Compare commits

..
Author SHA1 Message Date
Jonas Zohren d8b3003273 docs: WIP: Adjust suggested value ranges to Timo's suggestion and when to change them
Signed-off-by: Jonas Zohren <git-pbkyr@jzohren.de>
2021-09-15 12:45:13 +02:00
Jonas Zohren edd690c583 docs: WIP: Added some suggested range for numeric values
Signed-off-by: Jonas Zohren <git-pbkyr@jzohren.de>
2021-09-15 12:38:17 +02:00
Jonas Zohren c2068a434e docs: WIP: More config options
Signed-off-by: Jonas Zohren <git-pbkyr@jzohren.de>
2021-09-15 12:21:43 +02:00
Jonas Zohren b981d3d039 Move some documentation to docs/ and start overhaul 2021-09-15 11:16:19 +02:00
Jonas Zohren 80a2711f45 Move some documentation to docs/ and start overhaul 2021-09-15 11:16:19 +02:00
Jonas Zohren 2f4885c2a1 Remove the "register an account with element" test
Broke due to a timeout and Timo does not like broken tests.
Less testing means less failing tests.

Also, hopefully sytest is less broken now.
2021-09-15 11:16:18 +02:00
Timo Kösters 3fec75a7b6 improvement: more efficient sqlite 2021-09-15 11:16:18 +02:00
Timo Kösters 5094f8e7f1 improvement: deduplicate watchers 2021-09-15 11:16:18 +02:00
Jonas Platte cd19f02d3b Clean up (mostly automated with cargo clippy --fix) 2021-09-15 11:16:18 +02:00
Jonas Platte dbfe16bdb2 Upgrade ruma 2021-09-15 11:16:18 +02:00
Jonas Platte 5bdeb971d8 Get rid of more unnecessary intermediate collections 2021-09-15 11:16:18 +02:00
Daniel Wiesenberg a38b23d56d Add mautrix-signal appservice instructions 2021-09-15 11:16:18 +02:00
Jonathan de Jong 25a81c480b fix sync not firing on new events in room 2021-09-15 11:16:16 +02:00
Rasmus Thomsen 1282405f1d fix(database): handle errors in config parsin or database creation
Showing the user a backtrace can be pretty confusing, so just printing
a nice error message makes errors easier to understand for end users.

fixes #121
2021-09-15 11:16:16 +02:00
Jonas Platte ca4b676b4c Remove unused dependencies 2021-09-15 11:16:16 +02:00
charludo 3db558b2f8 These lines get *generated* by certbot. Having them in the file before running certbot results in an apache2 error, and putting them in afterwards is not necessary, since certbot places them there on its own. 2021-09-15 11:16:16 +02:00
Daniel Wiesenberg 5ea10bbd8e Fix deprecated/removed Traefik label 2021-09-15 11:16:16 +02:00
Timo Kösters 31b2728551 fix: make sure old events don't sneek into the timeline 2021-09-15 11:16:16 +02:00
Jonas Zohren d48b96fda6 chore(CI): Adjust CI for master and next branch development model
- Build release builds for branches "master" and "next"
- Push docker images under different tags, depending on why the pipeline started
  - branch master: push to `latest`
  - branch next: push to `next`
  - tag: push to `$TAG_NAME`

Signed-off-by: Jonas Zohren <git-pbkyr@jzohren.de>
2021-09-15 11:16:16 +02:00
Timo Kösters fe83c80a65 docs: add thanks to 2021-09-15 11:16:16 +02:00
Jonas Zohren 98ce5f62d5 chore: Also run CI on git tags, not only new commits.
Signed-off-by: Jonas Zohren <git-pbkyr@jzohren.de>
2021-09-15 11:16:16 +02:00
Timo Kösters 870c178c8b fix: avoid panic 2021-09-15 11:16:16 +02:00
Timo Kösters 1b9f56e48c chore: update dependencies 2021-09-15 11:16:16 +02:00
Timo Kösters 3143d468c7 revert Arc<EventId> commit 2021-09-15 11:16:15 +02:00
Timo Kösters ac594f5502 docs: update readme and version 2021-09-15 11:16:15 +02:00
Jonas Platte 75701f75e4 Remove unneeded intermediary container collection 2021-09-15 11:16:09 +02:00
Timo Kösters 186722c51f fix: batch key fetching 2021-09-15 11:16:09 +02:00
Kurt Roeckx fc3ff86ad9 fixup! Get required keys in batch when joining a room 2021-09-15 11:16:09 +02:00
Kurt Roeckx a8d181e00e fixup! Get required keys in batch when joining a room 2021-09-15 11:16:09 +02:00
Kurt Roeckx 61adef8f2c Let our server signing be valid for 1 week
It was valid for 2 minutes, which can result in the server being
asked too much for something that never changes.
2021-09-15 11:16:09 +02:00
Kurt Roeckx 13c9f2fb50 Get required keys in batch when joining a room
We now ask the trusted server for all keys in 1 request, instead of
asking each server individual for it's own keys.
2021-09-15 11:16:09 +02:00
Daniel Wiesenberg 5d279159e3 Add Traefik setup, incl. step-by-step, to docker README. 2021-09-15 11:16:09 +02:00
Jonas Platte e14db60b6c Fix a bunch of clippy lints 2021-09-15 11:16:09 +02:00
Timo Kösters 4cba464faf improvement: make pdu cache capacity configurable 2021-09-15 11:16:09 +02:00
Timo Kösters 828611c0e4 improvement: stop prev event fetching if too many events fail 2021-09-15 11:16:09 +02:00
Timo Kösters e206b0ca8f docs: documentation for every endpoint 2021-09-15 11:16:09 +02:00
Daniel Wiesenberg 4c971d7695 Fix healthcheck.sh permissions and rearange ci dockerfile 2021-09-15 11:16:09 +02:00
Jonathan de Jong 8aed9b7711 registration default true 2021-09-15 11:16:09 +02:00
Daniel Wiesenberg 3f766d2a13 Use $CI_COMMIT_SHORT_SHA for GIT_REF
Using `$CI_COMMIT_REF_NAME` means we get `master` for every image build,
which is not very useful/informative. Using `$CI_COMMIT_SHORT_SHA`,
on the other hand, makes it possible to see exactly from which commit an
image was built.
2021-09-15 11:16:09 +02:00
Daniel Wiesenberg 6d6ce0f697 Change healthcheck in ci dockerfile 2021-09-15 11:16:09 +02:00
Daniel Wiesenberg 8b4389e0c2 Move docker healthcheck into dedicated script. 2021-09-15 11:16:09 +02:00
Jonathan de Jong fc29ad088c add tls 2021-09-15 11:16:09 +02:00
Timo Kösters b0aa4fc7f5 improvement: call /state_ids less often by using state res 2021-09-15 11:16:09 +02:00
Timo Kösters 73ae29c0de fix: inviting dendrite users 2021-09-15 11:16:09 +02:00
Timo Kösters 8c7ae60518 improvement: don't do state updates if the event was soft failed 2021-09-15 11:16:09 +02:00
Timo Kösters 4d6af6ee88 fix: make appservices more efficient 2021-09-15 11:16:09 +02:00
Timo Kösters 4ede0dc2d9 add warning if calculated event id != requested event id 2021-09-15 11:16:09 +02:00
Timo Kösters 4ce02e8ff6 fix: don't retry soft failed events 2021-09-15 11:16:09 +02:00
Devin Ragotzy 4ac40dec9a Use Arc<EventId> in place of most EventIds 2021-09-15 11:16:09 +02:00
Timo Kösters 51b96b3538 fix: server resolution 2021-09-15 11:16:09 +02:00
Timo Kösters 5b904fbef4 fix: server resolution with well-known files 2021-09-15 11:16:09 +02:00
Timo Kösters 6c0079f2c4 improvement: less IO for auth chains 2021-09-15 11:16:09 +02:00
Timo Kösters 3a588c4561 fix: improve key fetching 2021-09-15 11:16:09 +02:00
Timo Kösters f22ad5dfba chore: bump dependencies 2021-09-15 11:16:09 +02:00
Timo Kösters 2b50afcc0a improvement: persist cached auth chains in db 2021-09-15 11:16:09 +02:00
Timo Kösters 8216abc8c5 fix: better migration, force e2ee device key updates 2021-09-15 11:16:09 +02:00
Timo Kösters 981d054b0c fix: wrong soft fail check, too many events in /sync state response 2021-09-15 11:16:09 +02:00
Timo Kösters 72e2b643bb fix: e2ee over federation
to device events were not being sent
2021-09-15 11:16:09 +02:00
Timo Kösters ee4b08c185 improvement: more efficient auth chain cache 2021-09-15 11:16:09 +02:00
Timo Kösters 896427b67f improvement: better e2ee over fed, faster incoming event handling 2021-09-15 11:16:09 +02:00
Jonas Zohren 95989241db Only apply max. optimizations in CI builds.
The average german man has a life expectancy of 78.7 years, or 689884.2 hours.
Assuming that Timo is 20 years old, he has rougly 514564.2 hours left on planet earth.
Also assuming that cross release builds took him 25 minutes before,
but 2-2.5x of that with the current release compilation config he wasted roughly an hour waiting for it to complete.
If he continued to work on Conduit for 20 more years (or 175320 hours),
and makes a release compilation about once per day, this means 7305 hours or 304 days wasted waiting for the rust compiler.
By cutting that back down to the original settings, he get's 182 days of his life back.
That's about 0.63% of his remaining life.
182 joyful days he can spend with family and loved ones.
2021-09-15 11:16:09 +02:00
Timo Kösters 10da08e260 fix: room upgrades 2021-09-15 11:16:09 +02:00
Timo Kösters 6255406de0 fix: correct create event warnings 2021-09-15 11:16:09 +02:00
Kurt Roeckx 4c44a7adf5 Implement federation/v1/send_join 2021-09-15 11:16:09 +02:00
Tom Smeding 48fb8c5d4d Consistent and escaped response in get_pdu
1. The fallback text of the get_pdu admin room command response message
   now contains the same text as the formatted_body content (namely, the
   json instead of Debug-formatting of a serde type).
2. The formatted_body content of the get_pdu response is now
   html-escaped.
2021-09-15 11:16:09 +02:00
Timo Kösters d058f6522e fmt 2021-09-15 11:16:09 +02:00
Timo Kösters 8784df0d3a improvement: limit prev event fetching 2021-09-15 11:16:09 +02:00
Timo Kösters 524b4960cd fix: don't save empty tokens 2021-09-15 11:16:09 +02:00
Timo Kösters 0176d40347 improvement: faster incoming transaction handling 2021-09-15 11:16:09 +02:00
Timo Kösters b74b8c48ce improvement: better prev event fetching, perf improvements 2021-09-15 11:16:09 +02:00
Timo Kösters 8a4b7cf4bd fix: faster room joins 2021-09-15 11:16:09 +02:00
Timo Kösters 4a9483c07c fix: don't load endless prev events and fix room join bug 2021-09-15 11:16:09 +02:00
Timo Kösters 8f1a41dcd2 improvement: better sqlite 2021-09-15 11:16:09 +02:00
Timo Kösters 6f58af0b99 improvement: state info cache 2021-09-15 11:16:09 +02:00
Timo Kösters ac9dce2bdf fix: fetch event multiple times 2021-09-15 11:16:09 +02:00
Timo Kösters 197a111748 fix: also fetch prev events that are outliers already 2021-09-15 11:16:09 +02:00
Timo Kösters b9f138450c fix: fetch more than one prev event 2021-09-15 11:16:09 +02:00
Timo Kösters 1aecadf990 fix: don't use recursion for prev events 2021-09-15 11:16:09 +02:00
Timo Kösters 7514aa5d95 fmt 2021-09-15 11:16:09 +02:00
Timo Kösters c0c783fdec remove prev event fetch limit 2021-09-15 11:16:09 +02:00
Timo Kösters b02e4bf46c fix: delta calculation 2021-09-15 11:16:09 +02:00
Timo Kösters 87f5f17134 fix: room exists panic 2021-09-15 11:16:09 +02:00
Timo Kösters f1aea98384 finish implementing better state store 2021-09-15 11:16:09 +02:00
Timo Kösters caebb13a98 improvement: migrations, batch inserts 2021-09-15 11:16:09 +02:00
Timo Kösters 0b073c6534 WIP improvement: much better state storage 2021-09-15 11:16:09 +02:00
Timo Kösters f29a6d7945 fix: long prev event fetch times for huge rooms 2021-09-15 11:16:09 +02:00
Timo Kösters 8696e627b0 less warnings 2021-09-15 11:16:09 +02:00
Timo Kösters 201de77db4 improvement: cache for short event ids 2021-09-15 11:16:09 +02:00
Timo Kösters 33e84c74ac improvement: use u64s in auth chain cache 2021-09-15 11:16:09 +02:00
Timo Kösters 39d1f86050 improvement: smaller cache, better prev event fetching 2021-09-15 11:16:09 +02:00
Jonas Zohren afa71756c0 WIP: change layout of config options documentation 2021-09-07 11:36:28 +02:00
Jonas Zohren d0e06b49da Switch to mdbook
Signed-off-by: Jonas Zohren <git-pbkyr@jzohren.de>
2021-08-19 16:46:03 +02:00
Jonas Zohren 5547c0f579 Merge remote-tracking branch 'origin/docs-overhaul' into docs-overhaul
# Conflicts:
#	mkdocs.yml
2021-08-19 12:37:05 +02:00
Jonas Zohren 66bc3958ff Document configuration options in own file
Signed-off-by: Jonas Zohren <git-pbkyr@jzohren.de>
2021-08-19 12:36:29 +02:00
Jonas Zohren 7e48567d4b Move some documentation to docs/ and start overhaul 2021-08-19 11:39:30 +02:00
Jonas Zohren 10ee0a53f9 Move some documentation to docs/ and start overhaul 2021-07-23 16:56:16 +02:00
25 changed files with 471 additions and 132 deletions
+18 -4
View File
@@ -3,6 +3,7 @@ stages:
- build docker image
- test
- upload artifacts
- publish
variables:
GIT_SUBMODULE_STRATEGY: recursive
@@ -75,12 +76,12 @@ build:release:cargo:aarch64-unknown-linux-gnu:
extends: .build-cargo-shared-settings
variables:
TARGET: "aarch64-unknown-linux-gnu"
NEEDED_PACKAGES: "build-essential gcc-10-aarch64-linux-gnu g++-aarch64-linux-gnu libc6-dev-arm64-cross"
NEEDED_PACKAGES: "build-essential gcc-8-aarch64-linux-gnu g++-aarch64-linux-gnu libc6-dev-arm64-cross"
CARGO_TARGET_AARCH64_UNKNOWN_LINUX_GNU_LINKER: aarch64-linux-gnu-gcc
CC_aarch64_unknown_linux_gnu: aarch64-linux-gnu-gcc
CXX_aarch64_unknown_linux_gnu: aarch64-linux-gnu-g++
TARGET_CC: "/usr/bin/aarch64-linux-gnu-gcc-10"
TARGET_AR: "/usr/bin/aarch64-linux-gnu-gcc-ar-10"
TARGET_CC: "/usr/bin/aarch64-linux-gnu-gcc-8"
TARGET_AR: "/usr/bin/aarch64-linux-gnu-gcc-ar-8"
artifacts:
name: "conduit-aarch64-unknown-linux-gnu"
paths:
@@ -222,7 +223,6 @@ build:docker:next:
- >
/kaniko/executor
$KANIKO_CACHE_ARGS
--force
--context $CI_PROJECT_DIR
--build-arg CREATED=$(date -u +'%Y-%m-%dT%H:%M:%SZ')
--build-arg VERSION=$(grep -m1 -o '[0-9].[0-9].[0-9]' Cargo.toml)
@@ -370,3 +370,17 @@ publish:package:
- 'curl --header "JOB-TOKEN: $CI_JOB_TOKEN" --upload-file conduit-x86_64-unknown-linux-gnu.deb "${BASE_URL}/conduit-x86_64-unknown-linux-gnu.deb"'
pages:
stage: "publish"
image:
name: "squidfunk/mkdocs-material"
entrypoint: [""]
tags: ["docker"]
needs: []
rules:
- if: '$CI_COMMIT_BRANCH == $CI_DEFAULT_BRANCH'
script:
- "mkdocs build --site-dir public"
artifacts:
paths:
- "public"
-86
View File
@@ -1,86 +0,0 @@
# Setting up Appservices
## Getting help
If you run into any problems while setting up an Appservice, write an email to `timo@koesters.xyz`, ask us in `#conduit:matrix.org` or [open an issue on GitLab](https://gitlab.com/famedly/conduit/-/issues/new).
## Set up the appservice - general instructions
Follow whatever instructions are given by the appservice. This usually includes
downloading, changing its config (setting domain, homeserver url, port etc.)
and later starting it.
At some point the appservice guide should ask you to add a registration yaml
file to the homeserver. In Synapse you would do this by adding the path to the
homeserver.yaml, but in Conduit you can do this from within Matrix:
First, go into the #admins room of your homeserver. The first person that
registered on the homeserver automatically joins it. Then send a message into
the room like this:
@conduit:your.server.name: register_appservice
```
paste
the
contents
of
the
yaml
registration
here
```
You can confirm it worked by sending a message like this:
`@conduit:your.server.name: list_appservices`
The @conduit bot should answer with `Appservices (1): your-bridge`
Then you are done. Conduit will send messages to the appservices and the
appservice can send requests to the homeserver. You don't need to restart
Conduit, but if it doesn't work, restarting while the appservice is running
could help.
## Appservice-specific instructions
### Tested appservices
These appservices have been tested and work with Conduit without any extra steps:
- [matrix-appservice-discord](https://github.com/Half-Shot/matrix-appservice-discord)
- [mautrix-hangouts](https://github.com/mautrix/hangouts/)
- [mautrix-telegram](https://github.com/mautrix/telegram/)
### [mautrix-signal](https://github.com/mautrix/signal)
There are a few things you need to do, in order for the Signal bridge (at least
up to version `0.2.0`) to work. How you do this depends on whether you use
Docker or `virtualenv` to run it. In either case you need to modify
[portal.py](https://github.com/mautrix/signal/blob/master/mautrix_signal/portal.py).
Do this **before** following the bridge installation guide.
1. **Create a copy of `portal.py`**. Go to
[portal.py](https://github.com/mautrix/signal/blob/master/mautrix_signal/portal.py)
at [mautrix-signal](https://github.com/mautrix/signal) (make sure you change to
the correct commit/version of mautrix-signal you're using) and copy its
content. Create a new `portal.py` on your system and paste the content in.
2. **Patch the copy**. Exact line numbers may be slightly different, look nearby if they don't match:
- [Line 1020](https://github.com/mautrix/signal/blob/4ea831536f154aba6419d13292479eb383ea3308/mautrix_signal/portal.py#L1020)
```diff
--- levels.users[self.main_intent.mxid] = 9001 if is_initial else 100
+++ levels.users[self.main_intent.mxid] = 100 if is_initial else 100
```
- [Between lines 1041 and 1042](https://github.com/mautrix/signal/blob/4ea831536f154aba6419d13292479eb383ea3308/mautrix_signal/portal.py#L1041-L1042) add a new line:
```diff
"type": str(EventType.ROOM_POWER_LEVELS),
+++ "state_key": "",
"content": power_levels.serialize(),
```
3. **Deploy the patch**. This is different depending on how you have `mautrix-signal` deployed:
- [*If using virtualenv*] Copy your patched `portal.py` to `./lib/python3.7/site-packages/mautrix_signal/portal.py` (the exact version of Python may be different on your system).
- [*If using Docker*] Map the patched `portal.py` into the `mautrix-signal` container:
```yaml
volumes:
- ./your/path/on/host/portal.py:/usr/lib/python3.9/site-packages/mautrix_signal/portal.py
```
4. Now continue with the [bridge installation instructions ](https://docs.mau.fi/bridges/index.html) and the general bridge notes above.
-11
View File
@@ -1,11 +0,0 @@
Install docker:
```
$ sudo apt install docker
$ sudo usermod -aG docker $USER
$ exec sudo su -l $USER
$ sudo systemctl start docker
$ cargo install cross
$ cross build --release --target armv7-unknown-linux-musleabihf
```
The cross-compiled binary is at target/armv7-unknown-linux-musleabihf/release/conduit
+16
View File
@@ -0,0 +1,16 @@
[book]
title = "Conduit Docs"
author = "The Conduit contributors"
description = "Conduit is a simple, fast and reliable chat server for the Matrix protocol"
language = "en"
src = "docs"
[rust]
edition = "2018"
[build]
build-dir = "public"
create-missing = true
[output.html.search]
limit-results = 15
+1 -1
View File
@@ -39,7 +39,7 @@ trusted_servers = ["matrix.org"]
#log = "info,state_res=warn,rocket=off,_=off,sled=off"
#workers = 4 # default: cpu core count * 2
address = "127.0.0.1" # This makes sure Conduit can only be reached using the reverse proxy, if you are running conduit from docker, set this to 0.0.0.0
address = "127.0.0.1" # This makes sure Conduit can only be reached using the reverse proxy
proxy = "none" # more examples can be found at src/database/proxy.rs:6
+16
View File
@@ -0,0 +1,16 @@
# Summary
[Home](index.md)
- [Installation](installation.md)
- [Prerequisites](installation/prerequisites.md)
- [From Binaries](installation/manual.md)
- [From Packages](installation/packages.md)
- [With Docker](installation/docker.md)
- [Configuration](configuration.md)
- [Config options](configuration/configuration.md)
- [Appservices](configuration/appservices.md)
- [Contribute](contribute.md)
- [Basics](development/basics.md)
- [Cross compilation](development/cross-compilation.md)
- [Tests & CI](development/tests-ci.md)
+1
View File
@@ -0,0 +1 @@
# Configuration
+2
View File
@@ -0,0 +1,2 @@
# Configuring Conduit
+220
View File
@@ -0,0 +1,220 @@
# Configuring Conduit
Conduit can be configured via a config file (conventionally called Conduit.toml) or environment variables. If a config
file exists and environment variables are set, environment variables overwrite config options.
You absolutely need to set the environment variable `CONDUIT_CONFIG_FILE` to either point to a config file (
e.g. `CONDUIT_CONFIG_FILE=/etc/conduit/Conduit.toml`) or to an empty string (`CONDUIT_CONFIG_FILE=''`) if you want to
configure Conduit with just environment variables.
## Mandatory config options
Mandatory variables must be configured in order for Conduit to run properly.
### Server Name
- Config file key: `server_name`
- Envirnoment variable: `CONDUIT_SERVER_NAME`
- Default value: _None, you will need to choose your own._
The server_name is the name of this server. It is used as a suffix for user and room ids. Example: If you set it
to `conduit.rs`, your usernames will look like `@somebody:conduit.rs`.
The Conduit server needs to be reachable at https://your.server.name/ on port 443 (client-server) and 8448 (
server-server) OR you can create /.well-known files to redirect requests. See
the [Client-Server specs](https://matrix.org/docs/spec/client_server/latest#get-well-known-matrix-client) and
the [Server-Server specs](https://matrix.org/docs/spec/server_server/r0.1.4#get-well-known-matrix-server) for more
information.
### Database Path
- Config file key: `database_path`
- Envirnoment variable: `CONDUIT_DATABASE_PATH`
- Default value: _None, but many people like to use `/var/lib/conduit/`_.
A **directory** where Conduit stores its database and media files. This directory must exist, have enough free space and
be readable and writable by the user Conduit is running as.
What does _enough free space_ mean? It heavily on the amount of messages your Conduit server will see and the amount and
size of media files users on your Conduit server send. As a rule of thumb, you should have at least 10 GB of free space
left. You should be comfortable for quite some time with 50 GB.
### TCP Port
- Config file key: `port`
- Environment variable: `CONDUIT_PORT`
- Default value: _None, but many people like to use `6167`_.
The TCP port Conduit will listen on for connections. The port needs to be free (no other program is listeing on it).
Conduit does currently (2021-09) not offer HTTPS by itself. Only unencrypted HTTP requests will be accepted on this
port. Unless you know what you are doing, this port should not be exposed to the internet. Instead, use a reverse proxy
capable of doing TLS to offer your Conduit server to the internet via HTTPS. See [TODO] for example configurations.
## Optional configuration options
These config options come with defaults and don't need to be configured for Conduit to run. That said, you should still
check them to make sure that your Conduit server behaves like you want it to do.
### Maximum request size
- Config file key: `max_request_size`
- Environment variable: `CONDUIT_MAX_REQUEST_SIZE`
- Default value: `20_000_000` (~= 20 MB)
The maximum size in bytes for incoming requests to Conduit. You can use underscores to improve readability.
This will effectively limit the size for images, videos and other files users on your Conduit server can send.
### Allow Registration?
- Config file key: `allow_registration`
- Environment variable: `CONDUIT_ALLOW_REGISTRATION`
- Default value: `true`
- Possible values: `true`, `false`
It this is set to `false`, no new users can register accounts on your Conduit server. Already registered users will not
be affected from this setting and can continue to user your server.
The first user to ever register on your Conduit server will be considered the admin account and is automatically invited
into the admin room.
### Allow Encryption?
- Config file key: `allow_encryption`
- Environment variable: `CONDUIT_ALLOW_ENCRYPTION`
- Default value: `true`
- Possible values: `true`, `false`
If this is set to `false`, Conduit disables the ability for users to create encrypted chats. Existing encrypted chats
may continue to work.
### Allow federation?
- Config file key: `allow_federation`
- Environment variable: `CONDUIT_ALLOW_FEDERATION`
- Default value: `false`
- Possible values: `true`, `false`
Federation means that users from different Matrix servers can chat with each other. E.g. `@mathew:matrix.org` can chat
with `@timo:conduit.rs`.
If this option is set to `false`, users on your Conduit server can only talk with other users on your Conduit server.
Federation with other servers needs to happen over HTTPS, so make sure you have set up a reverse proxy.
### Jaeger Tracing
- Config file key: `allow_jaeger`
- Environment variable: `CONDUIT_ALLOW_JAEGER`
- Default value: `false`
- Possible values: `true`, `false`
Enable Jaeger to support monitoring and troubleshooting through Jaeger.
If you don't know what Jaeger is, you can safely leave this set to `false`.
### Trusted servers
- Config file key: `trusted_servers`
- Environment variable: `CONDUIT_TRUSTED_SERVERS`
- Default value: `[]`
- Possible values: JSON-Array of server domains, e.g. `["matrix.org"]` or `["matrix.org", "conduit.rs"]`.
Matrix servers have so-called "server keys", which authenticate messages from their users. Because your Conduit server
might not know the server keys from every server it encounters, it can ask a _trusted server_ for them. This speeds
things up for rooms with people from a lot of different servers.
You should only set this to include trustworthy servers. Most people consider a good default to be `["matrix.org"]`.
Only relevant if you have federation enabled.
### Limit amount of concurrent requests
- Config file key: `max_concurrent_requests`
- Environment variable: `CONDUIT_MAX_CONCURRENT_REQUESTS`
- Default value: `100`
- Suggested values: `1` - `1000` (u16)
How many requests Conduit can make at the same time. This affects federation with other Matrix servers, push
notifications and app_services.
// TODO Timo: When does it make sense to change this?
### Configure logging
- Config file key: `log`
- Environment variable: `CONDUIT_LOG`
- Default value: `info,state_res=warn,rocket=off,_=off,sled=off`
Configures which kind of messages Conduit logs.
> // TODO: Better and more thorough explanation
### Worker threads
- Config file key: `workers`
- Environment variable: `CONDUIT_WORKERS`
- Default value: cpu core count * 2
- Possible values: // TODO
> // TODO: Which thing exactly threads? What not?
### Listening address
- Config file key: `address`
- Environment variable: `CONDUIT_ADDRESS`
- Default value: `127.0.0.1`
- Possible values: Valid IP addresses.
Which IP address conduit is listening on. 127.0.0.1 means that Conduit can only be accessed from the same server or
through a reverse proxy on that server. If you want it to be accessible from any network interface (which you should
not, because other matrix servers should talk to your Conduit via a reverse proxy and not directly), you can set it
to `0.0.0.0`.
### Database cache capacity
- Config file key: `db_cache_capacity_mb`
- Environment variable: `CONDUIT_DB_CACHE_CAPACITY_MB`
- Default value: `200`
- Possible values: `true`, `false`
The total amount of memory (RAM) that the database cache will be able to use.
> // TODO: this needs clearification: In RAM or on disk and for what exactly?
### PDU cache capacity
- Config file key: `pdu_cache_capacity`
- Environment variable: `CONDUIT_PDU_CACHE_CAPACITY`
- Default value: `100_000`
- Suggested values: `1_000` - `1_000_000` (u32)
The total capacity (read: number of items) the pdu cache can hold in memory. Setting this to a lower number may slow
Conduit down, as it must fetch more data from the database. Increasing it will mean that Conduit will start to use more
memory as the cache slowly fills up.
### SQLite WAL clean interval
- Config file key: `sqlite_wal_clean_second_interval`
- Environment variable: `CONDUIT_SQLITE_WAL_CLEAN_SECOND_INTERVAL`
- Default value: `60` (every 60 seconds)
- Suggested values: `1` - `3600` (u32)
How often the WAL file should be cleaned up. The WAL file will be written to until cleaned up, after which it restarts
writing from the beginning.
The file's size will correspond to how long it could write to it in one go. (e.g. if conduit writes 100MB of data to the
database inbetween that period, the file will grow to 100MB). You can read more about that in
the [SQLite Docs](https://www.sqlite.org/draft/wal.html).
Reducing this down too much can offset the benefits of using a WAL at all. However, having this too high can result in a
large WAL file.
Only relevant when using SQLite as the database.
### Still undocumented config options
- `tracing_flame`
- `proxy`
- `jwt_secret`
+1
View File
@@ -0,0 +1 @@
# Contribute
+1
View File
@@ -0,0 +1 @@
# Basics
+13
View File
@@ -0,0 +1,13 @@
# Cross compilation
Install docker:
```bash
sudo apt install docker
sudo usermod -aG docker $USER
exec sudo su -l $USER
sudo systemctl start docker
cargo install cross
cross build --release --target armv7-unknown-linux-musleabihf
```
The cross-compiled binary is at target/armv7-unknown-linux-musleabihf/release/conduit
+1
View File
@@ -0,0 +1 @@
# Tests & CI
+52
View File
@@ -0,0 +1,52 @@
# Conduit
Conduit is a simple, fast and reliable chat server for the [Matrix] protocol written in [Rust].
-----
> Note: This project is work-in-progress. Do *not* rely on it yet.
## What is Matrix?
[Matrix] is an open network for secure and decentralized
communication. It allows you to chat with friends even if they are using
another servers and client. You can even use bridges to communicate with users
outside of Matrix, like a community on Discord or your family on Hangouts.
## Why Conduit?
Conduit is an open-source server implementation of the [Matrix
Specification] with a focus on easy setup and low
system requirements, making it very easy to set up.
Other server implementations try to be extremely scalable, which makes sense if
the goal is to support millions of users on a single instance, but makes
smaller deployments a lot more inefficient. Conduit tries to keep it simple but
takes full advantage of that, for example by using an in-memory database for
[huge performance gains](https://github.com/timokoesters/romeo-and-juliet-benchmark).
The future for Conduit in peer-to-peer Matrix (every client contains a server)
is also bright.
Conduit tries to be reliable by using the Rust programming language and paying
close attention to error handling to make sure that evil clients, misbehaving
servers or even a partially broken database will not cause the whole server to
stop working.
## Chat with us!
We have a room on Matrix: [#conduit:matrix.org](https://matrix.to/#/#conduit:matrix.org)
You can also contact us using:
- Matrix: [@timo:koesters.xyz](https://matrix.to/#/@timo:koesters.xyz)
- Email: [conduit@koesters.xyz](mailto:conduit@koesters.xyz)
## Donate
Liberapay: <https://liberapay.com/timokoesters/>\
Bitcoin: `bc1qnnykf986tw49ur7wx9rpw2tevpsztvar5x8w4n`
[Matrix]: https://matrix.org/
[Rust]: https://rust-lang.org
[Matrix Specification]: https://matrix.org/docs/spec
+1
View File
@@ -0,0 +1 @@
# Installation
View File
+2
View File
@@ -0,0 +1,2 @@
:::warning Test
:::
+12
View File
@@ -0,0 +1,12 @@
# Distribution packages
## Debian / Ubuntu
[@paul:luon.net](https://matrix.to/#/@paul:luon.net) plans to package Conduit for Debian as soon as it reaches 1.0.
Until it is available in the official repos, you can install the development version of it manually:
```bash
sudo apt-get install ca-certificates
wget --https-only -O /tmp/conduit.deb https://gitlab.com/famedly/conduit/-/jobs/artifacts/master/raw/conduit-x86_64-unknown-linux-gnu.deb?job=build:cargo-deb:x86_64-unknown-linux-gnu
sudo dpkg -i /tmp/conduit.deb
```
+31
View File
@@ -0,0 +1,31 @@
# Prerequisites for running Conduit
You'll need:
- A domain. Commonly cost about $10/year.
- A Linux server with a stable internet connection, at least 500 MB of RAM and some disk space for messages and
attachments. Commonly start at $5/month.
- Some basic knowledge about using a shell, SSH and configuring and protecting a server.
## A word of caution:
Don't underestimate the toll of administrating your own server.
Conduit can't protect your conversations if your server gets compromised or deleted.
Make sure that you got:
- Automatic security updates
- On Ubuntu/Debian: Set up unattended-upgrades
- On RHEL/CentOS: Have a look at yum-cron
- A firewall blocking all but the needed incoming ports
- ufw is an easy interface for the linux firewall
- Protection against automatic attacks
- fail2ban scans logs and bans IPs which try to brute force their way into your server.
- Disable ssh login for root and switch from password to key based authentication.
- Automated backups
- Most VPS hosting companies offer whole server backups for a small fee.
- Or run your own backup with something like borg.
- A way to get notified if your disk fills up.
- If you send too much cat videos to your friends, Conduit might at some point become unable to
store any important messages.
+40
View File
@@ -0,0 +1,40 @@
# About Matrix Homeservers
Matrix homeservers manage its users chats. Every Matrix username includes the homserver it belongs to:
`@alice:matrix.org` means that the `matrix.org` homeserver hosts a user called `@alice`.
Every time someone chats with Alice, the `matrix.org` homeserver stores these messages.
When `@alice:matrix.org` talks with `@adelaide:matrix.org`, that's easy. Both users use the same server.
But how can `@bob:vector.tld`, who uses the `vector.tld` homeserver, exchange messages with `@alice:matrix.org`?
This is where it get's a bit more complicated.
## Matrix Homeserver discovery
The Matrix specification specifies multiple ways how servers can discover and then talk to each other.
Let's look at the most common one:
### .well-known files
At first, the only information a server has about a user (e.g. `@bob:vector.tld`) is its homeserver name: `vector.tld`.
It then makes a HTTP GET request to `https://vector.tld/.well-known/matrix/server`.
In the ideal case, this file contains a content like this:
```json
{
"m.server": "matrix.vector.tld:443"
}
```
This translates to: The matrix homeserver software for users with a username ending on `vector.tld`
can be found at the address `matrix.vector.tld` at port 443 (which is the common port for HTTPS).
The homeserver on it's quest to find `@bob:vector.tld` now contacts `matrix.vector.tld:443` and is then
able to exchange chat messages with it.
### Why so complicated?
Organizations often don't want to run their Matrix server on the same machine that hosts their website,
but `@foo:matrix.evil.corp` usernames are ugly and everyone wants to be `@foo:evil.corp`.
To solve that problem, Matrix implements this extra step via a .well-known file or a DNS entry.
+20
View File
@@ -0,0 +1,20 @@
site_name: Conduit Docs
site_description: Conduit is a simple, fast and reliable chat server for the Matrix protocol
theme:
name: material
repo_url: https://gitlab.com/famedly/conduit
nav:
- Home: index.md
- Installation:
- Prerequisites: installation/prerequisites.md
- Distribution Packages: installation/packages.md
- Manual: installation/manual.md
- Docker: installation/docker.md
- Configuration:
- Conduit.toml: configuration/conduit.toml.md
- Appservices: configuration/appservices.md
- Development:
- Basics: development/basics.md
- Cross compilation: development/cross-compilation.md
- Tests & CI: development/tests-ci.md
+5 -4
View File
@@ -1092,13 +1092,14 @@ pub(crate) async fn invite_helper<'a>(
"Could not accept incoming PDU as timeline event.",
))?;
let servers = db
for server in db
.rooms
.room_servers(room_id)
.filter_map(|r| r.ok())
.filter(|server| &**server != db.globals.server_name());
db.sending.send_pdu(servers, &pdu_id)?;
.filter(|server| &**server != db.globals.server_name())
{
db.sending.send_pdu(&server, &pdu_id)?;
}
return Ok(());
}
+5 -4
View File
@@ -2104,12 +2104,13 @@ impl Rooms {
// where events in the current room state do not exist
self.set_room_state(room_id, statehashid)?;
let servers = self
for server in self
.room_servers(room_id)
.filter_map(|r| r.ok())
.filter(|server| &**server != db.globals.server_name());
db.sending.send_pdu(servers, &pdu_id)?;
.filter(|server| &**server != db.globals.server_name())
{
db.sending.send_pdu(&server, &pdu_id)?;
}
for appservice in db.appservice.all()? {
if self.appservice_in_room(room_id, &appservice, db)? {
+8 -18
View File
@@ -84,7 +84,7 @@ pub enum SendingEventType {
pub struct Sending {
/// The state for a given state hash.
pub(super) servername_educount: Arc<dyn Tree>, // EduCount: Count of last EDU sync
pub(super) servernameevent_data: Arc<dyn Tree>, // ServernameEvent = (+ / $)SenderKey / ServerName / UserId + PduId / Id (for edus), Data = EDU content
pub(super) servernameevent_data: Arc<dyn Tree>, // ServernamEvent = (+ / $)SenderKey / ServerName / UserId + PduId / Id (for edus), Data = EDU content
pub(super) servercurrentevent_data: Arc<dyn Tree>, // ServerCurrentEvents = (+ / $)ServerName / UserId + PduId / Id (for edus), Data = EDU content
pub(super) maximum_requests: Arc<Semaphore>,
pub sender: mpsc::UnboundedSender<(Vec<u8>, Vec<u8>)>,
@@ -423,23 +423,13 @@ impl Sending {
Ok(())
}
#[tracing::instrument(skip(self, servers, pdu_id))]
pub fn send_pdu<I: Iterator<Item = Box<ServerName>>>(
&self,
servers: I,
pdu_id: &[u8],
) -> Result<()> {
let mut batch = servers.map(|server| {
let mut key = server.as_bytes().to_vec();
key.push(0xff);
key.extend_from_slice(pdu_id);
self.sender.unbounded_send((key.clone(), vec![])).unwrap();
(key, Vec::new())
});
self.servernameevent_data.insert_batch(&mut batch)?;
#[tracing::instrument(skip(self, server, pdu_id))]
pub fn send_pdu(&self, server: &ServerName, pdu_id: &[u8]) -> Result<()> {
let mut key = server.as_bytes().to_vec();
key.push(0xff);
key.extend_from_slice(pdu_id);
self.servernameevent_data.insert(&key, &[])?;
self.sender.unbounded_send((key, vec![])).unwrap();
Ok(())
}
+5 -4
View File
@@ -2880,13 +2880,14 @@ async fn create_join_event(
db,
)?;
let servers = db
for server in db
.rooms
.room_servers(room_id)
.filter_map(|r| r.ok())
.filter(|server| &**server != db.globals.server_name());
db.sending.send_pdu(servers, &pdu_id)?;
.filter(|server| &**server != db.globals.server_name())
{
db.sending.send_pdu(&server, &pdu_id)?;
}
db.flush()?;