fix(reliability): supervisor restores hop baseline unless a role holds the pause

payload/user/remote_access/pager-webui/server.py

@@ -3445,11 +3445,16 @@ def _read_hop():
     return out.strip()
 
 
+_HOP_PAUSE_HELD = False  # True while a Mark VIII role/attack holds hop off
+
+
 def _pause_hop():
+    global _HOP_PAUSE_HELD
     if _read_hop() != '0':
         device_run(['uci', 'set', 'pineapd.wlan1mon.hop=0'])
         device_run(['uci', 'commit', 'pineapd'])
         device_run(['/etc/init.d/pineapd', 'reload'])
+    _HOP_PAUSE_HELD = True
 
 
 def _resume_hop():
@@ -3468,6 +3473,8 @@ def _resume_hop():
     device_run(['uci', 'set', 'pineapd.wlan1mon.hop=%s' % wanted])
     device_run(['uci', 'commit', 'pineapd'])
     device_run(['/etc/init.d/pineapd', 'reload'])
+    global _HOP_PAUSE_HELD
+    _HOP_PAUSE_HELD = False
 
 
 def _remove_radio1_ap():
@@ -4789,6 +4796,20 @@ def health_check():
                     msg='reconciler re-applied: %s' % ','.join(rep['changed']))
         except Exception:
             pass
+        # Restore the hopping baseline unless a Mark VIII role/attack holds
+        # the pause intentionally (_resume_hop self-guards against fighting a
+        # live Pager hop enable).
+        if not _HOP_PAUSE_HELD:
+            try:
+                before = _read_hop()
+                if before == '0':
+                    import mk8_events
+                    _resume_hop()
+                    mk8_events.log_event(
+                        'guard_fix',
+                        msg='restored wlan1mon.hop baseline (was %s)' % before)
+            except Exception:
+                pass
     h['mem_percent'] = _mem_percent()
     if h['mem_percent'] >= MEM_WARN_PERCENT:
         h['mem_streak'] = h.get('mem_streak', 0) + 1