I have 3 storage servers that I would like to use as gluster servers for VM hosting and some Maildir hosting. these 3 servers will be connected to a bunch of hypervisors servers. I'll create a distributed replicated volume with SATA disks and ZFS (to use SLOG) and another distributed replicated volume shared with NFS-Ganesha to use as Dovecot mailbox storage. The dovecot server is a VM running on the same gluster nodes where ganesha is. Is this an issue ? I could try to store Dovecot VM on local disks on hypervisors without using the same Gluster nodes for VM hosting, then access ganesha directly from inside the VM. By using a shared NFS server, I can bring up multiple dovecot servers even without live migration, because there is always at least another dovecot VM ready to run on a different server accessing the same NFS share (stored on gluster). Any advice ?