Skip to content

Commit 2f29569

Browse files
committed
Optimize ChannelMonitor persistence on block connections.
Currently, every block connection triggers the persistence of all ChannelMonitors with an updated best_block. This approach poses challenges for large node operators managing thousands of channels. Furthermore, it leads to a thundering herd problem (https://en.wikipedia.org/wiki/Thundering_herd_problem), overwhelming the storage with simultaneous requests. To address this issue, we now persist ChannelMonitors at a regular cadence, spreading their persistence across blocks to mitigate spikes in write operations. Outcome: After doing this, Ldk's IO footprint should be reduced by ~50 times. The processing time required to sync each block will be significantly reduced, particularly for nodes with 1000s of channels, as write latency plays a significant role in this process. As a result, the Node/ChainMonitor will be blocked for a shorter duration, leading to further efficiency gains.
1 parent 1912d8d commit 2f29569

File tree

3 files changed

+114
-12
lines changed

3 files changed

+114
-12
lines changed

lightning/src/chain/chainmonitor.rs

Lines changed: 101 additions & 12 deletions
Original file line numberDiff line numberDiff line change
@@ -44,6 +44,7 @@ use crate::prelude::*;
4444
use crate::sync::{RwLock, RwLockReadGuard, Mutex, MutexGuard};
4545
use core::ops::Deref;
4646
use core::sync::atomic::{AtomicUsize, Ordering};
47+
use bitcoin::hashes::Hash;
4748
use bitcoin::secp256k1::PublicKey;
4849

4950
/// `Persist` defines behavior for persisting channel monitors: this could mean
@@ -263,7 +264,7 @@ where C::Target: chain::Filter,
263264
for funding_outpoint in funding_outpoints.iter() {
264265
let monitor_lock = self.monitors.read().unwrap();
265266
if let Some(monitor_state) = monitor_lock.get(funding_outpoint) {
266-
if self.update_monitor_with_chain_data(header, txdata, &process, funding_outpoint, &monitor_state).is_err() {
267+
if self.update_monitor_with_chain_data(header, best_height, txdata, &process, funding_outpoint, &monitor_state).is_err() {
267268
// Take the monitors lock for writing so that we poison it and any future
268269
// operations going forward fail immediately.
269270
core::mem::drop(monitor_lock);
@@ -278,7 +279,7 @@ where C::Target: chain::Filter,
278279
let monitor_states = self.monitors.write().unwrap();
279280
for (funding_outpoint, monitor_state) in monitor_states.iter() {
280281
if !funding_outpoints.contains(funding_outpoint) {
281-
if self.update_monitor_with_chain_data(header, txdata, &process, funding_outpoint, &monitor_state).is_err() {
282+
if self.update_monitor_with_chain_data(header, best_height, txdata, &process, funding_outpoint, &monitor_state).is_err() {
282283
log_error!(self.logger, "{}", err_str);
283284
panic!("{}", err_str);
284285
}
@@ -297,14 +298,23 @@ where C::Target: chain::Filter,
297298
}
298299

299300
fn update_monitor_with_chain_data<FN>(
300-
&self, header: &Header, txdata: &TransactionData, process: FN, funding_outpoint: &OutPoint,
301-
monitor_state: &MonitorHolder<ChannelSigner>
301+
&self, header: &Header, best_height: Option<u32>, txdata: &TransactionData, process: FN, funding_outpoint: &OutPoint,
302+
monitor_state: &MonitorHolder<ChannelSigner>,
302303
) -> Result<(), ()> where FN: Fn(&ChannelMonitor<ChannelSigner>, &TransactionData) -> Vec<TransactionOutputs> {
303304
let monitor = &monitor_state.monitor;
304305
let logger = WithChannelMonitor::from(&self.logger, &monitor, None);
305-
let mut txn_outputs;
306-
{
307-
txn_outputs = process(monitor, txdata);
306+
307+
let mut txn_outputs = process(monitor, txdata);
308+
309+
let get_partition_key = |funding_outpoint: &OutPoint| {
310+
let funding_txid_hash = funding_outpoint.txid.to_raw_hash();
311+
let funding_txid_hash_bytes = funding_txid_hash.as_byte_array();
312+
let funding_txid_u32 = u32::from_be_bytes([funding_txid_hash_bytes[0], funding_txid_hash_bytes[1], funding_txid_hash_bytes[2], funding_txid_hash_bytes[3]]);
313+
funding_txid_u32.wrapping_add(best_height.unwrap_or_default())
314+
};
315+
const CHAINSYNC_MONITOR_PARTITION_FACTOR: u32 = 50; // ~ 8hours
316+
let has_pending_claims = monitor_state.monitor.has_pending_claims();
317+
if has_pending_claims || get_partition_key(funding_outpoint) % CHAINSYNC_MONITOR_PARTITION_FACTOR == 0 {
308318
log_trace!(logger, "Syncing Channel Monitor for channel {}", log_funding_info!(monitor));
309319
match self.persister.update_persisted_channel(*funding_outpoint, None, monitor) {
310320
ChannelMonitorUpdateStatus::Completed =>
@@ -313,10 +323,10 @@ where C::Target: chain::Filter,
313323
),
314324
ChannelMonitorUpdateStatus::InProgress => {
315325
log_trace!(logger, "Channel Monitor sync for channel {} in progress.", log_funding_info!(monitor));
316-
},
326+
}
317327
ChannelMonitorUpdateStatus::UnrecoverableError => {
318328
return Err(());
319-
},
329+
}
320330
}
321331
}
322332

@@ -870,14 +880,17 @@ impl<ChannelSigner: EcdsaChannelSigner, C: Deref, T: Deref, F: Deref, L: Deref,
870880

871881
#[cfg(test)]
872882
mod tests {
873-
use crate::check_added_monitors;
883+
use crate::{check_added_monitors, check_closed_event};
874884
use crate::{expect_payment_path_successful, get_event_msg};
875885
use crate::{get_htlc_update_msgs, get_revoke_commit_msgs};
876886
use crate::chain::{ChannelMonitorUpdateStatus, Watch};
877-
use crate::events::{Event, MessageSendEvent, MessageSendEventsProvider};
887+
use crate::chain::channelmonitor::ANTI_REORG_DELAY;
888+
use crate::events::{ClosureReason, Event, MessageSendEvent, MessageSendEventsProvider};
878889
use crate::ln::functional_test_utils::*;
879890
use crate::ln::msgs::ChannelMessageHandler;
880891

892+
const CHAINSYNC_MONITOR_PARTITION_FACTOR: u32 = 50;
893+
881894
#[test]
882895
fn test_async_ooo_offchain_updates() {
883896
// Test that if we have multiple offchain updates being persisted and they complete
@@ -983,6 +996,79 @@ mod tests {
983996
check_added_monitors!(nodes[0], 1);
984997
}
985998

999+
#[test]
1000+
fn test_chainsync_triggers_distributed_monitor_persistence() {
1001+
let chanmon_cfgs = create_chanmon_cfgs(3);
1002+
let node_cfgs = create_node_cfgs(3, &chanmon_cfgs);
1003+
let node_chanmgrs = create_node_chanmgrs(3, &node_cfgs, &[None, None, None]);
1004+
let nodes = create_network(3, &node_cfgs, &node_chanmgrs);
1005+
1006+
// Use FullBlockViaListen to avoid duplicate calls to process_chain_data and skips_blocks() in
1007+
// case of other connect_styles.
1008+
*nodes[0].connect_style.borrow_mut() = ConnectStyle::FullBlockViaListen;
1009+
*nodes[1].connect_style.borrow_mut() = ConnectStyle::FullBlockViaListen;
1010+
*nodes[2].connect_style.borrow_mut() = ConnectStyle::FullBlockViaListen;
1011+
1012+
let _channel_1 = create_announced_chan_between_nodes(&nodes, 0, 1).2;
1013+
let channel_2 = create_announced_chan_between_nodes_with_value(&nodes, 0, 2, 1_000_000, 0).2;
1014+
1015+
chanmon_cfgs[0].persister.chain_sync_monitor_persistences.lock().unwrap().clear();
1016+
chanmon_cfgs[1].persister.chain_sync_monitor_persistences.lock().unwrap().clear();
1017+
chanmon_cfgs[2].persister.chain_sync_monitor_persistences.lock().unwrap().clear();
1018+
1019+
connect_blocks(&nodes[0], CHAINSYNC_MONITOR_PARTITION_FACTOR * 2);
1020+
connect_blocks(&nodes[1], CHAINSYNC_MONITOR_PARTITION_FACTOR * 2);
1021+
connect_blocks(&nodes[2], CHAINSYNC_MONITOR_PARTITION_FACTOR * 2);
1022+
1023+
// Connecting [`DEFAULT_CHAINSYNC_PARTITION_FACTOR`] * 2 blocks should trigger only 2 writes
1024+
// per monitor/channel.
1025+
assert_eq!(2 * 2, chanmon_cfgs[0].persister.chain_sync_monitor_persistences.lock().unwrap().len());
1026+
assert_eq!(2, chanmon_cfgs[1].persister.chain_sync_monitor_persistences.lock().unwrap().len());
1027+
assert_eq!(2, chanmon_cfgs[2].persister.chain_sync_monitor_persistences.lock().unwrap().len());
1028+
1029+
// Test that monitors with pending_claims are persisted on every block.
1030+
// Now, close channel_2 i.e. b/w node-0 and node-2 to create pending_claim in node[0].
1031+
nodes[0].node.force_close_broadcasting_latest_txn(&channel_2, &nodes[2].node.get_our_node_id(), "Channel force-closed".to_string()).unwrap();
1032+
check_closed_event!(&nodes[0], 1, ClosureReason::HolderForceClosed { broadcasted_latest_txn: Some(true) }, false,
1033+
[nodes[2].node.get_our_node_id()], 1000000);
1034+
check_closed_broadcast(&nodes[0], 1, true);
1035+
let close_tx = nodes[0].tx_broadcaster.txn_broadcasted.lock().unwrap().split_off(0);
1036+
assert_eq!(close_tx.len(), 1);
1037+
1038+
mine_transaction(&nodes[2], &close_tx[0]);
1039+
check_added_monitors(&nodes[2], 1);
1040+
check_closed_broadcast(&nodes[2], 1, true);
1041+
check_closed_event!(&nodes[2], 1, ClosureReason::CommitmentTxConfirmed, false,
1042+
[nodes[0].node.get_our_node_id()], 1000000);
1043+
1044+
chanmon_cfgs[0].persister.chain_sync_monitor_persistences.lock().unwrap().clear();
1045+
chanmon_cfgs[2].persister.chain_sync_monitor_persistences.lock().unwrap().clear();
1046+
1047+
// For channel_2, there should be a monitor write for every block connection.
1048+
// We connect [`DEFAULT_CHAINSYNC_MONITOR_PARTITION_FACTOR`] blocks since we don't know when
1049+
// channel_1 monitor persistence will occur, with [`DEFAULT_CHAINSYNC_MONITOR_PARTITION_FACTOR`]
1050+
// it will be persisted exactly once.
1051+
connect_blocks(&nodes[0], CHAINSYNC_MONITOR_PARTITION_FACTOR);
1052+
connect_blocks(&nodes[2], CHAINSYNC_MONITOR_PARTITION_FACTOR);
1053+
1054+
// DEFAULT_CHAINSYNC_MONITOR_PARTITION_FACTOR writes for channel_2 due to pending_claim, 1 for
1055+
// channel_1
1056+
assert_eq!((CHAINSYNC_MONITOR_PARTITION_FACTOR + 1) as usize, chanmon_cfgs[0].persister.chain_sync_monitor_persistences.lock().unwrap().len());
1057+
// For node[2], there is no pending_claim
1058+
assert_eq!(1, chanmon_cfgs[2].persister.chain_sync_monitor_persistences.lock().unwrap().len());
1059+
1060+
// Confirm claim for node[0] with ANTI_REORG_DELAY and reset monitor write counter.
1061+
mine_transaction(&nodes[0], &close_tx[0]);
1062+
connect_blocks(&nodes[0], ANTI_REORG_DELAY - 1);
1063+
check_added_monitors(&nodes[0], 1);
1064+
chanmon_cfgs[0].persister.chain_sync_monitor_persistences.lock().unwrap().clear();
1065+
1066+
// Again connect 1 full cycle of DEFAULT_CHAINSYNC_MONITOR_PARTITION_FACTOR blocks, it should only
1067+
// result in 1 write per monitor/channel.
1068+
connect_blocks(&nodes[0], CHAINSYNC_MONITOR_PARTITION_FACTOR);
1069+
assert_eq!(2, chanmon_cfgs[0].persister.chain_sync_monitor_persistences.lock().unwrap().len());
1070+
}
1071+
9861072
#[test]
9871073
#[cfg(feature = "std")]
9881074
fn update_during_chainsync_poisons_channel() {
@@ -991,12 +1077,15 @@ mod tests {
9911077
let node_chanmgrs = create_node_chanmgrs(2, &node_cfgs, &[None, None]);
9921078
let nodes = create_network(2, &node_cfgs, &node_chanmgrs);
9931079
create_announced_chan_between_nodes(&nodes, 0, 1);
1080+
*nodes[0].connect_style.borrow_mut() = ConnectStyle::FullBlockViaListen;
9941081

9951082
chanmon_cfgs[0].persister.set_update_ret(ChannelMonitorUpdateStatus::UnrecoverableError);
9961083

9971084
assert!(std::panic::catch_unwind(|| {
9981085
// Returning an UnrecoverableError should always panic immediately
999-
connect_blocks(&nodes[0], 1);
1086+
// Connecting [`DEFAULT_CHAINSYNC_PARTITION_FACTOR`] blocks so that we trigger some persistence
1087+
// after accounting for block-height based partitioning/distribution.
1088+
connect_blocks(&nodes[0], CHAINSYNC_MONITOR_PARTITION_FACTOR);
10001089
}).is_err());
10011090
assert!(std::panic::catch_unwind(|| {
10021091
// ...and also poison our locks causing later use to panic as well

lightning/src/chain/channelmonitor.rs

Lines changed: 6 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1812,6 +1812,12 @@ impl<Signer: EcdsaChannelSigner> ChannelMonitor<Signer> {
18121812
);
18131813
}
18141814

1815+
/// Returns true if the monitor has pending claim requests that are not fully confirmed yet.
1816+
pub fn has_pending_claims(&self) -> bool
1817+
{
1818+
self.inner.lock().unwrap().onchain_tx_handler.has_pending_claims()
1819+
}
1820+
18151821
/// Triggers rebroadcasts of pending claims from a force-closed channel after a transaction
18161822
/// signature generation failure.
18171823
pub fn signer_unblocked<B: Deref, F: Deref, L: Deref>(

lightning/src/chain/onchaintx.rs

Lines changed: 7 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -535,6 +535,13 @@ impl<ChannelSigner: EcdsaChannelSigner> OnchainTxHandler<ChannelSigner> {
535535
}
536536
}
537537

538+
/// Returns true if we are currently tracking any pending claim requests that are not fully
539+
/// confirmed yet.
540+
pub(super) fn has_pending_claims(&self) -> bool
541+
{
542+
self.pending_claim_requests.len() != 0
543+
}
544+
538545
/// Lightning security model (i.e being able to redeem/timeout HTLC or penalize counterparty
539546
/// onchain) lays on the assumption of claim transactions getting confirmed before timelock
540547
/// expiration (CSV or CLTV following cases). In case of high-fee spikes, claim tx may get stuck

0 commit comments

Comments
 (0)