Merge of #7751

Author: mergify[bot]

beacon_node/beacon_chain/src/builder.rs

@@ -899,6 +899,7 @@ where
         let genesis_time = head_snapshot.beacon_state.genesis_time();
         let canonical_head = CanonicalHead::new(fork_choice, Arc::new(head_snapshot));
         let shuffling_cache_size = self.chain_config.shuffling_cache_size;
+        let complete_blob_backfill = self.chain_config.complete_blob_backfill;
 
         // Calculate the weak subjectivity point in which to backfill blocks to.
         let genesis_backfill_slot = if self.chain_config.genesis_backfill {
@@ -1013,6 +1014,7 @@ where
             genesis_backfill_slot,
             data_availability_checker: Arc::new(
                 DataAvailabilityChecker::new(
+                    complete_blob_backfill,
                     slot_clock,
                     self.kzg.clone(),
                     store,

beacon_node/beacon_chain/src/chain_config.rs

@@ -86,6 +86,8 @@ pub struct ChainConfig {
     /// If using a weak-subjectivity sync, whether we should download blocks all the way back to
     /// genesis.
     pub genesis_backfill: bool,
+    /// EXPERIMENTAL: backfill blobs and data columns beyond the data availability window.
+    pub complete_blob_backfill: bool,
     /// Whether to send payload attributes every slot, regardless of connected proposers.
     ///
     /// This is useful for block builders and testing.
@@ -144,6 +146,7 @@ impl Default for ChainConfig {
             optimistic_finalized_sync: true,
             shuffling_cache_size: crate::shuffling_cache::DEFAULT_CACHE_SIZE,
             genesis_backfill: false,
+            complete_blob_backfill: false,
             always_prepare_payload: false,
             epochs_per_migration: crate::migrate::DEFAULT_EPOCHS_PER_MIGRATION,
             enable_light_client_server: true,

beacon_node/beacon_chain/src/data_availability_checker.rs

@@ -78,6 +78,7 @@ pub const STATE_LRU_CAPACITY: usize = STATE_LRU_CAPACITY_NON_ZERO.get();
 /// proposer. Having a capacity > 1 is an optimization to prevent sync lookup from having re-fetch
 /// data during moments of unstable network conditions.
 pub struct DataAvailabilityChecker<T: BeaconChainTypes> {
+    complete_blob_backfill: bool,
     availability_cache: Arc<DataAvailabilityCheckerInner<T>>,
     slot_clock: T::SlotClock,
     kzg: Arc<Kzg>,
@@ -116,6 +117,7 @@ impl<E: EthSpec> Debug for Availability<E> {
 
 impl<T: BeaconChainTypes> DataAvailabilityChecker<T> {
     pub fn new(
+        complete_blob_backfill: bool,
         slot_clock: T::SlotClock,
         kzg: Arc<Kzg>,
         store: BeaconStore<T>,
@@ -129,6 +131,7 @@ impl<T: BeaconChainTypes> DataAvailabilityChecker<T> {
             spec.clone(),
         )?;
         Ok(Self {
+            complete_blob_backfill,
             availability_cache: Arc::new(inner),
             slot_clock,
             kzg,
@@ -518,9 +521,15 @@ impl<T: BeaconChainTypes> DataAvailabilityChecker<T> {
     /// The epoch at which we require a data availability check in block processing.
     /// `None` if the `Deneb` fork is disabled.
     pub fn data_availability_boundary(&self) -> Option<Epoch> {
-        let current_epoch = self.slot_clock.now()?.epoch(T::EthSpec::slots_per_epoch());
-        self.spec
-            .min_epoch_data_availability_boundary(current_epoch)
+        let fork_epoch = self.spec.deneb_fork_epoch?;
+
+        if self.complete_blob_backfill {
+            Some(fork_epoch)
+        } else {
+            let current_epoch = self.slot_clock.now()?.epoch(T::EthSpec::slots_per_epoch());
+            self.spec
+                .min_epoch_data_availability_boundary(current_epoch)
+        }
     }
 
     /// Returns true if the given epoch lies within the da boundary and false otherwise.
@@ -1076,7 +1085,15 @@ mod test {
         let kzg = get_kzg(&spec);
         let store = Arc::new(HotColdDB::open_ephemeral(<_>::default(), spec.clone()).unwrap());
         let custody_context = Arc::new(CustodyContext::new(false));
-        DataAvailabilityChecker::new(slot_clock, kzg, store, custody_context, spec)
-            .expect("should initialise data availability checker")
+        let complete_blob_backfill = false;
+        DataAvailabilityChecker::new(
+            complete_blob_backfill,
+            slot_clock,
+            kzg,
+            store,
+            custody_context,
+            spec,
+        )
+        .expect("should initialise data availability checker")
     }
 }

beacon_node/src/cli.rs

@@ -401,6 +401,16 @@ pub fn cli_app() -> Command {
                 .help_heading(FLAG_HEADER)
                 .display_order(0)
         )
+        .arg(
+            Arg::new("complete-blob-backfill")
+                .long("complete-blob-backfill")
+                .help("Download all blobs back to the Deneb fork epoch. This will likely result in \
+                       the node banning most of its peers.")
+                .action(ArgAction::SetTrue)
+                .help_heading(FLAG_HEADER)
+                .display_order(0)
+                .hide(true)
+        )
         .arg(
             Arg::new("enable-private-discovery")
                 .long("enable-private-discovery")

beacon_node/src/config.rs

@@ -825,6 +825,14 @@ pub fn get_config<E: EthSpec>(
         client_config.chain.genesis_backfill = true;
     }
 
+    client_config.chain.complete_blob_backfill = cli_args.get_flag("complete-blob-backfill");
+
+    // Ensure `prune_blobs` is false whenever complete-blob-backfill is set. This overrides any
+    // setting of `--prune-blobs true` applied earlier in flag parsing.
+    if client_config.chain.complete_blob_backfill {
+        client_config.store.prune_blobs = false;
+    }
+
     // Backfill sync rate-limiting
     client_config.beacon_processor.enable_backfill_rate_limiting =
         !cli_args.get_flag("disable-backfill-rate-limiting");

lighthouse/tests/beacon_node.rs

@@ -392,6 +392,37 @@ fn genesis_backfill_with_historic_flag() {
         .with_config(|config| assert!(config.chain.genesis_backfill));
 }
 
+#[test]
+fn complete_blob_backfill_default() {
+    CommandLineTest::new()
+        .run_with_zero_port()
+        .with_config(|config| assert!(!config.chain.complete_blob_backfill));
+}
+
+#[test]
+fn complete_blob_backfill_flag() {
+    CommandLineTest::new()
+        .flag("complete-blob-backfill", None)
+        .run_with_zero_port()
+        .with_config(|config| {
+            assert!(config.chain.complete_blob_backfill);
+            assert!(!config.store.prune_blobs);
+        });
+}
+
+// Even if `--prune-blobs true` is provided, `--complete-blob-backfill` should override it to false.
+#[test]
+fn complete_blob_backfill_and_prune_blobs_true() {
+    CommandLineTest::new()
+        .flag("complete-blob-backfill", None)
+        .flag("prune-blobs", Some("true"))
+        .run_with_zero_port()
+        .with_config(|config| {
+            assert!(config.chain.complete_blob_backfill);
+            assert!(!config.store.prune_blobs);
+        });
+}
+
 // Tests for Eth1 flags.
 // DEPRECATED but should not crash
 #[test]