diff --git a/awx/main/models/ha.py b/awx/main/models/ha.py
index 3a6b7740a2..5f9588f627 100644
--- a/awx/main/models/ha.py
+++ b/awx/main/models/ha.py
@@ -207,7 +207,7 @@ class Instance(HasPolicyEditsMixin, BaseModel):
             return True
         if ref_time is None:
             ref_time = now()
-        grace_period = settings.CLUSTER_NODE_HEARTBEAT_PERIOD * 2
+        grace_period = settings.CLUSTER_NODE_HEARTBEAT_PERIOD * settings.CLUSTER_NODE_MISSED_HEARTBEAT_TOLERANCE
         if self.node_type in ('execution', 'hop'):
             grace_period += settings.RECEPTOR_SERVICE_ADVERTISEMENT_PERIOD
         return self.last_seen < ref_time - timedelta(seconds=grace_period)
diff --git a/awx/settings/defaults.py b/awx/settings/defaults.py
index 499225a17d..a24fe6f090 100644
--- a/awx/settings/defaults.py
+++ b/awx/settings/defaults.py
@@ -432,6 +432,10 @@ os.environ.setdefault('DJANGO_LIVE_TEST_SERVER_ADDRESS', 'localhost:9013-9199')
 
 # heartbeat period can factor into some forms of logic, so it is maintained as a setting here
 CLUSTER_NODE_HEARTBEAT_PERIOD = 60
+
+# Number of missed heartbeats until a node gets marked as lost
+CLUSTER_NODE_MISSED_HEARTBEAT_TOLERANCE = 2
+
 RECEPTOR_SERVICE_ADVERTISEMENT_PERIOD = 60  # https://github.com/ansible/receptor/blob/aa1d589e154d8a0cb99a220aff8f98faf2273be6/pkg/netceptor/netceptor.go#L34
 EXECUTION_NODE_REMEDIATION_CHECKS = 60 * 30  # once every 30 minutes check if an execution node errors have been resolved