chore(downloader): replace database with header provider (#5472)

Cargo.lock

@@ -5890,6 +5890,7 @@ dependencies = [
  "reth-interfaces",
  "reth-metrics",
  "reth-primitives",
+ "reth-provider",
  "reth-tasks",
  "reth-tracing",
  "tempfile",

bin/reth/src/chain/import.rs

@@ -147,7 +147,11 @@ impl ImportCommand {
             .into_task();
 
         let body_downloader = BodiesDownloaderBuilder::from(config.stages.bodies)
-            .build(file_client.clone(), consensus.clone(), db.clone())
+            .build(
+                file_client.clone(),
+                consensus.clone(),
+                ProviderFactory::new(db.clone(), self.chain.clone()),
+            )
             .into_task();
 
         let (tip_tx, tip_rx) = watch::channel(B256::ZERO);

bin/reth/src/debug_cmd/execution.rs

@@ -102,7 +102,11 @@ impl Command {
             .into_task_with(task_executor);
 
         let body_downloader = BodiesDownloaderBuilder::from(config.stages.bodies)
-            .build(client, Arc::clone(&consensus), db.clone())
+            .build(
+                client,
+                Arc::clone(&consensus),
+                ProviderFactory::new(db.clone(), self.chain.clone()),
+            )
             .into_task_with(task_executor);
 
         let stage_conf = &config.stages;

bin/reth/src/node/mod.rs

@@ -617,7 +617,11 @@ impl<Ext: RethCliExt> NodeCommand<Ext> {
             .into_task_with(task_executor);
 
         let body_downloader = BodiesDownloaderBuilder::from(config.stages.bodies)
-            .build(client, Arc::clone(&consensus), db.clone())
+            .build(
+                client,
+                Arc::clone(&consensus),
+                ProviderFactory::new(db.clone(), self.chain.clone()),
+            )
             .into_task_with(task_executor);
 
         let pipeline = self

bin/reth/src/stage/run.rs

@@ -163,6 +163,9 @@ impl Command {
 
                     let default_peers_path = data_dir.known_peers_path();
 
+                    let provider_factory =
+                        Arc::new(ProviderFactory::new(db.clone(), self.chain.clone()));
+
                     let network = self
                         .network
                         .network_config(
@@ -171,7 +174,7 @@ impl Command {
                             p2p_secret_key,
                             default_peers_path,
                         )
-                        .build(Arc::new(ProviderFactory::new(db.clone(), self.chain.clone())))
+                        .build(provider_factory.clone())
                         .start_network()
                         .await?;
                     let fetch_client = Arc::new(network.fetch_client().await?);
@@ -187,9 +190,8 @@ impl Command {
                                 config.stages.bodies.downloader_min_concurrent_requests..=
                                     config.stages.bodies.downloader_max_concurrent_requests,
                             )
-                            .build(fetch_client, consensus.clone(), db.clone()),
+                            .build(fetch_client, consensus.clone(), provider_factory),
                     );
-
                     (Box::new(stage), None)
                 }
                 StageEnum::Senders => (Box::new(SenderRecoveryStage::new(batch_size)), None),

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

@@ -455,6 +455,8 @@ where
     pub fn build(self) -> (TestBeaconConsensusEngine<Client>, TestEnv<Arc<DatabaseEnv>>) {
         reth_tracing::init_test_tracing();
         let db = create_test_rw_db();
+        let provider_factory =
+            ProviderFactory::new(db.clone(), self.base_config.chain_spec.clone());
 
         let consensus: Arc<dyn Consensus> = match self.base_config.consensus {
             TestConsensusConfig::Real => {
@@ -496,7 +498,7 @@ where
                     .into_task();
 
                 let body_downloader = BodiesDownloaderBuilder::default()
-                    .build(client.clone(), consensus.clone(), db.clone())
+                    .build(client.clone(), consensus.clone(), provider_factory.clone())
                     .into_task();
 
                 Pipeline::builder().add_stages(DefaultStages::new(
@@ -527,9 +529,8 @@ where
         let tree = ShareableBlockchainTree::new(
             BlockchainTree::new(externals, config, None).expect("failed to create tree"),
         );
-        let shareable_db = ProviderFactory::new(db.clone(), self.base_config.chain_spec.clone());
         let latest = self.base_config.chain_spec.genesis_header().seal_slow();
-        let blockchain_provider = BlockchainProvider::with_latest(shareable_db, tree, latest);
+        let blockchain_provider = BlockchainProvider::with_latest(provider_factory, tree, latest);
 
         let pruner = Pruner::new(
             db.clone(),

crates/interfaces/src/p2p/error.rs

@@ -1,5 +1,5 @@
 use super::headers::client::HeadersRequest;
-use crate::{consensus::ConsensusError, db};
+use crate::{consensus::ConsensusError, provider::ProviderError};
 use reth_network_api::ReputationChangeKind;
 use reth_primitives::{
     BlockHashOrNumber, BlockNumber, GotExpected, GotExpectedBoxed, Header, WithPeerId, B256,
@@ -177,9 +177,9 @@ pub enum DownloadError {
     /// Error while executing the request.
     #[error(transparent)]
     RequestError(#[from] RequestError),
-    /// Error while reading data from database.
+    /// Provider error.
     #[error(transparent)]
-    DatabaseError(#[from] db::DatabaseError),
+    Provider(#[from] ProviderError),
 }
 
 #[cfg(test)]

crates/net/downloaders/Cargo.toml

@@ -12,8 +12,8 @@ description = "Implementations of various block downloaders"
 # reth
 reth-interfaces.workspace = true
 reth-primitives.workspace = true
-reth-db.workspace = true
 reth-tasks.workspace = true
+reth-provider.workspace = true
 
 # async
 futures.workspace = true
@@ -33,6 +33,7 @@ rayon.workspace = true
 thiserror.workspace = true
 
 # optional deps for the test-utils feature
+reth-db = { workspace = true, optional = true }
 alloy-rlp = { workspace = true, optional = true }
 tempfile = { workspace = true, optional = true }
 itertools = { workspace = true, optional = true }
@@ -50,4 +51,4 @@ itertools.workspace = true
 tempfile.workspace = true
 
 [features]
-test-utils = ["dep:alloy-rlp", "dep:tempfile", "dep:itertools", "reth-interfaces/test-utils"]
+test-utils = ["dep:alloy-rlp", "dep:tempfile", "dep:itertools", "reth-db/test-utils", "reth-interfaces/test-utils"]

crates/net/downloaders/src/bodies/bodies.rs

@@ -2,7 +2,6 @@ use super::queue::BodiesRequestQueue;
 use crate::{bodies::task::TaskDownloader, metrics::BodyDownloaderMetrics};
 use futures::Stream;
 use futures_util::StreamExt;
-use reth_db::{cursor::DbCursorRO, database::Database, tables, transaction::DbTx};
 use reth_interfaces::{
     consensus::Consensus,
     p2p::{
@@ -15,6 +14,7 @@ use reth_interfaces::{
     },
 };
 use reth_primitives::{BlockNumber, SealedHeader};
+use reth_provider::HeaderProvider;
 use reth_tasks::{TaskSpawner, TokioTaskExecutor};
 use std::{
     cmp::Ordering,
@@ -27,22 +27,18 @@ use std::{
 };
 use tracing::info;
 
-/// The scope for headers downloader metrics.
-pub const BODIES_DOWNLOADER_SCOPE: &str = "downloaders.bodies";
-
 /// Downloads bodies in batches.
 ///
 /// All blocks in a batch are fetched at the same time.
 #[must_use = "Stream does nothing unless polled"]
 #[derive(Debug)]
-pub struct BodiesDownloader<B: BodiesClient, DB> {
+pub struct BodiesDownloader<B: BodiesClient, Provider> {
     /// The bodies client
     client: Arc<B>,
     /// The consensus client
     consensus: Arc<dyn Consensus>,
-    // TODO: make this a [HeaderProvider]
     /// The database handle
-    db: DB,
+    provider: Provider,
     /// The maximum number of non-empty blocks per one request
     request_limit: u64,
     /// The maximum number of block bodies returned at once from the stream
@@ -67,10 +63,10 @@ pub struct BodiesDownloader<B: BodiesClient, DB> {
     metrics: BodyDownloaderMetrics,
 }
 
-impl<B, DB> BodiesDownloader<B, DB>
+impl<B, Provider> BodiesDownloader<B, Provider>
 where
     B: BodiesClient + 'static,
-    DB: Database + Unpin + 'static,
+    Provider: HeaderProvider + Unpin + 'static,
 {
     /// Returns the next contiguous request.
     fn next_headers_request(&mut self) -> DownloadResult<Option<Vec<SealedHeader>>> {
@@ -103,47 +99,29 @@ where
             return Ok(None)
         }
 
-        // Collection of results
-        let mut headers = Vec::new();
-
-        // Non empty headers count
-        let mut non_empty_headers = 0;
-        let mut current_block_num = *range.start();
-
-        // Acquire cursors over canonical and header tables
-        let tx = self.db.tx()?;
-        let mut canonical_cursor = tx.cursor_read::<tables::CanonicalHeaders>()?;
-        let mut header_cursor = tx.cursor_read::<tables::Headers>()?;
-
         // Collect headers while
         //      1. Current block number is in range
         //      2. The number of non empty headers is less than maximum
         //      3. The total number of headers is less than the stream batch size (this is only
         //         relevant if the range consists entirely of empty headers)
-        while range.contains(&current_block_num) &&
+        let mut collected = 0;
+        let mut non_empty_headers = 0;
+        let headers = self.provider.sealed_headers_while(range.clone(), |header| {
+            let should_take = range.contains(&header.number) &&
                 non_empty_headers < max_non_empty &&
-            headers.len() < self.stream_batch_size
+                collected < self.stream_batch_size;
-        {
-            // Find the block hash.
-            let (number, hash) = canonical_cursor
-                .seek_exact(current_block_num)?
-                .ok_or(DownloadError::MissingHeader { block_number: current_block_num })?;
-            // Find the block header.
-            let (_, header) = header_cursor
-                .seek_exact(number)?
-                .ok_or(DownloadError::MissingHeader { block_number: number })?;
 
-            // If the header is not empty, increment the counter
+            if should_take {
+                collected += 1;
                 if !header.is_empty() {
                     non_empty_headers += 1;
                 }
-
+                true
-            // Add header to the result collection
+            } else {
-            headers.push(header.seal(hash));
+                false
-
-            // Increment current block number
-            current_block_num += 1;
             }
+        })?;
+
         Ok(Some(headers).filter(|h| !h.is_empty()))
     }
 
@@ -286,10 +264,10 @@ where
     }
 }
 
-impl<B, DB> BodiesDownloader<B, DB>
+impl<B, Provider> BodiesDownloader<B, Provider>
 where
     B: BodiesClient + 'static,
-    DB: Database + Unpin + 'static,
+    Provider: HeaderProvider + Unpin + 'static,
     Self: BodyDownloader + 'static,
 {
     /// Spawns the downloader task via [tokio::task::spawn]
@@ -306,10 +284,10 @@ where
     }
 }
 
-impl<B, DB> BodyDownloader for BodiesDownloader<B, DB>
+impl<B, Provider> BodyDownloader for BodiesDownloader<B, Provider>
 where
     B: BodiesClient + 'static,
-    DB: Database + Unpin + 'static,
+    Provider: HeaderProvider + Unpin + 'static,
 {
     /// Set a new download range (exclusive).
     ///
@@ -354,10 +332,10 @@ where
     }
 }
 
-impl<B, DB> Stream for BodiesDownloader<B, DB>
+impl<B, Provider> Stream for BodiesDownloader<B, Provider>
 where
     B: BodiesClient + 'static,
-    DB: Database + Unpin + 'static,
+    Provider: HeaderProvider + Unpin + 'static,
 {
     type Item = BodyDownloaderResult;
 
@@ -557,15 +535,15 @@ impl BodiesDownloaderBuilder {
     }
 
     /// Consume self and return the concurrent downloader.
-    pub fn build<B, DB>(
+    pub fn build<B, Provider>(
         self,
         client: B,
         consensus: Arc<dyn Consensus>,
-        db: DB,
+        provider: Provider,
-    ) -> BodiesDownloader<B, DB>
+    ) -> BodiesDownloader<B, Provider>
     where
         B: BodiesClient + 'static,
-        DB: Database,
+        Provider: HeaderProvider,
     {
         let Self {
             request_limit,
@@ -578,7 +556,7 @@ impl BodiesDownloaderBuilder {
         BodiesDownloader {
             client: Arc::new(client),
             consensus,
-            db,
+            provider,
             request_limit,
             stream_batch_size,
             max_buffered_blocks_size_bytes,
@@ -605,7 +583,8 @@ mod tests {
     use futures_util::stream::StreamExt;
     use reth_db::test_utils::create_test_rw_db;
     use reth_interfaces::test_utils::{generators, generators::random_block_range, TestConsensus};
-    use reth_primitives::{BlockBody, B256};
+    use reth_primitives::{BlockBody, B256, MAINNET};
+    use reth_provider::ProviderFactory;
     use std::{collections::HashMap, sync::Arc};
 
     // Check that the blocks are emitted in order of block number, not in order of
@@ -624,7 +603,7 @@ mod tests {
         let mut downloader = BodiesDownloaderBuilder::default().build(
             client.clone(),
             Arc::new(TestConsensus::default()),
-            db,
+            ProviderFactory::new(db, MAINNET.clone()),
         );
         downloader.set_download_range(0..=19).expect("failed to set download range");
 
@@ -659,9 +638,12 @@ mod tests {
 
         let request_limit = 10;
         let client = Arc::new(TestBodiesClient::default().with_bodies(bodies.clone()));
-        let mut downloader = BodiesDownloaderBuilder::default()
+        let mut downloader =
-            .with_request_limit(request_limit)
+            BodiesDownloaderBuilder::default().with_request_limit(request_limit).build(
-            .build(client.clone(), Arc::new(TestConsensus::default()), db);
+                client.clone(),
+                Arc::new(TestConsensus::default()),
+                ProviderFactory::new(db, MAINNET.clone()),
+            );
         downloader.set_download_range(0..=199).expect("failed to set download range");
 
         let _ = downloader.collect::<Vec<_>>().await;
@@ -686,7 +668,11 @@ mod tests {
         let mut downloader = BodiesDownloaderBuilder::default()
             .with_stream_batch_size(stream_batch_size)
             .with_request_limit(request_limit)
-            .build(client.clone(), Arc::new(TestConsensus::default()), db);
+            .build(
+                client.clone(),
+                Arc::new(TestConsensus::default()),
+                ProviderFactory::new(db, MAINNET.clone()),
+            );
 
         let mut range_start = 0;
         while range_start < 100 {
@@ -715,7 +701,7 @@ mod tests {
         let mut downloader = BodiesDownloaderBuilder::default().with_stream_batch_size(100).build(
             client.clone(),
             Arc::new(TestConsensus::default()),
-            db,
+            ProviderFactory::new(db, MAINNET.clone()),
         );
 
         // Set and download the first range
@@ -752,7 +738,11 @@ mod tests {
             .with_stream_batch_size(10)
             .with_request_limit(1)
             .with_max_buffered_blocks_size_bytes(1)
-            .build(client.clone(), Arc::new(TestConsensus::default()), db);
+            .build(
+                client.clone(),
+                Arc::new(TestConsensus::default()),
+                ProviderFactory::new(db, MAINNET.clone()),
+            );
 
         // Set and download the entire range
         downloader.set_download_range(0..=199).expect("failed to set download range");
@@ -779,7 +769,11 @@ mod tests {
         let mut downloader = BodiesDownloaderBuilder::default()
             .with_request_limit(3)
             .with_stream_batch_size(100)
-            .build(client.clone(), Arc::new(TestConsensus::default()), db);
+            .build(
+                client.clone(),
+                Arc::new(TestConsensus::default()),
+                ProviderFactory::new(db, MAINNET.clone()),
+            );
 
         // Download the requested range
         downloader.set_download_range(0..=99).expect("failed to set download range");

crates/net/downloaders/src/bodies/task.rs

@@ -42,16 +42,17 @@ impl TaskDownloader {
     /// # Example
     ///
     /// ```
-    /// use reth_db::database::Database;
     /// use reth_downloaders::bodies::{bodies::BodiesDownloaderBuilder, task::TaskDownloader};
     /// use reth_interfaces::{consensus::Consensus, p2p::bodies::client::BodiesClient};
+    /// use reth_provider::HeaderProvider;
     /// use std::sync::Arc;
-    /// fn t<B: BodiesClient + 'static, DB: Database + 'static>(
+    ///
+    /// fn t<B: BodiesClient + 'static, Provider: HeaderProvider + Unpin + 'static>(
     ///     client: Arc<B>,
     ///     consensus: Arc<dyn Consensus>,
-    ///     db: Arc<DB>,
+    ///     provider: Provider,
     /// ) {
-    ///     let downloader = BodiesDownloaderBuilder::default().build(client, consensus, db);
+    ///     let downloader = BodiesDownloaderBuilder::default().build(client, consensus, provider);
     ///     let downloader = TaskDownloader::spawn(downloader);
     /// }
     /// ```
@@ -170,6 +171,8 @@ mod tests {
     use assert_matches::assert_matches;
     use reth_db::test_utils::create_test_rw_db;
     use reth_interfaces::{p2p::error::DownloadError, test_utils::TestConsensus};
+    use reth_primitives::MAINNET;
+    use reth_provider::ProviderFactory;
     use std::sync::Arc;
 
     #[tokio::test(flavor = "multi_thread")]
@@ -187,7 +190,7 @@ mod tests {
         let downloader = BodiesDownloaderBuilder::default().build(
             client.clone(),
             Arc::new(TestConsensus::default()),
-            db,
+            ProviderFactory::new(db, MAINNET.clone()),
         );
         let mut downloader = TaskDownloader::spawn(downloader);
 
@@ -209,7 +212,7 @@ mod tests {
         let downloader = BodiesDownloaderBuilder::default().build(
             Arc::new(TestBodiesClient::default()),
             Arc::new(TestConsensus::default()),
-            db,
+            ProviderFactory::new(db, MAINNET.clone()),
         );
         let mut downloader = TaskDownloader::spawn(downloader);
 

crates/net/downloaders/src/test_utils/file_client.rs

@@ -267,7 +267,8 @@ mod tests {
         },
         test_utils::TestConsensus,
     };
-    use reth_primitives::SealedHeader;
+    use reth_primitives::{SealedHeader, MAINNET};
+    use reth_provider::ProviderFactory;
     use std::{
         io::{Read, Seek, SeekFrom, Write},
         sync::Arc,
@@ -291,7 +292,7 @@ mod tests {
         let mut downloader = BodiesDownloaderBuilder::default().build(
             client.clone(),
             Arc::new(TestConsensus::default()),
-            db,
+            ProviderFactory::new(db, MAINNET.clone()),
         );
         downloader.set_download_range(0..=19).expect("failed to set download range");
 
@@ -373,7 +374,7 @@ mod tests {
         let mut downloader = BodiesDownloaderBuilder::default().build(
             client.clone(),
             Arc::new(TestConsensus::default()),
-            db,
+            ProviderFactory::new(db, MAINNET.clone()),
         );
         downloader.set_download_range(0..=19).expect("failed to set download range");
 

crates/stages/src/lib.rs

@@ -36,7 +36,7 @@
 //! # let bodies_downloader = BodiesDownloaderBuilder::default().build(
 //! #    Arc::new(TestBodiesClient { responder: |_| Ok((PeerId::ZERO, vec![]).into()) }),
 //! #    consensus.clone(),
-//! #    db.clone()
+//! #    ProviderFactory::new(db.clone(), MAINNET.clone())
 //! # );
 //! # let (tip_tx, tip_rx) = watch::channel(B256::default());
 //! # let factory = Factory::new(chain_spec.clone());

crates/stages/src/stages/bodies.rs

@@ -484,7 +484,7 @@ mod tests {
             tables,
             test_utils::TempDatabase,
             transaction::{DbTx, DbTxMut},
-            DatabaseEnv,
+            DatabaseEnv, DatabaseError,
         };
         use reth_interfaces::{
             p2p::{
@@ -494,7 +494,7 @@ mod tests {
                     response::BlockResponse,
                 },
                 download::DownloadClient,
-                error::DownloadResult,
+                error::{DownloadError, DownloadResult},
                 priority::Priority,
             },
             test_utils::{
@@ -780,11 +780,13 @@ mod tests {
                 &mut self,
                 range: RangeInclusive<BlockNumber>,
             ) -> DownloadResult<()> {
-                self.headers =
+                self.headers = VecDeque::from(
-                    VecDeque::from(self.db.view(|tx| -> DownloadResult<Vec<SealedHeader>> {
+                    self.db
+                        .view(|tx| -> Result<Vec<SealedHeader>, DatabaseError> {
                             let mut header_cursor = tx.cursor_read::<tables::Headers>()?;
 
-                        let mut canonical_cursor = tx.cursor_read::<tables::CanonicalHeaders>()?;
+                            let mut canonical_cursor =
+                                tx.cursor_read::<tables::CanonicalHeaders>()?;
                             let walker = canonical_cursor.walk_range(range)?;
 
                             let mut headers = Vec::default();
@@ -795,7 +797,10 @@ mod tests {
                                 headers.push(header.seal(hash));
                             }
                             Ok(headers)
-                    })??);
+                        })
+                        .map_err(|err| DownloadError::Provider(err.into()))?
+                        .map_err(|err| DownloadError::Provider(err.into()))?,
+                );
                 Ok(())
             }
         }