Print

Print


Hi 

Thinking about the UCL "diskless" brokering jobs issue (discussed in this meeting). Even if I don't care where the data comes from (ie it doesn't have to be QMUL) - still it would be odd to broker jobs to UCL based  on the data on their storage element which may be in some random place. 
Ideally I would broker jobs based on either data at a list of sites that I choose e.g. QMUL; or on those "networkly close" ; or UK . Don't think I can do that with current atlas panda system. Maybe with future (Jedi) but I need to check. 

I think for tests I can force it though , and / or not care if it picks datasets that are supposedly at UCL and gets them from elsewhere /whereever.
cheers
Wahid

On 14 Mar 2014, at 14:44, Adam Huffman <[log in to unmask]> wrote:

> I've uploaded minutes of today's meeting to:
> 
> https://indico.cern.ch/event/308031/material/minutes/0.txt
> 
> Cheers,
> Adam
> 
> On Fri, Mar 14, 2014 at 1:14 PM, Adam Huffman <[log in to unmask]> wrote:
>> Just a reminder that we'll be holding a technical meeting today at
>> 2pm. I've created an Indico event, mostly copied from the last one:
>> 
>> https://indico.cern.ch/event/308031/
>> 
>> Cheers,
>> Adam
> 


-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.