Hello
At March GDP Ale diGi gave a talk on ATLAS activities: h.ttps://indico.cern.ch/getFile.py/access?contribId=3&sessionId=5&resId=1&materialId=slides&confId=155066
At slide 5 you can see that new reconstruction jobs require vmem > 3.5GB.
ATLAS ADC team wants to understand what CPU capacity is available to run reconstruction jobs consuming>3.5 GB vmem and how many sites have problems to run these jobs.
Atlas is asking for the limits on vmem of 4GB. SWAP should be set to 2*RAM size. Physical memory requirement are not changed (2 GB).
ATLAS has assure that there is no problem for the sites with physical memory 2GB and no vmem limit.
Cloud support teams were asked to fill the table with information for each site on
Number of jobs slots
HS06 (for these number of job slots)
RAM/Slot
VMem/Slot
Kill on overusage
% use by ATLAS
Timescale
For example, your site has 100 jobs slots and 40 jobs slots have vmem >3.5 and 40 job slots have vmem >4 GB you should put three entries in the table (the third one for jobs with vmem < 3.5 GB) .
IC, Brunel (CMS sites) and Durham do not run reconstruction jobs. They do not need to fill the table.
Could you please provide this information by Monday noon.
Thanks
Elena
__________________________________________________
Dr Elena Korolkova
Email: [log in to unmask]
Tel.: +44 (0)114 2223553
Fax: +44 (0)114 2223555
Department of Physics and Astronomy
University of Sheffield
Sheffield, S3 7RH, United Kingdom
|