lbry-sdk/lbry/blob_exchange/downloader.py

142 lines
6.6 KiB
Python
Raw Normal View History

2019-01-22 18:52:32 +01:00
import asyncio
import typing
import logging
2019-06-21 02:55:47 +02:00
from lbry.utils import cache_concurrent
from lbry.blob_exchange.client import request_blob
from lbry.dht.node import get_kademlia_peers_from_hosts
2019-01-22 18:52:32 +01:00
if typing.TYPE_CHECKING:
2019-06-21 02:55:47 +02:00
from lbry.conf import Config
from lbry.dht.node import Node
from lbry.dht.peer import KademliaPeer
from lbry.blob.blob_manager import BlobManager
from lbry.blob.blob_file import AbstractBlob
from lbry.blob_exchange.client import BlobExchangeClientProtocol
2019-01-22 18:52:32 +01:00
log = logging.getLogger(__name__)
class BlobDownloader:
BAN_FACTOR = 2.0 # fixme: when connection manager gets implemented, move it out from here
2019-03-28 19:51:55 +01:00
2019-08-02 19:14:41 +02:00
def __init__(self, loop: asyncio.AbstractEventLoop, config: 'Config', blob_manager: 'BlobManager',
peer_queue: asyncio.Queue):
2019-01-22 18:52:32 +01:00
self.loop = loop
self.config = config
2019-01-22 18:52:32 +01:00
self.blob_manager = blob_manager
self.peer_queue = peer_queue
2019-01-31 19:05:00 +01:00
self.active_connections: typing.Dict['KademliaPeer', asyncio.Task] = {} # active request_blob calls
self.ignored: typing.Dict['KademliaPeer', int] = {}
2019-01-31 18:28:59 +01:00
self.scores: typing.Dict['KademliaPeer', int] = {}
self.failures: typing.Dict['KademliaPeer', int] = {}
self.connection_failures: typing.Set['KademliaPeer'] = set()
self.connections: typing.Dict['KademliaPeer', 'BlobExchangeClientProtocol'] = {}
2019-05-06 21:04:38 +02:00
self.is_running = asyncio.Event(loop=self.loop)
2019-01-22 18:52:32 +01:00
def should_race_continue(self, blob: 'AbstractBlob'):
2019-07-25 03:32:20 +02:00
max_probes = self.config.max_connections_per_download * (1 if self.connections else 10)
if len(self.active_connections) >= max_probes:
2019-02-08 05:28:03 +01:00
return False
return not (blob.get_is_verified() or not blob.is_writeable())
2019-02-08 05:28:03 +01:00
2019-07-25 00:59:12 +02:00
async def request_blob_from_peer(self, blob: 'AbstractBlob', peer: 'KademliaPeer', connection_id: int = 0,
just_probe: bool = False):
2019-02-04 22:03:08 +01:00
if blob.get_is_verified():
return
2019-02-08 07:05:53 +01:00
start = self.loop.time()
bytes_received, protocol = await request_blob(
2019-07-25 00:59:12 +02:00
self.loop, blob if not just_probe else None, peer.address, peer.tcp_port, self.config.peer_connect_timeout,
self.config.blob_download_timeout, connected_protocol=self.connections.get(peer),
connection_id=connection_id, connection_manager=self.blob_manager.connection_manager
2019-02-04 22:03:08 +01:00
)
if not bytes_received and not protocol and peer not in self.connection_failures:
self.connection_failures.add(peer)
if not protocol and peer not in self.ignored:
self.ignored[peer] = self.loop.time()
2019-02-04 22:03:08 +01:00
log.debug("drop peer %s:%i", peer.address, peer.tcp_port)
self.failures[peer] = self.failures.get(peer, 0) + 1
if peer in self.connections:
del self.connections[peer]
elif protocol:
2019-02-04 22:03:08 +01:00
log.debug("keep peer %s:%i", peer.address, peer.tcp_port)
self.failures[peer] = 0
self.connections[peer] = protocol
elapsed = self.loop.time() - start
self.scores[peer] = bytes_received / elapsed if bytes_received and elapsed else 1
2019-01-22 18:52:32 +01:00
async def new_peer_or_finished(self):
active_tasks = list(self.active_connections.values()) + [asyncio.sleep(1)]
await asyncio.wait(active_tasks, loop=self.loop, return_when='FIRST_COMPLETED')
2019-01-22 18:52:32 +01:00
2019-02-08 00:11:28 +01:00
def cleanup_active(self):
if not self.active_connections and not self.connections:
self.clearbanned()
2019-02-08 00:11:28 +01:00
to_remove = [peer for (peer, task) in self.active_connections.items() if task.done()]
for peer in to_remove:
del self.active_connections[peer]
def clearbanned(self):
now = self.loop.time()
self.ignored = {
peer: when for (peer, when) in self.ignored.items()
2019-05-01 23:04:45 +02:00
if (now - when) < min(30.0, (self.failures.get(peer, 0) ** self.BAN_FACTOR))
}
2019-03-31 03:05:46 +02:00
@cache_concurrent
2019-05-06 02:22:10 +02:00
async def download_blob(self, blob_hash: str, length: typing.Optional[int] = None,
connection_id: int = 0) -> 'AbstractBlob':
blob = self.blob_manager.get_blob(blob_hash, length)
if blob.get_is_verified():
return blob
2019-05-06 21:04:38 +02:00
self.is_running.set()
2019-01-22 18:52:32 +01:00
try:
2019-05-06 21:04:38 +02:00
while not blob.get_is_verified() and self.is_running.is_set():
batch: typing.Set['KademliaPeer'] = set(self.connections.keys())
while not self.peer_queue.empty():
2019-05-01 23:04:45 +02:00
batch.update(self.peer_queue.get_nowait())
2019-02-08 00:11:28 +01:00
log.debug(
"%s running, %d peers, %d ignored, %d active, %s connections", blob_hash[:6],
2019-07-25 03:32:20 +02:00
len(batch), len(self.ignored), len(self.active_connections), len(self.connections)
2019-02-08 00:11:28 +01:00
)
2019-05-01 23:04:45 +02:00
for peer in sorted(batch, key=lambda peer: self.scores.get(peer, 0), reverse=True):
if peer in self.ignored:
continue
if peer in self.active_connections or not self.should_race_continue(blob):
2019-07-25 03:32:20 +02:00
continue
log.debug("request %s from %s:%i", blob_hash[:8], peer.address, peer.tcp_port)
t = self.loop.create_task(self.request_blob_from_peer(blob, peer, connection_id))
2019-07-25 03:32:20 +02:00
self.active_connections[peer] = t
self.peer_queue.put_nowait(list(batch))
await self.new_peer_or_finished()
self.cleanup_active()
2019-02-04 22:09:50 +01:00
log.debug("downloaded %s", blob_hash[:8])
return blob
finally:
blob.close()
if self.loop.is_running():
self.loop.call_soon(self.cleanup_active)
def close(self):
self.connection_failures.clear()
self.scores.clear()
self.ignored.clear()
2019-05-06 21:04:38 +02:00
self.is_running.clear()
for protocol in self.connections.values():
protocol.close()
async def download_blob(loop, config: 'Config', blob_manager: 'BlobManager', dht_node: 'Node',
blob_hash: str) -> 'AbstractBlob':
2019-01-31 18:28:59 +01:00
search_queue = asyncio.Queue(loop=loop, maxsize=config.max_connections_per_download)
search_queue.put_nowait(blob_hash)
peer_queue, accumulate_task = dht_node.accumulate_peers(search_queue)
fixed_peers = None if not config.fixed_peers else await get_kademlia_peers_from_hosts(config.fixed_peers)
if fixed_peers:
loop.call_later(config.fixed_peer_delay, peer_queue.put_nowait, fixed_peers)
downloader = BlobDownloader(loop, config, blob_manager, peer_queue)
try:
return await downloader.download_blob(blob_hash)
finally:
if accumulate_task and not accumulate_task.done():
accumulate_task.cancel()
downloader.close()