This is a bare-bones guide to preparing an experiment for correlation on the CCC incarnation on juw32/juw31:

1. The machine name is ccsops (login as jops).
2. The experiment directories are: /ccs/expr/{EXP} [{EXP} = experiment name in capitals]
3. log2vex works as usual
   • log files live in: /ccs/var/log2vex/logexp_date/{EXP_yyyymmdd}/
     • a copy of the skd file will need to live here for log2vex to be able to make the correlation vex-file (read from skd).
     • the FS logs will need to live here if you want log2vex to be able to make the vex-file “from the logs”.
   • gps files live in: /ccs/var/log2vex/gps/
     • gps files from the station(s) will need to live here for the date of the experiment; otherwise log2vex will extrapolate based on files that do live there when completing the $CLOCKS section of the vex-file.
   • EOP files live in: /ccs/var/log2vex/iers/
     • up-to-date enough EOP files from IERS (either observed or predicted) to cover the dates of the experiment (-1, +1) need to live here for log2vex to be able to complete the $EOP section of the vex-file.
4. Prior to log2vex being able to work, the experiment needs to be added to a file called MASTER_PROJECTS.LIS in the directory /ccs/var/log2vex/
   • you can see the format from the other entries; try to keep this in chronologic order.
   • there are other steps required to keep the bookkeeping programs happy, so let the head Sci.Ops.Supp. know that you've done this, or better let him/her know you'd like an experiment added so that you don't have to do this yourself.
5. There are still some tweaks that are not fully automated in log2vex:
   1. antenna specification
      • Wb_array vs. Wb_single_dish(Wb7):
        • position in $SITE (/ccs/expr/Wb7.site contains the ITRF2000 Wb7 position (epoch 1997)
        • clock offset in $CLOCKS (subtract 1015us)
        • e-Bob's vex_editor (in the jops path) also takes care of these (use the tab “Wb conversion”)
      • Jb: in principle, either Jb1 or Jb2 could observe, regardless of which is in the skd file
      • Cm/Jv/Da: similarly, other MERLIN stations could replace Cm (there are currently no automated ways to handle this)
   2. Disk start times (in $TAPELOG_OBS) may reflect part of the data_good (usually for Tr, but sometimes for others); this could affect the first scan of the experiment or of each disk, since those scans would not be “wholly” contained within a single recording medium (solution: nudge the VSN start time to match the scan start time).
   3. sometimes the scan length for an individual station will exceed the length of the scan itself (edit the data stop field)
   4. other situations in more complicated experiments require some degree of manual intervention:
      • sub-netting
      • a station that happened to observe a significantly different schedule
      • VLBA participation
6. vexsum.pl works as usual
7. Starting runjob: “start_runjob.sh &”
   • runjob works as usual
8. Starting JCCS: “start_procs.sh {list of SUs to start} &”
   • SU list in format, e.g., “0d 1d 8d 12d” (the “d” for disk playback)
   1. start_procs.sh lives in /ccs/bin, which is in jops' path
   2. The same start_procs.sh edits would be required prior to starting JCCS (all these are found near the beginning of the start_procs.sh file):
      • turning on the CRM servo (for 16MHz subbands), e.g., export ENABLE_CRM_SERVO=yes (comment out if not wanted)
      • setting the destimation data disk on juw26, e.g., export DATA_DIR=${DATA_DIR:-/juw26_3/jops/data}
      • masking out correlator crates, e.g., export CORRSRCMASK=0x00000ff (the variable is a group of 8 hex-charactrer bit-masks, one for each of the 4-board “segments”, where everything is counted from 0 from right to left – the example says to use all of crate 0 only; CORRSRCMASK=0xff0600c0 would mean use all of crate 3, boards 1&2 in crate 2, none of crate 1, and boards 6&7 of crate 0.)
   3. Individual icons to not appear on screen for all the processes; instead they are automatically put minimized into the menu bar at the bottom of the screens. You know that things are ready to go once the weight plots appear, as before. You have to minimize these yourself if you want them so.
9. Shutting down JCCS: kill_procs.sh
   • For safety, you could still run psi & ipcs on ccsops and juw26