r2 - 22 Aug 2011 - 13:05:57 - RichardMountYou are here: TWiki >  AtlasSoftware Web > Minutes19Aug2011

Minutes19Aug2011 RAC Minutes, August 19, 2011 - DRAFT

Members (*=present, #=apologies)

*Richard Mount, *Jim Cochran (Analysis Support Group Chair), Alexei Klimentov (ATLAS ADC), Ian Hinchliffe (Physics Advisor), Rik Yoshida (Analysis Support Group Deputy), *Michael Ernst (U.S. Facilities Manager), #Rob Gardner (Integration Coordinator), *Kaushik De, (U.S. Operations Manager), #Armen Vartapetian (U.S. Operations Deputy)

Ex-Officio: Torre Wenaus, Stephane Willocq, Mike Tuts, Howard Gordon

Approval or Correction of Minutes

Minutes of previous meetings were approved.

Operations Report (Kaushik)

Production fell to the floor about a week ago and is still there. Kaushik (just back from vacation) had not had time to contact Borut and find out why, and when it would start again. Michael said the BNL had been heavily used by production and reprocessing.

Analysis was continuing at the usual, somewhat sporadic level.

Richard agreed to email usatlas-l to remind people that additional production was most welcome.

With respect to storage, the Tier 2s were fine - plenty of headroom. At BNL space was tighter.

Michael confirmed that DATADISK was running close to its maximum of 5 PB. Some cleaning, largely removal of dark data, should allow BNL to continue for the immediate future. 3.8 PB of the BNL DATADISK was occupied by "primary" data. Only 0,75 PB was secondary. A new procurement was being planned using funds already available. Michael stressed - without disagreement from anyone present - that no site could operate up to the limit of available space.

LocalGroupDisk (Richard)

Richard promised to draft an email to usatlas-l encouraging the use of LOCALGROUPDISK. The essential messages were that usage was encouraged so that policies for LOCALGROUPDISK could be developed, the types of usage could be understood and the scale of future needs could be planned. A page withing the RAC Twiki would be set up to reflect the ongoing development of policies and would be linked in the email. The RAC would be given a few days to comment on the draft.

The request from Doug Benjamin for 15 GB of LOCALGROUPDISK space at each Tier-2 site for use by T3s was briefly discussed and approved. It was pointed out that this request fell well below the threshold for automatic approval when requests were submitted by DaTRI.

US Analysis Job Priority

A mechanism had been implemented some time ago that would augment the priority in the US of analysis jobs submitted with an OSG certificate to reflect the substantial unpledged US CPU resources available for analysis. This mechanism resulted in jobs appearing in "localpool". See http://panda.cern.ch/server/pandamon/query?jobsummary=cloud&select=analysis&hours=12&specialHandling=localpool. Relatively few jobs were being directed to localpool. Alden was volunteered to look into this to see if the algorithm was working as designed.

Kaushik noted that, in the current data distribution scheme, the US did not host a full copy of many datasets (e.g. AOD). Thus only a fraction of a large US-originated analysis task could be accelerated by the localpool mechanism. There was no provision in PD2P to copy datasets to the US to overcome this issue. Also, PD2P was not currently allowed to write to upledged disk resources, so even if the US offered unpledged disk space to hold the in-demand copies this would not help.


Jim has teaching at the RAC meeting time every second week. This will be taken into account as a weak constraint on RAC meeting scheduling.

Action Items

  1. 5/6/2011: Kaushik: Let Alden know he has been vounteered to investigate the localpool mechanism - New.
  2. 5/6/2011: Richard: Draft an email encouraging the use of LOCALGROUPDISK - New.
  3. 5/6/2011: Richard: Draft a web page giving an overview of LOCALGROUPDISK availbility, policies, monitoring etc. - New.
  4. 5/6/2011: Richard: Draft a web page explaining how users should exploit the various SpaceTokens - Ongoing will be complete with the additon of the LOCALGROUPDISK description.
  5. 9/24/2020: Kaushik: Organize the nature and timing of the effort to create an automated mechanism to give US-requested additional production priority access to US non-pledged resources. On hold - manual system works just fine for now

About This Site

Please note that this site is a content mirror of the BNL US ATLAS TWiki. To edit the content of this page, click the Edit this page button at the top of the page and log in with your US ATLAS computing account name and password.


Powered by TWiki
This site is powered by the TWiki collaboration platformCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback