[Users] Migrating from KVM to oVirt 3.1 fails - corrupt OVF

[Ayal Baron]

----- Original Message -----
> On 07/20/2012 09:19 PM, Trey Dockendorf wrote:
> > On Fri, Jul 20, 2012 at 11:32 AM, Itamar Heim <iheim at redhat.com>
> > wrote:
> >> On 07/20/2012 07:21 PM, Trey Dockendorf wrote:
> >>>
> >>> On Fri, Jul 20, 2012 at 3:52 AM, Itamar Heim <iheim at redhat.com>
> >>> wrote:
> >>>>
> >>>> On 07/20/2012 02:08 AM, Trey Dockendorf wrote:
> >>>>>
> >>>>>
> >>>>> On Thu, Jul 19, 2012 at 4:00 AM, Matthew Booth
> >>>>> <mbooth at redhat.com>
> >>>>> wrote:
> >>>>>>
> >>>>>>
> >>>>>> On 18/07/12 23:52, Itamar Heim wrote:
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>> On 07/18/2012 06:00 PM, Trey Dockendorf wrote:
> >>>>>>>>
> >>>>>>>>
> >>>>>>>>
> >>>>>>>> I'm attempting to fine-tune the process of getting my
> >>>>>>>> KVM/Libvirt
> >>>>>>>> managed VMs over into my new oVirt infrastructure, and the
> >>>>>>>> virt-v2v
> >>>>>>>> import is failing in the WUI with "Failed to read VM
> >>>>>>>> 'dh-imager01'
> >>>>>>>> OVF, it may be corrupted".  I've attached both engine and
> >>>>>>>> vdsm logs
> >>>>>>>> that are a snapshot from when I ran the virt-v2v command
> >>>>>>>> until I saw
> >>>>>>>> the failure under Events.
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>> matt - any thoughts?
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>>> Nothing springs to mind immediately, but it sounds like v2v is
> >>>>>> producing
> >>>>>> an
> >>>>>> invalid OVF. If somebody can diagnose what the problem with
> >>>>>> the OVF is
> >>>>>> I
> >>>>>> can
> >>>>>> fix v2v.
> >>>>>>
> >>>>>> Matt
> >>>>>>
> >>>>>>
> >>>>>>>
> >>>>>>>>
> >>>>>>>> virt-v2v command used...
> >>>>>>>>
> >>>>>>>> # virt-v2v -i libvirtxml -o rhev -os
> >>>>>>>> dc-vmarchitect.tamu.edu:/exportdomain dh-imager01.xml
> >>>>>>>> dh-imager01_sys.qcow2: 100%
> >>>>>>>>
> >>>>>>>>
> >>>>>>>>
> >>>>>>>> [===========================================================================================================================================================================================================]D
> >>>>>>>>
> >>>>>>>> 0h00m37s
> >>>>>>>> virt-v2v: dh-imager01 configured with virtio drivers.
> >>>>>>>>
> >>>>>>>> The xml has been modified numerous times based on past
> >>>>>>>> mailing list
> >>>>>>>> comments to have VNC and Network information removed, but
> >>>>>>>> still the
> >>>>>>>> same failure.  I've attached the latest XML that was used in
> >>>>>>>> the
> >>>>>>>> log's
> >>>>>>>> failure as dh-imager01.xml.  I've also tried passing hte
> >>>>>>>> bridge
> >>>>>>>> device
> >>>>>>>> (ovirtmgmt) in the above command with same failure results.
> >>>>>>>>
> >>>>>>>> Node and Engine are both CentOS 6.2, with vdsm-4.10.0-4 and
> >>>>>>>> ovirt-engine-3.1 respectively.
> >>>>>>>>
> >>>>>>>> Please let me know what other configuration information
> >>>>>>>> could be
> >>>>>>>> helpful to debug / troubleshoot this.
> >>>>>>>>
> >>>>>>>> Are there any other methods besides a virt-v2v migration
> >>>>>>>> that can
> >>>>>>>> allow me to use my previous KVM VMs within oVirt?
> >>>>>>>>
> >>>>>>>>
> >>>>>>>> Thanks
> >>>>>>>> - Trey
> >>>>>>>>
> >>>>>>>>
> >>>>>>>>
> >>>>>>>> _______________________________________________
> >>>>>>>> Users mailing list
> >>>>>>>> Users at ovirt.org
> >>>>>>>> http://lists.ovirt.org/mailman/listinfo/users
> >>>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>
> >>>>>>
> >>>>>> --
> >>>>>> Matthew Booth, RHCA, RHCSS
> >>>>>> Red Hat Engineering, Virtualisation Team
> >>>>>>
> >>>>>> GPG ID:  D33C3490
> >>>>>> GPG FPR: 3733 612D 2D05 5458 8A8A 1600 3441 EA19 D33C 3490
> >>>>>>
> >>>>>>
> >>>>>
> >>>>> Attached is the virt-v2v generated ovf that's in my NFS export
> >>>>> domain
> >>>>>
> >>>>> Any other means to get KVM/libvirt/virt-manager based VMs into
> >>>>> oVirt?
> >>>>> Possibly something as crude as provisioning new VMs with oVirt
> >>>>> then
> >>>>> replacing the virtual hard drives?
> >>>>>
> >>>>
> >>>> this would work - just create the VM on an NFS storage domain
> >>>> with a disk
> >>>> the same size as origin, and copy over the disk you had.
> >>>> a bit trickier for iscsi, so i'd do this with nfs.
> >>>>
> >>>>> Thanks
> >>>>> - Trey
> >>>>>
> >>>>
> >>>>
> >>>
> >>> Why is it trickier with iSCSI?  Currently the only Data Center I
> >>> have
> >>> functioning in oVirt only has iSCSI storage available.
> >>>
> >>
> >> with iscsi, you will have to create the disks as pre-allocated,
> >> and use DD
> >> to overwrite them.
> >> NFS doesn't have to be pre-allocated.
> >> and since you are using pre-allocated, you need to use the RAW
> >> format iirc
> >>
> >>
> >
> > Currently most of my KVM VMs are qcow2, so converting them to raw
> > would not be a problem.  However, why is DD necessary?  Why can't I
> > overwrite the <image_name>.img with my *.img file ?  Since I've
> > used
> > mostly qcow2 in my time with KVM/libvirt I may lack some
> > understanding
> > of how to correctly handle raw images.
> 
> in both cases there aren't any .img files.
> you can convert you qcow2 to raw before copying them over to iscsi or
> nfs using qemu-img convert.
> it is not necessary, but will save you failing on small details
> between
> the two.
> using the export domain is safest, even though it doubles the amount
> of IO
> 
> >
> > Would a qcow2 image with preallocation=metadata be possible on an
> > iSCSI data store?
> 
> ayal?
> 

nope.  metadata preallocation means that each logical block has a corresponding physical block.
With files this is fine as you can seek wherever you want and the file will remain sparse.  With block devices this makes little sense as the second the guest accesses a block which is mapped to an unallocated physical block we'd have to allocate all the area up to that point.
(btw, qemu-img will fail if you try to create such an image on a block device)