Whenever EDPM os-reboot role reboots the node, that disrupts workloads during adoption. We need to make this role smarter, like,
if it detects some change in the reboot-worthy parameters (osdpns ansible vars),
it might have some "safety switch" that would switch between:

• if changes detected -> reboot
• if changes detected -> fail (early, not after the point of no return) with a message

The latter point is where we need this new validations step to come into play, before

the "real deployment" happens (which is a point of no return for EDPM).

Validations should do the following:

• compare tripleo wallaby (OSP17.1) specific configuration to not cause conflicts (hostname FQDN/shortname), lost features (derived parameters), other changed cloud operational modes (stable uuids), with the configured ansible vars in osdpns(es);
• also detect config changes that require rebooting a node (like kernelags changes on its way from tripleo to rhoso);
• give end users chance to adjust osdnps and re-iterate

Unfortunatly, tools like os-diff, or tripleo validations framework unlikely to help with that.

DoD proposal after discussions:

• Implement a role in edpm-ansible to perform validation that important settings are not changing during Adoption.
  • Use this to validate service hostnames
  • Use this to validate other settings: kernel args, tuned profiles
• Add a validation step that runs the validation role into the Adoption docs and tests (a separate EDPM Deployment with just a single service for the validation). This comes before the point of no return from rollback perspective, last thing before we start adoption of the data plane nodes.
• Tweak os-reboot role so that:
  • The reboot prevention detection is smarter for Adoption nodes, avoids rebooting any nodes during Adoption.
  • The reboot can be explicitly disabled through a variable.

Acceptance criteria:

• The epdm-ansible validation role has been properly added. 
• The validation role has been added in the adoption test-suite downstream CI before EDPM Adoption.
• Verify os-reboot role will not reboot any node during adoption. [ explicit variable to control enable/disable rebooting]
• The documented procedure is verified with standalone and HA OSP env.