Futures

Futures¶

Dask supports a real-time task framework that extends Python’s concurrent.futures interface. Dask futures allow you to scale generic Python workflows across a Dask cluster with minimal code changes.

This interface is good for arbitrary task scheduling like dask.delayed, but is immediate rather than lazy, which provides some more flexibility in situations where the computations may evolve over time. These features depend on the second generation task scheduler found in dask.distributed (which, despite its name, runs very well on a single machine).

Examples¶

Visit https://examples.dask.org/futures.html to see and run examples using futures with Dask.

Start Dask Client¶

You must start a Client to use the futures interface. This tracks state among the various worker processes or threads:

from dask.distributed import Client

client = Client()  # start local workers as processes
# or
client = Client(processes=False)  # start local workers as threads

If you have Bokeh installed, then this starts up a diagnostic dashboard at http://localhost:8787 .

Submit Tasks¶

`Client.submit`(func, *args[, key, workers, ...])	Submit a function application to the scheduler
`Client.map`(func, *iterables[, key, workers, ...])	Map a function on a sequence of arguments
`Future.result`([timeout])	Wait until computation completes, gather result to local process.

You can submit individual tasks using the submit method:

def inc(x):
    return x + 1

def add(x, y):
    return x + y

a = client.submit(inc, 10)  # calls inc(10) in background thread or process
b = client.submit(inc, 20)  # calls inc(20) in background thread or process

The submit function returns a Future, which refers to a remote result. This result may not yet be completed:

>>> a
<Future: status: pending, key: inc-b8aaf26b99466a7a1980efa1ade6701d>

Eventually it will complete. The result stays in the remote thread/process/worker until you ask for it back explicitly:

>>> a
<Future: status: finished, type: int, key: inc-b8aaf26b99466a7a1980efa1ade6701d>

>>> a.result()  # blocks until task completes and data arrives
11

You can pass futures as inputs to submit. Dask automatically handles dependency tracking; once all input futures have completed, they will be moved onto a single worker (if necessary), and then the computation that depends on them will be started. You do not need to wait for inputs to finish before submitting a new task; Dask will handle this automatically:

c = client.submit(add, a, b)  # calls add on the results of a and b

Similar to Python’s map, you can use Client.map to call the same function and many inputs:

futures = client.map(inc, range(1000))

However, note that each task comes with about 1ms of overhead. If you want to map a function over a large number of inputs, then you might consider dask.bag or dask.dataframe instead.

Move Data¶

`Future.result`([timeout])	Wait until computation completes, gather result to local process.
`Client.gather`(futures[, errors, direct, ...])	Gather futures from distributed memory
`Client.scatter`(data[, workers, broadcast, ...])	Scatter data into distributed memory

Given any future, you can call the .result method to gather the result. This will block until the future is done computing and then transfer the result back to your local process if necessary:

>>> c.result()
32

You can gather many results concurrently using the Client.gather method. This can be more efficient than calling .result() on each future sequentially:

>>> # results = [future.result() for future in futures]
>>> results = client.gather(futures)  # this can be faster

If you have important local data that you want to include in your computation, you can either include it as a normal input to a submit or map call:

>>> df = pd.read_csv('training-data.csv')
>>> future = client.submit(my_function, df)

Or you can scatter it explicitly. Scattering moves your data to a worker and returns a future pointing to that data:

>>> remote_df = client.scatter(df)
>>> remote_df
<Future: status: finished, type: DataFrame, key: bbd0ca93589c56ea14af49cba470006e>

>>> future = client.submit(my_function, remote_df)

Both of these accomplish the same result, but using scatter can sometimes be faster. This is especially true if you use processes or distributed workers (where data transfer is necessary) and you want to use df in many computations. Scattering the data beforehand avoids excessive data movement.

Calling scatter on a list scatters all elements individually. Dask will spread these elements evenly throughout workers in a round-robin fashion:

>>> client.scatter([1, 2, 3])
[<Future: status: finished, type: int, key: c0a8a20f903a4915b94db8de3ea63195>,
 <Future: status: finished, type: int, key: 58e78e1b34eb49a68c65b54815d1b158>,
 <Future: status: finished, type: int, key: d3395e15f605bc35ab1bac6341a285e2>]

References, Cancellation, and Exceptions¶

`Future.cancel`([reason, msg])	Cancel the request to run this future
`Future.exception`([timeout])	Return the exception of a failed task
`Future.traceback`([timeout])	Return the traceback of a failed task
`Client.cancel`(futures[, asynchronous, ...])	Cancel running futures This stops future tasks from being scheduled if they have not yet run and deletes them if they have already run.

Dask will only compute and hold onto results for which there are active futures. In this way, your local variables define what is active in Dask. When a future is garbage collected by your local Python session, Dask will feel free to delete that data or stop ongoing computations that were trying to produce it:

>>> del future  # deletes remote data once future is garbage collected

You can also explicitly cancel a task using the Future.cancel or Client.cancel methods:

>>> future.cancel()  # deletes data even if other futures point to it

If a future fails, then Dask will raise the remote exceptions and tracebacks if you try to get the result:

def div(x, y):
    return x / y

>>> a = client.submit(div, 1, 0)  # 1 / 0 raises a ZeroDivisionError
>>> a
<Future: status: error, key: div-3601743182196fb56339e584a2bf1039>

>>> a.result()
      1 def div(x, y):
----> 2     return x / y

ZeroDivisionError: division by zero

All futures that depend on an erred future also err with the same exception:

>>> b = client.submit(inc, a)
>>> b
<Future: status: error, key: inc-15e2e4450a0227fa38ede4d6b1a952db>

You can collect the exception or traceback explicitly with the Future.exception or Future.traceback methods.

Waiting on Futures¶

`as_completed`([futures, loop, with_results, ...])	Return futures in the order in which they complete
`wait`(fs[, timeout, return_when])	Wait until all/any futures are finished

You can wait on a future or collection of futures using the wait function:

from dask.distributed import wait

>>> wait(futures)

This blocks until all futures are finished or have erred.

You can also iterate over the futures as they complete using the as_completed function:

from dask.distributed import as_completed

futures = client.map(score, x_values)

best = -1
for future in as_completed(futures):
   y = future.result()
   if y > best:
       best = y

For greater efficiency, you can also ask as_completed to gather the results in the background:

for future, result in as_completed(futures, with_results=True):
    # y = future.result()  # don't need this
   ...

Or collect all futures in batches that had arrived since the last iteration:

for batch in as_completed(futures, with_results=True).batches():
   for future, result in batch:
       ...

Additionally, for iterative algorithms, you can add more futures into the as_completed iterator during iteration:

seq = as_completed(futures)

for future in seq:
    y = future.result()
    if condition(y):
        new_future = client.submit(...)
        seq.add(new_future)  # add back into the loop

or use seq.update(futures) to add multiple futures at once.

Fire and Forget¶

fire_and_forget(obj)

Run tasks at least once, even if we release the futures

Sometimes we don’t care about gathering the result of a task, and only care about side effects that it might have like writing a result to a file:

>>> a = client.submit(load, filename)
>>> b = client.submit(process, a)
>>> c = client.submit(write, b, out_filename)

As noted above, Dask will stop work that doesn’t have any active futures. It thinks that because no one has a pointer to this data that no one cares. You can tell Dask to compute a task anyway, even if there are no active futures, using the fire_and_forget function:

from dask.distributed import fire_and_forget

>>> fire_and_forget(c)

This is particularly useful when a future may go out of scope, for example, as part of a function:

def process(filename):
    out_filename = 'out-' + filename
    a = client.submit(load, filename)
    b = client.submit(process, a)
    c = client.submit(write, b, out_filename)
    fire_and_forget(c)
    return  # here we lose the reference to c, but that's now ok

for filename in filenames:
    process(filename)

Submit task and retrieve results from a different process¶

Sometimes we care about retrieving a result but not necessarily from the same process.

from distributed import Variable

var = Variable("my-result")
fut = client.submit(...)
var.set(fut)

Using a Variable instructs dask to remember the result of this task under the given name so that it can be retrieved later without having to keep the Client alive in the meantime.

var = Variable("my-result")
fut = var.get()
result = fut.result()

Submit Tasks from Tasks¶

`get_client`([address, timeout, resolve_address])	Get a client while within a task.
`rejoin`()	Have this thread rejoin the ThreadPoolExecutor
`secede`()	Have this task secede from the worker's thread pool

This is an advanced feature and is rarely necessary in the common case.

Tasks can launch other tasks by getting their own client. This enables complex and highly dynamic workloads:

from dask.distributed import get_client

def my_function(x):
    ...

    # Get locally created client
    client = get_client()

    # Do normal client operations, asking cluster for computation
    a = client.submit(...)
    b = client.submit(...)
    a, b = client.gather([a, b])

    return a + b

It also allows you to set up long running tasks that watch other resources like sockets or physical sensors:

def monitor(device):
   client = get_client()
   while True:
       data = device.read_data()
       future = client.submit(process, data)
       fire_and_forget(future)

for device in devices:
    fire_and_forget(client.submit(monitor))

However, each running task takes up a single thread, and so if you launch many tasks that launch other tasks, then it is possible to deadlock the system if you are not careful. You can call the secede function from within a task to have it remove itself from the dedicated thread pool into an administrative thread that does not take up a slot within the Dask worker:

from dask.distributed import get_client, secede

def monitor(device):
   client = get_client()
   secede()  # remove this task from the thread pool
   while True:
       data = device.read_data()
       future = client.submit(process, data)
       fire_and_forget(future)

If you intend to do more work in the same thread after waiting on client work, you may want to explicitly block until the thread is able to rejoin the thread pool. This allows some control over the number of threads that are created and stops too many threads from being active at once, over-saturating your hardware:

def f(n):  # assume that this runs as a task
   client = get_client()

   secede()  # secede while we wait for results to come back
   futures = client.map(func, range(n))
   results = client.gather(futures)

   rejoin()  # block until a slot is open in the thread pool
   result = analyze(results)
   return result

Alternatively, you can just use the normal compute function within a task. This will automatically call secede and rejoin appropriately:

def f(name, fn):
    df = dd.read_csv(fn)  # note that this is a dask collection
    result = df[df.name == name].count()

    # This calls secede
    # Then runs the computation on the cluster (including this worker)
    # Then blocks on rejoin, and finally delivers the answer
    result = result.compute()

    return result

Coordination Primitives¶

`Queue`([name, client, maxsize])	Distributed Queue
`Variable`([name, client])	Distributed Global Variable
`Lock`([name, client, scheduler_rpc, loop])	Distributed Centralized Lock
`Event`([name, client])	Distributed Centralized Event equivalent to asyncio.Event
`Semaphore`([max_leases, name, scheduler_rpc, ...])	This semaphore will track leases on the scheduler which can be acquired and released by an instance of this class.
`Pub`	Publish data with Publish-Subscribe pattern
`Sub`	Subscribe to a Publish/Subscribe topic

Sometimes situations arise where tasks, workers, or clients need to coordinate with each other in ways beyond normal task scheduling with futures. In these cases Dask provides additional primitives to help in complex situations.

Dask provides distributed versions of coordination primitives like locks, events, queues, global variables, and pub-sub systems that, where appropriate, match their in-memory counterparts. These can be used to control access to external resources, track progress of ongoing computations, or share data in side-channels between many workers, clients, and tasks sensibly.

These features are rarely necessary for common use of Dask. We recommend that beginning users stick with using the simpler futures found above (like Client.submit and Client.gather) rather than embracing needlessly complex techniques.

Queues¶

Queue([name, client, maxsize])

Distributed Queue

Dask queues follow the API for the standard Python Queue, but now move futures or small messages between clients. Queues serialize sensibly and reconnect themselves on remote clients if necessary:

from dask.distributed import Queue

def load_and_submit(filename):
    data = load(filename)
    client = get_client()
    future = client.submit(process, data)
    queue.put(future)

client = Client()

queue = Queue()

for filename in filenames:
    future = client.submit(load_and_submit, filename)
    fire_and_forget(future)

while True:
    future = queue.get()
    print(future.result())

Queues can also send small pieces of information, anything that is msgpack encodable (ints, strings, bools, lists, dicts, etc.). This can be useful to send back small scores or administrative messages:

def func(x):
    try:
       ...
    except Exception as e:
        error_queue.put(str(e))

error_queue = Queue()

Queues are mediated by the central scheduler, and so they are not ideal for sending large amounts of data (everything you send will be routed through a central point). They are well suited to move around small bits of metadata, or futures. These futures may point to much larger pieces of data safely:

>>> x = ... # my large numpy array

# Don't do this!
>>> q.put(x)

# Do this instead
>>> future = client.scatter(x)
>>> q.put(future)

# Or use futures for metadata
>>> q.put({'status': 'OK', 'stage=': 1234})

If you’re looking to move large amounts of data between workers, then you might also want to consider the Pub/Sub system described a few sections below.

Global Variables¶

Variable([name, client])

Distributed Global Variable

Variables are like Queues in that they communicate futures and small data between clients. However, variables hold only a single value. You can get or set that value at any time:

>>> var = Variable('stopping-criterion')
>>> var.set(False)

>>> var.get()
False

This is often used to signal stopping criteria or current parameters between clients.

If you want to share large pieces of information, then scatter the data first:

>>> parameters = np.array(...)
>>> future = client.scatter(parameters)
>>> var.set(future)

Locks¶

Lock([name, client, scheduler_rpc, loop])

Distributed Centralized Lock

You can also hold onto cluster-wide locks using the Lock object. Dask Locks have the same API as normal threading.Lock objects, except that they work across the cluster:

from dask.distributed import Lock
lock = Lock()

with lock:
    # access protected resource

You can manage several locks at the same time. Lock can either be given a consistent name or you can pass the lock object around itself.

Using a consistent name is convenient when you want to lock some known named resource:

from dask.distributed import Lock

def load(fn):
    with Lock('the-production-database'):
        # read data from filename using some sensitive source
        return ...

futures = client.map(load, filenames)

Passing around a lock works as well and is easier when you want to create short-term locks for a particular situation:

from dask.distributed import Lock
lock = Lock()

def load(fn, lock=None):
    with lock:
        # read data from filename using some sensitive source
        return ...

futures = client.map(load, filenames, lock=lock)

This can be useful if you want to control concurrent access to some external resource like a database or un-thread-safe library.

Events¶

Event([name, client])

Distributed Centralized Event equivalent to asyncio.Event

Dask Events mimic asyncio.Event objects, but on a cluster scope. They hold a single flag which can be set or cleared. Clients can wait until the event flag is set. Different from a Lock, every client can set or clear the flag and there is no “ownership” of an event.

You can use events to e.g. synchronize multiple clients:

# One one client
from dask.distributed import Event

event = Event("my-event-1")
event.wait()

The call to wait will block until the event is set, e.g. in another client

# In another client
from dask.distributed import Event

event = Event("my-event-1")

# do some work

event.set()

Events can be set, cleared and waited on multiple times. Every waiter referencing the same event name will be notified on event set (and not only the first one as in the case of a lock):

from dask.distributed import Event

def wait_for_event(x):
   event = Event("my-event")

   event.wait()
   # at this point, all function calls
   # are in sync once the event is set

futures = client.map(wait_for_event, range(10))

Event("my-event").set()
client.gather(futures)

Semaphore¶

Semaphore([max_leases, name, scheduler_rpc, ...])

This semaphore will track leases on the scheduler which can be acquired and released by an instance of this class.

Similar to the single-valued Lock it is also possible to use a cluster-wide semaphore to coordinate and limit access to a sensitive resource like a database.

from dask.distributed import Semaphore

sem = Semaphore(max_leases=2, name="database")

def access_limited(val, sem):
   with sem:
      # Interact with the DB
      return

futures = client.map(access_limited, range(10), sem=sem)
client.gather(futures)
sem.close()

Actors¶

Actors allow workers to manage rapidly changing state without coordinating with the central scheduler. This has the advantage of reducing latency (worker-to-worker roundtrip latency is around 1ms), reducing pressure on the centralized scheduler (workers can coordinate actors entirely among each other), and also enabling workflows that require stateful or in-place memory manipulation.

However, these benefits come at a cost. The scheduler is unaware of actors and so they don’t benefit from diagnostics, load balancing, or resilience. Once an actor is running on a worker it is forever tied to that worker. If that worker becomes overburdened or dies, then there is no opportunity to recover the workload.

Because Actors avoid the central scheduler they can be high-performing, but not resilient.

Example: Counter¶

An actor is a class containing both state and methods that is submitted to a worker:

class Counter:
    n = 0

    def __init__(self):
        self.n = 0

    def increment(self):
        self.n += 1
        return self.n

from dask.distributed import Client
client = Client()

future = client.submit(Counter, actor=True)
counter = future.result()

>>> counter
<Actor: Counter, key=Counter-afa1cdfb6b4761e616fa2cfab42398c8>

Method calls on this object produce ActorFutures, which are similar to normal Futures, but interact only with the worker holding the Actor:

>>> future = counter.increment()
>>> future
<ActorFuture>

>>> future.result()
1

Attribute access is synchronous and blocking:

>>> counter.n
1

Example: Parameter Server¶

This example will perform the following minimization with a parameter server:

\[\min_{p\in\mathbb{R}^{1000}} \sum_{i=1}^{1000} (p_i - 1)^2\]

This is a simple minimization that will serve as an illustrative example.

The Dask Actor will serve as the parameter server that will hold the model. The client will calculate the gradient of the loss function above.

import numpy as np

from dask.distributed import Client
client = Client(processes=False)

class ParameterServer:
    def __init__(self):
        self.data = dict()

    def put(self, key, value):
        self.data[key] = value

    def get(self, key):
        return self.data[key]

def train(params, lr=0.1):
    grad = 2 * (params - 1)  # gradient of (params - 1)**2
    new_params = params - lr * grad
    return new_params

ps_future = client.submit(ParameterServer, actor=True)
ps = ps_future.result()

ps.put('parameters', np.random.default_rng().random(1000))
for k in range(20):
    params = ps.get('parameters').result()
    new_params = train(params)
    ps.put('parameters', new_params)
    print(new_params.mean())
    # k=0: "0.5988202981316124"
    # k=10: "0.9569236575164062"

This example works, and the loss function is minimized. The (simple) equation above is minimize, so each \(p_i\) converges to 1. If desired, this example could be adapted to machine learning with a more complex function to minimize.

Asynchronous Operation¶

All operations that require talking to the remote worker are awaitable:

async def f():
    future = client.submit(Counter, actor=True)
    counter = await future  # gather actor object locally

    counter.increment()  # send off a request asynchronously
    await counter.increment()  # or wait until it was received

    n = await counter.n  # attribute access also must be awaited

Generally, all I/O operations that trigger computations (e.g. to_parquet) should be done using the compute=False parameter to avoid asynchronous blocking:

await client.compute(ddf.to_parquet('/tmp/some.parquet', compute=False))

API¶

Client

`Client`([address, loop, timeout, ...])	Connect to and submit computation to a Dask cluster
`Client.cancel`(futures[, asynchronous, ...])	Cancel running futures This stops future tasks from being scheduled if they have not yet run and deletes them if they have already run.
`Client.compute`(collections[, sync, ...])	Compute dask collections on cluster
`Client.gather`(futures[, errors, direct, ...])	Gather futures from distributed memory
`Client.get`(dsk, keys[, workers, ...])	Compute dask graph
`Client.get_dataset`(name[, default])	Get named dataset from the scheduler if present.
`Client.get_executor`(**kwargs)	Return a concurrent.futures Executor for submitting tasks on this Client
`Client.has_what`([workers])	Which keys are held by which workers
`Client.list_datasets`(**kwargs)	List named datasets available on the scheduler
`Client.map`(func, *iterables[, key, workers, ...])	Map a function on a sequence of arguments
`Client.ncores`([workers])	The number of threads/cores available on each worker node
`Client.persist`(collections[, ...])	Persist dask collections on cluster
`Client.profile`([key, start, stop, workers, ...])	Collect statistical profiling information about recent work
`Client.publish_dataset`(args, *kwargs)	Publish named datasets to scheduler
`Client.rebalance`([futures, workers])	Rebalance data within network
`Client.replicate`(futures[, n, workers, ...])	Set replication of futures within network
`Client.restart`([timeout, wait_for_workers])	Restart all workers.
`Client.run`(function, *args[, workers, wait, ...])	Run a function on all workers outside of task scheduling system
`Client.run_on_scheduler`(function, *args, ...)	Run a function on the scheduler process
`Client.scatter`(data[, workers, broadcast, ...])	Scatter data into distributed memory
`Client.shutdown`()	Shut down the connected scheduler and workers
`Client.scheduler_info`(**kwargs)	Basic information about the workers in the cluster
`Client.submit`(func, *args[, key, workers, ...])	Submit a function application to the scheduler
`Client.unpublish_dataset`(name, **kwargs)	Remove named datasets from scheduler
`Client.upload_file`(filename[, load])	Upload local package to scheduler and workers
`Client.who_has`([futures])	The workers storing each future's data

Future

`Future`(key[, client, state, _id])	A remotely running computation
`Future.add_done_callback`(fn)	Call callback on future when future has finished
`Future.cancel`([reason, msg])	Cancel the request to run this future
`Future.cancelled`()	Returns True if the future has been cancelled
`Future.done`()	Returns whether or not the computation completed.
`Future.exception`([timeout])	Return the exception of a failed task
`Future.result`([timeout])	Wait until computation completes, gather result to local process.
`Future.traceback`([timeout])	Return the traceback of a failed task

Functions

`as_completed`([futures, loop, with_results, ...])	Return futures in the order in which they complete
`fire_and_forget`(obj)	Run tasks at least once, even if we release the futures
`get_client`([address, timeout, resolve_address])	Get a client while within a task.
`secede`()	Have this task secede from the worker's thread pool
`rejoin`()	Have this thread rejoin the ThreadPoolExecutor
`wait`(fs[, timeout, return_when])	Wait until all/any futures are finished
`print`(*args[, sep, end, file, flush])	A drop-in replacement of the built-in `print` function for remote printing from workers to clients.
`warn`(message[, category, stacklevel, source])	A drop-in replacement of the built-in `warnings.warn()` function for issuing warnings remotely from workers to clients.

distributed.as_completed(futures=None, loop=None, with_results=False, raise_errors=True, *, timeout=None)[source]¶

Return futures in the order in which they complete

This returns an iterator that yields the input future objects in the order in which they complete. Calling next on the iterator will block until the next future completes, irrespective of order.

Additionally, you can also add more futures to this object during computation with the .add method

Parameters

futures: Collection of futures: A list of Future objects to be iterated over in the order in which they complete
with_results: bool (False): Whether to wait and include results of futures as well; in this case as_completed yields a tuple of (future, result)
raise_errors: bool (True): Whether we should raise when the result of a future raises an exception; only affects behavior when with_results=True.
timeout: int (optional): The returned iterator raises a dask.distributed.TimeoutError if __next__() or __anext__() is called and the result isn’t available after timeout seconds from the original call to as_completed(). If timeout is not specified or None, there is no limit to the wait time.

Examples

>>> x, y, z = client.map(inc, [1, 2, 3])  
>>> for future in as_completed([x, y, z]):  
...     print(future.result())  
3
2
4

Add more futures during computation

>>> x, y, z = client.map(inc, [1, 2, 3])  
>>> ac = as_completed([x, y, z])  
>>> for future in ac:  
...     print(future.result())  
...     if random.random() < 0.5:  
...         ac.add(c.submit(double, future))  
4
2
8
3
6
12
24

Optionally wait until the result has been gathered as well

>>> ac = as_completed([x, y, z], with_results=True)  
>>> for future, result in ac:  
...     print(result)  
2
4
3

distributed.fire_and_forget(obj)[source]¶

Run tasks at least once, even if we release the futures

Under normal operation Dask will not run any tasks for which there is not an active future (this avoids unnecessary work in many situations). However sometimes you want to just fire off a task, not track its future, and expect it to finish eventually. You can use this function on a future or collection of futures to ask Dask to complete the task even if no active client is tracking it.

The results will not be kept in memory after the task completes (unless there is an active future) so this is only useful for tasks that depend on side effects.

Parameters

objFuture, list, dict, dask collection: The futures that you want to run at least once

Examples

>>> fire_and_forget(client.submit(func, *args))  

distributed.get_client(address=None, timeout=None, resolve_address=True) → Client[source]¶

Get a client while within a task.

This client connects to the same scheduler to which the worker is connected

Parameters

addressstr, optional: The address of the scheduler to connect to. Defaults to the scheduler the worker is connected to.
timeoutint or str: Timeout (in seconds) for getting the Client. Defaults to the distributed.comm.timeouts.connect configuration value.
resolve_addressbool, default True: Whether to resolve address to its canonical form.

Returns

Client

See also

get_worker
worker_client
secede

Examples

>>> def f():
...     client = get_client(timeout="10s")
...     futures = client.map(lambda x: x + 1, range(10))  # spawn many tasks
...     results = client.gather(futures)
...     return sum(results)

>>> future = client.submit(f)  
>>> future.result()  
55

distributed.secede()[source]¶

Have this task secede from the worker’s thread pool

This opens up a new scheduling slot and a new thread for a new task. This enables the client to schedule tasks on this node, which is especially useful while waiting for other jobs to finish (e.g., with client.gather).

See also

get_client
get_worker

Examples

>>> def mytask(x):
...     # do some work
...     client = get_client()
...     futures = client.map(...)  # do some remote work
...     secede()  # while that work happens, remove ourself from the pool
...     return client.gather(futures)  # return gathered results

distributed.rejoin()[source]¶

Have this thread rejoin the ThreadPoolExecutor

This will block until a new slot opens up in the executor. The next thread to finish a task will leave the pool to allow this one to join.

See also

secede: leave the thread pool

distributed.wait(fs, timeout=None, return_when='ALL_COMPLETED')[source]¶

Wait until all/any futures are finished

Parameters

fsList[Future]
timeoutnumber, string, optional: Time after which to raise a dask.distributed.TimeoutError. Can be a string like "10 minutes" or a number of seconds to wait.
return_whenstr, optional: One of ALL_COMPLETED or FIRST_COMPLETED

Returns

Named tuple of completed, not completed

distributed.print(*args, sep: str | None = ' ', end: str | None = '\n', file: TextIO | None = None, flush: bool = False) → None[source]¶

A drop-in replacement of the built-in print function for remote printing from workers to clients. If called from outside a dask worker, its arguments are passed directly to builtins.print(). If called by code running on a worker, then in addition to printing locally, any clients connected (possibly remotely) to the scheduler managing this worker will receive an event instructing them to print the same output to their own standard output or standard error streams. For example, the user can perform simple debugging of remote computations by including calls to this print function in the submitted code and inspecting the output in a local Jupyter notebook or interpreter session.

All arguments behave the same as those of builtins.print(), with the exception that the file keyword argument, if specified, must either be sys.stdout or sys.stderr; arbitrary file-like objects are not allowed.

All non-keyword arguments are converted to strings using str() and written to the stream, separated by sep and followed by end. Both sep and end must be strings; they can also be None, which means to use the default values. If no objects are given, print() will just write end.

Parameters

sepstr, optional: String inserted between values, default a space.
endstr, optional: String appended after the last value, default a newline.
filesys.stdout or sys.stderr, optional: Defaults to the current sys.stdout.
flushbool, default False: Whether to forcibly flush the stream.

Examples

>>> from dask.distributed import Client, print
>>> client = distributed.Client(...)
>>> def worker_function():
...     print("Hello from worker!")
>>> client.submit(worker_function)
<Future: finished, type: NoneType, key: worker_function-...>
Hello from worker!

distributed.warn(message: str | Warning, category: type[Warning] | None = <class 'UserWarning'>, stacklevel: int = 1, source: typing.Any = None) → None[source]¶

A drop-in replacement of the built-in warnings.warn() function for issuing warnings remotely from workers to clients.

If called from outside a dask worker, its arguments are passed directly to warnings.warn(). If called by code running on a worker, then in addition to emitting a warning locally, any clients connected (possibly remotely) to the scheduler managing this worker will receive an event instructing them to emit the same warning (subject to their own local filters, etc.). When implementing computations that may run on a worker, the user can call this warn function to ensure that any remote client sessions will see their warnings, for example in a Jupyter output cell.

While all of the arguments are respected by the locally emitted warning (with same meanings as in warnings.warn()), stacklevel and source are ignored by clients because they would not be meaningful in the client’s thread.

Examples

>>> from dask.distributed import Client, warn
>>> client = Client()
>>> def do_warn():
...    warn("A warning from a worker.")
>>> client.submit(do_warn).result()
/path/to/distributed/client.py:678: UserWarning: A warning from a worker.

class distributed.Client(address=None, loop=None, timeout=_NoDefault.no_default, set_as_default=True, scheduler_file=None, security=None, asynchronous=False, name=None, heartbeat_interval=None, serializers=None, deserializers=None, extensions={'pubsub': <class 'distributed.pubsub.PubSubClientExtension'>}, direct_to_workers=None, connection_limit=512, **kwargs)[source]¶

Connect to and submit computation to a Dask cluster

The Client connects users to a Dask cluster. It provides an asynchronous user interface around functions and futures. This class resembles executors in concurrent.futures but also allows Future objects within submit/map calls. When a Client is instantiated it takes over all dask.compute and dask.persist calls by default.

It is also common to create a Client without specifying the scheduler address , like Client(). In this case the Client creates a LocalCluster in the background and connects to that. Any extra keywords are passed from Client to LocalCluster in this case. See the LocalCluster documentation for more information.

Parameters

address: string, or Cluster: This can be the address of a Scheduler server like a string '127.0.0.1:8786' or a cluster object like LocalCluster()
loop: The event loop
timeout: int (defaults to configuration ``distributed.comm.timeouts.connect``): Timeout duration for initial connection to the scheduler
set_as_default: bool (True): Use this Client as the global dask scheduler
scheduler_file: string (optional): Path to a file with scheduler information if available
security: Security or bool, optional: Optional security information. If creating a local cluster can also pass in True, in which case temporary self-signed credentials will be created automatically.
asynchronous: bool (False by default): Set to True if using this client within async/await functions or within Tornado gen.coroutines. Otherwise this should remain False for normal use.
name: string (optional): Gives the client a name that will be included in logs generated on the scheduler for matters relating to this client
heartbeat_interval: int (optional): Time in milliseconds between heartbeats to scheduler
serializers: Iterable of approaches to use when serializing the object. See Serialization for more.
deserializers: Iterable of approaches to use when deserializing the object. See Serialization for more.
extensionslist: The extensions
direct_to_workers: bool (optional): Whether or not to connect directly to the workers, or to ask the scheduler to serve as intermediary.
connection_limitint: The number of open comms to maintain at once in the connection pool
**kwargs:: If you do not pass a scheduler address, Client will create a LocalCluster object, passing any extra keyword arguments.

See also

distributed.scheduler.Scheduler: Internal scheduler
distributed.LocalCluster

Examples

Provide cluster’s scheduler node address on initialization:

>>> client = Client('127.0.0.1:8786')  

Use submit method to send individual computations to the cluster

>>> a = client.submit(add, 1, 2)  
>>> b = client.submit(add, 10, 20)  

Continue using submit or map on results to build up larger computations

>>> c = client.submit(add, a, b)  

Gather results with the gather method.

>>> client.gather(c)  
33

You can also call Client with no arguments in order to create your own local cluster.

>>> client = Client()  # makes your own local "cluster" 

Extra keywords will be passed directly to LocalCluster

>>> client = Client(n_workers=2, threads_per_worker=4)  

property amm¶: Convenience accessors for the Active Memory Manager

as_current()[source]¶: Thread-local, Task-local context manager that causes the Client.current class method to return self. Any Future objects deserialized inside this context manager will be automatically attached to this Client.

benchmark_hardware() → dict[source]¶

Run a benchmark on the workers for memory, disk, and network bandwidths

Returns

result: dict: A dictionary mapping the names “disk”, “memory”, and “network” to dictionaries mapping sizes to bandwidths. These bandwidths are averaged over many workers running computations across the cluster.

call_stack(futures=None, keys=None)[source]¶

The actively running call stack of all relevant keys

You can specify data of interest either by providing futures or collections in the futures= keyword or a list of explicit keys in the keys= keyword. If neither are provided then all call stacks will be returned.

Parameters

futureslist (optional): List of futures, defaults to all data
keyslist (optional): List of key names, defaults to all data

Examples

>>> df = dd.read_parquet(...).persist()  
>>> client.call_stack(df)  # call on collections

>>> client.call_stack()  # Or call with no arguments for all activity  

cancel(futures, asynchronous=None, force=False, reason=None, msg=None)[source]¶

Cancel running futures This stops future tasks from being scheduled if they have not yet run and deletes them if they have already run. After calling, this result and all dependent results will no longer be accessible

Parameters

futuresList[Future]: The list of Futures
asynchronous: bool: If True the client is in asynchronous mode
forceboolean (False): Cancel this future even if other clients desire it
reason: str: Reason for cancelling the futures
msgstr: Message that will be attached to the cancelled future

close(timeout=_NoDefault.no_default)[source]¶

Close this client

Clients will also close automatically when your Python session ends

If you started a client without arguments like Client() then this will also close the local cluster that was started at the same time.

Parameters

timeoutnumber: Time in seconds after which to raise a dask.distributed.TimeoutError

See also

Client.restart

static collections_to_dsk(collections, *args, **kwargs)[source]¶: Convert many collections into a single dask graph, after optimization

compute(collections, sync=False, optimize_graph=True, workers=None, allow_other_workers=False, resources=None, retries=0, priority=0, fifo_timeout='60s', actors=None, traverse=True, **kwargs)[source]¶

Compute dask collections on cluster

Parameters

collectionsiterable of dask objects or single dask object: Collections like dask.array or dataframe or dask.value objects
syncbool (optional): Returns Futures if False (default) or concrete values if True
optimize_graphbool: Whether or not to optimize the underlying graphs
workersstring or iterable of strings: A set of worker hostnames on which computations may be performed. Leave empty to default to all workers (common case)
allow_other_workersbool (defaults to False): Used with workers. Indicates whether or not the computations may be performed on workers that are not in the workers set(s).
retriesint (default to 0): Number of allowed automatic retries if computing a result fails
priorityNumber: Optional prioritization of task. Zero is default. Higher priorities take precedence
fifo_timeouttimedelta str (defaults to ’60s’): Allowed amount of time between calls to consider the same priority
traversebool (defaults to True): By default dask traverses builtin python collections looking for dask objects passed to compute. For large collections this can be expensive. If none of the arguments contain any dask objects, set traverse=False to avoid doing this traversal.
resourcesdict (defaults to {}): Defines the resources each instance of this mapped task requires on the worker; e.g. {'GPU': 2}. See worker resources for details on defining resources.
actorsbool or dict (default None): Whether these tasks should exist on the worker as stateful actors. Specified on a global (True/False) or per-task ({'x': True, 'y': False}) basis. See Actors for additional details.
**kwargs: Options to pass to the graph optimize calls

Returns

List of Futures if input is a sequence, or a single future otherwise

See also

Client.get: Normal synchronous dask.get function

Examples

>>> from dask import delayed
>>> from operator import add
>>> x = delayed(add)(1, 2)
>>> y = delayed(add)(x, x)
>>> xx, yy = client.compute([x, y])  
>>> xx  
<Future: status: finished, key: add-8f6e709446674bad78ea8aeecfee188e>
>>> xx.result()  
3
>>> yy.result()  
6

Also support single arguments

>>> xx = client.compute(x)  

classmethod current(allow_global=True)[source]¶

When running within the context of as_client, return the context-local current client. Otherwise, return the latest initialised Client. If no Client instances exist, raise ValueError. If allow_global is set to False, raise ValueError if running outside of the as_client context manager.

Parameters

allow_globalbool: If True returns the default client

Returns

Client: The current client

Raises

ValueError: If there is no client set, a ValueError is raised

See also

default_client

property dashboard_link¶

Link to the scheduler’s dashboard.

Returns

str: Dashboard URL.

Examples

Opening the dashboard in your default web browser:

>>> import webbrowser
>>> from distributed import Client
>>> client = Client()
>>> webbrowser.open(client.dashboard_link)

dump_cluster_state(filename: str = 'dask-cluster-dump', write_from_scheduler: bool | None = None, exclude: collections.abc.Collection[str] = (), format: Literal['msgpack', 'yaml'] = 'msgpack', **storage_options)[source]¶

Extract a dump of the entire cluster state and persist to disk or a URL. This is intended for debugging purposes only.

Warning: Memory usage on the scheduler (and client, if writing the dump locally) can be large. On a large or long-running cluster, this can take several minutes. The scheduler may be unresponsive while the dump is processed.

Results will be stored in a dict:

{
    "scheduler": {...},  # scheduler state
    "workers": {
        worker_addr: {...},  # worker state
        ...
    }
    "versions": {
        "scheduler": {...},
        "workers": {
            worker_addr: {...},
            ...
        }
    }
}

Parameters

filename:

The path or URL to write to. The appropriate file suffix (.msgpack.gz or .yaml) will be appended automatically.

Must be a path supported by fsspec.open() (like s3://my-bucket/cluster-dump, or cluster-dumps/dump). See write_from_scheduler to control whether the dump is written directly to filename from the scheduler, or sent back to the client over the network, then written locally.

write_from_scheduler:

If None (default), infer based on whether filename looks like a URL or a local path: True if the filename contains :// (like s3://my-bucket/cluster-dump), False otherwise (like local_dir/cluster-dump).

If True, write cluster state directly to filename from the scheduler. If filename is a local path, the dump will be written to that path on the scheduler’s filesystem, so be careful if the scheduler is running on ephemeral hardware. Useful when the scheduler is attached to a network filesystem or persistent disk, or for writing to buckets.

If False, transfer cluster state from the scheduler back to the client over the network, then write it to filename. This is much less efficient for large dumps, but useful when the scheduler doesn’t have access to any persistent storage.

exclude:

A collection of attribute names which are supposed to be excluded from the dump, e.g. to exclude code, tracebacks, logs, etc.

Defaults to exclude run_spec, which is the serialized user code. This is typically not required for debugging. To allow serialization of this, pass an empty tuple.

format:

Either "msgpack" or "yaml". If msgpack is used (default), the output will be stored in a gzipped file as msgpack.

To read:

import gzip, msgpack
with gzip.open("filename") as fd:
    state = msgpack.unpack(fd)

or:

import yaml
try:
    from yaml import CLoader as Loader
except ImportError:
    from yaml import Loader
with open("filename") as fd:
    state = yaml.load(fd, Loader=Loader)

**storage_options:

Any additional arguments to fsspec.open() when writing to a URL.

forward_logging(logger_name=None, level=0)[source]¶

Begin forwarding the given logger (by default the root) and all loggers under it from worker tasks to the client process. Whenever the named logger handles a LogRecord on the worker-side, the record will be serialized, sent to the client, and handled by the logger with the same name on the client-side.

Note that worker-side loggers will only handle LogRecords if their level is set appropriately, and the client-side logger will only emit the forwarded LogRecord if its own level is likewise set appropriately. For example, if your submitted task logs a DEBUG message to logger “foo”, then in order for forward_logging() to cause that message to be emitted in your client session, you must ensure that the logger “foo” have its level set to DEBUG (or lower) in the worker process and in the client process.

Parameters

logger_namestr, optional

The name of the logger to begin forwarding. The usual rules of the logging module’s hierarchical naming system apply. For example, if name is "foo", then not only "foo", but also "foo.bar", "foo.baz", etc. will be forwarded. If name is None, this indicates the root logger, and so all loggers will be forwarded.

Note that a logger will only forward a given LogRecord if the logger’s level is sufficient for the LogRecord to be handled at all.

levelstr | int, optional

Optionally restrict forwarding to LogRecords of this level or higher, even if the forwarded logger’s own level is lower.

Examples

For purposes of the examples, suppose we configure client-side logging as a user might: with a single StreamHandler attached to the root logger with an output level of INFO and a simple output format:

import logging
import distributed
import io, yaml

TYPICAL_LOGGING_CONFIG = '''
version: 1
handlers:
  console:
    class : logging.StreamHandler
    formatter: default
    level   : INFO
formatters:
  default:
    format: '%(asctime)s %(levelname)-8s [worker %(worker)s] %(name)-15s %(message)s'
    datefmt: '%Y-%m-%d %H:%M:%S'
root:
  handlers:
    - console
'''
config = yaml.safe_load(io.StringIO(TYPICAL_LOGGING_CONFIG))
logging.config.dictConfig(config)

Now create a client and begin forwarding the root logger from workers back to our local client process.

>>> client = distributed.Client()
>>> client.forward_logging()  # forward the root logger at any handled level

Then submit a task that does some error logging on a worker. We see output from the client-side StreamHandler.

>>> def do_error():
...     logging.getLogger("user.module").error("Hello error")
...     return 42
>>> client.submit(do_error).result()
2022-11-09 03:43:25 ERROR    [worker tcp://127.0.0.1:34783] user.module     Hello error
42

Note how an attribute "worker" is also added by dask to the forwarded LogRecord, which our custom formatter uses. This is useful for identifying exactly which worker logged the error.

One nuance worth highlighting: even though our client-side root logger is configured with a level of INFO, the worker-side root loggers still have their default level of ERROR because we haven’t done any explicit logging configuration on the workers. Therefore worker-side INFO logs will not be forwarded because they never even get handled in the first place.

>>> def do_info_1():
...     # no output on the client side
...     logging.getLogger("user.module").info("Hello info the first time")
...     return 84
>>> client.submit(do_info_1).result()
84

It is necessary to set the client-side logger’s level to INFO before the info message will be handled and forwarded to the client. In other words, the “effective” level of the client-side forwarded logging is the maximum of each logger’s client-side and worker-side levels.

>>> def do_info_2():
...     logger = logging.getLogger("user.module")
...     logger.setLevel(logging.INFO)
...     # now produces output on the client side
...     logger.info("Hello info the second time")
...     return 84
>>> client.submit(do_info_2).result()
2022-11-09 03:57:39 INFO     [worker tcp://127.0.0.1:42815] user.module     Hello info the second time
84

futures_of(futures)[source]¶

Wrapper method of futures_of

Parameters

futurestuple: The futures

gather(futures, errors='raise', direct=None, asynchronous=None)[source]¶

Gather futures from distributed memory

Accepts a future, nested container of futures, iterator, or queue. The return type will match the input type.

Parameters

futuresCollection of futures: This can be a possibly nested collection of Future objects. Collections can be lists, sets, or dictionaries
errorsstring: Either ‘raise’ or ‘skip’ if we should raise if a future has erred or skip its inclusion in the output collection
directboolean: Whether or not to connect directly to the workers, or to ask the scheduler to serve as intermediary. This can also be set when creating the Client.
asynchronous: bool: If True the client is in asynchronous mode

Returns

results: a collection of the same type as the input, but now with
gathered results rather than futures

See also

Client.scatter: Send data out to cluster

Examples

>>> from operator import add  
>>> c = Client('127.0.0.1:8787')  
>>> x = c.submit(add, 1, 2)  
>>> c.gather(x)  
3
>>> c.gather([x, [x], x])  # support lists and dicts 
[3, [3], 3]

get(dsk, keys, workers=None, allow_other_workers=None, resources=None, sync=True, asynchronous=None, direct=None, retries=None, priority=0, fifo_timeout='60s', actors=None, **kwargs)[source]¶

Compute dask graph

Parameters

dskdict
keysobject, or nested lists of objects
workersstring or iterable of strings: A set of worker addresses or hostnames on which computations may be performed. Leave empty to default to all workers (common case)
allow_other_workersbool (defaults to False): Used with workers. Indicates whether or not the computations may be performed on workers that are not in the workers set(s).
resourcesdict (defaults to {}): Defines the resources each instance of this mapped task requires on the worker; e.g. {'GPU': 2}. See worker resources for details on defining resources.
syncbool (optional): Returns Futures if False or concrete values if True (default).
asynchronous: bool: If True the client is in asynchronous mode
directbool: Whether or not to connect directly to the workers, or to ask the scheduler to serve as intermediary. This can also be set when creating the Client.
retriesint (default to 0): Number of allowed automatic retries if computing a result fails
priorityNumber: Optional prioritization of task. Zero is default. Higher priorities take precedence
fifo_timeouttimedelta str (defaults to ’60s’): Allowed amount of time between calls to consider the same priority
actorsbool or dict (default None): Whether these tasks should exist on the worker as stateful actors. Specified on a global (True/False) or per-task ({'x': True, 'y': False}) basis. See Actors for additional details.

Returns

results: If ‘sync’ is True, returns the results. Otherwise, returns the known data packed If ‘sync’ is False, returns the known data. Otherwise, returns the results

See also

Client.compute: Compute asynchronous collections

Examples

>>> from operator import add  
>>> c = Client('127.0.0.1:8787')  
>>> c.get({'x': (add, 1, 2)}, 'x')  
3

get_dataset(name, default=_NoDefault.no_default, **kwargs)[source]¶

Get named dataset from the scheduler if present. Return the default or raise a KeyError if not present.

Parameters

namestr: name of the dataset to retrieve
defaultstr: optional, not set by default If set, do not raise a KeyError if the name is not present but return this default
kwargsdict: additional keyword arguments to _get_dataset

Returns

The dataset from the scheduler, if present

See also

Client.publish_dataset
Client.list_datasets

get_events(topic: str | None = None)[source]¶

Retrieve structured topic logs

Parameters

topicstr, optional: Name of topic log to retrieve events for. If no topic is provided, then logs for all topics will be returned.

get_executor(**kwargs)[source]¶

Return a concurrent.futures Executor for submitting tasks on this Client

Parameters

**kwargs: Any submit()- or map()- compatible arguments, such as workers or resources.

Returns

ClientExecutor: An Executor object that’s fully compatible with the concurrent.futures API.

get_metadata(keys, default=_NoDefault.no_default)[source]¶

Get arbitrary metadata from scheduler

See set_metadata for the full docstring with examples

Parameters

keyskey or list: Key to access. If a list then gets within a nested collection
defaultoptional: If the key does not exist then return this value instead. If not provided then this raises a KeyError if the key is not present

See also

Client.set_metadata

get_scheduler_logs(n=None)[source]¶

Get logs from scheduler

Parameters

nint: Number of logs to retrieve. Maxes out at 10000 by default, configurable via the distributed.admin.log-length configuration value.

Returns

Logs in reversed order (newest first)

get_task_stream(start=None, stop=None, count=None, plot=False, filename='task-stream.html', bokeh_resources=None)[source]¶

Get task stream data from scheduler

This collects the data present in the diagnostic “Task Stream” plot on the dashboard. It includes the start, stop, transfer, and deserialization time of every task for a particular duration.

Note that the task stream diagnostic does not run by default. You may wish to call this function once before you start work to ensure that things start recording, and then again after you have completed.

Parameters

startNumber or string: When you want to start recording If a number it should be the result of calling time() If a string then it should be a time difference before now, like ’60s’ or ‘500 ms’
stopNumber or string: When you want to stop recording
countint: The number of desired records, ignored if both start and stop are specified
plotboolean, str: If true then also return a Bokeh figure If plot == ‘save’ then save the figure to a file
filenamestr (optional): The filename to save to if you set plot='save'
bokeh_resourcesbokeh.resources.Resources (optional): Specifies if the resource component is INLINE or CDN

Returns

L: List[Dict]

See also

get_task_stream: a context manager version of this method

Examples

>>> client.get_task_stream()  # prime plugin if not already connected
>>> x.compute()  # do some work
>>> client.get_task_stream()
[{'task': ...,
  'type': ...,
  'thread': ...,
  ...}]

Pass the plot=True or plot='save' keywords to get back a Bokeh figure

>>> data, figure = client.get_task_stream(plot='save', filename='myfile.html')

Alternatively consider the context manager

>>> from dask.distributed import get_task_stream
>>> with get_task_stream() as ts:
...     x.compute()
>>> ts.data
[...]

get_versions(check: bool = False, packages: collections.abc.Sequence[str] | None = None) → distributed.client.VersionsDict | collections.abc.Coroutine[Any, Any, distributed.client.VersionsDict][source]¶

Return version info for the scheduler, all workers and myself

Parameters

check: raise ValueError if all required & optional packages do not match
packages: Extra package names to check

Examples

>>> c.get_versions()  

>>> c.get_versions(packages=['sklearn', 'geopandas'])  

get_worker_logs(n=None, workers=None, nanny=False)[source]¶

Get logs from workers

Parameters

nint: Number of logs to retrieve. Maxes out at 10000 by default, configurable via the distributed.admin.log-length configuration value.
workersiterable: List of worker addresses to retrieve. Gets all workers by default.
nannybool, default False: Whether to get the logs from the workers (False) or the nannies (True). If specified, the addresses in workers should still be the worker addresses, not the nanny addresses.

Returns

Dictionary mapping worker address to logs.
Logs are returned in reversed order (newest first)

has_what(workers=None, **kwargs)[source]¶

Which keys are held by which workers

This returns the keys of the data that are held in each worker’s memory.

Parameters

workerslist (optional): A list of worker addresses, defaults to all
**kwargsdict: Optional keyword arguments for the remote function

See also

Client.who_has
Client.nthreads
Client.processing

Examples

>>> x, y, z = c.map(inc, [1, 2, 3])  
>>> wait([x, y, z])  
>>> c.has_what()  
{'192.168.1.141:46784': ['inc-1c8dd6be1c21646c71f76c16d09304ea',
                         'inc-fd65c238a7ea60f6a01bf4c8a5fcf44b',
                         'inc-1e297fc27658d7b67b3a758f16bcf47a']}

list_datasets(**kwargs)[source]¶

List named datasets available on the scheduler

See also

Client.publish_dataset
Client.get_dataset

log_event(topic: str | collections.abc.Collection[str], msg: Any)[source]¶

Log an event under a given topic

Parameters

topicstr, list[str]: Name of the topic under which to log an event. To log the same event under multiple topics, pass a list of topic names.
msg: Event message to log. Note this must be msgpack serializable.

Examples

>>> from time import time
>>> client.log_event("current-time", time())

map(func: Callable, *iterables: collections.abc.Collection, key: str | list | None = None, workers: str | collections.abc.Iterable[str] | None = None, retries: int | None = None, resources: dict[str, Any] | None = None, priority: int = 0, allow_other_workers: bool = False, fifo_timeout: str = '100 ms', actor: bool = False, actors: bool = False, pure: bool = True, batch_size=None, **kwargs)[source]¶

Map a function on a sequence of arguments

Arguments can be normal objects or Futures

Parameters

funccallable: Callable to be scheduled for execution. If func returns a coroutine, it will be run on the main event loop of a worker. Otherwise func will be run in a worker’s task executor pool (see Worker.executors for more information.)
iterablesIterables: List-like objects to map over. They should have the same length.
keystr, list: Prefix for task names if string. Explicit names if list.
workersstring or iterable of strings: A set of worker hostnames on which computations may be performed. Leave empty to default to all workers (common case)
retriesint (default to 0): Number of allowed automatic retries if a task fails
resourcesdict (defaults to {}): Defines the resources each instance of this mapped task requires on the worker; e.g. {'GPU': 2}. See worker resources for details on defining resources.
priorityNumber: Optional prioritization of task. Zero is default. Higher priorities take precedence
allow_other_workersbool (defaults to False): Used with workers. Indicates whether or not the computations may be performed on workers that are not in the workers set(s).
fifo_timeoutstr timedelta (default ‘100ms’): Allowed amount of time between calls to consider the same priority
actorbool (default False): Whether these tasks should exist on the worker as stateful actors. See Actors for additional details.
actorsbool (default False): Alias for actor
purebool (defaults to True): Whether or not the function is pure. Set pure=False for impure functions like np.random.random. Note that if both actor and pure kwargs are set to True, then the value of pure will be reverted to False, since an actor is stateful. See Pure Functions by Default for more details.
batch_sizeint, optional (default: just one batch whose size is the entire iterable): Submit tasks to the scheduler in batches of (at most) batch_size. The tradeoff in batch size is that large batches avoid more per-batch overhead, but batches that are too big can take a long time to submit and unreasonably delay the cluster from starting its processing.
**kwargsdict: Extra keyword arguments to send to the function. Large values will be included explicitly in the task graph.

Returns

List, iterator, or Queue of futures, depending on the type of the
inputs.

See also

Client.submit: Submit a single function

Notes

The current implementation of a task graph resolution searches for occurrences of key and replaces it with a corresponding Future result. That can lead to unwanted substitution of strings passed as arguments to a task if these strings match some key that already exists on a cluster. To avoid these situations it is required to use unique values if a key is set manually. See https://github.com/dask/dask/issues/9969 to track progress on resolving this issue.

Examples

>>> L = client.map(func, sequence)  

nbytes(keys=None, summary=True, **kwargs)[source]¶

The bytes taken up by each key on the cluster

This is as measured by sys.getsizeof which may not accurately reflect the true cost.

Parameters

keyslist (optional): A list of keys, defaults to all keys
summaryboolean, (optional): Summarize keys into key types
**kwargsdict: Optional keyword arguments for the remote function

See also

Client.who_has

Examples

>>> x, y, z = c.map(inc, [1, 2, 3])  
>>> c.nbytes(summary=False)  
{'inc-1c8dd6be1c21646c71f76c16d09304ea': 28,
 'inc-1e297fc27658d7b67b3a758f16bcf47a': 28,
 'inc-fd65c238a7ea60f6a01bf4c8a5fcf44b': 28}

>>> c.nbytes(summary=True)  
{'inc': 84}

ncores(workers=None, **kwargs)¶

The number of threads/cores available on each worker node

Parameters

workerslist (optional): A list of workers that we care about specifically. Leave empty to receive information about all workers.
**kwargsdict: Optional keyword arguments for the remote function

See also

Client.who_has
Client.has_what

Examples

>>> c.nthreads()  
{'192.168.1.141:46784': 8,
 '192.167.1.142:47548': 8,
 '192.167.1.143:47329': 8,
 '192.167.1.144:37297': 8}

normalize_collection(collection)[source]¶

Replace collection’s tasks by already existing futures if they exist

This normalizes the tasks within a collections task graph against the known futures within the scheduler. It returns a copy of the collection with a task graph that includes the overlapping futures.

Parameters

collectiondask object: Collection like dask.array or dataframe or dask.value objects

Returns

collectiondask object: Collection with its tasks replaced with any existing futures.

See also

Client.persist: trigger computation of collection’s tasks

Examples

>>> len(x.__dask_graph__())  # x is a dask collection with 100 tasks  
100
>>> set(client.futures).intersection(x.__dask_graph__())  # some overlap exists  
10

>>> x = client.normalize_collection(x)  
>>> len(x.__dask_graph__())  # smaller computational graph  
20

nthreads(workers=None, **kwargs)[source]¶

The number of threads/cores available on each worker node

Parameters

workerslist (optional): A list of workers that we care about specifically. Leave empty to receive information about all workers.
**kwargsdict: Optional keyword arguments for the remote function

See also

Client.who_has
Client.has_what

Examples

>>> c.nthreads()  
{'192.168.1.141:46784': 8,
 '192.167.1.142:47548': 8,
 '192.167.1.143:47329': 8,
 '192.167.1.144:37297': 8}

persist(collections, optimize_graph=True, workers=None, allow_other_workers=None, resources=None, retries=None, priority=0, fifo_timeout='60s', actors=None, **kwargs)[source]¶

Persist dask collections on cluster

Starts computation of the collection on the cluster in the background. Provides a new dask collection that is semantically identical to the previous one, but now based off of futures currently in execution.

Parameters

collectionssequence or single dask object: Collections like dask.array or dataframe or dask.value objects
optimize_graphbool: Whether or not to optimize the underlying graphs
workersstring or iterable of strings: A set of worker hostnames on which computations may be performed. Leave empty to default to all workers (common case)
allow_other_workersbool (defaults to False): Used with workers. Indicates whether or not the computations may be performed on workers that are not in the workers set(s).
retriesint (default to 0): Number of allowed automatic retries if computing a result fails
priorityNumber: Optional prioritization of task. Zero is default. Higher priorities take precedence
fifo_timeouttimedelta str (defaults to ’60s’): Allowed amount of time between calls to consider the same priority
resourcesdict (defaults to {}): Defines the resources each instance of this mapped task requires on the worker; e.g. {'GPU': 2}. See worker resources for details on defining resources.
actorsbool or dict (default None): Whether these tasks should exist on the worker as stateful actors. Specified on a global (True/False) or per-task ({'x': True, 'y': False}) basis. See Actors for additional details.
**kwargs: Options to pass to the graph optimize calls

Returns

List of collections, or single collection, depending on type of input.

See also

Client.compute

Examples

>>> xx = client.persist(x)  
>>> xx, yy = client.persist([x, y])  

processing(workers=None)[source]¶

The tasks currently running on each worker

Parameters

workerslist (optional): A list of worker addresses, defaults to all

See also

Client.who_has
Client.has_what
Client.nthreads

Examples

>>> x, y, z = c.map(inc, [1, 2, 3])  
>>> c.processing()  
{'192.168.1.141:46784': ['inc-1c8dd6be1c21646c71f76c16d09304ea',
                         'inc-fd65c238a7ea60f6a01bf4c8a5fcf44b',
                         'inc-1e297fc27658d7b67b3a758f16bcf47a']}

profile(key=None, start=None, stop=None, workers=None, merge_workers=True, plot=False, filename=None, server=False, scheduler=False)[source]¶

Collect statistical profiling information about recent work

Parameters

keystr: Key prefix to select, this is typically a function name like ‘inc’ Leave as None to collect all data
starttime
stoptime
workerslist: List of workers to restrict profile information
serverbool: If true, return the profile of the worker’s administrative thread rather than the worker threads. This is useful when profiling Dask itself, rather than user code.
schedulerbool: If true, return the profile information from the scheduler’s administrative thread rather than the workers. This is useful when profiling Dask’s scheduling itself.
plotboolean or string: Whether or not to return a plot object
filenamestr: Filename to save the plot

Examples

>>> client.profile()  # call on collections
>>> client.profile(filename='dask-profile.html')  # save to html file

publish_dataset(*args, **kwargs)[source]¶

Publish named datasets to scheduler

This stores a named reference to a dask collection or list of futures on the scheduler. These references are available to other Clients which can download the collection or futures with get_dataset.

Datasets are not immediately computed. You may wish to call Client.persist prior to publishing a dataset.

Parameters

argslist of objects to publish as name
kwargsdict: named collections to publish on the scheduler

Returns

None

See also

Client.list_datasets
Client.get_dataset
Client.unpublish_dataset
Client.persist

Examples

Publishing client:

>>> df = dd.read_csv('s3://...')  
>>> df = c.persist(df) 
>>> c.publish_dataset(my_dataset=df)  

Alternative invocation >>> c.publish_dataset(df, name=’my_dataset’)

Receiving client:

>>> c.list_datasets()  
['my_dataset']
>>> df2 = c.get_dataset('my_dataset')  

rebalance(futures=None, workers=None, **kwargs)[source]¶

Rebalance data within network

Move data between workers to roughly balance memory burden. This either affects a subset of the keys/workers or the entire network, depending on keyword arguments.

For details on the algorithm and configuration options, refer to the matching scheduler-side method rebalance().

Warning

This operation is generally not well tested against normal operation of the scheduler. It is not recommended to use it while waiting on computations.

Parameters

futureslist, optional: A list of futures to balance, defaults all data
workerslist, optional: A list of workers on which to balance, defaults to all workers
**kwargsdict: Optional keyword arguments for the function

register_plugin(plugin: distributed.diagnostics.plugin.NannyPlugin | distributed.diagnostics.plugin.SchedulerPlugin | distributed.diagnostics.plugin.WorkerPlugin, name: str | None = None, idempotent: bool | None = None)[source]¶

See https://distributed.readthedocs.io/en/latest/plugins.html

Parameters

plugin: A nanny, scheduler, or worker plugin to register.
name: Name for the plugin; if None, a name is taken from the plugin instance or automatically generated if not present.
idempotent: Do not re-register if a plugin of the given name already exists. If None, plugin.idempotent is taken if defined, False otherwise.

register_scheduler_plugin(plugin: distributed.diagnostics.plugin.SchedulerPlugin, name: str | None = None, idempotent: bool | None = None)[source]¶

Deprecated since version 2023.9.2: Use Client.register_plugin() instead.

See https://distributed.readthedocs.io/en/latest/plugins.html#scheduler-plugins

Parameters

pluginSchedulerPlugin: SchedulerPlugin instance to pass to the scheduler.
namestr: Name for the plugin; if None, a name is taken from the plugin instance or automatically generated if not present.
idempotentbool: Do not re-register if a plugin of the given name already exists.

register_worker_callbacks(setup=None)[source]¶

Registers a setup callback function for all current and future workers.

This registers a new setup function for workers in this cluster. The function will run immediately on all currently connected workers. It will also be run upon connection by any workers that are added in the future. Multiple setup functions can be registered - these will be called in the order they were added.

If the function takes an input argument named dask_worker then that variable will be populated with the worker itself.

Parameters

setupcallable(dask_worker: Worker) -> None: Function to register and run on all workers

register_worker_plugin(plugin: distributed.diagnostics.plugin.NannyPlugin | distributed.diagnostics.plugin.WorkerPlugin, name: str | None = None, nanny: bool | None = None)[source]¶

Registers a lifecycle worker plugin for all current and future workers.

Deprecated since version 2023.9.2: Use Client.register_plugin() instead.

This registers a new object to handle setup, task state transitions and teardown for workers in this cluster. The plugin will instantiate itself on all currently connected workers. It will also be run on any worker that connects in the future.

The plugin may include methods setup, teardown, transition, and release_key. See the dask.distributed.WorkerPlugin class or the examples below for the interface and docstrings. It must be serializable with the pickle or cloudpickle modules.

If the plugin has a name attribute, or if the name= keyword is used then that will control idempotency. If a plugin with that name has already been registered, then it will be removed and replaced by the new one.

For alternatives to plugins, you may also wish to look into preload scripts.

Parameters

pluginWorkerPlugin or NannyPlugin: WorkerPlugin or NannyPlugin instance to register.
namestr, optional: A name for the plugin. Registering a plugin with the same name will have no effect. If plugin has no name attribute a random name is used.
nannybool, optional: Whether to register the plugin with workers or nannies.

See also

distributed.WorkerPlugin
unregister_worker_plugin

Examples

>>> class MyPlugin(WorkerPlugin):
...     def __init__(self, *args, **kwargs):
...         pass  # the constructor is up to you
...     def setup(self, worker: dask.distributed.Worker):
...         pass
...     def teardown(self, worker: dask.distributed.Worker):
...         pass
...     def transition(self, key: str, start: str, finish: str,
...                    **kwargs):
...         pass
...     def release_key(self, key: str, state: str, cause: str | None, reason: None, report: bool):
...         pass

>>> plugin = MyPlugin(1, 2, 3)
>>> client.register_plugin(plugin)

You can get access to the plugin with the get_worker function

>>> client.register_plugin(other_plugin, name='my-plugin')
>>> def f():
...    worker = get_worker()
...    plugin = worker.plugins['my-plugin']
...    return plugin.my_state

>>> future = client.run(f)

replicate(futures, n=None, workers=None, branching_factor=2, **kwargs)[source]¶

Set replication of futures within network

Copy data onto many workers. This helps to broadcast frequently accessed data and can improve resilience.

This performs a tree copy of the data throughout the network individually on each piece of data. This operation blocks until complete. It does not guarantee replication of data to future workers.

Note

This method is incompatible with the Active Memory Manager’s ReduceReplicas policy. If you wish to use it, you must first disable the policy or disable the AMM entirely.

Parameters

futureslist of futures: Futures we wish to replicate
nint, optional: Number of processes on the cluster on which to replicate the data. Defaults to all.
workerslist of worker addresses: Workers on which we want to restrict the replication. Defaults to all.
branching_factorint, optional: The number of workers that can copy data in each generation
**kwargsdict: Optional keyword arguments for the remote function

See also

Client.rebalance

Examples

>>> x = c.submit(func, *args)  
>>> c.replicate([x])  # send to all workers  
>>> c.replicate([x], n=3)  # send to three workers  
>>> c.replicate([x], workers=['alice', 'bob'])  # send to specific  
>>> c.replicate([x], n=1, workers=['alice', 'bob'])  # send to one of specific workers  
>>> c.replicate([x], n=1)  # reduce replications 

restart(timeout: typing.Union[str, int, float, typing.Literal[<no_default>]] = _NoDefault.no_default, wait_for_workers: bool = True)[source]¶

Restart all workers. Reset local state. Optionally wait for workers to return.

Workers without nannies are shut down, hoping an external deployment system will restart them. Therefore, if not using nannies and your deployment system does not automatically restart workers, restart will just shut down all workers, then time out!

After restart, all connected workers are new, regardless of whether TimeoutError was raised. Any workers that failed to shut down in time are removed, and may or may not shut down on their own in the future.

Parameters

timeout:: How long to wait for workers to shut down and come back, if wait_for_workers is True, otherwise just how long to wait for workers to shut down. Raises asyncio.TimeoutError if this is exceeded.
wait_for_workers:: Whether to wait for all workers to reconnect, or just for them to shut down (default True). Use restart(wait_for_workers=False) combined with Client.wait_for_workers() for granular control over how many workers to wait for.

See also

Scheduler.restart
Client.restart_workers

restart_workers(workers: list[str], timeout: typing.Union[str, int, float, typing.Literal[<no_default>]] = _NoDefault.no_default, raise_for_error: bool = True)[source]¶

Restart a specified set of workers

Note

Only workers being monitored by a distributed.Nanny can be restarted. See Nanny.restart for more details.

Parameters

workerslist[str]: Workers to restart. This can be a list of worker addresses, names, or a both.
timeoutint | float | None: Number of seconds to wait
raise_for_error: bool (default True): Whether to raise a TimeoutError if restarting worker(s) doesn’t finish within timeout, or another exception caused from restarting worker(s).

Returns

dict[str, “OK” | “removed” | “timed out”]: Mapping of worker and restart status, the keys will match the original values passed in via workers.

See also

Client.restart

Notes

This method differs from Client.restart() in that this method simply restarts the specified set of workers, while Client.restart will restart all workers and also reset local state on the cluster (e.g. all keys are released).

Additionally, this method does not gracefully handle tasks that are being executed when a worker is restarted. These tasks may fail or have their suspicious count incremented.

Examples

You can get information about active workers using the following:

>>> workers = client.scheduler_info()['workers']

From that list you may want to select some workers to restart

>>> client.restart_workers(workers=['tcp://address:port', ...])

retire_workers(workers: list[str] | None = None, close_workers: bool = True, **kwargs)[source]¶

Retire certain workers on the scheduler

See distributed.Scheduler.retire_workers() for the full docstring.

Parameters

workers
close_workers
**kwargsdict: Optional keyword arguments for the remote function

See also

dask.distributed.Scheduler.retire_workers

Examples

You can get information about active workers using the following:

>>> workers = client.scheduler_info()['workers']

From that list you may want to select some workers to close

>>> client.retire_workers(workers=['tcp://address:port', ...])

retry(futures, asynchronous=None)[source]¶

Retry failed futures

Parameters

futureslist of Futures: The list of Futures
asynchronous: bool: If True the client is in asynchronous mode

run(function, *args, workers: list[str] | None = None, wait: bool = True, nanny: bool = False, on_error: Literal['raise', 'return', 'ignore'] = 'raise', **kwargs)[source]¶

Run a function on all workers outside of task scheduling system

This calls a function on all currently known workers immediately, blocks until those results come back, and returns the results asynchronously as a dictionary keyed by worker address. This method is generally used for side effects such as collecting diagnostic information or installing libraries.

If your function takes an input argument named dask_worker then that variable will be populated with the worker itself.

Parameters

functioncallable

The function to run

*argstuple

Optional arguments for the remote function

**kwargsdict

Optional keyword arguments for the remote function

workerslist

Workers on which to run the function. Defaults to all known workers.

waitboolean (optional)

If the function is asynchronous whether or not to wait until that function finishes.

nannybool, default False

Whether to run function on the nanny. By default, the function is run on the worker process. If specified, the addresses in workers should still be the worker addresses, not the nanny addresses.

on_error: “raise” | “return” | “ignore”

If the function raises an error on a worker:

raise: (default) Re-raise the exception on the client. The output from other workers will be lost.
return: Return the Exception object instead of the function output for the worker
ignore: Ignore the exception and remove the worker from the result dict

Examples

>>> c.run(os.getpid)  
{'192.168.0.100:9000': 1234,
 '192.168.0.101:9000': 4321,
 '192.168.0.102:9000': 5555}

Restrict computation to particular workers with the workers= keyword argument.

>>> c.run(os.getpid, workers=['192.168.0.100:9000',
...                           '192.168.0.101:9000'])  
{'192.168.0.100:9000': 1234,
 '192.168.0.101:9000': 4321}

>>> def get_status(dask_worker):
...     return dask_worker.status

>>> c.run(get_status)  
{'192.168.0.100:9000': 'running',
 '192.168.0.101:9000': 'running}

Run asynchronous functions in the background:

>>> async def print_state(dask_worker):  
...    while True:
...        print(dask_worker.status)
...        await asyncio.sleep(1)

>>> c.run(print_state, wait=False)  

run_on_scheduler(function, *args, **kwargs)[source]¶

Run a function on the scheduler process

This is typically used for live debugging. The function should take a keyword argument dask_scheduler=, which will be given the scheduler object itself.

Parameters

functioncallable: The function to run on the scheduler process
*argstuple: Optional arguments for the function
**kwargsdict: Optional keyword arguments for the function

See also

Client.run: Run a function on all workers

Examples

>>> def get_number_of_tasks(dask_scheduler=None):
...     return len(dask_scheduler.tasks)

>>> client.run_on_scheduler(get_number_of_tasks)  
100

Run asynchronous functions in the background:

>>> async def print_state(dask_scheduler):  
...    while True:
...        print(dask_scheduler.status)
...        await asyncio.sleep(1)

>>> c.run(print_state, wait=False)  

scatter(data, workers=None, broadcast=False, direct=None, hash=True, timeout=_NoDefault.no_default, asynchronous=None)[source]¶

Scatter data into distributed memory

This moves data from the local client process into the workers of the distributed scheduler. Note that it is often better to submit jobs to your workers to have them load the data rather than loading data locally and then scattering it out to them.

Parameters

datalist, dict, or object: Data to scatter out to workers. Output type matches input type.
workerslist of tuples (optional): Optionally constrain locations of data. Specify workers as hostname/port pairs, e.g. ('127.0.0.1', 8787).
broadcastbool (defaults to False): Whether to send each data element to all workers. By default we round-robin based on number of cores.

Note

Setting this flag to True is incompatible with the Active Memory Manager’s ReduceReplicas policy. If you wish to use it, you must first disable the policy or disable the AMM entirely.
directbool (defaults to automatically check): Whether or not to connect directly to the workers, or to ask the scheduler to serve as intermediary. This can also be set when creating the Client.
hashbool (optional): Whether or not to hash data to determine key. If False then this uses a random key
timeoutnumber, optional: Time in seconds after which to raise a dask.distributed.TimeoutError
asynchronous: bool: If True the client is in asynchronous mode

Returns

List, dict, iterator, or queue of futures matching the type of input.

See also

Client.gather: Gather data back to local process

Notes

Scattering a dictionary uses dict keys to create Future keys. The current implementation of a task graph resolution searches for occurrences of key and replaces it with a corresponding Future result. That can lead to unwanted substitution of strings passed as arguments to a task if these strings match some key that already exists on a cluster. To avoid these situations it is required to use unique values if a key is set manually. See https://github.com/dask/dask/issues/9969 to track progress on resolving this issue.

Examples

>>> c = Client('127.0.0.1:8787')  
>>> c.scatter(1) 
<Future: status: finished, key: c0a8a20f903a4915b94db8de3ea63195>

>>> c.scatter([1, 2, 3])  
[<Future: status: finished, key: c0a8a20f903a4915b94db8de3ea63195>,
 <Future: status: finished, key: 58e78e1b34eb49a68c65b54815d1b158>,
 <Future: status: finished, key: d3395e15f605bc35ab1bac6341a285e2>]

>>> c.scatter({'x': 1, 'y': 2, 'z': 3})  
{'x': <Future: status: finished, key: x>,
 'y': <Future: status: finished, key: y>,
 'z': <Future: status: finished, key: z>}

Constrain location of data to subset of workers

>>> c.scatter([1, 2, 3], workers=[('hostname', 8788)])   

Broadcast data to all workers

>>> [future] = c.scatter([element], broadcast=True)  

Send scattered data to parallelized function using client futures interface

>>> data = c.scatter(data, broadcast=True)  
>>> res = [c.submit(func, data, i) for i in range(100)]

scheduler_info(**kwargs)[source]¶

Basic information about the workers in the cluster

Parameters

**kwargsdict: Optional keyword arguments for the remote function

Examples

>>> c.scheduler_info()  
{'id': '2de2b6da-69ee-11e6-ab6a-e82aea155996',
 'services': {},
 'type': 'Scheduler',
 'workers': {'127.0.0.1:40575': {'active': 0,
                                 'last-seen': 1472038237.4845693,
                                 'name': '127.0.0.1:40575',
                                 'services': {},
                                 'stored': 0,
                                 'time-delay': 0.0061032772064208984}}}

set_metadata(key, value)[source]¶

Set arbitrary metadata in the scheduler

This allows you to store small amounts of data on the central scheduler process for administrative purposes. Data should be msgpack serializable (ints, strings, lists, dicts)

If the key corresponds to a task then that key will be cleaned up when the task is forgotten by the scheduler.

If the key is a list then it will be assumed that you want to index into a nested dictionary structure using those keys. For example if you call the following:

>>> client.set_metadata(['a', 'b', 'c'], 123)

Then this is the same as setting

>>> scheduler.task_metadata['a']['b']['c'] = 123

The lower level dictionaries will be created on demand.

See also

get_metadata

Examples

>>> client.set_metadata('x', 123)  
>>> client.get_metadata('x')  
123

>>> client.set_metadata(['x', 'y'], 123)  
>>> client.get_metadata('x')  
{'y': 123}

>>> client.set_metadata(['x', 'w', 'z'], 456)  
>>> client.get_metadata('x')  
{'y': 123, 'w': {'z': 456}}

>>> client.get_metadata(['x', 'w'])  
{'z': 456}

shutdown()[source]¶

Shut down the connected scheduler and workers

Note, this may disrupt other clients that may be using the same scheduler and workers.

See also

Client.close: close only this client

start(**kwargs)[source]¶: Start scheduler running in separate thread

story(*keys_or_stimuli, on_error='raise')[source]¶: Returns a cluster-wide story for the given keys or stimulus_id’s

submit(func, *args, key=None, workers=None, resources=None, retries=None, priority=0, fifo_timeout='100 ms', allow_other_workers=False, actor=False, actors=False, pure=True, **kwargs)[source]¶

Submit a function application to the scheduler

Parameters

funccallable: Callable to be scheduled as func(*args **kwargs). If func returns a coroutine, it will be run on the main event loop of a worker. Otherwise func will be run in a worker’s task executor pool (see Worker.executors for more information.)
*argstuple: Optional positional arguments
keystr: Unique identifier for the task. Defaults to function-name and hash
workersstring or iterable of strings: A set of worker addresses or hostnames on which computations may be performed. Leave empty to default to all workers (common case)
resourcesdict (defaults to {}): Defines the resources each instance of this mapped task requires on the worker; e.g. {'GPU': 2}. See worker resources for details on defining resources.
retriesint (default to 0): Number of allowed automatic retries if the task fails
priorityNumber: Optional prioritization of task. Zero is default. Higher priorities take precedence
fifo_timeoutstr timedelta (default ‘100ms’): Allowed amount of time between calls to consider the same priority
allow_other_workersbool (defaults to False): Used with workers. Indicates whether or not the computations may be performed on workers that are not in the workers set(s).
actorbool (default False): Whether this task should exist on the worker as a stateful actor. See Actors for additional details.
actorsbool (default False): Alias for actor
purebool (defaults to True): Whether or not the function is pure. Set pure=False for impure functions like np.random.random. Note that if both actor and pure kwargs are set to True, then the value of pure will be reverted to False, since an actor is stateful. See Pure Functions by Default for more details.
**kwargs

Returns

Future: If running in asynchronous mode, returns the future. Otherwise returns the concrete value

Raises

TypeError: If ‘func’ is not callable, a TypeError is raised
ValueError: If ‘allow_other_workers’is True and ‘workers’ is None, a ValueError is raised

See also

Client.map: Submit on many arguments at once

Notes

Examples

>>> c = client.submit(add, a, b)  

subscribe_topic(topic, handler)[source]¶

Subscribe to a topic and execute a handler for every received event

Parameters

topic: str: The topic name
handler: callable or coroutine function: A handler called for every received event. The handler must accept a single argument event which is a tuple (timestamp, msg) where timestamp refers to the clock on the scheduler.

See also

dask.distributed.Client.unsubscribe_topic
dask.distributed.Client.get_events
dask.distributed.Client.log_event

Examples

>>> import logging
>>> logger = logging.getLogger("myLogger")  # Log config not shown
>>> client.subscribe_topic("topic-name", lambda: logger.info)

unforward_logging(logger_name=None)[source]¶: Stop forwarding the given logger (default root) from worker tasks to the client process.

unpublish_dataset(name, **kwargs)[source]¶

Remove named datasets from scheduler

Parameters

namestr: The name of the dataset to unpublish

See also

Client.publish_dataset

Examples

>>> c.list_datasets()  
['my_dataset']
>>> c.unpublish_dataset('my_dataset')  
>>> c.list_datasets()  
[]

unregister_scheduler_plugin(name)[source]¶

Unregisters a scheduler plugin

See https://distributed.readthedocs.io/en/latest/plugins.html#scheduler-plugins

Parameters

namestr: Name of the plugin to unregister. See the Client.register_scheduler_plugin() docstring for more information.

See also

register_scheduler_plugin

Examples

>>> class MyPlugin(SchedulerPlugin):
...     def __init__(self, *args, **kwargs):
...         pass  # the constructor is up to you
...     async def start(self, scheduler: Scheduler) -> None:
...         pass
...     async def before_close(self) -> None:
...         pass
...     async def close(self) -> None:
...         pass
...     def restart(self, scheduler: Scheduler) -> None:
...         pass

>>> plugin = MyPlugin(1, 2, 3)
>>> client.register_plugin(plugin, name='foo')
>>> client.unregister_scheduler_plugin(name='foo')

unregister_worker_plugin(name, nanny=None)[source]¶

Unregisters a lifecycle worker plugin

This unregisters an existing worker plugin. As part of the unregistration process the plugin’s teardown method will be called.

Parameters

namestr: Name of the plugin to unregister. See the Client.register_plugin() docstring for more information.

See also

register_plugin

Examples

>>> class MyPlugin(WorkerPlugin):
...     def __init__(self, *args, **kwargs):
...         pass  # the constructor is up to you
...     def setup(self, worker: dask.distributed.Worker):
...         pass
...     def teardown(self, worker: dask.distributed.Worker):
...         pass
...     def transition(self, key: str, start: str, finish: str, **kwargs):
...         pass
...     def release_key(self, key: str, state: str, cause: str | None, reason: None, report: bool):
...         pass

>>> plugin = MyPlugin(1, 2, 3)
>>> client.register_plugin(plugin, name='foo')
>>> client.unregister_worker_plugin(name='foo')

unsubscribe_topic(topic)[source]¶

Unsubscribe from a topic and remove event handler

See also

dask.distributed.Client.subscribe_topic
dask.distributed.Client.get_events
dask.distributed.Client.log_event

upload_file(filename, load: bool = True)[source]¶

Upload local package to scheduler and workers

This sends a local file up to the scheduler and all worker nodes. This file is placed into the working directory of each node, see config option temporary-directory (defaults to tempfile.gettempdir()).

This directory will be added to the Python’s system path so any .py, .egg or .zip files will be importable.

Parameters

filenamestring: Filename of .py, .egg, or .zip file to send to workers
loadbool, optional: Whether or not to import the module as part of the upload process. Defaults to True.

Examples

>>> client.upload_file('mylibrary.egg')  
>>> from mylibrary import myfunc  
>>> L = client.map(myfunc, seq)  
>>>
>>> # Where did that file go? Use `dask_worker.local_directory`.
>>> def where_is_mylibrary(dask_worker):
>>>     path = pathlib.Path(dask_worker.local_directory) / 'mylibrary.egg'
>>>     assert path.exists()
>>>     return str(path)
>>>
>>> client.run(where_is_mylibrary)  

wait_for_workers(n_workers: int, timeout: float | None = None) → None[source]¶

Blocking call to wait for n workers before continuing

Parameters

n_workersint: The number of workers
timeoutnumber, optional: Time in seconds after which to raise a dask.distributed.TimeoutError

who_has(futures=None, **kwargs)[source]¶

The workers storing each future’s data

Parameters

futureslist (optional): A list of futures, defaults to all data
**kwargsdict: Optional keyword arguments for the remote function

See also

Client.has_what
Client.nthreads

Examples

>>> x, y, z = c.map(inc, [1, 2, 3])  
>>> wait([x, y, z])  
>>> c.who_has()  
{'inc-1c8dd6be1c21646c71f76c16d09304ea': ['192.168.1.141:46784'],
 'inc-1e297fc27658d7b67b3a758f16bcf47a': ['192.168.1.141:46784'],
 'inc-fd65c238a7ea60f6a01bf4c8a5fcf44b': ['192.168.1.141:46784']}

>>> c.who_has([x, y])  
{'inc-1c8dd6be1c21646c71f76c16d09304ea': ['192.168.1.141:46784'],
 'inc-1e297fc27658d7b67b3a758f16bcf47a': ['192.168.1.141:46784']}

write_scheduler_file(scheduler_file)[source]¶

Write the scheduler information to a json file.

This facilitates easy sharing of scheduler information using a file system. The scheduler file can be used to instantiate a second Client using the same scheduler.

Parameters

scheduler_filestr: Path to a write the scheduler file.

Examples

>>> client = Client()  
>>> client.write_scheduler_file('scheduler.json')  
# connect to previous client's scheduler
>>> client2 = Client(scheduler_file='scheduler.json')  

class distributed.Future(key, client=None, state=None, _id=None)[source]¶

A remotely running computation

A Future is a local proxy to a result running on a remote worker. A user manages future objects in the local Python process to determine what happens in the larger cluster.

Note

Users should not instantiate futures manually. This can lead to state corruption and deadlocking clusters.

Parameters

key: str, or tuple: Key of remote data to which this future refers
client: Client: Client that should own this future. Defaults to _get_global_client()
inform: bool: Do we inform the scheduler that we need an update on this future
state: FutureState: The state of the future

See also

Client: Creates futures

Examples

Futures typically emerge from Client computations

>>> my_future = client.submit(add, 1, 2)  

We can track the progress and results of a future

>>> my_future  
<Future: status: finished, key: add-8f6e709446674bad78ea8aeecfee188e>

We can get the result or the exception and traceback from the future

>>> my_future.result()  

add_done_callback(fn)[source]¶

Call callback on future when future has finished

The callback fn should take the future as its only argument. This will be called regardless of if the future completes successfully, errs, or is cancelled

The callback is executed in a separate thread.

Parameters

fncallable: The method or function to be called

cancel(reason=None, msg=None, **kwargs)[source]¶

Cancel the request to run this future

See also

Client.cancel

cancelled()[source]¶

Returns True if the future has been cancelled

Returns

bool: True if the future was ‘cancelled’, otherwise False

done()[source]¶

Returns whether or not the computation completed.

Returns

bool: True if the computation is complete, otherwise False

exception(timeout=None, **kwargs)[source]¶

Return the exception of a failed task

Parameters

timeoutnumber, optional: Time in seconds after which to raise a dask.distributed.TimeoutError
**kwargsdict: Optional keyword arguments for the function

Returns

Exception: The exception that was raised If timeout seconds are elapsed before returning, a dask.distributed.TimeoutError is raised.

See also

Future.traceback

property executor¶

Returns the executor, which is the client.

Returns

Client: The executor

release()[source]¶

Notes

This method can be called from different threads (see e.g. Client.get() or Future.__del__())

result(timeout=None)[source]¶

Wait until computation completes, gather result to local process.

Parameters

timeoutnumber, optional: Time in seconds after which to raise a dask.distributed.TimeoutError

Returns

result: The result of the computation. Or a coroutine if the client is asynchronous.

Raises

dask.distributed.TimeoutError: If timeout seconds are elapsed before returning, a dask.distributed.TimeoutError is raised.

retry(**kwargs)[source]¶

Retry this future if it has failed

See also

Client.retry

property status¶

Returns the status

Returns

str: The status

traceback(timeout=None, **kwargs)[source]¶

Return the traceback of a failed task

This returns a traceback object. You can inspect this object using the traceback module. Alternatively if you call future.result() this traceback will accompany the raised exception.

Parameters

timeoutnumber, optional: Time in seconds after which to raise a dask.distributed.TimeoutError If timeout seconds are elapsed before returning, a dask.distributed.TimeoutError is raised.

Returns

traceback: The traceback object. Or a coroutine if the client is asynchronous.

See also

Future.exception

Examples

>>> import traceback  
>>> tb = future.traceback()  
>>> traceback.format_tb(tb)  
[...]

property type¶: Returns the type

class distributed.Queue(name=None, client=None, maxsize=0)[source]¶

Distributed Queue

This allows multiple clients to share futures or small bits of data between each other with a multi-producer/multi-consumer queue. All metadata is sequentialized through the scheduler.

Elements of the Queue must be either Futures or msgpack-encodable data (ints, strings, lists, dicts). All data is sent through the scheduler so it is wise not to send large objects. To share large objects scatter the data and share the future instead.

Warning

This object is experimental

Parameters

name: string (optional): Name used by other clients and the scheduler to identify the queue. If not given, a random name will be generated.
client: Client (optional): Client used for communication with the scheduler. If not given, the default global client will be used.
maxsize: int (optional): Number of items allowed in the queue. If 0 (the default), the queue size is unbounded.

See also

Variable: shared variable between clients

Examples

>>> from dask.distributed import Client, Queue  
>>> client = Client()  
>>> queue = Queue('x')  
>>> future = client.submit(f, x)  
>>> queue.put(future)  

get(timeout=None, batch=False, **kwargs)[source]¶

Get data from the queue

Parameters

timeoutnumber or string or timedelta, optional: Time in seconds to wait before timing out. Instead of number of seconds, it is also possible to specify a timedelta in string format, e.g. “200ms”.
batchboolean, int (optional): If True then return all elements currently waiting in the queue. If an integer than return that many elements from the queue If False (default) then return one item at a time

put(value, timeout=None, **kwargs)[source]¶

Put data into the queue

Parameters

timeoutnumber or string or timedelta, optional: Time in seconds to wait before timing out. Instead of number of seconds, it is also possible to specify a timedelta in string format, e.g. “200ms”.

qsize(**kwargs)[source]¶: Current number of elements in the queue

class distributed.Variable(name=None, client=None)[source]¶

Distributed Global Variable

This allows multiple clients to share futures and data between each other with a single mutable variable. All metadata is sequentialized through the scheduler. Race conditions can occur.

Values must be either Futures or msgpack-encodable data (ints, lists, strings, etc..) All data will be kept and sent through the scheduler, so it is wise not to send too much. If you want to share a large amount of data then scatter it and share the future instead.

Parameters

name: string (optional): Name used by other clients and the scheduler to identify the variable. If not given, a random name will be generated.
client: Client (optional): Client used for communication with the scheduler. If not given, the default global client will be used.

See also

Queue: shared multi-producer/multi-consumer queue between clients

Examples

>>> from dask.distributed import Client, Variable 
>>> client = Client()  
>>> x = Variable('x')  
>>> x.set(123)  # docttest: +SKIP
>>> x.get()  # docttest: +SKIP
123
>>> future = client.submit(f, x)  
>>> x.set(future)  

delete()[source]¶

Delete this variable

Caution, this affects all clients currently pointing to this variable.

get(timeout=None, **kwargs)[source]¶

Get the value of this variable

Parameters

timeoutnumber or string or timedelta, optional: Time in seconds to wait before timing out. Instead of number of seconds, it is also possible to specify a timedelta in string format, e.g. “200ms”.

set(value, timeout='30 s', **kwargs)[source]¶

Set the value of this variable

Parameters

valueFuture or object: Must be either a Future or a msgpack-encodable value

class distributed.Lock(name=None, client=<object object>, scheduler_rpc=None, loop=None)[source]¶

Distributed Centralized Lock

Warning

This is using the distributed.Semaphore as a backend, which is susceptible to lease overbooking. For the Lock this means that if a lease is timing out, two or more instances could acquire the lock at the same time. To disable lease timeouts, set distributed.scheduler.locks.lease-timeout to inf, e.g.

with dask.config.set({"distributed.scheduler.locks.lease-timeout": "inf"}):
    lock = Lock("x")
    ...

Note, that without lease timeouts, the Lock may deadlock in case of cluster downscaling or worker failures.

Parameters

name: string (optional): Name of the lock to acquire. Choosing the same name allows two disconnected processes to coordinate a lock. If not given, a random name will be generated.
client: Client (optional): Client to use for communication with the scheduler. If not given, the default global client will be used.

Examples

>>> lock = Lock('x')  
>>> lock.acquire(timeout=1)  
>>> # do things with protected resource
>>> lock.release()  

acquire(blocking=True, timeout=None)[source]¶

Acquire the lock

Parameters

blockingbool, optional: If false, don’t wait on the lock in the scheduler at all.
timeoutstring or number or timedelta, optional: Seconds to wait on the lock in the scheduler. This does not include local coroutine time, network transfer time, etc.. It is forbidden to specify a timeout when blocking is false. Instead of number of seconds, it is also possible to specify a timedelta in string format, e.g. “200ms”.

Returns

True or False whether or not it successfully acquired the lock

Examples

>>> lock = Lock('x')  
>>> lock.acquire(timeout="1s")  

class distributed.Event(name=None, client=None)[source]¶

Distributed Centralized Event equivalent to asyncio.Event

An event stores a single flag, which is set to false on start. The flag can be set to true (using the set() call) or back to false (with the clear() call). Every call to wait() blocks until the event flag is set to true.

Parameters

name: string (optional): Name of the event. Choosing the same name allows two disconnected processes to coordinate an event. If not given, a random name will be generated.
client: Client (optional): Client to use for communication with the scheduler. If not given, the default global client will be used.

Examples

>>> event_1 = Event('a')  
>>> event_1.wait(timeout=1)  
>>> # in another process
>>> event_2 = Event('a')  
>>> event_2.set() 
>>> # now event_1 will stop waiting

clear()[source]¶

Clear the event (set its flag to false).

All waiters will now block.

is_set()[source]¶: Check if the event is set

set()[source]¶

Set the event (set its flag to false).

All waiters will now be released.

wait(timeout=None)[source]¶

Wait until the event is set.

Parameters

timeoutnumber or string or timedelta, optional: Seconds to wait on the event in the scheduler. This does not include local coroutine time, network transfer time, etc.. Instead of number of seconds, it is also possible to specify a timedelta in string format, e.g. “200ms”.

Returns

True if the event was set of false, if a timeout happened

Examples

>>> event = Event('a')  
>>> event.wait(timeout="1s")  

class distributed.Semaphore(max_leases=1, name=None, scheduler_rpc=None, loop=None)[source]¶

This semaphore will track leases on the scheduler which can be acquired and released by an instance of this class. If the maximum amount of leases are already acquired, it is not possible to acquire more and the caller waits until another lease has been released.

The lifetime or leases are controlled using a timeout. This timeout is refreshed in regular intervals by the Client of this instance and provides protection from deadlocks or resource starvation in case of worker failure. The timeout can be controlled using the configuration option distributed.scheduler.locks.lease-timeout and the interval in which the scheduler verifies the timeout is set using the option distributed.scheduler.locks.lease-validation-interval.

A noticeable difference to the Semaphore of the python standard library is that this implementation does not allow to release more often than it was acquired. If this happens, a warning is emitted but the internal state is not modified.

Warning

This implementation is susceptible to lease overbooking in case of lease timeouts. It is advised to monitor log information and adjust above configuration options to suitable values for the user application.

Parameters

max_leases: int (optional): The maximum amount of leases that may be granted at the same time. This effectively sets an upper limit to the amount of parallel access to a specific resource. Defaults to 1.
name: string (optional): Name of the semaphore to acquire. Choosing the same name allows two disconnected processes to coordinate. If not given, a random name will be generated.
register: bool: If True, register the semaphore with the scheduler. This needs to be done before any leases can be acquired. If not done during initialization, this can also be done by calling the register method of this class. When registering, this needs to be awaited.
scheduler_rpc: ConnectionPool: The ConnectionPool to connect to the scheduler. If None is provided, it uses the worker or client pool. This parameter is mostly used for testing.
loop: IOLoop: The event loop this instance is using. If None is provided, reuse the loop of the active worker or client.

Notes

If a client attempts to release the semaphore but doesn’t have a lease acquired, this will raise an exception.

dask executes functions by default assuming they are pure, when using semaphore acquire/releases inside such a function, it must be noted that there are in fact side-effects, thus, the function can no longer be considered pure. If this is not taken into account, this may lead to unexpected behavior.

Examples

>>> from distributed import Semaphore
... sem = Semaphore(max_leases=2, name='my_database')
...
... def access_resource(s, sem):
...     # This automatically acquires a lease from the semaphore (if available) which will be
...     # released when leaving the context manager.
...     with sem:
...         pass
...
... futures = client.map(access_resource, range(10), sem=sem)
... client.gather(futures)
... # Once done, close the semaphore to clean up the state on scheduler side.
... sem.close()

acquire(timeout=None)[source]¶

Acquire a semaphore.

If the internal counter is greater than zero, decrement it by one and return True immediately. If it is zero, wait until a release() is called and return True.

Parameters

timeoutnumber or string or timedelta, optional: Seconds to wait on acquiring the semaphore. This does not include local coroutine time, network transfer time, etc.. Instead of number of seconds, it is also possible to specify a timedelta in string format, e.g. “200ms”.

get_value()[source]¶: Return the number of currently registered leases.

release()[source]¶

Release the semaphore.

Returns

bool: This value indicates whether a lease was released immediately or not. Note that a user should not retry this operation. Under certain circumstances (e.g. scheduler overload) the lease may not be released immediately, but it will always be automatically released after a specific interval configured using “distributed.scheduler.locks.lease-validation-interval” and “distributed.scheduler.locks.lease-timeout”.

Best Practices

Machine Learning

Futures

Contents

Futures¶

Examples¶

Start Dask Client¶

Submit Tasks¶

Move Data¶

References, Cancellation, and Exceptions¶

Waiting on Futures¶

Fire and Forget¶

Submit task and retrieve results from a different process¶

Submit Tasks from Tasks¶

Coordination Primitives¶

Queues¶

Global Variables¶

Locks¶

Events¶

Semaphore¶

Publish-Subscribe¶

Actors¶

Example: Counter¶

Example: Parameter Server¶

Asynchronous Operation¶

API¶