fix: ensure relay recovery loop can reclaim stuck messages before expiry

CHANGELOG.md

@@ -7,6 +7,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+### Fixed
+- Relay recovery loop could not reclaim stuck messages after a pod restart because `DefaultWebhookExpiry` (30s) was shorter than `pendingIdleTimeout` (60s). Fixed by raising `DefaultWebhookExpiry` to 2 minutes and lowering `pendingIdleTimeout` to 30s, ensuring the recovery loop always has time to reclaim a stuck message before it expires.
+
 ## [0.2.13] - 2026-06-11
 
 ### Changed

internal/relay/manager.go

@@ -17,8 +17,10 @@ var (
 )
 
 const (
-	// DefaultWebhookExpiry is the default time after which a queued webhook expires
-	DefaultWebhookExpiry = 30 * time.Second
+	// DefaultWebhookExpiry is the default time after which a queued webhook expires.
+	// Must be greater than pendingIdleTimeout + defaultRecoveryInterval so that the
+	// recovery loop can reclaim stuck messages before they expire.
+	DefaultWebhookExpiry = 2 * time.Minute
 )
 
 // Webhook represents a webhook request to be delivered via relay

internal/relay/redis_manager.go

@@ -35,7 +35,7 @@ const (
 
 	// Recovery settings
 	defaultRecoveryInterval = 30 * time.Second // How often to check for stuck messages
-	pendingIdleTimeout      = 60 * time.Second // How long a message can be pending before reclaim
+	pendingIdleTimeout      = 30 * time.Second // How long a message can be pending before reclaim
 	maxDeliveryAttempts     = 3                // Max retries before dead letter
 )
 

internal/relay/redis_manager_test.go

@@ -908,9 +908,9 @@ func TestRedisManager_Deliver_SetsExpiry(t *testing.T) {
 		if webhook.ExpiresAt == 0 {
 			t.Error("expected ExpiresAt to be set")
 		}
-		// Should be about 30 seconds in the future
-		expectedMin := time.Now().Add(25 * time.Second).Unix()
-		expectedMax := time.Now().Add(35 * time.Second).Unix()
+		// Should be about 2 minutes in the future
+		expectedMin := time.Now().Add(115 * time.Second).Unix()
+		expectedMax := time.Now().Add(125 * time.Second).Unix()
 		if webhook.ExpiresAt < expectedMin || webhook.ExpiresAt > expectedMax {
 			t.Errorf("ExpiresAt %d not in expected range [%d, %d]", webhook.ExpiresAt, expectedMin, expectedMax)
 		}