feat: add prune for ancient sidecars data (#95)

crates/config/src/config.rs

@@ -348,14 +348,16 @@ impl Default for IndexHistoryConfig {
 pub struct PruneConfig {
     /// Minimum pruning interval measured in blocks.
     pub block_interval: usize,
+    /// The number of recent sidecars to keep in the static file provider.
+    pub recent_sidecars_kept_blocks: usize,
     /// Pruning configuration for every part of the data that can be pruned.
     #[serde(alias = "parts")]
     pub segments: PruneModes,
 }
 
 impl Default for PruneConfig {
     fn default() -> Self {
-        Self { block_interval: 5, segments: PruneModes::none() }
+        Self { block_interval: 5, recent_sidecars_kept_blocks: 0, segments: PruneModes::none() }
     }
 }
 

crates/consensus/beacon/src/engine/test_utils.rs

@@ -410,6 +410,7 @@ where
             self.base_config.chain_spec.prune_delete_limit,
             None,
             watch::channel(FinishedExExHeight::NoExExs).1,
+            0,
         );
 
         let mut hooks = EngineHooks::new();

crates/engine/tree/src/persistence.rs

@@ -200,6 +200,7 @@ mod tests {
             0,
             None,
             finished_exex_height_rx,
+            0,
         );
 
         PersistenceHandle::spawn_services(provider, pruner)

crates/node/core/src/args/pruning.rs

@@ -23,6 +23,7 @@ impl PruningArgs {
         }
         Some(PruneConfig {
             block_interval: 5,
+            recent_sidecars_kept_blocks: 0,
             segments: PruneModes {
                 sender_recovery: Some(PruneMode::Full),
                 transaction_lookup: None,

crates/prune/prune/src/builder.rs

@@ -21,6 +21,8 @@ pub struct PrunerBuilder {
     timeout: Option<Duration>,
     /// The finished height of all `ExEx`'s.
     finished_exex_height: watch::Receiver<FinishedExExHeight>,
+    /// The number of recent sidecars to keep in the static file provider.
+    recent_sidecars_kept_blocks: usize,
 }
 
 impl PrunerBuilder {
@@ -32,6 +34,7 @@ impl PrunerBuilder {
         Self::default()
             .block_interval(pruner_config.block_interval)
             .segments(pruner_config.segments)
+            .recent_sidecars_kept_blocks(pruner_config.recent_sidecars_kept_blocks)
     }
 
     /// Sets the minimum pruning interval measured in blocks.
@@ -70,6 +73,12 @@ impl PrunerBuilder {
         self
     }
 
+    /// Sets the number of recent sidecars to keep in the static file provider.
+    pub const fn recent_sidecars_kept_blocks(mut self, recent_sidecars_kept_blocks: usize) -> Self {
+        self.recent_sidecars_kept_blocks = recent_sidecars_kept_blocks;
+        self
+    }
+
     /// Builds a [Pruner] from the current configuration with the given provider factory.
     pub fn build_with_provider_factory<DB: Database>(
         self,
@@ -87,6 +96,7 @@ impl PrunerBuilder {
             self.delete_limit,
             self.timeout,
             self.finished_exex_height,
+            self.recent_sidecars_kept_blocks,
         )
     }
 
@@ -100,6 +110,7 @@ impl PrunerBuilder {
             self.delete_limit,
             self.timeout,
             self.finished_exex_height,
+            self.recent_sidecars_kept_blocks,
         )
     }
 }
@@ -112,6 +123,8 @@ impl Default for PrunerBuilder {
             delete_limit: MAINNET.prune_delete_limit,
             timeout: None,
             finished_exex_height: watch::channel(FinishedExExHeight::NoExExs).1,
+            recent_sidecars_kept_blocks: 0, /* not enabled by default
+                                             * recent_sidecars_kept_blocks: 518400, // 18 days */
         }
     }
 }

crates/prune/prune/src/metrics.rs

@@ -10,6 +10,8 @@ use std::collections::HashMap;
 pub(crate) struct Metrics {
     /// Pruning duration
     pub(crate) duration_seconds: Histogram,
+    /// The height of the oldest sidecars
+    pub(crate) oldest_sidecars_height: Gauge,
     #[metric(skip)]
     prune_segments: HashMap<PruneSegment, PrunerSegmentMetrics>,
 }

crates/prune/prune/src/pruner.rs

@@ -9,8 +9,13 @@ use reth_db_api::database::Database;
 use reth_exex_types::FinishedExExHeight;
 use reth_provider::{DatabaseProviderRW, ProviderFactory, PruneCheckpointReader};
 use reth_prune_types::{PruneLimiter, PruneProgress, PruneSegment, PrunerOutput};
+use reth_static_file_types::{find_fixed_range, StaticFileSegment};
 use reth_tokio_util::{EventSender, EventStream};
-use std::time::{Duration, Instant};
+use std::{
+    fs,
+    path::Path,
+    time::{Duration, Instant},
+};
 use tokio::sync::watch;
 use tracing::debug;
 
@@ -41,6 +46,8 @@ pub struct Pruner<DB, PF> {
     timeout: Option<Duration>,
     /// The finished height of all `ExEx`'s.
     finished_exex_height: watch::Receiver<FinishedExExHeight>,
+    /// The number of recent sidecars to keep in the static file provider.
+    recent_sidecars_kept_blocks: usize,
     #[doc(hidden)]
     metrics: Metrics,
     event_sender: EventSender<PrunerEvent>,
@@ -54,6 +61,7 @@ impl<DB> Pruner<DB, ()> {
         delete_limit: usize,
         timeout: Option<Duration>,
         finished_exex_height: watch::Receiver<FinishedExExHeight>,
+        recent_sidecars_kept_blocks: usize,
     ) -> Self {
         Self {
             provider_factory: (),
@@ -63,6 +71,7 @@ impl<DB> Pruner<DB, ()> {
             delete_limit,
             timeout,
             finished_exex_height,
+            recent_sidecars_kept_blocks,
             metrics: Metrics::default(),
             event_sender: Default::default(),
         }
@@ -71,13 +80,15 @@ impl<DB> Pruner<DB, ()> {
 
 impl<DB: Database> Pruner<DB, ProviderFactory<DB>> {
     /// Crates a new pruner with the given provider factory.
+    #[allow(clippy::too_many_arguments)]
     pub fn new(
         provider_factory: ProviderFactory<DB>,
         segments: Vec<Box<dyn Segment<DB>>>,
         min_block_interval: usize,
         delete_limit: usize,
         timeout: Option<Duration>,
         finished_exex_height: watch::Receiver<FinishedExExHeight>,
+        recent_sidecars_kept_blocks: usize,
     ) -> Self {
         Self {
             provider_factory,
@@ -87,6 +98,7 @@ impl<DB: Database> Pruner<DB, ProviderFactory<DB>> {
             delete_limit,
             timeout,
             finished_exex_height,
+            recent_sidecars_kept_blocks,
             metrics: Metrics::default(),
             event_sender: Default::default(),
         }
@@ -129,6 +141,8 @@ impl<DB: Database, S> Pruner<DB, S> {
         let (stats, deleted_entries, output) =
             self.prune_segments(provider, tip_block_number, &mut limiter)?;
 
+        self.prune_ancient_sidecars(provider, tip_block_number);
+
         self.previous_tip_block_number = Some(tip_block_number);
 
         let elapsed = start.elapsed();
@@ -294,6 +308,80 @@ impl<DB: Database, S> Pruner<DB, S> {
             }
         }
     }
+
+    /// Prunes ancient sidecars data from the static file provider.
+    pub fn prune_ancient_sidecars(
+        &mut self,
+        provider: &DatabaseProviderRW<DB>,
+        tip_block_number: BlockNumber,
+    ) {
+        if self.recent_sidecars_kept_blocks == 0 {
+            return
+        }
+
+        let static_file_provider = provider.static_file_provider();
+
+        let prune_target_block =
+            tip_block_number.saturating_sub(self.recent_sidecars_kept_blocks as u64);
+        let mut range_start = find_fixed_range(prune_target_block).start();
+
+        if range_start == 0 {
+            return
+        }
+
+        debug!(
+            target: "pruner",
+            %tip_block_number,
+            "Ancient sidecars pruning started",
+        );
+
+        while range_start > 0 {
+            let range = find_fixed_range(range_start - 1);
+            let path =
+                static_file_provider.path().join(StaticFileSegment::Sidecars.filename(&range));
+
+            if path.exists() {
+                delete_static_files(&path);
+                self.metrics.oldest_sidecars_height.set(range.end() as f64 + 1_f64);
+            } else {
+                debug!(target: "pruner", path = %path.display(), "Static file not found, skipping");
+                break
+            }
+
+            range_start = range.start();
+        }
+
+        debug!(
+            target: "pruner",
+            %tip_block_number,
+            "Ancient sidecars pruning finished",
+        );
+    }
+}
+
+fn delete_static_files(path: &Path) {
+    // Delete the main file
+    if let Err(err) = fs::remove_file(path) {
+        debug!(target: "pruner", path = %path.display(), %err, "Failed to remove file");
+    } else {
+        debug!(target: "pruner", path = %path.display(), "Removed file");
+    }
+
+    // Delete the .conf file
+    let conf_path = path.with_extension("conf");
+    if let Err(err) = fs::remove_file(&conf_path) {
+        debug!(target: "pruner", path = %conf_path.display(), %err, "Failed to remove .conf file");
+    } else {
+        debug!(target: "pruner", path = %conf_path.display(), "Removed .conf file");
+    }
+
+    // Delete the .off file
+    let off_path = path.with_extension("off");
+    if let Err(err) = fs::remove_file(&off_path) {
+        debug!(target: "pruner", path = %off_path.display(), %err, "Failed to remove .off file");
+    } else {
+        debug!(target: "pruner", path = %off_path.display(), "Removed .off file");
+    }
 }
 
 impl<DB: Database> Pruner<DB, ()> {
@@ -347,6 +435,7 @@ mod tests {
             0,
             None,
             finished_exex_height_rx,
+            0,
         );
 
         // No last pruned block number was set before

crates/prune/prune/src/segments/mod.rs

@@ -14,7 +14,7 @@ use reth_prune_types::{
 };
 pub use set::SegmentSet;
 pub use static_file::{
-    Headers as StaticFileHeaders, Receipts as StaticFileReceipts,
+    Headers as StaticFileHeaders, Receipts as StaticFileReceipts, Sidecars as StaticFileSidecars,
     Transactions as StaticFileTransactions,
 };
 use std::{fmt::Debug, ops::RangeInclusive};

crates/prune/prune/src/segments/set.rs

@@ -1,6 +1,6 @@
 use crate::segments::{
-    AccountHistory, ReceiptsByLogs, Segment, SenderRecovery, StorageHistory, TransactionLookup,
-    UserReceipts,
+    AccountHistory, ReceiptsByLogs, Segment, SenderRecovery, StaticFileSidecars, StorageHistory,
+    TransactionLookup, UserReceipts,
 };
 use reth_db_api::database::Database;
 use reth_provider::providers::StaticFileProvider;
@@ -60,7 +60,9 @@ impl<DB: Database> SegmentSet<DB> {
             // Static file transactions
             .segment(StaticFileTransactions::new(static_file_provider.clone()))
             // Static file receipts
-            .segment(StaticFileReceipts::new(static_file_provider))
+            .segment(StaticFileReceipts::new(static_file_provider.clone()))
+            // Static file receipts
+            .segment(StaticFileSidecars::new(static_file_provider))
             // Account history
             .segment_opt(account_history.map(AccountHistory::new))
             // Storage history

crates/prune/prune/src/segments/static_file/mod.rs

@@ -1,7 +1,9 @@
 mod headers;
 mod receipts;
+mod sidecars;
 mod transactions;
 
 pub use headers::Headers;
 pub use receipts::Receipts;
+pub use sidecars::Sidecars;
 pub use transactions::Transactions;