Attending:
	Brian, Jeremy, John B, Matt D, Rob, Robert, Steve, Elena, David, Gang,
	Matt RB, Sam, John H, Gareth, Jens, Chris, Ewan, Tom, and (later) Pete

Special guest star:
	Jon Perkin from T2K

Jon had prepared some slides to present to the group.  The executive summary of the discussion was:

* syncat - data is available but not necessarily in the most useful format.
  In general need to do both ways.

* Resource allocation - OK, written into GridPP5, along with ILC and (maybe) SnoPlus

* Problems with use of DIRAC particularly from TRIUMF

T2K's infrastructure is kind of stitched together of WLCG stuff in which
RAL is the T1 and non-gLite stuff for which TRIUMF is the T1, and with
scripts glueing stuff together (as ever).

User level activity varies.  There are in fact 500 T2Kers, with about
103 in the VO, and 10 active users, and 2-3 who do infrastructure stuff...

Expecting 100-200 TB per production cycle, and 2-3 production cycles per year.

Currently caching metadata (BSD=beam summary data), but expecting to move
to database.  The LFC is for finding out where stuff is; it is not a filesystem.

Regarding syncat - need to audit disk usage.  Getting "raw" XML but this
isn't necessarily the best option - need SURL and filesize.

The EMI syncat is not syncat - the latter is a synchronisation facility
for synchronising with the catalogue, implemented (only) for DPM, and
meaningful only if all SEs used by the VO support it.

Maybe also opportunity to do some dissemination to acknowledge GridPP's
contribution to T2K.

Your rapporteur couldn't get the chat copied and pasted out of evo so
had to take screenshots and paste them together.  Gotta love the GIMP...

See http://storage.esc.rl.ac.uk/weekly/20140205-chatlog.png