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.
46 -t <sec> Runtime in seconds
47 -l Generate per-job latency logs
48 -w Generate per-job bandwidth logs
49 -o <file> Log output to file
50 -m Minimal (terse) output
52 -v Print version information and exit
54 Any parameters following the options will be assumed to be job files.
55 You can add as many as you want, each job file will be regarded as a
56 separate group and fio will stonewall it's execution.
62 Only a few options can be controlled with command line parameters,
63 generally it's a lot easier to just write a simple job file to describe
64 the workload. The job file format is in the ini style format, as it's
65 easy to read and write for the user.
67 The job file parameters are:
69 name=x Use 'x' as the identifier for this job.
70 directory=x Use 'x' as the top level directory for storing files
71 rw=x 'x' may be: read, randread, write, randwrite,
72 rw (read-write mix), randrw (read-write random mix)
73 rwmixcycle=x Base cycle for switching between read and write
75 rwmixread=x 'x' percentage of rw mix ios will be reads. If
76 rwmixwrite is also given, the last of the two will
77 be used if they don't add up to 100%.
78 rwmixwrite=x 'x' percentage of rw mix ios will be writes. See
80 rand_repeatable=x The sequence of random io blocks can be repeatable
81 across runs, if 'x' is 1.
82 size=x Set file size to x bytes (x string can include k/m/g)
83 ioengine=x 'x' may be: aio/libaio/linuxaio for Linux aio,
84 posixaio for POSIX aio, sync for regular read/write io,
85 mmap for mmap'ed io, splice for using splice/vmsplice,
86 or sgio for direct SG_IO io. The latter only works on
87 Linux on SCSI (or SCSI-like devices, such as
88 usb-storage or sata/libata driven) devices.
89 iodepth=x For async io, allow 'x' ios in flight
90 overwrite=x If 'x', layout a write file first.
91 nrfiles=x Spread io load over 'x' number of files per job,
93 prio=x Run io at prio X, 0-7 is the kernel allowed range
94 prioclass=x Run io at prio class X
95 bs=x Use 'x' for thread blocksize. May include k/m postfix.
96 bsrange=x-y Mix thread block sizes randomly between x and y. May
97 also include k/m postfix.
98 direct=x 1 for direct IO, 0 for buffered IO
99 thinktime=x "Think" x usec after each io
100 rate=x Throttle rate to x KiB/sec
101 ratemin=x Quit if rate of x KiB/sec can't be met
102 ratecycle=x ratemin averaged over x msecs
103 cpumask=x Only allow job to run on CPUs defined by mask.
104 fsync=x If writing, fsync after every x blocks have been written
105 startdelay=x Start this thread x seconds after startup
106 timeout=x Terminate x seconds after startup. Can include a
107 normal time suffix if not given in seconds, such as
108 'm' for minutes, 'h' for hours, and 'd' for days.
109 offset=x Start io at offset x (x string can include k/m/g)
110 invalidate=x Invalidate page cache for file prior to doing io
111 sync=x Use sync writes if x and writing
112 mem=x If x == malloc, use malloc for buffers. If x == shm,
113 use shm for buffers. If x == mmap, use anon mmap.
114 exitall When one thread quits, terminate the others
115 bwavgtime=x Average bandwidth stats over an x msec window.
116 create_serialize=x If 'x', serialize file creation.
117 create_fsync=x If 'x', run fsync() after file creation.
118 end_fsync=x If 'x', run fsync() after end-of-job.
119 loops=x Run the job 'x' number of times.
120 verify=x If 'x' == md5, use md5 for verifies. If 'x' == crc32,
121 use crc32 for verifies. md5 is 'safer', but crc32 is
122 a lot faster. Only makes sense for writing to a file.
123 stonewall Wait for preceeding jobs to end before running.
124 numjobs=x Create 'x' similar entries for this job
125 thread Use pthreads instead of forked jobs
127 zoneskip=y Zone options must be paired. If given, the job
128 will skip y bytes for every x read/written. This
129 can be used to gauge hard drive speed over the entire
130 platter, without reading everything. Both x/y can
131 include k/m/g suffix.
132 iolog=x Open and read io pattern from file 'x'. The file must
133 contain one io action per line in the following format:
135 where with rw=0/1 for read/write, and the offset
136 and length entries being in bytes.
137 write_iolog=x Write an iolog to file 'x' in the same format as iolog.
138 The iolog options are exclusive, if both given the
139 read iolog will be performed.
140 lockmem=x Lock down x amount of memory on the machine, to
141 simulate a machine with less memory available. x can
142 include k/m/g suffix.
143 nice=x Run job at given nice value.
144 exec_prerun=x Run 'x' before job io is begun.
145 exec_postrun=x Run 'x' after job io has finished.
146 ioscheduler=x Use ioscheduler 'x' for this job.
147 cpuload=x For a CPU io thread, percentage of CPU time to attempt
149 cpuchunks=x Split burn cycles into pieces of x.
152 Examples using a job file
153 -------------------------
155 Example 1) Two random readers
157 Lets say we want to simulate two threads reading randomly from a file
158 each. They will be doing IO in 4KiB chunks, using raw (O_DIRECT) IO.
159 Since they share most parameters, we'll put those in the [global]
160 section. Job 1 will use a 128MiB file, job 2 will use a 256MiB file.
165 ioengine=sync ; regular read/write(2), the default
178 Generally the [] bracketed name specifies a file name, but the "global"
179 keyword is reserved for setting options that are inherited by each
180 subsequent job description. It's possible to have several [global]
181 sections in the job file, each one adds options that are inherited by
182 jobs defined below it. The name can also point to a block device, such
183 as /dev/sda. To run the above job file, simply do:
187 Example 2) Many random writers
189 Say we want to exercise the IO subsystem some more. We'll define 64
190 threads doing random buffered writes. We'll let each thread use async io
191 with a depth of 4 ios in flight. A job file would then look like this:
208 This will create files.[0-63] and perform the random writes to them.
210 There are endless ways to define jobs, the examples/ directory contains
214 Interpreting the output
215 -----------------------
217 fio spits out a lot of output. While running, fio will display the
218 status of the jobs created. An example of that would be:
220 Threads running: 1: [_r] [24.79% done] [eta 00h:01m:31s]
222 The characters inside the square brackets denote the current status of
223 each thread. The possible values (in typical life cycle order) are:
227 P Thread setup, but not started.
229 I Thread initialized, waiting.
230 R Running, doing sequential reads.
231 r Running, doing random reads.
232 W Running, doing sequential writes.
233 w Running, doing random writes.
234 M Running, doing mixed sequential reads/writes.
235 m Running, doing mixed random reads/writes.
236 F Running, currently waiting for fsync()
237 V Running, doing verification of written data.
238 E Thread exited, not reaped by main thread yet.
241 The other values are fairly self explanatory - number of threads
242 currently running and doing io, and the estimated completion percentage
243 and time for the running group. It's impossible to estimate runtime
244 of the following groups (if any).
246 When fio is done (or interrupted by ctrl-c), it will show the data for
247 each thread, group of threads, and disks in that order. For each data
248 direction, the output looks like:
250 Client1 (g=0): err= 0:
251 write: io= 32MiB, bw= 666KiB/s, runt= 50320msec
252 slat (msec): min= 0, max= 136, avg= 0.03, dev= 1.92
253 clat (msec): min= 0, max= 631, avg=48.50, dev=86.82
254 bw (KiB/s) : min= 0, max= 1196, per=51.00%, avg=664.02, dev=681.68
255 cpu : usr=1.49%, sys=0.25%, ctx=7969
257 The client number is printed, along with the group id and error of that
258 thread. Below is the io statistics, here for writes. In the order listed,
261 io= Number of megabytes io performed
262 bw= Average bandwidth rate
263 runt= The runtime of that thread
264 slat= Submission latency (avg being the average, dev being the
265 standard deviation). This is the time it took to submit
266 the io. For sync io, the slat is really the completion
267 latency, since queue/complete is one operation there.
268 clat= Completion latency. Same names as slat, this denotes the
269 time from submission to completion of the io pieces. For
270 sync io, clat will usually be equal (or very close) to 0,
271 as the time from submit to complete is basically just
272 CPU time (io has already been done, see slat explanation).
273 bw= Bandwidth. Same names as the xlat stats, but also includes
274 an approximate percentage of total aggregate bandwidth
275 this thread received in this group. This last value is
276 only really useful if the threads in this group are on the
277 same disk, since they are then competing for disk access.
278 cpu= CPU usage. User and system time, along with the number
279 of context switches this thread went through.
281 After each client has been listed, the group statistics are printed. They
284 Run status group 0 (all jobs):
285 READ: io=64MiB, aggrb=22178, minb=11355, maxb=11814, mint=2840msec, maxt=2955msec
286 WRITE: io=64MiB, aggrb=1302, minb=666, maxb=669, mint=50093msec, maxt=50320msec
288 For each data direction, it prints:
290 io= Number of megabytes io performed.
291 aggrb= Aggregate bandwidth of threads in this group.
292 minb= The minimum average bandwidth a thread saw.
293 maxb= The maximum average bandwidth a thread saw.
294 mint= The smallest runtime of the threads in that group.
295 maxt= The longest runtime of the threads in that group.
297 And finally, the disk statistics are printed. They will look like this:
299 Disk stats (read/write):
300 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
302 Each value is printed for both reads and writes, with reads first. The
305 ios= Number of ios performed by all groups.
306 merge= Number of merges io the io scheduler.
307 ticks= Number of ticks we kept the disk busy.
308 io_queue= Total time spent in the disk queue.
309 util= The disk utilization. A value of 100% means we kept the disk
310 busy constantly, 50% would be a disk idling half of the time.
316 For scripted usage where you typically want to generate tables or graphs
317 of the results, fio can output the results in a comma seperated format.
318 The format is one long line of values, such as:
320 client1,0,0,936,331,2894,0,0,0.000000,0.000000,1,170,22.115385,34.290410,16,714,84.252874%,366.500000,566.417819,3496,1237,2894,0,0,0.000000,0.000000,0,246,6.671625,21.436952,0,2534,55.465300%,1406.600000,2008.044216,0.000000%,0.431928%,1109
322 Split up, the format is as follows:
324 jobname, groupid, error
326 KiB IO, bandwidth (KiB/sec), runtime (msec)
327 Submission latency: min, max, mean, deviation
328 Completion latency: min, max, mean, deviation
329 Bw: min, max, aggreate percentage of total, mean, deviation
331 KiB IO, bandwidth (KiB/sec), runtime (msec)
332 Submission latency: min, max, mean, deviation
333 Completion latency: min, max, mean, deviation
334 Bw: min, max, aggreate percentage of total, mean, deviation
335 CPU usage: user, system, context switches
341 Fio was written by Jens Axboe <axboe@kernel.dk> to enable flexible testing
342 of the Linux IO subsystem and schedulers. He got tired of writing
343 specific test applications to simulate a given workload, and found that
344 the existing io benchmark/test tools out there weren't flexible enough
345 to do what he wanted.
347 Jens Axboe <axboe@kernel.dk> 20060905