Scheduler Plugins

class distributed.diagnostics.plugin.SchedulerPlugin[source]

Interface to extend the Scheduler

The scheduler operates by triggering and responding to events like task_finished, update_graph, task_erred, etc..

A plugin enables custom code to run at each of those same events. The scheduler will run the analogous methods on this class when each event is triggered. This runs user code within the scheduler thread that can perform arbitrary operations in synchrony with the scheduler itself.

Plugins are often used for diagnostics and measurement, but have full access to the scheduler and could in principle affect core scheduling.

To implement a plugin implement some of the methods of this class and add the plugin to the scheduler with Scheduler.add_plugin(myplugin).


>>> class Counter(SchedulerPlugin):
...     def __init__(self):
...         self.counter = 0
...     def transition(self, key, start, finish, *args, **kwargs):
...         if start == 'processing' and finish == 'memory':
...             self.counter += 1
...     def restart(self, scheduler):
...         self.counter = 0
>>> c = Counter()
>>> scheduler.add_plugin(c)  # doctest: +SKIP
add_worker(self, scheduler=None, worker=None, **kwargs)[source]

Run when a new worker enters the cluster

remove_worker(self, scheduler=None, worker=None, **kwargs)[source]

Run when a worker leaves the cluster

restart(self, scheduler, **kwargs)[source]

Run when the scheduler restarts itself

transition(self, key, start, finish, *args, **kwargs)[source]

Run whenever a task changes state

key: string
start: string

Start state of the transition. One of released, waiting, processing, memory, error.

finish: string

Final state of the transition.

*args, **kwargs: More options passed when transitioning

This may include worker ID, compute time, etc.

update_graph(self, scheduler, dsk=None, keys=None, restrictions=None, **kwargs)[source]

Run when a new graph / tasks enter the scheduler

RabbitMQ Example

RabbitMQ is a distributed messaging queue that we can use to post updates about task transitions. By posting transitions to RabbitMQ, we allow other machines to do the processing of transitions and keep scheduler processing to a minimum. See the RabbitMQ tutorial for more information on RabbitMQ and how to consume the messages.

import json
from distributed.diagnostics.plugin import SchedulerPlugin
import pika

class RabbitMQPlugin(SchedulerPlugin):
    def __init__(self):
        # Update host to be your RabbitMQ host
        self.connection = pika.BlockingConnection(
            pika.ConnectionParameters(host='localhost')) ='dask_task_status', durable=True)

    def transition(self, key, start, finish, *args, **kwargs):
        message = dict(
                delivery_mode=2,  # make message persistent

def dask_setup(scheduler):
    plugin = RabbitMQPlugin()

Run with: dask-scheduler --preload <>

Accessing Full Task State

If you would like to access the full distributed.scheduler.TaskState stored in the scheduler you can do this by passing and storing a reference to the scheduler as so:

from distributed.diagnostics.plugin import SchedulerPlugin

class MyPlugin(SchedulerPlugin):
    def __init__(self, scheduler):
         self.scheduler = scheduler

    def transition(self, key, start, finish, *args, **kwargs):
         # Get full TaskState
         ts = self.scheduler.tasks[key]

def dask_setup(scheduler):
    plugin = MyPlugin(scheduler)