recipes

Documentation

Index

• func DumpMaps(recipes []MetricQueryRecipe)
• type MetricQueryRecipe
• type MetricQueryRecipeBase
  • func (mqrb *MetricQueryRecipeBase) GetNamespace() string
  • func (mqrb *MetricQueryRecipeBase) GetResultMaps() MultiResultMap
• type MetricQueryRecipeSimple
  • func (mqrs *MetricQueryRecipeSimple) Run(conn db.Conn) ([]db.ScannedResultSet, error)
• type MetricQueryRecipeTemplated
  • func (mqrt *MetricQueryRecipeTemplated) Run(conn db.Conn) ([]db.ScannedResultSet, error)
• type MultiResultMap
• type NamedResultMap
  • func (nrm NamedResultMap) ShouldSkip() bool
• type ResultMap

Functions

func DumpMaps

func DumpMaps(recipes []MetricQueryRecipe)

Types

type MetricQueryRecipe

type MetricQueryRecipe interface {
	// Return the basename associated with this recipe; all metrics yielded
	// will be prefixed with this
	GetNamespace() string
	// Returns a map to be used in interpreting the result of Run(): each
	// column in the resultset should be looked up in this map to determine
	// how to handle it.
	GetResultMaps() MultiResultMap
	// Run executes one or more queries and returns one or more resultsets.
	// There need not be a one-to-one mapping.
	Run(db.Conn) ([]db.ScannedResultSet, error)
}

MetricQueryRecipe is a collection of rules that specify how to produce a set of related metrics given a DB connection. In other words, it knows how to issue query(s) and interpret the results.

type MetricQueryRecipeBase

type MetricQueryRecipeBase struct {
	// Namespace is typically the name of a table to query, and it provides
	// the base name of derived metrics (after the driver prefix).
	Namespace string
	// ResultMaps maps column names in resultsets to the ColumnMapping
	// that should be used to build a metric.
	Resultmaps MultiResultMap
}

MetricQueryRecipeBase is common to all recipes.

func (*MetricQueryRecipeBase) GetNamespace

func (mqrb *MetricQueryRecipeBase) GetNamespace() string

GetNamespace implements MetricQueryRecipe.

func (*MetricQueryRecipeBase) GetResultMaps

func (mqrb *MetricQueryRecipeBase) GetResultMaps() MultiResultMap

GetResultMaps implements MetricQueryRecipe.

type MetricQueryRecipeSimple

type MetricQueryRecipeSimple struct {
	*MetricQueryRecipeBase
	// sqlquery is what should be executed
	Queries []string
}

func (*MetricQueryRecipeSimple) Run

func (mqrs *MetricQueryRecipeSimple) Run(conn db.Conn) ([]db.ScannedResultSet, error)

type MetricQueryRecipeTemplated

type MetricQueryRecipeTemplated struct {
	*MetricQueryRecipeBase
	// Rangequery returns a single-column resultset over which to iterate
	Rangequery string
	// All templates are executed in the context of a range over the rangequery results.
	// Each resulting string is executed as an SQL query, and the resulting resultsets
	// are returned by the ResultSet (after filtering out empty resultsets.)
	Queries []*template.Template
}

func (*MetricQueryRecipeTemplated) Run

func (mqrt *MetricQueryRecipeTemplated) Run(conn db.Conn) ([]db.ScannedResultSet, error)

type MultiResultMap

type MultiResultMap []NamedResultMap

MultiResultMap describes how to handle a list of resultsets, as can be returned by some queries in DBs like Sybase and SQL Server. Resultset don't intrically have names, just positions, but we're imposing a name so that we can craft metrics based on it.

type NamedResultMap

type NamedResultMap struct {
	ResultMap
	Name string
}

NamedResultMap associates a name with a ResultMap based on configuration.

func (NamedResultMap) ShouldSkip

func (nrm NamedResultMap) ShouldSkip() bool

ShouldSkip returns true if this resultset should be ignored.

type ResultMap

type ResultMap map[string]common.ColumnMapping

ResultMap describes how to handle a single resultset, i.e. how each column should be interpreted in terms of metrics.