[Logging] Truncate extras values (#4759)

javanlacerda · web-flow · commit 1e1c2b6576f7 · 2025-04-09T10:01:54.000-03:00
It truncates the extras value for logging, as it can be used to log full
exception stacktracing, blowing our log size threshold.

---------

Signed-off-by: Javan Lacerda &lt;javanlacerda@google.com&gt;
diff --git a/src/clusterfuzz/_internal/metrics/logs.py b/src/clusterfuzz/_internal/metrics/logs.py
@@ -182,7 +182,7 @@ def get_logging_config_dict(name):
 
 def truncate(msg, limit):
   """We need to truncate the message in the middle if it gets too long."""
-  if len(msg) <= limit:
+  if not isinstance(msg, str) or len(msg) <= limit:
     return msg
 
   half = limit // 2
@@ -225,7 +225,12 @@ def format(self, record: logging.LogRecord) -> str:
       entry['task_payload'] = initial_payload
 
     entry['location'] = getattr(record, 'location', {'error': True})
-    entry['extras'] = getattr(record, 'extras', {})
+    # This is needed to truncate the extras value, as it can be used
+    # to log exceptions stacktrace.
+    entry['extras'] = {
+        k: truncate(v, STACKDRIVER_LOG_MESSAGE_LIMIT)
+        for k, v in getattr(record, 'extras', {}).items()
+    }
     update_entry_with_exc(entry, record.exc_info)
 
     if not entry['extras']:
diff --git a/src/clusterfuzz/_internal/tests/core/metrics/logs_test.py b/src/clusterfuzz/_internal/tests/core/metrics/logs_test.py
@@ -190,6 +190,10 @@ def get_record(self):
         levelname='INFO',
         exc_info='exc_info',
         created=10,
+        # This extras field is needed because the call
+        # getattr(record, 'extras', {}) returns None and not the
+        # default for the case of running against a mock
+        extras={},
         location={
             'path': 'path',
             'line': 123,
@@ -208,16 +212,16 @@ def test_format_record(self):
         'message': 'log message',
         'created': '1970-01-01T00:00:10Z',
         'docker_image': '',
+        'extras': {
+            'a': 1
+        },
         'severity': 'INFO',
         'bot_name': 'linux-bot',
         'task_payload': 'fuzz fuzzer1 job1',
         'fuzz_target': 'fuzz_target1',
         'name': 'logger_name',
         'pid': 1337,
         'release': 'prod',
-        'extras': {
-            'a': 1,
-        },
         'location': {
             'path': 'path',
             'line': 123,
@@ -231,7 +235,6 @@ def test_format_record(self):
   def test_no_extras(self):
     """Test format record with no extras."""
     record = self.get_record()
-    record.extras = None
     self.assertEqual({
         'message': 'log message',
         'created': '1970-01-01T00:00:10Z',
@@ -255,7 +258,6 @@ def test_worker_bot_name(self):
     """Test format record with a worker bot name."""
     os.environ['WORKER_BOT_NAME'] = 'worker'
     record = self.get_record()
-    record.extras = None
 
     self.assertEqual({
         'docker_image': '',
