Documentation ¶
Overview ¶
Package stackdriver contains the OpenCensus exporters for Stackdriver Monitoring and Stackdriver Tracing.
This exporter can be used to send metrics to Stackdriver Monitoring and traces to Stackdriver trace.
The package uses Application Default Credentials to authenticate by default. See: https://developers.google.com/identity/protocols/application-default-credentials
Alternatively, pass the authentication options in both the MonitoringClientOptions and the TraceClientOptions fields of Options.
Stackdriver Monitoring ¶
This exporter support exporting OpenCensus views to Stackdriver Monitoring. Each registered view becomes a metric in Stackdriver Monitoring, with the tags becoming labels.
The aggregation function determines the metric kind: LastValue aggregations generate Gauge metrics and all other aggregations generate Cumulative metrics.
In order to be able to push your stats to Stackdriver Monitoring, you must:
- Create a Cloud project: https://support.google.com/cloud/answer/6251787?hl=en
- Enable billing: https://support.google.com/cloud/answer/6288653#new-billing
- Enable the Stackdriver Monitoring API: https://console.cloud.google.com/apis/dashboard
These steps enable the API but don't require that your app is hosted on Google Cloud Platform.
Stackdriver Trace ¶
This exporter supports exporting Trace Spans to Stackdriver Trace. It also supports the Google "Cloud Trace" propagation format header.
Example (Defaults) ¶
package main import ( "log" "net/http" "contrib.go.opencensus.io/exporter/stackdriver" "contrib.go.opencensus.io/exporter/stackdriver/propagation" "go.opencensus.io/plugin/ochttp" "go.opencensus.io/stats/view" "go.opencensus.io/trace" ) func main() { exporter, err := stackdriver.NewExporter(stackdriver.Options{ProjectID: "google-project-id"}) if err != nil { log.Fatal(err) } // Export to Stackdriver Monitoring. if err = exporter.StartMetricsExporter(); err != nil { log.Fatal(err) } // Subscribe views to see stats in Stackdriver Monitoring. if err := view.Register( ochttp.ClientRoundtripLatencyDistribution, ochttp.ClientReceivedBytesDistribution, ); err != nil { log.Fatal(err) } // Export to Stackdriver Trace. trace.RegisterExporter(exporter) // Automatically add a Stackdriver trace header to outgoing requests: client := &http.Client{ Transport: &ochttp.Transport{ Propagation: &propagation.HTTPFormat{}, }, } _ = client // use client // All outgoing requests from client will include a Stackdriver Trace header. // See the ochttp package for how to handle incoming requests. }
Output:
Example (GKE) ¶
package main import ( "log" "os" "cloud.google.com/go/compute/metadata" "contrib.go.opencensus.io/exporter/stackdriver" "google.golang.org/genproto/googleapis/api/monitoredres" ) func main() { // This example shows how to set up a Stackdriver exporter suitable for // monitoring a GKE container. instanceID, err := metadata.InstanceID() if err != nil { log.Println("Error getting instance ID:", err) instanceID = "unknown" } zone, err := metadata.Zone() if err != nil { log.Println("Error getting zone:", err) zone = "unknown" } exporter, err := stackdriver.NewExporter(stackdriver.Options{ ProjectID: "google-project-id", // Set a MonitoredResource that represents a GKE container. Resource: &monitoredres.MonitoredResource{ Type: "gke_container", Labels: map[string]string{ "project_id": "google-project-id", "cluster_name": "my-cluster-name", "instance_id": instanceID, "zone": zone, // See: https://kubernetes.io/docs/tasks/inject-data-application/environment-variable-expose-pod-information/ "namespace_id": os.Getenv("MY_POD_NAMESPACE"), "pod_id": os.Getenv("MY_POD_NAME"), "container_name": os.Getenv("MY_CONTAINER_NAME"), }, }, // Set DefaultMonitoringLabels to avoid getting the default "opencensus_task" // label. For this to be valid, this exporter should be the only writer // to the metrics against this gke_container MonitoredResource. In this case, // it means you should only have one process writing to Stackdriver from this // container. DefaultMonitoringLabels: &stackdriver.Labels{}, }) if err != nil { log.Fatal(err) } // Register so that views are exported. if err = exporter.StartMetricsExporter(); err != nil { log.Fatal(err) } }
Output:
Index ¶
- func DefaultMapResource(res *resource.Resource) *monitoredrespb.MonitoredResource
- type Exporter
- func (e *Exporter) Close() error
- func (e *Exporter) ExportMetrics(ctx context.Context, metrics []*metricdata.Metric) error
- func (e *Exporter) ExportMetricsProto(ctx context.Context, node *commonpb.Node, rsc *resourcepb.Resource, ...) error
- func (e *Exporter) ExportSpan(sd *trace.SpanData)
- func (e *Exporter) ExportView(vd *view.Data)
- func (e *Exporter) Flush()
- func (e *Exporter) PushMetricsProto(ctx context.Context, node *commonpb.Node, rsc *resourcepb.Resource, ...) (int, error)
- func (e *Exporter) PushTraceSpans(ctx context.Context, node *commonpb.Node, rsc *resourcepb.Resource, ...) (int, error)
- func (e *Exporter) StartMetricsExporter() error
- func (e *Exporter) StopMetricsExporter()
- func (e *Exporter) ViewToMetricDescriptor(ctx context.Context, v *view.View) (*metricpb.MetricDescriptor, error)
- type Labels
- type Options
Examples ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
func DefaultMapResource ¶ added in v0.9.2
func DefaultMapResource(res *resource.Resource) *monitoredrespb.MonitoredResource
DefaultMapResource implements default resource mapping for well-known resource types
Types ¶
type Exporter ¶
type Exporter struct {
// contains filtered or unexported fields
}
Exporter is a stats and trace exporter that uploads data to Stackdriver.
You can create a single Exporter and register it as both a trace exporter (to export to Stackdriver Trace) and a stats exporter (to integrate with Stackdriver Monitoring).
func NewExporter ¶
NewExporter creates a new Exporter that implements both stats.Exporter and trace.Exporter.
func (*Exporter) ExportMetrics ¶ added in v0.10.0
ExportMetrics exports OpenCensus Metrics to Stackdriver Monitoring
func (*Exporter) ExportMetricsProto ¶ added in v0.10.0
func (e *Exporter) ExportMetricsProto(ctx context.Context, node *commonpb.Node, rsc *resourcepb.Resource, metrics []*metricspb.Metric) error
ExportMetricsProto exports OpenCensus Metrics Proto to Stackdriver Monitoring synchronously, without de-duping or adding proto metrics to the bundler.
func (*Exporter) ExportSpan ¶
ExportSpan exports a SpanData to Stackdriver Trace.
func (*Exporter) ExportView ¶
ExportView exports to the Stackdriver Monitoring if view data has one or more rows. Deprecated: use ExportMetrics and StartMetricsExporter instead.
func (*Exporter) Flush ¶
func (e *Exporter) Flush()
Flush waits for exported data to be uploaded.
This is useful if your program is ending and you do not want to lose recent stats or spans.
func (*Exporter) PushMetricsProto ¶ added in v0.12.6
func (e *Exporter) PushMetricsProto(ctx context.Context, node *commonpb.Node, rsc *resourcepb.Resource, metrics []*metricspb.Metric) (int, error)
PushMetricsProto simliar with ExportMetricsProto but returns the number of dropped timeseries.
func (*Exporter) PushTraceSpans ¶ added in v0.13.0
func (e *Exporter) PushTraceSpans(ctx context.Context, node *commonpb.Node, rsc *resourcepb.Resource, spans []*trace.SpanData) (int, error)
PushTraceSpans exports a bundle of OpenCensus Spans. Returns number of dropped spans.
func (*Exporter) StartMetricsExporter ¶ added in v0.10.0
StartMetricsExporter starts exporter by creating an interval reader that reads metrics from all registered producers at set interval and exports them. Use StopMetricsExporter to stop exporting metrics. Previously, it required registering exporter to export stats collected by opencensus.
exporter := stackdriver.NewExporter(stackdriver.Option{}) view.RegisterExporter(exporter)
Now, it requires to call StartMetricsExporter() to export stats and metrics collected by opencensus.
exporter := stackdriver.NewExporter(stackdriver.Option{}) exporter.StartMetricsExporter() defer exporter.StopMetricsExporter()
Both approach should not be used simultaneously. Otherwise it may result into unknown behavior. Previous approach continues to work as before but will not report newly define metrics such as gauges.
func (*Exporter) StopMetricsExporter ¶ added in v0.10.0
func (e *Exporter) StopMetricsExporter()
StopMetricsExporter stops exporter from exporting metrics.
func (*Exporter) ViewToMetricDescriptor ¶ added in v0.13.5
func (e *Exporter) ViewToMetricDescriptor(ctx context.Context, v *view.View) (*metricpb.MetricDescriptor, error)
ViewToMetricDescriptor converts an OpenCensus view to a MetricDescriptor.
This is useful for cases when you want to use your Go code as source of truth of metric descriptors. You can extract or define views in a central place, then call this method to generate MetricDescriptors.
type Labels ¶ added in v0.4.0
type Labels struct {
// contains filtered or unexported fields
}
Labels represents a set of Stackdriver Monitoring labels.
type Options ¶
type Options struct { // ProjectID is the identifier of the Stackdriver // project the user is uploading the stats data to. // If not set, this will default to your "Application Default Credentials". // For details see: https://developers.google.com/accounts/docs/application-default-credentials. // // It will be used in the project_id label of a Stackdriver monitored // resource if the resource does not inherently belong to a specific // project, e.g. on-premise resource like k8s_container or generic_task. ProjectID string // Location is the identifier of the GCP or AWS cloud region/zone in which // the data for a resource is stored. // If not set, it will default to the location provided by the metadata server. // // It will be used in the location label of a Stackdriver monitored resource // if the resource does not inherently belong to a specific project, e.g. // on-premise resource like k8s_container or generic_task. Location string // OnError is the hook to be called when there is // an error uploading the stats or tracing data. // If no custom hook is set, errors are logged. // Optional. OnError func(err error) // MonitoringClientOptions are additional options to be passed // to the underlying Stackdriver Monitoring API client. // Optional. MonitoringClientOptions []option.ClientOption // TraceClientOptions are additional options to be passed // to the underlying Stackdriver Trace API client. // Optional. TraceClientOptions []option.ClientOption // BundleDelayThreshold determines the max amount of time // the exporter can wait before uploading view data or trace spans to // the backend. // Optional. BundleDelayThreshold time.Duration // BundleCountThreshold determines how many view data events or trace spans // can be buffered before batch uploading them to the backend. // Optional. BundleCountThreshold int // TraceSpansBufferMaxBytes is the maximum size (in bytes) of spans that // will be buffered in memory before being dropped. // // If unset, a default of 8MB will be used. TraceSpansBufferMaxBytes int // Resource sets the MonitoredResource against which all views will be // recorded by this exporter. // // All Stackdriver metrics created by this exporter are custom metrics, // so only a limited number of MonitoredResource types are supported, see: // https://cloud.google.com/monitoring/custom-metrics/creating-metrics#which-resource // // An important consideration when setting the Resource here is that // Stackdriver Monitoring only allows a single writer per // TimeSeries, see: https://cloud.google.com/monitoring/api/v3/metrics-details#intro-time-series // A TimeSeries is uniquely defined by the metric type name // (constructed from the view name and the MetricPrefix), the Resource field, // and the set of label key/value pairs (in OpenCensus terminology: tag). // // If no custom Resource is set, a default MonitoredResource // with type global and no resource labels will be used. If you explicitly // set this field, you may also want to set custom DefaultMonitoringLabels. // // Deprecated: Use MonitoredResource instead. Resource *monitoredrespb.MonitoredResource // MonitoredResource sets the MonitoredResource against which all views will be // recorded by this exporter. // // All Stackdriver metrics created by this exporter are custom metrics, // so only a limited number of MonitoredResource types are supported, see: // https://cloud.google.com/monitoring/custom-metrics/creating-metrics#which-resource // // An important consideration when setting the MonitoredResource here is that // Stackdriver Monitoring only allows a single writer per // TimeSeries, see: https://cloud.google.com/monitoring/api/v3/metrics-details#intro-time-series // A TimeSeries is uniquely defined by the metric type name // (constructed from the view name and the MetricPrefix), the MonitoredResource field, // and the set of label key/value pairs (in OpenCensus terminology: tag). // // If no custom MonitoredResource is set AND if Resource is also not set then // a default MonitoredResource with type global and no resource labels will be used. // If you explicitly set this field, you may also want to set custom DefaultMonitoringLabels. // // This field replaces Resource field. If this is set then it will override the // Resource field. // Optional, but encouraged. MonitoredResource monitoredresource.Interface // ResourceDetector provides a hook to discover arbitrary resource information. // // The translation function provided in MapResource must be able to conver the // the resource information to a Stackdriver monitored resource. // // If this field is unset, resource type and tags will automatically be discovered through // the OC_RESOURCE_TYPE and OC_RESOURCE_LABELS environment variables. ResourceDetector resource.Detector // MapResource converts a OpenCensus resource to a Stackdriver monitored resource. // // If this field is unset, DefaultMapResource will be used which encodes a set of default // conversions from auto-detected resources to well-known Stackdriver monitored resources. MapResource func(*resource.Resource) *monitoredrespb.MonitoredResource // MetricPrefix overrides the prefix of a Stackdriver metric names. // Optional. If unset defaults to "custom.googleapis.com/opencensus/". // If GetMetricPrefix is non-nil, this option is ignored. MetricPrefix string // GetMetricDisplayName allows customizing the display name for the metric // associated with the given view. By default it will be: // MetricPrefix + view.Name GetMetricDisplayName func(view *view.View) string // GetMetricType allows customizing the metric type for the given view. // By default, it will be: // "custom.googleapis.com/opencensus/" + view.Name // // See: https://cloud.google.com/monitoring/api/ref_v3/rest/v3/projects.metricDescriptors#MetricDescriptor // Depreacted. Use GetMetricPrefix instead. GetMetricType func(view *view.View) string // GetMetricPrefix allows customizing the metric prefix for the given metric name. // If it is not set, MetricPrefix is used. If MetricPrefix is not set, it defaults to: // "custom.googleapis.com/opencensus/" // // See: https://cloud.google.com/monitoring/api/ref_v3/rest/v3/projects.metricDescriptors#MetricDescriptor GetMetricPrefix func(name string) string // DefaultTraceAttributes will be appended to every span that is exported to // Stackdriver Trace. DefaultTraceAttributes map[string]interface{} // DefaultMonitoringLabels are labels added to every metric created by this // exporter in Stackdriver Monitoring. // // If unset, this defaults to a single label with key "opencensus_task" and // value "go-<pid>@<hostname>". This default ensures that the set of labels // together with the default Resource (global) are unique to this // process, as required by Stackdriver Monitoring. // // If you set DefaultMonitoringLabels, make sure that the Resource field // together with these labels is unique to the // current process. This is to ensure that there is only a single writer to // each TimeSeries in Stackdriver. // // Set this to &Labels{} (a pointer to an empty Labels) to avoid getting the // default "opencensus_task" label. You should only do this if you know that // the Resource you set uniquely identifies this Go process. DefaultMonitoringLabels *Labels // Context allows you to provide a custom context for API calls. // // This context will be used several times: first, to create Stackdriver // trace and metric clients, and then every time a new batch of traces or // stats needs to be uploaded. // // Do not set a timeout on this context. Instead, set the Timeout option. // // If unset, context.Background() will be used. Context context.Context // SkipCMD enforces to skip all the CreateMetricDescriptor calls. // These calls are important in order to configure the unit of the metrics, // but in some cases all the exported metrics are builtin (unit is configured) // or the unit is not important. SkipCMD bool // Timeout for all API calls. If not set, defaults to 12 seconds. Timeout time.Duration // ReportingInterval sets the interval between reporting metrics. // If it is set to zero then default value is used. ReportingInterval time.Duration // NumberOfWorkers sets the number of go rountines that send requests // to Stackdriver Monitoring and Trace. The minimum number of workers is 1. NumberOfWorkers int // ResourceByDescriptor may be provided to supply monitored resource dynamically // based on the metric Descriptor. Most users will not need to set this, // but should instead set ResourceDetector. // // The MonitoredResource and ResourceDetector fields are ignored if this // field is set to a non-nil value. // // The ResourceByDescriptor is called to derive monitored resources from // metric.Descriptor and the label map associated with the time-series. // If any label is used for the derived resource then it will be removed // from the label map. The remaining labels in the map are returned to // be used with the time-series. // // If the func set to this field does not return valid resource even for one // time-series then it will result into an error for the entire CreateTimeSeries request // which may contain more than one time-series. ResourceByDescriptor func(*metricdata.Descriptor, map[string]string) (map[string]string, monitoredresource.Interface) // Override the user agent value supplied to Monitoring APIs and included as an // attribute in trace data. UserAgent string }
Options contains options for configuring the exporter.
Source Files ¶
Directories ¶
Path | Synopsis |
---|---|
examples
|
|
stats
Command stackdriver is an example program that collects data for video size.
|
Command stackdriver is an example program that collects data for video size. |
internal
|
|
Package propagation implement X-Cloud-Trace-Context header propagation used by Google Cloud products.
|
Package propagation implement X-Cloud-Trace-Context header propagation used by Google Cloud products. |