-
Notifications
You must be signed in to change notification settings - Fork 1.7k
/
srv0srv.cc
3733 lines (2965 loc) · 111 KB
/
srv0srv.cc
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
/*****************************************************************************
Copyright (c) 1995, 2016, Oracle and/or its affiliates. All Rights Reserved.
Copyright (c) 2008, 2009 Google Inc.
Copyright (c) 2009, Percona Inc.
Copyright (c) 2013, 2017, MariaDB Corporation Ab. All Rights Reserved.
Portions of this file contain modifications contributed and copyrighted by
Google, Inc. Those modifications are gratefully acknowledged and are described
briefly in the InnoDB documentation. The contributions by Google are
incorporated with their permission, and subject to the conditions contained in
the file COPYING.Google.
Portions of this file contain modifications contributed and copyrighted
by Percona Inc.. Those modifications are
gratefully acknowledged and are described briefly in the InnoDB
documentation. The contributions by Percona Inc. are incorporated with
their permission, and subject to the conditions contained in the file
COPYING.Percona.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
Foundation; version 2 of the License.
This program is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
this program; if not, write to the Free Software Foundation, Inc.,
51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
/**************************************************//**
@file srv/srv0srv.cc
The database server main program
Created 10/8/1995 Heikki Tuuri
*******************************************************/
/* Dummy comment */
#include "srv0srv.h"
#include "ut0mem.h"
#include "ut0ut.h"
#include "os0proc.h"
#include "mem0mem.h"
#include "mem0pool.h"
#include "sync0sync.h"
#include "que0que.h"
#include "log0online.h"
#include "log0recv.h"
#include "pars0pars.h"
#include "usr0sess.h"
#include "lock0lock.h"
#include "trx0purge.h"
#include "ibuf0ibuf.h"
#include "buf0flu.h"
#include "buf0lru.h"
#include "btr0sea.h"
#include "dict0load.h"
#include "dict0boot.h"
#include "srv0start.h"
#include "row0mysql.h"
#include "row0log.h"
#include "ha_prototypes.h"
#include "trx0i_s.h"
#include "os0sync.h" /* for HAVE_ATOMIC_BUILTINS */
#include "srv0mon.h"
#include "ut0crc32.h"
#include "os0file.h"
#include "btr0defragment.h"
#include "mysql/plugin.h"
#include "mysql/service_thd_wait.h"
#include "fil0fil.h"
#include "fil0crypt.h"
#include "fil0pagecompress.h"
#include <my_rdtsc.h>
#include "btr0scrub.h"
/* prototypes of new functions added to ha_innodb.cc for kill_idle_transaction */
ibool innobase_thd_is_idle(const void* thd);
ib_int64_t innobase_thd_get_start_time(const void* thd);
void innobase_thd_kill(ulong thd_id);
ulong innobase_thd_get_thread_id(const void* thd);
/* prototypes for new functions added to ha_innodb.cc */
ibool innobase_get_slow_log();
#ifdef WITH_WSREP
extern int wsrep_debug;
extern int wsrep_trx_is_aborting(void *thd_ptr);
#endif
/* The following counter is incremented whenever there is some user activity
in the server */
UNIV_INTERN ulint srv_activity_count = 0;
/* The following is the maximum allowed duration of a lock wait. */
UNIV_INTERN ulong srv_fatal_semaphore_wait_threshold = DEFAULT_SRV_FATAL_SEMAPHORE_TIMEOUT;
/**/
UNIV_INTERN long long srv_kill_idle_transaction = 0;
/* How much data manipulation language (DML) statements need to be delayed,
in microseconds, in order to reduce the lagging of the purge thread. */
UNIV_INTERN ulint srv_dml_needed_delay = 0;
UNIV_INTERN bool srv_monitor_active;
UNIV_INTERN bool srv_error_monitor_active;
UNIV_INTERN bool srv_buf_dump_thread_active;
UNIV_INTERN bool srv_dict_stats_thread_active;
UNIV_INTERN my_bool srv_scrub_log;
UNIV_INTERN const char* srv_main_thread_op_info = "";
/** Prefix used by MySQL to indicate pre-5.1 table name encoding */
const char srv_mysql50_table_name_prefix[10] = "#mysql50#";
/* Server parameters which are read from the initfile */
/* The following three are dir paths which are catenated before file
names, where the file name itself may also contain a path */
UNIV_INTERN char* srv_data_home = NULL;
/** Rollback files directory, can be absolute. */
UNIV_INTERN char* srv_undo_dir = NULL;
/** The number of tablespaces to use for rollback segments. */
UNIV_INTERN ulong srv_undo_tablespaces = 8;
/** The number of UNDO tablespaces that are open and ready to use. */
UNIV_INTERN ulint srv_undo_tablespaces_open = 8;
/* The number of rollback segments to use */
UNIV_INTERN ulong srv_undo_logs = 1;
#ifdef UNIV_LOG_ARCHIVE
UNIV_INTERN char* srv_arch_dir = NULL;
UNIV_INTERN ulong srv_log_arch_expire_sec = 0;
#endif /* UNIV_LOG_ARCHIVE */
/** Set if InnoDB must operate in read-only mode. We don't do any
recovery and open all tables in RO mode instead of RW mode. We don't
sync the max trx id to disk either. */
UNIV_INTERN my_bool srv_read_only_mode;
/** store to its own file each table created by an user; data
dictionary tables are in the system tablespace 0 */
UNIV_INTERN my_bool srv_file_per_table;
/** The file format to use on new *.ibd files. */
UNIV_INTERN ulint srv_file_format = 0;
/** Whether to check file format during startup. A value of
UNIV_FORMAT_MAX + 1 means no checking ie. FALSE. The default is to
set it to the highest format we support. */
UNIV_INTERN ulint srv_max_file_format_at_startup = UNIV_FORMAT_MAX;
/** Set if InnoDB operates in read-only mode or innodb-force-recovery
is greater than SRV_FORCE_NO_TRX_UNDO. */
UNIV_INTERN my_bool high_level_read_only;
#if UNIV_FORMAT_A
# error "UNIV_FORMAT_A must be 0!"
#endif
/** Place locks to records only i.e. do not use next-key locking except
on duplicate key checking and foreign key checking */
UNIV_INTERN ibool srv_locks_unsafe_for_binlog = FALSE;
/** Sort buffer size in index creation */
UNIV_INTERN ulong srv_sort_buf_size = 1048576;
/** Maximum modification log file size for online index creation */
UNIV_INTERN unsigned long long srv_online_max_size;
/* If this flag is TRUE, then we will use the native aio of the
OS (provided we compiled Innobase with it in), otherwise we will
use simulated aio we build below with threads.
Currently we support native aio on windows and linux */
UNIV_INTERN my_bool srv_use_native_aio = TRUE;
UNIV_INTERN my_bool srv_numa_interleave = FALSE;
/* Default compression level if page compression is used and no compression
level is set for the table*/
UNIV_INTERN long srv_compress_zlib_level = 6;
/* If this flag is TRUE, then we will use fallocate(PUCH_HOLE)
to the pages */
UNIV_INTERN my_bool srv_use_trim = FALSE;
/* If this flag is TRUE, then we will use posix fallocate for file extentsion */
UNIV_INTERN my_bool srv_use_posix_fallocate = FALSE;
/* If this flag is TRUE, then we disable doublewrite buffer */
UNIV_INTERN my_bool srv_use_atomic_writes = FALSE;
/* If this flag IS TRUE, then we use this algorithm for page compressing the pages */
UNIV_INTERN ulong innodb_compression_algorithm = PAGE_ZLIB_ALGORITHM;
/* Number of threads used for multi-threaded flush */
UNIV_INTERN long srv_mtflush_threads = MTFLUSH_DEFAULT_WORKER;
/* If this flag is TRUE, then we will use multi threaded flush. */
UNIV_INTERN my_bool srv_use_mtflush = FALSE;
#ifdef __WIN__
/* Windows native condition variables. We use runtime loading / function
pointers, because they are not available on Windows Server 2003 and
Windows XP/2000.
We use condition for events on Windows if possible, even if os_event
resembles Windows kernel event object well API-wise. The reason is
performance, kernel objects are heavyweights and WaitForSingleObject() is a
performance killer causing calling thread to context switch. Besides, Innodb
is preallocating large number (often millions) of os_events. With kernel event
objects it takes a big chunk out of non-paged pool, which is better suited
for tasks like IO than for storing idle event objects. */
UNIV_INTERN ibool srv_use_native_conditions = FALSE;
#endif /* __WIN__ */
UNIV_INTERN ulint srv_n_data_files = 0;
UNIV_INTERN char** srv_data_file_names = NULL;
/* size in database pages */
UNIV_INTERN ulint* srv_data_file_sizes = NULL;
/** Whether the redo log tracking is currently enabled. Note that it is
possible for the log tracker thread to be running and the tracking to be
disabled */
UNIV_INTERN my_bool srv_track_changed_pages = FALSE;
UNIV_INTERN ulonglong srv_max_bitmap_file_size = 100 * 1024 * 1024;
UNIV_INTERN ulonglong srv_max_changed_pages = 0;
/** When TRUE, fake change transcations take S rather than X row locks.
When FALSE, row locks are not taken at all. */
UNIV_INTERN my_bool srv_fake_changes_locks = TRUE;
/* if TRUE, then we auto-extend the last data file */
UNIV_INTERN ibool srv_auto_extend_last_data_file = FALSE;
/* if != 0, this tells the max size auto-extending may increase the
last data file size */
UNIV_INTERN ulint srv_last_file_size_max = 0;
/* If the last data file is auto-extended, we add this
many pages to it at a time */
UNIV_INTERN ulong srv_auto_extend_increment = 8;
UNIV_INTERN ulint* srv_data_file_is_raw_partition = NULL;
/* If the following is TRUE we do not allow inserts etc. This protects
the user from forgetting the 'newraw' keyword to my.cnf */
UNIV_INTERN ibool srv_created_new_raw = FALSE;
UNIV_INTERN char* srv_log_group_home_dir = NULL;
UNIV_INTERN ulong srv_n_log_files = SRV_N_LOG_FILES_MAX;
/* size in database pages */
UNIV_INTERN ib_uint64_t srv_log_file_size = IB_UINT64_MAX;
UNIV_INTERN ib_uint64_t srv_log_file_size_requested;
/* size in database pages */
UNIV_INTERN ulint srv_log_buffer_size = ULINT_MAX;
UNIV_INTERN uint srv_flush_log_at_timeout = 1;
UNIV_INTERN ulong srv_page_size = UNIV_PAGE_SIZE_DEF;
UNIV_INTERN ulong srv_page_size_shift = UNIV_PAGE_SIZE_SHIFT_DEF;
UNIV_INTERN char srv_use_global_flush_log_at_trx_commit = TRUE;
/* Try to flush dirty pages so as to avoid IO bursts at
the checkpoints. */
UNIV_INTERN char srv_adaptive_flushing = TRUE;
UNIV_INTERN ulong srv_show_locks_held = 10;
UNIV_INTERN ulong srv_show_verbose_locks = 0;
/** Maximum number of times allowed to conditionally acquire
mutex before switching to blocking wait on the mutex */
#define MAX_MUTEX_NOWAIT 20
/** Check whether the number of failed nonblocking mutex
acquisition attempts exceeds maximum allowed value. If so,
srv_printf_innodb_monitor() will request mutex acquisition
with mutex_enter(), which will wait until it gets the mutex. */
#define MUTEX_NOWAIT(mutex_skipped) ((mutex_skipped) < MAX_MUTEX_NOWAIT)
#ifdef WITH_INNODB_DISALLOW_WRITES
UNIV_INTERN os_event_t srv_allow_writes_event;
#endif /* WITH_INNODB_DISALLOW_WRITES */
/** The sort order table of the MySQL latin1_swedish_ci character set
collation */
UNIV_INTERN const byte* srv_latin1_ordering;
/* use os/external memory allocator */
UNIV_INTERN my_bool srv_use_sys_malloc = TRUE;
/* requested size in kilobytes */
UNIV_INTERN ulint srv_buf_pool_size = ULINT_MAX;
/* requested number of buffer pool instances */
UNIV_INTERN ulint srv_buf_pool_instances = 1;
/* number of locks to protect buf_pool->page_hash */
UNIV_INTERN ulong srv_n_page_hash_locks = 16;
/** Scan depth for LRU flush batch i.e.: number of blocks scanned*/
UNIV_INTERN ulong srv_LRU_scan_depth = 1024;
/** whether or not to flush neighbors of a block */
UNIV_INTERN ulong srv_flush_neighbors = 1;
/* previously requested size */
UNIV_INTERN ulint srv_buf_pool_old_size;
/* current size in kilobytes */
UNIV_INTERN ulint srv_buf_pool_curr_size = 0;
/* dump that may % of each buffer pool during BP dump */
UNIV_INTERN ulong srv_buf_pool_dump_pct;
/* size in bytes */
UNIV_INTERN ulint srv_mem_pool_size = ULINT_MAX;
UNIV_INTERN ulint srv_lock_table_size = ULINT_MAX;
/* Defragmentation */
UNIV_INTERN my_bool srv_defragment = FALSE;
UNIV_INTERN uint srv_defragment_n_pages = 7;
UNIV_INTERN uint srv_defragment_stats_accuracy = 0;
UNIV_INTERN uint srv_defragment_fill_factor_n_recs = 20;
UNIV_INTERN double srv_defragment_fill_factor = 0.9;
UNIV_INTERN uint srv_defragment_frequency =
SRV_DEFRAGMENT_FREQUENCY_DEFAULT;
UNIV_INTERN ulonglong srv_defragment_interval = 0;
/** Query thread preflush algorithm */
UNIV_INTERN ulong srv_foreground_preflush
= SRV_FOREGROUND_PREFLUSH_EXP_BACKOFF;
/** The maximum time limit for a single LRU tail flush iteration by the page
cleaner thread */
UNIV_INTERN ulint srv_cleaner_max_lru_time = 1000;
/** The maximum time limit for a single flush list flush iteration by the page
cleaner thread */
UNIV_INTERN ulint srv_cleaner_max_flush_time = 1000;
/** Page cleaner flush list flush batches are further divided into this chunk
size */
UNIV_INTERN ulint srv_cleaner_flush_chunk_size = 100;
/** Page cleaner LRU list flush batches are further divided into this chunk
size */
UNIV_INTERN ulint srv_cleaner_lru_chunk_size = 100;
/** If free list length is lower than this percentage of srv_LRU_scan_depth,
page cleaner LRU flushes will issue flush batches to the same instance in a
row */
UNIV_INTERN ulint srv_cleaner_free_list_lwm = 10;
/** If TRUE, page cleaner heuristics use evicted instead of flushed page counts
for its heuristics */
UNIV_INTERN my_bool srv_cleaner_eviction_factor = FALSE;
/** Page cleaner LSN age factor formula option */
UNIV_INTERN ulong srv_cleaner_lsn_age_factor
= SRV_CLEANER_LSN_AGE_FACTOR_HIGH_CHECKPOINT;
/** Empty free list for a query thread handling algorithm option */
UNIV_INTERN ulong srv_empty_free_list_algorithm
= SRV_EMPTY_FREE_LIST_BACKOFF;
UNIV_INTERN ulong srv_idle_flush_pct = 100;
/* This parameter is deprecated. Use srv_n_io_[read|write]_threads
instead. */
UNIV_INTERN ulint srv_n_file_io_threads = ULINT_MAX;
UNIV_INTERN ulint srv_n_read_io_threads = ULINT_MAX;
UNIV_INTERN ulint srv_n_write_io_threads = ULINT_MAX;
/* Switch to enable random read ahead. */
UNIV_INTERN my_bool srv_random_read_ahead = FALSE;
/* The log block size */
UNIV_INTERN ulint srv_log_block_size = 0;
/* User settable value of the number of pages that must be present
in the buffer cache and accessed sequentially for InnoDB to trigger a
readahead request. */
UNIV_INTERN ulong srv_read_ahead_threshold = 56;
#ifdef UNIV_LOG_ARCHIVE
UNIV_INTERN ibool srv_log_archive_on = FALSE;
#endif /* UNIV_LOG_ARCHIVE */
/* This parameter is used to throttle the number of insert buffers that are
merged in a batch. By increasing this parameter on a faster disk you can
possibly reduce the number of I/O operations performed to complete the
merge operation. The value of this parameter is used as is by the
background loop when the system is idle (low load), on a busy system
the parameter is scaled down by a factor of 4, this is to avoid putting
a heavier load on the I/O sub system. */
UNIV_INTERN ulong srv_insert_buffer_batch_size = 20;
UNIV_INTERN char* srv_file_flush_method_str = NULL;
UNIV_INTERN ulint srv_unix_file_flush_method = SRV_UNIX_FSYNC;
UNIV_INTERN ulint srv_win_file_flush_method = SRV_WIN_IO_UNBUFFERED;
UNIV_INTERN ulint srv_max_n_open_files = 300;
/* Number of IO operations per second the server can do */
UNIV_INTERN ulong srv_io_capacity = 200;
UNIV_INTERN ulong srv_max_io_capacity = 400;
/* The InnoDB main thread tries to keep the ratio of modified pages
in the buffer pool to all database pages in the buffer pool smaller than
the following number. But it is not guaranteed that the value stays below
that during a time of heavy update/insert activity. */
UNIV_INTERN double srv_max_buf_pool_modified_pct = 75.0;
UNIV_INTERN double srv_max_dirty_pages_pct_lwm = 50.0;
/* This is the percentage of log capacity at which adaptive flushing,
if enabled, will kick in. */
UNIV_INTERN double srv_adaptive_flushing_lwm = 10.0;
/* Number of iterations over which adaptive flushing is averaged. */
UNIV_INTERN ulong srv_flushing_avg_loops = 30;
/* The tid of the cleaner thread */
UNIV_INTERN os_tid_t srv_cleaner_tid;
/* The tid of the LRU manager thread */
UNIV_INTERN os_tid_t srv_lru_manager_tid;
/* The tids of the purge threads */
UNIV_INTERN os_tid_t srv_purge_tids[SRV_MAX_N_PURGE_THREADS];
/* The tids of the I/O threads */
UNIV_INTERN os_tid_t srv_io_tids[SRV_MAX_N_IO_THREADS];
/* The tid of the master thread */
UNIV_INTERN os_tid_t srv_master_tid;
/* The relative scheduling priority of the cleaner and LRU manager threads */
UNIV_INTERN ulint srv_sched_priority_cleaner = 19;
/* The relative scheduling priority of the purge threads */
UNIV_INTERN ulint srv_sched_priority_purge = 19;
/* The relative scheduling priority of the I/O threads */
UNIV_INTERN ulint srv_sched_priority_io = 19;
/* The relative scheduling priority of the master thread */
UNIV_INTERN ulint srv_sched_priority_master = 19;
/* The relative priority of the current thread. If 0, low priority; if 1, high
priority. */
UNIV_INTERN UNIV_THREAD_LOCAL ulint srv_current_thread_priority = 0;
/* The relative priority of the purge coordinator and worker threads. */
UNIV_INTERN my_bool srv_purge_thread_priority = FALSE;
/* The relative priority of the I/O threads. */
UNIV_INTERN my_bool srv_io_thread_priority = FALSE;
/* The relative priority of the cleaner thread. */
UNIV_INTERN my_bool srv_cleaner_thread_priority = FALSE;
/* The relative priority of the master thread. */
UNIV_INTERN my_bool srv_master_thread_priority = FALSE;
/* The number of purge threads to use.*/
UNIV_INTERN ulong srv_n_purge_threads = 1;
/* the number of pages to purge in one batch */
UNIV_INTERN ulong srv_purge_batch_size = 20;
/* Internal setting for "innodb_stats_method". Decides how InnoDB treats
NULL value when collecting statistics. By default, it is set to
SRV_STATS_NULLS_EQUAL(0), ie. all NULL value are treated equal */
UNIV_INTERN ulong srv_innodb_stats_method = SRV_STATS_NULLS_EQUAL;
UNIV_INTERN srv_stats_t srv_stats;
/* structure to pass status variables to MySQL */
UNIV_INTERN export_var_t export_vars;
/** Normally 0. When nonzero, skip some phases of crash recovery,
starting from SRV_FORCE_IGNORE_CORRUPT, so that data can be recovered
by SELECT or mysqldump. When this is nonzero, we do not allow any user
modifications to the data. */
UNIV_INTERN ulong srv_force_recovery;
/** Print all user-level transactions deadlocks to mysqld stderr */
UNIV_INTERN my_bool srv_print_all_deadlocks = FALSE;
/* Produce a stacktrace on long semaphore wait */
UNIV_INTERN my_bool srv_use_stacktrace = FALSE;
/** Enable INFORMATION_SCHEMA.innodb_cmp_per_index */
UNIV_INTERN my_bool srv_cmp_per_index_enabled = FALSE;
/* If the following is set to 1 then we do not run purge and insert buffer
merge to completion before shutdown. If it is set to 2, do not even flush the
buffer pool to data files at the shutdown: we effectively 'crash'
InnoDB (but lose no committed transactions). */
UNIV_INTERN ulint srv_fast_shutdown = 0;
/* Generate a innodb_status.<pid> file */
UNIV_INTERN ibool srv_innodb_status = FALSE;
/* Optimize prefix index queries to skip cluster index lookup when possible */
/* Enables or disables this prefix optimization. Disabled by default. */
UNIV_INTERN my_bool srv_prefix_index_cluster_optimization = 0;
/* When estimating number of different key values in an index, sample
this many index pages, there are 2 ways to calculate statistics:
* persistent stats that are calculated by ANALYZE TABLE and saved
in the innodb database.
* quick transient stats, that are used if persistent stats for the given
table/index are not found in the innodb database */
UNIV_INTERN unsigned long long srv_stats_transient_sample_pages = 8;
UNIV_INTERN my_bool srv_stats_persistent = TRUE;
UNIV_INTERN unsigned long long srv_stats_persistent_sample_pages = 20;
UNIV_INTERN my_bool srv_stats_auto_recalc = TRUE;
/* The number of rows modified before we calculate new statistics (default 0
= current limits) */
UNIV_INTERN unsigned long long srv_stats_modified_counter = 0;
/* Enable traditional statistic calculation based on number of configured
pages default true. */
UNIV_INTERN my_bool srv_stats_sample_traditional = TRUE;
UNIV_INTERN ibool srv_use_doublewrite_buf = TRUE;
/** doublewrite buffer is 1MB is size i.e.: it can hold 128 16K pages.
The following parameter is the size of the buffer that is used for
batch flushing i.e.: LRU flushing and flush_list flushing. The rest
of the pages are used for single page flushing. */
UNIV_INTERN ulong srv_doublewrite_batch_size = 120;
UNIV_INTERN ulong srv_replication_delay = 0;
UNIV_INTERN ulong srv_pass_corrupt_table = 0; /* 0:disable 1:enable */
UNIV_INTERN ulong srv_log_checksum_algorithm =
SRV_CHECKSUM_ALGORITHM_INNODB;
/*-------------------------------------------*/
#ifdef HAVE_MEMORY_BARRIER
/* No idea to wait long with memory barriers */
UNIV_INTERN ulong srv_n_spin_wait_rounds = 15;
#else
UNIV_INTERN ulong srv_n_spin_wait_rounds = 30;
#endif
UNIV_INTERN ulong srv_spin_wait_delay = 6;
UNIV_INTERN ibool srv_priority_boost = TRUE;
#ifdef UNIV_DEBUG
UNIV_INTERN ibool srv_print_thread_releases = FALSE;
UNIV_INTERN ibool srv_print_lock_waits = FALSE;
UNIV_INTERN ibool srv_print_buf_io = FALSE;
UNIV_INTERN ibool srv_print_log_io = FALSE;
UNIV_INTERN ibool srv_print_latch_waits = FALSE;
#endif /* UNIV_DEBUG */
static ulint srv_n_rows_inserted_old = 0;
static ulint srv_n_rows_updated_old = 0;
static ulint srv_n_rows_deleted_old = 0;
static ulint srv_n_rows_read_old = 0;
static ulint srv_n_system_rows_inserted_old = 0;
static ulint srv_n_system_rows_updated_old = 0;
static ulint srv_n_system_rows_deleted_old = 0;
static ulint srv_n_system_rows_read_old = 0;
UNIV_INTERN ulint srv_truncated_status_writes = 0;
UNIV_INTERN ulint srv_available_undo_logs = 0;
UNIV_INTERN ib_uint64_t srv_page_compression_saved = 0;
UNIV_INTERN ib_uint64_t srv_page_compression_trim_sect512 = 0;
UNIV_INTERN ib_uint64_t srv_page_compression_trim_sect4096 = 0;
UNIV_INTERN ib_uint64_t srv_index_pages_written = 0;
UNIV_INTERN ib_uint64_t srv_non_index_pages_written = 0;
UNIV_INTERN ib_uint64_t srv_pages_page_compressed = 0;
UNIV_INTERN ib_uint64_t srv_page_compressed_trim_op = 0;
UNIV_INTERN ib_uint64_t srv_page_compressed_trim_op_saved = 0;
UNIV_INTERN ib_uint64_t srv_index_page_decompressed = 0;
/* Ensure status variables are on separate cache lines */
#ifdef __powerpc__
#define CACHE_LINE_SIZE 128
#else
#define CACHE_LINE_SIZE 64
#endif
#define CACHE_ALIGNED MY_ATTRIBUTE((aligned (CACHE_LINE_SIZE)))
UNIV_INTERN byte
counters_pad_start[CACHE_LINE_SIZE] MY_ATTRIBUTE((unused)) = {0};
UNIV_INTERN ulint srv_read_views_memory CACHE_ALIGNED = 0;
UNIV_INTERN ulint srv_descriptors_memory CACHE_ALIGNED = 0;
UNIV_INTERN byte
counters_pad_end[CACHE_LINE_SIZE] MY_ATTRIBUTE((unused)) = {0};
/* Set the following to 0 if you want InnoDB to write messages on
stderr on startup/shutdown. */
UNIV_INTERN ibool srv_print_verbose_log = TRUE;
UNIV_INTERN my_bool srv_print_innodb_monitor = FALSE;
UNIV_INTERN my_bool srv_print_innodb_lock_monitor = FALSE;
UNIV_INTERN ibool srv_print_innodb_tablespace_monitor = FALSE;
UNIV_INTERN ibool srv_print_innodb_table_monitor = FALSE;
/** If this flag is set tables without primary key are not allowed */
UNIV_INTERN my_bool srv_force_primary_key = FALSE;
/* Array of English strings describing the current state of an
i/o handler thread */
UNIV_INTERN const char* srv_io_thread_op_info[SRV_MAX_N_IO_THREADS];
UNIV_INTERN const char* srv_io_thread_function[SRV_MAX_N_IO_THREADS];
UNIV_INTERN time_t srv_last_monitor_time;
UNIV_INTERN ib_mutex_t srv_innodb_monitor_mutex;
/* Mutex for locking srv_monitor_file. Not created if srv_read_only_mode */
UNIV_INTERN ib_mutex_t srv_monitor_file_mutex;
#ifdef UNIV_PFS_MUTEX
# ifndef HAVE_ATOMIC_BUILTINS
/* Key to register server_mutex with performance schema */
UNIV_INTERN mysql_pfs_key_t server_mutex_key;
# endif /* !HAVE_ATOMIC_BUILTINS */
/** Key to register srv_innodb_monitor_mutex with performance schema */
UNIV_INTERN mysql_pfs_key_t srv_innodb_monitor_mutex_key;
/** Key to register srv_monitor_file_mutex with performance schema */
UNIV_INTERN mysql_pfs_key_t srv_monitor_file_mutex_key;
/** Key to register srv_dict_tmpfile_mutex with performance schema */
UNIV_INTERN mysql_pfs_key_t srv_dict_tmpfile_mutex_key;
/** Key to register the mutex with performance schema */
UNIV_INTERN mysql_pfs_key_t srv_misc_tmpfile_mutex_key;
/** Key to register srv_sys_t::mutex with performance schema */
UNIV_INTERN mysql_pfs_key_t srv_sys_mutex_key;
/** Key to register srv_sys_t::tasks_mutex with performance schema */
UNIV_INTERN mysql_pfs_key_t srv_sys_tasks_mutex_key;
#endif /* UNIV_PFS_MUTEX */
/** Temporary file for innodb monitor output */
UNIV_INTERN FILE* srv_monitor_file;
/** Mutex for locking srv_dict_tmpfile. Not created if srv_read_only_mode.
This mutex has a very high rank; threads reserving it should not
be holding any InnoDB latches. */
UNIV_INTERN ib_mutex_t srv_dict_tmpfile_mutex;
/** Temporary file for output from the data dictionary */
UNIV_INTERN FILE* srv_dict_tmpfile;
/** Mutex for locking srv_misc_tmpfile. Not created if srv_read_only_mode.
This mutex has a very low rank; threads reserving it should not
acquire any further latches or sleep before releasing this one. */
UNIV_INTERN ib_mutex_t srv_misc_tmpfile_mutex;
/** Temporary file for miscellanous diagnostic output */
UNIV_INTERN FILE* srv_misc_tmpfile;
UNIV_INTERN ulint srv_main_thread_process_no = 0;
UNIV_INTERN ulint srv_main_thread_id = 0;
/* The following counts are used by the srv_master_thread. */
/** Iterations of the loop bounded by 'srv_active' label. */
static ulint srv_main_active_loops = 0;
/** Iterations of the loop bounded by the 'srv_idle' label. */
static ulint srv_main_idle_loops = 0;
/** Iterations of the loop bounded by the 'srv_shutdown' label. */
static ulint srv_main_shutdown_loops = 0;
/** Log writes involving flush. */
static ulint srv_log_writes_and_flush = 0;
/* This is only ever touched by the master thread. It records the
time when the last flush of log file has happened. The master
thread ensures that we flush the log files at least once per
second. */
static time_t srv_last_log_flush_time;
/** Enable semaphore request instrumentation */
UNIV_INTERN my_bool srv_instrument_semaphores = FALSE;
/* Interval in seconds at which various tasks are performed by the
master thread when server is active. In order to balance the workload,
we should try to keep intervals such that they are not multiple of
each other. For example, if we have intervals for various tasks
defined as 5, 10, 15, 60 then all tasks will be performed when
current_time % 60 == 0 and no tasks will be performed when
current_time % 5 != 0. */
# define SRV_MASTER_CHECKPOINT_INTERVAL (7)
# define SRV_MASTER_PURGE_INTERVAL (10)
#ifdef MEM_PERIODIC_CHECK
# define SRV_MASTER_MEM_VALIDATE_INTERVAL (13)
#endif /* MEM_PERIODIC_CHECK */
# define SRV_MASTER_DICT_LRU_INTERVAL (47)
/** Buffer pool dump status frequence in percentages */
UNIV_INTERN ulong srv_buf_dump_status_frequency = 0;
/** Acquire the system_mutex. */
#define srv_sys_mutex_enter() do { \
mutex_enter(&srv_sys->mutex); \
} while (0)
/** Test if the system mutex is owned. */
#define srv_sys_mutex_own() (mutex_own(&srv_sys->mutex) \
&& !srv_read_only_mode)
/** Release the system mutex. */
#define srv_sys_mutex_exit() do { \
mutex_exit(&srv_sys->mutex); \
} while (0)
#define fetch_lock_wait_timeout(trx) \
((trx)->lock.allowed_to_wait \
? thd_lock_wait_timeout((trx)->mysql_thd) \
: 0)
/** Simulate compression failures. */
UNIV_INTERN uint srv_simulate_comp_failures = 0;
/*
IMPLEMENTATION OF THE SERVER MAIN PROGRAM
=========================================
There is the following analogue between this database
server and an operating system kernel:
DB concept equivalent OS concept
---------- ---------------------
transaction -- process;
query thread -- thread;
lock -- semaphore;
kernel -- kernel;
query thread execution:
(a) without lock mutex
reserved -- process executing in user mode;
(b) with lock mutex reserved
-- process executing in kernel mode;
The server has several backgroind threads all running at the same
priority as user threads. It periodically checks if here is anything
happening in the server which requires intervention of the master
thread. Such situations may be, for example, when flushing of dirty
blocks is needed in the buffer pool or old version of database rows
have to be cleaned away (purged). The user can configure a separate
dedicated purge thread(s) too, in which case the master thread does not
do any purging.
The threads which we call user threads serve the queries of the MySQL
server. They run at normal priority.
When there is no activity in the system, also the master thread
suspends itself to wait for an event making the server totally silent.
There is still one complication in our server design. If a
background utility thread obtains a resource (e.g., mutex) needed by a user
thread, and there is also some other user activity in the system,
the user thread may have to wait indefinitely long for the
resource, as the OS does not schedule a background thread if
there is some other runnable user thread. This problem is called
priority inversion in real-time programming.
One solution to the priority inversion problem would be to keep record
of which thread owns which resource and in the above case boost the
priority of the background thread so that it will be scheduled and it
can release the resource. This solution is called priority inheritance
in real-time programming. A drawback of this solution is that the overhead
of acquiring a mutex increases slightly, maybe 0.2 microseconds on a 100
MHz Pentium, because the thread has to call os_thread_get_curr_id. This may
be compared to 0.5 microsecond overhead for a mutex lock-unlock pair. Note
that the thread cannot store the information in the resource , say mutex,
itself, because competing threads could wipe out the information if it is
stored before acquiring the mutex, and if it stored afterwards, the
information is outdated for the time of one machine instruction, at least.
(To be precise, the information could be stored to lock_word in mutex if
the machine supports atomic swap.)
The above solution with priority inheritance may become actual in the
future, currently we do not implement any priority twiddling solution.
Our general aim is to reduce the contention of all mutexes by making
them more fine grained.
The thread table contains information of the current status of each
thread existing in the system, and also the event semaphores used in
suspending the master thread and utility threads when they have nothing
to do. The thread table can be seen as an analogue to the process table
in a traditional Unix implementation. */
/** The server system struct */
struct srv_sys_t{
ib_mutex_t tasks_mutex; /*!< variable protecting the
tasks queue */
UT_LIST_BASE_NODE_T(que_thr_t)
tasks; /*!< task queue */
ib_mutex_t mutex; /*!< variable protecting the
fields below. */
ulint n_sys_threads; /*!< size of the sys_threads
array */
srv_slot_t* sys_threads; /*!< server thread table */
ulint n_threads_active[SRV_MASTER + 1];
/*!< number of threads active
in a thread class */
srv_stats_t::ulint_ctr_1_t
activity_count; /*!< For tracking server
activity */
srv_stats_t::ulint_ctr_1_t
ibuf_merge_activity_count;/*!< For tracking change
buffer merge activity, a subset
of overall server activity */
};
#ifndef HAVE_ATOMIC_BUILTINS
/** Mutex protecting some server global variables. */
UNIV_INTERN ib_mutex_t server_mutex;
#endif /* !HAVE_ATOMIC_BUILTINS */
static srv_sys_t* srv_sys = NULL;
/** Event to signal the monitor thread. */
UNIV_INTERN os_event_t srv_monitor_event;
/** Event to signal the error thread */
UNIV_INTERN os_event_t srv_error_event;
/** Event to signal the buffer pool dump/load thread */
UNIV_INTERN os_event_t srv_buf_dump_event;
/** The buffer pool dump/load file name */
UNIV_INTERN char* srv_buf_dump_filename;
/** Boolean config knobs that tell InnoDB to dump the buffer pool at shutdown
and/or load it during startup. */
UNIV_INTERN char srv_buffer_pool_dump_at_shutdown = FALSE;
UNIV_INTERN char srv_buffer_pool_load_at_startup = FALSE;
/** Slot index in the srv_sys->sys_threads array for the purge thread. */
static const ulint SRV_PURGE_SLOT = 1;
/** Slot index in the srv_sys->sys_threads array for the master thread. */
static const ulint SRV_MASTER_SLOT = 0;
UNIV_INTERN os_event_t srv_checkpoint_completed_event;
UNIV_INTERN os_event_t srv_redo_log_tracked_event;
/** Whether the redo log tracker thread has been started. Does not take into
account whether the tracking is currently enabled (see srv_track_changed_pages
for that) */
UNIV_INTERN bool srv_redo_log_thread_started = false;
/*********************************************************************//**
Prints counters for work done by srv_master_thread. */
static
void
srv_print_master_thread_info(
/*=========================*/
FILE *file) /* in: output stream */
{
fprintf(file, "srv_master_thread loops: %lu srv_active, "
"%lu srv_shutdown, %lu srv_idle\n",
srv_main_active_loops,
srv_main_shutdown_loops,
srv_main_idle_loops);
fprintf(file, "srv_master_thread log flush and writes: %lu\n",
srv_log_writes_and_flush);
}
/*********************************************************************//**
Sets the info describing an i/o thread current state. */
UNIV_INTERN
void
srv_set_io_thread_op_info(
/*======================*/
ulint i, /*!< in: the 'segment' of the i/o thread */
const char* str) /*!< in: constant char string describing the
state */
{
ut_a(i < SRV_MAX_N_IO_THREADS);
srv_io_thread_op_info[i] = str;
}
/*********************************************************************//**
Resets the info describing an i/o thread current state. */
UNIV_INTERN
void
srv_reset_io_thread_op_info()
/*=========================*/
{
for (ulint i = 0; i < UT_ARR_SIZE(srv_io_thread_op_info); ++i) {
srv_io_thread_op_info[i] = "not started yet";
}
}
#ifdef UNIV_DEBUG
/*********************************************************************//**
Validates the type of a thread table slot.
@return TRUE if ok */
static
ibool
srv_thread_type_validate(
/*=====================*/
srv_thread_type type) /*!< in: thread type */
{
switch (type) {
case SRV_NONE:
break;
case SRV_WORKER:
case SRV_PURGE:
case SRV_MASTER:
return(TRUE);
}
ut_error;
return(FALSE);
}
#endif /* UNIV_DEBUG */
/*********************************************************************//**
Gets the type of a thread table slot.
@return thread type */
static
srv_thread_type
srv_slot_get_type(
/*==============*/
const srv_slot_t* slot) /*!< in: thread slot */
{
srv_thread_type type = slot->type;
ut_ad(srv_thread_type_validate(type));
return(type);
}
/*********************************************************************//**
Reserves a slot in the thread table for the current thread.
@return reserved slot */
static
srv_slot_t*
srv_reserve_slot(
/*=============*/
srv_thread_type type) /*!< in: type of the thread */
{
srv_slot_t* slot = 0;
srv_sys_mutex_enter();
ut_ad(srv_thread_type_validate(type));
switch (type) {
case SRV_MASTER:
slot = &srv_sys->sys_threads[SRV_MASTER_SLOT];
break;
case SRV_PURGE:
slot = &srv_sys->sys_threads[SRV_PURGE_SLOT];
break;
case SRV_WORKER:
/* Find an empty slot, skip the master and purge slots. */
for (slot = &srv_sys->sys_threads[2];
slot->in_use;
++slot) {
ut_a(slot < &srv_sys->sys_threads[
srv_sys->n_sys_threads]);
}
break;
case SRV_NONE:
ut_error;
}
ut_a(!slot->in_use);
slot->in_use = TRUE;
slot->suspended = FALSE;
slot->type = type;
ut_ad(srv_slot_get_type(slot) == type);
++srv_sys->n_threads_active[type];
srv_sys_mutex_exit();
return(slot);
}
/*********************************************************************//**
Suspends the calling thread to wait for the event in its thread slot.
@return the current signal count of the event. */
static
ib_int64_t
srv_suspend_thread_low(
/*===================*/
srv_slot_t* slot) /*!< in/out: thread slot */
{
ut_ad(!srv_read_only_mode);
ut_ad(srv_sys_mutex_own());
ut_ad(slot->in_use);