Add support for O_ATOMIC
[fio.git] / HOWTO
CommitLineData
71bfa161
JA
1Table of contents
2-----------------
3
41. Overview
52. How fio works
63. Running fio
74. Job file format
85. Detailed list of parameters
96. Normal output
107. Terse output
25c8b9d7 118. Trace file format
43f09da1 129. CPU idleness profiling
71bfa161
JA
13
141.0 Overview and history
15------------------------
16fio was originally written to save me the hassle of writing special test
17case programs when I wanted to test a specific workload, either for
18performance reasons or to find/reproduce a bug. The process of writing
19such a test app can be tiresome, especially if you have to do it often.
20Hence I needed a tool that would be able to simulate a given io workload
21without resorting to writing a tailored test case again and again.
22
23A test work load is difficult to define, though. There can be any number
24of processes or threads involved, and they can each be using their own
25way of generating io. You could have someone dirtying large amounts of
26memory in an memory mapped file, or maybe several threads issuing
27reads using asynchronous io. fio needed to be flexible enough to
28simulate both of these cases, and many more.
29
302.0 How fio works
31-----------------
32The first step in getting fio to simulate a desired io workload, is
33writing a job file describing that specific setup. A job file may contain
34any number of threads and/or files - the typical contents of the job file
35is a global section defining shared parameters, and one or more job
36sections describing the jobs involved. When run, fio parses this file
37and sets everything up as described. If we break down a job from top to
38bottom, it contains the following basic parameters:
39
40 IO type Defines the io pattern issued to the file(s).
41 We may only be reading sequentially from this
42 file(s), or we may be writing randomly. Or even
43 mixing reads and writes, sequentially or randomly.
44
45 Block size In how large chunks are we issuing io? This may be
46 a single value, or it may describe a range of
47 block sizes.
48
49 IO size How much data are we going to be reading/writing.
50
51 IO engine How do we issue io? We could be memory mapping the
52 file, we could be using regular read/write, we
d0ff85df 53 could be using splice, async io, syslet, or even
71bfa161
JA
54 SG (SCSI generic sg).
55
6c219763 56 IO depth If the io engine is async, how large a queuing
71bfa161
JA
57 depth do we want to maintain?
58
59 IO type Should we be doing buffered io, or direct/raw io?
60
61 Num files How many files are we spreading the workload over.
62
63 Num threads How many threads or processes should we spread
64 this workload over.
66c098b8 65
71bfa161
JA
66The above are the basic parameters defined for a workload, in addition
67there's a multitude of parameters that modify other aspects of how this
68job behaves.
69
70
713.0 Running fio
72---------------
73See the README file for command line parameters, there are only a few
74of them.
75
76Running fio is normally the easiest part - you just give it the job file
77(or job files) as parameters:
78
79$ fio job_file
80
81and it will start doing what the job_file tells it to do. You can give
82more than one job file on the command line, fio will serialize the running
83of those files. Internally that is the same as using the 'stonewall'
84parameter described the the parameter section.
85
b4692828
JA
86If the job file contains only one job, you may as well just give the
87parameters on the command line. The command line parameters are identical
88to the job parameters, with a few extra that control global parameters
89(see README). For example, for the job file parameter iodepth=2, the
c2b1e753
JA
90mirror command line option would be --iodepth 2 or --iodepth=2. You can
91also use the command line for giving more than one job entry. For each
92--name option that fio sees, it will start a new job with that name.
93Command line entries following a --name entry will apply to that job,
94until there are no more entries or a new --name entry is seen. This is
95similar to the job file options, where each option applies to the current
96job until a new [] job entry is seen.
b4692828 97
71bfa161
JA
98fio does not need to run as root, except if the files or devices specified
99in the job section requires that. Some other options may also be restricted,
6c219763 100such as memory locking, io scheduler switching, and decreasing the nice value.
71bfa161
JA
101
102
1034.0 Job file format
104-------------------
105As previously described, fio accepts one or more job files describing
106what it is supposed to do. The job file format is the classic ini file,
107where the names enclosed in [] brackets define the job name. You are free
108to use any ascii name you want, except 'global' which has special meaning.
109A global section sets defaults for the jobs described in that file. A job
110may override a global section parameter, and a job file may even have
111several global sections if so desired. A job is only affected by a global
65db0851
JA
112section residing above it. If the first character in a line is a ';' or a
113'#', the entire line is discarded as a comment.
71bfa161 114
3c54bc46 115So let's look at a really simple job file that defines two processes, each
b22989b9 116randomly reading from a 128MB file.
71bfa161
JA
117
118; -- start job file --
119[global]
120rw=randread
121size=128m
122
123[job1]
124
125[job2]
126
127; -- end job file --
128
129As you can see, the job file sections themselves are empty as all the
130described parameters are shared. As no filename= option is given, fio
c2b1e753
JA
131makes up a filename for each of the jobs as it sees fit. On the command
132line, this job would look as follows:
133
134$ fio --name=global --rw=randread --size=128m --name=job1 --name=job2
135
71bfa161 136
3c54bc46 137Let's look at an example that has a number of processes writing randomly
71bfa161
JA
138to files.
139
140; -- start job file --
141[random-writers]
142ioengine=libaio
143iodepth=4
144rw=randwrite
145bs=32k
146direct=0
147size=64m
148numjobs=4
149
150; -- end job file --
151
152Here we have no global section, as we only have one job defined anyway.
153We want to use async io here, with a depth of 4 for each file. We also
b22989b9 154increased the buffer size used to 32KB and define numjobs to 4 to
71bfa161 155fork 4 identical jobs. The result is 4 processes each randomly writing
b22989b9 156to their own 64MB file. Instead of using the above job file, you could
b4692828
JA
157have given the parameters on the command line. For this case, you would
158specify:
159
160$ fio --name=random-writers --ioengine=libaio --iodepth=4 --rw=randwrite --bs=32k --direct=0 --size=64m --numjobs=4
71bfa161 161
74929ac2
JA
1624.1 Environment variables
163-------------------------
164
3c54bc46
AC
165fio also supports environment variable expansion in job files. Any
166substring of the form "${VARNAME}" as part of an option value (in other
167words, on the right of the `='), will be expanded to the value of the
168environment variable called VARNAME. If no such environment variable
169is defined, or VARNAME is the empty string, the empty string will be
170substituted.
171
172As an example, let's look at a sample fio invocation and job file:
173
174$ SIZE=64m NUMJOBS=4 fio jobfile.fio
175
176; -- start job file --
177[random-writers]
178rw=randwrite
179size=${SIZE}
180numjobs=${NUMJOBS}
181; -- end job file --
182
183This will expand to the following equivalent job file at runtime:
184
185; -- start job file --
186[random-writers]
187rw=randwrite
188size=64m
189numjobs=4
190; -- end job file --
191
71bfa161
JA
192fio ships with a few example job files, you can also look there for
193inspiration.
194
74929ac2
JA
1954.2 Reserved keywords
196---------------------
197
198Additionally, fio has a set of reserved keywords that will be replaced
199internally with the appropriate value. Those keywords are:
200
201$pagesize The architecture page size of the running system
202$mb_memory Megabytes of total memory in the system
203$ncpus Number of online available CPUs
204
205These can be used on the command line or in the job file, and will be
206automatically substituted with the current system values when the job
892a6ffc
JA
207is run. Simple math is also supported on these keywords, so you can
208perform actions like:
209
210size=8*$mb_memory
211
212and get that properly expanded to 8 times the size of memory in the
213machine.
74929ac2 214
71bfa161
JA
215
2165.0 Detailed list of parameters
217-------------------------------
218
219This section describes in details each parameter associated with a job.
220Some parameters take an option of a given type, such as an integer or
221a string. The following types are used:
222
223str String. This is a sequence of alpha characters.
b09da8fa 224time Integer with possible time suffix. In seconds unless otherwise
e417fd66
JA
225 specified, use eg 10m for 10 minutes. Accepts s/m/h for seconds,
226 minutes, and hours.
b09da8fa
JA
227int SI integer. A whole number value, which may contain a suffix
228 describing the base of the number. Accepted suffixes are k/m/g/t/p,
229 meaning kilo, mega, giga, tera, and peta. The suffix is not case
57fc29fa
JA
230 sensitive, and you may also include trailing 'b' (eg 'kb' is the same
231 as 'k'). So if you want to specify 4096, you could either write
b09da8fa 232 out '4096' or just give 4k. The suffixes signify base 2 values, so
57fc29fa
JA
233 1024 is 1k and 1024k is 1m and so on, unless the suffix is explicitly
234 set to a base 10 value using 'kib', 'mib', 'gib', etc. If that is the
235 case, then 1000 is used as the multiplier. This can be handy for
236 disks, since manufacturers generally use base 10 values when listing
237 the capacity of a drive. If the option accepts an upper and lower
238 range, use a colon ':' or minus '-' to separate such values. May also
239 include a prefix to indicate numbers base. If 0x is used, the number
240 is assumed to be hexadecimal. See irange.
71bfa161
JA
241bool Boolean. Usually parsed as an integer, however only defined for
242 true and false (1 and 0).
b09da8fa 243irange Integer range with suffix. Allows value range to be given, such
bf9a3edb 244 as 1024-4096. A colon may also be used as the separator, eg
0c9baf91
JA
245 1k:4k. If the option allows two sets of ranges, they can be
246 specified with a ',' or '/' delimiter: 1k-4k/8k-32k. Also see
f7fa2653 247 int.
83349190 248float_list A list of floating numbers, separated by a ':' character.
71bfa161
JA
249
250With the above in mind, here follows the complete list of fio job
251parameters.
252
253name=str ASCII name of the job. This may be used to override the
254 name printed by fio for this job. Otherwise the job
c2b1e753 255 name is used. On the command line this parameter has the
6c219763 256 special purpose of also signaling the start of a new
c2b1e753 257 job.
71bfa161 258
61697c37
JA
259description=str Text description of the job. Doesn't do anything except
260 dump this text description when this job is run. It's
261 not parsed.
262
3776041e 263directory=str Prefix filenames with this directory. Used to place files
71bfa161
JA
264 in a different location than "./".
265
266filename=str Fio normally makes up a filename based on the job name,
267 thread number, and file number. If you want to share
268 files between threads in a job or several jobs, specify
ed92ac0c 269 a filename for each of them to override the default. If
414c2a3e 270 the ioengine used is 'net', the filename is the host, port,
0fd666bf 271 and protocol to use in the format of =host,port,protocol.
414c2a3e
JA
272 See ioengine=net for more. If the ioengine is file based, you
273 can specify a number of files by separating the names with a
274 ':' colon. So if you wanted a job to open /dev/sda and /dev/sdb
275 as the two working files, you would use
30a4588a
JA
276 filename=/dev/sda:/dev/sdb. On Windows, disk devices are
277 accessed as \\.\PhysicalDrive0 for the first device,
278 \\.\PhysicalDrive1 for the second etc. Note: Windows and
279 FreeBSD prevent write access to areas of the disk containing
280 in-use data (e.g. filesystems).
281 If the wanted filename does need to include a colon, then
282 escape that with a '\' character. For instance, if the filename
283 is "/dev/dsk/foo@3,0:c", then you would use
284 filename="/dev/dsk/foo@3,0\:c". '-' is a reserved name, meaning
285 stdin or stdout. Which of the two depends on the read/write
286 direction set.
71bfa161 287
de98bd30
JA
288filename_format=str
289 If sharing multiple files between jobs, it is usually necessary
290 to have fio generate the exact names that you want. By default,
291 fio will name a file based on the default file format
292 specification of jobname.jobnumber.filenumber. With this
293 option, that can be customized. Fio will recognize and replace
294 the following keywords in this string:
295
296 $jobname
297 The name of the worker thread or process.
298
299 $jobnum
300 The incremental number of the worker thread or
301 process.
302
303 $filenum
304 The incremental number of the file for that worker
305 thread or process.
306
307 To have dependent jobs share a set of files, this option can
308 be set to have fio generate filenames that are shared between
309 the two. For instance, if testfiles.$filenum is specified,
310 file number 4 for any job will be named testfiles.4. The
311 default of $jobname.$jobnum.$filenum will be used if
312 no other format specifier is given.
313
bbf6b540
JA
314opendir=str Tell fio to recursively add any file it can find in this
315 directory and down the file system tree.
316
3776041e 317lockfile=str Fio defaults to not locking any files before it does
4d4e80f2
JA
318 IO to them. If a file or file descriptor is shared, fio
319 can serialize IO to that file to make the end result
320 consistent. This is usual for emulating real workloads that
321 share files. The lock modes are:
322
323 none No locking. The default.
324 exclusive Only one thread/process may do IO,
325 excluding all others.
326 readwrite Read-write locking on the file. Many
327 readers may access the file at the
328 same time, but writes get exclusive
329 access.
330
d3aad8f2 331readwrite=str
71bfa161
JA
332rw=str Type of io pattern. Accepted values are:
333
334 read Sequential reads
335 write Sequential writes
336 randwrite Random writes
337 randread Random reads
10b023db 338 rw,readwrite Sequential mixed reads and writes
71bfa161
JA
339 randrw Random mixed reads and writes
340
341 For the mixed io types, the default is to split them 50/50.
342 For certain types of io the result may still be skewed a bit,
211097b2 343 since the speed may be different. It is possible to specify
38dad62d
JA
344 a number of IO's to do before getting a new offset, this is
345 one by appending a ':<nr>' to the end of the string given.
346 For a random read, it would look like 'rw=randread:8' for
059b0802 347 passing in an offset modifier with a value of 8. If the
ddb754db 348 suffix is used with a sequential IO pattern, then the value
059b0802
JA
349 specified will be added to the generated offset for each IO.
350 For instance, using rw=write:4k will skip 4k for every
351 write. It turns sequential IO into sequential IO with holes.
352 See the 'rw_sequencer' option.
38dad62d
JA
353
354rw_sequencer=str If an offset modifier is given by appending a number to
355 the rw=<str> line, then this option controls how that
356 number modifies the IO offset being generated. Accepted
357 values are:
358
359 sequential Generate sequential offset
360 identical Generate the same offset
361
362 'sequential' is only useful for random IO, where fio would
363 normally generate a new random offset for every IO. If you
364 append eg 8 to randread, you would get a new random offset for
211097b2
JA
365 every 8 IO's. The result would be a seek for only every 8
366 IO's, instead of for every IO. Use rw=randread:8 to specify
38dad62d
JA
367 that. As sequential IO is already sequential, setting
368 'sequential' for that would not result in any differences.
369 'identical' behaves in a similar fashion, except it sends
370 the same offset 8 number of times before generating a new
371 offset.
71bfa161 372
90fef2d1
JA
373kb_base=int The base unit for a kilobyte. The defacto base is 2^10, 1024.
374 Storage manufacturers like to use 10^3 or 1000 as a base
375 ten unit instead, for obvious reasons. Allow values are
376 1024 or 1000, with 1024 being the default.
377
771e58be
JA
378unified_rw_reporting=bool Fio normally reports statistics on a per
379 data direction basis, meaning that read, write, and trim are
380 accounted and reported separately. If this option is set,
381 the fio will sum the results and report them as "mixed"
382 instead.
383
ee738499
JA
384randrepeat=bool For random IO workloads, seed the generator in a predictable
385 way so that results are repeatable across repetitions.
386
2615cc4b
JA
387use_os_rand=bool Fio can either use the random generator supplied by the OS
388 to generator random offsets, or it can use it's own internal
389 generator (based on Tausworthe). Default is to use the
390 internal generator, which is often of better quality and
391 faster.
392
a596f047
EG
393fallocate=str Whether pre-allocation is performed when laying down files.
394 Accepted values are:
395
396 none Do not pre-allocate space
397 posix Pre-allocate via posix_fallocate()
398 keep Pre-allocate via fallocate() with
399 FALLOC_FL_KEEP_SIZE set
400 0 Backward-compatible alias for 'none'
401 1 Backward-compatible alias for 'posix'
402
403 May not be available on all supported platforms. 'keep' is only
404 available on Linux.If using ZFS on Solaris this must be set to
405 'none' because ZFS doesn't support it. Default: 'posix'.
7bc8c2cf 406
d2f3ac35
JA
407fadvise_hint=bool By default, fio will use fadvise() to advise the kernel
408 on what IO patterns it is likely to issue. Sometimes you
409 want to test specific IO patterns without telling the
410 kernel about it, in which case you can disable this option.
411 If set, fio will use POSIX_FADV_SEQUENTIAL for sequential
412 IO and POSIX_FADV_RANDOM for random IO.
413
f7fa2653 414size=int The total size of file io for this job. Fio will run until
7616cafe
JA
415 this many bytes has been transferred, unless runtime is
416 limited by other options (such as 'runtime', for instance).
3776041e 417 Unless specific nrfiles and filesize options are given,
7616cafe 418 fio will divide this size between the available files
d6667268
JA
419 specified by the job. If not set, fio will use the full
420 size of the given files or devices. If the the files
7bb59102
JA
421 do not exist, size must be given. It is also possible to
422 give size as a percentage between 1 and 100. If size=20%
423 is given, fio will use 20% of the full size of the given
424 files or devices.
71bfa161 425
f7fa2653 426filesize=int Individual file sizes. May be a range, in which case fio
9c60ce64
JA
427 will select sizes for files at random within the given range
428 and limited to 'size' in total (if that is given). If not
429 given, each created file is the same size.
430
74586c1e
JA
431fill_device=bool
432fill_fs=bool Sets size to something really large and waits for ENOSPC (no
aa31f1f1 433 space left on device) as the terminating condition. Only makes
de98bd30 434 sense with sequential write. For a read workload, the mount
4f12432e
JA
435 point will be filled first then IO started on the result. This
436 option doesn't make sense if operating on a raw device node,
437 since the size of that is already known by the file system.
438 Additionally, writing beyond end-of-device will not return
439 ENOSPC there.
aa31f1f1 440
f7fa2653
JA
441blocksize=int
442bs=int The block size used for the io units. Defaults to 4k. Values
443 can be given for both read and writes. If a single int is
444 given, it will apply to both. If a second int is specified
f90eff5a 445 after a comma, it will apply to writes only. In other words,
d9472271
JA
446 the format is either bs=read_and_write or bs=read,write,trim.
447 bs=4k,8k will thus use 4k blocks for reads, 8k blocks for
448 writes, and 8k for trims. You can terminate the list with
449 a trailing comma. bs=4k,8k, would use the default value for
450 trims.. If you only wish to set the write size, you
787f7e95
JA
451 can do so by passing an empty read size - bs=,8k will set
452 8k for writes and leave the read default value.
a00735e6 453
2b7a01d0
JA
454blockalign=int
455ba=int At what boundary to align random IO offsets. Defaults to
456 the same as 'blocksize' the minimum blocksize given.
457 Minimum alignment is typically 512b for using direct IO,
458 though it usually depends on the hardware block size. This
459 option is mutually exclusive with using a random map for
460 files, so it will turn off that option.
461
d3aad8f2 462blocksize_range=irange
71bfa161
JA
463bsrange=irange Instead of giving a single block size, specify a range
464 and fio will mix the issued io block sizes. The issued
465 io unit will always be a multiple of the minimum value
f90eff5a
JA
466 given (also see bs_unaligned). Applies to both reads and
467 writes, however a second range can be given after a comma.
468 See bs=.
a00735e6 469
564ca972
JA
470bssplit=str Sometimes you want even finer grained control of the
471 block sizes issued, not just an even split between them.
472 This option allows you to weight various block sizes,
473 so that you are able to define a specific amount of
474 block sizes issued. The format for this option is:
475
476 bssplit=blocksize/percentage:blocksize/percentage
477
478 for as many block sizes as needed. So if you want to define
479 a workload that has 50% 64k blocks, 10% 4k blocks, and
480 40% 32k blocks, you would write:
481
482 bssplit=4k/10:64k/50:32k/40
483
484 Ordering does not matter. If the percentage is left blank,
485 fio will fill in the remaining values evenly. So a bssplit
486 option like this one:
487
488 bssplit=4k/50:1k/:32k/
489
490 would have 50% 4k ios, and 25% 1k and 32k ios. The percentages
491 always add up to 100, if bssplit is given a range that adds
492 up to more, it will error out.
493
720e84ad
JA
494 bssplit also supports giving separate splits to reads and
495 writes. The format is identical to what bs= accepts. You
496 have to separate the read and write parts with a comma. So
497 if you want a workload that has 50% 2k reads and 50% 4k reads,
498 while having 90% 4k writes and 10% 8k writes, you would
499 specify:
500
501 bssplit=2k/50:4k/50,4k/90,8k/10
502
d3aad8f2 503blocksize_unaligned
690adba3
JA
504bs_unaligned If this option is given, any byte size value within bsrange
505 may be used as a block range. This typically wont work with
506 direct IO, as that normally requires sector alignment.
71bfa161 507
6aca9b3d
JA
508bs_is_seq_rand If this option is set, fio will use the normal read,write
509 blocksize settings as sequential,random instead. Any random
510 read or write will use the WRITE blocksize settings, and any
511 sequential read or write will use the READ blocksize setting.
512
e9459e5a
JA
513zero_buffers If this option is given, fio will init the IO buffers to
514 all zeroes. The default is to fill them with random data.
515
5973cafb
JA
516refill_buffers If this option is given, fio will refill the IO buffers
517 on every submit. The default is to only fill it at init
518 time and reuse that data. Only makes sense if zero_buffers
41ccd845
JA
519 isn't specified, naturally. If data verification is enabled,
520 refill_buffers is also automatically enabled.
5973cafb 521
fd68418e
JA
522scramble_buffers=bool If refill_buffers is too costly and the target is
523 using data deduplication, then setting this option will
524 slightly modify the IO buffer contents to defeat normal
525 de-dupe attempts. This is not enough to defeat more clever
526 block compression attempts, but it will stop naive dedupe of
527 blocks. Default: true.
528
c5751c62
JA
529buffer_compress_percentage=int If this is set, then fio will attempt to
530 provide IO buffer content (on WRITEs) that compress to
531 the specified level. Fio does this by providing a mix of
532 random data and zeroes. Note that this is per block size
533 unit, for file/disk wide compression level that matches
534 this setting, you'll also want to set refill_buffers.
535
536buffer_compress_chunk=int See buffer_compress_percentage. This
537 setting allows fio to manage how big the ranges of random
538 data and zeroed data is. Without this set, fio will
539 provide buffer_compress_percentage of blocksize random
540 data, followed by the remaining zeroed. With this set
541 to some chunk size smaller than the block size, fio can
542 alternate random and zeroed data throughout the IO
543 buffer.
544
71bfa161
JA
545nrfiles=int Number of files to use for this job. Defaults to 1.
546
390b1537
JA
547openfiles=int Number of files to keep open at the same time. Defaults to
548 the same as nrfiles, can be set smaller to limit the number
549 simultaneous opens.
550
5af1c6f3
JA
551file_service_type=str Defines how fio decides which file from a job to
552 service next. The following types are defined:
553
554 random Just choose a file at random.
555
556 roundrobin Round robin over open files. This
557 is the default.
558
a086c257
JA
559 sequential Finish one file before moving on to
560 the next. Multiple files can still be
561 open depending on 'openfiles'.
562
1907dbc6
JA
563 The string can have a number appended, indicating how
564 often to switch to a new file. So if option random:4 is
565 given, fio will switch to a new random file after 4 ios
566 have been issued.
567
71bfa161
JA
568ioengine=str Defines how the job issues io to the file. The following
569 types are defined:
570
571 sync Basic read(2) or write(2) io. lseek(2) is
572 used to position the io location.
573
a31041ea 574 psync Basic pread(2) or pwrite(2) io.
575
e05af9e5 576 vsync Basic readv(2) or writev(2) IO.
1d2af02a 577
a46c5e01
JA
578 psyncv Basic preadv(2) or pwritev(2) IO.
579
15d182aa
JA
580 libaio Linux native asynchronous io. Note that Linux
581 may only support queued behaviour with
582 non-buffered IO (set direct=1 or buffered=0).
de890a1e 583 This engine defines engine specific options.
71bfa161
JA
584
585 posixaio glibc posix asynchronous io.
586
417f0068
JA
587 solarisaio Solaris native asynchronous io.
588
03e20d68
BC
589 windowsaio Windows native asynchronous io.
590
71bfa161
JA
591 mmap File is memory mapped and data copied
592 to/from using memcpy(3).
593
594 splice splice(2) is used to transfer the data and
595 vmsplice(2) to transfer data from user
596 space to the kernel.
597
d0ff85df
JA
598 syslet-rw Use the syslet system calls to make
599 regular read/write async.
600
71bfa161 601 sg SCSI generic sg v3 io. May either be
6c219763 602 synchronous using the SG_IO ioctl, or if
71bfa161
JA
603 the target is an sg character device
604 we use read(2) and write(2) for asynchronous
605 io.
606
a94ea28b
JA
607 null Doesn't transfer any data, just pretends
608 to. This is mainly used to exercise fio
609 itself and for debugging/testing purposes.
610
ed92ac0c 611 net Transfer over the network to given host:port.
de890a1e
SL
612 Depending on the protocol used, the hostname,
613 port, listen and filename options are used to
614 specify what sort of connection to make, while
615 the protocol option determines which protocol
616 will be used.
617 This engine defines engine specific options.
ed92ac0c 618
9cce02e8
JA
619 netsplice Like net, but uses splice/vmsplice to
620 map data and send/receive.
de890a1e 621 This engine defines engine specific options.
9cce02e8 622
53aec0a4 623 cpuio Doesn't transfer any data, but burns CPU
ba0fbe10
JA
624 cycles according to the cpuload= and
625 cpucycle= options. Setting cpuload=85
626 will cause that job to do nothing but burn
36ecec83
GP
627 85% of the CPU. In case of SMP machines,
628 use numjobs=<no_of_cpu> to get desired CPU
629 usage, as the cpuload only loads a single
630 CPU at the desired rate.
ba0fbe10 631
e9a1806f
JA
632 guasi The GUASI IO engine is the Generic Userspace
633 Asyncronous Syscall Interface approach
634 to async IO. See
635
636 http://www.xmailserver.org/guasi-lib.html
637
638 for more info on GUASI.
639
21b8aee8 640 rdma The RDMA I/O engine supports both RDMA
eb52fa3f
BVA
641 memory semantics (RDMA_WRITE/RDMA_READ) and
642 channel semantics (Send/Recv) for the
643 InfiniBand, RoCE and iWARP protocols.
21b8aee8 644
d54fce84
DM
645 falloc IO engine that does regular fallocate to
646 simulate data transfer as fio ioengine.
647 DDIR_READ does fallocate(,mode = keep_size,)
0981fd71 648 DDIR_WRITE does fallocate(,mode = 0)
d54fce84
DM
649 DDIR_TRIM does fallocate(,mode = punch_hole)
650
651 e4defrag IO engine that does regular EXT4_IOC_MOVE_EXT
652 ioctls to simulate defragment activity in
653 request to DDIR_WRITE event
0981fd71 654
8a7bd877
JA
655 external Prefix to specify loading an external
656 IO engine object file. Append the engine
657 filename, eg ioengine=external:/tmp/foo.o
658 to load ioengine foo.o in /tmp.
659
71bfa161
JA
660iodepth=int This defines how many io units to keep in flight against
661 the file. The default is 1 for each file defined in this
662 job, can be overridden with a larger value for higher
ee72ca09
JA
663 concurrency. Note that increasing iodepth beyond 1 will not
664 affect synchronous ioengines (except for small degress when
9b836561 665 verify_async is in use). Even async engines may impose OS
ee72ca09
JA
666 restrictions causing the desired depth not to be achieved.
667 This may happen on Linux when using libaio and not setting
668 direct=1, since buffered IO is not async on that OS. Keep an
669 eye on the IO depth distribution in the fio output to verify
670 that the achieved depth is as expected. Default: 1.
71bfa161 671
4950421a 672iodepth_batch_submit=int
cb5ab512 673iodepth_batch=int This defines how many pieces of IO to submit at once.
89e820f6
JA
674 It defaults to 1 which means that we submit each IO
675 as soon as it is available, but can be raised to submit
676 bigger batches of IO at the time.
cb5ab512 677
4950421a
JA
678iodepth_batch_complete=int This defines how many pieces of IO to retrieve
679 at once. It defaults to 1 which means that we'll ask
680 for a minimum of 1 IO in the retrieval process from
681 the kernel. The IO retrieval will go on until we
682 hit the limit set by iodepth_low. If this variable is
683 set to 0, then fio will always check for completed
684 events before queuing more IO. This helps reduce
685 IO latency, at the cost of more retrieval system calls.
686
e916b390
JA
687iodepth_low=int The low water mark indicating when to start filling
688 the queue again. Defaults to the same as iodepth, meaning
689 that fio will attempt to keep the queue full at all times.
690 If iodepth is set to eg 16 and iodepth_low is set to 4, then
691 after fio has filled the queue of 16 requests, it will let
692 the depth drain down to 4 before starting to fill it again.
693
71bfa161 694direct=bool If value is true, use non-buffered io. This is usually
9b836561 695 O_DIRECT. Note that ZFS on Solaris doesn't support direct io.
93bcfd20 696 On Windows the synchronous ioengines don't support direct io.
76a43db4 697
d01612f3
CM
698atomic=bool If value is true, attempt to use atomic direct IO. Atomic
699 writes are guaranteed to be stable once acknowledged by
700 the operating system. Only Linux supports O_ATOMIC right
701 now.
702
76a43db4
JA
703buffered=bool If value is true, use buffered io. This is the opposite
704 of the 'direct' option. Defaults to true.
71bfa161 705
f7fa2653 706offset=int Start io at the given offset in the file. The data before
71bfa161
JA
707 the given offset will not be touched. This effectively
708 caps the file size at real_size - offset.
709
214ac7e0
DE
710offset_increment=int If this is provided, then the real offset becomes
711 the offset + offset_increment * thread_number, where the
712 thread number is a counter that starts at 0 and is incremented
713 for each job. This option is useful if there are several jobs
714 which are intended to operate on a file in parallel in disjoint
715 segments, with even spacing between the starting points.
716
ddf24e42
JA
717number_ios=int Fio will normally perform IOs until it has exhausted the size
718 of the region set by size=, or if it exhaust the allocated
719 time (or hits an error condition). With this setting, the
720 range/size can be set independently of the number of IOs to
721 perform. When fio reaches this number, it will exit normally
722 and report status.
723
71bfa161
JA
724fsync=int If writing to a file, issue a sync of the dirty data
725 for every number of blocks given. For example, if you give
726 32 as a parameter, fio will sync the file for every 32
727 writes issued. If fio is using non-buffered io, we may
728 not sync the file. The exception is the sg io engine, which
6c219763 729 synchronizes the disk cache anyway.
71bfa161 730
e76b1da4 731fdatasync=int Like fsync= but uses fdatasync() to only sync data and not
5f9099ea 732 metadata blocks.
93bcfd20 733 In FreeBSD and Windows there is no fdatasync(), this falls back to
e72fa4d4 734 using fsync()
5f9099ea 735
e76b1da4
JA
736sync_file_range=str:val Use sync_file_range() for every 'val' number of
737 write operations. Fio will track range of writes that
738 have happened since the last sync_file_range() call. 'str'
739 can currently be one or more of:
740
741 wait_before SYNC_FILE_RANGE_WAIT_BEFORE
742 write SYNC_FILE_RANGE_WRITE
743 wait_after SYNC_FILE_RANGE_WAIT_AFTER
744
745 So if you do sync_file_range=wait_before,write:8, fio would
746 use SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE for
747 every 8 writes. Also see the sync_file_range(2) man page.
748 This option is Linux specific.
749
5036fc1e
JA
750overwrite=bool If true, writes to a file will always overwrite existing
751 data. If the file doesn't already exist, it will be
752 created before the write phase begins. If the file exists
753 and is large enough for the specified write phase, nothing
754 will be done.
71bfa161 755
dbd11ead 756end_fsync=bool If true, fsync file contents when a write stage has completed.
71bfa161 757
ebb1415f
JA
758fsync_on_close=bool If true, fio will fsync() a dirty file on close.
759 This differs from end_fsync in that it will happen on every
760 file close, not just at the end of the job.
761
71bfa161
JA
762rwmixread=int How large a percentage of the mix should be reads.
763
764rwmixwrite=int How large a percentage of the mix should be writes. If both
765 rwmixread and rwmixwrite is given and the values do not add
766 up to 100%, the latter of the two will be used to override
c35dd7a6
JA
767 the first. This may interfere with a given rate setting,
768 if fio is asked to limit reads or writes to a certain rate.
769 If that is the case, then the distribution may be skewed.
71bfa161 770
92d42d69
JA
771random_distribution=str:float By default, fio will use a completely uniform
772 random distribution when asked to perform random IO. Sometimes
773 it is useful to skew the distribution in specific ways,
774 ensuring that some parts of the data is more hot than others.
775 fio includes the following distribution models:
776
777 random Uniform random distribution
778 zipf Zipf distribution
779 pareto Pareto distribution
780
781 When using a zipf or pareto distribution, an input value
782 is also needed to define the access pattern. For zipf, this
783 is the zipf theta. For pareto, it's the pareto power. Fio
784 includes a test program, genzipf, that can be used visualize
785 what the given input values will yield in terms of hit rates.
786 If you wanted to use zipf with a theta of 1.2, you would use
787 random_distribution=zipf:1.2 as the option. If a non-uniform
788 model is used, fio will disable use of the random map.
789
211c9b89
JA
790percentage_random=int For a random workload, set how big a percentage should
791 be random. This defaults to 100%, in which case the workload
792 is fully random. It can be set from anywhere from 0 to 100.
793 Setting it to 0 would make the workload fully sequential. Any
794 setting in between will result in a random mix of sequential
d9472271
JA
795 and random IO, at the given percentages. It is possible to
796 set different values for reads, writes, and trim. To do so,
797 simply use a comma separated list. See blocksize.
211c9b89 798
bb8895e0
JA
799norandommap Normally fio will cover every block of the file when doing
800 random IO. If this option is given, fio will just get a
801 new random offset without looking at past io history. This
802 means that some blocks may not be read or written, and that
803 some blocks may be read/written more than once. This option
8347239a
JA
804 is mutually exclusive with verify= if and only if multiple
805 blocksizes (via bsrange=) are used, since fio only tracks
806 complete rewrites of blocks.
bb8895e0 807
0408c206
JA
808softrandommap=bool See norandommap. If fio runs with the random block map
809 enabled and it fails to allocate the map, if this option is
810 set it will continue without a random block map. As coverage
811 will not be as complete as with random maps, this option is
2b386d25
JA
812 disabled by default.
813
e8b1961d
JA
814random_generator=str Fio supports the following engines for generating
815 IO offsets for random IO:
816
817 tausworthe Strong 2^88 cycle random number generator
818 lfsr Linear feedback shift register generator
819
820 Tausworthe is a strong random number generator, but it
821 requires tracking on the side if we want to ensure that
822 blocks are only read or written once. LFSR guarantees
823 that we never generate the same offset twice, and it's
824 also less computationally expensive. It's not a true
825 random generator, however, though for IO purposes it's
826 typically good enough. LFSR only works with single
827 block sizes, not with workloads that use multiple block
828 sizes. If used with such a workload, fio may read or write
829 some blocks multiple times.
43f09da1 830
71bfa161
JA
831nice=int Run the job with the given nice value. See man nice(2).
832
833prio=int Set the io priority value of this job. Linux limits us to
834 a positive value between 0 and 7, with 0 being the highest.
835 See man ionice(1).
836
837prioclass=int Set the io priority class. See man ionice(1).
838
839thinktime=int Stall the job x microseconds after an io has completed before
840 issuing the next. May be used to simulate processing being
48097d5c
JA
841 done by an application. See thinktime_blocks and
842 thinktime_spin.
843
844thinktime_spin=int
845 Only valid if thinktime is set - pretend to spend CPU time
846 doing something with the data received, before falling back
847 to sleeping for the rest of the period specified by
848 thinktime.
9c1f7434 849
4d01ece6 850thinktime_blocks=int
9c1f7434
JA
851 Only valid if thinktime is set - control how many blocks
852 to issue, before waiting 'thinktime' usecs. If not set,
853 defaults to 1 which will make fio wait 'thinktime' usecs
4d01ece6
JA
854 after every block. This effectively makes any queue depth
855 setting redundant, since no more than 1 IO will be queued
856 before we have to complete it and do our thinktime. In
857 other words, this setting effectively caps the queue depth
858 if the latter is larger.
71bfa161 859
581e7141 860rate=int Cap the bandwidth used by this job. The number is in bytes/sec,
b09da8fa 861 the normal suffix rules apply. You can use rate=500k to limit
581e7141
JA
862 reads and writes to 500k each, or you can specify read and
863 writes separately. Using rate=1m,500k would limit reads to
864 1MB/sec and writes to 500KB/sec. Capping only reads or
865 writes can be done with rate=,500k or rate=500k,. The former
866 will only limit writes (to 500KB/sec), the latter will only
867 limit reads.
71bfa161
JA
868
869ratemin=int Tell fio to do whatever it can to maintain at least this
4e991c23 870 bandwidth. Failing to meet this requirement, will cause
581e7141
JA
871 the job to exit. The same format as rate is used for
872 read vs write separation.
4e991c23
JA
873
874rate_iops=int Cap the bandwidth to this number of IOPS. Basically the same
875 as rate, just specified independently of bandwidth. If the
876 job is given a block size range instead of a fixed value,
581e7141 877 the smallest block size is used as the metric. The same format
de8f6de9 878 as rate is used for read vs write separation.
4e991c23
JA
879
880rate_iops_min=int If fio doesn't meet this rate of IO, it will cause
581e7141 881 the job to exit. The same format as rate is used for read vs
de8f6de9 882 write separation.
71bfa161 883
15501535
JA
884max_latency=int If set, fio will exit the job if it exceeds this maximum
885 latency. It will exit with an ETIME error.
886
71bfa161 887ratecycle=int Average bandwidth for 'rate' and 'ratemin' over this number
6c219763 888 of milliseconds.
71bfa161
JA
889
890cpumask=int Set the CPU affinity of this job. The parameter given is a
a08bc17f
JA
891 bitmask of allowed CPU's the job may run on. So if you want
892 the allowed CPUs to be 1 and 5, you would pass the decimal
893 value of (1 << 1 | 1 << 5), or 34. See man
7dbb6eba 894 sched_setaffinity(2). This may not work on all supported
b0ea08ce
JA
895 operating systems or kernel versions. This option doesn't
896 work well for a higher CPU count than what you can store in
897 an integer mask, so it can only control cpus 1-32. For
898 boxes with larger CPU counts, use cpus_allowed.
71bfa161 899
d2e268b0
JA
900cpus_allowed=str Controls the same options as cpumask, but it allows a text
901 setting of the permitted CPUs instead. So to use CPUs 1 and
62a7273d
JA
902 5, you would specify cpus_allowed=1,5. This options also
903 allows a range of CPUs. Say you wanted a binding to CPUs
904 1, 5, and 8-15, you would set cpus_allowed=1,5,8-15.
d2e268b0 905
d0b937ed
YR
906numa_cpu_nodes=str Set this job running on spcified NUMA nodes' CPUs. The
907 arguments allow comma delimited list of cpu numbers,
908 A-B ranges, or 'all'. Note, to enable numa options support,
67bf9823 909 fio must be built on a system with libnuma-dev(el) installed.
d0b937ed
YR
910
911numa_mem_policy=str Set this job's memory policy and corresponding NUMA
912 nodes. Format of the argements:
913 <mode>[:<nodelist>]
914 `mode' is one of the following memory policy:
915 default, prefer, bind, interleave, local
916 For `default' and `local' memory policy, no node is
917 needed to be specified.
918 For `prefer', only one node is allowed.
919 For `bind' and `interleave', it allow comma delimited
920 list of numbers, A-B ranges, or 'all'.
921
e417fd66 922startdelay=time Start this job the specified number of seconds after fio
71bfa161
JA
923 has started. Only useful if the job file contains several
924 jobs, and you want to delay starting some jobs to a certain
925 time.
926
e417fd66 927runtime=time Tell fio to terminate processing after the specified number
71bfa161
JA
928 of seconds. It can be quite hard to determine for how long
929 a specified job will run, so this parameter is handy to
930 cap the total runtime to a given time.
931
cf4464ca 932time_based If set, fio will run for the duration of the runtime
bf9a3edb 933 specified even if the file(s) are completely read or
cf4464ca
JA
934 written. It will simply loop over the same workload
935 as many times as the runtime allows.
936
e417fd66 937ramp_time=time If set, fio will run the specified workload for this amount
721938ae
JA
938 of time before logging any performance numbers. Useful for
939 letting performance settle before logging results, thus
b29ee5b3
JA
940 minimizing the runtime required for stable results. Note
941 that the ramp_time is considered lead in time for a job,
942 thus it will increase the total runtime if a special timeout
943 or runtime is specified.
721938ae 944
71bfa161
JA
945invalidate=bool Invalidate the buffer/page cache parts for this file prior
946 to starting io. Defaults to true.
947
948sync=bool Use sync io for buffered writes. For the majority of the
949 io engines, this means using O_SYNC.
950
d3aad8f2 951iomem=str
71bfa161
JA
952mem=str Fio can use various types of memory as the io unit buffer.
953 The allowed values are:
954
955 malloc Use memory from malloc(3) as the buffers.
956
957 shm Use shared memory as the buffers. Allocated
958 through shmget(2).
959
74b025b0
JA
960 shmhuge Same as shm, but use huge pages as backing.
961
313cb206
JA
962 mmap Use mmap to allocate buffers. May either be
963 anonymous memory, or can be file backed if
964 a filename is given after the option. The
965 format is mem=mmap:/path/to/file.
71bfa161 966
d0bdaf49
JA
967 mmaphuge Use a memory mapped huge file as the buffer
968 backing. Append filename after mmaphuge, ala
969 mem=mmaphuge:/hugetlbfs/file
970
71bfa161 971 The area allocated is a function of the maximum allowed
5394ae5f
JA
972 bs size for the job, multiplied by the io depth given. Note
973 that for shmhuge and mmaphuge to work, the system must have
974 free huge pages allocated. This can normally be checked
975 and set by reading/writing /proc/sys/vm/nr_hugepages on a
b22989b9 976 Linux system. Fio assumes a huge page is 4MB in size. So
5394ae5f
JA
977 to calculate the number of huge pages you need for a given
978 job file, add up the io depth of all jobs (normally one unless
979 iodepth= is used) and multiply by the maximum bs set. Then
980 divide that number by the huge page size. You can see the
981 size of the huge pages in /proc/meminfo. If no huge pages
982 are allocated by having a non-zero number in nr_hugepages,
56bb17f2 983 using mmaphuge or shmhuge will fail. Also see hugepage-size.
5394ae5f
JA
984
985 mmaphuge also needs to have hugetlbfs mounted and the file
986 location should point there. So if it's mounted in /huge,
987 you would use mem=mmaphuge:/huge/somefile.
71bfa161 988
d529ee19
JA
989iomem_align=int This indiciates the memory alignment of the IO memory buffers.
990 Note that the given alignment is applied to the first IO unit
991 buffer, if using iodepth the alignment of the following buffers
992 are given by the bs used. In other words, if using a bs that is
993 a multiple of the page sized in the system, all buffers will
994 be aligned to this value. If using a bs that is not page
995 aligned, the alignment of subsequent IO memory buffers is the
996 sum of the iomem_align and bs used.
997
f7fa2653 998hugepage-size=int
56bb17f2 999 Defines the size of a huge page. Must at least be equal
b22989b9 1000 to the system setting, see /proc/meminfo. Defaults to 4MB.
c51074e7
JA
1001 Should probably always be a multiple of megabytes, so using
1002 hugepage-size=Xm is the preferred way to set this to avoid
1003 setting a non-pow-2 bad value.
56bb17f2 1004
71bfa161
JA
1005exitall When one job finishes, terminate the rest. The default is
1006 to wait for each job to finish, sometimes that is not the
1007 desired action.
1008
1009bwavgtime=int Average the calculated bandwidth over the given time. Value
6c219763 1010 is specified in milliseconds.
71bfa161 1011
c8eeb9df
JA
1012iopsavgtime=int Average the calculated IOPS over the given time. Value
1013 is specified in milliseconds.
1014
71bfa161
JA
1015create_serialize=bool If true, serialize the file creating for the jobs.
1016 This may be handy to avoid interleaving of data
1017 files, which may greatly depend on the filesystem
1018 used and even the number of processors in the system.
1019
1020create_fsync=bool fsync the data file after creation. This is the
1021 default.
1022
814452bd
JA
1023create_on_open=bool Don't pre-setup the files for IO, just create open()
1024 when it's time to do IO to that file.
1025
25460cf6
JA
1026create_only=bool If true, fio will only run the setup phase of the job.
1027 If files need to be laid out or updated on disk, only
1028 that will be done. The actual job contents are not
1029 executed.
1030
afad68f7 1031pre_read=bool If this is given, files will be pre-read into memory before
34f1c044
JA
1032 starting the given IO operation. This will also clear
1033 the 'invalidate' flag, since it is pointless to pre-read
9c0d2241
JA
1034 and then drop the cache. This will only work for IO engines
1035 that are seekable, since they allow you to read the same data
1036 multiple times. Thus it will not work on eg network or splice
1037 IO.
afad68f7 1038
e545a6ce 1039unlink=bool Unlink the job files when done. Not the default, as repeated
bf9a3edb
JA
1040 runs of that job would then waste time recreating the file
1041 set again and again.
71bfa161
JA
1042
1043loops=int Run the specified number of iterations of this job. Used
1044 to repeat the same workload a given number of times. Defaults
1045 to 1.
1046
68e1f29a 1047do_verify=bool Run the verify phase after a write phase. Only makes sense if
e84c73a8
SL
1048 verify is set. Defaults to 1.
1049
71bfa161
JA
1050verify=str If writing to a file, fio can verify the file contents
1051 after each iteration of the job. The allowed values are:
1052
1053 md5 Use an md5 sum of the data area and store
1054 it in the header of each block.
1055
17dc34df
JA
1056 crc64 Use an experimental crc64 sum of the data
1057 area and store it in the header of each
1058 block.
1059
bac39e0e
JA
1060 crc32c Use a crc32c sum of the data area and store
1061 it in the header of each block.
1062
3845591f 1063 crc32c-intel Use hardware assisted crc32c calcuation
0539d758
JA
1064 provided on SSE4.2 enabled processors. Falls
1065 back to regular software crc32c, if not
1066 supported by the system.
3845591f 1067
71bfa161
JA
1068 crc32 Use a crc32 sum of the data area and store
1069 it in the header of each block.
1070
969f7ed3
JA
1071 crc16 Use a crc16 sum of the data area and store
1072 it in the header of each block.
1073
17dc34df
JA
1074 crc7 Use a crc7 sum of the data area and store
1075 it in the header of each block.
1076
cd14cc10
JA
1077 sha512 Use sha512 as the checksum function.
1078
1079 sha256 Use sha256 as the checksum function.
1080
7c353ceb
JA
1081 sha1 Use optimized sha1 as the checksum function.
1082
7437ee87
SL
1083 meta Write extra information about each io
1084 (timestamp, block number etc.). The block
996093bb 1085 number is verified. See also verify_pattern.
7437ee87 1086
36690c9b
JA
1087 null Only pretend to verify. Useful for testing
1088 internals with ioengine=null, not for much
1089 else.
1090
6c219763 1091 This option can be used for repeated burn-in tests of a
71bfa161 1092 system to make sure that the written data is also
b892dc08
JA
1093 correctly read back. If the data direction given is
1094 a read or random read, fio will assume that it should
1095 verify a previously written file. If the data direction
1096 includes any form of write, the verify will be of the
1097 newly written data.
71bfa161 1098
160b966d
JA
1099verifysort=bool If set, fio will sort written verify blocks when it deems
1100 it faster to read them back in a sorted manner. This is
1101 often the case when overwriting an existing file, since
1102 the blocks are already laid out in the file system. You
1103 can ignore this option unless doing huge amounts of really
1104 fast IO where the red-black tree sorting CPU time becomes
1105 significant.
3f9f4e26 1106
f7fa2653 1107verify_offset=int Swap the verification header with data somewhere else
546a9142
SL
1108 in the block before writing. Its swapped back before
1109 verifying.
1110
f7fa2653 1111verify_interval=int Write the verification header at a finer granularity
3f9f4e26
SL
1112 than the blocksize. It will be written for chunks the
1113 size of header_interval. blocksize should divide this
1114 evenly.
90059d65 1115
0e92f873 1116verify_pattern=str If set, fio will fill the io buffers with this
e28218f3
SL
1117 pattern. Fio defaults to filling with totally random
1118 bytes, but sometimes it's interesting to fill with a known
1119 pattern for io verification purposes. Depending on the
1120 width of the pattern, fio will fill 1/2/3/4 bytes of the
0e92f873
RR
1121 buffer at the time(it can be either a decimal or a hex number).
1122 The verify_pattern if larger than a 32-bit quantity has to
996093bb
JA
1123 be a hex number that starts with either "0x" or "0X". Use
1124 with verify=meta.
e28218f3 1125
68e1f29a 1126verify_fatal=bool Normally fio will keep checking the entire contents
a12a3b4d
JA
1127 before quitting on a block verification failure. If this
1128 option is set, fio will exit the job on the first observed
1129 failure.
e8462bd8 1130
b463e936
JA
1131verify_dump=bool If set, dump the contents of both the original data
1132 block and the data block we read off disk to files. This
1133 allows later analysis to inspect just what kind of data
ef71e317 1134 corruption occurred. Off by default.
b463e936 1135
e8462bd8
JA
1136verify_async=int Fio will normally verify IO inline from the submitting
1137 thread. This option takes an integer describing how many
1138 async offload threads to create for IO verification instead,
1139 causing fio to offload the duty of verifying IO contents
c85c324c
JA
1140 to one or more separate threads. If using this offload
1141 option, even sync IO engines can benefit from using an
1142 iodepth setting higher than 1, as it allows them to have
1143 IO in flight while verifies are running.
e8462bd8
JA
1144
1145verify_async_cpus=str Tell fio to set the given CPU affinity on the
1146 async IO verification threads. See cpus_allowed for the
1147 format used.
6f87418f
JA
1148
1149verify_backlog=int Fio will normally verify the written contents of a
1150 job that utilizes verify once that job has completed. In
1151 other words, everything is written then everything is read
1152 back and verified. You may want to verify continually
1153 instead for a variety of reasons. Fio stores the meta data
1154 associated with an IO block in memory, so for large
1155 verify workloads, quite a bit of memory would be used up
1156 holding this meta data. If this option is enabled, fio
f42195a3
JA
1157 will write only N blocks before verifying these blocks.
1158
6f87418f
JA
1159verify_backlog_batch=int Control how many blocks fio will verify
1160 if verify_backlog is set. If not set, will default to
1161 the value of verify_backlog (meaning the entire queue
f42195a3
JA
1162 is read back and verified). If verify_backlog_batch is
1163 less than verify_backlog then not all blocks will be verified,
1164 if verify_backlog_batch is larger than verify_backlog, some
1165 blocks will be verified more than once.
66c098b8 1166
d392365e 1167stonewall
de8f6de9 1168wait_for_previous Wait for preceding jobs in the job file to exit, before
71bfa161 1169 starting this one. Can be used to insert serialization
b3d62a75
JA
1170 points in the job file. A stone wall also implies starting
1171 a new reporting group.
1172
abcab6af 1173new_group Start a new reporting group. See: group_reporting.
71bfa161
JA
1174
1175numjobs=int Create the specified number of clones of this job. May be
1176 used to setup a larger number of threads/processes doing
abcab6af
AV
1177 the same thing. Each thread is reported separately; to see
1178 statistics for all clones as a whole, use group_reporting in
1179 conjunction with new_group.
1180
1181group_reporting It may sometimes be interesting to display statistics for
04b2f799
JA
1182 groups of jobs as a whole instead of for each individual job.
1183 This is especially true if 'numjobs' is used; looking at
1184 individual thread/process output quickly becomes unwieldy.
1185 To see the final report per-group instead of per-job, use
1186 'group_reporting'. Jobs in a file will be part of the same
1187 reporting group, unless if separated by a stonewall, or by
1188 using 'new_group'.
71bfa161
JA
1189
1190thread fio defaults to forking jobs, however if this option is
1191 given, fio will use pthread_create(3) to create threads
1192 instead.
1193
f7fa2653 1194zonesize=int Divide a file into zones of the specified size. See zoneskip.
71bfa161 1195
f7fa2653 1196zoneskip=int Skip the specified number of bytes when zonesize data has
71bfa161
JA
1197 been read. The two zone options can be used to only do
1198 io on zones of a file.
1199
076efc7c 1200write_iolog=str Write the issued io patterns to the specified file. See
5b42a488
SH
1201 read_iolog. Specify a separate file for each job, otherwise
1202 the iologs will be interspersed and the file may be corrupt.
71bfa161 1203
076efc7c 1204read_iolog=str Open an iolog with the specified file name and replay the
71bfa161 1205 io patterns it contains. This can be used to store a
6df8adaa
JA
1206 workload and replay it sometime later. The iolog given
1207 may also be a blktrace binary file, which allows fio
1208 to replay a workload captured by blktrace. See blktrace
1209 for how to capture such logging data. For blktrace replay,
1210 the file needs to be turned into a blkparse binary data
ea3e51c3 1211 file first (blkparse <device> -o /dev/null -d file_for_fio.bin).
66c098b8 1212
64bbb865 1213replay_no_stall=int When replaying I/O with read_iolog the default behavior
62776229
JA
1214 is to attempt to respect the time stamps within the log and
1215 replay them with the appropriate delay between IOPS. By
1216 setting this variable fio will not respect the timestamps and
1217 attempt to replay them as fast as possible while still
1218 respecting ordering. The result is the same I/O pattern to a
1219 given device, but different timings.
71bfa161 1220
d1c46c04
DN
1221replay_redirect=str While replaying I/O patterns using read_iolog the
1222 default behavior is to replay the IOPS onto the major/minor
1223 device that each IOP was recorded from. This is sometimes
de8f6de9 1224 undesirable because on a different machine those major/minor
d1c46c04
DN
1225 numbers can map to a different device. Changing hardware on
1226 the same system can also result in a different major/minor
1227 mapping. Replay_redirect causes all IOPS to be replayed onto
1228 the single specified device regardless of the device it was
1229 recorded from. i.e. replay_redirect=/dev/sdc would cause all
1230 IO in the blktrace to be replayed onto /dev/sdc. This means
1231 multiple devices will be replayed onto a single, if the trace
1232 contains multiple devices. If you want multiple devices to be
1233 replayed concurrently to multiple redirected devices you must
1234 blkparse your trace into separate traces and replay them with
1235 independent fio invocations. Unfortuantely this also breaks
1236 the strict time ordering between multiple device accesses.
1237
e3cedca7 1238write_bw_log=str If given, write a bandwidth log of the jobs in this job
71bfa161 1239 file. Can be used to store data of the bandwidth of the
e0da9bc2
JA
1240 jobs in their lifetime. The included fio_generate_plots
1241 script uses gnuplot to turn these text files into nice
ddb754db
LAG
1242 graphs. See write_lat_log for behaviour of given
1243 filename. For this option, the suffix is _bw.log.
71bfa161 1244
e3cedca7 1245write_lat_log=str Same as write_bw_log, except that this option stores io
02af0988
JA
1246 submission, completion, and total latencies instead. If no
1247 filename is given with this option, the default filename of
1248 "jobname_type.log" is used. Even if the filename is given,
1249 fio will still append the type of log. So if one specifies
e3cedca7
JA
1250
1251 write_lat_log=foo
1252
d5d94597 1253 The actual log names will be foo_slat.log, foo_clat.log,
02af0988
JA
1254 and foo_lat.log. This helps fio_generate_plot fine the logs
1255 automatically.
71bfa161 1256
b8bc8cba
JA
1257write_iops_log=str Same as write_bw_log, but writes IOPS. If no filename is
1258 given with this option, the default filename of
1259 "jobname_type.log" is used. Even if the filename is given,
1260 fio will still append the type of log.
1261
1262log_avg_msec=int By default, fio will log an entry in the iops, latency,
1263 or bw log for every IO that completes. When writing to the
1264 disk log, that can quickly grow to a very large size. Setting
1265 this option makes fio average the each log entry over the
1266 specified period of time, reducing the resolution of the log.
1267 Defaults to 0.
1268
f7fa2653 1269lockmem=int Pin down the specified amount of memory with mlock(2). Can
71bfa161
JA
1270 potentially be used instead of removing memory or booting
1271 with less memory to simulate a smaller amount of memory.
81c6b6cd 1272 The amount specified is per worker.
71bfa161
JA
1273
1274exec_prerun=str Before running this job, issue the command specified
74c8c488
JA
1275 through system(3). Output is redirected in a file called
1276 jobname.prerun.txt.
71bfa161
JA
1277
1278exec_postrun=str After the job completes, issue the command specified
74c8c488
JA
1279 though system(3). Output is redirected in a file called
1280 jobname.postrun.txt.
71bfa161
JA
1281
1282ioscheduler=str Attempt to switch the device hosting the file to the specified
1283 io scheduler before running.
1284
0a839f30
JA
1285disk_util=bool Generate disk utilization statistics, if the platform
1286 supports it. Defaults to on.
1287
02af0988 1288disable_lat=bool Disable measurements of total latency numbers. Useful
9520ebb9
JA
1289 only for cutting back the number of calls to gettimeofday,
1290 as that does impact performance at really high IOPS rates.
1291 Note that to really get rid of a large amount of these
1292 calls, this option must be used with disable_slat and
1293 disable_bw as well.
1294
02af0988
JA
1295disable_clat=bool Disable measurements of completion latency numbers. See
1296 disable_lat.
1297
9520ebb9 1298disable_slat=bool Disable measurements of submission latency numbers. See
02af0988 1299 disable_slat.
9520ebb9
JA
1300
1301disable_bw=bool Disable measurements of throughput/bandwidth numbers. See
02af0988 1302 disable_lat.
9520ebb9 1303
83349190
YH
1304clat_percentiles=bool Enable the reporting of percentiles of
1305 completion latencies.
1306
1307percentile_list=float_list Overwrite the default list of percentiles
1308 for completion latencies. Each number is a floating
1309 number in the range (0,100], and the maximum length of
1310 the list is 20. Use ':' to separate the numbers, and
1311 list the numbers in ascending order. For example,
1312 --percentile_list=99.5:99.9 will cause fio to report
1313 the values of completion latency below which 99.5% and
1314 99.9% of the observed latencies fell, respectively.
1315
23893646
JA
1316clocksource=str Use the given clocksource as the base of timing. The
1317 supported options are:
1318
1319 gettimeofday gettimeofday(2)
1320
1321 clock_gettime clock_gettime(2)
1322
1323 cpu Internal CPU clock source
1324
1325 cpu is the preferred clocksource if it is reliable, as it
1326 is very fast (and fio is heavy on time calls). Fio will
1327 automatically use this clocksource if it's supported and
1328 considered reliable on the system it is running on, unless
1329 another clocksource is specifically set. For x86/x86-64 CPUs,
1330 this means supporting TSC Invariant.
1331
993bf48b
JA
1332gtod_reduce=bool Enable all of the gettimeofday() reducing options
1333 (disable_clat, disable_slat, disable_bw) plus reduce
1334 precision of the timeout somewhat to really shrink
1335 the gettimeofday() call count. With this option enabled,
1336 we only do about 0.4% of the gtod() calls we would have
1337 done if all time keeping was enabled.
1338
be4ecfdf
JA
1339gtod_cpu=int Sometimes it's cheaper to dedicate a single thread of
1340 execution to just getting the current time. Fio (and
1341 databases, for instance) are very intensive on gettimeofday()
1342 calls. With this option, you can set one CPU aside for
1343 doing nothing but logging current time to a shared memory
1344 location. Then the other threads/processes that run IO
1345 workloads need only copy that segment, instead of entering
1346 the kernel with a gettimeofday() call. The CPU set aside
1347 for doing these time calls will be excluded from other
1348 uses. Fio will manually clear it from the CPU mask of other
1349 jobs.
a696fa2a 1350
06842027 1351continue_on_error=str Normally fio will exit the job on the first observed
f2bba182
RR
1352 failure. If this option is set, fio will continue the job when
1353 there is a 'non-fatal error' (EIO or EILSEQ) until the runtime
1354 is exceeded or the I/O size specified is completed. If this
1355 option is used, there are two more stats that are appended,
1356 the total error count and the first error. The error field
1357 given in the stats is the first error that was hit during the
1358 run.
be4ecfdf 1359
06842027
SL
1360 The allowed values are:
1361
1362 none Exit on any IO or verify errors.
1363
1364 read Continue on read errors, exit on all others.
1365
1366 write Continue on write errors, exit on all others.
1367
1368 io Continue on any IO error, exit on all others.
1369
1370 verify Continue on verify errors, exit on all others.
1371
1372 all Continue on all errors.
1373
1374 0 Backward-compatible alias for 'none'.
1375
1376 1 Backward-compatible alias for 'all'.
1377
8b28bd41
DM
1378ignore_error=str Sometimes you want to ignore some errors during test
1379 in that case you can specify error list for each error type.
1380 ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST
1381 errors for given error type is separated with ':'. Error
1382 may be symbol ('ENOSPC', 'ENOMEM') or integer.
1383 Example:
1384 ignore_error=EAGAIN,ENOSPC:122
66c098b8
BC
1385 This option will ignore EAGAIN from READ, and ENOSPC and
1386 122(EDQUOT) from WRITE.
8b28bd41
DM
1387
1388error_dump=bool If set dump every error even if it is non fatal, true
1389 by default. If disabled only fatal error will be dumped
66c098b8 1390
6adb38a1
JA
1391cgroup=str Add job to this control group. If it doesn't exist, it will
1392 be created. The system must have a mounted cgroup blkio
1393 mount point for this to work. If your system doesn't have it
1394 mounted, you can do so with:
a696fa2a
JA
1395
1396 # mount -t cgroup -o blkio none /cgroup
1397
a696fa2a
JA
1398cgroup_weight=int Set the weight of the cgroup to this value. See
1399 the documentation that comes with the kernel, allowed values
1400 are in the range of 100..1000.
71bfa161 1401
7de87099
VG
1402cgroup_nodelete=bool Normally fio will delete the cgroups it has created after
1403 the job completion. To override this behavior and to leave
1404 cgroups around after the job completion, set cgroup_nodelete=1.
1405 This can be useful if one wants to inspect various cgroup
1406 files after job completion. Default: false
1407
e0b0d892
JA
1408uid=int Instead of running as the invoking user, set the user ID to
1409 this value before the thread/process does any work.
1410
1411gid=int Set group ID, see uid.
1412
9e684a49
DE
1413flow_id=int The ID of the flow. If not specified, it defaults to being a
1414 global flow. See flow.
1415
1416flow=int Weight in token-based flow control. If this value is used, then
1417 there is a 'flow counter' which is used to regulate the
1418 proportion of activity between two or more jobs. fio attempts
1419 to keep this flow counter near zero. The 'flow' parameter
1420 stands for how much should be added or subtracted to the flow
1421 counter on each iteration of the main I/O loop. That is, if
1422 one job has flow=8 and another job has flow=-1, then there
1423 will be a roughly 1:8 ratio in how much one runs vs the other.
1424
1425flow_watermark=int The maximum value that the absolute value of the flow
1426 counter is allowed to reach before the job must wait for a
1427 lower value of the counter.
1428
1429flow_sleep=int The period of time, in microseconds, to wait after the flow
1430 watermark has been exceeded before retrying operations
1431
de890a1e
SL
1432In addition, there are some parameters which are only valid when a specific
1433ioengine is in use. These are used identically to normal parameters, with the
1434caveat that when used on the command line, they must come after the ioengine
1435that defines them is selected.
1436
1437[libaio] userspace_reap Normally, with the libaio engine in use, fio will use
1438 the io_getevents system call to reap newly returned events.
1439 With this flag turned on, the AIO ring will be read directly
1440 from user-space to reap events. The reaping mode is only
1441 enabled when polling for a minimum of 0 events (eg when
1442 iodepth_batch_complete=0).
1443
0353050f
JA
1444[cpu] cpuload=int Attempt to use the specified percentage of CPU cycles.
1445
1446[cpu] cpuchunks=int Split the load into cycles of the given time. In
1447 microseconds.
1448
de890a1e
SL
1449[netsplice] hostname=str
1450[net] hostname=str The host name or IP address to use for TCP or UDP based IO.
1451 If the job is a TCP listener or UDP reader, the hostname is not
b511c9aa
SB
1452 used and must be omitted unless it is a valid UDP multicast
1453 address.
de890a1e
SL
1454
1455[netsplice] port=int
1456[net] port=int The TCP or UDP port to bind to or connect to.
1457
b93b6a2e
SB
1458[netsplice] interface=str
1459[net] interface=str The IP address of the network interface used to send or
1460 receive UDP multicast
1461
d3a623de
SB
1462[netsplice] ttl=int
1463[net] ttl=int Time-to-live value for outgoing UDP multicast packets.
1464 Default: 1
1465
1d360ffb
JA
1466[netsplice] nodelay=bool
1467[net] nodelay=bool Set TCP_NODELAY on TCP connections.
1468
de890a1e
SL
1469[netsplice] protocol=str
1470[netsplice] proto=str
1471[net] protocol=str
1472[net] proto=str The network protocol to use. Accepted values are:
1473
1474 tcp Transmission control protocol
f5cc3d0e 1475 udp User datagram protocol
de890a1e
SL
1476 unix UNIX domain socket
1477
1478 When the protocol is TCP or UDP, the port must also be given,
1479 as well as the hostname if the job is a TCP listener or UDP
1480 reader. For unix sockets, the normal filename option should be
1481 used and the port is invalid.
1482
1483[net] listen For TCP network connections, tell fio to listen for incoming
1484 connections rather than initiating an outgoing connection. The
1485 hostname must be omitted if this option is used.
b511c9aa 1486[net] pingpong Normaly a network writer will just continue writing data, and
7aeb1e94
JA
1487 a network reader will just consume packages. If pingpong=1
1488 is set, a writer will send its normal payload to the reader,
1489 then wait for the reader to send the same payload back. This
1490 allows fio to measure network latencies. The submission
1491 and completion latencies then measure local time spent
1492 sending or receiving, and the completion latency measures
1493 how long it took for the other end to receive and send back.
b511c9aa
SB
1494 For UDP multicast traffic pingpong=1 should only be set for a
1495 single reader when multiple readers are listening to the same
1496 address.
7aeb1e94 1497
d54fce84
DM
1498[e4defrag] donorname=str
1499 File will be used as a block donor(swap extents between files)
1500[e4defrag] inplace=int
66c098b8 1501 Configure donor file blocks allocation strategy
d54fce84
DM
1502 0(default): Preallocate donor's file on init
1503 1 : allocate space immidietly inside defragment event,
1504 and free right after event
1505
de890a1e
SL
1506
1507
71bfa161
JA
15086.0 Interpreting the output
1509---------------------------
1510
1511fio spits out a lot of output. While running, fio will display the
1512status of the jobs created. An example of that would be:
1513
73c8b082 1514Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
71bfa161
JA
1515
1516The characters inside the square brackets denote the current status of
1517each thread. The possible values (in typical life cycle order) are:
1518
1519Idle Run
1520---- ---
1521P Thread setup, but not started.
1522C Thread created.
9c6f6316 1523I Thread initialized, waiting or generating necessary data.
b0f65863 1524 p Thread running pre-reading file(s).
71bfa161
JA
1525 R Running, doing sequential reads.
1526 r Running, doing random reads.
1527 W Running, doing sequential writes.
1528 w Running, doing random writes.
1529 M Running, doing mixed sequential reads/writes.
1530 m Running, doing mixed random reads/writes.
1531 F Running, currently waiting for fsync()
fc6bd43c 1532 V Running, doing verification of written data.
71bfa161 1533E Thread exited, not reaped by main thread yet.
4f7e57a4
JA
1534_ Thread reaped, or
1535X Thread reaped, exited with an error.
a5e371a6 1536K Thread reaped, exited due to signal.
71bfa161
JA
1537
1538The other values are fairly self explanatory - number of threads
c9f60304
JA
1539currently running and doing io, rate of io since last check (read speed
1540listed first, then write speed), and the estimated completion percentage
1541and time for the running group. It's impossible to estimate runtime of
4f7e57a4
JA
1542the following groups (if any). Note that the string is displayed in order,
1543so it's possible to tell which of the jobs are currently doing what. The
1544first character is the first job defined in the job file, and so forth.
71bfa161
JA
1545
1546When fio is done (or interrupted by ctrl-c), it will show the data for
1547each thread, group of threads, and disks in that order. For each data
1548direction, the output looks like:
1549
1550Client1 (g=0): err= 0:
35649e58 1551 write: io= 32MB, bw= 666KB/s, iops=89 , runt= 50320msec
6104ddb6
JA
1552 slat (msec): min= 0, max= 136, avg= 0.03, stdev= 1.92
1553 clat (msec): min= 0, max= 631, avg=48.50, stdev=86.82
b22989b9 1554 bw (KB/s) : min= 0, max= 1196, per=51.00%, avg=664.02, stdev=681.68
e7823a94 1555 cpu : usr=1.49%, sys=0.25%, ctx=7969, majf=0, minf=17
71619dc2 1556 IO depths : 1=0.1%, 2=0.3%, 4=0.5%, 8=99.0%, 16=0.0%, 32=0.0%, >32=0.0%
838bc709
JA
1557 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
1558 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
30061b97 1559 issued r/w: total=0/32768, short=0/0
8abdce66
JA
1560 lat (msec): 2=1.6%, 4=0.0%, 10=3.2%, 20=12.8%, 50=38.4%, 100=24.8%,
1561 lat (msec): 250=15.2%, 500=0.0%, 750=0.0%, 1000=0.0%, >=2048=0.0%
71bfa161
JA
1562
1563The client number is printed, along with the group id and error of that
1564thread. Below is the io statistics, here for writes. In the order listed,
1565they denote:
1566
1567io= Number of megabytes io performed
1568bw= Average bandwidth rate
35649e58 1569iops= Average IOs performed per second
71bfa161 1570runt= The runtime of that thread
72fbda2a 1571 slat= Submission latency (avg being the average, stdev being the
71bfa161
JA
1572 standard deviation). This is the time it took to submit
1573 the io. For sync io, the slat is really the completion
8a35c71e 1574 latency, since queue/complete is one operation there. This
bf9a3edb 1575 value can be in milliseconds or microseconds, fio will choose
8a35c71e 1576 the most appropriate base and print that. In the example
0d237712
LAG
1577 above, milliseconds is the best scale. Note: in --minimal mode
1578 latencies are always expressed in microseconds.
71bfa161
JA
1579 clat= Completion latency. Same names as slat, this denotes the
1580 time from submission to completion of the io pieces. For
1581 sync io, clat will usually be equal (or very close) to 0,
1582 as the time from submit to complete is basically just
1583 CPU time (io has already been done, see slat explanation).
1584 bw= Bandwidth. Same names as the xlat stats, but also includes
1585 an approximate percentage of total aggregate bandwidth
1586 this thread received in this group. This last value is
1587 only really useful if the threads in this group are on the
1588 same disk, since they are then competing for disk access.
1589cpu= CPU usage. User and system time, along with the number
e7823a94
JA
1590 of context switches this thread went through, usage of
1591 system and user time, and finally the number of major
1592 and minor page faults.
71619dc2
JA
1593IO depths= The distribution of io depths over the job life time. The
1594 numbers are divided into powers of 2, so for example the
1595 16= entries includes depths up to that value but higher
1596 than the previous entry. In other words, it covers the
1597 range from 16 to 31.
838bc709
JA
1598IO submit= How many pieces of IO were submitting in a single submit
1599 call. Each entry denotes that amount and below, until
1600 the previous entry - eg, 8=100% mean that we submitted
1601 anywhere in between 5-8 ios per submit call.
1602IO complete= Like the above submit number, but for completions instead.
30061b97
JA
1603IO issued= The number of read/write requests issued, and how many
1604 of them were short.
ec118304
JA
1605IO latencies= The distribution of IO completion latencies. This is the
1606 time from when IO leaves fio and when it gets completed.
1607 The numbers follow the same pattern as the IO depths,
1608 meaning that 2=1.6% means that 1.6% of the IO completed
8abdce66
JA
1609 within 2 msecs, 20=12.8% means that 12.8% of the IO
1610 took more than 10 msecs, but less than (or equal to) 20 msecs.
71bfa161
JA
1611
1612After each client has been listed, the group statistics are printed. They
1613will look like this:
1614
1615Run status group 0 (all jobs):
b22989b9
JA
1616 READ: io=64MB, aggrb=22178, minb=11355, maxb=11814, mint=2840msec, maxt=2955msec
1617 WRITE: io=64MB, aggrb=1302, minb=666, maxb=669, mint=50093msec, maxt=50320msec
71bfa161
JA
1618
1619For each data direction, it prints:
1620
1621io= Number of megabytes io performed.
1622aggrb= Aggregate bandwidth of threads in this group.
1623minb= The minimum average bandwidth a thread saw.
1624maxb= The maximum average bandwidth a thread saw.
1625mint= The smallest runtime of the threads in that group.
1626maxt= The longest runtime of the threads in that group.
1627
1628And finally, the disk statistics are printed. They will look like this:
1629
1630Disk stats (read/write):
1631 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
1632
1633Each value is printed for both reads and writes, with reads first. The
1634numbers denote:
1635
1636ios= Number of ios performed by all groups.
1637merge= Number of merges io the io scheduler.
1638ticks= Number of ticks we kept the disk busy.
1639io_queue= Total time spent in the disk queue.
1640util= The disk utilization. A value of 100% means we kept the disk
1641 busy constantly, 50% would be a disk idling half of the time.
1642
8423bd11
JA
1643It is also possible to get fio to dump the current output while it is
1644running, without terminating the job. To do that, send fio the USR1 signal.
06464907
JA
1645You can also get regularly timed dumps by using the --status-interval
1646parameter, or by creating a file in /tmp named fio-dump-status. If fio
1647sees this file, it will unlink it and dump the current output status.
8423bd11 1648
71bfa161
JA
1649
16507.0 Terse output
1651----------------
1652
1653For scripted usage where you typically want to generate tables or graphs
6af019c9 1654of the results, fio can output the results in a semicolon separated format.
71bfa161
JA
1655The format is one long line of values, such as:
1656
562c2d2f
DN
16572;card0;0;0;7139336;121836;60004;1;10109;27.932460;116.933948;220;126861;3495.446807;1085.368601;226;126864;3523.635629;1089.012448;24063;99944;50.275485%;59818.274627;5540.657370;7155060;122104;60004;1;8338;29.086342;117.839068;388;128077;5032.488518;1234.785715;391;128085;5061.839412;1236.909129;23436;100928;50.287926%;59964.832030;5644.844189;14.595833%;19.394167%;123706;0;7313;0.1%;0.1%;0.1%;0.1%;0.1%;0.1%;100.0%;0.00%;0.00%;0.00%;0.00%;0.00%;0.00%;0.01%;0.02%;0.05%;0.16%;6.04%;40.40%;52.68%;0.64%;0.01%;0.00%;0.01%;0.00%;0.00%;0.00%;0.00%;0.00%
1658A description of this job goes here.
1659
1660The job description (if provided) follows on a second line.
71bfa161 1661
525c2bfa
JA
1662To enable terse output, use the --minimal command line option. The first
1663value is the version of the terse output format. If the output has to
1664be changed for some reason, this number will be incremented by 1 to
1665signify that change.
6820cb3b 1666
71bfa161
JA
1667Split up, the format is as follows:
1668
5e726d0a 1669 terse version, fio version, jobname, groupid, error
71bfa161 1670 READ status:
312b4af2 1671 Total IO (KB), bandwidth (KB/sec), IOPS, runtime (msec)
de196b82
JA
1672 Submission latency: min, max, mean, deviation (usec)
1673 Completion latency: min, max, mean, deviation (usec)
1db92cb6 1674 Completion latency percentiles: 20 fields (see below)
de196b82 1675 Total latency: min, max, mean, deviation (usec)
0d237712 1676 Bw (KB/s): min, max, aggregate percentage of total, mean, deviation
71bfa161 1677 WRITE status:
312b4af2 1678 Total IO (KB), bandwidth (KB/sec), IOPS, runtime (msec)
de196b82
JA
1679 Submission latency: min, max, mean, deviation (usec)
1680 Completion latency: min, max, mean, deviation (usec)
1db92cb6 1681 Completion latency percentiles: 20 fields (see below)
de196b82 1682 Total latency: min, max, mean, deviation (usec)
0d237712 1683 Bw (KB/s): min, max, aggregate percentage of total, mean, deviation
046ee302 1684 CPU usage: user, system, context switches, major faults, minor faults
2270890c 1685 IO depths: <=1, 2, 4, 8, 16, 32, >=64
562c2d2f
DN
1686 IO latencies microseconds: <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
1687 IO latencies milliseconds: <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
f2f788dd
JA
1688 Disk utilization: Disk name, Read ios, write ios,
1689 Read merges, write merges,
1690 Read ticks, write ticks,
3d7cd9b4 1691 Time spent in queue, disk utilization percentage
de8f6de9 1692 Additional Info (dependent on continue_on_error, default off): total # errors, first error code
66c098b8 1693
de8f6de9 1694 Additional Info (dependent on description being set): Text description
25c8b9d7 1695
1db92cb6
JA
1696Completion latency percentiles can be a grouping of up to 20 sets, so
1697for the terse output fio writes all of them. Each field will look like this:
1698
1699 1.00%=6112
1700
1701which is the Xth percentile, and the usec latency associated with it.
1702
f2f788dd
JA
1703For disk utilization, all disks used by fio are shown. So for each disk
1704there will be a disk utilization section.
1705
25c8b9d7
PD
1706
17078.0 Trace file format
1708---------------------
66c098b8 1709There are two trace file format that you can encounter. The older (v1) format
25c8b9d7
PD
1710is unsupported since version 1.20-rc3 (March 2008). It will still be described
1711below in case that you get an old trace and want to understand it.
1712
1713In any case the trace is a simple text file with a single action per line.
1714
1715
17168.1 Trace file format v1
1717------------------------
1718Each line represents a single io action in the following format:
1719
1720rw, offset, length
1721
1722where rw=0/1 for read/write, and the offset and length entries being in bytes.
1723
1724This format is not supported in Fio versions => 1.20-rc3.
1725
1726
17278.2 Trace file format v2
1728------------------------
1729The second version of the trace file format was added in Fio version 1.17.
1730It allows to access more then one file per trace and has a bigger set of
1731possible file actions.
1732
1733The first line of the trace file has to be:
1734
1735fio version 2 iolog
1736
1737Following this can be lines in two different formats, which are described below.
1738
1739The file management format:
1740
1741filename action
1742
1743The filename is given as an absolute path. The action can be one of these:
1744
1745add Add the given filename to the trace
66c098b8 1746open Open the file with the given filename. The filename has to have
25c8b9d7
PD
1747 been added with the add action before.
1748close Close the file with the given filename. The file has to have been
1749 opened before.
1750
1751
1752The file io action format:
1753
1754filename action offset length
1755
1756The filename is given as an absolute path, and has to have been added and opened
66c098b8 1757before it can be used with this format. The offset and length are given in
25c8b9d7
PD
1758bytes. The action can be one of these:
1759
1760wait Wait for 'offset' microseconds. Everything below 100 is discarded.
1761read Read 'length' bytes beginning from 'offset'
1762write Write 'length' bytes beginning from 'offset'
1763sync fsync() the file
1764datasync fdatasync() the file
1765trim trim the given file from the given 'offset' for 'length' bytes
f2a2ce0e
HL
1766
1767
17689.0 CPU idleness profiling
06464907 1769--------------------------
f2a2ce0e
HL
1770In some cases, we want to understand CPU overhead in a test. For example,
1771we test patches for the specific goodness of whether they reduce CPU usage.
1772fio implements a balloon approach to create a thread per CPU that runs at
1773idle priority, meaning that it only runs when nobody else needs the cpu.
1774By measuring the amount of work completed by the thread, idleness of each
1775CPU can be derived accordingly.
1776
1777An unit work is defined as touching a full page of unsigned characters. Mean
1778and standard deviation of time to complete an unit work is reported in "unit
1779work" section. Options can be chosen to report detailed percpu idleness or
1780overall system idleness by aggregating percpu stats.