Merge branch 'main' into zs/put-overwrite-policy

1 files changed, 193 insertions, 63 deletions

diff --git a/src/zenstore/blockstore.cpp b/src/zenstore/blockstore.cpp
index e976c061d..7b56c64bd 100644
--- a/src/zenstore/blockstore.cpp
+++ b/src/zenstore/blockstore.cpp
@@ -70,7 +70,7 @@ BlockStoreFile::Open()
 			return false;
 		}
 		ZEN_WARN("Failed to open cas block '{}', reason: '{}', retries left: {}.", m_Path, Ec.message(), RetriesLeft);
-		Sleep(100 - (3 - RetriesLeft) * 100);  // Total 600 ms
+		Sleep(100 + (3 - RetriesLeft) * 100);  // Total 600 ms
 		RetriesLeft--;
 		return true;
 	});
@@ -85,7 +85,7 @@ BlockStoreFile::Create(uint64_t InitialSize)
 	ZEN_TRACE_CPU("BlockStoreFile::Create");
 
 	auto ParentPath = m_Path.parent_path();
-	if (!std::filesystem::is_directory(ParentPath))
+	if (!IsDir(ParentPath))
 	{
 		CreateDirectories(ParentPath);
 	}
@@ -153,14 +153,28 @@ void
 BlockStoreFile::Write(const void* Data, uint64_t Size, uint64_t FileOffset)
 {
 	ZEN_TRACE_CPU("BlockStoreFile::Write");
+#if ZEN_BUILD_DEBUG
+	if (uint64_t CachedFileSize = m_CachedFileSize.load(); CachedFileSize > 0)
+	{
+		ZEN_ASSERT(FileOffset + Size <= CachedFileSize);
+	}
+#endif	// ZEN_BUILD_DEBUG
 	m_File.Write(Data, Size, FileOffset);
 }
 
 void
-BlockStoreFile::Flush()
+BlockStoreFile::Flush(uint64_t FinalSize)
 {
 	ZEN_TRACE_CPU("BlockStoreFile::Flush");
 	m_File.Flush();
+	if (FinalSize != (uint64_t)-1)
+	{
+		uint64_t ExpectedSize = 0;
+		if (!m_CachedFileSize.compare_exchange_weak(ExpectedSize, FinalSize))
+		{
+			ZEN_ASSERT(m_CachedFileSize.load() == FinalSize);
+		}
+	}
 }
 
 BasicFile&
@@ -215,7 +229,7 @@ IsMetaDataValid(const std::filesystem::path& BlockPath, const std::filesystem::p
 	}
 	if (MetaWriteTime < BlockWriteTime)
 	{
-		std::filesystem::remove(MetaPath, Ec);
+		RemoveFile(MetaPath, Ec);
 		return false;
 	}
 	return true;
@@ -239,7 +253,7 @@ BlockStoreFile::MetaSize() const
 	if (IsMetaDataValid(m_Path, MetaPath))
 	{
 		std::error_code DummyEc;
-		if (uint64_t Size = std::filesystem::file_size(MetaPath, DummyEc); !DummyEc)
+		if (uint64_t Size = FileSizeFromPath(MetaPath, DummyEc); !DummyEc)
 		{
 			return Size;
 		}
@@ -252,7 +266,7 @@ BlockStoreFile::RemoveMeta()
 {
 	std::filesystem::path MetaPath = GetMetaPath();
 	std::error_code		  DummyEc;
-	std::filesystem::remove(MetaPath, DummyEc);
+	RemoveFile(MetaPath, DummyEc);
 }
 
 std::filesystem::path
@@ -272,6 +286,14 @@ BlockStore::BlockStore()
 
 BlockStore::~BlockStore()
 {
+	try
+	{
+		Close();
+	}
+	catch (const std::exception& Ex)
+	{
+		ZEN_ERROR("~BlockStore() failed with: ", Ex.what());
+	}
 }
 
 void
@@ -291,8 +313,9 @@ BlockStore::Initialize(const std::filesystem::path& BlocksBasePath, uint64_t Max
 	m_MaxBlockSize	 = MaxBlockSize;
 	m_MaxBlockCount	 = MaxBlockCount;
 
-	if (std::filesystem::is_directory(m_BlocksBasePath))
+	if (IsDir(m_BlocksBasePath))
 	{
+		std::vector<std::filesystem::path> EmptyBlockFiles;
 		uint32_t						   NextBlockIndex = 0;
 		std::vector<std::filesystem::path> FoldersToScan;
 		FoldersToScan.push_back(m_BlocksBasePath);
@@ -320,6 +343,12 @@ BlockStore::Initialize(const std::filesystem::path& BlocksBasePath, uint64_t Max
 					{
 						continue;
 					}
+					if (Entry.file_size() == 0)
+					{
+						EmptyBlockFiles.push_back(Path);
+						continue;
+					}
+
 					Ref<BlockStoreFile> BlockFile{new BlockStoreFile(Path)};
 					BlockFile->Open();
 					m_TotalSize.fetch_add(BlockFile->TotalSize(), std::memory_order::relaxed);
@@ -333,6 +362,17 @@ BlockStore::Initialize(const std::filesystem::path& BlocksBasePath, uint64_t Max
 			}
 			++FolderOffset;
 		}
+
+		for (const std::filesystem::path& EmptyBlockFile : EmptyBlockFiles)
+		{
+			std::error_code Ec;
+			RemoveFile(EmptyBlockFile, Ec);
+			if (Ec)
+			{
+				ZEN_WARN("Unable to remove empty block file {}. Reason: {}", EmptyBlockFile, Ec.message());
+			}
+		}
+
 		m_WriteBlockIndex.store(NextBlockIndex, std::memory_order_release);
 	}
 	else
@@ -341,7 +381,7 @@ BlockStore::Initialize(const std::filesystem::path& BlocksBasePath, uint64_t Max
 	}
 }
 
-void
+BlockStore::BlockIndexSet
 BlockStore::SyncExistingBlocksOnDisk(const BlockIndexSet& KnownBlocks)
 {
 	ZEN_MEMSCOPE(GetBlocksTag());
@@ -349,8 +389,8 @@ BlockStore::SyncExistingBlocksOnDisk(const BlockIndexSet& KnownBlocks)
 
 	RwLock::ExclusiveLockScope InsertLock(m_InsertLock);
 
-	tsl::robin_set<uint32_t> MissingBlocks;
-	tsl::robin_set<uint32_t> DeleteBlocks;
+	BlockIndexSet MissingBlocks;
+	BlockIndexSet DeleteBlocks;
 	DeleteBlocks.reserve(m_ChunkBlocks.size());
 	for (auto It : m_ChunkBlocks)
 	{
@@ -369,13 +409,6 @@ BlockStore::SyncExistingBlocksOnDisk(const BlockIndexSet& KnownBlocks)
 			MissingBlocks.insert(BlockIndex);
 		}
 	}
-	for (std::uint32_t BlockIndex : MissingBlocks)
-	{
-		std::filesystem::path BlockPath = GetBlockPath(m_BlocksBasePath, BlockIndex);
-		Ref<BlockStoreFile>	  NewBlockFile(new BlockStoreFile(BlockPath));
-		NewBlockFile->Create(0);
-		m_ChunkBlocks[BlockIndex] = NewBlockFile;
-	}
 	for (std::uint32_t BlockIndex : DeleteBlocks)
 	{
 		std::filesystem::path BlockPath = GetBlockPath(m_BlocksBasePath, BlockIndex);
@@ -386,6 +419,7 @@ BlockStore::SyncExistingBlocksOnDisk(const BlockIndexSet& KnownBlocks)
 		}
 		m_ChunkBlocks.erase(BlockIndex);
 	}
+	return MissingBlocks;
 }
 
 BlockStore::BlockEntryCountMap
@@ -500,7 +534,7 @@ BlockStore::GetFreeBlockIndex(uint32_t ProbeIndex, RwLock::ExclusiveLockScope&,
 		{
 			OutBlockPath = GetBlockPath(m_BlocksBasePath, ProbeIndex);
 			std::error_code Ec;
-			bool			Exists = std::filesystem::exists(OutBlockPath, Ec);
+			bool			Exists = IsFile(OutBlockPath, Ec);
 			if (Ec)
 			{
 				ZEN_WARN("Failed to probe existence of file '{}' when trying to allocate a new block. Reason: '{}'",
@@ -540,7 +574,7 @@ BlockStore::WriteChunk(const void* Data, uint64_t Size, uint32_t Alignment, cons
 	{
 		if (m_WriteBlock)
 		{
-			m_WriteBlock->Flush();
+			m_WriteBlock->Flush(m_CurrentInsertOffset);
 			m_WriteBlock = nullptr;
 		}
 
@@ -578,7 +612,7 @@ BlockStore::WriteChunk(const void* Data, uint64_t Size, uint32_t Alignment, cons
 }
 
 void
-BlockStore::WriteChunks(std::span<IoBuffer> Datas, uint32_t Alignment, const WriteChunksCallback& Callback)
+BlockStore::WriteChunks(std::span<const IoBuffer> Datas, uint32_t Alignment, const WriteChunksCallback& Callback)
 {
 	ZEN_MEMSCOPE(GetBlocksTag());
 	ZEN_TRACE_CPU("BlockStore::WriteChunks");
@@ -674,6 +708,27 @@ BlockStore::WriteChunks(std::span<IoBuffer> Datas, uint32_t Alignment, const Wri
 	}
 }
 
+bool
+BlockStore::HasChunk(const BlockStoreLocation& Location) const
+{
+	ZEN_TRACE_CPU("BlockStore::TryGetChunk");
+	RwLock::SharedLockScope InsertLock(m_InsertLock);
+	if (auto BlockIt = m_ChunkBlocks.find(Location.BlockIndex); BlockIt != m_ChunkBlocks.end())
+	{
+		if (const Ref<BlockStoreFile>& Block = BlockIt->second; Block)
+		{
+			InsertLock.ReleaseNow();
+
+			const uint64_t BlockSize = Block->FileSize();
+			if (Location.Offset + Location.Size <= BlockSize)
+			{
+				return true;
+			}
+		}
+	}
+	return false;
+}
+
 IoBuffer
 BlockStore::TryGetChunk(const BlockStoreLocation& Location) const
 {
@@ -706,7 +761,7 @@ BlockStore::Flush(bool ForceNewBlock)
 		{
 			if (m_WriteBlock)
 			{
-				m_WriteBlock->Flush();
+				m_WriteBlock->Flush(m_CurrentInsertOffset);
 			}
 			m_WriteBlock		  = nullptr;
 			m_CurrentInsertOffset = 0;
@@ -735,6 +790,8 @@ BlockStore::IterateBlock(std::span<const BlockStoreLocation>   ChunkLocations,
 		return true;
 	}
 
+	ZEN_ASSERT(ChunkLocations.size() >= InChunkIndexes.size());
+
 	if (LargeSizeLimit == 0)
 	{
 		LargeSizeLimit = DefaultIterateSmallChunkWindowSize;
@@ -746,7 +803,10 @@ BlockStore::IterateBlock(std::span<const BlockStoreLocation>   ChunkLocations,
 
 	IterateSmallChunkWindowSize = Min((LargeSizeLimit + IterateSmallChunkMaxGapSize) * ChunkLocations.size(), IterateSmallChunkWindowSize);
 
-	uint32_t			BlockIndex = ChunkLocations[InChunkIndexes[0]].BlockIndex;
+	const size_t FirstLocationIndex = InChunkIndexes[0];
+	ZEN_ASSERT(FirstLocationIndex < ChunkLocations.size());
+
+	const uint32_t		BlockIndex = ChunkLocations[FirstLocationIndex].BlockIndex;
 	std::vector<size_t> ChunkIndexes(InChunkIndexes.begin(), InChunkIndexes.end());
 	std::sort(ChunkIndexes.begin(), ChunkIndexes.end(), [&](size_t IndexA, size_t IndexB) -> bool {
 		return ChunkLocations[IndexA].Offset < ChunkLocations[IndexB].Offset;
@@ -756,8 +816,9 @@ BlockStore::IterateBlock(std::span<const BlockStoreLocation>   ChunkLocations,
 						 IterateSmallChunkWindowSize,
 						 IterateSmallChunkMaxGapSize,
 						 &ChunkLocations](uint64_t BlockFileSize, std::span<const size_t> ChunkIndexes, size_t StartIndexOffset) -> size_t {
-		size_t					  ChunkCount	= 0;
-		size_t					  StartIndex	= ChunkIndexes[StartIndexOffset];
+		size_t ChunkCount = 0;
+		size_t StartIndex = ChunkIndexes[StartIndexOffset];
+		ZEN_ASSERT(StartIndex < ChunkLocations.size());
 		const BlockStoreLocation& StartLocation = ChunkLocations[StartIndex];
 		uint64_t				  StartOffset	= StartLocation.Offset;
 		uint64_t				  LastEnd		= StartOffset + StartLocation.Size;
@@ -810,22 +871,26 @@ BlockStore::IterateBlock(std::span<const BlockStoreLocation>   ChunkLocations,
 		ZEN_ASSERT(BlockFile);
 		InsertLock.ReleaseNow();
 
+		const size_t BlockSize = BlockFile->FileSize();
+
 		IoBuffer ReadBuffer;
 		void*	 BufferBase = nullptr;
 
 		size_t LocationIndexOffset = 0;
 		while (LocationIndexOffset < ChunkIndexes.size())
 		{
-			size_t					  ChunkIndex	= ChunkIndexes[LocationIndexOffset];
+			size_t ChunkIndex = ChunkIndexes[LocationIndexOffset];
+			ZEN_ASSERT(ChunkIndex < ChunkLocations.size());
 			const BlockStoreLocation& FirstLocation = ChunkLocations[ChunkIndex];
+			ZEN_ASSERT(FirstLocation.BlockIndex == BlockIndex);
 
-			const size_t BlockSize	= BlockFile->FileSize();
 			const size_t RangeCount = GetNextRange(BlockSize, ChunkIndexes, LocationIndexOffset);
 			if (RangeCount > 1)
 			{
-				size_t					  LastChunkIndex = ChunkIndexes[LocationIndexOffset + RangeCount - 1];
-				const BlockStoreLocation& LastLocation	 = ChunkLocations[LastChunkIndex];
-				uint64_t				  Size			 = LastLocation.Offset + LastLocation.Size - FirstLocation.Offset;
+				size_t LastChunkIndex = ChunkIndexes[LocationIndexOffset + RangeCount - 1];
+				ZEN_ASSERT(LastChunkIndex < ChunkLocations.size());
+				const BlockStoreLocation& LastLocation = ChunkLocations[LastChunkIndex];
+				uint64_t				  Size		   = LastLocation.Offset + LastLocation.Size - FirstLocation.Offset;
 				if (ReadBuffer.GetSize() < Size)
 				{
 					ReadBuffer = IoBuffer(Min(Size * 2, IterateSmallChunkWindowSize));
@@ -834,8 +899,9 @@ BlockStore::IterateBlock(std::span<const BlockStoreLocation>   ChunkLocations,
 				BlockFile->Read(BufferBase, Size, FirstLocation.Offset);
 				for (size_t RangeIndex = 0; RangeIndex < RangeCount; ++RangeIndex)
 				{
-					size_t					  NextChunkIndex = ChunkIndexes[LocationIndexOffset + RangeIndex];
-					const BlockStoreLocation& ChunkLocation	 = ChunkLocations[NextChunkIndex];
+					size_t NextChunkIndex = ChunkIndexes[LocationIndexOffset + RangeIndex];
+					ZEN_ASSERT(NextChunkIndex < ChunkLocations.size());
+					const BlockStoreLocation& ChunkLocation = ChunkLocations[NextChunkIndex];
 					if (ChunkLocation.Size == 0 || ((ChunkLocation.Offset + ChunkLocation.Size) > BlockSize))
 					{
 						ZEN_LOG_SCOPE("chunk [{},{}] out of bounds (block #{} file size = {})",
@@ -958,6 +1024,7 @@ BlockStore::CompactBlocks(const BlockStoreCompactState&	  CompactState,
 
 	uint32_t		 NewBlockIndex = 0;
 	MovedChunksArray MovedChunks;
+	ChunkIndexArray	 ScrubbedChunks;
 
 	uint64_t AddedSize	 = 0;
 	uint64_t RemovedSize = 0;
@@ -986,14 +1053,16 @@ BlockStore::CompactBlocks(const BlockStoreCompactState&	  CompactState,
 
 	auto ReportChanges = [&]() -> bool {
 		bool Continue = true;
-		if (!MovedChunks.empty() || RemovedSize > 0)
+		if (!MovedChunks.empty() || !ScrubbedChunks.empty() || RemovedSize > 0)
 		{
-			Continue = ChangeCallback(MovedChunks, RemovedSize > AddedSize ? RemovedSize - AddedSize : 0);
+			Continue = ChangeCallback(MovedChunks, ScrubbedChunks, RemovedSize > AddedSize ? RemovedSize - AddedSize : 0);
 			DeletedSize += RemovedSize;
+			m_TotalSize.fetch_add(AddedSize);
 			RemovedSize = 0;
 			AddedSize	= 0;
 			MovedCount += MovedChunks.size();
 			MovedChunks.clear();
+			ScrubbedChunks.clear();
 		}
 		return Continue;
 	};
@@ -1022,6 +1091,7 @@ BlockStore::CompactBlocks(const BlockStoreCompactState&	  CompactState,
 						 LogPrefix,
 						 m_BlocksBasePath,
 						 BlockIndex);
+				ScrubbedChunks.insert(ScrubbedChunks.end(), KeepChunkIndexes.begin(), KeepChunkIndexes.end());
 				return true;
 			}
 			if (!It->second)
@@ -1030,6 +1100,7 @@ BlockStore::CompactBlocks(const BlockStoreCompactState&	  CompactState,
 						 LogPrefix,
 						 m_BlocksBasePath,
 						 BlockIndex);
+				ScrubbedChunks.insert(ScrubbedChunks.end(), KeepChunkIndexes.begin(), KeepChunkIndexes.end());
 				return true;
 			}
 			OldBlockFile = It->second;
@@ -1051,11 +1122,10 @@ BlockStore::CompactBlocks(const BlockStoreCompactState&	  CompactState,
 			std::sort(SortedChunkIndexes.begin(), SortedChunkIndexes.end(), [&ChunkLocations](size_t Lhs, size_t Rhs) {
 				return ChunkLocations[Lhs].Offset < ChunkLocations[Rhs].Offset;
 			});
-			BasicFileBuffer SourceFileBuffer(OldBlockFile->GetBasicFile(), Min(65536u, OldBlockSize));
+			BasicFileBuffer SourceFileBuffer(OldBlockFile->GetBasicFile(), Min(256u * 1024u, OldBlockSize));
 
-			uint64_t			 WrittenBytesToBlock = 0;
-			uint64_t			 MovedFromBlock		 = 0;
-			std::vector<uint8_t> Chunk;
+			uint64_t			 MovedFromBlock = 0;
+			std::vector<uint8_t> ChunkBuffer;
 			for (const size_t& ChunkIndex : SortedChunkIndexes)
 			{
 				const BlockStoreLocation ChunkLocation = ChunkLocations[ChunkIndex];
@@ -1070,19 +1140,29 @@ BlockStore::CompactBlocks(const BlockStoreCompactState&	  CompactState,
 						ChunkLocation.Size,
 						OldBlockFile->GetPath(),
 						OldBlockSize);
+					ScrubbedChunks.push_back(ChunkIndex);
 					continue;
 				}
 
-				Chunk.resize(ChunkLocation.Size);
-				SourceFileBuffer.Read(Chunk.data(), Chunk.size(), ChunkLocation.Offset);
+				MemoryView ChunkView = SourceFileBuffer.MakeView(ChunkLocation.Size, ChunkLocation.Offset);
+				if (ChunkView.GetSize() != ChunkLocation.Size)
+				{
+					ChunkBuffer.resize(ChunkLocation.Size);
+					SourceFileBuffer.Read(ChunkBuffer.data(), ChunkLocation.Size, ChunkLocation.Offset);
+					ChunkView = MemoryView(ChunkBuffer.data(), ChunkLocation.Size);
+				}
 
-				if ((WriteOffset + Chunk.size()) > m_MaxBlockSize)
+				if ((WriteOffset + ChunkView.GetSize()) > m_MaxBlockSize)
 				{
-					TargetFileBuffer.reset();
+					if (TargetFileBuffer)
+					{
+						TargetFileBuffer->Flush();
+						TargetFileBuffer.reset();
+					}
 					if (NewBlockFile)
 					{
 						ZEN_ASSERT_SLOW(NewBlockFile->IsOpen());
-						NewBlockFile->Flush();
+						NewBlockFile->Flush(WriteOffset);
 						uint64_t NewBlockSize = NewBlockFile->FileSize();
 						MovedSize += NewBlockSize;
 						NewBlockFile = nullptr;
@@ -1161,22 +1241,23 @@ BlockStore::CompactBlocks(const BlockStoreCompactState&	  CompactState,
 								 NiceBytes(Space.Free + ReclaimedSpace));
 					}
 					NewBlockFile->Create(m_MaxBlockSize);
-					NewBlockIndex = NextBlockIndex;
-					WriteOffset	  = 0;
-					AddedSize += WrittenBytesToBlock;
-					WrittenBytesToBlock = 0;
-					TargetFileBuffer	= std::make_unique<BasicFileWriter>(NewBlockFile->GetBasicFile(), Min(65536u, m_MaxBlockSize));
+					NewBlockIndex	 = NextBlockIndex;
+					WriteOffset		 = 0;
+					TargetFileBuffer = std::make_unique<BasicFileWriter>(NewBlockFile->GetBasicFile(), Min(256u * 1024u, m_MaxBlockSize));
 				}
 
-				TargetFileBuffer->Write(Chunk.data(), ChunkLocation.Size, WriteOffset);
+				const uint64_t OldWriteOffset = WriteOffset;
+				WriteOffset					  = TargetFileBuffer->AlignTo(PayloadAlignment);
+
+				TargetFileBuffer->Write(ChunkView.GetData(), ChunkLocation.Size, WriteOffset);
 				MovedChunks.push_back(
 					{ChunkIndex, {.BlockIndex = NewBlockIndex, .Offset = gsl::narrow<uint32_t>(WriteOffset), .Size = ChunkLocation.Size}});
-				WrittenBytesToBlock = WriteOffset + ChunkLocation.Size;
+				WriteOffset += ChunkLocation.Size;
 
 				MovedFromBlock += RoundUp(ChunkLocation.Offset + ChunkLocation.Size, PayloadAlignment) - ChunkLocation.Offset;
-				WriteOffset = RoundUp(WriteOffset + ChunkLocation.Size, PayloadAlignment);
+				uint64_t WrittenBytes = WriteOffset - OldWriteOffset;
+				AddedSize += WrittenBytes;
 			}
-			AddedSize += WrittenBytesToBlock;
 			ZEN_INFO("{}moved {} chunks ({}) from '{}' to new block, freeing {}",
 					 LogPrefix,
 					 KeepChunkIndexes.size(),
@@ -1209,10 +1290,16 @@ BlockStore::CompactBlocks(const BlockStoreCompactState&	  CompactState,
 		return true;
 	});
 
+	if (TargetFileBuffer)
+	{
+		TargetFileBuffer->Flush();
+		TargetFileBuffer.reset();
+	}
+
 	if (NewBlockFile)
 	{
 		ZEN_ASSERT_SLOW(NewBlockFile->IsOpen());
-		NewBlockFile->Flush();
+		NewBlockFile->Flush(WriteOffset);
 		uint64_t NewBlockSize = NewBlockFile->FileSize();
 		MovedSize += NewBlockSize;
 		NewBlockFile = nullptr;
@@ -1343,6 +1430,8 @@ TEST_CASE("blockstore.blockfile")
 		CHECK(std::string(Boop) == "boop");
 		File1.Flush();
 		CHECK(File1.FileSize() == 10);
+		File1.Flush(10);
+		CHECK(File1.FileSize() == 10);
 	}
 	{
 		BlockStoreFile File1(RootDirectory / "1");
@@ -1375,14 +1464,14 @@ TEST_CASE("blockstore.blockfile")
 			BoopChunk = File1.GetChunk(5, 5);
 		}
 
-		CHECK(std::filesystem::exists(RootDirectory / "1"));
+		CHECK(IsFile(RootDirectory / "1"));
 
 		const char* Data = static_cast<const char*>(DataChunk.GetData());
 		CHECK(std::string(Data) == "data");
 		const char* Boop = static_cast<const char*>(BoopChunk.GetData());
 		CHECK(std::string(Boop) == "boop");
 	}
-	CHECK(std::filesystem::exists(RootDirectory / "1"));
+	CHECK(IsFile(RootDirectory / "1"));
 
 	{
 		IoBuffer DataChunk;
@@ -1401,7 +1490,7 @@ TEST_CASE("blockstore.blockfile")
 		const char* Boop = static_cast<const char*>(BoopChunk.GetData());
 		CHECK(std::string(Boop) == "boop");
 	}
-	CHECK(!std::filesystem::exists(RootDirectory / "1"));
+	CHECK(!IsFile(RootDirectory / "1"));
 }
 
 namespace blockstore::impl {
@@ -1800,7 +1889,7 @@ TEST_CASE("blockstore.compact.blocks")
 		Store.CompactBlocks(
 			State,
 			Alignment,
-			[&](const BlockStore::MovedChunksArray&, uint64_t) {
+			[&](const BlockStore::MovedChunksArray&, const BlockStore::ChunkIndexArray&, uint64_t) {
 				CHECK(false);
 				return true;
 			},
@@ -1825,9 +1914,10 @@ TEST_CASE("blockstore.compact.blocks")
 		Store.CompactBlocks(
 			State,
 			Alignment,
-			[&](const BlockStore::MovedChunksArray& Moved, uint64_t Removed) {
+			[&](const BlockStore::MovedChunksArray& Moved, const BlockStore::ChunkIndexArray& Scrubbed, uint64_t Removed) {
 				RemovedSize += Removed;
 				CHECK(Moved.empty());
+				CHECK(Scrubbed.empty());
 				return true;
 			},
 			[]() { return 0; });
@@ -1850,9 +1940,10 @@ TEST_CASE("blockstore.compact.blocks")
 		Store.CompactBlocks(
 			State,
 			Alignment,
-			[&](const BlockStore::MovedChunksArray& Moved, uint64_t Removed) {
+			[&](const BlockStore::MovedChunksArray& Moved, const BlockStore::ChunkIndexArray& Scrubbed, uint64_t Removed) {
 				RemovedSize += Removed;
 				CHECK(Moved.empty());
+				CHECK(Scrubbed.empty());
 				return true;
 			},
 			[]() { return 0; });
@@ -1860,7 +1951,7 @@ TEST_CASE("blockstore.compact.blocks")
 		CHECK_LE(Store.TotalSize(), 1088);
 		CHECK_GT(Store.TotalSize(), 0);
 	}
-	SUBCASE("keep everthing")
+	SUBCASE("keep everything")
 	{
 		Store.Flush(true);
 
@@ -1873,7 +1964,7 @@ TEST_CASE("blockstore.compact.blocks")
 		Store.CompactBlocks(
 			State,
 			Alignment,
-			[&](const BlockStore::MovedChunksArray&, uint64_t) {
+			[&](const BlockStore::MovedChunksArray&, const BlockStore::ChunkIndexArray&, uint64_t) {
 				CHECK(false);
 				return true;
 			},
@@ -1904,8 +1995,9 @@ TEST_CASE("blockstore.compact.blocks")
 		Store.CompactBlocks(
 			State,
 			Alignment,
-			[&](const BlockStore::MovedChunksArray& Moved, uint64_t Removed) {
+			[&](const BlockStore::MovedChunksArray& Moved, const BlockStore::ChunkIndexArray& Scrubbed, uint64_t Removed) {
 				CHECK(Moved.empty());
+				CHECK(Scrubbed.empty());
 				RemovedSize += Removed;
 				return true;
 			},
@@ -1939,7 +2031,8 @@ TEST_CASE("blockstore.compact.blocks")
 		Store.CompactBlocks(
 			State,
 			Alignment,
-			[&](const BlockStore::MovedChunksArray& Moved, uint64_t Removed) {
+			[&](const BlockStore::MovedChunksArray& Moved, const BlockStore::ChunkIndexArray& Scrubbed, uint64_t Removed) {
+				CHECK(Scrubbed.empty());
 				for (const auto& Move : Moved)
 				{
 					const BlockStoreLocation& OldLocation = State.GetLocation(Move.first);
@@ -2016,7 +2109,8 @@ TEST_CASE("blockstore.compact.blocks")
 		Store.CompactBlocks(
 			State,
 			Alignment,
-			[&](const BlockStore::MovedChunksArray& Moved, uint64_t Removed) {
+			[&](const BlockStore::MovedChunksArray& Moved, const BlockStore::ChunkIndexArray& Scrubbed, uint64_t Removed) {
+				CHECK(Scrubbed.empty());
 				for (const auto& Move : Moved)
 				{
 					const BlockStoreLocation& OldLocation = State.GetLocation(Move.first);
@@ -2051,6 +2145,42 @@ TEST_CASE("blockstore.compact.blocks")
 		}
 		CHECK_LT(Store.TotalSize(), PreSize);
 	}
+	SUBCASE("scrub")
+	{
+		Store.Flush(true);
+
+		BlockStoreCompactState State;
+		for (const BlockStoreLocation& Location : ChunkLocations)
+		{
+			State.IncludeBlock(Location.BlockIndex);
+			CHECK(State.AddKeepLocation(Location));
+		}
+		State.IncludeBlock(0);
+		State.IncludeBlock(999);
+		std::vector<size_t> ExpectedScrubbedIndexes;
+		ExpectedScrubbedIndexes.push_back(ChunkLocations.size() + 0);
+		State.AddKeepLocation(BlockStoreLocation{.BlockIndex = 0, .Offset = 2000, .Size = 322});
+		ExpectedScrubbedIndexes.push_back(ChunkLocations.size() + 1);
+		State.AddKeepLocation(BlockStoreLocation{.BlockIndex = 0, .Offset = 10, .Size = 3220});
+		ExpectedScrubbedIndexes.push_back(ChunkLocations.size() + 2);
+		State.AddKeepLocation(BlockStoreLocation{.BlockIndex = 999, .Offset = 2, .Size = 40});
+
+		std::vector<size_t> ScrubbedIndexes;
+
+		Store.CompactBlocks(
+			State,
+			Alignment,
+			[&](const BlockStore::MovedChunksArray&, const BlockStore::ChunkIndexArray& ScrubbedArray, uint64_t) {
+				ScrubbedIndexes.insert(ScrubbedIndexes.end(), ScrubbedArray.begin(), ScrubbedArray.end());
+				return true;
+			},
+			[]() {
+				CHECK(false);
+				return 0;
+			});
+		std::sort(ScrubbedIndexes.begin(), ScrubbedIndexes.end());
+		CHECK_EQ(ExpectedScrubbedIndexes, ScrubbedIndexes);
+	}
 }
 
 #endif