4 The first step in getting fio to simulate a desired I/O workload, is writing a
5 job file describing that specific setup. A job file may contain any number of
6 threads and/or files -- the typical contents of the job file is a *global*
7 section defining shared parameters, and one or more job sections describing the
8 jobs involved. When run, fio parses this file and sets everything up as
9 described. If we break down a job from top to bottom, it contains the following
14 Defines the I/O pattern issued to the file(s). We may only be reading
15 sequentially from this file(s), or we may be writing randomly. Or even
16 mixing reads and writes, sequentially or randomly.
17 Should we be doing buffered I/O, or direct/raw I/O?
21 In how large chunks are we issuing I/O? This may be a single value,
22 or it may describe a range of block sizes.
26 How much data are we going to be reading/writing.
30 How do we issue I/O? We could be memory mapping the file, we could be
31 using regular read/write, we could be using splice, async I/O, or even
36 If the I/O engine is async, how large a queuing depth do we want to
42 How many files are we spreading the workload over.
44 `Threads, processes and job synchronization`_
46 How many threads or processes should we spread this workload over.
48 The above are the basic parameters defined for a workload, in addition there's a
49 multitude of parameters that modify other aspects of how this job behaves.
55 .. option:: --debug=type
57 Enable verbose tracing `type` of various fio actions. May be ``all`` for all types
58 or individual types separated by a comma (e.g. ``--debug=file,mem`` will
59 enable file and memory debugging). Currently, additional logging is
63 Dump info related to processes.
65 Dump info related to file actions.
67 Dump info related to I/O queuing.
69 Dump info related to memory allocations.
71 Dump info related to blktrace setup.
73 Dump info related to I/O verification.
75 Enable all debug options.
77 Dump info related to random offset generation.
79 Dump info related to option matching and parsing.
81 Dump info related to disk utilization updates.
83 Dump info only related to job number x.
85 Dump info only related to mutex up/down ops.
87 Dump info related to profile extensions.
89 Dump info related to internal time keeping.
91 Dump info related to networking connections.
93 Dump info related to I/O rate switching.
95 Dump info related to log compress/decompress.
97 Dump info related to steadystate detection.
99 Dump info related to the helper thread.
101 Dump info related to support for zoned block devices.
103 Show available debug options.
105 .. option:: --parse-only
107 Parse options only, don't start any I/O.
109 .. option:: --merge-blktrace-only
111 Merge blktraces only, don't start any I/O.
113 .. option:: --output=filename
115 Write output to file `filename`.
117 .. option:: --output-format=format
119 Set the reporting `format` to `normal`, `terse`, `json`, or `json+`. Multiple
120 formats can be selected, separated by a comma. `terse` is a CSV based
121 format. `json+` is like `json`, except it adds a full dump of the latency
124 .. option:: --bandwidth-log
126 Generate aggregate bandwidth logs.
128 .. option:: --minimal
130 Print statistics in a terse, semicolon-delimited format.
132 .. option:: --append-terse
134 Print statistics in selected mode AND terse, semicolon-delimited format.
135 **Deprecated**, use :option:`--output-format` instead to select multiple
138 .. option:: --terse-version=version
140 Set terse `version` output format (default 3, or 2 or 4 or 5).
142 .. option:: --version
144 Print version information and exit.
148 Print a summary of the command line options and exit.
150 .. option:: --cpuclock-test
152 Perform test and validation of internal CPU clock.
154 .. option:: --crctest=[test]
156 Test the speed of the built-in checksumming functions. If no argument is
157 given, all of them are tested. Alternatively, a comma separated list can
158 be passed, in which case the given ones are tested.
160 .. option:: --cmdhelp=command
162 Print help information for `command`. May be ``all`` for all commands.
164 .. option:: --enghelp=[ioengine[,command]]
166 List all commands defined by `ioengine`, or print help for `command`
167 defined by `ioengine`. If no `ioengine` is given, list all
170 .. option:: --showcmd=jobfile
172 Convert `jobfile` to a set of command-line options.
174 .. option:: --readonly
176 Turn on safety read-only checks, preventing writes and trims. The
177 ``--readonly`` option is an extra safety guard to prevent users from
178 accidentally starting a write or trim workload when that is not desired.
179 Fio will only modify the device under test if
180 `rw=write/randwrite/rw/randrw/trim/randtrim/trimwrite` is given. This
181 safety net can be used as an extra precaution.
183 .. option:: --eta=when
185 Specifies when real-time ETA estimate should be printed. `when` may be
186 `always`, `never` or `auto`. `auto` is the default, it prints ETA
187 when requested if the output is a TTY. `always` disregards the output
188 type, and prints ETA when requested. `never` never prints ETA.
190 .. option:: --eta-interval=time
192 By default, fio requests client ETA status roughly every second. With
193 this option, the interval is configurable. Fio imposes a minimum
194 allowed time to avoid flooding the console, less than 250 msec is
197 .. option:: --eta-newline=time
199 Force a new line for every `time` period passed. When the unit is omitted,
200 the value is interpreted in seconds.
202 .. option:: --status-interval=time
204 Force a full status dump of cumulative (from job start) values at `time`
205 intervals. This option does *not* provide per-period measurements. So
206 values such as bandwidth are running averages. When the time unit is omitted,
207 `time` is interpreted in seconds. Note that using this option with
208 ``--output-format=json`` will yield output that technically isn't valid
209 json, since the output will be collated sets of valid json. It will need
210 to be split into valid sets of json after the run.
212 .. option:: --section=name
214 Only run specified section `name` in job file. Multiple sections can be specified.
215 The ``--section`` option allows one to combine related jobs into one file.
216 E.g. one job file could define light, moderate, and heavy sections. Tell
217 fio to run only the "heavy" section by giving ``--section=heavy``
218 command line option. One can also specify the "write" operations in one
219 section and "verify" operation in another section. The ``--section`` option
220 only applies to job sections. The reserved *global* section is always
223 .. option:: --alloc-size=kb
225 Allocate additional internal smalloc pools of size `kb` in KiB. The
226 ``--alloc-size`` option increases shared memory set aside for use by fio.
227 If running large jobs with randommap enabled, fio can run out of memory.
228 Smalloc is an internal allocator for shared structures from a fixed size
229 memory pool and can grow to 16 pools. The pool size defaults to 16MiB.
231 NOTE: While running :file:`.fio_smalloc.*` backing store files are visible
234 .. option:: --warnings-fatal
236 All fio parser warnings are fatal, causing fio to exit with an
239 .. option:: --max-jobs=nr
241 Set the maximum number of threads/processes to support to `nr`.
242 NOTE: On Linux, it may be necessary to increase the shared-memory
243 limit (:file:`/proc/sys/kernel/shmmax`) if fio runs into errors while
246 .. option:: --server=args
248 Start a backend server, with `args` specifying what to listen to.
249 See `Client/Server`_ section.
251 .. option:: --daemonize=pidfile
253 Background a fio server, writing the pid to the given `pidfile` file.
255 .. option:: --client=hostname
257 Instead of running the jobs locally, send and run them on the given `hostname`
258 or set of `hostname`\s. See `Client/Server`_ section.
260 .. option:: --remote-config=file
262 Tell fio server to load this local `file`.
264 .. option:: --idle-prof=option
266 Report CPU idleness. `option` is one of the following:
269 Run unit work calibration only and exit.
272 Show aggregate system idleness and unit work.
275 As **system** but also show per CPU idleness.
277 .. option:: --inflate-log=log
279 Inflate and output compressed `log`.
281 .. option:: --trigger-file=file
283 Execute trigger command when `file` exists.
285 .. option:: --trigger-timeout=time
287 Execute trigger at this `time`.
289 .. option:: --trigger=command
291 Set this `command` as local trigger.
293 .. option:: --trigger-remote=command
295 Set this `command` as remote trigger.
297 .. option:: --aux-path=path
299 Use the directory specified by `path` for generated state files instead
300 of the current working directory.
302 Any parameters following the options will be assumed to be job files, unless
303 they match a job file parameter. Multiple job files can be listed and each job
304 file will be regarded as a separate group. Fio will :option:`stonewall`
305 execution between each group.
311 As previously described, fio accepts one or more job files describing what it is
312 supposed to do. The job file format is the classic ini file, where the names
313 enclosed in [] brackets define the job name. You are free to use any ASCII name
314 you want, except *global* which has special meaning. Following the job name is
315 a sequence of zero or more parameters, one per line, that define the behavior of
316 the job. If the first character in a line is a ';' or a '#', the entire line is
317 discarded as a comment.
319 A *global* section sets defaults for the jobs described in that file. A job may
320 override a *global* section parameter, and a job file may even have several
321 *global* sections if so desired. A job is only affected by a *global* section
324 The :option:`--cmdhelp` option also lists all options. If used with a `command`
325 argument, :option:`--cmdhelp` will detail the given `command`.
327 See the `examples/` directory for inspiration on how to write job files. Note
328 the copyright and license requirements currently apply to `examples/` files.
330 So let's look at a really simple job file that defines two processes, each
331 randomly reading from a 128MiB file:
335 ; -- start job file --
346 As you can see, the job file sections themselves are empty as all the described
347 parameters are shared. As no :option:`filename` option is given, fio makes up a
348 `filename` for each of the jobs as it sees fit. On the command line, this job
349 would look as follows::
351 $ fio --name=global --rw=randread --size=128m --name=job1 --name=job2
354 Let's look at an example that has a number of processes writing randomly to
359 ; -- start job file --
370 Here we have no *global* section, as we only have one job defined anyway. We
371 want to use async I/O here, with a depth of 4 for each file. We also increased
372 the buffer size used to 32KiB and define numjobs to 4 to fork 4 identical
373 jobs. The result is 4 processes each randomly writing to their own 64MiB
374 file. Instead of using the above job file, you could have given the parameters
375 on the command line. For this case, you would specify::
377 $ fio --name=random-writers --ioengine=libaio --iodepth=4 --rw=randwrite --bs=32k --direct=0 --size=64m --numjobs=4
379 When fio is utilized as a basis of any reasonably large test suite, it might be
380 desirable to share a set of standardized settings across multiple job files.
381 Instead of copy/pasting such settings, any section may pull in an external
382 :file:`filename.fio` file with *include filename* directive, as in the following
385 ; -- start job file including.fio --
389 include glob-include.fio
396 include test-include.fio
397 ; -- end job file including.fio --
401 ; -- start job file glob-include.fio --
404 ; -- end job file glob-include.fio --
408 ; -- start job file test-include.fio --
411 ; -- end job file test-include.fio --
413 Settings pulled into a section apply to that section only (except *global*
414 section). Include directives may be nested in that any included file may contain
415 further include directive(s). Include files may not contain [] sections.
418 Environment variables
419 ~~~~~~~~~~~~~~~~~~~~~
421 Fio also supports environment variable expansion in job files. Any sub-string of
422 the form ``${VARNAME}`` as part of an option value (in other words, on the right
423 of the '='), will be expanded to the value of the environment variable called
424 `VARNAME`. If no such environment variable is defined, or `VARNAME` is the
425 empty string, the empty string will be substituted.
427 As an example, let's look at a sample fio invocation and job file::
429 $ SIZE=64m NUMJOBS=4 fio jobfile.fio
433 ; -- start job file --
440 This will expand to the following equivalent job file at runtime:
444 ; -- start job file --
451 Fio ships with a few example job files, you can also look there for inspiration.
456 Additionally, fio has a set of reserved keywords that will be replaced
457 internally with the appropriate value. Those keywords are:
461 The architecture page size of the running system.
465 Megabytes of total memory in the system.
469 Number of online available CPUs.
471 These can be used on the command line or in the job file, and will be
472 automatically substituted with the current system values when the job is
473 run. Simple math is also supported on these keywords, so you can perform actions
478 and get that properly expanded to 8 times the size of memory in the machine.
484 This section describes in details each parameter associated with a job. Some
485 parameters take an option of a given type, such as an integer or a
486 string. Anywhere a numeric value is required, an arithmetic expression may be
487 used, provided it is surrounded by parentheses. Supported operators are:
496 For time values in expressions, units are microseconds by default. This is
497 different than for time values not in expressions (not enclosed in
498 parentheses). The following types are used:
505 String: A sequence of alphanumeric characters.
508 Integer with possible time suffix. Without a unit value is interpreted as
509 seconds unless otherwise specified. Accepts a suffix of 'd' for days, 'h' for
510 hours, 'm' for minutes, 's' for seconds, 'ms' (or 'msec') for milliseconds and
511 'us' (or 'usec') for microseconds. For example, use 10m for 10 minutes.
516 Integer. A whole number value, which may contain an integer prefix
517 and an integer suffix:
519 [*integer prefix*] **number** [*integer suffix*]
521 The optional *integer prefix* specifies the number's base. The default
522 is decimal. *0x* specifies hexadecimal.
524 The optional *integer suffix* specifies the number's units, and includes an
525 optional unit prefix and an optional unit. For quantities of data, the
526 default unit is bytes. For quantities of time, the default unit is seconds
527 unless otherwise specified.
529 With :option:`kb_base`\=1000, fio follows international standards for unit
530 prefixes. To specify power-of-10 decimal values defined in the
531 International System of Units (SI):
533 * *K* -- means kilo (K) or 1000
534 * *M* -- means mega (M) or 1000**2
535 * *G* -- means giga (G) or 1000**3
536 * *T* -- means tera (T) or 1000**4
537 * *P* -- means peta (P) or 1000**5
539 To specify power-of-2 binary values defined in IEC 80000-13:
541 * *Ki* -- means kibi (Ki) or 1024
542 * *Mi* -- means mebi (Mi) or 1024**2
543 * *Gi* -- means gibi (Gi) or 1024**3
544 * *Ti* -- means tebi (Ti) or 1024**4
545 * *Pi* -- means pebi (Pi) or 1024**5
547 For Zone Block Device Mode:
550 With :option:`kb_base`\=1024 (the default), the unit prefixes are opposite
551 from those specified in the SI and IEC 80000-13 standards to provide
552 compatibility with old scripts. For example, 4k means 4096.
554 For quantities of data, an optional unit of 'B' may be included
555 (e.g., 'kB' is the same as 'k').
557 The *integer suffix* is not case sensitive (e.g., m/mi mean mebi/mega,
558 not milli). 'b' and 'B' both mean byte, not bit.
560 Examples with :option:`kb_base`\=1000:
562 * *4 KiB*: 4096, 4096b, 4096B, 4ki, 4kib, 4kiB, 4Ki, 4KiB
563 * *1 MiB*: 1048576, 1mi, 1024ki
564 * *1 MB*: 1000000, 1m, 1000k
565 * *1 TiB*: 1099511627776, 1ti, 1024gi, 1048576mi
566 * *1 TB*: 1000000000, 1t, 1000m, 1000000k
568 Examples with :option:`kb_base`\=1024 (default):
570 * *4 KiB*: 4096, 4096b, 4096B, 4k, 4kb, 4kB, 4K, 4KB
571 * *1 MiB*: 1048576, 1m, 1024k
572 * *1 MB*: 1000000, 1mi, 1000ki
573 * *1 TiB*: 1099511627776, 1t, 1024g, 1048576m
574 * *1 TB*: 1000000000, 1ti, 1000mi, 1000000ki
576 To specify times (units are not case sensitive):
580 * *M* -- means minutes
581 * *s* -- or sec means seconds (default)
582 * *ms* -- or *msec* means milliseconds
583 * *us* -- or *usec* means microseconds
585 If the option accepts an upper and lower range, use a colon ':' or
586 minus '-' to separate such values. See :ref:`irange <irange>`.
587 If the lower value specified happens to be larger than the upper value
588 the two values are swapped.
593 Boolean. Usually parsed as an integer, however only defined for
594 true and false (1 and 0).
599 Integer range with suffix. Allows value range to be given, such as
600 1024-4096. A colon may also be used as the separator, e.g. 1k:4k. If the
601 option allows two sets of ranges, they can be specified with a ',' or '/'
602 delimiter: 1k-4k/8k-32k. Also see :ref:`int <int>`.
605 A list of floating point numbers, separated by a ':' character.
607 With the above in mind, here follows the complete list of fio job parameters.
613 .. option:: kb_base=int
615 Select the interpretation of unit prefixes in input parameters.
618 Inputs comply with IEC 80000-13 and the International
619 System of Units (SI). Use:
621 - power-of-2 values with IEC prefixes (e.g., KiB)
622 - power-of-10 values with SI prefixes (e.g., kB)
625 Compatibility mode (default). To avoid breaking old scripts:
627 - power-of-2 values with SI prefixes
628 - power-of-10 values with IEC prefixes
630 See :option:`bs` for more details on input parameters.
632 Outputs always use correct prefixes. Most outputs include both
635 bw=2383.3kB/s (2327.4KiB/s)
637 If only one value is reported, then kb_base selects the one to use:
639 **1000** -- SI prefixes
641 **1024** -- IEC prefixes
643 .. option:: unit_base=int
645 Base unit for reporting. Allowed values are:
648 Use auto-detection (default).
660 ASCII name of the job. This may be used to override the name printed by fio
661 for this job. Otherwise the job name is used. On the command line this
662 parameter has the special purpose of also signaling the start of a new job.
664 .. option:: description=str
666 Text description of the job. Doesn't do anything except dump this text
667 description when this job is run. It's not parsed.
669 .. option:: loops=int
671 Run the specified number of iterations of this job. Used to repeat the same
672 workload a given number of times. Defaults to 1.
674 .. option:: numjobs=int
676 Create the specified number of clones of this job. Each clone of job
677 is spawned as an independent thread or process. May be used to setup a
678 larger number of threads/processes doing the same thing. Each thread is
679 reported separately; to see statistics for all clones as a whole, use
680 :option:`group_reporting` in conjunction with :option:`new_group`.
681 See :option:`--max-jobs`. Default: 1.
684 Time related parameters
685 ~~~~~~~~~~~~~~~~~~~~~~~
687 .. option:: runtime=time
689 Tell fio to terminate processing after the specified period of time. It
690 can be quite hard to determine for how long a specified job will run, so
691 this parameter is handy to cap the total runtime to a given time. When
692 the unit is omitted, the value is interpreted in seconds.
694 .. option:: time_based
696 If set, fio will run for the duration of the :option:`runtime` specified
697 even if the file(s) are completely read or written. It will simply loop over
698 the same workload as many times as the :option:`runtime` allows.
700 .. option:: startdelay=irange(time)
702 Delay the start of job for the specified amount of time. Can be a single
703 value or a range. When given as a range, each thread will choose a value
704 randomly from within the range. Value is in seconds if a unit is omitted.
706 .. option:: ramp_time=time
708 If set, fio will run the specified workload for this amount of time before
709 logging any performance numbers. Useful for letting performance settle
710 before logging results, thus minimizing the runtime required for stable
711 results. Note that the ``ramp_time`` is considered lead in time for a job,
712 thus it will increase the total runtime if a special timeout or
713 :option:`runtime` is specified. When the unit is omitted, the value is
716 .. option:: clocksource=str
718 Use the given clocksource as the base of timing. The supported options are:
721 :manpage:`gettimeofday(2)`
724 :manpage:`clock_gettime(2)`
727 Internal CPU clock source
729 cpu is the preferred clocksource if it is reliable, as it is very fast (and
730 fio is heavy on time calls). Fio will automatically use this clocksource if
731 it's supported and considered reliable on the system it is running on,
732 unless another clocksource is specifically set. For x86/x86-64 CPUs, this
733 means supporting TSC Invariant.
735 .. option:: gtod_reduce=bool
737 Enable all of the :manpage:`gettimeofday(2)` reducing options
738 (:option:`disable_clat`, :option:`disable_slat`, :option:`disable_bw_measurement`) plus
739 reduce precision of the timeout somewhat to really shrink the
740 :manpage:`gettimeofday(2)` call count. With this option enabled, we only do
741 about 0.4% of the :manpage:`gettimeofday(2)` calls we would have done if all
742 time keeping was enabled.
744 .. option:: gtod_cpu=int
746 Sometimes it's cheaper to dedicate a single thread of execution to just
747 getting the current time. Fio (and databases, for instance) are very
748 intensive on :manpage:`gettimeofday(2)` calls. With this option, you can set
749 one CPU aside for doing nothing but logging current time to a shared memory
750 location. Then the other threads/processes that run I/O workloads need only
751 copy that segment, instead of entering the kernel with a
752 :manpage:`gettimeofday(2)` call. The CPU set aside for doing these time
753 calls will be excluded from other uses. Fio will manually clear it from the
754 CPU mask of other jobs.
760 .. option:: directory=str
762 Prefix filenames with this directory. Used to place files in a different
763 location than :file:`./`. You can specify a number of directories by
764 separating the names with a ':' character. These directories will be
765 assigned equally distributed to job clones created by :option:`numjobs` as
766 long as they are using generated filenames. If specific `filename(s)` are
767 set fio will use the first listed directory, and thereby matching the
768 `filename` semantic (which generates a file for each clone if not
769 specified, but lets all clones use the same file if set).
771 See the :option:`filename` option for information on how to escape "``:``"
772 characters within the directory path itself.
774 Note: To control the directory fio will use for internal state files
775 use :option:`--aux-path`.
777 .. option:: filename=str
779 Fio normally makes up a `filename` based on the job name, thread number, and
780 file number (see :option:`filename_format`). If you want to share files
781 between threads in a job or several
782 jobs with fixed file paths, specify a `filename` for each of them to override
783 the default. If the ioengine is file based, you can specify a number of files
784 by separating the names with a ':' colon. So if you wanted a job to open
785 :file:`/dev/sda` and :file:`/dev/sdb` as the two working files, you would use
786 ``filename=/dev/sda:/dev/sdb``. This also means that whenever this option is
787 specified, :option:`nrfiles` is ignored. The size of regular files specified
788 by this option will be :option:`size` divided by number of files unless an
789 explicit size is specified by :option:`filesize`.
791 Each colon in the wanted path must be escaped with a ``\``
792 character. For instance, if the path is :file:`/dev/dsk/foo@3,0:c` then you
793 would use ``filename=/dev/dsk/foo@3,0\:c`` and if the path is
794 :file:`F:\\filename` then you would use ``filename=F\:\filename``.
796 On Windows, disk devices are accessed as :file:`\\\\.\\PhysicalDrive0` for
797 the first device, :file:`\\\\.\\PhysicalDrive1` for the second etc.
798 Note: Windows and FreeBSD prevent write access to areas
799 of the disk containing in-use data (e.g. filesystems).
801 The filename "`-`" is a reserved name, meaning *stdin* or *stdout*. Which
802 of the two depends on the read/write direction set.
804 .. option:: filename_format=str
806 If sharing multiple files between jobs, it is usually necessary to have fio
807 generate the exact names that you want. By default, fio will name a file
808 based on the default file format specification of
809 :file:`jobname.jobnumber.filenumber`. With this option, that can be
810 customized. Fio will recognize and replace the following keywords in this
814 The name of the worker thread or process.
816 IP of the fio process when using client/server mode.
818 The incremental number of the worker thread or process.
820 The incremental number of the file for that worker thread or
823 To have dependent jobs share a set of files, this option can be set to have
824 fio generate filenames that are shared between the two. For instance, if
825 :file:`testfiles.$filenum` is specified, file number 4 for any job will be
826 named :file:`testfiles.4`. The default of :file:`$jobname.$jobnum.$filenum`
827 will be used if no other format specifier is given.
829 If you specify a path then the directories will be created up to the
830 main directory for the file. So for example if you specify
831 ``filename_format=a/b/c/$jobnum`` then the directories a/b/c will be
832 created before the file setup part of the job. If you specify
833 :option:`directory` then the path will be relative that directory,
834 otherwise it is treated as the absolute path.
836 .. option:: unique_filename=bool
838 To avoid collisions between networked clients, fio defaults to prefixing any
839 generated filenames (with a directory specified) with the source of the
840 client connecting. To disable this behavior, set this option to 0.
842 .. option:: opendir=str
844 Recursively open any files below directory `str`.
846 .. option:: lockfile=str
848 Fio defaults to not locking any files before it does I/O to them. If a file
849 or file descriptor is shared, fio can serialize I/O to that file to make the
850 end result consistent. This is usual for emulating real workloads that share
851 files. The lock modes are:
854 No locking. The default.
856 Only one thread or process may do I/O at a time, excluding all
859 Read-write locking on the file. Many readers may
860 access the file at the same time, but writes get exclusive access.
862 .. option:: nrfiles=int
864 Number of files to use for this job. Defaults to 1. The size of files
865 will be :option:`size` divided by this unless explicit size is specified by
866 :option:`filesize`. Files are created for each thread separately, and each
867 file will have a file number within its name by default, as explained in
868 :option:`filename` section.
871 .. option:: openfiles=int
873 Number of files to keep open at the same time. Defaults to the same as
874 :option:`nrfiles`, can be set smaller to limit the number simultaneous
877 .. option:: file_service_type=str
879 Defines how fio decides which file from a job to service next. The following
883 Choose a file at random.
886 Round robin over opened files. This is the default.
889 Finish one file before moving on to the next. Multiple files can
890 still be open depending on :option:`openfiles`.
893 Use a *Zipf* distribution to decide what file to access.
896 Use a *Pareto* distribution to decide what file to access.
899 Use a *Gaussian* (normal) distribution to decide what file to
905 For *random*, *roundrobin*, and *sequential*, a postfix can be appended to
906 tell fio how many I/Os to issue before switching to a new file. For example,
907 specifying ``file_service_type=random:8`` would cause fio to issue
908 8 I/Os before selecting a new file at random. For the non-uniform
909 distributions, a floating point postfix can be given to influence how the
910 distribution is skewed. See :option:`random_distribution` for a description
911 of how that would work.
913 .. option:: ioscheduler=str
915 Attempt to switch the device hosting the file to the specified I/O scheduler
918 .. option:: create_serialize=bool
920 If true, serialize the file creation for the jobs. This may be handy to
921 avoid interleaving of data files, which may greatly depend on the filesystem
922 used and even the number of processors in the system. Default: true.
924 .. option:: create_fsync=bool
926 :manpage:`fsync(2)` the data file after creation. This is the default.
928 .. option:: create_on_open=bool
930 If true, don't pre-create files but allow the job's open() to create a file
931 when it's time to do I/O. Default: false -- pre-create all necessary files
934 .. option:: create_only=bool
936 If true, fio will only run the setup phase of the job. If files need to be
937 laid out or updated on disk, only that will be done -- the actual job contents
938 are not executed. Default: false.
940 .. option:: allow_file_create=bool
942 If true, fio is permitted to create files as part of its workload. If this
943 option is false, then fio will error out if
944 the files it needs to use don't already exist. Default: true.
946 .. option:: allow_mounted_write=bool
948 If this isn't set, fio will abort jobs that are destructive (e.g. that write)
949 to what appears to be a mounted device or partition. This should help catch
950 creating inadvertently destructive tests, not realizing that the test will
951 destroy data on the mounted file system. Note that some platforms don't allow
952 writing against a mounted device regardless of this option. Default: false.
954 .. option:: pre_read=bool
956 If this is given, files will be pre-read into memory before starting the
957 given I/O operation. This will also clear the :option:`invalidate` flag,
958 since it is pointless to pre-read and then drop the cache. This will only
959 work for I/O engines that are seek-able, since they allow you to read the
960 same data multiple times. Thus it will not work on non-seekable I/O engines
961 (e.g. network, splice). Default: false.
963 .. option:: unlink=bool
965 Unlink the job files when done. Not the default, as repeated runs of that
966 job would then waste time recreating the file set again and again. Default:
969 .. option:: unlink_each_loop=bool
971 Unlink job files after each iteration or loop. Default: false.
973 .. option:: zonemode=str
978 The :option:`zonerange`, :option:`zonesize`,
979 :option `zonecapacity` and option:`zoneskip`
980 parameters are ignored.
982 I/O happens in a single zone until
983 :option:`zonesize` bytes have been transferred.
984 After that number of bytes has been
985 transferred processing of the next zone
986 starts. :option `zonecapacity` is ignored.
988 Zoned block device mode. I/O happens
989 sequentially in each zone, even if random I/O
990 has been selected. Random I/O happens across
991 all zones instead of being restricted to a
992 single zone. The :option:`zoneskip` parameter
993 is ignored. :option:`zonerange` and
994 :option:`zonesize` must be identical.
996 .. option:: zonerange=int
998 Size of a single zone. See also :option:`zonesize` and
1001 .. option:: zonesize=int
1003 For :option:`zonemode` =strided, this is the number of bytes to
1004 transfer before skipping :option:`zoneskip` bytes. If this parameter
1005 is smaller than :option:`zonerange` then only a fraction of each zone
1006 with :option:`zonerange` bytes will be accessed. If this parameter is
1007 larger than :option:`zonerange` then each zone will be accessed
1008 multiple times before skipping to the next zone.
1010 For :option:`zonemode` =zbd, this is the size of a single zone. The
1011 :option:`zonerange` parameter is ignored in this mode.
1014 .. option:: zonecapacity=int
1016 For :option:`zonemode` =zbd, this defines the capacity of a single zone,
1017 which is the accessible area starting from the zone start address.
1018 This parameter only applies when using :option:`zonemode` =zbd in
1019 combination with regular block devices. If not specified it defaults to
1020 the zone size. If the target device is a zoned block device, the zone
1021 capacity is obtained from the device information and this option is
1024 .. option:: zoneskip=int
1026 For :option:`zonemode` =strided, the number of bytes to skip after
1027 :option:`zonesize` bytes of data have been transferred. This parameter
1028 must be zero for :option:`zonemode` =zbd.
1030 .. option:: read_beyond_wp=bool
1032 This parameter applies to :option:`zonemode` =zbd only.
1034 Zoned block devices are block devices that consist of multiple zones.
1035 Each zone has a type, e.g. conventional or sequential. A conventional
1036 zone can be written at any offset that is a multiple of the block
1037 size. Sequential zones must be written sequentially. The position at
1038 which a write must occur is called the write pointer. A zoned block
1039 device can be either drive managed, host managed or host aware. For
1040 host managed devices the host must ensure that writes happen
1041 sequentially. Fio recognizes host managed devices and serializes
1042 writes to sequential zones for these devices.
1044 If a read occurs in a sequential zone beyond the write pointer then
1045 the zoned block device will complete the read without reading any data
1046 from the storage medium. Since such reads lead to unrealistically high
1047 bandwidth and IOPS numbers fio only reads beyond the write pointer if
1048 explicitly told to do so. Default: false.
1050 .. option:: max_open_zones=int
1052 When running a random write test across an entire drive many more
1053 zones will be open than in a typical application workload. Hence this
1054 command line option that allows to limit the number of open zones. The
1055 number of open zones is defined as the number of zones to which write
1056 commands are issued.
1058 .. option:: zone_reset_threshold=float
1060 A number between zero and one that indicates the ratio of logical
1061 blocks with data to the total number of logical blocks in the test
1062 above which zones should be reset periodically.
1064 .. option:: zone_reset_frequency=float
1066 A number between zero and one that indicates how often a zone reset
1067 should be issued if the zone reset threshold has been exceeded. A zone
1068 reset is submitted after each (1 / zone_reset_frequency) write
1069 requests. This and the previous parameter can be used to simulate
1070 garbage collection activity.
1076 .. option:: direct=bool
1078 If value is true, use non-buffered I/O. This is usually O_DIRECT. Note that
1079 OpenBSD and ZFS on Solaris don't support direct I/O. On Windows the synchronous
1080 ioengines don't support direct I/O. Default: false.
1082 .. option:: atomic=bool
1084 If value is true, attempt to use atomic direct I/O. Atomic writes are
1085 guaranteed to be stable once acknowledged by the operating system. Only
1086 Linux supports O_ATOMIC right now.
1088 .. option:: buffered=bool
1090 If value is true, use buffered I/O. This is the opposite of the
1091 :option:`direct` option. Defaults to true.
1093 .. option:: readwrite=str, rw=str
1095 Type of I/O pattern. Accepted values are:
1102 Sequential trims (Linux block devices and SCSI
1103 character devices only).
1109 Random trims (Linux block devices and SCSI
1110 character devices only).
1112 Sequential mixed reads and writes.
1114 Random mixed reads and writes.
1116 Sequential trim+write sequences. Blocks will be trimmed first,
1117 then the same blocks will be written to.
1119 Fio defaults to read if the option is not specified. For the mixed I/O
1120 types, the default is to split them 50/50. For certain types of I/O the
1121 result may still be skewed a bit, since the speed may be different.
1123 It is possible to specify the number of I/Os to do before getting a new
1124 offset by appending ``:<nr>`` to the end of the string given. For a
1125 random read, it would look like ``rw=randread:8`` for passing in an offset
1126 modifier with a value of 8. If the suffix is used with a sequential I/O
1127 pattern, then the *<nr>* value specified will be **added** to the generated
1128 offset for each I/O turning sequential I/O into sequential I/O with holes.
1129 For instance, using ``rw=write:4k`` will skip 4k for every write. Also see
1130 the :option:`rw_sequencer` option.
1132 .. option:: rw_sequencer=str
1134 If an offset modifier is given by appending a number to the ``rw=<str>``
1135 line, then this option controls how that number modifies the I/O offset
1136 being generated. Accepted values are:
1139 Generate sequential offset.
1141 Generate the same offset.
1143 ``sequential`` is only useful for random I/O, where fio would normally
1144 generate a new random offset for every I/O. If you append e.g. 8 to randread,
1145 you would get a new random offset for every 8 I/Os. The result would be a
1146 seek for only every 8 I/Os, instead of for every I/O. Use ``rw=randread:8``
1147 to specify that. As sequential I/O is already sequential, setting
1148 ``sequential`` for that would not result in any differences. ``identical``
1149 behaves in a similar fashion, except it sends the same offset 8 number of
1150 times before generating a new offset.
1152 .. option:: unified_rw_reporting=str
1154 Fio normally reports statistics on a per data direction basis, meaning that
1155 reads, writes, and trims are accounted and reported separately. This option
1156 determines whether fio reports the results normally, summed together, or as
1158 Accepted values are:
1161 Normal statistics reporting.
1164 Statistics are summed per data direction and reported together.
1167 Statistics are reported normally, followed by the mixed statistics.
1170 Backward-compatible alias for **none**.
1173 Backward-compatible alias for **mixed**.
1178 .. option:: randrepeat=bool
1180 Seed the random number generator used for random I/O patterns in a
1181 predictable way so the pattern is repeatable across runs. Default: true.
1183 .. option:: allrandrepeat=bool
1185 Seed all random number generators in a predictable way so results are
1186 repeatable across runs. Default: false.
1188 .. option:: randseed=int
1190 Seed the random number generators based on this seed value, to be able to
1191 control what sequence of output is being generated. If not set, the random
1192 sequence depends on the :option:`randrepeat` setting.
1194 .. option:: fallocate=str
1196 Whether pre-allocation is performed when laying down files.
1197 Accepted values are:
1200 Do not pre-allocate space.
1203 Use a platform's native pre-allocation call but fall back to
1204 **none** behavior if it fails/is not implemented.
1207 Pre-allocate via :manpage:`posix_fallocate(3)`.
1210 Pre-allocate via :manpage:`fallocate(2)` with
1211 FALLOC_FL_KEEP_SIZE set.
1214 Extend file to final size via :manpage:`ftruncate(2)`
1215 instead of allocating.
1218 Backward-compatible alias for **none**.
1221 Backward-compatible alias for **posix**.
1223 May not be available on all supported platforms. **keep** is only available
1224 on Linux. If using ZFS on Solaris this cannot be set to **posix**
1225 because ZFS doesn't support pre-allocation. Default: **native** if any
1226 pre-allocation methods except **truncate** are available, **none** if not.
1228 Note that using **truncate** on Windows will interact surprisingly
1229 with non-sequential write patterns. When writing to a file that has
1230 been extended by setting the end-of-file information, Windows will
1231 backfill the unwritten portion of the file up to that offset with
1232 zeroes before issuing the new write. This means that a single small
1233 write to the end of an extended file will stall until the entire
1234 file has been filled with zeroes.
1236 .. option:: fadvise_hint=str
1238 Use :manpage:`posix_fadvise(2)` or :manpage:`posix_fadvise(2)` to
1239 advise the kernel on what I/O patterns are likely to be issued.
1240 Accepted values are:
1243 Backwards-compatible hint for "no hint".
1246 Backwards compatible hint for "advise with fio workload type". This
1247 uses **FADV_RANDOM** for a random workload, and **FADV_SEQUENTIAL**
1248 for a sequential workload.
1251 Advise using **FADV_SEQUENTIAL**.
1254 Advise using **FADV_RANDOM**.
1256 .. option:: write_hint=str
1258 Use :manpage:`fcntl(2)` to advise the kernel what life time to expect
1259 from a write. Only supported on Linux, as of version 4.13. Accepted
1263 No particular life time associated with this file.
1266 Data written to this file has a short life time.
1269 Data written to this file has a medium life time.
1272 Data written to this file has a long life time.
1275 Data written to this file has a very long life time.
1277 The values are all relative to each other, and no absolute meaning
1278 should be associated with them.
1280 .. option:: offset=int
1282 Start I/O at the provided offset in the file, given as either a fixed size in
1283 bytes, zones or a percentage. If a percentage is given, the generated offset will be
1284 aligned to the minimum ``blocksize`` or to the value of ``offset_align`` if
1285 provided. Data before the given offset will not be touched. This
1286 effectively caps the file size at `real_size - offset`. Can be combined with
1287 :option:`size` to constrain the start and end range of the I/O workload.
1288 A percentage can be specified by a number between 1 and 100 followed by '%',
1289 for example, ``offset=20%`` to specify 20%. In ZBD mode, value can be set as
1290 number of zones using 'z'.
1292 .. option:: offset_align=int
1294 If set to non-zero value, the byte offset generated by a percentage ``offset``
1295 is aligned upwards to this value. Defaults to 0 meaning that a percentage
1296 offset is aligned to the minimum block size.
1298 .. option:: offset_increment=int
1300 If this is provided, then the real offset becomes `offset + offset_increment
1301 * thread_number`, where the thread number is a counter that starts at 0 and
1302 is incremented for each sub-job (i.e. when :option:`numjobs` option is
1303 specified). This option is useful if there are several jobs which are
1304 intended to operate on a file in parallel disjoint segments, with even
1305 spacing between the starting points. Percentages can be used for this option.
1306 If a percentage is given, the generated offset will be aligned to the minimum
1307 ``blocksize`` or to the value of ``offset_align`` if provided. In ZBD mode, value can
1308 also be set as number of zones using 'z'.
1310 .. option:: number_ios=int
1312 Fio will normally perform I/Os until it has exhausted the size of the region
1313 set by :option:`size`, or if it exhaust the allocated time (or hits an error
1314 condition). With this setting, the range/size can be set independently of
1315 the number of I/Os to perform. When fio reaches this number, it will exit
1316 normally and report status. Note that this does not extend the amount of I/O
1317 that will be done, it will only stop fio if this condition is met before
1318 other end-of-job criteria.
1320 .. option:: fsync=int
1322 If writing to a file, issue an :manpage:`fsync(2)` (or its equivalent) of
1323 the dirty data for every number of blocks given. For example, if you give 32
1324 as a parameter, fio will sync the file after every 32 writes issued. If fio is
1325 using non-buffered I/O, we may not sync the file. The exception is the sg
1326 I/O engine, which synchronizes the disk cache anyway. Defaults to 0, which
1327 means fio does not periodically issue and wait for a sync to complete. Also
1328 see :option:`end_fsync` and :option:`fsync_on_close`.
1330 .. option:: fdatasync=int
1332 Like :option:`fsync` but uses :manpage:`fdatasync(2)` to only sync data and
1333 not metadata blocks. In Windows, FreeBSD, DragonFlyBSD or OSX there is no
1334 :manpage:`fdatasync(2)` so this falls back to using :manpage:`fsync(2)`.
1335 Defaults to 0, which means fio does not periodically issue and wait for a
1336 data-only sync to complete.
1338 .. option:: write_barrier=int
1340 Make every `N-th` write a barrier write.
1342 .. option:: sync_file_range=str:int
1344 Use :manpage:`sync_file_range(2)` for every `int` number of write
1345 operations. Fio will track range of writes that have happened since the last
1346 :manpage:`sync_file_range(2)` call. `str` can currently be one or more of:
1349 SYNC_FILE_RANGE_WAIT_BEFORE
1351 SYNC_FILE_RANGE_WRITE
1353 SYNC_FILE_RANGE_WAIT_AFTER
1355 So if you do ``sync_file_range=wait_before,write:8``, fio would use
1356 ``SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE`` for every 8
1357 writes. Also see the :manpage:`sync_file_range(2)` man page. This option is
1360 .. option:: overwrite=bool
1362 If true, writes to a file will always overwrite existing data. If the file
1363 doesn't already exist, it will be created before the write phase begins. If
1364 the file exists and is large enough for the specified write phase, nothing
1365 will be done. Default: false.
1367 .. option:: end_fsync=bool
1369 If true, :manpage:`fsync(2)` file contents when a write stage has completed.
1372 .. option:: fsync_on_close=bool
1374 If true, fio will :manpage:`fsync(2)` a dirty file on close. This differs
1375 from :option:`end_fsync` in that it will happen on every file close, not
1376 just at the end of the job. Default: false.
1378 .. option:: rwmixread=int
1380 Percentage of a mixed workload that should be reads. Default: 50.
1382 .. option:: rwmixwrite=int
1384 Percentage of a mixed workload that should be writes. If both
1385 :option:`rwmixread` and :option:`rwmixwrite` is given and the values do not
1386 add up to 100%, the latter of the two will be used to override the
1387 first. This may interfere with a given rate setting, if fio is asked to
1388 limit reads or writes to a certain rate. If that is the case, then the
1389 distribution may be skewed. Default: 50.
1391 .. option:: random_distribution=str:float[:float][,str:float][,str:float]
1393 By default, fio will use a completely uniform random distribution when asked
1394 to perform random I/O. Sometimes it is useful to skew the distribution in
1395 specific ways, ensuring that some parts of the data is more hot than others.
1396 fio includes the following distribution models:
1399 Uniform random distribution
1408 Normal (Gaussian) distribution
1411 Zoned random distribution
1414 Zone absolute random distribution
1416 When using a **zipf** or **pareto** distribution, an input value is also
1417 needed to define the access pattern. For **zipf**, this is the `Zipf
1418 theta`. For **pareto**, it's the `Pareto power`. Fio includes a test
1419 program, :command:`fio-genzipf`, that can be used visualize what the given input
1420 values will yield in terms of hit rates. If you wanted to use **zipf** with
1421 a `theta` of 1.2, you would use ``random_distribution=zipf:1.2`` as the
1422 option. If a non-uniform model is used, fio will disable use of the random
1423 map. For the **normal** distribution, a normal (Gaussian) deviation is
1424 supplied as a value between 0 and 100.
1426 The second, optional float is allowed for **pareto**, **zipf** and **normal** distributions.
1427 It allows to set base of distribution in non-default place, giving more control
1428 over most probable outcome. This value is in range [0-1] which maps linearly to
1429 range of possible random values.
1430 Defaults are: random for **pareto** and **zipf**, and 0.5 for **normal**.
1431 If you wanted to use **zipf** with a `theta` of 1.2 centered on 1/4 of allowed value range,
1432 you would use ``random_distibution=zipf:1.2:0.25``.
1434 For a **zoned** distribution, fio supports specifying percentages of I/O
1435 access that should fall within what range of the file or device. For
1436 example, given a criteria of:
1438 * 60% of accesses should be to the first 10%
1439 * 30% of accesses should be to the next 20%
1440 * 8% of accesses should be to the next 30%
1441 * 2% of accesses should be to the next 40%
1443 we can define that through zoning of the random accesses. For the above
1444 example, the user would do::
1446 random_distribution=zoned:60/10:30/20:8/30:2/40
1448 A **zoned_abs** distribution works exactly like the **zoned**, except
1449 that it takes absolute sizes. For example, let's say you wanted to
1450 define access according to the following criteria:
1452 * 60% of accesses should be to the first 20G
1453 * 30% of accesses should be to the next 100G
1454 * 10% of accesses should be to the next 500G
1456 we can define an absolute zoning distribution with:
1458 random_distribution=zoned_abs=60/20G:30/100G:10/500g
1460 For both **zoned** and **zoned_abs**, fio supports defining up to
1463 Similarly to how :option:`bssplit` works for setting ranges and
1464 percentages of block sizes. Like :option:`bssplit`, it's possible to
1465 specify separate zones for reads, writes, and trims. If just one set
1466 is given, it'll apply to all of them. This goes for both **zoned**
1467 **zoned_abs** distributions.
1469 .. option:: percentage_random=int[,int][,int]
1471 For a random workload, set how big a percentage should be random. This
1472 defaults to 100%, in which case the workload is fully random. It can be set
1473 from anywhere from 0 to 100. Setting it to 0 would make the workload fully
1474 sequential. Any setting in between will result in a random mix of sequential
1475 and random I/O, at the given percentages. Comma-separated values may be
1476 specified for reads, writes, and trims as described in :option:`blocksize`.
1478 .. option:: norandommap
1480 Normally fio will cover every block of the file when doing random I/O. If
1481 this option is given, fio will just get a new random offset without looking
1482 at past I/O history. This means that some blocks may not be read or written,
1483 and that some blocks may be read/written more than once. If this option is
1484 used with :option:`verify` and multiple blocksizes (via :option:`bsrange`),
1485 only intact blocks are verified, i.e., partially-overwritten blocks are
1486 ignored. With an async I/O engine and an I/O depth > 1, it is possible for
1487 the same block to be overwritten, which can cause verification errors. Either
1488 do not use norandommap in this case, or also use the lfsr random generator.
1490 .. option:: softrandommap=bool
1492 See :option:`norandommap`. If fio runs with the random block map enabled and
1493 it fails to allocate the map, if this option is set it will continue without
1494 a random block map. As coverage will not be as complete as with random maps,
1495 this option is disabled by default.
1497 .. option:: random_generator=str
1499 Fio supports the following engines for generating I/O offsets for random I/O:
1502 Strong 2^88 cycle random number generator.
1504 Linear feedback shift register generator.
1506 Strong 64-bit 2^258 cycle random number generator.
1508 **tausworthe** is a strong random number generator, but it requires tracking
1509 on the side if we want to ensure that blocks are only read or written
1510 once. **lfsr** guarantees that we never generate the same offset twice, and
1511 it's also less computationally expensive. It's not a true random generator,
1512 however, though for I/O purposes it's typically good enough. **lfsr** only
1513 works with single block sizes, not with workloads that use multiple block
1514 sizes. If used with such a workload, fio may read or write some blocks
1515 multiple times. The default value is **tausworthe**, unless the required
1516 space exceeds 2^32 blocks. If it does, then **tausworthe64** is
1517 selected automatically.
1523 .. option:: blocksize=int[,int][,int], bs=int[,int][,int]
1525 The block size in bytes used for I/O units. Default: 4096. A single value
1526 applies to reads, writes, and trims. Comma-separated values may be
1527 specified for reads, writes, and trims. A value not terminated in a comma
1528 applies to subsequent types.
1533 means 256k for reads, writes and trims.
1536 means 8k for reads, 32k for writes and trims.
1539 means 8k for reads, 32k for writes, and default for trims.
1542 means default for reads, 8k for writes and trims.
1545 means default for reads, 8k for writes, and default for trims.
1547 .. option:: blocksize_range=irange[,irange][,irange], bsrange=irange[,irange][,irange]
1549 A range of block sizes in bytes for I/O units. The issued I/O unit will
1550 always be a multiple of the minimum size, unless
1551 :option:`blocksize_unaligned` is set.
1553 Comma-separated ranges may be specified for reads, writes, and trims as
1554 described in :option:`blocksize`.
1556 Example: ``bsrange=1k-4k,2k-8k``.
1558 .. option:: bssplit=str[,str][,str]
1560 Sometimes you want even finer grained control of the block sizes
1561 issued, not just an even split between them. This option allows you to
1562 weight various block sizes, so that you are able to define a specific
1563 amount of block sizes issued. The format for this option is::
1565 bssplit=blocksize/percentage:blocksize/percentage
1567 for as many block sizes as needed. So if you want to define a workload
1568 that has 50% 64k blocks, 10% 4k blocks, and 40% 32k blocks, you would
1571 bssplit=4k/10:64k/50:32k/40
1573 Ordering does not matter. If the percentage is left blank, fio will
1574 fill in the remaining values evenly. So a bssplit option like this one::
1576 bssplit=4k/50:1k/:32k/
1578 would have 50% 4k ios, and 25% 1k and 32k ios. The percentages always
1579 add up to 100, if bssplit is given a range that adds up to more, it
1582 Comma-separated values may be specified for reads, writes, and trims as
1583 described in :option:`blocksize`.
1585 If you want a workload that has 50% 2k reads and 50% 4k reads, while
1586 having 90% 4k writes and 10% 8k writes, you would specify::
1588 bssplit=2k/50:4k/50,4k/90:8k/10
1590 Fio supports defining up to 64 different weights for each data
1593 .. option:: blocksize_unaligned, bs_unaligned
1595 If set, fio will issue I/O units with any size within
1596 :option:`blocksize_range`, not just multiples of the minimum size. This
1597 typically won't work with direct I/O, as that normally requires sector
1600 .. option:: bs_is_seq_rand=bool
1602 If this option is set, fio will use the normal read,write blocksize settings
1603 as sequential,random blocksize settings instead. Any random read or write
1604 will use the WRITE blocksize settings, and any sequential read or write will
1605 use the READ blocksize settings.
1607 .. option:: blockalign=int[,int][,int], ba=int[,int][,int]
1609 Boundary to which fio will align random I/O units. Default:
1610 :option:`blocksize`. Minimum alignment is typically 512b for using direct
1611 I/O, though it usually depends on the hardware block size. This option is
1612 mutually exclusive with using a random map for files, so it will turn off
1613 that option. Comma-separated values may be specified for reads, writes, and
1614 trims as described in :option:`blocksize`.
1620 .. option:: zero_buffers
1622 Initialize buffers with all zeros. Default: fill buffers with random data.
1624 .. option:: refill_buffers
1626 If this option is given, fio will refill the I/O buffers on every
1627 submit. Only makes sense if :option:`zero_buffers` isn't specified,
1628 naturally. Defaults to being unset i.e., the buffer is only filled at
1629 init time and the data in it is reused when possible but if any of
1630 :option:`verify`, :option:`buffer_compress_percentage` or
1631 :option:`dedupe_percentage` are enabled then `refill_buffers` is also
1632 automatically enabled.
1634 .. option:: scramble_buffers=bool
1636 If :option:`refill_buffers` is too costly and the target is using data
1637 deduplication, then setting this option will slightly modify the I/O buffer
1638 contents to defeat normal de-dupe attempts. This is not enough to defeat
1639 more clever block compression attempts, but it will stop naive dedupe of
1640 blocks. Default: true.
1642 .. option:: buffer_compress_percentage=int
1644 If this is set, then fio will attempt to provide I/O buffer content
1645 (on WRITEs) that compresses to the specified level. Fio does this by
1646 providing a mix of random data followed by fixed pattern data. The
1647 fixed pattern is either zeros, or the pattern specified by
1648 :option:`buffer_pattern`. If the `buffer_pattern` option is used, it
1649 might skew the compression ratio slightly. Setting
1650 `buffer_compress_percentage` to a value other than 100 will also
1651 enable :option:`refill_buffers` in order to reduce the likelihood that
1652 adjacent blocks are so similar that they over compress when seen
1653 together. See :option:`buffer_compress_chunk` for how to set a finer or
1654 coarser granularity for the random/fixed data region. Defaults to unset
1655 i.e., buffer data will not adhere to any compression level.
1657 .. option:: buffer_compress_chunk=int
1659 This setting allows fio to manage how big the random/fixed data region
1660 is when using :option:`buffer_compress_percentage`. When
1661 `buffer_compress_chunk` is set to some non-zero value smaller than the
1662 block size, fio can repeat the random/fixed region throughout the I/O
1663 buffer at the specified interval (which particularly useful when
1664 bigger block sizes are used for a job). When set to 0, fio will use a
1665 chunk size that matches the block size resulting in a single
1666 random/fixed region within the I/O buffer. Defaults to 512. When the
1667 unit is omitted, the value is interpreted in bytes.
1669 .. option:: buffer_pattern=str
1671 If set, fio will fill the I/O buffers with this pattern or with the contents
1672 of a file. If not set, the contents of I/O buffers are defined by the other
1673 options related to buffer contents. The setting can be any pattern of bytes,
1674 and can be prefixed with 0x for hex values. It may also be a string, where
1675 the string must then be wrapped with ``""``. Or it may also be a filename,
1676 where the filename must be wrapped with ``''`` in which case the file is
1677 opened and read. Note that not all the file contents will be read if that
1678 would cause the buffers to overflow. So, for example::
1680 buffer_pattern='filename'
1684 buffer_pattern="abcd"
1692 buffer_pattern=0xdeadface
1694 Also you can combine everything together in any order::
1696 buffer_pattern=0xdeadface"abcd"-12'filename'
1698 .. option:: dedupe_percentage=int
1700 If set, fio will generate this percentage of identical buffers when
1701 writing. These buffers will be naturally dedupable. The contents of the
1702 buffers depend on what other buffer compression settings have been set. It's
1703 possible to have the individual buffers either fully compressible, or not at
1704 all -- this option only controls the distribution of unique buffers. Setting
1705 this option will also enable :option:`refill_buffers` to prevent every buffer
1708 .. option:: invalidate=bool
1710 Invalidate the buffer/page cache parts of the files to be used prior to
1711 starting I/O if the platform and file type support it. Defaults to true.
1712 This will be ignored if :option:`pre_read` is also specified for the
1715 .. option:: sync=str
1717 Whether, and what type, of synchronous I/O to use for writes. The allowed
1721 Do not use synchronous IO, the default.
1727 Use synchronous file IO. For the majority of I/O engines,
1728 this means using O_SYNC.
1734 Use synchronous data IO. For the majority of I/O engines,
1735 this means using O_DSYNC.
1738 .. option:: iomem=str, mem=str
1740 Fio can use various types of memory as the I/O unit buffer. The allowed
1744 Use memory from :manpage:`malloc(3)` as the buffers. Default memory
1748 Use shared memory as the buffers. Allocated through
1749 :manpage:`shmget(2)`.
1752 Same as shm, but use huge pages as backing.
1755 Use :manpage:`mmap(2)` to allocate buffers. May either be anonymous memory, or can
1756 be file backed if a filename is given after the option. The format
1757 is `mem=mmap:/path/to/file`.
1760 Use a memory mapped huge file as the buffer backing. Append filename
1761 after mmaphuge, ala `mem=mmaphuge:/hugetlbfs/file`.
1764 Same as mmap, but use a MMAP_SHARED mapping.
1767 Use GPU memory as the buffers for GPUDirect RDMA benchmark.
1768 The :option:`ioengine` must be `rdma`.
1770 The area allocated is a function of the maximum allowed bs size for the job,
1771 multiplied by the I/O depth given. Note that for **shmhuge** and
1772 **mmaphuge** to work, the system must have free huge pages allocated. This
1773 can normally be checked and set by reading/writing
1774 :file:`/proc/sys/vm/nr_hugepages` on a Linux system. Fio assumes a huge page
1775 is 4MiB in size. So to calculate the number of huge pages you need for a
1776 given job file, add up the I/O depth of all jobs (normally one unless
1777 :option:`iodepth` is used) and multiply by the maximum bs set. Then divide
1778 that number by the huge page size. You can see the size of the huge pages in
1779 :file:`/proc/meminfo`. If no huge pages are allocated by having a non-zero
1780 number in `nr_hugepages`, using **mmaphuge** or **shmhuge** will fail. Also
1781 see :option:`hugepage-size`.
1783 **mmaphuge** also needs to have hugetlbfs mounted and the file location
1784 should point there. So if it's mounted in :file:`/huge`, you would use
1785 `mem=mmaphuge:/huge/somefile`.
1787 .. option:: iomem_align=int, mem_align=int
1789 This indicates the memory alignment of the I/O memory buffers. Note that
1790 the given alignment is applied to the first I/O unit buffer, if using
1791 :option:`iodepth` the alignment of the following buffers are given by the
1792 :option:`bs` used. In other words, if using a :option:`bs` that is a
1793 multiple of the page sized in the system, all buffers will be aligned to
1794 this value. If using a :option:`bs` that is not page aligned, the alignment
1795 of subsequent I/O memory buffers is the sum of the :option:`iomem_align` and
1798 .. option:: hugepage-size=int
1800 Defines the size of a huge page. Must at least be equal to the system
1801 setting, see :file:`/proc/meminfo`. Defaults to 4MiB. Should probably
1802 always be a multiple of megabytes, so using ``hugepage-size=Xm`` is the
1803 preferred way to set this to avoid setting a non-pow-2 bad value.
1805 .. option:: lockmem=int
1807 Pin the specified amount of memory with :manpage:`mlock(2)`. Can be used to
1808 simulate a smaller amount of memory. The amount specified is per worker.
1814 .. option:: size=int
1816 The total size of file I/O for each thread of this job. Fio will run until
1817 this many bytes has been transferred, unless runtime is limited by other options
1818 (such as :option:`runtime`, for instance, or increased/decreased by :option:`io_size`).
1819 Fio will divide this size between the available files determined by options
1820 such as :option:`nrfiles`, :option:`filename`, unless :option:`filesize` is
1821 specified by the job. If the result of division happens to be 0, the size is
1822 set to the physical size of the given files or devices if they exist.
1823 If this option is not specified, fio will use the full size of the given
1824 files or devices. If the files do not exist, size must be given. It is also
1825 possible to give size as a percentage between 1 and 100. If ``size=20%`` is
1826 given, fio will use 20% of the full size of the given files or devices.
1827 In ZBD mode, value can also be set as number of zones using 'z'.
1828 Can be combined with :option:`offset` to constrain the start and end range
1829 that I/O will be done within.
1831 .. option:: io_size=int, io_limit=int
1833 Normally fio operates within the region set by :option:`size`, which means
1834 that the :option:`size` option sets both the region and size of I/O to be
1835 performed. Sometimes that is not what you want. With this option, it is
1836 possible to define just the amount of I/O that fio should do. For instance,
1837 if :option:`size` is set to 20GiB and :option:`io_size` is set to 5GiB, fio
1838 will perform I/O within the first 20GiB but exit when 5GiB have been
1839 done. The opposite is also possible -- if :option:`size` is set to 20GiB,
1840 and :option:`io_size` is set to 40GiB, then fio will do 40GiB of I/O within
1841 the 0..20GiB region.
1843 .. option:: filesize=irange(int)
1845 Individual file sizes. May be a range, in which case fio will select sizes
1846 for files at random within the given range and limited to :option:`size` in
1847 total (if that is given). If not given, each created file is the same size.
1848 This option overrides :option:`size` in terms of file size, which means
1849 this value is used as a fixed size or possible range of each file.
1851 .. option:: file_append=bool
1853 Perform I/O after the end of the file. Normally fio will operate within the
1854 size of a file. If this option is set, then fio will append to the file
1855 instead. This has identical behavior to setting :option:`offset` to the size
1856 of a file. This option is ignored on non-regular files.
1858 .. option:: fill_device=bool, fill_fs=bool
1860 Sets size to something really large and waits for ENOSPC (no space left on
1861 device) as the terminating condition. Only makes sense with sequential
1862 write. For a read workload, the mount point will be filled first then I/O
1863 started on the result. This option doesn't make sense if operating on a raw
1864 device node, since the size of that is already known by the file system.
1865 Additionally, writing beyond end-of-device will not return ENOSPC there.
1871 .. option:: ioengine=str
1873 Defines how the job issues I/O to the file. The following types are defined:
1876 Basic :manpage:`read(2)` or :manpage:`write(2)`
1877 I/O. :manpage:`lseek(2)` is used to position the I/O location.
1878 See :option:`fsync` and :option:`fdatasync` for syncing write I/Os.
1881 Basic :manpage:`pread(2)` or :manpage:`pwrite(2)` I/O. Default on
1882 all supported operating systems except for Windows.
1885 Basic :manpage:`readv(2)` or :manpage:`writev(2)` I/O. Will emulate
1886 queuing by coalescing adjacent I/Os into a single submission.
1889 Basic :manpage:`preadv(2)` or :manpage:`pwritev(2)` I/O.
1892 Basic :manpage:`preadv2(2)` or :manpage:`pwritev2(2)` I/O.
1895 Fast Linux native asynchronous I/O. Supports async IO
1896 for both direct and buffered IO.
1897 This engine defines engine specific options.
1900 Linux native asynchronous I/O. Note that Linux may only support
1901 queued behavior with non-buffered I/O (set ``direct=1`` or
1903 This engine defines engine specific options.
1906 POSIX asynchronous I/O using :manpage:`aio_read(3)` and
1907 :manpage:`aio_write(3)`.
1910 Solaris native asynchronous I/O.
1913 Windows native asynchronous I/O. Default on Windows.
1916 File is memory mapped with :manpage:`mmap(2)` and data copied
1917 to/from using :manpage:`memcpy(3)`.
1920 :manpage:`splice(2)` is used to transfer the data and
1921 :manpage:`vmsplice(2)` to transfer data from user space to the
1925 SCSI generic sg v3 I/O. May either be synchronous using the SG_IO
1926 ioctl, or if the target is an sg character device we use
1927 :manpage:`read(2)` and :manpage:`write(2)` for asynchronous
1928 I/O. Requires :option:`filename` option to specify either block or
1929 character devices. This engine supports trim operations.
1930 The sg engine includes engine specific options.
1933 Doesn't transfer any data, just pretends to. This is mainly used to
1934 exercise fio itself and for debugging/testing purposes.
1937 Transfer over the network to given ``host:port``. Depending on the
1938 :option:`protocol` used, the :option:`hostname`, :option:`port`,
1939 :option:`listen` and :option:`filename` options are used to specify
1940 what sort of connection to make, while the :option:`protocol` option
1941 determines which protocol will be used. This engine defines engine
1945 Like **net**, but uses :manpage:`splice(2)` and
1946 :manpage:`vmsplice(2)` to map data and send/receive.
1947 This engine defines engine specific options.
1950 Doesn't transfer any data, but burns CPU cycles according to the
1951 :option:`cpuload`, :option:`cpuchunks` and :option:`cpumode` options.
1952 Setting :option:`cpuload`\=85 will cause that job to do nothing but burn 85%
1953 of the CPU. In case of SMP machines, use :option:`numjobs`\=<nr_of_cpu>
1954 to get desired CPU usage, as the cpuload only loads a
1955 single CPU at the desired rate. A job never finishes unless there is
1956 at least one non-cpuio job.
1957 Setting :option:`cpumode`\=qsort replace the default noop instructions loop
1958 by a qsort algorithm to consume more energy.
1961 The RDMA I/O engine supports both RDMA memory semantics
1962 (RDMA_WRITE/RDMA_READ) and channel semantics (Send/Recv) for the
1963 InfiniBand, RoCE and iWARP protocols. This engine defines engine
1967 I/O engine that does regular fallocate to simulate data transfer as
1971 does fallocate(,mode = FALLOC_FL_KEEP_SIZE,).
1974 does fallocate(,mode = 0).
1977 does fallocate(,mode = FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE).
1980 I/O engine that sends :manpage:`ftruncate(2)` operations in response
1981 to write (DDIR_WRITE) events. Each ftruncate issued sets the file's
1982 size to the current block offset. :option:`blocksize` is ignored.
1985 I/O engine that does regular EXT4_IOC_MOVE_EXT ioctls to simulate
1986 defragment activity in request to DDIR_WRITE event.
1989 I/O engine supporting direct access to Ceph Reliable Autonomic
1990 Distributed Object Store (RADOS) via librados. This ioengine
1991 defines engine specific options.
1994 I/O engine supporting direct access to Ceph Rados Block Devices
1995 (RBD) via librbd without the need to use the kernel rbd driver. This
1996 ioengine defines engine specific options.
1999 I/O engine supporting GET/PUT requests over HTTP(S) with libcurl to
2000 a WebDAV or S3 endpoint. This ioengine defines engine specific options.
2002 This engine only supports direct IO of iodepth=1; you need to scale this
2003 via numjobs. blocksize defines the size of the objects to be created.
2005 TRIM is translated to object deletion.
2008 Using GlusterFS libgfapi sync interface to direct access to
2009 GlusterFS volumes without having to go through FUSE. This ioengine
2010 defines engine specific options.
2013 Using GlusterFS libgfapi async interface to direct access to
2014 GlusterFS volumes without having to go through FUSE. This ioengine
2015 defines engine specific options.
2018 Read and write through Hadoop (HDFS). The :option:`filename` option
2019 is used to specify host,port of the hdfs name-node to connect. This
2020 engine interprets offsets a little differently. In HDFS, files once
2021 created cannot be modified so random writes are not possible. To
2022 imitate this the libhdfs engine expects a bunch of small files to be
2023 created over HDFS and will randomly pick a file from them
2024 based on the offset generated by fio backend (see the example
2025 job file to create such files, use ``rw=write`` option). Please
2026 note, it may be necessary to set environment variables to work
2027 with HDFS/libhdfs properly. Each job uses its own connection to
2031 Read, write and erase an MTD character device (e.g.,
2032 :file:`/dev/mtd0`). Discards are treated as erases. Depending on the
2033 underlying device type, the I/O may have to go in a certain pattern,
2034 e.g., on NAND, writing sequentially to erase blocks and discarding
2035 before overwriting. The `trimwrite` mode works well for this
2039 Read and write using filesystem DAX to a file on a filesystem
2040 mounted with DAX on a persistent memory device through the PMDK
2044 Read and write using device DAX to a persistent memory device (e.g.,
2045 /dev/dax0.0) through the PMDK libpmem library.
2048 Prefix to specify loading an external I/O engine object file. Append
2049 the engine filename, e.g. ``ioengine=external:/tmp/foo.o`` to load
2050 ioengine :file:`foo.o` in :file:`/tmp`. The path can be either
2051 absolute or relative. See :file:`engines/skeleton_external.c` for
2052 details of writing an external I/O engine.
2055 Simply create the files and do no I/O to them. You still need to
2056 set `filesize` so that all the accounting still occurs, but no
2057 actual I/O will be done other than creating the file.
2060 Simply do stat() and do no I/O to the file. You need to set 'filesize'
2061 and 'nrfiles', so that files will be created.
2062 This engine is to measure file lookup and meta data access.
2065 Simply delete the files by unlink() and do no I/O to them. You need to set 'filesize'
2066 and 'nrfiles', so that the files will be created.
2067 This engine is to measure file delete.
2070 Read and write using mmap I/O to a file on a filesystem
2071 mounted with DAX on a persistent memory device through the PMDK
2075 Synchronous read and write using DDN's Infinite Memory Engine (IME).
2076 This engine is very basic and issues calls to IME whenever an IO is
2080 Synchronous read and write using DDN's Infinite Memory Engine (IME).
2081 This engine uses iovecs and will try to stack as much IOs as possible
2082 (if the IOs are "contiguous" and the IO depth is not exceeded)
2083 before issuing a call to IME.
2086 Asynchronous read and write using DDN's Infinite Memory Engine (IME).
2087 This engine will try to stack as much IOs as possible by creating
2088 requests for IME. FIO will then decide when to commit these requests.
2090 Read and write iscsi lun with libiscsi.
2092 Read and write a Network Block Device (NBD).
2095 I/O engine supporting libcufile synchronous access to nvidia-fs and a
2096 GPUDirect Storage-supported filesystem. This engine performs
2097 I/O without transferring buffers between user-space and the kernel,
2098 unless :option:`verify` is set or :option:`cuda_io` is `posix`.
2099 :option:`iomem` must not be `cudamalloc`. This ioengine defines
2100 engine specific options.
2102 I/O engine supporting asynchronous read and write operations to the
2103 DAOS File System (DFS) via libdfs.
2105 I/O engine specific parameters
2106 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2108 In addition, there are some parameters which are only valid when a specific
2109 :option:`ioengine` is in use. These are used identically to normal parameters,
2110 with the caveat that when used on the command line, they must come after the
2111 :option:`ioengine` that defines them is selected.
2113 .. option:: cmdprio_percentage=int : [io_uring] [libaio]
2115 Set the percentage of I/O that will be issued with higher priority by setting
2116 the priority bit. Non-read I/O is likely unaffected by ``cmdprio_percentage``.
2117 This option cannot be used with the `prio` or `prioclass` options. For this
2118 option to set the priority bit properly, NCQ priority must be supported and
2119 enabled and :option:`direct`\=1 option must be used. fio must also be run as
2122 .. option:: fixedbufs : [io_uring]
2124 If fio is asked to do direct IO, then Linux will map pages for each
2125 IO call, and release them when IO is done. If this option is set, the
2126 pages are pre-mapped before IO is started. This eliminates the need to
2127 map and release for each IO. This is more efficient, and reduces the
2130 .. option:: hipri : [io_uring]
2132 If this option is set, fio will attempt to use polled IO completions.
2133 Normal IO completions generate interrupts to signal the completion of
2134 IO, polled completions do not. Hence they are require active reaping
2135 by the application. The benefits are more efficient IO for high IOPS
2136 scenarios, and lower latencies for low queue depth IO.
2138 .. option:: registerfiles : [io_uring]
2140 With this option, fio registers the set of files being used with the
2141 kernel. This avoids the overhead of managing file counts in the kernel,
2142 making the submission and completion part more lightweight. Required
2143 for the below :option:`sqthread_poll` option.
2145 .. option:: sqthread_poll : [io_uring]
2147 Normally fio will submit IO by issuing a system call to notify the
2148 kernel of available items in the SQ ring. If this option is set, the
2149 act of submitting IO will be done by a polling thread in the kernel.
2150 This frees up cycles for fio, at the cost of using more CPU in the
2153 .. option:: sqthread_poll_cpu : [io_uring]
2155 When :option:`sqthread_poll` is set, this option provides a way to
2156 define which CPU should be used for the polling thread.
2158 .. option:: userspace_reap : [libaio]
2160 Normally, with the libaio engine in use, fio will use the
2161 :manpage:`io_getevents(2)` system call to reap newly returned events. With
2162 this flag turned on, the AIO ring will be read directly from user-space to
2163 reap events. The reaping mode is only enabled when polling for a minimum of
2164 0 events (e.g. when :option:`iodepth_batch_complete` `=0`).
2166 .. option:: hipri : [pvsync2]
2168 Set RWF_HIPRI on I/O, indicating to the kernel that it's of higher priority
2171 .. option:: hipri_percentage : [pvsync2]
2173 When hipri is set this determines the probability of a pvsync2 I/O being high
2174 priority. The default is 100%.
2176 .. option:: nowait : [pvsync2] [libaio] [io_uring]
2178 By default if a request cannot be executed immediately (e.g. resource starvation,
2179 waiting on locks) it is queued and the initiating process will be blocked until
2180 the required resource becomes free.
2182 This option sets the RWF_NOWAIT flag (supported from the 4.14 Linux kernel) and
2183 the call will return instantly with EAGAIN or a partial result rather than waiting.
2185 It is useful to also use ignore_error=EAGAIN when using this option.
2187 Note: glibc 2.27, 2.28 have a bug in syscall wrappers preadv2, pwritev2.
2188 They return EOPNOTSUP instead of EAGAIN.
2190 For cached I/O, using this option usually means a request operates only with
2191 cached data. Currently the RWF_NOWAIT flag does not supported for cached write.
2193 For direct I/O, requests will only succeed if cache invalidation isn't required,
2194 file blocks are fully allocated and the disk request could be issued immediately.
2196 .. option:: cpuload=int : [cpuio]
2198 Attempt to use the specified percentage of CPU cycles. This is a mandatory
2199 option when using cpuio I/O engine.
2201 .. option:: cpuchunks=int : [cpuio]
2203 Split the load into cycles of the given time. In microseconds.
2205 .. option:: exit_on_io_done=bool : [cpuio]
2207 Detect when I/O threads are done, then exit.
2209 .. option:: namenode=str : [libhdfs]
2211 The hostname or IP address of a HDFS cluster namenode to contact.
2213 .. option:: port=int
2217 The listening port of the HFDS cluster namenode.
2221 The TCP or UDP port to bind to or connect to. If this is used with
2222 :option:`numjobs` to spawn multiple instances of the same job type, then
2223 this will be the starting port number since fio will use a range of
2228 The port to use for RDMA-CM communication. This should be the same value
2229 on the client and the server side.
2231 .. option:: hostname=str : [netsplice] [net] [rdma]
2233 The hostname or IP address to use for TCP, UDP or RDMA-CM based I/O. If the job
2234 is a TCP listener or UDP reader, the hostname is not used and must be omitted
2235 unless it is a valid UDP multicast address.
2237 .. option:: serverip=str : [librpma_*]
2239 The IP address to be used for RDMA-CM based I/O.
2241 .. option:: direct_write_to_pmem=bool : [librpma_*]
2243 Set to 1 only when Direct Write to PMem from the remote host is possible.
2244 Otherwise, set to 0.
2246 .. option:: busy_wait_polling=bool : [librpma_*_server]
2248 Set to 0 to wait for completion instead of busy-wait polling completion.
2251 .. option:: interface=str : [netsplice] [net]
2253 The IP address of the network interface used to send or receive UDP
2256 .. option:: ttl=int : [netsplice] [net]
2258 Time-to-live value for outgoing UDP multicast packets. Default: 1.
2260 .. option:: nodelay=bool : [netsplice] [net]
2262 Set TCP_NODELAY on TCP connections.
2264 .. option:: protocol=str, proto=str : [netsplice] [net]
2266 The network protocol to use. Accepted values are:
2269 Transmission control protocol.
2271 Transmission control protocol V6.
2273 User datagram protocol.
2275 User datagram protocol V6.
2279 When the protocol is TCP or UDP, the port must also be given, as well as the
2280 hostname if the job is a TCP listener or UDP reader. For unix sockets, the
2281 normal :option:`filename` option should be used and the port is invalid.
2283 .. option:: listen : [netsplice] [net]
2285 For TCP network connections, tell fio to listen for incoming connections
2286 rather than initiating an outgoing connection. The :option:`hostname` must
2287 be omitted if this option is used.
2289 .. option:: pingpong : [netsplice] [net]
2291 Normally a network writer will just continue writing data, and a network
2292 reader will just consume packages. If ``pingpong=1`` is set, a writer will
2293 send its normal payload to the reader, then wait for the reader to send the
2294 same payload back. This allows fio to measure network latencies. The
2295 submission and completion latencies then measure local time spent sending or
2296 receiving, and the completion latency measures how long it took for the
2297 other end to receive and send back. For UDP multicast traffic
2298 ``pingpong=1`` should only be set for a single reader when multiple readers
2299 are listening to the same address.
2301 .. option:: window_size : [netsplice] [net]
2303 Set the desired socket buffer size for the connection.
2305 .. option:: mss : [netsplice] [net]
2307 Set the TCP maximum segment size (TCP_MAXSEG).
2309 .. option:: donorname=str : [e4defrag]
2311 File will be used as a block donor (swap extents between files).
2313 .. option:: inplace=int : [e4defrag]
2315 Configure donor file blocks allocation strategy:
2318 Default. Preallocate donor's file on init.
2320 Allocate space immediately inside defragment event, and free right
2323 .. option:: clustername=str : [rbd,rados]
2325 Specifies the name of the Ceph cluster.
2327 .. option:: rbdname=str : [rbd]
2329 Specifies the name of the RBD.
2331 .. option:: pool=str : [rbd,rados]
2333 Specifies the name of the Ceph pool containing RBD or RADOS data.
2335 .. option:: clientname=str : [rbd,rados]
2337 Specifies the username (without the 'client.' prefix) used to access the
2338 Ceph cluster. If the *clustername* is specified, the *clientname* shall be
2339 the full *type.id* string. If no type. prefix is given, fio will add
2340 'client.' by default.
2342 .. option:: busy_poll=bool : [rbd,rados]
2344 Poll store instead of waiting for completion. Usually this provides better
2345 throughput at cost of higher(up to 100%) CPU utilization.
2347 .. option:: touch_objects=bool : [rados]
2349 During initialization, touch (create if do not exist) all objects (files).
2350 Touching all objects affects ceph caches and likely impacts test results.
2353 .. option:: skip_bad=bool : [mtd]
2355 Skip operations against known bad blocks.
2357 .. option:: hdfsdirectory : [libhdfs]
2359 libhdfs will create chunk in this HDFS directory.
2361 .. option:: chunk_size : [libhdfs]
2363 The size of the chunk to use for each file.
2365 .. option:: verb=str : [rdma]
2367 The RDMA verb to use on this side of the RDMA ioengine connection. Valid
2368 values are write, read, send and recv. These correspond to the equivalent
2369 RDMA verbs (e.g. write = rdma_write etc.). Note that this only needs to be
2370 specified on the client side of the connection. See the examples folder.
2372 .. option:: bindname=str : [rdma]
2374 The name to use to bind the local RDMA-CM connection to a local RDMA device.
2375 This could be a hostname or an IPv4 or IPv6 address. On the server side this
2376 will be passed into the rdma_bind_addr() function and on the client site it
2377 will be used in the rdma_resolve_add() function. This can be useful when
2378 multiple paths exist between the client and the server or in certain loopback
2381 .. option:: stat_type=str : [filestat]
2383 Specify stat system call type to measure lookup/getattr performance.
2384 Default is **stat** for :manpage:`stat(2)`.
2386 .. option:: readfua=bool : [sg]
2388 With readfua option set to 1, read operations include
2389 the force unit access (fua) flag. Default is 0.
2391 .. option:: writefua=bool : [sg]
2393 With writefua option set to 1, write operations include
2394 the force unit access (fua) flag. Default is 0.
2396 .. option:: sg_write_mode=str : [sg]
2398 Specify the type of write commands to issue. This option can take three values:
2401 This is the default where write opcodes are issued as usual.
2403 Issue WRITE AND VERIFY commands. The BYTCHK bit is set to 0. This
2404 directs the device to carry out a medium verification with no data
2405 comparison. The writefua option is ignored with this selection.
2407 Issue WRITE SAME commands. This transfers a single block to the device
2408 and writes this same block of data to a contiguous sequence of LBAs
2409 beginning at the specified offset. fio's block size parameter specifies
2410 the amount of data written with each command. However, the amount of data
2411 actually transferred to the device is equal to the device's block
2412 (sector) size. For a device with 512 byte sectors, blocksize=8k will
2413 write 16 sectors with each command. fio will still generate 8k of data
2414 for each command but only the first 512 bytes will be used and
2415 transferred to the device. The writefua option is ignored with this
2418 .. option:: hipri : [sg]
2420 If this option is set, fio will attempt to use polled IO completions.
2421 This will have a similar effect as (io_uring)hipri. Only SCSI READ and
2422 WRITE commands will have the SGV4_FLAG_HIPRI set (not UNMAP (trim) nor
2423 VERIFY). Older versions of the Linux sg driver that do not support
2424 hipri will simply ignore this flag and do normal IO. The Linux SCSI
2425 Low Level Driver (LLD) that "owns" the device also needs to support
2426 hipri (also known as iopoll and mq_poll). The MegaRAID driver is an
2427 example of a SCSI LLD. Default: clear (0) which does normal
2428 (interrupted based) IO.
2430 .. option:: http_host=str : [http]
2432 Hostname to connect to. For S3, this could be the bucket hostname.
2433 Default is **localhost**
2435 .. option:: http_user=str : [http]
2437 Username for HTTP authentication.
2439 .. option:: http_pass=str : [http]
2441 Password for HTTP authentication.
2443 .. option:: https=str : [http]
2445 Enable HTTPS instead of http. *on* enables HTTPS; *insecure*
2446 will enable HTTPS, but disable SSL peer verification (use with
2447 caution!). Default is **off**
2449 .. option:: http_mode=str : [http]
2451 Which HTTP access mode to use: *webdav*, *swift*, or *s3*.
2452 Default is **webdav**
2454 .. option:: http_s3_region=str : [http]
2456 The S3 region/zone string.
2457 Default is **us-east-1**
2459 .. option:: http_s3_key=str : [http]
2463 .. option:: http_s3_keyid=str : [http]
2465 The S3 key/access id.
2467 .. option:: http_swift_auth_token=str : [http]
2469 The Swift auth token. See the example configuration file on how
2472 .. option:: http_verbose=int : [http]
2474 Enable verbose requests from libcurl. Useful for debugging. 1
2475 turns on verbose logging from libcurl, 2 additionally enables
2476 HTTP IO tracing. Default is **0**
2478 .. option:: uri=str : [nbd]
2480 Specify the NBD URI of the server to test. The string
2481 is a standard NBD URI
2482 (see https://github.com/NetworkBlockDevice/nbd/tree/master/doc).
2483 Example URIs: nbd://localhost:10809
2484 nbd+unix:///?socket=/tmp/socket
2485 nbds://tlshost/exportname
2487 .. option:: gpu_dev_ids=str : [libcufile]
2489 Specify the GPU IDs to use with CUDA. This is a colon-separated list of
2490 int. GPUs are assigned to workers roundrobin. Default is 0.
2492 .. option:: cuda_io=str : [libcufile]
2494 Specify the type of I/O to use with CUDA. Default is **cufile**.
2497 Use libcufile and nvidia-fs. This option performs I/O directly
2498 between a GPUDirect Storage filesystem and GPU buffers,
2499 avoiding use of a bounce buffer. If :option:`verify` is set,
2500 cudaMemcpy is used to copy verificaton data between RAM and GPU.
2501 Verification data is copied from RAM to GPU before a write
2502 and from GPU to RAM after a read. :option:`direct` must be 1.
2504 Use POSIX to perform I/O with a RAM buffer, and use cudaMemcpy
2505 to transfer data between RAM and the GPUs. Data is copied from
2506 GPU to RAM before a write and copied from RAM to GPU after a
2507 read. :option:`verify` does not affect use of cudaMemcpy.
2509 .. option:: pool=str : [dfs]
2511 Specify the UUID of the DAOS pool to connect to.
2513 .. option:: cont=str : [dfs]
2515 Specify the UUID of the DAOS container to open.
2517 .. option:: chunk_size=int : [dfs]
2519 Specificy a different chunk size (in bytes) for the dfs file.
2520 Use DAOS container's chunk size by default.
2522 .. option:: object_class=str : [dfs]
2524 Specificy a different object class for the dfs file.
2525 Use DAOS container's object class by default.
2530 .. option:: iodepth=int
2532 Number of I/O units to keep in flight against the file. Note that
2533 increasing *iodepth* beyond 1 will not affect synchronous ioengines (except
2534 for small degrees when :option:`verify_async` is in use). Even async
2535 engines may impose OS restrictions causing the desired depth not to be
2536 achieved. This may happen on Linux when using libaio and not setting
2537 :option:`direct`\=1, since buffered I/O is not async on that OS. Keep an
2538 eye on the I/O depth distribution in the fio output to verify that the
2539 achieved depth is as expected. Default: 1.
2541 .. option:: iodepth_batch_submit=int, iodepth_batch=int
2543 This defines how many pieces of I/O to submit at once. It defaults to 1
2544 which means that we submit each I/O as soon as it is available, but can be
2545 raised to submit bigger batches of I/O at the time. If it is set to 0 the
2546 :option:`iodepth` value will be used.
2548 .. option:: iodepth_batch_complete_min=int, iodepth_batch_complete=int
2550 This defines how many pieces of I/O to retrieve at once. It defaults to 1
2551 which means that we'll ask for a minimum of 1 I/O in the retrieval process
2552 from the kernel. The I/O retrieval will go on until we hit the limit set by
2553 :option:`iodepth_low`. If this variable is set to 0, then fio will always
2554 check for completed events before queuing more I/O. This helps reduce I/O
2555 latency, at the cost of more retrieval system calls.
2557 .. option:: iodepth_batch_complete_max=int
2559 This defines maximum pieces of I/O to retrieve at once. This variable should
2560 be used along with :option:`iodepth_batch_complete_min`\=int variable,
2561 specifying the range of min and max amount of I/O which should be
2562 retrieved. By default it is equal to the :option:`iodepth_batch_complete_min`
2567 iodepth_batch_complete_min=1
2568 iodepth_batch_complete_max=<iodepth>
2570 which means that we will retrieve at least 1 I/O and up to the whole
2571 submitted queue depth. If none of I/O has been completed yet, we will wait.
2575 iodepth_batch_complete_min=0
2576 iodepth_batch_complete_max=<iodepth>
2578 which means that we can retrieve up to the whole submitted queue depth, but
2579 if none of I/O has been completed yet, we will NOT wait and immediately exit
2580 the system call. In this example we simply do polling.
2582 .. option:: iodepth_low=int
2584 The low water mark indicating when to start filling the queue
2585 again. Defaults to the same as :option:`iodepth`, meaning that fio will
2586 attempt to keep the queue full at all times. If :option:`iodepth` is set to
2587 e.g. 16 and *iodepth_low* is set to 4, then after fio has filled the queue of
2588 16 requests, it will let the depth drain down to 4 before starting to fill
2591 .. option:: serialize_overlap=bool
2593 Serialize in-flight I/Os that might otherwise cause or suffer from data races.
2594 When two or more I/Os are submitted simultaneously, there is no guarantee that
2595 the I/Os will be processed or completed in the submitted order. Further, if
2596 two or more of those I/Os are writes, any overlapping region between them can
2597 become indeterminate/undefined on certain storage. These issues can cause
2598 verification to fail erratically when at least one of the racing I/Os is
2599 changing data and the overlapping region has a non-zero size. Setting
2600 ``serialize_overlap`` tells fio to avoid provoking this behavior by explicitly
2601 serializing in-flight I/Os that have a non-zero overlap. Note that setting
2602 this option can reduce both performance and the :option:`iodepth` achieved.
2604 This option only applies to I/Os issued for a single job except when it is
2605 enabled along with :option:`io_submit_mode`\=offload. In offload mode, fio
2606 will check for overlap among all I/Os submitted by offload jobs with :option:`serialize_overlap`
2611 .. option:: io_submit_mode=str
2613 This option controls how fio submits the I/O to the I/O engine. The default
2614 is `inline`, which means that the fio job threads submit and reap I/O
2615 directly. If set to `offload`, the job threads will offload I/O submission
2616 to a dedicated pool of I/O threads. This requires some coordination and thus
2617 has a bit of extra overhead, especially for lower queue depth I/O where it
2618 can increase latencies. The benefit is that fio can manage submission rates
2619 independently of the device completion rates. This avoids skewed latency
2620 reporting if I/O gets backed up on the device side (the coordinated omission
2621 problem). Note that this option cannot reliably be used with async IO
2628 .. option:: thinktime=time
2630 Stall the job for the specified period of time after an I/O has completed before issuing the
2631 next. May be used to simulate processing being done by an application.
2632 When the unit is omitted, the value is interpreted in microseconds. See
2633 :option:`thinktime_blocks` and :option:`thinktime_spin`.
2635 .. option:: thinktime_spin=time
2637 Only valid if :option:`thinktime` is set - pretend to spend CPU time doing
2638 something with the data received, before falling back to sleeping for the
2639 rest of the period specified by :option:`thinktime`. When the unit is
2640 omitted, the value is interpreted in microseconds.
2642 .. option:: thinktime_blocks=int
2644 Only valid if :option:`thinktime` is set - control how many blocks to issue,
2645 before waiting :option:`thinktime` usecs. If not set, defaults to 1 which will make
2646 fio wait :option:`thinktime` usecs after every block. This effectively makes any
2647 queue depth setting redundant, since no more than 1 I/O will be queued
2648 before we have to complete it and do our :option:`thinktime`. In other words, this
2649 setting effectively caps the queue depth if the latter is larger.
2651 .. option:: thinktime_blocks_type=str
2653 Only valid if :option:`thinktime` is set - control how :option:`thinktime_blocks`
2654 triggers. The default is `complete`, which triggers thinktime when fio completes
2655 :option:`thinktime_blocks` blocks. If this is set to `issue`, then the trigger happens
2658 .. option:: rate=int[,int][,int]
2660 Cap the bandwidth used by this job. The number is in bytes/sec, the normal
2661 suffix rules apply. Comma-separated values may be specified for reads,
2662 writes, and trims as described in :option:`blocksize`.
2664 For example, using `rate=1m,500k` would limit reads to 1MiB/sec and writes to
2665 500KiB/sec. Capping only reads or writes can be done with `rate=,500k` or
2666 `rate=500k,` where the former will only limit writes (to 500KiB/sec) and the
2667 latter will only limit reads.
2669 .. option:: rate_min=int[,int][,int]
2671 Tell fio to do whatever it can to maintain at least this bandwidth. Failing
2672 to meet this requirement will cause the job to exit. Comma-separated values
2673 may be specified for reads, writes, and trims as described in
2674 :option:`blocksize`.
2676 .. option:: rate_iops=int[,int][,int]
2678 Cap the bandwidth to this number of IOPS. Basically the same as
2679 :option:`rate`, just specified independently of bandwidth. If the job is
2680 given a block size range instead of a fixed value, the smallest block size
2681 is used as the metric. Comma-separated values may be specified for reads,
2682 writes, and trims as described in :option:`blocksize`.
2684 .. option:: rate_iops_min=int[,int][,int]
2686 If fio doesn't meet this rate of I/O, it will cause the job to exit.
2687 Comma-separated values may be specified for reads, writes, and trims as
2688 described in :option:`blocksize`.
2690 .. option:: rate_process=str
2692 This option controls how fio manages rated I/O submissions. The default is
2693 `linear`, which submits I/O in a linear fashion with fixed delays between
2694 I/Os that gets adjusted based on I/O completion rates. If this is set to
2695 `poisson`, fio will submit I/O based on a more real world random request
2696 flow, known as the Poisson process
2697 (https://en.wikipedia.org/wiki/Poisson_point_process). The lambda will be
2698 10^6 / IOPS for the given workload.
2700 .. option:: rate_ignore_thinktime=bool
2702 By default, fio will attempt to catch up to the specified rate setting,
2703 if any kind of thinktime setting was used. If this option is set, then
2704 fio will ignore the thinktime and continue doing IO at the specified
2705 rate, instead of entering a catch-up mode after thinktime is done.
2711 .. option:: latency_target=time
2713 If set, fio will attempt to find the max performance point that the given
2714 workload will run at while maintaining a latency below this target. When
2715 the unit is omitted, the value is interpreted in microseconds. See
2716 :option:`latency_window` and :option:`latency_percentile`.
2718 .. option:: latency_window=time
2720 Used with :option:`latency_target` to specify the sample window that the job
2721 is run at varying queue depths to test the performance. When the unit is
2722 omitted, the value is interpreted in microseconds.
2724 .. option:: latency_percentile=float
2726 The percentage of I/Os that must fall within the criteria specified by
2727 :option:`latency_target` and :option:`latency_window`. If not set, this
2728 defaults to 100.0, meaning that all I/Os must be equal or below to the value
2729 set by :option:`latency_target`.
2731 .. option:: latency_run=bool
2733 Used with :option:`latency_target`. If false (default), fio will find
2734 the highest queue depth that meets :option:`latency_target` and exit. If
2735 true, fio will continue running and try to meet :option:`latency_target`
2736 by adjusting queue depth.
2738 .. option:: max_latency=time[,time][,time]
2740 If set, fio will exit the job with an ETIMEDOUT error if it exceeds this
2741 maximum latency. When the unit is omitted, the value is interpreted in
2742 microseconds. Comma-separated values may be specified for reads, writes,
2743 and trims as described in :option:`blocksize`.
2745 .. option:: rate_cycle=int
2747 Average bandwidth for :option:`rate` and :option:`rate_min` over this number
2748 of milliseconds. Defaults to 1000.
2754 .. option:: write_iolog=str
2756 Write the issued I/O patterns to the specified file. See
2757 :option:`read_iolog`. Specify a separate file for each job, otherwise the
2758 iologs will be interspersed and the file may be corrupt.
2760 .. option:: read_iolog=str
2762 Open an iolog with the specified filename and replay the I/O patterns it
2763 contains. This can be used to store a workload and replay it sometime
2764 later. The iolog given may also be a blktrace binary file, which allows fio
2765 to replay a workload captured by :command:`blktrace`. See
2766 :manpage:`blktrace(8)` for how to capture such logging data. For blktrace
2767 replay, the file needs to be turned into a blkparse binary data file first
2768 (``blkparse <device> -o /dev/null -d file_for_fio.bin``).
2769 You can specify a number of files by separating the names with a ':'
2770 character. See the :option:`filename` option for information on how to
2771 escape ':' characters within the file names. These files will
2772 be sequentially assigned to job clones created by :option:`numjobs`.
2773 '-' is a reserved name, meaning read from stdin, notably if
2774 :option:`filename` is set to '-' which means stdin as well, then
2775 this flag can't be set to '-'.
2777 .. option:: read_iolog_chunked=bool
2779 Determines how iolog is read. If false(default) entire :option:`read_iolog`
2780 will be read at once. If selected true, input from iolog will be read
2781 gradually. Useful when iolog is very large, or it is generated.
2783 .. option:: merge_blktrace_file=str
2785 When specified, rather than replaying the logs passed to :option:`read_iolog`,
2786 the logs go through a merge phase which aggregates them into a single
2787 blktrace. The resulting file is then passed on as the :option:`read_iolog`
2788 parameter. The intention here is to make the order of events consistent.
2789 This limits the influence of the scheduler compared to replaying multiple
2790 blktraces via concurrent jobs.
2792 .. option:: merge_blktrace_scalars=float_list
2794 This is a percentage based option that is index paired with the list of
2795 files passed to :option:`read_iolog`. When merging is performed, scale
2796 the time of each event by the corresponding amount. For example,
2797 ``--merge_blktrace_scalars="50:100"`` runs the first trace in halftime
2798 and the second trace in realtime. This knob is separately tunable from
2799 :option:`replay_time_scale` which scales the trace during runtime and
2800 does not change the output of the merge unlike this option.
2802 .. option:: merge_blktrace_iters=float_list
2804 This is a whole number option that is index paired with the list of files
2805 passed to :option:`read_iolog`. When merging is performed, run each trace
2806 for the specified number of iterations. For example,
2807 ``--merge_blktrace_iters="2:1"`` runs the first trace for two iterations
2808 and the second trace for one iteration.
2810 .. option:: replay_no_stall=bool
2812 When replaying I/O with :option:`read_iolog` the default behavior is to
2813 attempt to respect the timestamps within the log and replay them with the
2814 appropriate delay between IOPS. By setting this variable fio will not
2815 respect the timestamps and attempt to replay them as fast as possible while
2816 still respecting ordering. The result is the same I/O pattern to a given
2817 device, but different timings.
2819 .. option:: replay_time_scale=int
2821 When replaying I/O with :option:`read_iolog`, fio will honor the
2822 original timing in the trace. With this option, it's possible to scale
2823 the time. It's a percentage option, if set to 50 it means run at 50%
2824 the original IO rate in the trace. If set to 200, run at twice the
2825 original IO rate. Defaults to 100.
2827 .. option:: replay_redirect=str
2829 While replaying I/O patterns using :option:`read_iolog` the default behavior
2830 is to replay the IOPS onto the major/minor device that each IOP was recorded
2831 from. This is sometimes undesirable because on a different machine those
2832 major/minor numbers can map to a different device. Changing hardware on the
2833 same system can also result in a different major/minor mapping.
2834 ``replay_redirect`` causes all I/Os to be replayed onto the single specified
2835 device regardless of the device it was recorded
2836 from. i.e. :option:`replay_redirect`\= :file:`/dev/sdc` would cause all I/O
2837 in the blktrace or iolog to be replayed onto :file:`/dev/sdc`. This means
2838 multiple devices will be replayed onto a single device, if the trace
2839 contains multiple devices. If you want multiple devices to be replayed
2840 concurrently to multiple redirected devices you must blkparse your trace
2841 into separate traces and replay them with independent fio invocations.
2842 Unfortunately this also breaks the strict time ordering between multiple
2845 .. option:: replay_align=int
2847 Force alignment of the byte offsets in a trace to this value. The value
2848 must be a power of 2.
2850 .. option:: replay_scale=int
2852 Scale byte offsets down by this factor when replaying traces. Should most
2853 likely use :option:`replay_align` as well.
2855 .. option:: replay_skip=str
2857 Sometimes it's useful to skip certain IO types in a replay trace.
2858 This could be, for instance, eliminating the writes in the trace.
2859 Or not replaying the trims/discards, if you are redirecting to
2860 a device that doesn't support them. This option takes a comma
2861 separated list of read, write, trim, sync.
2864 Threads, processes and job synchronization
2865 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2869 Fio defaults to creating jobs by using fork, however if this option is
2870 given, fio will create jobs by using POSIX Threads' function
2871 :manpage:`pthread_create(3)` to create threads instead.
2873 .. option:: wait_for=str
2875 If set, the current job won't be started until all workers of the specified
2876 waitee job are done.
2878 ``wait_for`` operates on the job name basis, so there are a few
2879 limitations. First, the waitee must be defined prior to the waiter job
2880 (meaning no forward references). Second, if a job is being referenced as a
2881 waitee, it must have a unique name (no duplicate waitees).
2883 .. option:: nice=int
2885 Run the job with the given nice value. See man :manpage:`nice(2)`.
2887 On Windows, values less than -15 set the process class to "High"; -1 through
2888 -15 set "Above Normal"; 1 through 15 "Below Normal"; and above 15 "Idle"
2891 .. option:: prio=int
2893 Set the I/O priority value of this job. Linux limits us to a positive value
2894 between 0 and 7, with 0 being the highest. See man
2895 :manpage:`ionice(1)`. Refer to an appropriate manpage for other operating
2896 systems since meaning of priority may differ. For per-command priority
2897 setting, see I/O engine specific `cmdprio_percentage` and `hipri_percentage`
2900 .. option:: prioclass=int
2902 Set the I/O priority class. See man :manpage:`ionice(1)`. For per-command
2903 priority setting, see I/O engine specific `cmdprio_percentage` and
2904 `hipri_percentage` options.
2906 .. option:: cpus_allowed=str
2908 Controls the same options as :option:`cpumask`, but accepts a textual
2909 specification of the permitted CPUs instead and CPUs are indexed from 0. So
2910 to use CPUs 0 and 5 you would specify ``cpus_allowed=0,5``. This option also
2911 allows a range of CPUs to be specified -- say you wanted a binding to CPUs
2912 0, 5, and 8 to 15, you would set ``cpus_allowed=0,5,8-15``.
2914 On Windows, when ``cpus_allowed`` is unset only CPUs from fio's current
2915 processor group will be used and affinity settings are inherited from the
2916 system. An fio build configured to target Windows 7 makes options that set
2917 CPUs processor group aware and values will set both the processor group
2918 and a CPU from within that group. For example, on a system where processor
2919 group 0 has 40 CPUs and processor group 1 has 32 CPUs, ``cpus_allowed``
2920 values between 0 and 39 will bind CPUs from processor group 0 and
2921 ``cpus_allowed`` values between 40 and 71 will bind CPUs from processor
2922 group 1. When using ``cpus_allowed_policy=shared`` all CPUs specified by a
2923 single ``cpus_allowed`` option must be from the same processor group. For
2924 Windows fio builds not built for Windows 7, CPUs will only be selected from
2925 (and be relative to) whatever processor group fio happens to be running in
2926 and CPUs from other processor groups cannot be used.
2928 .. option:: cpus_allowed_policy=str
2930 Set the policy of how fio distributes the CPUs specified by
2931 :option:`cpus_allowed` or :option:`cpumask`. Two policies are supported:
2934 All jobs will share the CPU set specified.
2936 Each job will get a unique CPU from the CPU set.
2938 **shared** is the default behavior, if the option isn't specified. If
2939 **split** is specified, then fio will assign one cpu per job. If not
2940 enough CPUs are given for the jobs listed, then fio will roundrobin the CPUs
2943 .. option:: cpumask=int
2945 Set the CPU affinity of this job. The parameter given is a bit mask of
2946 allowed CPUs the job may run on. So if you want the allowed CPUs to be 1
2947 and 5, you would pass the decimal value of (1 << 1 | 1 << 5), or 34. See man
2948 :manpage:`sched_setaffinity(2)`. This may not work on all supported
2949 operating systems or kernel versions. This option doesn't work well for a
2950 higher CPU count than what you can store in an integer mask, so it can only
2951 control cpus 1-32. For boxes with larger CPU counts, use
2952 :option:`cpus_allowed`.
2954 .. option:: numa_cpu_nodes=str
2956 Set this job running on specified NUMA nodes' CPUs. The arguments allow
2957 comma delimited list of cpu numbers, A-B ranges, or `all`. Note, to enable
2958 NUMA options support, fio must be built on a system with libnuma-dev(el)
2961 .. option:: numa_mem_policy=str
2963 Set this job's memory policy and corresponding NUMA nodes. Format of the
2968 ``mode`` is one of the following memory policies: ``default``, ``prefer``,
2969 ``bind``, ``interleave`` or ``local``. For ``default`` and ``local`` memory
2970 policies, no node needs to be specified. For ``prefer``, only one node is
2971 allowed. For ``bind`` and ``interleave`` the ``nodelist`` may be as
2972 follows: a comma delimited list of numbers, A-B ranges, or `all`.
2974 .. option:: cgroup=str
2976 Add job to this control group. If it doesn't exist, it will be created. The
2977 system must have a mounted cgroup blkio mount point for this to work. If
2978 your system doesn't have it mounted, you can do so with::
2980 # mount -t cgroup -o blkio none /cgroup
2982 .. option:: cgroup_weight=int
2984 Set the weight of the cgroup to this value. See the documentation that comes
2985 with the kernel, allowed values are in the range of 100..1000.
2987 .. option:: cgroup_nodelete=bool
2989 Normally fio will delete the cgroups it has created after the job
2990 completion. To override this behavior and to leave cgroups around after the
2991 job completion, set ``cgroup_nodelete=1``. This can be useful if one wants
2992 to inspect various cgroup files after job completion. Default: false.
2994 .. option:: flow_id=int
2996 The ID of the flow. If not specified, it defaults to being a global
2997 flow. See :option:`flow`.
2999 .. option:: flow=int
3001 Weight in token-based flow control. If this value is used, then there is a
3002 'flow counter' which is used to regulate the proportion of activity between
3003 two or more jobs. Fio attempts to keep this flow counter near zero. The
3004 ``flow`` parameter stands for how much should be added or subtracted to the
3005 flow counter on each iteration of the main I/O loop. That is, if one job has
3006 ``flow=8`` and another job has ``flow=-1``, then there will be a roughly 1:8
3007 ratio in how much one runs vs the other.
3009 .. option:: flow_sleep=int
3011 The period of time, in microseconds, to wait after the flow counter
3012 has exceeded its proportion before retrying operations.
3014 .. option:: stonewall, wait_for_previous
3016 Wait for preceding jobs in the job file to exit, before starting this
3017 one. Can be used to insert serialization points in the job file. A stone
3018 wall also implies starting a new reporting group, see
3019 :option:`group_reporting`.
3023 By default, fio will continue running all other jobs when one job finishes.
3024 Sometimes this is not the desired action. Setting ``exitall`` will instead
3025 make fio terminate all jobs in the same group, as soon as one job of that
3028 .. option:: exit_what
3030 By default, fio will continue running all other jobs when one job finishes.
3031 Sometimes this is not the desired action. Setting ``exit_all`` will
3032 instead make fio terminate all jobs in the same group. The option
3033 ``exit_what`` allows to control which jobs get terminated when ``exitall`` is
3034 enabled. The default is ``group`` and does not change the behaviour of
3035 ``exitall``. The setting ``all`` terminates all jobs. The setting ``stonewall``
3036 terminates all currently running jobs across all groups and continues execution
3037 with the next stonewalled group.
3039 .. option:: exec_prerun=str
3041 Before running this job, issue the command specified through
3042 :manpage:`system(3)`. Output is redirected in a file called
3043 :file:`jobname.prerun.txt`.
3045 .. option:: exec_postrun=str
3047 After the job completes, issue the command specified though
3048 :manpage:`system(3)`. Output is redirected in a file called
3049 :file:`jobname.postrun.txt`.
3053 Instead of running as the invoking user, set the user ID to this value
3054 before the thread/process does any work.
3058 Set group ID, see :option:`uid`.
3064 .. option:: verify_only
3066 Do not perform specified workload, only verify data still matches previous
3067 invocation of this workload. This option allows one to check data multiple
3068 times at a later date without overwriting it. This option makes sense only
3069 for workloads that write data, and does not support workloads with the
3070 :option:`time_based` option set.
3072 .. option:: do_verify=bool
3074 Run the verify phase after a write phase. Only valid if :option:`verify` is
3077 .. option:: verify=str
3079 If writing to a file, fio can verify the file contents after each iteration
3080 of the job. Each verification method also implies verification of special
3081 header, which is written to the beginning of each block. This header also
3082 includes meta information, like offset of the block, block number, timestamp
3083 when block was written, etc. :option:`verify` can be combined with
3084 :option:`verify_pattern` option. The allowed values are:
3087 Use an md5 sum of the data area and store it in the header of
3091 Use an experimental crc64 sum of the data area and store it in the
3092 header of each block.
3095 Use a crc32c sum of the data area and store it in the header of
3096 each block. This will automatically use hardware acceleration
3097 (e.g. SSE4.2 on an x86 or CRC crypto extensions on ARM64) but will
3098 fall back to software crc32c if none is found. Generally the
3099 fastest checksum fio supports when hardware accelerated.
3105 Use a crc32 sum of the data area and store it in the header of each
3109 Use a crc16 sum of the data area and store it in the header of each
3113 Use a crc7 sum of the data area and store it in the header of each
3117 Use xxhash as the checksum function. Generally the fastest software
3118 checksum that fio supports.
3121 Use sha512 as the checksum function.
3124 Use sha256 as the checksum function.
3127 Use optimized sha1 as the checksum function.
3130 Use optimized sha3-224 as the checksum function.
3133 Use optimized sha3-256 as the checksum function.
3136 Use optimized sha3-384 as the checksum function.
3139 Use optimized sha3-512 as the checksum function.
3142 This option is deprecated, since now meta information is included in
3143 generic verification header and meta verification happens by
3144 default. For detailed information see the description of the
3145 :option:`verify` setting. This option is kept because of
3146 compatibility's sake with old configurations. Do not use it.
3149 Verify a strict pattern. Normally fio includes a header with some
3150 basic information and checksumming, but if this option is set, only
3151 the specific pattern set with :option:`verify_pattern` is verified.
3154 Only pretend to verify. Useful for testing internals with
3155 :option:`ioengine`\=null, not for much else.
3157 This option can be used for repeated burn-in tests of a system to make sure
3158 that the written data is also correctly read back. If the data direction
3159 given is a read or random read, fio will assume that it should verify a
3160 previously written file. If the data direction includes any form of write,
3161 the verify will be of the newly written data.
3163 To avoid false verification errors, do not use the norandommap option when
3164 verifying data with async I/O engines and I/O depths > 1. Or use the
3165 norandommap and the lfsr random generator together to avoid writing to the
3166 same offset with muliple outstanding I/Os.
3168 .. option:: verify_offset=int
3170 Swap the verification header with data somewhere else in the block before
3171 writing. It is swapped back before verifying.
3173 .. option:: verify_interval=int
3175 Write the verification header at a finer granularity than the
3176 :option:`blocksize`. It will be written for chunks the size of
3177 ``verify_interval``. :option:`blocksize` should divide this evenly.
3179 .. option:: verify_pattern=str
3181 If set, fio will fill the I/O buffers with this pattern. Fio defaults to
3182 filling with totally random bytes, but sometimes it's interesting to fill
3183 with a known pattern for I/O verification purposes. Depending on the width
3184 of the pattern, fio will fill 1/2/3/4 bytes of the buffer at the time (it can
3185 be either a decimal or a hex number). The ``verify_pattern`` if larger than
3186 a 32-bit quantity has to be a hex number that starts with either "0x" or
3187 "0X". Use with :option:`verify`. Also, ``verify_pattern`` supports %o
3188 format, which means that for each block offset will be written and then
3189 verified back, e.g.::
3193 Or use combination of everything::
3195 verify_pattern=0xff%o"abcd"-12
3197 .. option:: verify_fatal=bool
3199 Normally fio will keep checking the entire contents before quitting on a
3200 block verification failure. If this option is set, fio will exit the job on
3201 the first observed failure. Default: false.
3203 .. option:: verify_dump=bool
3205 If set, dump the contents of both the original data block and the data block
3206 we read off disk to files. This allows later analysis to inspect just what
3207 kind of data corruption occurred. Off by default.
3209 .. option:: verify_async=int
3211 Fio will normally verify I/O inline from the submitting thread. This option
3212 takes an integer describing how many async offload threads to create for I/O
3213 verification instead, causing fio to offload the duty of verifying I/O
3214 contents to one or more separate threads. If using this offload option, even
3215 sync I/O engines can benefit from using an :option:`iodepth` setting higher
3216 than 1, as it allows them to have I/O in flight while verifies are running.
3217 Defaults to 0 async threads, i.e. verification is not asynchronous.
3219 .. option:: verify_async_cpus=str
3221 Tell fio to set the given CPU affinity on the async I/O verification
3222 threads. See :option:`cpus_allowed` for the format used.
3224 .. option:: verify_backlog=int
3226 Fio will normally verify the written contents of a job that utilizes verify
3227 once that job has completed. In other words, everything is written then
3228 everything is read back and verified. You may want to verify continually
3229 instead for a variety of reasons. Fio stores the meta data associated with
3230 an I/O block in memory, so for large verify workloads, quite a bit of memory
3231 would be used up holding this meta data. If this option is enabled, fio will
3232 write only N blocks before verifying these blocks.
3234 .. option:: verify_backlog_batch=int
3236 Control how many blocks fio will verify if :option:`verify_backlog` is
3237 set. If not set, will default to the value of :option:`verify_backlog`
3238 (meaning the entire queue is read back and verified). If
3239 ``verify_backlog_batch`` is less than :option:`verify_backlog` then not all
3240 blocks will be verified, if ``verify_backlog_batch`` is larger than
3241 :option:`verify_backlog`, some blocks will be verified more than once.
3243 .. option:: verify_state_save=bool
3245 When a job exits during the write phase of a verify workload, save its
3246 current state. This allows fio to replay up until that point, if the verify
3247 state is loaded for the verify read phase. The format of the filename is,
3250 <type>-<jobname>-<jobindex>-verify.state.
3252 <type> is "local" for a local run, "sock" for a client/server socket
3253 connection, and "ip" (192.168.0.1, for instance) for a networked
3254 client/server connection. Defaults to true.
3256 .. option:: verify_state_load=bool
3258 If a verify termination trigger was used, fio stores the current write state
3259 of each thread. This can be used at verification time so that fio knows how
3260 far it should verify. Without this information, fio will run a full
3261 verification pass, according to the settings in the job file used. Default
3264 .. option:: trim_percentage=int
3266 Number of verify blocks to discard/trim.
3268 .. option:: trim_verify_zero=bool
3270 Verify that trim/discarded blocks are returned as zeros.
3272 .. option:: trim_backlog=int
3274 Trim after this number of blocks are written.
3276 .. option:: trim_backlog_batch=int
3278 Trim this number of I/O blocks.
3280 .. option:: experimental_verify=bool
3282 Enable experimental verification.
3287 .. option:: steadystate=str:float, ss=str:float
3289 Define the criterion and limit for assessing steady state performance. The
3290 first parameter designates the criterion whereas the second parameter sets
3291 the threshold. When the criterion falls below the threshold for the
3292 specified duration, the job will stop. For example, `iops_slope:0.1%` will
3293 direct fio to terminate the job when the least squares regression slope
3294 falls below 0.1% of the mean IOPS. If :option:`group_reporting` is enabled
3295 this will apply to all jobs in the group. Below is the list of available
3296 steady state assessment criteria. All assessments are carried out using only
3297 data from the rolling collection window. Threshold limits can be expressed
3298 as a fixed value or as a percentage of the mean in the collection window.
3300 When using this feature, most jobs should include the :option:`time_based`
3301 and :option:`runtime` options or the :option:`loops` option so that fio does not
3302 stop running after it has covered the full size of the specified file(s) or device(s).
3305 Collect IOPS data. Stop the job if all individual IOPS measurements
3306 are within the specified limit of the mean IOPS (e.g., ``iops:2``
3307 means that all individual IOPS values must be within 2 of the mean,
3308 whereas ``iops:0.2%`` means that all individual IOPS values must be
3309 within 0.2% of the mean IOPS to terminate the job).
3312 Collect IOPS data and calculate the least squares regression
3313 slope. Stop the job if the slope falls below the specified limit.
3316 Collect bandwidth data. Stop the job if all individual bandwidth
3317 measurements are within the specified limit of the mean bandwidth.
3320 Collect bandwidth data and calculate the least squares regression
3321 slope. Stop the job if the slope falls below the specified limit.
3323 .. option:: steadystate_duration=time, ss_dur=time
3325 A rolling window of this duration will be used to judge whether steady state
3326 has been reached. Data will be collected once per second. The default is 0
3327 which disables steady state detection. When the unit is omitted, the
3328 value is interpreted in seconds.
3330 .. option:: steadystate_ramp_time=time, ss_ramp=time
3332 Allow the job to run for the specified duration before beginning data
3333 collection for checking the steady state job termination criterion. The
3334 default is 0. When the unit is omitted, the value is interpreted in seconds.
3337 Measurements and reporting
3338 ~~~~~~~~~~~~~~~~~~~~~~~~~~
3340 .. option:: per_job_logs=bool
3342 If set, this generates bw/clat/iops log with per file private filenames. If
3343 not set, jobs with identical names will share the log filename. Default:
3346 .. option:: group_reporting
3348 It may sometimes be interesting to display statistics for groups of jobs as
3349 a whole instead of for each individual job. This is especially true if
3350 :option:`numjobs` is used; looking at individual thread/process output
3351 quickly becomes unwieldy. To see the final report per-group instead of
3352 per-job, use :option:`group_reporting`. Jobs in a file will be part of the
3353 same reporting group, unless if separated by a :option:`stonewall`, or by
3354 using :option:`new_group`.
3356 .. option:: new_group
3358 Start a new reporting group. See: :option:`group_reporting`. If not given,
3359 all jobs in a file will be part of the same reporting group, unless
3360 separated by a :option:`stonewall`.
3362 .. option:: stats=bool
3364 By default, fio collects and shows final output results for all jobs
3365 that run. If this option is set to 0, then fio will ignore it in
3366 the final stat output.
3368 .. option:: write_bw_log=str
3370 If given, write a bandwidth log for this job. Can be used to store data of
3371 the bandwidth of the jobs in their lifetime.
3373 If no str argument is given, the default filename of
3374 :file:`jobname_type.x.log` is used. Even when the argument is given, fio
3375 will still append the type of log. So if one specifies::
3379 The actual log name will be :file:`foo_bw.x.log` where `x` is the index
3380 of the job (`1..N`, where `N` is the number of jobs). If
3381 :option:`per_job_logs` is false, then the filename will not include the
3384 The included :command:`fio_generate_plots` script uses :command:`gnuplot` to turn these
3385 text files into nice graphs. See `Log File Formats`_ for how data is
3386 structured within the file.
3388 .. option:: write_lat_log=str
3390 Same as :option:`write_bw_log`, except this option creates I/O
3391 submission (e.g., :file:`name_slat.x.log`), completion (e.g.,
3392 :file:`name_clat.x.log`), and total (e.g., :file:`name_lat.x.log`)
3393 latency files instead. See :option:`write_bw_log` for details about
3394 the filename format and `Log File Formats`_ for how data is structured
3397 .. option:: write_hist_log=str
3399 Same as :option:`write_bw_log` but writes an I/O completion latency
3400 histogram file (e.g., :file:`name_hist.x.log`) instead. Note that this
3401 file will be empty unless :option:`log_hist_msec` has also been set.
3402 See :option:`write_bw_log` for details about the filename format and
3403 `Log File Formats`_ for how data is structured within the file.
3405 .. option:: write_iops_log=str
3407 Same as :option:`write_bw_log`, but writes an IOPS file (e.g.
3408 :file:`name_iops.x.log`) instead. Because fio defaults to individual
3409 I/O logging, the value entry in the IOPS log will be 1 unless windowed
3410 logging (see :option:`log_avg_msec`) has been enabled. See
3411 :option:`write_bw_log` for details about the filename format and `Log
3412 File Formats`_ for how data is structured within the file.
3414 .. option:: log_avg_msec=int
3416 By default, fio will log an entry in the iops, latency, or bw log for every
3417 I/O that completes. When writing to the disk log, that can quickly grow to a
3418 very large size. Setting this option makes fio average the each log entry
3419 over the specified period of time, reducing the resolution of the log. See
3420 :option:`log_max_value` as well. Defaults to 0, logging all entries.
3421 Also see `Log File Formats`_.
3423 .. option:: log_hist_msec=int
3425 Same as :option:`log_avg_msec`, but logs entries for completion latency
3426 histograms. Computing latency percentiles from averages of intervals using
3427 :option:`log_avg_msec` is inaccurate. Setting this option makes fio log
3428 histogram entries over the specified period of time, reducing log sizes for
3429 high IOPS devices while retaining percentile accuracy. See
3430 :option:`log_hist_coarseness` and :option:`write_hist_log` as well.
3431 Defaults to 0, meaning histogram logging is disabled.
3433 .. option:: log_hist_coarseness=int
3435 Integer ranging from 0 to 6, defining the coarseness of the resolution of
3436 the histogram logs enabled with :option:`log_hist_msec`. For each increment
3437 in coarseness, fio outputs half as many bins. Defaults to 0, for which
3438 histogram logs contain 1216 latency bins. See :option:`write_hist_log`
3439 and `Log File Formats`_.
3441 .. option:: log_max_value=bool
3443 If :option:`log_avg_msec` is set, fio logs the average over that window. If
3444 you instead want to log the maximum value, set this option to 1. Defaults to
3445 0, meaning that averaged values are logged.
3447 .. option:: log_offset=bool
3449 If this is set, the iolog options will include the byte offset for the I/O
3450 entry as well as the other data values. Defaults to 0 meaning that
3451 offsets are not present in logs. Also see `Log File Formats`_.
3453 .. option:: log_compression=int
3455 If this is set, fio will compress the I/O logs as it goes, to keep the
3456 memory footprint lower. When a log reaches the specified size, that chunk is
3457 removed and compressed in the background. Given that I/O logs are fairly
3458 highly compressible, this yields a nice memory savings for longer runs. The
3459 downside is that the compression will consume some background CPU cycles, so
3460 it may impact the run. This, however, is also true if the logging ends up
3461 consuming most of the system memory. So pick your poison. The I/O logs are
3462 saved normally at the end of a run, by decompressing the chunks and storing
3463 them in the specified log file. This feature depends on the availability of
3466 .. option:: log_compression_cpus=str
3468 Define the set of CPUs that are allowed to handle online log compression for
3469 the I/O jobs. This can provide better isolation between performance
3470 sensitive jobs, and background compression work. See
3471 :option:`cpus_allowed` for the format used.
3473 .. option:: log_store_compressed=bool
3475 If set, fio will store the log files in a compressed format. They can be
3476 decompressed with fio, using the :option:`--inflate-log` command line
3477 parameter. The files will be stored with a :file:`.fz` suffix.
3479 .. option:: log_unix_epoch=bool
3481 If set, fio will log Unix timestamps to the log files produced by enabling
3482 write_type_log for each log type, instead of the default zero-based
3485 .. option:: block_error_percentiles=bool
3487 If set, record errors in trim block-sized units from writes and trims and
3488 output a histogram of how many trims it took to get to errors, and what kind
3489 of error was encountered.
3491 .. option:: bwavgtime=int
3493 Average the calculated bandwidth over the given time. Value is specified in
3494 milliseconds. If the job also does bandwidth logging through
3495 :option:`write_bw_log`, then the minimum of this option and
3496 :option:`log_avg_msec` will be used. Default: 500ms.
3498 .. option:: iopsavgtime=int
3500 Average the calculated IOPS over the given time. Value is specified in
3501 milliseconds. If the job also does IOPS logging through
3502 :option:`write_iops_log`, then the minimum of this option and
3503 :option:`log_avg_msec` will be used. Default: 500ms.
3505 .. option:: disk_util=bool
3507 Generate disk utilization statistics, if the platform supports it.
3510 .. option:: disable_lat=bool
3512 Disable measurements of total latency numbers. Useful only for cutting back
3513 the number of calls to :manpage:`gettimeofday(2)`, as that does impact
3514 performance at really high IOPS rates. Note that to really get rid of a
3515 large amount of these calls, this option must be used with
3516 :option:`disable_slat` and :option:`disable_bw_measurement` as well.
3518 .. option:: disable_clat=bool
3520 Disable measurements of completion latency numbers. See
3521 :option:`disable_lat`.
3523 .. option:: disable_slat=bool
3525 Disable measurements of submission latency numbers. See
3526 :option:`disable_lat`.
3528 .. option:: disable_bw_measurement=bool, disable_bw=bool
3530 Disable measurements of throughput/bandwidth numbers. See
3531 :option:`disable_lat`.
3533 .. option:: slat_percentiles=bool
3535 Report submission latency percentiles. Submission latency is not recorded
3536 for synchronous ioengines.
3538 .. option:: clat_percentiles=bool
3540 Report completion latency percentiles.
3542 .. option:: lat_percentiles=bool
3544 Report total latency percentiles. Total latency is the sum of submission
3545 latency and completion latency.
3547 .. option:: percentile_list=float_list
3549 Overwrite the default list of percentiles for latencies and the block error
3550 histogram. Each number is a floating point number in the range (0,100], and
3551 the maximum length of the list is 20. Use ``:`` to separate the numbers. For
3552 example, ``--percentile_list=99.5:99.9`` will cause fio to report the
3553 latency durations below which 99.5% and 99.9% of the observed latencies fell,
3556 .. option:: significant_figures=int
3558 If using :option:`--output-format` of `normal`, set the significant
3559 figures to this value. Higher values will yield more precise IOPS and
3560 throughput units, while lower values will round. Requires a minimum
3561 value of 1 and a maximum value of 10. Defaults to 4.
3567 .. option:: exitall_on_error
3569 When one job finishes in error, terminate the rest. The default is to wait
3570 for each job to finish.
3572 .. option:: continue_on_error=str
3574 Normally fio will exit the job on the first observed failure. If this option
3575 is set, fio will continue the job when there is a 'non-fatal error' (EIO or
3576 EILSEQ) until the runtime is exceeded or the I/O size specified is
3577 completed. If this option is used, there are two more stats that are
3578 appended, the total error count and the first error. The error field given
3579 in the stats is the first error that was hit during the run.
3581 The allowed values are:
3584 Exit on any I/O or verify errors.
3587 Continue on read errors, exit on all others.
3590 Continue on write errors, exit on all others.
3593 Continue on any I/O error, exit on all others.
3596 Continue on verify errors, exit on all others.
3599 Continue on all errors.
3602 Backward-compatible alias for 'none'.
3605 Backward-compatible alias for 'all'.
3607 .. option:: ignore_error=str
3609 Sometimes you want to ignore some errors during test in that case you can
3610 specify error list for each error type, instead of only being able to
3611 ignore the default 'non-fatal error' using :option:`continue_on_error`.
3612 ``ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST`` errors for
3613 given error type is separated with ':'. Error may be symbol ('ENOSPC',
3614 'ENOMEM') or integer. Example::
3616 ignore_error=EAGAIN,ENOSPC:122
3618 This option will ignore EAGAIN from READ, and ENOSPC and 122(EDQUOT) from
3619 WRITE. This option works by overriding :option:`continue_on_error` with
3620 the list of errors for each error type if any.
3622 .. option:: error_dump=bool
3624 If set dump every error even if it is non fatal, true by default. If
3625 disabled only fatal error will be dumped.
3627 Running predefined workloads
3628 ----------------------------
3630 Fio includes predefined profiles that mimic the I/O workloads generated by
3633 .. option:: profile=str
3635 The predefined workload to run. Current profiles are:
3638 Threaded I/O bench (tiotest/tiobench) like workload.
3641 Aerospike Certification Tool (ACT) like workload.
3643 To view a profile's additional options use :option:`--cmdhelp` after specifying
3644 the profile. For example::
3646 $ fio --profile=act --cmdhelp
3651 .. option:: device-names=str
3656 .. option:: load=int
3659 ACT load multiplier. Default: 1.
3661 .. option:: test-duration=time
3664 How long the entire test takes to run. When the unit is omitted, the value
3665 is given in seconds. Default: 24h.
3667 .. option:: threads-per-queue=int
3670 Number of read I/O threads per device. Default: 8.
3672 .. option:: read-req-num-512-blocks=int
3675 Number of 512B blocks to read at the time. Default: 3.
3677 .. option:: large-block-op-kbytes=int
3680 Size of large block ops in KiB (writes). Default: 131072.
3685 Set to run ACT prep phase.
3687 Tiobench profile options
3688 ~~~~~~~~~~~~~~~~~~~~~~~~
3690 .. option:: size=str
3695 .. option:: block=int
3698 Block size in bytes. Default: 4096.
3700 .. option:: numruns=int
3710 .. option:: threads=int
3715 Interpreting the output
3716 -----------------------
3719 Example output was based on the following:
3720 TZ=UTC fio --iodepth=8 --ioengine=null --size=100M --time_based \
3721 --rate=1256k --bs=14K --name=quick --runtime=1s --name=mixed \
3722 --runtime=2m --rw=rw
3724 Fio spits out a lot of output. While running, fio will display the status of the
3725 jobs created. An example of that would be::
3727 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]
3729 The characters inside the first set of square brackets denote the current status of
3730 each thread. The first character is the first job defined in the job file, and so
3731 forth. The possible values (in typical life cycle order) are:
3733 +------+-----+-----------------------------------------------------------+
3735 +======+=====+===========================================================+
3736 | P | | Thread setup, but not started. |
3737 +------+-----+-----------------------------------------------------------+
3738 | C | | Thread created. |
3739 +------+-----+-----------------------------------------------------------+
3740 | I | | Thread initialized, waiting or generating necessary data. |
3741 +------+-----+-----------------------------------------------------------+
3742 | | p | Thread running pre-reading file(s). |
3743 +------+-----+-----------------------------------------------------------+
3744 | | / | Thread is in ramp period. |
3745 +------+-----+-----------------------------------------------------------+
3746 | | R | Running, doing sequential reads. |
3747 +------+-----+-----------------------------------------------------------+
3748 | | r | Running, doing random reads. |
3749 +------+-----+-----------------------------------------------------------+
3750 | | W | Running, doing sequential writes. |
3751 +------+-----+-----------------------------------------------------------+
3752 | | w | Running, doing random writes. |
3753 +------+-----+-----------------------------------------------------------+
3754 | | M | Running, doing mixed sequential reads/writes. |
3755 +------+-----+-----------------------------------------------------------+
3756 | | m | Running, doing mixed random reads/writes. |
3757 +------+-----+-----------------------------------------------------------+
3758 | | D | Running, doing sequential trims. |
3759 +------+-----+-----------------------------------------------------------+
3760 | | d | Running, doing random trims. |
3761 +------+-----+-----------------------------------------------------------+
3762 | | F | Running, currently waiting for :manpage:`fsync(2)`. |
3763 +------+-----+-----------------------------------------------------------+
3764 | | V | Running, doing verification of written data. |
3765 +------+-----+-----------------------------------------------------------+
3766 | f | | Thread finishing. |
3767 +------+-----+-----------------------------------------------------------+
3768 | E | | Thread exited, not reaped by main thread yet. |
3769 +------+-----+-----------------------------------------------------------+
3770 | _ | | Thread reaped. |
3771 +------+-----+-----------------------------------------------------------+
3772 | X | | Thread reaped, exited with an error. |
3773 +------+-----+-----------------------------------------------------------+
3774 | K | | Thread reaped, exited due to signal. |
3775 +------+-----+-----------------------------------------------------------+
3778 Example output was based on the following:
3779 TZ=UTC fio --iodepth=8 --ioengine=null --size=100M --runtime=58m \
3780 --time_based --rate=2512k --bs=256K --numjobs=10 \
3781 --name=readers --rw=read --name=writers --rw=write
3783 Fio will condense the thread string as not to take up more space on the command
3784 line than needed. For instance, if you have 10 readers and 10 writers running,
3785 the output would look like this::
3787 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]
3789 Note that the status string is displayed in order, so it's possible to tell which of
3790 the jobs are currently doing what. In the example above this means that jobs 1--10
3791 are readers and 11--20 are writers.
3793 The other values are fairly self explanatory -- number of threads currently
3794 running and doing I/O, the number of currently open files (f=), the estimated
3795 completion percentage, the rate of I/O since last check (read speed listed first,
3796 then write speed and optionally trim speed) in terms of bandwidth and IOPS,
3797 and time to completion for the current running group. It's impossible to estimate
3798 runtime of the following groups (if any).
3801 Example output was based on the following:
3802 TZ=UTC fio --iodepth=16 --ioengine=posixaio --filename=/tmp/fiofile \
3803 --direct=1 --size=100M --time_based --runtime=50s --rate_iops=89 \
3804 --bs=7K --name=Client1 --rw=write
3806 When fio is done (or interrupted by :kbd:`Ctrl-C`), it will show the data for
3807 each thread, group of threads, and disks in that order. For each overall thread (or
3808 group) the output looks like::
3810 Client1: (groupid=0, jobs=1): err= 0: pid=16109: Sat Jun 24 12:07:54 2017
3811 write: IOPS=88, BW=623KiB/s (638kB/s)(30.4MiB/50032msec)
3812 slat (nsec): min=500, max=145500, avg=8318.00, stdev=4781.50
3813 clat (usec): min=170, max=78367, avg=4019.02, stdev=8293.31
3814 lat (usec): min=174, max=78375, avg=4027.34, stdev=8291.79
3815 clat percentiles (usec):
3816 | 1.00th=[ 302], 5.00th=[ 326], 10.00th=[ 343], 20.00th=[ 363],
3817 | 30.00th=[ 392], 40.00th=[ 404], 50.00th=[ 416], 60.00th=[ 445],
3818 | 70.00th=[ 816], 80.00th=[ 6718], 90.00th=[12911], 95.00th=[21627],
3819 | 99.00th=[43779], 99.50th=[51643], 99.90th=[68682], 99.95th=[72877],
3821 bw ( KiB/s): min= 532, max= 686, per=0.10%, avg=622.87, stdev=24.82, samples= 100
3822 iops : min= 76, max= 98, avg=88.98, stdev= 3.54, samples= 100
3823 lat (usec) : 250=0.04%, 500=64.11%, 750=4.81%, 1000=2.79%
3824 lat (msec) : 2=4.16%, 4=1.84%, 10=4.90%, 20=11.33%, 50=5.37%
3825 lat (msec) : 100=0.65%
3826 cpu : usr=0.27%, sys=0.18%, ctx=12072, majf=0, minf=21
3827 IO depths : 1=85.0%, 2=13.1%, 4=1.8%, 8=0.1%, 16=0.0%, 32=0.0%, >=64=0.0%
3828 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
3829 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
3830 issued rwt: total=0,4450,0, short=0,0,0, dropped=0,0,0
3831 latency : target=0, window=0, percentile=100.00%, depth=8
3833 The job name (or first job's name when using :option:`group_reporting`) is printed,
3834 along with the group id, count of jobs being aggregated, last error id seen (which
3835 is 0 when there are no errors), pid/tid of that thread and the time the job/group
3836 completed. Below are the I/O statistics for each data direction performed (showing
3837 writes in the example above). In the order listed, they denote:
3840 The string before the colon shows the I/O direction the statistics
3841 are for. **IOPS** is the average I/Os performed per second. **BW**
3842 is the average bandwidth rate shown as: value in power of 2 format
3843 (value in power of 10 format). The last two values show: (**total
3844 I/O performed** in power of 2 format / **runtime** of that thread).
3847 Submission latency (**min** being the minimum, **max** being the
3848 maximum, **avg** being the average, **stdev** being the standard
3849 deviation). This is the time it took to submit the I/O. For
3850 sync I/O this row is not displayed as the slat is really the
3851 completion latency (since queue/complete is one operation there).
3852 This value can be in nanoseconds, microseconds or milliseconds ---
3853 fio will choose the most appropriate base and print that (in the
3854 example above nanoseconds was the best scale). Note: in :option:`--minimal` mode
3855 latencies are always expressed in microseconds.
3858 Completion latency. Same names as slat, this denotes the time from
3859 submission to completion of the I/O pieces. For sync I/O, clat will
3860 usually be equal (or very close) to 0, as the time from submit to
3861 complete is basically just CPU time (I/O has already been done, see slat
3865 Total latency. Same names as slat and clat, this denotes the time from
3866 when fio created the I/O unit to completion of the I/O operation.
3869 Bandwidth statistics based on samples. Same names as the xlat stats,
3870 but also includes the number of samples taken (**samples**) and an
3871 approximate percentage of total aggregate bandwidth this thread
3872 received in its group (**per**). This last value is only really
3873 useful if the threads in this group are on the same disk, since they
3874 are then competing for disk access.
3877 IOPS statistics based on samples. Same names as bw.
3879 **lat (nsec/usec/msec)**
3880 The distribution of I/O completion latencies. This is the time from when
3881 I/O leaves fio and when it gets completed. Unlike the separate
3882 read/write/trim sections above, the data here and in the remaining
3883 sections apply to all I/Os for the reporting group. 250=0.04% means that
3884 0.04% of the I/Os completed in under 250us. 500=64.11% means that 64.11%
3885 of the I/Os required 250 to 499us for completion.
3888 CPU usage. User and system time, along with the number of context
3889 switches this thread went through, usage of system and user time, and
3890 finally the number of major and minor page faults. The CPU utilization
3891 numbers are averages for the jobs in that reporting group, while the
3892 context and fault counters are summed.
3895 The distribution of I/O depths over the job lifetime. The numbers are
3896 divided into powers of 2 and each entry covers depths from that value
3897 up to those that are lower than the next entry -- e.g., 16= covers
3898 depths from 16 to 31. Note that the range covered by a depth
3899 distribution entry can be different to the range covered by the
3900 equivalent submit/complete distribution entry.
3903 How many pieces of I/O were submitting in a single submit call. Each
3904 entry denotes that amount and below, until the previous entry -- e.g.,
3905 16=100% means that we submitted anywhere between 9 to 16 I/Os per submit
3906 call. Note that the range covered by a submit distribution entry can
3907 be different to the range covered by the equivalent depth distribution
3911 Like the above submit number, but for completions instead.
3914 The number of read/write/trim requests issued, and how many of them were
3918 These values are for :option:`latency_target` and related options. When
3919 these options are engaged, this section describes the I/O depth required
3920 to meet the specified latency target.
3923 Example output was based on the following:
3924 TZ=UTC fio --ioengine=null --iodepth=2 --size=100M --numjobs=2 \
3925 --rate_process=poisson --io_limit=32M --name=read --bs=128k \
3926 --rate=11M --name=write --rw=write --bs=2k --rate=700k
3928 After each client has been listed, the group statistics are printed. They
3929 will look like this::
3931 Run status group 0 (all jobs):
3932 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
3933 WRITE: bw=1231KiB/s (1261kB/s), 616KiB/s-621KiB/s (630kB/s-636kB/s), io=64.0MiB (67.1MB), run=52747-53223msec
3935 For each data direction it prints:
3938 Aggregate bandwidth of threads in this group followed by the
3939 minimum and maximum bandwidth of all the threads in this group.
3940 Values outside of brackets are power-of-2 format and those
3941 within are the equivalent value in a power-of-10 format.
3943 Aggregate I/O performed of all threads in this group. The
3944 format is the same as bw.
3946 The smallest and longest runtimes of the threads in this group.
3948 And finally, the disk statistics are printed. This is Linux specific. They will look like this::
3950 Disk stats (read/write):
3951 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
3953 Each value is printed for both reads and writes, with reads first. The
3957 Number of I/Os performed by all groups.
3959 Number of merges performed by the I/O scheduler.
3961 Number of ticks we kept the disk busy.
3963 Total time spent in the disk queue.
3965 The disk utilization. A value of 100% means we kept the disk
3966 busy constantly, 50% would be a disk idling half of the time.
3968 It is also possible to get fio to dump the current output while it is running,
3969 without terminating the job. To do that, send fio the **USR1** signal. You can
3970 also get regularly timed dumps by using the :option:`--status-interval`
3971 parameter, or by creating a file in :file:`/tmp` named
3972 :file:`fio-dump-status`. If fio sees this file, it will unlink it and dump the
3973 current output status.
3979 For scripted usage where you typically want to generate tables or graphs of the
3980 results, fio can output the results in a semicolon separated format. The format
3981 is one long line of values, such as::
3983 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%
3984 A description of this job goes here.
3986 The job description (if provided) follows on a second line for terse v2.
3987 It appears on the same line for other terse versions.
3989 To enable terse output, use the :option:`--minimal` or
3990 :option:`--output-format`\=terse command line options. The
3991 first value is the version of the terse output format. If the output has to be
3992 changed for some reason, this number will be incremented by 1 to signify that
3995 Split up, the format is as follows (comments in brackets denote when a
3996 field was introduced or whether it's specific to some terse version):
4000 terse version, fio version [v3], jobname, groupid, error
4004 Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec)
4005 Submission latency: min, max, mean, stdev (usec)
4006 Completion latency: min, max, mean, stdev (usec)
4007 Completion latency percentiles: 20 fields (see below)
4008 Total latency: min, max, mean, stdev (usec)
4009 Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev, number of samples [v5]
4010 IOPS [v5]: min, max, mean, stdev, number of samples
4016 Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec)
4017 Submission latency: min, max, mean, stdev (usec)
4018 Completion latency: min, max, mean, stdev (usec)
4019 Completion latency percentiles: 20 fields (see below)
4020 Total latency: min, max, mean, stdev (usec)
4021 Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev, number of samples [v5]
4022 IOPS [v5]: min, max, mean, stdev, number of samples
4024 TRIM status [all but version 3]:
4026 Fields are similar to READ/WRITE status.
4030 user, system, context switches, major faults, minor faults
4034 <=1, 2, 4, 8, 16, 32, >=64
4036 I/O latencies microseconds::
4038 <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
4040 I/O latencies milliseconds::
4042 <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
4044 Disk utilization [v3]::
4046 disk name, read ios, write ios, read merges, write merges, read ticks, write ticks,
4047 time spent in queue, disk utilization percentage
4049 Additional Info (dependent on continue_on_error, default off)::
4051 total # errors, first error code
4053 Additional Info (dependent on description being set)::
4057 Completion latency percentiles can be a grouping of up to 20 sets, so for the
4058 terse output fio writes all of them. Each field will look like this::
4062 which is the Xth percentile, and the `usec` latency associated with it.
4064 For `Disk utilization`, all disks used by fio are shown. So for each disk there
4065 will be a disk utilization section.
4067 Below is a single line containing short names for each of the fields in the
4068 minimal output v3, separated by semicolons::
4070 terse_version_3;fio_version;jobname;groupid;error;read_kb;read_bandwidth_kb;read_iops;read_runtime_ms;read_slat_min_us;read_slat_max_us;read_slat_mean_us;read_slat_dev_us;read_clat_min_us;read_clat_max_us;read_clat_mean_us;read_clat_dev_us;read_clat_pct01;read_clat_pct02;read_clat_pct03;read_clat_pct04;read_clat_pct05;read_clat_pct06;read_clat_pct07;read_clat_pct08;read_clat_pct09;read_clat_pct10;read_clat_pct11;read_clat_pct12;read_clat_pct13;read_clat_pct14;read_clat_pct15;read_clat_pct16;read_clat_pct17;read_clat_pct18;read_clat_pct19;read_clat_pct20;read_tlat_min_us;read_lat_max_us;read_lat_mean_us;read_lat_dev_us;read_bw_min_kb;read_bw_max_kb;read_bw_agg_pct;read_bw_mean_kb;read_bw_dev_kb;write_kb;write_bandwidth_kb;write_iops;write_runtime_ms;write_slat_min_us;write_slat_max_us;write_slat_mean_us;write_slat_dev_us;write_clat_min_us;write_clat_max_us;write_clat_mean_us;write_clat_dev_us;write_clat_pct01;write_clat_pct02;write_clat_pct03;write_clat_pct04;write_clat_pct05;write_clat_pct06;write_clat_pct07;write_clat_pct08;write_clat_pct09;write_clat_pct10;write_clat_pct11;write_clat_pct12;write_clat_pct13;write_clat_pct14;write_clat_pct15;write_clat_pct16;write_clat_pct17;write_clat_pct18;write_clat_pct19;write_clat_pct20;write_tlat_min_us;write_lat_max_us;write_lat_mean_us;write_lat_dev_us;write_bw_min_kb;write_bw_max_kb;write_bw_agg_pct;write_bw_mean_kb;write_bw_dev_kb;cpu_user;cpu_sys;cpu_csw;cpu_mjf;cpu_minf;iodepth_1;iodepth_2;iodepth_4;iodepth_8;iodepth_16;iodepth_32;iodepth_64;lat_2us;lat_4us;lat_10us;lat_20us;lat_50us;lat_100us;lat_250us;lat_500us;lat_750us;lat_1000us;lat_2ms;lat_4ms;lat_10ms;lat_20ms;lat_50ms;lat_100ms;lat_250ms;lat_500ms;lat_750ms;lat_1000ms;lat_2000ms;lat_over_2000ms;disk_name;disk_read_iops;disk_write_iops;disk_read_merges;disk_write_merges;disk_read_ticks;write_ticks;disk_queue_time;disk_util
4072 In client/server mode terse output differs from what appears when jobs are run
4073 locally. Disk utilization data is omitted from the standard terse output and
4074 for v3 and later appears on its own separate line at the end of each terse
4081 The `json` output format is intended to be both human readable and convenient
4082 for automated parsing. For the most part its sections mirror those of the
4083 `normal` output. The `runtime` value is reported in msec and the `bw` value is
4084 reported in 1024 bytes per second units.
4090 The `json+` output format is identical to the `json` output format except that it
4091 adds a full dump of the completion latency bins. Each `bins` object contains a
4092 set of (key, value) pairs where keys are latency durations and values count how
4093 many I/Os had completion latencies of the corresponding duration. For example,
4096 "bins" : { "87552" : 1, "89600" : 1, "94720" : 1, "96768" : 1, "97792" : 1, "99840" : 1, "100864" : 2, "103936" : 6, "104960" : 534, "105984" : 5995, "107008" : 7529, ... }
4098 This data indicates that one I/O required 87,552ns to complete, two I/Os required
4099 100,864ns to complete, and 7529 I/Os required 107,008ns to complete.
4101 Also included with fio is a Python script `fio_jsonplus_clat2csv` that takes
4102 json+ output and generates CSV-formatted latency data suitable for plotting.
4104 The latency durations actually represent the midpoints of latency intervals.
4105 For details refer to :file:`stat.h`.
4111 There are two trace file format that you can encounter. The older (v1) format is
4112 unsupported since version 1.20-rc3 (March 2008). It will still be described
4113 below in case that you get an old trace and want to understand it.
4115 In any case the trace is a simple text file with a single action per line.
4118 Trace file format v1
4119 ~~~~~~~~~~~~~~~~~~~~
4121 Each line represents a single I/O action in the following format::
4125 where `rw=0/1` for read/write, and the `offset` and `length` entries being in bytes.
4127 This format is not supported in fio versions >= 1.20-rc3.
4130 Trace file format v2
4131 ~~~~~~~~~~~~~~~~~~~~
4133 The second version of the trace file format was added in fio version 1.17. It
4134 allows to access more then one file per trace and has a bigger set of possible
4137 The first line of the trace file has to be::
4141 Following this can be lines in two different formats, which are described below.
4143 The file management format::
4147 The `filename` is given as an absolute path. The `action` can be one of these:
4150 Add the given `filename` to the trace.
4152 Open the file with the given `filename`. The `filename` has to have
4153 been added with the **add** action before.
4155 Close the file with the given `filename`. The file has to have been
4159 The file I/O action format::
4161 filename action offset length
4163 The `filename` is given as an absolute path, and has to have been added and
4164 opened before it can be used with this format. The `offset` and `length` are
4165 given in bytes. The `action` can be one of these:
4168 Wait for `offset` microseconds. Everything below 100 is discarded.
4169 The time is relative to the previous `wait` statement.
4171 Read `length` bytes beginning from `offset`.
4173 Write `length` bytes beginning from `offset`.
4175 :manpage:`fsync(2)` the file.
4177 :manpage:`fdatasync(2)` the file.
4179 Trim the given file from the given `offset` for `length` bytes.
4182 I/O Replay - Merging Traces
4183 ---------------------------
4185 Colocation is a common practice used to get the most out of a machine.
4186 Knowing which workloads play nicely with each other and which ones don't is
4187 a much harder task. While fio can replay workloads concurrently via multiple
4188 jobs, it leaves some variability up to the scheduler making results harder to
4189 reproduce. Merging is a way to make the order of events consistent.
4191 Merging is integrated into I/O replay and done when a
4192 :option:`merge_blktrace_file` is specified. The list of files passed to
4193 :option:`read_iolog` go through the merge process and output a single file
4194 stored to the specified file. The output file is passed on as if it were the
4195 only file passed to :option:`read_iolog`. An example would look like::
4197 $ fio --read_iolog="<file1>:<file2>" --merge_blktrace_file="<output_file>"
4199 Creating only the merged file can be done by passing the command line argument
4200 :option:`--merge-blktrace-only`.
4202 Scaling traces can be done to see the relative impact of any particular trace
4203 being slowed down or sped up. :option:`merge_blktrace_scalars` takes in a colon
4204 separated list of percentage scalars. It is index paired with the files passed
4205 to :option:`read_iolog`.
4207 With scaling, it may be desirable to match the running time of all traces.
4208 This can be done with :option:`merge_blktrace_iters`. It is index paired with
4209 :option:`read_iolog` just like :option:`merge_blktrace_scalars`.
4211 In an example, given two traces, A and B, each 60s long. If we want to see
4212 the impact of trace A issuing IOs twice as fast and repeat trace A over the
4213 runtime of trace B, the following can be done::
4215 $ fio --read_iolog="<trace_a>:"<trace_b>" --merge_blktrace_file"<output_file>" --merge_blktrace_scalars="50:100" --merge_blktrace_iters="2:1"
4217 This runs trace A at 2x the speed twice for approximately the same runtime as
4218 a single run of trace B.
4221 CPU idleness profiling
4222 ----------------------
4224 In some cases, we want to understand CPU overhead in a test. For example, we
4225 test patches for the specific goodness of whether they reduce CPU usage.
4226 Fio implements a balloon approach to create a thread per CPU that runs at idle
4227 priority, meaning that it only runs when nobody else needs the cpu.
4228 By measuring the amount of work completed by the thread, idleness of each CPU
4229 can be derived accordingly.
4231 An unit work is defined as touching a full page of unsigned characters. Mean and
4232 standard deviation of time to complete an unit work is reported in "unit work"
4233 section. Options can be chosen to report detailed percpu idleness or overall
4234 system idleness by aggregating percpu stats.
4237 Verification and triggers
4238 -------------------------
4240 Fio is usually run in one of two ways, when data verification is done. The first
4241 is a normal write job of some sort with verify enabled. When the write phase has
4242 completed, fio switches to reads and verifies everything it wrote. The second
4243 model is running just the write phase, and then later on running the same job
4244 (but with reads instead of writes) to repeat the same I/O patterns and verify
4245 the contents. Both of these methods depend on the write phase being completed,
4246 as fio otherwise has no idea how much data was written.
4248 With verification triggers, fio supports dumping the current write state to
4249 local files. Then a subsequent read verify workload can load this state and know
4250 exactly where to stop. This is useful for testing cases where power is cut to a
4251 server in a managed fashion, for instance.
4253 A verification trigger consists of two things:
4255 1) Storing the write state of each job.
4256 2) Executing a trigger command.
4258 The write state is relatively small, on the order of hundreds of bytes to single
4259 kilobytes. It contains information on the number of completions done, the last X
4262 A trigger is invoked either through creation ('touch') of a specified file in
4263 the system, or through a timeout setting. If fio is run with
4264 :option:`--trigger-file`\= :file:`/tmp/trigger-file`, then it will continually
4265 check for the existence of :file:`/tmp/trigger-file`. When it sees this file, it
4266 will fire off the trigger (thus saving state, and executing the trigger
4269 For client/server runs, there's both a local and remote trigger. If fio is
4270 running as a server backend, it will send the job states back to the client for
4271 safe storage, then execute the remote trigger, if specified. If a local trigger
4272 is specified, the server will still send back the write state, but the client
4273 will then execute the trigger.
4275 Verification trigger example
4276 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
4278 Let's say we want to run a powercut test on the remote Linux machine 'server'.
4279 Our write workload is in :file:`write-test.fio`. We want to cut power to 'server' at
4280 some point during the run, and we'll run this test from the safety or our local
4281 machine, 'localbox'. On the server, we'll start the fio backend normally::
4283 server# fio --server
4285 and on the client, we'll fire off the workload::
4287 localbox$ fio --client=server --trigger-file=/tmp/my-trigger --trigger-remote="bash -c \"echo b > /proc/sysrq-triger\""
4289 We set :file:`/tmp/my-trigger` as the trigger file, and we tell fio to execute::
4291 echo b > /proc/sysrq-trigger
4293 on the server once it has received the trigger and sent us the write state. This
4294 will work, but it's not **really** cutting power to the server, it's merely
4295 abruptly rebooting it. If we have a remote way of cutting power to the server
4296 through IPMI or similar, we could do that through a local trigger command
4297 instead. Let's assume we have a script that does IPMI reboot of a given hostname,
4298 ipmi-reboot. On localbox, we could then have run fio with a local trigger
4301 localbox$ fio --client=server --trigger-file=/tmp/my-trigger --trigger="ipmi-reboot server"
4303 For this case, fio would wait for the server to send us the write state, then
4304 execute ``ipmi-reboot server`` when that happened.
4306 Loading verify state
4307 ~~~~~~~~~~~~~~~~~~~~
4309 To load stored write state, a read verification job file must contain the
4310 :option:`verify_state_load` option. If that is set, fio will load the previously
4311 stored state. For a local fio run this is done by loading the files directly,
4312 and on a client/server run, the server backend will ask the client to send the
4313 files over and load them from there.
4319 Fio supports a variety of log file formats, for logging latencies, bandwidth,
4320 and IOPS. The logs share a common format, which looks like this:
4322 *time* (`msec`), *value*, *data direction*, *block size* (`bytes`),
4323 *offset* (`bytes`), *command priority*
4325 *Time* for the log entry is always in milliseconds. The *value* logged depends
4326 on the type of log, it will be one of the following:
4329 Value is latency in nsecs
4335 *Data direction* is one of the following:
4344 The entry's *block size* is always in bytes. The *offset* is the position in bytes
4345 from the start of the file for that particular I/O. The logging of the offset can be
4346 toggled with :option:`log_offset`.
4348 *Command priority* is 0 for normal priority and 1 for high priority. This is controlled
4349 by the ioengine specific :option:`cmdprio_percentage`.
4351 Fio defaults to logging every individual I/O but when windowed logging is set
4352 through :option:`log_avg_msec`, either the average (by default) or the maximum
4353 (:option:`log_max_value` is set) *value* seen over the specified period of time
4354 is recorded. Each *data direction* seen within the window period will aggregate
4355 its values in a separate row. Further, when using windowed logging the *block
4356 size* and *offset* entries will always contain 0.
4362 Normally fio is invoked as a stand-alone application on the machine where the
4363 I/O workload should be generated. However, the backend and frontend of fio can
4364 be run separately i.e., the fio server can generate an I/O workload on the "Device
4365 Under Test" while being controlled by a client on another machine.
4367 Start the server on the machine which has access to the storage DUT::
4371 where `args` defines what fio listens to. The arguments are of the form
4372 ``type,hostname`` or ``IP,port``. *type* is either ``ip`` (or ip4) for TCP/IP
4373 v4, ``ip6`` for TCP/IP v6, or ``sock`` for a local unix domain socket.
4374 *hostname* is either a hostname or IP address, and *port* is the port to listen
4375 to (only valid for TCP/IP, not a local socket). Some examples:
4379 Start a fio server, listening on all interfaces on the default port (8765).
4381 2) ``fio --server=ip:hostname,4444``
4383 Start a fio server, listening on IP belonging to hostname and on port 4444.
4385 3) ``fio --server=ip6:::1,4444``
4387 Start a fio server, listening on IPv6 localhost ::1 and on port 4444.
4389 4) ``fio --server=,4444``
4391 Start a fio server, listening on all interfaces on port 4444.
4393 5) ``fio --server=1.2.3.4``
4395 Start a fio server, listening on IP 1.2.3.4 on the default port.
4397 6) ``fio --server=sock:/tmp/fio.sock``
4399 Start a fio server, listening on the local socket :file:`/tmp/fio.sock`.
4401 Once a server is running, a "client" can connect to the fio server with::
4403 fio <local-args> --client=<server> <remote-args> <job file(s)>
4405 where `local-args` are arguments for the client where it is running, `server`
4406 is the connect string, and `remote-args` and `job file(s)` are sent to the
4407 server. The `server` string follows the same format as it does on the server
4408 side, to allow IP/hostname/socket and port strings.
4410 Fio can connect to multiple servers this way::
4412 fio --client=<server1> <job file(s)> --client=<server2> <job file(s)>
4414 If the job file is located on the fio server, then you can tell the server to
4415 load a local file as well. This is done by using :option:`--remote-config` ::
4417 fio --client=server --remote-config /path/to/file.fio
4419 Then fio will open this local (to the server) job file instead of being passed
4420 one from the client.
4422 If you have many servers (example: 100 VMs/containers), you can input a pathname
4423 of a file containing host IPs/names as the parameter value for the
4424 :option:`--client` option. For example, here is an example :file:`host.list`
4425 file containing 2 hostnames::
4427 host1.your.dns.domain
4428 host2.your.dns.domain
4430 The fio command would then be::
4432 fio --client=host.list <job file(s)>
4434 In this mode, you cannot input server-specific parameters or job files -- all
4435 servers receive the same job file.
4437 In order to let ``fio --client`` runs use a shared filesystem from multiple
4438 hosts, ``fio --client`` now prepends the IP address of the server to the
4439 filename. For example, if fio is using the directory :file:`/mnt/nfs/fio` and is
4440 writing filename :file:`fileio.tmp`, with a :option:`--client` `hostfile`
4441 containing two hostnames ``h1`` and ``h2`` with IP addresses 192.168.10.120 and
4442 192.168.10.121, then fio will create two files::
4444 /mnt/nfs/fio/192.168.10.120.fileio.tmp
4445 /mnt/nfs/fio/192.168.10.121.fileio.tmp
4447 Terse output in client/server mode will differ slightly from what is produced
4448 when fio is run in stand-alone mode. See the terse output section for details.