ZFS Raid array eats alot of space

[mditin]

Hello
I have a strange problem where RaidZ(1) takes away alot of space.
I tried setting up Raid array with mdadm and I found out that its not supported. So I tried seeting it up from 3x4TB disks. In Proxmox WebUI it shows as 12TB space. So if I set it to RaidZ it should be 7.9TB as its shofn under node Storage... so far everything looking good. But when I try to add disk to VM (Ubuntu server) I can only arrange 5350TiB and disk is shown as full. When I check space on server it only show up as 5.3TB space.

there should be at least 7.5TB, or am I wrong?
Server is running as media centre, and disk space is important to, but I still want to have redundancy.

Specifications:
MB: MSI B350
CPU: Pyzen5 2400g
RAM: 16gb
SSD: Nvme Samsung 860 250gb
HDD: 3x Seagate Ironwolf 4TB

 

[leesteken]

Indeed, ZFS raidZ1 has huge padding overhead, especially with a small number of drives and a small volblocksize. See Dunuin's excellent analysis and tests about this on this forum.

 

[Dunuin]

Short:
Go to "Datacenter -> Storage -> YourStorageName -> Edit" and set the "Block size" to at least 16K for a 3-disk raidz1. Then backup and restore all your VMs, so new VMs get created overwriting the old VMs, as the volblocksize can only be set at the creation of a zvol.

For more please search this forum for "padding overhead".