Attending: Steve, Jens (chair+mins), Daniel, John H, John B, Chris B, Winnie, Brian, Elena, Sam, David, Matt

Apologies: Simon


0. Bloggerational outposts

   No operational issues reported.

1. Loose ends: storage accounting and information, SNIA NVM (now PM)
summit [1], GridPP, "small" VOs. Reliability engineering for storage
systems?

Yesterday's pre-GDB was on benchmarking [2] but seems to focus on
compute? Today's GDB has got some iiiinteresting things particularly for
T1, though, such as EOS, synchronisation, and CTA.

   On storage publishing and storage accounting records.  There are two kinds of records, snapshots and incremental, or rather the former
   attempts to describe the current data volumes within a given time window, and the latter how much has moved during a time window.
   Most of the storage focuses on the former.

   There is a portal which attempts to aggregate information but it does not appear to treat the VOMSy stuff correctly, because DPM
   can assign accounting records (which contain a "group") also to VOMS roles and capabilities.  While the role is a subset of the members
   of the group, the data assigned to the role is also a subset of data accounted for the group.  Filecount is optional.

   The start and end time suggests a cloudy type accounting model, where the period of time could be equal to a billing period; in
   our case it is apparently sort of repurposed (ugh) for when the snapshot was created and the validity, ie the time before the next
   update can be expected,.

2. Anything else on the XMA hardware test and/or ZFS?

3. Anything else

[1] http://www.snia.org/pm-summit
[2] http://indico.cern.ch/event/578967/
[3] http://indico.cern.ch/event/578983/


John Hill: (08/02/2017 10:01:20)
No!
I run a daily cron script, rather than a daemon
Chris Brew: (10:22 AM)
CMS bare: 140,040,521,858,893
CMS Prod: 1,083,354,723,141,185
So no the plain role doesn't include the totals for the roles
Jens (RAL): (10:27 AM)
Thanks Chirs
-> Chris (!)
Daniel Peter Traynor: (10:27 AM)
https://www.backblaze.com/blog/hard-drive-benchmark-stats-2016/