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

Dan:
   work to replace StoRM front end [storm front?]; AMD 96 thread with 100 Gb NICs.  StoRM should be
   nginx based.  Can spread load with GridPP and WebDAV.  May go to 40 Gb on head node before 100Gb;
   site link is 20Gb but can saturate in spikes; ordinarily traffic is 10Gb or below.  Looking at
   switches.  CentOS running Lustre.  HDFS could be deployed on Lustre with Slurm.  Expecting upgrade
   in about 1y from now.

Duncan:
   Working on fixing perfsonar, also used by others.  Resolving IPv6 DNS issues with Tim and Matt.
   Looking at remote jobs accessing data between QMUL and IC, cf. earlier reports on Bham jobs accessing
   Manchester.

Teng:
   Moving servers (physically) for EOS and Rucio.  Also working on CEPH and OpenStack, performance testing.
   LSST and DUNE using Rucio monitoring.  Elastic search data for database networks.

Ste:
   JohnB only 10% so mostly steady as she goes.  220TB disk servers - may populate with ZFS or RAID...
   Also points out need for storage accounting?  Shouldn't need APEL.  Also considering moving to HTCondor
   CE.

Sam:
   Evaluation of CEPH.  Also working with DPM TPC troubleshooting - maybe due to differences between nodes
   in cluster which is not homogeneous.  Also DOMA QoS follow-up - "requirements gathering" through, probably
   a site survey.

Matt:
   Aim to get DPM on CentOS7 pool nodes working before Xmas.  New kit expected after Xmas hols, with 25 Gb
   NICs - upgrade of JANET link likely.  Also work with DOMA TPC.  Success with xroot config - rewrote it.

Winnie:
   long standing ticket with inconsistent storage for CMS, assigned to Luke.  Waiting for Sam to feed back
   on xroot only spec.
   Teng reports inconsistencies at Edinburgh, too.  Matt had a checking script for ATLAS, validating the
   disk copies against what should be there, but lost it.

Brian:
   Troubleshooting RAL FTS with IPv6 routing.  ATLAS want to get rid of their useful monitoring and replace
   it with less useful but prettier monitoring, like the main FTS one.

Jens:
   IRIS WGs on AAI, hybrid clouds about to meet, if only virtually.  Still need to look at Storage Acct records,
   there is also a new GLUE schema about to appear.  Interested in the Rucio+DIRAC as we discussed, our tutorial
   is out of date (slightly, not a lot), but maybe we need a tutorial on Rucio that we could use with IRIS.