[Qemu-discuss] corrupted QCOW2 image

[David Magda]

Hello,

We've run into a corrupted QCOW2 disk image on one of our VMs. I can
consistently trigger it by running an 'apt-get update' on the Debian 5
guest, which causes a SIGSEGV to be sent to the qemu-kvm process.

I've tried converting it, but get the following error:

    $ /srv/mcvmm/qemu-kvm-1.0/bin/qemu-img convert -p -f qcow2 -O qcow2 -o
preallocation=metadata www2-col.img.nonpre www2-col_prealloc.img
    qemu-img: error while reading sector 226750464: Input/output error
    $ /srv/mcvmm/qemu-kvm-1.0/bin/qemu-img info  www2-col.img.nonpre
    image: www2-col.img.nonpre
    file format: qcow2
    virtual size: 150G (161061273600 bytes)
    disk size: 17G
    cluster_size: 4096

A "check" reports lots of errors:

    9947 errors were found on the image.
    Data may be corrupted, or further writes to the image may corrupt it.

    2578 leaked clusters were found on the image.
    This means waste of disk space, but no harm to data.

    3046 internal errors have occurred during the check.

    An error has occurred during the check: Success
    The check is not complete and may have missed error.

I can attach the output of 'qemu-img check' (970KB) if that's of any use.

The guest is running Debian 5:

    Linux www2-col 2.6.32-bpo.5-amd64 #1 SMP Wed Jul 20 09:10:04 UTC 2011
x86_64 GNU/Linux

The hosting platform is running Debian 6:

    Linux vm20 2.6.39-bpo.2-amd64 #1 SMP Tue Jul 26 10:35:23 UTC 2011
x86_64 GNU/Linux

Storage is on NFS from an Isilon 6.5.4 cluster.

What are our options for fixing this disk image?

Thanks for any info.