6. Supabase Database Storage on NAS via NFS#

Status: Accepted

Context#

Supabase PostgreSQL needs persistent storage that survives pod restarts and rescheduling. An existing NAS on the LAN already serves NFS exports for LLM models (rkllama, llamacpp).

Example: in the author’s cluster, the NAS is at 192.168.1.3.

Decision#

Use a static NFS PersistentVolume/PersistentVolumeClaim pointing to /bigdisk/OpenBrain on the NAS, following the established rkllama/llamacpp pattern.

Consequences#

  • Data survives pod restarts, rescheduling, and node reboots

  • Depends on NAS availability (single point of failure)

  • No dynamic provisioning — PV/PVC are committed to the repo

  • persistentVolumeReclaimPolicy: Retain prevents accidental data loss