blob: fa54763dbc43fca5876f287fac908384d82b3d89 [file] [log] [blame]
Aaron Carrolld60e92d2007-09-17 10:32:59 +02001.TH fio 1 "September 2007" "User Manual"
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
15.BI \-\-output \fR=\fPfilename
16Write output to \fIfilename\fR.
17.TP
18.BI \-\-timeout \fR=\fPtimeout
19Limit run time to \fItimeout\fR seconds.
20.TP
21.B \-\-latency\-log
22Generate per-job latency logs.
23.TP
24.B \-\-bandwidth\-log
25Generate per-job bandwidth logs.
26.TP
27.B \-\-minimal
Aaron Carrolld1429b52007-09-18 08:10:57 +020028Print statistics in a terse, semicolon-delimited format.
Aaron Carrolld60e92d2007-09-17 10:32:59 +020029.TP
30.BI \-\-showcmd \fR=\fPjobfile
31Convert \fIjobfile\fR to a set of command-line options.
32.TP
33.B \-\-readonly
34Enable read-only safety checks.
35.TP
36.BI \-\-eta \fR=\fPwhen
37Specifies when real-time ETA estimate should be printed. \fIwhen\fR may
38be one of `always', `never' or `auto'.
39.TP
Aaron Carrollc0a5d352008-02-26 23:10:39 +010040.BI \-\-section \fR=\fPsec
41Only run section \fIsec\fR from job file.
42.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +020043.BI \-\-cmdhelp \fR=\fPcommand
44Print help information for \fIcommand\fR. May be `all' for all commands.
45.TP
Jens Axboe91837882008-02-05 12:02:07 +010046.BI \-\-debug \fR=\fPtype
47Enable verbose tracing of various fio actions. May be `all' for all types
48or individual types seperated by a comma (eg --debug=io,file). `help' will
49list all available tracing options.
50.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +020051.B \-\-help
52Display usage information and exit.
53.TP
54.B \-\-version
55Display version information and exit.
56.SH "JOB FILE FORMAT"
57Job files are in `ini' format. They consist of one or more
58job definitions, which begin with a job name in square brackets and
59extend to the next job name. The job name can be any ASCII string
60except `global', which has a special meaning. Following the job name is
61a sequence of zero or more parameters, one per line, that define the
62behavior of the job. Any line starting with a `;' or `#' character is
Aaron Carrolld1429b52007-09-18 08:10:57 +020063considered a comment and ignored.
Aaron Carrolld9956b62007-11-16 12:12:45 +010064.P
65If \fIjobfile\fR is specified as `-', the job file will be read from
66standard input.
Aaron Carrolld60e92d2007-09-17 10:32:59 +020067.SS "Global Section"
68The global section contains default parameters for jobs specified in the
69job file. A job is only affected by global sections residing above it,
70and there may be any number of global sections. Specific job definitions
71may override any parameter set in global sections.
72.SH "JOB PARAMETERS"
73.SS Types
74Some parameters may take arguments of a specific type. The types used are:
75.TP
76.I str
77String: a sequence of alphanumeric characters.
78.TP
79.I int
Aaron Carrolld60e92d2007-09-17 10:32:59 +020080SI integer: a whole number, possibly containing a suffix denoting the base unit
81of the value. Accepted suffixes are `k', 'M' and 'G', denoting kilo (1024),
Jens Axboef7fa2652009-03-09 14:20:20 +010082mega (1024*1024) and giga (1024*1024*1024) respectively. If prefixed with '0x',
83the value is assumed to be base 16 (hexadecimal).
Aaron Carrolld60e92d2007-09-17 10:32:59 +020084.TP
85.I bool
86Boolean: a true or false value. `0' denotes false, `1' denotes true.
87.TP
88.I irange
89Integer range: a range of integers specified in the format
Aaron Carrolld1429b52007-09-18 08:10:57 +020090\fIlower\fR:\fIupper\fR or \fIlower\fR\-\fIupper\fR. \fIlower\fR and
91\fIupper\fR may contain a suffix as described above. If an option allows two
92sets of ranges, they are separated with a `,' or `/' character. For example:
93`8\-8k/8M\-4G'.
Aaron Carrolld60e92d2007-09-17 10:32:59 +020094.SS "Parameter List"
95.TP
96.BI name \fR=\fPstr
Aaron Carrolld9956b62007-11-16 12:12:45 +010097May be used to override the job name. On the command line, this parameter
Aaron Carrolld60e92d2007-09-17 10:32:59 +020098has the special purpose of signalling the start of a new job.
99.TP
100.BI description \fR=\fPstr
101Human-readable description of the job. It is printed when the job is run, but
102otherwise has no special purpose.
103.TP
104.BI directory \fR=\fPstr
105Prefix filenames with this directory. Used to place files in a location other
106than `./'.
107.TP
108.BI filename \fR=\fPstr
109.B fio
110normally makes up a file name based on the job name, thread number, and file
Aaron Carrolld1429b52007-09-18 08:10:57 +0200111number. If you want to share files between threads in a job or several jobs,
112specify a \fIfilename\fR for each of them to override the default. If the I/O
113engine used is `net', \fIfilename\fR is the host and port to connect to in the
114format \fIhost\fR/\fIport\fR. If the I/O engine is file-based, you can specify
115a number of files by separating the names with a `:' character. `\-' is a
116reserved name, meaning stdin or stdout, depending on the read/write direction
117set.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200118.TP
Jens Axboe3ce9dca2009-06-10 08:55:21 +0200119.BI lockfile \fR=\fPstr
120Fio defaults to not locking any files before it does IO to them. If a file or
121file descriptor is shared, fio can serialize IO to that file to make the end
122result consistent. This is usual for emulating real workloads that share files.
123The lock modes are:
124.RS
125.RS
126.TP
127.B none
128No locking. This is the default.
129.TP
130.B exclusive
131Only one thread or process may do IO at the time, excluding all others.
132.TP
133.B readwrite
134Read-write locking on the file. Many readers may access the file at the same
135time, but writes get exclusive access.
136.RE
137.P
138The option may be post-fixed with a lock batch number. If set, then each
139thread/process may do that amount of IOs to the file before giving up the lock.
140Since lock acquisition is expensive, batching the lock/unlocks will speed up IO.
141.RE
142.P
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200143.BI opendir \fR=\fPstr
144Recursively open any files below directory \fIstr\fR.
145.TP
146.BI readwrite \fR=\fPstr "\fR,\fP rw" \fR=\fPstr
147Type of I/O pattern. Accepted values are:
148.RS
149.RS
150.TP
151.B read
Aaron Carrolld1429b52007-09-18 08:10:57 +0200152Sequential reads.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200153.TP
154.B write
Aaron Carrolld1429b52007-09-18 08:10:57 +0200155Sequential writes.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200156.TP
157.B randread
Aaron Carrolld1429b52007-09-18 08:10:57 +0200158Random reads.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200159.TP
160.B randwrite
Aaron Carrolld1429b52007-09-18 08:10:57 +0200161Random writes.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200162.TP
163.B rw
Aaron Carrolld1429b52007-09-18 08:10:57 +0200164Mixed sequential reads and writes.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200165.TP
166.B randrw
Aaron Carrolld1429b52007-09-18 08:10:57 +0200167Mixed random reads and writes.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200168.RE
169.P
Aaron Carrolld1429b52007-09-18 08:10:57 +0200170For mixed I/O, the default split is 50/50. For random I/O, the number of I/Os
171to perform before getting a new offset can be specified by appending
172`:\fIint\fR' to the pattern type. The default is 1.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200173.RE
174.TP
175.BI randrepeat \fR=\fPbool
176Seed the random number generator in a predictable way so results are repeatable
Aaron Carrolld1429b52007-09-18 08:10:57 +0200177across runs. Default: true.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200178.TP
179.BI fadvise_hint \fR=\fPbool
Aaron Carrolld1429b52007-09-18 08:10:57 +0200180Disable use of \fIposix_fadvise\fR\|(2) to advise the kernel what I/O patterns
181are likely to be issued. Default: true.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200182.TP
Jens Axboef7fa2652009-03-09 14:20:20 +0100183.BI size \fR=\fPint
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200184Total size of I/O for this job. \fBfio\fR will run until this many bytes have
185been transfered, unless limited by other options (\fBruntime\fR, for instance).
186Unless \fBnr_files\fR and \fBfilesize\fR options are given, this amount will be
187divided between the available files for the job.
188.TP
Jens Axboe3ce9dca2009-06-10 08:55:21 +0200189.BI fill_device \fR=\fPbool
190Sets size to something really large and waits for ENOSPC (no space left on
191device) as the terminating condition. Only makes sense with sequential write.
192For a read workload, the mount point will be filled first then IO started on
193the result.
194.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200195.BI filesize \fR=\fPirange
196Individual file sizes. May be a range, in which case \fBfio\fR will select sizes
Aaron Carrolld1429b52007-09-18 08:10:57 +0200197for files at random within the given range, limited to \fBsize\fR in total (if
198that is given). If \fBfilesize\fR is not specified, each created file is the
199same size.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200200.TP
Jens Axboef7fa2652009-03-09 14:20:20 +0100201.BI blocksize \fR=\fPint[,int] "\fR,\fB bs" \fR=\fPint[,int]
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200202Block size for I/O units. Default: 4k. Values for reads and writes can be
203specified seperately in the format \fIread\fR,\fIwrite\fR, either of
204which may be empty to leave that value at its default.
205.TP
Jens Axboe91837882008-02-05 12:02:07 +0100206.BI blocksize_range \fR=\fPirange[,irange] "\fR,\fB bsrange" \fR=\fPirange[,irange]
Aaron Carrolld1429b52007-09-18 08:10:57 +0200207Specify a range of I/O block sizes. The issued I/O unit will always be a
208multiple of the minimum size, unless \fBblocksize_unaligned\fR is set. Applies
Jens Axboe91837882008-02-05 12:02:07 +0100209to both reads and writes if only one range is given, but can be specified
210seperately with a comma seperating the values. Example: bsrange=1k-4k,2k-8k.
211Also (see \fBblocksize\fR).
212.TP
213.BI bssplit \fR=\fPstr
214This option allows even finer grained control of the block sizes issued,
215not just even splits between them. With this option, you can weight various
216block sizes for exact control of the issued IO for a job that has mixed
217block sizes. The format of the option is bssplit=blocksize/percentage,
218optionally adding as many definitions as needed seperated by a colon.
219Example: bssplit=4k/10:64k/50:32k/40 would issue 50% 64k blocks, 10% 4k
Jens Axboec83cdd32009-04-24 14:23:59 +0200220blocks and 40% 32k blocks. \fBbssplit\fR also supports giving separate
221splits to reads and writes. The format is identical to what the
222\fBbs\fR option accepts, the read and write parts are separated with a
223comma.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200224.TP
225.B blocksize_unaligned\fR,\fP bs_unaligned
Aaron Carrolld1429b52007-09-18 08:10:57 +0200226If set, any size in \fBblocksize_range\fR may be used. This typically won't
227work with direct I/O, as that normally requires sector alignment.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200228.TP
Jens Axboe2b7a01d2009-03-11 11:00:13 +0100229.BI blockalign \fR=\fPint[,int] "\fR,\fB ba" \fR=\fPint[,int]
Martin Steigerwald639ce0f2009-05-20 11:33:49 +0200230At what boundary to align random IO offsets. Defaults to the same as 'blocksize'
231the minimum blocksize given. Minimum alignment is typically 512b
Jens Axboe2b7a01d2009-03-11 11:00:13 +0100232for using direct IO, though it usually depends on the hardware block size.
233This option is mutually exclusive with using a random map for files, so it
234will turn off that option.
Jens Axboe43602662009-03-14 20:08:47 +0100235.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200236.B zero_buffers
237Initialise buffers with all zeros. Default: fill buffers with random data.
238.TP
Jens Axboe901bb992009-03-14 20:17:36 +0100239.B refill_buffers
240If this option is given, fio will refill the IO buffers on every submit. The
241default is to only fill it at init time and reuse that data. Only makes sense
242if zero_buffers isn't specified, naturally. If data verification is enabled,
243refill_buffers is also automatically enabled.
244.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200245.BI nrfiles \fR=\fPint
246Number of files to use for this job. Default: 1.
247.TP
248.BI openfiles \fR=\fPint
249Number of files to keep open at the same time. Default: \fBnrfiles\fR.
250.TP
251.BI file_service_type \fR=\fPstr
252Defines how files to service are selected. The following types are defined:
253.RS
254.RS
255.TP
256.B random
257Choose a file at random
258.TP
259.B roundrobin
260Round robin over open files (default).
Jens Axboe6b7f6852009-03-09 14:22:56 +0100261.B sequential
262Do each file in the set sequentially.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200263.RE
264.P
265The number of I/Os to issue before switching a new file can be specified by
266appending `:\fIint\fR' to the service type.
267.RE
268.TP
269.BI ioengine \fR=\fPstr
270Defines how the job issues I/O. The following types are defined:
271.RS
272.RS
273.TP
274.B sync
275Basic \fIread\fR\|(2) or \fIwrite\fR\|(2) I/O. \fIfseek\fR\|(2) is used to
276position the I/O location.
277.TP
gurudas paia31041e2007-10-23 15:12:30 +0200278.B psync
279Basic \fIpread\fR\|(2) or \fIpwrite\fR\|(2) I/O.
280.TP
Jens Axboe91837882008-02-05 12:02:07 +0100281.B vsync
282Basic \fIreadv\fR\|(2) or \fIwritev\fR\|(2) I/O. Will emulate queuing by
283coalescing adjacents IOs into a single submission.
284.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200285.B libaio
286Linux native asynchronous I/O.
287.TP
288.B posixaio
289glibc POSIX asynchronous I/O using \fIaio_read\fR\|(3) and \fIaio_write\fR\|(3).
290.TP
291.B mmap
Aaron Carrolld1429b52007-09-18 08:10:57 +0200292File is memory mapped with \fImmap\fR\|(2) and data copied using
293\fImemcpy\fR\|(3).
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200294.TP
295.B splice
Aaron Carrolld1429b52007-09-18 08:10:57 +0200296\fIsplice\fR\|(2) is used to transfer the data and \fIvmsplice\fR\|(2) to
297transfer data from user-space to the kernel.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200298.TP
299.B syslet-rw
300Use the syslet system calls to make regular read/write asynchronous.
301.TP
302.B sg
303SCSI generic sg v3 I/O. May be either synchronous using the SG_IO ioctl, or if
Aaron Carrolld1429b52007-09-18 08:10:57 +0200304the target is an sg character device, we use \fIread\fR\|(2) and
305\fIwrite\fR\|(2) for asynchronous I/O.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200306.TP
307.B null
308Doesn't transfer any data, just pretends to. Mainly used to exercise \fBfio\fR
309itself and for debugging and testing purposes.
310.TP
311.B net
312Transfer over the network. \fBfilename\fR must be set appropriately to
313`\fIhost\fR/\fIport\fR' regardless of data direction. If receiving, only the
314\fIport\fR argument is used.
315.TP
316.B netsplice
317Like \fBnet\fR, but uses \fIsplice\fR\|(2) and \fIvmsplice\fR\|(2) to map data
318and send/receive.
319.TP
gurudas pai53aec0a2007-10-05 13:20:18 +0200320.B cpuio
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200321Doesn't transfer any data, but burns CPU cycles according to \fBcpuload\fR and
322\fBcpucycles\fR parameters.
323.TP
324.B guasi
325The GUASI I/O engine is the Generic Userspace Asynchronous Syscall Interface
326approach to asycnronous I/O.
Aaron Carrolld1429b52007-09-18 08:10:57 +0200327.br
328See <http://www.xmailserver.org/guasi\-lib.html>.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200329.TP
330.B external
331Loads an external I/O engine object file. Append the engine filename as
332`:\fIenginepath\fR'.
333.RE
334.RE
335.TP
336.BI iodepth \fR=\fPint
337Number of I/O units to keep in flight against the file. Default: 1.
338.TP
339.BI iodepth_batch \fR=\fPint
340Number of I/Os to submit at once. Default: \fBiodepth\fR.
341.TP
Jens Axboe3ce9dca2009-06-10 08:55:21 +0200342.BI iodepth_batch_complete \fR=\fPint
343This defines how many pieces of IO to retrieve at once. It defaults to 1 which
344 means that we'll ask for a minimum of 1 IO in the retrieval process from the
345kernel. The IO retrieval will go on until we hit the limit set by
346\fBiodepth_low\fR. If this variable is set to 0, then fio will always check for
347completed events before queuing more IO. This helps reduce IO latency, at the
348cost of more retrieval system calls.
349.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200350.BI iodepth_low \fR=\fPint
351Low watermark indicating when to start filling the queue again. Default:
352\fBiodepth\fR.
353.TP
354.BI direct \fR=\fPbool
355If true, use non-buffered I/O (usually O_DIRECT). Default: false.
356.TP
357.BI buffered \fR=\fPbool
358If true, use buffered I/O. This is the opposite of the \fBdirect\fR parameter.
359Default: true.
360.TP
Jens Axboef7fa2652009-03-09 14:20:20 +0100361.BI offset \fR=\fPint
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200362Offset in the file to start I/O. Data before the offset will not be touched.
363.TP
364.BI fsync \fR=\fPint
Aaron Carrolld1429b52007-09-18 08:10:57 +0200365How many I/Os to perform before issuing an \fBfsync\fR\|(2) of dirty data. If
3660, don't sync. Default: 0.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200367.TP
Jens Axboe5f9099e2009-06-16 22:40:26 +0200368.BI fdatasync \fR=\fPint
369Like \fBfsync\fR, but uses \fBfdatasync\fR\|(2) instead to only sync the
370data parts of the file. Default: 0.
371.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200372.BI overwrite \fR=\fPbool
Aaron Carrolld1429b52007-09-18 08:10:57 +0200373If writing, setup the file first and do overwrites. Default: false.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200374.TP
375.BI end_fsync \fR=\fPbool
Aaron Carrolld1429b52007-09-18 08:10:57 +0200376Sync file contents when job exits. Default: false.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200377.TP
378.BI fsync_on_close \fR=\fPbool
379If true, sync file contents on close. This differs from \fBend_fsync\fR in that
Aaron Carrolld1429b52007-09-18 08:10:57 +0200380it will happen on every close, not just at the end of the job. Default: false.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200381.TP
382.BI rwmixcycle \fR=\fPint
383How many milliseconds before switching between reads and writes for a mixed
384workload. Default: 500ms.
385.TP
386.BI rwmixread \fR=\fPint
387Percentage of a mixed workload that should be reads. Default: 50.
388.TP
389.BI rwmixwrite \fR=\fPint
Aaron Carrolld1429b52007-09-18 08:10:57 +0200390Percentage of a mixed workload that should be writes. If \fBrwmixread\fR and
Jens Axboec35dd7a2009-06-10 08:39:16 +0200391\fBrwmixwrite\fR are given and do not sum to 100%, the latter of the two
392overrides the first. This may interfere with a given rate setting, if fio is
393asked to limit reads or writes to a certain rate. If that is the case, then
394the distribution may be skewed. Default: 50.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200395.TP
396.B norandommap
397Normally \fBfio\fR will cover every block of the file when doing random I/O. If
398this parameter is given, a new offset will be chosen without looking at past
399I/O history. This parameter is mutually exclusive with \fBverify\fR.
400.TP
Jens Axboe3ce9dca2009-06-10 08:55:21 +0200401.B softrandommap
402See \fBnorandommap\fR. If fio runs with the random block map enabled and it
403fails to allocate the map, if this option is set it will continue without a
404random block map. As coverage will not be as complete as with random maps, this
405option is disabled by default.
406.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200407.BI nice \fR=\fPint
408Run job with given nice value. See \fInice\fR\|(2).
409.TP
410.BI prio \fR=\fPint
411Set I/O priority value of this job between 0 (highest) and 7 (lowest). See
412\fIionice\fR\|(1).
413.TP
414.BI prioclass \fR=\fPint
415Set I/O priority class. See \fIionice\fR\|(1).
416.TP
417.BI thinktime \fR=\fPint
418Stall job for given number of microseconds between issuing I/Os.
419.TP
420.BI thinktime_spin \fR=\fPint
421Pretend to spend CPU time for given number of microseconds, sleeping the rest
422of the time specified by \fBthinktime\fR. Only valid if \fBthinktime\fR is set.
423.TP
424.BI thinktime_blocks \fR=\fPint
425Number of blocks to issue before waiting \fBthinktime\fR microseconds.
426Default: 1.
427.TP
428.BI rate \fR=\fPint
Jens Axboec35dd7a2009-06-10 08:39:16 +0200429Cap bandwidth used by this job. The number is in bytes/sec, the normal postfix
430rules apply. You can use \fBrate\fR=500k to limit reads and writes to 500k each,
431or you can specify read and writes separately. Using \fBrate\fR=1m,500k would
432limit reads to 1MB/sec and writes to 500KB/sec. Capping only reads or writes
433can be done with \fBrate\fR=,500k or \fBrate\fR=500k,. The former will only
434limit writes (to 500KB/sec), the latter will only limit reads.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200435.TP
436.BI ratemin \fR=\fPint
437Tell \fBfio\fR to do whatever it can to maintain at least the given bandwidth.
Jens Axboec35dd7a2009-06-10 08:39:16 +0200438Failing to meet this requirement will cause the job to exit. The same format
439as \fBrate\fR is used for read vs write separation.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200440.TP
441.BI rate_iops \fR=\fPint
Jens Axboec35dd7a2009-06-10 08:39:16 +0200442Cap the bandwidth to this number of IOPS. Basically the same as rate, just
443specified independently of bandwidth. The same format as \fBrate\fR is used for
444read vs write seperation. If \fBblocksize\fR is a range, the smallest block
445size is used as the metric.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200446.TP
447.BI rate_iops_min \fR=\fPint
Jens Axboec35dd7a2009-06-10 08:39:16 +0200448If this rate of I/O is not met, the job will exit. The same format as \fBrate\fR
449is used for read vs write seperation.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200450.TP
451.BI ratecycle \fR=\fPint
452Average bandwidth for \fBrate\fR and \fBratemin\fR over this number of
453milliseconds. Default: 1000ms.
454.TP
455.BI cpumask \fR=\fPint
456Set CPU affinity for this job. \fIint\fR is a bitmask of allowed CPUs the job
457may run on. See \fBsched_setaffinity\fR\|(2).
458.TP
459.BI cpus_allowed \fR=\fPstr
460Same as \fBcpumask\fR, but allows a comma-delimited list of CPU numbers.
461.TP
462.BI startdelay \fR=\fPint
463Delay start of job for the specified number of seconds.
464.TP
465.BI runtime \fR=\fPint
466Terminate processing after the specified number of seconds.
467.TP
468.B time_based
469If given, run for the specified \fBruntime\fR duration even if the files are
470completely read or written. The same workload will be repeated as many times
471as \fBruntime\fR allows.
472.TP
Jens Axboe901bb992009-03-14 20:17:36 +0100473.BI ramp_time \fR=\fPint
474If set, fio will run the specified workload for this amount of time before
475logging any performance numbers. Useful for letting performance settle before
476logging results, thus minimizing the runtime required for stable results. Note
Jens Axboec35dd7a2009-06-10 08:39:16 +0200477that the \fBramp_time\fR is considered lead in time for a job, thus it will
478increase the total runtime if a special timeout or runtime is specified.
Jens Axboe901bb992009-03-14 20:17:36 +0100479.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200480.BI invalidate \fR=\fPbool
481Invalidate buffer-cache for the file prior to starting I/O. Default: true.
482.TP
483.BI sync \fR=\fPbool
484Use synchronous I/O for buffered writes. For the majority of I/O engines,
Aaron Carrolld1429b52007-09-18 08:10:57 +0200485this means using O_SYNC. Default: false.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200486.TP
487.BI iomem \fR=\fPstr "\fR,\fP mem" \fR=\fPstr
488Allocation method for I/O unit buffer. Allowed values are:
489.RS
490.RS
491.TP
492.B malloc
493Allocate memory with \fImalloc\fR\|(3).
494.TP
495.B shm
496Use shared memory buffers allocated through \fIshmget\fR\|(2).
497.TP
498.B shmhuge
499Same as \fBshm\fR, but use huge pages as backing.
500.TP
501.B mmap
502Use \fImmap\fR\|(2) for allocation. Uses anonymous memory unless a filename
503is given after the option in the format `:\fIfile\fR'.
504.TP
505.B mmaphuge
506Same as \fBmmap\fR, but use huge files as backing.
507.RE
508.P
509The amount of memory allocated is the maximum allowed \fBblocksize\fR for the
510job multiplied by \fBiodepth\fR. For \fBshmhuge\fR or \fBmmaphuge\fR to work,
511the system must have free huge pages allocated. \fBmmaphuge\fR also needs to
512have hugetlbfs mounted, and \fIfile\fR must point there.
513.RE
514.TP
Jens Axboed529ee12009-07-01 10:33:03 +0200515.BI iomem_align \fR=\fPint
516This indiciates the memory alignment of the IO memory buffers. Note that the
517given alignment is applied to the first IO unit buffer, if using \fBiodepth\fR
518the alignment of the following buffers are given by the \fBbs\fR used. In
519other words, if using a \fBbs\fR that is a multiple of the page sized in the
520system, all buffers will be aligned to this value. If using a \fBbs\fR that
521is not page aligned, the alignment of subsequent IO memory buffers is the
522sum of the \fBiomem_align\fR and \fBbs\fR used.
523.TP
Jens Axboef7fa2652009-03-09 14:20:20 +0100524.BI hugepage\-size \fR=\fPint
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200525Defines the size of a huge page. Must be at least equal to the system setting.
526Should be a multiple of 1MiB. Default: 4MiB.
527.TP
528.B exitall
529Terminate all jobs when one finishes. Default: wait for each job to finish.
530.TP
531.BI bwavgtime \fR=\fPint
532Average bandwidth calculations over the given time in milliseconds. Default:
533500ms.
534.TP
535.BI create_serialize \fR=\fPbool
Aaron Carrolld1429b52007-09-18 08:10:57 +0200536If true, serialize file creation for the jobs. Default: true.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200537.TP
538.BI create_fsync \fR=\fPbool
539\fIfsync\fR\|(2) data file after creation. Default: true.
540.TP
Jens Axboe6b7f6852009-03-09 14:22:56 +0100541.BI create_on_open \fR=\fPbool
542If true, the files are not created until they are opened for IO by the job.
543.TP
Jens Axboee9f48472009-06-03 12:14:08 +0200544.BI pre_read \fR=\fPbool
545If this is given, files will be pre-read into memory before starting the given
546IO operation. This will also clear the \fR \fBinvalidate\fR flag, since it is
Jens Axboe9c0d2242009-07-01 12:26:28 +0200547pointless to pre-read and then drop the cache. This will only work for IO
548engines that are seekable, since they allow you to read the same data
549multiple times. Thus it will not work on eg network or splice IO.
Jens Axboee9f48472009-06-03 12:14:08 +0200550.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200551.BI unlink \fR=\fPbool
552Unlink job files when done. Default: false.
553.TP
554.BI loops \fR=\fPint
555Specifies the number of iterations (runs of the same workload) of this job.
556Default: 1.
557.TP
558.BI do_verify \fR=\fPbool
559Run the verify phase after a write phase. Only valid if \fBverify\fR is set.
560Default: true.
561.TP
562.BI verify \fR=\fPstr
563Method of verifying file contents after each iteration of the job. Allowed
564values are:
565.RS
566.RS
567.TP
568.B md5 crc16 crc32 crc64 crc7 sha256 sha512
569Store appropriate checksum in the header of each block.
570.TP
571.B meta
572Write extra information about each I/O (timestamp, block number, etc.). The
573block number is verified.
574.TP
575.B pattern
576Fill I/O buffers with a specific pattern that is used to verify. The pattern is
577specified by appending `:\fIint\fR' to the parameter. \fIint\fR cannot be larger
578than 32-bits.
579.TP
580.B null
581Pretend to verify. Used for testing internals.
582.RE
583.RE
584.TP
585.BI verify_sort \fR=\fPbool
586If true, written verify blocks are sorted if \fBfio\fR deems it to be faster to
587read them back in a sorted manner. Default: true.
588.TP
Jens Axboef7fa2652009-03-09 14:20:20 +0100589.BI verify_offset \fR=\fPint
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200590Swap the verification header with data somewhere else in the block before
Aaron Carrolld1429b52007-09-18 08:10:57 +0200591writing. It is swapped back before verifying.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200592.TP
Jens Axboef7fa2652009-03-09 14:20:20 +0100593.BI verify_interval \fR=\fPint
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200594Write the verification header for this number of bytes, which should divide
595\fBblocksize\fR. Default: \fBblocksize\fR.
596.TP
597.BI verify_fatal \fR=\fPbool
598If true, exit the job on the first observed verification failure. Default:
599false.
600.TP
Jens Axboee8462bd2009-07-06 12:59:04 +0200601.BI verify_async \fR=\fPint
602Fio will normally verify IO inline from the submitting thread. This option
603takes an integer describing how many async offload threads to create for IO
604verification instead, causing fio to offload the duty of verifying IO contents
Jens Axboec85c3242009-07-06 14:12:57 +0200605to one or more separate threads. If using this offload option, even sync IO
606engines can benefit from using an \fBiodepth\fR setting higher than 1, as it
607allows them to have IO in flight while verifies are running.
Jens Axboee8462bd2009-07-06 12:59:04 +0200608.TP
609.BI verify_async_cpus \fR=\fPstr
610Tell fio to set the given CPU affinity on the async IO verification threads.
611See \fBcpus_allowed\fP for the format used.
612.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200613.B stonewall
Aaron Carrolld1429b52007-09-18 08:10:57 +0200614Wait for preceeding jobs in the job file to exit before starting this one.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200615\fBstonewall\fR implies \fBnew_group\fR.
616.TP
617.B new_group
618Start a new reporting group. If not given, all jobs in a file will be part
619of the same reporting group, unless separated by a stonewall.
620.TP
621.BI numjobs \fR=\fPint
622Number of clones (processes/threads performing the same workload) of this job.
623Default: 1.
624.TP
625.B group_reporting
626If set, display per-group reports instead of per-job when \fBnumjobs\fR is
627specified.
628.TP
629.B thread
630Use threads created with \fBpthread_create\fR\|(3) instead of processes created
631with \fBfork\fR\|(2).
632.TP
Jens Axboef7fa2652009-03-09 14:20:20 +0100633.BI zonesize \fR=\fPint
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200634Divide file into zones of the specified size in bytes. See \fBzoneskip\fR.
635.TP
Jens Axboef7fa2652009-03-09 14:20:20 +0100636.BI zoneskip \fR=\fPint
Aaron Carrolld1429b52007-09-18 08:10:57 +0200637Skip the specified number of bytes when \fBzonesize\fR bytes of data have been
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200638read.
639.TP
640.BI write_iolog \fR=\fPstr
641Write the issued I/O patterns to the specified file.
642.TP
643.BI read_iolog \fR=\fPstr
644Replay the I/O patterns contained in the specified file generated by
645\fBwrite_iolog\fR, or may be a \fBblktrace\fR binary file.
646.TP
Jens Axboe901bb992009-03-14 20:17:36 +0100647.B write_bw_log \fR=\fPstr
648If given, write a bandwidth log of the jobs in this job file. Can be used to
649store data of the bandwidth of the jobs in their lifetime. The included
650fio_generate_plots script uses gnuplot to turn these text files into nice
651graphs. See \fBwrite_log_log\fR for behaviour of given filename. For this
652option, the postfix is _bw.log.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200653.TP
654.B write_lat_log
Jens Axboe901bb992009-03-14 20:17:36 +0100655Same as \fBwrite_bw_log\fR, but writes I/O completion latencies. If no
656filename is given with this option, the default filename of "jobname_type.log"
657is used. Even if the filename is given, fio will still append the type of log.
658.TP
659.B disable_clat \fR=\fPbool
660Disable measurements of completion latency numbers. Useful only for cutting
661back the number of calls to gettimeofday, as that does impact performance at
662really high IOPS rates. Note that to really get rid of a large amount of these
663calls, this option must be used with disable_slat and disable_bw as well.
664.TP
665.B disable_slat \fR=\fPbool
666Disable measurements of submission latency numbers. See \fBdisable_clat\fR.
667.TP
668.B disable_bw_measurement \fR=\fPbool
669Disable measurements of throughput/bandwidth numbers. See \fBdisable_clat\fR.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200670.TP
Jens Axboef7fa2652009-03-09 14:20:20 +0100671.BI lockmem \fR=\fPint
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200672Pin the specified amount of memory with \fBmlock\fR\|(2). Can be used to
673simulate a smaller amount of memory.
674.TP
675.BI exec_prerun \fR=\fPstr
676Before running the job, execute the specified command with \fBsystem\fR\|(3).
677.TP
678.BI exec_postrun \fR=\fPstr
679Same as \fBexec_prerun\fR, but the command is executed after the job completes.
680.TP
681.BI ioscheduler \fR=\fPstr
682Attempt to switch the device hosting the file to the specified I/O scheduler.
683.TP
684.BI cpuload \fR=\fPint
685If the job is a CPU cycle-eater, attempt to use the specified percentage of
686CPU cycles.
687.TP
688.BI cpuchunks \fR=\fPint
689If the job is a CPU cycle-eater, split the load into cycles of the
690given time in milliseconds.
691.TP
692.BI disk_util \fR=\fPbool
Aaron Carrolld1429b52007-09-18 08:10:57 +0200693Generate disk utilization statistics if the platform supports it. Default: true.
Jens Axboe901bb992009-03-14 20:17:36 +0100694.TP
695.BI gtod_reduce \fR=\fPbool
696Enable all of the gettimeofday() reducing options (disable_clat, disable_slat,
697disable_bw) plus reduce precision of the timeout somewhat to really shrink the
698gettimeofday() call count. With this option enabled, we only do about 0.4% of
699the gtod() calls we would have done if all time keeping was enabled.
700.TP
701.BI gtod_cpu \fR=\fPint
702Sometimes it's cheaper to dedicate a single thread of execution to just getting
703the current time. Fio (and databases, for instance) are very intensive on
704gettimeofday() calls. With this option, you can set one CPU aside for doing
705nothing but logging current time to a shared memory location. Then the other
706threads/processes that run IO workloads need only copy that segment, instead of
707entering the kernel with a gettimeofday() call. The CPU set aside for doing
708these time calls will be excluded from other uses. Fio will manually clear it
709from the CPU mask of other jobs.
Radha Ramachandranf2bba182009-06-15 08:40:16 +0200710.TP
711.BI continue_on_error \fR=\fPbool
712Normally fio will exit the job on the first observed failure. If this option is
713set, fio will continue the job when there is a 'non-fatal error'
714(\fBEIO\fR or \fBEILSEQ\fR) until the runtime is exceeded or the I/O size
715specified is completed. If this option is used, there are two more stats that
716are appended, the total error count and the first error. The error field given
717in the stats is the first error that was hit during the run.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200718.SH OUTPUT
Aaron Carrolld1429b52007-09-18 08:10:57 +0200719While running, \fBfio\fR will display the status of the created jobs. For
720example:
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200721.RS
Aaron Carrolld1429b52007-09-18 08:10:57 +0200722.P
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200723Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
724.RE
725.P
Aaron Carrolld1429b52007-09-18 08:10:57 +0200726The characters in the first set of brackets denote the current status of each
727threads. The possible values are:
728.P
729.PD 0
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200730.RS
731.TP
732.B P
733Setup but not started.
734.TP
735.B C
736Thread created.
737.TP
738.B I
739Initialized, waiting.
740.TP
741.B R
742Running, doing sequential reads.
743.TP
744.B r
745Running, doing random reads.
746.TP
747.B W
748Running, doing sequential writes.
749.TP
750.B w
751Running, doing random writes.
752.TP
753.B M
754Running, doing mixed sequential reads/writes.
755.TP
756.B m
757Running, doing mixed random reads/writes.
758.TP
759.B F
760Running, currently waiting for \fBfsync\fR\|(2).
761.TP
762.B V
763Running, verifying written data.
764.TP
765.B E
766Exited, not reaped by main thread.
767.TP
768.B \-
769Exited, thread reaped.
770.RE
Aaron Carrolld1429b52007-09-18 08:10:57 +0200771.PD
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200772.P
773The second set of brackets shows the estimated completion percentage of
774the current group. The third set shows the read and write I/O rate,
775respectively. Finally, the estimated run time of the job is displayed.
776.P
777When \fBfio\fR completes (or is interrupted by Ctrl-C), it will show data
778for each thread, each group of threads, and each disk, in that order.
779.P
780Per-thread statistics first show the threads client number, group-id, and
781error code. The remaining figures are as follows:
782.RS
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200783.TP
784.B io
785Number of megabytes of I/O performed.
786.TP
787.B bw
788Average data rate (bandwidth).
789.TP
790.B runt
791Threads run time.
792.TP
793.B slat
794Submission latency minimum, maximum, average and standard deviation. This is
795the time it took to submit the I/O.
796.TP
797.B clat
798Completion latency minimum, maximum, average and standard deviation. This
799is the time between submission and completion.
800.TP
801.B bw
802Bandwidth minimum, maximum, percentage of aggregate bandwidth received, average
803and standard deviation.
804.TP
805.B cpu
806CPU usage statistics. Includes user and system time, number of context switches
807this thread went through and number of major and minor page faults.
808.TP
809.B IO depths
810Distribution of I/O depths. Each depth includes everything less than (or equal)
811to it, but greater than the previous depth.
812.TP
813.B IO issued
814Number of read/write requests issued, and number of short read/write requests.
815.TP
816.B IO latencies
817Distribution of I/O completion latencies. The numbers follow the same pattern
818as \fBIO depths\fR.
819.RE
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200820.P
821The group statistics show:
Aaron Carrolld1429b52007-09-18 08:10:57 +0200822.PD 0
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200823.RS
824.TP
825.B io
826Number of megabytes I/O performed.
827.TP
828.B aggrb
829Aggregate bandwidth of threads in the group.
830.TP
831.B minb
832Minimum average bandwidth a thread saw.
833.TP
834.B maxb
835Maximum average bandwidth a thread saw.
836.TP
837.B mint
Aaron Carrolld1429b52007-09-18 08:10:57 +0200838Shortest runtime of threads in the group.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200839.TP
840.B maxt
841Longest runtime of threads in the group.
842.RE
Aaron Carrolld1429b52007-09-18 08:10:57 +0200843.PD
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200844.P
845Finally, disk statistics are printed with reads first:
Aaron Carrolld1429b52007-09-18 08:10:57 +0200846.PD 0
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200847.RS
848.TP
849.B ios
850Number of I/Os performed by all groups.
851.TP
852.B merge
853Number of merges in the I/O scheduler.
854.TP
855.B ticks
856Number of ticks we kept the disk busy.
857.TP
858.B io_queue
859Total time spent in the disk queue.
860.TP
861.B util
862Disk utilization.
863.RE
Aaron Carrolld1429b52007-09-18 08:10:57 +0200864.PD
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200865.SH TERSE OUTPUT
866If the \fB\-\-minimal\fR option is given, the results will be printed in a
867semicolon-delimited format suitable for scripted use. The fields are:
868.P
869.RS
870.B jobname, groupid, error
871.P
872Read status:
873.RS
874.B KiB I/O, bandwidth \fR(KiB/s)\fP, runtime \fR(ms)\fP
875.P
876Submission latency:
877.RS
878.B min, max, mean, standard deviation
879.RE
880Completion latency:
881.RS
882.B min, max, mean, standard deviation
883.RE
884Bandwidth:
885.RS
886.B min, max, aggregate percentage of total, mean, standard deviation
887.RE
888.RE
889.P
890Write status:
891.RS
892.B KiB I/O, bandwidth \fR(KiB/s)\fP, runtime \fR(ms)\fP
893.P
894Submission latency:
895.RS
896.B min, max, mean, standard deviation
897.RE
898Completion latency:
899.RS
900.B min, max, mean, standard deviation
901.RE
902Bandwidth:
903.RS
904.B min, max, aggregate percentage of total, mean, standard deviation
905.RE
906.RE
907.P
Aaron Carrolld1429b52007-09-18 08:10:57 +0200908CPU usage:
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200909.RS
Carl Henrik Lundebd2626f2008-06-12 09:17:46 +0200910.B user, system, context switches, major page faults, minor page faults
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200911.RE
912.P
913IO depth distribution:
914.RS
915.B <=1, 2, 4, 8, 16, 32, >=64
916.RE
917.P
918IO latency distribution (ms):
919.RS
920.B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, >=2000
921.RE
922.P
923.B text description
924.RE
925.SH AUTHORS
926.B fio
927was written by Jens Axboe <jens.axboe@oracle.com>.
Aaron Carrolld1429b52007-09-18 08:10:57 +0200928.br
929This man page was written by Aaron Carroll <aaronc@cse.unsw.edu.au> based
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200930on documentation by Jens Axboe.
931.SH "REPORTING BUGS"
Jens Axboe482900c2009-06-02 12:15:51 +0200932Report bugs to the \fBfio\fR mailing list <fio@vger.kernel.org>.
Aaron Carrolld1429b52007-09-18 08:10:57 +0200933See \fBREADME\fR.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200934.SH "SEE ALSO"
Aaron Carrolld1429b52007-09-18 08:10:57 +0200935For further documentation see \fBHOWTO\fR and \fBREADME\fR.
936.br
937Sample jobfiles are available in the \fBexamples\fR directory.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200938