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 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 Show available debug options.
99 .. option:: --parse-only
101 Parse options only, don\'t start any I/O.
103 .. option:: --output=filename
105 Write output to file `filename`.
107 .. option:: --bandwidth-log
109 Generate aggregate bandwidth logs.
111 .. option:: --minimal
113 Print statistics in a terse, semicolon-delimited format.
115 .. option:: --append-terse
117 Print statistics in selected mode AND terse, semicolon-delimited format.
118 **deprecated**, use :option:`--output-format` instead to select multiple
121 .. option:: --output-format=type
123 Set the reporting format to `normal`, `terse`, `json`, or `json+`. Multiple
124 formats can be selected, separated by a comma. `terse` is a CSV based
125 format. `json+` is like `json`, except it adds a full dump of the latency
128 .. option:: --terse-version=type
130 Set terse version output format (default 3, or 2 or 4 or 5).
132 .. option:: --version
134 Print version info and exit.
140 .. option:: --cpuclock-test
142 Perform test and validation of internal CPU clock.
144 .. option:: --crctest=test
146 Test the speed of the builtin checksumming functions. If no argument is
147 given, all of them are tested. Or a comma separated list can be passed, in
148 which case the given ones are tested.
150 .. option:: --cmdhelp=command
152 Print help information for `command`. May be ``all`` for all commands.
154 .. option:: --enghelp=[ioengine[,command]]
156 List all commands defined by :option:`ioengine`, or print help for `command`
157 defined by :option:`ioengine`. If no :option:`ioengine` is given, list all
160 .. option:: --showcmd=jobfile
162 Turn a job file into command line options.
164 .. option:: --readonly
166 Turn on safety read-only checks, preventing writes. The ``--readonly``
167 option is an extra safety guard to prevent users from accidentally starting
168 a write workload when that is not desired. Fio will only write if
169 `rw=write/randwrite/rw/randrw` is given. This extra safety net can be used
170 as an extra precaution as ``--readonly`` will also enable a write check in
171 the I/O engine core to prevent writes due to unknown user space bug(s).
173 .. option:: --eta=when
175 When real-time ETA estimate should be printed. May be `always`, `never` or
178 .. option:: --eta-newline=time
180 Force a new line for every `time` period passed. When the unit is omitted,
181 the value is interpreted in seconds.
183 .. option:: --status-interval=time
185 Force full status dump every `time` period passed. When the unit is
186 omitted, the value is interpreted in seconds.
188 .. option:: --section=name
190 Only run specified section in job file. Multiple sections can be specified.
191 The ``--section`` option allows one to combine related jobs into one file.
192 E.g. one job file could define light, moderate, and heavy sections. Tell
193 fio to run only the "heavy" section by giving ``--section=heavy``
194 command line option. One can also specify the "write" operations in one
195 section and "verify" operation in another section. The ``--section`` option
196 only applies to job sections. The reserved *global* section is always
199 .. option:: --alloc-size=kb
201 Set the internal smalloc pool to this size in kb (def 1024). The
202 ``--alloc-size`` switch allows one to use a larger pool size for smalloc.
203 If running large jobs with randommap enabled, fio can run out of memory.
204 Smalloc is an internal allocator for shared structures from a fixed size
205 memory pool. The pool size defaults to 16M and can grow to 8 pools.
207 NOTE: While running :file:`.fio_smalloc.*` backing store files are visible
210 .. option:: --warnings-fatal
212 All fio parser warnings are fatal, causing fio to exit with an
215 .. option:: --max-jobs=nr
217 Maximum number of threads/processes to support.
219 .. option:: --server=args
221 Start a backend server, with `args` specifying what to listen to.
222 See `Client/Server`_ section.
224 .. option:: --daemonize=pidfile
226 Background a fio server, writing the pid to the given `pidfile` file.
228 .. option:: --client=hostname
230 Instead of running the jobs locally, send and run them on the given host or
231 set of hosts. See `Client/Server`_ section.
233 .. option:: --remote-config=file
235 Tell fio server to load this local file.
237 .. option:: --idle-prof=option
239 Report cpu idleness on a system or percpu basis
240 ``--idle-prof=system,percpu`` or
241 run unit work calibration only ``--idle-prof=calibrate``.
243 .. option:: --inflate-log=log
245 Inflate and output compressed log.
247 .. option:: --trigger-file=file
249 Execute trigger cmd when file exists.
251 .. option:: --trigger-timeout=t
253 Execute trigger at this time.
255 .. option:: --trigger=cmd
257 Set this command as local trigger.
259 .. option:: --trigger-remote=cmd
261 Set this command as remote trigger.
263 .. option:: --aux-path=path
265 Use this path for fio state generated files.
267 Any parameters following the options will be assumed to be job files, unless
268 they match a job file parameter. Multiple job files can be listed and each job
269 file will be regarded as a separate group. Fio will :option:`stonewall`
270 execution between each group.
276 As previously described, fio accepts one or more job files describing what it is
277 supposed to do. The job file format is the classic ini file, where the names
278 enclosed in [] brackets define the job name. You are free to use any ASCII name
279 you want, except *global* which has special meaning. Following the job name is
280 a sequence of zero or more parameters, one per line, that define the behavior of
281 the job. If the first character in a line is a ';' or a '#', the entire line is
282 discarded as a comment.
284 A *global* section sets defaults for the jobs described in that file. A job may
285 override a *global* section parameter, and a job file may even have several
286 *global* sections if so desired. A job is only affected by a *global* section
289 The :option:`--cmdhelp` option also lists all options. If used with an `option`
290 argument, :option:`--cmdhelp` will detail the given `option`.
292 See the `examples/` directory for inspiration on how to write job files. Note
293 the copyright and license requirements currently apply to `examples/` files.
295 So let's look at a really simple job file that defines two processes, each
296 randomly reading from a 128MiB file:
300 ; -- start job file --
311 As you can see, the job file sections themselves are empty as all the described
312 parameters are shared. As no :option:`filename` option is given, fio makes up a
313 `filename` for each of the jobs as it sees fit. On the command line, this job
314 would look as follows::
316 $ fio --name=global --rw=randread --size=128m --name=job1 --name=job2
319 Let's look at an example that has a number of processes writing randomly to
324 ; -- start job file --
335 Here we have no *global* section, as we only have one job defined anyway. We
336 want to use async I/O here, with a depth of 4 for each file. We also increased
337 the buffer size used to 32KiB and define numjobs to 4 to fork 4 identical
338 jobs. The result is 4 processes each randomly writing to their own 64MiB
339 file. Instead of using the above job file, you could have given the parameters
340 on the command line. For this case, you would specify::
342 $ fio --name=random-writers --ioengine=libaio --iodepth=4 --rw=randwrite --bs=32k --direct=0 --size=64m --numjobs=4
344 When fio is utilized as a basis of any reasonably large test suite, it might be
345 desirable to share a set of standardized settings across multiple job files.
346 Instead of copy/pasting such settings, any section may pull in an external
347 :file:`filename.fio` file with *include filename* directive, as in the following
350 ; -- start job file including.fio --
354 include glob-include.fio
361 include test-include.fio
362 ; -- end job file including.fio --
366 ; -- start job file glob-include.fio --
369 ; -- end job file glob-include.fio --
373 ; -- start job file test-include.fio --
376 ; -- end job file test-include.fio --
378 Settings pulled into a section apply to that section only (except *global*
379 section). Include directives may be nested in that any included file may contain
380 further include directive(s). Include files may not contain [] sections.
383 Environment variables
384 ~~~~~~~~~~~~~~~~~~~~~
386 Fio also supports environment variable expansion in job files. Any sub-string of
387 the form ``${VARNAME}`` as part of an option value (in other words, on the right
388 of the '='), will be expanded to the value of the environment variable called
389 `VARNAME`. If no such environment variable is defined, or `VARNAME` is the
390 empty string, the empty string will be substituted.
392 As an example, let's look at a sample fio invocation and job file::
394 $ SIZE=64m NUMJOBS=4 fio jobfile.fio
398 ; -- start job file --
405 This will expand to the following equivalent job file at runtime:
409 ; -- start job file --
416 Fio ships with a few example job files, you can also look there for inspiration.
421 Additionally, fio has a set of reserved keywords that will be replaced
422 internally with the appropriate value. Those keywords are:
426 The architecture page size of the running system.
430 Megabytes of total memory in the system.
434 Number of online available CPUs.
436 These can be used on the command line or in the job file, and will be
437 automatically substituted with the current system values when the job is
438 run. Simple math is also supported on these keywords, so you can perform actions
443 and get that properly expanded to 8 times the size of memory in the machine.
449 This section describes in details each parameter associated with a job. Some
450 parameters take an option of a given type, such as an integer or a
451 string. Anywhere a numeric value is required, an arithmetic expression may be
452 used, provided it is surrounded by parentheses. Supported operators are:
461 For time values in expressions, units are microseconds by default. This is
462 different than for time values not in expressions (not enclosed in
463 parentheses). The following types are used:
470 String. This is a sequence of alpha characters.
473 Integer with possible time suffix. Without a unit value is interpreted as
474 seconds unless otherwise specified. Accepts a suffix of 'd' for days, 'h' for
475 hours, 'm' for minutes, 's' for seconds, 'ms' (or 'msec') for milliseconds and
476 'us' (or 'usec') for microseconds. For example, use 10m for 10 minutes.
481 Integer. A whole number value, which may contain an integer prefix
482 and an integer suffix:
484 [*integer prefix*] **number** [*integer suffix*]
486 The optional *integer prefix* specifies the number's base. The default
487 is decimal. *0x* specifies hexadecimal.
489 The optional *integer suffix* specifies the number's units, and includes an
490 optional unit prefix and an optional unit. For quantities of data, the
491 default unit is bytes. For quantities of time, the default unit is seconds
492 unless otherwise specified.
494 With :option:`kb_base`\=1000, fio follows international standards for unit
495 prefixes. To specify power-of-10 decimal values defined in the
496 International System of Units (SI):
498 * *Ki* -- means kilo (K) or 1000
499 * *Mi* -- means mega (M) or 1000**2
500 * *Gi* -- means giga (G) or 1000**3
501 * *Ti* -- means tera (T) or 1000**4
502 * *Pi* -- means peta (P) or 1000**5
504 To specify power-of-2 binary values defined in IEC 80000-13:
506 * *k* -- means kibi (Ki) or 1024
507 * *M* -- means mebi (Mi) or 1024**2
508 * *G* -- means gibi (Gi) or 1024**3
509 * *T* -- means tebi (Ti) or 1024**4
510 * *P* -- means pebi (Pi) or 1024**5
512 With :option:`kb_base`\=1024 (the default), the unit prefixes are opposite
513 from those specified in the SI and IEC 80000-13 standards to provide
514 compatibility with old scripts. For example, 4k means 4096.
516 For quantities of data, an optional unit of 'B' may be included
517 (e.g., 'kB' is the same as 'k').
519 The *integer suffix* is not case sensitive (e.g., m/mi mean mebi/mega,
520 not milli). 'b' and 'B' both mean byte, not bit.
522 Examples with :option:`kb_base`\=1000:
524 * *4 KiB*: 4096, 4096b, 4096B, 4ki, 4kib, 4kiB, 4Ki, 4KiB
525 * *1 MiB*: 1048576, 1mi, 1024ki
526 * *1 MB*: 1000000, 1m, 1000k
527 * *1 TiB*: 1099511627776, 1ti, 1024gi, 1048576mi
528 * *1 TB*: 1000000000, 1t, 1000m, 1000000k
530 Examples with :option:`kb_base`\=1024 (default):
532 * *4 KiB*: 4096, 4096b, 4096B, 4k, 4kb, 4kB, 4K, 4KB
533 * *1 MiB*: 1048576, 1m, 1024k
534 * *1 MB*: 1000000, 1mi, 1000ki
535 * *1 TiB*: 1099511627776, 1t, 1024g, 1048576m
536 * *1 TB*: 1000000000, 1ti, 1000mi, 1000000ki
538 To specify times (units are not case sensitive):
542 * *M* -- means minutes
543 * *s* -- or sec means seconds (default)
544 * *ms* -- or *msec* means milliseconds
545 * *us* -- or *usec* means microseconds
547 If the option accepts an upper and lower range, use a colon ':' or
548 minus '-' to separate such values. See :ref:`irange <irange>`.
549 If the lower value specified happens to be larger than the upper value
550 the two values are swapped.
555 Boolean. Usually parsed as an integer, however only defined for
556 true and false (1 and 0).
561 Integer range with suffix. Allows value range to be given, such as
562 1024-4096. A colon may also be used as the separator, e.g. 1k:4k. If the
563 option allows two sets of ranges, they can be specified with a ',' or '/'
564 delimiter: 1k-4k/8k-32k. Also see :ref:`int <int>`.
567 A list of floating point numbers, separated by a ':' character.
573 .. option:: kb_base=int
575 Select the interpretation of unit prefixes in input parameters.
578 Inputs comply with IEC 80000-13 and the International
579 System of Units (SI). Use:
581 - power-of-2 values with IEC prefixes (e.g., KiB)
582 - power-of-10 values with SI prefixes (e.g., kB)
585 Compatibility mode (default). To avoid breaking old scripts:
587 - power-of-2 values with SI prefixes
588 - power-of-10 values with IEC prefixes
590 See :option:`bs` for more details on input parameters.
592 Outputs always use correct prefixes. Most outputs include both
595 bw=2383.3kB/s (2327.4KiB/s)
597 If only one value is reported, then kb_base selects the one to use:
599 **1000** -- SI prefixes
601 **1024** -- IEC prefixes
603 .. option:: unit_base=int
605 Base unit for reporting. Allowed values are:
608 Use auto-detection (default).
615 With the above in mind, here follows the complete list of fio job parameters.
623 ASCII name of the job. This may be used to override the name printed by fio
624 for this job. Otherwise the job name is used. On the command line this
625 parameter has the special purpose of also signaling the start of a new job.
627 .. option:: description=str
629 Text description of the job. Doesn't do anything except dump this text
630 description when this job is run. It's not parsed.
632 .. option:: loops=int
634 Run the specified number of iterations of this job. Used to repeat the same
635 workload a given number of times. Defaults to 1.
637 .. option:: numjobs=int
639 Create the specified number of clones of this job. Each clone of job
640 is spawned as an independent thread or process. May be used to setup a
641 larger number of threads/processes doing the same thing. Each thread is
642 reported separately; to see statistics for all clones as a whole, use
643 :option:`group_reporting` in conjunction with :option:`new_group`.
644 See :option:`--max-jobs`. Default: 1.
647 Time related parameters
648 ~~~~~~~~~~~~~~~~~~~~~~~
650 .. option:: runtime=time
652 Tell fio to terminate processing after the specified period of time. It
653 can be quite hard to determine for how long a specified job will run, so
654 this parameter is handy to cap the total runtime to a given time. When
655 the unit is omitted, the value is intepreted in seconds.
657 .. option:: time_based
659 If set, fio will run for the duration of the :option:`runtime` specified
660 even if the file(s) are completely read or written. It will simply loop over
661 the same workload as many times as the :option:`runtime` allows.
663 .. option:: startdelay=irange(time)
665 Delay the start of job for the specified amount of time. Can be a single
666 value or a range. When given as a range, each thread will choose a value
667 randomly from within the range. Value is in seconds if a unit is omitted.
669 .. option:: ramp_time=time
671 If set, fio will run the specified workload for this amount of time before
672 logging any performance numbers. Useful for letting performance settle
673 before logging results, thus minimizing the runtime required for stable
674 results. Note that the ``ramp_time`` is considered lead in time for a job,
675 thus it will increase the total runtime if a special timeout or
676 :option:`runtime` is specified. When the unit is omitted, the value is
679 .. option:: clocksource=str
681 Use the given clocksource as the base of timing. The supported options are:
684 :manpage:`gettimeofday(2)`
687 :manpage:`clock_gettime(2)`
690 Internal CPU clock source
692 cpu is the preferred clocksource if it is reliable, as it is very fast (and
693 fio is heavy on time calls). Fio will automatically use this clocksource if
694 it's supported and considered reliable on the system it is running on,
695 unless another clocksource is specifically set. For x86/x86-64 CPUs, this
696 means supporting TSC Invariant.
698 .. option:: gtod_reduce=bool
700 Enable all of the :manpage:`gettimeofday(2)` reducing options
701 (:option:`disable_clat`, :option:`disable_slat`, :option:`disable_bw_measurement`) plus
702 reduce precision of the timeout somewhat to really shrink the
703 :manpage:`gettimeofday(2)` call count. With this option enabled, we only do
704 about 0.4% of the :manpage:`gettimeofday(2)` calls we would have done if all
705 time keeping was enabled.
707 .. option:: gtod_cpu=int
709 Sometimes it's cheaper to dedicate a single thread of execution to just
710 getting the current time. Fio (and databases, for instance) are very
711 intensive on :manpage:`gettimeofday(2)` calls. With this option, you can set
712 one CPU aside for doing nothing but logging current time to a shared memory
713 location. Then the other threads/processes that run I/O workloads need only
714 copy that segment, instead of entering the kernel with a
715 :manpage:`gettimeofday(2)` call. The CPU set aside for doing these time
716 calls will be excluded from other uses. Fio will manually clear it from the
717 CPU mask of other jobs.
723 .. option:: directory=str
725 Prefix filenames with this directory. Used to place files in a different
726 location than :file:`./`. You can specify a number of directories by
727 separating the names with a ':' character. These directories will be
728 assigned equally distributed to job clones created by :option:`numjobs` as
729 long as they are using generated filenames. If specific `filename(s)` are
730 set fio will use the first listed directory, and thereby matching the
731 `filename` semantic which generates a file each clone if not specified, but
732 let all clones use the same if set.
734 See the :option:`filename` option for information on how to escape "``:``" and
735 "``\``" characters within the directory path itself.
737 .. option:: filename=str
739 Fio normally makes up a `filename` based on the job name, thread number, and
740 file number (see :option:`filename_format`). If you want to share files
741 between threads in a job or several
742 jobs with fixed file paths, specify a `filename` for each of them to override
743 the default. If the ioengine is file based, you can specify a number of files
744 by separating the names with a ':' colon. So if you wanted a job to open
745 :file:`/dev/sda` and :file:`/dev/sdb` as the two working files, you would use
746 ``filename=/dev/sda:/dev/sdb``. This also means that whenever this option is
747 specified, :option:`nrfiles` is ignored. The size of regular files specified
748 by this option will be :option:`size` divided by number of files unless an
749 explicit size is specified by :option:`filesize`.
751 Each colon and backslash in the wanted path must be escaped with a ``\``
752 character. For instance, if the path is :file:`/dev/dsk/foo@3,0:c` then you
753 would use ``filename=/dev/dsk/foo@3,0\:c`` and if the path is
754 :file:`F:\\filename` then you would use ``filename=F\:\\filename``.
756 On Windows, disk devices are accessed as :file:`\\\\.\\PhysicalDrive0` for
757 the first device, :file:`\\\\.\\PhysicalDrive1` for the second etc.
758 Note: Windows and FreeBSD prevent write access to areas
759 of the disk containing in-use data (e.g. filesystems).
761 The filename "`-`" is a reserved name, meaning *stdin* or *stdout*. Which
762 of the two depends on the read/write direction set.
764 .. option:: filename_format=str
766 If sharing multiple files between jobs, it is usually necessary to have fio
767 generate the exact names that you want. By default, fio will name a file
768 based on the default file format specification of
769 :file:`jobname.jobnumber.filenumber`. With this option, that can be
770 customized. Fio will recognize and replace the following keywords in this
774 The name of the worker thread or process.
776 The incremental number of the worker thread or process.
778 The incremental number of the file for that worker thread or
781 To have dependent jobs share a set of files, this option can be set to have
782 fio generate filenames that are shared between the two. For instance, if
783 :file:`testfiles.$filenum` is specified, file number 4 for any job will be
784 named :file:`testfiles.4`. The default of :file:`$jobname.$jobnum.$filenum`
785 will be used if no other format specifier is given.
787 .. option:: unique_filename=bool
789 To avoid collisions between networked clients, fio defaults to prefixing any
790 generated filenames (with a directory specified) with the source of the
791 client connecting. To disable this behavior, set this option to 0.
793 .. option:: opendir=str
795 Recursively open any files below directory `str`.
797 .. option:: lockfile=str
799 Fio defaults to not locking any files before it does I/O to them. If a file
800 or file descriptor is shared, fio can serialize I/O to that file to make the
801 end result consistent. This is usual for emulating real workloads that share
802 files. The lock modes are:
805 No locking. The default.
807 Only one thread or process may do I/O at a time, excluding all
810 Read-write locking on the file. Many readers may
811 access the file at the same time, but writes get exclusive access.
813 .. option:: nrfiles=int
815 Number of files to use for this job. Defaults to 1. The size of files
816 will be :option:`size` divided by this unless explicit size is specified by
817 :option:`filesize`. Files are created for each thread separately, and each
818 file will have a file number within its name by default, as explained in
819 :option:`filename` section.
822 .. option:: openfiles=int
824 Number of files to keep open at the same time. Defaults to the same as
825 :option:`nrfiles`, can be set smaller to limit the number simultaneous
828 .. option:: file_service_type=str
830 Defines how fio decides which file from a job to service next. The following
834 Choose a file at random.
837 Round robin over opened files. This is the default.
840 Finish one file before moving on to the next. Multiple files can
841 still be open depending on 'openfiles'.
844 Use a *Zipf* distribution to decide what file to access.
847 Use a *Pareto* distribution to decide what file to access.
850 Use a *Gaussian* (normal) distribution to decide what file to
853 For *random*, *roundrobin*, and *sequential*, a postfix can be appended to
854 tell fio how many I/Os to issue before switching to a new file. For example,
855 specifying ``file_service_type=random:8`` would cause fio to issue
856 8 I/Os before selecting a new file at random. For the non-uniform
857 distributions, a floating point postfix can be given to influence how the
858 distribution is skewed. See :option:`random_distribution` for a description
859 of how that would work.
861 .. option:: ioscheduler=str
863 Attempt to switch the device hosting the file to the specified I/O scheduler
866 .. option:: create_serialize=bool
868 If true, serialize the file creation for the jobs. This may be handy to
869 avoid interleaving of data files, which may greatly depend on the filesystem
870 used and even the number of processors in the system. Default: true.
872 .. option:: create_fsync=bool
874 fsync the data file after creation. This is the default.
876 .. option:: create_on_open=bool
878 Don't pre-setup the files for I/O, just create open() when it's time to do
879 I/O to that file. Default: false.
881 .. option:: create_only=bool
883 If true, fio will only run the setup phase of the job. If files need to be
884 laid out or updated on disk, only that will be done -- the actual job contents
885 are not executed. Default: false.
887 .. option:: allow_file_create=bool
889 If true, fio is permitted to create files as part of its workload. This is
890 the default behavior. If this option is false, then fio will error out if
891 the files it needs to use don't already exist. Default: true.
893 .. option:: allow_mounted_write=bool
895 If this isn't set, fio will abort jobs that are destructive (e.g. that write)
896 to what appears to be a mounted device or partition. This should help catch
897 creating inadvertently destructive tests, not realizing that the test will
898 destroy data on the mounted file system. Note that some platforms don't allow
899 writing against a mounted device regardless of this option. Default: false.
901 .. option:: pre_read=bool
903 If this is given, files will be pre-read into memory before starting the
904 given I/O operation. This will also clear the :option:`invalidate` flag,
905 since it is pointless to pre-read and then drop the cache. This will only
906 work for I/O engines that are seek-able, since they allow you to read the
907 same data multiple times. Thus it will not work on non-seekable I/O engines
908 (e.g. network, splice). Default: false.
910 .. option:: unlink=bool
912 Unlink the job files when done. Not the default, as repeated runs of that
913 job would then waste time recreating the file set again and again. Default:
916 .. option:: unlink_each_loop=bool
918 Unlink job files after each iteration or loop. Default: false.
920 .. option:: zonesize=int
922 Divide a file into zones of the specified size. See :option:`zoneskip`.
924 .. option:: zonerange=int
926 Give size of an I/O zone. See :option:`zoneskip`.
928 .. option:: zoneskip=int
930 Skip the specified number of bytes when :option:`zonesize` data has been
931 read. The two zone options can be used to only do I/O on zones of a file.
937 .. option:: direct=bool
939 If value is true, use non-buffered I/O. This is usually O_DIRECT. Note that
940 ZFS on Solaris doesn't support direct I/O. On Windows the synchronous
941 ioengines don't support direct I/O. Default: false.
943 .. option:: atomic=bool
945 If value is true, attempt to use atomic direct I/O. Atomic writes are
946 guaranteed to be stable once acknowledged by the operating system. Only
947 Linux supports O_ATOMIC right now.
949 .. option:: buffered=bool
951 If value is true, use buffered I/O. This is the opposite of the
952 :option:`direct` option. Defaults to true.
954 .. option:: readwrite=str, rw=str
956 Type of I/O pattern. Accepted values are:
963 Sequential trims (Linux block devices only).
969 Random trims (Linux block devices only).
971 Sequential mixed reads and writes.
973 Random mixed reads and writes.
975 Sequential trim+write sequences. Blocks will be trimmed first,
976 then the same blocks will be written to.
978 Fio defaults to read if the option is not specified. For the mixed I/O
979 types, the default is to split them 50/50. For certain types of I/O the
980 result may still be skewed a bit, since the speed may be different. It is
981 possible to specify a number of I/O's to do before getting a new offset,
982 this is done by appending a ``:<nr>`` to the end of the string given. For a
983 random read, it would look like ``rw=randread:8`` for passing in an offset
984 modifier with a value of 8. If the suffix is used with a sequential I/O
985 pattern, then the value specified will be added to the generated offset for
986 each I/O. For instance, using ``rw=write:4k`` will skip 4k for every
987 write. It turns sequential I/O into sequential I/O with holes. See the
988 :option:`rw_sequencer` option.
990 .. option:: rw_sequencer=str
992 If an offset modifier is given by appending a number to the ``rw=<str>``
993 line, then this option controls how that number modifies the I/O offset
994 being generated. Accepted values are:
997 Generate sequential offset.
999 Generate the same offset.
1001 ``sequential`` is only useful for random I/O, where fio would normally
1002 generate a new random offset for every I/O. If you append e.g. 8 to randread,
1003 you would get a new random offset for every 8 I/O's. The result would be a
1004 seek for only every 8 I/O's, instead of for every I/O. Use ``rw=randread:8``
1005 to specify that. As sequential I/O is already sequential, setting
1006 ``sequential`` for that would not result in any differences. ``identical``
1007 behaves in a similar fashion, except it sends the same offset 8 number of
1008 times before generating a new offset.
1010 .. option:: unified_rw_reporting=bool
1012 Fio normally reports statistics on a per data direction basis, meaning that
1013 reads, writes, and trims are accounted and reported separately. If this
1014 option is set fio sums the results and report them as "mixed" instead.
1016 .. option:: randrepeat=bool
1018 Seed the random number generator used for random I/O patterns in a
1019 predictable way so the pattern is repeatable across runs. Default: true.
1021 .. option:: allrandrepeat=bool
1023 Seed all random number generators in a predictable way so results are
1024 repeatable across runs. Default: false.
1026 .. option:: randseed=int
1028 Seed the random number generators based on this seed value, to be able to
1029 control what sequence of output is being generated. If not set, the random
1030 sequence depends on the :option:`randrepeat` setting.
1032 .. option:: fallocate=str
1034 Whether pre-allocation is performed when laying down files.
1035 Accepted values are:
1038 Do not pre-allocate space.
1041 Pre-allocate via :manpage:`posix_fallocate(3)`.
1044 Pre-allocate via :manpage:`fallocate(2)` with
1045 FALLOC_FL_KEEP_SIZE set.
1048 Backward-compatible alias for **none**.
1051 Backward-compatible alias for **posix**.
1053 May not be available on all supported platforms. **keep** is only available
1054 on Linux. If using ZFS on Solaris this must be set to **none** because ZFS
1055 doesn't support it. Default: **posix**.
1057 .. option:: fadvise_hint=str
1059 Use :manpage:`posix_fadvise(2)` to advise the kernel on what I/O patterns
1060 are likely to be issued. Accepted values are:
1063 Backwards-compatible hint for "no hint".
1066 Backwards compatible hint for "advise with fio workload type". This
1067 uses **FADV_RANDOM** for a random workload, and **FADV_SEQUENTIAL**
1068 for a sequential workload.
1071 Advise using **FADV_SEQUENTIAL**.
1074 Advise using **FADV_RANDOM**.
1076 .. option:: fadvise_stream=int
1078 Use :manpage:`posix_fadvise(2)` to advise the kernel what stream ID the
1079 writes issued belong to. Only supported on Linux. Note, this option may
1080 change going forward.
1082 .. option:: offset=int
1084 Start I/O at the provided offset in the file, given as either a fixed size or
1085 a percentage. If a percentage is given, the next ``blockalign``-ed offset
1086 will be used. Data before the given offset will not be touched. This
1087 effectively caps the file size at `real_size - offset`. Can be combined with
1088 :option:`size` to constrain the start and end range of the I/O workload.
1090 .. option:: offset_increment=int
1092 If this is provided, then the real offset becomes `offset + offset_increment
1093 * thread_number`, where the thread number is a counter that starts at 0 and
1094 is incremented for each sub-job (i.e. when :option:`numjobs` option is
1095 specified). This option is useful if there are several jobs which are
1096 intended to operate on a file in parallel disjoint segments, with even
1097 spacing between the starting points.
1099 .. option:: number_ios=int
1101 Fio will normally perform I/Os until it has exhausted the size of the region
1102 set by :option:`size`, or if it exhaust the allocated time (or hits an error
1103 condition). With this setting, the range/size can be set independently of
1104 the number of I/Os to perform. When fio reaches this number, it will exit
1105 normally and report status. Note that this does not extend the amount of I/O
1106 that will be done, it will only stop fio if this condition is met before
1107 other end-of-job criteria.
1109 .. option:: fsync=int
1111 If writing to a file, issue a sync of the dirty data for every number of
1112 blocks given. For example, if you give 32 as a parameter, fio will sync the
1113 file for every 32 writes issued. If fio is using non-buffered I/O, we may
1114 not sync the file. The exception is the sg I/O engine, which synchronizes
1115 the disk cache anyway. Defaults to 0, which means no sync every certain
1118 .. option:: fdatasync=int
1120 Like :option:`fsync` but uses :manpage:`fdatasync(2)` to only sync data and
1121 not metadata blocks. In Windows, FreeBSD, and DragonFlyBSD there is no
1122 :manpage:`fdatasync(2)`, this falls back to using :manpage:`fsync(2)`.
1123 Defaults to 0, which means no sync data every certain number of writes.
1125 .. option:: write_barrier=int
1127 Make every `N-th` write a barrier write.
1129 .. option:: sync_file_range=str:val
1131 Use :manpage:`sync_file_range(2)` for every `val` number of write
1132 operations. Fio will track range of writes that have happened since the last
1133 :manpage:`sync_file_range(2)` call. `str` can currently be one or more of:
1136 SYNC_FILE_RANGE_WAIT_BEFORE
1138 SYNC_FILE_RANGE_WRITE
1140 SYNC_FILE_RANGE_WAIT_AFTER
1142 So if you do ``sync_file_range=wait_before,write:8``, fio would use
1143 ``SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE`` for every 8
1144 writes. Also see the :manpage:`sync_file_range(2)` man page. This option is
1147 .. option:: overwrite=bool
1149 If true, writes to a file will always overwrite existing data. If the file
1150 doesn't already exist, it will be created before the write phase begins. If
1151 the file exists and is large enough for the specified write phase, nothing
1152 will be done. Default: false.
1154 .. option:: end_fsync=bool
1156 If true, :manpage:`fsync(2)` file contents when a write stage has completed.
1159 .. option:: fsync_on_close=bool
1161 If true, fio will :manpage:`fsync(2)` a dirty file on close. This differs
1162 from :option:`end_fsync` in that it will happen on every file close, not
1163 just at the end of the job. Default: false.
1165 .. option:: rwmixread=int
1167 Percentage of a mixed workload that should be reads. Default: 50.
1169 .. option:: rwmixwrite=int
1171 Percentage of a mixed workload that should be writes. If both
1172 :option:`rwmixread` and :option:`rwmixwrite` is given and the values do not
1173 add up to 100%, the latter of the two will be used to override the
1174 first. This may interfere with a given rate setting, if fio is asked to
1175 limit reads or writes to a certain rate. If that is the case, then the
1176 distribution may be skewed. Default: 50.
1178 .. option:: random_distribution=str:float[,str:float][,str:float]
1180 By default, fio will use a completely uniform random distribution when asked
1181 to perform random I/O. Sometimes it is useful to skew the distribution in
1182 specific ways, ensuring that some parts of the data is more hot than others.
1183 fio includes the following distribution models:
1186 Uniform random distribution
1195 Normal (Gaussian) distribution
1198 Zoned random distribution
1200 When using a **zipf** or **pareto** distribution, an input value is also
1201 needed to define the access pattern. For **zipf**, this is the `zipf
1202 theta`. For **pareto**, it's the `Pareto power`. Fio includes a test
1203 program, :command:`genzipf`, that can be used visualize what the given input
1204 values will yield in terms of hit rates. If you wanted to use **zipf** with
1205 a `theta` of 1.2, you would use ``random_distribution=zipf:1.2`` as the
1206 option. If a non-uniform model is used, fio will disable use of the random
1207 map. For the **gauss** distribution, a normal deviation is supplied as a
1208 value between 0 and 100.
1210 For a **zoned** distribution, fio supports specifying percentages of I/O
1211 access that should fall within what range of the file or device. For
1212 example, given a criteria of:
1214 * 60% of accesses should be to the first 10%
1215 * 30% of accesses should be to the next 20%
1216 * 8% of accesses should be to to the next 30%
1217 * 2% of accesses should be to the next 40%
1219 we can define that through zoning of the random accesses. For the above
1220 example, the user would do::
1222 random_distribution=zoned:60/10:30/20:8/30:2/40
1224 similarly to how :option:`bssplit` works for setting ranges and percentages
1225 of block sizes. Like :option:`bssplit`, it's possible to specify separate
1226 zones for reads, writes, and trims. If just one set is given, it'll apply to
1229 .. option:: percentage_random=int[,int][,int]
1231 For a random workload, set how big a percentage should be random. This
1232 defaults to 100%, in which case the workload is fully random. It can be set
1233 from anywhere from 0 to 100. Setting it to 0 would make the workload fully
1234 sequential. Any setting in between will result in a random mix of sequential
1235 and random I/O, at the given percentages. Comma-separated values may be
1236 specified for reads, writes, and trims as described in :option:`blocksize`.
1238 .. option:: norandommap
1240 Normally fio will cover every block of the file when doing random I/O. If
1241 this option is given, fio will just get a new random offset without looking
1242 at past I/O history. This means that some blocks may not be read or written,
1243 and that some blocks may be read/written more than once. If this option is
1244 used with :option:`verify` and multiple blocksizes (via :option:`bsrange`),
1245 only intact blocks are verified, i.e., partially-overwritten blocks are
1248 .. option:: softrandommap=bool
1250 See :option:`norandommap`. If fio runs with the random block map enabled and
1251 it fails to allocate the map, if this option is set it will continue without
1252 a random block map. As coverage will not be as complete as with random maps,
1253 this option is disabled by default.
1255 .. option:: random_generator=str
1257 Fio supports the following engines for generating
1258 I/O offsets for random I/O:
1261 Strong 2^88 cycle random number generator
1263 Linear feedback shift register generator
1265 Strong 64-bit 2^258 cycle random number generator
1267 **tausworthe** is a strong random number generator, but it requires tracking
1268 on the side if we want to ensure that blocks are only read or written
1269 once. **LFSR** guarantees that we never generate the same offset twice, and
1270 it's also less computationally expensive. It's not a true random generator,
1271 however, though for I/O purposes it's typically good enough. **LFSR** only
1272 works with single block sizes, not with workloads that use multiple block
1273 sizes. If used with such a workload, fio may read or write some blocks
1274 multiple times. The default value is **tausworthe**, unless the required
1275 space exceeds 2^32 blocks. If it does, then **tausworthe64** is
1276 selected automatically.
1282 .. option:: blocksize=int[,int][,int], bs=int[,int][,int]
1284 The block size in bytes used for I/O units. Default: 4096. A single value
1285 applies to reads, writes, and trims. Comma-separated values may be
1286 specified for reads, writes, and trims. A value not terminated in a comma
1287 applies to subsequent types.
1292 means 256k for reads, writes and trims.
1295 means 8k for reads, 32k for writes and trims.
1298 means 8k for reads, 32k for writes, and default for trims.
1301 means default for reads, 8k for writes and trims.
1304 means default for reads, 8k for writes, and default for trims.
1306 .. option:: blocksize_range=irange[,irange][,irange], bsrange=irange[,irange][,irange]
1308 A range of block sizes in bytes for I/O units. The issued I/O unit will
1309 always be a multiple of the minimum size, unless
1310 :option:`blocksize_unaligned` is set.
1312 Comma-separated ranges may be specified for reads, writes, and trims as
1313 described in :option:`blocksize`.
1315 Example: ``bsrange=1k-4k,2k-8k``.
1317 .. option:: bssplit=str[,str][,str]
1319 Sometimes you want even finer grained control of the block sizes issued, not
1320 just an even split between them. This option allows you to weight various
1321 block sizes, so that you are able to define a specific amount of block sizes
1322 issued. The format for this option is::
1324 bssplit=blocksize/percentage:blocksize/percentage
1326 for as many block sizes as needed. So if you want to define a workload that
1327 has 50% 64k blocks, 10% 4k blocks, and 40% 32k blocks, you would write::
1329 bssplit=4k/10:64k/50:32k/40
1331 Ordering does not matter. If the percentage is left blank, fio will fill in
1332 the remaining values evenly. So a bssplit option like this one::
1334 bssplit=4k/50:1k/:32k/
1336 would have 50% 4k ios, and 25% 1k and 32k ios. The percentages always add up
1337 to 100, if bssplit is given a range that adds up to more, it will error out.
1339 Comma-separated values may be specified for reads, writes, and trims as
1340 described in :option:`blocksize`.
1342 If you want a workload that has 50% 2k reads and 50% 4k reads, while having
1343 90% 4k writes and 10% 8k writes, you would specify::
1345 bssplit=2k/50:4k/50,4k/90,8k/10
1347 .. option:: blocksize_unaligned, bs_unaligned
1349 If set, fio will issue I/O units with any size within
1350 :option:`blocksize_range`, not just multiples of the minimum size. This
1351 typically won't work with direct I/O, as that normally requires sector
1354 .. option:: bs_is_seq_rand
1356 If this option is set, fio will use the normal read,write blocksize settings
1357 as sequential,random blocksize settings instead. Any random read or write
1358 will use the WRITE blocksize settings, and any sequential read or write will
1359 use the READ blocksize settings.
1361 .. option:: blockalign=int[,int][,int], ba=int[,int][,int]
1363 Boundary to which fio will align random I/O units. Default:
1364 :option:`blocksize`. Minimum alignment is typically 512b for using direct
1365 I/O, though it usually depends on the hardware block size. This option is
1366 mutually exclusive with using a random map for files, so it will turn off
1367 that option. Comma-separated values may be specified for reads, writes, and
1368 trims as described in :option:`blocksize`.
1374 .. option:: zero_buffers
1376 Initialize buffers with all zeros. Default: fill buffers with random data.
1378 .. option:: refill_buffers
1380 If this option is given, fio will refill the I/O buffers on every
1381 submit. The default is to only fill it at init time and reuse that
1382 data. Only makes sense if zero_buffers isn't specified, naturally. If data
1383 verification is enabled, `refill_buffers` is also automatically enabled.
1385 .. option:: scramble_buffers=bool
1387 If :option:`refill_buffers` is too costly and the target is using data
1388 deduplication, then setting this option will slightly modify the I/O buffer
1389 contents to defeat normal de-dupe attempts. This is not enough to defeat
1390 more clever block compression attempts, but it will stop naive dedupe of
1391 blocks. Default: true.
1393 .. option:: buffer_compress_percentage=int
1395 If this is set, then fio will attempt to provide I/O buffer content (on
1396 WRITEs) that compress to the specified level. Fio does this by providing a
1397 mix of random data and a fixed pattern. The fixed pattern is either zeroes,
1398 or the pattern specified by :option:`buffer_pattern`. If the pattern option
1399 is used, it might skew the compression ratio slightly. Note that this is per
1400 block size unit, for file/disk wide compression level that matches this
1401 setting, you'll also want to set :option:`refill_buffers`.
1403 .. option:: buffer_compress_chunk=int
1405 See :option:`buffer_compress_percentage`. This setting allows fio to manage
1406 how big the ranges of random data and zeroed data is. Without this set, fio
1407 will provide :option:`buffer_compress_percentage` of blocksize random data,
1408 followed by the remaining zeroed. With this set to some chunk size smaller
1409 than the block size, fio can alternate random and zeroed data throughout the
1412 .. option:: buffer_pattern=str
1414 If set, fio will fill the I/O buffers with this pattern or with the contents
1415 of a file. If not set, the contents of I/O buffers are defined by the other
1416 options related to buffer contents. The setting can be any pattern of bytes,
1417 and can be prefixed with 0x for hex values. It may also be a string, where
1418 the string must then be wrapped with ``""``. Or it may also be a filename,
1419 where the filename must be wrapped with ``''`` in which case the file is
1420 opened and read. Note that not all the file contents will be read if that
1421 would cause the buffers to overflow. So, for example::
1423 buffer_pattern='filename'
1427 buffer_pattern="abcd"
1435 buffer_pattern=0xdeadface
1437 Also you can combine everything together in any order::
1439 buffer_pattern=0xdeadface"abcd"-12'filename'
1441 .. option:: dedupe_percentage=int
1443 If set, fio will generate this percentage of identical buffers when
1444 writing. These buffers will be naturally dedupable. The contents of the
1445 buffers depend on what other buffer compression settings have been set. It's
1446 possible to have the individual buffers either fully compressible, or not at
1447 all. This option only controls the distribution of unique buffers.
1449 .. option:: invalidate=bool
1451 Invalidate the buffer/page cache parts for this file prior to starting
1452 I/O if the platform and file type support it. Defaults to true.
1453 This will be ignored if :option:`pre_read` is also specified for the
1456 .. option:: sync=bool
1458 Use synchronous I/O for buffered writes. For the majority of I/O engines,
1459 this means using O_SYNC. Default: false.
1461 .. option:: iomem=str, mem=str
1463 Fio can use various types of memory as the I/O unit buffer. The allowed
1467 Use memory from :manpage:`malloc(3)` as the buffers. Default memory
1471 Use shared memory as the buffers. Allocated through
1472 :manpage:`shmget(2)`.
1475 Same as shm, but use huge pages as backing.
1478 Use mmap to allocate buffers. May either be anonymous memory, or can
1479 be file backed if a filename is given after the option. The format
1480 is `mem=mmap:/path/to/file`.
1483 Use a memory mapped huge file as the buffer backing. Append filename
1484 after mmaphuge, ala `mem=mmaphuge:/hugetlbfs/file`.
1487 Same as mmap, but use a MMAP_SHARED mapping.
1490 Use GPU memory as the buffers for GPUDirect RDMA benchmark.
1492 The area allocated is a function of the maximum allowed bs size for the job,
1493 multiplied by the I/O depth given. Note that for **shmhuge** and
1494 **mmaphuge** to work, the system must have free huge pages allocated. This
1495 can normally be checked and set by reading/writing
1496 :file:`/proc/sys/vm/nr_hugepages` on a Linux system. Fio assumes a huge page
1497 is 4MiB in size. So to calculate the number of huge pages you need for a
1498 given job file, add up the I/O depth of all jobs (normally one unless
1499 :option:`iodepth` is used) and multiply by the maximum bs set. Then divide
1500 that number by the huge page size. You can see the size of the huge pages in
1501 :file:`/proc/meminfo`. If no huge pages are allocated by having a non-zero
1502 number in `nr_hugepages`, using **mmaphuge** or **shmhuge** will fail. Also
1503 see :option:`hugepage-size`.
1505 **mmaphuge** also needs to have hugetlbfs mounted and the file location
1506 should point there. So if it's mounted in :file:`/huge`, you would use
1507 `mem=mmaphuge:/huge/somefile`.
1509 .. option:: iomem_align=int
1511 This indicates the memory alignment of the I/O memory buffers. Note that
1512 the given alignment is applied to the first I/O unit buffer, if using
1513 :option:`iodepth` the alignment of the following buffers are given by the
1514 :option:`bs` used. In other words, if using a :option:`bs` that is a
1515 multiple of the page sized in the system, all buffers will be aligned to
1516 this value. If using a :option:`bs` that is not page aligned, the alignment
1517 of subsequent I/O memory buffers is the sum of the :option:`iomem_align` and
1520 .. option:: hugepage-size=int
1522 Defines the size of a huge page. Must at least be equal to the system
1523 setting, see :file:`/proc/meminfo`. Defaults to 4MiB. Should probably
1524 always be a multiple of megabytes, so using ``hugepage-size=Xm`` is the
1525 preferred way to set this to avoid setting a non-pow-2 bad value.
1527 .. option:: lockmem=int
1529 Pin the specified amount of memory with :manpage:`mlock(2)`. Can be used to
1530 simulate a smaller amount of memory. The amount specified is per worker.
1536 .. option:: size=int
1538 The total size of file I/O for each thread of this job. Fio will run until
1539 this many bytes has been transferred, unless runtime is limited by other options
1540 (such as :option:`runtime`, for instance, or increased/decreased by :option:`io_size`).
1541 Fio will divide this size between the available files determined by options
1542 such as :option:`nrfiles`, :option:`filename`, unless :option:`filesize` is
1543 specified by the job. If the result of division happens to be 0, the size is
1544 set to the physical size of the given files or devices if they exist.
1545 If this option is not specified, fio will use the full size of the given
1546 files or devices. If the files do not exist, size must be given. It is also
1547 possible to give size as a percentage between 1 and 100. If ``size=20%`` is
1548 given, fio will use 20% of the full size of the given files or devices.
1549 Can be combined with :option:`offset` to constrain the start and end range
1550 that I/O will be done within.
1552 .. option:: io_size=int, io_limit=int
1554 Normally fio operates within the region set by :option:`size`, which means
1555 that the :option:`size` option sets both the region and size of I/O to be
1556 performed. Sometimes that is not what you want. With this option, it is
1557 possible to define just the amount of I/O that fio should do. For instance,
1558 if :option:`size` is set to 20GiB and :option:`io_size` is set to 5GiB, fio
1559 will perform I/O within the first 20GiB but exit when 5GiB have been
1560 done. The opposite is also possible -- if :option:`size` is set to 20GiB,
1561 and :option:`io_size` is set to 40GiB, then fio will do 40GiB of I/O within
1562 the 0..20GiB region.
1564 .. option:: filesize=int
1566 Individual file sizes. May be a range, in which case fio will select sizes
1567 for files at random within the given range and limited to :option:`size` in
1568 total (if that is given). If not given, each created file is the same size.
1569 This option overrides :option:`size` in terms of file size, which means
1570 this value is used as a fixed size or possible range of each file.
1572 .. option:: file_append=bool
1574 Perform I/O after the end of the file. Normally fio will operate within the
1575 size of a file. If this option is set, then fio will append to the file
1576 instead. This has identical behavior to setting :option:`offset` to the size
1577 of a file. This option is ignored on non-regular files.
1579 .. option:: fill_device=bool, fill_fs=bool
1581 Sets size to something really large and waits for ENOSPC (no space left on
1582 device) as the terminating condition. Only makes sense with sequential
1583 write. For a read workload, the mount point will be filled first then I/O
1584 started on the result. This option doesn't make sense if operating on a raw
1585 device node, since the size of that is already known by the file system.
1586 Additionally, writing beyond end-of-device will not return ENOSPC there.
1592 .. option:: ioengine=str
1594 Defines how the job issues I/O to the file. The following types are defined:
1597 Basic :manpage:`read(2)` or :manpage:`write(2)`
1598 I/O. :manpage:`lseek(2)` is used to position the I/O location.
1599 See :option:`fsync` and :option:`fdatasync` for syncing write I/Os.
1602 Basic :manpage:`pread(2)` or :manpage:`pwrite(2)` I/O. Default on
1603 all supported operating systems except for Windows.
1606 Basic :manpage:`readv(2)` or :manpage:`writev(2)` I/O. Will emulate
1607 queuing by coalescing adjacent I/Os into a single submission.
1610 Basic :manpage:`preadv(2)` or :manpage:`pwritev(2)` I/O.
1613 Basic :manpage:`preadv2(2)` or :manpage:`pwritev2(2)` I/O.
1616 Linux native asynchronous I/O. Note that Linux may only support
1617 queued behaviour with non-buffered I/O (set ``direct=1`` or
1619 This engine defines engine specific options.
1622 POSIX asynchronous I/O using :manpage:`aio_read(3)` and
1623 :manpage:`aio_write(3)`.
1626 Solaris native asynchronous I/O.
1629 Windows native asynchronous I/O. Default on Windows.
1632 File is memory mapped with :manpage:`mmap(2)` and data copied
1633 to/from using :manpage:`memcpy(3)`.
1636 :manpage:`splice(2)` is used to transfer the data and
1637 :manpage:`vmsplice(2)` to transfer data from user space to the
1641 SCSI generic sg v3 I/O. May either be synchronous using the SG_IO
1642 ioctl, or if the target is an sg character device we use
1643 :manpage:`read(2)` and :manpage:`write(2)` for asynchronous
1644 I/O. Requires filename option to specify either block or character
1648 Doesn't transfer any data, just pretends to. This is mainly used to
1649 exercise fio itself and for debugging/testing purposes.
1652 Transfer over the network to given ``host:port``. Depending on the
1653 :option:`protocol` used, the :option:`hostname`, :option:`port`,
1654 :option:`listen` and :option:`filename` options are used to specify
1655 what sort of connection to make, while the :option:`protocol` option
1656 determines which protocol will be used. This engine defines engine
1660 Like **net**, but uses :manpage:`splice(2)` and
1661 :manpage:`vmsplice(2)` to map data and send/receive.
1662 This engine defines engine specific options.
1665 Doesn't transfer any data, but burns CPU cycles according to the
1666 :option:`cpuload` and :option:`cpuchunks` options. Setting
1667 :option:`cpuload`\=85 will cause that job to do nothing but burn 85%
1668 of the CPU. In case of SMP machines, use :option:`numjobs`
1669 =<no_of_cpu> to get desired CPU usage, as the cpuload only loads a
1670 single CPU at the desired rate. A job never finishes unless there is
1671 at least one non-cpuio job.
1674 The GUASI I/O engine is the Generic Userspace Asyncronous Syscall
1675 Interface approach to async I/O. See
1677 http://www.xmailserver.org/guasi-lib.html
1679 for more info on GUASI.
1682 The RDMA I/O engine supports both RDMA memory semantics
1683 (RDMA_WRITE/RDMA_READ) and channel semantics (Send/Recv) for the
1684 InfiniBand, RoCE and iWARP protocols.
1687 I/O engine that does regular fallocate to simulate data transfer as
1691 does fallocate(,mode = FALLOC_FL_KEEP_SIZE,).
1694 does fallocate(,mode = 0).
1697 does fallocate(,mode = FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE).
1700 I/O engine that sends :manpage:`ftruncate(2)` operations in response
1701 to write (DDIR_WRITE) events. Each ftruncate issued sets the file's
1702 size to the current block offset. Block size is ignored.
1705 I/O engine that does regular EXT4_IOC_MOVE_EXT ioctls to simulate
1706 defragment activity in request to DDIR_WRITE event.
1709 I/O engine supporting direct access to Ceph Rados Block Devices
1710 (RBD) via librbd without the need to use the kernel rbd driver. This
1711 ioengine defines engine specific options.
1714 Using Glusterfs libgfapi sync interface to direct access to
1715 Glusterfs volumes without having to go through FUSE. This ioengine
1716 defines engine specific options.
1719 Using Glusterfs libgfapi async interface to direct access to
1720 Glusterfs volumes without having to go through FUSE. This ioengine
1721 defines engine specific options.
1724 Read and write through Hadoop (HDFS). The :file:`filename` option
1725 is used to specify host,port of the hdfs name-node to connect. This
1726 engine interprets offsets a little differently. In HDFS, files once
1727 created cannot be modified. So random writes are not possible. To
1728 imitate this, libhdfs engine expects bunch of small files to be
1729 created over HDFS, and engine will randomly pick a file out of those
1730 files based on the offset generated by fio backend. (see the example
1731 job file to create such files, use ``rw=write`` option). Please
1732 note, you might want to set necessary environment variables to work
1733 with hdfs/libhdfs properly. Each job uses its own connection to
1737 Read, write and erase an MTD character device (e.g.,
1738 :file:`/dev/mtd0`). Discards are treated as erases. Depending on the
1739 underlying device type, the I/O may have to go in a certain pattern,
1740 e.g., on NAND, writing sequentially to erase blocks and discarding
1741 before overwriting. The writetrim mode works well for this
1745 Read and write using filesystem DAX to a file on a filesystem
1746 mounted with DAX on a persistent memory device through the NVML
1750 Read and write using device DAX to a persistent memory device (e.g.,
1751 /dev/dax0.0) through the NVML libpmem library.
1754 Prefix to specify loading an external I/O engine object file. Append
1755 the engine filename, e.g. ``ioengine=external:/tmp/foo.o`` to load
1756 ioengine :file:`foo.o` in :file:`/tmp`.
1759 I/O engine specific parameters
1760 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1762 In addition, there are some parameters which are only valid when a specific
1763 ioengine is in use. These are used identically to normal parameters, with the
1764 caveat that when used on the command line, they must come after the
1765 :option:`ioengine` that defines them is selected.
1767 .. option:: userspace_reap : [libaio]
1769 Normally, with the libaio engine in use, fio will use the
1770 :manpage:`io_getevents(2)` system call to reap newly returned events. With
1771 this flag turned on, the AIO ring will be read directly from user-space to
1772 reap events. The reaping mode is only enabled when polling for a minimum of
1773 0 events (e.g. when :option:`iodepth_batch_complete` `=0`).
1775 .. option:: hipri : [pvsync2]
1777 Set RWF_HIPRI on I/O, indicating to the kernel that it's of higher priority
1780 .. option:: cpuload=int : [cpuio]
1782 Attempt to use the specified percentage of CPU cycles. This is a mandatory
1783 option when using cpuio I/O engine.
1785 .. option:: cpuchunks=int : [cpuio]
1787 Split the load into cycles of the given time. In microseconds.
1789 .. option:: exit_on_io_done=bool : [cpuio]
1791 Detect when I/O threads are done, then exit.
1793 .. option:: hostname=str : [netsplice] [net]
1795 The host name or IP address to use for TCP or UDP based I/O. If the job is
1796 a TCP listener or UDP reader, the host name is not used and must be omitted
1797 unless it is a valid UDP multicast address.
1799 .. option:: namenode=str : [libhdfs]
1801 The host name or IP address of a HDFS cluster namenode to contact.
1803 .. option:: port=int
1807 The TCP or UDP port to bind to or connect to. If this is used with
1808 :option:`numjobs` to spawn multiple instances of the same job type, then
1809 this will be the starting port number since fio will use a range of
1814 the listening port of the HFDS cluster namenode.
1816 .. option:: interface=str : [netsplice] [net]
1818 The IP address of the network interface used to send or receive UDP
1821 .. option:: ttl=int : [netsplice] [net]
1823 Time-to-live value for outgoing UDP multicast packets. Default: 1.
1825 .. option:: nodelay=bool : [netsplice] [net]
1827 Set TCP_NODELAY on TCP connections.
1829 .. option:: protocol=str : [netsplice] [net]
1831 .. option:: proto=str : [netsplice] [net]
1833 The network protocol to use. Accepted values are:
1836 Transmission control protocol.
1838 Transmission control protocol V6.
1840 User datagram protocol.
1842 User datagram protocol V6.
1846 When the protocol is TCP or UDP, the port must also be given, as well as the
1847 hostname if the job is a TCP listener or UDP reader. For unix sockets, the
1848 normal filename option should be used and the port is invalid.
1850 .. option:: listen : [net]
1852 For TCP network connections, tell fio to listen for incoming connections
1853 rather than initiating an outgoing connection. The :option:`hostname` must
1854 be omitted if this option is used.
1856 .. option:: pingpong : [net]
1858 Normally a network writer will just continue writing data, and a network
1859 reader will just consume packages. If ``pingpong=1`` is set, a writer will
1860 send its normal payload to the reader, then wait for the reader to send the
1861 same payload back. This allows fio to measure network latencies. The
1862 submission and completion latencies then measure local time spent sending or
1863 receiving, and the completion latency measures how long it took for the
1864 other end to receive and send back. For UDP multicast traffic
1865 ``pingpong=1`` should only be set for a single reader when multiple readers
1866 are listening to the same address.
1868 .. option:: window_size : [net]
1870 Set the desired socket buffer size for the connection.
1872 .. option:: mss : [net]
1874 Set the TCP maximum segment size (TCP_MAXSEG).
1876 .. option:: donorname=str : [e4defrag]
1878 File will be used as a block donor(swap extents between files).
1880 .. option:: inplace=int : [e4defrag]
1882 Configure donor file blocks allocation strategy:
1885 Default. Preallocate donor's file on init.
1887 Allocate space immediately inside defragment event, and free right
1890 .. option:: clustername=str : [rbd]
1892 Specifies the name of the Ceph cluster.
1894 .. option:: rbdname=str : [rbd]
1896 Specifies the name of the RBD.
1898 .. option:: pool=str : [rbd]
1900 Specifies the name of the Ceph pool containing RBD.
1902 .. option:: clientname=str : [rbd]
1904 Specifies the username (without the 'client.' prefix) used to access the
1905 Ceph cluster. If the *clustername* is specified, the *clientname* shall be
1906 the full *type.id* string. If no type. prefix is given, fio will add
1907 'client.' by default.
1909 .. option:: skip_bad=bool : [mtd]
1911 Skip operations against known bad blocks.
1913 .. option:: hdfsdirectory : [libhdfs]
1915 libhdfs will create chunk in this HDFS directory.
1917 .. option:: chunk_size : [libhdfs]
1919 the size of the chunk to use for each file.
1925 .. option:: iodepth=int
1927 Number of I/O units to keep in flight against the file. Note that
1928 increasing *iodepth* beyond 1 will not affect synchronous ioengines (except
1929 for small degrees when :option:`verify_async` is in use). Even async
1930 engines may impose OS restrictions causing the desired depth not to be
1931 achieved. This may happen on Linux when using libaio and not setting
1932 :option:`direct`\=1, since buffered I/O is not async on that OS. Keep an
1933 eye on the I/O depth distribution in the fio output to verify that the
1934 achieved depth is as expected. Default: 1.
1936 .. option:: iodepth_batch_submit=int, iodepth_batch=int
1938 This defines how many pieces of I/O to submit at once. It defaults to 1
1939 which means that we submit each I/O as soon as it is available, but can be
1940 raised to submit bigger batches of I/O at the time. If it is set to 0 the
1941 :option:`iodepth` value will be used.
1943 .. option:: iodepth_batch_complete_min=int, iodepth_batch_complete=int
1945 This defines how many pieces of I/O to retrieve at once. It defaults to 1
1946 which means that we'll ask for a minimum of 1 I/O in the retrieval process
1947 from the kernel. The I/O retrieval will go on until we hit the limit set by
1948 :option:`iodepth_low`. If this variable is set to 0, then fio will always
1949 check for completed events before queuing more I/O. This helps reduce I/O
1950 latency, at the cost of more retrieval system calls.
1952 .. option:: iodepth_batch_complete_max=int
1954 This defines maximum pieces of I/O to retrieve at once. This variable should
1955 be used along with :option:`iodepth_batch_complete_min`\=int variable,
1956 specifying the range of min and max amount of I/O which should be
1957 retrieved. By default it is equal to :option:`iodepth_batch_complete_min`
1962 iodepth_batch_complete_min=1
1963 iodepth_batch_complete_max=<iodepth>
1965 which means that we will retrieve at least 1 I/O and up to the whole
1966 submitted queue depth. If none of I/O has been completed yet, we will wait.
1970 iodepth_batch_complete_min=0
1971 iodepth_batch_complete_max=<iodepth>
1973 which means that we can retrieve up to the whole submitted queue depth, but
1974 if none of I/O has been completed yet, we will NOT wait and immediately exit
1975 the system call. In this example we simply do polling.
1977 .. option:: iodepth_low=int
1979 The low water mark indicating when to start filling the queue
1980 again. Defaults to the same as :option:`iodepth`, meaning that fio will
1981 attempt to keep the queue full at all times. If :option:`iodepth` is set to
1982 e.g. 16 and *iodepth_low* is set to 4, then after fio has filled the queue of
1983 16 requests, it will let the depth drain down to 4 before starting to fill
1986 .. option:: io_submit_mode=str
1988 This option controls how fio submits the I/O to the I/O engine. The default
1989 is `inline`, which means that the fio job threads submit and reap I/O
1990 directly. If set to `offload`, the job threads will offload I/O submission
1991 to a dedicated pool of I/O threads. This requires some coordination and thus
1992 has a bit of extra overhead, especially for lower queue depth I/O where it
1993 can increase latencies. The benefit is that fio can manage submission rates
1994 independently of the device completion rates. This avoids skewed latency
1995 reporting if I/O gets back up on the device side (the coordinated omission
2002 .. option:: thinktime=time
2004 Stall the job for the specified period of time after an I/O has completed before issuing the
2005 next. May be used to simulate processing being done by an application.
2006 When the unit is omitted, the value is interpreted in microseconds. See
2007 :option:`thinktime_blocks` and :option:`thinktime_spin`.
2009 .. option:: thinktime_spin=time
2011 Only valid if :option:`thinktime` is set - pretend to spend CPU time doing
2012 something with the data received, before falling back to sleeping for the
2013 rest of the period specified by :option:`thinktime`. When the unit is
2014 omitted, the value is interpreted in microseconds.
2016 .. option:: thinktime_blocks=int
2018 Only valid if :option:`thinktime` is set - control how many blocks to issue,
2019 before waiting `thinktime` usecs. If not set, defaults to 1 which will make
2020 fio wait `thinktime` usecs after every block. This effectively makes any
2021 queue depth setting redundant, since no more than 1 I/O will be queued
2022 before we have to complete it and do our thinktime. In other words, this
2023 setting effectively caps the queue depth if the latter is larger.
2025 .. option:: rate=int[,int][,int]
2027 Cap the bandwidth used by this job. The number is in bytes/sec, the normal
2028 suffix rules apply. Comma-separated values may be specified for reads,
2029 writes, and trims as described in :option:`blocksize`.
2031 .. option:: rate_min=int[,int][,int]
2033 Tell fio to do whatever it can to maintain at least this bandwidth. Failing
2034 to meet this requirement will cause the job to exit. Comma-separated values
2035 may be specified for reads, writes, and trims as described in
2036 :option:`blocksize`.
2038 .. option:: rate_iops=int[,int][,int]
2040 Cap the bandwidth to this number of IOPS. Basically the same as
2041 :option:`rate`, just specified independently of bandwidth. If the job is
2042 given a block size range instead of a fixed value, the smallest block size
2043 is used as the metric. Comma-separated values may be specified for reads,
2044 writes, and trims as described in :option:`blocksize`.
2046 .. option:: rate_iops_min=int[,int][,int]
2048 If fio doesn't meet this rate of I/O, it will cause the job to exit.
2049 Comma-separated values may be specified for reads, writes, and trims as
2050 described in :option:`blocksize`.
2052 .. option:: rate_process=str
2054 This option controls how fio manages rated I/O submissions. The default is
2055 `linear`, which submits I/O in a linear fashion with fixed delays between
2056 I/Os that gets adjusted based on I/O completion rates. If this is set to
2057 `poisson`, fio will submit I/O based on a more real world random request
2058 flow, known as the Poisson process
2059 (https://en.wikipedia.org/wiki/Poisson_point_process). The lambda will be
2060 10^6 / IOPS for the given workload.
2066 .. option:: latency_target=time
2068 If set, fio will attempt to find the max performance point that the given
2069 workload will run at while maintaining a latency below this target. When
2070 the unit is omitted, the value is interpreted in microseconds. See
2071 :option:`latency_window` and :option:`latency_percentile`.
2073 .. option:: latency_window=time
2075 Used with :option:`latency_target` to specify the sample window that the job
2076 is run at varying queue depths to test the performance. When the unit is
2077 omitted, the value is interpreted in microseconds.
2079 .. option:: latency_percentile=float
2081 The percentage of I/Os that must fall within the criteria specified by
2082 :option:`latency_target` and :option:`latency_window`. If not set, this
2083 defaults to 100.0, meaning that all I/Os must be equal or below to the value
2084 set by :option:`latency_target`.
2086 .. option:: max_latency=time
2088 If set, fio will exit the job with an ETIMEDOUT error if it exceeds this
2089 maximum latency. When the unit is omitted, the value is interpreted in
2092 .. option:: rate_cycle=int
2094 Average bandwidth for :option:`rate` and :option:`rate_min` over this number
2095 of milliseconds. Defaults to 1000.
2101 .. option:: write_iolog=str
2103 Write the issued I/O patterns to the specified file. See
2104 :option:`read_iolog`. Specify a separate file for each job, otherwise the
2105 iologs will be interspersed and the file may be corrupt.
2107 .. option:: read_iolog=str
2109 Open an iolog with the specified file name and replay the I/O patterns it
2110 contains. This can be used to store a workload and replay it sometime
2111 later. The iolog given may also be a blktrace binary file, which allows fio
2112 to replay a workload captured by :command:`blktrace`. See
2113 :manpage:`blktrace(8)` for how to capture such logging data. For blktrace
2114 replay, the file needs to be turned into a blkparse binary data file first
2115 (``blkparse <device> -o /dev/null -d file_for_fio.bin``).
2117 .. option:: replay_no_stall=int
2119 When replaying I/O with :option:`read_iolog` the default behavior is to
2120 attempt to respect the time stamps within the log and replay them with the
2121 appropriate delay between IOPS. By setting this variable fio will not
2122 respect the timestamps and attempt to replay them as fast as possible while
2123 still respecting ordering. The result is the same I/O pattern to a given
2124 device, but different timings.
2126 .. option:: replay_redirect=str
2128 While replaying I/O patterns using :option:`read_iolog` the default behavior
2129 is to replay the IOPS onto the major/minor device that each IOP was recorded
2130 from. This is sometimes undesirable because on a different machine those
2131 major/minor numbers can map to a different device. Changing hardware on the
2132 same system can also result in a different major/minor mapping.
2133 ``replay_redirect`` causes all IOPS to be replayed onto the single specified
2134 device regardless of the device it was recorded
2135 from. i.e. :option:`replay_redirect`\= :file:`/dev/sdc` would cause all I/O
2136 in the blktrace or iolog to be replayed onto :file:`/dev/sdc`. This means
2137 multiple devices will be replayed onto a single device, if the trace
2138 contains multiple devices. If you want multiple devices to be replayed
2139 concurrently to multiple redirected devices you must blkparse your trace
2140 into separate traces and replay them with independent fio invocations.
2141 Unfortunately this also breaks the strict time ordering between multiple
2144 .. option:: replay_align=int
2146 Force alignment of I/O offsets and lengths in a trace to this power of 2
2149 .. option:: replay_scale=int
2151 Scale sector offsets down by this factor when replaying traces.
2154 Threads, processes and job synchronization
2155 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2159 Fio defaults to forking jobs, however if this option is given, fio will use
2160 POSIX Threads function :manpage:`pthread_create(3)` to create threads instead
2161 of forking processes.
2163 .. option:: wait_for=str
2165 Specifies the name of the already defined job to wait for. Single waitee
2166 name only may be specified. If set, the job won't be started until all
2167 workers of the waitee job are done.
2169 ``wait_for`` operates on the job name basis, so there are a few
2170 limitations. First, the waitee must be defined prior to the waiter job
2171 (meaning no forward references). Second, if a job is being referenced as a
2172 waitee, it must have a unique name (no duplicate waitees).
2174 .. option:: nice=int
2176 Run the job with the given nice value. See man :manpage:`nice(2)`.
2178 On Windows, values less than -15 set the process class to "High"; -1 through
2179 -15 set "Above Normal"; 1 through 15 "Below Normal"; and above 15 "Idle"
2182 .. option:: prio=int
2184 Set the I/O priority value of this job. Linux limits us to a positive value
2185 between 0 and 7, with 0 being the highest. See man
2186 :manpage:`ionice(1)`. Refer to an appropriate manpage for other operating
2187 systems since meaning of priority may differ.
2189 .. option:: prioclass=int
2191 Set the I/O priority class. See man :manpage:`ionice(1)`.
2193 .. option:: cpumask=int
2195 Set the CPU affinity of this job. The parameter given is a bitmask of
2196 allowed CPU's the job may run on. So if you want the allowed CPUs to be 1
2197 and 5, you would pass the decimal value of (1 << 1 | 1 << 5), or 34. See man
2198 :manpage:`sched_setaffinity(2)`. This may not work on all supported
2199 operating systems or kernel versions. This option doesn't work well for a
2200 higher CPU count than what you can store in an integer mask, so it can only
2201 control cpus 1-32. For boxes with larger CPU counts, use
2202 :option:`cpus_allowed`.
2204 .. option:: cpus_allowed=str
2206 Controls the same options as :option:`cpumask`, but it allows a text setting
2207 of the permitted CPUs instead. So to use CPUs 1 and 5, you would specify
2208 ``cpus_allowed=1,5``. This options also allows a range of CPUs. Say you
2209 wanted a binding to CPUs 1, 5, and 8-15, you would set
2210 ``cpus_allowed=1,5,8-15``.
2212 .. option:: cpus_allowed_policy=str
2214 Set the policy of how fio distributes the CPUs specified by
2215 :option:`cpus_allowed` or cpumask. Two policies are supported:
2218 All jobs will share the CPU set specified.
2220 Each job will get a unique CPU from the CPU set.
2222 **shared** is the default behaviour, if the option isn't specified. If
2223 **split** is specified, then fio will will assign one cpu per job. If not
2224 enough CPUs are given for the jobs listed, then fio will roundrobin the CPUs
2227 .. option:: numa_cpu_nodes=str
2229 Set this job running on specified NUMA nodes' CPUs. The arguments allow
2230 comma delimited list of cpu numbers, A-B ranges, or `all`. Note, to enable
2231 numa options support, fio must be built on a system with libnuma-dev(el)
2234 .. option:: numa_mem_policy=str
2236 Set this job's memory policy and corresponding NUMA nodes. Format of the
2241 ``mode`` is one of the following memory policy: ``default``, ``prefer``,
2242 ``bind``, ``interleave``, ``local`` For ``default`` and ``local`` memory
2243 policy, no node is needed to be specified. For ``prefer``, only one node is
2244 allowed. For ``bind`` and ``interleave``, it allow comma delimited list of
2245 numbers, A-B ranges, or `all`.
2247 .. option:: cgroup=str
2249 Add job to this control group. If it doesn't exist, it will be created. The
2250 system must have a mounted cgroup blkio mount point for this to work. If
2251 your system doesn't have it mounted, you can do so with::
2253 # mount -t cgroup -o blkio none /cgroup
2255 .. option:: cgroup_weight=int
2257 Set the weight of the cgroup to this value. See the documentation that comes
2258 with the kernel, allowed values are in the range of 100..1000.
2260 .. option:: cgroup_nodelete=bool
2262 Normally fio will delete the cgroups it has created after the job
2263 completion. To override this behavior and to leave cgroups around after the
2264 job completion, set ``cgroup_nodelete=1``. This can be useful if one wants
2265 to inspect various cgroup files after job completion. Default: false.
2267 .. option:: flow_id=int
2269 The ID of the flow. If not specified, it defaults to being a global
2270 flow. See :option:`flow`.
2272 .. option:: flow=int
2274 Weight in token-based flow control. If this value is used, then there is a
2275 'flow counter' which is used to regulate the proportion of activity between
2276 two or more jobs. Fio attempts to keep this flow counter near zero. The
2277 ``flow`` parameter stands for how much should be added or subtracted to the
2278 flow counter on each iteration of the main I/O loop. That is, if one job has
2279 ``flow=8`` and another job has ``flow=-1``, then there will be a roughly 1:8
2280 ratio in how much one runs vs the other.
2282 .. option:: flow_watermark=int
2284 The maximum value that the absolute value of the flow counter is allowed to
2285 reach before the job must wait for a lower value of the counter.
2287 .. option:: flow_sleep=int
2289 The period of time, in microseconds, to wait after the flow watermark has
2290 been exceeded before retrying operations.
2292 .. option:: stonewall, wait_for_previous
2294 Wait for preceding jobs in the job file to exit, before starting this
2295 one. Can be used to insert serialization points in the job file. A stone
2296 wall also implies starting a new reporting group, see
2297 :option:`group_reporting`.
2301 When one job finishes, terminate the rest. The default is to wait for each
2302 job to finish, sometimes that is not the desired action.
2304 .. option:: exec_prerun=str
2306 Before running this job, issue the command specified through
2307 :manpage:`system(3)`. Output is redirected in a file called
2308 :file:`jobname.prerun.txt`.
2310 .. option:: exec_postrun=str
2312 After the job completes, issue the command specified though
2313 :manpage:`system(3)`. Output is redirected in a file called
2314 :file:`jobname.postrun.txt`.
2318 Instead of running as the invoking user, set the user ID to this value
2319 before the thread/process does any work.
2323 Set group ID, see :option:`uid`.
2329 .. option:: verify_only
2331 Do not perform specified workload, only verify data still matches previous
2332 invocation of this workload. This option allows one to check data multiple
2333 times at a later date without overwriting it. This option makes sense only
2334 for workloads that write data, and does not support workloads with the
2335 :option:`time_based` option set.
2337 .. option:: do_verify=bool
2339 Run the verify phase after a write phase. Only valid if :option:`verify` is
2342 .. option:: verify=str
2344 If writing to a file, fio can verify the file contents after each iteration
2345 of the job. Each verification method also implies verification of special
2346 header, which is written to the beginning of each block. This header also
2347 includes meta information, like offset of the block, block number, timestamp
2348 when block was written, etc. :option:`verify` can be combined with
2349 :option:`verify_pattern` option. The allowed values are:
2352 Use an md5 sum of the data area and store it in the header of
2356 Use an experimental crc64 sum of the data area and store it in the
2357 header of each block.
2360 Use a crc32c sum of the data area and store it in the header of each
2364 Use hardware assisted crc32c calculation provided on SSE4.2 enabled
2365 processors. Falls back to regular software crc32c, if not supported
2369 Use a crc32 sum of the data area and store it in the header of each
2373 Use a crc16 sum of the data area and store it in the header of each
2377 Use a crc7 sum of the data area and store it in the header of each
2381 Use xxhash as the checksum function. Generally the fastest software
2382 checksum that fio supports.
2385 Use sha512 as the checksum function.
2388 Use sha256 as the checksum function.
2391 Use optimized sha1 as the checksum function.
2394 Use optimized sha3-224 as the checksum function.
2397 Use optimized sha3-256 as the checksum function.
2400 Use optimized sha3-384 as the checksum function.
2403 Use optimized sha3-512 as the checksum function.
2406 This option is deprecated, since now meta information is included in
2407 generic verification header and meta verification happens by
2408 default. For detailed information see the description of the
2409 :option:`verify` setting. This option is kept because of
2410 compatibility's sake with old configurations. Do not use it.
2413 Verify a strict pattern. Normally fio includes a header with some
2414 basic information and checksumming, but if this option is set, only
2415 the specific pattern set with :option:`verify_pattern` is verified.
2418 Only pretend to verify. Useful for testing internals with
2419 :option:`ioengine`\=null, not for much else.
2421 This option can be used for repeated burn-in tests of a system to make sure
2422 that the written data is also correctly read back. If the data direction
2423 given is a read or random read, fio will assume that it should verify a
2424 previously written file. If the data direction includes any form of write,
2425 the verify will be of the newly written data.
2427 .. option:: verifysort=bool
2429 If true, fio will sort written verify blocks when it deems it faster to read
2430 them back in a sorted manner. This is often the case when overwriting an
2431 existing file, since the blocks are already laid out in the file system. You
2432 can ignore this option unless doing huge amounts of really fast I/O where
2433 the red-black tree sorting CPU time becomes significant. Default: true.
2435 .. option:: verifysort_nr=int
2437 Pre-load and sort verify blocks for a read workload.
2439 .. option:: verify_offset=int
2441 Swap the verification header with data somewhere else in the block before
2442 writing. It is swapped back before verifying.
2444 .. option:: verify_interval=int
2446 Write the verification header at a finer granularity than the
2447 :option:`blocksize`. It will be written for chunks the size of
2448 ``verify_interval``. :option:`blocksize` should divide this evenly.
2450 .. option:: verify_pattern=str
2452 If set, fio will fill the I/O buffers with this pattern. Fio defaults to
2453 filling with totally random bytes, but sometimes it's interesting to fill
2454 with a known pattern for I/O verification purposes. Depending on the width
2455 of the pattern, fio will fill 1/2/3/4 bytes of the buffer at the time(it can
2456 be either a decimal or a hex number). The ``verify_pattern`` if larger than
2457 a 32-bit quantity has to be a hex number that starts with either "0x" or
2458 "0X". Use with :option:`verify`. Also, ``verify_pattern`` supports %o
2459 format, which means that for each block offset will be written and then
2460 verified back, e.g.::
2464 Or use combination of everything::
2466 verify_pattern=0xff%o"abcd"-12
2468 .. option:: verify_fatal=bool
2470 Normally fio will keep checking the entire contents before quitting on a
2471 block verification failure. If this option is set, fio will exit the job on
2472 the first observed failure. Default: false.
2474 .. option:: verify_dump=bool
2476 If set, dump the contents of both the original data block and the data block
2477 we read off disk to files. This allows later analysis to inspect just what
2478 kind of data corruption occurred. Off by default.
2480 .. option:: verify_async=int
2482 Fio will normally verify I/O inline from the submitting thread. This option
2483 takes an integer describing how many async offload threads to create for I/O
2484 verification instead, causing fio to offload the duty of verifying I/O
2485 contents to one or more separate threads. If using this offload option, even
2486 sync I/O engines can benefit from using an :option:`iodepth` setting higher
2487 than 1, as it allows them to have I/O in flight while verifies are running.
2488 Defaults to 0 async threads, i.e. verification is not asynchronous.
2490 .. option:: verify_async_cpus=str
2492 Tell fio to set the given CPU affinity on the async I/O verification
2493 threads. See :option:`cpus_allowed` for the format used.
2495 .. option:: verify_backlog=int
2497 Fio will normally verify the written contents of a job that utilizes verify
2498 once that job has completed. In other words, everything is written then
2499 everything is read back and verified. You may want to verify continually
2500 instead for a variety of reasons. Fio stores the meta data associated with
2501 an I/O block in memory, so for large verify workloads, quite a bit of memory
2502 would be used up holding this meta data. If this option is enabled, fio will
2503 write only N blocks before verifying these blocks.
2505 .. option:: verify_backlog_batch=int
2507 Control how many blocks fio will verify if :option:`verify_backlog` is
2508 set. If not set, will default to the value of :option:`verify_backlog`
2509 (meaning the entire queue is read back and verified). If
2510 ``verify_backlog_batch`` is less than :option:`verify_backlog` then not all
2511 blocks will be verified, if ``verify_backlog_batch`` is larger than
2512 :option:`verify_backlog`, some blocks will be verified more than once.
2514 .. option:: verify_state_save=bool
2516 When a job exits during the write phase of a verify workload, save its
2517 current state. This allows fio to replay up until that point, if the verify
2518 state is loaded for the verify read phase. The format of the filename is,
2521 <type>-<jobname>-<jobindex>-verify.state.
2523 <type> is "local" for a local run, "sock" for a client/server socket
2524 connection, and "ip" (192.168.0.1, for instance) for a networked
2525 client/server connection. Defaults to true.
2527 .. option:: verify_state_load=bool
2529 If a verify termination trigger was used, fio stores the current write state
2530 of each thread. This can be used at verification time so that fio knows how
2531 far it should verify. Without this information, fio will run a full
2532 verification pass, according to the settings in the job file used. Default
2535 .. option:: trim_percentage=int
2537 Number of verify blocks to discard/trim.
2539 .. option:: trim_verify_zero=bool
2541 Verify that trim/discarded blocks are returned as zeroes.
2543 .. option:: trim_backlog=int
2545 Verify that trim/discarded blocks are returned as zeroes.
2547 .. option:: trim_backlog_batch=int
2549 Trim this number of I/O blocks.
2551 .. option:: experimental_verify=bool
2553 Enable experimental verification.
2559 .. option:: steadystate=str:float, ss=str:float
2561 Define the criterion and limit for assessing steady state performance. The
2562 first parameter designates the criterion whereas the second parameter sets
2563 the threshold. When the criterion falls below the threshold for the
2564 specified duration, the job will stop. For example, `iops_slope:0.1%` will
2565 direct fio to terminate the job when the least squares regression slope
2566 falls below 0.1% of the mean IOPS. If :option:`group_reporting` is enabled
2567 this will apply to all jobs in the group. Below is the list of available
2568 steady state assessment criteria. All assessments are carried out using only
2569 data from the rolling collection window. Threshold limits can be expressed
2570 as a fixed value or as a percentage of the mean in the collection window.
2573 Collect IOPS data. Stop the job if all individual IOPS measurements
2574 are within the specified limit of the mean IOPS (e.g., ``iops:2``
2575 means that all individual IOPS values must be within 2 of the mean,
2576 whereas ``iops:0.2%`` means that all individual IOPS values must be
2577 within 0.2% of the mean IOPS to terminate the job).
2580 Collect IOPS data and calculate the least squares regression
2581 slope. Stop the job if the slope falls below the specified limit.
2584 Collect bandwidth data. Stop the job if all individual bandwidth
2585 measurements are within the specified limit of the mean bandwidth.
2588 Collect bandwidth data and calculate the least squares regression
2589 slope. Stop the job if the slope falls below the specified limit.
2591 .. option:: steadystate_duration=time, ss_dur=time
2593 A rolling window of this duration will be used to judge whether steady state
2594 has been reached. Data will be collected once per second. The default is 0
2595 which disables steady state detection. When the unit is omitted, the
2596 value is interpreted in seconds.
2598 .. option:: steadystate_ramp_time=time, ss_ramp=time
2600 Allow the job to run for the specified duration before beginning data
2601 collection for checking the steady state job termination criterion. The
2602 default is 0. When the unit is omitted, the value is interpreted in seconds.
2605 Measurements and reporting
2606 ~~~~~~~~~~~~~~~~~~~~~~~~~~
2608 .. option:: per_job_logs=bool
2610 If set, this generates bw/clat/iops log with per file private filenames. If
2611 not set, jobs with identical names will share the log filename. Default:
2614 .. option:: group_reporting
2616 It may sometimes be interesting to display statistics for groups of jobs as
2617 a whole instead of for each individual job. This is especially true if
2618 :option:`numjobs` is used; looking at individual thread/process output
2619 quickly becomes unwieldy. To see the final report per-group instead of
2620 per-job, use :option:`group_reporting`. Jobs in a file will be part of the
2621 same reporting group, unless if separated by a :option:`stonewall`, or by
2622 using :option:`new_group`.
2624 .. option:: new_group
2626 Start a new reporting group. See: :option:`group_reporting`. If not given,
2627 all jobs in a file will be part of the same reporting group, unless
2628 separated by a :option:`stonewall`.
2632 By default, fio collects and shows final output results for all jobs
2633 that run. If this option is set to 0, then fio will ignore it in
2634 the final stat output.
2636 .. option:: write_bw_log=str
2638 If given, write a bandwidth log for this job. Can be used to store data of
2639 the bandwidth of the jobs in their lifetime. The included
2640 :command:`fio_generate_plots` script uses :command:`gnuplot` to turn these
2641 text files into nice graphs. See :option:`write_lat_log` for behaviour of
2642 given filename. For this option, the postfix is :file:`_bw.x.log`, where `x`
2643 is the index of the job (`1..N`, where `N` is the number of jobs). If
2644 :option:`per_job_logs` is false, then the filename will not include the job
2645 index. See `Log File Formats`_.
2647 .. option:: write_lat_log=str
2649 Same as :option:`write_bw_log`, except that this option stores I/O
2650 submission, completion, and total latencies instead. If no filename is given
2651 with this option, the default filename of :file:`jobname_type.log` is
2652 used. Even if the filename is given, fio will still append the type of
2653 log. So if one specifies::
2657 The actual log names will be :file:`foo_slat.x.log`, :file:`foo_clat.x.log`,
2658 and :file:`foo_lat.x.log`, where `x` is the index of the job (1..N, where N
2659 is the number of jobs). This helps :command:`fio_generate_plot` find the
2660 logs automatically. If :option:`per_job_logs` is false, then the filename
2661 will not include the job index. See `Log File Formats`_.
2663 .. option:: write_hist_log=str
2665 Same as :option:`write_lat_log`, but writes I/O completion latency
2666 histograms. If no filename is given with this option, the default filename
2667 of :file:`jobname_clat_hist.x.log` is used, where `x` is the index of the
2668 job (1..N, where `N` is the number of jobs). Even if the filename is given,
2669 fio will still append the type of log. If :option:`per_job_logs` is false,
2670 then the filename will not include the job index. See `Log File Formats`_.
2672 .. option:: write_iops_log=str
2674 Same as :option:`write_bw_log`, but writes IOPS. If no filename is given
2675 with this option, the default filename of :file:`jobname_type.x.log` is
2676 used,where `x` is the index of the job (1..N, where `N` is the number of
2677 jobs). Even if the filename is given, fio will still append the type of
2678 log. If :option:`per_job_logs` is false, then the filename will not include
2679 the job index. See `Log File Formats`_.
2681 .. option:: log_avg_msec=int
2683 By default, fio will log an entry in the iops, latency, or bw log for every
2684 I/O that completes. When writing to the disk log, that can quickly grow to a
2685 very large size. Setting this option makes fio average the each log entry
2686 over the specified period of time, reducing the resolution of the log. See
2687 :option:`log_max_value` as well. Defaults to 0, logging all entries.
2689 .. option:: log_hist_msec=int
2691 Same as :option:`log_avg_msec`, but logs entries for completion latency
2692 histograms. Computing latency percentiles from averages of intervals using
2693 :option:`log_avg_msec` is inaccurate. Setting this option makes fio log
2694 histogram entries over the specified period of time, reducing log sizes for
2695 high IOPS devices while retaining percentile accuracy. See
2696 :option:`log_hist_coarseness` as well. Defaults to 0, meaning histogram
2697 logging is disabled.
2699 .. option:: log_hist_coarseness=int
2701 Integer ranging from 0 to 6, defining the coarseness of the resolution of
2702 the histogram logs enabled with :option:`log_hist_msec`. For each increment
2703 in coarseness, fio outputs half as many bins. Defaults to 0, for which
2704 histogram logs contain 1216 latency bins. See `Log File Formats`_.
2706 .. option:: log_max_value=bool
2708 If :option:`log_avg_msec` is set, fio logs the average over that window. If
2709 you instead want to log the maximum value, set this option to 1. Defaults to
2710 0, meaning that averaged values are logged.
2712 .. option:: log_offset=int
2714 If this is set, the iolog options will include the byte offset for the I/O
2715 entry as well as the other data values.
2717 .. option:: log_compression=int
2719 If this is set, fio will compress the I/O logs as it goes, to keep the
2720 memory footprint lower. When a log reaches the specified size, that chunk is
2721 removed and compressed in the background. Given that I/O logs are fairly
2722 highly compressible, this yields a nice memory savings for longer runs. The
2723 downside is that the compression will consume some background CPU cycles, so
2724 it may impact the run. This, however, is also true if the logging ends up
2725 consuming most of the system memory. So pick your poison. The I/O logs are
2726 saved normally at the end of a run, by decompressing the chunks and storing
2727 them in the specified log file. This feature depends on the availability of
2730 .. option:: log_compression_cpus=str
2732 Define the set of CPUs that are allowed to handle online log compression for
2733 the I/O jobs. This can provide better isolation between performance
2734 sensitive jobs, and background compression work.
2736 .. option:: log_store_compressed=bool
2738 If set, fio will store the log files in a compressed format. They can be
2739 decompressed with fio, using the :option:`--inflate-log` command line
2740 parameter. The files will be stored with a :file:`.fz` suffix.
2742 .. option:: log_unix_epoch=bool
2744 If set, fio will log Unix timestamps to the log files produced by enabling
2745 write_type_log for each log type, instead of the default zero-based
2748 .. option:: block_error_percentiles=bool
2750 If set, record errors in trim block-sized units from writes and trims and
2751 output a histogram of how many trims it took to get to errors, and what kind
2752 of error was encountered.
2754 .. option:: bwavgtime=int
2756 Average the calculated bandwidth over the given time. Value is specified in
2757 milliseconds. If the job also does bandwidth logging through
2758 :option:`write_bw_log`, then the minimum of this option and
2759 :option:`log_avg_msec` will be used. Default: 500ms.
2761 .. option:: iopsavgtime=int
2763 Average the calculated IOPS over the given time. Value is specified in
2764 milliseconds. If the job also does IOPS logging through
2765 :option:`write_iops_log`, then the minimum of this option and
2766 :option:`log_avg_msec` will be used. Default: 500ms.
2768 .. option:: disk_util=bool
2770 Generate disk utilization statistics, if the platform supports it.
2773 .. option:: disable_lat=bool
2775 Disable measurements of total latency numbers. Useful only for cutting back
2776 the number of calls to :manpage:`gettimeofday(2)`, as that does impact
2777 performance at really high IOPS rates. Note that to really get rid of a
2778 large amount of these calls, this option must be used with
2779 :option:`disable_slat` and :option:`disable_bw_measurement` as well.
2781 .. option:: disable_clat=bool
2783 Disable measurements of completion latency numbers. See
2784 :option:`disable_lat`.
2786 .. option:: disable_slat=bool
2788 Disable measurements of submission latency numbers. See
2789 :option:`disable_slat`.
2791 .. option:: disable_bw_measurement=bool, disable_bw=bool
2793 Disable measurements of throughput/bandwidth numbers. See
2794 :option:`disable_lat`.
2796 .. option:: clat_percentiles=bool
2798 Enable the reporting of percentiles of completion latencies.
2800 .. option:: percentile_list=float_list
2802 Overwrite the default list of percentiles for completion latencies and the
2803 block error histogram. Each number is a floating number in the range
2804 (0,100], and the maximum length of the list is 20. Use ``:`` to separate the
2805 numbers, and list the numbers in ascending order. For example,
2806 ``--percentile_list=99.5:99.9`` will cause fio to report the values of
2807 completion latency below which 99.5% and 99.9% of the observed latencies
2814 .. option:: exitall_on_error
2816 When one job finishes in error, terminate the rest. The default is to wait
2817 for each job to finish.
2819 .. option:: continue_on_error=str
2821 Normally fio will exit the job on the first observed failure. If this option
2822 is set, fio will continue the job when there is a 'non-fatal error' (EIO or
2823 EILSEQ) until the runtime is exceeded or the I/O size specified is
2824 completed. If this option is used, there are two more stats that are
2825 appended, the total error count and the first error. The error field given
2826 in the stats is the first error that was hit during the run.
2828 The allowed values are:
2831 Exit on any I/O or verify errors.
2834 Continue on read errors, exit on all others.
2837 Continue on write errors, exit on all others.
2840 Continue on any I/O error, exit on all others.
2843 Continue on verify errors, exit on all others.
2846 Continue on all errors.
2849 Backward-compatible alias for 'none'.
2852 Backward-compatible alias for 'all'.
2854 .. option:: ignore_error=str
2856 Sometimes you want to ignore some errors during test in that case you can
2857 specify error list for each error type, instead of only being able to
2858 ignore the default 'non-fatal error' using :option:`continue_on_error`.
2859 ``ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST`` errors for
2860 given error type is separated with ':'. Error may be symbol ('ENOSPC',
2861 'ENOMEM') or integer. Example::
2863 ignore_error=EAGAIN,ENOSPC:122
2865 This option will ignore EAGAIN from READ, and ENOSPC and 122(EDQUOT) from
2866 WRITE. This option works by overriding :option:`continue_on_error` with
2867 the list of errors for each error type if any.
2869 .. option:: error_dump=bool
2871 If set dump every error even if it is non fatal, true by default. If
2872 disabled only fatal error will be dumped.
2874 Running predefined workloads
2875 ----------------------------
2877 Fio includes predefined profiles that mimic the I/O workloads generated by
2880 .. option:: profile=str
2882 The predefined workload to run. Current profiles are:
2885 Threaded I/O bench (tiotest/tiobench) like workload.
2888 Aerospike Certification Tool (ACT) like workload.
2890 To view a profile's additional options use :option:`--cmdhelp` after specifying
2891 the profile. For example::
2893 $ fio --profile=act --cmdhelp
2898 .. option:: device-names=str
2903 .. option:: load=int
2906 ACT load multiplier. Default: 1.
2908 .. option:: test-duration=time
2911 How long the entire test takes to run. When the unit is omitted, the value
2912 is given in seconds. Default: 24h.
2914 .. option:: threads-per-queue=int
2917 Number of read IO threads per device. Default: 8.
2919 .. option:: read-req-num-512-blocks=int
2922 Number of 512B blocks to read at the time. Default: 3.
2924 .. option:: large-block-op-kbytes=int
2927 Size of large block ops in KiB (writes). Default: 131072.
2932 Set to run ACT prep phase.
2934 Tiobench profile options
2935 ~~~~~~~~~~~~~~~~~~~~~~~~
2937 .. option:: size=str
2942 .. option:: block=int
2945 Block size in bytes. Default: 4096.
2947 .. option:: numruns=int
2957 .. option:: threads=int
2962 Interpreting the output
2963 -----------------------
2965 Fio spits out a lot of output. While running, fio will display the status of the
2966 jobs created. An example of that would be::
2968 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]
2970 The characters inside the square brackets denote the current status of each
2971 thread. The possible values (in typical life cycle order) are:
2973 +------+-----+-----------------------------------------------------------+
2975 +======+=====+===========================================================+
2976 | P | | Thread setup, but not started. |
2977 +------+-----+-----------------------------------------------------------+
2978 | C | | Thread created. |
2979 +------+-----+-----------------------------------------------------------+
2980 | I | | Thread initialized, waiting or generating necessary data. |
2981 +------+-----+-----------------------------------------------------------+
2982 | | p | Thread running pre-reading file(s). |
2983 +------+-----+-----------------------------------------------------------+
2984 | | R | Running, doing sequential reads. |
2985 +------+-----+-----------------------------------------------------------+
2986 | | r | Running, doing random reads. |
2987 +------+-----+-----------------------------------------------------------+
2988 | | W | Running, doing sequential writes. |
2989 +------+-----+-----------------------------------------------------------+
2990 | | w | Running, doing random writes. |
2991 +------+-----+-----------------------------------------------------------+
2992 | | M | Running, doing mixed sequential reads/writes. |
2993 +------+-----+-----------------------------------------------------------+
2994 | | m | Running, doing mixed random reads/writes. |
2995 +------+-----+-----------------------------------------------------------+
2996 | | F | Running, currently waiting for :manpage:`fsync(2)` |
2997 +------+-----+-----------------------------------------------------------+
2998 | | V | Running, doing verification of written data. |
2999 +------+-----+-----------------------------------------------------------+
3000 | E | | Thread exited, not reaped by main thread yet. |
3001 +------+-----+-----------------------------------------------------------+
3002 | _ | | Thread reaped, or |
3003 +------+-----+-----------------------------------------------------------+
3004 | X | | Thread reaped, exited with an error. |
3005 +------+-----+-----------------------------------------------------------+
3006 | K | | Thread reaped, exited due to signal. |
3007 +------+-----+-----------------------------------------------------------+
3009 Fio will condense the thread string as not to take up more space on the command
3010 line as is needed. For instance, if you have 10 readers and 10 writers running,
3011 the output would look like this::
3013 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]
3015 Fio will still maintain the ordering, though. So the above means that jobs 1..10
3016 are readers, and 11..20 are writers.
3018 The other values are fairly self explanatory -- number of threads currently
3019 running and doing I/O, the number of currently open files (f=), the rate of I/O
3020 since last check (read speed listed first, then write speed and optionally trim
3021 speed), and the estimated completion percentage and time for the current
3022 running group. It's impossible to estimate runtime of the following groups (if
3023 any). Note that the string is displayed in order, so it's possible to tell which
3024 of the jobs are currently doing what. The first character is the first job
3025 defined in the job file, and so forth.
3027 When fio is done (or interrupted by :kbd:`ctrl-c`), it will show the data for
3028 each thread, group of threads, and disks in that order. For each data direction,
3029 the output looks like::
3031 Client1 (g=0): err= 0:
3032 write: io= 32MiB, bw= 666KiB/s, iops=89 , runt= 50320msec
3033 slat (msec): min= 0, max= 136, avg= 0.03, stdev= 1.92
3034 clat (msec): min= 0, max= 631, avg=48.50, stdev=86.82
3035 bw (KiB/s) : min= 0, max= 1196, per=51.00%, avg=664.02, stdev=681.68
3036 cpu : usr=1.49%, sys=0.25%, ctx=7969, majf=0, minf=17
3037 IO depths : 1=0.1%, 2=0.3%, 4=0.5%, 8=99.0%, 16=0.0%, 32=0.0%, >32=0.0%
3038 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
3039 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
3040 issued r/w: total=0/32768, short=0/0
3041 lat (msec): 2=1.6%, 4=0.0%, 10=3.2%, 20=12.8%, 50=38.4%, 100=24.8%,
3042 lat (msec): 250=15.2%, 500=0.0%, 750=0.0%, 1000=0.0%, >=2048=0.0%
3044 The client number is printed, along with the group id and error of that
3045 thread. Below is the I/O statistics, here for writes. In the order listed, they
3049 Number of megabytes I/O performed.
3052 Average bandwidth rate.
3055 Average I/Os performed per second.
3058 The runtime of that thread.
3061 Submission latency (avg being the average, stdev being the standard
3062 deviation). This is the time it took to submit the I/O. For sync I/O,
3063 the slat is really the completion latency, since queue/complete is one
3064 operation there. This value can be in milliseconds or microseconds, fio
3065 will choose the most appropriate base and print that. In the example
3066 above, milliseconds is the best scale. Note: in :option:`--minimal` mode
3067 latencies are always expressed in microseconds.
3070 Completion latency. Same names as slat, this denotes the time from
3071 submission to completion of the I/O pieces. For sync I/O, clat will
3072 usually be equal (or very close) to 0, as the time from submit to
3073 complete is basically just CPU time (I/O has already been done, see slat
3077 Bandwidth. Same names as the xlat stats, but also includes an
3078 approximate percentage of total aggregate bandwidth this thread received
3079 in this group. This last value is only really useful if the threads in
3080 this group are on the same disk, since they are then competing for disk
3084 CPU usage. User and system time, along with the number of context
3085 switches this thread went through, usage of system and user time, and
3086 finally the number of major and minor page faults. The CPU utilization
3087 numbers are averages for the jobs in that reporting group, while the
3088 context and fault counters are summed.
3091 The distribution of I/O depths over the job life time. The numbers are
3092 divided into powers of 2, so for example the 16= entries includes depths
3093 up to that value but higher than the previous entry. In other words, it
3094 covers the range from 16 to 31.
3097 How many pieces of I/O were submitting in a single submit call. Each
3098 entry denotes that amount and below, until the previous entry -- e.g.,
3099 8=100% mean that we submitted anywhere in between 5-8 I/Os per submit
3103 Like the above submit number, but for completions instead.
3106 The number of read/write requests issued, and how many of them were
3110 The distribution of I/O completion latencies. This is the time from when
3111 I/O leaves fio and when it gets completed. The numbers follow the same
3112 pattern as the I/O depths, meaning that 2=1.6% means that 1.6% of the
3113 I/O completed within 2 msecs, 20=12.8% means that 12.8% of the I/O took
3114 more than 10 msecs, but less than (or equal to) 20 msecs.
3116 After each client has been listed, the group statistics are printed. They
3117 will look like this::
3119 Run status group 0 (all jobs):
3120 READ: io=64MB, aggrb=22178, minb=11355, maxb=11814, mint=2840msec, maxt=2955msec
3121 WRITE: io=64MB, aggrb=1302, minb=666, maxb=669, mint=50093msec, maxt=50320msec
3123 For each data direction, it prints:
3126 Number of megabytes I/O performed.
3128 Aggregate bandwidth of threads in this group.
3130 The minimum average bandwidth a thread saw.
3132 The maximum average bandwidth a thread saw.
3134 The smallest runtime of the threads in that group.
3136 The longest runtime of the threads in that group.
3138 And finally, the disk statistics are printed. They will look like this::
3140 Disk stats (read/write):
3141 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
3143 Each value is printed for both reads and writes, with reads first. The
3147 Number of I/Os performed by all groups.
3149 Number of merges I/O the I/O scheduler.
3151 Number of ticks we kept the disk busy.
3153 Total time spent in the disk queue.
3155 The disk utilization. A value of 100% means we kept the disk
3156 busy constantly, 50% would be a disk idling half of the time.
3158 It is also possible to get fio to dump the current output while it is running,
3159 without terminating the job. To do that, send fio the **USR1** signal. You can
3160 also get regularly timed dumps by using the :option:`--status-interval`
3161 parameter, or by creating a file in :file:`/tmp` named
3162 :file:`fio-dump-status`. If fio sees this file, it will unlink it and dump the
3163 current output status.
3169 For scripted usage where you typically want to generate tables or graphs of the
3170 results, fio can output the results in a semicolon separated format. The format
3171 is one long line of values, such as::
3173 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%
3174 A description of this job goes here.
3176 The job description (if provided) follows on a second line.
3178 To enable terse output, use the :option:`--minimal` command line option. The
3179 first value is the version of the terse output format. If the output has to be
3180 changed for some reason, this number will be incremented by 1 to signify that
3183 Split up, the format is as follows (comments in brackets denote when a
3184 field was introduced or whether its specific to some terse version):
3188 terse version, fio version [v3], jobname, groupid, error
3192 Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec)
3193 Submission latency: min, max, mean, stdev (usec)
3194 Completion latency: min, max, mean, stdev (usec)
3195 Completion latency percentiles: 20 fields (see below)
3196 Total latency: min, max, mean, stdev (usec)
3197 Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev, number of samples [v5]
3198 IOPS [v5]: min, max, mean, stdev, number of samples
3204 Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec)
3205 Submission latency: min, max, mean, stdev (usec)
3206 Completion latency: min, max, mean, stdev (usec)
3207 Completion latency percentiles: 20 fields (see below)
3208 Total latency: min, max, mean, stdev (usec)
3209 Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev, number of samples [v5]
3210 IOPS [v5]: min, max, mean, stdev, number of samples
3212 TRIM status [all but version 3]:
3214 Fields are similar to READ/WRITE status.
3218 user, system, context switches, major faults, minor faults
3222 <=1, 2, 4, 8, 16, 32, >=64
3224 I/O latencies microseconds::
3226 <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
3228 I/O latencies milliseconds::
3230 <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
3232 Disk utilization [v3]::
3234 Disk name, Read ios, write ios,
3235 Read merges, write merges,
3236 Read ticks, write ticks,
3237 Time spent in queue, disk utilization percentage
3239 Additional Info (dependent on continue_on_error, default off)::
3241 total # errors, first error code
3243 Additional Info (dependent on description being set)::
3247 Completion latency percentiles can be a grouping of up to 20 sets, so for the
3248 terse output fio writes all of them. Each field will look like this::
3252 which is the Xth percentile, and the `usec` latency associated with it.
3254 For disk utilization, all disks used by fio are shown. So for each disk there
3255 will be a disk utilization section.
3257 Below is a single line containing short names for each of the fields in the
3258 minimal output v3, separated by semicolons:
3260 terse_version_3;fio_version;jobname;groupid;error;read_kb;read_bandwidth;read_iops;read_runtime_ms;read_slat_min;read_slat_max;read_slat_mean;read_slat_dev;read_clat_max;read_clat_min;read_clat_mean;read_clat_dev;read_clat_pct01;read_clat_pct02;read_clat_pct03;read_clat_pct04;read_clat_pct05;read_clat_pct06;read_clat_pct07;read_clat_pct08;read_clat_pct09;read_clat_pct10;read_clat_pct11;read_clat_pct12;read_clat_pct13;read_clat_pct14;read_clat_pct15;read_clat_pct16;read_clat_pct17;read_clat_pct18;read_clat_pct19;read_clat_pct20;read_tlat_min;read_lat_max;read_lat_mean;read_lat_dev;read_bw_min;read_bw_max;read_bw_agg_pct;read_bw_mean;read_bw_dev;write_kb;write_bandwidth;write_iops;write_runtime_ms;write_slat_min;write_slat_max;write_slat_mean;write_slat_dev;write_clat_max;write_clat_min;write_clat_mean;write_clat_dev;write_clat_pct01;write_clat_pct02;write_clat_pct03;write_clat_pct04;write_clat_pct05;write_clat_pct06;write_clat_pct07;write_clat_pct08;write_clat_pct09;write_clat_pct10;write_clat_pct11;write_clat_pct12;write_clat_pct13;write_clat_pct14;write_clat_pct15;write_clat_pct16;write_clat_pct17;write_clat_pct18;write_clat_pct19;write_clat_pct20;write_tlat_min;write_lat_max;write_lat_mean;write_lat_dev;write_bw_min;write_bw_max;write_bw_agg_pct;write_bw_mean;write_bw_dev;cpu_user;cpu_sys;cpu_csw;cpu_mjf;pu_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
3266 There are two trace file format that you can encounter. The older (v1) format is
3267 unsupported since version 1.20-rc3 (March 2008). It will still be described
3268 below in case that you get an old trace and want to understand it.
3270 In any case the trace is a simple text file with a single action per line.
3273 Trace file format v1
3274 ~~~~~~~~~~~~~~~~~~~~
3276 Each line represents a single I/O action in the following format::
3280 where `rw=0/1` for read/write, and the offset and length entries being in bytes.
3282 This format is not supported in fio versions => 1.20-rc3.
3285 Trace file format v2
3286 ~~~~~~~~~~~~~~~~~~~~
3288 The second version of the trace file format was added in fio version 1.17. It
3289 allows to access more then one file per trace and has a bigger set of possible
3292 The first line of the trace file has to be::
3296 Following this can be lines in two different formats, which are described below.
3298 The file management format::
3302 The filename is given as an absolute path. The action can be one of these:
3305 Add the given filename to the trace.
3307 Open the file with the given filename. The filename has to have
3308 been added with the **add** action before.
3310 Close the file with the given filename. The file has to have been
3314 The file I/O action format::
3316 filename action offset length
3318 The `filename` is given as an absolute path, and has to have been added and
3319 opened before it can be used with this format. The `offset` and `length` are
3320 given in bytes. The `action` can be one of these:
3323 Wait for `offset` microseconds. Everything below 100 is discarded.
3324 The time is relative to the previous `wait` statement.
3326 Read `length` bytes beginning from `offset`.
3328 Write `length` bytes beginning from `offset`.
3330 :manpage:`fsync(2)` the file.
3332 :manpage:`fdatasync(2)` the file.
3334 Trim the given file from the given `offset` for `length` bytes.
3336 CPU idleness profiling
3337 ----------------------
3339 In some cases, we want to understand CPU overhead in a test. For example, we
3340 test patches for the specific goodness of whether they reduce CPU usage.
3341 Fio implements a balloon approach to create a thread per CPU that runs at idle
3342 priority, meaning that it only runs when nobody else needs the cpu.
3343 By measuring the amount of work completed by the thread, idleness of each CPU
3344 can be derived accordingly.
3346 An unit work is defined as touching a full page of unsigned characters. Mean and
3347 standard deviation of time to complete an unit work is reported in "unit work"
3348 section. Options can be chosen to report detailed percpu idleness or overall
3349 system idleness by aggregating percpu stats.
3352 Verification and triggers
3353 -------------------------
3355 Fio is usually run in one of two ways, when data verification is done. The first
3356 is a normal write job of some sort with verify enabled. When the write phase has
3357 completed, fio switches to reads and verifies everything it wrote. The second
3358 model is running just the write phase, and then later on running the same job
3359 (but with reads instead of writes) to repeat the same I/O patterns and verify
3360 the contents. Both of these methods depend on the write phase being completed,
3361 as fio otherwise has no idea how much data was written.
3363 With verification triggers, fio supports dumping the current write state to
3364 local files. Then a subsequent read verify workload can load this state and know
3365 exactly where to stop. This is useful for testing cases where power is cut to a
3366 server in a managed fashion, for instance.
3368 A verification trigger consists of two things:
3370 1) Storing the write state of each job.
3371 2) Executing a trigger command.
3373 The write state is relatively small, on the order of hundreds of bytes to single
3374 kilobytes. It contains information on the number of completions done, the last X
3377 A trigger is invoked either through creation ('touch') of a specified file in
3378 the system, or through a timeout setting. If fio is run with
3379 :option:`--trigger-file`\= :file:`/tmp/trigger-file`, then it will continually
3380 check for the existence of :file:`/tmp/trigger-file`. When it sees this file, it
3381 will fire off the trigger (thus saving state, and executing the trigger
3384 For client/server runs, there's both a local and remote trigger. If fio is
3385 running as a server backend, it will send the job states back to the client for
3386 safe storage, then execute the remote trigger, if specified. If a local trigger
3387 is specified, the server will still send back the write state, but the client
3388 will then execute the trigger.
3390 Verification trigger example
3391 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3393 Let's say we want to run a powercut test on the remote machine 'server'. Our
3394 write workload is in :file:`write-test.fio`. We want to cut power to 'server' at
3395 some point during the run, and we'll run this test from the safety or our local
3396 machine, 'localbox'. On the server, we'll start the fio backend normally::
3398 server# fio --server
3400 and on the client, we'll fire off the workload::
3402 localbox$ fio --client=server --trigger-file=/tmp/my-trigger --trigger-remote="bash -c \"echo b > /proc/sysrq-triger\""
3404 We set :file:`/tmp/my-trigger` as the trigger file, and we tell fio to execute::
3406 echo b > /proc/sysrq-trigger
3408 on the server once it has received the trigger and sent us the write state. This
3409 will work, but it's not **really** cutting power to the server, it's merely
3410 abruptly rebooting it. If we have a remote way of cutting power to the server
3411 through IPMI or similar, we could do that through a local trigger command
3412 instead. Let's assume we have a script that does IPMI reboot of a given hostname,
3413 ipmi-reboot. On localbox, we could then have run fio with a local trigger
3416 localbox$ fio --client=server --trigger-file=/tmp/my-trigger --trigger="ipmi-reboot server"
3418 For this case, fio would wait for the server to send us the write state, then
3419 execute ``ipmi-reboot server`` when that happened.
3421 Loading verify state
3422 ~~~~~~~~~~~~~~~~~~~~
3424 To load stored write state, a read verification job file must contain the
3425 :option:`verify_state_load` option. If that is set, fio will load the previously
3426 stored state. For a local fio run this is done by loading the files directly,
3427 and on a client/server run, the server backend will ask the client to send the
3428 files over and load them from there.
3434 Fio supports a variety of log file formats, for logging latencies, bandwidth,
3435 and IOPS. The logs share a common format, which looks like this:
3437 *time* (`msec`), *value*, *data direction*, *offset*
3439 Time for the log entry is always in milliseconds. The *value* logged depends
3440 on the type of log, it will be one of the following:
3443 Value is latency in usecs
3449 *Data direction* is one of the following:
3458 The *offset* is the offset, in bytes, from the start of the file, for that
3459 particular I/O. The logging of the offset can be toggled with
3460 :option:`log_offset`.
3462 If windowed logging is enabled through :option:`log_avg_msec` then fio doesn't
3463 log individual I/Os. Instead of logs the average values over the specified period
3464 of time. Since 'data direction' and 'offset' are per-I/O values, they aren't
3465 applicable if windowed logging is enabled. If windowed logging is enabled and
3466 :option:`log_max_value` is set, then fio logs maximum values in that window
3467 instead of averages.
3473 Normally fio is invoked as a stand-alone application on the machine where the
3474 I/O workload should be generated. However, the frontend and backend of fio can
3475 be run separately. Ie the fio server can generate an I/O workload on the "Device
3476 Under Test" while being controlled from another machine.
3478 Start the server on the machine which has access to the storage DUT::
3482 where args defines what fio listens to. The arguments are of the form
3483 ``type,hostname`` or ``IP,port``. *type* is either ``ip`` (or ip4) for TCP/IP
3484 v4, ``ip6`` for TCP/IP v6, or ``sock`` for a local unix domain socket.
3485 *hostname* is either a hostname or IP address, and *port* is the port to listen
3486 to (only valid for TCP/IP, not a local socket). Some examples:
3490 Start a fio server, listening on all interfaces on the default port (8765).
3492 2) ``fio --server=ip:hostname,4444``
3494 Start a fio server, listening on IP belonging to hostname and on port 4444.
3496 3) ``fio --server=ip6:::1,4444``
3498 Start a fio server, listening on IPv6 localhost ::1 and on port 4444.
3500 4) ``fio --server=,4444``
3502 Start a fio server, listening on all interfaces on port 4444.
3504 5) ``fio --server=1.2.3.4``
3506 Start a fio server, listening on IP 1.2.3.4 on the default port.
3508 6) ``fio --server=sock:/tmp/fio.sock``
3510 Start a fio server, listening on the local socket /tmp/fio.sock.
3512 Once a server is running, a "client" can connect to the fio server with::
3514 fio <local-args> --client=<server> <remote-args> <job file(s)>
3516 where `local-args` are arguments for the client where it is running, `server`
3517 is the connect string, and `remote-args` and `job file(s)` are sent to the
3518 server. The `server` string follows the same format as it does on the server
3519 side, to allow IP/hostname/socket and port strings.
3521 Fio can connect to multiple servers this way::
3523 fio --client=<server1> <job file(s)> --client=<server2> <job file(s)>
3525 If the job file is located on the fio server, then you can tell the server to
3526 load a local file as well. This is done by using :option:`--remote-config` ::
3528 fio --client=server --remote-config /path/to/file.fio
3530 Then fio will open this local (to the server) job file instead of being passed
3531 one from the client.
3533 If you have many servers (example: 100 VMs/containers), you can input a pathname
3534 of a file containing host IPs/names as the parameter value for the
3535 :option:`--client` option. For example, here is an example :file:`host.list`
3536 file containing 2 hostnames::
3538 host1.your.dns.domain
3539 host2.your.dns.domain
3541 The fio command would then be::
3543 fio --client=host.list <job file(s)>
3545 In this mode, you cannot input server-specific parameters or job files -- all
3546 servers receive the same job file.
3548 In order to let ``fio --client`` runs use a shared filesystem from multiple
3549 hosts, ``fio --client`` now prepends the IP address of the server to the
3550 filename. For example, if fio is using the directory :file:`/mnt/nfs/fio` and is
3551 writing filename :file:`fileio.tmp`, with a :option:`--client` `hostfile`
3552 containing two hostnames ``h1`` and ``h2`` with IP addresses 192.168.10.120 and
3553 192.168.10.121, then fio will create two files::
3555 /mnt/nfs/fio/192.168.10.120.fileio.tmp
3556 /mnt/nfs/fio/192.168.10.121.fileio.tmp