treat node-exporter metrics separately

SumoLogic · Jul 19, 2023 · ff3a264 · ff3a264
1 parent db99de5
commit ff3a264
Showing 1 changed file with 11 additions and 4 deletions.
diff --git a/deploy/helm/sumologic/values.yaml b/deploy/helm/sumologic/values.yaml
@@ -1490,7 +1490,7 @@ kube-prometheus-stack:
 
   ## disable default recording rules
   defaultRules:
-    create: false
+    create: true
 
   ## k8s pre-1.14 prometheus recording rules
   additionalPrometheusRulesMap:
@@ -2149,14 +2149,21 @@ kube-prometheus-stack:
               replacement: ${1}
               action: replace
 
-      remoteWrite:
-        ## kube-state-metrics
+        ## infrastructure metrics
         - url: http://$(METADATA_METRICS_SVC).$(NAMESPACE).svc.cluster.local.:9888/prometheus.metrics
           remoteTimeout: 5s
           writeRelabelConfigs:
             - action: keep
-              regex: (?:kube-state-metrics|kubelet|kube-scheduler|apiserver|node-exporter|coredns|kube-etcd)
+              regex: (?:kube-state-metrics|kubelet|kube-scheduler|apiserver|coredns|kube-etcd)
               sourceLabels: [job]
+
+        ## This needs to be separate because we have a bunch of recording rule based metrics for Nodes which fall into this job
+        - url: http://$(METADATA_METRICS_SVC).$(NAMESPACE).svc.cluster.local.:9888/prometheus.metrics.node
+          remoteTimeout: 5s
+          writeRelabelConfigs:
+            - action: keep
+              regex: node-exporter;(?:node_load1|node_load5|node_load15|node_cpu_seconds_total)
+              sourceLabels: [job, __name__]
         ## prometheus operator rules
         ## :kube_pod_info_node_count:
         ## :node_cpu_saturation_load1: