4 The first step in getting fio to simulate a desired I/O workload, is writing a
5 job file describing that specific setup. A job file may contain any number of
6 threads and/or files -- the typical contents of the job file is a *global*
7 section defining shared parameters, and one or more job sections describing the
8 jobs involved. When run, fio parses this file and sets everything up as
9 described. If we break down a job from top to bottom, it contains the following
14 Defines the I/O pattern issued to the file(s). We may only be reading
15 sequentially from this file(s), or we may be writing randomly. Or even
16 mixing reads and writes, sequentially or randomly.
17 Should we be doing buffered I/O, or direct/raw I/O?
21 In how large chunks are we issuing I/O? This may be a single value,
22 or it may describe a range of block sizes.
26 How much data are we going to be reading/writing.
30 How do we issue I/O? We could be memory mapping the file, we could be
31 using regular read/write, we could be using splice, async I/O, or even
36 If the I/O engine is async, how large a queuing depth do we want to
42 How many files are we spreading the workload over.
44 `Threads, processes and job synchronization`_
46 How many threads or processes should we spread this workload over.
48 The above are the basic parameters defined for a workload, in addition there's a
49 multitude of parameters that modify other aspects of how this job behaves.
55 .. option:: --debug=type
57 Enable verbose tracing `type` of various fio actions. May be ``all`` for all types
58 or individual types separated by a comma (e.g. ``--debug=file,mem`` will
59 enable file and memory debugging). Currently, additional logging is
63 Dump info related to processes.
65 Dump info related to file actions.
67 Dump info related to I/O queuing.
69 Dump info related to memory allocations.
71 Dump info related to blktrace setup.
73 Dump info related to I/O verification.
75 Enable all debug options.
77 Dump info related to random offset generation.
79 Dump info related to option matching and parsing.
81 Dump info related to disk utilization updates.
83 Dump info only related to job number x.
85 Dump info only related to mutex up/down ops.
87 Dump info related to profile extensions.
89 Dump info related to internal time keeping.
91 Dump info related to networking connections.
93 Dump info related to I/O rate switching.
95 Dump info related to log compress/decompress.
97 Dump info related to steadystate detection.
99 Dump info related to the helper thread.
101 Dump info related to support for zoned block devices.
103 Show available debug options.
105 .. option:: --parse-only
107 Parse options only, don't start any I/O.
109 .. option:: --merge-blktrace-only
111 Merge blktraces only, don't start any I/O.
113 .. option:: --output=filename
115 Write output to file `filename`.
117 .. option:: --output-format=format
119 Set the reporting `format` to `normal`, `terse`, `json`, or `json+`. Multiple
120 formats can be selected, separated by a comma. `terse` is a CSV based
121 format. `json+` is like `json`, except it adds a full dump of the latency
124 .. option:: --bandwidth-log
126 Generate aggregate bandwidth logs.
128 .. option:: --minimal
130 Print statistics in a terse, semicolon-delimited format.
132 .. option:: --append-terse
134 Print statistics in selected mode AND terse, semicolon-delimited format.
135 **Deprecated**, use :option:`--output-format` instead to select multiple
138 .. option:: --terse-version=version
140 Set terse `version` output format (default 3, or 2 or 4 or 5).
142 .. option:: --version
144 Print version information and exit.
148 Print a summary of the command line options and exit.
150 .. option:: --cpuclock-test
152 Perform test and validation of internal CPU clock.
154 .. option:: --crctest=[test]
156 Test the speed of the built-in checksumming functions. If no argument is
157 given, all of them are tested. Alternatively, a comma separated list can
158 be passed, in which case the given ones are tested.
160 .. option:: --cmdhelp=command
162 Print help information for `command`. May be ``all`` for all commands.
164 .. option:: --enghelp=[ioengine[,command]]
166 List all commands defined by `ioengine`, or print help for `command`
167 defined by `ioengine`. If no `ioengine` is given, list all
170 .. option:: --showcmd=jobfile
172 Convert `jobfile` to a set of command-line options.
174 .. option:: --readonly
176 Turn on safety read-only checks, preventing writes and trims. The
177 ``--readonly`` option is an extra safety guard to prevent users from
178 accidentally starting a write or trim workload when that is not desired.
179 Fio will only modify the device under test if
180 `rw=write/randwrite/rw/randrw/trim/randtrim/trimwrite` is given. This
181 safety net can be used as an extra precaution.
183 .. option:: --eta=when
185 Specifies when real-time ETA estimate should be printed. `when` may be
186 `always`, `never` or `auto`. `auto` is the default, it prints ETA
187 when requested if the output is a TTY. `always` disregards the output
188 type, and prints ETA when requested. `never` never prints ETA.
190 .. option:: --eta-interval=time
192 By default, fio requests client ETA status roughly every second. With
193 this option, the interval is configurable. Fio imposes a minimum
194 allowed time to avoid flooding the console, less than 250 msec is
197 .. option:: --eta-newline=time
199 Force a new line for every `time` period passed. When the unit is omitted,
200 the value is interpreted in seconds.
202 .. option:: --status-interval=time
204 Force a full status dump of cumulative (from job start) values at `time`
205 intervals. This option does *not* provide per-period measurements. So
206 values such as bandwidth are running averages. When the time unit is omitted,
207 `time` is interpreted in seconds. Note that using this option with
208 ``--output-format=json`` will yield output that technically isn't valid
209 json, since the output will be collated sets of valid json. It will need
210 to be split into valid sets of json after the run.
212 .. option:: --section=name
214 Only run specified section `name` in job file. Multiple sections can be specified.
215 The ``--section`` option allows one to combine related jobs into one file.
216 E.g. one job file could define light, moderate, and heavy sections. Tell
217 fio to run only the "heavy" section by giving ``--section=heavy``
218 command line option. One can also specify the "write" operations in one
219 section and "verify" operation in another section. The ``--section`` option
220 only applies to job sections. The reserved *global* section is always
223 .. option:: --alloc-size=kb
225 Allocate additional internal smalloc pools of size `kb` in KiB. The
226 ``--alloc-size`` option increases shared memory set aside for use by fio.
227 If running large jobs with randommap enabled, fio can run out of memory.
228 Smalloc is an internal allocator for shared structures from a fixed size
229 memory pool and can grow to 16 pools. The pool size defaults to 16MiB.
231 NOTE: While running :file:`.fio_smalloc.*` backing store files are visible
234 .. option:: --warnings-fatal
236 All fio parser warnings are fatal, causing fio to exit with an
239 .. option:: --max-jobs=nr
241 Set the maximum number of threads/processes to support to `nr`.
242 NOTE: On Linux, it may be necessary to increase the shared-memory
243 limit (:file:`/proc/sys/kernel/shmmax`) if fio runs into errors while
246 .. option:: --server=args
248 Start a backend server, with `args` specifying what to listen to.
249 See `Client/Server`_ section.
251 .. option:: --daemonize=pidfile
253 Background a fio server, writing the pid to the given `pidfile` file.
255 .. option:: --client=hostname
257 Instead of running the jobs locally, send and run them on the given `hostname`
258 or set of `hostname`\s. See `Client/Server`_ section.
260 .. option:: --remote-config=file
262 Tell fio server to load this local `file`.
264 .. option:: --idle-prof=option
266 Report CPU idleness. `option` is one of the following:
269 Run unit work calibration only and exit.
272 Show aggregate system idleness and unit work.
275 As **system** but also show per CPU idleness.
277 .. option:: --inflate-log=log
279 Inflate and output compressed `log`.
281 .. option:: --trigger-file=file
283 Execute trigger command when `file` exists.
285 .. option:: --trigger-timeout=time
287 Execute trigger at this `time`.
289 .. option:: --trigger=command
291 Set this `command` as local trigger.
293 .. option:: --trigger-remote=command
295 Set this `command` as remote trigger.
297 .. option:: --aux-path=path
299 Use the directory specified by `path` for generated state files instead
300 of the current working directory.
302 Any parameters following the options will be assumed to be job files, unless
303 they match a job file parameter. Multiple job files can be listed and each job
304 file will be regarded as a separate group. Fio will :option:`stonewall`
305 execution between each group.
311 As previously described, fio accepts one or more job files describing what it is
312 supposed to do. The job file format is the classic ini file, where the names
313 enclosed in [] brackets define the job name. You are free to use any ASCII name
314 you want, except *global* which has special meaning. Following the job name is
315 a sequence of zero or more parameters, one per line, that define the behavior of
316 the job. If the first character in a line is a ';' or a '#', the entire line is
317 discarded as a comment.
319 A *global* section sets defaults for the jobs described in that file. A job may
320 override a *global* section parameter, and a job file may even have several
321 *global* sections if so desired. A job is only affected by a *global* section
324 The :option:`--cmdhelp` option also lists all options. If used with a `command`
325 argument, :option:`--cmdhelp` will detail the given `command`.
327 See the `examples/` directory for inspiration on how to write job files. Note
328 the copyright and license requirements currently apply to `examples/` files.
330 So let's look at a really simple job file that defines two processes, each
331 randomly reading from a 128MiB file:
335 ; -- start job file --
346 As you can see, the job file sections themselves are empty as all the described
347 parameters are shared. As no :option:`filename` option is given, fio makes up a
348 `filename` for each of the jobs as it sees fit. On the command line, this job
349 would look as follows::
351 $ fio --name=global --rw=randread --size=128m --name=job1 --name=job2
354 Let's look at an example that has a number of processes writing randomly to
359 ; -- start job file --
370 Here we have no *global* section, as we only have one job defined anyway. We
371 want to use async I/O here, with a depth of 4 for each file. We also increased
372 the buffer size used to 32KiB and define numjobs to 4 to fork 4 identical
373 jobs. The result is 4 processes each randomly writing to their own 64MiB
374 file. Instead of using the above job file, you could have given the parameters
375 on the command line. For this case, you would specify::
377 $ fio --name=random-writers --ioengine=libaio --iodepth=4 --rw=randwrite --bs=32k --direct=0 --size=64m --numjobs=4
379 When fio is utilized as a basis of any reasonably large test suite, it might be
380 desirable to share a set of standardized settings across multiple job files.
381 Instead of copy/pasting such settings, any section may pull in an external
382 :file:`filename.fio` file with *include filename* directive, as in the following
385 ; -- start job file including.fio --
389 include glob-include.fio
396 include test-include.fio
397 ; -- end job file including.fio --
401 ; -- start job file glob-include.fio --
404 ; -- end job file glob-include.fio --
408 ; -- start job file test-include.fio --
411 ; -- end job file test-include.fio --
413 Settings pulled into a section apply to that section only (except *global*
414 section). Include directives may be nested in that any included file may contain
415 further include directive(s). Include files may not contain [] sections.
418 Environment variables
419 ~~~~~~~~~~~~~~~~~~~~~
421 Fio also supports environment variable expansion in job files. Any sub-string of
422 the form ``${VARNAME}`` as part of an option value (in other words, on the right
423 of the '='), will be expanded to the value of the environment variable called
424 `VARNAME`. If no such environment variable is defined, or `VARNAME` is the
425 empty string, the empty string will be substituted.
427 As an example, let's look at a sample fio invocation and job file::
429 $ SIZE=64m NUMJOBS=4 fio jobfile.fio
433 ; -- start job file --
440 This will expand to the following equivalent job file at runtime:
444 ; -- start job file --
451 Fio ships with a few example job files, you can also look there for inspiration.
456 Additionally, fio has a set of reserved keywords that will be replaced
457 internally with the appropriate value. Those keywords are:
461 The architecture page size of the running system.
465 Megabytes of total memory in the system.
469 Number of online available CPUs.
471 These can be used on the command line or in the job file, and will be
472 automatically substituted with the current system values when the job is
473 run. Simple math is also supported on these keywords, so you can perform actions
478 and get that properly expanded to 8 times the size of memory in the machine.
484 This section describes in details each parameter associated with a job. Some
485 parameters take an option of a given type, such as an integer or a
486 string. Anywhere a numeric value is required, an arithmetic expression may be
487 used, provided it is surrounded by parentheses. Supported operators are:
496 For time values in expressions, units are microseconds by default. This is
497 different than for time values not in expressions (not enclosed in
498 parentheses). The following types are used:
505 String: A sequence of alphanumeric characters.
508 Integer with possible time suffix. Without a unit value is interpreted as
509 seconds unless otherwise specified. Accepts a suffix of 'd' for days, 'h' for
510 hours, 'm' for minutes, 's' for seconds, 'ms' (or 'msec') for milliseconds and
511 'us' (or 'usec') for microseconds. For example, use 10m for 10 minutes.
516 Integer. A whole number value, which may contain an integer prefix
517 and an integer suffix:
519 [*integer prefix*] **number** [*integer suffix*]
521 The optional *integer prefix* specifies the number's base. The default
522 is decimal. *0x* specifies hexadecimal.
524 The optional *integer suffix* specifies the number's units, and includes an
525 optional unit prefix and an optional unit. For quantities of data, the
526 default unit is bytes. For quantities of time, the default unit is seconds
527 unless otherwise specified.
529 With :option:`kb_base`\=1000, fio follows international standards for unit
530 prefixes. To specify power-of-10 decimal values defined in the
531 International System of Units (SI):
533 * *K* -- means kilo (K) or 1000
534 * *M* -- means mega (M) or 1000**2
535 * *G* -- means giga (G) or 1000**3
536 * *T* -- means tera (T) or 1000**4
537 * *P* -- means peta (P) or 1000**5
539 To specify power-of-2 binary values defined in IEC 80000-13:
541 * *Ki* -- means kibi (Ki) or 1024
542 * *Mi* -- means mebi (Mi) or 1024**2
543 * *Gi* -- means gibi (Gi) or 1024**3
544 * *Ti* -- means tebi (Ti) or 1024**4
545 * *Pi* -- means pebi (Pi) or 1024**5
547 With :option:`kb_base`\=1024 (the default), the unit prefixes are opposite
548 from those specified in the SI and IEC 80000-13 standards to provide
549 compatibility with old scripts. For example, 4k means 4096.
551 For quantities of data, an optional unit of 'B' may be included
552 (e.g., 'kB' is the same as 'k').
554 The *integer suffix* is not case sensitive (e.g., m/mi mean mebi/mega,
555 not milli). 'b' and 'B' both mean byte, not bit.
557 Examples with :option:`kb_base`\=1000:
559 * *4 KiB*: 4096, 4096b, 4096B, 4ki, 4kib, 4kiB, 4Ki, 4KiB
560 * *1 MiB*: 1048576, 1mi, 1024ki
561 * *1 MB*: 1000000, 1m, 1000k
562 * *1 TiB*: 1099511627776, 1ti, 1024gi, 1048576mi
563 * *1 TB*: 1000000000, 1t, 1000m, 1000000k
565 Examples with :option:`kb_base`\=1024 (default):
567 * *4 KiB*: 4096, 4096b, 4096B, 4k, 4kb, 4kB, 4K, 4KB
568 * *1 MiB*: 1048576, 1m, 1024k
569 * *1 MB*: 1000000, 1mi, 1000ki
570 * *1 TiB*: 1099511627776, 1t, 1024g, 1048576m
571 * *1 TB*: 1000000000, 1ti, 1000mi, 1000000ki
573 To specify times (units are not case sensitive):
577 * *M* -- means minutes
578 * *s* -- or sec means seconds (default)
579 * *ms* -- or *msec* means milliseconds
580 * *us* -- or *usec* means microseconds
582 If the option accepts an upper and lower range, use a colon ':' or
583 minus '-' to separate such values. See :ref:`irange <irange>`.
584 If the lower value specified happens to be larger than the upper value
585 the two values are swapped.
590 Boolean. Usually parsed as an integer, however only defined for
591 true and false (1 and 0).
596 Integer range with suffix. Allows value range to be given, such as
597 1024-4096. A colon may also be used as the separator, e.g. 1k:4k. If the
598 option allows two sets of ranges, they can be specified with a ',' or '/'
599 delimiter: 1k-4k/8k-32k. Also see :ref:`int <int>`.
602 A list of floating point numbers, separated by a ':' character.
604 With the above in mind, here follows the complete list of fio job parameters.
610 .. option:: kb_base=int
612 Select the interpretation of unit prefixes in input parameters.
615 Inputs comply with IEC 80000-13 and the International
616 System of Units (SI). Use:
618 - power-of-2 values with IEC prefixes (e.g., KiB)
619 - power-of-10 values with SI prefixes (e.g., kB)
622 Compatibility mode (default). To avoid breaking old scripts:
624 - power-of-2 values with SI prefixes
625 - power-of-10 values with IEC prefixes
627 See :option:`bs` for more details on input parameters.
629 Outputs always use correct prefixes. Most outputs include both
632 bw=2383.3kB/s (2327.4KiB/s)
634 If only one value is reported, then kb_base selects the one to use:
636 **1000** -- SI prefixes
638 **1024** -- IEC prefixes
640 .. option:: unit_base=int
642 Base unit for reporting. Allowed values are:
645 Use auto-detection (default).
657 ASCII name of the job. This may be used to override the name printed by fio
658 for this job. Otherwise the job name is used. On the command line this
659 parameter has the special purpose of also signaling the start of a new job.
661 .. option:: description=str
663 Text description of the job. Doesn't do anything except dump this text
664 description when this job is run. It's not parsed.
666 .. option:: loops=int
668 Run the specified number of iterations of this job. Used to repeat the same
669 workload a given number of times. Defaults to 1.
671 .. option:: numjobs=int
673 Create the specified number of clones of this job. Each clone of job
674 is spawned as an independent thread or process. May be used to setup a
675 larger number of threads/processes doing the same thing. Each thread is
676 reported separately; to see statistics for all clones as a whole, use
677 :option:`group_reporting` in conjunction with :option:`new_group`.
678 See :option:`--max-jobs`. Default: 1.
681 Time related parameters
682 ~~~~~~~~~~~~~~~~~~~~~~~
684 .. option:: runtime=time
686 Tell fio to terminate processing after the specified period of time. It
687 can be quite hard to determine for how long a specified job will run, so
688 this parameter is handy to cap the total runtime to a given time. When
689 the unit is omitted, the value is interpreted in seconds.
691 .. option:: time_based
693 If set, fio will run for the duration of the :option:`runtime` specified
694 even if the file(s) are completely read or written. It will simply loop over
695 the same workload as many times as the :option:`runtime` allows.
697 .. option:: startdelay=irange(time)
699 Delay the start of job for the specified amount of time. Can be a single
700 value or a range. When given as a range, each thread will choose a value
701 randomly from within the range. Value is in seconds if a unit is omitted.
703 .. option:: ramp_time=time
705 If set, fio will run the specified workload for this amount of time before
706 logging any performance numbers. Useful for letting performance settle
707 before logging results, thus minimizing the runtime required for stable
708 results. Note that the ``ramp_time`` is considered lead in time for a job,
709 thus it will increase the total runtime if a special timeout or
710 :option:`runtime` is specified. When the unit is omitted, the value is
713 .. option:: clocksource=str
715 Use the given clocksource as the base of timing. The supported options are:
718 :manpage:`gettimeofday(2)`
721 :manpage:`clock_gettime(2)`
724 Internal CPU clock source
726 cpu is the preferred clocksource if it is reliable, as it is very fast (and
727 fio is heavy on time calls). Fio will automatically use this clocksource if
728 it's supported and considered reliable on the system it is running on,
729 unless another clocksource is specifically set. For x86/x86-64 CPUs, this
730 means supporting TSC Invariant.
732 .. option:: gtod_reduce=bool
734 Enable all of the :manpage:`gettimeofday(2)` reducing options
735 (:option:`disable_clat`, :option:`disable_slat`, :option:`disable_bw_measurement`) plus
736 reduce precision of the timeout somewhat to really shrink the
737 :manpage:`gettimeofday(2)` call count. With this option enabled, we only do
738 about 0.4% of the :manpage:`gettimeofday(2)` calls we would have done if all
739 time keeping was enabled.
741 .. option:: gtod_cpu=int
743 Sometimes it's cheaper to dedicate a single thread of execution to just
744 getting the current time. Fio (and databases, for instance) are very
745 intensive on :manpage:`gettimeofday(2)` calls. With this option, you can set
746 one CPU aside for doing nothing but logging current time to a shared memory
747 location. Then the other threads/processes that run I/O workloads need only
748 copy that segment, instead of entering the kernel with a
749 :manpage:`gettimeofday(2)` call. The CPU set aside for doing these time
750 calls will be excluded from other uses. Fio will manually clear it from the
751 CPU mask of other jobs.
757 .. option:: directory=str
759 Prefix filenames with this directory. Used to place files in a different
760 location than :file:`./`. You can specify a number of directories by
761 separating the names with a ':' character. These directories will be
762 assigned equally distributed to job clones created by :option:`numjobs` as
763 long as they are using generated filenames. If specific `filename(s)` are
764 set fio will use the first listed directory, and thereby matching the
765 `filename` semantic (which generates a file for each clone if not
766 specified, but lets all clones use the same file if set).
768 See the :option:`filename` option for information on how to escape "``:``" and
769 "``\``" characters within the directory path itself.
771 Note: To control the directory fio will use for internal state files
772 use :option:`--aux-path`.
774 .. option:: filename=str
776 Fio normally makes up a `filename` based on the job name, thread number, and
777 file number (see :option:`filename_format`). If you want to share files
778 between threads in a job or several
779 jobs with fixed file paths, specify a `filename` for each of them to override
780 the default. If the ioengine is file based, you can specify a number of files
781 by separating the names with a ':' colon. So if you wanted a job to open
782 :file:`/dev/sda` and :file:`/dev/sdb` as the two working files, you would use
783 ``filename=/dev/sda:/dev/sdb``. This also means that whenever this option is
784 specified, :option:`nrfiles` is ignored. The size of regular files specified
785 by this option will be :option:`size` divided by number of files unless an
786 explicit size is specified by :option:`filesize`.
788 Each colon and backslash in the wanted path must be escaped with a ``\``
789 character. For instance, if the path is :file:`/dev/dsk/foo@3,0:c` then you
790 would use ``filename=/dev/dsk/foo@3,0\:c`` and if the path is
791 :file:`F:\\filename` then you would use ``filename=F\:\\filename``.
793 On Windows, disk devices are accessed as :file:`\\\\.\\PhysicalDrive0` for
794 the first device, :file:`\\\\.\\PhysicalDrive1` for the second etc.
795 Note: Windows and FreeBSD prevent write access to areas
796 of the disk containing in-use data (e.g. filesystems).
798 The filename "`-`" is a reserved name, meaning *stdin* or *stdout*. Which
799 of the two depends on the read/write direction set.
801 .. option:: filename_format=str
803 If sharing multiple files between jobs, it is usually necessary to have fio
804 generate the exact names that you want. By default, fio will name a file
805 based on the default file format specification of
806 :file:`jobname.jobnumber.filenumber`. With this option, that can be
807 customized. Fio will recognize and replace the following keywords in this
811 The name of the worker thread or process.
813 The incremental number of the worker thread or process.
815 The incremental number of the file for that worker thread or
818 To have dependent jobs share a set of files, this option can be set to have
819 fio generate filenames that are shared between the two. For instance, if
820 :file:`testfiles.$filenum` is specified, file number 4 for any job will be
821 named :file:`testfiles.4`. The default of :file:`$jobname.$jobnum.$filenum`
822 will be used if no other format specifier is given.
824 If you specify a path then the directories will be created up to the
825 main directory for the file. So for example if you specify
826 ``filename_format=a/b/c/$jobnum`` then the directories a/b/c will be
827 created before the file setup part of the job. If you specify
828 :option:`directory` then the path will be relative that directory,
829 otherwise it is treated as the absolute path.
831 .. option:: unique_filename=bool
833 To avoid collisions between networked clients, fio defaults to prefixing any
834 generated filenames (with a directory specified) with the source of the
835 client connecting. To disable this behavior, set this option to 0.
837 .. option:: opendir=str
839 Recursively open any files below directory `str`.
841 .. option:: lockfile=str
843 Fio defaults to not locking any files before it does I/O to them. If a file
844 or file descriptor is shared, fio can serialize I/O to that file to make the
845 end result consistent. This is usual for emulating real workloads that share
846 files. The lock modes are:
849 No locking. The default.
851 Only one thread or process may do I/O at a time, excluding all
854 Read-write locking on the file. Many readers may
855 access the file at the same time, but writes get exclusive access.
857 .. option:: nrfiles=int
859 Number of files to use for this job. Defaults to 1. The size of files
860 will be :option:`size` divided by this unless explicit size is specified by
861 :option:`filesize`. Files are created for each thread separately, and each
862 file will have a file number within its name by default, as explained in
863 :option:`filename` section.
866 .. option:: openfiles=int
868 Number of files to keep open at the same time. Defaults to the same as
869 :option:`nrfiles`, can be set smaller to limit the number simultaneous
872 .. option:: file_service_type=str
874 Defines how fio decides which file from a job to service next. The following
878 Choose a file at random.
881 Round robin over opened files. This is the default.
884 Finish one file before moving on to the next. Multiple files can
885 still be open depending on :option:`openfiles`.
888 Use a *Zipf* distribution to decide what file to access.
891 Use a *Pareto* distribution to decide what file to access.
894 Use a *Gaussian* (normal) distribution to decide what file to
900 For *random*, *roundrobin*, and *sequential*, a postfix can be appended to
901 tell fio how many I/Os to issue before switching to a new file. For example,
902 specifying ``file_service_type=random:8`` would cause fio to issue
903 8 I/Os before selecting a new file at random. For the non-uniform
904 distributions, a floating point postfix can be given to influence how the
905 distribution is skewed. See :option:`random_distribution` for a description
906 of how that would work.
908 .. option:: ioscheduler=str
910 Attempt to switch the device hosting the file to the specified I/O scheduler
913 .. option:: create_serialize=bool
915 If true, serialize the file creation for the jobs. This may be handy to
916 avoid interleaving of data files, which may greatly depend on the filesystem
917 used and even the number of processors in the system. Default: true.
919 .. option:: create_fsync=bool
921 :manpage:`fsync(2)` the data file after creation. This is the default.
923 .. option:: create_on_open=bool
925 If true, don't pre-create files but allow the job's open() to create a file
926 when it's time to do I/O. Default: false -- pre-create all necessary files
929 .. option:: create_only=bool
931 If true, fio will only run the setup phase of the job. If files need to be
932 laid out or updated on disk, only that will be done -- the actual job contents
933 are not executed. Default: false.
935 .. option:: allow_file_create=bool
937 If true, fio is permitted to create files as part of its workload. If this
938 option is false, then fio will error out if
939 the files it needs to use don't already exist. Default: true.
941 .. option:: allow_mounted_write=bool
943 If this isn't set, fio will abort jobs that are destructive (e.g. that write)
944 to what appears to be a mounted device or partition. This should help catch
945 creating inadvertently destructive tests, not realizing that the test will
946 destroy data on the mounted file system. Note that some platforms don't allow
947 writing against a mounted device regardless of this option. Default: false.
949 .. option:: pre_read=bool
951 If this is given, files will be pre-read into memory before starting the
952 given I/O operation. This will also clear the :option:`invalidate` flag,
953 since it is pointless to pre-read and then drop the cache. This will only
954 work for I/O engines that are seek-able, since they allow you to read the
955 same data multiple times. Thus it will not work on non-seekable I/O engines
956 (e.g. network, splice). Default: false.
958 .. option:: unlink=bool
960 Unlink the job files when done. Not the default, as repeated runs of that
961 job would then waste time recreating the file set again and again. Default:
964 .. option:: unlink_each_loop=bool
966 Unlink job files after each iteration or loop. Default: false.
968 .. option:: zonemode=str
973 The :option:`zonerange`, :option:`zonesize` and
974 :option:`zoneskip` parameters are ignored.
976 I/O happens in a single zone until
977 :option:`zonesize` bytes have been transferred.
978 After that number of bytes has been
979 transferred processing of the next zone
982 Zoned block device mode. I/O happens
983 sequentially in each zone, even if random I/O
984 has been selected. Random I/O happens across
985 all zones instead of being restricted to a
986 single zone. The :option:`zoneskip` parameter
987 is ignored. :option:`zonerange` and
988 :option:`zonesize` must be identical.
990 .. option:: zonerange=int
992 Size of a single zone. See also :option:`zonesize` and
995 .. option:: zonesize=int
997 For :option:`zonemode` =strided, this is the number of bytes to
998 transfer before skipping :option:`zoneskip` bytes. If this parameter
999 is smaller than :option:`zonerange` then only a fraction of each zone
1000 with :option:`zonerange` bytes will be accessed. If this parameter is
1001 larger than :option:`zonerange` then each zone will be accessed
1002 multiple times before skipping to the next zone.
1004 For :option:`zonemode` =zbd, this is the size of a single zone. The
1005 :option:`zonerange` parameter is ignored in this mode.
1007 .. option:: zoneskip=int
1009 For :option:`zonemode` =strided, the number of bytes to skip after
1010 :option:`zonesize` bytes of data have been transferred. This parameter
1011 must be zero for :option:`zonemode` =zbd.
1013 .. option:: read_beyond_wp=bool
1015 This parameter applies to :option:`zonemode` =zbd only.
1017 Zoned block devices are block devices that consist of multiple zones.
1018 Each zone has a type, e.g. conventional or sequential. A conventional
1019 zone can be written at any offset that is a multiple of the block
1020 size. Sequential zones must be written sequentially. The position at
1021 which a write must occur is called the write pointer. A zoned block
1022 device can be either drive managed, host managed or host aware. For
1023 host managed devices the host must ensure that writes happen
1024 sequentially. Fio recognizes host managed devices and serializes
1025 writes to sequential zones for these devices.
1027 If a read occurs in a sequential zone beyond the write pointer then
1028 the zoned block device will complete the read without reading any data
1029 from the storage medium. Since such reads lead to unrealistically high
1030 bandwidth and IOPS numbers fio only reads beyond the write pointer if
1031 explicitly told to do so. Default: false.
1033 .. option:: max_open_zones=int
1035 When running a random write test across an entire drive many more
1036 zones will be open than in a typical application workload. Hence this
1037 command line option that allows to limit the number of open zones. The
1038 number of open zones is defined as the number of zones to which write
1039 commands are issued.
1041 .. option:: zone_reset_threshold=float
1043 A number between zero and one that indicates the ratio of logical
1044 blocks with data to the total number of logical blocks in the test
1045 above which zones should be reset periodically.
1047 .. option:: zone_reset_frequency=float
1049 A number between zero and one that indicates how often a zone reset
1050 should be issued if the zone reset threshold has been exceeded. A zone
1051 reset is submitted after each (1 / zone_reset_frequency) write
1052 requests. This and the previous parameter can be used to simulate
1053 garbage collection activity.
1059 .. option:: direct=bool
1061 If value is true, use non-buffered I/O. This is usually O_DIRECT. Note that
1062 OpenBSD and ZFS on Solaris don't support direct I/O. On Windows the synchronous
1063 ioengines don't support direct I/O. Default: false.
1065 .. option:: atomic=bool
1067 If value is true, attempt to use atomic direct I/O. Atomic writes are
1068 guaranteed to be stable once acknowledged by the operating system. Only
1069 Linux supports O_ATOMIC right now.
1071 .. option:: buffered=bool
1073 If value is true, use buffered I/O. This is the opposite of the
1074 :option:`direct` option. Defaults to true.
1076 .. option:: readwrite=str, rw=str
1078 Type of I/O pattern. Accepted values are:
1085 Sequential trims (Linux block devices and SCSI
1086 character devices only).
1092 Random trims (Linux block devices and SCSI
1093 character devices only).
1095 Sequential mixed reads and writes.
1097 Random mixed reads and writes.
1099 Sequential trim+write sequences. Blocks will be trimmed first,
1100 then the same blocks will be written to.
1102 Fio defaults to read if the option is not specified. For the mixed I/O
1103 types, the default is to split them 50/50. For certain types of I/O the
1104 result may still be skewed a bit, since the speed may be different.
1106 It is possible to specify the number of I/Os to do before getting a new
1107 offset by appending ``:<nr>`` to the end of the string given. For a
1108 random read, it would look like ``rw=randread:8`` for passing in an offset
1109 modifier with a value of 8. If the suffix is used with a sequential I/O
1110 pattern, then the *<nr>* value specified will be **added** to the generated
1111 offset for each I/O turning sequential I/O into sequential I/O with holes.
1112 For instance, using ``rw=write:4k`` will skip 4k for every write. Also see
1113 the :option:`rw_sequencer` option.
1115 .. option:: rw_sequencer=str
1117 If an offset modifier is given by appending a number to the ``rw=<str>``
1118 line, then this option controls how that number modifies the I/O offset
1119 being generated. Accepted values are:
1122 Generate sequential offset.
1124 Generate the same offset.
1126 ``sequential`` is only useful for random I/O, where fio would normally
1127 generate a new random offset for every I/O. If you append e.g. 8 to randread,
1128 you would get a new random offset for every 8 I/Os. The result would be a
1129 seek for only every 8 I/Os, instead of for every I/O. Use ``rw=randread:8``
1130 to specify that. As sequential I/O is already sequential, setting
1131 ``sequential`` for that would not result in any differences. ``identical``
1132 behaves in a similar fashion, except it sends the same offset 8 number of
1133 times before generating a new offset.
1135 .. option:: unified_rw_reporting=bool
1137 Fio normally reports statistics on a per data direction basis, meaning that
1138 reads, writes, and trims are accounted and reported separately. If this
1139 option is set fio sums the results and report them as "mixed" instead.
1141 .. option:: randrepeat=bool
1143 Seed the random number generator used for random I/O patterns in a
1144 predictable way so the pattern is repeatable across runs. Default: true.
1146 .. option:: allrandrepeat=bool
1148 Seed all random number generators in a predictable way so results are
1149 repeatable across runs. Default: false.
1151 .. option:: randseed=int
1153 Seed the random number generators based on this seed value, to be able to
1154 control what sequence of output is being generated. If not set, the random
1155 sequence depends on the :option:`randrepeat` setting.
1157 .. option:: fallocate=str
1159 Whether pre-allocation is performed when laying down files.
1160 Accepted values are:
1163 Do not pre-allocate space.
1166 Use a platform's native pre-allocation call but fall back to
1167 **none** behavior if it fails/is not implemented.
1170 Pre-allocate via :manpage:`posix_fallocate(3)`.
1173 Pre-allocate via :manpage:`fallocate(2)` with
1174 FALLOC_FL_KEEP_SIZE set.
1177 Extend file to final size via :manpage:`ftruncate(2)`
1178 instead of allocating.
1181 Backward-compatible alias for **none**.
1184 Backward-compatible alias for **posix**.
1186 May not be available on all supported platforms. **keep** is only available
1187 on Linux. If using ZFS on Solaris this cannot be set to **posix**
1188 because ZFS doesn't support pre-allocation. Default: **native** if any
1189 pre-allocation methods except **truncate** are available, **none** if not.
1191 Note that using **truncate** on Windows will interact surprisingly
1192 with non-sequential write patterns. When writing to a file that has
1193 been extended by setting the end-of-file information, Windows will
1194 backfill the unwritten portion of the file up to that offset with
1195 zeroes before issuing the new write. This means that a single small
1196 write to the end of an extended file will stall until the entire
1197 file has been filled with zeroes.
1199 .. option:: fadvise_hint=str
1201 Use :manpage:`posix_fadvise(2)` or :manpage:`posix_fadvise(2)` to
1202 advise the kernel on what I/O patterns are likely to be issued.
1203 Accepted values are:
1206 Backwards-compatible hint for "no hint".
1209 Backwards compatible hint for "advise with fio workload type". This
1210 uses **FADV_RANDOM** for a random workload, and **FADV_SEQUENTIAL**
1211 for a sequential workload.
1214 Advise using **FADV_SEQUENTIAL**.
1217 Advise using **FADV_RANDOM**.
1219 .. option:: write_hint=str
1221 Use :manpage:`fcntl(2)` to advise the kernel what life time to expect
1222 from a write. Only supported on Linux, as of version 4.13. Accepted
1226 No particular life time associated with this file.
1229 Data written to this file has a short life time.
1232 Data written to this file has a medium life time.
1235 Data written to this file has a long life time.
1238 Data written to this file has a very long life time.
1240 The values are all relative to each other, and no absolute meaning
1241 should be associated with them.
1243 .. option:: offset=int
1245 Start I/O at the provided offset in the file, given as either a fixed size in
1246 bytes or a percentage. If a percentage is given, the generated offset will be
1247 aligned to the minimum ``blocksize`` or to the value of ``offset_align`` if
1248 provided. Data before the given offset will not be touched. This
1249 effectively caps the file size at `real_size - offset`. Can be combined with
1250 :option:`size` to constrain the start and end range of the I/O workload.
1251 A percentage can be specified by a number between 1 and 100 followed by '%',
1252 for example, ``offset=20%`` to specify 20%.
1254 .. option:: offset_align=int
1256 If set to non-zero value, the byte offset generated by a percentage ``offset``
1257 is aligned upwards to this value. Defaults to 0 meaning that a percentage
1258 offset is aligned to the minimum block size.
1260 .. option:: offset_increment=int
1262 If this is provided, then the real offset becomes `offset + offset_increment
1263 * thread_number`, where the thread number is a counter that starts at 0 and
1264 is incremented for each sub-job (i.e. when :option:`numjobs` option is
1265 specified). This option is useful if there are several jobs which are
1266 intended to operate on a file in parallel disjoint segments, with even
1267 spacing between the starting points. Percentages can be used for this option.
1268 If a percentage is given, the generated offset will be aligned to the minimum
1269 ``blocksize`` or to the value of ``offset_align`` if provided.
1271 .. option:: number_ios=int
1273 Fio will normally perform I/Os until it has exhausted the size of the region
1274 set by :option:`size`, or if it exhaust the allocated time (or hits an error
1275 condition). With this setting, the range/size can be set independently of
1276 the number of I/Os to perform. When fio reaches this number, it will exit
1277 normally and report status. Note that this does not extend the amount of I/O
1278 that will be done, it will only stop fio if this condition is met before
1279 other end-of-job criteria.
1281 .. option:: fsync=int
1283 If writing to a file, issue an :manpage:`fsync(2)` (or its equivalent) of
1284 the dirty data for every number of blocks given. For example, if you give 32
1285 as a parameter, fio will sync the file after every 32 writes issued. If fio is
1286 using non-buffered I/O, we may not sync the file. The exception is the sg
1287 I/O engine, which synchronizes the disk cache anyway. Defaults to 0, which
1288 means fio does not periodically issue and wait for a sync to complete. Also
1289 see :option:`end_fsync` and :option:`fsync_on_close`.
1291 .. option:: fdatasync=int
1293 Like :option:`fsync` but uses :manpage:`fdatasync(2)` to only sync data and
1294 not metadata blocks. In Windows, FreeBSD, DragonFlyBSD or OSX there is no
1295 :manpage:`fdatasync(2)` so this falls back to using :manpage:`fsync(2)`.
1296 Defaults to 0, which means fio does not periodically issue and wait for a
1297 data-only sync to complete.
1299 .. option:: write_barrier=int
1301 Make every `N-th` write a barrier write.
1303 .. option:: sync_file_range=str:int
1305 Use :manpage:`sync_file_range(2)` for every `int` number of write
1306 operations. Fio will track range of writes that have happened since the last
1307 :manpage:`sync_file_range(2)` call. `str` can currently be one or more of:
1310 SYNC_FILE_RANGE_WAIT_BEFORE
1312 SYNC_FILE_RANGE_WRITE
1314 SYNC_FILE_RANGE_WAIT_AFTER
1316 So if you do ``sync_file_range=wait_before,write:8``, fio would use
1317 ``SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE`` for every 8
1318 writes. Also see the :manpage:`sync_file_range(2)` man page. This option is
1321 .. option:: overwrite=bool
1323 If true, writes to a file will always overwrite existing data. If the file
1324 doesn't already exist, it will be created before the write phase begins. If
1325 the file exists and is large enough for the specified write phase, nothing
1326 will be done. Default: false.
1328 .. option:: end_fsync=bool
1330 If true, :manpage:`fsync(2)` file contents when a write stage has completed.
1333 .. option:: fsync_on_close=bool
1335 If true, fio will :manpage:`fsync(2)` a dirty file on close. This differs
1336 from :option:`end_fsync` in that it will happen on every file close, not
1337 just at the end of the job. Default: false.
1339 .. option:: rwmixread=int
1341 Percentage of a mixed workload that should be reads. Default: 50.
1343 .. option:: rwmixwrite=int
1345 Percentage of a mixed workload that should be writes. If both
1346 :option:`rwmixread` and :option:`rwmixwrite` is given and the values do not
1347 add up to 100%, the latter of the two will be used to override the
1348 first. This may interfere with a given rate setting, if fio is asked to
1349 limit reads or writes to a certain rate. If that is the case, then the
1350 distribution may be skewed. Default: 50.
1352 .. option:: random_distribution=str:float[,str:float][,str:float]
1354 By default, fio will use a completely uniform random distribution when asked
1355 to perform random I/O. Sometimes it is useful to skew the distribution in
1356 specific ways, ensuring that some parts of the data is more hot than others.
1357 fio includes the following distribution models:
1360 Uniform random distribution
1369 Normal (Gaussian) distribution
1372 Zoned random distribution
1375 Zone absolute random distribution
1377 When using a **zipf** or **pareto** distribution, an input value is also
1378 needed to define the access pattern. For **zipf**, this is the `Zipf
1379 theta`. For **pareto**, it's the `Pareto power`. Fio includes a test
1380 program, :command:`fio-genzipf`, that can be used visualize what the given input
1381 values will yield in terms of hit rates. If you wanted to use **zipf** with
1382 a `theta` of 1.2, you would use ``random_distribution=zipf:1.2`` as the
1383 option. If a non-uniform model is used, fio will disable use of the random
1384 map. For the **normal** distribution, a normal (Gaussian) deviation is
1385 supplied as a value between 0 and 100.
1387 For a **zoned** distribution, fio supports specifying percentages of I/O
1388 access that should fall within what range of the file or device. For
1389 example, given a criteria of:
1391 * 60% of accesses should be to the first 10%
1392 * 30% of accesses should be to the next 20%
1393 * 8% of accesses should be to the next 30%
1394 * 2% of accesses should be to the next 40%
1396 we can define that through zoning of the random accesses. For the above
1397 example, the user would do::
1399 random_distribution=zoned:60/10:30/20:8/30:2/40
1401 A **zoned_abs** distribution works exactly like the **zoned**, except
1402 that it takes absolute sizes. For example, let's say you wanted to
1403 define access according to the following criteria:
1405 * 60% of accesses should be to the first 20G
1406 * 30% of accesses should be to the next 100G
1407 * 10% of accesses should be to the next 500G
1409 we can define an absolute zoning distribution with:
1411 random_distribution=zoned_abs=60/20G:30/100G:10/500g
1413 For both **zoned** and **zoned_abs**, fio supports defining up to
1416 Similarly to how :option:`bssplit` works for setting ranges and
1417 percentages of block sizes. Like :option:`bssplit`, it's possible to
1418 specify separate zones for reads, writes, and trims. If just one set
1419 is given, it'll apply to all of them. This goes for both **zoned**
1420 **zoned_abs** distributions.
1422 .. option:: percentage_random=int[,int][,int]
1424 For a random workload, set how big a percentage should be random. This
1425 defaults to 100%, in which case the workload is fully random. It can be set
1426 from anywhere from 0 to 100. Setting it to 0 would make the workload fully
1427 sequential. Any setting in between will result in a random mix of sequential
1428 and random I/O, at the given percentages. Comma-separated values may be
1429 specified for reads, writes, and trims as described in :option:`blocksize`.
1431 .. option:: norandommap
1433 Normally fio will cover every block of the file when doing random I/O. If
1434 this option is given, fio will just get a new random offset without looking
1435 at past I/O history. This means that some blocks may not be read or written,
1436 and that some blocks may be read/written more than once. If this option is
1437 used with :option:`verify` and multiple blocksizes (via :option:`bsrange`),
1438 only intact blocks are verified, i.e., partially-overwritten blocks are
1439 ignored. With an async I/O engine and an I/O depth > 1, it is possible for
1440 the same block to be overwritten, which can cause verification errors. Either
1441 do not use norandommap in this case, or also use the lfsr random generator.
1443 .. option:: softrandommap=bool
1445 See :option:`norandommap`. If fio runs with the random block map enabled and
1446 it fails to allocate the map, if this option is set it will continue without
1447 a random block map. As coverage will not be as complete as with random maps,
1448 this option is disabled by default.
1450 .. option:: random_generator=str
1452 Fio supports the following engines for generating I/O offsets for random I/O:
1455 Strong 2^88 cycle random number generator.
1457 Linear feedback shift register generator.
1459 Strong 64-bit 2^258 cycle random number generator.
1461 **tausworthe** is a strong random number generator, but it requires tracking
1462 on the side if we want to ensure that blocks are only read or written
1463 once. **lfsr** guarantees that we never generate the same offset twice, and
1464 it's also less computationally expensive. It's not a true random generator,
1465 however, though for I/O purposes it's typically good enough. **lfsr** only
1466 works with single block sizes, not with workloads that use multiple block
1467 sizes. If used with such a workload, fio may read or write some blocks
1468 multiple times. The default value is **tausworthe**, unless the required
1469 space exceeds 2^32 blocks. If it does, then **tausworthe64** is
1470 selected automatically.
1476 .. option:: blocksize=int[,int][,int], bs=int[,int][,int]
1478 The block size in bytes used for I/O units. Default: 4096. A single value
1479 applies to reads, writes, and trims. Comma-separated values may be
1480 specified for reads, writes, and trims. A value not terminated in a comma
1481 applies to subsequent types.
1486 means 256k for reads, writes and trims.
1489 means 8k for reads, 32k for writes and trims.
1492 means 8k for reads, 32k for writes, and default for trims.
1495 means default for reads, 8k for writes and trims.
1498 means default for reads, 8k for writes, and default for trims.
1500 .. option:: blocksize_range=irange[,irange][,irange], bsrange=irange[,irange][,irange]
1502 A range of block sizes in bytes for I/O units. The issued I/O unit will
1503 always be a multiple of the minimum size, unless
1504 :option:`blocksize_unaligned` is set.
1506 Comma-separated ranges may be specified for reads, writes, and trims as
1507 described in :option:`blocksize`.
1509 Example: ``bsrange=1k-4k,2k-8k``.
1511 .. option:: bssplit=str[,str][,str]
1513 Sometimes you want even finer grained control of the block sizes
1514 issued, not just an even split between them. This option allows you to
1515 weight various block sizes, so that you are able to define a specific
1516 amount of block sizes issued. The format for this option is::
1518 bssplit=blocksize/percentage:blocksize/percentage
1520 for as many block sizes as needed. So if you want to define a workload
1521 that has 50% 64k blocks, 10% 4k blocks, and 40% 32k blocks, you would
1524 bssplit=4k/10:64k/50:32k/40
1526 Ordering does not matter. If the percentage is left blank, fio will
1527 fill in the remaining values evenly. So a bssplit option like this one::
1529 bssplit=4k/50:1k/:32k/
1531 would have 50% 4k ios, and 25% 1k and 32k ios. The percentages always
1532 add up to 100, if bssplit is given a range that adds up to more, it
1535 Comma-separated values may be specified for reads, writes, and trims as
1536 described in :option:`blocksize`.
1538 If you want a workload that has 50% 2k reads and 50% 4k reads, while
1539 having 90% 4k writes and 10% 8k writes, you would specify::
1541 bssplit=2k/50:4k/50,4k/90:8k/10
1543 Fio supports defining up to 64 different weights for each data
1546 .. option:: blocksize_unaligned, bs_unaligned
1548 If set, fio will issue I/O units with any size within
1549 :option:`blocksize_range`, not just multiples of the minimum size. This
1550 typically won't work with direct I/O, as that normally requires sector
1553 .. option:: bs_is_seq_rand=bool
1555 If this option is set, fio will use the normal read,write blocksize settings
1556 as sequential,random blocksize settings instead. Any random read or write
1557 will use the WRITE blocksize settings, and any sequential read or write will
1558 use the READ blocksize settings.
1560 .. option:: blockalign=int[,int][,int], ba=int[,int][,int]
1562 Boundary to which fio will align random I/O units. Default:
1563 :option:`blocksize`. Minimum alignment is typically 512b for using direct
1564 I/O, though it usually depends on the hardware block size. This option is
1565 mutually exclusive with using a random map for files, so it will turn off
1566 that option. Comma-separated values may be specified for reads, writes, and
1567 trims as described in :option:`blocksize`.
1573 .. option:: zero_buffers
1575 Initialize buffers with all zeros. Default: fill buffers with random data.
1577 .. option:: refill_buffers
1579 If this option is given, fio will refill the I/O buffers on every
1580 submit. Only makes sense if :option:`zero_buffers` isn't specified,
1581 naturally. Defaults to being unset i.e., the buffer is only filled at
1582 init time and the data in it is reused when possible but if any of
1583 :option:`verify`, :option:`buffer_compress_percentage` or
1584 :option:`dedupe_percentage` are enabled then `refill_buffers` is also
1585 automatically enabled.
1587 .. option:: scramble_buffers=bool
1589 If :option:`refill_buffers` is too costly and the target is using data
1590 deduplication, then setting this option will slightly modify the I/O buffer
1591 contents to defeat normal de-dupe attempts. This is not enough to defeat
1592 more clever block compression attempts, but it will stop naive dedupe of
1593 blocks. Default: true.
1595 .. option:: buffer_compress_percentage=int
1597 If this is set, then fio will attempt to provide I/O buffer content
1598 (on WRITEs) that compresses to the specified level. Fio does this by
1599 providing a mix of random data followed by fixed pattern data. The
1600 fixed pattern is either zeros, or the pattern specified by
1601 :option:`buffer_pattern`. If the `buffer_pattern` option is used, it
1602 might skew the compression ratio slightly. Setting
1603 `buffer_compress_percentage` to a value other than 100 will also
1604 enable :option:`refill_buffers` in order to reduce the likelihood that
1605 adjacent blocks are so similar that they over compress when seen
1606 together. See :option:`buffer_compress_chunk` for how to set a finer or
1607 coarser granularity for the random/fixed data region. Defaults to unset
1608 i.e., buffer data will not adhere to any compression level.
1610 .. option:: buffer_compress_chunk=int
1612 This setting allows fio to manage how big the random/fixed data region
1613 is when using :option:`buffer_compress_percentage`. When
1614 `buffer_compress_chunk` is set to some non-zero value smaller than the
1615 block size, fio can repeat the random/fixed region throughout the I/O
1616 buffer at the specified interval (which particularly useful when
1617 bigger block sizes are used for a job). When set to 0, fio will use a
1618 chunk size that matches the block size resulting in a single
1619 random/fixed region within the I/O buffer. Defaults to 512. When the
1620 unit is omitted, the value is interpreted in bytes.
1622 .. option:: buffer_pattern=str
1624 If set, fio will fill the I/O buffers with this pattern or with the contents
1625 of a file. If not set, the contents of I/O buffers are defined by the other
1626 options related to buffer contents. The setting can be any pattern of bytes,
1627 and can be prefixed with 0x for hex values. It may also be a string, where
1628 the string must then be wrapped with ``""``. Or it may also be a filename,
1629 where the filename must be wrapped with ``''`` in which case the file is
1630 opened and read. Note that not all the file contents will be read if that
1631 would cause the buffers to overflow. So, for example::
1633 buffer_pattern='filename'
1637 buffer_pattern="abcd"
1645 buffer_pattern=0xdeadface
1647 Also you can combine everything together in any order::
1649 buffer_pattern=0xdeadface"abcd"-12'filename'
1651 .. option:: dedupe_percentage=int
1653 If set, fio will generate this percentage of identical buffers when
1654 writing. These buffers will be naturally dedupable. The contents of the
1655 buffers depend on what other buffer compression settings have been set. It's
1656 possible to have the individual buffers either fully compressible, or not at
1657 all -- this option only controls the distribution of unique buffers. Setting
1658 this option will also enable :option:`refill_buffers` to prevent every buffer
1661 .. option:: invalidate=bool
1663 Invalidate the buffer/page cache parts of the files to be used prior to
1664 starting I/O if the platform and file type support it. Defaults to true.
1665 This will be ignored if :option:`pre_read` is also specified for the
1668 .. option:: sync=bool
1670 Use synchronous I/O for buffered writes. For the majority of I/O engines,
1671 this means using O_SYNC. Default: false.
1673 .. option:: iomem=str, mem=str
1675 Fio can use various types of memory as the I/O unit buffer. The allowed
1679 Use memory from :manpage:`malloc(3)` as the buffers. Default memory
1683 Use shared memory as the buffers. Allocated through
1684 :manpage:`shmget(2)`.
1687 Same as shm, but use huge pages as backing.
1690 Use :manpage:`mmap(2)` to allocate buffers. May either be anonymous memory, or can
1691 be file backed if a filename is given after the option. The format
1692 is `mem=mmap:/path/to/file`.
1695 Use a memory mapped huge file as the buffer backing. Append filename
1696 after mmaphuge, ala `mem=mmaphuge:/hugetlbfs/file`.
1699 Same as mmap, but use a MMAP_SHARED mapping.
1702 Use GPU memory as the buffers for GPUDirect RDMA benchmark.
1703 The :option:`ioengine` must be `rdma`.
1705 The area allocated is a function of the maximum allowed bs size for the job,
1706 multiplied by the I/O depth given. Note that for **shmhuge** and
1707 **mmaphuge** to work, the system must have free huge pages allocated. This
1708 can normally be checked and set by reading/writing
1709 :file:`/proc/sys/vm/nr_hugepages` on a Linux system. Fio assumes a huge page
1710 is 4MiB in size. So to calculate the number of huge pages you need for a
1711 given job file, add up the I/O depth of all jobs (normally one unless
1712 :option:`iodepth` is used) and multiply by the maximum bs set. Then divide
1713 that number by the huge page size. You can see the size of the huge pages in
1714 :file:`/proc/meminfo`. If no huge pages are allocated by having a non-zero
1715 number in `nr_hugepages`, using **mmaphuge** or **shmhuge** will fail. Also
1716 see :option:`hugepage-size`.
1718 **mmaphuge** also needs to have hugetlbfs mounted and the file location
1719 should point there. So if it's mounted in :file:`/huge`, you would use
1720 `mem=mmaphuge:/huge/somefile`.
1722 .. option:: iomem_align=int, mem_align=int
1724 This indicates the memory alignment of the I/O memory buffers. Note that
1725 the given alignment is applied to the first I/O unit buffer, if using
1726 :option:`iodepth` the alignment of the following buffers are given by the
1727 :option:`bs` used. In other words, if using a :option:`bs` that is a
1728 multiple of the page sized in the system, all buffers will be aligned to
1729 this value. If using a :option:`bs` that is not page aligned, the alignment
1730 of subsequent I/O memory buffers is the sum of the :option:`iomem_align` and
1733 .. option:: hugepage-size=int
1735 Defines the size of a huge page. Must at least be equal to the system
1736 setting, see :file:`/proc/meminfo`. Defaults to 4MiB. Should probably
1737 always be a multiple of megabytes, so using ``hugepage-size=Xm`` is the
1738 preferred way to set this to avoid setting a non-pow-2 bad value.
1740 .. option:: lockmem=int
1742 Pin the specified amount of memory with :manpage:`mlock(2)`. Can be used to
1743 simulate a smaller amount of memory. The amount specified is per worker.
1749 .. option:: size=int
1751 The total size of file I/O for each thread of this job. Fio will run until
1752 this many bytes has been transferred, unless runtime is limited by other options
1753 (such as :option:`runtime`, for instance, or increased/decreased by :option:`io_size`).
1754 Fio will divide this size between the available files determined by options
1755 such as :option:`nrfiles`, :option:`filename`, unless :option:`filesize` is
1756 specified by the job. If the result of division happens to be 0, the size is
1757 set to the physical size of the given files or devices if they exist.
1758 If this option is not specified, fio will use the full size of the given
1759 files or devices. If the files do not exist, size must be given. It is also
1760 possible to give size as a percentage between 1 and 100. If ``size=20%`` is
1761 given, fio will use 20% of the full size of the given files or devices.
1762 Can be combined with :option:`offset` to constrain the start and end range
1763 that I/O will be done within.
1765 .. option:: io_size=int, io_limit=int
1767 Normally fio operates within the region set by :option:`size`, which means
1768 that the :option:`size` option sets both the region and size of I/O to be
1769 performed. Sometimes that is not what you want. With this option, it is
1770 possible to define just the amount of I/O that fio should do. For instance,
1771 if :option:`size` is set to 20GiB and :option:`io_size` is set to 5GiB, fio
1772 will perform I/O within the first 20GiB but exit when 5GiB have been
1773 done. The opposite is also possible -- if :option:`size` is set to 20GiB,
1774 and :option:`io_size` is set to 40GiB, then fio will do 40GiB of I/O within
1775 the 0..20GiB region.
1777 .. option:: filesize=irange(int)
1779 Individual file sizes. May be a range, in which case fio will select sizes
1780 for files at random within the given range and limited to :option:`size` in
1781 total (if that is given). If not given, each created file is the same size.
1782 This option overrides :option:`size` in terms of file size, which means
1783 this value is used as a fixed size or possible range of each file.
1785 .. option:: file_append=bool
1787 Perform I/O after the end of the file. Normally fio will operate within the
1788 size of a file. If this option is set, then fio will append to the file
1789 instead. This has identical behavior to setting :option:`offset` to the size
1790 of a file. This option is ignored on non-regular files.
1792 .. option:: fill_device=bool, fill_fs=bool
1794 Sets size to something really large and waits for ENOSPC (no space left on
1795 device) as the terminating condition. Only makes sense with sequential
1796 write. For a read workload, the mount point will be filled first then I/O
1797 started on the result. This option doesn't make sense if operating on a raw
1798 device node, since the size of that is already known by the file system.
1799 Additionally, writing beyond end-of-device will not return ENOSPC there.
1805 .. option:: ioengine=str
1807 Defines how the job issues I/O to the file. The following types are defined:
1810 Basic :manpage:`read(2)` or :manpage:`write(2)`
1811 I/O. :manpage:`lseek(2)` is used to position the I/O location.
1812 See :option:`fsync` and :option:`fdatasync` for syncing write I/Os.
1815 Basic :manpage:`pread(2)` or :manpage:`pwrite(2)` I/O. Default on
1816 all supported operating systems except for Windows.
1819 Basic :manpage:`readv(2)` or :manpage:`writev(2)` I/O. Will emulate
1820 queuing by coalescing adjacent I/Os into a single submission.
1823 Basic :manpage:`preadv(2)` or :manpage:`pwritev(2)` I/O.
1826 Basic :manpage:`preadv2(2)` or :manpage:`pwritev2(2)` I/O.
1829 Fast Linux native asynchronous I/O. Supports async IO
1830 for both direct and buffered IO.
1831 This engine defines engine specific options.
1834 Linux native asynchronous I/O. Note that Linux may only support
1835 queued behavior with non-buffered I/O (set ``direct=1`` or
1837 This engine defines engine specific options.
1840 POSIX asynchronous I/O using :manpage:`aio_read(3)` and
1841 :manpage:`aio_write(3)`.
1844 Solaris native asynchronous I/O.
1847 Windows native asynchronous I/O. Default on Windows.
1850 File is memory mapped with :manpage:`mmap(2)` and data copied
1851 to/from using :manpage:`memcpy(3)`.
1854 :manpage:`splice(2)` is used to transfer the data and
1855 :manpage:`vmsplice(2)` to transfer data from user space to the
1859 SCSI generic sg v3 I/O. May either be synchronous using the SG_IO
1860 ioctl, or if the target is an sg character device we use
1861 :manpage:`read(2)` and :manpage:`write(2)` for asynchronous
1862 I/O. Requires :option:`filename` option to specify either block or
1863 character devices. This engine supports trim operations.
1864 The sg engine includes engine specific options.
1867 Doesn't transfer any data, just pretends to. This is mainly used to
1868 exercise fio itself and for debugging/testing purposes.
1871 Transfer over the network to given ``host:port``. Depending on the
1872 :option:`protocol` used, the :option:`hostname`, :option:`port`,
1873 :option:`listen` and :option:`filename` options are used to specify
1874 what sort of connection to make, while the :option:`protocol` option
1875 determines which protocol will be used. This engine defines engine
1879 Like **net**, but uses :manpage:`splice(2)` and
1880 :manpage:`vmsplice(2)` to map data and send/receive.
1881 This engine defines engine specific options.
1884 Doesn't transfer any data, but burns CPU cycles according to the
1885 :option:`cpuload` and :option:`cpuchunks` options. Setting
1886 :option:`cpuload`\=85 will cause that job to do nothing but burn 85%
1887 of the CPU. In case of SMP machines, use :option:`numjobs`\=<nr_of_cpu>
1888 to get desired CPU usage, as the cpuload only loads a
1889 single CPU at the desired rate. A job never finishes unless there is
1890 at least one non-cpuio job.
1893 The GUASI I/O engine is the Generic Userspace Asynchronous Syscall
1894 Interface approach to async I/O. See
1896 http://www.xmailserver.org/guasi-lib.html
1898 for more info on GUASI.
1901 The RDMA I/O engine supports both RDMA memory semantics
1902 (RDMA_WRITE/RDMA_READ) and channel semantics (Send/Recv) for the
1903 InfiniBand, RoCE and iWARP protocols. This engine defines engine
1907 I/O engine that does regular fallocate to simulate data transfer as
1911 does fallocate(,mode = FALLOC_FL_KEEP_SIZE,).
1914 does fallocate(,mode = 0).
1917 does fallocate(,mode = FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE).
1920 I/O engine that sends :manpage:`ftruncate(2)` operations in response
1921 to write (DDIR_WRITE) events. Each ftruncate issued sets the file's
1922 size to the current block offset. :option:`blocksize` is ignored.
1925 I/O engine that does regular EXT4_IOC_MOVE_EXT ioctls to simulate
1926 defragment activity in request to DDIR_WRITE event.
1929 I/O engine supporting direct access to Ceph Reliable Autonomic
1930 Distributed Object Store (RADOS) via librados. This ioengine
1931 defines engine specific options.
1934 I/O engine supporting direct access to Ceph Rados Block Devices
1935 (RBD) via librbd without the need to use the kernel rbd driver. This
1936 ioengine defines engine specific options.
1939 I/O engine supporting GET/PUT requests over HTTP(S) with libcurl to
1940 a WebDAV or S3 endpoint. This ioengine defines engine specific options.
1942 This engine only supports direct IO of iodepth=1; you need to scale this
1943 via numjobs. blocksize defines the size of the objects to be created.
1945 TRIM is translated to object deletion.
1948 Using GlusterFS libgfapi sync interface to direct access to
1949 GlusterFS volumes without having to go through FUSE. This ioengine
1950 defines engine specific options.
1953 Using GlusterFS libgfapi async interface to direct access to
1954 GlusterFS volumes without having to go through FUSE. This ioengine
1955 defines engine specific options.
1958 Read and write through Hadoop (HDFS). The :option:`filename` option
1959 is used to specify host,port of the hdfs name-node to connect. This
1960 engine interprets offsets a little differently. In HDFS, files once
1961 created cannot be modified so random writes are not possible. To
1962 imitate this the libhdfs engine expects a bunch of small files to be
1963 created over HDFS and will randomly pick a file from them
1964 based on the offset generated by fio backend (see the example
1965 job file to create such files, use ``rw=write`` option). Please
1966 note, it may be necessary to set environment variables to work
1967 with HDFS/libhdfs properly. Each job uses its own connection to
1971 Read, write and erase an MTD character device (e.g.,
1972 :file:`/dev/mtd0`). Discards are treated as erases. Depending on the
1973 underlying device type, the I/O may have to go in a certain pattern,
1974 e.g., on NAND, writing sequentially to erase blocks and discarding
1975 before overwriting. The `trimwrite` mode works well for this
1979 Read and write using filesystem DAX to a file on a filesystem
1980 mounted with DAX on a persistent memory device through the PMDK
1984 Read and write using device DAX to a persistent memory device (e.g.,
1985 /dev/dax0.0) through the PMDK libpmem library.
1988 Prefix to specify loading an external I/O engine object file. Append
1989 the engine filename, e.g. ``ioengine=external:/tmp/foo.o`` to load
1990 ioengine :file:`foo.o` in :file:`/tmp`. The path can be either
1991 absolute or relative. See :file:`engines/skeleton_external.c` for
1992 details of writing an external I/O engine.
1995 Simply create the files and do no I/O to them. You still need to
1996 set `filesize` so that all the accounting still occurs, but no
1997 actual I/O will be done other than creating the file.
2000 Read and write using mmap I/O to a file on a filesystem
2001 mounted with DAX on a persistent memory device through the PMDK
2005 Synchronous read and write using DDN's Infinite Memory Engine (IME).
2006 This engine is very basic and issues calls to IME whenever an IO is
2010 Synchronous read and write using DDN's Infinite Memory Engine (IME).
2011 This engine uses iovecs and will try to stack as much IOs as possible
2012 (if the IOs are "contiguous" and the IO depth is not exceeded)
2013 before issuing a call to IME.
2016 Asynchronous read and write using DDN's Infinite Memory Engine (IME).
2017 This engine will try to stack as much IOs as possible by creating
2018 requests for IME. FIO will then decide when to commit these requests.
2020 Read and write iscsi lun with libiscsi.
2022 Read and write a Network Block Device (NBD).
2024 I/O engine specific parameters
2025 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2027 In addition, there are some parameters which are only valid when a specific
2028 :option:`ioengine` is in use. These are used identically to normal parameters,
2029 with the caveat that when used on the command line, they must come after the
2030 :option:`ioengine` that defines them is selected.
2032 .. option:: hipri : [io_uring]
2034 If this option is set, fio will attempt to use polled IO completions.
2035 Normal IO completions generate interrupts to signal the completion of
2036 IO, polled completions do not. Hence they are require active reaping
2037 by the application. The benefits are more efficient IO for high IOPS
2038 scenarios, and lower latencies for low queue depth IO.
2040 .. option:: fixedbufs : [io_uring]
2042 If fio is asked to do direct IO, then Linux will map pages for each
2043 IO call, and release them when IO is done. If this option is set, the
2044 pages are pre-mapped before IO is started. This eliminates the need to
2045 map and release for each IO. This is more efficient, and reduces the
2048 .. option:: registerfiles : [io_uring]
2049 With this option, fio registers the set of files being used with the
2050 kernel. This avoids the overhead of managing file counts in the kernel,
2051 making the submission and completion part more lightweight. Required
2052 for the below :option:`sqthread_poll` option.
2054 .. option:: sqthread_poll : [io_uring]
2056 Normally fio will submit IO by issuing a system call to notify the
2057 kernel of available items in the SQ ring. If this option is set, the
2058 act of submitting IO will be done by a polling thread in the kernel.
2059 This frees up cycles for fio, at the cost of using more CPU in the
2062 .. option:: sqthread_poll_cpu : [io_uring]
2064 When :option:`sqthread_poll` is set, this option provides a way to
2065 define which CPU should be used for the polling thread.
2067 .. option:: userspace_reap : [libaio]
2069 Normally, with the libaio engine in use, fio will use the
2070 :manpage:`io_getevents(2)` system call to reap newly returned events. With
2071 this flag turned on, the AIO ring will be read directly from user-space to
2072 reap events. The reaping mode is only enabled when polling for a minimum of
2073 0 events (e.g. when :option:`iodepth_batch_complete` `=0`).
2075 .. option:: hipri : [pvsync2]
2077 Set RWF_HIPRI on I/O, indicating to the kernel that it's of higher priority
2080 .. option:: hipri_percentage : [pvsync2]
2082 When hipri is set this determines the probability of a pvsync2 I/O being high
2083 priority. The default is 100%.
2085 .. option:: cpuload=int : [cpuio]
2087 Attempt to use the specified percentage of CPU cycles. This is a mandatory
2088 option when using cpuio I/O engine.
2090 .. option:: cpuchunks=int : [cpuio]
2092 Split the load into cycles of the given time. In microseconds.
2094 .. option:: exit_on_io_done=bool : [cpuio]
2096 Detect when I/O threads are done, then exit.
2098 .. option:: namenode=str : [libhdfs]
2100 The hostname or IP address of a HDFS cluster namenode to contact.
2102 .. option:: port=int
2106 The listening port of the HFDS cluster namenode.
2110 The TCP or UDP port to bind to or connect to. If this is used with
2111 :option:`numjobs` to spawn multiple instances of the same job type, then
2112 this will be the starting port number since fio will use a range of
2117 The port to use for RDMA-CM communication. This should be the same value
2118 on the client and the server side.
2120 .. option:: hostname=str : [netsplice] [net] [rdma]
2122 The hostname or IP address to use for TCP, UDP or RDMA-CM based I/O. If the job
2123 is a TCP listener or UDP reader, the hostname is not used and must be omitted
2124 unless it is a valid UDP multicast address.
2126 .. option:: interface=str : [netsplice] [net]
2128 The IP address of the network interface used to send or receive UDP
2131 .. option:: ttl=int : [netsplice] [net]
2133 Time-to-live value for outgoing UDP multicast packets. Default: 1.
2135 .. option:: nodelay=bool : [netsplice] [net]
2137 Set TCP_NODELAY on TCP connections.
2139 .. option:: protocol=str, proto=str : [netsplice] [net]
2141 The network protocol to use. Accepted values are:
2144 Transmission control protocol.
2146 Transmission control protocol V6.
2148 User datagram protocol.
2150 User datagram protocol V6.
2154 When the protocol is TCP or UDP, the port must also be given, as well as the
2155 hostname if the job is a TCP listener or UDP reader. For unix sockets, the
2156 normal :option:`filename` option should be used and the port is invalid.
2158 .. option:: listen : [netsplice] [net]
2160 For TCP network connections, tell fio to listen for incoming connections
2161 rather than initiating an outgoing connection. The :option:`hostname` must
2162 be omitted if this option is used.
2164 .. option:: pingpong : [netsplice] [net]
2166 Normally a network writer will just continue writing data, and a network
2167 reader will just consume packages. If ``pingpong=1`` is set, a writer will
2168 send its normal payload to the reader, then wait for the reader to send the
2169 same payload back. This allows fio to measure network latencies. The
2170 submission and completion latencies then measure local time spent sending or
2171 receiving, and the completion latency measures how long it took for the
2172 other end to receive and send back. For UDP multicast traffic
2173 ``pingpong=1`` should only be set for a single reader when multiple readers
2174 are listening to the same address.
2176 .. option:: window_size : [netsplice] [net]
2178 Set the desired socket buffer size for the connection.
2180 .. option:: mss : [netsplice] [net]
2182 Set the TCP maximum segment size (TCP_MAXSEG).
2184 .. option:: donorname=str : [e4defrag]
2186 File will be used as a block donor (swap extents between files).
2188 .. option:: inplace=int : [e4defrag]
2190 Configure donor file blocks allocation strategy:
2193 Default. Preallocate donor's file on init.
2195 Allocate space immediately inside defragment event, and free right
2198 .. option:: clustername=str : [rbd,rados]
2200 Specifies the name of the Ceph cluster.
2202 .. option:: rbdname=str : [rbd]
2204 Specifies the name of the RBD.
2206 .. option:: pool=str : [rbd,rados]
2208 Specifies the name of the Ceph pool containing RBD or RADOS data.
2210 .. option:: clientname=str : [rbd,rados]
2212 Specifies the username (without the 'client.' prefix) used to access the
2213 Ceph cluster. If the *clustername* is specified, the *clientname* shall be
2214 the full *type.id* string. If no type. prefix is given, fio will add
2215 'client.' by default.
2217 .. option:: busy_poll=bool : [rbd,rados]
2219 Poll store instead of waiting for completion. Usually this provides better
2220 throughput at cost of higher(up to 100%) CPU utilization.
2222 .. option:: skip_bad=bool : [mtd]
2224 Skip operations against known bad blocks.
2226 .. option:: hdfsdirectory : [libhdfs]
2228 libhdfs will create chunk in this HDFS directory.
2230 .. option:: chunk_size : [libhdfs]
2232 The size of the chunk to use for each file.
2234 .. option:: verb=str : [rdma]
2236 The RDMA verb to use on this side of the RDMA ioengine connection. Valid
2237 values are write, read, send and recv. These correspond to the equivalent
2238 RDMA verbs (e.g. write = rdma_write etc.). Note that this only needs to be
2239 specified on the client side of the connection. See the examples folder.
2241 .. option:: bindname=str : [rdma]
2243 The name to use to bind the local RDMA-CM connection to a local RDMA device.
2244 This could be a hostname or an IPv4 or IPv6 address. On the server side this
2245 will be passed into the rdma_bind_addr() function and on the client site it
2246 will be used in the rdma_resolve_add() function. This can be useful when
2247 multiple paths exist between the client and the server or in certain loopback
2250 .. option:: readfua=bool : [sg]
2252 With readfua option set to 1, read operations include
2253 the force unit access (fua) flag. Default is 0.
2255 .. option:: writefua=bool : [sg]
2257 With writefua option set to 1, write operations include
2258 the force unit access (fua) flag. Default is 0.
2260 .. option:: sg_write_mode=str : [sg]
2262 Specify the type of write commands to issue. This option can take three values:
2265 This is the default where write opcodes are issued as usual.
2267 Issue WRITE AND VERIFY commands. The BYTCHK bit is set to 0. This
2268 directs the device to carry out a medium verification with no data
2269 comparison. The writefua option is ignored with this selection.
2271 Issue WRITE SAME commands. This transfers a single block to the device
2272 and writes this same block of data to a contiguous sequence of LBAs
2273 beginning at the specified offset. fio's block size parameter specifies
2274 the amount of data written with each command. However, the amount of data
2275 actually transferred to the device is equal to the device's block
2276 (sector) size. For a device with 512 byte sectors, blocksize=8k will
2277 write 16 sectors with each command. fio will still generate 8k of data
2278 for each command but only the first 512 bytes will be used and
2279 transferred to the device. The writefua option is ignored with this
2282 .. option:: http_host=str : [http]
2284 Hostname to connect to. For S3, this could be the bucket hostname.
2285 Default is **localhost**
2287 .. option:: http_user=str : [http]
2289 Username for HTTP authentication.
2291 .. option:: http_pass=str : [http]
2293 Password for HTTP authentication.
2295 .. option:: https=str : [http]
2297 Enable HTTPS instead of http. *on* enables HTTPS; *insecure*
2298 will enable HTTPS, but disable SSL peer verification (use with
2299 caution!). Default is **off**
2301 .. option:: http_mode=str : [http]
2303 Which HTTP access mode to use: *webdav*, *swift*, or *s3*.
2304 Default is **webdav**
2306 .. option:: http_s3_region=str : [http]
2308 The S3 region/zone string.
2309 Default is **us-east-1**
2311 .. option:: http_s3_key=str : [http]
2315 .. option:: http_s3_keyid=str : [http]
2317 The S3 key/access id.
2319 .. option:: http_swift_auth_token=str : [http]
2321 The Swift auth token. See the example configuration file on how
2324 .. option:: http_verbose=int : [http]
2326 Enable verbose requests from libcurl. Useful for debugging. 1
2327 turns on verbose logging from libcurl, 2 additionally enables
2328 HTTP IO tracing. Default is **0**
2330 .. option:: uri=str : [nbd]
2332 Specify the NBD URI of the server to test. The string
2333 is a standard NBD URI
2334 (see https://github.com/NetworkBlockDevice/nbd/tree/master/doc).
2335 Example URIs: nbd://localhost:10809
2336 nbd+unix:///?socket=/tmp/socket
2337 nbds://tlshost/exportname
2342 .. option:: iodepth=int
2344 Number of I/O units to keep in flight against the file. Note that
2345 increasing *iodepth* beyond 1 will not affect synchronous ioengines (except
2346 for small degrees when :option:`verify_async` is in use). Even async
2347 engines may impose OS restrictions causing the desired depth not to be
2348 achieved. This may happen on Linux when using libaio and not setting
2349 :option:`direct`\=1, since buffered I/O is not async on that OS. Keep an
2350 eye on the I/O depth distribution in the fio output to verify that the
2351 achieved depth is as expected. Default: 1.
2353 .. option:: iodepth_batch_submit=int, iodepth_batch=int
2355 This defines how many pieces of I/O to submit at once. It defaults to 1
2356 which means that we submit each I/O as soon as it is available, but can be
2357 raised to submit bigger batches of I/O at the time. If it is set to 0 the
2358 :option:`iodepth` value will be used.
2360 .. option:: iodepth_batch_complete_min=int, iodepth_batch_complete=int
2362 This defines how many pieces of I/O to retrieve at once. It defaults to 1
2363 which means that we'll ask for a minimum of 1 I/O in the retrieval process
2364 from the kernel. The I/O retrieval will go on until we hit the limit set by
2365 :option:`iodepth_low`. If this variable is set to 0, then fio will always
2366 check for completed events before queuing more I/O. This helps reduce I/O
2367 latency, at the cost of more retrieval system calls.
2369 .. option:: iodepth_batch_complete_max=int
2371 This defines maximum pieces of I/O to retrieve at once. This variable should
2372 be used along with :option:`iodepth_batch_complete_min`\=int variable,
2373 specifying the range of min and max amount of I/O which should be
2374 retrieved. By default it is equal to the :option:`iodepth_batch_complete_min`
2379 iodepth_batch_complete_min=1
2380 iodepth_batch_complete_max=<iodepth>
2382 which means that we will retrieve at least 1 I/O and up to the whole
2383 submitted queue depth. If none of I/O has been completed yet, we will wait.
2387 iodepth_batch_complete_min=0
2388 iodepth_batch_complete_max=<iodepth>
2390 which means that we can retrieve up to the whole submitted queue depth, but
2391 if none of I/O has been completed yet, we will NOT wait and immediately exit
2392 the system call. In this example we simply do polling.
2394 .. option:: iodepth_low=int
2396 The low water mark indicating when to start filling the queue
2397 again. Defaults to the same as :option:`iodepth`, meaning that fio will
2398 attempt to keep the queue full at all times. If :option:`iodepth` is set to
2399 e.g. 16 and *iodepth_low* is set to 4, then after fio has filled the queue of
2400 16 requests, it will let the depth drain down to 4 before starting to fill
2403 .. option:: serialize_overlap=bool
2405 Serialize in-flight I/Os that might otherwise cause or suffer from data races.
2406 When two or more I/Os are submitted simultaneously, there is no guarantee that
2407 the I/Os will be processed or completed in the submitted order. Further, if
2408 two or more of those I/Os are writes, any overlapping region between them can
2409 become indeterminate/undefined on certain storage. These issues can cause
2410 verification to fail erratically when at least one of the racing I/Os is
2411 changing data and the overlapping region has a non-zero size. Setting
2412 ``serialize_overlap`` tells fio to avoid provoking this behavior by explicitly
2413 serializing in-flight I/Os that have a non-zero overlap. Note that setting
2414 this option can reduce both performance and the :option:`iodepth` achieved.
2416 This option only applies to I/Os issued for a single job except when it is
2417 enabled along with :option:`io_submit_mode`\=offload. In offload mode, fio
2418 will check for overlap among all I/Os submitted by offload jobs with :option:`serialize_overlap`
2423 .. option:: io_submit_mode=str
2425 This option controls how fio submits the I/O to the I/O engine. The default
2426 is `inline`, which means that the fio job threads submit and reap I/O
2427 directly. If set to `offload`, the job threads will offload I/O submission
2428 to a dedicated pool of I/O threads. This requires some coordination and thus
2429 has a bit of extra overhead, especially for lower queue depth I/O where it
2430 can increase latencies. The benefit is that fio can manage submission rates
2431 independently of the device completion rates. This avoids skewed latency
2432 reporting if I/O gets backed up on the device side (the coordinated omission
2439 .. option:: thinktime=time
2441 Stall the job for the specified period of time after an I/O has completed before issuing the
2442 next. May be used to simulate processing being done by an application.
2443 When the unit is omitted, the value is interpreted in microseconds. See
2444 :option:`thinktime_blocks` and :option:`thinktime_spin`.
2446 .. option:: thinktime_spin=time
2448 Only valid if :option:`thinktime` is set - pretend to spend CPU time doing
2449 something with the data received, before falling back to sleeping for the
2450 rest of the period specified by :option:`thinktime`. When the unit is
2451 omitted, the value is interpreted in microseconds.
2453 .. option:: thinktime_blocks=int
2455 Only valid if :option:`thinktime` is set - control how many blocks to issue,
2456 before waiting :option:`thinktime` usecs. If not set, defaults to 1 which will make
2457 fio wait :option:`thinktime` usecs after every block. This effectively makes any
2458 queue depth setting redundant, since no more than 1 I/O will be queued
2459 before we have to complete it and do our :option:`thinktime`. In other words, this
2460 setting effectively caps the queue depth if the latter is larger.
2462 .. option:: rate=int[,int][,int]
2464 Cap the bandwidth used by this job. The number is in bytes/sec, the normal
2465 suffix rules apply. Comma-separated values may be specified for reads,
2466 writes, and trims as described in :option:`blocksize`.
2468 For example, using `rate=1m,500k` would limit reads to 1MiB/sec and writes to
2469 500KiB/sec. Capping only reads or writes can be done with `rate=,500k` or
2470 `rate=500k,` where the former will only limit writes (to 500KiB/sec) and the
2471 latter will only limit reads.
2473 .. option:: rate_min=int[,int][,int]
2475 Tell fio to do whatever it can to maintain at least this bandwidth. Failing
2476 to meet this requirement will cause the job to exit. Comma-separated values
2477 may be specified for reads, writes, and trims as described in
2478 :option:`blocksize`.
2480 .. option:: rate_iops=int[,int][,int]
2482 Cap the bandwidth to this number of IOPS. Basically the same as
2483 :option:`rate`, just specified independently of bandwidth. If the job is
2484 given a block size range instead of a fixed value, the smallest block size
2485 is used as the metric. Comma-separated values may be specified for reads,
2486 writes, and trims as described in :option:`blocksize`.
2488 .. option:: rate_iops_min=int[,int][,int]
2490 If fio doesn't meet this rate of I/O, it will cause the job to exit.
2491 Comma-separated values may be specified for reads, writes, and trims as
2492 described in :option:`blocksize`.
2494 .. option:: rate_process=str
2496 This option controls how fio manages rated I/O submissions. The default is
2497 `linear`, which submits I/O in a linear fashion with fixed delays between
2498 I/Os that gets adjusted based on I/O completion rates. If this is set to
2499 `poisson`, fio will submit I/O based on a more real world random request
2500 flow, known as the Poisson process
2501 (https://en.wikipedia.org/wiki/Poisson_point_process). The lambda will be
2502 10^6 / IOPS for the given workload.
2504 .. option:: rate_ignore_thinktime=bool
2506 By default, fio will attempt to catch up to the specified rate setting,
2507 if any kind of thinktime setting was used. If this option is set, then
2508 fio will ignore the thinktime and continue doing IO at the specified
2509 rate, instead of entering a catch-up mode after thinktime is done.
2515 .. option:: latency_target=time
2517 If set, fio will attempt to find the max performance point that the given
2518 workload will run at while maintaining a latency below this target. When
2519 the unit is omitted, the value is interpreted in microseconds. See
2520 :option:`latency_window` and :option:`latency_percentile`.
2522 .. option:: latency_window=time
2524 Used with :option:`latency_target` to specify the sample window that the job
2525 is run at varying queue depths to test the performance. When the unit is
2526 omitted, the value is interpreted in microseconds.
2528 .. option:: latency_percentile=float
2530 The percentage of I/Os that must fall within the criteria specified by
2531 :option:`latency_target` and :option:`latency_window`. If not set, this
2532 defaults to 100.0, meaning that all I/Os must be equal or below to the value
2533 set by :option:`latency_target`.
2535 .. option:: max_latency=time
2537 If set, fio will exit the job with an ETIMEDOUT error if it exceeds this
2538 maximum latency. When the unit is omitted, the value is interpreted in
2541 .. option:: rate_cycle=int
2543 Average bandwidth for :option:`rate` and :option:`rate_min` over this number
2544 of milliseconds. Defaults to 1000.
2550 .. option:: write_iolog=str
2552 Write the issued I/O patterns to the specified file. See
2553 :option:`read_iolog`. Specify a separate file for each job, otherwise the
2554 iologs will be interspersed and the file may be corrupt.
2556 .. option:: read_iolog=str
2558 Open an iolog with the specified filename and replay the I/O patterns it
2559 contains. This can be used to store a workload and replay it sometime
2560 later. The iolog given may also be a blktrace binary file, which allows fio
2561 to replay a workload captured by :command:`blktrace`. See
2562 :manpage:`blktrace(8)` for how to capture such logging data. For blktrace
2563 replay, the file needs to be turned into a blkparse binary data file first
2564 (``blkparse <device> -o /dev/null -d file_for_fio.bin``).
2565 You can specify a number of files by separating the names with a ':'
2566 character. See the :option:`filename` option for information on how to
2567 escape ':' and '\' characters within the file names. These files will
2568 be sequentially assigned to job clones created by :option:`numjobs`.
2570 .. option:: read_iolog_chunked=bool
2572 Determines how iolog is read. If false(default) entire :option:`read_iolog`
2573 will be read at once. If selected true, input from iolog will be read
2574 gradually. Useful when iolog is very large, or it is generated.
2576 .. option:: merge_blktrace_file=str
2578 When specified, rather than replaying the logs passed to :option:`read_iolog`,
2579 the logs go through a merge phase which aggregates them into a single
2580 blktrace. The resulting file is then passed on as the :option:`read_iolog`
2581 parameter. The intention here is to make the order of events consistent.
2582 This limits the influence of the scheduler compared to replaying multiple
2583 blktraces via concurrent jobs.
2585 .. option:: merge_blktrace_scalars=float_list
2587 This is a percentage based option that is index paired with the list of
2588 files passed to :option:`read_iolog`. When merging is performed, scale
2589 the time of each event by the corresponding amount. For example,
2590 ``--merge_blktrace_scalars="50:100"`` runs the first trace in halftime
2591 and the second trace in realtime. This knob is separately tunable from
2592 :option:`replay_time_scale` which scales the trace during runtime and
2593 does not change the output of the merge unlike this option.
2595 .. option:: merge_blktrace_iters=float_list
2597 This is a whole number option that is index paired with the list of files
2598 passed to :option:`read_iolog`. When merging is performed, run each trace
2599 for the specified number of iterations. For example,
2600 ``--merge_blktrace_iters="2:1"`` runs the first trace for two iterations
2601 and the second trace for one iteration.
2603 .. option:: replay_no_stall=bool
2605 When replaying I/O with :option:`read_iolog` the default behavior is to
2606 attempt to respect the timestamps within the log and replay them with the
2607 appropriate delay between IOPS. By setting this variable fio will not
2608 respect the timestamps and attempt to replay them as fast as possible while
2609 still respecting ordering. The result is the same I/O pattern to a given
2610 device, but different timings.
2612 .. option:: replay_time_scale=int
2614 When replaying I/O with :option:`read_iolog`, fio will honor the
2615 original timing in the trace. With this option, it's possible to scale
2616 the time. It's a percentage option, if set to 50 it means run at 50%
2617 the original IO rate in the trace. If set to 200, run at twice the
2618 original IO rate. Defaults to 100.
2620 .. option:: replay_redirect=str
2622 While replaying I/O patterns using :option:`read_iolog` the default behavior
2623 is to replay the IOPS onto the major/minor device that each IOP was recorded
2624 from. This is sometimes undesirable because on a different machine those
2625 major/minor numbers can map to a different device. Changing hardware on the
2626 same system can also result in a different major/minor mapping.
2627 ``replay_redirect`` causes all I/Os to be replayed onto the single specified
2628 device regardless of the device it was recorded
2629 from. i.e. :option:`replay_redirect`\= :file:`/dev/sdc` would cause all I/O
2630 in the blktrace or iolog to be replayed onto :file:`/dev/sdc`. This means
2631 multiple devices will be replayed onto a single device, if the trace
2632 contains multiple devices. If you want multiple devices to be replayed
2633 concurrently to multiple redirected devices you must blkparse your trace
2634 into separate traces and replay them with independent fio invocations.
2635 Unfortunately this also breaks the strict time ordering between multiple
2638 .. option:: replay_align=int
2640 Force alignment of the byte offsets in a trace to this value. The value
2641 must be a power of 2.
2643 .. option:: replay_scale=int
2645 Scale byte offsets down by this factor when replaying traces. Should most
2646 likely use :option:`replay_align` as well.
2648 .. option:: replay_skip=str
2650 Sometimes it's useful to skip certain IO types in a replay trace.
2651 This could be, for instance, eliminating the writes in the trace.
2652 Or not replaying the trims/discards, if you are redirecting to
2653 a device that doesn't support them. This option takes a comma
2654 separated list of read, write, trim, sync.
2657 Threads, processes and job synchronization
2658 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2662 Fio defaults to creating jobs by using fork, however if this option is
2663 given, fio will create jobs by using POSIX Threads' function
2664 :manpage:`pthread_create(3)` to create threads instead.
2666 .. option:: wait_for=str
2668 If set, the current job won't be started until all workers of the specified
2669 waitee job are done.
2671 ``wait_for`` operates on the job name basis, so there are a few
2672 limitations. First, the waitee must be defined prior to the waiter job
2673 (meaning no forward references). Second, if a job is being referenced as a
2674 waitee, it must have a unique name (no duplicate waitees).
2676 .. option:: nice=int
2678 Run the job with the given nice value. See man :manpage:`nice(2)`.
2680 On Windows, values less than -15 set the process class to "High"; -1 through
2681 -15 set "Above Normal"; 1 through 15 "Below Normal"; and above 15 "Idle"
2684 .. option:: prio=int
2686 Set the I/O priority value of this job. Linux limits us to a positive value
2687 between 0 and 7, with 0 being the highest. See man
2688 :manpage:`ionice(1)`. Refer to an appropriate manpage for other operating
2689 systems since meaning of priority may differ.
2691 .. option:: prioclass=int
2693 Set the I/O priority class. See man :manpage:`ionice(1)`.
2695 .. option:: cpus_allowed=str
2697 Controls the same options as :option:`cpumask`, but accepts a textual
2698 specification of the permitted CPUs instead and CPUs are indexed from 0. So
2699 to use CPUs 0 and 5 you would specify ``cpus_allowed=0,5``. This option also
2700 allows a range of CPUs to be specified -- say you wanted a binding to CPUs
2701 0, 5, and 8 to 15, you would set ``cpus_allowed=0,5,8-15``.
2703 On Windows, when ``cpus_allowed`` is unset only CPUs from fio's current
2704 processor group will be used and affinity settings are inherited from the
2705 system. An fio build configured to target Windows 7 makes options that set
2706 CPUs processor group aware and values will set both the processor group
2707 and a CPU from within that group. For example, on a system where processor
2708 group 0 has 40 CPUs and processor group 1 has 32 CPUs, ``cpus_allowed``
2709 values between 0 and 39 will bind CPUs from processor group 0 and
2710 ``cpus_allowed`` values between 40 and 71 will bind CPUs from processor
2711 group 1. When using ``cpus_allowed_policy=shared`` all CPUs specified by a
2712 single ``cpus_allowed`` option must be from the same processor group. For
2713 Windows fio builds not built for Windows 7, CPUs will only be selected from
2714 (and be relative to) whatever processor group fio happens to be running in
2715 and CPUs from other processor groups cannot be used.
2717 .. option:: cpus_allowed_policy=str
2719 Set the policy of how fio distributes the CPUs specified by
2720 :option:`cpus_allowed` or :option:`cpumask`. Two policies are supported:
2723 All jobs will share the CPU set specified.
2725 Each job will get a unique CPU from the CPU set.
2727 **shared** is the default behavior, if the option isn't specified. If
2728 **split** is specified, then fio will assign one cpu per job. If not
2729 enough CPUs are given for the jobs listed, then fio will roundrobin the CPUs
2732 .. option:: cpumask=int
2734 Set the CPU affinity of this job. The parameter given is a bit mask of
2735 allowed CPUs the job may run on. So if you want the allowed CPUs to be 1
2736 and 5, you would pass the decimal value of (1 << 1 | 1 << 5), or 34. See man
2737 :manpage:`sched_setaffinity(2)`. This may not work on all supported
2738 operating systems or kernel versions. This option doesn't work well for a
2739 higher CPU count than what you can store in an integer mask, so it can only
2740 control cpus 1-32. For boxes with larger CPU counts, use
2741 :option:`cpus_allowed`.
2743 .. option:: numa_cpu_nodes=str
2745 Set this job running on specified NUMA nodes' CPUs. The arguments allow
2746 comma delimited list of cpu numbers, A-B ranges, or `all`. Note, to enable
2747 NUMA options support, fio must be built on a system with libnuma-dev(el)
2750 .. option:: numa_mem_policy=str
2752 Set this job's memory policy and corresponding NUMA nodes. Format of the
2757 ``mode`` is one of the following memory policies: ``default``, ``prefer``,
2758 ``bind``, ``interleave`` or ``local``. For ``default`` and ``local`` memory
2759 policies, no node needs to be specified. For ``prefer``, only one node is
2760 allowed. For ``bind`` and ``interleave`` the ``nodelist`` may be as
2761 follows: a comma delimited list of numbers, A-B ranges, or `all`.
2763 .. option:: cgroup=str
2765 Add job to this control group. If it doesn't exist, it will be created. The
2766 system must have a mounted cgroup blkio mount point for this to work. If
2767 your system doesn't have it mounted, you can do so with::
2769 # mount -t cgroup -o blkio none /cgroup
2771 .. option:: cgroup_weight=int
2773 Set the weight of the cgroup to this value. See the documentation that comes
2774 with the kernel, allowed values are in the range of 100..1000.
2776 .. option:: cgroup_nodelete=bool
2778 Normally fio will delete the cgroups it has created after the job
2779 completion. To override this behavior and to leave cgroups around after the
2780 job completion, set ``cgroup_nodelete=1``. This can be useful if one wants
2781 to inspect various cgroup files after job completion. Default: false.
2783 .. option:: flow_id=int
2785 The ID of the flow. If not specified, it defaults to being a global
2786 flow. See :option:`flow`.
2788 .. option:: flow=int
2790 Weight in token-based flow control. If this value is used, then there is a
2791 'flow counter' which is used to regulate the proportion of activity between
2792 two or more jobs. Fio attempts to keep this flow counter near zero. The
2793 ``flow`` parameter stands for how much should be added or subtracted to the
2794 flow counter on each iteration of the main I/O loop. That is, if one job has
2795 ``flow=8`` and another job has ``flow=-1``, then there will be a roughly 1:8
2796 ratio in how much one runs vs the other.
2798 .. option:: flow_watermark=int
2800 The maximum value that the absolute value of the flow counter is allowed to
2801 reach before the job must wait for a lower value of the counter.
2803 .. option:: flow_sleep=int
2805 The period of time, in microseconds, to wait after the flow watermark has
2806 been exceeded before retrying operations.
2808 .. option:: stonewall, wait_for_previous
2810 Wait for preceding jobs in the job file to exit, before starting this
2811 one. Can be used to insert serialization points in the job file. A stone
2812 wall also implies starting a new reporting group, see
2813 :option:`group_reporting`.
2817 By default, fio will continue running all other jobs when one job finishes.
2818 Sometimes this is not the desired action. Setting ``exitall`` will instead
2819 make fio terminate all jobs in the same group, as soon as one job of that
2822 .. option:: exit_what
2824 By default, fio will continue running all other jobs when one job finishes.
2825 Sometimes this is not the desired action. Setting ``exit_all`` will
2826 instead make fio terminate all jobs in the same group. The option
2827 ``exit_what`` allows to control which jobs get terminated when ``exitall`` is
2828 enabled. The default is ``group`` and does not change the behaviour of
2829 ``exitall``. The setting ``all`` terminates all jobs. The setting ``stonewall``
2830 terminates all currently running jobs across all groups and continues execution
2831 with the next stonewalled group.
2833 .. option:: exec_prerun=str
2835 Before running this job, issue the command specified through
2836 :manpage:`system(3)`. Output is redirected in a file called
2837 :file:`jobname.prerun.txt`.
2839 .. option:: exec_postrun=str
2841 After the job completes, issue the command specified though
2842 :manpage:`system(3)`. Output is redirected in a file called
2843 :file:`jobname.postrun.txt`.
2847 Instead of running as the invoking user, set the user ID to this value
2848 before the thread/process does any work.
2852 Set group ID, see :option:`uid`.
2858 .. option:: verify_only
2860 Do not perform specified workload, only verify data still matches previous
2861 invocation of this workload. This option allows one to check data multiple
2862 times at a later date without overwriting it. This option makes sense only
2863 for workloads that write data, and does not support workloads with the
2864 :option:`time_based` option set.
2866 .. option:: do_verify=bool
2868 Run the verify phase after a write phase. Only valid if :option:`verify` is
2871 .. option:: verify=str
2873 If writing to a file, fio can verify the file contents after each iteration
2874 of the job. Each verification method also implies verification of special
2875 header, which is written to the beginning of each block. This header also
2876 includes meta information, like offset of the block, block number, timestamp
2877 when block was written, etc. :option:`verify` can be combined with
2878 :option:`verify_pattern` option. The allowed values are:
2881 Use an md5 sum of the data area and store it in the header of
2885 Use an experimental crc64 sum of the data area and store it in the
2886 header of each block.
2889 Use a crc32c sum of the data area and store it in the header of
2890 each block. This will automatically use hardware acceleration
2891 (e.g. SSE4.2 on an x86 or CRC crypto extensions on ARM64) but will
2892 fall back to software crc32c if none is found. Generally the
2893 fastest checksum fio supports when hardware accelerated.
2899 Use a crc32 sum of the data area and store it in the header of each
2903 Use a crc16 sum of the data area and store it in the header of each
2907 Use a crc7 sum of the data area and store it in the header of each
2911 Use xxhash as the checksum function. Generally the fastest software
2912 checksum that fio supports.
2915 Use sha512 as the checksum function.
2918 Use sha256 as the checksum function.
2921 Use optimized sha1 as the checksum function.
2924 Use optimized sha3-224 as the checksum function.
2927 Use optimized sha3-256 as the checksum function.
2930 Use optimized sha3-384 as the checksum function.
2933 Use optimized sha3-512 as the checksum function.
2936 This option is deprecated, since now meta information is included in
2937 generic verification header and meta verification happens by
2938 default. For detailed information see the description of the
2939 :option:`verify` setting. This option is kept because of
2940 compatibility's sake with old configurations. Do not use it.
2943 Verify a strict pattern. Normally fio includes a header with some
2944 basic information and checksumming, but if this option is set, only
2945 the specific pattern set with :option:`verify_pattern` is verified.
2948 Only pretend to verify. Useful for testing internals with
2949 :option:`ioengine`\=null, not for much else.
2951 This option can be used for repeated burn-in tests of a system to make sure
2952 that the written data is also correctly read back. If the data direction
2953 given is a read or random read, fio will assume that it should verify a
2954 previously written file. If the data direction includes any form of write,
2955 the verify will be of the newly written data.
2957 To avoid false verification errors, do not use the norandommap option when
2958 verifying data with async I/O engines and I/O depths > 1. Or use the
2959 norandommap and the lfsr random generator together to avoid writing to the
2960 same offset with muliple outstanding I/Os.
2962 .. option:: verify_offset=int
2964 Swap the verification header with data somewhere else in the block before
2965 writing. It is swapped back before verifying.
2967 .. option:: verify_interval=int
2969 Write the verification header at a finer granularity than the
2970 :option:`blocksize`. It will be written for chunks the size of
2971 ``verify_interval``. :option:`blocksize` should divide this evenly.
2973 .. option:: verify_pattern=str
2975 If set, fio will fill the I/O buffers with this pattern. Fio defaults to
2976 filling with totally random bytes, but sometimes it's interesting to fill
2977 with a known pattern for I/O verification purposes. Depending on the width
2978 of the pattern, fio will fill 1/2/3/4 bytes of the buffer at the time (it can
2979 be either a decimal or a hex number). The ``verify_pattern`` if larger than
2980 a 32-bit quantity has to be a hex number that starts with either "0x" or
2981 "0X". Use with :option:`verify`. Also, ``verify_pattern`` supports %o
2982 format, which means that for each block offset will be written and then
2983 verified back, e.g.::
2987 Or use combination of everything::
2989 verify_pattern=0xff%o"abcd"-12
2991 .. option:: verify_fatal=bool
2993 Normally fio will keep checking the entire contents before quitting on a
2994 block verification failure. If this option is set, fio will exit the job on
2995 the first observed failure. Default: false.
2997 .. option:: verify_dump=bool
2999 If set, dump the contents of both the original data block and the data block
3000 we read off disk to files. This allows later analysis to inspect just what
3001 kind of data corruption occurred. Off by default.
3003 .. option:: verify_async=int
3005 Fio will normally verify I/O inline from the submitting thread. This option
3006 takes an integer describing how many async offload threads to create for I/O
3007 verification instead, causing fio to offload the duty of verifying I/O
3008 contents to one or more separate threads. If using this offload option, even
3009 sync I/O engines can benefit from using an :option:`iodepth` setting higher
3010 than 1, as it allows them to have I/O in flight while verifies are running.
3011 Defaults to 0 async threads, i.e. verification is not asynchronous.
3013 .. option:: verify_async_cpus=str
3015 Tell fio to set the given CPU affinity on the async I/O verification
3016 threads. See :option:`cpus_allowed` for the format used.
3018 .. option:: verify_backlog=int
3020 Fio will normally verify the written contents of a job that utilizes verify
3021 once that job has completed. In other words, everything is written then
3022 everything is read back and verified. You may want to verify continually
3023 instead for a variety of reasons. Fio stores the meta data associated with
3024 an I/O block in memory, so for large verify workloads, quite a bit of memory
3025 would be used up holding this meta data. If this option is enabled, fio will
3026 write only N blocks before verifying these blocks.
3028 .. option:: verify_backlog_batch=int
3030 Control how many blocks fio will verify if :option:`verify_backlog` is
3031 set. If not set, will default to the value of :option:`verify_backlog`
3032 (meaning the entire queue is read back and verified). If
3033 ``verify_backlog_batch`` is less than :option:`verify_backlog` then not all
3034 blocks will be verified, if ``verify_backlog_batch`` is larger than
3035 :option:`verify_backlog`, some blocks will be verified more than once.
3037 .. option:: verify_state_save=bool
3039 When a job exits during the write phase of a verify workload, save its
3040 current state. This allows fio to replay up until that point, if the verify
3041 state is loaded for the verify read phase. The format of the filename is,
3044 <type>-<jobname>-<jobindex>-verify.state.
3046 <type> is "local" for a local run, "sock" for a client/server socket
3047 connection, and "ip" (192.168.0.1, for instance) for a networked
3048 client/server connection. Defaults to true.
3050 .. option:: verify_state_load=bool
3052 If a verify termination trigger was used, fio stores the current write state
3053 of each thread. This can be used at verification time so that fio knows how
3054 far it should verify. Without this information, fio will run a full
3055 verification pass, according to the settings in the job file used. Default
3058 .. option:: trim_percentage=int
3060 Number of verify blocks to discard/trim.
3062 .. option:: trim_verify_zero=bool
3064 Verify that trim/discarded blocks are returned as zeros.
3066 .. option:: trim_backlog=int
3068 Trim after this number of blocks are written.
3070 .. option:: trim_backlog_batch=int
3072 Trim this number of I/O blocks.
3074 .. option:: experimental_verify=bool
3076 Enable experimental verification.
3081 .. option:: steadystate=str:float, ss=str:float
3083 Define the criterion and limit for assessing steady state performance. The
3084 first parameter designates the criterion whereas the second parameter sets
3085 the threshold. When the criterion falls below the threshold for the
3086 specified duration, the job will stop. For example, `iops_slope:0.1%` will
3087 direct fio to terminate the job when the least squares regression slope
3088 falls below 0.1% of the mean IOPS. If :option:`group_reporting` is enabled
3089 this will apply to all jobs in the group. Below is the list of available
3090 steady state assessment criteria. All assessments are carried out using only
3091 data from the rolling collection window. Threshold limits can be expressed
3092 as a fixed value or as a percentage of the mean in the collection window.
3094 When using this feature, most jobs should include the :option:`time_based`
3095 and :option:`runtime` options or the :option:`loops` option so that fio does not
3096 stop running after it has covered the full size of the specified file(s) or device(s).
3099 Collect IOPS data. Stop the job if all individual IOPS measurements
3100 are within the specified limit of the mean IOPS (e.g., ``iops:2``
3101 means that all individual IOPS values must be within 2 of the mean,
3102 whereas ``iops:0.2%`` means that all individual IOPS values must be
3103 within 0.2% of the mean IOPS to terminate the job).
3106 Collect IOPS data and calculate the least squares regression
3107 slope. Stop the job if the slope falls below the specified limit.
3110 Collect bandwidth data. Stop the job if all individual bandwidth
3111 measurements are within the specified limit of the mean bandwidth.
3114 Collect bandwidth data and calculate the least squares regression
3115 slope. Stop the job if the slope falls below the specified limit.
3117 .. option:: steadystate_duration=time, ss_dur=time
3119 A rolling window of this duration will be used to judge whether steady state
3120 has been reached. Data will be collected once per second. The default is 0
3121 which disables steady state detection. When the unit is omitted, the
3122 value is interpreted in seconds.
3124 .. option:: steadystate_ramp_time=time, ss_ramp=time
3126 Allow the job to run for the specified duration before beginning data
3127 collection for checking the steady state job termination criterion. The
3128 default is 0. When the unit is omitted, the value is interpreted in seconds.
3131 Measurements and reporting
3132 ~~~~~~~~~~~~~~~~~~~~~~~~~~
3134 .. option:: per_job_logs=bool
3136 If set, this generates bw/clat/iops log with per file private filenames. If
3137 not set, jobs with identical names will share the log filename. Default:
3140 .. option:: group_reporting
3142 It may sometimes be interesting to display statistics for groups of jobs as
3143 a whole instead of for each individual job. This is especially true if
3144 :option:`numjobs` is used; looking at individual thread/process output
3145 quickly becomes unwieldy. To see the final report per-group instead of
3146 per-job, use :option:`group_reporting`. Jobs in a file will be part of the
3147 same reporting group, unless if separated by a :option:`stonewall`, or by
3148 using :option:`new_group`.
3150 .. option:: new_group
3152 Start a new reporting group. See: :option:`group_reporting`. If not given,
3153 all jobs in a file will be part of the same reporting group, unless
3154 separated by a :option:`stonewall`.
3156 .. option:: stats=bool
3158 By default, fio collects and shows final output results for all jobs
3159 that run. If this option is set to 0, then fio will ignore it in
3160 the final stat output.
3162 .. option:: write_bw_log=str
3164 If given, write a bandwidth log for this job. Can be used to store data of
3165 the bandwidth of the jobs in their lifetime.
3167 If no str argument is given, the default filename of
3168 :file:`jobname_type.x.log` is used. Even when the argument is given, fio
3169 will still append the type of log. So if one specifies::
3173 The actual log name will be :file:`foo_bw.x.log` where `x` is the index
3174 of the job (`1..N`, where `N` is the number of jobs). If
3175 :option:`per_job_logs` is false, then the filename will not include the
3178 The included :command:`fio_generate_plots` script uses :command:`gnuplot` to turn these
3179 text files into nice graphs. See `Log File Formats`_ for how data is
3180 structured within the file.
3182 .. option:: write_lat_log=str
3184 Same as :option:`write_bw_log`, except this option creates I/O
3185 submission (e.g., :file:`name_slat.x.log`), completion (e.g.,
3186 :file:`name_clat.x.log`), and total (e.g., :file:`name_lat.x.log`)
3187 latency files instead. See :option:`write_bw_log` for details about
3188 the filename format and `Log File Formats`_ for how data is structured
3191 .. option:: write_hist_log=str
3193 Same as :option:`write_bw_log` but writes an I/O completion latency
3194 histogram file (e.g., :file:`name_hist.x.log`) instead. Note that this
3195 file will be empty unless :option:`log_hist_msec` has also been set.
3196 See :option:`write_bw_log` for details about the filename format and
3197 `Log File Formats`_ for how data is structured within the file.
3199 .. option:: write_iops_log=str
3201 Same as :option:`write_bw_log`, but writes an IOPS file (e.g.
3202 :file:`name_iops.x.log`) instead. Because fio defaults to individual
3203 I/O logging, the value entry in the IOPS log will be 1 unless windowed
3204 logging (see :option:`log_avg_msec`) has been enabled. See
3205 :option:`write_bw_log` for details about the filename format and `Log
3206 File Formats`_ for how data is structured within the file.
3208 .. option:: log_avg_msec=int
3210 By default, fio will log an entry in the iops, latency, or bw log for every
3211 I/O that completes. When writing to the disk log, that can quickly grow to a
3212 very large size. Setting this option makes fio average the each log entry
3213 over the specified period of time, reducing the resolution of the log. See
3214 :option:`log_max_value` as well. Defaults to 0, logging all entries.
3215 Also see `Log File Formats`_.
3217 .. option:: log_hist_msec=int
3219 Same as :option:`log_avg_msec`, but logs entries for completion latency
3220 histograms. Computing latency percentiles from averages of intervals using
3221 :option:`log_avg_msec` is inaccurate. Setting this option makes fio log
3222 histogram entries over the specified period of time, reducing log sizes for
3223 high IOPS devices while retaining percentile accuracy. See
3224 :option:`log_hist_coarseness` and :option:`write_hist_log` as well.
3225 Defaults to 0, meaning histogram logging is disabled.
3227 .. option:: log_hist_coarseness=int
3229 Integer ranging from 0 to 6, defining the coarseness of the resolution of
3230 the histogram logs enabled with :option:`log_hist_msec`. For each increment
3231 in coarseness, fio outputs half as many bins. Defaults to 0, for which
3232 histogram logs contain 1216 latency bins. See :option:`write_hist_log`
3233 and `Log File Formats`_.
3235 .. option:: log_max_value=bool
3237 If :option:`log_avg_msec` is set, fio logs the average over that window. If
3238 you instead want to log the maximum value, set this option to 1. Defaults to
3239 0, meaning that averaged values are logged.
3241 .. option:: log_offset=bool
3243 If this is set, the iolog options will include the byte offset for the I/O
3244 entry as well as the other data values. Defaults to 0 meaning that
3245 offsets are not present in logs. Also see `Log File Formats`_.
3247 .. option:: log_compression=int
3249 If this is set, fio will compress the I/O logs as it goes, to keep the
3250 memory footprint lower. When a log reaches the specified size, that chunk is
3251 removed and compressed in the background. Given that I/O logs are fairly
3252 highly compressible, this yields a nice memory savings for longer runs. The
3253 downside is that the compression will consume some background CPU cycles, so
3254 it may impact the run. This, however, is also true if the logging ends up
3255 consuming most of the system memory. So pick your poison. The I/O logs are
3256 saved normally at the end of a run, by decompressing the chunks and storing
3257 them in the specified log file. This feature depends on the availability of
3260 .. option:: log_compression_cpus=str
3262 Define the set of CPUs that are allowed to handle online log compression for
3263 the I/O jobs. This can provide better isolation between performance
3264 sensitive jobs, and background compression work. See
3265 :option:`cpus_allowed` for the format used.
3267 .. option:: log_store_compressed=bool
3269 If set, fio will store the log files in a compressed format. They can be
3270 decompressed with fio, using the :option:`--inflate-log` command line
3271 parameter. The files will be stored with a :file:`.fz` suffix.
3273 .. option:: log_unix_epoch=bool
3275 If set, fio will log Unix timestamps to the log files produced by enabling
3276 write_type_log for each log type, instead of the default zero-based
3279 .. option:: block_error_percentiles=bool
3281 If set, record errors in trim block-sized units from writes and trims and
3282 output a histogram of how many trims it took to get to errors, and what kind
3283 of error was encountered.
3285 .. option:: bwavgtime=int
3287 Average the calculated bandwidth over the given time. Value is specified in
3288 milliseconds. If the job also does bandwidth logging through
3289 :option:`write_bw_log`, then the minimum of this option and
3290 :option:`log_avg_msec` will be used. Default: 500ms.
3292 .. option:: iopsavgtime=int
3294 Average the calculated IOPS over the given time. Value is specified in
3295 milliseconds. If the job also does IOPS logging through
3296 :option:`write_iops_log`, then the minimum of this option and
3297 :option:`log_avg_msec` will be used. Default: 500ms.
3299 .. option:: disk_util=bool
3301 Generate disk utilization statistics, if the platform supports it.
3304 .. option:: disable_lat=bool
3306 Disable measurements of total latency numbers. Useful only for cutting back
3307 the number of calls to :manpage:`gettimeofday(2)`, as that does impact
3308 performance at really high IOPS rates. Note that to really get rid of a
3309 large amount of these calls, this option must be used with
3310 :option:`disable_slat` and :option:`disable_bw_measurement` as well.
3312 .. option:: disable_clat=bool
3314 Disable measurements of completion latency numbers. See
3315 :option:`disable_lat`.
3317 .. option:: disable_slat=bool
3319 Disable measurements of submission latency numbers. See
3320 :option:`disable_lat`.
3322 .. option:: disable_bw_measurement=bool, disable_bw=bool
3324 Disable measurements of throughput/bandwidth numbers. See
3325 :option:`disable_lat`.
3327 .. option:: clat_percentiles=bool
3329 Enable the reporting of percentiles of completion latencies. This
3330 option is mutually exclusive with :option:`lat_percentiles`.
3332 .. option:: lat_percentiles=bool
3334 Enable the reporting of percentiles of I/O latencies. This is similar
3335 to :option:`clat_percentiles`, except that this includes the
3336 submission latency. This option is mutually exclusive with
3337 :option:`clat_percentiles`.
3339 .. option:: percentile_list=float_list
3341 Overwrite the default list of percentiles for completion latencies and
3342 the block error histogram. Each number is a floating number in the
3343 range (0,100], and the maximum length of the list is 20. Use ``:`` to
3344 separate the numbers, and list the numbers in ascending order. For
3345 example, ``--percentile_list=99.5:99.9`` will cause fio to report the
3346 values of completion latency below which 99.5% and 99.9% of the observed
3347 latencies fell, respectively.
3349 .. option:: significant_figures=int
3351 If using :option:`--output-format` of `normal`, set the significant
3352 figures to this value. Higher values will yield more precise IOPS and
3353 throughput units, while lower values will round. Requires a minimum
3354 value of 1 and a maximum value of 10. Defaults to 4.
3360 .. option:: exitall_on_error
3362 When one job finishes in error, terminate the rest. The default is to wait
3363 for each job to finish.
3365 .. option:: continue_on_error=str
3367 Normally fio will exit the job on the first observed failure. If this option
3368 is set, fio will continue the job when there is a 'non-fatal error' (EIO or
3369 EILSEQ) until the runtime is exceeded or the I/O size specified is
3370 completed. If this option is used, there are two more stats that are
3371 appended, the total error count and the first error. The error field given
3372 in the stats is the first error that was hit during the run.
3374 The allowed values are:
3377 Exit on any I/O or verify errors.
3380 Continue on read errors, exit on all others.
3383 Continue on write errors, exit on all others.
3386 Continue on any I/O error, exit on all others.
3389 Continue on verify errors, exit on all others.
3392 Continue on all errors.
3395 Backward-compatible alias for 'none'.
3398 Backward-compatible alias for 'all'.
3400 .. option:: ignore_error=str
3402 Sometimes you want to ignore some errors during test in that case you can
3403 specify error list for each error type, instead of only being able to
3404 ignore the default 'non-fatal error' using :option:`continue_on_error`.
3405 ``ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST`` errors for
3406 given error type is separated with ':'. Error may be symbol ('ENOSPC',
3407 'ENOMEM') or integer. Example::
3409 ignore_error=EAGAIN,ENOSPC:122
3411 This option will ignore EAGAIN from READ, and ENOSPC and 122(EDQUOT) from
3412 WRITE. This option works by overriding :option:`continue_on_error` with
3413 the list of errors for each error type if any.
3415 .. option:: error_dump=bool
3417 If set dump every error even if it is non fatal, true by default. If
3418 disabled only fatal error will be dumped.
3420 Running predefined workloads
3421 ----------------------------
3423 Fio includes predefined profiles that mimic the I/O workloads generated by
3426 .. option:: profile=str
3428 The predefined workload to run. Current profiles are:
3431 Threaded I/O bench (tiotest/tiobench) like workload.
3434 Aerospike Certification Tool (ACT) like workload.
3436 To view a profile's additional options use :option:`--cmdhelp` after specifying
3437 the profile. For example::
3439 $ fio --profile=act --cmdhelp
3444 .. option:: device-names=str
3449 .. option:: load=int
3452 ACT load multiplier. Default: 1.
3454 .. option:: test-duration=time
3457 How long the entire test takes to run. When the unit is omitted, the value
3458 is given in seconds. Default: 24h.
3460 .. option:: threads-per-queue=int
3463 Number of read I/O threads per device. Default: 8.
3465 .. option:: read-req-num-512-blocks=int
3468 Number of 512B blocks to read at the time. Default: 3.
3470 .. option:: large-block-op-kbytes=int
3473 Size of large block ops in KiB (writes). Default: 131072.
3478 Set to run ACT prep phase.
3480 Tiobench profile options
3481 ~~~~~~~~~~~~~~~~~~~~~~~~
3483 .. option:: size=str
3488 .. option:: block=int
3491 Block size in bytes. Default: 4096.
3493 .. option:: numruns=int
3503 .. option:: threads=int
3508 Interpreting the output
3509 -----------------------
3512 Example output was based on the following:
3513 TZ=UTC fio --iodepth=8 --ioengine=null --size=100M --time_based \
3514 --rate=1256k --bs=14K --name=quick --runtime=1s --name=mixed \
3515 --runtime=2m --rw=rw
3517 Fio spits out a lot of output. While running, fio will display the status of the
3518 jobs created. An example of that would be::
3520 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]
3522 The characters inside the first set of square brackets denote the current status of
3523 each thread. The first character is the first job defined in the job file, and so
3524 forth. The possible values (in typical life cycle order) are:
3526 +------+-----+-----------------------------------------------------------+
3528 +======+=====+===========================================================+
3529 | P | | Thread setup, but not started. |
3530 +------+-----+-----------------------------------------------------------+
3531 | C | | Thread created. |
3532 +------+-----+-----------------------------------------------------------+
3533 | I | | Thread initialized, waiting or generating necessary data. |
3534 +------+-----+-----------------------------------------------------------+
3535 | | p | Thread running pre-reading file(s). |
3536 +------+-----+-----------------------------------------------------------+
3537 | | / | Thread is in ramp period. |
3538 +------+-----+-----------------------------------------------------------+
3539 | | R | Running, doing sequential reads. |
3540 +------+-----+-----------------------------------------------------------+
3541 | | r | Running, doing random reads. |
3542 +------+-----+-----------------------------------------------------------+
3543 | | W | Running, doing sequential writes. |
3544 +------+-----+-----------------------------------------------------------+
3545 | | w | Running, doing random writes. |
3546 +------+-----+-----------------------------------------------------------+
3547 | | M | Running, doing mixed sequential reads/writes. |
3548 +------+-----+-----------------------------------------------------------+
3549 | | m | Running, doing mixed random reads/writes. |
3550 +------+-----+-----------------------------------------------------------+
3551 | | D | Running, doing sequential trims. |
3552 +------+-----+-----------------------------------------------------------+
3553 | | d | Running, doing random trims. |
3554 +------+-----+-----------------------------------------------------------+
3555 | | F | Running, currently waiting for :manpage:`fsync(2)`. |
3556 +------+-----+-----------------------------------------------------------+
3557 | | V | Running, doing verification of written data. |
3558 +------+-----+-----------------------------------------------------------+
3559 | f | | Thread finishing. |
3560 +------+-----+-----------------------------------------------------------+
3561 | E | | Thread exited, not reaped by main thread yet. |
3562 +------+-----+-----------------------------------------------------------+
3563 | _ | | Thread reaped. |
3564 +------+-----+-----------------------------------------------------------+
3565 | X | | Thread reaped, exited with an error. |
3566 +------+-----+-----------------------------------------------------------+
3567 | K | | Thread reaped, exited due to signal. |
3568 +------+-----+-----------------------------------------------------------+
3571 Example output was based on the following:
3572 TZ=UTC fio --iodepth=8 --ioengine=null --size=100M --runtime=58m \
3573 --time_based --rate=2512k --bs=256K --numjobs=10 \
3574 --name=readers --rw=read --name=writers --rw=write
3576 Fio will condense the thread string as not to take up more space on the command
3577 line than needed. For instance, if you have 10 readers and 10 writers running,
3578 the output would look like this::
3580 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]
3582 Note that the status string is displayed in order, so it's possible to tell which of
3583 the jobs are currently doing what. In the example above this means that jobs 1--10
3584 are readers and 11--20 are writers.
3586 The other values are fairly self explanatory -- number of threads currently
3587 running and doing I/O, the number of currently open files (f=), the estimated
3588 completion percentage, the rate of I/O since last check (read speed listed first,
3589 then write speed and optionally trim speed) in terms of bandwidth and IOPS,
3590 and time to completion for the current running group. It's impossible to estimate
3591 runtime of the following groups (if any).
3594 Example output was based on the following:
3595 TZ=UTC fio --iodepth=16 --ioengine=posixaio --filename=/tmp/fiofile \
3596 --direct=1 --size=100M --time_based --runtime=50s --rate_iops=89 \
3597 --bs=7K --name=Client1 --rw=write
3599 When fio is done (or interrupted by :kbd:`Ctrl-C`), it will show the data for
3600 each thread, group of threads, and disks in that order. For each overall thread (or
3601 group) the output looks like::
3603 Client1: (groupid=0, jobs=1): err= 0: pid=16109: Sat Jun 24 12:07:54 2017
3604 write: IOPS=88, BW=623KiB/s (638kB/s)(30.4MiB/50032msec)
3605 slat (nsec): min=500, max=145500, avg=8318.00, stdev=4781.50
3606 clat (usec): min=170, max=78367, avg=4019.02, stdev=8293.31
3607 lat (usec): min=174, max=78375, avg=4027.34, stdev=8291.79
3608 clat percentiles (usec):
3609 | 1.00th=[ 302], 5.00th=[ 326], 10.00th=[ 343], 20.00th=[ 363],
3610 | 30.00th=[ 392], 40.00th=[ 404], 50.00th=[ 416], 60.00th=[ 445],
3611 | 70.00th=[ 816], 80.00th=[ 6718], 90.00th=[12911], 95.00th=[21627],
3612 | 99.00th=[43779], 99.50th=[51643], 99.90th=[68682], 99.95th=[72877],
3614 bw ( KiB/s): min= 532, max= 686, per=0.10%, avg=622.87, stdev=24.82, samples= 100
3615 iops : min= 76, max= 98, avg=88.98, stdev= 3.54, samples= 100
3616 lat (usec) : 250=0.04%, 500=64.11%, 750=4.81%, 1000=2.79%
3617 lat (msec) : 2=4.16%, 4=1.84%, 10=4.90%, 20=11.33%, 50=5.37%
3618 lat (msec) : 100=0.65%
3619 cpu : usr=0.27%, sys=0.18%, ctx=12072, majf=0, minf=21
3620 IO depths : 1=85.0%, 2=13.1%, 4=1.8%, 8=0.1%, 16=0.0%, 32=0.0%, >=64=0.0%
3621 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
3622 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
3623 issued rwt: total=0,4450,0, short=0,0,0, dropped=0,0,0
3624 latency : target=0, window=0, percentile=100.00%, depth=8
3626 The job name (or first job's name when using :option:`group_reporting`) is printed,
3627 along with the group id, count of jobs being aggregated, last error id seen (which
3628 is 0 when there are no errors), pid/tid of that thread and the time the job/group
3629 completed. Below are the I/O statistics for each data direction performed (showing
3630 writes in the example above). In the order listed, they denote:
3633 The string before the colon shows the I/O direction the statistics
3634 are for. **IOPS** is the average I/Os performed per second. **BW**
3635 is the average bandwidth rate shown as: value in power of 2 format
3636 (value in power of 10 format). The last two values show: (**total
3637 I/O performed** in power of 2 format / **runtime** of that thread).
3640 Submission latency (**min** being the minimum, **max** being the
3641 maximum, **avg** being the average, **stdev** being the standard
3642 deviation). This is the time it took to submit the I/O. For
3643 sync I/O this row is not displayed as the slat is really the
3644 completion latency (since queue/complete is one operation there).
3645 This value can be in nanoseconds, microseconds or milliseconds ---
3646 fio will choose the most appropriate base and print that (in the
3647 example above nanoseconds was the best scale). Note: in :option:`--minimal` mode
3648 latencies are always expressed in microseconds.
3651 Completion latency. Same names as slat, this denotes the time from
3652 submission to completion of the I/O pieces. For sync I/O, clat will
3653 usually be equal (or very close) to 0, as the time from submit to
3654 complete is basically just CPU time (I/O has already been done, see slat
3658 Total latency. Same names as slat and clat, this denotes the time from
3659 when fio created the I/O unit to completion of the I/O operation.
3662 Bandwidth statistics based on samples. Same names as the xlat stats,
3663 but also includes the number of samples taken (**samples**) and an
3664 approximate percentage of total aggregate bandwidth this thread
3665 received in its group (**per**). This last value is only really
3666 useful if the threads in this group are on the same disk, since they
3667 are then competing for disk access.
3670 IOPS statistics based on samples. Same names as bw.
3672 **lat (nsec/usec/msec)**
3673 The distribution of I/O completion latencies. This is the time from when
3674 I/O leaves fio and when it gets completed. Unlike the separate
3675 read/write/trim sections above, the data here and in the remaining
3676 sections apply to all I/Os for the reporting group. 250=0.04% means that
3677 0.04% of the I/Os completed in under 250us. 500=64.11% means that 64.11%
3678 of the I/Os required 250 to 499us for completion.
3681 CPU usage. User and system time, along with the number of context
3682 switches this thread went through, usage of system and user time, and
3683 finally the number of major and minor page faults. The CPU utilization
3684 numbers are averages for the jobs in that reporting group, while the
3685 context and fault counters are summed.
3688 The distribution of I/O depths over the job lifetime. The numbers are
3689 divided into powers of 2 and each entry covers depths from that value
3690 up to those that are lower than the next entry -- e.g., 16= covers
3691 depths from 16 to 31. Note that the range covered by a depth
3692 distribution entry can be different to the range covered by the
3693 equivalent submit/complete distribution entry.
3696 How many pieces of I/O were submitting in a single submit call. Each
3697 entry denotes that amount and below, until the previous entry -- e.g.,
3698 16=100% means that we submitted anywhere between 9 to 16 I/Os per submit
3699 call. Note that the range covered by a submit distribution entry can
3700 be different to the range covered by the equivalent depth distribution
3704 Like the above submit number, but for completions instead.
3707 The number of read/write/trim requests issued, and how many of them were
3711 These values are for :option:`latency_target` and related options. When
3712 these options are engaged, this section describes the I/O depth required
3713 to meet the specified latency target.
3716 Example output was based on the following:
3717 TZ=UTC fio --ioengine=null --iodepth=2 --size=100M --numjobs=2 \
3718 --rate_process=poisson --io_limit=32M --name=read --bs=128k \
3719 --rate=11M --name=write --rw=write --bs=2k --rate=700k
3721 After each client has been listed, the group statistics are printed. They
3722 will look like this::
3724 Run status group 0 (all jobs):
3725 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
3726 WRITE: bw=1231KiB/s (1261kB/s), 616KiB/s-621KiB/s (630kB/s-636kB/s), io=64.0MiB (67.1MB), run=52747-53223msec
3728 For each data direction it prints:
3731 Aggregate bandwidth of threads in this group followed by the
3732 minimum and maximum bandwidth of all the threads in this group.
3733 Values outside of brackets are power-of-2 format and those
3734 within are the equivalent value in a power-of-10 format.
3736 Aggregate I/O performed of all threads in this group. The
3737 format is the same as bw.
3739 The smallest and longest runtimes of the threads in this group.
3741 And finally, the disk statistics are printed. This is Linux specific. They will look like this::
3743 Disk stats (read/write):
3744 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
3746 Each value is printed for both reads and writes, with reads first. The
3750 Number of I/Os performed by all groups.
3752 Number of merges performed by the I/O scheduler.
3754 Number of ticks we kept the disk busy.
3756 Total time spent in the disk queue.
3758 The disk utilization. A value of 100% means we kept the disk
3759 busy constantly, 50% would be a disk idling half of the time.
3761 It is also possible to get fio to dump the current output while it is running,
3762 without terminating the job. To do that, send fio the **USR1** signal. You can
3763 also get regularly timed dumps by using the :option:`--status-interval`
3764 parameter, or by creating a file in :file:`/tmp` named
3765 :file:`fio-dump-status`. If fio sees this file, it will unlink it and dump the
3766 current output status.
3772 For scripted usage where you typically want to generate tables or graphs of the
3773 results, fio can output the results in a semicolon separated format. The format
3774 is one long line of values, such as::
3776 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%
3777 A description of this job goes here.
3779 The job description (if provided) follows on a second line for terse v2.
3780 It appears on the same line for other terse versions.
3782 To enable terse output, use the :option:`--minimal` or
3783 :option:`--output-format`\=terse command line options. The
3784 first value is the version of the terse output format. If the output has to be
3785 changed for some reason, this number will be incremented by 1 to signify that
3788 Split up, the format is as follows (comments in brackets denote when a
3789 field was introduced or whether it's specific to some terse version):
3793 terse version, fio version [v3], jobname, groupid, error
3797 Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec)
3798 Submission latency: min, max, mean, stdev (usec)
3799 Completion latency: min, max, mean, stdev (usec)
3800 Completion latency percentiles: 20 fields (see below)
3801 Total latency: min, max, mean, stdev (usec)
3802 Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev, number of samples [v5]
3803 IOPS [v5]: min, max, mean, stdev, number of samples
3809 Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec)
3810 Submission latency: min, max, mean, stdev (usec)
3811 Completion latency: min, max, mean, stdev (usec)
3812 Completion latency percentiles: 20 fields (see below)
3813 Total latency: min, max, mean, stdev (usec)
3814 Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev, number of samples [v5]
3815 IOPS [v5]: min, max, mean, stdev, number of samples
3817 TRIM status [all but version 3]:
3819 Fields are similar to READ/WRITE status.
3823 user, system, context switches, major faults, minor faults
3827 <=1, 2, 4, 8, 16, 32, >=64
3829 I/O latencies microseconds::
3831 <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
3833 I/O latencies milliseconds::
3835 <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
3837 Disk utilization [v3]::
3839 disk name, read ios, write ios, read merges, write merges, read ticks, write ticks,
3840 time spent in queue, disk utilization percentage
3842 Additional Info (dependent on continue_on_error, default off)::
3844 total # errors, first error code
3846 Additional Info (dependent on description being set)::
3850 Completion latency percentiles can be a grouping of up to 20 sets, so for the
3851 terse output fio writes all of them. Each field will look like this::
3855 which is the Xth percentile, and the `usec` latency associated with it.
3857 For `Disk utilization`, all disks used by fio are shown. So for each disk there
3858 will be a disk utilization section.
3860 Below is a single line containing short names for each of the fields in the
3861 minimal output v3, separated by semicolons::
3863 terse_version_3;fio_version;jobname;groupid;error;read_kb;read_bandwidth;read_iops;read_runtime_ms;read_slat_min;read_slat_max;read_slat_mean;read_slat_dev;read_clat_min;read_clat_max;read_clat_mean;read_clat_dev;read_clat_pct01;read_clat_pct02;read_clat_pct03;read_clat_pct04;read_clat_pct05;read_clat_pct06;read_clat_pct07;read_clat_pct08;read_clat_pct09;read_clat_pct10;read_clat_pct11;read_clat_pct12;read_clat_pct13;read_clat_pct14;read_clat_pct15;read_clat_pct16;read_clat_pct17;read_clat_pct18;read_clat_pct19;read_clat_pct20;read_tlat_min;read_lat_max;read_lat_mean;read_lat_dev;read_bw_min;read_bw_max;read_bw_agg_pct;read_bw_mean;read_bw_dev;write_kb;write_bandwidth;write_iops;write_runtime_ms;write_slat_min;write_slat_max;write_slat_mean;write_slat_dev;write_clat_min;write_clat_max;write_clat_mean;write_clat_dev;write_clat_pct01;write_clat_pct02;write_clat_pct03;write_clat_pct04;write_clat_pct05;write_clat_pct06;write_clat_pct07;write_clat_pct08;write_clat_pct09;write_clat_pct10;write_clat_pct11;write_clat_pct12;write_clat_pct13;write_clat_pct14;write_clat_pct15;write_clat_pct16;write_clat_pct17;write_clat_pct18;write_clat_pct19;write_clat_pct20;write_tlat_min;write_lat_max;write_lat_mean;write_lat_dev;write_bw_min;write_bw_max;write_bw_agg_pct;write_bw_mean;write_bw_dev;cpu_user;cpu_sys;cpu_csw;cpu_mjf;cpu_minf;iodepth_1;iodepth_2;iodepth_4;iodepth_8;iodepth_16;iodepth_32;iodepth_64;lat_2us;lat_4us;lat_10us;lat_20us;lat_50us;lat_100us;lat_250us;lat_500us;lat_750us;lat_1000us;lat_2ms;lat_4ms;lat_10ms;lat_20ms;lat_50ms;lat_100ms;lat_250ms;lat_500ms;lat_750ms;lat_1000ms;lat_2000ms;lat_over_2000ms;disk_name;disk_read_iops;disk_write_iops;disk_read_merges;disk_write_merges;disk_read_ticks;write_ticks;disk_queue_time;disk_util
3865 In client/server mode terse output differs from what appears when jobs are run
3866 locally. Disk utilization data is omitted from the standard terse output and
3867 for v3 and later appears on its own separate line at the end of each terse
3874 The `json` output format is intended to be both human readable and convenient
3875 for automated parsing. For the most part its sections mirror those of the
3876 `normal` output. The `runtime` value is reported in msec and the `bw` value is
3877 reported in 1024 bytes per second units.
3883 The `json+` output format is identical to the `json` output format except that it
3884 adds a full dump of the completion latency bins. Each `bins` object contains a
3885 set of (key, value) pairs where keys are latency durations and values count how
3886 many I/Os had completion latencies of the corresponding duration. For example,
3889 "bins" : { "87552" : 1, "89600" : 1, "94720" : 1, "96768" : 1, "97792" : 1, "99840" : 1, "100864" : 2, "103936" : 6, "104960" : 534, "105984" : 5995, "107008" : 7529, ... }
3891 This data indicates that one I/O required 87,552ns to complete, two I/Os required
3892 100,864ns to complete, and 7529 I/Os required 107,008ns to complete.
3894 Also included with fio is a Python script `fio_jsonplus_clat2csv` that takes
3895 json+ output and generates CSV-formatted latency data suitable for plotting.
3897 The latency durations actually represent the midpoints of latency intervals.
3898 For details refer to :file:`stat.h`.
3904 There are two trace file format that you can encounter. The older (v1) format is
3905 unsupported since version 1.20-rc3 (March 2008). It will still be described
3906 below in case that you get an old trace and want to understand it.
3908 In any case the trace is a simple text file with a single action per line.
3911 Trace file format v1
3912 ~~~~~~~~~~~~~~~~~~~~
3914 Each line represents a single I/O action in the following format::
3918 where `rw=0/1` for read/write, and the `offset` and `length` entries being in bytes.
3920 This format is not supported in fio versions >= 1.20-rc3.
3923 Trace file format v2
3924 ~~~~~~~~~~~~~~~~~~~~
3926 The second version of the trace file format was added in fio version 1.17. It
3927 allows to access more then one file per trace and has a bigger set of possible
3930 The first line of the trace file has to be::
3934 Following this can be lines in two different formats, which are described below.
3936 The file management format::
3940 The `filename` is given as an absolute path. The `action` can be one of these:
3943 Add the given `filename` to the trace.
3945 Open the file with the given `filename`. The `filename` has to have
3946 been added with the **add** action before.
3948 Close the file with the given `filename`. The file has to have been
3952 The file I/O action format::
3954 filename action offset length
3956 The `filename` is given as an absolute path, and has to have been added and
3957 opened before it can be used with this format. The `offset` and `length` are
3958 given in bytes. The `action` can be one of these:
3961 Wait for `offset` microseconds. Everything below 100 is discarded.
3962 The time is relative to the previous `wait` statement.
3964 Read `length` bytes beginning from `offset`.
3966 Write `length` bytes beginning from `offset`.
3968 :manpage:`fsync(2)` the file.
3970 :manpage:`fdatasync(2)` the file.
3972 Trim the given file from the given `offset` for `length` bytes.
3975 I/O Replay - Merging Traces
3976 ---------------------------
3978 Colocation is a common practice used to get the most out of a machine.
3979 Knowing which workloads play nicely with each other and which ones don't is
3980 a much harder task. While fio can replay workloads concurrently via multiple
3981 jobs, it leaves some variability up to the scheduler making results harder to
3982 reproduce. Merging is a way to make the order of events consistent.
3984 Merging is integrated into I/O replay and done when a
3985 :option:`merge_blktrace_file` is specified. The list of files passed to
3986 :option:`read_iolog` go through the merge process and output a single file
3987 stored to the specified file. The output file is passed on as if it were the
3988 only file passed to :option:`read_iolog`. An example would look like::
3990 $ fio --read_iolog="<file1>:<file2>" --merge_blktrace_file="<output_file>"
3992 Creating only the merged file can be done by passing the command line argument
3993 :option:`merge-blktrace-only`.
3995 Scaling traces can be done to see the relative impact of any particular trace
3996 being slowed down or sped up. :option:`merge_blktrace_scalars` takes in a colon
3997 separated list of percentage scalars. It is index paired with the files passed
3998 to :option:`read_iolog`.
4000 With scaling, it may be desirable to match the running time of all traces.
4001 This can be done with :option:`merge_blktrace_iters`. It is index paired with
4002 :option:`read_iolog` just like :option:`merge_blktrace_scalars`.
4004 In an example, given two traces, A and B, each 60s long. If we want to see
4005 the impact of trace A issuing IOs twice as fast and repeat trace A over the
4006 runtime of trace B, the following can be done::
4008 $ fio --read_iolog="<trace_a>:"<trace_b>" --merge_blktrace_file"<output_file>" --merge_blktrace_scalars="50:100" --merge_blktrace_iters="2:1"
4010 This runs trace A at 2x the speed twice for approximately the same runtime as
4011 a single run of trace B.
4014 CPU idleness profiling
4015 ----------------------
4017 In some cases, we want to understand CPU overhead in a test. For example, we
4018 test patches for the specific goodness of whether they reduce CPU usage.
4019 Fio implements a balloon approach to create a thread per CPU that runs at idle
4020 priority, meaning that it only runs when nobody else needs the cpu.
4021 By measuring the amount of work completed by the thread, idleness of each CPU
4022 can be derived accordingly.
4024 An unit work is defined as touching a full page of unsigned characters. Mean and
4025 standard deviation of time to complete an unit work is reported in "unit work"
4026 section. Options can be chosen to report detailed percpu idleness or overall
4027 system idleness by aggregating percpu stats.
4030 Verification and triggers
4031 -------------------------
4033 Fio is usually run in one of two ways, when data verification is done. The first
4034 is a normal write job of some sort with verify enabled. When the write phase has
4035 completed, fio switches to reads and verifies everything it wrote. The second
4036 model is running just the write phase, and then later on running the same job
4037 (but with reads instead of writes) to repeat the same I/O patterns and verify
4038 the contents. Both of these methods depend on the write phase being completed,
4039 as fio otherwise has no idea how much data was written.
4041 With verification triggers, fio supports dumping the current write state to
4042 local files. Then a subsequent read verify workload can load this state and know
4043 exactly where to stop. This is useful for testing cases where power is cut to a
4044 server in a managed fashion, for instance.
4046 A verification trigger consists of two things:
4048 1) Storing the write state of each job.
4049 2) Executing a trigger command.
4051 The write state is relatively small, on the order of hundreds of bytes to single
4052 kilobytes. It contains information on the number of completions done, the last X
4055 A trigger is invoked either through creation ('touch') of a specified file in
4056 the system, or through a timeout setting. If fio is run with
4057 :option:`--trigger-file`\= :file:`/tmp/trigger-file`, then it will continually
4058 check for the existence of :file:`/tmp/trigger-file`. When it sees this file, it
4059 will fire off the trigger (thus saving state, and executing the trigger
4062 For client/server runs, there's both a local and remote trigger. If fio is
4063 running as a server backend, it will send the job states back to the client for
4064 safe storage, then execute the remote trigger, if specified. If a local trigger
4065 is specified, the server will still send back the write state, but the client
4066 will then execute the trigger.
4068 Verification trigger example
4069 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4071 Let's say we want to run a powercut test on the remote Linux machine 'server'.
4072 Our write workload is in :file:`write-test.fio`. We want to cut power to 'server' at
4073 some point during the run, and we'll run this test from the safety or our local
4074 machine, 'localbox'. On the server, we'll start the fio backend normally::
4076 server# fio --server
4078 and on the client, we'll fire off the workload::
4080 localbox$ fio --client=server --trigger-file=/tmp/my-trigger --trigger-remote="bash -c \"echo b > /proc/sysrq-triger\""
4082 We set :file:`/tmp/my-trigger` as the trigger file, and we tell fio to execute::
4084 echo b > /proc/sysrq-trigger
4086 on the server once it has received the trigger and sent us the write state. This
4087 will work, but it's not **really** cutting power to the server, it's merely
4088 abruptly rebooting it. If we have a remote way of cutting power to the server
4089 through IPMI or similar, we could do that through a local trigger command
4090 instead. Let's assume we have a script that does IPMI reboot of a given hostname,
4091 ipmi-reboot. On localbox, we could then have run fio with a local trigger
4094 localbox$ fio --client=server --trigger-file=/tmp/my-trigger --trigger="ipmi-reboot server"
4096 For this case, fio would wait for the server to send us the write state, then
4097 execute ``ipmi-reboot server`` when that happened.
4099 Loading verify state
4100 ~~~~~~~~~~~~~~~~~~~~
4102 To load stored write state, a read verification job file must contain the
4103 :option:`verify_state_load` option. If that is set, fio will load the previously
4104 stored state. For a local fio run this is done by loading the files directly,
4105 and on a client/server run, the server backend will ask the client to send the
4106 files over and load them from there.
4112 Fio supports a variety of log file formats, for logging latencies, bandwidth,
4113 and IOPS. The logs share a common format, which looks like this:
4115 *time* (`msec`), *value*, *data direction*, *block size* (`bytes`),
4118 *Time* for the log entry is always in milliseconds. The *value* logged depends
4119 on the type of log, it will be one of the following:
4122 Value is latency in nsecs
4128 *Data direction* is one of the following:
4137 The entry's *block size* is always in bytes. The *offset* is the position in bytes
4138 from the start of the file for that particular I/O. The logging of the offset can be
4139 toggled with :option:`log_offset`.
4141 Fio defaults to logging every individual I/O but when windowed logging is set
4142 through :option:`log_avg_msec`, either the average (by default) or the maximum
4143 (:option:`log_max_value` is set) *value* seen over the specified period of time
4144 is recorded. Each *data direction* seen within the window period will aggregate
4145 its values in a separate row. Further, when using windowed logging the *block
4146 size* and *offset* entries will always contain 0.
4152 Normally fio is invoked as a stand-alone application on the machine where the
4153 I/O workload should be generated. However, the backend and frontend of fio can
4154 be run separately i.e., the fio server can generate an I/O workload on the "Device
4155 Under Test" while being controlled by a client on another machine.
4157 Start the server on the machine which has access to the storage DUT::
4161 where `args` defines what fio listens to. The arguments are of the form
4162 ``type,hostname`` or ``IP,port``. *type* is either ``ip`` (or ip4) for TCP/IP
4163 v4, ``ip6`` for TCP/IP v6, or ``sock`` for a local unix domain socket.
4164 *hostname* is either a hostname or IP address, and *port* is the port to listen
4165 to (only valid for TCP/IP, not a local socket). Some examples:
4169 Start a fio server, listening on all interfaces on the default port (8765).
4171 2) ``fio --server=ip:hostname,4444``
4173 Start a fio server, listening on IP belonging to hostname and on port 4444.
4175 3) ``fio --server=ip6:::1,4444``
4177 Start a fio server, listening on IPv6 localhost ::1 and on port 4444.
4179 4) ``fio --server=,4444``
4181 Start a fio server, listening on all interfaces on port 4444.
4183 5) ``fio --server=1.2.3.4``
4185 Start a fio server, listening on IP 1.2.3.4 on the default port.
4187 6) ``fio --server=sock:/tmp/fio.sock``
4189 Start a fio server, listening on the local socket :file:`/tmp/fio.sock`.
4191 Once a server is running, a "client" can connect to the fio server with::
4193 fio <local-args> --client=<server> <remote-args> <job file(s)>
4195 where `local-args` are arguments for the client where it is running, `server`
4196 is the connect string, and `remote-args` and `job file(s)` are sent to the
4197 server. The `server` string follows the same format as it does on the server
4198 side, to allow IP/hostname/socket and port strings.
4200 Fio can connect to multiple servers this way::
4202 fio --client=<server1> <job file(s)> --client=<server2> <job file(s)>
4204 If the job file is located on the fio server, then you can tell the server to
4205 load a local file as well. This is done by using :option:`--remote-config` ::
4207 fio --client=server --remote-config /path/to/file.fio
4209 Then fio will open this local (to the server) job file instead of being passed
4210 one from the client.
4212 If you have many servers (example: 100 VMs/containers), you can input a pathname
4213 of a file containing host IPs/names as the parameter value for the
4214 :option:`--client` option. For example, here is an example :file:`host.list`
4215 file containing 2 hostnames::
4217 host1.your.dns.domain
4218 host2.your.dns.domain
4220 The fio command would then be::
4222 fio --client=host.list <job file(s)>
4224 In this mode, you cannot input server-specific parameters or job files -- all
4225 servers receive the same job file.
4227 In order to let ``fio --client`` runs use a shared filesystem from multiple
4228 hosts, ``fio --client`` now prepends the IP address of the server to the
4229 filename. For example, if fio is using the directory :file:`/mnt/nfs/fio` and is
4230 writing filename :file:`fileio.tmp`, with a :option:`--client` `hostfile`
4231 containing two hostnames ``h1`` and ``h2`` with IP addresses 192.168.10.120 and
4232 192.168.10.121, then fio will create two files::
4234 /mnt/nfs/fio/192.168.10.120.fileio.tmp
4235 /mnt/nfs/fio/192.168.10.121.fileio.tmp
4237 Terse output in client/server mode will differ slightly from what is produced
4238 when fio is run in stand-alone mode. See the terse output section for details.