[Openstack-operators] [horizon] auth error when failing over to backup node when load balancing

Joel Cooklin joel.cooklin at gmail.com
Thu Jul 18 02:39:27 UTC 2013


When I authenticate against 01 and then force it to fail so that 02 takes
the next request (on refresh).  I get the error that 'dict' object has no
attribute 'user'.

I am running grizzly/stable and both member nodes have the same SECRET_KEY.
 While debugging I have found that the token is extracted from the session
OK in openstack_auth/backend.py(48) however it does not have an attribute
'user' as it does when authenticating against the original node (01).  The
token pulled from the session does however have a 'user' with a more
comprehensive structure in it than is expected if the 'user' attribute
exists.

ipdb> l
     24     request.session['user_id'] = user.id
     25     request.session['region_endpoint'] = user.endpoint
     26     request.session['services_region'] = user.services_region
     27
     28
---> 29 def create_user_from_token(request, token, endpoint,
services_region=None):
     30     #import ipdb; ipdb.set_trace()
     31     return User(id=token.user['id'],
     32                 token=token,
     33                 user=token.user['name'],
     34                 user_domain_id=token.user_domain_id,

ipdb> token.user
*** AttributeError: 'dict' object has no attribute 'user'

Any help or advice would be greatly appreciated.

Thanks,
Joel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-operators/attachments/20130717/4bacb70e/attachment.html>


More information about the OpenStack-operators mailing list