rpcpool / yellowstone-vixen

Solana program parsing toolkit
MIT License
54 stars 13 forks source link

Yellowstone Vixen

Yellowstone Vixen allows dApp developers to build program-aware change event streams for Solana. It provides the building blocks, such as a runtime, parser specification, and handler specification, to create custom indexes for specific programs, accounts, and transactions. Vixen enables developers to assemble program parsers to process real-time change events from Dragon's Mouth, converting them into program-aware structures. These structures can then be stored in a database or used in other data pipelines.

Table of Contents

  1. Objectives
  2. Requirements
  3. Example
  4. Dragon's Mouth
  5. Developers

Objectives

  1. Cost Efficiency: Utilizing Dragon's Mouth, multiple Vixen instances can share a single geyser stream. With various filter options, storage costs focus on what's essential for the dApp.
  2. Operational Simplicity: Vixen requires minimal configuration and dependency on other systems, making it easy to operate.
  3. Observability: Operators can monitor the health of their installation, gaining insights into lag, throughput, and error rates.
  4. Composability: Program parsers are developed as separate modules (cargo crates), enabling programs to include other parsers needed to deserialize cross-program invocations (CPI).

Requirements

  1. Parser: A module responsible for transforming raw Solana data into a program-specific format.
  2. Handler: A module that processes the parsed data, performing tasks such as logging, storing in a database, or triggering other actions.
  3. HandlerManager: Manages multiple handlers for different types of data (e.g., accounts, transactions).
  4. Configuration: A TOML file specifying the settings and parameters for Vixen.

Example

This example demonstrates how a developer can implement a generic parsing pipeline with Vixen. The examples are located in the /examples directory.

To run the example, follow these steps:

  1. Navigate to the prometheus example directory:

    cd examples/prometheus
  2. Execute the following command to run the example with the specified configuration:

    RUST_LOG=info cargo run -- --config "$(pwd)/../../Vixen.toml"

An example configuration file is available at Vixen.example.toml. Copy this file and modify it as needed to create your Vixen.toml.

Explanation

In this example, you need to implement specific components to create a functional parsing pipeline:

pub struct CustomParser;

impl vixen_core::Parser for CustomParser {
    type Input = AccountUpdate;
    type Output = CustomParsedData; // Replace with the actual data type

    fn prefilter(&self) -> Prefilter {
        Prefilter::builder()
            .account_owners([CUSTOM_PROGRAM_ID]) // Replace with the actual program ID
            .build()
            .unwrap()
    }

    async fn parse(&self, acct: &AccountUpdate) -> ParseResult<Self::Output> {
        // Implement parsing logic here
        // Example: Ok(CustomParsedData::from(acct))
        unimplemented!()
    }
}
pub struct CustomHandler;

impl<H: std::fmt::Debug + Sync> vixen::Handler<H> for CustomHandler {
    async fn handle(&self, value: &H) -> vixen::HandlerResult<()> {
        // Implement handling logic here
        // Example: tracing::info!(?value);
        unimplemented!()
    }
}
use std::path::PathBuf;

use clap::Parser as _;
use tracing_subscriber::{layer::SubscriberExt, util::SubscriberInitExt};
use yellowstone_vixen::{self as vixen, Pipeline};
use yellowstone_vixen_parser::{
    token_extension_program::{
        AccountParser as TokenExtensionProgramAccParser,
        InstructionParser as TokenExtensionProgramIxParser,
    },
    token_program::{
        AccountParser as TokenProgramAccParser, InstructionParser as TokenProgramIxParser,
    },
};

fn main() {
    tracing_subscriber::registry()
        .with(tracing_subscriber::EnvFilter::from_default_env())
        .with(tracing_subscriber::fmt::layer())
        .init();

    let Opts { config } = Opts::parse();
    let config = std::fs::read_to_string(config).expect("Error reading config file");
    let config = toml::from_str(&config).expect("Error parsing config");

    vixen::Runtime::builder()
        .account(Pipeline::new(TokenExtensionProgramAccParser, [Handler]))
        .account(Pipeline::new(TokenProgramAccParser, [Handler]))
        .instruction(Pipeline::new(TokenExtensionProgramIxParser, [Handler]))
        .instruction(Pipeline::new(TokenProgramIxParser, [Handler]))
        .build(config)
        .run();
}

Yellowstone Vixen Mock

This crate includes a mock feature designed for testing parsers. It is intended solely for testing purposes. For more details, refer to the mock documentation.

Metrics Support

Prometheus

Vixen also supports Prometheus for metrics. To enable Prometheus, set the prometheus feature in the Cargo.toml file:

[dependencies]
yellowstone-vixen = { version = "0.0.0", features = ["prometheus"] }
fn main() {
    tracing_subscriber::registry()
        .with(tracing_subscriber::EnvFilter::from_default_env())
        .with(tracing_subscriber::fmt::layer())
        .init();

    let Opts { config } = Opts::parse();
    let config = std::fs::read_to_string(config).expect("Error reading config file");
    let config = toml::from_str(&config).expect("Error parsing config");

    vixen::Runtime::builder()
        .account(Pipeline::new(TokenExtensionProgramAccParser, [Handler]))
        .account(Pipeline::new(TokenProgramAccParser, [Handler]))
        .instruction(Pipeline::new(TokenExtensionProgramIxParser, [Handler]))
        .instruction(Pipeline::new(TokenProgramIxParser, [Handler]))
        .metrics(vixen::metrics::Prometheus)
        .build(config)
        .run();
}

Prometheus metrics are served on the /metrics endpoint. To collect metrics, we have setup a prometheus server as a docker container. You can access the metrics at http://localhost:9090 after running the prometheus server using docker-compose.

Docker Setup for Metrics

To run prometheus, you need to have docker and docker-compose installed on your machine. To start the services, run the following command:

sudo docker-compose up

Dragon's Mouth

Dragon's Mouth can be self-hosted as a Geyser plugin or used via a commercial vendor. For more details, refer to the Yellowstone Dragon's Mouth documentation and Yellowstone repository.

Developers

This project is developed by ABK Labs and Triton One.