fix NodeGpuCountChanged definition issue

microsoft · Nov 10, 2020 · b2fa0b5 · b2fa0b5
1 parent efdd5f2
commit b2fa0b5
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/src/prometheus/deploy/alerting/gpu.rules b/src/prometheus/deploy/alerting/gpu.rules
@@ -71,7 +71,7 @@ groups:
           summary: "found nvidia used by zombie container in {{$labels.instance}}"
 
       - alert: NodeGpuCountChanged
-        expr: changes(node:gpu_utilization:count[5m]) > 0
+        expr: node:gpu_utilization:count != on (instance) configured_gpu_count
         labels:
           severity: fatal
         annotations: