Port aleph-node 8.2 testnet to mainnet branch (#776)

* Request justification from forks

* Reverted changes to cargo lock introduced in 6ae50ca (#773)

* A0-1635: Limit nonfinalized block production (#769) (#774)

* A0-1635: Limit nonfinalized block production (#769)

* Limit nonfinalized block production

* Better saturation handling

* Add log when delaying block production

* Disable two tests that were operating under outdated assumptions

* Disable one more test

* Turn off the likely broken test

* Changed node version

* Fixed leftover conflicted line in lock file

* Reverted unwanted change to runtime/Cargo.toml

* Updated Cargo.lock

Co-authored-by: Michal Swietek <[email protected]>
Co-authored-by: timorl <[email protected]>

Author: 3 people

.github/workflows/e2e-tests-main-devnet.yml

@@ -454,7 +454,7 @@ jobs:
       run-e2e-rewards-stake-change,
       run-e2e-rewards-change-stake-force-new-era,
       run-e2e-rewards-points-basic,
-      run-e2e-authorities-are-staking,
+#      run-e2e-authorities-are-staking,
       run-e2e-ban-automatic,
       run-e2e-version-upgrade,
     ]

Cargo.lock

@@ -1,6 +1,6 @@
 [package]
 name = "aleph-node"
-version = "0.8.0+mainnet"
+version = "0.8.2+mainnet"
 authors = ["Cardinal Cryptography"]
 description = "Aleph node binary"
 edition = "2021"
@@ -43,6 +43,8 @@ sc-network = { git = "https://github.com/Cardinal-Cryptography/substrate.git", b
 sc-transaction-pool = { git = "https://github.com/Cardinal-Cryptography/substrate.git", branch = "aleph-v0.9.28" }
 sp-transaction-pool = { git = "https://github.com/Cardinal-Cryptography/substrate.git", branch = "aleph-v0.9.28" }
 sc-transaction-pool-api = { git = "https://github.com/Cardinal-Cryptography/substrate.git", branch = "aleph-v0.9.28" }
+sp-arithmetic = { git = "https://github.com/Cardinal-Cryptography/substrate.git", branch = "aleph-v0.9.28" }
+sc-consensus-slots = { git = "https://github.com/Cardinal-Cryptography/substrate.git", branch = "aleph-v0.9.28" }
 sc-consensus-aura = { git = "https://github.com/Cardinal-Cryptography/substrate.git", branch = "aleph-v0.9.28" }
 sp-consensus-aura = { git = "https://github.com/Cardinal-Cryptography/substrate.git", branch = "aleph-v0.9.28" }
 sp-consensus = { git = "https://github.com/Cardinal-Cryptography/substrate.git", branch = "aleph-v0.9.28" }

bin/node/Cargo.toml

@@ -3,6 +3,7 @@ use std::path::PathBuf;
 use aleph_primitives::DEFAULT_UNIT_CREATION_DELAY;
 use clap::{ArgGroup, Parser};
 use finality_aleph::UnitCreationDelay;
+use log::warn;
 
 #[derive(Debug, Parser, Clone)]
 #[clap(group(ArgGroup::new("backup")))]
@@ -33,6 +34,12 @@ pub struct AlephCli {
     /// with `--no-backup`, but note that that limits crash recoverability.
     #[clap(long, value_name = "PATH", group = "backup")]
     backup_path: Option<PathBuf>,
+
+    /// The maximum number of nonfinalized blocks, after which block production should be locally
+    /// stopped. DO NOT CHANGE THIS, PRODUCING MORE OR FEWER BLOCKS MIGHT BE CONSIDERED MALICIOUS
+    /// BEHAVIOUR AND PUNISHED ACCORDINGLY!
+    #[clap(long, default_value_t = 20)]
+    max_nonfinalized_blocks: u32,
 }
 
 impl AlephCli {
@@ -55,4 +62,11 @@ impl AlephCli {
     pub fn no_backup(&self) -> bool {
         self.no_backup
     }
+
+    pub fn max_nonfinalized_blocks(&self) -> u32 {
+        if self.max_nonfinalized_blocks != 20 {
+            warn!("Running block production with a value of max-nonfinalized-blocks {}, which is not the default of 20. THIS MIGHT BE CONSIDERED MALICIOUS BEHAVIOUR AND RESULT IN PENALTIES!", self.max_nonfinalized_blocks);
+        }
+        self.max_nonfinalized_blocks
+    }
 }

bin/node/src/aleph_cli.rs

@@ -13,16 +13,19 @@ use finality_aleph::{
 };
 use futures::channel::mpsc;
 use log::warn;
-use sc_client_api::ExecutorProvider;
+use sc_client_api::{Backend, ExecutorProvider, HeaderBackend};
 use sc_consensus_aura::{ImportQueueParams, SlotProportion, StartAuraParams};
+use sc_consensus_slots::BackoffAuthoringBlocksStrategy;
 use sc_network::NetworkService;
 use sc_service::{
     error::Error as ServiceError, Configuration, KeystoreContainer, NetworkStarter, RpcHandlers,
     TFullClient, TaskManager,
 };
 use sc_telemetry::{Telemetry, TelemetryWorker};
 use sp_api::ProvideRuntimeApi;
-use sp_consensus_aura::sr25519::AuthorityPair as AuraPair;
+use sp_arithmetic::traits::BaseArithmetic;
+use sp_blockchain::Backend as _;
+use sp_consensus_aura::{sr25519::AuthorityPair as AuraPair, Slot};
 use sp_runtime::{
     generic::BlockId,
     traits::{Block as BlockT, Header as HeaderT, Zero},
@@ -34,7 +37,31 @@ type FullClient = sc_service::TFullClient<Block, RuntimeApi, AlephExecutor>;
 type FullBackend = sc_service::TFullBackend<Block>;
 type FullSelectChain = sc_consensus::LongestChain<FullBackend, Block>;
 
-fn get_backup_path(aleph_config: &AlephCli, base_path: &Path) -> Option<PathBuf> {
+struct LimitNonfinalized(u32);
+
+impl<N: BaseArithmetic> BackoffAuthoringBlocksStrategy<N> for LimitNonfinalized {
+    fn should_backoff(
+        &self,
+        chain_head_number: N,
+        _chain_head_slot: Slot,
+        finalized_number: N,
+        _slow_now: Slot,
+        _logging_target: &str,
+    ) -> bool {
+        let nonfinalized_blocks: u32 = chain_head_number
+            .saturating_sub(finalized_number)
+            .unique_saturated_into();
+        match nonfinalized_blocks >= self.0 {
+            true => {
+                warn!("We have {} nonfinalized blocks, with the limit being {}, delaying block production.", nonfinalized_blocks, self.0);
+                true
+            }
+            false => false,
+        }
+    }
+}
+
+fn backup_path(aleph_config: &AlephCli, base_path: &Path) -> Option<PathBuf> {
     if aleph_config.no_backup() {
         return None;
     }
@@ -264,7 +291,7 @@ pub fn new_authority(
         .extra_sets
         .push(finality_aleph::peers_set_config(Protocol::Validator));
 
-    let backup_path = get_backup_path(
+    let backup_path = backup_path(
         &aleph_config,
         config
             .base_path
@@ -288,12 +315,12 @@ pub fn new_authority(
     );
 
     let force_authoring = config.force_authoring;
-    let backoff_authoring_blocks: Option<()> = None;
+    let backoff_authoring_blocks = Some(LimitNonfinalized(aleph_config.max_nonfinalized_blocks()));
     let prometheus_registry = config.prometheus_registry().cloned();
 
     let (_rpc_handlers, network, network_starter) = setup(
         config,
-        backend,
+        backend.clone(),
         &keystore_container,
         import_queue,
         transaction_pool.clone(),
@@ -353,9 +380,11 @@ pub fn new_authority(
     if aleph_config.external_addresses().is_empty() {
         panic!("Cannot run a validator node without external addresses, stopping.");
     }
+    let blockchain_backend = BlockchainBackendImpl { backend };
     let aleph_config = AlephConfig {
         network,
         client,
+        blockchain_backend,
         select_chain,
         session_period,
         millisecs_per_block,
@@ -393,7 +422,7 @@ pub fn new_full(
         other: (_, justification_tx, justification_rx, mut telemetry, metrics),
     } = new_partial(&config)?;
 
-    let backup_path = get_backup_path(
+    let backup_path = backup_path(
         &aleph_config,
         config
             .base_path
@@ -404,7 +433,7 @@ pub fn new_full(
 
     let (_rpc_handlers, network, network_starter) = setup(
         config,
-        backend,
+        backend.clone(),
         &keystore_container,
         import_queue,
         transaction_pool,
@@ -428,9 +457,11 @@ pub fn new_full(
             .unwrap(),
     );
 
+    let blockchain_backend = BlockchainBackendImpl { backend };
     let aleph_config = AlephConfig {
         network,
         client,
+        blockchain_backend,
         select_chain,
         session_period,
         millisecs_per_block,
@@ -453,3 +484,24 @@ pub fn new_full(
     network_starter.start_network();
     Ok(task_manager)
 }
+
+struct BlockchainBackendImpl {
+    backend: Arc<FullBackend>,
+}
+impl finality_aleph::BlockchainBackend<Block> for BlockchainBackendImpl {
+    fn children(&self, parent_hash: <Block as BlockT>::Hash) -> Vec<<Block as BlockT>::Hash> {
+        self.backend
+            .blockchain()
+            .children(parent_hash)
+            .unwrap_or_default()
+    }
+    fn info(&self) -> sp_blockchain::Info<Block> {
+        self.backend.blockchain().info()
+    }
+    fn header(
+        &self,
+        block_id: sp_api::BlockId<Block>,
+    ) -> sp_blockchain::Result<Option<<Block as BlockT>::Header>> {
+        self.backend.blockchain().header(block_id)
+    }
+}

bin/node/src/service.rs

@@ -1,6 +1,6 @@
 [package]
 name = "finality-aleph"
-version = "0.5.0"
+version = "0.5.3"
 authors = ["Cardinal Cryptography"]
 edition = "2021"
 license = "Apache 2.0"

finality-aleph/Cargo.toml

@@ -1,12 +1,8 @@
-use std::{
-    sync::Arc,
-    time::{Duration, Instant},
-};
+use std::time::{Duration, Instant};
 
 use futures::{channel::mpsc, Stream, StreamExt};
 use futures_timer::Delay;
 use log::{debug, error};
-use sc_client_api::HeaderBackend;
 use sp_api::BlockT;
 use sp_runtime::traits::Header;
 use tokio::time::timeout;
@@ -17,53 +13,52 @@ use crate::{
         requester::BlockRequester, JustificationHandlerConfig, JustificationNotification,
         JustificationRequestScheduler, SessionInfo, SessionInfoProvider, Verifier,
     },
-    network, Metrics, STATUS_REPORT_INTERVAL,
+    network, BlockchainBackend, Metrics, STATUS_REPORT_INTERVAL,
 };
 
-pub struct JustificationHandler<B, V, RB, C, S, SI, F>
+pub struct JustificationHandler<B, V, RB, S, SI, F, BB>
 where
     B: BlockT,
     V: Verifier<B>,
     RB: network::RequestBlocks<B> + 'static,
-    C: HeaderBackend<B> + Send + Sync + 'static,
     S: JustificationRequestScheduler,
     SI: SessionInfoProvider<B, V>,
     F: BlockFinalizer<B>,
+    BB: BlockchainBackend<B> + 'static,
 {
     session_info_provider: SI,
-    block_requester: BlockRequester<B, RB, C, S, F, V>,
+    block_requester: BlockRequester<B, RB, S, F, V, BB>,
     verifier_timeout: Duration,
     notification_timeout: Duration,
 }
 
-impl<B, V, RB, C, S, SI, F> JustificationHandler<B, V, RB, C, S, SI, F>
+impl<B, V, RB, S, SI, F, BB> JustificationHandler<B, V, RB, S, SI, F, BB>
 where
     B: BlockT,
     V: Verifier<B>,
     RB: network::RequestBlocks<B> + 'static,
-    C: HeaderBackend<B> + Send + Sync + 'static,
     S: JustificationRequestScheduler,
     SI: SessionInfoProvider<B, V>,
     F: BlockFinalizer<B>,
+    BB: BlockchainBackend<B> + 'static,
 {
     pub fn new(
         session_info_provider: SI,
         block_requester: RB,
-        client: Arc<C>,
+        blockchain_backend: BB,
         finalizer: F,
         justification_request_scheduler: S,
         metrics: Option<Metrics<<B::Header as Header>::Hash>>,
-        justification_handler_config: JustificationHandlerConfig<B>,
+        justification_handler_config: JustificationHandlerConfig,
     ) -> Self {
         Self {
             session_info_provider,
             block_requester: BlockRequester::new(
                 block_requester,
-                client,
+                blockchain_backend,
                 finalizer,
                 justification_request_scheduler,
                 metrics,
-                justification_handler_config.min_allowed_delay,
             ),
             verifier_timeout: justification_handler_config.verifier_timeout,
             notification_timeout: justification_handler_config.notification_timeout,

finality-aleph/src/justification/handler.rs

@@ -55,36 +55,29 @@ pub struct JustificationNotification<Block: BlockT> {
 }
 
 #[derive(Clone)]
-pub struct JustificationHandlerConfig<B: BlockT> {
+pub struct JustificationHandlerConfig {
     /// How long should we wait when the session verifier is not yet available.
     verifier_timeout: Duration,
     /// How long should we wait for any notification.
     notification_timeout: Duration,
-    ///Distance (in amount of blocks) between the best and the block we want to request justification
-    min_allowed_delay: NumberFor<B>,
 }
 
-impl<B: BlockT> Default for JustificationHandlerConfig<B> {
+impl Default for JustificationHandlerConfig {
     fn default() -> Self {
         Self {
             verifier_timeout: Duration::from_millis(500),
-            notification_timeout: Duration::from_millis(1000),
-            min_allowed_delay: 3u32.into(),
+            // request justifications slightly more frequently than they're created
+            notification_timeout: Duration::from_millis(800),
         }
     }
 }
 
 #[cfg(test)]
-impl<B: BlockT> JustificationHandlerConfig<B> {
-    pub fn new(
-        verifier_timeout: Duration,
-        notification_timeout: Duration,
-        min_allowed_delay: NumberFor<B>,
-    ) -> Self {
+impl JustificationHandlerConfig {
+    pub fn new(verifier_timeout: Duration, notification_timeout: Duration) -> Self {
         Self {
             verifier_timeout,
             notification_timeout,
-            min_allowed_delay,
         }
     }
 }