Sung, James, Joel

Aims

  • (CS22p6) CS v20 400 rerun (Sung)
    • Run everything
  • (CS22p4) 100m CS South Island  (Sung as James)
    • Run everything
  • (CS22p11) Faults Wellington Region (Joel)
    • Run complete
    • Backup results
    • Merge / plot ts
  • Hikurangi planar / curved simulations (James)
  • MPI for IM_calc (Joel) - PR
  • WCT Bug with max retries (Joel)

Picked up after SP

  • KISTI PBS update breaking workflow - temporary code fix in place. Complaints lodged and system reportedly fixed (untested)
  • KISTI 2023 accounts need to be created for direct data migration (by Jan 15)

Highlights

Used the full Maui allocation 103% - new allocation granted

27 Dec

26 Dec

25 Dec

24 Dec

23 Dec

22 Dec

21 Dec

20 Dec

19 Dec 2022:  Aiming  80% usage 

Completed


Issues

  • James (x2319a03) and Jason (x2319a05) accounts exceeded inode quota. Manually executed cleanup.sh scripts (should have been in task_config.yaml): LF data cleaned up TARed ok, .out/.err files lost.
  • PBS server replaced with PBS1, reporting the job id #int.pbs1, not #int.pbs, enforcing .pbs1 suffix for qstat query, which breaks the current pbs.py workflow.
  • HF crashing on new versions on Maui (input file error)
  • Kisti - tasks disappearing from DB for CS  : v22p6 maybe related. EMOD 1622/11855 completed, IM_calc 1714/11814 completed.  The progress is very slow

Outstanding

Initial Plan for next sprint - in priority order

  • Investigate possible cs22p4 sim data loss (cleanup.sh failure) (Sung)
  • (CS22p6) CS v20 400 rerun (Sung)
  • (CS22p4) 100m CS South Island  (Sung as James)
  • (CS22p11) Faults Wellington Region merge/plot ts (Joel)
  • Hikurangi planar / curved simulations (James)
  • MPI for IM_calc (Joel)
  • (CS22p11) Setup for rest of faults to run to slowburn Core Hours

Notes

Jason's status document: https://ucliveac-my.sharepoint.com/:w:/g/personal/jason_motha_canterbury_ac_nz/Ec7hr85qan9Lt0mmvbwiBTABelx6QiF1uLwbaZLp2SASTw?e=mTHiNY


Other – Jason's list of FFA backlog tasks

  • Add a step to the workflow for realisation summary script (will be a fault based step eventually) (and plotting)
  • Fix Rrup and Empirical calculation in the automated workflow (Low Priority)
  • Add realisation name to generic slurms (and IM_calc)


  • No labels