dedup

Documentation

Index

• func NewBoundedSeriesIterator(it chunkenc.Iterator, mint, maxt int64) *boundedSeriesIterator
• func NewChunkSeriesMerger() storage.VerticalChunkSeriesMergeFunc
• func NewOverlapSplit(set storepb.SeriesSet) storepb.SeriesSet
• func NewSeriesSet(set storage.SeriesSet, f string) storage.SeriesSet

Functions

func NewBoundedSeriesIterator

func NewBoundedSeriesIterator(it chunkenc.Iterator, mint, maxt int64) *boundedSeriesIterator

func NewChunkSeriesMerger

func NewChunkSeriesMerger() storage.VerticalChunkSeriesMergeFunc

NewChunkSeriesMerger merges several chunk series into one. Deduplication is based on penalty based deduplication algorithm without handling counter reset.

func NewOverlapSplit

func NewOverlapSplit(set storepb.SeriesSet) storepb.SeriesSet

NewOverlapSplit splits overlapping chunks into separate series entry, so existing algorithm can work as usual. We cannot do this in dedup.SeriesSet as it iterates over samples already. TODO(bwplotka): Remove when we move to per chunk deduplication code. We expect non-duplicated series with sorted chunks by min time (possibly overlapped).

func NewSeriesSet

func NewSeriesSet(set storage.SeriesSet, f string) storage.SeriesSet

NewSeriesSet returns seriesSet that deduplicates the same series. The series in series set are expected be sorted by all labels.