I'm kind of new to Proxmox. I have been playing with it on and off for the last year or so. What I was hoping to eventually do, was to virtualize my existing TrueNAS file server and my PiHole DHCP/DNS server and ad blocker.
TrueNAS is running on a ASUS server motherboard with a XEON E-2236 CPU, 64GB of ECC memory, 2 x 232GB NVMe sticks in a RAIDZ1 configuration for boot and 4 x 4TB Ironwolf drives in a RAIDZ2 configuration for data.
The PiHole is running on a Raspberry Pi 3 Model B+.
What I have tried to do is to reconfigure my server to run Proxmox, by replacing the two 232GB NVMe sticks with two 1TB NVMe sticks. I then installed Proxmox VE 8.1-2. By the way, I tried using Proxmox VE 8.2-1, but it would not install on my server. Anyway, 8.1-2 installed so I configured it to run off my 2 1TB NVMe drives configured as a ZFS RAIDZ1 (Mirror). Everything went well.
Once up and running, I upgraded Proxmox to 8.2-1 and then created and configured my two VMs; one for TrueNAS Scale 24.04.1.1 using disk passthru for the 4 x 4TB drives, and one for Debian 12.5 for the PiHole.
I then ran my server this way for the past three weeks, tweaking things here and there, but mostly happy with the way things were running.
Then I noticed that although i had allocated only 4 CPUs and 16GB of memory for the TrueNAS VM and 2 CPUs and 2GB of memory for the PiHole, when I looked at the memory usage under the Proxmox server summary page, I saw that almost all of my 64GB was showing as being in use. That worried me a little as I wanted to create a couple of new VMs for some other purposes (the whole reason for going to Proxmox).
I knew from previous experience that ZFS uses a lot of memory and in my TrueNAS VM all 16GB of memory was showing as being used. So what I was wondering was if my small ZFS boot pool was consuming all the memory that wasn't being used. And if that is the case, should I not use ZFS for my boot pool.
I'm not too committed at this time and it is still early enough to back out, but I would like to continue with this. Is ZFS a good or bad idea for the boot pool. Or should I stick with using extfs in a non-mirrored setup. This is a home server that only myself and immediate family use, so nothing is mission critical. i have set up automatic backups of my VMs and have tried doing recoveries with no problems at all; just takes a couple of minutes. And the install/reinstall of Proxmox is also quite trivial (with the exception of 8.2 not being able to install on my hardware). So if Proxmox fails or the NVMe boot drive fails, recovery is not too big a deal.
Just looking for some input. All comments welcome. Thanks
TrueNAS is running on a ASUS server motherboard with a XEON E-2236 CPU, 64GB of ECC memory, 2 x 232GB NVMe sticks in a RAIDZ1 configuration for boot and 4 x 4TB Ironwolf drives in a RAIDZ2 configuration for data.
The PiHole is running on a Raspberry Pi 3 Model B+.
What I have tried to do is to reconfigure my server to run Proxmox, by replacing the two 232GB NVMe sticks with two 1TB NVMe sticks. I then installed Proxmox VE 8.1-2. By the way, I tried using Proxmox VE 8.2-1, but it would not install on my server. Anyway, 8.1-2 installed so I configured it to run off my 2 1TB NVMe drives configured as a ZFS RAIDZ1 (Mirror). Everything went well.
Once up and running, I upgraded Proxmox to 8.2-1 and then created and configured my two VMs; one for TrueNAS Scale 24.04.1.1 using disk passthru for the 4 x 4TB drives, and one for Debian 12.5 for the PiHole.
I then ran my server this way for the past three weeks, tweaking things here and there, but mostly happy with the way things were running.
Then I noticed that although i had allocated only 4 CPUs and 16GB of memory for the TrueNAS VM and 2 CPUs and 2GB of memory for the PiHole, when I looked at the memory usage under the Proxmox server summary page, I saw that almost all of my 64GB was showing as being in use. That worried me a little as I wanted to create a couple of new VMs for some other purposes (the whole reason for going to Proxmox).
I knew from previous experience that ZFS uses a lot of memory and in my TrueNAS VM all 16GB of memory was showing as being used. So what I was wondering was if my small ZFS boot pool was consuming all the memory that wasn't being used. And if that is the case, should I not use ZFS for my boot pool.
I'm not too committed at this time and it is still early enough to back out, but I would like to continue with this. Is ZFS a good or bad idea for the boot pool. Or should I stick with using extfs in a non-mirrored setup. This is a home server that only myself and immediate family use, so nothing is mission critical. i have set up automatic backups of my VMs and have tried doing recoveries with no problems at all; just takes a couple of minutes. And the install/reinstall of Proxmox is also quite trivial (with the exception of 8.2 not being able to install on my hardware). So if Proxmox fails or the NVMe boot drive fails, recovery is not too big a deal.
Just looking for some input. All comments welcome. Thanks