[Enhancement] set empty delta column group when new segment generate (S…

…tarRocks#25506) Fixes StarRocks#20436 Set empty delta column group when new segment generate Signed-off-by: luohaha <[email protected]>
abc982627271 · Jun 20, 2023 · 5fe3afc · 5fe3afc
1 parent de0fc73
commit 5fe3afc
Show file tree

Hide file tree

Showing 4 changed files with 65 additions and 1 deletion.
diff --git a/be/src/storage/tablet_updates.cpp b/be/src/storage/tablet_updates.cpp
@@ -1302,6 +1302,8 @@ void TabletUpdates::_apply_normal_rowset_commit(const EditVersionInfo& version_i
         for (auto& delvec_pair : new_del_vecs) {
             tsid.segment_id = delvec_pair.first;
             manager->set_cached_del_vec(tsid, delvec_pair.second);
+            // try to set empty dcg cache, for improving latency when reading
+            manager->set_cached_empty_delta_column_group(_tablet.data_dir()->get_meta(), tsid);
         }
         // 5. apply memory
         _next_log_id++;
@@ -1836,6 +1838,8 @@ void TabletUpdates::_apply_compaction_commit(const EditVersionInfo& version_info
         for (auto& delvec_pair : delvecs) {
             tsid.segment_id = delvec_pair.first;
             manager->set_cached_del_vec(tsid, delvec_pair.second);
+            // try to set empty dcg cache, for improving latency when reading
+            manager->set_cached_empty_delta_column_group(_tablet.data_dir()->get_meta(), tsid);
         }
         // 5. apply memory
         _next_log_id++;

diff --git a/be/src/storage/update_manager.cpp b/be/src/storage/update_manager.cpp
@@ -243,7 +243,7 @@ StatusOr<size_t> UpdateManager::clear_delta_column_group_before_version(KVStore*
     std::vector<std::pair<TabletSegmentId, int64_t>> clear_dcgs;
     const int64_t begin_ms = UnixMillis();
     auto is_timeout = [begin_ms]() {
-        if (UnixMillis() > begin_ms + 100) { // only hold cache_clock for 100ms max.
+        if (UnixMillis() > begin_ms + 10) { // only hold cache_clock for 10ms max.
             return true;
         } else {
             return false;
@@ -283,6 +283,46 @@ void UpdateManager::clear_cached_delta_column_group(const std::vector<TabletSegm
     }
 }
 
+Status UpdateManager::set_cached_empty_delta_column_group(KVStore* meta, const TabletSegmentId& tsid) {
+    {
+        std::lock_guard<std::mutex> lg(_delta_column_group_cache_lock);
+        auto itr = _delta_column_group_cache.find(tsid);
+        if (itr != _delta_column_group_cache.end()) {
+            // already exist, not need to cache
+            return Status::OK();
+        }
+    }
+    // find from rocksdb
+    DeltaColumnGroupList new_dcgs;
+    RETURN_IF_ERROR(
+            TabletMetaManager::get_delta_column_group(meta, tsid.tablet_id, tsid.segment_id, INT64_MAX, &new_dcgs));
+    std::lock_guard<std::mutex> lg(_delta_column_group_cache_lock);
+    auto itr = _delta_column_group_cache.find(tsid);
+    if (itr != _delta_column_group_cache.end()) {
+        // already exist, not need to cache
+        return Status::OK();
+    }
+    if (new_dcgs.empty()) {
+        // only set empty dcgs
+        _delta_column_group_cache[tsid] = new_dcgs;
+    }
+    return Status::OK();
+}
+
+bool UpdateManager::get_cached_delta_column_group(const TabletSegmentId& tsid, int64_t version,
+                                                  DeltaColumnGroupList* dcgs) {
+    // find in delta column group cache
+    std::lock_guard<std::mutex> lg(_delta_column_group_cache_lock);
+    auto itr = _delta_column_group_cache.find(tsid);
+    if (itr != _delta_column_group_cache.end()) {
+        search_delta_column_groups_by_version(itr->second, version, dcgs);
+        // hit cache
+        return true;
+    }
+    // miss cache
+    return false;
+}
+
 Status UpdateManager::set_cached_delta_column_group(KVStore* meta, const TabletSegmentId& tsid,
                                                     const DeltaColumnGroupPtr& dcg) {
     {

diff --git a/be/src/storage/update_manager.h b/be/src/storage/update_manager.h
@@ -101,6 +101,10 @@ class UpdateManager {
 
     Status set_cached_delta_column_group(KVStore* meta, const TabletSegmentId& tsid, const DeltaColumnGroupPtr& dcg);
 
+    Status set_cached_empty_delta_column_group(KVStore* meta, const TabletSegmentId& tsid);
+
+    bool get_cached_delta_column_group(const TabletSegmentId& tsid, int64_t version, DeltaColumnGroupList* dcgs);
+
     void clear_cache();
 
     void clear_cached_del_vec(const std::vector<TabletSegmentId>& tsids);

diff --git a/be/test/storage/update_manager_test.cpp b/be/test/storage/update_manager_test.cpp
@@ -196,4 +196,20 @@ TEST_F(UpdateManagerTest, testExpireEntry) {
     ASSERT_EQ(peak_size - expiring_size, remaining_size);
 }
 
+TEST_F(UpdateManagerTest, testSetEmptyCachedDeltaColumnGroup) {
+    srand(time(nullptr));
+    create_tablet(rand(), rand());
+    TabletSegmentId tsid;
+    tsid.tablet_id = _tablet->tablet_id();
+    tsid.segment_id = 1;
+    _update_manager->set_cached_empty_delta_column_group(_tablet->data_dir()->get_meta(), tsid);
+    // search this empty dcg
+    DeltaColumnGroupList dcgs;
+    // search in cache
+    ASSERT_TRUE(_update_manager->get_cached_delta_column_group(tsid, 1, &dcgs));
+    ASSERT_TRUE(dcgs.empty());
+    _update_manager->get_delta_column_group(_tablet->data_dir()->get_meta(), tsid, 1, &dcgs);
+    ASSERT_TRUE(dcgs.empty());
+}
+
 } // namespace starrocks