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.
60 If your compile fails with an error like this:
63 In file included from fio.h:23,
65 os/os.h:15:20: error: libaio.h: No such file or directory
66 In file included from gettime.c:8:
67 fio.h:119: error: field 'iocb' has incomplete type
68 make: *** [gettime.o] Error 1
70 Check that you have the libaio development package installed. On RPM
71 based distros, it's typically called libaio-devel.
78 --debug Enable some debugging options (see below)
79 --output Write output to file
80 --timeout Runtime in seconds
81 --latency-log Generate per-job latency logs
82 --bandwidth-log Generate per-job bandwidth logs
83 --minimal Minimal (terse) output
84 --version Print version info and exit
85 --help Print this page
86 --cmdhelp=cmd Print command help, "all" for all of them
87 --showcmd Turn a job file into command line options
88 --readonly Turn on safety read-only checks, preventing writes
89 --eta=when When ETA estimate should be printed
90 May be "always", "never" or "auto"
91 --section=name Only run specified section in job file
92 --alloc-size=kb Set smalloc pool to this size in kb (def 1024)
95 Any parameters following the options will be assumed to be job files,
96 unless they match a job file parameter. You can add as many as you want,
97 each job file will be regarded as a separate group and fio will stonewall
100 The --readonly switch is an extra safety guard to prevent accidentically
101 turning on a write setting when that is not desired. Fio will only write
102 if rw=write/randwrite/rw/randrw is given, but this extra safety net can
103 be used as an extra precaution. It will also enable a write check in the
104 io engine core to prevent an accidental write due to a fio bug.
106 The debug switch allows adding options that trigger certain logging
107 options in fio. Currently the options are:
109 process Dump info related to processes
110 file Dump info related to file actions
111 io Dump info related to IO queuing
112 mem Dump info related to memory allocations
113 blktrace Dump info related to blktrace setup
114 verify Dump info related to IO verification
115 all Enable all debug options
116 random Dump info related to random offset generation
117 parse Dump info related to option matching and parsing
118 diskutil Dump info related to disk utilization updates
119 job:x Dump info only related to job number x
120 ? or help Show available debug options.
122 You can specify as many as you want, eg --debug=file,mem will enable
123 file and memory debugging.
125 The section switch is meant to make it easier to ship a bigger job file
126 instead of several smaller ones. Say you define a job file with light,
127 moderate, and heavy parts. Then you can ask fio to run the given part
128 only by giving it a --section=heavy command line option. The section
129 option only applies to job sections, the reserved 'global' section is
130 always parsed and taken into account.
132 Fio has an internal allocator for shared memory called smalloc. It
133 allocates shared structures from this pool. The pool defaults to 1024k
134 in size, and can grow to 32 pools. If running large jobs with randommap
135 enabled it can run out of memory, in which case the --alloc-size switch
136 is handy for starting with a larger pool size.
142 See the HOWTO file for a more detailed description of parameters and what
143 they mean. This file contains the terse version. You can describe big and
144 complex setups with the command line, but generally it's a lot easier to
145 just write a simple job file to describe the workload. The job file format
146 is in the ini style format, as that is easy to read and write for the user.
148 The job file parameters are:
150 name=x Use 'x' as the identifier for this job.
151 description=x 'x' is a text description of the job.
152 directory=x Use 'x' as the top level directory for storing files
153 filename=x Force the use of 'x' as the filename for all files
154 in this thread. If not given, fio will make up
155 a suitable filename based on the thread and file
157 rw=x 'x' may be: read, randread, write, randwrite,
158 rw (read-write mix), randrw (read-write random mix)
159 rwmixcycle=x Base cycle for switching between read and write
161 rwmixread=x 'x' percentage of rw mix ios will be reads. If
162 rwmixwrite is also given, the last of the two will
163 be used if they don't add up to 100%.
164 rwmixwrite=x 'x' percentage of rw mix ios will be writes. See
166 rand_repeatable=x The sequence of random io blocks can be repeatable
167 across runs, if 'x' is 1.
168 size=x Set file size to x bytes (x string can include k/m/g)
169 ioengine=x 'x' may be: aio/libaio/linuxaio for Linux aio,
170 posixaio for POSIX aio, sync for regular read/write io,
171 psync for regular pread/pwrite io, vsync for regular
172 readv/writev (with queuing emulation) mmap for mmap'ed
173 io, syslet-rw for syslet driven read/write, splice for
174 using splice/vmsplice, sgio for direct SG_IO io, net
175 for network io, or cpuio for a cycler burner load. sgio
176 only works on Linux on SCSI (or SCSI-like devices, such
177 as usb-storage or sata/libata driven) devices. Fio also
178 has a null io engine, which is mainly used for testing
181 iodepth=x For async io, allow 'x' ios in flight
182 overwrite=x If 'x', layout a write file first.
183 nrfiles=x Spread io load over 'x' number of files per job,
185 prio=x Run io at prio X, 0-7 is the kernel allowed range
186 prioclass=x Run io at prio class X
187 bs=x Use 'x' for thread blocksize. May include k/m postfix.
188 bsrange=x-y Mix thread block sizes randomly between x and y. May
189 also include k/m postfix.
190 direct=x 1 for direct IO, 0 for buffered IO
191 thinktime=x "Think" x usec after each io
192 rate=x Throttle rate to x KiB/sec
193 ratemin=x Quit if rate of x KiB/sec can't be met
194 ratecycle=x ratemin averaged over x msecs
195 cpumask=x Only allow job to run on CPUs defined by mask.
196 cpus_allowed=x Like 'cpumask', but allow text setting of CPU affinity.
197 fsync=x If writing with buffered IO, fsync after every
198 'x' blocks have been written.
199 end_fsync=x If 'x', run fsync() after end-of-job.
200 startdelay=x Start this thread x seconds after startup
201 runtime=x Terminate x seconds after startup. Can include a
202 normal time suffix if not given in seconds, such as
203 'm' for minutes, 'h' for hours, and 'd' for days.
204 offset=x Start io at offset x (x string can include k/m/g)
205 invalidate=x Invalidate page cache for file prior to doing io
206 sync=x Use sync writes if x and writing buffered IO.
207 mem=x If x == malloc, use malloc for buffers. If x == shm,
208 use shared memory for buffers. If x == mmap, use
210 exitall When one thread quits, terminate the others
211 bwavgtime=x Average bandwidth stats over an x msec window.
212 create_serialize=x If 'x', serialize file creation.
213 create_fsync=x If 'x', run fsync() after file creation.
214 unlink If set, unlink files when done.
215 loops=x Run the job 'x' number of times.
216 verify=x If 'x' == md5, use md5 for verifies. If 'x' == crc32,
217 use crc32 for verifies. md5 is 'safer', but crc32 is
218 a lot faster. Only makes sense for writing to a file.
219 stonewall Wait for preceeding jobs to end before running.
220 numjobs=x Create 'x' similar entries for this job
221 thread Use pthreads instead of forked jobs
223 zoneskip=y Zone options must be paired. If given, the job
224 will skip y bytes for every x read/written. This
225 can be used to gauge hard drive speed over the entire
226 platter, without reading everything. Both x/y can
227 include k/m/g suffix.
228 iolog=x Open and read io pattern from file 'x'. The file must
229 contain one io action per line in the following format:
231 where with rw=0/1 for read/write, and the offset
232 and length entries being in bytes.
233 write_iolog=x Write an iolog to file 'x' in the same format as iolog.
234 The iolog options are exclusive, if both given the
235 read iolog will be performed.
236 write_bw_log Write a bandwidth log.
237 write_lat_log Write a latency log.
238 lockmem=x Lock down x amount of memory on the machine, to
239 simulate a machine with less memory available. x can
240 include k/m/g suffix.
241 nice=x Run job at given nice value.
242 exec_prerun=x Run 'x' before job io is begun.
243 exec_postrun=x Run 'x' after job io has finished.
244 ioscheduler=x Use ioscheduler 'x' for this job.
245 cpuload=x For a CPU io thread, percentage of CPU time to attempt
247 cpuchunks=x Split burn cycles into pieces of x usecs.
253 Fio was written by Jens Axboe <axboe@kernel.dk> to enable flexible testing
254 of the Linux IO subsystem and schedulers. He got tired of writing
255 specific test applications to simulate a given workload, and found that
256 the existing io benchmark/test tools out there weren't flexible enough
257 to do what he wanted.
259 Jens Axboe <axboe@kernel.dk> 20060905