reanimator/bee
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 1 Wiki Activity

fix: pcie gen, nccl binary, netconf sudo, boot noise, firmware cleanup

Browse Source 
- nvidia collector: read pcie.link.gen.current/max from nvidia-smi instead
  of sysfs to avoid false Gen1 readings when GPU is in ASPM idle state
- build: remove bee-nccl-gpu-stress from rm -f list so shell script from
  overlay is not silently dropped from the ISO
- smoketest: add explicit checks for bee-gpu-burn, bee-john-gpu-stress,
  bee-nccl-gpu-stress, all_reduce_perf
- netconf: re-exec via sudo when not root to fix RTNETLINK/resolv.conf errors
- auto/config: reduce loglevel 7→3 to show clean systemd output on boot
- auto/config: blacklist snd_hda_intel and related audio modules (unused on servers)
- package-lists: remove firmware-intel-sound and firmware-amd-graphics from
  base list; move firmware-amd-graphics to bee-amd variant only
- bible-local: mark memtest ADR resolved, document working solution

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
Michael Chus
2026-04-01 21:25:23 +03:00
parent 2baf3be640
commit eb60100297
9 changed files with 148 additions and 25 deletions
Download Patch File Download Diff File Expand all files Collapse all files

80
audit/internal/collector/nvidia.go
View File

@@ -13,14 +13,18 @@ import (
const nvidiaVendorID = 0x10de
type nvidiaGPUInfo struct {
BDF string
Serial string
VBIOS string
TemperatureC *float64
PowerW *float64
ECCUncorrected *int64
ECCCorrected *int64
HWSlowdown *bool
BDF string
Serial string
VBIOS string
TemperatureC *float64
PowerW *float64
ECCUncorrected *int64
ECCCorrected *int64
HWSlowdown *bool
PCIeLinkGenCurrent *int
PCIeLinkGenMax *int
PCIeLinkWidthCur *int
PCIeLinkWidthMax *int
}
// enrichPCIeWithNVIDIA enriches NVIDIA PCIe devices with data from nvidia-smi.
@@ -94,7 +98,7 @@ func enrichPCIeWithNVIDIAData(devs []schema.HardwarePCIeDevice, gpuByBDF map[str
func queryNVIDIAGPUs() (map[string]nvidiaGPUInfo, error) {
out, err := exec.Command(
"nvidia-smi",
"--query-gpu=index,pci.bus_id,serial,vbios_version,temperature.gpu,power.draw,ecc.errors.uncorrected.aggregate.total,ecc.errors.corrected.aggregate.total,clocks_throttle_reasons.hw_slowdown",
"--query-gpu=index,pci.bus_id,serial,vbios_version,temperature.gpu,power.draw,ecc.errors.uncorrected.aggregate.total,ecc.errors.corrected.aggregate.total,clocks_throttle_reasons.hw_slowdown,pcie.link.gen.current,pcie.link.gen.max,pcie.link.width.current,pcie.link.width.max",
"--format=csv,noheader,nounits",
).Output()
if err != nil {
@@ -118,8 +122,8 @@ func parseNVIDIASMIQuery(raw string) (map[string]nvidiaGPUInfo, error) {
if len(rec) == 0 {
continue
}
if len(rec) < 9 {
return nil, fmt.Errorf("unexpected nvidia-smi columns: got %d, want 9", len(rec))
if len(rec) < 13 {
return nil, fmt.Errorf("unexpected nvidia-smi columns: got %d, want 13", len(rec))
}
bdf := normalizePCIeBDF(rec[1])
@@ -128,14 +132,18 @@ func parseNVIDIASMIQuery(raw string) (map[string]nvidiaGPUInfo, error) {
}
info := nvidiaGPUInfo{
BDF: bdf,
Serial: strings.TrimSpace(rec[2]),
VBIOS: strings.TrimSpace(rec[3]),
TemperatureC: parseMaybeFloat(rec[4]),
PowerW: parseMaybeFloat(rec[5]),
ECCUncorrected: parseMaybeInt64(rec[6]),
ECCCorrected: parseMaybeInt64(rec[7]),
HWSlowdown: parseMaybeBool(rec[8]),
BDF: bdf,
Serial: strings.TrimSpace(rec[2]),
VBIOS: strings.TrimSpace(rec[3]),
TemperatureC: parseMaybeFloat(rec[4]),
PowerW: parseMaybeFloat(rec[5]),
ECCUncorrected: parseMaybeInt64(rec[6]),
ECCCorrected: parseMaybeInt64(rec[7]),
HWSlowdown: parseMaybeBool(rec[8]),
PCIeLinkGenCurrent: parseMaybeInt(rec[9]),
PCIeLinkGenMax: parseMaybeInt(rec[10]),
PCIeLinkWidthCur: parseMaybeInt(rec[11]),
PCIeLinkWidthMax: parseMaybeInt(rec[12]),
}
result[bdf] = info
}
@@ -167,6 +175,22 @@ func parseMaybeInt64(v string) *int64 {
return &n
}
func parseMaybeInt(v string) *int {
v = strings.TrimSpace(v)
if v == "" || strings.EqualFold(v, "n/a") || strings.EqualFold(v, "not supported") || strings.EqualFold(v, "[not supported]") {
return nil
}
n, err := strconv.Atoi(v)
if err != nil {
return nil
}
return &n
}
func pcieLinkGenLabel(gen int) string {
return fmt.Sprintf("Gen%d", gen)
}
func parseMaybeBool(v string) *bool {
v = strings.TrimSpace(strings.ToLower(v))
switch v {
@@ -231,4 +255,22 @@ func injectNVIDIATelemetry(dev *schema.HardwarePCIeDevice, info nvidiaGPUInfo) {
if info.HWSlowdown != nil {
dev.HWSlowdown = info.HWSlowdown
}
// Override PCIe link speed/width with nvidia-smi driver values.
// sysfs current_link_speed reflects the instantaneous physical link state and
// can show Gen1 when the GPU is idle due to ASPM power management. The driver
// knows the negotiated speed regardless of the current power state.
if info.PCIeLinkGenCurrent != nil {
s := pcieLinkGenLabel(*info.PCIeLinkGenCurrent)
dev.LinkSpeed = &s
}
if info.PCIeLinkGenMax != nil {
s := pcieLinkGenLabel(*info.PCIeLinkGenMax)
dev.MaxLinkSpeed = &s
}
if info.PCIeLinkWidthCur != nil {
dev.LinkWidth = info.PCIeLinkWidthCur
}
if info.PCIeLinkWidthMax != nil {
dev.MaxLinkWidth = info.PCIeLinkWidthMax
}
}

8
audit/internal/collector/nvidia_test.go
View File

@@ -6,7 +6,7 @@ import (
)
func TestParseNVIDIASMIQuery(t *testing.T) {
raw := "0, 00000000:65:00.0, GPU-SERIAL-1, 96.00.1F.00.02, 54, 210.33, 0, 5, Not Active\n"
raw := "0, 00000000:65:00.0, GPU-SERIAL-1, 96.00.1F.00.02, 54, 210.33, 0, 5, Not Active, 4, 4, 16, 16\n"
byBDF, err := parseNVIDIASMIQuery(raw)
if err != nil {
t.Fatalf("parse failed: %v", err)
@@ -28,6 +28,12 @@ func TestParseNVIDIASMIQuery(t *testing.T) {
if gpu.HWSlowdown == nil || *gpu.HWSlowdown {
t.Fatalf("hw slowdown: got %v, want false", gpu.HWSlowdown)
}
if gpu.PCIeLinkGenCurrent == nil || *gpu.PCIeLinkGenCurrent != 4 {
t.Fatalf("pcie link gen current: got %v, want 4", gpu.PCIeLinkGenCurrent)
}
if gpu.PCIeLinkGenMax == nil || *gpu.PCIeLinkGenMax != 4 {
t.Fatalf("pcie link gen max: got %v, want 4", gpu.PCIeLinkGenMax)
}
}
func TestNormalizePCIeBDF(t *testing.T) {