Horizon dashboard dies periodically
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
MicroStack |
Triaged
|
High
|
Unassigned |
Bug Description
We're running MicroStack on a test system, and while it's generally working very well, the dashboard dies from time to time, and networking dies on some (but not all) instances at the same time. The instances themselves all still appear to be running, but we lose connectivity with some and not others.
Sometimes the system runs for weeks at a time before failing. Most recently, the dashboard died within 18 hours of the system being restarted.
System version is ubuntu 20.04.
Microstack version is microstack ussuri 218 latest/beta installed in devmode.
When the dashboard is dead, the traceback is displayed as follows (hostname obscured):
-------------
Environment:
Request Method: GET
Request URL: https:/
Django Version: 2.2.12
Python Version: 3.8.5
Installed Applications:
['openstack_
'openstack_
'openstack_
'openstack_
'openstack_
'django.
'django.
'django.
'django.
'django.
'django.
'django_pyscss',
'debreach',
'openstack_
'compressor',
'horizon',
'openstack_auth']
Installed Middleware:
('openstack_
'debreach.
'django.
'django.
'django.
'django.
'horizon.
'django.
'horizon.
'horizon.
'django.
'django.
'openstack_
'openstack_
Traceback:
File "/snap/
34. response = get_response(
File "/snap/
145. response = self.process_
File "/snap/
143. response = response.render()
File "/snap/
106. self.content = self.rendered_
File "/snap/
81. template = self.resolve_
File "/snap/
63. return select_
File "/snap/
42. return engine.
File "/snap/
34. return Template(
File "/snap/
143. template, origin = self.find_
File "/snap/
125. template = loader.
File "/snap/
18. for origin in self.get_
File "/snap/
149. os.path.
File "/usr/lib/
379. cwd = os.getcwd()
Exception Type: FileNotFoundError at /auth/login/
Exception Value: [Errno 2] No such file or directory
-------------
The same issue appears to have been reported by someone else over at https:/
Hi Tony,
Thank you for reporting this. I can confirm that I'm able to recreate it as well so I will triage this bug now. I just restarted the dashboard though so of course it is back to working ok as a result which makes it harder to debug. I will dig in more the next time I hit it.
Thanks,
Corey