On Fri 03 Jan 2014 04:20:19 PM IST, Michal Skrivanek wrote:
On 3 Jan 2014, at 15:04, Itamar Heim wrote:
> On 01/03/2014 12:22 PM, Sven Kieske wrote:
>> Hi,
>>
>> I'm a little surprised by this development technique.
>
> its not a development technique. its a bug in upgrade or export/import of the change
to the much more powerful config file based OsInfo implementation in 3.3.
> though i thought we already fixed it.
>
> michal/roy - isn't this fixed already?
It is fixed for a long time. I think it's a TZ problem, not really osinfo.
I may not recall this correctly but I think the problematic code wasn't even
released, it was broken just for couple of weeks, that's why I'm curious what is
the exact release where it was exported to confirm it's not related to osinfo
unification of "Other".
>
>
>>
>> In my world, when you change data formats in a not compatible way
>> you should also write some sort of transition code to
>> convert the old data to the new data format for all possible
>> cases.
>>
>> And if this is not possible for some reason, at least document
>> this prominent in the release notes.
>>
>> In which version did this change occur?
>>
>> With such bad behaviour, I doubt we will ever get to something
>> like a stable release.
>>
>> I'm sorry when I missed the part of the release notes where this
>> is described and I'm happy if I'm totally wrong and just didn't
>> look good enough to find it. Please point me to some docs which
>> mention this behaviour.
>>
>> Am 03.01.2014 10:54, schrieb Patrick Hurrelmann:
>>> Hi Dan,
>>>
>>> I had the very problem myself. The fix for it is quite easy, but
>>> requires manual editing of one database table.
>>>
>>> In table "vm_static" find your non-starting vms (they propably all
have
>>> an empty string set as timezone in column "time_zone") and update
that
>>> column to null. There was a recent change in the timezone code and it
>>> now fails when the timezone is an empty string, but works fine if it null.
>>>
>>> Regards
>>> Patrick
>>>
>>
>
Not an osinfo issue, this is a bug in input validation. and here is the
fix [1]
[1]
http://gerrit.ovirt.org/22989