blob: e48a16a19e59484fdedfb10a56b3afb3113f9de1 [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
Martin Steigerwald5982a922011-06-27 16:07:24 +020048or individual types separated by a comma (eg \-\-debug=io,file). `help' will
Jens Axboe91837882008-02-05 12:02:07 +010049list 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
Jens Axboeb09da8f2009-07-17 23:16:17 +020081of the value. Accepted suffixes are `k', 'M', 'G', 'T', and 'P', denoting
82kilo (1024), mega (1024^2), giga (1024^3), tera (1024^4), and peta (1024^5)
83respectively. The suffix is not case sensitive. If prefixed with '0x', the
Martin Steigerwald5982a922011-06-27 16:07:24 +020084value is assumed to be base 16 (hexadecimal). A suffix may include a trailing 'b',
85for instance 'kb' is identical to 'k'. You can specify a base 10 value
Jens Axboe57fc29f2010-06-23 22:24:07 +020086by using 'KiB', 'MiB', 'GiB', etc. This is useful for disk drives where
87values are often given in base 10 values. Specifying '30GiB' will get you
8830*1000^3 bytes.
Aaron Carrolld60e92d2007-09-17 10:32:59 +020089.TP
90.I bool
91Boolean: a true or false value. `0' denotes false, `1' denotes true.
92.TP
93.I irange
94Integer range: a range of integers specified in the format
Aaron Carrolld1429b52007-09-18 08:10:57 +020095\fIlower\fR:\fIupper\fR or \fIlower\fR\-\fIupper\fR. \fIlower\fR and
96\fIupper\fR may contain a suffix as described above. If an option allows two
97sets of ranges, they are separated with a `,' or `/' character. For example:
98`8\-8k/8M\-4G'.
Aaron Carrolld60e92d2007-09-17 10:32:59 +020099.SS "Parameter List"
100.TP
101.BI name \fR=\fPstr
Aaron Carrolld9956b62007-11-16 12:12:45 +0100102May be used to override the job name. On the command line, this parameter
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200103has the special purpose of signalling the start of a new job.
104.TP
105.BI description \fR=\fPstr
106Human-readable description of the job. It is printed when the job is run, but
107otherwise has no special purpose.
108.TP
109.BI directory \fR=\fPstr
110Prefix filenames with this directory. Used to place files in a location other
111than `./'.
112.TP
113.BI filename \fR=\fPstr
114.B fio
115normally makes up a file name based on the job name, thread number, and file
Aaron Carrolld1429b52007-09-18 08:10:57 +0200116number. If you want to share files between threads in a job or several jobs,
117specify a \fIfilename\fR for each of them to override the default. If the I/O
118engine used is `net', \fIfilename\fR is the host and port to connect to in the
119format \fIhost\fR/\fIport\fR. If the I/O engine is file-based, you can specify
120a number of files by separating the names with a `:' character. `\-' is a
121reserved name, meaning stdin or stdout, depending on the read/write direction
122set.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200123.TP
Jens Axboe3ce9dca2009-06-10 08:55:21 +0200124.BI lockfile \fR=\fPstr
125Fio defaults to not locking any files before it does IO to them. If a file or
126file descriptor is shared, fio can serialize IO to that file to make the end
127result consistent. This is usual for emulating real workloads that share files.
128The lock modes are:
129.RS
130.RS
131.TP
132.B none
133No locking. This is the default.
134.TP
135.B exclusive
136Only one thread or process may do IO at the time, excluding all others.
137.TP
138.B readwrite
139Read-write locking on the file. Many readers may access the file at the same
140time, but writes get exclusive access.
141.RE
142.P
143The option may be post-fixed with a lock batch number. If set, then each
144thread/process may do that amount of IOs to the file before giving up the lock.
145Since lock acquisition is expensive, batching the lock/unlocks will speed up IO.
146.RE
147.P
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200148.BI opendir \fR=\fPstr
149Recursively open any files below directory \fIstr\fR.
150.TP
151.BI readwrite \fR=\fPstr "\fR,\fP rw" \fR=\fPstr
152Type of I/O pattern. Accepted values are:
153.RS
154.RS
155.TP
156.B read
Aaron Carrolld1429b52007-09-18 08:10:57 +0200157Sequential reads.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200158.TP
159.B write
Aaron Carrolld1429b52007-09-18 08:10:57 +0200160Sequential writes.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200161.TP
162.B randread
Aaron Carrolld1429b52007-09-18 08:10:57 +0200163Random reads.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200164.TP
165.B randwrite
Aaron Carrolld1429b52007-09-18 08:10:57 +0200166Random writes.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200167.TP
168.B rw
Aaron Carrolld1429b52007-09-18 08:10:57 +0200169Mixed sequential reads and writes.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200170.TP
171.B randrw
Aaron Carrolld1429b52007-09-18 08:10:57 +0200172Mixed random reads and writes.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200173.RE
174.P
Jens Axboe38dad622010-07-20 14:46:00 -0600175For mixed I/O, the default split is 50/50. For certain types of io the result
176may still be skewed a bit, since the speed may be different. It is possible to
177specify a number of IO's to do before getting a new offset, this is one by
178appending a `:\fI<nr>\fR to the end of the string given. For a random read, it
179would look like \fBrw=randread:8\fR for passing in an offset modifier with a
180value of 8. See the \fBrw_sequencer\fR option.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200181.RE
182.TP
Jens Axboe38dad622010-07-20 14:46:00 -0600183.BI rw_sequencer \fR=\fPstr
184If an offset modifier is given by appending a number to the \fBrw=<str>\fR line,
185then this option controls how that number modifies the IO offset being
186generated. Accepted values are:
187.RS
188.RS
189.TP
190.B sequential
191Generate sequential offset
192.TP
193.B identical
194Generate the same offset
195.RE
196.P
197\fBsequential\fR is only useful for random IO, where fio would normally
198generate a new random offset for every IO. If you append eg 8 to randread, you
199would get a new random offset for every 8 IO's. The result would be a seek for
200only every 8 IO's, instead of for every IO. Use \fBrw=randread:8\fR to specify
201that. As sequential IO is already sequential, setting \fBsequential\fR for that
202would not result in any differences. \fBidentical\fR behaves in a similar
203fashion, except it sends the same offset 8 number of times before generating a
204new offset.
205.RE
206.P
207.TP
Jens Axboe90fef2d2009-07-17 22:33:32 +0200208.BI kb_base \fR=\fPint
209The base unit for a kilobyte. The defacto base is 2^10, 1024. Storage
210manufacturers like to use 10^3 or 1000 as a base ten unit instead, for obvious
211reasons. Allow values are 1024 or 1000, with 1024 being the default.
212.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200213.BI randrepeat \fR=\fPbool
214Seed the random number generator in a predictable way so results are repeatable
Aaron Carrolld1429b52007-09-18 08:10:57 +0200215across runs. Default: true.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200216.TP
Jens Axboe2615cc42011-03-28 09:35:09 +0200217.BI use_os_rand \fR=\fPbool
218Fio can either use the random generator supplied by the OS to generator random
219offsets, or it can use it's own internal generator (based on Tausworthe).
220Default is to use the internal generator, which is often of better quality and
221faster. Default: false.
222.TP
Eric Gourioua596f042011-06-17 09:11:45 +0200223.BI fallocate \fR=\fPstr
224Whether pre-allocation is performed when laying down files. Accepted values
225are:
226.RS
227.RS
228.TP
229.B none
230Do not pre-allocate space.
231.TP
232.B posix
233Pre-allocate via posix_fallocate().
234.TP
235.B keep
236Pre-allocate via fallocate() with FALLOC_FL_KEEP_SIZE set.
237.TP
238.B 0
239Backward-compatible alias for 'none'.
240.TP
241.B 1
242Backward-compatible alias for 'posix'.
243.RE
244.P
245May not be available on all supported platforms. 'keep' is only
246available on Linux. If using ZFS on Solaris this must be set to 'none'
247because ZFS doesn't support it. Default: 'posix'.
248.RE
Jens Axboe7bc8c2c2010-01-28 11:31:31 +0100249.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200250.BI fadvise_hint \fR=\fPbool
Aaron Carrolld1429b52007-09-18 08:10:57 +0200251Disable use of \fIposix_fadvise\fR\|(2) to advise the kernel what I/O patterns
252are likely to be issued. Default: true.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200253.TP
Jens Axboef7fa2652009-03-09 14:20:20 +0100254.BI size \fR=\fPint
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200255Total size of I/O for this job. \fBfio\fR will run until this many bytes have
256been transfered, unless limited by other options (\fBruntime\fR, for instance).
Jens Axboed7c8be02010-11-25 08:21:39 +0100257Unless \fBnrfiles\fR and \fBfilesize\fR options are given, this amount will be
Jens Axboed6667262010-06-25 11:32:48 +0200258divided between the available files for the job. If not set, fio will use the
259full size of the given files or devices. If the the files do not exist, size
Jens Axboe7bb59102011-07-12 19:47:03 +0200260must be given. It is also possible to give size as a percentage between 1 and
261100. If size=20% is given, fio will use 20% of the full size of the given files
262or devices.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200263.TP
Jens Axboe74586c12011-01-20 10:16:03 -0700264.BI fill_device \fR=\fPbool "\fR,\fB fill_fs" \fR=\fPbool
Jens Axboe3ce9dca2009-06-10 08:55:21 +0200265Sets size to something really large and waits for ENOSPC (no space left on
266device) as the terminating condition. Only makes sense with sequential write.
267For a read workload, the mount point will be filled first then IO started on
Jens Axboe4f124322011-01-19 15:35:26 -0700268the result. This option doesn't make sense if operating on a raw device node,
269since the size of that is already known by the file system. Additionally,
270writing beyond end-of-device will not return ENOSPC there.
Jens Axboe3ce9dca2009-06-10 08:55:21 +0200271.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200272.BI filesize \fR=\fPirange
273Individual file sizes. May be a range, in which case \fBfio\fR will select sizes
Aaron Carrolld1429b52007-09-18 08:10:57 +0200274for files at random within the given range, limited to \fBsize\fR in total (if
275that is given). If \fBfilesize\fR is not specified, each created file is the
276same size.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200277.TP
Jens Axboef7fa2652009-03-09 14:20:20 +0100278.BI blocksize \fR=\fPint[,int] "\fR,\fB bs" \fR=\fPint[,int]
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200279Block size for I/O units. Default: 4k. Values for reads and writes can be
Jens Axboe656ebab2010-04-13 10:39:14 +0200280specified separately in the format \fIread\fR,\fIwrite\fR, either of
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200281which may be empty to leave that value at its default.
282.TP
Jens Axboe91837882008-02-05 12:02:07 +0100283.BI blocksize_range \fR=\fPirange[,irange] "\fR,\fB bsrange" \fR=\fPirange[,irange]
Aaron Carrolld1429b52007-09-18 08:10:57 +0200284Specify a range of I/O block sizes. The issued I/O unit will always be a
285multiple of the minimum size, unless \fBblocksize_unaligned\fR is set. Applies
Jens Axboe91837882008-02-05 12:02:07 +0100286to both reads and writes if only one range is given, but can be specified
Jens Axboe656ebab2010-04-13 10:39:14 +0200287separately with a comma seperating the values. Example: bsrange=1k-4k,2k-8k.
Jens Axboe91837882008-02-05 12:02:07 +0100288Also (see \fBblocksize\fR).
289.TP
290.BI bssplit \fR=\fPstr
291This option allows even finer grained control of the block sizes issued,
292not just even splits between them. With this option, you can weight various
293block sizes for exact control of the issued IO for a job that has mixed
294block sizes. The format of the option is bssplit=blocksize/percentage,
Martin Steigerwald5982a922011-06-27 16:07:24 +0200295optionally adding as many definitions as needed separated by a colon.
Jens Axboe91837882008-02-05 12:02:07 +0100296Example: bssplit=4k/10:64k/50:32k/40 would issue 50% 64k blocks, 10% 4k
Jens Axboec83cdd32009-04-24 14:23:59 +0200297blocks and 40% 32k blocks. \fBbssplit\fR also supports giving separate
298splits to reads and writes. The format is identical to what the
299\fBbs\fR option accepts, the read and write parts are separated with a
300comma.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200301.TP
302.B blocksize_unaligned\fR,\fP bs_unaligned
Aaron Carrolld1429b52007-09-18 08:10:57 +0200303If set, any size in \fBblocksize_range\fR may be used. This typically won't
304work with direct I/O, as that normally requires sector alignment.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200305.TP
Jens Axboe2b7a01d2009-03-11 11:00:13 +0100306.BI blockalign \fR=\fPint[,int] "\fR,\fB ba" \fR=\fPint[,int]
Martin Steigerwald639ce0f2009-05-20 11:33:49 +0200307At what boundary to align random IO offsets. Defaults to the same as 'blocksize'
308the minimum blocksize given. Minimum alignment is typically 512b
Jens Axboe2b7a01d2009-03-11 11:00:13 +0100309for using direct IO, though it usually depends on the hardware block size.
310This option is mutually exclusive with using a random map for files, so it
311will turn off that option.
Jens Axboe43602662009-03-14 20:08:47 +0100312.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200313.B zero_buffers
314Initialise buffers with all zeros. Default: fill buffers with random data.
315.TP
Jens Axboe901bb992009-03-14 20:17:36 +0100316.B refill_buffers
317If this option is given, fio will refill the IO buffers on every submit. The
318default is to only fill it at init time and reuse that data. Only makes sense
319if zero_buffers isn't specified, naturally. If data verification is enabled,
320refill_buffers is also automatically enabled.
321.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200322.BI nrfiles \fR=\fPint
323Number of files to use for this job. Default: 1.
324.TP
325.BI openfiles \fR=\fPint
326Number of files to keep open at the same time. Default: \fBnrfiles\fR.
327.TP
328.BI file_service_type \fR=\fPstr
329Defines how files to service are selected. The following types are defined:
330.RS
331.RS
332.TP
333.B random
334Choose a file at random
335.TP
336.B roundrobin
337Round robin over open files (default).
Jens Axboe6b7f6852009-03-09 14:22:56 +0100338.B sequential
339Do each file in the set sequentially.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200340.RE
341.P
342The number of I/Os to issue before switching a new file can be specified by
343appending `:\fIint\fR' to the service type.
344.RE
345.TP
346.BI ioengine \fR=\fPstr
347Defines how the job issues I/O. The following types are defined:
348.RS
349.RS
350.TP
351.B sync
352Basic \fIread\fR\|(2) or \fIwrite\fR\|(2) I/O. \fIfseek\fR\|(2) is used to
353position the I/O location.
354.TP
gurudas paia31041e2007-10-23 15:12:30 +0200355.B psync
356Basic \fIpread\fR\|(2) or \fIpwrite\fR\|(2) I/O.
357.TP
Jens Axboe91837882008-02-05 12:02:07 +0100358.B vsync
359Basic \fIreadv\fR\|(2) or \fIwritev\fR\|(2) I/O. Will emulate queuing by
360coalescing adjacents IOs into a single submission.
361.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200362.B libaio
363Linux native asynchronous I/O.
364.TP
365.B posixaio
Bruce Cran03e20d62011-01-02 20:14:54 +0100366POSIX asynchronous I/O using \fIaio_read\fR\|(3) and \fIaio_write\fR\|(3).
367.TP
368.B solarisaio
369Solaris native asynchronous I/O.
370.TP
371.B windowsaio
372Windows native asynchronous I/O.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200373.TP
374.B mmap
Aaron Carrolld1429b52007-09-18 08:10:57 +0200375File is memory mapped with \fImmap\fR\|(2) and data copied using
376\fImemcpy\fR\|(3).
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200377.TP
378.B splice
Aaron Carrolld1429b52007-09-18 08:10:57 +0200379\fIsplice\fR\|(2) is used to transfer the data and \fIvmsplice\fR\|(2) to
380transfer data from user-space to the kernel.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200381.TP
382.B syslet-rw
383Use the syslet system calls to make regular read/write asynchronous.
384.TP
385.B sg
386SCSI generic sg v3 I/O. May be either synchronous using the SG_IO ioctl, or if
Aaron Carrolld1429b52007-09-18 08:10:57 +0200387the target is an sg character device, we use \fIread\fR\|(2) and
388\fIwrite\fR\|(2) for asynchronous I/O.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200389.TP
390.B null
391Doesn't transfer any data, just pretends to. Mainly used to exercise \fBfio\fR
392itself and for debugging and testing purposes.
393.TP
394.B net
395Transfer over the network. \fBfilename\fR must be set appropriately to
396`\fIhost\fR/\fIport\fR' regardless of data direction. If receiving, only the
397\fIport\fR argument is used.
398.TP
399.B netsplice
400Like \fBnet\fR, but uses \fIsplice\fR\|(2) and \fIvmsplice\fR\|(2) to map data
401and send/receive.
402.TP
gurudas pai53aec0a2007-10-05 13:20:18 +0200403.B cpuio
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200404Doesn't transfer any data, but burns CPU cycles according to \fBcpuload\fR and
405\fBcpucycles\fR parameters.
406.TP
407.B guasi
408The GUASI I/O engine is the Generic Userspace Asynchronous Syscall Interface
409approach to asycnronous I/O.
Aaron Carrolld1429b52007-09-18 08:10:57 +0200410.br
411See <http://www.xmailserver.org/guasi\-lib.html>.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200412.TP
ren yufei21b8aee2011-08-01 10:01:57 +0200413.B rdma
414The RDMA I/O engine supports both RDMA memory semantic(RDMA_WRITE/RDMA_READ)
415and channel semantic(Send/Recv) in InfiniBand, RoCE and iWarp environment.
416.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200417.B external
418Loads an external I/O engine object file. Append the engine filename as
419`:\fIenginepath\fR'.
420.RE
421.RE
422.TP
423.BI iodepth \fR=\fPint
Sebastian Kayser8489dae2010-12-01 22:28:47 +0100424Number of I/O units to keep in flight against the file. Note that increasing
425iodepth beyond 1 will not affect synchronous ioengines (except for small
Jens Axboeee72ca02010-12-02 20:05:37 +0100426degress when verify_async is in use). Even async engines my impose OS
427restrictions causing the desired depth not to be achieved. This may happen on
428Linux when using libaio and not setting \fBdirect\fR=1, since buffered IO is
429not async on that OS. Keep an eye on the IO depth distribution in the
430fio output to verify that the achieved depth is as expected. Default: 1.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200431.TP
432.BI iodepth_batch \fR=\fPint
433Number of I/Os to submit at once. Default: \fBiodepth\fR.
434.TP
Jens Axboe3ce9dca2009-06-10 08:55:21 +0200435.BI iodepth_batch_complete \fR=\fPint
436This defines how many pieces of IO to retrieve at once. It defaults to 1 which
437 means that we'll ask for a minimum of 1 IO in the retrieval process from the
438kernel. The IO retrieval will go on until we hit the limit set by
439\fBiodepth_low\fR. If this variable is set to 0, then fio will always check for
440completed events before queuing more IO. This helps reduce IO latency, at the
441cost of more retrieval system calls.
442.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200443.BI iodepth_low \fR=\fPint
444Low watermark indicating when to start filling the queue again. Default:
445\fBiodepth\fR.
446.TP
447.BI direct \fR=\fPbool
448If true, use non-buffered I/O (usually O_DIRECT). Default: false.
449.TP
450.BI buffered \fR=\fPbool
451If true, use buffered I/O. This is the opposite of the \fBdirect\fR parameter.
452Default: true.
453.TP
Jens Axboef7fa2652009-03-09 14:20:20 +0100454.BI offset \fR=\fPint
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200455Offset in the file to start I/O. Data before the offset will not be touched.
456.TP
457.BI fsync \fR=\fPint
Aaron Carrolld1429b52007-09-18 08:10:57 +0200458How many I/Os to perform before issuing an \fBfsync\fR\|(2) of dirty data. If
4590, don't sync. Default: 0.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200460.TP
Jens Axboe5f9099e2009-06-16 22:40:26 +0200461.BI fdatasync \fR=\fPint
462Like \fBfsync\fR, but uses \fBfdatasync\fR\|(2) instead to only sync the
463data parts of the file. Default: 0.
464.TP
Jens Axboee76b1da2010-03-09 20:49:54 +0100465.BI sync_file_range \fR=\fPstr:int
466Use sync_file_range() for every \fRval\fP number of write operations. Fio will
467track range of writes that have happened since the last sync_file_range() call.
468\fRstr\fP can currently be one or more of:
469.RS
470.TP
471.B wait_before
472SYNC_FILE_RANGE_WAIT_BEFORE
473.TP
474.B write
475SYNC_FILE_RANGE_WRITE
476.TP
477.B wait_after
478SYNC_FILE_RANGE_WRITE
479.TP
480.RE
481.P
482So if you do sync_file_range=wait_before,write:8, fio would use
483\fBSYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE\fP for every 8 writes.
484Also see the sync_file_range(2) man page. This option is Linux specific.
485.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200486.BI overwrite \fR=\fPbool
Aaron Carrolld1429b52007-09-18 08:10:57 +0200487If writing, setup the file first and do overwrites. Default: false.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200488.TP
489.BI end_fsync \fR=\fPbool
Aaron Carrolld1429b52007-09-18 08:10:57 +0200490Sync file contents when job exits. Default: false.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200491.TP
492.BI fsync_on_close \fR=\fPbool
493If true, sync file contents on close. This differs from \fBend_fsync\fR in that
Aaron Carrolld1429b52007-09-18 08:10:57 +0200494it will happen on every close, not just at the end of the job. Default: false.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200495.TP
496.BI rwmixcycle \fR=\fPint
497How many milliseconds before switching between reads and writes for a mixed
498workload. Default: 500ms.
499.TP
500.BI rwmixread \fR=\fPint
501Percentage of a mixed workload that should be reads. Default: 50.
502.TP
503.BI rwmixwrite \fR=\fPint
Aaron Carrolld1429b52007-09-18 08:10:57 +0200504Percentage of a mixed workload that should be writes. If \fBrwmixread\fR and
Jens Axboec35dd7a2009-06-10 08:39:16 +0200505\fBrwmixwrite\fR are given and do not sum to 100%, the latter of the two
506overrides the first. This may interfere with a given rate setting, if fio is
507asked to limit reads or writes to a certain rate. If that is the case, then
508the distribution may be skewed. Default: 50.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200509.TP
510.B norandommap
511Normally \fBfio\fR will cover every block of the file when doing random I/O. If
512this parameter is given, a new offset will be chosen without looking at past
513I/O history. This parameter is mutually exclusive with \fBverify\fR.
514.TP
Jens Axboe3ce9dca2009-06-10 08:55:21 +0200515.B softrandommap
516See \fBnorandommap\fR. If fio runs with the random block map enabled and it
517fails to allocate the map, if this option is set it will continue without a
518random block map. As coverage will not be as complete as with random maps, this
519option is disabled by default.
520.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200521.BI nice \fR=\fPint
522Run job with given nice value. See \fInice\fR\|(2).
523.TP
524.BI prio \fR=\fPint
525Set I/O priority value of this job between 0 (highest) and 7 (lowest). See
526\fIionice\fR\|(1).
527.TP
528.BI prioclass \fR=\fPint
529Set I/O priority class. See \fIionice\fR\|(1).
530.TP
531.BI thinktime \fR=\fPint
532Stall job for given number of microseconds between issuing I/Os.
533.TP
534.BI thinktime_spin \fR=\fPint
535Pretend to spend CPU time for given number of microseconds, sleeping the rest
536of the time specified by \fBthinktime\fR. Only valid if \fBthinktime\fR is set.
537.TP
538.BI thinktime_blocks \fR=\fPint
539Number of blocks to issue before waiting \fBthinktime\fR microseconds.
540Default: 1.
541.TP
542.BI rate \fR=\fPint
Jens Axboec35dd7a2009-06-10 08:39:16 +0200543Cap bandwidth used by this job. The number is in bytes/sec, the normal postfix
544rules apply. You can use \fBrate\fR=500k to limit reads and writes to 500k each,
545or you can specify read and writes separately. Using \fBrate\fR=1m,500k would
546limit reads to 1MB/sec and writes to 500KB/sec. Capping only reads or writes
547can be done with \fBrate\fR=,500k or \fBrate\fR=500k,. The former will only
548limit writes (to 500KB/sec), the latter will only limit reads.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200549.TP
550.BI ratemin \fR=\fPint
551Tell \fBfio\fR to do whatever it can to maintain at least the given bandwidth.
Jens Axboec35dd7a2009-06-10 08:39:16 +0200552Failing to meet this requirement will cause the job to exit. The same format
553as \fBrate\fR is used for read vs write separation.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200554.TP
555.BI rate_iops \fR=\fPint
Jens Axboec35dd7a2009-06-10 08:39:16 +0200556Cap the bandwidth to this number of IOPS. Basically the same as rate, just
557specified independently of bandwidth. The same format as \fBrate\fR is used for
558read vs write seperation. If \fBblocksize\fR is a range, the smallest block
559size is used as the metric.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200560.TP
561.BI rate_iops_min \fR=\fPint
Jens Axboec35dd7a2009-06-10 08:39:16 +0200562If this rate of I/O is not met, the job will exit. The same format as \fBrate\fR
563is used for read vs write seperation.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200564.TP
565.BI ratecycle \fR=\fPint
566Average bandwidth for \fBrate\fR and \fBratemin\fR over this number of
567milliseconds. Default: 1000ms.
568.TP
569.BI cpumask \fR=\fPint
570Set CPU affinity for this job. \fIint\fR is a bitmask of allowed CPUs the job
571may run on. See \fBsched_setaffinity\fR\|(2).
572.TP
573.BI cpus_allowed \fR=\fPstr
574Same as \fBcpumask\fR, but allows a comma-delimited list of CPU numbers.
575.TP
576.BI startdelay \fR=\fPint
577Delay start of job for the specified number of seconds.
578.TP
579.BI runtime \fR=\fPint
580Terminate processing after the specified number of seconds.
581.TP
582.B time_based
583If given, run for the specified \fBruntime\fR duration even if the files are
584completely read or written. The same workload will be repeated as many times
585as \fBruntime\fR allows.
586.TP
Jens Axboe901bb992009-03-14 20:17:36 +0100587.BI ramp_time \fR=\fPint
588If set, fio will run the specified workload for this amount of time before
589logging any performance numbers. Useful for letting performance settle before
590logging results, thus minimizing the runtime required for stable results. Note
Jens Axboec35dd7a2009-06-10 08:39:16 +0200591that the \fBramp_time\fR is considered lead in time for a job, thus it will
592increase the total runtime if a special timeout or runtime is specified.
Jens Axboe901bb992009-03-14 20:17:36 +0100593.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200594.BI invalidate \fR=\fPbool
595Invalidate buffer-cache for the file prior to starting I/O. Default: true.
596.TP
597.BI sync \fR=\fPbool
598Use synchronous I/O for buffered writes. For the majority of I/O engines,
Aaron Carrolld1429b52007-09-18 08:10:57 +0200599this means using O_SYNC. Default: false.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200600.TP
601.BI iomem \fR=\fPstr "\fR,\fP mem" \fR=\fPstr
602Allocation method for I/O unit buffer. Allowed values are:
603.RS
604.RS
605.TP
606.B malloc
607Allocate memory with \fImalloc\fR\|(3).
608.TP
609.B shm
610Use shared memory buffers allocated through \fIshmget\fR\|(2).
611.TP
612.B shmhuge
613Same as \fBshm\fR, but use huge pages as backing.
614.TP
615.B mmap
616Use \fImmap\fR\|(2) for allocation. Uses anonymous memory unless a filename
617is given after the option in the format `:\fIfile\fR'.
618.TP
619.B mmaphuge
620Same as \fBmmap\fR, but use huge files as backing.
621.RE
622.P
623The amount of memory allocated is the maximum allowed \fBblocksize\fR for the
624job multiplied by \fBiodepth\fR. For \fBshmhuge\fR or \fBmmaphuge\fR to work,
625the system must have free huge pages allocated. \fBmmaphuge\fR also needs to
Jens Axboe2e266ba2009-09-14 08:56:53 +0200626have hugetlbfs mounted, and \fIfile\fR must point there. At least on Linux,
627huge pages must be manually allocated. See \fB/proc/sys/vm/nr_hugehages\fR
628and the documentation for that. Normally you just need to echo an appropriate
629number, eg echoing 8 will ensure that the OS has 8 huge pages ready for
630use.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200631.RE
632.TP
Jens Axboed529ee12009-07-01 10:33:03 +0200633.BI iomem_align \fR=\fPint
634This indiciates the memory alignment of the IO memory buffers. Note that the
635given alignment is applied to the first IO unit buffer, if using \fBiodepth\fR
636the alignment of the following buffers are given by the \fBbs\fR used. In
637other words, if using a \fBbs\fR that is a multiple of the page sized in the
638system, all buffers will be aligned to this value. If using a \fBbs\fR that
639is not page aligned, the alignment of subsequent IO memory buffers is the
640sum of the \fBiomem_align\fR and \fBbs\fR used.
641.TP
Jens Axboef7fa2652009-03-09 14:20:20 +0100642.BI hugepage\-size \fR=\fPint
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200643Defines the size of a huge page. Must be at least equal to the system setting.
Jens Axboeb22989b2009-07-17 22:29:23 +0200644Should be a multiple of 1MB. Default: 4MB.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200645.TP
646.B exitall
647Terminate all jobs when one finishes. Default: wait for each job to finish.
648.TP
649.BI bwavgtime \fR=\fPint
650Average bandwidth calculations over the given time in milliseconds. Default:
651500ms.
652.TP
653.BI create_serialize \fR=\fPbool
Aaron Carrolld1429b52007-09-18 08:10:57 +0200654If true, serialize file creation for the jobs. Default: true.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200655.TP
656.BI create_fsync \fR=\fPbool
657\fIfsync\fR\|(2) data file after creation. Default: true.
658.TP
Jens Axboe6b7f6852009-03-09 14:22:56 +0100659.BI create_on_open \fR=\fPbool
660If true, the files are not created until they are opened for IO by the job.
661.TP
Jens Axboee9f48472009-06-03 12:14:08 +0200662.BI pre_read \fR=\fPbool
663If this is given, files will be pre-read into memory before starting the given
664IO operation. This will also clear the \fR \fBinvalidate\fR flag, since it is
Jens Axboe9c0d2242009-07-01 12:26:28 +0200665pointless to pre-read and then drop the cache. This will only work for IO
666engines that are seekable, since they allow you to read the same data
667multiple times. Thus it will not work on eg network or splice IO.
Jens Axboee9f48472009-06-03 12:14:08 +0200668.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200669.BI unlink \fR=\fPbool
670Unlink job files when done. Default: false.
671.TP
672.BI loops \fR=\fPint
673Specifies the number of iterations (runs of the same workload) of this job.
674Default: 1.
675.TP
676.BI do_verify \fR=\fPbool
677Run the verify phase after a write phase. Only valid if \fBverify\fR is set.
678Default: true.
679.TP
680.BI verify \fR=\fPstr
681Method of verifying file contents after each iteration of the job. Allowed
682values are:
683.RS
684.RS
685.TP
Jens Axboeb892dc02009-09-05 20:37:35 +0200686.B md5 crc16 crc32 crc32c crc32c-intel crc64 crc7 sha256 sha512 sha1
Jens Axboe0539d752010-06-21 15:22:56 +0200687Store appropriate checksum in the header of each block. crc32c-intel is
688hardware accelerated SSE4.2 driven, falls back to regular crc32c if
689not supported by the system.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200690.TP
691.B meta
692Write extra information about each I/O (timestamp, block number, etc.). The
Jens Axboe996093b2010-06-24 08:37:13 +0200693block number is verified. See \fBverify_pattern\fR as well.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200694.TP
695.B null
696Pretend to verify. Used for testing internals.
697.RE
Jens Axboeb892dc02009-09-05 20:37:35 +0200698
699This option can be used for repeated burn-in tests of a system to make sure
700that the written data is also correctly read back. If the data direction given
701is a read or random read, fio will assume that it should verify a previously
702written file. If the data direction includes any form of write, the verify will
703be of the newly written data.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200704.RE
705.TP
706.BI verify_sort \fR=\fPbool
707If true, written verify blocks are sorted if \fBfio\fR deems it to be faster to
708read them back in a sorted manner. Default: true.
709.TP
Jens Axboef7fa2652009-03-09 14:20:20 +0100710.BI verify_offset \fR=\fPint
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200711Swap the verification header with data somewhere else in the block before
Aaron Carrolld1429b52007-09-18 08:10:57 +0200712writing. It is swapped back before verifying.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200713.TP
Jens Axboef7fa2652009-03-09 14:20:20 +0100714.BI verify_interval \fR=\fPint
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200715Write the verification header for this number of bytes, which should divide
716\fBblocksize\fR. Default: \fBblocksize\fR.
717.TP
Jens Axboe996093b2010-06-24 08:37:13 +0200718.BI verify_pattern \fR=\fPstr
719If set, fio will fill the io buffers with this pattern. Fio defaults to filling
720with totally random bytes, but sometimes it's interesting to fill with a known
721pattern for io verification purposes. Depending on the width of the pattern,
722fio will fill 1/2/3/4 bytes of the buffer at the time(it can be either a
723decimal or a hex number). The verify_pattern if larger than a 32-bit quantity
724has to be a hex number that starts with either "0x" or "0X". Use with
725\fBverify\fP=meta.
726.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200727.BI verify_fatal \fR=\fPbool
728If true, exit the job on the first observed verification failure. Default:
729false.
730.TP
Jens Axboeb463e932011-01-12 09:03:23 +0100731.BI verify_dump \fR=\fPbool
732If set, dump the contents of both the original data block and the data block we
733read off disk to files. This allows later analysis to inspect just what kind of
734data corruption occurred. On by default.
735.TP
Jens Axboee8462bd2009-07-06 12:59:04 +0200736.BI verify_async \fR=\fPint
737Fio will normally verify IO inline from the submitting thread. This option
738takes an integer describing how many async offload threads to create for IO
739verification instead, causing fio to offload the duty of verifying IO contents
Jens Axboec85c3242009-07-06 14:12:57 +0200740to one or more separate threads. If using this offload option, even sync IO
741engines can benefit from using an \fBiodepth\fR setting higher than 1, as it
742allows them to have IO in flight while verifies are running.
Jens Axboee8462bd2009-07-06 12:59:04 +0200743.TP
744.BI verify_async_cpus \fR=\fPstr
745Tell fio to set the given CPU affinity on the async IO verification threads.
746See \fBcpus_allowed\fP for the format used.
747.TP
Jens Axboe6f874182010-06-21 12:53:26 +0200748.BI verify_backlog \fR=\fPint
749Fio will normally verify the written contents of a job that utilizes verify
750once that job has completed. In other words, everything is written then
751everything is read back and verified. You may want to verify continually
752instead for a variety of reasons. Fio stores the meta data associated with an
753IO block in memory, so for large verify workloads, quite a bit of memory would
David Nellans092f7072010-10-26 08:08:42 -0600754be used up holding this meta data. If this option is enabled, fio will write
755only N blocks before verifying these blocks.
Jens Axboe6f874182010-06-21 12:53:26 +0200756.TP
757.BI verify_backlog_batch \fR=\fPint
758Control how many blocks fio will verify if verify_backlog is set. If not set,
759will default to the value of \fBverify_backlog\fR (meaning the entire queue is
David Nellans092f7072010-10-26 08:08:42 -0600760read back and verified). If \fBverify_backlog_batch\fR is less than
761\fBverify_backlog\fR then not all blocks will be verified, if
762\fBverify_backlog_batch\fR is larger than \fBverify_backlog\fR, some blocks
763will be verified more than once.
Jens Axboe6f874182010-06-21 12:53:26 +0200764.TP
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200765.B stonewall
Martin Steigerwald5982a922011-06-27 16:07:24 +0200766Wait for preceding jobs in the job file to exit before starting this one.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200767\fBstonewall\fR implies \fBnew_group\fR.
768.TP
769.B new_group
770Start a new reporting group. If not given, all jobs in a file will be part
771of the same reporting group, unless separated by a stonewall.
772.TP
773.BI numjobs \fR=\fPint
774Number of clones (processes/threads performing the same workload) of this job.
775Default: 1.
776.TP
777.B group_reporting
778If set, display per-group reports instead of per-job when \fBnumjobs\fR is
779specified.
780.TP
781.B thread
782Use threads created with \fBpthread_create\fR\|(3) instead of processes created
783with \fBfork\fR\|(2).
784.TP
Jens Axboef7fa2652009-03-09 14:20:20 +0100785.BI zonesize \fR=\fPint
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200786Divide file into zones of the specified size in bytes. See \fBzoneskip\fR.
787.TP
Jens Axboef7fa2652009-03-09 14:20:20 +0100788.BI zoneskip \fR=\fPint
Aaron Carrolld1429b52007-09-18 08:10:57 +0200789Skip the specified number of bytes when \fBzonesize\fR bytes of data have been
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200790read.
791.TP
792.BI write_iolog \fR=\fPstr
Stefan Hajnoczi5b42a482011-01-08 20:28:41 +0100793Write the issued I/O patterns to the specified file. Specify a separate file
794for each job, otherwise the iologs will be interspersed and the file may be
795corrupt.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200796.TP
797.BI read_iolog \fR=\fPstr
798Replay the I/O patterns contained in the specified file generated by
799\fBwrite_iolog\fR, or may be a \fBblktrace\fR binary file.
800.TP
David Nellans64bbb862010-08-24 22:13:30 +0200801.BI replay_no_stall \fR=\fPint
802While replaying I/O patterns using \fBread_iolog\fR the default behavior
803attempts to respect timing information between I/Os. Enabling
804\fBreplay_no_stall\fR causes I/Os to be replayed as fast as possible while
805still respecting ordering.
806.TP
David Nellansd1c46c02010-08-31 21:20:47 +0200807.BI replay_redirect \fR=\fPstr
808While replaying I/O patterns using \fBread_iolog\fR the default behavior
809is to replay the IOPS onto the major/minor device that each IOP was recorded
810from. Setting \fBreplay_redirect\fR causes all IOPS to be replayed onto the
811single specified device regardless of the device it was recorded from.
812.TP
Jens Axboe901bb992009-03-14 20:17:36 +0100813.B write_bw_log \fR=\fPstr
814If given, write a bandwidth log of the jobs in this job file. Can be used to
815store data of the bandwidth of the jobs in their lifetime. The included
816fio_generate_plots script uses gnuplot to turn these text files into nice
817graphs. See \fBwrite_log_log\fR for behaviour of given filename. For this
818option, the postfix is _bw.log.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200819.TP
David Nellans64bbb862010-08-24 22:13:30 +0200820.B write_lat_log \fR=\fPstr
Jens Axboe901bb992009-03-14 20:17:36 +0100821Same as \fBwrite_bw_log\fR, but writes I/O completion latencies. If no
822filename is given with this option, the default filename of "jobname_type.log"
823is used. Even if the filename is given, fio will still append the type of log.
824.TP
Jens Axboe02af0982010-06-24 09:59:34 +0200825.B disable_lat \fR=\fPbool
826Disable measurements of total latency numbers. Useful only for cutting
Jens Axboe901bb992009-03-14 20:17:36 +0100827back the number of calls to gettimeofday, as that does impact performance at
828really high IOPS rates. Note that to really get rid of a large amount of these
829calls, this option must be used with disable_slat and disable_bw as well.
830.TP
Jens Axboe02af0982010-06-24 09:59:34 +0200831.B disable_clat \fR=\fPbool
Steven Noonanc95f9da2011-06-22 09:47:09 +0200832Disable measurements of completion latency numbers. See \fBdisable_lat\fR.
Jens Axboe02af0982010-06-24 09:59:34 +0200833.TP
Jens Axboe901bb992009-03-14 20:17:36 +0100834.B disable_slat \fR=\fPbool
Jens Axboe02af0982010-06-24 09:59:34 +0200835Disable measurements of submission latency numbers. See \fBdisable_lat\fR.
Jens Axboe901bb992009-03-14 20:17:36 +0100836.TP
837.B disable_bw_measurement \fR=\fPbool
Jens Axboe02af0982010-06-24 09:59:34 +0200838Disable measurements of throughput/bandwidth numbers. See \fBdisable_lat\fR.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200839.TP
Jens Axboef7fa2652009-03-09 14:20:20 +0100840.BI lockmem \fR=\fPint
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200841Pin the specified amount of memory with \fBmlock\fR\|(2). Can be used to
842simulate a smaller amount of memory.
843.TP
844.BI exec_prerun \fR=\fPstr
845Before running the job, execute the specified command with \fBsystem\fR\|(3).
846.TP
847.BI exec_postrun \fR=\fPstr
848Same as \fBexec_prerun\fR, but the command is executed after the job completes.
849.TP
850.BI ioscheduler \fR=\fPstr
851Attempt to switch the device hosting the file to the specified I/O scheduler.
852.TP
853.BI cpuload \fR=\fPint
854If the job is a CPU cycle-eater, attempt to use the specified percentage of
855CPU cycles.
856.TP
857.BI cpuchunks \fR=\fPint
858If the job is a CPU cycle-eater, split the load into cycles of the
859given time in milliseconds.
860.TP
861.BI disk_util \fR=\fPbool
Aaron Carrolld1429b52007-09-18 08:10:57 +0200862Generate disk utilization statistics if the platform supports it. Default: true.
Jens Axboe901bb992009-03-14 20:17:36 +0100863.TP
864.BI gtod_reduce \fR=\fPbool
865Enable all of the gettimeofday() reducing options (disable_clat, disable_slat,
866disable_bw) plus reduce precision of the timeout somewhat to really shrink the
867gettimeofday() call count. With this option enabled, we only do about 0.4% of
868the gtod() calls we would have done if all time keeping was enabled.
869.TP
870.BI gtod_cpu \fR=\fPint
871Sometimes it's cheaper to dedicate a single thread of execution to just getting
872the current time. Fio (and databases, for instance) are very intensive on
873gettimeofday() calls. With this option, you can set one CPU aside for doing
874nothing but logging current time to a shared memory location. Then the other
875threads/processes that run IO workloads need only copy that segment, instead of
876entering the kernel with a gettimeofday() call. The CPU set aside for doing
877these time calls will be excluded from other uses. Fio will manually clear it
878from the CPU mask of other jobs.
Radha Ramachandranf2bba182009-06-15 08:40:16 +0200879.TP
Jens Axboea696fa22009-12-04 10:05:02 +0100880.BI cgroup \fR=\fPstr
881Add job to this control group. If it doesn't exist, it will be created.
Jens Axboe6adb38a2009-12-07 08:01:26 +0100882The system must have a mounted cgroup blkio mount point for this to work. If
883your system doesn't have it mounted, you can do so with:
884
Martin Steigerwald5982a922011-06-27 16:07:24 +0200885# mount \-t cgroup \-o blkio none /cgroup
Jens Axboea696fa22009-12-04 10:05:02 +0100886.TP
887.BI cgroup_weight \fR=\fPint
888Set the weight of the cgroup to this value. See the documentation that comes
889with the kernel, allowed values are in the range of 100..1000.
Jens Axboee0b0d892009-12-08 10:10:14 +0100890.TP
Vivek Goyal7de87092010-03-31 22:55:15 +0200891.BI cgroup_nodelete \fR=\fPbool
892Normally fio will delete the cgroups it has created after the job completion.
893To override this behavior and to leave cgroups around after the job completion,
894set cgroup_nodelete=1. This can be useful if one wants to inspect various
895cgroup files after job completion. Default: false
896.TP
Jens Axboee0b0d892009-12-08 10:10:14 +0100897.BI uid \fR=\fPint
898Instead of running as the invoking user, set the user ID to this value before
899the thread/process does any work.
900.TP
901.BI gid \fR=\fPint
902Set group ID, see \fBuid\fR.
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200903.SH OUTPUT
Aaron Carrolld1429b52007-09-18 08:10:57 +0200904While running, \fBfio\fR will display the status of the created jobs. For
905example:
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200906.RS
Aaron Carrolld1429b52007-09-18 08:10:57 +0200907.P
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200908Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
909.RE
910.P
Aaron Carrolld1429b52007-09-18 08:10:57 +0200911The characters in the first set of brackets denote the current status of each
912threads. The possible values are:
913.P
914.PD 0
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200915.RS
916.TP
917.B P
918Setup but not started.
919.TP
920.B C
921Thread created.
922.TP
923.B I
924Initialized, waiting.
925.TP
926.B R
927Running, doing sequential reads.
928.TP
929.B r
930Running, doing random reads.
931.TP
932.B W
933Running, doing sequential writes.
934.TP
935.B w
936Running, doing random writes.
937.TP
938.B M
939Running, doing mixed sequential reads/writes.
940.TP
941.B m
942Running, doing mixed random reads/writes.
943.TP
944.B F
945Running, currently waiting for \fBfsync\fR\|(2).
946.TP
947.B V
948Running, verifying written data.
949.TP
950.B E
951Exited, not reaped by main thread.
952.TP
953.B \-
954Exited, thread reaped.
955.RE
Aaron Carrolld1429b52007-09-18 08:10:57 +0200956.PD
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200957.P
958The second set of brackets shows the estimated completion percentage of
959the current group. The third set shows the read and write I/O rate,
960respectively. Finally, the estimated run time of the job is displayed.
961.P
962When \fBfio\fR completes (or is interrupted by Ctrl-C), it will show data
963for each thread, each group of threads, and each disk, in that order.
964.P
965Per-thread statistics first show the threads client number, group-id, and
966error code. The remaining figures are as follows:
967.RS
Aaron Carrolld60e92d2007-09-17 10:32:59 +0200968.TP
969.B io
970Number of megabytes of I/O performed.
971.TP
972.B bw
973Average data rate (bandwidth).
974.TP
975.B runt
976Threads run time.
977.TP
978.B slat
979Submission latency minimum, maximum, average and standard deviation. This is
980the time it took to submit the I/O.
981.TP
982.B clat
983Completion latency minimum, maximum, average and standard deviation. This
984is the time between submission and completion.
985.TP
986.B bw
987Bandwidth minimum, maximum, percentage of aggregate bandwidth received, average
988and standard deviation.
989.TP
990.B cpu
991CPU usage statistics. Includes user and system time, number of context switches
992this thread went through and number of major and minor page faults.
993.TP
994.B IO depths
995Distribution of I/O depths. Each depth includes everything less than (or equal)
996to it, but greater than the previous depth.
997.TP
998.B IO issued
999Number of read/write requests issued, and number of short read/write requests.
1000.TP
1001.B IO latencies
1002Distribution of I/O completion latencies. The numbers follow the same pattern
1003as \fBIO depths\fR.
1004.RE
Aaron Carrolld60e92d2007-09-17 10:32:59 +02001005.P
1006The group statistics show:
Aaron Carrolld1429b52007-09-18 08:10:57 +02001007.PD 0
Aaron Carrolld60e92d2007-09-17 10:32:59 +02001008.RS
1009.TP
1010.B io
1011Number of megabytes I/O performed.
1012.TP
1013.B aggrb
1014Aggregate bandwidth of threads in the group.
1015.TP
1016.B minb
1017Minimum average bandwidth a thread saw.
1018.TP
1019.B maxb
1020Maximum average bandwidth a thread saw.
1021.TP
1022.B mint
Aaron Carrolld1429b52007-09-18 08:10:57 +02001023Shortest runtime of threads in the group.
Aaron Carrolld60e92d2007-09-17 10:32:59 +02001024.TP
1025.B maxt
1026Longest runtime of threads in the group.
1027.RE
Aaron Carrolld1429b52007-09-18 08:10:57 +02001028.PD
Aaron Carrolld60e92d2007-09-17 10:32:59 +02001029.P
1030Finally, disk statistics are printed with reads first:
Aaron Carrolld1429b52007-09-18 08:10:57 +02001031.PD 0
Aaron Carrolld60e92d2007-09-17 10:32:59 +02001032.RS
1033.TP
1034.B ios
1035Number of I/Os performed by all groups.
1036.TP
1037.B merge
1038Number of merges in the I/O scheduler.
1039.TP
1040.B ticks
1041Number of ticks we kept the disk busy.
1042.TP
1043.B io_queue
1044Total time spent in the disk queue.
1045.TP
1046.B util
1047Disk utilization.
1048.RE
Aaron Carrolld1429b52007-09-18 08:10:57 +02001049.PD
Aaron Carrolld60e92d2007-09-17 10:32:59 +02001050.SH TERSE OUTPUT
1051If the \fB\-\-minimal\fR option is given, the results will be printed in a
David Nellans562c2d22010-09-23 08:38:17 +02001052semicolon-delimited format suitable for scripted use - a job description
1053(if provided) follows on a new line. Note that the first
Jens Axboe525c2bf2010-06-30 15:22:21 +02001054number in the line is the version number. If the output has to be changed
1055for some reason, this number will be incremented by 1 to signify that
1056change. The fields are:
Aaron Carrolld60e92d2007-09-17 10:32:59 +02001057.P
1058.RS
Jens Axboe525c2bf2010-06-30 15:22:21 +02001059.B version, jobname, groupid, error
Aaron Carrolld60e92d2007-09-17 10:32:59 +02001060.P
1061Read status:
1062.RS
Jens Axboeb22989b2009-07-17 22:29:23 +02001063.B KB I/O, bandwidth \fR(KB/s)\fP, runtime \fR(ms)\fP
Aaron Carrolld60e92d2007-09-17 10:32:59 +02001064.P
1065Submission latency:
1066.RS
1067.B min, max, mean, standard deviation
1068.RE
1069Completion latency:
1070.RS
1071.B min, max, mean, standard deviation
1072.RE
Jens Axboe525c2bf2010-06-30 15:22:21 +02001073Total latency:
1074.RS
1075.B min, max, mean, standard deviation
1076.RE
Aaron Carrolld60e92d2007-09-17 10:32:59 +02001077Bandwidth:
1078.RS
1079.B min, max, aggregate percentage of total, mean, standard deviation
1080.RE
1081.RE
1082.P
1083Write status:
1084.RS
Jens Axboeb22989b2009-07-17 22:29:23 +02001085.B KB I/O, bandwidth \fR(KB/s)\fP, runtime \fR(ms)\fP
Aaron Carrolld60e92d2007-09-17 10:32:59 +02001086.P
1087Submission latency:
1088.RS
1089.B min, max, mean, standard deviation
1090.RE
1091Completion latency:
1092.RS
1093.B min, max, mean, standard deviation
1094.RE
Jens Axboe525c2bf2010-06-30 15:22:21 +02001095Total latency:
1096.RS
1097.B min, max, mean, standard deviation
1098.RE
Aaron Carrolld60e92d2007-09-17 10:32:59 +02001099Bandwidth:
1100.RS
1101.B min, max, aggregate percentage of total, mean, standard deviation
1102.RE
1103.RE
1104.P
Aaron Carrolld1429b52007-09-18 08:10:57 +02001105CPU usage:
Aaron Carrolld60e92d2007-09-17 10:32:59 +02001106.RS
Carl Henrik Lundebd2626f2008-06-12 09:17:46 +02001107.B user, system, context switches, major page faults, minor page faults
Aaron Carrolld60e92d2007-09-17 10:32:59 +02001108.RE
1109.P
1110IO depth distribution:
1111.RS
1112.B <=1, 2, 4, 8, 16, 32, >=64
1113.RE
1114.P
David Nellans562c2d22010-09-23 08:38:17 +02001115IO latency distribution:
Aaron Carrolld60e92d2007-09-17 10:32:59 +02001116.RS
David Nellans562c2d22010-09-23 08:38:17 +02001117Microseconds:
1118.RS
1119.B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
1120.RE
1121Milliseconds:
1122.RS
1123.B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
1124.RE
Aaron Carrolld60e92d2007-09-17 10:32:59 +02001125.RE
1126.P
Martin Steigerwald5982a922011-06-27 16:07:24 +02001127Error Info (dependent on continue_on_error, default off):
David Nellans562c2d22010-09-23 08:38:17 +02001128.RS
1129.B total # errors, first error code
1130.RE
1131.P
1132.B text description (if provided in config - appears on newline)
Aaron Carrolld60e92d2007-09-17 10:32:59 +02001133.RE
1134.SH AUTHORS
1135.B fio
Jens Axboeaa58d252010-06-09 09:49:38 +02001136was written by Jens Axboe <jens.axboe@oracle.com>,
1137now Jens Axboe <jaxboe@fusionio.com>.
Aaron Carrolld1429b52007-09-18 08:10:57 +02001138.br
1139This man page was written by Aaron Carroll <aaronc@cse.unsw.edu.au> based
Aaron Carrolld60e92d2007-09-17 10:32:59 +02001140on documentation by Jens Axboe.
1141.SH "REPORTING BUGS"
Jens Axboe482900c2009-06-02 12:15:51 +02001142Report bugs to the \fBfio\fR mailing list <fio@vger.kernel.org>.
Aaron Carrolld1429b52007-09-18 08:10:57 +02001143See \fBREADME\fR.
Aaron Carrolld60e92d2007-09-17 10:32:59 +02001144.SH "SEE ALSO"
Aaron Carrolld1429b52007-09-18 08:10:57 +02001145For further documentation see \fBHOWTO\fR and \fBREADME\fR.
1146.br
1147Sample jobfiles are available in the \fBexamples\fR directory.
Aaron Carrolld60e92d2007-09-17 10:32:59 +02001148