feat(primitives): increase minimum pruning distance (#4750)

bin/reth/src/args/pruning_args.rs

@@ -11,8 +11,8 @@ use std::sync::Arc;
 #[derive(Debug, Args, PartialEq, Default)]
 #[command(next_help_heading = "Pruning")]
 pub struct PruningArgs {
-    /// Run full node. Only the most recent 128 block states are stored. This flag takes
-    /// priority over pruning configuration in reth.toml.
+    /// Run full node. Only the most recent [`MINIMUM_PRUNING_DISTANCE`] block states are stored.
+    /// This flag takes priority over pruning configuration in reth.toml.
     #[arg(long, default_value_t = false)]
     pub full: bool,
 }

crates/primitives/src/prune/mode.rs

@@ -71,7 +71,7 @@ mod tests {
 
     #[test]
     fn test_prune_target_block() {
-        let tip = 1000;
+        let tip = 20000;
         let min_blocks = MINIMUM_PRUNING_DISTANCE;
         let prune_part = PrunePart::Receipts;
 
@@ -91,7 +91,6 @@ mod tests {
                 PruneMode::Before(tip - MINIMUM_PRUNING_DISTANCE - 1),
                 Ok(Some(tip - MINIMUM_PRUNING_DISTANCE - 2)),
             ),
-            // MINIMUM_PRUNING_DISTANCE is 128
             (PruneMode::Before(tip - 1), Err(PrunePartError::Configuration(prune_part))),
         ];
 
@@ -113,7 +112,7 @@ mod tests {
 
     #[test]
     fn test_should_prune() {
-        let tip = 1000;
+        let tip = 20000;
         let should_prune = true;
 
         let tests = vec![

crates/primitives/src/prune/target.rs

@@ -5,8 +5,12 @@ use crate::{
 use paste::paste;
 use serde::{Deserialize, Serialize};
 
-/// Minimum distance necessary from the tip so blockchain tree can work correctly.
-pub const MINIMUM_PRUNING_DISTANCE: u64 = 128;
+/// Minimum distance from the tip necessary for the node to work correctly:
+/// 1. Minimum 2 epochs (32 blocks per epoch) required to handle any reorg according to the
+///    consensus protocol.
+/// 2. Another 10k blocks to have a room for maneuver in case when things go wrong and a manual
+///    unwind is required.
+pub const MINIMUM_PRUNING_DISTANCE: u64 = 32 * 2 + 10_000;
 
 /// Pruning configuration for every part of the data that can be pruned.
 #[derive(Debug, Clone, Default, Deserialize, Eq, PartialEq, Serialize)]
@@ -22,19 +26,19 @@ pub struct PruneModes {
     /// and offers improved performance.
     #[serde(
         skip_serializing_if = "Option::is_none",
-        deserialize_with = "deserialize_opt_prune_mode_with_min_blocks::<64, _>"
+        deserialize_with = "deserialize_opt_prune_mode_with_min_blocks::<MINIMUM_PRUNING_DISTANCE, _>"
     )]
     pub receipts: Option<PruneMode>,
     /// Account History pruning configuration.
     #[serde(
         skip_serializing_if = "Option::is_none",
-        deserialize_with = "deserialize_opt_prune_mode_with_min_blocks::<64, _>"
+        deserialize_with = "deserialize_opt_prune_mode_with_min_blocks::<MINIMUM_PRUNING_DISTANCE, _>"
     )]
     pub account_history: Option<PruneMode>,
     /// Storage History pruning configuration.
     #[serde(
         skip_serializing_if = "Option::is_none",
-        deserialize_with = "deserialize_opt_prune_mode_with_min_blocks::<64, _>"
+        deserialize_with = "deserialize_opt_prune_mode_with_min_blocks::<MINIMUM_PRUNING_DISTANCE, _>"
     )]
     pub storage_history: Option<PruneMode>,
     /// Receipts pruning configuration by retaining only those receipts that contain logs emitted
@@ -101,8 +105,8 @@ impl PruneModes {
     impl_prune_parts!(
         (sender_recovery, SenderRecovery, None),
         (transaction_lookup, TransactionLookup, None),
-        (receipts, Receipts, Some(64)),
-        (account_history, AccountHistory, Some(64)),
-        (storage_history, StorageHistory, Some(64))
+        (receipts, Receipts, Some(MINIMUM_PRUNING_DISTANCE)),
+        (account_history, AccountHistory, Some(MINIMUM_PRUNING_DISTANCE)),
+        (storage_history, StorageHistory, Some(MINIMUM_PRUNING_DISTANCE))
     );
 }

crates/prune/src/pruner.rs

@@ -416,7 +416,8 @@ impl<DB: Database> Pruner<DB> {
         tip_block_number: BlockNumber,
     ) -> PrunerResult {
         // Contract log filtering removes every receipt possible except the ones in the list. So,
-        // for the other receipts it's as if they had a `PruneMode::Distance()` of 128.
+        // for the other receipts it's as if they had a `PruneMode::Distance()` of
+        // `MINIMUM_PRUNING_DISTANCE`.
         let to_block = PruneMode::Distance(MINIMUM_PRUNING_DISTANCE)
             .prune_target_block(
                 tip_block_number,
@@ -1523,15 +1524,19 @@ mod tests {
         let tx = TestTransaction::default();
         let mut rng = generators::rng();
 
-        let tip = 300;
-        let blocks = random_block_range(&mut rng, 0..=tip, B256::ZERO, 1..5);
+        let tip = 20000;
+        let blocks = [
+            random_block_range(&mut rng, 0..=100, B256::ZERO, 1..5),
+            random_block_range(&mut rng, (100 + 1)..=(tip - 100), B256::ZERO, 0..1),
+            random_block_range(&mut rng, (tip - 100 + 1)..=tip, B256::ZERO, 1..5),
+        ]
+        .concat();
         tx.insert_blocks(blocks.iter(), None).expect("insert blocks");
 
         let mut receipts = Vec::new();
 
         let (deposit_contract_addr, _) = random_eoa_account(&mut rng);
         for block in &blocks {
-            assert!(!block.body.is_empty());
             for (txi, transaction) in block.body.iter().enumerate() {
                 let mut receipt = random_receipt(&mut rng, transaction, Some(1));
                 receipt.logs.push(random_log(
@@ -1569,7 +1574,7 @@ mod tests {
                     ..Default::default()
                 },
                 // Less than total amount of blocks to prune to test the batching logic
-                PruneBatchSizes::default().with_storage_history(10),
+                PruneBatchSizes::default().with_receipts(10),
             );
 
             let result = pruner.prune_receipts_by_logs(&provider, tip);

crates/stages/src/stages/index_account_history.rs

@@ -424,7 +424,7 @@ mod tests {
         .unwrap();
 
         // run
-        let input = ExecInput { target: Some(100), ..Default::default() };
+        let input = ExecInput { target: Some(20000), ..Default::default() };
         let mut stage = IndexAccountHistoryStage {
             prune_modes: PruneModes {
                 account_history: Some(PruneMode::Before(36)),
@@ -435,15 +435,15 @@ mod tests {
         let factory = ProviderFactory::new(tx.tx.as_ref(), MAINNET.clone());
         let provider = factory.provider_rw().unwrap();
         let out = stage.execute(&provider, input).await.unwrap();
-        assert_eq!(out, ExecOutput { checkpoint: StageCheckpoint::new(100), done: true });
+        assert_eq!(out, ExecOutput { checkpoint: StageCheckpoint::new(20000), done: true });
         provider.commit().unwrap();
 
         // verify
         let table = cast(tx.table::<tables::AccountHistory>().unwrap());
         assert_eq!(table, BTreeMap::from([(shard(u64::MAX), vec![36, 100])]));
 
         // unwind
-        unwind(&tx, 100, 0).await;
+        unwind(&tx, 20000, 0).await;
 
         // verify initial state
         let table = tx.table::<tables::AccountHistory>().unwrap();

crates/stages/src/stages/index_storage_history.rs

@@ -437,7 +437,7 @@ mod tests {
         .unwrap();
 
         // run
-        let input = ExecInput { target: Some(100), ..Default::default() };
+        let input = ExecInput { target: Some(20000), ..Default::default() };
         let mut stage = IndexStorageHistoryStage {
             prune_modes: PruneModes {
                 storage_history: Some(PruneMode::Before(36)),
@@ -448,15 +448,15 @@ mod tests {
         let factory = ProviderFactory::new(tx.tx.as_ref(), MAINNET.clone());
         let provider = factory.provider_rw().unwrap();
         let out = stage.execute(&provider, input).await.unwrap();
-        assert_eq!(out, ExecOutput { checkpoint: StageCheckpoint::new(100), done: true });
+        assert_eq!(out, ExecOutput { checkpoint: StageCheckpoint::new(20000), done: true });
         provider.commit().unwrap();
 
         // verify
         let table = cast(tx.table::<tables::StorageHistory>().unwrap());
         assert_eq!(table, BTreeMap::from([(shard(u64::MAX), vec![36, 100]),]));
 
         // unwind
-        unwind(&tx, 100, 0).await;
+        unwind(&tx, 20000, 0).await;
 
         // verify initial state
         let table = tx.table::<tables::StorageHistory>().unwrap();