Possible flow is described in the enhancement: https://github.com/openshift/enhancements/blob/master/enhancements/microshift/microshift-updateability-ostree.md#previous-commit-was-booted-only-once-and-and-new-commit-fails-to-back-up-the-data

In other words: new deployment fails to backup the data of previous deployment, then, after 'greenboot reboot', it should attempt to back up the data again (even though "previous boot was unhealthy"). If it fails and system rolls back, "original deployment" should make an attempt to backup the data again.

In yet another words: if the system was healthy, data should be backed up and that is the most important thing - we don't want to delete it, restore other backup on top of that, etc. Data needs to be backed up.