17dfdb63d4
This patch series refactors how we handle allocations in the relay to make it easier to forward a failure to the `Server`. Each allocation runs in a separate task (to allow for parallelization). If the allocation fails, this channel is automatically closed. Previously, this would erroneously trigger a `debug_assert!`. Now, we invoke a callback on `Server` to allow it to clean up its internal resources for the allocation. At the same time, we simplify the buffering around data that is destined for a certain allocation. Instead of having an additional buffer in the event-loop, we increase the channel size to 10. Any exceeding items will be dropped to avoid memory growth. This means that the `Server` is never blocked on a slow allocation. Given that we are running on top of an unreliable protocol anyway, I'd say this is fine.
relay
This crate houses a minimalistic STUN & TURN server.
Features
We aim to support the following feature set:
- STUN binding requests
- TURN allocate requests
- TURN refresh requests
- TURN channel bind requests
- TURN channel data requests
Relaying of data through other means such as DATA frames is not supported.
Building
You can build the server using: cargo build --release --bin relay
Running
For an up-to-date documentation on the available configurations options and a detailed help text, run cargo run --bin relay -- --help.
All command-line options can be overridden using environment variables.
Those variables are listed in the --help output at the bottom of each command.
The relay listens on port 3478.
This is the standard port for STUN/TURN and not configurable.
Additionally, the relay needs to have access to the port range 49152 - 65535 for the allocations.
Portal connection
When given a portal endpoint, the relay will connect to it and wait for an init message before commencing relay operations.
Design
The relay is designed in a sans-IO fashion, meaning the core components do not cause side effects but operate as pure, synchronous state machines. They take in data and emit commands: wake me at this point in time, send these bytes to this peer, etc.
This allows us to very easily unit-test all kinds of scenarios because all inputs are simple values.
The main server runs in a single task and spawns one additional task for each allocation. Incoming data that needs to be relayed is forwarded to the main task where it gets authenticated and relayed on success.