[horus] Developing standards for downtime (#405)

apache · Sep 30, 2024 · da49844 · da49844
1 parent cb2691d
commit da49844
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/manifests/horus/horus.yaml b/manifests/horus/horus.yaml
@@ -78,7 +78,7 @@ nodeDownTime:
   promQueryTimeSecond: 60
   abnormalityQL:
     - 100 - (avg by (node) (rate(node_cpu_seconds_total{mode="idle"}[5m])) * 100) > 80
-    - (avg by (node) (node_memory_MemAvailable_bytes / node_memory_MemTotal_bytes) * 100) < 20
+    - (avg by (node) (node_memory_MemFree_bytes / node_memory_MemTotal_bytes )) * 100 < 10
     - node_filesystem_avail_bytes{mountpoint="/"} / node_filesystem_size_bytes{mountpoint="/"} * 100 < 15
   nodeNameToIPs:
     node_os_info{node="%s"}