*: avoid missed Alertmanager targets (#6455)

This change makes sure that nearly-identical Alertmanager configurations
aren't merged together.

The config's identifier was the MD5 hash of the configuration serialized
to JSON but because `relabel.Regexp` has no public field and doesn't
implement the JSON.Marshaler interface, it was always serialized to
"{}".

In practice, the identifier can be based on the index of the
configuration in the list.

Signed-off-by: Simon Pasquier <spasquie@redhat.com>

cmd/prometheus/main.go

@@ -16,8 +16,6 @@ package main
 
 import (
 	"context"
-	"crypto/md5"
-	"encoding/json"
 	"fmt"
 	"net"
 	"net/http"
@@ -435,13 +433,8 @@ func main() {
 		notifierManager.ApplyConfig,
 		func(cfg *config.Config) error {
 			c := make(map[string]sd_config.ServiceDiscoveryConfig)
-			for _, v := range cfg.AlertingConfig.AlertmanagerConfigs {
+			for k, v := range cfg.AlertingConfig.AlertmanagerConfigs.ToMap() {
-				// AlertmanagerConfigs doesn't hold an unique identifier so we use the config hash as the identifier.
+				c[k] = v.ServiceDiscoveryConfig
-				b, err := json.Marshal(v)
-				if err != nil {
-					return err
-				}
-				c[fmt.Sprintf("%x", md5.Sum(b))] = v.ServiceDiscoveryConfig
 			}
 			return discoveryManagerNotify.ApplyConfig(c)
 		},

config/config.go

@@ -432,8 +432,8 @@ func (c *ScrapeConfig) UnmarshalYAML(unmarshal func(interface{}) error) error {
 
 // AlertingConfig configures alerting and alertmanager related configs.
 type AlertingConfig struct {
-	AlertRelabelConfigs []*relabel.Config     `yaml:"alert_relabel_configs,omitempty"`
+	AlertRelabelConfigs []*relabel.Config   `yaml:"alert_relabel_configs,omitempty"`
-	AlertmanagerConfigs []*AlertmanagerConfig `yaml:"alertmanagers,omitempty"`
+	AlertmanagerConfigs AlertmanagerConfigs `yaml:"alertmanagers,omitempty"`
 }
 
 // UnmarshalYAML implements the yaml.Unmarshaler interface.
@@ -454,6 +454,18 @@ func (c *AlertingConfig) UnmarshalYAML(unmarshal func(interface{}) error) error
 	return nil
 }
 
+// AlertmanagerConfigs is a slice of *AlertmanagerConfig.
+type AlertmanagerConfigs []*AlertmanagerConfig
+
+// ToMap converts a slice of *AlertmanagerConfig to a map.
+func (a AlertmanagerConfigs) ToMap() map[string]*AlertmanagerConfig {
+	ret := make(map[string]*AlertmanagerConfig)
+	for i := range a {
+		ret[fmt.Sprintf("config-%d", i)] = a[i]
+	}
+	return ret
+}
+
 // AlertmanagerAPIVersion represents a version of the
 // github.com/prometheus/alertmanager/api, e.g. 'v1' or 'v2'.
 type AlertmanagerAPIVersion string

notifier/notifier.go

@@ -16,7 +16,6 @@ package notifier
 import (
 	"bytes"
 	"context"
-	"crypto/md5"
 	"encoding/json"
 	"fmt"
 	"io"
@@ -262,20 +261,13 @@ func (n *Manager) ApplyConfig(conf *config.Config) error {
 
 	amSets := make(map[string]*alertmanagerSet)
 
-	for _, cfg := range conf.AlertingConfig.AlertmanagerConfigs {
+	for k, cfg := range conf.AlertingConfig.AlertmanagerConfigs.ToMap() {
-		ams, err := newAlertmanagerSet(cfg, n.logger)
+		ams, err := newAlertmanagerSet(cfg, n.logger, n.metrics)
 		if err != nil {
 			return err
 		}
 
-		ams.metrics = n.metrics
+		amSets[k] = ams
-
-		// The config hash is used for the map lookup identifier.
-		b, err := json.Marshal(cfg)
-		if err != nil {
-			return err
-		}
-		amSets[fmt.Sprintf("%x", md5.Sum(b))] = ams
 	}
 
 	n.alertmanagers = amSets
@@ -638,15 +630,16 @@ type alertmanagerSet struct {
 	logger     log.Logger
 }
 
-func newAlertmanagerSet(cfg *config.AlertmanagerConfig, logger log.Logger) (*alertmanagerSet, error) {
+func newAlertmanagerSet(cfg *config.AlertmanagerConfig, logger log.Logger, metrics *alertMetrics) (*alertmanagerSet, error) {
 	client, err := config_util.NewClientFromConfig(cfg.HTTPClientConfig, "alertmanager", false)
 	if err != nil {
 		return nil, err
 	}
 	s := &alertmanagerSet{
-		client: client,
+		client:  client,
-		cfg:    cfg,
+		cfg:     cfg,
-		logger: logger,
+		logger:  logger,
+		metrics: metrics,
 	}
 	return s, nil
 }

notifier/notifier_test.go

@@ -16,7 +16,6 @@ package notifier
 import (
 	"bytes"
 	"context"
-	"crypto/md5"
 	"encoding/json"
 	"fmt"
 	"io/ioutil"
@@ -467,6 +466,7 @@ alerting:
 	if err := yaml.UnmarshalStrict([]byte(s), cfg); err != nil {
 		t.Fatalf("Unable to load YAML config: %s", err)
 	}
+	testutil.Equals(t, 1, len(cfg.AlertingConfig.AlertmanagerConfigs))
 
 	if err := n.ApplyConfig(cfg); err != nil {
 		t.Fatalf("Error Applying the config:%v", err)
@@ -474,18 +474,16 @@ alerting:
 
 	tgs := make(map[string][]*targetgroup.Group)
 	for _, tt := range tests {
-
+		for k := range cfg.AlertingConfig.AlertmanagerConfigs.ToMap() {
-		b, err := json.Marshal(cfg.AlertingConfig.AlertmanagerConfigs[0])
+			tgs[k] = []*targetgroup.Group{
-		if err != nil {
+				tt.in,
-			t.Fatalf("Error creating config hash:%v", err)
+			}
-		}
+			break
-		tgs[fmt.Sprintf("%x", md5.Sum(b))] = []*targetgroup.Group{
-			tt.in,
 		}
 		n.reload(tgs)
 		res := n.Alertmanagers()[0].String()
 
-		testutil.Equals(t, res, tt.out)
+		testutil.Equals(t, tt.out, res)
 	}
 
 }
@@ -522,6 +520,7 @@ alerting:
 	if err := yaml.UnmarshalStrict([]byte(s), cfg); err != nil {
 		t.Fatalf("Unable to load YAML config: %s", err)
 	}
+	testutil.Equals(t, 1, len(cfg.AlertingConfig.AlertmanagerConfigs))
 
 	if err := n.ApplyConfig(cfg); err != nil {
 		t.Fatalf("Error Applying the config:%v", err)
@@ -529,20 +528,18 @@ alerting:
 
 	tgs := make(map[string][]*targetgroup.Group)
 	for _, tt := range tests {
+		for k := range cfg.AlertingConfig.AlertmanagerConfigs.ToMap() {
+			tgs[k] = []*targetgroup.Group{
+				tt.in,
+			}
+			break
+		}
 
-		b, err := json.Marshal(cfg.AlertingConfig.AlertmanagerConfigs[0])
-		if err != nil {
-			t.Fatalf("Error creating config hash:%v", err)
-		}
-		tgs[fmt.Sprintf("%x", md5.Sum(b))] = []*targetgroup.Group{
-			tt.in,
-		}
 		n.reload(tgs)
 		res := n.DroppedAlertmanagers()[0].String()
 
 		testutil.Equals(t, res, tt.out)
 	}
-
 }
 
 func makeInputTargetGroup() *targetgroup.Group {