Hi *
It is good that you guys are discussing this here now. We just had a discussion in the WLCG Grid Deployment Board on how important it is to get the validation of EMI-2 and/or UMD worker node distributions done. Site participation and sharing of knowledge is important.
http://indico.cern.ch/conferenceDisplay.py?confId=155072
Talk by Markus Schulz (the talk labelled GDBglite32.... )
JT
On Sep 12, 2012, at 15:31 , Christoph Wissing wrote:
> Hi Mario,
>
> On 09/12/2012 03:22 PM, Mario Kadastik wrote:
>>>> This is not part of 1.13.0.
>>>> Nevertheless, the fix of this problem is already applied on src and will
>>>> be part of the next Update 1.14.0 .
>>>>
>>>
>>> OK. I think it would be really helpful if we could have that update
>>> out quite soon - even as a very small update with just that bug fix
>>> in. We've got some small deployments of the EMI 2 WN in production
>>> in the UK, and the effects of this bug make it rather tricky to tell
>>> whether there are any other problems, particularly with VO software.
>>
>>
>> We have a large installation (ca 4000+ cores) with EMI-2.2 installed in Estonia. The bug did affect us and to get past it we made a small fix setting one environment variable for OPS tests. However for us the only two VO's are CMS and OPS and CMS doesn't use lcg-cr etc so the bug doesn't affect it and the environment variable fix was only to make ops tests happy. If you have VO's that use lcg-cr, then of course you're screwed.
>
> As far as I know CMS uses lcg-cp within CRAB2 for staging out user files. I did not look into the details of the bug. Almost for sure it does not apply for the CRAB use case, which neither uses the BDII nor the LFC together with lcg-cp.
>
> Some testing of CRAB at the Estonia site was successful at least.
>
> Cheers, Christoph
>
>
>
>> Mario Kadastik, PhD
>> Researcher
>>
>> ---
>> "Physics is like sex, sure it may have practical reasons, but that's not why we do it"
>> -- Richard P. Feynman
|