AI_INFN Technical Meeting

Monday 16 Sept 2024, 16:00 → 17:10 Europe/Rome

Virtual meeting room (zoom): https://l.infn.it/ai-infn-meeting

Date: 2024-09-16

News:

• The agenda of the Workshop “Computing@CSN5” applications and innovations at INFN is
  online
• The AI_INFN Platform has been unavailable from 11/9 evening to 12/9 evening due to a major technical incident with the CNAF internal network. The problem was limited to the network: no data was lost.

Tracked developments:

 Automation of RKE2 deployments in INFN Cloud

• NTR
• Giuseppe è stato invitato ad un meeting con INFN Cloud in cui Giuseppe ha presentato un modulo puppet per tirar su il dependency tracker. Inclusa la parte di IaaC. Perplessità:
  • è l’ennesimo servizio da servire centralmente;
  • va capito se si può fare con Harbor e non lo farebbe per una fase iniziale ma solo per le immagini pubblicate.
• Vogliamo continuare su risorse AI_INFN?
• L’update di kubernetes è stato affiancato con un Playbook Ansible, Giuseppe farà un seminario.

 Develop monitoring and accounting infrastructure (R. Petrini)

• [Low priority] We need to update the grafana instance to accomodate metrics for the hackathon platform

 Environment setup (M. Barbetti, S. Giagu, S. Bordoni, L. Cappelli)

• NTR

 Offloading tests with virtual kubelets (G. Bianchini, D. Ciangottini)

• Started the activity to submit jobs to CINECA Leonardo through the INFN Platform, we are debugging.
• Started a discussion with CNAF Tier-1 support to submit jobs there, a possibly viable option was identified, exercises to validate the technical requirements will start soon.
• Activity with Cloud@Veneto has been paused after migration to interlink 0.3 and should be revived.

A helm chart with monitoring infrastructure is being prepared by Giulio.

 Acquisto FPGA

• Macchinone 4 has been installed with:
  • 1 RTX 5000
  • 2 AMD Xilinx Alveo V70
• We have started the very first exercises with OpenStack to provision the FPGA in pass-through, but very limited effort has been available in August (volunteers?)

 Advanced Hackathon

• The First Advanced Hackathon of AI_INFN will be in Padua, on 26, 27 and 28 November 2024.

• A preliminary indico page has been created.

• The call for exercises is open. We have 3 volunteers for 4 slots:

  • Quantum Machine Learning
  • AI for muography, became unavailable
  • AI for medical imaging

• Reminder: deadline for 23/9 to define the exercises.

Status legend

 Active
 Priority
 Problems
 Postponed or Blocked by others
 Completed

16:00 → 16:15 News and setup

Speaker: Lucio Anderlini (Istituto Nazionale di Fisica Nucleare)

16:15 → 16:50 Discussion on tasks and priorities

Speaker: All

16:50 → 17:00 Any other business