Wrong network is used for keystone endpoint

[Mike Scherbakov]

Fuel 6.1 (master), ISO build around ~233

keystone endpoint-list:
http://paste.openstack.org/show/200708/

 bc40e7c959be42eeb43135b971293d92 | RegionOne | http://172.16.47.11:5000/v2.0 | http://172.16.47.11:5000/v2.0 | http://172.16.47.11:35357/v2.0 | 903c65eb6fcb4abdb18eb4f369dcb695

You can see here, that both internal & admin IPs are from public network, not from internal. It has to be 192.168.0.2 here.

It was originally discovered by Alex Shaposhnikov on the scale lab, so I'm opening this bug on his behalf.

[Aleksandr Didenko (adidenko)]

Here's output from ISO "build_number": "255"

root@node-6:~# keystone endpoint-list | grep 5000
| 3cf3cc1fc1a6428690ad09820c6fbbb6 | RegionOne | http://10.109.1.2:5000/v2.0 | http://192.168.0.2:5000/v2.0 | http://192.168.0.2:35357/v2.0 | 3e1cc6ac16934f5c9a1f883de4821ac0 |

I've checked manifests, in current master we use management_vip for admin and internal addresses in endpoint configuration.

[Mike Scherbakov (mihgen)]

Thanks Alex, I've passed it to Alex S. to take a look.

[Aleksandr Shaposhnikov (alashai8)]

Thanks guys for investigation but it was our scripts fault, we changed endpoint in prepare job to be able to manipulate keystone objects from outside which isn't correct way of doing that. Since keystone/haproxy/iptables got fixed in new releases this way finaly lead as to dead end :)