[system-tests] 6.1 upgrade jobs a skipped due to changing simple cases on ha

[Andrey Sledzinskiy]

Our 6.1 upgrade jobs are skipped because we deploy 6.0 ISO with 6.0 fuel-main but we use test group 'ceph_ha_one_controller_compact' that is present only in 6.1, in 6.0 it's 'ceph_multinode_compact'
I can see here 3 ways of handling it:
1. Deploy both 6.0 and 6.1 iso with 6.1 code from fuel-main (devops responsibility) - main drawback is that 6.1 fuel-main code can become incompatible with 6.0 ISO
2. Add 'ceph_ha_one_controller_compact' test to 6.0 for upgrades only (fuel-qa responsibility)
3. Backport this patch - https://review.openstack.org/#/c/150791/ (fuel-qa responsibility). The main drawback - we support simple mode in 6.0 and probably don't want to remove simple tests in 6.0

[Artem Panchenko (apanchenko-8)]

In my opinion the second was is most preferred, because it has only one slight disadvantage - increasing code amount in system tests.

[Andrey Sledzinskiy (asledzinskiy)]

Nastya, please comment too. In case of second option issue should be assigned to fuel-qa

[Dennis Dmitriev (ddmitriev)]

If we completely disable testing of upgrades from 6.0 to 6.1 for simple clusters, how we should provide the possibility to upgrade simple environments to 6.1 ?

Will it be skipped or disabled in that way?

[Nastya Urlapova (aurlapova)]

After discussion with devs I suggest to verify the next case:
1. deploy 6.0 simple env
2. upgrade master to 6.1
3. try to add/delete nodes in OS env - for verifying how our nailgun will be manipulate with 6.0 env

[Dennis Dmitriev (ddmitriev)]

Use the old snapshot name 'ceph_multinode_compact' for tests in 'upgrade' group, because fuel-main/6.0 makes snapshot with this name for upgrade jobs.

https://review.openstack.org/#/c/153584/