Squashed commit of the following (sigp#8188):

commit aa64356
Author: Pawan Dhananjay <[email protected]>
Date:   Fri Oct 10 15:30:45 2025 -0700

    lint

commit 9fab592
Author: Pawan Dhananjay <[email protected]>
Date:   Fri Oct 10 15:10:05 2025 -0700

    Get claude to write tests

commit bd181a3
Author: Pawan Dhananjay <[email protected]>
Date:   Fri Oct 10 13:34:41 2025 -0700

    Only persist custody columns

Author: jimmygchen

beacon_node/beacon_chain/src/beacon_chain.rs

@@ -3957,7 +3957,7 @@ impl<T: BeaconChainTypes> BeaconChain<T> {
         // See https://github.com/sigp/lighthouse/issues/2028
         let (_, signed_block, block_data) = signed_block.deconstruct();
 
-        match self.get_blobs_or_columns_store_op(block_root, block_data) {
+        match self.get_blobs_or_columns_store_op(block_root, signed_block.slot(), block_data) {
             Ok(Some(blobs_or_columns_store_op)) => {
                 ops.push(blobs_or_columns_store_op);
             }
@@ -7163,6 +7163,7 @@ impl<T: BeaconChainTypes> BeaconChain<T> {
     pub(crate) fn get_blobs_or_columns_store_op(
         &self,
         block_root: Hash256,
+        block_slot: Slot,
         block_data: AvailableBlockData<T::EthSpec>,
     ) -> Result<Option<StoreOp<'_, T::EthSpec>>, String> {
         match block_data {
@@ -7175,7 +7176,15 @@ impl<T: BeaconChainTypes> BeaconChain<T> {
                 );
                 Ok(Some(StoreOp::PutBlobs(block_root, blobs)))
             }
-            AvailableBlockData::DataColumns(data_columns) => {
+            AvailableBlockData::DataColumns(mut data_columns) => {
+                let columns_to_custody = self.custody_columns_for_epoch(Some(
+                    block_slot.epoch(T::EthSpec::slots_per_epoch()),
+                ));
+                // Supernodes need to persist all sampled custody columns
+                if columns_to_custody.len() != self.spec.number_of_custody_groups as usize {
+                    data_columns
+                        .retain(|data_column| columns_to_custody.contains(&data_column.index));
+                }
                 debug!(
                     %block_root,
                     count = data_columns.len(),

beacon_node/beacon_chain/src/historical_blocks.rs

@@ -140,7 +140,7 @@ impl<T: BeaconChainTypes> BeaconChain<T> {
 
             // Store the blobs or data columns too
             if let Some(op) = self
-                .get_blobs_or_columns_store_op(block_root, block_data)
+                .get_blobs_or_columns_store_op(block_root, block.slot(), block_data)
                 .map_err(|e| {
                     HistoricalBlockError::StoreError(StoreError::DBError {
                         message: format!("get_blobs_or_columns_store_op error {e:?}"),

beacon_node/beacon_chain/tests/store_tests.rs

@@ -13,7 +13,8 @@ use beacon_chain::test_utils::{
 use beacon_chain::{
     BeaconChain, BeaconChainError, BeaconChainTypes, BeaconSnapshot, BlockError, ChainConfig,
     NotifyExecutionLayer, ServerSentEventHandler, WhenSlotSkipped,
-    data_availability_checker::MaybeAvailableBlock, historical_blocks::HistoricalBlockError,
+    data_availability_checker::{AvailableBlockData, MaybeAvailableBlock},
+    historical_blocks::HistoricalBlockError,
     migrate::MigratorConfig,
 };
 use logging::create_test_tracing_subscriber;
@@ -4137,6 +4138,195 @@ async fn replay_from_split_state() {
     assert_eq!(state.slot(), split.slot);
 }
 
+/// Test that regular nodes filter and store only custody columns when processing blocks with data columns.
+#[tokio::test]
+async fn test_custody_column_filtering_regular_node() {
+    let db_path = tempdir().unwrap();
+    let store = get_store(&db_path);
+    let harness = get_harness(store.clone(), LOW_VALIDATOR_COUNT);
+
+    // Skip test if PeerDAS is not scheduled
+    if !harness.spec.is_peer_das_scheduled() {
+        return;
+    }
+
+    // Generate a block with data columns using public test utils
+    let fork_name = harness.spec.fork_name_at_epoch(Epoch::new(0));
+    let mut rng = rand::rng();
+    let (signed_block, all_data_columns) =
+        beacon_chain::test_utils::generate_rand_block_and_data_columns::<E>(
+            fork_name,
+            beacon_chain::test_utils::NumBlobs::Number(3), // Generate 3 blobs to produce data columns
+            &mut rng,
+            &harness.spec,
+        );
+
+    // Skip test if no data columns are generated
+    if all_data_columns.is_empty() {
+        return;
+    }
+
+    let block_root = signed_block.canonical_root();
+    let slot = signed_block.slot();
+
+    // Get custody columns for this epoch - regular nodes only store a subset
+    let custody_columns = harness
+        .chain
+        .custody_columns_for_epoch(Some(slot.epoch(E::slots_per_epoch())));
+
+    // Verify that custody columns is a proper subset (not all columns for regular node)
+    assert!(
+        custody_columns.len() < harness.spec.number_of_custody_groups as usize,
+        "Regular node should not custody all columns"
+    );
+
+    // Create AvailableBlock with all data columns
+    let available_block_data = AvailableBlockData::DataColumns(all_data_columns.clone());
+    let available_block = AvailableBlock::__new_for_testing(
+        block_root,
+        Arc::new(signed_block),
+        available_block_data,
+        harness.spec.clone(),
+    );
+
+    // Import the block through the historical block batch import which calls the custody filtering internally
+    harness
+        .chain
+        .import_historical_block_batch(vec![available_block])
+        .expect("should import block with data columns");
+
+    // Check what actually got stored in the database
+    let stored_column_keys = store
+        .get_data_column_keys(block_root)
+        .expect("should get stored column keys");
+
+    // Verify only custody columns were stored
+    let stored_column_indices: std::collections::HashSet<_> =
+        stored_column_keys.into_iter().collect();
+    let expected_column_indices: std::collections::HashSet<_> =
+        custody_columns.iter().copied().collect();
+
+    assert_eq!(
+        stored_column_indices, expected_column_indices,
+        "Regular node should only store custody columns"
+    );
+
+    // Verify no non-custody columns are included
+    let all_column_indices: std::collections::HashSet<_> =
+        all_data_columns.iter().map(|dc| dc.index).collect();
+    let non_custody_columns: std::collections::HashSet<_> = all_column_indices
+        .difference(&expected_column_indices)
+        .collect();
+
+    for &non_custody_column in &non_custody_columns {
+        assert!(
+            !stored_column_indices.contains(non_custody_column),
+            "Non-custody column {} should not be stored",
+            non_custody_column
+        );
+    }
+}
+
+/// Test that supernodes store all data columns when processing blocks with data columns.
+#[tokio::test]
+async fn test_custody_column_filtering_supernode() {
+    let db_path = tempdir().unwrap();
+    let store = get_store(&db_path);
+    let harness = get_harness_import_all_data_columns(store.clone(), LOW_VALIDATOR_COUNT);
+
+    // Skip test if PeerDAS is not scheduled
+    if !harness.spec.is_peer_das_scheduled() {
+        return;
+    }
+
+    // Generate a block with data columns using public test utils
+    let fork_name = harness.spec.fork_name_at_epoch(Epoch::new(0));
+    let mut rng = rand::rng();
+    let (signed_block, all_data_columns) =
+        beacon_chain::test_utils::generate_rand_block_and_data_columns::<E>(
+            fork_name,
+            beacon_chain::test_utils::NumBlobs::Number(3), // Generate 3 blobs to produce data columns
+            &mut rng,
+            &harness.spec,
+        );
+
+    // Skip test if no data columns are generated
+    if all_data_columns.is_empty() {
+        return;
+    }
+
+    let block_root = signed_block.canonical_root();
+
+    // Create AvailableBlock with all data columns
+    let available_block_data = AvailableBlockData::DataColumns(all_data_columns.clone());
+    let available_block = AvailableBlock::__new_for_testing(
+        block_root,
+        Arc::new(signed_block),
+        available_block_data,
+        harness.spec.clone(),
+    );
+
+    // Import the block through the historical block batch import
+    harness
+        .chain
+        .import_historical_block_batch(vec![available_block])
+        .expect("should import block with data columns");
+
+    // Check what actually got stored in the database
+    let stored_column_keys = store
+        .get_data_column_keys(block_root)
+        .expect("should get stored column keys");
+
+    // Verify ALL data columns were stored for supernodes
+    let stored_column_indices: std::collections::HashSet<_> =
+        stored_column_keys.into_iter().collect();
+    let all_column_indices: std::collections::HashSet<_> =
+        all_data_columns.iter().map(|dc| dc.index).collect();
+
+    assert_eq!(
+        stored_column_indices, all_column_indices,
+        "Supernode should store ALL data columns"
+    );
+
+    // Verify the count matches
+    assert_eq!(
+        stored_column_indices.len(),
+        all_data_columns.len(),
+        "Supernode should store the same number of columns as generated"
+    );
+
+    // Verify each stored column can be retrieved and matches original data
+    let stored_columns = store
+        .get_data_columns(&block_root)
+        .expect("should get data columns")
+        .expect("data columns should exist");
+
+    assert_eq!(
+        stored_columns.len(),
+        all_data_columns.len(),
+        "All data columns should be retrievable"
+    );
+
+    for original_column in &all_data_columns {
+        let matching_stored = stored_columns
+            .iter()
+            .find(|stored| stored.index == original_column.index);
+
+        assert!(
+            matching_stored.is_some(),
+            "Column {} should be present in stored data",
+            original_column.index
+        );
+
+        let stored_column = matching_stored.unwrap();
+        assert_eq!(
+            stored_column, original_column,
+            "Stored column {} should match original",
+            original_column.index
+        );
+    }
+}
+
 /// Checks that two chains are the same, for the purpose of these tests.
 ///
 /// Several fields that are hard/impossible to check are ignored (e.g., the store).