/usr/lib/python2.7/dist-packages/kafka/client.py is in python-kafka 0.9.3-2.
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 | import binascii
import collections
import copy
import functools
import itertools
import logging
import time
import kafka.common
from kafka.common import (TopicAndPartition, BrokerMetadata,
ConnectionError, FailedPayloadsError,
KafkaTimeoutError, KafkaUnavailableError,
LeaderNotAvailableError, UnknownTopicOrPartitionError,
NotLeaderForPartitionError, ReplicaNotAvailableError)
from kafka.conn import collect_hosts, KafkaConnection, DEFAULT_SOCKET_TIMEOUT_SECONDS
from kafka.protocol import KafkaProtocol
log = logging.getLogger("kafka")
class KafkaClient(object):
CLIENT_ID = b"kafka-python"
ID_GEN = itertools.count()
# 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):
# We need one connection to bootstrap
self.client_id = client_id
self.timeout = timeout
self.hosts = collect_hosts(hosts)
# create connections only when we need them
self.conns = {}
self.brokers = {} # broker_id -> BrokerMetadata
self.topics_to_brokers = {} # TopicAndPartition -> BrokerMetadata
self.topic_partitions = {} # topic -> partition -> PartitionMetadata
self.load_metadata_for_topics() # bootstrap with all metadata
##################
# Private API #
##################
def _get_conn(self, host, port):
"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] = KafkaConnection(
host,
port,
timeout=self.timeout
)
return self.conns[host_key]
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 = TopicAndPartition(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[topic]:
raise UnknownTopicOrPartitionError(key)
# If there's no leader for the partition, raise
meta = self.topic_partitions[topic][partition]
if meta.leader == -1:
raise LeaderNotAvailableError(meta)
# Otherwise return the BrokerMetadata
return self.brokers[meta.leader]
def _next_id(self):
"""
Generate a new correlation id
"""
return next(KafkaClient.ID_GEN)
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.
"""
for (host, port) in self.hosts:
requestId = self._next_id()
try:
conn = self._get_conn(host, port)
request = encoder_fn(client_id=self.client_id,
correlation_id=requestId,
payloads=payloads)
conn.send(requestId, request)
response = conn.recv(requestId)
return decoder_fn(response)
except Exception:
log.exception("Could not send request [%r] to server %s:%i, "
"trying next server" % (requestId, host, port))
raise KafkaUnavailableError("All servers failed to process request")
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
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
"""
# Group the requests by topic+partition
original_keys = []
payloads_by_broker = collections.defaultdict(list)
for payload in payloads:
leader = self._get_leader_for_partition(payload.topic,
payload.partition)
payloads_by_broker[leader].append(payload)
original_keys.append((payload.topic, payload.partition))
# Accumulate the responses in a dictionary
acc = {}
# keep a list of payloads that were failed to be sent to brokers
failed_payloads = []
# For each broker, send the list of request payloads
for broker, payloads in payloads_by_broker.items():
conn = self._get_conn(broker.host.decode('utf-8'), broker.port)
requestId = self._next_id()
request = encoder_fn(client_id=self.client_id,
correlation_id=requestId, payloads=payloads)
failed = False
# Send the request, recv the response
try:
conn.send(requestId, request)
if decoder_fn is None:
continue
try:
response = conn.recv(requestId)
except ConnectionError as e:
log.warning("Could not receive response to request [%s] "
"from server %s: %s", binascii.b2a_hex(request), conn, e)
failed = True
except ConnectionError as e:
log.warning("Could not send request [%s] to server %s: %s",
binascii.b2a_hex(request), conn, e)
failed = True
if failed:
failed_payloads += payloads
self.reset_all_metadata()
continue
for response in decoder_fn(response):
acc[(response.topic, response.partition)] = response
if failed_payloads:
raise FailedPayloadsError(failed_payloads)
# Order the accumulated responses by the original key order
return (acc[k] for k in original_keys) if acc else ()
def __repr__(self):
return '<KafkaClient client_id=%s>' % (self.client_id)
def _raise_on_response_error(self, resp):
try:
kafka.common.check_error(resp)
except (UnknownTopicOrPartitionError, NotLeaderForPartitionError):
self.reset_topic_metadata(resp.topic)
raise
#################
# Public API #
#################
def close(self):
for conn in self.conns.values():
conn.close()
def copy(self):
"""
Create an inactive copy of the client object
A reinit() has to be done on the copy before it can be used again
"""
c = copy.deepcopy(self)
for key in c.conns:
c.conns[key] = self.conns[key].copy()
return c
def reinit(self):
for conn in self.conns.values():
conn.reinit()
def reset_topic_metadata(self, *topics):
for topic in topics:
try:
partitions = self.topic_partitions[topic]
except KeyError:
continue
for partition in partitions:
self.topics_to_brokers.pop(TopicAndPartition(topic, partition), None)
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 None
return list(self.topic_partitions[topic])
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))
try:
self.load_metadata_for_topics(topic)
except LeaderNotAvailableError:
pass
except UnknownTopicOrPartitionError:
# Server is not configured to auto-create
# retrying in this case will not help
raise
time.sleep(.5)
def load_metadata_for_topics(self, *topics):
"""
Fetch broker and topic-partition metadata from the server,
and update internal data:
broker list, topic/partition list, and topic/parition -> broker map
This method should be called after receiving any error
Arguments:
*topics (optional): If a list of topics is provided,
the metadata refresh will be limited to the specified topics only.
Exceptions:
----------
If the broker is configured to not auto-create topics,
expect UnknownTopicOrPartitionError for topics that don't exist
If the broker is configured to auto-create topics,
expect LeaderNotAvailableError for new topics
until partitions have been initialized.
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)
"""
resp = self.send_metadata_request(topics)
log.debug("Broker metadata: %s", resp.brokers)
log.debug("Topic metadata: %s", resp.topics)
self.brokers = dict([(broker.nodeId, broker)
for broker in resp.brokers])
for topic_metadata in resp.topics:
topic = topic_metadata.topic
partitions = topic_metadata.partitions
self.reset_topic_metadata(topic)
# Errors expected for new topics
try:
kafka.common.check_error(topic_metadata)
except (UnknownTopicOrPartitionError, LeaderNotAvailableError) as e:
# Raise if the topic was passed in explicitly
if topic in topics:
raise
# Otherwise, just log a warning
log.error("Error loading topic metadata for %s: %s", topic, type(e))
continue
self.topic_partitions[topic] = {}
for partition_metadata in partitions:
partition = partition_metadata.partition
leader = partition_metadata.leader
self.topic_partitions[topic][partition] = partition_metadata
# Populate topics_to_brokers dict
topic_part = TopicAndPartition(topic, partition)
# Check for partition errors
try:
kafka.common.check_error(partition_metadata)
# If No Leader, topics_to_brokers topic_partition -> None
except 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
except ReplicaNotAvailableError:
log.warning('Some (non-leader) replicas not available for topic %s partition %d', topic, partition)
# 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 dont 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
)
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_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
fail_on_error: boolean, should we raise an Exception if we
encounter an API error?
callback: function, instead of returning the ProduceResponse,
first pass it through this function
Returns:
list of ProduceResponse or callback(ProduceResponse), 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)
out = []
for resp in resps:
if fail_on_error is True:
self._raise_on_response_error(resp)
if callback is not None:
out.append(callback(resp))
else:
out.append(resp)
return out
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)
out = []
for resp in resps:
if fail_on_error is True:
self._raise_on_response_error(resp)
if callback is not None:
out.append(callback(resp))
else:
out.append(resp)
return out
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)
out = []
for resp in resps:
if fail_on_error is True:
self._raise_on_response_error(resp)
if callback is not None:
out.append(callback(resp))
else:
out.append(resp)
return out
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)
out = []
for resp in resps:
if fail_on_error is True:
self._raise_on_response_error(resp)
if callback is not None:
out.append(callback(resp))
else:
out.append(resp)
return out
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)
out = []
for resp in resps:
if fail_on_error is True:
self._raise_on_response_error(resp)
if callback is not None:
out.append(callback(resp))
else:
out.append(resp)
return out
|