search

DUNE / data-mgmt-ops

3 stars 3 forks source link

Data Challenge: Rucio Readiness: Headroom on Rucio RSE #117

Closed StevenCTimm closed 1 year ago

StevenCTimm commented 2 years ago

Before the data challenge, clean out ~100TB of headroom on each of the big UK RSE (QMUL, RAL-PP, LANCASTER, MANCHESTER, RAL_ECHO) and make sure there is also 100TB of headroom at PRAGUE, NIKHEF, IN2P3, PIC.

Also arrange for 500TB of headroom at BNL (currently available) and at least 200TB headroom at the new CERN RSE for staging to/from CTA if necessary.

500TB of data expected in all:

Proposed split-up during the processing stage: 1 500TB copy to the distributed 5 UK sites 1 500TB copy to the distributed 4 non-UK European sites 1 500TB copy to BNL (from FNAL) 1 500TB copy stays resident at CERN EOS As much of the 500TB that stays in disk cache stays at FNAL Thus workflow allocator should have at least 5 places available from which to grab each file.

StevenCTimm commented 2 years ago

Have to be sure that the Rucio reaper is working on all sites where we need it.