Swiss Grid Operations Meeting on 2022-05-12 at 09:00

• Zoom link:
  • https://ethz.zoom.us/j/67184143478

*Next meeting:

F2F 2 June 2022

Minutes:

EGI:

still investigating with EnhanceR how to deal with the fee and the 10% FTE

ATLAS:

heavy under-deliver ~35% of the pledge in April

share under 19%

storage at bern decommissioned and integrated into nordugrid (resources are still physically in bern)

CMS:

Mont Fort added to the CMS workflows

no open tickets

overall situation looks OK

A number of file transfer are failing: issue identify in the way davs manages the VOMS attributes in third party transfer. Only newly transferred files are added to the storage space, existing data cannot be added. This lead to a micromanagement of the free space. Just today we may have received a solution for the purpose of managing small reservation / deleting old data

CMS cache storage space usage: proposed a new space allocation to meet the required 75% required to be managed by cps-central

Tickets are proactively taken care from CSCS: thanks !

LHCb:

All fine on Daint. Usual 10% failing pilots that disappeared at the beginning of last month. Good but unclear what happened...

Mont Fort: some troubles to submit with the standard python script, but they figured out that the was a typo.

CSCS:

Daint: pretty bad month… confirming the numbers found by ATLAS (not so bad for the others)

Why ATLAS is so bad ? unclear but related to the reco jobs.

Mon Fort:

April 10 nodes with 128 cores 512 + 4 nvidia A100 + 10kHS06

May 32 nodes with 256 cores 512 GB/RAM + 90kHS06

Share: 66% of the resources to ATLAS to catch up + 15% / 15% CMS/LHCb

Temporary adding cluster on Mont Gele

same config as Mont Fort but with HDD=-based storage

Additional +90kHS06

ALPS down time in July to replace some network components suggest to keep Daint for a while with reduced capacity until July migration. Is that OK ?

Action items and points for discussion:

increase resources for TALAS on Mont Fort

Daint: around 20% of the resources (it won’t fix the reco jobs)

Summing the two should help

Must add asap Slurm and ARC monitoring on Mont Fort Mont Gele

If ATLAS sees reco jobs not running, it doesn’t send anything else because assumes that the site is full

ATLAS

• ATLAST2ReportApril2022.pdf: ATLAST2ReportApril2022.pdf

CMS

LHCb

T2 Sites reports

CSCS

UNIBE

T3 Sites reports

PSI

UniGe

EGI / WLCG

• New CERN Grid CA introduced on 25th April. dCache restart needed => done @CSCS
  • the following versions of dCache do not need a restart:
    • 6.2.40 and higher
    • 7.0.29 and higher
    • 7.1.20 and higher
    • 7.2.4 and higher

• New benchmark replacing HEP-SPEC06

  • The benchmark HEPSCORE is going to replace the old HEP-SPEC06
  • Preparing plans with WLCG and the EGI Accounting team for deploying the new benchmark
  • There will be a transition period during which both the benchmark will be published by the sites and used to normalise the data to allow comparison between the two
  • APEL is working on a version where the accounting records contains 2 benchmarks
  • ARC needs to develop accordingly
  • Expected timescale for rollout: Q4 2022 (tentative)

Review of open tickets

https://ggus.eu/index.php?mode=ticket_search&supportunit=NGI_CH&status=open&timeframe=any&orderticketsby=REQUEST_ID&orderhow=desc&search_submit=GO

4 of 4 Tickets
Ticket-ID	Type	VO	Site	Priority	Resp. Unit	Status	Last Update	Subject	Scope
156799		lhcb	CSCS-LCG2	very urgent	NGI_CH	in progress	2022-05-04	Pilots Failed at CSCS-LCG2	EGI
156213		atlas	CSCS-LCG2	less urgent	NGI_CH	in progress	2022-03-07	CSCS-LCG2: Non-operational storage ...	EGI
154102		dune	UNIBE-LHEP	less urgent	NGI_CH	on hold	2021-12-22	Local accounting for DUNE jobs at ...	EGI
150373		dune	UNIBE-LHEP	less urgent	NGI_CH	on hold	2021-12-22	Enable DUNE queue for CPU and future ...	EGI

• Attendants
• CSCS:
• CMS:
• ATLAS:
• LHCb:
• EGI:

• 2022-04.ReportCSCS.pdf: 2022-04.ReportCSCS.pdf