Files
pezkuwi-subxt/polkadot/node/network/bridge/src/network.rs
T
Aaro Altonen 80616f6d03 Integrate litep2p into Polkadot SDK (#2944)
[litep2p](https://github.com/altonen/litep2p) is a libp2p-compatible P2P
networking library. It supports all of the features of `rust-libp2p`
that are currently being utilized by Polkadot SDK.

Compared to `rust-libp2p`, `litep2p` has a quite different architecture
which is why the new `litep2p` network backend is only able to use a
little of the existing code in `sc-network`. The design has been mainly
influenced by how we'd wish to structure our networking-related code in
Polkadot SDK: independent higher-levels protocols directly communicating
with the network over links that support bidirectional backpressure. A
good example would be `NotificationHandle`/`RequestResponseHandle`
abstractions which allow, e.g., `SyncingEngine` to directly communicate
with peers to announce/request blocks.

I've tried running `polkadot --network-backend litep2p` with a few
different peer configurations and there is a noticeable reduction in
networking CPU usage. For high load (`--out-peers 200`), networking CPU
usage goes down from ~110% to ~30% (80 pp) and for normal load
(`--out-peers 40`), the usage goes down from ~55% to ~18% (37 pp).

These should not be taken as final numbers because:

a) there are still some low-hanging optimization fruits, such as
enabling [receive window
auto-tuning](https://github.com/libp2p/rust-yamux/pull/176), integrating
`Peerset` more closely with `litep2p` or improving memory usage of the
WebSocket transport
b) fixing bugs/instabilities that incorrectly cause `litep2p` to do less
work will increase the networking CPU usage
c) verification in a more diverse set of tests/conditions is needed

Nevertheless, these numbers should give an early estimate for CPU usage
of the new networking backend.

This PR consists of three separate changes:
* introduce a generic `PeerId` (wrapper around `Multihash`) so that we
don't have use `NetworkService::PeerId` in every part of the code that
uses a `PeerId`
* introduce `NetworkBackend` trait, implement it for the libp2p network
stack and make Polkadot SDK generic over `NetworkBackend`
  * implement `NetworkBackend` for litep2p

The new library should be considered experimental which is why
`rust-libp2p` will remain as the default option for the time being. This
PR currently depends on the master branch of `litep2p` but I'll cut a
new release for the library once all review comments have been
addresses.

---------

Signed-off-by: Alexandru Vasile <alexandru.vasile@parity.io>
Co-authored-by: Dmitry Markin <dmitry@markin.tech>
Co-authored-by: Alexandru Vasile <60601340+lexnv@users.noreply.github.com>
Co-authored-by: Alexandru Vasile <alexandru.vasile@parity.io>
2024-04-08 16:44:13 +00:00

353 lines
10 KiB
Rust

// Copyright (C) Parity Technologies (UK) Ltd.
// This file is part of Polkadot.
// Polkadot is free software: you can redistribute it and/or modify
// it under the terms of the GNU General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
// Polkadot is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU General Public License for more details.
// You should have received a copy of the GNU General Public License
// along with Polkadot. If not, see <http://www.gnu.org/licenses/>.
use std::{
collections::{HashMap, HashSet},
sync::Arc,
};
use async_trait::async_trait;
use parking_lot::Mutex;
use parity_scale_codec::Encode;
use sc_network::{
config::parse_addr, multiaddr::Multiaddr, service::traits::NetworkService, types::ProtocolName,
IfDisconnected, MessageSink, OutboundFailure, ReputationChange, RequestFailure,
};
use polkadot_node_network_protocol::{
peer_set::{CollationVersion, PeerSet, ProtocolVersion, ValidationVersion},
request_response::{OutgoingRequest, Recipient, ReqProtocolNames, Requests},
v1 as protocol_v1, v2 as protocol_v2, v3 as protocol_v3, PeerId,
};
use polkadot_primitives::AuthorityDiscoveryId;
use crate::{metrics::Metrics, validator_discovery::AuthorityDiscovery, WireMessage};
// network bridge network abstraction log target
const LOG_TARGET: &'static str = "parachain::network-bridge-net";
// Helper function to send a validation v1 message to a list of peers.
// Messages are always sent via the main protocol, even legacy protocol messages.
pub(crate) fn send_validation_message_v1(
peers: Vec<PeerId>,
message: WireMessage<protocol_v1::ValidationProtocol>,
metrics: &Metrics,
notification_sinks: &Arc<Mutex<HashMap<(PeerSet, PeerId), Box<dyn MessageSink>>>>,
) {
gum::trace!(target: LOG_TARGET, ?peers, ?message, "Sending validation v1 message to peers",);
send_message(
peers,
PeerSet::Validation,
ValidationVersion::V1.into(),
message,
metrics,
notification_sinks,
);
}
// Helper function to send a validation v3 message to a list of peers.
// Messages are always sent via the main protocol, even legacy protocol messages.
pub(crate) fn send_validation_message_v3(
peers: Vec<PeerId>,
message: WireMessage<protocol_v3::ValidationProtocol>,
metrics: &Metrics,
notification_sinks: &Arc<Mutex<HashMap<(PeerSet, PeerId), Box<dyn MessageSink>>>>,
) {
gum::trace!(target: LOG_TARGET, ?peers, ?message, "Sending validation v3 message to peers",);
send_message(
peers,
PeerSet::Validation,
ValidationVersion::V3.into(),
message,
metrics,
notification_sinks,
);
}
// Helper function to send a validation v2 message to a list of peers.
// Messages are always sent via the main protocol, even legacy protocol messages.
pub(crate) fn send_validation_message_v2(
peers: Vec<PeerId>,
message: WireMessage<protocol_v2::ValidationProtocol>,
metrics: &Metrics,
notification_sinks: &Arc<Mutex<HashMap<(PeerSet, PeerId), Box<dyn MessageSink>>>>,
) {
send_message(
peers,
PeerSet::Validation,
ValidationVersion::V2.into(),
message,
metrics,
notification_sinks,
);
}
// Helper function to send a collation v1 message to a list of peers.
// Messages are always sent via the main protocol, even legacy protocol messages.
pub(crate) fn send_collation_message_v1(
peers: Vec<PeerId>,
message: WireMessage<protocol_v1::CollationProtocol>,
metrics: &Metrics,
notification_sinks: &Arc<Mutex<HashMap<(PeerSet, PeerId), Box<dyn MessageSink>>>>,
) {
send_message(
peers,
PeerSet::Collation,
CollationVersion::V1.into(),
message,
metrics,
notification_sinks,
);
}
// Helper function to send a collation v2 message to a list of peers.
// Messages are always sent via the main protocol, even legacy protocol messages.
pub(crate) fn send_collation_message_v2(
peers: Vec<PeerId>,
message: WireMessage<protocol_v2::CollationProtocol>,
metrics: &Metrics,
notification_sinks: &Arc<Mutex<HashMap<(PeerSet, PeerId), Box<dyn MessageSink>>>>,
) {
send_message(
peers,
PeerSet::Collation,
CollationVersion::V2.into(),
message,
metrics,
notification_sinks,
);
}
/// Lower level function that sends a message to the network using the main protocol version.
///
/// This function is only used internally by the network-bridge, which is responsible to only send
/// messages that are compatible with the passed peer set, as that is currently not enforced by
/// this function. These are messages of type `WireMessage` parameterized on the matching type.
fn send_message<M>(
mut peers: Vec<PeerId>,
peer_set: PeerSet,
version: ProtocolVersion,
message: M,
metrics: &super::Metrics,
network_notification_sinks: &Arc<Mutex<HashMap<(PeerSet, PeerId), Box<dyn MessageSink>>>>,
) where
M: Encode + Clone,
{
if peers.is_empty() {
return
}
let message = {
let encoded = message.encode();
metrics.on_notification_sent(peer_set, version, encoded.len(), peers.len());
metrics.on_message(std::any::type_name::<M>());
encoded
};
let notification_sinks = network_notification_sinks.lock();
gum::trace!(
target: LOG_TARGET,
?peers,
?peer_set,
?version,
?message,
"Sending message to peers",
);
// optimization: avoid cloning the message for the last peer in the
// list. The message payload can be quite large. If the underlying
// network used `Bytes` this would not be necessary.
//
// peer may have gotten disconnect by the time `send_message()` is called
// at which point the the sink is not available.
let last_peer = peers.pop();
peers.into_iter().for_each(|peer| {
if let Some(sink) = notification_sinks.get(&(peer_set, peer)) {
sink.send_sync_notification(message.clone());
}
});
if let Some(peer) = last_peer {
if let Some(sink) = notification_sinks.get(&(peer_set, peer)) {
sink.send_sync_notification(message.clone());
}
}
}
/// An abstraction over networking for the purposes of this subsystem.
#[async_trait]
pub trait Network: Clone + Send + 'static {
/// Ask the network to keep a substream open with these nodes and not disconnect from them
/// until removed from the protocol's peer set.
/// Note that `out_peers` setting has no effect on this.
async fn set_reserved_peers(
&mut self,
protocol: ProtocolName,
multiaddresses: HashSet<Multiaddr>,
) -> Result<(), String>;
/// Removes the peers for the protocol's peer set (both reserved and non-reserved).
async fn remove_from_peers_set(
&mut self,
protocol: ProtocolName,
peers: Vec<PeerId>,
) -> Result<(), String>;
/// Send a request to a remote peer.
async fn start_request<AD: AuthorityDiscovery>(
&self,
authority_discovery: &mut AD,
req: Requests,
req_protocol_names: &ReqProtocolNames,
if_disconnected: IfDisconnected,
);
/// Report a given peer as either beneficial (+) or costly (-) according to the given scalar.
fn report_peer(&self, who: PeerId, rep: ReputationChange);
/// Disconnect a given peer from the protocol specified without harming reputation.
fn disconnect_peer(&self, who: PeerId, protocol: ProtocolName);
/// Get peer role.
fn peer_role(&self, who: PeerId, handshake: Vec<u8>) -> Option<sc_network::ObservedRole>;
}
#[async_trait]
impl Network for Arc<dyn NetworkService> {
async fn set_reserved_peers(
&mut self,
protocol: ProtocolName,
multiaddresses: HashSet<Multiaddr>,
) -> Result<(), String> {
<dyn NetworkService>::set_reserved_peers(&**self, protocol, multiaddresses)
}
async fn remove_from_peers_set(
&mut self,
protocol: ProtocolName,
peers: Vec<PeerId>,
) -> Result<(), String> {
<dyn NetworkService>::remove_peers_from_reserved_set(&**self, protocol, peers)
}
fn report_peer(&self, who: PeerId, rep: ReputationChange) {
<dyn NetworkService>::report_peer(&**self, who, rep);
}
fn disconnect_peer(&self, who: PeerId, protocol: ProtocolName) {
<dyn NetworkService>::disconnect_peer(&**self, who, protocol);
}
async fn start_request<AD: AuthorityDiscovery>(
&self,
authority_discovery: &mut AD,
req: Requests,
req_protocol_names: &ReqProtocolNames,
if_disconnected: IfDisconnected,
) {
let (protocol, OutgoingRequest { peer, payload, pending_response, fallback_request }) =
req.encode_request();
let peer_id = match peer {
Recipient::Peer(peer_id) => Some(peer_id),
Recipient::Authority(authority) => {
gum::trace!(
target: LOG_TARGET,
?authority,
"Searching for peer id to connect to authority",
);
let mut found_peer_id = None;
// Note: `get_addresses_by_authority_id` searched in a cache, and it thus expected
// to be very quick.
for addr in authority_discovery
.get_addresses_by_authority_id(authority)
.await
.into_iter()
.flat_map(|list| list.into_iter())
{
let (peer_id, addr) = match parse_addr(addr) {
Ok(v) => v,
Err(_) => continue,
};
<dyn NetworkService>::add_known_address(&**self, peer_id, addr);
found_peer_id = Some(peer_id);
}
found_peer_id
},
};
let peer_id = match peer_id {
None => {
gum::debug!(target: LOG_TARGET, "Discovering authority failed");
match pending_response
.send(Err(RequestFailure::Network(OutboundFailure::DialFailure)))
{
Err(_) => {
gum::debug!(target: LOG_TARGET, "Sending failed request response failed.")
},
Ok(_) => {},
}
return
},
Some(peer_id) => peer_id,
};
gum::trace!(
target: LOG_TARGET,
%peer_id,
protocol = %req_protocol_names.get_name(protocol),
fallback_protocol = ?fallback_request.as_ref().map(|(_, p)| req_protocol_names.get_name(*p)),
?if_disconnected,
"Starting request",
);
<dyn NetworkService>::start_request(
&**self,
peer_id,
req_protocol_names.get_name(protocol),
payload,
fallback_request.map(|(r, p)| (r, req_protocol_names.get_name(p))),
pending_response,
if_disconnected,
);
}
fn peer_role(&self, who: PeerId, handshake: Vec<u8>) -> Option<sc_network::ObservedRole> {
<dyn NetworkService>::peer_role(&**self, who, handshake)
}
}
/// We assume one `peer_id` per `authority_id`.
pub async fn get_peer_id_by_authority_id<AD: AuthorityDiscovery>(
authority_discovery: &mut AD,
authority: AuthorityDiscoveryId,
) -> Option<PeerId> {
// Note: `get_addresses_by_authority_id` searched in a cache, and it thus expected
// to be very quick.
authority_discovery
.get_addresses_by_authority_id(authority)
.await
.into_iter()
.flat_map(|list| list.into_iter())
.find_map(|addr| parse_addr(addr).ok().map(|(p, _)| p))
}