Update 1.2.2.2 beta

AppImage/ProxMenux-Monitor.AppImage.sha256

@@ -1 +1 @@
-30c427079c3b2b829ece1b801f045c76b8dd7d16286b5ce243f56a5df9928eb8  ProxMenux-1.2.2.2-beta.AppImage
+6bd898c3801bdf27b983ca5c107421d485e163f0c339cbc5f5cc90c9bb26a4d6  ProxMenux-1.2.2.2-beta.AppImage

AppImage/components/node-metrics-charts.tsx

@@ -66,11 +66,17 @@ const CustomMemoryTooltip = ({ active, payload, label }: any) => {
   return null
 }
 
+interface MetricsError {
+  headline: string
+  details?: string
+  suggestion?: string
+}
+
 export function NodeMetricsCharts() {
   const [timeframe, setTimeframe] = useState("day")
   const [data, setData] = useState<NodeMetricsData[]>([])
   const [loading, setLoading] = useState(true)
-  const [error, setError] = useState<string | null>(null)
+  const [error, setError] = useState<MetricsError | null>(null)
   const isMobile = useIsMobile()
 
   const [visibleLines, setVisibleLines] = useState({
@@ -160,9 +166,16 @@ export function NodeMetricsCharts() {
       setData(transformedData)
     } catch (err: any) {
       console.error("Error fetching node metrics:", err)
-      console.error("Error message:", err.message)
+      // fetchApi attaches the parsed JSON body to err.body. The metrics
-      console.error("Error stack:", err.stack)
+      // endpoint enriches 503 responses with `details` (Proxmox-side
-      setError(err.message || "Error loading metrics")
+      // diagnostic) and `suggestion` (how to fix). Pull them through so
+      // the user sees actionable text instead of a bare "503".
+      const body = err?.body
+      setError({
+        headline: body?.error || err?.message || "Error loading metrics",
+        details: body?.details,
+        suggestion: body?.suggestion,
+      })
     } finally {
       setLoading(false)
     }
@@ -231,24 +244,36 @@ export function NodeMetricsCharts() {
   }
 
   if (error) {
+    // Both panels carry the same error — render an identical card on
+    // each side. The headline is the short cause, the details block
+    // explains it's a Proxmox-host issue (not a Monitor bug), and the
+    // suggestion is the exact command the operator should run.
+    const errorCard = (
+      <Card className="bg-card border-border">
+        <CardContent className="p-6">
+          <div className="flex flex-col items-start justify-center h-[300px] gap-2 px-2 overflow-auto">
+            <p className="text-sm font-semibold text-red-400">{error.headline}</p>
+            {error.details && (
+              <p className="text-xs text-muted-foreground leading-relaxed">{error.details}</p>
+            )}
+            {error.suggestion && (
+              <div className="w-full mt-2">
+                <p className="text-[10px] uppercase tracking-wide text-muted-foreground mb-1">
+                  Suggested fix on the Proxmox host
+                </p>
+                <code className="block text-xs bg-background/60 border border-border rounded px-2 py-1.5 font-mono break-all">
+                  {error.suggestion}
+                </code>
+              </div>
+            )}
+          </div>
+        </CardContent>
+      </Card>
+    )
     return (
       <div className="grid grid-cols-1 lg:grid-cols-2 gap-6">
-        <Card className="bg-card border-border">
+        {errorCard}
-          <CardContent className="p-6">
+        {errorCard}
-            <div className="flex flex-col items-center justify-center h-[300px] gap-2">
-              <p className="text-muted-foreground text-sm">Metrics data not available yet</p>
-              <p className="text-xs text-red-500">{error}</p>
-            </div>
-          </CardContent>
-        </Card>
-        <Card className="bg-card border-border">
-          <CardContent className="p-6">
-            <div className="flex flex-col items-center justify-center h-[300px] gap-2">
-              <p className="text-muted-foreground text-sm">Metrics data not available yet</p>
-              <p className="text-xs text-red-500">{error}</p>
-            </div>
-          </CardContent>
-        </Card>
       </div>
     )
   }

AppImage/lib/api-config.ts

@@ -138,6 +138,12 @@ export async function fetchApi<T>(endpoint: string, options?: RequestInit): Prom
       // return `{error: "..."}` on failure (e.g. /api/vms/<id>/control
       // includes the pvesh stderr — telling the user "no space left on
       // device" is infinitely more useful than the raw status text).
+      //
+      // We also attach the FULL parsed JSON body to the thrown Error
+      // as `.body` so callers that want the optional `details` /
+      // `suggestion` fields (e.g. /api/node/metrics) can render them
+      // without re-fetching. Callers that just read `err.message`
+      // keep working exactly as before.
       try {
         const ct = response.headers.get("content-type") || ""
         if (ct.includes("application/json")) {
@@ -145,7 +151,10 @@ export async function fetchApi<T>(endpoint: string, options?: RequestInit): Prom
           const detail =
             (body && (body.error || body.message)) || ""
           if (detail) {
-            throw new Error(detail)
+            const e: Error & { body?: unknown; status?: number } = new Error(detail)
+            e.body = body
+            e.status = response.status
+            throw e
           }
         }
       } catch (parseErr) {

AppImage/scripts/flask_server.py

@@ -9415,7 +9415,13 @@ def _update_smart_cron():
         test_type = schedule.get('test_type', 'short')
         retention = schedule.get('retention', 10)
         
-        cmd = f'/usr/local/share/proxmenux/scripts/smart-scheduled-test.sh --schedule-id {schedule_id} --test-type {test_type} --retention {retention}'
+        # The installer keeps the storage helpers under
+        # /usr/local/share/proxmenux/scripts/storage/, mirroring the
+        # repo layout. The previous path (without /storage/) generated
+        # cron lines that pointed at a non-existent file, so the timer
+        # fired, cron logged nothing, and the operator saw "no
+        # scheduled tests in history" with no obvious clue why.
+        cmd = f'/usr/local/share/proxmenux/scripts/storage/smart-scheduled-test.sh --schedule-id {schedule_id} --test-type {test_type} --retention {retention}'
         if disks != ['all']:
             cmd += f" --disks '{','.join(disks)}'"
         
@@ -9999,6 +10005,57 @@ def api_node_metrics():
                                     '--timeframe', timeframe, '--output-format', 'json'],
                                    capture_output=True, text=True, timeout=10)
         
+        # Detect well-known Proxmox-side failures BEFORE trying to parse
+        # the JSON. These are PVE host problems (rrdcached down, RRD file
+        # corrupt, node-name mismatch). None of them are caused by the
+        # Monitor itself — surface a specific message so the operator
+        # doesn't blame ProxMenux for a Proxmox-host data-store issue.
+        if rrd_result.returncode != 0:
+            stderr_str = (rrd_result.stderr or '') + (rrd_result.stdout or '')
+            stderr_lower = stderr_str.lower()
+            if 'mmaping file' in stderr_lower and 'invalid argument' in stderr_lower:
+                # Corrupt RRD file on disk. Operator must recreate it.
+                return jsonify({
+                    'error': 'Proxmox RRD database is corrupt',
+                    'details': (
+                        'The host metrics file Proxmox keeps under '
+                        '/var/lib/rrdcached/db/pve-node-9.0/ failed to '
+                        'memory-map (Invalid argument). This is a Proxmox-side '
+                        'data-store issue, not a Monitor bug.'
+                    ),
+                    'suggestion': (
+                        'Stop pvestatd + pve-cluster + rrdcached, move the '
+                        'broken RRD aside, restart the services. Proxmox will '
+                        'rebuild the RRD from scratch (history is lost).'
+                    ),
+                    'raw': stderr_str.strip()[:500],
+                }), 503
+            if 'no such file' in stderr_lower or 'no such node' in stderr_lower or 'does not exist' in stderr_lower:
+                return jsonify({
+                    'error': 'Proxmox node name mismatch',
+                    'details': (
+                        f"pvesh could not find node '{local_node}'. The "
+                        'usual cause is that the host was renamed after '
+                        'Proxmox was installed, so /etc/pve/nodes/ still '
+                        'carries the old name. This is a Proxmox-side '
+                        'config issue, not a Monitor bug.'
+                    ),
+                    'suggestion': 'Compare `hostname` with `ls /etc/pve/nodes/` — they must match.',
+                    'raw': stderr_str.strip()[:500],
+                }), 503
+            if 'rrd' in stderr_lower or 'empty' in stderr_lower:
+                return jsonify({
+                    'error': 'Proxmox RRD data not available',
+                    'details': 'The RRD database appears empty. Proxmox may not have collected metrics yet (fresh install) or rrdcached was down at boot.',
+                    'suggestion': 'systemctl restart rrdcached pvestatd ; wait ~5 min and reload this page.',
+                    'raw': stderr_str.strip()[:500],
+                }), 503
+            return jsonify({
+                'error': 'Proxmox metrics command failed',
+                'details': 'pvesh exited non-zero. Check Proxmox host status.',
+                'raw': stderr_str.strip()[:500],
+            }), 503
+
         if rrd_result.returncode == 0:
             rrd_data = json.loads(rrd_result.stdout)
 
@@ -10044,26 +10101,19 @@ def api_node_metrics():
             }
             _node_metrics_cache_set(timeframe, payload)
             return jsonify(payload)
-        else:
+        # Note: the old `else` branch that handled rrd_result.returncode != 0
-            # Check if RRD file is empty or corrupted
+        # was removed — the early-return block above now catches every
-            stderr_lower = rrd_result.stderr.lower() if rrd_result.stderr else ''
+        # non-zero exit BEFORE we ever attempt json.loads(), so reaching
-            if 'rrd' in stderr_lower or 'no such file' in stderr_lower or 'empty' in stderr_lower:
+        # this point with returncode != 0 is impossible.
-                return jsonify({
-                    'error': 'RRD data not available',
-                    'details': 'The RRD database file may be empty or corrupted. This can happen if rrdcached was not running properly after Proxmox installation.',
-                    'suggestion': 'Try restarting rrdcached: systemctl restart rrdcached'
-                }), 503  # Service Unavailable - more appropriate than 500
-            return jsonify({'error': f'Failed to get RRD data: {rrd_result.stderr}'}), 500
 
     except json.JSONDecodeError:
         # pvesh returned invalid JSON - likely empty RRD
         return jsonify({
-            'error': 'RRD data not available',
+            'error': 'Proxmox RRD data not available',
-            'details': 'Unable to parse metrics data. The RRD database may be empty or corrupted.',
+            'details': 'pvesh returned non-JSON output. The RRD database is likely empty (fresh install where pvestatd has not run yet) or the rrdcached daemon is down.',
-            'suggestion': 'Try restarting rrdcached: systemctl restart rrdcached'
+            'suggestion': 'systemctl restart rrdcached pvestatd ; wait ~5 min and reload.',
         }), 503
     except Exception as e:
-
         return jsonify({'error': str(e)}), 500
 
 @app.route('/api/logs/counts', methods=['GET'])
@@ -16711,6 +16761,17 @@ if __name__ == '__main__':
     import sys
     import logging
 
+    # Regenerate /etc/cron.d/proxmenux-smart from the current schedule
+    # config on every Monitor start. Pre-existing installs may carry
+    # a cron file with the broken script path (without /storage/),
+    # left over from before the path-fix in this build. Rewriting it
+    # at startup is idempotent and silently heals those hosts on the
+    # first restart after the upgrade — no operator action needed.
+    try:
+        _update_smart_cron()
+    except Exception as _e:
+        print(f"[startup] smart cron regen failed: {_e}", file=sys.stderr)
+    
     # Custom filter to suppress TLS handshake noise when running HTTP
     # (browsers may cache HTTPS and keep sending TLS ClientHello to an HTTP server)
     class TLSNoiseFilter(logging.Filter):