ERROR: Snapshot -27 (File too large)

MiracleToast

New Member
Dec 16, 2018
4
0
1
56
Hello,

Have encountered a strange error with a VM inside our v4.4 instance. Created a snapshot, and now are unable to roll back:

/usr/bin/qemu-img snapshot -a BeforeUpdatesNov2018 /vd01/images/104/vm-104-disk-1.qcow2
qemu-img: Could not apply snapshot 'BeforeUpdatesNov2018': -27 (File too large)


The 'info' command doesn't report any errors w/ the snapshot chain:

/vd01/images/104# qemu-img info ./vm-104-disk-1.qcow2
image: ./vm-104-disk-1.qcow2
file format: qcow2
virtual size: 60G (64424509440 bytes)
disk size: 30G
cluster_size: 65536
Snapshot list:
ID TAG VM SIZE DATE VM CLOCK
1 BeforeRemovingBellevue20180514 0 2018-05-14 14:13:35 7768:01:48.746
2 BeforeUpdatesNov2018 0 2018-11-14 14:34:0512185:22:18.149
3 BeforeRollback 0 2018-12-13 11:13:08 00:00:00.000
Format specific information:
compat: 1.1
lazy refcounts: false
refcount bits: 16
corrupt: false


The snapshot .raw seems ok:
qemu-img info ./vm-104-state-BeforeUpdatesNov2018.raw
image: ./vm-104-state-BeforeUpdatesNov2018.raw
file format: raw
virtual size: 3.7G (4007542784 bytes)
disk size: 3.7G



We are able to create a new snapshot (the 'BeforeRollback') (#3), and it works as expected -- created fine, and I can successfully rollback ok to snapshot#3. But, the critical snapshot point is #2 (or even #1 at this point). Both are throwing the 'File too large' error.

Running Proxmox v4.4-24,
qemu-img version 2.9.1pve-qemu-kvm_2.9.1-9~pve4
Copyright (c) 2003-2017 Fabrice Bellard and the QEMU Project developers


Any ideas how we can revert our system back to the snapshot, or, somehow build a new image based on that November snapshot?


Thanks!
 
Hi Wolfgang,

We copied the files to another Proxmox instance running 5.2-1 and ran the same command. Still not working, but the error msg has a little more detail:

root@hercules:/vm/data4/test# qemu-img --version
qemu-img version 2.11.1pve-qemu-kvm_2.11.1-5
Copyright (c) 2003-2017 Fabrice Bellard and the QEMU Project developers

root@hercules:/vm/data4/test# qemu-img info ./vm-104-disk-1.qcow2
image: ./vm-104-disk-1.qcow2
file format: qcow2
virtual size: 60G (64424509440 bytes)
disk size: 71G
cluster_size: 65536
Snapshot list:
ID TAG VM SIZE DATE VM CLOCK
1 BeforeRemovingBellevue20180514 0 2018-05-14 14:13:35 7768:01:48.746
2 BeforeUpdatesNov2018 0 2018-11-14 14:34:0512185:22:18.149
3 BeforeRollback 0 2018-12-13 11:13:08 00:00:00.000
Format specific information:
compat: 1.1
lazy refcounts: false
refcount bits: 16
corrupt: false

root@hercules:/vm/data4/test# qemu-img snapshot -a BeforeUpdatesNov2018 /vm/data4/test/vm-104-disk-1.qcow2
qemu-img: Could not apply snapshot 'BeforeUpdatesNov2018': Failed to load snapshot: File too large
qemu-img: Failed to flush the refcount block cache: File too large
root@hercules:/vm/data4/test#

root@hercules:/vm/data4/test# qemu-img check ./vm-104-disk-1.qcow2
qemu-img: Check failed: Cannot allocate memory


This server has lots of memory and disk space, and storage is running on hardware raid. Is there a way to allocate more resources to qemu-img?
 
For fun, I also setup a clean system and compiled the latest version of qemu 3.1 -- it's showing the same error when I tried to apply the snapshot:

'Failed to flush the refcount block cache: File too large'
 
Hi Wolfgang,

Do you have any suggestions for how we can rollback our image? Why would we be able to create a snapshot that can't be reverted?

Appreciate any help on this.
 
Is it possible to delete another snapshot in this image?