Hi all, I've recently bought a used Mac Mini 2018, i7 3.2ghz, 64gb ram, 1tb ssd and I've managed to install Proxmox version 8.4. I don't really like using cloud images so I've built a vm template using packer and deployed 7 vms for my testing env all by Terraform.
I've unfortunately noticed that there's some IO delay and I've already seen on the forum that the problem seems to be related to consumer ssd, but I'm tring to post my config and my questions anyway, maybe I can receive a fresh advice or answer, or someone recently managed to have the same problem and found a solution
Proxmox is installed flat from it's original bootable media, but I selected XFS as base partition, the other settings are the defaults
The vms are all Debian12 installed from official ISO
The vms config
Now I have 7 Vms, 6 nodes k3s cluster (3 masters 3 workers), 1 idle test node, and some services like ha proxy, keepalived
The k3s are empty, so the machines are not on heavy load
In this situation I have a 5-10% constant IO delay, is that normal? I made any mistake on the configs or I can modify something to optimize?
I also tried to use the optimized T2 kernel from https://github.com/AdityaGarg8/pve-edge-kernel-t2 but nothing changed.
Any clue are appreciated and feel free to ask for some other information, logs or anything, thanks!
I've unfortunately noticed that there's some IO delay and I've already seen on the forum that the problem seems to be related to consumer ssd, but I'm tring to post my config and my questions anyway, maybe I can receive a fresh advice or answer, or someone recently managed to have the same problem and found a solution
Proxmox is installed flat from it's original bootable media, but I selected XFS as base partition, the other settings are the defaults
The vms are all Debian12 installed from official ISO
The vms config
Code:
#Packer ephemeral build VM
agent: enabled=1,fstrim_cloned_disks=0,type=virtio
balloon: 0
bios: ovmf
boot: order=scsi0;net0
cicustom: user=local:snippets/user-data-testnode.yaml
cores: 1
cpu: cputype=host
cpuunits: 1024
efidisk0: local-lvm:vm-150-disk-0,efitype=4m,pre-enrolled-keys=1,size=4M
kvm: 1
machine: q35
memory: 2048
meta: creation-qemu=9.2.0,ctime=1752154337
name: testnode
net0: virtio=AA:BB:CC:DD:EE:50,bridge=vmbr0,firewall=0
numa: 0
onboot: 1
ostype: l26
scsi0: local-lvm:vm-150-disk-1,aio=io_uring,backup=1,cache=none,discard=on,iothread=1,replicate=0,size=50G,ssd=1
scsi10: local-lvm:vm-150-cloudinit,media=cdrom
scsihw: virtio-scsi-single
smbios1: uuid=f6b9568e-66fd-4555-a27e-52d7f1092edd
sockets: 1
vmgenid: 4e9bb656-75b1-45a0-8c60-01cafd93aabdNow I have 7 Vms, 6 nodes k3s cluster (3 masters 3 workers), 1 idle test node, and some services like ha proxy, keepalived
The k3s are empty, so the machines are not on heavy load
In this situation I have a 5-10% constant IO delay, is that normal? I made any mistake on the configs or I can modify something to optimize?
I also tried to use the optimized T2 kernel from https://github.com/AdityaGarg8/pve-edge-kernel-t2 but nothing changed.
Any clue are appreciated and feel free to ask for some other information, logs or anything, thanks!