Attending:
Jens (chair+mins), Sam, Rob, JohnB, JohnH, Winnie, Steve, Raul, Brian, Govind, Matt

0. No operational issues; Elena might have had one but didn't join the call

Blog posts are looking good, both number-wise and content-wise.

1. The Globus Toolkit withdrawal of support is known to storageelementware developers; DPM probably relies still on GT for its
   GridFTP implementation, dCache's may be independent given that, famously, their xroot implementation is.  Would be nice to
   retain support for GridFTP; however, supporting the Globus code ourselves (= within the DPM/StoRM/dCache communities) may be
   tricky even "just" for security patches.

   Lydia used Globus Connect and it worked fine as you can see in the writeup, but they might diverge, e.g. as GC moves away
   from client certificates.

2. hepsysman highlights include ZFS, IPv6, and just getting together to show off general gripping gridpp goodness [1]

3. WLCG workshop [2] had lots of good stuff and some notable things like people asking questions about the irrelevant parts of the talks...

   Clearly relevant for future evolutions of T2s; whether the remote access is necessary and sufficient (these are two very different questions);
   and also WLCG is noted for its openness in the process - e.g. Oliver asking for contributions to task forces for stuff like authorisation and
   benchmarking.

   Of course for a workshop in general there is a bit of a tradeoff between showing up with an idea or a quick PoC and asking if someone else
   is interested vs showing up with finished and polished work and showing everyone how you've done it.  (Note also the distinctions between
   having meetings/discussions about something vs doing Real Work(tm)).  There's probably a role for both ends of the spectrum, and perhaps moreso
   for things in between.

   Apropos Real Work(tm), Jens managed to write some code for the signing machine to support user-submitted (as opposed to wizard-edited) multi-SAN
   certificate requests, as used by the SEs of, currently, RALPP and RALT1; Imperial use it for DIRAC, etc. (use case presented at hepsysman).
   Being tested this week, with a bit of progress/luck we could have it ready this week.

   Also encouraging was the ability to discuss things that didn't go so well, and not just highlight the success stories.

   For people who were not able to attend, they should look through Oliver's excellent introductory introduction to the data session Tue morn.[3]

   As Brian mentioned there is a data pre-gdb planned for Sep 12; link has just appeared as we speak [4].

   All in all an excellent workshop with lots of good stuff that we will be spending more time following up on, great weather in Manchester
   (unusually), and good to see that others look to WLCG to learn how to do (resp., not to do) things, to share experience and build on the
   good stuff.  And a good GridPP presence.

   ... which reminds Jens about the storage accounting - 

4. AOB

   NOB

$. References

[1] http://hepwww.rl.ac.uk/SYSMAN/June2017/main.html
[2] https://indico.cern.ch/event/609911
[3] https://indico.cern.ch/event/609911/contributions/2605027/
[4] https://indico.cern.ch/event/578974/


Chat log:

Steve Jones: (28/06/2017 10:11:18)
It sounds like he's transmitting from Mercury.
Brian Davies @RAL-LCG2: (10:12 AM)
terry froy talk at WLCG also for ipv6 makes me think people need to thinka about their adress plan. TF was saying that having too large subnets can also lead to issues ...
will typoe form now on..
http://gridpp-storage.blogspot.co.uk/2017/06/a-storage-view-from-wlcg-2017.html
ALICE have always been worried for latency between WN and storage due to streaming methodology
Particlur of note also is network requirements for T2s 
prelimenary announcement of pre-GDB storage for Sep 12. perhaps focus on tape evolution
bye!