VLASS SE Imaging Operations Meeting 20-9-22

Date and Time

22 Sep 202016:00 EDT / 14:00 MDT

Video: https://zoom.us/j/5646363780?pwd=VGxrUGFrTTUrUmZHRGQ1L3poMTN6dz09 

Main page: VLASS SE Imaging operations

Roadmap: SE Imaging Operations Roadmap

External processing page: VLASS SE External Processing

HTCondor area: High Throughput Computing with HTCondor

VIP Script timing info: https://docs.google.com/spreadsheets/d/1yGpNNexL9YdEST3-_zyqtNqK2UxSNvmzKFOeRyanOHQ/edit#gid=0

All tickets pertaining to VLASS SE operations should have the VLASS-SE-OPS label.  Add VLASS-SE-DISCUSS label if it should be brought up in this meeting

Open Issues

• Need to determine whether pybdsf can be run from within CASA,  possibly it will work with CASA6 derivative. 
• Possibly a problem with the smaller Josh provided data set with CASA6
• Need to have smaller data set's cfcache's preloaded on VLASS nodes

Decisions

• Josh Marvil's test data set includes requisite CFCaches.  It reduces flops and memory via fewer channels and smaller image size, completes with 8-way parallelzation through VIP workflow in ~24hours.

Agenda

• Pipeline
  • Initial CASA6 based tests show pronounced speed up, SCG to help investigate.
• SSA
  

  • Is the proposed directory structure (/lustre/aoc/cluster/pipeline/vlass_auto/cache/pims/EB/CAL/DEC/RA) acceptable to everyone

  • What people want to see as directory components for EBs, calibrations, RA and dec for the phase center. Less essential since it will be mostly managed by the tool

  • what syntax do people want to use on the CLI to input this information. 

• SCG
  • Processing 47 + 7 repeat square degrees in AWS

Action items

•  Brian Kentor Laszlo Szucs [X] to point to failed execution for Josh's smaller data set
• James Robnett [X] to provide qsub call for 2x8 parallelization