John H, Brian, John B, Matt, Robert, Elena, Stephen, Wahid, Duncan, Jens, Govind, Chris, Sam. Apologies: Dave 0. We are still short of bloggin' effort! So far, we have a bloggin' total of 1 (Brian) for this quarter. Should we add people? Chris. Blog about upgrade experiecnes? Perfsonar and rates, network monitoring 1. Hepsysman summary? https://indico.cern.ch/conferenceDisplay.py?confId=211206 Brian reported (as in the blog post!) DPM tools in EPEL; updates are not as "instant" as before. Short talks were useful, particularly for work where we hadn't had much time to do a lot of work. Also upgrades (RHEL). Meeting was very grid oriented. People who don't like that should have offered to talk... so grid stuff is useful, or we're better at volunteering to speak :-) Chris talked about "small" VOs. Not the low level "how do we do things," more in the high level of "how should we do things." And the gotchas. WebDAV may also be the way to go. DPM has imrpoved support for WebDAV, particularly as of 1.8.4 as it moves into DMLite. LFC supports WebDAV? Could we publish WebDAV endpoints in GLUE? DPM publishes GridFTP and RFIO but not the other protocols - automatically. WebDAV and HTTP would be useful - and NFS? Haven't tested writing. And the semantics for replicating data. How much disk space is used ('du' on a filesystem) - T2K want to know - if the filesystem were mounted, it would be possible to du the filesustem - but is this a good idea? Are they interested in how much space they're using, or how much they have left? They say it's the former. DPM toolkit had a tool which summed the space used, written by Sam, but was discontinued when DPM claimed to get support elsewhere. This data may also not be published in the information systems for a shared pool. T2K is also not small any more, which would make the query heavier. Would they have their own pool, though? They could have a space token which would determine their usage (on DPM). Should the storage systems be able to track how much space is actually used eg in subdirectories? It would be worth doing some tests... what would we need. On StoRM (ie Lustre), could either run a separate WebDAV (like xroot), or run the du separately. Expected in a future release of StoRM?? 2. DMLite again... what is needed to determine how much we can support? Source is useful as "documentation", but to support it properly we'd need to be able to build it. There is a DPM workshop around ~3 Dec (first Mon&Tue) 3. Upgrade to EMI status... http://www2.ph.ed.ac.uk/~wbhimji/SRMMonitoring/ DPM unsupport from 1 Dec.; deadline for upgrading is end of Jan 2013. DPM and dCache sites ought to be upgrading. Sites will get ticketed anyway. Sites will not be "decertified" till Jan 31. Wahid set DMLite to "yes" and it didn't cause any problems. No longer publishing unset values for versions. 1.8.2 are definitely gLite; GLA and DUR are definitely EMI with fixed information systems 4. Round table - been a while since we last did this. 5. Technology watch furlong pebble ... is it time to have a look in the crystal ball? CEPH? iRODS? Error correction? Raul reports having a test server at Brunel; Sam has previously had one at Glasgow. $. GDB is today... http://indico.cern.ch/conferenceDisplay.py?confId=155074 We also need GLUE 2. http://indico.cern.ch/getFile.py/access?contribId=15&resId=1&materialId=slides&confId=155074 $+1. AOB Meeting at IC on Friday afternoon on object storage; Chris plans to go. Not sure if it is open? Organised by Dell. [10:02:05] Duncan Rand joined [10:03:11] Govind Songara joined [10:04:29] PPRC QMUL joined [10:06:22] Wahid Bhimji they are in EPEL [10:06:48] Sam Skipsey joined [10:07:40] Jens Jensen https://indico.cern.ch/conferenceDisplay.py?confId=211206 [10:08:43] Jens Jensen Alessandra's talk [10:08:44] Jens Jensen https://indico.cern.ch/getFile.py/access?contribId=2&sessionId=0&resId=3&materialId=slides&confId=211206 [10:08:48] Wahid Bhimji hmm she put [10:08:48] Wahid Bhimji GridPP storage group failure to properly distribute DPM tools since they moved [10:09:30] Wahid Bhimji I don't think its a big problem (now) - but potentially we could upload to both. To me it makes sense to get some version of the tools from the same repos as the rest of DPM [10:10:14] John Bland the agenda was way over 50% grid before I even got to look at the agenda [10:11:48] Wahid Bhimji it does [10:12:15] Wahid Bhimji lfc does too - but you have to swtich it on [10:13:37] Wahid Bhimji what did you ask me to check [10:14:05] Wahid Bhimji srm.glite.ecdf.ed.ac.uk is 1.8.4 and has both xrootd and Webdav so you can check the publishing if you want [10:21:11] Duncan Rand any vo using more than ~10 TB or so should use a space token in my opinion [10:22:44] Sam Skipsey (WRT publishing endpoints: ECDF's SE is publishing the following endpoints: rfio, gridftp, srmv1, srmv2 . There is no xrootd or https endpoint published.) [10:22:52] Sam Skipsey (Something in 1.8.5, you might hope) [10:23:56] Wahid Bhimji well lets see - I'll try it [10:24:54] Wahid Bhimji yes we can do that [10:25:27] John Bland gotta go [10:25:29] John Bland left [10:30:28] Duncan Rand Does dpm-drain work well these days? [10:31:20] Wahid Bhimji Duncan: it should do drain - but people still report problems. [10:32:11] Elena Korolkova I I was draining disk server two weeks ago, It stopped many times [10:32:22] Duncan Rand what problems [10:32:54] Elena Korolkova draining stopped [10:33:34] Elena Korolkova But I didn't put disk in RDONLY and I was draining couple of TBs [10:38:46] Jens Jensen http://www2.ph.ed.ac.uk/~wbhimji/SRMMonitoring/ [10:41:33] Wahid Bhimji I have my Tips page [10:41:33] Wahid Bhimji https://www.gridpp.ac.uk/wiki/DPMUpgradeTips [10:41:41] Wahid Bhimji if anyone wants to add their experiences [10:41:48] Jens Jensen http://www2.ph.ed.ac.uk/~wbhimji/SRMMonitoring/ [10:45:09] Brian Davies https://www.gridpp.ac.uk/wiki/UK_EMI2_Deployment