xref: /openbmc/qemu/docs/interop/bitmaps.rst (revision 09a274d8)
1..
2   Copyright 2015 John Snow <jsnow@redhat.com> and Red Hat, Inc.
3   All rights reserved.
4
5   This file is licensed via The FreeBSD Documentation License, the full
6   text of which is included at the end of this document.
7
8====================================
9Dirty Bitmaps and Incremental Backup
10====================================
11
12-  Dirty Bitmaps are objects that track which data needs to be backed up
13   for the next incremental backup.
14
15-  Dirty bitmaps can be created at any time and attached to any node
16   (not just complete drives).
17
18.. contents::
19
20Dirty Bitmap Names
21------------------
22
23-  A dirty bitmap's name is unique to the node, but bitmaps attached to
24   different nodes can share the same name.
25
26-  Dirty bitmaps created for internal use by QEMU may be anonymous and
27   have no name, but any user-created bitmaps must have a name. There
28   can be any number of anonymous bitmaps per node.
29
30-  The name of a user-created bitmap must not be empty ("").
31
32Bitmap Modes
33------------
34
35-  A bitmap can be "frozen," which means that it is currently in-use by
36   a backup operation and cannot be deleted, renamed, written to, reset,
37   etc.
38
39-  The normal operating mode for a bitmap is "active."
40
41Basic QMP Usage
42---------------
43
44Supported Commands
45~~~~~~~~~~~~~~~~~~
46
47- ``block-dirty-bitmap-add``
48- ``block-dirty-bitmap-remove``
49- ``block-dirty-bitmap-clear``
50
51Creation
52~~~~~~~~
53
54-  To create a new bitmap, enabled, on the drive with id=drive0:
55
56.. code:: json
57
58    { "execute": "block-dirty-bitmap-add",
59      "arguments": {
60        "node": "drive0",
61        "name": "bitmap0"
62      }
63    }
64
65-  This bitmap will have a default granularity that matches the cluster
66   size of its associated drive, if available, clamped to between [4KiB,
67   64KiB]. The current default for qcow2 is 64KiB.
68
69-  To create a new bitmap that tracks changes in 32KiB segments:
70
71.. code:: json
72
73    { "execute": "block-dirty-bitmap-add",
74      "arguments": {
75        "node": "drive0",
76        "name": "bitmap0",
77        "granularity": 32768
78      }
79    }
80
81Deletion
82~~~~~~~~
83
84-  Bitmaps that are frozen cannot be deleted.
85
86-  Deleting the bitmap does not impact any other bitmaps attached to the
87   same node, nor does it affect any backups already created from this
88   node.
89
90-  Because bitmaps are only unique to the node to which they are
91   attached, you must specify the node/drive name here, too.
92
93.. code:: json
94
95    { "execute": "block-dirty-bitmap-remove",
96      "arguments": {
97        "node": "drive0",
98        "name": "bitmap0"
99      }
100    }
101
102Resetting
103~~~~~~~~~
104
105-  Resetting a bitmap will clear all information it holds.
106
107-  An incremental backup created from an empty bitmap will copy no data,
108   as if nothing has changed.
109
110.. code:: json
111
112    { "execute": "block-dirty-bitmap-clear",
113      "arguments": {
114        "node": "drive0",
115        "name": "bitmap0"
116      }
117    }
118
119Transactions
120------------
121
122Justification
123~~~~~~~~~~~~~
124
125Bitmaps can be safely modified when the VM is paused or halted by using
126the basic QMP commands. For instance, you might perform the following
127actions:
128
1291. Boot the VM in a paused state.
1302. Create a full drive backup of drive0.
1313. Create a new bitmap attached to drive0.
1324. Resume execution of the VM.
1335. Incremental backups are ready to be created.
134
135At this point, the bitmap and drive backup would be correctly in sync,
136and incremental backups made from this point forward would be correctly
137aligned to the full drive backup.
138
139This is not particularly useful if we decide we want to start
140incremental backups after the VM has been running for a while, for which
141we will need to perform actions such as the following:
142
1431. Boot the VM and begin execution.
1442. Using a single transaction, perform the following operations:
145
146   -  Create ``bitmap0``.
147   -  Create a full drive backup of ``drive0``.
148
1493. Incremental backups are now ready to be created.
150
151Supported Bitmap Transactions
152~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
153
154-  ``block-dirty-bitmap-add``
155-  ``block-dirty-bitmap-clear``
156
157The usages are identical to their respective QMP commands, but see below
158for examples.
159
160Example: New Incremental Backup
161~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
162
163As outlined in the justification, perhaps we want to create a new
164incremental backup chain attached to a drive.
165
166.. code:: json
167
168    { "execute": "transaction",
169      "arguments": {
170        "actions": [
171          {"type": "block-dirty-bitmap-add",
172           "data": {"node": "drive0", "name": "bitmap0"} },
173          {"type": "drive-backup",
174           "data": {"device": "drive0", "target": "/path/to/full_backup.img",
175                    "sync": "full", "format": "qcow2"} }
176        ]
177      }
178    }
179
180Example: New Incremental Backup Anchor Point
181~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
182
183Maybe we just want to create a new full backup with an existing bitmap
184and want to reset the bitmap to track the new chain.
185
186.. code:: json
187
188    { "execute": "transaction",
189      "arguments": {
190        "actions": [
191          {"type": "block-dirty-bitmap-clear",
192           "data": {"node": "drive0", "name": "bitmap0"} },
193          {"type": "drive-backup",
194           "data": {"device": "drive0", "target": "/path/to/new_full_backup.img",
195                    "sync": "full", "format": "qcow2"} }
196        ]
197      }
198    }
199
200Incremental Backups
201-------------------
202
203The star of the show.
204
205**Nota Bene!** Only incremental backups of entire drives are supported
206for now. So despite the fact that you can attach a bitmap to any
207arbitrary node, they are only currently useful when attached to the root
208node. This is because drive-backup only supports drives/devices instead
209of arbitrary nodes.
210
211Example: First Incremental Backup
212~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
213
2141. Create a full backup and sync it to the dirty bitmap, as in the
215   transactional examples above; or with the VM offline, manually create
216   a full copy and then create a new bitmap before the VM begins
217   execution.
218
219   -  Let's assume the full backup is named ``full_backup.img``.
220   -  Let's assume the bitmap you created is ``bitmap0`` attached to
221      ``drive0``.
222
2232. Create a destination image for the incremental backup that utilizes
224   the full backup as a backing image.
225
226   -  Let's assume the new incremental image is named
227      ``incremental.0.img``.
228
229   .. code:: bash
230
231       $ qemu-img create -f qcow2 incremental.0.img -b full_backup.img -F qcow2
232
2333. Issue the incremental backup command:
234
235   .. code:: json
236
237       { "execute": "drive-backup",
238         "arguments": {
239           "device": "drive0",
240           "bitmap": "bitmap0",
241           "target": "incremental.0.img",
242           "format": "qcow2",
243           "sync": "incremental",
244           "mode": "existing"
245         }
246       }
247
248Example: Second Incremental Backup
249~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
250
2511. Create a new destination image for the incremental backup that points
252   to the previous one, e.g.: ``incremental.1.img``
253
254   .. code:: bash
255
256       $ qemu-img create -f qcow2 incremental.1.img -b incremental.0.img -F qcow2
257
2582. Issue a new incremental backup command. The only difference here is
259   that we have changed the target image below.
260
261   .. code:: json
262
263       { "execute": "drive-backup",
264         "arguments": {
265           "device": "drive0",
266           "bitmap": "bitmap0",
267           "target": "incremental.1.img",
268           "format": "qcow2",
269           "sync": "incremental",
270           "mode": "existing"
271         }
272       }
273
274Errors
275------
276
277-  In the event of an error that occurs after a backup job is
278   successfully launched, either by a direct QMP command or a QMP
279   transaction, the user will receive a ``BLOCK_JOB_COMPLETE`` event with
280   a failure message, accompanied by a ``BLOCK_JOB_ERROR`` event.
281
282-  In the case of an event being cancelled, the user will receive a
283   ``BLOCK_JOB_CANCELLED`` event instead of a pair of COMPLETE and ERROR
284   events.
285
286-  In either case, the incremental backup data contained within the
287   bitmap is safely rolled back, and the data within the bitmap is not
288   lost. The image file created for the failed attempt can be safely
289   deleted.
290
291-  Once the underlying problem is fixed (e.g. more storage space is
292   freed up), you can simply retry the incremental backup command with
293   the same bitmap.
294
295Example
296~~~~~~~
297
2981. Create a target image:
299
300   .. code:: bash
301
302       $ qemu-img create -f qcow2 incremental.0.img -b full_backup.img -F qcow2
303
3042. Attempt to create an incremental backup via QMP:
305
306   .. code:: json
307
308       { "execute": "drive-backup",
309         "arguments": {
310           "device": "drive0",
311           "bitmap": "bitmap0",
312           "target": "incremental.0.img",
313           "format": "qcow2",
314           "sync": "incremental",
315           "mode": "existing"
316         }
317       }
318
3193. Receive an event notifying us of failure:
320
321   .. code:: json
322
323       { "timestamp": { "seconds": 1424709442, "microseconds": 844524 },
324         "data": { "speed": 0, "offset": 0, "len": 67108864,
325                   "error": "No space left on device",
326                   "device": "drive1", "type": "backup" },
327         "event": "BLOCK_JOB_COMPLETED" }
328
3294. Delete the failed incremental, and re-create the image.
330
331   .. code:: bash
332
333       $ rm incremental.0.img
334       $ qemu-img create -f qcow2 incremental.0.img -b full_backup.img -F qcow2
335
3365. Retry the command after fixing the underlying problem, such as
337   freeing up space on the backup volume:
338
339   .. code:: json
340
341       { "execute": "drive-backup",
342         "arguments": {
343           "device": "drive0",
344           "bitmap": "bitmap0",
345           "target": "incremental.0.img",
346           "format": "qcow2",
347           "sync": "incremental",
348           "mode": "existing"
349         }
350       }
351
3526. Receive confirmation that the job completed successfully:
353
354   .. code:: json
355
356       { "timestamp": { "seconds": 1424709668, "microseconds": 526525 },
357         "data": { "device": "drive1", "type": "backup",
358                   "speed": 0, "len": 67108864, "offset": 67108864},
359         "event": "BLOCK_JOB_COMPLETED" }
360
361Partial Transactional Failures
362~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
363
364-  Sometimes, a transaction will succeed in launching and return
365   success, but then later the backup jobs themselves may fail. It is
366   possible that a management application may have to deal with a
367   partial backup failure after a successful transaction.
368
369-  If multiple backup jobs are specified in a single transaction, when
370   one of them fails, it will not interact with the other backup jobs in
371   any way.
372
373-  The job(s) that succeeded will clear the dirty bitmap associated with
374   the operation, but the job(s) that failed will not. It is not "safe"
375   to delete any incremental backups that were created successfully in
376   this scenario, even though others failed.
377
378Example
379^^^^^^^
380
381-  QMP example highlighting two backup jobs:
382
383   .. code:: json
384
385       { "execute": "transaction",
386         "arguments": {
387           "actions": [
388             { "type": "drive-backup",
389               "data": { "device": "drive0", "bitmap": "bitmap0",
390                         "format": "qcow2", "mode": "existing",
391                         "sync": "incremental", "target": "d0-incr-1.qcow2" } },
392             { "type": "drive-backup",
393               "data": { "device": "drive1", "bitmap": "bitmap1",
394                         "format": "qcow2", "mode": "existing",
395                         "sync": "incremental", "target": "d1-incr-1.qcow2" } },
396           ]
397         }
398       }
399
400-  QMP example response, highlighting one success and one failure:
401
402   -  Acknowledgement that the Transaction was accepted and jobs were
403      launched:
404
405      .. code:: json
406
407          { "return": {} }
408
409   -  Later, QEMU sends notice that the first job was completed:
410
411      .. code:: json
412
413          { "timestamp": { "seconds": 1447192343, "microseconds": 615698 },
414            "data": { "device": "drive0", "type": "backup",
415                       "speed": 0, "len": 67108864, "offset": 67108864 },
416            "event": "BLOCK_JOB_COMPLETED"
417          }
418
419   -  Later yet, QEMU sends notice that the second job has failed:
420
421      .. code:: json
422
423          { "timestamp": { "seconds": 1447192399, "microseconds": 683015 },
424            "data": { "device": "drive1", "action": "report",
425                      "operation": "read" },
426            "event": "BLOCK_JOB_ERROR" }
427
428      .. code:: json
429
430          { "timestamp": { "seconds": 1447192399, "microseconds":
431          685853 }, "data": { "speed": 0, "offset": 0, "len": 67108864,
432          "error": "Input/output error", "device": "drive1", "type":
433          "backup" }, "event": "BLOCK_JOB_COMPLETED" }
434
435-  In the above example, ``d0-incr-1.qcow2`` is valid and must be kept,
436   but ``d1-incr-1.qcow2`` is invalid and should be deleted. If a VM-wide
437   incremental backup of all drives at a point-in-time is to be made,
438   new backups for both drives will need to be made, taking into account
439   that a new incremental backup for drive0 needs to be based on top of
440   ``d0-incr-1.qcow2``.
441
442Grouped Completion Mode
443~~~~~~~~~~~~~~~~~~~~~~~
444
445-  While jobs launched by transactions normally complete or fail on
446   their own, it is possible to instruct them to complete or fail
447   together as a group.
448
449-  QMP transactions take an optional properties structure that can
450   affect the semantics of the transaction.
451
452-  The "completion-mode" transaction property can be either "individual"
453   which is the default, legacy behavior described above, or "grouped,"
454   a new behavior detailed below.
455
456-  Delayed Completion: In grouped completion mode, no jobs will report
457   success until all jobs are ready to report success.
458
459-  Grouped failure: If any job fails in grouped completion mode, all
460   remaining jobs will be cancelled. Any incremental backups will
461   restore their dirty bitmap objects as if no backup command was ever
462   issued.
463
464   -  Regardless of if QEMU reports a particular incremental backup job
465      as CANCELLED or as an ERROR, the in-memory bitmap will be
466      restored.
467
468Example
469^^^^^^^
470
471-  Here's the same example scenario from above with the new property:
472
473   .. code:: json
474
475       { "execute": "transaction",
476         "arguments": {
477           "actions": [
478             { "type": "drive-backup",
479               "data": { "device": "drive0", "bitmap": "bitmap0",
480                         "format": "qcow2", "mode": "existing",
481                         "sync": "incremental", "target": "d0-incr-1.qcow2" } },
482             { "type": "drive-backup",
483               "data": { "device": "drive1", "bitmap": "bitmap1",
484                         "format": "qcow2", "mode": "existing",
485                         "sync": "incremental", "target": "d1-incr-1.qcow2" } },
486           ],
487           "properties": {
488             "completion-mode": "grouped"
489           }
490         }
491       }
492
493-  QMP example response, highlighting a failure for ``drive2``:
494
495   -  Acknowledgement that the Transaction was accepted and jobs were
496      launched:
497
498      .. code:: json
499
500          { "return": {} }
501
502   -  Later, QEMU sends notice that the second job has errored out, but
503      that the first job was also cancelled:
504
505      .. code:: json
506
507          { "timestamp": { "seconds": 1447193702, "microseconds": 632377 },
508            "data": { "device": "drive1", "action": "report",
509                      "operation": "read" },
510            "event": "BLOCK_JOB_ERROR" }
511
512      .. code:: json
513
514          { "timestamp": { "seconds": 1447193702, "microseconds": 640074 },
515            "data": { "speed": 0, "offset": 0, "len": 67108864,
516                      "error": "Input/output error",
517                      "device": "drive1", "type": "backup" },
518            "event": "BLOCK_JOB_COMPLETED" }
519
520      .. code:: json
521
522          { "timestamp": { "seconds": 1447193702, "microseconds": 640163 },
523            "data": { "device": "drive0", "type": "backup", "speed": 0,
524                      "len": 67108864, "offset": 16777216 },
525            "event": "BLOCK_JOB_CANCELLED" }
526
527.. raw:: html
528
529   <!--
530   The FreeBSD Documentation License
531
532   Redistribution and use in source (Markdown) and 'compiled' forms (SGML, HTML,
533   PDF, PostScript, RTF and so forth) with or without modification, are permitted
534   provided that the following conditions are met:
535
536   Redistributions of source code (Markdown) must retain the above copyright
537   notice, this list of conditions and the following disclaimer of this file
538   unmodified.
539
540   Redistributions in compiled form (transformed to other DTDs, converted to PDF,
541   PostScript, RTF and other formats) must reproduce the above copyright notice,
542   this list of conditions and the following disclaimer in the documentation and/or
543   other materials provided with the distribution.
544
545   THIS DOCUMENTATION IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
546   AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
547   IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR  PURPOSE ARE
548   DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS  BE LIABLE
549   FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
550   DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
551   SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
552   CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
553   OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
554   THIS DOCUMENTATION, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
555   -->
556