-
Feature Request
-
Resolution: Unresolved
-
Major
-
None
Feature Request Overview
Provide a supported orchestration pattern that consumes the replication and promotion capability to re-instantiate workloads in a different AZ quickly. The workflow should coordinate quiescing, final sync, group promote, instance re-creation in the target AZ, volume attachment, and start. Supported entry points include Heat resources and actions, OpenStackClient commands, and guidance for Ansible or Terraform. Glance is not part of this workflow.
Business justification
- Minimize downtime during planned moves and maintenance by automating a repeatable cutover sequence.
- Reduce operator toil by standardizing the end-to-end runbook tenants execute themselves under policy.
- Improve compliance and change control with testable, auditable orchestration steps.
- Accelerate migrations from VCF by providing a familiar, deterministic move pattern.
- Align with IaC by enabling one-click or single-command stack updates across AZs.
- Increase infrastructure agility by enabling predictable capacity rebalancing without image packaging.
Functional requirements{}
- Heat resources and actions to: request quiesce, wait for replication states, trigger final sync, promote volume groups atomically, and proceed with server creation or update in the target AZ.
- Idempotent orchestration with clear preconditions (for example, detach or quiesce) and rollback guidance if any step fails.
- OpenStackClient subcommands or well-documented command sequences to execute the same flow outside Heat.
- Eventing and status reporting so pipelines can wait on milestones and surface progress to users.
- No reliance on Glance export/import. Data movement is through Cinder replication only.
- Documentation and examples for Heat, plus reference playbooks or modules for Ansible and patterns for Terraform.
Describe the customer impact
Tenants can execute a predictable, low-touch cutover from AZ A to AZ B using previously replicated volumes. Downtime is limited to the final sync and attach window. Operations teams avoid bespoke runbooks and tickets, while application teams gain a standard, auditable method to relocate workloads with consistent outcomes.
Point of contact
