tapsoob 0.1.10
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.gitignore +4 -0
- data/Gemfile +4 -0
- data/Gemfile.lock +24 -0
- data/README.md +66 -0
- data/Rakefile +0 -0
- data/bin/schema +54 -0
- data/bin/tapsoob +6 -0
- data/lib/tapsoob.rb +9 -0
- data/lib/tapsoob/chunksize.rb +53 -0
- data/lib/tapsoob/cli.rb +145 -0
- data/lib/tapsoob/config.rb +33 -0
- data/lib/tapsoob/data_stream.rb +350 -0
- data/lib/tapsoob/errors.rb +16 -0
- data/lib/tapsoob/log.rb +16 -0
- data/lib/tapsoob/operation.rb +468 -0
- data/lib/tapsoob/progress_bar.rb +236 -0
- data/lib/tapsoob/railtie.rb +11 -0
- data/lib/tapsoob/schema.rb +83 -0
- data/lib/tapsoob/utils.rb +179 -0
- data/lib/tapsoob/version.rb +4 -0
- data/lib/tasks/tapsoob.rake +59 -0
- data/tapsoob.gemspec +30 -0
- metadata +138 -0
@@ -0,0 +1,16 @@
|
|
1
|
+
# -*- encoding : utf-8 -*-
|
2
|
+
module Tapsoob
|
3
|
+
class BaseError < StandardError
|
4
|
+
attr_reader :original_backtrace
|
5
|
+
|
6
|
+
def initialize(message, opts = {})
|
7
|
+
@original_backtrace = opts.delete(:backtrace)
|
8
|
+
super(message)
|
9
|
+
end
|
10
|
+
end
|
11
|
+
|
12
|
+
class NotImplemented < BaseError; end
|
13
|
+
class DuplicatePrimaryKeyError < BaseError; end
|
14
|
+
class CorruptedData < BaseError; end
|
15
|
+
class InvalidData < BaseError; end
|
16
|
+
end
|
data/lib/tapsoob/log.rb
ADDED
@@ -0,0 +1,16 @@
|
|
1
|
+
# -*- encoding : utf-8 -*-
|
2
|
+
module Tapsoob
|
3
|
+
def self.log=(log)
|
4
|
+
@@log = log
|
5
|
+
end
|
6
|
+
|
7
|
+
def self.log
|
8
|
+
@@log ||= begin
|
9
|
+
require 'logger'
|
10
|
+
log = Logger.new($stderr)
|
11
|
+
log.level = Logger::ERROR
|
12
|
+
log.datetime_format = "%Y-%m-%d %H:%M:%S"
|
13
|
+
log
|
14
|
+
end
|
15
|
+
end
|
16
|
+
end
|
@@ -0,0 +1,468 @@
|
|
1
|
+
# -*- encoding : utf-8 -*-
|
2
|
+
require 'sequel'
|
3
|
+
|
4
|
+
require 'tapsoob/progress_bar'
|
5
|
+
require 'tapsoob/schema'
|
6
|
+
require 'tapsoob/data_stream'
|
7
|
+
|
8
|
+
module Tapsoob
|
9
|
+
class Operation
|
10
|
+
attr_reader :database_url, :dump_path, :opts
|
11
|
+
|
12
|
+
def initialize(database_url, dump_path, opts={})
|
13
|
+
@database_url = database_url
|
14
|
+
@dump_path = dump_path
|
15
|
+
@opts = opts
|
16
|
+
@exiting = false
|
17
|
+
end
|
18
|
+
|
19
|
+
def file_prefix
|
20
|
+
"op"
|
21
|
+
end
|
22
|
+
|
23
|
+
def skip_schema?
|
24
|
+
!!opts[:skip_schema]
|
25
|
+
end
|
26
|
+
|
27
|
+
def indexes_first?
|
28
|
+
!!opts[:indexes_first]
|
29
|
+
end
|
30
|
+
|
31
|
+
def table_filter
|
32
|
+
opts[:table_filter]
|
33
|
+
end
|
34
|
+
|
35
|
+
def exclude_tables
|
36
|
+
opts[:exclude_tables] || []
|
37
|
+
end
|
38
|
+
|
39
|
+
def apply_table_filter(tables)
|
40
|
+
return tables unless table_filter || exclude_tables
|
41
|
+
|
42
|
+
re = table_filter ? Regexp.new(table_filter) : nil
|
43
|
+
if tables.kind_of?(Hash)
|
44
|
+
ntables = {}
|
45
|
+
tables.each do |t, d|
|
46
|
+
if !exclude_tables.include?(t.to_s) && (!re || !re.match(t.to_s).nil?)
|
47
|
+
ntables[t] = d
|
48
|
+
end
|
49
|
+
end
|
50
|
+
ntables
|
51
|
+
else
|
52
|
+
tables.reject { |t| exclude_tables.include?(t.to_s) || (re && re.match(t.to_s).nil?) }
|
53
|
+
end
|
54
|
+
end
|
55
|
+
|
56
|
+
def log
|
57
|
+
Tapsoob.log
|
58
|
+
end
|
59
|
+
|
60
|
+
def store_session
|
61
|
+
file = "#{file_prefix}_#{Time.now.strftime("%Y%m%d%H%M")}.dat"
|
62
|
+
puts "\nSaving session to #{file}..."
|
63
|
+
File.open(file, 'w') do |f|
|
64
|
+
f.write(JSON.generate(to_hash))
|
65
|
+
end
|
66
|
+
end
|
67
|
+
|
68
|
+
def to_hash
|
69
|
+
{
|
70
|
+
:klass => self.class.to_s,
|
71
|
+
:database_url => database_url,
|
72
|
+
:stream_state => stream_state,
|
73
|
+
:completed_tables => completed_tables,
|
74
|
+
:table_filter => table_filter,
|
75
|
+
}
|
76
|
+
end
|
77
|
+
|
78
|
+
def exiting?
|
79
|
+
!!@exiting
|
80
|
+
end
|
81
|
+
|
82
|
+
def setup_signal_trap
|
83
|
+
trap("INT") {
|
84
|
+
puts "\nCompleting current action..."
|
85
|
+
@exiting = true
|
86
|
+
}
|
87
|
+
|
88
|
+
trap("TERM") {
|
89
|
+
puts "\nCompleting current action..."
|
90
|
+
@exiting = true
|
91
|
+
}
|
92
|
+
end
|
93
|
+
|
94
|
+
def resuming?
|
95
|
+
opts[:resume] == true
|
96
|
+
end
|
97
|
+
|
98
|
+
def default_chunksize
|
99
|
+
opts[:default_chunksize]
|
100
|
+
end
|
101
|
+
|
102
|
+
def completed_tables
|
103
|
+
opts[:completed_tables] ||= []
|
104
|
+
end
|
105
|
+
|
106
|
+
def stream_state
|
107
|
+
opts[:stream_state] ||= {}
|
108
|
+
end
|
109
|
+
|
110
|
+
def stream_state=(val)
|
111
|
+
opts[:stream_state] = val
|
112
|
+
end
|
113
|
+
|
114
|
+
def db
|
115
|
+
@db ||= Sequel.connect(database_url)
|
116
|
+
end
|
117
|
+
|
118
|
+
def format_number(num)
|
119
|
+
num.to_s.gsub(/(\d)(?=(\d\d\d)+(?!\d))/, "\\1,")
|
120
|
+
end
|
121
|
+
|
122
|
+
def catch_errors(&blk)
|
123
|
+
begin
|
124
|
+
blk.call
|
125
|
+
rescue Exception => e
|
126
|
+
raise e
|
127
|
+
end
|
128
|
+
end
|
129
|
+
|
130
|
+
def self.factory(type, database_url, dump_path, opts)
|
131
|
+
type = :resume if opts[:resume]
|
132
|
+
klass = case type
|
133
|
+
when :pull then Tapsoob::Pull
|
134
|
+
when :push then Tapsoob::Push
|
135
|
+
when :resume then eval(opts[:klass])
|
136
|
+
else raise "Unknown Operation Type -> #{type}"
|
137
|
+
end
|
138
|
+
|
139
|
+
klass.new(database_url, dump_path, opts)
|
140
|
+
end
|
141
|
+
end
|
142
|
+
|
143
|
+
class Pull < Operation
|
144
|
+
def file_prefix
|
145
|
+
"pull"
|
146
|
+
end
|
147
|
+
|
148
|
+
def to_hash
|
149
|
+
super.merge(:remote_tables_info => remote_tables_info)
|
150
|
+
end
|
151
|
+
|
152
|
+
def run
|
153
|
+
catch_errors do
|
154
|
+
unless resuming?
|
155
|
+
pull_schema if !skip_schema?
|
156
|
+
pull_indexes if indexes_first? && !skip_schema?
|
157
|
+
end
|
158
|
+
setup_signal_trap
|
159
|
+
pull_partial_data if resuming?
|
160
|
+
pull_data
|
161
|
+
pull_indexes if !indexes_first? && !skip_schema?
|
162
|
+
pull_reset_sequences
|
163
|
+
end
|
164
|
+
end
|
165
|
+
|
166
|
+
def pull_schema
|
167
|
+
puts "Receiving schema"
|
168
|
+
|
169
|
+
progress = ProgressBar.new('Schema', tables.size)
|
170
|
+
tables.each do |table_name, count|
|
171
|
+
schema_data = Tapsoob::Schema.dump_table(database_url, table_name)
|
172
|
+
log.debug "Table: #{table_name}\n#{schema_data}\n"
|
173
|
+
output = Tapsoob::Utils.export_schema(dump_path, table_name, schema_data)
|
174
|
+
puts output if output
|
175
|
+
progress.inc(1)
|
176
|
+
end
|
177
|
+
progress.finish
|
178
|
+
end
|
179
|
+
|
180
|
+
def pull_data
|
181
|
+
puts "Receiving data"
|
182
|
+
|
183
|
+
puts "#{tables.size} tables, #{format_number(record_count)} records"
|
184
|
+
|
185
|
+
tables.each do |table_name, count|
|
186
|
+
progress = ProgressBar.new(table_name.to_s, count)
|
187
|
+
stream = Tapsoob::DataStream.factory(db, {
|
188
|
+
:chunksize => default_chunksize,
|
189
|
+
:table_name => table_name
|
190
|
+
})
|
191
|
+
pull_data_from_table(stream, progress)
|
192
|
+
end
|
193
|
+
end
|
194
|
+
|
195
|
+
def pull_partial_data
|
196
|
+
return if stream_state == {}
|
197
|
+
|
198
|
+
table_name = stream_state[:table_name]
|
199
|
+
record_count = tables[table_name.to_s]
|
200
|
+
puts "Resuming #{table_name}, #{format_number(record_count)} records"
|
201
|
+
|
202
|
+
progress = ProgressBar.new(table_name.to_s, record_count)
|
203
|
+
stream = Tapsoob::DataStream.factory(db, stream_state)
|
204
|
+
pull_data_from_table(stream, progress)
|
205
|
+
end
|
206
|
+
|
207
|
+
def pull_data_from_table(stream, progress)
|
208
|
+
loop do
|
209
|
+
begin
|
210
|
+
exit 0 if exiting?
|
211
|
+
|
212
|
+
size = stream.fetch_database(dump_path)
|
213
|
+
break if stream.complete?
|
214
|
+
progress.inc(size) unless exiting?
|
215
|
+
stream.error = false
|
216
|
+
self.stream_state = stream.to_hash
|
217
|
+
rescue Tapsoob::CorruptedData => e
|
218
|
+
puts "Corrupted Data Received #{e.message}, retrying..."
|
219
|
+
stream.error = true
|
220
|
+
next
|
221
|
+
end
|
222
|
+
end
|
223
|
+
|
224
|
+
progress.finish
|
225
|
+
completed_tables << stream.table_name.to_s
|
226
|
+
self.stream_state = {}
|
227
|
+
end
|
228
|
+
|
229
|
+
def tables
|
230
|
+
h = {}
|
231
|
+
tables_info.each do |table_name, count|
|
232
|
+
next if completed_tables.include?(table_name.to_s)
|
233
|
+
h[table_name.to_s] = count
|
234
|
+
end
|
235
|
+
h
|
236
|
+
end
|
237
|
+
|
238
|
+
def record_count
|
239
|
+
tables_info.values.inject(:+)
|
240
|
+
end
|
241
|
+
|
242
|
+
def tables_info
|
243
|
+
opts[:tables_info] ||= fetch_tables_info
|
244
|
+
end
|
245
|
+
|
246
|
+
def fetch_tables_info
|
247
|
+
tables = db.tables
|
248
|
+
|
249
|
+
data = {}
|
250
|
+
apply_table_filter(tables).each do |table_name|
|
251
|
+
data[table_name] = db[table_name].count
|
252
|
+
end
|
253
|
+
data
|
254
|
+
end
|
255
|
+
|
256
|
+
def self.factory(db, state)
|
257
|
+
if defined?(Sequel::MySQL) && Sequel::MySQL.respond_to?(:convert_invalid_date_time=)
|
258
|
+
Sequel::MySQL.convert_invalid_date_time = :nil
|
259
|
+
end
|
260
|
+
|
261
|
+
if state.has_key?(:klass)
|
262
|
+
return eval(state[:klass]).new(db, state)
|
263
|
+
end
|
264
|
+
|
265
|
+
if Taps::Utils.single_integer_primary_key(db, state[:table_name].to_sym)
|
266
|
+
DataStreamKeyed.new(db, state)
|
267
|
+
else
|
268
|
+
DataStream.new(db, state)
|
269
|
+
end
|
270
|
+
end
|
271
|
+
|
272
|
+
def pull_indexes
|
273
|
+
puts "Receiving indexes"
|
274
|
+
|
275
|
+
idxs = JSON.parse(Tapsoob::Utils.schema_bin(:indexes_individual, database_url))
|
276
|
+
|
277
|
+
apply_table_filter(idxs).each do |table, indexes|
|
278
|
+
next unless indexes.size > 0
|
279
|
+
progress = ProgressBar.new(table, indexes.size)
|
280
|
+
indexes.each do |idx|
|
281
|
+
output = Tapsoob::Utils.export_indexes(dump_path, table, idx)
|
282
|
+
puts output if output
|
283
|
+
progress.inc(1)
|
284
|
+
end
|
285
|
+
progress.finish
|
286
|
+
end
|
287
|
+
end
|
288
|
+
|
289
|
+
def pull_reset_sequences
|
290
|
+
puts "Resetting sequences"
|
291
|
+
|
292
|
+
output = Tapsoob::Utils.schema_bin(:reset_db_sequences, database_url)
|
293
|
+
puts output if output
|
294
|
+
end
|
295
|
+
end
|
296
|
+
|
297
|
+
class Push < Operation
|
298
|
+
def file_prefix
|
299
|
+
"push"
|
300
|
+
end
|
301
|
+
|
302
|
+
def to_hash
|
303
|
+
super.merge(:local_tables_info => local_tables_info)
|
304
|
+
end
|
305
|
+
|
306
|
+
def run
|
307
|
+
catch_errors do
|
308
|
+
unless resuming?
|
309
|
+
push_schema if !skip_schema?
|
310
|
+
push_indexes if indexes_first? && !skip_schema?
|
311
|
+
end
|
312
|
+
setup_signal_trap
|
313
|
+
push_partial_data if resuming?
|
314
|
+
push_data
|
315
|
+
push_indexes if !indexes_first? && !skip_schema?
|
316
|
+
push_reset_sequences
|
317
|
+
end
|
318
|
+
end
|
319
|
+
|
320
|
+
def push_indexes
|
321
|
+
idxs = {}
|
322
|
+
table_idxs = Dir.glob(File.join(dump_path, "indexes", "*.json")).map { |path| File.basename(path, '.json') }
|
323
|
+
table_idxs.each do |table_idx|
|
324
|
+
idxs[table_idx] = JSON.parse(File.read(File.join(dump_path, "indexes", "#{table_idx}.json")))
|
325
|
+
end
|
326
|
+
|
327
|
+
return unless idxs.size > 0
|
328
|
+
|
329
|
+
puts "Sending indexes"
|
330
|
+
|
331
|
+
apply_table_filter(idxs).each do |table, indexes|
|
332
|
+
next unless indexes.size > 0
|
333
|
+
progress = ProgressBar.new(table, indexes.size)
|
334
|
+
indexes.each do |idx|
|
335
|
+
Tapsoob::Utils.load_indexes(database_url, idx)
|
336
|
+
progress.inc(1)
|
337
|
+
end
|
338
|
+
progress.finish
|
339
|
+
end
|
340
|
+
end
|
341
|
+
|
342
|
+
def push_schema
|
343
|
+
puts "Sending schema"
|
344
|
+
|
345
|
+
progress = ProgressBar.new('Schema', tables.size)
|
346
|
+
tables.each do |table, count|
|
347
|
+
log.debug "Loading '#{table}' schema\n"
|
348
|
+
Tapsoob::Utils.load_schema(dump_path, database_url, table)
|
349
|
+
progress.inc(1)
|
350
|
+
end
|
351
|
+
progress.finish
|
352
|
+
end
|
353
|
+
|
354
|
+
def push_reset_sequences
|
355
|
+
puts "Resetting sequences"
|
356
|
+
|
357
|
+
Tapsoob::Utils.schema_bin(:reset_db_sequences, database_url)
|
358
|
+
end
|
359
|
+
|
360
|
+
def push_partial_data
|
361
|
+
return if stream_state == {}
|
362
|
+
|
363
|
+
table_name = stream_state[:table_name]
|
364
|
+
record_count = tables[table_name.to_s]
|
365
|
+
puts "Resuming #{table_name}, #{format_number(record_count)} records"
|
366
|
+
progress = ProgressBar.new(table_name.to_s, record_count)
|
367
|
+
stream = Tapsoob::DataStream.factory(db, stream_state)
|
368
|
+
push_data_from_file(stream, progress)
|
369
|
+
end
|
370
|
+
|
371
|
+
def push_data
|
372
|
+
puts "Sending data"
|
373
|
+
|
374
|
+
puts "#{tables.size} tables, #{format_number(record_count)} records"
|
375
|
+
|
376
|
+
tables.each do |table_name, count|
|
377
|
+
stream = Tapsoob::DataStream.factory(db,
|
378
|
+
:table_name => table_name,
|
379
|
+
:chunksize => default_chunksize)
|
380
|
+
progress = ProgressBar.new(table_name.to_s, count)
|
381
|
+
push_data_from_file(stream, progress)
|
382
|
+
end
|
383
|
+
end
|
384
|
+
|
385
|
+
def push_data_from_file(stream, progress)
|
386
|
+
loop do
|
387
|
+
if exiting?
|
388
|
+
store_session
|
389
|
+
exit 0
|
390
|
+
end
|
391
|
+
|
392
|
+
row_size = 0
|
393
|
+
chunksize = stream.state[:chunksize]
|
394
|
+
|
395
|
+
begin
|
396
|
+
chunksize = Tapsoob::Utils.calculate_chunksize(chunksize) do |c|
|
397
|
+
stream.state[:chunksize] = c.to_i
|
398
|
+
encoded_data, row_size, elapsed_time = nil
|
399
|
+
d1 = c.time_delta do
|
400
|
+
encoded_data, row_size, elapsed_time = stream.fetch({ :type => "file", :source => dump_path })
|
401
|
+
end
|
402
|
+
break if stream.complete?
|
403
|
+
|
404
|
+
data = nil
|
405
|
+
d2 = c.time_delta do
|
406
|
+
data = {
|
407
|
+
:state => stream.to_hash,
|
408
|
+
:checksum => Tapsoob::Utils.checksum(encoded_data).to_s
|
409
|
+
}
|
410
|
+
end
|
411
|
+
|
412
|
+
size = stream.fetch_data_in_database({ :encoded_data => encoded_data, :checksum => data[:checksum] })
|
413
|
+
self.stream_state = stream.to_hash
|
414
|
+
|
415
|
+
c.idle_secs = (d1 + d2)
|
416
|
+
|
417
|
+
elapsed_time
|
418
|
+
end
|
419
|
+
rescue Tapsoob::CorruptedData => e
|
420
|
+
# retry the same data, it got corrupted somehow.
|
421
|
+
next
|
422
|
+
rescue Tapsoob::DuplicatePrimaryKeyError => e
|
423
|
+
# verify the stream and retry it
|
424
|
+
stream.verify_stream
|
425
|
+
stream = JSON.generate({ :state => stream.to_hash })
|
426
|
+
next
|
427
|
+
end
|
428
|
+
stream.state[:chunksize] = chunksize
|
429
|
+
|
430
|
+
progress.inc(row_size)
|
431
|
+
|
432
|
+
stream.increment(row_size)
|
433
|
+
break if stream.complete?
|
434
|
+
end
|
435
|
+
|
436
|
+
progress.finish
|
437
|
+
completed_tables << stream.table_name.to_s
|
438
|
+
self.stream_state = {}
|
439
|
+
end
|
440
|
+
|
441
|
+
def local_tables_info
|
442
|
+
opts[:local_tables_info] ||= fetch_local_tables_info
|
443
|
+
end
|
444
|
+
|
445
|
+
def tables
|
446
|
+
h = {}
|
447
|
+
local_tables_info.each do |table_name, count|
|
448
|
+
next if completed_tables.include?(table_name.to_s)
|
449
|
+
h[table_name.to_s] = count
|
450
|
+
end
|
451
|
+
h
|
452
|
+
end
|
453
|
+
|
454
|
+
def record_count
|
455
|
+
@record_count ||= local_tables_info.values.inject(0) { |a,c| a += c }
|
456
|
+
end
|
457
|
+
|
458
|
+
def fetch_local_tables_info
|
459
|
+
tables_with_counts = {}
|
460
|
+
tbls = Dir.glob(File.join(dump_path, "data", "*")).map { |path| File.basename(path, ".json") }
|
461
|
+
tbls.each do |table|
|
462
|
+
data = JSON.parse(File.read(File.join(dump_path, "data", "#{table}.json")))
|
463
|
+
tables_with_counts[table] = data.size
|
464
|
+
end
|
465
|
+
apply_table_filter(tables_with_counts)
|
466
|
+
end
|
467
|
+
end
|
468
|
+
end
|