Minutes of the storage EVO meeting 11 May 2011 Present: Glasgow: Sam, David Sheffield: Elena Lancaster: Matt Liverpool: John, Rob Edinburgh: Wahid Oxford: Ewan QMUL: Chris RHUL: Govind RAL T1: Brian, Jens (chair+mins) T2K: Jon Perkin 0. Review of actions... See below. 1. Space tokens for T2K Main thing is to check whether data can be moved into a space token without moving it. Sam will test this (or lead the activity) - ACTION. It may be worth trying it to see if it works - Sam will do this if feasible (ACTION). Wahid later reported that he had contact J.-P. but had no response yet; Sam will also try contacting J.-P. Jon Perkin joined towards the end of the meeting, by when we had finished the agenda and could go on to a general discussion with T2K. Matt suggested whether it was useful to have a test space token: it may be, but this is not essential. It is important for T2K to not have to rename files (at the SURL level) in the LFC. There is no specific requirement for how much space should go into the token: in fact T2K have a lot of data already, but some of it turns out to be old processing data - Jon now has a script for cleaning it up. What are the site pledges to T2K? Ewan points out that small VOs are best served in a shared pool, so they can make opportunistic use of the capacity - otherwise if they do not use the space inside the token, they will leave empty space. OTOH some small VOs are not so small any more - eg T2K. The upshot of this is that we will set up a single space token - eg T2KDATA - for T2K at all supporting GridPP sites, including the T1, once we have finished the DPM testing. (IC, QM, Lanc., Liv., Oxf., Shef.) Apparently jobs are not running at Oxford due to problems installing software. 2. Update on metrics Final versions of the metrics should be no surprise to anyone, as we have discussed them several times, including at the http://storage.esc.rl.ac.uk/weekly/20110504-minutes.txt These now have been sent to Pete - Jens will send to the list. The new GridPP4 metrics are: 1. Number of VO requested changes not implemented in a timely fashion across whole infrastructure (1wk for small changes, 3wks for medium size, 6wks for large.) - any GridPP VO TARGET 0 2. Number of experiments whose data transfer rates from/to RAL to/from T2s do not meet their (realistic) requirements TARGET 0 3. FTS data transfer success rate TARGET 85% (Was 70% in GridPP) 4. Conference paper (CHEP, AHM, or similar (or better)) produced each year describing GridPP developments and innovations in data management and storage area TARGET 4/year 5. Engage with storage and data management experts within WLCG/EGI/EMI/NGS or similar, or industry, to reinforce GridPP's recognised competence in this area (talks given, meetings, virtual meetings) - but this will include notable email discussions, eg ones with a agenda and outcome. TARGET 4/Quarter 6. Number of incidents not resolved within 1wk after being reported on list TARGET 0 7. Number of blog posts TARGET 8 Note that we are failing the blog entries for this quarter, unless people start posting soon - of course we should only post useful and relevant stuff. 2a. Speaking of meetings (as a metric), Wahid had attended the GDB. So far it was mostly about information services which Flavia is no longer doing, it is now one Lorenzo Disi. http://indico.cern.ch/conferenceDisplay.py?confId=106644 3. Impending meetings and relevant conferences: Who is going? Should we publish something? HEPSYSMAN: http://hepwww.rl.ac.uk/sysman/ This will be at RAL at the end of June, beg July. All hands (deadline 23rd): much cloud stuff. NeISS study very nice because the CFP specifically talks about social science. Hopefully, they will also extend the deadline. WLCG workshop @ DESY https://indico.desy.de/conferenceTimeTable.py?confId=4019#20110711 While the T1 has not yet clarified who can travel to it, it turns out that the PMB has not clarified. 4. Tribal events - CMS at CERN? There is indeed a CMS event but the focus is on T1, not something relevant to us. 5. AOB HEPiX report next week, from Martin Bly (T1). Jens is out next week, and MAY not be able to connect. 427 19/01/2011 Send Areca stuff to Sam Ewan High Open Let's close this. 430 19/01/2011 Test the rest of Chris's StoRM (cf 425) ALL Med Open And this. We discussed it, but there is probably not much else to do. 431 27/04/2011 Contact NeISS Sam Med Open Done. 432 27/04/2011 Find out who is currently pheno and get in touch Jens Med Open Half done - still need to get back to them. 433 27/04/2011 Contact T2K and see if talking to us a month ago helped Jens Med Open Elena did this - see discussion below. 434 04/05/2011 Volunteer Ian or Martin to report on storage at HEPiX Jens Med Open Done - Martin will speak at next week's storage meeting. 435 04/05/2011 Kickstart cloud storage plan in wiki Jens Med Open Done, see Cloud Storage Plan or something in the wiki. [09:55:48] Elena Korolkova Good morning, Sam [09:57:14] Sam Skipsey Morning. [09:59:24] Elena Korolkova joined [09:59:25] Sam Skipsey joined [09:59:26] John Bland joined [10:00:01] David Crooks joined [10:00:09] Elena Korolkova Morning [10:00:42] Rob Fay joined [10:01:16] Brian Davies joined [10:01:19] Queen Mary, U London London, U.K. joined [10:01:55] Matthew Doidge joined [10:02:04] Queen Mary, U London London, U.K. I can hear you [10:02:26] Queen Mary, U London London, U.K. left [10:02:39] Queen Mary, U London London, U.K. joined [10:02:45] Elena Korolkova We are supporting mice [10:03:03] Ewan Mac Mahon joined [10:03:53] Sam Skipsey NEISS, Elena, not Mice. [10:04:44] Govind Songara joined [10:05:23] Elena Korolkova /dpm/shef.ac.uk/home/t2k.org/nd280//raw/ND280/ND280/00007000_00007999/ [10:06:25] Elena Korolkova QM also has t2k data [10:06:36] Elena Korolkova and they are storm [10:09:09] Queen Mary, U London London, U.K. QMUL has around 60TB of T2k data [10:10:28] Wahid Bhimji joined [10:10:47] Wahid Bhimji sorry to be late - I was in the gdb on the srm v bdii option [10:11:01] Wahid Bhimji I meant "issue" not option [10:14:00] Matthew Doidge We'd like our t2k data tokenised too. [10:16:39] Elena Korolkova t2k has data [10:16:58] Elena Korolkova so it's not negligibale for t2k [10:19:34] Queen Mary, U London London, U.K. StoRM seems to treat spacetokens as QUOTAs [10:19:38] Ewan Mac Mahon As far as Oxford goes, I'm quite happy to give them some space tokened space. Within reason. [10:20:01] Elena Korolkova t2k should come with a name [10:20:24] Ewan Mac Mahon Probably 'T2KxxxDISK' [10:21:00] Elena Korolkova should we keep the name from the path /dpm/shef.ac.uk/home/t2k.org/nd280//raw/ND280/ND280? [10:21:54] Ewan Mac Mahon Wasn't there something about ND280 not being the whole of t2k though? [10:22:05] Sam Skipsey Well, ND280 is very specific to one detector. [10:22:12] Ewan Mac Mahon Do we want a generic t2k space token, or an ND280 one? [10:22:18] Elena Korolkova it there are data from nd280 and SK [10:22:47] Ewan Mac Mahon It might make sense to space token them separately, or it may not. I think that really would require VO input. [10:22:52] Elena Korolkova SuperK data I think is not available for all EU sites [10:23:10] Sam Skipsey Right, so I think it needs to go back to T2K for discussion. [10:25:32] Queen Mary, U London London, U.K. Can you post your URL please [10:26:47] Wahid Bhimji http://www2.ph.ed.ac.uk/~wbhimji/SRMMonitoringAtlas/ [10:27:00] Wahid Bhimji http://bourricot.cern.ch/dq2/accounting/bdii_vs_srm/ [10:27:10] Jonathan Perkin joined [10:28:09] Jonathan Perkin Hi all, sorry for my lateness [10:28:27] Jonathan Perkin Just reading back thru some of the previous comments [10:33:08] Elena Korolkova left [10:33:37] Elena Korolkova joined [10:36:41] Wahid Bhimji ah yes thats true [10:36:52] Govind Songara left [10:36:57] Wahid Bhimji you may want more than one later.... [10:37:02] Queen Mary, U London London, U.K. http://gstat-prod.cern.ch/gstat/site/UKI-LT2-QMUL/ will give you the amount of t2k data at QMUL [10:42:24] Ewan Mac Mahon I think the best way to do this is to have a wrapper around lcg-cr that your user jobs use that will then do the right thing. [10:42:35] Ewan Mac Mahon Which is (more or less) what ATLAS do. [10:47:06] Matthew Doidge itseems okay for lancaster [10:47:39] Ewan Mac Mahon Really? [10:47:45] Ewan Mac Mahon We should fix that. 10:54:20] Jonathan Perkin Thanks eveyone [10:54:23] David Crooks left [10:54:25] John Bland left [10:54:26] Rob Fay left [10:54:26] Elena Korolkova left