Minutes of NAF User Committee meeting from 09.09.2009 ----------------------------------------------------- Present/Phone: Steve Aplin (ILC), Wolfgang Ehrenfeld (ATLAS), Andreas Gellrich (NAF), Carsten Hof (CMS), Yves Kemp (IT), Niels Meyer (ILC), Hartmut Stadie (CMS), Alexey Zhelezov (LHCb) Excused: Andreas Haupt (IT), Jan Erik Sundermann (ATLAS) 1. News from the chair Telephone conference system: The DESY telephone conference system will be replace by Meeting Place. The new phone number is 040 8998 1333 and the id is 98775 for the next meetings. NUC face-to-face meeting: After some discussion we will have the NUC face-to-face meeting on Wednesday, 11th November 2009 at 10am at DESY, Hamburg. This is before the annual Terascale Alliance workshop, which starts at 1.30pm. Every experiment has around 10 minutes for a report. It should briefly cover past activities on the NAF, plans and activities for the future and feedback of what worked, what not and what needs improvement. Topics which require or benefit from discussions with NAF experts will be collected in the October meeting in order to prepare the discussion. The presentations from the experiments will be complemented with a short perspective talk from the NAF team. Annual NAF user meeting: During the annual Terascale Alliance workshop we have up to 2 hours starting from 5.30pm on Thursday for a NAF User meeting. As last year this is a place to get direct feedback and discussion with the users. Every experiment should prepare one use case (preferred on analysis work) to start the discussion. WE will advertise the meeting to the users in due time. Report for the DESY PRC: WE and Yves Kemp were asked to prepare some pages about the NAF for the PRC review in November. WE will ask the experiments for help if necessary. 2. Status report: Status report was given by Yves. See the agenda for the report. Below a few highlights from the discussion are listed: Main topic was the migration from SL4 to SL5. So far everything went smoothly. For further discussion see action items. 3. Action items: Only items with updates, see action item list for complete list of items. - NAF SL5 migration At the moment 50% of the batch resources are migrated to SL5. CMS, ILC and LHCb are ready to move to SL5 for the batch system. ATLAS has some problems with SL5 and needs more time. After discussion it was agreed to have 20% of the batch resources running SL4 until the end of September 2009, and then only 10% until end of October 2009. If the situation for ATLAS improves very fast, this process can be speed up. The aim is to remove all SL4 batch nodes by the end of 2009. The situation for the work group server is different. CERN might change the default for their interactive system early October 2009. ILC is ready to move all work group servers to SL5 now, LHCb is also ready but needs one backup server on SL4. CMS wants to wait for a full SL5 based software release and ATLAS aims for one month after the CERN switch to sort out all SL5 problems on data management and distributed analysis software. The NAF admins can only switch the default for all experiments at the same time. This discussion will be continued at the next meeting. - batch monitoring: waiting time by experiment Plots added to web page. Item closed. - auto reply for ticket system (naf-helpdes@desy.de and naf@desy.de) Mails to naf@desy.de will trigger immediately an auto reply to the user with the assigned ticket ID. This will not be introduced for naf-helpdesk@desy.de as supporters are usually quite fast in first contact. Item closed. 4. Feedback from CMS about user directories: CMS raised the questions about the ideal storage for user files from code development and submission logs. Compile times using AFS home, AFS scratch or lustre are slightly shorter on AFS, but all solutions have some short comings. What is the best strategy? Advice from NAF needed. 5. AOB: - from CMS: Users complained that working on the work group server seems slow. Do we have work group server monitoring? Network? NAF should report next meeting. - question from ATLAS about non naf.desy.de tokens/kerberos tickets: These can not be automatically prolonged or passed on to the batch system.