feat(topo): per-socket bar layout with disks under their controller

Rework the /topo diagram from side-by-side stacked cards into one tall vertical bar per CPU socket with everything attached to it branching off sideways (socket 0 left/branches right, socket 1 right/branches left). Disks are now parented under the storage controller they physically hang off (SATA/AHCI, SAS HBA, RAID) — itself a NUMA-affine PCIe device under one socket — instead of a synthetic catch-all node. The disk->controller link is read from a new storage-controllers.txt techdump (platform.StorageControllerMapScript, a /sys/block walk); disks with no resolvable controller fall back to an "Other" bar. No board/root node. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SYjrG6bVmeG1Z2Wmc8kg3o
2026-09-03 12:15:49 +03:00
parent b1ab866f58
commit 319f08ac4b
5 changed files with 634 additions and 100 deletions
@@ -152,6 +152,187 @@ func TestTopoPageRendersArbitraryPSUAndFirmwareCountsAsFlexRows(t *testing.T) {
 	}
 }

+func TestTopoPageRendersStorageDisksGroupedByType(t *testing.T) {
+	dir := t.TempDir()
+	path := filepath.Join(dir, "audit.json")
+
+	okStatus := "OK"
+	warnStatus := "WARNING"
+	ssd := "SSD"
+	nvme := "NVMe"
+	model := "SAMSUNG MZ7L3960HCJR-00B7C"
+	size := 960
+
+	ingest := schema.HardwareIngestRequest{
+		CollectedAt: "2026-03-15T00:00:00Z",
+		Hardware: schema.HardwareSnapshot{
+			Storage: []schema.HardwareStorage{
+				{HardwareComponentStatus: schema.HardwareComponentStatus{Status: &okStatus}, Type: &ssd, Model: &model, SizeGB: &size},
+				{HardwareComponentStatus: schema.HardwareComponentStatus{Status: &okStatus}, Type: &ssd, Model: &model, SizeGB: &size},
+				{HardwareComponentStatus: schema.HardwareComponentStatus{Status: &warnStatus}, Type: &nvme, SizeGB: &size},
+			},
+		},
+	}
+	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, "SSD ×2") {
+		t.Fatalf("topo page missing grouped SSD x2 card: %s", body)
+	}
+	if !strings.Contains(body, "NVMe") {
+		t.Fatalf("topo page missing NVMe disk card: %s", body)
+	}
+	// SSD card totals capacity across both disks.
+	if !strings.Contains(body, "1.9 TB total") {
+		t.Fatalf("topo page missing SSD total capacity: %s", body)
+	}
+	// The degraded NVMe disk must escalate that card's status.
+	nvmeIdx := strings.Index(body, ">NVMe<")
+	if nvmeIdx < 0 || !strings.Contains(body[nvmeIdx:nvmeIdx+400], "Warning") {
+		t.Fatalf("topo page missing NVMe warning status: %s", body)
+	}
+	// With no storage-controller techdump, disks have no resolvable socket and
+	// land under the "Other" bar rather than being glued to a CPU.
+	if !strings.Contains(body, ">Other<") {
+		t.Fatalf("topo page missing Other bar for unattached disks: %s", body)
+	}
+	clicks := strings.Count(body, `openComponentDetail('storage')`) + strings.Count(body, `openComponentDetail(&#39;storage&#39;)`)
+	if clicks != 2 {
+		t.Fatalf("expected one clickable card per disk-type group, got %d: %s", clicks, body)
+	}
+}
+
+func TestParseStorageControllerMap(t *testing.T) {
+	raw := "sda hctl=2:0:0:0 ctrl=0000:00:17.0\n" +
+		"sdb hctl=3:0:0:0 ctrl=0000:00:17.0\n" +
+		"nvme0n1 hctl= ctrl=0000:65:00.0\n"
+	got := parseStorageControllerMap(raw)
+	if got["2:0:0:0"] != "0000:00:17.0" || got["3:0:0:0"] != "0000:00:17.0" {
+		t.Fatalf("SATA disks not mapped to controller: %#v", got)
+	}
+	if _, ok := got["nvme0n1"]; ok {
+		t.Fatalf("NVMe line with empty hctl must be skipped: %#v", got)
+	}
+}
+
+func TestTopoPageParentsDisksUnderTheirControllerSocket(t *testing.T) {
+	dir := t.TempDir()
+	path := filepath.Join(dir, "audit.json")
+	techdump := filepath.Join(dir, "techdump")
+	if err := os.MkdirAll(techdump, 0755); err != nil {
+		t.Fatal(err)
+	}
+	// Both SSDs hang off the SATA controller at 0000:00:17.0, which is a
+	// NUMA-node-0 PCIe device -> they must render under CPU 0, not "Other".
+	if err := os.WriteFile(filepath.Join(techdump, "storage-controllers.txt"),
+		[]byte("sda hctl=2:0:0:0 ctrl=0000:00:17.0\nsdb hctl=3:0:0:0 ctrl=0000:00:17.0\n"), 0644); err != nil {
+		t.Fatal(err)
+	}
+
+	socket0, socket1 := 0, 1
+	numa0 := 0
+	okStatus := "OK"
+	sataClass := "SATA controller"
+	sataBDF := "0000:00:17.0"
+	hctlA, hctlB := "2:0:0:0", "3:0:0:0"
+	ssd := "SSD"
+	size := 960
+
+	ingest := schema.HardwareIngestRequest{
+		CollectedAt: "2026-03-15T00:00:00Z",
+		Hardware: schema.HardwareSnapshot{
+			CPUs: []schema.HardwareCPU{
+				{HardwareComponentStatus: schema.HardwareComponentStatus{Status: &okStatus}, Socket: &socket0},
+				{HardwareComponentStatus: schema.HardwareComponentStatus{Status: &okStatus}, Socket: &socket1},
+			},
+			PCIeDevices: []schema.HardwarePCIeDevice{
+				{HardwareComponentStatus: schema.HardwareComponentStatus{Status: &okStatus}, Slot: &sataBDF, DeviceClass: &sataClass, NUMANode: &numa0},
+			},
+			Storage: []schema.HardwareStorage{
+				{HardwareComponentStatus: schema.HardwareComponentStatus{Status: &okStatus}, Slot: &hctlA, Type: &ssd, SizeGB: &size},
+				{HardwareComponentStatus: schema.HardwareComponentStatus{Status: &okStatus}, Slot: &hctlB, Type: &ssd, SizeGB: &size},
+			},
+		},
+	}
+	data, _ := json.Marshal(ingest)
+	if err := os.WriteFile(path, data, 0644); err != nil {
+		t.Fatal(err)
+	}
+
+	handler := NewHandler(HandlerOptions{AuditPath: path, ExportDir: dir})
+	rec := httptest.NewRecorder()
+	handler.ServeHTTP(rec, httptest.NewRequest(http.MethodGet, "/topo", nil))
+	body := rec.Body.String()
+
+	if strings.Contains(body, ">Other<") {
+		t.Fatalf("disks resolved to a socket — no Other bar expected: %s", body)
+	}
+	if !strings.Contains(body, "SATA ctrl") {
+		t.Fatalf("topo page missing the SATA controller branch node: %s", body)
+	}
+	if !strings.Contains(body, "SSD ×2") {
+		t.Fatalf("topo page missing SSD disk group under the controller: %s", body)
+	}
+	// controller branch must render before its SSD sub-node
+	if strings.Index(body, "SATA ctrl") > strings.Index(body, "SSD ×2") {
+		t.Fatalf("controller node should render before its disks: %s", body)
+	}
+}
+
+func TestTopoPageRendersOneBarPerSocketNoRoot(t *testing.T) {
+	dir := t.TempDir()
+	path := filepath.Join(dir, "audit.json")
+
+	socket0, socket1 := 0, 1
+	okStatus := "OK"
+	board := "AS-4125GS-TNRT"
+	vendor := "Supermicro"
+
+	ingest := schema.HardwareIngestRequest{
+		CollectedAt: "2026-03-15T00:00:00Z",
+		Hardware: schema.HardwareSnapshot{
+			Board: schema.HardwareBoard{ProductName: &board, Manufacturer: &vendor},
+			CPUs: []schema.HardwareCPU{
+				{HardwareComponentStatus: schema.HardwareComponentStatus{Status: &okStatus}, Socket: &socket0},
+				{HardwareComponentStatus: schema.HardwareComponentStatus{Status: &okStatus}, Socket: &socket1},
+			},
+		},
+	}
+	data, _ := json.Marshal(ingest)
+	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))
+	body := rec.Body.String()
+
+	if !strings.Contains(body, ">CPU 0<") || !strings.Contains(body, ">CPU 1<") {
+		t.Fatalf("topo diagram missing a bar per CPU socket: %s", body)
+	}
+	// No board/root node in the diagram itself (board identity is the
+	// Firmware row lower down).
+	svg := body[strings.Index(body, "<svg"):]
+	svg = svg[:strings.Index(svg, "</svg>")]
+	if strings.Contains(svg, board) {
+		t.Fatalf("board node must not render inside the topology diagram: %s", svg)
+	}
+}
+
 func TestTopoPageLinkedFromNav(t *testing.T) {
 	handler := NewHandler(HandlerOptions{})
 	rec := httptest.NewRecorder()