If those k8s clusters don't yet exist, then you could install harvester on them baremetal and deploy k8s clusters from there so downstream clusters can use the storage. It's possible to have data-only Longhorn nodes in one cluster, but I'm not sure how to do Longhorn externally outside of DR volumes:
https://longhorn.io/docs/1.6.0/snapshots-and-backups/setup-disaster-recovery-volumes/
miniature-salesclerk-33951
03/11/2024, 10:42 PM
Depending on how many physical servers/disks you have, ceph RBD might be a better fit for this.