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://brick.kernel.dk/data/git/fio.git
19 Snapshots are frequently generated and they include the git meta data as
20 well. You can download them here:
22 http://brick.kernel.dk/snaps/
24 Pascal Bleser <guru@unixtech.be> has fio RPMs in his repository, you
27 http://linux01.gwdg.de/~pbleser/rpm-navigation.php?cat=System/fio
33 Just type 'make' and 'make install'. If on FreeBSD, for now you have to
34 specify the FreeBSD Makefile with -f, eg:
36 $ make -f Makefile.Freebsd && make -f Makefile.FreeBSD install
38 Likewise with OpenSolaris, use the Makefile.solaris to compile there.
39 This might change in the future if I opt for an autoconf type setup.
47 -b block size in KiB for each io
48 -t <sec> Runtime in seconds
49 -r For random io, sequence must be repeatable
50 -R <on> If one thread fails to meet rate, quit all
51 -o <on> Use direct IO is 1, buffered if 0
52 -l Generate per-job latency logs
53 -w Generate per-job bandwidth logs
54 -f <file> Read <file> for job descriptions
55 -O <file> Log output to file
57 -v Print version information and exit
63 Only a few options can be controlled with command line parameters,
64 generally it's a lot easier to just write a simple job file to describe
65 the workload. The job file format is in the ini style format, as it's
66 easy to read and write for the user.
68 The job file parameters are:
70 name=x Use 'x' as the identifier for this job.
71 directory=x Use 'x' as the top level directory for storing files
72 rw=x 'x' may be: read, randread, write, randwrite,
73 rw (read-write mix), randrw (read-write random mix)
74 rwmixcycle=x Base cycle for switching between read and write
76 rwmixread=x 'x' percentage of rw mix ios will be reads. If
77 rwmixwrite is also given, the last of the two will
78 be used if they don't add up to 100%.
79 rwmixwrite=x 'x' percentage of rw mix ios will be writes. See
81 size=x Set file size to x bytes (x string can include k/m/g)
82 ioengine=x 'x' may be: aio/libaio/linuxaio for Linux aio,
83 posixaio for POSIX aio, sync for regular read/write io,
84 mmap for mmap'ed io, splice for using splice/vmsplice,
85 or sgio for direct SG_IO io. The latter only works on
86 Linux on SCSI (or SCSI-like devices, such as
87 usb-storage or sata/libata driven) devices.
88 iodepth=x For async io, allow 'x' ios in flight
89 overwrite=x If 'x', layout a write file first.
90 prio=x Run io at prio X, 0-7 is the kernel allowed range
91 prioclass=x Run io at prio class X
92 bs=x Use 'x' for thread blocksize. May include k/m postfix.
93 bsrange=x-y Mix thread block sizes randomly between x and y. May
94 also include k/m postfix.
95 direct=x 1 for direct IO, 0 for buffered IO
96 thinktime=x "Think" x usec after each io
97 rate=x Throttle rate to x KiB/sec
98 ratemin=x Quit if rate of x KiB/sec can't be met
99 ratecycle=x ratemin averaged over x msecs
100 cpumask=x Only allow job to run on CPUs defined by mask.
101 fsync=x If writing, fsync after every x blocks have been written
102 startdelay=x Start this thread x seconds after startup
103 timeout=x Terminate x seconds after startup
104 offset=x Start io at offset x (x string can include k/m/g)
105 invalidate=x Invalidate page cache for file prior to doing io
106 sync=x Use sync writes if x and writing
107 mem=x If x == malloc, use malloc for buffers. If x == shm,
108 use shm for buffers. If x == mmap, use anon mmap.
109 exitall When one thread quits, terminate the others
110 bwavgtime=x Average bandwidth stats over an x msec window.
111 create_serialize=x If 'x', serialize file creation.
112 create_fsync=x If 'x', run fsync() after file creation.
113 end_fsync=x If 'x', run fsync() after end-of-job.
114 loops=x Run the job 'x' number of times.
115 verify=x If 'x' == md5, use md5 for verifies. If 'x' == crc32,
116 use crc32 for verifies. md5 is 'safer', but crc32 is
117 a lot faster. Only makes sense for writing to a file.
118 stonewall Wait for preceeding jobs to end before running.
119 numjobs=x Create 'x' similar entries for this job
120 thread Use pthreads instead of forked jobs
122 zoneskip=y Zone options must be paired. If given, the job
123 will skip y bytes for every x read/written. This
124 can be used to gauge hard drive speed over the entire
125 platter, without reading everything. Both x/y can
126 include k/m/g suffix.
127 iolog=x Open and read io pattern from file 'x'. The file must
128 contain one io action per line in the following format:
130 where with rw=0/1 for read/write, and the offset
131 and length entries being in bytes.
132 write_iolog=x Write an iolog to file 'x' in the same format as iolog.
133 The iolog options are exclusive, if both given the
134 read iolog will be performed.
135 lockmem=x Lock down x amount of memory on the machine, to
136 simulate a machine with less memory available. x can
137 include k/m/g suffix.
138 nice=x Run job at given nice value.
139 exec_prerun=x Run 'x' before job io is begun.
140 exec_postrun=x Run 'x' after job io has finished.
141 ioscheduler=x Use ioscheduler 'x' for this job.
144 Examples using a job file
145 -------------------------
147 Example 1) Two random readers
149 Lets say we want to simulate two threads reading randomly from a file
150 each. They will be doing IO in 4KiB chunks, using raw (O_DIRECT) IO.
151 Since they share most parameters, we'll put those in the [global]
152 section. Job 1 will use a 128MiB file, job 2 will use a 256MiB file.
157 ioengine=sync ; regular read/write(2), the default
170 Generally the [] bracketed name specifies a file name, but the "global"
171 keyword is reserved for setting options that are inherited by each
172 subsequent job description. It's possible to have several [global]
173 sections in the job file, each one adds options that are inherited by
174 jobs defined below it. The name can also point to a block device, such
175 as /dev/sda. To run the above job file, simply do:
179 Example 2) Many random writers
181 Say we want to exercise the IO subsystem some more. We'll define 64
182 threads doing random buffered writes. We'll let each thread use async io
183 with a depth of 4 ios in flight. A job file would then look like this:
200 This will create files.[0-63] and perform the random writes to them.
202 There are endless ways to define jobs, the examples/ directory contains
206 Interpreting the output
207 -----------------------
209 fio spits out a lot of output. While running, fio will display the
210 status of the jobs created. An example of that would be:
212 Threads now running: 2 : [ww] [5.73% done]
214 The characters inside the square brackets denote the current status of
215 each thread. The possible values (in typical life cycle order) are:
219 P Thread setup, but not started.
221 I Thread initialized, waiting.
222 R Running, doing sequential reads.
223 r Running, doing random reads.
224 W Running, doing sequential writes.
225 w Running, doing random writes.
226 M Running, doing mixed sequential reads/writes.
227 m Running, doing mixed random reads/writes.
228 F Running, currently waiting for fsync()
229 V Running, doing verification of written data.
230 E Thread exited, not reaped by main thread yet.
233 The other values are fairly self explanatory - number of threads
234 currently running and doing io, and the estimated completion percentage
237 When fio is done (or interrupted by ctrl-c), it will show the data for
238 each thread, group of threads, and disks in that order. For each data
239 direction, the output looks like:
241 Client1 (g=0): err= 0:
242 write: io= 32MiB, bw= 666KiB/s, runt= 50320msec
243 slat (msec): min= 0, max= 136, avg= 0.03, dev= 1.92
244 clat (msec): min= 0, max= 631, avg=48.50, dev=86.82
245 bw (KiB/s) : min= 0, max= 1196, per=51.00%, avg=664.02, dev=681.68
246 cpu : usr=1.49%, sys=0.25%, ctx=7969
248 The client number is printed, along with the group id and error of that
249 thread. Below is the io statistics, here for writes. In the order listed,
252 io= Number of megabytes io performed
253 bw= Average bandwidth rate
254 runt= The runtime of that thread
255 slat= Submission latency (avg being the average, dev being the
256 standard deviation). This is the time it took to submit
257 the io. For sync io, the slat is really the completion
258 latency, since queue/complete is one operation there.
259 clat= Completion latency. Same names as slat, this denotes the
260 time from submission to completion of the io pieces. For
261 sync io, clat will usually be equal (or very close) to 0,
262 as the time from submit to complete is basically just
263 CPU time (io has already been done, see slat explanation).
264 bw= Bandwidth. Same names as the xlat stats, but also includes
265 an approximate percentage of total aggregate bandwidth
266 this thread received in this group. This last value is
267 only really useful if the threads in this group are on the
268 same disk, since they are then competing for disk access.
269 cpu= CPU usage. User and system time, along with the number
270 of context switches this thread went through.
272 After each client has been listed, the group statistics are printed. They
275 Run status group 0 (all jobs):
276 READ: io=64MiB, aggrb=22178, minb=11355, maxb=11814, mint=2840msec, maxt=2955msec
277 WRITE: io=64MiB, aggrb=1302, minb=666, maxb=669, mint=50093msec, maxt=50320msec
279 For each data direction, it prints:
281 io= Number of megabytes io performed.
282 aggrb= Aggregate bandwidth of threads in this group.
283 minb= The minimum average bandwidth a thread saw.
284 maxb= The maximum average bandwidth a thread saw.
285 mint= The smallest runtime of the threads in that group.
286 maxt= The longest runtime of the threads in that group.
288 And finally, the disk statistics are printed. They will look like this:
290 Disk stats (read/write):
291 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
293 Each value is printed for both reads and writes, with reads first. The
296 ios= Number of ios performed by all groups.
297 merge= Number of merges io the io scheduler.
298 ticks= Number of ticks we kept the disk busy.
299 io_queue= Total time spent in the disk queue.
300 util= The disk utilization. A value of 100% means we kept the disk
301 busy constantly, 50% would be a disk idling half of the time.
307 Fio was written by Jens Axboe <axboe@suse.de> to enable flexible testing
308 of the Linux IO subsystem and schedulers. He got tired of writing
309 specific test applications to simulate a given workload, and found that
310 the existing io benchmark/test tools out there weren't flexible enough
311 to do what he wanted.
313 Jens Axboe <axboe@suse.de> 20060609