Fix principal cause caps and bind measured recovery to incident policy

.github/scripts/tests/notification_recovery_fixture.py

@@ -0,0 +1,86 @@
+"""Pinned, inert recovery review. No operational module imports or threads.
+Run with python3 -S under bwrap --unshare-net; DBs and export use TMPDIR.
+"""
+import ast, contextlib, datetime, hashlib, io, json, os, pathlib, sqlite3, subprocess, sys, tarfile, tempfile, threading, types, typing
+from unittest.mock import patch
+REPO=pathlib.Path('/home/martino/projects/proxmox/notification-maintainer-followup')
+BASE=datetime.datetime(2026,9,30,20,0,0).timestamp()
+class Clock(datetime.datetime):
+    epoch=BASE
+    tick=0.0
+    @classmethod
+    def now(cls,tz=None):
+        value=cls.fromtimestamp(cls.epoch,tz)
+        cls.epoch+=cls.tick
+        return value
+TIME=types.SimpleNamespace(time=lambda:Clock.epoch)
+
+def extract(path,name,owner,ns):
+    tree=ast.parse(path.read_text())
+    nodes=tree.body if owner is None else next(n.body for n in tree.body if isinstance(n,ast.ClassDef) and n.name==owner)
+    node=next(n for n in nodes if isinstance(n,(ast.FunctionDef,ast.AsyncFunctionDef)) and n.name==name)
+    node.decorator_list=[]
+    exec(compile(ast.Module(body=[node],type_ignores=[]),str(path),'exec'),ns)
+    return ns[name]
+
+from notification_fixture import SCRIPTS as scripts
+pns=dict(vars(typing),datetime=Clock,timedelta=datetime.timedelta,json=json,sqlite3=sqlite3,contextmanager=contextlib.contextmanager,re=__import__('re'),_re_disk_base=__import__('re'))
+pns['disk_base_name']=extract(scripts/'health_persistence.py','disk_base_name',None,pns)
+methods={n:extract(scripts/'health_persistence.py',n,'HealthPersistence',pns) for n in ('_get_conn','_db_connection','_init_database','record_error','_record_error_impl','resolve_error','_resolve_error_impl','get_recovery_evidence','_record_event','_entity_from_details','clear_error','get_active_errors','is_error_active','is_error_acknowledged','_get_setting_impl','get_setting','set_setting','acknowledge_error','_acknowledge_error_impl','get_excluded_interface_names')}
+def make_store(directory):
+    s=types.SimpleNamespace(db_path=pathlib.Path(directory)/'health.sqlite',_db_lock=threading.RLock(),DEFAULT_SUPPRESSION_HOURS=24,CATEGORY_SETTING_MAP={})
+    for n,f in methods.items():
+        if n=='_entity_from_details':setattr(s,n,f)
+        elif n=='_db_connection':setattr(s,n,types.MethodType(contextlib.contextmanager(f),s))
+        else:setattr(s,n,types.MethodType(f,s))
+    s._init_database()
+    return s
+def sql(s,query,args=()):
+    with s._db_connection() as c:
+        data=c.execute(query,args).fetchall();c.commit();return data
+def record(s,key='cpu_usage',category='cpu',reason='CPU high',details=None):
+    Clock.epoch=BASE-600
+    with patch.dict(sys.modules,{'os':types.SimpleNamespace(path=types.SimpleNamespace(exists=lambda p:True))}):s.record_error(key,category,'WARNING',reason,details)
+    Clock.epoch=BASE-10
+    with patch.dict(sys.modules,{'os':types.SimpleNamespace(path=types.SimpleNamespace(exists=lambda p:True))}):s.record_error(key,category,'WARNING',reason,details)
+    Clock.epoch=BASE
+    assert s.get_active_errors()
+    return sql(s,'SELECT first_seen FROM errors WHERE error_key=?',(key,))[0][0]
+def cpu(s,current=20,history=None,warning=85,critical=95):
+    ns=dict(vars(typing),time=TIME,os=types.SimpleNamespace(cpu_count=lambda:4),health_persistence=s,psutil=types.SimpleNamespace(cpu_percent=lambda **kw:current,cpu_count=lambda:4))
+    fn=extract(scripts/'health_monitor.py','_check_cpu_with_hysteresis','HealthMonitor',ns)
+    if history is None:history=[{'value':20,'time':BASE-i*10} for i in range(1,11)]
+    target=types.SimpleNamespace(state_history={'cpu_usage':list(history)},CPU_WARNING=85,CPU_CRITICAL=95,CPU_RECOVERY=75,CPU_WARNING_DURATION=300,CPU_CRITICAL_DURATION=300,CPU_RECOVERY_DURATION=120,_check_cpu_temperature=lambda:None)
+    refresh=extract(scripts/'health_monitor.py','_refresh_thresholds','HealthMonitor',{})
+    with patch.dict(sys.modules,{'health_thresholds':types.SimpleNamespace(get=lambda section,key:({'warning':warning,'critical':critical}.get(key) if section=='cpu' else None))}):refresh(target)
+    return fn(target)
+def poll(s,first,key='cpu_usage',category='cpu',reason='CPU high',details=None,first_done=True,foreign=False,restored=False):
+    events=[]
+    meta={'category':category,'reason':reason,'severity':'WARNING','first_seen':first,'details':details}
+    c=types.SimpleNamespace(_hostname='node-a',_ENTITY_MAP={'cpu':('node',''),'pve_services':('node',''),'network':('node','')},_first_poll_done=first_done,_known_errors={key:meta},_notified_severity={key:'WARNING'},_last_notified={key:BASE-1},SAME_ERROR_COOLDOWN=86400,_get_cooldown_from_db=lambda *a:BASE-1,_queue=types.SimpleNamespace(put=events.append),_guest_storage_error_is_now_foreign=lambda *a:foreign,_save_known_errors_meta=lambda:None)
+    ns=dict(vars(typing),time=TIME,json=json,re=__import__('re'),NotificationEvent=lambda *a,**kw:types.SimpleNamespace(event_type=a[0],severity=a[1],data=a[2],**kw),startup_grace=types.SimpleNamespace(should_suppress_category=lambda *a:False))
+    c._guest_storage_error_is_now_foreign=extract(scripts/'notification_events.py','_guest_storage_error_is_now_foreign','PollingCollector',ns)
+    fn=extract(scripts/'notification_events.py','_check_persistent_health','PollingCollector',ns)
+    with patch.dict(sys.modules,{'health_persistence':types.SimpleNamespace(health_persistence=s),'flask_server':types.SimpleNamespace(get_proxmox_node_name=lambda:'node-a',get_cached_pvesh_cluster_resources_vm=lambda:[{'vmid':100,'type':'lxc','node':'node-b' if foreign else 'node-a'}]),'datetime':types.SimpleNamespace(**{**vars(datetime),'datetime':Clock})}):
+        if restored:
+            c._KNOWN_ERRORS_SETTING_KEY='pollingcollector_known_errors_v1'
+            s.set_setting(c._KNOWN_ERRORS_SETTING_KEY,json.dumps(c._known_errors));c._known_errors={}
+            extract(scripts/'notification_events.py','_load_known_errors_meta','PollingCollector',ns)(c)
+            c._first_poll_done=bool(c._known_errors)
+        fn(c)
+    return [e.data for e in events],c
+def service(store, rc=0, stdout='active\n', raised=False, services=('pvedaemon',), clustered=False):
+    calls=[]
+    def run(argv, **kw):
+        calls.append((argv, kw))
+        assert argv[:2] == ['systemctl', 'is-active']
+        if raised: raise TimeoutError('inert timeout')
+        return types.SimpleNamespace(returncode=rc, stdout=stdout)
+    ns=dict(vars(typing),time=TIME,os=types.SimpleNamespace(path=types.SimpleNamespace(exists=lambda p:clustered)),subprocess=types.SimpleNamespace(run=run),health_persistence=store)
+    result=extract(scripts/'health_monitor.py','_check_pve_services','HealthMonitor',ns)(types.SimpleNamespace(PVE_SERVICES=list(services)))
+    return result,calls
+
+@contextlib.contextmanager
+def case():
+    Clock.epoch=BASE
+    with tempfile.TemporaryDirectory(prefix='recovery-review-db-',dir=os.environ['TMPDIR']) as d:yield make_store(d)

.github/scripts/tests/test_notification_recovery_corrections.py

@@ -0,0 +1,225 @@
+"""Native initializer, measurement methods, SQL writers/readers and collector."""
+import unittest
+from notification_recovery_fixture import case, Clock, BASE, cpu, sql, poll
+
+class RecoveryCorrectionTests(unittest.TestCase):
+    def test_supported_low_warning_current_violation_is_neutral(self):
+        with case() as store:
+            Clock.epoch = BASE - 600
+            initial = cpu(store, 80, [{'value':80, 'time':Clock.epoch-i*5} for i in range(1,26)], warning=50)
+            self.assertEqual(initial['status'], 'WARNING')
+            first = sql(store, 'SELECT first_seen FROM errors')[0][0]
+            Clock.epoch = BASE
+            result = cpu(store, 60, [{'value':60, 'time':BASE-i*5} for i in range(1,26)], warning=50)
+            events, _ = poll(store, first, reason=initial['reason'])
+            # Preserve operational clear/hysteresis behavior, not its factual claim.
+            self.assertEqual(result['status'], 'OK')
+            self.assertFalse(events[0]['is_recovery'], events)
+            self.assertIsNone(store.get_recovery_evidence('cpu_usage', first))
+
+    def test_original_policy_survives_repeated_native_updates(self):
+        import json
+        for later_warning in (85, 40):
+            with self.subTest(later_warning=later_warning), case() as store:
+                Clock.epoch = BASE - 600
+                cpu(store, 80, [{'value':80, 'time':Clock.epoch-i*5} for i in range(1,26)], warning=50)
+                first = sql(store, 'SELECT first_seen FROM errors')[0][0]
+                for step in (400, 200):
+                    Clock.epoch = BASE-step
+                    cpu(store, 90, [{'value':90, 'time':Clock.epoch-i*5} for i in range(1,26)], warning=later_warning)
+                original = json.loads(sql(store, 'SELECT details FROM errors')[0][0])['cpu_policy']
+                self.assertEqual(original['warning'], 50)
+                Clock.epoch = BASE
+                cpu(store, 20, warning=later_warning)
+                self.assertIsNone(store.get_recovery_evidence('cpu_usage', first))
+                self.assertFalse(poll(store, first)[0][0]['is_recovery'])
+
+    def test_clock_rollback_new_row_generic_clear_cannot_inherit_old_proof(self):
+        for rollback, reuse_first in ((True,False),(False,False),(True,True)):
+            with self.subTest(rollback=rollback,reuse_first=reuse_first),case() as store:
+                Clock.epoch = BASE-600
+                cpu(store, 90, [{'value':90, 'time':Clock.epoch-i*5} for i in range(1,26)])
+                first = sql(store, 'SELECT first_seen FROM errors')[0][0]
+                Clock.epoch = BASE; Clock.tick = .001
+                try: cpu(store)
+                finally: Clock.tick = 0
+                self.assertTrue(store.get_recovery_evidence('cpu_usage', first))
+                store.acknowledge_error('cpu_usage', suppression_hours=-1)
+                store.clear_error('cpu_usage')
+                Clock.epoch = BASE-100 if rollback else BASE+1
+                store.record_error('cpu_usage','cpu','WARNING','new incident after clock step',{})
+                second = sql(store, 'SELECT first_seen FROM errors')[0][0]
+                self.assertNotEqual(first, second)
+                if reuse_first:
+                    # Restored malformed snapshot with reused wall-clock identity;
+                    # native row id and latest closure still prevent replay.
+                    sql(store,'UPDATE errors SET first_seen=?',(first,));second=first
+                Clock.epoch = BASE+.0005 if rollback else BASE+2
+                store.clear_error('cpu_usage'); Clock.epoch = BASE+3
+                self.assertIsNone(store.get_recovery_evidence('cpu_usage', second))
+                self.assertFalse(poll(store, second)[0][0]['is_recovery'])
+
+    def test_malformed_native_and_manual_proof_is_neutral_at_all_consumers(self):
+        import copy, json, time
+        from unittest.mock import patch
+        from notification_fixture import LANGUAGES
+        from notification_final_fixture import deliver
+        with case() as store:
+            Clock.epoch = BASE-600
+            cpu(store, 90, [{'value':90,'time':Clock.epoch-i*5} for i in range(1,26)])
+            first = sql(store,'SELECT first_seen FROM errors')[0][0]
+            Clock.epoch = BASE; cpu(store)
+            event_data = json.loads(sql(store,"SELECT data FROM events WHERE event_type='resolved'")[0][0])
+            data = poll(store,first)[0][0]
+            for field, bad in [('checked_at','bad'),('checked_at',True),('checked_at',float('inf')),
+                               ('checked_at',10**400),('value',60),('max_sample',float('nan')),
+                               ('normal_samples',True),('normal_samples',9),('checked_at',BASE+1),('checked_at',BASE-7201),
+                               ('policy',{'warning':False,'critical':95,'recovery':75}),
+                               ('policy',{'warning':96,'critical':95,'recovery':75}),
+                               ('policy',{'warning':85,'critical':95}),
+                               ('policy',{'warning':85,'critical':95,'recovery':float('nan')}),
+                               ('policy',{'warning':85,'critical':95,'recovery':75,'extra':0})]:
+                broken = copy.deepcopy(event_data)
+                broken['check_evidence'][field] = bad
+                if field == 'value': broken['check_evidence']['policy']['warning'] = 50
+                sql(store,"UPDATE events SET data=? WHERE event_type='resolved'",(json.dumps(broken),))
+                with self.subTest(field=field,bad=str(bad)),patch('health_recovery.time.time',return_value=BASE):
+                    self.assertIsNone(store.get_recovery_evidence('cpu_usage',first))
+                    for language in LANGUAGES:
+                        for manual in (False,True):
+                            result = deliver('error_resolved',{**data,'check_evidence':broken['check_evidence']},'OK',language,manual=manual)
+                            self.assertNotIn('background:#f0fdf4;', result['html'])
+            # Caller content is trusted, not authenticated native proof; even
+            # well-shaped assertions require a valid time/type/numeric contract.
+            for proof in ({'check':'cpu_usage'}, {'check':'cpu_usage','checked_at':time.time()+1}):
+                self.assertNotIn('background:#f0fdf4;', deliver('error_resolved',{**data,'check_evidence':proof},'OK',manual=True)['html'])
+
+    def test_exact_service_active_native_clear_reaches_recovery_consumers(self):
+        from unittest.mock import patch
+        from notification_recovery_fixture import service
+        from notification_fixture import LANGUAGES
+        from notification_final_fixture import deliver
+        with case() as store:
+            Clock.epoch = BASE-600
+            service(store,3,'inactive\n')
+            first = sql(store,'SELECT first_seen FROM errors')[0][0]
+            Clock.epoch = BASE
+            result, calls = service(store)
+            self.assertEqual(result['status'],'OK')
+            self.assertEqual(calls,[(['systemctl','is-active','pvedaemon'], {'capture_output':True,'text':True,'timeout':2})])
+            proof = store.get_recovery_evidence('pve_service_pvedaemon',first)
+            self.assertTrue(proof)
+            data = poll(store,first,'pve_service_pvedaemon','pve_services','PVE service pvedaemon is inactive',details={'service':'pvedaemon'})[0][0]
+            self.assertTrue(data['is_recovery'])
+            with patch('health_recovery.time.time',return_value=BASE):
+                for language in LANGUAGES:
+                    for manual in (False,True):
+                        rendered = deliver('error_resolved',data,'OK',language,manual=manual)
+                        self.assertIn('background:#f0fdf4;',rendered['html'])
+                        self.assertIn('pvedaemon', rendered['text'])
+
+    def test_cpu_positive_default_low_policy_and_neutral_history_controls(self):
+        from notification_recovery_fixture import record
+        for warning,current,history,legacy,expected in (
+            (85,20,None,False,True), (50,20,None,False,True),
+            (85,20,None,True,False), (85,99,[],False,False),
+            (85,20,[],False,False),
+            (85,20,[{'value':20,'time':BASE+i*5} for i in range(1,10)],False,False),
+            (85,20,[{'value':20,'time':BASE-121-i} for i in range(12)],False,False),
+            (85,99,None,False,False),
+            (85,float('nan'),None,False,False), (85,float('inf'),None,False,False),
+            (85,True,None,False,False), (85,10**400,None,False,False)):
+            with self.subTest(warning=warning,current=str(current),legacy=legacy), case() as store:
+                if legacy: first = record(store)
+                else:
+                    Clock.epoch=BASE-600
+                    cpu(store,90,[{'value':90,'time':Clock.epoch-i*5} for i in range(1,26)],warning=warning)
+                    first=sql(store,'SELECT first_seen FROM errors')[0][0]
+                Clock.epoch=BASE; cpu(store,current,history,warning=warning)
+                self.assertEqual(bool(store.get_recovery_evidence('cpu_usage',first)),expected)
+
+    def test_service_unavailable_removed_overall_ok_and_ack_controls(self):
+        from notification_recovery_fixture import service, record
+        for rc,stdout,raised in ((3,'inactive\n',False),(4,'unknown\n',False),(0,'active extra\n',False),(1,'active\n',False),(0,'',True)):
+            with self.subTest(rc=rc,stdout=stdout,raised=raised),case() as store:
+                Clock.epoch=BASE-600; service(store,3,'inactive\n')
+                first=sql(store,'SELECT first_seen FROM errors')[0][0]
+                Clock.epoch=BASE; service(store,rc,stdout,raised)
+                self.assertIsNone(store.get_recovery_evidence('pve_service_pvedaemon',first))
+                self.assertFalse(poll(store,first,'pve_service_pvedaemon','pve_services')[0])
+        for removed in ('pvedaemon','corosync'):
+            with self.subTest(removed=removed),case() as store:
+                first=record(store,'pve_service_'+removed,'pve_services','service inactive',{'service':removed})
+                result,calls=service(store,services=(),clustered=False)
+                self.assertEqual(result['status'],'OK'); self.assertEqual(calls,[])
+                store.clear_error('pve_service_'+removed)
+                self.assertFalse(poll(store,first,'pve_service_'+removed,'pve_services')[0][0]['is_recovery'])
+        with case() as store:
+            first=record(store,'pve_service_corosync','pve_services','corosync inactive',{'service':'corosync'})
+            result,calls=service(store,services=('pvedaemon',),clustered=False)
+            self.assertEqual(result['status'],'OK')
+            self.assertEqual([c[0][-1] for c in calls],['pvedaemon'])
+            self.assertTrue(store.is_error_active('pve_service_corosync'))
+            self.assertIsNone(store.get_recovery_evidence('pve_service_corosync',first))
+        with case() as store:
+            first=record(store,'pve_service_pvedaemon','pve_services','service inactive')
+            store.acknowledge_error('pve_service_pvedaemon',suppression_hours=-1)
+            store.clear_error('pve_service_pvedaemon',check_evidence={'check':'pve_service_pvedaemon','checked_at':BASE,'service':'pvedaemon','state':'active','returncode':0})
+            self.assertEqual(sql(store,'SELECT id FROM errors'),[])
+            self.assertIsNone(store.get_recovery_evidence('pve_service_pvedaemon',first))
+
+    def test_native_binding_latest_closure_rollbacks_and_consistent_ack_read(self):
+        import contextlib, json, sqlite3
+        for mutation in ('row_id','first_seen','closure','last_seen','latest_clear','latest_resolve','ack','event_insert_failure','ack_before_join'):
+            with self.subTest(mutation=mutation),case() as store:
+                Clock.epoch=BASE-600
+                cpu(store,90,[{'value':90,'time':Clock.epoch-i*5} for i in range(1,26)])
+                first=sql(store,'SELECT first_seen FROM errors')[0][0]
+                Clock.epoch=BASE
+                if mutation=='event_insert_failure':
+                    sql(store,"CREATE TRIGGER fail_resolve BEFORE INSERT ON events WHEN NEW.event_type='resolved' BEGIN SELECT RAISE(ABORT,'fixture'); END")
+                    self.assertEqual(cpu(store)['status'],'UNKNOWN')
+                    self.assertIsNone(sql(store,'SELECT resolved_at FROM errors')[0][0])
+                    continue
+                cpu(store); self.assertTrue(store.get_recovery_evidence('cpu_usage',first))
+                if mutation in ('row_id','first_seen','closure'):
+                    data=json.loads(sql(store,"SELECT data FROM events WHERE event_type='resolved'")[0][0])
+                    field={'row_id':'id','first_seen':'first_seen','closure':'resolved_at'}[mutation]
+                    data['incident'][field]='wrong'
+                    sql(store,"UPDATE events SET data=? WHERE event_type='resolved'",(json.dumps(data),))
+                elif mutation=='last_seen': sql(store,'UPDATE errors SET last_seen=?',(Clock.fromtimestamp(BASE+1).isoformat(),))
+                elif mutation.startswith('latest_'):
+                    sql(store,"INSERT INTO events(event_type,error_key,timestamp,data) VALUES(?,'cpu_usage',?,'{}')",('cleared' if mutation=='latest_clear' else 'resolved',Clock.now().isoformat()))
+                elif mutation=='ack': store.acknowledge_error('cpu_usage',suppression_hours=-1)
+                else:
+                    original=store._db_connection; triggered=[]
+                    class Proxy:
+                        def __init__(self,connection):self.connection=connection
+                        def __getattr__(self,name):return getattr(self.connection,name)
+                        def execute(self,query,args=()):
+                            if 'FROM errors e JOIN events' in query and not triggered:
+                                triggered.append(True)
+                                store.acknowledge_error('cpu_usage',suppression_hours=-1)
+                            return self.connection.execute(query,args)
+                    @contextlib.contextmanager
+                    def interleaved(**kw):
+                        with original(**kw) as conn: yield Proxy(conn)
+                    store._db_connection=interleaved
+                self.assertIsNone(store.get_recovery_evidence('cpu_usage',first))
+                if mutation=='ack_before_join': self.assertTrue(triggered)
+        with case() as store:
+            sql(store,"INSERT INTO errors(error_key,category,severity,reason,first_seen,last_seen) VALUES('cpu_usage','cpu','WARNING','fixture','x','x')")
+            with self.assertRaises(sqlite3.IntegrityError):
+                sql(store,"INSERT INTO errors(error_key,category,severity,reason,first_seen,last_seen) VALUES('cpu_usage','cpu','WARNING','duplicate','x','x')")
+
+    def test_malformed_history_declines_proof_without_changing_operational_clear(self):
+        with case() as store:
+            Clock.epoch=BASE-600
+            cpu(store,90,[{'value':90,'time':Clock.epoch-i*5} for i in range(1,26)])
+            first=sql(store,'SELECT first_seen FROM errors')[0][0]
+            Clock.epoch=BASE
+            result=cpu(store,20,[{'value':10**400,'time':BASE-i*5} for i in range(1,10)])
+            self.assertEqual(result['status'],'OK')
+            self.assertIsNone(store.get_recovery_evidence('cpu_usage',first))
+
+if __name__ == '__main__': unittest.main()

.github/scripts/tests/test_notification_recovery_evidence.py

@@ -1,100 +1,68 @@
-"""Fresh existing-check provenance; extracted consumers, real disposable SQLite."""
-import contextlib
-import datetime
+"""Fresh existing-check provenance; native initializer and disposable SQLite."""
 import json
-import os
-import sqlite3
-import tempfile
 import time
-import types
 import unittest
 from unittest.mock import patch
-from notification_fixture import extract, SCRIPTS, templates, LANGUAGES
+from notification_fixture import templates, LANGUAGES
 from notification_final_fixture import deliver
+from notification_recovery_fixture import case, Clock, BASE, cpu, sql, poll
+
+
+def original_cpu(store):
+    Clock.epoch = BASE-600
+    result = cpu(store, 90, [{'value':90,'time':Clock.epoch-i*5} for i in range(1,26)])
+    assert result['status'] == 'WARNING'
+    Clock.epoch = BASE
+    return sql(store, 'SELECT first_seen FROM errors')[0][0]
+
 
 class RecoveryEvidenceTests(unittest.TestCase):
     def test_cpu_success_provenance_is_persisted_only_after_normal_samples(self):
-        events=[]
-        with tempfile.TemporaryDirectory() as scratch:
-            db=scratch+'/health.sqlite'
-            conn=sqlite3.connect(db)
-            conn.execute('CREATE TABLE errors(id INTEGER PRIMARY KEY,error_key TEXT,details TEXT,resolved_at TEXT,resolution_type TEXT,resolution_reason TEXT)')
-            conn.execute("INSERT INTO errors(error_key,details) VALUES ('cpu_usage','{}')")
-            conn.commit();conn.close()
-            @contextlib.contextmanager
-            def connection():
-                c=sqlite3.connect(db)
-                try:yield c
-                finally:c.close()
-            ns={'datetime':datetime.datetime,'json':json}
-            resolve=extract(SCRIPTS/'health_persistence.py','_resolve_error_impl','HealthPersistence',ns)
-            store=types.SimpleNamespace(_db_connection=connection,_entity_from_details=lambda details:'',_record_event=lambda cursor,kind,key,data:events.append(data))
-            store.resolve_error=lambda key,reason,**kw:resolve(store,key,reason,**kw)
-            ns={'Dict':dict,'Any':object,'os':os,'time':time,'health_persistence':store,'psutil':types.SimpleNamespace(cpu_percent=lambda **kw:20,cpu_count=lambda:4)}
-            check=extract(SCRIPTS/'health_monitor.py','_check_cpu_with_hysteresis','HealthMonitor',ns)
-            target=types.SimpleNamespace(state_history={'cpu_usage':[{'value':20,'time':time.time()-i*10} for i in range(10)]},CPU_CRITICAL=95,CPU_WARNING=85,CPU_RECOVERY=75,CPU_CRITICAL_DURATION=300,CPU_WARNING_DURATION=300,CPU_RECOVERY_DURATION=120,_check_cpu_temperature=lambda:None)
-            result=check(target)
-            self.assertEqual(result['status'],'OK')
-            self.assertTrue(events[-1].get('check_evidence'),events)
-            proof=events[-1]['check_evidence']
-            self.assertEqual(proof['check'],'cpu_usage')
-            self.assertGreaterEqual(proof['checked_at'],time.time()-5)
-            # Existing generic resolve callers (cleanup/exclusion) get no proof.
-            conn=sqlite3.connect(db);conn.execute('UPDATE errors SET resolved_at=NULL');conn.commit();conn.close()
-            resolve(store,'cpu_usage','No longer present')
-            self.assertFalse(events[-1].get('check_evidence'))
+        with case() as store:
+            first = original_cpu(store)
+            self.assertEqual(cpu(store)['status'], 'OK')
+            proof = store.get_recovery_evidence('cpu_usage', first)
+            self.assertTrue(proof)
+            self.assertEqual(proof['check'], 'cpu_usage')
+            self.assertEqual(proof['checked_at'], BASE)
+            # A generic closure never gains proof; use another actual native row.
+            store.record_error('pve_service_test','pve_services','CRITICAL','inactive')
+            store.resolve_error('pve_service_test','No longer present')
+            self.assertFalse(json.loads(sql(store,"SELECT data FROM events ORDER BY id DESC LIMIT 1")[0][0]).get('check_evidence'))
 
     def test_recovery_query_requires_fresh_same_incident_proof(self):
-        with tempfile.TemporaryDirectory() as scratch:
-            db=scratch+'/health.sqlite'
-            conn=sqlite3.connect(db)
-            conn.execute('CREATE TABLE errors(id INTEGER PRIMARY KEY,error_key TEXT,first_seen TEXT,last_seen TEXT,resolved_at TEXT,acknowledged INTEGER)')
-            conn.execute('CREATE TABLE events(id INTEGER PRIMARY KEY,event_type TEXT,error_key TEXT,timestamp TEXT,data TEXT)')
-            now=datetime.datetime.now(); first=(now-datetime.timedelta(minutes=10)).isoformat(); last=(now-datetime.timedelta(minutes=1)).isoformat(); resolved=now.isoformat()
-            proof={'check':'cpu_usage','checked_at':now.timestamp()}
-            conn.execute('INSERT INTO errors VALUES(1,?,?,?,?,0)',('cpu_usage',first,last,resolved))
-            conn.execute('INSERT INTO events VALUES(1,?,?,?,?)',('resolved','cpu_usage',resolved,json.dumps({'check_evidence':proof})))
-            conn.commit();conn.close()
-            @contextlib.contextmanager
-            def connection(**kwargs):
-                c=sqlite3.connect(db)
-                try:yield c
-                finally:c.close()
-            ns={'datetime':datetime.datetime,'json':json,'time':time}
-            tree=(SCRIPTS/'health_persistence.py').read_text()
-            query=extract(SCRIPTS/'health_persistence.py','get_recovery_evidence','HealthPersistence',ns) if 'def get_recovery_evidence(' in tree else lambda *args:None
-            store=types.SimpleNamespace(_db_connection=connection)
-            self.assertEqual(query(store,'cpu_usage',first),proof)
-            self.assertIsNone(query(store,'cpu_usage','different incident'))
-            for field,value in [('acknowledged',1),('resolved_at',None),('last_seen',(now+datetime.timedelta(seconds=1)).isoformat())]:
-                conn=sqlite3.connect(db);conn.execute(f'UPDATE errors SET {field}=?',(value,));conn.commit();conn.close()
-                self.assertIsNone(query(store,'cpu_usage',first))
-                conn=sqlite3.connect(db);conn.execute('UPDATE errors SET acknowledged=0,resolved_at=?,last_seen=?',(resolved,last));conn.commit();conn.close()
-            for bad in (None,{'check':'cpu_usage','checked_at':now.timestamp()-7201},{'check':'storage_removed','checked_at':now.timestamp()},{'check':'cpu_usage','checked_at':float('inf')},{'check':'cpu_usage','checked_at':10**400}):
-                conn=sqlite3.connect(db);conn.execute('UPDATE events SET data=?',(json.dumps({'check_evidence':bad}),));conn.commit();conn.close()
-                self.assertIsNone(query(store,'cpu_usage',first))
+        with case() as store:
+            first = original_cpu(store); cpu(store)
+            proof = store.get_recovery_evidence('cpu_usage',first)
+            self.assertTrue(proof)
+            self.assertIsNone(store.get_recovery_evidence('cpu_usage','different incident'))
+            saved = sql(store,'SELECT last_seen,resolved_at FROM errors')[0]
+            for field,value in [('acknowledged',1),('resolved_at',None),('last_seen',Clock.fromtimestamp(BASE+1).isoformat())]:
+                sql(store,f'UPDATE errors SET {field}=?',(value,))
+                self.assertIsNone(store.get_recovery_evidence('cpu_usage',first))
+                sql(store,'UPDATE errors SET acknowledged=0,last_seen=?,resolved_at=?',saved)
+            data = json.loads(sql(store,"SELECT data FROM events WHERE event_type='resolved'")[0][0])
+            for bad in (None,{'check':'cpu_usage','checked_at':BASE-7201}, {'check':'storage_removed','checked_at':BASE}, {'check':'cpu_usage','checked_at':float('inf')}, {'check':'cpu_usage','checked_at':10**400}):
+                sql(store,"UPDATE events SET data=? WHERE event_type='resolved'",(json.dumps({**data,'check_evidence':bad}),))
+                self.assertIsNone(store.get_recovery_evidence('cpu_usage',first))
 
     def test_poller_and_all_consumers_distinguish_proven_recovery_from_disappearance(self):
-        import sys
-        ns={'time':time,'json':json,'Dict':dict,'NotificationEvent':lambda *a,**kw:types.SimpleNamespace(event_type=a[0],severity=a[1],data=a[2])}
-        poll=extract(SCRIPTS/'notification_events.py','_check_persistent_health','PollingCollector',ns)
-        for proof in (None,{'check':'cpu_usage','checked_at':time.time()}):
-            events=[]
-            store=types.SimpleNamespace(get_active_errors=lambda:[],is_error_acknowledged=lambda key:False,get_recovery_evidence=lambda *a:proof)
-            collector=types.SimpleNamespace(_hostname='node-a',_ENTITY_MAP={'cpu':('node','')},_first_poll_done=True,_known_errors={'cpu_usage':{'category':'cpu','reason':'CPU high','severity':'WARNING','first_seen':'2026-09-30T00:00:00'}},_notified_severity={'cpu_usage':'WARNING'},_last_notified={'cpu_usage':1},_queue=types.SimpleNamespace(put=events.append),_guest_storage_error_is_now_foreign=lambda *a:False,_save_known_errors_meta=lambda:None)
-            with patch.dict(sys.modules,{'health_persistence':types.SimpleNamespace(health_persistence=store)}):poll(collector)
-            self.assertEqual(len(events),1)
-            event=events[0]
-            self.assertEqual(event.data.get('recovery_outcome'),'resolved' if proof else 'no_longer_reported')
-            self.assertEqual(event.data['is_recovery'],bool(proof))
-            for lang in LANGUAGES:
-                for manual in (False,True):
-                    result=deliver(event.event_type,event.data,event.severity,lang,manual=manual)
-                    if proof:
-                        self.assertIn(templates.runtime_message('healthRecovery.title',lang,hostname='node-a',category='cpu',entity_suffix=''),result['title'])
-                        self.assertIn('background:#f0fdf4;',result['html'])
-                    else:self.assertNotIn('background:#f0fdf4;',result['html'])
-                    self.assertEqual(result['text'].count(event.data['reason']),1)
+        for proved in (False,True):
+            with case() as store:
+                first = original_cpu(store)
+                if proved: cpu(store)
+                else: store.resolve_error('cpu_usage','No longer present')
+                data = poll(store,first,reason='CPU high')[0][0]
+                self.assertEqual(data['is_recovery'],proved)
+                self.assertEqual(data['recovery_outcome'],'resolved' if proved else 'no_longer_reported')
+                with patch('health_recovery.time.time',return_value=BASE):
+                    for lang in LANGUAGES:
+                        for manual in (False,True):
+                            result = deliver('error_resolved',data,'OK',lang,manual=manual)
+                            self.assertEqual('background:#f0fdf4;' in result['html'],proved)
+                            if proved:
+                                self.assertIn(templates.runtime_message('healthRecovery.title',lang,hostname='node-a',category='cpu',entity_suffix=''),result['title'])
+                            self.assertEqual(result['text'].count(data['reason']),1)
 
     def test_manual_recovery_flag_alone_is_not_authoritative_evidence(self):
         data={'hostname':'node-a','category':'cpu','reason':'Observation disappeared','duration':'1h','original_severity':'WARNING','recovery_outcome':'resolved'}

.github/scripts/tests/test_notification_review_corrections.py

@@ -0,0 +1,24 @@
+"""Review regressions at actual consumer seams; no operational imports."""
+import unittest
+from notification_fixture import receive, LANGUAGES
+from notification_final_fixture import deliver
+
+class ReviewCorrectionTests(unittest.TestCase):
+    def test_subject_equivalent_late_error_is_reserved_before_cap(self):
+        cause = 'job-end hook denied'
+        # Frozen native Perl notifier/log-reader output; Rust table source-modeled.
+        raw = '\nDetails\n=======\nVMID    Name    Status    Time     Size    Filename    \n100     web     err       1m 1s    0 B     null        \n\nTotal running time: 1m 1s\nTotal size: 0 B\n\nLogs\n====\nvzdump --all 1 --storage PBS --mode snapshot\n\n100: 2026-09-29 17:00:00 ERROR: earlier diagnostic 0\n100: 2026-09-29 17:00:00 ERROR: earlier diagnostic 1\n100: 2026-09-29 17:00:00 ERROR: earlier diagnostic 2\n100: 2026-09-29 17:00:00 ERROR: earlier diagnostic 3\n100: 2026-09-29 17:00:00 ERROR: earlier diagnostic 4\n100: 2026-09-29 17:00:00 ERROR: earlier diagnostic 5\n100: 2026-09-29 17:00:00 ERROR: earlier diagnostic 6\n100: 2026-09-29 17:00:00 ERROR: earlier diagnostic 7\n100: 2026-09-29 17:00:00 ERROR: earlier diagnostic 8\n100: 2026-09-29 17:00:00 ERROR: earlier diagnostic 9\n100: 2026-09-29 17:00:00 ERROR: earlier diagnostic 10\n100: 2026-09-29 17:00:00 ERROR: earlier diagnostic 11\n100: 2026-09-29 17:00:00 ERROR: job-end hook denied\n\n\n'
+        event = receive(raw, 'error', 'vzdump backup status (raw-host): backup failed: ' + cause)
+        for language in LANGUAGES:
+            for manual in (False, True):
+                with self.subTest(language=language, manual=manual):
+                    result = deliver(event.event_type, {**event.data, 'hostname':'alias {rack.location}'}, event.severity, language, manual=manual)
+                    self.assertEqual(result['text'].count(cause), 1)
+                    self.assertNotIn('raw-host', result['text'])
+                    diagnostics = [line for line in result['body'].splitlines() if 'ERROR:' in line]
+                    self.assertLessEqual(len(diagnostics), 8)
+                    self.assertLessEqual(len('\n'.join(diagnostics)), 1024)
+                    self.assertTrue(all(len(line) <= 512 for line in diagnostics))
+                    self.assertEqual(result['data']['pve_message'], raw)
+
+if __name__ == '__main__': unittest.main()

AppImage/scripts/health_monitor.py

@@ -1427,6 +1427,8 @@ class HealthMonitor:
                         'details': f'Sustained for {actual_duration}s above {self.CPU_CRITICAL}%.',
                         'cpu_percent': cpu_percent,
                         'duration': actual_duration,
+                        'cpu_policy': {'warning': self.CPU_WARNING, 'critical': self.CPU_CRITICAL,
+                                       'recovery': self.CPU_RECOVERY},
                     },
                 )
             elif len(warning_samples) >= WARNING_MIN_SAMPLES and len(recovery_samples) < RECOVERY_MIN_SAMPLES:
@@ -1446,6 +1448,8 @@ class HealthMonitor:
                         'details': f'Sustained for {actual_duration}s above {self.CPU_WARNING}%.',
                         'cpu_percent': cpu_percent,
                         'duration': actual_duration,
+                        'cpu_policy': {'warning': self.CPU_WARNING, 'critical': self.CPU_CRITICAL,
+                                       'recovery': self.CPU_RECOVERY},
                     },
                 )
             else:
@@ -1453,8 +1457,21 @@ class HealthMonitor:
                 reason = None
                 # CPU is normal - auto-resolve any existing CPU errors
                 evidence = None
-                if cpu_percent < self.CPU_RECOVERY and len(recovery_samples) >= RECOVERY_MIN_SAMPLES:
-                    evidence = {'check': 'cpu_usage', 'checked_at': current_time}
+                # Presentation proof is stricter than operational hysteresis:
+                # a supported warning can be below the fixed recovery cutoff.
+                from health_recovery import _finite_number
+                criterion = min(self.CPU_WARNING, self.CPU_RECOVERY)
+                normal_samples = [entry for entry in self.state_history[state_key]
+                    if _finite_number(entry['value']) and 0 <= entry['value'] < criterion
+                    and _finite_number(entry['time'])
+                    and 0 <= current_time - entry['time'] <= self.CPU_RECOVERY_DURATION]
+                if (_finite_number(cpu_percent) and 0 <= cpu_percent < criterion
+                        and len(normal_samples) >= RECOVERY_MIN_SAMPLES):
+                    evidence = {'check': 'cpu_usage', 'checked_at': current_time,
+                                'value': cpu_percent, 'normal_samples': len(normal_samples),
+                                'max_sample': max(entry['value'] for entry in normal_samples),
+                                'policy': {'warning': self.CPU_WARNING, 'critical': self.CPU_CRITICAL,
+                                           'recovery': self.CPU_RECOVERY}}
                 health_persistence.resolve_error('cpu_usage', 'CPU usage returned to normal',
                                                  check_evidence=evidence)
             
@@ -3904,6 +3921,7 @@ class HealthMonitor:
             
             failed_services = []
             service_details = {}
+            active_evidence = {}
 
             for service in services_to_check:
                 try:
@@ -3918,6 +3936,10 @@ class HealthMonitor:
                     if result.returncode != 0 or status != 'active':
                         failed_services.append(service)
                         service_details[service] = status or 'inactive'
+                    else:
+                        active_evidence[service] = {'check': f'pve_service_{service}',
+                            'checked_at': time.time(), 'service': service, 'state': status,
+                            'returncode': result.returncode}
                 except Exception:
                     failed_services.append(service)
                     service_details[service] = 'error'
@@ -3932,7 +3954,7 @@ class HealthMonitor:
                 if svc not in failed_services:
                     error_key = f'pve_service_{svc}'
                     if health_persistence.is_error_active(error_key):
-                        health_persistence.clear_error(error_key)
+                        health_persistence.clear_error(error_key, check_evidence=active_evidence.get(svc))
             
             # Build checks dict with status per service
             checks = {}

AppImage/scripts/health_persistence.py

@@ -600,7 +600,7 @@ class HealthPersistence:
 
             cursor.execute('''
                 SELECT id, acknowledged, resolved_at, category, severity, first_seen,
-                       notification_sent, suppression_hours, acknowledged_at
+                       notification_sent, suppression_hours, acknowledged_at, details
                 FROM errors WHERE error_key = ?
             ''', (error_key,))
             existing = cursor.fetchone()
@@ -609,7 +609,7 @@ class HealthPersistence:
 
             if existing:
                 (err_id, ack, resolved_at, old_cat, old_severity, first_seen,
-                 notif_sent, stored_suppression, acknowledged_at) = existing
+                 notif_sent, stored_suppression, acknowledged_at, old_details_json) = existing
 
                 if ack == 1:
                     # SAFETY OVERRIDE: Critical CPU temperature ALWAYS re-triggers
@@ -680,6 +680,18 @@ class HealthPersistence:
                     conn.commit()
                     return event_info
 
+                # Original CPU policy is immutable for this row's incident.
+                # Never upgrade a legacy row from later/current settings.
+                if error_key == 'cpu_usage':
+                    try:
+                        old_details = json.loads(old_details_json or '{}')
+                    except (ValueError, TypeError):
+                        old_details = {}
+                    details = dict(details) if isinstance(details, dict) else {}
+                    details.pop('cpu_policy', None)
+                    if isinstance(old_details, dict) and 'cpu_policy' in old_details:
+                        details['cpu_policy'] = old_details['cpu_policy']
+                    details_json = json.dumps(details)
                 # Not acknowledged - update existing active error
                 cursor.execute('''
                     UPDATE errors
@@ -776,7 +788,7 @@ class HealthPersistence:
                 # was created — otherwise "Storage 'Tuxis' unavailable"
                 # comes back as "Resolved - Storage" with no identity.
                 cursor.execute(
-                    'SELECT details FROM errors WHERE error_key = ? ORDER BY id DESC LIMIT 1',
+                    'SELECT details, id, first_seen, resolved_at FROM errors WHERE error_key = ? ORDER BY id DESC LIMIT 1',
                     (error_key,),
                 )
                 row = cursor.fetchone()
@@ -786,11 +798,18 @@ class HealthPersistence:
                         stored_details = json.loads(row[0])
                     except Exception:
                         stored_details = None
+                # Legacy/mismatched original policy cannot certify normality.
+                if error_key == 'cpu_usage' and (not isinstance(stored_details, dict)
+                        or not isinstance(check_evidence, dict)
+                        or check_evidence.get('policy') != stored_details.get('cpu_policy')
+                        or not stored_details.get('cpu_policy')):
+                    check_evidence = None
                 self._record_event(cursor, 'resolved', error_key, {
                     'reason': reason,
                     # Only explicit current-check callers attach this proof.
                     # Generic resolve/cleanup remains neutral.
                     'check_evidence': check_evidence,
+                    'incident': {'id': row[1], 'first_seen': row[2], 'resolved_at': row[3]},
                     'entity': self._entity_from_details(stored_details),
                     'details': stored_details or {},
                 })
@@ -800,41 +819,40 @@ class HealthPersistence:
     def get_recovery_evidence(self, error_key: str, first_seen: str):
         """Return fresh same-incident native check proof, never absence of errors.
 
-        Initially only the host CPU check has a stable condition identity. Other
+        Host CPU and exact per-service active checks carry provenance. Other
         checks, generic clears, excluded/deleted records and legacy events stay
         neutral until they have equivalent per-condition provenance.
         """
-        if error_key != 'cpu_usage' or not first_seen:
+        if not isinstance(error_key, str) or not first_seen or not (
+                error_key == 'cpu_usage' or error_key.startswith('pve_service_')):
             return None
         try:
-            with self._db_connection() as conn:
+            # One SQLite statement is one consistent row/ack/closure snapshot.
+            # Latest closure by event id, never search past a generic clear.
+            with self._db_lock, self._db_connection() as conn:
                 row = conn.execute('''
-                    SELECT first_seen, last_seen, resolved_at, acknowledged
-                    FROM errors WHERE error_key = ? ORDER BY id DESC LIMIT 1
+                    SELECT e.first_seen, e.last_seen, e.resolved_at, e.acknowledged,
+                           e.id, v.timestamp, v.data
+                    FROM errors e JOIN events v ON v.id = (
+                        SELECT id FROM events WHERE error_key = e.error_key
+                        AND event_type IN ('resolved', 'cleared') ORDER BY id DESC LIMIT 1
+                    ) WHERE e.error_key = ?
                 ''', (error_key,)).fetchone()
-                if not row or row[0] != first_seen or not row[2] or row[3]:
-                    return None
-                event = conn.execute('''
-                    SELECT timestamp, data FROM events
-                    WHERE error_key = ? AND event_type = 'resolved'
-                    ORDER BY id DESC LIMIT 1
-                ''', (error_key,)).fetchone()
-            if not event:
+            if not row or row[0] != first_seen or not row[2] or row[3]:
                 return None
-            proof = json.loads(event[1]).get('check_evidence')
-            if not isinstance(proof, dict) or proof.get('check') != error_key:
+            event_data = json.loads(row[6])
+            if event_data.get('incident') != {'id': row[4], 'first_seen': row[0], 'resolved_at': row[2]}:
                 return None
-            checked = proof.get('checked_at')
-            if not isinstance(checked, (int, float)) or isinstance(checked, bool):
+            proof = event_data.get('check_evidence')
+            from health_recovery import valid_check_evidence
+            if not valid_check_evidence(error_key, proof, now=datetime.now().timestamp()):
                 return None
-            checked = float(checked)
-            # Reuse the collector's existing two-hour freshness boundary.
-            now = datetime.now().timestamp()
-            if not 0 <= now - checked <= 7200:
+            if error_key == 'cpu_usage' and proof.get('policy') != event_data.get('details', {}).get('cpu_policy'):
                 return None
+            checked = float(proof['checked_at'])
             last_seen = datetime.fromisoformat(row[1]).timestamp()
             resolved = datetime.fromisoformat(row[2]).timestamp()
-            recorded = datetime.fromisoformat(event[0]).timestamp()
+            recorded = datetime.fromisoformat(row[5]).timestamp()
             if not last_seen <= checked <= resolved <= recorded:
                 return None
             return proof
@@ -906,7 +924,7 @@ class HealthPersistence:
         
         return False
     
-    def clear_error(self, error_key: str):
+    def clear_error(self, error_key: str, *, check_evidence=None):
         """
         Remove/resolve a specific error immediately.
         Used when the condition that caused the error no longer exists
@@ -928,7 +946,7 @@ class HealthPersistence:
 
             # Check if this error was acknowledged (dismissed)
             cursor.execute('''
-                SELECT acknowledged FROM errors WHERE error_key = ?
+                SELECT acknowledged, id, first_seen FROM errors WHERE error_key = ?
             ''', (error_key,))
             row = cursor.fetchone()
 
@@ -947,7 +965,10 @@ class HealthPersistence:
                 ''', (now, error_key))
 
                 if cursor.rowcount > 0:
-                    self._record_event(cursor, 'cleared', error_key, {'reason': 'condition_resolved'})
+                    self._record_event(cursor, 'cleared', error_key, {
+                        'reason': 'condition_resolved', 'check_evidence': check_evidence,
+                        'incident': {'id': row[1], 'first_seen': row[2], 'resolved_at': now},
+                    })
 
             conn.commit()
     

AppImage/scripts/health_recovery.py

@@ -0,0 +1,51 @@
+"""Bounded recovery metadata admission, without importing monitor singletons.
+
+Native persistence additionally binds the exact incident and closure. Manual
+notifications remain authenticated caller assertions: shape validation cannot
+establish that an asserted measurement actually happened.
+"""
+import math
+import time
+from typing import TypeGuard
+
+
+def _finite_number(value) -> TypeGuard[int | float]:
+    try:
+        return isinstance(value, (int, float)) and not isinstance(value, bool) and math.isfinite(value)
+    except (OverflowError, ValueError, TypeError):
+        return False
+
+
+def valid_check_evidence(error_key, proof, *, now=None):
+    """Validate a supported measurement contract, not its external authenticity."""
+    if not isinstance(proof, dict) or proof.get('check') != error_key:
+        return False
+    checked = proof.get('checked_at')
+    now = time.time() if now is None else now
+    if not _finite_number(checked) or not _finite_number(now) or not 0 <= now-checked <= 7200:
+        return False
+    if error_key == 'cpu_usage':
+        policy = proof.get('policy')
+        if not isinstance(policy, dict) or set(policy) != {'warning', 'critical', 'recovery'}:
+            return False
+        if not all(_finite_number(v) and 1 <= v <= 100 for v in policy.values()):
+            return False
+        if policy['warning'] > policy['critical']:
+            return False
+        value, maximum, count = proof.get('value'), proof.get('max_sample'), proof.get('normal_samples')
+        return (_finite_number(value) and _finite_number(maximum)
+                and 0 <= value <= maximum < min(policy['warning'], policy['recovery'])
+                and isinstance(count, int) and not isinstance(count, bool) and count >= 10)
+    if isinstance(error_key, str) and error_key.startswith('pve_service_'):
+        service = error_key[len('pve_service_'):]
+        return (bool(service) and proof.get('service') == service
+                and proof.get('state') == 'active'
+                and type(proof.get('returncode')) is int and proof['returncode'] == 0)
+    return False
+
+
+def presents_recovery(data):
+    """One presentation predicate shared by template, icon and email badge."""
+    return (isinstance(data, dict) and data.get('recovery_outcome') == 'resolved'
+            and data.get('is_recovery') is True
+            and valid_check_evidence(data.get('error_key'), data.get('check_evidence')))

AppImage/scripts/notification_channels.py

@@ -1038,10 +1038,8 @@ class EmailChannel(NotificationChannel):
         # Determine group for section header
         event_type = data.get('_event_type', '')
         if event_type == 'error_resolved':
-            if (data.get('recovery_outcome') == 'resolved'
-                    and data.get('is_recovery') is True
-                    and isinstance(data.get('check_evidence'), dict)
-                    and data['check_evidence'].get('check') == 'cpu_usage'):
+            from health_recovery import presents_recovery
+            if presents_recovery(data):
                 sev.update(self._SEV_STYLE['OK'])
                 sev['label'] = _runtime_notification_text('healthRecovery.status', data)
             else:
@@ -1070,8 +1068,11 @@ class EmailChannel(NotificationChannel):
                      or backup_email or data.get('_restore_summary') or data.get('_backup_summary'))
         temp_cell_wrap = 'word-wrap:break-word;overflow-wrap:break-word;word-break:break-word;' if wrap_body else ''
         temp_table_layout = 'table-layout:fixed;' if wrap_body else ''
-        backup_title_wrap = temp_cell_wrap if backup_email else ''
-        backup_metadata_layout = 'table-layout:fixed;' if backup_email else ''
+        # Recovery exposes the same literal host context as backup notices.
+        # Keep wrapping event-scoped; unrelated mail remains byte-identical.
+        context_email = backup_email or event_type == 'error_resolved'
+        backup_title_wrap = temp_cell_wrap if context_email else ''
+        backup_metadata_layout = 'table-layout:fixed;' if context_email else ''
         section_label = _runtime_text(f'email.groups.{group}', data)
         report_label = _runtime_text('email.report', data, group=section_label)
         host_label = _runtime_text('email.host', data)

AppImage/scripts/notification_templates.py

@@ -2070,10 +2070,8 @@ def render_template(event_type: str, data: Dict[str, Any],
             return ''
 
     safe_vars = _SafeDict(variables)
-    if (event_type == 'error_resolved' and data.get('recovery_outcome') == 'resolved'
-            and data.get('is_recovery') is True
-            and isinstance(data.get('check_evidence'), dict)
-            and data['check_evidence'].get('check') == 'cpu_usage'):
+    from health_recovery import presents_recovery
+    if event_type == 'error_resolved' and presents_recovery(data):
         safe_vars['_health_title'] = runtime_message('healthRecovery.title', language, **variables)
         safe_vars['_health_body'] = runtime_message('healthRecovery.body', language, **variables)
         template['title'] = '{_health_title}'
@@ -2089,13 +2087,14 @@ def render_template(event_type: str, data: Dict[str, Any],
     # parse the table/logs and format a rich body instead of the sparse template.
     pve_message = data.get('pve_message', '')
     backup_diagnostics = []
+    principal_cause = None
 
-    def bounded_backup_diagnostics(lines):
+    def bounded_backup_diagnostics(lines, principal_cause=None):
         # 1024 chars matches the repository's small-channel message convention;
         # 8 lines keeps repeated producer warnings readable. Inventory/title
         # size is separate: this is not a one-Telegram-message guarantee.
         unique = list(dict.fromkeys(line for line in lines if line.strip()))
-        principal = next((line for line in unique if re.search(r'\b(?:ERROR:|TASK ERROR:)', line, re.IGNORECASE)), None)
+        principal = principal_cause or next((line for line in unique if re.search(r'\b(?:ERROR:|TASK ERROR:)', line, re.IGNORECASE)), None)
         if principal:
             unique.remove(principal)
             unique.insert(0, principal)
@@ -2189,16 +2188,18 @@ def render_template(event_type: str, data: Dict[str, Any],
             source_subject = cause.group(1).strip() if cause else ''
             if source_subject.lower() == 'multiple problems':
                 source_subject = ''
-        cause_in_diagnostics = any(
-            line.strip() == source_subject or
-            re.split(r'\b(?:TASK ERROR:|ERROR:)\s*', line, maxsplit=1, flags=re.IGNORECASE)[-1].strip() == source_subject
-            for line in backup_diagnostics)
-        if source_subject and source_subject not in body_text and not cause_in_diagnostics:
-            # A unique job/setup cause must survive a warning-heavy report.
-            backup_diagnostics.insert(0, source_subject)
+        if source_subject and source_subject not in body_text:
+            # Reserve the subject-equivalent diagnostic BEFORE the cap. Finding
+            # it in uncapped logs is not enough: that late line could be omitted.
+            principal_cause = next((line for line in backup_diagnostics
+                if line.strip() == source_subject or
+                re.split(r'\b(?:TASK ERROR:|ERROR:)\s*', line, maxsplit=1, flags=re.IGNORECASE)[-1].strip() == source_subject), None)
+            if not principal_cause:
+                principal_cause = source_subject
+                backup_diagnostics.insert(0, source_subject)
 
     if backup_diagnostics:
-        body_text += '\n' + bounded_backup_diagnostics(backup_diagnostics)
+        body_text += '\n' + bounded_backup_diagnostics(backup_diagnostics, principal_cause)
 
     # Clean up: collapse runs of 3+ blank lines into 1, remove trailing whitespace
     import re as _re
@@ -2531,10 +2532,8 @@ def enrich_with_emojis(event_type: str, title: str, body: str,
     severity = data.get('severity', 'INFO')
     
     icon = EVENT_EMOJI.get(event_type) or CATEGORY_EMOJI.get(group) or SEVERITY_ICONS.get(severity, '')
-    if (event_type == 'error_resolved' and data.get('recovery_outcome') == 'resolved'
-            and data.get('is_recovery') is True
-            and isinstance(data.get('check_evidence'), dict)
-            and data['check_evidence'].get('check') == 'cpu_usage'):
+    from health_recovery import presents_recovery
+    if event_type == 'error_resolved' and presents_recovery(data):
         icon = '✅'
     if event_type == 'backup_complete':
         icon = {