The open and composable observability and data visualization platform. Visualize metrics, logs, and traces from multiple sources like Prometheus, Loki, Elasticsearch, InfluxDB, Postgres and many more.
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
 
 
 
grafana/pkg/tsdb/cloudmonitoring/time_series_query.go

363 lines
12 KiB

package cloudmonitoring
import (
"bytes"
"context"
"encoding/json"
"fmt"
"net/url"
"path"
"strconv"
"strings"
"time"
"go.opentelemetry.io/otel/attribute"
"github.com/grafana/grafana-plugin-sdk-go/backend"
"github.com/grafana/grafana-plugin-sdk-go/data"
"github.com/grafana/grafana/pkg/infra/tracing"
"github.com/grafana/grafana/pkg/tsdb/intervalv2"
)
func (timeSeriesQuery cloudMonitoringTimeSeriesQuery) run(ctx context.Context, req *backend.QueryDataRequest,
s *Service, dsInfo datasourceInfo, tracer tracing.Tracer) (*backend.DataResponse, cloudMonitoringResponse, string, error) {
dr := &backend.DataResponse{}
projectName := timeSeriesQuery.ProjectName
if projectName == "" {
var err error
projectName, err = s.getDefaultProject(ctx, dsInfo)
if err != nil {
dr.Error = err
return dr, cloudMonitoringResponse{}, "", nil
}
slog.Info("No project name set on query, using project name from datasource", "projectName", projectName)
}
intervalCalculator := intervalv2.NewCalculator(intervalv2.CalculatorOptions{})
interval := intervalCalculator.Calculate(req.Queries[0].TimeRange, time.Duration(timeSeriesQuery.IntervalMS/1000)*time.Second, req.Queries[0].MaxDataPoints)
from := req.Queries[0].TimeRange.From
to := req.Queries[0].TimeRange.To
timeFormat := "2006/01/02-15:04:05"
timeSeriesQuery.Query += fmt.Sprintf(" | graph_period %s | within d'%s', d'%s'", interval.Text, from.UTC().Format(timeFormat), to.UTC().Format(timeFormat))
buf, err := json.Marshal(map[string]interface{}{
"query": timeSeriesQuery.Query,
})
if err != nil {
dr.Error = err
return dr, cloudMonitoringResponse{}, "", nil
}
r, err := s.createRequest(ctx, &dsInfo, path.Join("/v3/projects", projectName, "timeSeries:query"), bytes.NewBuffer(buf))
if err != nil {
dr.Error = err
return dr, cloudMonitoringResponse{}, "", nil
}
ctx, span := tracer.Start(ctx, "cloudMonitoring MQL query")
span.SetAttributes("query", timeSeriesQuery.Query, attribute.Key("query").String(timeSeriesQuery.Query))
span.SetAttributes("from", req.Queries[0].TimeRange.From, attribute.Key("from").String(req.Queries[0].TimeRange.From.String()))
span.SetAttributes("until", req.Queries[0].TimeRange.To, attribute.Key("until").String(req.Queries[0].TimeRange.To.String()))
defer span.End()
tracer.Inject(ctx, r.Header, span)
r = r.WithContext(ctx)
res, err := dsInfo.services[cloudMonitor].client.Do(r)
if err != nil {
dr.Error = err
return dr, cloudMonitoringResponse{}, "", nil
}
d, err := unmarshalResponse(res)
if err != nil {
dr.Error = err
return dr, cloudMonitoringResponse{}, "", nil
}
return dr, d, timeSeriesQuery.Query, nil
}
func (timeSeriesQuery cloudMonitoringTimeSeriesQuery) parseResponse(queryRes *backend.DataResponse,
response cloudMonitoringResponse, executedQueryString string) error {
frames := data.Frames{}
for _, series := range response.TimeSeriesData {
seriesLabels := make(map[string]string)
frame := data.NewFrameOfFieldTypes("", len(series.PointData), data.FieldTypeTime, data.FieldTypeFloat64)
frame.RefID = timeSeriesQuery.RefID
frame.Meta = &data.FrameMeta{
ExecutedQueryString: executedQueryString,
}
labels := make(map[string]string)
for n, d := range response.TimeSeriesDescriptor.LabelDescriptors {
key := toSnakeCase(d.Key)
key = strings.Replace(key, ".", ".label.", 1)
labelValue := series.LabelValues[n]
switch d.ValueType {
case "BOOL":
strVal := strconv.FormatBool(labelValue.BoolValue)
labels[key] = strVal
seriesLabels[key] = strVal
case "INT64":
labels[key] = labelValue.Int64Value
seriesLabels[key] = labelValue.Int64Value
default:
labels[key] = labelValue.StringValue
seriesLabels[key] = labelValue.StringValue
}
}
for n, d := range response.TimeSeriesDescriptor.PointDescriptors {
// If more than 1 pointdescriptor was returned, three aggregations are returned per time series - min, mean and max.
// This is a because the period for the given table is less than half the duration which is used in the graph_period MQL function.
// See https://cloud.google.com/monitoring/mql/reference#graph_period-tabop
// When this is the case, we'll just ignore the min and max and use the mean value in the frame
if len(response.TimeSeriesDescriptor.PointDescriptors) > 1 && !strings.HasSuffix(d.Key, ".mean") {
continue
}
labels["metric.name"] = d.Key
seriesLabels["metric.name"] = d.Key
defaultMetricName := d.Key
// process non-distribution series
if d.ValueType != "DISTRIBUTION" {
// reverse the order to be ascending
for i := len(series.PointData) - 1; i >= 0; i-- {
point := series.PointData[i]
value := point.Values[n].DoubleValue
if d.ValueType == "INT64" {
parsedValue, err := strconv.ParseFloat(point.Values[n].Int64Value, 64)
if err == nil {
value = parsedValue
}
} else if d.ValueType == "BOOL" {
if point.Values[n].BoolValue {
value = 1
} else {
value = 0
}
}
frame.SetRow(len(series.PointData)-1-i, series.PointData[i].TimeInterval.EndTime, value)
}
metricName := formatLegendKeys(d.Key, defaultMetricName, seriesLabels, nil,
&cloudMonitoringTimeSeriesFilter{
ProjectName: timeSeriesQuery.ProjectName, AliasBy: timeSeriesQuery.AliasBy,
})
dataField := frame.Fields[1]
dataField.Name = metricName
dataField.Labels = seriesLabels
setDisplayNameAsFieldName(dataField)
frames = append(frames, frame)
continue
}
// process distribution series
buckets := make(map[int]*data.Frame)
// reverse the order to be ascending
for i := len(series.PointData) - 1; i >= 0; i-- {
point := series.PointData[i]
if len(point.Values[n].DistributionValue.BucketCounts) == 0 {
continue
}
maxKey := 0
for i := 0; i < len(point.Values[n].DistributionValue.BucketCounts); i++ {
value, err := strconv.ParseFloat(point.Values[n].DistributionValue.BucketCounts[i], 64)
if err != nil {
continue
}
if _, ok := buckets[i]; !ok {
// set lower bounds
// https://cloud.google.com/monitoring/api/ref_v3/rest/v3/TimeSeries#Distribution
bucketBound := calcBucketBound(point.Values[n].DistributionValue.BucketOptions, i)
additionalLabels := map[string]string{"bucket": bucketBound}
timeField := data.NewField(data.TimeSeriesTimeFieldName, nil, []time.Time{})
valueField := data.NewField(data.TimeSeriesValueFieldName, nil, []float64{})
frameName := formatLegendKeys(d.Key, defaultMetricName, nil, additionalLabels, &cloudMonitoringTimeSeriesFilter{ProjectName: timeSeriesQuery.ProjectName, AliasBy: timeSeriesQuery.AliasBy})
valueField.Name = frameName
valueField.Labels = seriesLabels
setDisplayNameAsFieldName(valueField)
buckets[i] = &data.Frame{
Name: frameName,
Fields: []*data.Field{
timeField,
valueField,
},
RefID: timeSeriesQuery.RefID,
}
if maxKey < i {
maxKey = i
}
}
buckets[i].AppendRow(point.TimeInterval.EndTime, value)
}
// fill empty bucket
for i := 0; i < maxKey; i++ {
if _, ok := buckets[i]; !ok {
bucketBound := calcBucketBound(point.Values[n].DistributionValue.BucketOptions, i)
additionalLabels := data.Labels{"bucket": bucketBound}
timeField := data.NewField(data.TimeSeriesTimeFieldName, nil, []time.Time{})
valueField := data.NewField(data.TimeSeriesValueFieldName, nil, []float64{})
frameName := formatLegendKeys(d.Key, defaultMetricName, seriesLabels, additionalLabels, &cloudMonitoringTimeSeriesFilter{ProjectName: timeSeriesQuery.ProjectName, AliasBy: timeSeriesQuery.AliasBy})
valueField.Name = frameName
valueField.Labels = seriesLabels
setDisplayNameAsFieldName(valueField)
buckets[i] = &data.Frame{
Name: frameName,
Fields: []*data.Field{
timeField,
valueField,
},
RefID: timeSeriesQuery.RefID,
}
}
}
}
for i := 0; i < len(buckets); i++ {
frames = append(frames, buckets[i])
}
}
customFrameMeta := map[string]interface{}{}
customFrameMeta["labels"] = labels
if frame.Meta != nil {
frame.Meta.Custom = customFrameMeta
} else {
frame.SetMeta(&data.FrameMeta{Custom: customFrameMeta})
}
}
if len(response.TimeSeriesData) > 0 {
dl := timeSeriesQuery.buildDeepLink()
frames = addConfigData(frames, dl, response.Unit)
}
queryRes.Frames = frames
return nil
}
func (timeSeriesQuery cloudMonitoringTimeSeriesQuery) parseToAnnotations(queryRes *backend.DataResponse,
data cloudMonitoringResponse, title, text string) error {
annotations := make([]map[string]string, 0)
for _, series := range data.TimeSeriesData {
metricLabels := make(map[string]string)
resourceLabels := make(map[string]string)
for n, d := range data.TimeSeriesDescriptor.LabelDescriptors {
key := toSnakeCase(d.Key)
labelValue := series.LabelValues[n]
value := ""
switch d.ValueType {
case "BOOL":
strVal := strconv.FormatBool(labelValue.BoolValue)
value = strVal
case "INT64":
value = labelValue.Int64Value
default:
value = labelValue.StringValue
}
if strings.Index(key, "metric.") == 0 {
key = key[len("metric."):]
metricLabels[key] = value
} else if strings.Index(key, "resource.") == 0 {
key = key[len("resource."):]
resourceLabels[key] = value
}
}
for n, d := range data.TimeSeriesDescriptor.PointDescriptors {
// reverse the order to be ascending
for i := len(series.PointData) - 1; i >= 0; i-- {
point := series.PointData[i]
value := strconv.FormatFloat(point.Values[n].DoubleValue, 'f', 6, 64)
if d.ValueType == "STRING" {
value = point.Values[n].StringValue
}
annotation := make(map[string]string)
annotation["time"] = point.TimeInterval.EndTime.UTC().Format(time.RFC3339)
annotation["title"] = formatAnnotationText(title, value, d.MetricKind, metricLabels, resourceLabels)
annotation["tags"] = ""
annotation["text"] = formatAnnotationText(text, value, d.MetricKind, metricLabels, resourceLabels)
annotations = append(annotations, annotation)
}
}
}
timeSeriesQuery.transformAnnotationToFrame(annotations, queryRes)
return nil
}
func (timeSeriesQuery cloudMonitoringTimeSeriesQuery) buildDeepLink() string {
u, err := url.Parse("https://console.cloud.google.com/monitoring/metrics-explorer")
if err != nil {
slog.Error("Failed to generate deep link: unable to parse metrics explorer URL", "projectName", timeSeriesQuery.ProjectName, "query", timeSeriesQuery.RefID)
return ""
}
q := u.Query()
q.Set("project", timeSeriesQuery.ProjectName)
q.Set("Grafana_deeplink", "true")
pageState := map[string]interface{}{
"xyChart": map[string]interface{}{
"constantLines": []string{},
"dataSets": []map[string]interface{}{
{
"timeSeriesQuery": timeSeriesQuery.Query,
"targetAxis": "Y1",
"plotType": "LINE",
},
},
"timeshiftDuration": "0s",
"y1Axis": map[string]string{
"label": "y1Axis",
"scale": "LINEAR",
},
},
"timeSelection": map[string]string{
"timeRange": "custom",
"start": timeSeriesQuery.timeRange.From.Format(time.RFC3339Nano),
"end": timeSeriesQuery.timeRange.To.Format(time.RFC3339Nano),
},
}
blob, err := json.Marshal(pageState)
if err != nil {
slog.Error("Failed to generate deep link", "pageState", pageState, "ProjectName", timeSeriesQuery.ProjectName, "query", timeSeriesQuery.RefID)
return ""
}
q.Set("pageState", string(blob))
u.RawQuery = q.Encode()
accountChooserURL, err := url.Parse("https://accounts.google.com/AccountChooser")
if err != nil {
slog.Error("Failed to generate deep link: unable to parse account chooser URL", "ProjectName", timeSeriesQuery.ProjectName, "query", timeSeriesQuery.RefID)
return ""
}
accountChooserQuery := accountChooserURL.Query()
accountChooserQuery.Set("continue", u.String())
accountChooserURL.RawQuery = accountChooserQuery.Encode()
return accountChooserURL.String()
}
func (timeSeriesQuery cloudMonitoringTimeSeriesQuery) getRefID() string {
return timeSeriesQuery.RefID
}