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
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
|
# frozen_string_literal: true
module Gitlab
module Database
module MigrationHelpers
include Migrations::BackgroundMigrationHelpers
# https://www.postgresql.org/docs/current/sql-syntax-lexical.html#SQL-SYNTAX-IDENTIFIERS
MAX_IDENTIFIER_NAME_LENGTH = 63
PERMITTED_TIMESTAMP_COLUMNS = %i[created_at updated_at deleted_at].to_set.freeze
DEFAULT_TIMESTAMP_COLUMNS = %i[created_at updated_at].freeze
# Adds `created_at` and `updated_at` columns with timezone information.
#
# This method is an improved version of Rails' built-in method `add_timestamps`.
#
# By default, adds `created_at` and `updated_at` columns, but these can be specified as:
#
# add_timestamps_with_timezone(:my_table, columns: [:created_at, :deleted_at])
#
# This allows you to create just the timestamps you need, saving space.
#
# Available options are:
# :default - The default value for the column.
# :null - When set to `true` the column will allow NULL values.
# The default is to not allow NULL values.
# :columns - the column names to create. Must be one
# of `Gitlab::Database::MigrationHelpers::PERMITTED_TIMESTAMP_COLUMNS`.
# Default value: `DEFAULT_TIMESTAMP_COLUMNS`
#
# All options are optional.
def add_timestamps_with_timezone(table_name, options = {})
options[:null] = false if options[:null].nil?
columns = options.fetch(:columns, DEFAULT_TIMESTAMP_COLUMNS)
default_value = options[:default]
validate_not_in_transaction!(:add_timestamps_with_timezone, 'with default value') if default_value
columns.each do |column_name|
validate_timestamp_column_name!(column_name)
# If default value is presented, use `add_column_with_default` method instead.
if default_value
add_column_with_default(
table_name,
column_name,
:datetime_with_timezone,
default: default_value,
allow_null: options[:null]
)
else
add_column(table_name, column_name, :datetime_with_timezone, options)
end
end
end
# To be used in the `#down` method of migrations that
# use `#add_timestamps_with_timezone`.
#
# Available options are:
# :columns - the column names to remove. Must be one
# Default value: `DEFAULT_TIMESTAMP_COLUMNS`
#
# All options are optional.
def remove_timestamps(table_name, options = {})
columns = options.fetch(:columns, DEFAULT_TIMESTAMP_COLUMNS)
columns.each do |column_name|
remove_column(table_name, column_name)
end
end
# Creates a new index, concurrently
#
# Example:
#
# add_concurrent_index :users, :some_column
#
# See Rails' `add_index` for more info on the available arguments.
def add_concurrent_index(table_name, column_name, options = {})
if transaction_open?
raise 'add_concurrent_index can not be run inside a transaction, ' \
'you can disable transactions by calling disable_ddl_transaction! ' \
'in the body of your migration class'
end
options = options.merge({ algorithm: :concurrently })
if index_exists?(table_name, column_name, options)
Gitlab::AppLogger.warn "Index not created because it already exists (this may be due to an aborted migration or similar): table_name: #{table_name}, column_name: #{column_name}"
return
end
disable_statement_timeout do
add_index(table_name, column_name, options)
end
end
# Removes an existed index, concurrently
#
# Example:
#
# remove_concurrent_index :users, :some_column
#
# See Rails' `remove_index` for more info on the available arguments.
def remove_concurrent_index(table_name, column_name, options = {})
if transaction_open?
raise 'remove_concurrent_index can not be run inside a transaction, ' \
'you can disable transactions by calling disable_ddl_transaction! ' \
'in the body of your migration class'
end
options = options.merge({ algorithm: :concurrently })
unless index_exists?(table_name, column_name, options)
Gitlab::AppLogger.warn "Index not removed because it does not exist (this may be due to an aborted migration or similar): table_name: #{table_name}, column_name: #{column_name}"
return
end
disable_statement_timeout do
remove_index(table_name, options.merge({ column: column_name }))
end
end
# Removes an existing index, concurrently
#
# Example:
#
# remove_concurrent_index :users, "index_X_by_Y"
#
# See Rails' `remove_index` for more info on the available arguments.
def remove_concurrent_index_by_name(table_name, index_name, options = {})
if transaction_open?
raise 'remove_concurrent_index_by_name can not be run inside a transaction, ' \
'you can disable transactions by calling disable_ddl_transaction! ' \
'in the body of your migration class'
end
index_name = index_name[:name] if index_name.is_a?(Hash)
raise 'remove_concurrent_index_by_name must get an index name as the second argument' if index_name.blank?
options = options.merge({ algorithm: :concurrently })
unless index_exists_by_name?(table_name, index_name)
Gitlab::AppLogger.warn "Index not removed because it does not exist (this may be due to an aborted migration or similar): table_name: #{table_name}, index_name: #{index_name}"
return
end
disable_statement_timeout do
remove_index(table_name, options.merge({ name: index_name }))
end
end
# Adds a foreign key with only minimal locking on the tables involved.
#
# This method only requires minimal locking
#
# source - The source table containing the foreign key.
# target - The target table the key points to.
# column - The name of the column to create the foreign key on.
# on_delete - The action to perform when associated data is removed,
# defaults to "CASCADE".
# name - The name of the foreign key.
#
def add_concurrent_foreign_key(source, target, column:, on_delete: :cascade, name: nil, validate: true)
# Transactions would result in ALTER TABLE locks being held for the
# duration of the transaction, defeating the purpose of this method.
if transaction_open?
raise 'add_concurrent_foreign_key can not be run inside a transaction'
end
options = {
column: column,
on_delete: on_delete,
name: name.presence || concurrent_foreign_key_name(source, column)
}
if foreign_key_exists?(source, target, options)
warning_message = "Foreign key not created because it exists already " \
"(this may be due to an aborted migration or similar): " \
"source: #{source}, target: #{target}, column: #{options[:column]}, "\
"name: #{options[:name]}, on_delete: #{options[:on_delete]}"
Gitlab::AppLogger.warn warning_message
else
# Using NOT VALID allows us to create a key without immediately
# validating it. This means we keep the ALTER TABLE lock only for a
# short period of time. The key _is_ enforced for any newly created
# data.
with_lock_retries do
execute <<-EOF.strip_heredoc
ALTER TABLE #{source}
ADD CONSTRAINT #{options[:name]}
FOREIGN KEY (#{options[:column]})
REFERENCES #{target} (id)
#{on_delete_statement(options[:on_delete])}
NOT VALID;
EOF
end
end
# Validate the existing constraint. This can potentially take a very
# long time to complete, but fortunately does not lock the source table
# while running.
# Disable this check by passing `validate: false` to the method call
# The check will be enforced for new data (inserts) coming in,
# but validating existing data is delayed.
#
# Note this is a no-op in case the constraint is VALID already
if validate
disable_statement_timeout do
execute("ALTER TABLE #{source} VALIDATE CONSTRAINT #{options[:name]};")
end
end
end
def validate_foreign_key(source, column, name: nil)
fk_name = name || concurrent_foreign_key_name(source, column)
unless foreign_key_exists?(source, name: fk_name)
raise missing_schema_object_message(source, "foreign key", fk_name)
end
disable_statement_timeout do
execute("ALTER TABLE #{source} VALIDATE CONSTRAINT #{fk_name};")
end
end
def foreign_key_exists?(source, target = nil, **options)
foreign_keys(source).any? do |foreign_key|
tables_match?(target.to_s, foreign_key.to_table.to_s) &&
options_match?(foreign_key.options, options)
end
end
# Returns the name for a concurrent foreign key.
#
# PostgreSQL constraint names have a limit of 63 bytes. The logic used
# here is based on Rails' foreign_key_name() method, which unfortunately
# is private so we can't rely on it directly.
#
# prefix:
# - The default prefix is `fk_` for backward compatibility with the existing
# concurrent foreign key helpers.
# - For standard rails foreign keys the prefix is `fk_rails_`
#
def concurrent_foreign_key_name(table, column, prefix: 'fk_')
identifier = "#{table}_#{column}_fk"
hashed_identifier = Digest::SHA256.hexdigest(identifier).first(10)
"#{prefix}#{hashed_identifier}"
end
# Long-running migrations may take more than the timeout allowed by
# the database. Disable the session's statement timeout to ensure
# migrations don't get killed prematurely.
#
# There are two possible ways to disable the statement timeout:
#
# - Per transaction (this is the preferred and default mode)
# - Per connection (requires a cleanup after the execution)
#
# When using a per connection disable statement, code must be inside
# a block so we can automatically execute `RESET ALL` after block finishes
# otherwise the statement will still be disabled until connection is dropped
# or `RESET ALL` is executed
def disable_statement_timeout
if block_given?
if statement_timeout_disabled?
# Don't do anything if the statement_timeout is already disabled
# Allows for nested calls of disable_statement_timeout without
# resetting the timeout too early (before the outer call ends)
yield
else
begin
execute('SET statement_timeout TO 0')
yield
ensure
execute('RESET ALL')
end
end
else
unless transaction_open?
raise <<~ERROR
Cannot call disable_statement_timeout() without a transaction open or outside of a transaction block.
If you don't want to use a transaction wrap your code in a block call:
disable_statement_timeout { # code that requires disabled statement here }
This will make sure statement_timeout is disabled before and reset after the block execution is finished.
ERROR
end
execute('SET LOCAL statement_timeout TO 0')
end
end
# Executes the block with a retry mechanism that alters the +lock_timeout+ and +sleep_time+ between attempts.
# The timings can be controlled via the +timing_configuration+ parameter.
# If the lock was not acquired within the retry period, a last attempt is made without using +lock_timeout+.
#
# ==== Examples
# # Invoking without parameters
# with_lock_retries do
# drop_table :my_table
# end
#
# # Invoking with custom +timing_configuration+
# t = [
# [1.second, 1.second],
# [2.seconds, 2.seconds]
# ]
#
# with_lock_retries(timing_configuration: t) do
# drop_table :my_table # this will be retried twice
# end
#
# # Disabling the retries using an environment variable
# > export DISABLE_LOCK_RETRIES=true
#
# with_lock_retries do
# drop_table :my_table # one invocation, it will not retry at all
# end
#
# ==== Parameters
# * +timing_configuration+ - [[ActiveSupport::Duration, ActiveSupport::Duration], ...] lock timeout for the block, sleep time before the next iteration, defaults to `Gitlab::Database::WithLockRetries::DEFAULT_TIMING_CONFIGURATION`
# * +logger+ - [Gitlab::JsonLogger]
# * +env+ - [Hash] custom environment hash, see the example with `DISABLE_LOCK_RETRIES`
def with_lock_retries(**args, &block)
merged_args = {
klass: self.class,
logger: Gitlab::BackgroundMigration::Logger
}.merge(args)
Gitlab::Database::WithLockRetries.new(merged_args).run(&block)
end
def true_value
Database.true_value
end
def false_value
Database.false_value
end
# Updates the value of a column in batches.
#
# This method updates the table in batches of 5% of the total row count.
# A `batch_size` option can also be passed to set this to a fixed number.
# This method will continue updating rows until no rows remain.
#
# When given a block this method will yield two values to the block:
#
# 1. An instance of `Arel::Table` for the table that is being updated.
# 2. The query to run as an Arel object.
#
# By supplying a block one can add extra conditions to the queries being
# executed. Note that the same block is used for _all_ queries.
#
# Example:
#
# update_column_in_batches(:projects, :foo, 10) do |table, query|
# query.where(table[:some_column].eq('hello'))
# end
#
# This would result in this method updating only rows where
# `projects.some_column` equals "hello".
#
# table - The name of the table.
# column - The name of the column to update.
# value - The value for the column.
#
# The `value` argument is typically a literal. To perform a computed
# update, an Arel literal can be used instead:
#
# update_value = Arel.sql('bar * baz')
#
# update_column_in_batches(:projects, :foo, update_value) do |table, query|
# query.where(table[:some_column].eq('hello'))
# end
#
# Rubocop's Metrics/AbcSize metric is disabled for this method as Rubocop
# determines this method to be too complex while there's no way to make it
# less "complex" without introducing extra methods (which actually will
# make things _more_ complex).
#
# `batch_column_name` option is for tables without primary key, in this
# case another unique integer column can be used. Example: :user_id
#
# rubocop: disable Metrics/AbcSize
def update_column_in_batches(table, column, value, batch_size: nil, batch_column_name: :id)
if transaction_open?
raise 'update_column_in_batches can not be run inside a transaction, ' \
'you can disable transactions by calling disable_ddl_transaction! ' \
'in the body of your migration class'
end
table = Arel::Table.new(table)
count_arel = table.project(Arel.star.count.as('count'))
count_arel = yield table, count_arel if block_given?
total = exec_query(count_arel.to_sql).to_a.first['count'].to_i
return if total == 0
if batch_size.nil?
# Update in batches of 5% until we run out of any rows to update.
batch_size = ((total / 100.0) * 5.0).ceil
max_size = 1000
# The upper limit is 1000 to ensure we don't lock too many rows. For
# example, for "merge_requests" even 1% of the table is around 35 000
# rows for GitLab.com.
batch_size = max_size if batch_size > max_size
end
start_arel = table.project(table[batch_column_name]).order(table[batch_column_name].asc).take(1)
start_arel = yield table, start_arel if block_given?
start_id = exec_query(start_arel.to_sql).to_a.first[batch_column_name.to_s].to_i
loop do
stop_arel = table.project(table[batch_column_name])
.where(table[batch_column_name].gteq(start_id))
.order(table[batch_column_name].asc)
.take(1)
.skip(batch_size)
stop_arel = yield table, stop_arel if block_given?
stop_row = exec_query(stop_arel.to_sql).to_a.first
update_arel = Arel::UpdateManager.new
.table(table)
.set([[table[column], value]])
.where(table[batch_column_name].gteq(start_id))
if stop_row
stop_id = stop_row[batch_column_name.to_s].to_i
start_id = stop_id
update_arel = update_arel.where(table[batch_column_name].lt(stop_id))
end
update_arel = yield table, update_arel if block_given?
execute(update_arel.to_sql)
# There are no more rows left to update.
break unless stop_row
end
end
# Adds a column with a default value without locking an entire table.
#
# @deprecated With PostgreSQL 11, adding columns with a default does not lead to a table rewrite anymore.
# As such, this method is not needed anymore and the default `add_column` helper should be used.
# This helper is subject to be removed in a >13.0 release.
def add_column_with_default(table, column, type, default:, limit: nil, allow_null: false)
raise 'Deprecated: add_column_with_default does not support being passed blocks anymore' if block_given?
add_column(table, column, type, default: default, limit: limit, null: allow_null)
end
# Renames a column without requiring downtime.
#
# Concurrent renames work by using database triggers to ensure both the
# old and new column are in sync. However, this method will _not_ remove
# the triggers or the old column automatically; this needs to be done
# manually in a post-deployment migration. This can be done using the
# method `cleanup_concurrent_column_rename`.
#
# table - The name of the database table containing the column.
# old - The old column name.
# new - The new column name.
# type - The type of the new column. If no type is given the old column's
# type is used.
# batch_column_name - option is for tables without primary key, in this
# case another unique integer column can be used. Example: :user_id
def rename_column_concurrently(table, old, new, type: nil, type_cast_function: nil, batch_column_name: :id)
unless column_exists?(table, batch_column_name)
raise "Column #{batch_column_name} does not exist on #{table}"
end
if transaction_open?
raise 'rename_column_concurrently can not be run inside a transaction'
end
check_trigger_permissions!(table)
create_column_from(table, old, new, type: type, batch_column_name: batch_column_name, type_cast_function: type_cast_function)
install_rename_triggers(table, old, new)
end
# Reverses operations performed by rename_column_concurrently.
#
# This method takes care of removing previously installed triggers as well
# as removing the new column.
#
# table - The name of the database table.
# old - The name of the old column.
# new - The name of the new column.
def undo_rename_column_concurrently(table, old, new)
trigger_name = rename_trigger_name(table, old, new)
check_trigger_permissions!(table)
remove_rename_triggers_for_postgresql(table, trigger_name)
remove_column(table, new)
end
# Installs triggers in a table that keep a new column in sync with an old
# one.
#
# table - The name of the table to install the trigger in.
# old_column - The name of the old column.
# new_column - The name of the new column.
def install_rename_triggers(table, old_column, new_column)
trigger_name = rename_trigger_name(table, old_column, new_column)
quoted_table = quote_table_name(table)
quoted_old = quote_column_name(old_column)
quoted_new = quote_column_name(new_column)
install_rename_triggers_for_postgresql(
trigger_name,
quoted_table,
quoted_old,
quoted_new
)
end
# Changes the type of a column concurrently.
#
# table - The table containing the column.
# column - The name of the column to change.
# new_type - The new column type.
def change_column_type_concurrently(table, column, new_type, type_cast_function: nil)
temp_column = "#{column}_for_type_change"
rename_column_concurrently(table, column, temp_column, type: new_type, type_cast_function: type_cast_function)
end
# Performs cleanup of a concurrent type change.
#
# table - The table containing the column.
# column - The name of the column to change.
# new_type - The new column type.
def cleanup_concurrent_column_type_change(table, column)
temp_column = "#{column}_for_type_change"
transaction do
# This has to be performed in a transaction as otherwise we might have
# inconsistent data.
cleanup_concurrent_column_rename(table, column, temp_column)
rename_column(table, temp_column, column)
end
end
# Cleans up a concurrent column name.
#
# This method takes care of removing previously installed triggers as well
# as removing the old column.
#
# table - The name of the database table.
# old - The name of the old column.
# new - The name of the new column.
def cleanup_concurrent_column_rename(table, old, new)
trigger_name = rename_trigger_name(table, old, new)
check_trigger_permissions!(table)
remove_rename_triggers_for_postgresql(table, trigger_name)
remove_column(table, old)
end
# Reverses the operations performed by cleanup_concurrent_column_rename.
#
# This method adds back the old_column removed
# by cleanup_concurrent_column_rename.
# It also adds back the (old_column > new_column) trigger that is removed
# by cleanup_concurrent_column_rename.
#
# table - The name of the database table containing the column.
# old - The old column name.
# new - The new column name.
# type - The type of the old column. If no type is given the new column's
# type is used.
# batch_column_name - option is for tables without primary key, in this
# case another unique integer column can be used. Example: :user_id
def undo_cleanup_concurrent_column_rename(table, old, new, type: nil, batch_column_name: :id)
unless column_exists?(table, batch_column_name)
raise "Column #{batch_column_name} does not exist on #{table}"
end
if transaction_open?
raise 'undo_cleanup_concurrent_column_rename can not be run inside a transaction'
end
check_trigger_permissions!(table)
create_column_from(table, new, old, type: type, batch_column_name: batch_column_name)
install_rename_triggers(table, old, new)
end
# Changes the column type of a table using a background migration.
#
# Because this method uses a background migration it's more suitable for
# large tables. For small tables it's better to use
# `change_column_type_concurrently` since it can complete its work in a
# much shorter amount of time and doesn't rely on Sidekiq.
#
# Example usage:
#
# class Issue < ActiveRecord::Base
# self.table_name = 'issues'
#
# include EachBatch
#
# def self.to_migrate
# where('closed_at IS NOT NULL')
# end
# end
#
# change_column_type_using_background_migration(
# Issue.to_migrate,
# :closed_at,
# :datetime_with_timezone
# )
#
# Reverting a migration like this is done exactly the same way, just with
# a different type to migrate to (e.g. `:datetime` in the above example).
#
# relation - An ActiveRecord relation to use for scheduling jobs and
# figuring out what table we're modifying. This relation _must_
# have the EachBatch module included.
#
# column - The name of the column for which the type will be changed.
#
# new_type - The new type of the column.
#
# batch_size - The number of rows to schedule in a single background
# migration.
#
# interval - The time interval between every background migration.
def change_column_type_using_background_migration(
relation,
column,
new_type,
batch_size: 10_000,
interval: 10.minutes
)
unless relation.model < EachBatch
raise TypeError, 'The relation must include the EachBatch module'
end
temp_column = "#{column}_for_type_change"
table = relation.table_name
max_index = 0
add_column(table, temp_column, new_type)
install_rename_triggers(table, column, temp_column)
# Schedule the jobs that will copy the data from the old column to the
# new one. Rows with NULL values in our source column are skipped since
# the target column is already NULL at this point.
relation.where.not(column => nil).each_batch(of: batch_size) do |batch, index|
start_id, end_id = batch.pluck('MIN(id), MAX(id)').first
max_index = index
migrate_in(
index * interval,
'CopyColumn',
[table, column, temp_column, start_id, end_id]
)
end
# Schedule the renaming of the column to happen (initially) 1 hour after
# the last batch finished.
migrate_in(
(max_index * interval) + 1.hour,
'CleanupConcurrentTypeChange',
[table, column, temp_column]
)
if perform_background_migration_inline?
# To ensure the schema is up to date immediately we perform the
# migration inline in dev / test environments.
Gitlab::BackgroundMigration.steal('CopyColumn')
Gitlab::BackgroundMigration.steal('CleanupConcurrentTypeChange')
end
end
# Renames a column using a background migration.
#
# Because this method uses a background migration it's more suitable for
# large tables. For small tables it's better to use
# `rename_column_concurrently` since it can complete its work in a much
# shorter amount of time and doesn't rely on Sidekiq.
#
# Example usage:
#
# rename_column_using_background_migration(
# :users,
# :feed_token,
# :rss_token
# )
#
# table - The name of the database table containing the column.
#
# old - The old column name.
#
# new - The new column name.
#
# type - The type of the new column. If no type is given the old column's
# type is used.
#
# batch_size - The number of rows to schedule in a single background
# migration.
#
# interval - The time interval between every background migration.
def rename_column_using_background_migration(
table,
old_column,
new_column,
type: nil,
batch_size: 10_000,
interval: 10.minutes
)
check_trigger_permissions!(table)
old_col = column_for(table, old_column)
new_type = type || old_col.type
max_index = 0
add_column(table, new_column, new_type,
limit: old_col.limit,
precision: old_col.precision,
scale: old_col.scale)
# We set the default value _after_ adding the column so we don't end up
# updating any existing data with the default value. This isn't
# necessary since we copy over old values further down.
change_column_default(table, new_column, old_col.default) if old_col.default
install_rename_triggers(table, old_column, new_column)
model = Class.new(ActiveRecord::Base) do
self.table_name = table
include ::EachBatch
end
# Schedule the jobs that will copy the data from the old column to the
# new one. Rows with NULL values in our source column are skipped since
# the target column is already NULL at this point.
model.where.not(old_column => nil).each_batch(of: batch_size) do |batch, index|
start_id, end_id = batch.pluck('MIN(id), MAX(id)').first
max_index = index
migrate_in(
index * interval,
'CopyColumn',
[table, old_column, new_column, start_id, end_id]
)
end
# Schedule the renaming of the column to happen (initially) 1 hour after
# the last batch finished.
migrate_in(
(max_index * interval) + 1.hour,
'CleanupConcurrentRename',
[table, old_column, new_column]
)
if perform_background_migration_inline?
# To ensure the schema is up to date immediately we perform the
# migration inline in dev / test environments.
Gitlab::BackgroundMigration.steal('CopyColumn')
Gitlab::BackgroundMigration.steal('CleanupConcurrentRename')
end
end
# Performs a concurrent column rename when using PostgreSQL.
def install_rename_triggers_for_postgresql(trigger, table, old, new)
execute <<-EOF.strip_heredoc
CREATE OR REPLACE FUNCTION #{trigger}()
RETURNS trigger AS
$BODY$
BEGIN
NEW.#{new} := NEW.#{old};
RETURN NEW;
END;
$BODY$
LANGUAGE 'plpgsql'
VOLATILE
EOF
execute <<-EOF.strip_heredoc
DROP TRIGGER IF EXISTS #{trigger}
ON #{table}
EOF
execute <<-EOF.strip_heredoc
CREATE TRIGGER #{trigger}
BEFORE INSERT OR UPDATE
ON #{table}
FOR EACH ROW
EXECUTE FUNCTION #{trigger}()
EOF
end
# Removes the triggers used for renaming a PostgreSQL column concurrently.
def remove_rename_triggers_for_postgresql(table, trigger)
execute("DROP TRIGGER IF EXISTS #{trigger} ON #{table}")
execute("DROP FUNCTION IF EXISTS #{trigger}()")
end
# Returns the (base) name to use for triggers when renaming columns.
def rename_trigger_name(table, old, new)
'trigger_' + Digest::SHA256.hexdigest("#{table}_#{old}_#{new}").first(12)
end
# Returns an Array containing the indexes for the given column
def indexes_for(table, column)
column = column.to_s
indexes(table).select { |index| index.columns.include?(column) }
end
# Returns an Array containing the foreign keys for the given column.
def foreign_keys_for(table, column)
column = column.to_s
foreign_keys(table).select { |fk| fk.column == column }
end
# Copies all indexes for the old column to a new column.
#
# table - The table containing the columns and indexes.
# old - The old column.
# new - The new column.
def copy_indexes(table, old, new)
old = old.to_s
new = new.to_s
indexes_for(table, old).each do |index|
new_columns = index.columns.map do |column|
column == old ? new : column
end
# This is necessary as we can't properly rename indexes such as
# "ci_taggings_idx".
unless index.name.include?(old)
raise "The index #{index.name} can not be copied as it does not "\
"mention the old column. You have to rename this index manually first."
end
name = index.name.gsub(old, new)
options = {
unique: index.unique,
name: name,
length: index.lengths,
order: index.orders
}
options[:using] = index.using if index.using
options[:where] = index.where if index.where
unless index.opclasses.blank?
opclasses = index.opclasses.dup
# Copy the operator classes for the old column (if any) to the new
# column.
opclasses[new] = opclasses.delete(old) if opclasses[old]
options[:opclasses] = opclasses
end
add_concurrent_index(table, new_columns, options)
end
end
# Copies all foreign keys for the old column to the new column.
#
# table - The table containing the columns and indexes.
# old - The old column.
# new - The new column.
def copy_foreign_keys(table, old, new)
foreign_keys_for(table, old).each do |fk|
add_concurrent_foreign_key(fk.from_table,
fk.to_table,
column: new,
on_delete: fk.on_delete)
end
end
# Returns the column for the given table and column name.
def column_for(table, name)
name = name.to_s
column = columns(table).find { |column| column.name == name }
raise(missing_schema_object_message(table, "column", name)) if column.nil?
column
end
# This will replace the first occurrence of a string in a column with
# the replacement using `regexp_replace`
def replace_sql(column, pattern, replacement)
quoted_pattern = Arel::Nodes::Quoted.new(pattern.to_s)
quoted_replacement = Arel::Nodes::Quoted.new(replacement.to_s)
replace = Arel::Nodes::NamedFunction.new(
"regexp_replace", [column, quoted_pattern, quoted_replacement]
)
Arel::Nodes::SqlLiteral.new(replace.to_sql)
end
def remove_foreign_key_if_exists(*args)
if foreign_key_exists?(*args)
remove_foreign_key(*args)
end
end
def remove_foreign_key_without_error(*args)
remove_foreign_key(*args)
rescue ArgumentError
end
def sidekiq_queue_migrate(queue_from, to:)
while sidekiq_queue_length(queue_from) > 0
Sidekiq.redis do |conn|
conn.rpoplpush "queue:#{queue_from}", "queue:#{to}"
end
end
end
def sidekiq_queue_length(queue_name)
Sidekiq.redis do |conn|
conn.llen("queue:#{queue_name}")
end
end
def check_trigger_permissions!(table)
unless Grant.create_and_execute_trigger?(table)
dbname = Database.database_name
user = Database.username
raise <<-EOF
Your database user is not allowed to create, drop, or execute triggers on the
table #{table}.
If you are using PostgreSQL you can solve this by logging in to the GitLab
database (#{dbname}) using a super user and running:
ALTER #{user} WITH SUPERUSER
This query will grant the user super user permissions, ensuring you don't run
into similar problems in the future (e.g. when new tables are created).
EOF
end
end
# Fetches indexes on a column by name for postgres.
#
# This will include indexes using an expression on the column, for example:
# `CREATE INDEX CONCURRENTLY index_name ON table (LOWER(column));`
#
# We can remove this when upgrading to Rails 5 with an updated `index_exists?`:
# - https://github.com/rails/rails/commit/edc2b7718725016e988089b5fb6d6fb9d6e16882
#
# Or this can be removed when we no longer support postgres < 9.5, so we
# can use `CREATE INDEX IF NOT EXISTS`.
def index_exists_by_name?(table, index)
# We can't fall back to the normal `index_exists?` method because that
# does not find indexes without passing a column name.
if indexes(table).map(&:name).include?(index.to_s)
true
else
postgres_exists_by_name?(table, index)
end
end
def postgres_exists_by_name?(table, name)
index_sql = <<~SQL
SELECT COUNT(*)
FROM pg_index
JOIN pg_class i ON (indexrelid=i.oid)
JOIN pg_class t ON (indrelid=t.oid)
WHERE i.relname = '#{name}' AND t.relname = '#{table}'
SQL
connection.select_value(index_sql).to_i > 0
end
def create_or_update_plan_limit(limit_name, plan_name, limit_value)
limit_name_quoted = quote_column_name(limit_name)
plan_name_quoted = quote(plan_name)
limit_value_quoted = quote(limit_value)
execute <<~SQL
INSERT INTO plan_limits (plan_id, #{limit_name_quoted})
SELECT id, #{limit_value_quoted} FROM plans WHERE name = #{plan_name_quoted} LIMIT 1
ON CONFLICT (plan_id) DO UPDATE SET #{limit_name_quoted} = EXCLUDED.#{limit_name_quoted};
SQL
end
# Note this should only be used with very small tables
def backfill_iids(table)
sql = <<-END
UPDATE #{table}
SET iid = #{table}_with_calculated_iid.iid_num
FROM (
SELECT id, ROW_NUMBER() OVER (PARTITION BY project_id ORDER BY id ASC) AS iid_num FROM #{table}
) AS #{table}_with_calculated_iid
WHERE #{table}.id = #{table}_with_calculated_iid.id
END
execute(sql)
end
# Returns the name for a check constraint
#
# type:
# - Any value, as long as it is unique
# - Constraint names are unique per table in Postgres, and, additionally,
# we can have multiple check constraints over a column
# So we use the (table, column, type) triplet as a unique name
# - e.g. we use 'max_length' when adding checks for text limits
# or 'not_null' when adding a NOT NULL constraint
#
def check_constraint_name(table, column, type)
identifier = "#{table}_#{column}_check_#{type}"
# Check concurrent_foreign_key_name() for info on why we use a hash
hashed_identifier = Digest::SHA256.hexdigest(identifier).first(10)
"check_#{hashed_identifier}"
end
def check_constraint_exists?(table, constraint_name)
# Constraint names are unique per table in Postgres, not per schema
# Two tables can have constraints with the same name, so we filter by
# the table name in addition to using the constraint_name
check_sql = <<~SQL
SELECT COUNT(*)
FROM pg_constraint
JOIN pg_class ON pg_constraint.conrelid = pg_class.oid
WHERE pg_constraint.contype = 'c'
AND pg_constraint.conname = '#{constraint_name}'
AND pg_class.relname = '#{table}'
SQL
connection.select_value(check_sql) > 0
end
# Adds a check constraint to a table
#
# This method is the generic helper for adding any check constraint
# More specialized helpers may use it (e.g. add_text_limit or add_not_null)
#
# This method only requires minimal locking:
# - The constraint is added using NOT VALID
# This allows us to add the check constraint without validating it
# - The check will be enforced for new data (inserts) coming in
# - If `validate: true` the constraint is also validated
# Otherwise, validate_check_constraint() can be used at a later stage
# - Check comments on add_concurrent_foreign_key for more info
#
# table - The table the constraint will be added to
# check - The check clause to add
# e.g. 'char_length(name) <= 5' or 'store IS NOT NULL'
# constraint_name - The name of the check constraint (otherwise auto-generated)
# Should be unique per table (not per column)
# validate - Whether to validate the constraint in this call
#
def add_check_constraint(table, check, constraint_name, validate: true)
validate_check_constraint_name!(constraint_name)
# Transactions would result in ALTER TABLE locks being held for the
# duration of the transaction, defeating the purpose of this method.
if transaction_open?
raise 'add_check_constraint can not be run inside a transaction'
end
if check_constraint_exists?(table, constraint_name)
warning_message = <<~MESSAGE
Check constraint was not created because it exists already
(this may be due to an aborted migration or similar)
table: #{table}, check: #{check}, constraint name: #{constraint_name}
MESSAGE
Gitlab::AppLogger.warn warning_message
else
# Only add the constraint without validating it
# Even though it is fast, ADD CONSTRAINT requires an EXCLUSIVE lock
# Use with_lock_retries to make sure that this operation
# will not timeout on tables accessed by many processes
with_lock_retries do
execute <<-EOF.strip_heredoc
ALTER TABLE #{table}
ADD CONSTRAINT #{constraint_name}
CHECK ( #{check} )
NOT VALID;
EOF
end
end
if validate
validate_check_constraint(table, constraint_name)
end
end
def validate_check_constraint(table, constraint_name)
validate_check_constraint_name!(constraint_name)
unless check_constraint_exists?(table, constraint_name)
raise missing_schema_object_message(table, "check constraint", constraint_name)
end
disable_statement_timeout do
# VALIDATE CONSTRAINT only requires a SHARE UPDATE EXCLUSIVE LOCK
# It only conflicts with other validations and creating indexes
execute("ALTER TABLE #{table} VALIDATE CONSTRAINT #{constraint_name};")
end
end
def remove_check_constraint(table, constraint_name)
validate_check_constraint_name!(constraint_name)
# DROP CONSTRAINT requires an EXCLUSIVE lock
# Use with_lock_retries to make sure that this will not timeout
with_lock_retries do
execute <<-EOF.strip_heredoc
ALTER TABLE #{table}
DROP CONSTRAINT IF EXISTS #{constraint_name}
EOF
end
end
# Migration Helpers for adding limit to text columns
def add_text_limit(table, column, limit, constraint_name: nil, validate: true)
add_check_constraint(
table,
"char_length(#{column}) <= #{limit}",
text_limit_name(table, column, name: constraint_name),
validate: validate
)
end
def validate_text_limit(table, column, constraint_name: nil)
validate_check_constraint(table, text_limit_name(table, column, name: constraint_name))
end
def remove_text_limit(table, column, constraint_name: nil)
remove_check_constraint(table, text_limit_name(table, column, name: constraint_name))
end
def check_text_limit_exists?(table, column, constraint_name: nil)
check_constraint_exists?(table, text_limit_name(table, column, name: constraint_name))
end
# Migration Helpers for managing not null constraints
def add_not_null_constraint(table, column, constraint_name: nil, validate: true)
if column_is_nullable?(table, column)
add_check_constraint(
table,
"#{column} IS NOT NULL",
not_null_constraint_name(table, column, name: constraint_name),
validate: validate
)
else
warning_message = <<~MESSAGE
NOT NULL check constraint was not created:
column #{table}.#{column} is already defined as `NOT NULL`
MESSAGE
Gitlab::AppLogger.warn warning_message
end
end
def validate_not_null_constraint(table, column, constraint_name: nil)
validate_check_constraint(
table,
not_null_constraint_name(table, column, name: constraint_name)
)
end
def remove_not_null_constraint(table, column, constraint_name: nil)
remove_check_constraint(
table,
not_null_constraint_name(table, column, name: constraint_name)
)
end
def check_not_null_constraint_exists?(table, column, constraint_name: nil)
check_constraint_exists?(
table,
not_null_constraint_name(table, column, name: constraint_name)
)
end
def create_extension(extension)
execute('CREATE EXTENSION IF NOT EXISTS %s' % extension)
rescue ActiveRecord::StatementInvalid => e
dbname = Database.database_name
user = Database.username
warn(<<~MSG) if e.to_s =~ /permission denied/
GitLab requires the PostgreSQL extension '#{extension}' installed in database '#{dbname}', but
the database user is not allowed to install the extension.
You can either install the extension manually using a database superuser:
CREATE EXTENSION IF NOT EXISTS #{extension}
Or, you can solve this by logging in to the GitLab
database (#{dbname}) using a superuser and running:
ALTER #{user} WITH SUPERUSER
This query will grant the user superuser permissions, ensuring any database extensions
can be installed through migrations.
For more information, refer to https://docs.gitlab.com/ee/install/postgresql_extensions.html.
MSG
raise
end
def drop_extension(extension)
execute('DROP EXTENSION IF EXISTS %s' % extension)
rescue ActiveRecord::StatementInvalid => e
dbname = Database.database_name
user = Database.username
warn(<<~MSG) if e.to_s =~ /permission denied/
This migration attempts to drop the PostgreSQL extension '#{extension}'
installed in database '#{dbname}', but the database user is not allowed
to drop the extension.
You can either drop the extension manually using a database superuser:
DROP EXTENSION IF EXISTS #{extension}
Or, you can solve this by logging in to the GitLab
database (#{dbname}) using a superuser and running:
ALTER #{user} WITH SUPERUSER
This query will grant the user superuser permissions, ensuring any database extensions
can be dropped through migrations.
For more information, refer to https://docs.gitlab.com/ee/install/postgresql_extensions.html.
MSG
raise
end
private
def validate_check_constraint_name!(constraint_name)
if constraint_name.to_s.length > MAX_IDENTIFIER_NAME_LENGTH
raise "The maximum allowed constraint name is #{MAX_IDENTIFIER_NAME_LENGTH} characters"
end
end
def statement_timeout_disabled?
# This is a string of the form "100ms" or "0" when disabled
connection.select_value('SHOW statement_timeout') == "0"
end
def column_is_nullable?(table, column)
# Check if table.column has not been defined with NOT NULL
check_sql = <<~SQL
SELECT c.is_nullable
FROM information_schema.columns c
WHERE c.table_name = '#{table}'
AND c.column_name = '#{column}'
SQL
connection.select_value(check_sql) == 'YES'
end
def text_limit_name(table, column, name: nil)
name.presence || check_constraint_name(table, column, 'max_length')
end
def not_null_constraint_name(table, column, name: nil)
name.presence || check_constraint_name(table, column, 'not_null')
end
def missing_schema_object_message(table, type, name)
<<~MESSAGE
Could not find #{type} "#{name}" on table "#{table}" which was referenced during the migration.
This issue could be caused by the database schema straying from the expected state.
To resolve this issue, please verify:
1. all previous migrations have completed
2. the database objects used in this migration match the Rails definition in schema.rb or structure.sql
MESSAGE
end
def tables_match?(target_table, foreign_key_table)
target_table.blank? || foreign_key_table == target_table
end
def options_match?(foreign_key_options, options)
options.all? { |k, v| foreign_key_options[k].to_s == v.to_s }
end
def on_delete_statement(on_delete)
return '' if on_delete.blank?
return 'ON DELETE SET NULL' if on_delete == :nullify
"ON DELETE #{on_delete.upcase}"
end
def create_column_from(table, old, new, type: nil, batch_column_name: :id, type_cast_function: nil)
old_col = column_for(table, old)
new_type = type || old_col.type
add_column(table, new, new_type,
limit: old_col.limit,
precision: old_col.precision,
scale: old_col.scale)
# We set the default value _after_ adding the column so we don't end up
# updating any existing data with the default value. This isn't
# necessary since we copy over old values further down.
change_column_default(table, new, old_col.default) unless old_col.default.nil?
old_value = Arel::Table.new(table)[old]
if type_cast_function.present?
old_value = Arel::Nodes::NamedFunction.new(type_cast_function, [old_value])
end
update_column_in_batches(table, new, old_value, batch_column_name: batch_column_name)
add_not_null_constraint(table, new) unless old_col.null
copy_indexes(table, old, new)
copy_foreign_keys(table, old, new)
end
def validate_timestamp_column_name!(column_name)
return if PERMITTED_TIMESTAMP_COLUMNS.member?(column_name)
raise <<~MESSAGE
Illegal timestamp column name! Got #{column_name}.
Must be one of: #{PERMITTED_TIMESTAMP_COLUMNS.to_a}
MESSAGE
end
def validate_not_in_transaction!(method_name, modifier = nil)
return unless transaction_open?
raise <<~ERROR
#{["`#{method_name}`", modifier].compact.join(' ')} cannot be run inside a transaction.
You can disable transactions by calling `disable_ddl_transaction!` in the body of
your migration class
ERROR
end
end
end
end
|