@nxtedition/rocksdb 7.1.20 → 7.1.21

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

@@ -1055,10 +1055,10 @@ Status DBImpl::CompactRangeInternal(const CompactRangeOptions& options,
       SelectColumnFamiliesForAtomicFlush(&cfds);
       mutex_.Unlock();
       s = AtomicFlushMemTables(cfds, fo, FlushReason::kManualCompaction,
-                               false /* writes_stopped */);
+                               false /* entered_write_thread */);
     } else {
       s = FlushMemTable(cfd, fo, FlushReason::kManualCompaction,
-                        false /* writes_stopped*/);
+                        false /* entered_write_thread */);
     }
     if (!s.ok()) {
       LogFlush(immutable_db_options_.info_log);
@@ -2016,9 +2016,16 @@ void DBImpl::GenerateFlushRequest(const autovector<ColumnFamilyData*>& cfds,
 
 Status DBImpl::FlushMemTable(ColumnFamilyData* cfd,
                              const FlushOptions& flush_options,
-                             FlushReason flush_reason, bool writes_stopped) {
+                             FlushReason flush_reason,
+                             bool entered_write_thread) {
   // This method should not be called if atomic_flush is true.
   assert(!immutable_db_options_.atomic_flush);
+  if (!flush_options.wait && write_controller_.IsStopped()) {
+    std::ostringstream oss;
+    oss << "Writes have been stopped, thus unable to perform manual flush. "
+           "Please try again later after writes are resumed";
+    return Status::TryAgain(oss.str());
+  }
   Status s;
   if (!flush_options.allow_write_stall) {
     bool flush_needed = true;
@@ -2029,6 +2036,7 @@ Status DBImpl::FlushMemTable(ColumnFamilyData* cfd,
     }
   }
 
+  const bool needs_to_join_write_thread = !entered_write_thread;
   autovector<FlushRequest> flush_reqs;
   autovector<uint64_t> memtable_ids_to_wait;
   {
@@ -2037,7 +2045,7 @@ Status DBImpl::FlushMemTable(ColumnFamilyData* cfd,
 
     WriteThread::Writer w;
     WriteThread::Writer nonmem_w;
-    if (!writes_stopped) {
+    if (needs_to_join_write_thread) {
       write_thread_.EnterUnbatched(&w, &mutex_);
       if (two_write_queues_) {
         nonmem_write_thread_.EnterUnbatched(&nonmem_w, &mutex_);
@@ -2120,7 +2128,7 @@ Status DBImpl::FlushMemTable(ColumnFamilyData* cfd,
       MaybeScheduleFlushOrCompaction();
     }
 
-    if (!writes_stopped) {
+    if (needs_to_join_write_thread) {
       write_thread_.ExitUnbatched(&w);
       if (two_write_queues_) {
         nonmem_write_thread_.ExitUnbatched(&nonmem_w);
@@ -2156,7 +2164,14 @@ Status DBImpl::FlushMemTable(ColumnFamilyData* cfd,
 Status DBImpl::AtomicFlushMemTables(
     const autovector<ColumnFamilyData*>& column_family_datas,
     const FlushOptions& flush_options, FlushReason flush_reason,
-    bool writes_stopped) {
+    bool entered_write_thread) {
+  assert(immutable_db_options_.atomic_flush);
+  if (!flush_options.wait && write_controller_.IsStopped()) {
+    std::ostringstream oss;
+    oss << "Writes have been stopped, thus unable to perform manual flush. "
+           "Please try again later after writes are resumed";
+    return Status::TryAgain(oss.str());
+  }
   Status s;
   if (!flush_options.allow_write_stall) {
     int num_cfs_to_flush = 0;
@@ -2173,6 +2188,7 @@ Status DBImpl::AtomicFlushMemTables(
       return s;
     }
   }
+  const bool needs_to_join_write_thread = !entered_write_thread;
   FlushRequest flush_req;
   autovector<ColumnFamilyData*> cfds;
   {
@@ -2181,7 +2197,7 @@ Status DBImpl::AtomicFlushMemTables(
 
     WriteThread::Writer w;
     WriteThread::Writer nonmem_w;
-    if (!writes_stopped) {
+    if (needs_to_join_write_thread) {
       write_thread_.EnterUnbatched(&w, &mutex_);
       if (two_write_queues_) {
         nonmem_write_thread_.EnterUnbatched(&nonmem_w, &mutex_);
@@ -2229,7 +2245,7 @@ Status DBImpl::AtomicFlushMemTables(
       MaybeScheduleFlushOrCompaction();
     }
 
-    if (!writes_stopped) {
+    if (needs_to_join_write_thread) {
       write_thread_.ExitUnbatched(&w);
       if (two_write_queues_) {
         nonmem_write_thread_.ExitUnbatched(&nonmem_w);
@@ -3263,7 +3279,6 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
     TEST_SYNC_POINT_CALLBACK("DBImpl::BackgroundCompaction:BeforeCompaction",
                              c->column_family_data());
     assert(c->num_input_files(1) == 0);
-    assert(c->level() == 0);
     assert(c->column_family_data()->ioptions()->compaction_style ==
            kCompactionStyleFIFO);
 
@@ -3756,9 +3771,11 @@ void DBImpl::InstallSuperVersionAndScheduleWork(
   // triggered soon anyway.
   bottommost_files_mark_threshold_ = kMaxSequenceNumber;
   for (auto* my_cfd : *versions_->GetColumnFamilySet()) {
-    bottommost_files_mark_threshold_ = std::min(
-        bottommost_files_mark_threshold_,
-        my_cfd->current()->storage_info()->bottommost_files_mark_threshold());
+    if (!my_cfd->ioptions()->allow_ingest_behind) {
+      bottommost_files_mark_threshold_ = std::min(
+          bottommost_files_mark_threshold_,
+          my_cfd->current()->storage_info()->bottommost_files_mark_threshold());
+    }
   }
 
   // Whenever we install new SuperVersion, we might need to issue new flushes or

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

@@ -31,18 +31,6 @@ Status DBImpl::TEST_SwitchWAL() {
   return s;
 }
 
-bool DBImpl::TEST_WALBufferIsEmpty(bool lock) {
-  if (lock) {
-    log_write_mutex_.Lock();
-  }
-  log::Writer* cur_log_writer = logs_.back().writer;
-  auto res = cur_log_writer->TEST_BufferIsEmpty();
-  if (lock) {
-    log_write_mutex_.Unlock();
-  }
-  return res;
-}
-
 uint64_t DBImpl::TEST_MaxNextLevelOverlappingBytes(
     ColumnFamilyHandle* column_family) {
   ColumnFamilyData* cfd;

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

@@ -193,11 +193,14 @@ void DBImpl::FindObsoleteFiles(JobContext* job_context, bool force,
       }
     }
 
+    IOOptions io_opts;
+    io_opts.do_not_recurse = true;
     for (auto& path : paths) {
       // set of all files in the directory. We'll exclude files that are still
       // alive in the subsequent processings.
       std::vector<std::string> files;
-      Status s = env_->GetChildren(path, &files);
+      Status s = immutable_db_options_.fs->GetChildren(
+          path, io_opts, &files, /*IODebugContext*=*/nullptr);
       s.PermitUncheckedError();  // TODO: What should we do on error?
       for (const std::string& file : files) {
         uint64_t number;
@@ -223,7 +226,9 @@ void DBImpl::FindObsoleteFiles(JobContext* job_context, bool force,
     // Add log files in wal_dir
     if (!immutable_db_options_.IsWalDirSameAsDBPath(dbname_)) {
       std::vector<std::string> log_files;
-      Status s = env_->GetChildren(immutable_db_options_.wal_dir, &log_files);
+      Status s = immutable_db_options_.fs->GetChildren(
+          immutable_db_options_.wal_dir, io_opts, &log_files,
+          /*IODebugContext*=*/nullptr);
       s.PermitUncheckedError();  // TODO: What should we do on error?
       for (const std::string& log_file : log_files) {
         job_context->full_scan_candidate_files.emplace_back(
@@ -235,8 +240,9 @@ void DBImpl::FindObsoleteFiles(JobContext* job_context, bool force,
     if (!immutable_db_options_.db_log_dir.empty() &&
         immutable_db_options_.db_log_dir != dbname_) {
       std::vector<std::string> info_log_files;
-      Status s =
-          env_->GetChildren(immutable_db_options_.db_log_dir, &info_log_files);
+      Status s = immutable_db_options_.fs->GetChildren(
+          immutable_db_options_.db_log_dir, io_opts, &info_log_files,
+          /*IODebugContext*=*/nullptr);
       s.PermitUncheckedError();  // TODO: What should we do on error?
       for (std::string& log_file : info_log_files) {
         job_context->full_scan_candidate_files.emplace_back(

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

@@ -163,8 +163,11 @@ DBOptions SanitizeOptions(const std::string& dbname, const DBOptions& src,
     // DeleteScheduler::CleanupDirectory on the same dir later, it will be
     // safe
     std::vector<std::string> filenames;
+    IOOptions io_opts;
+    io_opts.do_not_recurse = true;
     auto wal_dir = immutable_db_options.GetWalDir();
-    Status s = result.env->GetChildren(wal_dir, &filenames);
+    Status s = immutable_db_options.fs->GetChildren(
+        wal_dir, io_opts, &filenames, /*IODebugContext*=*/nullptr);
     s.PermitUncheckedError();  //**TODO: What to do on error?
     for (std::string& filename : filenames) {
       if (filename.find(".log.trash", filename.length() -
@@ -432,7 +435,10 @@ Status DBImpl::Recover(
       s = env_->FileExists(current_fname);
     } else {
       s = Status::NotFound();
-      Status io_s = env_->GetChildren(dbname_, &files_in_dbname);
+      IOOptions io_opts;
+      io_opts.do_not_recurse = true;
+      Status io_s = immutable_db_options_.fs->GetChildren(
+          dbname_, io_opts, &files_in_dbname, /*IODebugContext*=*/nullptr);
       if (!io_s.ok()) {
         s = io_s;
         files_in_dbname.clear();
@@ -498,7 +504,10 @@ Status DBImpl::Recover(
     }
   } else if (immutable_db_options_.best_efforts_recovery) {
     assert(files_in_dbname.empty());
-    Status s = env_->GetChildren(dbname_, &files_in_dbname);
+    IOOptions io_opts;
+    io_opts.do_not_recurse = true;
+    Status s = immutable_db_options_.fs->GetChildren(
+        dbname_, io_opts, &files_in_dbname, /*IODebugContext*=*/nullptr);
     if (s.IsNotFound()) {
       return Status::InvalidArgument(dbname_,
                                      "does not exist (open for read only)");
@@ -570,7 +579,10 @@ Status DBImpl::Recover(
     // produced by an older version of rocksdb.
     auto wal_dir = immutable_db_options_.GetWalDir();
     if (!immutable_db_options_.best_efforts_recovery) {
-      s = env_->GetChildren(wal_dir, &files_in_wal_dir);
+      IOOptions io_opts;
+      io_opts.do_not_recurse = true;
+      s = immutable_db_options_.fs->GetChildren(
+          wal_dir, io_opts, &files_in_wal_dir, /*IODebugContext*=*/nullptr);
     }
     if (s.IsNotFound()) {
       return Status::InvalidArgument("wal_dir not found", wal_dir);
@@ -678,7 +690,10 @@ Status DBImpl::Recover(
       } else if (normalized_dbname == normalized_wal_dir) {
         filenames = std::move(files_in_wal_dir);
       } else {
-        s = env_->GetChildren(GetName(), &filenames);
+        IOOptions io_opts;
+        io_opts.do_not_recurse = true;
+        s = immutable_db_options_.fs->GetChildren(
+            GetName(), io_opts, &filenames, /*IODebugContext*=*/nullptr);
       }
     }
     if (s.ok()) {
@@ -1929,18 +1944,6 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
 
   if (s.ok()) {
     for (auto cfd : *impl->versions_->GetColumnFamilySet()) {
-      if (cfd->ioptions()->compaction_style == kCompactionStyleFIFO) {
-        auto* vstorage = cfd->current()->storage_info();
-        for (int i = 1; i < vstorage->num_levels(); ++i) {
-          int num_files = vstorage->NumLevelFiles(i);
-          if (num_files > 0) {
-            s = Status::InvalidArgument(
-                "Not all files are at level 0. Cannot "
-                "open with FIFO compaction style.");
-            break;
-          }
-        }
-      }
       if (!cfd->mem()->IsSnapshotSupported()) {
         impl->is_snapshot_supported_ = false;
       }
@@ -2022,9 +2025,13 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
     // Remove duplicate paths.
     std::sort(paths.begin(), paths.end());
     paths.erase(std::unique(paths.begin(), paths.end()), paths.end());
+    IOOptions io_opts;
+    io_opts.do_not_recurse = true;
     for (auto& path : paths) {
       std::vector<std::string> existing_files;
-      impl->immutable_db_options_.env->GetChildren(path, &existing_files)
+      impl->immutable_db_options_.fs
+          ->GetChildren(path, io_opts, &existing_files,
+                        /*IODebugContext*=*/nullptr)
           .PermitUncheckedError();  //**TODO: What do to on error?
       for (auto& file_name : existing_files) {
         uint64_t file_number;
@@ -2060,10 +2067,16 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
     ROCKS_LOG_HEADER(impl->immutable_db_options_.info_log, "DB pointer %p",
                      impl);
     LogFlush(impl->immutable_db_options_.info_log);
-    assert(impl->TEST_WALBufferIsEmpty());
-    // If the assert above fails then we need to FlushWAL before returning
-    // control back to the user.
-    if (!persist_options_status.ok()) {
+    if (!impl->WALBufferIsEmpty()) {
+      s = impl->FlushWAL(false);
+      if (s.ok()) {
+        // Sync is needed otherwise WAL buffered data might get lost after a
+        // power reset.
+        log::Writer* log_writer = impl->logs_.back().writer;
+        s = log_writer->file()->Sync(impl->immutable_db_options_.use_fsync);
+      }
+    }
+    if (s.ok() && !persist_options_status.ok()) {
       s = Status::IOError(
           "DB::Open() failed --- Unable to persist Options file",
           persist_options_status.ToString());

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

@@ -100,7 +100,11 @@ Status DBImplSecondary::FindNewLogNumbers(std::vector<uint64_t>* logs) {
   assert(logs != nullptr);
   std::vector<std::string> filenames;
   Status s;
-  s = env_->GetChildren(immutable_db_options_.GetWalDir(), &filenames);
+  IOOptions io_opts;
+  io_opts.do_not_recurse = true;
+  s = immutable_db_options_.fs->GetChildren(immutable_db_options_.GetWalDir(),
+                                            io_opts, &filenames,
+                                            /*IODebugContext*=*/nullptr);
   if (s.IsNotFound()) {
     return Status::InvalidArgument("Failed to open wal_dir",
                                    immutable_db_options_.GetWalDir());

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

@@ -111,6 +111,17 @@ Status DBImpl::DeleteRange(const WriteOptions& write_options,
   return DB::DeleteRange(write_options, column_family, begin_key, end_key);
 }
 
+Status DBImpl::DeleteRange(const WriteOptions& write_options,
+                           ColumnFamilyHandle* column_family,
+                           const Slice& begin_key, const Slice& end_key,
+                           const Slice& ts) {
+  const Status s = FailIfTsMismatchCf(column_family, ts, /*ts_for_read=*/false);
+  if (!s.ok()) {
+    return s;
+  }
+  return DB::DeleteRange(write_options, column_family, begin_key, end_key, ts);
+}
+
 void DBImpl::SetRecoverableStatePreReleaseCallback(
     PreReleaseCallback* callback) {
   recoverable_state_pre_release_callback_.reset(callback);
@@ -1749,6 +1760,7 @@ Status DBImpl::DelayWrite(uint64_t num_bytes,
                  &time_delayed);
     uint64_t delay =
         write_controller_.GetDelay(immutable_db_options_.clock, num_bytes);
+    TEST_SYNC_POINT("DBImpl::DelayWrite:Start");
     if (delay > 0) {
       if (write_options.no_slowdown) {
         return Status::Incomplete("Write stall");
@@ -1758,8 +1770,8 @@ Status DBImpl::DelayWrite(uint64_t num_bytes,
       // Notify write_thread_ about the stall so it can setup a barrier and
       // fail any pending writers with no_slowdown
       write_thread_.BeginWriteStall();
-      TEST_SYNC_POINT("DBImpl::DelayWrite:BeginWriteStallDone");
       mutex_.Unlock();
+      TEST_SYNC_POINT("DBImpl::DelayWrite:BeginWriteStallDone");
       // We will delay the write until we have slept for `delay` microseconds
       // or we don't need a delay anymore. We check for cancellation every 1ms
       // (slightly longer because WriteController minimum delay is 1ms, in
@@ -1784,7 +1796,8 @@ Status DBImpl::DelayWrite(uint64_t num_bytes,
     // might wait here indefinitely as the background compaction may never
     // finish successfully, resulting in the stall condition lasting
     // indefinitely
-    while (error_handler_.GetBGError().ok() && write_controller_.IsStopped()) {
+    while (error_handler_.GetBGError().ok() && write_controller_.IsStopped() &&
+           !shutting_down_.load(std::memory_order_relaxed)) {
       if (write_options.no_slowdown) {
         return Status::Incomplete("Write stall");
       }
@@ -1810,9 +1823,13 @@ Status DBImpl::DelayWrite(uint64_t num_bytes,
   // proceed
   Status s;
   if (write_controller_.IsStopped()) {
-    // If writes are still stopped, it means we bailed due to a background
-    // error
-    s = Status::Incomplete(error_handler_.GetBGError().ToString());
+    if (!shutting_down_.load(std::memory_order_relaxed)) {
+      // If writes are still stopped and db not shutdown, it means we bailed
+      // due to a background error
+      s = Status::Incomplete(error_handler_.GetBGError().ToString());
+    } else {
+      s = Status::ShutdownInProgress("stalled writes");
+    }
   }
   if (error_handler_.IsDBStopped()) {
     s = error_handler_.GetBGError();
@@ -2361,6 +2378,24 @@ Status DB::DeleteRange(const WriteOptions& opt,
   return Write(opt, &batch);
 }
 
+Status DB::DeleteRange(const WriteOptions& opt,
+                       ColumnFamilyHandle* column_family,
+                       const Slice& begin_key, const Slice& end_key,
+                       const Slice& ts) {
+  ColumnFamilyHandle* default_cf = DefaultColumnFamily();
+  assert(default_cf);
+  const Comparator* const default_cf_ucmp = default_cf->GetComparator();
+  assert(default_cf_ucmp);
+  WriteBatch batch(0 /* reserved_bytes */, 0 /* max_bytes */,
+                   opt.protection_bytes_per_key,
+                   default_cf_ucmp->timestamp_size());
+  Status s = batch.DeleteRange(column_family, begin_key, end_key, ts);
+  if (!s.ok()) {
+    return s;
+  }
+  return Write(opt, &batch);
+}
+
 Status DB::Merge(const WriteOptions& opt, ColumnFamilyHandle* column_family,
                  const Slice& key, const Slice& value) {
   WriteBatch batch(0 /* reserved_bytes */, 0 /* max_bytes */,

package/deps/rocksdb/rocksdb/db/db_iter.cc

@@ -89,6 +89,7 @@ DBIter::DBIter(Env* _env, const ReadOptions& read_options,
   if (iter_.iter()) {
     iter_.iter()->SetPinnedItersMgr(&pinned_iters_mgr_);
   }
+  status_.PermitUncheckedError();
   assert(timestamp_size_ == user_comparator_.timestamp_size());
 }
 

package/deps/rocksdb/rocksdb/db/db_iter_stress_test.cc

@@ -652,6 +652,7 @@ TEST_F(DBIteratorStressTest, StressTest) {
 }  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {
+  ROCKSDB_NAMESPACE::port::InstallStackTraceHandler();
   ::testing::InitGoogleTest(&argc, argv);
   ParseCommandLineFlags(&argc, &argv, true);
   return RUN_ALL_TESTS();

package/deps/rocksdb/rocksdb/db/db_iter_test.cc

@@ -3187,6 +3187,7 @@ TEST_F(DBIteratorTest, ReverseToForwardWithDisappearingKeys) {
 }  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {
+  ROCKSDB_NAMESPACE::port::InstallStackTraceHandler();
   ::testing::InitGoogleTest(&argc, argv);
   return RUN_ALL_TESTS();
 }

package/deps/rocksdb/rocksdb/db/db_iterator_test.cc

@@ -3232,6 +3232,28 @@ TEST_F(DBIteratorTest, BackwardIterationOnInplaceUpdateMemtable) {
   }
 }
 
+TEST_F(DBIteratorTest, IteratorRefreshReturnSV) {
+  Options options = CurrentOptions();
+  options.disable_auto_compactions = true;
+  DestroyAndReopen(options);
+  ASSERT_OK(
+      db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), "a", "z"));
+  std::unique_ptr<Iterator> iter{db_->NewIterator(ReadOptions())};
+  SyncPoint::GetInstance()->SetCallBack(
+      "ArenaWrappedDBIter::Refresh:SV", [&](void*) {
+        ASSERT_OK(db_->Put(WriteOptions(), "dummy", "new SV"));
+        // This makes the local SV obselete.
+        ASSERT_OK(Flush());
+        SyncPoint::GetInstance()->DisableProcessing();
+      });
+  SyncPoint::GetInstance()->EnableProcessing();
+  ASSERT_OK(iter->Refresh());
+  iter.reset();
+  // iter used to not cleanup SV, so the Close() below would hit an assertion
+  // error.
+  Close();
+}
+
 }  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {

package/deps/rocksdb/rocksdb/db/db_kv_checksum_test.cc

@@ -879,6 +879,7 @@ TEST_P(DbMemtableKVChecksumTest, FlushWithCorruptAfterMemtableInsert) {
 }  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {
+  ROCKSDB_NAMESPACE::port::InstallStackTraceHandler();
   ::testing::InitGoogleTest(&argc, argv);
   return RUN_ALL_TESTS();
 }

package/deps/rocksdb/rocksdb/db/db_logical_block_size_cache_test.cc

@@ -515,6 +515,7 @@ TEST_F(DBLogicalBlockSizeCacheTest, MultiDBWithSamePaths) {
 #endif  // OS_LINUX
 
 int main(int argc, char** argv) {
+  ROCKSDB_NAMESPACE::port::InstallStackTraceHandler();
   ::testing::InitGoogleTest(&argc, argv);
   return RUN_ALL_TESTS();
 }

package/deps/rocksdb/rocksdb/db/db_range_del_test.cc

@@ -13,6 +13,7 @@
 
 namespace ROCKSDB_NAMESPACE {
 
+// TODO(cbi): parameterize the test to cover user-defined timestamp cases
 class DBRangeDelTest : public DBTestBase {
  public:
   DBRangeDelTest() : DBTestBase("db_range_del_test", /*env_do_fsync=*/false) {}
@@ -1029,7 +1030,11 @@ TEST_F(DBRangeDelTest, CompactionTreatsSplitInputLevelDeletionAtomically) {
   options.level0_file_num_compaction_trigger = kNumFilesPerLevel;
   options.memtable_factory.reset(
       test::NewSpecialSkipListFactory(2 /* num_entries_flush */));
-  options.target_file_size_base = kValueBytes;
+  // max file size could be 2x of target file size, so set it to half of that
+  options.target_file_size_base = kValueBytes / 2;
+  // disable dynamic_file_size, as it will cut L1 files into more files (than
+  // kNumFilesPerLevel).
+  options.level_compaction_dynamic_file_size = false;
   options.max_compaction_bytes = 1500;
   // i == 0: CompactFiles
   // i == 1: CompactRange
@@ -1100,6 +1105,12 @@ TEST_F(DBRangeDelTest, RangeTombstoneEndKeyAsSstableUpperBound) {
       test::NewSpecialSkipListFactory(2 /* num_entries_flush */));
   options.target_file_size_base = kValueBytes;
   options.disable_auto_compactions = true;
+  // disable it for now, otherwise the L1 files are going be cut before data 1:
+  // L1: [0]   [1,4]
+  // L2: [0,0]
+  // because the grandparent file is between [0]->[1] and it's size is more than
+  // 1/8 of target size (4k).
+  options.level_compaction_dynamic_file_size = false;
 
   DestroyAndReopen(options);
 
@@ -1723,9 +1734,8 @@ TEST_F(DBRangeDelTest, OverlappedKeys) {
 
   ASSERT_OK(dbfull()->TEST_CompactRange(1, nullptr, nullptr, nullptr,
                                         true /* disallow_trivial_move */));
-  ASSERT_EQ(
-      3, NumTableFilesAtLevel(
-             2));  // L1->L2 compaction size is limited to max_compaction_bytes
+  // L1->L2 compaction size is limited to max_compaction_bytes
+  ASSERT_EQ(3, NumTableFilesAtLevel(2));
   ASSERT_EQ(0, NumTableFilesAtLevel(1));
 }
 
@@ -1974,6 +1984,40 @@ TEST_F(DBRangeDelTest, TombstoneFromCurrentLevel) {
   delete iter;
 }
 
+class TombstoneTestSstPartitioner : public SstPartitioner {
+ public:
+  const char* Name() const override { return "SingleKeySstPartitioner"; }
+
+  PartitionerResult ShouldPartition(
+      const PartitionerRequest& request) override {
+    if (cmp->Compare(*request.current_user_key, DBTestBase::Key(5)) == 0) {
+      return kRequired;
+    } else {
+      return kNotRequired;
+    }
+  }
+
+  bool CanDoTrivialMove(const Slice& /*smallest_user_key*/,
+                        const Slice& /*largest_user_key*/) override {
+    return false;
+  }
+
+  const Comparator* cmp = BytewiseComparator();
+};
+
+class TombstoneTestSstPartitionerFactory : public SstPartitionerFactory {
+ public:
+  static const char* kClassName() {
+    return "TombstoneTestSstPartitionerFactory";
+  }
+  const char* Name() const override { return kClassName(); }
+
+  std::unique_ptr<SstPartitioner> CreatePartitioner(
+      const SstPartitioner::Context& /* context */) const override {
+    return std::unique_ptr<SstPartitioner>(new TombstoneTestSstPartitioner());
+  }
+};
+
 TEST_F(DBRangeDelTest, TombstoneAcrossFileBoundary) {
   // Verify that a range tombstone across file boundary covers keys from older
   // levels. Test set up:
@@ -1987,11 +2031,17 @@ TEST_F(DBRangeDelTest, TombstoneAcrossFileBoundary) {
   options.target_file_size_base = 2 * 1024;
   options.max_compaction_bytes = 2 * 1024;
 
+  // Make sure L1 files are split before "5"
+  auto factory = std::make_shared<TombstoneTestSstPartitionerFactory>();
+  options.sst_partitioner_factory = factory;
+
   DestroyAndReopen(options);
 
   Random rnd(301);
   // L2
-  ASSERT_OK(db_->Put(WriteOptions(), Key(5), rnd.RandomString(1 << 10)));
+  // the file should be smaller than max_compaction_bytes, otherwise the file
+  // will be cut before 7.
+  ASSERT_OK(db_->Put(WriteOptions(), Key(5), rnd.RandomString(1 << 9)));
   ASSERT_OK(db_->Flush(FlushOptions()));
   MoveFilesToLevel(2);
   ASSERT_EQ(1, NumTableFilesAtLevel(2));
@@ -2687,6 +2737,23 @@ TEST_F(DBRangeDelTest, PrefixSentinelKey) {
   delete iter;
 }
 
+TEST_F(DBRangeDelTest, RefreshMemtableIter) {
+  Options options = CurrentOptions();
+  options.disable_auto_compactions = true;
+  DestroyAndReopen(options);
+  ASSERT_OK(
+      db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), "a", "z"));
+  ReadOptions ro;
+  ro.read_tier = kMemtableTier;
+  std::unique_ptr<Iterator> iter{db_->NewIterator(ro)};
+  ASSERT_OK(Flush());
+  // First refresh reinits iter, which had a bug where
+  // iter.memtable_range_tombstone_iter_ was not set to nullptr, and caused
+  // subsequent refresh to double free.
+  ASSERT_OK(iter->Refresh());
+  ASSERT_OK(iter->Refresh());
+}
+
 #endif  // ROCKSDB_LITE
 
 }  // namespace ROCKSDB_NAMESPACE