[Users] Missing VM disk images after yum update (addressed but not solved)

[Ricky Schneberger]

On 2012-08-15 19:52, Itamar Heim wrote:
> On 08/15/2012 04:44 PM, Ricky Schneberger wrote:
>> After my last "yum update" I got some strange errors in my cluster.
>>
>> Today when I tried to start a VM i got a pop-up that says:
>>
>> " Error:
>> Cannot run VM without at least one bootable disk.
>> Alternatives:
>> -Create a disk for this VM, and rerun the VM.
>> -Change the boot sequence using the Edit VM command (Boot Option
>> Sub-Tab).
>> -Use the Run-Once command to select a different boot option and rerun
>> the VM."
>>
>> When I went to the Virtual disks tab there was no disk left. On the
>> snapshots tab the still exist, and some of them seems even to be "OK"
>>
>> I then check my NFS-storage to se if they still exist even there, and
>> they was (Puh!)!
>> I tried to mount the images with kpartx and everything went fine.
>> So far so good. The data is still there. GOOD!
>>
>> But, how can I get things together again? And....what happened when I
>> did the yum update?
>>
>>
>>
>>
>> _______________________________________________
>> Users mailing list
>> Users at ovirt.org
>> http://lists.ovirt.org/mailman/listinfo/users
>>
> 
> which server did you upgrade (engine or host)?
> which rpms where updated?
> for engine, did you run yum update or engine-upgrade?

Well, I must end this question with my "solution" of the problem.

As some of my WMs didn’t start after a "yum update" with strange errors
in the logs I did as follow, as my prime goal was to reuse the disk images.

1. In DB-table "vm-static" I search out the key "vm-guid" for the
virtual machine

2. in DB-table "vm-device" I match my "vm-guid-ID" against field "vm-Id"
and row "disk" in the field "device". From that I got the disk-ID for
that WM.

3. After that I got to my NFS-storage and rsynced the disk image and the
META-file to a backup folder.

4. I check that it was possible to mount the image with kpartx

5. Then I just erased the WM from ovirt and checked that everything went
away.

6. A new machine with a new disk (equal size as the old one) was created

7. I copied the name of the new created disk-image including the
META-file and then deleted it.

8. After that I copied the old disk image to the right storage pool on
the NFS-server and renamed it with the earlier copied name.

9. Finally I start the newly created VM with the old disk  and
everything went up.

//Ricky
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0xB88C0B63.asc
Type: application/pgp-keys
Size: 8073 bytes
Desc: not available
URL: <http://lists.ovirt.org/pipermail/users/attachments/20120827/c3d651b3/attachment-0001.bin>