docs(rm): document why nvmllib is stored on health checker struct

ArangoGutierrez · ArangoGutierrez · commit f4d7e99548b8 · 2026-03-11T15:49:01.000+01:00
Address reviewer question about why the NVML interface is duplicated
on the health checker rather than accessed via nvmlResourceManager.

Signed-off-by: Carlos Eduardo Arango Gutierrez &lt;eduardoa@nvidia.com&gt;
diff --git a/internal/rm/health.go b/internal/rm/health.go
@@ -42,6 +42,9 @@ const (
 )
 
 type nvmlDeviceHealthChecker struct {
+	// nvmllib is the NVML interface used to query device handles during event
+	// monitoring. Stored here rather than accessed via nvmlResourceManager to
+	// keep the health checker decoupled and independently testable.
 	nvmllib           nvml.Interface
 	devices           Devices
 	parentToDeviceMap map[string]*Device

Original file line number	Diff line number	Diff line change
`@@ -42,6 +42,9 @@ const (`
`42`	`42`	`)`
`43`	`43`
`44`	`44`	`type nvmlDeviceHealthChecker struct {`
	`45`	`+ // nvmllib is the NVML interface used to query device handles during event`
	`46`	`+ // monitoring. Stored here rather than accessed via nvmlResourceManager to`
	`47`	`+ // keep the health checker decoupled and independently testable.`
`45`	`48`	`nvmllib nvml.Interface`
`46`	`49`	`devices Devices`
`47`	`50`	`parentToDeviceMap map[string]*Device`