Attending: Jens (Chair+mins), Luke, Marcus, Rob C, John H, Sam, John B, Daniel, Tom, Brian, Duncan, Winnie, Govind

Special welcome to Rob Currie who has just joined the Edinburgh team

0. Operational blog posts

   We need more blog posts!  About interesting stuff!

   Operational issue: Marcus has a problem with xroot transfers failing due to checksum problems.  This is job output data,
   moved with xrdcp; curiously, the checksum should not yet be supported by DPM, yet it does not seem to cause problems at
   other sites.  Marcus and Sam compared software installed, to see if there were any notable differences.  Brian thinks it
   should be working again (configuration on the ATLAS end?)

   Operational issue: Luke reports suspiciously truncated file transfers *out* of Bristol with xroot; files get truncated
   at precisely 2^31-1 bytes.  This would be AAA data (for CMS).  No data is written the xroot; and reading and writing with
   GridFTP is successful, the problem occurs only with xroot.  Underlying storage is HDFS, so fairly unique in this respect;
   Nebraska also have HDFS but use BeSTMan.

1. Any news/updates on - accounting for ATLAS, DPM workshop, cache testing?

   Duncan gave a quick feedback report on the data transfer workshop organised by JISC in October (targeting mostly campus
   networkers, so relevant for most sites).  About 60-70 people present;

   https://www.jisc.ac.uk/events/campus-network-engineering-for-data-intensive-science-workshop-19-oct-2016

   Phil Mayers from IC gave a talk on setting up what is essentially "Science DMZ" at IC; has also presented (elsewhere)
   work on IPv6 for IC.
   JISC also continues to work with WLCG sites on things such as firewalls, performance etc.  Currently work with Bristol
   where there will be a DMZ meeting this week.

   Interestingly more disciplines are starting to move Real Data(tm) (as opposed to talking about it); e.g. biomed,
   climate, and of course DiRAC.

   "Data transfer zone" considered a better term than "Science DMZ", also Data Transfer Node is the node hosting the
   endpoint for the transfer; this is the stuff that we have talked about in recent weeks where there is a proposal to
   write up a, er, writeup with others, including the chap from the NZ NREN.  Obviously open to anyone in GridPP who
   is willing and able to contribute.


   As regards the ATLAS information thing, Paul Millar posted an update to the OGF GLUE list proposing steps which could
   meet ATLAS's requirements within GLUE, but with small changes (so with GLUE2.1 instead of GLUE2.0).

   DPM workshop - Sam working on two presentations, one a general UK update and the other on upgrading DPM where we in the
   UK have tended to take a more suspicious^W careful approach.

   On the caching stuff, Alessandra should be chasing an update; not much else is known.

2. GridPP as a data infrastructure - what to do next

   Explanation next week! (maybe)

3. If there is time, we can do a quick round table - e.g. what would you
   be working on if you had time?

   ... ok there wasn't time for this - maybe next time!

4. AOB

   Brian - chasing sites with SL5 disk servers, because SL5 will turn into a pumpkin some time in March/April '17
   and it may take some time to drain them before they do.


Lukasz Kreczko: (16/11/2016 10:08:59)
https://ggus.eu/index.php?mode=ticket_info&ticket_id=125052
Daniel Peter Traynor: (10:09 AM)
32 bit limit somewahere?
Marcus Ebert: (10:09 AM)
which files system is used for the storage there?
Lukasz Kreczko: (10:13 AM)
ll test.root 
-rw-r--r-- 1 phxlk users 2.0G Nov 16 10:12 test.root
[phxlk@soolin tmp]$ ls -l test.root 
-rw-r--r-- 1 phxlk users 2147483647 Nov 16 10:12 test.root
Daniel Peter Traynor: (10:13 AM)
is this is a stand alone xrootd serevr or part of dpm?
Lukasz Kreczko: (10:13 AM)
DMLite, yes
Jens Jensen: (10:14 AM)
.. precisely 2^31-1
Lukasz Kreczko: (10:14 AM)
the file is
xrdcp -d2 root://xrootd-cms.infn.it//store/mc/TTI2023Upg14D/PYTHIA6_Tauola_TTbar_TuneZ2star_14TeV/GEN-SIM-DIGI-RAW/DES23_62_V1-v1/70000/2E469FE2-FC92-E611-BE04-0CC47A4D99A0.root /tmp/test.root
Marcus Ebert: (10:18 AM)
for sites using xrootd with DPM, it would be interesting to see which of these commands work on the own site with a proxy of a VO supported at the site:
xrdcp -f --cksum adler32:print file root://storage//.... 

xrdcp -f --cksum adler32:source file root://storage//....

xrdcp -f -adler32 file root://storage//.... 

xrdcp -version 

Lukasz Kreczko: (10:20 AM)
xrdcp -f --cksum adler32:print root://xrootd-cms.infn.it//store/mc/TTI2023Upg14D/PYTHIA6_Tauola_TTbar_TuneZ2star_14TeV/GEN-SIM-DIGI-RAW/DES23_62_V1-v1/70000/2E469FE2-FC92-E611-BE04-0CC47A4D99A0.root /tmp/test.root
[2GB/2GB][100%][==================================================][56.89MB/s] s] 
adler32: 9b4dc426 /tmp/test.root 2147483647
Tom Whyntie: (10:39 AM)
Thanks! Bye