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