Attending: Jens (chair+mins), RobC, JohnH, David, Teng, Brian, Chris, Sam, Govind, Winnie, Dan

0. O B P

   We have a blog post from Matt!  Hurrah!

   Brian mentions keeping track of site upgrades; in particular "large T2s" are expected to provide IPv6 in only about two months' time;
   and sites need to bring their storage to baseline again.  The monitoring page (see chat) is maintained but we used to have automated
   querying of the information systems (previously known as Greig's monitoring, then Wahid's monitoring).  The information is in the
   information system provided they publish correctly.

1. Discussion of last week's storage workshop (and hepsysman)

   Feedback was that it was a useful meeting and we got a fair bit of stuff done in an afternoon and it was useful to have people together
   and thanks to Glasgow for hosting it.

   Most popular topics were:
   - upgrades, particularly DPM - should be a topic for wiki?
   - IPv6, dual stack - [and when do we get to single stack IPv6?]
     IPv6 makes "things go slow" because the service is trying IPv6 and failing and only later falling back to IPv4 which works.

   Less popular (and possibly more exotic) topics include:
   - Data lake - is it a buzzword (yes) but also a place to put "all your data" and collaboratively explore it, with expert support.
     Probably useful to get some feel for what's out there with the large CSPs, since the general feeling at the Crick workshop was
     that hybrids are the way forward (i.e. HPC and cloud, HPC and BYOC).  May also be relevant in the context of UKT0.
   - Replacement for GridFTP?  xroot or HTTP?   xrootd can support HTTP.  The xroot protocol is not likely to be used outside of HEP
     (and despite being openish is not a standard strictly speaking), but then HEP has long traditions of doing its own thing.
     Again the UKT0 argument - if GridPP is seen as the shining example of a DTZ etc, we still do some things in ways which do
     not translate easily to other communities.  Maybe also worth revisiting the document known as Lydia's document.
   - SciTokens - popped up elsewhere in presentation from NCSA, so still needs more looking at...
   - CASTOR information provider - now required by CERN to publish tape data (and long standing GGUS ticket against RALT1); progress
     made over Christmas
   - ZFS (topic revisited regularly in storage group)

   Need to pick topics to follow up on at GridPP40.  Suggestions are:
   - IPv6
   - The general GridPP<->cloud<->traditional approach to data (incl UKT0?)

2. Wiki pages and documentation - are keydocs sufficient?

   Pages are there to be useful and too many people only grudgingly do the minimal edits on their pages.  Maybe worth having someone else
   assess whether the pages are actually useful?  Or, of course, everyone can contribute, not just the ones named

3. AOB

   Jens probably in GVA next week

Brian Davies @RAL-LCG2: (24/01/2018 10:01:09)
https://www.gridpp.ac.uk/wiki/Storage_site_status
Govind: (10:08 AM)
it is blurry
David Crooks: (10:09 AM)
If you double click it it should open up in a larger size
Paige Winslowe Lacesso: (10:11 AM)
I'm not hearing anything?1
David Crooks: (10:11 AM)
Jens is talking - he sounds OK to me
Paige Winslowe Lacesso: (10:11 AM)
exit+rejoin = audio now fine
Chris Brew: (10:20 AM)
Does http offer third party transfers and multiple streams?
Matt Doidge: (10:36 AM)
Duncan asked me to do a ipv6 talk at GridPP
Paige Winslowe Lacesso: (10:40 AM)
THANKS JENS