indexheader

package
v0.26.0 Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: Apr 9, 2022 License: Apache-2.0 Imports: 27 Imported by: 0

Documentation

Index

Constants

View Source 
const (
	// BinaryFormatV1 represents first version of index-header file.
	BinaryFormatV1 = 1

	// MagicIndex are 4 bytes at the head of an index-header file.
	MagicIndex = 0xBAAAD792
)

Variables

View Source 
var NotFoundRangeErr = errors.New("range not found")

NotFoundRangeErr is an error returned by PostingsOffset when there is no posting for given name and value pairs.

Functions

func WriteBinary 

func WriteBinary(ctx context.Context, bkt objstore.BucketReader, id ulid.ULID, filename string) (err error)

WriteBinary build index-header file from the pieces of index in object storage.

Types

type BinaryReader 

type BinaryReader struct {
	// contains filtered or unexported fields
}

func NewBinaryReader 

func NewBinaryReader(ctx context.Context, logger log.Logger, bkt objstore.BucketReader, dir string, id ulid.ULID, postingOffsetsInMemSampling int) (*BinaryReader, error)

NewBinaryReader loads or builds new index-header if not present on disk.

func (*BinaryReader) Close 

func (r *BinaryReader) Close() error

func (*BinaryReader) IndexVersion 

func (r *BinaryReader) IndexVersion() (int, error)

func (*BinaryReader) LabelNames 

func (r *BinaryReader) LabelNames() ([]string, error)

func (*BinaryReader) LabelValues 

func (r *BinaryReader) LabelValues(name string) ([]string, error)

func (*BinaryReader) LookupSymbol 

func (r *BinaryReader) LookupSymbol(o uint32) (string, error)

func (*BinaryReader) PostingsOffset 

func (r *BinaryReader) PostingsOffset(name, value string) (index.Range, error)

TODO(bwplotka): Get advantage of multi value offset fetch.

type BinaryTOC 

type BinaryTOC struct {
	// Symbols holds start to the same symbols section as index related to this index header.
	Symbols uint64
	// PostingsOffsetTable holds start to the same Postings Offset Table section as index related to this index header.
	PostingsOffsetTable uint64
}

BinaryTOC is a table of content for index-header file.

type FileWriter 

type FileWriter struct {
	// contains filtered or unexported fields
}

func NewFileWriter 

func NewFileWriter(name string, size int) (*FileWriter, error)

TODO(bwplotka): Added size to method, upstream this.

func (*FileWriter) AddPadding 

func (fw *FileWriter) AddPadding(size int) error

AddPadding adds zero byte padding until the file size is a multiple size.

func (*FileWriter) Close 

func (fw *FileWriter) Close() error

func (*FileWriter) Flush 

func (fw *FileWriter) Flush() error

func (*FileWriter) Pos 

func (fw *FileWriter) Pos() uint64

func (*FileWriter) Remove 

func (fw *FileWriter) Remove() error

func (*FileWriter) Write 

func (fw *FileWriter) Write(bufs ...[]byte) error

func (*FileWriter) WriteAt 

func (fw *FileWriter) WriteAt(buf []byte, pos uint64) error

type LazyBinaryReader 

type LazyBinaryReader struct {
	// contains filtered or unexported fields
}

LazyBinaryReader wraps BinaryReader and loads (mmap) the index-header only upon the first Reader function is called.

func NewLazyBinaryReader 

func NewLazyBinaryReader(
	ctx context.Context,
	logger log.Logger,
	bkt objstore.BucketReader,
	dir string,
	id ulid.ULID,
	postingOffsetsInMemSampling int,
	metrics *LazyBinaryReaderMetrics,
	onClosed func(*LazyBinaryReader),
) (*LazyBinaryReader, error)

NewLazyBinaryReader makes a new LazyBinaryReader. If the index-header does not exist on the local disk at dir location, this function will build it downloading required sections from the full index stored in the bucket. However, this function doesn't load (mmap) the index-header; it will be loaded at first Reader function call.

func (*LazyBinaryReader) Close 

func (r *LazyBinaryReader) Close() error

Close implements Reader. It unloads the index-header from memory (releasing the mmap area), but a subsequent call to any other Reader function will automatically reload it.

func (*LazyBinaryReader) IndexVersion 

func (r *LazyBinaryReader) IndexVersion() (int, error)

IndexVersion implements Reader.

func (*LazyBinaryReader) LabelNames 

func (r *LazyBinaryReader) LabelNames() ([]string, error)

LabelNames implements Reader.

func (*LazyBinaryReader) LabelValues 

func (r *LazyBinaryReader) LabelValues(name string) ([]string, error)

LabelValues implements Reader.

func (*LazyBinaryReader) LookupSymbol 

func (r *LazyBinaryReader) LookupSymbol(o uint32) (string, error)

LookupSymbol implements Reader.

func (*LazyBinaryReader) PostingsOffset 

func (r *LazyBinaryReader) PostingsOffset(name, value string) (index.Range, error)

PostingsOffset implements Reader.

type LazyBinaryReaderMetrics 

type LazyBinaryReaderMetrics struct {
	// contains filtered or unexported fields
}

LazyBinaryReaderMetrics holds metrics tracked by LazyBinaryReader.

func NewLazyBinaryReaderMetrics 

func NewLazyBinaryReaderMetrics(reg prometheus.Registerer) *LazyBinaryReaderMetrics

NewLazyBinaryReaderMetrics makes new LazyBinaryReaderMetrics.

type Reader 

type Reader interface {
	io.Closer

	// IndexVersion returns version of index.
	IndexVersion() (int, error)

	// PostingsOffset returns start and end offsets of postings for given name and value.
	// The end offset might be bigger than the actual posting ending, but not larger than the whole index file.
	// NotFoundRangeErr is returned when no index can be found for given name and value.
	// TODO(bwplotka): Move to PostingsOffsets(name string, value ...string) []index.Range and benchmark.
	PostingsOffset(name string, value string) (index.Range, error)

	// LookupSymbol returns string based on given reference.
	// Error is return if the symbol can't be found.
	LookupSymbol(o uint32) (string, error)

	// LabelValues returns all label values for given label name or error.
	// If no values are found for label name, or label name does not exists,
	// then empty string is returned and no error.
	LabelValues(name string) ([]string, error)

	// LabelNames returns all label names in sorted order.
	LabelNames() ([]string, error)
}

Reader is an interface allowing to read essential, minimal number of index fields from the small portion of index file called header.

type ReaderPool 

type ReaderPool struct {
	// contains filtered or unexported fields
}

ReaderPool is used to istantiate new index-header readers and keep track of them. When the lazy reader is enabled, the pool keeps track of all instantiated readers and automatically close them once the idle timeout is reached. A closed lazy reader will be automatically re-opened upon next usage.

func NewReaderPool 

func NewReaderPool(logger log.Logger, lazyReaderEnabled bool, lazyReaderIdleTimeout time.Duration, metrics *ReaderPoolMetrics) *ReaderPool

NewReaderPool makes a new ReaderPool.

func (*ReaderPool) Close 

func (p *ReaderPool) Close()

Close the pool and stop checking for idle readers. No reader tracked by this pool will be closed. It's the caller responsibility to close readers.

func (*ReaderPool) NewBinaryReader 

func (p *ReaderPool) NewBinaryReader(ctx context.Context, logger log.Logger, bkt objstore.BucketReader, dir string, id ulid.ULID, postingOffsetsInMemSampling int) (Reader, error)

NewBinaryReader creates and returns a new binary reader. If the pool has been configured with lazy reader enabled, this function will return a lazy reader. The returned lazy reader is tracked by the pool and automatically closed once the idle timeout expires.

type ReaderPoolMetrics 

type ReaderPoolMetrics struct {
	// contains filtered or unexported fields
}

ReaderPoolMetrics holds metrics tracked by ReaderPool.

func NewReaderPoolMetrics 

func NewReaderPoolMetrics(reg prometheus.Registerer) *ReaderPoolMetrics

NewReaderPoolMetrics makes new ReaderPoolMetrics.

Source Files

View all Source files

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.