Single-VM OpenStack Sandbox: Cinder NFS Storage Reboot Verification Completed

Successfully verified Cinder NFS block storage connectivity and data integrity after reboot in a single-VM OpenStack sandbox
Confirmed that the cause of the guest VM's DHCP non-response is a bug in the upstream openstack-helm chart involving a missing dhcp-agent template hostPath
Encountered image re-release issues due to Quay registry daily builds, causing db-init Job CrashLoop due to missing pymysql
Applied a fundamental solution by pushing to an internal Harbor mirror to block direct impact from external registry daily builds

Root Cause and Resolution

Confirmed that inability to access the guest floating IP from the host Mac is the sole limitation of the single-VM architecture
An error in the systemd unit's Type=oneshot setting blocked cloud-init execution, causing an issue where SSH access was unavailable
Resolved external internet connectivity by removing the dead default route caused by the low metric of the dummy gateway inside Lima

GitOps (ArgoCD + Flux) and Secret management (Vault + ESO) flows are configured identically to the production environment
The difference in the cinder backend driver is a one-line change in cinder.conf, and the API code can be applied directly to production
Direct calls to the cinder backend API, Manila, and Octavia are scheduled for verification in a multi-node environment