Restrict stats display to admin users and enhance stats information

- Modified analysis page to only show stats sidebar for admin users
- Enhanced /api/stats endpoint to include cluster information for admins
- Added GPU backend detection summary to stats
- Updated JavaScript to display comprehensive system and cluster stats
- Stats now show local resource usage and cluster status for administrators

Note: Full job-specific worker stats (showing resources from the machine executing each specific job) would require additional development to track job-to-worker mappings and implement worker resource reporting.

templates/analyze.html

@@ -58,29 +58,42 @@
     </style>
 <script>
         function updateStats() {
-            fetch('/stats')
+            fetch('/api/stats')
                 .then(response => response.json())
                 .then(data => {
-                    let html = '<h3>GPU Stats</h3>';
-                    html += `<p style="color: ${data.status === 'Idle' ? 'green' : 'orange'};">Status: ${data.status}</p>`;
-                    if (data.elapsed > 0) {
-                        html += `<p>Elapsed: ${data.elapsed.toFixed(1)}s</p>`;
+                    let html = '<h3>System Stats</h3>';
+
+                    // GPU Information
+                    if (data.gpu_info) {
+                        html += '<h4>GPU Information</h4>';
+                        html += `<p>CUDA: ${data.gpu_info.cuda_available ? 'Available' : 'Not available'} (${data.gpu_info.cuda_devices} devices)</p>`;
+                        html += `<p>ROCm: ${data.gpu_info.rocm_available ? 'Available' : 'Not available'} (${data.gpu_info.rocm_devices} devices)</p>`;
+                        html += `<p>Available backends: ${data.gpu_info.available_backends.join(', ') || 'None'}</p>`;
                     }
+
+                    // Local GPU stats
                     if (data.gpu_count > 0) {
+                        html += '<h4>Local GPU Usage</h4>';
                         data.gpus.forEach((gpu, i) => {
                             let memPercent = (gpu.memory_used / gpu.memory_total * 100).toFixed(1);
                             html += `<p>GPU ${i}: ${gpu.name}<br>Memory: <progress value="${gpu.memory_used}" max="${gpu.memory_total}"></progress> ${gpu.memory_used.toFixed(2)} / ${gpu.memory_total.toFixed(2)} GB (${memPercent}%)<br>Utilization: ${gpu.utilization}%</p>`;
                         });
-                    } else {
-                        html += '<p>No GPUs detected</p>';
                     }
+
+                    // CPU and RAM
+                    html += '<h4>Local Resources</h4>';
                     html += `<p>CPU: ${data.cpu_percent.toFixed(1)}%</p>`;
                     html += `<p>RAM: ${data.ram_used.toFixed(2)} / ${data.ram_total.toFixed(2)} GB</p>`;
-                    document.getElementById('stats').innerHTML = html;
-                    if (data.result) {
-                        document.getElementById('result_div').innerHTML = '<h3>Result:</h3><p>' + data.result + '</p>';
-                        document.getElementById('result_div').style.display = 'block';
+
+                    // Cluster stats (admin only)
+                    if (data.cluster_clients !== undefined) {
+                        html += '<h4>Cluster Status</h4>';
+                        html += `<p>Connected clients: ${data.cluster_clients}</p>`;
+                        html += `<p>Active processes: ${data.active_processes || 0}</p>`;
+                        html += `<p>GPU-enabled clients: ${data.gpu_clients || 0}</p>`;
                     }
+
+                    document.getElementById('stats').innerHTML = html;
                 })
                 .catch(e => {
                     document.getElementById('stats').innerHTML = '<p>Error loading stats</p>';
@@ -246,9 +259,11 @@
             {% endif %}
         </div>
 
+        {% if user.get('role') == 'admin' %}
         <div class="sidebar">
             <div id="stats" class="stats">Loading stats...</div>
         </div>
+        {% endif %}
     </div>
 
     <!-- File Browser Modal -->

vidai/api.py

@@ -80,9 +80,23 @@ def api_stats():
     import torch
     import time
 
-    data = {'status': 'Idle'}  # Simplified - in real implementation, get from backend
+    user = request.api_user
+    data = {'status': 'Idle'}
+
+    # For admin users, show more detailed stats including cluster information
+    if user.get('role') == 'admin':
+        # Check if we have cluster master information
+        try:
+            from .cluster_master import cluster_master
+            # Add cluster stats if available
+            data['cluster_clients'] = len(cluster_master.clients)
+            data['active_processes'] = len(cluster_master.processes)
+            data['gpu_clients'] = sum(1 for c in cluster_master.clients.values()
+                                    if c['gpu_info'].get('cuda_available') or c['gpu_info'].get('rocm_available'))
+        except:
+            pass
 
-    # GPU stats
+    # GPU stats (local machine)
     if torch.cuda.is_available():
         data['gpu_count'] = torch.cuda.device_count()
         data['gpus'] = []
@@ -91,18 +105,31 @@ def api_stats():
                 'name': torch.cuda.get_device_name(i),
                 'memory_used': torch.cuda.memory_allocated(i) / 1024**3,  # GB
                 'memory_total': torch.cuda.get_device_properties(i).total_memory / 1024**3,
-                'utilization': 0  # Would need pynvml for actual utilization
+                'utilization': 0,  # Would need pynvml for actual utilization
+                'backend': 'cuda'
             }
             data['gpus'].append(gpu)
     else:
         data['gpu_count'] = 0
+        data['gpus'] = []
 
-    # CPU and RAM
+    # CPU and RAM (local machine)
     data['cpu_percent'] = psutil.cpu_percent()
     ram = psutil.virtual_memory()
     data['ram_used'] = ram.used / 1024**3
     data['ram_total'] = ram.total / 1024**3
 
+    # Add GPU info summary
+    from .compat import detect_gpu_backends
+    gpu_info = detect_gpu_backends()
+    data['gpu_info'] = {
+        'cuda_available': gpu_info['cuda'],
+        'rocm_available': gpu_info['rocm'],
+        'cuda_devices': gpu_info['cuda_devices'],
+        'rocm_devices': gpu_info['rocm_devices'],
+        'available_backends': [k for k, v in gpu_info.items() if k.endswith('_available') and v]
+    }
+
     return json.dumps(data)
 
 @api_bp.route('/api/analyze', methods=['POST'])