How jito-foundation/jito-solana works
Project type and purpose
• This is a full-stack, distributed blockchain node implementation: Jito’s fork of the Solana validator.
• It encompasses the core validator engine (networking, consensus, transaction pipeline, ledger storage, runtime), plus CLI tools, client libraries, programs, benchmarks, and integration with external services (Bigtable, metrics, etc.).
File structure
• Top-level crates capture the major logical areas:
– validator/cli/clients (e.g. “cli”, “rpc-client”, “tpu-client”, “banks-client”)
– networking (e.g. “gossip”, “turbine”, “streamer”, “quic-client”, “udp-client”)
– transaction pipeline & consensus (“core”, “runtime”, “banking-stage”, “consensus”, “poh”, “replay-stage”, “snapshot-packager-service”)
– storage (“ledger”, “accounts-db”, “blockstore”, “storage-bigtable”)
– runtime/program execution (“program-runtime”, “bpf_loader”, “program-test”, “runtime-plugin”)
– tooling, benchmarks, CI/CD, docs, and scripts.
• Patterns: stage-based pipeline in “core” (“fetch-stage” → “sigverify-stage” → “banking-stage” → “replay-stage” → “snapshot”), actor-style services, and CRDT gossip (CRDS) for cluster state.
README for additional insights
• Confirms purpose: “Jito’s fork of the Solana validator.”
• Describes build/test workflow, but leaves detailed architecture to code and docs/architecture.md.
Detail and essential elements
• Every stage in the transaction processing pipeline
• Interactions between consensus & pipeline & storage
• Don’t omit the RPC/API layer or external services
• Important patterns (CRDT, actor model, async)