Implemented move-in-place for large CAS payloads

1 files changed, 108 insertions, 0 deletions

diff --git a/zenstore/filecas.cpp b/zenstore/filecas.cpp
index 84a06c3be..0011ddec6 100644
--- a/zenstore/filecas.cpp
+++ b/zenstore/filecas.cpp
@@ -3,11 +3,14 @@
 #include "FileCas.h"
 
 #include <zencore/except.h>
+#include <zencore/filesystem.h>
+#include <zencore/fmtutils.h>
 #include <zencore/memory.h>
 #include <zencore/string.h>
 #include <zencore/thread.h>
 #include <zencore/uid.h>
 
+#include <spdlog/spdlog.h>
 #include <gsl/gsl-lite.hpp>
 
 #include <functional>
@@ -25,6 +28,8 @@ struct IUnknown;  // Workaround for "combaseapi.h(229): error C2187: syntax erro
 
 namespace zen {
 
+using namespace fmt::literals;
+
 WideStringBuilderBase&
 FileCasStrategy::MakeShardedPath(WideStringBuilderBase& ShardedPath, const IoHash& ChunkHash, size_t& OutShard2len)
 {
@@ -58,6 +63,109 @@ FileCasStrategy::MakeShardedPath(WideStringBuilderBase& ShardedPath, const IoHas
 CasStore::InsertResult
 FileCasStrategy::InsertChunk(IoBuffer Chunk, const IoHash& ChunkHash)
 {
+	IoBufferFileReference FileRef;
+	if (Chunk.GetFileReference(/* out */ FileRef))
+	{
+		size_t							 Shard2len = 0;
+		ExtendableWideStringBuilder<128> ShardedPath;
+		ShardedPath.Append(m_Config.RootDirectory.c_str());
+		ShardedPath.Append(std::filesystem::path::preferred_separator);
+		MakeShardedPath(ShardedPath, ChunkHash, /* out */ Shard2len);
+
+		auto DeletePayloadFileOnClose = [&] {
+			FILE_DISPOSITION_INFO Fdi{};
+			Fdi.DeleteFile = TRUE;
+			BOOL Success   = SetFileInformationByHandle(FileRef.FileHandle, FileDispositionInfo, &Fdi, sizeof Fdi);
+
+			if (!Success)
+			{
+				spdlog::warn("Failed to flag temporary payload file for deletion: '{}'", PathFromHandle(FileRef.FileHandle));
+			}
+		};
+
+		// See if file already exists
+		//
+		// Future improvement: maintain Bloom filter to avoid expensive file system probes?
+
+		{
+			CAtlFile PayloadFile;
+
+			HRESULT hRes = PayloadFile.Create(ShardedPath.c_str(), GENERIC_READ, FILE_SHARE_READ, OPEN_EXISTING);
+
+			if (SUCCEEDED(hRes))
+			{
+				// If we succeeded in opening the file then we don't need to do anything else because it already exists and should contain
+				// the content we were about to insert
+
+				// We do need to ensure the file goes away on close, however
+
+				DeletePayloadFileOnClose();
+
+				return CasStore::InsertResult{.New = false};
+			}
+		}
+
+		std::filesystem::path FullPath(ShardedPath.c_str());
+
+		std::filesystem::path FilePath = FullPath.parent_path();
+		std::wstring		  FileName = FullPath.native();
+
+		const DWORD		  BufferSize = sizeof(FILE_RENAME_INFO) + gsl::narrow<DWORD>(FileName.size() * sizeof(WCHAR));
+		FILE_RENAME_INFO* RenameInfo = reinterpret_cast<FILE_RENAME_INFO*>(Memory::Alloc(BufferSize));
+		memset(RenameInfo, 0, BufferSize);
+
+		RenameInfo->ReplaceIfExists = FALSE;
+		RenameInfo->FileNameLength	= gsl::narrow<DWORD>(FileName.size());
+		memcpy(RenameInfo->FileName, FileName.c_str(), FileName.size() * sizeof(WCHAR));
+		RenameInfo->FileName[FileName.size()] = 0;
+
+		// Try to move file into place
+
+		BOOL Success = SetFileInformationByHandle(FileRef.FileHandle, FileRenameInfo, RenameInfo, BufferSize);
+
+		if (!Success)
+		{
+			CAtlFile DirHandle;
+
+			auto InternalCreateDirectoryHandle = [&] {
+				return DirHandle.Create(FilePath.c_str(),
+										GENERIC_READ | GENERIC_WRITE,
+										FILE_SHARE_READ | FILE_SHARE_WRITE,
+										OPEN_EXISTING,
+										FILE_FLAG_BACKUP_SEMANTICS);
+			};
+
+			HRESULT hRes = InternalCreateDirectoryHandle();
+
+			if (FAILED(hRes))
+			{
+				zen::CreateDirectories(FilePath.c_str());
+
+				hRes = InternalCreateDirectoryHandle();
+			}
+
+			if (FAILED(hRes))
+			{
+				throw WindowsException(hRes, "Failed to open shard directory '{}'"_format(FilePath));
+			}
+
+			// Retry
+
+			Success = SetFileInformationByHandle(FileRef.FileHandle, FileRenameInfo, RenameInfo, BufferSize);
+		}
+
+		Memory::Free(RenameInfo);
+
+		if (Success)
+		{
+			return CasStore::InsertResult{.New = true};
+		}
+
+		spdlog::warn("rename of CAS payload file failed, falling back to regular write for {}", ChunkHash);
+
+		DeletePayloadFileOnClose();
+	}
+
 	return InsertChunk(Chunk.Data(), Chunk.Size(), ChunkHash);
 }