I’ve found that when I’m copying data between NFS-shared datasets on my (singular) TrueNAS machine that the files are going over-the-network… is there any way to smarten this up?
I saw one other relevant post about this on here from back in May.. I’m using Dolphin on KDE and mounting the shares as nfs4 in fstab.
The TrueNAS is on the latest CORE - 13.3U2 if memory serves. (I’m in the process of accruing parts for an entire new NAS, scheduled to be complete in the next couple months, which the latest CE will then go on - hence not having moved to CE yet)
You are not missing anything. It cannot work any other way. NFS has no idea that you are copying. The client performs a read request for one share and a write request for another one.
To perform a local copy, use ssh and the command line on the NAS to copy the files. Midnight Commander is available as mc. You could also use rsync locally. rsync -a quite nicely preservers file ownership and access rights when run as root.
NFS was developed in 1984, was intended to be “stateless”, has no idea about user based authentication but blindly assumes that clients and servers are all under the same administrative control, UIDs and GIDs are globally identical across all systems, and users do not have root privileges on their local workstations/clients.
At least up to NFS version 3, which introduced locking and other enhancements, but kept the “UIDs must be global” and “no authentication besides admin control at setup time”.
Version 4 and newer is to my knowledge intended to improve some of these design constraints, but I have not put any work into researching that, yet. NFS v3 is good enough to host VM disk images and that is the only use case I frequently have.
There’s a reason why everyone is settling on SMB - even Apple.
SMB was supposed to have solved this a long while ago with Server-Side Copy - but it’s just not a thing for NFS…?