proxmox questions

[ekke]

I got a SM x10srh with both onboard lsi and 10 sata . I would prefer to passthrough both controllers to a vm, FreeNAS. Is it possible to passthrough the onboard sata ports? or are they tied to the chipset so it's not possible?

and another question have anyone tried eve-NG, Eve-NG , under proxmox?

 

[T_Minus]

You don't have to "Pass Through" anything in ProxMox, you can run/manage ZFS at the host level, and you should.

Regarding passing through 'sata ports'... it depends on the motherboard, chipset, generation, etc... Some of the older boards I know you can, not sure about the one you have. I just did it recently with an Intel E5 v1/2 system too as a 'test', and it worked. If you can check the box to pass through you can do it iirc on my SM x10 I can't.

 

[ekke]

You don't have to "Pass Through" anything in ProxMox, you can run/manage ZFS at the host level, and you should.

Regarding passing through 'sata ports'... it depends on the motherboard, chipset, generation, etc... Some of the older boards I know you can, not sure about the one you have. I just did it recently with an Intel E5 v1/2 system too as a 'test', and it worked. If you can check the box to pass through you can do it iirc on my SM x10 I can't.

Freenas is so Nice to work with regarding filesharing and ZFS in linux in userland isnt efficiant(?). it feeds hackish ,freenas is a more polished solution.

I Will be using ZFS in proxmox for vm storage tho

Skickat från min FRD-L09 via Tapatalk

 

[Jon Massey]

Yes, best way I've found is to pass the disks through by id (/dev/disk/by-id/...) using virtio.

 

[ekke]

but thats a big nono what I have read. zfs needs direct access to discs. no ? the freenas people would kill me.

 

[Jon Massey]

That is giving them direct access to the disks, just not the controller. Doesn't seem to make any difference.

 

[_alex]

Not sure if it's really better to manage zfs in proxmox. putting it into vm gives more flexibility (i.e. pti=off) and keeps zfs from exhaustive ressource usage within the hypervisor.

I also would feel more comfortable knowing zfs is limited on N (pinned) CPU / RAM and running on it's own (lts or even state-of-the-art) kernel and in a Version that doesn't change with hypervisor updates.