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
|
.. include:: ../global.rst.inc
.. highlight:: none
.. _data-structures:
Data structures and file formats
================================
This page documents the internal data structures and storage
mechanisms of Borg. It is partly based on `mailing list
discussion about internals`_ and also on static code analysis.
.. todo:: Clarify terms, perhaps create a glossary.
ID (client?) vs. key (repository?),
chunks (blob of data in repo?) vs. object (blob of data in repo, referred to from another object?),
.. _repository:
Repository
----------
.. Some parts of this description were taken from the Repository docstring
Borg stores its data in a `Repository`, which is a file system based
transactional key-value store. Thus the repository does not know about
the concept of archives or items.
Each repository has the following file structure:
README
simple text file telling that this is a Borg repository
config
repository configuration
data/
directory where the actual data is stored
hints.%d
hints for repository compaction
index.%d
repository index
lock.roster and lock.exclusive/*
used by the locking system to manage shared and exclusive locks
Transactionality is achieved by using a log (aka journal) to record changes. The log is a series of numbered files
called segments_. Each segment is a series of log entries. The segment number together with the offset of each
entry relative to its segment start establishes an ordering of the log entries. This is the "definition" of
time for the purposes of the log.
.. _config-file:
Config file
~~~~~~~~~~~
Each repository has a ``config`` file which is a ``INI``-style file
and looks like this::
[repository]
version = 1
segments_per_dir = 1000
max_segment_size = 524288000
id = 57d6c1d52ce76a836b532b0e42e677dec6af9fca3673db511279358828a21ed6
This is where the ``repository.id`` is stored. It is a unique
identifier for repositories. It will not change if you move the
repository around so you can make a local transfer then decide to move
the repository to another (even remote) location at a later time.
Keys
~~~~
Repository keys are byte-strings of fixed length (32 bytes), they
don't have a particular meaning (except for the Manifest_).
Normally the keys are computed like this::
key = id = id_hash(unencrypted_data)
The id_hash function depends on the :ref:`encryption mode <borg_init>`.
As the id / key is used for deduplication, id_hash must be a cryptographically
strong hash or MAC.
Segments
~~~~~~~~
Objects referenced by a key are stored inline in files (`segments`) of approx.
500 MB size in numbered subdirectories of ``repo/data``. The number of segments
per directory is controlled by the value of ``segments_per_dir``. If you change
this value in a non-empty repository, you may also need to relocate the segment
files manually.
A segment starts with a magic number (``BORG_SEG`` as an eight byte ASCII string),
followed by a number of log entries. Each log entry consists of:
* 32-bit size of the entry
* CRC32 of the entire entry (for a PUT this includes the data)
* entry tag: PUT, DELETE or COMMIT
* PUT and DELETE follow this with the 32 byte key
* PUT follow the key with the data
Those files are strictly append-only and modified only once.
Tag is either ``PUT``, ``DELETE``, or ``COMMIT``.
When an object is written to the repository a ``PUT`` entry is written
to the file containing the object id and data. If an object is deleted
a ``DELETE`` entry is appended with the object id.
A ``COMMIT`` tag is written when a repository transaction is
committed. The segment number of the segment containing
a commit is the **transaction ID**.
When a repository is opened any ``PUT`` or ``DELETE`` operations not
followed by a ``COMMIT`` tag are discarded since they are part of a
partial/uncommitted transaction.
The size of individual segments is limited to 4 GiB, since the offset of entries
within segments is stored in a 32-bit unsigned integer in the repository index.
Index, hints and integrity
~~~~~~~~~~~~~~~~~~~~~~~~~~
The **repository index** is stored in ``index.<TRANSACTION_ID>`` and is used to
determine an object's location in the repository. It is a HashIndex_,
a hash table using open addressing. It maps object keys_ to two
unsigned 32-bit integers; the first integer gives the segment number,
the second indicates the offset of the object's entry within the segment.
The **hints file** is a msgpacked file named ``hints.<TRANSACTION_ID>``.
It contains:
* version
* list of segments
* compact
The **integrity file** is a msgpacked file named ``integrity.<TRANSACTION_ID>``.
It contains checksums of the index and hints files and is described in the
:ref:`Checksumming data structures <integrity_repo>` section below.
If the index or hints are corrupted, they are re-generated automatically.
If they are outdated, segments are replayed from the index state to the currently
committed transaction.
Compaction
~~~~~~~~~~
For a given key only the last entry regarding the key, which is called current (all other entries are called
superseded), is relevant: If there is no entry or the last entry is a DELETE then the key does not exist.
Otherwise the last PUT defines the value of the key.
By superseding a PUT (with either another PUT or a DELETE) the log entry becomes obsolete. A segment containing
such obsolete entries is called sparse, while a segment containing no such entries is called compact.
Since writing a ``DELETE`` tag does not actually delete any data and
thus does not free disk space any log-based data store will need a
compaction strategy (somewhat analogous to a garbage collector).
Borg uses a simple forward compacting algorithm,
which avoids modifying existing segments.
Compaction runs when a commit is issued with ``compact=True`` parameter, e.g.
by the ``borg compact`` command (unless the :ref:`append_only_mode` is active).
One client transaction can manifest as multiple physical transactions,
since compaction is transacted, too, and Borg does not distinguish between the two::
Perspective| Time -->
-----------+--------------
Client | Begin transaction - Modify Data - Commit | <client waits for repository> (done)
Repository | Begin transaction - Modify Data - Commit | Compact segments - Commit | (done)
The compaction algorithm requires two inputs in addition to the segments themselves:
(i) Which segments are sparse, to avoid scanning all segments (impractical).
Further, Borg uses a conditional compaction strategy: Only those
segments that exceed a threshold sparsity are compacted.
To implement the threshold condition efficiently, the sparsity has
to be stored as well. Therefore, Borg stores a mapping ``(segment
id,) -> (number of sparse bytes,)``.
The 1.0.x series used a simpler non-conditional algorithm,
which only required the list of sparse segments. Thus,
it only stored a list, not the mapping described above.
(ii) Each segment's reference count, which indicates how many live objects are in a segment.
This is not strictly required to perform the algorithm. Rather, it is used to validate
that a segment is unused before deleting it. If the algorithm is incorrect, or the reference
count was not accounted correctly, then an assertion failure occurs.
These two pieces of information are stored in the hints file (`hints.N`)
next to the index (`index.N`).
When loading a hints file, Borg checks the version contained in the file.
The 1.0.x series writes version 1 of the format (with the segments list instead
of the mapping, mentioned above). Since Borg 1.0.4, version 2 is read as well.
The 1.1.x series writes version 2 of the format and reads either version.
When reading a version 1 hints file, Borg 1.1.x will
read all sparse segments to determine their sparsity.
This process may take some time if a repository has been kept in append-only mode
or ``borg compact`` has not been used for a longer time, which both has caused
the number of sparse segments to grow.
Compaction processes sparse segments from oldest to newest; sparse segments
which don't contain enough deleted data to justify compaction are skipped. This
avoids doing e.g. 500 MB of writing current data to a new segment when only
a couple kB were deleted in a segment.
Segments that are compacted are read in entirety. Current entries are written to
a new segment, while superseded entries are omitted. After each segment an intermediary
commit is written to the new segment. Then, the old segment is deleted
(asserting that the reference count diminished to zero), freeing disk space.
A simplified example (excluding conditional compaction and with simpler
commit logic) showing the principal operation of compaction:
.. figure:: compaction.png
:figwidth: 100%
:width: 100%
(The actual algorithm is more complex to avoid various consistency issues, refer to
the ``borg.repository`` module for more comments and documentation on these issues.)
.. _internals_storage_quota:
Storage quotas
~~~~~~~~~~~~~~
Quotas are implemented at the Repository level. The active quota of a repository
is determined by the ``storage_quota`` `config` entry or a run-time override (via :ref:`borg_serve`).
The currently used quota is stored in the hints file. Operations (PUT and DELETE) during
a transaction modify the currently used quota:
- A PUT adds the size of the *log entry* to the quota,
i.e. the length of the data plus the 41 byte header.
- A DELETE subtracts the size of the deleted log entry from the quota,
which includes the header.
Thus, PUT and DELETE are symmetric and cancel each other out precisely.
The quota does not track on-disk size overheads (due to conditional compaction
or append-only mode). In normal operation the inclusion of the log entry headers
in the quota act as a faithful proxy for index and hints overheads.
By tracking effective content size, the client can *always* recover from a full quota
by deleting archives. This would not be possible if the quota tracked on-disk size,
since journaling DELETEs requires extra disk space before space is freed.
Tracking effective size on the other hand accounts DELETEs immediately as freeing quota.
.. rubric:: Enforcing the quota
The storage quota is meant as a robust mechanism for service providers, therefore
:ref:`borg_serve` has to enforce it without loopholes (e.g. modified clients).
The following sections refer to using quotas on remotely accessed repositories.
For local access, consider *client* and *serve* the same.
Accordingly, quotas cannot be enforced with local access,
since the quota can be changed in the repository config.
The quota is enforcible only if *all* :ref:`borg_serve` versions
accessible to clients support quotas (see next section). Further, quota is
per repository. Therefore, ensure clients can only access a defined set of repositories
with their quotas set, using ``--restrict-to-repository``.
If the client exceeds the storage quota the ``StorageQuotaExceeded`` exception is
raised. Normally a client could ignore such an exception and just send a ``commit()``
command anyway, circumventing the quota. However, when ``StorageQuotaExceeded`` is raised,
it is stored in the ``transaction_doomed`` attribute of the repository.
If the transaction is doomed, then commit will re-raise this exception, aborting the commit.
The transaction_doomed indicator is reset on a rollback (which erases the quota-exceeding
state).
.. rubric:: Compatibility with older servers and enabling quota after-the-fact
If no quota data is stored in the hints file, Borg assumes zero quota is used.
Thus, if a repository with an enabled quota is written to with an older ``borg serve``
version that does not understand quotas, then the quota usage will be erased.
The client version is irrelevant to the storage quota and has no part in it.
The form of error messages due to exceeding quota varies with client versions.
A similar situation arises when upgrading from a Borg release that did not have quotas.
Borg will start tracking quota use from the time of the upgrade, starting at zero.
If the quota shall be enforced accurately in these cases, either
- delete the ``index.N`` and ``hints.N`` files, forcing Borg to rebuild both,
re-acquiring quota data in the process, or
-
|