1fio
2---
3
4fio is a tool that will spawn a number of threads or processes doing a
5particular type of io action as specified by the user. fio takes a
6number of global parameters, each inherited by the thread unless
7otherwise parameters given to them overriding that setting is given.
8The typical use of fio is to write a job file matching the io load
9one wants to simulate.
10
11
12Source
13------
14
15fio resides in a git repo, the canonical place is:
16
17 git://git.kernel.dk/fio.git
18
19When inside a corporate firewall, git:// URL sometimes does not work.
20If git:// does not work, use the http protocol instead:
21
22 http://git.kernel.dk/fio.git
23
24Snapshots are frequently generated and include the git meta data as well.
25Snapshots can download from:
26
27 http://brick.kernel.dk/snaps/
28
29
30Binary packages
31---------------
32
33Debian:
34Starting with Debian "Squeeze", fio packages are part of the official
35Debian repository. http://packages.debian.org/search?keywords=fio
36
37Ubuntu:
38Starting with Ubuntu 10.04 LTS (aka "Lucid Lynx"), fio packages are part
39of the Ubuntu "universe" repository.
40http://packages.ubuntu.com/search?keywords=fio
41
42Red Hat, CentOS & Co:
43Dag Wieërs has RPMs for Red Hat related distros, find them here:
44http://dag.wieers.com/rpm/packages/fio/
45
46Mandriva:
47Mandriva has integrated fio into their package repository, so installing
48on that distro should be as easy as typing 'urpmi fio'.
49
50Solaris:
51Packages for Solaris are available from OpenCSW. Install their pkgutil
52tool (http://www.opencsw.org/get-it/pkgutil/) and then install fio via
53'pkgutil -i fio'.
54
55Windows:
56Bruce Cran <bruce@cran.org.uk> has fio packages for Windows at
57http://www.bluestop.org/fio/ .
58
59
60Mailing list
61------------
62
63The fio project mailing list is meant for anything related to fio including
64general discussion, bug reporting, questions, and development.
65
66An automated mail detailing recent commits is automatically sent to the
67list at most daily. The list address is fio@vger.kernel.org, subscribe
68by sending an email to majordomo@vger.kernel.org with
69
70 subscribe fio
71
72in the body of the email. Archives can be found here:
73
74 http://www.spinics.net/lists/fio/
75
76and archives for the old list can be found here:
77
78 http://maillist.kernel.dk/fio-devel/
79
80
81Building
82--------
83
84Just type 'configure', 'make' and 'make install'.
85
86Note that GNU make is required. On BSD it's available from devel/gmake;
87on Solaris it's in the SUNWgmake package. On platforms where GNU make
88isn't the default, type 'gmake' instead of 'make'.
89
90Configure will print the enabled options. Note that on Linux based
91platforms, the libaio development packages must be installed to use
92the libaio engine. Depending on distro, it is usually called
93libaio-devel or libaio-dev.
94
95For gfio, gtk 2.18 (or newer), associated glib threads, and cairo are required
96to be installed. gfio isn't built automatically and can be enabled
97with a --enable-gfio option to configure.
98
99To build FIO with a cross-compiler:
100 $ make clean
101 $ make CROSS_COMPILE=/path/to/toolchain/prefix
102Configure will attempt to determine the target platform automatically.
103
104
105Windows
106-------
107
108On Windows, Cygwin (http://www.cygwin.com/) is required in order to
109build fio. To create an MSI installer package install WiX 3.8 from
110http://wixtoolset.org and run dobuild.cmd from the
111os/windows directory.
112
113How to compile fio on 64-bit Windows:
114
115 1. Install Cygwin (http://www.cygwin.com/). Install 'make' and all
116 packages starting with 'mingw64-i686' and 'mingw64-x86_64'.
117 2. Open the Cygwin Terminal.
118 3. Go to the fio directory (source files).
119 4. Run 'make clean && make -j'.
120
121To build fio on 32-bit Windows, run './configure --build-32bit-win' before 'make'.
122
123It's recommended that once built or installed, fio be run in a Command Prompt
124or other 'native' console such as console2, since there are known to be display
125and signal issues when running it under a Cygwin shell
126(see http://code.google.com/p/mintty/issues/detail?id=56 for details).
127
128
129Command line
130------------
131
132$ fio
133 --debug Enable some debugging options (see below)
134 --parse-only Parse options only, don't start any IO
135 --output Write output to file
136 --runtime Runtime in seconds
137 --latency-log Generate per-job latency logs
138 --bandwidth-log Generate per-job bandwidth logs
139 --minimal Minimal (terse) output
140 --output-format=type Output format (terse,json,normal)
141 --terse-version=type Terse version output format (default 3, or 2 or 4).
142 --version Print version info and exit
143 --help Print this page
144 --cpuclock-test Perform test/validation of CPU clock
145 --crctest[=test] Test speed of checksum functions
146 --cmdhelp=cmd Print command help, "all" for all of them
147 --enghelp=engine Print ioengine help, or list available ioengines
148 --enghelp=engine,cmd Print help for an ioengine cmd
149 --showcmd Turn a job file into command line options
150 --readonly Turn on safety read-only checks, preventing
151 writes
152 --eta=when When ETA estimate should be printed
153 May be "always", "never" or "auto"
154 --eta-newline=time Force a new line for every 'time' period passed
155 --status-interval=t Force full status dump every 't' 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).
166
167
168Any parameters following the options will be assumed to be job files,
169unless they match a job file parameter. Multiple job files can be listed
170and each job file will be regarded as a separate group. fio will stonewall
171execution between each group.
172
173The --readonly option is an extra safety guard to prevent users from
174accidentally starting a write workload when that is not desired. Fio
175will only write if rw=write/randwrite/rw/randrw is given. This extra
176safety net can be used as an extra precaution as --readonly will also
177enable a write check in the io engine core to prevent writes due to
178unknown user space bug(s).
179
180The --debug option triggers additional logging by fio.
181Currently, additional logging is available for:
182
183 process Dump info related to processes
184 file Dump info related to file actions
185 io Dump info related to IO queuing
186 mem Dump info related to memory allocations
187 blktrace Dump info related to blktrace setup
188 verify Dump info related to IO verification
189 all Enable all debug options
190 random Dump info related to random offset generation
191 parse Dump info related to option matching and parsing
192 diskutil Dump info related to disk utilization updates
193 job:x Dump info only related to job number x
194 mutex Dump info only related to mutex up/down ops
195 profile Dump info related to profile extensions
196 time Dump info related to internal time keeping
197 net Dump info related to networking connections
198 rate Dump info related to IO rate switching
199 ? or help Show available debug options.
200
201One can specify multiple debug options: e.g. --debug=file,mem will enable
202file and memory debugging.
203
204The --section option allows one to combine related jobs into one file.
205E.g. one job file could define light, moderate, and heavy sections. Tell fio to
206run only the "heavy" section by giving --section=heavy command line option.
207One can also specify the "write" operations in one section and "verify"
208operation in another section. The --section option only applies to job
209sections. The reserved 'global' section is always parsed and used.
210
211The --alloc-size switch allows one to use a larger pool size for smalloc.
212If running large jobs with randommap enabled, fio can run out of memory.
213Smalloc is an internal allocator for shared structures from a fixed size
214memory pool. The pool size defaults to 1024k and can grow to 128 pools.
215
216NOTE: While running .fio_smalloc.* backing store files are visible in /tmp.
217
218
219Job file
220--------
221
222See the HOWTO file for a complete description of job file syntax and
223parameters. The --cmdhelp option also lists all options. If used with
224an option argument, --cmdhelp will detail the given option. The job file
225format is in the ini style format, as that is easy for the user to review
226and modify.
227
228This README contains the terse version. Job files can describe big and
229complex setups that are not possible with the command line. Job files
230are a good practice even for simple jobs since the file provides an
231easily accessed record of the workload and can include comments.
232
233See the examples/ directory for inspiration on how to write job files. Note
234the copyright and license requirements currently apply to examples/ files.
235
236
237Client/server
238------------
239
240Normally fio is invoked as a stand-alone application on the machine
241where the IO workload should be generated. However, the frontend and
242backend of fio can be run separately. Ie the fio server can generate
243an IO workload on the "Device Under Test" while being controlled from
244another machine.
245
246Start the server on the machine which has access to the storage DUT:
247
248fio --server=args
249
250where args defines what fio listens to. The arguments are of the form
251'type,hostname or IP,port'. 'type' is either 'ip' (or ip4) for TCP/IP v4,
252'ip6' for TCP/IP v6, or 'sock' for a local unix domain socket.
253'hostname' is either a hostname or IP address, and 'port' is the port to
254listen to (only valid for TCP/IP, not a local socket). Some examples:
255
2561) fio --server
257
258 Start a fio server, listening on all interfaces on the default port (8765).
259
2602) fio --server=ip:hostname,4444
261
262 Start a fio server, listening on IP belonging to hostname and on port 4444.
263
2643) fio --server=ip6:::1,4444
265
266 Start a fio server, listening on IPv6 localhost ::1 and on port 4444.
267
2684) fio --server=,4444
269
270 Start a fio server, listening on all interfaces on port 4444.
271
2725) fio --server=1.2.3.4
273
274 Start a fio server, listening on IP 1.2.3.4 on the default port.
275
2766) fio --server=sock:/tmp/fio.sock
277
278 Start a fio server, listening on the local socket /tmp/fio.sock.
279
280Once a server is running, a "client" can connect to the fio server with:
281
282fio --local-args --client=<server> --remote-args <job file(s)>
283
284where --local-args are arguments for the client where it is
285running, 'server' is the connect string, and --remote-args and <job file(s)>
286are sent to the server. The 'server' string follows the same format as it
287does on the server side, to allow IP/hostname/socket and port strings.
288
289Fio can connect to multiple servers this way:
290
291fio --client=<server1> <job file(s)> --client=<server2> <job file(s)>
292
293
294Platforms
295---------
296
297Fio works on (at least) Linux, Solaris, AIX, HP-UX, OSX, NetBSD, OpenBSD,
298Windows and FreeBSD. Some features and/or options may only be available on
299some of the platforms, typically because those features only apply to that
300platform (like the solarisaio engine, or the splice engine on Linux).
301
302Some features are not available on FreeBSD/Solaris even if they could be
303implemented, I'd be happy to take patches for that. An example of that is
304disk utility statistics and (I think) huge page support, support for that
305does exist in FreeBSD/Solaris.
306
307Fio uses pthread mutexes for signalling and locking and FreeBSD does not
308support process shared pthread mutexes. As a result, only threads are
309supported on FreeBSD. This could be fixed with sysv ipc locking or
310other locking alternatives.
311
312Other *BSD platforms are untested, but fio should work there almost out
313of the box. Since I don't do test runs or even compiles on those platforms,
314your mileage may vary. Sending me patches for other platforms is greatly
315appreciated. There's a lot of value in having the same test/benchmark tool
316available on all platforms.
317
318Note that POSIX aio is not enabled by default on AIX. Messages like these:
319
320 Symbol resolution failed for /usr/lib/libc.a(posix_aio.o) because:
321 Symbol _posix_kaio_rdwr (number 2) is not exported from dependent module /unix.
322
323indicate one needs to enable POSIX aio. Run the following commands as root:
324
325 # lsdev -C -l posix_aio0
326 posix_aio0 Defined Posix Asynchronous I/O
327 # cfgmgr -l posix_aio0
328 # lsdev -C -l posix_aio0
329 posix_aio0 Available Posix Asynchronous I/O
330
331POSIX aio should work now. To make the change permanent:
332
333 # chdev -l posix_aio0 -P -a autoconfig='available'
334 posix_aio0 changed
335
336
337Author
338------
339
340Fio was written by Jens Axboe <axboe@kernel.dk> to enable flexible testing
341of the Linux IO subsystem and schedulers. He got tired of writing
342specific test applications to simulate a given workload, and found that
343the existing io benchmark/test tools out there weren't flexible enough
344to do what he wanted.
345
346Jens Axboe <axboe@kernel.dk> 20060905
347
348