From 2bfe4fae256532181b1035961f8f281d6fbba539 Mon Sep 17 00:00:00 2001
From: Alexandre Iooss <erdnaxe@crans.org>
Date: Sun, 26 Jan 2020 21:31:48 +0100
Subject: [PATCH] Mesure load rather than CPU time usage

---
 roles/prometheus/templates/prometheus/alert.rules.yml.j2 | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/roles/prometheus/templates/prometheus/alert.rules.yml.j2 b/roles/prometheus/templates/prometheus/alert.rules.yml.j2
index ce7f711b..2b358276 100644
--- a/roles/prometheus/templates/prometheus/alert.rules.yml.j2
+++ b/roles/prometheus/templates/prometheus/alert.rules.yml.j2
@@ -43,13 +43,13 @@ groups:
       summary: "Presque plus d'inodes disponibles ({{ $value }}% restant) dans {{ $labels.mountpoint }} sur {{ $labels.instance }}."
 
   # Alert for high CPU usage
-  - alert: CpuUsage
-    expr: (100 - avg by (instance) (irate(node_cpu_seconds_total{mode="idle"}[5m])) * 100) > 75
+  - alert: CpuBusy
+    expr: node_load5 > 3
     for: 10m
     labels:
       severity: warning
     annotations:
-      summary: "CPU sur {{ $labels.instance }} à {{ $value }}%."
+      summary: "Charge sur {{ $labels.instance }} à {{ $value }}."
 
   # Check mdadm software RAID
   - alert: SoftwareRAIDDegraded