Fix Windows CPU count
[fio.git] / HOWTO
CommitLineData
f80dba8d
MT
1How fio works
2-------------
3
4The first step in getting fio to simulate a desired I/O workload, is writing a
5job file describing that specific setup. A job file may contain any number of
6threads and/or files -- the typical contents of the job file is a *global*
7section defining shared parameters, and one or more job sections describing the
8jobs involved. When run, fio parses this file and sets everything up as
9described. If we break down a job from top to bottom, it contains the following
10basic parameters:
11
12`I/O type`_
13
14 Defines the I/O pattern issued to the file(s). We may only be reading
15 sequentially from this file(s), or we may be writing randomly. Or even
16 mixing reads and writes, sequentially or randomly.
17 Should we be doing buffered I/O, or direct/raw I/O?
18
19`Block size`_
20
21 In how large chunks are we issuing I/O? This may be a single value,
22 or it may describe a range of block sizes.
23
24`I/O size`_
25
26 How much data are we going to be reading/writing.
27
28`I/O engine`_
29
30 How do we issue I/O? We could be memory mapping the file, we could be
31 using regular read/write, we could be using splice, async I/O, or even
32 SG (SCSI generic sg).
33
34`I/O depth`_
35
36 If the I/O engine is async, how large a queuing depth do we want to
37 maintain?
38
39
40`Target file/device`_
41
42 How many files are we spreading the workload over.
43
44`Threads, processes and job synchronization`_
45
46 How many threads or processes should we spread this workload over.
47
48The above are the basic parameters defined for a workload, in addition there's a
49multitude of parameters that modify other aspects of how this job behaves.
50
51
52Command line options
53--------------------
54
55.. option:: --debug=type
56
f50fbdda 57 Enable verbose tracing `type` of various fio actions. May be ``all`` for all types
b034c0dd
TK
58 or individual types separated by a comma (e.g. ``--debug=file,mem`` will
59 enable file and memory debugging). Currently, additional logging is
60 available for:
f80dba8d 61
b034c0dd 62 *process*
f80dba8d 63 Dump info related to processes.
b034c0dd 64 *file*
f80dba8d 65 Dump info related to file actions.
b034c0dd 66 *io*
f80dba8d 67 Dump info related to I/O queuing.
b034c0dd 68 *mem*
f80dba8d 69 Dump info related to memory allocations.
b034c0dd 70 *blktrace*
f80dba8d 71 Dump info related to blktrace setup.
b034c0dd 72 *verify*
f80dba8d 73 Dump info related to I/O verification.
b034c0dd 74 *all*
f80dba8d 75 Enable all debug options.
b034c0dd 76 *random*
f80dba8d 77 Dump info related to random offset generation.
b034c0dd 78 *parse*
f80dba8d 79 Dump info related to option matching and parsing.
b034c0dd 80 *diskutil*
f80dba8d 81 Dump info related to disk utilization updates.
b034c0dd 82 *job:x*
f80dba8d 83 Dump info only related to job number x.
b034c0dd 84 *mutex*
f80dba8d 85 Dump info only related to mutex up/down ops.
b034c0dd 86 *profile*
f80dba8d 87 Dump info related to profile extensions.
b034c0dd 88 *time*
f80dba8d 89 Dump info related to internal time keeping.
b034c0dd 90 *net*
f80dba8d 91 Dump info related to networking connections.
b034c0dd 92 *rate*
f80dba8d 93 Dump info related to I/O rate switching.
b034c0dd 94 *compress*
f80dba8d 95 Dump info related to log compress/decompress.
b034c0dd 96 *?* or *help*
f80dba8d
MT
97 Show available debug options.
98
99.. option:: --parse-only
100
25cd4b95 101 Parse options only, don't start any I/O.
f80dba8d 102
b9921d1a
DZ
103.. option:: --merge-blktrace-only
104
105 Merge blktraces only, don't start any I/O.
106
f80dba8d
MT
107.. option:: --output=filename
108
109 Write output to file `filename`.
110
f50fbdda 111.. option:: --output-format=format
b8f7e412 112
f50fbdda 113 Set the reporting `format` to `normal`, `terse`, `json`, or `json+`. Multiple
b8f7e412
TK
114 formats can be selected, separated by a comma. `terse` is a CSV based
115 format. `json+` is like `json`, except it adds a full dump of the latency
116 buckets.
117
f80dba8d
MT
118.. option:: --bandwidth-log
119
120 Generate aggregate bandwidth logs.
121
122.. option:: --minimal
123
124 Print statistics in a terse, semicolon-delimited format.
125
126.. option:: --append-terse
127
b034c0dd
TK
128 Print statistics in selected mode AND terse, semicolon-delimited format.
129 **Deprecated**, use :option:`--output-format` instead to select multiple
130 formats.
f80dba8d 131
f50fbdda 132.. option:: --terse-version=version
f80dba8d 133
f50fbdda 134 Set terse `version` output format (default 3, or 2 or 4 or 5).
f80dba8d
MT
135
136.. option:: --version
137
b8f7e412 138 Print version information and exit.
f80dba8d
MT
139
140.. option:: --help
141
113f0e7c 142 Print a summary of the command line options and exit.
f80dba8d
MT
143
144.. option:: --cpuclock-test
145
146 Perform test and validation of internal CPU clock.
147
113f0e7c 148.. option:: --crctest=[test]
f80dba8d 149
b034c0dd
TK
150 Test the speed of the built-in checksumming functions. If no argument is
151 given, all of them are tested. Alternatively, a comma separated list can
152 be passed, in which case the given ones are tested.
f80dba8d
MT
153
154.. option:: --cmdhelp=command
155
156 Print help information for `command`. May be ``all`` for all commands.
157
158.. option:: --enghelp=[ioengine[,command]]
159
f50fbdda
TK
160 List all commands defined by `ioengine`, or print help for `command`
161 defined by `ioengine`. If no `ioengine` is given, list all
b034c0dd 162 available ioengines.
f80dba8d
MT
163
164.. option:: --showcmd=jobfile
165
b8f7e412 166 Convert `jobfile` to a set of command-line options.
f80dba8d
MT
167
168.. option:: --readonly
169
4027b2a1
VF
170 Turn on safety read-only checks, preventing writes and trims. The
171 ``--readonly`` option is an extra safety guard to prevent users from
172 accidentally starting a write or trim workload when that is not desired.
173 Fio will only modify the device under test if
174 `rw=write/randwrite/rw/randrw/trim/randtrim/trimwrite` is given. This
175 safety net can be used as an extra precaution.
f80dba8d
MT
176
177.. option:: --eta=when
178
b8f7e412 179 Specifies when real-time ETA estimate should be printed. `when` may be
db37d890
JA
180 `always`, `never` or `auto`. `auto` is the default, it prints ETA
181 when requested if the output is a TTY. `always` disregards the output
182 type, and prints ETA when requested. `never` never prints ETA.
183
184.. option:: --eta-interval=time
185
186 By default, fio requests client ETA status roughly every second. With
187 this option, the interval is configurable. Fio imposes a minimum
188 allowed time to avoid flooding the console, less than 250 msec is
189 not supported.
f80dba8d
MT
190
191.. option:: --eta-newline=time
192
947e0fe0
SW
193 Force a new line for every `time` period passed. When the unit is omitted,
194 the value is interpreted in seconds.
f80dba8d
MT
195
196.. option:: --status-interval=time
197
aa6cb459
VF
198 Force a full status dump of cumulative (from job start) values at `time`
199 intervals. This option does *not* provide per-period measurements. So
200 values such as bandwidth are running averages. When the time unit is omitted,
c1f4de8a
JA
201 `time` is interpreted in seconds. Note that using this option with
202 ``--output-format=json`` will yield output that technically isn't valid
203 json, since the output will be collated sets of valid json. It will need
204 to be split into valid sets of json after the run.
f80dba8d
MT
205
206.. option:: --section=name
207
b034c0dd
TK
208 Only run specified section `name` in job file. Multiple sections can be specified.
209 The ``--section`` option allows one to combine related jobs into one file.
210 E.g. one job file could define light, moderate, and heavy sections. Tell
211 fio to run only the "heavy" section by giving ``--section=heavy``
212 command line option. One can also specify the "write" operations in one
213 section and "verify" operation in another section. The ``--section`` option
214 only applies to job sections. The reserved *global* section is always
215 parsed and used.
f80dba8d
MT
216
217.. option:: --alloc-size=kb
218
b034c0dd
TK
219 Set the internal smalloc pool size to `kb` in KiB. The
220 ``--alloc-size`` switch allows one to use a larger pool size for smalloc.
221 If running large jobs with randommap enabled, fio can run out of memory.
222 Smalloc is an internal allocator for shared structures from a fixed size
223 memory pool and can grow to 16 pools. The pool size defaults to 16MiB.
f80dba8d 224
b034c0dd
TK
225 NOTE: While running :file:`.fio_smalloc.*` backing store files are visible
226 in :file:`/tmp`.
f80dba8d
MT
227
228.. option:: --warnings-fatal
229
b034c0dd
TK
230 All fio parser warnings are fatal, causing fio to exit with an
231 error.
f80dba8d
MT
232
233.. option:: --max-jobs=nr
234
f50fbdda 235 Set the maximum number of threads/processes to support to `nr`.
818322cc 236 NOTE: On Linux, it may be necessary to increase the shared-memory
71aa48eb 237 limit (:file:`/proc/sys/kernel/shmmax`) if fio runs into errors while
818322cc 238 creating jobs.
f80dba8d
MT
239
240.. option:: --server=args
241
b034c0dd
TK
242 Start a backend server, with `args` specifying what to listen to.
243 See `Client/Server`_ section.
f80dba8d
MT
244
245.. option:: --daemonize=pidfile
246
b034c0dd 247 Background a fio server, writing the pid to the given `pidfile` file.
f80dba8d
MT
248
249.. option:: --client=hostname
250
f50fbdda 251 Instead of running the jobs locally, send and run them on the given `hostname`
71aa48eb 252 or set of `hostname`\s. See `Client/Server`_ section.
f80dba8d
MT
253
254.. option:: --remote-config=file
255
f50fbdda 256 Tell fio server to load this local `file`.
f80dba8d
MT
257
258.. option:: --idle-prof=option
259
b8f7e412 260 Report CPU idleness. `option` is one of the following:
113f0e7c
SW
261
262 **calibrate**
263 Run unit work calibration only and exit.
264
265 **system**
266 Show aggregate system idleness and unit work.
267
268 **percpu**
269 As **system** but also show per CPU idleness.
f80dba8d
MT
270
271.. option:: --inflate-log=log
272
f50fbdda 273 Inflate and output compressed `log`.
f80dba8d
MT
274
275.. option:: --trigger-file=file
276
f50fbdda 277 Execute trigger command when `file` exists.
f80dba8d 278
f50fbdda 279.. option:: --trigger-timeout=time
f80dba8d 280
f50fbdda 281 Execute trigger at this `time`.
f80dba8d 282
f50fbdda 283.. option:: --trigger=command
f80dba8d 284
f50fbdda 285 Set this `command` as local trigger.
f80dba8d 286
f50fbdda 287.. option:: --trigger-remote=command
f80dba8d 288
f50fbdda 289 Set this `command` as remote trigger.
f80dba8d
MT
290
291.. option:: --aux-path=path
292
f4401bf8
SW
293 Use the directory specified by `path` for generated state files instead
294 of the current working directory.
f80dba8d
MT
295
296Any parameters following the options will be assumed to be job files, unless
297they match a job file parameter. Multiple job files can be listed and each job
298file will be regarded as a separate group. Fio will :option:`stonewall`
299execution between each group.
300
301
302Job file format
303---------------
304
305As previously described, fio accepts one or more job files describing what it is
306supposed to do. The job file format is the classic ini file, where the names
c60ebc45 307enclosed in [] brackets define the job name. You are free to use any ASCII name
f80dba8d
MT
308you want, except *global* which has special meaning. Following the job name is
309a sequence of zero or more parameters, one per line, that define the behavior of
310the job. If the first character in a line is a ';' or a '#', the entire line is
311discarded as a comment.
312
313A *global* section sets defaults for the jobs described in that file. A job may
314override a *global* section parameter, and a job file may even have several
315*global* sections if so desired. A job is only affected by a *global* section
316residing above it.
317
f50fbdda
TK
318The :option:`--cmdhelp` option also lists all options. If used with a `command`
319argument, :option:`--cmdhelp` will detail the given `command`.
f80dba8d
MT
320
321See the `examples/` directory for inspiration on how to write job files. Note
322the copyright and license requirements currently apply to `examples/` files.
323
324So let's look at a really simple job file that defines two processes, each
325randomly reading from a 128MiB file:
326
327.. code-block:: ini
328
329 ; -- start job file --
330 [global]
331 rw=randread
332 size=128m
333
334 [job1]
335
336 [job2]
337
338 ; -- end job file --
339
340As you can see, the job file sections themselves are empty as all the described
341parameters are shared. As no :option:`filename` option is given, fio makes up a
342`filename` for each of the jobs as it sees fit. On the command line, this job
343would look as follows::
344
345$ fio --name=global --rw=randread --size=128m --name=job1 --name=job2
346
347
348Let's look at an example that has a number of processes writing randomly to
349files:
350
351.. code-block:: ini
352
353 ; -- start job file --
354 [random-writers]
355 ioengine=libaio
356 iodepth=4
357 rw=randwrite
358 bs=32k
359 direct=0
360 size=64m
361 numjobs=4
362 ; -- end job file --
363
364Here we have no *global* section, as we only have one job defined anyway. We
365want to use async I/O here, with a depth of 4 for each file. We also increased
366the buffer size used to 32KiB and define numjobs to 4 to fork 4 identical
367jobs. The result is 4 processes each randomly writing to their own 64MiB
368file. Instead of using the above job file, you could have given the parameters
369on the command line. For this case, you would specify::
370
371$ fio --name=random-writers --ioengine=libaio --iodepth=4 --rw=randwrite --bs=32k --direct=0 --size=64m --numjobs=4
372
373When fio is utilized as a basis of any reasonably large test suite, it might be
374desirable to share a set of standardized settings across multiple job files.
375Instead of copy/pasting such settings, any section may pull in an external
376:file:`filename.fio` file with *include filename* directive, as in the following
377example::
378
379 ; -- start job file including.fio --
380 [global]
381 filename=/tmp/test
382 filesize=1m
383 include glob-include.fio
384
385 [test]
386 rw=randread
387 bs=4k
388 time_based=1
389 runtime=10
390 include test-include.fio
391 ; -- end job file including.fio --
392
393.. code-block:: ini
394
395 ; -- start job file glob-include.fio --
396 thread=1
397 group_reporting=1
398 ; -- end job file glob-include.fio --
399
400.. code-block:: ini
401
402 ; -- start job file test-include.fio --
403 ioengine=libaio
404 iodepth=4
405 ; -- end job file test-include.fio --
406
407Settings pulled into a section apply to that section only (except *global*
408section). Include directives may be nested in that any included file may contain
409further include directive(s). Include files may not contain [] sections.
410
411
412Environment variables
413~~~~~~~~~~~~~~~~~~~~~
414
415Fio also supports environment variable expansion in job files. Any sub-string of
416the form ``${VARNAME}`` as part of an option value (in other words, on the right
417of the '='), will be expanded to the value of the environment variable called
418`VARNAME`. If no such environment variable is defined, or `VARNAME` is the
419empty string, the empty string will be substituted.
420
421As an example, let's look at a sample fio invocation and job file::
422
423$ SIZE=64m NUMJOBS=4 fio jobfile.fio
424
425.. code-block:: ini
426
427 ; -- start job file --
428 [random-writers]
429 rw=randwrite
430 size=${SIZE}
431 numjobs=${NUMJOBS}
432 ; -- end job file --
433
434This will expand to the following equivalent job file at runtime:
435
436.. code-block:: ini
437
438 ; -- start job file --
439 [random-writers]
440 rw=randwrite
441 size=64m
442 numjobs=4
443 ; -- end job file --
444
445Fio ships with a few example job files, you can also look there for inspiration.
446
447Reserved keywords
448~~~~~~~~~~~~~~~~~
449
450Additionally, fio has a set of reserved keywords that will be replaced
451internally with the appropriate value. Those keywords are:
452
453**$pagesize**
454
455 The architecture page size of the running system.
456
457**$mb_memory**
458
459 Megabytes of total memory in the system.
460
461**$ncpus**
462
463 Number of online available CPUs.
464
465These can be used on the command line or in the job file, and will be
466automatically substituted with the current system values when the job is
467run. Simple math is also supported on these keywords, so you can perform actions
468like::
469
b034c0dd 470 size=8*$mb_memory
f80dba8d
MT
471
472and get that properly expanded to 8 times the size of memory in the machine.
473
474
475Job file parameters
476-------------------
477
478This section describes in details each parameter associated with a job. Some
479parameters take an option of a given type, such as an integer or a
480string. Anywhere a numeric value is required, an arithmetic expression may be
481used, provided it is surrounded by parentheses. Supported operators are:
482
483 - addition (+)
484 - subtraction (-)
485 - multiplication (*)
486 - division (/)
487 - modulus (%)
488 - exponentiation (^)
489
490For time values in expressions, units are microseconds by default. This is
491different than for time values not in expressions (not enclosed in
492parentheses). The following types are used:
493
494
495Parameter types
496~~~~~~~~~~~~~~~
497
498**str**
b034c0dd 499 String: A sequence of alphanumeric characters.
f80dba8d
MT
500
501**time**
008d0feb
SW
502 Integer with possible time suffix. Without a unit value is interpreted as
503 seconds unless otherwise specified. Accepts a suffix of 'd' for days, 'h' for
504 hours, 'm' for minutes, 's' for seconds, 'ms' (or 'msec') for milliseconds and
505 'us' (or 'usec') for microseconds. For example, use 10m for 10 minutes.
f80dba8d
MT
506
507.. _int:
508
509**int**
510 Integer. A whole number value, which may contain an integer prefix
511 and an integer suffix:
512
b034c0dd 513 [*integer prefix*] **number** [*integer suffix*]
f80dba8d
MT
514
515 The optional *integer prefix* specifies the number's base. The default
516 is decimal. *0x* specifies hexadecimal.
517
518 The optional *integer suffix* specifies the number's units, and includes an
519 optional unit prefix and an optional unit. For quantities of data, the
947e0fe0
SW
520 default unit is bytes. For quantities of time, the default unit is seconds
521 unless otherwise specified.
f80dba8d 522
9207a0cb 523 With :option:`kb_base`\=1000, fio follows international standards for unit
f80dba8d
MT
524 prefixes. To specify power-of-10 decimal values defined in the
525 International System of Units (SI):
526
eccce61a
TK
527 * *K* -- means kilo (K) or 1000
528 * *M* -- means mega (M) or 1000**2
529 * *G* -- means giga (G) or 1000**3
530 * *T* -- means tera (T) or 1000**4
531 * *P* -- means peta (P) or 1000**5
f80dba8d
MT
532
533 To specify power-of-2 binary values defined in IEC 80000-13:
534
eccce61a
TK
535 * *Ki* -- means kibi (Ki) or 1024
536 * *Mi* -- means mebi (Mi) or 1024**2
537 * *Gi* -- means gibi (Gi) or 1024**3
538 * *Ti* -- means tebi (Ti) or 1024**4
539 * *Pi* -- means pebi (Pi) or 1024**5
f80dba8d 540
9207a0cb 541 With :option:`kb_base`\=1024 (the default), the unit prefixes are opposite
f80dba8d
MT
542 from those specified in the SI and IEC 80000-13 standards to provide
543 compatibility with old scripts. For example, 4k means 4096.
544
545 For quantities of data, an optional unit of 'B' may be included
b8f7e412 546 (e.g., 'kB' is the same as 'k').
f80dba8d
MT
547
548 The *integer suffix* is not case sensitive (e.g., m/mi mean mebi/mega,
549 not milli). 'b' and 'B' both mean byte, not bit.
550
9207a0cb 551 Examples with :option:`kb_base`\=1000:
f80dba8d
MT
552
553 * *4 KiB*: 4096, 4096b, 4096B, 4ki, 4kib, 4kiB, 4Ki, 4KiB
554 * *1 MiB*: 1048576, 1mi, 1024ki
555 * *1 MB*: 1000000, 1m, 1000k
556 * *1 TiB*: 1099511627776, 1ti, 1024gi, 1048576mi
557 * *1 TB*: 1000000000, 1t, 1000m, 1000000k
558
9207a0cb 559 Examples with :option:`kb_base`\=1024 (default):
f80dba8d
MT
560
561 * *4 KiB*: 4096, 4096b, 4096B, 4k, 4kb, 4kB, 4K, 4KB
562 * *1 MiB*: 1048576, 1m, 1024k
563 * *1 MB*: 1000000, 1mi, 1000ki
564 * *1 TiB*: 1099511627776, 1t, 1024g, 1048576m
565 * *1 TB*: 1000000000, 1ti, 1000mi, 1000000ki
566
567 To specify times (units are not case sensitive):
568
569 * *D* -- means days
570 * *H* -- means hours
4502cb42 571 * *M* -- means minutes
f80dba8d
MT
572 * *s* -- or sec means seconds (default)
573 * *ms* -- or *msec* means milliseconds
574 * *us* -- or *usec* means microseconds
575
576 If the option accepts an upper and lower range, use a colon ':' or
577 minus '-' to separate such values. See :ref:`irange <irange>`.
4502cb42
SW
578 If the lower value specified happens to be larger than the upper value
579 the two values are swapped.
f80dba8d
MT
580
581.. _bool:
582
583**bool**
584 Boolean. Usually parsed as an integer, however only defined for
585 true and false (1 and 0).
586
587.. _irange:
588
589**irange**
590 Integer range with suffix. Allows value range to be given, such as
c60ebc45 591 1024-4096. A colon may also be used as the separator, e.g. 1k:4k. If the
f80dba8d
MT
592 option allows two sets of ranges, they can be specified with a ',' or '/'
593 delimiter: 1k-4k/8k-32k. Also see :ref:`int <int>`.
594
595**float_list**
596 A list of floating point numbers, separated by a ':' character.
597
f5c3bcf2
TK
598With the above in mind, here follows the complete list of fio job parameters.
599
f80dba8d
MT
600
601Units
602~~~~~
603
604.. option:: kb_base=int
605
606 Select the interpretation of unit prefixes in input parameters.
607
608 **1000**
609 Inputs comply with IEC 80000-13 and the International
610 System of Units (SI). Use:
611
612 - power-of-2 values with IEC prefixes (e.g., KiB)
613 - power-of-10 values with SI prefixes (e.g., kB)
614
615 **1024**
616 Compatibility mode (default). To avoid breaking old scripts:
617
618 - power-of-2 values with SI prefixes
619 - power-of-10 values with IEC prefixes
620
621 See :option:`bs` for more details on input parameters.
622
623 Outputs always use correct prefixes. Most outputs include both
624 side-by-side, like::
625
626 bw=2383.3kB/s (2327.4KiB/s)
627
628 If only one value is reported, then kb_base selects the one to use:
629
630 **1000** -- SI prefixes
631
632 **1024** -- IEC prefixes
633
634.. option:: unit_base=int
635
636 Base unit for reporting. Allowed values are:
637
638 **0**
639 Use auto-detection (default).
640 **8**
641 Byte based.
642 **1**
643 Bit based.
644
645
f80dba8d
MT
646Job description
647~~~~~~~~~~~~~~~
648
649.. option:: name=str
650
651 ASCII name of the job. This may be used to override the name printed by fio
652 for this job. Otherwise the job name is used. On the command line this
653 parameter has the special purpose of also signaling the start of a new job.
654
655.. option:: description=str
656
657 Text description of the job. Doesn't do anything except dump this text
658 description when this job is run. It's not parsed.
659
660.. option:: loops=int
661
662 Run the specified number of iterations of this job. Used to repeat the same
663 workload a given number of times. Defaults to 1.
664
665.. option:: numjobs=int
666
79591fa9
TK
667 Create the specified number of clones of this job. Each clone of job
668 is spawned as an independent thread or process. May be used to setup a
f80dba8d
MT
669 larger number of threads/processes doing the same thing. Each thread is
670 reported separately; to see statistics for all clones as a whole, use
671 :option:`group_reporting` in conjunction with :option:`new_group`.
a47b697c 672 See :option:`--max-jobs`. Default: 1.
f80dba8d
MT
673
674
675Time related parameters
676~~~~~~~~~~~~~~~~~~~~~~~
677
678.. option:: runtime=time
679
f75ede1d 680 Tell fio to terminate processing after the specified period of time. It
f80dba8d 681 can be quite hard to determine for how long a specified job will run, so
f75ede1d 682 this parameter is handy to cap the total runtime to a given time. When
804c0839 683 the unit is omitted, the value is interpreted in seconds.
f80dba8d
MT
684
685.. option:: time_based
686
687 If set, fio will run for the duration of the :option:`runtime` specified
688 even if the file(s) are completely read or written. It will simply loop over
689 the same workload as many times as the :option:`runtime` allows.
690
a881438b 691.. option:: startdelay=irange(time)
f80dba8d 692
947e0fe0
SW
693 Delay the start of job for the specified amount of time. Can be a single
694 value or a range. When given as a range, each thread will choose a value
695 randomly from within the range. Value is in seconds if a unit is omitted.
f80dba8d
MT
696
697.. option:: ramp_time=time
698
699 If set, fio will run the specified workload for this amount of time before
700 logging any performance numbers. Useful for letting performance settle
701 before logging results, thus minimizing the runtime required for stable
702 results. Note that the ``ramp_time`` is considered lead in time for a job,
703 thus it will increase the total runtime if a special timeout or
f75ede1d
SW
704 :option:`runtime` is specified. When the unit is omitted, the value is
705 given in seconds.
f80dba8d
MT
706
707.. option:: clocksource=str
708
709 Use the given clocksource as the base of timing. The supported options are:
710
711 **gettimeofday**
712 :manpage:`gettimeofday(2)`
713
714 **clock_gettime**
715 :manpage:`clock_gettime(2)`
716
717 **cpu**
718 Internal CPU clock source
719
720 cpu is the preferred clocksource if it is reliable, as it is very fast (and
721 fio is heavy on time calls). Fio will automatically use this clocksource if
722 it's supported and considered reliable on the system it is running on,
723 unless another clocksource is specifically set. For x86/x86-64 CPUs, this
724 means supporting TSC Invariant.
725
726.. option:: gtod_reduce=bool
727
728 Enable all of the :manpage:`gettimeofday(2)` reducing options
f75ede1d 729 (:option:`disable_clat`, :option:`disable_slat`, :option:`disable_bw_measurement`) plus
f80dba8d
MT
730 reduce precision of the timeout somewhat to really shrink the
731 :manpage:`gettimeofday(2)` call count. With this option enabled, we only do
732 about 0.4% of the :manpage:`gettimeofday(2)` calls we would have done if all
733 time keeping was enabled.
734
735.. option:: gtod_cpu=int
736
737 Sometimes it's cheaper to dedicate a single thread of execution to just
738 getting the current time. Fio (and databases, for instance) are very
739 intensive on :manpage:`gettimeofday(2)` calls. With this option, you can set
740 one CPU aside for doing nothing but logging current time to a shared memory
741 location. Then the other threads/processes that run I/O workloads need only
742 copy that segment, instead of entering the kernel with a
743 :manpage:`gettimeofday(2)` call. The CPU set aside for doing these time
744 calls will be excluded from other uses. Fio will manually clear it from the
745 CPU mask of other jobs.
746
747
748Target file/device
749~~~~~~~~~~~~~~~~~~
750
751.. option:: directory=str
752
753 Prefix filenames with this directory. Used to place files in a different
754 location than :file:`./`. You can specify a number of directories by
755 separating the names with a ':' character. These directories will be
02dd2689 756 assigned equally distributed to job clones created by :option:`numjobs` as
f80dba8d
MT
757 long as they are using generated filenames. If specific `filename(s)` are
758 set fio will use the first listed directory, and thereby matching the
f4401bf8
SW
759 `filename` semantic (which generates a file for each clone if not
760 specified, but lets all clones use the same file if set).
f80dba8d 761
02dd2689
SW
762 See the :option:`filename` option for information on how to escape "``:``" and
763 "``\``" characters within the directory path itself.
f80dba8d 764
f4401bf8
SW
765 Note: To control the directory fio will use for internal state files
766 use :option:`--aux-path`.
767
f80dba8d
MT
768.. option:: filename=str
769
770 Fio normally makes up a `filename` based on the job name, thread number, and
02dd2689
SW
771 file number (see :option:`filename_format`). If you want to share files
772 between threads in a job or several
79591fa9
TK
773 jobs with fixed file paths, specify a `filename` for each of them to override
774 the default. If the ioengine is file based, you can specify a number of files
775 by separating the names with a ':' colon. So if you wanted a job to open
776 :file:`/dev/sda` and :file:`/dev/sdb` as the two working files, you would use
777 ``filename=/dev/sda:/dev/sdb``. This also means that whenever this option is
778 specified, :option:`nrfiles` is ignored. The size of regular files specified
02dd2689 779 by this option will be :option:`size` divided by number of files unless an
79591fa9
TK
780 explicit size is specified by :option:`filesize`.
781
02dd2689
SW
782 Each colon and backslash in the wanted path must be escaped with a ``\``
783 character. For instance, if the path is :file:`/dev/dsk/foo@3,0:c` then you
784 would use ``filename=/dev/dsk/foo@3,0\:c`` and if the path is
785 :file:`F:\\filename` then you would use ``filename=F\:\\filename``.
786
f80dba8d
MT
787 On Windows, disk devices are accessed as :file:`\\\\.\\PhysicalDrive0` for
788 the first device, :file:`\\\\.\\PhysicalDrive1` for the second etc.
789 Note: Windows and FreeBSD prevent write access to areas
02dd2689
SW
790 of the disk containing in-use data (e.g. filesystems).
791
792 The filename "`-`" is a reserved name, meaning *stdin* or *stdout*. Which
793 of the two depends on the read/write direction set.
f80dba8d
MT
794
795.. option:: filename_format=str
796
797 If sharing multiple files between jobs, it is usually necessary to have fio
798 generate the exact names that you want. By default, fio will name a file
799 based on the default file format specification of
800 :file:`jobname.jobnumber.filenumber`. With this option, that can be
801 customized. Fio will recognize and replace the following keywords in this
802 string:
803
804 **$jobname**
805 The name of the worker thread or process.
806 **$jobnum**
807 The incremental number of the worker thread or process.
808 **$filenum**
809 The incremental number of the file for that worker thread or
810 process.
811
812 To have dependent jobs share a set of files, this option can be set to have
813 fio generate filenames that are shared between the two. For instance, if
814 :file:`testfiles.$filenum` is specified, file number 4 for any job will be
815 named :file:`testfiles.4`. The default of :file:`$jobname.$jobnum.$filenum`
816 will be used if no other format specifier is given.
817
645943c0
JB
818 If you specify a path then the directories will be created up to the
819 main directory for the file. So for example if you specify
820 ``filename_format=a/b/c/$jobnum`` then the directories a/b/c will be
821 created before the file setup part of the job. If you specify
822 :option:`directory` then the path will be relative that directory,
823 otherwise it is treated as the absolute path.
824
f80dba8d
MT
825.. option:: unique_filename=bool
826
827 To avoid collisions between networked clients, fio defaults to prefixing any
828 generated filenames (with a directory specified) with the source of the
829 client connecting. To disable this behavior, set this option to 0.
830
831.. option:: opendir=str
832
833 Recursively open any files below directory `str`.
834
835.. option:: lockfile=str
836
837 Fio defaults to not locking any files before it does I/O to them. If a file
838 or file descriptor is shared, fio can serialize I/O to that file to make the
839 end result consistent. This is usual for emulating real workloads that share
840 files. The lock modes are:
841
842 **none**
843 No locking. The default.
844 **exclusive**
845 Only one thread or process may do I/O at a time, excluding all
846 others.
847 **readwrite**
848 Read-write locking on the file. Many readers may
849 access the file at the same time, but writes get exclusive access.
850
851.. option:: nrfiles=int
852
79591fa9
TK
853 Number of files to use for this job. Defaults to 1. The size of files
854 will be :option:`size` divided by this unless explicit size is specified by
855 :option:`filesize`. Files are created for each thread separately, and each
856 file will have a file number within its name by default, as explained in
857 :option:`filename` section.
858
f80dba8d
MT
859
860.. option:: openfiles=int
861
862 Number of files to keep open at the same time. Defaults to the same as
863 :option:`nrfiles`, can be set smaller to limit the number simultaneous
864 opens.
865
866.. option:: file_service_type=str
867
868 Defines how fio decides which file from a job to service next. The following
869 types are defined:
870
871 **random**
872 Choose a file at random.
873
874 **roundrobin**
875 Round robin over opened files. This is the default.
876
877 **sequential**
878 Finish one file before moving on to the next. Multiple files can
f50fbdda 879 still be open depending on :option:`openfiles`.
f80dba8d
MT
880
881 **zipf**
c60ebc45 882 Use a *Zipf* distribution to decide what file to access.
f80dba8d
MT
883
884 **pareto**
c60ebc45 885 Use a *Pareto* distribution to decide what file to access.
f80dba8d 886
dd3503d3 887 **normal**
c60ebc45 888 Use a *Gaussian* (normal) distribution to decide what file to
f80dba8d
MT
889 access.
890
dd3503d3
SW
891 **gauss**
892 Alias for normal.
893
f80dba8d
MT
894 For *random*, *roundrobin*, and *sequential*, a postfix can be appended to
895 tell fio how many I/Os to issue before switching to a new file. For example,
896 specifying ``file_service_type=random:8`` would cause fio to issue
897 8 I/Os before selecting a new file at random. For the non-uniform
898 distributions, a floating point postfix can be given to influence how the
899 distribution is skewed. See :option:`random_distribution` for a description
900 of how that would work.
901
902.. option:: ioscheduler=str
903
904 Attempt to switch the device hosting the file to the specified I/O scheduler
905 before running.
906
907.. option:: create_serialize=bool
908
909 If true, serialize the file creation for the jobs. This may be handy to
910 avoid interleaving of data files, which may greatly depend on the filesystem
a47b697c 911 used and even the number of processors in the system. Default: true.
f80dba8d
MT
912
913.. option:: create_fsync=bool
914
22413915 915 :manpage:`fsync(2)` the data file after creation. This is the default.
f80dba8d
MT
916
917.. option:: create_on_open=bool
918
730bd7d9
SW
919 If true, don't pre-create files but allow the job's open() to create a file
920 when it's time to do I/O. Default: false -- pre-create all necessary files
921 when the job starts.
f80dba8d
MT
922
923.. option:: create_only=bool
924
925 If true, fio will only run the setup phase of the job. If files need to be
4502cb42 926 laid out or updated on disk, only that will be done -- the actual job contents
a47b697c 927 are not executed. Default: false.
f80dba8d
MT
928
929.. option:: allow_file_create=bool
930
730bd7d9
SW
931 If true, fio is permitted to create files as part of its workload. If this
932 option is false, then fio will error out if
f80dba8d
MT
933 the files it needs to use don't already exist. Default: true.
934
935.. option:: allow_mounted_write=bool
936
c60ebc45 937 If this isn't set, fio will abort jobs that are destructive (e.g. that write)
f80dba8d
MT
938 to what appears to be a mounted device or partition. This should help catch
939 creating inadvertently destructive tests, not realizing that the test will
b1db0375
TK
940 destroy data on the mounted file system. Note that some platforms don't allow
941 writing against a mounted device regardless of this option. Default: false.
f80dba8d
MT
942
943.. option:: pre_read=bool
944
945 If this is given, files will be pre-read into memory before starting the
946 given I/O operation. This will also clear the :option:`invalidate` flag,
947 since it is pointless to pre-read and then drop the cache. This will only
948 work for I/O engines that are seek-able, since they allow you to read the
a47b697c
SW
949 same data multiple times. Thus it will not work on non-seekable I/O engines
950 (e.g. network, splice). Default: false.
f80dba8d
MT
951
952.. option:: unlink=bool
953
954 Unlink the job files when done. Not the default, as repeated runs of that
a47b697c
SW
955 job would then waste time recreating the file set again and again. Default:
956 false.
f80dba8d
MT
957
958.. option:: unlink_each_loop=bool
959
a47b697c 960 Unlink job files after each iteration or loop. Default: false.
f80dba8d 961
7b865a2f
BVA
962.. option:: zonemode=str
963
964 Accepted values are:
965
966 **none**
967 The :option:`zonerange`, :option:`zonesize` and
968 :option:`zoneskip` parameters are ignored.
969 **strided**
970 I/O happens in a single zone until
971 :option:`zonesize` bytes have been transferred.
972 After that number of bytes has been
973 transferred processing of the next zone
974 starts.
975 **zbd**
976 Zoned block device mode. I/O happens
977 sequentially in each zone, even if random I/O
978 has been selected. Random I/O happens across
979 all zones instead of being restricted to a
980 single zone. The :option:`zoneskip` parameter
981 is ignored. :option:`zonerange` and
982 :option:`zonesize` must be identical.
983
5faddc64 984.. option:: zonerange=int
f80dba8d 985
7b865a2f
BVA
986 Size of a single zone. See also :option:`zonesize` and
987 :option:`zoneskip`.
f80dba8d 988
5faddc64 989.. option:: zonesize=int
f80dba8d 990
7b865a2f
BVA
991 For :option:`zonemode` =strided, this is the number of bytes to
992 transfer before skipping :option:`zoneskip` bytes. If this parameter
993 is smaller than :option:`zonerange` then only a fraction of each zone
994 with :option:`zonerange` bytes will be accessed. If this parameter is
995 larger than :option:`zonerange` then each zone will be accessed
996 multiple times before skipping to the next zone.
997
998 For :option:`zonemode` =zbd, this is the size of a single zone. The
999 :option:`zonerange` parameter is ignored in this mode.
f80dba8d
MT
1000
1001.. option:: zoneskip=int
1002
7b865a2f
BVA
1003 For :option:`zonemode` =strided, the number of bytes to skip after
1004 :option:`zonesize` bytes of data have been transferred. This parameter
1005 must be zero for :option:`zonemode` =zbd.
f80dba8d 1006
bfbdd35b
BVA
1007.. option:: read_beyond_wp=bool
1008
1009 This parameter applies to :option:`zonemode` =zbd only.
1010
1011 Zoned block devices are block devices that consist of multiple zones.
1012 Each zone has a type, e.g. conventional or sequential. A conventional
1013 zone can be written at any offset that is a multiple of the block
1014 size. Sequential zones must be written sequentially. The position at
1015 which a write must occur is called the write pointer. A zoned block
1016 device can be either drive managed, host managed or host aware. For
1017 host managed devices the host must ensure that writes happen
1018 sequentially. Fio recognizes host managed devices and serializes
1019 writes to sequential zones for these devices.
1020
1021 If a read occurs in a sequential zone beyond the write pointer then
1022 the zoned block device will complete the read without reading any data
1023 from the storage medium. Since such reads lead to unrealistically high
1024 bandwidth and IOPS numbers fio only reads beyond the write pointer if
1025 explicitly told to do so. Default: false.
1026
59b07544
BVA
1027.. option:: max_open_zones=int
1028
1029 When running a random write test across an entire drive many more
1030 zones will be open than in a typical application workload. Hence this
1031 command line option that allows to limit the number of open zones. The
1032 number of open zones is defined as the number of zones to which write
1033 commands are issued.
1034
a7c2b6fc
BVA
1035.. option:: zone_reset_threshold=float
1036
1037 A number between zero and one that indicates the ratio of logical
1038 blocks with data to the total number of logical blocks in the test
1039 above which zones should be reset periodically.
1040
1041.. option:: zone_reset_frequency=float
1042
1043 A number between zero and one that indicates how often a zone reset
1044 should be issued if the zone reset threshold has been exceeded. A zone
1045 reset is submitted after each (1 / zone_reset_frequency) write
1046 requests. This and the previous parameter can be used to simulate
1047 garbage collection activity.
1048
f80dba8d
MT
1049
1050I/O type
1051~~~~~~~~
1052
1053.. option:: direct=bool
1054
1055 If value is true, use non-buffered I/O. This is usually O_DIRECT. Note that
8e889110 1056 OpenBSD and ZFS on Solaris don't support direct I/O. On Windows the synchronous
f80dba8d
MT
1057 ioengines don't support direct I/O. Default: false.
1058
1059.. option:: atomic=bool
1060
1061 If value is true, attempt to use atomic direct I/O. Atomic writes are
1062 guaranteed to be stable once acknowledged by the operating system. Only
1063 Linux supports O_ATOMIC right now.
1064
1065.. option:: buffered=bool
1066
1067 If value is true, use buffered I/O. This is the opposite of the
1068 :option:`direct` option. Defaults to true.
1069
1070.. option:: readwrite=str, rw=str
1071
1072 Type of I/O pattern. Accepted values are:
1073
1074 **read**
1075 Sequential reads.
1076 **write**
1077 Sequential writes.
1078 **trim**
3740cfc8
VF
1079 Sequential trims (Linux block devices and SCSI
1080 character devices only).
f80dba8d
MT
1081 **randread**
1082 Random reads.
2831be97
SW
1083 **randwrite**
1084 Random writes.
f80dba8d 1085 **randtrim**
3740cfc8
VF
1086 Random trims (Linux block devices and SCSI
1087 character devices only).
f80dba8d
MT
1088 **rw,readwrite**
1089 Sequential mixed reads and writes.
1090 **randrw**
1091 Random mixed reads and writes.
1092 **trimwrite**
1093 Sequential trim+write sequences. Blocks will be trimmed first,
1094 then the same blocks will be written to.
1095
1096 Fio defaults to read if the option is not specified. For the mixed I/O
1097 types, the default is to split them 50/50. For certain types of I/O the
730bd7d9
SW
1098 result may still be skewed a bit, since the speed may be different.
1099
1100 It is possible to specify the number of I/Os to do before getting a new
1101 offset by appending ``:<nr>`` to the end of the string given. For a
f80dba8d
MT
1102 random read, it would look like ``rw=randread:8`` for passing in an offset
1103 modifier with a value of 8. If the suffix is used with a sequential I/O
730bd7d9
SW
1104 pattern, then the *<nr>* value specified will be **added** to the generated
1105 offset for each I/O turning sequential I/O into sequential I/O with holes.
1106 For instance, using ``rw=write:4k`` will skip 4k for every write. Also see
1107 the :option:`rw_sequencer` option.
f80dba8d
MT
1108
1109.. option:: rw_sequencer=str
1110
1111 If an offset modifier is given by appending a number to the ``rw=<str>``
1112 line, then this option controls how that number modifies the I/O offset
1113 being generated. Accepted values are:
1114
1115 **sequential**
1116 Generate sequential offset.
1117 **identical**
1118 Generate the same offset.
1119
1120 ``sequential`` is only useful for random I/O, where fio would normally
c60ebc45 1121 generate a new random offset for every I/O. If you append e.g. 8 to randread,
007c7be9
SW
1122 you would get a new random offset for every 8 I/Os. The result would be a
1123 seek for only every 8 I/Os, instead of for every I/O. Use ``rw=randread:8``
f80dba8d
MT
1124 to specify that. As sequential I/O is already sequential, setting
1125 ``sequential`` for that would not result in any differences. ``identical``
1126 behaves in a similar fashion, except it sends the same offset 8 number of
1127 times before generating a new offset.
1128
1129.. option:: unified_rw_reporting=bool
1130
1131 Fio normally reports statistics on a per data direction basis, meaning that
1132 reads, writes, and trims are accounted and reported separately. If this
1133 option is set fio sums the results and report them as "mixed" instead.
1134
1135.. option:: randrepeat=bool
1136
1137 Seed the random number generator used for random I/O patterns in a
1138 predictable way so the pattern is repeatable across runs. Default: true.
1139
1140.. option:: allrandrepeat=bool
1141
1142 Seed all random number generators in a predictable way so results are
1143 repeatable across runs. Default: false.
1144
1145.. option:: randseed=int
1146
1147 Seed the random number generators based on this seed value, to be able to
1148 control what sequence of output is being generated. If not set, the random
1149 sequence depends on the :option:`randrepeat` setting.
1150
1151.. option:: fallocate=str
1152
1153 Whether pre-allocation is performed when laying down files.
1154 Accepted values are:
1155
1156 **none**
1157 Do not pre-allocate space.
1158
2c3e17be
SW
1159 **native**
1160 Use a platform's native pre-allocation call but fall back to
1161 **none** behavior if it fails/is not implemented.
1162
f80dba8d
MT
1163 **posix**
1164 Pre-allocate via :manpage:`posix_fallocate(3)`.
1165
1166 **keep**
1167 Pre-allocate via :manpage:`fallocate(2)` with
1168 FALLOC_FL_KEEP_SIZE set.
1169
1170 **0**
1171 Backward-compatible alias for **none**.
1172
1173 **1**
1174 Backward-compatible alias for **posix**.
1175
1176 May not be available on all supported platforms. **keep** is only available
2c3e17be
SW
1177 on Linux. If using ZFS on Solaris this cannot be set to **posix**
1178 because ZFS doesn't support pre-allocation. Default: **native** if any
1179 pre-allocation methods are available, **none** if not.
f80dba8d
MT
1180
1181.. option:: fadvise_hint=str
1182
c712c97a
JA
1183 Use :manpage:`posix_fadvise(2)` or :manpage:`posix_fadvise(2)` to
1184 advise the kernel on what I/O patterns are likely to be issued.
1185 Accepted values are:
f80dba8d
MT
1186
1187 **0**
1188 Backwards-compatible hint for "no hint".
1189
1190 **1**
1191 Backwards compatible hint for "advise with fio workload type". This
1192 uses **FADV_RANDOM** for a random workload, and **FADV_SEQUENTIAL**
1193 for a sequential workload.
1194
1195 **sequential**
1196 Advise using **FADV_SEQUENTIAL**.
1197
1198 **random**
1199 Advise using **FADV_RANDOM**.
1200
8f4b9f24 1201.. option:: write_hint=str
f80dba8d 1202
8f4b9f24
JA
1203 Use :manpage:`fcntl(2)` to advise the kernel what life time to expect
1204 from a write. Only supported on Linux, as of version 4.13. Accepted
1205 values are:
1206
1207 **none**
1208 No particular life time associated with this file.
1209
1210 **short**
1211 Data written to this file has a short life time.
1212
1213 **medium**
1214 Data written to this file has a medium life time.
1215
1216 **long**
1217 Data written to this file has a long life time.
1218
1219 **extreme**
1220 Data written to this file has a very long life time.
1221
1222 The values are all relative to each other, and no absolute meaning
1223 should be associated with them.
f80dba8d
MT
1224
1225.. option:: offset=int
1226
82dbb8cb 1227 Start I/O at the provided offset in the file, given as either a fixed size in
83c8b093
JF
1228 bytes or a percentage. If a percentage is given, the generated offset will be
1229 aligned to the minimum ``blocksize`` or to the value of ``offset_align`` if
1230 provided. Data before the given offset will not be touched. This
89978a6b
BW
1231 effectively caps the file size at `real_size - offset`. Can be combined with
1232 :option:`size` to constrain the start and end range of the I/O workload.
44bb1142
TK
1233 A percentage can be specified by a number between 1 and 100 followed by '%',
1234 for example, ``offset=20%`` to specify 20%.
f80dba8d 1235
83c8b093
JF
1236.. option:: offset_align=int
1237
1238 If set to non-zero value, the byte offset generated by a percentage ``offset``
1239 is aligned upwards to this value. Defaults to 0 meaning that a percentage
1240 offset is aligned to the minimum block size.
1241
f80dba8d
MT
1242.. option:: offset_increment=int
1243
1244 If this is provided, then the real offset becomes `offset + offset_increment
1245 * thread_number`, where the thread number is a counter that starts at 0 and
1246 is incremented for each sub-job (i.e. when :option:`numjobs` option is
1247 specified). This option is useful if there are several jobs which are
1248 intended to operate on a file in parallel disjoint segments, with even
1249 spacing between the starting points.
1250
1251.. option:: number_ios=int
1252
c60ebc45 1253 Fio will normally perform I/Os until it has exhausted the size of the region
f80dba8d
MT
1254 set by :option:`size`, or if it exhaust the allocated time (or hits an error
1255 condition). With this setting, the range/size can be set independently of
c60ebc45 1256 the number of I/Os to perform. When fio reaches this number, it will exit
f80dba8d
MT
1257 normally and report status. Note that this does not extend the amount of I/O
1258 that will be done, it will only stop fio if this condition is met before
1259 other end-of-job criteria.
1260
1261.. option:: fsync=int
1262
730bd7d9
SW
1263 If writing to a file, issue an :manpage:`fsync(2)` (or its equivalent) of
1264 the dirty data for every number of blocks given. For example, if you give 32
1265 as a parameter, fio will sync the file after every 32 writes issued. If fio is
1266 using non-buffered I/O, we may not sync the file. The exception is the sg
1267 I/O engine, which synchronizes the disk cache anyway. Defaults to 0, which
1268 means fio does not periodically issue and wait for a sync to complete. Also
1269 see :option:`end_fsync` and :option:`fsync_on_close`.
f80dba8d
MT
1270
1271.. option:: fdatasync=int
1272
1273 Like :option:`fsync` but uses :manpage:`fdatasync(2)` to only sync data and
000a5f1c 1274 not metadata blocks. In Windows, FreeBSD, and DragonFlyBSD there is no
730bd7d9
SW
1275 :manpage:`fdatasync(2)` so this falls back to using :manpage:`fsync(2)`.
1276 Defaults to 0, which means fio does not periodically issue and wait for a
1277 data-only sync to complete.
f80dba8d
MT
1278
1279.. option:: write_barrier=int
1280
2831be97 1281 Make every `N-th` write a barrier write.
f80dba8d 1282
f50fbdda 1283.. option:: sync_file_range=str:int
f80dba8d 1284
f50fbdda 1285 Use :manpage:`sync_file_range(2)` for every `int` number of write
f80dba8d
MT
1286 operations. Fio will track range of writes that have happened since the last
1287 :manpage:`sync_file_range(2)` call. `str` can currently be one or more of:
1288
1289 **wait_before**
1290 SYNC_FILE_RANGE_WAIT_BEFORE
1291 **write**
1292 SYNC_FILE_RANGE_WRITE
1293 **wait_after**
1294 SYNC_FILE_RANGE_WAIT_AFTER
1295
1296 So if you do ``sync_file_range=wait_before,write:8``, fio would use
1297 ``SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE`` for every 8
1298 writes. Also see the :manpage:`sync_file_range(2)` man page. This option is
1299 Linux specific.
1300
1301.. option:: overwrite=bool
1302
1303 If true, writes to a file will always overwrite existing data. If the file
1304 doesn't already exist, it will be created before the write phase begins. If
1305 the file exists and is large enough for the specified write phase, nothing
a47b697c 1306 will be done. Default: false.
f80dba8d
MT
1307
1308.. option:: end_fsync=bool
1309
a47b697c
SW
1310 If true, :manpage:`fsync(2)` file contents when a write stage has completed.
1311 Default: false.
f80dba8d
MT
1312
1313.. option:: fsync_on_close=bool
1314
1315 If true, fio will :manpage:`fsync(2)` a dirty file on close. This differs
a47b697c
SW
1316 from :option:`end_fsync` in that it will happen on every file close, not
1317 just at the end of the job. Default: false.
f80dba8d
MT
1318
1319.. option:: rwmixread=int
1320
1321 Percentage of a mixed workload that should be reads. Default: 50.
1322
1323.. option:: rwmixwrite=int
1324
1325 Percentage of a mixed workload that should be writes. If both
1326 :option:`rwmixread` and :option:`rwmixwrite` is given and the values do not
1327 add up to 100%, the latter of the two will be used to override the
1328 first. This may interfere with a given rate setting, if fio is asked to
1329 limit reads or writes to a certain rate. If that is the case, then the
1330 distribution may be skewed. Default: 50.
1331
1332.. option:: random_distribution=str:float[,str:float][,str:float]
1333
1334 By default, fio will use a completely uniform random distribution when asked
1335 to perform random I/O. Sometimes it is useful to skew the distribution in
1336 specific ways, ensuring that some parts of the data is more hot than others.
1337 fio includes the following distribution models:
1338
1339 **random**
1340 Uniform random distribution
1341
1342 **zipf**
1343 Zipf distribution
1344
1345 **pareto**
1346 Pareto distribution
1347
b2f4b559 1348 **normal**
c60ebc45 1349 Normal (Gaussian) distribution
f80dba8d
MT
1350
1351 **zoned**
1352 Zoned random distribution
1353
59466396
JA
1354 **zoned_abs**
1355 Zone absolute random distribution
1356
f80dba8d 1357 When using a **zipf** or **pareto** distribution, an input value is also
f50fbdda 1358 needed to define the access pattern. For **zipf**, this is the `Zipf
c60ebc45 1359 theta`. For **pareto**, it's the `Pareto power`. Fio includes a test
f50fbdda 1360 program, :command:`fio-genzipf`, that can be used visualize what the given input
f80dba8d
MT
1361 values will yield in terms of hit rates. If you wanted to use **zipf** with
1362 a `theta` of 1.2, you would use ``random_distribution=zipf:1.2`` as the
1363 option. If a non-uniform model is used, fio will disable use of the random
b2f4b559
SW
1364 map. For the **normal** distribution, a normal (Gaussian) deviation is
1365 supplied as a value between 0 and 100.
f80dba8d
MT
1366
1367 For a **zoned** distribution, fio supports specifying percentages of I/O
1368 access that should fall within what range of the file or device. For
1369 example, given a criteria of:
1370
f50fbdda
TK
1371 * 60% of accesses should be to the first 10%
1372 * 30% of accesses should be to the next 20%
1373 * 8% of accesses should be to the next 30%
1374 * 2% of accesses should be to the next 40%
f80dba8d
MT
1375
1376 we can define that through zoning of the random accesses. For the above
1377 example, the user would do::
1378
1379 random_distribution=zoned:60/10:30/20:8/30:2/40
1380
59466396
JA
1381 A **zoned_abs** distribution works exactly like the **zoned**, except
1382 that it takes absolute sizes. For example, let's say you wanted to
1383 define access according to the following criteria:
1384
1385 * 60% of accesses should be to the first 20G
1386 * 30% of accesses should be to the next 100G
1387 * 10% of accesses should be to the next 500G
1388
1389 we can define an absolute zoning distribution with:
1390
1391 random_distribution=zoned_abs=60/20G:30/100G:10/500g
1392
6a16ece8
JA
1393 For both **zoned** and **zoned_abs**, fio supports defining up to
1394 256 separate zones.
1395
59466396
JA
1396 Similarly to how :option:`bssplit` works for setting ranges and
1397 percentages of block sizes. Like :option:`bssplit`, it's possible to
1398 specify separate zones for reads, writes, and trims. If just one set
1399 is given, it'll apply to all of them. This goes for both **zoned**
1400 **zoned_abs** distributions.
f80dba8d
MT
1401
1402.. option:: percentage_random=int[,int][,int]
1403
1404 For a random workload, set how big a percentage should be random. This
1405 defaults to 100%, in which case the workload is fully random. It can be set
1406 from anywhere from 0 to 100. Setting it to 0 would make the workload fully
1407 sequential. Any setting in between will result in a random mix of sequential
1408 and random I/O, at the given percentages. Comma-separated values may be
1409 specified for reads, writes, and trims as described in :option:`blocksize`.
1410
1411.. option:: norandommap
1412
1413 Normally fio will cover every block of the file when doing random I/O. If
1414 this option is given, fio will just get a new random offset without looking
1415 at past I/O history. This means that some blocks may not be read or written,
1416 and that some blocks may be read/written more than once. If this option is
1417 used with :option:`verify` and multiple blocksizes (via :option:`bsrange`),
1418 only intact blocks are verified, i.e., partially-overwritten blocks are
47e6a6e5
SW
1419 ignored. With an async I/O engine and an I/O depth > 1, it is possible for
1420 the same block to be overwritten, which can cause verification errors. Either
1421 do not use norandommap in this case, or also use the lfsr random generator.
f80dba8d
MT
1422
1423.. option:: softrandommap=bool
1424
1425 See :option:`norandommap`. If fio runs with the random block map enabled and
1426 it fails to allocate the map, if this option is set it will continue without
1427 a random block map. As coverage will not be as complete as with random maps,
1428 this option is disabled by default.
1429
1430.. option:: random_generator=str
1431
f50fbdda 1432 Fio supports the following engines for generating I/O offsets for random I/O:
f80dba8d
MT
1433
1434 **tausworthe**
f50fbdda 1435 Strong 2^88 cycle random number generator.
f80dba8d 1436 **lfsr**
f50fbdda 1437 Linear feedback shift register generator.
f80dba8d 1438 **tausworthe64**
f50fbdda 1439 Strong 64-bit 2^258 cycle random number generator.
f80dba8d
MT
1440
1441 **tausworthe** is a strong random number generator, but it requires tracking
1442 on the side if we want to ensure that blocks are only read or written
f50fbdda 1443 once. **lfsr** guarantees that we never generate the same offset twice, and
f80dba8d 1444 it's also less computationally expensive. It's not a true random generator,
f50fbdda 1445 however, though for I/O purposes it's typically good enough. **lfsr** only
f80dba8d
MT
1446 works with single block sizes, not with workloads that use multiple block
1447 sizes. If used with such a workload, fio may read or write some blocks
1448 multiple times. The default value is **tausworthe**, unless the required
1449 space exceeds 2^32 blocks. If it does, then **tausworthe64** is
1450 selected automatically.
1451
1452
1453Block size
1454~~~~~~~~~~
1455
1456.. option:: blocksize=int[,int][,int], bs=int[,int][,int]
1457
1458 The block size in bytes used for I/O units. Default: 4096. A single value
1459 applies to reads, writes, and trims. Comma-separated values may be
1460 specified for reads, writes, and trims. A value not terminated in a comma
1461 applies to subsequent types.
1462
1463 Examples:
1464
1465 **bs=256k**
1466 means 256k for reads, writes and trims.
1467
1468 **bs=8k,32k**
1469 means 8k for reads, 32k for writes and trims.
1470
1471 **bs=8k,32k,**
1472 means 8k for reads, 32k for writes, and default for trims.
1473
1474 **bs=,8k**
1475 means default for reads, 8k for writes and trims.
1476
1477 **bs=,8k,**
b443ae44 1478 means default for reads, 8k for writes, and default for trims.
f80dba8d
MT
1479
1480.. option:: blocksize_range=irange[,irange][,irange], bsrange=irange[,irange][,irange]
1481
1482 A range of block sizes in bytes for I/O units. The issued I/O unit will
1483 always be a multiple of the minimum size, unless
1484 :option:`blocksize_unaligned` is set.
1485
1486 Comma-separated ranges may be specified for reads, writes, and trims as
1487 described in :option:`blocksize`.
1488
1489 Example: ``bsrange=1k-4k,2k-8k``.
1490
1491.. option:: bssplit=str[,str][,str]
1492
6a16ece8
JA
1493 Sometimes you want even finer grained control of the block sizes
1494 issued, not just an even split between them. This option allows you to
1495 weight various block sizes, so that you are able to define a specific
1496 amount of block sizes issued. The format for this option is::
f80dba8d
MT
1497
1498 bssplit=blocksize/percentage:blocksize/percentage
1499
6a16ece8
JA
1500 for as many block sizes as needed. So if you want to define a workload
1501 that has 50% 64k blocks, 10% 4k blocks, and 40% 32k blocks, you would
1502 write::
f80dba8d
MT
1503
1504 bssplit=4k/10:64k/50:32k/40
1505
6a16ece8
JA
1506 Ordering does not matter. If the percentage is left blank, fio will
1507 fill in the remaining values evenly. So a bssplit option like this one::
f80dba8d
MT
1508
1509 bssplit=4k/50:1k/:32k/
1510
6a16ece8
JA
1511 would have 50% 4k ios, and 25% 1k and 32k ios. The percentages always
1512 add up to 100, if bssplit is given a range that adds up to more, it
1513 will error out.
f80dba8d
MT
1514
1515 Comma-separated values may be specified for reads, writes, and trims as
1516 described in :option:`blocksize`.
1517
6a16ece8
JA
1518 If you want a workload that has 50% 2k reads and 50% 4k reads, while
1519 having 90% 4k writes and 10% 8k writes, you would specify::
f80dba8d 1520
cf04b906 1521 bssplit=2k/50:4k/50,4k/90:8k/10
f80dba8d 1522
6a16ece8
JA
1523 Fio supports defining up to 64 different weights for each data
1524 direction.
1525
f80dba8d
MT
1526.. option:: blocksize_unaligned, bs_unaligned
1527
1528 If set, fio will issue I/O units with any size within
1529 :option:`blocksize_range`, not just multiples of the minimum size. This
1530 typically won't work with direct I/O, as that normally requires sector
1531 alignment.
1532
589e88b7 1533.. option:: bs_is_seq_rand=bool
f80dba8d
MT
1534
1535 If this option is set, fio will use the normal read,write blocksize settings
1536 as sequential,random blocksize settings instead. Any random read or write
1537 will use the WRITE blocksize settings, and any sequential read or write will
1538 use the READ blocksize settings.
1539
1540.. option:: blockalign=int[,int][,int], ba=int[,int][,int]
1541
1542 Boundary to which fio will align random I/O units. Default:
1543 :option:`blocksize`. Minimum alignment is typically 512b for using direct
1544 I/O, though it usually depends on the hardware block size. This option is
1545 mutually exclusive with using a random map for files, so it will turn off
1546 that option. Comma-separated values may be specified for reads, writes, and
1547 trims as described in :option:`blocksize`.
1548
1549
1550Buffers and memory
1551~~~~~~~~~~~~~~~~~~
1552
1553.. option:: zero_buffers
1554
1555 Initialize buffers with all zeros. Default: fill buffers with random data.
1556
1557.. option:: refill_buffers
1558
1559 If this option is given, fio will refill the I/O buffers on every
72592780
SW
1560 submit. Only makes sense if :option:`zero_buffers` isn't specified,
1561 naturally. Defaults to being unset i.e., the buffer is only filled at
1562 init time and the data in it is reused when possible but if any of
1563 :option:`verify`, :option:`buffer_compress_percentage` or
1564 :option:`dedupe_percentage` are enabled then `refill_buffers` is also
1565 automatically enabled.
f80dba8d
MT
1566
1567.. option:: scramble_buffers=bool
1568
1569 If :option:`refill_buffers` is too costly and the target is using data
1570 deduplication, then setting this option will slightly modify the I/O buffer
1571 contents to defeat normal de-dupe attempts. This is not enough to defeat
1572 more clever block compression attempts, but it will stop naive dedupe of
1573 blocks. Default: true.
1574
1575.. option:: buffer_compress_percentage=int
1576
72592780
SW
1577 If this is set, then fio will attempt to provide I/O buffer content
1578 (on WRITEs) that compresses to the specified level. Fio does this by
1579 providing a mix of random data followed by fixed pattern data. The
1580 fixed pattern is either zeros, or the pattern specified by
1581 :option:`buffer_pattern`. If the `buffer_pattern` option is used, it
1582 might skew the compression ratio slightly. Setting
1583 `buffer_compress_percentage` to a value other than 100 will also
1584 enable :option:`refill_buffers` in order to reduce the likelihood that
1585 adjacent blocks are so similar that they over compress when seen
1586 together. See :option:`buffer_compress_chunk` for how to set a finer or
1587 coarser granularity for the random/fixed data region. Defaults to unset
1588 i.e., buffer data will not adhere to any compression level.
f80dba8d
MT
1589
1590.. option:: buffer_compress_chunk=int
1591
72592780
SW
1592 This setting allows fio to manage how big the random/fixed data region
1593 is when using :option:`buffer_compress_percentage`. When
1594 `buffer_compress_chunk` is set to some non-zero value smaller than the
1595 block size, fio can repeat the random/fixed region throughout the I/O
1596 buffer at the specified interval (which particularly useful when
1597 bigger block sizes are used for a job). When set to 0, fio will use a
1598 chunk size that matches the block size resulting in a single
1599 random/fixed region within the I/O buffer. Defaults to 512. When the
1600 unit is omitted, the value is interpreted in bytes.
f80dba8d
MT
1601
1602.. option:: buffer_pattern=str
1603
a1554f65
SB
1604 If set, fio will fill the I/O buffers with this pattern or with the contents
1605 of a file. If not set, the contents of I/O buffers are defined by the other
1606 options related to buffer contents. The setting can be any pattern of bytes,
1607 and can be prefixed with 0x for hex values. It may also be a string, where
1608 the string must then be wrapped with ``""``. Or it may also be a filename,
1609 where the filename must be wrapped with ``''`` in which case the file is
1610 opened and read. Note that not all the file contents will be read if that
1611 would cause the buffers to overflow. So, for example::
1612
1613 buffer_pattern='filename'
1614
1615 or::
f80dba8d
MT
1616
1617 buffer_pattern="abcd"
1618
1619 or::
1620
1621 buffer_pattern=-12
1622
1623 or::
1624
1625 buffer_pattern=0xdeadface
1626
1627 Also you can combine everything together in any order::
1628
a1554f65 1629 buffer_pattern=0xdeadface"abcd"-12'filename'
f80dba8d
MT
1630
1631.. option:: dedupe_percentage=int
1632
1633 If set, fio will generate this percentage of identical buffers when
1634 writing. These buffers will be naturally dedupable. The contents of the
1635 buffers depend on what other buffer compression settings have been set. It's
1636 possible to have the individual buffers either fully compressible, or not at
72592780
SW
1637 all -- this option only controls the distribution of unique buffers. Setting
1638 this option will also enable :option:`refill_buffers` to prevent every buffer
1639 being identical.
f80dba8d
MT
1640
1641.. option:: invalidate=bool
1642
730bd7d9
SW
1643 Invalidate the buffer/page cache parts of the files to be used prior to
1644 starting I/O if the platform and file type support it. Defaults to true.
21c1b29e
TK
1645 This will be ignored if :option:`pre_read` is also specified for the
1646 same job.
f80dba8d
MT
1647
1648.. option:: sync=bool
1649
1650 Use synchronous I/O for buffered writes. For the majority of I/O engines,
1651 this means using O_SYNC. Default: false.
1652
1653.. option:: iomem=str, mem=str
1654
1655 Fio can use various types of memory as the I/O unit buffer. The allowed
1656 values are:
1657
1658 **malloc**
1659 Use memory from :manpage:`malloc(3)` as the buffers. Default memory
1660 type.
1661
1662 **shm**
1663 Use shared memory as the buffers. Allocated through
1664 :manpage:`shmget(2)`.
1665
1666 **shmhuge**
1667 Same as shm, but use huge pages as backing.
1668
1669 **mmap**
22413915 1670 Use :manpage:`mmap(2)` to allocate buffers. May either be anonymous memory, or can
f80dba8d
MT
1671 be file backed if a filename is given after the option. The format
1672 is `mem=mmap:/path/to/file`.
1673
1674 **mmaphuge**
1675 Use a memory mapped huge file as the buffer backing. Append filename
1676 after mmaphuge, ala `mem=mmaphuge:/hugetlbfs/file`.
1677
1678 **mmapshared**
1679 Same as mmap, but use a MMAP_SHARED mapping.
1680
03553853
YR
1681 **cudamalloc**
1682 Use GPU memory as the buffers for GPUDirect RDMA benchmark.
f50fbdda 1683 The :option:`ioengine` must be `rdma`.
03553853 1684
f80dba8d
MT
1685 The area allocated is a function of the maximum allowed bs size for the job,
1686 multiplied by the I/O depth given. Note that for **shmhuge** and
1687 **mmaphuge** to work, the system must have free huge pages allocated. This
1688 can normally be checked and set by reading/writing
1689 :file:`/proc/sys/vm/nr_hugepages` on a Linux system. Fio assumes a huge page
1690 is 4MiB in size. So to calculate the number of huge pages you need for a
1691 given job file, add up the I/O depth of all jobs (normally one unless
1692 :option:`iodepth` is used) and multiply by the maximum bs set. Then divide
1693 that number by the huge page size. You can see the size of the huge pages in
1694 :file:`/proc/meminfo`. If no huge pages are allocated by having a non-zero
1695 number in `nr_hugepages`, using **mmaphuge** or **shmhuge** will fail. Also
1696 see :option:`hugepage-size`.
1697
1698 **mmaphuge** also needs to have hugetlbfs mounted and the file location
1699 should point there. So if it's mounted in :file:`/huge`, you would use
1700 `mem=mmaphuge:/huge/somefile`.
1701
f50fbdda 1702.. option:: iomem_align=int, mem_align=int
f80dba8d
MT
1703
1704 This indicates the memory alignment of the I/O memory buffers. Note that
1705 the given alignment is applied to the first I/O unit buffer, if using
1706 :option:`iodepth` the alignment of the following buffers are given by the
1707 :option:`bs` used. In other words, if using a :option:`bs` that is a
1708 multiple of the page sized in the system, all buffers will be aligned to
1709 this value. If using a :option:`bs` that is not page aligned, the alignment
1710 of subsequent I/O memory buffers is the sum of the :option:`iomem_align` and
1711 :option:`bs` used.
1712
1713.. option:: hugepage-size=int
1714
1715 Defines the size of a huge page. Must at least be equal to the system
1716 setting, see :file:`/proc/meminfo`. Defaults to 4MiB. Should probably
1717 always be a multiple of megabytes, so using ``hugepage-size=Xm`` is the
1718 preferred way to set this to avoid setting a non-pow-2 bad value.
1719
1720.. option:: lockmem=int
1721
1722 Pin the specified amount of memory with :manpage:`mlock(2)`. Can be used to
1723 simulate a smaller amount of memory. The amount specified is per worker.
1724
1725
1726I/O size
1727~~~~~~~~
1728
1729.. option:: size=int
1730
79591fa9
TK
1731 The total size of file I/O for each thread of this job. Fio will run until
1732 this many bytes has been transferred, unless runtime is limited by other options
1733 (such as :option:`runtime`, for instance, or increased/decreased by :option:`io_size`).
1734 Fio will divide this size between the available files determined by options
1735 such as :option:`nrfiles`, :option:`filename`, unless :option:`filesize` is
1736 specified by the job. If the result of division happens to be 0, the size is
c4aa2d08 1737 set to the physical size of the given files or devices if they exist.
79591fa9 1738 If this option is not specified, fio will use the full size of the given
f80dba8d
MT
1739 files or devices. If the files do not exist, size must be given. It is also
1740 possible to give size as a percentage between 1 and 100. If ``size=20%`` is
1741 given, fio will use 20% of the full size of the given files or devices.
9d25d068
SW
1742 Can be combined with :option:`offset` to constrain the start and end range
1743 that I/O will be done within.
f80dba8d
MT
1744
1745.. option:: io_size=int, io_limit=int
1746
1747 Normally fio operates within the region set by :option:`size`, which means
1748 that the :option:`size` option sets both the region and size of I/O to be
1749 performed. Sometimes that is not what you want. With this option, it is
1750 possible to define just the amount of I/O that fio should do. For instance,
1751 if :option:`size` is set to 20GiB and :option:`io_size` is set to 5GiB, fio
1752 will perform I/O within the first 20GiB but exit when 5GiB have been
1753 done. The opposite is also possible -- if :option:`size` is set to 20GiB,
1754 and :option:`io_size` is set to 40GiB, then fio will do 40GiB of I/O within
1755 the 0..20GiB region.
1756
7fdd97ca 1757.. option:: filesize=irange(int)
f80dba8d
MT
1758
1759 Individual file sizes. May be a range, in which case fio will select sizes
1760 for files at random within the given range and limited to :option:`size` in
1761 total (if that is given). If not given, each created file is the same size.
79591fa9
TK
1762 This option overrides :option:`size` in terms of file size, which means
1763 this value is used as a fixed size or possible range of each file.
f80dba8d
MT
1764
1765.. option:: file_append=bool
1766
1767 Perform I/O after the end of the file. Normally fio will operate within the
1768 size of a file. If this option is set, then fio will append to the file
1769 instead. This has identical behavior to setting :option:`offset` to the size
1770 of a file. This option is ignored on non-regular files.
1771
1772.. option:: fill_device=bool, fill_fs=bool
1773
1774 Sets size to something really large and waits for ENOSPC (no space left on
1775 device) as the terminating condition. Only makes sense with sequential
1776 write. For a read workload, the mount point will be filled first then I/O
1777 started on the result. This option doesn't make sense if operating on a raw
1778 device node, since the size of that is already known by the file system.
1779 Additionally, writing beyond end-of-device will not return ENOSPC there.
1780
1781
1782I/O engine
1783~~~~~~~~~~
1784
1785.. option:: ioengine=str
1786
1787 Defines how the job issues I/O to the file. The following types are defined:
1788
1789 **sync**
1790 Basic :manpage:`read(2)` or :manpage:`write(2)`
1791 I/O. :manpage:`lseek(2)` is used to position the I/O location.
54227e6b 1792 See :option:`fsync` and :option:`fdatasync` for syncing write I/Os.
f80dba8d
MT
1793
1794 **psync**
1795 Basic :manpage:`pread(2)` or :manpage:`pwrite(2)` I/O. Default on
1796 all supported operating systems except for Windows.
1797
1798 **vsync**
1799 Basic :manpage:`readv(2)` or :manpage:`writev(2)` I/O. Will emulate
c60ebc45 1800 queuing by coalescing adjacent I/Os into a single submission.
f80dba8d
MT
1801
1802 **pvsync**
1803 Basic :manpage:`preadv(2)` or :manpage:`pwritev(2)` I/O.
1804
1805 **pvsync2**
1806 Basic :manpage:`preadv2(2)` or :manpage:`pwritev2(2)` I/O.
1807
1808 **libaio**
1809 Linux native asynchronous I/O. Note that Linux may only support
22413915 1810 queued behavior with non-buffered I/O (set ``direct=1`` or
f80dba8d
MT
1811 ``buffered=0``).
1812 This engine defines engine specific options.
1813
1814 **posixaio**
1815 POSIX asynchronous I/O using :manpage:`aio_read(3)` and
1816 :manpage:`aio_write(3)`.
1817
1818 **solarisaio**
1819 Solaris native asynchronous I/O.
1820
1821 **windowsaio**
1822 Windows native asynchronous I/O. Default on Windows.
1823
1824 **mmap**
1825 File is memory mapped with :manpage:`mmap(2)` and data copied
1826 to/from using :manpage:`memcpy(3)`.
1827
1828 **splice**
1829 :manpage:`splice(2)` is used to transfer the data and
1830 :manpage:`vmsplice(2)` to transfer data from user space to the
1831 kernel.
1832
1833 **sg**
1834 SCSI generic sg v3 I/O. May either be synchronous using the SG_IO
1835 ioctl, or if the target is an sg character device we use
1836 :manpage:`read(2)` and :manpage:`write(2)` for asynchronous
f50fbdda 1837 I/O. Requires :option:`filename` option to specify either block or
3740cfc8 1838 character devices. This engine supports trim operations.
52b81b7c 1839 The sg engine includes engine specific options.
f80dba8d
MT
1840
1841 **null**
1842 Doesn't transfer any data, just pretends to. This is mainly used to
1843 exercise fio itself and for debugging/testing purposes.
1844
1845 **net**
1846 Transfer over the network to given ``host:port``. Depending on the
1847 :option:`protocol` used, the :option:`hostname`, :option:`port`,
1848 :option:`listen` and :option:`filename` options are used to specify
1849 what sort of connection to make, while the :option:`protocol` option
1850 determines which protocol will be used. This engine defines engine
1851 specific options.
1852
1853 **netsplice**
1854 Like **net**, but uses :manpage:`splice(2)` and
1855 :manpage:`vmsplice(2)` to map data and send/receive.
1856 This engine defines engine specific options.
1857
1858 **cpuio**
1859 Doesn't transfer any data, but burns CPU cycles according to the
1860 :option:`cpuload` and :option:`cpuchunks` options. Setting
9207a0cb 1861 :option:`cpuload`\=85 will cause that job to do nothing but burn 85%
71aa48eb 1862 of the CPU. In case of SMP machines, use :option:`numjobs`\=<nr_of_cpu>
f50fbdda 1863 to get desired CPU usage, as the cpuload only loads a
f80dba8d
MT
1864 single CPU at the desired rate. A job never finishes unless there is
1865 at least one non-cpuio job.
1866
1867 **guasi**
804c0839 1868 The GUASI I/O engine is the Generic Userspace Asynchronous Syscall
f80dba8d
MT
1869 Interface approach to async I/O. See
1870
1871 http://www.xmailserver.org/guasi-lib.html
1872
1873 for more info on GUASI.
1874
1875 **rdma**
1876 The RDMA I/O engine supports both RDMA memory semantics
1877 (RDMA_WRITE/RDMA_READ) and channel semantics (Send/Recv) for the
609ac152
SB
1878 InfiniBand, RoCE and iWARP protocols. This engine defines engine
1879 specific options.
f80dba8d
MT
1880
1881 **falloc**
1882 I/O engine that does regular fallocate to simulate data transfer as
1883 fio ioengine.
1884
1885 DDIR_READ
1886 does fallocate(,mode = FALLOC_FL_KEEP_SIZE,).
1887
1888 DDIR_WRITE
1889 does fallocate(,mode = 0).
1890
1891 DDIR_TRIM
1892 does fallocate(,mode = FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE).
1893
761cd093
SW
1894 **ftruncate**
1895 I/O engine that sends :manpage:`ftruncate(2)` operations in response
1896 to write (DDIR_WRITE) events. Each ftruncate issued sets the file's
f50fbdda 1897 size to the current block offset. :option:`blocksize` is ignored.
761cd093 1898
f80dba8d
MT
1899 **e4defrag**
1900 I/O engine that does regular EXT4_IOC_MOVE_EXT ioctls to simulate
1901 defragment activity in request to DDIR_WRITE event.
1902
f3f96717
IF
1903 **rados**
1904 I/O engine supporting direct access to Ceph Reliable Autonomic
1905 Distributed Object Store (RADOS) via librados. This ioengine
1906 defines engine specific options.
1907
f80dba8d
MT
1908 **rbd**
1909 I/O engine supporting direct access to Ceph Rados Block Devices
1910 (RBD) via librbd without the need to use the kernel rbd driver. This
1911 ioengine defines engine specific options.
1912
c2f6a13d
LMB
1913 **http**
1914 I/O engine supporting GET/PUT requests over HTTP(S) with libcurl to
1915 a WebDAV or S3 endpoint. This ioengine defines engine specific options.
1916
1917 This engine only supports direct IO of iodepth=1; you need to scale this
1918 via numjobs. blocksize defines the size of the objects to be created.
1919
1920 TRIM is translated to object deletion.
1921
f80dba8d 1922 **gfapi**
ac8ca2af
SW
1923 Using GlusterFS libgfapi sync interface to direct access to
1924 GlusterFS volumes without having to go through FUSE. This ioengine
f80dba8d
MT
1925 defines engine specific options.
1926
1927 **gfapi_async**
ac8ca2af
SW
1928 Using GlusterFS libgfapi async interface to direct access to
1929 GlusterFS volumes without having to go through FUSE. This ioengine
f80dba8d
MT
1930 defines engine specific options.
1931
1932 **libhdfs**
f50fbdda 1933 Read and write through Hadoop (HDFS). The :option:`filename` option
f80dba8d
MT
1934 is used to specify host,port of the hdfs name-node to connect. This
1935 engine interprets offsets a little differently. In HDFS, files once
e25c0c91
SW
1936 created cannot be modified so random writes are not possible. To
1937 imitate this the libhdfs engine expects a bunch of small files to be
1938 created over HDFS and will randomly pick a file from them
1939 based on the offset generated by fio backend (see the example
f80dba8d 1940 job file to create such files, use ``rw=write`` option). Please
e25c0c91
SW
1941 note, it may be necessary to set environment variables to work
1942 with HDFS/libhdfs properly. Each job uses its own connection to
f80dba8d
MT
1943 HDFS.
1944
1945 **mtd**
1946 Read, write and erase an MTD character device (e.g.,
1947 :file:`/dev/mtd0`). Discards are treated as erases. Depending on the
1948 underlying device type, the I/O may have to go in a certain pattern,
1949 e.g., on NAND, writing sequentially to erase blocks and discarding
c298ee71 1950 before overwriting. The `trimwrite` mode works well for this
f80dba8d
MT
1951 constraint.
1952
1953 **pmemblk**
1954 Read and write using filesystem DAX to a file on a filesystem
363a5f65 1955 mounted with DAX on a persistent memory device through the PMDK
f80dba8d
MT
1956 libpmemblk library.
1957
1958 **dev-dax**
1959 Read and write using device DAX to a persistent memory device (e.g.,
363a5f65 1960 /dev/dax0.0) through the PMDK libpmem library.
f80dba8d
MT
1961
1962 **external**
1963 Prefix to specify loading an external I/O engine object file. Append
c60ebc45 1964 the engine filename, e.g. ``ioengine=external:/tmp/foo.o`` to load
d243fd6d
TK
1965 ioengine :file:`foo.o` in :file:`/tmp`. The path can be either
1966 absolute or relative. See :file:`engines/skeleton_external.c` for
1967 details of writing an external I/O engine.
f80dba8d 1968
1216cc5a 1969 **filecreate**
b71968b1 1970 Simply create the files and do no I/O to them. You still need to
1216cc5a 1971 set `filesize` so that all the accounting still occurs, but no
b71968b1 1972 actual I/O will be done other than creating the file.
f80dba8d 1973
ae0db592
TI
1974 **libpmem**
1975 Read and write using mmap I/O to a file on a filesystem
363a5f65 1976 mounted with DAX on a persistent memory device through the PMDK
ae0db592
TI
1977 libpmem library.
1978
a40e7a59
GB
1979 **ime_psync**
1980 Synchronous read and write using DDN's Infinite Memory Engine (IME).
1981 This engine is very basic and issues calls to IME whenever an IO is
1982 queued.
1983
1984 **ime_psyncv**
1985 Synchronous read and write using DDN's Infinite Memory Engine (IME).
1986 This engine uses iovecs and will try to stack as much IOs as possible
1987 (if the IOs are "contiguous" and the IO depth is not exceeded)
1988 before issuing a call to IME.
1989
1990 **ime_aio**
1991 Asynchronous read and write using DDN's Infinite Memory Engine (IME).
1992 This engine will try to stack as much IOs as possible by creating
1993 requests for IME. FIO will then decide when to commit these requests.
1994
f80dba8d
MT
1995I/O engine specific parameters
1996~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1997
1998In addition, there are some parameters which are only valid when a specific
f50fbdda
TK
1999:option:`ioengine` is in use. These are used identically to normal parameters,
2000with the caveat that when used on the command line, they must come after the
f80dba8d
MT
2001:option:`ioengine` that defines them is selected.
2002
2003.. option:: userspace_reap : [libaio]
2004
2005 Normally, with the libaio engine in use, fio will use the
2006 :manpage:`io_getevents(2)` system call to reap newly returned events. With
2007 this flag turned on, the AIO ring will be read directly from user-space to
2008 reap events. The reaping mode is only enabled when polling for a minimum of
c60ebc45 2009 0 events (e.g. when :option:`iodepth_batch_complete` `=0`).
f80dba8d 2010
9d25d068 2011.. option:: hipri : [pvsync2]
f80dba8d
MT
2012
2013 Set RWF_HIPRI on I/O, indicating to the kernel that it's of higher priority
2014 than normal.
2015
a0679ce5
SB
2016.. option:: hipri_percentage : [pvsync2]
2017
f50fbdda 2018 When hipri is set this determines the probability of a pvsync2 I/O being high
a0679ce5
SB
2019 priority. The default is 100%.
2020
f80dba8d
MT
2021.. option:: cpuload=int : [cpuio]
2022
da19cdb4
TK
2023 Attempt to use the specified percentage of CPU cycles. This is a mandatory
2024 option when using cpuio I/O engine.
f80dba8d
MT
2025
2026.. option:: cpuchunks=int : [cpuio]
2027
2028 Split the load into cycles of the given time. In microseconds.
2029
2030.. option:: exit_on_io_done=bool : [cpuio]
2031
2032 Detect when I/O threads are done, then exit.
2033
f80dba8d
MT
2034.. option:: namenode=str : [libhdfs]
2035
22413915 2036 The hostname or IP address of a HDFS cluster namenode to contact.
f80dba8d
MT
2037
2038.. option:: port=int
2039
f50fbdda
TK
2040 [libhdfs]
2041
2042 The listening port of the HFDS cluster namenode.
2043
f80dba8d
MT
2044 [netsplice], [net]
2045
2046 The TCP or UDP port to bind to or connect to. If this is used with
2047 :option:`numjobs` to spawn multiple instances of the same job type, then
2048 this will be the starting port number since fio will use a range of
2049 ports.
2050
609ac152
SB
2051 [rdma]
2052
2053 The port to use for RDMA-CM communication. This should be the same value
2054 on the client and the server side.
2055
2056.. option:: hostname=str : [netsplice] [net] [rdma]
f80dba8d 2057
609ac152
SB
2058 The hostname or IP address to use for TCP, UDP or RDMA-CM based I/O. If the job
2059 is a TCP listener or UDP reader, the hostname is not used and must be omitted
f50fbdda 2060 unless it is a valid UDP multicast address.
f80dba8d
MT
2061
2062.. option:: interface=str : [netsplice] [net]
2063
2064 The IP address of the network interface used to send or receive UDP
2065 multicast.
2066
2067.. option:: ttl=int : [netsplice] [net]
2068
2069 Time-to-live value for outgoing UDP multicast packets. Default: 1.
2070
2071.. option:: nodelay=bool : [netsplice] [net]
2072
2073 Set TCP_NODELAY on TCP connections.
2074
f50fbdda 2075.. option:: protocol=str, proto=str : [netsplice] [net]
f80dba8d
MT
2076
2077 The network protocol to use. Accepted values are:
2078
2079 **tcp**
2080 Transmission control protocol.
2081 **tcpv6**
2082 Transmission control protocol V6.
2083 **udp**
2084 User datagram protocol.
2085 **udpv6**
2086 User datagram protocol V6.
2087 **unix**
2088 UNIX domain socket.
2089
2090 When the protocol is TCP or UDP, the port must also be given, as well as the
2091 hostname if the job is a TCP listener or UDP reader. For unix sockets, the
f50fbdda 2092 normal :option:`filename` option should be used and the port is invalid.
f80dba8d 2093
e9184ec1 2094.. option:: listen : [netsplice] [net]
f80dba8d
MT
2095
2096 For TCP network connections, tell fio to listen for incoming connections
2097 rather than initiating an outgoing connection. The :option:`hostname` must
2098 be omitted if this option is used.
2099
e9184ec1 2100.. option:: pingpong : [netsplice] [net]
f80dba8d
MT
2101
2102 Normally a network writer will just continue writing data, and a network
2103 reader will just consume packages. If ``pingpong=1`` is set, a writer will
2104 send its normal payload to the reader, then wait for the reader to send the
2105 same payload back. This allows fio to measure network latencies. The
2106 submission and completion latencies then measure local time spent sending or
2107 receiving, and the completion latency measures how long it took for the
2108 other end to receive and send back. For UDP multicast traffic
2109 ``pingpong=1`` should only be set for a single reader when multiple readers
2110 are listening to the same address.
2111
e9184ec1 2112.. option:: window_size : [netsplice] [net]
f80dba8d
MT
2113
2114 Set the desired socket buffer size for the connection.
2115
e9184ec1 2116.. option:: mss : [netsplice] [net]
f80dba8d
MT
2117
2118 Set the TCP maximum segment size (TCP_MAXSEG).
2119
2120.. option:: donorname=str : [e4defrag]
2121
730bd7d9 2122 File will be used as a block donor (swap extents between files).
f80dba8d
MT
2123
2124.. option:: inplace=int : [e4defrag]
2125
2126 Configure donor file blocks allocation strategy:
2127
2128 **0**
2129 Default. Preallocate donor's file on init.
2130 **1**
2b455dbf 2131 Allocate space immediately inside defragment event, and free right
f80dba8d
MT
2132 after event.
2133
f3f96717 2134.. option:: clustername=str : [rbd,rados]
f80dba8d
MT
2135
2136 Specifies the name of the Ceph cluster.
2137
2138.. option:: rbdname=str : [rbd]
2139
2140 Specifies the name of the RBD.
2141
f3f96717 2142.. option:: pool=str : [rbd,rados]
f80dba8d 2143
f3f96717 2144 Specifies the name of the Ceph pool containing RBD or RADOS data.
f80dba8d 2145
f3f96717 2146.. option:: clientname=str : [rbd,rados]
f80dba8d
MT
2147
2148 Specifies the username (without the 'client.' prefix) used to access the
2149 Ceph cluster. If the *clustername* is specified, the *clientname* shall be
2150 the full *type.id* string. If no type. prefix is given, fio will add
2151 'client.' by default.
2152
f3f96717
IF
2153.. option:: busy_poll=bool : [rbd,rados]
2154
2155 Poll store instead of waiting for completion. Usually this provides better
2156 throughput at cost of higher(up to 100%) CPU utilization.
2157
f80dba8d
MT
2158.. option:: skip_bad=bool : [mtd]
2159
2160 Skip operations against known bad blocks.
2161
2162.. option:: hdfsdirectory : [libhdfs]
2163
2164 libhdfs will create chunk in this HDFS directory.
2165
2166.. option:: chunk_size : [libhdfs]
2167
2b455dbf 2168 The size of the chunk to use for each file.
f80dba8d 2169
609ac152
SB
2170.. option:: verb=str : [rdma]
2171
2172 The RDMA verb to use on this side of the RDMA ioengine connection. Valid
2173 values are write, read, send and recv. These correspond to the equivalent
2174 RDMA verbs (e.g. write = rdma_write etc.). Note that this only needs to be
2175 specified on the client side of the connection. See the examples folder.
2176
2177.. option:: bindname=str : [rdma]
2178
2179 The name to use to bind the local RDMA-CM connection to a local RDMA device.
2180 This could be a hostname or an IPv4 or IPv6 address. On the server side this
2181 will be passed into the rdma_bind_addr() function and on the client site it
2182 will be used in the rdma_resolve_add() function. This can be useful when
2183 multiple paths exist between the client and the server or in certain loopback
2184 configurations.
f80dba8d 2185
52b81b7c
KD
2186.. option:: readfua=bool : [sg]
2187
2188 With readfua option set to 1, read operations include
2189 the force unit access (fua) flag. Default is 0.
2190
2191.. option:: writefua=bool : [sg]
2192
2193 With writefua option set to 1, write operations include
2194 the force unit access (fua) flag. Default is 0.
2195
2c3a9150 2196.. option:: sg_write_mode=str : [sg]
3740cfc8 2197
2c3a9150
VF
2198 Specify the type of write commands to issue. This option can take three values:
2199
2200 **write**
2201 This is the default where write opcodes are issued as usual.
2202 **verify**
2203 Issue WRITE AND VERIFY commands. The BYTCHK bit is set to 0. This
2204 directs the device to carry out a medium verification with no data
2205 comparison. The writefua option is ignored with this selection.
2206 **same**
2207 Issue WRITE SAME commands. This transfers a single block to the device
2208 and writes this same block of data to a contiguous sequence of LBAs
2209 beginning at the specified offset. fio's block size parameter specifies
2210 the amount of data written with each command. However, the amount of data
2211 actually transferred to the device is equal to the device's block
2212 (sector) size. For a device with 512 byte sectors, blocksize=8k will
2213 write 16 sectors with each command. fio will still generate 8k of data
2214 for each command but only the first 512 bytes will be used and
2215 transferred to the device. The writefua option is ignored with this
2216 selection.
52b81b7c 2217
c2f6a13d
LMB
2218.. option:: http_host=str : [http]
2219
2220 Hostname to connect to. For S3, this could be the bucket hostname.
2221 Default is **localhost**
2222
2223.. option:: http_user=str : [http]
2224
2225 Username for HTTP authentication.
2226
2227.. option:: http_pass=str : [http]
2228
2229 Password for HTTP authentication.
2230
09fd2966 2231.. option:: https=str : [http]
c2f6a13d 2232
09fd2966
LMB
2233 Enable HTTPS instead of http. *on* enables HTTPS; *insecure*
2234 will enable HTTPS, but disable SSL peer verification (use with
2235 caution!). Default is **off**
c2f6a13d 2236
09fd2966 2237.. option:: http_mode=str : [http]
c2f6a13d 2238
09fd2966
LMB
2239 Which HTTP access mode to use: *webdav*, *swift*, or *s3*.
2240 Default is **webdav**
c2f6a13d
LMB
2241
2242.. option:: http_s3_region=str : [http]
2243
2244 The S3 region/zone string.
2245 Default is **us-east-1**
2246
2247.. option:: http_s3_key=str : [http]
2248
2249 The S3 secret key.
2250
2251.. option:: http_s3_keyid=str : [http]
2252
2253 The S3 key/access id.
2254
09fd2966
LMB
2255.. option:: http_swift_auth_token=str : [http]
2256
2257 The Swift auth token. See the example configuration file on how
2258 to retrieve this.
2259
c2f6a13d
LMB
2260.. option:: http_verbose=int : [http]
2261
2262 Enable verbose requests from libcurl. Useful for debugging. 1
2263 turns on verbose logging from libcurl, 2 additionally enables
2264 HTTP IO tracing. Default is **0**
2265
f80dba8d
MT
2266I/O depth
2267~~~~~~~~~
2268
2269.. option:: iodepth=int
2270
2271 Number of I/O units to keep in flight against the file. Note that
2272 increasing *iodepth* beyond 1 will not affect synchronous ioengines (except
c60ebc45 2273 for small degrees when :option:`verify_async` is in use). Even async
f80dba8d
MT
2274 engines may impose OS restrictions causing the desired depth not to be
2275 achieved. This may happen on Linux when using libaio and not setting
9207a0cb 2276 :option:`direct`\=1, since buffered I/O is not async on that OS. Keep an
f80dba8d
MT
2277 eye on the I/O depth distribution in the fio output to verify that the
2278 achieved depth is as expected. Default: 1.
2279
2280.. option:: iodepth_batch_submit=int, iodepth_batch=int
2281
2282 This defines how many pieces of I/O to submit at once. It defaults to 1
2283 which means that we submit each I/O as soon as it is available, but can be
2284 raised to submit bigger batches of I/O at the time. If it is set to 0 the
2285 :option:`iodepth` value will be used.
2286
2287.. option:: iodepth_batch_complete_min=int, iodepth_batch_complete=int
2288
2289 This defines how many pieces of I/O to retrieve at once. It defaults to 1
2290 which means that we'll ask for a minimum of 1 I/O in the retrieval process
2291 from the kernel. The I/O retrieval will go on until we hit the limit set by
2292 :option:`iodepth_low`. If this variable is set to 0, then fio will always
2293 check for completed events before queuing more I/O. This helps reduce I/O
2294 latency, at the cost of more retrieval system calls.
2295
2296.. option:: iodepth_batch_complete_max=int
2297
2298 This defines maximum pieces of I/O to retrieve at once. This variable should
9207a0cb 2299 be used along with :option:`iodepth_batch_complete_min`\=int variable,
f80dba8d 2300 specifying the range of min and max amount of I/O which should be
730bd7d9 2301 retrieved. By default it is equal to the :option:`iodepth_batch_complete_min`
f80dba8d
MT
2302 value.
2303
2304 Example #1::
2305
2306 iodepth_batch_complete_min=1
2307 iodepth_batch_complete_max=<iodepth>
2308
2309 which means that we will retrieve at least 1 I/O and up to the whole
2310 submitted queue depth. If none of I/O has been completed yet, we will wait.
2311
2312 Example #2::
2313
2314 iodepth_batch_complete_min=0
2315 iodepth_batch_complete_max=<iodepth>
2316
2317 which means that we can retrieve up to the whole submitted queue depth, but
2318 if none of I/O has been completed yet, we will NOT wait and immediately exit
2319 the system call. In this example we simply do polling.
2320
2321.. option:: iodepth_low=int
2322
2323 The low water mark indicating when to start filling the queue
2324 again. Defaults to the same as :option:`iodepth`, meaning that fio will
2325 attempt to keep the queue full at all times. If :option:`iodepth` is set to
c60ebc45 2326 e.g. 16 and *iodepth_low* is set to 4, then after fio has filled the queue of
f80dba8d
MT
2327 16 requests, it will let the depth drain down to 4 before starting to fill
2328 it again.
2329
997b5680
SW
2330.. option:: serialize_overlap=bool
2331
2332 Serialize in-flight I/Os that might otherwise cause or suffer from data races.
2333 When two or more I/Os are submitted simultaneously, there is no guarantee that
2334 the I/Os will be processed or completed in the submitted order. Further, if
2335 two or more of those I/Os are writes, any overlapping region between them can
2336 become indeterminate/undefined on certain storage. These issues can cause
2337 verification to fail erratically when at least one of the racing I/Os is
2338 changing data and the overlapping region has a non-zero size. Setting
2339 ``serialize_overlap`` tells fio to avoid provoking this behavior by explicitly
2340 serializing in-flight I/Os that have a non-zero overlap. Note that setting
ee21ebee 2341 this option can reduce both performance and the :option:`iodepth` achieved.
3d6a6f04
VF
2342
2343 This option only applies to I/Os issued for a single job except when it is
2344 enabled along with :option:`io_submit_mode`=offload. In offload mode, fio
2345 will check for overlap among all I/Os submitted by offload jobs with :option:`serialize_overlap`
307f2246 2346 enabled.
3d6a6f04
VF
2347
2348 Default: false.
997b5680 2349
f80dba8d
MT
2350.. option:: io_submit_mode=str
2351
2352 This option controls how fio submits the I/O to the I/O engine. The default
2353 is `inline`, which means that the fio job threads submit and reap I/O
2354 directly. If set to `offload`, the job threads will offload I/O submission
2355 to a dedicated pool of I/O threads. This requires some coordination and thus
2356 has a bit of extra overhead, especially for lower queue depth I/O where it
2357 can increase latencies. The benefit is that fio can manage submission rates
2358 independently of the device completion rates. This avoids skewed latency
730bd7d9 2359 reporting if I/O gets backed up on the device side (the coordinated omission
f80dba8d
MT
2360 problem).
2361
2362
2363I/O rate
2364~~~~~~~~
2365
a881438b 2366.. option:: thinktime=time
f80dba8d 2367
f75ede1d
SW
2368 Stall the job for the specified period of time after an I/O has completed before issuing the
2369 next. May be used to simulate processing being done by an application.
947e0fe0 2370 When the unit is omitted, the value is interpreted in microseconds. See
f80dba8d
MT
2371 :option:`thinktime_blocks` and :option:`thinktime_spin`.
2372
a881438b 2373.. option:: thinktime_spin=time
f80dba8d
MT
2374
2375 Only valid if :option:`thinktime` is set - pretend to spend CPU time doing
2376 something with the data received, before falling back to sleeping for the
f75ede1d 2377 rest of the period specified by :option:`thinktime`. When the unit is
947e0fe0 2378 omitted, the value is interpreted in microseconds.
f80dba8d
MT
2379
2380.. option:: thinktime_blocks=int
2381
2382 Only valid if :option:`thinktime` is set - control how many blocks to issue,
f50fbdda
TK
2383 before waiting :option:`thinktime` usecs. If not set, defaults to 1 which will make
2384 fio wait :option:`thinktime` usecs after every block. This effectively makes any
f80dba8d 2385 queue depth setting redundant, since no more than 1 I/O will be queued
f50fbdda 2386 before we have to complete it and do our :option:`thinktime`. In other words, this
f80dba8d 2387 setting effectively caps the queue depth if the latter is larger.
71bfa161 2388
f80dba8d 2389.. option:: rate=int[,int][,int]
71bfa161 2390
f80dba8d
MT
2391 Cap the bandwidth used by this job. The number is in bytes/sec, the normal
2392 suffix rules apply. Comma-separated values may be specified for reads,
2393 writes, and trims as described in :option:`blocksize`.
71bfa161 2394
b25b3464
SW
2395 For example, using `rate=1m,500k` would limit reads to 1MiB/sec and writes to
2396 500KiB/sec. Capping only reads or writes can be done with `rate=,500k` or
2397 `rate=500k,` where the former will only limit writes (to 500KiB/sec) and the
2398 latter will only limit reads.
2399
f80dba8d 2400.. option:: rate_min=int[,int][,int]
71bfa161 2401
f80dba8d
MT
2402 Tell fio to do whatever it can to maintain at least this bandwidth. Failing
2403 to meet this requirement will cause the job to exit. Comma-separated values
2404 may be specified for reads, writes, and trims as described in
2405 :option:`blocksize`.
71bfa161 2406
f80dba8d 2407.. option:: rate_iops=int[,int][,int]
71bfa161 2408
f80dba8d
MT
2409 Cap the bandwidth to this number of IOPS. Basically the same as
2410 :option:`rate`, just specified independently of bandwidth. If the job is
2411 given a block size range instead of a fixed value, the smallest block size
2412 is used as the metric. Comma-separated values may be specified for reads,
2413 writes, and trims as described in :option:`blocksize`.
71bfa161 2414
f80dba8d 2415.. option:: rate_iops_min=int[,int][,int]
71bfa161 2416
f80dba8d
MT
2417 If fio doesn't meet this rate of I/O, it will cause the job to exit.
2418 Comma-separated values may be specified for reads, writes, and trims as
2419 described in :option:`blocksize`.
71bfa161 2420
f80dba8d 2421.. option:: rate_process=str
66c098b8 2422
f80dba8d
MT
2423 This option controls how fio manages rated I/O submissions. The default is
2424 `linear`, which submits I/O in a linear fashion with fixed delays between
c60ebc45 2425 I/Os that gets adjusted based on I/O completion rates. If this is set to
f80dba8d
MT
2426 `poisson`, fio will submit I/O based on a more real world random request
2427 flow, known as the Poisson process
2428 (https://en.wikipedia.org/wiki/Poisson_point_process). The lambda will be
2429 10^6 / IOPS for the given workload.
71bfa161 2430
1a9bf814
JA
2431.. option:: rate_ignore_thinktime=bool
2432
2433 By default, fio will attempt to catch up to the specified rate setting,
2434 if any kind of thinktime setting was used. If this option is set, then
2435 fio will ignore the thinktime and continue doing IO at the specified
2436 rate, instead of entering a catch-up mode after thinktime is done.
2437
71bfa161 2438
f80dba8d
MT
2439I/O latency
2440~~~~~~~~~~~
71bfa161 2441
a881438b 2442.. option:: latency_target=time
71bfa161 2443
f80dba8d 2444 If set, fio will attempt to find the max performance point that the given
f75ede1d 2445 workload will run at while maintaining a latency below this target. When
947e0fe0 2446 the unit is omitted, the value is interpreted in microseconds. See
f75ede1d 2447 :option:`latency_window` and :option:`latency_percentile`.
71bfa161 2448
a881438b 2449.. option:: latency_window=time
71bfa161 2450
f80dba8d 2451 Used with :option:`latency_target` to specify the sample window that the job
f75ede1d 2452 is run at varying queue depths to test the performance. When the unit is
947e0fe0 2453 omitted, the value is interpreted in microseconds.
b4692828 2454
f80dba8d 2455.. option:: latency_percentile=float
71bfa161 2456
c60ebc45 2457 The percentage of I/Os that must fall within the criteria specified by
f80dba8d 2458 :option:`latency_target` and :option:`latency_window`. If not set, this
c60ebc45 2459 defaults to 100.0, meaning that all I/Os must be equal or below to the value
f80dba8d 2460 set by :option:`latency_target`.
71bfa161 2461
a881438b 2462.. option:: max_latency=time
71bfa161 2463
f75ede1d 2464 If set, fio will exit the job with an ETIMEDOUT error if it exceeds this
947e0fe0 2465 maximum latency. When the unit is omitted, the value is interpreted in
f75ede1d 2466 microseconds.
71bfa161 2467
f80dba8d 2468.. option:: rate_cycle=int
71bfa161 2469
f80dba8d 2470 Average bandwidth for :option:`rate` and :option:`rate_min` over this number
a47b697c 2471 of milliseconds. Defaults to 1000.
71bfa161 2472
71bfa161 2473
f80dba8d
MT
2474I/O replay
2475~~~~~~~~~~
71bfa161 2476
f80dba8d 2477.. option:: write_iolog=str
c2b1e753 2478
f80dba8d
MT
2479 Write the issued I/O patterns to the specified file. See
2480 :option:`read_iolog`. Specify a separate file for each job, otherwise the
2481 iologs will be interspersed and the file may be corrupt.
c2b1e753 2482
f80dba8d 2483.. option:: read_iolog=str
71bfa161 2484
22413915 2485 Open an iolog with the specified filename and replay the I/O patterns it
f80dba8d
MT
2486 contains. This can be used to store a workload and replay it sometime
2487 later. The iolog given may also be a blktrace binary file, which allows fio
2488 to replay a workload captured by :command:`blktrace`. See
2489 :manpage:`blktrace(8)` for how to capture such logging data. For blktrace
2490 replay, the file needs to be turned into a blkparse binary data file first
2491 (``blkparse <device> -o /dev/null -d file_for_fio.bin``).
78439a18
JA
2492 You can specify a number of files by separating the names with a ':'
2493 character. See the :option:`filename` option for information on how to
2494 escape ':' and '\' characters within the file names. These files will
2495 be sequentially assigned to job clones created by :option:`numjobs`.
71bfa161 2496
77be374d
AK
2497.. option:: read_iolog_chunked=bool
2498
2499 Determines how iolog is read. If false(default) entire :option:`read_iolog`
2500 will be read at once. If selected true, input from iolog will be read
2501 gradually. Useful when iolog is very large, or it is generated.
2502
b9921d1a
DZ
2503.. option:: merge_blktrace_file=str
2504
2505 When specified, rather than replaying the logs passed to :option:`read_iolog`,
2506 the logs go through a merge phase which aggregates them into a single
2507 blktrace. The resulting file is then passed on as the :option:`read_iolog`
2508 parameter. The intention here is to make the order of events consistent.
2509 This limits the influence of the scheduler compared to replaying multiple
2510 blktraces via concurrent jobs.
2511
87a48ada
DZ
2512.. option:: merge_blktrace_scalars=float_list
2513
2514 This is a percentage based option that is index paired with the list of
2515 files passed to :option:`read_iolog`. When merging is performed, scale
2516 the time of each event by the corresponding amount. For example,
2517 ``--merge_blktrace_scalars="50:100"`` runs the first trace in halftime
2518 and the second trace in realtime. This knob is separately tunable from
2519 :option:`replay_time_scale` which scales the trace during runtime and
2520 does not change the output of the merge unlike this option.
2521
55bfd8c8
DZ
2522.. option:: merge_blktrace_iters=float_list
2523
2524 This is a whole number option that is index paired with the list of files
2525 passed to :option:`read_iolog`. When merging is performed, run each trace
2526 for the specified number of iterations. For example,
2527 ``--merge_blktrace_iters="2:1"`` runs the first trace for two iterations
2528 and the second trace for one iteration.
2529
589e88b7 2530.. option:: replay_no_stall=bool
71bfa161 2531
f80dba8d 2532 When replaying I/O with :option:`read_iolog` the default behavior is to
22413915 2533 attempt to respect the timestamps within the log and replay them with the
f80dba8d
MT
2534 appropriate delay between IOPS. By setting this variable fio will not
2535 respect the timestamps and attempt to replay them as fast as possible while
2536 still respecting ordering. The result is the same I/O pattern to a given
2537 device, but different timings.
71bfa161 2538
6dd7fa77
JA
2539.. option:: replay_time_scale=int
2540
2541 When replaying I/O with :option:`read_iolog`, fio will honor the
2542 original timing in the trace. With this option, it's possible to scale
2543 the time. It's a percentage option, if set to 50 it means run at 50%
2544 the original IO rate in the trace. If set to 200, run at twice the
2545 original IO rate. Defaults to 100.
2546
f80dba8d 2547.. option:: replay_redirect=str
b4692828 2548
f80dba8d
MT
2549 While replaying I/O patterns using :option:`read_iolog` the default behavior
2550 is to replay the IOPS onto the major/minor device that each IOP was recorded
2551 from. This is sometimes undesirable because on a different machine those
2552 major/minor numbers can map to a different device. Changing hardware on the
2553 same system can also result in a different major/minor mapping.
730bd7d9 2554 ``replay_redirect`` causes all I/Os to be replayed onto the single specified
f80dba8d 2555 device regardless of the device it was recorded
9207a0cb 2556 from. i.e. :option:`replay_redirect`\= :file:`/dev/sdc` would cause all I/O
f80dba8d
MT
2557 in the blktrace or iolog to be replayed onto :file:`/dev/sdc`. This means
2558 multiple devices will be replayed onto a single device, if the trace
2559 contains multiple devices. If you want multiple devices to be replayed
2560 concurrently to multiple redirected devices you must blkparse your trace
2561 into separate traces and replay them with independent fio invocations.
2562 Unfortunately this also breaks the strict time ordering between multiple
2563 device accesses.
71bfa161 2564
f80dba8d 2565.. option:: replay_align=int
74929ac2 2566
350a535d
DZ
2567 Force alignment of the byte offsets in a trace to this value. The value
2568 must be a power of 2.
3c54bc46 2569
f80dba8d 2570.. option:: replay_scale=int
3c54bc46 2571
350a535d
DZ
2572 Scale byte offsets down by this factor when replaying traces. Should most
2573 likely use :option:`replay_align` as well.
3c54bc46 2574
38f68906
JA
2575.. option:: replay_skip=str
2576
2577 Sometimes it's useful to skip certain IO types in a replay trace.
2578 This could be, for instance, eliminating the writes in the trace.
2579 Or not replaying the trims/discards, if you are redirecting to
2580 a device that doesn't support them. This option takes a comma
2581 separated list of read, write, trim, sync.
2582
3c54bc46 2583
f80dba8d
MT
2584Threads, processes and job synchronization
2585~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3c54bc46 2586
f80dba8d 2587.. option:: thread
3c54bc46 2588
730bd7d9
SW
2589 Fio defaults to creating jobs by using fork, however if this option is
2590 given, fio will create jobs by using POSIX Threads' function
2591 :manpage:`pthread_create(3)` to create threads instead.
71bfa161 2592
f80dba8d 2593.. option:: wait_for=str
74929ac2 2594
730bd7d9
SW
2595 If set, the current job won't be started until all workers of the specified
2596 waitee job are done.
74929ac2 2597
f80dba8d
MT
2598 ``wait_for`` operates on the job name basis, so there are a few
2599 limitations. First, the waitee must be defined prior to the waiter job
2600 (meaning no forward references). Second, if a job is being referenced as a
2601 waitee, it must have a unique name (no duplicate waitees).
74929ac2 2602
f80dba8d 2603.. option:: nice=int
892a6ffc 2604
f80dba8d 2605 Run the job with the given nice value. See man :manpage:`nice(2)`.
892a6ffc 2606
f80dba8d
MT
2607 On Windows, values less than -15 set the process class to "High"; -1 through
2608 -15 set "Above Normal"; 1 through 15 "Below Normal"; and above 15 "Idle"
2609 priority class.
74929ac2 2610
f80dba8d 2611.. option:: prio=int
71bfa161 2612
f80dba8d
MT
2613 Set the I/O priority value of this job. Linux limits us to a positive value
2614 between 0 and 7, with 0 being the highest. See man
2615 :manpage:`ionice(1)`. Refer to an appropriate manpage for other operating
2616 systems since meaning of priority may differ.
71bfa161 2617
f80dba8d 2618.. option:: prioclass=int
d59aa780 2619
f80dba8d 2620 Set the I/O priority class. See man :manpage:`ionice(1)`.
d59aa780 2621
f80dba8d 2622.. option:: cpus_allowed=str
6d500c2e 2623
730bd7d9 2624 Controls the same options as :option:`cpumask`, but accepts a textual
b570e037
SW
2625 specification of the permitted CPUs instead and CPUs are indexed from 0. So
2626 to use CPUs 0 and 5 you would specify ``cpus_allowed=0,5``. This option also
2627 allows a range of CPUs to be specified -- say you wanted a binding to CPUs
2628 0, 5, and 8 to 15, you would set ``cpus_allowed=0,5,8-15``.
2629
2630 On Windows, when ``cpus_allowed`` is unset only CPUs from fio's current
2631 processor group will be used and affinity settings are inherited from the
2632 system. An fio build configured to target Windows 7 makes options that set
2633 CPUs processor group aware and values will set both the processor group
2634 and a CPU from within that group. For example, on a system where processor
2635 group 0 has 40 CPUs and processor group 1 has 32 CPUs, ``cpus_allowed``
2636 values between 0 and 39 will bind CPUs from processor group 0 and
2637 ``cpus_allowed`` values between 40 and 71 will bind CPUs from processor
2638 group 1. When using ``cpus_allowed_policy=shared`` all CPUs specified by a
2639 single ``cpus_allowed`` option must be from the same processor group. For
2640 Windows fio builds not built for Windows 7, CPUs will only be selected from
2641 (and be relative to) whatever processor group fio happens to be running in
2642 and CPUs from other processor groups cannot be used.
6d500c2e 2643
f80dba8d 2644.. option:: cpus_allowed_policy=str
6d500c2e 2645
f80dba8d 2646 Set the policy of how fio distributes the CPUs specified by
730bd7d9 2647 :option:`cpus_allowed` or :option:`cpumask`. Two policies are supported:
6d500c2e 2648
f80dba8d
MT
2649 **shared**
2650 All jobs will share the CPU set specified.
2651 **split**
2652 Each job will get a unique CPU from the CPU set.
6d500c2e 2653
22413915 2654 **shared** is the default behavior, if the option isn't specified. If
f80dba8d
MT
2655 **split** is specified, then fio will will assign one cpu per job. If not
2656 enough CPUs are given for the jobs listed, then fio will roundrobin the CPUs
2657 in the set.
6d500c2e 2658
b570e037
SW
2659.. option:: cpumask=int
2660
2661 Set the CPU affinity of this job. The parameter given is a bit mask of
2662 allowed CPUs the job may run on. So if you want the allowed CPUs to be 1
2663 and 5, you would pass the decimal value of (1 << 1 | 1 << 5), or 34. See man
2664 :manpage:`sched_setaffinity(2)`. This may not work on all supported
2665 operating systems or kernel versions. This option doesn't work well for a
2666 higher CPU count than what you can store in an integer mask, so it can only
2667 control cpus 1-32. For boxes with larger CPU counts, use
2668 :option:`cpus_allowed`.
2669
f80dba8d 2670.. option:: numa_cpu_nodes=str
6d500c2e 2671
f80dba8d
MT
2672 Set this job running on specified NUMA nodes' CPUs. The arguments allow
2673 comma delimited list of cpu numbers, A-B ranges, or `all`. Note, to enable
ac8ca2af 2674 NUMA options support, fio must be built on a system with libnuma-dev(el)
f80dba8d 2675 installed.
61b9861d 2676
f80dba8d 2677.. option:: numa_mem_policy=str
61b9861d 2678
f80dba8d
MT
2679 Set this job's memory policy and corresponding NUMA nodes. Format of the
2680 arguments::
5c94b008 2681
f80dba8d 2682 <mode>[:<nodelist>]
ce35b1ec 2683
804c0839 2684 ``mode`` is one of the following memory policies: ``default``, ``prefer``,
730bd7d9
SW
2685 ``bind``, ``interleave`` or ``local``. For ``default`` and ``local`` memory
2686 policies, no node needs to be specified. For ``prefer``, only one node is
2687 allowed. For ``bind`` and ``interleave`` the ``nodelist`` may be as
2688 follows: a comma delimited list of numbers, A-B ranges, or `all`.
71bfa161 2689
f80dba8d 2690.. option:: cgroup=str
390b1537 2691
f80dba8d
MT
2692 Add job to this control group. If it doesn't exist, it will be created. The
2693 system must have a mounted cgroup blkio mount point for this to work. If
2694 your system doesn't have it mounted, you can do so with::
5af1c6f3 2695
f80dba8d 2696 # mount -t cgroup -o blkio none /cgroup
5af1c6f3 2697
f80dba8d 2698.. option:: cgroup_weight=int
5af1c6f3 2699
f80dba8d
MT
2700 Set the weight of the cgroup to this value. See the documentation that comes
2701 with the kernel, allowed values are in the range of 100..1000.
a086c257 2702
f80dba8d 2703.. option:: cgroup_nodelete=bool
8c07860d 2704
f80dba8d
MT
2705 Normally fio will delete the cgroups it has created after the job
2706 completion. To override this behavior and to leave cgroups around after the
2707 job completion, set ``cgroup_nodelete=1``. This can be useful if one wants
2708 to inspect various cgroup files after job completion. Default: false.
8c07860d 2709
f80dba8d 2710.. option:: flow_id=int
8c07860d 2711
f80dba8d
MT
2712 The ID of the flow. If not specified, it defaults to being a global
2713 flow. See :option:`flow`.
1907dbc6 2714
f80dba8d 2715.. option:: flow=int
71bfa161 2716
f80dba8d
MT
2717 Weight in token-based flow control. If this value is used, then there is a
2718 'flow counter' which is used to regulate the proportion of activity between
2719 two or more jobs. Fio attempts to keep this flow counter near zero. The
2720 ``flow`` parameter stands for how much should be added or subtracted to the
2721 flow counter on each iteration of the main I/O loop. That is, if one job has
2722 ``flow=8`` and another job has ``flow=-1``, then there will be a roughly 1:8
2723 ratio in how much one runs vs the other.
71bfa161 2724
f80dba8d 2725.. option:: flow_watermark=int
a31041ea 2726
f80dba8d
MT
2727 The maximum value that the absolute value of the flow counter is allowed to
2728 reach before the job must wait for a lower value of the counter.
82407585 2729
f80dba8d 2730.. option:: flow_sleep=int
82407585 2731
f80dba8d
MT
2732 The period of time, in microseconds, to wait after the flow watermark has
2733 been exceeded before retrying operations.
82407585 2734
f80dba8d 2735.. option:: stonewall, wait_for_previous
82407585 2736
f80dba8d
MT
2737 Wait for preceding jobs in the job file to exit, before starting this
2738 one. Can be used to insert serialization points in the job file. A stone
2739 wall also implies starting a new reporting group, see
2740 :option:`group_reporting`.
2741
2742.. option:: exitall
2743
730bd7d9
SW
2744 By default, fio will continue running all other jobs when one job finishes
2745 but sometimes this is not the desired action. Setting ``exitall`` will
2746 instead make fio terminate all other jobs when one job finishes.
f80dba8d
MT
2747
2748.. option:: exec_prerun=str
2749
2750 Before running this job, issue the command specified through
2751 :manpage:`system(3)`. Output is redirected in a file called
2752 :file:`jobname.prerun.txt`.
2753
2754.. option:: exec_postrun=str
2755
2756 After the job completes, issue the command specified though
2757 :manpage:`system(3)`. Output is redirected in a file called
2758 :file:`jobname.postrun.txt`.
2759
2760.. option:: uid=int
2761
2762 Instead of running as the invoking user, set the user ID to this value
2763 before the thread/process does any work.
2764
2765.. option:: gid=int
2766
2767 Set group ID, see :option:`uid`.
2768
2769
2770Verification
2771~~~~~~~~~~~~
2772
2773.. option:: verify_only
2774
2775 Do not perform specified workload, only verify data still matches previous
2776 invocation of this workload. This option allows one to check data multiple
2777 times at a later date without overwriting it. This option makes sense only
2778 for workloads that write data, and does not support workloads with the
2779 :option:`time_based` option set.
2780
2781.. option:: do_verify=bool
2782
2783 Run the verify phase after a write phase. Only valid if :option:`verify` is
2784 set. Default: true.
2785
2786.. option:: verify=str
2787
2788 If writing to a file, fio can verify the file contents after each iteration
2789 of the job. Each verification method also implies verification of special
2790 header, which is written to the beginning of each block. This header also
2791 includes meta information, like offset of the block, block number, timestamp
2792 when block was written, etc. :option:`verify` can be combined with
2793 :option:`verify_pattern` option. The allowed values are:
2794
2795 **md5**
2796 Use an md5 sum of the data area and store it in the header of
2797 each block.
2798
2799 **crc64**
2800 Use an experimental crc64 sum of the data area and store it in the
2801 header of each block.
2802
2803 **crc32c**
a5896300
SW
2804 Use a crc32c sum of the data area and store it in the header of
2805 each block. This will automatically use hardware acceleration
2806 (e.g. SSE4.2 on an x86 or CRC crypto extensions on ARM64) but will
2807 fall back to software crc32c if none is found. Generally the
804c0839 2808 fastest checksum fio supports when hardware accelerated.
f80dba8d
MT
2809
2810 **crc32c-intel**
a5896300 2811 Synonym for crc32c.
f80dba8d
MT
2812
2813 **crc32**
2814 Use a crc32 sum of the data area and store it in the header of each
2815 block.
2816
2817 **crc16**
2818 Use a crc16 sum of the data area and store it in the header of each
2819 block.
2820
2821 **crc7**
2822 Use a crc7 sum of the data area and store it in the header of each
2823 block.
2824
2825 **xxhash**
2826 Use xxhash as the checksum function. Generally the fastest software
2827 checksum that fio supports.
2828
2829 **sha512**
2830 Use sha512 as the checksum function.
2831
2832 **sha256**
2833 Use sha256 as the checksum function.
2834
2835 **sha1**
2836 Use optimized sha1 as the checksum function.
82407585 2837
ae3a5acc
JA
2838 **sha3-224**
2839 Use optimized sha3-224 as the checksum function.
2840
2841 **sha3-256**
2842 Use optimized sha3-256 as the checksum function.
2843
2844 **sha3-384**
2845 Use optimized sha3-384 as the checksum function.
2846
2847 **sha3-512**
2848 Use optimized sha3-512 as the checksum function.
2849
f80dba8d
MT
2850 **meta**
2851 This option is deprecated, since now meta information is included in
2852 generic verification header and meta verification happens by
2853 default. For detailed information see the description of the
2854 :option:`verify` setting. This option is kept because of
2855 compatibility's sake with old configurations. Do not use it.
2856
2857 **pattern**
2858 Verify a strict pattern. Normally fio includes a header with some
2859 basic information and checksumming, but if this option is set, only
2860 the specific pattern set with :option:`verify_pattern` is verified.
2861
2862 **null**
2863 Only pretend to verify. Useful for testing internals with
9207a0cb 2864 :option:`ioengine`\=null, not for much else.
f80dba8d
MT
2865
2866 This option can be used for repeated burn-in tests of a system to make sure
2867 that the written data is also correctly read back. If the data direction
2868 given is a read or random read, fio will assume that it should verify a
2869 previously written file. If the data direction includes any form of write,
2870 the verify will be of the newly written data.
2871
47e6a6e5
SW
2872 To avoid false verification errors, do not use the norandommap option when
2873 verifying data with async I/O engines and I/O depths > 1. Or use the
2874 norandommap and the lfsr random generator together to avoid writing to the
2875 same offset with muliple outstanding I/Os.
2876
f80dba8d
MT
2877.. option:: verify_offset=int
2878
2879 Swap the verification header with data somewhere else in the block before
2880 writing. It is swapped back before verifying.
2881
2882.. option:: verify_interval=int
2883
2884 Write the verification header at a finer granularity than the
2885 :option:`blocksize`. It will be written for chunks the size of
2886 ``verify_interval``. :option:`blocksize` should divide this evenly.
2887
2888.. option:: verify_pattern=str
2889
2890 If set, fio will fill the I/O buffers with this pattern. Fio defaults to
2891 filling with totally random bytes, but sometimes it's interesting to fill
2892 with a known pattern for I/O verification purposes. Depending on the width
730bd7d9 2893 of the pattern, fio will fill 1/2/3/4 bytes of the buffer at the time (it can
f80dba8d
MT
2894 be either a decimal or a hex number). The ``verify_pattern`` if larger than
2895 a 32-bit quantity has to be a hex number that starts with either "0x" or
2896 "0X". Use with :option:`verify`. Also, ``verify_pattern`` supports %o
2897 format, which means that for each block offset will be written and then
2898 verified back, e.g.::
61b9861d
RP
2899
2900 verify_pattern=%o
2901
f80dba8d
MT
2902 Or use combination of everything::
2903
61b9861d 2904 verify_pattern=0xff%o"abcd"-12
e28218f3 2905
f80dba8d
MT
2906.. option:: verify_fatal=bool
2907
2908 Normally fio will keep checking the entire contents before quitting on a
2909 block verification failure. If this option is set, fio will exit the job on
2910 the first observed failure. Default: false.
2911
2912.. option:: verify_dump=bool
2913
2914 If set, dump the contents of both the original data block and the data block
2915 we read off disk to files. This allows later analysis to inspect just what
2916 kind of data corruption occurred. Off by default.
2917
2918.. option:: verify_async=int
2919
2920 Fio will normally verify I/O inline from the submitting thread. This option
2921 takes an integer describing how many async offload threads to create for I/O
2922 verification instead, causing fio to offload the duty of verifying I/O
2923 contents to one or more separate threads. If using this offload option, even
2924 sync I/O engines can benefit from using an :option:`iodepth` setting higher
2925 than 1, as it allows them to have I/O in flight while verifies are running.
d7e6ea1c 2926 Defaults to 0 async threads, i.e. verification is not asynchronous.
f80dba8d
MT
2927
2928.. option:: verify_async_cpus=str
2929
2930 Tell fio to set the given CPU affinity on the async I/O verification
2931 threads. See :option:`cpus_allowed` for the format used.
2932
2933.. option:: verify_backlog=int
2934
2935 Fio will normally verify the written contents of a job that utilizes verify
2936 once that job has completed. In other words, everything is written then
2937 everything is read back and verified. You may want to verify continually
2938 instead for a variety of reasons. Fio stores the meta data associated with
2939 an I/O block in memory, so for large verify workloads, quite a bit of memory
2940 would be used up holding this meta data. If this option is enabled, fio will
2941 write only N blocks before verifying these blocks.
2942
2943.. option:: verify_backlog_batch=int
2944
2945 Control how many blocks fio will verify if :option:`verify_backlog` is
2946 set. If not set, will default to the value of :option:`verify_backlog`
2947 (meaning the entire queue is read back and verified). If
2948 ``verify_backlog_batch`` is less than :option:`verify_backlog` then not all
2949 blocks will be verified, if ``verify_backlog_batch`` is larger than
2950 :option:`verify_backlog`, some blocks will be verified more than once.
2951
2952.. option:: verify_state_save=bool
2953
2954 When a job exits during the write phase of a verify workload, save its
2955 current state. This allows fio to replay up until that point, if the verify
2956 state is loaded for the verify read phase. The format of the filename is,
2957 roughly::
2958
f50fbdda 2959 <type>-<jobname>-<jobindex>-verify.state.
f80dba8d
MT
2960
2961 <type> is "local" for a local run, "sock" for a client/server socket
2962 connection, and "ip" (192.168.0.1, for instance) for a networked
d7e6ea1c 2963 client/server connection. Defaults to true.
f80dba8d
MT
2964
2965.. option:: verify_state_load=bool
2966
2967 If a verify termination trigger was used, fio stores the current write state
2968 of each thread. This can be used at verification time so that fio knows how
2969 far it should verify. Without this information, fio will run a full
a47b697c
SW
2970 verification pass, according to the settings in the job file used. Default
2971 false.
f80dba8d
MT
2972
2973.. option:: trim_percentage=int
2974
2975 Number of verify blocks to discard/trim.
2976
2977.. option:: trim_verify_zero=bool
2978
22413915 2979 Verify that trim/discarded blocks are returned as zeros.
f80dba8d
MT
2980
2981.. option:: trim_backlog=int
2982
5cfd1e9a 2983 Trim after this number of blocks are written.
f80dba8d
MT
2984
2985.. option:: trim_backlog_batch=int
2986
2987 Trim this number of I/O blocks.
2988
2989.. option:: experimental_verify=bool
2990
2991 Enable experimental verification.
2992
f80dba8d
MT
2993Steady state
2994~~~~~~~~~~~~
2995
2996.. option:: steadystate=str:float, ss=str:float
2997
2998 Define the criterion and limit for assessing steady state performance. The
2999 first parameter designates the criterion whereas the second parameter sets
3000 the threshold. When the criterion falls below the threshold for the
3001 specified duration, the job will stop. For example, `iops_slope:0.1%` will
3002 direct fio to terminate the job when the least squares regression slope
3003 falls below 0.1% of the mean IOPS. If :option:`group_reporting` is enabled
3004 this will apply to all jobs in the group. Below is the list of available
3005 steady state assessment criteria. All assessments are carried out using only
3006 data from the rolling collection window. Threshold limits can be expressed
3007 as a fixed value or as a percentage of the mean in the collection window.
3008
1cb049d9
VF
3009 When using this feature, most jobs should include the :option:`time_based`
3010 and :option:`runtime` options or the :option:`loops` option so that fio does not
3011 stop running after it has covered the full size of the specified file(s) or device(s).
3012
f80dba8d
MT
3013 **iops**
3014 Collect IOPS data. Stop the job if all individual IOPS measurements
3015 are within the specified limit of the mean IOPS (e.g., ``iops:2``
3016 means that all individual IOPS values must be within 2 of the mean,
3017 whereas ``iops:0.2%`` means that all individual IOPS values must be
3018 within 0.2% of the mean IOPS to terminate the job).
3019
3020 **iops_slope**
3021 Collect IOPS data and calculate the least squares regression
3022 slope. Stop the job if the slope falls below the specified limit.
3023
3024 **bw**
3025 Collect bandwidth data. Stop the job if all individual bandwidth
3026 measurements are within the specified limit of the mean bandwidth.
3027
3028 **bw_slope**
3029 Collect bandwidth data and calculate the least squares regression
3030 slope. Stop the job if the slope falls below the specified limit.
3031
3032.. option:: steadystate_duration=time, ss_dur=time
3033
3034 A rolling window of this duration will be used to judge whether steady state
3035 has been reached. Data will be collected once per second. The default is 0
f75ede1d 3036 which disables steady state detection. When the unit is omitted, the
947e0fe0 3037 value is interpreted in seconds.
f80dba8d
MT
3038
3039.. option:: steadystate_ramp_time=time, ss_ramp=time
3040
3041 Allow the job to run for the specified duration before beginning data
3042 collection for checking the steady state job termination criterion. The
947e0fe0 3043 default is 0. When the unit is omitted, the value is interpreted in seconds.
f80dba8d
MT
3044
3045
3046Measurements and reporting
3047~~~~~~~~~~~~~~~~~~~~~~~~~~
3048
3049.. option:: per_job_logs=bool
3050
3051 If set, this generates bw/clat/iops log with per file private filenames. If
3052 not set, jobs with identical names will share the log filename. Default:
3053 true.
3054
3055.. option:: group_reporting
3056
3057 It may sometimes be interesting to display statistics for groups of jobs as
3058 a whole instead of for each individual job. This is especially true if
3059 :option:`numjobs` is used; looking at individual thread/process output
3060 quickly becomes unwieldy. To see the final report per-group instead of
3061 per-job, use :option:`group_reporting`. Jobs in a file will be part of the
3062 same reporting group, unless if separated by a :option:`stonewall`, or by
3063 using :option:`new_group`.
3064
3065.. option:: new_group
3066
3067 Start a new reporting group. See: :option:`group_reporting`. If not given,
3068 all jobs in a file will be part of the same reporting group, unless
3069 separated by a :option:`stonewall`.
3070
589e88b7 3071.. option:: stats=bool
8243be59
JA
3072
3073 By default, fio collects and shows final output results for all jobs
3074 that run. If this option is set to 0, then fio will ignore it in
3075 the final stat output.
3076
f80dba8d
MT
3077.. option:: write_bw_log=str
3078
3079 If given, write a bandwidth log for this job. Can be used to store data of
074f0817 3080 the bandwidth of the jobs in their lifetime.
f80dba8d 3081
074f0817
SW
3082 If no str argument is given, the default filename of
3083 :file:`jobname_type.x.log` is used. Even when the argument is given, fio
3084 will still append the type of log. So if one specifies::
3085
3086 write_bw_log=foo
f80dba8d 3087
074f0817
SW
3088 The actual log name will be :file:`foo_bw.x.log` where `x` is the index
3089 of the job (`1..N`, where `N` is the number of jobs). If
3090 :option:`per_job_logs` is false, then the filename will not include the
3091 `.x` job index.
e3cedca7 3092
074f0817
SW
3093 The included :command:`fio_generate_plots` script uses :command:`gnuplot` to turn these
3094 text files into nice graphs. See `Log File Formats`_ for how data is
3095 structured within the file.
3096
3097.. option:: write_lat_log=str
e3cedca7 3098
074f0817 3099 Same as :option:`write_bw_log`, except this option creates I/O
77b7e675
SW
3100 submission (e.g., :file:`name_slat.x.log`), completion (e.g.,
3101 :file:`name_clat.x.log`), and total (e.g., :file:`name_lat.x.log`)
074f0817
SW
3102 latency files instead. See :option:`write_bw_log` for details about
3103 the filename format and `Log File Formats`_ for how data is structured
3104 within the files.
be4ecfdf 3105
f80dba8d 3106.. option:: write_hist_log=str
06842027 3107
074f0817 3108 Same as :option:`write_bw_log` but writes an I/O completion latency
77b7e675 3109 histogram file (e.g., :file:`name_hist.x.log`) instead. Note that this
074f0817
SW
3110 file will be empty unless :option:`log_hist_msec` has also been set.
3111 See :option:`write_bw_log` for details about the filename format and
3112 `Log File Formats`_ for how data is structured within the file.
06842027 3113
f80dba8d 3114.. option:: write_iops_log=str
06842027 3115
074f0817 3116 Same as :option:`write_bw_log`, but writes an IOPS file (e.g.
15417073
SW
3117 :file:`name_iops.x.log`) instead. Because fio defaults to individual
3118 I/O logging, the value entry in the IOPS log will be 1 unless windowed
3119 logging (see :option:`log_avg_msec`) has been enabled. See
3120 :option:`write_bw_log` for details about the filename format and `Log
3121 File Formats`_ for how data is structured within the file.
06842027 3122
f80dba8d 3123.. option:: log_avg_msec=int
06842027 3124
f80dba8d
MT
3125 By default, fio will log an entry in the iops, latency, or bw log for every
3126 I/O that completes. When writing to the disk log, that can quickly grow to a
3127 very large size. Setting this option makes fio average the each log entry
3128 over the specified period of time, reducing the resolution of the log. See
3129 :option:`log_max_value` as well. Defaults to 0, logging all entries.
6fc82095 3130 Also see `Log File Formats`_.
06842027 3131
f80dba8d 3132.. option:: log_hist_msec=int
06842027 3133
f80dba8d
MT
3134 Same as :option:`log_avg_msec`, but logs entries for completion latency
3135 histograms. Computing latency percentiles from averages of intervals using
c60ebc45 3136 :option:`log_avg_msec` is inaccurate. Setting this option makes fio log
f80dba8d
MT
3137 histogram entries over the specified period of time, reducing log sizes for
3138 high IOPS devices while retaining percentile accuracy. See
074f0817
SW
3139 :option:`log_hist_coarseness` and :option:`write_hist_log` as well.
3140 Defaults to 0, meaning histogram logging is disabled.
06842027 3141
f80dba8d 3142.. option:: log_hist_coarseness=int
06842027 3143
f80dba8d
MT
3144 Integer ranging from 0 to 6, defining the coarseness of the resolution of
3145 the histogram logs enabled with :option:`log_hist_msec`. For each increment
3146 in coarseness, fio outputs half as many bins. Defaults to 0, for which
074f0817
SW
3147 histogram logs contain 1216 latency bins. See :option:`write_hist_log`
3148 and `Log File Formats`_.
8b28bd41 3149
f80dba8d 3150.. option:: log_max_value=bool
66c098b8 3151
f80dba8d
MT
3152 If :option:`log_avg_msec` is set, fio logs the average over that window. If
3153 you instead want to log the maximum value, set this option to 1. Defaults to
3154 0, meaning that averaged values are logged.
a696fa2a 3155
589e88b7 3156.. option:: log_offset=bool
a696fa2a 3157
f80dba8d 3158 If this is set, the iolog options will include the byte offset for the I/O
5a83478f
SW
3159 entry as well as the other data values. Defaults to 0 meaning that
3160 offsets are not present in logs. Also see `Log File Formats`_.
71bfa161 3161
f80dba8d 3162.. option:: log_compression=int
7de87099 3163
f80dba8d
MT
3164 If this is set, fio will compress the I/O logs as it goes, to keep the
3165 memory footprint lower. When a log reaches the specified size, that chunk is
3166 removed and compressed in the background. Given that I/O logs are fairly
3167 highly compressible, this yields a nice memory savings for longer runs. The
3168 downside is that the compression will consume some background CPU cycles, so
3169 it may impact the run. This, however, is also true if the logging ends up
3170 consuming most of the system memory. So pick your poison. The I/O logs are
3171 saved normally at the end of a run, by decompressing the chunks and storing
3172 them in the specified log file. This feature depends on the availability of
3173 zlib.
e0b0d892 3174
f80dba8d 3175.. option:: log_compression_cpus=str
e0b0d892 3176
f80dba8d
MT
3177 Define the set of CPUs that are allowed to handle online log compression for
3178 the I/O jobs. This can provide better isolation between performance
0cf90a62
SW
3179 sensitive jobs, and background compression work. See
3180 :option:`cpus_allowed` for the format used.
9e684a49 3181
f80dba8d 3182.. option:: log_store_compressed=bool
9e684a49 3183
f80dba8d
MT
3184 If set, fio will store the log files in a compressed format. They can be
3185 decompressed with fio, using the :option:`--inflate-log` command line
3186 parameter. The files will be stored with a :file:`.fz` suffix.
9e684a49 3187
f80dba8d 3188.. option:: log_unix_epoch=bool
9e684a49 3189
f80dba8d
MT
3190 If set, fio will log Unix timestamps to the log files produced by enabling
3191 write_type_log for each log type, instead of the default zero-based
3192 timestamps.
3193
3194.. option:: block_error_percentiles=bool
3195
3196 If set, record errors in trim block-sized units from writes and trims and
3197 output a histogram of how many trims it took to get to errors, and what kind
3198 of error was encountered.
3199
3200.. option:: bwavgtime=int
3201
3202 Average the calculated bandwidth over the given time. Value is specified in
3203 milliseconds. If the job also does bandwidth logging through
3204 :option:`write_bw_log`, then the minimum of this option and
3205 :option:`log_avg_msec` will be used. Default: 500ms.
3206
3207.. option:: iopsavgtime=int
3208
3209 Average the calculated IOPS over the given time. Value is specified in
3210 milliseconds. If the job also does IOPS logging through
3211 :option:`write_iops_log`, then the minimum of this option and
3212 :option:`log_avg_msec` will be used. Default: 500ms.
3213
3214.. option:: disk_util=bool
3215
3216 Generate disk utilization statistics, if the platform supports it.
3217 Default: true.
3218
3219.. option:: disable_lat=bool
3220
3221 Disable measurements of total latency numbers. Useful only for cutting back
3222 the number of calls to :manpage:`gettimeofday(2)`, as that does impact
3223 performance at really high IOPS rates. Note that to really get rid of a
3224 large amount of these calls, this option must be used with
f75ede1d 3225 :option:`disable_slat` and :option:`disable_bw_measurement` as well.
f80dba8d
MT
3226
3227.. option:: disable_clat=bool
3228
3229 Disable measurements of completion latency numbers. See
3230 :option:`disable_lat`.
3231
3232.. option:: disable_slat=bool
3233
3234 Disable measurements of submission latency numbers. See
f50fbdda 3235 :option:`disable_lat`.
f80dba8d 3236
f75ede1d 3237.. option:: disable_bw_measurement=bool, disable_bw=bool
f80dba8d
MT
3238
3239 Disable measurements of throughput/bandwidth numbers. See
3240 :option:`disable_lat`.
3241
3242.. option:: clat_percentiles=bool
3243
b599759b
JA
3244 Enable the reporting of percentiles of completion latencies. This
3245 option is mutually exclusive with :option:`lat_percentiles`.
3246
3247.. option:: lat_percentiles=bool
3248
b71968b1 3249 Enable the reporting of percentiles of I/O latencies. This is similar
b599759b
JA
3250 to :option:`clat_percentiles`, except that this includes the
3251 submission latency. This option is mutually exclusive with
3252 :option:`clat_percentiles`.
f80dba8d
MT
3253
3254.. option:: percentile_list=float_list
3255
c32ba107
JA
3256 Overwrite the default list of percentiles for completion latencies and
3257 the block error histogram. Each number is a floating number in the
3258 range (0,100], and the maximum length of the list is 20. Use ``:`` to
3259 separate the numbers, and list the numbers in ascending order. For
3260 example, ``--percentile_list=99.5:99.9`` will cause fio to report the
3261 values of completion latency below which 99.5% and 99.9% of the observed
3262 latencies fell, respectively.
f80dba8d 3263
e883cb35
JF
3264.. option:: significant_figures=int
3265
c32ba107
JA
3266 If using :option:`--output-format` of `normal`, set the significant
3267 figures to this value. Higher values will yield more precise IOPS and
3268 throughput units, while lower values will round. Requires a minimum
3269 value of 1 and a maximum value of 10. Defaults to 4.
e883cb35 3270
f80dba8d
MT
3271
3272Error handling
3273~~~~~~~~~~~~~~
3274
3275.. option:: exitall_on_error
3276
3277 When one job finishes in error, terminate the rest. The default is to wait
3278 for each job to finish.
3279
3280.. option:: continue_on_error=str
3281
3282 Normally fio will exit the job on the first observed failure. If this option
3283 is set, fio will continue the job when there is a 'non-fatal error' (EIO or
3284 EILSEQ) until the runtime is exceeded or the I/O size specified is
3285 completed. If this option is used, there are two more stats that are
3286 appended, the total error count and the first error. The error field given
3287 in the stats is the first error that was hit during the run.
3288
3289 The allowed values are:
3290
3291 **none**
3292 Exit on any I/O or verify errors.
3293
3294 **read**
3295 Continue on read errors, exit on all others.
3296
3297 **write**
3298 Continue on write errors, exit on all others.
3299
3300 **io**
3301 Continue on any I/O error, exit on all others.
3302
3303 **verify**
3304 Continue on verify errors, exit on all others.
3305
3306 **all**
3307 Continue on all errors.
3308
3309 **0**
3310 Backward-compatible alias for 'none'.
3311
3312 **1**
3313 Backward-compatible alias for 'all'.
3314
3315.. option:: ignore_error=str
3316
3317 Sometimes you want to ignore some errors during test in that case you can
a35ef7cb
TK
3318 specify error list for each error type, instead of only being able to
3319 ignore the default 'non-fatal error' using :option:`continue_on_error`.
f80dba8d
MT
3320 ``ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST`` errors for
3321 given error type is separated with ':'. Error may be symbol ('ENOSPC',
3322 'ENOMEM') or integer. Example::
3323
3324 ignore_error=EAGAIN,ENOSPC:122
3325
3326 This option will ignore EAGAIN from READ, and ENOSPC and 122(EDQUOT) from
a35ef7cb
TK
3327 WRITE. This option works by overriding :option:`continue_on_error` with
3328 the list of errors for each error type if any.
f80dba8d
MT
3329
3330.. option:: error_dump=bool
3331
3332 If set dump every error even if it is non fatal, true by default. If
3333 disabled only fatal error will be dumped.
3334
f75ede1d
SW
3335Running predefined workloads
3336----------------------------
3337
3338Fio includes predefined profiles that mimic the I/O workloads generated by
3339other tools.
3340
3341.. option:: profile=str
3342
3343 The predefined workload to run. Current profiles are:
3344
3345 **tiobench**
3346 Threaded I/O bench (tiotest/tiobench) like workload.
3347
3348 **act**
3349 Aerospike Certification Tool (ACT) like workload.
3350
3351To view a profile's additional options use :option:`--cmdhelp` after specifying
3352the profile. For example::
3353
f50fbdda 3354 $ fio --profile=act --cmdhelp
f75ede1d
SW
3355
3356Act profile options
3357~~~~~~~~~~~~~~~~~~~
3358
3359.. option:: device-names=str
3360 :noindex:
3361
3362 Devices to use.
3363
3364.. option:: load=int
3365 :noindex:
3366
3367 ACT load multiplier. Default: 1.
3368
3369.. option:: test-duration=time
3370 :noindex:
3371
947e0fe0
SW
3372 How long the entire test takes to run. When the unit is omitted, the value
3373 is given in seconds. Default: 24h.
f75ede1d
SW
3374
3375.. option:: threads-per-queue=int
3376 :noindex:
3377
f50fbdda 3378 Number of read I/O threads per device. Default: 8.
f75ede1d
SW
3379
3380.. option:: read-req-num-512-blocks=int
3381 :noindex:
3382
3383 Number of 512B blocks to read at the time. Default: 3.
3384
3385.. option:: large-block-op-kbytes=int
3386 :noindex:
3387
3388 Size of large block ops in KiB (writes). Default: 131072.
3389
3390.. option:: prep
3391 :noindex:
3392
3393 Set to run ACT prep phase.
3394
3395Tiobench profile options
3396~~~~~~~~~~~~~~~~~~~~~~~~
3397
3398.. option:: size=str
3399 :noindex:
3400
f50fbdda 3401 Size in MiB.
f75ede1d
SW
3402
3403.. option:: block=int
3404 :noindex:
3405
3406 Block size in bytes. Default: 4096.
3407
3408.. option:: numruns=int
3409 :noindex:
3410
3411 Number of runs.
3412
3413.. option:: dir=str
3414 :noindex:
3415
3416 Test directory.
3417
3418.. option:: threads=int
3419 :noindex:
3420
3421 Number of threads.
f80dba8d
MT
3422
3423Interpreting the output
3424-----------------------
3425
36214730
SW
3426..
3427 Example output was based on the following:
3428 TZ=UTC fio --iodepth=8 --ioengine=null --size=100M --time_based \
3429 --rate=1256k --bs=14K --name=quick --runtime=1s --name=mixed \
3430 --runtime=2m --rw=rw
3431
f80dba8d
MT
3432Fio spits out a lot of output. While running, fio will display the status of the
3433jobs created. An example of that would be::
3434
9d25d068 3435 Jobs: 1 (f=1): [_(1),M(1)][24.8%][r=20.5MiB/s,w=23.5MiB/s][r=82,w=94 IOPS][eta 01m:31s]
f80dba8d 3436
36214730
SW
3437The characters inside the first set of square brackets denote the current status of
3438each thread. The first character is the first job defined in the job file, and so
3439forth. The possible values (in typical life cycle order) are:
f80dba8d
MT
3440
3441+------+-----+-----------------------------------------------------------+
3442| Idle | Run | |
3443+======+=====+===========================================================+
3444| P | | Thread setup, but not started. |
3445+------+-----+-----------------------------------------------------------+
3446| C | | Thread created. |
3447+------+-----+-----------------------------------------------------------+
3448| I | | Thread initialized, waiting or generating necessary data. |
3449+------+-----+-----------------------------------------------------------+
3450| | p | Thread running pre-reading file(s). |
3451+------+-----+-----------------------------------------------------------+
36214730
SW
3452| | / | Thread is in ramp period. |
3453+------+-----+-----------------------------------------------------------+
f80dba8d
MT
3454| | R | Running, doing sequential reads. |
3455+------+-----+-----------------------------------------------------------+
3456| | r | Running, doing random reads. |
3457+------+-----+-----------------------------------------------------------+
3458| | W | Running, doing sequential writes. |
3459+------+-----+-----------------------------------------------------------+
3460| | w | Running, doing random writes. |
3461+------+-----+-----------------------------------------------------------+
3462| | M | Running, doing mixed sequential reads/writes. |
3463+------+-----+-----------------------------------------------------------+
3464| | m | Running, doing mixed random reads/writes. |
3465+------+-----+-----------------------------------------------------------+
36214730
SW
3466| | D | Running, doing sequential trims. |
3467+------+-----+-----------------------------------------------------------+
3468| | d | Running, doing random trims. |
3469+------+-----+-----------------------------------------------------------+
3470| | F | Running, currently waiting for :manpage:`fsync(2)`. |
f80dba8d
MT
3471+------+-----+-----------------------------------------------------------+
3472| | V | Running, doing verification of written data. |
3473+------+-----+-----------------------------------------------------------+
36214730
SW
3474| f | | Thread finishing. |
3475+------+-----+-----------------------------------------------------------+
f80dba8d
MT
3476| E | | Thread exited, not reaped by main thread yet. |
3477+------+-----+-----------------------------------------------------------+
36214730 3478| _ | | Thread reaped. |
f80dba8d
MT
3479+------+-----+-----------------------------------------------------------+
3480| X | | Thread reaped, exited with an error. |
3481+------+-----+-----------------------------------------------------------+
3482| K | | Thread reaped, exited due to signal. |
3483+------+-----+-----------------------------------------------------------+
3484
36214730
SW
3485..
3486 Example output was based on the following:
3487 TZ=UTC fio --iodepth=8 --ioengine=null --size=100M --runtime=58m \
3488 --time_based --rate=2512k --bs=256K --numjobs=10 \
3489 --name=readers --rw=read --name=writers --rw=write
3490
f80dba8d 3491Fio will condense the thread string as not to take up more space on the command
36214730 3492line than needed. For instance, if you have 10 readers and 10 writers running,
f80dba8d
MT
3493the output would look like this::
3494
9d25d068 3495 Jobs: 20 (f=20): [R(10),W(10)][4.0%][r=20.5MiB/s,w=23.5MiB/s][r=82,w=94 IOPS][eta 57m:36s]
f80dba8d 3496
36214730
SW
3497Note that the status string is displayed in order, so it's possible to tell which of
3498the jobs are currently doing what. In the example above this means that jobs 1--10
3499are readers and 11--20 are writers.
f80dba8d
MT
3500
3501The other values are fairly self explanatory -- number of threads currently
36214730
SW
3502running and doing I/O, the number of currently open files (f=), the estimated
3503completion percentage, the rate of I/O since last check (read speed listed first,
f50fbdda
TK
3504then write speed and optionally trim speed) in terms of bandwidth and IOPS,
3505and time to completion for the current running group. It's impossible to estimate
3506runtime of the following groups (if any).
36214730
SW
3507
3508..
3509 Example output was based on the following:
3510 TZ=UTC fio --iodepth=16 --ioengine=posixaio --filename=/tmp/fiofile \
3511 --direct=1 --size=100M --time_based --runtime=50s --rate_iops=89 \
3512 --bs=7K --name=Client1 --rw=write
3513
3514When fio is done (or interrupted by :kbd:`Ctrl-C`), it will show the data for
3515each thread, group of threads, and disks in that order. For each overall thread (or
3516group) the output looks like::
3517
3518 Client1: (groupid=0, jobs=1): err= 0: pid=16109: Sat Jun 24 12:07:54 2017
3519 write: IOPS=88, BW=623KiB/s (638kB/s)(30.4MiB/50032msec)
3520 slat (nsec): min=500, max=145500, avg=8318.00, stdev=4781.50
3521 clat (usec): min=170, max=78367, avg=4019.02, stdev=8293.31
3522 lat (usec): min=174, max=78375, avg=4027.34, stdev=8291.79
3523 clat percentiles (usec):
3524 | 1.00th=[ 302], 5.00th=[ 326], 10.00th=[ 343], 20.00th=[ 363],
3525 | 30.00th=[ 392], 40.00th=[ 404], 50.00th=[ 416], 60.00th=[ 445],
3526 | 70.00th=[ 816], 80.00th=[ 6718], 90.00th=[12911], 95.00th=[21627],
3527 | 99.00th=[43779], 99.50th=[51643], 99.90th=[68682], 99.95th=[72877],
3528 | 99.99th=[78119]
3529 bw ( KiB/s): min= 532, max= 686, per=0.10%, avg=622.87, stdev=24.82, samples= 100
3530 iops : min= 76, max= 98, avg=88.98, stdev= 3.54, samples= 100
29092211
VF
3531 lat (usec) : 250=0.04%, 500=64.11%, 750=4.81%, 1000=2.79%
3532 lat (msec) : 2=4.16%, 4=1.84%, 10=4.90%, 20=11.33%, 50=5.37%
3533 lat (msec) : 100=0.65%
36214730
SW
3534 cpu : usr=0.27%, sys=0.18%, ctx=12072, majf=0, minf=21
3535 IO depths : 1=85.0%, 2=13.1%, 4=1.8%, 8=0.1%, 16=0.0%, 32=0.0%, >=64=0.0%
3536 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
3537 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
3538 issued rwt: total=0,4450,0, short=0,0,0, dropped=0,0,0
3539 latency : target=0, window=0, percentile=100.00%, depth=8
3540
3541The job name (or first job's name when using :option:`group_reporting`) is printed,
3542along with the group id, count of jobs being aggregated, last error id seen (which
3543is 0 when there are no errors), pid/tid of that thread and the time the job/group
3544completed. Below are the I/O statistics for each data direction performed (showing
3545writes in the example above). In the order listed, they denote:
3546
3547**read/write/trim**
3548 The string before the colon shows the I/O direction the statistics
3549 are for. **IOPS** is the average I/Os performed per second. **BW**
3550 is the average bandwidth rate shown as: value in power of 2 format
3551 (value in power of 10 format). The last two values show: (**total
3552 I/O performed** in power of 2 format / **runtime** of that thread).
f80dba8d
MT
3553
3554**slat**
36214730
SW
3555 Submission latency (**min** being the minimum, **max** being the
3556 maximum, **avg** being the average, **stdev** being the standard
3557 deviation). This is the time it took to submit the I/O. For
3558 sync I/O this row is not displayed as the slat is really the
3559 completion latency (since queue/complete is one operation there).
3560 This value can be in nanoseconds, microseconds or milliseconds ---
3561 fio will choose the most appropriate base and print that (in the
3562 example above nanoseconds was the best scale). Note: in :option:`--minimal` mode
0d237712 3563 latencies are always expressed in microseconds.
f80dba8d
MT
3564
3565**clat**
3566 Completion latency. Same names as slat, this denotes the time from
3567 submission to completion of the I/O pieces. For sync I/O, clat will
3568 usually be equal (or very close) to 0, as the time from submit to
3569 complete is basically just CPU time (I/O has already been done, see slat
3570 explanation).
3571
29092211
VF
3572**lat**
3573 Total latency. Same names as slat and clat, this denotes the time from
3574 when fio created the I/O unit to completion of the I/O operation.
3575
f80dba8d 3576**bw**
36214730
SW
3577 Bandwidth statistics based on samples. Same names as the xlat stats,
3578 but also includes the number of samples taken (**samples**) and an
3579 approximate percentage of total aggregate bandwidth this thread
3580 received in its group (**per**). This last value is only really
3581 useful if the threads in this group are on the same disk, since they
3582 are then competing for disk access.
3583
3584**iops**
3585 IOPS statistics based on samples. Same names as bw.
f80dba8d 3586
29092211
VF
3587**lat (nsec/usec/msec)**
3588 The distribution of I/O completion latencies. This is the time from when
3589 I/O leaves fio and when it gets completed. Unlike the separate
3590 read/write/trim sections above, the data here and in the remaining
3591 sections apply to all I/Os for the reporting group. 250=0.04% means that
3592 0.04% of the I/Os completed in under 250us. 500=64.11% means that 64.11%
3593 of the I/Os required 250 to 499us for completion.
3594
f80dba8d
MT
3595**cpu**
3596 CPU usage. User and system time, along with the number of context
3597 switches this thread went through, usage of system and user time, and
3598 finally the number of major and minor page faults. The CPU utilization
3599 numbers are averages for the jobs in that reporting group, while the
23a8e176 3600 context and fault counters are summed.
f80dba8d
MT
3601
3602**IO depths**
a2140525
SW
3603 The distribution of I/O depths over the job lifetime. The numbers are
3604 divided into powers of 2 and each entry covers depths from that value
3605 up to those that are lower than the next entry -- e.g., 16= covers
3606 depths from 16 to 31. Note that the range covered by a depth
3607 distribution entry can be different to the range covered by the
3608 equivalent submit/complete distribution entry.
f80dba8d
MT
3609
3610**IO submit**
3611 How many pieces of I/O were submitting in a single submit call. Each
c60ebc45 3612 entry denotes that amount and below, until the previous entry -- e.g.,
a2140525
SW
3613 16=100% means that we submitted anywhere between 9 to 16 I/Os per submit
3614 call. Note that the range covered by a submit distribution entry can
3615 be different to the range covered by the equivalent depth distribution
3616 entry.
f80dba8d
MT
3617
3618**IO complete**
3619 Like the above submit number, but for completions instead.
3620
36214730
SW
3621**IO issued rwt**
3622 The number of read/write/trim requests issued, and how many of them were
3623 short or dropped.
f80dba8d 3624
29092211 3625**IO latency**
ee21ebee 3626 These values are for :option:`latency_target` and related options. When
29092211
VF
3627 these options are engaged, this section describes the I/O depth required
3628 to meet the specified latency target.
71bfa161 3629
36214730
SW
3630..
3631 Example output was based on the following:
3632 TZ=UTC fio --ioengine=null --iodepth=2 --size=100M --numjobs=2 \
3633 --rate_process=poisson --io_limit=32M --name=read --bs=128k \
3634 --rate=11M --name=write --rw=write --bs=2k --rate=700k
3635
71bfa161 3636After each client has been listed, the group statistics are printed. They
f80dba8d 3637will look like this::
71bfa161 3638
f80dba8d 3639 Run status group 0 (all jobs):
36214730
SW
3640 READ: bw=20.9MiB/s (21.9MB/s), 10.4MiB/s-10.8MiB/s (10.9MB/s-11.3MB/s), io=64.0MiB (67.1MB), run=2973-3069msec
3641 WRITE: bw=1231KiB/s (1261kB/s), 616KiB/s-621KiB/s (630kB/s-636kB/s), io=64.0MiB (67.1MB), run=52747-53223msec
71bfa161 3642
36214730 3643For each data direction it prints:
71bfa161 3644
36214730
SW
3645**bw**
3646 Aggregate bandwidth of threads in this group followed by the
3647 minimum and maximum bandwidth of all the threads in this group.
3648 Values outside of brackets are power-of-2 format and those
3649 within are the equivalent value in a power-of-10 format.
f80dba8d 3650**io**
36214730
SW
3651 Aggregate I/O performed of all threads in this group. The
3652 format is the same as bw.
3653**run**
3654 The smallest and longest runtimes of the threads in this group.
71bfa161 3655
f50fbdda 3656And finally, the disk statistics are printed. This is Linux specific. They will look like this::
71bfa161 3657
f80dba8d
MT
3658 Disk stats (read/write):
3659 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
71bfa161
JA
3660
3661Each value is printed for both reads and writes, with reads first. The
3662numbers denote:
3663
f80dba8d 3664**ios**
c60ebc45 3665 Number of I/Os performed by all groups.
f80dba8d 3666**merge**
007c7be9 3667 Number of merges performed by the I/O scheduler.
f80dba8d
MT
3668**ticks**
3669 Number of ticks we kept the disk busy.
36214730 3670**in_queue**
f80dba8d
MT
3671 Total time spent in the disk queue.
3672**util**
3673 The disk utilization. A value of 100% means we kept the disk
71bfa161
JA
3674 busy constantly, 50% would be a disk idling half of the time.
3675
f80dba8d
MT
3676It is also possible to get fio to dump the current output while it is running,
3677without terminating the job. To do that, send fio the **USR1** signal. You can
3678also get regularly timed dumps by using the :option:`--status-interval`
3679parameter, or by creating a file in :file:`/tmp` named
3680:file:`fio-dump-status`. If fio sees this file, it will unlink it and dump the
3681current output status.
8423bd11 3682
71bfa161 3683
f80dba8d
MT
3684Terse output
3685------------
71bfa161 3686
f80dba8d
MT
3687For scripted usage where you typically want to generate tables or graphs of the
3688results, fio can output the results in a semicolon separated format. The format
3689is one long line of values, such as::
71bfa161 3690
f80dba8d
MT
3691 2;card0;0;0;7139336;121836;60004;1;10109;27.932460;116.933948;220;126861;3495.446807;1085.368601;226;126864;3523.635629;1089.012448;24063;99944;50.275485%;59818.274627;5540.657370;7155060;122104;60004;1;8338;29.086342;117.839068;388;128077;5032.488518;1234.785715;391;128085;5061.839412;1236.909129;23436;100928;50.287926%;59964.832030;5644.844189;14.595833%;19.394167%;123706;0;7313;0.1%;0.1%;0.1%;0.1%;0.1%;0.1%;100.0%;0.00%;0.00%;0.00%;0.00%;0.00%;0.00%;0.01%;0.02%;0.05%;0.16%;6.04%;40.40%;52.68%;0.64%;0.01%;0.00%;0.01%;0.00%;0.00%;0.00%;0.00%;0.00%
3692 A description of this job goes here.
562c2d2f
DN
3693
3694The job description (if provided) follows on a second line.
71bfa161 3695
a7f77fa6
SW
3696To enable terse output, use the :option:`--minimal` or
3697:option:`--output-format`\=terse command line options. The
f80dba8d
MT
3698first value is the version of the terse output format. If the output has to be
3699changed for some reason, this number will be incremented by 1 to signify that
3700change.
6820cb3b 3701
a2c95580 3702Split up, the format is as follows (comments in brackets denote when a
007c7be9 3703field was introduced or whether it's specific to some terse version):
71bfa161 3704
f80dba8d
MT
3705 ::
3706
f50fbdda 3707 terse version, fio version [v3], jobname, groupid, error
f80dba8d
MT
3708
3709 READ status::
3710
3711 Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec)
3712 Submission latency: min, max, mean, stdev (usec)
3713 Completion latency: min, max, mean, stdev (usec)
3714 Completion latency percentiles: 20 fields (see below)
3715 Total latency: min, max, mean, stdev (usec)
f50fbdda
TK
3716 Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev, number of samples [v5]
3717 IOPS [v5]: min, max, mean, stdev, number of samples
f80dba8d
MT
3718
3719 WRITE status:
3720
3721 ::
3722
3723 Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec)
3724 Submission latency: min, max, mean, stdev (usec)
247823cc 3725 Completion latency: min, max, mean, stdev (usec)
f80dba8d
MT
3726 Completion latency percentiles: 20 fields (see below)
3727 Total latency: min, max, mean, stdev (usec)
f50fbdda
TK
3728 Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev, number of samples [v5]
3729 IOPS [v5]: min, max, mean, stdev, number of samples
a2c95580
AH
3730
3731 TRIM status [all but version 3]:
3732
f50fbdda 3733 Fields are similar to READ/WRITE status.
f80dba8d
MT
3734
3735 CPU usage::
3736
3737 user, system, context switches, major faults, minor faults
3738
3739 I/O depths::
3740
3741 <=1, 2, 4, 8, 16, 32, >=64
3742
3743 I/O latencies microseconds::
3744
3745 <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
3746
3747 I/O latencies milliseconds::
3748
3749 <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
3750
a2c95580 3751 Disk utilization [v3]::
f80dba8d 3752
f50fbdda
TK
3753 disk name, read ios, write ios, read merges, write merges, read ticks, write ticks,
3754 time spent in queue, disk utilization percentage
f80dba8d
MT
3755
3756 Additional Info (dependent on continue_on_error, default off)::
3757
3758 total # errors, first error code
3759
3760 Additional Info (dependent on description being set)::
3761
3762 Text description
3763
3764Completion latency percentiles can be a grouping of up to 20 sets, so for the
3765terse output fio writes all of them. Each field will look like this::
1db92cb6 3766
f50fbdda 3767 1.00%=6112
1db92cb6 3768
f80dba8d 3769which is the Xth percentile, and the `usec` latency associated with it.
1db92cb6 3770
f50fbdda 3771For `Disk utilization`, all disks used by fio are shown. So for each disk there
f80dba8d 3772will be a disk utilization section.
f2f788dd 3773
2fc26c3d 3774Below is a single line containing short names for each of the fields in the
2831be97 3775minimal output v3, separated by semicolons::
2fc26c3d 3776
f50fbdda 3777 terse_version_3;fio_version;jobname;groupid;error;read_kb;read_bandwidth;read_iops;read_runtime_ms;read_slat_min;read_slat_max;read_slat_mean;read_slat_dev;read_clat_min;read_clat_max;read_clat_mean;read_clat_dev;read_clat_pct01;read_clat_pct02;read_clat_pct03;read_clat_pct04;read_clat_pct05;read_clat_pct06;read_clat_pct07;read_clat_pct08;read_clat_pct09;read_clat_pct10;read_clat_pct11;read_clat_pct12;read_clat_pct13;read_clat_pct14;read_clat_pct15;read_clat_pct16;read_clat_pct17;read_clat_pct18;read_clat_pct19;read_clat_pct20;read_tlat_min;read_lat_max;read_lat_mean;read_lat_dev;read_bw_min;read_bw_max;read_bw_agg_pct;read_bw_mean;read_bw_dev;write_kb;write_bandwidth;write_iops;write_runtime_ms;write_slat_min;write_slat_max;write_slat_mean;write_slat_dev;write_clat_min;write_clat_max;write_clat_mean;write_clat_dev;write_clat_pct01;write_clat_pct02;write_clat_pct03;write_clat_pct04;write_clat_pct05;write_clat_pct06;write_clat_pct07;write_clat_pct08;write_clat_pct09;write_clat_pct10;write_clat_pct11;write_clat_pct12;write_clat_pct13;write_clat_pct14;write_clat_pct15;write_clat_pct16;write_clat_pct17;write_clat_pct18;write_clat_pct19;write_clat_pct20;write_tlat_min;write_lat_max;write_lat_mean;write_lat_dev;write_bw_min;write_bw_max;write_bw_agg_pct;write_bw_mean;write_bw_dev;cpu_user;cpu_sys;cpu_csw;cpu_mjf;cpu_minf;iodepth_1;iodepth_2;iodepth_4;iodepth_8;iodepth_16;iodepth_32;iodepth_64;lat_2us;lat_4us;lat_10us;lat_20us;lat_50us;lat_100us;lat_250us;lat_500us;lat_750us;lat_1000us;lat_2ms;lat_4ms;lat_10ms;lat_20ms;lat_50ms;lat_100ms;lat_250ms;lat_500ms;lat_750ms;lat_1000ms;lat_2000ms;lat_over_2000ms;disk_name;disk_read_iops;disk_write_iops;disk_read_merges;disk_write_merges;disk_read_ticks;write_ticks;disk_queue_time;disk_util
2fc26c3d 3778
25c8b9d7 3779
44c82dba
VF
3780JSON output
3781------------
3782
3783The `json` output format is intended to be both human readable and convenient
3784for automated parsing. For the most part its sections mirror those of the
3785`normal` output. The `runtime` value is reported in msec and the `bw` value is
3786reported in 1024 bytes per second units.
3787
3788
d29c4a91
VF
3789JSON+ output
3790------------
3791
3792The `json+` output format is identical to the `json` output format except that it
3793adds a full dump of the completion latency bins. Each `bins` object contains a
3794set of (key, value) pairs where keys are latency durations and values count how
3795many I/Os had completion latencies of the corresponding duration. For example,
3796consider:
3797