Move histogram validation code to model/histogram

Signed-off-by: Linas Medziunas <linas.medziunas@gmail.com>
This commit is contained in:
Linas Medziunas 2023-10-13 10:58:48 +03:00
parent 1cd6c1cde5
commit 1f8aea11d6
7 changed files with 377 additions and 301 deletions

View file

@ -0,0 +1,52 @@
// Copyright 2023 The Prometheus Authors
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package histogram
// GenerateBigTestHistograms generates a slice of histograms with given number of buckets each.
func GenerateBigTestHistograms(numHistograms, numBuckets int) []*Histogram {
numSpans := numBuckets / 10
bucketsPerSide := numBuckets / 2
spanLength := uint32(bucketsPerSide / numSpans)
// Given all bucket deltas are 1, sum bucketsPerSide + 1.
observationCount := bucketsPerSide * (1 + bucketsPerSide)
var histograms []*Histogram
for i := 0; i < numHistograms; i++ {
h := &Histogram{
Count: uint64(i + observationCount),
ZeroCount: uint64(i),
ZeroThreshold: 1e-128,
Sum: 18.4 * float64(i+1),
Schema: 2,
NegativeSpans: make([]Span, numSpans),
PositiveSpans: make([]Span, numSpans),
NegativeBuckets: make([]int64, bucketsPerSide),
PositiveBuckets: make([]int64, bucketsPerSide),
}
for j := 0; j < numSpans; j++ {
s := Span{Offset: 1, Length: spanLength}
h.NegativeSpans[j] = s
h.PositiveSpans[j] = s
}
for j := 0; j < bucketsPerSide; j++ {
h.NegativeBuckets[j] = 1
h.PositiveBuckets[j] = 1
}
histograms = append(histograms, h)
}
return histograms
}

136
model/histogram/validate.go Normal file
View file

@ -0,0 +1,136 @@
// Copyright 2023 The Prometheus Authors
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package histogram
import (
"fmt"
"math"
"github.com/pkg/errors"
)
var (
ErrHistogramCountNotBigEnough = errors.New("histogram's observation count should be at least the number of observations found in the buckets")
ErrHistogramCountMismatch = errors.New("histogram's observation count should equal the number of observations found in the buckets (in absence of NaN)")
ErrHistogramNegativeBucketCount = errors.New("histogram has a bucket whose observation count is negative")
ErrHistogramSpanNegativeOffset = errors.New("histogram has a span whose offset is negative")
ErrHistogramSpansBucketsMismatch = errors.New("histogram spans specify different number of buckets than provided")
)
func ValidateHistogram(h *Histogram) error {
if err := checkHistogramSpans(h.NegativeSpans, len(h.NegativeBuckets)); err != nil {
return errors.Wrap(err, "negative side")
}
if err := checkHistogramSpans(h.PositiveSpans, len(h.PositiveBuckets)); err != nil {
return errors.Wrap(err, "positive side")
}
var nCount, pCount uint64
err := checkHistogramBuckets(h.NegativeBuckets, &nCount, true)
if err != nil {
return errors.Wrap(err, "negative side")
}
err = checkHistogramBuckets(h.PositiveBuckets, &pCount, true)
if err != nil {
return errors.Wrap(err, "positive side")
}
sumOfBuckets := nCount + pCount + h.ZeroCount
if math.IsNaN(h.Sum) {
if sumOfBuckets > h.Count {
return errors.Wrap(
ErrHistogramCountNotBigEnough,
fmt.Sprintf("%d observations found in buckets, but the Count field is %d", sumOfBuckets, h.Count),
)
}
} else {
if sumOfBuckets != h.Count {
return errors.Wrap(
ErrHistogramCountMismatch,
fmt.Sprintf("%d observations found in buckets, but the Count field is %d", sumOfBuckets, h.Count),
)
}
}
return nil
}
func ValidateFloatHistogram(h *FloatHistogram) error {
if err := checkHistogramSpans(h.NegativeSpans, len(h.NegativeBuckets)); err != nil {
return errors.Wrap(err, "negative side")
}
if err := checkHistogramSpans(h.PositiveSpans, len(h.PositiveBuckets)); err != nil {
return errors.Wrap(err, "positive side")
}
var nCount, pCount float64
err := checkHistogramBuckets(h.NegativeBuckets, &nCount, false)
if err != nil {
return errors.Wrap(err, "negative side")
}
err = checkHistogramBuckets(h.PositiveBuckets, &pCount, false)
if err != nil {
return errors.Wrap(err, "positive side")
}
// We do not check for h.Count being at least as large as the sum of the
// counts in the buckets because floating point precision issues can
// create false positives here.
return nil
}
func checkHistogramSpans(spans []Span, numBuckets int) error {
var spanBuckets int
for n, span := range spans {
if n > 0 && span.Offset < 0 {
return errors.Wrap(
ErrHistogramSpanNegativeOffset,
fmt.Sprintf("span number %d with offset %d", n+1, span.Offset),
)
}
spanBuckets += int(span.Length)
}
if spanBuckets != numBuckets {
return errors.Wrap(
ErrHistogramSpansBucketsMismatch,
fmt.Sprintf("spans need %d buckets, have %d buckets", spanBuckets, numBuckets),
)
}
return nil
}
func checkHistogramBuckets[BC BucketCount, IBC InternalBucketCount](buckets []IBC, count *BC, deltas bool) error {
if len(buckets) == 0 {
return nil
}
var last IBC
for i := 0; i < len(buckets); i++ {
var c IBC
if deltas {
c = last + buckets[i]
} else {
c = buckets[i]
}
if c < 0 {
return errors.Wrap(
ErrHistogramNegativeBucketCount,
fmt.Sprintf("bucket number %d has observation count of %v", i+1, c),
)
}
last = c
*count += BC(c)
}
return nil
}

View file

@ -0,0 +1,175 @@
// Copyright 2023 The Prometheus Authors
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package histogram
import (
"math"
"testing"
"github.com/stretchr/testify/require"
)
func TestHistogramValidation(t *testing.T) {
tests := map[string]struct {
h *Histogram
errMsg string
skipFloat bool
}{
"valid histogram": {
h: &Histogram{
Count: 12,
ZeroCount: 2,
ZeroThreshold: 0.001,
Sum: 19.4,
Schema: 1,
PositiveSpans: []Span{
{Offset: 0, Length: 2},
{Offset: 1, Length: 2},
},
PositiveBuckets: []int64{1, 1, -1, 0},
NegativeSpans: []Span{
{Offset: 0, Length: 2},
{Offset: 1, Length: 2},
},
NegativeBuckets: []int64{1, 1, -1, 0},
},
},
"valid histogram with NaN observations that has its Count (4) higher than the actual total of buckets (2 + 1)": {
// This case is possible if NaN values (which do not fall into any bucket) are observed.
h: &Histogram{
ZeroCount: 2,
Count: 4,
Sum: math.NaN(),
PositiveSpans: []Span{{Offset: 0, Length: 1}},
PositiveBuckets: []int64{1},
},
},
"rejects histogram without NaN observations that has its Count (4) higher than the actual total of buckets (2 + 1)": {
h: &Histogram{
ZeroCount: 2,
Count: 4,
Sum: 333,
PositiveSpans: []Span{{Offset: 0, Length: 1}},
PositiveBuckets: []int64{1},
},
errMsg: `3 observations found in buckets, but the Count field is 4: histogram's observation count should equal the number of observations found in the buckets (in absence of NaN)`,
skipFloat: true,
},
"rejects histogram that has too few negative buckets": {
h: &Histogram{
NegativeSpans: []Span{{Offset: 0, Length: 1}},
NegativeBuckets: []int64{},
},
errMsg: `negative side: spans need 1 buckets, have 0 buckets: histogram spans specify different number of buckets than provided`,
},
"rejects histogram that has too few positive buckets": {
h: &Histogram{
PositiveSpans: []Span{{Offset: 0, Length: 1}},
PositiveBuckets: []int64{},
},
errMsg: `positive side: spans need 1 buckets, have 0 buckets: histogram spans specify different number of buckets than provided`,
},
"rejects histogram that has too many negative buckets": {
h: &Histogram{
NegativeSpans: []Span{{Offset: 0, Length: 1}},
NegativeBuckets: []int64{1, 2},
},
errMsg: `negative side: spans need 1 buckets, have 2 buckets: histogram spans specify different number of buckets than provided`,
},
"rejects histogram that has too many positive buckets": {
h: &Histogram{
PositiveSpans: []Span{{Offset: 0, Length: 1}},
PositiveBuckets: []int64{1, 2},
},
errMsg: `positive side: spans need 1 buckets, have 2 buckets: histogram spans specify different number of buckets than provided`,
},
"rejects a histogram that has a negative span with a negative offset": {
h: &Histogram{
NegativeSpans: []Span{{Offset: -1, Length: 1}, {Offset: -1, Length: 1}},
NegativeBuckets: []int64{1, 2},
},
errMsg: `negative side: span number 2 with offset -1: histogram has a span whose offset is negative`,
},
"rejects a histogram which has a positive span with a negative offset": {
h: &Histogram{
PositiveSpans: []Span{{Offset: -1, Length: 1}, {Offset: -1, Length: 1}},
PositiveBuckets: []int64{1, 2},
},
errMsg: `positive side: span number 2 with offset -1: histogram has a span whose offset is negative`,
},
"rejects a histogram that has a negative bucket with a negative count": {
h: &Histogram{
NegativeSpans: []Span{{Offset: -1, Length: 1}},
NegativeBuckets: []int64{-1},
},
errMsg: `negative side: bucket number 1 has observation count of -1: histogram has a bucket whose observation count is negative`,
},
"rejects a histogram that has a positive bucket with a negative count": {
h: &Histogram{
PositiveSpans: []Span{{Offset: -1, Length: 1}},
PositiveBuckets: []int64{-1},
},
errMsg: `positive side: bucket number 1 has observation count of -1: histogram has a bucket whose observation count is negative`,
},
"rejects a histogram that has a lower count than count in buckets": {
h: &Histogram{
Count: 0,
NegativeSpans: []Span{{Offset: -1, Length: 1}},
PositiveSpans: []Span{{Offset: -1, Length: 1}},
NegativeBuckets: []int64{1},
PositiveBuckets: []int64{1},
},
errMsg: `2 observations found in buckets, but the Count field is 0: histogram's observation count should equal the number of observations found in the buckets (in absence of NaN)`,
skipFloat: true,
},
"rejects a histogram that doesn't count the zero bucket in its count": {
h: &Histogram{
Count: 2,
ZeroCount: 1,
NegativeSpans: []Span{{Offset: -1, Length: 1}},
PositiveSpans: []Span{{Offset: -1, Length: 1}},
NegativeBuckets: []int64{1},
PositiveBuckets: []int64{1},
},
errMsg: `3 observations found in buckets, but the Count field is 2: histogram's observation count should equal the number of observations found in the buckets (in absence of NaN)`,
skipFloat: true,
},
}
for testName, tc := range tests {
t.Run(testName, func(t *testing.T) {
if err := ValidateHistogram(tc.h); tc.errMsg != "" {
require.EqualError(t, err, tc.errMsg)
} else {
require.NoError(t, err)
}
if tc.skipFloat {
return
}
if err := ValidateFloatHistogram(tc.h.ToFloat()); tc.errMsg != "" {
require.EqualError(t, err, tc.errMsg)
} else {
require.NoError(t, err)
}
})
}
}
func BenchmarkHistogramValidation(b *testing.B) {
histograms := GenerateBigTestHistograms(b.N, 500)
b.ResetTimer()
for _, h := range histograms {
require.NoError(b, ValidateHistogram(h))
}
}

View file

@ -37,17 +37,12 @@ var (
// ErrTooOldSample is when out of order support is enabled but the sample is outside the time window allowed.
ErrTooOldSample = errors.New("too old sample")
// ErrDuplicateSampleForTimestamp is when the sample has same timestamp but different value.
ErrDuplicateSampleForTimestamp = errors.New("duplicate sample for timestamp")
ErrOutOfOrderExemplar = errors.New("out of order exemplar")
ErrDuplicateExemplar = errors.New("duplicate exemplar")
ErrExemplarLabelLength = fmt.Errorf("label length for exemplar exceeds maximum of %d UTF-8 characters", exemplar.ExemplarMaxLabelSetLength)
ErrExemplarsDisabled = fmt.Errorf("exemplar storage is disabled or max exemplars is less than or equal to 0")
ErrNativeHistogramsDisabled = fmt.Errorf("native histograms are disabled")
ErrHistogramCountNotBigEnough = errors.New("histogram's observation count should be at least the number of observations found in the buckets")
ErrHistogramCountMismatch = errors.New("histogram's observation count should equal the number of observations found in the buckets (in absence of NaN)")
ErrHistogramNegativeBucketCount = errors.New("histogram has a bucket whose observation count is negative")
ErrHistogramSpanNegativeOffset = errors.New("histogram has a span whose offset is negative")
ErrHistogramSpansBucketsMismatch = errors.New("histogram spans specify different number of buckets than provided")
ErrDuplicateSampleForTimestamp = errors.New("duplicate sample for timestamp")
ErrOutOfOrderExemplar = errors.New("out of order exemplar")
ErrDuplicateExemplar = errors.New("duplicate exemplar")
ErrExemplarLabelLength = fmt.Errorf("label length for exemplar exceeds maximum of %d UTF-8 characters", exemplar.ExemplarMaxLabelSetLength)
ErrExemplarsDisabled = fmt.Errorf("exemplar storage is disabled or max exemplars is less than or equal to 0")
ErrNativeHistogramsDisabled = fmt.Errorf("native histograms are disabled")
)
// SeriesRef is a generic series reference. In prometheus it is either a

View file

@ -883,13 +883,13 @@ func (a *appender) AppendExemplar(ref storage.SeriesRef, _ labels.Labels, e exem
func (a *appender) AppendHistogram(ref storage.SeriesRef, l labels.Labels, t int64, h *histogram.Histogram, fh *histogram.FloatHistogram) (storage.SeriesRef, error) {
if h != nil {
if err := tsdb.ValidateHistogram(h); err != nil {
if err := histogram.ValidateHistogram(h); err != nil {
return 0, err
}
}
if fh != nil {
if err := tsdb.ValidateFloatHistogram(fh); err != nil {
if err := histogram.ValidateFloatHistogram(fh); err != nil {
return 0, err
}
}

View file

@ -521,13 +521,13 @@ func (a *headAppender) AppendHistogram(ref storage.SeriesRef, lset labels.Labels
}
if h != nil {
if err := ValidateHistogram(h); err != nil {
if err := histogram.ValidateHistogram(h); err != nil {
return 0, err
}
}
if fh != nil {
if err := ValidateFloatHistogram(fh); err != nil {
if err := histogram.ValidateFloatHistogram(fh); err != nil {
return 0, err
}
}
@ -642,113 +642,6 @@ func (a *headAppender) UpdateMetadata(ref storage.SeriesRef, lset labels.Labels,
return ref, nil
}
func ValidateHistogram(h *histogram.Histogram) error {
if err := checkHistogramSpans(h.NegativeSpans, len(h.NegativeBuckets)); err != nil {
return errors.Wrap(err, "negative side")
}
if err := checkHistogramSpans(h.PositiveSpans, len(h.PositiveBuckets)); err != nil {
return errors.Wrap(err, "positive side")
}
var nCount, pCount uint64
err := checkHistogramBuckets(h.NegativeBuckets, &nCount, true)
if err != nil {
return errors.Wrap(err, "negative side")
}
err = checkHistogramBuckets(h.PositiveBuckets, &pCount, true)
if err != nil {
return errors.Wrap(err, "positive side")
}
sumOfBuckets := nCount + pCount + h.ZeroCount
if math.IsNaN(h.Sum) {
if sumOfBuckets > h.Count {
return errors.Wrap(
storage.ErrHistogramCountNotBigEnough,
fmt.Sprintf("%d observations found in buckets, but the Count field is %d", sumOfBuckets, h.Count),
)
}
} else {
if sumOfBuckets != h.Count {
return errors.Wrap(
storage.ErrHistogramCountMismatch,
fmt.Sprintf("%d observations found in buckets, but the Count field is %d", sumOfBuckets, h.Count),
)
}
}
return nil
}
func ValidateFloatHistogram(h *histogram.FloatHistogram) error {
if err := checkHistogramSpans(h.NegativeSpans, len(h.NegativeBuckets)); err != nil {
return errors.Wrap(err, "negative side")
}
if err := checkHistogramSpans(h.PositiveSpans, len(h.PositiveBuckets)); err != nil {
return errors.Wrap(err, "positive side")
}
var nCount, pCount float64
err := checkHistogramBuckets(h.NegativeBuckets, &nCount, false)
if err != nil {
return errors.Wrap(err, "negative side")
}
err = checkHistogramBuckets(h.PositiveBuckets, &pCount, false)
if err != nil {
return errors.Wrap(err, "positive side")
}
// We do not check for h.Count being at least as large as the sum of the
// counts in the buckets because floating point precision issues can
// create false positives here.
return nil
}
func checkHistogramSpans(spans []histogram.Span, numBuckets int) error {
var spanBuckets int
for n, span := range spans {
if n > 0 && span.Offset < 0 {
return errors.Wrap(
storage.ErrHistogramSpanNegativeOffset,
fmt.Sprintf("span number %d with offset %d", n+1, span.Offset),
)
}
spanBuckets += int(span.Length)
}
if spanBuckets != numBuckets {
return errors.Wrap(
storage.ErrHistogramSpansBucketsMismatch,
fmt.Sprintf("spans need %d buckets, have %d buckets", spanBuckets, numBuckets),
)
}
return nil
}
func checkHistogramBuckets[BC histogram.BucketCount, IBC histogram.InternalBucketCount](buckets []IBC, count *BC, deltas bool) error {
if len(buckets) == 0 {
return nil
}
var last IBC
for i := 0; i < len(buckets); i++ {
var c IBC
if deltas {
c = last + buckets[i]
} else {
c = buckets[i]
}
if c < 0 {
return errors.Wrap(
storage.ErrHistogramNegativeBucketCount,
fmt.Sprintf("bucket number %d has observation count of %v", i+1, c),
)
}
last = c
*count += BC(c)
}
return nil
}
var _ storage.GetRef = &headAppender{}
func (a *headAppender) GetRef(lset labels.Labels, hash uint64) (storage.SeriesRef, labels.Labels) {

View file

@ -4894,181 +4894,6 @@ func TestReplayAfterMmapReplayError(t *testing.T) {
require.NoError(t, h.Close())
}
func TestHistogramValidation(t *testing.T) {
tests := map[string]struct {
h *histogram.Histogram
errMsg string
skipFloat bool
}{
"valid histogram": {
h: tsdbutil.GenerateTestHistograms(1)[0],
},
"valid histogram with NaN observations that has its Count (4) higher than the actual total of buckets (2 + 1)": {
// This case is possible if NaN values (which do not fall into any bucket) are observed.
h: &histogram.Histogram{
ZeroCount: 2,
Count: 4,
Sum: math.NaN(),
PositiveSpans: []histogram.Span{{Offset: 0, Length: 1}},
PositiveBuckets: []int64{1},
},
},
"rejects histogram without NaN observations that has its Count (4) higher than the actual total of buckets (2 + 1)": {
h: &histogram.Histogram{
ZeroCount: 2,
Count: 4,
Sum: 333,
PositiveSpans: []histogram.Span{{Offset: 0, Length: 1}},
PositiveBuckets: []int64{1},
},
errMsg: `3 observations found in buckets, but the Count field is 4: histogram's observation count should equal the number of observations found in the buckets (in absence of NaN)`,
skipFloat: true,
},
"rejects histogram that has too few negative buckets": {
h: &histogram.Histogram{
NegativeSpans: []histogram.Span{{Offset: 0, Length: 1}},
NegativeBuckets: []int64{},
},
errMsg: `negative side: spans need 1 buckets, have 0 buckets: histogram spans specify different number of buckets than provided`,
},
"rejects histogram that has too few positive buckets": {
h: &histogram.Histogram{
PositiveSpans: []histogram.Span{{Offset: 0, Length: 1}},
PositiveBuckets: []int64{},
},
errMsg: `positive side: spans need 1 buckets, have 0 buckets: histogram spans specify different number of buckets than provided`,
},
"rejects histogram that has too many negative buckets": {
h: &histogram.Histogram{
NegativeSpans: []histogram.Span{{Offset: 0, Length: 1}},
NegativeBuckets: []int64{1, 2},
},
errMsg: `negative side: spans need 1 buckets, have 2 buckets: histogram spans specify different number of buckets than provided`,
},
"rejects histogram that has too many positive buckets": {
h: &histogram.Histogram{
PositiveSpans: []histogram.Span{{Offset: 0, Length: 1}},
PositiveBuckets: []int64{1, 2},
},
errMsg: `positive side: spans need 1 buckets, have 2 buckets: histogram spans specify different number of buckets than provided`,
},
"rejects a histogram that has a negative span with a negative offset": {
h: &histogram.Histogram{
NegativeSpans: []histogram.Span{{Offset: -1, Length: 1}, {Offset: -1, Length: 1}},
NegativeBuckets: []int64{1, 2},
},
errMsg: `negative side: span number 2 with offset -1: histogram has a span whose offset is negative`,
},
"rejects a histogram which has a positive span with a negative offset": {
h: &histogram.Histogram{
PositiveSpans: []histogram.Span{{Offset: -1, Length: 1}, {Offset: -1, Length: 1}},
PositiveBuckets: []int64{1, 2},
},
errMsg: `positive side: span number 2 with offset -1: histogram has a span whose offset is negative`,
},
"rejects a histogram that has a negative bucket with a negative count": {
h: &histogram.Histogram{
NegativeSpans: []histogram.Span{{Offset: -1, Length: 1}},
NegativeBuckets: []int64{-1},
},
errMsg: `negative side: bucket number 1 has observation count of -1: histogram has a bucket whose observation count is negative`,
},
"rejects a histogram that has a positive bucket with a negative count": {
h: &histogram.Histogram{
PositiveSpans: []histogram.Span{{Offset: -1, Length: 1}},
PositiveBuckets: []int64{-1},
},
errMsg: `positive side: bucket number 1 has observation count of -1: histogram has a bucket whose observation count is negative`,
},
"rejects a histogram that has a lower count than count in buckets": {
h: &histogram.Histogram{
Count: 0,
NegativeSpans: []histogram.Span{{Offset: -1, Length: 1}},
PositiveSpans: []histogram.Span{{Offset: -1, Length: 1}},
NegativeBuckets: []int64{1},
PositiveBuckets: []int64{1},
},
errMsg: `2 observations found in buckets, but the Count field is 0: histogram's observation count should equal the number of observations found in the buckets (in absence of NaN)`,
skipFloat: true,
},
"rejects a histogram that doesn't count the zero bucket in its count": {
h: &histogram.Histogram{
Count: 2,
ZeroCount: 1,
NegativeSpans: []histogram.Span{{Offset: -1, Length: 1}},
PositiveSpans: []histogram.Span{{Offset: -1, Length: 1}},
NegativeBuckets: []int64{1},
PositiveBuckets: []int64{1},
},
errMsg: `3 observations found in buckets, but the Count field is 2: histogram's observation count should equal the number of observations found in the buckets (in absence of NaN)`,
skipFloat: true,
},
}
for testName, tc := range tests {
t.Run(testName, func(t *testing.T) {
if err := ValidateHistogram(tc.h); tc.errMsg != "" {
require.EqualError(t, err, tc.errMsg)
} else {
require.NoError(t, err)
}
if tc.skipFloat {
return
}
if err := ValidateFloatHistogram(tc.h.ToFloat()); tc.errMsg != "" {
require.EqualError(t, err, tc.errMsg)
} else {
require.NoError(t, err)
}
})
}
}
func BenchmarkHistogramValidation(b *testing.B) {
histograms := generateBigTestHistograms(b.N, 500)
b.ResetTimer()
for _, h := range histograms {
require.NoError(b, ValidateHistogram(h))
}
}
func generateBigTestHistograms(numHistograms, numBuckets int) []*histogram.Histogram {
numSpans := numBuckets / 10
bucketsPerSide := numBuckets / 2
spanLength := uint32(bucketsPerSide / numSpans)
// Given all bucket deltas are 1, sum bucketsPerSide + 1.
observationCount := bucketsPerSide * (1 + bucketsPerSide)
var histograms []*histogram.Histogram
for i := 0; i < numHistograms; i++ {
h := &histogram.Histogram{
Count: uint64(i + observationCount),
ZeroCount: uint64(i),
ZeroThreshold: 1e-128,
Sum: 18.4 * float64(i+1),
Schema: 2,
NegativeSpans: make([]histogram.Span, numSpans),
PositiveSpans: make([]histogram.Span, numSpans),
NegativeBuckets: make([]int64, bucketsPerSide),
PositiveBuckets: make([]int64, bucketsPerSide),
}
for j := 0; j < numSpans; j++ {
s := histogram.Span{Offset: 1, Length: spanLength}
h.NegativeSpans[j] = s
h.PositiveSpans[j] = s
}
for j := 0; j < bucketsPerSide; j++ {
h.NegativeBuckets[j] = 1
h.PositiveBuckets[j] = 1
}
histograms = append(histograms, h)
}
return histograms
}
func TestOOOAppendWithNoSeries(t *testing.T) {
dir := t.TempDir()
wal, err := wlog.NewSize(nil, nil, filepath.Join(dir, "wal"), 32768, wlog.CompressionSnappy)
@ -5409,7 +5234,7 @@ func BenchmarkCuttingHeadHistogramChunks(b *testing.B) {
numSamples = 50000
numBuckets = 100
)
samples := generateBigTestHistograms(numSamples, numBuckets)
samples := histogram.GenerateBigTestHistograms(numSamples, numBuckets)
h, _ := newTestHead(b, DefaultBlockDuration, wlog.CompressionNone, false)
defer func() {
@ -5473,7 +5298,7 @@ func TestCuttingNewHeadChunks(t *testing.T) {
"small histograms": {
numTotalSamples: 240,
histValFunc: func() func(i int) *histogram.Histogram {
hists := generateBigTestHistograms(240, 10)
hists := histogram.GenerateBigTestHistograms(240, 10)
return func(i int) *histogram.Histogram {
return hists[i]
}
@ -5489,7 +5314,7 @@ func TestCuttingNewHeadChunks(t *testing.T) {
"large histograms": {
numTotalSamples: 240,
histValFunc: func() func(i int) *histogram.Histogram {
hists := generateBigTestHistograms(240, 100)
hists := histogram.GenerateBigTestHistograms(240, 100)
return func(i int) *histogram.Histogram {
return hists[i]
}
@ -5512,7 +5337,7 @@ func TestCuttingNewHeadChunks(t *testing.T) {
// per chunk.
numTotalSamples: 11,
histValFunc: func() func(i int) *histogram.Histogram {
hists := generateBigTestHistograms(11, 100000)
hists := histogram.GenerateBigTestHistograms(11, 100000)
return func(i int) *histogram.Histogram {
return hists[i]
}