fix comments

Signed-off-by: Yijie Qin <qinyijie@amazon.com>
prometheus · Nov 1, 2022 · adf583d · adf583d
1 parent 839b227
commit adf583d
Show file tree

Hide file tree

Showing 3 changed files with 22 additions and 8 deletions.
diff --git a/notify/notify.go b/notify/notify.go
@@ -265,7 +265,7 @@ func NewMetrics(r prometheus.Registerer) *Metrics {
 			Namespace: "alertmanager",
 			Name:      "notifications_failed_total",
 			Help:      "The total number of failed notifications.",
-		}, []string{"integration", "statusCode"}),
+		}, []string{"integration", "code"}),
 		numNotificationRequestsTotal: prometheus.NewCounterVec(prometheus.CounterOpts{
 			Namespace: "alertmanager",
 			Name:      "notification_requests_total",
@@ -300,6 +300,10 @@ func NewMetrics(r prometheus.Registerer) *Metrics {
 		m.numNotificationRequestsTotal.WithLabelValues(integration)
 		m.numNotificationRequestsFailedTotal.WithLabelValues(integration)
 		m.notificationLatencySeconds.WithLabelValues(integration)
+
+		for _, code := range PossibleFailureStatusCategory {
+			m.numTotalFailedNotifications.WithLabelValues(integration, code)
+		}
 	}
 	r.MustRegister(
 		m.numNotifications, m.numTotalFailedNotifications,

diff --git a/notify/notify_test.go b/notify/notify_test.go
@@ -429,10 +429,11 @@ func TestRetryStageWithErrorCode(t *testing.T) {
 		codelabel     string
 		expectedCount int
 	}{
-		"for 400": {errorcode: 400, codelabel: "4xx", expectedCount: 1},
-		"for 402": {errorcode: 402, codelabel: "4xx", expectedCount: 1},
-		"for 500": {errorcode: 500, codelabel: "5xx", expectedCount: 1},
-		"for 502": {errorcode: 502, codelabel: "5xx", expectedCount: 1},
+		"for 400":               {errorcode: 400, codelabel: failure4xxCategoryCode, expectedCount: 1},
+		"for 402":               {errorcode: 402, codelabel: failure4xxCategoryCode, expectedCount: 1},
+		"for 500":               {errorcode: 500, codelabel: failure5xxCategoryCode, expectedCount: 1},
+		"for 502":               {errorcode: 502, codelabel: failure5xxCategoryCode, expectedCount: 1},
+		"for expected code 100": {errorcode: 100, codelabel: failure5xxCategoryCode, expectedCount: 1},
 	}
 	for _, testData := range testcases {
 		fail, retry := true, false

diff --git a/notify/util.go b/notify/util.go
@@ -30,9 +30,18 @@ import (
 	"github.com/prometheus/alertmanager/types"
 )
 
+const (
+	failure4xxCategoryCode     = "4xx"
+	failure5xxCategoryCode     = "5xx"
+	failureUnknownCategoryCode = "unknown"
+)
+
 // UserAgentHeader is the default User-Agent for notification requests
 var UserAgentHeader = fmt.Sprintf("Alertmanager/%s", version.Version)
 
+// PossibleFailureStatusCategory is a list of possible failure status code category
+var PossibleFailureStatusCategory = []string{failure4xxCategoryCode, failure5xxCategoryCode, failureUnknownCategoryCode}
+
 // RedactURL removes the URL part from an error of *url.Error type.
 func RedactURL(err error) error {
 	e, ok := err.(*url.Error)
@@ -172,13 +181,13 @@ func readAll(r io.Reader) string {
 // other than 4xx and 5xx input status will return an error.
 func getFailureStatusCodeCategory(statusCode int) (string, error) {
 	if statusCode/100 == 4 {
-		return "4xx", nil
+		return failure4xxCategoryCode, nil
 	}
 	if statusCode/100 == 5 {
-		return "5xx", nil
+		return failure5xxCategoryCode, nil
 	}
 
-	return "", fmt.Errorf("unexpected status code %v", statusCode)
+	return failureUnknownCategoryCode, fmt.Errorf("unexpected status code %v", statusCode)
 }
 
 // Retrier knows when to retry an HTTP request to a receiver. 2xx status codes