cache reference tracking (#455)

- Feature: Add caching of referenced CId content for structured cache records, this avoid disk thrashing when gathering references for GC
  - disabled by default, enable with `--cache-reference-cache-enabled`
- Improvement: Faster collection of referenced CId content in project store

1 files changed, 45 insertions, 36 deletions

diff --git a/src/zenserver/cache/structuredcachestore.cpp b/src/zenserver/cache/structuredcachestore.cpp
index fe0b84f33..48463fcd8 100644
--- a/src/zenserver/cache/structuredcachestore.cpp
+++ b/src/zenserver/cache/structuredcachestore.cpp
@@ -61,13 +61,14 @@ IsKnownBadBucketName(std::string_view Bucket)
 ZenCacheNamespace::ZenCacheNamespace(GcManager&								  Gc,
 									 JobQueue&								  JobQueue,
 									 const std::filesystem::path&			  RootDir,
+									 bool									  EnableReferenceCaching,
 									 const ZenCacheMemoryLayer::Configuration MemLayerConfig)
 : GcStorage(Gc)
 , GcContributor(Gc)
 , m_RootDir(RootDir)
 , m_JobQueue(JobQueue)
 , m_MemLayer(m_JobQueue, MemLayerConfig)
-, m_DiskLayer(RootDir)
+, m_DiskLayer(RootDir, EnableReferenceCaching)
 {
 	ZEN_INFO("initializing structured cache at '{}'", RootDir);
 	CreateDirectories(RootDir);
@@ -116,7 +117,7 @@ ZenCacheNamespace::Get(std::string_view InBucket, const IoHash& HashKey, ZenCach
 }
 
 void
-ZenCacheNamespace::Put(std::string_view InBucket, const IoHash& HashKey, const ZenCacheValue& Value)
+ZenCacheNamespace::Put(std::string_view InBucket, const IoHash& HashKey, const ZenCacheValue& Value, std::span<IoHash> References)
 {
 	ZEN_TRACE_CPU("Z$::Namespace::Put");
 
@@ -126,7 +127,7 @@ ZenCacheNamespace::Put(std::string_view InBucket, const IoHash& HashKey, const Z
 
 	ZEN_ASSERT(Value.Value.Size());
 
-	m_DiskLayer.Put(InBucket, HashKey, Value);
+	m_DiskLayer.Put(InBucket, HashKey, Value, References);
 
 	if (Value.Value.Size() <= m_DiskLayerSizeThreshold)
 	{
@@ -321,6 +322,7 @@ ZenCacheStore::ZenCacheStore(GcManager&				 Gc,
 			std::make_unique<ZenCacheNamespace>(Gc,
 												m_JobQueue,
 												m_Configuration.BasePath / fmt::format("{}{}", NamespaceDiskPrefix, NamespaceName),
+												m_Configuration.EnableReferenceCaching,
 												m_Configuration.MemLayerConfig);
 	}
 }
@@ -476,7 +478,8 @@ ZenCacheStore::Put(const CacheRequestContext& Context,
 				   std::string_view			  Namespace,
 				   std::string_view			  Bucket,
 				   const IoHash&			  HashKey,
-				   const ZenCacheValue&		  Value)
+				   const ZenCacheValue&		  Value,
+				   std::span<IoHash>		  References)
 {
 	// Ad hoc rejection of known bad usage patterns for DDC bucket names
 
@@ -512,7 +515,7 @@ ZenCacheStore::Put(const CacheRequestContext& Context,
 
 	if (ZenCacheNamespace* Store = GetNamespace(Namespace); Store)
 	{
-		Store->Put(Bucket, HashKey, Value);
+		Store->Put(Bucket, HashKey, Value, References);
 		m_WriteCount++;
 		return;
 	}
@@ -625,6 +628,7 @@ ZenCacheStore::GetNamespace(std::string_view Namespace)
 		std::make_unique<ZenCacheNamespace>(m_Gc,
 											m_JobQueue,
 											m_Configuration.BasePath / fmt::format("{}{}", NamespaceDiskPrefix, Namespace),
+											m_Configuration.EnableReferenceCaching,
 											m_Configuration.MemLayerConfig));
 	return NewNamespace.first->second.get();
 }
@@ -803,7 +807,7 @@ TEST_CASE("z$.store")
 	GcManager Gc;
 
 	auto			  JobQueue = MakeJobQueue(1, "testqueue");
-	ZenCacheNamespace Zcs(Gc, *JobQueue, TempDir.Path() / "cache");
+	ZenCacheNamespace Zcs(Gc, *JobQueue, TempDir.Path() / "cache", false);
 
 	const int kIterationCount = 100;
 
@@ -819,7 +823,7 @@ TEST_CASE("z$.store")
 		Value.Value = Obj.GetBuffer().AsIoBuffer();
 		Value.Value.SetContentType(ZenContentType::kCbObject);
 
-		Zcs.Put("test_bucket"sv, Key, Value);
+		Zcs.Put("test_bucket"sv, Key, Value, {});
 	}
 
 	for (int i = 0; i < kIterationCount; ++i)
@@ -859,7 +863,7 @@ TEST_CASE("z$.size")
 
 		{
 			GcManager		  Gc;
-			ZenCacheNamespace Zcs(Gc, *JobQueue, TempDir.Path() / "cache");
+			ZenCacheNamespace Zcs(Gc, *JobQueue, TempDir.Path() / "cache", false);
 
 			CbObject CacheValue = CreateCacheValue(Zcs.DiskLayerThreshold() - 256);
 
@@ -869,7 +873,10 @@ TEST_CASE("z$.size")
 			for (size_t Key = 0; Key < Count; ++Key)
 			{
 				const size_t Bucket = Key % 4;
-				Zcs.Put(fmt::format("test_bucket-{}", Bucket), IoHash::HashBuffer(&Key, sizeof(uint32_t)), ZenCacheValue{.Value = Buffer});
+				Zcs.Put(fmt::format("test_bucket-{}", Bucket),
+						IoHash::HashBuffer(&Key, sizeof(uint32_t)),
+						ZenCacheValue{.Value = Buffer},
+						{});
 			}
 
 			CacheSize = Zcs.StorageSize();
@@ -879,7 +886,7 @@ TEST_CASE("z$.size")
 
 		{
 			GcManager		  Gc;
-			ZenCacheNamespace Zcs(Gc, *JobQueue, TempDir.Path() / "cache");
+			ZenCacheNamespace Zcs(Gc, *JobQueue, TempDir.Path() / "cache", false);
 
 			const GcStorageSize SerializedSize = Zcs.StorageSize();
 			CHECK_EQ(SerializedSize.MemorySize, 0);
@@ -902,7 +909,7 @@ TEST_CASE("z$.size")
 
 		{
 			GcManager		  Gc;
-			ZenCacheNamespace Zcs(Gc, *JobQueue, TempDir.Path() / "cache");
+			ZenCacheNamespace Zcs(Gc, *JobQueue, TempDir.Path() / "cache", false);
 
 			CbObject CacheValue = CreateCacheValue(Zcs.DiskLayerThreshold() + 64);
 
@@ -912,7 +919,7 @@ TEST_CASE("z$.size")
 			for (size_t Key = 0; Key < Count; ++Key)
 			{
 				const size_t Bucket = Key % 4;
-				Zcs.Put(fmt::format("test_bucket-{}", Bucket), IoHash::HashBuffer(&Key, sizeof(uint32_t)), {.Value = Buffer});
+				Zcs.Put(fmt::format("test_bucket-{}", Bucket), IoHash::HashBuffer(&Key, sizeof(uint32_t)), {.Value = Buffer}, {});
 			}
 
 			CacheSize = Zcs.StorageSize();
@@ -922,7 +929,7 @@ TEST_CASE("z$.size")
 
 		{
 			GcManager		  Gc;
-			ZenCacheNamespace Zcs(Gc, *JobQueue, TempDir.Path() / "cache");
+			ZenCacheNamespace Zcs(Gc, *JobQueue, TempDir.Path() / "cache", false);
 
 			const GcStorageSize SerializedSize = Zcs.StorageSize();
 			CHECK_EQ(SerializedSize.MemorySize, 0);
@@ -961,7 +968,7 @@ TEST_CASE("z$.gc")
 
 		{
 			GcManager		  Gc;
-			ZenCacheNamespace Zcs(Gc, *JobQueue, TempDir.Path() / "cache");
+			ZenCacheNamespace Zcs(Gc, *JobQueue, TempDir.Path() / "cache", true);
 			const auto		  Bucket = "teardrinker"sv;
 
 			// Create a cache record
@@ -978,7 +985,7 @@ TEST_CASE("z$.gc")
 			IoBuffer Buffer = Record.Save().GetBuffer().AsIoBuffer();
 			Buffer.SetContentType(ZenContentType::kCbObject);
 
-			Zcs.Put(Bucket, Key, {.Value = Buffer});
+			Zcs.Put(Bucket, Key, {.Value = Buffer}, Cids);
 
 			std::vector<IoHash> Keep;
 
@@ -998,7 +1005,7 @@ TEST_CASE("z$.gc")
 		// Expect timestamps to be serialized
 		{
 			GcManager			Gc;
-			ZenCacheNamespace	Zcs(Gc, *JobQueue, TempDir.Path() / "cache");
+			ZenCacheNamespace	Zcs(Gc, *JobQueue, TempDir.Path() / "cache", true);
 			std::vector<IoHash> Keep;
 
 			// Collect garbage with 1 hour max cache duration
@@ -1019,7 +1026,7 @@ TEST_CASE("z$.gc")
 	{
 		ScopedTemporaryDirectory TempDir;
 		GcManager				 Gc;
-		ZenCacheNamespace		 Zcs(Gc, *JobQueue, TempDir.Path() / "cache");
+		ZenCacheNamespace		 Zcs(Gc, *JobQueue, TempDir.Path() / "cache", true);
 		const auto				 Bucket		 = "fortysixandtwo"sv;
 		const GcClock::TimePoint CurrentTime = GcClock::Now();
 
@@ -1028,7 +1035,7 @@ TEST_CASE("z$.gc")
 		for (const auto& Key : Keys)
 		{
 			IoBuffer Value = testutils::CreateBinaryCacheValue(128 << 10);
-			Zcs.Put(Bucket, Key, {.Value = Value});
+			Zcs.Put(Bucket, Key, {.Value = Value}, {});
 		}
 
 		{
@@ -1065,7 +1072,7 @@ TEST_CASE("z$.gc")
 		ScopedTemporaryDirectory TempDir;
 		GcManager				 Gc;
 		{
-			ZenCacheNamespace Zcs(Gc, *JobQueue, TempDir.Path() / "cache");
+			ZenCacheNamespace Zcs(Gc, *JobQueue, TempDir.Path() / "cache", true);
 			const auto		  Bucket = "rightintwo"sv;
 
 			std::vector<IoHash> Keys{CreateKey(1), CreateKey(2), CreateKey(3)};
@@ -1073,7 +1080,7 @@ TEST_CASE("z$.gc")
 			for (const auto& Key : Keys)
 			{
 				IoBuffer Value = testutils::CreateBinaryCacheValue(128);
-				Zcs.Put(Bucket, Key, {.Value = Value});
+				Zcs.Put(Bucket, Key, {.Value = Value}, {});
 			}
 
 			{
@@ -1107,7 +1114,7 @@ TEST_CASE("z$.gc")
 			}
 		}
 		{
-			ZenCacheNamespace Zcs(Gc, *JobQueue, TempDir.Path() / "cache");
+			ZenCacheNamespace Zcs(Gc, *JobQueue, TempDir.Path() / "cache", true);
 			CHECK_EQ(0, Zcs.StorageSize().DiskSize);
 		}
 	}
@@ -1164,14 +1171,14 @@ TEST_CASE("z$.threadedinsert")	// * doctest::skip(true))
 		WorkerThreadPool  ThreadPool(4);
 		GcManager		  Gc;
 		auto			  JobQueue = MakeJobQueue(1, "testqueue");
-		ZenCacheNamespace Zcs(Gc, *JobQueue, TempDir.Path());
+		ZenCacheNamespace Zcs(Gc, *JobQueue, TempDir.Path(), true);
 
 		{
 			std::atomic<size_t> WorkCompleted = 0;
 			for (const auto& Chunk : Chunks)
 			{
 				ThreadPool.ScheduleWork([&Zcs, &WorkCompleted, &Chunk]() {
-					Zcs.Put(Chunk.second.Bucket, Chunk.first, {.Value = Chunk.second.Buffer});
+					Zcs.Put(Chunk.second.Bucket, Chunk.first, {.Value = Chunk.second.Buffer}, {});
 					WorkCompleted.fetch_add(1);
 				});
 			}
@@ -1232,7 +1239,7 @@ TEST_CASE("z$.threadedinsert")	// * doctest::skip(true))
 			for (const auto& Chunk : NewChunks)
 			{
 				ThreadPool.ScheduleWork([&Zcs, &WorkCompleted, Chunk, &AddedChunkCount]() {
-					Zcs.Put(Chunk.second.Bucket, Chunk.first, {.Value = Chunk.second.Buffer});
+					Zcs.Put(Chunk.second.Bucket, Chunk.first, {.Value = Chunk.second.Buffer}, {});
 					AddedChunkCount.fetch_add(1);
 					WorkCompleted.fetch_add(1);
 				});
@@ -1395,14 +1402,14 @@ TEST_CASE("z$.namespaces")
 		Buffer.SetContentType(ZenContentType::kCbObject);
 
 		ZenCacheValue PutValue = {.Value = Buffer};
-		Zcs.Put(Context, ZenCacheStore::DefaultNamespace, Bucket, Key1, PutValue);
+		Zcs.Put(Context, ZenCacheStore::DefaultNamespace, Bucket, Key1, PutValue, {});
 
 		ZenCacheValue GetValue;
 		CHECK(Zcs.Get(Context, ZenCacheStore::DefaultNamespace, Bucket, Key1, GetValue));
 		CHECK(!Zcs.Get(Context, CustomNamespace, Bucket, Key1, GetValue));
 
 		// This should just be dropped as we don't allow creating of namespaces on the fly
-		Zcs.Put(Context, CustomNamespace, Bucket, Key1, PutValue);
+		Zcs.Put(Context, CustomNamespace, Bucket, Key1, PutValue, {});
 		CHECK(!Zcs.Get(Context, CustomNamespace, Bucket, Key1, GetValue));
 	}
 
@@ -1418,7 +1425,7 @@ TEST_CASE("z$.namespaces")
 		IoBuffer Buffer2 = CacheValue2.GetBuffer().AsIoBuffer();
 		Buffer2.SetContentType(ZenContentType::kCbObject);
 		ZenCacheValue PutValue2 = {.Value = Buffer2};
-		Zcs.Put(Context, CustomNamespace, Bucket, Key2, PutValue2);
+		Zcs.Put(Context, CustomNamespace, Bucket, Key2, PutValue2, {});
 
 		ZenCacheValue GetValue;
 		CHECK(!Zcs.Get(Context, ZenCacheStore::DefaultNamespace, Bucket, Key2, GetValue));
@@ -1460,7 +1467,7 @@ TEST_CASE("z$.drop.bucket")
 		Buffer.SetContentType(ZenContentType::kCbObject);
 
 		ZenCacheValue PutValue = {.Value = Buffer};
-		Zcs.Put(Context, Namespace, Bucket, Key, PutValue);
+		Zcs.Put(Context, Namespace, Bucket, Key, PutValue, {});
 		return Key;
 	};
 	auto GetValue = [&Context](ZenCacheStore& Zcs, std::string_view Namespace, std::string_view Bucket, const IoHash& Key) {
@@ -1533,7 +1540,7 @@ TEST_CASE("z$.drop.namespace")
 		Buffer.SetContentType(ZenContentType::kCbObject);
 
 		ZenCacheValue PutValue = {.Value = Buffer};
-		Zcs.Put(Context, Namespace, Bucket, Key, PutValue);
+		Zcs.Put(Context, Namespace, Bucket, Key, PutValue, {});
 		return Key;
 	};
 	auto GetValue = [&Context](ZenCacheStore& Zcs, std::string_view Namespace, std::string_view Bucket, const IoHash& Key) {
@@ -1610,7 +1617,7 @@ TEST_CASE("z$.blocked.disklayer.put")
 
 	GcManager		  Gc;
 	auto			  JobQueue = MakeJobQueue(1, "testqueue");
-	ZenCacheNamespace Zcs(Gc, *JobQueue, TempDir.Path() / "cache");
+	ZenCacheNamespace Zcs(Gc, *JobQueue, TempDir.Path() / "cache", false);
 
 	CbObject CacheValue = CreateCacheValue(64 * 1024 + 64);
 
@@ -1619,7 +1626,7 @@ TEST_CASE("z$.blocked.disklayer.put")
 
 	size_t Key	   = Buffer.Size();
 	IoHash HashKey = IoHash::HashBuffer(&Key, sizeof(uint32_t));
-	Zcs.Put("test_bucket", HashKey, {.Value = Buffer});
+	Zcs.Put("test_bucket", HashKey, {.Value = Buffer}, {});
 
 	ZenCacheValue BufferGet;
 	CHECK(Zcs.Get("test_bucket", HashKey, BufferGet));
@@ -1629,7 +1636,7 @@ TEST_CASE("z$.blocked.disklayer.put")
 	Buffer2.SetContentType(ZenContentType::kCbObject);
 
 	// We should be able to overwrite even if the file is open for read
-	Zcs.Put("test_bucket", HashKey, {.Value = Buffer2});
+	Zcs.Put("test_bucket", HashKey, {.Value = Buffer2}, {});
 
 	MemoryView OldView = BufferGet.Value.GetView();
 
@@ -1705,7 +1712,7 @@ TEST_CASE("z$.scrub")
 	GcManager			  Gc;
 	CidStore			  CidStore(Gc);
 	auto				  JobQueue = MakeJobQueue(1, "testqueue");
-	ZenCacheNamespace	  Zcs(Gc, *JobQueue, TempDir.Path() / "cache");
+	ZenCacheNamespace	  Zcs(Gc, *JobQueue, TempDir.Path() / "cache", true);
 	CidStoreConfiguration CidConfig = {.RootDirectory = TempDir.Path() / "cas", .TinyValueThreshold = 1024, .HugeValueThreshold = 4096};
 	CidStore.Initialize(CidConfig);
 
@@ -1713,12 +1720,14 @@ TEST_CASE("z$.scrub")
 		[&](bool IsStructured, std::string_view BucketName, const std::vector<IoHash>& Cids, const std::vector<size_t>& AttachmentSizes) {
 			for (const IoHash& Cid : Cids)
 			{
-				CacheRecord Record = CreateCacheRecord(IsStructured, BucketName, Cid, AttachmentSizes);
-				Zcs.Put("mybucket", Cid, {.Value = Record.Record});
+				CacheRecord			Record = CreateCacheRecord(IsStructured, BucketName, Cid, AttachmentSizes);
+				std::vector<IoHash> AttachmentHashes;
 				for (const CompressedBuffer& Attachment : Record.Attachments)
 				{
-					CidStore.AddChunk(Attachment.GetCompressed().Flatten().AsIoBuffer(), Attachment.DecodeRawHash());
+					AttachmentHashes.push_back(Attachment.DecodeRawHash());
+					CidStore.AddChunk(Attachment.GetCompressed().Flatten().AsIoBuffer(), AttachmentHashes.back());
 				}
+				Zcs.Put("mybucket", Cid, {.Value = Record.Record}, AttachmentHashes);
 			}
 		};