Hi, I recently made a TrueNAS CORE VM on my proxmox instance, and it has been going quite well. Until recently, one of my friends had got the same server as mine, and same proxmox, and installed truenas SCALE, and when they did a file transfer, they got speeds up to nearly a gigabit per second.
This is extremely strange in my case, (i am using core) as I only get 300 MBPS when doing the same thing he did, transfer from an ssd on a machine over SMB to a mechanical disk. He gets a gigabit, I dont.
I ran an iperf3 server on my truenas core instance and between a PC that has a 10 gb network card, and I noticed the speeds were the same as the ones I get during a transfer over SMB.
htt ps://imgur.com/a/aR4bsQj
This is strange, I have the truenas core instance connected to its own SFP 10GBE port.
Whats the bottleneck here? Why is there a bottleneck in the first place? Do I need to install drivers or something since its a VM? My friend is running scale in the same way im doing, just using scale.
I am using disk passthrough with their disk ID’s on proxmox, so I dont know why this is happeneing. Do I need to get SCALE instead of CORE?
His setup
htt ps://imgur.com/a/ANpR7ds
Mine
h ttps://imgur.com/a/PWRuXpk
Thanks in advance, let me know if anyone has any more questions about my setup.
This is a problem. Maybe not performance, but is a reliability problem. This is known to potentially cause irrecoverable ZFS pool corruption. The only known safe way to supply disks to a TrueNAS instance using Proxmox, is through disk controller pass through.
Now to be clear, I am no expert on running TrueNAS as a VM, (under either Proxmox or VMWare). Nor do I know Proxmox. Just passing along what others have shown to be be both best practices, and known to avoid some of the problems of running TrueNAS as a VM under Proxmox.
I have seen multiple people, how does this have to affect preformance reliability? I dont think I can pass through my disk controller without using every disk in truenas, which I cannot do since I have disks that proxmox needs to use as well.
Is the preformance issue some sort of driver issue? Scale gives me 10 gigabit results on iperf3, core dosent.
I think this might be a driver issue as I had the same issue with a windows machine once, but im not sure what to do considering that a fresh core VM runs fine, along with a SCALE VM, and I dont want to wipe my instance if this is going to pop up later. Still need help
Your 300MB/s transfer speed could be fine, you would have to provide info on your hardware, how your drives are setup and the size of the files you are transferring to tell. Also I ran TN on proxmox and eventually passing disks will bite you in the butt. SAS/Sata HBA’s are cheap if your config can support one.
I have all NVMe in the Proxmox as it is the best for VMs.
All HDDs go via HBA to CORE, because I did not like SCALE in Proxmox.
Backup Proxmox to the CORE VM. ( I know, I know. Please don’t make comments. I know )
All in all, try SCALE. If it works better, keep it. Don’t like it, boot that CORE back up. What is the risk with your data I hear ?, well, I hope you have a backup of your storage.
That’s it. No one will tell you the why or how ( too many variables ), but you can find out and share your experience.