diff --git a/internal/alerts/active_cleanup.go b/internal/alerts/active_cleanup.go
index 16772a8ca..a35c6e5b9 100644
--- a/internal/alerts/active_cleanup.go
+++ b/internal/alerts/active_cleanup.go
@@ -310,16 +310,7 @@ func (m *Manager) CleanupAlertsForNodes(existingNodes map[string]bool) {
 
 	if removedCount > 0 {
 		log.Debug().Int("removed", removedCount).Int("remaining", len(m.activeAlerts)).Msg("cleaned up alerts for non-existent nodes")
-		go func() {
-			defer func() {
-				if r := recover(); r != nil {
-					log.Error().Interface("panic", r).Msg("panic in SaveActiveAlerts goroutine (cleanup)")
-				}
-			}()
-			if err := m.SaveActiveAlerts(); err != nil {
-				log.Error().Err(err).Msg("failed to save alerts after cleanup")
-			}
-		}()
+		m.saveActiveAlertsAsync("node cleanup")
 	} else {
 		log.Debug().Msg("no alerts needed cleanup")
 	}
@@ -358,14 +349,5 @@ func (m *Manager) ClearActiveAlerts() {
 
 	log.Info().Msg("cleared all active and pending alerts")
 
-	go func() {
-		defer func() {
-			if r := recover(); r != nil {
-				log.Error().Interface("panic", r).Msg("panic in SaveActiveAlerts goroutine (clear)")
-			}
-		}()
-		if err := m.SaveActiveAlerts(); err != nil {
-			log.Error().Err(err).Msg("failed to persist cleared alerts")
-		}
-	}()
+	m.saveActiveAlertsAsync("clear active alerts")
 }
diff --git a/internal/alerts/active_persistence.go b/internal/alerts/active_persistence.go
index d04690489..bc28e24f5 100644
--- a/internal/alerts/active_persistence.go
+++ b/internal/alerts/active_persistence.go
@@ -14,15 +14,30 @@ import (
 	"github.com/rs/zerolog/log"
 )
 
-func asyncSaveActiveAlerts(reason string, save func() error) {
+func (m *Manager) saveActiveAlertsAsync(context string) {
+	m.stopMu.RLock()
+	if m.stopping {
+		m.stopMu.RUnlock()
+		return
+	}
+	m.workerWG.Add(1)
+	m.stopMu.RUnlock()
+
 	go func() {
+		defer m.workerWG.Done()
 		defer func() {
 			if r := recover(); r != nil {
-				log.Error().Interface("panic", r).Str("reason", reason).Msg("panic in SaveActiveAlerts goroutine")
+				log.Error().
+					Interface("panic", r).
+					Str("context", context).
+					Msg("Panic in SaveActiveAlerts goroutine")
 			}
 		}()
-		if err := save(); err != nil {
-			log.Error().Err(err).Str("reason", reason).Msg("failed to save active alerts")
+		if err := m.SaveActiveAlerts(); err != nil {
+			log.Error().
+				Err(err).
+				Str("context", context).
+				Msg("Failed to save active alerts")
 		}
 	}()
 }
@@ -107,25 +122,6 @@ func (m *Manager) SaveActiveAlerts() error {
 	return nil
 }
 
-func (m *Manager) saveActiveAlertsAsync(context string) {
-	go func() {
-		defer func() {
-			if r := recover(); r != nil {
-				log.Error().
-					Interface("panic", r).
-					Str("context", context).
-					Msg("Panic in SaveActiveAlerts goroutine")
-			}
-		}()
-		if err := m.SaveActiveAlerts(); err != nil {
-			log.Error().
-				Err(err).
-				Str("context", context).
-				Msg("Failed to save active alerts")
-		}
-	}()
-}
-
 // LoadActiveAlerts restores active alerts from disk.
 func (m *Manager) LoadActiveAlerts() error {
 	m.mu.Lock()
diff --git a/internal/alerts/alerts_test.go b/internal/alerts/alerts_test.go
index 1d8fc5baa..08edaab96 100644
--- a/internal/alerts/alerts_test.go
+++ b/internal/alerts/alerts_test.go
@@ -101,28 +101,8 @@ func newTestManager(t *testing.T) *Manager {
 	m := NewManager()
 
 	// Restore env var and release mutex when test completes.
-	// We also stop the history manager's background goroutines (but not the
-	// full manager Stop which includes a 100ms sleep) to prevent writes to
-	// the temp directory after the test completes.
 	t.Cleanup(func() {
-		// Stop the history manager to halt background save routines
-		m.historyManager.Stop()
-		// Close escalation channel to stop that goroutine too
-		select {
-		case <-m.escalationStop:
-			// Already closed
-		default:
-			close(m.escalationStop)
-		}
-		// Close cleanup channel
-		select {
-		case <-m.cleanupStop:
-			// Already closed
-		default:
-			close(m.cleanupStop)
-		}
-		// Brief pause to let goroutines finish any in-flight operations
-		time.Sleep(10 * time.Millisecond)
+		m.Stop()
 
 		if hadOld {
 			os.Setenv("PULSE_DATA_DIR", oldVal)
diff --git a/internal/alerts/backup_snapshot.go b/internal/alerts/backup_snapshot.go
index 0d4aa4c68..23aaa47eb 100644
--- a/internal/alerts/backup_snapshot.go
+++ b/internal/alerts/backup_snapshot.go
@@ -525,7 +525,7 @@ func (m *Manager) CheckSnapshotsForInstance(instanceName string, snapshots []mod
 			},
 		})
 		if result.Transition != nil && result.Transition.Kind == alertspecs.EvaluationTransitionActivated {
-			asyncSaveActiveAlerts("snapshot", m.SaveActiveAlerts)
+			m.saveActiveAlertsAsync("snapshot")
 		}
 	}
 
@@ -913,7 +913,7 @@ func (m *Manager) CheckBackupsWithInventory(
 			},
 		})
 		if result.Transition != nil && result.Transition.Kind == alertspecs.EvaluationTransitionActivated {
-			asyncSaveActiveAlerts("backup", m.SaveActiveAlerts)
+			m.saveActiveAlertsAsync("backup")
 		}
 	}
 
diff --git a/internal/alerts/canonical_metric.go b/internal/alerts/canonical_metric.go
index a3cbc6eed..62bb5b5e9 100644
--- a/internal/alerts/canonical_metric.go
+++ b/internal/alerts/canonical_metric.go
@@ -113,7 +113,7 @@ func (m *Manager) evaluateCanonicalMetricAlert(spec alertspecs.ResourceAlertSpec
 	migratedAlertIdentity := false
 	defer func() {
 		if migratedAlertIdentity {
-			asyncSaveActiveAlerts("canonical guest metric node move", m.SaveActiveAlerts)
+			m.saveActiveAlertsAsync("canonical guest metric node move")
 		}
 	}()
 	defer m.mu.Unlock()
@@ -237,7 +237,7 @@ func (m *Manager) evaluateCanonicalMetricAlert(spec alertspecs.ResourceAlertSpec
 			m.recentAlerts[trackingKey] = alert
 			m.historyManager.AddAlert(*alert)
 
-			asyncSaveActiveAlerts("canonical metric create", m.SaveActiveAlerts)
+			m.saveActiveAlertsAsync("canonical metric create")
 
 			if alertForAICallback := m.getAlertForAICallback(); alertForAICallback != nil {
 				alertCopy := cloneAlertForOutput(alert)
@@ -315,7 +315,7 @@ func (m *Manager) evaluateCanonicalMetricAlert(spec alertspecs.ResourceAlertSpec
 			ResolvedTime: observedAt,
 		}
 		m.removeActiveAlertNoLock(storageKey)
-		asyncSaveActiveAlerts("canonical metric resolution", m.SaveActiveAlerts)
+		m.saveActiveAlertsAsync("canonical metric resolution")
 		m.addRecentlyResolvedWithPrimaryLock(resolvedAlert)
 		m.safeCallResolvedAlertCallback(existingAlert, storageKey, true)
 	}
diff --git a/internal/alerts/config_runtime.go b/internal/alerts/config_runtime.go
index 53ce1ad19..8be490b27 100644
--- a/internal/alerts/config_runtime.go
+++ b/internal/alerts/config_runtime.go
@@ -538,16 +538,7 @@ func (m *Manager) reevaluateActiveAlertsLocked() {
 	}
 
 	if len(alertsToResolve) > 0 {
-		go func() {
-			defer func() {
-				if r := recover(); r != nil {
-					log.Error().Interface("panic", r).Msg("panic in SaveActiveAlerts goroutine (config update)")
-				}
-			}()
-			if err := m.SaveActiveAlerts(); err != nil {
-				log.Error().Err(err).Msg("failed to save active alerts after config update")
-			}
-		}()
+		m.saveActiveAlertsAsync("config update")
 	}
 }
 
diff --git a/internal/alerts/history.go b/internal/alerts/history.go
index fd423efe3..4e5bc3480 100644
--- a/internal/alerts/history.go
+++ b/internal/alerts/history.go
@@ -54,6 +54,7 @@ type HistoryManager struct {
 	saveInterval time.Duration
 	stopChan     chan struct{}
 	saveTicker   *time.Ticker
+	workerWG     sync.WaitGroup
 	callbacks    []AlertCallback // Called when alerts are added
 }
 
@@ -118,7 +119,11 @@ func NewHistoryManager(dataDir string) *HistoryManager {
 	hm.startPeriodicSave()
 
 	// Start cleanup routine
-	go hm.cleanupRoutine()
+	hm.workerWG.Add(1)
+	go func() {
+		defer hm.workerWG.Done()
+		hm.cleanupRoutine()
+	}()
 
 	return hm
 }
@@ -462,7 +467,9 @@ func syncDir(dirPath string) error {
 func (hm *HistoryManager) startPeriodicSave() {
 	hm.saveTicker = time.NewTicker(hm.saveInterval)
 
+	hm.workerWG.Add(1)
 	go func() {
+		defer hm.workerWG.Done()
 		for {
 			select {
 			case <-hm.saveTicker.C:
@@ -643,6 +650,8 @@ func (hm *HistoryManager) Stop() {
 			hm.saveTicker.Stop()
 		}
 
+		hm.workerWG.Wait()
+
 		// Save one final time
 		if err := hm.saveHistory(); err != nil {
 			log.Error().Err(err).Msg("Failed to save alert history on shutdown")
diff --git a/internal/alerts/lifecycle.go b/internal/alerts/lifecycle.go
index 72ee3560d..a35386ece 100644
--- a/internal/alerts/lifecycle.go
+++ b/internal/alerts/lifecycle.go
@@ -1,23 +1,24 @@
 package alerts
 
 import (
-	"time"
-
 	"github.com/rs/zerolog/log"
 )
 
 // Stop stops the alert manager and saves history
 func (m *Manager) Stop() {
 	m.stopOnce.Do(func() {
+		m.stopMu.Lock()
+		m.stopping = true
 		closeSignalChannel(m.escalationStop)
 		closeSignalChannel(m.cleanupStop)
+		m.stopMu.Unlock()
+
+		m.workerWG.Wait()
+
 		if m.historyManager != nil {
 			m.historyManager.Stop()
 		}
 
-		// Give background goroutines time to exit cleanly
-		time.Sleep(100 * time.Millisecond)
-
 		// Save active alerts before stopping
 		if err := m.SaveActiveAlerts(); err != nil {
 			log.Error().Err(err).Msg("Failed to save active alerts on stop")
diff --git a/internal/alerts/manager.go b/internal/alerts/manager.go
index c52137ff7..34f8008e5 100644
--- a/internal/alerts/manager.go
+++ b/internal/alerts/manager.go
@@ -86,6 +86,9 @@ type Manager struct {
 	quietHoursLoc *time.Location
 	now           func() time.Time
 	stopOnce      sync.Once
+	stopMu        sync.RWMutex
+	stopping      bool
+	workerWG      sync.WaitGroup
 }
 
 type ackRecord struct {
@@ -151,14 +154,20 @@ func NewManagerWithDataDir(dataDir string) *Manager {
 		log.Error().Err(err).Msg("failed to load active alerts")
 	}
 
-	// Start escalation checker
-	go m.escalationChecker()
-
-	// Start periodic save of active alerts
-	go m.periodicSaveAlerts()
-
-	// Start periodic cleanup of stale tracking map entries
-	go m.trackingMapCleanup()
+	// Start background workers.
+	m.workerWG.Add(3)
+	go func() {
+		defer m.workerWG.Done()
+		m.escalationChecker()
+	}()
+	go func() {
+		defer m.workerWG.Done()
+		m.periodicSaveAlerts()
+	}()
+	go func() {
+		defer m.workerWG.Done()
+		m.trackingMapCleanup()
+	}()
 
 	return m
 }
diff --git a/internal/alerts/metric_runtime.go b/internal/alerts/metric_runtime.go
index 566a974ef..876786d98 100644
--- a/internal/alerts/metric_runtime.go
+++ b/internal/alerts/metric_runtime.go
@@ -175,7 +175,7 @@ func (m *Manager) checkMetric(resourceID, resourceName, node, instance, resource
 	migratedAlertIdentity := false
 	defer func() {
 		if migratedAlertIdentity {
-			asyncSaveActiveAlerts("guest metric node move", m.SaveActiveAlerts)
+			m.saveActiveAlertsAsync("guest metric node move")
 		}
 	}()
 	defer m.mu.Unlock()
@@ -342,17 +342,7 @@ func (m *Manager) checkMetric(resourceID, resourceName, node, instance, resource
 			m.recentAlerts[trackingKey] = alert
 			m.historyManager.AddAlert(*alert)
 
-			// Save active alerts after adding new one
-			go func() {
-				defer func() {
-					if r := recover(); r != nil {
-						log.Error().Interface("panic", r).Msg("panic in SaveActiveAlerts goroutine")
-					}
-				}()
-				if err := m.SaveActiveAlerts(); err != nil {
-					log.Error().Err(err).Msg("failed to save active alerts after creation")
-				}
-			}()
+			m.saveActiveAlertsAsync("metric create")
 
 			log.Warn().
 				Str("alertID", alertID).
@@ -505,17 +495,7 @@ func (m *Manager) checkMetric(resourceID, resourceName, node, instance, resource
 				// Remove from active alerts
 				m.removeActiveAlertNoLock(alertID)
 
-				// Save active alerts after resolution
-				go func() {
-					defer func() {
-						if r := recover(); r != nil {
-							log.Error().Interface("panic", r).Msg("panic in SaveActiveAlerts goroutine (resolution)")
-						}
-					}()
-					if err := m.SaveActiveAlerts(); err != nil {
-						log.Error().Err(err).Msg("failed to save active alerts after resolution")
-					}
-				}()
+				m.saveActiveAlertsAsync("metric resolution")
 
 				// Add to recently resolved while preventing lock-order inversions
 				m.addRecentlyResolvedWithPrimaryLock(resolvedAlert)
diff --git a/internal/alerts/tracking_cleanup.go b/internal/alerts/tracking_cleanup.go
index b07b91458..e7951ec86 100644
--- a/internal/alerts/tracking_cleanup.go
+++ b/internal/alerts/tracking_cleanup.go
@@ -223,16 +223,7 @@ func (m *Manager) cleanupStaleMaps() {
 	}
 
 	if staleResolved > 0 {
-		go func() {
-			defer func() {
-				if r := recover(); r != nil {
-					log.Error().Interface("panic", r).Msg("panic in SaveActiveAlerts goroutine (stale cleanup)")
-				}
-			}()
-			if err := m.SaveActiveAlerts(); err != nil {
-				log.Error().Err(err).Msg("failed to save active alerts after stale cleanup")
-			}
-		}()
+		m.saveActiveAlertsAsync("stale cleanup")
 		log.Info().
 			Int("count", staleResolved).
 			Msg("Auto-resolved stale alerts")
diff --git a/internal/alerts/unified_eval_parity_test.go b/internal/alerts/unified_eval_parity_test.go
index bceabde67..0c620a41d 100644
--- a/internal/alerts/unified_eval_parity_test.go
+++ b/internal/alerts/unified_eval_parity_test.go
@@ -2,7 +2,6 @@ package alerts
 
 import (
 	"testing"
-	"time"
 
 	"github.com/rcourtman/pulse-go-rewrite/internal/models"
 )
@@ -11,14 +10,8 @@ func newUnifiedEvalParityManager(t *testing.T) *Manager {
 	t.Helper()
 	m := NewManagerWithDataDir(t.TempDir())
 
-	// Stop background workers to avoid post-test writes into temp dirs.
 	t.Cleanup(func() {
-		if m.historyManager != nil {
-			m.historyManager.Stop()
-		}
-		closeSignalChannel(m.escalationStop)
-		closeSignalChannel(m.cleanupStop)
-		time.Sleep(10 * time.Millisecond)
+		m.Stop()
 	})
 
 	return m