Trigger compaction to the next level if the data age exceeds periodic_compaction_seconds (#12175)

Summary:
Currently, the data are always compacted to the same level if exceed periodic_compaction_seconds which may confuse users, so we change it to allow trigger compaction to the next level here. It's a behavior change to users, and may affect users
who have disabled their ttl or ttl > periodic_compaction_seconds.

Relate issue: https://github.com/facebook/rocksdb/issues/12165

Pull Request resolved: https://github.com/facebook/rocksdb/pull/12175

Reviewed By: ajkr

Differential Revision: D52446722

Pulled By: cbi42

fbshipit-source-id: ccd3d2c6434ed77055735a03408d4a62d119342f

db/compaction/compaction_picker_level.cc

@@ -45,6 +45,13 @@ bool LevelCompactionPicker::NeedsCompaction(
 }
 
 namespace {
+
+enum class CompactToNextLevel {
+  kNo,   // compact to the same level as the input file
+  kYes,  // compact to the next level except the last level to the same level
+  kSkipLastLevel,  // compact to the next level but skip the last level
+};
+
 // A class to build a leveled compaction step-by-step.
 class LevelCompactionBuilder {
  public:
@@ -115,9 +122,10 @@ class LevelCompactionBuilder {
   // level_files is a vector of (level, file metadata) in ascending order of
   // level. If compact_to_next_level is true, compact the file to the next
   // level, otherwise, compact to the same level as the input file.
+  // If skip_last_level is true, skip the last level.
   void PickFileToCompact(
       const autovector<std::pair<int, FileMetaData*>>& level_files,
-      bool compact_to_next_level);
+      CompactToNextLevel compact_to_next_level);
 
   const std::string& cf_name_;
   VersionStorageInfo* vstorage_;
@@ -149,20 +157,24 @@ class LevelCompactionBuilder {
 
 void LevelCompactionBuilder::PickFileToCompact(
     const autovector<std::pair<int, FileMetaData*>>& level_files,
-    bool compact_to_next_level) {
+    CompactToNextLevel compact_to_next_level) {
   for (auto& level_file : level_files) {
     // If it's being compacted it has nothing to do here.
     // If this assert() fails that means that some function marked some
     // files as being_compacted, but didn't call ComputeCompactionScore()
     assert(!level_file.second->being_compacted);
     start_level_ = level_file.first;
-    if ((compact_to_next_level &&
+    if ((compact_to_next_level == CompactToNextLevel::kSkipLastLevel &&
          start_level_ == vstorage_->num_non_empty_levels() - 1) ||
         (start_level_ == 0 &&
          !compaction_picker_->level0_compactions_in_progress()->empty())) {
       continue;
     }
-    if (compact_to_next_level) {
+
+    // Compact to the next level only if the file is not in the last level and
+    // compact_to_next_level is kYes or kSkipLastLevel.
+    if (compact_to_next_level != CompactToNextLevel::kNo &&
+        (start_level_ < vstorage_->num_non_empty_levels() - 1)) {
       output_level_ =
           (start_level_ == 0) ? vstorage_->base_level() : start_level_ + 1;
     } else {
@@ -248,7 +260,8 @@ void LevelCompactionBuilder::SetupInitialFiles() {
   }
 
   // Bottommost Files Compaction on deleting tombstones
-  PickFileToCompact(vstorage_->BottommostFilesMarkedForCompaction(), false);
+  PickFileToCompact(vstorage_->BottommostFilesMarkedForCompaction(),
+                    CompactToNextLevel::kNo);
   if (!start_level_inputs_.empty()) {
     compaction_reason_ = CompactionReason::kBottommostFiles;
     return;
@@ -274,21 +287,26 @@ void LevelCompactionBuilder::SetupInitialFiles() {
     }
   }
 
-  PickFileToCompact(vstorage_->ExpiredTtlFiles(), true);
+  PickFileToCompact(vstorage_->ExpiredTtlFiles(),
+                    CompactToNextLevel::kSkipLastLevel);
   if (!start_level_inputs_.empty()) {
     compaction_reason_ = CompactionReason::kTtl;
     return;
   }
 
   // Periodic Compaction
-  PickFileToCompact(vstorage_->FilesMarkedForPeriodicCompaction(), false);
+  PickFileToCompact(vstorage_->FilesMarkedForPeriodicCompaction(),
+                    ioptions_.level_compaction_dynamic_level_bytes
+                        ? CompactToNextLevel::kYes
+                        : CompactToNextLevel::kNo);
   if (!start_level_inputs_.empty()) {
     compaction_reason_ = CompactionReason::kPeriodicCompaction;
     return;
   }
 
   // Forced blob garbage collection
-  PickFileToCompact(vstorage_->FilesMarkedForForcedBlobGC(), false);
+  PickFileToCompact(vstorage_->FilesMarkedForForcedBlobGC(),
+                    CompactToNextLevel::kNo);
   if (!start_level_inputs_.empty()) {
     compaction_reason_ = CompactionReason::kForcedBlobGC;
     return;

include/rocksdb/advanced_options.h

@@ -914,7 +914,9 @@ struct AdvancedColumnFamilyOptions {
   //
   // Leveled: files older than `periodic_compaction_seconds` will be picked up
   //    for compaction and will be re-written to the same level as they were
-  //    before.
+  //    before if level_compaction_dynamic_level_bytes is disabled. Otherwise,
+  //    it will rewrite files to the next level except for the last level files
+  //    to the same level.
   //
   // FIFO: not supported. Setting this option has no effect for FIFO compaction.
   //

unreleased_history/README.txt

@@ -48,7 +48,7 @@ Updating HISTORY.md with release notes
 --------------------------------------
 
 The script unreleased_history/release.sh does this. Run the script before
-updating version.h to the next develpment release, so that the script will pick
+updating version.h to the next development release, so that the script will pick
 up the version being released. You might want to start with
 
 $ DRY_RUN=1 unreleased_history/release.sh | less
@@ -68,6 +68,6 @@ First, it was common to hit unnecessary merge conflicts when adding entries to
 HISTORY.md, which slowed development. Second, when a PR was opened before a
 release cut and landed after the release cut, it was easy to add the HISTORY
 entry to the wrong version's history. This new setup completely fixes both of
-those issues, with perhaps slighly more initial work to create each entry.
+those issues, with perhaps slightly more initial work to create each entry.
 There is also now an extra step in using `git blame` to map a release note
 to its source code implementation, but that is a relatively rare operation.

unreleased_history/behavior_changes/change_behavior_of_compact_to_next_level.md

@@ -0,0 +1 @@
+Files will be compacted to the next level if the data age exceeds periodic_compaction_seconds except for the last level.