1. 29 Jul, 2019 1 commit
  2. 24 Jul, 2019 1 commit
    • Paul Rich's avatar
      Lock doesn't need to be held around this state fetch · d658914d
      Paul Rich authored
      This is to improve Cobalt responsiveness during state update.  Combined
      with the fix for aig/cobalt#177, this should allow for much faster
      scheduling cadences as well and should significantly reduce the delay in
      a number of commands.
      
      There is an additional fix here where the check for a starting job
      during the cleanup update wasn't working.  This was due to a hidden type
      mismatch.  This was exposed by both speedups put together.
      d658914d
  3. 22 Jul, 2019 1 commit
    • Paul Rich's avatar
      Adding config option for sleeptime for componnents · 567eeec9
      Paul Rich authored
      This allows the site to configure the base sleeptime between automatic
      checks on a per-component basis.
      
      This removes the argument and relies on the config file.  Nothing was
      setting the option prior to this, and this is a more consistient way to
      handle this configuration.
      567eeec9
  4. 21 Jun, 2019 1 commit
  5. 17 May, 2019 2 commits
  6. 16 May, 2019 1 commit
  7. 15 Mar, 2019 1 commit
  8. 14 Mar, 2019 1 commit
  9. 12 Feb, 2019 1 commit
    • Paul Rich's avatar
      Start time estimate with unit tests and manpage update · 89183eab
      Paul Rich authored
      This adds doc, unit tests and the starttime estimate.  This provides
      a naive estimate based on machine hours ahead of a job, score-wise,
      between the running jobs and currently eligible jobs.  Jobs in hold
      do not get estimates, and putting a job in hold removes the estimate.
      
      This also enforces a minimum interval ahead for start times.  This
      is adjustable in the cobalt config.
      
      This is a queue-data only estimate.  Cobalt is not prebuliding a
      schedudle and the scheduler behavior is not impacted by this.
      
      This will only show up by default with qstat -fl.  This may also
      be requested via the header flag.  This is to prevent current user
      script breakage.
      89183eab
  10. 19 Dec, 2018 1 commit
  11. 10 Dec, 2018 1 commit
  12. 07 Dec, 2018 2 commits
  13. 06 Dec, 2018 1 commit
  14. 29 Nov, 2018 1 commit
  15. 28 Nov, 2018 2 commits
  16. 27 Nov, 2018 9 commits
  17. 30 Oct, 2018 1 commit
  18. 26 Oct, 2018 2 commits
  19. 16 Oct, 2018 1 commit
    • Paul Rich's avatar
      Adjustments to accounting log for reservations from consumer feedback · d1d96cd2
      Paul Rich authored
      Got feedback on the reservation accounting record data.  Changes
      1) The default resource is "NOTSET" chosen to avoid future parsing
         it's easy to spot that you didn't set this in cobalt.conf.
      2) active_id is added to Y and F records (confirm and finsh for
         reservations.) While not strictly necessary since a reservation
         shouldn't be active at that point, it can provide a useful event
         binding in the event of an in-flight reservation modification.
      d1d96cd2
  20. 03 Oct, 2018 2 commits
  21. 02 Oct, 2018 1 commit
  22. 01 Oct, 2018 1 commit
    • Paul Rich's avatar
      Removing kill_job message · 7a9baaab
      Paul Rich authored
      This is already handled by the 'D' message.  'D' will be getting a flag
      to indicate a force-delete.
      7a9baaab
  23. 28 Sep, 2018 1 commit
  24. 26 Sep, 2018 1 commit
  25. 25 Sep, 2018 2 commits
    • Paul Rich's avatar
      Fixing a traceback from cyclic reservations · 92a4a333
      Paul Rich authored
      If the is_over check cycled the reservation, the running flag wasn't
      getting set to False, causing a traceback to occur when the is_active
      check happens later, as stime was set to None at this point (stime is
      None in a reservation that has deactivated, and is only set while
      active).
      
      We would then get a loop where the reservation would never progress
      because the log message would fail due to stime being set to None (which
      it should be for a dormant reservation).
      92a4a333
    • Paul Rich's avatar
      f6718c32
  26. 18 Sep, 2018 1 commit
    • Paul Rich's avatar
      Owner now is the list rather than first character · 2f542278
      Paul Rich authored
      self.users is a ':'-delimited list as a string.  This was pulling the
      first character for "owener".  At this point, since any user on a
      reservation can run commands, "owners" and "users" are the same thing
      here.
      2f542278