Путеводитель по Руководству Linux

  User  |  Syst  |  Libr  |  Device  |  Files  |  Other  |  Admin  |  Head  |



   pmSeriesLabelValues    ( 3 )

быстрые, масштабируемые метаданные временных рядов (fast, scalable time series metadata)

   Дубль

(статьи: pmSeriesDescs - быстрые, масштабируемые метаданные временных рядов )

Имя (Name)

pmSeriesDescs, pmSeriesLabels, pmSeriesLabelValues, pmSeriesInstances, pmSeriesMetrics, pmSeriesSources - fast, scalable time series metadata


Синопсис C (C Synopsis)

#include <pcp/pmwebapi.h>

int pmSeriesDescs(pmSeriesSettings *sp, int count, sds *series, void *arg); int pmSeriesLabels(pmSeriesSettings *sp, int count, sds *series, void *arg); int pmSeriesLabelValues(pmSeriesSettings *sp, int count, sds *labels, void *arg); int pmSeriesInstances(pmSeriesSettings *sp, int count, sds *series, void *arg); int pmSeriesMetrics(pmSeriesSettings *sp, int count, sds *series, void *arg); int pmSeriesSources(pmSeriesSettings *sp, int count, sds *series, void *arg);

cc ... -lpcp_web


Описание (Description)

Each performance metric and data source forming part of the Performance Co-Pilot (PCP) fast, scalable time series service has certain properties (metadata) associated with it. These properties can be queried using the interfaces described here.

This functionality is provided through asynchronous APIs, which function in an event-driven fashion where callbacks are invoked for each metadata structure being returned.

As a general pattern, all interfaces in these APIs that need to invoke callbacks provided by the calling program (see pmSeriesSetup(3)) will take an opaque (void * pointer) parameter, arg. This pointer will be passed through unchanged and is typically used to access a data structure maintaining state within the calling program.

With a couple of exceptions, these interfaces tend to operate in one of two modes. Firstly, one invocation is available to query metadata for a specific (non-zero) count of time series. A second, (zero) count, mode returns all available metadata of that type, across all time series.

In order to retrieve metric descriptor information for one or more time series, the pmSeriesDescs interface is used. For each valid series identifier provided, a callback will be invoked providing the metric units, semantics, and type as well as other identifiers associated with it (pmID, InDom and source identifier). When this call is issued with a zero value for count, It is an error to pass a zero or negative value of count into this interface.

To extract label names and values, the pmSeriesLabels interface is used. With a non-zero value for count valid series identifiers, this routine will invoke a callback once for each label name and once for each label name:value pair, for labels associated with those time series. If called with a count of zero, this interface will return (via the label name callback) all label names that have observed to date - not associated with specific time series. The pmSeriesLabelValues interface is comparable to this latter mode, except will return all label values that have been observed to date for the given array of labels names (not associated with any specific time series identifiers).

pmSeriesInstances can be used to find metadata about instance domains and instance identifiers associated with a given (non- zero) count of time series identifiers. If a zero value is passed for count then all instance names observed to date will be returned.

pmSeriesMetrics can be used to find metric names associated with a given (non-zero) count of time series identifiers. If a zero value is passed for count then all metric names observed to date will be returned.

pmSeriesSources can be used to find metadata about the source of metrics - that is, host names and archive file paths. If a zero value is passed for count then all metric sources observed to date will be returned.


Диагностика (Diagnostic)

Where these functions return a status code, this is always zero on success. On failure a negative PMAPI error code is returned.


Смотри также (See also)

pmproxy(1), pmlogger(1), pmseries(1), pmSeriesQuery(3), pmSeriesSetup(3), PMAPI(3) and PMWEBAPI(3).