Change subsystem to type label.

Signed-off-by: Harkishen-Singh <[email protected]>

As per the design doc, now type can have ['metric', 'trace'].
This commit ensures the same, leaving subsystem for ['metric_batcher',
'copier'] which will be implemented in another PR that will be
responsible to update all metric path based metrics.

Author: harkishen

CHANGELOG.md

@@ -14,6 +14,9 @@ We use the following categories for changes:
 
 ## [Unreleased]
 
+### Added
+- Add Prometheus metrics support for Tracing [#1102]
+
 ## [0.9.0] - 2022-02-02
 
 ### Added

pkg/jaeger/query/metrics.go

@@ -13,13 +13,14 @@ import (
 )
 
 var (
+	// TODO (harkishen): update telemetry module to extract metric data from series in new consistent metrics.
 	traceRequestsExec = prometheus.NewCounter(prometheus.CounterOpts{
 		Namespace: util.PromNamespace,
 		Subsystem: "trace",
 		Name:      "query_requests_executed_total",
 		Help:      "Total number of query requests successfully executed by /getTrace and /fetchTraces API.",
 	})
-	// Even though this is handled by promscale_query_requests_total{subsystem="trace", handler="get_dependencies", code="200"}
+	// Even though this is handled by promscale_query_requests_total{type="trace", handler="get_dependencies", code="200"}
 	// yet we will have to keep this metric for telemetry as extracting the underlying series from a metric will require
 	// changing telemetry arch that tracks the all prometheus metrics, just for this metric, which is not worth.
 	dependencyRequestsExec = prometheus.NewCounter(prometheus.CounterOpts{

pkg/jaeger/query/query.go

@@ -14,6 +14,7 @@ import (
 	"github.com/jaegertracing/jaeger/model"
 	"github.com/jaegertracing/jaeger/storage/dependencystore"
 	"github.com/jaegertracing/jaeger/storage/spanstore"
+
 	"github.com/timescale/promscale/pkg/log"
 	"github.com/timescale/promscale/pkg/pgmodel/metrics"
 	"github.com/timescale/promscale/pkg/pgxconn"
@@ -44,85 +45,98 @@ func (p *Query) SpanWriter() spanstore.Writer {
 }
 
 func (p *Query) GetTrace(ctx context.Context, traceID model.TraceID) (*model.Trace, error) {
-	metrics.RequestsTotal.With(prometheus.Labels{"subsystem": "trace", "handler": "get_trace", "code": ""}).Inc()
+	code := "2xx"
 	start := time.Now()
+	defer func() {
+		metrics.RequestsTotal.With(prometheus.Labels{"type": "trace", "handler": "Get_Trace", "code": code}).Inc()
+		metrics.RequestsDuration.With(prometheus.Labels{"type": "trace", "handler": "Get_Trace", "code": code}).Observe(time.Since(start).Seconds())
+	}()
 	res, err := getTrace(ctx, p.conn, traceID)
-	if err == nil {
-		metrics.RequestsTotal.With(prometheus.Labels{"subsystem": "trace", "handler": "get_trace", "code": "200"}).Inc()
-		traceRequestsExec.Add(1)
-		metrics.RequestsDuration.With(prometheus.Labels{"subsystem": "trace", "handler": "get_trace"}).Observe(time.Since(start).Seconds())
-	} else {
-		metrics.RequestsTotal.With(prometheus.Labels{"subsystem": "trace", "handler": "get_trace", "code": "500"}).Inc()
+	if err != nil {
+		code = "5xx"
+		return nil, logError(err)
 	}
-	return res, logError(err)
+	traceRequestsExec.Add(1)
+	return res, nil
 }
 
 func (p *Query) GetServices(ctx context.Context) ([]string, error) {
-	metrics.RequestsTotal.With(prometheus.Labels{"subsystem": "trace", "handler": "get_services", "code": ""}).Inc()
+	code := "2xx"
 	start := time.Now()
+	defer func() {
+		metrics.RequestsTotal.With(prometheus.Labels{"type": "trace", "handler": "Get_Services", "code": code}).Inc()
+		metrics.RequestsDuration.With(prometheus.Labels{"type": "trace", "handler": "Get_Services", "code": code}).Observe(time.Since(start).Seconds())
+	}()
 	res, err := getServices(ctx, p.conn)
-	if err == nil {
-		metrics.RequestsTotal.With(prometheus.Labels{"subsystem": "trace", "handler": "get_services", "code": "200"}).Inc()
-		metrics.RequestsDuration.With(prometheus.Labels{"subsystem": "trace", "handler": "get_services"}).Observe(time.Since(start).Seconds())
-	} else {
-		metrics.RequestsTotal.With(prometheus.Labels{"subsystem": "trace", "handler": "get_services", "code": "500"}).Inc()
+	if err != nil {
+		code = "5xx"
+		return nil, logError(err)
 	}
-	return res, logError(err)
+	return res, nil
 }
 
 func (p *Query) GetOperations(ctx context.Context, query spanstore.OperationQueryParameters) ([]spanstore.Operation, error) {
-	metrics.RequestsTotal.With(prometheus.Labels{"subsystem": "trace", "handler": "get_operations", "code": ""}).Inc()
+	code := "2xx"
 	start := time.Now()
+	defer func() {
+		metrics.RequestsTotal.With(prometheus.Labels{"type": "trace", "handler": "Get_Operations", "code": code}).Inc()
+		metrics.RequestsDuration.With(prometheus.Labels{"type": "trace", "handler": "Get_Operations", "code": code}).Observe(time.Since(start).Seconds())
+	}()
 	res, err := getOperations(ctx, p.conn, query)
-	if err == nil {
-		metrics.RequestsTotal.With(prometheus.Labels{"subsystem": "trace", "handler": "get_operations", "code": "200"}).Inc()
-		metrics.RequestsDuration.With(prometheus.Labels{"subsystem": "trace", "handler": "get_operations"}).Observe(time.Since(start).Seconds())
-	} else {
-		metrics.RequestsTotal.With(prometheus.Labels{"subsystem": "trace", "handler": "get_operations", "code": "500"}).Inc()
+	if err != nil {
+		code = "5xx"
+		return nil, logError(err)
 	}
-	return res, logError(err)
+	return res, nil
 }
 
 func (p *Query) FindTraces(ctx context.Context, query *spanstore.TraceQueryParameters) ([]*model.Trace, error) {
-	metrics.RequestsTotal.With(prometheus.Labels{"subsystem": "trace", "handler": "find_traces", "code": ""}).Inc()
+	code := "2xx"
 	start := time.Now()
+	defer func() {
+		metrics.RequestsTotal.With(prometheus.Labels{"type": "trace", "handler": "Find_Traces", "code": code}).Inc()
+		metrics.RequestsDuration.With(prometheus.Labels{"type": "trace", "handler": "Find_Traces", "code": code}).Observe(time.Since(start).Seconds())
+	}()
 	res, err := findTraces(ctx, p.conn, query)
-	if err == nil {
-		metrics.RequestsTotal.With(prometheus.Labels{"subsystem": "trace", "handler": "find_traces", "code": "200"}).Inc()
-		traceRequestsExec.Add(1)
-		metrics.RequestsDuration.With(prometheus.Labels{"subsystem": "trace", "handler": "find_traces"}).Observe(time.Since(start).Seconds())
-	} else {
-		metrics.RequestsTotal.With(prometheus.Labels{"subsystem": "trace", "handler": "find_traces", "code": "500"}).Inc()
+	if err != nil {
+		code = "5xx"
+		return nil, logError(err)
 	}
-	return res, logError(err)
+	traceRequestsExec.Add(1)
+	return res, nil
 }
 
 func (p *Query) FindTraceIDs(ctx context.Context, query *spanstore.TraceQueryParameters) ([]model.TraceID, error) {
-	metrics.RequestsTotal.With(prometheus.Labels{"subsystem": "trace", "handler": "find_trace_ids", "code": ""}).Inc()
+	code := "2xx"
 	start := time.Now()
+	defer func() {
+		metrics.RequestsTotal.With(prometheus.Labels{"type": "trace", "handler": "Find_Trace_IDs", "code": code}).Inc()
+		metrics.RequestsDuration.With(prometheus.Labels{"type": "trace", "handler": "Find_Trace_IDs", "code": code}).Observe(time.Since(start).Seconds())
+	}()
 	res, err := findTraceIDs(ctx, p.conn, query)
-	if err == nil {
-		metrics.RequestsTotal.With(prometheus.Labels{"subsystem": "trace", "handler": "find_trace_ids", "code": "200"}).Inc()
-		traceRequestsExec.Add(1)
-		metrics.RequestsDuration.With(prometheus.Labels{"subsystem": "trace", "handler": "find_trace_ids"}).Observe(time.Since(start).Seconds())
-	} else {
-		metrics.RequestsTotal.With(prometheus.Labels{"subsystem": "trace", "handler": "find_trace_ids", "code": "500"}).Inc()
+	if err != nil {
+		code = "5xx"
+		return nil, logError(err)
 	}
-	return res, logError(err)
+	traceRequestsExec.Add(1)
+	return res, nil
 }
 
 func (p *Query) GetDependencies(ctx context.Context, endTs time.Time, lookback time.Duration) ([]model.DependencyLink, error) {
-	metrics.RequestsTotal.With(prometheus.Labels{"subsystem": "trace", "handler": "get_dependencies", "code": ""}).Inc()
+	code := "2xx"
 	start := time.Now()
+	defer func() {
+		metrics.RequestsTotal.With(prometheus.Labels{"type": "trace", "handler": "Get_Dependencies", "code": code}).Inc()
+		metrics.RequestsDuration.With(prometheus.Labels{"type": "trace", "handler": "Get_Dependencies", "code": code}).Observe(time.Since(start).Seconds())
+	}()
+
 	res, err := getDependencies(ctx, p.conn, endTs, lookback)
-	if err == nil {
-		metrics.RequestsTotal.With(prometheus.Labels{"subsystem": "trace", "handler": "get_dependencies", "code": "200"}).Inc()
-		dependencyRequestsExec.Add(1)
-		metrics.RequestsDuration.With(prometheus.Labels{"subsystem": "trace", "handler": "get_dependencies"}).Observe(time.Since(start).Seconds())
-	} else {
-		metrics.RequestsTotal.With(prometheus.Labels{"subsystem": "trace", "handler": "get_dependencies", "code": "500"}).Inc()
+	if err != nil {
+		code = "5xx"
+		return nil, logError(err)
 	}
-	return res, logError(err)
+	dependencyRequestsExec.Add(1)
+	return res, nil
 }
 
 func logError(err error) error {