Confused... VM FileServer using ZFS on the host.

[palazo]

Hello Guys, i was using proxmox VE from long time ago, using clusters etc, on different ways.

Now i need to make a File Server and i want use one of the Host who have 3 HDD Disk 8TB on a zfs raizZ1-0 on the PVE host.

The first thing I thought of was create raw images for the VM that would work as a file server, but after doing some tests I realized that since the size of the images is so large, backups can become very difficult to perform or restore.

I have read several posts that warn of different difficulties in case of doing a passthrowg of the disks.

I would like to know what can be the best way to use the entire 3 disks in a VM, using the ZFS pool that was created on the host.

Or if it would be better to passthrough each of the disks to the vm and install ZFS on the VM by creating the pool inside it.

I don't want to use NFS to share the disks from the host to the VM, since I just want the VM to function as a samba fileserver.

 

[leesteken]

Run a Linux fileserver in a unprivileged container? Then you can bind mount any host filesystem.

 

[palazo]

Run a Linux fileserver in a unprivileged container? Then you can bind mount any host filesystem.

Thx for your fast response; didn't use containers already, tried with one, seems works pretty good with subvols of the pools.
The backups of the subvols seems really fast and ez to manage on the resources section.

The only thing im looking now is i dont like that uid:gid mapping.
On the container it shows like root uid is 0.
On the host if i watch the subVol file or folder it shows like is 100000, so the real UID on the Disk is 100000.
Could be hard if for some reason i need to move the disks from the host server.
Any way to have same uid:gid on countainer and host?

 

[palazo]

Thx for your fast response; didn't use containers already, tried with one, seems works pretty good with subvols of the pools.
The backups of the subvols seems really fast and ez to manage on the resources section.

The only thing im looking now is i dont like that uid:gid mapping.
On the container it shows like root uid is 0.
On the host if i watch the subVol file or folder it shows like is 100000, so the real UID on the Disk is 100000.
Could be hard if for some reason i need to move the disks from the host server.
Any way to have same uid:gid on countainer and host?

Im go to respond to my self; seems if i make the container privileged i can do that mounts without UID and GID mappings.
And mount the zfs entirely.
So if for some reason i need to move the disks to other host i can do it without any troubles.