Planner: Estimate to recognize modifyCount when all TopN collected (#…

…55077)

close #47400

pkg/planner/cardinality/row_count_column.go

@@ -124,7 +124,7 @@ func GetRowCountByIntColumnRanges(sctx context.PlanContext, coll *statistics.His
 }
 
 // equalRowCountOnColumn estimates the row count by a slice of Range and a Datum.
-func equalRowCountOnColumn(sctx context.PlanContext, c *statistics.Column, val types.Datum, encodedVal []byte, realtimeRowCount int64) (result float64, err error) {
+func equalRowCountOnColumn(sctx context.PlanContext, c *statistics.Column, val types.Datum, encodedVal []byte, realtimeRowCount, modifyCount int64) (result float64, err error) {
 	if sctx.GetSessionVars().StmtCtx.EnableOptimizerDebugTrace {
 		debugtrace.EnterContextCommon(sctx)
 		debugtrace.RecordAnyValuesWithNames(sctx, "Value", val.String(), "Encoded", encodedVal)
@@ -172,7 +172,11 @@ func equalRowCountOnColumn(sctx context.PlanContext, c *statistics.Column, val t
 	// 3. use uniform distribution assumption for the rest (even when this value is not covered by the range of stats)
 	histNDV := float64(c.Histogram.NDV - int64(c.TopN.Num()))
 	if histNDV <= 0 {
-		return 0, nil
+		// If the table hasn't been modified, it's safe to return 0. Otherwise, the TopN could be stale - return 1.
+		if modifyCount == 0 {
+			return 0, nil
+		}
+		return 1, nil
 	}
 	return c.Histogram.NotNullCount() / histNDV, nil
 }
@@ -224,7 +228,7 @@ func GetColumnRowCount(sctx context.PlanContext, c *statistics.Column, ranges []
 					continue
 				}
 				var cnt float64
-				cnt, err = equalRowCountOnColumn(sctx, c, lowVal, lowEncoded, realtimeRowCount)
+				cnt, err = equalRowCountOnColumn(sctx, c, lowVal, lowEncoded, realtimeRowCount, modifyCount)
 				if err != nil {
 					return 0, errors.Trace(err)
 				}
@@ -245,7 +249,7 @@ func GetColumnRowCount(sctx context.PlanContext, c *statistics.Column, ranges []
 			// case 2: it's a small range && using ver1 stats
 			if rangeVals != nil {
 				for _, val := range rangeVals {
-					cnt, err := equalRowCountOnColumn(sctx, c, val, lowEncoded, realtimeRowCount)
+					cnt, err := equalRowCountOnColumn(sctx, c, val, lowEncoded, realtimeRowCount, modifyCount)
 					if err != nil {
 						return 0, err
 					}
@@ -269,7 +273,7 @@ func GetColumnRowCount(sctx context.PlanContext, c *statistics.Column, ranges []
 		// And because we use (2, MaxValue] to represent expressions like a > 2 and use [MinNotNull, 3) to represent
 		//   expressions like b < 3, we need to exclude the special values.
 		if rg.LowExclude && !lowVal.IsNull() && lowVal.Kind() != types.KindMaxValue && lowVal.Kind() != types.KindMinNotNull {
-			lowCnt, err := equalRowCountOnColumn(sctx, c, lowVal, lowEncoded, realtimeRowCount)
+			lowCnt, err := equalRowCountOnColumn(sctx, c, lowVal, lowEncoded, realtimeRowCount, modifyCount)
 			if err != nil {
 				return 0, errors.Trace(err)
 			}
@@ -280,7 +284,7 @@ func GetColumnRowCount(sctx context.PlanContext, c *statistics.Column, ranges []
 			cnt += float64(c.NullCount)
 		}
 		if !rg.HighExclude && highVal.Kind() != types.KindMaxValue && highVal.Kind() != types.KindMinNotNull {
-			highCnt, err := equalRowCountOnColumn(sctx, c, highVal, highEncoded, realtimeRowCount)
+			highCnt, err := equalRowCountOnColumn(sctx, c, highVal, highEncoded, realtimeRowCount, modifyCount)
 			if err != nil {
 				return 0, errors.Trace(err)
 			}
@@ -376,7 +380,7 @@ func ColumnEqualRowCount(sctx context.PlanContext, t *statistics.Table, value ty
 	if err != nil {
 		return 0, err
 	}
-	result, err := equalRowCountOnColumn(sctx, c, value, encodedVal, t.ModifyCount)
+	result, err := equalRowCountOnColumn(sctx, c, value, encodedVal, t.RealtimeCount, t.ModifyCount)
 	result *= c.GetIncreaseFactor(t.RealtimeCount)
 	return result, errors.Trace(err)
 }

pkg/planner/cardinality/row_count_index.go

@@ -256,7 +256,7 @@ func getIndexRowCountForStatsV2(sctx context.PlanContext, idx *statistics.Index,
 					}
 					continue
 				}
-				count = equalRowCountOnIndex(sctx, idx, lb, realtimeRowCount)
+				count = equalRowCountOnIndex(sctx, idx, lb, realtimeRowCount, modifyCount)
 				// If the current table row count has changed, we should scale the row count accordingly.
 				count *= idx.GetIncreaseFactor(realtimeRowCount)
 				if debugTrace {
@@ -356,7 +356,7 @@ func getIndexRowCountForStatsV2(sctx context.PlanContext, idx *statistics.Index,
 
 var nullKeyBytes, _ = codec.EncodeKey(time.UTC, nil, types.NewDatum(nil))
 
-func equalRowCountOnIndex(sctx context.PlanContext, idx *statistics.Index, b []byte, realtimeRowCount int64) (result float64) {
+func equalRowCountOnIndex(sctx context.PlanContext, idx *statistics.Index, b []byte, realtimeRowCount, modifyCount int64) (result float64) {
 	if sctx.GetSessionVars().StmtCtx.EnableOptimizerDebugTrace {
 		debugtrace.EnterContextCommon(sctx)
 		debugtrace.RecordAnyValuesWithNames(sctx, "Encoded Value", b)
@@ -397,7 +397,11 @@ func equalRowCountOnIndex(sctx context.PlanContext, idx *statistics.Index, b []b
 	// 3. use uniform distribution assumption for the rest (even when this value is not covered by the range of stats)
 	histNDV := float64(idx.Histogram.NDV - int64(idx.TopN.Num()))
 	if histNDV <= 0 {
-		return 0
+		// If the table hasn't been modified, it's safe to return 0. Otherwise, the TopN could be stale - return 1.
+		if modifyCount == 0 {
+			return 0
+		}
+		return 1
 	}
 	return idx.Histogram.NotNullCount() / histNDV
 }