We need a test scenario to verify that downgrading is blocked.

We can build a "fake 4.15" commit by building RPMs:

MICROSHIFT_VERSION=4.15.0-0 make rpms 

Flow:

1. Deploy/install system with fake 4.15 MicroShift
2. Wait for system to be healthy (wait for greenboot to end and be green)
   1. /var/lib/microshift/version should contain 4.15.0
3. Rebase and reboot system to "latest MicroShift from source" (same like 4.15.0 but without overriding the version)
4. System should reboot 3 times attempting to get healthy
5. 4th reboot should be to the original "fake 4.15.0" deployment (roll back)
6. System should be healthy

Things we can check additionally:

• microshift logs should contain information about invalid "upgrade"
• there should be a backup for "4.15.0" deployment (created on first boot of 4.14 deployment)

FYI: version file should never contain 4.14