When deploying nova with spice, sometime Spice port binding is failing

[Kevin Lefevre]

When using SPICE instead of NOVNC for console. Sometimes nova failed to boot the instances.

That's why instances ended up with two IPs because nova tries to boot up the instance on one host, it fails and then it tries another.

Failure are du to an error with spice:

```
2017-05-23 14:38:00.490 86286 ERROR nova.compute.manager [instance: 155d90d1-755b-4200-b4ad-abe7e103d393] libvirtError: internal error: process exited while connecting to monitor: ((null):107797): Spice-Warning **: reds.c:2463:reds_init_socket: reds_init_socket: binding socket to 10.16.17.102:5907 failed
2017-05-23 14:38:00.490 86286 ERROR nova.compute.manager [instance: 155d90d1-755b-4200-b4ad-abe7e103d393] 2017-05-23T14:37:59.998191Z qemu-system-x86_64: failed to initialize spice server
2017-05-23 14:38:00.490 86286 ERROR nova.compute.manager [instance: 155d90d1-755b-4200-b4ad-abe7e103d393]
2017-05-23 14:38:00.492 86286 INFO nova.compute.manager [req-68858e22-60c2-4224-ba2a-990c9a0bc6bf 405d7ea45c0243d8bf84478f2cbfca18 f3ab786dd69043659850c021d976516c - - -] [instance: 155d90d1-755b-4200-b4ad-abe7e103d393] Terminating instance
2017-05-23 14:38:00.505 86286 INFO nova.virt.libvirt.driver [-] [instance: 155d90d1-755b-4200-b4ad-abe7e103d393] Instance destroyed successfully.
```

[Jean-Philippe Evrard (jean-philippe-evrard)]

That is not enough for us to debug what's going on in your environment.

It doesn't happen in our gates, which are using spice, so I suspect this is not a bug, but a configuration issue.

Could you tell us more about your machine 10.16.17.102 ? What is that? How is your osa configured?
Which branch are you running?

[Jean-Philippe Evrard (jean-philippe-evrard)]

According to the bug reporter "i'm not sure it is related to osa".