I have TrueNas Scale setup on my Proxmox HomeServer. From there I run a pod with Jellyfin, and one with NextCloud. I also have an SMB share setup. In the last few days, TrueNas has started crashing as soon as I try to upload or download larger volumes of data. ( in excess of about 50MB)
I was previously able to move about 0.5 TB of movies through the SMB share without any issues. On another VM I run nginx to route traffic, but the issue occurs when uploading/downloading data locally as well, so I don’t think it has any impact.
My system is running on an old gaming PC from 2014, no fancy server equipement in there.
Hardware:
Asus Z97-PRO Gamer Motherboard
intel core i7-4790
2x8 GB of 1333MHZ RAM (not sure what make anymore) It should run in dual-channel
2x1TB Red SSD (latest model) running in Raid 1 think (says mirror-0 in the shell status) - This is where proxmox is installed
3x5TB Red HDD (units from 2022) running in Raidz1-0 - I have about 8TB storage available with that setup, 6 of which are used by TrueNas. The remaining amount is currently unused
I’m happy to provide further details, not sure what else is relevant. I have TrueNas 24.04.2 installed. Tried to upgrade to the latest stable version a few times, but the update doesn’t run when the VM is started manually, and it crashed when applied normally. My best guess is that it crashed while downloading the files (which is why the vm doesn’t reboot automatically), and thus the update build is never complete.
Any idea what could be causing this issue? As far as I know, nothing substantial changed when it started happening.
Hello,
Thanks for answering!
I’m not entirely sure, but since I only passed a ‘single drive’, so to speak, and I can only manager the one drive from TrueNas I assume I passed the SATA controller with all disks attached to it. I take it that’s not the right way to do it?
If you have other disks on the SATA controller that Proxmox uses, then you will have passed only the disk through.
I don’t use Proxmox myself, but I think that the current recommendations are:
Install TrueNAS Scale native / bare metal, and use TrueNAS virtualisation for guess O/S; or
If you are going to virtualise TrueNAS (under Proxmox),then you should pass through the SATA/SAS controller and all attached disks to TrueNAS and allow TrueNAS to put ZFS on them.
There is insufficient detail in what you have said so far to be sure, but it sounds like you may have had both Proxmox and TrueNAS writing to the same ZFS system - which is highly likely to cause metadata corruption and the loss of the pool.
But actually, since you have said that you have a 3x RAIDZ1 that is used by TrueNAS, it sounds like you may have created a ZVol in Proxmox and passed that through to TrueNAS as a virtual drive.
Can you provide some screen shots of your Proxmox settings for the TrueNAS VM?
Hello,
I’ve been reading up more and found similar recommendations. I’m currently trying to get the ‘most value’ out of an old desktop, so I don’t have the option to pass the whole controller without first buying a separate one. (I’ll keep this as a last resort. But then I also read that running non-ECC RAM could cause problems so I feel like I’m grapsing at straws.
I’m quite sure I did this:
“But actually, since you have said that you have a 3x RAIDZ1 that is used by TrueNAS, it sounds like you may have created a ZVol in Proxmox and passed that through to TrueNAS as a virtual drive.”
I’ll add some screenshots here. In the meantime I’ll wipe my installation and pass the disks directly to TrueNas see if that works better.
