from __future__ import annotations import asyncio import logging import uuid from collections import defaultdict from dask.utils import parse_timedelta from distributed.client import Future from distributed.utils import wait_for from distributed.worker import get_client logger = logging.getLogger(__name__) class QueueExtension: """An extension for the scheduler to manage queues This adds the following routes to the scheduler * queue_create * queue_release * queue_put * queue_get * queue_size """ def __init__(self, scheduler): self.scheduler = scheduler self.queues = dict() self.client_refcount = dict() self.future_refcount = defaultdict(int) self.scheduler.handlers.update( { "queue_create": self.create, "queue_put": self.put, "queue_get": self.get, "queue_qsize": self.qsize, } ) self.scheduler.stream_handlers.update( {"queue-future-release": self.future_release, "queue_release": self.release} ) def create(self, name=None, client=None, maxsize=0): logger.debug(f"Queue name: {name}") if name not in self.queues: self.queues[name] = asyncio.Queue(maxsize=maxsize) self.client_refcount[name] = 1 else: self.client_refcount[name] += 1 def release(self, name=None, client=None): if name not in self.queues: return self.client_refcount[name] -= 1 if self.client_refcount[name] == 0: del self.client_refcount[name] futures = self.queues[name]._queue del self.queues[name] keys = [d["value"] for d in futures if d["type"] == "Future"] if keys: self.scheduler.client_releases_keys(keys=keys, client="queue-%s" % name) async def put(self, name=None, key=None, data=None, client=None, timeout=None): if key is not None: record = {"type": "Future", "value": key} self.future_refcount[name, key] += 1 self.scheduler.client_desires_keys(keys=[key], client="queue-%s" % name) else: record = {"type": "msgpack", "value": data} await wait_for(self.queues[name].put(record), timeout=timeout) def future_release(self, name=None, key=None, client=None): self.future_refcount[name, key] -= 1 if self.future_refcount[name, key] == 0: self.scheduler.client_releases_keys(keys=[key], client="queue-%s" % name) del self.future_refcount[name, key] async def get(self, name=None, client=None, timeout=None, batch=False): def process(record): """Add task status if known""" if record["type"] == "Future": record = record.copy() key = record["value"] ts = self.scheduler.tasks.get(key) state = ts.state if ts is not None else "lost" record["state"] = state if state == "erred": record["exception"] = ts.exception_blame.exception record["traceback"] = ts.exception_blame.traceback return record if batch: q = self.queues[name] out = [] if batch is True: while not q.empty(): record = await q.get() out.append(record) else: if timeout is not None: msg = ( "Dask queues don't support simultaneous use of " "integer batch sizes and timeouts" ) raise NotImplementedError(msg) for _ in range(batch): record = await q.get() out.append(record) out = [process(o) for o in out] return out else: record = await wait_for(self.queues[name].get(), timeout=timeout) record = process(record) return record def qsize(self, name=None, client=None): return self.queues[name].qsize() class Queue: """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. Examples -------- >>> from dask.distributed import Client, Queue # doctest: +SKIP >>> client = Client() # doctest: +SKIP >>> queue = Queue('x') # doctest: +SKIP >>> future = client.submit(f, x) # doctest: +SKIP >>> queue.put(future) # doctest: +SKIP See Also -------- Variable: shared variable between clients """ def __init__(self, name=None, client=None, maxsize=0): self._client = client self.name = name or "queue-" + uuid.uuid4().hex self.maxsize = maxsize self._maybe_start() def _maybe_start(self): if self.client: if self.client.asynchronous: self._started = asyncio.ensure_future(self._start()) else: self.client.sync(self._start) @property def client(self): if not self._client: try: self._client = get_client() except ValueError: pass return self._client def _verify_running(self): if not self.client: raise RuntimeError( f"{type(self)} object not properly initialized. This can happen" " if the object is being deserialized outside of the context of" " a Client or Worker." ) async def _start(self): await self.client.scheduler.queue_create(name=self.name, maxsize=self.maxsize) return self def __await__(self): self._maybe_start() if hasattr(self, "_started"): return self._started.__await__() else: async def _(): return self return _().__await__() async def _put(self, value, timeout=None): if isinstance(value, Future): await self.client.scheduler.queue_put( key=value.key, timeout=timeout, name=self.name ) else: await self.client.scheduler.queue_put( data=value, timeout=timeout, name=self.name ) def put(self, value, timeout=None, **kwargs): """Put data into the queue Parameters ---------- timeout : number 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". """ self._verify_running() timeout = parse_timedelta(timeout) return self.client.sync(self._put, value, timeout=timeout, **kwargs) def get(self, timeout=None, batch=False, **kwargs): """Get data from the queue Parameters ---------- timeout : number 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". batch : boolean, 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 """ self._verify_running() timeout = parse_timedelta(timeout) return self.client.sync(self._get, timeout=timeout, batch=batch, **kwargs) def qsize(self, **kwargs): """Current number of elements in the queue""" self._verify_running() return self.client.sync(self._qsize, **kwargs) async def _get(self, timeout=None, batch=False): resp = await self.client.scheduler.queue_get( timeout=timeout, name=self.name, batch=batch ) def process(d): if d["type"] == "Future": value = Future(d["value"], self.client, inform=True, state=d["state"]) if d["state"] == "erred": value._state.set_error(d["exception"], d["traceback"]) self.client._send_to_scheduler( {"op": "queue-future-release", "name": self.name, "key": d["value"]} ) else: value = d["value"] return value if batch is False: result = process(resp) else: result = list(map(process, resp)) return result async def _qsize(self): result = await self.client.scheduler.queue_qsize(name=self.name) return result def close(self): self._verify_running() if self.client.status == "running": # TODO: can leave zombie futures self.client._send_to_scheduler({"op": "queue_release", "name": self.name}) def __reduce__(self): return type(self), (self.name, None, self.maxsize)