Merge pull request #544 from q3k/feature/hide-instance-auth

Hide HTTP auth parts from instance label

retrieval/target.go

@@ -17,6 +17,7 @@ import (
 	"fmt"
 	"math/rand"
 	"net/http"
+	"net/url"
 	"os"
 	"strings"
 	"sync"
@@ -114,6 +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
+	// 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
@@ -186,8 +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.URL())
-	durationMetric[InstanceLabel] = clientmodel.LabelValue(t.URL())
+	healthMetric[InstanceLabel] = clientmodel.LabelValue(t.InstanceIdentifier())
+	durationMetric[InstanceLabel] = clientmodel.LabelValue(t.InstanceIdentifier())
 
 	healthValue := clientmodel.SampleValue(0)
 	if healthy {
@@ -320,7 +323,7 @@ func (t *target) scrape(ingester extraction.Ingester) (err error) {
 		return err
 	}
 
-	baseLabels := clientmodel.LabelSet{InstanceLabel: clientmodel.LabelValue(t.URL())}
+	baseLabels := clientmodel.LabelSet{InstanceLabel: clientmodel.LabelValue(t.InstanceIdentifier())}
 	for baseLabel, baseValue := range t.baseLabels {
 		baseLabels[baseLabel] = baseValue
 	}
@@ -363,6 +366,28 @@ func (t *target) URL() string {
 	return t.url
 }
 
+// InstanceIdentifier implements Target.
+func (t *target) InstanceIdentifier() string {
+	u, err := url.Parse(t.url)
+	if err != nil {
+		glog.Warningf("Could not parse instance URL when generating identifier, using raw URL: %s", err)
+		return t.url
+	}
+	// 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.
 func (t *target) GlobalURL() string {
 	url := t.url

retrieval/target_test.go

@@ -34,6 +34,26 @@ func (i *collectResultIngester) Ingest(s clientmodel.Samples) error {
 	return nil
 }
 
+func TestTargetHidesURLAuth(t *testing.T) {
+	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.")
+	}
+
+	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)
+		}
+	}
+}
+
 func TestTargetScrapeUpdatesState(t *testing.T) {
 	testTarget := target{
 		state:      Unknown,
@@ -93,7 +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,
@@ -108,7 +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

@@ -42,6 +42,10 @@ func (t fakeTarget) URL() string {
 	return "fake"
 }
 
+func (t fakeTarget) InstanceIdentifier() string {
+	return "fake"
+}
+
 func (t fakeTarget) GlobalURL() string {
 	return t.URL()
 }