Comment 1 for bug 1160442

Note: most of the interesting details are in the thread of: https://answers.launchpad.net/quantum/+question/225158

Basically, when a large number of vms are booted at once, each of which should have a single nic (and thus a single quantum port), every once and a while one of the VMs gets two quantum ports allocated. This is then visible to the tenant as two fixed IPs on the VM (since both quantum ports have a device_id of the same VM).

It may be that the allocate_for_instance logic is running twice for the same VM-id. unclear if this is happening on the same nova-compute host, or multiple. Since the ports are created back-to-back, my best guess is that its on the same nova-compute.

xinxin-shu, can you also provide the quantum-server log that covers the time period when the two ports are created? I'm trying to understand if there is an error on the quantum side that prompts nova to try to create the port again.