And at the GDB even Michel was compalining about the torque problem. We
have been discussing about this for weeks. Tiziana said she would send a
new broadcast making it clear that if there is a valid reason sites can
postpone the CE upgrade. And Simone commented about that with relieve too.
Also you have put your face with the GDB to upgrade by the end of
October all CEs and I promised the atlas computing coordinator we
wouldn't be disruptive.
Do you want to start an internal war on how to do it now?
cheers
alessandra
On 16/10/2012 07:39, Alessandra Forti wrote:
> Hi Jeremy,
>
> were we at the same Tuesday meeting? I said clearly we had problems
>
> 1) wth the fact that the best way is to upgrade the CE with the WNs
> because of torque,
> b) that I was having a problem with the maui/torque configuration (now
> solved but not a week ago),
> c) with the fact that we need to integrate other hardware work in the
> WNs.
>
> We also discussed in a private thread other sites that might have a
> problem upgrading by the end of October because of torque or manpower
> and we would have to argue with EGI if necessary, this was before the
> GDB.
>
> All of you keep on coming back at me as if I wanted to delay let's be
> clear that if I hadn't done most of the background communication with
> atlas we would still be talking about March also for the WN and quite
> few sites would have to do disruptive double work to upgrade before
> torque and then the middleware because of how well this was handled by
> all the parts.
>
> cheers
> alessandra
>
>
> On 16/10/2012 02:55, Jeremy Coles wrote:
>> Hi Alessandra/All
>>
>>> At the last GDB it was decided, and Tiziana agreed, the cream CE can
>>> be postponed until 30 November if the site has a clear plan to
>>> upgrade it with the WNs this is to avoid chaos and help sites with
>>> the torque problem.
>> This is not quite how I would word it. All sites are *expected* to
>> move to EMI CEs by the end of October. From the start of November
>> sites will receive tickets indicating a possible suspension if they
>> still run a gLite 3.2 CE at that time. If a site responds to the
>> ticket with a clear plan to upgrade in the following weeks or a solid
>> technical reason as to why they cannot, then the grace period for
>> that site *may* be extended. In addition, if any unsupported
>> middleware is found to be vulnerable the site will have 24hrs to
>> remove it. As an NGI our own plans in the UK ought to be stricter
>> than the last possible option offered by EGI/WLCG if there is no
>> technical problem stopping us bringing things forward, or rather
>> pushing them back again.
>>
>> At last Tuesday's (9th) ops meeting I went round each site asking in
>> turn about site plans for moving to EMI CEs and rechecked at the end
>> my understanding that no site would have a problem with the end of
>> October timescale. Since nobody came back at that point I would be
>> reluctant to now say we can set a new deadline a month later just
>> because the GDB discussions presented a possible extension. The UK is
>> already trailing the curve on CE upgrades.
>>
>> Jeremy
>>
>>
>>
>>
>>
>> On 15 Oct 2012, at 20:45, Alessandra Forti wrote:
>>
>>> To be fair the first broadcast for the retirement of glite services
>>> was in March. The first extension to allow more testing to the VOs
>>> for WNs and UIs was in March. You can look at the broadcasts
>>> archives looking for Tiziana. The 25/9/2012 is the first broadcast
>>> where they threatened sites with suspension I think.
>>>
>>> https://operations-portal.egi.eu/broadcast/archive
>>>
>>> but here are some links
>>>
>>> End of security updates in February 6th
>>> https://operations-portal.egi.eu/broadcast/archive/id/595
>>>
>>> extension for WN and UI March 15th
>>> https://operations-portal.egi.eu/broadcast/archive/id/627
>>>
>>> glite decommissioning plan march 15th
>>> https://operations-portal.egi.eu/broadcast/archive/id/628
>>>
>>> there are other two on August 31st
>>> https://operations-portal.egi.eu/broadcast/archive/id/723
>>>
>>> and September 3rd
>>> https://operations-portal.egi.eu/broadcast/archive/id/725
>>>
>>> Let's say there were black holes sucking communication out of every
>>> single interaction between the parties......
>>>
>>> At the last GDB it was decided, and Tiziana agreed, the cream CE can
>>> be postponed until 30 November if the site has a clear plan to
>>> upgrade it with the WNs this is to avoid chaos and help sites with
>>> the torque problem. Tickets will still be opened anyway and need to
>>> be answered with the plan.
>>>
>>> cheers
>>> alessandra
>>>
>>>
>>> On 15/10/2012 20:04, Stephen Burke wrote:
>>>> Testbed Support for GridPP member institutes [mailto:TB-
>>>>> [log in to unmask]] On Behalf Of Ewan MacMahon said:
>>>>> The trouble with that, and with the more detailed gLite 3.2 specific
>>>>> page at:
>>>>> http://glite.cern.ch/R3.2/
>>>>> is that it lists the gLite cream as being unsupported since the
>>>>> end of
>>>>> April. Now, that may very well be true, but it hasn't been
>>>>> particularly
>>>>> sensible to push the upgrade to EMI since it's been buggy (and
>>>>> indeed,
>>>>> still is).
>>>> ? Maybe I'm misremembering but I think people have been installing
>>>> the EMI 1 CREAM for a long time - the EMI *2* CREAM has had some
>>>> problems but that's a different matter. Bear in mind that EGEE
>>>> ended in spring 2010 ...
>>>>
>>>>> In any case, the end-of-support dates there are clearly not
>>>>> the same thing as the must-upgrade-by dates that we're talking about
>>>>> now.
>>>> Well, that's something that has changed - the WMS vulnerabilities
>>>> concentrated minds. In principle people should not have been
>>>> running services with no security support but no-one was pushing it
>>>> until it became a live issue. I think the suggestion from now on is
>>>> that sites will only have a month beyond the end of support.
>>>>
>>>>> Well, I'm not so sure. We've been talking about it in a lot of
>>>>> meetings
>>>>> in terms of the September deadline relating to upgrading away from
>>>>> gLite 3.1 components, and the end of October for migrating away from
>>>>> gLite 3.2 bits (except the extended support ones). That's either
>>>>> changed, or it was never true in the first place.
>>>> I think the latter. It is true that the message has been a bit
>>>> softer for glite 3.2 - the suggestion was that sites running 3.1
>>>> would definitely be suspended on 1st November whereas sites with
>>>> 3.2 services would get some additional wiggle room, but the basic
>>>> deadline was the same.
>>>>
>>>>> Either way, there's
>>>>> been a lack of well documented clarity, and the broadcast of the 25th
>>>>> is still, I think, the first occasion that some of this was written
>>>>> down and properly disseminated.
>>>> EGI OMB minutes from 17th July:
>>>>
>>>> http://indico.egi.eu/indico/getFile.py/access?resId=0&materialId=minutes&confId=723
>>>>
>>>>
>>>> "DECISION. The OMB approves the decommissioning of gLite 3.2
>>>> unsupported products by 30 Sep 2012."
>>>>
>>>> Stephen
>>>>
>>>
>>> --
>>> Facts aren't facts if they come from the wrong people. (Paul Krugman)
>
>
--
Facts aren't facts if they come from the wrong people. (Paul Krugman)
|