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=jobfile
172 Convert `jobfile` 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
1067 If this option is used, fio will ignore the maximum number of open
1068 zones limit of the zoned block device in use, thus allowing the
1069 option :option:`max_open_zones` value to be larger than the device
1070 reported limit. Default: false.
1072 .. option:: zone_reset_threshold=float
1074 A number between zero and one that indicates the ratio of logical
1075 blocks with data to the total number of logical blocks in the test
1076 above which zones should be reset periodically.
1078 .. option:: zone_reset_frequency=float
1080 A number between zero and one that indicates how often a zone reset
1081 should be issued if the zone reset threshold has been exceeded. A zone
1082 reset is submitted after each (1 / zone_reset_frequency) write
1083 requests. This and the previous parameter can be used to simulate
1084 garbage collection activity.
1090 .. option:: direct=bool
1092 If value is true, use non-buffered I/O. This is usually O_DIRECT. Note that
1093 OpenBSD and ZFS on Solaris don't support direct I/O. On Windows the synchronous
1094 ioengines don't support direct I/O. Default: false.
1096 .. option:: atomic=bool
1098 If value is true, attempt to use atomic direct I/O. Atomic writes are
1099 guaranteed to be stable once acknowledged by the operating system. Only
1100 Linux supports O_ATOMIC right now.
1102 .. option:: buffered=bool
1104 If value is true, use buffered I/O. This is the opposite of the
1105 :option:`direct` option. Defaults to true.
1107 .. option:: readwrite=str, rw=str
1109 Type of I/O pattern. Accepted values are:
1116 Sequential trims (Linux block devices and SCSI
1117 character devices only).
1123 Random trims (Linux block devices and SCSI
1124 character devices only).
1126 Sequential mixed reads and writes.
1128 Random mixed reads and writes.
1130 Sequential trim+write sequences. Blocks will be trimmed first,
1131 then the same blocks will be written to.
1133 Fio defaults to read if the option is not specified. For the mixed I/O
1134 types, the default is to split them 50/50. For certain types of I/O the
1135 result may still be skewed a bit, since the speed may be different.
1137 It is possible to specify the number of I/Os to do before getting a new
1138 offset by appending ``:<nr>`` to the end of the string given. For a
1139 random read, it would look like ``rw=randread:8`` for passing in an offset
1140 modifier with a value of 8. If the suffix is used with a sequential I/O
1141 pattern, then the *<nr>* value specified will be **added** to the generated
1142 offset for each I/O turning sequential I/O into sequential I/O with holes.
1143 For instance, using ``rw=write:4k`` will skip 4k for every write. Also see
1144 the :option:`rw_sequencer` option.
1146 .. option:: rw_sequencer=str
1148 If an offset modifier is given by appending a number to the ``rw=<str>``
1149 line, then this option controls how that number modifies the I/O offset
1150 being generated. Accepted values are:
1153 Generate sequential offset.
1155 Generate the same offset.
1157 ``sequential`` is only useful for random I/O, where fio would normally
1158 generate a new random offset for every I/O. If you append e.g. 8 to randread,
1159 you would get a new random offset for every 8 I/Os. The result would be a
1160 seek for only every 8 I/Os, instead of for every I/O. Use ``rw=randread:8``
1161 to specify that. As sequential I/O is already sequential, setting
1162 ``sequential`` for that would not result in any differences. ``identical``
1163 behaves in a similar fashion, except it sends the same offset 8 number of
1164 times before generating a new offset.
1166 .. option:: unified_rw_reporting=str
1168 Fio normally reports statistics on a per data direction basis, meaning that
1169 reads, writes, and trims are accounted and reported separately. This option
1170 determines whether fio reports the results normally, summed together, or as
1172 Accepted values are:
1175 Normal statistics reporting.
1178 Statistics are summed per data direction and reported together.
1181 Statistics are reported normally, followed by the mixed statistics.
1184 Backward-compatible alias for **none**.
1187 Backward-compatible alias for **mixed**.
1192 .. option:: randrepeat=bool
1194 Seed the random number generator used for random I/O patterns in a
1195 predictable way so the pattern is repeatable across runs. Default: true.
1197 .. option:: allrandrepeat=bool
1199 Seed all random number generators in a predictable way so results are
1200 repeatable across runs. Default: false.
1202 .. option:: randseed=int
1204 Seed the random number generators based on this seed value, to be able to
1205 control what sequence of output is being generated. If not set, the random
1206 sequence depends on the :option:`randrepeat` setting.
1208 .. option:: fallocate=str
1210 Whether pre-allocation is performed when laying down files.
1211 Accepted values are:
1214 Do not pre-allocate space.
1217 Use a platform's native pre-allocation call but fall back to
1218 **none** behavior if it fails/is not implemented.
1221 Pre-allocate via :manpage:`posix_fallocate(3)`.
1224 Pre-allocate via :manpage:`fallocate(2)` with
1225 FALLOC_FL_KEEP_SIZE set.
1228 Extend file to final size via :manpage:`ftruncate(2)`
1229 instead of allocating.
1232 Backward-compatible alias for **none**.
1235 Backward-compatible alias for **posix**.
1237 May not be available on all supported platforms. **keep** is only available
1238 on Linux. If using ZFS on Solaris this cannot be set to **posix**
1239 because ZFS doesn't support pre-allocation. Default: **native** if any
1240 pre-allocation methods except **truncate** are available, **none** if not.
1242 Note that using **truncate** on Windows will interact surprisingly
1243 with non-sequential write patterns. When writing to a file that has
1244 been extended by setting the end-of-file information, Windows will
1245 backfill the unwritten portion of the file up to that offset with
1246 zeroes before issuing the new write. This means that a single small
1247 write to the end of an extended file will stall until the entire
1248 file has been filled with zeroes.
1250 .. option:: fadvise_hint=str
1252 Use :manpage:`posix_fadvise(2)` or :manpage:`posix_fadvise(2)` to
1253 advise the kernel on what I/O patterns are likely to be issued.
1254 Accepted values are:
1257 Backwards-compatible hint for "no hint".
1260 Backwards compatible hint for "advise with fio workload type". This
1261 uses **FADV_RANDOM** for a random workload, and **FADV_SEQUENTIAL**
1262 for a sequential workload.
1265 Advise using **FADV_SEQUENTIAL**.
1268 Advise using **FADV_RANDOM**.
1270 .. option:: write_hint=str
1272 Use :manpage:`fcntl(2)` to advise the kernel what life time to expect
1273 from a write. Only supported on Linux, as of version 4.13. Accepted
1277 No particular life time associated with this file.
1280 Data written to this file has a short life time.
1283 Data written to this file has a medium life time.
1286 Data written to this file has a long life time.
1289 Data written to this file has a very long life time.
1291 The values are all relative to each other, and no absolute meaning
1292 should be associated with them.
1294 .. option:: offset=int
1296 Start I/O at the provided offset in the file, given as either a fixed size in
1297 bytes, zones or a percentage. If a percentage is given, the generated offset will be
1298 aligned to the minimum ``blocksize`` or to the value of ``offset_align`` if
1299 provided. Data before the given offset will not be touched. This
1300 effectively caps the file size at `real_size - offset`. Can be combined with
1301 :option:`size` to constrain the start and end range of the I/O workload.
1302 A percentage can be specified by a number between 1 and 100 followed by '%',
1303 for example, ``offset=20%`` to specify 20%. In ZBD mode, value can be set as
1304 number of zones using 'z'.
1306 .. option:: offset_align=int
1308 If set to non-zero value, the byte offset generated by a percentage ``offset``
1309 is aligned upwards to this value. Defaults to 0 meaning that a percentage
1310 offset is aligned to the minimum block size.
1312 .. option:: offset_increment=int
1314 If this is provided, then the real offset becomes `offset + offset_increment
1315 * thread_number`, where the thread number is a counter that starts at 0 and
1316 is incremented for each sub-job (i.e. when :option:`numjobs` option is
1317 specified). This option is useful if there are several jobs which are
1318 intended to operate on a file in parallel disjoint segments, with even
1319 spacing between the starting points. Percentages can be used for this option.
1320 If a percentage is given, the generated offset will be aligned to the minimum
1321 ``blocksize`` or to the value of ``offset_align`` if provided. In ZBD mode, value can
1322 also be set as number of zones using 'z'.
1324 .. option:: number_ios=int
1326 Fio will normally perform I/Os until it has exhausted the size of the region
1327 set by :option:`size`, or if it exhaust the allocated time (or hits an error
1328 condition). With this setting, the range/size can be set independently of
1329 the number of I/Os to perform. When fio reaches this number, it will exit
1330 normally and report status. Note that this does not extend the amount of I/O
1331 that will be done, it will only stop fio if this condition is met before
1332 other end-of-job criteria.
1334 .. option:: fsync=int
1336 If writing to a file, issue an :manpage:`fsync(2)` (or its equivalent) of
1337 the dirty data for every number of blocks given. For example, if you give 32
1338 as a parameter, fio will sync the file after every 32 writes issued. If fio is
1339 using non-buffered I/O, we may not sync the file. The exception is the sg
1340 I/O engine, which synchronizes the disk cache anyway. Defaults to 0, which
1341 means fio does not periodically issue and wait for a sync to complete. Also
1342 see :option:`end_fsync` and :option:`fsync_on_close`.
1344 .. option:: fdatasync=int
1346 Like :option:`fsync` but uses :manpage:`fdatasync(2)` to only sync data and
1347 not metadata blocks. In Windows, FreeBSD, DragonFlyBSD or OSX there is no
1348 :manpage:`fdatasync(2)` so this falls back to using :manpage:`fsync(2)`.
1349 Defaults to 0, which means fio does not periodically issue and wait for a
1350 data-only sync to complete.
1352 .. option:: write_barrier=int
1354 Make every `N-th` write a barrier write.
1356 .. option:: sync_file_range=str:int
1358 Use :manpage:`sync_file_range(2)` for every `int` number of write
1359 operations. Fio will track range of writes that have happened since the last
1360 :manpage:`sync_file_range(2)` call. `str` can currently be one or more of:
1363 SYNC_FILE_RANGE_WAIT_BEFORE
1365 SYNC_FILE_RANGE_WRITE
1367 SYNC_FILE_RANGE_WAIT_AFTER
1369 So if you do ``sync_file_range=wait_before,write:8``, fio would use
1370 ``SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE`` for every 8
1371 writes. Also see the :manpage:`sync_file_range(2)` man page. This option is
1374 .. option:: overwrite=bool
1376 If true, writes to a file will always overwrite existing data. If the file
1377 doesn't already exist, it will be created before the write phase begins. If
1378 the file exists and is large enough for the specified write phase, nothing
1379 will be done. Default: false.
1381 .. option:: end_fsync=bool
1383 If true, :manpage:`fsync(2)` file contents when a write stage has completed.
1386 .. option:: fsync_on_close=bool
1388 If true, fio will :manpage:`fsync(2)` a dirty file on close. This differs
1389 from :option:`end_fsync` in that it will happen on every file close, not
1390 just at the end of the job. Default: false.
1392 .. option:: rwmixread=int
1394 Percentage of a mixed workload that should be reads. Default: 50.
1396 .. option:: rwmixwrite=int
1398 Percentage of a mixed workload that should be writes. If both
1399 :option:`rwmixread` and :option:`rwmixwrite` is given and the values do not
1400 add up to 100%, the latter of the two will be used to override the
1401 first. This may interfere with a given rate setting, if fio is asked to
1402 limit reads or writes to a certain rate. If that is the case, then the
1403 distribution may be skewed. Default: 50.
1405 .. option:: random_distribution=str:float[:float][,str:float][,str:float]
1407 By default, fio will use a completely uniform random distribution when asked
1408 to perform random I/O. Sometimes it is useful to skew the distribution in
1409 specific ways, ensuring that some parts of the data is more hot than others.
1410 fio includes the following distribution models:
1413 Uniform random distribution
1422 Normal (Gaussian) distribution
1425 Zoned random distribution
1428 Zone absolute random distribution
1430 When using a **zipf** or **pareto** distribution, an input value is also
1431 needed to define the access pattern. For **zipf**, this is the `Zipf
1432 theta`. For **pareto**, it's the `Pareto power`. Fio includes a test
1433 program, :command:`fio-genzipf`, that can be used visualize what the given input
1434 values will yield in terms of hit rates. If you wanted to use **zipf** with
1435 a `theta` of 1.2, you would use ``random_distribution=zipf:1.2`` as the
1436 option. If a non-uniform model is used, fio will disable use of the random
1437 map. For the **normal** distribution, a normal (Gaussian) deviation is
1438 supplied as a value between 0 and 100.
1440 The second, optional float is allowed for **pareto**, **zipf** and **normal** distributions.
1441 It allows to set base of distribution in non-default place, giving more control
1442 over most probable outcome. This value is in range [0-1] which maps linearly to
1443 range of possible random values.
1444 Defaults are: random for **pareto** and **zipf**, and 0.5 for **normal**.
1445 If you wanted to use **zipf** with a `theta` of 1.2 centered on 1/4 of allowed value range,
1446 you would use ``random_distibution=zipf:1.2:0.25``.
1448 For a **zoned** distribution, fio supports specifying percentages of I/O
1449 access that should fall within what range of the file or device. For
1450 example, given a criteria of:
1452 * 60% of accesses should be to the first 10%
1453 * 30% of accesses should be to the next 20%
1454 * 8% of accesses should be to the next 30%
1455 * 2% of accesses should be to the next 40%
1457 we can define that through zoning of the random accesses. For the above
1458 example, the user would do::
1460 random_distribution=zoned:60/10:30/20:8/30:2/40
1462 A **zoned_abs** distribution works exactly like the **zoned**, except
1463 that it takes absolute sizes. For example, let's say you wanted to
1464 define access according to the following criteria:
1466 * 60% of accesses should be to the first 20G
1467 * 30% of accesses should be to the next 100G
1468 * 10% of accesses should be to the next 500G
1470 we can define an absolute zoning distribution with:
1472 random_distribution=zoned_abs=60/20G:30/100G:10/500g
1474 For both **zoned** and **zoned_abs**, fio supports defining up to
1477 Similarly to how :option:`bssplit` works for setting ranges and
1478 percentages of block sizes. Like :option:`bssplit`, it's possible to
1479 specify separate zones for reads, writes, and trims. If just one set
1480 is given, it'll apply to all of them. This goes for both **zoned**
1481 **zoned_abs** distributions.
1483 .. option:: percentage_random=int[,int][,int]
1485 For a random workload, set how big a percentage should be random. This
1486 defaults to 100%, in which case the workload is fully random. It can be set
1487 from anywhere from 0 to 100. Setting it to 0 would make the workload fully
1488 sequential. Any setting in between will result in a random mix of sequential
1489 and random I/O, at the given percentages. Comma-separated values may be
1490 specified for reads, writes, and trims as described in :option:`blocksize`.
1492 .. option:: norandommap
1494 Normally fio will cover every block of the file when doing random I/O. If
1495 this option is given, fio will just get a new random offset without looking
1496 at past I/O history. This means that some blocks may not be read or written,
1497 and that some blocks may be read/written more than once. If this option is
1498 used with :option:`verify` and multiple blocksizes (via :option:`bsrange`),
1499 only intact blocks are verified, i.e., partially-overwritten blocks are
1500 ignored. With an async I/O engine and an I/O depth > 1, it is possible for
1501 the same block to be overwritten, which can cause verification errors. Either
1502 do not use norandommap in this case, or also use the lfsr random generator.
1504 .. option:: softrandommap=bool
1506 See :option:`norandommap`. If fio runs with the random block map enabled and
1507 it fails to allocate the map, if this option is set it will continue without
1508 a random block map. As coverage will not be as complete as with random maps,
1509 this option is disabled by default.
1511 .. option:: random_generator=str
1513 Fio supports the following engines for generating I/O offsets for random I/O:
1516 Strong 2^88 cycle random number generator.
1518 Linear feedback shift register generator.
1520 Strong 64-bit 2^258 cycle random number generator.
1522 **tausworthe** is a strong random number generator, but it requires tracking
1523 on the side if we want to ensure that blocks are only read or written
1524 once. **lfsr** guarantees that we never generate the same offset twice, and
1525 it's also less computationally expensive. It's not a true random generator,
1526 however, though for I/O purposes it's typically good enough. **lfsr** only
1527 works with single block sizes, not with workloads that use multiple block
1528 sizes. If used with such a workload, fio may read or write some blocks
1529 multiple times. The default value is **tausworthe**, unless the required
1530 space exceeds 2^32 blocks. If it does, then **tausworthe64** is
1531 selected automatically.
1537 .. option:: blocksize=int[,int][,int], bs=int[,int][,int]
1539 The block size in bytes used for I/O units. Default: 4096. A single value
1540 applies to reads, writes, and trims. Comma-separated values may be
1541 specified for reads, writes, and trims. A value not terminated in a comma
1542 applies to subsequent types.
1547 means 256k for reads, writes and trims.
1550 means 8k for reads, 32k for writes and trims.
1553 means 8k for reads, 32k for writes, and default for trims.
1556 means default for reads, 8k for writes and trims.
1559 means default for reads, 8k for writes, and default for trims.
1561 .. option:: blocksize_range=irange[,irange][,irange], bsrange=irange[,irange][,irange]
1563 A range of block sizes in bytes for I/O units. The issued I/O unit will
1564 always be a multiple of the minimum size, unless
1565 :option:`blocksize_unaligned` is set.
1567 Comma-separated ranges may be specified for reads, writes, and trims as
1568 described in :option:`blocksize`.
1570 Example: ``bsrange=1k-4k,2k-8k``.
1572 .. option:: bssplit=str[,str][,str]
1574 Sometimes you want even finer grained control of the block sizes
1575 issued, not just an even split between them. This option allows you to
1576 weight various block sizes, so that you are able to define a specific
1577 amount of block sizes issued. The format for this option is::
1579 bssplit=blocksize/percentage:blocksize/percentage
1581 for as many block sizes as needed. So if you want to define a workload
1582 that has 50% 64k blocks, 10% 4k blocks, and 40% 32k blocks, you would
1585 bssplit=4k/10:64k/50:32k/40
1587 Ordering does not matter. If the percentage is left blank, fio will
1588 fill in the remaining values evenly. So a bssplit option like this one::
1590 bssplit=4k/50:1k/:32k/
1592 would have 50% 4k ios, and 25% 1k and 32k ios. The percentages always
1593 add up to 100, if bssplit is given a range that adds up to more, it
1596 Comma-separated values may be specified for reads, writes, and trims as
1597 described in :option:`blocksize`.
1599 If you want a workload that has 50% 2k reads and 50% 4k reads, while
1600 having 90% 4k writes and 10% 8k writes, you would specify::
1602 bssplit=2k/50:4k/50,4k/90:8k/10
1604 Fio supports defining up to 64 different weights for each data
1607 .. option:: blocksize_unaligned, bs_unaligned
1609 If set, fio will issue I/O units with any size within
1610 :option:`blocksize_range`, not just multiples of the minimum size. This
1611 typically won't work with direct I/O, as that normally requires sector
1614 .. option:: bs_is_seq_rand=bool
1616 If this option is set, fio will use the normal read,write blocksize settings
1617 as sequential,random blocksize settings instead. Any random read or write
1618 will use the WRITE blocksize settings, and any sequential read or write will
1619 use the READ blocksize settings.
1621 .. option:: blockalign=int[,int][,int], ba=int[,int][,int]
1623 Boundary to which fio will align random I/O units. Default:
1624 :option:`blocksize`. Minimum alignment is typically 512b for using direct
1625 I/O, though it usually depends on the hardware block size. This option is
1626 mutually exclusive with using a random map for files, so it will turn off
1627 that option. Comma-separated values may be specified for reads, writes, and
1628 trims as described in :option:`blocksize`.
1634 .. option:: zero_buffers
1636 Initialize buffers with all zeros. Default: fill buffers with random data.
1638 .. option:: refill_buffers
1640 If this option is given, fio will refill the I/O buffers on every
1641 submit. Only makes sense if :option:`zero_buffers` isn't specified,
1642 naturally. Defaults to being unset i.e., the buffer is only filled at
1643 init time and the data in it is reused when possible but if any of
1644 :option:`verify`, :option:`buffer_compress_percentage` or
1645 :option:`dedupe_percentage` are enabled then `refill_buffers` is also
1646 automatically enabled.
1648 .. option:: scramble_buffers=bool
1650 If :option:`refill_buffers` is too costly and the target is using data
1651 deduplication, then setting this option will slightly modify the I/O buffer
1652 contents to defeat normal de-dupe attempts. This is not enough to defeat
1653 more clever block compression attempts, but it will stop naive dedupe of
1654 blocks. Default: true.
1656 .. option:: buffer_compress_percentage=int
1658 If this is set, then fio will attempt to provide I/O buffer content
1659 (on WRITEs) that compresses to the specified level. Fio does this by
1660 providing a mix of random data followed by fixed pattern data. The
1661 fixed pattern is either zeros, or the pattern specified by
1662 :option:`buffer_pattern`. If the `buffer_pattern` option is used, it
1663 might skew the compression ratio slightly. Setting
1664 `buffer_compress_percentage` to a value other than 100 will also
1665 enable :option:`refill_buffers` in order to reduce the likelihood that
1666 adjacent blocks are so similar that they over compress when seen
1667 together. See :option:`buffer_compress_chunk` for how to set a finer or
1668 coarser granularity for the random/fixed data region. Defaults to unset
1669 i.e., buffer data will not adhere to any compression level.
1671 .. option:: buffer_compress_chunk=int
1673 This setting allows fio to manage how big the random/fixed data region
1674 is when using :option:`buffer_compress_percentage`. When
1675 `buffer_compress_chunk` is set to some non-zero value smaller than the
1676 block size, fio can repeat the random/fixed region throughout the I/O
1677 buffer at the specified interval (which particularly useful when
1678 bigger block sizes are used for a job). When set to 0, fio will use a
1679 chunk size that matches the block size resulting in a single
1680 random/fixed region within the I/O buffer. Defaults to 512. When the
1681 unit is omitted, the value is interpreted in bytes.
1683 .. option:: buffer_pattern=str
1685 If set, fio will fill the I/O buffers with this pattern or with the contents
1686 of a file. If not set, the contents of I/O buffers are defined by the other
1687 options related to buffer contents. The setting can be any pattern of bytes,
1688 and can be prefixed with 0x for hex values. It may also be a string, where
1689 the string must then be wrapped with ``""``. Or it may also be a filename,
1690 where the filename must be wrapped with ``''`` in which case the file is
1691 opened and read. Note that not all the file contents will be read if that
1692 would cause the buffers to overflow. So, for example::
1694 buffer_pattern='filename'
1698 buffer_pattern="abcd"
1706 buffer_pattern=0xdeadface
1708 Also you can combine everything together in any order::
1710 buffer_pattern=0xdeadface"abcd"-12'filename'
1712 .. option:: dedupe_percentage=int
1714 If set, fio will generate this percentage of identical buffers when
1715 writing. These buffers will be naturally dedupable. The contents of the
1716 buffers depend on what other buffer compression settings have been set. It's
1717 possible to have the individual buffers either fully compressible, or not at
1718 all -- this option only controls the distribution of unique buffers. Setting
1719 this option will also enable :option:`refill_buffers` to prevent every buffer
1722 .. option:: dedupe_mode=str
1724 If ``dedupe_percentage=<int>`` is given, then this option controls how fio
1725 generates the dedupe buffers.
1728 Generate dedupe buffers by repeating previous writes
1730 Generate dedupe buffers from working set
1732 ``repeat`` is the default option for fio. Dedupe buffers are generated
1733 by repeating previous unique write.
1735 ``working_set`` is a more realistic workload.
1736 With ``working_set``, ``dedupe_working_set_percentage=<int>`` should be provided.
1737 Given that, fio will use the initial unique write buffers as its working set.
1738 Upon deciding to dedupe, fio will randomly choose a buffer from the working set.
1739 Note that by using ``working_set`` the dedupe percentage will converge
1740 to the desired over time while ``repeat`` maintains the desired percentage
1743 .. option:: dedupe_working_set_percentage=int
1745 If ``dedupe_mode=<str>`` is set to ``working_set``, then this controls
1746 the percentage of size of the file or device used as the buffers
1747 fio will choose to generate the dedupe buffers from
1749 Note that size needs to be explicitly provided and only 1 file per
1752 .. option:: invalidate=bool
1754 Invalidate the buffer/page cache parts of the files to be used prior to
1755 starting I/O if the platform and file type support it. Defaults to true.
1756 This will be ignored if :option:`pre_read` is also specified for the
1759 .. option:: sync=str
1761 Whether, and what type, of synchronous I/O to use for writes. The allowed
1765 Do not use synchronous IO, the default.
1771 Use synchronous file IO. For the majority of I/O engines,
1772 this means using O_SYNC.
1778 Use synchronous data IO. For the majority of I/O engines,
1779 this means using O_DSYNC.
1782 .. option:: iomem=str, mem=str
1784 Fio can use various types of memory as the I/O unit buffer. The allowed
1788 Use memory from :manpage:`malloc(3)` as the buffers. Default memory
1792 Use shared memory as the buffers. Allocated through
1793 :manpage:`shmget(2)`.
1796 Same as shm, but use huge pages as backing.
1799 Use :manpage:`mmap(2)` to allocate buffers. May either be anonymous memory, or can
1800 be file backed if a filename is given after the option. The format
1801 is `mem=mmap:/path/to/file`.
1804 Use a memory mapped huge file as the buffer backing. Append filename
1805 after mmaphuge, ala `mem=mmaphuge:/hugetlbfs/file`.
1808 Same as mmap, but use a MMAP_SHARED mapping.
1811 Use GPU memory as the buffers for GPUDirect RDMA benchmark.
1812 The :option:`ioengine` must be `rdma`.
1814 The area allocated is a function of the maximum allowed bs size for the job,
1815 multiplied by the I/O depth given. Note that for **shmhuge** and
1816 **mmaphuge** to work, the system must have free huge pages allocated. This
1817 can normally be checked and set by reading/writing
1818 :file:`/proc/sys/vm/nr_hugepages` on a Linux system. Fio assumes a huge page
1819 is 4MiB in size. So to calculate the number of huge pages you need for a
1820 given job file, add up the I/O depth of all jobs (normally one unless
1821 :option:`iodepth` is used) and multiply by the maximum bs set. Then divide
1822 that number by the huge page size. You can see the size of the huge pages in
1823 :file:`/proc/meminfo`. If no huge pages are allocated by having a non-zero
1824 number in `nr_hugepages`, using **mmaphuge** or **shmhuge** will fail. Also
1825 see :option:`hugepage-size`.
1827 **mmaphuge** also needs to have hugetlbfs mounted and the file location
1828 should point there. So if it's mounted in :file:`/huge`, you would use
1829 `mem=mmaphuge:/huge/somefile`.
1831 .. option:: iomem_align=int, mem_align=int
1833 This indicates the memory alignment of the I/O memory buffers. Note that
1834 the given alignment is applied to the first I/O unit buffer, if using
1835 :option:`iodepth` the alignment of the following buffers are given by the
1836 :option:`bs` used. In other words, if using a :option:`bs` that is a
1837 multiple of the page sized in the system, all buffers will be aligned to
1838 this value. If using a :option:`bs` that is not page aligned, the alignment
1839 of subsequent I/O memory buffers is the sum of the :option:`iomem_align` and
1842 .. option:: hugepage-size=int
1844 Defines the size of a huge page. Must at least be equal to the system
1845 setting, see :file:`/proc/meminfo`. Defaults to 4MiB. Should probably
1846 always be a multiple of megabytes, so using ``hugepage-size=Xm`` is the
1847 preferred way to set this to avoid setting a non-pow-2 bad value.
1849 .. option:: lockmem=int
1851 Pin the specified amount of memory with :manpage:`mlock(2)`. Can be used to
1852 simulate a smaller amount of memory. The amount specified is per worker.
1858 .. option:: size=int
1860 The total size of file I/O for each thread of this job. Fio will run until
1861 this many bytes has been transferred, unless runtime is limited by other options
1862 (such as :option:`runtime`, for instance, or increased/decreased by :option:`io_size`).
1863 Fio will divide this size between the available files determined by options
1864 such as :option:`nrfiles`, :option:`filename`, unless :option:`filesize` is
1865 specified by the job. If the result of division happens to be 0, the size is
1866 set to the physical size of the given files or devices if they exist.
1867 If this option is not specified, fio will use the full size of the given
1868 files or devices. If the files do not exist, size must be given. It is also
1869 possible to give size as a percentage between 1 and 100. If ``size=20%`` is
1870 given, fio will use 20% of the full size of the given files or devices.
1871 In ZBD mode, value can also be set as number of zones using 'z'.
1872 Can be combined with :option:`offset` to constrain the start and end range
1873 that I/O will be done within.
1875 .. option:: io_size=int, io_limit=int
1877 Normally fio operates within the region set by :option:`size`, which means
1878 that the :option:`size` option sets both the region and size of I/O to be
1879 performed. Sometimes that is not what you want. With this option, it is
1880 possible to define just the amount of I/O that fio should do. For instance,
1881 if :option:`size` is set to 20GiB and :option:`io_size` is set to 5GiB, fio
1882 will perform I/O within the first 20GiB but exit when 5GiB have been
1883 done. The opposite is also possible -- if :option:`size` is set to 20GiB,
1884 and :option:`io_size` is set to 40GiB, then fio will do 40GiB of I/O within
1885 the 0..20GiB region.
1887 .. option:: filesize=irange(int)
1889 Individual file sizes. May be a range, in which case fio will select sizes
1890 for files at random within the given range and limited to :option:`size` in
1891 total (if that is given). If not given, each created file is the same size.
1892 This option overrides :option:`size` in terms of file size, which means
1893 this value is used as a fixed size or possible range of each file.
1895 .. option:: file_append=bool
1897 Perform I/O after the end of the file. Normally fio will operate within the
1898 size of a file. If this option is set, then fio will append to the file
1899 instead. This has identical behavior to setting :option:`offset` to the size
1900 of a file. This option is ignored on non-regular files.
1902 .. option:: fill_device=bool, fill_fs=bool
1904 Sets size to something really large and waits for ENOSPC (no space left on
1905 device) or EDQUOT (disk quota exceeded)
1906 as the terminating condition. Only makes sense with sequential
1907 write. For a read workload, the mount point will be filled first then I/O
1908 started on the result. This option doesn't make sense if operating on a raw
1909 device node, since the size of that is already known by the file system.
1910 Additionally, writing beyond end-of-device will not return ENOSPC there.
1916 .. option:: ioengine=str
1918 Defines how the job issues I/O to the file. The following types are defined:
1921 Basic :manpage:`read(2)` or :manpage:`write(2)`
1922 I/O. :manpage:`lseek(2)` is used to position the I/O location.
1923 See :option:`fsync` and :option:`fdatasync` for syncing write I/Os.
1926 Basic :manpage:`pread(2)` or :manpage:`pwrite(2)` I/O. Default on
1927 all supported operating systems except for Windows.
1930 Basic :manpage:`readv(2)` or :manpage:`writev(2)` I/O. Will emulate
1931 queuing by coalescing adjacent I/Os into a single submission.
1934 Basic :manpage:`preadv(2)` or :manpage:`pwritev(2)` I/O.
1937 Basic :manpage:`preadv2(2)` or :manpage:`pwritev2(2)` I/O.
1940 Fast Linux native asynchronous I/O. Supports async IO
1941 for both direct and buffered IO.
1942 This engine defines engine specific options.
1945 Linux native asynchronous I/O. Note that Linux may only support
1946 queued behavior with non-buffered I/O (set ``direct=1`` or
1948 This engine defines engine specific options.
1951 POSIX asynchronous I/O using :manpage:`aio_read(3)` and
1952 :manpage:`aio_write(3)`.
1955 Solaris native asynchronous I/O.
1958 Windows native asynchronous I/O. Default on Windows.
1961 File is memory mapped with :manpage:`mmap(2)` and data copied
1962 to/from using :manpage:`memcpy(3)`.
1965 :manpage:`splice(2)` is used to transfer the data and
1966 :manpage:`vmsplice(2)` to transfer data from user space to the
1970 SCSI generic sg v3 I/O. May either be synchronous using the SG_IO
1971 ioctl, or if the target is an sg character device we use
1972 :manpage:`read(2)` and :manpage:`write(2)` for asynchronous
1973 I/O. Requires :option:`filename` option to specify either block or
1974 character devices. This engine supports trim operations.
1975 The sg engine includes engine specific options.
1978 Read, write, trim and ZBC/ZAC operations to a zoned
1979 block device using libzbc library. The target can be
1980 either an SG character device or a block device file.
1983 Doesn't transfer any data, just pretends to. This is mainly used to
1984 exercise fio itself and for debugging/testing purposes.
1987 Transfer over the network to given ``host:port``. Depending on the
1988 :option:`protocol` used, the :option:`hostname`, :option:`port`,
1989 :option:`listen` and :option:`filename` options are used to specify
1990 what sort of connection to make, while the :option:`protocol` option
1991 determines which protocol will be used. This engine defines engine
1995 Like **net**, but uses :manpage:`splice(2)` and
1996 :manpage:`vmsplice(2)` to map data and send/receive.
1997 This engine defines engine specific options.
2000 Doesn't transfer any data, but burns CPU cycles according to the
2001 :option:`cpuload`, :option:`cpuchunks` and :option:`cpumode` options.
2002 Setting :option:`cpuload`\=85 will cause that job to do nothing but burn 85%
2003 of the CPU. In case of SMP machines, use :option:`numjobs`\=<nr_of_cpu>
2004 to get desired CPU usage, as the cpuload only loads a
2005 single CPU at the desired rate. A job never finishes unless there is
2006 at least one non-cpuio job.
2007 Setting :option:`cpumode`\=qsort replace the default noop instructions loop
2008 by a qsort algorithm to consume more energy.
2011 The RDMA I/O engine supports both RDMA memory semantics
2012 (RDMA_WRITE/RDMA_READ) and channel semantics (Send/Recv) for the
2013 InfiniBand, RoCE and iWARP protocols. This engine defines engine
2017 I/O engine that does regular fallocate to simulate data transfer as
2021 does fallocate(,mode = FALLOC_FL_KEEP_SIZE,).
2024 does fallocate(,mode = 0).
2027 does fallocate(,mode = FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE).
2030 I/O engine that sends :manpage:`ftruncate(2)` operations in response
2031 to write (DDIR_WRITE) events. Each ftruncate issued sets the file's
2032 size to the current block offset. :option:`blocksize` is ignored.
2035 I/O engine that does regular EXT4_IOC_MOVE_EXT ioctls to simulate
2036 defragment activity in request to DDIR_WRITE event.
2039 I/O engine supporting direct access to Ceph Reliable Autonomic
2040 Distributed Object Store (RADOS) via librados. This ioengine
2041 defines engine specific options.
2044 I/O engine supporting direct access to Ceph Rados Block Devices
2045 (RBD) via librbd without the need to use the kernel rbd driver. This
2046 ioengine defines engine specific options.
2049 I/O engine supporting GET/PUT requests over HTTP(S) with libcurl to
2050 a WebDAV or S3 endpoint. This ioengine defines engine specific options.
2052 This engine only supports direct IO of iodepth=1; you need to scale this
2053 via numjobs. blocksize defines the size of the objects to be created.
2055 TRIM is translated to object deletion.
2058 Using GlusterFS libgfapi sync interface to direct access to
2059 GlusterFS volumes without having to go through FUSE. This ioengine
2060 defines engine specific options.
2063 Using GlusterFS libgfapi async interface to direct access to
2064 GlusterFS volumes without having to go through FUSE. This ioengine
2065 defines engine specific options.
2068 Read and write through Hadoop (HDFS). The :option:`filename` option
2069 is used to specify host,port of the hdfs name-node to connect. This
2070 engine interprets offsets a little differently. In HDFS, files once
2071 created cannot be modified so random writes are not possible. To
2072 imitate this the libhdfs engine expects a bunch of small files to be
2073 created over HDFS and will randomly pick a file from them
2074 based on the offset generated by fio backend (see the example
2075 job file to create such files, use ``rw=write`` option). Please
2076 note, it may be necessary to set environment variables to work
2077 with HDFS/libhdfs properly. Each job uses its own connection to
2081 Read, write and erase an MTD character device (e.g.,
2082 :file:`/dev/mtd0`). Discards are treated as erases. Depending on the
2083 underlying device type, the I/O may have to go in a certain pattern,
2084 e.g., on NAND, writing sequentially to erase blocks and discarding
2085 before overwriting. The `trimwrite` mode works well for this
2089 Read and write using filesystem DAX to a file on a filesystem
2090 mounted with DAX on a persistent memory device through the PMDK
2094 Read and write using device DAX to a persistent memory device (e.g.,
2095 /dev/dax0.0) through the PMDK libpmem library.
2098 Prefix to specify loading an external I/O engine object file. Append
2099 the engine filename, e.g. ``ioengine=external:/tmp/foo.o`` to load
2100 ioengine :file:`foo.o` in :file:`/tmp`. The path can be either
2101 absolute or relative. See :file:`engines/skeleton_external.c` for
2102 details of writing an external I/O engine.
2105 Simply create the files and do no I/O to them. You still need to
2106 set `filesize` so that all the accounting still occurs, but no
2107 actual I/O will be done other than creating the file.
2110 Simply do stat() and do no I/O to the file. You need to set 'filesize'
2111 and 'nrfiles', so that files will be created.
2112 This engine is to measure file lookup and meta data access.
2115 Simply delete the files by unlink() and do no I/O to them. You need to set 'filesize'
2116 and 'nrfiles', so that the files will be created.
2117 This engine is to measure file delete.
2120 Read and write using mmap I/O to a file on a filesystem
2121 mounted with DAX on a persistent memory device through the PMDK
2125 Synchronous read and write using DDN's Infinite Memory Engine (IME).
2126 This engine is very basic and issues calls to IME whenever an IO is
2130 Synchronous read and write using DDN's Infinite Memory Engine (IME).
2131 This engine uses iovecs and will try to stack as much IOs as possible
2132 (if the IOs are "contiguous" and the IO depth is not exceeded)
2133 before issuing a call to IME.
2136 Asynchronous read and write using DDN's Infinite Memory Engine (IME).
2137 This engine will try to stack as much IOs as possible by creating
2138 requests for IME. FIO will then decide when to commit these requests.
2140 Read and write iscsi lun with libiscsi.
2142 Read and write a Network Block Device (NBD).
2145 I/O engine supporting libcufile synchronous access to nvidia-fs and a
2146 GPUDirect Storage-supported filesystem. This engine performs
2147 I/O without transferring buffers between user-space and the kernel,
2148 unless :option:`verify` is set or :option:`cuda_io` is `posix`.
2149 :option:`iomem` must not be `cudamalloc`. This ioengine defines
2150 engine specific options.
2152 I/O engine supporting asynchronous read and write operations to the
2153 DAOS File System (DFS) via libdfs.
2156 I/O engine supporting asynchronous read and write operations to
2157 NFS filesystems from userspace via libnfs. This is useful for
2158 achieving higher concurrency and thus throughput than is possible
2162 Execute 3rd party tools. Could be used to perform monitoring during jobs runtime.
2164 I/O engine specific parameters
2165 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2167 In addition, there are some parameters which are only valid when a specific
2168 :option:`ioengine` is in use. These are used identically to normal parameters,
2169 with the caveat that when used on the command line, they must come after the
2170 :option:`ioengine` that defines them is selected.
2172 .. option:: cmdprio_percentage=int[,int] : [io_uring] [libaio]
2174 Set the percentage of I/O that will be issued with the highest priority.
2175 Default: 0. A single value applies to reads and writes. Comma-separated
2176 values may be specified for reads and writes. For this option to be
2177 effective, NCQ priority must be supported and enabled, and `direct=1'
2178 option must be used. fio must also be run as the root user. Unlike
2179 slat/clat/lat stats, which can be tracked and reported independently, per
2180 priority stats only track and report a single type of latency. By default,
2181 completion latency (clat) will be reported, if :option:`lat_percentiles` is
2182 set, total latency (lat) will be reported.
2184 .. option:: cmdprio_class=int[,int] : [io_uring] [libaio]
2186 Set the I/O priority class to use for I/Os that must be issued with
2187 a priority when :option:`cmdprio_percentage` or
2188 :option:`cmdprio_bssplit` is set. If not specified when
2189 :option:`cmdprio_percentage` or :option:`cmdprio_bssplit` is set,
2190 this defaults to the highest priority class. A single value applies
2191 to reads and writes. Comma-separated values may be specified for
2192 reads and writes. See :manpage:`ionice(1)`. See also the
2193 :option:`prioclass` option.
2195 .. option:: cmdprio=int[,int] : [io_uring] [libaio]
2197 Set the I/O priority value to use for I/Os that must be issued with
2198 a priority when :option:`cmdprio_percentage` or
2199 :option:`cmdprio_bssplit` is set. If not specified when
2200 :option:`cmdprio_percentage` or :option:`cmdprio_bssplit` is set,
2202 Linux limits us to a positive value between 0 and 7, with 0 being the
2203 highest. A single value applies to reads and writes. Comma-separated
2204 values may be specified for reads and writes. See :manpage:`ionice(1)`.
2205 Refer to an appropriate manpage for other operating systems since
2206 meaning of priority may differ. See also the :option:`prio` option.
2208 .. option:: cmdprio_bssplit=str[,str] : [io_uring] [libaio]
2209 To get a finer control over I/O priority, this option allows
2210 specifying the percentage of IOs that must have a priority set
2211 depending on the block size of the IO. This option is useful only
2212 when used together with the :option:`bssplit` option, that is,
2213 multiple different block sizes are used for reads and writes.
2214 The format for this option is the same as the format of the
2215 :option:`bssplit` option, with the exception that values for
2216 trim IOs are ignored. This option is mutually exclusive with the
2217 :option:`cmdprio_percentage` option.
2219 .. option:: fixedbufs : [io_uring]
2221 If fio is asked to do direct IO, then Linux will map pages for each
2222 IO call, and release them when IO is done. If this option is set, the
2223 pages are pre-mapped before IO is started. This eliminates the need to
2224 map and release for each IO. This is more efficient, and reduces the
2227 .. option:: hipri : [io_uring]
2229 If this option is set, fio will attempt to use polled IO completions.
2230 Normal IO completions generate interrupts to signal the completion of
2231 IO, polled completions do not. Hence they are require active reaping
2232 by the application. The benefits are more efficient IO for high IOPS
2233 scenarios, and lower latencies for low queue depth IO.
2235 .. option:: registerfiles : [io_uring]
2237 With this option, fio registers the set of files being used with the
2238 kernel. This avoids the overhead of managing file counts in the kernel,
2239 making the submission and completion part more lightweight. Required
2240 for the below :option:`sqthread_poll` option.
2242 .. option:: sqthread_poll : [io_uring]
2244 Normally fio will submit IO by issuing a system call to notify the
2245 kernel of available items in the SQ ring. If this option is set, the
2246 act of submitting IO will be done by a polling thread in the kernel.
2247 This frees up cycles for fio, at the cost of using more CPU in the
2250 .. option:: sqthread_poll_cpu : [io_uring]
2252 When :option:`sqthread_poll` is set, this option provides a way to
2253 define which CPU should be used for the polling thread.
2255 .. option:: userspace_reap : [libaio]
2257 Normally, with the libaio engine in use, fio will use the
2258 :manpage:`io_getevents(2)` system call to reap newly returned events. With
2259 this flag turned on, the AIO ring will be read directly from user-space to
2260 reap events. The reaping mode is only enabled when polling for a minimum of
2261 0 events (e.g. when :option:`iodepth_batch_complete` `=0`).
2263 .. option:: hipri : [pvsync2]
2265 Set RWF_HIPRI on I/O, indicating to the kernel that it's of higher priority
2268 .. option:: hipri_percentage : [pvsync2]
2270 When hipri is set this determines the probability of a pvsync2 I/O being high
2271 priority. The default is 100%.
2273 .. option:: nowait : [pvsync2] [libaio] [io_uring]
2275 By default if a request cannot be executed immediately (e.g. resource starvation,
2276 waiting on locks) it is queued and the initiating process will be blocked until
2277 the required resource becomes free.
2279 This option sets the RWF_NOWAIT flag (supported from the 4.14 Linux kernel) and
2280 the call will return instantly with EAGAIN or a partial result rather than waiting.
2282 It is useful to also use ignore_error=EAGAIN when using this option.
2284 Note: glibc 2.27, 2.28 have a bug in syscall wrappers preadv2, pwritev2.
2285 They return EOPNOTSUP instead of EAGAIN.
2287 For cached I/O, using this option usually means a request operates only with
2288 cached data. Currently the RWF_NOWAIT flag does not supported for cached write.
2290 For direct I/O, requests will only succeed if cache invalidation isn't required,
2291 file blocks are fully allocated and the disk request could be issued immediately.
2293 .. option:: cpuload=int : [cpuio]
2295 Attempt to use the specified percentage of CPU cycles. This is a mandatory
2296 option when using cpuio I/O engine.
2298 .. option:: cpuchunks=int : [cpuio]
2300 Split the load into cycles of the given time. In microseconds.
2302 .. option:: exit_on_io_done=bool : [cpuio]
2304 Detect when I/O threads are done, then exit.
2306 .. option:: namenode=str : [libhdfs]
2308 The hostname or IP address of a HDFS cluster namenode to contact.
2310 .. option:: port=int
2314 The listening port of the HFDS cluster namenode.
2318 The TCP or UDP port to bind to or connect to. If this is used with
2319 :option:`numjobs` to spawn multiple instances of the same job type, then
2320 this will be the starting port number since fio will use a range of
2325 The port to use for RDMA-CM communication. This should be the same value
2326 on the client and the server side.
2328 .. option:: hostname=str : [netsplice] [net] [rdma]
2330 The hostname or IP address to use for TCP, UDP or RDMA-CM based I/O. If the job
2331 is a TCP listener or UDP reader, the hostname is not used and must be omitted
2332 unless it is a valid UDP multicast address.
2334 .. option:: serverip=str : [librpma_*]
2336 The IP address to be used for RDMA-CM based I/O.
2338 .. option:: direct_write_to_pmem=bool : [librpma_*]
2340 Set to 1 only when Direct Write to PMem from the remote host is possible.
2341 Otherwise, set to 0.
2343 .. option:: busy_wait_polling=bool : [librpma_*_server]
2345 Set to 0 to wait for completion instead of busy-wait polling completion.
2348 .. option:: interface=str : [netsplice] [net]
2350 The IP address of the network interface used to send or receive UDP
2353 .. option:: ttl=int : [netsplice] [net]
2355 Time-to-live value for outgoing UDP multicast packets. Default: 1.
2357 .. option:: nodelay=bool : [netsplice] [net]
2359 Set TCP_NODELAY on TCP connections.
2361 .. option:: protocol=str, proto=str : [netsplice] [net]
2363 The network protocol to use. Accepted values are:
2366 Transmission control protocol.
2368 Transmission control protocol V6.
2370 User datagram protocol.
2372 User datagram protocol V6.
2376 When the protocol is TCP or UDP, the port must also be given, as well as the
2377 hostname if the job is a TCP listener or UDP reader. For unix sockets, the
2378 normal :option:`filename` option should be used and the port is invalid.
2380 .. option:: listen : [netsplice] [net]
2382 For TCP network connections, tell fio to listen for incoming connections
2383 rather than initiating an outgoing connection. The :option:`hostname` must
2384 be omitted if this option is used.
2386 .. option:: pingpong : [netsplice] [net]
2388 Normally a network writer will just continue writing data, and a network
2389 reader will just consume packages. If ``pingpong=1`` is set, a writer will
2390 send its normal payload to the reader, then wait for the reader to send the
2391 same payload back. This allows fio to measure network latencies. The
2392 submission and completion latencies then measure local time spent sending or
2393 receiving, and the completion latency measures how long it took for the
2394 other end to receive and send back. For UDP multicast traffic
2395 ``pingpong=1`` should only be set for a single reader when multiple readers
2396 are listening to the same address.
2398 .. option:: window_size : [netsplice] [net]
2400 Set the desired socket buffer size for the connection.
2402 .. option:: mss : [netsplice] [net]
2404 Set the TCP maximum segment size (TCP_MAXSEG).
2406 .. option:: donorname=str : [e4defrag]
2408 File will be used as a block donor (swap extents between files).
2410 .. option:: inplace=int : [e4defrag]
2412 Configure donor file blocks allocation strategy:
2415 Default. Preallocate donor's file on init.
2417 Allocate space immediately inside defragment event, and free right
2420 .. option:: clustername=str : [rbd,rados]
2422 Specifies the name of the Ceph cluster.
2424 .. option:: rbdname=str : [rbd]
2426 Specifies the name of the RBD.
2428 .. option:: pool=str : [rbd,rados]
2430 Specifies the name of the Ceph pool containing RBD or RADOS data.
2432 .. option:: clientname=str : [rbd,rados]
2434 Specifies the username (without the 'client.' prefix) used to access the
2435 Ceph cluster. If the *clustername* is specified, the *clientname* shall be
2436 the full *type.id* string. If no type. prefix is given, fio will add
2437 'client.' by default.
2439 .. option:: busy_poll=bool : [rbd,rados]
2441 Poll store instead of waiting for completion. Usually this provides better
2442 throughput at cost of higher(up to 100%) CPU utilization.
2444 .. option:: touch_objects=bool : [rados]
2446 During initialization, touch (create if do not exist) all objects (files).
2447 Touching all objects affects ceph caches and likely impacts test results.
2450 .. option:: skip_bad=bool : [mtd]
2452 Skip operations against known bad blocks.
2454 .. option:: hdfsdirectory : [libhdfs]
2456 libhdfs will create chunk in this HDFS directory.
2458 .. option:: chunk_size : [libhdfs]
2460 The size of the chunk to use for each file.
2462 .. option:: verb=str : [rdma]
2464 The RDMA verb to use on this side of the RDMA ioengine connection. Valid
2465 values are write, read, send and recv. These correspond to the equivalent
2466 RDMA verbs (e.g. write = rdma_write etc.). Note that this only needs to be
2467 specified on the client side of the connection. See the examples folder.
2469 .. option:: bindname=str : [rdma]
2471 The name to use to bind the local RDMA-CM connection to a local RDMA device.
2472 This could be a hostname or an IPv4 or IPv6 address. On the server side this
2473 will be passed into the rdma_bind_addr() function and on the client site it
2474 will be used in the rdma_resolve_add() function. This can be useful when
2475 multiple paths exist between the client and the server or in certain loopback
2478 .. option:: stat_type=str : [filestat]
2480 Specify stat system call type to measure lookup/getattr performance.
2481 Default is **stat** for :manpage:`stat(2)`.
2483 .. option:: readfua=bool : [sg]
2485 With readfua option set to 1, read operations include
2486 the force unit access (fua) flag. Default is 0.
2488 .. option:: writefua=bool : [sg]
2490 With writefua option set to 1, write operations include
2491 the force unit access (fua) flag. Default is 0.
2493 .. option:: sg_write_mode=str : [sg]
2495 Specify the type of write commands to issue. This option can take three values:
2498 This is the default where write opcodes are issued as usual.
2499 **write_and_verify**
2500 Issue WRITE AND VERIFY commands. The BYTCHK bit is set to 0. This
2501 directs the device to carry out a medium verification with no data
2502 comparison. The writefua option is ignored with this selection.
2504 This option is deprecated. Use write_and_verify instead.
2506 Issue WRITE SAME commands. This transfers a single block to the device
2507 and writes this same block of data to a contiguous sequence of LBAs
2508 beginning at the specified offset. fio's block size parameter specifies
2509 the amount of data written with each command. However, the amount of data
2510 actually transferred to the device is equal to the device's block
2511 (sector) size. For a device with 512 byte sectors, blocksize=8k will
2512 write 16 sectors with each command. fio will still generate 8k of data
2513 for each command but only the first 512 bytes will be used and
2514 transferred to the device. The writefua option is ignored with this
2517 This option is deprecated. Use write_same instead.
2519 Issue WRITE SAME(16) commands as above but with the No Data Output
2520 Buffer (NDOB) bit set. No data will be transferred to the device with
2521 this bit set. Data written will be a pre-determined pattern such as
2523 **verify_bytchk_00**
2524 Issue VERIFY commands with BYTCHK set to 00. This directs the
2525 device to carry out a medium verification with no data comparison.
2526 **verify_bytchk_01**
2527 Issue VERIFY commands with BYTCHK set to 01. This directs the device to
2528 compare the data on the device with the data transferred to the device.
2529 **verify_bytchk_11**
2530 Issue VERIFY commands with BYTCHK set to 11. This transfers a
2531 single block to the device and compares the contents of this block with the
2532 data on the device beginning at the specified offset. fio's block size
2533 parameter specifies the total amount of data compared with this command.
2534 However, only one block (sector) worth of data is transferred to the device.
2535 This is similar to the WRITE SAME command except that data is compared instead
2538 .. option:: hipri : [sg]
2540 If this option is set, fio will attempt to use polled IO completions.
2541 This will have a similar effect as (io_uring)hipri. Only SCSI READ and
2542 WRITE commands will have the SGV4_FLAG_HIPRI set (not UNMAP (trim) nor
2543 VERIFY). Older versions of the Linux sg driver that do not support
2544 hipri will simply ignore this flag and do normal IO. The Linux SCSI
2545 Low Level Driver (LLD) that "owns" the device also needs to support
2546 hipri (also known as iopoll and mq_poll). The MegaRAID driver is an
2547 example of a SCSI LLD. Default: clear (0) which does normal
2548 (interrupted based) IO.
2550 .. option:: http_host=str : [http]
2552 Hostname to connect to. For S3, this could be the bucket hostname.
2553 Default is **localhost**
2555 .. option:: http_user=str : [http]
2557 Username for HTTP authentication.
2559 .. option:: http_pass=str : [http]
2561 Password for HTTP authentication.
2563 .. option:: https=str : [http]
2565 Enable HTTPS instead of http. *on* enables HTTPS; *insecure*
2566 will enable HTTPS, but disable SSL peer verification (use with
2567 caution!). Default is **off**
2569 .. option:: http_mode=str : [http]
2571 Which HTTP access mode to use: *webdav*, *swift*, or *s3*.
2572 Default is **webdav**
2574 .. option:: http_s3_region=str : [http]
2576 The S3 region/zone string.
2577 Default is **us-east-1**
2579 .. option:: http_s3_key=str : [http]
2583 .. option:: http_s3_keyid=str : [http]
2585 The S3 key/access id.
2587 .. option:: http_swift_auth_token=str : [http]
2589 The Swift auth token. See the example configuration file on how
2592 .. option:: http_verbose=int : [http]
2594 Enable verbose requests from libcurl. Useful for debugging. 1
2595 turns on verbose logging from libcurl, 2 additionally enables
2596 HTTP IO tracing. Default is **0**
2598 .. option:: uri=str : [nbd]
2600 Specify the NBD URI of the server to test. The string
2601 is a standard NBD URI
2602 (see https://github.com/NetworkBlockDevice/nbd/tree/master/doc).
2603 Example URIs: nbd://localhost:10809
2604 nbd+unix:///?socket=/tmp/socket
2605 nbds://tlshost/exportname
2607 .. option:: gpu_dev_ids=str : [libcufile]
2609 Specify the GPU IDs to use with CUDA. This is a colon-separated list of
2610 int. GPUs are assigned to workers roundrobin. Default is 0.
2612 .. option:: cuda_io=str : [libcufile]
2614 Specify the type of I/O to use with CUDA. Default is **cufile**.
2617 Use libcufile and nvidia-fs. This option performs I/O directly
2618 between a GPUDirect Storage filesystem and GPU buffers,
2619 avoiding use of a bounce buffer. If :option:`verify` is set,
2620 cudaMemcpy is used to copy verificaton data between RAM and GPU.
2621 Verification data is copied from RAM to GPU before a write
2622 and from GPU to RAM after a read. :option:`direct` must be 1.
2624 Use POSIX to perform I/O with a RAM buffer, and use cudaMemcpy
2625 to transfer data between RAM and the GPUs. Data is copied from
2626 GPU to RAM before a write and copied from RAM to GPU after a
2627 read. :option:`verify` does not affect use of cudaMemcpy.
2629 .. option:: pool=str : [dfs]
2631 Specify the label or UUID of the DAOS pool to connect to.
2633 .. option:: cont=str : [dfs]
2635 Specify the label or UUID of the DAOS container to open.
2637 .. option:: chunk_size=int : [dfs]
2639 Specificy a different chunk size (in bytes) for the dfs file.
2640 Use DAOS container's chunk size by default.
2642 .. option:: object_class=str : [dfs]
2644 Specificy a different object class for the dfs file.
2645 Use DAOS container's object class by default.
2647 .. option:: nfs_url=str : [nfs]
2649 URL in libnfs format, eg nfs://<server|ipv4|ipv6>/path[?arg=val[&arg=val]*]
2650 Refer to the libnfs README for more details.
2652 .. option:: program=str : [exec]
2654 Specify the program to execute.
2656 .. option:: arguments=str : [exec]
2658 Specify arguments to pass to program.
2659 Some special variables can be expanded to pass fio's job details to the program.
2662 Replaced by the duration of the job in seconds.
2664 Replaced by the name of the job.
2666 .. option:: grace_time=int : [exec]
2668 Specify the time between the SIGTERM and SIGKILL signals. Default is 1 second.
2670 .. option:: std_redirect=bool : [exec]
2672 If set, stdout and stderr streams are redirected to files named from the job name. Default is true.
2677 .. option:: iodepth=int
2679 Number of I/O units to keep in flight against the file. Note that
2680 increasing *iodepth* beyond 1 will not affect synchronous ioengines (except
2681 for small degrees when :option:`verify_async` is in use). Even async
2682 engines may impose OS restrictions causing the desired depth not to be
2683 achieved. This may happen on Linux when using libaio and not setting
2684 :option:`direct`\=1, since buffered I/O is not async on that OS. Keep an
2685 eye on the I/O depth distribution in the fio output to verify that the
2686 achieved depth is as expected. Default: 1.
2688 .. option:: iodepth_batch_submit=int, iodepth_batch=int
2690 This defines how many pieces of I/O to submit at once. It defaults to 1
2691 which means that we submit each I/O as soon as it is available, but can be
2692 raised to submit bigger batches of I/O at the time. If it is set to 0 the
2693 :option:`iodepth` value will be used.
2695 .. option:: iodepth_batch_complete_min=int, iodepth_batch_complete=int
2697 This defines how many pieces of I/O to retrieve at once. It defaults to 1
2698 which means that we'll ask for a minimum of 1 I/O in the retrieval process
2699 from the kernel. The I/O retrieval will go on until we hit the limit set by
2700 :option:`iodepth_low`. If this variable is set to 0, then fio will always
2701 check for completed events before queuing more I/O. This helps reduce I/O
2702 latency, at the cost of more retrieval system calls.
2704 .. option:: iodepth_batch_complete_max=int
2706 This defines maximum pieces of I/O to retrieve at once. This variable should
2707 be used along with :option:`iodepth_batch_complete_min`\=int variable,
2708 specifying the range of min and max amount of I/O which should be
2709 retrieved. By default it is equal to the :option:`iodepth_batch_complete_min`
2714 iodepth_batch_complete_min=1
2715 iodepth_batch_complete_max=<iodepth>
2717 which means that we will retrieve at least 1 I/O and up to the whole
2718 submitted queue depth. If none of I/O has been completed yet, we will wait.
2722 iodepth_batch_complete_min=0
2723 iodepth_batch_complete_max=<iodepth>
2725 which means that we can retrieve up to the whole submitted queue depth, but
2726 if none of I/O has been completed yet, we will NOT wait and immediately exit
2727 the system call. In this example we simply do polling.
2729 .. option:: iodepth_low=int
2731 The low water mark indicating when to start filling the queue
2732 again. Defaults to the same as :option:`iodepth`, meaning that fio will
2733 attempt to keep the queue full at all times. If :option:`iodepth` is set to
2734 e.g. 16 and *iodepth_low* is set to 4, then after fio has filled the queue of
2735 16 requests, it will let the depth drain down to 4 before starting to fill
2738 .. option:: serialize_overlap=bool
2740 Serialize in-flight I/Os that might otherwise cause or suffer from data races.
2741 When two or more I/Os are submitted simultaneously, there is no guarantee that
2742 the I/Os will be processed or completed in the submitted order. Further, if
2743 two or more of those I/Os are writes, any overlapping region between them can
2744 become indeterminate/undefined on certain storage. These issues can cause
2745 verification to fail erratically when at least one of the racing I/Os is
2746 changing data and the overlapping region has a non-zero size. Setting
2747 ``serialize_overlap`` tells fio to avoid provoking this behavior by explicitly
2748 serializing in-flight I/Os that have a non-zero overlap. Note that setting
2749 this option can reduce both performance and the :option:`iodepth` achieved.
2751 This option only applies to I/Os issued for a single job except when it is
2752 enabled along with :option:`io_submit_mode`\=offload. In offload mode, fio
2753 will check for overlap among all I/Os submitted by offload jobs with :option:`serialize_overlap`
2758 .. option:: io_submit_mode=str
2760 This option controls how fio submits the I/O to the I/O engine. The default
2761 is `inline`, which means that the fio job threads submit and reap I/O
2762 directly. If set to `offload`, the job threads will offload I/O submission
2763 to a dedicated pool of I/O threads. This requires some coordination and thus
2764 has a bit of extra overhead, especially for lower queue depth I/O where it
2765 can increase latencies. The benefit is that fio can manage submission rates
2766 independently of the device completion rates. This avoids skewed latency
2767 reporting if I/O gets backed up on the device side (the coordinated omission
2768 problem). Note that this option cannot reliably be used with async IO
2775 .. option:: thinktime=time
2777 Stall the job for the specified period of time after an I/O has completed before issuing the
2778 next. May be used to simulate processing being done by an application.
2779 When the unit is omitted, the value is interpreted in microseconds. See
2780 :option:`thinktime_blocks`, :option:`thinktime_iotime` and :option:`thinktime_spin`.
2782 .. option:: thinktime_spin=time
2784 Only valid if :option:`thinktime` is set - pretend to spend CPU time doing
2785 something with the data received, before falling back to sleeping for the
2786 rest of the period specified by :option:`thinktime`. When the unit is
2787 omitted, the value is interpreted in microseconds.
2789 .. option:: thinktime_blocks=int
2791 Only valid if :option:`thinktime` is set - control how many blocks to issue,
2792 before waiting :option:`thinktime` usecs. If not set, defaults to 1 which will make
2793 fio wait :option:`thinktime` usecs after every block. This effectively makes any
2794 queue depth setting redundant, since no more than 1 I/O will be queued
2795 before we have to complete it and do our :option:`thinktime`. In other words, this
2796 setting effectively caps the queue depth if the latter is larger.
2798 .. option:: thinktime_blocks_type=str
2800 Only valid if :option:`thinktime` is set - control how :option:`thinktime_blocks`
2801 triggers. The default is `complete`, which triggers thinktime when fio completes
2802 :option:`thinktime_blocks` blocks. If this is set to `issue`, then the trigger happens
2805 .. option:: thinktime_iotime=time
2807 Only valid if :option:`thinktime` is set - control :option:`thinktime`
2808 interval by time. The :option:`thinktime` stall is repeated after IOs
2809 are executed for :option:`thinktime_iotime`. For example,
2810 ``--thinktime_iotime=9s --thinktime=1s`` repeat 10-second cycle with IOs
2811 for 9 seconds and stall for 1 second. When the unit is omitted,
2812 :option:`thinktime_iotime` is interpreted as a number of seconds. If
2813 this option is used together with :option:`thinktime_blocks`, the
2814 :option:`thinktime` stall is repeated after :option:`thinktime_iotime`
2815 or after :option:`thinktime_blocks` IOs, whichever happens first.
2817 .. option:: rate=int[,int][,int]
2819 Cap the bandwidth used by this job. The number is in bytes/sec, the normal
2820 suffix rules apply. Comma-separated values may be specified for reads,
2821 writes, and trims as described in :option:`blocksize`.
2823 For example, using `rate=1m,500k` would limit reads to 1MiB/sec and writes to
2824 500KiB/sec. Capping only reads or writes can be done with `rate=,500k` or
2825 `rate=500k,` where the former will only limit writes (to 500KiB/sec) and the
2826 latter will only limit reads.
2828 .. option:: rate_min=int[,int][,int]
2830 Tell fio to do whatever it can to maintain at least this bandwidth. Failing
2831 to meet this requirement will cause the job to exit. Comma-separated values
2832 may be specified for reads, writes, and trims as described in
2833 :option:`blocksize`.
2835 .. option:: rate_iops=int[,int][,int]
2837 Cap the bandwidth to this number of IOPS. Basically the same as
2838 :option:`rate`, just specified independently of bandwidth. If the job is
2839 given a block size range instead of a fixed value, the smallest block size
2840 is used as the metric. Comma-separated values may be specified for reads,
2841 writes, and trims as described in :option:`blocksize`.
2843 .. option:: rate_iops_min=int[,int][,int]
2845 If fio doesn't meet this rate of I/O, it will cause the job to exit.
2846 Comma-separated values may be specified for reads, writes, and trims as
2847 described in :option:`blocksize`.
2849 .. option:: rate_process=str
2851 This option controls how fio manages rated I/O submissions. The default is
2852 `linear`, which submits I/O in a linear fashion with fixed delays between
2853 I/Os that gets adjusted based on I/O completion rates. If this is set to
2854 `poisson`, fio will submit I/O based on a more real world random request
2855 flow, known as the Poisson process
2856 (https://en.wikipedia.org/wiki/Poisson_point_process). The lambda will be
2857 10^6 / IOPS for the given workload.
2859 .. option:: rate_ignore_thinktime=bool
2861 By default, fio will attempt to catch up to the specified rate setting,
2862 if any kind of thinktime setting was used. If this option is set, then
2863 fio will ignore the thinktime and continue doing IO at the specified
2864 rate, instead of entering a catch-up mode after thinktime is done.
2870 .. option:: latency_target=time
2872 If set, fio will attempt to find the max performance point that the given
2873 workload will run at while maintaining a latency below this target. When
2874 the unit is omitted, the value is interpreted in microseconds. See
2875 :option:`latency_window` and :option:`latency_percentile`.
2877 .. option:: latency_window=time
2879 Used with :option:`latency_target` to specify the sample window that the job
2880 is run at varying queue depths to test the performance. When the unit is
2881 omitted, the value is interpreted in microseconds.
2883 .. option:: latency_percentile=float
2885 The percentage of I/Os that must fall within the criteria specified by
2886 :option:`latency_target` and :option:`latency_window`. If not set, this
2887 defaults to 100.0, meaning that all I/Os must be equal or below to the value
2888 set by :option:`latency_target`.
2890 .. option:: latency_run=bool
2892 Used with :option:`latency_target`. If false (default), fio will find
2893 the highest queue depth that meets :option:`latency_target` and exit. If
2894 true, fio will continue running and try to meet :option:`latency_target`
2895 by adjusting queue depth.
2897 .. option:: max_latency=time[,time][,time]
2899 If set, fio will exit the job with an ETIMEDOUT error if it exceeds this
2900 maximum latency. When the unit is omitted, the value is interpreted in
2901 microseconds. Comma-separated values may be specified for reads, writes,
2902 and trims as described in :option:`blocksize`.
2904 .. option:: rate_cycle=int
2906 Average bandwidth for :option:`rate` and :option:`rate_min` over this number
2907 of milliseconds. Defaults to 1000.
2913 .. option:: write_iolog=str
2915 Write the issued I/O patterns to the specified file. See
2916 :option:`read_iolog`. Specify a separate file for each job, otherwise the
2917 iologs will be interspersed and the file may be corrupt.
2919 .. option:: read_iolog=str
2921 Open an iolog with the specified filename and replay the I/O patterns it
2922 contains. This can be used to store a workload and replay it sometime
2923 later. The iolog given may also be a blktrace binary file, which allows fio
2924 to replay a workload captured by :command:`blktrace`. See
2925 :manpage:`blktrace(8)` for how to capture such logging data. For blktrace
2926 replay, the file needs to be turned into a blkparse binary data file first
2927 (``blkparse <device> -o /dev/null -d file_for_fio.bin``).
2928 You can specify a number of files by separating the names with a ':'
2929 character. See the :option:`filename` option for information on how to
2930 escape ':' characters within the file names. These files will
2931 be sequentially assigned to job clones created by :option:`numjobs`.
2932 '-' is a reserved name, meaning read from stdin, notably if
2933 :option:`filename` is set to '-' which means stdin as well, then
2934 this flag can't be set to '-'.
2936 .. option:: read_iolog_chunked=bool
2938 Determines how iolog is read. If false(default) entire :option:`read_iolog`
2939 will be read at once. If selected true, input from iolog will be read
2940 gradually. Useful when iolog is very large, or it is generated.
2942 .. option:: merge_blktrace_file=str
2944 When specified, rather than replaying the logs passed to :option:`read_iolog`,
2945 the logs go through a merge phase which aggregates them into a single
2946 blktrace. The resulting file is then passed on as the :option:`read_iolog`
2947 parameter. The intention here is to make the order of events consistent.
2948 This limits the influence of the scheduler compared to replaying multiple
2949 blktraces via concurrent jobs.
2951 .. option:: merge_blktrace_scalars=float_list
2953 This is a percentage based option that is index paired with the list of
2954 files passed to :option:`read_iolog`. When merging is performed, scale
2955 the time of each event by the corresponding amount. For example,
2956 ``--merge_blktrace_scalars="50:100"`` runs the first trace in halftime
2957 and the second trace in realtime. This knob is separately tunable from
2958 :option:`replay_time_scale` which scales the trace during runtime and
2959 does not change the output of the merge unlike this option.
2961 .. option:: merge_blktrace_iters=float_list
2963 This is a whole number option that is index paired with the list of files
2964 passed to :option:`read_iolog`. When merging is performed, run each trace
2965 for the specified number of iterations. For example,
2966 ``--merge_blktrace_iters="2:1"`` runs the first trace for two iterations
2967 and the second trace for one iteration.
2969 .. option:: replay_no_stall=bool
2971 When replaying I/O with :option:`read_iolog` the default behavior is to
2972 attempt to respect the timestamps within the log and replay them with the
2973 appropriate delay between IOPS. By setting this variable fio will not
2974 respect the timestamps and attempt to replay them as fast as possible while
2975 still respecting ordering. The result is the same I/O pattern to a given
2976 device, but different timings.
2978 .. option:: replay_time_scale=int
2980 When replaying I/O with :option:`read_iolog`, fio will honor the
2981 original timing in the trace. With this option, it's possible to scale
2982 the time. It's a percentage option, if set to 50 it means run at 50%
2983 the original IO rate in the trace. If set to 200, run at twice the
2984 original IO rate. Defaults to 100.
2986 .. option:: replay_redirect=str
2988 While replaying I/O patterns using :option:`read_iolog` the default behavior
2989 is to replay the IOPS onto the major/minor device that each IOP was recorded
2990 from. This is sometimes undesirable because on a different machine those
2991 major/minor numbers can map to a different device. Changing hardware on the
2992 same system can also result in a different major/minor mapping.
2993 ``replay_redirect`` causes all I/Os to be replayed onto the single specified
2994 device regardless of the device it was recorded
2995 from. i.e. :option:`replay_redirect`\= :file:`/dev/sdc` would cause all I/O
2996 in the blktrace or iolog to be replayed onto :file:`/dev/sdc`. This means
2997 multiple devices will be replayed onto a single device, if the trace
2998 contains multiple devices. If you want multiple devices to be replayed
2999 concurrently to multiple redirected devices you must blkparse your trace
3000 into separate traces and replay them with independent fio invocations.
3001 Unfortunately this also breaks the strict time ordering between multiple
3004 .. option:: replay_align=int
3006 Force alignment of the byte offsets in a trace to this value. The value
3007 must be a power of 2.
3009 .. option:: replay_scale=int
3011 Scale byte offsets down by this factor when replaying traces. Should most
3012 likely use :option:`replay_align` as well.
3014 .. option:: replay_skip=str
3016 Sometimes it's useful to skip certain IO types in a replay trace.
3017 This could be, for instance, eliminating the writes in the trace.
3018 Or not replaying the trims/discards, if you are redirecting to
3019 a device that doesn't support them. This option takes a comma
3020 separated list of read, write, trim, sync.
3023 Threads, processes and job synchronization
3024 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3028 Fio defaults to creating jobs by using fork, however if this option is
3029 given, fio will create jobs by using POSIX Threads' function
3030 :manpage:`pthread_create(3)` to create threads instead.
3032 .. option:: wait_for=str
3034 If set, the current job won't be started until all workers of the specified
3035 waitee job are done.
3037 ``wait_for`` operates on the job name basis, so there are a few
3038 limitations. First, the waitee must be defined prior to the waiter job
3039 (meaning no forward references). Second, if a job is being referenced as a
3040 waitee, it must have a unique name (no duplicate waitees).
3042 .. option:: nice=int
3044 Run the job with the given nice value. See man :manpage:`nice(2)`.
3046 On Windows, values less than -15 set the process class to "High"; -1 through
3047 -15 set "Above Normal"; 1 through 15 "Below Normal"; and above 15 "Idle"
3050 .. option:: prio=int
3052 Set the I/O priority value of this job. Linux limits us to a positive value
3053 between 0 and 7, with 0 being the highest. See man
3054 :manpage:`ionice(1)`. Refer to an appropriate manpage for other operating
3055 systems since meaning of priority may differ. For per-command priority
3056 setting, see I/O engine specific :option:`cmdprio_percentage` and
3057 :option:`cmdprio` options.
3059 .. option:: prioclass=int
3061 Set the I/O priority class. See man :manpage:`ionice(1)`. For per-command
3062 priority setting, see I/O engine specific :option:`cmdprio_percentage`
3063 and :option:`cmdprio_class` options.
3065 .. option:: cpus_allowed=str
3067 Controls the same options as :option:`cpumask`, but accepts a textual
3068 specification of the permitted CPUs instead and CPUs are indexed from 0. So
3069 to use CPUs 0 and 5 you would specify ``cpus_allowed=0,5``. This option also
3070 allows a range of CPUs to be specified -- say you wanted a binding to CPUs
3071 0, 5, and 8 to 15, you would set ``cpus_allowed=0,5,8-15``.
3073 On Windows, when ``cpus_allowed`` is unset only CPUs from fio's current
3074 processor group will be used and affinity settings are inherited from the
3075 system. An fio build configured to target Windows 7 makes options that set
3076 CPUs processor group aware and values will set both the processor group
3077 and a CPU from within that group. For example, on a system where processor
3078 group 0 has 40 CPUs and processor group 1 has 32 CPUs, ``cpus_allowed``
3079 values between 0 and 39 will bind CPUs from processor group 0 and
3080 ``cpus_allowed`` values between 40 and 71 will bind CPUs from processor
3081 group 1. When using ``cpus_allowed_policy=shared`` all CPUs specified by a
3082 single ``cpus_allowed`` option must be from the same processor group. For
3083 Windows fio builds not built for Windows 7, CPUs will only be selected from
3084 (and be relative to) whatever processor group fio happens to be running in
3085 and CPUs from other processor groups cannot be used.
3087 .. option:: cpus_allowed_policy=str
3089 Set the policy of how fio distributes the CPUs specified by
3090 :option:`cpus_allowed` or :option:`cpumask`. Two policies are supported:
3093 All jobs will share the CPU set specified.
3095 Each job will get a unique CPU from the CPU set.
3097 **shared** is the default behavior, if the option isn't specified. If
3098 **split** is specified, then fio will assign one cpu per job. If not
3099 enough CPUs are given for the jobs listed, then fio will roundrobin the CPUs
3102 .. option:: cpumask=int
3104 Set the CPU affinity of this job. The parameter given is a bit mask of
3105 allowed CPUs the job may run on. So if you want the allowed CPUs to be 1
3106 and 5, you would pass the decimal value of (1 << 1 | 1 << 5), or 34. See man
3107 :manpage:`sched_setaffinity(2)`. This may not work on all supported
3108 operating systems or kernel versions. This option doesn't work well for a
3109 higher CPU count than what you can store in an integer mask, so it can only
3110 control cpus 1-32. For boxes with larger CPU counts, use
3111 :option:`cpus_allowed`.
3113 .. option:: numa_cpu_nodes=str
3115 Set this job running on specified NUMA nodes' CPUs. The arguments allow
3116 comma delimited list of cpu numbers, A-B ranges, or `all`. Note, to enable
3117 NUMA options support, fio must be built on a system with libnuma-dev(el)
3120 .. option:: numa_mem_policy=str
3122 Set this job's memory policy and corresponding NUMA nodes. Format of the
3127 ``mode`` is one of the following memory policies: ``default``, ``prefer``,
3128 ``bind``, ``interleave`` or ``local``. For ``default`` and ``local`` memory
3129 policies, no node needs to be specified. For ``prefer``, only one node is
3130 allowed. For ``bind`` and ``interleave`` the ``nodelist`` may be as
3131 follows: a comma delimited list of numbers, A-B ranges, or `all`.
3133 .. option:: cgroup=str
3135 Add job to this control group. If it doesn't exist, it will be created. The
3136 system must have a mounted cgroup blkio mount point for this to work. If
3137 your system doesn't have it mounted, you can do so with::
3139 # mount -t cgroup -o blkio none /cgroup
3141 .. option:: cgroup_weight=int
3143 Set the weight of the cgroup to this value. See the documentation that comes
3144 with the kernel, allowed values are in the range of 100..1000.
3146 .. option:: cgroup_nodelete=bool
3148 Normally fio will delete the cgroups it has created after the job
3149 completion. To override this behavior and to leave cgroups around after the
3150 job completion, set ``cgroup_nodelete=1``. This can be useful if one wants
3151 to inspect various cgroup files after job completion. Default: false.
3153 .. option:: flow_id=int
3155 The ID of the flow. If not specified, it defaults to being a global
3156 flow. See :option:`flow`.
3158 .. option:: flow=int
3160 Weight in token-based flow control. If this value is used, then there is a
3161 'flow counter' which is used to regulate the proportion of activity between
3162 two or more jobs. Fio attempts to keep this flow counter near zero. The
3163 ``flow`` parameter stands for how much should be added or subtracted to the
3164 flow counter on each iteration of the main I/O loop. That is, if one job has
3165 ``flow=8`` and another job has ``flow=-1``, then there will be a roughly 1:8
3166 ratio in how much one runs vs the other.
3168 .. option:: flow_sleep=int
3170 The period of time, in microseconds, to wait after the flow counter
3171 has exceeded its proportion before retrying operations.
3173 .. option:: stonewall, wait_for_previous
3175 Wait for preceding jobs in the job file to exit, before starting this
3176 one. Can be used to insert serialization points in the job file. A stone
3177 wall also implies starting a new reporting group, see
3178 :option:`group_reporting`.
3182 By default, fio will continue running all other jobs when one job finishes.
3183 Sometimes this is not the desired action. Setting ``exitall`` will instead
3184 make fio terminate all jobs in the same group, as soon as one job of that
3187 .. option:: exit_what
3189 By default, fio will continue running all other jobs when one job finishes.
3190 Sometimes this is not the desired action. Setting ``exit_all`` will
3191 instead make fio terminate all jobs in the same group. The option
3192 ``exit_what`` allows to control which jobs get terminated when ``exitall`` is
3193 enabled. The default is ``group`` and does not change the behaviour of
3194 ``exitall``. The setting ``all`` terminates all jobs. The setting ``stonewall``
3195 terminates all currently running jobs across all groups and continues execution
3196 with the next stonewalled group.
3198 .. option:: exec_prerun=str
3200 Before running this job, issue the command specified through
3201 :manpage:`system(3)`. Output is redirected in a file called
3202 :file:`jobname.prerun.txt`.
3204 .. option:: exec_postrun=str
3206 After the job completes, issue the command specified though
3207 :manpage:`system(3)`. Output is redirected in a file called
3208 :file:`jobname.postrun.txt`.
3212 Instead of running as the invoking user, set the user ID to this value
3213 before the thread/process does any work.
3217 Set group ID, see :option:`uid`.
3223 .. option:: verify_only
3225 Do not perform specified workload, only verify data still matches previous
3226 invocation of this workload. This option allows one to check data multiple
3227 times at a later date without overwriting it. This option makes sense only
3228 for workloads that write data, and does not support workloads with the
3229 :option:`time_based` option set.
3231 .. option:: do_verify=bool
3233 Run the verify phase after a write phase. Only valid if :option:`verify` is
3236 .. option:: verify=str
3238 If writing to a file, fio can verify the file contents after each iteration
3239 of the job. Each verification method also implies verification of special
3240 header, which is written to the beginning of each block. This header also
3241 includes meta information, like offset of the block, block number, timestamp
3242 when block was written, etc. :option:`verify` can be combined with
3243 :option:`verify_pattern` option. The allowed values are:
3246 Use an md5 sum of the data area and store it in the header of
3250 Use an experimental crc64 sum of the data area and store it in the
3251 header of each block.
3254 Use a crc32c sum of the data area and store it in the header of
3255 each block. This will automatically use hardware acceleration
3256 (e.g. SSE4.2 on an x86 or CRC crypto extensions on ARM64) but will
3257 fall back to software crc32c if none is found. Generally the
3258 fastest checksum fio supports when hardware accelerated.
3264 Use a crc32 sum of the data area and store it in the header of each
3268 Use a crc16 sum of the data area and store it in the header of each
3272 Use a crc7 sum of the data area and store it in the header of each
3276 Use xxhash as the checksum function. Generally the fastest software
3277 checksum that fio supports.
3280 Use sha512 as the checksum function.
3283 Use sha256 as the checksum function.
3286 Use optimized sha1 as the checksum function.
3289 Use optimized sha3-224 as the checksum function.
3292 Use optimized sha3-256 as the checksum function.
3295 Use optimized sha3-384 as the checksum function.
3298 Use optimized sha3-512 as the checksum function.
3301 This option is deprecated, since now meta information is included in
3302 generic verification header and meta verification happens by
3303 default. For detailed information see the description of the
3304 :option:`verify` setting. This option is kept because of
3305 compatibility's sake with old configurations. Do not use it.
3308 Verify a strict pattern. Normally fio includes a header with some
3309 basic information and checksumming, but if this option is set, only
3310 the specific pattern set with :option:`verify_pattern` is verified.
3313 Only pretend to verify. Useful for testing internals with
3314 :option:`ioengine`\=null, not for much else.
3316 This option can be used for repeated burn-in tests of a system to make sure
3317 that the written data is also correctly read back. If the data direction
3318 given is a read or random read, fio will assume that it should verify a
3319 previously written file. If the data direction includes any form of write,
3320 the verify will be of the newly written data.
3322 To avoid false verification errors, do not use the norandommap option when
3323 verifying data with async I/O engines and I/O depths > 1. Or use the
3324 norandommap and the lfsr random generator together to avoid writing to the
3325 same offset with muliple outstanding I/Os.
3327 .. option:: verify_offset=int
3329 Swap the verification header with data somewhere else in the block before
3330 writing. It is swapped back before verifying.
3332 .. option:: verify_interval=int
3334 Write the verification header at a finer granularity than the
3335 :option:`blocksize`. It will be written for chunks the size of
3336 ``verify_interval``. :option:`blocksize` should divide this evenly.
3338 .. option:: verify_pattern=str
3340 If set, fio will fill the I/O buffers with this pattern. Fio defaults to
3341 filling with totally random bytes, but sometimes it's interesting to fill
3342 with a known pattern for I/O verification purposes. Depending on the width
3343 of the pattern, fio will fill 1/2/3/4 bytes of the buffer at the time (it can
3344 be either a decimal or a hex number). The ``verify_pattern`` if larger than
3345 a 32-bit quantity has to be a hex number that starts with either "0x" or
3346 "0X". Use with :option:`verify`. Also, ``verify_pattern`` supports %o
3347 format, which means that for each block offset will be written and then
3348 verified back, e.g.::
3352 Or use combination of everything::
3354 verify_pattern=0xff%o"abcd"-12
3356 .. option:: verify_fatal=bool
3358 Normally fio will keep checking the entire contents before quitting on a
3359 block verification failure. If this option is set, fio will exit the job on
3360 the first observed failure. Default: false.
3362 .. option:: verify_dump=bool
3364 If set, dump the contents of both the original data block and the data block
3365 we read off disk to files. This allows later analysis to inspect just what
3366 kind of data corruption occurred. Off by default.
3368 .. option:: verify_async=int
3370 Fio will normally verify I/O inline from the submitting thread. This option
3371 takes an integer describing how many async offload threads to create for I/O
3372 verification instead, causing fio to offload the duty of verifying I/O
3373 contents to one or more separate threads. If using this offload option, even
3374 sync I/O engines can benefit from using an :option:`iodepth` setting higher
3375 than 1, as it allows them to have I/O in flight while verifies are running.
3376 Defaults to 0 async threads, i.e. verification is not asynchronous.
3378 .. option:: verify_async_cpus=str
3380 Tell fio to set the given CPU affinity on the async I/O verification
3381 threads. See :option:`cpus_allowed` for the format used.
3383 .. option:: verify_backlog=int
3385 Fio will normally verify the written contents of a job that utilizes verify
3386 once that job has completed. In other words, everything is written then
3387 everything is read back and verified. You may want to verify continually
3388 instead for a variety of reasons. Fio stores the meta data associated with
3389 an I/O block in memory, so for large verify workloads, quite a bit of memory
3390 would be used up holding this meta data. If this option is enabled, fio will
3391 write only N blocks before verifying these blocks.
3393 .. option:: verify_backlog_batch=int
3395 Control how many blocks fio will verify if :option:`verify_backlog` is
3396 set. If not set, will default to the value of :option:`verify_backlog`
3397 (meaning the entire queue is read back and verified). If
3398 ``verify_backlog_batch`` is less than :option:`verify_backlog` then not all
3399 blocks will be verified, if ``verify_backlog_batch`` is larger than
3400 :option:`verify_backlog`, some blocks will be verified more than once.
3402 .. option:: verify_state_save=bool
3404 When a job exits during the write phase of a verify workload, save its
3405 current state. This allows fio to replay up until that point, if the verify
3406 state is loaded for the verify read phase. The format of the filename is,
3409 <type>-<jobname>-<jobindex>-verify.state.
3411 <type> is "local" for a local run, "sock" for a client/server socket
3412 connection, and "ip" (192.168.0.1, for instance) for a networked
3413 client/server connection. Defaults to true.
3415 .. option:: verify_state_load=bool
3417 If a verify termination trigger was used, fio stores the current write state
3418 of each thread. This can be used at verification time so that fio knows how
3419 far it should verify. Without this information, fio will run a full
3420 verification pass, according to the settings in the job file used. Default
3423 .. option:: trim_percentage=int
3425 Number of verify blocks to discard/trim.
3427 .. option:: trim_verify_zero=bool
3429 Verify that trim/discarded blocks are returned as zeros.
3431 .. option:: trim_backlog=int
3433 Trim after this number of blocks are written.
3435 .. option:: trim_backlog_batch=int
3437 Trim this number of I/O blocks.
3439 .. option:: experimental_verify=bool
3441 Enable experimental verification.
3446 .. option:: steadystate=str:float, ss=str:float
3448 Define the criterion and limit for assessing steady state performance. The
3449 first parameter designates the criterion whereas the second parameter sets
3450 the threshold. When the criterion falls below the threshold for the
3451 specified duration, the job will stop. For example, `iops_slope:0.1%` will
3452 direct fio to terminate the job when the least squares regression slope
3453 falls below 0.1% of the mean IOPS. If :option:`group_reporting` is enabled
3454 this will apply to all jobs in the group. Below is the list of available
3455 steady state assessment criteria. All assessments are carried out using only
3456 data from the rolling collection window. Threshold limits can be expressed
3457 as a fixed value or as a percentage of the mean in the collection window.
3459 When using this feature, most jobs should include the :option:`time_based`
3460 and :option:`runtime` options or the :option:`loops` option so that fio does not
3461 stop running after it has covered the full size of the specified file(s) or device(s).
3464 Collect IOPS data. Stop the job if all individual IOPS measurements
3465 are within the specified limit of the mean IOPS (e.g., ``iops:2``
3466 means that all individual IOPS values must be within 2 of the mean,
3467 whereas ``iops:0.2%`` means that all individual IOPS values must be
3468 within 0.2% of the mean IOPS to terminate the job).
3471 Collect IOPS data and calculate the least squares regression
3472 slope. Stop the job if the slope falls below the specified limit.
3475 Collect bandwidth data. Stop the job if all individual bandwidth
3476 measurements are within the specified limit of the mean bandwidth.
3479 Collect bandwidth data and calculate the least squares regression
3480 slope. Stop the job if the slope falls below the specified limit.
3482 .. option:: steadystate_duration=time, ss_dur=time
3484 A rolling window of this duration will be used to judge whether steady state
3485 has been reached. Data will be collected once per second. The default is 0
3486 which disables steady state detection. When the unit is omitted, the
3487 value is interpreted in seconds.
3489 .. option:: steadystate_ramp_time=time, ss_ramp=time
3491 Allow the job to run for the specified duration before beginning data
3492 collection for checking the steady state job termination criterion. The
3493 default is 0. When the unit is omitted, the value is interpreted in seconds.
3496 Measurements and reporting
3497 ~~~~~~~~~~~~~~~~~~~~~~~~~~
3499 .. option:: per_job_logs=bool
3501 If set, this generates bw/clat/iops log with per file private filenames. If
3502 not set, jobs with identical names will share the log filename. Default:
3505 .. option:: group_reporting
3507 It may sometimes be interesting to display statistics for groups of jobs as
3508 a whole instead of for each individual job. This is especially true if
3509 :option:`numjobs` is used; looking at individual thread/process output
3510 quickly becomes unwieldy. To see the final report per-group instead of
3511 per-job, use :option:`group_reporting`. Jobs in a file will be part of the
3512 same reporting group, unless if separated by a :option:`stonewall`, or by
3513 using :option:`new_group`.
3515 .. option:: new_group
3517 Start a new reporting group. See: :option:`group_reporting`. If not given,
3518 all jobs in a file will be part of the same reporting group, unless
3519 separated by a :option:`stonewall`.
3521 .. option:: stats=bool
3523 By default, fio collects and shows final output results for all jobs
3524 that run. If this option is set to 0, then fio will ignore it in
3525 the final stat output.
3527 .. option:: write_bw_log=str
3529 If given, write a bandwidth log for this job. Can be used to store data of
3530 the bandwidth of the jobs in their lifetime.
3532 If no str argument is given, the default filename of
3533 :file:`jobname_type.x.log` is used. Even when the argument is given, fio
3534 will still append the type of log. So if one specifies::
3538 The actual log name will be :file:`foo_bw.x.log` where `x` is the index
3539 of the job (`1..N`, where `N` is the number of jobs). If
3540 :option:`per_job_logs` is false, then the filename will not include the
3543 The included :command:`fio_generate_plots` script uses :command:`gnuplot` to turn these
3544 text files into nice graphs. See `Log File Formats`_ for how data is
3545 structured within the file.
3547 .. option:: write_lat_log=str
3549 Same as :option:`write_bw_log`, except this option creates I/O
3550 submission (e.g., :file:`name_slat.x.log`), completion (e.g.,
3551 :file:`name_clat.x.log`), and total (e.g., :file:`name_lat.x.log`)
3552 latency files instead. See :option:`write_bw_log` for details about
3553 the filename format and `Log File Formats`_ for how data is structured
3556 .. option:: write_hist_log=str
3558 Same as :option:`write_bw_log` but writes an I/O completion latency
3559 histogram file (e.g., :file:`name_hist.x.log`) instead. Note that this
3560 file will be empty unless :option:`log_hist_msec` has also been set.
3561 See :option:`write_bw_log` for details about the filename format and
3562 `Log File Formats`_ for how data is structured within the file.
3564 .. option:: write_iops_log=str
3566 Same as :option:`write_bw_log`, but writes an IOPS file (e.g.
3567 :file:`name_iops.x.log`) instead. Because fio defaults to individual
3568 I/O logging, the value entry in the IOPS log will be 1 unless windowed
3569 logging (see :option:`log_avg_msec`) has been enabled. See
3570 :option:`write_bw_log` for details about the filename format and `Log
3571 File Formats`_ for how data is structured within the file.
3573 .. option:: log_entries=int
3575 By default, fio will log an entry in the iops, latency, or bw log for
3576 every I/O that completes. The initial number of I/O log entries is 1024.
3577 When the log entries are all used, new log entries are dynamically
3578 allocated. This dynamic log entry allocation may negatively impact
3579 time-related statistics such as I/O tail latencies (e.g. 99.9th percentile
3580 completion latency). This option allows specifying a larger initial
3581 number of log entries to avoid run-time allocations of new log entries,
3582 resulting in more precise time-related I/O statistics.
3583 Also see :option:`log_avg_msec`. Defaults to 1024.
3585 .. option:: log_avg_msec=int
3587 By default, fio will log an entry in the iops, latency, or bw log for every
3588 I/O that completes. When writing to the disk log, that can quickly grow to a
3589 very large size. Setting this option makes fio average the each log entry
3590 over the specified period of time, reducing the resolution of the log. See
3591 :option:`log_max_value` as well. Defaults to 0, logging all entries.
3592 Also see `Log File Formats`_.
3594 .. option:: log_hist_msec=int
3596 Same as :option:`log_avg_msec`, but logs entries for completion latency
3597 histograms. Computing latency percentiles from averages of intervals using
3598 :option:`log_avg_msec` is inaccurate. Setting this option makes fio log
3599 histogram entries over the specified period of time, reducing log sizes for
3600 high IOPS devices while retaining percentile accuracy. See
3601 :option:`log_hist_coarseness` and :option:`write_hist_log` as well.
3602 Defaults to 0, meaning histogram logging is disabled.
3604 .. option:: log_hist_coarseness=int
3606 Integer ranging from 0 to 6, defining the coarseness of the resolution of
3607 the histogram logs enabled with :option:`log_hist_msec`. For each increment
3608 in coarseness, fio outputs half as many bins. Defaults to 0, for which
3609 histogram logs contain 1216 latency bins. See :option:`write_hist_log`
3610 and `Log File Formats`_.
3612 .. option:: log_max_value=bool
3614 If :option:`log_avg_msec` is set, fio logs the average over that window. If
3615 you instead want to log the maximum value, set this option to 1. Defaults to
3616 0, meaning that averaged values are logged.
3618 .. option:: log_offset=bool
3620 If this is set, the iolog options will include the byte offset for the I/O
3621 entry as well as the other data values. Defaults to 0 meaning that
3622 offsets are not present in logs. Also see `Log File Formats`_.
3624 .. option:: log_compression=int
3626 If this is set, fio will compress the I/O logs as it goes, to keep the
3627 memory footprint lower. When a log reaches the specified size, that chunk is
3628 removed and compressed in the background. Given that I/O logs are fairly
3629 highly compressible, this yields a nice memory savings for longer runs. The
3630 downside is that the compression will consume some background CPU cycles, so
3631 it may impact the run. This, however, is also true if the logging ends up
3632 consuming most of the system memory. So pick your poison. The I/O logs are
3633 saved normally at the end of a run, by decompressing the chunks and storing
3634 them in the specified log file. This feature depends on the availability of
3637 .. option:: log_compression_cpus=str
3639 Define the set of CPUs that are allowed to handle online log compression for
3640 the I/O jobs. This can provide better isolation between performance
3641 sensitive jobs, and background compression work. See
3642 :option:`cpus_allowed` for the format used.
3644 .. option:: log_store_compressed=bool
3646 If set, fio will store the log files in a compressed format. They can be
3647 decompressed with fio, using the :option:`--inflate-log` command line
3648 parameter. The files will be stored with a :file:`.fz` suffix.
3650 .. option:: log_unix_epoch=bool
3652 If set, fio will log Unix timestamps to the log files produced by enabling
3653 write_type_log for each log type, instead of the default zero-based
3656 .. option:: block_error_percentiles=bool
3658 If set, record errors in trim block-sized units from writes and trims and
3659 output a histogram of how many trims it took to get to errors, and what kind
3660 of error was encountered.
3662 .. option:: bwavgtime=int
3664 Average the calculated bandwidth over the given time. Value is specified in
3665 milliseconds. If the job also does bandwidth logging through
3666 :option:`write_bw_log`, then the minimum of this option and
3667 :option:`log_avg_msec` will be used. Default: 500ms.
3669 .. option:: iopsavgtime=int
3671 Average the calculated IOPS over the given time. Value is specified in
3672 milliseconds. If the job also does IOPS logging through
3673 :option:`write_iops_log`, then the minimum of this option and
3674 :option:`log_avg_msec` will be used. Default: 500ms.
3676 .. option:: disk_util=bool
3678 Generate disk utilization statistics, if the platform supports it.
3681 .. option:: disable_lat=bool
3683 Disable measurements of total latency numbers. Useful only for cutting back
3684 the number of calls to :manpage:`gettimeofday(2)`, as that does impact
3685 performance at really high IOPS rates. Note that to really get rid of a
3686 large amount of these calls, this option must be used with
3687 :option:`disable_slat` and :option:`disable_bw_measurement` as well.
3689 .. option:: disable_clat=bool
3691 Disable measurements of completion latency numbers. See
3692 :option:`disable_lat`.
3694 .. option:: disable_slat=bool
3696 Disable measurements of submission latency numbers. See
3697 :option:`disable_lat`.
3699 .. option:: disable_bw_measurement=bool, disable_bw=bool
3701 Disable measurements of throughput/bandwidth numbers. See
3702 :option:`disable_lat`.
3704 .. option:: slat_percentiles=bool
3706 Report submission latency percentiles. Submission latency is not recorded
3707 for synchronous ioengines.
3709 .. option:: clat_percentiles=bool
3711 Report completion latency percentiles.
3713 .. option:: lat_percentiles=bool
3715 Report total latency percentiles. Total latency is the sum of submission
3716 latency and completion latency.
3718 .. option:: percentile_list=float_list
3720 Overwrite the default list of percentiles for latencies and the block error
3721 histogram. Each number is a floating point number in the range (0,100], and
3722 the maximum length of the list is 20. Use ``:`` to separate the numbers. For
3723 example, ``--percentile_list=99.5:99.9`` will cause fio to report the
3724 latency durations below which 99.5% and 99.9% of the observed latencies fell,
3727 .. option:: significant_figures=int
3729 If using :option:`--output-format` of `normal`, set the significant
3730 figures to this value. Higher values will yield more precise IOPS and
3731 throughput units, while lower values will round. Requires a minimum
3732 value of 1 and a maximum value of 10. Defaults to 4.
3738 .. option:: exitall_on_error
3740 When one job finishes in error, terminate the rest. The default is to wait
3741 for each job to finish.
3743 .. option:: continue_on_error=str
3745 Normally fio will exit the job on the first observed failure. If this option
3746 is set, fio will continue the job when there is a 'non-fatal error' (EIO or
3747 EILSEQ) until the runtime is exceeded or the I/O size specified is
3748 completed. If this option is used, there are two more stats that are
3749 appended, the total error count and the first error. The error field given
3750 in the stats is the first error that was hit during the run.
3752 The allowed values are:
3755 Exit on any I/O or verify errors.
3758 Continue on read errors, exit on all others.
3761 Continue on write errors, exit on all others.
3764 Continue on any I/O error, exit on all others.
3767 Continue on verify errors, exit on all others.
3770 Continue on all errors.
3773 Backward-compatible alias for 'none'.
3776 Backward-compatible alias for 'all'.
3778 .. option:: ignore_error=str
3780 Sometimes you want to ignore some errors during test in that case you can
3781 specify error list for each error type, instead of only being able to
3782 ignore the default 'non-fatal error' using :option:`continue_on_error`.
3783 ``ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST`` errors for
3784 given error type is separated with ':'. Error may be symbol ('ENOSPC',
3785 'ENOMEM') or integer. Example::
3787 ignore_error=EAGAIN,ENOSPC:122
3789 This option will ignore EAGAIN from READ, and ENOSPC and 122(EDQUOT) from
3790 WRITE. This option works by overriding :option:`continue_on_error` with
3791 the list of errors for each error type if any.
3793 .. option:: error_dump=bool
3795 If set dump every error even if it is non fatal, true by default. If
3796 disabled only fatal error will be dumped.
3798 Running predefined workloads
3799 ----------------------------
3801 Fio includes predefined profiles that mimic the I/O workloads generated by
3804 .. option:: profile=str
3806 The predefined workload to run. Current profiles are:
3809 Threaded I/O bench (tiotest/tiobench) like workload.
3812 Aerospike Certification Tool (ACT) like workload.
3814 To view a profile's additional options use :option:`--cmdhelp` after specifying
3815 the profile. For example::
3817 $ fio --profile=act --cmdhelp
3822 .. option:: device-names=str
3827 .. option:: load=int
3830 ACT load multiplier. Default: 1.
3832 .. option:: test-duration=time
3835 How long the entire test takes to run. When the unit is omitted, the value
3836 is given in seconds. Default: 24h.
3838 .. option:: threads-per-queue=int
3841 Number of read I/O threads per device. Default: 8.
3843 .. option:: read-req-num-512-blocks=int
3846 Number of 512B blocks to read at the time. Default: 3.
3848 .. option:: large-block-op-kbytes=int
3851 Size of large block ops in KiB (writes). Default: 131072.
3856 Set to run ACT prep phase.
3858 Tiobench profile options
3859 ~~~~~~~~~~~~~~~~~~~~~~~~
3861 .. option:: size=str
3866 .. option:: block=int
3869 Block size in bytes. Default: 4096.
3871 .. option:: numruns=int
3881 .. option:: threads=int
3886 Interpreting the output
3887 -----------------------
3890 Example output was based on the following:
3891 TZ=UTC fio --iodepth=8 --ioengine=null --size=100M --time_based \
3892 --rate=1256k --bs=14K --name=quick --runtime=1s --name=mixed \
3893 --runtime=2m --rw=rw
3895 Fio spits out a lot of output. While running, fio will display the status of the
3896 jobs created. An example of that would be::
3898 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]
3900 The characters inside the first set of square brackets denote the current status of
3901 each thread. The first character is the first job defined in the job file, and so
3902 forth. The possible values (in typical life cycle order) are:
3904 +------+-----+-----------------------------------------------------------+
3906 +======+=====+===========================================================+
3907 | P | | Thread setup, but not started. |
3908 +------+-----+-----------------------------------------------------------+
3909 | C | | Thread created. |
3910 +------+-----+-----------------------------------------------------------+
3911 | I | | Thread initialized, waiting or generating necessary data. |
3912 +------+-----+-----------------------------------------------------------+
3913 | | p | Thread running pre-reading file(s). |
3914 +------+-----+-----------------------------------------------------------+
3915 | | / | Thread is in ramp period. |
3916 +------+-----+-----------------------------------------------------------+
3917 | | R | Running, doing sequential reads. |
3918 +------+-----+-----------------------------------------------------------+
3919 | | r | Running, doing random reads. |
3920 +------+-----+-----------------------------------------------------------+
3921 | | W | Running, doing sequential writes. |
3922 +------+-----+-----------------------------------------------------------+
3923 | | w | Running, doing random writes. |
3924 +------+-----+-----------------------------------------------------------+
3925 | | M | Running, doing mixed sequential reads/writes. |
3926 +------+-----+-----------------------------------------------------------+
3927 | | m | Running, doing mixed random reads/writes. |
3928 +------+-----+-----------------------------------------------------------+
3929 | | D | Running, doing sequential trims. |
3930 +------+-----+-----------------------------------------------------------+
3931 | | d | Running, doing random trims. |
3932 +------+-----+-----------------------------------------------------------+
3933 | | F | Running, currently waiting for :manpage:`fsync(2)`. |
3934 +------+-----+-----------------------------------------------------------+
3935 | | V | Running, doing verification of written data. |
3936 +------+-----+-----------------------------------------------------------+
3937 | f | | Thread finishing. |
3938 +------+-----+-----------------------------------------------------------+
3939 | E | | Thread exited, not reaped by main thread yet. |
3940 +------+-----+-----------------------------------------------------------+
3941 | _ | | Thread reaped. |
3942 +------+-----+-----------------------------------------------------------+
3943 | X | | Thread reaped, exited with an error. |
3944 +------+-----+-----------------------------------------------------------+
3945 | K | | Thread reaped, exited due to signal. |
3946 +------+-----+-----------------------------------------------------------+
3949 Example output was based on the following:
3950 TZ=UTC fio --iodepth=8 --ioengine=null --size=100M --runtime=58m \
3951 --time_based --rate=2512k --bs=256K --numjobs=10 \
3952 --name=readers --rw=read --name=writers --rw=write
3954 Fio will condense the thread string as not to take up more space on the command
3955 line than needed. For instance, if you have 10 readers and 10 writers running,
3956 the output would look like this::
3958 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]
3960 Note that the status string is displayed in order, so it's possible to tell which of
3961 the jobs are currently doing what. In the example above this means that jobs 1--10
3962 are readers and 11--20 are writers.
3964 The other values are fairly self explanatory -- number of threads currently
3965 running and doing I/O, the number of currently open files (f=), the estimated
3966 completion percentage, the rate of I/O since last check (read speed listed first,
3967 then write speed and optionally trim speed) in terms of bandwidth and IOPS,
3968 and time to completion for the current running group. It's impossible to estimate
3969 runtime of the following groups (if any).
3972 Example output was based on the following:
3973 TZ=UTC fio --iodepth=16 --ioengine=posixaio --filename=/tmp/fiofile \
3974 --direct=1 --size=100M --time_based --runtime=50s --rate_iops=89 \
3975 --bs=7K --name=Client1 --rw=write
3977 When fio is done (or interrupted by :kbd:`Ctrl-C`), it will show the data for
3978 each thread, group of threads, and disks in that order. For each overall thread (or
3979 group) the output looks like::
3981 Client1: (groupid=0, jobs=1): err= 0: pid=16109: Sat Jun 24 12:07:54 2017
3982 write: IOPS=88, BW=623KiB/s (638kB/s)(30.4MiB/50032msec)
3983 slat (nsec): min=500, max=145500, avg=8318.00, stdev=4781.50
3984 clat (usec): min=170, max=78367, avg=4019.02, stdev=8293.31
3985 lat (usec): min=174, max=78375, avg=4027.34, stdev=8291.79
3986 clat percentiles (usec):
3987 | 1.00th=[ 302], 5.00th=[ 326], 10.00th=[ 343], 20.00th=[ 363],
3988 | 30.00th=[ 392], 40.00th=[ 404], 50.00th=[ 416], 60.00th=[ 445],
3989 | 70.00th=[ 816], 80.00th=[ 6718], 90.00th=[12911], 95.00th=[21627],
3990 | 99.00th=[43779], 99.50th=[51643], 99.90th=[68682], 99.95th=[72877],
3992 bw ( KiB/s): min= 532, max= 686, per=0.10%, avg=622.87, stdev=24.82, samples= 100
3993 iops : min= 76, max= 98, avg=88.98, stdev= 3.54, samples= 100
3994 lat (usec) : 250=0.04%, 500=64.11%, 750=4.81%, 1000=2.79%
3995 lat (msec) : 2=4.16%, 4=1.84%, 10=4.90%, 20=11.33%, 50=5.37%
3996 lat (msec) : 100=0.65%
3997 cpu : usr=0.27%, sys=0.18%, ctx=12072, majf=0, minf=21
3998 IO depths : 1=85.0%, 2=13.1%, 4=1.8%, 8=0.1%, 16=0.0%, 32=0.0%, >=64=0.0%
3999 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
4000 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
4001 issued rwt: total=0,4450,0, short=0,0,0, dropped=0,0,0
4002 latency : target=0, window=0, percentile=100.00%, depth=8
4004 The job name (or first job's name when using :option:`group_reporting`) is printed,
4005 along with the group id, count of jobs being aggregated, last error id seen (which
4006 is 0 when there are no errors), pid/tid of that thread and the time the job/group
4007 completed. Below are the I/O statistics for each data direction performed (showing
4008 writes in the example above). In the order listed, they denote:
4011 The string before the colon shows the I/O direction the statistics
4012 are for. **IOPS** is the average I/Os performed per second. **BW**
4013 is the average bandwidth rate shown as: value in power of 2 format
4014 (value in power of 10 format). The last two values show: (**total
4015 I/O performed** in power of 2 format / **runtime** of that thread).
4018 Submission latency (**min** being the minimum, **max** being the
4019 maximum, **avg** being the average, **stdev** being the standard
4020 deviation). This is the time it took to submit the I/O. For
4021 sync I/O this row is not displayed as the slat is really the
4022 completion latency (since queue/complete is one operation there).
4023 This value can be in nanoseconds, microseconds or milliseconds ---
4024 fio will choose the most appropriate base and print that (in the
4025 example above nanoseconds was the best scale). Note: in :option:`--minimal` mode
4026 latencies are always expressed in microseconds.
4029 Completion latency. Same names as slat, this denotes the time from
4030 submission to completion of the I/O pieces. For sync I/O, clat will
4031 usually be equal (or very close) to 0, as the time from submit to
4032 complete is basically just CPU time (I/O has already been done, see slat
4036 Total latency. Same names as slat and clat, this denotes the time from
4037 when fio created the I/O unit to completion of the I/O operation.
4040 Bandwidth statistics based on samples. Same names as the xlat stats,
4041 but also includes the number of samples taken (**samples**) and an
4042 approximate percentage of total aggregate bandwidth this thread
4043 received in its group (**per**). This last value is only really
4044 useful if the threads in this group are on the same disk, since they
4045 are then competing for disk access.
4048 IOPS statistics based on samples. Same names as bw.
4050 **lat (nsec/usec/msec)**
4051 The distribution of I/O completion latencies. This is the time from when
4052 I/O leaves fio and when it gets completed. Unlike the separate
4053 read/write/trim sections above, the data here and in the remaining
4054 sections apply to all I/Os for the reporting group. 250=0.04% means that
4055 0.04% of the I/Os completed in under 250us. 500=64.11% means that 64.11%
4056 of the I/Os required 250 to 499us for completion.
4059 CPU usage. User and system time, along with the number of context
4060 switches this thread went through, usage of system and user time, and
4061 finally the number of major and minor page faults. The CPU utilization
4062 numbers are averages for the jobs in that reporting group, while the
4063 context and fault counters are summed.
4066 The distribution of I/O depths over the job lifetime. The numbers are
4067 divided into powers of 2 and each entry covers depths from that value
4068 up to those that are lower than the next entry -- e.g., 16= covers
4069 depths from 16 to 31. Note that the range covered by a depth
4070 distribution entry can be different to the range covered by the
4071 equivalent submit/complete distribution entry.
4074 How many pieces of I/O were submitting in a single submit call. Each
4075 entry denotes that amount and below, until the previous entry -- e.g.,
4076 16=100% means that we submitted anywhere between 9 to 16 I/Os per submit
4077 call. Note that the range covered by a submit distribution entry can
4078 be different to the range covered by the equivalent depth distribution
4082 Like the above submit number, but for completions instead.
4085 The number of read/write/trim requests issued, and how many of them were
4089 These values are for :option:`latency_target` and related options. When
4090 these options are engaged, this section describes the I/O depth required
4091 to meet the specified latency target.
4094 Example output was based on the following:
4095 TZ=UTC fio --ioengine=null --iodepth=2 --size=100M --numjobs=2 \
4096 --rate_process=poisson --io_limit=32M --name=read --bs=128k \
4097 --rate=11M --name=write --rw=write --bs=2k --rate=700k
4099 After each client has been listed, the group statistics are printed. They
4100 will look like this::
4102 Run status group 0 (all jobs):
4103 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
4104 WRITE: bw=1231KiB/s (1261kB/s), 616KiB/s-621KiB/s (630kB/s-636kB/s), io=64.0MiB (67.1MB), run=52747-53223msec
4106 For each data direction it prints:
4109 Aggregate bandwidth of threads in this group followed by the
4110 minimum and maximum bandwidth of all the threads in this group.
4111 Values outside of brackets are power-of-2 format and those
4112 within are the equivalent value in a power-of-10 format.
4114 Aggregate I/O performed of all threads in this group. The
4115 format is the same as bw.
4117 The smallest and longest runtimes of the threads in this group.
4119 And finally, the disk statistics are printed. This is Linux specific. They will look like this::
4121 Disk stats (read/write):
4122 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
4124 Each value is printed for both reads and writes, with reads first. The
4128 Number of I/Os performed by all groups.
4130 Number of merges performed by the I/O scheduler.
4132 Number of ticks we kept the disk busy.
4134 Total time spent in the disk queue.
4136 The disk utilization. A value of 100% means we kept the disk
4137 busy constantly, 50% would be a disk idling half of the time.
4139 It is also possible to get fio to dump the current output while it is running,
4140 without terminating the job. To do that, send fio the **USR1** signal. You can
4141 also get regularly timed dumps by using the :option:`--status-interval`
4142 parameter, or by creating a file in :file:`/tmp` named
4143 :file:`fio-dump-status`. If fio sees this file, it will unlink it and dump the
4144 current output status.
4150 For scripted usage where you typically want to generate tables or graphs of the
4151 results, fio can output the results in a semicolon separated format. The format
4152 is one long line of values, such as::
4154 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%
4155 A description of this job goes here.
4157 The job description (if provided) follows on a second line for terse v2.
4158 It appears on the same line for other terse versions.
4160 To enable terse output, use the :option:`--minimal` or
4161 :option:`--output-format`\=terse command line options. The
4162 first value is the version of the terse output format. If the output has to be
4163 changed for some reason, this number will be incremented by 1 to signify that
4166 Split up, the format is as follows (comments in brackets denote when a
4167 field was introduced or whether it's specific to some terse version):
4171 terse version, fio version [v3], jobname, groupid, error
4175 Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec)
4176 Submission latency: min, max, mean, stdev (usec)
4177 Completion latency: min, max, mean, stdev (usec)
4178 Completion latency percentiles: 20 fields (see below)
4179 Total latency: min, max, mean, stdev (usec)
4180 Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev, number of samples [v5]
4181 IOPS [v5]: min, max, mean, stdev, number of samples
4187 Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec)
4188 Submission latency: min, max, mean, stdev (usec)
4189 Completion latency: min, max, mean, stdev (usec)
4190 Completion latency percentiles: 20 fields (see below)
4191 Total latency: min, max, mean, stdev (usec)
4192 Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev, number of samples [v5]
4193 IOPS [v5]: min, max, mean, stdev, number of samples
4195 TRIM status [all but version 3]:
4197 Fields are similar to READ/WRITE status.
4201 user, system, context switches, major faults, minor faults
4205 <=1, 2, 4, 8, 16, 32, >=64
4207 I/O latencies microseconds::
4209 <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
4211 I/O latencies milliseconds::
4213 <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
4215 Disk utilization [v3]::
4217 disk name, read ios, write ios, read merges, write merges, read ticks, write ticks,
4218 time spent in queue, disk utilization percentage
4220 Additional Info (dependent on continue_on_error, default off)::
4222 total # errors, first error code
4224 Additional Info (dependent on description being set)::
4228 Completion latency percentiles can be a grouping of up to 20 sets, so for the
4229 terse output fio writes all of them. Each field will look like this::
4233 which is the Xth percentile, and the `usec` latency associated with it.
4235 For `Disk utilization`, all disks used by fio are shown. So for each disk there
4236 will be a disk utilization section.
4238 Below is a single line containing short names for each of the fields in the
4239 minimal output v3, separated by semicolons::
4241 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
4243 In client/server mode terse output differs from what appears when jobs are run
4244 locally. Disk utilization data is omitted from the standard terse output and
4245 for v3 and later appears on its own separate line at the end of each terse
4252 The `json` output format is intended to be both human readable and convenient
4253 for automated parsing. For the most part its sections mirror those of the
4254 `normal` output. The `runtime` value is reported in msec and the `bw` value is
4255 reported in 1024 bytes per second units.
4261 The `json+` output format is identical to the `json` output format except that it
4262 adds a full dump of the completion latency bins. Each `bins` object contains a
4263 set of (key, value) pairs where keys are latency durations and values count how
4264 many I/Os had completion latencies of the corresponding duration. For example,
4267 "bins" : { "87552" : 1, "89600" : 1, "94720" : 1, "96768" : 1, "97792" : 1, "99840" : 1, "100864" : 2, "103936" : 6, "104960" : 534, "105984" : 5995, "107008" : 7529, ... }
4269 This data indicates that one I/O required 87,552ns to complete, two I/Os required
4270 100,864ns to complete, and 7529 I/Os required 107,008ns to complete.
4272 Also included with fio is a Python script `fio_jsonplus_clat2csv` that takes
4273 json+ output and generates CSV-formatted latency data suitable for plotting.
4275 The latency durations actually represent the midpoints of latency intervals.
4276 For details refer to :file:`stat.h`.
4282 There are two trace file format that you can encounter. The older (v1) format is
4283 unsupported since version 1.20-rc3 (March 2008). It will still be described
4284 below in case that you get an old trace and want to understand it.
4286 In any case the trace is a simple text file with a single action per line.
4289 Trace file format v1
4290 ~~~~~~~~~~~~~~~~~~~~
4292 Each line represents a single I/O action in the following format::
4296 where `rw=0/1` for read/write, and the `offset` and `length` entries being in bytes.
4298 This format is not supported in fio versions >= 1.20-rc3.
4301 Trace file format v2
4302 ~~~~~~~~~~~~~~~~~~~~
4304 The second version of the trace file format was added in fio version 1.17. It
4305 allows to access more then one file per trace and has a bigger set of possible
4308 The first line of the trace file has to be::
4312 Following this can be lines in two different formats, which are described below.
4314 The file management format::
4318 The `filename` is given as an absolute path. The `action` can be one of these:
4321 Add the given `filename` to the trace.
4323 Open the file with the given `filename`. The `filename` has to have
4324 been added with the **add** action before.
4326 Close the file with the given `filename`. The file has to have been
4330 The file I/O action format::
4332 filename action offset length
4334 The `filename` is given as an absolute path, and has to have been added and
4335 opened before it can be used with this format. The `offset` and `length` are
4336 given in bytes. The `action` can be one of these:
4339 Wait for `offset` microseconds. Everything below 100 is discarded.
4340 The time is relative to the previous `wait` statement.
4342 Read `length` bytes beginning from `offset`.
4344 Write `length` bytes beginning from `offset`.
4346 :manpage:`fsync(2)` the file.
4348 :manpage:`fdatasync(2)` the file.
4350 Trim the given file from the given `offset` for `length` bytes.
4353 I/O Replay - Merging Traces
4354 ---------------------------
4356 Colocation is a common practice used to get the most out of a machine.
4357 Knowing which workloads play nicely with each other and which ones don't is
4358 a much harder task. While fio can replay workloads concurrently via multiple
4359 jobs, it leaves some variability up to the scheduler making results harder to
4360 reproduce. Merging is a way to make the order of events consistent.
4362 Merging is integrated into I/O replay and done when a
4363 :option:`merge_blktrace_file` is specified. The list of files passed to
4364 :option:`read_iolog` go through the merge process and output a single file
4365 stored to the specified file. The output file is passed on as if it were the
4366 only file passed to :option:`read_iolog`. An example would look like::
4368 $ fio --read_iolog="<file1>:<file2>" --merge_blktrace_file="<output_file>"
4370 Creating only the merged file can be done by passing the command line argument
4371 :option:`--merge-blktrace-only`.
4373 Scaling traces can be done to see the relative impact of any particular trace
4374 being slowed down or sped up. :option:`merge_blktrace_scalars` takes in a colon
4375 separated list of percentage scalars. It is index paired with the files passed
4376 to :option:`read_iolog`.
4378 With scaling, it may be desirable to match the running time of all traces.
4379 This can be done with :option:`merge_blktrace_iters`. It is index paired with
4380 :option:`read_iolog` just like :option:`merge_blktrace_scalars`.
4382 In an example, given two traces, A and B, each 60s long. If we want to see
4383 the impact of trace A issuing IOs twice as fast and repeat trace A over the
4384 runtime of trace B, the following can be done::
4386 $ fio --read_iolog="<trace_a>:"<trace_b>" --merge_blktrace_file"<output_file>" --merge_blktrace_scalars="50:100" --merge_blktrace_iters="2:1"
4388 This runs trace A at 2x the speed twice for approximately the same runtime as
4389 a single run of trace B.
4392 CPU idleness profiling
4393 ----------------------
4395 In some cases, we want to understand CPU overhead in a test. For example, we
4396 test patches for the specific goodness of whether they reduce CPU usage.
4397 Fio implements a balloon approach to create a thread per CPU that runs at idle
4398 priority, meaning that it only runs when nobody else needs the cpu.
4399 By measuring the amount of work completed by the thread, idleness of each CPU
4400 can be derived accordingly.
4402 An unit work is defined as touching a full page of unsigned characters. Mean and
4403 standard deviation of time to complete an unit work is reported in "unit work"
4404 section. Options can be chosen to report detailed percpu idleness or overall
4405 system idleness by aggregating percpu stats.
4408 Verification and triggers
4409 -------------------------
4411 Fio is usually run in one of two ways, when data verification is done. The first
4412 is a normal write job of some sort with verify enabled. When the write phase has
4413 completed, fio switches to reads and verifies everything it wrote. The second
4414 model is running just the write phase, and then later on running the same job
4415 (but with reads instead of writes) to repeat the same I/O patterns and verify
4416 the contents. Both of these methods depend on the write phase being completed,
4417 as fio otherwise has no idea how much data was written.
4419 With verification triggers, fio supports dumping the current write state to
4420 local files. Then a subsequent read verify workload can load this state and know
4421 exactly where to stop. This is useful for testing cases where power is cut to a
4422 server in a managed fashion, for instance.
4424 A verification trigger consists of two things:
4426 1) Storing the write state of each job.
4427 2) Executing a trigger command.
4429 The write state is relatively small, on the order of hundreds of bytes to single
4430 kilobytes. It contains information on the number of completions done, the last X
4433 A trigger is invoked either through creation ('touch') of a specified file in
4434 the system, or through a timeout setting. If fio is run with
4435 :option:`--trigger-file`\= :file:`/tmp/trigger-file`, then it will continually
4436 check for the existence of :file:`/tmp/trigger-file`. When it sees this file, it
4437 will fire off the trigger (thus saving state, and executing the trigger
4440 For client/server runs, there's both a local and remote trigger. If fio is
4441 running as a server backend, it will send the job states back to the client for
4442 safe storage, then execute the remote trigger, if specified. If a local trigger
4443 is specified, the server will still send back the write state, but the client
4444 will then execute the trigger.
4446 Verification trigger example
4447 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4449 Let's say we want to run a powercut test on the remote Linux machine 'server'.
4450 Our write workload is in :file:`write-test.fio`. We want to cut power to 'server' at
4451 some point during the run, and we'll run this test from the safety or our local
4452 machine, 'localbox'. On the server, we'll start the fio backend normally::
4454 server# fio --server
4456 and on the client, we'll fire off the workload::
4458 localbox$ fio --client=server --trigger-file=/tmp/my-trigger --trigger-remote="bash -c \"echo b > /proc/sysrq-triger\""
4460 We set :file:`/tmp/my-trigger` as the trigger file, and we tell fio to execute::
4462 echo b > /proc/sysrq-trigger
4464 on the server once it has received the trigger and sent us the write state. This
4465 will work, but it's not **really** cutting power to the server, it's merely
4466 abruptly rebooting it. If we have a remote way of cutting power to the server
4467 through IPMI or similar, we could do that through a local trigger command
4468 instead. Let's assume we have a script that does IPMI reboot of a given hostname,
4469 ipmi-reboot. On localbox, we could then have run fio with a local trigger
4472 localbox$ fio --client=server --trigger-file=/tmp/my-trigger --trigger="ipmi-reboot server"
4474 For this case, fio would wait for the server to send us the write state, then
4475 execute ``ipmi-reboot server`` when that happened.
4477 Loading verify state
4478 ~~~~~~~~~~~~~~~~~~~~
4480 To load stored write state, a read verification job file must contain the
4481 :option:`verify_state_load` option. If that is set, fio will load the previously
4482 stored state. For a local fio run this is done by loading the files directly,
4483 and on a client/server run, the server backend will ask the client to send the
4484 files over and load them from there.
4490 Fio supports a variety of log file formats, for logging latencies, bandwidth,
4491 and IOPS. The logs share a common format, which looks like this:
4493 *time* (`msec`), *value*, *data direction*, *block size* (`bytes`),
4494 *offset* (`bytes`), *command priority*
4496 *Time* for the log entry is always in milliseconds. The *value* logged depends
4497 on the type of log, it will be one of the following:
4500 Value is latency in nsecs
4506 *Data direction* is one of the following:
4515 The entry's *block size* is always in bytes. The *offset* is the position in bytes
4516 from the start of the file for that particular I/O. The logging of the offset can be
4517 toggled with :option:`log_offset`.
4519 *Command priority* is 0 for normal priority and 1 for high priority. This is controlled
4520 by the ioengine specific :option:`cmdprio_percentage`.
4522 Fio defaults to logging every individual I/O but when windowed logging is set
4523 through :option:`log_avg_msec`, either the average (by default) or the maximum
4524 (:option:`log_max_value` is set) *value* seen over the specified period of time
4525 is recorded. Each *data direction* seen within the window period will aggregate
4526 its values in a separate row. Further, when using windowed logging the *block
4527 size* and *offset* entries will always contain 0.
4533 Normally fio is invoked as a stand-alone application on the machine where the
4534 I/O workload should be generated. However, the backend and frontend of fio can
4535 be run separately i.e., the fio server can generate an I/O workload on the "Device
4536 Under Test" while being controlled by a client on another machine.
4538 Start the server on the machine which has access to the storage DUT::
4542 where `args` defines what fio listens to. The arguments are of the form
4543 ``type,hostname`` or ``IP,port``. *type* is either ``ip`` (or ip4) for TCP/IP
4544 v4, ``ip6`` for TCP/IP v6, or ``sock`` for a local unix domain socket.
4545 *hostname* is either a hostname or IP address, and *port* is the port to listen
4546 to (only valid for TCP/IP, not a local socket). Some examples:
4550 Start a fio server, listening on all interfaces on the default port (8765).
4552 2) ``fio --server=ip:hostname,4444``
4554 Start a fio server, listening on IP belonging to hostname and on port 4444.
4556 3) ``fio --server=ip6:::1,4444``
4558 Start a fio server, listening on IPv6 localhost ::1 and on port 4444.
4560 4) ``fio --server=,4444``
4562 Start a fio server, listening on all interfaces on port 4444.
4564 5) ``fio --server=1.2.3.4``
4566 Start a fio server, listening on IP 1.2.3.4 on the default port.
4568 6) ``fio --server=sock:/tmp/fio.sock``
4570 Start a fio server, listening on the local socket :file:`/tmp/fio.sock`.
4572 Once a server is running, a "client" can connect to the fio server with::
4574 fio <local-args> --client=<server> <remote-args> <job file(s)>
4576 where `local-args` are arguments for the client where it is running, `server`
4577 is the connect string, and `remote-args` and `job file(s)` are sent to the
4578 server. The `server` string follows the same format as it does on the server
4579 side, to allow IP/hostname/socket and port strings.
4581 Fio can connect to multiple servers this way::
4583 fio --client=<server1> <job file(s)> --client=<server2> <job file(s)>
4585 If the job file is located on the fio server, then you can tell the server to
4586 load a local file as well. This is done by using :option:`--remote-config` ::
4588 fio --client=server --remote-config /path/to/file.fio
4590 Then fio will open this local (to the server) job file instead of being passed
4591 one from the client.
4593 If you have many servers (example: 100 VMs/containers), you can input a pathname
4594 of a file containing host IPs/names as the parameter value for the
4595 :option:`--client` option. For example, here is an example :file:`host.list`
4596 file containing 2 hostnames::
4598 host1.your.dns.domain
4599 host2.your.dns.domain
4601 The fio command would then be::
4603 fio --client=host.list <job file(s)>
4605 In this mode, you cannot input server-specific parameters or job files -- all
4606 servers receive the same job file.
4608 In order to let ``fio --client`` runs use a shared filesystem from multiple
4609 hosts, ``fio --client`` now prepends the IP address of the server to the
4610 filename. For example, if fio is using the directory :file:`/mnt/nfs/fio` and is
4611 writing filename :file:`fileio.tmp`, with a :option:`--client` `hostfile`
4612 containing two hostnames ``h1`` and ``h2`` with IP addresses 192.168.10.120 and
4613 192.168.10.121, then fio will create two files::
4615 /mnt/nfs/fio/192.168.10.120.fileio.tmp
4616 /mnt/nfs/fio/192.168.10.121.fileio.tmp
4618 Terse output in client/server mode will differ slightly from what is produced
4619 when fio is run in stand-alone mode. See the terse output section for details.