fix(nvidia): surface Xid 79/154 GPU bus-fall-off as a physical-reboot-required signal

nvidia-bug-report.sh appends .gz to --output-file when gzip is available,
which silently produced empty nvidia-bug-report.txt in support bundles
(cat looked for the uncompressed name that never existed).

Also: a GPU that falls off the PCIe/NVLink bus (Xid 79) or gets flagged
for Node Reboot Required (Xid 154) mid-SAT-run left every downstream test
failing with generic, unrelated-looking errors (CUDA "unknown error",
"unable to determine device handle") with no indication the GPU needed a
physical power-cycle to recover. Detect these codes from SAT run logs and
surface a plain-English "physical reboot required" message in the task's
failure detail, the persisted component-status DB, a dashboard banner on
the Hardware Summary card, and topology diagram GPU-node severity.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

audit/internal/webui/pages.go

@@ -138,6 +138,7 @@ function openComponentDetail(type) {
 func renderDashboard(opts HandlerOptions) string {
 	var b strings.Builder
 	b.WriteString(renderAuditStatusBanner(opts))
+	b.WriteString(renderTimeSyncCard())
 	b.WriteString(renderHardwareSummaryCard(opts))
 	b.WriteString(renderHealthCard(opts))
 	b.WriteString(renderMetrics())
@@ -152,6 +153,45 @@ setInterval(function(){
 	return b.String()
 }
 
+// renderTimeSyncCard shows the server's current time and a button that syncs
+// the host clock and timezone to whatever the client's browser reports.
+func renderTimeSyncCard() string {
+	return `<div class="card" style="margin-bottom:16px">
+  <div class="card-head card-head-actions">
+    <span>System Time</span>
+    <div class="card-head-buttons">
+      <button id="time-sync-btn" class="btn btn-primary btn-sm" onclick="timeSyncRun()">&#8635; Sync with this browser</button>
+    </div>
+  </div>
+  <div class="card-body">
+    <span id="time-sync-status" style="font-size:13px;color:var(--muted)"></span>
+  </div>
+</div>
+<script>
+function timeSyncRun() {
+  var btn = document.getElementById('time-sync-btn');
+  var status = document.getElementById('time-sync-status');
+  btn.disabled = true;
+  var tz = Intl.DateTimeFormat().resolvedOptions().timeZone || '';
+  fetch('/api/system/time-sync', {
+    method: 'POST',
+    headers: {'Content-Type': 'application/json'},
+    body: JSON.stringify({timezone: tz, epoch_ms: Date.now()})
+  })
+    .then(function(r) { if (!r.ok) return r.text().then(function(t){throw new Error(t || r.statusText);}); return r.json(); })
+    .then(function(d) {
+      status.style.color = 'var(--ok-fg,#2c662d)';
+      status.textContent = '✓ Synced to ' + tz + ' at ' + new Date().toLocaleString();
+    })
+    .catch(function(err) {
+      status.style.color = 'var(--crit-fg,#9f3a38)';
+      status.textContent = '✗ Sync failed: ' + err.message;
+    })
+    .finally(function() { btn.disabled = false; });
+}
+</script>`
+}
+
 // renderAuditStatusBanner shows a live progress banner when an audit task is
 // running and auto-reloads the page when it completes.
 func renderAuditStatusBanner(opts HandlerOptions) string {
@@ -242,10 +282,16 @@ func renderHardwareSummaryCard(opts HandlerOptions) string {
 	if db, err := app.OpenComponentStatusDB(filepath.Join(opts.ExportDir, "component-status.json")); err == nil {
 		records = db.All()
 	}
+	gpuRecords := matchedRecords(records, nil, []string{"pcie:gpu:"})
 
 	var b strings.Builder
 	b.WriteString(`<div class="card"` + cardID + `><div class="card-head">Hardware Summary</div><div class="card-body">`)
 
+	if reason, needsReboot := gpuNeedsPhysicalReboot(gpuRecords); needsReboot {
+		fmt.Fprintf(&b, `<div class="alert alert-crit">&#9888; GPU needs a physical reboot/power-cycle — further GPU tests will keep failing until then: %s</div>`,
+			html.EscapeString(reason))
+	}
+
 	// Server identity block above the component table.
 	{
 		var model, serial string
@@ -297,8 +343,7 @@ func renderHardwareSummaryCard(opts HandlerOptions) string {
 	writeRow("Storage", hwDescribeStorage(hw),
 		renderComponentChips(matchedRecords(records, []string{"storage:all"}, []string{"storage:"})), "storage")
 
-	writeRow("GPU", hwDescribeGPU(hw),
-		renderComponentChips(matchedRecords(records, nil, []string{"pcie:gpu:"})), "gpu")
+	writeRow("GPU", hwDescribeGPU(hw), renderComponentChips(gpuRecords), "gpu")
 
 	psuMatched := matchedRecords(records, nil, []string{"psu:"})
 	if len(psuMatched) == 0 && len(hw.PowerSupplies) > 0 {
@@ -986,6 +1031,25 @@ func runtimeIssueDescriptions(issues []schema.RuntimeIssue, codes ...string) str
 	return strings.Join(messages, "; ")
 }
 
+// gpuNeedsPhysicalReboot reports whether any GPU component record carries a
+// hardware-fault reason that a driver reset can't clear (e.g. Xid 79 "GPU
+// has fallen off the bus", Xid 154 "Node Reboot Required" — see
+// collector.xidHardwareFaultMessages). Those errors mean every subsequent
+// GPU SAT job will keep failing until the node is physically power-cycled,
+// so this drives a dashboard banner that says so up front instead of making
+// an operator burn another test cycle to rediscover it.
+func gpuNeedsPhysicalReboot(records []app.ComponentStatusRecord) (reason string, needsReboot bool) {
+	for _, rec := range records {
+		if !strings.EqualFold(strings.TrimSpace(rec.Status), "Critical") {
+			continue
+		}
+		if strings.Contains(strings.ToLower(rec.ErrorSummary), "reboot") {
+			return rec.ErrorSummary, true
+		}
+	}
+	return "", false
+}
+
 // chipLetterClass maps a component status to a single display letter and CSS class.
 func chipLetterClass(status string) (letter, cls string) {
 	switch strings.ToUpper(strings.TrimSpace(status)) {