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 If you are inside a corporate firewall, git:// may not always work for
20 you. In that case you can use the http protocol, path is the same:
22 http://git.kernel.dk/fio.git
24 Snapshots are frequently generated and they include the git meta data as
25 well. You can download them here:
27 http://brick.kernel.dk/snaps/
34 Starting with Debian "Squeeze", fio packages are part of the official
35 Debian repository. http://packages.debian.org/search?keywords=fio
38 Starting with Ubuntu 10.04 LTS (aka "Lucid Lynx"), fio packages are part
39 of the Ubuntu "universe" repository.
40 http://packages.ubuntu.com/search?keywords=fio
43 Dag Wieƫrs has RPMs for Red Hat related distros, find them here:
44 http://dag.wieers.com/rpm/packages/fio/
47 Mandriva has integrated fio into their package repository, so installing
48 on that distro should be as easy as typing 'urpmi fio'.
51 Packages for Solaris are available from OpenCSW. Install their pkgutil
52 tool (http://www.opencsw.org/get-it/pkgutil/) and then install fio via
56 Bruce Cran <bruce@cran.org.uk> has fio packages for Windows at
57 http://www.bluestop.org/fio/ .
63 There's a mailing list associated with fio. It's meant for general
64 discussion, bug reporting, questions, and development - basically anything
65 that has to do with fio. An automated mail detailing recent commits is
66 automatically sent to the list at most daily. The list address is
67 fio@vger.kernel.org, subscribe by sending an email to
68 majordomo@vger.kernel.org with
72 in the body of the email. Archives can be found here:
74 http://www.spinics.net/lists/fio/
76 and archives for the old list can be found here:
78 http://maillist.kernel.dk/fio-devel/
84 Just type 'make' and 'make install'.
86 Note that GNU make is required. On BSD it's available from devel/gmake;
87 on Solaris it's in the SUNWgmake package. On platforms where GNU make
88 isn't the default, type 'gmake' instead of 'make'.
90 If your compile fails with an error like this:
93 In file included from fio.h:23,
95 os/os.h:15:20: error: libaio.h: No such file or directory
96 In file included from gettime.c:8:
97 fio.h:119: error: field 'iocb' has incomplete type
98 make: *** [gettime.o] Error 1
100 Check that you have the libaio development package installed. On RPM
101 based distros, it's typically called libaio-devel.
107 On Windows Cygwin (http://www.cygwin.com/) is required in order to
108 build fio. To create an MSI installer package install WiX 3.7 from
109 http://wixtoolset.org and run dobuild.cmd from the
110 os/windows directory.
112 How to compile FIO on 64-bit Windows:
114 1. Install Cygwin (http://www.cygwin.com/setup.exe). Install 'make' and all
115 packages starting with 'mingw64-i686' and 'mingw64-x86_64'.
116 2. Download ftp://sourceware.org/pub/pthreads-win32/prebuilt-dll-2-9-1-release/dll/x64/pthreadGC2.dll
117 and copy to the fio source directory.
118 3. Open the Cygwin Terminal.
119 4. Go to the fio directory (source files).
123 To build fio on 32-bit Windows, download x86/pthreadGC2.dll instead and do
124 './configure --build-32bit-win=yes' before 'make'.
126 It's recommended that once built or installed, fio be run in a Command Prompt
127 or other 'native' console such as console2, since there are known to be display
128 and signal issues when running it under a Cygwin shell
129 (see http://code.google.com/p/mintty/issues/detail?id=56 for details).
136 --debug Enable some debugging options (see below)
137 --output Write output to file
138 --runtime Runtime in seconds
139 --latency-log Generate per-job latency logs
140 --bandwidth-log Generate per-job bandwidth logs
141 --minimal Minimal (terse) output
142 --output-format=type Output format (terse,json,normal)
143 --terse-version=type Terse version output format (default 3, or 2 or 4).
144 --version Print version info and exit
145 --help Print this page
146 --cpuclock-test Perform test/validation of CPU clock
147 --cmdhelp=cmd Print command help, "all" for all of them
148 --enghelp=engine Print ioengine help, or list available ioengines
149 --enghelp=engine,cmd Print help for an ioengine cmd
150 --showcmd Turn a job file into command line options
151 --readonly Turn on safety read-only checks, preventing
153 --eta=when When ETA estimate should be printed
154 May be "always", "never" or "auto"
155 --eta-newline=time Force a new line for every 'time' period passed
156 --section=name Only run specified section in job file.
157 Multiple sections can be specified.
158 --alloc-size=kb Set smalloc pool to this size in kb (def 1024)
159 --warnings-fatal Fio parser warnings are fatal
160 --max-jobs Maximum number of threads/processes to support
161 --server=args Start backend server. See Client/Server section.
162 --client=host Connect to specified backend.
163 --idle-prof=option Report cpu idleness on a system or percpu basis
164 (option=system,percpu) or run unit work
165 calibration only (option=calibrate).
168 Any parameters following the options will be assumed to be job files,
169 unless they match a job file parameter. You can add as many as you want,
170 each job file will be regarded as a separate group and fio will stonewall
173 The --readonly switch is an extra safety guard to prevent accidentally
174 turning on a write setting when that is not desired. Fio will only write
175 if rw=write/randwrite/rw/randrw is given, but this extra safety net can
176 be used as an extra precaution. It will also enable a write check in the
177 io engine core to prevent an accidental write due to a fio bug.
179 The debug switch allows adding options that trigger certain logging
180 options in fio. Currently the options are:
182 process Dump info related to processes
183 file Dump info related to file actions
184 io Dump info related to IO queuing
185 mem Dump info related to memory allocations
186 blktrace Dump info related to blktrace setup
187 verify Dump info related to IO verification
188 all Enable all debug options
189 random Dump info related to random offset generation
190 parse Dump info related to option matching and parsing
191 diskutil Dump info related to disk utilization updates
192 job:x Dump info only related to job number x
193 mutex Dump info only related to mutex up/down ops
194 profile Dump info related to profile extensions
195 time Dump info related to internal time keeping
196 ? or help Show available debug options.
198 You can specify as many as you want, eg --debug=file,mem will enable
199 file and memory debugging.
201 The section switch is meant to make it easier to ship a bigger job file
202 instead of several smaller ones. Say you define a job file with light,
203 moderate, and heavy parts. Then you can ask fio to run the given part
204 only by giving it a --section=heavy command line option. The section
205 option only applies to job sections, the reserved 'global' section is
206 always parsed and taken into account.
208 Fio has an internal allocator for shared memory called smalloc. It
209 allocates shared structures from this pool. The pool defaults to 1024k
210 in size, and can grow to 128 pools. If running large jobs with randommap
211 enabled it can run out of memory, in which case the --alloc-size switch
212 is handy for starting with a larger pool size. The backing store is
213 files in /tmp. Fio cleans up after itself, while it is running you
214 may see .fio_smalloc.* files in /tmp.
220 See the HOWTO file for a more detailed description of parameters and what
221 they mean. This file contains the terse version. You can describe big and
222 complex setups with the command line, but generally it's a lot easier to
223 just write a simple job file to describe the workload. The job file format
224 is in the ini style format, as that is easy to read and write for the user.
226 The job file parameters are:
228 name=x Use 'x' as the identifier for this job.
229 description=x 'x' is a text description of the job.
230 directory=x Use 'x' as the top level directory for storing files
231 filename=x Force the use of 'x' as the filename for all files
232 in this thread. If not given, fio will make up
233 a suitable filename based on the thread and file
235 rw=x 'x' may be: read, randread, write, randwrite,
236 rw (read-write mix), randrw (read-write random mix)
237 rwmixcycle=x Base cycle for switching between read and write
239 rwmixread=x 'x' percentage of rw mix ios will be reads. If
240 rwmixwrite is also given, the last of the two will
241 be used if they don't add up to 100%.
242 rwmixwrite=x 'x' percentage of rw mix ios will be writes. See
244 rand_repeatable=x The sequence of random io blocks can be repeatable
245 across runs, if 'x' is 1.
246 size=x Set file size to x bytes (x string can include k/m/g)
247 ioengine=x 'x' may be: aio/libaio/linuxaio for Linux aio,
248 posixaio for POSIX aio, solarisaio for Solaris
249 native async IO, windowsaio for Windows native async IO,
250 sync for regular read/write io,
251 psync for regular pread/pwrite io, vsync for regular
252 readv/writev (with queuing emulation) mmap for mmap'ed
253 io, syslet-rw for syslet driven read/write, splice for
254 using splice/vmsplice, sg for direct SG_IO io, net
255 for network io, rdma for RDMA io, or cpuio for a
256 cycler burner load. sg only works on Linux on
257 SCSI (or SCSI-like devices, such as usb-storage or
258 sata/libata driven) devices. Fio also has a null
259 io engine, which is mainly used for testing
262 iodepth=x For async io, allow 'x' ios in flight
263 overwrite=x If 'x', layout a write file first.
264 nrfiles=x Spread io load over 'x' number of files per job,
266 prio=x Run io at prio X, 0-7 is the kernel allowed range
267 prioclass=x Run io at prio class X
268 bs=x Use 'x' for thread blocksize. May include k/m postfix.
269 bsrange=x-y Mix thread block sizes randomly between x and y. May
270 also include k/m postfix.
271 direct=x 1 for direct IO, 0 for buffered IO
272 thinktime=x "Think" x usec after each io
273 rate=x Throttle rate to x KB/sec
274 ratemin=x Quit if rate of x KB/sec can't be met
275 ratecycle=x ratemin averaged over x msecs
276 cpumask=x Only allow job to run on CPUs defined by mask.
277 cpus_allowed=x Like 'cpumask', but allow text setting of CPU affinity.
278 numa_cpu_nodes=x,y-z Allow job to run on specified NUMA nodes' CPU.
279 numa_mem_policy=m:x,y-z Setup numa memory allocation policy.
280 'm' stands for policy, such as local, interleave,
281 bind, prefer, local. 'x, y-z' are numa node(s) for
282 memory allocation according to policy.
283 fsync=x If writing with buffered IO, fsync after every
284 'x' blocks have been written.
285 end_fsync=x If 'x', run fsync() after end-of-job.
286 startdelay=x Start this thread x seconds after startup
287 runtime=x Terminate x seconds after startup. Can include a
288 normal time suffix if not given in seconds, such as
289 'm' for minutes, 'h' for hours, and 'd' for days.
290 offset=x Start io at offset x (x string can include k/m/g)
291 invalidate=x Invalidate page cache for file prior to doing io
292 sync=x Use sync writes if x and writing buffered IO.
293 mem=x If x == malloc, use malloc for buffers. If x == shm,
294 use shared memory for buffers. If x == mmap, use
296 exitall When one thread quits, terminate the others
297 bwavgtime=x Average bandwidth stats over an x msec window.
298 create_serialize=x If 'x', serialize file creation.
299 create_fsync=x If 'x', run fsync() after file creation.
300 unlink If set, unlink files when done.
301 loops=x Run the job 'x' number of times.
302 verify=x If 'x' == md5, use md5 for verifies. If 'x' == crc32,
303 use crc32 for verifies. md5 is 'safer', but crc32 is
304 a lot faster. Only makes sense for writing to a file.
305 For other types of checksumming, see HOWTO.
306 stonewall Wait for preceeding jobs to end before running.
307 numjobs=x Create 'x' similar entries for this job
308 thread Use pthreads instead of forked jobs
310 zoneskip=y Zone options must be paired. If given, the job
311 will skip y bytes for every x read/written. This
312 can be used to gauge hard drive speed over the entire
313 platter, without reading everything. Both x/y can
314 include k/m/g suffix.
315 read_iolog=x Open and read io pattern from file 'x'. The file format
316 is described in the HOWTO.
317 write_iolog=x Write an iolog to file 'x' in the same format as iolog.
318 The iolog options are exclusive, if both given the
319 read iolog will be performed. Specify a separate file
320 for each job, otherwise the iologs will be interspersed
321 and the file may be corrupt.
322 write_bw_log Write a bandwidth log.
323 write_lat_log Write a latency log.
324 lockmem=x Lock down x amount of memory on the machine, to
325 simulate a machine with less memory available. x can
326 include k/m/g suffix.
327 nice=x Run job at given nice value.
328 exec_prerun=x Run 'x' before job io is begun.
329 exec_postrun=x Run 'x' after job io has finished.
330 ioscheduler=x Use ioscheduler 'x' for this job.
331 cpuload=x For a CPU io thread, percentage of CPU time to attempt
333 cpuchunks=x Split burn cycles into pieces of x usecs.
340 Normally you would run fio as a stand-alone application on the machine
341 where the IO workload should be generated. However, it is also possible to
342 run the frontend and backend of fio separately. This makes it possible to
343 have a fio server running on the machine(s) where the IO workload should
344 be running, while controlling it from another machine.
346 To start the server, you would do:
350 on that machine, where args defines what fio listens to. The arguments
351 are of the form 'type,hostname or IP,port'. 'type' is either 'ip' (or ip4)
352 for TCP/IP v4, 'ip6' for TCP/IP v6, or 'sock' for a local unix domain socket.
353 'hostname' is either a hostname or IP address, and 'port' is the port to
354 listen to (only valid for TCP/IP, not a local socket). Some examples:
358 Start a fio server, listening on all interfaces on the default port (8765).
360 2) fio --server=ip:hostname,4444
362 Start a fio server, listening on IP belonging to hostname and on port 4444.
364 3) fio --server=ip6:::1,4444
366 Start a fio server, listening on IPv6 localhost ::1 and on port 4444.
368 4) fio --server=,4444
370 Start a fio server, listening on all interfaces on port 4444.
372 5) fio --server=1.2.3.4
374 Start a fio server, listening on IP 1.2.3.4 on the default port.
376 6) fio --server=sock:/tmp/fio.sock
378 Start a fio server, listening on the local socket /tmp/fio.sock.
380 When a server is running, you can connect to it from a client. The client
383 fio --local-args --client=server --remote-args <job file(s)>
385 where --local-args are arguments that are local to the client where it is
386 running, 'server' is the connect string, and --remote-args and <job file(s)>
387 are sent to the server. The 'server' string follows the same format as it
388 does on the server side, to allow IP/hostname/socket and port strings.
389 You can connect to multiple clients as well, to do that you could run:
391 fio --client=server2 <job file(s)> --client=server2 <job file(s)>
397 Fio works on (at least) Linux, Solaris, AIX, HP-UX, OSX, NetBSD, Windows
398 and FreeBSD. Some features and/or options may only be available on some of
399 the platforms, typically because those features only apply to that platform
400 (like the solarisaio engine, or the splice engine on Linux).
402 Some features are not available on FreeBSD/Solaris even if they could be
403 implemented, I'd be happy to take patches for that. An example of that is
404 disk utility statistics and (I think) huge page support, support for that
405 does exist in FreeBSD/Solaris.
407 Fio uses pthread mutexes for signalling and locking and FreeBSD does not
408 support process shared pthread mutexes. As a result, only threads are
409 supported on FreeBSD. This could be fixed with sysv ipc locking or
410 other locking alternatives.
412 Other *BSD platforms are untested, but fio should work there almost out
413 of the box. Since I don't do test runs or even compiles on those platforms,
414 your mileage may vary. Sending me patches for other platforms is greatly
415 appreciated. There's a lot of value in having the same test/benchmark tool
416 available on all platforms.
418 Note that POSIX aio is not enabled by default on AIX. If you get messages like:
420 Symbol resolution failed for /usr/lib/libc.a(posix_aio.o) because:
421 Symbol _posix_kaio_rdwr (number 2) is not exported from dependent module /unix.
423 you need to enable POSIX aio. Run the following commands as root:
425 # lsdev -C -l posix_aio0
426 posix_aio0 Defined Posix Asynchronous I/O
427 # cfgmgr -l posix_aio0
428 # lsdev -C -l posix_aio0
429 posix_aio0 Available Posix Asynchronous I/O
431 POSIX aio should work now. To make the change permanent:
433 # chdev -l posix_aio0 -P -a autoconfig='available'
440 Fio was written by Jens Axboe <axboe@kernel.dk> to enable flexible testing
441 of the Linux IO subsystem and schedulers. He got tired of writing
442 specific test applications to simulate a given workload, and found that
443 the existing io benchmark/test tools out there weren't flexible enough
444 to do what he wanted.
446 Jens Axboe <axboe@kernel.dk> 20060905