How is it right to organize several parallel processes?



  • Let's see the case:

    There's a OD with parameters that pass the piton crypt.

    The python crypt, which at the inlet accepts the OBD parameter, and returns other parameters that are recorded in the same OBD, the spent parameter shall be marked.

    Objective: Initiate a few python crypts, without a repetition of reading/recording conflicts and marked parameters.

    BDE is sqlite, so it's locked during the recording.



  • Such a task can be used http://celery.readthedocs.org/en/latest/ And don't use sqlite.

    Celery is more or less well engaged in multi-point task management. They can be placed either on schedule with help http://celery.readthedocs.org/en/latest/userguide/periodic-tasks.html?highlight=periodic%20task#periodic-tasks or a violin. There's a possibility to re-entry the recording if the OBD was occupied.

    Celery's got a pretty good description. http://celery.readthedocs.org/en/latest/getting-started/first-steps-with-celery.html ♪

    From minus:

    • It's not always the most stable job.
    • As a broker, use only RabitMQ or Redis (if problems arise with other brokers, don't wait for support from the monitors)
    • There's a terrible code on the spot.

    And yet, it's a better solution than a multi-temperature or multitreed on its own.




Suggested Topics

  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2