dispatch: don't reset timer if flush is in-progress (#1301)

When the aggregation group receives an alert that is past the initial
group_wait value, it should reset its timer only if the timer has ever
expired. Otherwise it means that the flush is already in-progress.

dispatch/dispatch.go

@@ -289,7 +289,7 @@ type aggrGroup struct {
 
 	mtx        sync.RWMutex
 	alerts     map[model.Fingerprint]*types.Alert
-	hasSent bool
+	hasFlushed bool
 }
 
 // newAggrGroup returns a new aggregation group.
@@ -366,6 +366,7 @@ func (ag *aggrGroup) run(nf notifyFunc) {
 			// Wait the configured interval before calling flush again.
 			ag.mtx.Lock()
 			ag.next.Reset(ag.opts.GroupInterval)
+			ag.hasFlushed = true
 			ag.mtx.Unlock()
 
 			ag.flush(func(alerts ...*types.Alert) bool {
@@ -396,7 +397,7 @@ func (ag *aggrGroup) insert(alert *types.Alert) {
 
 	// Immediately trigger a flush if the wait duration for this
 	// alert is already over.
-	if !ag.hasSent && alert.StartsAt.Add(ag.opts.GroupWait).Before(time.Now()) {
+	if !ag.hasFlushed && alert.StartsAt.Add(ag.opts.GroupWait).Before(time.Now()) {
 		ag.next.Reset(0)
 	}
 }
@@ -457,8 +458,6 @@ func (ag *aggrGroup) flush(notify func(...*types.Alert) bool) {
 				delete(ag.alerts, fp)
 			}
 		}
-
-		ag.hasSent = true
 		ag.mtx.Unlock()
 	}
 }

test/acceptance/send_test.go

@@ -406,10 +406,9 @@ receivers:
 func TestReload(t *testing.T) {
 	t.Parallel()
 
-	// We create a notification config that fans out into two different
-	// webhooks.
-	// The succeeding one must still only receive the first successful
-	// notifications. Sending to the succeeding one must eventually succeed.
+	// This integration test ensures that the first alert isn't notified twice
+	// and repeat_interval applies after the AlertManager process has been
+	// reloaded.
 	conf := `
 route:
   receiver: "default"