Repository navigation
Django Server Hangs During Startup/Shutdown Due to Blocking LogtailHandler.flush() #35
Description
Problem
Django server hangs indefinitely during startup/shutdown when LogtailHandler is enabled. When commenting out the logtail configuration, the server works fine. This issue occurs consistently but not always - depends on timing and network conditions.
Root Cause
The LogtailHandler.flush() method blocks indefinitely in an infinite loop without timeout:
# logtail/flusher.py lines 80-84
def flush(self):
self._flushing = True
while not self._clean or not self.pipe.empty():
time.sleep(self.check_interval) # Infinite loop!
self._flushing = FalseReproduction
- Add LogtailHandler to Django logging configuration
- Start Django server:
python manage.py runserver - Stop server with Ctrl+C
- Result: Server hangs and must be force-killed
Workaround: Commenting out logtail handler makes Django work normally.
Error When Force-Interrupted
Traceback (most recent call last):
File "/usr/lib/python3.11/logging/__init__.py", line 2185, in shutdown
h.flush()
File "logtail/flusher.py", line 83, in flush
time.sleep(self.check_interval)
KeyboardInterrupt:
Environment
- Python 3.11, Django 3.2.8+, logtail 0.3.3
Proposed Fix
Add a flush_timeout parameter (default 5 seconds) to prevent infinite waiting:
def flush(self):
self._flushing = True
start_time = time.time()
while not self._clean or not self.pipe.empty():
if time.time() - start_time > self.flush_timeout:
break # Exit after timeout
time.sleep(self.check_interval)
self._flushing = FalseImpact
- Prevents graceful Django shutdown
- Affects development workflow and deployment scripts
- Forces developers to kill processes manually
Activity
dbartenstein commentedon Sep 29, 2025
@vikram-typeset is your issue maybe solved with version 0.3.4? https://github.com/logtail/logtail-python/releases/tag/v0.3.4
dbartenstein commentedon Oct 13, 2025
@PetrHeinz what’s your take on this?
PetrHeinz commentedon Oct 13, 2025
Hi @dbartenstein,
Thank you for such a detailed report 🙌 We have recently added a configurable timeout in v0.3.4, which should make sure that server doesn't hang indefinitely.
Would that solve this issue as well? Would you be so kind as to upgrade from v0.3.3 🙏
The default timeout might still be a little too high for the development workflow though... What do you think?
dbartenstein commentedon Oct 13, 2025
Wow @PetrHeinz,
thanks for your swift reply! 🙇 That’s highly appreciated.
The Kudos go out to @vikram-typeset for the detailed report - I just sent a ping ;-)
We’re currently using Mezmo (previously logdna) and face this issue:
Unfortunately there has been no reaction by Mezmo. And yes, we’re currently evaluating betterstack :-)
As soon as we try out betterstack with Django, we’ll give you an update whether v0.3.3 does the job.