ettindang: Jens, Brian, John B, Steve, Rob, Duncan, Pete, David, Chris B, Ewan, Adam, John H opaligeos: Wahid who is busy polishing Higg's medals. 0. We have six blog posts for the quarter which isn't great but it isn't bad either: three by Brian, three by Jens. Reminder to others to blog about the good stuff they're doing (if any). 1. Review of stuff for the pre-GDB next week. It seems to have two themes, data preservation and the "protocol zoo". Regarding the first, there used to be a DPHEP working group led by DESY. This group is now led by Jamie Shiers, and (as far as we know) intends to pick up its work in the context of RDA (rd-alliance.org). There is an RDA meeting coming up some time in March somewhere. Regarding the second, Brian points out there is input from the ATLAS Jamboree with relevant slides presented at the dteam meeting yesterday [1]. Future directions - do we intend to make use of object stores; how is local access to be done. Ideally we'd want common protocols for the VOs, so we don't have to support so many - but then, we (GridPP) are also supporting VOs which are not represented in the meeting, such as MICE, T2K, NA62, and SnoPlus. Two philosophies: Either the VOs use what we give them, or we decide to support them and then give them what they want. The purpose of the grid middleware was that it would be easy(ish) to support several interdisciplinary VOs on the same middleware. Small VOs also don't have the luxury of adapting their code to run on exotic middleware; large VOs can always throw a few PhD students at the problem. So the main thing is that we shouldn't forget that we still need to support non-LHC VOs, and they typically need a rich middleware with replica catalogues and brokers and stuff, as they don't have that themselves. [1]. https://indico.cern.ch/event/360123/ 2. Any major things we still need to do prior to run 2? * Upgrading DPMs or StoRMs? Brian thinks the dCaches are OK. - as a corollary, is 1.8.9 safe, and recommended? * Upgrading to SL6? At least head nodes. If anyone has not done that yet. * IPv6 (not likely on this timescale) * Space: while run1 started from empty (or near enough), run 2 starts from full sites. May need the option to delete space - but note the deletion rate may be slower than you think it should be. Also cleaning dark data, and consistency checking (with catalogues) CHAT log Chris Brew: (07/01/2015 10:05:02) mike not working Jens Jensen: (10:06 AM) http://indico.cern.ch/event/319817/