4 The first step in getting fio to simulate a desired I/O workload, is writing a
5 job file describing that specific setup. A job file may contain any number of
6 threads and/or files -- the typical contents of the job file is a *global*
7 section defining shared parameters, and one or more job sections describing the
8 jobs involved. When run, fio parses this file and sets everything up as
9 described. If we break down a job from top to bottom, it contains the following
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?
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.
26 How much data are we going to be reading/writing.
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
36 If the I/O engine is async, how large a queuing depth do we want to
42 How many files are we spreading the workload over.
44 `Threads, processes and job synchronization`_
46 How many threads or processes should we spread this workload over.
48 The above are the basic parameters defined for a workload, in addition there's a
49 multitude of parameters that modify other aspects of how this job behaves.
55 .. option:: --debug=type
57 Enable verbose tracing `type` of various fio actions. May be ``all`` for all types
58 or individual types separated by a comma (e.g. ``--debug=file,mem`` will
59 enable file and memory debugging). Currently, additional logging is
63 Dump info related to processes.
65 Dump info related to file actions.
67 Dump info related to I/O queuing.
69 Dump info related to memory allocations.
71 Dump info related to blktrace setup.
73 Dump info related to I/O verification.
75 Enable all debug options.
77 Dump info related to random offset generation.
79 Dump info related to option matching and parsing.
81 Dump info related to disk utilization updates.
83 Dump info only related to job number x.
85 Dump info only related to mutex up/down ops.
87 Dump info related to profile extensions.
89 Dump info related to internal time keeping.
91 Dump info related to networking connections.
93 Dump info related to I/O rate switching.
95 Dump info related to log compress/decompress.
97 Dump info related to steadystate detection.
99 Dump info related to the helper thread.
101 Dump info related to support for zoned block devices.
103 Show available debug options.
105 .. option:: --parse-only
107 Parse options only, don't start any I/O.
109 .. option:: --merge-blktrace-only
111 Merge blktraces only, don't start any I/O.
113 .. option:: --output=filename
115 Write output to file `filename`.
117 .. option:: --output-format=format
119 Set the reporting `format` to `normal`, `terse`, `json`, or `json+`. Multiple
120 formats can be selected, separated by a comma. `terse` is a CSV based
121 format. `json+` is like `json`, except it adds a full dump of the latency
124 .. option:: --bandwidth-log
126 Generate aggregate bandwidth logs.
128 .. option:: --minimal
130 Print statistics in a terse, semicolon-delimited format.
132 .. option:: --append-terse
134 Print statistics in selected mode AND terse, semicolon-delimited format.
135 **Deprecated**, use :option:`--output-format` instead to select multiple
138 .. option:: --terse-version=version
140 Set terse `version` output format (default 3, or 2 or 4 or 5).
142 .. option:: --version
144 Print version information and exit.
148 Print a summary of the command line options and exit.
150 .. option:: --cpuclock-test
152 Perform test and validation of internal CPU clock.
154 .. option:: --crctest=[test]
156 Test the speed of the built-in checksumming functions. If no argument is
157 given, all of them are tested. Alternatively, a comma separated list can
158 be passed, in which case the given ones are tested.
160 .. option:: --cmdhelp=command
162 Print help information for `command`. May be ``all`` for all commands.
164 .. option:: --enghelp=[ioengine[,command]]
166 List all commands defined by `ioengine`, or print help for `command`
167 defined by `ioengine`. If no `ioengine` is given, list all
170 .. option:: --showcmd
172 Convert given job files to a set of command-line options.
174 .. option:: --readonly
176 Turn on safety read-only checks, preventing writes and trims. The
177 ``--readonly`` option is an extra safety guard to prevent users from
178 accidentally starting a write or trim workload when that is not desired.
179 Fio will only modify the device under test if
180 `rw=write/randwrite/rw/randrw/trim/randtrim/trimwrite` is given. This
181 safety net can be used as an extra precaution.
183 .. option:: --eta=when
185 Specifies when real-time ETA estimate should be printed. `when` may be
186 `always`, `never` or `auto`. `auto` is the default, it prints ETA
187 when requested if the output is a TTY. `always` disregards the output
188 type, and prints ETA when requested. `never` never prints ETA.
190 .. option:: --eta-interval=time
192 By default, fio requests client ETA status roughly every second. With
193 this option, the interval is configurable. Fio imposes a minimum
194 allowed time to avoid flooding the console, less than 250 msec is
197 .. option:: --eta-newline=time
199 Force a new line for every `time` period passed. When the unit is omitted,
200 the value is interpreted in seconds.
202 .. option:: --status-interval=time
204 Force a full status dump of cumulative (from job start) values at `time`
205 intervals. This option does *not* provide per-period measurements. So
206 values such as bandwidth are running averages. When the time unit is omitted,
207 `time` is interpreted in seconds. Note that using this option with
208 ``--output-format=json`` will yield output that technically isn't valid
209 json, since the output will be collated sets of valid json. It will need
210 to be split into valid sets of json after the run.
212 .. option:: --section=name
214 Only run specified section `name` in job file. Multiple sections can be specified.
215 The ``--section`` option allows one to combine related jobs into one file.
216 E.g. one job file could define light, moderate, and heavy sections. Tell
217 fio to run only the "heavy" section by giving ``--section=heavy``
218 command line option. One can also specify the "write" operations in one
219 section and "verify" operation in another section. The ``--section`` option
220 only applies to job sections. The reserved *global* section is always
223 .. option:: --alloc-size=kb
225 Allocate additional internal smalloc pools of size `kb` in KiB. The
226 ``--alloc-size`` option increases shared memory set aside for use by fio.
227 If running large jobs with randommap enabled, fio can run out of memory.
228 Smalloc is an internal allocator for shared structures from a fixed size
229 memory pool and can grow to 16 pools. The pool size defaults to 16MiB.
231 NOTE: While running :file:`.fio_smalloc.*` backing store files are visible
234 .. option:: --warnings-fatal
236 All fio parser warnings are fatal, causing fio to exit with an
239 .. option:: --max-jobs=nr
241 Set the maximum number of threads/processes to support to `nr`.
242 NOTE: On Linux, it may be necessary to increase the shared-memory
243 limit (:file:`/proc/sys/kernel/shmmax`) if fio runs into errors while
246 .. option:: --server=args
248 Start a backend server, with `args` specifying what to listen to.
249 See `Client/Server`_ section.
251 .. option:: --daemonize=pidfile
253 Background a fio server, writing the pid to the given `pidfile` file.
255 .. option:: --client=hostname
257 Instead of running the jobs locally, send and run them on the given `hostname`
258 or set of `hostname`\s. See `Client/Server`_ section.
260 .. option:: --remote-config=file
262 Tell fio server to load this local `file`.
264 .. option:: --idle-prof=option
266 Report CPU idleness. `option` is one of the following:
269 Run unit work calibration only and exit.
272 Show aggregate system idleness and unit work.
275 As **system** but also show per CPU idleness.
277 .. option:: --inflate-log=log
279 Inflate and output compressed `log`.
281 .. option:: --trigger-file=file
283 Execute trigger command when `file` exists.
285 .. option:: --trigger-timeout=time
287 Execute trigger at this `time`.
289 .. option:: --trigger=command
291 Set this `command` as local trigger.
293 .. option:: --trigger-remote=command
295 Set this `command` as remote trigger.
297 .. option:: --aux-path=path
299 Use the directory specified by `path` for generated state files instead
300 of the current working directory.
302 Any parameters following the options will be assumed to be job files, unless
303 they match a job file parameter. Multiple job files can be listed and each job
304 file will be regarded as a separate group. Fio will :option:`stonewall`
305 execution between each group.
311 As previously described, fio accepts one or more job files describing what it is
312 supposed to do. The job file format is the classic ini file, where the names
313 enclosed in [] brackets define the job name. You are free to use any ASCII name
314 you want, except *global* which has special meaning. Following the job name is
315 a sequence of zero or more parameters, one per line, that define the behavior of
316 the job. If the first character in a line is a ';' or a '#', the entire line is
317 discarded as a comment.
319 A *global* section sets defaults for the jobs described in that file. A job may
320 override a *global* section parameter, and a job file may even have several
321 *global* sections if so desired. A job is only affected by a *global* section
324 The :option:`--cmdhelp` option also lists all options. If used with a `command`
325 argument, :option:`--cmdhelp` will detail the given `command`.
327 See the `examples/` directory for inspiration on how to write job files. Note
328 the copyright and license requirements currently apply to `examples/` files.
330 So let's look at a really simple job file that defines two processes, each
331 randomly reading from a 128MiB file:
335 ; -- start job file --
346 As you can see, the job file sections themselves are empty as all the described
347 parameters are shared. As no :option:`filename` option is given, fio makes up a
348 `filename` for each of the jobs as it sees fit. On the command line, this job
349 would look as follows::
351 $ fio --name=global --rw=randread --size=128m --name=job1 --name=job2
354 Let's look at an example that has a number of processes writing randomly to
359 ; -- start job file --
370 Here we have no *global* section, as we only have one job defined anyway. We
371 want to use async I/O here, with a depth of 4 for each file. We also increased
372 the buffer size used to 32KiB and define numjobs to 4 to fork 4 identical
373 jobs. The result is 4 processes each randomly writing to their own 64MiB
374 file. Instead of using the above job file, you could have given the parameters
375 on the command line. For this case, you would specify::
377 $ fio --name=random-writers --ioengine=libaio --iodepth=4 --rw=randwrite --bs=32k --direct=0 --size=64m --numjobs=4
379 When fio is utilized as a basis of any reasonably large test suite, it might be
380 desirable to share a set of standardized settings across multiple job files.
381 Instead of copy/pasting such settings, any section may pull in an external
382 :file:`filename.fio` file with *include filename* directive, as in the following
385 ; -- start job file including.fio --
389 include glob-include.fio
396 include test-include.fio
397 ; -- end job file including.fio --
401 ; -- start job file glob-include.fio --
404 ; -- end job file glob-include.fio --
408 ; -- start job file test-include.fio --
411 ; -- end job file test-include.fio --
413 Settings pulled into a section apply to that section only (except *global*
414 section). Include directives may be nested in that any included file may contain
415 further include directive(s). Include files may not contain [] sections.
418 Environment variables
419 ~~~~~~~~~~~~~~~~~~~~~
421 Fio also supports environment variable expansion in job files. Any sub-string of
422 the form ``${VARNAME}`` as part of an option value (in other words, on the right
423 of the '='), will be expanded to the value of the environment variable called
424 `VARNAME`. If no such environment variable is defined, or `VARNAME` is the
425 empty string, the empty string will be substituted.
427 As an example, let's look at a sample fio invocation and job file::
429 $ SIZE=64m NUMJOBS=4 fio jobfile.fio
433 ; -- start job file --
440 This will expand to the following equivalent job file at runtime:
444 ; -- start job file --
451 Fio ships with a few example job files, you can also look there for inspiration.
456 Additionally, fio has a set of reserved keywords that will be replaced
457 internally with the appropriate value. Those keywords are:
461 The architecture page size of the running system.
465 Megabytes of total memory in the system.
469 Number of online available CPUs.
471 These can be used on the command line or in the job file, and will be
472 automatically substituted with the current system values when the job is
473 run. Simple math is also supported on these keywords, so you can perform actions
478 and get that properly expanded to 8 times the size of memory in the machine.
484 This section describes in details each parameter associated with a job. Some
485 parameters take an option of a given type, such as an integer or a
486 string. Anywhere a numeric value is required, an arithmetic expression may be
487 used, provided it is surrounded by parentheses. Supported operators are:
496 For time values in expressions, units are microseconds by default. This is
497 different than for time values not in expressions (not enclosed in
498 parentheses). The following types are used:
505 String: A sequence of alphanumeric characters.
508 Integer with possible time suffix. Without a unit value is interpreted as
509 seconds unless otherwise specified. Accepts a suffix of 'd' for days, 'h' for
510 hours, 'm' for minutes, 's' for seconds, 'ms' (or 'msec') for milliseconds and
511 'us' (or 'usec') for microseconds. For example, use 10m for 10 minutes.
516 Integer. A whole number value, which may contain an integer prefix
517 and an integer suffix:
519 [*integer prefix*] **number** [*integer suffix*]
521 The optional *integer prefix* specifies the number's base. The default
522 is decimal. *0x* specifies hexadecimal.
524 The optional *integer suffix* specifies the number's units, and includes an
525 optional unit prefix and an optional unit. For quantities of data, the
526 default unit is bytes. For quantities of time, the default unit is seconds
527 unless otherwise specified.
529 With :option:`kb_base`\=1000, fio follows international standards for unit
530 prefixes. To specify power-of-10 decimal values defined in the
531 International System of Units (SI):
533 * *K* -- means kilo (K) or 1000
534 * *M* -- means mega (M) or 1000**2
535 * *G* -- means giga (G) or 1000**3
536 * *T* -- means tera (T) or 1000**4
537 * *P* -- means peta (P) or 1000**5
539 To specify power-of-2 binary values defined in IEC 80000-13:
541 * *Ki* -- means kibi (Ki) or 1024
542 * *Mi* -- means mebi (Mi) or 1024**2
543 * *Gi* -- means gibi (Gi) or 1024**3
544 * *Ti* -- means tebi (Ti) or 1024**4
545 * *Pi* -- means pebi (Pi) or 1024**5
547 For Zone Block Device Mode:
550 With :option:`kb_base`\=1024 (the default), the unit prefixes are opposite
551 from those specified in the SI and IEC 80000-13 standards to provide
552 compatibility with old scripts. For example, 4k means 4096.
554 For quantities of data, an optional unit of 'B' may be included
555 (e.g., 'kB' is the same as 'k').
557 The *integer suffix* is not case sensitive (e.g., m/mi mean mebi/mega,
558 not milli). 'b' and 'B' both mean byte, not bit.
560 Examples with :option:`kb_base`\=1000:
562 * *4 KiB*: 4096, 4096b, 4096B, 4ki, 4kib, 4kiB, 4Ki, 4KiB
563 * *1 MiB*: 1048576, 1mi, 1024ki
564 * *1 MB*: 1000000, 1m, 1000k
565 * *1 TiB*: 1099511627776, 1ti, 1024gi, 1048576mi
566 * *1 TB*: 1000000000, 1t, 1000m, 1000000k
568 Examples with :option:`kb_base`\=1024 (default):
570 * *4 KiB*: 4096, 4096b, 4096B, 4k, 4kb, 4kB, 4K, 4KB
571 * *1 MiB*: 1048576, 1m, 1024k
572 * *1 MB*: 1000000, 1mi, 1000ki
573 * *1 TiB*: 1099511627776, 1t, 1024g, 1048576m
574 * *1 TB*: 1000000000, 1ti, 1000mi, 1000000ki
576 To specify times (units are not case sensitive):
580 * *M* -- means minutes
581 * *s* -- or sec means seconds (default)
582 * *ms* -- or *msec* means milliseconds
583 * *us* -- or *usec* means microseconds
585 If the option accepts an upper and lower range, use a colon ':' or
586 minus '-' to separate such values. See :ref:`irange <irange>`.
587 If the lower value specified happens to be larger than the upper value
588 the two values are swapped.
593 Boolean. Usually parsed as an integer, however only defined for
594 true and false (1 and 0).
599 Integer range with suffix. Allows value range to be given, such as
600 1024-4096. A colon may also be used as the separator, e.g. 1k:4k. If the
601 option allows two sets of ranges, they can be specified with a ',' or '/'
602 delimiter: 1k-4k/8k-32k. Also see :ref:`int <int>`.
605 A list of floating point numbers, separated by a ':' character.
607 With the above in mind, here follows the complete list of fio job parameters.
613 .. option:: kb_base=int
615 Select the interpretation of unit prefixes in input parameters.
618 Inputs comply with IEC 80000-13 and the International
619 System of Units (SI). Use:
621 - power-of-2 values with IEC prefixes (e.g., KiB)
622 - power-of-10 values with SI prefixes (e.g., kB)
625 Compatibility mode (default). To avoid breaking old scripts:
627 - power-of-2 values with SI prefixes
628 - power-of-10 values with IEC prefixes
630 See :option:`bs` for more details on input parameters.
632 Outputs always use correct prefixes. Most outputs include both
635 bw=2383.3kB/s (2327.4KiB/s)
637 If only one value is reported, then kb_base selects the one to use:
639 **1000** -- SI prefixes
641 **1024** -- IEC prefixes
643 .. option:: unit_base=int
645 Base unit for reporting. Allowed values are:
648 Use auto-detection (default).
660 ASCII name of the job. This may be used to override the name printed by fio
661 for this job. Otherwise the job name is used. On the command line this
662 parameter has the special purpose of also signaling the start of a new job.
664 .. option:: description=str
666 Text description of the job. Doesn't do anything except dump this text
667 description when this job is run. It's not parsed.
669 .. option:: loops=int
671 Run the specified number of iterations of this job. Used to repeat the same
672 workload a given number of times. Defaults to 1.
674 .. option:: numjobs=int
676 Create the specified number of clones of this job. Each clone of job
677 is spawned as an independent thread or process. May be used to setup a
678 larger number of threads/processes doing the same thing. Each thread is
679 reported separately; to see statistics for all clones as a whole, use
680 :option:`group_reporting` in conjunction with :option:`new_group`.
681 See :option:`--max-jobs`. Default: 1.
684 Time related parameters
685 ~~~~~~~~~~~~~~~~~~~~~~~
687 .. option:: runtime=time
689 Tell fio to terminate processing after the specified period of time. It
690 can be quite hard to determine for how long a specified job will run, so
691 this parameter is handy to cap the total runtime to a given time. When
692 the unit is omitted, the value is interpreted in seconds.
694 .. option:: time_based
696 If set, fio will run for the duration of the :option:`runtime` specified
697 even if the file(s) are completely read or written. It will simply loop over
698 the same workload as many times as the :option:`runtime` allows.
700 .. option:: startdelay=irange(time)
702 Delay the start of job for the specified amount of time. Can be a single
703 value or a range. When given as a range, each thread will choose a value
704 randomly from within the range. Value is in seconds if a unit is omitted.
706 .. option:: ramp_time=time
708 If set, fio will run the specified workload for this amount of time before
709 logging any performance numbers. Useful for letting performance settle
710 before logging results, thus minimizing the runtime required for stable
711 results. Note that the ``ramp_time`` is considered lead in time for a job,
712 thus it will increase the total runtime if a special timeout or
713 :option:`runtime` is specified. When the unit is omitted, the value is
716 .. option:: clocksource=str
718 Use the given clocksource as the base of timing. The supported options are:
721 :manpage:`gettimeofday(2)`
724 :manpage:`clock_gettime(2)`
727 Internal CPU clock source
729 cpu is the preferred clocksource if it is reliable, as it is very fast (and
730 fio is heavy on time calls). Fio will automatically use this clocksource if
731 it's supported and considered reliable on the system it is running on,
732 unless another clocksource is specifically set. For x86/x86-64 CPUs, this
733 means supporting TSC Invariant.
735 .. option:: gtod_reduce=bool
737 Enable all of the :manpage:`gettimeofday(2)` reducing options
738 (:option:`disable_clat`, :option:`disable_slat`, :option:`disable_bw_measurement`) plus
739 reduce precision of the timeout somewhat to really shrink the
740 :manpage:`gettimeofday(2)` call count. With this option enabled, we only do
741 about 0.4% of the :manpage:`gettimeofday(2)` calls we would have done if all
742 time keeping was enabled.
744 .. option:: gtod_cpu=int
746 Sometimes it's cheaper to dedicate a single thread of execution to just
747 getting the current time. Fio (and databases, for instance) are very
748 intensive on :manpage:`gettimeofday(2)` calls. With this option, you can set
749 one CPU aside for doing nothing but logging current time to a shared memory
750 location. Then the other threads/processes that run I/O workloads need only
751 copy that segment, instead of entering the kernel with a
752 :manpage:`gettimeofday(2)` call. The CPU set aside for doing these time
753 calls will be excluded from other uses. Fio will manually clear it from the
754 CPU mask of other jobs.
760 .. option:: directory=str
762 Prefix filenames with this directory. Used to place files in a different
763 location than :file:`./`. You can specify a number of directories by
764 separating the names with a ':' character. These directories will be
765 assigned equally distributed to job clones created by :option:`numjobs` as
766 long as they are using generated filenames. If specific `filename(s)` are
767 set fio will use the first listed directory, and thereby matching the
768 `filename` semantic (which generates a file for each clone if not
769 specified, but lets all clones use the same file if set).
771 See the :option:`filename` option for information on how to escape "``:``"
772 characters within the directory path itself.
774 Note: To control the directory fio will use for internal state files
775 use :option:`--aux-path`.
777 .. option:: filename=str
779 Fio normally makes up a `filename` based on the job name, thread number, and
780 file number (see :option:`filename_format`). If you want to share files
781 between threads in a job or several
782 jobs with fixed file paths, specify a `filename` for each of them to override
783 the default. If the ioengine is file based, you can specify a number of files
784 by separating the names with a ':' colon. So if you wanted a job to open
785 :file:`/dev/sda` and :file:`/dev/sdb` as the two working files, you would use
786 ``filename=/dev/sda:/dev/sdb``. This also means that whenever this option is
787 specified, :option:`nrfiles` is ignored. The size of regular files specified
788 by this option will be :option:`size` divided by number of files unless an
789 explicit size is specified by :option:`filesize`.
791 Each colon in the wanted path must be escaped with a ``\``
792 character. For instance, if the path is :file:`/dev/dsk/foo@3,0:c` then you
793 would use ``filename=/dev/dsk/foo@3,0\:c`` and if the path is
794 :file:`F:\\filename` then you would use ``filename=F\:\filename``.
796 On Windows, disk devices are accessed as :file:`\\\\.\\PhysicalDrive0` for
797 the first device, :file:`\\\\.\\PhysicalDrive1` for the second etc.
798 Note: Windows and FreeBSD prevent write access to areas
799 of the disk containing in-use data (e.g. filesystems).
801 The filename "`-`" is a reserved name, meaning *stdin* or *stdout*. Which
802 of the two depends on the read/write direction set.
804 .. option:: filename_format=str
806 If sharing multiple files between jobs, it is usually necessary to have fio
807 generate the exact names that you want. By default, fio will name a file
808 based on the default file format specification of
809 :file:`jobname.jobnumber.filenumber`. With this option, that can be
810 customized. Fio will recognize and replace the following keywords in this
814 The name of the worker thread or process.
816 IP of the fio process when using client/server mode.
818 The incremental number of the worker thread or process.
820 The incremental number of the file for that worker thread or
823 To have dependent jobs share a set of files, this option can be set to have
824 fio generate filenames that are shared between the two. For instance, if
825 :file:`testfiles.$filenum` is specified, file number 4 for any job will be
826 named :file:`testfiles.4`. The default of :file:`$jobname.$jobnum.$filenum`
827 will be used if no other format specifier is given.
829 If you specify a path then the directories will be created up to the
830 main directory for the file. So for example if you specify
831 ``filename_format=a/b/c/$jobnum`` then the directories a/b/c will be
832 created before the file setup part of the job. If you specify
833 :option:`directory` then the path will be relative that directory,
834 otherwise it is treated as the absolute path.
836 .. option:: unique_filename=bool
838 To avoid collisions between networked clients, fio defaults to prefixing any
839 generated filenames (with a directory specified) with the source of the
840 client connecting. To disable this behavior, set this option to 0.
842 .. option:: opendir=str
844 Recursively open any files below directory `str`.
846 .. option:: lockfile=str
848 Fio defaults to not locking any files before it does I/O to them. If a file
849 or file descriptor is shared, fio can serialize I/O to that file to make the
850 end result consistent. This is usual for emulating real workloads that share
851 files. The lock modes are:
854 No locking. The default.
856 Only one thread or process may do I/O at a time, excluding all
859 Read-write locking on the file. Many readers may
860 access the file at the same time, but writes get exclusive access.
862 .. option:: nrfiles=int
864 Number of files to use for this job. Defaults to 1. The size of files
865 will be :option:`size` divided by this unless explicit size is specified by
866 :option:`filesize`. Files are created for each thread separately, and each
867 file will have a file number within its name by default, as explained in
868 :option:`filename` section.
871 .. option:: openfiles=int
873 Number of files to keep open at the same time. Defaults to the same as
874 :option:`nrfiles`, can be set smaller to limit the number simultaneous
877 .. option:: file_service_type=str
879 Defines how fio decides which file from a job to service next. The following
883 Choose a file at random.
886 Round robin over opened files. This is the default.
889 Finish one file before moving on to the next. Multiple files can
890 still be open depending on :option:`openfiles`.
893 Use a *Zipf* distribution to decide what file to access.
896 Use a *Pareto* distribution to decide what file to access.
899 Use a *Gaussian* (normal) distribution to decide what file to
905 For *random*, *roundrobin*, and *sequential*, a postfix can be appended to
906 tell fio how many I/Os to issue before switching to a new file. For example,
907 specifying ``file_service_type=random:8`` would cause fio to issue
908 8 I/Os before selecting a new file at random. For the non-uniform
909 distributions, a floating point postfix can be given to influence how the
910 distribution is skewed. See :option:`random_distribution` for a description
911 of how that would work.
913 .. option:: ioscheduler=str
915 Attempt to switch the device hosting the file to the specified I/O scheduler
918 .. option:: create_serialize=bool
920 If true, serialize the file creation for the jobs. This may be handy to
921 avoid interleaving of data files, which may greatly depend on the filesystem
922 used and even the number of processors in the system. Default: true.
924 .. option:: create_fsync=bool
926 :manpage:`fsync(2)` the data file after creation. This is the default.
928 .. option:: create_on_open=bool
930 If true, don't pre-create files but allow the job's open() to create a file
931 when it's time to do I/O. Default: false -- pre-create all necessary files
934 .. option:: create_only=bool
936 If true, fio will only run the setup phase of the job. If files need to be
937 laid out or updated on disk, only that will be done -- the actual job contents
938 are not executed. Default: false.
940 .. option:: allow_file_create=bool
942 If true, fio is permitted to create files as part of its workload. If this
943 option is false, then fio will error out if
944 the files it needs to use don't already exist. Default: true.
946 .. option:: allow_mounted_write=bool
948 If this isn't set, fio will abort jobs that are destructive (e.g. that write)
949 to what appears to be a mounted device or partition. This should help catch
950 creating inadvertently destructive tests, not realizing that the test will
951 destroy data on the mounted file system. Note that some platforms don't allow
952 writing against a mounted device regardless of this option. Default: false.
954 .. option:: pre_read=bool
956 If this is given, files will be pre-read into memory before starting the
957 given I/O operation. This will also clear the :option:`invalidate` flag,
958 since it is pointless to pre-read and then drop the cache. This will only
959 work for I/O engines that are seek-able, since they allow you to read the
960 same data multiple times. Thus it will not work on non-seekable I/O engines
961 (e.g. network, splice). Default: false.
963 .. option:: unlink=bool
965 Unlink the job files when done. Not the default, as repeated runs of that
966 job would then waste time recreating the file set again and again. Default:
969 .. option:: unlink_each_loop=bool
971 Unlink job files after each iteration or loop. Default: false.
973 .. option:: zonemode=str
978 The :option:`zonerange`, :option:`zonesize`,
979 :option `zonecapacity` and option:`zoneskip`
980 parameters are ignored.
982 I/O happens in a single zone until
983 :option:`zonesize` bytes have been transferred.
984 After that number of bytes has been
985 transferred processing of the next zone
986 starts. :option `zonecapacity` is ignored.
988 Zoned block device mode. I/O happens
989 sequentially in each zone, even if random I/O
990 has been selected. Random I/O happens across
991 all zones instead of being restricted to a
992 single zone. The :option:`zoneskip` parameter
993 is ignored. :option:`zonerange` and
994 :option:`zonesize` must be identical.
995 Trim is handled using a zone reset operation.
996 Trim only considers non-empty sequential write
997 required and sequential write preferred zones.
999 .. option:: zonerange=int
1001 Size of a single zone. See also :option:`zonesize` and
1004 .. option:: zonesize=int
1006 For :option:`zonemode` =strided, this is the number of bytes to
1007 transfer before skipping :option:`zoneskip` bytes. If this parameter
1008 is smaller than :option:`zonerange` then only a fraction of each zone
1009 with :option:`zonerange` bytes will be accessed. If this parameter is
1010 larger than :option:`zonerange` then each zone will be accessed
1011 multiple times before skipping to the next zone.
1013 For :option:`zonemode` =zbd, this is the size of a single zone. The
1014 :option:`zonerange` parameter is ignored in this mode.
1017 .. option:: zonecapacity=int
1019 For :option:`zonemode` =zbd, this defines the capacity of a single zone,
1020 which is the accessible area starting from the zone start address.
1021 This parameter only applies when using :option:`zonemode` =zbd in
1022 combination with regular block devices. If not specified it defaults to
1023 the zone size. If the target device is a zoned block device, the zone
1024 capacity is obtained from the device information and this option is
1027 .. option:: zoneskip=int
1029 For :option:`zonemode` =strided, the number of bytes to skip after
1030 :option:`zonesize` bytes of data have been transferred. This parameter
1031 must be zero for :option:`zonemode` =zbd.
1033 .. option:: read_beyond_wp=bool
1035 This parameter applies to :option:`zonemode` =zbd only.
1037 Zoned block devices are block devices that consist of multiple zones.
1038 Each zone has a type, e.g. conventional or sequential. A conventional
1039 zone can be written at any offset that is a multiple of the block
1040 size. Sequential zones must be written sequentially. The position at
1041 which a write must occur is called the write pointer. A zoned block
1042 device can be either drive managed, host managed or host aware. For
1043 host managed devices the host must ensure that writes happen
1044 sequentially. Fio recognizes host managed devices and serializes
1045 writes to sequential zones for these devices.
1047 If a read occurs in a sequential zone beyond the write pointer then
1048 the zoned block device will complete the read without reading any data
1049 from the storage medium. Since such reads lead to unrealistically high
1050 bandwidth and IOPS numbers fio only reads beyond the write pointer if
1051 explicitly told to do so. Default: false.
1053 .. option:: max_open_zones=int
1055 When running a random write test across an entire drive many more
1056 zones will be open than in a typical application workload. Hence this
1057 command line option that allows to limit the number of open zones. The
1058 number of open zones is defined as the number of zones to which write
1059 commands are issued.
1061 .. option:: job_max_open_zones=int
1063 Limit on the number of simultaneously opened zones per single
1066 .. option:: ignore_zone_limits=bool
1068 If this option is used, fio will ignore the maximum number of open
1069 zones limit of the zoned block device in use, thus allowing the
1070 option :option:`max_open_zones` value to be larger than the device
1071 reported limit. Default: false.
1073 .. option:: zone_reset_threshold=float
1075 A number between zero and one that indicates the ratio of logical
1076 blocks with data to the total number of logical blocks in the test
1077 above which zones should be reset periodically.
1079 .. option:: zone_reset_frequency=float
1081 A number between zero and one that indicates how often a zone reset
1082 should be issued if the zone reset threshold has been exceeded. A zone
1083 reset is submitted after each (1 / zone_reset_frequency) write
1084 requests. This and the previous parameter can be used to simulate
1085 garbage collection activity.
1091 .. option:: direct=bool
1093 If value is true, use non-buffered I/O. This is usually O_DIRECT. Note that
1094 OpenBSD and ZFS on Solaris don't support direct I/O. On Windows the synchronous
1095 ioengines don't support direct I/O. Default: false.
1097 .. option:: atomic=bool
1099 If value is true, attempt to use atomic direct I/O. Atomic writes are
1100 guaranteed to be stable once acknowledged by the operating system. Only
1101 Linux supports O_ATOMIC right now.
1103 .. option:: buffered=bool
1105 If value is true, use buffered I/O. This is the opposite of the
1106 :option:`direct` option. Defaults to true.
1108 .. option:: readwrite=str, rw=str
1110 Type of I/O pattern. Accepted values are:
1117 Sequential trims (Linux block devices and SCSI
1118 character devices only).
1124 Random trims (Linux block devices and SCSI
1125 character devices only).
1127 Sequential mixed reads and writes.
1129 Random mixed reads and writes.
1131 Sequential trim+write sequences. Blocks will be trimmed first,
1132 then the same blocks will be written to. So if ``io_size=64K``
1133 is specified, Fio will trim a total of 64K bytes and also
1134 write 64K bytes on the same trimmed blocks. This behaviour
1135 will be consistent with ``number_ios`` or other Fio options
1136 limiting the total bytes or number of I/O's.
1138 Fio defaults to read if the option is not specified. For the mixed I/O
1139 types, the default is to split them 50/50. For certain types of I/O the
1140 result may still be skewed a bit, since the speed may be different.
1142 It is possible to specify the number of I/Os to do before getting a new
1143 offset by appending ``:<nr>`` to the end of the string given. For a
1144 random read, it would look like ``rw=randread:8`` for passing in an offset
1145 modifier with a value of 8. If the suffix is used with a sequential I/O
1146 pattern, then the *<nr>* value specified will be **added** to the generated
1147 offset for each I/O turning sequential I/O into sequential I/O with holes.
1148 For instance, using ``rw=write:4k`` will skip 4k for every write. Also see
1149 the :option:`rw_sequencer` option.
1151 .. option:: rw_sequencer=str
1153 If an offset modifier is given by appending a number to the ``rw=<str>``
1154 line, then this option controls how that number modifies the I/O offset
1155 being generated. Accepted values are:
1158 Generate sequential offset.
1160 Generate the same offset.
1162 ``sequential`` is only useful for random I/O, where fio would normally
1163 generate a new random offset for every I/O. If you append e.g. 8 to randread,
1164 you would get a new random offset for every 8 I/Os. The result would be a
1165 seek for only every 8 I/Os, instead of for every I/O. Use ``rw=randread:8``
1166 to specify that. As sequential I/O is already sequential, setting
1167 ``sequential`` for that would not result in any differences. ``identical``
1168 behaves in a similar fashion, except it sends the same offset 8 number of
1169 times before generating a new offset.
1171 .. option:: unified_rw_reporting=str
1173 Fio normally reports statistics on a per data direction basis, meaning that
1174 reads, writes, and trims are accounted and reported separately. This option
1175 determines whether fio reports the results normally, summed together, or as
1177 Accepted values are:
1180 Normal statistics reporting.
1183 Statistics are summed per data direction and reported together.
1186 Statistics are reported normally, followed by the mixed statistics.
1189 Backward-compatible alias for **none**.
1192 Backward-compatible alias for **mixed**.
1197 .. option:: randrepeat=bool
1199 Seed the random number generator used for random I/O patterns in a
1200 predictable way so the pattern is repeatable across runs. Default: true.
1202 .. option:: allrandrepeat=bool
1204 Seed all random number generators in a predictable way so results are
1205 repeatable across runs. Default: false.
1207 .. option:: randseed=int
1209 Seed the random number generators based on this seed value, to be able to
1210 control what sequence of output is being generated. If not set, the random
1211 sequence depends on the :option:`randrepeat` setting.
1213 .. option:: fallocate=str
1215 Whether pre-allocation is performed when laying down files.
1216 Accepted values are:
1219 Do not pre-allocate space.
1222 Use a platform's native pre-allocation call but fall back to
1223 **none** behavior if it fails/is not implemented.
1226 Pre-allocate via :manpage:`posix_fallocate(3)`.
1229 Pre-allocate via :manpage:`fallocate(2)` with
1230 FALLOC_FL_KEEP_SIZE set.
1233 Extend file to final size via :manpage:`ftruncate(2)`
1234 instead of allocating.
1237 Backward-compatible alias for **none**.
1240 Backward-compatible alias for **posix**.
1242 May not be available on all supported platforms. **keep** is only available
1243 on Linux. If using ZFS on Solaris this cannot be set to **posix**
1244 because ZFS doesn't support pre-allocation. Default: **native** if any
1245 pre-allocation methods except **truncate** are available, **none** if not.
1247 Note that using **truncate** on Windows will interact surprisingly
1248 with non-sequential write patterns. When writing to a file that has
1249 been extended by setting the end-of-file information, Windows will
1250 backfill the unwritten portion of the file up to that offset with
1251 zeroes before issuing the new write. This means that a single small
1252 write to the end of an extended file will stall until the entire
1253 file has been filled with zeroes.
1255 .. option:: fadvise_hint=str
1257 Use :manpage:`posix_fadvise(2)` or :manpage:`posix_fadvise(2)` to
1258 advise the kernel on what I/O patterns are likely to be issued.
1259 Accepted values are:
1262 Backwards-compatible hint for "no hint".
1265 Backwards compatible hint for "advise with fio workload type". This
1266 uses **FADV_RANDOM** for a random workload, and **FADV_SEQUENTIAL**
1267 for a sequential workload.
1270 Advise using **FADV_SEQUENTIAL**.
1273 Advise using **FADV_RANDOM**.
1275 .. option:: write_hint=str
1277 Use :manpage:`fcntl(2)` to advise the kernel what life time to expect
1278 from a write. Only supported on Linux, as of version 4.13. Accepted
1282 No particular life time associated with this file.
1285 Data written to this file has a short life time.
1288 Data written to this file has a medium life time.
1291 Data written to this file has a long life time.
1294 Data written to this file has a very long life time.
1296 The values are all relative to each other, and no absolute meaning
1297 should be associated with them.
1299 .. option:: offset=int
1301 Start I/O at the provided offset in the file, given as either a fixed size in
1302 bytes, zones or a percentage. If a percentage is given, the generated offset will be
1303 aligned to the minimum ``blocksize`` or to the value of ``offset_align`` if
1304 provided. Data before the given offset will not be touched. This
1305 effectively caps the file size at `real_size - offset`. Can be combined with
1306 :option:`size` to constrain the start and end range of the I/O workload.
1307 A percentage can be specified by a number between 1 and 100 followed by '%',
1308 for example, ``offset=20%`` to specify 20%. In ZBD mode, value can be set as
1309 number of zones using 'z'.
1311 .. option:: offset_align=int
1313 If set to non-zero value, the byte offset generated by a percentage ``offset``
1314 is aligned upwards to this value. Defaults to 0 meaning that a percentage
1315 offset is aligned to the minimum block size.
1317 .. option:: offset_increment=int
1319 If this is provided, then the real offset becomes `offset + offset_increment
1320 * thread_number`, where the thread number is a counter that starts at 0 and
1321 is incremented for each sub-job (i.e. when :option:`numjobs` option is
1322 specified). This option is useful if there are several jobs which are
1323 intended to operate on a file in parallel disjoint segments, with even
1324 spacing between the starting points. Percentages can be used for this option.
1325 If a percentage is given, the generated offset will be aligned to the minimum
1326 ``blocksize`` or to the value of ``offset_align`` if provided. In ZBD mode, value can
1327 also be set as number of zones using 'z'.
1329 .. option:: number_ios=int
1331 Fio will normally perform I/Os until it has exhausted the size of the region
1332 set by :option:`size`, or if it exhaust the allocated time (or hits an error
1333 condition). With this setting, the range/size can be set independently of
1334 the number of I/Os to perform. When fio reaches this number, it will exit
1335 normally and report status. Note that this does not extend the amount of I/O
1336 that will be done, it will only stop fio if this condition is met before
1337 other end-of-job criteria.
1339 .. option:: fsync=int
1341 If writing to a file, issue an :manpage:`fsync(2)` (or its equivalent) of
1342 the dirty data for every number of blocks given. For example, if you give 32
1343 as a parameter, fio will sync the file after every 32 writes issued. If fio is
1344 using non-buffered I/O, we may not sync the file. The exception is the sg
1345 I/O engine, which synchronizes the disk cache anyway. Defaults to 0, which
1346 means fio does not periodically issue and wait for a sync to complete. Also
1347 see :option:`end_fsync` and :option:`fsync_on_close`.
1349 .. option:: fdatasync=int
1351 Like :option:`fsync` but uses :manpage:`fdatasync(2)` to only sync data and
1352 not metadata blocks. In Windows, DragonFlyBSD or OSX there is no
1353 :manpage:`fdatasync(2)` so this falls back to using :manpage:`fsync(2)`.
1354 Defaults to 0, which means fio does not periodically issue and wait for a
1355 data-only sync to complete.
1357 .. option:: write_barrier=int
1359 Make every `N-th` write a barrier write.
1361 .. option:: sync_file_range=str:int
1363 Use :manpage:`sync_file_range(2)` for every `int` number of write
1364 operations. Fio will track range of writes that have happened since the last
1365 :manpage:`sync_file_range(2)` call. `str` can currently be one or more of:
1368 SYNC_FILE_RANGE_WAIT_BEFORE
1370 SYNC_FILE_RANGE_WRITE
1372 SYNC_FILE_RANGE_WAIT_AFTER
1374 So if you do ``sync_file_range=wait_before,write:8``, fio would use
1375 ``SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE`` for every 8
1376 writes. Also see the :manpage:`sync_file_range(2)` man page. This option is
1379 .. option:: overwrite=bool
1381 If true, writes to a file will always overwrite existing data. If the file
1382 doesn't already exist, it will be created before the write phase begins. If
1383 the file exists and is large enough for the specified write phase, nothing
1384 will be done. Default: false.
1386 .. option:: end_fsync=bool
1388 If true, :manpage:`fsync(2)` file contents when a write stage has completed.
1391 .. option:: fsync_on_close=bool
1393 If true, fio will :manpage:`fsync(2)` a dirty file on close. This differs
1394 from :option:`end_fsync` in that it will happen on every file close, not
1395 just at the end of the job. Default: false.
1397 .. option:: rwmixread=int
1399 Percentage of a mixed workload that should be reads. Default: 50.
1401 .. option:: rwmixwrite=int
1403 Percentage of a mixed workload that should be writes. If both
1404 :option:`rwmixread` and :option:`rwmixwrite` is given and the values do not
1405 add up to 100%, the latter of the two will be used to override the
1406 first. This may interfere with a given rate setting, if fio is asked to
1407 limit reads or writes to a certain rate. If that is the case, then the
1408 distribution may be skewed. Default: 50.
1410 .. option:: random_distribution=str:float[:float][,str:float][,str:float]
1412 By default, fio will use a completely uniform random distribution when asked
1413 to perform random I/O. Sometimes it is useful to skew the distribution in
1414 specific ways, ensuring that some parts of the data is more hot than others.
1415 fio includes the following distribution models:
1418 Uniform random distribution
1427 Normal (Gaussian) distribution
1430 Zoned random distribution
1433 Zone absolute random distribution
1435 When using a **zipf** or **pareto** distribution, an input value is also
1436 needed to define the access pattern. For **zipf**, this is the `Zipf
1437 theta`. For **pareto**, it's the `Pareto power`. Fio includes a test
1438 program, :command:`fio-genzipf`, that can be used visualize what the given input
1439 values will yield in terms of hit rates. If you wanted to use **zipf** with
1440 a `theta` of 1.2, you would use ``random_distribution=zipf:1.2`` as the
1441 option. If a non-uniform model is used, fio will disable use of the random
1442 map. For the **normal** distribution, a normal (Gaussian) deviation is
1443 supplied as a value between 0 and 100.
1445 The second, optional float is allowed for **pareto**, **zipf** and **normal** distributions.
1446 It allows to set base of distribution in non-default place, giving more control
1447 over most probable outcome. This value is in range [0-1] which maps linearly to
1448 range of possible random values.
1449 Defaults are: random for **pareto** and **zipf**, and 0.5 for **normal**.
1450 If you wanted to use **zipf** with a `theta` of 1.2 centered on 1/4 of allowed value range,
1451 you would use ``random_distribution=zipf:1.2:0.25``.
1453 For a **zoned** distribution, fio supports specifying percentages of I/O
1454 access that should fall within what range of the file or device. For
1455 example, given a criteria of:
1457 * 60% of accesses should be to the first 10%
1458 * 30% of accesses should be to the next 20%
1459 * 8% of accesses should be to the next 30%
1460 * 2% of accesses should be to the next 40%
1462 we can define that through zoning of the random accesses. For the above
1463 example, the user would do::
1465 random_distribution=zoned:60/10:30/20:8/30:2/40
1467 A **zoned_abs** distribution works exactly like the **zoned**, except
1468 that it takes absolute sizes. For example, let's say you wanted to
1469 define access according to the following criteria:
1471 * 60% of accesses should be to the first 20G
1472 * 30% of accesses should be to the next 100G
1473 * 10% of accesses should be to the next 500G
1475 we can define an absolute zoning distribution with:
1477 random_distribution=zoned_abs=60/20G:30/100G:10/500g
1479 For both **zoned** and **zoned_abs**, fio supports defining up to
1482 Similarly to how :option:`bssplit` works for setting ranges and
1483 percentages of block sizes. Like :option:`bssplit`, it's possible to
1484 specify separate zones for reads, writes, and trims. If just one set
1485 is given, it'll apply to all of them. This goes for both **zoned**
1486 **zoned_abs** distributions.
1488 .. option:: percentage_random=int[,int][,int]
1490 For a random workload, set how big a percentage should be random. This
1491 defaults to 100%, in which case the workload is fully random. It can be set
1492 from anywhere from 0 to 100. Setting it to 0 would make the workload fully
1493 sequential. Any setting in between will result in a random mix of sequential
1494 and random I/O, at the given percentages. Comma-separated values may be
1495 specified for reads, writes, and trims as described in :option:`blocksize`.
1497 .. option:: norandommap
1499 Normally fio will cover every block of the file when doing random I/O. If
1500 this option is given, fio will just get a new random offset without looking
1501 at past I/O history. This means that some blocks may not be read or written,
1502 and that some blocks may be read/written more than once. If this option is
1503 used with :option:`verify` and multiple blocksizes (via :option:`bsrange`),
1504 only intact blocks are verified, i.e., partially-overwritten blocks are
1505 ignored. With an async I/O engine and an I/O depth > 1, it is possible for
1506 the same block to be overwritten, which can cause verification errors. Either
1507 do not use norandommap in this case, or also use the lfsr random generator.
1509 .. option:: softrandommap=bool
1511 See :option:`norandommap`. If fio runs with the random block map enabled and
1512 it fails to allocate the map, if this option is set it will continue without
1513 a random block map. As coverage will not be as complete as with random maps,
1514 this option is disabled by default.
1516 .. option:: random_generator=str
1518 Fio supports the following engines for generating I/O offsets for random I/O:
1521 Strong 2^88 cycle random number generator.
1523 Linear feedback shift register generator.
1525 Strong 64-bit 2^258 cycle random number generator.
1527 **tausworthe** is a strong random number generator, but it requires tracking
1528 on the side if we want to ensure that blocks are only read or written
1529 once. **lfsr** guarantees that we never generate the same offset twice, and
1530 it's also less computationally expensive. It's not a true random generator,
1531 however, though for I/O purposes it's typically good enough. **lfsr** only
1532 works with single block sizes, not with workloads that use multiple block
1533 sizes. If used with such a workload, fio may read or write some blocks
1534 multiple times. The default value is **tausworthe**, unless the required
1535 space exceeds 2^32 blocks. If it does, then **tausworthe64** is
1536 selected automatically.
1542 .. option:: blocksize=int[,int][,int], bs=int[,int][,int]
1544 The block size in bytes used for I/O units. Default: 4096. A single value
1545 applies to reads, writes, and trims. Comma-separated values may be
1546 specified for reads, writes, and trims. A value not terminated in a comma
1547 applies to subsequent types.
1552 means 256k for reads, writes and trims.
1555 means 8k for reads, 32k for writes and trims.
1558 means 8k for reads, 32k for writes, and default for trims.
1561 means default for reads, 8k for writes and trims.
1564 means default for reads, 8k for writes, and default for trims.
1566 .. option:: blocksize_range=irange[,irange][,irange], bsrange=irange[,irange][,irange]
1568 A range of block sizes in bytes for I/O units. The issued I/O unit will
1569 always be a multiple of the minimum size, unless
1570 :option:`blocksize_unaligned` is set.
1572 Comma-separated ranges may be specified for reads, writes, and trims as
1573 described in :option:`blocksize`.
1575 Example: ``bsrange=1k-4k,2k-8k``.
1577 .. option:: bssplit=str[,str][,str]
1579 Sometimes you want even finer grained control of the block sizes
1580 issued, not just an even split between them. This option allows you to
1581 weight various block sizes, so that you are able to define a specific
1582 amount of block sizes issued. The format for this option is::
1584 bssplit=blocksize/percentage:blocksize/percentage
1586 for as many block sizes as needed. So if you want to define a workload
1587 that has 50% 64k blocks, 10% 4k blocks, and 40% 32k blocks, you would
1590 bssplit=4k/10:64k/50:32k/40
1592 Ordering does not matter. If the percentage is left blank, fio will
1593 fill in the remaining values evenly. So a bssplit option like this one::
1595 bssplit=4k/50:1k/:32k/
1597 would have 50% 4k ios, and 25% 1k and 32k ios. The percentages always
1598 add up to 100, if bssplit is given a range that adds up to more, it
1601 Comma-separated values may be specified for reads, writes, and trims as
1602 described in :option:`blocksize`.
1604 If you want a workload that has 50% 2k reads and 50% 4k reads, while
1605 having 90% 4k writes and 10% 8k writes, you would specify::
1607 bssplit=2k/50:4k/50,4k/90:8k/10
1609 Fio supports defining up to 64 different weights for each data
1612 .. option:: blocksize_unaligned, bs_unaligned
1614 If set, fio will issue I/O units with any size within
1615 :option:`blocksize_range`, not just multiples of the minimum size. This
1616 typically won't work with direct I/O, as that normally requires sector
1619 .. option:: bs_is_seq_rand=bool
1621 If this option is set, fio will use the normal read,write blocksize settings
1622 as sequential,random blocksize settings instead. Any random read or write
1623 will use the WRITE blocksize settings, and any sequential read or write will
1624 use the READ blocksize settings.
1626 .. option:: blockalign=int[,int][,int], ba=int[,int][,int]
1628 Boundary to which fio will align random I/O units. Default:
1629 :option:`blocksize`. Minimum alignment is typically 512b for using direct
1630 I/O, though it usually depends on the hardware block size. This option is
1631 mutually exclusive with using a random map for files, so it will turn off
1632 that option. Comma-separated values may be specified for reads, writes, and
1633 trims as described in :option:`blocksize`.
1639 .. option:: zero_buffers
1641 Initialize buffers with all zeros. Default: fill buffers with random data.
1643 .. option:: refill_buffers
1645 If this option is given, fio will refill the I/O buffers on every
1646 submit. Only makes sense if :option:`zero_buffers` isn't specified,
1647 naturally. Defaults to being unset i.e., the buffer is only filled at
1648 init time and the data in it is reused when possible but if any of
1649 :option:`verify`, :option:`buffer_compress_percentage` or
1650 :option:`dedupe_percentage` are enabled then `refill_buffers` is also
1651 automatically enabled.
1653 .. option:: scramble_buffers=bool
1655 If :option:`refill_buffers` is too costly and the target is using data
1656 deduplication, then setting this option will slightly modify the I/O buffer
1657 contents to defeat normal de-dupe attempts. This is not enough to defeat
1658 more clever block compression attempts, but it will stop naive dedupe of
1659 blocks. Default: true.
1661 .. option:: buffer_compress_percentage=int
1663 If this is set, then fio will attempt to provide I/O buffer content
1664 (on WRITEs) that compresses to the specified level. Fio does this by
1665 providing a mix of random data followed by fixed pattern data. The
1666 fixed pattern is either zeros, or the pattern specified by
1667 :option:`buffer_pattern`. If the `buffer_pattern` option is used, it
1668 might skew the compression ratio slightly. Setting
1669 `buffer_compress_percentage` to a value other than 100 will also
1670 enable :option:`refill_buffers` in order to reduce the likelihood that
1671 adjacent blocks are so similar that they over compress when seen
1672 together. See :option:`buffer_compress_chunk` for how to set a finer or
1673 coarser granularity for the random/fixed data region. Defaults to unset
1674 i.e., buffer data will not adhere to any compression level.
1676 .. option:: buffer_compress_chunk=int
1678 This setting allows fio to manage how big the random/fixed data region
1679 is when using :option:`buffer_compress_percentage`. When
1680 `buffer_compress_chunk` is set to some non-zero value smaller than the
1681 block size, fio can repeat the random/fixed region throughout the I/O
1682 buffer at the specified interval (which particularly useful when
1683 bigger block sizes are used for a job). When set to 0, fio will use a
1684 chunk size that matches the block size resulting in a single
1685 random/fixed region within the I/O buffer. Defaults to 512. When the
1686 unit is omitted, the value is interpreted in bytes.
1688 .. option:: buffer_pattern=str
1690 If set, fio will fill the I/O buffers with this pattern or with the contents
1691 of a file. If not set, the contents of I/O buffers are defined by the other
1692 options related to buffer contents. The setting can be any pattern of bytes,
1693 and can be prefixed with 0x for hex values. It may also be a string, where
1694 the string must then be wrapped with ``""``. Or it may also be a filename,
1695 where the filename must be wrapped with ``''`` in which case the file is
1696 opened and read. Note that not all the file contents will be read if that
1697 would cause the buffers to overflow. So, for example::
1699 buffer_pattern='filename'
1703 buffer_pattern="abcd"
1711 buffer_pattern=0xdeadface
1713 Also you can combine everything together in any order::
1715 buffer_pattern=0xdeadface"abcd"-12'filename'
1717 .. option:: dedupe_percentage=int
1719 If set, fio will generate this percentage of identical buffers when
1720 writing. These buffers will be naturally dedupable. The contents of the
1721 buffers depend on what other buffer compression settings have been set. It's
1722 possible to have the individual buffers either fully compressible, or not at
1723 all -- this option only controls the distribution of unique buffers. Setting
1724 this option will also enable :option:`refill_buffers` to prevent every buffer
1727 .. option:: dedupe_mode=str
1729 If ``dedupe_percentage=<int>`` is given, then this option controls how fio
1730 generates the dedupe buffers.
1733 Generate dedupe buffers by repeating previous writes
1735 Generate dedupe buffers from working set
1737 ``repeat`` is the default option for fio. Dedupe buffers are generated
1738 by repeating previous unique write.
1740 ``working_set`` is a more realistic workload.
1741 With ``working_set``, ``dedupe_working_set_percentage=<int>`` should be provided.
1742 Given that, fio will use the initial unique write buffers as its working set.
1743 Upon deciding to dedupe, fio will randomly choose a buffer from the working set.
1744 Note that by using ``working_set`` the dedupe percentage will converge
1745 to the desired over time while ``repeat`` maintains the desired percentage
1748 .. option:: dedupe_working_set_percentage=int
1750 If ``dedupe_mode=<str>`` is set to ``working_set``, then this controls
1751 the percentage of size of the file or device used as the buffers
1752 fio will choose to generate the dedupe buffers from
1754 Note that size needs to be explicitly provided and only 1 file per
1757 .. option:: dedupe_global=bool
1759 This controls whether the deduplication buffers will be shared amongst
1760 all jobs that have this option set. The buffers are spread evenly between
1763 .. option:: invalidate=bool
1765 Invalidate the buffer/page cache parts of the files to be used prior to
1766 starting I/O if the platform and file type support it. Defaults to true.
1767 This will be ignored if :option:`pre_read` is also specified for the
1770 .. option:: sync=str
1772 Whether, and what type, of synchronous I/O to use for writes. The allowed
1776 Do not use synchronous IO, the default.
1782 Use synchronous file IO. For the majority of I/O engines,
1783 this means using O_SYNC.
1789 Use synchronous data IO. For the majority of I/O engines,
1790 this means using O_DSYNC.
1793 .. option:: iomem=str, mem=str
1795 Fio can use various types of memory as the I/O unit buffer. The allowed
1799 Use memory from :manpage:`malloc(3)` as the buffers. Default memory
1803 Use shared memory as the buffers. Allocated through
1804 :manpage:`shmget(2)`.
1807 Same as shm, but use huge pages as backing.
1810 Use :manpage:`mmap(2)` to allocate buffers. May either be anonymous memory, or can
1811 be file backed if a filename is given after the option. The format
1812 is `mem=mmap:/path/to/file`.
1815 Use a memory mapped huge file as the buffer backing. Append filename
1816 after mmaphuge, ala `mem=mmaphuge:/hugetlbfs/file`.
1819 Same as mmap, but use a MMAP_SHARED mapping.
1822 Use GPU memory as the buffers for GPUDirect RDMA benchmark.
1823 The :option:`ioengine` must be `rdma`.
1825 The area allocated is a function of the maximum allowed bs size for the job,
1826 multiplied by the I/O depth given. Note that for **shmhuge** and
1827 **mmaphuge** to work, the system must have free huge pages allocated. This
1828 can normally be checked and set by reading/writing
1829 :file:`/proc/sys/vm/nr_hugepages` on a Linux system. Fio assumes a huge page
1830 is 2 or 4MiB in size depending on the platform. So to calculate the
1831 number of huge pages you need for a given job file, add up the I/O
1832 depth of all jobs (normally one unless :option:`iodepth` is used) and
1833 multiply by the maximum bs set. Then divide that number by the huge
1834 page size. You can see the size of the huge pages in
1835 :file:`/proc/meminfo`. If no huge pages are allocated by having a
1836 non-zero number in `nr_hugepages`, using **mmaphuge** or **shmhuge**
1837 will fail. Also see :option:`hugepage-size`.
1839 **mmaphuge** also needs to have hugetlbfs mounted and the file location
1840 should point there. So if it's mounted in :file:`/huge`, you would use
1841 `mem=mmaphuge:/huge/somefile`.
1843 .. option:: iomem_align=int, mem_align=int
1845 This indicates the memory alignment of the I/O memory buffers. Note that
1846 the given alignment is applied to the first I/O unit buffer, if using
1847 :option:`iodepth` the alignment of the following buffers are given by the
1848 :option:`bs` used. In other words, if using a :option:`bs` that is a
1849 multiple of the page sized in the system, all buffers will be aligned to
1850 this value. If using a :option:`bs` that is not page aligned, the alignment
1851 of subsequent I/O memory buffers is the sum of the :option:`iomem_align` and
1854 .. option:: hugepage-size=int
1856 Defines the size of a huge page. Must at least be equal to the system
1857 setting, see :file:`/proc/meminfo` and
1858 :file:`/sys/kernel/mm/hugepages/`. Defaults to 2 or 4MiB depending on
1859 the platform. Should probably always be a multiple of megabytes, so
1860 using ``hugepage-size=Xm`` is the preferred way to set this to avoid
1861 setting a non-pow-2 bad value.
1863 .. option:: lockmem=int
1865 Pin the specified amount of memory with :manpage:`mlock(2)`. Can be used to
1866 simulate a smaller amount of memory. The amount specified is per worker.
1872 .. option:: size=int
1874 The total size of file I/O for each thread of this job. Fio will run until
1875 this many bytes has been transferred, unless runtime is limited by other options
1876 (such as :option:`runtime`, for instance, or increased/decreased by :option:`io_size`).
1877 Fio will divide this size between the available files determined by options
1878 such as :option:`nrfiles`, :option:`filename`, unless :option:`filesize` is
1879 specified by the job. If the result of division happens to be 0, the size is
1880 set to the physical size of the given files or devices if they exist.
1881 If this option is not specified, fio will use the full size of the given
1882 files or devices. If the files do not exist, size must be given. It is also
1883 possible to give size as a percentage between 1 and 100. If ``size=20%`` is
1884 given, fio will use 20% of the full size of the given files or devices.
1885 In ZBD mode, value can also be set as number of zones using 'z'.
1886 Can be combined with :option:`offset` to constrain the start and end range
1887 that I/O will be done within.
1889 .. option:: io_size=int, io_limit=int
1891 Normally fio operates within the region set by :option:`size`, which means
1892 that the :option:`size` option sets both the region and size of I/O to be
1893 performed. Sometimes that is not what you want. With this option, it is
1894 possible to define just the amount of I/O that fio should do. For instance,
1895 if :option:`size` is set to 20GiB and :option:`io_size` is set to 5GiB, fio
1896 will perform I/O within the first 20GiB but exit when 5GiB have been
1897 done. The opposite is also possible -- if :option:`size` is set to 20GiB,
1898 and :option:`io_size` is set to 40GiB, then fio will do 40GiB of I/O within
1899 the 0..20GiB region.
1901 .. option:: filesize=irange(int)
1903 Individual file sizes. May be a range, in which case fio will select sizes for
1904 files at random within the given range. If not given, each created file is the
1905 same size. This option overrides :option:`size` in terms of file size, i.e. if
1906 :option:`filesize` is specified then :option:`size` becomes merely the default
1907 for :option:`io_size` and has no effect at all if :option:`io_size` is set
1910 .. option:: file_append=bool
1912 Perform I/O after the end of the file. Normally fio will operate within the
1913 size of a file. If this option is set, then fio will append to the file
1914 instead. This has identical behavior to setting :option:`offset` to the size
1915 of a file. This option is ignored on non-regular files.
1917 .. option:: fill_device=bool, fill_fs=bool
1919 Sets size to something really large and waits for ENOSPC (no space left on
1920 device) or EDQUOT (disk quota exceeded)
1921 as the terminating condition. Only makes sense with sequential
1922 write. For a read workload, the mount point will be filled first then I/O
1923 started on the result. This option doesn't make sense if operating on a raw
1924 device node, since the size of that is already known by the file system.
1925 Additionally, writing beyond end-of-device will not return ENOSPC there.
1931 .. option:: ioengine=str
1933 Defines how the job issues I/O to the file. The following types are defined:
1936 Basic :manpage:`read(2)` or :manpage:`write(2)`
1937 I/O. :manpage:`lseek(2)` is used to position the I/O location.
1938 See :option:`fsync` and :option:`fdatasync` for syncing write I/Os.
1941 Basic :manpage:`pread(2)` or :manpage:`pwrite(2)` I/O. Default on
1942 all supported operating systems except for Windows.
1945 Basic :manpage:`readv(2)` or :manpage:`writev(2)` I/O. Will emulate
1946 queuing by coalescing adjacent I/Os into a single submission.
1949 Basic :manpage:`preadv(2)` or :manpage:`pwritev(2)` I/O.
1952 Basic :manpage:`preadv2(2)` or :manpage:`pwritev2(2)` I/O.
1955 Fast Linux native asynchronous I/O. Supports async IO
1956 for both direct and buffered IO.
1957 This engine defines engine specific options.
1960 Fast Linux native asynchronous I/O for pass through commands.
1961 This engine defines engine specific options.
1964 Linux native asynchronous I/O. Note that Linux may only support
1965 queued behavior with non-buffered I/O (set ``direct=1`` or
1967 This engine defines engine specific options.
1970 POSIX asynchronous I/O using :manpage:`aio_read(3)` and
1971 :manpage:`aio_write(3)`.
1974 Solaris native asynchronous I/O.
1977 Windows native asynchronous I/O. Default on Windows.
1980 File is memory mapped with :manpage:`mmap(2)` and data copied
1981 to/from using :manpage:`memcpy(3)`.
1984 :manpage:`splice(2)` is used to transfer the data and
1985 :manpage:`vmsplice(2)` to transfer data from user space to the
1989 SCSI generic sg v3 I/O. May either be synchronous using the SG_IO
1990 ioctl, or if the target is an sg character device we use
1991 :manpage:`read(2)` and :manpage:`write(2)` for asynchronous
1992 I/O. Requires :option:`filename` option to specify either block or
1993 character devices. This engine supports trim operations.
1994 The sg engine includes engine specific options.
1997 Read, write, trim and ZBC/ZAC operations to a zoned
1998 block device using libzbc library. The target can be
1999 either an SG character device or a block device file.
2002 Doesn't transfer any data, just pretends to. This is mainly used to
2003 exercise fio itself and for debugging/testing purposes.
2006 Transfer over the network to given ``host:port``. Depending on the
2007 :option:`protocol` used, the :option:`hostname`, :option:`port`,
2008 :option:`listen` and :option:`filename` options are used to specify
2009 what sort of connection to make, while the :option:`protocol` option
2010 determines which protocol will be used. This engine defines engine
2014 Like **net**, but uses :manpage:`splice(2)` and
2015 :manpage:`vmsplice(2)` to map data and send/receive.
2016 This engine defines engine specific options.
2019 Doesn't transfer any data, but burns CPU cycles according to the
2020 :option:`cpuload`, :option:`cpuchunks` and :option:`cpumode` options.
2021 Setting :option:`cpuload`\=85 will cause that job to do nothing but burn 85%
2022 of the CPU. In case of SMP machines, use :option:`numjobs`\=<nr_of_cpu>
2023 to get desired CPU usage, as the cpuload only loads a
2024 single CPU at the desired rate. A job never finishes unless there is
2025 at least one non-cpuio job.
2026 Setting :option:`cpumode`\=qsort replace the default noop instructions loop
2027 by a qsort algorithm to consume more energy.
2030 The RDMA I/O engine supports both RDMA memory semantics
2031 (RDMA_WRITE/RDMA_READ) and channel semantics (Send/Recv) for the
2032 InfiniBand, RoCE and iWARP protocols. This engine defines engine
2036 I/O engine that does regular fallocate to simulate data transfer as
2040 does fallocate(,mode = FALLOC_FL_KEEP_SIZE,).
2043 does fallocate(,mode = 0).
2046 does fallocate(,mode = FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE).
2049 I/O engine that sends :manpage:`ftruncate(2)` operations in response
2050 to write (DDIR_WRITE) events. Each ftruncate issued sets the file's
2051 size to the current block offset. :option:`blocksize` is ignored.
2054 I/O engine that does regular EXT4_IOC_MOVE_EXT ioctls to simulate
2055 defragment activity in request to DDIR_WRITE event.
2058 I/O engine supporting direct access to Ceph Reliable Autonomic
2059 Distributed Object Store (RADOS) via librados. This ioengine
2060 defines engine specific options.
2063 I/O engine supporting direct access to Ceph Rados Block Devices
2064 (RBD) via librbd without the need to use the kernel rbd driver. This
2065 ioengine defines engine specific options.
2068 I/O engine supporting GET/PUT requests over HTTP(S) with libcurl to
2069 a WebDAV or S3 endpoint. This ioengine defines engine specific options.
2071 This engine only supports direct IO of iodepth=1; you need to scale this
2072 via numjobs. blocksize defines the size of the objects to be created.
2074 TRIM is translated to object deletion.
2077 Using GlusterFS libgfapi sync interface to direct access to
2078 GlusterFS volumes without having to go through FUSE. This ioengine
2079 defines engine specific options.
2082 Using GlusterFS libgfapi async interface to direct access to
2083 GlusterFS volumes without having to go through FUSE. This ioengine
2084 defines engine specific options.
2087 Read and write through Hadoop (HDFS). The :option:`filename` option
2088 is used to specify host,port of the hdfs name-node to connect. This
2089 engine interprets offsets a little differently. In HDFS, files once
2090 created cannot be modified so random writes are not possible. To
2091 imitate this the libhdfs engine expects a bunch of small files to be
2092 created over HDFS and will randomly pick a file from them
2093 based on the offset generated by fio backend (see the example
2094 job file to create such files, use ``rw=write`` option). Please
2095 note, it may be necessary to set environment variables to work
2096 with HDFS/libhdfs properly. Each job uses its own connection to
2100 Read, write and erase an MTD character device (e.g.,
2101 :file:`/dev/mtd0`). Discards are treated as erases. Depending on the
2102 underlying device type, the I/O may have to go in a certain pattern,
2103 e.g., on NAND, writing sequentially to erase blocks and discarding
2104 before overwriting. The `trimwrite` mode works well for this
2108 Read and write using filesystem DAX to a file on a filesystem
2109 mounted with DAX on a persistent memory device through the PMDK
2113 Read and write using device DAX to a persistent memory device (e.g.,
2114 /dev/dax0.0) through the PMDK libpmem library.
2117 Prefix to specify loading an external I/O engine object file. Append
2118 the engine filename, e.g. ``ioengine=external:/tmp/foo.o`` to load
2119 ioengine :file:`foo.o` in :file:`/tmp`. The path can be either
2120 absolute or relative. See :file:`engines/skeleton_external.c` for
2121 details of writing an external I/O engine.
2124 Simply create the files and do no I/O to them. You still need to
2125 set `filesize` so that all the accounting still occurs, but no
2126 actual I/O will be done other than creating the file.
2129 Simply do stat() and do no I/O to the file. You need to set 'filesize'
2130 and 'nrfiles', so that files will be created.
2131 This engine is to measure file lookup and meta data access.
2134 Simply delete the files by unlink() and do no I/O to them. You need to set 'filesize'
2135 and 'nrfiles', so that the files will be created.
2136 This engine is to measure file delete.
2139 Read and write using mmap I/O to a file on a filesystem
2140 mounted with DAX on a persistent memory device through the PMDK
2144 Synchronous read and write using DDN's Infinite Memory Engine (IME).
2145 This engine is very basic and issues calls to IME whenever an IO is
2149 Synchronous read and write using DDN's Infinite Memory Engine (IME).
2150 This engine uses iovecs and will try to stack as much IOs as possible
2151 (if the IOs are "contiguous" and the IO depth is not exceeded)
2152 before issuing a call to IME.
2155 Asynchronous read and write using DDN's Infinite Memory Engine (IME).
2156 This engine will try to stack as much IOs as possible by creating
2157 requests for IME. FIO will then decide when to commit these requests.
2160 Read and write iscsi lun with libiscsi.
2163 Read and write a Network Block Device (NBD).
2166 I/O engine supporting libcufile synchronous access to nvidia-fs and a
2167 GPUDirect Storage-supported filesystem. This engine performs
2168 I/O without transferring buffers between user-space and the kernel,
2169 unless :option:`verify` is set or :option:`cuda_io` is `posix`.
2170 :option:`iomem` must not be `cudamalloc`. This ioengine defines
2171 engine specific options.
2174 I/O engine supporting asynchronous read and write operations to the
2175 DAOS File System (DFS) via libdfs.
2178 I/O engine supporting asynchronous read and write operations to
2179 NFS filesystems from userspace via libnfs. This is useful for
2180 achieving higher concurrency and thus throughput than is possible
2184 Execute 3rd party tools. Could be used to perform monitoring during jobs runtime.
2187 I/O engine using the xNVMe C API, for NVMe devices. The xnvme engine provides
2188 flexibility to access GNU/Linux Kernel NVMe driver via libaio, IOCTLs, io_uring,
2189 the SPDK NVMe driver, or your own custom NVMe driver. The xnvme engine includes
2190 engine specific options. (See https://xnvme.io).
2192 I/O engine specific parameters
2193 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2195 In addition, there are some parameters which are only valid when a specific
2196 :option:`ioengine` is in use. These are used identically to normal parameters,
2197 with the caveat that when used on the command line, they must come after the
2198 :option:`ioengine` that defines them is selected.
2200 .. option:: cmdprio_percentage=int[,int] : [io_uring] [libaio]
2202 Set the percentage of I/O that will be issued with the highest priority.
2203 Default: 0. A single value applies to reads and writes. Comma-separated
2204 values may be specified for reads and writes. For this option to be
2205 effective, NCQ priority must be supported and enabled, and the :option:`direct`
2206 option must be set. fio must also be run as the root user. Unlike
2207 slat/clat/lat stats, which can be tracked and reported independently, per
2208 priority stats only track and report a single type of latency. By default,
2209 completion latency (clat) will be reported, if :option:`lat_percentiles` is
2210 set, total latency (lat) will be reported.
2212 .. option:: cmdprio_class=int[,int] : [io_uring] [libaio]
2214 Set the I/O priority class to use for I/Os that must be issued with
2215 a priority when :option:`cmdprio_percentage` or
2216 :option:`cmdprio_bssplit` is set. If not specified when
2217 :option:`cmdprio_percentage` or :option:`cmdprio_bssplit` is set,
2218 this defaults to the highest priority class. A single value applies
2219 to reads and writes. Comma-separated values may be specified for
2220 reads and writes. See :manpage:`ionice(1)`. See also the
2221 :option:`prioclass` option.
2223 .. option:: cmdprio=int[,int] : [io_uring] [libaio]
2225 Set the I/O priority value to use for I/Os that must be issued with
2226 a priority when :option:`cmdprio_percentage` or
2227 :option:`cmdprio_bssplit` is set. If not specified when
2228 :option:`cmdprio_percentage` or :option:`cmdprio_bssplit` is set,
2230 Linux limits us to a positive value between 0 and 7, with 0 being the
2231 highest. A single value applies to reads and writes. Comma-separated
2232 values may be specified for reads and writes. See :manpage:`ionice(1)`.
2233 Refer to an appropriate manpage for other operating systems since
2234 meaning of priority may differ. See also the :option:`prio` option.
2236 .. option:: cmdprio_bssplit=str[,str] : [io_uring] [libaio]
2238 To get a finer control over I/O priority, this option allows
2239 specifying the percentage of IOs that must have a priority set
2240 depending on the block size of the IO. This option is useful only
2241 when used together with the :option:`bssplit` option, that is,
2242 multiple different block sizes are used for reads and writes.
2244 The first accepted format for this option is the same as the format of
2245 the :option:`bssplit` option:
2247 cmdprio_bssplit=blocksize/percentage:blocksize/percentage
2249 In this case, each entry will use the priority class and priority
2250 level defined by the options :option:`cmdprio_class` and
2251 :option:`cmdprio` respectively.
2253 The second accepted format for this option is:
2255 cmdprio_bssplit=blocksize/percentage/class/level:blocksize/percentage/class/level
2257 In this case, the priority class and priority level is defined inside
2258 each entry. In comparison with the first accepted format, the second
2259 accepted format does not restrict all entries to have the same priority
2260 class and priority level.
2262 For both formats, only the read and write data directions are supported,
2263 values for trim IOs are ignored. This option is mutually exclusive with
2264 the :option:`cmdprio_percentage` option.
2266 .. option:: fixedbufs : [io_uring] [io_uring_cmd]
2268 If fio is asked to do direct IO, then Linux will map pages for each
2269 IO call, and release them when IO is done. If this option is set, the
2270 pages are pre-mapped before IO is started. This eliminates the need to
2271 map and release for each IO. This is more efficient, and reduces the
2274 .. option:: nonvectored : [io_uring] [io_uring_cmd]
2276 With this option, fio will use non-vectored read/write commands, where
2277 address must contain the address directly. Default is -1.
2279 .. option:: force_async=int : [io_uring] [io_uring_cmd]
2281 Normal operation for io_uring is to try and issue an sqe as
2282 non-blocking first, and if that fails, execute it in an async manner.
2283 With this option set to N, then every N request fio will ask sqe to
2284 be issued in an async manner. Default is 0.
2286 .. option:: registerfiles : [io_uring] [io_uring_cmd]
2288 With this option, fio registers the set of files being used with the
2289 kernel. This avoids the overhead of managing file counts in the kernel,
2290 making the submission and completion part more lightweight. Required
2291 for the below :option:`sqthread_poll` option.
2293 .. option:: sqthread_poll : [io_uring] [io_uring_cmd] [xnvme]
2295 Normally fio will submit IO by issuing a system call to notify the
2296 kernel of available items in the SQ ring. If this option is set, the
2297 act of submitting IO will be done by a polling thread in the kernel.
2298 This frees up cycles for fio, at the cost of using more CPU in the
2301 .. option:: sqthread_poll_cpu : [io_uring] [io_uring_cmd]
2303 When :option:`sqthread_poll` is set, this option provides a way to
2304 define which CPU should be used for the polling thread.
2306 .. option:: cmd_type=str : [io_uring_cmd]
2308 Specifies the type of uring passthrough command to be used. Supported
2309 value is nvme. Default is nvme.
2313 [io_uring] [io_uring_cmd] [xnvme]
2315 If this option is set, fio will attempt to use polled IO completions.
2316 Normal IO completions generate interrupts to signal the completion of
2317 IO, polled completions do not. Hence they are require active reaping
2318 by the application. The benefits are more efficient IO for high IOPS
2319 scenarios, and lower latencies for low queue depth IO.
2323 Set RWF_HIPRI on I/O, indicating to the kernel that it's of higher priority
2328 If this option is set, fio will attempt to use polled IO completions.
2329 This will have a similar effect as (io_uring)hipri. Only SCSI READ and
2330 WRITE commands will have the SGV4_FLAG_HIPRI set (not UNMAP (trim) nor
2331 VERIFY). Older versions of the Linux sg driver that do not support
2332 hipri will simply ignore this flag and do normal IO. The Linux SCSI
2333 Low Level Driver (LLD) that "owns" the device also needs to support
2334 hipri (also known as iopoll and mq_poll). The MegaRAID driver is an
2335 example of a SCSI LLD. Default: clear (0) which does normal
2336 (interrupted based) IO.
2338 .. option:: userspace_reap : [libaio]
2340 Normally, with the libaio engine in use, fio will use the
2341 :manpage:`io_getevents(2)` system call to reap newly returned events. With
2342 this flag turned on, the AIO ring will be read directly from user-space to
2343 reap events. The reaping mode is only enabled when polling for a minimum of
2344 0 events (e.g. when :option:`iodepth_batch_complete` `=0`).
2346 .. option:: hipri_percentage : [pvsync2]
2348 When hipri is set this determines the probability of a pvsync2 I/O being high
2349 priority. The default is 100%.
2351 .. option:: nowait : [pvsync2] [libaio] [io_uring]
2353 By default if a request cannot be executed immediately (e.g. resource starvation,
2354 waiting on locks) it is queued and the initiating process will be blocked until
2355 the required resource becomes free.
2357 This option sets the RWF_NOWAIT flag (supported from the 4.14 Linux kernel) and
2358 the call will return instantly with EAGAIN or a partial result rather than waiting.
2360 It is useful to also use ignore_error=EAGAIN when using this option.
2362 Note: glibc 2.27, 2.28 have a bug in syscall wrappers preadv2, pwritev2.
2363 They return EOPNOTSUP instead of EAGAIN.
2365 For cached I/O, using this option usually means a request operates only with
2366 cached data. Currently the RWF_NOWAIT flag does not supported for cached write.
2368 For direct I/O, requests will only succeed if cache invalidation isn't required,
2369 file blocks are fully allocated and the disk request could be issued immediately.
2371 .. option:: cpuload=int : [cpuio]
2373 Attempt to use the specified percentage of CPU cycles. This is a mandatory
2374 option when using cpuio I/O engine.
2376 .. option:: cpuchunks=int : [cpuio]
2378 Split the load into cycles of the given time. In microseconds.
2380 .. option:: cpumode=str : [cpuio]
2382 Specify how to stress the CPU. It can take these two values:
2385 This is the default where the CPU executes noop instructions.
2387 Replace the default noop instructions loop with a qsort algorithm to
2388 consume more energy.
2390 .. option:: exit_on_io_done=bool : [cpuio]
2392 Detect when I/O threads are done, then exit.
2394 .. option:: namenode=str : [libhdfs]
2396 The hostname or IP address of a HDFS cluster namenode to contact.
2398 .. option:: port=int
2402 The listening port of the HFDS cluster namenode.
2406 The TCP or UDP port to bind to or connect to. If this is used with
2407 :option:`numjobs` to spawn multiple instances of the same job type, then
2408 this will be the starting port number since fio will use a range of
2413 The port to use for RDMA-CM communication. This should be the same value
2414 on the client and the server side.
2416 .. option:: hostname=str : [netsplice] [net] [rdma]
2418 The hostname or IP address to use for TCP, UDP or RDMA-CM based I/O. If the job
2419 is a TCP listener or UDP reader, the hostname is not used and must be omitted
2420 unless it is a valid UDP multicast address.
2422 .. option:: serverip=str : [librpma_*]
2424 The IP address to be used for RDMA-CM based I/O.
2426 .. option:: direct_write_to_pmem=bool : [librpma_*]
2428 Set to 1 only when Direct Write to PMem from the remote host is possible.
2429 Otherwise, set to 0.
2431 .. option:: busy_wait_polling=bool : [librpma_*_server]
2433 Set to 0 to wait for completion instead of busy-wait polling completion.
2436 .. option:: interface=str : [netsplice] [net]
2438 The IP address of the network interface used to send or receive UDP
2441 .. option:: ttl=int : [netsplice] [net]
2443 Time-to-live value for outgoing UDP multicast packets. Default: 1.
2445 .. option:: nodelay=bool : [netsplice] [net]
2447 Set TCP_NODELAY on TCP connections.
2449 .. option:: protocol=str, proto=str : [netsplice] [net]
2451 The network protocol to use. Accepted values are:
2454 Transmission control protocol.
2456 Transmission control protocol V6.
2458 User datagram protocol.
2460 User datagram protocol V6.
2464 When the protocol is TCP or UDP, the port must also be given, as well as the
2465 hostname if the job is a TCP listener or UDP reader. For unix sockets, the
2466 normal :option:`filename` option should be used and the port is invalid.
2468 .. option:: listen : [netsplice] [net]
2470 For TCP network connections, tell fio to listen for incoming connections
2471 rather than initiating an outgoing connection. The :option:`hostname` must
2472 be omitted if this option is used.
2474 .. option:: pingpong : [netsplice] [net]
2476 Normally a network writer will just continue writing data, and a network
2477 reader will just consume packages. If ``pingpong=1`` is set, a writer will
2478 send its normal payload to the reader, then wait for the reader to send the
2479 same payload back. This allows fio to measure network latencies. The
2480 submission and completion latencies then measure local time spent sending or
2481 receiving, and the completion latency measures how long it took for the
2482 other end to receive and send back. For UDP multicast traffic
2483 ``pingpong=1`` should only be set for a single reader when multiple readers
2484 are listening to the same address.
2486 .. option:: window_size : [netsplice] [net]
2488 Set the desired socket buffer size for the connection.
2490 .. option:: mss : [netsplice] [net]
2492 Set the TCP maximum segment size (TCP_MAXSEG).
2494 .. option:: donorname=str : [e4defrag]
2496 File will be used as a block donor (swap extents between files).
2498 .. option:: inplace=int : [e4defrag]
2500 Configure donor file blocks allocation strategy:
2503 Default. Preallocate donor's file on init.
2505 Allocate space immediately inside defragment event, and free right
2508 .. option:: clustername=str : [rbd,rados]
2510 Specifies the name of the Ceph cluster.
2512 .. option:: rbdname=str : [rbd]
2514 Specifies the name of the RBD.
2516 .. option:: clientname=str : [rbd,rados]
2518 Specifies the username (without the 'client.' prefix) used to access the
2519 Ceph cluster. If the *clustername* is specified, the *clientname* shall be
2520 the full *type.id* string. If no type. prefix is given, fio will add
2521 'client.' by default.
2523 .. option:: conf=str : [rados]
2525 Specifies the configuration path of ceph cluster, so conf file does not
2526 have to be /etc/ceph/ceph.conf.
2528 .. option:: busy_poll=bool : [rbd,rados]
2530 Poll store instead of waiting for completion. Usually this provides better
2531 throughput at cost of higher(up to 100%) CPU utilization.
2533 .. option:: touch_objects=bool : [rados]
2535 During initialization, touch (create if do not exist) all objects (files).
2536 Touching all objects affects ceph caches and likely impacts test results.
2539 .. option:: pool=str :
2543 Specifies the name of the Ceph pool containing RBD or RADOS data.
2547 Specify the label or UUID of the DAOS pool to connect to.
2549 .. option:: cont=str : [dfs]
2551 Specify the label or UUID of the DAOS container to open.
2553 .. option:: chunk_size=int
2557 Specify a different chunk size (in bytes) for the dfs file.
2558 Use DAOS container's chunk size by default.
2562 The size of the chunk to use for each file.
2564 .. option:: object_class=str : [dfs]
2566 Specify a different object class for the dfs file.
2567 Use DAOS container's object class by default.
2569 .. option:: skip_bad=bool : [mtd]
2571 Skip operations against known bad blocks.
2573 .. option:: hdfsdirectory : [libhdfs]
2575 libhdfs will create chunk in this HDFS directory.
2577 .. option:: verb=str : [rdma]
2579 The RDMA verb to use on this side of the RDMA ioengine connection. Valid
2580 values are write, read, send and recv. These correspond to the equivalent
2581 RDMA verbs (e.g. write = rdma_write etc.). Note that this only needs to be
2582 specified on the client side of the connection. See the examples folder.
2584 .. option:: bindname=str : [rdma]
2586 The name to use to bind the local RDMA-CM connection to a local RDMA device.
2587 This could be a hostname or an IPv4 or IPv6 address. On the server side this
2588 will be passed into the rdma_bind_addr() function and on the client site it
2589 will be used in the rdma_resolve_add() function. This can be useful when
2590 multiple paths exist between the client and the server or in certain loopback
2593 .. option:: stat_type=str : [filestat]
2595 Specify stat system call type to measure lookup/getattr performance.
2596 Default is **stat** for :manpage:`stat(2)`.
2598 .. option:: readfua=bool : [sg]
2600 With readfua option set to 1, read operations include
2601 the force unit access (fua) flag. Default is 0.
2603 .. option:: writefua=bool : [sg]
2605 With writefua option set to 1, write operations include
2606 the force unit access (fua) flag. Default is 0.
2608 .. option:: sg_write_mode=str : [sg]
2610 Specify the type of write commands to issue. This option can take three values:
2613 This is the default where write opcodes are issued as usual.
2614 **write_and_verify**
2615 Issue WRITE AND VERIFY commands. The BYTCHK bit is set to 0. This
2616 directs the device to carry out a medium verification with no data
2617 comparison. The writefua option is ignored with this selection.
2619 This option is deprecated. Use write_and_verify instead.
2621 Issue WRITE SAME commands. This transfers a single block to the device
2622 and writes this same block of data to a contiguous sequence of LBAs
2623 beginning at the specified offset. fio's block size parameter specifies
2624 the amount of data written with each command. However, the amount of data
2625 actually transferred to the device is equal to the device's block
2626 (sector) size. For a device with 512 byte sectors, blocksize=8k will
2627 write 16 sectors with each command. fio will still generate 8k of data
2628 for each command but only the first 512 bytes will be used and
2629 transferred to the device. The writefua option is ignored with this
2632 This option is deprecated. Use write_same instead.
2634 Issue WRITE SAME(16) commands as above but with the No Data Output
2635 Buffer (NDOB) bit set. No data will be transferred to the device with
2636 this bit set. Data written will be a pre-determined pattern such as
2639 Issue WRITE STREAM(16) commands. Use the **stream_id** option to specify
2640 the stream identifier.
2641 **verify_bytchk_00**
2642 Issue VERIFY commands with BYTCHK set to 00. This directs the
2643 device to carry out a medium verification with no data comparison.
2644 **verify_bytchk_01**
2645 Issue VERIFY commands with BYTCHK set to 01. This directs the device to
2646 compare the data on the device with the data transferred to the device.
2647 **verify_bytchk_11**
2648 Issue VERIFY commands with BYTCHK set to 11. This transfers a
2649 single block to the device and compares the contents of this block with the
2650 data on the device beginning at the specified offset. fio's block size
2651 parameter specifies the total amount of data compared with this command.
2652 However, only one block (sector) worth of data is transferred to the device.
2653 This is similar to the WRITE SAME command except that data is compared instead
2656 .. option:: stream_id=int : [sg]
2658 Set the stream identifier for WRITE STREAM commands. If this is set to 0 (which is not
2659 a valid stream identifier) fio will open a stream and then close it when done. Default
2662 .. option:: http_host=str : [http]
2664 Hostname to connect to. For S3, this could be the bucket hostname.
2665 Default is **localhost**
2667 .. option:: http_user=str : [http]
2669 Username for HTTP authentication.
2671 .. option:: http_pass=str : [http]
2673 Password for HTTP authentication.
2675 .. option:: https=str : [http]
2677 Enable HTTPS instead of http. *on* enables HTTPS; *insecure*
2678 will enable HTTPS, but disable SSL peer verification (use with
2679 caution!). Default is **off**
2681 .. option:: http_mode=str : [http]
2683 Which HTTP access mode to use: *webdav*, *swift*, or *s3*.
2684 Default is **webdav**
2686 .. option:: http_s3_region=str : [http]
2688 The S3 region/zone string.
2689 Default is **us-east-1**
2691 .. option:: http_s3_key=str : [http]
2695 .. option:: http_s3_keyid=str : [http]
2697 The S3 key/access id.
2699 .. option:: http_s3_sse_customer_key=str : [http]
2701 The encryption customer key in SSE server side.
2703 .. option:: http_s3_sse_customer_algorithm=str : [http]
2705 The encryption customer algorithm in SSE server side.
2706 Default is **AES256**
2708 .. option:: http_s3_storage_class=str : [http]
2710 Which storage class to access. User-customizable settings.
2711 Default is **STANDARD**
2713 .. option:: http_swift_auth_token=str : [http]
2715 The Swift auth token. See the example configuration file on how
2718 .. option:: http_verbose=int : [http]
2720 Enable verbose requests from libcurl. Useful for debugging. 1
2721 turns on verbose logging from libcurl, 2 additionally enables
2722 HTTP IO tracing. Default is **0**
2724 .. option:: uri=str : [nbd]
2726 Specify the NBD URI of the server to test. The string
2727 is a standard NBD URI
2728 (see https://github.com/NetworkBlockDevice/nbd/tree/master/doc).
2729 Example URIs: nbd://localhost:10809
2730 nbd+unix:///?socket=/tmp/socket
2731 nbds://tlshost/exportname
2733 .. option:: gpu_dev_ids=str : [libcufile]
2735 Specify the GPU IDs to use with CUDA. This is a colon-separated list of
2736 int. GPUs are assigned to workers roundrobin. Default is 0.
2738 .. option:: cuda_io=str : [libcufile]
2740 Specify the type of I/O to use with CUDA. Default is **cufile**.
2743 Use libcufile and nvidia-fs. This option performs I/O directly
2744 between a GPUDirect Storage filesystem and GPU buffers,
2745 avoiding use of a bounce buffer. If :option:`verify` is set,
2746 cudaMemcpy is used to copy verificaton data between RAM and GPU.
2747 Verification data is copied from RAM to GPU before a write
2748 and from GPU to RAM after a read. :option:`direct` must be 1.
2750 Use POSIX to perform I/O with a RAM buffer, and use cudaMemcpy
2751 to transfer data between RAM and the GPUs. Data is copied from
2752 GPU to RAM before a write and copied from RAM to GPU after a
2753 read. :option:`verify` does not affect use of cudaMemcpy.
2755 .. option:: nfs_url=str : [nfs]
2757 URL in libnfs format, eg nfs://<server|ipv4|ipv6>/path[?arg=val[&arg=val]*]
2758 Refer to the libnfs README for more details.
2760 .. option:: program=str : [exec]
2762 Specify the program to execute.
2764 .. option:: arguments=str : [exec]
2766 Specify arguments to pass to program.
2767 Some special variables can be expanded to pass fio's job details to the program.
2770 Replaced by the duration of the job in seconds.
2772 Replaced by the name of the job.
2774 .. option:: grace_time=int : [exec]
2776 Specify the time between the SIGTERM and SIGKILL signals. Default is 1 second.
2778 .. option:: std_redirect=bool : [exec]
2780 If set, stdout and stderr streams are redirected to files named from the job name. Default is true.
2782 .. option:: xnvme_async=str : [xnvme]
2784 Select the xnvme async command interface. This can take these values.
2787 This is default and use to emulate asynchronous I/O by using a
2788 single thread to create a queue pair on top of a synchronous
2789 I/O interface using the NVMe driver IOCTL.
2791 Emulate an asynchronous I/O interface with a pool of userspace
2792 threads on top of a synchronous I/O interface using the NVMe
2793 driver IOCTL. By default four threads are used.
2795 Linux native asynchronous I/O interface which supports both
2796 direct and buffered I/O.
2798 Fast Linux native asynchronous I/O interface for NVMe pass
2799 through commands. This only works with NVMe character device
2802 Use Linux aio for Asynchronous I/O.
2804 Use the posix asynchronous I/O interface to perform one or
2805 more I/O operations asynchronously.
2807 Do not transfer any data; just pretend to. This is mainly used
2808 for introspective performance evaluation.
2810 .. option:: xnvme_sync=str : [xnvme]
2812 Select the xnvme synchronous command interface. This can take these values.
2815 This is default and uses Linux NVMe Driver ioctl() for
2818 This supports regular as well as vectored pread() and pwrite()
2821 This is the same as psync except that it also supports zone
2822 management commands using Linux block layer IOCTLs.
2824 .. option:: xnvme_admin=str : [xnvme]
2826 Select the xnvme admin command interface. This can take these values.
2829 This is default and uses linux NVMe Driver ioctl() for admin
2832 Use Linux Block Layer ioctl() and sysfs for admin commands.
2834 .. option:: xnvme_dev_nsid=int : [xnvme]
2836 xnvme namespace identifier for userspace NVMe driver, such as SPDK.
2838 .. option:: xnvme_iovec=int : [xnvme]
2840 If this option is set. xnvme will use vectored read/write commands.
2845 .. option:: iodepth=int
2847 Number of I/O units to keep in flight against the file. Note that
2848 increasing *iodepth* beyond 1 will not affect synchronous ioengines (except
2849 for small degrees when :option:`verify_async` is in use). Even async
2850 engines may impose OS restrictions causing the desired depth not to be
2851 achieved. This may happen on Linux when using libaio and not setting
2852 :option:`direct`\=1, since buffered I/O is not async on that OS. Keep an
2853 eye on the I/O depth distribution in the fio output to verify that the
2854 achieved depth is as expected. Default: 1.
2856 .. option:: iodepth_batch_submit=int, iodepth_batch=int
2858 This defines how many pieces of I/O to submit at once. It defaults to 1
2859 which means that we submit each I/O as soon as it is available, but can be
2860 raised to submit bigger batches of I/O at the time. If it is set to 0 the
2861 :option:`iodepth` value will be used.
2863 .. option:: iodepth_batch_complete_min=int, iodepth_batch_complete=int
2865 This defines how many pieces of I/O to retrieve at once. It defaults to 1
2866 which means that we'll ask for a minimum of 1 I/O in the retrieval process
2867 from the kernel. The I/O retrieval will go on until we hit the limit set by
2868 :option:`iodepth_low`. If this variable is set to 0, then fio will always
2869 check for completed events before queuing more I/O. This helps reduce I/O
2870 latency, at the cost of more retrieval system calls.
2872 .. option:: iodepth_batch_complete_max=int
2874 This defines maximum pieces of I/O to retrieve at once. This variable should
2875 be used along with :option:`iodepth_batch_complete_min`\=int variable,
2876 specifying the range of min and max amount of I/O which should be
2877 retrieved. By default it is equal to the :option:`iodepth_batch_complete_min`
2882 iodepth_batch_complete_min=1
2883 iodepth_batch_complete_max=<iodepth>
2885 which means that we will retrieve at least 1 I/O and up to the whole
2886 submitted queue depth. If none of I/O has been completed yet, we will wait.
2890 iodepth_batch_complete_min=0
2891 iodepth_batch_complete_max=<iodepth>
2893 which means that we can retrieve up to the whole submitted queue depth, but
2894 if none of I/O has been completed yet, we will NOT wait and immediately exit
2895 the system call. In this example we simply do polling.
2897 .. option:: iodepth_low=int
2899 The low water mark indicating when to start filling the queue
2900 again. Defaults to the same as :option:`iodepth`, meaning that fio will
2901 attempt to keep the queue full at all times. If :option:`iodepth` is set to
2902 e.g. 16 and *iodepth_low* is set to 4, then after fio has filled the queue of
2903 16 requests, it will let the depth drain down to 4 before starting to fill
2906 .. option:: serialize_overlap=bool
2908 Serialize in-flight I/Os that might otherwise cause or suffer from data races.
2909 When two or more I/Os are submitted simultaneously, there is no guarantee that
2910 the I/Os will be processed or completed in the submitted order. Further, if
2911 two or more of those I/Os are writes, any overlapping region between them can
2912 become indeterminate/undefined on certain storage. These issues can cause
2913 verification to fail erratically when at least one of the racing I/Os is
2914 changing data and the overlapping region has a non-zero size. Setting
2915 ``serialize_overlap`` tells fio to avoid provoking this behavior by explicitly
2916 serializing in-flight I/Os that have a non-zero overlap. Note that setting
2917 this option can reduce both performance and the :option:`iodepth` achieved.
2919 This option only applies to I/Os issued for a single job except when it is
2920 enabled along with :option:`io_submit_mode`\=offload. In offload mode, fio
2921 will check for overlap among all I/Os submitted by offload jobs with :option:`serialize_overlap`
2926 .. option:: io_submit_mode=str
2928 This option controls how fio submits the I/O to the I/O engine. The default
2929 is `inline`, which means that the fio job threads submit and reap I/O
2930 directly. If set to `offload`, the job threads will offload I/O submission
2931 to a dedicated pool of I/O threads. This requires some coordination and thus
2932 has a bit of extra overhead, especially for lower queue depth I/O where it
2933 can increase latencies. The benefit is that fio can manage submission rates
2934 independently of the device completion rates. This avoids skewed latency
2935 reporting if I/O gets backed up on the device side (the coordinated omission
2936 problem). Note that this option cannot reliably be used with async IO
2943 .. option:: thinktime=time
2945 Stall the job for the specified period of time after an I/O has completed before issuing the
2946 next. May be used to simulate processing being done by an application.
2947 When the unit is omitted, the value is interpreted in microseconds. See
2948 :option:`thinktime_blocks`, :option:`thinktime_iotime` and :option:`thinktime_spin`.
2950 .. option:: thinktime_spin=time
2952 Only valid if :option:`thinktime` is set - pretend to spend CPU time doing
2953 something with the data received, before falling back to sleeping for the
2954 rest of the period specified by :option:`thinktime`. When the unit is
2955 omitted, the value is interpreted in microseconds.
2957 .. option:: thinktime_blocks=int
2959 Only valid if :option:`thinktime` is set - control how many blocks to issue,
2960 before waiting :option:`thinktime` usecs. If not set, defaults to 1 which will make
2961 fio wait :option:`thinktime` usecs after every block. This effectively makes any
2962 queue depth setting redundant, since no more than 1 I/O will be queued
2963 before we have to complete it and do our :option:`thinktime`. In other words, this
2964 setting effectively caps the queue depth if the latter is larger.
2966 .. option:: thinktime_blocks_type=str
2968 Only valid if :option:`thinktime` is set - control how :option:`thinktime_blocks`
2969 triggers. The default is `complete`, which triggers thinktime when fio completes
2970 :option:`thinktime_blocks` blocks. If this is set to `issue`, then the trigger happens
2973 .. option:: thinktime_iotime=time
2975 Only valid if :option:`thinktime` is set - control :option:`thinktime`
2976 interval by time. The :option:`thinktime` stall is repeated after IOs
2977 are executed for :option:`thinktime_iotime`. For example,
2978 ``--thinktime_iotime=9s --thinktime=1s`` repeat 10-second cycle with IOs
2979 for 9 seconds and stall for 1 second. When the unit is omitted,
2980 :option:`thinktime_iotime` is interpreted as a number of seconds. If
2981 this option is used together with :option:`thinktime_blocks`, the
2982 :option:`thinktime` stall is repeated after :option:`thinktime_iotime`
2983 or after :option:`thinktime_blocks` IOs, whichever happens first.
2985 .. option:: rate=int[,int][,int]
2987 Cap the bandwidth used by this job. The number is in bytes/sec, the normal
2988 suffix rules apply. Comma-separated values may be specified for reads,
2989 writes, and trims as described in :option:`blocksize`.
2991 For example, using `rate=1m,500k` would limit reads to 1MiB/sec and writes to
2992 500KiB/sec. Capping only reads or writes can be done with `rate=,500k` or
2993 `rate=500k,` where the former will only limit writes (to 500KiB/sec) and the
2994 latter will only limit reads.
2996 .. option:: rate_min=int[,int][,int]
2998 Tell fio to do whatever it can to maintain at least this bandwidth. Failing
2999 to meet this requirement will cause the job to exit. Comma-separated values
3000 may be specified for reads, writes, and trims as described in
3001 :option:`blocksize`.
3003 .. option:: rate_iops=int[,int][,int]
3005 Cap the bandwidth to this number of IOPS. Basically the same as
3006 :option:`rate`, just specified independently of bandwidth. If the job is
3007 given a block size range instead of a fixed value, the smallest block size
3008 is used as the metric. Comma-separated values may be specified for reads,
3009 writes, and trims as described in :option:`blocksize`.
3011 .. option:: rate_iops_min=int[,int][,int]
3013 If fio doesn't meet this rate of I/O, it will cause the job to exit.
3014 Comma-separated values may be specified for reads, writes, and trims as
3015 described in :option:`blocksize`.
3017 .. option:: rate_process=str
3019 This option controls how fio manages rated I/O submissions. The default is
3020 `linear`, which submits I/O in a linear fashion with fixed delays between
3021 I/Os that gets adjusted based on I/O completion rates. If this is set to
3022 `poisson`, fio will submit I/O based on a more real world random request
3023 flow, known as the Poisson process
3024 (https://en.wikipedia.org/wiki/Poisson_point_process). The lambda will be
3025 10^6 / IOPS for the given workload.
3027 .. option:: rate_ignore_thinktime=bool
3029 By default, fio will attempt to catch up to the specified rate setting,
3030 if any kind of thinktime setting was used. If this option is set, then
3031 fio will ignore the thinktime and continue doing IO at the specified
3032 rate, instead of entering a catch-up mode after thinktime is done.
3038 .. option:: latency_target=time
3040 If set, fio will attempt to find the max performance point that the given
3041 workload will run at while maintaining a latency below this target. When
3042 the unit is omitted, the value is interpreted in microseconds. See
3043 :option:`latency_window` and :option:`latency_percentile`.
3045 .. option:: latency_window=time
3047 Used with :option:`latency_target` to specify the sample window that the job
3048 is run at varying queue depths to test the performance. When the unit is
3049 omitted, the value is interpreted in microseconds.
3051 .. option:: latency_percentile=float
3053 The percentage of I/Os that must fall within the criteria specified by
3054 :option:`latency_target` and :option:`latency_window`. If not set, this
3055 defaults to 100.0, meaning that all I/Os must be equal or below to the value
3056 set by :option:`latency_target`.
3058 .. option:: latency_run=bool
3060 Used with :option:`latency_target`. If false (default), fio will find
3061 the highest queue depth that meets :option:`latency_target` and exit. If
3062 true, fio will continue running and try to meet :option:`latency_target`
3063 by adjusting queue depth.
3065 .. option:: max_latency=time[,time][,time]
3067 If set, fio will exit the job with an ETIMEDOUT error if it exceeds this
3068 maximum latency. When the unit is omitted, the value is interpreted in
3069 microseconds. Comma-separated values may be specified for reads, writes,
3070 and trims as described in :option:`blocksize`.
3072 .. option:: rate_cycle=int
3074 Average bandwidth for :option:`rate` and :option:`rate_min` over this number
3075 of milliseconds. Defaults to 1000.
3081 .. option:: write_iolog=str
3083 Write the issued I/O patterns to the specified file. See
3084 :option:`read_iolog`. Specify a separate file for each job, otherwise the
3085 iologs will be interspersed and the file may be corrupt. This file will
3086 be opened in append mode.
3088 .. option:: read_iolog=str
3090 Open an iolog with the specified filename and replay the I/O patterns it
3091 contains. This can be used to store a workload and replay it sometime
3092 later. The iolog given may also be a blktrace binary file, which allows fio
3093 to replay a workload captured by :command:`blktrace`. See
3094 :manpage:`blktrace(8)` for how to capture such logging data. For blktrace
3095 replay, the file needs to be turned into a blkparse binary data file first
3096 (``blkparse <device> -o /dev/null -d file_for_fio.bin``).
3097 You can specify a number of files by separating the names with a ':'
3098 character. See the :option:`filename` option for information on how to
3099 escape ':' characters within the file names. These files will
3100 be sequentially assigned to job clones created by :option:`numjobs`.
3101 '-' is a reserved name, meaning read from stdin, notably if
3102 :option:`filename` is set to '-' which means stdin as well, then
3103 this flag can't be set to '-'.
3105 .. option:: read_iolog_chunked=bool
3107 Determines how iolog is read. If false(default) entire :option:`read_iolog`
3108 will be read at once. If selected true, input from iolog will be read
3109 gradually. Useful when iolog is very large, or it is generated.
3111 .. option:: merge_blktrace_file=str
3113 When specified, rather than replaying the logs passed to :option:`read_iolog`,
3114 the logs go through a merge phase which aggregates them into a single
3115 blktrace. The resulting file is then passed on as the :option:`read_iolog`
3116 parameter. The intention here is to make the order of events consistent.
3117 This limits the influence of the scheduler compared to replaying multiple
3118 blktraces via concurrent jobs.
3120 .. option:: merge_blktrace_scalars=float_list
3122 This is a percentage based option that is index paired with the list of
3123 files passed to :option:`read_iolog`. When merging is performed, scale
3124 the time of each event by the corresponding amount. For example,
3125 ``--merge_blktrace_scalars="50:100"`` runs the first trace in halftime
3126 and the second trace in realtime. This knob is separately tunable from
3127 :option:`replay_time_scale` which scales the trace during runtime and
3128 does not change the output of the merge unlike this option.
3130 .. option:: merge_blktrace_iters=float_list
3132 This is a whole number option that is index paired with the list of files
3133 passed to :option:`read_iolog`. When merging is performed, run each trace
3134 for the specified number of iterations. For example,
3135 ``--merge_blktrace_iters="2:1"`` runs the first trace for two iterations
3136 and the second trace for one iteration.
3138 .. option:: replay_no_stall=bool
3140 When replaying I/O with :option:`read_iolog` the default behavior is to
3141 attempt to respect the timestamps within the log and replay them with the
3142 appropriate delay between IOPS. By setting this variable fio will not
3143 respect the timestamps and attempt to replay them as fast as possible while
3144 still respecting ordering. The result is the same I/O pattern to a given
3145 device, but different timings.
3147 .. option:: replay_time_scale=int
3149 When replaying I/O with :option:`read_iolog`, fio will honor the
3150 original timing in the trace. With this option, it's possible to scale
3151 the time. It's a percentage option, if set to 50 it means run at 50%
3152 the original IO rate in the trace. If set to 200, run at twice the
3153 original IO rate. Defaults to 100.
3155 .. option:: replay_redirect=str
3157 While replaying I/O patterns using :option:`read_iolog` the default behavior
3158 is to replay the IOPS onto the major/minor device that each IOP was recorded
3159 from. This is sometimes undesirable because on a different machine those
3160 major/minor numbers can map to a different device. Changing hardware on the
3161 same system can also result in a different major/minor mapping.
3162 ``replay_redirect`` causes all I/Os to be replayed onto the single specified
3163 device regardless of the device it was recorded
3164 from. i.e. :option:`replay_redirect`\= :file:`/dev/sdc` would cause all I/O
3165 in the blktrace or iolog to be replayed onto :file:`/dev/sdc`. This means
3166 multiple devices will be replayed onto a single device, if the trace
3167 contains multiple devices. If you want multiple devices to be replayed
3168 concurrently to multiple redirected devices you must blkparse your trace
3169 into separate traces and replay them with independent fio invocations.
3170 Unfortunately this also breaks the strict time ordering between multiple
3173 .. option:: replay_align=int
3175 Force alignment of the byte offsets in a trace to this value. The value
3176 must be a power of 2.
3178 .. option:: replay_scale=int
3180 Scale byte offsets down by this factor when replaying traces. Should most
3181 likely use :option:`replay_align` as well.
3183 .. option:: replay_skip=str
3185 Sometimes it's useful to skip certain IO types in a replay trace.
3186 This could be, for instance, eliminating the writes in the trace.
3187 Or not replaying the trims/discards, if you are redirecting to
3188 a device that doesn't support them. This option takes a comma
3189 separated list of read, write, trim, sync.
3192 Threads, processes and job synchronization
3193 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3197 Fio defaults to creating jobs by using fork, however if this option is
3198 given, fio will create jobs by using POSIX Threads' function
3199 :manpage:`pthread_create(3)` to create threads instead.
3201 .. option:: wait_for=str
3203 If set, the current job won't be started until all workers of the specified
3204 waitee job are done.
3206 ``wait_for`` operates on the job name basis, so there are a few
3207 limitations. First, the waitee must be defined prior to the waiter job
3208 (meaning no forward references). Second, if a job is being referenced as a
3209 waitee, it must have a unique name (no duplicate waitees).
3211 .. option:: nice=int
3213 Run the job with the given nice value. See man :manpage:`nice(2)`.
3215 On Windows, values less than -15 set the process class to "High"; -1 through
3216 -15 set "Above Normal"; 1 through 15 "Below Normal"; and above 15 "Idle"
3219 .. option:: prio=int
3221 Set the I/O priority value of this job. Linux limits us to a positive value
3222 between 0 and 7, with 0 being the highest. See man
3223 :manpage:`ionice(1)`. Refer to an appropriate manpage for other operating
3224 systems since meaning of priority may differ. For per-command priority
3225 setting, see I/O engine specific :option:`cmdprio_percentage` and
3226 :option:`cmdprio` options.
3228 .. option:: prioclass=int
3230 Set the I/O priority class. See man :manpage:`ionice(1)`. For per-command
3231 priority setting, see I/O engine specific :option:`cmdprio_percentage`
3232 and :option:`cmdprio_class` options.
3234 .. option:: cpus_allowed=str
3236 Controls the same options as :option:`cpumask`, but accepts a textual
3237 specification of the permitted CPUs instead and CPUs are indexed from 0. So
3238 to use CPUs 0 and 5 you would specify ``cpus_allowed=0,5``. This option also
3239 allows a range of CPUs to be specified -- say you wanted a binding to CPUs
3240 0, 5, and 8 to 15, you would set ``cpus_allowed=0,5,8-15``.
3242 On Windows, when ``cpus_allowed`` is unset only CPUs from fio's current
3243 processor group will be used and affinity settings are inherited from the
3244 system. An fio build configured to target Windows 7 makes options that set
3245 CPUs processor group aware and values will set both the processor group
3246 and a CPU from within that group. For example, on a system where processor
3247 group 0 has 40 CPUs and processor group 1 has 32 CPUs, ``cpus_allowed``
3248 values between 0 and 39 will bind CPUs from processor group 0 and
3249 ``cpus_allowed`` values between 40 and 71 will bind CPUs from processor
3250 group 1. When using ``cpus_allowed_policy=shared`` all CPUs specified by a
3251 single ``cpus_allowed`` option must be from the same processor group. For
3252 Windows fio builds not built for Windows 7, CPUs will only be selected from
3253 (and be relative to) whatever processor group fio happens to be running in
3254 and CPUs from other processor groups cannot be used.
3256 .. option:: cpus_allowed_policy=str
3258 Set the policy of how fio distributes the CPUs specified by
3259 :option:`cpus_allowed` or :option:`cpumask`. Two policies are supported:
3262 All jobs will share the CPU set specified.
3264 Each job will get a unique CPU from the CPU set.
3266 **shared** is the default behavior, if the option isn't specified. If
3267 **split** is specified, then fio will assign one cpu per job. If not
3268 enough CPUs are given for the jobs listed, then fio will roundrobin the CPUs
3271 .. option:: cpumask=int
3273 Set the CPU affinity of this job. The parameter given is a bit mask of
3274 allowed CPUs the job may run on. So if you want the allowed CPUs to be 1
3275 and 5, you would pass the decimal value of (1 << 1 | 1 << 5), or 34. See man
3276 :manpage:`sched_setaffinity(2)`. This may not work on all supported
3277 operating systems or kernel versions. This option doesn't work well for a
3278 higher CPU count than what you can store in an integer mask, so it can only
3279 control cpus 1-32. For boxes with larger CPU counts, use
3280 :option:`cpus_allowed`.
3282 .. option:: numa_cpu_nodes=str
3284 Set this job running on specified NUMA nodes' CPUs. The arguments allow
3285 comma delimited list of cpu numbers, A-B ranges, or `all`. Note, to enable
3286 NUMA options support, fio must be built on a system with libnuma-dev(el)
3289 .. option:: numa_mem_policy=str
3291 Set this job's memory policy and corresponding NUMA nodes. Format of the
3296 ``mode`` is one of the following memory policies: ``default``, ``prefer``,
3297 ``bind``, ``interleave`` or ``local``. For ``default`` and ``local`` memory
3298 policies, no node needs to be specified. For ``prefer``, only one node is
3299 allowed. For ``bind`` and ``interleave`` the ``nodelist`` may be as
3300 follows: a comma delimited list of numbers, A-B ranges, or `all`.
3302 .. option:: cgroup=str
3304 Add job to this control group. If it doesn't exist, it will be created. The
3305 system must have a mounted cgroup blkio mount point for this to work. If
3306 your system doesn't have it mounted, you can do so with::
3308 # mount -t cgroup -o blkio none /cgroup
3310 .. option:: cgroup_weight=int
3312 Set the weight of the cgroup to this value. See the documentation that comes
3313 with the kernel, allowed values are in the range of 100..1000.
3315 .. option:: cgroup_nodelete=bool
3317 Normally fio will delete the cgroups it has created after the job
3318 completion. To override this behavior and to leave cgroups around after the
3319 job completion, set ``cgroup_nodelete=1``. This can be useful if one wants
3320 to inspect various cgroup files after job completion. Default: false.
3322 .. option:: flow_id=int
3324 The ID of the flow. If not specified, it defaults to being a global
3325 flow. See :option:`flow`.
3327 .. option:: flow=int
3329 Weight in token-based flow control. If this value is used, then there is a
3330 'flow counter' which is used to regulate the proportion of activity between
3331 two or more jobs. Fio attempts to keep this flow counter near zero. The
3332 ``flow`` parameter stands for how much should be added or subtracted to the
3333 flow counter on each iteration of the main I/O loop. That is, if one job has
3334 ``flow=8`` and another job has ``flow=-1``, then there will be a roughly 1:8
3335 ratio in how much one runs vs the other.
3337 .. option:: flow_sleep=int
3339 The period of time, in microseconds, to wait after the flow counter
3340 has exceeded its proportion before retrying operations.
3342 .. option:: stonewall, wait_for_previous
3344 Wait for preceding jobs in the job file to exit, before starting this
3345 one. Can be used to insert serialization points in the job file. A stone
3346 wall also implies starting a new reporting group, see
3347 :option:`group_reporting`.
3351 By default, fio will continue running all other jobs when one job finishes.
3352 Sometimes this is not the desired action. Setting ``exitall`` will instead
3353 make fio terminate all jobs in the same group, as soon as one job of that
3356 .. option:: exit_what
3358 By default, fio will continue running all other jobs when one job finishes.
3359 Sometimes this is not the desired action. Setting ``exit_all`` will
3360 instead make fio terminate all jobs in the same group. The option
3361 ``exit_what`` allows to control which jobs get terminated when ``exitall`` is
3362 enabled. The default is ``group`` and does not change the behaviour of
3363 ``exitall``. The setting ``all`` terminates all jobs. The setting ``stonewall``
3364 terminates all currently running jobs across all groups and continues execution
3365 with the next stonewalled group.
3367 .. option:: exec_prerun=str
3369 Before running this job, issue the command specified through
3370 :manpage:`system(3)`. Output is redirected in a file called
3371 :file:`jobname.prerun.txt`.
3373 .. option:: exec_postrun=str
3375 After the job completes, issue the command specified though
3376 :manpage:`system(3)`. Output is redirected in a file called
3377 :file:`jobname.postrun.txt`.
3381 Instead of running as the invoking user, set the user ID to this value
3382 before the thread/process does any work.
3386 Set group ID, see :option:`uid`.
3392 .. option:: verify_only
3394 Do not perform specified workload, only verify data still matches previous
3395 invocation of this workload. This option allows one to check data multiple
3396 times at a later date without overwriting it. This option makes sense only
3397 for workloads that write data, and does not support workloads with the
3398 :option:`time_based` option set.
3400 .. option:: do_verify=bool
3402 Run the verify phase after a write phase. Only valid if :option:`verify` is
3405 .. option:: verify=str
3407 If writing to a file, fio can verify the file contents after each iteration
3408 of the job. Each verification method also implies verification of special
3409 header, which is written to the beginning of each block. This header also
3410 includes meta information, like offset of the block, block number, timestamp
3411 when block was written, etc. :option:`verify` can be combined with
3412 :option:`verify_pattern` option. The allowed values are:
3415 Use an md5 sum of the data area and store it in the header of
3419 Use an experimental crc64 sum of the data area and store it in the
3420 header of each block.
3423 Use a crc32c sum of the data area and store it in the header of
3424 each block. This will automatically use hardware acceleration
3425 (e.g. SSE4.2 on an x86 or CRC crypto extensions on ARM64) but will
3426 fall back to software crc32c if none is found. Generally the
3427 fastest checksum fio supports when hardware accelerated.
3433 Use a crc32 sum of the data area and store it in the header of each
3437 Use a crc16 sum of the data area and store it in the header of each
3441 Use a crc7 sum of the data area and store it in the header of each
3445 Use xxhash as the checksum function. Generally the fastest software
3446 checksum that fio supports.
3449 Use sha512 as the checksum function.
3452 Use sha256 as the checksum function.
3455 Use optimized sha1 as the checksum function.
3458 Use optimized sha3-224 as the checksum function.
3461 Use optimized sha3-256 as the checksum function.
3464 Use optimized sha3-384 as the checksum function.
3467 Use optimized sha3-512 as the checksum function.
3470 This option is deprecated, since now meta information is included in
3471 generic verification header and meta verification happens by
3472 default. For detailed information see the description of the
3473 :option:`verify` setting. This option is kept because of
3474 compatibility's sake with old configurations. Do not use it.
3477 Verify a strict pattern. Normally fio includes a header with some
3478 basic information and checksumming, but if this option is set, only
3479 the specific pattern set with :option:`verify_pattern` is verified.
3482 Only pretend to verify. Useful for testing internals with
3483 :option:`ioengine`\=null, not for much else.
3485 This option can be used for repeated burn-in tests of a system to make sure
3486 that the written data is also correctly read back. If the data direction
3487 given is a read or random read, fio will assume that it should verify a
3488 previously written file. If the data direction includes any form of write,
3489 the verify will be of the newly written data.
3491 To avoid false verification errors, do not use the norandommap option when
3492 verifying data with async I/O engines and I/O depths > 1. Or use the
3493 norandommap and the lfsr random generator together to avoid writing to the
3494 same offset with multiple outstanding I/Os.
3496 .. option:: verify_offset=int
3498 Swap the verification header with data somewhere else in the block before
3499 writing. It is swapped back before verifying.
3501 .. option:: verify_interval=int
3503 Write the verification header at a finer granularity than the
3504 :option:`blocksize`. It will be written for chunks the size of
3505 ``verify_interval``. :option:`blocksize` should divide this evenly.
3507 .. option:: verify_pattern=str
3509 If set, fio will fill the I/O buffers with this pattern. Fio defaults to
3510 filling with totally random bytes, but sometimes it's interesting to fill
3511 with a known pattern for I/O verification purposes. Depending on the width
3512 of the pattern, fio will fill 1/2/3/4 bytes of the buffer at the time (it can
3513 be either a decimal or a hex number). The ``verify_pattern`` if larger than
3514 a 32-bit quantity has to be a hex number that starts with either "0x" or
3515 "0X". Use with :option:`verify`. Also, ``verify_pattern`` supports %o
3516 format, which means that for each block offset will be written and then
3517 verified back, e.g.::
3521 Or use combination of everything::
3523 verify_pattern=0xff%o"abcd"-12
3525 .. option:: verify_fatal=bool
3527 Normally fio will keep checking the entire contents before quitting on a
3528 block verification failure. If this option is set, fio will exit the job on
3529 the first observed failure. Default: false.
3531 .. option:: verify_dump=bool
3533 If set, dump the contents of both the original data block and the data block
3534 we read off disk to files. This allows later analysis to inspect just what
3535 kind of data corruption occurred. Off by default.
3537 .. option:: verify_async=int
3539 Fio will normally verify I/O inline from the submitting thread. This option
3540 takes an integer describing how many async offload threads to create for I/O
3541 verification instead, causing fio to offload the duty of verifying I/O
3542 contents to one or more separate threads. If using this offload option, even
3543 sync I/O engines can benefit from using an :option:`iodepth` setting higher
3544 than 1, as it allows them to have I/O in flight while verifies are running.
3545 Defaults to 0 async threads, i.e. verification is not asynchronous.
3547 .. option:: verify_async_cpus=str
3549 Tell fio to set the given CPU affinity on the async I/O verification
3550 threads. See :option:`cpus_allowed` for the format used.
3552 .. option:: verify_backlog=int
3554 Fio will normally verify the written contents of a job that utilizes verify
3555 once that job has completed. In other words, everything is written then
3556 everything is read back and verified. You may want to verify continually
3557 instead for a variety of reasons. Fio stores the meta data associated with
3558 an I/O block in memory, so for large verify workloads, quite a bit of memory
3559 would be used up holding this meta data. If this option is enabled, fio will
3560 write only N blocks before verifying these blocks.
3562 .. option:: verify_backlog_batch=int
3564 Control how many blocks fio will verify if :option:`verify_backlog` is
3565 set. If not set, will default to the value of :option:`verify_backlog`
3566 (meaning the entire queue is read back and verified). If
3567 ``verify_backlog_batch`` is less than :option:`verify_backlog` then not all
3568 blocks will be verified, if ``verify_backlog_batch`` is larger than
3569 :option:`verify_backlog`, some blocks will be verified more than once.
3571 .. option:: verify_state_save=bool
3573 When a job exits during the write phase of a verify workload, save its
3574 current state. This allows fio to replay up until that point, if the verify
3575 state is loaded for the verify read phase. The format of the filename is,
3578 <type>-<jobname>-<jobindex>-verify.state.
3580 <type> is "local" for a local run, "sock" for a client/server socket
3581 connection, and "ip" (192.168.0.1, for instance) for a networked
3582 client/server connection. Defaults to true.
3584 .. option:: verify_state_load=bool
3586 If a verify termination trigger was used, fio stores the current write state
3587 of each thread. This can be used at verification time so that fio knows how
3588 far it should verify. Without this information, fio will run a full
3589 verification pass, according to the settings in the job file used. Default
3592 .. option:: trim_percentage=int
3594 Number of verify blocks to discard/trim.
3596 .. option:: trim_verify_zero=bool
3598 Verify that trim/discarded blocks are returned as zeros.
3600 .. option:: trim_backlog=int
3602 Trim after this number of blocks are written.
3604 .. option:: trim_backlog_batch=int
3606 Trim this number of I/O blocks.
3608 .. option:: experimental_verify=bool
3610 Enable experimental verification.
3615 .. option:: steadystate=str:float, ss=str:float
3617 Define the criterion and limit for assessing steady state performance. The
3618 first parameter designates the criterion whereas the second parameter sets
3619 the threshold. When the criterion falls below the threshold for the
3620 specified duration, the job will stop. For example, `iops_slope:0.1%` will
3621 direct fio to terminate the job when the least squares regression slope
3622 falls below 0.1% of the mean IOPS. If :option:`group_reporting` is enabled
3623 this will apply to all jobs in the group. Below is the list of available
3624 steady state assessment criteria. All assessments are carried out using only
3625 data from the rolling collection window. Threshold limits can be expressed
3626 as a fixed value or as a percentage of the mean in the collection window.
3628 When using this feature, most jobs should include the :option:`time_based`
3629 and :option:`runtime` options or the :option:`loops` option so that fio does not
3630 stop running after it has covered the full size of the specified file(s) or device(s).
3633 Collect IOPS data. Stop the job if all individual IOPS measurements
3634 are within the specified limit of the mean IOPS (e.g., ``iops:2``
3635 means that all individual IOPS values must be within 2 of the mean,
3636 whereas ``iops:0.2%`` means that all individual IOPS values must be
3637 within 0.2% of the mean IOPS to terminate the job).
3640 Collect IOPS data and calculate the least squares regression
3641 slope. Stop the job if the slope falls below the specified limit.
3644 Collect bandwidth data. Stop the job if all individual bandwidth
3645 measurements are within the specified limit of the mean bandwidth.
3648 Collect bandwidth data and calculate the least squares regression
3649 slope. Stop the job if the slope falls below the specified limit.
3651 .. option:: steadystate_duration=time, ss_dur=time
3653 A rolling window of this duration will be used to judge whether steady state
3654 has been reached. Data will be collected once per second. The default is 0
3655 which disables steady state detection. When the unit is omitted, the
3656 value is interpreted in seconds.
3658 .. option:: steadystate_ramp_time=time, ss_ramp=time
3660 Allow the job to run for the specified duration before beginning data
3661 collection for checking the steady state job termination criterion. The
3662 default is 0. When the unit is omitted, the value is interpreted in seconds.
3665 Measurements and reporting
3666 ~~~~~~~~~~~~~~~~~~~~~~~~~~
3668 .. option:: per_job_logs=bool
3670 If set, this generates bw/clat/iops log with per file private filenames. If
3671 not set, jobs with identical names will share the log filename. Default:
3674 .. option:: group_reporting
3676 It may sometimes be interesting to display statistics for groups of jobs as
3677 a whole instead of for each individual job. This is especially true if
3678 :option:`numjobs` is used; looking at individual thread/process output
3679 quickly becomes unwieldy. To see the final report per-group instead of
3680 per-job, use :option:`group_reporting`. Jobs in a file will be part of the
3681 same reporting group, unless if separated by a :option:`stonewall`, or by
3682 using :option:`new_group`.
3684 .. option:: new_group
3686 Start a new reporting group. See: :option:`group_reporting`. If not given,
3687 all jobs in a file will be part of the same reporting group, unless
3688 separated by a :option:`stonewall`.
3690 .. option:: stats=bool
3692 By default, fio collects and shows final output results for all jobs
3693 that run. If this option is set to 0, then fio will ignore it in
3694 the final stat output.
3696 .. option:: write_bw_log=str
3698 If given, write a bandwidth log for this job. Can be used to store data of
3699 the bandwidth of the jobs in their lifetime.
3701 If no str argument is given, the default filename of
3702 :file:`jobname_type.x.log` is used. Even when the argument is given, fio
3703 will still append the type of log. So if one specifies::
3707 The actual log name will be :file:`foo_bw.x.log` where `x` is the index
3708 of the job (`1..N`, where `N` is the number of jobs). If
3709 :option:`per_job_logs` is false, then the filename will not include the
3712 The included :command:`fio_generate_plots` script uses :command:`gnuplot` to turn these
3713 text files into nice graphs. See `Log File Formats`_ for how data is
3714 structured within the file.
3716 .. option:: write_lat_log=str
3718 Same as :option:`write_bw_log`, except this option creates I/O
3719 submission (e.g., :file:`name_slat.x.log`), completion (e.g.,
3720 :file:`name_clat.x.log`), and total (e.g., :file:`name_lat.x.log`)
3721 latency files instead. See :option:`write_bw_log` for details about
3722 the filename format and `Log File Formats`_ for how data is structured
3725 .. option:: write_hist_log=str
3727 Same as :option:`write_bw_log` but writes an I/O completion latency
3728 histogram file (e.g., :file:`name_hist.x.log`) instead. Note that this
3729 file will be empty unless :option:`log_hist_msec` has also been set.
3730 See :option:`write_bw_log` for details about the filename format and
3731 `Log File Formats`_ for how data is structured within the file.
3733 .. option:: write_iops_log=str
3735 Same as :option:`write_bw_log`, but writes an IOPS file (e.g.
3736 :file:`name_iops.x.log`) instead. Because fio defaults to individual
3737 I/O logging, the value entry in the IOPS log will be 1 unless windowed
3738 logging (see :option:`log_avg_msec`) has been enabled. See
3739 :option:`write_bw_log` for details about the filename format and `Log
3740 File Formats`_ for how data is structured within the file.
3742 .. option:: log_entries=int
3744 By default, fio will log an entry in the iops, latency, or bw log for
3745 every I/O that completes. The initial number of I/O log entries is 1024.
3746 When the log entries are all used, new log entries are dynamically
3747 allocated. This dynamic log entry allocation may negatively impact
3748 time-related statistics such as I/O tail latencies (e.g. 99.9th percentile
3749 completion latency). This option allows specifying a larger initial
3750 number of log entries to avoid run-time allocations of new log entries,
3751 resulting in more precise time-related I/O statistics.
3752 Also see :option:`log_avg_msec`. Defaults to 1024.
3754 .. option:: log_avg_msec=int
3756 By default, fio will log an entry in the iops, latency, or bw log for every
3757 I/O that completes. When writing to the disk log, that can quickly grow to a
3758 very large size. Setting this option makes fio average the each log entry
3759 over the specified period of time, reducing the resolution of the log. See
3760 :option:`log_max_value` as well. Defaults to 0, logging all entries.
3761 Also see `Log File Formats`_.
3763 .. option:: log_hist_msec=int
3765 Same as :option:`log_avg_msec`, but logs entries for completion latency
3766 histograms. Computing latency percentiles from averages of intervals using
3767 :option:`log_avg_msec` is inaccurate. Setting this option makes fio log
3768 histogram entries over the specified period of time, reducing log sizes for
3769 high IOPS devices while retaining percentile accuracy. See
3770 :option:`log_hist_coarseness` and :option:`write_hist_log` as well.
3771 Defaults to 0, meaning histogram logging is disabled.
3773 .. option:: log_hist_coarseness=int
3775 Integer ranging from 0 to 6, defining the coarseness of the resolution of
3776 the histogram logs enabled with :option:`log_hist_msec`. For each increment
3777 in coarseness, fio outputs half as many bins. Defaults to 0, for which
3778 histogram logs contain 1216 latency bins. See :option:`write_hist_log`
3779 and `Log File Formats`_.
3781 .. option:: log_max_value=bool
3783 If :option:`log_avg_msec` is set, fio logs the average over that window. If
3784 you instead want to log the maximum value, set this option to 1. Defaults to
3785 0, meaning that averaged values are logged.
3787 .. option:: log_offset=bool
3789 If this is set, the iolog options will include the byte offset for the I/O
3790 entry as well as the other data values. Defaults to 0 meaning that
3791 offsets are not present in logs. Also see `Log File Formats`_.
3793 .. option:: log_compression=int
3795 If this is set, fio will compress the I/O logs as it goes, to keep the
3796 memory footprint lower. When a log reaches the specified size, that chunk is
3797 removed and compressed in the background. Given that I/O logs are fairly
3798 highly compressible, this yields a nice memory savings for longer runs. The
3799 downside is that the compression will consume some background CPU cycles, so
3800 it may impact the run. This, however, is also true if the logging ends up
3801 consuming most of the system memory. So pick your poison. The I/O logs are
3802 saved normally at the end of a run, by decompressing the chunks and storing
3803 them in the specified log file. This feature depends on the availability of
3806 .. option:: log_compression_cpus=str
3808 Define the set of CPUs that are allowed to handle online log compression for
3809 the I/O jobs. This can provide better isolation between performance
3810 sensitive jobs, and background compression work. See
3811 :option:`cpus_allowed` for the format used.
3813 .. option:: log_store_compressed=bool
3815 If set, fio will store the log files in a compressed format. They can be
3816 decompressed with fio, using the :option:`--inflate-log` command line
3817 parameter. The files will be stored with a :file:`.fz` suffix.
3819 .. option:: log_unix_epoch=bool
3821 If set, fio will log Unix timestamps to the log files produced by enabling
3822 write_type_log for each log type, instead of the default zero-based
3825 .. option:: log_alternate_epoch=bool
3827 If set, fio will log timestamps based on the epoch used by the clock specified
3828 in the log_alternate_epoch_clock_id option, to the log files produced by
3829 enabling write_type_log for each log type, instead of the default zero-based
3832 .. option:: log_alternate_epoch_clock_id=int
3834 Specifies the clock_id to be used by clock_gettime to obtain the alternate epoch
3835 if either log_unix_epoch or log_alternate_epoch are true. Otherwise has no
3836 effect. Default value is 0, or CLOCK_REALTIME.
3838 .. option:: block_error_percentiles=bool
3840 If set, record errors in trim block-sized units from writes and trims and
3841 output a histogram of how many trims it took to get to errors, and what kind
3842 of error was encountered.
3844 .. option:: bwavgtime=int
3846 Average the calculated bandwidth over the given time. Value is specified in
3847 milliseconds. If the job also does bandwidth logging through
3848 :option:`write_bw_log`, then the minimum of this option and
3849 :option:`log_avg_msec` will be used. Default: 500ms.
3851 .. option:: iopsavgtime=int
3853 Average the calculated IOPS over the given time. Value is specified in
3854 milliseconds. If the job also does IOPS logging through
3855 :option:`write_iops_log`, then the minimum of this option and
3856 :option:`log_avg_msec` will be used. Default: 500ms.
3858 .. option:: disk_util=bool
3860 Generate disk utilization statistics, if the platform supports it.
3863 .. option:: disable_lat=bool
3865 Disable measurements of total latency numbers. Useful only for cutting back
3866 the number of calls to :manpage:`gettimeofday(2)`, as that does impact
3867 performance at really high IOPS rates. Note that to really get rid of a
3868 large amount of these calls, this option must be used with
3869 :option:`disable_slat` and :option:`disable_bw_measurement` as well.
3871 .. option:: disable_clat=bool
3873 Disable measurements of completion latency numbers. See
3874 :option:`disable_lat`.
3876 .. option:: disable_slat=bool
3878 Disable measurements of submission latency numbers. See
3879 :option:`disable_lat`.
3881 .. option:: disable_bw_measurement=bool, disable_bw=bool
3883 Disable measurements of throughput/bandwidth numbers. See
3884 :option:`disable_lat`.
3886 .. option:: slat_percentiles=bool
3888 Report submission latency percentiles. Submission latency is not recorded
3889 for synchronous ioengines.
3891 .. option:: clat_percentiles=bool
3893 Report completion latency percentiles.
3895 .. option:: lat_percentiles=bool
3897 Report total latency percentiles. Total latency is the sum of submission
3898 latency and completion latency.
3900 .. option:: percentile_list=float_list
3902 Overwrite the default list of percentiles for latencies and the block error
3903 histogram. Each number is a floating point number in the range (0,100], and
3904 the maximum length of the list is 20. Use ``:`` to separate the numbers. For
3905 example, ``--percentile_list=99.5:99.9`` will cause fio to report the
3906 latency durations below which 99.5% and 99.9% of the observed latencies fell,
3909 .. option:: significant_figures=int
3911 If using :option:`--output-format` of `normal`, set the significant
3912 figures to this value. Higher values will yield more precise IOPS and
3913 throughput units, while lower values will round. Requires a minimum
3914 value of 1 and a maximum value of 10. Defaults to 4.
3920 .. option:: exitall_on_error
3922 When one job finishes in error, terminate the rest. The default is to wait
3923 for each job to finish.
3925 .. option:: continue_on_error=str
3927 Normally fio will exit the job on the first observed failure. If this option
3928 is set, fio will continue the job when there is a 'non-fatal error' (EIO or
3929 EILSEQ) until the runtime is exceeded or the I/O size specified is
3930 completed. If this option is used, there are two more stats that are
3931 appended, the total error count and the first error. The error field given
3932 in the stats is the first error that was hit during the run.
3934 Note: a write error from the device may go unnoticed by fio when using
3935 buffered IO, as the write() (or similar) system call merely dirties the
3936 kernel pages, unless :option:`sync` or :option:`direct` is used. Device IO
3937 errors occur when the dirty data is actually written out to disk. If fully
3938 sync writes aren't desirable, :option:`fsync` or :option:`fdatasync` can be
3939 used as well. This is specific to writes, as reads are always synchronous.
3941 The allowed values are:
3944 Exit on any I/O or verify errors.
3947 Continue on read errors, exit on all others.
3950 Continue on write errors, exit on all others.
3953 Continue on any I/O error, exit on all others.
3956 Continue on verify errors, exit on all others.
3959 Continue on all errors.
3962 Backward-compatible alias for 'none'.
3965 Backward-compatible alias for 'all'.
3967 .. option:: ignore_error=str
3969 Sometimes you want to ignore some errors during test in that case you can
3970 specify error list for each error type, instead of only being able to
3971 ignore the default 'non-fatal error' using :option:`continue_on_error`.
3972 ``ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST`` errors for
3973 given error type is separated with ':'. Error may be symbol ('ENOSPC',
3974 'ENOMEM') or integer. Example::
3976 ignore_error=EAGAIN,ENOSPC:122
3978 This option will ignore EAGAIN from READ, and ENOSPC and 122(EDQUOT) from
3979 WRITE. This option works by overriding :option:`continue_on_error` with
3980 the list of errors for each error type if any.
3982 .. option:: error_dump=bool
3984 If set dump every error even if it is non fatal, true by default. If
3985 disabled only fatal error will be dumped.
3987 Running predefined workloads
3988 ----------------------------
3990 Fio includes predefined profiles that mimic the I/O workloads generated by
3993 .. option:: profile=str
3995 The predefined workload to run. Current profiles are:
3998 Threaded I/O bench (tiotest/tiobench) like workload.
4001 Aerospike Certification Tool (ACT) like workload.
4003 To view a profile's additional options use :option:`--cmdhelp` after specifying
4004 the profile. For example::
4006 $ fio --profile=act --cmdhelp
4011 .. option:: device-names=str
4016 .. option:: load=int
4019 ACT load multiplier. Default: 1.
4021 .. option:: test-duration=time
4024 How long the entire test takes to run. When the unit is omitted, the value
4025 is given in seconds. Default: 24h.
4027 .. option:: threads-per-queue=int
4030 Number of read I/O threads per device. Default: 8.
4032 .. option:: read-req-num-512-blocks=int
4035 Number of 512B blocks to read at the time. Default: 3.
4037 .. option:: large-block-op-kbytes=int
4040 Size of large block ops in KiB (writes). Default: 131072.
4045 Set to run ACT prep phase.
4047 Tiobench profile options
4048 ~~~~~~~~~~~~~~~~~~~~~~~~
4050 .. option:: size=str
4055 .. option:: block=int
4058 Block size in bytes. Default: 4096.
4060 .. option:: numruns=int
4070 .. option:: threads=int
4075 Interpreting the output
4076 -----------------------
4079 Example output was based on the following:
4080 TZ=UTC fio --iodepth=8 --ioengine=null --size=100M --time_based \
4081 --rate=1256k --bs=14K --name=quick --runtime=1s --name=mixed \
4082 --runtime=2m --rw=rw
4084 Fio spits out a lot of output. While running, fio will display the status of the
4085 jobs created. An example of that would be::
4087 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]
4089 The characters inside the first set of square brackets denote the current status of
4090 each thread. The first character is the first job defined in the job file, and so
4091 forth. The possible values (in typical life cycle order) are:
4093 +------+-----+-----------------------------------------------------------+
4095 +======+=====+===========================================================+
4096 | P | | Thread setup, but not started. |
4097 +------+-----+-----------------------------------------------------------+
4098 | C | | Thread created. |
4099 +------+-----+-----------------------------------------------------------+
4100 | I | | Thread initialized, waiting or generating necessary data. |
4101 +------+-----+-----------------------------------------------------------+
4102 | | p | Thread running pre-reading file(s). |
4103 +------+-----+-----------------------------------------------------------+
4104 | | / | Thread is in ramp period. |
4105 +------+-----+-----------------------------------------------------------+
4106 | | R | Running, doing sequential reads. |
4107 +------+-----+-----------------------------------------------------------+
4108 | | r | Running, doing random reads. |
4109 +------+-----+-----------------------------------------------------------+
4110 | | W | Running, doing sequential writes. |
4111 +------+-----+-----------------------------------------------------------+
4112 | | w | Running, doing random writes. |
4113 +------+-----+-----------------------------------------------------------+
4114 | | M | Running, doing mixed sequential reads/writes. |
4115 +------+-----+-----------------------------------------------------------+
4116 | | m | Running, doing mixed random reads/writes. |
4117 +------+-----+-----------------------------------------------------------+
4118 | | D | Running, doing sequential trims. |
4119 +------+-----+-----------------------------------------------------------+
4120 | | d | Running, doing random trims. |
4121 +------+-----+-----------------------------------------------------------+
4122 | | F | Running, currently waiting for :manpage:`fsync(2)`. |
4123 +------+-----+-----------------------------------------------------------+
4124 | | V | Running, doing verification of written data. |
4125 +------+-----+-----------------------------------------------------------+
4126 | f | | Thread finishing. |
4127 +------+-----+-----------------------------------------------------------+
4128 | E | | Thread exited, not reaped by main thread yet. |
4129 +------+-----+-----------------------------------------------------------+
4130 | _ | | Thread reaped. |
4131 +------+-----+-----------------------------------------------------------+
4132 | X | | Thread reaped, exited with an error. |
4133 +------+-----+-----------------------------------------------------------+
4134 | K | | Thread reaped, exited due to signal. |
4135 +------+-----+-----------------------------------------------------------+
4138 Example output was based on the following:
4139 TZ=UTC fio --iodepth=8 --ioengine=null --size=100M --runtime=58m \
4140 --time_based --rate=2512k --bs=256K --numjobs=10 \
4141 --name=readers --rw=read --name=writers --rw=write
4143 Fio will condense the thread string as not to take up more space on the command
4144 line than needed. For instance, if you have 10 readers and 10 writers running,
4145 the output would look like this::
4147 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]
4149 Note that the status string is displayed in order, so it's possible to tell which of
4150 the jobs are currently doing what. In the example above this means that jobs 1--10
4151 are readers and 11--20 are writers.
4153 The other values are fairly self explanatory -- number of threads currently
4154 running and doing I/O, the number of currently open files (f=), the estimated
4155 completion percentage, the rate of I/O since last check (read speed listed first,
4156 then write speed and optionally trim speed) in terms of bandwidth and IOPS,
4157 and time to completion for the current running group. It's impossible to estimate
4158 runtime of the following groups (if any).
4161 Example output was based on the following:
4162 TZ=UTC fio --iodepth=16 --ioengine=posixaio --filename=/tmp/fiofile \
4163 --direct=1 --size=100M --time_based --runtime=50s --rate_iops=89 \
4164 --bs=7K --name=Client1 --rw=write
4166 When fio is done (or interrupted by :kbd:`Ctrl-C`), it will show the data for
4167 each thread, group of threads, and disks in that order. For each overall thread (or
4168 group) the output looks like::
4170 Client1: (groupid=0, jobs=1): err= 0: pid=16109: Sat Jun 24 12:07:54 2017
4171 write: IOPS=88, BW=623KiB/s (638kB/s)(30.4MiB/50032msec)
4172 slat (nsec): min=500, max=145500, avg=8318.00, stdev=4781.50
4173 clat (usec): min=170, max=78367, avg=4019.02, stdev=8293.31
4174 lat (usec): min=174, max=78375, avg=4027.34, stdev=8291.79
4175 clat percentiles (usec):
4176 | 1.00th=[ 302], 5.00th=[ 326], 10.00th=[ 343], 20.00th=[ 363],
4177 | 30.00th=[ 392], 40.00th=[ 404], 50.00th=[ 416], 60.00th=[ 445],
4178 | 70.00th=[ 816], 80.00th=[ 6718], 90.00th=[12911], 95.00th=[21627],
4179 | 99.00th=[43779], 99.50th=[51643], 99.90th=[68682], 99.95th=[72877],
4181 bw ( KiB/s): min= 532, max= 686, per=0.10%, avg=622.87, stdev=24.82, samples= 100
4182 iops : min= 76, max= 98, avg=88.98, stdev= 3.54, samples= 100
4183 lat (usec) : 250=0.04%, 500=64.11%, 750=4.81%, 1000=2.79%
4184 lat (msec) : 2=4.16%, 4=1.84%, 10=4.90%, 20=11.33%, 50=5.37%
4185 lat (msec) : 100=0.65%
4186 cpu : usr=0.27%, sys=0.18%, ctx=12072, majf=0, minf=21
4187 IO depths : 1=85.0%, 2=13.1%, 4=1.8%, 8=0.1%, 16=0.0%, 32=0.0%, >=64=0.0%
4188 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
4189 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
4190 issued rwt: total=0,4450,0, short=0,0,0, dropped=0,0,0
4191 latency : target=0, window=0, percentile=100.00%, depth=8
4193 The job name (or first job's name when using :option:`group_reporting`) is printed,
4194 along with the group id, count of jobs being aggregated, last error id seen (which
4195 is 0 when there are no errors), pid/tid of that thread and the time the job/group
4196 completed. Below are the I/O statistics for each data direction performed (showing
4197 writes in the example above). In the order listed, they denote:
4200 The string before the colon shows the I/O direction the statistics
4201 are for. **IOPS** is the average I/Os performed per second. **BW**
4202 is the average bandwidth rate shown as: value in power of 2 format
4203 (value in power of 10 format). The last two values show: (**total
4204 I/O performed** in power of 2 format / **runtime** of that thread).
4207 Submission latency (**min** being the minimum, **max** being the
4208 maximum, **avg** being the average, **stdev** being the standard
4209 deviation). This is the time from when fio initialized the I/O
4210 to submission. For synchronous ioengines this includes the time
4211 up until just before the ioengine's queue function is called.
4212 For asynchronous ioengines this includes the time up through the
4213 completion of the ioengine's queue function (and commit function
4214 if it is defined). For sync I/O this row is not displayed as the
4215 slat is negligible. This value can be in nanoseconds,
4216 microseconds or milliseconds --- fio will choose the most
4217 appropriate base and print that (in the example above
4218 nanoseconds was the best scale). Note: in :option:`--minimal`
4219 mode latencies are always expressed in microseconds.
4222 Completion latency. Same names as slat, this denotes the time from
4223 submission to completion of the I/O pieces. For sync I/O, this
4224 represents the time from when the I/O was submitted to the
4225 operating system to when it was completed. For asynchronous
4226 ioengines this is the time from when the ioengine's queue (and
4227 commit if available) functions were completed to when the I/O's
4228 completion was reaped by fio.
4231 Total latency. Same names as slat and clat, this denotes the time from
4232 when fio created the I/O unit to completion of the I/O operation.
4233 It is the sum of submission and completion latency.
4236 Bandwidth statistics based on samples. Same names as the xlat stats,
4237 but also includes the number of samples taken (**samples**) and an
4238 approximate percentage of total aggregate bandwidth this thread
4239 received in its group (**per**). This last value is only really
4240 useful if the threads in this group are on the same disk, since they
4241 are then competing for disk access.
4244 IOPS statistics based on samples. Same names as bw.
4246 **lat (nsec/usec/msec)**
4247 The distribution of I/O completion latencies. This is the time from when
4248 I/O leaves fio and when it gets completed. Unlike the separate
4249 read/write/trim sections above, the data here and in the remaining
4250 sections apply to all I/Os for the reporting group. 250=0.04% means that
4251 0.04% of the I/Os completed in under 250us. 500=64.11% means that 64.11%
4252 of the I/Os required 250 to 499us for completion.
4255 CPU usage. User and system time, along with the number of context
4256 switches this thread went through, usage of system and user time, and
4257 finally the number of major and minor page faults. The CPU utilization
4258 numbers are averages for the jobs in that reporting group, while the
4259 context and fault counters are summed.
4262 The distribution of I/O depths over the job lifetime. The numbers are
4263 divided into powers of 2 and each entry covers depths from that value
4264 up to those that are lower than the next entry -- e.g., 16= covers
4265 depths from 16 to 31. Note that the range covered by a depth
4266 distribution entry can be different to the range covered by the
4267 equivalent submit/complete distribution entry.
4270 How many pieces of I/O were submitting in a single submit call. Each
4271 entry denotes that amount and below, until the previous entry -- e.g.,
4272 16=100% means that we submitted anywhere between 9 to 16 I/Os per submit
4273 call. Note that the range covered by a submit distribution entry can
4274 be different to the range covered by the equivalent depth distribution
4278 Like the above submit number, but for completions instead.
4281 The number of read/write/trim requests issued, and how many of them were
4285 These values are for :option:`latency_target` and related options. When
4286 these options are engaged, this section describes the I/O depth required
4287 to meet the specified latency target.
4290 Example output was based on the following:
4291 TZ=UTC fio --ioengine=null --iodepth=2 --size=100M --numjobs=2 \
4292 --rate_process=poisson --io_limit=32M --name=read --bs=128k \
4293 --rate=11M --name=write --rw=write --bs=2k --rate=700k
4295 After each client has been listed, the group statistics are printed. They
4296 will look like this::
4298 Run status group 0 (all jobs):
4299 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
4300 WRITE: bw=1231KiB/s (1261kB/s), 616KiB/s-621KiB/s (630kB/s-636kB/s), io=64.0MiB (67.1MB), run=52747-53223msec
4302 For each data direction it prints:
4305 Aggregate bandwidth of threads in this group followed by the
4306 minimum and maximum bandwidth of all the threads in this group.
4307 Values outside of brackets are power-of-2 format and those
4308 within are the equivalent value in a power-of-10 format.
4310 Aggregate I/O performed of all threads in this group. The
4311 format is the same as bw.
4313 The smallest and longest runtimes of the threads in this group.
4315 And finally, the disk statistics are printed. This is Linux specific. They will look like this::
4317 Disk stats (read/write):
4318 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
4320 Each value is printed for both reads and writes, with reads first. The
4324 Number of I/Os performed by all groups.
4326 Number of merges performed by the I/O scheduler.
4328 Number of ticks we kept the disk busy.
4330 Total time spent in the disk queue.
4332 The disk utilization. A value of 100% means we kept the disk
4333 busy constantly, 50% would be a disk idling half of the time.
4335 It is also possible to get fio to dump the current output while it is running,
4336 without terminating the job. To do that, send fio the **USR1** signal. You can
4337 also get regularly timed dumps by using the :option:`--status-interval`
4338 parameter, or by creating a file in :file:`/tmp` named
4339 :file:`fio-dump-status`. If fio sees this file, it will unlink it and dump the
4340 current output status.
4346 For scripted usage where you typically want to generate tables or graphs of the
4347 results, fio can output the results in a semicolon separated format. The format
4348 is one long line of values, such as::
4350 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%
4351 A description of this job goes here.
4353 The job description (if provided) follows on a second line for terse v2.
4354 It appears on the same line for other terse versions.
4356 To enable terse output, use the :option:`--minimal` or
4357 :option:`--output-format`\=terse command line options. The
4358 first value is the version of the terse output format. If the output has to be
4359 changed for some reason, this number will be incremented by 1 to signify that
4362 Split up, the format is as follows (comments in brackets denote when a
4363 field was introduced or whether it's specific to some terse version):
4367 terse version, fio version [v3], jobname, groupid, error
4371 Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec)
4372 Submission latency: min, max, mean, stdev (usec)
4373 Completion latency: min, max, mean, stdev (usec)
4374 Completion latency percentiles: 20 fields (see below)
4375 Total latency: min, max, mean, stdev (usec)
4376 Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev, number of samples [v5]
4377 IOPS [v5]: min, max, mean, stdev, number of samples
4383 Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec)
4384 Submission latency: min, max, mean, stdev (usec)
4385 Completion latency: min, max, mean, stdev (usec)
4386 Completion latency percentiles: 20 fields (see below)
4387 Total latency: min, max, mean, stdev (usec)
4388 Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev, number of samples [v5]
4389 IOPS [v5]: min, max, mean, stdev, number of samples
4391 TRIM status [all but version 3]:
4393 Fields are similar to READ/WRITE status.
4397 user, system, context switches, major faults, minor faults
4401 <=1, 2, 4, 8, 16, 32, >=64
4403 I/O latencies microseconds::
4405 <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
4407 I/O latencies milliseconds::
4409 <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
4411 Disk utilization [v3]::
4413 disk name, read ios, write ios, read merges, write merges, read ticks, write ticks,
4414 time spent in queue, disk utilization percentage
4416 Additional Info (dependent on continue_on_error, default off)::
4418 total # errors, first error code
4420 Additional Info (dependent on description being set)::
4424 Completion latency percentiles can be a grouping of up to 20 sets, so for the
4425 terse output fio writes all of them. Each field will look like this::
4429 which is the Xth percentile, and the `usec` latency associated with it.
4431 For `Disk utilization`, all disks used by fio are shown. So for each disk there
4432 will be a disk utilization section.
4434 Below is a single line containing short names for each of the fields in the
4435 minimal output v3, separated by semicolons::
4437 terse_version_3;fio_version;jobname;groupid;error;read_kb;read_bandwidth_kb;read_iops;read_runtime_ms;read_slat_min_us;read_slat_max_us;read_slat_mean_us;read_slat_dev_us;read_clat_min_us;read_clat_max_us;read_clat_mean_us;read_clat_dev_us;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_us;read_lat_max_us;read_lat_mean_us;read_lat_dev_us;read_bw_min_kb;read_bw_max_kb;read_bw_agg_pct;read_bw_mean_kb;read_bw_dev_kb;write_kb;write_bandwidth_kb;write_iops;write_runtime_ms;write_slat_min_us;write_slat_max_us;write_slat_mean_us;write_slat_dev_us;write_clat_min_us;write_clat_max_us;write_clat_mean_us;write_clat_dev_us;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_us;write_lat_max_us;write_lat_mean_us;write_lat_dev_us;write_bw_min_kb;write_bw_max_kb;write_bw_agg_pct;write_bw_mean_kb;write_bw_dev_kb;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
4439 In client/server mode terse output differs from what appears when jobs are run
4440 locally. Disk utilization data is omitted from the standard terse output and
4441 for v3 and later appears on its own separate line at the end of each terse
4448 The `json` output format is intended to be both human readable and convenient
4449 for automated parsing. For the most part its sections mirror those of the
4450 `normal` output. The `runtime` value is reported in msec and the `bw` value is
4451 reported in 1024 bytes per second units.
4457 The `json+` output format is identical to the `json` output format except that it
4458 adds a full dump of the completion latency bins. Each `bins` object contains a
4459 set of (key, value) pairs where keys are latency durations and values count how
4460 many I/Os had completion latencies of the corresponding duration. For example,
4463 "bins" : { "87552" : 1, "89600" : 1, "94720" : 1, "96768" : 1, "97792" : 1, "99840" : 1, "100864" : 2, "103936" : 6, "104960" : 534, "105984" : 5995, "107008" : 7529, ... }
4465 This data indicates that one I/O required 87,552ns to complete, two I/Os required
4466 100,864ns to complete, and 7529 I/Os required 107,008ns to complete.
4468 Also included with fio is a Python script `fio_jsonplus_clat2csv` that takes
4469 json+ output and generates CSV-formatted latency data suitable for plotting.
4471 The latency durations actually represent the midpoints of latency intervals.
4472 For details refer to :file:`stat.h`.
4478 There are two trace file format that you can encounter. The older (v1) format is
4479 unsupported since version 1.20-rc3 (March 2008). It will still be described
4480 below in case that you get an old trace and want to understand it.
4482 In any case the trace is a simple text file with a single action per line.
4485 Trace file format v1
4486 ~~~~~~~~~~~~~~~~~~~~
4488 Each line represents a single I/O action in the following format::
4492 where `rw=0/1` for read/write, and the `offset` and `length` entries being in bytes.
4494 This format is not supported in fio versions >= 1.20-rc3.
4497 Trace file format v2
4498 ~~~~~~~~~~~~~~~~~~~~
4500 The second version of the trace file format was added in fio version 1.17. It
4501 allows to access more then one file per trace and has a bigger set of possible
4504 The first line of the trace file has to be::
4508 Following this can be lines in two different formats, which are described below.
4510 The file management format::
4514 The `filename` is given as an absolute path. The `action` can be one of these:
4517 Add the given `filename` to the trace.
4519 Open the file with the given `filename`. The `filename` has to have
4520 been added with the **add** action before.
4522 Close the file with the given `filename`. The file has to have been
4526 The file I/O action format::
4528 filename action offset length
4530 The `filename` is given as an absolute path, and has to have been added and
4531 opened before it can be used with this format. The `offset` and `length` are
4532 given in bytes. The `action` can be one of these:
4535 Wait for `offset` microseconds. Everything below 100 is discarded.
4536 The time is relative to the previous `wait` statement. Note that
4537 action `wait` is not allowed as of version 3, as the same behavior
4538 can be achieved using timestamps.
4540 Read `length` bytes beginning from `offset`.
4542 Write `length` bytes beginning from `offset`.
4544 :manpage:`fsync(2)` the file.
4546 :manpage:`fdatasync(2)` the file.
4548 Trim the given file from the given `offset` for `length` bytes.
4551 Trace file format v3
4552 ~~~~~~~~~~~~~~~~~~~~
4554 The third version of the trace file format was added in fio version 3.31. It
4555 forces each action to have a timestamp associated with it.
4557 The first line of the trace file has to be::
4561 Following this can be lines in two different formats, which are described below.
4563 The file management format::
4565 timestamp filename action
4567 The file I/O action format::
4569 timestamp filename action offset length
4571 The `timestamp` is relative to the beginning of the run (ie starts at 0). The
4572 `filename`, `action`, `offset` and `length` are identical to version 2, except
4573 that version 3 does not allow the `wait` action.
4576 I/O Replay - Merging Traces
4577 ---------------------------
4579 Colocation is a common practice used to get the most out of a machine.
4580 Knowing which workloads play nicely with each other and which ones don't is
4581 a much harder task. While fio can replay workloads concurrently via multiple
4582 jobs, it leaves some variability up to the scheduler making results harder to
4583 reproduce. Merging is a way to make the order of events consistent.
4585 Merging is integrated into I/O replay and done when a
4586 :option:`merge_blktrace_file` is specified. The list of files passed to
4587 :option:`read_iolog` go through the merge process and output a single file
4588 stored to the specified file. The output file is passed on as if it were the
4589 only file passed to :option:`read_iolog`. An example would look like::
4591 $ fio --read_iolog="<file1>:<file2>" --merge_blktrace_file="<output_file>"
4593 Creating only the merged file can be done by passing the command line argument
4594 :option:`--merge-blktrace-only`.
4596 Scaling traces can be done to see the relative impact of any particular trace
4597 being slowed down or sped up. :option:`merge_blktrace_scalars` takes in a colon
4598 separated list of percentage scalars. It is index paired with the files passed
4599 to :option:`read_iolog`.
4601 With scaling, it may be desirable to match the running time of all traces.
4602 This can be done with :option:`merge_blktrace_iters`. It is index paired with
4603 :option:`read_iolog` just like :option:`merge_blktrace_scalars`.
4605 In an example, given two traces, A and B, each 60s long. If we want to see
4606 the impact of trace A issuing IOs twice as fast and repeat trace A over the
4607 runtime of trace B, the following can be done::
4609 $ fio --read_iolog="<trace_a>:"<trace_b>" --merge_blktrace_file"<output_file>" --merge_blktrace_scalars="50:100" --merge_blktrace_iters="2:1"
4611 This runs trace A at 2x the speed twice for approximately the same runtime as
4612 a single run of trace B.
4615 CPU idleness profiling
4616 ----------------------
4618 In some cases, we want to understand CPU overhead in a test. For example, we
4619 test patches for the specific goodness of whether they reduce CPU usage.
4620 Fio implements a balloon approach to create a thread per CPU that runs at idle
4621 priority, meaning that it only runs when nobody else needs the cpu.
4622 By measuring the amount of work completed by the thread, idleness of each CPU
4623 can be derived accordingly.
4625 An unit work is defined as touching a full page of unsigned characters. Mean and
4626 standard deviation of time to complete an unit work is reported in "unit work"
4627 section. Options can be chosen to report detailed percpu idleness or overall
4628 system idleness by aggregating percpu stats.
4631 Verification and triggers
4632 -------------------------
4634 Fio is usually run in one of two ways, when data verification is done. The first
4635 is a normal write job of some sort with verify enabled. When the write phase has
4636 completed, fio switches to reads and verifies everything it wrote. The second
4637 model is running just the write phase, and then later on running the same job
4638 (but with reads instead of writes) to repeat the same I/O patterns and verify
4639 the contents. Both of these methods depend on the write phase being completed,
4640 as fio otherwise has no idea how much data was written.
4642 With verification triggers, fio supports dumping the current write state to
4643 local files. Then a subsequent read verify workload can load this state and know
4644 exactly where to stop. This is useful for testing cases where power is cut to a
4645 server in a managed fashion, for instance.
4647 A verification trigger consists of two things:
4649 1) Storing the write state of each job.
4650 2) Executing a trigger command.
4652 The write state is relatively small, on the order of hundreds of bytes to single
4653 kilobytes. It contains information on the number of completions done, the last X
4656 A trigger is invoked either through creation ('touch') of a specified file in
4657 the system, or through a timeout setting. If fio is run with
4658 :option:`--trigger-file`\= :file:`/tmp/trigger-file`, then it will continually
4659 check for the existence of :file:`/tmp/trigger-file`. When it sees this file, it
4660 will fire off the trigger (thus saving state, and executing the trigger
4663 For client/server runs, there's both a local and remote trigger. If fio is
4664 running as a server backend, it will send the job states back to the client for
4665 safe storage, then execute the remote trigger, if specified. If a local trigger
4666 is specified, the server will still send back the write state, but the client
4667 will then execute the trigger.
4669 Verification trigger example
4670 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4672 Let's say we want to run a powercut test on the remote Linux machine 'server'.
4673 Our write workload is in :file:`write-test.fio`. We want to cut power to 'server' at
4674 some point during the run, and we'll run this test from the safety or our local
4675 machine, 'localbox'. On the server, we'll start the fio backend normally::
4677 server# fio --server
4679 and on the client, we'll fire off the workload::
4681 localbox$ fio --client=server --trigger-file=/tmp/my-trigger --trigger-remote="bash -c \"echo b > /proc/sysrq-triger\""
4683 We set :file:`/tmp/my-trigger` as the trigger file, and we tell fio to execute::
4685 echo b > /proc/sysrq-trigger
4687 on the server once it has received the trigger and sent us the write state. This
4688 will work, but it's not **really** cutting power to the server, it's merely
4689 abruptly rebooting it. If we have a remote way of cutting power to the server
4690 through IPMI or similar, we could do that through a local trigger command
4691 instead. Let's assume we have a script that does IPMI reboot of a given hostname,
4692 ipmi-reboot. On localbox, we could then have run fio with a local trigger
4695 localbox$ fio --client=server --trigger-file=/tmp/my-trigger --trigger="ipmi-reboot server"
4697 For this case, fio would wait for the server to send us the write state, then
4698 execute ``ipmi-reboot server`` when that happened.
4700 Loading verify state
4701 ~~~~~~~~~~~~~~~~~~~~
4703 To load stored write state, a read verification job file must contain the
4704 :option:`verify_state_load` option. If that is set, fio will load the previously
4705 stored state. For a local fio run this is done by loading the files directly,
4706 and on a client/server run, the server backend will ask the client to send the
4707 files over and load them from there.
4713 Fio supports a variety of log file formats, for logging latencies, bandwidth,
4714 and IOPS. The logs share a common format, which looks like this:
4716 *time* (`msec`), *value*, *data direction*, *block size* (`bytes`),
4717 *offset* (`bytes`), *command priority*
4719 *Time* for the log entry is always in milliseconds. The *value* logged depends
4720 on the type of log, it will be one of the following:
4723 Value is latency in nsecs
4729 *Data direction* is one of the following:
4738 The entry's *block size* is always in bytes. The *offset* is the position in bytes
4739 from the start of the file for that particular I/O. The logging of the offset can be
4740 toggled with :option:`log_offset`.
4742 *Command priority* is 0 for normal priority and 1 for high priority. This is controlled
4743 by the ioengine specific :option:`cmdprio_percentage`.
4745 Fio defaults to logging every individual I/O but when windowed logging is set
4746 through :option:`log_avg_msec`, either the average (by default) or the maximum
4747 (:option:`log_max_value` is set) *value* seen over the specified period of time
4748 is recorded. Each *data direction* seen within the window period will aggregate
4749 its values in a separate row. Further, when using windowed logging the *block
4750 size* and *offset* entries will always contain 0.
4756 Normally fio is invoked as a stand-alone application on the machine where the
4757 I/O workload should be generated. However, the backend and frontend of fio can
4758 be run separately i.e., the fio server can generate an I/O workload on the "Device
4759 Under Test" while being controlled by a client on another machine.
4761 Start the server on the machine which has access to the storage DUT::
4765 where `args` defines what fio listens to. The arguments are of the form
4766 ``type,hostname`` or ``IP,port``. *type* is either ``ip`` (or ip4) for TCP/IP
4767 v4, ``ip6`` for TCP/IP v6, or ``sock`` for a local unix domain socket.
4768 *hostname* is either a hostname or IP address, and *port* is the port to listen
4769 to (only valid for TCP/IP, not a local socket). Some examples:
4773 Start a fio server, listening on all interfaces on the default port (8765).
4775 2) ``fio --server=ip:hostname,4444``
4777 Start a fio server, listening on IP belonging to hostname and on port 4444.
4779 3) ``fio --server=ip6:::1,4444``
4781 Start a fio server, listening on IPv6 localhost ::1 and on port 4444.
4783 4) ``fio --server=,4444``
4785 Start a fio server, listening on all interfaces on port 4444.
4787 5) ``fio --server=1.2.3.4``
4789 Start a fio server, listening on IP 1.2.3.4 on the default port.
4791 6) ``fio --server=sock:/tmp/fio.sock``
4793 Start a fio server, listening on the local socket :file:`/tmp/fio.sock`.
4795 Once a server is running, a "client" can connect to the fio server with::
4797 fio <local-args> --client=<server> <remote-args> <job file(s)>
4799 where `local-args` are arguments for the client where it is running, `server`
4800 is the connect string, and `remote-args` and `job file(s)` are sent to the
4801 server. The `server` string follows the same format as it does on the server
4802 side, to allow IP/hostname/socket and port strings.
4804 Fio can connect to multiple servers this way::
4806 fio --client=<server1> <job file(s)> --client=<server2> <job file(s)>
4808 If the job file is located on the fio server, then you can tell the server to
4809 load a local file as well. This is done by using :option:`--remote-config` ::
4811 fio --client=server --remote-config /path/to/file.fio
4813 Then fio will open this local (to the server) job file instead of being passed
4814 one from the client.
4816 If you have many servers (example: 100 VMs/containers), you can input a pathname
4817 of a file containing host IPs/names as the parameter value for the
4818 :option:`--client` option. For example, here is an example :file:`host.list`
4819 file containing 2 hostnames::
4821 host1.your.dns.domain
4822 host2.your.dns.domain
4824 The fio command would then be::
4826 fio --client=host.list <job file(s)>
4828 In this mode, you cannot input server-specific parameters or job files -- all
4829 servers receive the same job file.
4831 In order to let ``fio --client`` runs use a shared filesystem from multiple
4832 hosts, ``fio --client`` now prepends the IP address of the server to the
4833 filename. For example, if fio is using the directory :file:`/mnt/nfs/fio` and is
4834 writing filename :file:`fileio.tmp`, with a :option:`--client` `hostfile`
4835 containing two hostnames ``h1`` and ``h2`` with IP addresses 192.168.10.120 and
4836 192.168.10.121, then fio will create two files::
4838 /mnt/nfs/fio/192.168.10.120.fileio.tmp
4839 /mnt/nfs/fio/192.168.10.121.fileio.tmp
4841 Terse output in client/server mode will differ slightly from what is produced
4842 when fio is run in stand-alone mode. See the terse output section for details.