fix: Add tolerations in helm chart.

castai · Dec 5, 2024 · 00342aa · 00342aa
1 parent 872ebe4
commit 00342aa
Show file tree

Hide file tree

Showing 2 changed files with 8 additions and 0 deletions.
diff --git a/charts/gpu-metrics-exporter/templates/daemonset.yaml b/charts/gpu-metrics-exporter/templates/daemonset.yaml
@@ -53,6 +53,10 @@ spec:
         {{- toYaml . | nindent 8 }}
       {{- end }}
       {{- end}}
+      {{- with .Values.gpuMetricsExporter.tolerations }}
+      tolerations:
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
       containers:
         - name: castai-gpu-metrics-exporter
           securityContext:

diff --git a/charts/gpu-metrics-exporter/values.yaml b/charts/gpu-metrics-exporter/values.yaml
@@ -27,6 +27,10 @@ gpuMetricsExporter:
     readOnlyRootFilesystem: true
     runAsNonRoot: true
   port: 6061
+  tolerations:
+    - key: "nvidia.com/gpu"
+      value: "true"
+      effect: "NoSchedule"
 
 dcgmExporter:
   enabled: true