Attending: Brian, Sam, Steve, Jens (chair+mins), JohnB, RobC, JohnH, David, Winnie, Govind


0. Operational blog posts

    - including discussion on list and whether we should update the
  (key) docs accordingly

  Brian's got the one on data loss and dCache; Jens's has turned amber, and Sam's remembered to make his document green.

1. GridPP contributions to September's pre-GDB, see list of topics:

    - https://indico.cern.ch/event/578974/

    - if people have hols coming up, not too soon to start agreeing what
   we aim to do. In particular, identify things you want to work on during
   your hols (if storage and data management related...)

    - topics could also include (perhaps) accounting (cf WLCG workshop),
   non-certificate access or gateway, caching(?),

   - we didn't get a firm commitment on topics, but more like a wishlist: storage accounting and resource reporting, discussed in
     ops meeting.
     Maybe something on the non-certificate authenticator (a la Globus), or Pathfinder.

   - caching - Sam's found the documentation for Brian B's thingy; turns out to be using xroot internally and setting up config
     for that.

   - Object stores and archival storage is mostly at RAL, but maybe there is scope for archival storage between T2s using disk?
     E.g. a distributed T2 that replicates data could achieve good durability?  Or you could archive on other stuff like vellum,
     or archival quality paper or blurays - Facebook apparently use bluray.

   - or something with VOs...


2. July/August (work)plans. In theory, these should be quieter months as
   people beetle off on holidays - so time to get some Real Work™ done?

    - Should we switch to fortnightly meetings during the rest of
   July+August?

    - yes.

3. If there's time, the round table update we didn't have time for last
   week.

   Brian - needs to do something on data loss and dCache key docs...

   David - getting storage online, ZFS transitions for remaining servers.  Older disk servers have RAID arrays, so need to
   transition from RAID to ZFS.

   Govind - upgrading pool nodes from 1.8 to 1.9 and puppetising.  Also interested in moving from SL6 to CentOS7

   John B - building work in machine room.  Coincidentally also working on SL6->CentOS7 migration, puppetising (from YAIM),
   and a rolling upgrade of pool nodes.

   John H - care and feeding of storage system; may look into 1.9.0 if there's time, but also thinking about the caching
   "solution" - although there is no timescale for it.

   Winnie - checking through Luke's documentation for storage systems.

   Rob - not much on prod'n.  Retiring small amounts of server capacity.  New storage from RDF showing odd socket timeout
   so aim to sherlockholmes that, ideally prior to deployment into production.

   Sam - in addition to David's list also working on the stashcache; parallel to what Catalin is investigating for OSG

   Steve - nothing to add beyond John's description.

   Jens - forgot to add myself didn't I.  Need to take some leave so I can get some work done; from a GridPP perspective would be
   nice to get time to work on the storage accounting.


4. AOB

   NOB

Brian: (19/07/2017 10:01:17)
yes
need opt look it up
John Hill: (10:03 AM)
LHC is currently still recovering from the TS
Brian: (10:04 AM)
blogpost not ready. clicked wrong buttom on blogspot, have reverted.,
youn cab already "archive" to disk. It all depends on which defintion or archive you mean.
vellum...;)
nope
still on it
queen's speech is still on vellum
ast least
I/O rates for ATLAS
Paige Winslowe Lacesso: (10:25 AM)
hahahahahahahahhahhahaha!
John Bland: (10:27 AM)
govind: our DPM puppet config works the same on SL6 or Centos7 so you can do the upgrades separately, we just took the decision that we wanted a clean install and if we're reinstalling we might as well go with Centos7.