Goal:

Implement the PodRemediator controller in the infra-operator to remediate workloads using local PVCs when worker nodes become unhealthy. The controller must integrate with Node Health Check (NHC) and Self Node Remediation (SNR), expose a PodRemediator CRD, and run in the openstack-operators namespace. The goal is to improve resilience for stateful workloads (e.g. Galera, RabbitMQ) without requiring changes inside application operators.

Acceptance Criteria:

• PodRemediator controller deployment runs in the openstack-operators namespace, aligned with other infra-operator controllers.
• RBAC is in place, granting the controller the required permissions
• When NHC and SNR are installed and the supplemental RBAC is applied, the PodRemediator CR reaches status.conditions[type=Ready].status=True.
• A from-scratch POC workflow exists and is automated:
  • build/use a custom infra-operator image that includes PodRemediator
  • apply CRD and RBAC, deploy PodRemediator CR
  • install and configure NHC/SNR
  • run a smoke test that validates PodRemediator is Ready and no remediation is triggered while nodes are healthy

• An E2E remediation scenario is automated in the repo (e.g. Ansible playbook + helper script):
  • create a stateful workload with a local PVC
  • simulate node failure (e.g. virsh destroy of the worker VM) so the node becomes NotReady
  • verify that PodRemediator deletes the PVC bound to that node
  • verify that the workload is recreated on another node with a new PVC