So I have a proxmox server on which I run a few VM’s and the other day it completely ran out of space. This was because of overprovisioning through thin volumes.

After much head scratching and metaphorically banging my head against a wall, here are the things I learnt.

Empty Trash

Local Trash

Make sure that have emptied the trash on the VM’s .Ubuntu has this issue and so might other distributions

Network Trash

If you have SAMBA enabled on your VM’s make sure that the Recycle Bin is not enabled. I have openmediavault running on a VM and I had to go through and disable the Recycling Bin. Make sure that the Recycle bin is emptied. They are hidden folders in the root of your shares.

Correct Driver & Settings

• When setting up the hard drive for your VM, make sure you use virtio-scsi (or just scsi on the web interface).
  • If you disk is already set up using IDE or VirtIO,
    • Delete it. Don’t worry, it’s only deleting the link. The disk itself will show up in the interface afterwards
    • Double click on the unattached disk and select SCSI and Discard
    • You might have to fix the references to the drive in the OS
• On the Device Screen, make sure discard is selected.

TRIM

Configure the OS to send TRIM commands to the drive

Linux

On Mount

You can pass the parameter discard to any mountpoint and the correct TRIM commands will be sent to the disk. HOWEVER, this is apparently a big performance hit.

To do the actual trim, run

$ fstrim /

OR to run fstrim on all supported drives

$ fstrim -a

Digital Ocean has a detailed post about setting TRIM and setting up a schedule etc.

Windows

My deepest apologies! Also, I don’t run Windows on any of my VM’s so I have no experience with it.