Files
bee/audit/internal/webui/page_topo_test.go
T
Mikhail ChusavitinandClaude Sonnet 5 a3377083aa webui: add /topo server topology page
New read-only visualization page: CPU sockets as anchor nodes, PCIe
devices (GPU/NIC/RAID) linked to their NUMA-affine socket with edge
color derived strictly from link_speed vs max_link_speed (not from the
device's own Status, which can also be overwritten by SAT-test results
on the same field), PSU/BMC as standalone boxes with no connecting
line, and a separate NVLink Topology card (live nvidia-smi topo -m /
nvlink -s/-e queries, not persisted to any contract).

GPU-GPU edges are drawn strictly from the actual bonded-pair list
parsed out of "nvidia-smi topo -m" (parseGPUPairAdjacency), not from
adjacent box position in the layout — an earlier ASCII mockup drew a
"chain" through unrelated GPUs, which a dedicated regression test now
guards against. A bonded pair spanning two different NUMA nodes is
flagged Warning on the edge and on both GPU boxes, per project
decision that this is an anomaly worth surfacing, not a neutral fact.

Zero changes to the ingest contract: this reverts the HardwareNVLinkPort/
HardwarePCIeDevice.NVLinks field shipped in v11.55 (33d6eee) along with
its collector/nvidia.go enrichment — that field risked a 400 from
Reanimator Core's strict decoder without an RFC, and isn't needed since
the topo page queries nvidia-smi directly instead of reading it from
audit.json. The v11.55 systemd fix (bee-nvidia.service ordering,
nv-hostengine restart) and the nvlink-status/-errors/dcgmi dumps in the
support bundle are untouched.

Also reorganizes support bundle collection per project convention:
system/ is now LiveCD-operational logs only (Xorg, services, console,
network/FS of the host itself); all server-hardware dumps (lspci,
NVIDIA/NVLink/DCGM, fabric manager, PCIe AER, ethtool, mstflint) move
to techdump/, deduplicating two entries already produced by
platform/techdump.go. Adds nvidia-bug-report.sh (previously only
collected inside the NVIDIA SAT pack) and lscpu to the always-on dump.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
2026-07-07 11:50:33 +03:00

246 lines
7.2 KiB
Go

package webui
import (
"encoding/json"
"net/http"
"net/http/httptest"
"os"
"path/filepath"
"strings"
"testing"
"bee/audit/internal/schema"
)
func TestTopoPageNoAuditDataGracefulFallback(t *testing.T) {
handler := NewHandler(HandlerOptions{})
rec := httptest.NewRecorder()
handler.ServeHTTP(rec, httptest.NewRequest(http.MethodGet, "/topo", nil))
if rec.Code != http.StatusOK {
t.Fatalf("status=%d", rec.Code)
}
body := rec.Body.String()
if !strings.Contains(body, "No audit data") {
t.Fatalf("topo page missing no-audit-data fallback: %s", body)
}
}
func TestTopoPageRendersCPUAndDegradedPCIeLink(t *testing.T) {
dir := t.TempDir()
path := filepath.Join(dir, "audit.json")
socket := 0
numaNode := 0
gen3, gen4 := "Gen3", "Gen4"
deviceClass := "VideoController"
model := "NVIDIA H100 80GB HBM3"
cpuModel := "Intel Xeon 6530"
okStatus := "OK"
ingest := schema.HardwareIngestRequest{
CollectedAt: "2026-03-15T00:00:00Z",
Hardware: schema.HardwareSnapshot{
CPUs: []schema.HardwareCPU{
{
HardwareComponentStatus: schema.HardwareComponentStatus{Status: &okStatus},
Socket: &socket,
Model: &cpuModel,
},
},
PCIeDevices: []schema.HardwarePCIeDevice{
{
DeviceClass: &deviceClass,
Model: &model,
NUMANode: &numaNode,
LinkSpeed: &gen3,
MaxLinkSpeed: &gen4,
},
},
},
}
data, err := json.Marshal(ingest)
if err != nil {
t.Fatal(err)
}
if err := os.WriteFile(path, data, 0644); err != nil {
t.Fatal(err)
}
handler := NewHandler(HandlerOptions{AuditPath: path})
rec := httptest.NewRecorder()
handler.ServeHTTP(rec, httptest.NewRequest(http.MethodGet, "/topo", nil))
if rec.Code != http.StatusOK {
t.Fatalf("status=%d", rec.Code)
}
body := rec.Body.String()
if !strings.Contains(body, "CPU 0") {
t.Fatalf("topo page missing CPU 0 box: %s", body)
}
if !strings.Contains(body, "GPU") {
t.Fatalf("topo page missing GPU box: %s", body)
}
if !strings.Contains(body, "var(--warn-fg)") {
t.Fatalf("topo page missing degraded-link warn edge color: %s", body)
}
}
func TestTopoPageLinkedFromNav(t *testing.T) {
handler := NewHandler(HandlerOptions{})
rec := httptest.NewRecorder()
handler.ServeHTTP(rec, httptest.NewRequest(http.MethodGet, "/", nil))
if rec.Code != http.StatusOK {
t.Fatalf("status=%d", rec.Code)
}
if !strings.Contains(rec.Body.String(), `href="/topo"`) {
t.Fatalf("nav missing /topo link")
}
}
func TestParseGPUPairAdjacencyRealTwoGPUDump(t *testing.T) {
// Real system/nvidia-smi-topo.txt from a support bundle for this exact
// server: two H100s directly bridged (NV17), spanning two NUMA nodes.
input := "\tGPU0\tGPU1\tNIC0\tNIC1\tCPU Affinity\tNUMA Affinity\tGPU NUMA ID\n" +
"GPU0\t X \tNV17\tSYS\tSYS\t0-23,48-71\t0\t\tN/A\n" +
"GPU1\tNV17\t X \tNODE\tNODE\t24-47,72-95\t1\t\tN/A\n" +
"NIC0\tSYS\tNODE\t X \tPIX\t\t\t\n" +
"NIC1\tSYS\tNODE\tPIX\t X \t\t\t\n"
pairs := parseGPUPairAdjacency(input)
if len(pairs) != 1 {
t.Fatalf("pairs=%d want 1 (%#v)", len(pairs), pairs)
}
if pairs[0].GPUA != 0 || pairs[0].GPUB != 1 || pairs[0].NVLinks != 17 {
t.Fatalf("pair=%#v want {0,1,17}", pairs[0])
}
}
func TestParseGPUPairAdjacencyDoesNotChainUnrelatedGPUs(t *testing.T) {
// 4 GPUs: only (0,1) and (2,3) are actually bonded. GPU1 and GPU2 must
// NOT get an edge just because they're adjacent in the layout.
input := "\tGPU0\tGPU1\tGPU2\tGPU3\n" +
"GPU0\t X \tNV18\tSYS\tSYS\n" +
"GPU1\tNV18\t X \tSYS\tSYS\n" +
"GPU2\tSYS\tSYS\t X \tNV18\n" +
"GPU3\tSYS\tSYS\tNV18\t X \n"
pairs := parseGPUPairAdjacency(input)
if len(pairs) != 2 {
t.Fatalf("pairs=%d want 2 (%#v)", len(pairs), pairs)
}
want := map[[2]int]bool{{0, 1}: true, {2, 3}: true}
for _, p := range pairs {
if !want[[2]int{p.GPUA, p.GPUB}] {
t.Fatalf("unexpected pair %#v (GPU1-GPU2 chaining bug?)", p)
}
}
}
func TestParseGPUPairAdjacencyEmptyOnNoMatrix(t *testing.T) {
if pairs := parseGPUPairAdjacency("no gpus here"); pairs != nil {
t.Fatalf("pairs=%v want nil", pairs)
}
}
func TestPcieGenRank(t *testing.T) {
if pcieGenRank("Gen5") <= pcieGenRank("Gen4") {
t.Fatalf("Gen5 should rank higher than Gen4")
}
if pcieGenRank("bogus") != 0 {
t.Fatalf("unparseable gen should rank 0")
}
}
func TestTopoEdgeColorVar(t *testing.T) {
gen3, gen4 := "Gen3", "Gen4"
degraded := schema.HardwarePCIeDevice{LinkSpeed: &gen3, MaxLinkSpeed: &gen4}
if got := topoEdgeColorVar(degraded); got != "var(--warn-fg)" {
t.Fatalf("degraded color=%q want warn", got)
}
full := schema.HardwarePCIeDevice{LinkSpeed: &gen4, MaxLinkSpeed: &gen4}
if got := topoEdgeColorVar(full); got != "var(--ok-fg)" {
t.Fatalf("full-speed color=%q want ok", got)
}
unknown := schema.HardwarePCIeDevice{}
if got := topoEdgeColorVar(unknown); got != "var(--muted)" {
t.Fatalf("unknown color=%q want muted", got)
}
}
func TestBuildSocketIndex(t *testing.T) {
s0, s1 := 0, 1
cpus := []schema.HardwareCPU{{Socket: &s1}, {Socket: &s0}}
idx := buildSocketIndex(cpus)
if idx[0] != 1 || idx[1] != 0 {
t.Fatalf("idx=%#v want {0:1, 1:0}", idx)
}
}
func TestIsRAIDControllerClass(t *testing.T) {
if !isRAIDControllerClass("StorageController") || !isRAIDControllerClass("MassStorageController") {
t.Fatalf("expected known RAID/storage classes to match")
}
if isRAIDControllerClass("VideoController") {
t.Fatalf("GPU class should not match RAID classifier")
}
}
func TestIsNICDeviceClassDev(t *testing.T) {
class := "EthernetController"
nic := schema.HardwarePCIeDevice{DeviceClass: &class}
if !isNICDeviceClassDev(nic) {
t.Fatalf("expected EthernetController to classify as NIC")
}
withMac := schema.HardwarePCIeDevice{MacAddresses: []string{"aa:bb:cc:dd:ee:ff"}}
if !isNICDeviceClassDev(withMac) {
t.Fatalf("expected device with MAC address to classify as NIC")
}
other := schema.HardwarePCIeDevice{}
if isNICDeviceClassDev(other) {
t.Fatalf("expected empty device to not classify as NIC")
}
}
func TestTopoStatusBadgeClassNilIsUnknown(t *testing.T) {
label, cls := topoStatusBadgeClass(nil)
if label != "?" || cls != "badge-unknown" {
t.Fatalf("nil status = (%q,%q) want (?, badge-unknown)", label, cls)
}
ok := "OK"
label, cls = topoStatusBadgeClass(&ok)
if label != "OK" || cls != "badge-ok" {
t.Fatalf("OK status = (%q,%q) want (OK, badge-ok)", label, cls)
}
}
func TestParseTopoNVLinkStatus(t *testing.T) {
input := `GPU 0: NVIDIA H100 80GB HBM3 (UUID: GPU-a59f6931-c099-8fba-a0b3-08469d86f140)
Link 0: 26.562 GB/s
Link 15: <inactive>
GPU 1: NVIDIA H100 80GB HBM3 (UUID: GPU-603fe750-0516-9db5-86ec-ea61af3fce35)
Link 0: 26.562 GB/s
`
got := parseTopoNVLinkStatus(input)
if len(got[0]) != 2 || got[0][1].Active {
t.Fatalf("gpu0=%#v want link15 inactive", got[0])
}
if len(got[1]) != 1 || got[1][0].SpeedGBs == nil || *got[1][0].SpeedGBs != 26.562 {
t.Fatalf("gpu1=%#v want link0 26.562 GB/s", got[1])
}
}
func TestParseTopoNVLinkErrors(t *testing.T) {
input := `GPU 0: NVIDIA H100 80GB HBM3 (UUID: GPU-a59f6931-c099-8fba-a0b3-08469d86f140)
Link 0: Replay Errors: 0
Link 0: Recovery Errors: 0
Link 0: CRC Errors: 0
Link 1: Replay Errors: 3
Link 1: Recovery Errors: 1
Link 1: CRC Errors: 2
`
got := parseTopoNVLinkErrors(input)
c := got[0][1]
if c[0] != 3 || c[1] != 1 || c[2] != 2 {
t.Fatalf("link1 counters=%#v want {3,1,2}", c)
}
}