Instance created with several IPs in mitaka

[Sergey Galkin]

Steps to reproduce

1. Build and deploy mitaka with ccp on 152 nodes with attached configs
2. Try to create new instance

Instance created with several IPs (2 times from 3 in my case) on the 2 new deployments after 'ccp cleanup'

[Sergey Galkin (sgalkin)]

Deployment one

[Sergey Galkin (sgalkin)]

Deployment two

[Sergey Galkin (sgalkin)]

fuel-ccp HEAD is

commit de44c2957f6145ffebc50479b3e005bbe515e776
Author: Andrey Pavlov <email address hidden>
Date: Tue Mar 21 22:50:25 2017 +0400

    Change rolling update strategy for deployments with aa

    Change-Id: I8ca60bbda77712d78c6b675c348c19bf075bead8

[Sergey Galkin (sgalkin)]

It seems this is a consequence of rescheduling
On example:
1. Create server with
openstack --insecure server create --flavor m1.tiny --nic net-id=ed2745e3-61fb-4fe5-a7f4-ae76404d7ff3 --image cirros test5
2. Port list for this server
(openstack) port list --server 329a6205-be97-4e20-a82b-033c143428cf
+--------------------------------------+------+-------------------+---------------------------------------------------------------------------+--------+
| ID | Name | MAC Address | Fixed IP Addresses | Status |
+--------------------------------------+------+-------------------+---------------------------------------------------------------------------+--------+
| 5b9c52af-68c5-47f1-8c3d-08694b7a868f | | fa:16:3e:ff:5f:13 | ip_address='10.50.0.40', subnet_id='5bdb31ff-6924-4dfd-af5e-90356a2e1aa8' | ACTIVE |
| 66186e03-90a2-45ff-8f06-0b7162eaca58 | | fa:16:3e:19:89:f0 | ip_address='10.50.0.38', subnet_id='5bdb31ff-6924-4dfd-af5e-90356a2e1aa8' | DOWN |
| ae3256bb-e4d7-40c6-8c1e-c6ea2bf60a82 | | fa:16:3e:8c:50:b3 | ip_address='10.50.0.37', subnet_id='5bdb31ff-6924-4dfd-af5e-90356a2e1aa8' | DOWN |
| bdad6ddc-1460-4abb-8f34-e22c5a97acf0 | | fa:16:3e:3c:dd:1e | ip_address='10.50.0.39', subnet_id='5bdb31ff-6924-4dfd-af5e-90356a2e1aa8' | DOWN |
+--------------------------------------+------+-------------------+---------------------------------------------------------------------------+--------+

Try to grep request id in all nova-computes pod

After
for i in $(kubectl -n ccp get pods | grep nova-compute | awk '{print $1}'); do echo $i; kubectl -n ccp logs $i | grep req-ce8a7ad0-de91-4678-b820-2478e890f299 && echo && echo "Found on pod $i" ; done
we have 4 reschedule
http://paste.openstack.org/show/604316/