips` table. ([\#6098](https://github.com/matrix-org/synapse/issues/6098))

Bugfixes
--------

- Don't create broken room when `power_level_content_override.users` does not contain `creator_id`. ([\#5633](https://github.com/matrix-org/synapse/issues/5633))
- Fix database index so that different backup versions can have the same sessions. ([\#5857](https://github.com/matrix-org/synapse/issues/5857))
- Fix Synapse looking for config options `password_reset_failure_template` and `password_reset_success_template`, when they are actually `password_reset_template_failure_html`, `password_reset_template_success_html`. ([\#5863](https://github.com/matrix-org/synapse/issues/5863))
- Fix stack overflow when recovering an appservice which had an outage. ([\#5885](https://github.com/matrix-org/synapse/issues/5885))
- Fix error message which referred to `public_base_url` instead of `public_baseurl`. Thanks to @aaronraimist for the fix! ([\#5909](https://github.com/matrix-org/synapse/issues/5909))
- Fix 404 for thumbnail download when `dynamic_thumbnails` is `false` and the thumbnail was dynamically generated. Fix reported by rkfg. ([\#5915](https://github.com/matrix-org/synapse/issues/5915))
- Fix a cache-invalidation bug for worker-based deployments. ([\#5920](https://github.com/matrix-org/synapse/issues/5920))
- Fix admin API for listing media in a room not being available with an external media repo. ([\#5966](https://github.com/matrix-org/synapse/issues/5966))
- Fix list media admin API always returning an error. ([\#5967](https://github.com/matrix-org/synapse/issues/5967))
- Fix room and user stats tracking. ([\#5971](https://github.com/matrix-org/synapse/issues/5971), [\#5998](https://github.com/matrix-org/synapse/issues/5998), [\#6029](https://github.com/matrix-org/synapse/issues/6029))
- Return a `M_MISSING_PARAM` if `sid` is not provided to `/account/3pid`. ([\#5995](https://github.com/matrix-org/synapse/issues/5995))
- `federation_certificate_verification_whitelist` now will not cause `TypeErrors` to be raised (a regression in 1.3). Additionally, it now supports internationalised domain names in their non-canonical representation. ([\#5996](https://github.com/matrix-org/synapse/issues/5996))
- Only count real users when checking for auto-creation of auto-join room. ([\#6004](https://github.com/matrix-org/synapse/issues/6004))
- Ensure support users can be registered even if MAU limit is reached. ([\#6020](https://github.com/matrix-org/synapse/issues/6020))
- Fix bug where login error was shown incorrectly on SSO fallback login. ([\#6024](https://github.com/matrix-org/synapse/issues/6024))
- Fix bug in calculating the federation retry backoff period. ([\#6025](https://github.com/matrix-org/synapse/issues/6025))
- Prevent exceptions being logged when extremity-cleanup events fail due to lack of user consent to the terms of service. ([\#6053](https://github.com/matrix-org/synapse/issues/6053))
- Remove POST method from password-reset `submit_token` endpoint until we implement `submit_url` functionality. ([\#6056](https://github.com/matrix-org/synapse/issues/6056))
- Fix logcontext spam on non-Linux platforms. ([\#6059](https://github.com/matrix-org/synapse/issues/6059))
- Ensure query parameters in email validation links are URL-encoded. ([\#6063](https://github.com/matrix-org/synapse/issues/6063))
- Fix a bug which caused SAML attribute maps to be overridden by defaults. ([\#6069](https://github.com/matrix-org/synapse/issues/6069))
- Fix the logged number of updated items for the `users_set_deactivated_flag` background update. ([\#6092](https://github.com/matrix-org/synapse/issues/6092))
- Add `sid` to `next_link` for email validation. ([\#6097](https://github.com/matrix-org/synapse/issues/6097))
- Threepid validity checks on msisdns should not be dependent on `threepid_behaviour_email`. ([\#6104](https://github.com/matrix-org/synapse/issues/6104))
- Ensure that servers which are not configured to support email address verification do not offer it in the registration flows. ([\#6107](https://github.com/matrix-org/synapse/issues/6107))

Updates to the Docker image
---------------------------

- Avoid changing `UID/GID` if they are already correct. ([\#5970](https://github.com/matrix-org/synapse/issues/5970))
- Provide `SYNAPSE_WORKER` envvar to specify python module. ([\#6058](https://github.com/matrix-org/synapse/issues/6058))

Improved Documentation
----------------------

- Convert documentation to markdown (from rst) ([\#5849](https://github.com/matrix-org/synapse/issues/5849))
- Update `INSTALL.md` to say that Python 2 is no longer supported. ([\#5953](https://github.com/matrix-org/synapse/issues/5953))
- Add developer documentation for using SAML2. ([\#6032](https://github.com/matrix-org/synapse/issues/6032))
- Add some notes on rolling back to v1.3.1. ([\#6049](https://github.com/matrix-org/synapse/issues/6049))
- Update the upgrade notes. ([\#6050](https://github.com/matrix-org/synapse/issues/6050))

Deprecations and Removals
-------------------------

- Remove shared-secret registration from `/_matrix/client/r0/register` endpoint. Contributed by Awesome Technologies Innovationslabor GmbH. ([\#5877](https://github.com/matrix-org/synapse/issues/5877))
- Deprecate the `trusted_third_party_id_servers` option. ([\#5875](https://github.com/matrix-org/synapse/issues/5875))

Internal Changes
----------------

- Lay the groundwork for structured logging output. ([\#5680](https://github.com/matrix-org/synapse/issues/5680))
- Retry well-known lookup before the cache expires, giving a grace period where the remote well-known can be down but we still use the old result. ([\#5844](https://github.com/matrix-org/synapse/issues/5844))
- Remove log line for debugging issue #5407. ([\#5860](https://github.com/matrix-org/synapse/issues/5860))
- Refactor the Appservice scheduler code. ([\#5886](https://github.com/matrix-org/synapse/issues/5886))
- Compatibility with v2 Identity Service APIs other than /lookup. ([\#5892](https://github.com/matrix-org/synapse/issues/5892), [\#6013](https://github.com/matrix-org/synapse/issues/6013))
- Stop populating some unused tables. ([\#5893](https://github.com/matrix-org/synapse/issues/5893), [\#6047](https://github.com/matrix-org/synapse/issues/6047))
- Add missing index on `users_in_public_rooms` to improve the performance of directory queries. ([\#5894](https://github.com/matrix-org/synapse/issues/5894))
- Improve the logging when we have an error when fetching signing keys. ([\#5896](https://github.com/matrix-org/synapse/issues/5896))
- Add support for database engine-specific schema deltas, based on file extension. ([\#5911](https://github.com/matrix-org/synapse/issues/5911))
- Update Buildkite pipeline to use plugins instead of buildkite-agent commands. ([\#5922](https://github.com/matrix-org/synapse/issues/5922))
- Add link in sample config to the logging config schema. ([\#5926](https://github.com/matrix-org/synapse/issues/5926))
- Remove unnecessary parentheses in return statements. ([\#5931](https://github.com/matrix-org/synapse/issues/5931))
- Remove unused `jenkins/prepare_sytest.sh` file. ([\#5938](https://github.com/matrix-org/synapse/issues/5938))
- Move Buildkite pipeline config to the pipelines repo. ([\#5943](https://github.com/matrix-org/synapse/issues/5943))
- Remove unnecessary return statements in the codebase which were the result of a regex run. ([\#5962](https://github.com/matrix-org/synapse/issues/5962))
- Remove left-over methods from v1 registration API. ([\#5963](https://github.com/matrix-org/synapse/issues/5963))
- Cleanup event auth type initialisation. ([\#5975](https://github.com/matrix-org/synapse/issues/5975))
- Clean up dependency checking at setup. ([\#5989](https://github.com/matrix-org/synapse/issues/5989))
- Update OpenTracing docs to use the unified `trace` method. ([\#5776](https://github.com/matrix-org/synapse/issues/5776))
- Small refactor of function arguments and docstrings in` RoomMemberHandler`. ([\#6009](https://github.com/matrix-org/synapse/issues/6009))
- Remove unused `origin` argument on `FederationHandler.add_display_name_to_third_party_invite`. ([\#6010](https://github.com/matrix-org/synapse/issues/6010))
- Add a `failure_ts` column to the `destinations` database table. ([\#6016](https://github.com/matrix-org/synapse/issues/6016), [\#6072](https://github.com/matrix-org/synapse/issues/6072))
- Clean up some code in the retry logic. ([\#6017](https://github.com/matrix-org/synapse/issues/6017))
- Fix the structured logging tests stomping on the global log configuration for subsequent tests. ([\#6023](https://github.com/matrix-org/synapse/issues/6023))
- Clean up the sample config for SAML authentication. ([\#6064](https://github.com/matrix-org/synapse/issues/6064))
- Change mailer logging to reflect Synapse doesn't just do chat notifications by email now. ([\#6075](https://github.com/matrix-org/synapse/issues/6075))
- Move last-seen info into devices table. ([\#6089](https://github.com/matrix-org/synapse/issues/6089))
- Remove unused parameter to `get_user_id_by_threepid`. ([\#6099](https://github.com/matrix-org/synapse/issues/6099))
- Refactor the user-interactive auth handling. ([\#6105](https://github.com/matrix-org/synapse/issues/6105))
- Refactor code for calculating registration flows. ([\#6106](https://github.com/matrix-org/synapse/issues/6106))

1 files changed, 249 insertions, 0 deletions

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