mudhorn/prometheus
1
0
Fork 0
mirror of https://github.com/prometheus/prometheus.git synced 2024-11-12 16:44:05 -08:00
Code Issues Actions 7 Packages Projects Releases Wiki Activity
prometheus/CHANGELOG.md
Brian Brazil 259847a6b1
Be smarter in how we look at matchers. (#572) 
* Add unittests for PostingsForMatcher.

* Selector methods are all stateless, don't need a reference.

* Be smarter in how we look at matchers.

Look at all matchers to see if a label can be empty.

Optimise Not handling, so i!="2" is a simple lookup
rather than an inverse postings list.

All all the Withouts together, rather than
having to subtract each from all postings.

Change the pre-expand the postings logic to always do it before doing a
Without only. Don't do that if it's already a list.

The initial goal here was that the oft-seen pattern
i=~"something.+",i!="foo",i!="bar" becomes more efficient.

benchmark                                                            old ns/op     new ns/op     delta
BenchmarkHeadPostingForMatchers/n="1"-4                              5888          6160          +4.62%
BenchmarkHeadPostingForMatchers/n="1",j="foo"-4                      7190          6640          -7.65%
BenchmarkHeadPostingForMatchers/j="foo",n="1"-4                      6038          5923          -1.90%
BenchmarkHeadPostingForMatchers/n="1",j!="foo"-4                     6030884       4850525       -19.57%
BenchmarkHeadPostingForMatchers/i=~".*"-4                            887377940     230329137     -74.04%
BenchmarkHeadPostingForMatchers/i=~".+"-4                            490316101     319931758     -34.75%
BenchmarkHeadPostingForMatchers/i=~""-4                              594961991     130279313     -78.10%
BenchmarkHeadPostingForMatchers/i!=""-4                              537542388     318751015     -40.70%
BenchmarkHeadPostingForMatchers/n="1",i=~".*",j="foo"-4              10460243      8565195       -18.12%
BenchmarkHeadPostingForMatchers/n="1",i=~".*",i!="2",j="foo"-4       44964267      8561546       -80.96%
BenchmarkHeadPostingForMatchers/n="1",i!="",j="foo"-4                42244885      29137737      -31.03%
BenchmarkHeadPostingForMatchers/n="1",i=~".+",j="foo"-4              35285834      32774584      -7.12%
BenchmarkHeadPostingForMatchers/n="1",i=~"1.+",j="foo"-4             8951047       8379024       -6.39%
BenchmarkHeadPostingForMatchers/n="1",i=~".+",i!="2",j="foo"-4       63813335      30672688      -51.93%
BenchmarkHeadPostingForMatchers/n="1",i=~".+",i!~"2.*",j="foo"-4     45381112      44924397      -1.01%

Signed-off-by: Brian Brazil <brian.brazil@robustperception.io>
2019-04-09 11:59:45 +01:00

4 KiB
Raw Blame History

master / unreleased

  • [CHANGE] tsdb now requires golang 1.12 or higher.
  • [REMOVED] chunks.NewReader is removed as it wasn't used anywhere.
  • [REMOVED] FromData is considered unused so was removed.
  • [FEATURE] Added option WALSegmentSize -1 to disable the WAL.
  • [BUGFIX] Bugfix in selectOverlappingDirs. Only return the first overlapping blocks.
  • [BUGFIX] Fsync the meta file to persist it on disk to avoid data loss in case of a host crash.
  • [BUGFIX] Fix fd and vm_area leak on error path in chunks.NewDirReader.
  • [BUGFIX] Fix fd and vm_area leak on error path in index.NewFileReader.
  • [ENHANCEMENT] Be smarter in how we look at matchers.
  • [ENHANCEMENT] PostListings and NotMatcher now public.

0.6.1

  • [BUGFIX] Update last after appending a non-overlapping chunk in chunks.MergeOverlappingChunks. #539

0.6.0

  • [CHANGE] AllowOverlappingBlock is now AllowOverlappingBlocks.

0.5.0

  • [FEATURE] Time-ovelapping blocks are now allowed. #370
    • Disabled by default and can be enabled via AllowOverlappingBlock option.
    • Added MergeChunks function in chunkenc/xor.go to merge 2 time-overlapping chunks.
    • Added MergeOverlappingChunks function in chunks/chunks.go to merge multiple time-overlapping Chunk Metas.
    • Added MinTime and MaxTime method for BlockReader.
  • [FEATURE] New dump command to tsdb tool to dump all samples.
  • [FEATURE] New encoding package for common binary encoding/decoding helpers.
    • Added to remove some code duplication.
  • [ENHANCEMENT] When closing the db any running compaction will be cancelled so it doesn't block.
    • NewLeveledCompactor takes a context.
  • [CHANGE] prometheus_tsdb_storage_blocks_bytes_total is now prometheus_tsdb_storage_blocks_bytes.
  • [BUGFIX] Improved Postings Merge performance. Fixes a regression from the the previous release.
  • [BUGFIX] LiveReader can get into an infinite loop on corrupt WALs.

0.4.0

  • [CHANGE] New WALSegmentSize option to override the DefaultOptions.WALSegmentSize. Added to allow using smaller wal files. For example using tmpfs on a RPI to minimise the SD card wear out from the constant WAL writes. As part of this change the DefaultOptions.WALSegmentSize constant was also exposed.
  • [CHANGE] Empty blocks are not written during compaction #374
  • [FEATURE] Size base retention through Options.MaxBytes. As part of this change:
    • Added new metrics - prometheus_tsdb_storage_blocks_bytes_total, prometheus_tsdb_size_retentions_total, prometheus_tsdb_time_retentions_total
    • New public interface SizeReader: Size() int64
    • OpenBlock signature changed to take a logger.
  • [REMOVED] PrefixMatcher is considered unused so was removed.
  • [CLEANUP] Options.WALFlushInterval is removed as it wasn't used anywhere.
  • [FEATURE] Add new LiveReader to WAL pacakge. Added to allow live tailing of a WAL segment, used by Prometheus Remote Write after refactor. The main difference between the new reader and the existing Reader is that for LiveReader a call to Next() that returns false does not mean that there will never be more data to read.

0.3.1

  • [BUGFIX] Fixed most windows test and some actual bugs for unclosed file readers.

0.3.0

  • [CHANGE] LastCheckpoint() used to return just the segment name and now it returns the full relative path.
  • [CHANGE] NewSegmentsRangeReader() can now read over miltiple wal ranges by using the new SegmentRange{} struct.
  • [CHANGE] CorruptionErr{} now also exposes the Segment Dir which is added when displaying any errors.
  • [CHANGE] Head.Init() is changed to Head.Init(minValidTime int64)
  • [CHANGE] SymbolTable() renamed to SymbolTableSize() to make the name consistent with the Block{ symbolTableSize uint64 } field.
  • [CHANGE] wal.Reader{} now exposes Segment() for the current segment being read and Offset() for the current offset.
  • [FEATURE] tsdbutil analyze subcomand to find churn, high cardinality, etc.