fix: complete hardware collection diagnostics

This commit is contained in:
2026-08-25 17:04:33 +03:00
parent 33ace33de5
commit ccc781856e
19 changed files with 311 additions and 81 deletions
@@ -43,6 +43,9 @@ func (s *System) RunNvidiaConfigCheckPack(ctx context.Context, baseDir string, l
if settings, err := s.ListNvidiaGPUSettings(); err != nil {
status.Notes = append(status.Notes, "nvidia-smi GPU settings unavailable (no driver, or no GPU present): "+err.Error())
} else {
if len(settings) == 0 {
status.NoGPUs = true
}
for _, g := range settings {
f := NvidiaGPUConfigFinding{
Index: g.Index, Name: g.Name,
@@ -127,6 +130,7 @@ type NvidiaConfigCheckStatus struct {
GPUs []NvidiaGPUConfigFinding `json:"gpus,omitempty"`
NVLinkPairs []NvidiaNVLinkPairFinding `json:"nvlink_pairs,omitempty"`
NoGPUs bool `json:"no_gpus"`
// Confidential Computing readiness — informational only, does not gate
// overall_status: an unconfigured/NOT_READY CC state is a
@@ -466,7 +470,10 @@ func renderNvidiaConfigCheckSummary(status NvidiaConfigCheckStatus) string {
fmt.Fprintf(&b, "cc_state=%s\n", status.CCState)
fmt.Fprintf(&b, "cpu_cc_capability=%s\n", status.CPUCCCapability)
fmt.Fprintf(&b, "gpu_cc_capability=%s\n", status.GPUCCCapability)
if len(status.Warnings) == 0 {
if status.NoGPUs {
fmt.Fprintln(&b, "overall_status=UNSUPPORTED")
fmt.Fprintln(&b, "reason=no_nvidia_gpus_detected")
} else if len(status.Warnings) == 0 {
fmt.Fprintln(&b, "overall_status=OK")
} else {
fmt.Fprintln(&b, "overall_status=FAILED")