stats: fix building too many buckets (pingcap#5978)

statistics/builder.go

@@ -110,10 +110,13 @@ func BuildColumn(ctx sessionctx.Context, numBuckets, id int64, collector *Sample
 		ndv = count
 	}
 	hg := NewHistogram(id, ndv, collector.NullCount, 0, tp, int(numBuckets))
-	valuesPerBucket := float64(count)/float64(numBuckets) + 1
 
 	// As we use samples to build the histogram, the bucket number and repeat should multiply a factor.
 	sampleFactor := float64(count) / float64(len(samples))
+	// Since bucket count is increased by sampleFactor, so the actual max values per bucket is
+	// floor(valuesPerBucket/sampleFactor)*sampleFactor, which may less than valuesPerBucket,
+	// thus we need to add a sampleFactor to avoid building too many buckets.
+	valuesPerBucket := float64(count)/float64(numBuckets) + sampleFactor
 	ndvFactor := float64(count) / float64(hg.NDV)
 	if ndvFactor > sampleFactor {
 		ndvFactor = sampleFactor

statistics/statistics_test.go

@@ -246,23 +246,23 @@ func (s *testStatisticsSuite) TestBuild(c *C) {
 	c.Check(err, IsNil)
 	checkRepeats(c, col)
 	col.PreCalculateScalar()
-	c.Check(col.Len(), Equals, 232)
+	c.Check(col.Len(), Equals, 226)
 	count := col.equalRowCount(types.NewIntDatum(1000))
 	c.Check(int(count), Equals, 0)
 	count = col.lessRowCount(types.NewIntDatum(1000))
 	c.Check(int(count), Equals, 10000)
 	count = col.lessRowCount(types.NewIntDatum(2000))
-	c.Check(int(count), Equals, 19995)
+	c.Check(int(count), Equals, 19999)
 	count = col.greaterRowCount(types.NewIntDatum(2000))
-	c.Check(int(count), Equals, 80003)
+	c.Check(int(count), Equals, 80000)
 	count = col.lessRowCount(types.NewIntDatum(200000000))
 	c.Check(int(count), Equals, 100000)
 	count = col.greaterRowCount(types.NewIntDatum(200000000))
 	c.Check(count, Equals, 0.0)
 	count = col.equalRowCount(types.NewIntDatum(200000000))
 	c.Check(count, Equals, 0.0)
 	count = col.betweenRowCount(types.NewIntDatum(3000), types.NewIntDatum(3500))
-	c.Check(int(count), Equals, 5008)
+	c.Check(int(count), Equals, 4994)
 	count = col.lessRowCount(types.NewIntDatum(1))
 	c.Check(int(count), Equals, 9)
 
@@ -280,6 +280,7 @@ func (s *testStatisticsSuite) TestBuild(c *C) {
 	col, err = BuildColumn(mock.NewContext(), 256, 2, collectors[0], types.NewFieldType(mysql.TypeLonglong))
 	c.Assert(err, IsNil)
 	checkRepeats(c, col)
+	c.Assert(col.Len(), Equals, 250)
 
 	tblCount, col, _, err := buildIndex(ctx, bucketCount, 1, ast.RecordSet(s.rc))
 	c.Check(err, IsNil)
@@ -497,12 +498,12 @@ func (s *testStatisticsSuite) TestColumnRange(c *C) {
 	ran[0].HighExclude = true
 	count, err = tbl.GetRowCountByColumnRanges(sc, 0, ran)
 	c.Assert(err, IsNil)
-	c.Assert(int(count), Equals, 9994)
+	c.Assert(int(count), Equals, 9998)
 	ran[0].LowExclude = false
 	ran[0].HighExclude = false
 	count, err = tbl.GetRowCountByColumnRanges(sc, 0, ran)
 	c.Assert(err, IsNil)
-	c.Assert(int(count), Equals, 9996)
+	c.Assert(int(count), Equals, 10000)
 	ran[0].LowVal[0] = ran[0].HighVal[0]
 	count, err = tbl.GetRowCountByColumnRanges(sc, 0, ran)
 	c.Assert(err, IsNil)