[upgrade][8.0] Resetting cluster after upgrade leads to mismatching interfaces naming schema
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Fuel for OpenStack |
Confirmed
|
Medium
|
Fuel Sustaining | ||
8.0.x |
Fix Committed
|
Medium
|
Oleg S. Gelbukh | ||
Mitaka |
Won't Fix
|
Medium
|
Registry Administrators | ||
Newton |
Confirmed
|
Medium
|
Fuel Sustaining |
Bug Description
Detailed bug description:
After resetting cluster nodes will be rebooted into ubuntu bootstrap which uses new interface naming scheme. However, in nailgun node still contains ethX-named interfaces in "network_data" section
Steps to reproduce:
Deploy Kilo cluster using 7.0 Fuel
Upgrade Fuel to 8.0
Reset already deployed cluster
Wait until nodes will became online and run network check
Expected results:
Network check should pass
Actual result:
Network verification failed with network_probe.py error
Reproducibility:
Always
Workaround:
Remove nodes which were added in 7.0 fuel from nailgun db using "fuel nodes --node-id X,Y --delete-from-db --force" and add it back to cluster
Impact:
The cluster will become unmanageable by the upgraded Master node.
Description of the environment:
Fuel 7.0 Release, Fuel 8.0 release, Kilo cluster on ubuntu.
Changed in fuel: | |
status: | New → Confirmed |
description: | updated |
Changed in fuel: | |
assignee: | Oleg S. Gelbukh (gelbuhos) → Fuel Octane Dev Team (fuel-octane) |
tags: |
added: release-notes-done removed: release-notes |
Changed in fuel: | |
assignee: | Fuel Octane (fuel-octane-team) → Fuel Sustaining (fuel-sustaining-team) |
This must be added as a known issue in corresponding section of Release Notes for version 8.0.