[PVE-User] Volume chain snapshotting

[PVE-User] Volume chain snapshotting

[Mark Schouten via pve-user]

[-- Attachment #1: Type: message/rfc822, Size: 4794 bytes --]

From: "Mark Schouten" <mark@tuxis.nl>
To: "Proxmox VE user list" <pve-user@lists.proxmox.com>
Cc: "Alexandre de Rumier" <aderumier@odiso.com>
Subject: Volume chain snapshotting
Date: Mon, 24 Nov 2025 14:42:16 +0000
Message-ID: <em9f436164-dd83-4d7d-95aa-383aa205d080@4f26fb63.com>

Hi all,

Since PVE9, we have Volume Chain Snapshotting. The documentation states: 
'Enabling or disabling this flag only affects newly created virtual disk 
volumes.’

Two questions:
1: Are there plans to make it possible to convert existing LVM Backed 
storages with images that already exist? Or is there an idea how we can 
convert existing images?
2: Why is this in status ’Technology Preview’? What are the risks that 
are in the back of the heads of the developers? Or is it mainly ’there 
might be bugs, but we expect dataconsistency to be OK’?

Thanks,

—
Mark Schouten
CTO, Tuxis B.V.
+31 318 200208 / mark@tuxis.nl

[-- Attachment #2: Type: text/plain, Size: 157 bytes --]

_______________________________________________
pve-user mailing list
pve-user@lists.proxmox.com
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-user

Re: [PVE-User] Volume chain snapshotting

[g.husson_proxmox-pve-user--- via pve-user]

[-- Attachment #1: Type: message/rfc822, Size: 5067 bytes --]

From: g.husson_proxmox-pve-user@liberasys.com
To: Mark Schouten via pve-user <pve-user@lists.proxmox.com>
Subject: Re: [PVE-User] Volume chain snapshotting
Date: Mon, 24 Nov 2025 16:49:21 +0100
Message-ID: <35630053-f685-4e5a-93a5-fb61e5cbd146@liberasys.com>

Hello Mark.

1. I may be wrong but I think that if your LVM storage has been 
initialised as THICK provisioning, it should be allright.

2. The best summary I have red about snapshots over iSCSI is : 
https://kb.blockbridge.com/technote/proxmox-qcow-snapshots-on-lvm/

Best regards,
Gautier Husson - Liberasys.


Le 24/11/2025 à 15:42, Mark Schouten via pve-user a écrit :
> Hi all,
>
>
>
> Since PVE9, we have Volume Chain Snapshotting. The documentation states:
> 'Enabling or disabling this flag only affects newly created virtual disk
> volumes.’
>
>
>
> Two questions:
>
> 1: Are there plans to make it possible to convert existing LVM Backed
> storages with images that already exist? Or is there an idea how we can
> convert existing images?
>
> 2: Why is this in status ’Technology Preview’? What are the risks that
> are in the back of the heads of the developers? Or is it mainly ’there
> might be bugs, but we expect dataconsistency to be OK’?
>
>
>
> Thanks,
>
>
> _______________________________________________
> pve-user mailing list
> pve-user@lists.proxmox.com
> https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-user





[-- Attachment #2: Type: text/plain, Size: 157 bytes --]

_______________________________________________
pve-user mailing list
pve-user@lists.proxmox.com
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-user

Re: [PVE-User] Volume chain snapshotting

[Fiona Ebner]

Am 24.11.25 um 4:55 PM schrieb g.husson_proxmox-pve-user--- via pve-user:
> 2. The best summary I have red about snapshots over iSCSI is : https://kb.blockbridge.com/technote/proxmox-qcow-snapshots-on-lvm/
> 
> Le 24/11/2025 à 15:42, Mark Schouten via pve-user a écrit :
>> Hi all,
>> 2: Why is this in status ’Technology Preview’? What are the risks that
>> are in the back of the heads of the developers? Or is it mainly ’there
>> might be bugs, but we expect dataconsistency to be OK’?

Yes, the blockbridge article is one of the things that still need to be
considered.

We did still have quite a few bug fixes between the 9.0 and 9.1 release,
see the qemu-server and libpve-storage-perl changelogs, mostly
relatively minor stuff, but also one case of LVM metadata corruption,
because of missing locking in some edge cases. So the feature is still
in the process of maturing. We cannot 100% guarantee no corruption, but
will be looked into immediately if something pops up (nothing else has
as of yet).

Another mostly cosmetic issue that would be nice to get fixed is size
reporting while the LVs are inactive, i.e.
https://bugzilla.proxmox.com/show_bug.cgi?id=6819

Best Regards,
Fiona


_______________________________________________
pve-user mailing list
pve-user@lists.proxmox.com
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-user

Re: [PVE-User] Volume chain snapshotting

[Fiona Ebner]

Am 24.11.25 um 4:31 PM schrieb DERUMIER, Alexandre via pve-user:
> 
> Hi Mark !
> 
>>> 1: Are there plans to make it possible to convert existing LVM Backed
>>> storages with images that already exist? Or is there an idea how we
>>> can convert existing images?
> 
> It should be possible to use move disk to do it. (Define same storage
> twice, without volume-chain  and another with with-volume chain, and
> move to qcow2).

Please don't! This can break shared LVM storages because it fails to
honor locking guarantees! LVM metadata corruption is possible with
concurrent operations and Proxmox VE won't be aware of it actually being
the same storage under the hood.

> Alternatively, I think it could be possible to simply enable it on your
> current storage (in storage.cfg directory), and do the move disk to
> qcow2 format.
> (but be carefull to not have existing internal snapshots, as it's not
> supported to have both kind of snapshots at the same time)

You can also just enable the flag for an existing LVM storage via the
web UI/API/CLI, no need for manual editing. You will not be able to
disable the flag anymore as long as qcow2 formatted images exist
however. Moving disks to the same storage while changing formats is a
possibility, yes.


_______________________________________________
pve-user mailing list
pve-user@lists.proxmox.com
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-user

Re: [PVE-User] Volume chain snapshotting

[DERUMIER, Alexandre via pve-user]

[-- Attachment #1: Type: message/rfc822, Size: 12818 bytes --]

From: "DERUMIER, Alexandre" <alexandre.derumier@groupe-cyllene.com>
To: "mark@tuxis.nl" <mark@tuxis.nl>, "pve-user@lists.proxmox.com" <pve-user@lists.proxmox.com>
Cc: "DERUMIER, Alexandre" <aderumier@odiso.com>
Subject: Re: Volume chain snapshotting
Date: Mon, 24 Nov 2025 15:31:28 +0000
Message-ID: <3bb492d1dacac4e1ad8bbc2aac3d145ab5064121.camel@groupe-cyllene.com>

Hi Mark !

>>1: Are there plans to make it possible to convert existing LVM Backed
>>storages with images that already exist? Or is there an idea how we
>>can convert existing images?

It should be possible to use move disk to do it. (Define same storage
twice, without volume-chain  and another with with-volume chain, and
move to qcow2).

Alternatively, I think it could be possible to simply enable it on your
current storage (in storage.cfg directory), and do the move disk to
qcow2 format.
(but be carefull to not have existing internal snapshots, as it's not
supported to have both kind of snapshots at the same time)








[-- Attachment #2: Type: text/plain, Size: 157 bytes --]

_______________________________________________
pve-user mailing list
pve-user@lists.proxmox.com
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-user