Memory usage bar stucks

[buye]

Hi all, I've a problem with a virtual machine with linux installed and the indicator or progress bar for the usage memory that you see inside the dashboard of proxmox of the specific machine. This memory usage constantly increase every day till arrive at 97% (the machine has 100gb assigned) and then it stucks and never decrease. If I go to the terminal inside the virtual machine and check the memory usage running htop the memory in use is about 15/20gb. What's going wrong? Is the indicator wrong or is a real memory usage? The problem is also that sometimes the machine seems to be blocked and never resume until I stop it and start it again. Have you any ideas?
Thanks
Carlo

 

[oguz]

hi,

can you please post the vm configuration? (qm config VMID)

it's possible that this is an issue with ballooning. do you have it enabled?

 

[buye]

hi,
this is the output of the command:
bootdisk: virtio0
cores: 8
description: virtio1%3A rshares_storage%3Avm-100-disk-1,backup=no,size=9500G
ide2: cdrom,media=cdrom
memory: 100000
name: zentyal
net0: virtio=1A:99:10:CB:4E:04,bridge=vmbr0
net1: virtio=6A:82:F9:75:C4:07,bridge=vmbr1
net2: virtio=56:C1:70:87:63:FE,bridge=vmbr2
onboot: 1
ostype: l26
parent: zentyal_20_09_2019
smbios1: uuid=f0f8c82e-06d5-4df1-b19a-b3ee807b4f26
sockets: 1
unused0: rimages:100/vm-100-disk-1.qcow2
virtio0: rimages:100/vm-100-disk-2.qcow2,cache=writeback,size=120G
virtio2: rimages:100/vm-100-disk-0.qcow2,backup=0,size=3500G

Ballooning is enabled

 

[buye]

Any idea?

 

[Morantz]

Are you talking about the memory usage reporting to the ProxMox server not the individual VMs? I noticed that even with just a couple of simple CT and VMs, with almost NO LOAD on them, the memory usage reported on the ProxMox server itself seems to be creeping upwards. It is in small amounts. If I check same stats on the individual instances running on that server they are NOT doing the same.

Seems to be an issue with the GUI. This is mostly visible after 12 hours or so... Longest I have gone is 48 hours without rebooting do to other testing. After ProxMox server reboot reported RAM usage is lower. (Been making changes to the hardware as parts arrive...)

 

[oguz]

hi,

as long as you don't see any visible performance issues, this should be normal behavior.

if you install & run stress-ng on your host, the memory on your host should go down (since you require more memory on the host)

this has to do with how qemu handles memory management

 

[stark]

I see the same thing - Proxmox dash showing all memory used but when I go into the VM, I see that only 2GB is used but the rest is used for caching which I would totally expect.

So from the Proxmox host's perspective, all ram is indeed being used.

Seems at least on my setup, it's working as designed.

Instead of htop, what does the free command show?