From 44742c8b56a018cdb7607ec9f0acf76a3bfd6877 Mon Sep 17 00:00:00 2001
From: cqy123456 <39671710+cqy123456@users.noreply.github.com>
Date: Thu, 5 Sep 2024 18:07:05 +0800
Subject: [PATCH] enhance: reduce mmap_rss after chunkcache warmup (#35974)

related pr: https://github.com/milvus-io/milvus/pull/35965

Signed-off-by: cqy123456 <qianya.cheng@zilliz.com>
---
 .../core/src/segcore/SegmentSealedImpl.cpp    |  8 +++++--
 internal/core/src/storage/ChunkCache.cpp      | 21 +++++++++++++++++--
 internal/core/src/storage/ChunkCache.h        |  3 ++-
 3 files changed, 27 insertions(+), 5 deletions(-)

diff --git a/internal/core/src/segcore/SegmentSealedImpl.cpp b/internal/core/src/segcore/SegmentSealedImpl.cpp
index 6154678391c81..e1f37e4b4c46f 100644
--- a/internal/core/src/segcore/SegmentSealedImpl.cpp
+++ b/internal/core/src/segcore/SegmentSealedImpl.cpp
@@ -152,9 +152,13 @@ SegmentSealedImpl::WarmupChunkCache(const FieldId field_id, bool mmap_enabled) {
     auto field_info = it->second;
 
     auto cc = storage::MmapManager::GetInstance().GetChunkCache();
+    bool mmap_rss_not_need = true;
     for (const auto& data_path : field_info.insert_files) {
-        auto column =
-            cc->Read(data_path, mmap_descriptor_, field_meta, mmap_enabled);
+        auto column = cc->Read(data_path,
+                               mmap_descriptor_,
+                               field_meta,
+                               mmap_enabled,
+                               mmap_rss_not_need);
     }
 }
 
diff --git a/internal/core/src/storage/ChunkCache.cpp b/internal/core/src/storage/ChunkCache.cpp
index 10871babf9198..54661453d9d5d 100644
--- a/internal/core/src/storage/ChunkCache.cpp
+++ b/internal/core/src/storage/ChunkCache.cpp
@@ -24,7 +24,8 @@ std::shared_ptr<ColumnBase>
 ChunkCache::Read(const std::string& filepath,
                  const MmapChunkDescriptorPtr& descriptor,
                  const FieldMeta& field_meta,
-                 bool mmap_enabled) {
+                 bool mmap_enabled,
+                 bool mmap_rss_not_need) {
     // use rlock to get future
     {
         std::shared_lock lck(mutex_);
@@ -64,6 +65,22 @@ ChunkCache::Read(const std::string& filepath,
         field_data = DownloadAndDecodeRemoteFile(cm_.get(), filepath);
         column = Mmap(
             field_data->GetFieldData(), descriptor, field_meta, mmap_enabled);
+        if (mmap_enabled && mmap_rss_not_need) {
+            auto ok = madvise(reinterpret_cast<void*>(
+                                  const_cast<char*>(column->MmappedData())),
+                              column->ByteSize(),
+                              ReadAheadPolicy_Map["dontneed"]);
+            if (ok != 0) {
+                LOG_WARN(
+                    "failed to madvise to the data file {}, addr {}, size {}, "
+                    "err: "
+                    "{}",
+                    filepath,
+                    static_cast<const void*>(column->MmappedData()),
+                    column->ByteSize(),
+                    strerror(errno));
+            }
+        }
         allocate_success = true;
     } catch (const SegcoreError& e) {
         err_code = e.get_error_code();
@@ -113,7 +130,7 @@ ChunkCache::Prefetch(const std::string& filepath) {
         LOG_WARN(
             "failed to madvise to the data file {}, addr {}, size {}, err: {}",
             filepath,
-            column->MmappedData(),
+            static_cast<const void*>(column->MmappedData()),
             column->ByteSize(),
             strerror(errno));
     }
diff --git a/internal/core/src/storage/ChunkCache.h b/internal/core/src/storage/ChunkCache.h
index b2b36c6ec4f30..64be0ae319540 100644
--- a/internal/core/src/storage/ChunkCache.h
+++ b/internal/core/src/storage/ChunkCache.h
@@ -48,7 +48,8 @@ class ChunkCache {
     Read(const std::string& filepath,
          const MmapChunkDescriptorPtr& descriptor,
          const FieldMeta& field_meta,
-         bool mmap_enabled);
+         bool mmap_enabled,
+         bool mmap_rss_not_need = false);
 
     void
     Remove(const std::string& filepath);