[SOLVED] LVM Thick with ISCSI PVE 9.0.3

IlNerd · Aug 6, 2025

Hi,
i'm trying to use snapshot on LVM Thick over ISCSi. The snapshot corretly work if the VM is powered OFF, if the VM is powered ON i recieve the following error

Code:

snapshotting 'drive-scsi0' (OS-LVM:vm-111-disk-0.qcow2)
external qemu snapshot
Creating a new current volume with xx as backing snap
  Renamed "vm-111-disk-0.qcow2" to "snap_vm-111-disk-0_xx.qcow2" in volume group "OS-LVM-G0"
  Rounding up size to full physical extent <120.02 GiB
  Logical volume "vm-111-disk-0.qcow2" created.
Formatting '/dev/OS-LVM-G0/vm-111-disk-0.qcow2', fmt=qcow2 cluster_size=131072 extended_l2=on preallocation=metadata compression_type=zlib size=128849018880 backing_file=snap_vm-111-disk-0_xx.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
blockdev replace current by xx
snapshot create failed: starting cleanup
TASK ERROR: VM 111 qmp command 'blockdev-reopen' failed - Failed to find node with node-name='drive-scsi0'

Then, if i try to take one other snapshot with the same name

Code:

snapshotting 'drive-scsi0' (OS-LVM:vm-111-disk-0.qcow2)
snapshot create failed: starting cleanup
TASK ERROR: error rename vm-111-disk-0.qcow2 to xx - target volume 'snap_vm-111-disk-0_xx.qcow2' already exists

So checking in the shell the img-info i have as output

Code:

root@pve02:~# qemu-img info /dev/OS-LVM-G0/vm-111-disk-0.qcow2
image: /dev/OS-LVM-G0/vm-111-disk-0.qcow2
file format: qcow2
virtual size: 120 GiB (128849018880 bytes)
disk size: 0 B
cluster_size: 131072
backing file: snap_vm-111-disk-0_xx.qcow2 (actual path: /dev/OS-LVM-G0/snap_vm-111-disk-0_xx.qcow2)
backing file format: qcow2
Format specific information:
    compat: 1.1
    compression type: zlib
    lazy refcounts: false
    refcount bits: 16
    corrupt: false
    extended l2: true
Child node '/file':
    filename: /dev/OS-LVM-G0/vm-111-disk-0.qcow2
    protocol type: host_device
    file length: 120 GiB (128869990400 bytes)
    disk size: 0 B

To fix this i have to use

Code:

qemu-img rebase -f qcow2 -b "" /dev/OS-LVM-G0/vm-111-disk-0.qcow2

Then the lvs is still there

Code:

lvs | grep snap_vm-111-disk-0
  snap_vm-111-disk-0_xx.qcow2 OS-LVM-G0 -wi-a----- <120,02g

and i have to remove it with

Code:

lvremove OS-LVM-G0/snap_vm-111-disk-0_xx.qcow2
Do you really want to remove active logical volume OS-LVM-G0/snap_vm-111-disk-0_xx.qcow2? [y/n]: y
  Logical volume "snap_vm-111-disk-0_xx.qcow2" successfully removed.

The /etc/pve/storage.cfg

Code:

lvm: OS-LVM
        vgname OS-LVM-G0
        base OceanStor:0.0.1.scsi-SHUAWEI_XSG1_2102354CMHTUN29100040001
        content rootdir,images
        saferemove 0
        shared 1
        snapshot-as-volume-chain 1

Am i missing something? I'm not able to snapshot powered ON VMs.

Thanks

Pifouney · Aug 6, 2025

Hey,

Did you realize fine a live snapshot without RAM snapshoting ? or not ?
Not really in peace with your case, never experienced myself. But if LUN host'mounted, maybe an option needed ( really no idea on what needed for LVM2 on LUN)

IlNerd · Aug 6, 2025

Tried both, same result

Code:

snapshotting 'drive-scsi0' (OS-LVM:vm-111-disk-0.qcow2)
external qemu snapshot
Creating a new current volume with xx as backing snap
  Renamed "vm-111-disk-0.qcow2" to "snap_vm-111-disk-0_xx.qcow2" in volume group "OS-LVM-G0"
  Rounding up size to full physical extent <120.02 GiB
  Logical volume "vm-111-disk-0.qcow2" created.
Formatting '/dev/OS-LVM-G0/vm-111-disk-0.qcow2', fmt=qcow2 cluster_size=131072 extended_l2=on preallocation=metadata compression_type=zlib size=128849018880 backing_file=snap_vm-111-disk-0_xx.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
blockdev replace current by xx
snapshot create failed: starting cleanup
TASK ERROR: VM 111 qmp command 'blockdev-reopen' failed - Failed to find node with node-name='drive-scsi0'

bbgeek17 · Aug 6, 2025

IlNerd said:
Tried both, same result

Is this a VM that existed before upgrade? If yes, post it's configuration here: qm config

Also, try to create a new VM (it does not have to have an OS), turn it on and try to snapshot.

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

IlNerd · Aug 6, 2025

Yes the VM is present since PVE 8.

Code:

qm config 111
agent: 1
bios: ovmf
boot: order=scsi0
cores: 4
cpu: x86-64-v2-AES
efidisk0: OS-LVM:vm-111-disk-1.qcow2,size=128K
ide2: none,media=cdrom
machine: pc-i440fx-8.1
memory: 8192
meta: creation-qemu=8.1.5,ctime=1718378025
name: W10-EoiP
net0: e1000e=00:0c:29:dc:ea:a6,bridge=vmbr0,tag=137
numa: 0
ostype: win10
scsi0: OS-LVM:vm-111-disk-0.qcow2,iothread=1,size=120G
scsihw: virtio-scsi-single
smbios1: uuid=564d2fe4-3a20-2e81-358c-5e5b2edceaa6
sockets: 1
tags: prod
vmgenid: 7418fccc-ee46-4059-9b30-ffece4c77b69

with a new vm the snapshot work correctly

Code:

()
snapshotting 'drive-scsi0' (OS-LVM:vm-118-disk-0.qcow2)
external qemu snapshot
Creating a new current volume with yy as backing snap
  Renamed "vm-118-disk-0.qcow2" to "snap_vm-118-disk-0_yy.qcow2" in volume group "OS-LVM-G0"
  Rounding up size to full physical extent <32.01 GiB
  Logical volume "vm-118-disk-0.qcow2" created.
Formatting '/dev/OS-LVM-G0/vm-118-disk-0.qcow2', fmt=qcow2 cluster_size=131072 extended_l2=on preallocation=metadata compression_type=zlib size=34359738368 backing_file=snap_vm-118-disk-0_yy.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
blockdev replace current by yy
blockdev-snapshot: reopen current with yy backing image
TASK OK

Code:

qm config 118
boot: order=scsi0;ide2;net0
cores: 1
cpu: x86-64-v2-AES
ide2: local:iso/mikrotik-7.16.1.iso,media=cdrom,size=64080K
memory: 2048
meta: creation-qemu=10.0.2,ctime=1754492853
name: Test
net0: virtio=BC:24:11:13:AD:80,bridge=vlan2,firewall=1
numa: 0
ostype: l26
parent: yy
scsi0: OS-LVM:vm-118-disk-0.qcow2,iothread=1,size=32G
scsihw: virtio-scsi-single
smbios1: uuid=d63f7c34-f15a-4469-ad2c-f4b9faee9440
sockets: 1
vmgenid: 8afc5627-e1c3-431b-bf16-36ce5c71e857

bbgeek17 · Aug 6, 2025

I suspect that the old VM is still running under old QEMU binary and needs a cold start.
Perhaps someone from Proxmox staff will confirm and will add this as a caveat.

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

IlNerd · Aug 6, 2025

Tried with cold start and it fail, now i'm tring to upgrade the VM version to 10, and i will update the post

Code:

machine: pc-i440fx-8.1

to

Code:

machine: pc-i440fx-10.0

IlNerd · Aug 6, 2025

bbgeek17 said:
I suspect that the old VM is still running under old QEMU binary and needs a cold start.
Perhaps someone from Proxmox staff will confirm and will add this as a caveat.

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

Confirmed, upgrading to 10.0 fixed the problem

Code:

qm config 111
agent: 1
bios: ovmf
boot: order=scsi0
cores: 1
cpu: x86-64-v2-AES
efidisk0: OS-LVM:vm-111-disk-1.qcow2,size=128K
ide2: none,media=cdrom
machine: pc-i440fx-10.0
memory: 8192
meta: creation-qemu=8.1.5,ctime=1718378025
name: W10-EoiP
net0: e1000e=00:0c:29:dc:ea:a6,bridge=vmbr0,tag=137
ostype: win10
scsi0: OS-LVM:vm-111-disk-0.qcow2,iothread=1,size=120G
scsihw: virtio-scsi-single
smbios1: uuid=564d2fe4-3a20-2e81-358c-5e5b2edceaa6
sockets: 2
tags: prod
vmgenid: 38c55b44-3abb-4567-8e64-24513a57f816

Code:

QEMU Guest Agent is not running - VM 111 qmp command 'guest-ping' failed - got timeout
snapshotting 'drive-scsi0' (OS-LVM:vm-111-disk-0.qcow2)
external qemu snapshot
Creating a new current volume with zz as backing snap
  Renamed "vm-111-disk-0.qcow2" to "snap_vm-111-disk-0_zz.qcow2" in volume group "OS-LVM-G0"
  Rounding up size to full physical extent <120.02 GiB
  Logical volume "vm-111-disk-0.qcow2" created.
Formatting '/dev/OS-LVM-G0/vm-111-disk-0.qcow2', fmt=qcow2 cluster_size=131072 extended_l2=on preallocation=metadata compression_type=zlib size=128849018880 backing_file=snap_vm-111-disk-0_zz.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
blockdev replace current by zz
blockdev-snapshot: reopen current with zz backing image
snapshotting 'drive-efidisk0' (OS-LVM:vm-111-disk-1.qcow2)
external qemu snapshot
Creating a new current volume with zz as backing snap
  Renamed "vm-111-disk-1.qcow2" to "snap_vm-111-disk-1_zz.qcow2" in volume group "OS-LVM-G0"
  Rounding up size to full physical extent 4.00 MiB
  Logical volume "vm-111-disk-1.qcow2" created.
Formatting '/dev/OS-LVM-G0/vm-111-disk-1.qcow2', fmt=qcow2 cluster_size=131072 extended_l2=on preallocation=metadata compression_type=zlib size=131072 backing_file=snap_vm-111-disk-1_zz.qcow2 backing_fmt=qcow2 lazy_refcounts=off refcount_bits=16
blockdev replace current by zz
blockdev-snapshot: reopen current with zz backing image
TASK OK

bbgeek17 · Aug 6, 2025

Great, you can mark this thread as solve to help others by editing first post and changing the prefix of the subject via drop-down

The important line change is likely:
machine: pc-i440fx-8.1 | machine: pc-i440fx-10.0

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

fiona · Aug 7, 2025

Hi,
thank you for the report! A patch to improve this was proposed on the development mailing list: https://lore.proxmox.com/pve-devel/20250807104832.51784-1-f.ebner@proxmox.com/T/#u

Search

Search

[SOLVED] LVM Thick with ISCSI PVE 9.0.3

IlNerd

New Member

Pifouney

Active Member

IlNerd

New Member

bbgeek17

Distinguished Member

IlNerd

New Member

bbgeek17

Distinguished Member

IlNerd

New Member

IlNerd

New Member

bbgeek17

Distinguished Member

fiona

Proxmox Staff Member

We value your privacy