Attending: Jens (chair+mins), Gareth, John B, Sam, Steve, Matt D, John H,
Winnie, Tom, Pete, Ewan, Elena, Raul, David, Brian (at 10:27)


0. You guessed it!  Operational blog posts.

   We are now half way through the quarter, and the number of blog posts
   is - zero!

   Is August a good month (apart from being on leave) to do some Real
   Work™ (and blog about it?)

   No operational issues

1. Round up of "new" and small VOs issues & status (data storage or
transfer or metadata related)?

   DiRAC - a very data focused VO (from our perspective), yet they currently
   have written only ~9 tapes in CASTOR - while the data is compressed, it is
   only ~70TB (not 50, it'd be 50 if they were C tapes but they are D).  Out
   of the 2,000 TB expected from Durham, that's not a lot... they are working
   on a document which describes how to set up the DiRAC end, and Lydia will
   be speaking at the GridPP35.

   VOs in general also update in ops, which focuses on how they submit jobs
   and stuff.  UKQCD @ Glasgow does more data oriented stuff, Craig has used
   the new GFAL tools; and LIGO, too, but Paul is currently busy.

   User representation on the GRIDPP-SUPPORT mailing list is very poor... we
   only have one (Lydia).

   LHCb - no problems with Tier 2s; eagerly awaiting trial of CEPH at RAL.

   We should have a standard plan for new VOs... but every time we have a new
   VO we need to update it because every VO is different from each other.  How
   much support is required to develop a data model?


2. "syncat" issues: current status; expectations from VOs.
   As a corollary, deletion rates? and cleaning up empty directories and
similar?

   ATLAS and CMS require namespace dumps of SEs.  ATLAS request that we put
   them into the SEs themselves, into the ATLAS namespace (so will require
   privileged access); CMS have an upload process.  Raul has been working with
   CMS and apparently they have been taking the feedback on board.  There will
   be a web site describing the processing that will be done with the
   (meta)data; once available, Raul will circulate the link to the list.


3. Update on the future T2D proposal - we're meant to finalise the
   Ewanian document.  The sort-of self imposed deadline (but giving time for
   the PMB to read it and process it) was mid August, so we should do
   something on it this week.


4. I am still catching up. Here are some potentially useful things to
have a chat about:
 - whether and how other groups can benefit from GridPP's expertise with
data?
 - as a corollary, moving data to/from GridPP and other infrastructures?
 - technology - could someone present something?
 - data related events (conferences and suchlike)
 - do we need to look at accounting again at some point?

   Technical things suggested mainly include the best way to set up a CEPH
   pool.  Talks for GridPP35 were discussed at ops yesterday; topics relevant
   to storage and data management (or rather in which the S&DM is relevant)
   include IPv6 and security and traceability.

   For JANET networkshop (next year), there will be something on end to end
   performance - where we should put some thoughts into what we could present;
   we should have some graphs and things - Duncan could do some perfsonar
   stuff?
   Also IPv6 of course where JANET support the universities and we have the
   requirements for deployment of IPv6.

5. AOB

   NOB


Tom Whyntie: (12/08/2015 10:02:59)
Just to say huge thanks to Steve for the feedback on the DIRAC instructions so far
Steve Jones: (10:03 AM)
50 TB not a lot! WHat counts as a lot?
Jens Jensen: (10:04 AM)
5 PB would be more
Ewan Mac Mahon: (10:06 AM)
A lot of the updates in ops has been that the new user has been busy/on leave and not got back to us.
It's tricky because I think it sounds like /we're/ doing OK, but it's a bit slow going in practice.
gridpp-support
It seems to be kicking people with CERN adddreses, but we need to follow up and see what's going on.
David Crooks: (10:14 AM)
Sorry I'm late in.
Ewan Mac Mahon: (10:15 AM)
np, Sam can give you your list of new actions afterwards.
raul: (10:16 AM)
yes
they are followi inbg
Ewan Mac Mahon: (10:32 AM)
We have two factor auth - you have a grid certificate, and you know how to use it.
raul: (10:33 AM)
Has anyone used Atlas Ripe measurements?
John Bland: (10:34 AM)
I haven't even looked at ripe for months. I assume it's still working.
raul: (10:35 AM)
It's quite useful to show network connectivity, but it is supoosed to have some measurement tooling
Jens Jensen: (10:36 AM)
https://indico.cern.ch/event/384358/
Ewan Mac Mahon: (10:37 AM)
It's been used for the occasional thing - I've done some basic 'ping a thing from a bunch of places' tests, Chris Walker diagnosed a routing problem to/from Taiwan (quite) a while back.
raul: (10:37 AM)
any paper/blog about that?
Ewan Mac Mahon: (10:37 AM)
But mostly I think the RIPE probes are best thought of as a contribution to the infrastructure, we tend to use perfSonar for 'internal' PP stuff.
And I'm not sure if it was published, talked about at meetings, certainly.
Tom Whyntie: (10:39 AM)
Is this a (public?) Google Doc?
Cool, tanks
*thanks