tweak iterate block parameters (#390)

* tweak block iteration chunk sizes

2 files changed, 50 insertions, 34 deletions

diff --git a/src/zenstore/cache/cachedisklayer.cpp b/src/zenstore/cache/cachedisklayer.cpp
index fbe559e58..8b120fda5 100644
--- a/src/zenstore/cache/cachedisklayer.cpp
+++ b/src/zenstore/cache/cachedisklayer.cpp
@@ -1449,7 +1449,7 @@ ZenCacheDiskLayer::CacheBucket::EndGetBatch(GetBatchHandle* Batch) noexcept
 			// Often we will find the metadata due to the thread setting the mem cached part doing it before us so it is worth
 			// checking if it is present once more before spending time fetching and setting the RawHash and RawSize in metadata
 
-			auto FillOne = [&](const DiskLocation& Location, size_t KeyIndex, IoBuffer&& Value) {
+			auto FillOne = [&](const DiskLocation& Location, size_t KeyIndex, IoBuffer&& Value, bool UsesTemporaryMemory) {
 				if (!Value)
 				{
 					return;
@@ -1472,6 +1472,12 @@ ZenCacheDiskLayer::CacheBucket::EndGetBatch(GetBatchHandle* Batch) noexcept
 					}
 				}
 
+				if (AddToMemCache || UsesTemporaryMemory)
+				{
+					// We need to own it if we want to add it to the memcache or the buffer is just a range of the block iteration buffer
+					OutValue.Value.MakeOwned();
+				}
+
 				if (SetMetaInfo)
 				{
 					// See ZenCacheDiskLayer::CacheBucket::Get - it sets the memcache part first and then if it needs to it set the
@@ -1543,35 +1549,42 @@ ZenCacheDiskLayer::CacheBucket::EndGetBatch(GetBatchHandle* Batch) noexcept
 			if (!InlineDiskLocations.empty())
 			{
 				ZEN_TRACE_CPU("Z$::Bucket::EndGetBatch::ReadInline");
-				m_BlockStore.IterateChunks(
-					std::span{begin(InlineBlockLocations), end(InlineBlockLocations)},
-					[&](uint32_t, std::span<const size_t> ChunkIndexes) -> bool {
-						// Only read into memory the IoBuffers we could potentially add to memcache
-						const uint64_t LargeChunkSizeLimit = Max(m_Configuration.MemCacheSizeThreshold, 1u * 1024u);
-						m_BlockStore.IterateBlock(
-							std::span{begin(InlineBlockLocations), end(InlineBlockLocations)},
-							ChunkIndexes,
-							[this, &FillOne, &InlineDiskLocations, &InlineKeyIndexes](size_t	  ChunkIndex,
-																					  const void* Data,
-																					  uint64_t	  Size) -> bool {
-								if (Data != nullptr)
-								{
-									FillOne(InlineDiskLocations[ChunkIndex],
-											InlineKeyIndexes[ChunkIndex],
-											IoBufferBuilder::MakeCloneFromMemory(Data, Size));
-								}
-								return true;
-							},
-							[this, &FillOne, &InlineDiskLocations, &InlineKeyIndexes](size_t		  ChunkIndex,
-																					  BlockStoreFile& File,
-																					  uint64_t		  Offset,
-																					  uint64_t		  Size) -> bool {
-								FillOne(InlineDiskLocations[ChunkIndex], InlineKeyIndexes[ChunkIndex], File.GetChunk(Offset, Size));
-								return true;
-							},
-							LargeChunkSizeLimit);
-						return true;
-					});
+				m_BlockStore.IterateChunks(std::span{begin(InlineBlockLocations), end(InlineBlockLocations)},
+										   [&](uint32_t, std::span<const size_t> ChunkIndexes) -> bool {
+											   // Up to 8KB or m_Configuration.MemCacheSizeThreshold depending on configuration
+											   const uint64_t LargeChunkSizeLimit =
+												   m_Configuration.MemCacheSizeThreshold == 0
+													   ? Min(m_Configuration.LargeObjectThreshold, 8u * 1024u)
+													   : Max(m_Configuration.MemCacheSizeThreshold, 8u * 1024u);
+
+											   m_BlockStore.IterateBlock(
+												   std::span{begin(InlineBlockLocations), end(InlineBlockLocations)},
+												   ChunkIndexes,
+												   [this, &FillOne, &InlineDiskLocations, &InlineKeyIndexes](size_t		 ChunkIndex,
+																											 const void* Data,
+																											 uint64_t	 Size) -> bool {
+													   if (Data != nullptr)
+													   {
+														   FillOne(InlineDiskLocations[ChunkIndex],
+																   InlineKeyIndexes[ChunkIndex],
+																   IoBufferBuilder::MakeFromMemory(MemoryView(Data, Size)),
+																   /*UsesTemporaryMemory*/ true);
+													   }
+													   return true;
+												   },
+												   [this, &FillOne, &InlineDiskLocations, &InlineKeyIndexes](size_t			 ChunkIndex,
+																											 BlockStoreFile& File,
+																											 uint64_t		 Offset,
+																											 uint64_t		 Size) -> bool {
+													   FillOne(InlineDiskLocations[ChunkIndex],
+															   InlineKeyIndexes[ChunkIndex],
+															   File.GetChunk(Offset, Size),
+															   /*UsesTemporaryMemory*/ false);
+													   return true;
+												   },
+												   LargeChunkSizeLimit);
+											   return true;
+										   });
 			}
 
 			if (!StandaloneDiskLocations.empty())
@@ -1581,7 +1594,7 @@ ZenCacheDiskLayer::CacheBucket::EndGetBatch(GetBatchHandle* Batch) noexcept
 				{
 					size_t				KeyIndex = StandaloneKeyIndexes[Index];
 					const DiskLocation& Location = StandaloneDiskLocations[Index];
-					FillOne(Location, KeyIndex, GetStandaloneCacheValue(Location, Batch->Keys[KeyIndex]));
+					FillOne(Location, KeyIndex, GetStandaloneCacheValue(Location, Batch->Keys[KeyIndex]), /*UsesTemporaryMemory*/ false);
 				}
 			}
 
@@ -3289,7 +3302,7 @@ ZenCacheDiskLayer::CacheBucket::GetReferences(const LoggerRef&	   Logger,
 						CaptureAttachments(ChunkIndex, File.GetChunk(Offset, Size).GetView());
 						return !IsCancelledFlag.load();
 					},
-					0);
+					32u * 1024);
 
 				if (Continue)
 				{
diff --git a/src/zenstore/cas.cpp b/src/zenstore/cas.cpp
index ed42f254e..460f0e10d 100644
--- a/src/zenstore/cas.cpp
+++ b/src/zenstore/cas.cpp
@@ -412,6 +412,7 @@ CasImpl::IterateChunks(std::span<IoHash>												 DecompressedIds,
 					   uint64_t															 LargeSizeLimit)
 {
 	ZEN_TRACE_CPU("CAS::IterateChunks");
+
 	if (!m_SmallStrategy.IterateChunks(
 			DecompressedIds,
 			[&](size_t Index, const IoBuffer& Payload) {
@@ -420,10 +421,11 @@ CasImpl::IterateChunks(std::span<IoHash>												 DecompressedIds,
 				return AsyncCallback(Index, Payload);
 			},
 			OptionalWorkerPool,
-			LargeSizeLimit))
+			LargeSizeLimit == 0 ? m_Config.HugeValueThreshold : Min(LargeSizeLimit, m_Config.HugeValueThreshold)))
 	{
 		return false;
 	}
+
 	if (!m_TinyStrategy.IterateChunks(
 			DecompressedIds,
 			[&](size_t Index, const IoBuffer& Payload) {
@@ -432,10 +434,11 @@ CasImpl::IterateChunks(std::span<IoHash>												 DecompressedIds,
 				return AsyncCallback(Index, Payload);
 			},
 			OptionalWorkerPool,
-			LargeSizeLimit))
+			LargeSizeLimit == 0 ? m_Config.TinyValueThreshold : Min(LargeSizeLimit, m_Config.TinyValueThreshold)))
 	{
 		return false;
 	}
+
 	if (!m_LargeStrategy.IterateChunks(
 			DecompressedIds,
 			[&](size_t Index, const IoBuffer& Payload) {