Move bulk of MigrateLegacyData to blockstore.cpp

1 files changed, 215 insertions, 4 deletions

diff --git a/zenstore/blockstore.cpp b/zenstore/blockstore.cpp
index 9961e734d..6f5578be8 100644
--- a/zenstore/blockstore.cpp
+++ b/zenstore/blockstore.cpp
@@ -437,7 +437,7 @@ BlockStore::ReclaimSpace(const ReclaimSnapshotState&			Snapshot,
 			{
 				DeletedSize += ChunkLocations[DeleteIndex].Size;
 			}
-			Callback(BlockIndex, {}, DeleteMap);
+			Callback({}, DeleteMap);
 			DeletedCount += DeleteMap.size();
 			{
 				RwLock::ExclusiveLockScope _i(m_InsertLock);
@@ -477,7 +477,7 @@ BlockStore::ReclaimSpace(const ReclaimSnapshotState&			Snapshot,
 					NewBlockFile->Flush();
 				}
 				{
-					Callback(0xfffffffful, MovedChunks, {});
+					Callback(MovedChunks, {});
 					MovedCount += KeepMap.size();
 					MovedChunks.clear();
 					RwLock::ExclusiveLockScope __(m_InsertLock);
@@ -558,7 +558,7 @@ BlockStore::ReclaimSpace(const ReclaimSnapshotState&			Snapshot,
 			DeletedSize += ChunkLocations[DeleteIndex].Size;
 		}
 
-		Callback(BlockIndex, MovedChunks, DeleteMap);
+		Callback(MovedChunks, DeleteMap);
 		MovedCount += KeepMap.size();
 		DeletedCount += DeleteMap.size();
 		MovedChunks.clear();
@@ -599,7 +599,7 @@ BlockStore::IterateChunks(const std::vector<BlockStoreLocation>& ChunkLocations,
 		IoBuffer			ReadBuffer{WindowSize};
 		void*				BufferBase = ReadBuffer.MutableData();
 
-		RwLock::SharedLockScope _(m_InsertLock);  // TODO: Refactor so we don't have to keep m_InsertLock all the time?
+		RwLock::SharedLockScope _(m_InsertLock);
 
 		for (const auto& Block : m_ChunkBlocks)
 		{
@@ -659,6 +659,217 @@ BlockStore::IterateChunks(const std::vector<BlockStoreLocation>& ChunkLocations,
 	}
 }
 
+bool
+BlockStore::Split(const std::vector<BlockStoreLocation>& ChunkLocations,
+				  const std::filesystem::path&			 SourceBlockFilePath,
+				  const std::filesystem::path&			 BlocksBasePath,
+				  uint64_t								 MaxBlockSize,
+				  uint64_t								 MaxBlockCount,
+				  size_t								 PayloadAlignment,
+				  bool									 CleanSource,
+				  const SplitCallback&					 Callback)
+{
+	std::error_code Error;
+	DiskSpace		Space = DiskSpaceInfo(BlocksBasePath.parent_path(), Error);
+	if (Error)
+	{
+		ZEN_ERROR("get disk space in {} FAILED, reason: '{}'", BlocksBasePath, Error.message());
+		return false;
+	}
+
+	if (Space.Free < MaxBlockSize)
+	{
+		ZEN_ERROR("legacy store migration from '{}' FAILED, required disk space {}, free {}",
+				  BlocksBasePath,
+				  MaxBlockSize,
+				  NiceBytes(Space.Free));
+		return false;
+	}
+
+	size_t TotalSize = 0;
+	for (const BlockStoreLocation& Location : ChunkLocations)
+	{
+		TotalSize += Location.Size;
+	}
+	size_t	 ChunkCount			   = ChunkLocations.size();
+	uint64_t RequiredDiskSpace	   = TotalSize + ((PayloadAlignment - 1) * ChunkCount);
+	uint64_t MaxRequiredBlockCount = RoundUp(RequiredDiskSpace, MaxBlockSize) / MaxBlockSize;
+	if (MaxRequiredBlockCount > MaxBlockCount)
+	{
+		ZEN_ERROR("legacy store migration from '{}' FAILED, required block count {}, possible {}",
+				  BlocksBasePath,
+				  MaxRequiredBlockCount,
+				  MaxBlockCount);
+		return false;
+	}
+
+	constexpr const uint64_t DiskReserve = 1ul << 28;
+
+	if (CleanSource)
+	{
+		if (Space.Free < (MaxBlockSize + DiskReserve))
+		{
+			ZEN_INFO("legacy store migration from '{}' aborted, not enough disk space available {} ({})",
+					 BlocksBasePath,
+					 NiceBytes(MaxBlockSize + DiskReserve),
+					 NiceBytes(Space.Free));
+			return false;
+		}
+	}
+	else
+	{
+		if (Space.Free < (RequiredDiskSpace + DiskReserve))
+		{
+			ZEN_INFO("legacy store migration from '{}' aborted, not enough disk space available {} ({})",
+					 BlocksBasePath,
+					 NiceBytes(RequiredDiskSpace + DiskReserve),
+					 NiceBytes(Space.Free));
+			return false;
+		}
+	}
+
+	uint32_t WriteBlockIndex = 0;
+	while (std::filesystem::exists(BlockStore::GetBlockPath(BlocksBasePath, WriteBlockIndex)))
+	{
+		++WriteBlockIndex;
+	}
+
+	BasicFile BlockFile;
+	BlockFile.Open(SourceBlockFilePath, CleanSource ? BasicFile::Mode::kWrite : BasicFile::Mode::kRead);
+
+	if (CleanSource && (MaxRequiredBlockCount < 2))
+	{
+		std::vector<std::pair<size_t, BlockStoreLocation>> Chunks;
+		Chunks.reserve(ChunkCount);
+		for (size_t Index = 0; Index < ChunkCount; ++Index)
+		{
+			const BlockStoreLocation& ChunkLocation = ChunkLocations[Index];
+			Chunks.push_back({Index, {.BlockIndex = WriteBlockIndex, .Offset = ChunkLocation.Offset, .Size = ChunkLocation.Size}});
+		}
+		std::filesystem::path BlockPath = BlockStore::GetBlockPath(BlocksBasePath, WriteBlockIndex);
+		CreateDirectories(BlockPath.parent_path());
+		BlockFile.Close();
+		std::filesystem::rename(SourceBlockFilePath, BlockPath);
+		Callback(Chunks);
+		return true;
+	}
+
+	std::vector<size_t> ChunkIndexes;
+	ChunkIndexes.reserve(ChunkCount);
+	for (size_t Index = 0; Index < ChunkCount; ++Index)
+	{
+		ChunkIndexes.push_back(Index);
+	}
+
+	std::sort(begin(ChunkIndexes), end(ChunkIndexes), [&ChunkLocations](size_t Lhs, size_t Rhs) {
+		const BlockStoreLocation& LhsLocation = ChunkLocations[Lhs];
+		const BlockStoreLocation& RhsLocation = ChunkLocations[Rhs];
+		return LhsLocation.Offset < RhsLocation.Offset;
+	});
+
+	uint64_t						BlockSize	= 0;
+	uint64_t						BlockOffset = 0;
+	std::vector<BlockStoreLocation> NewLocations;
+	struct BlockData
+	{
+		std::vector<std::pair<size_t, BlockStoreLocation>> Chunks;
+		uint64_t										   BlockOffset;
+		uint64_t										   BlockSize;
+		uint32_t										   BlockIndex;
+	};
+
+	std::vector<BlockData>							   BlockRanges;
+	std::vector<std::pair<size_t, BlockStoreLocation>> Chunks;
+	BlockRanges.reserve(MaxRequiredBlockCount);
+	for (const size_t& ChunkIndex : ChunkIndexes)
+	{
+		const BlockStoreLocation& LegacyChunkLocation = ChunkLocations[ChunkIndex];
+
+		uint64_t ChunkOffset = LegacyChunkLocation.Offset;
+		uint64_t ChunkSize	 = LegacyChunkLocation.Size;
+		uint64_t ChunkEnd	 = ChunkOffset + ChunkSize;
+
+		if (BlockSize == 0)
+		{
+			BlockOffset = ChunkOffset;
+		}
+		if ((ChunkEnd - BlockOffset) > MaxBlockSize)
+		{
+			BlockData BlockRange{.BlockOffset = BlockOffset, .BlockSize = BlockSize, .BlockIndex = WriteBlockIndex};
+			BlockRange.Chunks.swap(Chunks);
+			BlockRanges.push_back(BlockRange);
+
+			WriteBlockIndex++;
+			while (std::filesystem::exists(BlockStore::GetBlockPath(BlocksBasePath, WriteBlockIndex)))
+			{
+				++WriteBlockIndex;
+			}
+			BlockOffset = ChunkOffset;
+			BlockSize	= 0;
+		}
+		BlockSize						 = RoundUp(BlockSize, PayloadAlignment);
+		BlockStoreLocation ChunkLocation = {.BlockIndex = WriteBlockIndex, .Offset = ChunkOffset - BlockOffset, .Size = ChunkSize};
+		Chunks.push_back({ChunkIndex, ChunkLocation});
+		BlockSize = ChunkEnd - BlockOffset;
+	}
+	if (BlockSize > 0)
+	{
+		BlockRanges.push_back(
+			{.Chunks = std::move(Chunks), .BlockOffset = BlockOffset, .BlockSize = BlockSize, .BlockIndex = WriteBlockIndex});
+	}
+
+	Stopwatch WriteBlockTimer;
+
+	std::reverse(BlockRanges.begin(), BlockRanges.end());
+	std::vector<std::uint8_t> Buffer(1 << 28);
+	for (size_t Idx = 0; Idx < BlockRanges.size(); ++Idx)
+	{
+		const BlockData& BlockRange = BlockRanges[Idx];
+		if (Idx > 0)
+		{
+			uint64_t Remaining = BlockRange.BlockOffset + BlockRange.BlockSize;
+			uint64_t Completed = BlockOffset + BlockSize - Remaining;
+			uint64_t ETA	   = (WriteBlockTimer.GetElapsedTimeMs() * Remaining) / Completed;
+
+			ZEN_INFO("migrating store '{}' {}/{} blocks, remaining {} ({}) ETA: {}",
+					 BlocksBasePath,
+					 Idx,
+					 BlockRanges.size(),
+					 NiceBytes(BlockRange.BlockOffset + BlockRange.BlockSize),
+					 NiceBytes(BlockOffset + BlockSize),
+					 NiceTimeSpanMs(ETA));
+		}
+
+		std::filesystem::path BlockPath = BlockStore::GetBlockPath(BlocksBasePath, BlockRange.BlockIndex);
+		BlockStoreFile		  ChunkBlock(BlockPath);
+		ChunkBlock.Create(BlockRange.BlockSize);
+		uint64_t Offset = 0;
+		while (Offset < BlockRange.BlockSize)
+		{
+			uint64_t Size = BlockRange.BlockSize - Offset;
+			if (Size > Buffer.size())
+			{
+				Size = Buffer.size();
+			}
+			BlockFile.Read(Buffer.data(), Size, BlockRange.BlockOffset + Offset);
+			ChunkBlock.Write(Buffer.data(), Size, Offset);
+			Offset += Size;
+		}
+		ChunkBlock.Truncate(Offset);
+		ChunkBlock.Flush();
+
+		Callback(BlockRange.Chunks);
+
+		if (CleanSource)
+		{
+			BlockFile.SetFileSize(BlockRange.BlockOffset);
+		}
+	}
+	BlockFile.Close();
+
+	return true;
+}
+
 const char*
 BlockStore::GetBlockFileExtension()
 {