AttributeError, dogslow in peek
Issue #18
resolved
Types: AttributeError
Value: 'NoneType' object has no attribute '__name__'
Location: dogslow/__init__.py in peek , line 196
Stacktrace (most recent call last):
File "dogslow/__init__.py", line 196, in peek
extra['culprit'] = '%s in %s' % (module.__name__, frame.f_code.co_name)
Comments (8)
-
-
repo owner Do you have more info on this? Does this happen with Sentry logging only?
-
I use gunicorn with
worker_class = 'gevent'
-
I'm seeing this as well running gunicorn with eventlet on Gondor.io.
-
I see this with the
gevent
worker class, but not withsync
. FWIW, while the documentation does say that Dogslow is intended for synchronous workers, it doesn't claim that it won't work withgevent
. -
reporter python 2.7, uwsgi, dogslow==0.9.7
Types: KeyError Value: 139657006073600 Location: dogslow/__init__.py in peek , line 194 File "dogslow/__init__.py", line 194, in peek frame = sys._current_frames()[thread_id] request <WSGIRequest at 0x110671760> started datetime.datetime(2014, 2, 13, 15, 54, 47, 886870) thread_id 139657006073600
-
This is fixed in the latest RC release (0.9.8-rc3):
diff --git a/dogslow/__init__.py b/dogslow/__init__.py index 8d82da1..8c768fb 100644 --- a/dogslow/__init__.py +++ b/dogslow/__init__.py @@ -127,8 +140,9 @@ class WatchdogMiddleware(object): # This is a bizarre construct, `module` in `function`, but # this is how all stack traces are formatted. - extra['culprit'] = '%s in %s' % (module.__name__, - frame.f_code.co_name) + extra['culprit'] = '%s in %s' % ( + getattr(module, '__name__', '(unknown module)'), + frame.f_code.co_name) # We've got to simplify the stack, because raven only accepts # a list of 2-tuples of (frame, lineno).
-
- changed status to resolved
- Log in to comment
Same throube. Dogslow failed.