Merge pull request #1106 from NVIDIA/gpu-mode-labeler-optional

Raise a warning instead of an error if GPU mode labeler fails
NVIDIA · Jan 10, 2025 · f7dc5f1 · f7dc5f1
2 parents 61887fe + bd00e7d
commit f7dc5f1
Showing 1 changed file with 2 additions and 1 deletion.
diff --git a/internal/lm/nvml.go b/internal/lm/nvml.go
@@ -215,7 +215,8 @@ func isMPSCapable(manager resource.Manager) (bool, error) {
 func newGPUModeLabeler(devices []resource.Device) (Labeler, error) {
 	classes, err := getDeviceClasses(devices)
 	if err != nil {
-		return nil, err
+		klog.Warningf("Failed to create GPU mode labeler: failed to get device classes: %v", err)
+		return Labels{"nvidia.com/gpu.mode": "unknown"}, nil
 	}
 	gpuMode := getModeForClasses(classes)
 	labels := Labels{