Attending: Sam, Ste, Winnie, Brian, Dan, Duncan, Matt, Teng, Luke, Jens (chair+mins)



0. OBP

   Lancaster high load caused by DPM GridFTP, but which ones? FTS, users - AtLAS jobs use GridFTP, apparently, to
   upload.  Seems to be busy threads and limiting the number of threads only makes the individual threads busier.
   These look like FTS transfers, and FTS currently does need GridFTP for TPC.  On the user side, Winnie sherlockholmesed
   there were LSST jobs from Bristol accessing data at Lancaster through SRM (and hence GridFTP which is the default and
   hence only transfer protocol used)

1. Today (if you read this tomorrow, yesterday) has (resp., had) a pre-GDB on dynafed: https://indico.cern.ch/event/739894/, so if anyone attended and could give us a summary, that'd be much appreciated.

   Agenda includes a google doc with useful discussion.  NERSC and FZJ contributing.

   * Authorisations - using gridmap file(!?)  However, dynafed can use Macaroons.
   * Question about token pass-through to object store - 
   * dynafed needs to work in both directions (R/W), "not a problem for RAL" - but might be more interesting if it
     was a single global world-domination dynafed

On a not unrelated note, pre-GDBs are suspciously often followed by GDBs (https://indico.cern.ch/event/739880/), which promises final reports on the SSC, DPM workshop, the pre-GDB (as is customary), and the xroot workshop.

   There were in fact two pre-GDBs.  The other one was on Xroot proxy cache - https://indico.cern.ch/event/827556/

2. Storagey things for GridPP in August. We should decide, which, if any, of the numerous ongoing tests we aim to finish so we can report on  them. We should take advantage of the quieter (*cough*) summer months to do some Real Work™?

   * SRM less - Bristol as example.  For sites going SRM-less, if not updated in GOCDB, are the tests still running (and failing?)
     LZ enabled to use SRM-less at Bristol, but tend to use Imperial.  Lancaster could volunteer to go SRM-less, but without
     turning off GridFTP.
   * Glasgow CEPH trial
   * xroot proxy cache setup - particularly if using the SLATE/Kubernetes recipe.
   * xcache monitoring and analytics (Teng) - if using xcache, it is necessary to have a good view of its performance.
     (as Chris discovered)

3. AOB

   This Zoom thing seems to be working.  Jens should switch to the UKRI-provided account.


From Brian Davies @UKRI-STFC-RAL to Everyone:  10:07 AM
trying chat
From Me to Everyone:  10:07 AM
it works...
From Brian Davies @UKRI-STFC-RAL to Everyone:  10:08 AM
all your SAM tests will prbably fail
From Luke to Everyone:  10:10 AM
done :P
No SRM since 2017?
From Brian Davies @UKRI-STFC-RAL to Everyone:  10:11 AM
now you have no production ATLAS usage , your  srm log should be really low now so you can see who is stil using you.
From Matt Doidge to Everyone:  10:23 AM
http://atlas-agis.cern.ch/agis/service/detail/UKI-NORTHGRID-LANCS-HEP_SE_0/
From Sam Skipsey to Everyone:  10:48 AM
https://indico.cern.ch/event/827556/contributions/3463592/attachments/1875921/3089505/WLCGPreGDBMeeting_BHAM_Jul2019.pdf
https://indico.cern.ch/event/827556/
From Brian Davies @UKRI-STFC-RAL to Everyone:  10:50 AM
https://indico.cern.ch/event/739880/contributions/3470865/attachments/1877628/3092527/StatusDPM2019-gdb.pdf
DPM sunmmmary at GDB as well, bit rose tinted.....
From Luke to Everyone:  10:51 AM
also: breakout rooms