- This is the weekly GridPP ops & sites meeting - The intention is to run the meeting in Vidyo: https://vidyoportal.cern.ch/flex.html?roomdirect.html&key=zXhsqAxVnaT6 -- The PIN is 1234. To join via phone see http://information-technology.web.cern.ch/services/fe/howto/users-join-vidyo-meeting-phone for dial in numbers. -- The London (UK) service is on +44 (0)161 306 6802. Phone bridge ID 1001002 -- The meeting extension is 109308582. PIN 1234 Chair: Jeremy Minutes: Apologies:
GridPP Ops minutes 23 Jan 2018
LHCb: Broadcast about new cvmfs mount point (/cvmfs/lhcb-condb.cern.ch/) but
should be automatic if cvmfs repos are automounted.
Lost files ticket at IC, due to lost server. LHCb needs to follow up
CMS: xrootd problems. Tickets about xrootd related to presence of IPv6 support
at the site.
ATLAS: Storage overloading ticket at Glasgow? Increased number of
connections allowed per SE.
Ticket for Sheffield, closed and now reopened.
Ticket opened for RAL at the weekend, about ARC CE 03 instability
Deletion errors ticket at RAL.
RAL ticket about problem with transfers, due to overload?
IC configuration change so can write to QMUL disk. Should be ok
within existing QMUL 10Gb/s link.
Request for more capacity for SOLID
GridPP DIRAC status: only a couple of sites not run recently
GFAL vs DIRAC problem still being understood
Birmingham going to remove CREAM GridPP site
and rely on Vac GridPP site
Meetings and updates
(Points not already mentioned on this week's bulletin)
Do CMS and ATLAS Singularity requirements match? Glasgow will be
a test of this as they have both.
ATLAS CASTOR at RAL back and believed to be ok now.
https://wiki.egi.eu/wiki/SVG:Meltdown_and_Spectre_Vulnerabilities has links
Intel have fix for instability introduced by microcode changes
Sites are requested to monitor the situation eg via that wiki page and apply
appropriate updates on the timescales requested
Perfsonar tests show order of magnitude differences between sites. Some
sites have been contacted for more information.
Steve Lloyd's network tests also available:
Have to be careful because sites may optimise for SE to remote SE rather
than SE to remote WN at random site. eg WNs might be on NAT with a good
connection internally but poorer route to the WAN. Some experiments (eg
CMS and LHCb are streaming already. eg LHCb failover or when stripping
at larger Tier-2s where data is streamed in and out of the WNs without
using site storage.)
See detailed Bulletin comments
Please look at the agenda https://indico.cern.ch/event/651349/ for
links to slides
Daniela Bauer: (23/01/2018 11:04)
Raja Nandakumar: (11:12 AM)
Jeremy Coles: (11:16 AM)
John Hill: (11:19 AM)
Aren't RHUL in downtime?
Duncan Rand: (11:19 AM)
Yes.. they are in downtime for Network maintenance for last 5 days.
Jeremy Coles: (11:20 AM)
Yes. But intermittent.
David Crooks: (11:35 AM)
Mark Slater: (11:41 AM)
I'm afraid I've got to head off - email me of there's any bham specific stuff!
Jeremy Coles: (11:45 AM)
Paige Winslowe Lacesso: (11:59 AM)
Sorry sorry, must go - email me if any brizzle-specific data
Jeremy Coles: (12:01 PM)
Daniela Bauer: (12:21 PM)
Sorry, I've got to go.
Jeremy Coles: (12:21 PM)