4 The first step in getting fio to simulate a desired I/O workload, is writing a
5 job file describing that specific setup. A job file may contain any number of
6 threads and/or files -- the typical contents of the job file is a *global*
7 section defining shared parameters, and one or more job sections describing the
8 jobs involved. When run, fio parses this file and sets everything up as
9 described. If we break down a job from top to bottom, it contains the following
14 Defines the I/O pattern issued to the file(s). We may only be reading
15 sequentially from this file(s), or we may be writing randomly. Or even
16 mixing reads and writes, sequentially or randomly.
17 Should we be doing buffered I/O, or direct/raw I/O?
21 In how large chunks are we issuing I/O? This may be a single value,
22 or it may describe a range of block sizes.
26 How much data are we going to be reading/writing.
30 How do we issue I/O? We could be memory mapping the file, we could be
31 using regular read/write, we could be using splice, async I/O, or even
36 If the I/O engine is async, how large a queuing depth do we want to
42 How many files are we spreading the workload over.
44 `Threads, processes and job synchronization`_
46 How many threads or processes should we spread this workload over.
48 The above are the basic parameters defined for a workload, in addition there's a
49 multitude of parameters that modify other aspects of how this job behaves.
55 .. option:: --debug=type
57 Enable verbose tracing `type` of various fio actions. May be ``all`` for all types
58 or individual types separated by a comma (e.g. ``--debug=file,mem`` will
59 enable file and memory debugging). Currently, additional logging is
63 Dump info related to processes.
65 Dump info related to file actions.
67 Dump info related to I/O queuing.
69 Dump info related to memory allocations.
71 Dump info related to blktrace setup.
73 Dump info related to I/O verification.
75 Enable all debug options.
77 Dump info related to random offset generation.
79 Dump info related to option matching and parsing.
81 Dump info related to disk utilization updates.
83 Dump info only related to job number x.
85 Dump info only related to mutex up/down ops.
87 Dump info related to profile extensions.
89 Dump info related to internal time keeping.
91 Dump info related to networking connections.
93 Dump info related to I/O rate switching.
95 Dump info related to log compress/decompress.
97 Dump info related to steadystate detection.
99 Dump info related to the helper thread.
101 Dump info related to support for zoned block devices.
103 Show available debug options.
105 .. option:: --parse-only
107 Parse options only, don't start any I/O.
109 .. option:: --merge-blktrace-only
111 Merge blktraces only, don't start any I/O.
113 .. option:: --output=filename
115 Write output to file `filename`.
117 .. option:: --output-format=format
119 Set the reporting `format` to `normal`, `terse`, `json`, or `json+`. Multiple
120 formats can be selected, separated by a comma. `terse` is a CSV based
121 format. `json+` is like `json`, except it adds a full dump of the latency
124 .. option:: --bandwidth-log
126 Generate aggregate bandwidth logs.
128 .. option:: --minimal
130 Print statistics in a terse, semicolon-delimited format.
132 .. option:: --append-terse
134 Print statistics in selected mode AND terse, semicolon-delimited format.
135 **Deprecated**, use :option:`--output-format` instead to select multiple
138 .. option:: --terse-version=version
140 Set terse `version` output format (default 3, or 2 or 4 or 5).
142 .. option:: --version
144 Print version information and exit.
148 Print a summary of the command line options and exit.
150 .. option:: --cpuclock-test
152 Perform test and validation of internal CPU clock.
154 .. option:: --crctest=[test]
156 Test the speed of the built-in checksumming functions. If no argument is
157 given, all of them are tested. Alternatively, a comma separated list can
158 be passed, in which case the given ones are tested.
160 .. option:: --cmdhelp=command
162 Print help information for `command`. May be ``all`` for all commands.
164 .. option:: --enghelp=[ioengine[,command]]
166 List all commands defined by `ioengine`, or print help for `command`
167 defined by `ioengine`. If no `ioengine` is given, list all
170 .. option:: --showcmd
172 Convert given job files to a set of command-line options.
174 .. option:: --readonly
176 Turn on safety read-only checks, preventing writes and trims. The
177 ``--readonly`` option is an extra safety guard to prevent users from
178 accidentally starting a write or trim workload when that is not desired.
179 Fio will only modify the device under test if
180 `rw=write/randwrite/rw/randrw/trim/randtrim/trimwrite` is given. This
181 safety net can be used as an extra precaution.
183 .. option:: --eta=when
185 Specifies when real-time ETA estimate should be printed. `when` may be
186 `always`, `never` or `auto`. `auto` is the default, it prints ETA
187 when requested if the output is a TTY. `always` disregards the output
188 type, and prints ETA when requested. `never` never prints ETA.
190 .. option:: --eta-interval=time
192 By default, fio requests client ETA status roughly every second. With
193 this option, the interval is configurable. Fio imposes a minimum
194 allowed time to avoid flooding the console, less than 250 msec is
197 .. option:: --eta-newline=time
199 Force a new line for every `time` period passed. When the unit is omitted,
200 the value is interpreted in seconds.
202 .. option:: --status-interval=time
204 Force a full status dump of cumulative (from job start) values at `time`
205 intervals. This option does *not* provide per-period measurements. So
206 values such as bandwidth are running averages. When the time unit is omitted,
207 `time` is interpreted in seconds. Note that using this option with
208 ``--output-format=json`` will yield output that technically isn't valid
209 json, since the output will be collated sets of valid json. It will need
210 to be split into valid sets of json after the run.
212 .. option:: --section=name
214 Only run specified section `name` in job file. Multiple sections can be specified.
215 The ``--section`` option allows one to combine related jobs into one file.
216 E.g. one job file could define light, moderate, and heavy sections. Tell
217 fio to run only the "heavy" section by giving ``--section=heavy``
218 command line option. One can also specify the "write" operations in one
219 section and "verify" operation in another section. The ``--section`` option
220 only applies to job sections. The reserved *global* section is always
223 .. option:: --alloc-size=kb
225 Allocate additional internal smalloc pools of size `kb` in KiB. The
226 ``--alloc-size`` option increases shared memory set aside for use by fio.
227 If running large jobs with randommap enabled, fio can run out of memory.
228 Smalloc is an internal allocator for shared structures from a fixed size
229 memory pool and can grow to 16 pools. The pool size defaults to 16MiB.
231 NOTE: While running :file:`.fio_smalloc.*` backing store files are visible
234 .. option:: --warnings-fatal
236 All fio parser warnings are fatal, causing fio to exit with an
239 .. option:: --max-jobs=nr
241 Set the maximum number of threads/processes to support to `nr`.
242 NOTE: On Linux, it may be necessary to increase the shared-memory
243 limit (:file:`/proc/sys/kernel/shmmax`) if fio runs into errors while
246 .. option:: --server=args
248 Start a backend server, with `args` specifying what to listen to.
249 See `Client/Server`_ section.
251 .. option:: --daemonize=pidfile
253 Background a fio server, writing the pid to the given `pidfile` file.
255 .. option:: --client=hostname
257 Instead of running the jobs locally, send and run them on the given `hostname`
258 or set of `hostname`\s. See `Client/Server`_ section.
260 .. option:: --remote-config=file
262 Tell fio server to load this local `file`.
264 .. option:: --idle-prof=option
266 Report CPU idleness. `option` is one of the following:
269 Run unit work calibration only and exit.
272 Show aggregate system idleness and unit work.
275 As **system** but also show per CPU idleness.
277 .. option:: --inflate-log=log
279 Inflate and output compressed `log`.
281 .. option:: --trigger-file=file
283 Execute trigger command when `file` exists.
285 .. option:: --trigger-timeout=time
287 Execute trigger at this `time`.
289 .. option:: --trigger=command
291 Set this `command` as local trigger.
293 .. option:: --trigger-remote=command
295 Set this `command` as remote trigger.
297 .. option:: --aux-path=path
299 Use the directory specified by `path` for generated state files instead
300 of the current working directory.
302 Any parameters following the options will be assumed to be job files, unless
303 they match a job file parameter. Multiple job files can be listed and each job
304 file will be regarded as a separate group. Fio will :option:`stonewall`
305 execution between each group.
311 As previously described, fio accepts one or more job files describing what it is
312 supposed to do. The job file format is the classic ini file, where the names
313 enclosed in [] brackets define the job name. You are free to use any ASCII name
314 you want, except *global* which has special meaning. Following the job name is
315 a sequence of zero or more parameters, one per line, that define the behavior of
316 the job. If the first character in a line is a ';' or a '#', the entire line is
317 discarded as a comment.
319 A *global* section sets defaults for the jobs described in that file. A job may
320 override a *global* section parameter, and a job file may even have several
321 *global* sections if so desired. A job is only affected by a *global* section
324 The :option:`--cmdhelp` option also lists all options. If used with a `command`
325 argument, :option:`--cmdhelp` will detail the given `command`.
327 See the `examples/` directory for inspiration on how to write job files. Note
328 the copyright and license requirements currently apply to `examples/` files.
330 So let's look at a really simple job file that defines two processes, each
331 randomly reading from a 128MiB file:
335 ; -- start job file --
346 As you can see, the job file sections themselves are empty as all the described
347 parameters are shared. As no :option:`filename` option is given, fio makes up a
348 `filename` for each of the jobs as it sees fit. On the command line, this job
349 would look as follows::
351 $ fio --name=global --rw=randread --size=128m --name=job1 --name=job2
354 Let's look at an example that has a number of processes writing randomly to
359 ; -- start job file --
370 Here we have no *global* section, as we only have one job defined anyway. We
371 want to use async I/O here, with a depth of 4 for each file. We also increased
372 the buffer size used to 32KiB and define numjobs to 4 to fork 4 identical
373 jobs. The result is 4 processes each randomly writing to their own 64MiB
374 file. Instead of using the above job file, you could have given the parameters
375 on the command line. For this case, you would specify::
377 $ fio --name=random-writers --ioengine=libaio --iodepth=4 --rw=randwrite --bs=32k --direct=0 --size=64m --numjobs=4
379 When fio is utilized as a basis of any reasonably large test suite, it might be
380 desirable to share a set of standardized settings across multiple job files.
381 Instead of copy/pasting such settings, any section may pull in an external
382 :file:`filename.fio` file with *include filename* directive, as in the following
385 ; -- start job file including.fio --
389 include glob-include.fio
396 include test-include.fio
397 ; -- end job file including.fio --
401 ; -- start job file glob-include.fio --
404 ; -- end job file glob-include.fio --
408 ; -- start job file test-include.fio --
411 ; -- end job file test-include.fio --
413 Settings pulled into a section apply to that section only (except *global*
414 section). Include directives may be nested in that any included file may contain
415 further include directive(s). Include files may not contain [] sections.
418 Environment variables
419 ~~~~~~~~~~~~~~~~~~~~~
421 Fio also supports environment variable expansion in job files. Any sub-string of
422 the form ``${VARNAME}`` as part of an option value (in other words, on the right
423 of the '='), will be expanded to the value of the environment variable called
424 `VARNAME`. If no such environment variable is defined, or `VARNAME` is the
425 empty string, the empty string will be substituted.
427 As an example, let's look at a sample fio invocation and job file::
429 $ SIZE=64m NUMJOBS=4 fio jobfile.fio
433 ; -- start job file --
440 This will expand to the following equivalent job file at runtime:
444 ; -- start job file --
451 Fio ships with a few example job files, you can also look there for inspiration.
456 Additionally, fio has a set of reserved keywords that will be replaced
457 internally with the appropriate value. Those keywords are:
461 The architecture page size of the running system.
465 Megabytes of total memory in the system.
469 Number of online available CPUs.
471 These can be used on the command line or in the job file, and will be
472 automatically substituted with the current system values when the job is
473 run. Simple math is also supported on these keywords, so you can perform actions
478 and get that properly expanded to 8 times the size of memory in the machine.
484 This section describes in details each parameter associated with a job. Some
485 parameters take an option of a given type, such as an integer or a
486 string. Anywhere a numeric value is required, an arithmetic expression may be
487 used, provided it is surrounded by parentheses. Supported operators are:
496 For time values in expressions, units are microseconds by default. This is
497 different than for time values not in expressions (not enclosed in
498 parentheses). The following types are used:
505 String: A sequence of alphanumeric characters.
508 Integer with possible time suffix. Without a unit value is interpreted as
509 seconds unless otherwise specified. Accepts a suffix of 'd' for days, 'h' for
510 hours, 'm' for minutes, 's' for seconds, 'ms' (or 'msec') for milliseconds and
511 'us' (or 'usec') for microseconds. For example, use 10m for 10 minutes.
516 Integer. A whole number value, which may contain an integer prefix
517 and an integer suffix:
519 [*integer prefix*] **number** [*integer suffix*]
521 The optional *integer prefix* specifies the number's base. The default
522 is decimal. *0x* specifies hexadecimal.
524 The optional *integer suffix* specifies the number's units, and includes an
525 optional unit prefix and an optional unit. For quantities of data, the
526 default unit is bytes. For quantities of time, the default unit is seconds
527 unless otherwise specified.
529 With :option:`kb_base`\=1000, fio follows international standards for unit
530 prefixes. To specify power-of-10 decimal values defined in the
531 International System of Units (SI):
533 * *K* -- means kilo (K) or 1000
534 * *M* -- means mega (M) or 1000**2
535 * *G* -- means giga (G) or 1000**3
536 * *T* -- means tera (T) or 1000**4
537 * *P* -- means peta (P) or 1000**5
539 To specify power-of-2 binary values defined in IEC 80000-13:
541 * *Ki* -- means kibi (Ki) or 1024
542 * *Mi* -- means mebi (Mi) or 1024**2
543 * *Gi* -- means gibi (Gi) or 1024**3
544 * *Ti* -- means tebi (Ti) or 1024**4
545 * *Pi* -- means pebi (Pi) or 1024**5
547 For Zone Block Device Mode:
550 With :option:`kb_base`\=1024 (the default), the unit prefixes are opposite
551 from those specified in the SI and IEC 80000-13 standards to provide
552 compatibility with old scripts. For example, 4k means 4096.
554 For quantities of data, an optional unit of 'B' may be included
555 (e.g., 'kB' is the same as 'k').
557 The *integer suffix* is not case sensitive (e.g., m/mi mean mebi/mega,
558 not milli). 'b' and 'B' both mean byte, not bit.
560 Examples with :option:`kb_base`\=1000:
562 * *4 KiB*: 4096, 4096b, 4096B, 4ki, 4kib, 4kiB, 4Ki, 4KiB
563 * *1 MiB*: 1048576, 1mi, 1024ki
564 * *1 MB*: 1000000, 1m, 1000k
565 * *1 TiB*: 1099511627776, 1ti, 1024gi, 1048576mi
566 * *1 TB*: 1000000000, 1t, 1000m, 1000000k
568 Examples with :option:`kb_base`\=1024 (default):
570 * *4 KiB*: 4096, 4096b, 4096B, 4k, 4kb, 4kB, 4K, 4KB
571 * *1 MiB*: 1048576, 1m, 1024k
572 * *1 MB*: 1000000, 1mi, 1000ki
573 * *1 TiB*: 1099511627776, 1t, 1024g, 1048576m
574 * *1 TB*: 1000000000, 1ti, 1000mi, 1000000ki
576 To specify times (units are not case sensitive):
580 * *M* -- means minutes
581 * *s* -- or sec means seconds (default)
582 * *ms* -- or *msec* means milliseconds
583 * *us* -- or *usec* means microseconds
585 If the option accepts an upper and lower range, use a colon ':' or
586 minus '-' to separate such values. See :ref:`irange <irange>`.
587 If the lower value specified happens to be larger than the upper value
588 the two values are swapped.
593 Boolean. Usually parsed as an integer, however only defined for
594 true and false (1 and 0).
599 Integer range with suffix. Allows value range to be given, such as
600 1024-4096. A colon may also be used as the separator, e.g. 1k:4k. If the
601 option allows two sets of ranges, they can be specified with a ',' or '/'
602 delimiter: 1k-4k/8k-32k. Also see :ref:`int <int>`.
605 A list of floating point numbers, separated by a ':' character.
607 With the above in mind, here follows the complete list of fio job parameters.
613 .. option:: kb_base=int
615 Select the interpretation of unit prefixes in input parameters.
618 Inputs comply with IEC 80000-13 and the International
619 System of Units (SI). Use:
621 - power-of-2 values with IEC prefixes (e.g., KiB)
622 - power-of-10 values with SI prefixes (e.g., kB)
625 Compatibility mode (default). To avoid breaking old scripts:
627 - power-of-2 values with SI prefixes
628 - power-of-10 values with IEC prefixes
630 See :option:`bs` for more details on input parameters.
632 Outputs always use correct prefixes. Most outputs include both
635 bw=2383.3kB/s (2327.4KiB/s)
637 If only one value is reported, then kb_base selects the one to use:
639 **1000** -- SI prefixes
641 **1024** -- IEC prefixes
643 .. option:: unit_base=int
645 Base unit for reporting. Allowed values are:
648 Use auto-detection (default).
660 ASCII name of the job. This may be used to override the name printed by fio
661 for this job. Otherwise the job name is used. On the command line this
662 parameter has the special purpose of also signaling the start of a new job.
664 .. option:: description=str
666 Text description of the job. Doesn't do anything except dump this text
667 description when this job is run. It's not parsed.
669 .. option:: loops=int
671 Run the specified number of iterations of this job. Used to repeat the same
672 workload a given number of times. Defaults to 1.
674 .. option:: numjobs=int
676 Create the specified number of clones of this job. Each clone of job
677 is spawned as an independent thread or process. May be used to setup a
678 larger number of threads/processes doing the same thing. Each thread is
679 reported separately; to see statistics for all clones as a whole, use
680 :option:`group_reporting` in conjunction with :option:`new_group`.
681 See :option:`--max-jobs`. Default: 1.
684 Time related parameters
685 ~~~~~~~~~~~~~~~~~~~~~~~
687 .. option:: runtime=time
689 Tell fio to terminate processing after the specified period of time. It
690 can be quite hard to determine for how long a specified job will run, so
691 this parameter is handy to cap the total runtime to a given time. When
692 the unit is omitted, the value is interpreted in seconds.
694 .. option:: time_based
696 If set, fio will run for the duration of the :option:`runtime` specified
697 even if the file(s) are completely read or written. It will simply loop over
698 the same workload as many times as the :option:`runtime` allows.
700 .. option:: startdelay=irange(time)
702 Delay the start of job for the specified amount of time. Can be a single
703 value or a range. When given as a range, each thread will choose a value
704 randomly from within the range. Value is in seconds if a unit is omitted.
706 .. option:: ramp_time=time
708 If set, fio will run the specified workload for this amount of time before
709 logging any performance numbers. Useful for letting performance settle
710 before logging results, thus minimizing the runtime required for stable
711 results. Note that the ``ramp_time`` is considered lead in time for a job,
712 thus it will increase the total runtime if a special timeout or
713 :option:`runtime` is specified. When the unit is omitted, the value is
716 .. option:: clocksource=str
718 Use the given clocksource as the base of timing. The supported options are:
721 :manpage:`gettimeofday(2)`
724 :manpage:`clock_gettime(2)`
727 Internal CPU clock source
729 cpu is the preferred clocksource if it is reliable, as it is very fast (and
730 fio is heavy on time calls). Fio will automatically use this clocksource if
731 it's supported and considered reliable on the system it is running on,
732 unless another clocksource is specifically set. For x86/x86-64 CPUs, this
733 means supporting TSC Invariant.
735 .. option:: gtod_reduce=bool
737 Enable all of the :manpage:`gettimeofday(2)` reducing options
738 (:option:`disable_clat`, :option:`disable_slat`, :option:`disable_bw_measurement`) plus
739 reduce precision of the timeout somewhat to really shrink the
740 :manpage:`gettimeofday(2)` call count. With this option enabled, we only do
741 about 0.4% of the :manpage:`gettimeofday(2)` calls we would have done if all
742 time keeping was enabled.
744 .. option:: gtod_cpu=int
746 Sometimes it's cheaper to dedicate a single thread of execution to just
747 getting the current time. Fio (and databases, for instance) are very
748 intensive on :manpage:`gettimeofday(2)` calls. With this option, you can set
749 one CPU aside for doing nothing but logging current time to a shared memory
750 location. Then the other threads/processes that run I/O workloads need only
751 copy that segment, instead of entering the kernel with a
752 :manpage:`gettimeofday(2)` call. The CPU set aside for doing these time
753 calls will be excluded from other uses. Fio will manually clear it from the
754 CPU mask of other jobs.
760 .. option:: directory=str
762 Prefix filenames with this directory. Used to place files in a different
763 location than :file:`./`. You can specify a number of directories by
764 separating the names with a ':' character. These directories will be
765 assigned equally distributed to job clones created by :option:`numjobs` as
766 long as they are using generated filenames. If specific `filename(s)` are
767 set fio will use the first listed directory, and thereby matching the
768 `filename` semantic (which generates a file for each clone if not
769 specified, but lets all clones use the same file if set).
771 See the :option:`filename` option for information on how to escape "``:``"
772 characters within the directory path itself.
774 Note: To control the directory fio will use for internal state files
775 use :option:`--aux-path`.
777 .. option:: filename=str
779 Fio normally makes up a `filename` based on the job name, thread number, and
780 file number (see :option:`filename_format`). If you want to share files
781 between threads in a job or several
782 jobs with fixed file paths, specify a `filename` for each of them to override
783 the default. If the ioengine is file based, you can specify a number of files
784 by separating the names with a ':' colon. So if you wanted a job to open
785 :file:`/dev/sda` and :file:`/dev/sdb` as the two working files, you would use
786 ``filename=/dev/sda:/dev/sdb``. This also means that whenever this option is
787 specified, :option:`nrfiles` is ignored. The size of regular files specified
788 by this option will be :option:`size` divided by number of files unless an
789 explicit size is specified by :option:`filesize`.
791 Each colon in the wanted path must be escaped with a ``\``
792 character. For instance, if the path is :file:`/dev/dsk/foo@3,0:c` then you
793 would use ``filename=/dev/dsk/foo@3,0\:c`` and if the path is
794 :file:`F:\\filename` then you would use ``filename=F\:\filename``.
796 On Windows, disk devices are accessed as :file:`\\\\.\\PhysicalDrive0` for
797 the first device, :file:`\\\\.\\PhysicalDrive1` for the second etc.
798 Note: Windows and FreeBSD prevent write access to areas
799 of the disk containing in-use data (e.g. filesystems).
801 The filename "`-`" is a reserved name, meaning *stdin* or *stdout*. Which
802 of the two depends on the read/write direction set.
804 .. option:: filename_format=str
806 If sharing multiple files between jobs, it is usually necessary to have fio
807 generate the exact names that you want. By default, fio will name a file
808 based on the default file format specification of
809 :file:`jobname.jobnumber.filenumber`. With this option, that can be
810 customized. Fio will recognize and replace the following keywords in this
814 The name of the worker thread or process.
816 IP of the fio process when using client/server mode.
818 The incremental number of the worker thread or process.
820 The incremental number of the file for that worker thread or
823 To have dependent jobs share a set of files, this option can be set to have
824 fio generate filenames that are shared between the two. For instance, if
825 :file:`testfiles.$filenum` is specified, file number 4 for any job will be
826 named :file:`testfiles.4`. The default of :file:`$jobname.$jobnum.$filenum`
827 will be used if no other format specifier is given.
829 If you specify a path then the directories will be created up to the
830 main directory for the file. So for example if you specify
831 ``filename_format=a/b/c/$jobnum`` then the directories a/b/c will be
832 created before the file setup part of the job. If you specify
833 :option:`directory` then the path will be relative that directory,
834 otherwise it is treated as the absolute path.
836 .. option:: unique_filename=bool
838 To avoid collisions between networked clients, fio defaults to prefixing any
839 generated filenames (with a directory specified) with the source of the
840 client connecting. To disable this behavior, set this option to 0.
842 .. option:: opendir=str
844 Recursively open any files below directory `str`.
846 .. option:: lockfile=str
848 Fio defaults to not locking any files before it does I/O to them. If a file
849 or file descriptor is shared, fio can serialize I/O to that file to make the
850 end result consistent. This is usual for emulating real workloads that share
851 files. The lock modes are:
854 No locking. The default.
856 Only one thread or process may do I/O at a time, excluding all
859 Read-write locking on the file. Many readers may
860 access the file at the same time, but writes get exclusive access.
862 .. option:: nrfiles=int
864 Number of files to use for this job. Defaults to 1. The size of files
865 will be :option:`size` divided by this unless explicit size is specified by
866 :option:`filesize`. Files are created for each thread separately, and each
867 file will have a file number within its name by default, as explained in
868 :option:`filename` section.
871 .. option:: openfiles=int
873 Number of files to keep open at the same time. Defaults to the same as
874 :option:`nrfiles`, can be set smaller to limit the number simultaneous
877 .. option:: file_service_type=str
879 Defines how fio decides which file from a job to service next. The following
883 Choose a file at random.
886 Round robin over opened files. This is the default.
889 Finish one file before moving on to the next. Multiple files can
890 still be open depending on :option:`openfiles`.
893 Use a *Zipf* distribution to decide what file to access.
896 Use a *Pareto* distribution to decide what file to access.
899 Use a *Gaussian* (normal) distribution to decide what file to
905 For *random*, *roundrobin*, and *sequential*, a postfix can be appended to
906 tell fio how many I/Os to issue before switching to a new file. For example,
907 specifying ``file_service_type=random:8`` would cause fio to issue
908 8 I/Os before selecting a new file at random. For the non-uniform
909 distributions, a floating point postfix can be given to influence how the
910 distribution is skewed. See :option:`random_distribution` for a description
911 of how that would work.
913 .. option:: ioscheduler=str
915 Attempt to switch the device hosting the file to the specified I/O scheduler
918 .. option:: create_serialize=bool
920 If true, serialize the file creation for the jobs. This may be handy to
921 avoid interleaving of data files, which may greatly depend on the filesystem
922 used and even the number of processors in the system. Default: true.
924 .. option:: create_fsync=bool
926 :manpage:`fsync(2)` the data file after creation. This is the default.
928 .. option:: create_on_open=bool
930 If true, don't pre-create files but allow the job's open() to create a file
931 when it's time to do I/O. Default: false -- pre-create all necessary files
934 .. option:: create_only=bool
936 If true, fio will only run the setup phase of the job. If files need to be
937 laid out or updated on disk, only that will be done -- the actual job contents
938 are not executed. Default: false.
940 .. option:: allow_file_create=bool
942 If true, fio is permitted to create files as part of its workload. If this
943 option is false, then fio will error out if
944 the files it needs to use don't already exist. Default: true.
946 .. option:: allow_mounted_write=bool
948 If this isn't set, fio will abort jobs that are destructive (e.g. that write)
949 to what appears to be a mounted device or partition. This should help catch
950 creating inadvertently destructive tests, not realizing that the test will
951 destroy data on the mounted file system. Note that some platforms don't allow
952 writing against a mounted device regardless of this option. Default: false.
954 .. option:: pre_read=bool
956 If this is given, files will be pre-read into memory before starting the
957 given I/O operation. This will also clear the :option:`invalidate` flag,
958 since it is pointless to pre-read and then drop the cache. This will only
959 work for I/O engines that are seek-able, since they allow you to read the
960 same data multiple times. Thus it will not work on non-seekable I/O engines
961 (e.g. network, splice). Default: false.
963 .. option:: unlink=bool
965 Unlink the job files when done. Not the default, as repeated runs of that
966 job would then waste time recreating the file set again and again. Default:
969 .. option:: unlink_each_loop=bool
971 Unlink job files after each iteration or loop. Default: false.
973 .. option:: zonemode=str
978 The :option:`zonerange`, :option:`zonesize`,
979 :option `zonecapacity` and option:`zoneskip`
980 parameters are ignored.
982 I/O happens in a single zone until
983 :option:`zonesize` bytes have been transferred.
984 After that number of bytes has been
985 transferred processing of the next zone
986 starts. :option `zonecapacity` is ignored.
988 Zoned block device mode. I/O happens
989 sequentially in each zone, even if random I/O
990 has been selected. Random I/O happens across
991 all zones instead of being restricted to a
992 single zone. The :option:`zoneskip` parameter
993 is ignored. :option:`zonerange` and
994 :option:`zonesize` must be identical.
995 Trim is handled using a zone reset operation.
996 Trim only considers non-empty sequential write
997 required and sequential write preferred zones.
999 .. option:: zonerange=int
1001 Size of a single zone. See also :option:`zonesize` and
1004 .. option:: zonesize=int
1006 For :option:`zonemode` =strided, this is the number of bytes to
1007 transfer before skipping :option:`zoneskip` bytes. If this parameter
1008 is smaller than :option:`zonerange` then only a fraction of each zone
1009 with :option:`zonerange` bytes will be accessed. If this parameter is
1010 larger than :option:`zonerange` then each zone will be accessed
1011 multiple times before skipping to the next zone.
1013 For :option:`zonemode` =zbd, this is the size of a single zone. The
1014 :option:`zonerange` parameter is ignored in this mode.
1017 .. option:: zonecapacity=int
1019 For :option:`zonemode` =zbd, this defines the capacity of a single zone,
1020 which is the accessible area starting from the zone start address.
1021 This parameter only applies when using :option:`zonemode` =zbd in
1022 combination with regular block devices. If not specified it defaults to
1023 the zone size. If the target device is a zoned block device, the zone
1024 capacity is obtained from the device information and this option is
1027 .. option:: zoneskip=int
1029 For :option:`zonemode` =strided, the number of bytes to skip after
1030 :option:`zonesize` bytes of data have been transferred. This parameter
1031 must be zero for :option:`zonemode` =zbd.
1033 .. option:: read_beyond_wp=bool
1035 This parameter applies to :option:`zonemode` =zbd only.
1037 Zoned block devices are block devices that consist of multiple zones.
1038 Each zone has a type, e.g. conventional or sequential. A conventional
1039 zone can be written at any offset that is a multiple of the block
1040 size. Sequential zones must be written sequentially. The position at
1041 which a write must occur is called the write pointer. A zoned block
1042 device can be either drive managed, host managed or host aware. For
1043 host managed devices the host must ensure that writes happen
1044 sequentially. Fio recognizes host managed devices and serializes
1045 writes to sequential zones for these devices.
1047 If a read occurs in a sequential zone beyond the write pointer then
1048 the zoned block device will complete the read without reading any data
1049 from the storage medium. Since such reads lead to unrealistically high
1050 bandwidth and IOPS numbers fio only reads beyond the write pointer if
1051 explicitly told to do so. Default: false.
1053 .. option:: max_open_zones=int
1055 A zone of a zoned block device is in the open state when it is partially
1056 written (i.e. not all sectors of the zone have been written). Zoned
1057 block devices may have a limit on the total number of zones that can
1058 be simultaneously in the open state, that is, the number of zones that
1059 can be written to simultaneously. The :option:`max_open_zones` parameter
1060 limits the number of zones to which write commands are issued by all fio
1061 jobs, that is, limits the number of zones that will be in the open
1062 state. This parameter is relevant only if the :option:`zonemode` =zbd is
1063 used. The default value is always equal to maximum number of open zones
1064 of the target zoned block device and a value higher than this limit
1065 cannot be specified by users unless the option
1066 :option:`ignore_zone_limits` is specified. When
1067 :option:`ignore_zone_limits` is specified or the target device has no
1068 limit on the number of zones that can be in an open state,
1069 :option:`max_open_zones` can specify 0 to disable any limit on the
1070 number of zones that can be simultaneously written to by all jobs.
1072 .. option:: job_max_open_zones=int
1074 In the same manner as :option:`max_open_zones`, limit the number of open
1075 zones per fio job, that is, the number of zones that a single job can
1076 simultaneously write to. A value of zero indicates no limit.
1079 .. option:: ignore_zone_limits=bool
1081 If this option is used, fio will ignore the maximum number of open
1082 zones limit of the zoned block device in use, thus allowing the
1083 option :option:`max_open_zones` value to be larger than the device
1084 reported limit. Default: false.
1086 .. option:: zone_reset_threshold=float
1088 A number between zero and one that indicates the ratio of written bytes
1089 in the zones with write pointers in the IO range to the size of the IO
1090 range. When current ratio is above this ratio, zones are reset
1091 periodically as :option:`zone_reset_frequency` specifies. If there are
1092 multiple jobs when using this option, the IO range for all write jobs
1095 .. option:: zone_reset_frequency=float
1097 A number between zero and one that indicates how often a zone reset
1098 should be issued if the zone reset threshold has been exceeded. A zone
1099 reset is submitted after each (1 / zone_reset_frequency) write
1100 requests. This and the previous parameter can be used to simulate
1101 garbage collection activity.
1107 .. option:: direct=bool
1109 If value is true, use non-buffered I/O. This is usually O_DIRECT. Note that
1110 OpenBSD and ZFS on Solaris don't support direct I/O. On Windows the synchronous
1111 ioengines don't support direct I/O. Default: false.
1113 .. option:: buffered=bool
1115 If value is true, use buffered I/O. This is the opposite of the
1116 :option:`direct` option. Defaults to true.
1118 .. option:: readwrite=str, rw=str
1120 Type of I/O pattern. Accepted values are:
1127 Sequential trims (Linux block devices and SCSI
1128 character devices only).
1134 Random trims (Linux block devices and SCSI
1135 character devices only).
1137 Sequential mixed reads and writes.
1139 Random mixed reads and writes.
1141 Sequential trim+write sequences. Blocks will be trimmed first,
1142 then the same blocks will be written to. So if ``io_size=64K``
1143 is specified, Fio will trim a total of 64K bytes and also
1144 write 64K bytes on the same trimmed blocks. This behaviour
1145 will be consistent with ``number_ios`` or other Fio options
1146 limiting the total bytes or number of I/O's.
1148 Like trimwrite, but uses random offsets rather
1149 than sequential writes.
1151 Fio defaults to read if the option is not specified. For the mixed I/O
1152 types, the default is to split them 50/50. For certain types of I/O the
1153 result may still be skewed a bit, since the speed may be different.
1155 It is possible to specify the number of I/Os to do before getting a new
1156 offset by appending ``:<nr>`` to the end of the string given. For a
1157 random read, it would look like ``rw=randread:8`` for passing in an offset
1158 modifier with a value of 8. If the suffix is used with a sequential I/O
1159 pattern, then the *<nr>* value specified will be **added** to the generated
1160 offset for each I/O turning sequential I/O into sequential I/O with holes.
1161 For instance, using ``rw=write:4k`` will skip 4k for every write. Also see
1162 the :option:`rw_sequencer` option.
1164 .. option:: rw_sequencer=str
1166 If an offset modifier is given by appending a number to the ``rw=<str>``
1167 line, then this option controls how that number modifies the I/O offset
1168 being generated. Accepted values are:
1171 Generate sequential offset.
1173 Generate the same offset.
1175 ``sequential`` is only useful for random I/O, where fio would normally
1176 generate a new random offset for every I/O. If you append e.g. 8 to
1177 randread, i.e. ``rw=randread:8`` you would get a new random offset for
1178 every 8 I/Os. The result would be a sequence of 8 sequential offsets
1179 with a random starting point. However this behavior may change if a
1180 sequential I/O reaches end of the file. As sequential I/O is already
1181 sequential, setting ``sequential`` for that would not result in any
1182 difference. ``identical`` behaves in a similar fashion, except it sends
1183 the same offset 8 number of times before generating a new offset.
1188 rw_sequencer=sequential
1191 The generated sequence of offsets will look like this:
1192 4k, 8k, 12k, 16k, 20k, 24k, 28k, 32k, 92k, 96k, 100k, 104k, 108k,
1193 112k, 116k, 120k, 48k, 52k ...
1198 rw_sequencer=identical
1201 The generated sequence of offsets will look like this:
1202 4k, 4k, 4k, 4k, 4k, 4k, 4k, 4k, 92k, 92k, 92k, 92k, 92k, 92k, 92k, 92k,
1205 .. option:: unified_rw_reporting=str
1207 Fio normally reports statistics on a per data direction basis, meaning that
1208 reads, writes, and trims are accounted and reported separately. This option
1209 determines whether fio reports the results normally, summed together, or as
1211 Accepted values are:
1214 Normal statistics reporting.
1217 Statistics are summed per data direction and reported together.
1220 Statistics are reported normally, followed by the mixed statistics.
1223 Backward-compatible alias for **none**.
1226 Backward-compatible alias for **mixed**.
1231 .. option:: randrepeat=bool
1233 Seed the random number generator used for random I/O patterns in a
1234 predictable way so the pattern is repeatable across runs. Default: true.
1236 .. option:: allrandrepeat=bool
1238 Seed all random number generators in a predictable way so results are
1239 repeatable across runs. Default: false.
1241 .. option:: randseed=int
1243 Seed the random number generators based on this seed value, to be able to
1244 control what sequence of output is being generated. If not set, the random
1245 sequence depends on the :option:`randrepeat` setting.
1247 .. option:: fallocate=str
1249 Whether pre-allocation is performed when laying down files.
1250 Accepted values are:
1253 Do not pre-allocate space.
1256 Use a platform's native pre-allocation call but fall back to
1257 **none** behavior if it fails/is not implemented.
1260 Pre-allocate via :manpage:`posix_fallocate(3)`.
1263 Pre-allocate via :manpage:`fallocate(2)` with
1264 FALLOC_FL_KEEP_SIZE set.
1267 Extend file to final size via :manpage:`ftruncate(2)`
1268 instead of allocating.
1271 Backward-compatible alias for **none**.
1274 Backward-compatible alias for **posix**.
1276 May not be available on all supported platforms. **keep** is only available
1277 on Linux. If using ZFS on Solaris this cannot be set to **posix**
1278 because ZFS doesn't support pre-allocation. Default: **native** if any
1279 pre-allocation methods except **truncate** are available, **none** if not.
1281 Note that using **truncate** on Windows will interact surprisingly
1282 with non-sequential write patterns. When writing to a file that has
1283 been extended by setting the end-of-file information, Windows will
1284 backfill the unwritten portion of the file up to that offset with
1285 zeroes before issuing the new write. This means that a single small
1286 write to the end of an extended file will stall until the entire
1287 file has been filled with zeroes.
1289 .. option:: fadvise_hint=str
1291 Use :manpage:`posix_fadvise(2)` or :manpage:`posix_fadvise(2)` to
1292 advise the kernel on what I/O patterns are likely to be issued.
1293 Accepted values are:
1296 Backwards-compatible hint for "no hint".
1299 Backwards compatible hint for "advise with fio workload type". This
1300 uses **FADV_RANDOM** for a random workload, and **FADV_SEQUENTIAL**
1301 for a sequential workload.
1304 Advise using **FADV_SEQUENTIAL**.
1307 Advise using **FADV_RANDOM**.
1309 .. option:: write_hint=str
1311 Use :manpage:`fcntl(2)` to advise the kernel what life time to expect
1312 from a write. Only supported on Linux, as of version 4.13. Accepted
1316 No particular life time associated with this file.
1319 Data written to this file has a short life time.
1322 Data written to this file has a medium life time.
1325 Data written to this file has a long life time.
1328 Data written to this file has a very long life time.
1330 The values are all relative to each other, and no absolute meaning
1331 should be associated with them.
1333 .. option:: offset=int
1335 Start I/O at the provided offset in the file, given as either a fixed size in
1336 bytes, zones or a percentage. If a percentage is given, the generated offset will be
1337 aligned to the minimum ``blocksize`` or to the value of ``offset_align`` if
1338 provided. Data before the given offset will not be touched. This
1339 effectively caps the file size at `real_size - offset`. Can be combined with
1340 :option:`size` to constrain the start and end range of the I/O workload.
1341 A percentage can be specified by a number between 1 and 100 followed by '%',
1342 for example, ``offset=20%`` to specify 20%. In ZBD mode, value can be set as
1343 number of zones using 'z'.
1345 .. option:: offset_align=int
1347 If set to non-zero value, the byte offset generated by a percentage ``offset``
1348 is aligned upwards to this value. Defaults to 0 meaning that a percentage
1349 offset is aligned to the minimum block size.
1351 .. option:: offset_increment=int
1353 If this is provided, then the real offset becomes `offset + offset_increment
1354 * thread_number`, where the thread number is a counter that starts at 0 and
1355 is incremented for each sub-job (i.e. when :option:`numjobs` option is
1356 specified). This option is useful if there are several jobs which are
1357 intended to operate on a file in parallel disjoint segments, with even
1358 spacing between the starting points. Percentages can be used for this option.
1359 If a percentage is given, the generated offset will be aligned to the minimum
1360 ``blocksize`` or to the value of ``offset_align`` if provided. In ZBD mode, value can
1361 also be set as number of zones using 'z'.
1363 .. option:: number_ios=int
1365 Fio will normally perform I/Os until it has exhausted the size of the region
1366 set by :option:`size`, or if it exhaust the allocated time (or hits an error
1367 condition). With this setting, the range/size can be set independently of
1368 the number of I/Os to perform. When fio reaches this number, it will exit
1369 normally and report status. Note that this does not extend the amount of I/O
1370 that will be done, it will only stop fio if this condition is met before
1371 other end-of-job criteria.
1373 .. option:: fsync=int
1375 If writing to a file, issue an :manpage:`fsync(2)` (or its equivalent) of
1376 the dirty data for every number of blocks given. For example, if you give 32
1377 as a parameter, fio will sync the file after every 32 writes issued. If fio is
1378 using non-buffered I/O, we may not sync the file. The exception is the sg
1379 I/O engine, which synchronizes the disk cache anyway. Defaults to 0, which
1380 means fio does not periodically issue and wait for a sync to complete. Also
1381 see :option:`end_fsync` and :option:`fsync_on_close`.
1383 .. option:: fdatasync=int
1385 Like :option:`fsync` but uses :manpage:`fdatasync(2)` to only sync data and
1386 not metadata blocks. In Windows, DragonFlyBSD or OSX there is no
1387 :manpage:`fdatasync(2)` so this falls back to using :manpage:`fsync(2)`.
1388 Defaults to 0, which means fio does not periodically issue and wait for a
1389 data-only sync to complete.
1391 .. option:: write_barrier=int
1393 Make every `N-th` write a barrier write.
1395 .. option:: sync_file_range=str:int
1397 Use :manpage:`sync_file_range(2)` for every `int` number of write
1398 operations. Fio will track range of writes that have happened since the last
1399 :manpage:`sync_file_range(2)` call. `str` can currently be one or more of:
1402 SYNC_FILE_RANGE_WAIT_BEFORE
1404 SYNC_FILE_RANGE_WRITE
1406 SYNC_FILE_RANGE_WAIT_AFTER
1408 So if you do ``sync_file_range=wait_before,write:8``, fio would use
1409 ``SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE`` for every 8
1410 writes. Also see the :manpage:`sync_file_range(2)` man page. This option is
1413 .. option:: overwrite=bool
1415 If true, writes to a file will always overwrite existing data. If the file
1416 doesn't already exist, it will be created before the write phase begins. If
1417 the file exists and is large enough for the specified write phase, nothing
1418 will be done. Default: false.
1420 .. option:: end_fsync=bool
1422 If true, :manpage:`fsync(2)` file contents when a write stage has completed.
1425 .. option:: fsync_on_close=bool
1427 If true, fio will :manpage:`fsync(2)` a dirty file on close. This differs
1428 from :option:`end_fsync` in that it will happen on every file close, not
1429 just at the end of the job. Default: false.
1431 .. option:: rwmixread=int
1433 Percentage of a mixed workload that should be reads. Default: 50.
1435 .. option:: rwmixwrite=int
1437 Percentage of a mixed workload that should be writes. If both
1438 :option:`rwmixread` and :option:`rwmixwrite` is given and the values do not
1439 add up to 100%, the latter of the two will be used to override the
1440 first. This may interfere with a given rate setting, if fio is asked to
1441 limit reads or writes to a certain rate. If that is the case, then the
1442 distribution may be skewed. Default: 50.
1444 .. option:: random_distribution=str:float[:float][,str:float][,str:float]
1446 By default, fio will use a completely uniform random distribution when asked
1447 to perform random I/O. Sometimes it is useful to skew the distribution in
1448 specific ways, ensuring that some parts of the data is more hot than others.
1449 fio includes the following distribution models:
1452 Uniform random distribution
1461 Normal (Gaussian) distribution
1464 Zoned random distribution
1467 Zone absolute random distribution
1469 When using a **zipf** or **pareto** distribution, an input value is also
1470 needed to define the access pattern. For **zipf**, this is the `Zipf
1471 theta`. For **pareto**, it's the `Pareto power`. Fio includes a test
1472 program, :command:`fio-genzipf`, that can be used visualize what the given input
1473 values will yield in terms of hit rates. If you wanted to use **zipf** with
1474 a `theta` of 1.2, you would use ``random_distribution=zipf:1.2`` as the
1475 option. If a non-uniform model is used, fio will disable use of the random
1476 map. For the **normal** distribution, a normal (Gaussian) deviation is
1477 supplied as a value between 0 and 100.
1479 The second, optional float is allowed for **pareto**, **zipf** and **normal** distributions.
1480 It allows one to set base of distribution in non-default place, giving more control
1481 over most probable outcome. This value is in range [0-1] which maps linearly to
1482 range of possible random values.
1483 Defaults are: random for **pareto** and **zipf**, and 0.5 for **normal**.
1484 If you wanted to use **zipf** with a `theta` of 1.2 centered on 1/4 of allowed value range,
1485 you would use ``random_distribution=zipf:1.2:0.25``.
1487 For a **zoned** distribution, fio supports specifying percentages of I/O
1488 access that should fall within what range of the file or device. For
1489 example, given a criteria of:
1491 * 60% of accesses should be to the first 10%
1492 * 30% of accesses should be to the next 20%
1493 * 8% of accesses should be to the next 30%
1494 * 2% of accesses should be to the next 40%
1496 we can define that through zoning of the random accesses. For the above
1497 example, the user would do::
1499 random_distribution=zoned:60/10:30/20:8/30:2/40
1501 A **zoned_abs** distribution works exactly like the **zoned**, except
1502 that it takes absolute sizes. For example, let's say you wanted to
1503 define access according to the following criteria:
1505 * 60% of accesses should be to the first 20G
1506 * 30% of accesses should be to the next 100G
1507 * 10% of accesses should be to the next 500G
1509 we can define an absolute zoning distribution with:
1511 random_distribution=zoned_abs=60/20G:30/100G:10/500g
1513 For both **zoned** and **zoned_abs**, fio supports defining up to
1516 Similarly to how :option:`bssplit` works for setting ranges and
1517 percentages of block sizes. Like :option:`bssplit`, it's possible to
1518 specify separate zones for reads, writes, and trims. If just one set
1519 is given, it'll apply to all of them. This goes for both **zoned**
1520 **zoned_abs** distributions.
1522 .. option:: percentage_random=int[,int][,int]
1524 For a random workload, set how big a percentage should be random. This
1525 defaults to 100%, in which case the workload is fully random. It can be set
1526 from anywhere from 0 to 100. Setting it to 0 would make the workload fully
1527 sequential. Any setting in between will result in a random mix of sequential
1528 and random I/O, at the given percentages. Comma-separated values may be
1529 specified for reads, writes, and trims as described in :option:`blocksize`.
1531 .. option:: norandommap
1533 Normally fio will cover every block of the file when doing random I/O. If
1534 this option is given, fio will just get a new random offset without looking
1535 at past I/O history. This means that some blocks may not be read or written,
1536 and that some blocks may be read/written more than once. If this option is
1537 used with :option:`verify` and multiple blocksizes (via :option:`bsrange`),
1538 only intact blocks are verified, i.e., partially-overwritten blocks are
1539 ignored. With an async I/O engine and an I/O depth > 1, it is possible for
1540 the same block to be overwritten, which can cause verification errors. Either
1541 do not use norandommap in this case, or also use the lfsr random generator.
1543 .. option:: softrandommap=bool
1545 See :option:`norandommap`. If fio runs with the random block map enabled and
1546 it fails to allocate the map, if this option is set it will continue without
1547 a random block map. As coverage will not be as complete as with random maps,
1548 this option is disabled by default.
1550 .. option:: random_generator=str
1552 Fio supports the following engines for generating I/O offsets for random I/O:
1555 Strong 2^88 cycle random number generator.
1557 Linear feedback shift register generator.
1559 Strong 64-bit 2^258 cycle random number generator.
1561 **tausworthe** is a strong random number generator, but it requires tracking
1562 on the side if we want to ensure that blocks are only read or written
1563 once. **lfsr** guarantees that we never generate the same offset twice, and
1564 it's also less computationally expensive. It's not a true random generator,
1565 however, though for I/O purposes it's typically good enough. **lfsr** only
1566 works with single block sizes, not with workloads that use multiple block
1567 sizes. If used with such a workload, fio may read or write some blocks
1568 multiple times. The default value is **tausworthe**, unless the required
1569 space exceeds 2^32 blocks. If it does, then **tausworthe64** is
1570 selected automatically.
1576 .. option:: blocksize=int[,int][,int], bs=int[,int][,int]
1578 The block size in bytes used for I/O units. Default: 4096. A single value
1579 applies to reads, writes, and trims. Comma-separated values may be
1580 specified for reads, writes, and trims. A value not terminated in a comma
1581 applies to subsequent types.
1586 means 256k for reads, writes and trims.
1589 means 8k for reads, 32k for writes and trims.
1592 means 8k for reads, 32k for writes, and default for trims.
1595 means default for reads, 8k for writes and trims.
1598 means default for reads, 8k for writes, and default for trims.
1600 .. option:: blocksize_range=irange[,irange][,irange], bsrange=irange[,irange][,irange]
1602 A range of block sizes in bytes for I/O units. The issued I/O unit will
1603 always be a multiple of the minimum size, unless
1604 :option:`blocksize_unaligned` is set.
1606 Comma-separated ranges may be specified for reads, writes, and trims as
1607 described in :option:`blocksize`.
1609 Example: ``bsrange=1k-4k,2k-8k``.
1611 .. option:: bssplit=str[,str][,str]
1613 Sometimes you want even finer grained control of the block sizes
1614 issued, not just an even split between them. This option allows you to
1615 weight various block sizes, so that you are able to define a specific
1616 amount of block sizes issued. The format for this option is::
1618 bssplit=blocksize/percentage:blocksize/percentage
1620 for as many block sizes as needed. So if you want to define a workload
1621 that has 50% 64k blocks, 10% 4k blocks, and 40% 32k blocks, you would
1624 bssplit=4k/10:64k/50:32k/40
1626 Ordering does not matter. If the percentage is left blank, fio will
1627 fill in the remaining values evenly. So a bssplit option like this one::
1629 bssplit=4k/50:1k/:32k/
1631 would have 50% 4k ios, and 25% 1k and 32k ios. The percentages always
1632 add up to 100, if bssplit is given a range that adds up to more, it
1635 Comma-separated values may be specified for reads, writes, and trims as
1636 described in :option:`blocksize`.
1638 If you want a workload that has 50% 2k reads and 50% 4k reads, while
1639 having 90% 4k writes and 10% 8k writes, you would specify::
1641 bssplit=2k/50:4k/50,4k/90:8k/10
1643 Fio supports defining up to 64 different weights for each data
1646 .. option:: blocksize_unaligned, bs_unaligned
1648 If set, fio will issue I/O units with any size within
1649 :option:`blocksize_range`, not just multiples of the minimum size. This
1650 typically won't work with direct I/O, as that normally requires sector
1653 .. option:: bs_is_seq_rand=bool
1655 If this option is set, fio will use the normal read,write blocksize settings
1656 as sequential,random blocksize settings instead. Any random read or write
1657 will use the WRITE blocksize settings, and any sequential read or write will
1658 use the READ blocksize settings.
1660 .. option:: blockalign=int[,int][,int], ba=int[,int][,int]
1662 Boundary to which fio will align random I/O units. Default:
1663 :option:`blocksize`. Minimum alignment is typically 512b for using direct
1664 I/O, though it usually depends on the hardware block size. This option is
1665 mutually exclusive with using a random map for files, so it will turn off
1666 that option. Comma-separated values may be specified for reads, writes, and
1667 trims as described in :option:`blocksize`.
1673 .. option:: zero_buffers
1675 Initialize buffers with all zeros. Default: fill buffers with random data.
1677 .. option:: refill_buffers
1679 If this option is given, fio will refill the I/O buffers on every
1680 submit. Only makes sense if :option:`zero_buffers` isn't specified,
1681 naturally. Defaults to being unset i.e., the buffer is only filled at
1682 init time and the data in it is reused when possible but if any of
1683 :option:`verify`, :option:`buffer_compress_percentage` or
1684 :option:`dedupe_percentage` are enabled then `refill_buffers` is also
1685 automatically enabled.
1687 .. option:: scramble_buffers=bool
1689 If :option:`refill_buffers` is too costly and the target is using data
1690 deduplication, then setting this option will slightly modify the I/O buffer
1691 contents to defeat normal de-dupe attempts. This is not enough to defeat
1692 more clever block compression attempts, but it will stop naive dedupe of
1693 blocks. Default: true.
1695 .. option:: buffer_compress_percentage=int
1697 If this is set, then fio will attempt to provide I/O buffer content
1698 (on WRITEs) that compresses to the specified level. Fio does this by
1699 providing a mix of random data followed by fixed pattern data. The
1700 fixed pattern is either zeros, or the pattern specified by
1701 :option:`buffer_pattern`. If the `buffer_pattern` option is used, it
1702 might skew the compression ratio slightly. Setting
1703 `buffer_compress_percentage` to a value other than 100 will also
1704 enable :option:`refill_buffers` in order to reduce the likelihood that
1705 adjacent blocks are so similar that they over compress when seen
1706 together. See :option:`buffer_compress_chunk` for how to set a finer or
1707 coarser granularity for the random/fixed data region. Defaults to unset
1708 i.e., buffer data will not adhere to any compression level.
1710 .. option:: buffer_compress_chunk=int
1712 This setting allows fio to manage how big the random/fixed data region
1713 is when using :option:`buffer_compress_percentage`. When
1714 `buffer_compress_chunk` is set to some non-zero value smaller than the
1715 block size, fio can repeat the random/fixed region throughout the I/O
1716 buffer at the specified interval (which particularly useful when
1717 bigger block sizes are used for a job). When set to 0, fio will use a
1718 chunk size that matches the block size resulting in a single
1719 random/fixed region within the I/O buffer. Defaults to 512. When the
1720 unit is omitted, the value is interpreted in bytes.
1722 .. option:: buffer_pattern=str
1724 If set, fio will fill the I/O buffers with this pattern or with the contents
1725 of a file. If not set, the contents of I/O buffers are defined by the other
1726 options related to buffer contents. The setting can be any pattern of bytes,
1727 and can be prefixed with 0x for hex values. It may also be a string, where
1728 the string must then be wrapped with ``""``. Or it may also be a filename,
1729 where the filename must be wrapped with ``''`` in which case the file is
1730 opened and read. Note that not all the file contents will be read if that
1731 would cause the buffers to overflow. So, for example::
1733 buffer_pattern='filename'
1737 buffer_pattern="abcd"
1745 buffer_pattern=0xdeadface
1747 Also you can combine everything together in any order::
1749 buffer_pattern=0xdeadface"abcd"-12'filename'
1751 .. option:: dedupe_percentage=int
1753 If set, fio will generate this percentage of identical buffers when
1754 writing. These buffers will be naturally dedupable. The contents of the
1755 buffers depend on what other buffer compression settings have been set. It's
1756 possible to have the individual buffers either fully compressible, or not at
1757 all -- this option only controls the distribution of unique buffers. Setting
1758 this option will also enable :option:`refill_buffers` to prevent every buffer
1761 .. option:: dedupe_mode=str
1763 If ``dedupe_percentage=<int>`` is given, then this option controls how fio
1764 generates the dedupe buffers.
1767 Generate dedupe buffers by repeating previous writes
1769 Generate dedupe buffers from working set
1771 ``repeat`` is the default option for fio. Dedupe buffers are generated
1772 by repeating previous unique write.
1774 ``working_set`` is a more realistic workload.
1775 With ``working_set``, ``dedupe_working_set_percentage=<int>`` should be provided.
1776 Given that, fio will use the initial unique write buffers as its working set.
1777 Upon deciding to dedupe, fio will randomly choose a buffer from the working set.
1778 Note that by using ``working_set`` the dedupe percentage will converge
1779 to the desired over time while ``repeat`` maintains the desired percentage
1782 .. option:: dedupe_working_set_percentage=int
1784 If ``dedupe_mode=<str>`` is set to ``working_set``, then this controls
1785 the percentage of size of the file or device used as the buffers
1786 fio will choose to generate the dedupe buffers from
1788 Note that size needs to be explicitly provided and only 1 file per
1791 .. option:: dedupe_global=bool
1793 This controls whether the deduplication buffers will be shared amongst
1794 all jobs that have this option set. The buffers are spread evenly between
1797 .. option:: invalidate=bool
1799 Invalidate the buffer/page cache parts of the files to be used prior to
1800 starting I/O if the platform and file type support it. Defaults to true.
1801 This will be ignored if :option:`pre_read` is also specified for the
1804 .. option:: sync=str
1806 Whether, and what type, of synchronous I/O to use for writes. The allowed
1810 Do not use synchronous IO, the default.
1816 Use synchronous file IO. For the majority of I/O engines,
1817 this means using O_SYNC.
1823 Use synchronous data IO. For the majority of I/O engines,
1824 this means using O_DSYNC.
1827 .. option:: iomem=str, mem=str
1829 Fio can use various types of memory as the I/O unit buffer. The allowed
1833 Use memory from :manpage:`malloc(3)` as the buffers. Default memory
1837 Use shared memory as the buffers. Allocated through
1838 :manpage:`shmget(2)`.
1841 Same as shm, but use huge pages as backing.
1844 Use :manpage:`mmap(2)` to allocate buffers. May either be anonymous memory, or can
1845 be file backed if a filename is given after the option. The format
1846 is `mem=mmap:/path/to/file`.
1849 Use a memory mapped huge file as the buffer backing. Append filename
1850 after mmaphuge, ala `mem=mmaphuge:/hugetlbfs/file`.
1853 Same as mmap, but use a MMAP_SHARED mapping.
1856 Use GPU memory as the buffers for GPUDirect RDMA benchmark.
1857 The :option:`ioengine` must be `rdma`.
1859 The area allocated is a function of the maximum allowed bs size for the job,
1860 multiplied by the I/O depth given. Note that for **shmhuge** and
1861 **mmaphuge** to work, the system must have free huge pages allocated. This
1862 can normally be checked and set by reading/writing
1863 :file:`/proc/sys/vm/nr_hugepages` on a Linux system. Fio assumes a huge page
1864 is 2 or 4MiB in size depending on the platform. So to calculate the
1865 number of huge pages you need for a given job file, add up the I/O
1866 depth of all jobs (normally one unless :option:`iodepth` is used) and
1867 multiply by the maximum bs set. Then divide that number by the huge
1868 page size. You can see the size of the huge pages in
1869 :file:`/proc/meminfo`. If no huge pages are allocated by having a
1870 non-zero number in `nr_hugepages`, using **mmaphuge** or **shmhuge**
1871 will fail. Also see :option:`hugepage-size`.
1873 **mmaphuge** also needs to have hugetlbfs mounted and the file location
1874 should point there. So if it's mounted in :file:`/huge`, you would use
1875 `mem=mmaphuge:/huge/somefile`.
1877 .. option:: iomem_align=int, mem_align=int
1879 This indicates the memory alignment of the I/O memory buffers. Note that
1880 the given alignment is applied to the first I/O unit buffer, if using
1881 :option:`iodepth` the alignment of the following buffers are given by the
1882 :option:`bs` used. In other words, if using a :option:`bs` that is a
1883 multiple of the page sized in the system, all buffers will be aligned to
1884 this value. If using a :option:`bs` that is not page aligned, the alignment
1885 of subsequent I/O memory buffers is the sum of the :option:`iomem_align` and
1888 .. option:: hugepage-size=int
1890 Defines the size of a huge page. Must at least be equal to the system
1891 setting, see :file:`/proc/meminfo` and
1892 :file:`/sys/kernel/mm/hugepages/`. Defaults to 2 or 4MiB depending on
1893 the platform. Should probably always be a multiple of megabytes, so
1894 using ``hugepage-size=Xm`` is the preferred way to set this to avoid
1895 setting a non-pow-2 bad value.
1897 .. option:: lockmem=int
1899 Pin the specified amount of memory with :manpage:`mlock(2)`. Can be used to
1900 simulate a smaller amount of memory. The amount specified is per worker.
1906 .. option:: size=int
1908 The total size of file I/O for each thread of this job. Fio will run until
1909 this many bytes has been transferred, unless runtime is altered by other means
1910 such as (1) :option:`runtime`, (2) :option:`io_size` (3) :option:`number_ios`,
1911 (4) gaps/holes while doing I/O's such as ``rw=read:16K``, or (5) sequential
1912 I/O reaching end of the file which is possible when :option:`percentage_random`
1914 Fio will divide this size between the available files determined by options
1915 such as :option:`nrfiles`, :option:`filename`, unless :option:`filesize` is
1916 specified by the job. If the result of division happens to be 0, the size is
1917 set to the physical size of the given files or devices if they exist.
1918 If this option is not specified, fio will use the full size of the given
1919 files or devices. If the files do not exist, size must be given. It is also
1920 possible to give size as a percentage between 1 and 100. If ``size=20%`` is
1921 given, fio will use 20% of the full size of the given files or devices.
1922 In ZBD mode, value can also be set as number of zones using 'z'.
1923 Can be combined with :option:`offset` to constrain the start and end range
1924 that I/O will be done within.
1926 .. option:: io_size=int, io_limit=int
1928 Normally fio operates within the region set by :option:`size`, which means
1929 that the :option:`size` option sets both the region and size of I/O to be
1930 performed. Sometimes that is not what you want. With this option, it is
1931 possible to define just the amount of I/O that fio should do. For instance,
1932 if :option:`size` is set to 20GiB and :option:`io_size` is set to 5GiB, fio
1933 will perform I/O within the first 20GiB but exit when 5GiB have been
1934 done. The opposite is also possible -- if :option:`size` is set to 20GiB,
1935 and :option:`io_size` is set to 40GiB, then fio will do 40GiB of I/O within
1936 the 0..20GiB region.
1938 .. option:: filesize=irange(int)
1940 Individual file sizes. May be a range, in which case fio will select sizes for
1941 files at random within the given range. If not given, each created file is the
1942 same size. This option overrides :option:`size` in terms of file size, i.e. if
1943 :option:`filesize` is specified then :option:`size` becomes merely the default
1944 for :option:`io_size` and has no effect at all if :option:`io_size` is set
1947 .. option:: file_append=bool
1949 Perform I/O after the end of the file. Normally fio will operate within the
1950 size of a file. If this option is set, then fio will append to the file
1951 instead. This has identical behavior to setting :option:`offset` to the size
1952 of a file. This option is ignored on non-regular files.
1954 .. option:: fill_device=bool, fill_fs=bool
1956 Sets size to something really large and waits for ENOSPC (no space left on
1957 device) or EDQUOT (disk quota exceeded)
1958 as the terminating condition. Only makes sense with sequential
1959 write. For a read workload, the mount point will be filled first then I/O
1960 started on the result. This option doesn't make sense if operating on a raw
1961 device node, since the size of that is already known by the file system.
1962 Additionally, writing beyond end-of-device will not return ENOSPC there.
1968 .. option:: ioengine=str
1970 Defines how the job issues I/O to the file. The following types are defined:
1973 Basic :manpage:`read(2)` or :manpage:`write(2)`
1974 I/O. :manpage:`lseek(2)` is used to position the I/O location.
1975 See :option:`fsync` and :option:`fdatasync` for syncing write I/Os.
1978 Basic :manpage:`pread(2)` or :manpage:`pwrite(2)` I/O. Default on
1979 all supported operating systems except for Windows.
1982 Basic :manpage:`readv(2)` or :manpage:`writev(2)` I/O. Will emulate
1983 queuing by coalescing adjacent I/Os into a single submission.
1986 Basic :manpage:`preadv(2)` or :manpage:`pwritev(2)` I/O.
1989 Basic :manpage:`preadv2(2)` or :manpage:`pwritev2(2)` I/O.
1992 Fast Linux native asynchronous I/O. Supports async IO
1993 for both direct and buffered IO.
1994 This engine defines engine specific options.
1997 Fast Linux native asynchronous I/O for pass through commands.
1998 This engine defines engine specific options.
2001 Linux native asynchronous I/O. Note that Linux may only support
2002 queued behavior with non-buffered I/O (set ``direct=1`` or
2004 This engine defines engine specific options.
2007 POSIX asynchronous I/O using :manpage:`aio_read(3)` and
2008 :manpage:`aio_write(3)`.
2011 Solaris native asynchronous I/O.
2014 Windows native asynchronous I/O. Default on Windows.
2017 File is memory mapped with :manpage:`mmap(2)` and data copied
2018 to/from using :manpage:`memcpy(3)`.
2021 :manpage:`splice(2)` is used to transfer the data and
2022 :manpage:`vmsplice(2)` to transfer data from user space to the
2026 SCSI generic sg v3 I/O. May either be synchronous using the SG_IO
2027 ioctl, or if the target is an sg character device we use
2028 :manpage:`read(2)` and :manpage:`write(2)` for asynchronous
2029 I/O. Requires :option:`filename` option to specify either block or
2030 character devices. This engine supports trim operations.
2031 The sg engine includes engine specific options.
2034 Read, write, trim and ZBC/ZAC operations to a zoned
2035 block device using libzbc library. The target can be
2036 either an SG character device or a block device file.
2039 Doesn't transfer any data, just pretends to. This is mainly used to
2040 exercise fio itself and for debugging/testing purposes.
2043 Transfer over the network to given ``host:port``. Depending on the
2044 :option:`protocol` used, the :option:`hostname`, :option:`port`,
2045 :option:`listen` and :option:`filename` options are used to specify
2046 what sort of connection to make, while the :option:`protocol` option
2047 determines which protocol will be used. This engine defines engine
2051 Like **net**, but uses :manpage:`splice(2)` and
2052 :manpage:`vmsplice(2)` to map data and send/receive.
2053 This engine defines engine specific options.
2056 Doesn't transfer any data, but burns CPU cycles according to the
2057 :option:`cpuload`, :option:`cpuchunks` and :option:`cpumode` options.
2058 Setting :option:`cpuload`\=85 will cause that job to do nothing but burn 85%
2059 of the CPU. In case of SMP machines, use :option:`numjobs`\=<nr_of_cpu>
2060 to get desired CPU usage, as the cpuload only loads a
2061 single CPU at the desired rate. A job never finishes unless there is
2062 at least one non-cpuio job.
2063 Setting :option:`cpumode`\=qsort replace the default noop instructions loop
2064 by a qsort algorithm to consume more energy.
2067 The RDMA I/O engine supports both RDMA memory semantics
2068 (RDMA_WRITE/RDMA_READ) and channel semantics (Send/Recv) for the
2069 InfiniBand, RoCE and iWARP protocols. This engine defines engine
2073 I/O engine that does regular fallocate to simulate data transfer as
2077 does fallocate(,mode = FALLOC_FL_KEEP_SIZE,).
2080 does fallocate(,mode = 0).
2083 does fallocate(,mode = FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE).
2086 I/O engine that sends :manpage:`ftruncate(2)` operations in response
2087 to write (DDIR_WRITE) events. Each ftruncate issued sets the file's
2088 size to the current block offset. :option:`blocksize` is ignored.
2091 I/O engine that does regular EXT4_IOC_MOVE_EXT ioctls to simulate
2092 defragment activity in request to DDIR_WRITE event.
2095 I/O engine supporting direct access to Ceph Reliable Autonomic
2096 Distributed Object Store (RADOS) via librados. This ioengine
2097 defines engine specific options.
2100 I/O engine supporting direct access to Ceph Rados Block Devices
2101 (RBD) via librbd without the need to use the kernel rbd driver. This
2102 ioengine defines engine specific options.
2105 I/O engine supporting GET/PUT requests over HTTP(S) with libcurl to
2106 a WebDAV or S3 endpoint. This ioengine defines engine specific options.
2108 This engine only supports direct IO of iodepth=1; you need to scale this
2109 via numjobs. blocksize defines the size of the objects to be created.
2111 TRIM is translated to object deletion.
2114 Using GlusterFS libgfapi sync interface to direct access to
2115 GlusterFS volumes without having to go through FUSE. This ioengine
2116 defines engine specific options.
2119 Using GlusterFS libgfapi async interface to direct access to
2120 GlusterFS volumes without having to go through FUSE. This ioengine
2121 defines engine specific options.
2124 Read and write through Hadoop (HDFS). The :option:`filename` option
2125 is used to specify host,port of the hdfs name-node to connect. This
2126 engine interprets offsets a little differently. In HDFS, files once
2127 created cannot be modified so random writes are not possible. To
2128 imitate this the libhdfs engine expects a bunch of small files to be
2129 created over HDFS and will randomly pick a file from them
2130 based on the offset generated by fio backend (see the example
2131 job file to create such files, use ``rw=write`` option). Please
2132 note, it may be necessary to set environment variables to work
2133 with HDFS/libhdfs properly. Each job uses its own connection to
2137 Read, write and erase an MTD character device (e.g.,
2138 :file:`/dev/mtd0`). Discards are treated as erases. Depending on the
2139 underlying device type, the I/O may have to go in a certain pattern,
2140 e.g., on NAND, writing sequentially to erase blocks and discarding
2141 before overwriting. The `trimwrite` mode works well for this
2145 Read and write using device DAX to a persistent memory device (e.g.,
2146 /dev/dax0.0) through the PMDK libpmem library.
2149 Prefix to specify loading an external I/O engine object file. Append
2150 the engine filename, e.g. ``ioengine=external:/tmp/foo.o`` to load
2151 ioengine :file:`foo.o` in :file:`/tmp`. The path can be either
2152 absolute or relative. See :file:`engines/skeleton_external.c` for
2153 details of writing an external I/O engine.
2156 Simply create the files and do no I/O to them. You still need to
2157 set `filesize` so that all the accounting still occurs, but no
2158 actual I/O will be done other than creating the file.
2161 Simply do stat() and do no I/O to the file. You need to set 'filesize'
2162 and 'nrfiles', so that files will be created.
2163 This engine is to measure file lookup and meta data access.
2166 Simply delete the files by unlink() and do no I/O to them. You need to set 'filesize'
2167 and 'nrfiles', so that the files will be created.
2168 This engine is to measure file delete.
2171 Read and write using mmap I/O to a file on a filesystem
2172 mounted with DAX on a persistent memory device through the PMDK
2176 Synchronous read and write using DDN's Infinite Memory Engine (IME).
2177 This engine is very basic and issues calls to IME whenever an IO is
2181 Synchronous read and write using DDN's Infinite Memory Engine (IME).
2182 This engine uses iovecs and will try to stack as much IOs as possible
2183 (if the IOs are "contiguous" and the IO depth is not exceeded)
2184 before issuing a call to IME.
2187 Asynchronous read and write using DDN's Infinite Memory Engine (IME).
2188 This engine will try to stack as much IOs as possible by creating
2189 requests for IME. FIO will then decide when to commit these requests.
2192 Read and write iscsi lun with libiscsi.
2195 Read and write a Network Block Device (NBD).
2198 I/O engine supporting libcufile synchronous access to nvidia-fs and a
2199 GPUDirect Storage-supported filesystem. This engine performs
2200 I/O without transferring buffers between user-space and the kernel,
2201 unless :option:`verify` is set or :option:`cuda_io` is `posix`.
2202 :option:`iomem` must not be `cudamalloc`. This ioengine defines
2203 engine specific options.
2206 I/O engine supporting asynchronous read and write operations to the
2207 DAOS File System (DFS) via libdfs.
2210 I/O engine supporting asynchronous read and write operations to
2211 NFS filesystems from userspace via libnfs. This is useful for
2212 achieving higher concurrency and thus throughput than is possible
2216 Execute 3rd party tools. Could be used to perform monitoring during jobs runtime.
2219 I/O engine using the xNVMe C API, for NVMe devices. The xnvme engine provides
2220 flexibility to access GNU/Linux Kernel NVMe driver via libaio, IOCTLs, io_uring,
2221 the SPDK NVMe driver, or your own custom NVMe driver. The xnvme engine includes
2222 engine specific options. (See https://xnvme.io).
2225 Use the libblkio library
2226 (https://gitlab.com/libblkio/libblkio). The specific
2227 *driver* to use must be set using
2228 :option:`libblkio_driver`. If
2229 :option:`mem`/:option:`iomem` is not specified, memory
2230 allocation is delegated to libblkio (and so is
2231 guaranteed to work with the selected *driver*). One
2232 libblkio instance is used per process, so all jobs
2233 setting option :option:`thread` will share a single
2234 instance (with one queue per thread) and must specify
2235 compatible options. Note that some drivers don't allow
2236 several instances to access the same device or file
2237 simultaneously, but allow it for threads.
2239 I/O engine specific parameters
2240 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2242 In addition, there are some parameters which are only valid when a specific
2243 :option:`ioengine` is in use. These are used identically to normal parameters,
2244 with the caveat that when used on the command line, they must come after the
2245 :option:`ioengine` that defines them is selected.
2247 .. option:: cmdprio_percentage=int[,int] : [io_uring] [libaio]
2249 Set the percentage of I/O that will be issued with the highest priority.
2250 Default: 0. A single value applies to reads and writes. Comma-separated
2251 values may be specified for reads and writes. For this option to be
2252 effective, NCQ priority must be supported and enabled, and the :option:`direct`
2253 option must be set. fio must also be run as the root user. Unlike
2254 slat/clat/lat stats, which can be tracked and reported independently, per
2255 priority stats only track and report a single type of latency. By default,
2256 completion latency (clat) will be reported, if :option:`lat_percentiles` is
2257 set, total latency (lat) will be reported.
2259 .. option:: cmdprio_class=int[,int] : [io_uring] [libaio]
2261 Set the I/O priority class to use for I/Os that must be issued with
2262 a priority when :option:`cmdprio_percentage` or
2263 :option:`cmdprio_bssplit` is set. If not specified when
2264 :option:`cmdprio_percentage` or :option:`cmdprio_bssplit` is set,
2265 this defaults to the highest priority class. A single value applies
2266 to reads and writes. Comma-separated values may be specified for
2267 reads and writes. See :manpage:`ionice(1)`. See also the
2268 :option:`prioclass` option.
2270 .. option:: cmdprio=int[,int] : [io_uring] [libaio]
2272 Set the I/O priority value to use for I/Os that must be issued with
2273 a priority when :option:`cmdprio_percentage` or
2274 :option:`cmdprio_bssplit` is set. If not specified when
2275 :option:`cmdprio_percentage` or :option:`cmdprio_bssplit` is set,
2277 Linux limits us to a positive value between 0 and 7, with 0 being the
2278 highest. A single value applies to reads and writes. Comma-separated
2279 values may be specified for reads and writes. See :manpage:`ionice(1)`.
2280 Refer to an appropriate manpage for other operating systems since
2281 meaning of priority may differ. See also the :option:`prio` option.
2283 .. option:: cmdprio_bssplit=str[,str] : [io_uring] [libaio]
2285 To get a finer control over I/O priority, this option allows
2286 specifying the percentage of IOs that must have a priority set
2287 depending on the block size of the IO. This option is useful only
2288 when used together with the :option:`bssplit` option, that is,
2289 multiple different block sizes are used for reads and writes.
2291 The first accepted format for this option is the same as the format of
2292 the :option:`bssplit` option:
2294 cmdprio_bssplit=blocksize/percentage:blocksize/percentage
2296 In this case, each entry will use the priority class and priority
2297 level defined by the options :option:`cmdprio_class` and
2298 :option:`cmdprio` respectively.
2300 The second accepted format for this option is:
2302 cmdprio_bssplit=blocksize/percentage/class/level:blocksize/percentage/class/level
2304 In this case, the priority class and priority level is defined inside
2305 each entry. In comparison with the first accepted format, the second
2306 accepted format does not restrict all entries to have the same priority
2307 class and priority level.
2309 For both formats, only the read and write data directions are supported,
2310 values for trim IOs are ignored. This option is mutually exclusive with
2311 the :option:`cmdprio_percentage` option.
2313 .. option:: fixedbufs : [io_uring] [io_uring_cmd]
2315 If fio is asked to do direct IO, then Linux will map pages for each
2316 IO call, and release them when IO is done. If this option is set, the
2317 pages are pre-mapped before IO is started. This eliminates the need to
2318 map and release for each IO. This is more efficient, and reduces the
2321 .. option:: nonvectored=int : [io_uring] [io_uring_cmd]
2323 With this option, fio will use non-vectored read/write commands, where
2324 address must contain the address directly. Default is -1.
2326 .. option:: force_async=int : [io_uring] [io_uring_cmd]
2328 Normal operation for io_uring is to try and issue an sqe as
2329 non-blocking first, and if that fails, execute it in an async manner.
2330 With this option set to N, then every N request fio will ask sqe to
2331 be issued in an async manner. Default is 0.
2333 .. option:: registerfiles : [io_uring] [io_uring_cmd]
2335 With this option, fio registers the set of files being used with the
2336 kernel. This avoids the overhead of managing file counts in the kernel,
2337 making the submission and completion part more lightweight. Required
2338 for the below :option:`sqthread_poll` option.
2340 .. option:: sqthread_poll : [io_uring] [io_uring_cmd] [xnvme]
2342 Normally fio will submit IO by issuing a system call to notify the
2343 kernel of available items in the SQ ring. If this option is set, the
2344 act of submitting IO will be done by a polling thread in the kernel.
2345 This frees up cycles for fio, at the cost of using more CPU in the
2346 system. As submission is just the time it takes to fill in the sqe
2347 entries and any syscall required to wake up the idle kernel thread,
2348 fio will not report submission latencies.
2350 .. option:: sqthread_poll_cpu=int : [io_uring] [io_uring_cmd]
2352 When :option:`sqthread_poll` is set, this option provides a way to
2353 define which CPU should be used for the polling thread.
2355 .. option:: cmd_type=str : [io_uring_cmd]
2357 Specifies the type of uring passthrough command to be used. Supported
2358 value is nvme. Default is nvme.
2362 [io_uring] [io_uring_cmd] [xnvme]
2364 If this option is set, fio will attempt to use polled IO completions.
2365 Normal IO completions generate interrupts to signal the completion of
2366 IO, polled completions do not. Hence they are require active reaping
2367 by the application. The benefits are more efficient IO for high IOPS
2368 scenarios, and lower latencies for low queue depth IO.
2372 Use poll queues. This is incompatible with
2373 :option:`libblkio_wait_mode=eventfd <libblkio_wait_mode>` and
2374 :option:`libblkio_force_enable_completion_eventfd`.
2378 Set RWF_HIPRI on I/O, indicating to the kernel that it's of higher priority
2383 If this option is set, fio will attempt to use polled IO completions.
2384 This will have a similar effect as (io_uring)hipri. Only SCSI READ and
2385 WRITE commands will have the SGV4_FLAG_HIPRI set (not UNMAP (trim) nor
2386 VERIFY). Older versions of the Linux sg driver that do not support
2387 hipri will simply ignore this flag and do normal IO. The Linux SCSI
2388 Low Level Driver (LLD) that "owns" the device also needs to support
2389 hipri (also known as iopoll and mq_poll). The MegaRAID driver is an
2390 example of a SCSI LLD. Default: clear (0) which does normal
2391 (interrupted based) IO.
2393 .. option:: userspace_reap : [libaio]
2395 Normally, with the libaio engine in use, fio will use the
2396 :manpage:`io_getevents(2)` system call to reap newly returned events. With
2397 this flag turned on, the AIO ring will be read directly from user-space to
2398 reap events. The reaping mode is only enabled when polling for a minimum of
2399 0 events (e.g. when :option:`iodepth_batch_complete` `=0`).
2401 .. option:: hipri_percentage : [pvsync2]
2403 When hipri is set this determines the probability of a pvsync2 I/O being high
2404 priority. The default is 100%.
2406 .. option:: nowait=bool : [pvsync2] [libaio] [io_uring] [io_uring_cmd]
2408 By default if a request cannot be executed immediately (e.g. resource starvation,
2409 waiting on locks) it is queued and the initiating process will be blocked until
2410 the required resource becomes free.
2412 This option sets the RWF_NOWAIT flag (supported from the 4.14 Linux kernel) and
2413 the call will return instantly with EAGAIN or a partial result rather than waiting.
2415 It is useful to also use ignore_error=EAGAIN when using this option.
2417 Note: glibc 2.27, 2.28 have a bug in syscall wrappers preadv2, pwritev2.
2418 They return EOPNOTSUP instead of EAGAIN.
2420 For cached I/O, using this option usually means a request operates only with
2421 cached data. Currently the RWF_NOWAIT flag does not supported for cached write.
2423 For direct I/O, requests will only succeed if cache invalidation isn't required,
2424 file blocks are fully allocated and the disk request could be issued immediately.
2426 .. option:: cpuload=int : [cpuio]
2428 Attempt to use the specified percentage of CPU cycles. This is a mandatory
2429 option when using cpuio I/O engine.
2431 .. option:: cpuchunks=int : [cpuio]
2433 Split the load into cycles of the given time. In microseconds.
2435 .. option:: cpumode=str : [cpuio]
2437 Specify how to stress the CPU. It can take these two values:
2440 This is the default where the CPU executes noop instructions.
2442 Replace the default noop instructions loop with a qsort algorithm to
2443 consume more energy.
2445 .. option:: exit_on_io_done=bool : [cpuio]
2447 Detect when I/O threads are done, then exit.
2449 .. option:: namenode=str : [libhdfs]
2451 The hostname or IP address of a HDFS cluster namenode to contact.
2453 .. option:: port=int
2457 The listening port of the HFDS cluster namenode.
2461 The TCP or UDP port to bind to or connect to. If this is used with
2462 :option:`numjobs` to spawn multiple instances of the same job type, then
2463 this will be the starting port number since fio will use a range of
2468 The port to use for RDMA-CM communication. This should be the same value
2469 on the client and the server side.
2471 .. option:: hostname=str : [netsplice] [net] [rdma]
2473 The hostname or IP address to use for TCP, UDP or RDMA-CM based I/O. If the job
2474 is a TCP listener or UDP reader, the hostname is not used and must be omitted
2475 unless it is a valid UDP multicast address.
2477 .. option:: serverip=str : [librpma_*]
2479 The IP address to be used for RDMA-CM based I/O.
2481 .. option:: direct_write_to_pmem=bool : [librpma_*]
2483 Set to 1 only when Direct Write to PMem from the remote host is possible.
2484 Otherwise, set to 0.
2486 .. option:: busy_wait_polling=bool : [librpma_*_server]
2488 Set to 0 to wait for completion instead of busy-wait polling completion.
2491 .. option:: interface=str : [netsplice] [net]
2493 The IP address of the network interface used to send or receive UDP
2496 .. option:: ttl=int : [netsplice] [net]
2498 Time-to-live value for outgoing UDP multicast packets. Default: 1.
2500 .. option:: nodelay=bool : [netsplice] [net]
2502 Set TCP_NODELAY on TCP connections.
2504 .. option:: protocol=str, proto=str : [netsplice] [net]
2506 The network protocol to use. Accepted values are:
2509 Transmission control protocol.
2511 Transmission control protocol V6.
2513 User datagram protocol.
2515 User datagram protocol V6.
2519 When the protocol is TCP or UDP, the port must also be given, as well as the
2520 hostname if the job is a TCP listener or UDP reader. For unix sockets, the
2521 normal :option:`filename` option should be used and the port is invalid.
2523 .. option:: listen : [netsplice] [net]
2525 For TCP network connections, tell fio to listen for incoming connections
2526 rather than initiating an outgoing connection. The :option:`hostname` must
2527 be omitted if this option is used.
2529 .. option:: pingpong : [netsplice] [net]
2531 Normally a network writer will just continue writing data, and a network
2532 reader will just consume packages. If ``pingpong=1`` is set, a writer will
2533 send its normal payload to the reader, then wait for the reader to send the
2534 same payload back. This allows fio to measure network latencies. The
2535 submission and completion latencies then measure local time spent sending or
2536 receiving, and the completion latency measures how long it took for the
2537 other end to receive and send back. For UDP multicast traffic
2538 ``pingpong=1`` should only be set for a single reader when multiple readers
2539 are listening to the same address.
2541 .. option:: window_size : [netsplice] [net]
2543 Set the desired socket buffer size for the connection.
2545 .. option:: mss : [netsplice] [net]
2547 Set the TCP maximum segment size (TCP_MAXSEG).
2549 .. option:: donorname=str : [e4defrag]
2551 File will be used as a block donor (swap extents between files).
2553 .. option:: inplace=int : [e4defrag]
2555 Configure donor file blocks allocation strategy:
2558 Default. Preallocate donor's file on init.
2560 Allocate space immediately inside defragment event, and free right
2563 .. option:: clustername=str : [rbd,rados]
2565 Specifies the name of the Ceph cluster.
2567 .. option:: rbdname=str : [rbd]
2569 Specifies the name of the RBD.
2571 .. option:: clientname=str : [rbd,rados]
2573 Specifies the username (without the 'client.' prefix) used to access the
2574 Ceph cluster. If the *clustername* is specified, the *clientname* shall be
2575 the full *type.id* string. If no type. prefix is given, fio will add
2576 'client.' by default.
2578 .. option:: conf=str : [rados]
2580 Specifies the configuration path of ceph cluster, so conf file does not
2581 have to be /etc/ceph/ceph.conf.
2583 .. option:: busy_poll=bool : [rbd,rados]
2585 Poll store instead of waiting for completion. Usually this provides better
2586 throughput at cost of higher(up to 100%) CPU utilization.
2588 .. option:: touch_objects=bool : [rados]
2590 During initialization, touch (create if do not exist) all objects (files).
2591 Touching all objects affects ceph caches and likely impacts test results.
2594 .. option:: pool=str :
2598 Specifies the name of the Ceph pool containing RBD or RADOS data.
2602 Specify the label or UUID of the DAOS pool to connect to.
2604 .. option:: cont=str : [dfs]
2606 Specify the label or UUID of the DAOS container to open.
2608 .. option:: chunk_size=int
2612 Specify a different chunk size (in bytes) for the dfs file.
2613 Use DAOS container's chunk size by default.
2617 The size of the chunk to use for each file.
2619 .. option:: object_class=str : [dfs]
2621 Specify a different object class for the dfs file.
2622 Use DAOS container's object class by default.
2624 .. option:: skip_bad=bool : [mtd]
2626 Skip operations against known bad blocks.
2628 .. option:: hdfsdirectory : [libhdfs]
2630 libhdfs will create chunk in this HDFS directory.
2632 .. option:: verb=str : [rdma]
2634 The RDMA verb to use on this side of the RDMA ioengine connection. Valid
2635 values are write, read, send and recv. These correspond to the equivalent
2636 RDMA verbs (e.g. write = rdma_write etc.). Note that this only needs to be
2637 specified on the client side of the connection. See the examples folder.
2639 .. option:: bindname=str : [rdma]
2641 The name to use to bind the local RDMA-CM connection to a local RDMA device.
2642 This could be a hostname or an IPv4 or IPv6 address. On the server side this
2643 will be passed into the rdma_bind_addr() function and on the client site it
2644 will be used in the rdma_resolve_add() function. This can be useful when
2645 multiple paths exist between the client and the server or in certain loopback
2648 .. option:: stat_type=str : [filestat]
2650 Specify stat system call type to measure lookup/getattr performance.
2651 Default is **stat** for :manpage:`stat(2)`.
2653 .. option:: readfua=bool : [sg]
2655 With readfua option set to 1, read operations include
2656 the force unit access (fua) flag. Default is 0.
2658 .. option:: writefua=bool : [sg]
2660 With writefua option set to 1, write operations include
2661 the force unit access (fua) flag. Default is 0.
2663 .. option:: sg_write_mode=str : [sg]
2665 Specify the type of write commands to issue. This option can take three values:
2668 This is the default where write opcodes are issued as usual.
2669 **write_and_verify**
2670 Issue WRITE AND VERIFY commands. The BYTCHK bit is set to 0. This
2671 directs the device to carry out a medium verification with no data
2672 comparison. The writefua option is ignored with this selection.
2674 This option is deprecated. Use write_and_verify instead.
2676 Issue WRITE SAME commands. This transfers a single block to the device
2677 and writes this same block of data to a contiguous sequence of LBAs
2678 beginning at the specified offset. fio's block size parameter specifies
2679 the amount of data written with each command. However, the amount of data
2680 actually transferred to the device is equal to the device's block
2681 (sector) size. For a device with 512 byte sectors, blocksize=8k will
2682 write 16 sectors with each command. fio will still generate 8k of data
2683 for each command but only the first 512 bytes will be used and
2684 transferred to the device. The writefua option is ignored with this
2687 This option is deprecated. Use write_same instead.
2689 Issue WRITE SAME(16) commands as above but with the No Data Output
2690 Buffer (NDOB) bit set. No data will be transferred to the device with
2691 this bit set. Data written will be a pre-determined pattern such as
2694 Issue WRITE STREAM(16) commands. Use the **stream_id** option to specify
2695 the stream identifier.
2696 **verify_bytchk_00**
2697 Issue VERIFY commands with BYTCHK set to 00. This directs the
2698 device to carry out a medium verification with no data comparison.
2699 **verify_bytchk_01**
2700 Issue VERIFY commands with BYTCHK set to 01. This directs the device to
2701 compare the data on the device with the data transferred to the device.
2702 **verify_bytchk_11**
2703 Issue VERIFY commands with BYTCHK set to 11. This transfers a
2704 single block to the device and compares the contents of this block with the
2705 data on the device beginning at the specified offset. fio's block size
2706 parameter specifies the total amount of data compared with this command.
2707 However, only one block (sector) worth of data is transferred to the device.
2708 This is similar to the WRITE SAME command except that data is compared instead
2711 .. option:: stream_id=int : [sg]
2713 Set the stream identifier for WRITE STREAM commands. If this is set to 0 (which is not
2714 a valid stream identifier) fio will open a stream and then close it when done. Default
2717 .. option:: http_host=str : [http]
2719 Hostname to connect to. For S3, this could be the bucket hostname.
2720 Default is **localhost**
2722 .. option:: http_user=str : [http]
2724 Username for HTTP authentication.
2726 .. option:: http_pass=str : [http]
2728 Password for HTTP authentication.
2730 .. option:: https=str : [http]
2732 Enable HTTPS instead of http. *on* enables HTTPS; *insecure*
2733 will enable HTTPS, but disable SSL peer verification (use with
2734 caution!). Default is **off**
2736 .. option:: http_mode=str : [http]
2738 Which HTTP access mode to use: *webdav*, *swift*, or *s3*.
2739 Default is **webdav**
2741 .. option:: http_s3_region=str : [http]
2743 The S3 region/zone string.
2744 Default is **us-east-1**
2746 .. option:: http_s3_key=str : [http]
2750 .. option:: http_s3_keyid=str : [http]
2752 The S3 key/access id.
2754 .. option:: http_s3_sse_customer_key=str : [http]
2756 The encryption customer key in SSE server side.
2758 .. option:: http_s3_sse_customer_algorithm=str : [http]
2760 The encryption customer algorithm in SSE server side.
2761 Default is **AES256**
2763 .. option:: http_s3_storage_class=str : [http]
2765 Which storage class to access. User-customizable settings.
2766 Default is **STANDARD**
2768 .. option:: http_swift_auth_token=str : [http]
2770 The Swift auth token. See the example configuration file on how
2773 .. option:: http_verbose=int : [http]
2775 Enable verbose requests from libcurl. Useful for debugging. 1
2776 turns on verbose logging from libcurl, 2 additionally enables
2777 HTTP IO tracing. Default is **0**
2779 .. option:: uri=str : [nbd]
2781 Specify the NBD URI of the server to test. The string
2782 is a standard NBD URI
2783 (see https://github.com/NetworkBlockDevice/nbd/tree/master/doc).
2784 Example URIs: nbd://localhost:10809
2785 nbd+unix:///?socket=/tmp/socket
2786 nbds://tlshost/exportname
2788 .. option:: gpu_dev_ids=str : [libcufile]
2790 Specify the GPU IDs to use with CUDA. This is a colon-separated list of
2791 int. GPUs are assigned to workers roundrobin. Default is 0.
2793 .. option:: cuda_io=str : [libcufile]
2795 Specify the type of I/O to use with CUDA. Default is **cufile**.
2798 Use libcufile and nvidia-fs. This option performs I/O directly
2799 between a GPUDirect Storage filesystem and GPU buffers,
2800 avoiding use of a bounce buffer. If :option:`verify` is set,
2801 cudaMemcpy is used to copy verificaton data between RAM and GPU.
2802 Verification data is copied from RAM to GPU before a write
2803 and from GPU to RAM after a read. :option:`direct` must be 1.
2805 Use POSIX to perform I/O with a RAM buffer, and use cudaMemcpy
2806 to transfer data between RAM and the GPUs. Data is copied from
2807 GPU to RAM before a write and copied from RAM to GPU after a
2808 read. :option:`verify` does not affect use of cudaMemcpy.
2810 .. option:: nfs_url=str : [nfs]
2812 URL in libnfs format, eg nfs://<server|ipv4|ipv6>/path[?arg=val[&arg=val]*]
2813 Refer to the libnfs README for more details.
2815 .. option:: program=str : [exec]
2817 Specify the program to execute.
2819 .. option:: arguments=str : [exec]
2821 Specify arguments to pass to program.
2822 Some special variables can be expanded to pass fio's job details to the program.
2825 Replaced by the duration of the job in seconds.
2827 Replaced by the name of the job.
2829 .. option:: grace_time=int : [exec]
2831 Specify the time between the SIGTERM and SIGKILL signals. Default is 1 second.
2833 .. option:: std_redirect=bool : [exec]
2835 If set, stdout and stderr streams are redirected to files named from the job name. Default is true.
2837 .. option:: xnvme_async=str : [xnvme]
2839 Select the xnvme async command interface. This can take these values.
2842 This is default and use to emulate asynchronous I/O by using a
2843 single thread to create a queue pair on top of a synchronous
2844 I/O interface using the NVMe driver IOCTL.
2846 Emulate an asynchronous I/O interface with a pool of userspace
2847 threads on top of a synchronous I/O interface using the NVMe
2848 driver IOCTL. By default four threads are used.
2850 Linux native asynchronous I/O interface which supports both
2851 direct and buffered I/O.
2853 Fast Linux native asynchronous I/O interface for NVMe pass
2854 through commands. This only works with NVMe character device
2857 Use Linux aio for Asynchronous I/O.
2859 Use the posix asynchronous I/O interface to perform one or
2860 more I/O operations asynchronously.
2862 Use the user-space VFIO-based backend, implemented using
2863 libvfn instead of SPDK.
2865 Do not transfer any data; just pretend to. This is mainly used
2866 for introspective performance evaluation.
2868 .. option:: xnvme_sync=str : [xnvme]
2870 Select the xnvme synchronous command interface. This can take these values.
2873 This is default and uses Linux NVMe Driver ioctl() for
2876 This supports regular as well as vectored pread() and pwrite()
2879 This is the same as psync except that it also supports zone
2880 management commands using Linux block layer IOCTLs.
2882 .. option:: xnvme_admin=str : [xnvme]
2884 Select the xnvme admin command interface. This can take these values.
2887 This is default and uses linux NVMe Driver ioctl() for admin
2890 Use Linux Block Layer ioctl() and sysfs for admin commands.
2892 .. option:: xnvme_dev_nsid=int : [xnvme]
2894 xnvme namespace identifier for userspace NVMe driver, SPDK or vfio.
2896 .. option:: xnvme_dev_subnqn=str : [xnvme]
2898 Sets the subsystem NQN for fabrics. This is for xNVMe to utilize a
2899 fabrics target with multiple systems.
2901 .. option:: xnvme_mem=str : [xnvme]
2903 Select the xnvme memory backend. This can take these values.
2906 This is the default posix memory backend for linux NVMe driver.
2908 Use hugepages, instead of existing posix memory backend. The
2909 memory backend uses hugetlbfs. This require users to allocate
2910 hugepages, mount hugetlbfs and set an enviornment variable for
2913 Uses SPDK's memory allocator.
2915 Uses libvfn's memory allocator. This also specifies the use
2916 of libvfn backend instead of SPDK.
2918 .. option:: xnvme_iovec=int : [xnvme]
2920 If this option is set. xnvme will use vectored read/write commands.
2922 .. option:: libblkio_driver=str : [libblkio]
2924 The libblkio *driver* to use. Different drivers access devices through
2925 different underlying interfaces. Available drivers depend on the
2926 libblkio version in use and are listed at
2927 https://libblkio.gitlab.io/libblkio/blkio.html#drivers
2929 .. option:: libblkio_path=str : [libblkio]
2931 Sets the value of the driver-specific "path" property before connecting
2932 the libblkio instance, which identifies the target device or file on
2933 which to perform I/O. Its exact semantics are driver-dependent and not
2934 all drivers may support it; see
2935 https://libblkio.gitlab.io/libblkio/blkio.html#drivers
2937 .. option:: libblkio_pre_connect_props=str : [libblkio]
2939 A colon-separated list of additional libblkio properties to be set after
2940 creating but before connecting the libblkio instance. Each property must
2941 have the format ``<name>=<value>``. Colons can be escaped as ``\:``.
2942 These are set after the engine sets any other properties, so those can
2943 be overriden. Available properties depend on the libblkio version in use
2945 https://libblkio.gitlab.io/libblkio/blkio.html#properties
2947 .. option:: libblkio_num_entries=int : [libblkio]
2949 Sets the value of the driver-specific "num-entries" property before
2950 starting the libblkio instance. Its exact semantics are driver-dependent
2951 and not all drivers may support it; see
2952 https://libblkio.gitlab.io/libblkio/blkio.html#drivers
2954 .. option:: libblkio_queue_size=int : [libblkio]
2956 Sets the value of the driver-specific "queue-size" property before
2957 starting the libblkio instance. Its exact semantics are driver-dependent
2958 and not all drivers may support it; see
2959 https://libblkio.gitlab.io/libblkio/blkio.html#drivers
2961 .. option:: libblkio_pre_start_props=str : [libblkio]
2963 A colon-separated list of additional libblkio properties to be set after
2964 connecting but before starting the libblkio instance. Each property must
2965 have the format ``<name>=<value>``. Colons can be escaped as ``\:``.
2966 These are set after the engine sets any other properties, so those can
2967 be overriden. Available properties depend on the libblkio version in use
2969 https://libblkio.gitlab.io/libblkio/blkio.html#properties
2971 .. option:: libblkio_vectored : [libblkio]
2973 Submit vectored read and write requests.
2975 .. option:: libblkio_write_zeroes_on_trim : [libblkio]
2977 Submit trims as "write zeroes" requests instead of discard requests.
2979 .. option:: libblkio_wait_mode=str : [libblkio]
2981 How to wait for completions:
2984 Use a blocking call to ``blkioq_do_io()``.
2986 Use a blocking call to ``read()`` on the completion eventfd.
2988 Use a busy loop with a non-blocking call to ``blkioq_do_io()``.
2990 .. option:: libblkio_force_enable_completion_eventfd : [libblkio]
2992 Enable the queue's completion eventfd even when unused. This may impact
2993 performance. The default is to enable it only if
2994 :option:`libblkio_wait_mode=eventfd <libblkio_wait_mode>`.
2999 .. option:: iodepth=int
3001 Number of I/O units to keep in flight against the file. Note that
3002 increasing *iodepth* beyond 1 will not affect synchronous ioengines (except
3003 for small degrees when :option:`verify_async` is in use). Even async
3004 engines may impose OS restrictions causing the desired depth not to be
3005 achieved. This may happen on Linux when using libaio and not setting
3006 :option:`direct`\=1, since buffered I/O is not async on that OS. Keep an
3007 eye on the I/O depth distribution in the fio output to verify that the
3008 achieved depth is as expected. Default: 1.
3010 .. option:: iodepth_batch_submit=int, iodepth_batch=int
3012 This defines how many pieces of I/O to submit at once. It defaults to 1
3013 which means that we submit each I/O as soon as it is available, but can be
3014 raised to submit bigger batches of I/O at the time. If it is set to 0 the
3015 :option:`iodepth` value will be used.
3017 .. option:: iodepth_batch_complete_min=int, iodepth_batch_complete=int
3019 This defines how many pieces of I/O to retrieve at once. It defaults to 1
3020 which means that we'll ask for a minimum of 1 I/O in the retrieval process
3021 from the kernel. The I/O retrieval will go on until we hit the limit set by
3022 :option:`iodepth_low`. If this variable is set to 0, then fio will always
3023 check for completed events before queuing more I/O. This helps reduce I/O
3024 latency, at the cost of more retrieval system calls.
3026 .. option:: iodepth_batch_complete_max=int
3028 This defines maximum pieces of I/O to retrieve at once. This variable should
3029 be used along with :option:`iodepth_batch_complete_min`\=int variable,
3030 specifying the range of min and max amount of I/O which should be
3031 retrieved. By default it is equal to the :option:`iodepth_batch_complete_min`
3036 iodepth_batch_complete_min=1
3037 iodepth_batch_complete_max=<iodepth>
3039 which means that we will retrieve at least 1 I/O and up to the whole
3040 submitted queue depth. If none of I/O has been completed yet, we will wait.
3044 iodepth_batch_complete_min=0
3045 iodepth_batch_complete_max=<iodepth>
3047 which means that we can retrieve up to the whole submitted queue depth, but
3048 if none of I/O has been completed yet, we will NOT wait and immediately exit
3049 the system call. In this example we simply do polling.
3051 .. option:: iodepth_low=int
3053 The low water mark indicating when to start filling the queue
3054 again. Defaults to the same as :option:`iodepth`, meaning that fio will
3055 attempt to keep the queue full at all times. If :option:`iodepth` is set to
3056 e.g. 16 and *iodepth_low* is set to 4, then after fio has filled the queue of
3057 16 requests, it will let the depth drain down to 4 before starting to fill
3060 .. option:: serialize_overlap=bool
3062 Serialize in-flight I/Os that might otherwise cause or suffer from data races.
3063 When two or more I/Os are submitted simultaneously, there is no guarantee that
3064 the I/Os will be processed or completed in the submitted order. Further, if
3065 two or more of those I/Os are writes, any overlapping region between them can
3066 become indeterminate/undefined on certain storage. These issues can cause
3067 verification to fail erratically when at least one of the racing I/Os is
3068 changing data and the overlapping region has a non-zero size. Setting
3069 ``serialize_overlap`` tells fio to avoid provoking this behavior by explicitly
3070 serializing in-flight I/Os that have a non-zero overlap. Note that setting
3071 this option can reduce both performance and the :option:`iodepth` achieved.
3073 This option only applies to I/Os issued for a single job except when it is
3074 enabled along with :option:`io_submit_mode`\=offload. In offload mode, fio
3075 will check for overlap among all I/Os submitted by offload jobs with :option:`serialize_overlap`
3080 .. option:: io_submit_mode=str
3082 This option controls how fio submits the I/O to the I/O engine. The default
3083 is `inline`, which means that the fio job threads submit and reap I/O
3084 directly. If set to `offload`, the job threads will offload I/O submission
3085 to a dedicated pool of I/O threads. This requires some coordination and thus
3086 has a bit of extra overhead, especially for lower queue depth I/O where it
3087 can increase latencies. The benefit is that fio can manage submission rates
3088 independently of the device completion rates. This avoids skewed latency
3089 reporting if I/O gets backed up on the device side (the coordinated omission
3090 problem). Note that this option cannot reliably be used with async IO
3097 .. option:: thinktime=time
3099 Stall the job for the specified period of time after an I/O has completed before issuing the
3100 next. May be used to simulate processing being done by an application.
3101 When the unit is omitted, the value is interpreted in microseconds. See
3102 :option:`thinktime_blocks`, :option:`thinktime_iotime` and :option:`thinktime_spin`.
3104 .. option:: thinktime_spin=time
3106 Only valid if :option:`thinktime` is set - pretend to spend CPU time doing
3107 something with the data received, before falling back to sleeping for the
3108 rest of the period specified by :option:`thinktime`. When the unit is
3109 omitted, the value is interpreted in microseconds.
3111 .. option:: thinktime_blocks=int
3113 Only valid if :option:`thinktime` is set - control how many blocks to issue,
3114 before waiting :option:`thinktime` usecs. If not set, defaults to 1 which will make
3115 fio wait :option:`thinktime` usecs after every block. This effectively makes any
3116 queue depth setting redundant, since no more than 1 I/O will be queued
3117 before we have to complete it and do our :option:`thinktime`. In other words, this
3118 setting effectively caps the queue depth if the latter is larger.
3120 .. option:: thinktime_blocks_type=str
3122 Only valid if :option:`thinktime` is set - control how :option:`thinktime_blocks`
3123 triggers. The default is `complete`, which triggers thinktime when fio completes
3124 :option:`thinktime_blocks` blocks. If this is set to `issue`, then the trigger happens
3127 .. option:: thinktime_iotime=time
3129 Only valid if :option:`thinktime` is set - control :option:`thinktime`
3130 interval by time. The :option:`thinktime` stall is repeated after IOs
3131 are executed for :option:`thinktime_iotime`. For example,
3132 ``--thinktime_iotime=9s --thinktime=1s`` repeat 10-second cycle with IOs
3133 for 9 seconds and stall for 1 second. When the unit is omitted,
3134 :option:`thinktime_iotime` is interpreted as a number of seconds. If
3135 this option is used together with :option:`thinktime_blocks`, the
3136 :option:`thinktime` stall is repeated after :option:`thinktime_iotime`
3137 or after :option:`thinktime_blocks` IOs, whichever happens first.
3139 .. option:: rate=int[,int][,int]
3141 Cap the bandwidth used by this job. The number is in bytes/sec, the normal
3142 suffix rules apply. Comma-separated values may be specified for reads,
3143 writes, and trims as described in :option:`blocksize`.
3145 For example, using `rate=1m,500k` would limit reads to 1MiB/sec and writes to
3146 500KiB/sec. Capping only reads or writes can be done with `rate=,500k` or
3147 `rate=500k,` where the former will only limit writes (to 500KiB/sec) and the
3148 latter will only limit reads.
3150 .. option:: rate_min=int[,int][,int]
3152 Tell fio to do whatever it can to maintain at least this bandwidth. Failing
3153 to meet this requirement will cause the job to exit. Comma-separated values
3154 may be specified for reads, writes, and trims as described in
3155 :option:`blocksize`.
3157 .. option:: rate_iops=int[,int][,int]
3159 Cap the bandwidth to this number of IOPS. Basically the same as
3160 :option:`rate`, just specified independently of bandwidth. If the job is
3161 given a block size range instead of a fixed value, the smallest block size
3162 is used as the metric. Comma-separated values may be specified for reads,
3163 writes, and trims as described in :option:`blocksize`.
3165 .. option:: rate_iops_min=int[,int][,int]
3167 If fio doesn't meet this rate of I/O, it will cause the job to exit.
3168 Comma-separated values may be specified for reads, writes, and trims as
3169 described in :option:`blocksize`.
3171 .. option:: rate_process=str
3173 This option controls how fio manages rated I/O submissions. The default is
3174 `linear`, which submits I/O in a linear fashion with fixed delays between
3175 I/Os that gets adjusted based on I/O completion rates. If this is set to
3176 `poisson`, fio will submit I/O based on a more real world random request
3177 flow, known as the Poisson process
3178 (https://en.wikipedia.org/wiki/Poisson_point_process). The lambda will be
3179 10^6 / IOPS for the given workload.
3181 .. option:: rate_ignore_thinktime=bool
3183 By default, fio will attempt to catch up to the specified rate setting,
3184 if any kind of thinktime setting was used. If this option is set, then
3185 fio will ignore the thinktime and continue doing IO at the specified
3186 rate, instead of entering a catch-up mode after thinktime is done.
3192 .. option:: latency_target=time
3194 If set, fio will attempt to find the max performance point that the given
3195 workload will run at while maintaining a latency below this target. When
3196 the unit is omitted, the value is interpreted in microseconds. See
3197 :option:`latency_window` and :option:`latency_percentile`.
3199 .. option:: latency_window=time
3201 Used with :option:`latency_target` to specify the sample window that the job
3202 is run at varying queue depths to test the performance. When the unit is
3203 omitted, the value is interpreted in microseconds.
3205 .. option:: latency_percentile=float
3207 The percentage of I/Os that must fall within the criteria specified by
3208 :option:`latency_target` and :option:`latency_window`. If not set, this
3209 defaults to 100.0, meaning that all I/Os must be equal or below to the value
3210 set by :option:`latency_target`.
3212 .. option:: latency_run=bool
3214 Used with :option:`latency_target`. If false (default), fio will find
3215 the highest queue depth that meets :option:`latency_target` and exit. If
3216 true, fio will continue running and try to meet :option:`latency_target`
3217 by adjusting queue depth.
3219 .. option:: max_latency=time[,time][,time]
3221 If set, fio will exit the job with an ETIMEDOUT error if it exceeds this
3222 maximum latency. When the unit is omitted, the value is interpreted in
3223 microseconds. Comma-separated values may be specified for reads, writes,
3224 and trims as described in :option:`blocksize`.
3226 .. option:: rate_cycle=int
3228 Average bandwidth for :option:`rate` and :option:`rate_min` over this number
3229 of milliseconds. Defaults to 1000.
3235 .. option:: write_iolog=str
3237 Write the issued I/O patterns to the specified file. See
3238 :option:`read_iolog`. Specify a separate file for each job, otherwise the
3239 iologs will be interspersed and the file may be corrupt. This file will
3240 be opened in append mode.
3242 .. option:: read_iolog=str
3244 Open an iolog with the specified filename and replay the I/O patterns it
3245 contains. This can be used to store a workload and replay it sometime
3246 later. The iolog given may also be a blktrace binary file, which allows fio
3247 to replay a workload captured by :command:`blktrace`. See
3248 :manpage:`blktrace(8)` for how to capture such logging data. For blktrace
3249 replay, the file needs to be turned into a blkparse binary data file first
3250 (``blkparse <device> -o /dev/null -d file_for_fio.bin``).
3251 You can specify a number of files by separating the names with a ':'
3252 character. See the :option:`filename` option for information on how to
3253 escape ':' characters within the file names. These files will
3254 be sequentially assigned to job clones created by :option:`numjobs`.
3255 '-' is a reserved name, meaning read from stdin, notably if
3256 :option:`filename` is set to '-' which means stdin as well, then
3257 this flag can't be set to '-'.
3259 .. option:: read_iolog_chunked=bool
3261 Determines how iolog is read. If false(default) entire :option:`read_iolog`
3262 will be read at once. If selected true, input from iolog will be read
3263 gradually. Useful when iolog is very large, or it is generated.
3265 .. option:: merge_blktrace_file=str
3267 When specified, rather than replaying the logs passed to :option:`read_iolog`,
3268 the logs go through a merge phase which aggregates them into a single
3269 blktrace. The resulting file is then passed on as the :option:`read_iolog`
3270 parameter. The intention here is to make the order of events consistent.
3271 This limits the influence of the scheduler compared to replaying multiple
3272 blktraces via concurrent jobs.
3274 .. option:: merge_blktrace_scalars=float_list
3276 This is a percentage based option that is index paired with the list of
3277 files passed to :option:`read_iolog`. When merging is performed, scale
3278 the time of each event by the corresponding amount. For example,
3279 ``--merge_blktrace_scalars="50:100"`` runs the first trace in halftime
3280 and the second trace in realtime. This knob is separately tunable from
3281 :option:`replay_time_scale` which scales the trace during runtime and
3282 does not change the output of the merge unlike this option.
3284 .. option:: merge_blktrace_iters=float_list
3286 This is a whole number option that is index paired with the list of files
3287 passed to :option:`read_iolog`. When merging is performed, run each trace
3288 for the specified number of iterations. For example,
3289 ``--merge_blktrace_iters="2:1"`` runs the first trace for two iterations
3290 and the second trace for one iteration.
3292 .. option:: replay_no_stall=bool
3294 When replaying I/O with :option:`read_iolog` the default behavior is to
3295 attempt to respect the timestamps within the log and replay them with the
3296 appropriate delay between IOPS. By setting this variable fio will not
3297 respect the timestamps and attempt to replay them as fast as possible while
3298 still respecting ordering. The result is the same I/O pattern to a given
3299 device, but different timings.
3301 .. option:: replay_time_scale=int
3303 When replaying I/O with :option:`read_iolog`, fio will honor the
3304 original timing in the trace. With this option, it's possible to scale
3305 the time. It's a percentage option, if set to 50 it means run at 50%
3306 the original IO rate in the trace. If set to 200, run at twice the
3307 original IO rate. Defaults to 100.
3309 .. option:: replay_redirect=str
3311 While replaying I/O patterns using :option:`read_iolog` the default behavior
3312 is to replay the IOPS onto the major/minor device that each IOP was recorded
3313 from. This is sometimes undesirable because on a different machine those
3314 major/minor numbers can map to a different device. Changing hardware on the
3315 same system can also result in a different major/minor mapping.
3316 ``replay_redirect`` causes all I/Os to be replayed onto the single specified
3317 device regardless of the device it was recorded
3318 from. i.e. :option:`replay_redirect`\= :file:`/dev/sdc` would cause all I/O
3319 in the blktrace or iolog to be replayed onto :file:`/dev/sdc`. This means
3320 multiple devices will be replayed onto a single device, if the trace
3321 contains multiple devices. If you want multiple devices to be replayed
3322 concurrently to multiple redirected devices you must blkparse your trace
3323 into separate traces and replay them with independent fio invocations.
3324 Unfortunately this also breaks the strict time ordering between multiple
3327 .. option:: replay_align=int
3329 Force alignment of the byte offsets in a trace to this value. The value
3330 must be a power of 2.
3332 .. option:: replay_scale=int
3334 Scale byte offsets down by this factor when replaying traces. Should most
3335 likely use :option:`replay_align` as well.
3337 .. option:: replay_skip=str
3339 Sometimes it's useful to skip certain IO types in a replay trace.
3340 This could be, for instance, eliminating the writes in the trace.
3341 Or not replaying the trims/discards, if you are redirecting to
3342 a device that doesn't support them. This option takes a comma
3343 separated list of read, write, trim, sync.
3346 Threads, processes and job synchronization
3347 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3351 Fio defaults to creating jobs by using fork, however if this option is
3352 given, fio will create jobs by using POSIX Threads' function
3353 :manpage:`pthread_create(3)` to create threads instead.
3355 .. option:: wait_for=str
3357 If set, the current job won't be started until all workers of the specified
3358 waitee job are done.
3360 ``wait_for`` operates on the job name basis, so there are a few
3361 limitations. First, the waitee must be defined prior to the waiter job
3362 (meaning no forward references). Second, if a job is being referenced as a
3363 waitee, it must have a unique name (no duplicate waitees).
3365 .. option:: nice=int
3367 Run the job with the given nice value. See man :manpage:`nice(2)`.
3369 On Windows, values less than -15 set the process class to "High"; -1 through
3370 -15 set "Above Normal"; 1 through 15 "Below Normal"; and above 15 "Idle"
3373 .. option:: prio=int
3375 Set the I/O priority value of this job. Linux limits us to a positive value
3376 between 0 and 7, with 0 being the highest. See man
3377 :manpage:`ionice(1)`. Refer to an appropriate manpage for other operating
3378 systems since meaning of priority may differ. For per-command priority
3379 setting, see I/O engine specific :option:`cmdprio_percentage` and
3380 :option:`cmdprio` options.
3382 .. option:: prioclass=int
3384 Set the I/O priority class. See man :manpage:`ionice(1)`. For per-command
3385 priority setting, see I/O engine specific :option:`cmdprio_percentage`
3386 and :option:`cmdprio_class` options.
3388 .. option:: cpus_allowed=str
3390 Controls the same options as :option:`cpumask`, but accepts a textual
3391 specification of the permitted CPUs instead and CPUs are indexed from 0. So
3392 to use CPUs 0 and 5 you would specify ``cpus_allowed=0,5``. This option also
3393 allows a range of CPUs to be specified -- say you wanted a binding to CPUs
3394 0, 5, and 8 to 15, you would set ``cpus_allowed=0,5,8-15``.
3396 On Windows, when ``cpus_allowed`` is unset only CPUs from fio's current
3397 processor group will be used and affinity settings are inherited from the
3398 system. An fio build configured to target Windows 7 makes options that set
3399 CPUs processor group aware and values will set both the processor group
3400 and a CPU from within that group. For example, on a system where processor
3401 group 0 has 40 CPUs and processor group 1 has 32 CPUs, ``cpus_allowed``
3402 values between 0 and 39 will bind CPUs from processor group 0 and
3403 ``cpus_allowed`` values between 40 and 71 will bind CPUs from processor
3404 group 1. When using ``cpus_allowed_policy=shared`` all CPUs specified by a
3405 single ``cpus_allowed`` option must be from the same processor group. For
3406 Windows fio builds not built for Windows 7, CPUs will only be selected from
3407 (and be relative to) whatever processor group fio happens to be running in
3408 and CPUs from other processor groups cannot be used.
3410 .. option:: cpus_allowed_policy=str
3412 Set the policy of how fio distributes the CPUs specified by
3413 :option:`cpus_allowed` or :option:`cpumask`. Two policies are supported:
3416 All jobs will share the CPU set specified.
3418 Each job will get a unique CPU from the CPU set.
3420 **shared** is the default behavior, if the option isn't specified. If
3421 **split** is specified, then fio will assign one cpu per job. If not
3422 enough CPUs are given for the jobs listed, then fio will roundrobin the CPUs
3425 .. option:: cpumask=int
3427 Set the CPU affinity of this job. The parameter given is a bit mask of
3428 allowed CPUs the job may run on. So if you want the allowed CPUs to be 1
3429 and 5, you would pass the decimal value of (1 << 1 | 1 << 5), or 34. See man
3430 :manpage:`sched_setaffinity(2)`. This may not work on all supported
3431 operating systems or kernel versions. This option doesn't work well for a
3432 higher CPU count than what you can store in an integer mask, so it can only
3433 control cpus 1-32. For boxes with larger CPU counts, use
3434 :option:`cpus_allowed`.
3436 .. option:: numa_cpu_nodes=str
3438 Set this job running on specified NUMA nodes' CPUs. The arguments allow
3439 comma delimited list of cpu numbers, A-B ranges, or `all`. Note, to enable
3440 NUMA options support, fio must be built on a system with libnuma-dev(el)
3443 .. option:: numa_mem_policy=str
3445 Set this job's memory policy and corresponding NUMA nodes. Format of the
3450 ``mode`` is one of the following memory policies: ``default``, ``prefer``,
3451 ``bind``, ``interleave`` or ``local``. For ``default`` and ``local`` memory
3452 policies, no node needs to be specified. For ``prefer``, only one node is
3453 allowed. For ``bind`` and ``interleave`` the ``nodelist`` may be as
3454 follows: a comma delimited list of numbers, A-B ranges, or `all`.
3456 .. option:: cgroup=str
3458 Add job to this control group. If it doesn't exist, it will be created. The
3459 system must have a mounted cgroup blkio mount point for this to work. If
3460 your system doesn't have it mounted, you can do so with::
3462 # mount -t cgroup -o blkio none /cgroup
3464 .. option:: cgroup_weight=int
3466 Set the weight of the cgroup to this value. See the documentation that comes
3467 with the kernel, allowed values are in the range of 100..1000.
3469 .. option:: cgroup_nodelete=bool
3471 Normally fio will delete the cgroups it has created after the job
3472 completion. To override this behavior and to leave cgroups around after the
3473 job completion, set ``cgroup_nodelete=1``. This can be useful if one wants
3474 to inspect various cgroup files after job completion. Default: false.
3476 .. option:: flow_id=int
3478 The ID of the flow. If not specified, it defaults to being a global
3479 flow. See :option:`flow`.
3481 .. option:: flow=int
3483 Weight in token-based flow control. If this value is used, then fio
3484 regulates the activity between two or more jobs sharing the same
3485 flow_id. Fio attempts to keep each job activity proportional to other
3486 jobs' activities in the same flow_id group, with respect to requested
3487 weight per job. That is, if one job has `flow=3', another job has
3488 `flow=2' and another with `flow=1`, then there will be a roughly 3:2:1
3489 ratio in how much one runs vs the others.
3491 .. option:: flow_sleep=int
3493 The period of time, in microseconds, to wait after the flow counter
3494 has exceeded its proportion before retrying operations.
3496 .. option:: stonewall, wait_for_previous
3498 Wait for preceding jobs in the job file to exit, before starting this
3499 one. Can be used to insert serialization points in the job file. A stone
3500 wall also implies starting a new reporting group, see
3501 :option:`group_reporting`.
3505 By default, fio will continue running all other jobs when one job finishes.
3506 Sometimes this is not the desired action. Setting ``exitall`` will instead
3507 make fio terminate all jobs in the same group, as soon as one job of that
3510 .. option:: exit_what=str
3512 By default, fio will continue running all other jobs when one job finishes.
3513 Sometimes this is not the desired action. Setting ``exitall`` will
3514 instead make fio terminate all jobs in the same group. The option
3515 ``exit_what`` allows to control which jobs get terminated when ``exitall`` is
3516 enabled. The default is ``group`` and does not change the behaviour of
3517 ``exitall``. The setting ``all`` terminates all jobs. The setting ``stonewall``
3518 terminates all currently running jobs across all groups and continues execution
3519 with the next stonewalled group.
3521 .. option:: exec_prerun=str
3523 Before running this job, issue the command specified through
3524 :manpage:`system(3)`. Output is redirected in a file called
3525 :file:`jobname.prerun.txt`.
3527 .. option:: exec_postrun=str
3529 After the job completes, issue the command specified though
3530 :manpage:`system(3)`. Output is redirected in a file called
3531 :file:`jobname.postrun.txt`.
3535 Instead of running as the invoking user, set the user ID to this value
3536 before the thread/process does any work.
3540 Set group ID, see :option:`uid`.
3546 .. option:: verify_only
3548 Do not perform specified workload, only verify data still matches previous
3549 invocation of this workload. This option allows one to check data multiple
3550 times at a later date without overwriting it. This option makes sense only
3551 for workloads that write data, and does not support workloads with the
3552 :option:`time_based` option set.
3554 .. option:: do_verify=bool
3556 Run the verify phase after a write phase. Only valid if :option:`verify` is
3559 .. option:: verify=str
3561 If writing to a file, fio can verify the file contents after each iteration
3562 of the job. Each verification method also implies verification of special
3563 header, which is written to the beginning of each block. This header also
3564 includes meta information, like offset of the block, block number, timestamp
3565 when block was written, etc. :option:`verify` can be combined with
3566 :option:`verify_pattern` option. The allowed values are:
3569 Use an md5 sum of the data area and store it in the header of
3573 Use an experimental crc64 sum of the data area and store it in the
3574 header of each block.
3577 Use a crc32c sum of the data area and store it in the header of
3578 each block. This will automatically use hardware acceleration
3579 (e.g. SSE4.2 on an x86 or CRC crypto extensions on ARM64) but will
3580 fall back to software crc32c if none is found. Generally the
3581 fastest checksum fio supports when hardware accelerated.
3587 Use a crc32 sum of the data area and store it in the header of each
3591 Use a crc16 sum of the data area and store it in the header of each
3595 Use a crc7 sum of the data area and store it in the header of each
3599 Use xxhash as the checksum function. Generally the fastest software
3600 checksum that fio supports.
3603 Use sha512 as the checksum function.
3606 Use sha256 as the checksum function.
3609 Use optimized sha1 as the checksum function.
3612 Use optimized sha3-224 as the checksum function.
3615 Use optimized sha3-256 as the checksum function.
3618 Use optimized sha3-384 as the checksum function.
3621 Use optimized sha3-512 as the checksum function.
3624 This option is deprecated, since now meta information is included in
3625 generic verification header and meta verification happens by
3626 default. For detailed information see the description of the
3627 :option:`verify` setting. This option is kept because of
3628 compatibility's sake with old configurations. Do not use it.
3631 Verify a strict pattern. Normally fio includes a header with some
3632 basic information and checksumming, but if this option is set, only
3633 the specific pattern set with :option:`verify_pattern` is verified.
3636 Only pretend to verify. Useful for testing internals with
3637 :option:`ioengine`\=null, not for much else.
3639 This option can be used for repeated burn-in tests of a system to make sure
3640 that the written data is also correctly read back. If the data direction
3641 given is a read or random read, fio will assume that it should verify a
3642 previously written file. If the data direction includes any form of write,
3643 the verify will be of the newly written data.
3645 To avoid false verification errors, do not use the norandommap option when
3646 verifying data with async I/O engines and I/O depths > 1. Or use the
3647 norandommap and the lfsr random generator together to avoid writing to the
3648 same offset with multiple outstanding I/Os.
3650 .. option:: verify_offset=int
3652 Swap the verification header with data somewhere else in the block before
3653 writing. It is swapped back before verifying.
3655 .. option:: verify_interval=int
3657 Write the verification header at a finer granularity than the
3658 :option:`blocksize`. It will be written for chunks the size of
3659 ``verify_interval``. :option:`blocksize` should divide this evenly.
3661 .. option:: verify_pattern=str
3663 If set, fio will fill the I/O buffers with this pattern. Fio defaults to
3664 filling with totally random bytes, but sometimes it's interesting to fill
3665 with a known pattern for I/O verification purposes. Depending on the width
3666 of the pattern, fio will fill 1/2/3/4 bytes of the buffer at the time (it can
3667 be either a decimal or a hex number). The ``verify_pattern`` if larger than
3668 a 32-bit quantity has to be a hex number that starts with either "0x" or
3669 "0X". Use with :option:`verify`. Also, ``verify_pattern`` supports %o
3670 format, which means that for each block offset will be written and then
3671 verified back, e.g.::
3675 Or use combination of everything::
3677 verify_pattern=0xff%o"abcd"-12
3679 .. option:: verify_fatal=bool
3681 Normally fio will keep checking the entire contents before quitting on a
3682 block verification failure. If this option is set, fio will exit the job on
3683 the first observed failure. Default: false.
3685 .. option:: verify_dump=bool
3687 If set, dump the contents of both the original data block and the data block
3688 we read off disk to files. This allows later analysis to inspect just what
3689 kind of data corruption occurred. Off by default.
3691 .. option:: verify_async=int
3693 Fio will normally verify I/O inline from the submitting thread. This option
3694 takes an integer describing how many async offload threads to create for I/O
3695 verification instead, causing fio to offload the duty of verifying I/O
3696 contents to one or more separate threads. If using this offload option, even
3697 sync I/O engines can benefit from using an :option:`iodepth` setting higher
3698 than 1, as it allows them to have I/O in flight while verifies are running.
3699 Defaults to 0 async threads, i.e. verification is not asynchronous.
3701 .. option:: verify_async_cpus=str
3703 Tell fio to set the given CPU affinity on the async I/O verification
3704 threads. See :option:`cpus_allowed` for the format used.
3706 .. option:: verify_backlog=int
3708 Fio will normally verify the written contents of a job that utilizes verify
3709 once that job has completed. In other words, everything is written then
3710 everything is read back and verified. You may want to verify continually
3711 instead for a variety of reasons. Fio stores the meta data associated with
3712 an I/O block in memory, so for large verify workloads, quite a bit of memory
3713 would be used up holding this meta data. If this option is enabled, fio will
3714 write only N blocks before verifying these blocks.
3716 .. option:: verify_backlog_batch=int
3718 Control how many blocks fio will verify if :option:`verify_backlog` is
3719 set. If not set, will default to the value of :option:`verify_backlog`
3720 (meaning the entire queue is read back and verified). If
3721 ``verify_backlog_batch`` is less than :option:`verify_backlog` then not all
3722 blocks will be verified, if ``verify_backlog_batch`` is larger than
3723 :option:`verify_backlog`, some blocks will be verified more than once.
3725 .. option:: verify_state_save=bool
3727 When a job exits during the write phase of a verify workload, save its
3728 current state. This allows fio to replay up until that point, if the verify
3729 state is loaded for the verify read phase. The format of the filename is,
3732 <type>-<jobname>-<jobindex>-verify.state.
3734 <type> is "local" for a local run, "sock" for a client/server socket
3735 connection, and "ip" (192.168.0.1, for instance) for a networked
3736 client/server connection. Defaults to true.
3738 .. option:: verify_state_load=bool
3740 If a verify termination trigger was used, fio stores the current write state
3741 of each thread. This can be used at verification time so that fio knows how
3742 far it should verify. Without this information, fio will run a full
3743 verification pass, according to the settings in the job file used. Default
3746 .. option:: trim_percentage=int
3748 Number of verify blocks to discard/trim.
3750 .. option:: trim_verify_zero=bool
3752 Verify that trim/discarded blocks are returned as zeros.
3754 .. option:: trim_backlog=int
3756 Trim after this number of blocks are written.
3758 .. option:: trim_backlog_batch=int
3760 Trim this number of I/O blocks.
3762 .. option:: experimental_verify=bool
3764 Enable experimental verification. Standard verify records I/O metadata
3765 for later use during the verification phase. Experimental verify
3766 instead resets the file after the write phase and then replays I/Os for
3767 the verification phase.
3772 .. option:: steadystate=str:float, ss=str:float
3774 Define the criterion and limit for assessing steady state performance. The
3775 first parameter designates the criterion whereas the second parameter sets
3776 the threshold. When the criterion falls below the threshold for the
3777 specified duration, the job will stop. For example, `iops_slope:0.1%` will
3778 direct fio to terminate the job when the least squares regression slope
3779 falls below 0.1% of the mean IOPS. If :option:`group_reporting` is enabled
3780 this will apply to all jobs in the group. Below is the list of available
3781 steady state assessment criteria. All assessments are carried out using only
3782 data from the rolling collection window. Threshold limits can be expressed
3783 as a fixed value or as a percentage of the mean in the collection window.
3785 When using this feature, most jobs should include the :option:`time_based`
3786 and :option:`runtime` options or the :option:`loops` option so that fio does not
3787 stop running after it has covered the full size of the specified file(s) or device(s).
3790 Collect IOPS data. Stop the job if all individual IOPS measurements
3791 are within the specified limit of the mean IOPS (e.g., ``iops:2``
3792 means that all individual IOPS values must be within 2 of the mean,
3793 whereas ``iops:0.2%`` means that all individual IOPS values must be
3794 within 0.2% of the mean IOPS to terminate the job).
3797 Collect IOPS data and calculate the least squares regression
3798 slope. Stop the job if the slope falls below the specified limit.
3801 Collect bandwidth data. Stop the job if all individual bandwidth
3802 measurements are within the specified limit of the mean bandwidth.
3805 Collect bandwidth data and calculate the least squares regression
3806 slope. Stop the job if the slope falls below the specified limit.
3808 .. option:: steadystate_duration=time, ss_dur=time
3810 A rolling window of this duration will be used to judge whether steady state
3811 has been reached. Data will be collected once per second. The default is 0
3812 which disables steady state detection. When the unit is omitted, the
3813 value is interpreted in seconds.
3815 .. option:: steadystate_ramp_time=time, ss_ramp=time
3817 Allow the job to run for the specified duration before beginning data
3818 collection for checking the steady state job termination criterion. The
3819 default is 0. When the unit is omitted, the value is interpreted in seconds.
3822 Measurements and reporting
3823 ~~~~~~~~~~~~~~~~~~~~~~~~~~
3825 .. option:: per_job_logs=bool
3827 If set, this generates bw/clat/iops log with per file private filenames. If
3828 not set, jobs with identical names will share the log filename. Default:
3831 .. option:: group_reporting
3833 It may sometimes be interesting to display statistics for groups of jobs as
3834 a whole instead of for each individual job. This is especially true if
3835 :option:`numjobs` is used; looking at individual thread/process output
3836 quickly becomes unwieldy. To see the final report per-group instead of
3837 per-job, use :option:`group_reporting`. Jobs in a file will be part of the
3838 same reporting group, unless if separated by a :option:`stonewall`, or by
3839 using :option:`new_group`.
3841 .. option:: new_group
3843 Start a new reporting group. See: :option:`group_reporting`. If not given,
3844 all jobs in a file will be part of the same reporting group, unless
3845 separated by a :option:`stonewall`.
3847 .. option:: stats=bool
3849 By default, fio collects and shows final output results for all jobs
3850 that run. If this option is set to 0, then fio will ignore it in
3851 the final stat output.
3853 .. option:: write_bw_log=str
3855 If given, write a bandwidth log for this job. Can be used to store data of
3856 the bandwidth of the jobs in their lifetime.
3858 If no str argument is given, the default filename of
3859 :file:`jobname_type.x.log` is used. Even when the argument is given, fio
3860 will still append the type of log. So if one specifies::
3864 The actual log name will be :file:`foo_bw.x.log` where `x` is the index
3865 of the job (`1..N`, where `N` is the number of jobs). If
3866 :option:`per_job_logs` is false, then the filename will not include the
3869 The included :command:`fio_generate_plots` script uses :command:`gnuplot` to turn these
3870 text files into nice graphs. See `Log File Formats`_ for how data is
3871 structured within the file.
3873 .. option:: write_lat_log=str
3875 Same as :option:`write_bw_log`, except this option creates I/O
3876 submission (e.g., :file:`name_slat.x.log`), completion (e.g.,
3877 :file:`name_clat.x.log`), and total (e.g., :file:`name_lat.x.log`)
3878 latency files instead. See :option:`write_bw_log` for details about
3879 the filename format and `Log File Formats`_ for how data is structured
3882 .. option:: write_hist_log=str
3884 Same as :option:`write_bw_log` but writes an I/O completion latency
3885 histogram file (e.g., :file:`name_hist.x.log`) instead. Note that this
3886 file will be empty unless :option:`log_hist_msec` has also been set.
3887 See :option:`write_bw_log` for details about the filename format and
3888 `Log File Formats`_ for how data is structured within the file.
3890 .. option:: write_iops_log=str
3892 Same as :option:`write_bw_log`, but writes an IOPS file (e.g.
3893 :file:`name_iops.x.log`) instead. Because fio defaults to individual
3894 I/O logging, the value entry in the IOPS log will be 1 unless windowed
3895 logging (see :option:`log_avg_msec`) has been enabled. See
3896 :option:`write_bw_log` for details about the filename format and `Log
3897 File Formats`_ for how data is structured within the file.
3899 .. option:: log_entries=int
3901 By default, fio will log an entry in the iops, latency, or bw log for
3902 every I/O that completes. The initial number of I/O log entries is 1024.
3903 When the log entries are all used, new log entries are dynamically
3904 allocated. This dynamic log entry allocation may negatively impact
3905 time-related statistics such as I/O tail latencies (e.g. 99.9th percentile
3906 completion latency). This option allows specifying a larger initial
3907 number of log entries to avoid run-time allocations of new log entries,
3908 resulting in more precise time-related I/O statistics.
3909 Also see :option:`log_avg_msec`. Defaults to 1024.
3911 .. option:: log_avg_msec=int
3913 By default, fio will log an entry in the iops, latency, or bw log for every
3914 I/O that completes. When writing to the disk log, that can quickly grow to a
3915 very large size. Setting this option makes fio average the each log entry
3916 over the specified period of time, reducing the resolution of the log. See
3917 :option:`log_max_value` as well. Defaults to 0, logging all entries.
3918 Also see `Log File Formats`_.
3920 .. option:: log_hist_msec=int
3922 Same as :option:`log_avg_msec`, but logs entries for completion latency
3923 histograms. Computing latency percentiles from averages of intervals using
3924 :option:`log_avg_msec` is inaccurate. Setting this option makes fio log
3925 histogram entries over the specified period of time, reducing log sizes for
3926 high IOPS devices while retaining percentile accuracy. See
3927 :option:`log_hist_coarseness` and :option:`write_hist_log` as well.
3928 Defaults to 0, meaning histogram logging is disabled.
3930 .. option:: log_hist_coarseness=int
3932 Integer ranging from 0 to 6, defining the coarseness of the resolution of
3933 the histogram logs enabled with :option:`log_hist_msec`. For each increment
3934 in coarseness, fio outputs half as many bins. Defaults to 0, for which
3935 histogram logs contain 1216 latency bins. See :option:`write_hist_log`
3936 and `Log File Formats`_.
3938 .. option:: log_max_value=bool
3940 If :option:`log_avg_msec` is set, fio logs the average over that window. If
3941 you instead want to log the maximum value, set this option to 1. Defaults to
3942 0, meaning that averaged values are logged.
3944 .. option:: log_offset=bool
3946 If this is set, the iolog options will include the byte offset for the I/O
3947 entry as well as the other data values. Defaults to 0 meaning that
3948 offsets are not present in logs. Also see `Log File Formats`_.
3950 .. option:: log_compression=int
3952 If this is set, fio will compress the I/O logs as it goes, to keep the
3953 memory footprint lower. When a log reaches the specified size, that chunk is
3954 removed and compressed in the background. Given that I/O logs are fairly
3955 highly compressible, this yields a nice memory savings for longer runs. The
3956 downside is that the compression will consume some background CPU cycles, so
3957 it may impact the run. This, however, is also true if the logging ends up
3958 consuming most of the system memory. So pick your poison. The I/O logs are
3959 saved normally at the end of a run, by decompressing the chunks and storing
3960 them in the specified log file. This feature depends on the availability of
3963 .. option:: log_compression_cpus=str
3965 Define the set of CPUs that are allowed to handle online log compression for
3966 the I/O jobs. This can provide better isolation between performance
3967 sensitive jobs, and background compression work. See
3968 :option:`cpus_allowed` for the format used.
3970 .. option:: log_store_compressed=bool
3972 If set, fio will store the log files in a compressed format. They can be
3973 decompressed with fio, using the :option:`--inflate-log` command line
3974 parameter. The files will be stored with a :file:`.fz` suffix.
3976 .. option:: log_unix_epoch=bool
3978 If set, fio will log Unix timestamps to the log files produced by enabling
3979 write_type_log for each log type, instead of the default zero-based
3982 .. option:: log_alternate_epoch=bool
3984 If set, fio will log timestamps based on the epoch used by the clock specified
3985 in the log_alternate_epoch_clock_id option, to the log files produced by
3986 enabling write_type_log for each log type, instead of the default zero-based
3989 .. option:: log_alternate_epoch_clock_id=int
3991 Specifies the clock_id to be used by clock_gettime to obtain the alternate epoch
3992 if either log_unix_epoch or log_alternate_epoch are true. Otherwise has no
3993 effect. Default value is 0, or CLOCK_REALTIME.
3995 .. option:: block_error_percentiles=bool
3997 If set, record errors in trim block-sized units from writes and trims and
3998 output a histogram of how many trims it took to get to errors, and what kind
3999 of error was encountered.
4001 .. option:: bwavgtime=int
4003 Average the calculated bandwidth over the given time. Value is specified in
4004 milliseconds. If the job also does bandwidth logging through
4005 :option:`write_bw_log`, then the minimum of this option and
4006 :option:`log_avg_msec` will be used. Default: 500ms.
4008 .. option:: iopsavgtime=int
4010 Average the calculated IOPS over the given time. Value is specified in
4011 milliseconds. If the job also does IOPS logging through
4012 :option:`write_iops_log`, then the minimum of this option and
4013 :option:`log_avg_msec` will be used. Default: 500ms.
4015 .. option:: disk_util=bool
4017 Generate disk utilization statistics, if the platform supports it.
4020 .. option:: disable_lat=bool
4022 Disable measurements of total latency numbers. Useful only for cutting back
4023 the number of calls to :manpage:`gettimeofday(2)`, as that does impact
4024 performance at really high IOPS rates. Note that to really get rid of a
4025 large amount of these calls, this option must be used with
4026 :option:`disable_slat` and :option:`disable_bw_measurement` as well.
4028 .. option:: disable_clat=bool
4030 Disable measurements of completion latency numbers. See
4031 :option:`disable_lat`.
4033 .. option:: disable_slat=bool
4035 Disable measurements of submission latency numbers. See
4036 :option:`disable_lat`.
4038 .. option:: disable_bw_measurement=bool, disable_bw=bool
4040 Disable measurements of throughput/bandwidth numbers. See
4041 :option:`disable_lat`.
4043 .. option:: slat_percentiles=bool
4045 Report submission latency percentiles. Submission latency is not recorded
4046 for synchronous ioengines.
4048 .. option:: clat_percentiles=bool
4050 Report completion latency percentiles.
4052 .. option:: lat_percentiles=bool
4054 Report total latency percentiles. Total latency is the sum of submission
4055 latency and completion latency.
4057 .. option:: percentile_list=float_list
4059 Overwrite the default list of percentiles for latencies and the block error
4060 histogram. Each number is a floating point number in the range (0,100], and
4061 the maximum length of the list is 20. Use ``:`` to separate the numbers. For
4062 example, ``--percentile_list=99.5:99.9`` will cause fio to report the
4063 latency durations below which 99.5% and 99.9% of the observed latencies fell,
4066 .. option:: significant_figures=int
4068 If using :option:`--output-format` of `normal`, set the significant
4069 figures to this value. Higher values will yield more precise IOPS and
4070 throughput units, while lower values will round. Requires a minimum
4071 value of 1 and a maximum value of 10. Defaults to 4.
4077 .. option:: exitall_on_error
4079 When one job finishes in error, terminate the rest. The default is to wait
4080 for each job to finish.
4082 .. option:: continue_on_error=str
4084 Normally fio will exit the job on the first observed failure. If this option
4085 is set, fio will continue the job when there is a 'non-fatal error' (EIO or
4086 EILSEQ) until the runtime is exceeded or the I/O size specified is
4087 completed. If this option is used, there are two more stats that are
4088 appended, the total error count and the first error. The error field given
4089 in the stats is the first error that was hit during the run.
4091 Note: a write error from the device may go unnoticed by fio when using
4092 buffered IO, as the write() (or similar) system call merely dirties the
4093 kernel pages, unless :option:`sync` or :option:`direct` is used. Device IO
4094 errors occur when the dirty data is actually written out to disk. If fully
4095 sync writes aren't desirable, :option:`fsync` or :option:`fdatasync` can be
4096 used as well. This is specific to writes, as reads are always synchronous.
4098 The allowed values are:
4101 Exit on any I/O or verify errors.
4104 Continue on read errors, exit on all others.
4107 Continue on write errors, exit on all others.
4110 Continue on any I/O error, exit on all others.
4113 Continue on verify errors, exit on all others.
4116 Continue on all errors.
4119 Backward-compatible alias for 'none'.
4122 Backward-compatible alias for 'all'.
4124 .. option:: ignore_error=str
4126 Sometimes you want to ignore some errors during test in that case you can
4127 specify error list for each error type, instead of only being able to
4128 ignore the default 'non-fatal error' using :option:`continue_on_error`.
4129 ``ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST`` errors for
4130 given error type is separated with ':'. Error may be symbol ('ENOSPC',
4131 'ENOMEM') or integer. Example::
4133 ignore_error=EAGAIN,ENOSPC:122
4135 This option will ignore EAGAIN from READ, and ENOSPC and 122(EDQUOT) from
4136 WRITE. This option works by overriding :option:`continue_on_error` with
4137 the list of errors for each error type if any.
4139 .. option:: error_dump=bool
4141 If set dump every error even if it is non fatal, true by default. If
4142 disabled only fatal error will be dumped.
4144 Running predefined workloads
4145 ----------------------------
4147 Fio includes predefined profiles that mimic the I/O workloads generated by
4150 .. option:: profile=str
4152 The predefined workload to run. Current profiles are:
4155 Threaded I/O bench (tiotest/tiobench) like workload.
4158 Aerospike Certification Tool (ACT) like workload.
4160 To view a profile's additional options use :option:`--cmdhelp` after specifying
4161 the profile. For example::
4163 $ fio --profile=act --cmdhelp
4168 .. option:: device-names=str
4173 .. option:: load=int
4176 ACT load multiplier. Default: 1.
4178 .. option:: test-duration=time
4181 How long the entire test takes to run. When the unit is omitted, the value
4182 is given in seconds. Default: 24h.
4184 .. option:: threads-per-queue=int
4187 Number of read I/O threads per device. Default: 8.
4189 .. option:: read-req-num-512-blocks=int
4192 Number of 512B blocks to read at the time. Default: 3.
4194 .. option:: large-block-op-kbytes=int
4197 Size of large block ops in KiB (writes). Default: 131072.
4202 Set to run ACT prep phase.
4204 Tiobench profile options
4205 ~~~~~~~~~~~~~~~~~~~~~~~~
4207 .. option:: size=str
4212 .. option:: block=int
4215 Block size in bytes. Default: 4096.
4217 .. option:: numruns=int
4227 .. option:: threads=int
4232 Interpreting the output
4233 -----------------------
4236 Example output was based on the following:
4237 TZ=UTC fio --iodepth=8 --ioengine=null --size=100M --time_based \
4238 --rate=1256k --bs=14K --name=quick --runtime=1s --name=mixed \
4239 --runtime=2m --rw=rw
4241 Fio spits out a lot of output. While running, fio will display the status of the
4242 jobs created. An example of that would be::
4244 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]
4246 The characters inside the first set of square brackets denote the current status of
4247 each thread. The first character is the first job defined in the job file, and so
4248 forth. The possible values (in typical life cycle order) are:
4250 +------+-----+-----------------------------------------------------------+
4252 +======+=====+===========================================================+
4253 | P | | Thread setup, but not started. |
4254 +------+-----+-----------------------------------------------------------+
4255 | C | | Thread created. |
4256 +------+-----+-----------------------------------------------------------+
4257 | I | | Thread initialized, waiting or generating necessary data. |
4258 +------+-----+-----------------------------------------------------------+
4259 | | p | Thread running pre-reading file(s). |
4260 +------+-----+-----------------------------------------------------------+
4261 | | / | Thread is in ramp period. |
4262 +------+-----+-----------------------------------------------------------+
4263 | | R | Running, doing sequential reads. |
4264 +------+-----+-----------------------------------------------------------+
4265 | | r | Running, doing random reads. |
4266 +------+-----+-----------------------------------------------------------+
4267 | | W | Running, doing sequential writes. |
4268 +------+-----+-----------------------------------------------------------+
4269 | | w | Running, doing random writes. |
4270 +------+-----+-----------------------------------------------------------+
4271 | | M | Running, doing mixed sequential reads/writes. |
4272 +------+-----+-----------------------------------------------------------+
4273 | | m | Running, doing mixed random reads/writes. |
4274 +------+-----+-----------------------------------------------------------+
4275 | | D | Running, doing sequential trims. |
4276 +------+-----+-----------------------------------------------------------+
4277 | | d | Running, doing random trims. |
4278 +------+-----+-----------------------------------------------------------+
4279 | | F | Running, currently waiting for :manpage:`fsync(2)`. |
4280 +------+-----+-----------------------------------------------------------+
4281 | | V | Running, doing verification of written data. |
4282 +------+-----+-----------------------------------------------------------+
4283 | f | | Thread finishing. |
4284 +------+-----+-----------------------------------------------------------+
4285 | E | | Thread exited, not reaped by main thread yet. |
4286 +------+-----+-----------------------------------------------------------+
4287 | _ | | Thread reaped. |
4288 +------+-----+-----------------------------------------------------------+
4289 | X | | Thread reaped, exited with an error. |
4290 +------+-----+-----------------------------------------------------------+
4291 | K | | Thread reaped, exited due to signal. |
4292 +------+-----+-----------------------------------------------------------+
4295 Example output was based on the following:
4296 TZ=UTC fio --iodepth=8 --ioengine=null --size=100M --runtime=58m \
4297 --time_based --rate=2512k --bs=256K --numjobs=10 \
4298 --name=readers --rw=read --name=writers --rw=write
4300 Fio will condense the thread string as not to take up more space on the command
4301 line than needed. For instance, if you have 10 readers and 10 writers running,
4302 the output would look like this::
4304 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]
4306 Note that the status string is displayed in order, so it's possible to tell which of
4307 the jobs are currently doing what. In the example above this means that jobs 1--10
4308 are readers and 11--20 are writers.
4310 The other values are fairly self explanatory -- number of threads currently
4311 running and doing I/O, the number of currently open files (f=), the estimated
4312 completion percentage, the rate of I/O since last check (read speed listed first,
4313 then write speed and optionally trim speed) in terms of bandwidth and IOPS,
4314 and time to completion for the current running group. It's impossible to estimate
4315 runtime of the following groups (if any).
4318 Example output was based on the following:
4319 TZ=UTC fio --iodepth=16 --ioengine=posixaio --filename=/tmp/fiofile \
4320 --direct=1 --size=100M --time_based --runtime=50s --rate_iops=89 \
4321 --bs=7K --name=Client1 --rw=write
4323 When fio is done (or interrupted by :kbd:`Ctrl-C`), it will show the data for
4324 each thread, group of threads, and disks in that order. For each overall thread (or
4325 group) the output looks like::
4327 Client1: (groupid=0, jobs=1): err= 0: pid=16109: Sat Jun 24 12:07:54 2017
4328 write: IOPS=88, BW=623KiB/s (638kB/s)(30.4MiB/50032msec)
4329 slat (nsec): min=500, max=145500, avg=8318.00, stdev=4781.50
4330 clat (usec): min=170, max=78367, avg=4019.02, stdev=8293.31
4331 lat (usec): min=174, max=78375, avg=4027.34, stdev=8291.79
4332 clat percentiles (usec):
4333 | 1.00th=[ 302], 5.00th=[ 326], 10.00th=[ 343], 20.00th=[ 363],
4334 | 30.00th=[ 392], 40.00th=[ 404], 50.00th=[ 416], 60.00th=[ 445],
4335 | 70.00th=[ 816], 80.00th=[ 6718], 90.00th=[12911], 95.00th=[21627],
4336 | 99.00th=[43779], 99.50th=[51643], 99.90th=[68682], 99.95th=[72877],
4338 bw ( KiB/s): min= 532, max= 686, per=0.10%, avg=622.87, stdev=24.82, samples= 100
4339 iops : min= 76, max= 98, avg=88.98, stdev= 3.54, samples= 100
4340 lat (usec) : 250=0.04%, 500=64.11%, 750=4.81%, 1000=2.79%
4341 lat (msec) : 2=4.16%, 4=1.84%, 10=4.90%, 20=11.33%, 50=5.37%
4342 lat (msec) : 100=0.65%
4343 cpu : usr=0.27%, sys=0.18%, ctx=12072, majf=0, minf=21
4344 IO depths : 1=85.0%, 2=13.1%, 4=1.8%, 8=0.1%, 16=0.0%, 32=0.0%, >=64=0.0%
4345 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
4346 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
4347 issued rwt: total=0,4450,0, short=0,0,0, dropped=0,0,0
4348 latency : target=0, window=0, percentile=100.00%, depth=8
4350 The job name (or first job's name when using :option:`group_reporting`) is printed,
4351 along with the group id, count of jobs being aggregated, last error id seen (which
4352 is 0 when there are no errors), pid/tid of that thread and the time the job/group
4353 completed. Below are the I/O statistics for each data direction performed (showing
4354 writes in the example above). In the order listed, they denote:
4357 The string before the colon shows the I/O direction the statistics
4358 are for. **IOPS** is the average I/Os performed per second. **BW**
4359 is the average bandwidth rate shown as: value in power of 2 format
4360 (value in power of 10 format). The last two values show: (**total
4361 I/O performed** in power of 2 format / **runtime** of that thread).
4364 Submission latency (**min** being the minimum, **max** being the
4365 maximum, **avg** being the average, **stdev** being the standard
4366 deviation). This is the time from when fio initialized the I/O
4367 to submission. For synchronous ioengines this includes the time
4368 up until just before the ioengine's queue function is called.
4369 For asynchronous ioengines this includes the time up through the
4370 completion of the ioengine's queue function (and commit function
4371 if it is defined). For sync I/O this row is not displayed as the
4372 slat is negligible. This value can be in nanoseconds,
4373 microseconds or milliseconds --- fio will choose the most
4374 appropriate base and print that (in the example above
4375 nanoseconds was the best scale). Note: in :option:`--minimal`
4376 mode latencies are always expressed in microseconds.
4379 Completion latency. Same names as slat, this denotes the time from
4380 submission to completion of the I/O pieces. For sync I/O, this
4381 represents the time from when the I/O was submitted to the
4382 operating system to when it was completed. For asynchronous
4383 ioengines this is the time from when the ioengine's queue (and
4384 commit if available) functions were completed to when the I/O's
4385 completion was reaped by fio.
4388 Total latency. Same names as slat and clat, this denotes the time from
4389 when fio created the I/O unit to completion of the I/O operation.
4390 It is the sum of submission and completion latency.
4393 Bandwidth statistics based on samples. Same names as the xlat stats,
4394 but also includes the number of samples taken (**samples**) and an
4395 approximate percentage of total aggregate bandwidth this thread
4396 received in its group (**per**). This last value is only really
4397 useful if the threads in this group are on the same disk, since they
4398 are then competing for disk access.
4401 IOPS statistics based on samples. Same names as bw.
4403 **lat (nsec/usec/msec)**
4404 The distribution of I/O completion latencies. This is the time from when
4405 I/O leaves fio and when it gets completed. Unlike the separate
4406 read/write/trim sections above, the data here and in the remaining
4407 sections apply to all I/Os for the reporting group. 250=0.04% means that
4408 0.04% of the I/Os completed in under 250us. 500=64.11% means that 64.11%
4409 of the I/Os required 250 to 499us for completion.
4412 CPU usage. User and system time, along with the number of context
4413 switches this thread went through, usage of system and user time, and
4414 finally the number of major and minor page faults. The CPU utilization
4415 numbers are averages for the jobs in that reporting group, while the
4416 context and fault counters are summed.
4419 The distribution of I/O depths over the job lifetime. The numbers are
4420 divided into powers of 2 and each entry covers depths from that value
4421 up to those that are lower than the next entry -- e.g., 16= covers
4422 depths from 16 to 31. Note that the range covered by a depth
4423 distribution entry can be different to the range covered by the
4424 equivalent submit/complete distribution entry.
4427 How many pieces of I/O were submitting in a single submit call. Each
4428 entry denotes that amount and below, until the previous entry -- e.g.,
4429 16=100% means that we submitted anywhere between 9 to 16 I/Os per submit
4430 call. Note that the range covered by a submit distribution entry can
4431 be different to the range covered by the equivalent depth distribution
4435 Like the above submit number, but for completions instead.
4438 The number of read/write/trim requests issued, and how many of them were
4442 These values are for :option:`latency_target` and related options. When
4443 these options are engaged, this section describes the I/O depth required
4444 to meet the specified latency target.
4447 Example output was based on the following:
4448 TZ=UTC fio --ioengine=null --iodepth=2 --size=100M --numjobs=2 \
4449 --rate_process=poisson --io_limit=32M --name=read --bs=128k \
4450 --rate=11M --name=write --rw=write --bs=2k --rate=700k
4452 After each client has been listed, the group statistics are printed. They
4453 will look like this::
4455 Run status group 0 (all jobs):
4456 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
4457 WRITE: bw=1231KiB/s (1261kB/s), 616KiB/s-621KiB/s (630kB/s-636kB/s), io=64.0MiB (67.1MB), run=52747-53223msec
4459 For each data direction it prints:
4462 Aggregate bandwidth of threads in this group followed by the
4463 minimum and maximum bandwidth of all the threads in this group.
4464 Values outside of brackets are power-of-2 format and those
4465 within are the equivalent value in a power-of-10 format.
4467 Aggregate I/O performed of all threads in this group. The
4468 format is the same as bw.
4470 The smallest and longest runtimes of the threads in this group.
4472 And finally, the disk statistics are printed. This is Linux specific. They will look like this::
4474 Disk stats (read/write):
4475 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
4477 Each value is printed for both reads and writes, with reads first. The
4481 Number of I/Os performed by all groups.
4483 Number of merges performed by the I/O scheduler.
4485 Number of ticks we kept the disk busy.
4487 Total time spent in the disk queue.
4489 The disk utilization. A value of 100% means we kept the disk
4490 busy constantly, 50% would be a disk idling half of the time.
4492 It is also possible to get fio to dump the current output while it is running,
4493 without terminating the job. To do that, send fio the **USR1** signal. You can
4494 also get regularly timed dumps by using the :option:`--status-interval`
4495 parameter, or by creating a file in :file:`/tmp` named
4496 :file:`fio-dump-status`. If fio sees this file, it will unlink it and dump the
4497 current output status.
4503 For scripted usage where you typically want to generate tables or graphs of the
4504 results, fio can output the results in a semicolon separated format. The format
4505 is one long line of values, such as::
4507 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%
4508 A description of this job goes here.
4510 The job description (if provided) follows on a second line for terse v2.
4511 It appears on the same line for other terse versions.
4513 To enable terse output, use the :option:`--minimal` or
4514 :option:`--output-format`\=terse command line options. The
4515 first value is the version of the terse output format. If the output has to be
4516 changed for some reason, this number will be incremented by 1 to signify that
4519 Split up, the format is as follows (comments in brackets denote when a
4520 field was introduced or whether it's specific to some terse version):
4524 terse version, fio version [v3], jobname, groupid, error
4528 Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec)
4529 Submission latency: min, max, mean, stdev (usec)
4530 Completion latency: min, max, mean, stdev (usec)
4531 Completion latency percentiles: 20 fields (see below)
4532 Total latency: min, max, mean, stdev (usec)
4533 Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev, number of samples [v5]
4534 IOPS [v5]: min, max, mean, stdev, number of samples
4540 Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec)
4541 Submission latency: min, max, mean, stdev (usec)
4542 Completion latency: min, max, mean, stdev (usec)
4543 Completion latency percentiles: 20 fields (see below)
4544 Total latency: min, max, mean, stdev (usec)
4545 Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev, number of samples [v5]
4546 IOPS [v5]: min, max, mean, stdev, number of samples
4548 TRIM status [all but version 3]:
4550 Fields are similar to READ/WRITE status.
4554 user, system, context switches, major faults, minor faults
4558 <=1, 2, 4, 8, 16, 32, >=64
4560 I/O latencies microseconds::
4562 <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
4564 I/O latencies milliseconds::
4566 <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
4568 Disk utilization [v3]::
4570 disk name, read ios, write ios, read merges, write merges, read ticks, write ticks,
4571 time spent in queue, disk utilization percentage
4573 Additional Info (dependent on continue_on_error, default off)::
4575 total # errors, first error code
4577 Additional Info (dependent on description being set)::
4581 Completion latency percentiles can be a grouping of up to 20 sets, so for the
4582 terse output fio writes all of them. Each field will look like this::
4586 which is the Xth percentile, and the `usec` latency associated with it.
4588 For `Disk utilization`, all disks used by fio are shown. So for each disk there
4589 will be a disk utilization section.
4591 Below is a single line containing short names for each of the fields in the
4592 minimal output v3, separated by semicolons::
4594 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
4596 In client/server mode terse output differs from what appears when jobs are run
4597 locally. Disk utilization data is omitted from the standard terse output and
4598 for v3 and later appears on its own separate line at the end of each terse
4605 The `json` output format is intended to be both human readable and convenient
4606 for automated parsing. For the most part its sections mirror those of the
4607 `normal` output. The `runtime` value is reported in msec and the `bw` value is
4608 reported in 1024 bytes per second units.
4614 The `json+` output format is identical to the `json` output format except that it
4615 adds a full dump of the completion latency bins. Each `bins` object contains a
4616 set of (key, value) pairs where keys are latency durations and values count how
4617 many I/Os had completion latencies of the corresponding duration. For example,
4620 "bins" : { "87552" : 1, "89600" : 1, "94720" : 1, "96768" : 1, "97792" : 1, "99840" : 1, "100864" : 2, "103936" : 6, "104960" : 534, "105984" : 5995, "107008" : 7529, ... }
4622 This data indicates that one I/O required 87,552ns to complete, two I/Os required
4623 100,864ns to complete, and 7529 I/Os required 107,008ns to complete.
4625 Also included with fio is a Python script `fio_jsonplus_clat2csv` that takes
4626 json+ output and generates CSV-formatted latency data suitable for plotting.
4628 The latency durations actually represent the midpoints of latency intervals.
4629 For details refer to :file:`stat.h`.
4635 There are two trace file format that you can encounter. The older (v1) format is
4636 unsupported since version 1.20-rc3 (March 2008). It will still be described
4637 below in case that you get an old trace and want to understand it.
4639 In any case the trace is a simple text file with a single action per line.
4642 Trace file format v1
4643 ~~~~~~~~~~~~~~~~~~~~
4645 Each line represents a single I/O action in the following format::
4649 where `rw=0/1` for read/write, and the `offset` and `length` entries being in bytes.
4651 This format is not supported in fio versions >= 1.20-rc3.
4654 Trace file format v2
4655 ~~~~~~~~~~~~~~~~~~~~
4657 The second version of the trace file format was added in fio version 1.17. It
4658 allows one to access more than one file per trace and has a bigger set of possible
4661 The first line of the trace file has to be::
4665 Following this can be lines in two different formats, which are described below.
4667 The file management format::
4671 The `filename` is given as an absolute path. The `action` can be one of these:
4674 Add the given `filename` to the trace.
4676 Open the file with the given `filename`. The `filename` has to have
4677 been added with the **add** action before.
4679 Close the file with the given `filename`. The file has to have been
4683 The file I/O action format::
4685 filename action offset length
4687 The `filename` is given as an absolute path, and has to have been added and
4688 opened before it can be used with this format. The `offset` and `length` are
4689 given in bytes. The `action` can be one of these:
4692 Wait for `offset` microseconds. Everything below 100 is discarded.
4693 The time is relative to the previous `wait` statement. Note that
4694 action `wait` is not allowed as of version 3, as the same behavior
4695 can be achieved using timestamps.
4697 Read `length` bytes beginning from `offset`.
4699 Write `length` bytes beginning from `offset`.
4701 :manpage:`fsync(2)` the file.
4703 :manpage:`fdatasync(2)` the file.
4705 Trim the given file from the given `offset` for `length` bytes.
4708 Trace file format v3
4709 ~~~~~~~~~~~~~~~~~~~~
4711 The third version of the trace file format was added in fio version 3.31. It
4712 forces each action to have a timestamp associated with it.
4714 The first line of the trace file has to be::
4718 Following this can be lines in two different formats, which are described below.
4720 The file management format::
4722 timestamp filename action
4724 The file I/O action format::
4726 timestamp filename action offset length
4728 The `timestamp` is relative to the beginning of the run (ie starts at 0). The
4729 `filename`, `action`, `offset` and `length` are identical to version 2, except
4730 that version 3 does not allow the `wait` action.
4733 I/O Replay - Merging Traces
4734 ---------------------------
4736 Colocation is a common practice used to get the most out of a machine.
4737 Knowing which workloads play nicely with each other and which ones don't is
4738 a much harder task. While fio can replay workloads concurrently via multiple
4739 jobs, it leaves some variability up to the scheduler making results harder to
4740 reproduce. Merging is a way to make the order of events consistent.
4742 Merging is integrated into I/O replay and done when a
4743 :option:`merge_blktrace_file` is specified. The list of files passed to
4744 :option:`read_iolog` go through the merge process and output a single file
4745 stored to the specified file. The output file is passed on as if it were the
4746 only file passed to :option:`read_iolog`. An example would look like::
4748 $ fio --read_iolog="<file1>:<file2>" --merge_blktrace_file="<output_file>"
4750 Creating only the merged file can be done by passing the command line argument
4751 :option:`--merge-blktrace-only`.
4753 Scaling traces can be done to see the relative impact of any particular trace
4754 being slowed down or sped up. :option:`merge_blktrace_scalars` takes in a colon
4755 separated list of percentage scalars. It is index paired with the files passed
4756 to :option:`read_iolog`.
4758 With scaling, it may be desirable to match the running time of all traces.
4759 This can be done with :option:`merge_blktrace_iters`. It is index paired with
4760 :option:`read_iolog` just like :option:`merge_blktrace_scalars`.
4762 In an example, given two traces, A and B, each 60s long. If we want to see
4763 the impact of trace A issuing IOs twice as fast and repeat trace A over the
4764 runtime of trace B, the following can be done::
4766 $ fio --read_iolog="<trace_a>:"<trace_b>" --merge_blktrace_file"<output_file>" --merge_blktrace_scalars="50:100" --merge_blktrace_iters="2:1"
4768 This runs trace A at 2x the speed twice for approximately the same runtime as
4769 a single run of trace B.
4772 CPU idleness profiling
4773 ----------------------
4775 In some cases, we want to understand CPU overhead in a test. For example, we
4776 test patches for the specific goodness of whether they reduce CPU usage.
4777 Fio implements a balloon approach to create a thread per CPU that runs at idle
4778 priority, meaning that it only runs when nobody else needs the cpu.
4779 By measuring the amount of work completed by the thread, idleness of each CPU
4780 can be derived accordingly.
4782 An unit work is defined as touching a full page of unsigned characters. Mean and
4783 standard deviation of time to complete an unit work is reported in "unit work"
4784 section. Options can be chosen to report detailed percpu idleness or overall
4785 system idleness by aggregating percpu stats.
4788 Verification and triggers
4789 -------------------------
4791 Fio is usually run in one of two ways, when data verification is done. The first
4792 is a normal write job of some sort with verify enabled. When the write phase has
4793 completed, fio switches to reads and verifies everything it wrote. The second
4794 model is running just the write phase, and then later on running the same job
4795 (but with reads instead of writes) to repeat the same I/O patterns and verify
4796 the contents. Both of these methods depend on the write phase being completed,
4797 as fio otherwise has no idea how much data was written.
4799 With verification triggers, fio supports dumping the current write state to
4800 local files. Then a subsequent read verify workload can load this state and know
4801 exactly where to stop. This is useful for testing cases where power is cut to a
4802 server in a managed fashion, for instance.
4804 A verification trigger consists of two things:
4806 1) Storing the write state of each job.
4807 2) Executing a trigger command.
4809 The write state is relatively small, on the order of hundreds of bytes to single
4810 kilobytes. It contains information on the number of completions done, the last X
4813 A trigger is invoked either through creation ('touch') of a specified file in
4814 the system, or through a timeout setting. If fio is run with
4815 :option:`--trigger-file`\= :file:`/tmp/trigger-file`, then it will continually
4816 check for the existence of :file:`/tmp/trigger-file`. When it sees this file, it
4817 will fire off the trigger (thus saving state, and executing the trigger
4820 For client/server runs, there's both a local and remote trigger. If fio is
4821 running as a server backend, it will send the job states back to the client for
4822 safe storage, then execute the remote trigger, if specified. If a local trigger
4823 is specified, the server will still send back the write state, but the client
4824 will then execute the trigger.
4826 Verification trigger example
4827 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4829 Let's say we want to run a powercut test on the remote Linux machine 'server'.
4830 Our write workload is in :file:`write-test.fio`. We want to cut power to 'server' at
4831 some point during the run, and we'll run this test from the safety or our local
4832 machine, 'localbox'. On the server, we'll start the fio backend normally::
4834 server# fio --server
4836 and on the client, we'll fire off the workload::
4838 localbox$ fio --client=server --trigger-file=/tmp/my-trigger --trigger-remote="bash -c \"echo b > /proc/sysrq-triger\""
4840 We set :file:`/tmp/my-trigger` as the trigger file, and we tell fio to execute::
4842 echo b > /proc/sysrq-trigger
4844 on the server once it has received the trigger and sent us the write state. This
4845 will work, but it's not **really** cutting power to the server, it's merely
4846 abruptly rebooting it. If we have a remote way of cutting power to the server
4847 through IPMI or similar, we could do that through a local trigger command
4848 instead. Let's assume we have a script that does IPMI reboot of a given hostname,
4849 ipmi-reboot. On localbox, we could then have run fio with a local trigger
4852 localbox$ fio --client=server --trigger-file=/tmp/my-trigger --trigger="ipmi-reboot server"
4854 For this case, fio would wait for the server to send us the write state, then
4855 execute ``ipmi-reboot server`` when that happened.
4857 Loading verify state
4858 ~~~~~~~~~~~~~~~~~~~~
4860 To load stored write state, a read verification job file must contain the
4861 :option:`verify_state_load` option. If that is set, fio will load the previously
4862 stored state. For a local fio run this is done by loading the files directly,
4863 and on a client/server run, the server backend will ask the client to send the
4864 files over and load them from there.
4870 Fio supports a variety of log file formats, for logging latencies, bandwidth,
4871 and IOPS. The logs share a common format, which looks like this:
4873 *time* (`msec`), *value*, *data direction*, *block size* (`bytes`),
4874 *offset* (`bytes`), *command priority*
4876 *Time* for the log entry is always in milliseconds. The *value* logged depends
4877 on the type of log, it will be one of the following:
4880 Value is latency in nsecs
4886 *Data direction* is one of the following:
4895 The entry's *block size* is always in bytes. The *offset* is the position in bytes
4896 from the start of the file for that particular I/O. The logging of the offset can be
4897 toggled with :option:`log_offset`.
4899 *Command priority* is 0 for normal priority and 1 for high priority. This is controlled
4900 by the ioengine specific :option:`cmdprio_percentage`.
4902 Fio defaults to logging every individual I/O but when windowed logging is set
4903 through :option:`log_avg_msec`, either the average (by default) or the maximum
4904 (:option:`log_max_value` is set) *value* seen over the specified period of time
4905 is recorded. Each *data direction* seen within the window period will aggregate
4906 its values in a separate row. Further, when using windowed logging the *block
4907 size* and *offset* entries will always contain 0.
4913 Normally fio is invoked as a stand-alone application on the machine where the
4914 I/O workload should be generated. However, the backend and frontend of fio can
4915 be run separately i.e., the fio server can generate an I/O workload on the "Device
4916 Under Test" while being controlled by a client on another machine.
4918 Start the server on the machine which has access to the storage DUT::
4922 where `args` defines what fio listens to. The arguments are of the form
4923 ``type,hostname`` or ``IP,port``. *type* is either ``ip`` (or ip4) for TCP/IP
4924 v4, ``ip6`` for TCP/IP v6, or ``sock`` for a local unix domain socket.
4925 *hostname* is either a hostname or IP address, and *port* is the port to listen
4926 to (only valid for TCP/IP, not a local socket). Some examples:
4930 Start a fio server, listening on all interfaces on the default port (8765).
4932 2) ``fio --server=ip:hostname,4444``
4934 Start a fio server, listening on IP belonging to hostname and on port 4444.
4936 3) ``fio --server=ip6:::1,4444``
4938 Start a fio server, listening on IPv6 localhost ::1 and on port 4444.
4940 4) ``fio --server=,4444``
4942 Start a fio server, listening on all interfaces on port 4444.
4944 5) ``fio --server=1.2.3.4``
4946 Start a fio server, listening on IP 1.2.3.4 on the default port.
4948 6) ``fio --server=sock:/tmp/fio.sock``
4950 Start a fio server, listening on the local socket :file:`/tmp/fio.sock`.
4952 Once a server is running, a "client" can connect to the fio server with::
4954 fio <local-args> --client=<server> <remote-args> <job file(s)>
4956 where `local-args` are arguments for the client where it is running, `server`
4957 is the connect string, and `remote-args` and `job file(s)` are sent to the
4958 server. The `server` string follows the same format as it does on the server
4959 side, to allow IP/hostname/socket and port strings.
4961 Fio can connect to multiple servers this way::
4963 fio --client=<server1> <job file(s)> --client=<server2> <job file(s)>
4965 If the job file is located on the fio server, then you can tell the server to
4966 load a local file as well. This is done by using :option:`--remote-config` ::
4968 fio --client=server --remote-config /path/to/file.fio
4970 Then fio will open this local (to the server) job file instead of being passed
4971 one from the client.
4973 If you have many servers (example: 100 VMs/containers), you can input a pathname
4974 of a file containing host IPs/names as the parameter value for the
4975 :option:`--client` option. For example, here is an example :file:`host.list`
4976 file containing 2 hostnames::
4978 host1.your.dns.domain
4979 host2.your.dns.domain
4981 The fio command would then be::
4983 fio --client=host.list <job file(s)>
4985 In this mode, you cannot input server-specific parameters or job files -- all
4986 servers receive the same job file.
4988 In order to let ``fio --client`` runs use a shared filesystem from multiple
4989 hosts, ``fio --client`` now prepends the IP address of the server to the
4990 filename. For example, if fio is using the directory :file:`/mnt/nfs/fio` and is
4991 writing filename :file:`fileio.tmp`, with a :option:`--client` `hostfile`
4992 containing two hostnames ``h1`` and ``h2`` with IP addresses 192.168.10.120 and
4993 192.168.10.121, then fio will create two files::
4995 /mnt/nfs/fio/192.168.10.120.fileio.tmp
4996 /mnt/nfs/fio/192.168.10.121.fileio.tmp
4998 Terse output in client/server mode will differ slightly from what is produced
4999 when fio is run in stand-alone mode. See the terse output section for details.