Notes from DUNE Site meeting 4/13/20
2 GGUS tickets
One at Manchester—not responded to as yet
One at PIC—resolved—they didn’t know about voms1 change, need to make sure they are on list.
NERSC—had 600 simultaneous nodes going, filled up scratch to 167% 34 TB
Steve will pull the deism files back to FNAL.
Problem with CVMFS library only on workers, still don’t know why but can work around.
Also a different failure in reco still being investigated.
ETF testing —still going, lots of auth failures, need to be investigated
2 SE’s still to onboard, RAL-PP and NIKHEF
Brazil
CBPF test jobs now working
Still working on getting a project on the supercomputer
Other institutions likely to join
Each one will have to send technical contact and institutional one to Heidi.
(2 other institutions planning to join)
In past APS has had money to collaborate with Brazil
Request to OSG factory to put GLIDEIN_DUNESite on all our sites
Change FE at Fermilab to use the DUNE proxy everywhere
Sam4Users—tom hasn’t been able to make it go
Ken—might want to think about longer pilot lifetimes—difficult to get jobs for prod3 going right now because we need longer than 24hr for data reconstruction, or we need more memory
Single core < 2.5GB and 24 hr run time we can’t use right now.=
At least 1/3 of data reco takes > 24 hrs
Are all target data files where we need them? Seems to be ok
After the target 8 runs will go to 1GeV data reco.
First pass of data reco doing totally off site.