Change instance identifiers to be host:port

This changes the PublicURL function into InstanceIdentifier, which now
returns a simple <host>:<port> string instead of a full URL.

retrieval/target.go

@@ -115,8 +115,8 @@ type Target interface {
 	// points in this interface, this one is the best candidate to change given
 	// the ways to express the endpoint.
 	URL() string
-	// The URL without any auth data in it, used to label data
-	PublicURL() string
+	// Used to populate the `instance` label in metrics.
+	InstanceIdentifier() string
 	// The URL as seen from other hosts. References to localhost are resolved
 	// to the address of the prometheus server.
 	GlobalURL() string
@@ -189,8 +189,8 @@ func (t *target) recordScrapeHealth(ingester extraction.Ingester, timestamp clie
 	}
 	healthMetric[clientmodel.MetricNameLabel] = clientmodel.LabelValue(scrapeHealthMetricName)
 	durationMetric[clientmodel.MetricNameLabel] = clientmodel.LabelValue(scrapeDurationMetricName)
-	healthMetric[InstanceLabel] = clientmodel.LabelValue(t.PublicURL())
-	durationMetric[InstanceLabel] = clientmodel.LabelValue(t.PublicURL())
+	healthMetric[InstanceLabel] = clientmodel.LabelValue(t.InstanceIdentifier())
+	durationMetric[InstanceLabel] = clientmodel.LabelValue(t.InstanceIdentifier())
 
 	healthValue := clientmodel.SampleValue(0)
 	if healthy {
@@ -323,7 +323,7 @@ func (t *target) scrape(ingester extraction.Ingester) (err error) {
 		return err
 	}
 
-	baseLabels := clientmodel.LabelSet{InstanceLabel: clientmodel.LabelValue(t.PublicURL())}
+	baseLabels := clientmodel.LabelSet{InstanceLabel: clientmodel.LabelValue(t.InstanceIdentifier())}
 	for baseLabel, baseValue := range t.baseLabels {
 		baseLabels[baseLabel] = baseValue
 	}
@@ -366,15 +366,26 @@ func (t *target) URL() string {
 	return t.url
 }
 
-// PublicURL implements Target.
-func (t *target) PublicURL() string {
+// InstanceIdentifier implements Target.
+func (t *target) InstanceIdentifier() string {
 	u, err := url.Parse(t.url)
 	if err != nil {
-		glog.Warning("Could not parse URL for auth stripping (%s), returning it unchanged", err)
+		glog.Warningf("Could not parse instance URL when generating identifier, using raw URL: %s", err)
 		return t.url
 	}
-	u.User = nil
-	return u.String()
+	// If we are given a port in the host port, use that.
+	if strings.Contains(u.Host, ":") {
+		return u.Host
+	}
+	// Otherwise, deduce port based on protocol.
+	if u.Scheme == "http" {
+		return fmt.Sprintf("%s:80", u.Host)
+	} else if u.Scheme == "https" {
+		return fmt.Sprintf("%s:443", u.Host)
+	}
+
+	glog.Warningf("Unknown scheme %s when generating identifier, using raw URL.", u.Scheme)
+	return t.url
 }
 
 // GlobalURL implements Target.

retrieval/target_test.go

@@ -35,15 +35,22 @@ func (i *collectResultIngester) Ingest(s clientmodel.Samples) error {
 }
 
 func TestTargetHidesURLAuth(t *testing.T) {
-	testTarget := target{
-		state:      Unknown,
-		url:        "http://secret:data@host.com/query?args#fragment",
-		httpClient: utility.NewDeadlineClient(0),
+	testVectors := []string{"http://secret:data@host.com/query?args#fragment", "https://example.net/foo", "http://foo.com:31337/bar"}
+	testResults := []string{"host.com:80", "example.net:443", "foo.com:31337"}
+	if len(testVectors) != len(testResults) {
+		t.Errorf("Test vector length does not match test result length.")
 	}
-	expected := "http://host.com/query?args#fragment"
-	u := testTarget.PublicURL()
-	if u != expected {
-		t.Errorf("Expected PublicURL to be %v, actual %v", expected, u)
+
+	for i := 0; i < len(testVectors); i++ {
+		testTarget := target{
+			state:      Unknown,
+			url:        testVectors[i],
+			httpClient: utility.NewDeadlineClient(0),
+		}
+		u := testTarget.InstanceIdentifier()
+		if u != testResults[i] {
+			t.Errorf("Expected InstanceIdentifier to be %v, actual %v", testResults[i], u)
+		}
 	}
 }
 
@@ -106,7 +113,7 @@ func TestTargetRecordScrapeHealth(t *testing.T) {
 	expected := &clientmodel.Sample{
 		Metric: clientmodel.Metric{
 			clientmodel.MetricNameLabel: scrapeHealthMetricName,
-			InstanceLabel:               "http://example.url",
+			InstanceLabel:               "example.url:80",
 			clientmodel.JobLabel:        "testjob",
 		},
 		Timestamp: now,
@@ -121,7 +128,7 @@ func TestTargetRecordScrapeHealth(t *testing.T) {
 	expected = &clientmodel.Sample{
 		Metric: clientmodel.Metric{
 			clientmodel.MetricNameLabel: scrapeDurationMetricName,
-			InstanceLabel:               "http://example.url",
+			InstanceLabel:               "example.url:80",
 			clientmodel.JobLabel:        "testjob",
 		},
 		Timestamp: now,

retrieval/targetmanager_test.go

@@ -42,7 +42,7 @@ func (t fakeTarget) URL() string {
 	return "fake"
 }
 
-func (t fakeTarget) PublicURL() string {
+func (t fakeTarget) InstanceIdentifier() string {
 	return "fake"
 }