[Users] Restoring guest from export domain fails

[Karli Sjöberg]

21 okt 2012 kl. 08.54 skrev Itamar Heim:

> On 09/12/2012 03:33 PM, Karli Sjöberg wrote:
>> 
>> 12 sep 2012 kl. 13.51 skrev Eli Mesika:
>> 
>>> 
>>> 
>>> ----- Original Message -----
>>>> From: "Oved Ourfalli" <ovedo at redhat.com <mailto:ovedo at redhat.com>>
>>>> To: "Karli Sjöberg" <Karli.Sjoberg at slu.se <mailto:Karli.Sjoberg at slu.se>>
>>>> Cc: "users" <users at ovirt.org <mailto:users at ovirt.org>>
>>>> Sent: Wednesday, September 12, 2012 1:37:27 PM
>>>> Subject: Re: [Users] Restoring guest from export domain fails
>>>> 
>>>> Hi Karli,
>>>> 
>>>> Does it also happen in the latest upstream?
>>>> I looked at the call stack, and saw that this code was changed
>>>> recently, in commit 37a8bf354b9c9d348ccbee4ce60bd62d89a37198.
>>> 
>>> Yes, I saw that, but it seems that  auditInvalidInterfaces that throws
>>> the exception was also changed in that commit so I suspect that the
>>> commit change caused a regression and not that before it there was a
>>> bug and it fixed that ...
>> 
>> Yeah sure, I can report a bug for that, but how/when will/can this be
>> fixed for us? I mean, I am using oVirt-3.1.0-2 from the oVirt 3.1
>> release repo. Do I wait until there is an update for this issue, or can
>> you suggest something else I could try? Because without the guests being
>> imported, our oVirt environment is empty, and there are customers
>> waiting to start using their guests again.
>> 
>> And how exactly do I file this report? (First timer)
> 
> Hi Karli,
> 
> did you get this resolved?

I never got any instructions on how to file a report so I never did, but thanks to findings made by a Mr Thomas Scofield, I managed to resolve the issue on my own. Here´s a copy of the reply I made after resolving the issue:

"Thank you so much for catching this, it made my life a bunch more easier!

When we wanted to upgrade from oVirt-3.0 to 3.1, we followed the procedure of exporting all of the machines to export domain, crashed the entire system and started building it up again. When it was up and running, we tried importing the machines back, they all failed and we couldn´t figure anything out from just that non-saying error message and cryptic errors in engine log. But then I remembered reading something a while back about this, searched back and compared our error messages, and they matched. So I then dove into the storage and changed all of the references from the old name to the new, and importing was all smooth sailing from there:
# sed -i '' 's/oldnetname/newnetname/g' master/vms/*/*.ovf

Devs: I can´t remember, is this explained anywhere in the upgrading documents, that the bridge names need to be the same before and after, or that you need to modify that afterwards if they´ve changed?"

And thank you Itamar, just for caring:)

> 
> Thanks,
>   Itamar