Merge pull request #63 from vllm-project/local-dev/runtimeclass-2

[Add] customization support for runtimeClass
vllm-project · Feb 5, 2025 · 54a5d4f · 54a5d4f
2 parents ec71964 + 9e456bc
commit 54a5d4f
Show file tree

Hide file tree

Showing 2 changed files with 2 additions and 2 deletions.
diff --git a/helm/templates/deployment-vllm-multi.yaml b/helm/templates/deployment-vllm-multi.yaml
@@ -122,7 +122,7 @@ spec:
       {{- end }}
 
       {{- if .Values.servingEngineSpec.runtimeClassName }}
-      runtimeClassName: nvidia
+      runtimeClassName: {{ .Values.servingEngineSpec.runtimeClassName }}
       {{- end }}
       {{- if $modelSpec.nodeSelectorTerms}}
       affinity:

diff --git a/helm/values.yaml b/helm/values.yaml
@@ -130,7 +130,7 @@ servingEngineSpec:
   tolerations: []
 
   # -- RuntimeClassName configuration, set to "nvidia" if the model requires GPU
-  runtimeClassName: ""
+  runtimeClassName: "nvidia"
 
 routerSpec:
   # -- Number of replicas