fix(reliability): journal fail-safe serialization, per-line parse, review hardening

payload/user/remote_access/pager-webui/mk8_events.py

@@ -1,5 +1,5 @@
 """Mark VIII reliability event journal. JSONL on /mmc, rotated."""
-import json, os, threading
+import json, os, threading, time
 
 MK8_DIR = '/mmc/mk8'
 EVENTS_PATH = os.path.join(MK8_DIR, 'events.log')
@@ -7,11 +7,10 @@ MAX_BYTES = 5 * 1024 * 1024
 KEEP = 4
 _LOCK = threading.Lock()
 
-COUNTER_KINDS = ('boot', 'unexpected_boot', 'rollback', 'restart',
-                 'guard_fix')
 _COUNTER_KEYS = {'boot': 'boots', 'unexpected_boot': 'unexpected_boots',
                  'rollback': 'rollbacks', 'restart': 'restarts',
                  'guard_fix': 'guard_fixes'}
+COUNTER_KINDS = tuple(_COUNTER_KEYS)
 
 
 def _ensure_dir():
@@ -22,11 +21,23 @@ def _ensure_dir():
 
 
 def log_event(kind, sev='info', msg='', meta=None):
-    entry = {'ts': int(__import__('time').time()), 'kind': str(kind),
-             'sev': sev, 'msg': msg[:500]}
-    if meta:
+    """Append one journal entry. Never raises: a reliability journal that
+    can crash its caller would defeat its purpose. Single-writer per
+    process is assumed; there is no inter-process lock."""
+    try:
+        entry = {'ts': int(time.time()), 'kind': str(kind), 'sev': sev,
+                 'msg': str(msg)[:500]}
+        if meta is not None:
+            json.dumps(meta)
             entry['meta'] = meta
         line = json.dumps(entry) + '\n'
+    except Exception:
+        try:
+            line = json.dumps({'ts': int(time.time()), 'kind': str(kind),
+                               'sev': sev, 'msg': str(msg)[:500],
+                               'meta_repr': repr(meta)[:500]}) + '\n'
+        except Exception:
+            return
     with _LOCK:
         _ensure_dir()
         try:
@@ -52,8 +63,16 @@ def read_events(limit=100):
     for path in paths:
         try:
             with open(path) as f:
-                out.extend(json.loads(l) for l in f if l.strip())
-        except (OSError, ValueError):
+                for l in f:
+                    if not l.strip():
+                        continue
+                    try:
+                        row = json.loads(l)
+                    except ValueError:
+                        continue
+                    if isinstance(row, dict):
+                        out.append(row)
+        except OSError:
             continue
     out.sort(key=lambda r: r.get('ts', 0))
     return out[-limit:][::-1]

tests/test_mk8_events.py

@@ -11,6 +11,8 @@ class EventsTest(unittest.TestCase):
 
     def tearDown(self):
         mk8_events.MK8_DIR = self.old
+        mk8_events.EVENTS_PATH = os.path.join(self.old, 'events.log')
+        mk8_events.MAX_BYTES = 5 * 1024 * 1024
 
     def test_log_and_read_newest_first(self):
         mk8_events.log_event('boot', msg='first')