4 fio is a tool that will spawn a number of threads or processes doing a
5 particular type of io action as specified by the user. fio takes a
6 number of global parameters, each inherited by the thread unless
7 otherwise parameters given to them overriding that setting is given.
8 The typical use of fio is to write a job file matching the io load
15 fio resides in a git repo, the canonical place is:
17 git://git.kernel.dk/fio.git
19 The http protocol also works, path is the same.
21 Snapshots are frequently generated and they include the git meta data as
22 well. You can download them here:
24 http://brick.kernel.dk/snaps/
31 Starting with Debian "Squeeze", fio packages are part of the official
32 Debian repository. http://packages.debian.org/search?keywords=fio
35 Starting with Ubuntu 10.04 LTS (aka "Lucid Lynx"), fio packages are part
36 of the Ubuntu "universe" repository.
37 http://packages.ubuntu.com/search?keywords=fio
40 Pascal Bleser <guru@unixtech.be> has fio RPMs in his repository for SUSE
41 variants, you can find them here:
42 http://linux01.gwdg.de/~pbleser/rpm-navigation.php?cat=System/fio
45 Dag Wieƫrs has RPMs for Red Hat related distros, find them here:
46 http://dag.wieers.com/rpm/packages/fio/
49 Mandriva has integrated fio into their package repository, so installing
50 on that distro should be as easy as typing 'urpmi fio'.
53 Packages for Solaris are available from OpenCSW. Install their pkgutil
54 tool (http://www.opencsw.org/get-it/pkgutil/) and then install fio via
58 Bruce Cran <bruce@cran.org.uk> has fio packages for Windows at
59 http://www.bluestop.org/fio .
65 There's a mailing list associated with fio. It's meant for general
66 discussion, bug reporting, questions, and development - basically anything
67 that has to do with fio. An automated mail detailing recent commits is
68 automatically sent to the list at most daily. The list address is
69 fio@vger.kernel.org, subscribe by sending an email to
70 majordomo@vger.kernel.org with
74 in the body of the email. Archives can be found here:
76 http://www.spinics.net/lists/fio/
78 and archives for the old list can be found here:
80 http://maillist.kernel.dk/fio-devel/
86 Just type 'make' and 'make install'. If on BSD, for now you have to
87 specify the BSD Makefile with -f and use gmake (not make), eg:
89 $ gmake -f Makefile.FreeBSD && gmake -f Makefile.FreeBSD install
93 $ gmake -f Makefile.aix && gmake -f Makefile.aix install
95 Likewise with OpenSolaris, use the Makefile.solaris to compile there.
96 The OpenSolaris make should work fine. This might change in the
97 future if I opt for an autoconf type setup.
99 If your compile fails with an error like this:
102 In file included from fio.h:23,
104 os/os.h:15:20: error: libaio.h: No such file or directory
105 In file included from gettime.c:8:
106 fio.h:119: error: field 'iocb' has incomplete type
107 make: *** [gettime.o] Error 1
109 Check that you have the libaio development package installed. On RPM
110 based distros, it's typically called libaio-devel.
117 --debug Enable some debugging options (see below)
118 --output Write output to file
119 --timeout Runtime in seconds
120 --latency-log Generate per-job latency logs
121 --bandwidth-log Generate per-job bandwidth logs
122 --minimal Minimal (terse) output
123 --version Print version info and exit
124 --help Print this page
125 --cmdhelp=cmd Print command help, "all" for all of them
126 --showcmd Turn a job file into command line options
127 --readonly Turn on safety read-only checks, preventing writes
128 --eta=when When ETA estimate should be printed
129 May be "always", "never" or "auto"
130 --section=name Only run specified section in job file
131 --alloc-size=kb Set smalloc pool to this size in kb (def 1024)
132 --warnings-fatal Fio parser warnings are fatal
135 Any parameters following the options will be assumed to be job files,
136 unless they match a job file parameter. You can add as many as you want,
137 each job file will be regarded as a separate group and fio will stonewall
140 The --readonly switch is an extra safety guard to prevent accidentally
141 turning on a write setting when that is not desired. Fio will only write
142 if rw=write/randwrite/rw/randrw is given, but this extra safety net can
143 be used as an extra precaution. It will also enable a write check in the
144 io engine core to prevent an accidental write due to a fio bug.
146 The debug switch allows adding options that trigger certain logging
147 options in fio. Currently the options are:
149 process Dump info related to processes
150 file Dump info related to file actions
151 io Dump info related to IO queuing
152 mem Dump info related to memory allocations
153 blktrace Dump info related to blktrace setup
154 verify Dump info related to IO verification
155 all Enable all debug options
156 random Dump info related to random offset generation
157 parse Dump info related to option matching and parsing
158 diskutil Dump info related to disk utilization updates
159 job:x Dump info only related to job number x
160 mutex Dump info only related to mutex up/down ops
161 profile Dump info related to profile extensions
162 time Dump info related to internal time keeping
163 ? or help Show available debug options.
165 You can specify as many as you want, eg --debug=file,mem will enable
166 file and memory debugging.
168 The section switch is meant to make it easier to ship a bigger job file
169 instead of several smaller ones. Say you define a job file with light,
170 moderate, and heavy parts. Then you can ask fio to run the given part
171 only by giving it a --section=heavy command line option. The section
172 option only applies to job sections, the reserved 'global' section is
173 always parsed and taken into account.
175 Fio has an internal allocator for shared memory called smalloc. It
176 allocates shared structures from this pool. The pool defaults to 1024k
177 in size, and can grow to 128 pools. If running large jobs with randommap
178 enabled it can run out of memory, in which case the --alloc-size switch
179 is handy for starting with a larger pool size. The backing store is
180 files in /tmp. Fio cleans up after itself, while it is running you
181 may see .fio_smalloc.* files in /tmp.
187 See the HOWTO file for a more detailed description of parameters and what
188 they mean. This file contains the terse version. You can describe big and
189 complex setups with the command line, but generally it's a lot easier to
190 just write a simple job file to describe the workload. The job file format
191 is in the ini style format, as that is easy to read and write for the user.
193 The job file parameters are:
195 name=x Use 'x' as the identifier for this job.
196 description=x 'x' is a text description of the job.
197 directory=x Use 'x' as the top level directory for storing files
198 filename=x Force the use of 'x' as the filename for all files
199 in this thread. If not given, fio will make up
200 a suitable filename based on the thread and file
202 rw=x 'x' may be: read, randread, write, randwrite,
203 rw (read-write mix), randrw (read-write random mix)
204 rwmixcycle=x Base cycle for switching between read and write
206 rwmixread=x 'x' percentage of rw mix ios will be reads. If
207 rwmixwrite is also given, the last of the two will
208 be used if they don't add up to 100%.
209 rwmixwrite=x 'x' percentage of rw mix ios will be writes. See
211 rand_repeatable=x The sequence of random io blocks can be repeatable
212 across runs, if 'x' is 1.
213 size=x Set file size to x bytes (x string can include k/m/g)
214 ioengine=x 'x' may be: aio/libaio/linuxaio for Linux aio,
215 posixaio for POSIX aio, solarisaio for Solaris
216 native async IO, windowsaio for Windows native async IO,
217 sync for regular read/write io,
218 psync for regular pread/pwrite io, vsync for regular
219 readv/writev (with queuing emulation) mmap for mmap'ed
220 io, syslet-rw for syslet driven read/write, splice for
221 using splice/vmsplice, sg for direct SG_IO io, net
222 for network io, or cpuio for a cycler burner load. sg
223 only works on Linux on SCSI (or SCSI-like devices, such
224 as usb-storage or sata/libata driven) devices. Fio also
225 has a null io engine, which is mainly used for testing
228 iodepth=x For async io, allow 'x' ios in flight
229 overwrite=x If 'x', layout a write file first.
230 nrfiles=x Spread io load over 'x' number of files per job,
232 prio=x Run io at prio X, 0-7 is the kernel allowed range
233 prioclass=x Run io at prio class X
234 bs=x Use 'x' for thread blocksize. May include k/m postfix.
235 bsrange=x-y Mix thread block sizes randomly between x and y. May
236 also include k/m postfix.
237 direct=x 1 for direct IO, 0 for buffered IO
238 thinktime=x "Think" x usec after each io
239 rate=x Throttle rate to x KB/sec
240 ratemin=x Quit if rate of x KB/sec can't be met
241 ratecycle=x ratemin averaged over x msecs
242 cpumask=x Only allow job to run on CPUs defined by mask.
243 cpus_allowed=x Like 'cpumask', but allow text setting of CPU affinity.
244 fsync=x If writing with buffered IO, fsync after every
245 'x' blocks have been written.
246 end_fsync=x If 'x', run fsync() after end-of-job.
247 startdelay=x Start this thread x seconds after startup
248 runtime=x Terminate x seconds after startup. Can include a
249 normal time suffix if not given in seconds, such as
250 'm' for minutes, 'h' for hours, and 'd' for days.
251 offset=x Start io at offset x (x string can include k/m/g)
252 invalidate=x Invalidate page cache for file prior to doing io
253 sync=x Use sync writes if x and writing buffered IO.
254 mem=x If x == malloc, use malloc for buffers. If x == shm,
255 use shared memory for buffers. If x == mmap, use
257 exitall When one thread quits, terminate the others
258 bwavgtime=x Average bandwidth stats over an x msec window.
259 create_serialize=x If 'x', serialize file creation.
260 create_fsync=x If 'x', run fsync() after file creation.
261 unlink If set, unlink files when done.
262 loops=x Run the job 'x' number of times.
263 verify=x If 'x' == md5, use md5 for verifies. If 'x' == crc32,
264 use crc32 for verifies. md5 is 'safer', but crc32 is
265 a lot faster. Only makes sense for writing to a file.
266 For other types of checksumming, see HOWTO.
267 stonewall Wait for preceeding jobs to end before running.
268 numjobs=x Create 'x' similar entries for this job
269 thread Use pthreads instead of forked jobs
271 zoneskip=y Zone options must be paired. If given, the job
272 will skip y bytes for every x read/written. This
273 can be used to gauge hard drive speed over the entire
274 platter, without reading everything. Both x/y can
275 include k/m/g suffix.
276 iolog=x Open and read io pattern from file 'x'. The file must
277 contain one io action per line in the following format:
279 where with rw=0/1 for read/write, and the offset
280 and length entries being in bytes.
281 write_iolog=x Write an iolog to file 'x' in the same format as iolog.
282 The iolog options are exclusive, if both given the
283 read iolog will be performed. Specify a separate file
284 for each job, otherwise the iologs will be interspersed
285 and the file may be corrupt.
286 write_bw_log Write a bandwidth log.
287 write_lat_log Write a latency log.
288 lockmem=x Lock down x amount of memory on the machine, to
289 simulate a machine with less memory available. x can
290 include k/m/g suffix.
291 nice=x Run job at given nice value.
292 exec_prerun=x Run 'x' before job io is begun.
293 exec_postrun=x Run 'x' after job io has finished.
294 ioscheduler=x Use ioscheduler 'x' for this job.
295 cpuload=x For a CPU io thread, percentage of CPU time to attempt
297 cpuchunks=x Split burn cycles into pieces of x usecs.
304 Fio works on (at least) Linux, Solaris, AIX, OSX, NetBSD, Windows and FreeBSD.
305 Some features and/or options may only be available on some of the platforms,
306 typically because those features only apply to that platform (like the
307 solarisaio engine, or the splice engine on Linux).
309 Some features are not available on FreeBSD/Solaris even if they could be
310 implemented, I'd be happy to take patches for that. An example of that is
311 disk utility statistics and (I think) huge page support, support for that
312 does exist in FreeBSD/Solaris.
314 Fio uses pthread mutexes for signalling and locking and FreeBSD does not
315 support process shared pthread mutexes. As a result, only threads are
316 supported on FreeBSD. This could be fixed with sysv ipc locking or
317 other locking alternatives.
319 Other *BSD platforms are untested, but fio should work there almost out
320 of the box. Since I don't do test runs or even compiles on those platforms,
321 your mileage may vary. Sending me patches for other platforms is greatly
322 appreciated. There's a lot of value in having the same test/benchmark tool
323 available on all platforms.
325 Note that POSIX aio is not enabled by default on AIX. If you get messages like:
327 Symbol resolution failed for /usr/lib/libc.a(posix_aio.o) because:
328 Symbol _posix_kaio_rdwr (number 2) is not exported from dependent module /unix.
330 you need to enable POSIX aio. Run the following commands as root:
332 # lsdev -C -l posix_aio0
333 posix_aio0 Defined Posix Asynchronous I/O
334 # cfgmgr -l posix_aio0
335 # lsdev -C -l posix_aio0
336 posix_aio0 Available Posix Asynchronous I/O
338 POSIX aio should work now. To make the change permanent:
340 # chdev -l posix_aio0 -P -a autoconfig='available'
347 Fio was written by Jens Axboe <axboe@kernel.dk> to enable flexible testing
348 of the Linux IO subsystem and schedulers. He got tired of writing
349 specific test applications to simulate a given workload, and found that
350 the existing io benchmark/test tools out there weren't flexible enough
351 to do what he wanted.
353 Jens Axboe <axboe@kernel.dk> 20060905