cerc-io/lighthouse
16
0
Fork 0
Code Issues 1 Pull Requests Packages Projects Releases Wiki Activity

Tweak head syncing (#1845)

Browse Source 
## Issue Addressed

Fixes head syncing

## Proposed Changes

- Get back to statusing peers after removing chain segments and making the peer manager deal with status according to the Sync status, preventing an old known deadlock
- Also a bug where a chain would get removed if the optimistic batch succeeds being empty

## Additional Info

Tested on Medalla and looking good
This commit is contained in:
divma 2020-11-01 23:37:39 +00:00
parent 304793a6ab
commit 6c0c050fbb
7 changed files with 134 additions and 60 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
beacon_node/eth2_libp2p/src/peer_manager/mod.rs
View File

@ -3,6 +3,7 @@
pub use self::peerdb::*;
use crate::discovery::{subnet_predicate, Discovery, DiscoveryEvent, TARGET_SUBNET_PEERS};
use crate::rpc::{GoodbyeReason, MetaData, Protocol, RPCError, RPCResponseErrorCode};
use crate::types::SyncState;
use crate::{error, metrics};
use crate::{EnrExt, NetworkConfig, NetworkGlobals, PeerId, SubnetDiscovery};
use futures::prelude::*;
@ -844,6 +845,8 @@ impl<TSpec: EthSpec> Stream for PeerManager<TSpec> {
}
}
if !matches!(self.network_globals.sync_state(), SyncState::SyncingFinalized{..}|SyncState::SyncingHead{..})
{
loop {
match self.status_peers.poll_next_unpin(cx) {
Poll::Ready(Some(Ok(peer_id))) => {
@ -856,6 +859,7 @@ impl<TSpec: EthSpec> Stream for PeerManager<TSpec> {
Poll::Ready(None) | Poll::Pending => break,
}
}
}
if !self.events.is_empty() {
return Poll::Ready(Some(self.events.remove(0)));

6
beacon_node/eth2_libp2p/src/types/sync_state.rs
View File

@ -10,6 +10,9 @@ pub enum SyncState {
/// The node is performing a long-range (batch) sync over one or many head chains.
/// In this state parent lookups are disabled.
SyncingHead { start_slot: Slot, target_slot: Slot },
/// The node has identified the need for is sync operations and is transitioning to a syncing
/// state.
SyncTransition,
/// The node is up to date with all known peers and is connected to at least one
/// fully synced peer. In this state, parent lookups are enabled.
Synced,
@ -25,6 +28,7 @@ impl PartialEq for SyncState {
(SyncState::SyncingHead { .. }, SyncState::SyncingHead { .. }) => true,
(SyncState::Synced, SyncState::Synced) => true,
(SyncState::Stalled, SyncState::Stalled) => true,
(SyncState::SyncTransition, SyncState::SyncTransition) => true,
_ => false,
}
}
@ -36,6 +40,7 @@ impl SyncState {
match self {
SyncState::SyncingFinalized { .. } => true,
SyncState::SyncingHead { .. } => true,
SyncState::SyncTransition => true,
SyncState::Synced => false,
SyncState::Stalled => false,
}
@ -54,6 +59,7 @@ impl std::fmt::Display for SyncState {
SyncState::SyncingHead { .. } => write!(f, "Syncing Head Chain"),
SyncState::Synced { .. } => write!(f, "Synced"),
SyncState::Stalled { .. } => write!(f, "Stalled"),
SyncState::SyncTransition => write!(f, "Searching syncing peers"),
}
}
}

10
beacon_node/http_api/src/lib.rs
View File

@ -330,7 +330,7 @@ pub fn serve<T: BeaconChainTypes>(
)))
}
}
SyncState::SyncingHead { .. } => Ok(()),
SyncState::SyncingHead { .. } | SyncState::SyncTransition => Ok(()),
SyncState::Synced => Ok(()),
SyncState::Stalled => Err(warp_utils::reject::not_synced(
"sync is stalled".to_string(),
@ -1231,12 +1231,12 @@ pub fn serve<T: BeaconChainTypes>(
.and(network_globals.clone())
.and_then(|network_globals: Arc<NetworkGlobals<T::EthSpec>>| {
blocking_task(move || match *network_globals.sync_state.read() {
SyncState::SyncingFinalized { .. } | SyncState::SyncingHead { .. } => {
Ok(warp::reply::with_status(
SyncState::SyncingFinalized { .. }
| SyncState::SyncingHead { .. }
| SyncState::SyncTransition => Ok(warp::reply::with_status(
warp::reply(),
warp::http::StatusCode::PARTIAL_CONTENT,
))
}
)),
SyncState::Synced => Ok(warp::reply::with_status(
warp::reply(),
warp::http::StatusCode::OK,

5
beacon_node/network/src/sync/manager.rs
View File

@ -691,10 +691,7 @@ impl<T: BeaconChainTypes> SyncManager<T> {
{
SyncState::Synced
} else if peers.advanced_peers().next().is_some() {
SyncState::SyncingHead {
start_slot: head,
target_slot: current_slot,
}
SyncState::SyncTransition
} else if peers.synced_peers().next().is_none() {
SyncState::Stalled
} else {

34
beacon_node/network/src/sync/range_sync/chain.rs
View File

@ -91,6 +91,9 @@ pub struct SyncingChain<T: BeaconChainTypes> {
/// The current processing batch, if any.
current_processing_batch: Option<BatchId>,
/// Batches validated by this chain.
validated_batches: u8,
/// A multi-threaded, non-blocking processor for applying messages to the beacon chain.
beacon_processor_send: Sender<BeaconWorkEvent<T::EthSpec>>,
@ -140,6 +143,7 @@ impl<T: BeaconChainTypes> SyncingChain<T> {
attempted_optimistic_starts: HashSet::default(),
state: ChainSyncingState::Stopped,
current_processing_batch: None,
validated_batches: 0,
beacon_processor_send,
log: log.new(o!("chain" => id)),
}
@ -155,6 +159,16 @@ impl<T: BeaconChainTypes> SyncingChain<T> {
self.id
}
/// Peers currently syncing this chain.
pub fn peers<'a>(&'a self) -> impl Iterator<Item = PeerId> + 'a {
self.peers.keys().cloned()
}
/// Progress in epochs made by the chain
pub fn validated_epochs(&self) -> u64 {
self.validated_batches as u64 * EPOCHS_PER_BATCH
}
/// Removes a peer from the chain.
/// If the peer has active batches, those are considered failed and re-requested.
pub fn remove_peer(
@ -447,6 +461,7 @@ impl<T: BeaconChainTypes> SyncingChain<T> {
self.advance_chain(network, batch_id);
// we register so that on chain switching we don't try it again
self.attempted_optimistic_starts.insert(batch_id);
self.processing_target += EPOCHS_PER_BATCH;
} else if let Some(epoch) = self.optimistic_start {
// check if this batch corresponds to an optimistic batch. In this case, we
// reject it as an optimistic candidate since the batch was empty
@ -456,10 +471,10 @@ impl<T: BeaconChainTypes> SyncingChain<T> {
false, /* do not re-request */
"batch was empty",
)?;
}
}
} else {
self.processing_target += EPOCHS_PER_BATCH;
}
}
// check if the chain has completed syncing
if self.current_processed_slot() >= self.target_head_slot {
@ -574,6 +589,7 @@ impl<T: BeaconChainTypes> SyncingChain<T> {
let removed_batches = std::mem::replace(&mut self.batches, remaining_batches);
for (id, batch) in removed_batches.into_iter() {
self.validated_batches = self.validated_batches.saturating_add(1);
// only for batches awaiting validation can we be sure the last attempt is
// right, and thus, that any different attempt is wrong
match batch.state() {
@ -1024,6 +1040,7 @@ impl<T: BeaconChainTypes> slog::KV for SyncingChain<T> {
)?;
serializer.emit_usize("batches", self.batches.len())?;
serializer.emit_usize("peers", self.peers.len())?;
serializer.emit_u8("validated_batches", self.validated_batches)?;
slog::Result::Ok(())
}
}
@ -1037,7 +1054,7 @@ impl From<WrongBatchState> for RemoveChain {
impl std::fmt::Debug for RemoveChain {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
// needed to avoid Debuggins Strings
// needed to avoid Debugging Strings
match self {
RemoveChain::ChainCompleted => f.write_str("ChainCompleted"),
RemoveChain::EmptyPeerPool => f.write_str("EmptyPeerPool"),
@ -1047,3 +1064,12 @@ impl std::fmt::Debug for RemoveChain {
}
}
}
impl RemoveChain {
pub fn is_critical(&self) -> bool {
matches!(
self,
RemoveChain::WrongBatchState(..) | RemoveChain::WrongChainState(..)
)
}
}

19
beacon_node/network/src/sync/range_sync/chain_collection.rs
View File

@ -11,7 +11,7 @@ use crate::sync::PeerSyncInfo;
use beacon_chain::{BeaconChain, BeaconChainTypes};
use eth2_libp2p::PeerId;
use fnv::FnvHashMap;
use slog::{debug, error};
use slog::{crit, debug, error};
use smallvec::SmallVec;
use std::collections::hash_map::Entry;
use std::collections::HashMap;
@ -300,7 +300,7 @@ impl<T: BeaconChainTypes> ChainCollection<T> {
match old_id {
Some(Some(old_id)) => debug!(self.log, "Switching finalized chains";
"old_id" => old_id, &chain),
None => debug!(self.log, "Syncing new chain"; &chain),
None => debug!(self.log, "Syncing new finalized chain"; &chain),
Some(None) => {
// this is the same chain. We try to advance it.
}
@ -311,8 +311,12 @@ impl<T: BeaconChainTypes> ChainCollection<T> {
if let Err(remove_reason) = chain.start_syncing(network, local_epoch, local_head_epoch)
{
if remove_reason.is_critical() {
crit!(self.log, "Chain removed while switching chains"; "chain" => new_id, "reason" => ?remove_reason);
} else {
// this happens only if sending a batch over the `network` fails a lot
error!(self.log, "Chain removed while switching chains"; "chain" => new_id, "reason" => ?remove_reason);
}
self.finalized_chains.remove(&new_id);
self.on_chain_removed(&new_id, true);
}
@ -330,6 +334,7 @@ impl<T: BeaconChainTypes> ChainCollection<T> {
) {
// Include the awaiting head peers
for (peer_id, peer_sync_info) in awaiting_head_peers.drain() {
debug!(self.log, "including head peer");
self.add_peer_or_create_chain(
local_epoch,
peer_sync_info.head_root,
@ -368,7 +373,11 @@ impl<T: BeaconChainTypes> ChainCollection<T> {
chain.start_syncing(network, local_epoch, local_head_epoch)
{
self.head_chains.remove(&id);
if remove_reason.is_critical() {
crit!(self.log, "Chain removed while switching head chains"; "chain" => id, "reason" => ?remove_reason);
} else {
error!(self.log, "Chain removed while switching head chains"; "chain" => id, "reason" => ?remove_reason);
}
} else {
syncing_chains.push(id);
}
@ -482,7 +491,11 @@ impl<T: BeaconChainTypes> ChainCollection<T> {
debug_assert_eq!(chain.target_head_root, target_head_root);
debug_assert_eq!(chain.target_head_slot, target_head_slot);
if let Err(remove_reason) = chain.add_peer(network, peer) {
debug!(self.log, "Chain removed after adding peer"; "chain" => id, "reason" => ?remove_reason);
if remove_reason.is_critical() {
error!(self.log, "Chain removed after adding peer"; "chain" => id, "reason" => ?remove_reason);
} else {
error!(self.log, "Chain removed after adding peer"; "chain" => id, "reason" => ?remove_reason);
}
let chain = entry.remove();
self.on_chain_removed(&id, chain.is_syncing());
}

96
beacon_node/network/src/sync/range_sync/range.rs
View File

@ -19,7 +19,7 @@
//! need to be downloaded.
//!
//! A few interesting notes about finalized chain syncing:
//! - Only one finalized chain can sync at a time.
//! - Only one finalized chain can sync at a time
//! - The finalized chain with the largest peer pool takes priority.
//! - As one finalized chain completes, others are checked to see if we they can be continued,
//! otherwise they are removed.
@ -39,7 +39,7 @@
//! Each chain is downloaded in batches of blocks. The batched blocks are processed sequentially
//! and further batches are requested as current blocks are being processed.
use super::chain::ChainId;
use super::chain::{ChainId, RemoveChain, SyncingChain};
use super::chain_collection::ChainCollection;
use super::sync_type::RangeSyncType;
use crate::beacon_processor::WorkEvent as BeaconWorkEvent;
@ -49,7 +49,7 @@ use crate::sync::PeerSyncInfo;
use crate::sync::RequestId;
use beacon_chain::{BeaconChain, BeaconChainTypes};
use eth2_libp2p::PeerId;
use slog::{debug, error, trace};
use slog::{crit, debug, error, trace};
use std::collections::HashMap;
use std::sync::Arc;
use tokio::sync::mpsc;
@ -121,7 +121,8 @@ impl<T: BeaconChainTypes> RangeSync<T> {
.finalized_epoch
.start_slot(T::EthSpec::slots_per_epoch());
// NOTE: A peer that has been re-status'd may now exist in multiple finalized chains.
// NOTE: A peer that has been re-status'd may now exist in multiple finalized chains. This
// is OK since we since only one finalized chain at a time.
// determine which kind of sync to perform and set up the chains
match RangeSyncType::new(&self.beacon_chain, &local_info, &remote_info) {
@ -208,22 +209,22 @@ impl<T: BeaconChainTypes> RangeSync<T> {
chain.on_block_response(network, batch_id, &peer_id, request_id, beacon_block)
}) {
Ok((removed_chain, sync_type)) => {
if let Some(_removed_chain) = removed_chain {
debug!(self.log, "Chain removed after block response"; "sync_type" => ?sync_type, "chain_id" => chain_id);
// update the state of the collection
self.chains.update(
if let Some((removed_chain, remove_reason)) = removed_chain {
self.on_chain_removed(
removed_chain,
sync_type,
remove_reason,
network,
&mut self.awaiting_head_peers,
&self.beacon_processor_send,
"block response",
);
}
}
Err(_) => {
debug!(self.log, "BlocksByRange response for removed chain"; "chain" => chain_id)
trace!(self.log, "BlocksByRange response for removed chain"; "chain" => chain_id)
}
}
} else {
debug!(self.log, "Response/Error for non registered request"; "request_id" => request_id)
trace!(self.log, "Response/Error for non registered request"; "request_id" => request_id)
}
}
@ -241,17 +242,18 @@ impl<T: BeaconChainTypes> RangeSync<T> {
Ok((None, _sync_type)) => {
// Chain was found and not removed
}
Ok((Some(_removed_chain), sync_type)) => {
debug!(self.log, "Chain removed after processing result"; "chain" => chain_id, "sync_type" => ?sync_type);
self.chains.update(
Ok((Some((removed_chain, remove_reason)), sync_type)) => {
self.on_chain_removed(
removed_chain,
sync_type,
remove_reason,
network,
&mut self.awaiting_head_peers,
&self.beacon_processor_send,
"batch processing result",
);
}
Err(_) => {
debug!(self.log, "BlocksByRange response for removed chain"; "chain" => chain_id)
trace!(self.log, "BlocksByRange response for removed chain"; "chain" => chain_id)
}
}
}
@ -279,14 +281,16 @@ impl<T: BeaconChainTypes> RangeSync<T> {
.chains
.call_all(|chain| chain.remove_peer(peer_id, network))
{
debug!(self.log, "Chain removed after removing peer"; "sync_type" => ?sync_type, "chain" => removed_chain.get_id(), "reason" => ?remove_reason);
self.on_chain_removed(
removed_chain,
sync_type,
remove_reason,
network,
"peer removed",
);
// update the state of the collection
}
self.chains.update(
network,
&mut self.awaiting_head_peers,
&self.beacon_processor_send,
);
}
/// An RPC error has occurred.
@ -307,7 +311,40 @@ impl<T: BeaconChainTypes> RangeSync<T> {
}) {
Ok((removed_chain, sync_type)) => {
if let Some((removed_chain, remove_reason)) = removed_chain {
debug!(self.log, "Chain removed on rpc error"; "sync_type" => ?sync_type, "chain" => removed_chain.get_id(), "reason" => ?remove_reason);
self.on_chain_removed(
removed_chain,
sync_type,
remove_reason,
network,
"RPC error",
);
}
}
Err(_) => {
trace!(self.log, "BlocksByRange response for removed chain"; "chain" => chain_id)
}
}
} else {
trace!(self.log, "Response/Error for non registered request"; "request_id" => request_id)
}
}
fn on_chain_removed(
&mut self,
chain: SyncingChain<T>,
sync_type: RangeSyncType,
remove_reason: RemoveChain,
network: &mut SyncNetworkContext<T::EthSpec>,
op: &'static str,
) {
if remove_reason.is_critical() {
crit!(self.log, "Chain removed"; "sync_type" => ?sync_type, &chain, "reason" => ?remove_reason, "op" => op);
} else {
debug!(self.log, "Chain removed"; "sync_type" => ?sync_type, &chain, "reason" => ?remove_reason, "op" => op);
}
network.status_peers(self.beacon_chain.clone(), chain.peers());
// update the state of the collection
self.chains.update(
network,
@ -315,13 +352,4 @@ impl<T: BeaconChainTypes> RangeSync<T> {
&self.beacon_processor_send,
);
}
}
Err(_) => {
debug!(self.log, "BlocksByRange response for removed chain"; "chain" => chain_id)
}
}
} else {
debug!(self.log, "Response/Error for non registered request"; "request_id" => request_id)
}
}
}