Attendants: Xavi, Ignacio, Geoff, Elisabetta, Steve, Stu
* Question regarding cernbox as substitute for AFS home directories
* Position is to continue migrating certain use cases but no urgency to move out of AFS in cases of high performance shared file system needs.
* Strategy will be further defined through the year
* Bunny will be at CERN in the coming weeks and would be the right person to discuss regarding CC8 and the desktop support group
Storage space status update for NP02 and NP04: Usage predictions short and mid-term
Steve: 90% used quota (2PB physical). No reconstruction data there yet. Trying to reduce targeting 1PB
Elisabetta: Approx the same quota usage. Needed 50% of available space. Currently cleaning after each data transfer.
NP02 Status update towards data taking (and plans for the DC)
Elisabetta: Kick-off meeting with Steve. Practical details, formats, conventions etc. We have a lot of computing quota
in Lyon datacenter which will be used during the DC to include computing with MC files (using 3GB datafiles).
Set up UI machine with FTS. All machines seem to be in place.
* The defined goal is to sustain a constant dataflow of from NP02 up to EOS->EOS IT up to FNAL (dchache to castor) during 24hours
* The target transfer data is still not fixed, we are verifying what are the options.
* Looking for peak values would be also desirable
Steve 4.4 GB/sec reached, not peak. But the question is if data can be generated that fast
2GB/sec for single phase data was reached yesterday
Estimated time to run the DC to be performed before next Friday
* Elisabetta: There will be an intervention next week in EOS
Steve: We still have to do some operations on the Rucio client on our side. Moving to use upload files to configure the jobs
Elisabetta: After running the reco jobs we will write to local storage element in Lyon (new in this DC)
Xavi: Looking forward to see the effect of caching in the performance of the jobs. In WLCG we are running a working group focused on content delivery and latency hiding. Related to this we are organising a pre-GDB meeting on XCache focused on caching and latency hiding tests and prototypes. Once is official I'll send the link to the list.
Works in the cryostat is progressing. Sealing expected to happen in time for start filling in July.
We'll be there next week for DAQ week. We'll look into some issues.
The global objetive is to move to the DUNE model for the NP04 computing model
* DC data will be in a different Castor path. Do we need to contact the team?
Xavi: Yes please
Question regarding Castor -> CTA transition:
There is a prototype for ATLAS that ought to be ready for Q3-Q4 this year. CMS, LHCb and ALICE will be migrated during 2020. For the experiments currently living in castorpublic each experiment will be contacted individually by the Castor team at a later time, the migration will start only once the four LHC experiments have been moved to CTA.
* Monday is holiday at CERN and in France
* Next Meeting proposed for the 20th June 2019