/usr/lib/python2.7/dist-packages/kafka/client.py is in python-kafka 1.3.2-0ubuntu1.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 | from __future__ import absolute_import
import collections
import copy
import functools
import logging
import random
import time
from kafka.vendor import six
import kafka.errors
from kafka.errors import (UnknownError, ConnectionError, FailedPayloadsError,
KafkaTimeoutError, KafkaUnavailableError,
LeaderNotAvailableError, UnknownTopicOrPartitionError,
NotLeaderForPartitionError, ReplicaNotAvailableError)
from kafka.structs import TopicPartition, BrokerMetadata
from kafka.conn import (
collect_hosts, BrokerConnection,
ConnectionStates, get_ip_port_afi)
from kafka.protocol import KafkaProtocol
# New KafkaClient
# this is not exposed in top-level imports yet,
# due to conflicts with legacy SimpleConsumer / SimpleProducer usage
from kafka.client_async import KafkaClient
log = logging.getLogger(__name__)
# Legacy KafkaClient interface -- will be deprecated soon
class SimpleClient(object):
CLIENT_ID = b'kafka-python'
DEFAULT_SOCKET_TIMEOUT_SECONDS = 120
# NOTE: The timeout given to the client should always be greater than the
# one passed to SimpleConsumer.get_message(), otherwise you can get a
# socket timeout.
def __init__(self, hosts, client_id=CLIENT_ID,
timeout=DEFAULT_SOCKET_TIMEOUT_SECONDS,
correlation_id=0):
# We need one connection to bootstrap
self.client_id = client_id
self.timeout = timeout
self.hosts = collect_hosts(hosts)
self.correlation_id = correlation_id
self._conns = {}
self.brokers = {} # broker_id -> BrokerMetadata
self.topics_to_brokers = {} # TopicPartition -> BrokerMetadata
self.topic_partitions = {} # topic -> partition -> leader
self.load_metadata_for_topics() # bootstrap with all metadata
##################
# Private API #
##################
def _get_conn(self, host, port, afi):
"""Get or create a connection to a broker using host and port"""
host_key = (host, port)
if host_key not in self._conns:
self._conns[host_key] = BrokerConnection(
host, port, afi,
request_timeout_ms=self.timeout * 1000,
client_id=self.client_id
)
conn = self._conns[host_key]
conn.connect()
if conn.connected():
return conn
timeout = time.time() + self.timeout
while time.time() < timeout and conn.connecting():
if conn.connect() is ConnectionStates.CONNECTED:
break
else:
time.sleep(0.05)
else:
conn.close()
raise ConnectionError("%s:%s (%s)" % (host, port, afi))
return conn
def _get_leader_for_partition(self, topic, partition):
"""
Returns the leader for a partition or None if the partition exists
but has no leader.
UnknownTopicOrPartitionError will be raised if the topic or partition
is not part of the metadata.
LeaderNotAvailableError is raised if server has metadata, but there is
no current leader
"""
key = TopicPartition(topic, partition)
# Use cached metadata if it is there
if self.topics_to_brokers.get(key) is not None:
return self.topics_to_brokers[key]
# Otherwise refresh metadata
# If topic does not already exist, this will raise
# UnknownTopicOrPartitionError if not auto-creating
# LeaderNotAvailableError otherwise until partitions are created
self.load_metadata_for_topics(topic)
# If the partition doesn't actually exist, raise
if partition not in self.topic_partitions.get(topic, []):
raise UnknownTopicOrPartitionError(key)
# If there's no leader for the partition, raise
leader = self.topic_partitions[topic][partition]
if leader == -1:
raise LeaderNotAvailableError((topic, partition))
# Otherwise return the BrokerMetadata
return self.brokers[leader]
def _get_coordinator_for_group(self, group):
"""
Returns the coordinator broker for a consumer group.
GroupCoordinatorNotAvailableError will be raised if the coordinator
does not currently exist for the group.
GroupLoadInProgressError is raised if the coordinator is available
but is still loading offsets from the internal topic
"""
resp = self.send_consumer_metadata_request(group)
# If there's a problem with finding the coordinator, raise the
# provided error
kafka.errors.check_error(resp)
# Otherwise return the BrokerMetadata
return BrokerMetadata(resp.nodeId, resp.host, resp.port, None)
def _next_id(self):
"""Generate a new correlation id"""
# modulo to keep w/i int32
self.correlation_id = (self.correlation_id + 1) % 2**31
return self.correlation_id
def _send_broker_unaware_request(self, payloads, encoder_fn, decoder_fn):
"""
Attempt to send a broker-agnostic request to one of the available
brokers. Keep trying until you succeed.
"""
hosts = set()
for broker in self.brokers.values():
host, port, afi = get_ip_port_afi(broker.host)
hosts.add((host, broker.port, afi))
hosts.update(self.hosts)
hosts = list(hosts)
random.shuffle(hosts)
for (host, port, afi) in hosts:
try:
conn = self._get_conn(host, port, afi)
except ConnectionError:
log.warning("Skipping unconnected connection: %s:%s (AFI %s)",
host, port, afi)
continue
request = encoder_fn(payloads=payloads)
future = conn.send(request)
# Block
while not future.is_done:
conn.recv()
if future.failed():
log.error("Request failed: %s", future.exception)
continue
return decoder_fn(future.value)
raise KafkaUnavailableError('All servers failed to process request: %s' % hosts)
def _payloads_by_broker(self, payloads):
payloads_by_broker = collections.defaultdict(list)
for payload in payloads:
try:
leader = self._get_leader_for_partition(payload.topic, payload.partition)
except (KafkaUnavailableError, LeaderNotAvailableError,
UnknownTopicOrPartitionError):
leader = None
payloads_by_broker[leader].append(payload)
return dict(payloads_by_broker)
def _send_broker_aware_request(self, payloads, encoder_fn, decoder_fn):
"""
Group a list of request payloads by topic+partition and send them to
the leader broker for that partition using the supplied encode/decode
functions
Arguments:
payloads: list of object-like entities with a topic (str) and
partition (int) attribute; payloads with duplicate topic-partitions
are not supported.
encode_fn: a method to encode the list of payloads to a request body,
must accept client_id, correlation_id, and payloads as
keyword arguments
decode_fn: a method to decode a response body into response objects.
The response objects must be object-like and have topic
and partition attributes
Returns:
List of response objects in the same order as the supplied payloads
"""
# encoders / decoders do not maintain ordering currently
# so we need to keep this so we can rebuild order before returning
original_ordering = [(p.topic, p.partition) for p in payloads]
# Connection errors generally mean stale metadata
# although sometimes it means incorrect api request
# Unfortunately there is no good way to tell the difference
# so we'll just reset metadata on all errors to be safe
refresh_metadata = False
# For each broker, send the list of request payloads
# and collect the responses and errors
payloads_by_broker = self._payloads_by_broker(payloads)
responses = {}
def failed_payloads(payloads):
for payload in payloads:
topic_partition = (str(payload.topic), payload.partition)
responses[(topic_partition)] = FailedPayloadsError(payload)
# For each BrokerConnection keep the real socket so that we can use
# a select to perform unblocking I/O
connections_by_future = {}
for broker, broker_payloads in six.iteritems(payloads_by_broker):
if broker is None:
failed_payloads(broker_payloads)
continue
host, port, afi = get_ip_port_afi(broker.host)
try:
conn = self._get_conn(host, broker.port, afi)
except ConnectionError:
refresh_metadata = True
failed_payloads(broker_payloads)
continue
request = encoder_fn(payloads=broker_payloads)
# decoder_fn=None signal that the server is expected to not
# send a response. This probably only applies to
# ProduceRequest w/ acks = 0
expect_response = (decoder_fn is not None)
future = conn.send(request, expect_response=expect_response)
if future.failed():
refresh_metadata = True
failed_payloads(broker_payloads)
continue
if not expect_response:
for payload in broker_payloads:
topic_partition = (str(payload.topic), payload.partition)
responses[topic_partition] = None
continue
connections_by_future[future] = (conn, broker)
conn = None
while connections_by_future:
futures = list(connections_by_future.keys())
for future in futures:
if not future.is_done:
conn, _ = connections_by_future[future]
conn.recv()
continue
_, broker = connections_by_future.pop(future)
if future.failed():
refresh_metadata = True
failed_payloads(payloads_by_broker[broker])
else:
for payload_response in decoder_fn(future.value):
topic_partition = (str(payload_response.topic),
payload_response.partition)
responses[topic_partition] = payload_response
if refresh_metadata:
self.reset_all_metadata()
# Return responses in the same order as provided
return [responses[tp] for tp in original_ordering]
def _send_consumer_aware_request(self, group, payloads, encoder_fn, decoder_fn):
"""
Send a list of requests to the consumer coordinator for the group
specified using the supplied encode/decode functions. As the payloads
that use consumer-aware requests do not contain the group (e.g.
OffsetFetchRequest), all payloads must be for a single group.
Arguments:
group: the name of the consumer group (str) the payloads are for
payloads: list of object-like entities with topic (str) and
partition (int) attributes; payloads with duplicate
topic+partition are not supported.
encode_fn: a method to encode the list of payloads to a request body,
must accept client_id, correlation_id, and payloads as
keyword arguments
decode_fn: a method to decode a response body into response objects.
The response objects must be object-like and have topic
and partition attributes
Returns:
List of response objects in the same order as the supplied payloads
"""
# encoders / decoders do not maintain ordering currently
# so we need to keep this so we can rebuild order before returning
original_ordering = [(p.topic, p.partition) for p in payloads]
broker = self._get_coordinator_for_group(group)
# Send the list of request payloads and collect the responses and
# errors
responses = {}
requestId = self._next_id()
log.debug('Request %s to %s: %s', requestId, broker, payloads)
request = encoder_fn(client_id=self.client_id,
correlation_id=requestId, payloads=payloads)
# Send the request, recv the response
try:
host, port, afi = get_ip_port_afi(broker.host)
conn = self._get_conn(host, broker.port, afi)
conn.send(requestId, request)
except ConnectionError as e:
log.warning('ConnectionError attempting to send request %s '
'to server %s: %s', requestId, broker, e)
for payload in payloads:
topic_partition = (payload.topic, payload.partition)
responses[topic_partition] = FailedPayloadsError(payload)
# No exception, try to get response
else:
# decoder_fn=None signal that the server is expected to not
# send a response. This probably only applies to
# ProduceRequest w/ acks = 0
if decoder_fn is None:
log.debug('Request %s does not expect a response '
'(skipping conn.recv)', requestId)
for payload in payloads:
topic_partition = (payload.topic, payload.partition)
responses[topic_partition] = None
return []
try:
response = conn.recv(requestId)
except ConnectionError as e:
log.warning('ConnectionError attempting to receive a '
'response to request %s from server %s: %s',
requestId, broker, e)
for payload in payloads:
topic_partition = (payload.topic, payload.partition)
responses[topic_partition] = FailedPayloadsError(payload)
else:
_resps = []
for payload_response in decoder_fn(response):
topic_partition = (payload_response.topic,
payload_response.partition)
responses[topic_partition] = payload_response
_resps.append(payload_response)
log.debug('Response %s: %s', requestId, _resps)
# Return responses in the same order as provided
return [responses[tp] for tp in original_ordering]
def __repr__(self):
return '<KafkaClient client_id=%s>' % (self.client_id)
def _raise_on_response_error(self, resp):
# Response can be an unraised exception object (FailedPayloadsError)
if isinstance(resp, Exception):
raise resp
# Or a server api error response
try:
kafka.errors.check_error(resp)
except (UnknownTopicOrPartitionError, NotLeaderForPartitionError):
self.reset_topic_metadata(resp.topic)
raise
# Return False if no error to enable list comprehensions
return False
#################
# Public API #
#################
def close(self):
for conn in self._conns.values():
conn.close()
def copy(self):
"""
Create an inactive copy of the client object, suitable for passing
to a separate thread.
Note that the copied connections are not initialized, so reinit() must
be called on the returned copy.
"""
_conns = self._conns
self._conns = {}
c = copy.deepcopy(self)
self._conns = _conns
return c
def reinit(self):
timeout = time.time() + self.timeout
conns = set(self._conns.values())
for conn in conns:
conn.close()
conn.connect()
while time.time() < timeout:
for conn in list(conns):
conn.connect()
if conn.connected():
conns.remove(conn)
if not conns:
break
def reset_topic_metadata(self, *topics):
for topic in topics:
for topic_partition in list(self.topics_to_brokers.keys()):
if topic_partition.topic == topic:
del self.topics_to_brokers[topic_partition]
if topic in self.topic_partitions:
del self.topic_partitions[topic]
def reset_all_metadata(self):
self.topics_to_brokers.clear()
self.topic_partitions.clear()
def has_metadata_for_topic(self, topic):
return (
topic in self.topic_partitions
and len(self.topic_partitions[topic]) > 0
)
def get_partition_ids_for_topic(self, topic):
if topic not in self.topic_partitions:
return []
return sorted(list(self.topic_partitions[topic]))
@property
def topics(self):
return list(self.topic_partitions.keys())
def ensure_topic_exists(self, topic, timeout = 30):
start_time = time.time()
while not self.has_metadata_for_topic(topic):
if time.time() > start_time + timeout:
raise KafkaTimeoutError('Unable to create topic {0}'.format(topic))
self.load_metadata_for_topics(topic, ignore_leadernotavailable=True)
time.sleep(.5)
def load_metadata_for_topics(self, *topics, **kwargs):
"""Fetch broker and topic-partition metadata from the server.
Updates internal data: broker list, topic/partition list, and
topic/parition -> broker map. This method should be called after
receiving any error.
Note: Exceptions *will not* be raised in a full refresh (i.e. no topic
list). In this case, error codes will be logged as errors.
Partition-level errors will also not be raised here (a single partition
w/o a leader, for example).
Arguments:
*topics (optional): If a list of topics is provided,
the metadata refresh will be limited to the specified topics
only.
ignore_leadernotavailable (bool): suppress LeaderNotAvailableError
so that metadata is loaded correctly during auto-create.
Default: False.
Raises:
UnknownTopicOrPartitionError: Raised for topics that do not exist,
unless the broker is configured to auto-create topics.
LeaderNotAvailableError: Raised for topics that do not exist yet,
when the broker is configured to auto-create topics. Retry
after a short backoff (topics/partitions are initializing).
"""
if 'ignore_leadernotavailable' in kwargs:
ignore_leadernotavailable = kwargs['ignore_leadernotavailable']
else:
ignore_leadernotavailable = False
if topics:
self.reset_topic_metadata(*topics)
else:
self.reset_all_metadata()
resp = self.send_metadata_request(topics)
log.debug('Updating broker metadata: %s', resp.brokers)
log.debug('Updating topic metadata: %s', [topic for _, topic, _ in resp.topics])
self.brokers = dict([(nodeId, BrokerMetadata(nodeId, host, port, None))
for nodeId, host, port in resp.brokers])
for error, topic, partitions in resp.topics:
# Errors expected for new topics
if error:
error_type = kafka.errors.kafka_errors.get(error, UnknownError)
if error_type in (UnknownTopicOrPartitionError, LeaderNotAvailableError):
log.error('Error loading topic metadata for %s: %s (%s)',
topic, error_type, error)
if topic not in topics:
continue
elif (error_type is LeaderNotAvailableError and
ignore_leadernotavailable):
continue
raise error_type(topic)
self.topic_partitions[topic] = {}
for error, partition, leader, _, _ in partitions:
self.topic_partitions[topic][partition] = leader
# Populate topics_to_brokers dict
topic_part = TopicPartition(topic, partition)
# Check for partition errors
if error:
error_type = kafka.errors.kafka_errors.get(error, UnknownError)
# If No Leader, topics_to_brokers topic_partition -> None
if error_type is LeaderNotAvailableError:
log.error('No leader for topic %s partition %d', topic, partition)
self.topics_to_brokers[topic_part] = None
continue
# If one of the replicas is unavailable -- ignore
# this error code is provided for admin purposes only
# we never talk to replicas, only the leader
elif error_type is ReplicaNotAvailableError:
log.debug('Some (non-leader) replicas not available for topic %s partition %d', topic, partition)
else:
raise error_type(topic_part)
# If Known Broker, topic_partition -> BrokerMetadata
if leader in self.brokers:
self.topics_to_brokers[topic_part] = self.brokers[leader]
# If Unknown Broker, fake BrokerMetadata so we don't lose the id
# (not sure how this could happen. server could be in bad state)
else:
self.topics_to_brokers[topic_part] = BrokerMetadata(
leader, None, None, None
)
def send_metadata_request(self, payloads=[], fail_on_error=True,
callback=None):
encoder = KafkaProtocol.encode_metadata_request
decoder = KafkaProtocol.decode_metadata_response
return self._send_broker_unaware_request(payloads, encoder, decoder)
def send_consumer_metadata_request(self, payloads=[], fail_on_error=True,
callback=None):
encoder = KafkaProtocol.encode_consumer_metadata_request
decoder = KafkaProtocol.decode_consumer_metadata_response
return self._send_broker_unaware_request(payloads, encoder, decoder)
def send_produce_request(self, payloads=[], acks=1, timeout=1000,
fail_on_error=True, callback=None):
"""
Encode and send some ProduceRequests
ProduceRequests will be grouped by (topic, partition) and then
sent to a specific broker. Output is a list of responses in the
same order as the list of payloads specified
Arguments:
payloads (list of ProduceRequest): produce requests to send to kafka
ProduceRequest payloads must not contain duplicates for any
topic-partition.
acks (int, optional): how many acks the servers should receive from replica
brokers before responding to the request. If it is 0, the server
will not send any response. If it is 1, the server will wait
until the data is written to the local log before sending a
response. If it is -1, the server will wait until the message
is committed by all in-sync replicas before sending a response.
For any value > 1, the server will wait for this number of acks to
occur (but the server will never wait for more acknowledgements than
there are in-sync replicas). defaults to 1.
timeout (int, optional): maximum time in milliseconds the server can
await the receipt of the number of acks, defaults to 1000.
fail_on_error (bool, optional): raise exceptions on connection and
server response errors, defaults to True.
callback (function, optional): instead of returning the ProduceResponse,
first pass it through this function, defaults to None.
Returns:
list of ProduceResponses, or callback results if supplied, in the
order of input payloads
"""
encoder = functools.partial(
KafkaProtocol.encode_produce_request,
acks=acks,
timeout=timeout)
if acks == 0:
decoder = None
else:
decoder = KafkaProtocol.decode_produce_response
resps = self._send_broker_aware_request(payloads, encoder, decoder)
return [resp if not callback else callback(resp) for resp in resps
if resp is not None and
(not fail_on_error or not self._raise_on_response_error(resp))]
def send_fetch_request(self, payloads=[], fail_on_error=True,
callback=None, max_wait_time=100, min_bytes=4096):
"""
Encode and send a FetchRequest
Payloads are grouped by topic and partition so they can be pipelined
to the same brokers.
"""
encoder = functools.partial(KafkaProtocol.encode_fetch_request,
max_wait_time=max_wait_time,
min_bytes=min_bytes)
resps = self._send_broker_aware_request(
payloads, encoder,
KafkaProtocol.decode_fetch_response)
return [resp if not callback else callback(resp) for resp in resps
if not fail_on_error or not self._raise_on_response_error(resp)]
def send_offset_request(self, payloads=[], fail_on_error=True,
callback=None):
resps = self._send_broker_aware_request(
payloads,
KafkaProtocol.encode_offset_request,
KafkaProtocol.decode_offset_response)
return [resp if not callback else callback(resp) for resp in resps
if not fail_on_error or not self._raise_on_response_error(resp)]
def send_offset_commit_request(self, group, payloads=[],
fail_on_error=True, callback=None):
encoder = functools.partial(KafkaProtocol.encode_offset_commit_request,
group=group)
decoder = KafkaProtocol.decode_offset_commit_response
resps = self._send_broker_aware_request(payloads, encoder, decoder)
return [resp if not callback else callback(resp) for resp in resps
if not fail_on_error or not self._raise_on_response_error(resp)]
def send_offset_fetch_request(self, group, payloads=[],
fail_on_error=True, callback=None):
encoder = functools.partial(KafkaProtocol.encode_offset_fetch_request,
group=group)
decoder = KafkaProtocol.decode_offset_fetch_response
resps = self._send_broker_aware_request(payloads, encoder, decoder)
return [resp if not callback else callback(resp) for resp in resps
if not fail_on_error or not self._raise_on_response_error(resp)]
def send_offset_fetch_request_kafka(self, group, payloads=[],
fail_on_error=True, callback=None):
encoder = functools.partial(KafkaProtocol.encode_offset_fetch_request,
group=group, from_kafka=True)
decoder = KafkaProtocol.decode_offset_fetch_response
resps = self._send_consumer_aware_request(group, payloads, encoder, decoder)
return [resp if not callback else callback(resp) for resp in resps
if not fail_on_error or not self._raise_on_response_error(resp)]
|