Dear Andreas,
Andreas Unterkircher wrote:
> Hi Dimitris,
>
> as Konstantin wrote, the SAM tests have been updated since nearly a week.
> Now I quickly scanned the production SAM results and found at least 30
> sites that upgraded to lcg_util 1.7.2/GFAL 1.11.4 and which pass the
> relevant SAM tests (CE-sft-lcg-*). So I see no reason to take these rpms
> out of the release.
So it is my understanding that SAM results were not distorted by this
problem. Is that correct?
But I searched the update page and I could find this issue mentioned
anywhere. My only source of information so far is the discussion
triggered by Arnau. As a site admin I feel I should be made aware, in
this particular case proably with a broadcast or even better a big red
text in the updates description section. Or maybe I did not look in the
right place, please point it to me.
Cheers,
>
> Best regards,
> Andreas
>
>
>
> On Tue, 19 May 2009, Dimitris Zilaskos wrote:
>
>> Hi,
>>
>> Maarten Litmaath wrote:
>>> Hola Arnau,
>>>
>>>>> https://savannah.cern.ch/patch/?2983
>>>>>
>>>>> R3.1/SLC4/i386: GFAL 1.11.5 and lcg_util 1.7.4
>>>>>
>>>>> Those are later versions, which only have been certified for i386!
>>>>>
>>>>> The x86_64 patch is still in certification:
>>>>>
>>>>> https://savannah.cern.ch/patch/?2984
>>>>>
>>>>> For now your safest bet would be to downgrade to the previous
>>>>> GFAL/lcg_util.
>>>> Ok, but then, why is lcg_util 1.7.2 in production repo if it does not
>>>> work?
>>> Taking them out requires a rollback, which is a last-resort measure,
>> Shouldn't breakdown of a working site (as far as SAM are concerned) be a cause
>> for last resort measures? How for example production sites are supposed to
>> adhere to SLAs when they are provided production software known to be
>> problematic with the measurement tool(SAM) used for SLA related calculations?
>>
>>> whereas there have been very few complaints so far (downgrading the WN
>>> may not be a big deal on most sites).
>> I guess because people over the years have become naturally cautious NOT to
>> update no matter the rating of the update (high i think in this case)
>> allowing some time for what bugs were missed to show up.
>>
>> I also think that wasted effort (updating, getting burned, then downgrading)
>> is a big deal, there are various support overheads involved that are beyond
>> executing some commands, like explaining to manager "why we got red light
>> there" etc.
>>
>> Only the "-t" option is broken,
>>> which is not used by everyone, and many other things were fixed.
>>
>> It is used by SAM, and I would assume that whatever SAM is testing should also
>> be the minimum of the certification progress. If -t shouldn't be used, then
>> the SAM people should become aware in order to fix it. Maybe SAM tests should
>> be also executed on the certification infrastructure.
>>
>> Cheers,
>>
>>
>>
>>
>
--
============================================================================
Dimitris Zilaskos
GridAUTH Operations Centre @ Aristotle University of Thessaloniki , Greece
Tel: +302310998988 Fax: +302310994309
http://www.grid.auth.gr
============================================================================
|