radicle-heartwood-lfs/radicle-node/src/service.rs

1600 lines
57 KiB
Rust

#![allow(clippy::too_many_arguments)]
#![allow(clippy::collapsible_match)]
#![allow(clippy::collapsible_if)]
pub mod config;
pub mod filter;
pub mod message;
pub mod reactor;
pub mod routing;
pub mod session;
pub mod tracking;
use std::collections::hash_map::Entry;
use std::collections::{BTreeMap, HashMap, HashSet};
use std::ops::{Deref, DerefMut};
use std::sync::Arc;
use std::{fmt, net, str};
use crossbeam_channel as chan;
use fastrand::Rng;
use localtime::{LocalDuration, LocalTime};
use log::*;
use crate::address;
use crate::address::AddressBook;
use crate::clock::Timestamp;
use crate::crypto;
use crate::crypto::{Signer, Verified};
use crate::identity::{Doc, Id};
use crate::node;
use crate::node::{Address, Features, FetchResult};
use crate::prelude::*;
use crate::service::message::{Announcement, AnnouncementMessage, Ping};
use crate::service::message::{NodeAnnouncement, RefsAnnouncement};
use crate::storage;
use crate::storage::{Inventory, ReadRepository, RefUpdate, WriteStorage};
use crate::storage::{Namespaces, ReadStorage};
use crate::worker::FetchError;
use crate::Link;
pub use crate::node::NodeId;
pub use crate::service::config::{Config, Network};
pub use crate::service::message::{Message, ZeroBytes};
pub use crate::service::reactor::Fetch;
pub use crate::service::session::Session;
use self::gossip::Gossip;
use self::message::InventoryAnnouncement;
use self::reactor::Reactor;
/// Target number of peers to maintain connections to.
pub const TARGET_OUTBOUND_PEERS: usize = 8;
/// How often to run the "idle" task.
pub const IDLE_INTERVAL: LocalDuration = LocalDuration::from_secs(30);
/// How often to run the "announce" task.
pub const ANNOUNCE_INTERVAL: LocalDuration = LocalDuration::from_secs(30);
/// How often to run the "sync" task.
pub const SYNC_INTERVAL: LocalDuration = LocalDuration::from_secs(60);
/// How often to run the "prune" task.
pub const PRUNE_INTERVAL: LocalDuration = LocalDuration::from_mins(30);
/// Duration to wait on an unresponsive peer before dropping its connection.
pub const STALE_CONNECTION_TIMEOUT: LocalDuration = LocalDuration::from_secs(60);
/// How much time should pass after a peer was last active for a *ping* to be sent.
pub const KEEP_ALIVE_DELTA: LocalDuration = LocalDuration::from_secs(30);
/// Maximum time difference between the local time, and an announcement timestamp.
pub const MAX_TIME_DELTA: LocalDuration = LocalDuration::from_mins(60);
/// Maximum attempts to connect to a peer before we give up.
pub const MAX_CONNECTION_ATTEMPTS: usize = 3;
/// How far back from the present time should we request gossip messages when connecting to a peer.
pub const SUBSCRIBE_BACKLOG_DELTA: LocalDuration = LocalDuration::from_mins(60);
/// Maximum external address limit imposed by message size limits.
pub use message::ADDRESS_LIMIT;
/// Maximum inventory limit imposed by message size limits.
pub use message::INVENTORY_LIMIT;
/// Maximum number of project git references imposed by message size limits.
pub use message::REF_REMOTE_LIMIT;
/// A service event.
#[derive(Debug, Clone)]
pub enum Event {
RefsFetched {
remote: NodeId,
rid: Id,
updated: Vec<RefUpdate>,
},
}
/// General service error.
#[derive(thiserror::Error, Debug)]
pub enum Error {
#[error(transparent)]
Storage(#[from] storage::Error),
#[error(transparent)]
Routing(#[from] routing::Error),
#[error(transparent)]
Tracking(#[from] tracking::Error),
}
/// Function used to query internal service state.
pub type QueryState = dyn Fn(&dyn ServiceState) -> Result<(), CommandError> + Send + Sync;
/// Commands sent to the service by the operator.
pub enum Command {
/// Announce repository references for given repository to peers.
AnnounceRefs(Id),
/// Announce local inventory to peers.
SyncInventory(chan::Sender<bool>),
/// Connect to node with the given address.
Connect(NodeId, Address),
/// Lookup seeds for the given repository in the routing table.
Seeds(Id, chan::Sender<Vec<NodeId>>),
/// Fetch the given repository from the network.
Fetch(Id, NodeId, chan::Sender<FetchResult>),
/// Track the given repository.
TrackRepo(Id, chan::Sender<bool>),
/// Untrack the given repository.
UntrackRepo(Id, chan::Sender<bool>),
/// Track the given node.
TrackNode(NodeId, Option<String>, chan::Sender<bool>),
/// Untrack the given node.
UntrackNode(NodeId, chan::Sender<bool>),
/// Query the internal service state.
QueryState(Arc<QueryState>, chan::Sender<Result<(), CommandError>>),
}
impl fmt::Debug for Command {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
match self {
Self::AnnounceRefs(id) => write!(f, "AnnounceRefs({id})"),
Self::SyncInventory(_) => write!(f, "SyncInventory(..)"),
Self::Connect(id, addr) => write!(f, "Connect({id}, {addr})"),
Self::Seeds(id, _) => write!(f, "Seeds({id})"),
Self::Fetch(id, node, _) => write!(f, "Fetch({id}, {node})"),
Self::TrackRepo(id, _) => write!(f, "TrackRepo({id})"),
Self::UntrackRepo(id, _) => write!(f, "UntrackRepo({id})"),
Self::TrackNode(id, _, _) => write!(f, "TrackNode({id})"),
Self::UntrackNode(id, _) => write!(f, "UntrackNode({id})"),
Self::QueryState { .. } => write!(f, "QueryState(..)"),
}
}
}
/// Command-related errors.
#[derive(thiserror::Error, Debug)]
pub enum CommandError {
#[error(transparent)]
Storage(#[from] storage::Error),
#[error(transparent)]
Routing(#[from] routing::Error),
}
#[derive(Debug)]
pub struct Service<R, A, S, G> {
/// Service configuration.
config: Config,
/// Our cryptographic signer and key.
signer: G,
/// Project storage.
storage: S,
/// Network routing table. Keeps track of where projects are located.
routing: R,
/// Node address manager.
addresses: A,
/// Tracking policy configuration.
tracking: tracking::Config,
/// State relating to gossip.
gossip: Gossip,
/// Peer sessions, currently or recently connected.
sessions: Sessions,
/// Keeps track of node states.
nodes: BTreeMap<NodeId, Node>,
/// Clock. Tells the time.
clock: LocalTime,
/// Interface to the I/O reactor.
reactor: Reactor,
/// Source of entropy.
rng: Rng,
/// Whether our local inventory no long represents what we have announced to the network.
out_of_sync: bool,
/// Fetch requests initiated by user, which are waiting for results.
fetch_reqs: HashMap<Id, chan::Sender<FetchResult>>,
/// Current tracked repository bloom filter.
filter: Filter,
/// Last time the service was idle.
last_idle: LocalTime,
/// Last time the service synced.
last_sync: LocalTime,
/// Last time the service routing table was pruned.
last_prune: LocalTime,
/// Last time the service announced its inventory.
last_announce: LocalTime,
/// Time when the service was initialized.
start_time: LocalTime,
}
impl<R, A, S, G> Service<R, A, S, G>
where
G: crypto::Signer,
{
/// Get the local node id.
pub fn node_id(&self) -> NodeId {
*self.signer.public_key()
}
/// Get the local service time.
pub fn local_time(&self) -> LocalTime {
self.clock
}
}
impl<R, A, S, G> Service<R, A, S, G>
where
R: routing::Store,
A: address::Store,
S: WriteStorage + 'static,
G: Signer,
{
pub fn new(
config: Config,
clock: LocalTime,
routing: R,
storage: S,
addresses: A,
tracking: tracking::Config,
signer: G,
rng: Rng,
) -> Self {
let sessions = Sessions::new(rng.clone());
Self {
config,
storage,
addresses,
tracking,
signer,
rng,
clock,
routing,
gossip: Gossip::default(),
// FIXME: This should be loaded from the address store.
nodes: BTreeMap::new(),
reactor: Reactor::default(),
sessions,
out_of_sync: false,
fetch_reqs: HashMap::new(),
filter: Filter::empty(),
last_idle: LocalTime::default(),
last_sync: LocalTime::default(),
last_prune: LocalTime::default(),
last_announce: LocalTime::default(),
start_time: LocalTime::default(),
}
}
/// Return the next i/o action to execute.
#[allow(clippy::should_implement_trait)]
pub fn next(&mut self) -> Option<reactor::Io> {
self.reactor.next()
}
/// Track a repository.
/// Returns whether or not the tracking policy was updated.
pub fn track_repo(&mut self, id: &Id, scope: tracking::Scope) -> Result<bool, tracking::Error> {
self.out_of_sync = self.tracking.track_repo(id, scope)?;
self.filter.insert(id);
Ok(self.out_of_sync)
}
/// Untrack a repository.
/// Returns whether or not the tracking policy was updated.
/// Note that when untracking, we don't announce anything to the network. This is because by
/// simply not announcing it anymore, it will eventually be pruned by nodes.
pub fn untrack_repo(&mut self, id: &Id) -> Result<bool, tracking::Error> {
let updated = self.tracking.untrack_repo(id)?;
// Nb. This is potentially slow if we have lots of projects. We should probably
// only re-compute the filter when we've untracked a certain amount of projects
// and the filter is really out of date.
//
// TODO: Share this code with initialization code.
self.filter = Filter::new(
self.tracking
.repo_entries()?
.filter(|(_, _, policy)| *policy == tracking::Policy::Track)
.map(|(e, _, _)| e),
);
Ok(updated)
}
/// Check whether we are tracking a certain repository.
pub fn is_tracking(&self, id: &Id) -> Result<bool, tracking::Error> {
self.tracking.is_repo_tracked(id)
}
/// Find the closest `n` peers by proximity in tracking graphs.
/// Returns a sorted list from the closest peer to the furthest.
/// Peers with more trackings in common score score higher.
#[allow(unused)]
pub fn closest_peers(&self, n: usize) -> Vec<NodeId> {
todo!()
}
/// Get the address book instance.
pub fn addresses(&self) -> &A {
&self.addresses
}
/// Get the mutable address book instance.
pub fn addresses_mut(&mut self) -> &mut A {
&mut self.addresses
}
/// Get the storage instance.
pub fn storage(&self) -> &S {
&self.storage
}
/// Get the mutable storage instance.
pub fn storage_mut(&mut self) -> &mut S {
&mut self.storage
}
/// Get the tracking policy.
pub fn tracking(&self) -> &tracking::Config {
&self.tracking
}
/// Get the local signer.
pub fn signer(&self) -> &G {
&self.signer
}
/// Get I/O reactor.
pub fn reactor(&mut self) -> &mut Reactor {
&mut self.reactor
}
/// Lookup a project, both locally and in the routing table.
pub fn lookup(&self, id: Id) -> Result<Lookup, LookupError> {
let remote = self.routing.get(&id)?.iter().cloned().collect();
Ok(Lookup {
local: self.storage.get(&self.node_id(), id)?,
remote,
})
}
pub fn initialize(&mut self, time: LocalTime) -> Result<(), Error> {
debug!(target: "service", "Init @{}", time.as_millis());
self.start_time = time;
// Connect to configured peers.
let addrs = self.config.connect.clone();
for (id, addr) in addrs {
self.connect(id, addr);
}
// Ensure that our inventory is recorded in our routing table.
for id in self.storage.inventory()? {
self.routing.insert(id, self.node_id(), time.as_millis())?;
}
// Setup subscription filter for tracked repos.
self.filter = Filter::new(
self.tracking
.repo_entries()?
.filter(|(_, _, policy)| *policy == tracking::Policy::Track)
.map(|(e, _, _)| e),
);
Ok(())
}
pub fn tick(&mut self, now: LocalTime) {
trace!("Tick +{}", now - self.start_time);
self.clock = now;
}
pub fn wake(&mut self) {
let now = self.clock;
trace!("Wake +{}", now - self.start_time);
if now - self.last_idle >= IDLE_INTERVAL {
debug!(target: "service", "Running 'idle' task...");
self.keep_alive(&now);
self.disconnect_unresponsive_peers(&now);
self.maintain_connections();
self.reactor.wakeup(IDLE_INTERVAL);
self.last_idle = now;
}
if now - self.last_sync >= SYNC_INTERVAL {
debug!(target: "service", "Running 'sync' task...");
// TODO: What do we do here?
self.reactor.wakeup(SYNC_INTERVAL);
self.last_sync = now;
}
if now - self.last_announce >= ANNOUNCE_INTERVAL {
if self.out_of_sync {
if let Err(err) = self
.storage
.inventory()
.and_then(|i| self.announce_inventory(i))
{
error!("Error announcing inventory: {}", err);
}
}
self.reactor.wakeup(ANNOUNCE_INTERVAL);
self.last_announce = now;
}
if now - self.last_prune >= PRUNE_INTERVAL {
debug!(target: "service", "Running 'prune' task...");
if let Err(err) = self.prune_routing_entries(&now) {
error!("Error pruning routing entries: {}", err);
}
self.reactor.wakeup(PRUNE_INTERVAL);
self.last_prune = now;
}
}
pub fn command(&mut self, cmd: Command) {
debug!(target: "service", "Received command {:?}", cmd);
match cmd {
Command::Connect(id, addr) => {
self.connect(id, addr);
}
Command::Seeds(rid, resp) => {
let (connected, unconnected) = match self.routing.get(&rid) {
Ok(seeds) => seeds
.into_iter()
.filter(|node| *node != self.node_id())
.partition::<Vec<_>, _>(|node| self.sessions.is_negotiated(node)),
Err(err) => {
error!(target: "service", "Error reading routing table for {rid}: {err}");
drop(resp);
return;
}
};
debug!(
target: "service",
"Found {} connected seed(s) and {} unconnected seed(s) for {}",
connected.len(), unconnected.len(), rid
);
resp.send(connected).ok();
}
Command::Fetch(rid, seed, resp) => {
// TODO: Establish connections to unconnected seeds, and retry.
// TODO: Fetch requests should be queued and re-checked to see if they can
// be fulfilled everytime a new node connects.
self.fetch_reqs.insert(rid, resp);
self.fetch(rid, &seed);
}
Command::TrackRepo(rid, resp) => {
let tracked = self
.track_repo(&rid, tracking::Scope::All)
.expect("Service::command: error tracking repository");
// TODO: Try to fetch project if we weren't tracking it before.
resp.send(tracked).ok();
// Let all our peers know that we're interested in this repo from now on.
self.reactor.broadcast(
Message::subscribe(self.filter(), self.time(), Timestamp::MAX),
self.sessions.negotiated().map(|(_, s)| s),
);
}
Command::UntrackRepo(id, resp) => {
let untracked = self
.untrack_repo(&id)
.expect("Service::command: error untracking repository");
resp.send(untracked).ok();
}
Command::TrackNode(id, alias, resp) => {
let tracked = self
.tracking
.track_node(&id, alias.as_deref())
.expect("Service::command: error tracking node");
resp.send(tracked).ok();
}
Command::UntrackNode(id, resp) => {
let untracked = self
.tracking
.untrack_node(&id)
.expect("Service::command: error untracking node");
resp.send(untracked).ok();
}
Command::AnnounceRefs(id) => {
if let Err(err) = self.announce_refs(id, Namespaces::One(self.node_id())) {
error!("Error announcing refs: {}", err);
}
}
Command::SyncInventory(resp) => {
let updated = self
.sync_and_announce_inventory()
.expect("Service::command: error syncing and announcing inventory");
resp.send(!updated.is_empty()).ok();
}
Command::QueryState(query, sender) => {
sender.send(query(self)).ok();
}
}
}
pub fn fetch(&mut self, rid: Id, from: &NodeId) {
let Some(session) = self.sessions.get_mut(from) else {
error!(target: "service", "Session {from} does not exist; cannot initiate fetch");
return;
};
debug_assert!(session.is_negotiated());
let seed = session.id;
match session.fetch(rid) {
session::FetchResult::Ready(fetch) => {
debug!(target: "service", "Fetch initiated for {rid} with {seed}..");
self.reactor.write(session, fetch);
session.to_requesting(rid);
}
session::FetchResult::AlreadyFetching(other) => {
if other == rid {
debug!(target: "service", "Ignoring redundant attempt to fetch {rid} from {from}");
} else {
// TODO: If we can't fetch, it's because we're already fetching from
// this peer. So we need to queue the request, or find another peer.
error!(
target: "service",
"Dropping fetch for {rid} from {from}: another fetch is ongoing"
);
}
}
session::FetchResult::NotConnected => {
error!(target: "service", "Unable to fetch {rid} from peer {seed}: peer is not connected");
}
}
}
pub fn fetched(&mut self, fetch: Fetch, result: Result<Vec<RefUpdate>, FetchError>) {
let remote = fetch.remote;
let rid = fetch.rid;
let initiated = fetch.initiated;
if initiated {
let result = match result {
Ok(updated) => {
log::debug!(target: "service", "Fetched {rid} from {remote}");
self.reactor.event(Event::RefsFetched {
remote,
rid,
updated: updated.clone(),
});
FetchResult::Success { updated }
}
Err(err) => {
let reason = err.to_string();
error!(target: "service", "Fetch failed for {rid} from {remote}: {reason}");
// For now, we only disconnect the remote in case of timeout. In the future,
// there may be other reasons to disconnect.
if err.is_timeout() {
self.reactor
.disconnect(remote, DisconnectReason::Fetch(err));
}
FetchResult::Failed { reason }
}
};
if let Some(results) = self.fetch_reqs.get(&rid) {
log::debug!(target: "service", "Found existing fetch request, sending result..");
if results.send(result).is_err() {
log::error!(target: "service", "Error sending fetch result for {rid}..");
// FIXME: We should remove the channel even on success, once all seeds
// were fetched from. Otherwise an organic fetch will try to send on the
// channel.
self.fetch_reqs.remove(&rid);
} else {
log::debug!(target: "service", "Sent fetch result for {rid}..");
}
} else {
log::debug!(target: "service", "No fetch requests found for {rid}..");
// We only announce refs here when the fetch wasn't user-requested. This is
// because the user might want to announce his fork, once he has created one,
// or may choose to not announce anything.
if let Err(e) = self.announce_refs(rid, fetch.namespaces) {
error!(target: "service", "Failed to announce new refs: {e}");
}
}
// TODO: Since this fetch could be either a full clone or simply a ref update, we need
// to either announce new inventory, or new refs. Right now, we announce both in some
// cases.
// Announce the newly fetched repository to the network, if necessary.
if let Err(e) = self.sync_and_announce_inventory() {
error!(target: "service", "Failed to sync announce new inventory: {e}");
}
}
if let Some(session) = self.sessions.get_mut(&remote) {
// Transition session back to gossip protocol.
session.to_gossip();
// Drain any messages in the session's outbox, which might have accumulated during the
// fetch, and send them to the peer.
self.reactor.drain(session);
} else {
log::debug!(target: "service", "Session not found for {remote}");
}
}
pub fn accepted(&mut self, _addr: net::SocketAddr) {
// Inbound connection attempt.
}
pub fn attempted(&mut self, id: NodeId, addr: &Address) {
debug!(target: "service", "Attempted connection to {id} ({addr})");
let persistent = self.config.is_persistent(&id);
self.sessions
.entry(id)
.and_modify(|sess| sess.to_connecting())
.or_insert_with(|| Session::connecting(id, persistent, self.rng.clone()));
}
pub fn connected(&mut self, remote: NodeId, link: Link) {
info!(target: "service", "Connected to {} ({:?})", remote, link);
// For outbound connections, we are the first to say "Hello".
// For inbound connections, we wait for the remote to say "Hello" first.
if link.is_outbound() {
let filter = self.filter();
if let Some(peer) = self.sessions.get_mut(&remote) {
peer.to_connected(self.clock);
self.reactor.write_all(
peer,
gossip::handshake(
self.clock.as_millis(),
&self.storage,
&self.signer,
filter,
&self.config,
),
);
}
} else {
match self.sessions.entry(remote) {
Entry::Occupied(e) => {
warn!(
target: "service",
"Connecting peer {remote} already has a session open ({})", e.get()
);
}
Entry::Vacant(e) => {
e.insert(Session::connected(
remote,
Link::Inbound,
self.config.is_persistent(&remote),
self.rng.clone(),
self.clock,
));
}
}
}
}
pub fn disconnected(&mut self, remote: NodeId, reason: &DisconnectReason) {
let since = self.local_time();
debug!(target: "service", "Disconnected from {} ({})", remote, reason);
if let Some(session) = self.sessions.get_mut(&remote) {
session.to_disconnected(since);
// Attempt to re-connect to persistent peers.
if let Some(address) = self.config.peer(&remote) {
if session.attempts() < MAX_CONNECTION_ATTEMPTS {
if reason.is_dial_err() {
return;
}
if !reason.is_transient() {
return;
}
// TODO: Eventually we want a delay before attempting a reconnection,
// with exponential back-off.
debug!(target: "service",
"Reconnecting to {} (attempts={})...",
remote,
session.attempts()
);
// TODO: Try to reconnect only if the peer was attempted. A disconnect without
// even a successful attempt means that we're unlikely to be able to reconnect.
self.connect(remote, address.clone());
}
} else {
self.sessions.remove(&remote);
self.maintain_connections();
}
}
}
pub fn received_message(&mut self, remote: NodeId, message: Message) {
match self.handle_message(&remote, message) {
Err(session::Error::NotFound(id)) => {
error!("Session not found for {id}");
}
Err(err) => {
// If there's an error, stop processing messages from this peer.
// However, we still relay messages returned up to this point.
self.reactor
.disconnect(remote, DisconnectReason::Session(err));
// FIXME: The peer should be set in a state such that we don't
// process further messages.
}
Ok(()) => {}
}
}
/// Handle an announcement message.
///
/// Returns `true` if this announcement should be stored and relayed to connected peers,
/// and `false` if it should not.
pub fn handle_announcement(
&mut self,
relayer: &NodeId,
announcement: &Announcement,
) -> Result<bool, session::Error> {
if !announcement.verify() {
return Err(session::Error::Misbehavior);
}
let Announcement {
node: announcer,
message,
..
} = announcement;
// Ignore our own announcements, in case the relayer sent one by mistake.
if *announcer == self.node_id() {
return Ok(false);
}
let now = self.clock;
let timestamp = message.timestamp();
let relay = self.config.relay;
let peer = self.nodes.entry(*announcer).or_insert_with(Node::default);
// Don't allow messages from too far in the future.
if timestamp.saturating_sub(now.as_millis()) > MAX_TIME_DELTA.as_millis() as u64 {
return Err(session::Error::InvalidTimestamp(timestamp));
}
match message {
AnnouncementMessage::Inventory(message) => {
// Discard inventory messages we've already seen, otherwise update
// out last seen time.
if !peer.inventory_announced(timestamp) {
debug!(target: "service", "Ignoring stale inventory announcement from {announcer} (t={})", self.time());
return Ok(false);
}
match self.sync_routing(&message.inventory, *announcer, message.timestamp) {
Ok(updated) => {
if updated.is_empty() {
return Ok(false);
}
}
Err(e) => {
error!("Error processing inventory from {}: {}", announcer, e);
return Ok(false);
}
}
for id in message.inventory.as_slice() {
// TODO: Move this out (good luck with the borrow checker).
if let Some(sess) = self.sessions.get_mut(announcer) {
// If we are connected to the announcer of this inventory, update the peer's
// subscription filter to include all inventory items. This way, we'll
// relay messages relating to the peer's inventory.
if let Some(sub) = &mut sess.subscribe {
sub.filter.insert(id);
}
// If we're tracking and connected to the announcer, and we don't have
// the inventory, fetch it from the announcer.
if self.tracking.is_repo_tracked(id).expect(
"Service::handle_announcement: error accessing tracking configuration",
) {
// Only if we do not have the repository locally do we fetch here.
// If we do have it, only fetch after receiving a ref announcement.
if let Ok(true) = self.storage.contains(id) {
// Do nothing.
} else {
// We may hit this branch due to an error returned by storage.
// We attempt to fetch in case of error because it's likely
// the repository was corrupted, and fetching will fix it.
self.fetch(*id, announcer);
}
}
}
}
return Ok(relay);
}
// Process a peer inventory update announcement by (maybe) fetching.
AnnouncementMessage::Refs(message) => {
for (remote_id, theirs) in message.refs.iter() {
if theirs.verify(remote_id).is_err() {
warn!(target: "service", "Peer {relayer} relayed refs announcement with invalid signature for {remote_id}");
return Err(session::Error::Misbehavior);
}
}
// We update inventories when receiving ref announcements, as these could come
// from a new repository being initialized.
if let Ok(updated) = self
.routing
.insert(message.rid, *relayer, message.timestamp)
{
if updated {
info!(target: "service", "Routing table updated for {} with seed {relayer}", message.rid);
}
}
// TODO: Buffer/throttle fetches.
if self
.tracking
.is_repo_tracked(&message.rid)
.expect("Service::handle_announcement: error accessing tracking configuration")
{
// Discard announcement messages we've already seen, otherwise update
// our last seen time.
if !peer.refs_announced(message.rid, timestamp) {
debug!(target: "service", "Ignoring stale refs announcement from {announcer}");
return Ok(false);
}
// Refs can be relayed by peers who don't have the data in storage,
// therefore we only check whether we are connected to the *announcer*,
// which is required by the protocol to only announce refs it has.
if self.sessions.is_negotiated(announcer) {
match message.is_fresh(&self.storage) {
Ok(is_fresh) => {
if is_fresh {
// TODO: Only fetch if the refs announced are for peers we're tracking.
self.fetch(message.rid, announcer);
}
}
Err(e) => {
error!(target: "service", "Failed to check ref announcement freshness: {e}");
}
}
}
return Ok(relay);
} else {
debug!(
target: "service",
"Ignoring refs announcement from {announcer}: repository {} isn't tracked",
message.rid
);
}
}
AnnouncementMessage::Node(
ann @ NodeAnnouncement {
features,
alias,
addresses,
..
},
) => {
// Discard node messages we've already seen, otherwise update
// our last seen time.
if !peer.node_announced(timestamp) {
debug!(target: "service", "Ignoring stale node announcement from {announcer}");
return Ok(false);
}
if !ann.validate() {
warn!(target: "service", "Dropping node announcement from {announcer}: invalid proof-of-work");
return Ok(false);
}
let alias = match str::from_utf8(alias) {
Ok(s) => s,
Err(e) => {
warn!(target: "service", "Dropping node announcement from {announcer}: invalid alias: {e}");
return Ok(false);
}
};
// If this node isn't a seed, we're not interested in adding it
// to our address book, but other nodes may be, so we relay the message anyway.
if !features.has(Features::SEED) {
return Ok(relay);
}
match self.addresses.insert(
announcer,
*features,
alias,
timestamp,
addresses
.iter()
.map(|a| address::KnownAddress::new(a.clone(), address::Source::Peer)),
) {
Ok(updated) => {
// Only relay if we received new information.
if updated {
debug!(target: "service",
"Address store entry for node {announcer} updated at {timestamp}"
);
return Ok(relay);
}
}
Err(err) => {
// An error here is due to a fault in our address store.
error!("Error processing node announcement from {announcer}: {err}");
}
}
}
}
Ok(false)
}
pub fn handle_message(
&mut self,
remote: &NodeId,
message: Message,
) -> Result<(), session::Error> {
let filter = self.filter(); // TODO: Don't call this if it's not used.
let Some(peer) = self.sessions.get_mut(remote) else {
return Err(session::Error::NotFound(*remote));
};
peer.last_active = self.clock;
debug!(target: "service", "Received message {:?} from {}", &message, peer.id);
match (&mut peer.state, message) {
(
session::State::Connected {
protocol: session::Protocol::Fetch { .. },
..
},
_,
) => {
// This should never happen if the service is properly configured, since all
// incoming data is sent directly to the Git worker.
log::error!(target: "service", "Received gossip message from {remote} during git fetch");
return Err(session::Error::Misbehavior);
}
(session::State::Connected { initialized, .. }, Message::Initialize { .. }) => {
// Already initialized!
if *initialized {
debug!(
target: "service",
"Disconnecting peer {} for initializing already initialized session",
peer.id
);
return Err(session::Error::Misbehavior);
}
*initialized = true;
if peer.link.is_inbound() {
self.reactor.write_all(
peer,
gossip::handshake(
self.clock.as_millis(),
&self.storage,
&self.signer,
filter,
&self.config,
),
);
}
// Nb. we don't set the peer timestamp here, since it is going to be
// set after the first message is received only. Setting it here would
// mean that messages received right after the handshake could be ignored.
}
// Process a peer announcement.
(session::State::Connected { .. }, Message::Announcement(ann)) => {
let relayer = peer.id;
// Returning true here means that the message should be relayed.
if self.handle_announcement(&relayer, &ann)? {
self.gossip.received(ann.clone(), ann.message.timestamp());
// Choose peers we should relay this message to.
// 1. Don't relay to the peer who sent us this message.
// 2. Don't relay to the peer who signed this announcement.
let relay_to = self
.sessions
.negotiated()
.filter(|(id, _)| *id != remote && *id != &ann.node);
self.reactor.relay(ann.clone(), relay_to.map(|(_, p)| p));
return Ok(());
}
}
(session::State::Connected { .. }, Message::Subscribe(subscribe)) => {
for ann in self
.gossip
// Filter announcements by interest.
.filtered(&subscribe.filter, subscribe.since, subscribe.until)
// Don't send announcements authored by the remote, back to the remote.
.filter(|ann| &ann.node != remote)
{
self.reactor.write(peer, ann.into());
}
peer.subscribe = Some(subscribe);
}
(session::State::Connected { .. }, Message::Ping(Ping { ponglen, .. })) => {
// Ignore pings which ask for too much data.
if ponglen > Ping::MAX_PONG_ZEROES {
return Ok(());
}
self.reactor.write(
peer,
Message::Pong {
zeroes: ZeroBytes::new(ponglen),
},
);
}
(session::State::Connected { ping, .. }, Message::Pong { zeroes }) => {
if let session::PingState::AwaitingResponse(ponglen) = *ping {
if (ponglen as usize) == zeroes.len() {
*ping = session::PingState::Ok;
}
}
}
(session::State::Connected { .. }, Message::Fetch { rid }) => {
debug!(target: "service", "Fetch requested for {rid} from {remote}..");
// TODO: Check that we have the repo first?
// Accept the request and instruct the transport to handover the socket to the worker.
self.reactor.write(peer, Message::FetchOk { rid });
self.reactor.fetch(peer, rid, Namespaces::default(), false);
}
(session::State::Connected { protocol, .. }, Message::FetchOk { rid }) => {
if *protocol
!= (session::Protocol::Gossip {
requested: Some(rid),
})
{
// As long as we disconnect peers who don't respond to our fetch requests within
// the alloted time, this shouldn't happen by mistake.
error!(
"Received unexpected message `fetch-ok` from peer {}",
peer.id
);
return Err(session::Error::Misbehavior);
}
debug!(target: "service", "Fetch accepted for {rid} from {remote}..");
// Instruct the transport to handover the socket to the worker.
self.reactor.fetch(peer, rid, Namespaces::default(), true);
}
(session::State::Connecting { .. }, msg) => {
error!("Received {:?} from connecting peer {}", msg, peer.id);
}
(session::State::Disconnected { .. }, msg) => {
debug!(target: "service", "Ignoring {:?} from disconnected peer {}", msg, peer.id);
}
}
Ok(())
}
/// Sync, and if needed, announce our local inventory.
fn sync_and_announce_inventory(&mut self) -> Result<Vec<Id>, Error> {
let inventory = self.storage.inventory()?;
let updated = self.sync_routing(&inventory, self.node_id(), self.time())?;
if !updated.is_empty() {
self.announce_inventory(inventory)?;
}
Ok(updated)
}
/// Process a peer inventory announcement by updating our routing table.
/// This function expects the peer's full inventory, and prunes entries that are not in the
/// given inventory.
fn sync_routing(
&mut self,
inventory: &[Id],
from: NodeId,
timestamp: Timestamp,
) -> Result<Vec<Id>, Error> {
let mut updated = Vec::new();
let mut included = HashSet::new();
for proj_id in inventory {
included.insert(proj_id);
if self.routing.insert(*proj_id, from, timestamp)? {
info!(target: "service", "Routing table updated for {proj_id} with seed {from}");
if self
.tracking
.is_repo_tracked(proj_id)
.expect("Service::process_inventory: error accessing tracking configuration")
{
// TODO: We should fetch here if we're already connected, case this seed has
// refs we don't have.
}
updated.push(*proj_id);
}
}
for id in self.routing.get_resources(&from)?.into_iter() {
if !included.contains(&id) {
if self.routing.remove(&id, &from)? {
updated.push(id);
}
}
}
Ok(updated)
}
/// Announce local refs for given id.
fn announce_refs(&mut self, rid: Id, namespaces: Namespaces) -> Result<(), storage::Error> {
let repo = self.storage.repository(rid)?;
let peers = self.sessions.negotiated().map(|(_, p)| p);
let timestamp = self.time();
let mut refs = BoundedVec::<_, REF_REMOTE_LIMIT>::new();
match namespaces {
Namespaces::All => {
for (remote_id, remote) in repo.remotes()?.into_iter() {
if refs.push((remote_id, remote.refs.unverified())).is_err() {
warn!(
target: "service",
"refs announcement limit ({}) exceeded, peers will see only some of your repository references",
REF_REMOTE_LIMIT,
);
break;
}
}
}
Namespaces::One(pk) => refs
.push((pk, repo.remote(&pk)?.refs.unverified()))
// SAFETY: `REF_REMOTE_LIMIT` is greater than 1, thus the bounded vec can hold at least
// one remote.
.unwrap(),
}
let msg = AnnouncementMessage::from(RefsAnnouncement {
rid,
refs,
timestamp,
});
let ann = msg.signed(&self.signer);
self.reactor.broadcast(ann, peers);
Ok(())
}
fn connect(&mut self, node: NodeId, addr: Address) -> bool {
if self.sessions.is_unconnected(&node) {
self.reactor.connect(node, addr);
return true;
}
log::warn!(target: "service", "Attempted connection to peer {node} which already has a session");
false
}
/// Return a new filter object, based on our tracking policy.
fn filter(&self) -> Filter {
if self.config.policy == tracking::Policy::Track {
// TODO: Remove bits for blocked repos.
Filter::default()
} else {
self.filter.clone()
}
}
/// Get the current time.
fn time(&self) -> Timestamp {
self.clock.as_millis()
}
////////////////////////////////////////////////////////////////////////////
// Periodic tasks
////////////////////////////////////////////////////////////////////////////
/// Announce our inventory to all connected peers.
fn announce_inventory(&mut self, inventory: Vec<Id>) -> Result<(), storage::Error> {
let time = self.time();
let inv = Message::inventory(gossip::inventory(time, inventory), &self.signer);
for (_, sess) in self.sessions.negotiated() {
self.reactor.write(sess, inv.clone());
}
Ok(())
}
fn prune_routing_entries(&mut self, now: &LocalTime) -> Result<(), routing::Error> {
let count = self.routing.len()?;
if count <= self.config.limits.routing_max_size {
return Ok(());
}
let delta = count - self.config.limits.routing_max_size;
self.routing.prune(
(*now - self.config.limits.routing_max_age).as_millis(),
Some(delta),
)?;
Ok(())
}
fn disconnect_unresponsive_peers(&mut self, now: &LocalTime) {
let stale = self
.sessions
.negotiated()
.filter(|(_, session)| session.last_active < *now - STALE_CONNECTION_TIMEOUT);
for (_, session) in stale {
self.reactor.disconnect(
session.id,
DisconnectReason::Session(session::Error::Timeout),
);
}
}
/// Ensure connection health by pinging connected peers.
fn keep_alive(&mut self, now: &LocalTime) {
let inactive_sessions = self
.sessions
.negotiated_mut()
.filter(|(_, session)| *now - session.last_active >= KEEP_ALIVE_DELTA)
.map(|(_, session)| session);
for session in inactive_sessions {
session.ping(&mut self.reactor).ok();
}
}
fn choose_addresses(&mut self) -> Vec<(NodeId, Address)> {
let sessions = self
.sessions
.values()
.filter(|s| s.is_connected() && s.link.is_outbound())
.map(|s| (s.id, s))
.collect::<HashMap<_, _>>();
let wanted = TARGET_OUTBOUND_PEERS.saturating_sub(sessions.len());
if wanted == 0 {
return Vec::new();
}
self.addresses
.entries()
.unwrap()
.filter(|(node_id, _)| !sessions.contains_key(node_id))
.take(wanted)
.map(|(n, s)| (n, s.addr))
.collect()
}
fn maintain_connections(&mut self) {
let addrs = self.choose_addresses();
if addrs.is_empty() {
debug!(target: "service", "No eligible peers available to connect to");
}
for (id, addr) in addrs {
self.connect(id, addr.clone());
}
}
}
/// Gives read access to the service state.
pub trait ServiceState {
/// Get the connected peers.
fn sessions(&self) -> &Sessions;
/// Get the current inventory.
fn inventory(&self) -> Result<Inventory, storage::Error>;
/// Get a project from storage, using the local node's key.
fn get(&self, proj: Id) -> Result<Option<Doc<Verified>>, storage::ProjectError>;
/// Get the clock.
fn clock(&self) -> &LocalTime;
/// Get the clock mutably.
fn clock_mut(&mut self) -> &mut LocalTime;
/// Get service configuration.
fn config(&self) -> &Config;
/// Get reference to routing table.
fn routing(&self) -> &dyn routing::Store;
}
impl<R, A, S, G> ServiceState for Service<R, A, S, G>
where
R: routing::Store,
G: Signer,
S: ReadStorage,
{
fn sessions(&self) -> &Sessions {
&self.sessions
}
fn inventory(&self) -> Result<Inventory, storage::Error> {
self.storage.inventory()
}
fn get(&self, proj: Id) -> Result<Option<Doc<Verified>>, storage::ProjectError> {
self.storage.get(&self.node_id(), proj)
}
fn clock(&self) -> &LocalTime {
&self.clock
}
fn clock_mut(&mut self) -> &mut LocalTime {
&mut self.clock
}
fn config(&self) -> &Config {
&self.config
}
fn routing(&self) -> &dyn routing::Store {
&self.routing
}
}
/// Disconnect reason.
#[derive(Debug)]
pub enum DisconnectReason {
/// Error while dialing the remote. This error occures before a connection is
/// even established. Errors of this kind are usually not transient.
Dial(Arc<dyn std::error::Error + Sync + Send>),
/// Error with an underlying established connection. Sometimes, reconnecting
/// after such an error is possible.
Connection(Arc<dyn std::error::Error + Sync + Send>),
/// Error with a fetch.
Fetch(FetchError),
/// Session error.
Session(session::Error),
}
impl DisconnectReason {
pub fn is_dial_err(&self) -> bool {
matches!(self, Self::Dial(_))
}
pub fn is_connection_err(&self) -> bool {
matches!(self, Self::Connection(_))
}
pub fn is_transient(&self) -> bool {
match self {
Self::Dial(_) => false,
Self::Connection(_) => true,
Self::Session(..) => false,
Self::Fetch(_) => true,
}
}
}
impl fmt::Display for DisconnectReason {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
match self {
Self::Dial(err) => write!(f, "{err}"),
Self::Connection(err) => write!(f, "{err}"),
Self::Session(err) => write!(f, "error: {err}"),
Self::Fetch(err) => write!(f, "fetch: {err}"),
}
}
}
/// Result of a project lookup.
#[derive(Debug)]
pub struct Lookup {
/// Whether the project was found locally or not.
pub local: Option<Doc<Verified>>,
/// A list of remote peers on which the project is known to exist.
pub remote: Vec<NodeId>,
}
#[derive(thiserror::Error, Debug)]
pub enum LookupError {
#[error(transparent)]
Storage(#[from] storage::Error),
#[error(transparent)]
Routing(#[from] routing::Error),
#[error(transparent)]
Project(#[from] storage::ProjectError),
}
/// Information on a peer, that we may or may not be connected to.
#[derive(Default, Debug)]
pub struct Node {
/// Last ref announcements (per project).
pub last_refs: HashMap<Id, Timestamp>,
/// Last inventory announcement.
pub last_inventory: Timestamp,
/// Last node announcement.
pub last_node: Timestamp,
}
impl Node {
/// Process a refs announcement for the given node.
/// Returns `true` if the timestamp was updated.
pub fn refs_announced(&mut self, id: Id, t: Timestamp) -> bool {
match self.last_refs.entry(id) {
Entry::Vacant(e) => {
e.insert(t);
return true;
}
Entry::Occupied(mut e) => {
let last = e.get_mut();
if t > *last {
*last = t;
return true;
}
}
}
false
}
/// Process an inventory announcement for the given node.
/// Returns `true` if the timestamp was updated.
pub fn inventory_announced(&mut self, t: Timestamp) -> bool {
if t > self.last_inventory {
self.last_inventory = t;
return true;
}
false
}
/// Process a node announcement for the given node.
/// Returns `true` if the timestamp was updated.
pub fn node_announced(&mut self, t: Timestamp) -> bool {
if t > self.last_node {
self.last_node = t;
return true;
}
false
}
}
#[derive(Debug, Clone)]
/// Holds currently (or recently) connected peers.
pub struct Sessions(AddressBook<NodeId, Session>);
impl Sessions {
pub fn new(rng: Rng) -> Self {
Self(AddressBook::new(rng))
}
/// Iterator over fully negotiated peers.
pub fn negotiated(&self) -> impl Iterator<Item = (&NodeId, &Session)> + Clone {
self.0
.iter()
.filter_map(move |(id, sess)| match &sess.state {
session::State::Connected { .. } => Some((id, sess)),
_ => None,
})
}
/// Iterator over mutable fully negotiated peers.
pub fn negotiated_mut(&mut self) -> impl Iterator<Item = (&NodeId, &mut Session)> {
self.0.iter_mut().filter(move |(_, p)| p.is_connected())
}
/// Return whether this node has a fully established session.
pub fn is_negotiated(&self, id: &NodeId) -> bool {
self.0.get(id).map(|s| s.is_connected()).unwrap_or(false)
}
/// Return whether this node can be connected to.
pub fn is_unconnected(&self, id: &NodeId) -> bool {
self.0.get(id).map(|s| s.is_disconnected()).unwrap_or(true)
}
}
impl Deref for Sessions {
type Target = AddressBook<NodeId, Session>;
fn deref(&self) -> &Self::Target {
&self.0
}
}
impl DerefMut for Sessions {
fn deref_mut(&mut self) -> &mut Self::Target {
&mut self.0
}
}
mod gossip {
use super::*;
use crate::service::filter::Filter;
#[derive(Default, Debug)]
pub struct Gossip {
received: Vec<(Timestamp, Announcement)>,
}
impl Gossip {
// TODO: Overwrite old messages from the same node or project.
// TODO: Should "time" be this node's time, or the time inside the message?
pub fn received(&mut self, ann: Announcement, time: Timestamp) {
self.received.push((time, ann));
}
pub fn filtered<'a>(
&'a self,
filter: &'a Filter,
start: Timestamp,
end: Timestamp,
) -> impl Iterator<Item = Announcement> + '_ {
self.received
.iter()
.filter(move |(t, _)| *t >= start && *t < end)
.filter(move |(_, a)| a.matches(filter))
.cloned()
.map(|(_, ann)| ann)
}
}
pub fn handshake<G: Signer, S: ReadStorage>(
now: Timestamp,
storage: &S,
signer: &G,
filter: Filter,
config: &Config,
) -> Vec<Message> {
let inventory = match storage.inventory() {
Ok(i) => i,
Err(e) => {
error!("Error getting local inventory for handshake: {}", e);
// Other than crashing the node completely, there's nothing we can do
// here besides returning an empty inventory and logging an error.
vec![]
}
};
let mut msgs = vec![
Message::init(*signer.public_key()),
Message::inventory(gossip::inventory(now, inventory), signer),
Message::subscribe(
filter,
now - SUBSCRIBE_BACKLOG_DELTA.as_millis() as u64,
Timestamp::MAX,
),
];
if let Some(m) = gossip::node(now, config) {
msgs.push(Message::node(m, signer));
};
msgs
}
pub fn node(timestamp: Timestamp, config: &Config) -> Option<NodeAnnouncement> {
let features = node::Features::SEED;
let alias = config.alias();
let addresses: BoundedVec<_, ADDRESS_LIMIT> = config
.external_addresses
.clone()
.try_into()
.expect("external addresses are within the limit");
if addresses.is_empty() {
return None;
}
Some(
NodeAnnouncement {
features,
timestamp,
alias,
addresses,
nonce: 0,
}
.solve(),
)
}
pub fn inventory(timestamp: Timestamp, inventory: Vec<Id>) -> InventoryAnnouncement {
type Inventory = BoundedVec<Id, INVENTORY_LIMIT>;
if inventory.len() > Inventory::max() {
error!(
target: "service",
"inventory announcement limit ({}) exceeded, other nodes will see only some of your projects",
inventory.len()
);
}
InventoryAnnouncement {
inventory: BoundedVec::truncate(inventory),
timestamp,
}
}
}