@nxtedition/rocksdb 7.0.0-alpha.6 → 7.0.0

package/deps/rocksdb/rocksdb/db/db_impl/db_impl_write.cc

@@ -30,7 +30,7 @@ Status DBImpl::Put(const WriteOptions& o, ColumnFamilyHandle* column_family,
 
 Status DBImpl::Put(const WriteOptions& o, ColumnFamilyHandle* column_family,
                    const Slice& key, const Slice& ts, const Slice& val) {
-  const Status s = FailIfTsSizesMismatch(column_family, ts);
+  const Status s = FailIfTsMismatchCf(column_family, ts, /*ts_for_read=*/false);
   if (!s.ok()) {
     return s;
   }
@@ -63,7 +63,7 @@ Status DBImpl::Delete(const WriteOptions& write_options,
 Status DBImpl::Delete(const WriteOptions& write_options,
                       ColumnFamilyHandle* column_family, const Slice& key,
                       const Slice& ts) {
-  const Status s = FailIfTsSizesMismatch(column_family, ts);
+  const Status s = FailIfTsMismatchCf(column_family, ts, /*ts_for_read=*/false);
   if (!s.ok()) {
     return s;
   }
@@ -83,7 +83,7 @@ Status DBImpl::SingleDelete(const WriteOptions& write_options,
 Status DBImpl::SingleDelete(const WriteOptions& write_options,
                             ColumnFamilyHandle* column_family, const Slice& key,
                             const Slice& ts) {
-  const Status s = FailIfTsSizesMismatch(column_family, ts);
+  const Status s = FailIfTsMismatchCf(column_family, ts, /*ts_for_read=*/false);
   if (!s.ok()) {
     return s;
   }
@@ -126,7 +126,8 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
                          uint64_t* log_used, uint64_t log_ref,
                          bool disable_memtable, uint64_t* seq_used,
                          size_t batch_cnt,
-                         PreReleaseCallback* pre_release_callback) {
+                         PreReleaseCallback* pre_release_callback,
+                         PostMemTableCallback* post_memtable_callback) {
   assert(!seq_per_batch_ || batch_cnt != 0);
   if (my_batch == nullptr) {
     return Status::InvalidArgument("Batch is nullptr!");
@@ -185,6 +186,15 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
     return Status::NotSupported(
         "pipelined_writes is not compatible with unordered_write");
   }
+  if (immutable_db_options_.enable_pipelined_write &&
+      post_memtable_callback != nullptr) {
+    return Status::NotSupported(
+        "pipelined write currently does not honor post_memtable_callback");
+  }
+  if (seq_per_batch_ && post_memtable_callback != nullptr) {
+    return Status::NotSupported(
+        "seq_per_batch currently does not honor post_memtable_callback");
+  }
   // Otherwise IsLatestPersistentState optimization does not make sense
   assert(!WriteBatchInternal::IsLatestPersistentState(my_batch) ||
          disable_memtable);
@@ -241,7 +251,8 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
 
   PERF_TIMER_GUARD(write_pre_and_post_process_time);
   WriteThread::Writer w(write_options, my_batch, callback, log_ref,
-                        disable_memtable, batch_cnt, pre_release_callback);
+                        disable_memtable, batch_cnt, pre_release_callback,
+                        post_memtable_callback);
   StopWatch write_sw(immutable_db_options_.clock, stats_, DB_WRITE);
 
   write_thread_.JoinBatchGroup(&w);
@@ -268,6 +279,16 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
       // we're responsible for exit batch group
       // TODO(myabandeh): propagate status to write_group
       auto last_sequence = w.write_group->last_sequence;
+      for (auto* tmp_w : *(w.write_group)) {
+        assert(tmp_w);
+        if (tmp_w->post_memtable_callback) {
+          Status tmp_s =
+              (*tmp_w->post_memtable_callback)(last_sequence, disable_memtable);
+          // TODO: propagate the execution status of post_memtable_callback to
+          // caller.
+          assert(tmp_s.ok());
+        }
+      }
       versions_->SetLastSequence(last_sequence);
       MemTableInsertStatusCheck(w.status);
       write_thread_.ExitAsBatchGroupFollower(&w);
@@ -319,7 +340,11 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
 
     PERF_TIMER_START(write_pre_and_post_process_time);
   }
+
   log::Writer* log_writer = logs_.back().writer;
+  LogFileNumberSize& log_file_number_size = alive_log_files_.back();
+
+  assert(log_writer->get_log_number() == log_file_number_size.number);
 
   mutex_.Unlock();
 
@@ -419,7 +444,8 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
       if (status.ok() && !write_options.disableWAL) {
         PERF_TIMER_GUARD(write_wal_time);
         io_s = WriteToWAL(write_group, log_writer, log_used, need_log_sync,
-                          need_log_dir_sync, last_sequence + 1);
+                          need_log_dir_sync, last_sequence + 1,
+                          log_file_number_size);
       }
     } else {
       if (status.ok() && !write_options.disableWAL) {
@@ -545,6 +571,16 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
   }
   if (should_exit_batch_group) {
     if (status.ok()) {
+      for (auto* tmp_w : write_group) {
+        assert(tmp_w);
+        if (tmp_w->post_memtable_callback) {
+          Status tmp_s =
+              (*tmp_w->post_memtable_callback)(last_sequence, disable_memtable);
+          // TODO: propagate the execution status of post_memtable_callback to
+          // caller.
+          assert(tmp_s.ok());
+        }
+      }
       // Note: if we are to resume after non-OK statuses we need to revisit how
       // we reacts to non-OK statuses here.
       versions_->SetLastSequence(last_sequence);
@@ -586,6 +622,10 @@ Status DBImpl::PipelinedWriteImpl(const WriteOptions& write_options,
     w.status = PreprocessWrite(write_options, &need_log_sync, &write_context);
     PERF_TIMER_START(write_pre_and_post_process_time);
     log::Writer* log_writer = logs_.back().writer;
+    LogFileNumberSize& log_file_number_size = alive_log_files_.back();
+
+    assert(log_writer->get_log_number() == log_file_number_size.number);
+
     mutex_.Unlock();
 
     // This can set non-OK status if callback fail.
@@ -649,8 +689,9 @@ Status DBImpl::PipelinedWriteImpl(const WriteOptions& write_options,
                           wal_write_group.size - 1);
         RecordTick(stats_, WRITE_DONE_BY_OTHER, wal_write_group.size - 1);
       }
-      io_s = WriteToWAL(wal_write_group, log_writer, log_used, need_log_sync,
-                        need_log_dir_sync, current_sequence);
+      io_s =
+          WriteToWAL(wal_write_group, log_writer, log_used, need_log_sync,
+                     need_log_dir_sync, current_sequence, log_file_number_size);
       w.status = io_s;
     }
 
@@ -1041,12 +1082,18 @@ Status DBImpl::PreprocessWrite(const WriteOptions& write_options,
 
   PERF_TIMER_GUARD(write_scheduling_flushes_compactions_time);
 
-  assert(!single_column_family_mode_ ||
-         versions_->GetColumnFamilySet()->NumberOfColumnFamilies() == 1);
-  if (UNLIKELY(status.ok() && !single_column_family_mode_ &&
-               total_log_size_ > GetMaxTotalWalSize())) {
-    WaitForPendingWrites();
-    status = SwitchWAL(write_context);
+  if (UNLIKELY(status.ok() && total_log_size_ > GetMaxTotalWalSize())) {
+    assert(versions_);
+    const ColumnFamilySet* const column_families =
+        versions_->GetColumnFamilySet();
+    assert(column_families);
+    size_t num_cfs = column_families->NumberOfColumnFamilies();
+
+    assert(num_cfs >= 1);
+    if (num_cfs > 1) {
+      WaitForPendingWrites();
+      status = SwitchWAL(write_context);
+    }
   }
 
   if (UNLIKELY(status.ok() && write_buffer_manager_->ShouldFlush())) {
@@ -1172,17 +1219,9 @@ IOStatus DBImpl::WriteToWAL(const WriteBatch& merged_batch,
                             log::Writer* log_writer, uint64_t* log_used,
                             uint64_t* log_size,
                             Env::IOPriority rate_limiter_priority,
-                            bool with_db_mutex, bool with_log_mutex) {
+                            LogFileNumberSize& log_file_number_size) {
   assert(log_size != nullptr);
 
-  // Assert mutex explicitly.
-  if (with_db_mutex) {
-    mutex_.AssertHeld();
-  } else if (two_write_queues_) {
-    log_write_mutex_.AssertHeld();
-    assert(with_log_mutex);
-  }
-
   Slice log_entry = WriteBatchInternal::Contents(&merged_batch);
   *log_size = log_entry.size();
   // When two_write_queues_ WriteToWAL has to be protected from concurretn calls
@@ -1205,12 +1244,7 @@ IOStatus DBImpl::WriteToWAL(const WriteBatch& merged_batch,
     *log_used = logfile_number_;
   }
   total_log_size_ += log_entry.size();
-  if (with_db_mutex || with_log_mutex) {
-    assert(alive_log_files_tail_ == alive_log_files_.rbegin());
-    assert(alive_log_files_tail_ != alive_log_files_.rend());
-  }
-  LogFileNumberSize& last_alive_log = *alive_log_files_tail_;
-  last_alive_log.AddSize(*log_size);
+  log_file_number_size.AddSize(*log_size);
   log_empty_ = false;
   return io_s;
 }
@@ -1218,7 +1252,8 @@ IOStatus DBImpl::WriteToWAL(const WriteBatch& merged_batch,
 IOStatus DBImpl::WriteToWAL(const WriteThread::WriteGroup& write_group,
                             log::Writer* log_writer, uint64_t* log_used,
                             bool need_log_sync, bool need_log_dir_sync,
-                            SequenceNumber sequence) {
+                            SequenceNumber sequence,
+                            LogFileNumberSize& log_file_number_size) {
   IOStatus io_s;
   assert(!two_write_queues_);
   assert(!write_group.leader->disable_wal);
@@ -1239,7 +1274,8 @@ IOStatus DBImpl::WriteToWAL(const WriteThread::WriteGroup& write_group,
 
   uint64_t log_size;
   io_s = WriteToWAL(*merged_batch, log_writer, log_used, &log_size,
-                    write_group.leader->rate_limiter_priority);
+                    write_group.leader->rate_limiter_priority,
+                    log_file_number_size);
   if (to_be_cached_state) {
     cached_recoverable_state_ = *to_be_cached_state;
     cached_recoverable_state_empty_ = false;
@@ -1333,10 +1369,14 @@ IOStatus DBImpl::ConcurrentWriteToWAL(
   WriteBatchInternal::SetSequence(merged_batch, sequence);
 
   log::Writer* log_writer = logs_.back().writer;
+  LogFileNumberSize& log_file_number_size = alive_log_files_.back();
+
+  assert(log_writer->get_log_number() == log_file_number_size.number);
+
   uint64_t log_size;
   io_s = WriteToWAL(*merged_batch, log_writer, log_used, &log_size,
                     write_group.leader->rate_limiter_priority,
-                    /*with_db_mutex=*/false, /*with_log_mutex=*/true);
+                    log_file_number_size);
   if (to_be_cached_state) {
     cached_recoverable_state_ = *to_be_cached_state;
     cached_recoverable_state_empty_ = false;
@@ -1992,7 +2032,6 @@ Status DBImpl::SwitchMemtable(ColumnFamilyData* cfd, WriteContext* context) {
       log_dir_synced_ = false;
       logs_.emplace_back(logfile_number_, new_log);
       alive_log_files_.push_back(LogFileNumberSize(logfile_number_));
-      alive_log_files_tail_ = alive_log_files_.rbegin();
     }
     log_write_mutex_.Unlock();
   }