<!-- keep this as a security measure:
* Set ALLOWTOPICCHANGE =
TWikiAdminGroup,Main.LCGAdminGroup,Main.EgiGroup
* Set ALLOWTOPICRENAME =
TWikiAdminGroup,Main.LCGAdminGroup
#uncomment this if you want the page only be viewable by the internal people
#* Set ALLOWTOPICVIEW =
TWikiAdminGroup,Main.LCGAdminGroup,Main.ChippComputingBoardGroup
-->
Swiss Grid Operations Meeting on 2019-05-02 at 14:00
Site status
CSCS
Systems
- Phoenix complete shutdown
- Un-Racked all WN and Services nodes and cleaned from cables.
Storage
- - No special reports for the service
- - New (SE) servers installed, starting full data migration soon
- - SE servers will have GPFS as backend: this will give more performance and flexibility to the service and is the base for future steps (containers)
- - Planning the future of storage01-02: as first step they will be virtualized and then will be consolidated in one (v)server with external chimera (postgres) database
- - Servers moved to a new location, on a new network with no impact to the filesystem
- - Storage (SSD) moved to the new location
- - HDD Tier/pool moved from DDN SF12k to DELL SC9000
- - Will double the FC connections squeeze more performance (sometimes I see FC connections reaching the limit)
PSI
UNIBE-LHEP
- Stable running despite lack of maintenance
- Started decommissioning nodes, but re-deployment delayed, since Phoenix hardware is still at CSCS
- this will very likely affect our ability to meet the deadline for CentOs 7 migration (1st June)
- Monthly summary: Pledged: 42k, delivered 22k
- Ubelix contributing ~45% (23% typical)
- Running an average >2000 slots (2500 typical during the previous pledge period)
WC UNIBE-LHEP:
5month_UNIBE-LHEP:
- Accounting numbers (from scheduler) from last month
Swiss ATLAS statistics
- Hammercloud availability:
-
- ANALY_CSCS-HPC: 80%
- CSCS-LCG2-HPC_MCORE: 99%
- ANALY_UNIBE*: 100%
- UNIBE-* : 99%
- Running slots
- CSCS: 3.4k
- UNIBE: 2.1k (38%)
- CPU consumption
- Accounting Numbers from the ATLAS dashboard (April 2019) CSCS+UNIBE
- Delivered vs pledged
CSCS-LCG2: pledged 50k, delivered 39.4k
UNIBE-LHEP: pledged 42k, delivered 22k
- Update on VO shares at CSCS
- We have discussed this at length ion the past. In that occasion the issue seems to afflict mostly Phoenix and much less or not at all Piz Daint.
- Shares on Daint have been ok also after the decommissioning of Phoenix (40:40:20)
- Following the recent outages however, the shares are largely wrong, much worse problem that before
- RT ticket, Miguel following up
-
- Further ATLAS CSCS topics
- CentOS 7 migration
- deadline 1st June, awaiting clarification from ATLAS on what to do
- ARC monitoring
- Dino will join the NorduGrid conference in June to discuss his ideas
- There is alreday a feature request open since many weeks
UNIBE-ID
- Smooth operation of ARC-CE services for UNIBE-LHEP-UBELIX * currently lots of failed jobs due to memory excess
- Soon changing storage current storage in a downtime (one of the next two)
- current: 575TB GPFS based, new: 3.9PB still GPFS based
UNIGE
- ARC CE provisioning further delayed
NGI_CH
Other topics
- Topic1
- Topic2
Next meeting date:
A.O.B.
Attendants
- CSCS:
- CMS:
- ATLAS: Gianfranco
- LHCb:
- EGI: Gianfranco
Action items