Setting up Dataset in TrueNAS Core for Proxmox VM and CT Storage?

(I realize my current TrueNAS Core build is not ideal for VM storage–they should be on NVME drives and I don’t have enough of those to use, for one.)

I’m running a test TrueNAS Core install (my first TrueNAS Core install) on a Dell Optiplex 5040 with 4x1.92 TB enterprise SATA SSDs for storage, in a single pool with two mirror VDEVs. Aside from creating it and making sure ashift was set to 12 (the GUI doesn’t let you tweak this, so I was relying on it to be automatic), I haven’t tweaked the settings of the data storage pool in any way yet.

Aside from setting recordize and volblocksize attributes as necessary in my VM and CT storage datasets, are there any other attributes I need to change for remote Proxmox nodes on my LAN to use this storage? My goal is to set up the datasets, share them out to Proxmox, and not have to worry about tweaking them again. :stuck_out_tongue:

  1. I’ve previously been using NFS on other nodes for accessing my backup server and the ISO storage on my QNAP, but I’m open to using CIFS if it would work better with TrueNAS Core.
  2. There are some things that I’d like to use iSCSI for, as well.
  3. There are also some settings used in Core that I don’t recognize from Proxmox/ZFS on Linux. The boot pool and my storage pool both use an acltype of NFSv4, which I had never heard of before. I’m not sure if I need to change that if I’m going to use NFS and CIFS, or what impact it has on iSCSI.
  4. In addition to the above, is there anything else I need to tweak?

Thanks!

Make sure you have sync disabled on the datasets to get better write performance… Yes there are risk and I cover that in this video:

1 Like

Thanks, Tom. This video is just what I needed. :slight_smile:

I still haven’t gotten this done. My Proxmox servers are still mostly for hobby use (I suppose teh Unifi controller counts as production), and the last month has been full of … events.

I’m hoping to try setting this up this week. I need to watch the video again, but my understanding is that the main risk without sync enabled is the potential for data loss in the event of a power outage or drive/controller failure, correct? I’ve got a UPS and will have confidence in the hardware I have until (1) something fails or (2) I can get better hardware, so I think I’m good to go.