Attending: Sam, Dan, Jens (chair+mins), RobC, Vip, Winnie, Matt, Brian 0. OBP Unfortunately Tom wasn't available today after all; he got the dates wrong, so the advertised CEPH day report will have to wait till next week. On a related note, next week also has a GDB which (along with its concomitant pre) is mostly about benchmarking; however, Oliver has a slot on the Wednesday about the site storage survey - with "results, conclusions, actions", so another thing to look forward to. https://indico.cern.ch/event/739883/ 1. IanJ very kindly notified me that the writeup from the Rucio workshop in Sept. has been, er, written up. (cf notes from 20190918) https://indico.cern.ch/event/844100/timetable/?view=standard Included is a link to the "Rucio coding camp" which is in about a fortnight. We believe IanJ (RAL) is going, and JamesP (Ed) 2. Catching up on backlog instead, what happened to any, all, or none of (a) xroot upgrades for ATLAS? (b) proxy cache monitoring? (c) token-based authorisation (anyone experiments using?) (d) DPM sites' DOME upgrades, (e) FTS race condition (did it go away? was it in FTS or layered above?), (f) DPM checksums ("fixed in 1.13") OK, so many of these problems are related, and for DPM sites are fixed or improved in DPM 1.13. Matt reports that he'd already upgraded (these days version numbers are not published! which is unhelpful), and the imminent release is 1.13.2. Generally upgrading is safe, but Matt had to implement a hack to get the BDII back - the BDII is needed for DIRAC (or a manual config in DIRAC is required, which Daniela had previously made but it got lost when something was moved at Imperial). Apparently protocols are now xroot@something, so not understood by the scripts, and dpm-listspaces is already 2007 lines of not uncomplicated python, with lots of hacks added over time. Matt's solution was to publish a dummy xroot server and it fixed the BDII problem. No one outside of the US is believed to currently using authorisation tokens in production, except for ALICE who've done their own thing for a while, but the SciTokens or Macaroons - which are the standard types of tokens - are not used by any experiment in production in the UK. So we'll need to revisit this as they roll out - and possibly support the rollout/migration, as appropriate. As regards FTS, Brian informs us that the production instance is 3.9.1 but does not have the TPC modules installed for xroot or http; and the dev instance is busy being used in production by SKA and AENEAS. The current plan is that the latter will finish using the dev instance by the end of 2019, and in 2020 it can then be repurposed to exercise the TPC modules. Bottom line: safe to upgrade to 1.13 but keep an eye on the BDII if you support DIRAC. 3. AOB NOB From Me to Everyone: 10:03 AM https://indico.cern.ch/event/844100/timetable/?view=standard From Sam Skipsey to Everyone: 10:07 AM https://projectescape.eu From Brian Davies @UKRI-STFC-RAL to Everyone: 10:16 AM Accounting? bdii?apel/epel?