Add BackgroundProcessor for ChannelManager persistence and other

Other includes calling timer_chan_freshness_every_minute() and in the
future, possibly persisting channel graph data.

This struct is suitable for things that need to happen periodically and
can happen in the background.

Author: valentinewallace

Cargo.toml

@@ -5,6 +5,7 @@ members = [
     "lightning-block-sync",
     "lightning-net-tokio",
     "lightning-persister",
+    "background-processor",
 ]
 
 # Our tests do actual crypo and lots of work, the tradeoff for -O1 is well worth it.

background-processor/Cargo.toml

@@ -0,0 +1,15 @@
+[package]
+name = "background-processor"
+version = "0.1.0"
+authors = ["Valentine Wallace <[email protected]>"]
+edition = "2018"
+
+# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
+
+[dependencies]
+bitcoin = "0.24"
+lightning = { version = "0.0.12", path = "../lightning", features = ["allow_wallclock_use"] }
+lightning-persister = { version = "0.0.1", path = "../lightning-persister" }
+
+[dev-dependencies]
+lightning = { version = "0.0.12", path = "../lightning", features = ["_test_utils"] }

background-processor/src/lib.rs

@@ -0,0 +1,347 @@
+use lightning::chain;
+use lightning::chain::chaininterface::{BroadcasterInterface, FeeEstimator};
+use lightning::chain::keysinterface::{ChannelKeys, KeysInterface};
+use lightning::ln::channelmanager::ChannelManager;
+use lightning::{log_internal, log_trace};
+use lightning::util::logger::Logger;
+use lightning::util::ser::Writeable;
+use std::sync::Arc;
+use std::sync::atomic::{AtomicBool, Ordering};
+use std::thread;
+use std::thread::JoinHandle;
+use std::time::{Duration, Instant};
+
+/// BackgroundProcessor takes care of tasks that (1) need to happen periodically to keep
+/// Rust-Lightning running properly, and (2) either can or should be run in the background. Its
+/// responsibilities are:
+/// * Monitoring whether the ChannelManager needs to be re-persisted to disk, and if so,
+///   writing it to disk/backups by invoking the callback given to it at startup.
+///   ChannelManager persistence should be done in the background.
+/// * Calling `ChannelManager::timer_chan_freshness_every_min()` every minute (can be done in the
+///   background).
+///
+/// Note that if ChannelManager persistence fails and the persisted manager becomes out-of-date,
+/// then there is a risk of channels force-closing on startup when the manager realizes it's
+/// outdated. However, as long as `ChannelMonitor` backups are sound, no funds besides those used
+/// for unilateral chain closure fees are at risk.
+pub struct BackgroundProcessor {
+	stop_thread: Arc<AtomicBool>,
+	/// May be used to retrieve and handle the error if `BackgroundProcessor`'s thread
+	/// panics.
+	pub thread_handle: JoinHandle<()>
+}
+
+// Re-export these modules for the logging macros.
+mod util {
+	pub(crate) mod logger {
+		pub(crate) use lightning::util::logger::Record;
+	}
+}
+
+#[cfg(not(test))]
+const CHAN_FRESHNESS_TIMER: u64 = 60;
+#[cfg(test)]
+const CHAN_FRESHNESS_TIMER: u64 = 1;
+
+impl BackgroundProcessor {
+	/// Start the background thread that takes care of responsibilities (enumerated in the top-level
+	/// documentation). Marked as `must_use` because otherwise the result is dropped immediately,
+	/// resulting in the thread being terminated.
+	/// Important note: this thread will panic if invoking `persist_manager` results in an error (and
+	/// `start()` will need to be called again to restart the `BackgroundProcessor`).
+	/// There are 3 main options for handling this panic:
+	/// * wait on [`thread_handle`]'s `join()`, handle the error
+	/// * [configure] to abort on panic
+	/// * write a custom `persist_manager` to handle the error so it never gets returned to
+	///   `BackgroundProcessor`.
+	///
+	/// `persist_manager` is responsible for writing out the `ChannelManager` to disk, and/or uploading
+	/// to one or more backup services. See [`ChannelManager::write`] for writing out a `ChannelManager`.
+	/// See [`FilesystemPersister::persist_manager`] for Rust-Lightning's provided implementation.
+	///
+	/// [`thread_handle`]: struct.BackgroundProcessor.html#structfield.thread_handle
+	/// [configure]: https://doc.rust-lang.org/edition-guide/rust-2018/error-handling-and-panics/aborting-on-panic.html
+	/// [`ChannelManager::write`]: ../lightning/ln/channelmanager/struct.ChannelManager.html#method.write
+	/// [`FilesystemPersister::persist_manager`]: ../lightning_persister/struct.FilesystemPersister.html#impl
+	#[must_use]
+	pub fn start<PM, ChanSigner, M, T, K, F, L>(persist_manager: PM, manager: Arc<ChannelManager<ChanSigner, Arc<M>, Arc<T>, Arc<K>, Arc<F>, Arc<L>>>, logger: Arc<L>) -> Self
+	where ChanSigner: 'static + ChannelKeys + Writeable,
+	      M: 'static + chain::Watch<Keys=ChanSigner>,
+	      T: 'static + BroadcasterInterface,
+	      K: 'static + KeysInterface<ChanKeySigner=ChanSigner>,
+	      F: 'static + FeeEstimator,
+	      L: 'static + Logger,
+	      PM: 'static + Send + Fn(Arc<ChannelManager<ChanSigner, Arc<M>, Arc<T>, Arc<K>, Arc<F>, Arc<L>>>) -> Result<(), std::io::Error>,
+	{
+		let stop_thread = Arc::new(AtomicBool::new(false));
+		let stop_thread_clone = stop_thread.clone();
+		let handle = thread::spawn(move || {
+			let mut current_time = Instant::now();
+			loop {
+				let updates_available = manager.wait_timeout(Duration::from_millis(100));
+				if updates_available {
+					if let Err(e) = persist_manager(manager.clone()) {
+						panic!("Errored persisting manager: {}", e);
+					};
+				}
+				// If we see that the thread has been stopped, exit now.
+				if stop_thread.load(Ordering::Acquire) == true {
+					log_trace!(logger, "Terminating background processor.");
+					break;
+				}
+				if current_time.elapsed().as_secs() > CHAN_FRESHNESS_TIMER {
+					log_trace!(logger, "Calling manager's timer_chan_freshness_every_min");
+					manager.timer_chan_freshness_every_min();
+					current_time = Instant::now();
+				}
+			}
+		});
+		Self {
+			stop_thread: stop_thread_clone,
+			thread_handle: handle
+		}
+	}
+
+	/// Stop `BackgroundProcessor`'s thread.
+	pub fn stop(&self) {
+		self.stop_thread.store(true, Ordering::Release)
+	}
+}
+
+#[cfg(test)]
+mod tests {
+	use bitcoin::blockdata::constants::genesis_block;
+	use bitcoin::blockdata::transaction::{Transaction, TxOut};
+	use bitcoin::network::constants::Network;
+	use lightning::chain;
+	use lightning::chain::chaininterface::{BroadcasterInterface, FeeEstimator};
+	use lightning::chain::chainmonitor;
+	use lightning::chain::keysinterface::{ChannelKeys, InMemoryChannelKeys, KeysInterface, KeysManager};
+	use lightning::chain::transaction::OutPoint;
+	use lightning::get_event_msg;
+	use lightning::ln::channelmanager::{ChannelManager, SimpleArcChannelManager};
+	use lightning::ln::features::InitFeatures;
+	use lightning::ln::msgs::ChannelMessageHandler;
+	use lightning::util::config::UserConfig;
+	use lightning::util::events::{Event, EventsProvider, MessageSendEventsProvider, MessageSendEvent};
+	use lightning::util::logger::Logger;
+	use lightning::util::ser::Writeable;
+	use lightning::util::test_utils;
+	use lightning_persister::FilesystemPersister;
+	use std::fs;
+	use std::path::PathBuf;
+	use std::sync::{Arc, Mutex};
+	use std::time::Duration;
+	use super::BackgroundProcessor;
+
+	type ChainMonitor = chainmonitor::ChainMonitor<InMemoryChannelKeys, Arc<test_utils::TestChainSource>, Arc<test_utils::TestBroadcaster>, Arc<test_utils::TestFeeEstimator>, Arc<test_utils::TestLogger>, Arc<FilesystemPersister>>;
+
+	struct Node {
+		node: SimpleArcChannelManager<ChainMonitor, test_utils::TestBroadcaster, test_utils::TestFeeEstimator, test_utils::TestLogger>,
+		persister: Arc<FilesystemPersister>,
+		logger: Arc<test_utils::TestLogger>,
+	}
+
+	impl Drop for Node {
+		fn drop(&mut self) {
+			let data_dir = self.persister.get_data_dir();
+			match fs::remove_dir_all(data_dir.clone()) {
+				Err(e) => println!("Failed to remove test persister directory {}: {}", data_dir, e),
+				_ => {}
+			}
+		}
+	}
+
+	fn get_full_filepath(filepath: String, filename: String) -> String {
+		let mut path = PathBuf::from(filepath);
+		path.push(filename);
+		path.to_str().unwrap().to_string()
+	}
+
+	fn create_nodes(num_nodes: usize, persist_dir: String) -> Vec<Node> {
+		let mut nodes = Vec::new();
+		for i in 0..num_nodes {
+			let tx_broadcaster = Arc::new(test_utils::TestBroadcaster{txn_broadcasted: Mutex::new(Vec::new())});
+			let fee_estimator = Arc::new(test_utils::TestFeeEstimator { sat_per_kw: 253 });
+			let chain_source = Arc::new(test_utils::TestChainSource::new(Network::Testnet));
+			let logger = Arc::new(test_utils::TestLogger::with_id(format!("node {}", i)));
+			let persister = Arc::new(FilesystemPersister::new(format!("{}_persister_{}", persist_dir, i)));
+			let seed = [i as u8; 32];
+			let network = Network::Testnet;
+			let now = Duration::from_secs(genesis_block(network).header.time as u64);
+			let keys_manager = Arc::new(KeysManager::new(&seed, network, now.as_secs(), now.subsec_nanos()));
+			let chain_monitor = Arc::new(chainmonitor::ChainMonitor::new(Some(chain_source.clone()), tx_broadcaster.clone(), logger.clone(), fee_estimator.clone(), persister.clone()));
+			let manager = Arc::new(ChannelManager::new(Network::Testnet, fee_estimator.clone(), chain_monitor.clone(), tx_broadcaster, logger.clone(), keys_manager.clone(), UserConfig::default(), i));
+			let node = Node { node: manager, persister, logger };
+			nodes.push(node);
+		}
+		nodes
+	}
+
+	macro_rules! open_channel {
+		($node_a: expr, $node_b: expr, $channel_value: expr) => {{
+			$node_a.node.create_channel($node_b.node.get_our_node_id(), $channel_value, 100, 42, None).unwrap();
+			$node_b.node.handle_open_channel(&$node_a.node.get_our_node_id(), InitFeatures::known(), &get_event_msg!($node_a, MessageSendEvent::SendOpenChannel, $node_b.node.get_our_node_id()));
+			$node_a.node.handle_accept_channel(&$node_b.node.get_our_node_id(), InitFeatures::known(), &get_event_msg!($node_b, MessageSendEvent::SendAcceptChannel, $node_a.node.get_our_node_id()));
+			let events = $node_a.node.get_and_clear_pending_events();
+			assert_eq!(events.len(), 1);
+			let (temporary_channel_id, tx, funding_output) = match events[0] {
+				Event::FundingGenerationReady { ref temporary_channel_id, ref channel_value_satoshis, ref output_script, user_channel_id } => {
+					assert_eq!(*channel_value_satoshis, $channel_value);
+					assert_eq!(user_channel_id, 42);
+
+					let tx = Transaction { version: 1 as i32, lock_time: 0, input: Vec::new(), output: vec![TxOut {
+						value: *channel_value_satoshis, script_pubkey: output_script.clone(),
+					}]};
+					let funding_outpoint = OutPoint { txid: tx.txid(), index: 0 };
+					(*temporary_channel_id, tx, funding_outpoint)
+				},
+				_ => panic!("Unexpected event"),
+			};
+
+			$node_a.node.funding_transaction_generated(&temporary_channel_id, funding_output);
+			$node_b.node.handle_funding_created(&$node_a.node.get_our_node_id(), &get_event_msg!($node_a, MessageSendEvent::SendFundingCreated, $node_b.node.get_our_node_id()));
+			$node_a.node.handle_funding_signed(&$node_b.node.get_our_node_id(), &get_event_msg!($node_b, MessageSendEvent::SendFundingSigned, $node_a.node.get_our_node_id()));
+			tx
+		}}
+	}
+
+	#[test]
+	fn test_background_processor() {
+		// Test that when a new channel is created, the ChannelManager needs to be re-persisted with
+		// updates. Also test that when new updates are available, the manager signals that it needs
+		// re-persistence and is successfully re-persisted.
+		let nodes = create_nodes(2, "test_background_processor".to_string());
+
+		// Initiate the background processors to watch each node.
+		let data_dir_0 = nodes[0].persister.get_data_dir();
+		let data_dir_1 = nodes[1].persister.get_data_dir();
+		let callback_0 = move |node| FilesystemPersister::persist_manager(data_dir_0.clone(), node);
+		let callback_1 = move |node| FilesystemPersister::persist_manager(data_dir_1.clone(), node);
+		let _processor_0 = BackgroundProcessor::start(callback_0, nodes[0].node.clone(), nodes[0].logger.clone());
+		let _processor_1 = BackgroundProcessor::start(callback_1, nodes[1].node.clone(), nodes[1].logger.clone());
+
+		// Go through the channel creation process until each node should have something persisted.
+		let tx = open_channel!(nodes[0], nodes[1], 100000);
+
+		let mut done_persisting = false;
+		macro_rules! check_persisted_data {
+			($node: expr, $filepath: expr) => {
+				let bytes = loop {
+					match std::fs::read($filepath) {
+						Ok(bytes) => break bytes,
+						Err(_) => continue
+					}
+				};
+				let mut expected_bytes = Vec::new();
+				assert!($node.write(&mut expected_bytes).is_ok());
+				if bytes == expected_bytes {
+					done_persisting = true;
+				}
+			}
+		}
+
+		// Check that the initial channel manager data is persisted as expected.
+		let filepath_0 = get_full_filepath("test_background_processor_persister_0".to_string(), "manager".to_string());
+		let filepath_1 = get_full_filepath("test_background_processor_persister_1".to_string(), "manager".to_string());
+		loop {
+			check_persisted_data!(nodes[0].node, filepath_0.clone());
+			if done_persisting {
+				// Check that eventually BackgroundProcessor resets the condvar when everything's done persisting.
+				loop {
+					if !nodes[0].node.get_persistence_condvar_value() { break }
+				}
+				break
+			}
+		}
+		done_persisting = false;
+		loop {
+			check_persisted_data!(nodes[1].node, filepath_1.clone());
+			if done_persisting {
+				loop {
+					if !nodes[1].node.get_persistence_condvar_value() { break }
+				}
+				break
+			}
+		}
+
+		// Force-close the channel.
+		nodes[0].node.force_close_channel(&OutPoint { txid: tx.txid(), index: 0 }.to_channel_id());
+
+		// Check that the force-close updates are persisted.
+		done_persisting = false;
+		loop {
+			check_persisted_data!(nodes[0].node, filepath_0.clone());
+			if done_persisting {
+				loop {
+					if !nodes[0].node.get_persistence_condvar_value() { break }
+				}
+				break
+			}
+		}
+	}
+
+	#[test]
+	fn test_chan_freshness_called() {
+		// Test that ChannelManager's `timer_chan_freshness_every_min` is called every
+		// `CHAN_FRESHNESS_TIMER`.
+		let nodes = create_nodes(1, "test_chan_freshness_called".to_string());
+		let data_dir = nodes[0].persister.get_data_dir();
+		let callback = move |node| FilesystemPersister::persist_manager(data_dir.clone(), node);
+		let processor = BackgroundProcessor::start(callback, nodes[0].node.clone(), nodes[0].logger.clone());
+		loop {
+			let log_entries = nodes[0].logger.lines.lock().unwrap();
+			let desired_log = "Calling manager's timer_chan_freshness_every_min".to_string();
+			if log_entries.get(&("background_processor".to_string(), desired_log)).is_some() {
+				break
+			}
+		}
+		processor.stop();
+		processor.thread_handle.join().unwrap();
+	}
+
+	#[test]
+	fn test_persist_error() {
+		// Test that if we encounter an error during manager persistence, the thread panics.
+		fn persist_manager<ChanSigner, M, T, K, F, L>(_data: Arc<ChannelManager<ChanSigner, Arc<M>, Arc<T>, Arc<K>, Arc<F>, Arc<L>>>) -> Result<(), std::io::Error>
+		where ChanSigner: 'static + ChannelKeys + Writeable,
+		      M: 'static + chain::Watch<Keys=ChanSigner>,
+		      T: 'static + BroadcasterInterface,
+		      K: 'static + KeysInterface<ChanKeySigner=ChanSigner>,
+		      F: 'static + FeeEstimator,
+		      L: 'static + Logger,
+		{
+			Err(std::io::Error::new(std::io::ErrorKind::Other, "test"))
+		}
+
+		let nodes = create_nodes(2, "test_persist_error".to_string());
+		let bg_processor = BackgroundProcessor::start(persist_manager, nodes[0].node.clone(), nodes[0].logger.clone());
+		open_channel!(nodes[0], nodes[1], 100000);
+
+		match bg_processor.thread_handle.join() {
+			Err(e) => {
+				let error = e.downcast_ref::<String>().unwrap();
+				assert_eq!("Errored persisting manager: test", error);
+			}
+			_ => panic!("Unexpected result")
+		}
+	}
+
+	#[test]
+	fn test_stop_bg_thread() {
+		// Test that we can successfully stop the background thread through BackgroundProcessor::stop().
+		let nodes = create_nodes(1, "test_stop_bg_thread".to_string());
+		let data_dir = nodes[0].persister.get_data_dir();
+		let callback = move |node| FilesystemPersister::persist_manager(data_dir.clone(), node);
+		let bg_processor = BackgroundProcessor::start(callback, nodes[0].node.clone(), nodes[0].logger.clone());
+
+		// Stop the background processor and check that it terminated.
+		bg_processor.stop();
+		loop {
+			let log_entries = nodes[0].logger.lines.lock().unwrap();
+			if log_entries.get(&("background_processor".to_string(), "Terminating background processor.".to_string())).is_some() {
+				break
+			}
+		}
+	}
+}