Have you tried the CSI yet? How did it go? I haven't looked at Longhorn in awhile but it was pretty great for local disk when I checked it out. Early in my K8s journey I had a habit of scorching clusters, reviewing the destruction, resetting, and recreating. I'm a little impatient, so configs ended up in repos & I started using automation to rebuild everything. Which allowed me to move the data that I care about out of the cluster. Since the clusters were stateless from a data perspective, failure & recovery were way less painful.
---
I have a Turing Pi 2 sitting here and because compute modules are insanely expensive I don't have any to put in it. Someday the Turing RK1 modules will be available and I'm looking forward to trying in-cluster storage again. If that goes well hopefully I can use the majority of my external NAS volume for application & CSI layer backups.