feat(dkg): implement sync (#311)

* fix: correct sync proto

* feat: add read/write fixed size proto

* feat: initial implementation of sync

* fix: skip relay connection handler

* refactor: simplify sync client

* refactor: inline functions

* fix: don't need to self-manage pending dial

* fix: simplify code write/read proto

* feat: add events

* feat: implement example sync

* fix: clear server connected state

* fix: simplify code

* fix: sync example allow a peer is disconnected

* fix: clippy

* fix: retry connect when dial fail

* fix: Client cancel will stop outbound stream

* fix: remove decode and encode error

* fix: check stop_rx when return canceled

* fix: drop duplicated inbound sync

* fix: initial backoff on retry

* fix: don't retry on NegotiationFailed

* fix: check server error on await_all_Shutdown

* fix: add comments

* refactor: avoid double hash

* fix: fmt

* fix: address comments

* fix: set max msg size to 32MB for sync msg

* fix: update sync example

* fix: sync example now use pluto create cluster

* fix: bcast example now use pluto create cluster

* fix: update sync example to use the p2p_context in side node

* fix: change log level of relay event

* fix: remove PartialEq and Eq on error

* fix: address comments

* fix: clippy

* fix: duplicate key in cargo.toml

* fix: test

* fix: single lock for client state

* fix: finish should acquire the lock once

* fix: call server.clear_connected on ConnectionClosed

* fix: handler sleep more when the outbound is claimed

* fix: check all type of relay io error

* fix: add PeerRttObserved event

* fix: retry on NoAddresses also

* fix: make state.error one-shot, only retain the first error

* fix: handle NoAddresses retry properly, wait a time

* refactor: simplify retry logic

* fix: add more test for edge cases

* doc: add documentation for for sync package

* doc: add documentation for for sync package

* fix: compile error after merge

Author: iamquang95

crates/dkg/Cargo.toml

@@ -7,10 +7,10 @@ license.workspace = true
 publish.workspace = true
 
 [dependencies]
+bon.workspace = true
 prost.workspace = true
 prost-types.workspace = true
 thiserror.workspace = true
-bon.workspace = true
 libp2p.workspace = true
 futures.workspace = true
 tokio.workspace = true
@@ -51,7 +51,6 @@ pluto-cluster = { workspace = true, features = ["test-cluster"] }
 pluto-testutil.workspace = true
 pluto-tracing.workspace = true
 serde_json.workspace = true
-tokio-util.workspace = true
 tempfile.workspace = true
 wiremock.workspace = true
 

crates/dkg/examples/bcast.rs

@@ -1,63 +1,45 @@
-//! Relay-based example for the DKG reliable-broadcast protocol.
-//!
-//! This example follows the same high-level shape as the `bootnode` example:
-//! - load a local private key from a data directory
-//! - load cluster peers from `cluster-lock.json`
-//! - resolve relay URLs with `bootnode::new_relays`
-//! - create relay reservations and relay routing
-//! - run `bcast` over relay-mediated connectivity
+//! Example for the DKG reliable-broadcast protocol.
 //!
 //! To try it locally:
 //!
 //! ```text
-//! # Terminal 1: start a relay server
-//! cargo run -p pluto-relay-server --example relay_server
+//! # Data preparation: create a local 3-node cluster fixture
+//! cargo run -p pluto-cli -- create cluster \
+//!   --cluster-dir /tmp/pluto-bcast-demo \
+//!   --name bcast-demo \
+//!   --network holesky \
+//!   --nodes 3 \
+//!   --num-validators 1 \
+//!   --insecure-keys \
+//!   --fee-recipient-addresses 0x000000000000000000000000000000000000dead \
+//!   --withdrawal-addresses 0x000000000000000000000000000000000000dead
 //!
-//! # Terminals 2-4: run three node directories from the same cluster
+//! # Run in 3 terminals against the generated node directories
 //! cargo run -p pluto-dkg --example bcast -- \
-//!   --relays http://127.0.0.1:8888 \
-//!   --data-dir /path/to/node0
+//!   --relays https://0.relay.obol.tech,https://1.relay.obol.tech \
+//!   --data-dir /tmp/pluto-bcast-demo/node0
 //!
 //! cargo run -p pluto-dkg --example bcast -- \
-//!   --relays http://127.0.0.1:8888 \
-//!   --data-dir /path/to/node1
+//!   --relays https://0.relay.obol.tech,https://1.relay.obol.tech \
+//!   --data-dir /tmp/pluto-bcast-demo/node1
 //!
 //! cargo run -p pluto-dkg --example bcast -- \
-//!   --relays http://127.0.0.1:8888 \
-//!   --data-dir /path/to/node2
+//!   --relays https://0.relay.obol.tech,https://1.relay.obol.tech \
+//!   --data-dir /tmp/pluto-bcast-demo/node2
 //! ```
 //!
-//! Assumption:
-//! - the three data directories already exist
-//! - each one belongs to one node in the same cluster
-//!
-//! Required files in each data directory:
-//! - `charon-enr-private-key`
-//! - `cluster-lock.json`
+//! For stable local repros or CI-style runs, prefer a self-hosted relay
+//! instead of shared public relays.
 //!
-//! Expected flow:
-//! 1. Each node loads the same cluster peer order from the lock file.
-//! 2. Nodes resolve the configured relays and establish relay reservations.
-//! 3. The relay router dials known cluster peers through relay circuits.
-//! 4. Once all cluster peers are connected, each node sends one signed
-//!    `DemoTick`.
-//! 5. Receiver logs show inbound signature requests and final broadcasts.
-//!
-//! Success signals:
+//! What to expect:
+//! - all nodes must use the same `--relays` value
 //! - `Relay reservation accepted`
-//! - `Connection established` with `peer_type="CLUSTER"`
-//! - `Cluster connectivity update connected=2 expected=2`
 //! - `Sending broadcast`
 //! - `Received signature request`
 //! - `Received broadcast`
 //!
-//! Transient relay warnings can occur during startup and reconnects. The demo
-//! is healthy once all cluster peers are connected and the single broadcast is
-//! delivered.
-//!
 //! The demo uses a single fixed message ID because `bcast` is intended for
 //! one-shot protocol-step messages, not a changing heartbeat stream.
-#![allow(missing_docs)]
 
 use std::{
     collections::{HashMap, HashSet},
@@ -110,7 +92,8 @@ struct Args {
     relays: Vec<String>,
 
     /// Data directory containing `charon-enr-private-key` and
-    /// `cluster-lock.json`.
+    /// `cluster-lock.json`, typically one of the `nodeN/` directories produced
+    /// by `pluto create cluster`.
     #[arg(long)]
     data_dir: PathBuf,