ENG section meeting #5

513/R-068 (CERN)



Show room on map
ENG section meeting
Zoom Meeting ID
Simone Campana
Useful links
Join via phone
Zoom URL
    • 9:00 AM 9:10 AM
      News 10m

      Report for the IT-SC 20th July

      Technical Coordination


      • Still experiencing some synchronisation issues from Grappa to e-groups;

        • Some users getting only "<surname.name@cern.ch>" in the e-group "E-Mail Addresses" tab, others getting "Surname Name <surname.name@cern.ch>" => access issues through ADAMs

        • Number of users is wrong after Grappa/e-group sync. After removing all the members of the e-group, sync goes well; issue creating the problem seems fixed now, but no easy way to fix the existing ‘broken’ e-groups)

      • Phone/GSM issue: it was reported by two of the attendants to the meeting that the automatic switch to the CH country does not seem to work anymore. 

      • ASDF session last week: some presentation had only CodiMD version uploaded (need to remind speakers to add a static file)

      • Acron migration finished with the exception of a user on holidays (and 2FA in place)

      • Follow up on the Mac pdf-expert issue. Concern expressed: is Mac support under-staffed? (to be followed up internally)


      Small and Medium Experiments (SME) 

      • AMS requested clarification on the 2FA-WINs roll-out schedule and whether they were affected by that.

      • AMS: Started full reprocessing 3 weeks ago. Continue for 3 weeks more. No major problem. Activity being followed-up with the EOS team. 

      • NA61: Started data taking and successfully sending data to CTA. Planning a big reprocessing campaign and changing workflows to minimise recalls (in collaboration with the CTA team). Monte-carlo production continues. 

      • Compass: Data reconstruction working well, some data on tape was recalled over 3 days. 

      • ProtoDUNE is half way through a Data Challenge. Both detectors (VD and HD) reported heavy usage of FTS and EOS/CTA - which are performing very well.


      • Meeting scheduled with Computing Coordinators to better understand WebEOS usage and why it is a critical service.

      Theory department

      • The experience with the 1TB/256-core lxtheory9 node is so far very good. Some minor problems with the renewal of kerberos token from emacs sessions, being followed-up with experts.
      • Some issues signaled with the GPUs available in the batch service. The current allocation of 2GB per core is not enough for people doing Machine Learning activities, particularly when doing training. 


      • Linux desktop support model is still unclear for some use cases. Which are the privileged places/groups that get desktop support (e.g CCC), which is the eligibility criteria (e.g. Alice P2 desktops can ask for official support)? 

      • Some communities asked for more information and details about Microsoft and Google contracts. ie: How to onboard to Google Docs, Microsoft teams usage, zoom vs. MS teams, etc.

      • Volatile presentations in indico: would it make sense to have an auto-conversion service from CodiMD to PDF?

      • We are collecting information from experiments about the usage of GPUs: current usage and future plans. The aim is to know their needs, requirements and understand the workflows to better fit the type of GPUs and the way to expose them.

      WLCG Operations coordination

      • A dedicated meeting and further discussions were held between IT-PW-IAM, WLCG Ops Coordination and the IAM service development team at CNAF to agree plans for the evolution of our IAM instances in the next months, which were subsequently discussed in the WLCG Authorization WG and also feature in an update on the transition to tokens included in the July GDB agenda.

      • Collaborative efforts between IT-CD-CC and IT-GOV-ENG allowed a serious problem with our VOMS-Admin services to be identified, understood and corrected with little fallout. 

      • The proposal on how to address accounting and operational needs in SRR (Storage Resource Reporting) for the federated storage has been presented at the WLCG Ops Coordination meeting in July. The proposal suggests changes in the WLCG Storage Space Accounting application and will be validated on the NDGF federated storage.

      • Privacy notice for WLCG services. 

        • WLCG OPS coordination collects information from the T1 sites regarding publishing privacy notices. Some of them already publish privacy notice following the local regulation, some of them consider to customise WLCG generic privacy notice. 

        • The RoPO for CMS CRAB service has been created and submitted


      • A hardcoded work-around for GEANT CA certificates has been implemented in CRIC as a temporary fix for a bug in LDAP that omits whitespaces after commas in DN. This is not considered as a final solution because the problem might appear in other information coming from LDAP and in certificates coming from different CAs, therefore the fix on LDAP itself is desirable.

      • Migrated CRIC MONIT data sources from the PLAYGROUND MONIT project to the proper projects.

      • Multiple bug fixes and improvements have been implemented for monthly accounting reports and RRB plots generation

      EGL (Explorer of Grid Load)

      • The first official usage of the new dissemination tool EGL at the 5th of July event was very successful. The backend developed by the OPS team has provided all the necessary information and functionalities requested by Medialab who is responsible for the visualisation layer.

      • Work in close collaboration with Medialab is ongoing to further improve completeness and quality of data presented by EGL. Alice cores are now accounted more accurately and CRIC accounting information will be provided through the dedicated EGL backend.


      Collaborations, contributions with the experiments. Scientific Computing Activities.


      • Implemented ETF probe for LHCb to test WebDAV and GSIftp endpoints. Being tested, will get feedback from LHCb

      • Consolidated the ETF worker node tests for LHCb and removed some obsolete ones. Again, need to get feedback from LHCb before putting into production

      • (LHCb liaison) Prepared with the LHCb computing coordinator the LHCb request for “ARM and Power servers needed for the build of the LHCb software access to HPC centres to consolidate the integration of these centres in the WMS of the experiment” The request has been submitted to the IT steering committee and has been accepted.

      • (LHCb liaison) Preparation of the Job description for the Fellow on WLCG Monitoring improvement (LHCb monitoring and other experiments)


      • Started to make the code of the Hammercloud plugin for CMS’ CRAB3 Python3-compliant. 

      • Worked with a colleague from IT-CS on understanding residual IPv4 traffic in the CMS submission infrastructure. Issues identified and communicated to CMS.


      • Organised and chaired the session on Analytics Coordination in ADC at the ATLAS Software and Computing week. 


      • Domenico Giordano has taken over the role of co-chair of the HEP-SCORE Task force 

      • Work on Benchmarks progressed. Focus is still on CPUs, but there are now also prototype benchmarks from CMS, Madgraph and ML using GPUs.

      • Planning the HEPscore Workshop (19-20 September), draft agenda exists. 

      • Testing of the HEP Benchmark Suite in RedHat 8 started

      Data analytics

      • As a follow up of the presentation at the OpenInfra Summit in Berlin "Anomaly Detection in CERN Cloud" a couple of companies have asked to get more insights about the tools and methods we have adopted, with the purpose of investigating their adoption for their use cases


      • Two members of the team participated in the HSF workshop on the analysis ecosystem in Paris. We had discussions on how we could understand the impact of Analysis Facilities on resources and users quantitatively. These discussions continued after the workshop. 

      • Contributions to a proposal In the context of the CERN CIPEA programme. The project’s goal is to to optimise the power consumption of HEP applications. Therefore a set of metrics are needed to assess the power use. The project proposes to start by measuring the electrical power consumption of HEP software applications and compare the consumption on different architectures, e.g. CPU, GPU, etc. 

      • Summer student arrived, started to work on implementing different caching algorithms. 

    • 9:10 AM 10:00 AM
      Round Table 50m