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 Limit runtime. The test will run until it completes the configured I/O
690 workload or until it has run for this specified amount of time, whichever
691 occurs first. It can be quite hard to determine for how long a specified
692 job will run, so this parameter is handy to cap the total runtime to a
693 given time. When the unit is omitted, the value is interpreted in
696 .. option:: time_based
698 If set, fio will run for the duration of the :option:`runtime` specified
699 even if the file(s) are completely read or written. It will simply loop over
700 the same workload as many times as the :option:`runtime` allows.
702 .. option:: startdelay=irange(time)
704 Delay the start of job for the specified amount of time. Can be a single
705 value or a range. When given as a range, each thread will choose a value
706 randomly from within the range. Value is in seconds if a unit is omitted.
708 .. option:: ramp_time=time
710 If set, fio will run the specified workload for this amount of time before
711 logging any performance numbers. Useful for letting performance settle
712 before logging results, thus minimizing the runtime required for stable
713 results. Note that the ``ramp_time`` is considered lead in time for a job,
714 thus it will increase the total runtime if a special timeout or
715 :option:`runtime` is specified. When the unit is omitted, the value is
718 .. option:: clocksource=str
720 Use the given clocksource as the base of timing. The supported options are:
723 :manpage:`gettimeofday(2)`
726 :manpage:`clock_gettime(2)`
729 Internal CPU clock source
731 cpu is the preferred clocksource if it is reliable, as it is very fast (and
732 fio is heavy on time calls). Fio will automatically use this clocksource if
733 it's supported and considered reliable on the system it is running on,
734 unless another clocksource is specifically set. For x86/x86-64 CPUs, this
735 means supporting TSC Invariant.
737 .. option:: gtod_reduce=bool
739 Enable all of the :manpage:`gettimeofday(2)` reducing options
740 (:option:`disable_clat`, :option:`disable_slat`, :option:`disable_bw_measurement`) plus
741 reduce precision of the timeout somewhat to really shrink the
742 :manpage:`gettimeofday(2)` call count. With this option enabled, we only do
743 about 0.4% of the :manpage:`gettimeofday(2)` calls we would have done if all
744 time keeping was enabled.
746 .. option:: gtod_cpu=int
748 Sometimes it's cheaper to dedicate a single thread of execution to just
749 getting the current time. Fio (and databases, for instance) are very
750 intensive on :manpage:`gettimeofday(2)` calls. With this option, you can set
751 one CPU aside for doing nothing but logging current time to a shared memory
752 location. Then the other threads/processes that run I/O workloads need only
753 copy that segment, instead of entering the kernel with a
754 :manpage:`gettimeofday(2)` call. The CPU set aside for doing these time
755 calls will be excluded from other uses. Fio will manually clear it from the
756 CPU mask of other jobs.
762 .. option:: directory=str
764 Prefix filenames with this directory. Used to place files in a different
765 location than :file:`./`. You can specify a number of directories by
766 separating the names with a ':' character. These directories will be
767 assigned equally distributed to job clones created by :option:`numjobs` as
768 long as they are using generated filenames. If specific `filename(s)` are
769 set fio will use the first listed directory, and thereby matching the
770 `filename` semantic (which generates a file for each clone if not
771 specified, but lets all clones use the same file if set).
773 See the :option:`filename` option for information on how to escape "``:``"
774 characters within the directory path itself.
776 Note: To control the directory fio will use for internal state files
777 use :option:`--aux-path`.
779 .. option:: filename=str
781 Fio normally makes up a `filename` based on the job name, thread number, and
782 file number (see :option:`filename_format`). If you want to share files
783 between threads in a job or several
784 jobs with fixed file paths, specify a `filename` for each of them to override
785 the default. If the ioengine is file based, you can specify a number of files
786 by separating the names with a ':' colon. So if you wanted a job to open
787 :file:`/dev/sda` and :file:`/dev/sdb` as the two working files, you would use
788 ``filename=/dev/sda:/dev/sdb``. This also means that whenever this option is
789 specified, :option:`nrfiles` is ignored. The size of regular files specified
790 by this option will be :option:`size` divided by number of files unless an
791 explicit size is specified by :option:`filesize`.
793 Each colon in the wanted path must be escaped with a ``\``
794 character. For instance, if the path is :file:`/dev/dsk/foo@3,0:c` then you
795 would use ``filename=/dev/dsk/foo@3,0\:c`` and if the path is
796 :file:`F:\\filename` then you would use ``filename=F\:\filename``.
798 On Windows, disk devices are accessed as :file:`\\\\.\\PhysicalDrive0` for
799 the first device, :file:`\\\\.\\PhysicalDrive1` for the second etc.
800 Note: Windows and FreeBSD prevent write access to areas
801 of the disk containing in-use data (e.g. filesystems).
803 The filename "`-`" is a reserved name, meaning *stdin* or *stdout*. Which
804 of the two depends on the read/write direction set.
806 .. option:: filename_format=str
808 If sharing multiple files between jobs, it is usually necessary to have fio
809 generate the exact names that you want. By default, fio will name a file
810 based on the default file format specification of
811 :file:`jobname.jobnumber.filenumber`. With this option, that can be
812 customized. Fio will recognize and replace the following keywords in this
816 The name of the worker thread or process.
818 IP of the fio process when using client/server mode.
820 The incremental number of the worker thread or process.
822 The incremental number of the file for that worker thread or
825 To have dependent jobs share a set of files, this option can be set to have
826 fio generate filenames that are shared between the two. For instance, if
827 :file:`testfiles.$filenum` is specified, file number 4 for any job will be
828 named :file:`testfiles.4`. The default of :file:`$jobname.$jobnum.$filenum`
829 will be used if no other format specifier is given.
831 If you specify a path then the directories will be created up to the
832 main directory for the file. So for example if you specify
833 ``filename_format=a/b/c/$jobnum`` then the directories a/b/c will be
834 created before the file setup part of the job. If you specify
835 :option:`directory` then the path will be relative that directory,
836 otherwise it is treated as the absolute path.
838 .. option:: unique_filename=bool
840 To avoid collisions between networked clients, fio defaults to prefixing any
841 generated filenames (with a directory specified) with the source of the
842 client connecting. To disable this behavior, set this option to 0.
844 .. option:: opendir=str
846 Recursively open any files below directory `str`. This accepts only a
847 single directory and unlike related options, colons appearing in the
848 path must not be escaped.
850 .. option:: lockfile=str
852 Fio defaults to not locking any files before it does I/O to them. If a file
853 or file descriptor is shared, fio can serialize I/O to that file to make the
854 end result consistent. This is usual for emulating real workloads that share
855 files. The lock modes are:
858 No locking. The default.
860 Only one thread or process may do I/O at a time, excluding all
863 Read-write locking on the file. Many readers may
864 access the file at the same time, but writes get exclusive access.
866 .. option:: nrfiles=int
868 Number of files to use for this job. Defaults to 1. The size of files
869 will be :option:`size` divided by this unless explicit size is specified by
870 :option:`filesize`. Files are created for each thread separately, and each
871 file will have a file number within its name by default, as explained in
872 :option:`filename` section.
875 .. option:: openfiles=int
877 Number of files to keep open at the same time. Defaults to the same as
878 :option:`nrfiles`, can be set smaller to limit the number simultaneous
881 .. option:: file_service_type=str
883 Defines how fio decides which file from a job to service next. The following
887 Choose a file at random.
890 Round robin over opened files. This is the default.
893 Finish one file before moving on to the next. Multiple files can
894 still be open depending on :option:`openfiles`.
897 Use a *Zipf* distribution to decide what file to access.
900 Use a *Pareto* distribution to decide what file to access.
903 Use a *Gaussian* (normal) distribution to decide what file to
909 For *random*, *roundrobin*, and *sequential*, a postfix can be appended to
910 tell fio how many I/Os to issue before switching to a new file. For example,
911 specifying ``file_service_type=random:8`` would cause fio to issue
912 8 I/Os before selecting a new file at random. For the non-uniform
913 distributions, a floating point postfix can be given to influence how the
914 distribution is skewed. See :option:`random_distribution` for a description
915 of how that would work.
917 .. option:: ioscheduler=str
919 Attempt to switch the device hosting the file to the specified I/O scheduler
922 .. option:: create_serialize=bool
924 If true, serialize the file creation for the jobs. This may be handy to
925 avoid interleaving of data files, which may greatly depend on the filesystem
926 used and even the number of processors in the system. Default: true.
928 .. option:: create_fsync=bool
930 :manpage:`fsync(2)` the data file after creation. This is the default.
932 .. option:: create_on_open=bool
934 If true, don't pre-create files but allow the job's open() to create a file
935 when it's time to do I/O. Default: false -- pre-create all necessary files
938 .. option:: create_only=bool
940 If true, fio will only run the setup phase of the job. If files need to be
941 laid out or updated on disk, only that will be done -- the actual job contents
942 are not executed. Default: false.
944 .. option:: allow_file_create=bool
946 If true, fio is permitted to create files as part of its workload. If this
947 option is false, then fio will error out if
948 the files it needs to use don't already exist. Default: true.
950 .. option:: allow_mounted_write=bool
952 If this isn't set, fio will abort jobs that are destructive (e.g. that write)
953 to what appears to be a mounted device or partition. This should help catch
954 creating inadvertently destructive tests, not realizing that the test will
955 destroy data on the mounted file system. Note that some platforms don't allow
956 writing against a mounted device regardless of this option. Default: false.
958 .. option:: pre_read=bool
960 If this is given, files will be pre-read into memory before starting the
961 given I/O operation. This will also clear the :option:`invalidate` flag,
962 since it is pointless to pre-read and then drop the cache. This will only
963 work for I/O engines that are seek-able, since they allow you to read the
964 same data multiple times. Thus it will not work on non-seekable I/O engines
965 (e.g. network, splice). Default: false.
967 .. option:: unlink=bool
969 Unlink the job files when done. Not the default, as repeated runs of that
970 job would then waste time recreating the file set again and again. Default:
973 .. option:: unlink_each_loop=bool
975 Unlink job files after each iteration or loop. Default: false.
977 .. option:: zonemode=str
982 The :option:`zonerange`, :option:`zonesize`,
983 :option `zonecapacity` and option:`zoneskip`
984 parameters are ignored.
986 I/O happens in a single zone until
987 :option:`zonesize` bytes have been transferred.
988 After that number of bytes has been
989 transferred processing of the next zone
990 starts. :option `zonecapacity` is ignored.
992 Zoned block device mode. I/O happens
993 sequentially in each zone, even if random I/O
994 has been selected. Random I/O happens across
995 all zones instead of being restricted to a
996 single zone. The :option:`zoneskip` parameter
997 is ignored. :option:`zonerange` and
998 :option:`zonesize` must be identical.
999 Trim is handled using a zone reset operation.
1000 Trim only considers non-empty sequential write
1001 required and sequential write preferred zones.
1003 .. option:: zonerange=int
1005 Size of a single zone. See also :option:`zonesize` and
1008 .. option:: zonesize=int
1010 For :option:`zonemode` =strided, this is the number of bytes to
1011 transfer before skipping :option:`zoneskip` bytes. If this parameter
1012 is smaller than :option:`zonerange` then only a fraction of each zone
1013 with :option:`zonerange` bytes will be accessed. If this parameter is
1014 larger than :option:`zonerange` then each zone will be accessed
1015 multiple times before skipping to the next zone.
1017 For :option:`zonemode` =zbd, this is the size of a single zone. The
1018 :option:`zonerange` parameter is ignored in this mode.
1021 .. option:: zonecapacity=int
1023 For :option:`zonemode` =zbd, this defines the capacity of a single zone,
1024 which is the accessible area starting from the zone start address.
1025 This parameter only applies when using :option:`zonemode` =zbd in
1026 combination with regular block devices. If not specified it defaults to
1027 the zone size. If the target device is a zoned block device, the zone
1028 capacity is obtained from the device information and this option is
1031 .. option:: zoneskip=int
1033 For :option:`zonemode` =strided, the number of bytes to skip after
1034 :option:`zonesize` bytes of data have been transferred. This parameter
1035 must be zero for :option:`zonemode` =zbd.
1037 .. option:: read_beyond_wp=bool
1039 This parameter applies to :option:`zonemode` =zbd only.
1041 Zoned block devices are block devices that consist of multiple zones.
1042 Each zone has a type, e.g. conventional or sequential. A conventional
1043 zone can be written at any offset that is a multiple of the block
1044 size. Sequential zones must be written sequentially. The position at
1045 which a write must occur is called the write pointer. A zoned block
1046 device can be either drive managed, host managed or host aware. For
1047 host managed devices the host must ensure that writes happen
1048 sequentially. Fio recognizes host managed devices and serializes
1049 writes to sequential zones for these devices.
1051 If a read occurs in a sequential zone beyond the write pointer then
1052 the zoned block device will complete the read without reading any data
1053 from the storage medium. Since such reads lead to unrealistically high
1054 bandwidth and IOPS numbers fio only reads beyond the write pointer if
1055 explicitly told to do so. Default: false.
1057 .. option:: max_open_zones=int
1059 A zone of a zoned block device is in the open state when it is partially
1060 written (i.e. not all sectors of the zone have been written). Zoned
1061 block devices may have a limit on the total number of zones that can
1062 be simultaneously in the open state, that is, the number of zones that
1063 can be written to simultaneously. The :option:`max_open_zones` parameter
1064 limits the number of zones to which write commands are issued by all fio
1065 jobs, that is, limits the number of zones that will be in the open
1066 state. This parameter is relevant only if the :option:`zonemode` =zbd is
1067 used. The default value is always equal to maximum number of open zones
1068 of the target zoned block device and a value higher than this limit
1069 cannot be specified by users unless the option
1070 :option:`ignore_zone_limits` is specified. When
1071 :option:`ignore_zone_limits` is specified or the target device has no
1072 limit on the number of zones that can be in an open state,
1073 :option:`max_open_zones` can specify 0 to disable any limit on the
1074 number of zones that can be simultaneously written to by all jobs.
1076 .. option:: job_max_open_zones=int
1078 In the same manner as :option:`max_open_zones`, limit the number of open
1079 zones per fio job, that is, the number of zones that a single job can
1080 simultaneously write to. A value of zero indicates no limit.
1083 .. option:: ignore_zone_limits=bool
1085 If this option is used, fio will ignore the maximum number of open
1086 zones limit of the zoned block device in use, thus allowing the
1087 option :option:`max_open_zones` value to be larger than the device
1088 reported limit. Default: false.
1090 .. option:: zone_reset_threshold=float
1092 A number between zero and one that indicates the ratio of written bytes
1093 in the zones with write pointers in the IO range to the size of the IO
1094 range. When current ratio is above this ratio, zones are reset
1095 periodically as :option:`zone_reset_frequency` specifies. If there are
1096 multiple jobs when using this option, the IO range for all write jobs
1099 .. option:: zone_reset_frequency=float
1101 A number between zero and one that indicates how often a zone reset
1102 should be issued if the zone reset threshold has been exceeded. A zone
1103 reset is submitted after each (1 / zone_reset_frequency) write
1104 requests. This and the previous parameter can be used to simulate
1105 garbage collection activity.
1111 .. option:: direct=bool
1113 If value is true, use non-buffered I/O. This is usually O_DIRECT. Note that
1114 OpenBSD and ZFS on Solaris don't support direct I/O. On Windows the synchronous
1115 ioengines don't support direct I/O. Default: false.
1117 .. option:: buffered=bool
1119 If value is true, use buffered I/O. This is the opposite of the
1120 :option:`direct` option. Defaults to true.
1122 .. option:: readwrite=str, rw=str
1124 Type of I/O pattern. Accepted values are:
1131 Sequential trims (Linux block devices and SCSI
1132 character devices only).
1138 Random trims (Linux block devices and SCSI
1139 character devices only).
1141 Sequential mixed reads and writes.
1143 Random mixed reads and writes.
1145 Sequential trim+write sequences. Blocks will be trimmed first,
1146 then the same blocks will be written to. So if ``io_size=64K``
1147 is specified, Fio will trim a total of 64K bytes and also
1148 write 64K bytes on the same trimmed blocks. This behaviour
1149 will be consistent with ``number_ios`` or other Fio options
1150 limiting the total bytes or number of I/O's.
1152 Like trimwrite, but uses random offsets rather
1153 than sequential writes.
1155 Fio defaults to read if the option is not specified. For the mixed I/O
1156 types, the default is to split them 50/50. For certain types of I/O the
1157 result may still be skewed a bit, since the speed may be different.
1159 It is possible to specify the number of I/Os to do before getting a new
1160 offset by appending ``:<nr>`` to the end of the string given. For a
1161 random read, it would look like ``rw=randread:8`` for passing in an offset
1162 modifier with a value of 8. If the suffix is used with a sequential I/O
1163 pattern, then the *<nr>* value specified will be **added** to the generated
1164 offset for each I/O turning sequential I/O into sequential I/O with holes.
1165 For instance, using ``rw=write:4k`` will skip 4k for every write. Also see
1166 the :option:`rw_sequencer` option.
1168 .. option:: rw_sequencer=str
1170 If an offset modifier is given by appending a number to the ``rw=<str>``
1171 line, then this option controls how that number modifies the I/O offset
1172 being generated. Accepted values are:
1175 Generate sequential offset.
1177 Generate the same offset.
1179 ``sequential`` is only useful for random I/O, where fio would normally
1180 generate a new random offset for every I/O. If you append e.g. 8 to
1181 randread, i.e. ``rw=randread:8`` you would get a new random offset for
1182 every 8 I/Os. The result would be a sequence of 8 sequential offsets
1183 with a random starting point. However this behavior may change if a
1184 sequential I/O reaches end of the file. As sequential I/O is already
1185 sequential, setting ``sequential`` for that would not result in any
1186 difference. ``identical`` behaves in a similar fashion, except it sends
1187 the same offset 8 number of times before generating a new offset.
1192 rw_sequencer=sequential
1195 The generated sequence of offsets will look like this:
1196 4k, 8k, 12k, 16k, 20k, 24k, 28k, 32k, 92k, 96k, 100k, 104k, 108k,
1197 112k, 116k, 120k, 48k, 52k ...
1202 rw_sequencer=identical
1205 The generated sequence of offsets will look like this:
1206 4k, 4k, 4k, 4k, 4k, 4k, 4k, 4k, 92k, 92k, 92k, 92k, 92k, 92k, 92k, 92k,
1209 .. option:: unified_rw_reporting=str
1211 Fio normally reports statistics on a per data direction basis, meaning that
1212 reads, writes, and trims are accounted and reported separately. This option
1213 determines whether fio reports the results normally, summed together, or as
1215 Accepted values are:
1218 Normal statistics reporting.
1221 Statistics are summed per data direction and reported together.
1224 Statistics are reported normally, followed by the mixed statistics.
1227 Backward-compatible alias for **none**.
1230 Backward-compatible alias for **mixed**.
1235 .. option:: randrepeat=bool
1237 Seed all random number generators in a predictable way so the pattern
1238 is repeatable across runs. Default: true.
1240 .. option:: allrandrepeat=bool
1242 Alias for :option:`randrepeat`. Default: true.
1244 .. option:: randseed=int
1246 Seed the random number generators based on this seed value, to be able to
1247 control what sequence of output is being generated. If not set, the random
1248 sequence depends on the :option:`randrepeat` setting.
1250 .. option:: fallocate=str
1252 Whether pre-allocation is performed when laying down files.
1253 Accepted values are:
1256 Do not pre-allocate space.
1259 Use a platform's native pre-allocation call but fall back to
1260 **none** behavior if it fails/is not implemented.
1263 Pre-allocate via :manpage:`posix_fallocate(3)`.
1266 Pre-allocate via :manpage:`fallocate(2)` with
1267 FALLOC_FL_KEEP_SIZE set.
1270 Extend file to final size via :manpage:`ftruncate(2)`
1271 instead of allocating.
1274 Backward-compatible alias for **none**.
1277 Backward-compatible alias for **posix**.
1279 May not be available on all supported platforms. **keep** is only available
1280 on Linux. If using ZFS on Solaris this cannot be set to **posix**
1281 because ZFS doesn't support pre-allocation. Default: **native** if any
1282 pre-allocation methods except **truncate** are available, **none** if not.
1284 Note that using **truncate** on Windows will interact surprisingly
1285 with non-sequential write patterns. When writing to a file that has
1286 been extended by setting the end-of-file information, Windows will
1287 backfill the unwritten portion of the file up to that offset with
1288 zeroes before issuing the new write. This means that a single small
1289 write to the end of an extended file will stall until the entire
1290 file has been filled with zeroes.
1292 .. option:: fadvise_hint=str
1294 Use :manpage:`posix_fadvise(2)` or :manpage:`posix_fadvise(2)` to
1295 advise the kernel on what I/O patterns are likely to be issued.
1296 Accepted values are:
1299 Backwards-compatible hint for "no hint".
1302 Backwards compatible hint for "advise with fio workload type". This
1303 uses **FADV_RANDOM** for a random workload, and **FADV_SEQUENTIAL**
1304 for a sequential workload.
1307 Advise using **FADV_SEQUENTIAL**.
1310 Advise using **FADV_RANDOM**.
1313 Advise using **FADV_NOREUSE**. This may be a no-op on older Linux
1314 kernels. Since Linux 6.3, it provides a hint to the LRU algorithm.
1315 See the :manpage:`posix_fadvise(2)` man page.
1317 .. option:: write_hint=str
1319 Use :manpage:`fcntl(2)` to advise the kernel what life time to expect
1320 from a write. Only supported on Linux, as of version 4.13. Accepted
1324 No particular life time associated with this file.
1327 Data written to this file has a short life time.
1330 Data written to this file has a medium life time.
1333 Data written to this file has a long life time.
1336 Data written to this file has a very long life time.
1338 The values are all relative to each other, and no absolute meaning
1339 should be associated with them.
1341 .. option:: offset=int
1343 Start I/O at the provided offset in the file, given as either a fixed size in
1344 bytes, zones or a percentage. If a percentage is given, the generated offset will be
1345 aligned to the minimum ``blocksize`` or to the value of ``offset_align`` if
1346 provided. Data before the given offset will not be touched. This
1347 effectively caps the file size at `real_size - offset`. Can be combined with
1348 :option:`size` to constrain the start and end range of the I/O workload.
1349 A percentage can be specified by a number between 1 and 100 followed by '%',
1350 for example, ``offset=20%`` to specify 20%. In ZBD mode, value can be set as
1351 number of zones using 'z'.
1353 .. option:: offset_align=int
1355 If set to non-zero value, the byte offset generated by a percentage ``offset``
1356 is aligned upwards to this value. Defaults to 0 meaning that a percentage
1357 offset is aligned to the minimum block size.
1359 .. option:: offset_increment=int
1361 If this is provided, then the real offset becomes `offset + offset_increment
1362 * thread_number`, where the thread number is a counter that starts at 0 and
1363 is incremented for each sub-job (i.e. when :option:`numjobs` option is
1364 specified). This option is useful if there are several jobs which are
1365 intended to operate on a file in parallel disjoint segments, with even
1366 spacing between the starting points. Percentages can be used for this option.
1367 If a percentage is given, the generated offset will be aligned to the minimum
1368 ``blocksize`` or to the value of ``offset_align`` if provided. In ZBD mode, value can
1369 also be set as number of zones using 'z'.
1371 .. option:: number_ios=int
1373 Fio will normally perform I/Os until it has exhausted the size of the region
1374 set by :option:`size`, or if it exhaust the allocated time (or hits an error
1375 condition). With this setting, the range/size can be set independently of
1376 the number of I/Os to perform. When fio reaches this number, it will exit
1377 normally and report status. Note that this does not extend the amount of I/O
1378 that will be done, it will only stop fio if this condition is met before
1379 other end-of-job criteria.
1381 .. option:: fsync=int
1383 If writing to a file, issue an :manpage:`fsync(2)` (or its equivalent) of
1384 the dirty data for every number of blocks given. For example, if you give 32
1385 as a parameter, fio will sync the file after every 32 writes issued. If fio is
1386 using non-buffered I/O, we may not sync the file. The exception is the sg
1387 I/O engine, which synchronizes the disk cache anyway. Defaults to 0, which
1388 means fio does not periodically issue and wait for a sync to complete. Also
1389 see :option:`end_fsync` and :option:`fsync_on_close`.
1391 .. option:: fdatasync=int
1393 Like :option:`fsync` but uses :manpage:`fdatasync(2)` to only sync data and
1394 not metadata blocks. In Windows, DragonFlyBSD or OSX there is no
1395 :manpage:`fdatasync(2)` so this falls back to using :manpage:`fsync(2)`.
1396 Defaults to 0, which means fio does not periodically issue and wait for a
1397 data-only sync to complete.
1399 .. option:: write_barrier=int
1401 Make every `N-th` write a barrier write.
1403 .. option:: sync_file_range=str:int
1405 Use :manpage:`sync_file_range(2)` for every `int` number of write
1406 operations. Fio will track range of writes that have happened since the last
1407 :manpage:`sync_file_range(2)` call. `str` can currently be one or more of:
1410 SYNC_FILE_RANGE_WAIT_BEFORE
1412 SYNC_FILE_RANGE_WRITE
1414 SYNC_FILE_RANGE_WAIT_AFTER
1416 So if you do ``sync_file_range=wait_before,write:8``, fio would use
1417 ``SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE`` for every 8
1418 writes. Also see the :manpage:`sync_file_range(2)` man page. This option is
1421 .. option:: overwrite=bool
1423 If true, writes to a file will always overwrite existing data. If the file
1424 doesn't already exist, it will be created before the write phase begins. If
1425 the file exists and is large enough for the specified write phase, nothing
1426 will be done. Default: false.
1428 .. option:: end_fsync=bool
1430 If true, :manpage:`fsync(2)` file contents when a write stage has completed.
1433 .. option:: fsync_on_close=bool
1435 If true, fio will :manpage:`fsync(2)` a dirty file on close. This differs
1436 from :option:`end_fsync` in that it will happen on every file close, not
1437 just at the end of the job. Default: false.
1439 .. option:: rwmixread=int
1441 Percentage of a mixed workload that should be reads. Default: 50.
1443 .. option:: rwmixwrite=int
1445 Percentage of a mixed workload that should be writes. If both
1446 :option:`rwmixread` and :option:`rwmixwrite` is given and the values do not
1447 add up to 100%, the latter of the two will be used to override the
1448 first. This may interfere with a given rate setting, if fio is asked to
1449 limit reads or writes to a certain rate. If that is the case, then the
1450 distribution may be skewed. Default: 50.
1452 .. option:: random_distribution=str:float[:float][,str:float][,str:float]
1454 By default, fio will use a completely uniform random distribution when asked
1455 to perform random I/O. Sometimes it is useful to skew the distribution in
1456 specific ways, ensuring that some parts of the data is more hot than others.
1457 fio includes the following distribution models:
1460 Uniform random distribution
1469 Normal (Gaussian) distribution
1472 Zoned random distribution
1475 Zone absolute random distribution
1477 When using a **zipf** or **pareto** distribution, an input value is also
1478 needed to define the access pattern. For **zipf**, this is the `Zipf
1479 theta`. For **pareto**, it's the `Pareto power`. Fio includes a test
1480 program, :command:`fio-genzipf`, that can be used visualize what the given input
1481 values will yield in terms of hit rates. If you wanted to use **zipf** with
1482 a `theta` of 1.2, you would use ``random_distribution=zipf:1.2`` as the
1483 option. If a non-uniform model is used, fio will disable use of the random
1484 map. For the **normal** distribution, a normal (Gaussian) deviation is
1485 supplied as a value between 0 and 100.
1487 The second, optional float is allowed for **pareto**, **zipf** and **normal** distributions.
1488 It allows one to set base of distribution in non-default place, giving more control
1489 over most probable outcome. This value is in range [0-1] which maps linearly to
1490 range of possible random values.
1491 Defaults are: random for **pareto** and **zipf**, and 0.5 for **normal**.
1492 If you wanted to use **zipf** with a `theta` of 1.2 centered on 1/4 of allowed value range,
1493 you would use ``random_distribution=zipf:1.2:0.25``.
1495 For a **zoned** distribution, fio supports specifying percentages of I/O
1496 access that should fall within what range of the file or device. For
1497 example, given a criteria of:
1499 * 60% of accesses should be to the first 10%
1500 * 30% of accesses should be to the next 20%
1501 * 8% of accesses should be to the next 30%
1502 * 2% of accesses should be to the next 40%
1504 we can define that through zoning of the random accesses. For the above
1505 example, the user would do::
1507 random_distribution=zoned:60/10:30/20:8/30:2/40
1509 A **zoned_abs** distribution works exactly like the **zoned**, except
1510 that it takes absolute sizes. For example, let's say you wanted to
1511 define access according to the following criteria:
1513 * 60% of accesses should be to the first 20G
1514 * 30% of accesses should be to the next 100G
1515 * 10% of accesses should be to the next 500G
1517 we can define an absolute zoning distribution with:
1519 random_distribution=zoned_abs=60/20G:30/100G:10/500g
1521 For both **zoned** and **zoned_abs**, fio supports defining up to
1524 Similarly to how :option:`bssplit` works for setting ranges and
1525 percentages of block sizes. Like :option:`bssplit`, it's possible to
1526 specify separate zones for reads, writes, and trims. If just one set
1527 is given, it'll apply to all of them. This goes for both **zoned**
1528 **zoned_abs** distributions.
1530 .. option:: percentage_random=int[,int][,int]
1532 For a random workload, set how big a percentage should be random. This
1533 defaults to 100%, in which case the workload is fully random. It can be set
1534 from anywhere from 0 to 100. Setting it to 0 would make the workload fully
1535 sequential. Any setting in between will result in a random mix of sequential
1536 and random I/O, at the given percentages. Comma-separated values may be
1537 specified for reads, writes, and trims as described in :option:`blocksize`.
1539 .. option:: norandommap
1541 Normally fio will cover every block of the file when doing random I/O. If
1542 this option is given, fio will just get a new random offset without looking
1543 at past I/O history. This means that some blocks may not be read or written,
1544 and that some blocks may be read/written more than once. If this option is
1545 used with :option:`verify` and multiple blocksizes (via :option:`bsrange`),
1546 only intact blocks are verified, i.e., partially-overwritten blocks are
1547 ignored. With an async I/O engine and an I/O depth > 1, it is possible for
1548 the same block to be overwritten, which can cause verification errors. Either
1549 do not use norandommap in this case, or also use the lfsr random generator.
1551 .. option:: softrandommap=bool
1553 See :option:`norandommap`. If fio runs with the random block map enabled and
1554 it fails to allocate the map, if this option is set it will continue without
1555 a random block map. As coverage will not be as complete as with random maps,
1556 this option is disabled by default.
1558 .. option:: random_generator=str
1560 Fio supports the following engines for generating I/O offsets for random I/O:
1563 Strong 2^88 cycle random number generator.
1565 Linear feedback shift register generator.
1567 Strong 64-bit 2^258 cycle random number generator.
1569 **tausworthe** is a strong random number generator, but it requires tracking
1570 on the side if we want to ensure that blocks are only read or written
1571 once. **lfsr** guarantees that we never generate the same offset twice, and
1572 it's also less computationally expensive. It's not a true random generator,
1573 however, though for I/O purposes it's typically good enough. **lfsr** only
1574 works with single block sizes, not with workloads that use multiple block
1575 sizes. If used with such a workload, fio may read or write some blocks
1576 multiple times. The default value is **tausworthe**, unless the required
1577 space exceeds 2^32 blocks. If it does, then **tausworthe64** is
1578 selected automatically.
1584 .. option:: blocksize=int[,int][,int], bs=int[,int][,int]
1586 The block size in bytes used for I/O units. Default: 4096. A single value
1587 applies to reads, writes, and trims. Comma-separated values may be
1588 specified for reads, writes, and trims. A value not terminated in a comma
1589 applies to subsequent types.
1594 means 256k for reads, writes and trims.
1597 means 8k for reads, 32k for writes and trims.
1600 means 8k for reads, 32k for writes, and default for trims.
1603 means default for reads, 8k for writes and trims.
1606 means default for reads, 8k for writes, and default for trims.
1608 .. option:: blocksize_range=irange[,irange][,irange], bsrange=irange[,irange][,irange]
1610 A range of block sizes in bytes for I/O units. The issued I/O unit will
1611 always be a multiple of the minimum size, unless
1612 :option:`blocksize_unaligned` is set.
1614 Comma-separated ranges may be specified for reads, writes, and trims as
1615 described in :option:`blocksize`.
1617 Example: ``bsrange=1k-4k,2k-8k``.
1619 .. option:: bssplit=str[,str][,str]
1621 Sometimes you want even finer grained control of the block sizes
1622 issued, not just an even split between them. This option allows you to
1623 weight various block sizes, so that you are able to define a specific
1624 amount of block sizes issued. The format for this option is::
1626 bssplit=blocksize/percentage:blocksize/percentage
1628 for as many block sizes as needed. So if you want to define a workload
1629 that has 50% 64k blocks, 10% 4k blocks, and 40% 32k blocks, you would
1632 bssplit=4k/10:64k/50:32k/40
1634 Ordering does not matter. If the percentage is left blank, fio will
1635 fill in the remaining values evenly. So a bssplit option like this one::
1637 bssplit=4k/50:1k/:32k/
1639 would have 50% 4k ios, and 25% 1k and 32k ios. The percentages always
1640 add up to 100, if bssplit is given a range that adds up to more, it
1643 Comma-separated values may be specified for reads, writes, and trims as
1644 described in :option:`blocksize`.
1646 If you want a workload that has 50% 2k reads and 50% 4k reads, while
1647 having 90% 4k writes and 10% 8k writes, you would specify::
1649 bssplit=2k/50:4k/50,4k/90:8k/10
1651 Fio supports defining up to 64 different weights for each data
1654 .. option:: blocksize_unaligned, bs_unaligned
1656 If set, fio will issue I/O units with any size within
1657 :option:`blocksize_range`, not just multiples of the minimum size. This
1658 typically won't work with direct I/O, as that normally requires sector
1661 .. option:: bs_is_seq_rand=bool
1663 If this option is set, fio will use the normal read,write blocksize settings
1664 as sequential,random blocksize settings instead. Any random read or write
1665 will use the WRITE blocksize settings, and any sequential read or write will
1666 use the READ blocksize settings.
1668 .. option:: blockalign=int[,int][,int], ba=int[,int][,int]
1670 Boundary to which fio will align random I/O units. Default:
1671 :option:`blocksize`. Minimum alignment is typically 512b for using direct
1672 I/O, though it usually depends on the hardware block size. This option is
1673 mutually exclusive with using a random map for files, so it will turn off
1674 that option. Comma-separated values may be specified for reads, writes, and
1675 trims as described in :option:`blocksize`.
1681 .. option:: zero_buffers
1683 Initialize buffers with all zeros. Default: fill buffers with random data.
1685 .. option:: refill_buffers
1687 If this option is given, fio will refill the I/O buffers on every
1688 submit. Only makes sense if :option:`zero_buffers` isn't specified,
1689 naturally. Defaults to being unset i.e., the buffer is only filled at
1690 init time and the data in it is reused when possible but if any of
1691 :option:`verify`, :option:`buffer_compress_percentage` or
1692 :option:`dedupe_percentage` are enabled then `refill_buffers` is also
1693 automatically enabled.
1695 .. option:: scramble_buffers=bool
1697 If :option:`refill_buffers` is too costly and the target is using data
1698 deduplication, then setting this option will slightly modify the I/O buffer
1699 contents to defeat normal de-dupe attempts. This is not enough to defeat
1700 more clever block compression attempts, but it will stop naive dedupe of
1701 blocks. Default: true.
1703 .. option:: buffer_compress_percentage=int
1705 If this is set, then fio will attempt to provide I/O buffer content
1706 (on WRITEs) that compresses to the specified level. Fio does this by
1707 providing a mix of random data followed by fixed pattern data. The
1708 fixed pattern is either zeros, or the pattern specified by
1709 :option:`buffer_pattern`. If the `buffer_pattern` option is used, it
1710 might skew the compression ratio slightly. Setting
1711 `buffer_compress_percentage` to a value other than 100 will also
1712 enable :option:`refill_buffers` in order to reduce the likelihood that
1713 adjacent blocks are so similar that they over compress when seen
1714 together. See :option:`buffer_compress_chunk` for how to set a finer or
1715 coarser granularity for the random/fixed data region. Defaults to unset
1716 i.e., buffer data will not adhere to any compression level.
1718 .. option:: buffer_compress_chunk=int
1720 This setting allows fio to manage how big the random/fixed data region
1721 is when using :option:`buffer_compress_percentage`. When
1722 `buffer_compress_chunk` is set to some non-zero value smaller than the
1723 block size, fio can repeat the random/fixed region throughout the I/O
1724 buffer at the specified interval (which particularly useful when
1725 bigger block sizes are used for a job). When set to 0, fio will use a
1726 chunk size that matches the block size resulting in a single
1727 random/fixed region within the I/O buffer. Defaults to 512. When the
1728 unit is omitted, the value is interpreted in bytes.
1730 .. option:: buffer_pattern=str
1732 If set, fio will fill the I/O buffers with this pattern or with the contents
1733 of a file. If not set, the contents of I/O buffers are defined by the other
1734 options related to buffer contents. The setting can be any pattern of bytes,
1735 and can be prefixed with 0x for hex values. It may also be a string, where
1736 the string must then be wrapped with ``""``. Or it may also be a filename,
1737 where the filename must be wrapped with ``''`` in which case the file is
1738 opened and read. Note that not all the file contents will be read if that
1739 would cause the buffers to overflow. So, for example::
1741 buffer_pattern='filename'
1745 buffer_pattern="abcd"
1753 buffer_pattern=0xdeadface
1755 Also you can combine everything together in any order::
1757 buffer_pattern=0xdeadface"abcd"-12'filename'
1759 .. option:: dedupe_percentage=int
1761 If set, fio will generate this percentage of identical buffers when
1762 writing. These buffers will be naturally dedupable. The contents of the
1763 buffers depend on what other buffer compression settings have been set. It's
1764 possible to have the individual buffers either fully compressible, or not at
1765 all -- this option only controls the distribution of unique buffers. Setting
1766 this option will also enable :option:`refill_buffers` to prevent every buffer
1769 .. option:: dedupe_mode=str
1771 If ``dedupe_percentage=<int>`` is given, then this option controls how fio
1772 generates the dedupe buffers.
1775 Generate dedupe buffers by repeating previous writes
1777 Generate dedupe buffers from working set
1779 ``repeat`` is the default option for fio. Dedupe buffers are generated
1780 by repeating previous unique write.
1782 ``working_set`` is a more realistic workload.
1783 With ``working_set``, ``dedupe_working_set_percentage=<int>`` should be provided.
1784 Given that, fio will use the initial unique write buffers as its working set.
1785 Upon deciding to dedupe, fio will randomly choose a buffer from the working set.
1786 Note that by using ``working_set`` the dedupe percentage will converge
1787 to the desired over time while ``repeat`` maintains the desired percentage
1790 .. option:: dedupe_working_set_percentage=int
1792 If ``dedupe_mode=<str>`` is set to ``working_set``, then this controls
1793 the percentage of size of the file or device used as the buffers
1794 fio will choose to generate the dedupe buffers from
1796 Note that size needs to be explicitly provided and only 1 file per
1799 .. option:: dedupe_global=bool
1801 This controls whether the deduplication buffers will be shared amongst
1802 all jobs that have this option set. The buffers are spread evenly between
1805 .. option:: invalidate=bool
1807 Invalidate the buffer/page cache parts of the files to be used prior to
1808 starting I/O if the platform and file type support it. Defaults to true.
1809 This will be ignored if :option:`pre_read` is also specified for the
1812 .. option:: sync=str
1814 Whether, and what type, of synchronous I/O to use for writes. The allowed
1818 Do not use synchronous IO, the default.
1824 Use synchronous file IO. For the majority of I/O engines,
1825 this means using O_SYNC.
1831 Use synchronous data IO. For the majority of I/O engines,
1832 this means using O_DSYNC.
1835 .. option:: iomem=str, mem=str
1837 Fio can use various types of memory as the I/O unit buffer. The allowed
1841 Use memory from :manpage:`malloc(3)` as the buffers. Default memory
1845 Use shared memory as the buffers. Allocated through
1846 :manpage:`shmget(2)`.
1849 Same as shm, but use huge pages as backing.
1852 Use :manpage:`mmap(2)` to allocate buffers. May either be anonymous memory, or can
1853 be file backed if a filename is given after the option. The format
1854 is `mem=mmap:/path/to/file`.
1857 Use a memory mapped huge file as the buffer backing. Append filename
1858 after mmaphuge, ala `mem=mmaphuge:/hugetlbfs/file`.
1861 Same as mmap, but use a MMAP_SHARED mapping.
1864 Use GPU memory as the buffers for GPUDirect RDMA benchmark.
1865 The :option:`ioengine` must be `rdma`.
1867 The area allocated is a function of the maximum allowed bs size for the job,
1868 multiplied by the I/O depth given. Note that for **shmhuge** and
1869 **mmaphuge** to work, the system must have free huge pages allocated. This
1870 can normally be checked and set by reading/writing
1871 :file:`/proc/sys/vm/nr_hugepages` on a Linux system. Fio assumes a huge page
1872 is 2 or 4MiB in size depending on the platform. So to calculate the
1873 number of huge pages you need for a given job file, add up the I/O
1874 depth of all jobs (normally one unless :option:`iodepth` is used) and
1875 multiply by the maximum bs set. Then divide that number by the huge
1876 page size. You can see the size of the huge pages in
1877 :file:`/proc/meminfo`. If no huge pages are allocated by having a
1878 non-zero number in `nr_hugepages`, using **mmaphuge** or **shmhuge**
1879 will fail. Also see :option:`hugepage-size`.
1881 **mmaphuge** also needs to have hugetlbfs mounted and the file location
1882 should point there. So if it's mounted in :file:`/huge`, you would use
1883 `mem=mmaphuge:/huge/somefile`.
1885 .. option:: iomem_align=int, mem_align=int
1887 This indicates the memory alignment of the I/O memory buffers. Note that
1888 the given alignment is applied to the first I/O unit buffer, if using
1889 :option:`iodepth` the alignment of the following buffers are given by the
1890 :option:`bs` used. In other words, if using a :option:`bs` that is a
1891 multiple of the page sized in the system, all buffers will be aligned to
1892 this value. If using a :option:`bs` that is not page aligned, the alignment
1893 of subsequent I/O memory buffers is the sum of the :option:`iomem_align` and
1896 .. option:: hugepage-size=int
1898 Defines the size of a huge page. Must at least be equal to the system
1899 setting, see :file:`/proc/meminfo` and
1900 :file:`/sys/kernel/mm/hugepages/`. Defaults to 2 or 4MiB depending on
1901 the platform. Should probably always be a multiple of megabytes, so
1902 using ``hugepage-size=Xm`` is the preferred way to set this to avoid
1903 setting a non-pow-2 bad value.
1905 .. option:: lockmem=int
1907 Pin the specified amount of memory with :manpage:`mlock(2)`. Can be used to
1908 simulate a smaller amount of memory. The amount specified is per worker.
1914 .. option:: size=int
1916 The total size of file I/O for each thread of this job. Fio will run until
1917 this many bytes has been transferred, unless runtime is altered by other means
1918 such as (1) :option:`runtime`, (2) :option:`io_size` (3) :option:`number_ios`,
1919 (4) gaps/holes while doing I/O's such as ``rw=read:16K``, or (5) sequential
1920 I/O reaching end of the file which is possible when :option:`percentage_random`
1922 Fio will divide this size between the available files determined by options
1923 such as :option:`nrfiles`, :option:`filename`, unless :option:`filesize` is
1924 specified by the job. If the result of division happens to be 0, the size is
1925 set to the physical size of the given files or devices if they exist.
1926 If this option is not specified, fio will use the full size of the given
1927 files or devices. If the files do not exist, size must be given. It is also
1928 possible to give size as a percentage between 1 and 100. If ``size=20%`` is
1929 given, fio will use 20% of the full size of the given files or devices.
1930 In ZBD mode, value can also be set as number of zones using 'z'.
1931 Can be combined with :option:`offset` to constrain the start and end range
1932 that I/O will be done within.
1934 .. option:: io_size=int, io_limit=int
1936 Normally fio operates within the region set by :option:`size`, which means
1937 that the :option:`size` option sets both the region and size of I/O to be
1938 performed. Sometimes that is not what you want. With this option, it is
1939 possible to define just the amount of I/O that fio should do. For instance,
1940 if :option:`size` is set to 20GiB and :option:`io_size` is set to 5GiB, fio
1941 will perform I/O within the first 20GiB but exit when 5GiB have been
1942 done. The opposite is also possible -- if :option:`size` is set to 20GiB,
1943 and :option:`io_size` is set to 40GiB, then fio will do 40GiB of I/O within
1944 the 0..20GiB region.
1946 .. option:: filesize=irange(int)
1948 Individual file sizes. May be a range, in which case fio will select sizes for
1949 files at random within the given range. If not given, each created file is the
1950 same size. This option overrides :option:`size` in terms of file size, i.e. if
1951 :option:`filesize` is specified then :option:`size` becomes merely the default
1952 for :option:`io_size` and has no effect at all if :option:`io_size` is set
1955 .. option:: file_append=bool
1957 Perform I/O after the end of the file. Normally fio will operate within the
1958 size of a file. If this option is set, then fio will append to the file
1959 instead. This has identical behavior to setting :option:`offset` to the size
1960 of a file. This option is ignored on non-regular files.
1962 .. option:: fill_device=bool, fill_fs=bool
1964 Sets size to something really large and waits for ENOSPC (no space left on
1965 device) or EDQUOT (disk quota exceeded)
1966 as the terminating condition. Only makes sense with sequential
1967 write. For a read workload, the mount point will be filled first then I/O
1968 started on the result. This option doesn't make sense if operating on a raw
1969 device node, since the size of that is already known by the file system.
1970 Additionally, writing beyond end-of-device will not return ENOSPC there.
1976 .. option:: ioengine=str
1978 Defines how the job issues I/O to the file. The following types are defined:
1981 Basic :manpage:`read(2)` or :manpage:`write(2)`
1982 I/O. :manpage:`lseek(2)` is used to position the I/O location.
1983 See :option:`fsync` and :option:`fdatasync` for syncing write I/Os.
1986 Basic :manpage:`pread(2)` or :manpage:`pwrite(2)` I/O. Default on
1987 all supported operating systems except for Windows.
1990 Basic :manpage:`readv(2)` or :manpage:`writev(2)` I/O. Will emulate
1991 queuing by coalescing adjacent I/Os into a single submission.
1994 Basic :manpage:`preadv(2)` or :manpage:`pwritev(2)` I/O.
1997 Basic :manpage:`preadv2(2)` or :manpage:`pwritev2(2)` I/O.
2000 Fast Linux native asynchronous I/O. Supports async IO
2001 for both direct and buffered IO.
2002 This engine defines engine specific options.
2005 Fast Linux native asynchronous I/O for pass through commands.
2006 This engine defines engine specific options.
2009 Linux native asynchronous I/O. Note that Linux may only support
2010 queued behavior with non-buffered I/O (set ``direct=1`` or
2012 This engine defines engine specific options.
2015 POSIX asynchronous I/O using :manpage:`aio_read(3)` and
2016 :manpage:`aio_write(3)`.
2019 Solaris native asynchronous I/O.
2022 Windows native asynchronous I/O. Default on Windows.
2025 File is memory mapped with :manpage:`mmap(2)` and data copied
2026 to/from using :manpage:`memcpy(3)`.
2029 :manpage:`splice(2)` is used to transfer the data and
2030 :manpage:`vmsplice(2)` to transfer data from user space to the
2034 SCSI generic sg v3 I/O. May either be synchronous using the SG_IO
2035 ioctl, or if the target is an sg character device we use
2036 :manpage:`read(2)` and :manpage:`write(2)` for asynchronous
2037 I/O. Requires :option:`filename` option to specify either block or
2038 character devices. This engine supports trim operations.
2039 The sg engine includes engine specific options.
2042 Read, write, trim and ZBC/ZAC operations to a zoned
2043 block device using libzbc library. The target can be
2044 either an SG character device or a block device file.
2047 Doesn't transfer any data, just pretends to. This is mainly used to
2048 exercise fio itself and for debugging/testing purposes.
2051 Transfer over the network to given ``host:port``. Depending on the
2052 :option:`protocol` used, the :option:`hostname`, :option:`port`,
2053 :option:`listen` and :option:`filename` options are used to specify
2054 what sort of connection to make, while the :option:`protocol` option
2055 determines which protocol will be used. This engine defines engine
2059 Like **net**, but uses :manpage:`splice(2)` and
2060 :manpage:`vmsplice(2)` to map data and send/receive.
2061 This engine defines engine specific options.
2064 Doesn't transfer any data, but burns CPU cycles according to the
2065 :option:`cpuload`, :option:`cpuchunks` and :option:`cpumode` options.
2066 Setting :option:`cpuload`\=85 will cause that job to do nothing but burn 85%
2067 of the CPU. In case of SMP machines, use :option:`numjobs`\=<nr_of_cpu>
2068 to get desired CPU usage, as the cpuload only loads a
2069 single CPU at the desired rate. A job never finishes unless there is
2070 at least one non-cpuio job.
2071 Setting :option:`cpumode`\=qsort replace the default noop instructions loop
2072 by a qsort algorithm to consume more energy.
2075 The RDMA I/O engine supports both RDMA memory semantics
2076 (RDMA_WRITE/RDMA_READ) and channel semantics (Send/Recv) for the
2077 InfiniBand, RoCE and iWARP protocols. This engine defines engine
2081 I/O engine that does regular fallocate to simulate data transfer as
2085 does fallocate(,mode = FALLOC_FL_KEEP_SIZE,).
2088 does fallocate(,mode = 0).
2091 does fallocate(,mode = FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE).
2094 I/O engine that sends :manpage:`ftruncate(2)` operations in response
2095 to write (DDIR_WRITE) events. Each ftruncate issued sets the file's
2096 size to the current block offset. :option:`blocksize` is ignored.
2099 I/O engine that does regular EXT4_IOC_MOVE_EXT ioctls to simulate
2100 defragment activity in request to DDIR_WRITE event.
2103 I/O engine supporting direct access to Ceph Reliable Autonomic
2104 Distributed Object Store (RADOS) via librados. This ioengine
2105 defines engine specific options.
2108 I/O engine supporting direct access to Ceph Rados Block Devices
2109 (RBD) via librbd without the need to use the kernel rbd driver. This
2110 ioengine defines engine specific options.
2113 I/O engine supporting GET/PUT requests over HTTP(S) with libcurl to
2114 a WebDAV or S3 endpoint. This ioengine defines engine specific options.
2116 This engine only supports direct IO of iodepth=1; you need to scale this
2117 via numjobs. blocksize defines the size of the objects to be created.
2119 TRIM is translated to object deletion.
2122 Using GlusterFS libgfapi sync interface to direct access to
2123 GlusterFS volumes without having to go through FUSE. This ioengine
2124 defines engine specific options.
2127 Using GlusterFS libgfapi async interface to direct access to
2128 GlusterFS volumes without having to go through FUSE. This ioengine
2129 defines engine specific options.
2132 Read and write through Hadoop (HDFS). The :option:`filename` option
2133 is used to specify host,port of the hdfs name-node to connect. This
2134 engine interprets offsets a little differently. In HDFS, files once
2135 created cannot be modified so random writes are not possible. To
2136 imitate this the libhdfs engine expects a bunch of small files to be
2137 created over HDFS and will randomly pick a file from them
2138 based on the offset generated by fio backend (see the example
2139 job file to create such files, use ``rw=write`` option). Please
2140 note, it may be necessary to set environment variables to work
2141 with HDFS/libhdfs properly. Each job uses its own connection to
2145 Read, write and erase an MTD character device (e.g.,
2146 :file:`/dev/mtd0`). Discards are treated as erases. Depending on the
2147 underlying device type, the I/O may have to go in a certain pattern,
2148 e.g., on NAND, writing sequentially to erase blocks and discarding
2149 before overwriting. The `trimwrite` mode works well for this
2153 Read and write using device DAX to a persistent memory device (e.g.,
2154 /dev/dax0.0) through the PMDK libpmem library.
2157 Prefix to specify loading an external I/O engine object file. Append
2158 the engine filename, e.g. ``ioengine=external:/tmp/foo.o`` to load
2159 ioengine :file:`foo.o` in :file:`/tmp`. The path can be either
2160 absolute or relative. See :file:`engines/skeleton_external.c` for
2161 details of writing an external I/O engine.
2164 Simply create the files and do no I/O to them. You still need to
2165 set `filesize` so that all the accounting still occurs, but no
2166 actual I/O will be done other than creating the file.
2169 Simply do stat() and do no I/O to the file. You need to set 'filesize'
2170 and 'nrfiles', so that files will be created.
2171 This engine is to measure file lookup and meta data access.
2174 Simply delete the files by unlink() and do no I/O to them. You need to set 'filesize'
2175 and 'nrfiles', so that the files will be created.
2176 This engine is to measure file delete.
2179 Read and write using mmap I/O to a file on a filesystem
2180 mounted with DAX on a persistent memory device through the PMDK
2184 Synchronous read and write using DDN's Infinite Memory Engine (IME).
2185 This engine is very basic and issues calls to IME whenever an IO is
2189 Synchronous read and write using DDN's Infinite Memory Engine (IME).
2190 This engine uses iovecs and will try to stack as much IOs as possible
2191 (if the IOs are "contiguous" and the IO depth is not exceeded)
2192 before issuing a call to IME.
2195 Asynchronous read and write using DDN's Infinite Memory Engine (IME).
2196 This engine will try to stack as much IOs as possible by creating
2197 requests for IME. FIO will then decide when to commit these requests.
2200 Read and write iscsi lun with libiscsi.
2203 Read and write a Network Block Device (NBD).
2206 I/O engine supporting libcufile synchronous access to nvidia-fs and a
2207 GPUDirect Storage-supported filesystem. This engine performs
2208 I/O without transferring buffers between user-space and the kernel,
2209 unless :option:`verify` is set or :option:`cuda_io` is `posix`.
2210 :option:`iomem` must not be `cudamalloc`. This ioengine defines
2211 engine specific options.
2214 I/O engine supporting asynchronous read and write operations to the
2215 DAOS File System (DFS) via libdfs.
2218 I/O engine supporting asynchronous read and write operations to
2219 NFS filesystems from userspace via libnfs. This is useful for
2220 achieving higher concurrency and thus throughput than is possible
2224 Execute 3rd party tools. Could be used to perform monitoring during jobs runtime.
2227 I/O engine using the xNVMe C API, for NVMe devices. The xnvme engine provides
2228 flexibility to access GNU/Linux Kernel NVMe driver via libaio, IOCTLs, io_uring,
2229 the SPDK NVMe driver, or your own custom NVMe driver. The xnvme engine includes
2230 engine specific options. (See https://xnvme.io).
2233 Use the libblkio library
2234 (https://gitlab.com/libblkio/libblkio). The specific
2235 *driver* to use must be set using
2236 :option:`libblkio_driver`. If
2237 :option:`mem`/:option:`iomem` is not specified, memory
2238 allocation is delegated to libblkio (and so is
2239 guaranteed to work with the selected *driver*). One
2240 libblkio instance is used per process, so all jobs
2241 setting option :option:`thread` will share a single
2242 instance (with one queue per thread) and must specify
2243 compatible options. Note that some drivers don't allow
2244 several instances to access the same device or file
2245 simultaneously, but allow it for threads.
2247 I/O engine specific parameters
2248 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2250 In addition, there are some parameters which are only valid when a specific
2251 :option:`ioengine` is in use. These are used identically to normal parameters,
2252 with the caveat that when used on the command line, they must come after the
2253 :option:`ioengine` that defines them is selected.
2255 .. option:: cmdprio_percentage=int[,int] : [io_uring] [libaio]
2257 Set the percentage of I/O that will be issued with the highest priority.
2258 Default: 0. A single value applies to reads and writes. Comma-separated
2259 values may be specified for reads and writes. For this option to be
2260 effective, NCQ priority must be supported and enabled, and the :option:`direct`
2261 option must be set. fio must also be run as the root user. Unlike
2262 slat/clat/lat stats, which can be tracked and reported independently, per
2263 priority stats only track and report a single type of latency. By default,
2264 completion latency (clat) will be reported, if :option:`lat_percentiles` is
2265 set, total latency (lat) will be reported.
2267 .. option:: cmdprio_class=int[,int] : [io_uring] [libaio]
2269 Set the I/O priority class to use for I/Os that must be issued with
2270 a priority when :option:`cmdprio_percentage` or
2271 :option:`cmdprio_bssplit` is set. If not specified when
2272 :option:`cmdprio_percentage` or :option:`cmdprio_bssplit` is set,
2273 this defaults to the highest priority class. A single value applies
2274 to reads and writes. Comma-separated values may be specified for
2275 reads and writes. See :manpage:`ionice(1)`. See also the
2276 :option:`prioclass` option.
2278 .. option:: cmdprio=int[,int] : [io_uring] [libaio]
2280 Set the I/O priority value to use for I/Os that must be issued with
2281 a priority when :option:`cmdprio_percentage` or
2282 :option:`cmdprio_bssplit` is set. If not specified when
2283 :option:`cmdprio_percentage` or :option:`cmdprio_bssplit` is set,
2285 Linux limits us to a positive value between 0 and 7, with 0 being the
2286 highest. A single value applies to reads and writes. Comma-separated
2287 values may be specified for reads and writes. See :manpage:`ionice(1)`.
2288 Refer to an appropriate manpage for other operating systems since
2289 meaning of priority may differ. See also the :option:`prio` option.
2291 .. option:: cmdprio_bssplit=str[,str] : [io_uring] [libaio]
2293 To get a finer control over I/O priority, this option allows
2294 specifying the percentage of IOs that must have a priority set
2295 depending on the block size of the IO. This option is useful only
2296 when used together with the :option:`bssplit` option, that is,
2297 multiple different block sizes are used for reads and writes.
2299 The first accepted format for this option is the same as the format of
2300 the :option:`bssplit` option:
2302 cmdprio_bssplit=blocksize/percentage:blocksize/percentage
2304 In this case, each entry will use the priority class and priority
2305 level defined by the options :option:`cmdprio_class` and
2306 :option:`cmdprio` respectively.
2308 The second accepted format for this option is:
2310 cmdprio_bssplit=blocksize/percentage/class/level:blocksize/percentage/class/level
2312 In this case, the priority class and priority level is defined inside
2313 each entry. In comparison with the first accepted format, the second
2314 accepted format does not restrict all entries to have the same priority
2315 class and priority level.
2317 For both formats, only the read and write data directions are supported,
2318 values for trim IOs are ignored. This option is mutually exclusive with
2319 the :option:`cmdprio_percentage` option.
2321 .. option:: fixedbufs : [io_uring] [io_uring_cmd]
2323 If fio is asked to do direct IO, then Linux will map pages for each
2324 IO call, and release them when IO is done. If this option is set, the
2325 pages are pre-mapped before IO is started. This eliminates the need to
2326 map and release for each IO. This is more efficient, and reduces the
2329 .. option:: nonvectored=int : [io_uring] [io_uring_cmd]
2331 With this option, fio will use non-vectored read/write commands, where
2332 address must contain the address directly. Default is -1.
2334 .. option:: force_async=int : [io_uring] [io_uring_cmd]
2336 Normal operation for io_uring is to try and issue an sqe as
2337 non-blocking first, and if that fails, execute it in an async manner.
2338 With this option set to N, then every N request fio will ask sqe to
2339 be issued in an async manner. Default is 0.
2341 .. option:: registerfiles : [io_uring] [io_uring_cmd]
2343 With this option, fio registers the set of files being used with the
2344 kernel. This avoids the overhead of managing file counts in the kernel,
2345 making the submission and completion part more lightweight. Required
2346 for the below :option:`sqthread_poll` option.
2348 .. option:: sqthread_poll : [io_uring] [io_uring_cmd] [xnvme]
2350 Normally fio will submit IO by issuing a system call to notify the
2351 kernel of available items in the SQ ring. If this option is set, the
2352 act of submitting IO will be done by a polling thread in the kernel.
2353 This frees up cycles for fio, at the cost of using more CPU in the
2354 system. As submission is just the time it takes to fill in the sqe
2355 entries and any syscall required to wake up the idle kernel thread,
2356 fio will not report submission latencies.
2358 .. option:: sqthread_poll_cpu=int : [io_uring] [io_uring_cmd]
2360 When :option:`sqthread_poll` is set, this option provides a way to
2361 define which CPU should be used for the polling thread.
2363 .. option:: cmd_type=str : [io_uring_cmd]
2365 Specifies the type of uring passthrough command to be used. Supported
2366 value is nvme. Default is nvme.
2370 [io_uring] [io_uring_cmd] [xnvme]
2372 If this option is set, fio will attempt to use polled IO completions.
2373 Normal IO completions generate interrupts to signal the completion of
2374 IO, polled completions do not. Hence they are require active reaping
2375 by the application. The benefits are more efficient IO for high IOPS
2376 scenarios, and lower latencies for low queue depth IO.
2380 Use poll queues. This is incompatible with
2381 :option:`libblkio_wait_mode=eventfd <libblkio_wait_mode>` and
2382 :option:`libblkio_force_enable_completion_eventfd`.
2386 Set RWF_HIPRI on I/O, indicating to the kernel that it's of higher priority
2391 If this option is set, fio will attempt to use polled IO completions.
2392 This will have a similar effect as (io_uring)hipri. Only SCSI READ and
2393 WRITE commands will have the SGV4_FLAG_HIPRI set (not UNMAP (trim) nor
2394 VERIFY). Older versions of the Linux sg driver that do not support
2395 hipri will simply ignore this flag and do normal IO. The Linux SCSI
2396 Low Level Driver (LLD) that "owns" the device also needs to support
2397 hipri (also known as iopoll and mq_poll). The MegaRAID driver is an
2398 example of a SCSI LLD. Default: clear (0) which does normal
2399 (interrupted based) IO.
2401 .. option:: userspace_reap : [libaio]
2403 Normally, with the libaio engine in use, fio will use the
2404 :manpage:`io_getevents(2)` system call to reap newly returned events. With
2405 this flag turned on, the AIO ring will be read directly from user-space to
2406 reap events. The reaping mode is only enabled when polling for a minimum of
2407 0 events (e.g. when :option:`iodepth_batch_complete` `=0`).
2409 .. option:: hipri_percentage : [pvsync2]
2411 When hipri is set this determines the probability of a pvsync2 I/O being high
2412 priority. The default is 100%.
2414 .. option:: nowait=bool : [pvsync2] [libaio] [io_uring] [io_uring_cmd]
2416 By default if a request cannot be executed immediately (e.g. resource starvation,
2417 waiting on locks) it is queued and the initiating process will be blocked until
2418 the required resource becomes free.
2420 This option sets the RWF_NOWAIT flag (supported from the 4.14 Linux kernel) and
2421 the call will return instantly with EAGAIN or a partial result rather than waiting.
2423 It is useful to also use ignore_error=EAGAIN when using this option.
2425 Note: glibc 2.27, 2.28 have a bug in syscall wrappers preadv2, pwritev2.
2426 They return EOPNOTSUP instead of EAGAIN.
2428 For cached I/O, using this option usually means a request operates only with
2429 cached data. Currently the RWF_NOWAIT flag does not supported for cached write.
2431 For direct I/O, requests will only succeed if cache invalidation isn't required,
2432 file blocks are fully allocated and the disk request could be issued immediately.
2434 .. option:: fdp=bool : [io_uring_cmd]
2436 Enable Flexible Data Placement mode for write commands.
2438 .. option:: fdp_pli=str : [io_uring_cmd]
2440 Select which Placement ID Index/Indicies this job is allowed to use for
2441 writes. By default, the job will cycle through all available Placement
2442 IDs, so use this to isolate these identifiers to specific jobs. If you
2443 want fio to use placement identifier only at indices 0, 2 and 5 specify
2446 .. option:: cpuload=int : [cpuio]
2448 Attempt to use the specified percentage of CPU cycles. This is a mandatory
2449 option when using cpuio I/O engine.
2451 .. option:: cpuchunks=int : [cpuio]
2453 Split the load into cycles of the given time. In microseconds.
2455 .. option:: cpumode=str : [cpuio]
2457 Specify how to stress the CPU. It can take these two values:
2460 This is the default where the CPU executes noop instructions.
2462 Replace the default noop instructions loop with a qsort algorithm to
2463 consume more energy.
2465 .. option:: exit_on_io_done=bool : [cpuio]
2467 Detect when I/O threads are done, then exit.
2469 .. option:: namenode=str : [libhdfs]
2471 The hostname or IP address of a HDFS cluster namenode to contact.
2473 .. option:: port=int
2477 The listening port of the HFDS cluster namenode.
2481 The TCP or UDP port to bind to or connect to. If this is used with
2482 :option:`numjobs` to spawn multiple instances of the same job type, then
2483 this will be the starting port number since fio will use a range of
2488 The port to use for RDMA-CM communication. This should be the same value
2489 on the client and the server side.
2491 .. option:: hostname=str : [netsplice] [net] [rdma]
2493 The hostname or IP address to use for TCP, UDP or RDMA-CM based I/O. If the job
2494 is a TCP listener or UDP reader, the hostname is not used and must be omitted
2495 unless it is a valid UDP multicast address.
2497 .. option:: serverip=str : [librpma_*]
2499 The IP address to be used for RDMA-CM based I/O.
2501 .. option:: direct_write_to_pmem=bool : [librpma_*]
2503 Set to 1 only when Direct Write to PMem from the remote host is possible.
2504 Otherwise, set to 0.
2506 .. option:: busy_wait_polling=bool : [librpma_*_server]
2508 Set to 0 to wait for completion instead of busy-wait polling completion.
2511 .. option:: interface=str : [netsplice] [net]
2513 The IP address of the network interface used to send or receive UDP
2516 .. option:: ttl=int : [netsplice] [net]
2518 Time-to-live value for outgoing UDP multicast packets. Default: 1.
2520 .. option:: nodelay=bool : [netsplice] [net]
2522 Set TCP_NODELAY on TCP connections.
2524 .. option:: protocol=str, proto=str : [netsplice] [net]
2526 The network protocol to use. Accepted values are:
2529 Transmission control protocol.
2531 Transmission control protocol V6.
2533 User datagram protocol.
2535 User datagram protocol V6.
2539 When the protocol is TCP or UDP, the port must also be given, as well as the
2540 hostname if the job is a TCP listener or UDP reader. For unix sockets, the
2541 normal :option:`filename` option should be used and the port is invalid.
2543 .. option:: listen : [netsplice] [net]
2545 For TCP network connections, tell fio to listen for incoming connections
2546 rather than initiating an outgoing connection. The :option:`hostname` must
2547 be omitted if this option is used.
2549 .. option:: pingpong : [netsplice] [net]
2551 Normally a network writer will just continue writing data, and a network
2552 reader will just consume packages. If ``pingpong=1`` is set, a writer will
2553 send its normal payload to the reader, then wait for the reader to send the
2554 same payload back. This allows fio to measure network latencies. The
2555 submission and completion latencies then measure local time spent sending or
2556 receiving, and the completion latency measures how long it took for the
2557 other end to receive and send back. For UDP multicast traffic
2558 ``pingpong=1`` should only be set for a single reader when multiple readers
2559 are listening to the same address.
2561 .. option:: window_size : [netsplice] [net]
2563 Set the desired socket buffer size for the connection.
2565 .. option:: mss : [netsplice] [net]
2567 Set the TCP maximum segment size (TCP_MAXSEG).
2569 .. option:: donorname=str : [e4defrag]
2571 File will be used as a block donor (swap extents between files).
2573 .. option:: inplace=int : [e4defrag]
2575 Configure donor file blocks allocation strategy:
2578 Default. Preallocate donor's file on init.
2580 Allocate space immediately inside defragment event, and free right
2583 .. option:: clustername=str : [rbd,rados]
2585 Specifies the name of the Ceph cluster.
2587 .. option:: rbdname=str : [rbd]
2589 Specifies the name of the RBD.
2591 .. option:: clientname=str : [rbd,rados]
2593 Specifies the username (without the 'client.' prefix) used to access the
2594 Ceph cluster. If the *clustername* is specified, the *clientname* shall be
2595 the full *type.id* string. If no type. prefix is given, fio will add
2596 'client.' by default.
2598 .. option:: conf=str : [rados]
2600 Specifies the configuration path of ceph cluster, so conf file does not
2601 have to be /etc/ceph/ceph.conf.
2603 .. option:: busy_poll=bool : [rbd,rados]
2605 Poll store instead of waiting for completion. Usually this provides better
2606 throughput at cost of higher(up to 100%) CPU utilization.
2608 .. option:: touch_objects=bool : [rados]
2610 During initialization, touch (create if do not exist) all objects (files).
2611 Touching all objects affects ceph caches and likely impacts test results.
2614 .. option:: pool=str :
2618 Specifies the name of the Ceph pool containing RBD or RADOS data.
2622 Specify the label or UUID of the DAOS pool to connect to.
2624 .. option:: cont=str : [dfs]
2626 Specify the label or UUID of the DAOS container to open.
2628 .. option:: chunk_size=int
2632 Specify a different chunk size (in bytes) for the dfs file.
2633 Use DAOS container's chunk size by default.
2637 The size of the chunk to use for each file.
2639 .. option:: object_class=str : [dfs]
2641 Specify a different object class for the dfs file.
2642 Use DAOS container's object class by default.
2644 .. option:: skip_bad=bool : [mtd]
2646 Skip operations against known bad blocks.
2648 .. option:: hdfsdirectory : [libhdfs]
2650 libhdfs will create chunk in this HDFS directory.
2652 .. option:: verb=str : [rdma]
2654 The RDMA verb to use on this side of the RDMA ioengine connection. Valid
2655 values are write, read, send and recv. These correspond to the equivalent
2656 RDMA verbs (e.g. write = rdma_write etc.). Note that this only needs to be
2657 specified on the client side of the connection. See the examples folder.
2659 .. option:: bindname=str : [rdma]
2661 The name to use to bind the local RDMA-CM connection to a local RDMA device.
2662 This could be a hostname or an IPv4 or IPv6 address. On the server side this
2663 will be passed into the rdma_bind_addr() function and on the client site it
2664 will be used in the rdma_resolve_add() function. This can be useful when
2665 multiple paths exist between the client and the server or in certain loopback
2668 .. option:: stat_type=str : [filestat]
2670 Specify stat system call type to measure lookup/getattr performance.
2671 Default is **stat** for :manpage:`stat(2)`.
2673 .. option:: readfua=bool : [sg]
2675 With readfua option set to 1, read operations include
2676 the force unit access (fua) flag. Default is 0.
2678 .. option:: writefua=bool : [sg]
2680 With writefua option set to 1, write operations include
2681 the force unit access (fua) flag. Default is 0.
2683 .. option:: sg_write_mode=str : [sg]
2685 Specify the type of write commands to issue. This option can take three values:
2688 This is the default where write opcodes are issued as usual.
2689 **write_and_verify**
2690 Issue WRITE AND VERIFY commands. The BYTCHK bit is set to 0. This
2691 directs the device to carry out a medium verification with no data
2692 comparison. The writefua option is ignored with this selection.
2694 This option is deprecated. Use write_and_verify instead.
2696 Issue WRITE SAME commands. This transfers a single block to the device
2697 and writes this same block of data to a contiguous sequence of LBAs
2698 beginning at the specified offset. fio's block size parameter specifies
2699 the amount of data written with each command. However, the amount of data
2700 actually transferred to the device is equal to the device's block
2701 (sector) size. For a device with 512 byte sectors, blocksize=8k will
2702 write 16 sectors with each command. fio will still generate 8k of data
2703 for each command but only the first 512 bytes will be used and
2704 transferred to the device. The writefua option is ignored with this
2707 This option is deprecated. Use write_same instead.
2709 Issue WRITE SAME(16) commands as above but with the No Data Output
2710 Buffer (NDOB) bit set. No data will be transferred to the device with
2711 this bit set. Data written will be a pre-determined pattern such as
2714 Issue WRITE STREAM(16) commands. Use the **stream_id** option to specify
2715 the stream identifier.
2716 **verify_bytchk_00**
2717 Issue VERIFY commands with BYTCHK set to 00. This directs the
2718 device to carry out a medium verification with no data comparison.
2719 **verify_bytchk_01**
2720 Issue VERIFY commands with BYTCHK set to 01. This directs the device to
2721 compare the data on the device with the data transferred to the device.
2722 **verify_bytchk_11**
2723 Issue VERIFY commands with BYTCHK set to 11. This transfers a
2724 single block to the device and compares the contents of this block with the
2725 data on the device beginning at the specified offset. fio's block size
2726 parameter specifies the total amount of data compared with this command.
2727 However, only one block (sector) worth of data is transferred to the device.
2728 This is similar to the WRITE SAME command except that data is compared instead
2731 .. option:: stream_id=int : [sg]
2733 Set the stream identifier for WRITE STREAM commands. If this is set to 0 (which is not
2734 a valid stream identifier) fio will open a stream and then close it when done. Default
2737 .. option:: http_host=str : [http]
2739 Hostname to connect to. For S3, this could be the bucket hostname.
2740 Default is **localhost**
2742 .. option:: http_user=str : [http]
2744 Username for HTTP authentication.
2746 .. option:: http_pass=str : [http]
2748 Password for HTTP authentication.
2750 .. option:: https=str : [http]
2752 Enable HTTPS instead of http. *on* enables HTTPS; *insecure*
2753 will enable HTTPS, but disable SSL peer verification (use with
2754 caution!). Default is **off**
2756 .. option:: http_mode=str : [http]
2758 Which HTTP access mode to use: *webdav*, *swift*, or *s3*.
2759 Default is **webdav**
2761 .. option:: http_s3_region=str : [http]
2763 The S3 region/zone string.
2764 Default is **us-east-1**
2766 .. option:: http_s3_key=str : [http]
2770 .. option:: http_s3_keyid=str : [http]
2772 The S3 key/access id.
2774 .. option:: http_s3_sse_customer_key=str : [http]
2776 The encryption customer key in SSE server side.
2778 .. option:: http_s3_sse_customer_algorithm=str : [http]
2780 The encryption customer algorithm in SSE server side.
2781 Default is **AES256**
2783 .. option:: http_s3_storage_class=str : [http]
2785 Which storage class to access. User-customizable settings.
2786 Default is **STANDARD**
2788 .. option:: http_swift_auth_token=str : [http]
2790 The Swift auth token. See the example configuration file on how
2793 .. option:: http_verbose=int : [http]
2795 Enable verbose requests from libcurl. Useful for debugging. 1
2796 turns on verbose logging from libcurl, 2 additionally enables
2797 HTTP IO tracing. Default is **0**
2799 .. option:: uri=str : [nbd]
2801 Specify the NBD URI of the server to test. The string
2802 is a standard NBD URI
2803 (see https://github.com/NetworkBlockDevice/nbd/tree/master/doc).
2804 Example URIs: nbd://localhost:10809
2805 nbd+unix:///?socket=/tmp/socket
2806 nbds://tlshost/exportname
2808 .. option:: gpu_dev_ids=str : [libcufile]
2810 Specify the GPU IDs to use with CUDA. This is a colon-separated list of
2811 int. GPUs are assigned to workers roundrobin. Default is 0.
2813 .. option:: cuda_io=str : [libcufile]
2815 Specify the type of I/O to use with CUDA. Default is **cufile**.
2818 Use libcufile and nvidia-fs. This option performs I/O directly
2819 between a GPUDirect Storage filesystem and GPU buffers,
2820 avoiding use of a bounce buffer. If :option:`verify` is set,
2821 cudaMemcpy is used to copy verificaton data between RAM and GPU.
2822 Verification data is copied from RAM to GPU before a write
2823 and from GPU to RAM after a read. :option:`direct` must be 1.
2825 Use POSIX to perform I/O with a RAM buffer, and use cudaMemcpy
2826 to transfer data between RAM and the GPUs. Data is copied from
2827 GPU to RAM before a write and copied from RAM to GPU after a
2828 read. :option:`verify` does not affect use of cudaMemcpy.
2830 .. option:: nfs_url=str : [nfs]
2832 URL in libnfs format, eg nfs://<server|ipv4|ipv6>/path[?arg=val[&arg=val]*]
2833 Refer to the libnfs README for more details.
2835 .. option:: program=str : [exec]
2837 Specify the program to execute.
2839 .. option:: arguments=str : [exec]
2841 Specify arguments to pass to program.
2842 Some special variables can be expanded to pass fio's job details to the program.
2845 Replaced by the duration of the job in seconds.
2847 Replaced by the name of the job.
2849 .. option:: grace_time=int : [exec]
2851 Specify the time between the SIGTERM and SIGKILL signals. Default is 1 second.
2853 .. option:: std_redirect=bool : [exec]
2855 If set, stdout and stderr streams are redirected to files named from the job name. Default is true.
2857 .. option:: xnvme_async=str : [xnvme]
2859 Select the xnvme async command interface. This can take these values.
2862 This is default and use to emulate asynchronous I/O by using a
2863 single thread to create a queue pair on top of a synchronous
2864 I/O interface using the NVMe driver IOCTL.
2866 Emulate an asynchronous I/O interface with a pool of userspace
2867 threads on top of a synchronous I/O interface using the NVMe
2868 driver IOCTL. By default four threads are used.
2870 Linux native asynchronous I/O interface which supports both
2871 direct and buffered I/O.
2873 Fast Linux native asynchronous I/O interface for NVMe pass
2874 through commands. This only works with NVMe character device
2877 Use Linux aio for Asynchronous I/O.
2879 Use the posix asynchronous I/O interface to perform one or
2880 more I/O operations asynchronously.
2882 Use the user-space VFIO-based backend, implemented using
2883 libvfn instead of SPDK.
2885 Do not transfer any data; just pretend to. This is mainly used
2886 for introspective performance evaluation.
2888 .. option:: xnvme_sync=str : [xnvme]
2890 Select the xnvme synchronous command interface. This can take these values.
2893 This is default and uses Linux NVMe Driver ioctl() for
2896 This supports regular as well as vectored pread() and pwrite()
2899 This is the same as psync except that it also supports zone
2900 management commands using Linux block layer IOCTLs.
2902 .. option:: xnvme_admin=str : [xnvme]
2904 Select the xnvme admin command interface. This can take these values.
2907 This is default and uses linux NVMe Driver ioctl() for admin
2910 Use Linux Block Layer ioctl() and sysfs for admin commands.
2912 .. option:: xnvme_dev_nsid=int : [xnvme]
2914 xnvme namespace identifier for userspace NVMe driver, SPDK or vfio.
2916 .. option:: xnvme_dev_subnqn=str : [xnvme]
2918 Sets the subsystem NQN for fabrics. This is for xNVMe to utilize a
2919 fabrics target with multiple systems.
2921 .. option:: xnvme_mem=str : [xnvme]
2923 Select the xnvme memory backend. This can take these values.
2926 This is the default posix memory backend for linux NVMe driver.
2928 Use hugepages, instead of existing posix memory backend. The
2929 memory backend uses hugetlbfs. This require users to allocate
2930 hugepages, mount hugetlbfs and set an enviornment variable for
2933 Uses SPDK's memory allocator.
2935 Uses libvfn's memory allocator. This also specifies the use
2936 of libvfn backend instead of SPDK.
2938 .. option:: xnvme_iovec=int : [xnvme]
2940 If this option is set. xnvme will use vectored read/write commands.
2942 .. option:: libblkio_driver=str : [libblkio]
2944 The libblkio *driver* to use. Different drivers access devices through
2945 different underlying interfaces. Available drivers depend on the
2946 libblkio version in use and are listed at
2947 https://libblkio.gitlab.io/libblkio/blkio.html#drivers
2949 .. option:: libblkio_path=str : [libblkio]
2951 Sets the value of the driver-specific "path" property before connecting
2952 the libblkio instance, which identifies the target device or file on
2953 which to perform I/O. Its exact semantics are driver-dependent and not
2954 all drivers may support it; see
2955 https://libblkio.gitlab.io/libblkio/blkio.html#drivers
2957 .. option:: libblkio_pre_connect_props=str : [libblkio]
2959 A colon-separated list of additional libblkio properties to be set after
2960 creating but before connecting the libblkio instance. Each property must
2961 have the format ``<name>=<value>``. Colons can be escaped as ``\:``.
2962 These are set after the engine sets any other properties, so those can
2963 be overriden. Available properties depend on the libblkio version in use
2965 https://libblkio.gitlab.io/libblkio/blkio.html#properties
2967 .. option:: libblkio_num_entries=int : [libblkio]
2969 Sets the value of the driver-specific "num-entries" property before
2970 starting the libblkio instance. Its exact semantics are driver-dependent
2971 and not all drivers may support it; see
2972 https://libblkio.gitlab.io/libblkio/blkio.html#drivers
2974 .. option:: libblkio_queue_size=int : [libblkio]
2976 Sets the value of the driver-specific "queue-size" property before
2977 starting the libblkio instance. Its exact semantics are driver-dependent
2978 and not all drivers may support it; see
2979 https://libblkio.gitlab.io/libblkio/blkio.html#drivers
2981 .. option:: libblkio_pre_start_props=str : [libblkio]
2983 A colon-separated list of additional libblkio properties to be set after
2984 connecting but before starting the libblkio instance. Each property must
2985 have the format ``<name>=<value>``. Colons can be escaped as ``\:``.
2986 These are set after the engine sets any other properties, so those can
2987 be overriden. Available properties depend on the libblkio version in use
2989 https://libblkio.gitlab.io/libblkio/blkio.html#properties
2991 .. option:: libblkio_vectored : [libblkio]
2993 Submit vectored read and write requests.
2995 .. option:: libblkio_write_zeroes_on_trim : [libblkio]
2997 Submit trims as "write zeroes" requests instead of discard requests.
2999 .. option:: libblkio_wait_mode=str : [libblkio]
3001 How to wait for completions:
3004 Use a blocking call to ``blkioq_do_io()``.
3006 Use a blocking call to ``read()`` on the completion eventfd.
3008 Use a busy loop with a non-blocking call to ``blkioq_do_io()``.
3010 .. option:: libblkio_force_enable_completion_eventfd : [libblkio]
3012 Enable the queue's completion eventfd even when unused. This may impact
3013 performance. The default is to enable it only if
3014 :option:`libblkio_wait_mode=eventfd <libblkio_wait_mode>`.
3016 .. option:: no_completion_thread : [windowsaio]
3018 Avoid using a separate thread for completion polling.
3023 .. option:: iodepth=int
3025 Number of I/O units to keep in flight against the file. Note that
3026 increasing *iodepth* beyond 1 will not affect synchronous ioengines (except
3027 for small degrees when :option:`verify_async` is in use). Even async
3028 engines may impose OS restrictions causing the desired depth not to be
3029 achieved. This may happen on Linux when using libaio and not setting
3030 :option:`direct`\=1, since buffered I/O is not async on that OS. Keep an
3031 eye on the I/O depth distribution in the fio output to verify that the
3032 achieved depth is as expected. Default: 1.
3034 .. option:: iodepth_batch_submit=int, iodepth_batch=int
3036 This defines how many pieces of I/O to submit at once. It defaults to 1
3037 which means that we submit each I/O as soon as it is available, but can be
3038 raised to submit bigger batches of I/O at the time. If it is set to 0 the
3039 :option:`iodepth` value will be used.
3041 .. option:: iodepth_batch_complete_min=int, iodepth_batch_complete=int
3043 This defines how many pieces of I/O to retrieve at once. It defaults to 1
3044 which means that we'll ask for a minimum of 1 I/O in the retrieval process
3045 from the kernel. The I/O retrieval will go on until we hit the limit set by
3046 :option:`iodepth_low`. If this variable is set to 0, then fio will always
3047 check for completed events before queuing more I/O. This helps reduce I/O
3048 latency, at the cost of more retrieval system calls.
3050 .. option:: iodepth_batch_complete_max=int
3052 This defines maximum pieces of I/O to retrieve at once. This variable should
3053 be used along with :option:`iodepth_batch_complete_min`\=int variable,
3054 specifying the range of min and max amount of I/O which should be
3055 retrieved. By default it is equal to the :option:`iodepth_batch_complete_min`
3060 iodepth_batch_complete_min=1
3061 iodepth_batch_complete_max=<iodepth>
3063 which means that we will retrieve at least 1 I/O and up to the whole
3064 submitted queue depth. If none of I/O has been completed yet, we will wait.
3068 iodepth_batch_complete_min=0
3069 iodepth_batch_complete_max=<iodepth>
3071 which means that we can retrieve up to the whole submitted queue depth, but
3072 if none of I/O has been completed yet, we will NOT wait and immediately exit
3073 the system call. In this example we simply do polling.
3075 .. option:: iodepth_low=int
3077 The low water mark indicating when to start filling the queue
3078 again. Defaults to the same as :option:`iodepth`, meaning that fio will
3079 attempt to keep the queue full at all times. If :option:`iodepth` is set to
3080 e.g. 16 and *iodepth_low* is set to 4, then after fio has filled the queue of
3081 16 requests, it will let the depth drain down to 4 before starting to fill
3084 .. option:: serialize_overlap=bool
3086 Serialize in-flight I/Os that might otherwise cause or suffer from data races.
3087 When two or more I/Os are submitted simultaneously, there is no guarantee that
3088 the I/Os will be processed or completed in the submitted order. Further, if
3089 two or more of those I/Os are writes, any overlapping region between them can
3090 become indeterminate/undefined on certain storage. These issues can cause
3091 verification to fail erratically when at least one of the racing I/Os is
3092 changing data and the overlapping region has a non-zero size. Setting
3093 ``serialize_overlap`` tells fio to avoid provoking this behavior by explicitly
3094 serializing in-flight I/Os that have a non-zero overlap. Note that setting
3095 this option can reduce both performance and the :option:`iodepth` achieved.
3097 This option only applies to I/Os issued for a single job except when it is
3098 enabled along with :option:`io_submit_mode`\=offload. In offload mode, fio
3099 will check for overlap among all I/Os submitted by offload jobs with :option:`serialize_overlap`
3104 .. option:: io_submit_mode=str
3106 This option controls how fio submits the I/O to the I/O engine. The default
3107 is `inline`, which means that the fio job threads submit and reap I/O
3108 directly. If set to `offload`, the job threads will offload I/O submission
3109 to a dedicated pool of I/O threads. This requires some coordination and thus
3110 has a bit of extra overhead, especially for lower queue depth I/O where it
3111 can increase latencies. The benefit is that fio can manage submission rates
3112 independently of the device completion rates. This avoids skewed latency
3113 reporting if I/O gets backed up on the device side (the coordinated omission
3114 problem). Note that this option cannot reliably be used with async IO
3121 .. option:: thinktime=time
3123 Stall the job for the specified period of time after an I/O has completed before issuing the
3124 next. May be used to simulate processing being done by an application.
3125 When the unit is omitted, the value is interpreted in microseconds. See
3126 :option:`thinktime_blocks`, :option:`thinktime_iotime` and :option:`thinktime_spin`.
3128 .. option:: thinktime_spin=time
3130 Only valid if :option:`thinktime` is set - pretend to spend CPU time doing
3131 something with the data received, before falling back to sleeping for the
3132 rest of the period specified by :option:`thinktime`. When the unit is
3133 omitted, the value is interpreted in microseconds.
3135 .. option:: thinktime_blocks=int
3137 Only valid if :option:`thinktime` is set - control how many blocks to issue,
3138 before waiting :option:`thinktime` usecs. If not set, defaults to 1 which will make
3139 fio wait :option:`thinktime` usecs after every block. This effectively makes any
3140 queue depth setting redundant, since no more than 1 I/O will be queued
3141 before we have to complete it and do our :option:`thinktime`. In other words, this
3142 setting effectively caps the queue depth if the latter is larger.
3144 .. option:: thinktime_blocks_type=str
3146 Only valid if :option:`thinktime` is set - control how :option:`thinktime_blocks`
3147 triggers. The default is `complete`, which triggers thinktime when fio completes
3148 :option:`thinktime_blocks` blocks. If this is set to `issue`, then the trigger happens
3151 .. option:: thinktime_iotime=time
3153 Only valid if :option:`thinktime` is set - control :option:`thinktime`
3154 interval by time. The :option:`thinktime` stall is repeated after IOs
3155 are executed for :option:`thinktime_iotime`. For example,
3156 ``--thinktime_iotime=9s --thinktime=1s`` repeat 10-second cycle with IOs
3157 for 9 seconds and stall for 1 second. When the unit is omitted,
3158 :option:`thinktime_iotime` is interpreted as a number of seconds. If
3159 this option is used together with :option:`thinktime_blocks`, the
3160 :option:`thinktime` stall is repeated after :option:`thinktime_iotime`
3161 or after :option:`thinktime_blocks` IOs, whichever happens first.
3163 .. option:: rate=int[,int][,int]
3165 Cap the bandwidth used by this job. The number is in bytes/sec, the normal
3166 suffix rules apply. Comma-separated values may be specified for reads,
3167 writes, and trims as described in :option:`blocksize`.
3169 For example, using `rate=1m,500k` would limit reads to 1MiB/sec and writes to
3170 500KiB/sec. Capping only reads or writes can be done with `rate=,500k` or
3171 `rate=500k,` where the former will only limit writes (to 500KiB/sec) and the
3172 latter will only limit reads.
3174 .. option:: rate_min=int[,int][,int]
3176 Tell fio to do whatever it can to maintain at least this bandwidth. Failing
3177 to meet this requirement will cause the job to exit. Comma-separated values
3178 may be specified for reads, writes, and trims as described in
3179 :option:`blocksize`.
3181 .. option:: rate_iops=int[,int][,int]
3183 Cap the bandwidth to this number of IOPS. Basically the same as
3184 :option:`rate`, just specified independently of bandwidth. If the job is
3185 given a block size range instead of a fixed value, the smallest block size
3186 is used as the metric. Comma-separated values may be specified for reads,
3187 writes, and trims as described in :option:`blocksize`.
3189 .. option:: rate_iops_min=int[,int][,int]
3191 If fio doesn't meet this rate of I/O, it will cause the job to exit.
3192 Comma-separated values may be specified for reads, writes, and trims as
3193 described in :option:`blocksize`.
3195 .. option:: rate_process=str
3197 This option controls how fio manages rated I/O submissions. The default is
3198 `linear`, which submits I/O in a linear fashion with fixed delays between
3199 I/Os that gets adjusted based on I/O completion rates. If this is set to
3200 `poisson`, fio will submit I/O based on a more real world random request
3201 flow, known as the Poisson process
3202 (https://en.wikipedia.org/wiki/Poisson_point_process). The lambda will be
3203 10^6 / IOPS for the given workload.
3205 .. option:: rate_ignore_thinktime=bool
3207 By default, fio will attempt to catch up to the specified rate setting,
3208 if any kind of thinktime setting was used. If this option is set, then
3209 fio will ignore the thinktime and continue doing IO at the specified
3210 rate, instead of entering a catch-up mode after thinktime is done.
3212 .. option:: rate_cycle=int
3214 Average bandwidth for :option:`rate` and :option:`rate_min` over this number
3215 of milliseconds. Defaults to 1000.
3221 .. option:: latency_target=time
3223 If set, fio will attempt to find the max performance point that the given
3224 workload will run at while maintaining a latency below this target. When
3225 the unit is omitted, the value is interpreted in microseconds. See
3226 :option:`latency_window` and :option:`latency_percentile`.
3228 .. option:: latency_window=time
3230 Used with :option:`latency_target` to specify the sample window that the job
3231 is run at varying queue depths to test the performance. When the unit is
3232 omitted, the value is interpreted in microseconds.
3234 .. option:: latency_percentile=float
3236 The percentage of I/Os that must fall within the criteria specified by
3237 :option:`latency_target` and :option:`latency_window`. If not set, this
3238 defaults to 100.0, meaning that all I/Os must be equal or below to the value
3239 set by :option:`latency_target`.
3241 .. option:: latency_run=bool
3243 Used with :option:`latency_target`. If false (default), fio will find
3244 the highest queue depth that meets :option:`latency_target` and exit. If
3245 true, fio will continue running and try to meet :option:`latency_target`
3246 by adjusting queue depth.
3248 .. option:: max_latency=time[,time][,time]
3250 If set, fio will exit the job with an ETIMEDOUT error if it exceeds this
3251 maximum latency. When the unit is omitted, the value is interpreted in
3252 microseconds. Comma-separated values may be specified for reads, writes,
3253 and trims as described in :option:`blocksize`.
3259 .. option:: write_iolog=str
3261 Write the issued I/O patterns to the specified file. See
3262 :option:`read_iolog`. Specify a separate file for each job, otherwise the
3263 iologs will be interspersed and the file may be corrupt. This file will
3264 be opened in append mode.
3266 .. option:: read_iolog=str
3268 Open an iolog with the specified filename and replay the I/O patterns it
3269 contains. This can be used to store a workload and replay it sometime
3270 later. The iolog given may also be a blktrace binary file, which allows fio
3271 to replay a workload captured by :command:`blktrace`. See
3272 :manpage:`blktrace(8)` for how to capture such logging data. For blktrace
3273 replay, the file needs to be turned into a blkparse binary data file first
3274 (``blkparse <device> -o /dev/null -d file_for_fio.bin``).
3275 You can specify a number of files by separating the names with a ':'
3276 character. See the :option:`filename` option for information on how to
3277 escape ':' characters within the file names. These files will
3278 be sequentially assigned to job clones created by :option:`numjobs`.
3279 '-' is a reserved name, meaning read from stdin, notably if
3280 :option:`filename` is set to '-' which means stdin as well, then
3281 this flag can't be set to '-'.
3283 .. option:: read_iolog_chunked=bool
3285 Determines how iolog is read. If false(default) entire :option:`read_iolog`
3286 will be read at once. If selected true, input from iolog will be read
3287 gradually. Useful when iolog is very large, or it is generated.
3289 .. option:: merge_blktrace_file=str
3291 When specified, rather than replaying the logs passed to :option:`read_iolog`,
3292 the logs go through a merge phase which aggregates them into a single
3293 blktrace. The resulting file is then passed on as the :option:`read_iolog`
3294 parameter. The intention here is to make the order of events consistent.
3295 This limits the influence of the scheduler compared to replaying multiple
3296 blktraces via concurrent jobs.
3298 .. option:: merge_blktrace_scalars=float_list
3300 This is a percentage based option that is index paired with the list of
3301 files passed to :option:`read_iolog`. When merging is performed, scale
3302 the time of each event by the corresponding amount. For example,
3303 ``--merge_blktrace_scalars="50:100"`` runs the first trace in halftime
3304 and the second trace in realtime. This knob is separately tunable from
3305 :option:`replay_time_scale` which scales the trace during runtime and
3306 does not change the output of the merge unlike this option.
3308 .. option:: merge_blktrace_iters=float_list
3310 This is a whole number option that is index paired with the list of files
3311 passed to :option:`read_iolog`. When merging is performed, run each trace
3312 for the specified number of iterations. For example,
3313 ``--merge_blktrace_iters="2:1"`` runs the first trace for two iterations
3314 and the second trace for one iteration.
3316 .. option:: replay_no_stall=bool
3318 When replaying I/O with :option:`read_iolog` the default behavior is to
3319 attempt to respect the timestamps within the log and replay them with the
3320 appropriate delay between IOPS. By setting this variable fio will not
3321 respect the timestamps and attempt to replay them as fast as possible while
3322 still respecting ordering. The result is the same I/O pattern to a given
3323 device, but different timings.
3325 .. option:: replay_time_scale=int
3327 When replaying I/O with :option:`read_iolog`, fio will honor the
3328 original timing in the trace. With this option, it's possible to scale
3329 the time. It's a percentage option, if set to 50 it means run at 50%
3330 the original IO rate in the trace. If set to 200, run at twice the
3331 original IO rate. Defaults to 100.
3333 .. option:: replay_redirect=str
3335 While replaying I/O patterns using :option:`read_iolog` the default behavior
3336 is to replay the IOPS onto the major/minor device that each IOP was recorded
3337 from. This is sometimes undesirable because on a different machine those
3338 major/minor numbers can map to a different device. Changing hardware on the
3339 same system can also result in a different major/minor mapping.
3340 ``replay_redirect`` causes all I/Os to be replayed onto the single specified
3341 device regardless of the device it was recorded
3342 from. i.e. :option:`replay_redirect`\= :file:`/dev/sdc` would cause all I/O
3343 in the blktrace or iolog to be replayed onto :file:`/dev/sdc`. This means
3344 multiple devices will be replayed onto a single device, if the trace
3345 contains multiple devices. If you want multiple devices to be replayed
3346 concurrently to multiple redirected devices you must blkparse your trace
3347 into separate traces and replay them with independent fio invocations.
3348 Unfortunately this also breaks the strict time ordering between multiple
3351 .. option:: replay_align=int
3353 Force alignment of the byte offsets in a trace to this value. The value
3354 must be a power of 2.
3356 .. option:: replay_scale=int
3358 Scale byte offsets down by this factor when replaying traces. Should most
3359 likely use :option:`replay_align` as well.
3361 .. option:: replay_skip=str
3363 Sometimes it's useful to skip certain IO types in a replay trace.
3364 This could be, for instance, eliminating the writes in the trace.
3365 Or not replaying the trims/discards, if you are redirecting to
3366 a device that doesn't support them. This option takes a comma
3367 separated list of read, write, trim, sync.
3370 Threads, processes and job synchronization
3371 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3375 Fio defaults to creating jobs by using fork, however if this option is
3376 given, fio will create jobs by using POSIX Threads' function
3377 :manpage:`pthread_create(3)` to create threads instead.
3379 .. option:: wait_for=str
3381 If set, the current job won't be started until all workers of the specified
3382 waitee job are done.
3384 ``wait_for`` operates on the job name basis, so there are a few
3385 limitations. First, the waitee must be defined prior to the waiter job
3386 (meaning no forward references). Second, if a job is being referenced as a
3387 waitee, it must have a unique name (no duplicate waitees).
3389 .. option:: nice=int
3391 Run the job with the given nice value. See man :manpage:`nice(2)`.
3393 On Windows, values less than -15 set the process class to "High"; -1 through
3394 -15 set "Above Normal"; 1 through 15 "Below Normal"; and above 15 "Idle"
3397 .. option:: prio=int
3399 Set the I/O priority value of this job. Linux limits us to a positive value
3400 between 0 and 7, with 0 being the highest. See man
3401 :manpage:`ionice(1)`. Refer to an appropriate manpage for other operating
3402 systems since meaning of priority may differ. For per-command priority
3403 setting, see I/O engine specific :option:`cmdprio_percentage` and
3404 :option:`cmdprio` options.
3406 .. option:: prioclass=int
3408 Set the I/O priority class. See man :manpage:`ionice(1)`. For per-command
3409 priority setting, see I/O engine specific :option:`cmdprio_percentage`
3410 and :option:`cmdprio_class` options.
3412 .. option:: cpus_allowed=str
3414 Controls the same options as :option:`cpumask`, but accepts a textual
3415 specification of the permitted CPUs instead and CPUs are indexed from 0. So
3416 to use CPUs 0 and 5 you would specify ``cpus_allowed=0,5``. This option also
3417 allows a range of CPUs to be specified -- say you wanted a binding to CPUs
3418 0, 5, and 8 to 15, you would set ``cpus_allowed=0,5,8-15``.
3420 On Windows, when ``cpus_allowed`` is unset only CPUs from fio's current
3421 processor group will be used and affinity settings are inherited from the
3422 system. An fio build configured to target Windows 7 makes options that set
3423 CPUs processor group aware and values will set both the processor group
3424 and a CPU from within that group. For example, on a system where processor
3425 group 0 has 40 CPUs and processor group 1 has 32 CPUs, ``cpus_allowed``
3426 values between 0 and 39 will bind CPUs from processor group 0 and
3427 ``cpus_allowed`` values between 40 and 71 will bind CPUs from processor
3428 group 1. When using ``cpus_allowed_policy=shared`` all CPUs specified by a
3429 single ``cpus_allowed`` option must be from the same processor group. For
3430 Windows fio builds not built for Windows 7, CPUs will only be selected from
3431 (and be relative to) whatever processor group fio happens to be running in
3432 and CPUs from other processor groups cannot be used.
3434 .. option:: cpus_allowed_policy=str
3436 Set the policy of how fio distributes the CPUs specified by
3437 :option:`cpus_allowed` or :option:`cpumask`. Two policies are supported:
3440 All jobs will share the CPU set specified.
3442 Each job will get a unique CPU from the CPU set.
3444 **shared** is the default behavior, if the option isn't specified. If
3445 **split** is specified, then fio will assign one cpu per job. If not
3446 enough CPUs are given for the jobs listed, then fio will roundrobin the CPUs
3449 .. option:: cpumask=int
3451 Set the CPU affinity of this job. The parameter given is a bit mask of
3452 allowed CPUs the job may run on. So if you want the allowed CPUs to be 1
3453 and 5, you would pass the decimal value of (1 << 1 | 1 << 5), or 34. See man
3454 :manpage:`sched_setaffinity(2)`. This may not work on all supported
3455 operating systems or kernel versions. This option doesn't work well for a
3456 higher CPU count than what you can store in an integer mask, so it can only
3457 control cpus 1-32. For boxes with larger CPU counts, use
3458 :option:`cpus_allowed`.
3460 .. option:: numa_cpu_nodes=str
3462 Set this job running on specified NUMA nodes' CPUs. The arguments allow
3463 comma delimited list of cpu numbers, A-B ranges, or `all`. Note, to enable
3464 NUMA options support, fio must be built on a system with libnuma-dev(el)
3467 .. option:: numa_mem_policy=str
3469 Set this job's memory policy and corresponding NUMA nodes. Format of the
3474 ``mode`` is one of the following memory policies: ``default``, ``prefer``,
3475 ``bind``, ``interleave`` or ``local``. For ``default`` and ``local`` memory
3476 policies, no node needs to be specified. For ``prefer``, only one node is
3477 allowed. For ``bind`` and ``interleave`` the ``nodelist`` may be as
3478 follows: a comma delimited list of numbers, A-B ranges, or `all`.
3480 .. option:: cgroup=str
3482 Add job to this control group. If it doesn't exist, it will be created. The
3483 system must have a mounted cgroup blkio mount point for this to work. If
3484 your system doesn't have it mounted, you can do so with::
3486 # mount -t cgroup -o blkio none /cgroup
3488 .. option:: cgroup_weight=int
3490 Set the weight of the cgroup to this value. See the documentation that comes
3491 with the kernel, allowed values are in the range of 100..1000.
3493 .. option:: cgroup_nodelete=bool
3495 Normally fio will delete the cgroups it has created after the job
3496 completion. To override this behavior and to leave cgroups around after the
3497 job completion, set ``cgroup_nodelete=1``. This can be useful if one wants
3498 to inspect various cgroup files after job completion. Default: false.
3500 .. option:: flow_id=int
3502 The ID of the flow. If not specified, it defaults to being a global
3503 flow. See :option:`flow`.
3505 .. option:: flow=int
3507 Weight in token-based flow control. If this value is used, then fio
3508 regulates the activity between two or more jobs sharing the same
3509 flow_id. Fio attempts to keep each job activity proportional to other
3510 jobs' activities in the same flow_id group, with respect to requested
3511 weight per job. That is, if one job has `flow=3', another job has
3512 `flow=2' and another with `flow=1`, then there will be a roughly 3:2:1
3513 ratio in how much one runs vs the others.
3515 .. option:: flow_sleep=int
3517 The period of time, in microseconds, to wait after the flow counter
3518 has exceeded its proportion before retrying operations.
3520 .. option:: stonewall, wait_for_previous
3522 Wait for preceding jobs in the job file to exit, before starting this
3523 one. Can be used to insert serialization points in the job file. A stone
3524 wall also implies starting a new reporting group, see
3525 :option:`group_reporting`.
3529 By default, fio will continue running all other jobs when one job finishes.
3530 Sometimes this is not the desired action. Setting ``exitall`` will instead
3531 make fio terminate all jobs in the same group, as soon as one job of that
3534 .. option:: exit_what=str
3536 By default, fio will continue running all other jobs when one job finishes.
3537 Sometimes this is not the desired action. Setting ``exitall`` will
3538 instead make fio terminate all jobs in the same group. The option
3539 ``exit_what`` allows to control which jobs get terminated when ``exitall`` is
3540 enabled. The default is ``group`` and does not change the behaviour of
3541 ``exitall``. The setting ``all`` terminates all jobs. The setting ``stonewall``
3542 terminates all currently running jobs across all groups and continues execution
3543 with the next stonewalled group.
3545 .. option:: exec_prerun=str
3547 Before running this job, issue the command specified through
3548 :manpage:`system(3)`. Output is redirected in a file called
3549 :file:`jobname.prerun.txt`.
3551 .. option:: exec_postrun=str
3553 After the job completes, issue the command specified though
3554 :manpage:`system(3)`. Output is redirected in a file called
3555 :file:`jobname.postrun.txt`.
3559 Instead of running as the invoking user, set the user ID to this value
3560 before the thread/process does any work.
3564 Set group ID, see :option:`uid`.
3570 .. option:: verify_only
3572 Do not perform specified workload, only verify data still matches previous
3573 invocation of this workload. This option allows one to check data multiple
3574 times at a later date without overwriting it. This option makes sense only
3575 for workloads that write data, and does not support workloads with the
3576 :option:`time_based` option set.
3578 .. option:: do_verify=bool
3580 Run the verify phase after a write phase. Only valid if :option:`verify` is
3583 .. option:: verify=str
3585 If writing to a file, fio can verify the file contents after each iteration
3586 of the job. Each verification method also implies verification of special
3587 header, which is written to the beginning of each block. This header also
3588 includes meta information, like offset of the block, block number, timestamp
3589 when block was written, etc. :option:`verify` can be combined with
3590 :option:`verify_pattern` option. The allowed values are:
3593 Use an md5 sum of the data area and store it in the header of
3597 Use an experimental crc64 sum of the data area and store it in the
3598 header of each block.
3601 Use a crc32c sum of the data area and store it in the header of
3602 each block. This will automatically use hardware acceleration
3603 (e.g. SSE4.2 on an x86 or CRC crypto extensions on ARM64) but will
3604 fall back to software crc32c if none is found. Generally the
3605 fastest checksum fio supports when hardware accelerated.
3611 Use a crc32 sum of the data area and store it in the header of each
3615 Use a crc16 sum of the data area and store it in the header of each
3619 Use a crc7 sum of the data area and store it in the header of each
3623 Use xxhash as the checksum function. Generally the fastest software
3624 checksum that fio supports.
3627 Use sha512 as the checksum function.
3630 Use sha256 as the checksum function.
3633 Use optimized sha1 as the checksum function.
3636 Use optimized sha3-224 as the checksum function.
3639 Use optimized sha3-256 as the checksum function.
3642 Use optimized sha3-384 as the checksum function.
3645 Use optimized sha3-512 as the checksum function.
3648 This option is deprecated, since now meta information is included in
3649 generic verification header and meta verification happens by
3650 default. For detailed information see the description of the
3651 :option:`verify` setting. This option is kept because of
3652 compatibility's sake with old configurations. Do not use it.
3655 Verify a strict pattern. Normally fio includes a header with some
3656 basic information and checksumming, but if this option is set, only
3657 the specific pattern set with :option:`verify_pattern` is verified.
3660 Only pretend to verify. Useful for testing internals with
3661 :option:`ioengine`\=null, not for much else.
3663 This option can be used for repeated burn-in tests of a system to make sure
3664 that the written data is also correctly read back. If the data direction
3665 given is a read or random read, fio will assume that it should verify a
3666 previously written file. If the data direction includes any form of write,
3667 the verify will be of the newly written data.
3669 To avoid false verification errors, do not use the norandommap option when
3670 verifying data with async I/O engines and I/O depths > 1. Or use the
3671 norandommap and the lfsr random generator together to avoid writing to the
3672 same offset with multiple outstanding I/Os.
3674 .. option:: verify_offset=int
3676 Swap the verification header with data somewhere else in the block before
3677 writing. It is swapped back before verifying.
3679 .. option:: verify_interval=int
3681 Write the verification header at a finer granularity than the
3682 :option:`blocksize`. It will be written for chunks the size of
3683 ``verify_interval``. :option:`blocksize` should divide this evenly.
3685 .. option:: verify_pattern=str
3687 If set, fio will fill the I/O buffers with this pattern. Fio defaults to
3688 filling with totally random bytes, but sometimes it's interesting to fill
3689 with a known pattern for I/O verification purposes. Depending on the width
3690 of the pattern, fio will fill 1/2/3/4 bytes of the buffer at the time (it can
3691 be either a decimal or a hex number). The ``verify_pattern`` if larger than
3692 a 32-bit quantity has to be a hex number that starts with either "0x" or
3693 "0X". Use with :option:`verify`. Also, ``verify_pattern`` supports %o
3694 format, which means that for each block offset will be written and then
3695 verified back, e.g.::
3699 Or use combination of everything::
3701 verify_pattern=0xff%o"abcd"-12
3703 .. option:: verify_fatal=bool
3705 Normally fio will keep checking the entire contents before quitting on a
3706 block verification failure. If this option is set, fio will exit the job on
3707 the first observed failure. Default: false.
3709 .. option:: verify_dump=bool
3711 If set, dump the contents of both the original data block and the data block
3712 we read off disk to files. This allows later analysis to inspect just what
3713 kind of data corruption occurred. Off by default.
3715 .. option:: verify_async=int
3717 Fio will normally verify I/O inline from the submitting thread. This option
3718 takes an integer describing how many async offload threads to create for I/O
3719 verification instead, causing fio to offload the duty of verifying I/O
3720 contents to one or more separate threads. If using this offload option, even
3721 sync I/O engines can benefit from using an :option:`iodepth` setting higher
3722 than 1, as it allows them to have I/O in flight while verifies are running.
3723 Defaults to 0 async threads, i.e. verification is not asynchronous.
3725 .. option:: verify_async_cpus=str
3727 Tell fio to set the given CPU affinity on the async I/O verification
3728 threads. See :option:`cpus_allowed` for the format used.
3730 .. option:: verify_backlog=int
3732 Fio will normally verify the written contents of a job that utilizes verify
3733 once that job has completed. In other words, everything is written then
3734 everything is read back and verified. You may want to verify continually
3735 instead for a variety of reasons. Fio stores the meta data associated with
3736 an I/O block in memory, so for large verify workloads, quite a bit of memory
3737 would be used up holding this meta data. If this option is enabled, fio will
3738 write only N blocks before verifying these blocks.
3740 .. option:: verify_backlog_batch=int
3742 Control how many blocks fio will verify if :option:`verify_backlog` is
3743 set. If not set, will default to the value of :option:`verify_backlog`
3744 (meaning the entire queue is read back and verified). If
3745 ``verify_backlog_batch`` is less than :option:`verify_backlog` then not all
3746 blocks will be verified, if ``verify_backlog_batch`` is larger than
3747 :option:`verify_backlog`, some blocks will be verified more than once.
3749 .. option:: verify_state_save=bool
3751 When a job exits during the write phase of a verify workload, save its
3752 current state. This allows fio to replay up until that point, if the verify
3753 state is loaded for the verify read phase. The format of the filename is,
3756 <type>-<jobname>-<jobindex>-verify.state.
3758 <type> is "local" for a local run, "sock" for a client/server socket
3759 connection, and "ip" (192.168.0.1, for instance) for a networked
3760 client/server connection. Defaults to true.
3762 .. option:: verify_state_load=bool
3764 If a verify termination trigger was used, fio stores the current write state
3765 of each thread. This can be used at verification time so that fio knows how
3766 far it should verify. Without this information, fio will run a full
3767 verification pass, according to the settings in the job file used. Default
3770 .. option:: experimental_verify=bool
3772 Enable experimental verification. Standard verify records I/O metadata
3773 for later use during the verification phase. Experimental verify
3774 instead resets the file after the write phase and then replays I/Os for
3775 the verification phase.
3777 .. option:: trim_percentage=int
3779 Number of verify blocks to discard/trim.
3781 .. option:: trim_verify_zero=bool
3783 Verify that trim/discarded blocks are returned as zeros.
3785 .. option:: trim_backlog=int
3787 Trim after this number of blocks are written.
3789 .. option:: trim_backlog_batch=int
3791 Trim this number of I/O blocks.
3796 .. option:: steadystate=str:float, ss=str:float
3798 Define the criterion and limit for assessing steady state performance. The
3799 first parameter designates the criterion whereas the second parameter sets
3800 the threshold. When the criterion falls below the threshold for the
3801 specified duration, the job will stop. For example, `iops_slope:0.1%` will
3802 direct fio to terminate the job when the least squares regression slope
3803 falls below 0.1% of the mean IOPS. If :option:`group_reporting` is enabled
3804 this will apply to all jobs in the group. Below is the list of available
3805 steady state assessment criteria. All assessments are carried out using only
3806 data from the rolling collection window. Threshold limits can be expressed
3807 as a fixed value or as a percentage of the mean in the collection window.
3809 When using this feature, most jobs should include the :option:`time_based`
3810 and :option:`runtime` options or the :option:`loops` option so that fio does not
3811 stop running after it has covered the full size of the specified file(s) or device(s).
3814 Collect IOPS data. Stop the job if all individual IOPS measurements
3815 are within the specified limit of the mean IOPS (e.g., ``iops:2``
3816 means that all individual IOPS values must be within 2 of the mean,
3817 whereas ``iops:0.2%`` means that all individual IOPS values must be
3818 within 0.2% of the mean IOPS to terminate the job).
3821 Collect IOPS data and calculate the least squares regression
3822 slope. Stop the job if the slope falls below the specified limit.
3825 Collect bandwidth data. Stop the job if all individual bandwidth
3826 measurements are within the specified limit of the mean bandwidth.
3829 Collect bandwidth data and calculate the least squares regression
3830 slope. Stop the job if the slope falls below the specified limit.
3832 .. option:: steadystate_duration=time, ss_dur=time
3834 A rolling window of this duration will be used to judge whether steady
3835 state has been reached. Data will be collected every
3836 :option:`ss_interval`. The default is 0 which disables steady state
3837 detection. When the unit is omitted, the value is interpreted in
3840 .. option:: steadystate_ramp_time=time, ss_ramp=time
3842 Allow the job to run for the specified duration before beginning data
3843 collection for checking the steady state job termination criterion. The
3844 default is 0. When the unit is omitted, the value is interpreted in seconds.
3846 .. option:: steadystate_check_interval=time, ss_interval=time
3848 The values during the rolling window will be collected with a period of
3849 this value. If :option:`ss_interval` is 30s and :option:`ss_dur` is
3850 300s, 10 measurements will be taken. Default is 1s but that might not
3851 converge, especially for slower devices, so set this accordingly. When
3852 the unit is omitted, the value is interpreted in seconds.
3855 Measurements and reporting
3856 ~~~~~~~~~~~~~~~~~~~~~~~~~~
3858 .. option:: per_job_logs=bool
3860 If set, this generates bw/clat/iops log with per file private filenames. If
3861 not set, jobs with identical names will share the log filename. Default:
3864 .. option:: group_reporting
3866 It may sometimes be interesting to display statistics for groups of jobs as
3867 a whole instead of for each individual job. This is especially true if
3868 :option:`numjobs` is used; looking at individual thread/process output
3869 quickly becomes unwieldy. To see the final report per-group instead of
3870 per-job, use :option:`group_reporting`. Jobs in a file will be part of the
3871 same reporting group, unless if separated by a :option:`stonewall`, or by
3872 using :option:`new_group`.
3874 .. option:: new_group
3876 Start a new reporting group. See: :option:`group_reporting`. If not given,
3877 all jobs in a file will be part of the same reporting group, unless
3878 separated by a :option:`stonewall`.
3880 .. option:: stats=bool
3882 By default, fio collects and shows final output results for all jobs
3883 that run. If this option is set to 0, then fio will ignore it in
3884 the final stat output.
3886 .. option:: write_bw_log=str
3888 If given, write a bandwidth log for this job. Can be used to store data of
3889 the bandwidth of the jobs in their lifetime.
3891 If no str argument is given, the default filename of
3892 :file:`jobname_type.x.log` is used. Even when the argument is given, fio
3893 will still append the type of log. So if one specifies::
3897 The actual log name will be :file:`foo_bw.x.log` where `x` is the index
3898 of the job (`1..N`, where `N` is the number of jobs). If
3899 :option:`per_job_logs` is false, then the filename will not include the
3902 The included :command:`fio_generate_plots` script uses :command:`gnuplot` to turn these
3903 text files into nice graphs. See `Log File Formats`_ for how data is
3904 structured within the file.
3906 .. option:: write_lat_log=str
3908 Same as :option:`write_bw_log`, except this option creates I/O
3909 submission (e.g., :file:`name_slat.x.log`), completion (e.g.,
3910 :file:`name_clat.x.log`), and total (e.g., :file:`name_lat.x.log`)
3911 latency files instead. See :option:`write_bw_log` for details about
3912 the filename format and `Log File Formats`_ for how data is structured
3915 .. option:: write_hist_log=str
3917 Same as :option:`write_bw_log` but writes an I/O completion latency
3918 histogram file (e.g., :file:`name_hist.x.log`) instead. Note that this
3919 file will be empty unless :option:`log_hist_msec` has also been set.
3920 See :option:`write_bw_log` for details about the filename format and
3921 `Log File Formats`_ for how data is structured within the file.
3923 .. option:: write_iops_log=str
3925 Same as :option:`write_bw_log`, but writes an IOPS file (e.g.
3926 :file:`name_iops.x.log`) instead. Because fio defaults to individual
3927 I/O logging, the value entry in the IOPS log will be 1 unless windowed
3928 logging (see :option:`log_avg_msec`) has been enabled. See
3929 :option:`write_bw_log` for details about the filename format and `Log
3930 File Formats`_ for how data is structured within the file.
3932 .. option:: log_entries=int
3934 By default, fio will log an entry in the iops, latency, or bw log for
3935 every I/O that completes. The initial number of I/O log entries is 1024.
3936 When the log entries are all used, new log entries are dynamically
3937 allocated. This dynamic log entry allocation may negatively impact
3938 time-related statistics such as I/O tail latencies (e.g. 99.9th percentile
3939 completion latency). This option allows specifying a larger initial
3940 number of log entries to avoid run-time allocations of new log entries,
3941 resulting in more precise time-related I/O statistics.
3942 Also see :option:`log_avg_msec`. Defaults to 1024.
3944 .. option:: log_avg_msec=int
3946 By default, fio will log an entry in the iops, latency, or bw log for every
3947 I/O that completes. When writing to the disk log, that can quickly grow to a
3948 very large size. Setting this option makes fio average the each log entry
3949 over the specified period of time, reducing the resolution of the log. See
3950 :option:`log_max_value` as well. Defaults to 0, logging all entries.
3951 Also see `Log File Formats`_.
3953 .. option:: log_hist_msec=int
3955 Same as :option:`log_avg_msec`, but logs entries for completion latency
3956 histograms. Computing latency percentiles from averages of intervals using
3957 :option:`log_avg_msec` is inaccurate. Setting this option makes fio log
3958 histogram entries over the specified period of time, reducing log sizes for
3959 high IOPS devices while retaining percentile accuracy. See
3960 :option:`log_hist_coarseness` and :option:`write_hist_log` as well.
3961 Defaults to 0, meaning histogram logging is disabled.
3963 .. option:: log_hist_coarseness=int
3965 Integer ranging from 0 to 6, defining the coarseness of the resolution of
3966 the histogram logs enabled with :option:`log_hist_msec`. For each increment
3967 in coarseness, fio outputs half as many bins. Defaults to 0, for which
3968 histogram logs contain 1216 latency bins. See :option:`write_hist_log`
3969 and `Log File Formats`_.
3971 .. option:: log_max_value=bool
3973 If :option:`log_avg_msec` is set, fio logs the average over that window. If
3974 you instead want to log the maximum value, set this option to 1. Defaults to
3975 0, meaning that averaged values are logged.
3977 .. option:: log_offset=bool
3979 If this is set, the iolog options will include the byte offset for the I/O
3980 entry as well as the other data values. Defaults to 0 meaning that
3981 offsets are not present in logs. Also see `Log File Formats`_.
3983 .. option:: log_compression=int
3985 If this is set, fio will compress the I/O logs as it goes, to keep the
3986 memory footprint lower. When a log reaches the specified size, that chunk is
3987 removed and compressed in the background. Given that I/O logs are fairly
3988 highly compressible, this yields a nice memory savings for longer runs. The
3989 downside is that the compression will consume some background CPU cycles, so
3990 it may impact the run. This, however, is also true if the logging ends up
3991 consuming most of the system memory. So pick your poison. The I/O logs are
3992 saved normally at the end of a run, by decompressing the chunks and storing
3993 them in the specified log file. This feature depends on the availability of
3996 .. option:: log_compression_cpus=str
3998 Define the set of CPUs that are allowed to handle online log compression for
3999 the I/O jobs. This can provide better isolation between performance
4000 sensitive jobs, and background compression work. See
4001 :option:`cpus_allowed` for the format used.
4003 .. option:: log_store_compressed=bool
4005 If set, fio will store the log files in a compressed format. They can be
4006 decompressed with fio, using the :option:`--inflate-log` command line
4007 parameter. The files will be stored with a :file:`.fz` suffix.
4009 .. option:: log_unix_epoch=bool
4011 If set, fio will log Unix timestamps to the log files produced by enabling
4012 write_type_log for each log type, instead of the default zero-based
4015 .. option:: log_alternate_epoch=bool
4017 If set, fio will log timestamps based on the epoch used by the clock specified
4018 in the log_alternate_epoch_clock_id option, to the log files produced by
4019 enabling write_type_log for each log type, instead of the default zero-based
4022 .. option:: log_alternate_epoch_clock_id=int
4024 Specifies the clock_id to be used by clock_gettime to obtain the alternate epoch
4025 if either log_unix_epoch or log_alternate_epoch are true. Otherwise has no
4026 effect. Default value is 0, or CLOCK_REALTIME.
4028 .. option:: block_error_percentiles=bool
4030 If set, record errors in trim block-sized units from writes and trims and
4031 output a histogram of how many trims it took to get to errors, and what kind
4032 of error was encountered.
4034 .. option:: bwavgtime=int
4036 Average the calculated bandwidth over the given time. Value is specified in
4037 milliseconds. If the job also does bandwidth logging through
4038 :option:`write_bw_log`, then the minimum of this option and
4039 :option:`log_avg_msec` will be used. Default: 500ms.
4041 .. option:: iopsavgtime=int
4043 Average the calculated IOPS over the given time. Value is specified in
4044 milliseconds. If the job also does IOPS logging through
4045 :option:`write_iops_log`, then the minimum of this option and
4046 :option:`log_avg_msec` will be used. Default: 500ms.
4048 .. option:: disk_util=bool
4050 Generate disk utilization statistics, if the platform supports it.
4053 .. option:: disable_lat=bool
4055 Disable measurements of total latency numbers. Useful only for cutting back
4056 the number of calls to :manpage:`gettimeofday(2)`, as that does impact
4057 performance at really high IOPS rates. Note that to really get rid of a
4058 large amount of these calls, this option must be used with
4059 :option:`disable_slat` and :option:`disable_bw_measurement` as well.
4061 .. option:: disable_clat=bool
4063 Disable measurements of completion latency numbers. See
4064 :option:`disable_lat`.
4066 .. option:: disable_slat=bool
4068 Disable measurements of submission latency numbers. See
4069 :option:`disable_lat`.
4071 .. option:: disable_bw_measurement=bool, disable_bw=bool
4073 Disable measurements of throughput/bandwidth numbers. See
4074 :option:`disable_lat`.
4076 .. option:: slat_percentiles=bool
4078 Report submission latency percentiles. Submission latency is not recorded
4079 for synchronous ioengines.
4081 .. option:: clat_percentiles=bool
4083 Report completion latency percentiles.
4085 .. option:: lat_percentiles=bool
4087 Report total latency percentiles. Total latency is the sum of submission
4088 latency and completion latency.
4090 .. option:: percentile_list=float_list
4092 Overwrite the default list of percentiles for latencies and the block error
4093 histogram. Each number is a floating point number in the range (0,100], and
4094 the maximum length of the list is 20. Use ``:`` to separate the numbers. For
4095 example, ``--percentile_list=99.5:99.9`` will cause fio to report the
4096 latency durations below which 99.5% and 99.9% of the observed latencies fell,
4099 .. option:: significant_figures=int
4101 If using :option:`--output-format` of `normal`, set the significant
4102 figures to this value. Higher values will yield more precise IOPS and
4103 throughput units, while lower values will round. Requires a minimum
4104 value of 1 and a maximum value of 10. Defaults to 4.
4110 .. option:: exitall_on_error
4112 When one job finishes in error, terminate the rest. The default is to wait
4113 for each job to finish.
4115 .. option:: continue_on_error=str
4117 Normally fio will exit the job on the first observed failure. If this option
4118 is set, fio will continue the job when there is a 'non-fatal error' (EIO or
4119 EILSEQ) until the runtime is exceeded or the I/O size specified is
4120 completed. If this option is used, there are two more stats that are
4121 appended, the total error count and the first error. The error field given
4122 in the stats is the first error that was hit during the run.
4124 Note: a write error from the device may go unnoticed by fio when using
4125 buffered IO, as the write() (or similar) system call merely dirties the
4126 kernel pages, unless :option:`sync` or :option:`direct` is used. Device IO
4127 errors occur when the dirty data is actually written out to disk. If fully
4128 sync writes aren't desirable, :option:`fsync` or :option:`fdatasync` can be
4129 used as well. This is specific to writes, as reads are always synchronous.
4131 The allowed values are:
4134 Exit on any I/O or verify errors.
4137 Continue on read errors, exit on all others.
4140 Continue on write errors, exit on all others.
4143 Continue on any I/O error, exit on all others.
4146 Continue on verify errors, exit on all others.
4149 Continue on all errors.
4152 Backward-compatible alias for 'none'.
4155 Backward-compatible alias for 'all'.
4157 .. option:: ignore_error=str
4159 Sometimes you want to ignore some errors during test in that case you can
4160 specify error list for each error type, instead of only being able to
4161 ignore the default 'non-fatal error' using :option:`continue_on_error`.
4162 ``ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST`` errors for
4163 given error type is separated with ':'. Error may be symbol ('ENOSPC',
4164 'ENOMEM') or integer. Example::
4166 ignore_error=EAGAIN,ENOSPC:122
4168 This option will ignore EAGAIN from READ, and ENOSPC and 122(EDQUOT) from
4169 WRITE. This option works by overriding :option:`continue_on_error` with
4170 the list of errors for each error type if any.
4172 .. option:: error_dump=bool
4174 If set dump every error even if it is non fatal, true by default. If
4175 disabled only fatal error will be dumped.
4177 Running predefined workloads
4178 ----------------------------
4180 Fio includes predefined profiles that mimic the I/O workloads generated by
4183 .. option:: profile=str
4185 The predefined workload to run. Current profiles are:
4188 Threaded I/O bench (tiotest/tiobench) like workload.
4191 Aerospike Certification Tool (ACT) like workload.
4193 To view a profile's additional options use :option:`--cmdhelp` after specifying
4194 the profile. For example::
4196 $ fio --profile=act --cmdhelp
4201 .. option:: device-names=str
4206 .. option:: load=int
4209 ACT load multiplier. Default: 1.
4211 .. option:: test-duration=time
4214 How long the entire test takes to run. When the unit is omitted, the value
4215 is given in seconds. Default: 24h.
4217 .. option:: threads-per-queue=int
4220 Number of read I/O threads per device. Default: 8.
4222 .. option:: read-req-num-512-blocks=int
4225 Number of 512B blocks to read at the time. Default: 3.
4227 .. option:: large-block-op-kbytes=int
4230 Size of large block ops in KiB (writes). Default: 131072.
4235 Set to run ACT prep phase.
4237 Tiobench profile options
4238 ~~~~~~~~~~~~~~~~~~~~~~~~
4240 .. option:: size=str
4245 .. option:: block=int
4248 Block size in bytes. Default: 4096.
4250 .. option:: numruns=int
4260 .. option:: threads=int
4265 Interpreting the output
4266 -----------------------
4269 Example output was based on the following:
4270 TZ=UTC fio --iodepth=8 --ioengine=null --size=100M --time_based \
4271 --rate=1256k --bs=14K --name=quick --runtime=1s --name=mixed \
4272 --runtime=2m --rw=rw
4274 Fio spits out a lot of output. While running, fio will display the status of the
4275 jobs created. An example of that would be::
4277 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]
4279 The characters inside the first set of square brackets denote the current status of
4280 each thread. The first character is the first job defined in the job file, and so
4281 forth. The possible values (in typical life cycle order) are:
4283 +------+-----+-----------------------------------------------------------+
4285 +======+=====+===========================================================+
4286 | P | | Thread setup, but not started. |
4287 +------+-----+-----------------------------------------------------------+
4288 | C | | Thread created. |
4289 +------+-----+-----------------------------------------------------------+
4290 | I | | Thread initialized, waiting or generating necessary data. |
4291 +------+-----+-----------------------------------------------------------+
4292 | | p | Thread running pre-reading file(s). |
4293 +------+-----+-----------------------------------------------------------+
4294 | | / | Thread is in ramp period. |
4295 +------+-----+-----------------------------------------------------------+
4296 | | R | Running, doing sequential reads. |
4297 +------+-----+-----------------------------------------------------------+
4298 | | r | Running, doing random reads. |
4299 +------+-----+-----------------------------------------------------------+
4300 | | W | Running, doing sequential writes. |
4301 +------+-----+-----------------------------------------------------------+
4302 | | w | Running, doing random writes. |
4303 +------+-----+-----------------------------------------------------------+
4304 | | M | Running, doing mixed sequential reads/writes. |
4305 +------+-----+-----------------------------------------------------------+
4306 | | m | Running, doing mixed random reads/writes. |
4307 +------+-----+-----------------------------------------------------------+
4308 | | D | Running, doing sequential trims. |
4309 +------+-----+-----------------------------------------------------------+
4310 | | d | Running, doing random trims. |
4311 +------+-----+-----------------------------------------------------------+
4312 | | F | Running, currently waiting for :manpage:`fsync(2)`. |
4313 +------+-----+-----------------------------------------------------------+
4314 | | V | Running, doing verification of written data. |
4315 +------+-----+-----------------------------------------------------------+
4316 | f | | Thread finishing. |
4317 +------+-----+-----------------------------------------------------------+
4318 | E | | Thread exited, not reaped by main thread yet. |
4319 +------+-----+-----------------------------------------------------------+
4320 | _ | | Thread reaped. |
4321 +------+-----+-----------------------------------------------------------+
4322 | X | | Thread reaped, exited with an error. |
4323 +------+-----+-----------------------------------------------------------+
4324 | K | | Thread reaped, exited due to signal. |
4325 +------+-----+-----------------------------------------------------------+
4328 Example output was based on the following:
4329 TZ=UTC fio --iodepth=8 --ioengine=null --size=100M --runtime=58m \
4330 --time_based --rate=2512k --bs=256K --numjobs=10 \
4331 --name=readers --rw=read --name=writers --rw=write
4333 Fio will condense the thread string as not to take up more space on the command
4334 line than needed. For instance, if you have 10 readers and 10 writers running,
4335 the output would look like this::
4337 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]
4339 Note that the status string is displayed in order, so it's possible to tell which of
4340 the jobs are currently doing what. In the example above this means that jobs 1--10
4341 are readers and 11--20 are writers.
4343 The other values are fairly self explanatory -- number of threads currently
4344 running and doing I/O, the number of currently open files (f=), the estimated
4345 completion percentage, the rate of I/O since last check (read speed listed first,
4346 then write speed and optionally trim speed) in terms of bandwidth and IOPS,
4347 and time to completion for the current running group. It's impossible to estimate
4348 runtime of the following groups (if any).
4351 Example output was based on the following:
4352 TZ=UTC fio --iodepth=16 --ioengine=posixaio --filename=/tmp/fiofile \
4353 --direct=1 --size=100M --time_based --runtime=50s --rate_iops=89 \
4354 --bs=7K --name=Client1 --rw=write
4356 When fio is done (or interrupted by :kbd:`Ctrl-C`), it will show the data for
4357 each thread, group of threads, and disks in that order. For each overall thread (or
4358 group) the output looks like::
4360 Client1: (groupid=0, jobs=1): err= 0: pid=16109: Sat Jun 24 12:07:54 2017
4361 write: IOPS=88, BW=623KiB/s (638kB/s)(30.4MiB/50032msec)
4362 slat (nsec): min=500, max=145500, avg=8318.00, stdev=4781.50
4363 clat (usec): min=170, max=78367, avg=4019.02, stdev=8293.31
4364 lat (usec): min=174, max=78375, avg=4027.34, stdev=8291.79
4365 clat percentiles (usec):
4366 | 1.00th=[ 302], 5.00th=[ 326], 10.00th=[ 343], 20.00th=[ 363],
4367 | 30.00th=[ 392], 40.00th=[ 404], 50.00th=[ 416], 60.00th=[ 445],
4368 | 70.00th=[ 816], 80.00th=[ 6718], 90.00th=[12911], 95.00th=[21627],
4369 | 99.00th=[43779], 99.50th=[51643], 99.90th=[68682], 99.95th=[72877],
4371 bw ( KiB/s): min= 532, max= 686, per=0.10%, avg=622.87, stdev=24.82, samples= 100
4372 iops : min= 76, max= 98, avg=88.98, stdev= 3.54, samples= 100
4373 lat (usec) : 250=0.04%, 500=64.11%, 750=4.81%, 1000=2.79%
4374 lat (msec) : 2=4.16%, 4=1.84%, 10=4.90%, 20=11.33%, 50=5.37%
4375 lat (msec) : 100=0.65%
4376 cpu : usr=0.27%, sys=0.18%, ctx=12072, majf=0, minf=21
4377 IO depths : 1=85.0%, 2=13.1%, 4=1.8%, 8=0.1%, 16=0.0%, 32=0.0%, >=64=0.0%
4378 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
4379 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
4380 issued rwt: total=0,4450,0, short=0,0,0, dropped=0,0,0
4381 latency : target=0, window=0, percentile=100.00%, depth=8
4383 The job name (or first job's name when using :option:`group_reporting`) is printed,
4384 along with the group id, count of jobs being aggregated, last error id seen (which
4385 is 0 when there are no errors), pid/tid of that thread and the time the job/group
4386 completed. Below are the I/O statistics for each data direction performed (showing
4387 writes in the example above). In the order listed, they denote:
4390 The string before the colon shows the I/O direction the statistics
4391 are for. **IOPS** is the average I/Os performed per second. **BW**
4392 is the average bandwidth rate shown as: value in power of 2 format
4393 (value in power of 10 format). The last two values show: (**total
4394 I/O performed** in power of 2 format / **runtime** of that thread).
4397 Submission latency (**min** being the minimum, **max** being the
4398 maximum, **avg** being the average, **stdev** being the standard
4399 deviation). This is the time from when fio initialized the I/O
4400 to submission. For synchronous ioengines this includes the time
4401 up until just before the ioengine's queue function is called.
4402 For asynchronous ioengines this includes the time up through the
4403 completion of the ioengine's queue function (and commit function
4404 if it is defined). For sync I/O this row is not displayed as the
4405 slat is negligible. This value can be in nanoseconds,
4406 microseconds or milliseconds --- fio will choose the most
4407 appropriate base and print that (in the example above
4408 nanoseconds was the best scale). Note: in :option:`--minimal`
4409 mode latencies are always expressed in microseconds.
4412 Completion latency. Same names as slat, this denotes the time from
4413 submission to completion of the I/O pieces. For sync I/O, this
4414 represents the time from when the I/O was submitted to the
4415 operating system to when it was completed. For asynchronous
4416 ioengines this is the time from when the ioengine's queue (and
4417 commit if available) functions were completed to when the I/O's
4418 completion was reaped by fio.
4421 Total latency. Same names as slat and clat, this denotes the time from
4422 when fio created the I/O unit to completion of the I/O operation.
4423 It is the sum of submission and completion latency.
4426 Bandwidth statistics based on measurements from discrete
4427 intervals. Fio continuously monitors bytes transferred and I/O
4428 operations completed. By default fio calculates bandwidth in
4429 each half-second interval (see :option:`bwavgtime`) and reports
4430 descriptive statistics for the measurements here. Same names as
4431 the xlat stats, but also includes the number of samples taken
4432 (**samples**) and an approximate percentage of total aggregate
4433 bandwidth this thread received in its group (**per**). This
4434 last value is only really useful if the threads in this group
4435 are on the same disk, since they are then competing for disk
4439 IOPS statistics based on measurements from discrete intervals.
4440 For details see the description for bw above. See
4441 :option:`iopsavgtime` to control the duration of the intervals.
4442 Same values reported here as for bw except for percentage.
4444 **lat (nsec/usec/msec)**
4445 The distribution of I/O completion latencies. This is the time from when
4446 I/O leaves fio and when it gets completed. Unlike the separate
4447 read/write/trim sections above, the data here and in the remaining
4448 sections apply to all I/Os for the reporting group. 250=0.04% means that
4449 0.04% of the I/Os completed in under 250us. 500=64.11% means that 64.11%
4450 of the I/Os required 250 to 499us for completion.
4453 CPU usage. User and system time, along with the number of context
4454 switches this thread went through, usage of system and user time, and
4455 finally the number of major and minor page faults. The CPU utilization
4456 numbers are averages for the jobs in that reporting group, while the
4457 context and fault counters are summed.
4460 The distribution of I/O depths over the job lifetime. The numbers are
4461 divided into powers of 2 and each entry covers depths from that value
4462 up to those that are lower than the next entry -- e.g., 16= covers
4463 depths from 16 to 31. Note that the range covered by a depth
4464 distribution entry can be different to the range covered by the
4465 equivalent submit/complete distribution entry.
4468 How many pieces of I/O were submitting in a single submit call. Each
4469 entry denotes that amount and below, until the previous entry -- e.g.,
4470 16=100% means that we submitted anywhere between 9 to 16 I/Os per submit
4471 call. Note that the range covered by a submit distribution entry can
4472 be different to the range covered by the equivalent depth distribution
4476 Like the above submit number, but for completions instead.
4479 The number of read/write/trim requests issued, and how many of them were
4483 These values are for :option:`latency_target` and related options. When
4484 these options are engaged, this section describes the I/O depth required
4485 to meet the specified latency target.
4488 Example output was based on the following:
4489 TZ=UTC fio --ioengine=null --iodepth=2 --size=100M --numjobs=2 \
4490 --rate_process=poisson --io_limit=32M --name=read --bs=128k \
4491 --rate=11M --name=write --rw=write --bs=2k --rate=700k
4493 After each client has been listed, the group statistics are printed. They
4494 will look like this::
4496 Run status group 0 (all jobs):
4497 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
4498 WRITE: bw=1231KiB/s (1261kB/s), 616KiB/s-621KiB/s (630kB/s-636kB/s), io=64.0MiB (67.1MB), run=52747-53223msec
4500 For each data direction it prints:
4503 Aggregate bandwidth of threads in this group followed by the
4504 minimum and maximum bandwidth of all the threads in this group.
4505 Values outside of brackets are power-of-2 format and those
4506 within are the equivalent value in a power-of-10 format.
4508 Aggregate I/O performed of all threads in this group. The
4509 format is the same as bw.
4511 The smallest and longest runtimes of the threads in this group.
4513 And finally, the disk statistics are printed. This is Linux specific. They will look like this::
4515 Disk stats (read/write):
4516 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
4518 Each value is printed for both reads and writes, with reads first. The
4522 Number of I/Os performed by all groups.
4524 Number of merges performed by the I/O scheduler.
4526 Number of ticks we kept the disk busy.
4528 Total time spent in the disk queue.
4530 The disk utilization. A value of 100% means we kept the disk
4531 busy constantly, 50% would be a disk idling half of the time.
4533 It is also possible to get fio to dump the current output while it is running,
4534 without terminating the job. To do that, send fio the **USR1** signal. You can
4535 also get regularly timed dumps by using the :option:`--status-interval`
4536 parameter, or by creating a file in :file:`/tmp` named
4537 :file:`fio-dump-status`. If fio sees this file, it will unlink it and dump the
4538 current output status.
4544 For scripted usage where you typically want to generate tables or graphs of the
4545 results, fio can output the results in a semicolon separated format. The format
4546 is one long line of values, such as::
4548 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%
4549 A description of this job goes here.
4551 The job description (if provided) follows on a second line for terse v2.
4552 It appears on the same line for other terse versions.
4554 To enable terse output, use the :option:`--minimal` or
4555 :option:`--output-format`\=terse command line options. The
4556 first value is the version of the terse output format. If the output has to be
4557 changed for some reason, this number will be incremented by 1 to signify that
4560 Split up, the format is as follows (comments in brackets denote when a
4561 field was introduced or whether it's specific to some terse version):
4565 terse version, fio version [v3], jobname, groupid, error
4569 Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec)
4570 Submission latency: min, max, mean, stdev (usec)
4571 Completion latency: min, max, mean, stdev (usec)
4572 Completion latency percentiles: 20 fields (see below)
4573 Total latency: min, max, mean, stdev (usec)
4574 Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev, number of samples [v5]
4575 IOPS [v5]: min, max, mean, stdev, number of samples
4581 Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec)
4582 Submission latency: min, max, mean, stdev (usec)
4583 Completion latency: min, max, mean, stdev (usec)
4584 Completion latency percentiles: 20 fields (see below)
4585 Total latency: min, max, mean, stdev (usec)
4586 Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev, number of samples [v5]
4587 IOPS [v5]: min, max, mean, stdev, number of samples
4589 TRIM status [all but version 3]:
4591 Fields are similar to READ/WRITE status.
4595 user, system, context switches, major faults, minor faults
4599 <=1, 2, 4, 8, 16, 32, >=64
4601 I/O latencies microseconds::
4603 <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
4605 I/O latencies milliseconds::
4607 <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
4609 Disk utilization [v3]::
4611 disk name, read ios, write ios, read merges, write merges, read ticks, write ticks,
4612 time spent in queue, disk utilization percentage
4614 Additional Info (dependent on continue_on_error, default off)::
4616 total # errors, first error code
4618 Additional Info (dependent on description being set)::
4622 Completion latency percentiles can be a grouping of up to 20 sets, so for the
4623 terse output fio writes all of them. Each field will look like this::
4627 which is the Xth percentile, and the `usec` latency associated with it.
4629 For `Disk utilization`, all disks used by fio are shown. So for each disk there
4630 will be a disk utilization section.
4632 Below is a single line containing short names for each of the fields in the
4633 minimal output v3, separated by semicolons::
4635 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
4637 In client/server mode terse output differs from what appears when jobs are run
4638 locally. Disk utilization data is omitted from the standard terse output and
4639 for v3 and later appears on its own separate line at the end of each terse
4646 The `json` output format is intended to be both human readable and convenient
4647 for automated parsing. For the most part its sections mirror those of the
4648 `normal` output. The `runtime` value is reported in msec and the `bw` value is
4649 reported in 1024 bytes per second units.
4655 The `json+` output format is identical to the `json` output format except that it
4656 adds a full dump of the completion latency bins. Each `bins` object contains a
4657 set of (key, value) pairs where keys are latency durations and values count how
4658 many I/Os had completion latencies of the corresponding duration. For example,
4661 "bins" : { "87552" : 1, "89600" : 1, "94720" : 1, "96768" : 1, "97792" : 1, "99840" : 1, "100864" : 2, "103936" : 6, "104960" : 534, "105984" : 5995, "107008" : 7529, ... }
4663 This data indicates that one I/O required 87,552ns to complete, two I/Os required
4664 100,864ns to complete, and 7529 I/Os required 107,008ns to complete.
4666 Also included with fio is a Python script `fio_jsonplus_clat2csv` that takes
4667 json+ output and generates CSV-formatted latency data suitable for plotting.
4669 The latency durations actually represent the midpoints of latency intervals.
4670 For details refer to :file:`stat.h`.
4676 There are two trace file format that you can encounter. The older (v1) format is
4677 unsupported since version 1.20-rc3 (March 2008). It will still be described
4678 below in case that you get an old trace and want to understand it.
4680 In any case the trace is a simple text file with a single action per line.
4683 Trace file format v1
4684 ~~~~~~~~~~~~~~~~~~~~
4686 Each line represents a single I/O action in the following format::
4690 where `rw=0/1` for read/write, and the `offset` and `length` entries being in bytes.
4692 This format is not supported in fio versions >= 1.20-rc3.
4695 Trace file format v2
4696 ~~~~~~~~~~~~~~~~~~~~
4698 The second version of the trace file format was added in fio version 1.17. It
4699 allows one to access more than one file per trace and has a bigger set of possible
4702 The first line of the trace file has to be::
4706 Following this can be lines in two different formats, which are described below.
4708 The file management format::
4712 The `filename` is given as an absolute path. The `action` can be one of these:
4715 Add the given `filename` to the trace.
4717 Open the file with the given `filename`. The `filename` has to have
4718 been added with the **add** action before.
4720 Close the file with the given `filename`. The file has to have been
4724 The file I/O action format::
4726 filename action offset length
4728 The `filename` is given as an absolute path, and has to have been added and
4729 opened before it can be used with this format. The `offset` and `length` are
4730 given in bytes. The `action` can be one of these:
4733 Wait for `offset` microseconds. Everything below 100 is discarded.
4734 The time is relative to the previous `wait` statement. Note that
4735 action `wait` is not allowed as of version 3, as the same behavior
4736 can be achieved using timestamps.
4738 Read `length` bytes beginning from `offset`.
4740 Write `length` bytes beginning from `offset`.
4742 :manpage:`fsync(2)` the file.
4744 :manpage:`fdatasync(2)` the file.
4746 Trim the given file from the given `offset` for `length` bytes.
4749 Trace file format v3
4750 ~~~~~~~~~~~~~~~~~~~~
4752 The third version of the trace file format was added in fio version 3.31. It
4753 forces each action to have a timestamp associated with it.
4755 The first line of the trace file has to be::
4759 Following this can be lines in two different formats, which are described below.
4761 The file management format::
4763 timestamp filename action
4765 The file I/O action format::
4767 timestamp filename action offset length
4769 The `timestamp` is relative to the beginning of the run (ie starts at 0). The
4770 `filename`, `action`, `offset` and `length` are identical to version 2, except
4771 that version 3 does not allow the `wait` action.
4774 I/O Replay - Merging Traces
4775 ---------------------------
4777 Colocation is a common practice used to get the most out of a machine.
4778 Knowing which workloads play nicely with each other and which ones don't is
4779 a much harder task. While fio can replay workloads concurrently via multiple
4780 jobs, it leaves some variability up to the scheduler making results harder to
4781 reproduce. Merging is a way to make the order of events consistent.
4783 Merging is integrated into I/O replay and done when a
4784 :option:`merge_blktrace_file` is specified. The list of files passed to
4785 :option:`read_iolog` go through the merge process and output a single file
4786 stored to the specified file. The output file is passed on as if it were the
4787 only file passed to :option:`read_iolog`. An example would look like::
4789 $ fio --read_iolog="<file1>:<file2>" --merge_blktrace_file="<output_file>"
4791 Creating only the merged file can be done by passing the command line argument
4792 :option:`--merge-blktrace-only`.
4794 Scaling traces can be done to see the relative impact of any particular trace
4795 being slowed down or sped up. :option:`merge_blktrace_scalars` takes in a colon
4796 separated list of percentage scalars. It is index paired with the files passed
4797 to :option:`read_iolog`.
4799 With scaling, it may be desirable to match the running time of all traces.
4800 This can be done with :option:`merge_blktrace_iters`. It is index paired with
4801 :option:`read_iolog` just like :option:`merge_blktrace_scalars`.
4803 In an example, given two traces, A and B, each 60s long. If we want to see
4804 the impact of trace A issuing IOs twice as fast and repeat trace A over the
4805 runtime of trace B, the following can be done::
4807 $ fio --read_iolog="<trace_a>:"<trace_b>" --merge_blktrace_file"<output_file>" --merge_blktrace_scalars="50:100" --merge_blktrace_iters="2:1"
4809 This runs trace A at 2x the speed twice for approximately the same runtime as
4810 a single run of trace B.
4813 CPU idleness profiling
4814 ----------------------
4816 In some cases, we want to understand CPU overhead in a test. For example, we
4817 test patches for the specific goodness of whether they reduce CPU usage.
4818 Fio implements a balloon approach to create a thread per CPU that runs at idle
4819 priority, meaning that it only runs when nobody else needs the cpu.
4820 By measuring the amount of work completed by the thread, idleness of each CPU
4821 can be derived accordingly.
4823 An unit work is defined as touching a full page of unsigned characters. Mean and
4824 standard deviation of time to complete an unit work is reported in "unit work"
4825 section. Options can be chosen to report detailed percpu idleness or overall
4826 system idleness by aggregating percpu stats.
4829 Verification and triggers
4830 -------------------------
4832 Fio is usually run in one of two ways, when data verification is done. The first
4833 is a normal write job of some sort with verify enabled. When the write phase has
4834 completed, fio switches to reads and verifies everything it wrote. The second
4835 model is running just the write phase, and then later on running the same job
4836 (but with reads instead of writes) to repeat the same I/O patterns and verify
4837 the contents. Both of these methods depend on the write phase being completed,
4838 as fio otherwise has no idea how much data was written.
4840 With verification triggers, fio supports dumping the current write state to
4841 local files. Then a subsequent read verify workload can load this state and know
4842 exactly where to stop. This is useful for testing cases where power is cut to a
4843 server in a managed fashion, for instance.
4845 A verification trigger consists of two things:
4847 1) Storing the write state of each job.
4848 2) Executing a trigger command.
4850 The write state is relatively small, on the order of hundreds of bytes to single
4851 kilobytes. It contains information on the number of completions done, the last X
4854 A trigger is invoked either through creation ('touch') of a specified file in
4855 the system, or through a timeout setting. If fio is run with
4856 :option:`--trigger-file`\= :file:`/tmp/trigger-file`, then it will continually
4857 check for the existence of :file:`/tmp/trigger-file`. When it sees this file, it
4858 will fire off the trigger (thus saving state, and executing the trigger
4861 For client/server runs, there's both a local and remote trigger. If fio is
4862 running as a server backend, it will send the job states back to the client for
4863 safe storage, then execute the remote trigger, if specified. If a local trigger
4864 is specified, the server will still send back the write state, but the client
4865 will then execute the trigger.
4867 Verification trigger example
4868 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4870 Let's say we want to run a powercut test on the remote Linux machine 'server'.
4871 Our write workload is in :file:`write-test.fio`. We want to cut power to 'server' at
4872 some point during the run, and we'll run this test from the safety or our local
4873 machine, 'localbox'. On the server, we'll start the fio backend normally::
4875 server# fio --server
4877 and on the client, we'll fire off the workload::
4879 localbox$ fio --client=server --trigger-file=/tmp/my-trigger --trigger-remote="bash -c \"echo b > /proc/sysrq-triger\""
4881 We set :file:`/tmp/my-trigger` as the trigger file, and we tell fio to execute::
4883 echo b > /proc/sysrq-trigger
4885 on the server once it has received the trigger and sent us the write state. This
4886 will work, but it's not **really** cutting power to the server, it's merely
4887 abruptly rebooting it. If we have a remote way of cutting power to the server
4888 through IPMI or similar, we could do that through a local trigger command
4889 instead. Let's assume we have a script that does IPMI reboot of a given hostname,
4890 ipmi-reboot. On localbox, we could then have run fio with a local trigger
4893 localbox$ fio --client=server --trigger-file=/tmp/my-trigger --trigger="ipmi-reboot server"
4895 For this case, fio would wait for the server to send us the write state, then
4896 execute ``ipmi-reboot server`` when that happened.
4898 Loading verify state
4899 ~~~~~~~~~~~~~~~~~~~~
4901 To load stored write state, a read verification job file must contain the
4902 :option:`verify_state_load` option. If that is set, fio will load the previously
4903 stored state. For a local fio run this is done by loading the files directly,
4904 and on a client/server run, the server backend will ask the client to send the
4905 files over and load them from there.
4911 Fio supports a variety of log file formats, for logging latencies, bandwidth,
4912 and IOPS. The logs share a common format, which looks like this:
4914 *time* (`msec`), *value*, *data direction*, *block size* (`bytes`),
4915 *offset* (`bytes`), *command priority*
4917 *Time* for the log entry is always in milliseconds. The *value* logged depends
4918 on the type of log, it will be one of the following:
4921 Value is latency in nsecs
4927 *Data direction* is one of the following:
4936 The entry's *block size* is always in bytes. The *offset* is the position in bytes
4937 from the start of the file for that particular I/O. The logging of the offset can be
4938 toggled with :option:`log_offset`.
4940 *Command priority* is 0 for normal priority and 1 for high priority. This is controlled
4941 by the ioengine specific :option:`cmdprio_percentage`.
4943 Fio defaults to logging every individual I/O but when windowed logging is set
4944 through :option:`log_avg_msec`, either the average (by default) or the maximum
4945 (:option:`log_max_value` is set) *value* seen over the specified period of time
4946 is recorded. Each *data direction* seen within the window period will aggregate
4947 its values in a separate row. Further, when using windowed logging the *block
4948 size* and *offset* entries will always contain 0.
4954 Normally fio is invoked as a stand-alone application on the machine where the
4955 I/O workload should be generated. However, the backend and frontend of fio can
4956 be run separately i.e., the fio server can generate an I/O workload on the "Device
4957 Under Test" while being controlled by a client on another machine.
4959 Start the server on the machine which has access to the storage DUT::
4963 where `args` defines what fio listens to. The arguments are of the form
4964 ``type,hostname`` or ``IP,port``. *type* is either ``ip`` (or ip4) for TCP/IP
4965 v4, ``ip6`` for TCP/IP v6, or ``sock`` for a local unix domain socket.
4966 *hostname* is either a hostname or IP address, and *port* is the port to listen
4967 to (only valid for TCP/IP, not a local socket). Some examples:
4971 Start a fio server, listening on all interfaces on the default port (8765).
4973 2) ``fio --server=ip:hostname,4444``
4975 Start a fio server, listening on IP belonging to hostname and on port 4444.
4977 3) ``fio --server=ip6:::1,4444``
4979 Start a fio server, listening on IPv6 localhost ::1 and on port 4444.
4981 4) ``fio --server=,4444``
4983 Start a fio server, listening on all interfaces on port 4444.
4985 5) ``fio --server=1.2.3.4``
4987 Start a fio server, listening on IP 1.2.3.4 on the default port.
4989 6) ``fio --server=sock:/tmp/fio.sock``
4991 Start a fio server, listening on the local socket :file:`/tmp/fio.sock`.
4993 Once a server is running, a "client" can connect to the fio server with::
4995 fio <local-args> --client=<server> <remote-args> <job file(s)>
4997 where `local-args` are arguments for the client where it is running, `server`
4998 is the connect string, and `remote-args` and `job file(s)` are sent to the
4999 server. The `server` string follows the same format as it does on the server
5000 side, to allow IP/hostname/socket and port strings.
5002 Fio can connect to multiple servers this way::
5004 fio --client=<server1> <job file(s)> --client=<server2> <job file(s)>
5006 If the job file is located on the fio server, then you can tell the server to
5007 load a local file as well. This is done by using :option:`--remote-config` ::
5009 fio --client=server --remote-config /path/to/file.fio
5011 Then fio will open this local (to the server) job file instead of being passed
5012 one from the client.
5014 If you have many servers (example: 100 VMs/containers), you can input a pathname
5015 of a file containing host IPs/names as the parameter value for the
5016 :option:`--client` option. For example, here is an example :file:`host.list`
5017 file containing 2 hostnames::
5019 host1.your.dns.domain
5020 host2.your.dns.domain
5022 The fio command would then be::
5024 fio --client=host.list <job file(s)>
5026 In this mode, you cannot input server-specific parameters or job files -- all
5027 servers receive the same job file.
5029 In order to let ``fio --client`` runs use a shared filesystem from multiple
5030 hosts, ``fio --client`` now prepends the IP address of the server to the
5031 filename. For example, if fio is using the directory :file:`/mnt/nfs/fio` and is
5032 writing filename :file:`fileio.tmp`, with a :option:`--client` `hostfile`
5033 containing two hostnames ``h1`` and ``h2`` with IP addresses 192.168.10.120 and
5034 192.168.10.121, then fio will create two files::
5036 /mnt/nfs/fio/192.168.10.120.fileio.tmp
5037 /mnt/nfs/fio/192.168.10.121.fileio.tmp
5039 Terse output in client/server mode will differ slightly from what is produced
5040 when fio is run in stand-alone mode. See the terse output section for details.