Track disk io for reading disk index posting lists.

vespa-engine · Oct 22, 2024 · 5e4c750 · 5e4c750
1 parent bbc6c94
commit 5e4c750
Show file tree

Hide file tree

Showing 10 changed files with 129 additions and 14 deletions.
diff --git a/searchlib/src/tests/util/searchable_stats/searchable_stats_test.cpp b/searchlib/src/tests/util/searchable_stats/searchable_stats_test.cpp
@@ -41,17 +41,25 @@ TEST(SearchableStatsTest, stats_can_be_merged)
 TEST(SearchableStatsTest, field_stats_can_be_merged)
 {
     SearchableStats base_stats;
-    base_stats.add_field_stats("f1", FieldIndexStats().memory_usage({100, 40, 10, 5}).size_on_disk(1000)).
-        add_field_stats("f2", FieldIndexStats().memory_usage({400, 200, 60, 10}).size_on_disk(1500));
+    auto f1_stats = FieldIndexStats().memory_usage({100, 40, 10, 5}).size_on_disk(1000).
+        disk_io_stats(DiskIoStats().read_operations(1).read_bytes(1000));
+    auto f2_stats1 = FieldIndexStats().memory_usage({400, 200, 60, 10}).size_on_disk(1500).
+        disk_io_stats(DiskIoStats().read_operations(1).read_bytes(1000));
+    auto f2_stats2 = FieldIndexStats().memory_usage({300, 100, 40, 5}).size_on_disk(500).
+        disk_io_stats(DiskIoStats().read_operations(4).read_bytes(6000));
+    auto f2_stats3 = FieldIndexStats().memory_usage({700, 300, 100, 15}).size_on_disk(2000).
+        disk_io_stats(DiskIoStats().read_operations(5).read_bytes(7000));
+    auto f3_stats = FieldIndexStats().memory_usage({110, 50, 20, 12}).size_on_disk(500).
+        disk_io_stats(DiskIoStats().read_operations(1).read_bytes(1000));
+    base_stats.add_field_stats("f1", f1_stats).add_field_stats("f2", f2_stats1);
     SearchableStats added_stats;
-    added_stats.add_field_stats("f2", FieldIndexStats().memory_usage({300, 100, 40, 5}).size_on_disk(500)).
-        add_field_stats("f3", FieldIndexStats().memory_usage({110, 50, 20, 12}).size_on_disk(500));
+    added_stats.add_field_stats("f2", f2_stats2).add_field_stats("f3", f3_stats);
     SearchableStats act_stats = base_stats;
     act_stats.merge(added_stats);
     SearchableStats exp_stats;
-    exp_stats.add_field_stats("f1", FieldIndexStats().memory_usage({100, 40, 10, 5}).size_on_disk(1000)).
-        add_field_stats("f2", FieldIndexStats().memory_usage({700, 300, 100, 15}).size_on_disk(2000)).
-        add_field_stats("f3", FieldIndexStats().memory_usage({110, 50, 20, 12}).size_on_disk(500));
+    exp_stats.add_field_stats("f1", f1_stats).
+        add_field_stats("f2", f2_stats3).
+        add_field_stats("f3", f3_stats);
     EXPECT_EQ(exp_stats, act_stats);
 }
 

diff --git a/searchlib/src/vespa/searchlib/diskindex/field_index.cpp b/searchlib/src/vespa/searchlib/diskindex/field_index.cpp
@@ -27,13 +27,21 @@ const std::vector<std::string> field_file_names{
 
 }
 
+FieldIndex::LockedDiskIoStats::LockedDiskIoStats() noexcept
+    : DiskIoStats(),
+      _mutex()
+{
+}
+
+FieldIndex::LockedDiskIoStats::~LockedDiskIoStats() = default;
+
 FieldIndex::FieldIndex()
     : _posting_file(),
       _bit_vector_dict(),
       _dict(),
-      _size_on_disk(0)
+      _size_on_disk(0),
+      _disk_io_stats(std::make_shared<LockedDiskIoStats>())
 {
-
 }
 
 FieldIndex::FieldIndex(FieldIndex&&) = default;
@@ -146,6 +154,9 @@ FieldIndex::read_posting_list(const DictionaryLookupResult& lookup_result) const
         return {};
     }
     handle->_file->readPostingList(*handle);
+    if (handle->_read_bytes != 0) {
+        _disk_io_stats->add_read_operation(handle->_read_bytes);
+    }
     return handle;
 }
 
@@ -167,7 +178,8 @@ FieldIndex::get_field_length_info() const
 FieldIndexStats
 FieldIndex::get_stats() const
 {
-    return FieldIndexStats().size_on_disk(_size_on_disk);
+    auto disk_io_stats = _disk_io_stats->read_and_clear();
+    return FieldIndexStats().size_on_disk(_size_on_disk).disk_io_stats(disk_io_stats);
 }
 
 }
diff --git a/searchlib/src/vespa/searchlib/diskindex/field_index.h b/searchlib/src/vespa/searchlib/diskindex/field_index.h
@@ -22,10 +22,29 @@ class FieldIndex {
     using DiskPostingFileReal = Zc4PosOccRandRead;
     using DiskPostingFileDynamicKReal = ZcPosOccRandRead;
 
+    class LockedDiskIoStats : public DiskIoStats {
+        std::mutex _mutex;
+
+    public:
+        LockedDiskIoStats() noexcept;
+        ~LockedDiskIoStats();
+
+        void add_read_operation(uint64_t bytes) {
+            std::lock_guard guard(_mutex);
+            DiskIoStats::add_read_operation(bytes);
+        }
+
+        DiskIoStats read_and_clear() {
+            std::lock_guard guard(_mutex);
+            return DiskIoStats::read_and_clear();
+        }
+    };
+
     std::shared_ptr<DiskPostingFile> _posting_file;
     std::shared_ptr<BitVectorDictionary> _bit_vector_dict;
     std::unique_ptr<index::DictionaryFileRandRead> _dict;
     uint64_t _size_on_disk;
+    std::shared_ptr<LockedDiskIoStats> _disk_io_stats;
 
 public:
     FieldIndex();

diff --git a/searchlib/src/vespa/searchlib/diskindex/zcposoccrandread.cpp b/searchlib/src/vespa/searchlib/diskindex/zcposoccrandread.cpp
@@ -136,6 +136,7 @@ ZcPosOccRandRead::readPostingList(PostingListHandle &handle)
         handle._mem = static_cast<char *>(alignedBuffer) + padBefore;
         handle._allocMem = mallocStart;
         handle._allocSize = mallocLen;
+        handle._read_bytes = padBefore + vectorLen + padAfter;
     }
     handle._bitOffsetMem = (startOffset << 3) - _headerBitSize;
 }

diff --git a/searchlib/src/vespa/searchlib/index/postinglisthandle.h b/searchlib/src/vespa/searchlib/index/postinglisthandle.h
@@ -31,6 +31,7 @@ class PostingListHandle {
     const void *_mem;       // Memory backing posting list after read/mmap
     void *_allocMem;        // What to free after posting list
     size_t _allocSize;      // Size of allocated memory
+    uint64_t _read_bytes;   // Bytes read from disk (used by disk io stats)
 
     PostingListHandle()
     : _file(nullptr),
@@ -39,7 +40,8 @@ class PostingListHandle {
       _bitOffsetMem(0),
       _mem(nullptr),
       _allocMem(nullptr),
-      _allocSize(0)
+      _allocSize(0),
+      _read_bytes(0)
     { }
 
     ~PostingListHandle()
@@ -68,6 +70,7 @@ class PostingListHandle {
             _allocMem = nullptr;
         }
         _allocSize = 0;
+        _read_bytes = 0;
     }
 };
 

diff --git a/searchlib/src/vespa/searchlib/util/CMakeLists.txt b/searchlib/src/vespa/searchlib/util/CMakeLists.txt
@@ -5,6 +5,7 @@ vespa_add_library(searchlib_util OBJECT
     comprbuffer.cpp
     comprfile.cpp
     data_buffer_writer.cpp
+    disk_io_stats.cpp
     dirtraverse.cpp
     drainingbufferwriter.cpp
     field_index_stats.cpp

diff --git a/searchlib/src/vespa/searchlib/util/disk_io_stats.cpp b/searchlib/src/vespa/searchlib/util/disk_io_stats.cpp
@@ -0,0 +1,13 @@
+// Copyright Vespa.ai. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+
+#include "disk_io_stats.h"
+#include <ostream>
+
+namespace search {
+
+std::ostream& operator<<(std::ostream& os, const DiskIoStats& stats) {
+    os << "{read_operations: " << stats.read_operations() << ", read_bytes: " << stats.read_bytes() << "}";
+    return os;
+}
+
+}
diff --git a/searchlib/src/vespa/searchlib/util/disk_io_stats.h b/searchlib/src/vespa/searchlib/util/disk_io_stats.h
@@ -0,0 +1,48 @@
+// Copyright Vespa.ai. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+#pragma once
+
+#include <cstdint>
+#include <iosfwd>
+
+namespace search {
+
+/*
+ * Class tracking disk io.
+ */
+class DiskIoStats {
+    uint64_t _read_operations;
+    uint64_t _read_bytes;
+
+public:
+    DiskIoStats() noexcept
+        : _read_operations(0),
+          _read_bytes(0)
+    {}
+
+    void add_read_operation(uint64_t bytes) noexcept {
+        ++_read_operations;
+        _read_bytes += bytes;
+    }
+    void merge(const DiskIoStats& rhs) noexcept {
+        _read_operations += rhs._read_operations;
+        _read_bytes += rhs._read_bytes;
+    }
+    bool operator==(const DiskIoStats& rhs) const noexcept {
+        return _read_operations == rhs._read_operations &&
+               _read_bytes == rhs._read_bytes;
+    }
+    void clear() noexcept {
+        _read_operations = 0;
+        _read_bytes = 0;
+    }
+    DiskIoStats read_and_clear() noexcept { auto result = *this; clear(); return result; }
+
+    DiskIoStats& read_operations(uint64_t value) { _read_operations = value; return *this; }
+    DiskIoStats& read_bytes(uint64_t value) { _read_bytes = value; return *this; }
+    uint64_t read_operations() const noexcept { return _read_operations; }
+    uint64_t read_bytes() const noexcept { return _read_bytes; }
+};
+
+std::ostream& operator<<(std::ostream& os, const DiskIoStats& stats);
+
+}
diff --git a/searchlib/src/vespa/searchlib/util/field_index_stats.cpp b/searchlib/src/vespa/searchlib/util/field_index_stats.cpp
@@ -6,7 +6,8 @@
 namespace search {
 
 std::ostream& operator<<(std::ostream& os, const FieldIndexStats& stats) {
-    os << "{memory: " << stats.memory_usage() << ", disk: " << stats.size_on_disk() << "}";
+    os << "{memory: " << stats.memory_usage() << ", disk: " << stats.size_on_disk() <<
+    ", diskio: " << stats.disk_io_stats() << "}";
     return os;
 }
 

diff --git a/searchlib/src/vespa/searchlib/util/field_index_stats.h b/searchlib/src/vespa/searchlib/util/field_index_stats.h
@@ -1,7 +1,9 @@
 // Copyright Vespa.ai. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
 #pragma once
 
+#include "disk_io_stats.h"
 #include <vespa/vespalib/util/memoryusage.h>
+#include <iosfwd>
 
 namespace search {
 
@@ -13,11 +15,13 @@ class FieldIndexStats
 private:
     vespalib::MemoryUsage _memory_usage;
     size_t _size_on_disk; // in bytes
+    DiskIoStats _disk_io_stats;
 
 public:
     FieldIndexStats() noexcept
         : _memory_usage(),
-          _size_on_disk(0)
+          _size_on_disk(0),
+          _disk_io_stats()
     {}
     FieldIndexStats &memory_usage(const vespalib::MemoryUsage &usage) noexcept {
         _memory_usage = usage;
@@ -30,14 +34,19 @@ class FieldIndexStats
     }
     size_t size_on_disk() const noexcept { return _size_on_disk; }
 
+    FieldIndexStats& disk_io_stats(const DiskIoStats& stats) { _disk_io_stats = stats; return *this; }
+    const DiskIoStats& disk_io_stats() const noexcept { return _disk_io_stats; }
+
     void merge(const FieldIndexStats &rhs) noexcept {
         _memory_usage.merge(rhs._memory_usage);
         _size_on_disk += rhs._size_on_disk;
+        _disk_io_stats.merge(rhs._disk_io_stats);
     }
 
     bool operator==(const FieldIndexStats& rhs) const noexcept {
         return _memory_usage == rhs._memory_usage &&
-        _size_on_disk == rhs._size_on_disk;
+        _size_on_disk == rhs._size_on_disk &&
+        _disk_io_stats == rhs._disk_io_stats;
     }
 };