openstack/deb-python-kafka
Code Issues Proposed changes

Merge branch 'teach_producers_about_compression' into producer_compression

Browse Source 
Conflicts:
	servers/0.8.0/kafka-src
	test/test_unit.py
This commit is contained in:
Mark Roberts
2014-05-07 03:05:50 -07:00
parent b47bf781eb 671b74ab2e
commit 39796ec491
3 changed files with 49 additions and 25 deletions
Download Patch File Download Diff File Expand all files Collapse all files

42
kafka/producer.py
View File

@@ -11,7 +11,10 @@ from multiprocessing import Queue, Process
from kafka.common import ProduceRequest, TopicAndPartition
from kafka.partitioner import HashedPartitioner
from kafka.protocol import create_message
from kafka.protocol import (
CODEC_NONE, CODEC_GZIP, CODEC_SNAPPY, ALL_CODECS,
create_message, create_gzip_message, create_snappy_message,
)
log = logging.getLogger("kafka")
@@ -21,7 +24,7 @@ BATCH_SEND_MSG_COUNT = 20
STOP_ASYNC_PRODUCER = -1
def _send_upstream(queue, client, batch_time, batch_size,
def _send_upstream(queue, client, codec, batch_time, batch_size,
req_acks, ack_timeout):
"""
Listen on the queue for a specified number of messages or till
@@ -62,7 +65,14 @@ def _send_upstream(queue, client, batch_time, batch_size,
# Send collected requests upstream
reqs = []
for topic_partition, messages in msgset.items():
for topic_partition, msg in msgset.items():
if codec == CODEC_GZIP:
messages = [create_gzip_message(msg)]
elif codec == CODEC_SNAPPY:
messages = [create_snappy_message(msg)]
else:
messages = [create_message(m) for m in msg]
req = ProduceRequest(topic_partition.topic,
topic_partition.partition,
messages)
@@ -102,6 +112,7 @@ class Producer(object):
def __init__(self, client, async=False,
req_acks=ACK_AFTER_LOCAL_WRITE,
ack_timeout=DEFAULT_ACK_TIMEOUT,
codec=None,
batch_send=False,
batch_send_every_n=BATCH_SEND_MSG_COUNT,
batch_send_every_t=BATCH_SEND_DEFAULT_INTERVAL):
@@ -119,11 +130,17 @@ class Producer(object):
self.req_acks = req_acks
self.ack_timeout = ack_timeout
if codec is None:
codec = CODEC_NONE
assert codec in ALL_CODECS
self.codec = codec
if self.async:
self.queue = Queue() # Messages are sent through this queue
self.proc = Process(target=_send_upstream,
args=(self.queue,
self.client.copy(),
self.codec,
batch_send_every_t,
batch_send_every_n,
self.req_acks,
@@ -139,11 +156,16 @@ class Producer(object):
"""
if self.async:
for m in msg:
self.queue.put((TopicAndPartition(topic, partition),
create_message(m)))
self.queue.put((TopicAndPartition(topic, partition), m))
resp = []
else:
messages = [create_message(m) for m in msg]
if self.codec == CODEC_GZIP:
messages = [create_gzip_message(msg)]
elif self.codec == CODEC_SNAPPY:
messages = [create_snappy_message(msg)]
else:
messages = [create_message(m) for m in msg]
req = ProduceRequest(topic, partition, messages)
try:
resp = self.client.send_produce_request([req], acks=self.req_acks,
@@ -168,7 +190,7 @@ class Producer(object):
class SimpleProducer(Producer):
"""
A simple, round-robbin producer. Each message goes to exactly one partition
A simple, round-robin producer. Each message goes to exactly one partition
Params:
client - The Kafka client instance to use
@@ -189,6 +211,7 @@ class SimpleProducer(Producer):
def __init__(self, client, async=False,
req_acks=Producer.ACK_AFTER_LOCAL_WRITE,
ack_timeout=Producer.DEFAULT_ACK_TIMEOUT,
codec=None,
batch_send=False,
batch_send_every_n=BATCH_SEND_MSG_COUNT,
batch_send_every_t=BATCH_SEND_DEFAULT_INTERVAL,
@@ -196,7 +219,7 @@ class SimpleProducer(Producer):
self.partition_cycles = {}
self.random_start = random_start
super(SimpleProducer, self).__init__(client, async, req_acks,
ack_timeout, batch_send,
ack_timeout, codec, batch_send,
batch_send_every_n,
batch_send_every_t)
@@ -241,6 +264,7 @@ class KeyedProducer(Producer):
def __init__(self, client, partitioner=None, async=False,
req_acks=Producer.ACK_AFTER_LOCAL_WRITE,
ack_timeout=Producer.DEFAULT_ACK_TIMEOUT,
codec=None,
batch_send=False,
batch_send_every_n=BATCH_SEND_MSG_COUNT,
batch_send_every_t=BATCH_SEND_DEFAULT_INTERVAL):
@@ -250,7 +274,7 @@ class KeyedProducer(Producer):
self.partitioners = {}
super(KeyedProducer, self).__init__(client, async, req_acks,
ack_timeout, batch_send,
ack_timeout, codec, batch_send,
batch_send_every_n,
batch_send_every_t)

23
kafka/protocol.py
View File

@@ -18,6 +18,12 @@ from kafka.util import (
log = logging.getLogger("kafka")
ATTRIBUTE_CODEC_MASK = 0x03
CODEC_NONE = 0x00
CODEC_GZIP = 0x01
CODEC_SNAPPY = 0x02
ALL_CODECS = (CODEC_NONE, CODEC_GZIP, CODEC_SNAPPY)
class KafkaProtocol(object):
"""
@@ -32,11 +38,6 @@ class KafkaProtocol(object):
OFFSET_COMMIT_KEY = 8
OFFSET_FETCH_KEY = 9
ATTRIBUTE_CODEC_MASK = 0x03
CODEC_NONE = 0x00
CODEC_GZIP = 0x01
CODEC_SNAPPY = 0x02
###################
# Private API #
###################
@@ -150,17 +151,17 @@ class KafkaProtocol(object):
(key, cur) = read_int_string(data, cur)
(value, cur) = read_int_string(data, cur)
codec = att & KafkaProtocol.ATTRIBUTE_CODEC_MASK
codec = att & ATTRIBUTE_CODEC_MASK
if codec == KafkaProtocol.CODEC_NONE:
if codec == CODEC_NONE:
yield (offset, Message(magic, att, key, value))
elif codec == KafkaProtocol.CODEC_GZIP:
elif codec == CODEC_GZIP:
gz = gzip_decode(value)
for (offset, msg) in KafkaProtocol._decode_message_set_iter(gz):
yield (offset, msg)
elif codec == KafkaProtocol.CODEC_SNAPPY:
elif codec == CODEC_SNAPPY:
snp = snappy_decode(value)
for (offset, msg) in KafkaProtocol._decode_message_set_iter(snp):
yield (offset, msg)
@@ -543,7 +544,7 @@ def create_gzip_message(payloads, key=None):
[create_message(payload) for payload in payloads])
gzipped = gzip_encode(message_set)
codec = KafkaProtocol.ATTRIBUTE_CODEC_MASK & KafkaProtocol.CODEC_GZIP
codec = ATTRIBUTE_CODEC_MASK & CODEC_GZIP
return Message(0, 0x00 | codec, key, gzipped)
@@ -564,6 +565,6 @@ def create_snappy_message(payloads, key=None):
[create_message(payload) for payload in payloads])
snapped = snappy_encode(message_set)
codec = KafkaProtocol.ATTRIBUTE_CODEC_MASK & KafkaProtocol.CODEC_SNAPPY
codec = ATTRIBUTE_CODEC_MASK & CODEC_SNAPPY
return Message(0, 0x00 | codec, key, snapped)

9
test/test_protocol.py
View File

@@ -16,7 +16,8 @@ from kafka.codec import (
snappy_encode, snappy_decode
)
from kafka.protocol import (
create_gzip_message, create_message, create_snappy_message, KafkaProtocol
create_gzip_message, create_message, create_snappy_message, KafkaProtocol,
ATTRIBUTE_CODEC_MASK, CODEC_GZIP, CODEC_SNAPPY
)
class TestProtocol(unittest2.TestCase):
@@ -33,8 +34,7 @@ class TestProtocol(unittest2.TestCase):
payloads = ["v1", "v2"]
msg = create_gzip_message(payloads)
self.assertEqual(msg.magic, 0)
self.assertEqual(msg.attributes, KafkaProtocol.ATTRIBUTE_CODEC_MASK &
KafkaProtocol.CODEC_GZIP)
self.assertEqual(msg.attributes, ATTRIBUTE_CODEC_MASK & CODEC_GZIP)
self.assertEqual(msg.key, None)
# Need to decode to check since gzipped payload is non-deterministic
decoded = gzip_decode(msg.value)
@@ -63,8 +63,7 @@ class TestProtocol(unittest2.TestCase):
payloads = ["v1", "v2"]
msg = create_snappy_message(payloads)
self.assertEqual(msg.magic, 0)
self.assertEqual(msg.attributes, KafkaProtocol.ATTRIBUTE_CODEC_MASK &
KafkaProtocol.CODEC_SNAPPY)
self.assertEqual(msg.attributes, ATTRIBUTE_CODEC_MASK & CODEC_SNAPPY)
self.assertEqual(msg.key, None)
decoded = snappy_decode(msg.value)
expect = "".join([