prometheus/model/histogram/histogram.go
Filip Petkovski 10a82f87fd
Enable reusing memory when converting between histogram types
The 'ToFloat' method on integer histograms currently allocates new memory
each time it is called.

This commit adds an optional *FloatHistogram parameter that can be used
to reuse span and bucket slices. It is up to the caller to make sure the
input float histogram is not used anymore after the call.

Signed-off-by: Filip Petkovski <filip.petkovsky@gmail.com>
2023-12-08 10:22:59 +01:00

512 lines
16 KiB
Go

// Copyright 2021 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"
"strings"
"golang.org/x/exp/slices"
)
// CounterResetHint contains the known information about a counter reset,
// or alternatively that we are dealing with a gauge histogram, where counter resets do not apply.
type CounterResetHint byte
const (
UnknownCounterReset CounterResetHint = iota // UnknownCounterReset means we cannot say if this histogram signals a counter reset or not.
CounterReset // CounterReset means there was definitely a counter reset starting from this histogram.
NotCounterReset // NotCounterReset means there was definitely no counter reset with this histogram.
GaugeType // GaugeType means this is a gauge histogram, where counter resets do not happen.
)
// Histogram encodes a sparse, high-resolution histogram. See the design
// document for full details:
// https://docs.google.com/document/d/1cLNv3aufPZb3fNfaJgdaRBZsInZKKIHo9E6HinJVbpM/edit#
//
// The most tricky bit is how bucket indices represent real bucket boundaries.
// An example for schema 0 (by which each bucket is twice as wide as the
// previous bucket):
//
// Bucket boundaries → [-2,-1) [-1,-0.5) [-0.5,-0.25) ... [-0.001,0.001] ... (0.25,0.5] (0.5,1] (1,2] ....
// ↑ ↑ ↑ ↑ ↑ ↑ ↑
// Zero bucket (width e.g. 0.001) → | | | ZB | | |
// Positive bucket indices → | | | ... -1 0 1 2 3
// Negative bucket indices → 3 2 1 0 -1 ...
//
// Which bucket indices are actually used is determined by the spans.
type Histogram struct {
// Counter reset information.
CounterResetHint CounterResetHint
// Currently valid schema numbers are -4 <= n <= 8. They are all for
// base-2 bucket schemas, where 1 is a bucket boundary in each case, and
// then each power of two is divided into 2^n logarithmic buckets. Or
// in other words, each bucket boundary is the previous boundary times
// 2^(2^-n).
Schema int32
// Width of the zero bucket.
ZeroThreshold float64
// Observations falling into the zero bucket.
ZeroCount uint64
// Total number of observations.
Count uint64
// Sum of observations. This is also used as the stale marker.
Sum float64
// Spans for positive and negative buckets (see Span below).
PositiveSpans, NegativeSpans []Span
// Observation counts in buckets. The first element is an absolute
// count. All following ones are deltas relative to the previous
// element.
PositiveBuckets, NegativeBuckets []int64
}
// A Span defines a continuous sequence of buckets.
type Span struct {
// Gap to previous span (always positive), or starting index for the 1st
// span (which can be negative).
Offset int32
// Length of the span.
Length uint32
}
// Copy returns a deep copy of the Histogram.
func (h *Histogram) Copy() *Histogram {
c := *h
if len(h.PositiveSpans) != 0 {
c.PositiveSpans = make([]Span, len(h.PositiveSpans))
copy(c.PositiveSpans, h.PositiveSpans)
}
if len(h.NegativeSpans) != 0 {
c.NegativeSpans = make([]Span, len(h.NegativeSpans))
copy(c.NegativeSpans, h.NegativeSpans)
}
if len(h.PositiveBuckets) != 0 {
c.PositiveBuckets = make([]int64, len(h.PositiveBuckets))
copy(c.PositiveBuckets, h.PositiveBuckets)
}
if len(h.NegativeBuckets) != 0 {
c.NegativeBuckets = make([]int64, len(h.NegativeBuckets))
copy(c.NegativeBuckets, h.NegativeBuckets)
}
return &c
}
// String returns a string representation of the Histogram.
func (h *Histogram) String() string {
var sb strings.Builder
fmt.Fprintf(&sb, "{count:%d, sum:%g", h.Count, h.Sum)
var nBuckets []Bucket[uint64]
for it := h.NegativeBucketIterator(); it.Next(); {
bucket := it.At()
if bucket.Count != 0 {
nBuckets = append(nBuckets, it.At())
}
}
for i := len(nBuckets) - 1; i >= 0; i-- {
fmt.Fprintf(&sb, ", %s", nBuckets[i].String())
}
if h.ZeroCount != 0 {
fmt.Fprintf(&sb, ", %s", h.ZeroBucket().String())
}
for it := h.PositiveBucketIterator(); it.Next(); {
bucket := it.At()
if bucket.Count != 0 {
fmt.Fprintf(&sb, ", %s", bucket.String())
}
}
sb.WriteRune('}')
return sb.String()
}
// ZeroBucket returns the zero bucket.
func (h *Histogram) ZeroBucket() Bucket[uint64] {
return Bucket[uint64]{
Lower: -h.ZeroThreshold,
Upper: h.ZeroThreshold,
LowerInclusive: true,
UpperInclusive: true,
Count: h.ZeroCount,
}
}
// PositiveBucketIterator returns a BucketIterator to iterate over all positive
// buckets in ascending order (starting next to the zero bucket and going up).
func (h *Histogram) PositiveBucketIterator() BucketIterator[uint64] {
it := newRegularBucketIterator(h.PositiveSpans, h.PositiveBuckets, h.Schema, true)
return &it
}
// NegativeBucketIterator returns a BucketIterator to iterate over all negative
// buckets in descending order (starting next to the zero bucket and going down).
func (h *Histogram) NegativeBucketIterator() BucketIterator[uint64] {
it := newRegularBucketIterator(h.NegativeSpans, h.NegativeBuckets, h.Schema, false)
return &it
}
// CumulativeBucketIterator returns a BucketIterator to iterate over a
// cumulative view of the buckets. This method currently only supports
// Histograms without negative buckets and panics if the Histogram has negative
// buckets. It is currently only used for testing.
func (h *Histogram) CumulativeBucketIterator() BucketIterator[uint64] {
if len(h.NegativeBuckets) > 0 {
panic("CumulativeBucketIterator called on Histogram with negative buckets")
}
return &cumulativeBucketIterator{h: h, posSpansIdx: -1}
}
// Equals returns true if the given histogram matches exactly.
// Exact match is when there are no new buckets (even empty) and no missing buckets,
// and all the bucket values match. Spans can have different empty length spans in between,
// but they must represent the same bucket layout to match.
// Sum is compared based on its bit pattern because this method
// is about data equality rather than mathematical equality.
func (h *Histogram) Equals(h2 *Histogram) bool {
if h2 == nil {
return false
}
if h.Schema != h2.Schema || h.ZeroThreshold != h2.ZeroThreshold ||
h.ZeroCount != h2.ZeroCount || h.Count != h2.Count ||
math.Float64bits(h.Sum) != math.Float64bits(h2.Sum) {
return false
}
if !spansMatch(h.PositiveSpans, h2.PositiveSpans) {
return false
}
if !spansMatch(h.NegativeSpans, h2.NegativeSpans) {
return false
}
if !slices.Equal(h.PositiveBuckets, h2.PositiveBuckets) {
return false
}
if !slices.Equal(h.NegativeBuckets, h2.NegativeBuckets) {
return false
}
return true
}
// spansMatch returns true if both spans represent the same bucket layout
// after combining zero length spans with the next non-zero length span.
func spansMatch(s1, s2 []Span) bool {
if len(s1) == 0 && len(s2) == 0 {
return true
}
s1idx, s2idx := 0, 0
for {
if s1idx >= len(s1) {
return allEmptySpans(s2[s2idx:])
}
if s2idx >= len(s2) {
return allEmptySpans(s1[s1idx:])
}
currS1, currS2 := s1[s1idx], s2[s2idx]
s1idx++
s2idx++
if currS1.Length == 0 {
// This span is zero length, so we add consecutive such spans
// until we find a non-zero span.
for ; s1idx < len(s1) && s1[s1idx].Length == 0; s1idx++ {
currS1.Offset += s1[s1idx].Offset
}
if s1idx < len(s1) {
currS1.Offset += s1[s1idx].Offset
currS1.Length = s1[s1idx].Length
s1idx++
}
}
if currS2.Length == 0 {
// This span is zero length, so we add consecutive such spans
// until we find a non-zero span.
for ; s2idx < len(s2) && s2[s2idx].Length == 0; s2idx++ {
currS2.Offset += s2[s2idx].Offset
}
if s2idx < len(s2) {
currS2.Offset += s2[s2idx].Offset
currS2.Length = s2[s2idx].Length
s2idx++
}
}
if currS1.Length == 0 && currS2.Length == 0 {
// The last spans of both set are zero length. Previous spans match.
return true
}
if currS1.Offset != currS2.Offset || currS1.Length != currS2.Length {
return false
}
}
}
func allEmptySpans(s []Span) bool {
for _, ss := range s {
if ss.Length > 0 {
return false
}
}
return true
}
// Compact works like FloatHistogram.Compact. See there for detailed
// explanations.
func (h *Histogram) Compact(maxEmptyBuckets int) *Histogram {
h.PositiveBuckets, h.PositiveSpans = compactBuckets(
h.PositiveBuckets, h.PositiveSpans, maxEmptyBuckets, true,
)
h.NegativeBuckets, h.NegativeSpans = compactBuckets(
h.NegativeBuckets, h.NegativeSpans, maxEmptyBuckets, true,
)
return h
}
// ToFloat returns a FloatHistogram representation of the Histogram. It is a deep
// copy (e.g. spans are not shared). The function accepts a FloatHistogram as an
// argument whose memory will be reused and overwritten if provided. If this
// argument is nil, a new FloatHistogram will be allocated.
func (h *Histogram) ToFloat(fh *FloatHistogram) *FloatHistogram {
if fh == nil {
fh = &FloatHistogram{}
}
fh.CounterResetHint = h.CounterResetHint
fh.Schema = h.Schema
fh.ZeroThreshold = h.ZeroThreshold
fh.ZeroCount = float64(h.ZeroCount)
fh.Count = float64(h.Count)
fh.Sum = h.Sum
fh.PositiveSpans = resize(fh.PositiveSpans, len(h.PositiveSpans))
copy(fh.PositiveSpans, h.PositiveSpans)
fh.NegativeSpans = resize(fh.NegativeSpans, len(h.NegativeSpans))
copy(fh.NegativeSpans, h.NegativeSpans)
fh.PositiveBuckets = resize(fh.PositiveBuckets, len(h.PositiveBuckets))
var currentPositive float64
for i, b := range h.PositiveBuckets {
currentPositive += float64(b)
fh.PositiveBuckets[i] = currentPositive
}
fh.NegativeBuckets = resize(fh.NegativeBuckets, len(h.NegativeBuckets))
var currentNegative float64
for i, b := range h.NegativeBuckets {
currentNegative += float64(b)
fh.NegativeBuckets[i] = currentNegative
}
return fh
}
func resize[T any](items []T, n int) []T {
if len(items) < n {
return make([]T, n)
}
return items[:n]
}
// Validate validates consistency between span and bucket slices. Also, buckets are checked
// against negative values.
// For histograms that have not observed any NaN values (based on IsNaN(h.Sum) check), a
// strict h.Count = nCount + pCount + h.ZeroCount check is performed.
// Otherwise, only a lower bound check will be done (h.Count >= nCount + pCount + h.ZeroCount),
// because NaN observations do not increment the values of buckets (but they do increment
// the total h.Count).
func (h *Histogram) Validate() error {
if err := checkHistogramSpans(h.NegativeSpans, len(h.NegativeBuckets)); err != nil {
return fmt.Errorf("negative side: %w", err)
}
if err := checkHistogramSpans(h.PositiveSpans, len(h.PositiveBuckets)); err != nil {
return fmt.Errorf("positive side: %w", err)
}
var nCount, pCount uint64
err := checkHistogramBuckets(h.NegativeBuckets, &nCount, true)
if err != nil {
return fmt.Errorf("negative side: %w", err)
}
err = checkHistogramBuckets(h.PositiveBuckets, &pCount, true)
if err != nil {
return fmt.Errorf("positive side: %w", err)
}
sumOfBuckets := nCount + pCount + h.ZeroCount
if math.IsNaN(h.Sum) {
if sumOfBuckets > h.Count {
return fmt.Errorf("%d observations found in buckets, but the Count field is %d: %w", sumOfBuckets, h.Count, ErrHistogramCountNotBigEnough)
}
} else {
if sumOfBuckets != h.Count {
return fmt.Errorf("%d observations found in buckets, but the Count field is %d: %w", sumOfBuckets, h.Count, ErrHistogramCountMismatch)
}
}
return nil
}
type regularBucketIterator struct {
baseBucketIterator[uint64, int64]
}
func newRegularBucketIterator(spans []Span, buckets []int64, schema int32, positive bool) regularBucketIterator {
i := baseBucketIterator[uint64, int64]{
schema: schema,
spans: spans,
buckets: buckets,
positive: positive,
}
return regularBucketIterator{i}
}
func (r *regularBucketIterator) Next() bool {
if r.spansIdx >= len(r.spans) {
return false
}
span := r.spans[r.spansIdx]
// Seed currIdx for the first bucket.
if r.bucketsIdx == 0 {
r.currIdx = span.Offset
} else {
r.currIdx++
}
for r.idxInSpan >= span.Length {
// We have exhausted the current span and have to find a new
// one. We'll even handle pathologic spans of length 0.
r.idxInSpan = 0
r.spansIdx++
if r.spansIdx >= len(r.spans) {
return false
}
span = r.spans[r.spansIdx]
r.currIdx += span.Offset
}
r.currCount += r.buckets[r.bucketsIdx]
r.idxInSpan++
r.bucketsIdx++
return true
}
type cumulativeBucketIterator struct {
h *Histogram
posSpansIdx int // Index in h.PositiveSpans we are in. -1 means 0 bucket.
posBucketsIdx int // Index in h.PositiveBuckets.
idxInSpan uint32 // Index in the current span. 0 <= idxInSpan < span.Length.
initialized bool
currIdx int32 // The actual bucket index after decoding from spans.
currUpper float64 // The upper boundary of the current bucket.
currCount int64 // Current non-cumulative count for the current bucket. Does not apply for empty bucket.
currCumulativeCount uint64 // Current "cumulative" count for the current bucket.
// Between 2 spans there could be some empty buckets which
// still needs to be counted for cumulative buckets.
// When we hit the end of a span, we use this to iterate
// through the empty buckets.
emptyBucketCount int32
}
func (c *cumulativeBucketIterator) Next() bool {
if c.posSpansIdx == -1 {
// Zero bucket.
c.posSpansIdx++
if c.h.ZeroCount == 0 {
return c.Next()
}
c.currUpper = c.h.ZeroThreshold
c.currCount = int64(c.h.ZeroCount)
c.currCumulativeCount = uint64(c.currCount)
return true
}
if c.posSpansIdx >= len(c.h.PositiveSpans) {
return false
}
if c.emptyBucketCount > 0 {
// We are traversing through empty buckets at the moment.
c.currUpper = getBound(c.currIdx, c.h.Schema)
c.currIdx++
c.emptyBucketCount--
return true
}
span := c.h.PositiveSpans[c.posSpansIdx]
if c.posSpansIdx == 0 && !c.initialized {
// Initializing.
c.currIdx = span.Offset
// The first bucket is an absolute value and not a delta with Zero bucket.
c.currCount = 0
c.initialized = true
}
c.currCount += c.h.PositiveBuckets[c.posBucketsIdx]
c.currCumulativeCount += uint64(c.currCount)
c.currUpper = getBound(c.currIdx, c.h.Schema)
c.posBucketsIdx++
c.idxInSpan++
c.currIdx++
if c.idxInSpan >= span.Length {
// Move to the next span. This one is done.
c.posSpansIdx++
c.idxInSpan = 0
if c.posSpansIdx < len(c.h.PositiveSpans) {
c.emptyBucketCount = c.h.PositiveSpans[c.posSpansIdx].Offset
}
}
return true
}
func (c *cumulativeBucketIterator) At() Bucket[uint64] {
return Bucket[uint64]{
Upper: c.currUpper,
Lower: math.Inf(-1),
UpperInclusive: true,
LowerInclusive: true,
Count: c.currCumulativeCount,
Index: c.currIdx - 1,
}
}
// ReduceResolution reduces the histogram's spans, buckets into target schema.
// The target schema must be smaller than the current histogram's schema.
func (h *Histogram) ReduceResolution(targetSchema int32) *Histogram {
if targetSchema >= h.Schema {
panic(fmt.Errorf("cannot reduce resolution from schema %d to %d", h.Schema, targetSchema))
}
h.PositiveSpans, h.PositiveBuckets = reduceResolution(
h.PositiveSpans, h.PositiveBuckets, h.Schema, targetSchema, true, true,
)
h.NegativeSpans, h.NegativeBuckets = reduceResolution(
h.NegativeSpans, h.NegativeBuckets, h.Schema, targetSchema, true, true,
)
h.Schema = targetSchema
return h
}