libp2p-rs kad使用及调试方法
前言
libp2p-rs v0.2.0已经支持Kad-DHT，支持节点通过DHT网络发现节点和内容。
本文主要展示如何在libp2p-rs使用DHT，以及罗列出swarm和kad的部分API。
使用kad
step1: 创建swarm
let sec = secio::Config::new(keys.clone());
let mux = yamux::Config::new();
let tu = TransportUpgrade::new(TcpConfig::default(), mux, sec);
let mut swarm = Swarm::new(keys.public())
.with_transport(Box::new(tu))
.with_identify(IdentifyConfig::new(false));
swarm.listen_on(vec![listen_addr]).expect("listen on");
let mut swarm_control = swarm.control();
step2: 创建kad
let store = MemoryStore::new(swarm.local_peer_id().clone());
let kad = Kademlia::new(swarm.local_peer_id().clone(), store);
let kad_handler = kad.handler();
let mut kad_ctrl = kad.control();
step3: 注册kad
将swarm升级，使其具备routing功能。
// register handler to swarm
swarm = swarm.with_protocol(Box::new(kad_handler)).with_routing(Box::new(kad_ctrl.clone()));
step4: 启动kad和swarm
kad.start(swarm_control.clone());
swarm.start();
step5: 使用kad
先将bootstrap节点添加到peerstore和路由表，再启动bootstrap。
kad_control.add_node(bootstrap_peer, vec![bootstrap_addr]).await;
kad_control.bootstrap().await;
step6: 启用cli
集成cli，可调试swarm和kad。
let mut app = App::new("xCLI").version("v0.1").author("[email protected]");
app.add_subcommand_with_userdata(swarm_cli_commands(), Box::new(swarm_control.clone()));
app.add_subcommand_with_userdata(dht_cli_commands(), Box::new(kad_control.clone()));
app.run();
cli调试
以下只列出部分常用的调试命令，更多玩法请亲自下场解锁。
swarm connection
connection命令可以用于获取当前所有连接及其子流的信息，也可以获取和某个peer的连接信息。
# s con
CID DIR Remote-Peer-Id I/O Remote-Multiaddr
231 In Qmf3ZX3yHnmzaXFGH5G149HyrAeKRFantVnZ9gZdnuPv1U 2/0 /ip4/114.227.83.230/tcp/24792
(231 Sid(7) In /ipfs/kad/1.0.0)
(231 Sid(9) In /ipfs/kad/1.0.0)
161 In QmZC9dZPyJWXSB2Ao2ChGJMjfuFiT7TyKdsGsFEKVSqnnf 0/0 /ip4/212.102.37.201/tcp/4001
2185 In QmTmnqSEarcSLJxhehJRKX64pxSkeKn7jS2fDEZFjjt9Bn 1/0 /ip4/114.226.44.86/tcp/3109
(2185 Sid(7) In /ipfs/kad/1.0.0)
2069 In 12D3KooWPtfLkqAVMPP6FNufHvqxPYe55XuAdEZWUn2cPbLLAwuT 1/0 /ip4/111.16.39.80/tcp/17881
(2069 Sid(1) In /ipfs/kad/1.0.0)
2349 In QmVcXP4bnoCJkUUJinduuM68n5jfSjDj6sKaTjRhHoecpt 1/0 /ip4/83.248.150.24/tcp/42761
(2349 Sid(3) In /ipfs/kad/1.0.0)
492 In 12D3KooWKBkFNUCvyP5PbV2mAhrzcvPi4EPL4vD3CjieGGU9ZcQr 1/0 /ip4/203.145.95.60/tcp/64417
(492 Sid(1) In /ipfs/kad/1.0.0)
18 In QmP6waLA8S6M8WPoQ5tWPE6xpgtsJ44LGQcq7vDTUAmyob 1/0 /ip4/188.127.190.220/tcp/4001
(18 Sid(7) In /ipfs/kad/1.0.0)
# s con Qmf3ZX3yHnmzaXFGH5G149HyrAeKRFantVnZ9gZdnuPv1U
CID DIR Remote-Peer-Id I/O Remote-Multiaddr
231 In Qmf3ZX3yHnmzaXFGH5G149HyrAeKRFantVnZ9gZdnuPv1U 2/0 /ip4/114.227.83.230/tcp/24792
(231 Sid(7) In /ipfs/kad/1.0.0)
(231 Sid(9) In /ipfs/kad/1.0.0)
dht states
states用于统计迭代查询运行状况，也可以观察当前节点接收到的Kad请求的次数。
# d st
Total refreshes : 1
Successful queries : 4
Timeout queries : 0
Query details : QueryStats { requests: 59, success: 41, failure: 10, duration: 18.198653932s }
Kad rx messages : MessageStats { ping: 0, find_node: 17216, get_provider: 559, add_provider: 3667, get_value: 1, put_value: 27 }
dht dump
dump命令用于dump出路由表的信息，使用verbose可以打印详细信息。
# d dp
Index Entries Active
244 1 1
246 1 1
247 2 2
248 7 7
249 10 8
250 20 20
251 20 20
252 20 20
253 20 20
254 20 19
255 20 20
# d dp 1
Index Entries Active
244 1 1
Qme9PR5oDcSSGoS2He53RqaML4vinDD5CNgxxmV2qPefFP Conn(false) Some(52292.68894773s) Addrs([])
246 1 1
QmboRZYso6VdQ5yfXe1DAj9u8EqouZGUsf2inoqYDtzdf8 Conn(true) Some(4330.367016609s) Addrs([])
247 2 2
QmZsbivLpaVpWQ4Mum2nzbEcoXbH7QbftRkQCmmJiTqcUp Conn(false) Some(77688.875853187s) Addrs([])
QmZaCQ6anyaPuebhLeomzpyKRAY6GnNS5NCU8h7kSjwFKN Conn(false) Some(43775.12096365s) Addrs(["/ip4/127.0.0.1/tcp/4001", "/ip4/138.68.29.104/tcp/4001", "/ip4/10.46.0.6/tcp/4001", "/ip4/10.138.16.85/tcp/4001", "/ip6/::1/tcp/4001"])
API介绍
目前swarm和kad的API已经比较完善，以下只列出部分常用的API，想了解更多API请直接阅读源码。
swarm
peerstore
/// Gets the public key by peer_id.
pub fn get_key(&self, peer_id: &PeerId) -> Option<PublicKey>
/// Gets all multiaddr of a peer.
pub fn get_addrs(&self, peer_id: &PeerId) -> Option<Vec<Multiaddr>>
/// Adds a address to address_book by peer_id, if exists, update rtt.
pub fn add_addr(&self, peer_id: &PeerId, addr: Multiaddr, ttl: Duration)
/// Adds many new addresses if they're not already in the peer store.
pub fn add_addrs(&self, peer_id: &PeerId, addrs: Vec<Multiaddr>, ttl: Duration)
/// Clears all multiaddr of a peer from the peer store.
pub fn clear_addrs(&self, peer_id: &PeerId)
connection
/// Make a new connection towards the remote peer with addresses specified.
pub async fn connect_with_addrs(&mut self, peer_id: PeerId, addrs: Vec<Multiaddr>) -> Result<()>
/// Make a new connection towards the remote peer.
///
/// It will lookup the peer store for address of the peer, otherwise
/// initiate the routing interface for querying the addresses, if routing
/// is available.
pub async fn new_connection(&mut self, peer_id: PeerId) -> Result<()>
/// Make a new connection towards the remote peer, without using routing(Kad-DHT).
pub async fn new_connection_no_routing(&mut self, peer_id: PeerId) -> Result<()>
/// Close connection towards the remote peer.
pub async fn disconnect(&mut self, peer_id: PeerId) -> Result<()>
stream
/// Open a new outbound stream towards the remote peer.
///
/// It will lookup the peer store for address of the peer,
/// otherwise initiate the routing interface for address querying,
/// when routing is enabled. In the end, it will open an outgoing
/// sub-stream when the connection is eventually established.
pub async fn new_stream(&mut self, peer_id: PeerId, pids: Vec<ProtocolId>) -> Result<Substream>
/// Open a new outbound stream towards the remote peer, without routing.
pub async fn new_stream_no_routing(&mut self, peer_id: PeerId, pids: Vec<ProtocolId>) -> Result<Substream>
/// Open a new outbound stream towards the remote peer, without routing.
pub async fn new_stream_no_routing(&mut self, peer_id: PeerId, pids: Vec<ProtocolId>) -> Result<Substream>
kad
/// Add a node and its listening addresses to KBuckets.
pub async fn add_node(&mut self, peer_id: PeerId, addrs: Vec<Multiaddr>)
/// Add a node and its listening addresses to KBuckets.
pub async fn remove_node(&mut self, peer_id: PeerId)
/// Initiate bootstrapping.
///
/// In general it should be done only once upon Kad startup.
pub async fn bootstrap(&mut self)
/// Lookup the closer peers with the given key.
pub async fn lookup(&mut self, key: record::Key) -> Result<Vec<KadPeer>>
/// Lookup the given peer.
pub async fn find_peer(&mut self, peer_id: &PeerId) -> Result<KadPeer>
/// Put value in local and other peers which closest to the given key.
pub async fn put_value(&mut self, key: Vec<u8>, value: Vec<u8>) -> Result<()>
/// Get value from local and other peers which closest to the given key.
pub async fn get_value(&mut self, key: Vec<u8>) -> Result<Vec<u8>>
/// Announce to peers which closer to the given key that self provide content.
pub async fn provide(&mut self, key: Vec<u8>) -> Result<()>
/// Find peers who provide content.
pub async fn find_providers(&mut self, key: Vec<u8>, count: usize) -> Option<Vec<KadPeer>>
总结
在启动kad和swarm后，就可以通过句柄（controller），调用API。当前
swarm的API已经比较完善，目前已经可以支持Kad-DHT这种比较复杂的协议，这也为新协议的添加创造了条件。