Add postings list cache to M3DB index

scripts/development/m3_stack/m3dbnode.yml

@@ -41,6 +41,8 @@ db:
   cache:
     series:
       policy: lru
+    postingsList:
+      size: 262144
 
   commitlog:
     flushMaxBytes: 524288

src/cmd/services/m3dbnode/config/cache.go

@@ -22,10 +22,19 @@ package config
 
 import "github.com/m3db/m3/src/dbnode/storage/series"
 
+var (
+	defaultPostingsListCacheSize   = 2 << 17 // 262,144
+	defaultPostingsListCacheRegexp = true
+	defaultPostingsListCacheTerms  = true
+)
+
 // CacheConfigurations is the cache configurations.
 type CacheConfigurations struct {
 	// Series cache policy.
 	Series *SeriesCacheConfiguration `yaml:"series"`
+
+	// PostingsList cache policy.
+	PostingsList *PostingsListCacheConfiguration `yaml:"postingsList"`
 }
 
 // SeriesConfiguration returns the series cache configuration or default
@@ -38,6 +47,16 @@ func (c CacheConfigurations) SeriesConfiguration() SeriesCacheConfiguration {
 	return *c.Series
 }
 
+// PostingsListConfiguration returns the postings list cache configuration
+// or default if none is specified.
+func (c CacheConfigurations) PostingsListConfiguration() PostingsListCacheConfiguration {
+	if c.PostingsList == nil {
+		return PostingsListCacheConfiguration{}
+	}
+
+	return *c.PostingsList
+}
+
 // SeriesCacheConfiguration is the series cache configuration.
 type SeriesCacheConfiguration struct {
 	Policy series.CachePolicy                 `yaml:"policy"`
@@ -50,3 +69,40 @@ type LRUSeriesCachePolicyConfiguration struct {
 	MaxBlocks         uint `yaml:"maxBlocks" validate:"nonzero"`
 	EventsChannelSize uint `yaml:"eventsChannelSize" validate:"nonzero"`
 }
+
+// PostingsListCacheConfiguration is the postings list cache configuration.
+type PostingsListCacheConfiguration struct {
+	Size        *int  `yaml:"size"`
+	CacheRegexp *bool `yaml:"cacheRegexp"`
+	CacheTerms  *bool `yaml:"cacheTerms"`
+}
+
+// SizeOrDefault returns the provided size or the default value is none is
+// provided.
+func (p *PostingsListCacheConfiguration) SizeOrDefault() int {
+	if p.Size == nil {
+		return defaultPostingsListCacheSize
+	}
+
+	return *p.Size
+}
+
+// CacheRegexpOrDefault returns the provided cache regexp configuration value
+// or the default value is none is provided.
+func (p *PostingsListCacheConfiguration) CacheRegexpOrDefault() bool {
+	if p.CacheRegexp == nil {
+		return defaultPostingsListCacheRegexp
+	}
+
+	return *p.CacheRegexp
+}
+
+// CacheTermsOrDefault returns the provided cache terms configuration value
+// or the default value is none is provided.
+func (p *PostingsListCacheConfiguration) CacheTermsOrDefault() bool {
+	if p.CacheTerms == nil {
+		return defaultPostingsListCacheTerms
+	}
+
+	return *p.CacheTerms
+}

src/cmd/services/m3dbnode/config/config_test.go

@@ -40,6 +40,12 @@ db:
       level: info
       file: /var/log/m3dbnode.log
 
+  cache:
+      postingsList:
+          size: 100
+          cacheRegexp: false
+          cacheTerms: false
+
   metrics:
       prometheus:
           handlerPath: /metrics
@@ -381,6 +387,10 @@ func TestConfiguration(t *testing.T) {
   blockRetrieve: null
   cache:
     series: null
+    postingsList:
+      size: 100
+      cacheRegexp: false
+      cacheTerms: false
   fs:
     filePathPrefix: /var/lib/m3db
     writeBufferSize: 65536

src/dbnode/config/m3dbnode-cluster-template.yml

@@ -83,6 +83,8 @@ db:
   cache:
     series:
       policy: lru
+    postingsList:
+      size: 262144
 
   commitlog:
     flushMaxBytes: 524288

src/dbnode/config/m3dbnode-local-etcd.yml

@@ -65,6 +65,8 @@ db:
   cache:
     series:
       policy: lru
+    postingsList:
+      size: 262144
 
   commitlog:
     flushMaxBytes: 524288

src/dbnode/config/m3dbnode-local.yml

@@ -62,6 +62,8 @@ db:
   cache:
     series:
       policy: lru
+    postingsList:
+      size: 262144
 
   commitlog:
     flushMaxBytes: 524288

src/dbnode/integration/setup.go

@@ -166,7 +166,21 @@ func newTestSetup(t *testing.T, opts testOptions, fsOpts fs.Options) (*testSetup
 	if opts.WriteNewSeriesAsync() {
 		indexMode = index.InsertAsync
 	}
-	storageOpts = storageOpts.SetIndexOptions(storageOpts.IndexOptions().SetInsertMode(indexMode))
+
+	plCache, stopReporting, err := index.NewPostingsListCache(10, index.PostingsListCacheOptions{
+		InstrumentOptions: iOpts,
+	})
+	if err != nil {
+		return nil, fmt.Errorf("unable to create postings list cache: %v", err)
+	}
+	// Ok to run immediately since it just closes the background reporting loop. Only ok because
+	// this is a test setup, in production we would want the metrics.
+	stopReporting()
+
+	indexOpts := storageOpts.IndexOptions().
+		SetInsertMode(indexMode).
+		SetPostingsListCache(plCache)
+	storageOpts = storageOpts.SetIndexOptions(indexOpts)
 
 	runtimeOptsMgr := storageOpts.RuntimeOptionsManager()
 	runtimeOpts := runtimeOptsMgr.Get().

src/dbnode/persist/fs/index_read_segments.go

@@ -101,7 +101,8 @@ func ReadIndexSegments(
 			return nil, err
 		}
 
-		seg, err := newPersistentSegment(fileset, fsOpts.FSTOptions())
+		fstOpts := fsOpts.FSTOptions()
+		seg, err := newPersistentSegment(fileset, fstOpts)
 		if err != nil {
 			return nil, err
 		}

src/dbnode/persist/fs/types.go

@@ -313,65 +313,65 @@ type IndexFileSetReader interface {
 	Validate() error
 }
 
-// Options represents the options for filesystem persistence
+// Options represents the options for filesystem persistence.
 type Options interface {
-	// Validate will validate the options and return an error if not valid
+	// Validate will validate the options and return an error if not valid.
 	Validate() error
 
-	// SetClockOptions sets the clock options
+	// SetClockOptions sets the clock options.
 	SetClockOptions(value clock.Options) Options
 
-	// ClockOptions returns the clock options
+	// ClockOptions returns the clock options.
 	ClockOptions() clock.Options
 
-	// SetInstrumentOptions sets the instrumentation options
+	// SetInstrumentOptions sets the instrumentation options.
 	SetInstrumentOptions(value instrument.Options) Options
 
-	// InstrumentOptions returns the instrumentation options
+	// InstrumentOptions returns the instrumentation options.
 	InstrumentOptions() instrument.Options
 
-	// SetRuntimeOptionsManager sets the runtime options manager
+	// SetRuntimeOptionsManager sets the runtime options manager.
 	SetRuntimeOptionsManager(value runtime.OptionsManager) Options
 
-	// RuntimeOptionsManager returns the runtime options manager
+	// RuntimeOptionsManager returns the runtime options manager.
 	RuntimeOptionsManager() runtime.OptionsManager
 
-	// SetDecodingOptions sets the decoding options
+	// SetDecodingOptions sets the decoding options.
 	SetDecodingOptions(value msgpack.DecodingOptions) Options
 
-	// DecodingOptions returns the decoding options
+	// DecodingOptions returns the decoding options.
 	DecodingOptions() msgpack.DecodingOptions
 
-	// SetFilePathPrefix sets the file path prefix for sharded TSDB files
+	// SetFilePathPrefix sets the file path prefix for sharded TSDB files.
 	SetFilePathPrefix(value string) Options
 
-	// FilePathPrefix returns the file path prefix for sharded TSDB files
+	// FilePathPrefix returns the file path prefix for sharded TSDB files.
 	FilePathPrefix() string
 
-	// SetNewFileMode sets the new file mode
+	// SetNewFileMode sets the new file mode.
 	SetNewFileMode(value os.FileMode) Options
 
-	// NewFileMode returns the new file mode
+	// NewFileMode returns the new file mode.
 	NewFileMode() os.FileMode
 
-	// SetNewDirectoryMode sets the new directory mode
+	// SetNewDirectoryMode sets the new directory mode.
 	SetNewDirectoryMode(value os.FileMode) Options
 
-	// NewDirectoryMode returns the new directory mode
+	// NewDirectoryMode returns the new directory mode.
 	NewDirectoryMode() os.FileMode
 
-	// SetIndexSummariesPercent size sets the percent of index summaries to write
+	// SetIndexSummariesPercent size sets the percent of index summaries to write.
 	SetIndexSummariesPercent(value float64) Options
 
-	// IndexSummariesPercent size returns the percent of index summaries to write
+	// IndexSummariesPercent size returns the percent of index summaries to write.
 	IndexSummariesPercent() float64
 
 	// SetIndexBloomFilterFalsePositivePercent size sets the percent of false positive
 	// rate to use for the index bloom filter size and k hashes estimation
 	SetIndexBloomFilterFalsePositivePercent(value float64) Options
 
 	// IndexBloomFilterFalsePositivePercent size returns the percent of false positive
-	// rate to use for the index bloom filter size and k hashes estimation
+	// rate to use for the index bloom filter size and k hashes estimation.
 	IndexBloomFilterFalsePositivePercent() float64
 
 	// SetForceIndexSummariesMmapMemory sets whether the summaries files will be mmap'd
@@ -382,66 +382,66 @@ type Options interface {
 	// as an anonymous region, or as a file.
 	ForceIndexSummariesMmapMemory() bool
 
-	// SetForceBloomFilterMmapMemory sets whether the bloom filters will be mmap'd
+	// SetForceBloomFilterMmapMemory sets whether the bloom filters will be mmap'd.
 	// as an anonymous region, or as a file.
 	SetForceBloomFilterMmapMemory(value bool) Options
 
-	// ForceBloomFilterMmapMemory returns whether the bloom filters will be mmap'd
+	// ForceBloomFilterMmapMemory returns whether the bloom filters will be mmap'd.
 	// as an anonymous region, or as a file.
 	ForceBloomFilterMmapMemory() bool
 
-	// SetWriterBufferSize sets the buffer size for writing TSDB files
+	// SetWriterBufferSize sets the buffer size for writing TSDB files.
 	SetWriterBufferSize(value int) Options
 
-	// WriterBufferSize returns the buffer size for writing TSDB files
+	// WriterBufferSize returns the buffer size for writing TSDB files.
 	WriterBufferSize() int
 
-	// SetInfoReaderBufferSize sets the buffer size for reading TSDB info, digest and checkpoint files
+	// SetInfoReaderBufferSize sets the buffer size for reading TSDB info, digest and checkpoint files.
 	SetInfoReaderBufferSize(value int) Options
 
-	// InfoReaderBufferSize returns the buffer size for reading TSDB info, digest and checkpoint files
+	// InfoReaderBufferSize returns the buffer size for reading TSDB info, digest and checkpoint files.
 	InfoReaderBufferSize() int
 
-	// SetDataReaderBufferSize sets the buffer size for reading TSDB data and index files
+	// SetDataReaderBufferSize sets the buffer size for reading TSDB data and index files.
 	SetDataReaderBufferSize(value int) Options
 
-	// DataReaderBufferSize returns the buffer size for reading TSDB data and index files
+	// DataReaderBufferSize returns the buffer size for reading TSDB data and index files.
 	DataReaderBufferSize() int
 
-	// SetSeekReaderBufferSize size sets the buffer size for seeking TSDB files
+	// SetSeekReaderBufferSize size sets the buffer size for seeking TSDB files.
 	SetSeekReaderBufferSize(value int) Options
 
-	// SeekReaderBufferSize size returns the buffer size for seeking TSDB files
+	// SeekReaderBufferSize size returns the buffer size for seeking TSDB files.
 	SeekReaderBufferSize() int
 
-	// SetMmapEnableHugeTLB sets whether mmap huge pages are enabled when running on linux
+	// SetMmapEnableHugeTLB sets whether mmap huge pages are enabled when running on linux.
 	SetMmapEnableHugeTLB(value bool) Options
 
-	// MmapEnableHugeTLB returns whether mmap huge pages are enabled when running on linux
+	// MmapEnableHugeTLB returns whether mmap huge pages are enabled when running on linux.
 	MmapEnableHugeTLB() bool
 
-	// SetMmapHugeTLBThreshold sets the threshold when to use mmap huge pages for mmap'd files on linux
+	// SetMmapHugeTLBThreshold sets the threshold when to use mmap huge pages for mmap'd files on linux.
 	SetMmapHugeTLBThreshold(value int64) Options
 
-	// MmapHugeTLBThreshold returns the threshold when to use mmap huge pages for mmap'd files on linux
+	// MmapHugeTLBThreshold returns the threshold when to use mmap huge pages for mmap'd files on linux.
 	MmapHugeTLBThreshold() int64
 
-	// SetTagEncoderPool sets the tag encoder pool
+	// SetTagEncoderPool sets the tag encoder pool.
 	SetTagEncoderPool(value serialize.TagEncoderPool) Options
 
-	// TagEncoderPool returns the tag encoder pool
+	// TagEncoderPool returns the tag encoder pool.
 	TagEncoderPool() serialize.TagEncoderPool
 
-	// SetTagDecoderPool sets the tag decoder pool
+	// SetTagDecoderPool sets the tag decoder pool.
 	SetTagDecoderPool(value serialize.TagDecoderPool) Options
 
-	// TagDecoderPool returns the tag decoder pool
+	// TagDecoderPool returns the tag decoder pool.
 	TagDecoderPool() serialize.TagDecoderPool
 
-	// SetFStOptions sets the fst options
+	// SetFStOptions sets the fst options.
 	SetFSTOptions(value fst.Options) Options
 
-	// FSTOptions returns the fst options
+	// FSTOptions returns the fst options.
 	FSTOptions() fst.Options
 }