Minutes of storage group EVO meeting, 1 Aug 2007 First time in EVO. Present: Edinburgh: Greig Lancaster: Matt RAL Tier 1: Derek RAL Storage: Jens (chair+mins) Apologies: DESY: Owen (apology lasts till next week, as I recall). Glasgow: Graeme (till this week only IIRC) 0. Review of actions (see below) 1. Filesystems revisited, including distributed ones. Two sides to this question. First one is best practice for the care and feeding of your pool filesystems, as also raised by Winnie on the list now exactly two months ago (but there has been follow-ups before then :-) Derek lets a pool fsck when needed, requests for files on the pool will hang for a couple of hours potentially. Even journaling fs need occasional fscking. Maybe best practice is worth documenting - Winnie sent a short summary (not to the list though). There is also the filesystems WG in HEPiX of which Greig is a member, and now also Nick White. http://hepix.caspur.it/storage Other side of the question is distributed filesystems again. NGS interested, asked about SRM too, but SRM doesn't do this, nossir. 2. Space token deployment - and VOs and pools and stuff. This think has a description of the, er, space token descriptions: https://twiki.cern.ch/twiki/bin/view/LCG/GSSDATLAS https://twiki.cern.ch/twiki/bin/view/CMS/StorageClassesForFlavia https://twiki.cern.ch/twiki/bin/view/LCG/GSSDLHCB Experiments expect all storage at Tier2s to be REPLICA&ONLINE. Work ongoing to ensure that this gets published properly, although it is currently relevant only for the PPS. See also Maarten's example here: https://twiki.cern.ch/twiki/bin/view/LCG/GSSDGLUEExample 3. Overview of GridPP3 planning for storage, issues for discussion. No time for this really, postponed. 4. DONM Jens out next two Wednesdays, Greig will either run or cancel the meetings, as appropriate. 5. AOB Greig reports progress on two "bugs", one being a new SAM test to detect full SEs (which as you will recall previously had caused problems because often ops tests then fail): https://savannah.cern.ch/bugs/?func=detailitem&item_id=26046 Another is the ongoing item on testing SEs without relying on higher level services, which we of course have done in the past, e.g. with the tests that Dave Kant built for us. https://savannah.cern.ch/bugs/?func=detailitem&item_id=25249 ------------------------------------------------------------------------ 176 07/02/2007 Deploy SRM 2.2 for CASTOR at RAL RAL CASTOR team Open Ongoing. Chris K was expecting to make progress before tomorrow (2 Aug) but the real expert is Shaun who is back from leave next week. 191 28/02/2007 Investigate YAIM conf to publish SRM22 for dCache Owen Open Still open. Greig deployed it without YAIM, but remaining issue is the information system. Greig reports that Stephen is looking into this, although mainly as a GlueService. The SE publishes information twice, loosely speaking, once as a GlueService and once as a GlueSEControlProtocol. Plus some tools like GFAL depend on the deprecated 'port' attribute published in the StorageElement object. [I managed to scribble this URL off the chat window which doesn't seem to allow pasting...] http://jra1mw.cvs.cern.ch:8180/cgi-bin/jra1mw.cgi/glite-info-provider-service/src/glite-info-service?rev=1.1&content-type=text/vnd.viewcvs.markup 193 07/03/2007 Document RFIO testing in Wiki for (DPM) site metric Greig Open Improvements have been made to the system. It is now easier to deploy, and can be configured in more ways, e.g. for random vs sequential access. 199 14/03/2007 Investigate reading recently read files in CASTOR Jens/Derek Open Investigation believed to have been done as part of the Tier1/CASTOR team meeting with experiments. Closed. 215 27/06/2007 Report on DPM on Lustre Greig Open Ongoing, Stuart is having VOMS difficulties. 222 04/07/2007 Volunteer Phil (Durham) to join the list Jens Open No news. 224 04/07/2007 Investigate whether a new headset solves audio Jens Open Er, old one seems OK, perhaps EVO is better than VRVS or more tolerant of old headsets. Closed. 226 11/07/2007 Report on RFIO stress testing Duncan Open No news. 228 18/07/2007 Investigate any correlation between SAM failures and WN Matt Open Done. Turned out to be dodgy pnfs mount points on doors, as suggested by Chris B. 229 18/07/2007 Collect information for SRM2.2 configuration Greig Open Ongoing, see discussion above. ------------------------------------------------------------------------ NEW ACTIONS 231 01/08/2007 Update expiration mode etc on wiki Jens Open 232 01/08/2007 Fwd filesystem stuff to Nick White Jens Open