Lazy load origin children (kaspanet#518)

* Lazy load origin children

* remove redundant collect

* Delete level-0 relations for blocks which only belong to higher proof levels

* Comments

* Edit comment

consensus/src/model/services/reachability.rs

@@ -17,6 +17,7 @@ pub trait ReachabilityService {
     fn is_any_dag_ancestor_result(&self, list: &mut impl Iterator<Item = Hash>, queried: Hash) -> Result<bool>;
     fn get_next_chain_ancestor(&self, descendant: Hash, ancestor: Hash) -> Hash;
     fn get_chain_parent(&self, this: Hash) -> Hash;
+    fn has_reachability_data(&self, this: Hash) -> bool;
 }
 
 impl<T: ReachabilityStoreReader + ?Sized> ReachabilityService for T {
@@ -56,6 +57,10 @@ impl<T: ReachabilityStoreReader + ?Sized> ReachabilityService for T {
     fn get_chain_parent(&self, this: Hash) -> Hash {
         self.get_parent(this).unwrap()
     }
+
+    fn has_reachability_data(&self, this: Hash) -> bool {
+        self.has(this).unwrap()
+    }
 }
 
 /// Multi-threaded reachability service imp
@@ -108,6 +113,10 @@ impl<T: ReachabilityStoreReader + ?Sized> ReachabilityService for MTReachability
     fn get_chain_parent(&self, this: Hash) -> Hash {
         self.store.read().get_parent(this).unwrap()
     }
+
+    fn has_reachability_data(&self, this: Hash) -> bool {
+        self.store.read().has(this).unwrap()
+    }
 }
 
 impl<T: ReachabilityStoreReader + ?Sized> MTReachabilityService<T> {

consensus/src/pipeline/pruning_processor/processor.rs

@@ -268,6 +268,12 @@ impl PruningProcessor {
             .chain(data.ghostdag_blocks.iter().map(|gd| gd.hash))
             .chain(proof.iter().flatten().map(|h| h.hash))
             .collect();
+        let keep_level_zero_relations: BlockHashSet = std::iter::empty()
+            .chain(data.anticone.iter().copied())
+            .chain(data.daa_window_blocks.iter().map(|th| th.header.hash))
+            .chain(data.ghostdag_blocks.iter().map(|gd| gd.hash))
+            .chain(proof[0].iter().map(|h| h.hash))
+            .collect();
         let keep_headers: BlockHashSet = self.past_pruning_points();
 
         info!("Header and Block pruning: waiting for consensus write permissions...");
@@ -281,16 +287,16 @@ impl PruningProcessor {
         {
             let mut counter = 0;
             let mut batch = WriteBatch::default();
-            for kept in keep_relations.iter().copied() {
+            for kept in keep_level_zero_relations.iter().copied() {
                 let Some(ghostdag) = self.ghostdag_primary_store.get_data(kept).unwrap_option() else {
                     continue;
                 };
-                if ghostdag.unordered_mergeset().any(|h| !keep_relations.contains(&h)) {
+                if ghostdag.unordered_mergeset().any(|h| !keep_level_zero_relations.contains(&h)) {
                     let mut mutable_ghostdag: ExternalGhostdagData = ghostdag.as_ref().into();
-                    mutable_ghostdag.mergeset_blues.retain(|h| keep_relations.contains(h));
-                    mutable_ghostdag.mergeset_reds.retain(|h| keep_relations.contains(h));
-                    mutable_ghostdag.blues_anticone_sizes.retain(|k, _| keep_relations.contains(k));
-                    if !keep_relations.contains(&mutable_ghostdag.selected_parent) {
+                    mutable_ghostdag.mergeset_blues.retain(|h| keep_level_zero_relations.contains(h));
+                    mutable_ghostdag.mergeset_reds.retain(|h| keep_level_zero_relations.contains(h));
+                    mutable_ghostdag.blues_anticone_sizes.retain(|k, _| keep_level_zero_relations.contains(k));
+                    if !keep_level_zero_relations.contains(&mutable_ghostdag.selected_parent) {
                         mutable_ghostdag.selected_parent = ORIGIN;
                     }
                     counter += 1;
@@ -396,6 +402,19 @@ impl PruningProcessor {
                         // other parts of the code assume the existence of GD data etc.)
                         statuses_write.set_batch(&mut batch, current, StatusHeaderOnly).unwrap();
                     }
+
+                    // Delete level-0 relations for blocks which only belong to higher proof levels.
+                    // Note: it is also possible to delete level relations for level x > 0 for any block that only belongs
+                    // to proof levels higher than x, but this requires maintaining such per level usage mapping.
+                    // Since the main motivation of this deletion step is to reduce the
+                    // number of origin's children in level 0, and this is not a bottleneck in any other
+                    // level, we currently chose to only delete level-0 redundant relations.
+                    if !keep_level_zero_relations.contains(&current) {
+                        let mut staging_level_relations = StagingRelationsStore::new(&mut level_relations_write[0]);
+                        relations::delete_level_relations(MemoryWriter, &mut staging_level_relations, current).unwrap_option();
+                        staging_level_relations.commit(&mut batch).unwrap();
+                        self.ghostdag_stores[0].delete_batch(&mut batch, current).unwrap_option();
+                    }
                 } else {
                     // Count only blocks which get fully pruned including DAG relations
                     counter += 1;

consensus/src/processes/parents_builder.rs

@@ -10,8 +10,6 @@ use crate::model::{
     stores::{headers::HeaderStoreReader, reachability::ReachabilityStoreReader, relations::RelationsStoreReader},
 };
 
-use super::reachability::ReachabilityResultExtensions;
-
 #[derive(Clone)]
 pub struct ParentsManager<T: HeaderStoreReader, U: ReachabilityStoreReader, V: RelationsStoreReader> {
     max_block_level: BlockLevel,
@@ -52,10 +50,7 @@ impl<T: HeaderStoreReader, U: ReachabilityStoreReader, V: RelationsStoreReader>
             .expect("at least one of the parents is expected to be in the future of the pruning point");
         direct_parent_headers.swap(0, first_parent_in_future_of_pruning_point);
 
-        let origin_children = self.relations_service.get_children(ORIGIN).unwrap().read().iter().copied().collect_vec();
-        let origin_children_headers =
-            origin_children.iter().copied().map(|parent| self.headers_store.get_header(parent).unwrap()).collect_vec();
-
+        let mut origin_children_headers = None;
         let mut parents = Vec::with_capacity(self.max_block_level as usize);
 
         for block_level in 0..self.max_block_level {
@@ -97,11 +92,7 @@ impl<T: HeaderStoreReader, U: ReachabilityStoreReader, V: RelationsStoreReader>
             };
 
             for (i, parent) in grandparents.into_iter().enumerate() {
-                let is_in_origin_children_future = self
-                    .reachability_service
-                    .is_any_dag_ancestor_result(&mut origin_children.iter().copied(), parent)
-                    .unwrap_option()
-                    .is_some_and(|r| r);
+                let has_reachability_data = self.reachability_service.has_reachability_data(parent);
 
                 // Reference blocks are the blocks that are used in reachability queries to check if
                 // a candidate is in the future of another candidate. In most cases this is just the
@@ -110,13 +101,24 @@ impl<T: HeaderStoreReader, U: ReachabilityStoreReader, V: RelationsStoreReader>
                 // If we make sure to add a parent in the future of the pruning point first, we can
                 // know that any pruned candidate that is in the past of some blocks in the pruning
                 // point anticone should be a parent (in the relevant level) of one of
-                // the virtual genesis children in the pruning point anticone. So we can check which
-                // virtual genesis children have this block as parent and use those block as
+                // the origin children in the pruning point anticone. So we can check which
+                // origin children have this block as parent and use those block as
                 // reference blocks.
-                let reference_blocks = if is_in_origin_children_future {
+                let reference_blocks = if has_reachability_data {
                     smallvec![parent]
                 } else {
-                    let mut reference_blocks = SmallVec::with_capacity(origin_children.len());
+                    // Here we explicitly declare the type because otherwise Rust would make it mutable.
+                    let origin_children_headers: &Vec<_> = origin_children_headers.get_or_insert_with(|| {
+                        self.relations_service
+                            .get_children(ORIGIN)
+                            .unwrap()
+                            .read()
+                            .iter()
+                            .copied()
+                            .map(|parent| self.headers_store.get_header(parent).unwrap())
+                            .collect_vec()
+                    });
+                    let mut reference_blocks = SmallVec::with_capacity(origin_children_headers.len());
                     for child_header in origin_children_headers.iter() {
                         if self.parents_at_level(child_header, block_level).contains(&parent) {
                             reference_blocks.push(child_header.hash);
@@ -133,7 +135,7 @@ impl<T: HeaderStoreReader, U: ReachabilityStoreReader, V: RelationsStoreReader>
                     continue;
                 }
 
-                if !is_in_origin_children_future {
+                if !has_reachability_data {
                     continue;
                 }