[octane] Octane doesn't handle multinode (non-HA) environments in from 6.0
Bug #1660932 reported by
Roman Sokolkov
This bug affects 1 person
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Fuel for OpenStack |
Confirmed
|
High
|
Fuel Sustaining |
Bug Description
Octane's upgrade-env migrates cluster mode AS IS. For example, 6.0 had also "multinode" (non HA) mode, which is not available in later versions.
From 6.1 there is ONLY correct mode "ha_compact".
Steps to reproduce:
- Install 6.0 w/o HA
- upgrade Fuel master to 9.1
- Install octane and execute "upgrade-env"
Expected result:
Cluster will have correct mode "ha_compact"
Actual result:
Cluster has mode "multinode" and this breaks deployment
Changed in fuel: | |
assignee: | nobody → MOS Maintenance (mos-maintenance) |
assignee: | MOS Maintenance (mos-maintenance) → Fuel Sustaining (fuel-sustaining-team) |
milestone: | none → 9.x-updates |
tags: | added: area-python |
To post a comment you must log in.
This is happening because cluster-upgrade extension creates a new environment with exactly the same parameters, such as mode and net_provider, as an original one has. It seems as a lag in validation of users' data and operator possibilities. Several simple improvements can be made here:
* re-use validation of the mode parameter for for a new cluster to prevent of using incorrect modes for releases [1]
* add an additional optional parameter "mode" to provide ability to override the original one [2]
[1] https:/ /github. com/openstack/ fuel-web/ blob/master/ nailgun/ nailgun/ api/v1/ validators/ cluster. py#L122- L132 /github. com/openstack/ fuel-nailgun- extension- cluster- upgrade/ blob/master/ cluster_ upgrade/ validators. py#L32- L35
[2] https:/