Bond interface becomes inconsistent after deployment failure

[Igor Braginsky]

Detailed bug description:
After deployment has failed, bond interface of random node contains no interfaces. Interfaces, that should be included to bond are listed below as usual ones.

Steps to reproduce:
1 controller + 1 compute + 1 cinder.
Make deployment to fail because of any reason (Admin network lost connectivity in my case)
Remove problem that caused failure and check status of bond on all nodes

Expected results:
Bond remains consistent

Actual result:
bond interface of random node contains no interfaces. Interfaces, that should be included to bond are listed below as usual ones.

Reproducibility:
Not every deployment and not on every node. Sometimes occurs on controller, sometimes on compute.

Description of the environment:
  Operation system:
  Versions of components: MOS 8.0
  Reference architecture:
  Network model:
  Related projects installed: Mellanox plugin

[Oleksiy Molchanov (omolchanov)]

Please attach diagnostic snapshot.

[Oleksiy Molchanov (omolchanov)]

Marking as Invalid because of no activity for more than a month.

[Noam Angel (noama)]

i had this issue also. when deploying environment successfully and the reset the environment the bond interface on every node become messed up and you cant simply redeploy you need to remove nodes from env and then add them again and reconfigure bond.

[Noam Angel (noama)]

scenarios:

1. deploy environment with each node has bond interface with active backup mode.
2. deploy successful.
3. reset env.
4. wait for node to rediscover.
5. observe nodes interface.

expected:
same interfaces appear under bond interface.

actual result:
bond interface is empty. cant remove the bond or reassigned the salve interface to the same bond.
most remove the node from env in order to workaround the issue.

[Noam Angel (noama)]

Why have you changed bug description? It is a completely different scenario from what we tested

[Dmitry Pyzhov (dpyzhov)]

Noam, please create a new bug report for your case. Moving this bug back to invalid because we didn't get diagnostic snapshot for several months. Please reopen the bug if you have required information related to original case.