Edge cluster storage roles — homelab / farm stack¶

Purpose¶

Stable reference for where bytes live in the farm edge / homelab stack: central bulk store (e.g. NAS), distributed cluster volumes (e.g. Longhorn on k3s), object/backup repositories (e.g. restic targets), and ephemeral local caches on gateways—so analyses can say “PVC tier” or “central store” without re-explaining the whole platform.

Scope¶

• In scope: Role definitions, backup/DR implications at a high level, pointers to doctrine.
• Out of scope: Your hostname, disk SKU, capacity, or network path—those are deployment tables.

Knowns¶

• The wiki treats logical DB exports (e.g. PostgreSQL / farmOS) as a distinct concern from volume snapshots—see backup package.
• Hybrid central + Longhorn patterns are analyzed in the platform cluster.

Assumptions¶

• At least one versioned recovery path exists for farm records (not sync-only mirroring).

Related analyses¶

• Longhorn vs central storage architecture — homelab / farm platform
• Longhorn role in the homelab / farm platform
• Backup and disaster recovery package — smart farm stack
• Central vs local backup scope — farm edge stack
• Platform doctrine package — homelab / farm edge

Related source notes¶

• k3s / Longhorn / Rancher / Pi platform — official captures
• homelab backup stack — official captures

Outstanding unknowns¶

• Final ratio of central bulk to cluster-backed PVCs for your fleet size and power budget.
• Whether single-site loss requires a documented restore ordering beyond the generic DR package (operator worksheet).