diff --git a/examples/main.hmdl.json b/examples/main.hmdl.json
index 7b6abb1..bdbc795 100644
--- a/examples/main.hmdl.json
+++ b/examples/main.hmdl.json
@@ -2,7 +2,8 @@
 	"defaults": {
 		"schedule": {
 			"regular_interval": 10,
-			"attentive_interval": 1
+			"attentive_interval": 1,
+			"reminding_interval": 20
 		},
 		"threshold": 1,
 		"notifications": [
diff --git a/source/logic/conf.py b/source/logic/conf.py
index eba6e5f..19e407b 100644
--- a/source/logic/conf.py
+++ b/source/logic/conf.py
@@ -27,13 +27,14 @@ def conf_schema_annoy(
 	
 
 def conf_schema_interval(
+	allow_null,
 	default
 ):
 	return {
 		"anyOf": [
 			{
 				"description": "in seconds",
-				"type": "integer",
+				"type": ("integer" if allow_null else ["null", "integer"]),
 				"exclusiveMinimum": 0,
 			},
 			{
@@ -57,8 +58,9 @@ def conf_schema_schedule(
 		"type": "object",
 		"additionalProperties": False,
 		"properties": {
-			"regular_interval": conf_schema_interval(60 * 60),
-			"attentive_interval": conf_schema_interval(60 * 2),
+			"regular_interval": conf_schema_interval(False, (60 * 60)),
+			"attentive_interval": conf_schema_interval(False, (60 * 2)),
+			"reminding_interval": conf_schema_interval(True, (60 * 60 * 24)),
 		},
 		"required": [
 		],
@@ -193,22 +195,25 @@ def conf_schema_root(
 def conf_normalize_interval(
 	interval_raw
 ):
-	if (type(interval_raw) == int):
-		return interval_raw
-	elif (type(interval_raw) == str):
-		map_ = {
-			"minute": (60),
-			"hour": (60 * 60),
-			"day": (60 * 60 * 24),
-			"week": (60 * 60 * 24 * 7),
-		}
-		if (interval_raw not in map_):
-			raise ValueError("invalid string interval value: %s" % interval_raw)
-		else:
-			return map_[interval_raw]
+	if (interval_raw is None):
+		return None
 	else:
-		raise ValueError("invalid type for interval value")
-	
+		if (type(interval_raw) == int):
+			return interval_raw
+		elif (type(interval_raw) == str):
+			map_ = {
+				"minute": (60),
+				"hour": (60 * 60),
+				"day": (60 * 60 * 24),
+				"week": (60 * 60 * 24 * 7),
+			}
+			if (interval_raw not in map_):
+				raise ValueError("invalid string interval value: %s" % interval_raw)
+			else:
+				return map_[interval_raw]
+		else:
+			raise ValueError("invalid type for interval value")
+		
 
 def conf_normalize_schedule(
 	node
@@ -217,12 +222,14 @@ def conf_normalize_schedule(
 		{
 			"regular_interval": (60 * 60),
 			"attentive_interval": (60 * 2),
+			"reminding_interval": (60 * 60 * 24),
 		},
 		node
 	)
 	return {
 		"regular_interval": conf_normalize_interval(node["regular_interval"]),
 		"attentive_interval": conf_normalize_interval(node["attentive_interval"]),
+		"reminding_interval": conf_normalize_interval(node["reminding_interval"]),
 	}
 	
 
@@ -251,6 +258,7 @@ def conf_normalize_defaults(
 			"schedule": {
 				"regular_interval": (60 * 60),
 				"attentive_interval": (60 * 2),
+				"reminding_interval": (60 * 60 * 24),
 			},
 			"notifications": [
 			],
diff --git a/source/logic/main.py b/source/logic/main.py
index 3c3404c..4e08bc8 100644
--- a/source/logic/main.py
+++ b/source/logic/main.py
@@ -266,6 +266,24 @@ def main():
 										and
 										check_data["annoy"]
 									)
+									or
+									(
+										(count is None)
+										and
+										(
+											(old_item_state is not None)
+											and
+											(old_item_state["last_notification_timestamp"] is not None)
+											and
+											(check_data["schedule"]["reminding_interval"] is not None)
+											and
+											(
+												(timestamp - old_item_state["last_notification_timestamp"])
+												>=
+												check_data["schedule"]["reminding_interval"]
+											)
+										)
+									)
 								)
 								and
 								(
diff --git a/todo.md b/todo.md
index db8e275..9239ecb 100644
--- a/todo.md
+++ b/todo.md
@@ -1,8 +1,5 @@
-- parallele Zugriffe auf die Zustands-Datei verhindern
-	- mutex?
-	- auf DB umstellen?
+- auf DB umstellen?
 - Benachrichtigungen versenden, wenn ein Zustand sich wieder normalisiert hat (aber vorher über dem Schwellwert oft nicht OK war)
-- erneute Benachrichtigung über nicht-OK-Zustand nach einer Weile (siehe https://gitlab.greenscale.de/tools/heimdall/-/issues/3)
 - längere Statistiken über Metriken führen um auch Anstiege/Abfälle auszuwerten (z.B. "Speicherplatzverbrauch innerhalb einer Woche um 5GB gestiegen")
 - Selbst-Test
 - Benachrichtigungs-Kanäle: