SERVER-64659 Sorter::File accepts optional SorterFileStats for tracking file open/close operations

(cherry picked from commit 176747132c3be0b0e73adebf55ba4acf6211934f)

3 files changed, 74 insertions, 6 deletions

diff --git a/src/mongo/db/sorter/sorter.cpp b/src/mongo/db/sorter/sorter.cpp
index 2f2dc44f1c8..3f532270092 100644
--- a/src/mongo/db/sorter/sorter.cpp
+++ b/src/mongo/db/sorter/sorter.cpp
@@ -158,17 +158,25 @@ public:
     FileIterator(const std::string& fileName,
                  std::streampos fileStartOffset,
                  std::streampos fileEndOffset,
-                 const Settings& settings)
+                 const Settings& settings,
+                 SorterFileStats* stats)
         : _settings(settings),
           _done(false),
           _fileName(fileName),
           _fileStartOffset(fileStartOffset),
-          _fileEndOffset(fileEndOffset) {
+          _fileEndOffset(fileEndOffset),
+          _stats(stats) {
         uassert(16815,
                 str::stream() << "unexpected empty file: " << _fileName,
                 boost::filesystem::file_size(_fileName) != 0);
     }
 
+    ~FileIterator() {
+        if (_stats && _file.is_open()) {
+            _stats->closed.addAndFetch(1);
+        }
+    }
+
     void openSource() {
         _file.open(_fileName.c_str(), std::ios::in | std::ios::binary);
         uassert(16814,
@@ -180,6 +188,10 @@ public:
                 str::stream() << "error seeking starting offset of '" << _fileStartOffset
                               << "' in file \"" << _fileName << "\": " << myErrnoWithDescription(),
                 _file.good());
+
+        if (_stats) {
+            _stats->opened.addAndFetch(1);
+        }
     }
 
     void closeSource() {
@@ -188,6 +200,10 @@ public:
                 str::stream() << "error closing file \"" << _fileName
                               << "\": " << myErrnoWithDescription(),
                 !_file.fail());
+
+        if (_stats) {
+            _stats->closed.addAndFetch(1);
+        }
     }
 
     bool more() {
@@ -313,6 +329,9 @@ private:
     std::streampos _fileStartOffset;  // File offset at which the sorted data range starts.
     std::streampos _fileEndOffset;    // File offset at which the sorted data range ends.
     std::ifstream _file;
+
+    // If set, this points to an external metrics holder for tracking file open/close activity.
+    SorterFileStats* _stats;
 };
 
 /**
@@ -895,7 +914,7 @@ SortedFileWriter<Key, Value>::SortedFileWriter(const SortOptions& opts,
                                                const std::string& fileName,
                                                const std::streampos fileStartOffset,
                                                const Settings& settings)
-    : _settings(settings) {
+    : _settings(settings), _stats(opts.sorterFileStats) {
 
     // This should be checked by consumers, but if we get here don't allow writes.
     uassert(
@@ -924,6 +943,17 @@ SortedFileWriter<Key, Value>::SortedFileWriter(const SortOptions& opts,
 
     // throw on failure
     _file.exceptions(std::ios::failbit | std::ios::badbit | std::ios::eofbit);
+
+    if (_stats) {
+        _stats->opened.addAndFetch(1);
+    }
+}
+
+template <typename Key, typename Value>
+SortedFileWriter<Key, Value>::~SortedFileWriter() {
+    if (_stats && _file.is_open()) {
+        _stats->closed.addAndFetch(1);
+    }
 }
 
 template <typename Key, typename Value>
@@ -999,8 +1029,12 @@ SortIteratorInterface<Key, Value>* SortedFileWriter<Key, Value>::done() {
     _fileEndOffset = currentFileOffset < _fileStartOffset ? _fileStartOffset : currentFileOffset;
     _file.close();
 
+    if (_stats) {
+        _stats->closed.addAndFetch(1);
+    }
+
     return new sorter::FileIterator<Key, Value>(
-        _fileName, _fileStartOffset, _fileEndOffset, _settings);
+        _fileName, _fileStartOffset, _fileEndOffset, _settings, _stats);
 }
 
 //
diff --git a/src/mongo/db/sorter/sorter.h b/src/mongo/db/sorter/sorter.h
index fccb3eef115..edd8a5611d1 100644
--- a/src/mongo/db/sorter/sorter.h
+++ b/src/mongo/db/sorter/sorter.h
@@ -37,6 +37,7 @@
 #include <vector>
 
 #include "mongo/bson/util/builder.h"
+#include "mongo/platform/atomic_word.h"
 
 /**
  * This is the public API for the Sorter (both in-memory and external)
@@ -86,6 +87,14 @@
 namespace mongo {
 
 /**
+ * For collecting file usage metrics.
+ */
+struct SorterFileStats {
+    AtomicWord<long long> opened;
+    AtomicWord<long long> closed;
+};
+
+/**
  * Runtime options that control the Sorter's behavior
  */
 struct SortOptions {
@@ -103,7 +112,14 @@ struct SortOptions {
     // extSortAllowed is true.
     std::string tempDir;
 
-    SortOptions() : limit(0), maxMemoryUsageBytes(64 * 1024 * 1024), extSortAllowed(false) {}
+    // If set, allows us to observe Sorter file handle usage.
+    SorterFileStats* sorterFileStats;
+
+    SortOptions()
+        : limit(0),
+          maxMemoryUsageBytes(64 * 1024 * 1024),
+          extSortAllowed(false),
+          sorterFileStats(nullptr) {}
 
     // Fluent API to support expressions like SortOptions().Limit(1000).ExtSortAllowed(true)
 
@@ -126,6 +142,11 @@ struct SortOptions {
         tempDir = newTempDir;
         return *this;
     }
+
+    SortOptions& FileStats(SorterFileStats* newSorterFileStats) {
+        sorterFileStats = newSorterFileStats;
+        return *this;
+    }
 };
 
 /**
@@ -234,6 +255,8 @@ public:
                               const std::streampos fileStartOffset,
                               const Settings& settings = Settings());
 
+    ~SortedFileWriter();
+
     void addAlreadySorted(const Key&, const Value&);
 
     /**
@@ -265,6 +288,9 @@ private:
     // for the next SortedFileWriter instance using the same file.
     std::streampos _fileStartOffset;
     std::streampos _fileEndOffset;
+
+    // If set, this points to an external metrics holder for tracking file open/close activity.
+    SorterFileStats* _stats;
 };
 }  // namespace mongo
 
diff --git a/src/mongo/db/sorter/sorter_test.cpp b/src/mongo/db/sorter/sorter_test.cpp
index 18fcf91f0b1..9654d69dd42 100644
--- a/src/mongo/db/sorter/sorter_test.cpp
+++ b/src/mongo/db/sorter/sorter_test.cpp
@@ -279,7 +279,8 @@ class SortedFileWriterAndFileIteratorTests : public ScopedGlobalServiceContextFo
 public:
     void run() {
         unittest::TempDir tempDir("sortedFileWriterTests");
-        const SortOptions opts = SortOptions().TempDir(tempDir.path());
+        SorterFileStats sorterFileStats;
+        const SortOptions opts = SortOptions().TempDir(tempDir.path()).FileStats(&sorterFileStats);
         {  // small
             std::string fileName = opts.tempDir + "/" + nextFileName();
             SortedFileWriter<IntWrapper, IntWrapper> sorter(opts, fileName, 0);
@@ -293,6 +294,10 @@ public:
 
             ASSERT_TRUE(boost::filesystem::remove(fileName));
         }
+
+        ASSERT_EQ(sorterFileStats.opened.load(), 2);
+        ASSERT_EQ(sorterFileStats.closed.load(), 2);
+
         {  // big
             std::string fileName = opts.tempDir + "/" + nextFileName();
             SortedFileWriter<IntWrapper, IntWrapper> sorter(opts, fileName, 0);
@@ -305,6 +310,9 @@ public:
             ASSERT_TRUE(boost::filesystem::remove(fileName));
         }
 
+        ASSERT_EQ(sorterFileStats.opened.load(), 4);
+        ASSERT_EQ(sorterFileStats.closed.load(), 4);
+
         ASSERT(boost::filesystem::is_empty(tempDir.path()));
     }
 };