solana leader

solana leader library

use

• default crate surface is the core only.
• use LeaderEngine with your own slot source and schedule snapshot.
• call seed with the first live slot.
• call on_slot for each new slot.
• if a refresh is needed, fetch a new schedule snapshot and call replace_schedule.
• read the current leaders from the shared buffer.

features

• default features are empty
• core types are always available without extra features
• rpc adds solana_leader::rpc::ScheduleFetcher
• slot adds solana_leader::slot::SlotSubscriber
• slot-grpc adds grpc slot support and implies slot and rpc
• managed adds solana_leader::managed::{ManagedTrackerConfig, ManagedTracker} and implies rpc, slot, and slot-grpc

feature gates

• core only: cargo add solana-leader
• core with rpc fetch: cargo add solana-leader --features rpc
• core with slot subscription: cargo add solana-leader --features slot
• core with grpc slot subscription: cargo add solana-leader --features slot-grpc
• full managed tracker over websocket or grpc: cargo add solana-leader --features managed

cargo.toml

[dependencies]
solana-leader = "x.x.x"

[dependencies]
solana-leader = { version = "x.x.x", features = ["rpc"] }

[dependencies]
solana-leader = { version = "x.x.x", features = ["slot"] }

modules

• solana_leader core types and engine
• solana_leader::rpc cold-path schedule fetch
• solana_leader::slot slot sources and connection handling
• solana_leader::managed full tracker orchestration

api

• core: LeaderEngine, EngineEvent, LeaderBuffer, LeaderEntry, LeaderPubkey, ScheduleSnapshot
• rpc: ScheduleFetcher, FetcherError
• slot: SlotSubscriber, SlotSubscriberError, SlotSourceConfig
• slot grpc: GrpcCommitment
• managed: ManagedTrackerConfig, ManagedTracker

limits

• rpc supports http:// and https://
• websocket slot subscription supports ws:// and wss://
• tls uses rustls with native system roots and stays on the cold path
• managed uses synchronous i/o plus std threads for the cold path
• managed and the examples use spin-based polling on the fast path, so pin or isolate a core if you care about tail latency
• the low-latency hot path stays in LeaderEngine and LeaderBuffer

examples

• cargo run --example example --features managed -- [slot_url] [rpc_url] [--ahead N]
• cargo run --example example_tokio --features managed -- [slot_url] [rpc_url]
• cargo run --example slot_subscriber --features slot
• cargo run --example example_grpc --features slot-grpc -- [grpc_url] [rpc_url]