Wednesday 9th May - Agenda
Introduction (Michel Jouvin)
• GDB summary needed. Plan to put notes in wiki linked to agenda.
• Encourage cross-VO T2 deployment of perfSONAR.
TEG next steps (Ian Bird)
• The TEGs as such (big working groups) have finished. Still work to be done.
• Further discussions at CHEP – including future of DPM.
• Deployment of glexec becomes a priority (tarball is not a solved problem)
CRSG report to the C-RRB (Ian Bird)
• Higher rates and parking data plus more analysis means slightly more resources needed (ALICE, ATLAS and CMS)
• LHCb have a revised charm physics program
• A lot more pile-up in 2012
• Use of high-level trigger farms to help with processing
• T2 installed capacity still an issue – REBUS will be used. Need to check the figures published for each site.
LHCOPN/ONE status & directions (John Shade)
• LHCOPN functioning well. Alarms: https://cclhcopnmon.in2p3.fr/LHCOPN/report/.
• perfSONAR-PS and MDM interoperability not tested. Jason Zurawski offering workshop on toolkit for site managers – any interest?
• L3VPN operations https://twiki.cern.ch/twiki/bin/view/LHCONE/WebHome.
• Routing policies important – symmetric paths
• OpenFlow as a protocol. TRILL/SPB for resilience (replace P2P links)
• PerfSONAR is a dormant setup with some selected core sites/points involved with regular testing.
• Encourage sites to install – guidelines in twiki. Sites need to setup alarms for themselves.
Federated Identity Management for HEP (David Kelsey)
• Remove the ID management from the service (use single sign-on). Adding an attribute authority (e.g VOMS) adds complexity
• Spans many communities not just HEP – common requirements being discussed (e.g. open standards, attribute aggregation) including operational ones like traceability.
• Research communities are to perform a risk analysis of using IdM.
Procedure to follow for proposed new T1 sites (Ian Bird)
• Policy document linked from agenda – discussed in 2011
• Requires expt. Support and balance against high-standards of existing T1 services.
• Prepare detailed plan. Follow tests and meet required service levels. Reach full status after about 1 year and Overview Board approval
HEPiX (Helge Meinhard)
• 23rd-27th April. Over-packed agenda. New track on business continuity
• Fabric management changing. Many labs moving to puppet. Quattor healthy. Some sites moving monitoring away from Nagios.
• Cloud computing on the horizon of realism (Openstack and OpenNebula)
• Working groups (virtualization; IPv6; Storage; Benchmarking)
WLCG workshop (Jamie Shiers)
• wlcg workshop mailing list
• Draft agenda now online
Virtualized WNs and Clouds
HEPiX Virtualisation WG report (Tony Cass)
• Set up to facilitate the instantiation of user-generated VM images at HEPiX sites
• Image endorsement – technical constraints and policy discussed
• Framework for endorsers to publish has been developed
Update on WNosDeS (Davide Salomoni)
• Worker nodes on demand service: http://web/infn.it/wnodes
• Integrates Grid and Cloud provisioning through virtualization
• Upcoming – dynamic VLANs
EGI Federated Clouds Task Force (Matteo Turilli)
• Why? Users keen on personalized environments
• Federation testbed now live
• https://wiki.egi.eu/wiki/Fedcloud-tf:FederatedCloudsTaskForce
VM contextualization and image cataloguing in ATLAS (Fernando Megino)
• Why? Some sites support multiple VOs with different needs. To be ready for when cloud resources are offered.
• Work ongoing over 1 year. Still manual steps. Currently use CERNVM.
• Contextualization desired for Gangli and Condor installs.
• Comment: Getting credentials in fine but concern about site settings for syslog etc. being overridden. Condor should be via CVMFS.
|