statistics: do not depend on table information when calculating the t…

…able size (#56036) ref #55906
pingcap · Sep 14, 2024 · 3688a2b · 3688a2b
1 parent ae86e04
commit 3688a2b
Show file tree

Hide file tree

Showing 3 changed files with 21 additions and 58 deletions.
diff --git a/pkg/statistics/handle/autoanalyze/refresher/refresher.go b/pkg/statistics/handle/autoanalyze/refresher/refresher.go
@@ -347,7 +347,7 @@ func CreateTableAnalysisJob(
 	statistics.CheckAnalyzeVerOnTable(tblStats, &tableStatsVer)
 
 	changePercentage := CalculateChangePercentage(tblStats, autoAnalyzeRatio)
-	tableSize := calculateTableSize(tblInfo, tblStats)
+	tableSize := calculateTableSize(tblStats)
 	lastAnalysisDuration := GetTableLastAnalyzeDuration(tblStats, currentTs)
 	indexes := CheckIndexesNeedAnalyze(tblInfo, tblStats)
 
@@ -391,7 +391,7 @@ func CreateStaticPartitionAnalysisJob(
 	statistics.CheckAnalyzeVerOnTable(partitionStats, &tableStatsVer)
 
 	changePercentage := CalculateChangePercentage(partitionStats, autoAnalyzeRatio)
-	tableSize := calculateTableSize(globalTblInfo, partitionStats)
+	tableSize := calculateTableSize(partitionStats)
 	lastAnalysisDuration := GetTableLastAnalyzeDuration(partitionStats, currentTs)
 	indexes := CheckIndexesNeedAnalyze(globalTblInfo, partitionStats)
 
@@ -448,12 +448,11 @@ func CalculateChangePercentage(
 }
 
 func calculateTableSize(
-	tblInfo *model.TableInfo,
 	tblStats *statistics.Table,
 ) float64 {
 	tblCnt := float64(tblStats.RealtimeCount)
-	// TODO: Ignore unanalyzable columns.
-	colCnt := float64(len(tblInfo.Columns))
+	colCnt := float64(tblStats.ColAndIdxExistenceMap.ColNum())
+	intest.Assert(colCnt != 0, "Column count should not be 0")
 
 	return tblCnt * colCnt
 }
@@ -526,7 +525,7 @@ func createTableAnalysisJobForPartitions(
 	statistics.CheckAnalyzeVerOnTable(tblStats, &tableStatsVer)
 
 	averageChangePercentage, avgSize, minLastAnalyzeDuration, partitionNames := CalculateIndicatorsForPartitions(
-		tblInfo,
+		tblStats,
 		partitionStats,
 		autoAnalyzeRatio,
 		currentTs,
@@ -563,7 +562,7 @@ func createTableAnalysisJobForPartitions(
 // Size is the product of the number of rows and the number of columns.
 // Last analyze duration is the duration since the last analyze.
 func CalculateIndicatorsForPartitions(
-	tblInfo *model.TableInfo,
+	globalStats *statistics.Table,
 	partitionStats map[PartitionIDAndName]*statistics.Table,
 	autoAnalyzeRatio float64,
 	currentTs uint64,
@@ -577,7 +576,8 @@ func CalculateIndicatorsForPartitions(
 	totalSize := 0.0
 	count := 0.0
 	partitionNames = make([]string, 0, len(partitionStats))
-	cols := float64(len(tblInfo.Columns))
+	cols := float64(globalStats.ColAndIdxExistenceMap.ColNum())
+	intest.Assert(cols != 0, "Column count should not be 0")
 	totalLastAnalyzeDuration := time.Duration(0)
 
 	for pIDAndName, tblStats := range partitionStats {

diff --git a/pkg/statistics/handle/autoanalyze/refresher/refresher_test.go b/pkg/statistics/handle/autoanalyze/refresher/refresher_test.go
@@ -614,7 +614,7 @@ func TestCalculateIndicatorsForPartitions(t *testing.T) {
 	analyzedMap.InsertIndex(1, nil, true)
 	tests := []struct {
 		name                       string
-		tblInfo                    *model.TableInfo
+		globalStats                *statistics.Table
 		partitionStats             map[refresher.PartitionIDAndName]*statistics.Table
 		defs                       []model.PartitionDefinition
 		autoAnalyzeRatio           float64
@@ -626,22 +626,8 @@ func TestCalculateIndicatorsForPartitions(t *testing.T) {
 	}{
 		{
 			name: "Test Table not analyzed",
-			tblInfo: &model.TableInfo{
-				Indices: []*model.IndexInfo{
-					{
-						ID:    1,
-						Name:  pmodel.NewCIStr("index1"),
-						State: model.StatePublic,
-					},
-				},
-				Columns: []*model.ColumnInfo{
-					{
-						ID: 1,
-					},
-					{
-						ID: 2,
-					},
-				},
+			globalStats: &statistics.Table{
+				ColAndIdxExistenceMap: analyzedMap,
 			},
 			partitionStats: map[refresher.PartitionIDAndName]*statistics.Table{
 				{
@@ -684,22 +670,8 @@ func TestCalculateIndicatorsForPartitions(t *testing.T) {
 		},
 		{
 			name: "Test Table analyzed and only one partition meets the threshold",
-			tblInfo: &model.TableInfo{
-				Indices: []*model.IndexInfo{
-					{
-						ID:    1,
-						Name:  pmodel.NewCIStr("index1"),
-						State: model.StatePublic,
-					},
-				},
-				Columns: []*model.ColumnInfo{
-					{
-						ID: 1,
-					},
-					{
-						ID: 2,
-					},
-				},
+			globalStats: &statistics.Table{
+				ColAndIdxExistenceMap: analyzedMap,
 			},
 			partitionStats: map[refresher.PartitionIDAndName]*statistics.Table{
 				{
@@ -766,22 +738,8 @@ func TestCalculateIndicatorsForPartitions(t *testing.T) {
 		},
 		{
 			name: "No partition meets the threshold",
-			tblInfo: &model.TableInfo{
-				Indices: []*model.IndexInfo{
-					{
-						ID:    1,
-						Name:  pmodel.NewCIStr("index1"),
-						State: model.StatePublic,
-					},
-				},
-				Columns: []*model.ColumnInfo{
-					{
-						ID: 1,
-					},
-					{
-						ID: 2,
-					},
-				},
+			globalStats: &statistics.Table{
+				ColAndIdxExistenceMap: analyzedMap,
 			},
 			partitionStats: map[refresher.PartitionIDAndName]*statistics.Table{
 				{
@@ -855,7 +813,7 @@ func TestCalculateIndicatorsForPartitions(t *testing.T) {
 				gotAvgLastAnalyzeDuration,
 				gotPartitions :=
 				refresher.CalculateIndicatorsForPartitions(
-					tt.tblInfo,
+					tt.globalStats,
 					tt.partitionStats,
 					tt.autoAnalyzeRatio,
 					tt.currentTs,

diff --git a/pkg/statistics/table.go b/pkg/statistics/table.go
@@ -148,6 +148,11 @@ func (m *ColAndIdxExistenceMap) IsEmpty() bool {
 	return len(m.colInfoMap)+len(m.idxInfoMap) == 0
 }
 
+// ColNum returns the number of columns in the map.
+func (m *ColAndIdxExistenceMap) ColNum() int {
+	return len(m.colInfoMap)
+}
+
 // Clone deeply copies the map.
 func (m *ColAndIdxExistenceMap) Clone() *ColAndIdxExistenceMap {
 	mm := NewColAndIndexExistenceMap(len(m.colInfoMap), len(m.idxInfoMap))