Due to Corosync there is clearly a finite and rather small number of nodes that a single Proxmox cluster can support, yet Ceph clusters have no such size limit.
The only thing preventing multiple Proxmox clusters going hog-wild on a shared Ceph cluster is the certainty of overlap and collision of machine IDs and by extension the RBD disk names.
Each cluster will unknowingly create a machine using the next available ID and create RBD disks following the innocent convention vm-$id-disk-$k and so on. Even with external non-HCI storage support, a Proxmox cluster handling multiple Ceph clusters faces the challenge of ensured uniqueness.
What I'm looking for is a way to either pre-define a range of permitted VM IDs at the cluster level to ensure against overlap with multiple clusters, and an even better solution to offer in addition to this would also be to offer specifying a cluster-specific prefix on virtual disk names.
So if we have 2 clusters, pve01.domain.tld and pve02.domain.tld, all of the virtual disks in each cluster could be prefixed pve01- or pve02- or whatever the admin wants. I would be great if the virtual disk names were defined in a text file somewhere but I doubt it. Looking for some guidance.
The only thing preventing multiple Proxmox clusters going hog-wild on a shared Ceph cluster is the certainty of overlap and collision of machine IDs and by extension the RBD disk names.
Each cluster will unknowingly create a machine using the next available ID and create RBD disks following the innocent convention vm-$id-disk-$k and so on. Even with external non-HCI storage support, a Proxmox cluster handling multiple Ceph clusters faces the challenge of ensured uniqueness.
What I'm looking for is a way to either pre-define a range of permitted VM IDs at the cluster level to ensure against overlap with multiple clusters, and an even better solution to offer in addition to this would also be to offer specifying a cluster-specific prefix on virtual disk names.
So if we have 2 clusters, pve01.domain.tld and pve02.domain.tld, all of the virtual disks in each cluster could be prefixed pve01- or pve02- or whatever the admin wants. I would be great if the virtual disk names were defined in a text file somewhere but I doubt it. Looking for some guidance.
