Fix healthchecking on old devices

elezar · elezar · commit 8cd14472af22 · 2026-01-05T13:49:51.000+01:00
In cases where registering events for a device are not supported,
we should not mark the device as unhealthy, but skip the device instead.

Signed-off-by: Evan Lezar &lt;elezar@nvidia.com&gt;
diff --git a/internal/rm/health.go b/internal/rm/health.go
@@ -102,10 +102,10 @@ func (r *nvmlResourceManager) checkHealth(stop <-chan interface{}, devices Devic
 		}
 
 		ret = gpu.RegisterEvents(eventMask&supportedEvents, eventSet)
-		if ret == nvml.ERROR_NOT_SUPPORTED {
+		switch {
+		case ret == nvml.ERROR_NOT_SUPPORTED:
 			klog.Warningf("Device %v is too old to support healthchecking.", d.ID)
-		}
-		if ret != nvml.SUCCESS {
+		case ret != nvml.SUCCESS:
 			klog.Infof("Marking device %v as unhealthy: %v", d.ID, ret)
 			unhealthy <- d
 		}

Original file line number	Diff line number	Diff line change
`@@ -102,10 +102,10 @@ func (r *nvmlResourceManager) checkHealth(stop <-chan interface{}, devices Devic`
`102`	`102`	`}`
`103`	`103`
`104`	`104`	`ret = gpu.RegisterEvents(eventMask&supportedEvents, eventSet)`
`105`		`- if ret == nvml.ERROR_NOT_SUPPORTED {`
	`105`	`+ switch {`
	`106`	`+ case ret == nvml.ERROR_NOT_SUPPORTED:`
`106`	`107`	`klog.Warningf("Device %v is too old to support healthchecking.", d.ID)`
`107`		`- }`
`108`		`- if ret != nvml.SUCCESS {`
	`108`	`+ case ret != nvml.SUCCESS:`
`109`	`109`	`klog.Infof("Marking device %v as unhealthy: %v", d.ID, ret)`
`110`	`110`	`unhealthy <- d`
`111`	`111`	`}`