man: don't use non-breaking minuses when they're not necessary
[fio.git] / fio.1
CommitLineData
523bad63 1.TH fio 1 "August 2017" "User Manual"
d60e92d1
AC
2.SH NAME
3fio \- flexible I/O tester
4.SH SYNOPSIS
5.B fio
6[\fIoptions\fR] [\fIjobfile\fR]...
7.SH DESCRIPTION
8.B fio
9is a tool that will spawn a number of threads or processes doing a
10particular type of I/O action as specified by the user.
11The typical use of fio is to write a job file matching the I/O load
12one wants to simulate.
13.SH OPTIONS
14.TP
49da1240 15.BI \-\-debug \fR=\fPtype
7db7a5a0
TK
16Enable verbose tracing \fItype\fR of various fio actions. May be `all' for all \fItype\fRs
17or individual types separated by a comma (e.g. `\-\-debug=file,mem' will enable
bdd88be3
TK
18file and memory debugging). `help' will list all available tracing options.
19.TP
7db7a5a0 20.BI \-\-parse\-only
bdd88be3 21Parse options only, don't start any I/O.
49da1240 22.TP
b9921d1a
DZ
23.BI \-\-merge\-blktrace\-only
24Merge blktraces only, don't start any I/O.
25.TP
d60e92d1
AC
26.BI \-\-output \fR=\fPfilename
27Write output to \fIfilename\fR.
28.TP
7db7a5a0
TK
29.BI \-\-output\-format \fR=\fPformat
30Set the reporting \fIformat\fR to `normal', `terse', `json', or
31`json+'. Multiple formats can be selected, separate by a comma. `terse'
32is a CSV based format. `json+' is like `json', except it adds a full
513e37ee 33dump of the latency buckets.
e28ee21d 34.TP
7db7a5a0 35.BI \-\-bandwidth\-log
d23ae827 36Generate aggregate bandwidth logs.
d60e92d1 37.TP
7db7a5a0
TK
38.BI \-\-minimal
39Print statistics in a terse, semicolon\-delimited format.
d60e92d1 40.TP
7db7a5a0
TK
41.BI \-\-append\-terse
42Print statistics in selected mode AND terse, semicolon\-delimited format.
43\fBDeprecated\fR, use \fB\-\-output\-format\fR instead to select multiple formats.
f6a7df53 44.TP
065248bf 45.BI \-\-terse\-version \fR=\fPversion
7db7a5a0 46Set terse \fIversion\fR output format (default `3', or `2', `4', `5').
49da1240 47.TP
7db7a5a0 48.BI \-\-version
bdd88be3
TK
49Print version information and exit.
50.TP
7db7a5a0 51.BI \-\-help
bdd88be3 52Print a summary of the command line options and exit.
49da1240 53.TP
7db7a5a0 54.BI \-\-cpuclock\-test
bdd88be3 55Perform test and validation of internal CPU clock.
fec0f21c 56.TP
bdd88be3 57.BI \-\-crctest \fR=\fP[test]
7db7a5a0 58Test the speed of the built\-in checksumming functions. If no argument is given,
bdd88be3 59all of them are tested. Alternatively, a comma separated list can be passed, in which
fec0f21c
JA
60case the given ones are tested.
61.TP
49da1240 62.BI \-\-cmdhelp \fR=\fPcommand
bdd88be3 63Print help information for \fIcommand\fR. May be `all' for all commands.
49da1240 64.TP
7db7a5a0
TK
65.BI \-\-enghelp \fR=\fP[ioengine[,command]]
66List all commands defined by \fIioengine\fR, or print help for \fIcommand\fR
67defined by \fIioengine\fR. If no \fIioengine\fR is given, list all
68available ioengines.
de890a1e 69.TP
d60e92d1 70.BI \-\-showcmd \fR=\fPjobfile
7db7a5a0 71Convert \fIjobfile\fR to a set of command\-line options.
d60e92d1 72.TP
bdd88be3 73.BI \-\-readonly
4027b2a1 74Turn on safety read\-only checks, preventing writes and trims. The \fB\-\-readonly\fR
bdd88be3 75option is an extra safety guard to prevent users from accidentally starting
4027b2a1
VF
76a write or trim workload when that is not desired. Fio will only modify the
77device under test if `rw=write/randwrite/rw/randrw/trim/randtrim/trimwrite'
78is given. This safety net can be used as an extra precaution.
bdd88be3 79.TP
d60e92d1 80.BI \-\-eta \fR=\fPwhen
7db7a5a0 81Specifies when real\-time ETA estimate should be printed. \fIwhen\fR may
db37d890
JA
82be `always', `never' or `auto'. `auto' is the default, it prints ETA when
83requested if the output is a TTY. `always' disregards the output type, and
84prints ETA when requested. `never' never prints ETA.
85.TP
86.BI \-\-eta\-interval \fR=\fPtime
87By default, fio requests client ETA status roughly every second. With this
88option, the interval is configurable. Fio imposes a minimum allowed time to
89avoid flooding the console, less than 250 msec is not supported.
d60e92d1 90.TP
30b5d57f 91.BI \-\-eta\-newline \fR=\fPtime
bdd88be3
TK
92Force a new line for every \fItime\fR period passed. When the unit is omitted,
93the value is interpreted in seconds.
30b5d57f
JA
94.TP
95.BI \-\-status\-interval \fR=\fPtime
aa6cb459
VF
96Force a full status dump of cumulative (from job start) values at \fItime\fR
97intervals. This option does *not* provide per-period measurements. So
98values such as bandwidth are running averages. When the time unit is omitted,
c1f4de8a
JA
99\fItime\fR is interpreted in seconds. Note that using this option with
100`\-\-output-format=json' will yield output that technically isn't valid json,
101since the output will be collated sets of valid json. It will need to be split
102into valid sets of json after the run.
bdd88be3
TK
103.TP
104.BI \-\-section \fR=\fPname
105Only run specified section \fIname\fR in job file. Multiple sections can be specified.
7db7a5a0 106The \fB\-\-section\fR option allows one to combine related jobs into one file.
bdd88be3 107E.g. one job file could define light, moderate, and heavy sections. Tell
7db7a5a0 108fio to run only the "heavy" section by giving `\-\-section=heavy'
bdd88be3 109command line option. One can also specify the "write" operations in one
7db7a5a0 110section and "verify" operation in another section. The \fB\-\-section\fR option
bdd88be3
TK
111only applies to job sections. The reserved *global* section is always
112parsed and used.
c0a5d35e 113.TP
49da1240 114.BI \-\-alloc\-size \fR=\fPkb
4a419903
VF
115Allocate additional internal smalloc pools of size \fIkb\fR in KiB. The
116\fB\-\-alloc\-size\fR option increases shared memory set aside for use by fio.
bdd88be3
TK
117If running large jobs with randommap enabled, fio can run out of memory.
118Smalloc is an internal allocator for shared structures from a fixed size
119memory pool and can grow to 16 pools. The pool size defaults to 16MiB.
7db7a5a0
TK
120NOTE: While running `.fio_smalloc.*' backing store files are visible
121in `/tmp'.
d60e92d1 122.TP
49da1240
JA
123.BI \-\-warnings\-fatal
124All fio parser warnings are fatal, causing fio to exit with an error.
9183788d 125.TP
49da1240 126.BI \-\-max\-jobs \fR=\fPnr
7db7a5a0 127Set the maximum number of threads/processes to support to \fInr\fR.
7f4811bb
RNS
128NOTE: On Linux, it may be necessary to increase the shared-memory limit
129(`/proc/sys/kernel/shmmax') if fio runs into errors while creating jobs.
d60e92d1 130.TP
49da1240 131.BI \-\-server \fR=\fPargs
7db7a5a0
TK
132Start a backend server, with \fIargs\fR specifying what to listen to.
133See \fBCLIENT/SERVER\fR section.
f57a9c59 134.TP
49da1240 135.BI \-\-daemonize \fR=\fPpidfile
7db7a5a0 136Background a fio server, writing the pid to the given \fIpidfile\fR file.
49da1240 137.TP
bdd88be3 138.BI \-\-client \fR=\fPhostname
7db7a5a0
TK
139Instead of running the jobs locally, send and run them on the given \fIhostname\fR
140or set of \fIhostname\fRs. See \fBCLIENT/SERVER\fR section.
bdd88be3 141.TP
7db7a5a0
TK
142.BI \-\-remote\-config \fR=\fPfile
143Tell fio server to load this local \fIfile\fR.
f2a2ce0e
HL
144.TP
145.BI \-\-idle\-prof \fR=\fPoption
7db7a5a0 146Report CPU idleness. \fIoption\fR is one of the following:
bdd88be3
TK
147.RS
148.RS
149.TP
150.B calibrate
151Run unit work calibration only and exit.
152.TP
153.B system
154Show aggregate system idleness and unit work.
155.TP
156.B percpu
7db7a5a0 157As \fBsystem\fR but also show per CPU idleness.
bdd88be3
TK
158.RE
159.RE
160.TP
7db7a5a0
TK
161.BI \-\-inflate\-log \fR=\fPlog
162Inflate and output compressed \fIlog\fR.
bdd88be3 163.TP
7db7a5a0
TK
164.BI \-\-trigger\-file \fR=\fPfile
165Execute trigger command when \fIfile\fR exists.
bdd88be3 166.TP
7db7a5a0
TK
167.BI \-\-trigger\-timeout \fR=\fPtime
168Execute trigger at this \fItime\fR.
bdd88be3 169.TP
7db7a5a0
TK
170.BI \-\-trigger \fR=\fPcommand
171Set this \fIcommand\fR as local trigger.
bdd88be3 172.TP
7db7a5a0
TK
173.BI \-\-trigger\-remote \fR=\fPcommand
174Set this \fIcommand\fR as remote trigger.
bdd88be3 175.TP
7db7a5a0 176.BI \-\-aux\-path \fR=\fPpath
f4401bf8
SW
177Use the directory specified by \fIpath\fP for generated state files instead
178of the current working directory.
d60e92d1 179.SH "JOB FILE FORMAT"
7a14cf18
TK
180Any parameters following the options will be assumed to be job files, unless
181they match a job file parameter. Multiple job files can be listed and each job
7db7a5a0 182file will be regarded as a separate group. Fio will \fBstonewall\fR execution
7a14cf18
TK
183between each group.
184
185Fio accepts one or more job files describing what it is
186supposed to do. The job file format is the classic ini file, where the names
187enclosed in [] brackets define the job name. You are free to use any ASCII name
188you want, except *global* which has special meaning. Following the job name is
189a sequence of zero or more parameters, one per line, that define the behavior of
190the job. If the first character in a line is a ';' or a '#', the entire line is
191discarded as a comment.
192
193A *global* section sets defaults for the jobs described in that file. A job may
194override a *global* section parameter, and a job file may even have several
195*global* sections if so desired. A job is only affected by a *global* section
196residing above it.
197
7db7a5a0
TK
198The \fB\-\-cmdhelp\fR option also lists all options. If used with an \fIcommand\fR
199argument, \fB\-\-cmdhelp\fR will detail the given \fIcommand\fR.
7a14cf18 200
7db7a5a0
TK
201See the `examples/' directory for inspiration on how to write job files. Note
202the copyright and license requirements currently apply to
203`examples/' files.
07a2919d
NP
204
205Note that the maximum length of a line in the job file is 8192 bytes.
54eb4569
TK
206.SH "JOB FILE PARAMETERS"
207Some parameters take an option of a given type, such as an integer or a
208string. Anywhere a numeric value is required, an arithmetic expression may be
209used, provided it is surrounded by parentheses. Supported operators are:
d59aa780 210.RS
7db7a5a0 211.P
d59aa780 212.B addition (+)
7db7a5a0
TK
213.P
214.B subtraction (\-)
215.P
d59aa780 216.B multiplication (*)
7db7a5a0 217.P
d59aa780 218.B division (/)
7db7a5a0 219.P
d59aa780 220.B modulus (%)
7db7a5a0 221.P
d59aa780
JA
222.B exponentiation (^)
223.RE
d59aa780
JA
224.P
225For time values in expressions, units are microseconds by default. This is
226different than for time values not in expressions (not enclosed in
54eb4569
TK
227parentheses).
228.SH "PARAMETER TYPES"
229The following parameter types are used.
d60e92d1
AC
230.TP
231.I str
6b86fc18
TK
232String. A sequence of alphanumeric characters.
233.TP
234.I time
235Integer with possible time suffix. Without a unit value is interpreted as
236seconds unless otherwise specified. Accepts a suffix of 'd' for days, 'h' for
237hours, 'm' for minutes, 's' for seconds, 'ms' (or 'msec') for milliseconds and 'us'
238(or 'usec') for microseconds. For example, use 10m for 10 minutes.
d60e92d1
AC
239.TP
240.I int
6d500c2e
RE
241Integer. A whole number value, which may contain an integer prefix
242and an integer suffix.
0b43a833
TK
243.RS
244.RS
245.P
6b86fc18 246[*integer prefix*] **number** [*integer suffix*]
0b43a833
TK
247.RE
248.P
6b86fc18
TK
249The optional *integer prefix* specifies the number's base. The default
250is decimal. *0x* specifies hexadecimal.
0b43a833 251.P
6b86fc18
TK
252The optional *integer suffix* specifies the number's units, and includes an
253optional unit prefix and an optional unit. For quantities of data, the
254default unit is bytes. For quantities of time, the default unit is seconds
255unless otherwise specified.
0b43a833
TK
256.P
257With `kb_base=1000', fio follows international standards for unit
338f2db5 258prefixes. To specify power-of-10 decimal values defined in the
6b86fc18 259International System of Units (SI):
0b43a833
TK
260.RS
261.P
7db7a5a0 262.PD 0
eccce61a 263K means kilo (K) or 1000
7db7a5a0 264.P
eccce61a 265M means mega (M) or 1000**2
7db7a5a0 266.P
eccce61a 267G means giga (G) or 1000**3
7db7a5a0 268.P
eccce61a 269T means tera (T) or 1000**4
7db7a5a0 270.P
eccce61a 271P means peta (P) or 1000**5
7db7a5a0 272.PD
0b43a833
TK
273.RE
274.P
338f2db5 275To specify power-of-2 binary values defined in IEC 80000-13:
0b43a833
TK
276.RS
277.P
7db7a5a0 278.PD 0
eccce61a 279Ki means kibi (Ki) or 1024
7db7a5a0 280.P
eccce61a 281Mi means mebi (Mi) or 1024**2
7db7a5a0 282.P
eccce61a 283Gi means gibi (Gi) or 1024**3
7db7a5a0 284.P
eccce61a 285Ti means tebi (Ti) or 1024**4
7db7a5a0 286.P
eccce61a 287Pi means pebi (Pi) or 1024**5
7db7a5a0 288.PD
0b43a833
TK
289.RE
290.P
291With `kb_base=1024' (the default), the unit prefixes are opposite
338f2db5 292from those specified in the SI and IEC 80000-13 standards to provide
6b86fc18 293compatibility with old scripts. For example, 4k means 4096.
0b43a833 294.P
6b86fc18
TK
295For quantities of data, an optional unit of 'B' may be included
296(e.g., 'kB' is the same as 'k').
0b43a833 297.P
6b86fc18
TK
298The *integer suffix* is not case sensitive (e.g., m/mi mean mebi/mega,
299not milli). 'b' and 'B' both mean byte, not bit.
0b43a833
TK
300.P
301Examples with `kb_base=1000':
302.RS
303.P
7db7a5a0 304.PD 0
6d500c2e 3054 KiB: 4096, 4096b, 4096B, 4k, 4kb, 4kB, 4K, 4KB
7db7a5a0 306.P
6d500c2e 3071 MiB: 1048576, 1m, 1024k
7db7a5a0 308.P
6d500c2e 3091 MB: 1000000, 1mi, 1000ki
7db7a5a0 310.P
6d500c2e 3111 TiB: 1073741824, 1t, 1024m, 1048576k
7db7a5a0 312.P
6d500c2e 3131 TB: 1000000000, 1ti, 1000mi, 1000000ki
7db7a5a0 314.PD
0b43a833
TK
315.RE
316.P
317Examples with `kb_base=1024' (default):
318.RS
319.P
7db7a5a0 320.PD 0
6d500c2e 3214 KiB: 4096, 4096b, 4096B, 4k, 4kb, 4kB, 4K, 4KB
7db7a5a0 322.P
6d500c2e 3231 MiB: 1048576, 1m, 1024k
7db7a5a0 324.P
6d500c2e 3251 MB: 1000000, 1mi, 1000ki
7db7a5a0 326.P
6d500c2e 3271 TiB: 1073741824, 1t, 1024m, 1048576k
7db7a5a0 328.P
6d500c2e 3291 TB: 1000000000, 1ti, 1000mi, 1000000ki
7db7a5a0 330.PD
0b43a833
TK
331.RE
332.P
6d500c2e 333To specify times (units are not case sensitive):
0b43a833
TK
334.RS
335.P
7db7a5a0 336.PD 0
6d500c2e 337D means days
7db7a5a0 338.P
6d500c2e 339H means hours
7db7a5a0 340.P
6d500c2e 341M mean minutes
7db7a5a0 342.P
6d500c2e 343s or sec means seconds (default)
7db7a5a0 344.P
6d500c2e 345ms or msec means milliseconds
7db7a5a0 346.P
6d500c2e 347us or usec means microseconds
7db7a5a0 348.PD
0b43a833
TK
349.RE
350.P
6b86fc18 351If the option accepts an upper and lower range, use a colon ':' or
7db7a5a0 352minus '\-' to separate such values. See \fIirange\fR parameter type.
6b86fc18
TK
353If the lower value specified happens to be larger than the upper value
354the two values are swapped.
0b43a833 355.RE
d60e92d1
AC
356.TP
357.I bool
6b86fc18
TK
358Boolean. Usually parsed as an integer, however only defined for
359true and false (1 and 0).
d60e92d1
AC
360.TP
361.I irange
6b86fc18 362Integer range with suffix. Allows value range to be given, such as
7db7a5a0 3631024\-4096. A colon may also be used as the separator, e.g. 1k:4k. If the
6b86fc18 364option allows two sets of ranges, they can be specified with a ',' or '/'
7db7a5a0 365delimiter: 1k\-4k/8k\-32k. Also see \fIint\fR parameter type.
83349190
YH
366.TP
367.I float_list
6b86fc18 368A list of floating point numbers, separated by a ':' character.
523bad63 369.SH "JOB PARAMETERS"
54eb4569 370With the above in mind, here follows the complete list of fio job parameters.
523bad63 371.SS "Units"
d60e92d1 372.TP
523bad63
TK
373.BI kb_base \fR=\fPint
374Select the interpretation of unit prefixes in input parameters.
375.RS
376.RS
d60e92d1 377.TP
523bad63 378.B 1000
338f2db5 379Inputs comply with IEC 80000-13 and the International
523bad63
TK
380System of Units (SI). Use:
381.RS
382.P
383.PD 0
338f2db5 384\- power-of-2 values with IEC prefixes (e.g., KiB)
523bad63 385.P
338f2db5 386\- power-of-10 values with SI prefixes (e.g., kB)
523bad63
TK
387.PD
388.RE
389.TP
390.B 1024
391Compatibility mode (default). To avoid breaking old scripts:
392.P
393.RS
394.PD 0
338f2db5 395\- power-of-2 values with SI prefixes
523bad63 396.P
338f2db5 397\- power-of-10 values with IEC prefixes
523bad63
TK
398.PD
399.RE
400.RE
401.P
402See \fBbs\fR for more details on input parameters.
403.P
404Outputs always use correct prefixes. Most outputs include both
338f2db5 405side-by-side, like:
523bad63
TK
406.P
407.RS
408bw=2383.3kB/s (2327.4KiB/s)
409.RE
410.P
411If only one value is reported, then kb_base selects the one to use:
412.P
413.RS
414.PD 0
4151000 \-\- SI prefixes
416.P
4171024 \-\- IEC prefixes
418.PD
419.RE
420.RE
421.TP
422.BI unit_base \fR=\fPint
423Base unit for reporting. Allowed values are:
424.RS
425.RS
426.TP
427.B 0
338f2db5 428Use auto-detection (default).
523bad63
TK
429.TP
430.B 8
431Byte based.
432.TP
433.B 1
434Bit based.
435.RE
436.RE
437.SS "Job description"
438.TP
439.BI name \fR=\fPstr
440ASCII name of the job. This may be used to override the name printed by fio
441for this job. Otherwise the job name is used. On the command line this
442parameter has the special purpose of also signaling the start of a new job.
9cc8cb91 443.TP
d60e92d1 444.BI description \fR=\fPstr
523bad63
TK
445Text description of the job. Doesn't do anything except dump this text
446description when this job is run. It's not parsed.
447.TP
448.BI loops \fR=\fPint
449Run the specified number of iterations of this job. Used to repeat the same
450workload a given number of times. Defaults to 1.
451.TP
452.BI numjobs \fR=\fPint
453Create the specified number of clones of this job. Each clone of job
454is spawned as an independent thread or process. May be used to setup a
455larger number of threads/processes doing the same thing. Each thread is
456reported separately; to see statistics for all clones as a whole, use
457\fBgroup_reporting\fR in conjunction with \fBnew_group\fR.
458See \fB\-\-max\-jobs\fR. Default: 1.
459.SS "Time related parameters"
460.TP
461.BI runtime \fR=\fPtime
462Tell fio to terminate processing after the specified period of time. It
463can be quite hard to determine for how long a specified job will run, so
464this parameter is handy to cap the total runtime to a given time. When
f1dd3fb1 465the unit is omitted, the value is interpreted in seconds.
523bad63
TK
466.TP
467.BI time_based
468If set, fio will run for the duration of the \fBruntime\fR specified
469even if the file(s) are completely read or written. It will simply loop over
470the same workload as many times as the \fBruntime\fR allows.
471.TP
472.BI startdelay \fR=\fPirange(int)
473Delay the start of job for the specified amount of time. Can be a single
474value or a range. When given as a range, each thread will choose a value
475randomly from within the range. Value is in seconds if a unit is omitted.
476.TP
477.BI ramp_time \fR=\fPtime
478If set, fio will run the specified workload for this amount of time before
479logging any performance numbers. Useful for letting performance settle
480before logging results, thus minimizing the runtime required for stable
481results. Note that the \fBramp_time\fR is considered lead in time for a job,
482thus it will increase the total runtime if a special timeout or
483\fBruntime\fR is specified. When the unit is omitted, the value is
484given in seconds.
485.TP
486.BI clocksource \fR=\fPstr
487Use the given clocksource as the base of timing. The supported options are:
488.RS
489.RS
490.TP
491.B gettimeofday
492\fBgettimeofday\fR\|(2)
493.TP
494.B clock_gettime
495\fBclock_gettime\fR\|(2)
496.TP
497.B cpu
498Internal CPU clock source
499.RE
500.P
501\fBcpu\fR is the preferred clocksource if it is reliable, as it is very fast (and
502fio is heavy on time calls). Fio will automatically use this clocksource if
503it's supported and considered reliable on the system it is running on,
504unless another clocksource is specifically set. For x86/x86\-64 CPUs, this
505means supporting TSC Invariant.
506.RE
507.TP
508.BI gtod_reduce \fR=\fPbool
509Enable all of the \fBgettimeofday\fR\|(2) reducing options
510(\fBdisable_clat\fR, \fBdisable_slat\fR, \fBdisable_bw_measurement\fR) plus
511reduce precision of the timeout somewhat to really shrink the
512\fBgettimeofday\fR\|(2) call count. With this option enabled, we only do
513about 0.4% of the \fBgettimeofday\fR\|(2) calls we would have done if all
514time keeping was enabled.
515.TP
516.BI gtod_cpu \fR=\fPint
517Sometimes it's cheaper to dedicate a single thread of execution to just
518getting the current time. Fio (and databases, for instance) are very
519intensive on \fBgettimeofday\fR\|(2) calls. With this option, you can set
520one CPU aside for doing nothing but logging current time to a shared memory
521location. Then the other threads/processes that run I/O workloads need only
522copy that segment, instead of entering the kernel with a
523\fBgettimeofday\fR\|(2) call. The CPU set aside for doing these time
524calls will be excluded from other uses. Fio will manually clear it from the
525CPU mask of other jobs.
526.SS "Target file/device"
d60e92d1
AC
527.TP
528.BI directory \fR=\fPstr
523bad63
TK
529Prefix \fBfilename\fRs with this directory. Used to place files in a different
530location than `./'. You can specify a number of directories by
531separating the names with a ':' character. These directories will be
532assigned equally distributed to job clones created by \fBnumjobs\fR as
533long as they are using generated filenames. If specific \fBfilename\fR(s) are
534set fio will use the first listed directory, and thereby matching the
f4401bf8
SW
535\fBfilename\fR semantic (which generates a file for each clone if not
536specified, but lets all clones use the same file if set).
523bad63
TK
537.RS
538.P
ffc90a44 539See the \fBfilename\fR option for information on how to escape ':' and '\\'
523bad63 540characters within the directory path itself.
f4401bf8
SW
541.P
542Note: To control the directory fio will use for internal state files
543use \fB\-\-aux\-path\fR.
523bad63 544.RE
d60e92d1
AC
545.TP
546.BI filename \fR=\fPstr
523bad63
TK
547Fio normally makes up a \fBfilename\fR based on the job name, thread number, and
548file number (see \fBfilename_format\fR). If you want to share files
549between threads in a job or several
550jobs with fixed file paths, specify a \fBfilename\fR for each of them to override
551the default. If the ioengine is file based, you can specify a number of files
552by separating the names with a ':' colon. So if you wanted a job to open
553`/dev/sda' and `/dev/sdb' as the two working files, you would use
554`filename=/dev/sda:/dev/sdb'. This also means that whenever this option is
555specified, \fBnrfiles\fR is ignored. The size of regular files specified
556by this option will be \fBsize\fR divided by number of files unless an
557explicit size is specified by \fBfilesize\fR.
558.RS
559.P
ffc90a44 560Each colon and backslash in the wanted path must be escaped with a '\\'
523bad63
TK
561character. For instance, if the path is `/dev/dsk/foo@3,0:c' then you
562would use `filename=/dev/dsk/foo@3,0\\:c' and if the path is
ffc90a44 563`F:\\filename' then you would use `filename=F\\:\\\\filename'.
523bad63 564.P
ffc90a44
SW
565On Windows, disk devices are accessed as `\\\\.\\PhysicalDrive0' for
566the first device, `\\\\.\\PhysicalDrive1' for the second etc.
523bad63 567Note: Windows and FreeBSD prevent write access to areas
338f2db5 568of the disk containing in-use data (e.g. filesystems).
523bad63
TK
569.P
570The filename `\-' is a reserved name, meaning *stdin* or *stdout*. Which
571of the two depends on the read/write direction set.
572.RE
d60e92d1 573.TP
de98bd30 574.BI filename_format \fR=\fPstr
523bad63
TK
575If sharing multiple files between jobs, it is usually necessary to have fio
576generate the exact names that you want. By default, fio will name a file
de98bd30 577based on the default file format specification of
523bad63 578`jobname.jobnumber.filenumber'. With this option, that can be
de98bd30
JA
579customized. Fio will recognize and replace the following keywords in this
580string:
581.RS
582.RS
583.TP
584.B $jobname
585The name of the worker thread or process.
586.TP
587.B $jobnum
588The incremental number of the worker thread or process.
589.TP
590.B $filenum
591The incremental number of the file for that worker thread or process.
592.RE
593.P
523bad63
TK
594To have dependent jobs share a set of files, this option can be set to have
595fio generate filenames that are shared between the two. For instance, if
596`testfiles.$filenum' is specified, file number 4 for any job will be
597named `testfiles.4'. The default of `$jobname.$jobnum.$filenum'
de98bd30 598will be used if no other format specifier is given.
645943c0
JB
599.P
600If you specify a path then the directories will be created up to the main
601directory for the file. So for example if you specify `a/b/c/$jobnum` then the
602directories a/b/c will be created before the file setup part of the job. If you
603specify \fBdirectory\fR then the path will be relative that directory, otherwise
604it is treated as the absolute path.
de98bd30 605.RE
de98bd30 606.TP
922a5be8 607.BI unique_filename \fR=\fPbool
523bad63
TK
608To avoid collisions between networked clients, fio defaults to prefixing any
609generated filenames (with a directory specified) with the source of the
610client connecting. To disable this behavior, set this option to 0.
611.TP
612.BI opendir \fR=\fPstr
613Recursively open any files below directory \fIstr\fR.
922a5be8 614.TP
3ce9dcaf 615.BI lockfile \fR=\fPstr
523bad63
TK
616Fio defaults to not locking any files before it does I/O to them. If a file
617or file descriptor is shared, fio can serialize I/O to that file to make the
618end result consistent. This is usual for emulating real workloads that share
619files. The lock modes are:
3ce9dcaf
JA
620.RS
621.RS
622.TP
623.B none
523bad63 624No locking. The default.
3ce9dcaf
JA
625.TP
626.B exclusive
523bad63 627Only one thread or process may do I/O at a time, excluding all others.
3ce9dcaf
JA
628.TP
629.B readwrite
523bad63
TK
630Read\-write locking on the file. Many readers may
631access the file at the same time, but writes get exclusive access.
3ce9dcaf 632.RE
ce594fbe 633.RE
523bad63
TK
634.TP
635.BI nrfiles \fR=\fPint
636Number of files to use for this job. Defaults to 1. The size of files
637will be \fBsize\fR divided by this unless explicit size is specified by
638\fBfilesize\fR. Files are created for each thread separately, and each
639file will have a file number within its name by default, as explained in
640\fBfilename\fR section.
641.TP
642.BI openfiles \fR=\fPint
643Number of files to keep open at the same time. Defaults to the same as
644\fBnrfiles\fR, can be set smaller to limit the number simultaneous
645opens.
646.TP
647.BI file_service_type \fR=\fPstr
648Defines how fio decides which file from a job to service next. The following
649types are defined:
650.RS
651.RS
652.TP
653.B random
654Choose a file at random.
655.TP
656.B roundrobin
657Round robin over opened files. This is the default.
658.TP
659.B sequential
660Finish one file before moving on to the next. Multiple files can
661still be open depending on \fBopenfiles\fR.
662.TP
663.B zipf
664Use a Zipf distribution to decide what file to access.
665.TP
666.B pareto
667Use a Pareto distribution to decide what file to access.
668.TP
669.B normal
670Use a Gaussian (normal) distribution to decide what file to access.
671.TP
672.B gauss
673Alias for normal.
674.RE
3ce9dcaf 675.P
523bad63
TK
676For \fBrandom\fR, \fBroundrobin\fR, and \fBsequential\fR, a postfix can be appended to
677tell fio how many I/Os to issue before switching to a new file. For example,
678specifying `file_service_type=random:8' would cause fio to issue
338f2db5 6798 I/Os before selecting a new file at random. For the non-uniform
523bad63
TK
680distributions, a floating point postfix can be given to influence how the
681distribution is skewed. See \fBrandom_distribution\fR for a description
682of how that would work.
683.RE
684.TP
685.BI ioscheduler \fR=\fPstr
686Attempt to switch the device hosting the file to the specified I/O scheduler
687before running.
688.TP
689.BI create_serialize \fR=\fPbool
690If true, serialize the file creation for the jobs. This may be handy to
691avoid interleaving of data files, which may greatly depend on the filesystem
692used and even the number of processors in the system. Default: true.
693.TP
694.BI create_fsync \fR=\fPbool
695\fBfsync\fR\|(2) the data file after creation. This is the default.
696.TP
697.BI create_on_open \fR=\fPbool
338f2db5
SW
698If true, don't pre-create files but allow the job's open() to create a file
699when it's time to do I/O. Default: false \-\- pre-create all necessary files
523bad63
TK
700when the job starts.
701.TP
702.BI create_only \fR=\fPbool
703If true, fio will only run the setup phase of the job. If files need to be
704laid out or updated on disk, only that will be done \-\- the actual job contents
705are not executed. Default: false.
706.TP
707.BI allow_file_create \fR=\fPbool
708If true, fio is permitted to create files as part of its workload. If this
709option is false, then fio will error out if
710the files it needs to use don't already exist. Default: true.
711.TP
712.BI allow_mounted_write \fR=\fPbool
713If this isn't set, fio will abort jobs that are destructive (e.g. that write)
714to what appears to be a mounted device or partition. This should help catch
715creating inadvertently destructive tests, not realizing that the test will
716destroy data on the mounted file system. Note that some platforms don't allow
717writing against a mounted device regardless of this option. Default: false.
718.TP
719.BI pre_read \fR=\fPbool
338f2db5 720If this is given, files will be pre-read into memory before starting the
523bad63 721given I/O operation. This will also clear the \fBinvalidate\fR flag,
338f2db5
SW
722since it is pointless to pre-read and then drop the cache. This will only
723work for I/O engines that are seek-able, since they allow you to read the
724same data multiple times. Thus it will not work on non-seekable I/O engines
523bad63
TK
725(e.g. network, splice). Default: false.
726.TP
727.BI unlink \fR=\fPbool
728Unlink the job files when done. Not the default, as repeated runs of that
729job would then waste time recreating the file set again and again. Default:
730false.
731.TP
732.BI unlink_each_loop \fR=\fPbool
733Unlink job files after each iteration or loop. Default: false.
734.TP
7b865a2f
BVA
735.BI zonemode \fR=\fPstr
736Accepted values are:
737.RS
738.RS
739.TP
740.B none
741The \fBzonerange\fR, \fBzonesize\fR and \fBzoneskip\fR parameters are ignored.
742.TP
743.B strided
744I/O happens in a single zone until \fBzonesize\fR bytes have been transferred.
745After that number of bytes has been transferred processing of the next zone
746starts.
747.TP
748.B zbd
749Zoned block device mode. I/O happens sequentially in each zone, even if random
750I/O has been selected. Random I/O happens across all zones instead of being
751restricted to a single zone.
752.RE
753.RE
523bad63
TK
754.TP
755.BI zonerange \fR=\fPint
d4e058cd
DLM
756For \fBzonemode\fR=strided, this is the size of a single zone. See also
757\fBzonesize\fR and \fBzoneskip\fR.
758
759For \fBzonemode\fR=zbd, this parameter is ignored.
5faddc64
BVA
760.TP
761.BI zonesize \fR=\fPint
7b865a2f
BVA
762For \fBzonemode\fR=strided, this is the number of bytes to transfer before
763skipping \fBzoneskip\fR bytes. If this parameter is smaller than
764\fBzonerange\fR then only a fraction of each zone with \fBzonerange\fR bytes
765will be accessed. If this parameter is larger than \fBzonerange\fR then each
766zone will be accessed multiple times before skipping to the next zone.
767
d4e058cd
DLM
768For \fBzonemode\fR=zbd, this is the size of a single zone. The
769\fBzonerange\fR parameter is ignored in this mode. For a job accessing a
770zoned block device, the specified \fBzonesize\fR must be 0 or equal to the
771device zone size. For a regular block device or file, the specified
772\fBzonesize\fR must be at least 512B.
523bad63
TK
773.TP
774.BI zoneskip \fR=\fPint
7b865a2f 775For \fBzonemode\fR=strided, the number of bytes to skip after \fBzonesize\fR
4d37720a
DLM
776bytes of data have been transferred.
777
778For \fBzonemode\fR=zbd, the \fBzonesize\fR aligned number of bytes to skip
779once a zone is fully written (write workloads) or all written data in the
780zone have been read (read workloads). This parameter is valid only for
781sequential workloads and ignored for random workloads. For read workloads,
782see also \fBread_beyond_wp\fR.
5faddc64 783
bfbdd35b
BVA
784.TP
785.BI read_beyond_wp \fR=\fPbool
786This parameter applies to \fBzonemode=zbd\fR only.
787
788Zoned block devices are block devices that consist of multiple zones. Each
789zone has a type, e.g. conventional or sequential. A conventional zone can be
790written at any offset that is a multiple of the block size. Sequential zones
791must be written sequentially. The position at which a write must occur is
402f0887
DLM
792called the write pointer. A zoned block device can be either host managed or
793host aware. For host managed devices the host must ensure that writes happen
794sequentially. Fio recognizes host managed devices and serializes writes to
795sequential zones for these devices.
bfbdd35b
BVA
796
797If a read occurs in a sequential zone beyond the write pointer then the zoned
798block device will complete the read without reading any data from the storage
799medium. Since such reads lead to unrealistically high bandwidth and IOPS
800numbers fio only reads beyond the write pointer if explicitly told to do
801so. Default: false.
59b07544
BVA
802.TP
803.BI max_open_zones \fR=\fPint
804When running a random write test across an entire drive many more zones will be
805open than in a typical application workload. Hence this command line option
806that allows to limit the number of open zones. The number of open zones is
807defined as the number of zones to which write commands are issued.
a7c2b6fc
BVA
808.TP
809.BI zone_reset_threshold \fR=\fPfloat
810A number between zero and one that indicates the ratio of logical blocks with
811data to the total number of logical blocks in the test above which zones
812should be reset periodically.
813.TP
814.BI zone_reset_frequency \fR=\fPfloat
815A number between zero and one that indicates how often a zone reset should be
816issued if the zone reset threshold has been exceeded. A zone reset is
817submitted after each (1 / zone_reset_frequency) write requests. This and the
818previous parameter can be used to simulate garbage collection activity.
bfbdd35b 819
523bad63
TK
820.SS "I/O type"
821.TP
822.BI direct \fR=\fPbool
338f2db5 823If value is true, use non-buffered I/O. This is usually O_DIRECT. Note that
8e889110 824OpenBSD and ZFS on Solaris don't support direct I/O. On Windows the synchronous
523bad63
TK
825ioengines don't support direct I/O. Default: false.
826.TP
827.BI atomic \fR=\fPbool
828If value is true, attempt to use atomic direct I/O. Atomic writes are
829guaranteed to be stable once acknowledged by the operating system. Only
830Linux supports O_ATOMIC right now.
831.TP
832.BI buffered \fR=\fPbool
833If value is true, use buffered I/O. This is the opposite of the
834\fBdirect\fR option. Defaults to true.
d60e92d1
AC
835.TP
836.BI readwrite \fR=\fPstr "\fR,\fP rw" \fR=\fPstr
523bad63 837Type of I/O pattern. Accepted values are:
d60e92d1
AC
838.RS
839.RS
840.TP
841.B read
d1429b5c 842Sequential reads.
d60e92d1
AC
843.TP
844.B write
d1429b5c 845Sequential writes.
d60e92d1 846.TP
fa769d44 847.B trim
3740cfc8 848Sequential trims (Linux block devices and SCSI character devices only).
fa769d44 849.TP
d60e92d1 850.B randread
d1429b5c 851Random reads.
d60e92d1
AC
852.TP
853.B randwrite
d1429b5c 854Random writes.
d60e92d1 855.TP
fa769d44 856.B randtrim
3740cfc8 857Random trims (Linux block devices and SCSI character devices only).
fa769d44 858.TP
523bad63
TK
859.B rw,readwrite
860Sequential mixed reads and writes.
d60e92d1 861.TP
ff6bb260 862.B randrw
523bad63 863Random mixed reads and writes.
82a90686
JA
864.TP
865.B trimwrite
523bad63
TK
866Sequential trim+write sequences. Blocks will be trimmed first,
867then the same blocks will be written to.
d60e92d1
AC
868.RE
869.P
523bad63
TK
870Fio defaults to read if the option is not specified. For the mixed I/O
871types, the default is to split them 50/50. For certain types of I/O the
872result may still be skewed a bit, since the speed may be different.
873.P
874It is possible to specify the number of I/Os to do before getting a new
875offset by appending `:<nr>' to the end of the string given. For a
876random read, it would look like `rw=randread:8' for passing in an offset
877modifier with a value of 8. If the suffix is used with a sequential I/O
878pattern, then the `<nr>' value specified will be added to the generated
879offset for each I/O turning sequential I/O into sequential I/O with holes.
880For instance, using `rw=write:4k' will skip 4k for every write. Also see
881the \fBrw_sequencer\fR option.
d60e92d1
AC
882.RE
883.TP
38dad62d 884.BI rw_sequencer \fR=\fPstr
523bad63
TK
885If an offset modifier is given by appending a number to the `rw=\fIstr\fR'
886line, then this option controls how that number modifies the I/O offset
887being generated. Accepted values are:
38dad62d
JA
888.RS
889.RS
890.TP
891.B sequential
523bad63 892Generate sequential offset.
38dad62d
JA
893.TP
894.B identical
523bad63 895Generate the same offset.
38dad62d
JA
896.RE
897.P
523bad63
TK
898\fBsequential\fR is only useful for random I/O, where fio would normally
899generate a new random offset for every I/O. If you append e.g. 8 to randread,
900you would get a new random offset for every 8 I/Os. The result would be a
901seek for only every 8 I/Os, instead of for every I/O. Use `rw=randread:8'
902to specify that. As sequential I/O is already sequential, setting
903\fBsequential\fR for that would not result in any differences. \fBidentical\fR
904behaves in a similar fashion, except it sends the same offset 8 number of
905times before generating a new offset.
38dad62d 906.RE
90fef2d1 907.TP
771e58be
JA
908.BI unified_rw_reporting \fR=\fPbool
909Fio normally reports statistics on a per data direction basis, meaning that
523bad63
TK
910reads, writes, and trims are accounted and reported separately. If this
911option is set fio sums the results and report them as "mixed" instead.
771e58be 912.TP
d60e92d1 913.BI randrepeat \fR=\fPbool
523bad63
TK
914Seed the random number generator used for random I/O patterns in a
915predictable way so the pattern is repeatable across runs. Default: true.
56e2a5fc
CE
916.TP
917.BI allrandrepeat \fR=\fPbool
918Seed all random number generators in a predictable way so results are
523bad63 919repeatable across runs. Default: false.
d60e92d1 920.TP
04778baf
JA
921.BI randseed \fR=\fPint
922Seed the random number generators based on this seed value, to be able to
923control what sequence of output is being generated. If not set, the random
924sequence depends on the \fBrandrepeat\fR setting.
925.TP
a596f047 926.BI fallocate \fR=\fPstr
338f2db5 927Whether pre-allocation is performed when laying down files.
523bad63 928Accepted values are:
a596f047
EG
929.RS
930.RS
931.TP
932.B none
338f2db5 933Do not pre-allocate space.
a596f047 934.TP
2c3e17be 935.B native
338f2db5 936Use a platform's native pre-allocation call but fall back to
523bad63 937\fBnone\fR behavior if it fails/is not implemented.
2c3e17be 938.TP
a596f047 939.B posix
338f2db5 940Pre-allocate via \fBposix_fallocate\fR\|(3).
a596f047
EG
941.TP
942.B keep
338f2db5 943Pre-allocate via \fBfallocate\fR\|(2) with
523bad63 944FALLOC_FL_KEEP_SIZE set.
a596f047
EG
945.TP
946.B 0
338f2db5 947Backward-compatible alias for \fBnone\fR.
a596f047
EG
948.TP
949.B 1
338f2db5 950Backward-compatible alias for \fBposix\fR.
a596f047
EG
951.RE
952.P
523bad63
TK
953May not be available on all supported platforms. \fBkeep\fR is only available
954on Linux. If using ZFS on Solaris this cannot be set to \fBposix\fR
338f2db5
SW
955because ZFS doesn't support pre-allocation. Default: \fBnative\fR if any
956pre-allocation methods are available, \fBnone\fR if not.
a596f047 957.RE
7bc8c2cf 958.TP
ecb2083d 959.BI fadvise_hint \fR=\fPstr
c712c97a
JA
960Use \fBposix_fadvise\fR\|(2) or \fBposix_madvise\fR\|(2) to advise the kernel
961what I/O patterns are likely to be issued. Accepted values are:
ecb2083d
JA
962.RS
963.RS
964.TP
965.B 0
966Backwards compatible hint for "no hint".
967.TP
968.B 1
969Backwards compatible hint for "advise with fio workload type". This
523bad63 970uses FADV_RANDOM for a random workload, and FADV_SEQUENTIAL
ecb2083d
JA
971for a sequential workload.
972.TP
973.B sequential
523bad63 974Advise using FADV_SEQUENTIAL.
ecb2083d
JA
975.TP
976.B random
523bad63 977Advise using FADV_RANDOM.
ecb2083d
JA
978.RE
979.RE
d60e92d1 980.TP
8f4b9f24 981.BI write_hint \fR=\fPstr
523bad63
TK
982Use \fBfcntl\fR\|(2) to advise the kernel what life time to expect
983from a write. Only supported on Linux, as of version 4.13. Accepted
8f4b9f24
JA
984values are:
985.RS
986.RS
987.TP
988.B none
989No particular life time associated with this file.
990.TP
991.B short
992Data written to this file has a short life time.
993.TP
994.B medium
995Data written to this file has a medium life time.
996.TP
997.B long
998Data written to this file has a long life time.
999.TP
1000.B extreme
1001Data written to this file has a very long life time.
1002.RE
523bad63
TK
1003.P
1004The values are all relative to each other, and no absolute meaning
1005should be associated with them.
8f4b9f24 1006.RE
37659335 1007.TP
523bad63
TK
1008.BI offset \fR=\fPint
1009Start I/O at the provided offset in the file, given as either a fixed size in
83c8b093
JF
1010bytes or a percentage. If a percentage is given, the generated offset will be
1011aligned to the minimum \fBblocksize\fR or to the value of \fBoffset_align\fR if
1012provided. Data before the given offset will not be touched. This
523bad63
TK
1013effectively caps the file size at `real_size \- offset'. Can be combined with
1014\fBsize\fR to constrain the start and end range of the I/O workload.
1015A percentage can be specified by a number between 1 and 100 followed by '%',
1016for example, `offset=20%' to specify 20%.
6d500c2e 1017.TP
83c8b093
JF
1018.BI offset_align \fR=\fPint
1019If set to non-zero value, the byte offset generated by a percentage \fBoffset\fR
1020is aligned upwards to this value. Defaults to 0 meaning that a percentage
1021offset is aligned to the minimum block size.
1022.TP
523bad63
TK
1023.BI offset_increment \fR=\fPint
1024If this is provided, then the real offset becomes `\fBoffset\fR + \fBoffset_increment\fR
1025* thread_number', where the thread number is a counter that starts at 0 and
338f2db5 1026is incremented for each sub-job (i.e. when \fBnumjobs\fR option is
523bad63
TK
1027specified). This option is useful if there are several jobs which are
1028intended to operate on a file in parallel disjoint segments, with even
0b288ba1
VF
1029spacing between the starting points. Percentages can be used for this option.
1030If a percentage is given, the generated offset will be aligned to the minimum
1031\fBblocksize\fR or to the value of \fBoffset_align\fR if provided.
6d500c2e 1032.TP
523bad63
TK
1033.BI number_ios \fR=\fPint
1034Fio will normally perform I/Os until it has exhausted the size of the region
1035set by \fBsize\fR, or if it exhaust the allocated time (or hits an error
1036condition). With this setting, the range/size can be set independently of
1037the number of I/Os to perform. When fio reaches this number, it will exit
1038normally and report status. Note that this does not extend the amount of I/O
1039that will be done, it will only stop fio if this condition is met before
338f2db5 1040other end-of-job criteria.
d60e92d1 1041.TP
523bad63
TK
1042.BI fsync \fR=\fPint
1043If writing to a file, issue an \fBfsync\fR\|(2) (or its equivalent) of
1044the dirty data for every number of blocks given. For example, if you give 32
1045as a parameter, fio will sync the file after every 32 writes issued. If fio is
338f2db5 1046using non-buffered I/O, we may not sync the file. The exception is the sg
523bad63
TK
1047I/O engine, which synchronizes the disk cache anyway. Defaults to 0, which
1048means fio does not periodically issue and wait for a sync to complete. Also
1049see \fBend_fsync\fR and \fBfsync_on_close\fR.
6d500c2e 1050.TP
523bad63
TK
1051.BI fdatasync \fR=\fPint
1052Like \fBfsync\fR but uses \fBfdatasync\fR\|(2) to only sync data and
44f668d7 1053not metadata blocks. In Windows, FreeBSD, DragonFlyBSD or OSX there is no
523bad63
TK
1054\fBfdatasync\fR\|(2) so this falls back to using \fBfsync\fR\|(2).
1055Defaults to 0, which means fio does not periodically issue and wait for a
338f2db5 1056data-only sync to complete.
d60e92d1 1057.TP
523bad63
TK
1058.BI write_barrier \fR=\fPint
1059Make every N\-th write a barrier write.
901bb994 1060.TP
523bad63
TK
1061.BI sync_file_range \fR=\fPstr:int
1062Use \fBsync_file_range\fR\|(2) for every \fIint\fR number of write
1063operations. Fio will track range of writes that have happened since the last
1064\fBsync_file_range\fR\|(2) call. \fIstr\fR can currently be one or more of:
1065.RS
1066.RS
fd68418e 1067.TP
523bad63
TK
1068.B wait_before
1069SYNC_FILE_RANGE_WAIT_BEFORE
c5751c62 1070.TP
523bad63
TK
1071.B write
1072SYNC_FILE_RANGE_WRITE
c5751c62 1073.TP
523bad63
TK
1074.B wait_after
1075SYNC_FILE_RANGE_WRITE_AFTER
2fa5a241 1076.RE
523bad63
TK
1077.P
1078So if you do `sync_file_range=wait_before,write:8', fio would use
1079`SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE' for every 8
1080writes. Also see the \fBsync_file_range\fR\|(2) man page. This option is
1081Linux specific.
2fa5a241 1082.RE
ce35b1ec 1083.TP
523bad63
TK
1084.BI overwrite \fR=\fPbool
1085If true, writes to a file will always overwrite existing data. If the file
1086doesn't already exist, it will be created before the write phase begins. If
1087the file exists and is large enough for the specified write phase, nothing
1088will be done. Default: false.
5c94b008 1089.TP
523bad63
TK
1090.BI end_fsync \fR=\fPbool
1091If true, \fBfsync\fR\|(2) file contents when a write stage has completed.
1092Default: false.
d60e92d1 1093.TP
523bad63
TK
1094.BI fsync_on_close \fR=\fPbool
1095If true, fio will \fBfsync\fR\|(2) a dirty file on close. This differs
1096from \fBend_fsync\fR in that it will happen on every file close, not
1097just at the end of the job. Default: false.
d60e92d1 1098.TP
523bad63
TK
1099.BI rwmixread \fR=\fPint
1100Percentage of a mixed workload that should be reads. Default: 50.
1101.TP
1102.BI rwmixwrite \fR=\fPint
1103Percentage of a mixed workload that should be writes. If both
1104\fBrwmixread\fR and \fBrwmixwrite\fR is given and the values do not
1105add up to 100%, the latter of the two will be used to override the
1106first. This may interfere with a given rate setting, if fio is asked to
1107limit reads or writes to a certain rate. If that is the case, then the
1108distribution may be skewed. Default: 50.
1109.TP
1110.BI random_distribution \fR=\fPstr:float[,str:float][,str:float]
1111By default, fio will use a completely uniform random distribution when asked
1112to perform random I/O. Sometimes it is useful to skew the distribution in
1113specific ways, ensuring that some parts of the data is more hot than others.
1114fio includes the following distribution models:
d60e92d1
AC
1115.RS
1116.RS
1117.TP
1118.B random
523bad63 1119Uniform random distribution
8c07860d
JA
1120.TP
1121.B zipf
523bad63 1122Zipf distribution
8c07860d
JA
1123.TP
1124.B pareto
523bad63 1125Pareto distribution
8c07860d 1126.TP
dd3503d3 1127.B normal
523bad63 1128Normal (Gaussian) distribution
dd3503d3 1129.TP
523bad63
TK
1130.B zoned
1131Zoned random distribution
59466396
JA
1132.B zoned_abs
1133Zoned absolute random distribution
d60e92d1
AC
1134.RE
1135.P
523bad63
TK
1136When using a \fBzipf\fR or \fBpareto\fR distribution, an input value is also
1137needed to define the access pattern. For \fBzipf\fR, this is the `Zipf theta'.
1138For \fBpareto\fR, it's the `Pareto power'. Fio includes a test
1139program, \fBfio\-genzipf\fR, that can be used visualize what the given input
1140values will yield in terms of hit rates. If you wanted to use \fBzipf\fR with
1141a `theta' of 1.2, you would use `random_distribution=zipf:1.2' as the
1142option. If a non\-uniform model is used, fio will disable use of the random
1143map. For the \fBnormal\fR distribution, a normal (Gaussian) deviation is
1144supplied as a value between 0 and 100.
1145.P
1146For a \fBzoned\fR distribution, fio supports specifying percentages of I/O
1147access that should fall within what range of the file or device. For
1148example, given a criteria of:
d60e92d1 1149.RS
523bad63
TK
1150.P
1151.PD 0
115260% of accesses should be to the first 10%
1153.P
115430% of accesses should be to the next 20%
1155.P
11568% of accesses should be to the next 30%
1157.P
11582% of accesses should be to the next 40%
1159.PD
1160.RE
1161.P
1162we can define that through zoning of the random accesses. For the above
1163example, the user would do:
1164.RS
1165.P
1166random_distribution=zoned:60/10:30/20:8/30:2/40
1167.RE
1168.P
59466396
JA
1169A \fBzoned_abs\fR distribution works exactly like the\fBzoned\fR, except that
1170it takes absolute sizes. For example, let's say you wanted to define access
1171according to the following criteria:
1172.RS
1173.P
1174.PD 0
117560% of accesses should be to the first 20G
1176.P
117730% of accesses should be to the next 100G
1178.P
117910% of accesses should be to the next 500G
1180.PD
1181.RE
1182.P
1183we can define an absolute zoning distribution with:
1184.RS
1185.P
1186random_distribution=zoned:60/10:30/20:8/30:2/40
1187.RE
1188.P
6a16ece8
JA
1189For both \fBzoned\fR and \fBzoned_abs\fR, fio supports defining up to 256
1190separate zones.
1191.P
59466396 1192Similarly to how \fBbssplit\fR works for setting ranges and percentages
523bad63
TK
1193of block sizes. Like \fBbssplit\fR, it's possible to specify separate
1194zones for reads, writes, and trims. If just one set is given, it'll apply to
1195all of them.
1196.RE
1197.TP
1198.BI percentage_random \fR=\fPint[,int][,int]
1199For a random workload, set how big a percentage should be random. This
1200defaults to 100%, in which case the workload is fully random. It can be set
1201from anywhere from 0 to 100. Setting it to 0 would make the workload fully
1202sequential. Any setting in between will result in a random mix of sequential
338f2db5 1203and random I/O, at the given percentages. Comma-separated values may be
523bad63
TK
1204specified for reads, writes, and trims as described in \fBblocksize\fR.
1205.TP
1206.BI norandommap
1207Normally fio will cover every block of the file when doing random I/O. If
1208this option is given, fio will just get a new random offset without looking
1209at past I/O history. This means that some blocks may not be read or written,
1210and that some blocks may be read/written more than once. If this option is
1211used with \fBverify\fR and multiple blocksizes (via \fBbsrange\fR),
338f2db5 1212only intact blocks are verified, i.e., partially-overwritten blocks are
47e6a6e5
SW
1213ignored. With an async I/O engine and an I/O depth > 1, it is possible for
1214the same block to be overwritten, which can cause verification errors. Either
1215do not use norandommap in this case, or also use the lfsr random generator.
523bad63
TK
1216.TP
1217.BI softrandommap \fR=\fPbool
1218See \fBnorandommap\fR. If fio runs with the random block map enabled and
1219it fails to allocate the map, if this option is set it will continue without
1220a random block map. As coverage will not be as complete as with random maps,
1221this option is disabled by default.
1222.TP
1223.BI random_generator \fR=\fPstr
1224Fio supports the following engines for generating I/O offsets for random I/O:
1225.RS
1226.RS
1227.TP
1228.B tausworthe
1229Strong 2^88 cycle random number generator.
1230.TP
1231.B lfsr
1232Linear feedback shift register generator.
1233.TP
1234.B tausworthe64
1235Strong 64\-bit 2^258 cycle random number generator.
1236.RE
1237.P
1238\fBtausworthe\fR is a strong random number generator, but it requires tracking
1239on the side if we want to ensure that blocks are only read or written
1240once. \fBlfsr\fR guarantees that we never generate the same offset twice, and
1241it's also less computationally expensive. It's not a true random generator,
1242however, though for I/O purposes it's typically good enough. \fBlfsr\fR only
1243works with single block sizes, not with workloads that use multiple block
1244sizes. If used with such a workload, fio may read or write some blocks
1245multiple times. The default value is \fBtausworthe\fR, unless the required
1246space exceeds 2^32 blocks. If it does, then \fBtausworthe64\fR is
1247selected automatically.
1248.RE
1249.SS "Block size"
1250.TP
1251.BI blocksize \fR=\fPint[,int][,int] "\fR,\fB bs" \fR=\fPint[,int][,int]
1252The block size in bytes used for I/O units. Default: 4096. A single value
338f2db5 1253applies to reads, writes, and trims. Comma-separated values may be
523bad63
TK
1254specified for reads, writes, and trims. A value not terminated in a comma
1255applies to subsequent types. Examples:
1256.RS
1257.RS
1258.P
1259.PD 0
1260bs=256k means 256k for reads, writes and trims.
1261.P
1262bs=8k,32k means 8k for reads, 32k for writes and trims.
1263.P
1264bs=8k,32k, means 8k for reads, 32k for writes, and default for trims.
1265.P
1266bs=,8k means default for reads, 8k for writes and trims.
1267.P
1268bs=,8k, means default for reads, 8k for writes, and default for trims.
1269.PD
1270.RE
1271.RE
1272.TP
1273.BI blocksize_range \fR=\fPirange[,irange][,irange] "\fR,\fB bsrange" \fR=\fPirange[,irange][,irange]
1274A range of block sizes in bytes for I/O units. The issued I/O unit will
1275always be a multiple of the minimum size, unless
1276\fBblocksize_unaligned\fR is set.
338f2db5 1277Comma-separated ranges may be specified for reads, writes, and trims as
523bad63
TK
1278described in \fBblocksize\fR. Example:
1279.RS
1280.RS
1281.P
1282bsrange=1k\-4k,2k\-8k
1283.RE
1284.RE
1285.TP
1286.BI bssplit \fR=\fPstr[,str][,str]
1287Sometimes you want even finer grained control of the block sizes issued, not
1288just an even split between them. This option allows you to weight various
1289block sizes, so that you are able to define a specific amount of block sizes
1290issued. The format for this option is:
1291.RS
1292.RS
1293.P
1294bssplit=blocksize/percentage:blocksize/percentage
1295.RE
1296.P
1297for as many block sizes as needed. So if you want to define a workload that
1298has 50% 64k blocks, 10% 4k blocks, and 40% 32k blocks, you would write:
1299.RS
1300.P
1301bssplit=4k/10:64k/50:32k/40
1302.RE
1303.P
1304Ordering does not matter. If the percentage is left blank, fio will fill in
1305the remaining values evenly. So a bssplit option like this one:
1306.RS
1307.P
1308bssplit=4k/50:1k/:32k/
1309.RE
1310.P
1311would have 50% 4k ios, and 25% 1k and 32k ios. The percentages always add up
1312to 100, if bssplit is given a range that adds up to more, it will error out.
1313.P
338f2db5 1314Comma-separated values may be specified for reads, writes, and trims as
523bad63
TK
1315described in \fBblocksize\fR.
1316.P
1317If you want a workload that has 50% 2k reads and 50% 4k reads, while having
131890% 4k writes and 10% 8k writes, you would specify:
1319.RS
1320.P
cf04b906 1321bssplit=2k/50:4k/50,4k/90:8k/10
523bad63 1322.RE
6a16ece8
JA
1323.P
1324Fio supports defining up to 64 different weights for each data direction.
523bad63
TK
1325.RE
1326.TP
1327.BI blocksize_unaligned "\fR,\fB bs_unaligned"
1328If set, fio will issue I/O units with any size within
1329\fBblocksize_range\fR, not just multiples of the minimum size. This
1330typically won't work with direct I/O, as that normally requires sector
1331alignment.
1332.TP
1333.BI bs_is_seq_rand \fR=\fPbool
1334If this option is set, fio will use the normal read,write blocksize settings
1335as sequential,random blocksize settings instead. Any random read or write
1336will use the WRITE blocksize settings, and any sequential read or write will
1337use the READ blocksize settings.
1338.TP
1339.BI blockalign \fR=\fPint[,int][,int] "\fR,\fB ba" \fR=\fPint[,int][,int]
1340Boundary to which fio will align random I/O units. Default:
1341\fBblocksize\fR. Minimum alignment is typically 512b for using direct
1342I/O, though it usually depends on the hardware block size. This option is
1343mutually exclusive with using a random map for files, so it will turn off
338f2db5 1344that option. Comma-separated values may be specified for reads, writes, and
523bad63
TK
1345trims as described in \fBblocksize\fR.
1346.SS "Buffers and memory"
1347.TP
1348.BI zero_buffers
1349Initialize buffers with all zeros. Default: fill buffers with random data.
1350.TP
1351.BI refill_buffers
1352If this option is given, fio will refill the I/O buffers on every
1353submit. The default is to only fill it at init time and reuse that
1354data. Only makes sense if zero_buffers isn't specified, naturally. If data
1355verification is enabled, \fBrefill_buffers\fR is also automatically enabled.
1356.TP
1357.BI scramble_buffers \fR=\fPbool
1358If \fBrefill_buffers\fR is too costly and the target is using data
1359deduplication, then setting this option will slightly modify the I/O buffer
338f2db5 1360contents to defeat normal de-dupe attempts. This is not enough to defeat
523bad63
TK
1361more clever block compression attempts, but it will stop naive dedupe of
1362blocks. Default: true.
1363.TP
1364.BI buffer_compress_percentage \fR=\fPint
72592780
SW
1365If this is set, then fio will attempt to provide I/O buffer content
1366(on WRITEs) that compresses to the specified level. Fio does this by
1367providing a mix of random data followed by fixed pattern data. The
1368fixed pattern is either zeros, or the pattern specified by
1369\fBbuffer_pattern\fR. If the \fBbuffer_pattern\fR option is used, it
1370might skew the compression ratio slightly. Setting
1371\fBbuffer_compress_percentage\fR to a value other than 100 will also
1372enable \fBrefill_buffers\fR in order to reduce the likelihood that
1373adjacent blocks are so similar that they over compress when seen
1374together. See \fBbuffer_compress_chunk\fR for how to set a finer or
1375coarser granularity of the random/fixed data regions. Defaults to unset
1376i.e., buffer data will not adhere to any compression level.
523bad63
TK
1377.TP
1378.BI buffer_compress_chunk \fR=\fPint
72592780
SW
1379This setting allows fio to manage how big the random/fixed data region
1380is when using \fBbuffer_compress_percentage\fR. When
1381\fBbuffer_compress_chunk\fR is set to some non-zero value smaller than the
1382block size, fio can repeat the random/fixed region throughout the I/O
1383buffer at the specified interval (which particularly useful when
1384bigger block sizes are used for a job). When set to 0, fio will use a
1385chunk size that matches the block size resulting in a single
1386random/fixed region within the I/O buffer. Defaults to 512. When the
1387unit is omitted, the value is interpreted in bytes.
523bad63
TK
1388.TP
1389.BI buffer_pattern \fR=\fPstr
1390If set, fio will fill the I/O buffers with this pattern or with the contents
1391of a file. If not set, the contents of I/O buffers are defined by the other
1392options related to buffer contents. The setting can be any pattern of bytes,
1393and can be prefixed with 0x for hex values. It may also be a string, where
1394the string must then be wrapped with "". Or it may also be a filename,
1395where the filename must be wrapped with '' in which case the file is
1396opened and read. Note that not all the file contents will be read if that
1397would cause the buffers to overflow. So, for example:
1398.RS
1399.RS
1400.P
1401.PD 0
1402buffer_pattern='filename'
1403.P
1404or:
1405.P
1406buffer_pattern="abcd"
1407.P
1408or:
1409.P
1410buffer_pattern=\-12
1411.P
1412or:
1413.P
1414buffer_pattern=0xdeadface
1415.PD
1416.RE
1417.P
1418Also you can combine everything together in any order:
1419.RS
1420.P
1421buffer_pattern=0xdeadface"abcd"\-12'filename'
1422.RE
1423.RE
1424.TP
1425.BI dedupe_percentage \fR=\fPint
1426If set, fio will generate this percentage of identical buffers when
1427writing. These buffers will be naturally dedupable. The contents of the
1428buffers depend on what other buffer compression settings have been set. It's
1429possible to have the individual buffers either fully compressible, or not at
72592780
SW
1430all \-\- this option only controls the distribution of unique buffers. Setting
1431this option will also enable \fBrefill_buffers\fR to prevent every buffer
1432being identical.
523bad63
TK
1433.TP
1434.BI invalidate \fR=\fPbool
1435Invalidate the buffer/page cache parts of the files to be used prior to
1436starting I/O if the platform and file type support it. Defaults to true.
1437This will be ignored if \fBpre_read\fR is also specified for the
1438same job.
1439.TP
1440.BI sync \fR=\fPbool
1441Use synchronous I/O for buffered writes. For the majority of I/O engines,
1442this means using O_SYNC. Default: false.
1443.TP
1444.BI iomem \fR=\fPstr "\fR,\fP mem" \fR=\fPstr
1445Fio can use various types of memory as the I/O unit buffer. The allowed
1446values are:
1447.RS
1448.RS
1449.TP
1450.B malloc
1451Use memory from \fBmalloc\fR\|(3) as the buffers. Default memory type.
1452.TP
1453.B shm
1454Use shared memory as the buffers. Allocated through \fBshmget\fR\|(2).
1455.TP
1456.B shmhuge
1457Same as \fBshm\fR, but use huge pages as backing.
1458.TP
1459.B mmap
1460Use \fBmmap\fR\|(2) to allocate buffers. May either be anonymous memory, or can
1461be file backed if a filename is given after the option. The format
1462is `mem=mmap:/path/to/file'.
1463.TP
1464.B mmaphuge
1465Use a memory mapped huge file as the buffer backing. Append filename
1466after mmaphuge, ala `mem=mmaphuge:/hugetlbfs/file'.
1467.TP
1468.B mmapshared
1469Same as \fBmmap\fR, but use a MMAP_SHARED mapping.
1470.TP
1471.B cudamalloc
1472Use GPU memory as the buffers for GPUDirect RDMA benchmark.
1473The \fBioengine\fR must be \fBrdma\fR.
1474.RE
1475.P
1476The area allocated is a function of the maximum allowed bs size for the job,
1477multiplied by the I/O depth given. Note that for \fBshmhuge\fR and
1478\fBmmaphuge\fR to work, the system must have free huge pages allocated. This
1479can normally be checked and set by reading/writing
1480`/proc/sys/vm/nr_hugepages' on a Linux system. Fio assumes a huge page
1481is 4MiB in size. So to calculate the number of huge pages you need for a
1482given job file, add up the I/O depth of all jobs (normally one unless
1483\fBiodepth\fR is used) and multiply by the maximum bs set. Then divide
1484that number by the huge page size. You can see the size of the huge pages in
338f2db5 1485`/proc/meminfo'. If no huge pages are allocated by having a non-zero
523bad63
TK
1486number in `nr_hugepages', using \fBmmaphuge\fR or \fBshmhuge\fR will fail. Also
1487see \fBhugepage\-size\fR.
1488.P
1489\fBmmaphuge\fR also needs to have hugetlbfs mounted and the file location
1490should point there. So if it's mounted in `/huge', you would use
1491`mem=mmaphuge:/huge/somefile'.
1492.RE
1493.TP
1494.BI iomem_align \fR=\fPint "\fR,\fP mem_align" \fR=\fPint
1495This indicates the memory alignment of the I/O memory buffers. Note that
1496the given alignment is applied to the first I/O unit buffer, if using
1497\fBiodepth\fR the alignment of the following buffers are given by the
1498\fBbs\fR used. In other words, if using a \fBbs\fR that is a
1499multiple of the page sized in the system, all buffers will be aligned to
1500this value. If using a \fBbs\fR that is not page aligned, the alignment
1501of subsequent I/O memory buffers is the sum of the \fBiomem_align\fR and
1502\fBbs\fR used.
1503.TP
1504.BI hugepage\-size \fR=\fPint
1505Defines the size of a huge page. Must at least be equal to the system
1506setting, see `/proc/meminfo'. Defaults to 4MiB. Should probably
1507always be a multiple of megabytes, so using `hugepage\-size=Xm' is the
338f2db5 1508preferred way to set this to avoid setting a non-pow-2 bad value.
523bad63
TK
1509.TP
1510.BI lockmem \fR=\fPint
1511Pin the specified amount of memory with \fBmlock\fR\|(2). Can be used to
1512simulate a smaller amount of memory. The amount specified is per worker.
1513.SS "I/O size"
1514.TP
1515.BI size \fR=\fPint
1516The total size of file I/O for each thread of this job. Fio will run until
1517this many bytes has been transferred, unless runtime is limited by other options
1518(such as \fBruntime\fR, for instance, or increased/decreased by \fBio_size\fR).
1519Fio will divide this size between the available files determined by options
1520such as \fBnrfiles\fR, \fBfilename\fR, unless \fBfilesize\fR is
1521specified by the job. If the result of division happens to be 0, the size is
1522set to the physical size of the given files or devices if they exist.
1523If this option is not specified, fio will use the full size of the given
1524files or devices. If the files do not exist, size must be given. It is also
1525possible to give size as a percentage between 1 and 100. If `size=20%' is
1526given, fio will use 20% of the full size of the given files or devices.
1527Can be combined with \fBoffset\fR to constrain the start and end range
1528that I/O will be done within.
1529.TP
1530.BI io_size \fR=\fPint "\fR,\fB io_limit" \fR=\fPint
1531Normally fio operates within the region set by \fBsize\fR, which means
1532that the \fBsize\fR option sets both the region and size of I/O to be
1533performed. Sometimes that is not what you want. With this option, it is
1534possible to define just the amount of I/O that fio should do. For instance,
1535if \fBsize\fR is set to 20GiB and \fBio_size\fR is set to 5GiB, fio
1536will perform I/O within the first 20GiB but exit when 5GiB have been
1537done. The opposite is also possible \-\- if \fBsize\fR is set to 20GiB,
1538and \fBio_size\fR is set to 40GiB, then fio will do 40GiB of I/O within
1539the 0..20GiB region.
1540.TP
1541.BI filesize \fR=\fPirange(int)
1542Individual file sizes. May be a range, in which case fio will select sizes
1543for files at random within the given range and limited to \fBsize\fR in
1544total (if that is given). If not given, each created file is the same size.
1545This option overrides \fBsize\fR in terms of file size, which means
1546this value is used as a fixed size or possible range of each file.
1547.TP
1548.BI file_append \fR=\fPbool
1549Perform I/O after the end of the file. Normally fio will operate within the
1550size of a file. If this option is set, then fio will append to the file
1551instead. This has identical behavior to setting \fBoffset\fR to the size
338f2db5 1552of a file. This option is ignored on non-regular files.
523bad63
TK
1553.TP
1554.BI fill_device \fR=\fPbool "\fR,\fB fill_fs" \fR=\fPbool
1555Sets size to something really large and waits for ENOSPC (no space left on
1556device) as the terminating condition. Only makes sense with sequential
1557write. For a read workload, the mount point will be filled first then I/O
1558started on the result. This option doesn't make sense if operating on a raw
1559device node, since the size of that is already known by the file system.
338f2db5 1560Additionally, writing beyond end-of-device will not return ENOSPC there.
523bad63
TK
1561.SS "I/O engine"
1562.TP
1563.BI ioengine \fR=\fPstr
1564Defines how the job issues I/O to the file. The following types are defined:
1565.RS
1566.RS
1567.TP
1568.B sync
1569Basic \fBread\fR\|(2) or \fBwrite\fR\|(2)
1570I/O. \fBlseek\fR\|(2) is used to position the I/O location.
1571See \fBfsync\fR and \fBfdatasync\fR for syncing write I/Os.
1572.TP
1573.B psync
1574Basic \fBpread\fR\|(2) or \fBpwrite\fR\|(2) I/O. Default on
1575all supported operating systems except for Windows.
1576.TP
1577.B vsync
1578Basic \fBreadv\fR\|(2) or \fBwritev\fR\|(2) I/O. Will emulate
1579queuing by coalescing adjacent I/Os into a single submission.
1580.TP
1581.B pvsync
1582Basic \fBpreadv\fR\|(2) or \fBpwritev\fR\|(2) I/O.
a46c5e01 1583.TP
2cafffbe
JA
1584.B pvsync2
1585Basic \fBpreadv2\fR\|(2) or \fBpwritev2\fR\|(2) I/O.
1586.TP
d60e92d1 1587.B libaio
523bad63 1588Linux native asynchronous I/O. Note that Linux may only support
338f2db5 1589queued behavior with non-buffered I/O (set `direct=1' or
523bad63
TK
1590`buffered=0').
1591This engine defines engine specific options.
d60e92d1
AC
1592.TP
1593.B posixaio
523bad63
TK
1594POSIX asynchronous I/O using \fBaio_read\fR\|(3) and
1595\fBaio_write\fR\|(3).
03e20d68
BC
1596.TP
1597.B solarisaio
1598Solaris native asynchronous I/O.
1599.TP
1600.B windowsaio
38f8c318 1601Windows native asynchronous I/O. Default on Windows.
d60e92d1
AC
1602.TP
1603.B mmap
523bad63
TK
1604File is memory mapped with \fBmmap\fR\|(2) and data copied
1605to/from using \fBmemcpy\fR\|(3).
d60e92d1
AC
1606.TP
1607.B splice
523bad63
TK
1608\fBsplice\fR\|(2) is used to transfer the data and
1609\fBvmsplice\fR\|(2) to transfer data from user space to the
1610kernel.
d60e92d1 1611.TP
d60e92d1 1612.B sg
523bad63
TK
1613SCSI generic sg v3 I/O. May either be synchronous using the SG_IO
1614ioctl, or if the target is an sg character device we use
1615\fBread\fR\|(2) and \fBwrite\fR\|(2) for asynchronous
1616I/O. Requires \fBfilename\fR option to specify either block or
3740cfc8
VF
1617character devices. This engine supports trim operations. The
1618sg engine includes engine specific options.
d60e92d1
AC
1619.TP
1620.B null
523bad63
TK
1621Doesn't transfer any data, just pretends to. This is mainly used to
1622exercise fio itself and for debugging/testing purposes.
d60e92d1
AC
1623.TP
1624.B net
523bad63
TK
1625Transfer over the network to given `host:port'. Depending on the
1626\fBprotocol\fR used, the \fBhostname\fR, \fBport\fR,
1627\fBlisten\fR and \fBfilename\fR options are used to specify
1628what sort of connection to make, while the \fBprotocol\fR option
1629determines which protocol will be used. This engine defines engine
1630specific options.
d60e92d1
AC
1631.TP
1632.B netsplice
523bad63
TK
1633Like \fBnet\fR, but uses \fBsplice\fR\|(2) and
1634\fBvmsplice\fR\|(2) to map data and send/receive.
1635This engine defines engine specific options.
d60e92d1 1636.TP
53aec0a4 1637.B cpuio
523bad63
TK
1638Doesn't transfer any data, but burns CPU cycles according to the
1639\fBcpuload\fR and \fBcpuchunks\fR options. Setting
1640\fBcpuload\fR\=85 will cause that job to do nothing but burn 85%
1641of the CPU. In case of SMP machines, use `numjobs=<nr_of_cpu>'
1642to get desired CPU usage, as the cpuload only loads a
1643single CPU at the desired rate. A job never finishes unless there is
338f2db5 1644at least one non-cpuio job.
d60e92d1
AC
1645.TP
1646.B guasi
f1dd3fb1 1647The GUASI I/O engine is the Generic Userspace Asynchronous Syscall
338f2db5 1648Interface approach to async I/O. See \fIhttp://www.xmailserver.org/guasi-lib.html\fR
523bad63 1649for more info on GUASI.
d60e92d1 1650.TP
21b8aee8 1651.B rdma
523bad63
TK
1652The RDMA I/O engine supports both RDMA memory semantics
1653(RDMA_WRITE/RDMA_READ) and channel semantics (Send/Recv) for the
609ac152
SB
1654InfiniBand, RoCE and iWARP protocols. This engine defines engine
1655specific options.
d54fce84
DM
1656.TP
1657.B falloc
523bad63
TK
1658I/O engine that does regular fallocate to simulate data transfer as
1659fio ioengine.
1660.RS
1661.P
1662.PD 0
1663DDIR_READ does fallocate(,mode = FALLOC_FL_KEEP_SIZE,).
1664.P
1665DIR_WRITE does fallocate(,mode = 0).
1666.P
1667DDIR_TRIM does fallocate(,mode = FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE).
1668.PD
1669.RE
1670.TP
1671.B ftruncate
1672I/O engine that sends \fBftruncate\fR\|(2) operations in response
1673to write (DDIR_WRITE) events. Each ftruncate issued sets the file's
1674size to the current block offset. \fBblocksize\fR is ignored.
d54fce84
DM
1675.TP
1676.B e4defrag
523bad63
TK
1677I/O engine that does regular EXT4_IOC_MOVE_EXT ioctls to simulate
1678defragment activity in request to DDIR_WRITE event.
0d978694 1679.TP
d5f9b0ea
IF
1680.B rados
1681I/O engine supporting direct access to Ceph Reliable Autonomic Distributed
1682Object Store (RADOS) via librados. This ioengine defines engine specific
1683options.
1684.TP
0d978694 1685.B rbd
523bad63
TK
1686I/O engine supporting direct access to Ceph Rados Block Devices
1687(RBD) via librbd without the need to use the kernel rbd driver. This
1688ioengine defines engine specific options.
a7c386f4 1689.TP
c2f6a13d
LMB
1690.B http
1691I/O engine supporting GET/PUT requests over HTTP(S) with libcurl to
1692a WebDAV or S3 endpoint. This ioengine defines engine specific options.
1693
1694This engine only supports direct IO of iodepth=1; you need to scale this
1695via numjobs. blocksize defines the size of the objects to be created.
1696
1697TRIM is translated to object deletion.
1698.TP
a7c386f4 1699.B gfapi
523bad63
TK
1700Using GlusterFS libgfapi sync interface to direct access to
1701GlusterFS volumes without having to go through FUSE. This ioengine
1702defines engine specific options.
cc47f094 1703.TP
1704.B gfapi_async
523bad63
TK
1705Using GlusterFS libgfapi async interface to direct access to
1706GlusterFS volumes without having to go through FUSE. This ioengine
1707defines engine specific options.
1b10477b 1708.TP
b74e419e 1709.B libhdfs
523bad63
TK
1710Read and write through Hadoop (HDFS). The \fBfilename\fR option
1711is used to specify host,port of the hdfs name\-node to connect. This
1712engine interprets offsets a little differently. In HDFS, files once
1713created cannot be modified so random writes are not possible. To
1714imitate this the libhdfs engine expects a bunch of small files to be
1715created over HDFS and will randomly pick a file from them
1716based on the offset generated by fio backend (see the example
1717job file to create such files, use `rw=write' option). Please
1718note, it may be necessary to set environment variables to work
1719with HDFS/libhdfs properly. Each job uses its own connection to
1720HDFS.
65fa28ca
DE
1721.TP
1722.B mtd
523bad63
TK
1723Read, write and erase an MTD character device (e.g.,
1724`/dev/mtd0'). Discards are treated as erases. Depending on the
1725underlying device type, the I/O may have to go in a certain pattern,
1726e.g., on NAND, writing sequentially to erase blocks and discarding
1727before overwriting. The \fBtrimwrite\fR mode works well for this
65fa28ca 1728constraint.
5c4ef02e
JA
1729.TP
1730.B pmemblk
523bad63 1731Read and write using filesystem DAX to a file on a filesystem
363a5f65 1732mounted with DAX on a persistent memory device through the PMDK
523bad63 1733libpmemblk library.
104ee4de 1734.TP
523bad63
TK
1735.B dev\-dax
1736Read and write using device DAX to a persistent memory device (e.g.,
363a5f65 1737/dev/dax0.0) through the PMDK libpmem library.
d60e92d1 1738.TP
523bad63
TK
1739.B external
1740Prefix to specify loading an external I/O engine object file. Append
1741the engine filename, e.g. `ioengine=external:/tmp/foo.o' to load
d243fd6d
TK
1742ioengine `foo.o' in `/tmp'. The path can be either
1743absolute or relative. See `engines/skeleton_external.c' in the fio source for
1744details of writing an external I/O engine.
1216cc5a
JB
1745.TP
1746.B filecreate
b71968b1
SW
1747Simply create the files and do no I/O to them. You still need to set
1748\fBfilesize\fR so that all the accounting still occurs, but no actual I/O will be
1749done other than creating the file.
ae0db592
TI
1750.TP
1751.B libpmem
1752Read and write using mmap I/O to a file on a filesystem
363a5f65 1753mounted with DAX on a persistent memory device through the PMDK
ae0db592 1754libpmem library.
07751e10
JA
1755.TP
1756.B ime_psync
1757Synchronous read and write using DDN's Infinite Memory Engine (IME). This
1758engine is very basic and issues calls to IME whenever an IO is queued.
1759.TP
1760.B ime_psyncv
1761Synchronous read and write using DDN's Infinite Memory Engine (IME). This
1762engine uses iovecs and will try to stack as much IOs as possible (if the IOs
1763are "contiguous" and the IO depth is not exceeded) before issuing a call to IME.
1764.TP
1765.B ime_aio
1766Asynchronous read and write using DDN's Infinite Memory Engine (IME). This
1767engine will try to stack as much IOs as possible by creating requests for IME.
1768FIO will then decide when to commit these requests.
247ef2aa
KZ
1769.TP
1770.B libiscsi
1771Read and write iscsi lun with libiscsi.
d643a1e2
RJ
1772.TP
1773.B nbd
1774Synchronous read and write a Network Block Device (NBD).
523bad63
TK
1775.SS "I/O engine specific parameters"
1776In addition, there are some parameters which are only valid when a specific
1777\fBioengine\fR is in use. These are used identically to normal parameters,
1778with the caveat that when used on the command line, they must come after the
1779\fBioengine\fR that defines them is selected.
d60e92d1 1780.TP
029b42ac
JA
1781.BI (io_uring)hipri
1782If this option is set, fio will attempt to use polled IO completions. Normal IO
1783completions generate interrupts to signal the completion of IO, polled
1784completions do not. Hence they are require active reaping by the application.
1785The benefits are more efficient IO for high IOPS scenarios, and lower latencies
1786for low queue depth IO.
1787.TP
1788.BI (io_uring)fixedbufs
1789If fio is asked to do direct IO, then Linux will map pages for each IO call, and
1790release them when IO is done. If this option is set, the pages are pre-mapped
1791before IO is started. This eliminates the need to map and release for each IO.
1792This is more efficient, and reduces the IO latency as well.
1793.TP
5ffd5626
JA
1794.BI (io_uring)registerfiles
1795With this option, fio registers the set of files being used with the kernel.
1796This avoids the overhead of managing file counts in the kernel, making the
1797submission and completion part more lightweight. Required for the below
1798sqthread_poll option.
1799.TP
029b42ac
JA
1800.BI (io_uring)sqthread_poll
1801Normally fio will submit IO by issuing a system call to notify the kernel of
1802available items in the SQ ring. If this option is set, the act of submitting IO
1803will be done by a polling thread in the kernel. This frees up cycles for fio, at
1804the cost of using more CPU in the system.
1805.TP
1806.BI (io_uring)sqthread_poll_cpu
1807When `sqthread_poll` is set, this option provides a way to define which CPU
1808should be used for the polling thread.
1809.TP
523bad63
TK
1810.BI (libaio)userspace_reap
1811Normally, with the libaio engine in use, fio will use the
1812\fBio_getevents\fR\|(3) system call to reap newly returned events. With
338f2db5 1813this flag turned on, the AIO ring will be read directly from user-space to
523bad63
TK
1814reap events. The reaping mode is only enabled when polling for a minimum of
18150 events (e.g. when `iodepth_batch_complete=0').
3ce9dcaf 1816.TP
523bad63
TK
1817.BI (pvsync2)hipri
1818Set RWF_HIPRI on I/O, indicating to the kernel that it's of higher priority
1819than normal.
82407585 1820.TP
523bad63
TK
1821.BI (pvsync2)hipri_percentage
1822When hipri is set this determines the probability of a pvsync2 I/O being high
1823priority. The default is 100%.
d60e92d1 1824.TP
523bad63
TK
1825.BI (cpuio)cpuload \fR=\fPint
1826Attempt to use the specified percentage of CPU cycles. This is a mandatory
1827option when using cpuio I/O engine.
997b5680 1828.TP
523bad63
TK
1829.BI (cpuio)cpuchunks \fR=\fPint
1830Split the load into cycles of the given time. In microseconds.
1ad01bd1 1831.TP
523bad63
TK
1832.BI (cpuio)exit_on_io_done \fR=\fPbool
1833Detect when I/O threads are done, then exit.
d60e92d1 1834.TP
523bad63
TK
1835.BI (libhdfs)namenode \fR=\fPstr
1836The hostname or IP address of a HDFS cluster namenode to contact.
d01612f3 1837.TP
523bad63
TK
1838.BI (libhdfs)port
1839The listening port of the HFDS cluster namenode.
d60e92d1 1840.TP
523bad63
TK
1841.BI (netsplice,net)port
1842The TCP or UDP port to bind to or connect to. If this is used with
1843\fBnumjobs\fR to spawn multiple instances of the same job type, then
1844this will be the starting port number since fio will use a range of
1845ports.
d60e92d1 1846.TP
609ac152
SB
1847.BI (rdma)port
1848The port to use for RDMA-CM communication. This should be the same
1849value on the client and the server side.
1850.TP
1851.BI (netsplice,net, rdma)hostname \fR=\fPstr
1852The hostname or IP address to use for TCP, UDP or RDMA-CM based I/O.
1853If the job is a TCP listener or UDP reader, the hostname is not used
1854and must be omitted unless it is a valid UDP multicast address.
591e9e06 1855.TP
523bad63
TK
1856.BI (netsplice,net)interface \fR=\fPstr
1857The IP address of the network interface used to send or receive UDP
1858multicast.
ddf24e42 1859.TP
523bad63
TK
1860.BI (netsplice,net)ttl \fR=\fPint
1861Time\-to\-live value for outgoing UDP multicast packets. Default: 1.
d60e92d1 1862.TP
523bad63
TK
1863.BI (netsplice,net)nodelay \fR=\fPbool
1864Set TCP_NODELAY on TCP connections.
fa769d44 1865.TP
523bad63
TK
1866.BI (netsplice,net)protocol \fR=\fPstr "\fR,\fP proto" \fR=\fPstr
1867The network protocol to use. Accepted values are:
1868.RS
e76b1da4
JA
1869.RS
1870.TP
523bad63
TK
1871.B tcp
1872Transmission control protocol.
e76b1da4 1873.TP
523bad63
TK
1874.B tcpv6
1875Transmission control protocol V6.
e76b1da4 1876.TP
523bad63
TK
1877.B udp
1878User datagram protocol.
1879.TP
1880.B udpv6
1881User datagram protocol V6.
e76b1da4 1882.TP
523bad63
TK
1883.B unix
1884UNIX domain socket.
e76b1da4
JA
1885.RE
1886.P
523bad63
TK
1887When the protocol is TCP or UDP, the port must also be given, as well as the
1888hostname if the job is a TCP listener or UDP reader. For unix sockets, the
1889normal \fBfilename\fR option should be used and the port is invalid.
1890.RE
1891.TP
1892.BI (netsplice,net)listen
1893For TCP network connections, tell fio to listen for incoming connections
1894rather than initiating an outgoing connection. The \fBhostname\fR must
1895be omitted if this option is used.
1896.TP
1897.BI (netsplice,net)pingpong
1898Normally a network writer will just continue writing data, and a network
1899reader will just consume packages. If `pingpong=1' is set, a writer will
1900send its normal payload to the reader, then wait for the reader to send the
1901same payload back. This allows fio to measure network latencies. The
1902submission and completion latencies then measure local time spent sending or
1903receiving, and the completion latency measures how long it took for the
1904other end to receive and send back. For UDP multicast traffic
1905`pingpong=1' should only be set for a single reader when multiple readers
1906are listening to the same address.
1907.TP
1908.BI (netsplice,net)window_size \fR=\fPint
1909Set the desired socket buffer size for the connection.
e76b1da4 1910.TP
523bad63
TK
1911.BI (netsplice,net)mss \fR=\fPint
1912Set the TCP maximum segment size (TCP_MAXSEG).
d60e92d1 1913.TP
523bad63
TK
1914.BI (e4defrag)donorname \fR=\fPstr
1915File will be used as a block donor (swap extents between files).
d60e92d1 1916.TP
523bad63
TK
1917.BI (e4defrag)inplace \fR=\fPint
1918Configure donor file blocks allocation strategy:
1919.RS
1920.RS
d60e92d1 1921.TP
523bad63
TK
1922.B 0
1923Default. Preallocate donor's file on init.
d60e92d1 1924.TP
523bad63
TK
1925.B 1
1926Allocate space immediately inside defragment event, and free right
1927after event.
1928.RE
1929.RE
d60e92d1 1930.TP
d5f9b0ea 1931.BI (rbd,rados)clustername \fR=\fPstr
523bad63 1932Specifies the name of the Ceph cluster.
92d42d69 1933.TP
523bad63
TK
1934.BI (rbd)rbdname \fR=\fPstr
1935Specifies the name of the RBD.
92d42d69 1936.TP
d5f9b0ea
IF
1937.BI (rbd,rados)pool \fR=\fPstr
1938Specifies the name of the Ceph pool containing RBD or RADOS data.
92d42d69 1939.TP
d5f9b0ea 1940.BI (rbd,rados)clientname \fR=\fPstr
523bad63
TK
1941Specifies the username (without the 'client.' prefix) used to access the
1942Ceph cluster. If the \fBclustername\fR is specified, the \fBclientname\fR shall be
1943the full *type.id* string. If no type. prefix is given, fio will add 'client.'
1944by default.
92d42d69 1945.TP
d5f9b0ea
IF
1946.BI (rbd,rados)busy_poll \fR=\fPbool
1947Poll store instead of waiting for completion. Usually this provides better
1948throughput at cost of higher(up to 100%) CPU utilization.
1949.TP
c2f6a13d
LMB
1950.BI (http)http_host \fR=\fPstr
1951Hostname to connect to. For S3, this could be the bucket name. Default
1952is \fBlocalhost\fR
1953.TP
1954.BI (http)http_user \fR=\fPstr
1955Username for HTTP authentication.
1956.TP
1957.BI (http)http_pass \fR=\fPstr
1958Password for HTTP authentication.
1959.TP
09fd2966
LMB
1960.BI (http)https \fR=\fPstr
1961Whether to use HTTPS instead of plain HTTP. \fRon\fP enables HTTPS;
1962\fRinsecure\fP will enable HTTPS, but disable SSL peer verification (use
1963with caution!). Default is \fBoff\fR.
c2f6a13d 1964.TP
09fd2966
LMB
1965.BI (http)http_mode \fR=\fPstr
1966Which HTTP access mode to use: webdav, swift, or s3. Default is
1967\fBwebdav\fR.
c2f6a13d
LMB
1968.TP
1969.BI (http)http_s3_region \fR=\fPstr
1970The S3 region/zone to include in the request. Default is \fBus-east-1\fR.
1971.TP
1972.BI (http)http_s3_key \fR=\fPstr
1973The S3 secret key.
1974.TP
1975.BI (http)http_s3_keyid \fR=\fPstr
1976The S3 key/access id.
1977.TP
09fd2966
LMB
1978.BI (http)http_swift_auth_token \fR=\fPstr
1979The Swift auth token. See the example configuration file on how to
1980retrieve this.
1981.TP
c2f6a13d
LMB
1982.BI (http)http_verbose \fR=\fPint
1983Enable verbose requests from libcurl. Useful for debugging. 1 turns on
1984verbose logging from libcurl, 2 additionally enables HTTP IO tracing.
1985Default is \fB0\fR
1986.TP
523bad63
TK
1987.BI (mtd)skip_bad \fR=\fPbool
1988Skip operations against known bad blocks.
8116fd24 1989.TP
523bad63
TK
1990.BI (libhdfs)hdfsdirectory
1991libhdfs will create chunk in this HDFS directory.
e0a04ac1 1992.TP
523bad63
TK
1993.BI (libhdfs)chunk_size
1994The size of the chunk to use for each file.
609ac152
SB
1995.TP
1996.BI (rdma)verb \fR=\fPstr
1997The RDMA verb to use on this side of the RDMA ioengine
1998connection. Valid values are write, read, send and recv. These
1999correspond to the equivalent RDMA verbs (e.g. write = rdma_write
2000etc.). Note that this only needs to be specified on the client side of
2001the connection. See the examples folder.
2002.TP
2003.BI (rdma)bindname \fR=\fPstr
2004The name to use to bind the local RDMA-CM connection to a local RDMA
2005device. This could be a hostname or an IPv4 or IPv6 address. On the
2006server side this will be passed into the rdma_bind_addr() function and
2007on the client site it will be used in the rdma_resolve_add()
2008function. This can be useful when multiple paths exist between the
2009client and the server or in certain loopback configurations.
52b81b7c
KD
2010.TP
2011.BI (sg)readfua \fR=\fPbool
2012With readfua option set to 1, read operations include the force
2013unit access (fua) flag. Default: 0.
2014.TP
2015.BI (sg)writefua \fR=\fPbool
2016With writefua option set to 1, write operations include the force
2017unit access (fua) flag. Default: 0.
2c3a9150
VF
2018.TP
2019.BI (sg)sg_write_mode \fR=\fPstr
2020Specify the type of write commands to issue. This option can take three
2021values:
2022.RS
2023.RS
2024.TP
2025.B write (default)
2026Write opcodes are issued as usual
2027.TP
2028.B verify
2029Issue WRITE AND VERIFY commands. The BYTCHK bit is set to 0. This
2030directs the device to carry out a medium verification with no data
2031comparison. The writefua option is ignored with this selection.
2032.TP
2033.B same
2034Issue WRITE SAME commands. This transfers a single block to the device
2035and writes this same block of data to a contiguous sequence of LBAs
2036beginning at the specified offset. fio's block size parameter
2037specifies the amount of data written with each command. However, the
2038amount of data actually transferred to the device is equal to the
2039device's block (sector) size. For a device with 512 byte sectors,
2040blocksize=8k will write 16 sectors with each command. fio will still
2041generate 8k of data for each command butonly the first 512 bytes will
2042be used and transferred to the device. The writefua option is ignored
2043with this selection.
f2d6de5d
RJ
2044.RE
2045.RE
2046.TP
2047.BI (nbd)uri \fR=\fPstr
2048Specify the NBD URI of the server to test.
2049The string is a standard NBD URI (see
2050\fIhttps://github.com/NetworkBlockDevice/nbd/tree/master/doc\fR).
2051Example URIs:
2052.RS
2053.RS
2054.TP
2055\fInbd://localhost:10809\fR
2056.TP
2057\fInbd+unix:///?socket=/tmp/socket\fR
2058.TP
2059\fInbds://tlshost/exportname\fR
2c3a9150 2060
523bad63
TK
2061.SS "I/O depth"
2062.TP
2063.BI iodepth \fR=\fPint
2064Number of I/O units to keep in flight against the file. Note that
2065increasing \fBiodepth\fR beyond 1 will not affect synchronous ioengines (except
2066for small degrees when \fBverify_async\fR is in use). Even async
2067engines may impose OS restrictions causing the desired depth not to be
2068achieved. This may happen on Linux when using libaio and not setting
2069`direct=1', since buffered I/O is not async on that OS. Keep an
2070eye on the I/O depth distribution in the fio output to verify that the
2071achieved depth is as expected. Default: 1.
2072.TP
2073.BI iodepth_batch_submit \fR=\fPint "\fR,\fP iodepth_batch" \fR=\fPint
2074This defines how many pieces of I/O to submit at once. It defaults to 1
2075which means that we submit each I/O as soon as it is available, but can be
2076raised to submit bigger batches of I/O at the time. If it is set to 0 the
2077\fBiodepth\fR value will be used.
2078.TP
2079.BI iodepth_batch_complete_min \fR=\fPint "\fR,\fP iodepth_batch_complete" \fR=\fPint
2080This defines how many pieces of I/O to retrieve at once. It defaults to 1
2081which means that we'll ask for a minimum of 1 I/O in the retrieval process
2082from the kernel. The I/O retrieval will go on until we hit the limit set by
2083\fBiodepth_low\fR. If this variable is set to 0, then fio will always
2084check for completed events before queuing more I/O. This helps reduce I/O
2085latency, at the cost of more retrieval system calls.
2086.TP
2087.BI iodepth_batch_complete_max \fR=\fPint
2088This defines maximum pieces of I/O to retrieve at once. This variable should
2089be used along with \fBiodepth_batch_complete_min\fR=\fIint\fR variable,
2090specifying the range of min and max amount of I/O which should be
2091retrieved. By default it is equal to \fBiodepth_batch_complete_min\fR
2092value. Example #1:
e0a04ac1 2093.RS
e0a04ac1 2094.RS
e0a04ac1 2095.P
523bad63
TK
2096.PD 0
2097iodepth_batch_complete_min=1
e0a04ac1 2098.P
523bad63
TK
2099iodepth_batch_complete_max=<iodepth>
2100.PD
e0a04ac1
JA
2101.RE
2102.P
523bad63
TK
2103which means that we will retrieve at least 1 I/O and up to the whole
2104submitted queue depth. If none of I/O has been completed yet, we will wait.
2105Example #2:
e8b1961d 2106.RS
523bad63
TK
2107.P
2108.PD 0
2109iodepth_batch_complete_min=0
2110.P
2111iodepth_batch_complete_max=<iodepth>
2112.PD
e8b1961d
JA
2113.RE
2114.P
523bad63
TK
2115which means that we can retrieve up to the whole submitted queue depth, but
2116if none of I/O has been completed yet, we will NOT wait and immediately exit
2117the system call. In this example we simply do polling.
2118.RE
e8b1961d 2119.TP
523bad63
TK
2120.BI iodepth_low \fR=\fPint
2121The low water mark indicating when to start filling the queue
2122again. Defaults to the same as \fBiodepth\fR, meaning that fio will
2123attempt to keep the queue full at all times. If \fBiodepth\fR is set to
2124e.g. 16 and \fBiodepth_low\fR is set to 4, then after fio has filled the queue of
212516 requests, it will let the depth drain down to 4 before starting to fill
2126it again.
d60e92d1 2127.TP
523bad63
TK
2128.BI serialize_overlap \fR=\fPbool
2129Serialize in-flight I/Os that might otherwise cause or suffer from data races.
2130When two or more I/Os are submitted simultaneously, there is no guarantee that
2131the I/Os will be processed or completed in the submitted order. Further, if
2132two or more of those I/Os are writes, any overlapping region between them can
2133become indeterminate/undefined on certain storage. These issues can cause
2134verification to fail erratically when at least one of the racing I/Os is
2135changing data and the overlapping region has a non-zero size. Setting
2136\fBserialize_overlap\fR tells fio to avoid provoking this behavior by explicitly
2137serializing in-flight I/Os that have a non-zero overlap. Note that setting
2138this option can reduce both performance and the \fBiodepth\fR achieved.
3d6a6f04
VF
2139.RS
2140.P
2141This option only applies to I/Os issued for a single job except when it is
2142enabled along with \fBio_submit_mode\fR=offload. In offload mode, fio
2143will check for overlap among all I/Os submitted by offload jobs with \fBserialize_overlap\fR
307f2246 2144enabled.
3d6a6f04
VF
2145.P
2146Default: false.
2147.RE
d60e92d1 2148.TP
523bad63
TK
2149.BI io_submit_mode \fR=\fPstr
2150This option controls how fio submits the I/O to the I/O engine. The default
2151is `inline', which means that the fio job threads submit and reap I/O
2152directly. If set to `offload', the job threads will offload I/O submission
2153to a dedicated pool of I/O threads. This requires some coordination and thus
2154has a bit of extra overhead, especially for lower queue depth I/O where it
2155can increase latencies. The benefit is that fio can manage submission rates
2156independently of the device completion rates. This avoids skewed latency
2157reporting if I/O gets backed up on the device side (the coordinated omission
2158problem).
2159.SS "I/O rate"
d60e92d1 2160.TP
523bad63
TK
2161.BI thinktime \fR=\fPtime
2162Stall the job for the specified period of time after an I/O has completed before issuing the
2163next. May be used to simulate processing being done by an application.
2164When the unit is omitted, the value is interpreted in microseconds. See
2165\fBthinktime_blocks\fR and \fBthinktime_spin\fR.
d60e92d1 2166.TP
523bad63 2167.BI thinktime_spin \fR=\fPtime
338f2db5 2168Only valid if \fBthinktime\fR is set - pretend to spend CPU time doing
523bad63
TK
2169something with the data received, before falling back to sleeping for the
2170rest of the period specified by \fBthinktime\fR. When the unit is
2171omitted, the value is interpreted in microseconds.
d60e92d1
AC
2172.TP
2173.BI thinktime_blocks \fR=\fPint
338f2db5 2174Only valid if \fBthinktime\fR is set - control how many blocks to issue,
523bad63
TK
2175before waiting \fBthinktime\fR usecs. If not set, defaults to 1 which will make
2176fio wait \fBthinktime\fR usecs after every block. This effectively makes any
2177queue depth setting redundant, since no more than 1 I/O will be queued
2178before we have to complete it and do our \fBthinktime\fR. In other words, this
2179setting effectively caps the queue depth if the latter is larger.
d60e92d1 2180.TP
6d500c2e 2181.BI rate \fR=\fPint[,int][,int]
523bad63 2182Cap the bandwidth used by this job. The number is in bytes/sec, the normal
338f2db5 2183suffix rules apply. Comma-separated values may be specified for reads,
523bad63
TK
2184writes, and trims as described in \fBblocksize\fR.
2185.RS
2186.P
2187For example, using `rate=1m,500k' would limit reads to 1MiB/sec and writes to
2188500KiB/sec. Capping only reads or writes can be done with `rate=,500k' or
2189`rate=500k,' where the former will only limit writes (to 500KiB/sec) and the
2190latter will only limit reads.
2191.RE
d60e92d1 2192.TP
6d500c2e 2193.BI rate_min \fR=\fPint[,int][,int]
523bad63 2194Tell fio to do whatever it can to maintain at least this bandwidth. Failing
338f2db5 2195to meet this requirement will cause the job to exit. Comma-separated values
523bad63
TK
2196may be specified for reads, writes, and trims as described in
2197\fBblocksize\fR.
d60e92d1 2198.TP
6d500c2e 2199.BI rate_iops \fR=\fPint[,int][,int]
523bad63
TK
2200Cap the bandwidth to this number of IOPS. Basically the same as
2201\fBrate\fR, just specified independently of bandwidth. If the job is
2202given a block size range instead of a fixed value, the smallest block size
338f2db5 2203is used as the metric. Comma-separated values may be specified for reads,
523bad63 2204writes, and trims as described in \fBblocksize\fR.
d60e92d1 2205.TP
6d500c2e 2206.BI rate_iops_min \fR=\fPint[,int][,int]
523bad63 2207If fio doesn't meet this rate of I/O, it will cause the job to exit.
338f2db5 2208Comma-separated values may be specified for reads, writes, and trims as
523bad63 2209described in \fBblocksize\fR.
d60e92d1 2210.TP
6de65959 2211.BI rate_process \fR=\fPstr
523bad63
TK
2212This option controls how fio manages rated I/O submissions. The default is
2213`linear', which submits I/O in a linear fashion with fixed delays between
2214I/Os that gets adjusted based on I/O completion rates. If this is set to
2215`poisson', fio will submit I/O based on a more real world random request
6de65959 2216flow, known as the Poisson process
523bad63 2217(\fIhttps://en.wikipedia.org/wiki/Poisson_point_process\fR). The lambda will be
5d02b083 221810^6 / IOPS for the given workload.
1a9bf814
JA
2219.TP
2220.BI rate_ignore_thinktime \fR=\fPbool
2221By default, fio will attempt to catch up to the specified rate setting, if any
2222kind of thinktime setting was used. If this option is set, then fio will
2223ignore the thinktime and continue doing IO at the specified rate, instead of
2224entering a catch-up mode after thinktime is done.
523bad63 2225.SS "I/O latency"
ff6bb260 2226.TP
523bad63 2227.BI latency_target \fR=\fPtime
3e260a46 2228If set, fio will attempt to find the max performance point that the given
523bad63
TK
2229workload will run at while maintaining a latency below this target. When
2230the unit is omitted, the value is interpreted in microseconds. See
2231\fBlatency_window\fR and \fBlatency_percentile\fR.
3e260a46 2232.TP
523bad63 2233.BI latency_window \fR=\fPtime
3e260a46 2234Used with \fBlatency_target\fR to specify the sample window that the job
523bad63
TK
2235is run at varying queue depths to test the performance. When the unit is
2236omitted, the value is interpreted in microseconds.
3e260a46
JA
2237.TP
2238.BI latency_percentile \fR=\fPfloat
523bad63
TK
2239The percentage of I/Os that must fall within the criteria specified by
2240\fBlatency_target\fR and \fBlatency_window\fR. If not set, this
2241defaults to 100.0, meaning that all I/Os must be equal or below to the value
2242set by \fBlatency_target\fR.
2243.TP
2244.BI max_latency \fR=\fPtime
2245If set, fio will exit the job with an ETIMEDOUT error if it exceeds this
2246maximum latency. When the unit is omitted, the value is interpreted in
2247microseconds.
2248.TP
2249.BI rate_cycle \fR=\fPint
2250Average bandwidth for \fBrate\fR and \fBrate_min\fR over this number
2251of milliseconds. Defaults to 1000.
2252.SS "I/O replay"
2253.TP
2254.BI write_iolog \fR=\fPstr
2255Write the issued I/O patterns to the specified file. See
2256\fBread_iolog\fR. Specify a separate file for each job, otherwise the
2257iologs will be interspersed and the file may be corrupt.
2258.TP
2259.BI read_iolog \fR=\fPstr
2260Open an iolog with the specified filename and replay the I/O patterns it
2261contains. This can be used to store a workload and replay it sometime
2262later. The iolog given may also be a blktrace binary file, which allows fio
2263to replay a workload captured by blktrace. See
2264\fBblktrace\fR\|(8) for how to capture such logging data. For blktrace
2265replay, the file needs to be turned into a blkparse binary data file first
2266(`blkparse <device> \-o /dev/null \-d file_for_fio.bin').
c70c7f58
AK
2267You can specify a number of files by separating the names with a ':' character.
2268See the \fBfilename\fR option for information on how to escape ':' and '\'
2269characters within the file names. These files will be sequentially assigned to
2270job clones created by \fBnumjobs\fR.
3e260a46 2271.TP
98e7161c
AK
2272.BI read_iolog_chunked \fR=\fPbool
2273Determines how iolog is read. If false (default) entire \fBread_iolog\fR will
2274be read at once. If selected true, input from iolog will be read gradually.
2275Useful when iolog is very large, or it is generated.
2276.TP
b9921d1a
DZ
2277.BI merge_blktrace_file \fR=\fPstr
2278When specified, rather than replaying the logs passed to \fBread_iolog\fR,
2279the logs go through a merge phase which aggregates them into a single blktrace.
2280The resulting file is then passed on as the \fBread_iolog\fR parameter. The
2281intention here is to make the order of events consistent. This limits the
2282influence of the scheduler compared to replaying multiple blktraces via
2283concurrent jobs.
2284.TP
87a48ada
DZ
2285.BI merge_blktrace_scalars \fR=\fPfloat_list
2286This is a percentage based option that is index paired with the list of files
2287passed to \fBread_iolog\fR. When merging is performed, scale the time of each
2288event by the corresponding amount. For example,
2289`\-\-merge_blktrace_scalars="50:100"' runs the first trace in halftime and the
2290second trace in realtime. This knob is separately tunable from
2291\fBreplay_time_scale\fR which scales the trace during runtime and will not
2292change the output of the merge unlike this option.
2293.TP
55bfd8c8
DZ
2294.BI merge_blktrace_iters \fR=\fPfloat_list
2295This is a whole number option that is index paired with the list of files
2296passed to \fBread_iolog\fR. When merging is performed, run each trace for
2297the specified number of iterations. For example,
2298`\-\-merge_blktrace_iters="2:1"' runs the first trace for two iterations
2299and the second trace for one iteration.
2300.TP
523bad63
TK
2301.BI replay_no_stall \fR=\fPbool
2302When replaying I/O with \fBread_iolog\fR the default behavior is to
2303attempt to respect the timestamps within the log and replay them with the
2304appropriate delay between IOPS. By setting this variable fio will not
2305respect the timestamps and attempt to replay them as fast as possible while
2306still respecting ordering. The result is the same I/O pattern to a given
2307device, but different timings.
2308.TP
6dd7fa77
JA
2309.BI replay_time_scale \fR=\fPint
2310When replaying I/O with \fBread_iolog\fR, fio will honor the original timing
2311in the trace. With this option, it's possible to scale the time. It's a
2312percentage option, if set to 50 it means run at 50% the original IO rate in
2313the trace. If set to 200, run at twice the original IO rate. Defaults to 100.
2314.TP
523bad63
TK
2315.BI replay_redirect \fR=\fPstr
2316While replaying I/O patterns using \fBread_iolog\fR the default behavior
2317is to replay the IOPS onto the major/minor device that each IOP was recorded
2318from. This is sometimes undesirable because on a different machine those
2319major/minor numbers can map to a different device. Changing hardware on the
2320same system can also result in a different major/minor mapping.
2321\fBreplay_redirect\fR causes all I/Os to be replayed onto the single specified
2322device regardless of the device it was recorded
2323from. i.e. `replay_redirect=/dev/sdc' would cause all I/O
2324in the blktrace or iolog to be replayed onto `/dev/sdc'. This means
2325multiple devices will be replayed onto a single device, if the trace
2326contains multiple devices. If you want multiple devices to be replayed
2327concurrently to multiple redirected devices you must blkparse your trace
2328into separate traces and replay them with independent fio invocations.
2329Unfortunately this also breaks the strict time ordering between multiple
2330device accesses.
2331.TP
2332.BI replay_align \fR=\fPint
350a535d
DZ
2333Force alignment of the byte offsets in a trace to this value. The value
2334must be a power of 2.
523bad63
TK
2335.TP
2336.BI replay_scale \fR=\fPint
350a535d
DZ
2337Scale bye offsets down by this factor when replaying traces. Should most
2338likely use \fBreplay_align\fR as well.
523bad63
TK
2339.SS "Threads, processes and job synchronization"
2340.TP
38f68906
JA
2341.BI replay_skip \fR=\fPstr
2342Sometimes it's useful to skip certain IO types in a replay trace. This could
2343be, for instance, eliminating the writes in the trace. Or not replaying the
2344trims/discards, if you are redirecting to a device that doesn't support them.
2345This option takes a comma separated list of read, write, trim, sync.
2346.TP
523bad63
TK
2347.BI thread
2348Fio defaults to creating jobs by using fork, however if this option is
2349given, fio will create jobs by using POSIX Threads' function
2350\fBpthread_create\fR\|(3) to create threads instead.
2351.TP
2352.BI wait_for \fR=\fPstr
2353If set, the current job won't be started until all workers of the specified
2354waitee job are done.
2355.\" ignore blank line here from HOWTO as it looks normal without it
2356\fBwait_for\fR operates on the job name basis, so there are a few
2357limitations. First, the waitee must be defined prior to the waiter job
2358(meaning no forward references). Second, if a job is being referenced as a
2359waitee, it must have a unique name (no duplicate waitees).
2360.TP
2361.BI nice \fR=\fPint
2362Run the job with the given nice value. See man \fBnice\fR\|(2).
2363.\" ignore blank line here from HOWTO as it looks normal without it
2364On Windows, values less than \-15 set the process class to "High"; \-1 through
2365\-15 set "Above Normal"; 1 through 15 "Below Normal"; and above 15 "Idle"
2366priority class.
2367.TP
2368.BI prio \fR=\fPint
2369Set the I/O priority value of this job. Linux limits us to a positive value
2370between 0 and 7, with 0 being the highest. See man
2371\fBionice\fR\|(1). Refer to an appropriate manpage for other operating
2372systems since meaning of priority may differ.
2373.TP
2374.BI prioclass \fR=\fPint
2375Set the I/O priority class. See man \fBionice\fR\|(1).
15501535 2376.TP
d60e92d1 2377.BI cpus_allowed \fR=\fPstr
523bad63 2378Controls the same options as \fBcpumask\fR, but accepts a textual
b570e037
SW
2379specification of the permitted CPUs instead and CPUs are indexed from 0. So
2380to use CPUs 0 and 5 you would specify `cpus_allowed=0,5'. This option also
2381allows a range of CPUs to be specified \-\- say you wanted a binding to CPUs
23820, 5, and 8 to 15, you would set `cpus_allowed=0,5,8\-15'.
2383.RS
2384.P
2385On Windows, when `cpus_allowed' is unset only CPUs from fio's current
2386processor group will be used and affinity settings are inherited from the
2387system. An fio build configured to target Windows 7 makes options that set
2388CPUs processor group aware and values will set both the processor group
2389and a CPU from within that group. For example, on a system where processor
2390group 0 has 40 CPUs and processor group 1 has 32 CPUs, `cpus_allowed'
2391values between 0 and 39 will bind CPUs from processor group 0 and
2392`cpus_allowed' values between 40 and 71 will bind CPUs from processor
2393group 1. When using `cpus_allowed_policy=shared' all CPUs specified by a
2394single `cpus_allowed' option must be from the same processor group. For
2395Windows fio builds not built for Windows 7, CPUs will only be selected from
2396(and be relative to) whatever processor group fio happens to be running in
2397and CPUs from other processor groups cannot be used.
2398.RE
d60e92d1 2399.TP
c2acfbac 2400.BI cpus_allowed_policy \fR=\fPstr
523bad63
TK
2401Set the policy of how fio distributes the CPUs specified by
2402\fBcpus_allowed\fR or \fBcpumask\fR. Two policies are supported:
c2acfbac
JA
2403.RS
2404.RS
2405.TP
2406.B shared
2407All jobs will share the CPU set specified.
2408.TP
2409.B split
2410Each job will get a unique CPU from the CPU set.
2411.RE
2412.P
523bad63
TK
2413\fBshared\fR is the default behavior, if the option isn't specified. If
2414\fBsplit\fR is specified, then fio will will assign one cpu per job. If not
2415enough CPUs are given for the jobs listed, then fio will roundrobin the CPUs
2416in the set.
c2acfbac 2417.RE
c2acfbac 2418.TP
b570e037
SW
2419.BI cpumask \fR=\fPint
2420Set the CPU affinity of this job. The parameter given is a bit mask of
2421allowed CPUs the job may run on. So if you want the allowed CPUs to be 1
2422and 5, you would pass the decimal value of (1 << 1 | 1 << 5), or 34. See man
2423\fBsched_setaffinity\fR\|(2). This may not work on all supported
2424operating systems or kernel versions. This option doesn't work well for a
2425higher CPU count than what you can store in an integer mask, so it can only
2426control cpus 1\-32. For boxes with larger CPU counts, use
2427\fBcpus_allowed\fR.
2428.TP
d0b937ed 2429.BI numa_cpu_nodes \fR=\fPstr
cecbfd47 2430Set this job running on specified NUMA nodes' CPUs. The arguments allow
523bad63
TK
2431comma delimited list of cpu numbers, A\-B ranges, or `all'. Note, to enable
2432NUMA options support, fio must be built on a system with libnuma\-dev(el)
2433installed.
d0b937ed
YR
2434.TP
2435.BI numa_mem_policy \fR=\fPstr
523bad63
TK
2436Set this job's memory policy and corresponding NUMA nodes. Format of the
2437arguments:
39c7a2ca
VF
2438.RS
2439.RS
523bad63
TK
2440.P
2441<mode>[:<nodelist>]
39c7a2ca 2442.RE
523bad63 2443.P
f1dd3fb1 2444`mode' is one of the following memory policies: `default', `prefer',
523bad63
TK
2445`bind', `interleave' or `local'. For `default' and `local' memory
2446policies, no node needs to be specified. For `prefer', only one node is
2447allowed. For `bind' and `interleave' the `nodelist' may be as
2448follows: a comma delimited list of numbers, A\-B ranges, or `all'.
39c7a2ca
VF
2449.RE
2450.TP
523bad63
TK
2451.BI cgroup \fR=\fPstr
2452Add job to this control group. If it doesn't exist, it will be created. The
2453system must have a mounted cgroup blkio mount point for this to work. If
2454your system doesn't have it mounted, you can do so with:
d60e92d1
AC
2455.RS
2456.RS
d60e92d1 2457.P
523bad63
TK
2458# mount \-t cgroup \-o blkio none /cgroup
2459.RE
d60e92d1
AC
2460.RE
2461.TP
523bad63
TK
2462.BI cgroup_weight \fR=\fPint
2463Set the weight of the cgroup to this value. See the documentation that comes
2464with the kernel, allowed values are in the range of 100..1000.
d60e92d1 2465.TP
523bad63
TK
2466.BI cgroup_nodelete \fR=\fPbool
2467Normally fio will delete the cgroups it has created after the job
2468completion. To override this behavior and to leave cgroups around after the
2469job completion, set `cgroup_nodelete=1'. This can be useful if one wants
2470to inspect various cgroup files after job completion. Default: false.
c8eeb9df 2471.TP
523bad63
TK
2472.BI flow_id \fR=\fPint
2473The ID of the flow. If not specified, it defaults to being a global
2474flow. See \fBflow\fR.
d60e92d1 2475.TP
523bad63 2476.BI flow \fR=\fPint
338f2db5 2477Weight in token-based flow control. If this value is used, then there is
523bad63
TK
2478a 'flow counter' which is used to regulate the proportion of activity between
2479two or more jobs. Fio attempts to keep this flow counter near zero. The
2480\fBflow\fR parameter stands for how much should be added or subtracted to the
2481flow counter on each iteration of the main I/O loop. That is, if one job has
2482`flow=8' and another job has `flow=\-1', then there will be a roughly 1:8
2483ratio in how much one runs vs the other.
d60e92d1 2484.TP
523bad63
TK
2485.BI flow_watermark \fR=\fPint
2486The maximum value that the absolute value of the flow counter is allowed to
2487reach before the job must wait for a lower value of the counter.
6b7f6851 2488.TP
523bad63
TK
2489.BI flow_sleep \fR=\fPint
2490The period of time, in microseconds, to wait after the flow watermark has
2491been exceeded before retrying operations.
25460cf6 2492.TP
523bad63
TK
2493.BI stonewall "\fR,\fB wait_for_previous"
2494Wait for preceding jobs in the job file to exit, before starting this
2495one. Can be used to insert serialization points in the job file. A stone
2496wall also implies starting a new reporting group, see
2497\fBgroup_reporting\fR.
2378826d 2498.TP
523bad63
TK
2499.BI exitall
2500By default, fio will continue running all other jobs when one job finishes
2501but sometimes this is not the desired action. Setting \fBexitall\fR will
2502instead make fio terminate all other jobs when one job finishes.
e81ecca3 2503.TP
523bad63
TK
2504.BI exec_prerun \fR=\fPstr
2505Before running this job, issue the command specified through
2506\fBsystem\fR\|(3). Output is redirected in a file called `jobname.prerun.txt'.
e9f48479 2507.TP
523bad63
TK
2508.BI exec_postrun \fR=\fPstr
2509After the job completes, issue the command specified though
2510\fBsystem\fR\|(3). Output is redirected in a file called `jobname.postrun.txt'.
d60e92d1 2511.TP
523bad63
TK
2512.BI uid \fR=\fPint
2513Instead of running as the invoking user, set the user ID to this value
2514before the thread/process does any work.
39c1c323 2515.TP
523bad63
TK
2516.BI gid \fR=\fPint
2517Set group ID, see \fBuid\fR.
2518.SS "Verification"
d60e92d1 2519.TP
589e88b7 2520.BI verify_only
523bad63 2521Do not perform specified workload, only verify data still matches previous
5e4c7118 2522invocation of this workload. This option allows one to check data multiple
523bad63
TK
2523times at a later date without overwriting it. This option makes sense only
2524for workloads that write data, and does not support workloads with the
5e4c7118
JA
2525\fBtime_based\fR option set.
2526.TP
d60e92d1 2527.BI do_verify \fR=\fPbool
523bad63
TK
2528Run the verify phase after a write phase. Only valid if \fBverify\fR is
2529set. Default: true.
d60e92d1
AC
2530.TP
2531.BI verify \fR=\fPstr
523bad63
TK
2532If writing to a file, fio can verify the file contents after each iteration
2533of the job. Each verification method also implies verification of special
2534header, which is written to the beginning of each block. This header also
2535includes meta information, like offset of the block, block number, timestamp
2536when block was written, etc. \fBverify\fR can be combined with
2537\fBverify_pattern\fR option. The allowed values are:
d60e92d1
AC
2538.RS
2539.RS
2540.TP
523bad63
TK
2541.B md5
2542Use an md5 sum of the data area and store it in the header of
2543each block.
2544.TP
2545.B crc64
2546Use an experimental crc64 sum of the data area and store it in the
2547header of each block.
2548.TP
2549.B crc32c
2550Use a crc32c sum of the data area and store it in the header of
2551each block. This will automatically use hardware acceleration
2552(e.g. SSE4.2 on an x86 or CRC crypto extensions on ARM64) but will
2553fall back to software crc32c if none is found. Generally the
f1dd3fb1 2554fastest checksum fio supports when hardware accelerated.
523bad63
TK
2555.TP
2556.B crc32c\-intel
2557Synonym for crc32c.
2558.TP
2559.B crc32
2560Use a crc32 sum of the data area and store it in the header of each
2561block.
2562.TP
2563.B crc16
2564Use a crc16 sum of the data area and store it in the header of each
2565block.
2566.TP
2567.B crc7
2568Use a crc7 sum of the data area and store it in the header of each
2569block.
2570.TP
2571.B xxhash
2572Use xxhash as the checksum function. Generally the fastest software
2573checksum that fio supports.
2574.TP
2575.B sha512
2576Use sha512 as the checksum function.
2577.TP
2578.B sha256
2579Use sha256 as the checksum function.
2580.TP
2581.B sha1
2582Use optimized sha1 as the checksum function.
2583.TP
2584.B sha3\-224
2585Use optimized sha3\-224 as the checksum function.
2586.TP
2587.B sha3\-256
2588Use optimized sha3\-256 as the checksum function.
2589.TP
2590.B sha3\-384
2591Use optimized sha3\-384 as the checksum function.
2592.TP
2593.B sha3\-512
2594Use optimized sha3\-512 as the checksum function.
d60e92d1
AC
2595.TP
2596.B meta
523bad63
TK
2597This option is deprecated, since now meta information is included in
2598generic verification header and meta verification happens by
2599default. For detailed information see the description of the
2600\fBverify\fR setting. This option is kept because of
2601compatibility's sake with old configurations. Do not use it.
d60e92d1 2602.TP
59245381 2603.B pattern
523bad63
TK
2604Verify a strict pattern. Normally fio includes a header with some
2605basic information and checksumming, but if this option is set, only
2606the specific pattern set with \fBverify_pattern\fR is verified.
59245381 2607.TP
d60e92d1 2608.B null
523bad63
TK
2609Only pretend to verify. Useful for testing internals with
2610`ioengine=null', not for much else.
d60e92d1 2611.RE
523bad63
TK
2612.P
2613This option can be used for repeated burn\-in tests of a system to make sure
2614that the written data is also correctly read back. If the data direction
2615given is a read or random read, fio will assume that it should verify a
2616previously written file. If the data direction includes any form of write,
2617the verify will be of the newly written data.
47e6a6e5
SW
2618.P
2619To avoid false verification errors, do not use the norandommap option when
2620verifying data with async I/O engines and I/O depths > 1. Or use the
2621norandommap and the lfsr random generator together to avoid writing to the
2622same offset with muliple outstanding I/Os.
d60e92d1
AC
2623.RE
2624.TP
f7fa2653 2625.BI verify_offset \fR=\fPint
d60e92d1 2626Swap the verification header with data somewhere else in the block before
523bad63 2627writing. It is swapped back before verifying.
d60e92d1 2628.TP
f7fa2653 2629.BI verify_interval \fR=\fPint
523bad63
TK
2630Write the verification header at a finer granularity than the
2631\fBblocksize\fR. It will be written for chunks the size of
2632\fBverify_interval\fR. \fBblocksize\fR should divide this evenly.
d60e92d1 2633.TP
996093bb 2634.BI verify_pattern \fR=\fPstr
523bad63
TK
2635If set, fio will fill the I/O buffers with this pattern. Fio defaults to
2636filling with totally random bytes, but sometimes it's interesting to fill
2637with a known pattern for I/O verification purposes. Depending on the width
2638of the pattern, fio will fill 1/2/3/4 bytes of the buffer at the time (it can
2639be either a decimal or a hex number). The \fBverify_pattern\fR if larger than
2640a 32\-bit quantity has to be a hex number that starts with either "0x" or
2641"0X". Use with \fBverify\fR. Also, \fBverify_pattern\fR supports %o
2642format, which means that for each block offset will be written and then
2643verified back, e.g.:
2fa5a241
RP
2644.RS
2645.RS
523bad63
TK
2646.P
2647verify_pattern=%o
2fa5a241 2648.RE
523bad63 2649.P
2fa5a241 2650Or use combination of everything:
2fa5a241 2651.RS
523bad63
TK
2652.P
2653verify_pattern=0xff%o"abcd"\-12
2fa5a241
RP
2654.RE
2655.RE
996093bb 2656.TP
d60e92d1 2657.BI verify_fatal \fR=\fPbool
523bad63
TK
2658Normally fio will keep checking the entire contents before quitting on a
2659block verification failure. If this option is set, fio will exit the job on
2660the first observed failure. Default: false.
d60e92d1 2661.TP
b463e936 2662.BI verify_dump \fR=\fPbool
523bad63
TK
2663If set, dump the contents of both the original data block and the data block
2664we read off disk to files. This allows later analysis to inspect just what
2665kind of data corruption occurred. Off by default.
b463e936 2666.TP
e8462bd8 2667.BI verify_async \fR=\fPint
523bad63
TK
2668Fio will normally verify I/O inline from the submitting thread. This option
2669takes an integer describing how many async offload threads to create for I/O
2670verification instead, causing fio to offload the duty of verifying I/O
2671contents to one or more separate threads. If using this offload option, even
2672sync I/O engines can benefit from using an \fBiodepth\fR setting higher
2673than 1, as it allows them to have I/O in flight while verifies are running.
2674Defaults to 0 async threads, i.e. verification is not asynchronous.
e8462bd8
JA
2675.TP
2676.BI verify_async_cpus \fR=\fPstr
523bad63
TK
2677Tell fio to set the given CPU affinity on the async I/O verification
2678threads. See \fBcpus_allowed\fR for the format used.
e8462bd8 2679.TP
6f87418f
JA
2680.BI verify_backlog \fR=\fPint
2681Fio will normally verify the written contents of a job that utilizes verify
2682once that job has completed. In other words, everything is written then
2683everything is read back and verified. You may want to verify continually
523bad63
TK
2684instead for a variety of reasons. Fio stores the meta data associated with
2685an I/O block in memory, so for large verify workloads, quite a bit of memory
2686would be used up holding this meta data. If this option is enabled, fio will
2687write only N blocks before verifying these blocks.
6f87418f
JA
2688.TP
2689.BI verify_backlog_batch \fR=\fPint
523bad63
TK
2690Control how many blocks fio will verify if \fBverify_backlog\fR is
2691set. If not set, will default to the value of \fBverify_backlog\fR
2692(meaning the entire queue is read back and verified). If
2693\fBverify_backlog_batch\fR is less than \fBverify_backlog\fR then not all
2694blocks will be verified, if \fBverify_backlog_batch\fR is larger than
2695\fBverify_backlog\fR, some blocks will be verified more than once.
2696.TP
2697.BI verify_state_save \fR=\fPbool
2698When a job exits during the write phase of a verify workload, save its
2699current state. This allows fio to replay up until that point, if the verify
2700state is loaded for the verify read phase. The format of the filename is,
2701roughly:
2702.RS
2703.RS
2704.P
2705<type>\-<jobname>\-<jobindex>\-verify.state.
2706.RE
2707.P
2708<type> is "local" for a local run, "sock" for a client/server socket
2709connection, and "ip" (192.168.0.1, for instance) for a networked
2710client/server connection. Defaults to true.
2711.RE
2712.TP
2713.BI verify_state_load \fR=\fPbool
2714If a verify termination trigger was used, fio stores the current write state
2715of each thread. This can be used at verification time so that fio knows how
2716far it should verify. Without this information, fio will run a full
2717verification pass, according to the settings in the job file used. Default
2718false.
6f87418f 2719.TP
fa769d44
SW
2720.BI trim_percentage \fR=\fPint
2721Number of verify blocks to discard/trim.
2722.TP
2723.BI trim_verify_zero \fR=\fPbool
523bad63 2724Verify that trim/discarded blocks are returned as zeros.
fa769d44
SW
2725.TP
2726.BI trim_backlog \fR=\fPint
523bad63 2727Verify that trim/discarded blocks are returned as zeros.
fa769d44
SW
2728.TP
2729.BI trim_backlog_batch \fR=\fPint
523bad63 2730Trim this number of I/O blocks.
fa769d44
SW
2731.TP
2732.BI experimental_verify \fR=\fPbool
2733Enable experimental verification.
523bad63 2734.SS "Steady state"
fa769d44 2735.TP
523bad63
TK
2736.BI steadystate \fR=\fPstr:float "\fR,\fP ss" \fR=\fPstr:float
2737Define the criterion and limit for assessing steady state performance. The
2738first parameter designates the criterion whereas the second parameter sets
2739the threshold. When the criterion falls below the threshold for the
2740specified duration, the job will stop. For example, `iops_slope:0.1%' will
2741direct fio to terminate the job when the least squares regression slope
2742falls below 0.1% of the mean IOPS. If \fBgroup_reporting\fR is enabled
2743this will apply to all jobs in the group. Below is the list of available
2744steady state assessment criteria. All assessments are carried out using only
2745data from the rolling collection window. Threshold limits can be expressed
2746as a fixed value or as a percentage of the mean in the collection window.
2747.RS
1cb049d9
VF
2748.P
2749When using this feature, most jobs should include the \fBtime_based\fR
2750and \fBruntime\fR options or the \fBloops\fR option so that fio does not
2751stop running after it has covered the full size of the specified file(s)
2752or device(s).
2753.RS
523bad63 2754.RS
d60e92d1 2755.TP
523bad63
TK
2756.B iops
2757Collect IOPS data. Stop the job if all individual IOPS measurements
2758are within the specified limit of the mean IOPS (e.g., `iops:2'
2759means that all individual IOPS values must be within 2 of the mean,
2760whereas `iops:0.2%' means that all individual IOPS values must be
2761within 0.2% of the mean IOPS to terminate the job).
d60e92d1 2762.TP
523bad63
TK
2763.B iops_slope
2764Collect IOPS data and calculate the least squares regression
2765slope. Stop the job if the slope falls below the specified limit.
d60e92d1 2766.TP
523bad63
TK
2767.B bw
2768Collect bandwidth data. Stop the job if all individual bandwidth
2769measurements are within the specified limit of the mean bandwidth.
64bbb865 2770.TP
523bad63
TK
2771.B bw_slope
2772Collect bandwidth data and calculate the least squares regression
2773slope. Stop the job if the slope falls below the specified limit.
2774.RE
2775.RE
d1c46c04 2776.TP
523bad63
TK
2777.BI steadystate_duration \fR=\fPtime "\fR,\fP ss_dur" \fR=\fPtime
2778A rolling window of this duration will be used to judge whether steady state
2779has been reached. Data will be collected once per second. The default is 0
2780which disables steady state detection. When the unit is omitted, the
2781value is interpreted in seconds.
0c63576e 2782.TP
523bad63
TK
2783.BI steadystate_ramp_time \fR=\fPtime "\fR,\fP ss_ramp" \fR=\fPtime
2784Allow the job to run for the specified duration before beginning data
2785collection for checking the steady state job termination criterion. The
2786default is 0. When the unit is omitted, the value is interpreted in seconds.
2787.SS "Measurements and reporting"
0c63576e 2788.TP
3a5db920
JA
2789.BI per_job_logs \fR=\fPbool
2790If set, this generates bw/clat/iops log with per file private filenames. If
523bad63
TK
2791not set, jobs with identical names will share the log filename. Default:
2792true.
2793.TP
2794.BI group_reporting
2795It may sometimes be interesting to display statistics for groups of jobs as
2796a whole instead of for each individual job. This is especially true if
2797\fBnumjobs\fR is used; looking at individual thread/process output
338f2db5
SW
2798quickly becomes unwieldy. To see the final report per-group instead of
2799per-job, use \fBgroup_reporting\fR. Jobs in a file will be part of the
523bad63
TK
2800same reporting group, unless if separated by a \fBstonewall\fR, or by
2801using \fBnew_group\fR.
2802.TP
2803.BI new_group
2804Start a new reporting group. See: \fBgroup_reporting\fR. If not given,
2805all jobs in a file will be part of the same reporting group, unless
2806separated by a \fBstonewall\fR.
2807.TP
2808.BI stats \fR=\fPbool
2809By default, fio collects and shows final output results for all jobs
2810that run. If this option is set to 0, then fio will ignore it in
2811the final stat output.
3a5db920 2812.TP
836bad52 2813.BI write_bw_log \fR=\fPstr
523bad63 2814If given, write a bandwidth log for this job. Can be used to store data of
074f0817 2815the bandwidth of the jobs in their lifetime.
523bad63 2816.RS
074f0817
SW
2817.P
2818If no str argument is given, the default filename of
2819`jobname_type.x.log' is used. Even when the argument is given, fio
2820will still append the type of log. So if one specifies:
523bad63
TK
2821.RS
2822.P
074f0817 2823write_bw_log=foo
523bad63
TK
2824.RE
2825.P
074f0817
SW
2826The actual log name will be `foo_bw.x.log' where `x' is the index
2827of the job (1..N, where N is the number of jobs). If
2828\fBper_job_logs\fR is false, then the filename will not include the
2829`.x` job index.
2830.P
2831The included \fBfio_generate_plots\fR script uses gnuplot to turn these
2832text files into nice graphs. See the \fBLOG FILE FORMATS\fR section for how data is
2833structured within the file.
523bad63 2834.RE
901bb994 2835.TP
074f0817
SW
2836.BI write_lat_log \fR=\fPstr
2837Same as \fBwrite_bw_log\fR, except this option creates I/O
2838submission (e.g., `name_slat.x.log'), completion (e.g.,
2839`name_clat.x.log'), and total (e.g., `name_lat.x.log') latency
2840files instead. See \fBwrite_bw_log\fR for details about the
2841filename format and the \fBLOG FILE FORMATS\fR section for how data is structured
2842within the files.
2843.TP
1e613c9c 2844.BI write_hist_log \fR=\fPstr
074f0817
SW
2845Same as \fBwrite_bw_log\fR but writes an I/O completion latency
2846histogram file (e.g., `name_hist.x.log') instead. Note that this
2847file will be empty unless \fBlog_hist_msec\fR has also been set.
2848See \fBwrite_bw_log\fR for details about the filename format and
2849the \fBLOG FILE FORMATS\fR section for how data is structured
2850within the file.
1e613c9c 2851.TP
c8eeb9df 2852.BI write_iops_log \fR=\fPstr
074f0817 2853Same as \fBwrite_bw_log\fR, but writes an IOPS file (e.g.
15417073
SW
2854`name_iops.x.log`) instead. Because fio defaults to individual
2855I/O logging, the value entry in the IOPS log will be 1 unless windowed
2856logging (see \fBlog_avg_msec\fR) has been enabled. See
2857\fBwrite_bw_log\fR for details about the filename format and \fBLOG
2858FILE FORMATS\fR for how data is structured within the file.
c8eeb9df 2859.TP
b8bc8cba
JA
2860.BI log_avg_msec \fR=\fPint
2861By default, fio will log an entry in the iops, latency, or bw log for every
523bad63 2862I/O that completes. When writing to the disk log, that can quickly grow to a
b8bc8cba 2863very large size. Setting this option makes fio average the each log entry
e6989e10 2864over the specified period of time, reducing the resolution of the log. See
523bad63
TK
2865\fBlog_max_value\fR as well. Defaults to 0, logging all entries.
2866Also see \fBLOG FILE FORMATS\fR section.
b8bc8cba 2867.TP
1e613c9c 2868.BI log_hist_msec \fR=\fPint
523bad63
TK
2869Same as \fBlog_avg_msec\fR, but logs entries for completion latency
2870histograms. Computing latency percentiles from averages of intervals using
2871\fBlog_avg_msec\fR is inaccurate. Setting this option makes fio log
2872histogram entries over the specified period of time, reducing log sizes for
2873high IOPS devices while retaining percentile accuracy. See
074f0817
SW
2874\fBlog_hist_coarseness\fR and \fBwrite_hist_log\fR as well.
2875Defaults to 0, meaning histogram logging is disabled.
1e613c9c
KC
2876.TP
2877.BI log_hist_coarseness \fR=\fPint
523bad63
TK
2878Integer ranging from 0 to 6, defining the coarseness of the resolution of
2879the histogram logs enabled with \fBlog_hist_msec\fR. For each increment
2880in coarseness, fio outputs half as many bins. Defaults to 0, for which
2881histogram logs contain 1216 latency bins. See \fBLOG FILE FORMATS\fR section.
2882.TP
2883.BI log_max_value \fR=\fPbool
2884If \fBlog_avg_msec\fR is set, fio logs the average over that window. If
2885you instead want to log the maximum value, set this option to 1. Defaults to
28860, meaning that averaged values are logged.
1e613c9c 2887.TP
ae588852 2888.BI log_offset \fR=\fPbool
523bad63
TK
2889If this is set, the iolog options will include the byte offset for the I/O
2890entry as well as the other data values. Defaults to 0 meaning that
2891offsets are not present in logs. Also see \fBLOG FILE FORMATS\fR section.
ae588852 2892.TP
aee2ab67 2893.BI log_compression \fR=\fPint
523bad63
TK
2894If this is set, fio will compress the I/O logs as it goes, to keep the
2895memory footprint lower. When a log reaches the specified size, that chunk is
2896removed and compressed in the background. Given that I/O logs are fairly
2897highly compressible, this yields a nice memory savings for longer runs. The
2898downside is that the compression will consume some background CPU cycles, so
2899it may impact the run. This, however, is also true if the logging ends up
2900consuming most of the system memory. So pick your poison. The I/O logs are
2901saved normally at the end of a run, by decompressing the chunks and storing
2902them in the specified log file. This feature depends on the availability of
2903zlib.
aee2ab67 2904.TP
c08f9fe2 2905.BI log_compression_cpus \fR=\fPstr
523bad63
TK
2906Define the set of CPUs that are allowed to handle online log compression for
2907the I/O jobs. This can provide better isolation between performance
0cf90a62
SW
2908sensitive jobs, and background compression work. See \fBcpus_allowed\fR for
2909the format used.
c08f9fe2 2910.TP
b26317c9 2911.BI log_store_compressed \fR=\fPbool
c08f9fe2 2912If set, fio will store the log files in a compressed format. They can be
523bad63
TK
2913decompressed with fio, using the \fB\-\-inflate\-log\fR command line
2914parameter. The files will be stored with a `.fz' suffix.
b26317c9 2915.TP
3aea75b1
KC
2916.BI log_unix_epoch \fR=\fPbool
2917If set, fio will log Unix timestamps to the log files produced by enabling
338f2db5 2918write_type_log for each log type, instead of the default zero-based
3aea75b1
KC
2919timestamps.
2920.TP
66347cfa 2921.BI block_error_percentiles \fR=\fPbool
338f2db5 2922If set, record errors in trim block-sized units from writes and trims and
523bad63
TK
2923output a histogram of how many trims it took to get to errors, and what kind
2924of error was encountered.
d60e92d1 2925.TP
523bad63
TK
2926.BI bwavgtime \fR=\fPint
2927Average the calculated bandwidth over the given time. Value is specified in
2928milliseconds. If the job also does bandwidth logging through
2929\fBwrite_bw_log\fR, then the minimum of this option and
2930\fBlog_avg_msec\fR will be used. Default: 500ms.
d60e92d1 2931.TP
523bad63
TK
2932.BI iopsavgtime \fR=\fPint
2933Average the calculated IOPS over the given time. Value is specified in
2934milliseconds. If the job also does IOPS logging through
2935\fBwrite_iops_log\fR, then the minimum of this option and
2936\fBlog_avg_msec\fR will be used. Default: 500ms.
d60e92d1 2937.TP
d60e92d1 2938.BI disk_util \fR=\fPbool
523bad63
TK
2939Generate disk utilization statistics, if the platform supports it.
2940Default: true.
fa769d44 2941.TP
523bad63
TK
2942.BI disable_lat \fR=\fPbool
2943Disable measurements of total latency numbers. Useful only for cutting back
2944the number of calls to \fBgettimeofday\fR\|(2), as that does impact
2945performance at really high IOPS rates. Note that to really get rid of a
2946large amount of these calls, this option must be used with
2947\fBdisable_slat\fR and \fBdisable_bw_measurement\fR as well.
9e684a49 2948.TP
523bad63
TK
2949.BI disable_clat \fR=\fPbool
2950Disable measurements of completion latency numbers. See
2951\fBdisable_lat\fR.
9e684a49 2952.TP
523bad63
TK
2953.BI disable_slat \fR=\fPbool
2954Disable measurements of submission latency numbers. See
2955\fBdisable_lat\fR.
9e684a49 2956.TP
523bad63
TK
2957.BI disable_bw_measurement \fR=\fPbool "\fR,\fP disable_bw" \fR=\fPbool
2958Disable measurements of throughput/bandwidth numbers. See
2959\fBdisable_lat\fR.
9e684a49 2960.TP
83349190 2961.BI clat_percentiles \fR=\fPbool
b599759b
JA
2962Enable the reporting of percentiles of completion latencies. This option is
2963mutually exclusive with \fBlat_percentiles\fR.
2964.TP
2965.BI lat_percentiles \fR=\fPbool
b71968b1 2966Enable the reporting of percentiles of I/O latencies. This is similar to
b599759b
JA
2967\fBclat_percentiles\fR, except that this includes the submission latency.
2968This option is mutually exclusive with \fBclat_percentiles\fR.
83349190
YH
2969.TP
2970.BI percentile_list \fR=\fPfloat_list
66347cfa 2971Overwrite the default list of percentiles for completion latencies and the
523bad63
TK
2972block error histogram. Each number is a floating number in the range
2973(0,100], and the maximum length of the list is 20. Use ':' to separate the
2974numbers, and list the numbers in ascending order. For example,
2975`\-\-percentile_list=99.5:99.9' will cause fio to report the values of
2976completion latency below which 99.5% and 99.9% of the observed latencies
2977fell, respectively.
e883cb35
JF
2978.TP
2979.BI significant_figures \fR=\fPint
c32ba107
JA
2980If using \fB\-\-output\-format\fR of `normal', set the significant figures
2981to this value. Higher values will yield more precise IOPS and throughput
2982units, while lower values will round. Requires a minimum value of 1 and a
e883cb35 2983maximum value of 10. Defaults to 4.
523bad63 2984.SS "Error handling"
e4585935 2985.TP
523bad63
TK
2986.BI exitall_on_error
2987When one job finishes in error, terminate the rest. The default is to wait
2988for each job to finish.
e4585935 2989.TP
523bad63
TK
2990.BI continue_on_error \fR=\fPstr
2991Normally fio will exit the job on the first observed failure. If this option
338f2db5 2992is set, fio will continue the job when there is a 'non-fatal error' (EIO or
523bad63
TK
2993EILSEQ) until the runtime is exceeded or the I/O size specified is
2994completed. If this option is used, there are two more stats that are
2995appended, the total error count and the first error. The error field given
2996in the stats is the first error that was hit during the run.
2997The allowed values are:
2998.RS
2999.RS
046395d7 3000.TP
523bad63
TK
3001.B none
3002Exit on any I/O or verify errors.
de890a1e 3003.TP
523bad63
TK
3004.B read
3005Continue on read errors, exit on all others.
2cafffbe 3006.TP
523bad63
TK
3007.B write
3008Continue on write errors, exit on all others.
a0679ce5 3009.TP
523bad63
TK
3010.B io
3011Continue on any I/O error, exit on all others.
de890a1e 3012.TP
523bad63
TK
3013.B verify
3014Continue on verify errors, exit on all others.
de890a1e 3015.TP
523bad63
TK
3016.B all
3017Continue on all errors.
b93b6a2e 3018.TP
523bad63 3019.B 0
338f2db5 3020Backward-compatible alias for 'none'.
d3a623de 3021.TP
523bad63 3022.B 1
338f2db5 3023Backward-compatible alias for 'all'.
523bad63
TK
3024.RE
3025.RE
1d360ffb 3026.TP
523bad63
TK
3027.BI ignore_error \fR=\fPstr
3028Sometimes you want to ignore some errors during test in that case you can
3029specify error list for each error type, instead of only being able to
338f2db5 3030ignore the default 'non-fatal error' using \fBcontinue_on_error\fR.
523bad63
TK
3031`ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST' errors for
3032given error type is separated with ':'. Error may be symbol ('ENOSPC', 'ENOMEM')
3033or integer. Example:
de890a1e
SL
3034.RS
3035.RS
523bad63
TK
3036.P
3037ignore_error=EAGAIN,ENOSPC:122
3038.RE
3039.P
3040This option will ignore EAGAIN from READ, and ENOSPC and 122(EDQUOT) from
3041WRITE. This option works by overriding \fBcontinue_on_error\fR with
3042the list of errors for each error type if any.
3043.RE
de890a1e 3044.TP
523bad63
TK
3045.BI error_dump \fR=\fPbool
3046If set dump every error even if it is non fatal, true by default. If
3047disabled only fatal error will be dumped.
3048.SS "Running predefined workloads"
3049Fio includes predefined profiles that mimic the I/O workloads generated by
3050other tools.
49ccb8c1 3051.TP
523bad63
TK
3052.BI profile \fR=\fPstr
3053The predefined workload to run. Current profiles are:
3054.RS
3055.RS
de890a1e 3056.TP
523bad63
TK
3057.B tiobench
3058Threaded I/O bench (tiotest/tiobench) like workload.
49ccb8c1 3059.TP
523bad63
TK
3060.B act
3061Aerospike Certification Tool (ACT) like workload.
3062.RE
de890a1e
SL
3063.RE
3064.P
523bad63
TK
3065To view a profile's additional options use \fB\-\-cmdhelp\fR after specifying
3066the profile. For example:
3067.RS
3068.TP
3069$ fio \-\-profile=act \-\-cmdhelp
de890a1e 3070.RE
523bad63 3071.SS "Act profile options"
de890a1e 3072.TP
523bad63
TK
3073.BI device\-names \fR=\fPstr
3074Devices to use.
d54fce84 3075.TP
523bad63
TK
3076.BI load \fR=\fPint
3077ACT load multiplier. Default: 1.
7aeb1e94 3078.TP
523bad63
TK
3079.BI test\-duration\fR=\fPtime
3080How long the entire test takes to run. When the unit is omitted, the value
3081is given in seconds. Default: 24h.
1008602c 3082.TP
523bad63
TK
3083.BI threads\-per\-queue\fR=\fPint
3084Number of read I/O threads per device. Default: 8.
e5f34d95 3085.TP
523bad63
TK
3086.BI read\-req\-num\-512\-blocks\fR=\fPint
3087Number of 512B blocks to read at the time. Default: 3.
d54fce84 3088.TP
523bad63
TK
3089.BI large\-block\-op\-kbytes\fR=\fPint
3090Size of large block ops in KiB (writes). Default: 131072.
d54fce84 3091.TP
523bad63
TK
3092.BI prep
3093Set to run ACT prep phase.
3094.SS "Tiobench profile options"
6d500c2e 3095.TP
523bad63
TK
3096.BI size\fR=\fPstr
3097Size in MiB.
0d978694 3098.TP
523bad63
TK
3099.BI block\fR=\fPint
3100Block size in bytes. Default: 4096.
0d978694 3101.TP
523bad63
TK
3102.BI numruns\fR=\fPint
3103Number of runs.
0d978694 3104.TP
523bad63
TK
3105.BI dir\fR=\fPstr
3106Test directory.
65fa28ca 3107.TP
523bad63
TK
3108.BI threads\fR=\fPint
3109Number of threads.
d60e92d1 3110.SH OUTPUT
40943b9a
TK
3111Fio spits out a lot of output. While running, fio will display the status of the
3112jobs created. An example of that would be:
d60e92d1 3113.P
40943b9a
TK
3114.nf
3115 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]
3116.fi
d1429b5c 3117.P
40943b9a
TK
3118The characters inside the first set of square brackets denote the current status of
3119each thread. The first character is the first job defined in the job file, and so
3120forth. The possible values (in typical life cycle order) are:
d60e92d1
AC
3121.RS
3122.TP
40943b9a 3123.PD 0
d60e92d1 3124.B P
40943b9a 3125Thread setup, but not started.
d60e92d1
AC
3126.TP
3127.B C
3128Thread created.
3129.TP
3130.B I
40943b9a
TK
3131Thread initialized, waiting or generating necessary data.
3132.TP
522c29f6 3133.B p
338f2db5 3134Thread running pre-reading file(s).
40943b9a
TK
3135.TP
3136.B /
3137Thread is in ramp period.
d60e92d1
AC
3138.TP
3139.B R
3140Running, doing sequential reads.
3141.TP
3142.B r
3143Running, doing random reads.
3144.TP
3145.B W
3146Running, doing sequential writes.
3147.TP
3148.B w
3149Running, doing random writes.
3150.TP
3151.B M
3152Running, doing mixed sequential reads/writes.
3153.TP
3154.B m
3155Running, doing mixed random reads/writes.
3156.TP
40943b9a
TK
3157.B D
3158Running, doing sequential trims.
3159.TP
3160.B d
3161Running, doing random trims.
3162.TP
d60e92d1
AC
3163.B F
3164Running, currently waiting for \fBfsync\fR\|(2).
3165.TP
3166.B V
40943b9a
TK
3167Running, doing verification of written data.
3168.TP
3169.B f
3170Thread finishing.
d60e92d1
AC
3171.TP
3172.B E
40943b9a 3173Thread exited, not reaped by main thread yet.
d60e92d1
AC
3174.TP
3175.B \-
40943b9a
TK
3176Thread reaped.
3177.TP
3178.B X
3179Thread reaped, exited with an error.
3180.TP
3181.B K
3182Thread reaped, exited due to signal.
d1429b5c 3183.PD
40943b9a
TK
3184.RE
3185.P
3186Fio will condense the thread string as not to take up more space on the command
3187line than needed. For instance, if you have 10 readers and 10 writers running,
3188the output would look like this:
3189.P
3190.nf
3191 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]
3192.fi
d60e92d1 3193.P
40943b9a
TK
3194Note that the status string is displayed in order, so it's possible to tell which of
3195the jobs are currently doing what. In the example above this means that jobs 1\-\-10
3196are readers and 11\-\-20 are writers.
d60e92d1 3197.P
40943b9a
TK
3198The other values are fairly self explanatory \-\- number of threads currently
3199running and doing I/O, the number of currently open files (f=), the estimated
3200completion percentage, the rate of I/O since last check (read speed listed first,
3201then write speed and optionally trim speed) in terms of bandwidth and IOPS,
3202and time to completion for the current running group. It's impossible to estimate
3203runtime of the following groups (if any).
d60e92d1 3204.P
40943b9a
TK
3205When fio is done (or interrupted by Ctrl\-C), it will show the data for
3206each thread, group of threads, and disks in that order. For each overall thread (or
3207group) the output looks like:
3208.P
3209.nf
3210 Client1: (groupid=0, jobs=1): err= 0: pid=16109: Sat Jun 24 12:07:54 2017
3211 write: IOPS=88, BW=623KiB/s (638kB/s)(30.4MiB/50032msec)
3212 slat (nsec): min=500, max=145500, avg=8318.00, stdev=4781.50
3213 clat (usec): min=170, max=78367, avg=4019.02, stdev=8293.31
3214 lat (usec): min=174, max=78375, avg=4027.34, stdev=8291.79
3215 clat percentiles (usec):
3216 | 1.00th=[ 302], 5.00th=[ 326], 10.00th=[ 343], 20.00th=[ 363],
3217 | 30.00th=[ 392], 40.00th=[ 404], 50.00th=[ 416], 60.00th=[ 445],
3218 | 70.00th=[ 816], 80.00th=[ 6718], 90.00th=[12911], 95.00th=[21627],
3219 | 99.00th=[43779], 99.50th=[51643], 99.90th=[68682], 99.95th=[72877],
3220 | 99.99th=[78119]
3221 bw ( KiB/s): min= 532, max= 686, per=0.10%, avg=622.87, stdev=24.82, samples= 100
3222 iops : min= 76, max= 98, avg=88.98, stdev= 3.54, samples= 100
d3b9694d
VF
3223 lat (usec) : 250=0.04%, 500=64.11%, 750=4.81%, 1000=2.79%
3224 lat (msec) : 2=4.16%, 4=1.84%, 10=4.90%, 20=11.33%, 50=5.37%
3225 lat (msec) : 100=0.65%
40943b9a
TK
3226 cpu : usr=0.27%, sys=0.18%, ctx=12072, majf=0, minf=21
3227 IO depths : 1=85.0%, 2=13.1%, 4=1.8%, 8=0.1%, 16=0.0%, 32=0.0%, >=64=0.0%
3228 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
3229 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
3230 issued rwt: total=0,4450,0, short=0,0,0, dropped=0,0,0
3231 latency : target=0, window=0, percentile=100.00%, depth=8
3232.fi
3233.P
3234The job name (or first job's name when using \fBgroup_reporting\fR) is printed,
3235along with the group id, count of jobs being aggregated, last error id seen (which
3236is 0 when there are no errors), pid/tid of that thread and the time the job/group
3237completed. Below are the I/O statistics for each data direction performed (showing
3238writes in the example above). In the order listed, they denote:
d60e92d1 3239.RS
d60e92d1 3240.TP
40943b9a
TK
3241.B read/write/trim
3242The string before the colon shows the I/O direction the statistics
3243are for. \fIIOPS\fR is the average I/Os performed per second. \fIBW\fR
3244is the average bandwidth rate shown as: value in power of 2 format
3245(value in power of 10 format). The last two values show: (total
3246I/O performed in power of 2 format / \fIruntime\fR of that thread).
d60e92d1
AC
3247.TP
3248.B slat
40943b9a
TK
3249Submission latency (\fImin\fR being the minimum, \fImax\fR being the
3250maximum, \fIavg\fR being the average, \fIstdev\fR being the standard
3251deviation). This is the time it took to submit the I/O. For
3252sync I/O this row is not displayed as the slat is really the
3253completion latency (since queue/complete is one operation there).
3254This value can be in nanoseconds, microseconds or milliseconds \-\-\-
3255fio will choose the most appropriate base and print that (in the
3256example above nanoseconds was the best scale). Note: in \fB\-\-minimal\fR mode
3257latencies are always expressed in microseconds.
d60e92d1
AC
3258.TP
3259.B clat
40943b9a
TK
3260Completion latency. Same names as slat, this denotes the time from
3261submission to completion of the I/O pieces. For sync I/O, clat will
3262usually be equal (or very close) to 0, as the time from submit to
3263complete is basically just CPU time (I/O has already been done, see slat
3264explanation).
d60e92d1 3265.TP
d3b9694d
VF
3266.B lat
3267Total latency. Same names as slat and clat, this denotes the time from
3268when fio created the I/O unit to completion of the I/O operation.
3269.TP
d60e92d1 3270.B bw
40943b9a
TK
3271Bandwidth statistics based on samples. Same names as the xlat stats,
3272but also includes the number of samples taken (\fIsamples\fR) and an
3273approximate percentage of total aggregate bandwidth this thread
3274received in its group (\fIper\fR). This last value is only really
3275useful if the threads in this group are on the same disk, since they
3276are then competing for disk access.
3277.TP
3278.B iops
3279IOPS statistics based on samples. Same names as \fBbw\fR.
d60e92d1 3280.TP
d3b9694d
VF
3281.B lat (nsec/usec/msec)
3282The distribution of I/O completion latencies. This is the time from when
3283I/O leaves fio and when it gets completed. Unlike the separate
3284read/write/trim sections above, the data here and in the remaining
3285sections apply to all I/Os for the reporting group. 250=0.04% means that
32860.04% of the I/Os completed in under 250us. 500=64.11% means that 64.11%
3287of the I/Os required 250 to 499us for completion.
3288.TP
d60e92d1 3289.B cpu
40943b9a
TK
3290CPU usage. User and system time, along with the number of context
3291switches this thread went through, usage of system and user time, and
3292finally the number of major and minor page faults. The CPU utilization
3293numbers are averages for the jobs in that reporting group, while the
3294context and fault counters are summed.
d60e92d1
AC
3295.TP
3296.B IO depths
40943b9a
TK
3297The distribution of I/O depths over the job lifetime. The numbers are
3298divided into powers of 2 and each entry covers depths from that value
3299up to those that are lower than the next entry \-\- e.g., 16= covers
3300depths from 16 to 31. Note that the range covered by a depth
3301distribution entry can be different to the range covered by the
3302equivalent \fBsubmit\fR/\fBcomplete\fR distribution entry.
3303.TP
3304.B IO submit
3305How many pieces of I/O were submitting in a single submit call. Each
3306entry denotes that amount and below, until the previous entry \-\- e.g.,
330716=100% means that we submitted anywhere between 9 to 16 I/Os per submit
3308call. Note that the range covered by a \fBsubmit\fR distribution entry can
3309be different to the range covered by the equivalent depth distribution
3310entry.
3311.TP
3312.B IO complete
3313Like the above \fBsubmit\fR number, but for completions instead.
3314.TP
3315.B IO issued rwt
3316The number of \fBread/write/trim\fR requests issued, and how many of them were
3317short or dropped.
d60e92d1 3318.TP
d3b9694d 3319.B IO latency
ee21ebee 3320These values are for \fBlatency_target\fR and related options. When
d3b9694d
VF
3321these options are engaged, this section describes the I/O depth required
3322to meet the specified latency target.
d60e92d1 3323.RE
d60e92d1 3324.P
40943b9a
TK
3325After each client has been listed, the group statistics are printed. They
3326will look like this:
3327.P
3328.nf
3329 Run status group 0 (all jobs):
3330 READ: bw=20.9MiB/s (21.9MB/s), 10.4MiB/s\-10.8MiB/s (10.9MB/s\-11.3MB/s), io=64.0MiB (67.1MB), run=2973\-3069msec
3331 WRITE: bw=1231KiB/s (1261kB/s), 616KiB/s\-621KiB/s (630kB/s\-636kB/s), io=64.0MiB (67.1MB), run=52747\-53223msec
3332.fi
3333.P
3334For each data direction it prints:
d60e92d1
AC
3335.RS
3336.TP
40943b9a
TK
3337.B bw
3338Aggregate bandwidth of threads in this group followed by the
3339minimum and maximum bandwidth of all the threads in this group.
338f2db5
SW
3340Values outside of brackets are power-of-2 format and those
3341within are the equivalent value in a power-of-10 format.
d60e92d1 3342.TP
40943b9a
TK
3343.B io
3344Aggregate I/O performed of all threads in this group. The
3345format is the same as \fBbw\fR.
d60e92d1 3346.TP
40943b9a
TK
3347.B run
3348The smallest and longest runtimes of the threads in this group.
d60e92d1 3349.RE
d60e92d1 3350.P
40943b9a
TK
3351And finally, the disk statistics are printed. This is Linux specific.
3352They will look like this:
3353.P
3354.nf
3355 Disk stats (read/write):
3356 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
3357.fi
3358.P
3359Each value is printed for both reads and writes, with reads first. The
3360numbers denote:
d60e92d1
AC
3361.RS
3362.TP
3363.B ios
3364Number of I/Os performed by all groups.
3365.TP
3366.B merge
007c7be9 3367Number of merges performed by the I/O scheduler.
d60e92d1
AC
3368.TP
3369.B ticks
3370Number of ticks we kept the disk busy.
3371.TP
40943b9a 3372.B in_queue
d60e92d1
AC
3373Total time spent in the disk queue.
3374.TP
3375.B util
40943b9a
TK
3376The disk utilization. A value of 100% means we kept the disk
3377busy constantly, 50% would be a disk idling half of the time.
d60e92d1 3378.RE
8423bd11 3379.P
40943b9a
TK
3380It is also possible to get fio to dump the current output while it is running,
3381without terminating the job. To do that, send fio the USR1 signal. You can
3382also get regularly timed dumps by using the \fB\-\-status\-interval\fR
3383parameter, or by creating a file in `/tmp' named
3384`fio\-dump\-status'. If fio sees this file, it will unlink it and dump the
3385current output status.
d60e92d1 3386.SH TERSE OUTPUT
40943b9a
TK
3387For scripted usage where you typically want to generate tables or graphs of the
3388results, fio can output the results in a semicolon separated format. The format
3389is one long line of values, such as:
d60e92d1 3390.P
40943b9a
TK
3391.nf
3392 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%
3393 A description of this job goes here.
3394.fi
d60e92d1 3395.P
4e757af1
VF
3396The job description (if provided) follows on a second line for terse v2.
3397It appears on the same line for other terse versions.
d60e92d1 3398.P
40943b9a
TK
3399To enable terse output, use the \fB\-\-minimal\fR or
3400`\-\-output\-format=terse' command line options. The
3401first value is the version of the terse output format. If the output has to be
3402changed for some reason, this number will be incremented by 1 to signify that
3403change.
d60e92d1 3404.P
40943b9a
TK
3405Split up, the format is as follows (comments in brackets denote when a
3406field was introduced or whether it's specific to some terse version):
d60e92d1 3407.P
40943b9a
TK
3408.nf
3409 terse version, fio version [v3], jobname, groupid, error
3410.fi
525c2bfa 3411.RS
40943b9a
TK
3412.P
3413.B
3414READ status:
525c2bfa 3415.RE
40943b9a
TK
3416.P
3417.nf
3418 Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec)
3419 Submission latency: min, max, mean, stdev (usec)
3420 Completion latency: min, max, mean, stdev (usec)
3421 Completion latency percentiles: 20 fields (see below)
3422 Total latency: min, max, mean, stdev (usec)
3423 Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev, number of samples [v5]
3424 IOPS [v5]: min, max, mean, stdev, number of samples
3425.fi
d60e92d1 3426.RS
40943b9a
TK
3427.P
3428.B
3429WRITE status:
a2c95580 3430.RE
40943b9a
TK
3431.P
3432.nf
3433 Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec)
3434 Submission latency: min, max, mean, stdev (usec)
3435 Completion latency: min, max, mean, stdev (usec)
3436 Completion latency percentiles: 20 fields (see below)
3437 Total latency: min, max, mean, stdev (usec)
3438 Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev, number of samples [v5]
3439 IOPS [v5]: min, max, mean, stdev, number of samples
3440.fi
a2c95580 3441.RS
40943b9a
TK
3442.P
3443.B
3444TRIM status [all but version 3]:
d60e92d1
AC
3445.RE
3446.P
40943b9a
TK
3447.nf
3448 Fields are similar to \fBREAD/WRITE\fR status.
3449.fi
a2c95580 3450.RS
a2c95580 3451.P
40943b9a 3452.B
d1429b5c 3453CPU usage:
d60e92d1
AC
3454.RE
3455.P
40943b9a
TK
3456.nf
3457 user, system, context switches, major faults, minor faults
3458.fi
d60e92d1 3459.RS
40943b9a
TK
3460.P
3461.B
3462I/O depths:
d60e92d1
AC
3463.RE
3464.P
40943b9a
TK
3465.nf
3466 <=1, 2, 4, 8, 16, 32, >=64
3467.fi
562c2d2f 3468.RS
40943b9a
TK
3469.P
3470.B
3471I/O latencies microseconds:
562c2d2f 3472.RE
40943b9a
TK
3473.P
3474.nf
3475 <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
3476.fi
562c2d2f 3477.RS
40943b9a
TK
3478.P
3479.B
3480I/O latencies milliseconds:
562c2d2f
DN
3481.RE
3482.P
40943b9a
TK
3483.nf
3484 <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
3485.fi
f2f788dd 3486.RS
40943b9a
TK
3487.P
3488.B
3489Disk utilization [v3]:
f2f788dd
JA
3490.RE
3491.P
40943b9a
TK
3492.nf
3493 disk name, read ios, write ios, read merges, write merges, read ticks, write ticks, time spent in queue, disk utilization percentage
3494.fi
562c2d2f 3495.RS
d60e92d1 3496.P
40943b9a
TK
3497.B
3498Additional Info (dependent on continue_on_error, default off):
d60e92d1 3499.RE
2fc26c3d 3500.P
40943b9a
TK
3501.nf
3502 total # errors, first error code
3503.fi
2fc26c3d
IC
3504.RS
3505.P
40943b9a
TK
3506.B
3507Additional Info (dependent on description being set):
3508.RE
3509.P
2fc26c3d 3510.nf
40943b9a
TK
3511 Text description
3512.fi
3513.P
3514Completion latency percentiles can be a grouping of up to 20 sets, so for the
3515terse output fio writes all of them. Each field will look like this:
3516.P
3517.nf
3518 1.00%=6112
3519.fi
3520.P
3521which is the Xth percentile, and the `usec' latency associated with it.
3522.P
3523For \fBDisk utilization\fR, all disks used by fio are shown. So for each disk there
3524will be a disk utilization section.
3525.P
3526Below is a single line containing short names for each of the fields in the
3527minimal output v3, separated by semicolons:
3528.P
3529.nf
3530 terse_version_3;fio_version;jobname;groupid;error;read_kb;read_bandwidth;read_iops;read_runtime_ms;read_slat_min;read_slat_max;read_slat_mean;read_slat_dev;read_clat_min;read_clat_max;read_clat_mean;read_clat_dev;read_clat_pct01;read_clat_pct02;read_clat_pct03;read_clat_pct04;read_clat_pct05;read_clat_pct06;read_clat_pct07;read_clat_pct08;read_clat_pct09;read_clat_pct10;read_clat_pct11;read_clat_pct12;read_clat_pct13;read_clat_pct14;read_clat_pct15;read_clat_pct16;read_clat_pct17;read_clat_pct18;read_clat_pct19;read_clat_pct20;read_tlat_min;read_lat_max;read_lat_mean;read_lat_dev;read_bw_min;read_bw_max;read_bw_agg_pct;read_bw_mean;read_bw_dev;write_kb;write_bandwidth;write_iops;write_runtime_ms;write_slat_min;write_slat_max;write_slat_mean;write_slat_dev;write_clat_min;write_clat_max;write_clat_mean;write_clat_dev;write_clat_pct01;write_clat_pct02;write_clat_pct03;write_clat_pct04;write_clat_pct05;write_clat_pct06;write_clat_pct07;write_clat_pct08;write_clat_pct09;write_clat_pct10;write_clat_pct11;write_clat_pct12;write_clat_pct13;write_clat_pct14;write_clat_pct15;write_clat_pct16;write_clat_pct17;write_clat_pct18;write_clat_pct19;write_clat_pct20;write_tlat_min;write_lat_max;write_lat_mean;write_lat_dev;write_bw_min;write_bw_max;write_bw_agg_pct;write_bw_mean;write_bw_dev;cpu_user;cpu_sys;cpu_csw;cpu_mjf;cpu_minf;iodepth_1;iodepth_2;iodepth_4;iodepth_8;iodepth_16;iodepth_32;iodepth_64;lat_2us;lat_4us;lat_10us;lat_20us;lat_50us;lat_100us;lat_250us;lat_500us;lat_750us;lat_1000us;lat_2ms;lat_4ms;lat_10ms;lat_20ms;lat_50ms;lat_100ms;lat_250ms;lat_500ms;lat_750ms;lat_1000ms;lat_2000ms;lat_over_2000ms;disk_name;disk_read_iops;disk_write_iops;disk_read_merges;disk_write_merges;disk_read_ticks;write_ticks;disk_queue_time;disk_util
2fc26c3d 3531.fi
4e757af1
VF
3532.P
3533In client/server mode terse output differs from what appears when jobs are run
3534locally. Disk utilization data is omitted from the standard terse output and
3535for v3 and later appears on its own separate line at the end of each terse
3536reporting cycle.
44c82dba
VF
3537.SH JSON OUTPUT
3538The \fBjson\fR output format is intended to be both human readable and convenient
3539for automated parsing. For the most part its sections mirror those of the
3540\fBnormal\fR output. The \fBruntime\fR value is reported in msec and the \fBbw\fR value is
3541reported in 1024 bytes per second units.
3542.fi
d9e557ab
VF
3543.SH JSON+ OUTPUT
3544The \fBjson+\fR output format is identical to the \fBjson\fR output format except that it
3545adds a full dump of the completion latency bins. Each \fBbins\fR object contains a
3546set of (key, value) pairs where keys are latency durations and values count how
3547many I/Os had completion latencies of the corresponding duration. For example,
3548consider:
d9e557ab 3549.RS
40943b9a 3550.P
d9e557ab
VF
3551"bins" : { "87552" : 1, "89600" : 1, "94720" : 1, "96768" : 1, "97792" : 1, "99840" : 1, "100864" : 2, "103936" : 6, "104960" : 534, "105984" : 5995, "107008" : 7529, ... }
3552.RE
40943b9a 3553.P
d9e557ab
VF
3554This data indicates that one I/O required 87,552ns to complete, two I/Os required
3555100,864ns to complete, and 7529 I/Os required 107,008ns to complete.
40943b9a 3556.P
d9e557ab 3557Also included with fio is a Python script \fBfio_jsonplus_clat2csv\fR that takes
338f2db5 3558json+ output and generates CSV-formatted latency data suitable for plotting.
40943b9a 3559.P
d9e557ab 3560The latency durations actually represent the midpoints of latency intervals.
40943b9a 3561For details refer to `stat.h' in the fio source.
29dbd1e5 3562.SH TRACE FILE FORMAT
40943b9a
TK
3563There are two trace file format that you can encounter. The older (v1) format is
3564unsupported since version 1.20\-rc3 (March 2008). It will still be described
29dbd1e5 3565below in case that you get an old trace and want to understand it.
29dbd1e5 3566.P
40943b9a
TK
3567In any case the trace is a simple text file with a single action per line.
3568.TP
29dbd1e5 3569.B Trace file format v1
40943b9a 3570Each line represents a single I/O action in the following format:
29dbd1e5 3571.RS
40943b9a
TK
3572.RS
3573.P
29dbd1e5 3574rw, offset, length
29dbd1e5
JA
3575.RE
3576.P
40943b9a
TK
3577where `rw=0/1' for read/write, and the `offset' and `length' entries being in bytes.
3578.P
3579This format is not supported in fio versions >= 1.20\-rc3.
3580.RE
3581.TP
29dbd1e5 3582.B Trace file format v2
40943b9a
TK
3583The second version of the trace file format was added in fio version 1.17. It
3584allows to access more then one file per trace and has a bigger set of possible
3585file actions.
29dbd1e5 3586.RS
40943b9a 3587.P
29dbd1e5 3588The first line of the trace file has to be:
40943b9a
TK
3589.RS
3590.P
3591"fio version 2 iolog"
3592.RE
3593.P
29dbd1e5 3594Following this can be lines in two different formats, which are described below.
40943b9a
TK
3595.P
3596.B
29dbd1e5 3597The file management format:
40943b9a
TK
3598.RS
3599filename action
29dbd1e5 3600.P
40943b9a 3601The `filename' is given as an absolute path. The `action' can be one of these:
29dbd1e5
JA
3602.RS
3603.TP
3604.B add
40943b9a 3605Add the given `filename' to the trace.
29dbd1e5
JA
3606.TP
3607.B open
40943b9a
TK
3608Open the file with the given `filename'. The `filename' has to have
3609been added with the \fBadd\fR action before.
29dbd1e5
JA
3610.TP
3611.B close
40943b9a
TK
3612Close the file with the given `filename'. The file has to have been
3613\fBopen\fRed before.
3614.RE
29dbd1e5 3615.RE