WARNING: Nostrex is still alpha software, has bugs, and will have backwards incompatible database changes. Do not use it for a production system.
$ git clone git@github.com:RiverFinancial/nostrex.git
$ cd nostrex
$ docker-compose up -d
$ mix deps.get && mix ecto.migrate
$ iex -S mix phx.serverIf you're using asdf you will need to run asdf install after you're in the project root directory.
If you're not using the DB provided by docker compose you can create a local development configuration file called local.exs and configure the database values in there.
For example:
# config/local.exs
# local DB port
config :nostrex, Nostrex.Repo,
port: "5234",Nostrex is a Nostr relay written in Elixir using the Phoenix framework. It is designed to be highly scalable to handle the rapid adoption of the Nostr protocol. Nostrex has a number of qualities and features that make it a compelling relay implementation.
The Erlang OTP gives us the primitives necessary to build high-throughput, concurrent high-uptime message routing software. The OTP has been hardened and optimized over decades and a single beefy server should be able to handle millions of concurrent connections.
One of the major bottlenecks of scaling a relay are the database operations. Nostrex has already implemented a number of optimizations at the DB level, including partitioning by Event created_at timestamps to keep table sizes in check.
Nostrex takes full advantage of the Erlang OTP's ETS in-memory datastore as the global state for a fast-filtering algorithm that efficiently matches events to filters instead of brute-forcing and checking each new event against every registered filter. There is still a lot of room to improve this algo.
The Erlang OTP allowed Whatsapp to scale to billions of users with a small team of engineers. With the right optimizations I feel confident that a single beefy Nostr instance will be able to power millions of concurrent connections.
- Validate event signatures (waiting on Bitcoinex lib updates)
- Remove unused boilerplate code
- Add prometheus monitoring
- Implement NIP
- Cleanup messy tests and add proper socket integration tests
- Create better filter tests to uncover any remaining edge cases
- Load test with something like Tsung
- NIP 01 Basic protocol flow description
- NIP 02 Contact List and Petnames
- NIP 04 Encrypted Direct Message
- NIP 09 Event Deletion
- NIP 11 Relay Information Document
- NIP 12 Generic Tag Queries
- NIP 15 End of Stored Events Notice
- NIP 20 Command Results
- NIP 40 Expiration Timestamp