Rewrite cas log on gc

1 files changed, 55 insertions, 36 deletions

diff --git a/zenstore/compactcas.cpp b/zenstore/compactcas.cpp
index 133784cb6..3d7eb695b 100644
--- a/zenstore/compactcas.cpp
+++ b/zenstore/compactcas.cpp
@@ -78,10 +78,10 @@ CasContainerStrategy::InsertChunk(const void* ChunkData, size_t ChunkSize, const
 	m_CurrentInsertOffset = (m_CurrentInsertOffset + ChunkSize + m_PayloadAlignment - 1) & ~(m_PayloadAlignment - 1);
 
 	const CasDiskLocation Location{InsertOffset, ChunkSize};
-	CasDiskIndexEntry IndexEntry{.Key = ChunkHash, .Location = Location};
+	CasDiskIndexEntry	  IndexEntry{.Key = ChunkHash, .Location = Location};
 
-//	m_SmallObjectIndex.Write(&IndexEntry, sizeof(CasDiskIndexEntry), m_CurrentIndexOffset);
-//	m_CurrentIndexOffset += sizeof(CasDiskIndexEntry);
+	//	m_SmallObjectIndex.Write(&IndexEntry, sizeof(CasDiskIndexEntry), m_CurrentIndexOffset);
+	//	m_CurrentIndexOffset += sizeof(CasDiskIndexEntry);
 
 	RwLock::ExclusiveLockScope __(m_LocationMapLock);
 	m_LocationMap[ChunkHash] = Location;
@@ -144,7 +144,7 @@ void
 CasContainerStrategy::Flush()
 {
 	m_CasLog.Flush();
-//	m_SmallObjectIndex.Flush();
+	//	m_SmallObjectIndex.Flush();
 	m_SmallObjectFile.Flush();
 }
 
@@ -312,7 +312,7 @@ CasContainerStrategy::CollectGarbage(GcContext& GcCtx)
 	});
 
 	uint64_t					 NewTotalSize = 0;
-	std::vector<CasDiskLocation> ChunkLocations;  // Sorted by position
+	std::vector<CasDiskLocation> ChunkLocations;
 	ChunkLocations.reserve(ChunkHashes.size());
 	for (auto Entry : ChunkHashes)
 	{
@@ -347,6 +347,8 @@ CasContainerStrategy::CollectGarbage(GcContext& GcCtx)
 		m_TotalSize.fetch_sub(static_cast<uint64_t>(ChunkLocation.GetSize()));
 	}
 
+	// We can break here if we only want to remove items without compacting of space
+
 	std::vector<IoHash> MovedChunks;
 
 	uint64_t WriteOffset{};
@@ -430,6 +432,8 @@ CasContainerStrategy::CollectGarbage(GcContext& GcCtx)
 			m_CurrentInsertOffset = WriteOffset;
 		}
 
+		// We can break here if we want to do incremental GC
+
 		ChunkIndex++;
 	}
 
@@ -438,33 +442,48 @@ CasContainerStrategy::CollectGarbage(GcContext& GcCtx)
 		m_CurrentInsertOffset = 0;
 	}
 
-//	if (!MovedChunks.empty() || DeletedChunks.empty())
-//	{
-//		// Rewrite object index
-//		std::filesystem::path SidxPath	  = m_Config.RootDirectory / (m_ContainerBaseName + ".uidx");
-//		std::filesystem::path TmpSidxPath = m_Config.RootDirectory / (m_ContainerBaseName + ".gc.uidx");
-//		BasicFile			  TmpSmallObjectIndex;
-//		TmpSmallObjectIndex.Open(TmpSidxPath, true);
-//
-//		std::vector<CasDiskIndexEntry> Entries{m_LocationMap.size()};
-//		uint64_t					   EntryIndex = 0;
-//		for (auto& Entry : m_LocationMap)
-//		{
-//			CasDiskIndexEntry& IndexEntry = Entries[EntryIndex++];
-//			IndexEntry.Key				  = Entry.first;
-//			IndexEntry.Location			  = Entry.second;
-//		}
-//
-//		TmpSmallObjectIndex.Write(Entries.data(), Entries.size() * sizeof(CasDiskIndexEntry), 0);
-//		TmpSmallObjectIndex.Close();
-//
-//		m_SmallObjectIndex.Close();
-//		fs::remove(SidxPath);
-//		fs::rename(TmpSidxPath, SidxPath);
-//		m_SmallObjectIndex.Open(SidxPath, false);
-//	}
-
-	// TODO: Should we flush the cas log? It will grow forever...
+	if (!MovedChunks.empty() || DeletedChunks.empty())
+	{
+		//		// Rewrite object index
+		//		std::filesystem::path SidxPath	  = m_Config.RootDirectory / (m_ContainerBaseName + ".uidx");
+		//		std::filesystem::path TmpSidxPath = m_Config.RootDirectory / (m_ContainerBaseName + ".gc.uidx");
+		//		BasicFile			  TmpSmallObjectIndex;
+		//		TmpSmallObjectIndex.Open(TmpSidxPath, true);
+		//
+		//		std::vector<CasDiskIndexEntry> Entries{m_LocationMap.size()};
+		//		uint64_t					   EntryIndex = 0;
+		//		for (auto& Entry : m_LocationMap)
+		//		{
+		//			CasDiskIndexEntry& IndexEntry = Entries[EntryIndex++];
+		//			IndexEntry.Key				  = Entry.first;
+		//			IndexEntry.Location			  = Entry.second;
+		//		}
+		//
+		//		TmpSmallObjectIndex.Write(Entries.data(), Entries.size() * sizeof(CasDiskIndexEntry), 0);
+		//		TmpSmallObjectIndex.Close();
+		//
+		//		m_SmallObjectIndex.Close();
+		//		fs::remove(SidxPath);
+		//		fs::rename(TmpSidxPath, SidxPath);
+		//		m_SmallObjectIndex.Open(SidxPath, false);
+
+		// Rewrite cas log
+		std::filesystem::path		   SlogPath	   = m_Config.RootDirectory / (m_ContainerBaseName + ".ulog");
+		std::filesystem::path		   TmpSlogPath = m_Config.RootDirectory / (m_ContainerBaseName + ".gc.ulog");
+		TCasLogFile<CasDiskIndexEntry> TmpCasLog;
+		TmpCasLog.Open(TmpSlogPath, true);
+
+		for (auto& Entry : m_LocationMap)
+		{
+			CasDiskIndexEntry IndexEntry{.Key = Entry.first, .Location = Entry.second};
+			TmpCasLog.Append(IndexEntry);
+		}
+		TmpCasLog.Close();
+		m_CasLog.Close();
+		fs::remove(SlogPath);
+		fs::rename(TmpSlogPath, SlogPath);
+		m_CasLog.Open(SlogPath, false);
+	}
 
 	uint64_t CurrentSize = m_SmallObjectFile.FileSize();
 	ZEN_INFO("garbage collection complete '{}', space {} to {}, moved {} and delete {} chunks",
@@ -485,14 +504,14 @@ CasContainerStrategy::OpenContainer(bool IsNewStore)
 	std::filesystem::path SlogPath = m_Config.RootDirectory / (m_ContainerBaseName + ".ulog");
 
 	m_SmallObjectFile.Open(SobsPath, IsNewStore);
-//	m_SmallObjectIndex.Open(SidxPath, IsNewStore);
+	//	m_SmallObjectIndex.Open(SidxPath, IsNewStore);
 	m_CasLog.Open(SlogPath, IsNewStore);
 
 	// TODO: should validate integrity of container files here
 
 	m_CurrentInsertOffset = 0;
-//	m_CurrentIndexOffset  = 0;
-	m_TotalSize			  = 0;
+	//	m_CurrentIndexOffset  = 0;
+	m_TotalSize = 0;
 
 	m_LocationMap.clear();
 
@@ -518,7 +537,7 @@ CasContainerStrategy::OpenContainer(bool IsNewStore)
 	}
 
 	m_CurrentInsertOffset = (MaxFileOffset + m_PayloadAlignment - 1) & ~(m_PayloadAlignment - 1);
-//	m_CurrentIndexOffset = m_SmallObjectIndex.FileSize();
+	//	m_CurrentIndexOffset = m_SmallObjectIndex.FileSize();
 }
 
 //////////////////////////////////////////////////////////////////////////