feat: support PBS hosts and remote PBS monitoring

AppImage/scripts/health_monitor.py

@@ -51,6 +51,7 @@ DEBUG_PERF = False
 # ─── Startup Grace Period ────────────────────────────────────────────────────
 # Import centralized startup grace management for consistent behavior
 import startup_grace
+from pbs_manager import get_platform, PBS_SERVICES
 
 def _is_startup_health_grace() -> bool:
     """Check if we're within the startup health grace period (5 min).
@@ -466,6 +467,7 @@ class HealthMonitor:
 
     def __init__(self):
         """Initialize health monitor with state tracking"""
+        self.platform_type = get_platform().get("product")
         self.state_history = defaultdict(list)
         self.last_check_times = {}
         self.cached_results = {}
@@ -919,9 +921,9 @@ class HealthMonitor:
         _perf_log("services", (time.time() - _t) * 1000)
         details['services'] = services_status
         if services_status['status'] == 'CRITICAL':
-            critical_issues.append(f"PVE Services: {services_status.get('reason', 'Service failure')}")
+            critical_issues.append(f"Proxmox Services: {services_status.get('reason', 'Service failure')}")
         elif services_status['status'] == 'WARNING':
-            warning_issues.append(f"PVE Services: {services_status.get('reason', 'Service issue')}")
+            warning_issues.append(f"Proxmox Services: {services_status.get('reason', 'Service issue')}")
         
         # Priority 1.5: Proxmox Storage Check (External Module)
         _t = time.time()
@@ -2347,6 +2349,8 @@ class HealthMonitor:
     # The new _check_proxmox_storage function handles this logic better.
     def _check_proxmox_storages(self) -> Dict[str, Any]:
         """Check Proxmox-specific storages (only report problems)"""
+        if getattr(self, "platform_type", "pve") == "pbs":
+            return {'status': 'OK', 'reason': 'Not applicable on PBS', 'checks': {}, 'not_applicable': True}
         storages = {}
         
         try:
@@ -3515,6 +3519,8 @@ class HealthMonitor:
         Optimized VM/CT check - detects qmp failures and startup errors from logs.
         Improved detection of container and VM errors from journalctl.
         """
+        if getattr(self, "platform_type", "pve") == "pbs":
+            return {'status': 'OK', 'reason': 'Not applicable on PBS', 'checks': {}, 'not_applicable': True}
         try:
             # First: auto-resolve any persisted VM/CT errors where the guest
             # is now running.  This clears stale "Failed to start" / QMP
@@ -3678,6 +3684,8 @@ class HealthMonitor:
         Check VMs/CTs with persistent error tracking.
         Errors persist until VM starts or 48h elapsed.
         """
+        if getattr(self, "platform_type", "pve") == "pbs":
+            return {'status': 'OK', 'reason': 'Not applicable on PBS', 'checks': {}, 'not_applicable': True}
         try:
             issues = []
             vm_details = {}
@@ -3919,8 +3927,18 @@ class HealthMonitor:
         """
         try:
             # Build service list: base PVE services + corosync if clustered
-            services_to_check = list(self.PVE_SERVICES)
-            is_cluster = os.path.exists('/etc/corosync/corosync.conf')
+            pbs_host = getattr(self, "platform_type", "pve") == "pbs"
+            service_prefix = "pbs" if pbs_host else "pve"
+            product_name = "PBS" if pbs_host else "PVE"
+            services_to_check = list(PBS_SERVICES) if pbs_host else list(self.PVE_SERVICES)
+            if pbs_host:
+                # Old Monitor versions incorrectly treated absent PVE services
+                # as PBS failures; remove only those known inapplicable errors.
+                for service in self.PVE_SERVICES:
+                    old_key = f'pve_service_{service}'
+                    if health_persistence.is_error_active(old_key):
+                        health_persistence.clear_error(old_key)
+            is_cluster = not pbs_host and os.path.exists('/etc/corosync/corosync.conf')
             if is_cluster and 'corosync' not in services_to_check:
                 services_to_check.append('corosync')
             
@@ -3952,14 +3970,14 @@ class HealthMonitor:
             # cleanup loop). Audit Tier 5 — Health stack.
             for svc in services_to_check:
                 if svc not in failed_services:
-                    error_key = f'pve_service_{svc}'
+                    error_key = f'{service_prefix}_service_{svc}'
                     if health_persistence.is_error_active(error_key):
                         health_persistence.clear_error(error_key)
             
             # Build checks dict with status per service
             checks = {}
             for svc in services_to_check:
-                error_key = f'pve_service_{svc}'
+                error_key = f'{service_prefix}_service_{svc}'
                 if svc in failed_services:
                     state = service_details.get(svc, 'inactive')
                     checks[svc] = {
@@ -3987,12 +4005,12 @@ class HealthMonitor:
                 # Record each failed service in persistence, respecting dismiss
                 active_failed = []
                 for svc in failed_services:
-                    error_key = f'pve_service_{svc}'
+                    error_key = f'{service_prefix}_service_{svc}'
                     rec_result = health_persistence.record_error(
                         error_key=error_key,
                         category='pve_services',
                         severity='CRITICAL',
-                        reason=f'PVE service {svc} is {service_details.get(svc, "inactive")}',
+                        reason=f'{product_name} service {svc} is {service_details.get(svc, "inactive")}',
                         details={'service': svc, 'state': service_details.get(svc, 'inactive')}
                     )
                     if rec_result and rec_result.get('type') == 'skipped_acknowledged':
@@ -5816,6 +5834,8 @@ class HealthMonitor:
         - No persistent errors are recorded
         This prevents false positives when NFS/PBS/remote storage is still mounting.
         """
+        if getattr(self, "platform_type", "pve") == "pbs":
+            return None
         if not PROXMOX_STORAGE_AVAILABLE:
             return None
         
@@ -6205,6 +6225,8 @@ class HealthMonitor:
         Notification cooldown is the standard 1h via the storage
         category, so a CT teetering at 96% won't spam the channel.
         """
+        if getattr(self, "platform_type", "pve") == "pbs":
+            return None
         # Cheap pre-check: bail out when there are no running LXCs. On
         # nodes without CTs, the previous version still paid ~860ms per
         # cycle for a `pvesh /cluster/resources` call that produced no
@@ -6365,6 +6387,8 @@ class HealthMonitor:
         path, so a slow / dead guest agent can't stretch the health
         cycle.
         """
+        if getattr(self, "platform_type", "pve") == "pbs":
+            return None
         try:
             import flask_server  # deferred — avoids circular import
             resources = flask_server.get_cached_pvesh_cluster_resources_vm() or []
@@ -6524,6 +6548,8 @@ class HealthMonitor:
         check (already covered) because those are usually large data
         mounts whose fill rate is driven by user content, not log spam.
         """
+        if getattr(self, "platform_type", "pve") == "pbs":
+            return None
         if not MOUNT_MONITOR_AVAILABLE:
             return None
         try:
@@ -6627,6 +6653,8 @@ class HealthMonitor:
         check via _check_filesystem. Re-alerting on the same condition
         through two channels just clutters the inbox.
         """
+        if getattr(self, "platform_type", "pve") == "pbs":
+            return None
         if not PROXMOX_STORAGE_AVAILABLE:
             return None
         try: