prometheus-node-exporter: declare cpu/mem requests and limits

helm-charts/support/values.yaml

@@ -100,6 +100,26 @@ prometheus:
       - key: k8s.dask.org/dedicated
         value: worker
         effect: NoSchedule
+    # resources for the node-exporter was set after inspecting cpu and memory
+    # use in a few clusters using prometheus queries from
+    # https://gist.github.com/payam-int/edf977c6af603fee0ce1b05da7792fe7.
+    #
+    # - Memory query:
+    #   100 * (1 - ((avg_over_time(node_memory_MemFree_bytes[10m]) + avg_over_time(node_memory_Cached_bytes[10m]) + avg_over_time(node_memory_Buffers_bytes[10m])) / avg_over_time(node_memory_MemTotal_bytes[10m])))
+    # - CPU query:
+    #   (1 - avg(irate(node_cpu_seconds_total{mode="idle"}[10m])) by (instance)) * 100
+    #
+    # node-exporter is typically found using around 1-10m in CPU, peaking up to
+    # 1000m, and memory is typically below 25Mi but has been seen peaking to
+    # ~90Mi.
+    #
+    resources:
+      limits:
+        cpu: 1
+        memory: 128Mi
+      requests:
+        cpu: 10m
+        memory: 128Mi
 
   networkPolicy:
     enabled: true