Minutes of storage phone conference 07 June 2006 Present: Edinburgh: Greig Glasgow: Graeme Lancaster: Brian Durham: Mark Tier 1: Derek RAL Storage: Owen, Jiri, Jens (chair+mins) 0. Review of actions (see below) 1. Summary of the FNAL workshop, issues (Graeme) Lots of WSDL bashing. LHC expts need storage space reservations. There is a reserveSpace function call but they don't need many tokens so obtaining a space token may, in some implementations, like CASTOR, require operator intervention. CASTOR will need an Olof to allocate space, dCache will provide the call but maybe for sysadmins only, and DPM may provide the full dynamic storage allocation, although a later discussion between J.-P. and Shaun seems to indicate it's not certain. In any case, the space allocation is mainly a Tier 0 and 1 issue and they tend to run dCaches and CASTORs, not DPMs. Accessibility of a file will be encoded as "online" or "nearline" (only), and reliability as "replica", "custodial", or "other". Only certain combinations allowed: online-replica, online-custodial, and nearline-custodial. In the GLUE schema, should these values be published? Graeme doesn't think they should be used for selecting SEs, but that it should be possible to publish them in the next revision of the schema. Implementations are expected to provide SRM 2.2 and to interoperate by early November 2006. Experiments have declared that they will only use GFAL and FTS. 2. dCache issues Brian has now also installed a fresh dCache with xrootd doors. Chris Brew has the other one. New dCache client today from Owen. Owen also mentioned the need to remove the PNFS dependency on pool nodes, and the database dependency, also on pool nodes. ACTION Owen: document these bugs in Savannah. Question whether people want a new RPM for read-only xrootd or a tarball. The RPM can dep-exclude the other one. Owen should ask list. Release procedure - although there is one, it should be documented (ACTION). Input from action #110 may be useful. Incidentally, Jiri has now written the beer song for S2. 3. Storage VO discussion: So as mentioned below, Jens will circulate next version (ACTION). Jens also proposed "cheating" as an option - to publish available space shared between VOs as an appropriate fraction of the available space: Suppose CMS and Atlas are sharing 10 TB with 6 TB available, but with ratio 2:1. Then, if the SE publishes 4 TB for CMS and 2 for Atlas then it will look fine, even if either VO could write more than that into the space. 4. AOB Some collected under dCache above. ------------------------------------------------------------------------ ACTIONS 41 10/08/2005 Agree licence with DESY Jens Open No news. 53 12/10/2005 Find reasoanable % for SE uptime for SC4 Jeremy Open Reassigned. Follow up with GDB et al No news. 54 02/11/2005 Report on performance/scalability with pools on WNs Paul Open Progress. Will put in wiki. NB: dcache pools No news. 86 08/02/2006 Extend monitoring to do sites per VO and VOs per site Greig Open No news. 105 03/05/2006 Re-poke DESY or FNAL about SRM 2.1 for dCache Owen Open Reassigned to Owen who poked Tigran during the meeting. We should have something to test Really Soon Now(tm). Testing is important because if we find problems with 2.1.1 then they may also affect 2.2 so best to discover problems as early as possible. 110 23/05/2006 Locate EGEE release procedure Jens Open I've searched for it after the meeting, but without any luck. I've been through my email, JRA2's website and document repository, and googled for it... 111 23/05/2006 Test Owen's latest install Greig Open Urgent Done. 112 23/05/2006 Document Xen stuff in Wiki Jiri Open No news. Jiri proposed waiting for Xen3. Greig needs it most but will be at CERN next week for the T2 workshop, and then there's GridPP. 113 23/05/2006 Feedback on CPU/storage % and VO % and VO alloc to Jens ALL Open Some feedback had been received, from RHUL, Lancaster, Edinburgh. 114 31/05/2006 Send suggestions to the list for items that should be covered during the DPM/dCache sessions at the T2 workshop ALL Open Not much had been sent, but three items were mentioned: * Quotas * File system draining * Performance tweaks There should be other things - during the past two months there are several things we have discussed in these meetings and said "oh yes we mustn't forget to raise it at the workshop." 115 31/05/2006 Speak to dCache team about Tier-2 dCache configuration Owen Open Ongoing. 116 31/05/2006 Progress of Durham-MAN networking discussions. Mark Open We are aiming for 200 Mb/s. Question whether that is average, and what to do when expts send data in bursts? A good measure may be to calculate the peak as six times the average aggregated rate. ------------------------------------------------------------------------ NEW ACTIONS 117 07/06/2006 Document dCache release procedure Owen/Greig Open 118 07/06/2006 Document pool node dependency problems in savannah Owen Open 119 07/06/2006 Circulate next version of VO storage to list Jens Open