This project is mirrored from https://:*****@github.com/Ensembl/ensembl-hive.git. Pull mirroring updated .
  1. 17 May, 2017 2 commits
  2. 24 Apr, 2017 1 commit
  3. 12 Jan, 2017 1 commit
  4. 02 Dec, 2016 1 commit
  5. 06 Jul, 2016 6 commits
  6. 10 Jun, 2016 1 commit
    • Matthieu Muffato's avatar
      bugfix for the infinite loop on "unreachable" workers · c115c8a7
      Matthieu Muffato authored
      Queen::check_for_dead_workers() calls Valley::status_of_all_our_workers_by_meadow_signature()
      and checks there not only the status of the workers, but also which meadows
      are available.  In 2.4, this used to contain empty hashes for reachable
      meadows with no workers, but this behaviour was broken on master when I
      refactored parts of Valley and Meadow.  In case there was no running
      workers, there wouldn't even be an entry for the meadow, and Queen was
      considering the meadow to be unreachable.  If at the same time a worker is
      killed by LSF on this meadow, beekeeper wasn't able to understand it could
      bacct it.
      c115c8a7
  7. 24 May, 2016 1 commit
  8. 20 May, 2016 1 commit
    • Matthieu Muffato's avatar
      New Meadow/Valley interface that detects slow workers and mark them as pending · 14fcdaa5
      Matthieu Muffato authored
      This fixes @danstaines's issue:
      [Feb 11th at 09:49]
      I've got a problem with hive where I supply a registry, and the registry takes
      a couple of minutes to load (this is normal with 40k bacteria unfortunately).
      
      The registry takes a lot of time to load, so the worker is running from the LSF
      point-of-view but is not registered in the database yet (because the DBAdaptor
      object is not ready yet). Because of the discrepancy, beekeeper thinks that
      nothing is running and keep on submitting new workers (beyond the analysis
      capacity)
      
      My fix is to basically count those workers as pending. Because the
      Valley/Meadow had a method to get the running workers and another one to get
      the pending workers, I had to unify them into a single method which saves us
      one call to _bjobs_. Then the Valley adjusts the counts into a complex hash
      that contains accurate information about all the workers. This super-hash is
      then carried around in various places so that all the decisions are consistent
      to each other. Overall the Valley now does more things and the Meadows are a
      bit shorter
      
      The test-suite has been updated accordingly
      14fcdaa5
  9. 03 Feb, 2016 1 commit
  10. 29 Jan, 2016 1 commit
  11. 12 Mar, 2015 1 commit
  12. 23 Jan, 2015 1 commit
  13. 22 Jan, 2015 1 commit
  14. 13 Jan, 2015 1 commit
  15. 07 Jan, 2015 1 commit
  16. 29 Apr, 2014 1 commit
  17. 24 Jan, 2014 2 commits
  18. 09 Dec, 2013 1 commit
  19. 01 Jun, 2013 1 commit
  20. 28 May, 2013 1 commit
  21. 23 Nov, 2012 1 commit
  22. 20 Nov, 2012 2 commits
  23. 16 Nov, 2012 1 commit
  24. 31 May, 2012 1 commit
  25. 23 May, 2012 2 commits
  26. 22 May, 2012 1 commit
  27. 21 May, 2012 1 commit
  28. 18 May, 2012 2 commits
  29. 12 May, 2012 1 commit
  30. 11 May, 2012 1 commit