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 When inside a corporate firewall, git:// URL sometimes does not work.
20 If git:// does not work, use the http protocol instead:
22 http://git.kernel.dk/fio.git
24 Snapshots are frequently generated and include the git meta data as well.
25 Snapshots can download from:
27 http://brick.kernel.dk/snaps/
29 There are also two official mirrors. Both of these are automatically synced
30 with the main repository, when changes are pushed. If the main repo is down
31 for some reason, either one of these is safe to use as a backup:
33 git://git.kernel.org/pub/scm/linux/kernel/git/axboe/fio.git
34 https://git.kernel.org/pub/scm/linux/kernel/git/axboe/fio.git
38 git://github.com/axboe/fio.git
39 https://github.com/axboe/fio.git
46 Starting with Debian "Squeeze", fio packages are part of the official
47 Debian repository. http://packages.debian.org/search?keywords=fio
50 Starting with Ubuntu 10.04 LTS (aka "Lucid Lynx"), fio packages are part
51 of the Ubuntu "universe" repository.
52 http://packages.ubuntu.com/search?keywords=fio
55 Dag Wieƫrs has RPMs for Red Hat related distros, find them here:
56 http://dag.wieers.com/rpm/packages/fio/
59 Mandriva has integrated fio into their package repository, so installing
60 on that distro should be as easy as typing 'urpmi fio'.
63 Packages for Solaris are available from OpenCSW. Install their pkgutil
64 tool (http://www.opencsw.org/get-it/pkgutil/) and then install fio via
68 Rebecca Cran <rebecca+fio@bluestop.org> has fio packages for Windows at
69 http://www.bluestop.org/fio/ .
72 Packages for BSDs may be available from their binary package repositories.
73 Look for a package "fio" using their binary package managers.
79 The fio project mailing list is meant for anything related to fio including
80 general discussion, bug reporting, questions, and development.
82 An automated mail detailing recent commits is automatically sent to the
83 list at most daily. The list address is fio@vger.kernel.org, subscribe
84 by sending an email to majordomo@vger.kernel.org with
88 in the body of the email. Archives can be found here:
90 http://www.spinics.net/lists/fio/
92 and archives for the old list can be found here:
94 http://maillist.kernel.dk/fio-devel/
100 Just type './configure', 'make' and 'make install'.
102 Note that GNU make is required. On BSDs it's available from devel/gmake
103 within ports directory; on Solaris it's in the SUNWgmake package.
104 On platforms where GNU make isn't the default, type 'gmake' instead of 'make'.
106 Configure will print the enabled options. Note that on Linux based
107 platforms, the libaio development packages must be installed to use
108 the libaio engine. Depending on distro, it is usually called
109 libaio-devel or libaio-dev.
111 For gfio, gtk 2.18 (or newer), associated glib threads, and cairo are required
112 to be installed. gfio isn't built automatically and can be enabled
113 with a --enable-gfio option to configure.
115 To build FIO with a cross-compiler:
117 $ make CROSS_COMPILE=/path/to/toolchain/prefix
118 Configure will attempt to determine the target platform automatically.
120 It's possible to build fio for ESX as well, use the --esx switch to
127 On Windows, Cygwin (http://www.cygwin.com/) is required in order to
128 build fio. To create an MSI installer package install WiX 3.8 from
129 http://wixtoolset.org and run dobuild.cmd from the
130 os/windows directory.
132 How to compile fio on 64-bit Windows:
134 1. Install Cygwin (http://www.cygwin.com/). Install 'make' and all
135 packages starting with 'mingw64-i686' and 'mingw64-x86_64'.
136 2. Open the Cygwin Terminal.
137 3. Go to the fio directory (source files).
138 4. Run 'make clean && make -j'.
140 To build fio on 32-bit Windows, run './configure --build-32bit-win' before 'make'.
142 It's recommended that once built or installed, fio be run in a Command Prompt
143 or other 'native' console such as console2, since there are known to be display
144 and signal issues when running it under a Cygwin shell
145 (see http://code.google.com/p/mintty/issues/detail?id=56 for details).
152 --debug Enable some debugging options (see below)
153 --parse-only Parse options only, don't start any IO
154 --output Write output to file
155 --bandwidth-log Generate aggregate bandwidth logs
156 --minimal Minimal (terse) output
157 --output-format=type Output format (terse,json,json+,normal)
158 --terse-version=type Set terse version output format (default 3, or 2 or 4)
159 --version Print version info and exit
160 --help Print this page
161 --cpuclock-test Perform test/validation of CPU clock
162 --crctest=type Test speed of checksum functions
163 --cmdhelp=cmd Print command help, "all" for all of them
164 --enghelp=engine Print ioengine help, or list available ioengines
165 --enghelp=engine,cmd Print help for an ioengine cmd
166 --showcmd Turn a job file into command line options
167 --eta=when When ETA estimate should be printed
168 May be "always", "never" or "auto"
169 --eta-newline=time Force a new line for every 'time' period passed
170 --status-interval=t Force full status dump every 't' period passed
171 --readonly Turn on safety read-only checks, preventing writes
172 --section=name Only run specified section in job file.
173 Multiple sections can be specified.
174 --alloc-size=kb Set smalloc pool to this size in kb (def 16384)
175 --warnings-fatal Fio parser warnings are fatal
176 --max-jobs=nr Maximum number of threads/processes to support
177 --server=args Start a backend fio server. See Client/Server section.
178 --client=hostname Talk to remote backend(s) fio server at hostname
179 --daemonize=pidfile Background fio server, write pid to file
180 --remote-config=file Tell fio server to load this local job file
181 --idle-prof=option Report cpu idleness on a system or percpu basis
182 (option=system,percpu) or run unit work
183 calibration only (option=calibrate)
184 --inflate-log=log Inflate and output compressed log
185 --trigger-file=file Execute trigger cmd when file exists
186 --trigger-timeout=t Execute trigger af this time
187 --trigger=cmd Set this command as local trigger
188 --trigger-remote=cmd Set this command as remote trigger
189 --aux-path=path Use this path for fio state generated files
192 Any parameters following the options will be assumed to be job files,
193 unless they match a job file parameter. Multiple job files can be listed
194 and each job file will be regarded as a separate group. fio will stonewall
195 execution between each group.
197 The --readonly option is an extra safety guard to prevent users from
198 accidentally starting a write workload when that is not desired. Fio
199 will only write if rw=write/randwrite/rw/randrw is given. This extra
200 safety net can be used as an extra precaution as --readonly will also
201 enable a write check in the io engine core to prevent writes due to
202 unknown user space bug(s).
204 The --debug option triggers additional logging by fio.
205 Currently, additional logging is available for:
207 process Dump info related to processes
208 file Dump info related to file actions
209 io Dump info related to IO queuing
210 mem Dump info related to memory allocations
211 blktrace Dump info related to blktrace setup
212 verify Dump info related to IO verification
213 all Enable all debug options
214 random Dump info related to random offset generation
215 parse Dump info related to option matching and parsing
216 diskutil Dump info related to disk utilization updates
217 job:x Dump info only related to job number x
218 mutex Dump info only related to mutex up/down ops
219 profile Dump info related to profile extensions
220 time Dump info related to internal time keeping
221 net Dump info related to networking connections
222 rate Dump info related to IO rate switching
223 compress Dump info related to log compress/decompress
224 steadystate Dump info related to steady state detection
225 helperthread Dump info related to helper thread
226 ? or help Show available debug options.
228 One can specify multiple debug options: e.g. --debug=file,mem will enable
229 file and memory debugging.
231 The --section option allows one to combine related jobs into one file.
232 E.g. one job file could define light, moderate, and heavy sections. Tell fio to
233 run only the "heavy" section by giving --section=heavy command line option.
234 One can also specify the "write" operations in one section and "verify"
235 operation in another section. The --section option only applies to job
236 sections. The reserved 'global' section is always parsed and used.
238 The --alloc-size switch allows one to use a larger pool size for smalloc.
239 If running large jobs with randommap enabled, fio can run out of memory.
240 Smalloc is an internal allocator for shared structures from a fixed size
241 memory pool. The pool size defaults to 16MiB and can grow to 8 pools.
243 NOTE: While running .fio_smalloc.* backing store files are visible in /tmp.
249 See the HOWTO file for a complete description of job file syntax and
250 parameters. The --cmdhelp option also lists all options. If used with
251 an option argument, --cmdhelp will detail the given option. The job file
252 format is in the ini style format, as that is easy for the user to review
255 This README contains the terse version. Job files can describe big and
256 complex setups that are not possible with the command line. Job files
257 are a good practice even for simple jobs since the file provides an
258 easily accessed record of the workload and can include comments.
260 See the examples/ directory for inspiration on how to write job files. Note
261 the copyright and license requirements currently apply to examples/ files.
267 Normally fio is invoked as a stand-alone application on the machine
268 where the IO workload should be generated. However, the frontend and
269 backend of fio can be run separately. Ie the fio server can generate
270 an IO workload on the "Device Under Test" while being controlled from
273 Start the server on the machine which has access to the storage DUT:
277 where args defines what fio listens to. The arguments are of the form
278 'type,hostname or IP,port'. 'type' is either 'ip' (or ip4) for TCP/IP v4,
279 'ip6' for TCP/IP v6, or 'sock' for a local unix domain socket.
280 'hostname' is either a hostname or IP address, and 'port' is the port to
281 listen to (only valid for TCP/IP, not a local socket). Some examples:
285 Start a fio server, listening on all interfaces on the default port (8765).
287 2) fio --server=ip:hostname,4444
289 Start a fio server, listening on IP belonging to hostname and on port 4444.
291 3) fio --server=ip6:::1,4444
293 Start a fio server, listening on IPv6 localhost ::1 and on port 4444.
295 4) fio --server=,4444
297 Start a fio server, listening on all interfaces on port 4444.
299 5) fio --server=1.2.3.4
301 Start a fio server, listening on IP 1.2.3.4 on the default port.
303 6) fio --server=sock:/tmp/fio.sock
305 Start a fio server, listening on the local socket /tmp/fio.sock.
307 Once a server is running, a "client" can connect to the fio server with:
309 fio --local-args --client=<server> --remote-args <job file(s)>
311 where --local-args are arguments for the client where it is
312 running, 'server' is the connect string, and --remote-args and <job file(s)>
313 are sent to the server. The 'server' string follows the same format as it
314 does on the server side, to allow IP/hostname/socket and port strings.
316 Fio can connect to multiple servers this way:
318 fio --client=<server1> <job file(s)> --client=<server2> <job file(s)>
320 If the job file is located on the fio server, then you can tell the server
321 to load a local file as well. This is done by using --remote-config:
323 fio --client=server --remote-config /path/to/file.fio
325 Then fio will open this local (to the server) job file instead
326 of being passed one from the client.
328 If you have many servers (example: 100 VMs/containers),
329 you can input a pathname of a file containing host IPs/names as the parameter
330 value for the --client option. For example, here is an example "host.list"
331 file containing 2 hostnames:
333 host1.your.dns.domain
334 host2.your.dns.domain
336 The fio command would then be:
338 fio --client=host.list <job file(s)>
340 In this mode, you cannot input server-specific parameters or job files -- all
341 servers receive the same job file.
343 In order to let fio --client runs use a shared filesystem
344 from multiple hosts, fio --client now prepends the IP address of the
345 server to the filename. For example, if fio is using directory /mnt/nfs/fio
346 and is writing filename fileio.tmp, with a --client hostfile containing
347 two hostnames h1 and h2 with IP addresses 192.168.10.120 and 192.168.10.121,
348 then fio will create two files:
350 /mnt/nfs/fio/192.168.10.120.fileio.tmp
351 /mnt/nfs/fio/192.168.10.121.fileio.tmp
357 Fio works on (at least) Linux, Solaris, AIX, HP-UX, OSX, NetBSD, OpenBSD,
358 Windows, FreeBSD, and DragonFly. Some features and/or options may only be
359 available on some of the platforms, typically because those features only
360 apply to that platform (like the solarisaio engine, or the splice engine on
363 Some features are not available on FreeBSD/Solaris even if they could be
364 implemented, I'd be happy to take patches for that. An example of that is
365 disk utility statistics and (I think) huge page support, support for that
366 does exist in FreeBSD/Solaris.
368 Fio uses pthread mutexes for signalling and locking and FreeBSD does not
369 support process shared pthread mutexes. As a result, only threads are
370 supported on FreeBSD. This could be fixed with sysv ipc locking or
371 other locking alternatives.
373 Other *BSD platforms are untested, but fio should work there almost out
374 of the box. Since I don't do test runs or even compiles on those platforms,
375 your mileage may vary. Sending me patches for other platforms is greatly
376 appreciated. There's a lot of value in having the same test/benchmark tool
377 available on all platforms.
379 Note that POSIX aio is not enabled by default on AIX. Messages like these:
381 Symbol resolution failed for /usr/lib/libc.a(posix_aio.o) because:
382 Symbol _posix_kaio_rdwr (number 2) is not exported from dependent module /unix.
384 indicate one needs to enable POSIX aio. Run the following commands as root:
386 # lsdev -C -l posix_aio0
387 posix_aio0 Defined Posix Asynchronous I/O
388 # cfgmgr -l posix_aio0
389 # lsdev -C -l posix_aio0
390 posix_aio0 Available Posix Asynchronous I/O
392 POSIX aio should work now. To make the change permanent:
394 # chdev -l posix_aio0 -P -a autoconfig='available'
401 Fio was written by Jens Axboe <axboe@kernel.dk> to enable flexible testing
402 of the Linux IO subsystem and schedulers. He got tired of writing
403 specific test applications to simulate a given workload, and found that
404 the existing io benchmark/test tools out there weren't flexible enough
405 to do what he wanted.
407 Jens Axboe <axboe@kernel.dk> 20060905