##// END OF EJS Templates
Make logging unicode-aware...
Make logging unicode-aware Closes gh-1777

File last commit:

r6472:89a00dba
r7081:0eeff21b
Show More
parallel_task.txt
471 lines | 17.4 KiB | text/plain | TextLexer
MinRK
update API after sagedays29...
r3664 .. _parallel_task:
MinRK
clone parallel docs to parallelz
r3586
==========================
The IPython task interface
==========================
MinRK
dependency tweaks + dependency/scheduler docs
r3624 The task interface to the cluster presents the engines as a fault tolerant,
MinRK
initial draft of core zmq.parallel docs
r3591 dynamic load-balanced system of workers. Unlike the multiengine interface, in
MinRK
dependency tweaks + dependency/scheduler docs
r3624 the task interface the user have no direct access to individual engines. By
allowing the IPython scheduler to assign work, this interface is simultaneously
simpler and more powerful.
MinRK
clone parallel docs to parallelz
r3586
MinRK
dependency tweaks + dependency/scheduler docs
r3624 Best of all, the user can use both of these interfaces running at the same time
MinRK
initial draft of core zmq.parallel docs
r3591 to take advantage of their respective strengths. When the user can break up
the user's work into segments that do not depend on previous execution, the
task interface is ideal. But it also has more power and flexibility, allowing
the user to guide the distribution of jobs, without having to assign tasks to
MinRK
clone parallel docs to parallelz
r3586 engines explicitly.
Starting the IPython controller and engines
===========================================
To follow along with this tutorial, you will need to start the IPython
controller and four IPython engines. The simplest way of doing this is to use
MinRK
rebase IPython.parallel after removal of IPython.kernel...
r3672 the :command:`ipcluster` command::
MinRK
clone parallel docs to parallelz
r3586
MinRK
update docs to reflect relaxed syntax of argparse
r4608 $ ipcluster start -n 4
MinRK
clone parallel docs to parallelz
r3586
For more detailed information about starting the controller and engines, see
Fernando Perez
Lots more edits to release notes, nearing completion.
r4435 our :ref:`introduction <parallel_overview>` to using IPython for parallel computing.
MinRK
clone parallel docs to parallelz
r3586
MinRK
update parallel docs with some changes from scipy tutorial...
r5169 Creating a ``LoadBalancedView`` instance
========================================
MinRK
clone parallel docs to parallelz
r3586
MinRK
organize IPython.parallel into subpackages
r3673 The first step is to import the IPython :mod:`IPython.parallel`
MinRK
split get_results into get_result/result_status, add AsyncHubResult
r3639 module and then create a :class:`.Client` instance, and we will also be using
a :class:`LoadBalancedView`, here called `lview`:
MinRK
clone parallel docs to parallelz
r3586
.. sourcecode:: ipython
MinRK
move IPython.zmq.parallel to IPython.parallel
r3666 In [1]: from IPython.parallel import Client
MinRK
Doc tweaks and updates
r3663
MinRK
move IPython.zmq.parallel to IPython.parallel
r3666 In [2]: rc = Client()
MinRK
initial draft of core zmq.parallel docs
r3591
MinRK
clone parallel docs to parallelz
r3586
MinRK
initial draft of core zmq.parallel docs
r3591 This form assumes that the controller was started on localhost with default
configuration. If not, the location of the controller must be given as an
argument to the constructor:
MinRK
clone parallel docs to parallelz
r3586
.. sourcecode:: ipython
MinRK
initial draft of core zmq.parallel docs
r3591 # for a visible LAN controller listening on an external port:
MinRK
move IPython.zmq.parallel to IPython.parallel
r3666 In [2]: rc = Client('tcp://192.168.1.16:10101')
MinRK
Doc tweaks and updates
r3663 # or to connect with a specific profile you have set up:
MinRK
move IPython.zmq.parallel to IPython.parallel
r3666 In [3]: rc = Client(profile='mpi')
MinRK
Doc tweaks and updates
r3663
MinRK
update API after sagedays29...
r3664 For load-balanced execution, we will make use of a :class:`LoadBalancedView` object, which can
be constructed via the client's :meth:`load_balanced_view` method:
MinRK
Doc tweaks and updates
r3663
.. sourcecode:: ipython
MinRK
update API after sagedays29...
r3664 In [4]: lview = rc.load_balanced_view() # default load-balanced view
MinRK
Doc tweaks and updates
r3663
.. seealso::
MinRK
initial draft of core zmq.parallel docs
r3591
MinRK
Doc tweaks and updates
r3663 For more information, see the in-depth explanation of :ref:`Views <parallel_details>`.
MinRK
initial draft of core zmq.parallel docs
r3591
MinRK
clone parallel docs to parallelz
r3586
Quick and easy parallelism
==========================
In many cases, you simply want to apply a Python function to a sequence of
MinRK
initial draft of core zmq.parallel docs
r3591 objects, but *in parallel*. Like the multiengine interface, these can be
implemented via the task interface. The exact same tools can perform these
actions in load-balanced ways as well as multiplexed ways: a parallel version
of :func:`map` and :func:`@parallel` function decorator. If one specifies the
MinRK
API update involving map and load-balancing
r3635 argument `balanced=True`, then they are dynamically load balanced. Thus, if the
MinRK
initial draft of core zmq.parallel docs
r3591 execution time per item varies significantly, you should use the versions in
the task interface.
MinRK
clone parallel docs to parallelz
r3586
Parallel map
------------
MinRK
split get_results into get_result/result_status, add AsyncHubResult
r3639 To load-balance :meth:`map`,simply use a LoadBalancedView:
MinRK
clone parallel docs to parallelz
r3586
.. sourcecode:: ipython
MinRK
split get_results into get_result/result_status, add AsyncHubResult
r3639
In [62]: lview.block = True
MinRK
move parallel doc figures into 'figs' subdir...
r5168 In [63]: serial_result = map(lambda x:x**10, range(32))
MinRK
clone parallel docs to parallelz
r3586
MinRK
move parallel doc figures into 'figs' subdir...
r5168 In [64]: parallel_result = lview.map(lambda x:x**10, range(32))
MinRK
clone parallel docs to parallelz
r3586
MinRK
move parallel doc figures into 'figs' subdir...
r5168 In [65]: serial_result==parallel_result
Out[65]: True
MinRK
clone parallel docs to parallelz
r3586
Parallel function decorator
---------------------------
Parallel functions are just like normal function, but they can be called on
sequences and *in parallel*. The multiengine interface provides a decorator
that turns any Python function into a parallel function:
.. sourcecode:: ipython
MinRK
initial draft of core zmq.parallel docs
r3591 In [10]: @lview.parallel()
MinRK
clone parallel docs to parallelz
r3586 ....: def f(x):
....: return 10.0*x**4
MinRK
dependency tweaks + dependency/scheduler docs
r3624 ....:
MinRK
clone parallel docs to parallelz
r3586
MinRK
initial draft of core zmq.parallel docs
r3591 In [11]: f.map(range(32)) # this is done in parallel
Out[11]: [0.0,10.0,160.0,...]
MinRK
clone parallel docs to parallelz
r3586
MinRK
update API after sagedays29...
r3664 .. _parallel_dependencies:
MinRK
dependency tweaks + dependency/scheduler docs
r3624 Dependencies
============
Often, pure atomic load-balancing is too primitive for your work. In these cases, you
may want to associate some kind of `Dependency` that describes when, where, or whether
a task can be run. In IPython, we provide two types of dependencies:
`Functional Dependencies`_ and `Graph Dependencies`_
.. note::
It is important to note that the pure ZeroMQ scheduler does not support dependencies,
and you will see errors or warnings if you try to use dependencies with the pure
scheduler.
Functional Dependencies
-----------------------
Functional dependencies are used to determine whether a given engine is capable of running
a particular task. This is implemented via a special :class:`Exception` class,
MinRK
move IPython.zmq.parallel to IPython.parallel
r3666 :class:`UnmetDependency`, found in `IPython.parallel.error`. Its use is very simple:
MinRK
dependency tweaks + dependency/scheduler docs
r3624 if a task fails with an UnmetDependency exception, then the scheduler, instead of relaying
the error up to the client like any other error, catches the error, and submits the task
to a different engine. This will repeat indefinitely, and a task will never be submitted
to a given engine a second time.
You can manually raise the :class:`UnmetDependency` yourself, but IPython has provided
some decorators for facilitating this behavior.
There are two decorators and a class used for functional dependencies:
.. sourcecode:: ipython
MinRK
organize IPython.parallel into subpackages
r3673 In [9]: from IPython.parallel import depend, require, dependent
MinRK
dependency tweaks + dependency/scheduler docs
r3624
@require
********
The simplest sort of dependency is requiring that a Python module is available. The
``@require`` decorator lets you define a function that will only run on engines where names
you specify are importable:
.. sourcecode:: ipython
In [10]: @require('numpy', 'zmq')
MinRK
update parallel docs with some changes from scipy tutorial...
r5169 ....: def myfunc():
....: return dostuff()
MinRK
dependency tweaks + dependency/scheduler docs
r3624
Now, any time you apply :func:`myfunc`, the task will only run on a machine that has
MinRK
updates to docs and examples
r3670 numpy and pyzmq available, and when :func:`myfunc` is called, numpy and zmq will be imported.
MinRK
dependency tweaks + dependency/scheduler docs
r3624
@depend
*******
The ``@depend`` decorator lets you decorate any function with any *other* function to
evaluate the dependency. The dependency function will be called at the start of the task,
and if it returns ``False``, then the dependency will be considered unmet, and the task
will be assigned to another engine. If the dependency returns *anything other than
``False``*, the rest of the task will continue.
.. sourcecode:: ipython
In [10]: def platform_specific(plat):
MinRK
update parallel docs with some changes from scipy tutorial...
r5169 ....: import sys
....: return sys.platform == plat
MinRK
dependency tweaks + dependency/scheduler docs
r3624
In [11]: @depend(platform_specific, 'darwin')
MinRK
update parallel docs with some changes from scipy tutorial...
r5169 ....: def mactask():
....: do_mac_stuff()
MinRK
dependency tweaks + dependency/scheduler docs
r3624
In [12]: @depend(platform_specific, 'nt')
MinRK
update parallel docs with some changes from scipy tutorial...
r5169 ....: def wintask():
....: do_windows_stuff()
MinRK
dependency tweaks + dependency/scheduler docs
r3624
In this case, any time you apply ``mytask``, it will only run on an OSX machine.
``@depend`` is just like ``apply``, in that it has a ``@depend(f,*args,**kwargs)``
signature.
dependents
**********
You don't have to use the decorators on your tasks, if for instance you may want
to run tasks with a single function but varying dependencies, you can directly construct
the :class:`dependent` object that the decorators use:
.. sourcecode::ipython
In [13]: def mytask(*args):
MinRK
update parallel docs with some changes from scipy tutorial...
r5169 ....: dostuff()
MinRK
dependency tweaks + dependency/scheduler docs
r3624
In [14]: mactask = dependent(mytask, platform_specific, 'darwin')
# this is the same as decorating the declaration of mytask with @depend
# but you can do it again:
In [15]: wintask = dependent(mytask, platform_specific, 'nt')
# in general:
In [16]: t = dependent(f, g, *dargs, **dkwargs)
# is equivalent to:
In [17]: @depend(g, *dargs, **dkwargs)
MinRK
update parallel docs with some changes from scipy tutorial...
r5169 ....: def t(a,b,c):
....: # contents of f
MinRK
dependency tweaks + dependency/scheduler docs
r3624
Graph Dependencies
------------------
Sometimes you want to restrict the time and/or location to run a given task as a function
of the time and/or location of other tasks. This is implemented via a subclass of
:class:`set`, called a :class:`Dependency`. A Dependency is just a set of `msg_ids`
corresponding to tasks, and a few attributes to guide how to decide when the Dependency
has been met.
The switches we provide for interpreting whether a given dependency set has been met:
any|all
Whether the dependency is considered met if *any* of the dependencies are done, or
only after *all* of them have finished. This is set by a Dependency's :attr:`all`
boolean attribute, which defaults to ``True``.
MinRK
update API after sagedays29...
r3664 success [default: True]
Whether to consider tasks that succeeded as fulfilling dependencies.
failure [default : False]
Whether to consider tasks that failed as fulfilling dependencies.
using `failure=True,success=False` is useful for setting up cleanup tasks, to be run
only when tasks have failed.
Sometimes you want to run a task after another, but only if that task succeeded. In this case,
``success`` should be ``True`` and ``failure`` should be ``False``. However sometimes you may
not care whether the task succeeds, and always want the second task to run, in which case you
should use `success=failure=True`. The default behavior is to only use successes.
MinRK
dependency tweaks + dependency/scheduler docs
r3624
There are other switches for interpretation that are made at the *task* level. These are
specified via keyword arguments to the client's :meth:`apply` method.
after,follow
You may want to run a task *after* a given set of dependencies have been run and/or
run it *where* another set of dependencies are met. To support this, every task has an
`after` dependency to restrict time, and a `follow` dependency to restrict
destination.
timeout
You may also want to set a time-limit for how long the scheduler should wait before a
task's dependencies are met. This is done via a `timeout`, which defaults to 0, which
indicates that the task should never timeout. If the timeout is reached, and the
scheduler still hasn't been able to assign the task to an engine, the task will fail
with a :class:`DependencyTimeout`.
.. note::
Dependencies only work within the task scheduler. You cannot instruct a load-balanced
task to run after a job submitted via the MUX interface.
MinRK
update API after sagedays29...
r3664 The simplest form of Dependencies is with `all=True,success=True,failure=False`. In these cases,
MinRK
dependency tweaks + dependency/scheduler docs
r3624 you can skip using Dependency objects, and just pass msg_ids or AsyncResult objects as the
`follow` and `after` keywords to :meth:`client.apply`:
.. sourcecode:: ipython
In [14]: client.block=False
MinRK
update API after sagedays29...
r3664 In [15]: ar = lview.apply(f, args, kwargs)
MinRK
dependency tweaks + dependency/scheduler docs
r3624
MinRK
update API after sagedays29...
r3664 In [16]: ar2 = lview.apply(f2)
MinRK
dependency tweaks + dependency/scheduler docs
r3624
MinRK
update parallel docs with some changes from scipy tutorial...
r5169 In [17]: with lview.temp_flags(after=[ar,ar2]):
....: ar3 = lview.apply(f3)
MinRK
dependency tweaks + dependency/scheduler docs
r3624
MinRK
update parallel docs with some changes from scipy tutorial...
r5169 In [18]: with lview.temp_flags(follow=[ar], timeout=2.5)
....: ar4 = lview.apply(f3)
MinRK
dependency tweaks + dependency/scheduler docs
r3624
.. seealso::
Some parallel workloads can be described as a `Directed Acyclic Graph
<http://en.wikipedia.org/wiki/Directed_acyclic_graph>`_, or DAG. See :ref:`DAG
Dependencies <dag_dependencies>` for an example demonstrating how to use map a NetworkX DAG
onto task dependencies.
Impossible Dependencies
***********************
The schedulers do perform some analysis on graph dependencies to determine whether they
are not possible to be met. If the scheduler does discover that a dependency cannot be
met, then the task will fail with an :class:`ImpossibleDependency` error. This way, if the
scheduler realized that a task can never be run, it won't sit indefinitely in the
scheduler clogging the pipeline.
The basic cases that are checked:
* depending on nonexistent messages
* `follow` dependencies were run on more than one machine and `all=True`
MinRK
update API after sagedays29...
r3664 * any dependencies failed and `all=True,success=True,failures=False`
* all dependencies failed and `all=False,success=True,failure=False`
MinRK
dependency tweaks + dependency/scheduler docs
r3624
.. warning::
This analysis has not been proven to be rigorous, so it is likely possible for tasks
to become impossible to run in obscure situations, so a timeout may be a good choice.
MinRK
add db,resubmit/retries docs
r3876
Retries and Resubmit
====================
Retries
-------
Another flag for tasks is `retries`. This is an integer, specifying how many times
a task should be resubmitted after failure. This is useful for tasks that should still run
if their engine was shutdown, or may have some statistical chance of failing. The default
is to not retry tasks.
Resubmit
--------
Sometimes you may want to re-run a task. This could be because it failed for some reason, and
you have fixed the error, or because you want to restore the cluster to an interrupted state.
For this, the :class:`Client` has a :meth:`rc.resubmit` method. This simply takes one or more
msg_ids, and returns an :class:`AsyncHubResult` for the result(s). You cannot resubmit
a task that is pending - only those that have finished, either successful or unsuccessful.
MinRK
reflect revised apply_bound pattern
r3655 .. _parallel_schedulers:
MinRK
dependency tweaks + dependency/scheduler docs
r3624 Schedulers
==========
There are a variety of valid ways to determine where jobs should be assigned in a
load-balancing situation. In IPython, we support several standard schemes, and
MinRK
parallel docs, tests, default config updated to newconfig
r3990 even make it easy to define your own. The scheme can be selected via the ``scheme``
argument to :command:`ipcontroller`, or in the :attr:`TaskScheduler.schemename` attribute
MinRK
dependency tweaks + dependency/scheduler docs
r3624 of a controller config object.
The built-in routing schemes:
MinRK
reflect revised apply_bound pattern
r3655 To select one of these schemes, simply do::
Thomas Kluyver
Update command line args format in parallel docs section.
r4196 $ ipcontroller --scheme=<schemename>
MinRK
reflect revised apply_bound pattern
r3655 for instance:
Thomas Kluyver
Update command line args format in parallel docs section.
r4196 $ ipcontroller --scheme=lru
MinRK
reflect revised apply_bound pattern
r3655
MinRK
dependency tweaks + dependency/scheduler docs
r3624 lru: Least Recently Used
Always assign work to the least-recently-used engine. A close relative of
round-robin, it will be fair with respect to the number of tasks, agnostic
with respect to runtime of each task.
plainrandom: Plain Random
MinRK
reflect revised apply_bound pattern
r3655
MinRK
dependency tweaks + dependency/scheduler docs
r3624 Randomly picks an engine on which to run.
twobin: Two-Bin Random
MinRK
reflect revised apply_bound pattern
r3655 **Requires numpy**
MinRK
dependency tweaks + dependency/scheduler docs
r3624
Jason Grout
Small parallel doc typo
r3848 Pick two engines at random, and use the LRU of the two. This is known to be better
MinRK
dependency tweaks + dependency/scheduler docs
r3624 than plain random in many cases, but requires a small amount of computation.
leastload: Least Load
**This is the default scheme**
Always assign tasks to the engine with the fewest outstanding tasks (LRU breaks tie).
weighted: Weighted Two-Bin Random
MinRK
reflect revised apply_bound pattern
r3655 **Requires numpy**
MinRK
dependency tweaks + dependency/scheduler docs
r3624
Pick two engines at random using the number of outstanding tasks as inverse weights,
and use the one with the lower load.
MinRK
document TaskScheduler.hwm behavior
r5890 Greedy Assignment
-----------------
Tasks are assigned greedily as they are submitted. If their dependencies are
met, they will be assigned to an engine right away, and multiple tasks can be
assigned to an engine at a given time. This limit is set with the
``TaskScheduler.hwm`` (high water mark) configurable:
.. sourcecode:: python
# the most common choices are:
MinRK
document new default hwm value
r5919 c.TaskSheduler.hwm = 0 # (minimal latency, default in IPython ≤ 0.12)
MinRK
document TaskScheduler.hwm behavior
r5890 # or
MinRK
document new default hwm value
r5919 c.TaskScheduler.hwm = 1 # (most-informed balancing, default in > 0.12)
MinRK
document TaskScheduler.hwm behavior
r5890
MinRK
document new default hwm value
r5919 In IPython ≤ 0.12,the default is 0, or no-limit. That is, there is no limit to the number of
MinRK
document TaskScheduler.hwm behavior
r5890 tasks that can be outstanding on a given engine. This greatly benefits the
latency of execution, because network traffic can be hidden behind computation.
However, this means that workload is assigned without knowledge of how long
each task might take, and can result in poor load-balancing, particularly for
submitting a collection of heterogeneous tasks all at once. You can limit this
effect by setting hwm to a positive integer, 1 being maximum load-balancing (a
task will never be waiting if there is an idle engine), and any larger number
being a compromise between load-balance and latency-hiding.
MinRK
document new default hwm value
r5919 In practice, some users have been confused by having this optimization on by
default, and the default value has been changed to 1. This can be slower,
but has more obvious behavior and won't result in assigning too many tasks to
some engines in heterogeneous cases.
MinRK
dependency tweaks + dependency/scheduler docs
r3624
Pure ZMQ Scheduler
------------------
For maximum throughput, the 'pure' scheme is not Python at all, but a C-level
MinRK
use ROUTER/DEALER socket names instead of XREP/XREQ...
r4725 :class:`MonitoredQueue` from PyZMQ, which uses a ZeroMQ ``DEALER`` socket to perform all
MinRK
dependency tweaks + dependency/scheduler docs
r3624 load-balancing. This scheduler does not support any of the advanced features of the Python
:class:`.Scheduler`.
Disabled features when using the ZMQ Scheduler:
* Engine unregistration
Task farming will be disabled if an engine unregisters.
Further, if an engine is unregistered during computation, the scheduler may not recover.
* Dependencies
Since there is no Python logic inside the Scheduler, routing decisions cannot be made
based on message content.
* Early destination notification
The Python schedulers know which engine gets which task, and notify the Hub. This
allows graceful handling of Engines coming and going. There is no way to know
where ZeroMQ messages have gone, so there is no way to know what tasks are on which
engine until they *finish*. This makes recovery from engine shutdown very difficult.
MinRK
reflect revised apply_bound pattern
r3655
MinRK
dependency tweaks + dependency/scheduler docs
r3624
.. note::
TODO: performance comparisons
MinRK
add db,resubmit/retries docs
r3876
MinRK
clone parallel docs to parallelz
r3586 More details
============
MinRK
update API after sagedays29...
r3664 The :class:`LoadBalancedView` has many more powerful features that allow quite a bit
MinRK
clone parallel docs to parallelz
r3586 of flexibility in how tasks are defined and run. The next places to look are
in the following classes:
MinRK
organize IPython.parallel into subpackages
r3673 * :class:`~IPython.parallel.client.view.LoadBalancedView`
* :class:`~IPython.parallel.client.asyncresult.AsyncResult`
* :meth:`~IPython.parallel.client.view.LoadBalancedView.apply`
* :mod:`~IPython.parallel.controller.dependency`
MinRK
clone parallel docs to parallelz
r3586
The following is an overview of how to use these classes together:
MinRK
update API after sagedays29...
r3664 1. Create a :class:`Client` and :class:`LoadBalancedView`
MinRK
initial draft of core zmq.parallel docs
r3591 2. Define some functions to be run as tasks
3. Submit your tasks to using the :meth:`apply` method of your
MinRK
update API after sagedays29...
r3664 :class:`LoadBalancedView` instance.
MinRK
update parallel docs with some changes from scipy tutorial...
r5169 4. Use :meth:`.Client.get_result` to get the results of the
MinRK
initial draft of core zmq.parallel docs
r3591 tasks, or use the :meth:`AsyncResult.get` method of the results to wait
for and then receive the results.
MinRK
clone parallel docs to parallelz
r3586
MinRK
dependency tweaks + dependency/scheduler docs
r3624 .. seealso::
MinRK
clone parallel docs to parallelz
r3586
MinRK
dependency tweaks + dependency/scheduler docs
r3624 A demo of :ref:`DAG Dependencies <dag_dependencies>` with NetworkX and IPython.