search

jimfrimel / jfHWRF

My HWRF repository for tracking issues and tasks.
0 stars 0 forks source link

Review **Completion** task behavior #6

Open evankalina opened 5 years ago

evankalina commented 5 years ago

ie. 00 cycle ... is waiting on 06 ... cycle but that is a ways off ...need to get through all the hourly cycle ... ? should it still wait OR wait on NEXT cycle ?

ie. 01 cycle is complete ONLY Because 07 cycle does not exist ? Should it Check on next cycle, 02 ?? -c 201810090100 -t scrub_work_E99 dependencies

    AND is satisfied
      disk_archive_E99 of cycle 201810090100 is SUCCEEDED
      tape_archive_E99 of cycle 201810090100 is SUCCEEDED
      OR is satisfied
        NOT is satisfied
          cycle 201810090700 does not exist
evankalina commented 5 years ago

@jimfrimel Jason showed that this does not work quite yet. After 7 (hourly) cycles, the ICDA workflow will get stuck because not enough cycles are marked complete. The issue is related to the scrub_com and scrub_work tasks not running for the first set of five intercycles, even after the next (warm) synoptic cycle completes. I have some ideas on how to fix this; it might be pretty simple, but we need an existing ICDA experiment to test them (we can launch some fresh cycles with it and check the behavior). Let's talk once you're back :-)