Anyways i've commited this, hope it help a bit
https://abf.io/openmandriva/iso-build-tools/commit/0c8c30cd351f81484dcd587c40fd8aef459aac9d


2014-04-17 11:52 GMT+02:00 Tomasz Gajc <[email protected]>:

> Well i first build this one today:
>
> https://abf.io/platforms/openmandriva2014.0/products/82/product_build_lists/3339
> and when i saw that everything is fine with adding repos to urpmi, i've
> started normal builds,and rest story you know.
>
> I'll try again, build KDE isos for both archs, but i have feeling it will
> fail again. Today we have a release day, and it will be really WEAK if we
> release ISO without default urpmi medias set.
>
> 2014-04-17 11:47 GMT+02:00 Александр Бурмашев <[email protected]>:
>
> Thx, now i see it.
>> Is the problem still reproducible with the current builds ? Maybe build
>> happened right at the momnet of metadata being synced ?
>>
>>
>> 2014-04-17 13:39 GMT+04:00 Tomasz Gajc <[email protected]>:
>>
>> Alexander i'm speaking about
>>> https://abf.io/platforms/openmandriva2014.0/products/71/product_build_lists/3341
>>>
>>> ## Installing default urpmi repositories for end user ##
>>>
>>> adding medium "contrib updates"
>>>
>>>     $MIRRORLIST: media/main/release/media_info/synthesis.hdlist.cz
>>>
>>>
>>>     $MIRRORLIST: media/main/release/media_info/info.xml.lzma
>>>
>>>
>>> retrieval of 
>>> [http://mirror.yandex.ru/openmandriva/openmandriva2014.0/repository/i586/media/main/release/media_info/info.xml.lzma]
>>>  failed (md5sum mismatch)
>>>
>>> nothing to update (use urpmi.addmedia to add a media)
>>>
>>>
>>> When we use http://abf-downloads.rosalinux.ru/ as a master for urpmi
>>> repos, then above issue does not occur.
>>> I've checked anf info.xml.lzma md5sums are equall on both locations
>>> (abf, yandex).
>>>
>>>
>>>
>>>
>>> 2014-04-17 10:52 GMT+02:00 Александр Бурмашев <[email protected]>
>>> :
>>>
>>> Checked
>>>> https://abf.io/platforms/openmandriva2014.0/products/70/product_build_lists/3332
>>>> iso md5sum matches md5 file.
>>>> Integrated image check does not work, but the reason is not with iso
>>>> having a wrong md5 sum.
>>>> Starting it from debug shell manually produces PASS. So i think there
>>>> is something wrong with how it is started ( if started at all ).
>>>>
>>>>
>>>> 2014-04-17 12:08 GMT+04:00 Александр Бурмашев <[email protected]
>>>> >:
>>>>
>>>> Hello, what should be checked ? How md5 mismatch situation can be
>>>>> reproduced ?
>>>>>
>>>>>
>>>>> 2014-04-17 10:44 GMT+04:00 Tomasz Gajc <[email protected]>:
>>>>>
>>>>> Bero you are hero !
>>>>>>
>>>>>> I've checked media_info files from abf and a local mirror yandex.ruand 
>>>>>> md5sum/sha1sum were same for each file. So either something got changed
>>>>>> or this is really ABF issue. Denis, Alexander can you guys help us with
>>>>>> this issue ?
>>>>>>
>>>>>>
>>>>>> 2014-04-17 3:21 GMT+02:00 Bernhard Rosenkränzer <[email protected]>:
>>>>>>
>>>>>> Hi,
>>>>>>> the "md5sum mismatch" problems while building ISOs in ABF (not
>>>>>>> reproducable locally) persisted all day, so we didn't get to build the
>>>>>>> final rc iso.
>>>>>>> In the end, I've ended up disabling mirrorlist and just using the
>>>>>>> master server as a test - this does get around the mismatch error, but
>>>>>>> obviously isn't what we should be doing, and there's still a suspicious
>>>>>>> wget error in the build log.
>>>>>>>
>>>>>>> This is the resulting iso:
>>>>>>> https://abf.io/platforms/openmandriva2014.0/products/
>>>>>>> 70/product_build_lists/3332
>>>>>>>
>>>>>>> I'm too tired to wait for its download (3:20 am here), but please
>>>>>>> test it -- we need to know whether or not that workaround fixes things 
>>>>>>> at
>>>>>>> all (there may be some other problem, e.g. with firewalls on abf 
>>>>>>> machines
>>>>>>> not allowing any sort of access we need anymore).
>>>>>>>
>>>>>>> ttyl
>>>>>>> bero
>>>>>>>
>>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>
>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>>
>>>
>>
>>
>>
>>
>


Reply via email to