ruby-kafka 0.1.0.pre.beta4 → 0.1.0.pre.beta5
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/kafka.gemspec +1 -0
- data/lib/kafka/connection.rb +8 -12
- data/lib/kafka/message_buffer.rb +8 -4
- data/lib/kafka/producer.rb +10 -68
- data/lib/kafka/socket_with_timeout.rb +87 -0
- data/lib/kafka/transmission.rb +78 -0
- data/lib/kafka/version.rb +1 -1
- metadata +18 -2
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA1:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: 2ed5c1319300907fea512494cd9f4f07c3b15b70
|
4
|
+
data.tar.gz: e76e37ec50c63fbd8969478a38e6e804059d5a45
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: 878fc9de44046cee3895212d766f44d5011de1dd3fd05f0f745efd00d284dc2cdae6e5d2aed3cb1ae794210da5f4b76a993cbff03c1554339cb5f5fb088ab071
|
7
|
+
data.tar.gz: 243bb41e327d1f9d8a9d06eb0c1b6bc585a63d2607ab0b25fe44af0ff7cb936429949620a9a4e8b12b5f0838db59aebfab7f73a4e100e7f8763f566094592405
|
data/kafka.gemspec
CHANGED
data/lib/kafka/connection.rb
CHANGED
@@ -1,5 +1,5 @@
|
|
1
|
-
require "socket"
|
2
1
|
require "stringio"
|
2
|
+
require "kafka/socket_with_timeout"
|
3
3
|
require "kafka/protocol/request_message"
|
4
4
|
require "kafka/protocol/encoder"
|
5
5
|
require "kafka/protocol/decoder"
|
@@ -38,7 +38,7 @@ module Kafka
|
|
38
38
|
|
39
39
|
@logger.info "Opening connection to #{@host}:#{@port} with client id #{@client_id}..."
|
40
40
|
|
41
|
-
@socket =
|
41
|
+
@socket = SocketWithTimeout.new(@host, @port, timeout: @connect_timeout)
|
42
42
|
|
43
43
|
@encoder = Kafka::Protocol::Encoder.new(@socket)
|
44
44
|
@decoder = Kafka::Protocol::Decoder.new(@socket)
|
@@ -112,14 +112,12 @@ module Kafka
|
|
112
112
|
|
113
113
|
data = Kafka::Protocol::Encoder.encode_with(message)
|
114
114
|
|
115
|
-
unless IO.select(nil, [@socket], nil, @socket_timeout)
|
116
|
-
@logger.error "Timed out while writing request #{@correlation_id}"
|
117
|
-
raise ConnectionError
|
118
|
-
end
|
119
|
-
|
120
115
|
@encoder.write_bytes(data)
|
121
116
|
|
122
117
|
nil
|
118
|
+
rescue Errno::ETIMEDOUT
|
119
|
+
@logger.error "Timed out while writing request #{@correlation_id}"
|
120
|
+
raise ConnectionError
|
123
121
|
end
|
124
122
|
|
125
123
|
# Reads a response from the connection.
|
@@ -131,11 +129,6 @@ module Kafka
|
|
131
129
|
def read_response(response_class)
|
132
130
|
@logger.debug "Waiting for response #{@correlation_id} from #{to_s}"
|
133
131
|
|
134
|
-
unless IO.select([@socket], nil, nil, @socket_timeout)
|
135
|
-
@logger.error "Timed out while waiting for response #{@correlation_id}"
|
136
|
-
raise ConnectionError
|
137
|
-
end
|
138
|
-
|
139
132
|
bytes = @decoder.bytes
|
140
133
|
|
141
134
|
buffer = StringIO.new(bytes)
|
@@ -147,6 +140,9 @@ module Kafka
|
|
147
140
|
@logger.debug "Received response #{correlation_id} from #{to_s}"
|
148
141
|
|
149
142
|
return correlation_id, response
|
143
|
+
rescue Errno::ETIMEDOUT
|
144
|
+
@logger.error "Timed out while waiting for response #{@correlation_id}"
|
145
|
+
raise ConnectionError
|
150
146
|
end
|
151
147
|
end
|
152
148
|
end
|
data/lib/kafka/message_buffer.rb
CHANGED
@@ -4,15 +4,20 @@ module Kafka
|
|
4
4
|
class MessageBuffer
|
5
5
|
include Enumerable
|
6
6
|
|
7
|
+
attr_reader :size
|
8
|
+
|
7
9
|
def initialize
|
8
10
|
@buffer = {}
|
11
|
+
@size = 0
|
9
12
|
end
|
10
13
|
|
11
14
|
def write(message, topic:, partition:)
|
15
|
+
@size += 1
|
12
16
|
buffer_for(topic, partition) << message
|
13
17
|
end
|
14
18
|
|
15
19
|
def concat(messages, topic:, partition:)
|
20
|
+
@size += messages.count
|
16
21
|
buffer_for(topic, partition).concat(messages)
|
17
22
|
end
|
18
23
|
|
@@ -20,10 +25,6 @@ module Kafka
|
|
20
25
|
@buffer
|
21
26
|
end
|
22
27
|
|
23
|
-
def size
|
24
|
-
@buffer.values.inject(0) {|sum, messages| messages.values.flatten.size + sum }
|
25
|
-
end
|
26
|
-
|
27
28
|
def empty?
|
28
29
|
@buffer.empty?
|
29
30
|
end
|
@@ -43,6 +44,8 @@ module Kafka
|
|
43
44
|
#
|
44
45
|
# @return [nil]
|
45
46
|
def clear_messages(topic:, partition:)
|
47
|
+
@size -= @buffer[topic][partition].count
|
48
|
+
|
46
49
|
@buffer[topic].delete(partition)
|
47
50
|
@buffer.delete(topic) if @buffer[topic].empty?
|
48
51
|
end
|
@@ -52,6 +55,7 @@ module Kafka
|
|
52
55
|
# @return [nil]
|
53
56
|
def clear
|
54
57
|
@buffer = {}
|
58
|
+
@size = 0
|
55
59
|
end
|
56
60
|
|
57
61
|
private
|
data/lib/kafka/producer.rb
CHANGED
@@ -1,6 +1,7 @@
|
|
1
1
|
require "kafka/partitioner"
|
2
2
|
require "kafka/message_buffer"
|
3
3
|
require "kafka/protocol/message"
|
4
|
+
require "kafka/transmission"
|
4
5
|
|
5
6
|
module Kafka
|
6
7
|
|
@@ -161,11 +162,19 @@ module Kafka
|
|
161
162
|
def send_messages
|
162
163
|
attempt = 0
|
163
164
|
|
165
|
+
transmission = Transmission.new(
|
166
|
+
broker_pool: @broker_pool,
|
167
|
+
buffer: @buffer,
|
168
|
+
required_acks: @required_acks,
|
169
|
+
ack_timeout: @ack_timeout,
|
170
|
+
logger: @logger,
|
171
|
+
)
|
172
|
+
|
164
173
|
loop do
|
165
174
|
@logger.info "Sending #{@buffer.size} messages"
|
166
175
|
|
167
176
|
attempt += 1
|
168
|
-
|
177
|
+
transmission.send_messages
|
169
178
|
|
170
179
|
if @buffer.empty?
|
171
180
|
@logger.info "Successfully transmitted all messages"
|
@@ -208,72 +217,5 @@ module Kafka
|
|
208
217
|
def shutdown
|
209
218
|
@broker_pool.shutdown
|
210
219
|
end
|
211
|
-
|
212
|
-
private
|
213
|
-
|
214
|
-
def transmit_messages
|
215
|
-
messages_for_broker = {}
|
216
|
-
|
217
|
-
@buffer.each do |topic, partition, messages|
|
218
|
-
broker_id = @broker_pool.get_leader_id(topic, partition)
|
219
|
-
|
220
|
-
@logger.debug "Current leader for #{topic}/#{partition} is node #{broker_id}"
|
221
|
-
|
222
|
-
messages_for_broker[broker_id] ||= MessageBuffer.new
|
223
|
-
messages_for_broker[broker_id].concat(messages, topic: topic, partition: partition)
|
224
|
-
end
|
225
|
-
|
226
|
-
messages_for_broker.each do |broker_id, message_set|
|
227
|
-
begin
|
228
|
-
broker = @broker_pool.get_broker(broker_id)
|
229
|
-
|
230
|
-
response = broker.produce(
|
231
|
-
messages_for_topics: message_set.to_h,
|
232
|
-
required_acks: @required_acks,
|
233
|
-
timeout: @ack_timeout * 1000, # Kafka expects the timeout in milliseconds.
|
234
|
-
)
|
235
|
-
|
236
|
-
handle_response(response) if response
|
237
|
-
rescue ConnectionError => e
|
238
|
-
@logger.error "Could not connect to broker #{broker_id}: #{e}"
|
239
|
-
|
240
|
-
# Mark the broker pool as stale in order to force a cluster metadata refresh.
|
241
|
-
@broker_pool.mark_as_stale!
|
242
|
-
end
|
243
|
-
end
|
244
|
-
end
|
245
|
-
|
246
|
-
def handle_response(response)
|
247
|
-
response.each_partition do |topic_info, partition_info|
|
248
|
-
topic = topic_info.topic
|
249
|
-
partition = partition_info.partition
|
250
|
-
|
251
|
-
begin
|
252
|
-
Protocol.handle_error(partition_info.error_code)
|
253
|
-
rescue Kafka::CorruptMessage
|
254
|
-
@logger.error "Corrupt message when writing to #{topic}/#{partition}"
|
255
|
-
rescue Kafka::UnknownTopicOrPartition
|
256
|
-
@logger.error "Unknown topic or partition #{topic}/#{partition}"
|
257
|
-
rescue Kafka::LeaderNotAvailable
|
258
|
-
@logger.error "Leader currently not available for #{topic}/#{partition}"
|
259
|
-
@broker_pool.mark_as_stale!
|
260
|
-
rescue Kafka::NotLeaderForPartition
|
261
|
-
@logger.error "Broker not currently leader for #{topic}/#{partition}"
|
262
|
-
@broker_pool.mark_as_stale!
|
263
|
-
rescue Kafka::RequestTimedOut
|
264
|
-
@logger.error "Timed out while writing to #{topic}/#{partition}"
|
265
|
-
rescue Kafka::NotEnoughReplicas
|
266
|
-
@logger.error "Not enough in-sync replicas for #{topic}/#{partition}"
|
267
|
-
rescue Kafka::NotEnoughReplicasAfterAppend
|
268
|
-
@logger.error "Messages written, but to fewer in-sync replicas than required for #{topic}/#{partition}"
|
269
|
-
else
|
270
|
-
offset = partition_info.offset
|
271
|
-
@logger.info "Successfully sent messages for #{topic}/#{partition}; new offset is #{offset}"
|
272
|
-
|
273
|
-
# The messages were successfully written; clear them from the buffer.
|
274
|
-
@buffer.clear_messages(topic: topic, partition: partition)
|
275
|
-
end
|
276
|
-
end
|
277
|
-
end
|
278
220
|
end
|
279
221
|
end
|
@@ -0,0 +1,87 @@
|
|
1
|
+
require "socket"
|
2
|
+
|
3
|
+
module Kafka
|
4
|
+
|
5
|
+
# Opens sockets in a non-blocking fashion, ensuring that we're not stalling
|
6
|
+
# for long periods of time.
|
7
|
+
#
|
8
|
+
# It's possible to set timeouts for connecting to the server, for reading data,
|
9
|
+
# and for writing data. Whenever a timeout is exceeded, Errno::ETIMEDOUT is
|
10
|
+
# raised.
|
11
|
+
#
|
12
|
+
class SocketWithTimeout
|
13
|
+
|
14
|
+
# Opens a socket.
|
15
|
+
#
|
16
|
+
# @param host [String]
|
17
|
+
# @param port [Integer]
|
18
|
+
# @param timeout [Integer] the connection timeout, in seconds.
|
19
|
+
# @raise [Errno::ETIMEDOUT] if the timeout is exceeded.
|
20
|
+
def initialize(host, port, timeout: nil)
|
21
|
+
addr = Socket.getaddrinfo(host, nil)
|
22
|
+
sockaddr = Socket.pack_sockaddr_in(port, addr[0][3])
|
23
|
+
|
24
|
+
@socket = Socket.new(Socket.const_get(addr[0][0]), Socket::SOCK_STREAM, 0)
|
25
|
+
@socket.setsockopt(Socket::IPPROTO_TCP, Socket::TCP_NODELAY, 1)
|
26
|
+
|
27
|
+
begin
|
28
|
+
# Initiate the socket connection in the background. If it doesn't fail
|
29
|
+
# immediately it will raise an IO::WaitWritable (Errno::EINPROGRESS)
|
30
|
+
# indicating the connection is in progress.
|
31
|
+
@socket.connect_nonblock(sockaddr)
|
32
|
+
rescue IO::WaitWritable
|
33
|
+
# IO.select will block until the socket is writable or the timeout
|
34
|
+
# is exceeded, whichever comes first.
|
35
|
+
unless IO.select(nil, [@socket], nil, timeout)
|
36
|
+
# IO.select returns nil when the socket is not ready before timeout
|
37
|
+
# seconds have elapsed
|
38
|
+
@socket.close
|
39
|
+
raise Errno::ETIMEDOUT
|
40
|
+
end
|
41
|
+
|
42
|
+
begin
|
43
|
+
# Verify there is now a good connection.
|
44
|
+
@socket.connect_nonblock(sockaddr)
|
45
|
+
rescue Errno::EISCONN
|
46
|
+
# The socket is connected, we're good!
|
47
|
+
end
|
48
|
+
end
|
49
|
+
end
|
50
|
+
|
51
|
+
# Reads bytes from the socket, possible with a timeout.
|
52
|
+
#
|
53
|
+
# @param num_bytes [Integer] the number of bytes to read.
|
54
|
+
# @param timeout [Integer] the number of seconds to wait before timing out.
|
55
|
+
# @raise [Errno::ETIMEDOUT] if the timeout is exceeded.
|
56
|
+
# @return [String] the data that was read from the socket.
|
57
|
+
def read(num_bytes, timeout: nil)
|
58
|
+
unless IO.select([@socket], nil, nil, timeout)
|
59
|
+
raise Errno::ETIMEDOUT
|
60
|
+
end
|
61
|
+
|
62
|
+
@socket.read(num_bytes)
|
63
|
+
end
|
64
|
+
|
65
|
+
# Writes bytes to the socket, possible with a timeout.
|
66
|
+
#
|
67
|
+
# @param bytes [String] the data that should be written to the socket.
|
68
|
+
# @param timeout [Integer] the number of seconds to wait before timing out.
|
69
|
+
# @raise [Errno::ETIMEDOUT] if the timeout is exceeded.
|
70
|
+
# @return [Integer] the number of bytes written.
|
71
|
+
def write(bytes, timeout: nil)
|
72
|
+
unless IO.select(nil, [@socket], nil, timeout)
|
73
|
+
raise Errno::ETIMEDOUT
|
74
|
+
end
|
75
|
+
|
76
|
+
@socket.write(bytes)
|
77
|
+
end
|
78
|
+
|
79
|
+
def close
|
80
|
+
@socket.close
|
81
|
+
end
|
82
|
+
|
83
|
+
def set_encoding(encoding)
|
84
|
+
@socket.set_encoding(encoding)
|
85
|
+
end
|
86
|
+
end
|
87
|
+
end
|
@@ -0,0 +1,78 @@
|
|
1
|
+
module Kafka
|
2
|
+
class Transmission
|
3
|
+
def initialize(broker_pool:, buffer:, required_acks:, ack_timeout:, logger:)
|
4
|
+
@broker_pool = broker_pool
|
5
|
+
@buffer = buffer
|
6
|
+
@required_acks = required_acks
|
7
|
+
@ack_timeout = ack_timeout
|
8
|
+
@logger = logger
|
9
|
+
end
|
10
|
+
|
11
|
+
def send_messages
|
12
|
+
messages_for_broker = {}
|
13
|
+
|
14
|
+
@buffer.each do |topic, partition, messages|
|
15
|
+
broker_id = @broker_pool.get_leader_id(topic, partition)
|
16
|
+
|
17
|
+
@logger.debug "Current leader for #{topic}/#{partition} is node #{broker_id}"
|
18
|
+
|
19
|
+
messages_for_broker[broker_id] ||= MessageBuffer.new
|
20
|
+
messages_for_broker[broker_id].concat(messages, topic: topic, partition: partition)
|
21
|
+
end
|
22
|
+
|
23
|
+
messages_for_broker.each do |broker_id, message_set|
|
24
|
+
begin
|
25
|
+
broker = @broker_pool.get_broker(broker_id)
|
26
|
+
|
27
|
+
response = broker.produce(
|
28
|
+
messages_for_topics: message_set.to_h,
|
29
|
+
required_acks: @required_acks,
|
30
|
+
timeout: @ack_timeout * 1000, # Kafka expects the timeout in milliseconds.
|
31
|
+
)
|
32
|
+
|
33
|
+
handle_response(response) if response
|
34
|
+
rescue ConnectionError => e
|
35
|
+
@logger.error "Could not connect to broker #{broker_id}: #{e}"
|
36
|
+
|
37
|
+
# Mark the broker pool as stale in order to force a cluster metadata refresh.
|
38
|
+
@broker_pool.mark_as_stale!
|
39
|
+
end
|
40
|
+
end
|
41
|
+
end
|
42
|
+
|
43
|
+
private
|
44
|
+
|
45
|
+
def handle_response(response)
|
46
|
+
response.each_partition do |topic_info, partition_info|
|
47
|
+
topic = topic_info.topic
|
48
|
+
partition = partition_info.partition
|
49
|
+
|
50
|
+
begin
|
51
|
+
Protocol.handle_error(partition_info.error_code)
|
52
|
+
rescue Kafka::CorruptMessage
|
53
|
+
@logger.error "Corrupt message when writing to #{topic}/#{partition}"
|
54
|
+
rescue Kafka::UnknownTopicOrPartition
|
55
|
+
@logger.error "Unknown topic or partition #{topic}/#{partition}"
|
56
|
+
rescue Kafka::LeaderNotAvailable
|
57
|
+
@logger.error "Leader currently not available for #{topic}/#{partition}"
|
58
|
+
@broker_pool.mark_as_stale!
|
59
|
+
rescue Kafka::NotLeaderForPartition
|
60
|
+
@logger.error "Broker not currently leader for #{topic}/#{partition}"
|
61
|
+
@broker_pool.mark_as_stale!
|
62
|
+
rescue Kafka::RequestTimedOut
|
63
|
+
@logger.error "Timed out while writing to #{topic}/#{partition}"
|
64
|
+
rescue Kafka::NotEnoughReplicas
|
65
|
+
@logger.error "Not enough in-sync replicas for #{topic}/#{partition}"
|
66
|
+
rescue Kafka::NotEnoughReplicasAfterAppend
|
67
|
+
@logger.error "Messages written, but to fewer in-sync replicas than required for #{topic}/#{partition}"
|
68
|
+
else
|
69
|
+
offset = partition_info.offset
|
70
|
+
@logger.info "Successfully sent messages for #{topic}/#{partition}; new offset is #{offset}"
|
71
|
+
|
72
|
+
# The messages were successfully written; clear them from the buffer.
|
73
|
+
@buffer.clear_messages(topic: topic, partition: partition)
|
74
|
+
end
|
75
|
+
end
|
76
|
+
end
|
77
|
+
end
|
78
|
+
end
|
data/lib/kafka/version.rb
CHANGED
metadata
CHANGED
@@ -1,14 +1,14 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: ruby-kafka
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 0.1.0.pre.
|
4
|
+
version: 0.1.0.pre.beta5
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- Daniel Schierbeck
|
8
8
|
autorequire:
|
9
9
|
bindir: exe
|
10
10
|
cert_chain: []
|
11
|
-
date: 2016-02-
|
11
|
+
date: 2016-02-02 00:00:00.000000000 Z
|
12
12
|
dependencies:
|
13
13
|
- !ruby/object:Gem::Dependency
|
14
14
|
name: bundler
|
@@ -94,6 +94,20 @@ dependencies:
|
|
94
94
|
- - ">="
|
95
95
|
- !ruby/object:Gem::Version
|
96
96
|
version: '0'
|
97
|
+
- !ruby/object:Gem::Dependency
|
98
|
+
name: rspec-benchmark
|
99
|
+
requirement: !ruby/object:Gem::Requirement
|
100
|
+
requirements:
|
101
|
+
- - ">="
|
102
|
+
- !ruby/object:Gem::Version
|
103
|
+
version: '0'
|
104
|
+
type: :development
|
105
|
+
prerelease: false
|
106
|
+
version_requirements: !ruby/object:Gem::Requirement
|
107
|
+
requirements:
|
108
|
+
- - ">="
|
109
|
+
- !ruby/object:Gem::Version
|
110
|
+
version: '0'
|
97
111
|
description: |-
|
98
112
|
A client library for the Kafka distributed commit log.
|
99
113
|
|
@@ -133,6 +147,8 @@ files:
|
|
133
147
|
- lib/kafka/protocol/produce_response.rb
|
134
148
|
- lib/kafka/protocol/request_message.rb
|
135
149
|
- lib/kafka/protocol/topic_metadata_request.rb
|
150
|
+
- lib/kafka/socket_with_timeout.rb
|
151
|
+
- lib/kafka/transmission.rb
|
136
152
|
- lib/kafka/version.rb
|
137
153
|
- lib/ruby-kafka.rb
|
138
154
|
homepage: https://github.com/zendesk/ruby-kafka
|