add host_source label to track host identifier attribute (grafana#5152)

* add host_source label to track host identifier attribute

* update changelog

* fix lint

Author: rlankfo

CHANGELOG.md

@@ -75,6 +75,7 @@ configurable via the throughput_bytes_slo field, and it will populate op="traces
 * [ENHANCEMENT] Support TraceQL Metrics checks in Vulture [#4886](https://github.com/grafana/tempo/pull/4886) (@ruslan-mikhailov)
 * [ENHANCEMENT] Add memcached to the resources dashboard [#5049](https://github.com/grafana/tempo/pull/5049) (@javiermolinar)
 * [ENHANCEMENT] Query-frontend: logs add msg to the log line [#4975](https://github.com/grafana/tempo/pull/4975) (@jmichalek132)
+* [ENHANCEMENT] Host Info Processor: track host identifying resource attribute in metric [#5152](https://github.com/grafana/tempo/pull/5152) (@rlankfo)
 * [BUGFIX] Choose a default step for a gRPC streaming query range request if none is provided. [#4546](https://github.com/grafana/tempo/pull/4576) (@joe-elliott)
   Correctly copy exemplars for metrics like `| rate()` when gRPC streaming.
 * [BUGFIX] Make comparison to nil symmetric [#4869](https://github.com/grafana/tempo/pull/4869) (@stoewer)

modules/generator/processor/hostinfo/processor.go

@@ -14,6 +14,7 @@ const (
 
 	hostInfoMetric     = "traces_host_info"
 	hostIdentifierAttr = "grafana_host_id"
+	hostSourceAttr     = "host_source"
 )
 
 type Processor struct {
@@ -30,28 +31,30 @@ func (p *Processor) Name() string {
 	return Name
 }
 
-func (p *Processor) findHostIdentifier(resourceSpans *v1.ResourceSpans) string {
+func (p *Processor) findHostIdentifier(resourceSpans *v1.ResourceSpans) (string, string) {
 	attrs := resourceSpans.GetResource().GetAttributes()
 	for _, idAttr := range p.Cfg.HostIdentifiers {
 		for _, attr := range attrs {
-			if attr.GetKey() == idAttr {
+			hostSource := attr.GetKey()
+			if hostSource == idAttr {
 				if val := attr.GetValue(); val != nil {
 					if strVal := val.GetStringValue(); strVal != "" {
-						return strVal
+						return strVal, hostSource
 					}
 				}
 			}
 		}
 	}
-	return ""
+	return "", ""
 }
 
 func (p *Processor) PushSpans(_ context.Context, req *tempopb.PushSpansRequest) {
-	values := make([]string, 1)
+	values := make([]string, 2)
 	for i := range req.Batches {
 		resourceSpans := req.Batches[i]
-		if hostID := p.findHostIdentifier(resourceSpans); hostID != "" {
+		if hostID, hostSource := p.findHostIdentifier(resourceSpans); hostID != "" && hostSource != "" {
 			values[0] = hostID
+			values[1] = hostSource
 			labelValues := p.registry.NewLabelValueCombo(
 				p.labels,
 				values,
@@ -64,8 +67,9 @@ func (p *Processor) PushSpans(_ context.Context, req *tempopb.PushSpansRequest)
 func (p *Processor) Shutdown(_ context.Context) {}
 
 func New(cfg Config, reg registry.Registry, logger log.Logger) (*Processor, error) {
-	labels := make([]string, 1)
+	labels := make([]string, 2)
 	labels[0] = hostIdentifierAttr
+	labels[1] = hostSourceAttr
 	p := &Processor{
 		Cfg:        cfg,
 		logger:     logger,

modules/generator/processor/hostinfo/processor_test.go

@@ -17,8 +17,8 @@ import (
 
 func TestHostInfo(t *testing.T) {
 	testRegistry := registry.NewTestRegistry()
-
 	cfg := Config{}
+
 	cfg.RegisterFlagsAndApplyDefaults("", nil)
 	p, err := New(cfg, testRegistry, nil)
 	require.NoError(t, err)
@@ -42,11 +42,64 @@ func TestHostInfo(t *testing.T) {
 
 	lbls0 := labels.FromMap(map[string]string{
 		hostIdentifierAttr: "test0",
+		hostSourceAttr:     "host.id",
 	})
 	assert.Equal(t, 1.0, testRegistry.Query(hostInfoMetric, lbls0))
 
 	lbls1 := labels.FromMap(map[string]string{
 		hostIdentifierAttr: "test1",
+		hostSourceAttr:     "host.id",
 	})
 	assert.Equal(t, 1.0, testRegistry.Query(hostInfoMetric, lbls1))
 }
+
+func TestHostInfoHostSource(t *testing.T) {
+	testRegistry := registry.NewTestRegistry()
+
+	cfg := Config{}
+	cfg.RegisterFlagsAndApplyDefaults("", nil)
+	p, err := New(cfg, testRegistry, nil)
+	require.NoError(t, err)
+	require.Equal(t, p.Name(), Name)
+	defer p.Shutdown(context.TODO())
+
+	req := &tempopb.PushSpansRequest{
+		Batches: []*trace_v1.ResourceSpans{
+			test.MakeBatch(10, nil),
+			test.MakeBatch(10, nil),
+		},
+	}
+
+	for i, b := range req.Batches {
+		if i%2 == 0 {
+			b.Resource.Attributes = append(b.Resource.Attributes, []*common_v1.KeyValue{
+				{Key: "k8s.node.name", Value: &common_v1.AnyValue{Value: &common_v1.AnyValue_StringValue{StringValue: "test" + strconv.Itoa(i)}}},
+			}...)
+		}
+		b.Resource.Attributes = append(b.Resource.Attributes, []*common_v1.KeyValue{
+			{Key: "host.id", Value: &common_v1.AnyValue{Value: &common_v1.AnyValue_StringValue{StringValue: "test" + strconv.Itoa(i)}}},
+		}...)
+	}
+
+	p.PushSpans(context.Background(), req)
+
+	for i := range len(req.Batches) {
+		hostIDLbls := labels.FromMap(map[string]string{
+			hostIdentifierAttr: "test" + strconv.Itoa(i),
+			hostSourceAttr:     "host.id",
+		})
+		k8sNodeNameLbls := labels.FromMap(map[string]string{
+			hostIdentifierAttr: "test" + strconv.Itoa(i),
+			hostSourceAttr:     "k8s.node.name",
+		})
+
+		if i%2 == 0 {
+			assert.Equal(t, 0.0, testRegistry.Query(hostInfoMetric, hostIDLbls))
+			assert.Equal(t, 1.0, testRegistry.Query(hostInfoMetric, k8sNodeNameLbls))
+		} else {
+			assert.Equal(t, 1.0, testRegistry.Query(hostInfoMetric, hostIDLbls))
+			assert.Equal(t, 0.0, testRegistry.Query(hostInfoMetric, k8sNodeNameLbls))
+		}
+
+	}
+}