tsdb

package
v0.9.4-rc1 Latest Latest
Warning

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

Go to latest
Published: Sep 10, 2015 License: MIT Imports: 29 Imported by: 0

README

Line Protocol

The line protocol is a text based format for writing points to InfluxDB. Each line defines a single point. Multiple lines must be separated by the newline character \n. The format of the line consists of three parts:

[key] [fields] [timestamp]

Each section is separated by spaces. The minimum required point consists of a measurement name and at least one field. Points without a specified timestamp will be written using the server's local timestamp. Timestamps are assumed to be in nanoseconds unless a precision value is passed in the query string.

Key

The key is the measurement name and any optional tags separated by commas. Measurement names, tag keys, and tag values must escape any spaces or commas using a backslash (\). For example: \ and \,. All tag values are stored as strings and should not be surrounded in quotes.

Tags should be sorted by key before being sent for best performance. The sort should match that from the Go bytes.Compare function (http://golang.org/pkg/bytes/#Compare).

Examples
# measurement only
cpu

# measurment and tags
cpu,host=serverA,region=us-west

# measurment with commas
cpu\,01,host=serverA,region=us-west

# tag value with spaces
cpu,host=server\ A,region=us\ west

Fields

Fields are key-value metrics associated with the measurement. Every line must have at least one field. Multiple fields must be separated with commas and not spaces.

Field keys are always strings and follow the same syntactical rules as described above for tag keys and values. Field values can be one of four types. The first value written for a given field on a given measurement defines the type of that field for all series under that measurement.

  • integer - Numeric values that do not include a decimal and are followed by a trailing i when inserted (e.g. 1i, 345i, 2015i, -10i). Note that all values must have a trailing i. If they do not they will be written as floats.
  • float - Numeric values tha are not followed by a trailing i. (e.g. 1, 1.0, -3.14, 6.0+e5, 10).
  • boolean - A value indicating true or false. Valid boolean strings are (t, T, true, TRUE, f, F, false, and FALSE).
  • string - A text value. All string values must be surrounded in double-quotes ". If the string contains a double-quote or backslashes, it must be escaped with a backslash, e.g. \", \\.
# integer value
cpu value=1i

cpu value=1.1i # will result in a parse error

# float value
cpu_load value=1

cpu_load value=1.0

cpu_load value=1.2

# boolean value
error fatal=true

# string value
event msg="logged out"

# multiple values
cpu load=10,alert=true,reason="value above maximum threshold"

Timestamp

The timestamp section is optional but should be specified if possible. The value is an integer representing nanoseconds since the epoch. If the timestamp is not provided the point will inherit the server's local timestamp.

Some write APIs allow passing a lower precision. If the API supports a lower precision, the timestamp may also be an integer epoch in microseconds, milliseconds, seconds, minutes or hours.

Full Example

A full example is shown below.

cpu,host=server01,region=uswest value=1 1434055562000000000
cpu,host=server02,region=uswest value=3 1434055562000010000

In this example the first line shows a measurement of "cpu", there are two tags "host" and "region, the value is 1.0, and the timestamp is 1434055562000000000. Following this is a second line, also a point in the measurement "cpu" but belonging to a different "host".

cpu,host=server\ 01,region=uswest value=1,msg="all systems nominal"
cpu,host=server\ 01,region=us\,west value_int=1i

In these examples, the "host" is set to server 01. The field value associated with field key msg is double-quoted, as it is a string. The second example shows a region of us,west with the comma properly escaped. In the first example value is written as a floating point number. In the second, value_int is an integer.

Distributed Queries

Documentation

Overview

Package tsdb implements a durable time series database.

Index

Constants

View Source
const (
	// DefaultMaxWALSize is the default size of the WAL before it is flushed.
	DefaultMaxWALSize = 100 * 1024 * 1024 // 100MB

	// DefaultWALFlushInterval is the frequency the WAL will get flushed if
	// it doesn't reach its size threshold.
	DefaultWALFlushInterval = 10 * time.Minute

	// DefaultWALPartitionFlushDelay is the sleep time between WAL partition flushes.
	DefaultWALPartitionFlushDelay = 2 * time.Second

	// DefaultReadySeriesSize of 32KB specifies when a series is eligible to be flushed
	DefaultReadySeriesSize = 30 * 1024

	// DefaultCompactionThreshold flush and compact a partition once this ratio of keys are over the flush size
	DefaultCompactionThreshold = 0.5

	// DefaultMaxSeriesSize specifies the size at which a series will be forced to flush
	DefaultMaxSeriesSize = 1024 * 1024

	// DefaultFlushColdInterval specifies how long after a partition has been cold
	// for writes that a full flush and compaction are forced
	DefaultFlushColdInterval = 5 * time.Second

	// DefaultParititionSizeThreshold specifies when a partition gets to this size in
	// memory, we should slow down writes until it gets a chance to compact.
	// This will force clients to get backpressure if they're writing too fast. We need
	// this because the WAL can take writes much faster than the index. So eventually
	// we'll need to create backpressure, otherwise we'll fill up the memory and die.
	// This number multiplied by the parition count is roughly the max possible memory
	// size for the in-memory WAL cache.
	DefaultPartitionSizeThreshold = 20 * 1024 * 1024 // 20MB
)
View Source
const (
	// Return an error if the user is trying to select more than this number of points in a group by statement.
	// Most likely they specified a group by interval without time boundaries.
	MaxGroupByPoints = 100000

	// Since time is always selected, the column count when selecting only a single other value will be 2
	SelectColumnCountWithOneValue = 2

	// IgnoredChunkSize is what gets passed into Mapper.Begin for aggregate queries as they don't chunk points out
	IgnoredChunkSize = 0
)
View Source
const DefaultEngine = "bz1"

DefaultEngine is the default engine used by the shard when initializing.

Variables

View Source
var (
	// ErrInvalidQuery is returned when executing an unknown query type.
	ErrInvalidQuery = errors.New("invalid query")

	// ErrNotExecuted is returned when a statement is not executed in a query.
	// This can occur when a previous statement in the same query has errored.
	ErrNotExecuted = errors.New("not executed")
)
View Source
var (
	// ErrFieldOverflow is returned when too many fields are created on a measurement.
	ErrFieldOverflow = errors.New("field overflow")

	// ErrFieldTypeConflict is returned when a new field already exists with a different type.
	ErrFieldTypeConflict = errors.New("field type conflict")

	// ErrFieldNotFound is returned when a field cannot be found.
	ErrFieldNotFound = errors.New("field not found")

	// ErrFieldUnmappedID is returned when the system is presented, during decode, with a field ID
	// there is no mapping for.
	ErrFieldUnmappedID = errors.New("field ID not mapped")
)
View Source
var (
	// ErrFormatNotFound is returned when no format can be determined from a path.
	ErrFormatNotFound = errors.New("format not found")
)
View Source
var (
	ErrShardNotFound = fmt.Errorf("shard not found")
)

Functions

func DedupeEntries added in v0.9.3

func DedupeEntries(a [][]byte) [][]byte

DedupeEntries returns slices with unique keys (the first 8 bytes).

func ErrDatabaseNotFound

func ErrDatabaseNotFound(name string) error

func ErrMeasurementNotFound

func ErrMeasurementNotFound(name string) error

func IsRetryable

func IsRetryable(err error) bool

IsRetryable returns true if this error is temporary and could be retried

func MakeKey added in v0.9.3

func MakeKey(name []byte, tags Tags) []byte

func MarshalTags added in v0.9.3

func MarshalTags(tags map[string]string) []byte

used to convert the tag set to bytes for use as a lookup key

func MeasurementFromSeriesKey added in v0.9.3

func MeasurementFromSeriesKey(key string) string

func NewSnapshotWriter

func NewSnapshotWriter(meta []byte, store *Store) (*snapshot.Writer, error)

NewSnapshotWriter returns a new snapshot.Writer that will write metadata and the store's shards to an archive.

func NopWriteToCloser

func NopWriteToCloser(w io.WriterTo) interface {
	io.WriterTo
	io.Closer
}

NopWriteToCloser returns an io.WriterTo that implements io.Closer.

func ProcessAggregateDerivative added in v0.9.3

func ProcessAggregateDerivative(results [][]interface{}, isNonNegative bool, interval time.Duration) [][]interface{}

ProcessAggregateDerivative returns the derivatives of an aggregate result set

func RegisterEngine added in v0.9.3

func RegisterEngine(name string, fn NewEngineFunc)

RegisterEngine registers a storage engine initializer by name.

Types

type ByteSlices added in v0.9.3

type ByteSlices [][]byte

func (ByteSlices) Len added in v0.9.3

func (a ByteSlices) Len() int

func (ByteSlices) Less added in v0.9.3

func (a ByteSlices) Less(i, j int) bool

func (ByteSlices) Swap added in v0.9.3

func (a ByteSlices) Swap(i, j int)

type Config

type Config struct {
	Dir string `toml:"dir"`

	// WAL config options for b1 (introduced in 0.9.2)
	MaxWALSize             int           `toml:"max-wal-size"`
	WALFlushInterval       toml.Duration `toml:"wal-flush-interval"`
	WALPartitionFlushDelay toml.Duration `toml:"wal-partition-flush-delay"`

	// WAL configuration options for bz1 (introduced in 0.9.3)
	WALDir                    string        `toml:"wal-dir"`
	WALLoggingEnabled         bool          `toml:"wal-logging-enabled"`
	WALReadySeriesSize        int           `toml:"wal-ready-series-size"`
	WALCompactionThreshold    float64       `toml:"wal-compaction-threshold"`
	WALMaxSeriesSize          int           `toml:"wal-max-series-size"`
	WALFlushColdInterval      toml.Duration `toml:"wal-flush-cold-interval"`
	WALPartitionSizeThreshold uint64        `toml:"wal-partition-size-threshold"`
}

func NewConfig

func NewConfig() Config

type Cursor added in v0.9.3

type Cursor interface {
	Seek(seek []byte) (key, value []byte)
	Next() (key, value []byte)
	Direction() Direction
}

Cursor represents an iterator over a series.

func MultiCursor added in v0.9.3

func MultiCursor(d Direction, cursors ...Cursor) Cursor

MultiCursor returns a single cursor that combines the results of all cursors in order.

If the same key is returned from multiple cursors then the first cursor specified will take precendence. A key will only be returned once from the returned cursor.

type DatabaseIndex

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

DatabaseIndex is the in memory index of a collection of measurements, time series, and their tags. Exported functions are goroutine safe while un-exported functions assume the caller will use the appropriate locks

func NewDatabaseIndex

func NewDatabaseIndex() *DatabaseIndex

func (*DatabaseIndex) CreateMeasurementIndexIfNotExists added in v0.9.3

func (s *DatabaseIndex) CreateMeasurementIndexIfNotExists(name string) *Measurement

CreateMeasurementIndexIfNotExists creates or retrieves an in memory index object for the measurement

func (*DatabaseIndex) CreateSeriesIndexIfNotExists added in v0.9.3

func (s *DatabaseIndex) CreateSeriesIndexIfNotExists(measurementName string, series *Series) *Series

CreateSeriesIndexIfNotExists adds the series for the given measurement to the index and sets its ID or returns the existing series object

func (*DatabaseIndex) DropMeasurement

func (db *DatabaseIndex) DropMeasurement(name string)

DropMeasurement removes the measurement and all of its underlying series from the database index

func (*DatabaseIndex) DropSeries

func (db *DatabaseIndex) DropSeries(keys []string)

DropSeries removes the series keys and their tags from the index

func (*DatabaseIndex) Measurement

func (d *DatabaseIndex) Measurement(name string) *Measurement

Measurement returns the measurement object from the index by the name

func (*DatabaseIndex) MeasurementSeriesCounts

func (d *DatabaseIndex) MeasurementSeriesCounts() (nMeasurements int, nSeries int)

MeasurementSeriesCounts returns the number of measurements and series currently indexed by the database. Useful for reporting and monitoring.

func (*DatabaseIndex) Measurements

func (db *DatabaseIndex) Measurements() Measurements

Measurements returns a list of all measurements.

func (*DatabaseIndex) Series added in v0.9.3

func (d *DatabaseIndex) Series(key string) *Series

Series returns a series by key.

func (*DatabaseIndex) SeriesN added in v0.9.3

func (d *DatabaseIndex) SeriesN() int

SeriesN returns the number of series.

func (*DatabaseIndex) TagsForSeries added in v0.9.3

func (s *DatabaseIndex) TagsForSeries(key string) map[string]string

TagsForSeries returns the tag map for the passed in series

type Direction added in v0.9.4

type Direction bool

Direction represents a cursor navigation direction.

const (
	// Forward indicates that a cursor will move forward over its values.
	Forward Direction = true
	// Reverse indicates that a cursor will move backwards over its values.
	Reverse Direction = false
)

func (Direction) Forward added in v0.9.4

func (d Direction) Forward() bool

Forward returns true if direction is forward

func (Direction) Reverse added in v0.9.4

func (d Direction) Reverse() bool

Forward returns true if direction is reverse

func (Direction) String added in v0.9.4

func (d Direction) String() string

type Engine added in v0.9.3

type Engine interface {
	Open() error
	Close() error

	SetLogOutput(io.Writer)
	LoadMetadataIndex(index *DatabaseIndex, measurementFields map[string]*MeasurementFields) error

	Begin(writable bool) (Tx, error)
	WritePoints(points []Point, measurementFieldsToSave map[string]*MeasurementFields, seriesToCreate []*SeriesCreate) error
	DeleteSeries(keys []string) error
	DeleteMeasurement(name string, seriesKeys []string) error
	SeriesCount() (n int, err error)

	io.WriterTo
}

Engine represents a swappable storage engine for the shard.

func NewEngine added in v0.9.3

func NewEngine(path string, walPath string, options EngineOptions) (Engine, error)

NewEngine returns an instance of an engine based on its format. If the path does not exist then the DefaultFormat is used.

type EngineOptions added in v0.9.3

type EngineOptions struct {
	EngineVersion          string
	MaxWALSize             int
	WALFlushInterval       time.Duration
	WALPartitionFlushDelay time.Duration

	Config Config
}

EngineOptions represents the options used to initialize the engine.

func NewEngineOptions added in v0.9.3

func NewEngineOptions() EngineOptions

NewEngineOptions returns the default options.

type ErrAuthorize

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

ErrAuthorize represents an authorization error.

func NewErrAuthorize added in v0.9.2

func NewErrAuthorize(qe *QueryExecutor, q *influxql.Query, u, db, m string) *ErrAuthorize

newAuthorizationError returns a new instance of AuthorizationError.

func (ErrAuthorize) Error

func (e ErrAuthorize) Error() string

Error returns the text of the error.

type Executor added in v0.9.2

type Executor interface {
	Execute() <-chan *influxql.Row
}

Executor is an interface for a query executor.

type Field added in v0.9.3

type Field struct {
	ID   uint8             `json:"id,omitempty"`
	Name string            `json:"name,omitempty"`
	Type influxql.DataType `json:"type,omitempty"`
}

Field represents a series field.

type FieldCodec

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

FieldCodec provides encoding and decoding functionality for the fields of a given Measurement. It is a distinct type to avoid locking writes on this node while potentially long-running queries are executing.

It is not affected by changes to the Measurement object after codec creation. TODO: this shouldn't be exported. nothing outside the shard should know about field encodings.

However, this is here until tx.go and the engine get refactored into tsdb.

func NewFieldCodec added in v0.9.3

func NewFieldCodec(fields map[string]*Field) *FieldCodec

NewFieldCodec returns a FieldCodec for the given Measurement. Must be called with a RLock that protects the Measurement.

func (*FieldCodec) DecodeByID

func (f *FieldCodec) DecodeByID(targetID uint8, b []byte) (interface{}, error)

DecodeByID scans a byte slice for a field with the given ID, converts it to its expected type, and return that value. TODO: shouldn't be exported. refactor engine

func (*FieldCodec) DecodeByName added in v0.9.2

func (f *FieldCodec) DecodeByName(name string, b []byte) (interface{}, error)

DecodeByName scans a byte slice for a field with the given name, converts it to its expected type, and return that value.

func (*FieldCodec) DecodeFields

func (f *FieldCodec) DecodeFields(b []byte) (map[uint8]interface{}, error)

DecodeFields decodes a byte slice into a set of field ids and values.

func (*FieldCodec) DecodeFieldsWithNames

func (f *FieldCodec) DecodeFieldsWithNames(b []byte) (map[string]interface{}, error)

DecodeFieldsWithNames decodes a byte slice into a set of field names and values TODO: shouldn't be exported. refactor engine

func (*FieldCodec) EncodeFields

func (f *FieldCodec) EncodeFields(values map[string]interface{}) ([]byte, error)

EncodeFields converts a map of values with string keys to a byte slice of field IDs and values.

If a field exists in the codec, but its type is different, an error is returned. If a field is not present in the codec, the system panics.

func (*FieldCodec) FieldIDByName

func (f *FieldCodec) FieldIDByName(s string) (uint8, error)

TODO: this shouldn't be exported. remove when tx.go and engine.go get refactored into tsdb

type FieldCreate added in v0.9.3

type FieldCreate struct {
	Measurement string
	Field       *Field
}

struct to hold information for a field to create on a measurement

type Fields

type Fields map[string]interface{}

func (Fields) MarshalBinary

func (p Fields) MarshalBinary() []byte

MarshalBinary encodes all the fields to their proper type and returns the binary represenation NOTE: uint64 is specifically not supported due to potential overflow when we decode again later to an int64

type Mapper added in v0.9.2

type Mapper interface {
	Open() error
	SetRemote(m Mapper) error
	TagSets() []string
	Fields() []string
	NextChunk() (interface{}, error)
	Close()
}

Mapper is the interface all Mapper types must implement.

type MapperOutput added in v0.9.3

type MapperOutput struct {
	Name   string            `json:"name,omitempty"`
	Tags   map[string]string `json:"tags,omitempty"`
	Fields []string          `json:"fields,omitempty"` // Field names of returned data.
	Values []*MapperValue    `json:"values,omitempty"` // For aggregates contains a single value at [0]
	// contains filtered or unexported fields
}

type MapperValue added in v0.9.3

type MapperValue struct {
	Time  int64             `json:"time,omitempty"`  // Ignored for aggregate output.
	Value interface{}       `json:"value,omitempty"` // For aggregate, contains interval time multiple values.
	Tags  map[string]string `json:"tags,omitempty"`  // Meta tags for results
}

MapperValue is a complex type, which can encapsulate data from both raw and aggregate mappers. This currently allows marshalling and network system to remain simpler. For aggregate output Time is ignored, and actual Time-Value pairs are contained soley within the Value field.

type MapperValues added in v0.9.3

type MapperValues []*MapperValue

func (MapperValues) Len added in v0.9.3

func (a MapperValues) Len() int

func (MapperValues) Less added in v0.9.3

func (a MapperValues) Less(i, j int) bool

func (MapperValues) Swap added in v0.9.3

func (a MapperValues) Swap(i, j int)

type Measurement

type Measurement struct {
	Name string `json:"name,omitempty"`
	// contains filtered or unexported fields
}

Measurement represents a collection of time series in a database. It also contains in memory structures for indexing tags. Exported functions are goroutine safe while un-exported functions assume the caller will use the appropriate locks

func NewMeasurement

func NewMeasurement(name string, idx *DatabaseIndex) *Measurement

NewMeasurement allocates and initializes a new Measurement.

func (*Measurement) AddSeries

func (m *Measurement) AddSeries(s *Series) bool

AddSeries will add a series to the measurementIndex. Returns false if already present

func (*Measurement) DropSeries

func (m *Measurement) DropSeries(seriesID uint64)

DropSeries will remove a series from the measurementIndex.

func (*Measurement) FieldNames

func (m *Measurement) FieldNames() (a []string)

FieldNames returns a list of the measurement's field names

func (*Measurement) HasField

func (m *Measurement) HasField(name string) bool

HasField returns true if the measurement has a field by the given name

func (*Measurement) HasSeries

func (m *Measurement) HasSeries() bool

HasSeries returns true if there is at least 1 series under this measurement

func (*Measurement) HasTagKey

func (m *Measurement) HasTagKey(k string) bool

HasTagKey returns true if at least one series in this measurement has written a value for the passed in tag key

func (*Measurement) SeriesByID added in v0.9.3

func (m *Measurement) SeriesByID(id uint64) *Series

SeriesByID returns a series by identifier.

func (*Measurement) SeriesKeys

func (m *Measurement) SeriesKeys() []string

SeriesKeys returns the keys of every series in this measurement

func (*Measurement) SetFieldName added in v0.9.3

func (m *Measurement) SetFieldName(name string)

SetFieldName adds the field name to the measurement.

func (*Measurement) TagKeys

func (m *Measurement) TagKeys() []string

TagKeys returns a list of the measurement's tag names.

func (*Measurement) TagSets

func (m *Measurement) TagSets(stmt *influxql.SelectStatement, dimensions []string) ([]*influxql.TagSet, error)

tagSets returns the unique tag sets that exist for the given tag keys. This is used to determine what composite series will be created by a group by. i.e. "group by region" should return: {"region":"uswest"}, {"region":"useast"} or region, service returns {"region": "uswest", "service": "redis"}, {"region": "uswest", "service": "mysql"}, etc... This will also populate the TagSet objects with the series IDs that match each tagset and any influx filter expression that goes with the series TODO: this shouldn't be exported. However, until tx.go and the engine get refactored into tsdb, we need it.

func (*Measurement) TagValues added in v0.9.4

func (m *Measurement) TagValues(key string) []string

TagValues returns all the values for the given tag key

func (*Measurement) ValidateGroupBy added in v0.9.2

func (m *Measurement) ValidateGroupBy(stmt *influxql.SelectStatement) error

ValidateGroupBy ensures that the GROUP BY is not a field.

type MeasurementFields added in v0.9.3

type MeasurementFields struct {
	Fields map[string]*Field `json:"fields"`
	Codec  *FieldCodec
}

func (*MeasurementFields) CreateFieldIfNotExists added in v0.9.3

func (m *MeasurementFields) CreateFieldIfNotExists(name string, typ influxql.DataType) error

CreateFieldIfNotExists creates a new field with an autoincrementing ID. Returns an error if 255 fields have already been created on the measurement or the fields already exists with a different type.

func (*MeasurementFields) MarshalBinary added in v0.9.3

func (m *MeasurementFields) MarshalBinary() ([]byte, error)

MarshalBinary encodes the object to a binary format.

func (*MeasurementFields) UnmarshalBinary added in v0.9.3

func (m *MeasurementFields) UnmarshalBinary(buf []byte) error

UnmarshalBinary decodes the object from a binary format.

type Measurements

type Measurements []*Measurement

Measurements represents a list of *Measurement.

func (Measurements) Len

func (a Measurements) Len() int

func (Measurements) Less

func (a Measurements) Less(i, j int) bool

func (Measurements) Swap

func (a Measurements) Swap(i, j int)

type NewEngineFunc added in v0.9.3

type NewEngineFunc func(path string, walPath string, options EngineOptions) Engine

NewEngineFunc creates a new engine.

type Point

type Point interface {
	Name() string
	SetName(string)

	Tags() Tags
	AddTag(key, value string)
	SetTags(tags Tags)

	Fields() Fields
	AddField(name string, value interface{})

	Time() time.Time
	SetTime(t time.Time)
	UnixNano() int64

	HashID() uint64
	Key() []byte

	Data() []byte
	SetData(buf []byte)

	String() string
}

Point defines the values that will be written to the database

func NewPoint

func NewPoint(name string, tags Tags, fields Fields, time time.Time) Point

NewPoint returns a new point with the given measurement name, tags, fields and timestamp

func ParsePoints

func ParsePoints(buf []byte) ([]Point, error)

ParsePoints returns a slice of Points from a text representation of a point with each point separated by newlines.

func ParsePointsString

func ParsePointsString(buf string) ([]Point, error)

func ParsePointsWithPrecision

func ParsePointsWithPrecision(buf []byte, defaultTime time.Time, precision string) ([]Point, error)

type PointBatcher

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

PointBatcher accepts Points and will emit a batch of those points when either a) the batch reaches a certain size, or b) a certain time passes.

func NewPointBatcher

func NewPointBatcher(sz int, bp int, d time.Duration) *PointBatcher

NewPointBatcher returns a new PointBatcher. sz is the batching size, bp is the maximum number of batches that may be pending. d is the time after which a batch will be emitted after the first point is received for the batch, regardless of its size.

func (*PointBatcher) Flush

func (b *PointBatcher) Flush()

Flush instructs the batcher to emit any pending points in a batch, regardless of batch size. If there are no pending points, no batch is emitted.

func (*PointBatcher) In

func (b *PointBatcher) In() chan<- Point

In returns the channel to which points should be written.

func (*PointBatcher) Out

func (b *PointBatcher) Out() <-chan []Point

Out returns the channel from which batches should be read.

func (*PointBatcher) Start

func (b *PointBatcher) Start()

Start starts the batching process. Returns the in and out channels for points and point-batches respectively.

func (*PointBatcher) Stats

func (b *PointBatcher) Stats() *PointBatcherStats

Stats returns a PointBatcherStats object for the PointBatcher. While the each statistic should be closely correlated with each other statistic, it is not guaranteed.

func (*PointBatcher) Stop

func (b *PointBatcher) Stop()

type PointBatcherStats

type PointBatcherStats struct {
	BatchTotal   uint64 // Total count of batches transmitted.
	PointTotal   uint64 // Total count of points processed.
	SizeTotal    uint64 // Number of batches that reached size threshold.
	TimeoutTotal uint64 // Number of timeouts that occurred.
}

PointBatcherStats are the statistics each batcher tracks.

type Points added in v0.9.3

type Points []Point

Points represents a sortable list of points by timestamp.

func (Points) Len added in v0.9.3

func (a Points) Len() int

func (Points) Less added in v0.9.3

func (a Points) Less(i, j int) bool

func (Points) Swap added in v0.9.3

func (a Points) Swap(i, j int)

type QueryExecutor

type QueryExecutor struct {
	// The meta store for accessing and updating cluster and schema data.
	MetaStore interface {
		Database(name string) (*meta.DatabaseInfo, error)
		Databases() ([]meta.DatabaseInfo, error)
		User(name string) (*meta.UserInfo, error)
		AdminUserExists() (bool, error)
		Authenticate(username, password string) (*meta.UserInfo, error)
		RetentionPolicy(database, name string) (rpi *meta.RetentionPolicyInfo, err error)
		UserCount() (int, error)
		ShardGroupsByTimeRange(database, policy string, min, max time.Time) (a []meta.ShardGroupInfo, err error)
		NodeID() uint64
	}

	// Executes statements relating to meta data.
	MetaStatementExecutor interface {
		ExecuteStatement(stmt influxql.Statement) *influxql.Result
	}

	// Execute statements relating to statistics and diagnostics.
	MonitorStatementExecutor interface {
		ExecuteStatement(stmt influxql.Statement) *influxql.Result
	}

	// Maps shards for queries.
	ShardMapper interface {
		CreateMapper(shard meta.ShardInfo, stmt influxql.Statement, chunkSize int) (Mapper, error)
	}

	Logger *log.Logger

	// the local data store
	Store *Store
}

QueryExecutor executes every statement in an influxdb Query. It is responsible for coordinating between the local tsdb.Store, the meta.Store, and the other nodes in the cluster to run the query against their local tsdb.Stores. There should be one executor in a running process

func NewQueryExecutor

func NewQueryExecutor(store *Store) *QueryExecutor

NewQueryExecutor returns an initialized QueryExecutor

func (*QueryExecutor) Authorize

func (q *QueryExecutor) Authorize(u *meta.UserInfo, query *influxql.Query, database string) error

Authorize user u to execute query q on database. database can be "" for queries that do not require a database. If no user is provided it will return an error unless the query's first statement is to create a root user.

func (*QueryExecutor) ExecuteQuery

func (q *QueryExecutor) ExecuteQuery(query *influxql.Query, database string, chunkSize int) (<-chan *influxql.Result, error)

ExecuteQuery executes an InfluxQL query against the server. It sends results down the passed in chan and closes it when done. It will close the chan on the first statement that throws an error.

func (*QueryExecutor) PlanSelect added in v0.9.4

func (q *QueryExecutor) PlanSelect(stmt *influxql.SelectStatement, chunkSize int) (Executor, error)

Plan creates an execution plan for the given SelectStatement and returns an Executor.

func (*QueryExecutor) PlanShowMeasurements added in v0.9.4

func (q *QueryExecutor) PlanShowMeasurements(stmt *influxql.ShowMeasurementsStatement, database string, chunkSize int) (Executor, error)

PlanShowMeasurements creates an execution plan for the given SelectStatement and returns an Executor.

func (*QueryExecutor) SetLogger added in v0.9.4

func (q *QueryExecutor) SetLogger(l *log.Logger)

SetLogger sets the internal logger to the logger passed in.

type RawQueryDerivativeProcessor added in v0.9.3

type RawQueryDerivativeProcessor struct {
	LastValueFromPreviousChunk *MapperValue
	IsNonNegative              bool // Whether to drop negative differences
	DerivativeInterval         time.Duration
}

func (*RawQueryDerivativeProcessor) Process added in v0.9.3

func (rqdp *RawQueryDerivativeProcessor) Process(input []*MapperValue) []*MapperValue

type SelectExecutor added in v0.9.4

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

func NewSelectExecutor added in v0.9.4

func NewSelectExecutor(stmt *influxql.SelectStatement, mappers []Mapper, chunkSize int) *SelectExecutor

NewSelectExecutor returns a new SelectExecutor.

func (*SelectExecutor) Execute added in v0.9.4

func (e *SelectExecutor) Execute() <-chan *influxql.Row

Execute begins execution of the query and returns a channel to receive rows.

type SelectMapper added in v0.9.4

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

SelectMapper is for retrieving data for a query, from a given shard.

func NewSelectMapper added in v0.9.4

func NewSelectMapper(shard *Shard, stmt influxql.Statement, chunkSize int) *SelectMapper

NewSelectMapper returns a mapper for the given shard, which will return data for the SELECT statement.

func (*SelectMapper) Close added in v0.9.4

func (lm *SelectMapper) Close()

Close closes the mapper.

func (*SelectMapper) Fields added in v0.9.4

func (lm *SelectMapper) Fields() []string

Fields returns any SELECT fields. If this Mapper is not processing a SELECT query then an empty slice is returned.

func (*SelectMapper) NextChunk added in v0.9.4

func (lm *SelectMapper) NextChunk() (interface{}, error)

func (*SelectMapper) Open added in v0.9.4

func (lm *SelectMapper) Open() error

Open opens the local mapper.

func (*SelectMapper) SetRemote added in v0.9.4

func (lm *SelectMapper) SetRemote(m Mapper) error

func (*SelectMapper) TagSets added in v0.9.4

func (lm *SelectMapper) TagSets() []string

TagSets returns the list of TagSets for which this mapper has data.

type Series

type Series struct {
	Key  string
	Tags map[string]string
	// contains filtered or unexported fields
}

Series belong to a Measurement and represent unique time series in a database

func NewSeries added in v0.9.3

func NewSeries(key string, tags map[string]string) *Series

NewSeries returns an initialized series struct

func (*Series) InitializeShards added in v0.9.3

func (s *Series) InitializeShards()

func (*Series) MarshalBinary

func (s *Series) MarshalBinary() ([]byte, error)

MarshalBinary encodes the object to a binary format.

func (*Series) UnmarshalBinary

func (s *Series) UnmarshalBinary(buf []byte) error

UnmarshalBinary decodes the object from a binary format.

type SeriesCreate added in v0.9.3

type SeriesCreate struct {
	Measurement string
	Series      *Series
}

struct to hold information for a series to create

type SeriesIDs added in v0.9.3

type SeriesIDs []uint64

SeriesIDs is a convenience type for sorting, checking equality, and doing union and intersection of collections of series ids.

func (SeriesIDs) Equals added in v0.9.3

func (a SeriesIDs) Equals(other SeriesIDs) bool

Equals assumes that both are sorted.

func (SeriesIDs) Intersect added in v0.9.3

func (a SeriesIDs) Intersect(other SeriesIDs) SeriesIDs

Intersect returns a new collection of series ids in sorted order that is the intersection of the two. The two collections must already be sorted.

func (SeriesIDs) Len added in v0.9.3

func (a SeriesIDs) Len() int

func (SeriesIDs) Less added in v0.9.3

func (a SeriesIDs) Less(i, j int) bool

func (SeriesIDs) Reject added in v0.9.3

func (a SeriesIDs) Reject(other SeriesIDs) SeriesIDs

Reject returns a new collection of series ids in sorted order with the passed in set removed from the original. This is useful for the NOT operator. The two collections must already be sorted.

func (SeriesIDs) Swap added in v0.9.3

func (a SeriesIDs) Swap(i, j int)

func (SeriesIDs) Union added in v0.9.3

func (a SeriesIDs) Union(other SeriesIDs) SeriesIDs

Union returns a new collection of series ids in sorted order that is the union of the two. The two collections must already be sorted.

type Shard

type Shard struct {

	// The writer used by the logger.
	LogOutput io.Writer
	// contains filtered or unexported fields
}

Shard represents a self-contained time series database. An inverted index of the measurement and tag data is kept along with the raw time series data. Data can be split across many shards. The query engine in TSDB is responsible for combining the output of many shards into a single query result.

func NewShard

func NewShard(id uint64, index *DatabaseIndex, path string, walPath string, options EngineOptions) *Shard

NewShard returns a new initialized Shard. walPath doesn't apply to the b1 type index

func (*Shard) Close

func (s *Shard) Close() error

Close shuts down the shard's store.

func (*Shard) DeleteMeasurement added in v0.9.3

func (s *Shard) DeleteMeasurement(name string, seriesKeys []string) error

DeleteMeasurement deletes a measurement and all underlying series.

func (*Shard) DeleteSeries added in v0.9.3

func (s *Shard) DeleteSeries(keys []string) error

DeleteSeries deletes a list of series.

func (*Shard) DiskSize added in v0.9.4

func (s *Shard) DiskSize() (int64, error)

DiskSize returns the size on disk of this shard

func (*Shard) FieldCodec

func (s *Shard) FieldCodec(measurementName string) *FieldCodec

TODO: this is temporarily exported to make tx.go work. When the query engine gets refactored into the tsdb package this should be removed. No one outside tsdb should know the underlying field encoding scheme.

func (*Shard) Open

func (s *Shard) Open() error

open initializes and opens the shard's store.

func (*Shard) Path

func (s *Shard) Path() string

Path returns the path set on the shard when it was created.

func (*Shard) ReadOnlyTx added in v0.9.4

func (s *Shard) ReadOnlyTx() (Tx, error)

ReadOnlyTx returns a read-only transaction for the shard. The transaction must be rolled back to release resources.

func (*Shard) SeriesCount added in v0.9.1

func (s *Shard) SeriesCount() (int, error)

SeriesCount returns the number of series buckets on the shard.

func (*Shard) ValidateAggregateFieldsInStatement

func (s *Shard) ValidateAggregateFieldsInStatement(measurementName string, stmt *influxql.SelectStatement) error

func (*Shard) WritePoints

func (s *Shard) WritePoints(points []Point) error

WritePoints will write the raw data points and any new metadata to the index in the shard

func (*Shard) WriteTo added in v0.9.4

func (s *Shard) WriteTo(w io.Writer) (int64, error)

WriteTo writes the shard's data to w.

type ShowMeasurementsExecutor added in v0.9.4

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

ShowMeasurementsExecutor implements the Executor interface for a SHOW MEASUREMENTS statement.

func NewShowMeasurementsExecutor added in v0.9.4

func NewShowMeasurementsExecutor(stmt *influxql.ShowMeasurementsStatement, mappers []Mapper, chunkSize int) *ShowMeasurementsExecutor

NewShowMeasurementsExecutor returns a new ShowMeasurementsExecutor.

func (*ShowMeasurementsExecutor) Execute added in v0.9.4

func (e *ShowMeasurementsExecutor) Execute() <-chan *influxql.Row

Execute begins execution of the query and returns a channel to receive rows.

type ShowMeasurementsMapper added in v0.9.4

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

ShowMeasurementsMapper is a mapper for collecting measurement names from a shard.

func NewShowMeasurementsMapper added in v0.9.4

func NewShowMeasurementsMapper(shard *Shard, stmt *influxql.ShowMeasurementsStatement, chunkSize int) *ShowMeasurementsMapper

NewShowMeasurementsMapper returns a mapper for the given shard, which will return data for the meta statement.

func (*ShowMeasurementsMapper) Close added in v0.9.4

func (m *ShowMeasurementsMapper) Close()

Close closes the mapper.

func (*ShowMeasurementsMapper) Fields added in v0.9.4

func (m *ShowMeasurementsMapper) Fields() []string

Fields returns a list of field names for this mapper.

func (*ShowMeasurementsMapper) NextChunk added in v0.9.4

func (m *ShowMeasurementsMapper) NextChunk() (interface{}, error)

NextChunk returns the next chunk of measurement names.

func (*ShowMeasurementsMapper) Open added in v0.9.4

func (m *ShowMeasurementsMapper) Open() error

Open opens the mapper for use.

func (*ShowMeasurementsMapper) SetRemote added in v0.9.4

func (m *ShowMeasurementsMapper) SetRemote(remote Mapper) error

SetRemote sets the remote mapper to use.

func (*ShowMeasurementsMapper) TagSets added in v0.9.4

func (m *ShowMeasurementsMapper) TagSets() []string

TagSets is only implemented on this mapper to satisfy the Mapper interface.

type StatefulMapper added in v0.9.2

type StatefulMapper struct {
	Mapper
	// contains filtered or unexported fields
}

StatefulMapper encapsulates a Mapper and some state that the executor needs to track for that mapper.

func (*StatefulMapper) NextChunk added in v0.9.2

func (sm *StatefulMapper) NextChunk() (*MapperOutput, error)

NextChunk wraps a RawMapper and some state.

type Store

type Store struct {
	EngineOptions EngineOptions
	Logger        *log.Logger
	// contains filtered or unexported fields
}

func NewStore

func NewStore(path string) *Store

func (*Store) Close

func (s *Store) Close() error

func (*Store) CreateMapper added in v0.9.2

func (s *Store) CreateMapper(shardID uint64, stmt influxql.Statement, chunkSize int) (Mapper, error)

func (*Store) CreateShard

func (s *Store) CreateShard(database, retentionPolicy string, shardID uint64) error

func (*Store) DatabaseIndex

func (s *Store) DatabaseIndex(name string) *DatabaseIndex

func (*Store) DatabaseIndexN added in v0.9.3

func (s *Store) DatabaseIndexN() int

DatabaseIndexN returns the number of databases indicies in the store.

func (*Store) Databases added in v0.9.4

func (s *Store) Databases() []string

Databases returns all the databases in the indexes

func (*Store) DeleteDatabase

func (s *Store) DeleteDatabase(name string, shardIDs []uint64) error

DeleteDatabase will close all shards associated with a database and remove the directory and files from disk.

func (*Store) DeleteShard

func (s *Store) DeleteShard(shardID uint64) error

DeleteShard removes a shard from disk.

func (*Store) DiskSize added in v0.9.4

func (s *Store) DiskSize() (int64, error)

DiskSize returns the size of all the shard files in bytes. This size does not include the WAL size.

func (*Store) Measurement

func (s *Store) Measurement(database, name string) *Measurement

func (*Store) Open

func (s *Store) Open() error

func (*Store) Path

func (s *Store) Path() string

Path returns the store's root path.

func (*Store) Shard

func (s *Store) Shard(id uint64) *Shard

Shard returns a shard by id.

func (*Store) ShardIDs

func (s *Store) ShardIDs() []uint64

ShardIDs returns a slice of all ShardIDs under management.

func (*Store) ShardN added in v0.9.3

func (s *Store) ShardN() int

ShardN returns the number of shard in the store.

func (*Store) ValidateAggregateFieldsInStatement

func (s *Store) ValidateAggregateFieldsInStatement(shardID uint64, measurementName string, stmt *influxql.SelectStatement) error

func (*Store) WriteToShard

func (s *Store) WriteToShard(shardID uint64, points []Point) error

type TagFilter

type TagFilter struct {
	Op    influxql.Token
	Key   string
	Value string
	Regex *regexp.Regexp
}

TagFilter represents a tag filter when looking up other tags or measurements.

type Tags

type Tags map[string]string

func (Tags) HashKey added in v0.9.3

func (t Tags) HashKey() []byte

type Tx added in v0.9.3

type Tx interface {
	io.WriterTo

	Cursor(series string, direction Direction) Cursor
	Size() int64
	Commit() error
	Rollback() error
}

Tx represents a transaction.

Directories

Path Synopsis
b1
bz1
wal
Package WAL implements a write ahead log optimized for write throughput that can be put in front of the database index.
Package WAL implements a write ahead log optimized for write throughput that can be put in front of the database index.
Package internal is a generated protocol buffer package.
Package internal is a generated protocol buffer package.

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL