Hi Nova, While doing some test with my newly deployed devstack env today, it turns out that the default devstack deployment cannot cleanup networks after the retry attempt exceeded. This is because in the deployment with super-conductor and cell-conductor, the retry and cleanup logic is in cell-conductor [1], and by default the devstack didn't put Neutron endpoint info in nova_cell1.conf. And as the neutron endpoint is also not included in the context [2], so we can't find Neutron endpoint when try to cleanup network [3]. The solution is simple though, ether add Neutron endpoint info in nova_cell1.conf in devstack or change Nova code to support get auth from context. I think the latter one is better as in real deployment there could be many cells and by doing this can ignore config it all the time. Any particular consideration that Neutron is not included in [2]? Suggestions on how this should be fixed? I also registered a devstack bug to fix it in devstack [4]. [1] https://github.com/openstack/nova/blob/bccf26c93a973d000e4339843ce9256814286d10/nova/conductor/manager.py#L604 [2] https://github.com/openstack/nova/blob/9519601401ee116a9197fe3b5d571495a96912e9/nova/context.py#L121 [3] https://bugs.launchpad.net/nova/+bug/1747600 [4] https://bugs.launchpad.net/devstack/+bug/1747598 BR, Kevin Zheng -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20180206/b87d3c60/attachment.html>