ethernet order changed after bootstrap => ubuntu

[Gleb]

Very strange situation.
I have PXE boot on eth2 on all nodes. After provisioning Ubuntu with Fuel one of the nodes can't discovery DHCP. Then I reboot it
and saw that interface order changed. Now it can't boot with PXE anymore. After I changed PXE setting manually it booted with PXE but now it have eth3 interface in bootstrap and I don't have any idea how to fix it.

It looks like as provisioning change interface order not in kernel but in controller also. I understand that it sounds strange but I see the fact with my own eyes. Any ideas?

It doesn't duplicate bug #1269920
I already have those patches and they doesn't help.
In my case the situation is much worse. I can't revert right interface order even after cold reboot. (it sounds crazy, I know)

{"build_id": "2014-01-30_01-17-41", "ostf_sha": "338ddf840c229918d1df8c6597588b853d02de4c", "build_number": "68", "nailgun_sha": "6f9766853968f0bba68a0487b43fe67ba02a05f7", "fuelmain_sha": "7d8768f2ac7e1e54d16c135e4ebd64722e49179e", "astute_sha": "d002c3bf626cff96a1d4aec9eb92fc4d5f4542c4", "release": "4.1", "fuellib_sha": "6cb8a54d91bd552539970b01a80ba5635876bd2c"}

[Matthew Mosesohn (raytrac3r)]

This is already fixed in
https://review.openstack.org/68041
and
https://review.openstack.org/68038
Duplicate of https://launchpad.net/bugs/1269920

[Vladimir Kuklin (vkuklin)]

Which version of FUEL are you using? Where is the diagnostic snapshot of the environment? Which SHA commits of the components are returned by sending GET request to http://master_node_ip:8000/api/version ?

[Matthew Mosesohn (raytrac3r)]

Gleb, can you tell us what kind of hardware are you running on? Or what virtualization platform?

Also, if you can reproduce this, run "ip addr show" and report the results for one node in each of the following states: bootstrap, during install (if you can manage it. screenshot is fine), first boot, and reboot and try again.

Lastly, can you explain where each interface for a given node is mapped? Like eth0=pxe, eth1=mgmt+storage+private, eth2=public+floating, eth4=pxe (2 interfaces on pxe network)

[Gleb (gleb-q)]

I've figured out that was not a bug really. It was wrong PXE configuration + wrong commutation which I didn't detect before. When I started to investigate I assumed that it was setup right and then it brokes.
Today morning I realized that there is simplier explanation.
So we can just close this bug. Sorry for false alarm.