[PVE-User] lzo files conundrum

[Ronny Aasen]

On 11.03.2020 08:55, Alwin Antreich wrote:
> Hello Renato,
> 
> On Wed, Mar 11, 2020 at 07:35:21AM +0100, Renato Gallo via pve-user wrote:
>> Date: Wed, 11 Mar 2020 07:35:21 +0100 (CET)
>> From: Renato Gallo <renato at aster-lab.com>
>> To: pve-user at pve.proxmox.com
>> Cc: g noto <g.noto at way-interactive.com>
>> Subject: lzo files conundrum
>> X-Mailer: Zimbra 8.8.15_GA_3829 (ZimbraWebClient - FF68
>>   (Linux)/8.8.15_GA_3829)
>>
>> Hello,
>>
>> we have two machines same system 100 Gb of virtual disk both have the disks on a zfspool.
>> from the df -h point of view the disks use is quite similar.
>> we backup those machine regularly during the weekend using the scheduled backup of proxmox from the datastore menu.
>> we cannot understand why one machine have a .lzo file of 9.something gigas and the other has a backup .lzo file of around 40 gigas.
>> can you explain this to us ?
> Possibly thin-provisioning.
> 
> In the storage definition the _Thin provision_ needs to be set, to
> create thin-provisioned zVOLs from the start. And OFC, you will need to
> have the VM disks setup with SCSI and discard on to run a TRIM inside
> the VM.
> 

you can also run fstrim from inside the vm to manually trim filesystems.

Ronny