Make fio -Wshadow clean
[fio.git] / README
... / ...
CommitLineData
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
29There are also two official mirrors. Both of these are synced within
30an hour of commits landing at git.kernel.dk. So if the main repo is
31down for some reason, either one of those is safe to use:
32
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
35
36or
37
38 https://github.com/axboe/fio.git
39
40
41Binary packages
42---------------
43
44Debian:
45Starting with Debian "Squeeze", fio packages are part of the official
46Debian repository. http://packages.debian.org/search?keywords=fio
47
48Ubuntu:
49Starting with Ubuntu 10.04 LTS (aka "Lucid Lynx"), fio packages are part
50of the Ubuntu "universe" repository.
51http://packages.ubuntu.com/search?keywords=fio
52
53Red Hat, CentOS & Co:
54Dag Wieƫrs has RPMs for Red Hat related distros, find them here:
55http://dag.wieers.com/rpm/packages/fio/
56
57Mandriva:
58Mandriva has integrated fio into their package repository, so installing
59on that distro should be as easy as typing 'urpmi fio'.
60
61Solaris:
62Packages for Solaris are available from OpenCSW. Install their pkgutil
63tool (http://www.opencsw.org/get-it/pkgutil/) and then install fio via
64'pkgutil -i fio'.
65
66Windows:
67Bruce Cran <bruce@cran.org.uk> has fio packages for Windows at
68http://www.bluestop.org/fio/ .
69
70
71Mailing list
72------------
73
74The fio project mailing list is meant for anything related to fio including
75general discussion, bug reporting, questions, and development.
76
77An automated mail detailing recent commits is automatically sent to the
78list at most daily. The list address is fio@vger.kernel.org, subscribe
79by sending an email to majordomo@vger.kernel.org with
80
81 subscribe fio
82
83in the body of the email. Archives can be found here:
84
85 http://www.spinics.net/lists/fio/
86
87and archives for the old list can be found here:
88
89 http://maillist.kernel.dk/fio-devel/
90
91
92Building
93--------
94
95Just type 'configure', 'make' and 'make install'.
96
97Note that GNU make is required. On BSD it's available from devel/gmake;
98on Solaris it's in the SUNWgmake package. On platforms where GNU make
99isn't the default, type 'gmake' instead of 'make'.
100
101Configure will print the enabled options. Note that on Linux based
102platforms, the libaio development packages must be installed to use
103the libaio engine. Depending on distro, it is usually called
104libaio-devel or libaio-dev.
105
106For gfio, gtk 2.18 (or newer), associated glib threads, and cairo are required
107to be installed. gfio isn't built automatically and can be enabled
108with a --enable-gfio option to configure.
109
110To build FIO with a cross-compiler:
111 $ make clean
112 $ make CROSS_COMPILE=/path/to/toolchain/prefix
113Configure will attempt to determine the target platform automatically.
114
115It's possible to build fio for ESX as well, use the --esx switch to
116configure.
117
118
119Windows
120-------
121
122On Windows, Cygwin (http://www.cygwin.com/) is required in order to
123build fio. To create an MSI installer package install WiX 3.8 from
124http://wixtoolset.org and run dobuild.cmd from the
125os/windows directory.
126
127How to compile fio on 64-bit Windows:
128
129 1. Install Cygwin (http://www.cygwin.com/). Install 'make' and all
130 packages starting with 'mingw64-i686' and 'mingw64-x86_64'.
131 2. Open the Cygwin Terminal.
132 3. Go to the fio directory (source files).
133 4. Run 'make clean && make -j'.
134
135To build fio on 32-bit Windows, run './configure --build-32bit-win' before 'make'.
136
137It's recommended that once built or installed, fio be run in a Command Prompt
138or other 'native' console such as console2, since there are known to be display
139and signal issues when running it under a Cygwin shell
140(see http://code.google.com/p/mintty/issues/detail?id=56 for details).
141
142
143Command line
144------------
145
146$ fio
147 --debug Enable some debugging options (see below)
148 --parse-only Parse options only, don't start any IO
149 --output Write output to file
150 --runtime Runtime in seconds
151 --bandwidth-log Generate per-job bandwidth logs
152 --minimal Minimal (terse) output
153 --output-format=type Output format (terse,json,normal)
154 --terse-version=type Terse version output format (default 3, or 2 or 4).
155 --version Print version info and exit
156 --help Print this page
157 --cpuclock-test Perform test/validation of CPU clock
158 --crctest[=test] Test speed of checksum functions
159 --cmdhelp=cmd Print command help, "all" for all of them
160 --enghelp=engine Print ioengine help, or list available ioengines
161 --enghelp=engine,cmd Print help for an ioengine cmd
162 --showcmd Turn a job file into command line options
163 --readonly Turn on safety read-only checks, preventing
164 writes
165 --eta=when When ETA estimate should be printed
166 May be "always", "never" or "auto"
167 --eta-newline=time Force a new line for every 'time' period passed
168 --status-interval=t Force full status dump every 't' period passed
169 --section=name Only run specified section in job file.
170 Multiple sections can be specified.
171 --alloc-size=kb Set smalloc pool to this size in kb (def 1024)
172 --warnings-fatal Fio parser warnings are fatal
173 --max-jobs Maximum number of threads/processes to support
174 --server=args Start backend server. See Client/Server section.
175 --client=host Connect to specified backend.
176 --remote-config=file Tell fio server to load this local file
177 --idle-prof=option Report cpu idleness on a system or percpu basis
178 (option=system,percpu) or run unit work
179 calibration only (option=calibrate).
180 --inflate-log=log Inflate and output compressed log
181
182
183Any parameters following the options will be assumed to be job files,
184unless they match a job file parameter. Multiple job files can be listed
185and each job file will be regarded as a separate group. fio will stonewall
186execution between each group.
187
188The --readonly option is an extra safety guard to prevent users from
189accidentally starting a write workload when that is not desired. Fio
190will only write if rw=write/randwrite/rw/randrw is given. This extra
191safety net can be used as an extra precaution as --readonly will also
192enable a write check in the io engine core to prevent writes due to
193unknown user space bug(s).
194
195The --debug option triggers additional logging by fio.
196Currently, additional logging is available for:
197
198 process Dump info related to processes
199 file Dump info related to file actions
200 io Dump info related to IO queuing
201 mem Dump info related to memory allocations
202 blktrace Dump info related to blktrace setup
203 verify Dump info related to IO verification
204 all Enable all debug options
205 random Dump info related to random offset generation
206 parse Dump info related to option matching and parsing
207 diskutil Dump info related to disk utilization updates
208 job:x Dump info only related to job number x
209 mutex Dump info only related to mutex up/down ops
210 profile Dump info related to profile extensions
211 time Dump info related to internal time keeping
212 net Dump info related to networking connections
213 rate Dump info related to IO rate switching
214 compress Dump info related to log compress/decompress
215 ? or help Show available debug options.
216
217One can specify multiple debug options: e.g. --debug=file,mem will enable
218file and memory debugging.
219
220The --section option allows one to combine related jobs into one file.
221E.g. one job file could define light, moderate, and heavy sections. Tell fio to
222run only the "heavy" section by giving --section=heavy command line option.
223One can also specify the "write" operations in one section and "verify"
224operation in another section. The --section option only applies to job
225sections. The reserved 'global' section is always parsed and used.
226
227The --alloc-size switch allows one to use a larger pool size for smalloc.
228If running large jobs with randommap enabled, fio can run out of memory.
229Smalloc is an internal allocator for shared structures from a fixed size
230memory pool. The pool size defaults to 1024k and can grow to 128 pools.
231
232NOTE: While running .fio_smalloc.* backing store files are visible in /tmp.
233
234
235Job file
236--------
237
238See the HOWTO file for a complete description of job file syntax and
239parameters. The --cmdhelp option also lists all options. If used with
240an option argument, --cmdhelp will detail the given option. The job file
241format is in the ini style format, as that is easy for the user to review
242and modify.
243
244This README contains the terse version. Job files can describe big and
245complex setups that are not possible with the command line. Job files
246are a good practice even for simple jobs since the file provides an
247easily accessed record of the workload and can include comments.
248
249See the examples/ directory for inspiration on how to write job files. Note
250the copyright and license requirements currently apply to examples/ files.
251
252
253Client/server
254------------
255
256Normally fio is invoked as a stand-alone application on the machine
257where the IO workload should be generated. However, the frontend and
258backend of fio can be run separately. Ie the fio server can generate
259an IO workload on the "Device Under Test" while being controlled from
260another machine.
261
262Start the server on the machine which has access to the storage DUT:
263
264fio --server=args
265
266where args defines what fio listens to. The arguments are of the form
267'type,hostname or IP,port'. 'type' is either 'ip' (or ip4) for TCP/IP v4,
268'ip6' for TCP/IP v6, or 'sock' for a local unix domain socket.
269'hostname' is either a hostname or IP address, and 'port' is the port to
270listen to (only valid for TCP/IP, not a local socket). Some examples:
271
2721) fio --server
273
274 Start a fio server, listening on all interfaces on the default port (8765).
275
2762) fio --server=ip:hostname,4444
277
278 Start a fio server, listening on IP belonging to hostname and on port 4444.
279
2803) fio --server=ip6:::1,4444
281
282 Start a fio server, listening on IPv6 localhost ::1 and on port 4444.
283
2844) fio --server=,4444
285
286 Start a fio server, listening on all interfaces on port 4444.
287
2885) fio --server=1.2.3.4
289
290 Start a fio server, listening on IP 1.2.3.4 on the default port.
291
2926) fio --server=sock:/tmp/fio.sock
293
294 Start a fio server, listening on the local socket /tmp/fio.sock.
295
296Once a server is running, a "client" can connect to the fio server with:
297
298fio --local-args --client=<server> --remote-args <job file(s)>
299
300where --local-args are arguments for the client where it is
301running, 'server' is the connect string, and --remote-args and <job file(s)>
302are sent to the server. The 'server' string follows the same format as it
303does on the server side, to allow IP/hostname/socket and port strings.
304
305Fio can connect to multiple servers this way:
306
307fio --client=<server1> <job file(s)> --client=<server2> <job file(s)>
308
309If the job file is located on the fio server, then you can tell the server
310to load a local file as well. This is done by using --remote-config:
311
312fio --client=server --remote-config /path/to/file.fio
313
314Then the fio serer will open this local (to the server) job file instead
315of being passed one from the client.
316
317
318Platforms
319---------
320
321Fio works on (at least) Linux, Solaris, AIX, HP-UX, OSX, NetBSD, OpenBSD,
322Windows and FreeBSD. Some features and/or options may only be available on
323some of the platforms, typically because those features only apply to that
324platform (like the solarisaio engine, or the splice engine on Linux).
325
326Some features are not available on FreeBSD/Solaris even if they could be
327implemented, I'd be happy to take patches for that. An example of that is
328disk utility statistics and (I think) huge page support, support for that
329does exist in FreeBSD/Solaris.
330
331Fio uses pthread mutexes for signalling and locking and FreeBSD does not
332support process shared pthread mutexes. As a result, only threads are
333supported on FreeBSD. This could be fixed with sysv ipc locking or
334other locking alternatives.
335
336Other *BSD platforms are untested, but fio should work there almost out
337of the box. Since I don't do test runs or even compiles on those platforms,
338your mileage may vary. Sending me patches for other platforms is greatly
339appreciated. There's a lot of value in having the same test/benchmark tool
340available on all platforms.
341
342Note that POSIX aio is not enabled by default on AIX. Messages like these:
343
344 Symbol resolution failed for /usr/lib/libc.a(posix_aio.o) because:
345 Symbol _posix_kaio_rdwr (number 2) is not exported from dependent module /unix.
346
347indicate one needs to enable POSIX aio. Run the following commands as root:
348
349 # lsdev -C -l posix_aio0
350 posix_aio0 Defined Posix Asynchronous I/O
351 # cfgmgr -l posix_aio0
352 # lsdev -C -l posix_aio0
353 posix_aio0 Available Posix Asynchronous I/O
354
355POSIX aio should work now. To make the change permanent:
356
357 # chdev -l posix_aio0 -P -a autoconfig='available'
358 posix_aio0 changed
359
360
361Author
362------
363
364Fio was written by Jens Axboe <axboe@kernel.dk> to enable flexible testing
365of the Linux IO subsystem and schedulers. He got tired of writing
366specific test applications to simulate a given workload, and found that
367the existing io benchmark/test tools out there weren't flexible enough
368to do what he wanted.
369
370Jens Axboe <axboe@kernel.dk> 20060905
371