upgrade alerting

go.mod

@@ -87,7 +87,7 @@ require (
 	github.com/googleapis/gax-go/v2 v2.15.0 // @grafana/grafana-backend-group
 	github.com/gorilla/mux v1.8.1 // @grafana/grafana-backend-group
 	github.com/gorilla/websocket v1.5.4-0.20250319132907-e064f32e3674 // @grafana/grafana-app-platform-squad
-	github.com/grafana/alerting v0.0.0-20251119204204-77fa75125181 // @grafana/alerting-backend
+	github.com/grafana/alerting v0.0.0-20251128214501-29edef10542b // @grafana/alerting-backend
 	github.com/grafana/authlib v0.0.0-20250930082137-a40e2c2b094f // @grafana/identity-access-team
 	github.com/grafana/authlib/types v0.0.0-20251119142549-be091cf2f4d4 // @grafana/identity-access-team
 	github.com/grafana/dataplane/examples v0.0.1 // @grafana/observability-metrics
@@ -697,7 +697,7 @@ require (
 replace github.com/crewjam/saml => github.com/grafana/saml v0.4.15-0.20240917091248-ae3bbdad8a56
 
 // Use our fork of the upstream Alertmanager.
-replace github.com/prometheus/alertmanager => github.com/grafana/prometheus-alertmanager v0.25.1-0.20250911094103-5456b6e45604
+replace github.com/prometheus/alertmanager => github.com/grafana/prometheus-alertmanager v0.25.1-0.20251128214251-5dfe7baf90af
 
 exclude github.com/mattn/go-sqlite3 v2.0.3+incompatible
 

go.sum

@@ -1611,8 +1611,10 @@ github.com/gorilla/sessions v1.2.1 h1:DHd3rPN5lE3Ts3D8rKkQ8x/0kqfeNmBAaiSi+o7Fsg
 github.com/gorilla/sessions v1.2.1/go.mod h1:dk2InVEVJ0sfLlnXv9EAgkf6ecYs/i80K/zI+bUmuGM=
 github.com/gorilla/websocket v1.5.4-0.20250319132907-e064f32e3674 h1:JeSE6pjso5THxAzdVpqr6/geYxZytqFMBCOtn/ujyeo=
 github.com/gorilla/websocket v1.5.4-0.20250319132907-e064f32e3674/go.mod h1:r4w70xmWCQKmi1ONH4KIaBptdivuRPyosB9RmPlGEwA=
-github.com/grafana/alerting v0.0.0-20251119204204-77fa75125181 h1:nbxKRtrbuhvOYmI2RhOYauHRJCtpR+vTNIgg1lFUCws=
+github.com/grafana/alerting v0.0.0-20251119223942-7e109cc4f3e9 h1:qHY1ibKD/S5CyYi1VtUGKSq5kbgA7dTv/OYtlEMklqg=
-github.com/grafana/alerting v0.0.0-20251119204204-77fa75125181/go.mod h1:VtPNIFlEOJPPEc13Ax6ZTbNV3M/sAzLID72YjgzOPVA=
+github.com/grafana/alerting v0.0.0-20251119223942-7e109cc4f3e9/go.mod h1:VtPNIFlEOJPPEc13Ax6ZTbNV3M/sAzLID72YjgzOPVA=
+github.com/grafana/alerting v0.0.0-20251128214501-29edef10542b h1:QI2xO0UVmGtkv6CuQSCaoW0uw3SV7MDsC/Mwehl75II=
+github.com/grafana/alerting v0.0.0-20251128214501-29edef10542b/go.mod h1:QAWfUswv2DaC5dD45o3vXia/QDvLi17jT4u4UToeF7A=
 github.com/grafana/authlib v0.0.0-20250930082137-a40e2c2b094f h1:Cbm6OKkOcJ+7CSZsGsEJzktC/SIa5bxVeYKQLuYK86o=
 github.com/grafana/authlib v0.0.0-20250930082137-a40e2c2b094f/go.mod h1:axY0cdOg3q0TZHwpHnIz5x16xZ8ZBxJHShsSHHXcHQg=
 github.com/grafana/authlib/types v0.0.0-20251119142549-be091cf2f4d4 h1:Muoy+FMGrHj3GdFbvsMzUT7eusgii9PKf9L1ZaXDDbY=
@@ -1665,8 +1667,8 @@ github.com/grafana/nanogit v0.3.0 h1:XNEef+4Vi+465ZITJs/g/xgnDRJbWhhJ7iQrAnWZ0oQ
 github.com/grafana/nanogit v0.3.0/go.mod h1:6s6CCTpyMOHPpcUZaLGI+rgBEKdmxVbhqSGgCK13j7Y=
 github.com/grafana/otel-profiling-go v0.5.1 h1:stVPKAFZSa7eGiqbYuG25VcqYksR6iWvF3YH66t4qL8=
 github.com/grafana/otel-profiling-go v0.5.1/go.mod h1:ftN/t5A/4gQI19/8MoWurBEtC6gFw8Dns1sJZ9W4Tls=
-github.com/grafana/prometheus-alertmanager v0.25.1-0.20250911094103-5456b6e45604 h1:aXfUhVN/Ewfpbko2CCtL65cIiGgwStOo4lWH2b6gw2U=
+github.com/grafana/prometheus-alertmanager v0.25.1-0.20251128214251-5dfe7baf90af h1:X5p+fjoW+062S7GmTIyG3c7GcSVhCPo+/2+5Hgah6AQ=
-github.com/grafana/prometheus-alertmanager v0.25.1-0.20250911094103-5456b6e45604/go.mod h1:O/QP1BCm0HHIzbKvgMzqb5sSyH88rzkFk84F4TfJjBU=
+github.com/grafana/prometheus-alertmanager v0.25.1-0.20251128214251-5dfe7baf90af/go.mod h1:O/QP1BCm0HHIzbKvgMzqb5sSyH88rzkFk84F4TfJjBU=
 github.com/grafana/pyroscope-go/godeltaprof v0.1.9 h1:c1Us8i6eSmkW+Ez05d3co8kasnuOY813tbMN8i/a3Og=
 github.com/grafana/pyroscope-go/godeltaprof v0.1.9/go.mod h1:2+l7K7twW49Ct4wFluZD3tZ6e0SjanjcUUBPVD/UuGU=
 github.com/grafana/pyroscope/api v1.2.1-0.20250415190842-3ff7247547ae h1:35W3Wjp9KWnSoV/DuymmyIj5aHE0CYlDQ5m2KeXUPAc=

go.work.sum

@@ -1932,6 +1932,7 @@ golang.org/x/sync v0.13.0/go.mod h1:1dzgHSNfp02xaA81J2MS99Qcpr2w7fw1gpm99rleRqA=
 golang.org/x/sync v0.14.0/go.mod h1:1dzgHSNfp02xaA81J2MS99Qcpr2w7fw1gpm99rleRqA=
 golang.org/x/sync v0.15.0/go.mod h1:1dzgHSNfp02xaA81J2MS99Qcpr2w7fw1gpm99rleRqA=
 golang.org/x/sync v0.16.0/go.mod h1:1dzgHSNfp02xaA81J2MS99Qcpr2w7fw1gpm99rleRqA=
+golang.org/x/sync v0.17.0/go.mod h1:9KTHXmSnoGruLpwFjVSX0lNNA75CykiMECbovNTZqGI=
 golang.org/x/sys v0.0.0-20210112080510-489259a85091/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20210503080704-8803ae5d1324/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20210616045830-e2b7044e8c71/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=

packages/grafana-data/src/types/featureToggles.gen.ts

@@ -1201,4 +1201,8 @@ export interface FeatureToggles {
   * Adds support for Kubernetes alerting historian APIs
   */
   kubernetesAlertingHistorian?: boolean;
+  /**
+  * Use synchronized dispatch timer to minimize duplicate notifications across alertmanager HA pods
+  */
+  alertingSyncDispatchTimer?: boolean;
 }

pkg/services/featuremgmt/registry.go

@@ -1984,6 +1984,14 @@ var (
 			Owner:           grafanaAlertingSquad,
 			RequiresRestart: true,
 		},
+		{
+			Name:            "alertingSyncDispatchTimer",
+			Description:     "Use synchronized dispatch timer to minimize duplicate notifications across alertmanager HA pods",
+			Stage:           FeatureStageExperimental,
+			Owner:           grafanaAlertingSquad,
+			RequiresRestart: true,
+			HideFromDocs:    true,
+		},
 	}
 )
 

pkg/services/featuremgmt/toggles_gen.csv

@@ -269,3 +269,4 @@ ttlPluginInstanceManager,experimental,@grafana/plugins-platform-backend,false,fa
 lokiQueryLimitsContext,experimental,@grafana/observability-logs,false,false,true
 rudderstackUpgrade,experimental,@grafana/grafana-frontend-platform,false,false,true
 kubernetesAlertingHistorian,experimental,@grafana/alerting-squad,false,true,false
+alertingSyncDispatchTimer,experimental,@grafana/alerting-squad,false,true,false

pkg/services/featuremgmt/toggles_gen.go

@@ -765,4 +765,8 @@ const (
 	// FlagKubernetesAlertingHistorian
 	// Adds support for Kubernetes alerting historian APIs
 	FlagKubernetesAlertingHistorian = "kubernetesAlertingHistorian"
+
+	// FlagAlertingSyncDispatchTimer
+	// Use synchronized dispatch timer to minimize duplicate notifications across alertmanager HA pods
+	FlagAlertingSyncDispatchTimer = "alertingSyncDispatchTimer"
 )

pkg/services/featuremgmt/toggles_gen.json

@@ -498,6 +498,24 @@
         "codeowner": "@grafana/alerting-squad"
       }
     },
+    {
+      "metadata": {
+        "name": "alertingSyncDispatchTimer",
+        "resourceVersion": "1764342211090",
+        "creationTimestamp": "2025-11-19T19:22:33Z",
+        "deletionTimestamp": "2025-11-27T17:14:53Z",
+        "annotations": {
+          "grafana.app/updatedTimestamp": "2025-11-28 15:03:31.090159 +0000 UTC"
+        }
+      },
+      "spec": {
+        "description": "Use synchronized dispatch timer to minimize duplicate notifications across alertmanager HA pods",
+        "stage": "experimental",
+        "codeowner": "@grafana/alerting-squad",
+        "requiresRestart": true,
+        "hideFromDocs": true
+      }
+    },
     {
       "metadata": {
         "name": "alertingTriage",

pkg/services/ngalert/ngalert.go

@@ -222,6 +222,7 @@ func (ng *AlertNG) init() error {
 			ExternalURL:       ng.Cfg.AppURL,
 			SmtpConfig:        smtpCfg,
 			Timeout:           ng.Cfg.UnifiedAlerting.RemoteAlertmanager.Timeout,
+			DispatchTimer:     notifier.GetDispatchTimer(ng.FeatureToggles).String(),
 		}
 		autogenFn := func(ctx context.Context, logger log.Logger, orgID int64, cfg *definitions.PostableApiAlertingConfig, invalidReceiverAction notifier.InvalidReceiversAction) error {
 			return notifier.AddAutogenConfig(ctx, logger, ng.store, orgID, cfg, invalidReceiverAction, ng.FeatureToggles)

pkg/services/ngalert/notifier/alertmanager.go

@@ -32,6 +32,9 @@ const (
 
 	// How long we keep silences in the kvstore after they've expired.
 	silenceRetention = 5 * 24 * time.Hour
+
+	// How long we keep flushes in the kvstore after they've expired.
+	flushRetention = 5 * 24 * time.Hour
 )
 
 type AlertingStore interface {
@@ -43,8 +46,10 @@ type AlertingStore interface {
 type stateStore interface {
 	SaveSilences(ctx context.Context, st alertingNotify.State) (int64, error)
 	SaveNotificationLog(ctx context.Context, st alertingNotify.State) (int64, error)
+	SaveFlushLog(ctx context.Context, st alertingNotify.State) (int64, error)
 	GetSilences(ctx context.Context) (string, error)
 	GetNotificationLog(ctx context.Context) (string, error)
+	GetFlushLog(ctx context.Context) (string, error)
 }
 
 type alertmanager struct {
@@ -99,6 +104,10 @@ func NewAlertmanager(ctx context.Context, orgID int64, cfg *setting.Cfg, store A
 	if err != nil {
 		return nil, err
 	}
+	flushLog, err := stateStore.GetFlushLog(ctx)
+	if err != nil {
+		return nil, err
+	}
 
 	silencesOptions := maintenanceOptions{
 		initialState:         silences,
@@ -121,12 +130,29 @@ func NewAlertmanager(ctx context.Context, orgID int64, cfg *setting.Cfg, store A
 	}
 	l := log.New("ngalert.notifier")
 
+	dispatchTimer := GetDispatchTimer(featureToggles)
+
+	var flushLogOptions maintenanceOptions
+	if dispatchTimer == alertingNotify.DispatchTimerSync {
+		flushLogOptions = maintenanceOptions{
+			initialState:         flushLog,
+			retention:            flushRetention,
+			maintenanceFrequency: maintenanceInterval,
+			maintenanceFunc: func(state alertingNotify.State) (int64, error) {
+				// Detached context here is to make sure that when the service is shut down the persist operation is executed.
+				return stateStore.SaveFlushLog(context.Background(), state)
+			},
+		}
+	}
+
 	opts := alertingNotify.GrafanaAlertmanagerOpts{
 		ExternalURL:        cfg.AppURL,
 		AlertStoreCallback: nil,
 		PeerTimeout:        cfg.UnifiedAlerting.HAPeerTimeout,
 		Silences:           silencesOptions,
 		Nflog:              nflogOptions,
+		FlushLog:           flushLogOptions,
+		DispatchTimer:      dispatchTimer,
 		Limits: alertingNotify.Limits{
 			MaxSilences:         cfg.UnifiedAlerting.AlertmanagerMaxSilencesCount,
 			MaxSilenceSizeBytes: cfg.UnifiedAlerting.AlertmanagerMaxSilenceSizeBytes,

pkg/services/ngalert/notifier/dispatch_timer.go

@@ -0,0 +1,19 @@
+package notifier
+
+import (
+	alertingNotify "github.com/grafana/alerting/notify"
+	"github.com/grafana/grafana/pkg/infra/log"
+	"github.com/grafana/grafana/pkg/services/featuremgmt"
+)
+
+// GetDispatchTimer returns the appropriate dispatch timer based on feature toggles.
+func GetDispatchTimer(features featuremgmt.FeatureToggles) (dt alertingNotify.DispatchTimer) {
+	//nolint:staticcheck // not yet migrated to OpenFeature
+	enabled := features.IsEnabledGlobally(featuremgmt.FlagAlertingSyncDispatchTimer)
+	log.New("ngalert.dispatchTimer").Info("GetDispatchTimer called", "enabled", enabled, "result", dt.String())
+
+	if enabled {
+		dt = alertingNotify.DispatchTimerSync
+	}
+	return
+}

pkg/services/ngalert/notifier/file_store.go

@@ -15,6 +15,7 @@ const (
 	KVNamespace             = "alertmanager"
 	NotificationLogFilename = "notifications"
 	SilencesFilename        = "silences"
+	FlushLogFilename        = "flushes"
 )
 
 // FileStore is in charge of persisting the alertmanager files to the database.
@@ -42,6 +43,10 @@ func (fileStore *FileStore) GetNotificationLog(ctx context.Context) (string, err
 	return fileStore.contentFor(ctx, NotificationLogFilename)
 }
 
+func (fileStore *FileStore) GetFlushLog(ctx context.Context) (string, error) {
+	return fileStore.contentFor(ctx, FlushLogFilename)
+}
+
 // contentFor returns the content for the given Alertmanager kvstore key.
 func (fileStore *FileStore) contentFor(ctx context.Context, filename string) (string, error) {
 	// Then, let's attempt to read it from the database.
@@ -74,6 +79,11 @@ func (fileStore *FileStore) SaveNotificationLog(ctx context.Context, st alerting
 	return fileStore.persist(ctx, NotificationLogFilename, st)
 }
 
+// SaveFlushLog saves the flush log to the database and returns the size of the unencoded state.
+func (fileStore *FileStore) SaveFlushLog(ctx context.Context, st alertingNotify.State) (int64, error) {
+	return fileStore.persist(ctx, FlushLogFilename, st)
+}
+
 // persist takes care of persisting the binary representation of internal state to the database as a base64 encoded string.
 func (fileStore *FileStore) persist(ctx context.Context, filename string, st alertingNotify.State) (int64, error) {
 	var size int64

pkg/services/ngalert/notifier/state.go

@@ -5,5 +5,9 @@ func (am *alertmanager) MergeState(state ExternalState) error {
 	if err := am.Base.MergeNflog(state.Nflog); err != nil {
 		return err
 	}
-	return am.Base.MergeSilences(state.Silences)
+	if err := am.Base.MergeSilences(state.Silences); err != nil {
+		return err
+	}
+	return nil
+	// return am.Base.MergeFlushLog(state.FlushLog)
 }

pkg/services/ngalert/remote/alertmanager.go

@@ -84,6 +84,8 @@ type Alertmanager struct {
 
 	promoteConfig bool
 	externalURL   string
+
+	dispatchTimer string
 }
 
 type AlertmanagerConfig struct {
@@ -109,6 +111,9 @@ type AlertmanagerConfig struct {
 
 	// Timeout for the HTTP client.
 	Timeout time.Duration
+
+	// DispatchTimer specifies which timer to use in the dispatcher
+	DispatchTimer string
 }
 
 func (cfg *AlertmanagerConfig) Validate() error {
@@ -201,6 +206,7 @@ func NewAlertmanager(ctx context.Context, cfg AlertmanagerConfig, store stateSto
 		externalURL:   cfg.ExternalURL,
 		promoteConfig: cfg.PromoteConfig,
 		smtp:          cfg.SmtpConfig,
+		dispatchTimer: cfg.DispatchTimer,
 	}
 
 	// Parse the default configuration once and remember its hash so we can compare it later.
@@ -347,6 +353,7 @@ func (am *Alertmanager) buildConfiguration(ctx context.Context, raw []byte, crea
 		Promoted:      am.promoteConfig,
 		ExternalURL:   am.externalURL,
 		SmtpConfig:    am.smtp,
+		DispatchTimer: am.dispatchTimer,
 	}
 
 	cfgHash, err := calculateUserGrafanaConfigHash(payload)

pkg/services/ngalert/remote/client/alertmanager_configuration.go

@@ -37,6 +37,7 @@ type UserGrafanaConfig struct {
 	Promoted                  bool                      `json:"promoted"`
 	ExternalURL               string                    `json:"external_url"`
 	SmtpConfig                SmtpConfig                `json:"smtp_config"`
+	DispatchTimer             string                    `json:"dispatch_timer"`
 }
 
 func (mc *Mimir) GetGrafanaAlertmanagerConfig(ctx context.Context) (*UserGrafanaConfig, error) {