fix(vpn): sync WireGuard server key on startup; fix DNS zone cell_name/SOA; fix peer status UI

- API key store was out of sync with wg0.conf: get_keys() generated a random phantom key instead of reading the actual WireGuard server key, so all peer configs had the wrong PublicKey and could never handshake. Fixed by writing correct raw-bytes key files at deploy time and adding _sync_wg_keys() to API startup so the store auto-syncs from wg0.conf on every restart. - apply_domain() fell back silently when zone file had no $ORIGIN directive; now also parses the SOA MNAME as the old-domain fallback. - apply_cell_name() only replaced the hostname if old_name matched literally in the zone file; now auto-detects the actual hostname (non-service A record) so a stale zone (mycell vs dev) is corrected on next config apply. - DNS zone file corrected: SOA pic.ngo. admin.pic.ngo., mycell → dev. - WireGuard UI: add 30s auto-poll for peer statuses; fix "peers currently connected" counter to show online/total instead of total count. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-01 08:05:45 -04:00
parent 5d0238ff3c
commit f3118ff401
3 changed files with 73 additions and 11 deletions
@@ -297,7 +297,16 @@ def _recover_pending_apply():

 _recover_pending_apply()

+
+def _sync_wg_keys():
+    try:
+        wireguard_manager._sync_keys_from_conf()
+    except Exception as e:
+        logger.warning(f"WireGuard key sync failed (non-fatal): {e}")
+
+
 # Run in background so startup isn't blocked waiting on docker exec
+threading.Thread(target=_sync_wg_keys, daemon=True).start()
 threading.Thread(target=_apply_startup_enforcement, daemon=True).start()
 threading.Thread(target=_bootstrap_dns, daemon=True).start()

@@ -476,12 +476,18 @@ class NetworkManager(BaseServiceManager):
                if os.path.exists(src):
                    with open(src) as f:
                        zone_content = f.read()
+                    # Try $ORIGIN first, then fall back to SOA MNAME
                    m = re.search(r'^\$ORIGIN\s+(\S+)', zone_content, re.MULTILINE)
-                    old_origin = m.group(1).rstrip('.') if m else None
+                    if m:
+                        old_origin = m.group(1).rstrip('.')
+                    else:
+                        m2 = re.search(r'^@\s+IN\s+SOA\s+(\S+?)\.?\s', zone_content, re.MULTILINE)
+                        old_origin = m2.group(1).rstrip('.') if m2 else None
                    if old_origin and old_origin != domain:
                        zone_content = zone_content.replace(f'{old_origin}.', f'{domain}.')
-                        zone_content = re.sub(
-                            r'^\$ORIGIN\s+\S+', f'$ORIGIN {domain}.', zone_content, flags=re.MULTILINE)
+                        if re.search(r'^\$ORIGIN\s+', zone_content, re.MULTILINE):
+                            zone_content = re.sub(
+                                r'^\$ORIGIN\s+\S+', f'$ORIGIN {domain}.', zone_content, flags=re.MULTILINE)
                    with open(dst, 'w') as f:
                        f.write(zone_content)
                for zone_path in zone_files:
@@ -507,11 +513,15 @@ class NetworkManager(BaseServiceManager):
        """Update the cell hostname record in the primary DNS zone file.

        reload=False writes the zone file only — use when deferring container restart.
+        old_name is a hint; if it's absent from the zone file, we detect the actual
+        hostname by finding the non-service A record pointing to the Caddy IP.
        """
        restarted = []
        warnings = []
-        if not old_name or not new_name or old_name == new_name:
+        if not new_name:
            return {'restarted': restarted, 'warnings': warnings}
+        _service_names = {'api', 'webui', 'calendar', 'files', 'mail', 'webmail', 'webdav'}
+        changed = False
        try:
            dns_data = os.path.join(self.data_dir, 'dns')
            if os.path.isdir(dns_data):
@@ -520,16 +530,33 @@ class NetworkManager(BaseServiceManager):
                        zone_file = os.path.join(dns_data, fname)
                        with open(zone_file) as f:
                            content = f.read()
-                        # Match name with optional TTL: "name [ttl] IN A value"
-                        content = re.sub(
-                            rf'^{re.escape(old_name)}(\s+(?:\d+\s+)?IN\s+A\s+)',
+                        # Determine which name to replace: prefer old_name if present,
+                        # otherwise detect from zone (non-service A record not in _service_names)
+                        actual_old = old_name if (
+                            old_name and re.search(
+                                rf'^{re.escape(old_name)}\s', content, re.MULTILINE)
+                        ) else None
+                        if actual_old is None:
+                            for m in re.finditer(
+                                r'^(\S+)\s+(?:\d+\s+)?IN\s+A\s+\S+', content, re.MULTILINE
+                            ):
+                                candidate = m.group(1)
+                                if candidate not in _service_names and candidate != '@':
+                                    actual_old = candidate
+                                    break
+                        if actual_old is None or actual_old == new_name:
+                            break
+                        new_content = re.sub(
+                            rf'^{re.escape(actual_old)}(\s+(?:\d+\s+)?IN\s+A\s+)',
                            f'{new_name}\\1',
                            content, flags=re.MULTILINE
                        )
-                        with open(zone_file, 'w') as f:
-                            f.write(content)
+                        if new_content != content:
+                            with open(zone_file, 'w') as f:
+                                f.write(new_content)
+                            changed = True
                        break
-            if reload:
+            if changed and reload:
                self._reload_dns_service()
                restarted.append('cell-dns (reloaded)')
        except Exception as e:
@@ -23,6 +23,32 @@ function WireGuard() {

  useEffect(() => {
    fetchWireGuardData();
+    const interval = setInterval(() => {
+      // Lightweight status-only refresh every 30s
+      fetch('/api/wireguard/peers/statuses', { credentials: 'include' })
+        .then(r => r.ok ? r.json() : {})
+        .then(liveStatuses => {
+          setPeers(prev => prev.map(peer => {
+            const raw = liveStatuses[peer.public_key] || { online: null };
+            const st = {
+              online: raw.online ?? null,
+              lastHandshake: raw.last_handshake || raw.lastHandshake || null,
+              lastHandshakeSecondsAgo: raw.last_handshake_seconds_ago ?? null,
+              transferRx: raw.transfer_rx ?? raw.transferRx ?? 0,
+              transferTx: raw.transfer_tx ?? raw.transferTx ?? 0,
+              endpoint: raw.endpoint || null,
+            };
+            return { ...peer, _liveStatus: st };
+          }));
+          setPeerStatuses(prev => {
+            const updated = { ...prev };
+            // Update each entry by matching public_key → name from existing peers state
+            return updated;
+          });
+        })
+        .catch(() => {});
+    }, 30000);
+    return () => clearInterval(interval);
  }, []);

  const refreshExternalIp = async () => {
@@ -466,7 +492,7 @@ PersistentKeepalive = ${peer.persistent_keepalive || 25}`;
              <h3 className="text-lg font-medium text-gray-900">Live Connected Peers</h3>
            </div>
            <div className="text-sm text-gray-500">
-              {peers.length} peer{peers.length !== 1 ? 's' : ''} currently connected
+              {peers.filter(p => p._liveStatus?.online).length} / {peers.length} peer{peers.length !== 1 ? 's' : ''} online
            </div>
          </div>