Files
django-q2/README.rst
T
2015-06-18 18:59:31 +02:00

96 lines
2.6 KiB
ReStructuredText

Django Q
========
A multiprocessing task queue application for Django
---------------------------------------------------
Status
~~~~~~
In Alpha and Python 3 only (for now). Everything should work, but the
basic structure can still change. Main focus will be put on creating
tests now.
Architecture
~~~~~~~~~~~~
.. figure:: http://i.imgur.com/wTIeg2T.png
:alt: Django Q schema
Django Q schema
.. code:: python
async(func,*args,hook=None,**kwargs)
Signed Tasks
~~~~~~~~~~~~
Tasks are first pickled to Json and then signed using Django's own
signing module before being sent to a Redis list. This ensures that task
packages on the Redis server can only be excuted and read by clusters
and django servers who share the same secret key.
Optionally, packages can be compressed before transport by setting
``Q_COMPRESSED = True``
Pusher
~~~~~~
The pusher process continously checks the Redis list for new task
packages and pushes them on the Task Queue.
Worker
~~~~~~
A worker process checks the package signing, unpacks the task, executes
it and saves the return value. Irrespective of the failure or success of
any of these steps, the package is then pushed onto the Result Queue.
By default Django Q spawns a worker for each detected CPU on the host
system. This can be overridden by setting ``Q_WORKERS = n``. With *n*
being the numbe of desired worker processes.
Monitor
~~~~~~~
The result monitor checks the Result Queue for processed packages and
saves both failed and succesful packages to the Django database.
By default only the last 100 succesful packages are kept in the
database. This can be increased or decreased at will by settings
``Q_SAVE_LIMIT = n``. With *n* being the desired number of records. Set
``Q_SAVE_LIMIT = 0`` to save all results to the database. Failed
packages are always saved.
Sentinel
~~~~~~~~
The sentinel spawns all process and then checks the health of all
workers, including the pusher and the monitor. Reincarnating processes
if any may fail. In case of a stop signal, the sentinel will halt the
pusher and instruct the workers and monitor to finish the remaining
items , before exiting.
Hooks
~~~~~
Packages can be assigned a hook function, upon completion of the package
this function will be called with the Task object as the first argument.
Management command
~~~~~~~~~~~~~~~~~~
Start the cluster with ``./manage.py qcluster``
Admin integration
~~~~~~~~~~~~~~~~~
Django Q registers itself with the admin page to show failed and
succesful tasks. From there task results can be read or deleted. If
neccesary, failed tasks can be reintroduced to the queue.
Todo
~~~~
I'll add to this README while I'm developing the various parts.