Add options.compaction_measure_io_stats to print write I/O stats in compactions

Summary:
Add options.compaction_measure_io_stats to print out / pass to listener accumulated time spent on write calls. Example outputs in info logs:

2015/08/12-16:27:59.463944 7fd428bff700 (Original Log Time 2015/08/12-16:27:59.463922) EVENT_LOG_v1 {"time_micros": 1439422079463897, "job": 6, "event": "compaction_finished", "output_level": 1, "num_output_files": 4, "total_output_size": 6900525, "num_input_records": 111483, "num_output_records": 106877, "file_write_nanos": 15663206, "file_range_sync_nanos": 649588, "file_fsync_nanos": 349614797, "file_prepare_write_nanos": 1505812, "lsm_state": [2, 4, 0, 0, 0, 0, 0]}

Add two more counters in iostats_context.

Also add a parameter of db_bench.

Test Plan: Add a unit test. Also manually verify LOG outputs in db_bench

Subscribers: leveldb, dhruba

Differential Revision: https://reviews.facebook.net/D44115

db/compaction_job.cc

@@ -96,7 +96,7 @@ CompactionJob::CompactionJob(
     Directory* db_directory, Directory* output_directory, Statistics* stats,
     std::vector<SequenceNumber> existing_snapshots,
     std::shared_ptr<Cache> table_cache, EventLogger* event_logger,
-    bool paranoid_file_checks, const std::string& dbname,
+    bool paranoid_file_checks, bool measure_io_stats, const std::string& dbname,
     CompactionJobStats* compaction_job_stats)
     : job_id_(job_id),
       compact_(new CompactionState(compaction)),
@@ -115,7 +115,8 @@ CompactionJob::CompactionJob(
       existing_snapshots_(std::move(existing_snapshots)),
       table_cache_(std::move(table_cache)),
       event_logger_(event_logger),
-      paranoid_file_checks_(paranoid_file_checks) {
+      paranoid_file_checks_(paranoid_file_checks),
+      measure_io_stats_(measure_io_stats) {
   assert(log_buffer_ != nullptr);
   ThreadStatusUtil::SetColumnFamily(compact_->compaction->column_family_data());
   ThreadStatusUtil::SetThreadOperation(ThreadStatus::OP_COMPACTION);
@@ -270,6 +271,22 @@ Status CompactionJob::Run() {
 }
 
 Status CompactionJob::SubCompactionRun(Slice* start, Slice* end) {
+  PerfLevel prev_perf_level = PerfLevel::kEnableTime;
+  uint64_t prev_write_nanos = 0;
+  uint64_t prev_fsync_nanos = 0;
+  uint64_t prev_range_sync_nanos = 0;
+  uint64_t prev_prepare_write_nanos = 0;
+  bool enabled_io_stats = false;
+  if (measure_io_stats_ && compaction_job_stats_ != nullptr) {
+    prev_perf_level = GetPerfLevel();
+    SetPerfLevel(PerfLevel::kEnableTime);
+    prev_write_nanos = iostats_context.write_nanos;
+    prev_fsync_nanos = iostats_context.fsync_nanos;
+    prev_range_sync_nanos = iostats_context.range_sync_nanos;
+    prev_prepare_write_nanos = iostats_context.prepare_write_nanos;
+    enabled_io_stats = true;
+  }
+
   auto* compaction = compact_->compaction;
   const uint64_t start_micros = env_->NowMicros();
   int64_t imm_micros = 0;  // Micros spent doing imm_ compactions
@@ -286,6 +303,21 @@ Status CompactionJob::SubCompactionRun(Slice* start, Slice* end) {
 
   compaction_stats_.micros = env_->NowMicros() - start_micros - imm_micros;
   MeasureTime(stats_, COMPACTION_TIME, compaction_stats_.micros);
+
+  if (enabled_io_stats) {
+    // Not supporting parallel running yet
+    compaction_job_stats_->file_write_nanos +=
+        iostats_context.write_nanos - prev_write_nanos;
+    compaction_job_stats_->file_fsync_nanos +=
+        iostats_context.fsync_nanos - prev_fsync_nanos;
+    compaction_job_stats_->file_range_sync_nanos +=
+        iostats_context.range_sync_nanos - prev_range_sync_nanos;
+    compaction_job_stats_->file_prepare_write_nanos +=
+        iostats_context.prepare_write_nanos - prev_prepare_write_nanos;
+    if (prev_perf_level != PerfLevel::kEnableTime) {
+      SetPerfLevel(prev_perf_level);
+    }
+  }
   return status;
 }
 
@@ -340,6 +372,16 @@ void CompactionJob::Install(Status* status,
          << "total_output_size" << compact_->total_bytes
          << "num_input_records" << compact_->num_input_records
          << "num_output_records" << compact_->num_output_records;
+
+  if (measure_io_stats_ && compaction_job_stats_ != nullptr) {
+    stream << "file_write_nanos" << compaction_job_stats_->file_write_nanos;
+    stream << "file_range_sync_nanos"
+           << compaction_job_stats_->file_range_sync_nanos;
+    stream << "file_fsync_nanos" << compaction_job_stats_->file_fsync_nanos;
+    stream << "file_prepare_write_nanos"
+           << compaction_job_stats_->file_prepare_write_nanos;
+  }
+
   stream << "lsm_state";
   stream.StartArray();
   for (int level = 0; level < vstorage->num_levels(); ++level) {

db/compaction_job.h

@@ -57,8 +57,8 @@ class CompactionJob {
                 Directory* db_directory, Directory* output_directory,
                 Statistics* stats,
                 std::vector<SequenceNumber> existing_snapshots,
-                std::shared_ptr<Cache> table_cache,
-                EventLogger* event_logger, bool paranoid_file_checks,
+                std::shared_ptr<Cache> table_cache, EventLogger* event_logger,
+                bool paranoid_file_checks, bool measure_io_stats,
                 const std::string& dbname,
                 CompactionJobStats* compaction_job_stats);
 
@@ -153,6 +153,7 @@ class CompactionJob {
   EventLogger* event_logger_;
 
   bool paranoid_file_checks_;
+  bool measure_io_stats_;
   std::vector<Slice> sub_compaction_boundaries_;
 };
 

db/compaction_job_stats_test.cc

@@ -417,14 +417,25 @@ class CompactionJobStatsTest : public testing::Test,
 // test CompactionJobStatsTest.
 class CompactionJobStatsChecker : public EventListener {
  public:
-  CompactionJobStatsChecker() : compression_enabled_(false) {}
+  CompactionJobStatsChecker()
+      : compression_enabled_(false), verify_next_comp_io_stats_(false) {}
 
   size_t NumberOfUnverifiedStats() { return expected_stats_.size(); }
 
+  void set_verify_next_comp_io_stats(bool v) { verify_next_comp_io_stats_ = v; }
+
   // Once a compaction completed, this function will verify the returned
   // CompactionJobInfo with the oldest CompactionJobInfo added earlier
   // in "expected_stats_" which has not yet being used for verification.
   virtual void OnCompactionCompleted(DB *db, const CompactionJobInfo& ci) {
+    if (verify_next_comp_io_stats_) {
+      ASSERT_GT(ci.stats.file_write_nanos, 0);
+      ASSERT_GT(ci.stats.file_range_sync_nanos, 0);
+      ASSERT_GT(ci.stats.file_fsync_nanos, 0);
+      ASSERT_GT(ci.stats.file_prepare_write_nanos, 0);
+      verify_next_comp_io_stats_ = false;
+    }
+
     std::lock_guard<std::mutex> lock(mutex_);
     if (expected_stats_.size()) {
       Verify(ci.stats, expected_stats_.front());
@@ -498,10 +509,13 @@ class CompactionJobStatsChecker : public EventListener {
     compression_enabled_ = flag;
   }
 
+  bool verify_next_comp_io_stats() const { return verify_next_comp_io_stats_; }
+
  private:
   std::mutex mutex_;
   std::queue<CompactionJobStats> expected_stats_;
   bool compression_enabled_;
+  bool verify_next_comp_io_stats_;
 };
 
 // An EventListener which helps verify the compaction statistics in
@@ -643,7 +657,9 @@ TEST_P(CompactionJobStatsTest, CompactionJobStatsTest) {
   options.num_levels = 3;
   options.compression = kNoCompression;
   options.num_subcompactions = num_subcompactions_;
+  options.bytes_per_sync = 512 * 1024;
 
+  options.compaction_measure_io_stats = true;
   for (int test = 0; test < 2; ++test) {
     DestroyAndReopen(options);
     CreateAndReopenWithCF({"pikachu"}, options);
@@ -766,6 +782,7 @@ TEST_P(CompactionJobStatsTest, CompactionJobStatsTest) {
             num_keys_per_L0_file));
     ASSERT_EQ(stats_checker->NumberOfUnverifiedStats(), 1U);
     Compact(1, smallest_key, largest_key);
+
     num_L1_files = options.num_subcompactions > 1 ? 7 : 4;
     char L1_buf[4];
     snprintf(L1_buf, sizeof(L1_buf), "0,%d", num_L1_files);
@@ -777,6 +794,61 @@ TEST_P(CompactionJobStatsTest, CompactionJobStatsTest) {
     }
     stats_checker->EnableCompression(true);
     compression_ratio = kCompressionRatio;
+
+    for (int i = 0; i < 5; i++) {
+      ASSERT_OK(Put(1, Slice(Key(key_base + i, 10)),
+                    Slice(RandomString(&rnd, 512 * 1024, 1))));
+    }
+
+    ASSERT_OK(Flush(1));
+    reinterpret_cast<DBImpl*>(db_)->TEST_WaitForCompact();
+
+    stats_checker->set_verify_next_comp_io_stats(true);
+    std::atomic<bool> first_prepare_write(true);
+    rocksdb::SyncPoint::GetInstance()->SetCallBack(
+        "WritableFileWriter::Append:BeforePrepareWrite", [&](void* arg) {
+          if (first_prepare_write.load()) {
+            options.env->SleepForMicroseconds(3);
+            first_prepare_write.store(false);
+          }
+        });
+
+    std::atomic<bool> first_flush(true);
+    rocksdb::SyncPoint::GetInstance()->SetCallBack(
+        "WritableFileWriter::Flush:BeforeAppend", [&](void* arg) {
+          if (first_flush.load()) {
+            options.env->SleepForMicroseconds(3);
+            first_flush.store(false);
+          }
+        });
+
+    std::atomic<bool> first_sync(true);
+    rocksdb::SyncPoint::GetInstance()->SetCallBack(
+        "WritableFileWriter::SyncInternal:0", [&](void* arg) {
+          if (first_sync.load()) {
+            options.env->SleepForMicroseconds(3);
+            first_sync.store(false);
+          }
+        });
+
+    std::atomic<bool> first_range_sync(true);
+    rocksdb::SyncPoint::GetInstance()->SetCallBack(
+        "WritableFileWriter::RangeSync:0", [&](void* arg) {
+          if (first_range_sync.load()) {
+            options.env->SleepForMicroseconds(3);
+            first_range_sync.store(false);
+          }
+        });
+    rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+
+    Compact(1, smallest_key, largest_key);
+
+    ASSERT_TRUE(!stats_checker->verify_next_comp_io_stats());
+    ASSERT_TRUE(!first_prepare_write.load());
+    ASSERT_TRUE(!first_flush.load());
+    ASSERT_TRUE(!first_sync.load());
+    ASSERT_TRUE(!first_range_sync.load());
+    rocksdb::SyncPoint::GetInstance()->DisableProcessing();
   }
   ASSERT_EQ(stats_checker->NumberOfUnverifiedStats(), 0U);
 }

db/compaction_job_test.cc

@@ -246,10 +246,11 @@ class CompactionJobTest : public testing::Test {
     LogBuffer log_buffer(InfoLogLevel::INFO_LEVEL, db_options_.info_log.get());
     mutex_.Lock();
     EventLogger event_logger(db_options_.info_log.get());
-    CompactionJob compaction_job(
-        0, &compaction, db_options_, env_options_, versions_.get(),
-        &shutting_down_, &log_buffer, nullptr, nullptr, nullptr, {},
-        table_cache_, &event_logger, false, dbname_, &compaction_job_stats_);
+    CompactionJob compaction_job(0, &compaction, db_options_, env_options_,
+                                 versions_.get(), &shutting_down_, &log_buffer,
+                                 nullptr, nullptr, nullptr, {}, table_cache_,
+                                 &event_logger, false, false, dbname_,
+                                 &compaction_job_stats_);
 
     VerifyInitializationOfCompactionJobStats(compaction_job_stats_);
 

db/db_bench.cc

@@ -465,6 +465,9 @@ DEFINE_int32(transaction_sleep, 0,
              "Max microseconds to sleep in between "
              "reading and writing a value (used in RandomTransaction only). ");
 
+DEFINE_bool(compaction_measure_io_stats, false,
+            "Measure times spents on I/Os while in compactions. ");
+
 namespace {
 enum rocksdb::CompressionType StringToCompressionType(const char* ctype) {
   assert(ctype);
@@ -2379,6 +2382,7 @@ class Benchmark {
       exit(1);
     }
     options.max_successive_merges = FLAGS_max_successive_merges;
+    options.compaction_measure_io_stats = FLAGS_compaction_measure_io_stats;
 
     // set universal style compaction configurations, if applicable
     if (FLAGS_universal_size_ratio != 0) {

db/db_impl.cc

@@ -1669,7 +1669,8 @@ Status DBImpl::CompactFilesImpl(
       &shutting_down_, log_buffer, directories_.GetDbDir(),
       directories_.GetDataDir(c->output_path_id()), stats_, snapshots_.GetAll(),
       table_cache_, &event_logger_,
-      c->mutable_cf_options()->paranoid_file_checks, dbname_,
+      c->mutable_cf_options()->paranoid_file_checks,
+      c->mutable_cf_options()->compaction_measure_io_stats, dbname_,
       nullptr);  // Here we pass a nullptr for CompactionJobStats because
                  // CompactFiles does not trigger OnCompactionCompleted(),
                  // which is the only place where CompactionJobStats is
@@ -2683,7 +2684,8 @@ Status DBImpl::BackgroundCompaction(bool* madeProgress, JobContext* job_context,
         versions_.get(), &shutting_down_, log_buffer, directories_.GetDbDir(),
         directories_.GetDataDir(c->output_path_id()), stats_,
         snapshots_.GetAll(), table_cache_, &event_logger_,
-        c->mutable_cf_options()->paranoid_file_checks, dbname_,
+        c->mutable_cf_options()->paranoid_file_checks,
+        c->mutable_cf_options()->compaction_measure_io_stats, dbname_,
         &compaction_job_stats);
     compaction_job.Prepare();
 

include/rocksdb/compaction_job_stats.h

@@ -58,6 +58,21 @@ struct CompactionJobStats {
   // the key) encountered and written out.
   uint64_t num_corrupt_keys;
 
+  // Following counters are only populated if
+  // options.compaction_measure_io_stats = true;
+
+  // Time spent on file's Append() call.
+  uint64_t file_write_nanos;
+
+  // Time spent on sync file range.
+  uint64_t file_range_sync_nanos;
+
+  // Time spent on file fsync.
+  uint64_t file_fsync_nanos;
+
+  // Time spent on preparing file write (falocate, etc)
+  uint64_t file_prepare_write_nanos;
+
   // 0-terminated strings storing the first 8 bytes of the smallest and
   // largest key in the output.
   static const size_t kMaxPrefixLength = 8;

include/rocksdb/iostats_context.h

@@ -38,7 +38,10 @@ struct IOStatsContext {
   uint64_t read_nanos;
   // time spent in sync_file_range().
   uint64_t range_sync_nanos;
-
+  // time spent in fsync
+  uint64_t fsync_nanos;
+  // time spent in preparing write (fallocate etc).
+  uint64_t prepare_write_nanos;
   // time spent in Logger::Logv().
   uint64_t logger_nanos;
 };

include/rocksdb/options.h

@@ -15,7 +15,6 @@
 #include <memory>
 #include <vector>
 #include <limits>
-#include <stdint.h>
 #include <unordered_map>
 
 #include "rocksdb/version.h"
@@ -736,6 +735,10 @@ struct ColumnFamilyOptions {
   // Default: false
   bool paranoid_file_checks;
 
+  // Measure IO stats in compactions, if true.
+  // Default: false
+  bool compaction_measure_io_stats;
+
   // Create ColumnFamilyOptions with default values for all fields
   ColumnFamilyOptions();
   // Create ColumnFamilyOptions from Options

util/compaction_job_stats_impl.cc

@@ -33,6 +33,11 @@ void CompactionJobStats::Reset() {
   num_expired_deletion_records = 0;
 
   num_corrupt_keys = 0;
+
+  file_write_nanos = 0;
+  file_range_sync_nanos = 0;
+  file_fsync_nanos = 0;
+  file_prepare_write_nanos = 0;
 }
 
 #else

util/file_reader_writer.cc

@@ -43,7 +43,11 @@ Status WritableFileWriter::Append(const Slice& data) {
 
   TEST_KILL_RANDOM(rocksdb_kill_odds * REDUCE_ODDS2);
 
+  {
+    IOSTATS_TIMER_GUARD(prepare_write_nanos);
+    TEST_SYNC_POINT("WritableFileWriter::Append:BeforePrepareWrite");
     writable_file_->PrepareWrite(static_cast<size_t>(GetFileSize()), left);
+  }
   // if there is no space in the cache, then flush
   if (cursize_ + left > capacity_) {
     s = Flush();
@@ -105,6 +109,7 @@ Status WritableFileWriter::Flush() {
     size_t size = RequestToken(left);
     {
       IOSTATS_TIMER_GUARD(write_nanos);
+      TEST_SYNC_POINT("WritableFileWriter::Flush:BeforeAppend");
       Status s = writable_file_->Append(Slice(src, size));
       if (!s.ok()) {
         return s;
@@ -182,6 +187,8 @@ Status WritableFileWriter::SyncWithoutFlush(bool use_fsync) {
 
 Status WritableFileWriter::SyncInternal(bool use_fsync) {
   Status s;
+  IOSTATS_TIMER_GUARD(fsync_nanos);
+  TEST_SYNC_POINT("WritableFileWriter::SyncInternal:0");
   if (use_fsync) {
     s = writable_file_->Fsync();
   } else {
@@ -192,6 +199,7 @@ Status WritableFileWriter::SyncInternal(bool use_fsync) {
 
 Status WritableFileWriter::RangeSync(off_t offset, off_t nbytes) {
   IOSTATS_TIMER_GUARD(range_sync_nanos);
+  TEST_SYNC_POINT("WritableFileWriter::RangeSync:0");
   return writable_file_->RangeSync(offset, nbytes);
 }
 

util/iostats_context.cc

@@ -26,6 +26,8 @@ void IOStatsContext::Reset() {
   write_nanos = 0;
   read_nanos = 0;
   range_sync_nanos = 0;
+  prepare_write_nanos = 0;
+  fsync_nanos = 0;
   logger_nanos = 0;
 }
 
@@ -33,15 +35,10 @@ void IOStatsContext::Reset() {
 
 std::string IOStatsContext::ToString() const {
   std::ostringstream ss;
-  ss << OUTPUT(thread_pool_id)
-     << OUTPUT(bytes_read)
-     << OUTPUT(bytes_written)
-     << OUTPUT(open_nanos)
-     << OUTPUT(allocate_nanos)
-     << OUTPUT(write_nanos)
-     << OUTPUT(read_nanos)
-     << OUTPUT(range_sync_nanos)
-     << OUTPUT(logger_nanos);
+  ss << OUTPUT(thread_pool_id) << OUTPUT(bytes_read) << OUTPUT(bytes_written)
+     << OUTPUT(open_nanos) << OUTPUT(allocate_nanos) << OUTPUT(write_nanos)
+     << OUTPUT(read_nanos) << OUTPUT(range_sync_nanos) << OUTPUT(fsync_nanos)
+     << OUTPUT(prepare_write_nanos) << OUTPUT(logger_nanos);
 
   return ss.str();
 }

util/mutable_cf_options.h

@@ -43,7 +43,9 @@ struct MutableCFOptions {
         num_subcompactions(options.num_subcompactions),
         max_sequential_skip_in_iterations(
             options.max_sequential_skip_in_iterations),
-      paranoid_file_checks(options.paranoid_file_checks)
+        paranoid_file_checks(options.paranoid_file_checks),
+        compaction_measure_io_stats(options.compaction_measure_io_stats)
+
   {
     RefreshDerivedOptions(ioptions);
   }
@@ -73,8 +75,8 @@ struct MutableCFOptions {
         verify_checksums_in_compaction(false),
         num_subcompactions(1),
         max_sequential_skip_in_iterations(0),
-      paranoid_file_checks(false)
-  {}
+        paranoid_file_checks(false),
+        compaction_measure_io_stats(false) {}
 
   // Must be called after any change to MutableCFOptions
   void RefreshDerivedOptions(const ImmutableCFOptions& ioptions);
@@ -128,6 +130,7 @@ struct MutableCFOptions {
   // Misc options
   uint64_t max_sequential_skip_in_iterations;
   bool paranoid_file_checks;
+  bool compaction_measure_io_stats;
 
   // Derived options
   // Per-level target file size.

util/options.cc

@@ -121,7 +121,8 @@ ColumnFamilyOptions::ColumnFamilyOptions()
       max_successive_merges(0),
       min_partial_merge_operands(2),
       optimize_filters_for_hits(false),
-      paranoid_file_checks(false) {
+      paranoid_file_checks(false),
+      compaction_measure_io_stats(false) {
   assert(memtable_factory.get() != nullptr);
 }
 
@@ -185,7 +186,8 @@ ColumnFamilyOptions::ColumnFamilyOptions(const Options& options)
       max_successive_merges(options.max_successive_merges),
       min_partial_merge_operands(options.min_partial_merge_operands),
       optimize_filters_for_hits(options.optimize_filters_for_hits),
-      paranoid_file_checks(options.paranoid_file_checks) {
+      paranoid_file_checks(options.paranoid_file_checks),
+      compaction_measure_io_stats(options.compaction_measure_io_stats) {
   assert(memtable_factory.get() != nullptr);
   if (max_bytes_for_level_multiplier_additional.size() <
       static_cast<unsigned int>(num_levels)) {
@@ -514,6 +516,10 @@ void ColumnFamilyOptions::Dump(Logger* log) const {
          max_successive_merges);
     Warn(log, "               Options.optimize_fllters_for_hits: %d",
         optimize_filters_for_hits);
+    Warn(log, "               Options.paranoid_file_checks: %d",
+         paranoid_file_checks);
+    Warn(log, "               Options.compaction_measure_io_stats: %d",
+         compaction_measure_io_stats);
 }  // ColumnFamilyOptions::Dump
 
 void Options::Dump(Logger* log) const {

util/options_helper.cc

@@ -2,6 +2,7 @@
 //  This source code is licensed under the BSD-style license found in the
 //  LICENSE file in the root directory of this source tree. An additional grant
 //  of patent rights can be found in the PATENTS file in the same directory.
+#include "util/options_helper.h"
 
 #include <cassert>
 #include <cctype>
@@ -16,7 +17,6 @@
 #include "rocksdb/table.h"
 #include "table/block_based_table_factory.h"
 #include "util/logging.h"
-#include "util/options_helper.h"
 
 namespace rocksdb {
 
@@ -449,6 +449,8 @@ bool ParseColumnFamilyOption(const std::string& name, const std::string& value,
       new_options->min_partial_merge_operands = ParseUint32(value);
     } else if (name == "inplace_update_support") {
       new_options->inplace_update_support = ParseBoolean(name, value);
+    } else if (name == "compaction_measure_io_stats") {
+      new_options->compaction_measure_io_stats = ParseBoolean(name, value);
     } else if (name == "prefix_extractor") {
       const std::string kFixedPrefixName = "fixed:";
       const std::string kCappedPrefixName = "capped:";

util/options_test.cc

@@ -129,6 +129,7 @@ TEST_F(OptionsTest, GetOptionsFromMapTest) {
       {"filter_deletes", "0"},
       {"max_sequential_skip_in_iterations", "24"},
       {"inplace_update_support", "true"},
+      {"compaction_measure_io_stats", "true"},
       {"inplace_update_num_locks", "25"},
       {"memtable_prefix_bloom_bits", "26"},
       {"memtable_prefix_bloom_probes", "27"},