tailf2kafka 0.1.7 → 0.1.8
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/bin/tailf2kafka +33 -27
- data/bin/tailf2kafka.save +302 -0
- data/lib/tailf2kafka/version.rb +1 -1
- metadata +4 -2
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA1:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: 36ed03a088f2e37c9e35773e233159bc1c32cf2d
|
4
|
+
data.tar.gz: 203b2fd8a437e4d8baa555f46a246411eed0cb9e
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: 5554c921a0ce29a3bf4f3548e8b180a46545ace278163b4cebaff11255e709565730f5de25263fc4fa44636a345eb6eec2a187e7dd012dc16bdd109bbb02c484
|
7
|
+
data.tar.gz: 11bc611eac30966753a8327acc0a7f97304d6b1c8757893ef7316b6cf8f1acfc83fdc7a9981b46ccb720a63bb3e5940657ddc8d4ed3795b838d590c8f3c1d485
|
data/bin/tailf2kafka
CHANGED
@@ -205,7 +205,7 @@ def time_pattern_to_regexp(pattern)
|
|
205
205
|
end
|
206
206
|
end
|
207
207
|
|
208
|
-
#Scan existing files that match watched prefixes and start
|
208
|
+
#Scan existing files that match watched prefixes and start tailing them
|
209
209
|
@settings[:tailf][:files].each do |tailf_file|
|
210
210
|
tailf_file[:prefix] = File.expand_path(tailf_file[:prefix])
|
211
211
|
dir = File.dirname(tailf_file[:prefix])
|
@@ -257,46 +257,52 @@ end
|
|
257
257
|
|
258
258
|
@timers = Timers::Group.new
|
259
259
|
@uploads_timer = @timers.every(@flush_interval) { write_position_file }
|
260
|
-
@
|
260
|
+
@timers.every(60) { delete_old_tailed_files } if @delete_old_tailed_files
|
261
261
|
Thread.new { loop { @timers.wait } }
|
262
262
|
|
263
|
-
|
264
|
-
|
265
|
-
|
266
|
-
@
|
267
|
-
|
268
|
-
|
269
|
-
|
270
|
-
|
271
|
-
unless
|
272
|
-
@
|
273
|
-
|
274
|
-
|
263
|
+
def setup_watchers
|
264
|
+
@dirs.each_key do |dir|
|
265
|
+
|
266
|
+
@create_notifier.watch(dir, :create, :moved_to) do |event|
|
267
|
+
@mutex.synchronize do
|
268
|
+
path = "#{dir}/#{event.name}"
|
269
|
+
match = @dirs[dir].detect{|h| event.name.match(Regexp.new(h[:prefix] + time_pattern_to_regexp(h[:pattern]) + h[:suffix]))}
|
270
|
+
if match
|
271
|
+
unless File.directory?(path)
|
272
|
+
unless @threads.has_key?(path)
|
273
|
+
@logger.info("File #{event.name} was created in / moved into watched dir #{dir}")
|
274
|
+
@files[path] = { :pattern => match[:pattern], :topic => match[:topic], :inode => File.stat(path).ino, :offset => 0 }
|
275
|
+
@threads[path] = Thread.new { tailf(path) }
|
276
|
+
end
|
275
277
|
end
|
276
278
|
end
|
277
279
|
end
|
278
280
|
end
|
279
|
-
end
|
280
281
|
|
281
|
-
|
282
|
-
|
283
|
-
|
284
|
-
|
285
|
-
|
286
|
-
|
287
|
-
|
288
|
-
|
282
|
+
@delete_notifier.watch(dir, :delete, :moved_from) do |event|
|
283
|
+
@mutex.synchronize do
|
284
|
+
path = "#{dir}/#{event.name}"
|
285
|
+
if @threads.has_key?(path)
|
286
|
+
@logger.info("File #{event.name} was deleted / moved from watched dir #{dir}")
|
287
|
+
if @threads[path].alive?
|
288
|
+
@threads[path].terminate
|
289
|
+
@threads[path].join
|
290
|
+
end
|
291
|
+
@threads.delete(path)
|
292
|
+
@files[path][:fd].close unless @files[path][:fd].closed?
|
293
|
+
@files.delete(path)
|
289
294
|
end
|
290
|
-
@threads.delete(path)
|
291
|
-
@files[path][:fd].close unless @files[path][:fd].closed?
|
292
|
-
@files.delete(path)
|
293
295
|
end
|
294
296
|
end
|
295
|
-
end
|
296
297
|
|
298
|
+
end
|
297
299
|
end
|
298
300
|
|
301
|
+
setup_watchers
|
302
|
+
|
299
303
|
Thread.new { @create_notifier.run }
|
300
304
|
Thread.new { @delete_notifier.run }
|
301
305
|
|
306
|
+
@timers.every(60) { setup_watchers }
|
307
|
+
|
302
308
|
@tailf_notifier.run
|
@@ -0,0 +1,302 @@
|
|
1
|
+
#!/usr/bin/env ruby
|
2
|
+
|
3
|
+
require 'optparse'
|
4
|
+
require 'poseidon'
|
5
|
+
require 'yaml'
|
6
|
+
require 'hash_symbolizer'
|
7
|
+
require 'schash'
|
8
|
+
require 'rb-inotify'
|
9
|
+
require 'timers'
|
10
|
+
require 'socket'
|
11
|
+
require 'fileutils'
|
12
|
+
require 'logger'
|
13
|
+
require 'mixlib/shellout'
|
14
|
+
|
15
|
+
$stdout.sync = true
|
16
|
+
|
17
|
+
Thread.abort_on_exception = true
|
18
|
+
|
19
|
+
@config = nil
|
20
|
+
|
21
|
+
loglevels = {
|
22
|
+
:debug => Logger::DEBUG,
|
23
|
+
:info => Logger::INFO,
|
24
|
+
:warn => Logger::WARN,
|
25
|
+
:error => Logger::Error,
|
26
|
+
:fatal => Logger::FATAL,
|
27
|
+
:unknown => Logger::UNKNOWN
|
28
|
+
}
|
29
|
+
|
30
|
+
@loglevel = Logger::INFO
|
31
|
+
|
32
|
+
opts = OptionParser.new
|
33
|
+
opts.banner = "Usage: #{$0} [options]"
|
34
|
+
opts.on( '--config PATH', String, 'Path to settings config' ) { |c| @config = c }
|
35
|
+
opts.on( '--log-level [LEVEL]', [:debug, :info, :warn, :error, :fatal, :unknown] ) { |l| @loglevel = loglevels[l] }
|
36
|
+
opts.on( '-h', '--help', 'Display this screen' ) { puts opts; exit 0 }
|
37
|
+
opts.parse!
|
38
|
+
|
39
|
+
unless @config
|
40
|
+
puts opts
|
41
|
+
exit 1
|
42
|
+
end
|
43
|
+
|
44
|
+
@logger = Logger.new(STDOUT)
|
45
|
+
|
46
|
+
@settings = YAML.load_file(@config).symbolize_keys(true)
|
47
|
+
|
48
|
+
validator = Schash::Validator.new do
|
49
|
+
{
|
50
|
+
tailf: {
|
51
|
+
files: array_of({
|
52
|
+
topic: string,
|
53
|
+
prefix: string,
|
54
|
+
suffix: optional(string),
|
55
|
+
time_pattern: string,
|
56
|
+
}),
|
57
|
+
position_file: string,
|
58
|
+
flush_interval: integer,
|
59
|
+
max_batch_lines: integer,
|
60
|
+
from_begining: boolean,
|
61
|
+
delete_old_tailed_files: optional(boolean),
|
62
|
+
post_delete_command: optional(string),
|
63
|
+
},
|
64
|
+
kafka: {
|
65
|
+
brokers: array_of(string),
|
66
|
+
producer_type: match(/^(sync|async)$/),
|
67
|
+
produce: optional(boolean),
|
68
|
+
},
|
69
|
+
}
|
70
|
+
end
|
71
|
+
|
72
|
+
unless validator.validate(@settings).empty?
|
73
|
+
@logger.error("ERROR: bad settings")
|
74
|
+
@logger.error(validator.validate(@settings))
|
75
|
+
exit 1
|
76
|
+
end
|
77
|
+
|
78
|
+
@settings[:tailf][:files] = @settings[:tailf][:files].map{|h| h.symbolize_keys(true)}
|
79
|
+
|
80
|
+
@mutex = Mutex.new
|
81
|
+
|
82
|
+
@create_notifier = INotify::Notifier.new
|
83
|
+
@delete_notifier = INotify::Notifier.new
|
84
|
+
@tailf_notifier = INotify::Notifier.new
|
85
|
+
|
86
|
+
@dirs = {}
|
87
|
+
@files = {}
|
88
|
+
@threads = {}
|
89
|
+
@position_file = @settings[:tailf][:position_file]
|
90
|
+
@flush_interval = @settings[:tailf][:flush_interval]
|
91
|
+
@max_batch_lines = @settings[:tailf][:max_batch_lines]
|
92
|
+
@from_begining = @settings[:tailf][:from_begining]
|
93
|
+
@delete_old_tailed_files = @settings[:tailf].has_key?(:delete_old_tailed_files) ? @settings[:tailf][:delete_old_tailed_files] : false
|
94
|
+
@brokers = @settings[:kafka][:brokers]
|
95
|
+
@producer_type = @settings[:kafka][:producer_type].to_sym
|
96
|
+
@produce = @settings[:kafka].has_key?(:produce) ? @settings[:kafka][:produce] : true
|
97
|
+
|
98
|
+
def write_position_file
|
99
|
+
@mutex.synchronize do
|
100
|
+
File.open(@position_file, 'w') do |file|
|
101
|
+
@files.each do |path, attrs|
|
102
|
+
file.puts "#{path} #{attrs[:pattern]} #{attrs[:topic]} #{attrs[:inode]} #{attrs[:offset]}"
|
103
|
+
end
|
104
|
+
end
|
105
|
+
end
|
106
|
+
end
|
107
|
+
|
108
|
+
def load_position_file
|
109
|
+
if File.exist?(@position_file)
|
110
|
+
IO.readlines(@position_file).each do |line|
|
111
|
+
path, pattern, topic, inode, offset = line.split(' ')
|
112
|
+
#Load state only for that exist with same inode and were not truncated/rewinded.
|
113
|
+
if File.exists?(path) and File.stat(path).ino == inode.to_i and File.stat(path).size >= offset.to_i
|
114
|
+
@files[path] = { :pattern => pattern, :topic => topic, :inode => inode.to_i, :offset => offset.to_i }
|
115
|
+
end
|
116
|
+
end
|
117
|
+
end
|
118
|
+
write_position_file
|
119
|
+
end
|
120
|
+
|
121
|
+
load_position_file
|
122
|
+
|
123
|
+
@topics = @settings[:tailf][:files].map{|tailf_file| tailf_file[:topic]}
|
124
|
+
@producer = Poseidon::Producer.new(@brokers, "#{Socket.gethostname}", :type => @producer_type, :compression_codec => :snappy, :compressed_topics => @topics) if @produce
|
125
|
+
|
126
|
+
@producer_queue = SizedQueue.new(10)
|
127
|
+
|
128
|
+
@producer_thread = Thread.new do
|
129
|
+
loop do
|
130
|
+
batch = @producer_queue.pop
|
131
|
+
begin
|
132
|
+
@producer.send_messages(batch[:messages]) if @produce
|
133
|
+
rescue Poseidon::Errors::UnableToFetchMetadata
|
134
|
+
@logger.warn("Got Poseidon::Errors::UnableToFetchMetadata while trying to produce kafka messages, retrying in 1 second ...")
|
135
|
+
sleep 1
|
136
|
+
retry
|
137
|
+
end
|
138
|
+
@files[batch[:path]][:offset] = batch[:offset]
|
139
|
+
end
|
140
|
+
end
|
141
|
+
|
142
|
+
def kafka_produce(path, buffer, offset)
|
143
|
+
truncated = nil
|
144
|
+
|
145
|
+
messages = []
|
146
|
+
while msg = buffer.shift
|
147
|
+
unless msg[-1] == "\n"
|
148
|
+
if buffer.empty?
|
149
|
+
truncated = msg
|
150
|
+
else
|
151
|
+
msg = msg + buffer.shift
|
152
|
+
messages << Poseidon::MessageToSend.new(@files[path][:topic], msg.strip)
|
153
|
+
end
|
154
|
+
else
|
155
|
+
messages << Poseidon::MessageToSend.new(@files[path][:topic], msg.strip)
|
156
|
+
end
|
157
|
+
end
|
158
|
+
@producer_queue.push({ :path => path, :messages => messages, :offset => offset})
|
159
|
+
|
160
|
+
truncated
|
161
|
+
end
|
162
|
+
|
163
|
+
def tailf(path)
|
164
|
+
file = File.open(path, 'r')
|
165
|
+
@files[path][:fd] = file
|
166
|
+
file.seek(@files[path][:offset], IO::SEEK_SET)
|
167
|
+
|
168
|
+
truncated = nil
|
169
|
+
loop do #Fast read file in batches until we reach EOF upon which we start the tailf modify watcher
|
170
|
+
batch = file.each_line.take(@max_batch_lines)
|
171
|
+
break if batch.empty?
|
172
|
+
batch.unshift(truncated + batch.shift) if truncated
|
173
|
+
truncated = kafka_produce(path, batch, file.pos)
|
174
|
+
end
|
175
|
+
|
176
|
+
mutex = Mutex.new
|
177
|
+
@tailf_notifier.watch(path, :modify) do |event|
|
178
|
+
mutex.synchronize do
|
179
|
+
unless file.closed?
|
180
|
+
loop do
|
181
|
+
batch = file.each_line.take(@max_batch_lines)
|
182
|
+
break if batch.empty?
|
183
|
+
batch.unshift(truncated + batch.shift) if truncated
|
184
|
+
truncated = kafka_produce(path, batch, file.pos)
|
185
|
+
end
|
186
|
+
else
|
187
|
+
@logger.warn("watcher got modify event on closed file #{event.name}")
|
188
|
+
end
|
189
|
+
end
|
190
|
+
end
|
191
|
+
end
|
192
|
+
|
193
|
+
@time_regexp_hash = {
|
194
|
+
'Y' => '[0-9]{4}',
|
195
|
+
'm' => '[0-9]{2}',
|
196
|
+
'd' => '[0-9]{2}',
|
197
|
+
'H' => '[0-9]{2}',
|
198
|
+
'M' => '[0-9]{2}'
|
199
|
+
}
|
200
|
+
|
201
|
+
def time_pattern_to_regexp(pattern)
|
202
|
+
pattern.gsub(/%([^%])/) do
|
203
|
+
match = $1
|
204
|
+
@time_regexp_hash.has_key?(match) ? @time_regexp_hash[match] : match
|
205
|
+
end
|
206
|
+
end
|
207
|
+
|
208
|
+
#Scan existing files that match watched prefixes and start failing them
|
209
|
+
@settings[:tailf][:files].each do |tailf_file|
|
210
|
+
dir = File.dirname(tailf_file[:prefix])
|
211
|
+
if File.exists?(dir) and File.directory?(dir)
|
212
|
+
@dirs[dir] ||= []
|
213
|
+
@dirs[dir] << { :prefix => File.basename(tailf_file[:prefix]), :pattern => tailf_file[:time_pattern], :suffix => "#{tailf_file[:suffix]}", :topic => tailf_file[:topic]}
|
214
|
+
Dir.glob("#{tailf_file[:prefix]}*#{tailf_file[:suffix]}").each do |path|
|
215
|
+
if path.match(Regexp.new(time_pattern_to_regexp(tailf_file[:time_pattern])))
|
216
|
+
unless File.directory?(path)
|
217
|
+
#Populate state only if it was not loaded from position file
|
218
|
+
unless @files.has_key?(path)
|
219
|
+
@files[path] = { :pattern => tailf_file[:time_pattern], :topic => tailf_file[:topic], :inode => File.stat(path).ino, :offset => 0 }
|
220
|
+
@files[path][:offset] = File.stat(path).size unless @from_begining
|
221
|
+
end
|
222
|
+
@threads[path] = Thread.new { tailf(path) } unless @threads.has_key?(path)
|
223
|
+
end
|
224
|
+
end
|
225
|
+
end
|
226
|
+
end
|
227
|
+
end
|
228
|
+
|
229
|
+
def delete_old_tailed_files
|
230
|
+
@mutex.synchronize do
|
231
|
+
@files.each_key do |path|
|
232
|
+
unless path.match(Regexp.new(Time.now.strftime(@files[path][:pattern])))
|
233
|
+
if File.exists?(path) and File.stat(path).ino == @files[path][:inode] and File.stat(path).size == @files[path][:offset] and (Time.now - File.stat(path).mtime) > 30
|
234
|
+
@logger.info("Deleteing old time pattern fully kafka produced file #{path}")
|
235
|
+
FileUtils.rm_r(path)
|
236
|
+
if @settings[:tailf].has_key?(:post_delete_command)
|
237
|
+
@logger.info("Running post delete command => #{@settings[:tailf][:post_delete_command]}")
|
238
|
+
command = Mixlib::created
|
239
|
+
ShellOut.new(@settings[:tailf][:post_delete_command])
|
240
|
+
begin
|
241
|
+
command.run_command
|
242
|
+
if command.error?
|
243
|
+
@logger.error("Failed post delete command => #{@settings[:tailf][:post_delete_command]}")
|
244
|
+
@logger.info("STDOUT: #{command.stdout}")
|
245
|
+
@logger.info("STDERR: #{command.stderr}")
|
246
|
+
end
|
247
|
+
rescue => e
|
248
|
+
@logger.error("Failed post delete command => #{@settings[:tailf][:post_delete_command]}")
|
249
|
+
@logger.info(e.message)
|
250
|
+
end
|
251
|
+
end
|
252
|
+
end
|
253
|
+
end
|
254
|
+
end
|
255
|
+
end
|
256
|
+
end
|
257
|
+
|
258
|
+
@timers = Timers::Group.new
|
259
|
+
@uploads_timer = @timers.every(@flush_interval) { write_position_file }
|
260
|
+
@delete_old_tailed_files_timer = @timers.every(60) { delete_old_tailed_files } if @delete_old_tailed_files
|
261
|
+
Thread.new { loop { @timers.wait } }
|
262
|
+
|
263
|
+
@dirs.each_key do |dir|
|
264
|
+
|
265
|
+
@create_notifier.watch(dir, :create, :moved_to) do |event|
|
266
|
+
@mutex.synchronize do
|
267
|
+
path = "#{dir}/#{event.name}"
|
268
|
+
match = @dirs[dir].detect{|h| event.name.match(Regexp.new(h[:prefix] + time_pattern_to_regexp(h[:pattern]) + h[:suffix]))}
|
269
|
+
if match
|
270
|
+
unless File.directory?(path)
|
271
|
+
unless @threads.has_key?(path)
|
272
|
+
@logger.info("File #{event.name} was created in / moved into watched dir #{dir}")
|
273
|
+
@files[path] = { :pattern => match[:pattern], :topic => match[:topic], :inode => File.stat(path).ino, :offset => 0 }
|
274
|
+
@threads[path] = Thread.new { tailf(path) }
|
275
|
+
end
|
276
|
+
end
|
277
|
+
end
|
278
|
+
end
|
279
|
+
end
|
280
|
+
|
281
|
+
@delete_notifier.watch(dir, :delete, :moved_from) do |event|
|
282
|
+
@mutex.synchronize do
|
283
|
+
path = "#{dir}/#{event.name}"
|
284
|
+
if @threads.has_key?(path)
|
285
|
+
@logger.info("File #{event.name} was deleted / moved from watched dir #{dir}")
|
286
|
+
if @threads[path].alive?
|
287
|
+
@threads[path].terminate
|
288
|
+
@threads[path].join
|
289
|
+
end
|
290
|
+
@threads.delete(path)
|
291
|
+
@files[path][:fd].close unless @files[path][:fd].closed?
|
292
|
+
@files.delete(path)
|
293
|
+
end
|
294
|
+
end
|
295
|
+
end
|
296
|
+
|
297
|
+
end
|
298
|
+
|
299
|
+
Thread.new { @create_notifier.run }
|
300
|
+
Thread.new { @delete_notifier.run }
|
301
|
+
|
302
|
+
@tailf_notifier.run
|
data/lib/tailf2kafka/version.rb
CHANGED
metadata
CHANGED
@@ -1,14 +1,14 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: tailf2kafka
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 0.1.
|
4
|
+
version: 0.1.8
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- Alexander Piavlo
|
8
8
|
autorequire:
|
9
9
|
bindir: bin
|
10
10
|
cert_chain: []
|
11
|
-
date:
|
11
|
+
date: 2016-07-03 00:00:00.000000000 Z
|
12
12
|
dependencies:
|
13
13
|
- !ruby/object:Gem::Dependency
|
14
14
|
name: poseidon
|
@@ -128,12 +128,14 @@ email:
|
|
128
128
|
- devops@supersonic.com
|
129
129
|
executables:
|
130
130
|
- tailf2kafka
|
131
|
+
- tailf2kafka.save
|
131
132
|
extensions: []
|
132
133
|
extra_rdoc_files: []
|
133
134
|
files:
|
134
135
|
- LICENSE
|
135
136
|
- README.md
|
136
137
|
- bin/tailf2kafka
|
138
|
+
- bin/tailf2kafka.save
|
137
139
|
- lib/tailf2kafka.rb
|
138
140
|
- lib/tailf2kafka/version.rb
|
139
141
|
- tailf2kafka.gemspec
|