statistics/handle: refine the condition of dumping stats delta (#41133) (#41168)

ref #36004, ref #38189, ref #39785

Author: ti-chi-bot

statistics/handle/update.go

@@ -403,22 +403,35 @@ var (
 	dumpStatsMaxDuration = time.Hour
 )
 
-// needDumpStatsDelta returns true when only updates a small portion of the table and the time since last update
-// do not exceed one hour.
-func needDumpStatsDelta(h *Handle, id int64, item variable.TableDelta, currentTime time.Time) bool {
-	if item.InitTime.IsZero() {
-		item.InitTime = currentTime
+// needDumpStatsDelta checks whether to dump stats delta.
+// 1. If the table doesn't exist or is a mem table or system table, then return false.
+// 2. If the mode is DumpAll, then return true.
+// 3. If the stats delta haven't been dumped in the past hour, then return true.
+// 4. If the table stats is pseudo or empty or `Modify Count / Table Count` exceeds the threshold.
+func (h *Handle) needDumpStatsDelta(is infoschema.InfoSchema, mode dumpMode, id int64, item variable.TableDelta, currentTime time.Time) bool {
+	tbl, ok := h.getTableByPhysicalID(is, id)
+	if !ok {
+		return false
 	}
-	tbl, ok := h.statsCache.Load().(statsCache).Get(id)
+	dbInfo, ok := is.SchemaByTable(tbl.Meta())
 	if !ok {
-		// No need to dump if the stats is invalid.
 		return false
 	}
+	if util.IsMemOrSysDB(dbInfo.Name.L) {
+		return false
+	}
+	if mode == DumpAll {
+		return true
+	}
+	if item.InitTime.IsZero() {
+		item.InitTime = currentTime
+	}
 	if currentTime.Sub(item.InitTime) > dumpStatsMaxDuration {
 		// Dump the stats to kv at least once an hour.
 		return true
 	}
-	if tbl.Count == 0 || float64(item.Count)/float64(tbl.Count) > DumpStatsDeltaRatio {
+	statsTbl := h.GetPartitionStats(tbl.Meta(), id)
+	if statsTbl.Pseudo || statsTbl.Count == 0 || float64(item.Count)/float64(statsTbl.Count) > DumpStatsDeltaRatio {
 		// Dump the stats when there are many modifications.
 		return true
 	}
@@ -487,9 +500,15 @@ func (h *Handle) DumpStatsDeltaToKV(mode dumpMode) error {
 		h.globalMap.data = deltaMap
 		h.globalMap.Unlock()
 	}()
+	// TODO: pass in do.InfoSchema() to DumpStatsDeltaToKV.
+	is := func() infoschema.InfoSchema {
+		h.mu.Lock()
+		defer h.mu.Unlock()
+		return h.mu.ctx.GetDomainInfoSchema().(infoschema.InfoSchema)
+	}()
 	currentTime := time.Now()
 	for id, item := range deltaMap {
-		if mode == DumpDelta && !needDumpStatsDelta(h, id, item, currentTime) {
+		if !h.needDumpStatsDelta(is, mode, id, item, currentTime) {
 			continue
 		}
 		updated, err := h.dumpTableStatCountToKV(id, item)

statistics/handle/update_test.go

@@ -2691,20 +2691,42 @@ func TestFillMissingStatsMeta(t *testing.T) {
 	}
 
 	tk.MustExec("insert into t1 values (1, 2), (3, 4)")
-	require.NoError(t, h.DumpStatsDeltaToKV(handle.DumpAll))
+	require.NoError(t, h.DumpStatsDeltaToKV(handle.DumpDelta))
+	require.NoError(t, h.Update(is))
 	ver1 := checkStatsMeta(tbl1ID, "2", "2")
 	tk.MustExec("delete from t1 where a = 1")
-	require.NoError(t, h.DumpStatsDeltaToKV(handle.DumpAll))
+	require.NoError(t, h.DumpStatsDeltaToKV(handle.DumpDelta))
+	require.NoError(t, h.Update(is))
 	ver2 := checkStatsMeta(tbl1ID, "3", "1")
 	require.Greater(t, ver2, ver1)
 
 	tk.MustExec("insert into t2 values (1, 2), (3, 4)")
-	require.NoError(t, h.DumpStatsDeltaToKV(handle.DumpAll))
+	require.NoError(t, h.DumpStatsDeltaToKV(handle.DumpDelta))
+	require.NoError(t, h.Update(is))
 	checkStatsMeta(p0ID, "2", "2")
 	globalVer1 := checkStatsMeta(tbl2ID, "2", "2")
 	tk.MustExec("insert into t2 values (11, 12)")
-	require.NoError(t, h.DumpStatsDeltaToKV(handle.DumpAll))
+	require.NoError(t, h.DumpStatsDeltaToKV(handle.DumpDelta))
+	require.NoError(t, h.Update(is))
 	checkStatsMeta(p1ID, "1", "1")
 	globalVer2 := checkStatsMeta(tbl2ID, "3", "3")
 	require.Greater(t, globalVer2, globalVer1)
 }
+
+func TestNotDumpSysTable(t *testing.T) {
+	store, dom := testkit.CreateMockStoreAndDomain(t)
+	tk := testkit.NewTestKit(t, store)
+	tk.MustExec("use test")
+	tk.MustExec("create table t1 (a int, b int)")
+	h := dom.StatsHandle()
+	require.NoError(t, h.HandleDDLEvent(<-h.DDLEventCh()))
+	tk.MustQuery("select count(1) from mysql.stats_meta").Check(testkit.Rows("1"))
+	// After executing `delete from mysql.stats_meta`, a delta for mysql.stats_meta is created but it would not be dumped.
+	tk.MustExec("delete from mysql.stats_meta")
+	require.NoError(t, h.DumpStatsDeltaToKV(handle.DumpAll))
+	is := dom.InfoSchema()
+	tbl, err := is.TableByName(model.NewCIStr("mysql"), model.NewCIStr("stats_meta"))
+	require.NoError(t, err)
+	tblID := tbl.Meta().ID
+	tk.MustQuery(fmt.Sprintf("select * from mysql.stats_meta where table_id = %v", tblID)).Check(testkit.Rows())
+}