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 The http protocol also works, path is the same.
21 Snapshots are frequently generated and they include the git meta data as
22 well. You can download them here:
24 http://brick.kernel.dk/snaps/
26 Pascal Bleser <guru@unixtech.be> has fio RPMs in his repository for
27 SUSE variants, you can find them here:
29 http://linux01.gwdg.de/~pbleser/rpm-navigation.php?cat=System/fio
31 Dag Wieƫrs has RPMs for Red Hat related distros, find them here:
33 http://dag.wieers.com/rpm/packages/fio/
35 Mandriva has integrated fio into their package repository, so installing
36 on that distro should be as easy as typing 'urpmi fio'.
42 There's a mailing list associated with fio. It's meant for general
43 discussion, bug reporting, questions - basically anything that has to
44 do with fio. An automated mail detailing recent commits is automatically
45 sent to the list at most daily. The list address is fio-devel@kernel.dk,
46 subscribe by sending an empty email to fio-devel+subscribe@kernel.dk.
52 Just type 'make' and 'make install'. If on FreeBSD, for now you have to
53 specify the FreeBSD Makefile with -f, eg:
55 $ make -f Makefile.Freebsd && make -f Makefile.FreeBSD install
57 Likewise with OpenSolaris, use the Makefile.solaris to compile there.
58 This might change in the future if I opt for an autoconf type setup.
65 --debug Enable some debugging options (see below)
66 --output Write output to file
67 --timeout Runtime in seconds
68 --latency-log Generate per-job latency logs
69 --bandwidth-log Generate per-job bandwidth logs
70 --minimal Minimal (terse) output
71 --version Print version info and exit
72 --help Print this page
73 --cmdhelp=cmd Print command help, "all" for all of them
74 --showcmd Turn a job file into command line options
75 --readonly Turn on safety read-only checks, preventing writes
76 --eta=when When ETA estimate should be printed
77 May be "always", "never" or "auto"
78 --section=name Only run specified section in job file
81 Any parameters following the options will be assumed to be job files,
82 unless they match a job file parameter. You can add as many as you want,
83 each job file will be regarded as a separate group and fio will stonewall
86 The --readonly switch is an extra safety guard to prevent accidentically
87 turning on a write setting when that is not desired. Fio will only write
88 if rw=write/randwrite/rw/randrw is given, but this extra safety net can
89 be used as an extra precaution. It will also enable a write check in the
90 io engine core to prevent an accidental write due to a fio bug.
92 The debug switch allows adding options that trigger certain logging
93 options in fio. Currently the options are:
95 process Dump info related to processes
96 file Dump info related to file actions
97 io Dump info related to IO queuing
98 mem Dump info related to memory allocations
99 blktrace Dump info related to blktrace setup
100 verify Dump info related to IO verification
101 all Enable all debug options
102 random Dump info related to random offset generation
103 parse Dump info related to option matching and parsing
104 diskutil Dump info related to disk utilization updates
105 ? or help Show available debug options.
107 You can specify as many as you want, eg --debug=file,mem will enable
108 file and memory debugging.
110 The section switch is meant to make it easier to ship a bigger job file
111 instead of several smaller ones. Say you define a job file with light,
112 moderate, and heavy parts. Then you can ask fio to run the given part
113 only by giving it a --section=heavy command line option. The section
114 option only applies to job sections, the reserved 'global' section is
115 always parsed and taken into account.
121 See the HOWTO file for a more detailed description of parameters and what
122 they mean. This file contains the terse version. You can describe big and
123 complex setups with the command line, but generally it's a lot easier to
124 just write a simple job file to describe the workload. The job file format
125 is in the ini style format, as that is easy to read and write for the user.
127 The job file parameters are:
129 name=x Use 'x' as the identifier for this job.
130 description=x 'x' is a text description of the job.
131 directory=x Use 'x' as the top level directory for storing files
132 filename=x Force the use of 'x' as the filename for all files
133 in this thread. If not given, fio will make up
134 a suitable filename based on the thread and file
136 rw=x 'x' may be: read, randread, write, randwrite,
137 rw (read-write mix), randrw (read-write random mix)
138 rwmixcycle=x Base cycle for switching between read and write
140 rwmixread=x 'x' percentage of rw mix ios will be reads. If
141 rwmixwrite is also given, the last of the two will
142 be used if they don't add up to 100%.
143 rwmixwrite=x 'x' percentage of rw mix ios will be writes. See
145 rand_repeatable=x The sequence of random io blocks can be repeatable
146 across runs, if 'x' is 1.
147 size=x Set file size to x bytes (x string can include k/m/g)
148 ioengine=x 'x' may be: aio/libaio/linuxaio for Linux aio,
149 posixaio for POSIX aio, sync for regular read/write io,
150 psync for regular pread/pwrite io, vsync for regular
151 readv/writev (with queuing emulation) mmap for mmap'ed
152 io, syslet-rw for syslet driven read/write, splice for
153 using splice/vmsplice, sgio for direct SG_IO io, net
154 for network io, or cpuio for a cycler burner load. sgio
155 only works on Linux on SCSI (or SCSI-like devices, such
156 as usb-storage or sata/libata driven) devices. Fio also
157 has a null io engine, which is mainly used for testing
160 iodepth=x For async io, allow 'x' ios in flight
161 overwrite=x If 'x', layout a write file first.
162 nrfiles=x Spread io load over 'x' number of files per job,
164 prio=x Run io at prio X, 0-7 is the kernel allowed range
165 prioclass=x Run io at prio class X
166 bs=x Use 'x' for thread blocksize. May include k/m postfix.
167 bsrange=x-y Mix thread block sizes randomly between x and y. May
168 also include k/m postfix.
169 direct=x 1 for direct IO, 0 for buffered IO
170 thinktime=x "Think" x usec after each io
171 rate=x Throttle rate to x KiB/sec
172 ratemin=x Quit if rate of x KiB/sec can't be met
173 ratecycle=x ratemin averaged over x msecs
174 cpumask=x Only allow job to run on CPUs defined by mask.
175 cpus_allowed=x Like 'cpumask', but allow text setting of CPU affinity.
176 fsync=x If writing with buffered IO, fsync after every
177 'x' blocks have been written.
178 end_fsync=x If 'x', run fsync() after end-of-job.
179 startdelay=x Start this thread x seconds after startup
180 runtime=x Terminate x seconds after startup. Can include a
181 normal time suffix if not given in seconds, such as
182 'm' for minutes, 'h' for hours, and 'd' for days.
183 offset=x Start io at offset x (x string can include k/m/g)
184 invalidate=x Invalidate page cache for file prior to doing io
185 sync=x Use sync writes if x and writing buffered IO.
186 mem=x If x == malloc, use malloc for buffers. If x == shm,
187 use shared memory for buffers. If x == mmap, use
189 exitall When one thread quits, terminate the others
190 bwavgtime=x Average bandwidth stats over an x msec window.
191 create_serialize=x If 'x', serialize file creation.
192 create_fsync=x If 'x', run fsync() after file creation.
193 unlink If set, unlink files when done.
194 loops=x Run the job 'x' number of times.
195 verify=x If 'x' == md5, use md5 for verifies. If 'x' == crc32,
196 use crc32 for verifies. md5 is 'safer', but crc32 is
197 a lot faster. Only makes sense for writing to a file.
198 stonewall Wait for preceeding jobs to end before running.
199 numjobs=x Create 'x' similar entries for this job
200 thread Use pthreads instead of forked jobs
202 zoneskip=y Zone options must be paired. If given, the job
203 will skip y bytes for every x read/written. This
204 can be used to gauge hard drive speed over the entire
205 platter, without reading everything. Both x/y can
206 include k/m/g suffix.
207 iolog=x Open and read io pattern from file 'x'. The file must
208 contain one io action per line in the following format:
210 where with rw=0/1 for read/write, and the offset
211 and length entries being in bytes.
212 write_iolog=x Write an iolog to file 'x' in the same format as iolog.
213 The iolog options are exclusive, if both given the
214 read iolog will be performed.
215 write_bw_log Write a bandwidth log.
216 write_lat_log Write a latency log.
217 lockmem=x Lock down x amount of memory on the machine, to
218 simulate a machine with less memory available. x can
219 include k/m/g suffix.
220 nice=x Run job at given nice value.
221 exec_prerun=x Run 'x' before job io is begun.
222 exec_postrun=x Run 'x' after job io has finished.
223 ioscheduler=x Use ioscheduler 'x' for this job.
224 cpuload=x For a CPU io thread, percentage of CPU time to attempt
226 cpuchunks=x Split burn cycles into pieces of x usecs.
232 Fio was written by Jens Axboe <axboe@kernel.dk> to enable flexible testing
233 of the Linux IO subsystem and schedulers. He got tired of writing
234 specific test applications to simulate a given workload, and found that
235 the existing io benchmark/test tools out there weren't flexible enough
236 to do what he wanted.
238 Jens Axboe <axboe@kernel.dk> 20060905