2020-03-24 13:15:47 -07:00
|
|
|
// Copyright 2020 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 storage
|
|
|
|
|
|
|
|
import (
|
2020-06-24 06:41:52 -07:00
|
|
|
"math"
|
2020-03-24 13:15:47 -07:00
|
|
|
"sort"
|
|
|
|
|
Style cleanup of all the changes in sparsehistogram so far
A lot of this code was hacked together, literally during a
hackathon. This commit intends not to change the code substantially,
but just make the code obey the usual style practices.
A (possibly incomplete) list of areas:
* Generally address linter warnings.
* The `pgk` directory is deprecated as per dev-summit. No new packages should
be added to it. I moved the new `pkg/histogram` package to `model`
anticipating what's proposed in #9478.
* Make the naming of the Sparse Histogram more consistent. Including
abbreviations, there were just too many names for it: SparseHistogram,
Histogram, Histo, hist, his, shs, h. The idea is to call it "Histogram" in
general. Only add "Sparse" if it is needed to avoid confusion with
conventional Histograms (which is rare because the TSDB really has no notion
of conventional Histograms). Use abbreviations only in local scope, and then
really abbreviate (not just removing three out of seven letters like in
"Histo"). This is in the spirit of
https://github.com/golang/go/wiki/CodeReviewComments#variable-names
* Several other minor name changes.
* A lot of formatting of doc comments. For one, following
https://github.com/golang/go/wiki/CodeReviewComments#comment-sentences
, but also layout question, anticipating how things will look like
when rendered by `godoc` (even where `godoc` doesn't render them
right now because they are for unexported types or not a doc comment
at all but just a normal code comment - consistency is queen!).
* Re-enabled `TestQueryLog` and `TestEndopints` (they pass now,
leaving them disabled was presumably an oversight).
* Bucket iterator for histogram.Histogram is now created with a
method.
* HistogramChunk.iterator now allows iterator recycling. (I think
@dieterbe only commented it out because he was confused by the
question in the comment.)
* HistogramAppender.Append panics now because we decided to treat
staleness marker differently.
Signed-off-by: beorn7 <beorn@grafana.com>
2021-10-09 06:57:07 -07:00
|
|
|
"github.com/prometheus/prometheus/model/histogram"
|
2020-03-24 13:15:47 -07:00
|
|
|
"github.com/prometheus/prometheus/pkg/labels"
|
|
|
|
"github.com/prometheus/prometheus/tsdb/chunkenc"
|
|
|
|
"github.com/prometheus/prometheus/tsdb/chunks"
|
|
|
|
"github.com/prometheus/prometheus/tsdb/tsdbutil"
|
|
|
|
)
|
|
|
|
|
2020-07-31 08:03:02 -07:00
|
|
|
type SeriesEntry struct {
|
|
|
|
Lset labels.Labels
|
|
|
|
SampleIteratorFn func() chunkenc.Iterator
|
|
|
|
}
|
|
|
|
|
|
|
|
func (s *SeriesEntry) Labels() labels.Labels { return s.Lset }
|
|
|
|
func (s *SeriesEntry) Iterator() chunkenc.Iterator { return s.SampleIteratorFn() }
|
|
|
|
|
|
|
|
type ChunkSeriesEntry struct {
|
|
|
|
Lset labels.Labels
|
|
|
|
ChunkIteratorFn func() chunks.Iterator
|
|
|
|
}
|
|
|
|
|
|
|
|
func (s *ChunkSeriesEntry) Labels() labels.Labels { return s.Lset }
|
|
|
|
func (s *ChunkSeriesEntry) Iterator() chunks.Iterator { return s.ChunkIteratorFn() }
|
|
|
|
|
|
|
|
// NewListSeries returns series entry with iterator that allows to iterate over provided samples.
|
|
|
|
func NewListSeries(lset labels.Labels, s []tsdbutil.Sample) *SeriesEntry {
|
|
|
|
return &SeriesEntry{
|
|
|
|
Lset: lset,
|
|
|
|
SampleIteratorFn: func() chunkenc.Iterator {
|
|
|
|
return NewListSeriesIterator(samples(s))
|
|
|
|
},
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2020-08-06 12:27:37 -07:00
|
|
|
// NewListChunkSeriesFromSamples returns chunk series entry that allows to iterate over provided samples.
|
2020-07-31 08:03:02 -07:00
|
|
|
// NOTE: It uses inefficient chunks encoding implementation, not caring about chunk size.
|
|
|
|
func NewListChunkSeriesFromSamples(lset labels.Labels, samples ...[]tsdbutil.Sample) *ChunkSeriesEntry {
|
|
|
|
return &ChunkSeriesEntry{
|
|
|
|
Lset: lset,
|
|
|
|
ChunkIteratorFn: func() chunks.Iterator {
|
|
|
|
chks := make([]chunks.Meta, 0, len(samples))
|
|
|
|
for _, s := range samples {
|
|
|
|
chks = append(chks, tsdbutil.ChunkFromSamples(s))
|
|
|
|
}
|
|
|
|
return NewListChunkSeriesIterator(chks...)
|
|
|
|
},
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2020-03-24 13:15:47 -07:00
|
|
|
type listSeriesIterator struct {
|
2020-06-24 06:41:52 -07:00
|
|
|
samples Samples
|
2020-03-24 13:15:47 -07:00
|
|
|
idx int
|
|
|
|
}
|
|
|
|
|
2020-06-24 06:41:52 -07:00
|
|
|
type samples []tsdbutil.Sample
|
|
|
|
|
|
|
|
func (s samples) Get(i int) tsdbutil.Sample { return s[i] }
|
|
|
|
func (s samples) Len() int { return len(s) }
|
|
|
|
|
|
|
|
// Samples interface allows to work on arrays of types that are compatible with tsdbutil.Sample.
|
|
|
|
type Samples interface {
|
|
|
|
Get(i int) tsdbutil.Sample
|
|
|
|
Len() int
|
|
|
|
}
|
|
|
|
|
2020-07-31 08:03:02 -07:00
|
|
|
// NewListSeriesIterator returns listSeriesIterator that allows to iterate over provided samples.
|
2020-06-24 06:41:52 -07:00
|
|
|
func NewListSeriesIterator(samples Samples) chunkenc.Iterator {
|
2020-03-24 13:15:47 -07:00
|
|
|
return &listSeriesIterator{samples: samples, idx: -1}
|
|
|
|
}
|
|
|
|
|
|
|
|
func (it *listSeriesIterator) At() (int64, float64) {
|
2020-06-24 06:41:52 -07:00
|
|
|
s := it.samples.Get(it.idx)
|
2020-03-24 13:15:47 -07:00
|
|
|
return s.T(), s.V()
|
|
|
|
}
|
|
|
|
|
Style cleanup of all the changes in sparsehistogram so far
A lot of this code was hacked together, literally during a
hackathon. This commit intends not to change the code substantially,
but just make the code obey the usual style practices.
A (possibly incomplete) list of areas:
* Generally address linter warnings.
* The `pgk` directory is deprecated as per dev-summit. No new packages should
be added to it. I moved the new `pkg/histogram` package to `model`
anticipating what's proposed in #9478.
* Make the naming of the Sparse Histogram more consistent. Including
abbreviations, there were just too many names for it: SparseHistogram,
Histogram, Histo, hist, his, shs, h. The idea is to call it "Histogram" in
general. Only add "Sparse" if it is needed to avoid confusion with
conventional Histograms (which is rare because the TSDB really has no notion
of conventional Histograms). Use abbreviations only in local scope, and then
really abbreviate (not just removing three out of seven letters like in
"Histo"). This is in the spirit of
https://github.com/golang/go/wiki/CodeReviewComments#variable-names
* Several other minor name changes.
* A lot of formatting of doc comments. For one, following
https://github.com/golang/go/wiki/CodeReviewComments#comment-sentences
, but also layout question, anticipating how things will look like
when rendered by `godoc` (even where `godoc` doesn't render them
right now because they are for unexported types or not a doc comment
at all but just a normal code comment - consistency is queen!).
* Re-enabled `TestQueryLog` and `TestEndopints` (they pass now,
leaving them disabled was presumably an oversight).
* Bucket iterator for histogram.Histogram is now created with a
method.
* HistogramChunk.iterator now allows iterator recycling. (I think
@dieterbe only commented it out because he was confused by the
question in the comment.)
* HistogramAppender.Append panics now because we decided to treat
staleness marker differently.
Signed-off-by: beorn7 <beorn@grafana.com>
2021-10-09 06:57:07 -07:00
|
|
|
// AtHistogram always returns (0, histogram.Histogram{}) because there is no
|
|
|
|
// support for histogram values yet.
|
|
|
|
func (it *listSeriesIterator) AtHistogram() (int64, histogram.Histogram) {
|
|
|
|
return 0, histogram.Histogram{}
|
2021-06-29 07:38:46 -07:00
|
|
|
}
|
|
|
|
|
2021-06-30 07:48:13 -07:00
|
|
|
func (it *listSeriesIterator) ChunkEncoding() chunkenc.Encoding {
|
|
|
|
return chunkenc.EncXOR
|
|
|
|
}
|
|
|
|
|
2020-03-24 13:15:47 -07:00
|
|
|
func (it *listSeriesIterator) Next() bool {
|
|
|
|
it.idx++
|
2020-06-24 06:41:52 -07:00
|
|
|
return it.idx < it.samples.Len()
|
2020-03-24 13:15:47 -07:00
|
|
|
}
|
|
|
|
|
|
|
|
func (it *listSeriesIterator) Seek(t int64) bool {
|
|
|
|
if it.idx == -1 {
|
|
|
|
it.idx = 0
|
|
|
|
}
|
|
|
|
// Do binary search between current position and end.
|
2020-06-24 06:41:52 -07:00
|
|
|
it.idx = sort.Search(it.samples.Len()-it.idx, func(i int) bool {
|
|
|
|
s := it.samples.Get(i + it.idx)
|
2020-03-24 13:15:47 -07:00
|
|
|
return s.T() >= t
|
|
|
|
})
|
|
|
|
|
2020-06-24 06:41:52 -07:00
|
|
|
return it.idx < it.samples.Len()
|
2020-03-24 13:15:47 -07:00
|
|
|
}
|
|
|
|
|
|
|
|
func (it *listSeriesIterator) Err() error { return nil }
|
|
|
|
|
|
|
|
type listChunkSeriesIterator struct {
|
|
|
|
chks []chunks.Meta
|
2020-07-31 08:03:02 -07:00
|
|
|
idx int
|
2020-03-24 13:15:47 -07:00
|
|
|
}
|
|
|
|
|
2020-07-31 08:03:02 -07:00
|
|
|
// NewListChunkSeriesIterator returns listChunkSeriesIterator that allows to iterate over provided chunks.
|
2020-03-24 13:15:47 -07:00
|
|
|
func NewListChunkSeriesIterator(chks ...chunks.Meta) chunks.Iterator {
|
|
|
|
return &listChunkSeriesIterator{chks: chks, idx: -1}
|
|
|
|
}
|
|
|
|
|
|
|
|
func (it *listChunkSeriesIterator) At() chunks.Meta {
|
|
|
|
return it.chks[it.idx]
|
|
|
|
}
|
|
|
|
|
|
|
|
func (it *listChunkSeriesIterator) Next() bool {
|
|
|
|
it.idx++
|
|
|
|
return it.idx < len(it.chks)
|
|
|
|
}
|
|
|
|
|
|
|
|
func (it *listChunkSeriesIterator) Err() error { return nil }
|
|
|
|
|
|
|
|
type chunkSetToSeriesSet struct {
|
|
|
|
ChunkSeriesSet
|
|
|
|
|
|
|
|
chkIterErr error
|
|
|
|
sameSeriesChunks []Series
|
|
|
|
}
|
|
|
|
|
|
|
|
// NewSeriesSetFromChunkSeriesSet converts ChunkSeriesSet to SeriesSet by decoding chunks one by one.
|
|
|
|
func NewSeriesSetFromChunkSeriesSet(chk ChunkSeriesSet) SeriesSet {
|
|
|
|
return &chunkSetToSeriesSet{ChunkSeriesSet: chk}
|
|
|
|
}
|
|
|
|
|
|
|
|
func (c *chunkSetToSeriesSet) Next() bool {
|
|
|
|
if c.Err() != nil || !c.ChunkSeriesSet.Next() {
|
|
|
|
return false
|
|
|
|
}
|
|
|
|
|
|
|
|
iter := c.ChunkSeriesSet.At().Iterator()
|
|
|
|
c.sameSeriesChunks = c.sameSeriesChunks[:0]
|
|
|
|
|
|
|
|
for iter.Next() {
|
2020-07-31 08:03:02 -07:00
|
|
|
c.sameSeriesChunks = append(
|
|
|
|
c.sameSeriesChunks,
|
|
|
|
newChunkToSeriesDecoder(c.ChunkSeriesSet.At().Labels(), iter.At()),
|
|
|
|
)
|
2020-03-24 13:15:47 -07:00
|
|
|
}
|
|
|
|
|
|
|
|
if iter.Err() != nil {
|
|
|
|
c.chkIterErr = iter.Err()
|
|
|
|
return false
|
|
|
|
}
|
|
|
|
return true
|
|
|
|
}
|
|
|
|
|
|
|
|
func (c *chunkSetToSeriesSet) At() Series {
|
|
|
|
// Series composed of same chunks for the same series.
|
|
|
|
return ChainedSeriesMerge(c.sameSeriesChunks...)
|
|
|
|
}
|
|
|
|
|
|
|
|
func (c *chunkSetToSeriesSet) Err() error {
|
|
|
|
if c.chkIterErr != nil {
|
|
|
|
return c.chkIterErr
|
|
|
|
}
|
|
|
|
return c.ChunkSeriesSet.Err()
|
|
|
|
}
|
|
|
|
|
2020-07-31 08:03:02 -07:00
|
|
|
func newChunkToSeriesDecoder(labels labels.Labels, chk chunks.Meta) Series {
|
|
|
|
return &SeriesEntry{
|
|
|
|
Lset: labels,
|
|
|
|
SampleIteratorFn: func() chunkenc.Iterator {
|
|
|
|
// TODO(bwplotka): Can we provide any chunkenc buffer?
|
|
|
|
return chk.Chunk.Iterator(nil)
|
|
|
|
},
|
|
|
|
}
|
2020-03-24 13:15:47 -07:00
|
|
|
}
|
|
|
|
|
2020-06-24 06:41:52 -07:00
|
|
|
type seriesSetToChunkSet struct {
|
|
|
|
SeriesSet
|
|
|
|
}
|
|
|
|
|
|
|
|
// NewSeriesSetToChunkSet converts SeriesSet to ChunkSeriesSet by encoding chunks from samples.
|
|
|
|
func NewSeriesSetToChunkSet(chk SeriesSet) ChunkSeriesSet {
|
|
|
|
return &seriesSetToChunkSet{SeriesSet: chk}
|
|
|
|
}
|
|
|
|
|
|
|
|
func (c *seriesSetToChunkSet) Next() bool {
|
|
|
|
if c.Err() != nil || !c.SeriesSet.Next() {
|
|
|
|
return false
|
|
|
|
}
|
|
|
|
return true
|
|
|
|
}
|
|
|
|
|
|
|
|
func (c *seriesSetToChunkSet) At() ChunkSeries {
|
2021-05-19 05:01:35 -07:00
|
|
|
return NewSeriesToChunkEncoder(c.SeriesSet.At())
|
2020-06-24 06:41:52 -07:00
|
|
|
}
|
|
|
|
|
|
|
|
func (c *seriesSetToChunkSet) Err() error {
|
|
|
|
return c.SeriesSet.Err()
|
|
|
|
}
|
|
|
|
|
|
|
|
type seriesToChunkEncoder struct {
|
|
|
|
Series
|
|
|
|
}
|
|
|
|
|
2021-05-18 09:37:16 -07:00
|
|
|
const seriesToChunkEncoderSplit = 120
|
|
|
|
|
2021-05-19 05:01:35 -07:00
|
|
|
// NewSeriesToChunkEncoder encodes samples to chunks with 120 samples limit.
|
|
|
|
func NewSeriesToChunkEncoder(series Series) ChunkSeries {
|
|
|
|
return &seriesToChunkEncoder{series}
|
|
|
|
}
|
|
|
|
|
2020-06-24 06:41:52 -07:00
|
|
|
func (s *seriesToChunkEncoder) Iterator() chunks.Iterator {
|
|
|
|
chk := chunkenc.NewXORChunk()
|
|
|
|
app, err := chk.Appender()
|
|
|
|
if err != nil {
|
|
|
|
return errChunksIterator{err: err}
|
|
|
|
}
|
|
|
|
mint := int64(math.MaxInt64)
|
|
|
|
maxt := int64(math.MinInt64)
|
|
|
|
|
2021-05-18 09:37:16 -07:00
|
|
|
chks := []chunks.Meta{}
|
|
|
|
|
|
|
|
i := 0
|
2020-06-24 06:41:52 -07:00
|
|
|
seriesIter := s.Series.Iterator()
|
|
|
|
for seriesIter.Next() {
|
2021-05-18 09:37:16 -07:00
|
|
|
// Create a new chunk if too many samples in the current one.
|
|
|
|
if i >= seriesToChunkEncoderSplit {
|
|
|
|
chks = append(chks, chunks.Meta{
|
|
|
|
MinTime: mint,
|
|
|
|
MaxTime: maxt,
|
|
|
|
Chunk: chk,
|
|
|
|
})
|
|
|
|
chk = chunkenc.NewXORChunk()
|
|
|
|
app, err = chk.Appender()
|
|
|
|
if err != nil {
|
|
|
|
return errChunksIterator{err: err}
|
|
|
|
}
|
|
|
|
mint = int64(math.MaxInt64)
|
|
|
|
// maxt is immediately overwritten below which is why setting it here won't make a difference.
|
|
|
|
i = 0
|
|
|
|
}
|
|
|
|
|
2020-06-24 06:41:52 -07:00
|
|
|
t, v := seriesIter.At()
|
|
|
|
app.Append(t, v)
|
|
|
|
|
|
|
|
maxt = t
|
|
|
|
if mint == math.MaxInt64 {
|
|
|
|
mint = t
|
|
|
|
}
|
2021-05-18 09:37:16 -07:00
|
|
|
i++
|
2020-06-24 06:41:52 -07:00
|
|
|
}
|
|
|
|
if err := seriesIter.Err(); err != nil {
|
|
|
|
return errChunksIterator{err: err}
|
|
|
|
}
|
|
|
|
|
2021-05-18 09:37:16 -07:00
|
|
|
chks = append(chks, chunks.Meta{
|
2020-06-24 06:41:52 -07:00
|
|
|
MinTime: mint,
|
|
|
|
MaxTime: maxt,
|
|
|
|
Chunk: chk,
|
|
|
|
})
|
2021-05-18 09:37:16 -07:00
|
|
|
|
|
|
|
return NewListChunkSeriesIterator(chks...)
|
2020-06-24 06:41:52 -07:00
|
|
|
}
|
|
|
|
|
|
|
|
type errChunksIterator struct {
|
|
|
|
err error
|
|
|
|
}
|
|
|
|
|
|
|
|
func (e errChunksIterator) At() chunks.Meta { return chunks.Meta{} }
|
|
|
|
func (e errChunksIterator) Next() bool { return false }
|
|
|
|
func (e errChunksIterator) Err() error { return e.err }
|
2020-07-31 08:03:02 -07:00
|
|
|
|
|
|
|
// ExpandSamples iterates over all samples in the iterator, buffering all in slice.
|
|
|
|
// Optionally it takes samples constructor, useful when you want to compare sample slices with different
|
|
|
|
// sample implementations. if nil, sample type from this package will be used.
|
|
|
|
func ExpandSamples(iter chunkenc.Iterator, newSampleFn func(t int64, v float64) tsdbutil.Sample) ([]tsdbutil.Sample, error) {
|
|
|
|
if newSampleFn == nil {
|
|
|
|
newSampleFn = func(t int64, v float64) tsdbutil.Sample { return sample{t, v} }
|
|
|
|
}
|
|
|
|
|
|
|
|
var result []tsdbutil.Sample
|
|
|
|
for iter.Next() {
|
|
|
|
t, v := iter.At()
|
|
|
|
// NaNs can't be compared normally, so substitute for another value.
|
|
|
|
if math.IsNaN(v) {
|
|
|
|
v = -42
|
|
|
|
}
|
|
|
|
result = append(result, newSampleFn(t, v))
|
|
|
|
}
|
|
|
|
return result, iter.Err()
|
|
|
|
}
|
|
|
|
|
|
|
|
// ExpandChunks iterates over all chunks in the iterator, buffering all in slice.
|
|
|
|
func ExpandChunks(iter chunks.Iterator) ([]chunks.Meta, error) {
|
|
|
|
var result []chunks.Meta
|
|
|
|
for iter.Next() {
|
|
|
|
result = append(result, iter.At())
|
|
|
|
}
|
|
|
|
return result, iter.Err()
|
|
|
|
}
|