feat: report world state size on disk (#13706)

Closes #13200

Adds reporting for tracking the disk space currently used by the current
world state.

Author: natebeauregard

barretenberg/cpp/src/barretenberg/crypto/merkle_tree/lmdb_store/lmdb_tree_store.cpp

@@ -103,6 +103,7 @@ const std::string& LMDBTreeStore::get_name() const
 void LMDBTreeStore::get_stats(TreeDBStats& stats, ReadTransaction& tx)
 {
     stats.mapSize = _environment->get_map_size();
+    stats.physicalFileSize = _environment->get_data_file_size();
     stats.blocksDBStats = _blockDatabase->get_stats(tx);
     stats.leafPreimagesDBStats = _leafHashToPreImageDatabase->get_stats(tx);
     stats.leafIndicesDBStats = _leafKeyToIndexDatabase->get_stats(tx);

barretenberg/cpp/src/barretenberg/crypto/merkle_tree/lmdb_store/lmdb_tree_store.test.cpp

@@ -585,3 +585,49 @@ TEST_F(LMDBTreeStoreTest, can_write_and_retrieve_block_numbers_with_duplicate_in
         EXPECT_EQ(readBack, 5);
     }
 }
+
+TEST_F(LMDBTreeStoreTest, reports_physical_file_size)
+{
+    LMDBTreeStore store(_directory, "DB1", _mapSize, _maxReaders);
+    std::string dataDbPath = _directory + "/data.mdb";
+    size_t previousFileSize = 0;
+
+    for (size_t i = 0; i < 3; i++) {
+        {
+            BlockPayload blockData;
+            blockData.blockNumber = i;
+            blockData.root = VALUES[i];
+            blockData.size = 45 + (i * 97);
+
+            TreeMeta metaData;
+            metaData.committedSize = blockData.size;
+            metaData.size = blockData.size;
+            metaData.root = blockData.root;
+            metaData.depth = 32;
+            metaData.unfinalisedBlockHeight = i;
+            metaData.name = "NullifierTree";
+
+            // Write metadata and block data with different values each iteration
+            LMDBWriteTransaction::Ptr transaction = store.create_write_transaction();
+            store.write_meta_data(metaData, *transaction);
+            store.write_block_data(i, blockData, *transaction);
+            transaction->commit();
+        }
+
+        {
+            LMDBReadTransaction::Ptr transaction = store.create_read_transaction();
+            TreeDBStats stats;
+            store.get_stats(stats, *transaction);
+
+            EXPECT_TRUE(std::filesystem::exists(dataDbPath));
+
+            // Verify reported size matches actual file size
+            EXPECT_EQ(stats.physicalFileSize, std::filesystem::file_size(dataDbPath));
+
+            // Verify size is increasing due to the new DB writes
+            EXPECT_GT(stats.physicalFileSize, previousFileSize);
+
+            previousFileSize = stats.physicalFileSize;
+        }
+    }
+}

barretenberg/cpp/src/barretenberg/crypto/merkle_tree/types.hpp

@@ -67,23 +67,27 @@ const std::string BLOCK_INDICES_DB = "block indices";
 
 struct TreeDBStats {
     uint64_t mapSize;
+    uint64_t physicalFileSize;
     DBStats blocksDBStats;
     DBStats nodesDBStats;
     DBStats leafPreimagesDBStats;
     DBStats leafIndicesDBStats;
     DBStats blockIndicesDBStats;
 
     TreeDBStats() = default;
-    TreeDBStats(uint64_t mapSize)
+    TreeDBStats(uint64_t mapSize, uint64_t physicalFileSize)
         : mapSize(mapSize)
+        , physicalFileSize(physicalFileSize)
     {}
     TreeDBStats(uint64_t mapSize,
+                uint64_t physicalFileSize,
                 const DBStats& blockStats,
                 const DBStats& nodesStats,
                 const DBStats& leafPreimagesDBStats,
                 const DBStats& leafIndicesStats,
                 const DBStats& blockIndicesStats)
         : mapSize(mapSize)
+        , physicalFileSize(physicalFileSize)
         , blocksDBStats(blockStats)
         , nodesDBStats(nodesStats)
         , leafPreimagesDBStats(leafPreimagesDBStats)
@@ -95,11 +99,18 @@ struct TreeDBStats {
 
     ~TreeDBStats() = default;
 
-    MSGPACK_FIELDS(mapSize, blocksDBStats, nodesDBStats, leafPreimagesDBStats, leafIndicesDBStats, blockIndicesDBStats)
+    MSGPACK_FIELDS(mapSize,
+                   physicalFileSize,
+                   blocksDBStats,
+                   nodesDBStats,
+                   leafPreimagesDBStats,
+                   leafIndicesDBStats,
+                   blockIndicesDBStats)
 
     bool operator==(const TreeDBStats& other) const
     {
-        return mapSize == other.mapSize && blocksDBStats == other.blocksDBStats && nodesDBStats == other.nodesDBStats &&
+        return mapSize == other.mapSize && physicalFileSize == other.physicalFileSize &&
+               blocksDBStats == other.blocksDBStats && nodesDBStats == other.nodesDBStats &&
                leafPreimagesDBStats == other.leafPreimagesDBStats && leafIndicesDBStats == other.leafIndicesDBStats &&
                blockIndicesDBStats == other.blockIndicesDBStats;
     }
@@ -108,6 +119,7 @@ struct TreeDBStats {
     {
         if (this != &other) {
             mapSize = other.mapSize;
+            physicalFileSize = other.physicalFileSize;
             blocksDBStats = std::move(other.blocksDBStats);
             nodesDBStats = std::move(other.nodesDBStats);
             leafPreimagesDBStats = std::move(other.leafPreimagesDBStats);
@@ -121,9 +133,10 @@ struct TreeDBStats {
 
     friend std::ostream& operator<<(std::ostream& os, const TreeDBStats& stats)
     {
-        os << "Map Size: " << stats.mapSize << " Blocks DB " << stats.blocksDBStats << ", Nodes DB "
-           << stats.nodesDBStats << ", Leaf Pre-images DB " << stats.leafPreimagesDBStats << ", Leaf Indices DB "
-           << stats.leafIndicesDBStats << ", Block Indices DB " << stats.blockIndicesDBStats;
+        os << "Map Size: " << stats.mapSize << ", Physical File Size: " << stats.physicalFileSize << " Blocks DB "
+           << stats.blocksDBStats << ", Nodes DB " << stats.nodesDBStats << ", Leaf Pre-images DB "
+           << stats.leafPreimagesDBStats << ", Leaf Indices DB " << stats.leafIndicesDBStats << ", Block Indices DB "
+           << stats.blockIndicesDBStats;
         return os;
     }
 };

barretenberg/cpp/src/barretenberg/lmdblib/lmdb_environment.cpp

@@ -2,6 +2,7 @@
 #include "barretenberg/lmdblib/lmdb_helpers.hpp"
 #include "lmdb.h"
 #include <cstdint>
+#include <filesystem>
 #include <stdexcept>
 #include <sys/stat.h>
 
@@ -12,6 +13,7 @@ LMDBEnvironment::LMDBEnvironment(const std::string& directory,
                                  uint32_t maxNumDBs,
                                  uint32_t maxNumReaders)
     : _id(0)
+    , _directory(directory)
     , _readGuard(maxNumReaders)
     , _writeGuard(1) // LMDB only permits one write transaction at a time
 {
@@ -71,4 +73,13 @@ uint64_t LMDBEnvironment::get_map_size() const
     call_lmdb_func(mdb_env_info, _mdbEnv, &info);
     return info.me_mapsize;
 }
+
+uint64_t LMDBEnvironment::get_data_file_size() const
+{
+    std::string dataPath = (std::filesystem::path(_directory) / "data.mdb").string();
+    if (std::filesystem::exists(dataPath)) {
+        return std::filesystem::file_size(dataPath);
+    }
+    return 0;
+}
 } // namespace bb::lmdblib

barretenberg/cpp/src/barretenberg/lmdblib/lmdb_environment.hpp

@@ -48,8 +48,11 @@ class LMDBEnvironment {
 
     uint64_t get_map_size() const;
 
+    uint64_t get_data_file_size() const;
+
   private:
     std::atomic_uint64_t _id;
+    std::string _directory;
     MDB_env* _mdbEnv;
 
     struct ResourceGuard {
@@ -82,4 +85,4 @@ class LMDBEnvironment {
     ResourceGuard _readGuard;
     ResourceGuard _writeGuard;
 };
-} // namespace bb::lmdblib
+} // namespace bb::lmdblib

yarn-project/telemetry-client/src/metrics.ts

@@ -150,6 +150,7 @@ export const WORLD_STATE_SYNC_DURATION = 'aztec.world_state.sync.duration';
 export const WORLD_STATE_MERKLE_TREE_SIZE = 'aztec.world_state.merkle_tree_size';
 export const WORLD_STATE_DB_SIZE = 'aztec.world_state.db_size';
 export const WORLD_STATE_DB_MAP_SIZE = 'aztec.world_state.db_map_size';
+export const WORLD_STATE_DB_PHYSICAL_SIZE = 'aztec.world_state.db_physical_size';
 export const WORLD_STATE_TREE_SIZE = 'aztec.world_state.tree_size';
 export const WORLD_STATE_UNFINALISED_HEIGHT = 'aztec.world_state.unfinalised_height';
 export const WORLD_STATE_FINALISED_HEIGHT = 'aztec.world_state.finalised_height';

yarn-project/world-state/src/instrumentation/instrumentation.ts

@@ -31,6 +31,7 @@ const durationTrackDenylist = new Set<WorldStateMessageType>([
 
 export class WorldStateInstrumentation {
   private dbMapSize: Gauge;
+  private dbPhysicalSize: Gauge;
   private treeSize: Gauge;
   private unfinalisedHeight: Gauge;
   private finalisedHeight: Gauge;
@@ -50,6 +51,11 @@ export class WorldStateInstrumentation {
       valueType: ValueType.INT,
     });
 
+    this.dbPhysicalSize = meter.createGauge(Metrics.WORLD_STATE_DB_PHYSICAL_SIZE, {
+      description: `The current physical disk space used for each database`,
+      valueType: ValueType.INT,
+    });
+
     this.treeSize = meter.createGauge(Metrics.WORLD_STATE_TREE_SIZE, {
       description: `The current number of leaves in each merkle tree`,
       valueType: ValueType.INT,
@@ -96,6 +102,9 @@ export class WorldStateInstrumentation {
     this.dbMapSize.record(Number(treeDbStats.mapSize), {
       [Attributes.MERKLE_TREE_NAME]: MerkleTreeId[tree],
     });
+    this.dbPhysicalSize.record(Number(treeDbStats.physicalFileSize), {
+      [Attributes.MERKLE_TREE_NAME]: MerkleTreeId[tree],
+    });
     this.treeSize.record(Number(treeMeta.size), {
       [Attributes.MERKLE_TREE_NAME]: MerkleTreeId[tree],
     });

yarn-project/world-state/src/native/message.ts

@@ -96,6 +96,8 @@ export interface DBStats {
 export interface TreeDBStats {
   /** The configured max size of the DB mapping file (effectively the max possible size of the DB) */
   mapSize: bigint;
+  /** The physical file size of the database on disk */
+  physicalFileSize: bigint;
   /** Stats for the 'blocks' DB */
   blocksDBStats: DBStats;
   /** Stats for the 'nodes' DB */
@@ -151,6 +153,7 @@ export function buildEmptyDBStats() {
 export function buildEmptyTreeDBStats() {
   return {
     mapSize: 0n,
+    physicalFileSize: 0n,
     blocksDBStats: buildEmptyDBStats(),
     nodesDBStats: buildEmptyDBStats(),
     leafIndicesDBStats: buildEmptyDBStats(),
@@ -242,6 +245,7 @@ export function sanitiseTreeDBStats(stats: TreeDBStats) {
   stats.blockIndicesDBStats = sanitiseDBStats(stats.blockIndicesDBStats);
   stats.nodesDBStats = sanitiseDBStats(stats.nodesDBStats);
   stats.mapSize = BigInt(stats.mapSize);
+  stats.physicalFileSize = BigInt(stats.physicalFileSize);
   return stats;
 }