SVM Rollup Node

SVM rollup node which replays transactions. This repo uses a local Postgres rather than an actual DA layer. There are no execution layer or DA customizations in this repo.

Building

For how to build and test code, it is recommanded to see solana's succinct instructions.

For Apple chip build, there's a [note](./apple_ chip_build.md) that might help.

Design

The architecture of design as below:

There are two roles in our Layer 2: Execution Layer and Settlement Layer.

• Execution Layer contains:

  • Execution Node

    Handles all Layer 2 transactions, produce block, push block to DA.

  • Verification Node:

    Pull data from DA, reconstruct block and replay.

• Settlement Layer contains:

  • Full node:

    Check the challenge and push transaction data to DA.

  • Light node:

    Sync header data.

Progress

1. Execution Layer

graph TD
    A[Execution Node] --> |push block| B[DA]
    B --> |replay data| C[Verification Node]

Loading

For now, since Celestia is still unstable, we use PostgreSQL as a DA simulator, here is the execution flow:

Execution flow of Execution Layer

1.1 Execution Node

• The execution node produce blocks, we use accountsdb-plugin-postgres to save blocks into PostgreSQL database.

  Instructions:

  a. Build code with submdoule

  git clone https://github.com/Eclipse-Laboratories-Inc/settlement.git
  cd settlement
  cargo build --release
  

  b. Setup database

    For detailed documents of how to setup database, see here .

    Suppose we got a database named solana, a username solana with password 1234.

    Then we should create Schema Objects in our solana database. Our current directory is still solana-executor, so here is the command:

  psql -U solana -p 5432 -h localhost -d solana -f solana-accountsdb-plugin-postgres/scripts/create_schema.sql
  

    Let's explain the parameters in above command:

  -U -- username
  -p -- port of PostgreSQL server
  -h -- ip address pf PostgreSQL server
  -d -- database name
  -f -- the path of SQL script file we want to execute
  

  c. Configure plugin settings

   The plugin configure file is solana-accountsdb-plugin-postgres/scripts/geyser.json, we need change some settings in it:

  {
    "libpath": "../../target/release/libsolana_geyser_plugin_postgres.dylib",
    "host": "127.0.0.1",
    "user": "solana",
    "password": "1234",
    "dbname": "solana",
    "port": 5432,
    "threads": 20,
    "batch_size": 20,
    "panic_on_db_errors": true,
    "accounts_selector" : {
        "accounts" : ["*"]
    },
    "transaction_selector" : {
        "mentions" : ["*"]
    },
    "entry_selector" : true
  }

   The configuration details are:

  libpath -- Our `libsolana_geyser_plugin_postgres` lib, should be in `target/release/libsolana_geyser_plugin_postgres.dylib
  host -- PostgreSQL server ip address
  user -- Username of database
  password -- Paddword of database
  dbname -- Database name
  port -- Port of PostgreSQL server, in our case, 5432.
  

  d. Start execution node

   For now, we use Test Validator as our execution node, we start it with plugin configure file we just set.

  ./target/release/solana-test-validator --geyser-plugin-config ./solana-accountsdb-plugin-postgres/scripts/geyser.json

   Now our test validator start producing blocks, and all these data saved in PostgreSQL.

1.2 Verification Node

• Start verification node

  Prerequisites:

* Configuration file of geyser plugin.
* 'genesis.bin' directory path, we need to obtain 'genesis.bin' from execution node's ledger directory, and store it in verification node file system.

  Replay and verify shred from PG.

./target/release/shred-replay -c ./solana-accountsdb-plugin-postgres/scripts/geyser.json -l replay-ledger -g genesis_path

  The command means:

-c, --config <CONFIG>           Configuration file to use [default: config.json]
-g, --genesis <GENESIS_PATH>    Use GENESIS_PATH as genesis path [default: ledger]
-l, --ledger <DIR>              Use DIR as ledger location [default: ledger]

• Start verification node's RPC service

./target/release/solo_rpc -l replay-ledger

  -l means the verified ledger of verification node.

2. Settlement Layer

    TBD

Designs of under the hood

Core designs of Execution Layer

For Execution Node:

• Adding a entry_notifier related functions in geyser-plugin, which is used for writing entries into PG when validator actives.

For Verification Node:

• shred-replay-service querys entries from PG, converts them to shreds, inserts into blockstore.
• Use solana-ledger-tool's verify function, generating bank hash for whole blockstore.