Experiments ----------- LHCb - Nothing to report CMS - Bristol Fedex transfers failing. Authentication issue? ATLAS - All sites 90%+. Other VOs - LIGO awaiting cvmfs - LSST at three sites GridPP DIRAC - All sites passing test jobs today apart from Lancaster Vac (upgrading) and QMUL LCG (investigating). - http://www.gridpp.ac.uk/php/gridpp-dirac-sam.php?action=view Meetings & updates ------------------ See http://www.gridpp.ac.uk/wiki/Operations_Bulletin_Latest for items and comments Additional comments: - General updates - Call to get involved in DPM, attend meetings etc (see DPM user forum) - GridPP34 agenda: Ganga, Cloud&VM, GridPP5, storage, other topics? (please suggest) - See bulletin for ARGUS with latest Java problem and workaround Found by RAL in January... - EMI merging into UMD (UMD Test). Matters since EMI gets timely updates. UMD Untested same as EMI repo now. - Tier-1 status - Network outage caused by loss of power. Also ongoing router issue. - Storage and data management - Some users having problems using Puppet. Will improve docs and default configuration files. Also manual configuration instructions just in case. Canonical versions in PuppetForge. - Getting rid of RFIO. Want to provide space accounting within DPM rather than via SRM. GridFTP redirection will be tested once beta version working. - Security - setroubleshoot advisory. Believed that no UK sites affected, since not normally installed. - OpenSSL vulnerability low risk for the way we use OpenSSL. - Tools - REBUS ignores subclusters if they have the same core-count as another one! This is the intended functionality and they refuse to fix it (to avoid counting duplicates.) https://ggus.eu/?mode=ticket_info&ticket_id=112402 Actions review -------------- * https://www.gridpp.ac.uk/wiki/Operations_Team_Action_items - Puppet updates for all VOs. Looking at writing simpler module. pre-GDB and GDB review + discussion ----------------------------------- Agenda references: * pre-GDB: http://indico.cern.ch/event/319819/ * GDB: http://indico.cern.ch/event/319745/other-view?view=standard GDB - https://twiki.cern.ch/twiki/bin/view/LCG/GDBMeetingNotes20150311 - List of actions in progress: - https://twiki.cern.ch/twiki/bin/view/LCG/GDBActionInProgress pre-GDB - Discussion with other communities: - https://twiki.cern.ch/twiki/bin/view/LCG/GDBMeetingNotes20150210 - Cloud issues: https://twiki.cern.ch/twiki/bin/view/LCG/20150310PreGDB HEPiX - quick observations -------------------------- * Pointers to areas that stood out! * Agenda: https://indico.cern.ch/event/346931/timetable/#all.detailed - Eduardo's tutorial on IPv6 particularly useful - Great talk about dust from Julien Leduc (Wed) - CERN operational talk very thorough and useful (Mon) - Amazon talk implied spot prices for unreliable, killable jobs comparable in cost to our gold-plated Tier-1 services providing a much higher level of service. AOB --- - One of the GridPP34 discussion sessions will continue the technical meeting discussion of last week. This concerned arguments for and against scaling up the GridPP cloud resources. Present: Alessandra F; Andy W; Andrew L; Andrew M (minutes); Chris B; Dan T; Daniela B; David C; Elena K; Ewan M; Federico M; Gang Q; Gareth R; Gordon S; Ian L; Jeremy C (chair); John B; John H; Kashif M; LC; Oliver S; Winnie L; Pete G; Raja N; Rob F; Robert F; Sam S and Steve J. Apologies: Tom W; Matt D. Chat window ----------- Jeremy Coles: (31/03/2015 11:02) Andrew McNab is taking minutes today. Paige Winslowe Lacesso: (11:07 AM) I beg your pardon, pls repeat? Most likely Dr Kreczko taking care of it Daniela Bauer: (11:09 AM) https://cmsweb.cern.ch/phedex/debug/Activity::ErrorInfo?tofilter=.*&fromfilter=T2_UK_SGrid_Bristol&report_code=.*&xfer_code=.*&to_pfn=.*&from_pfn=.*&log_detail=.*&log_validate=.*&.submit=Update# That's the link to the errors (If the chat window doesn't mangle it) Jeremy Coles: (11:11 AM) https://www.gridpp.ac.uk/wiki/Operations_Bulletin_Latest Paige Winslowe Lacesso: (11:11 AM) No, it's ongoing, May be my cert, the certs, the fact that yaim is used vs puppet.... Alessandra Forti: (11:12 AM) YAIM is still supported for the UI Ewan Mac Mahon: (11:13 AM) I think DPM is basically fine. Daniela Bauer: (11:16 AM) My sound just died.... Ewan Mac Mahon: (11:16 AM) it sounds OK here, so I think that must be you. Daniela Bauer: (11:16 AM) I'm pretty sure it's me. I'll duess I'll restart Vidyo yet again.. Chris Brew: (11:18 AM) I am! Daniel Peter Traynor: (11:18 AM) i am Chris Brew: (11:18 AM) it was in the testing repo Ewan Mac Mahon: (11:21 AM) I wasn't completely clear what was happening, but it did sound like it was somewhere between a real change and a rebrand. Though of course, it's still worth noting that we don't use a huge amount of either UMD/EMI at all these days. A lot of major components are elsewhere. David Crooks: (11:48 AM) http://www.scotgrid.ac.uk/graphite/ Gareth Douglas Roy: (11:56 AM) https://ggus.eu/?mode=ticket_info&ticket_id=112402 Ewan Mac Mahon: (11:56 AM) I think the suggested change would be "don't do the really stupid thing". David Crooks: (11:58 AM) I've lost audio - need to restsrt Steve Jones: (11:58 AM) Spot on, Ewan. It's embarrasing, to say the least. Ewan Mac Mahon: (11:59 AM) Or possibly aiming at a different layer of the stack - don't trust the figures generated by the tool that does the really stupid thing. Jeremy Coles: (11:59 AM) https://www.gridpp.ac.uk/w/images/5/5f/Twhyntie_DRN000024-v1-0_DIRAC-CVMFS-CERNVM_mk01.jpg David Crooks: (11:59 AM) That soudns better now Gareth Douglas Roy: (11:59 AM) Well whats annoying is they have a tool that _does_the right thing.. so why isn't there a flag that says Gstat != Rebus do something Daniela Bauer: (12:00 PM) Dirac doesn't use the WMS so there's a stray box in that diagram Ewan Mac Mahon: (12:02 PM) The whole BDII-based 'what kit do you have' thing is fundamentally misconceived anyway - it doesn't work well with modern resources. A lot of the figures are based on massaging the advertising to generate outputs from the tools that are not wildly unhelpful, as opposed to actually doing the 'right thing' according to the design. Jeremy Coles: (12:02 PM) https://twiki.cern.ch/twiki/bin/view/LCG/GDBMeetingNotes20150311 Gareth Douglas Roy: (12:04 PM) Ewan, I agree... having had a number of tickets because our Glue is not correct, usually beacuse our Batch system is so busy it was timing out and publishing the default which there was absolutely nothing we could do to fix is annoying to say the least Jeremy Coles: (12:19 PM) https://indico.cern.ch/event/346931/timetable/#all.detailed Robert Wolfgang Frank: (12:19 PM) lots of sites still use torque / maui ... Ewan Mac Mahon: (12:21 PM) On the fedcloud points, I think all one can really say to the realisations that a federated service requires a decent federated identity system, and that making up your own brand spanking new exclusive wierdy interface in a space which already has a widely supported entrenched dominant one might be a barrier to adoption is No Shit Sherlock. David Crooks: (12:26 PM) Yeah, it was very good Federico Melaccio: (12:27 PM) I agree I would recommend the computer security update as well, it was very interesting and quite scary Ewan Mac Mahon: (12:30 PM) He's not kidding about 'exploded' BTW, the photos were quite striking. Lucia Morganti, on wednesday. David Crooks: (12:31 PM) https://indico.cern.ch/event/346931/session/5/contribution/57 Ewan Mac Mahon: (12:31 PM) It was interesting - not quite in the 'news you can use' category as yet, but some fascinating stuff. Gareth Douglas Roy: (12:32 PM) https://indico.cern.ch/event/346931/session/5/contribution/57/material/slides/0.pdf Ewan Mac Mahon: (12:33 PM) The bits for the demo are apparently all on github, so it should be possible to DIY. (but you're right, we should get him to do it at GridPP) Daniela Bauer: (12:35 PM) sorry, I have to go It's past 12:30 ... Happy Easter :-) Jeremy Coles: (12:36 PM) Just letting the discussion continue! Does anyone have an AOB item? Ewan Mac Mahon: (12:38 PM) Indeed. I think if anyone asks "why aren't you running on commercial clouds more" this gives us a really solid answer to that quetion. elena korolkova: (12:38 PM) Need to go too. Happy Easter Federico Melaccio: (12:39 PM) thanks and happy Easter