fio: make gui capture text ops
[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
71bfa161
JA
12
131.0 Overview and history
14------------------------
15fio was originally written to save me the hassle of writing special test
16case programs when I wanted to test a specific workload, either for
17performance reasons or to find/reproduce a bug. The process of writing
18such a test app can be tiresome, especially if you have to do it often.
19Hence I needed a tool that would be able to simulate a given io workload
20without resorting to writing a tailored test case again and again.
21
22A test work load is difficult to define, though. There can be any number
23of processes or threads involved, and they can each be using their own
24way of generating io. You could have someone dirtying large amounts of
25memory in an memory mapped file, or maybe several threads issuing
26reads using asynchronous io. fio needed to be flexible enough to
27simulate both of these cases, and many more.
28
292.0 How fio works
30-----------------
31The first step in getting fio to simulate a desired io workload, is
32writing a job file describing that specific setup. A job file may contain
33any number of threads and/or files - the typical contents of the job file
34is a global section defining shared parameters, and one or more job
35sections describing the jobs involved. When run, fio parses this file
36and sets everything up as described. If we break down a job from top to
37bottom, it contains the following basic parameters:
38
39 IO type Defines the io pattern issued to the file(s).
40 We may only be reading sequentially from this
41 file(s), or we may be writing randomly. Or even
42 mixing reads and writes, sequentially or randomly.
43
44 Block size In how large chunks are we issuing io? This may be
45 a single value, or it may describe a range of
46 block sizes.
47
48 IO size How much data are we going to be reading/writing.
49
50 IO engine How do we issue io? We could be memory mapping the
51 file, we could be using regular read/write, we
d0ff85df 52 could be using splice, async io, syslet, or even
71bfa161
JA
53 SG (SCSI generic sg).
54
6c219763 55 IO depth If the io engine is async, how large a queuing
71bfa161
JA
56 depth do we want to maintain?
57
58 IO type Should we be doing buffered io, or direct/raw io?
59
60 Num files How many files are we spreading the workload over.
61
62 Num threads How many threads or processes should we spread
63 this workload over.
64
65The above are the basic parameters defined for a workload, in addition
66there's a multitude of parameters that modify other aspects of how this
67job behaves.
68
69
703.0 Running fio
71---------------
72See the README file for command line parameters, there are only a few
73of them.
74
75Running fio is normally the easiest part - you just give it the job file
76(or job files) as parameters:
77
78$ fio job_file
79
80and it will start doing what the job_file tells it to do. You can give
81more than one job file on the command line, fio will serialize the running
82of those files. Internally that is the same as using the 'stonewall'
83parameter described the the parameter section.
84
b4692828
JA
85If the job file contains only one job, you may as well just give the
86parameters on the command line. The command line parameters are identical
87to the job parameters, with a few extra that control global parameters
88(see README). For example, for the job file parameter iodepth=2, the
c2b1e753
JA
89mirror command line option would be --iodepth 2 or --iodepth=2. You can
90also use the command line for giving more than one job entry. For each
91--name option that fio sees, it will start a new job with that name.
92Command line entries following a --name entry will apply to that job,
93until there are no more entries or a new --name entry is seen. This is
94similar to the job file options, where each option applies to the current
95job until a new [] job entry is seen.
b4692828 96
71bfa161
JA
97fio does not need to run as root, except if the files or devices specified
98in the job section requires that. Some other options may also be restricted,
6c219763 99such as memory locking, io scheduler switching, and decreasing the nice value.
71bfa161
JA
100
101
1024.0 Job file format
103-------------------
104As previously described, fio accepts one or more job files describing
105what it is supposed to do. The job file format is the classic ini file,
106where the names enclosed in [] brackets define the job name. You are free
107to use any ascii name you want, except 'global' which has special meaning.
108A global section sets defaults for the jobs described in that file. A job
109may override a global section parameter, and a job file may even have
110several global sections if so desired. A job is only affected by a global
65db0851
JA
111section residing above it. If the first character in a line is a ';' or a
112'#', the entire line is discarded as a comment.
71bfa161 113
3c54bc46 114So let's look at a really simple job file that defines two processes, each
b22989b9 115randomly reading from a 128MB file.
71bfa161
JA
116
117; -- start job file --
118[global]
119rw=randread
120size=128m
121
122[job1]
123
124[job2]
125
126; -- end job file --
127
128As you can see, the job file sections themselves are empty as all the
129described parameters are shared. As no filename= option is given, fio
c2b1e753
JA
130makes up a filename for each of the jobs as it sees fit. On the command
131line, this job would look as follows:
132
133$ fio --name=global --rw=randread --size=128m --name=job1 --name=job2
134
71bfa161 135
3c54bc46 136Let's look at an example that has a number of processes writing randomly
71bfa161
JA
137to files.
138
139; -- start job file --
140[random-writers]
141ioengine=libaio
142iodepth=4
143rw=randwrite
144bs=32k
145direct=0
146size=64m
147numjobs=4
148
149; -- end job file --
150
151Here we have no global section, as we only have one job defined anyway.
152We want to use async io here, with a depth of 4 for each file. We also
b22989b9 153increased the buffer size used to 32KB and define numjobs to 4 to
71bfa161 154fork 4 identical jobs. The result is 4 processes each randomly writing
b22989b9 155to their own 64MB file. Instead of using the above job file, you could
b4692828
JA
156have given the parameters on the command line. For this case, you would
157specify:
158
159$ fio --name=random-writers --ioengine=libaio --iodepth=4 --rw=randwrite --bs=32k --direct=0 --size=64m --numjobs=4
71bfa161 160
74929ac2
JA
1614.1 Environment variables
162-------------------------
163
3c54bc46
AC
164fio also supports environment variable expansion in job files. Any
165substring of the form "${VARNAME}" as part of an option value (in other
166words, on the right of the `='), will be expanded to the value of the
167environment variable called VARNAME. If no such environment variable
168is defined, or VARNAME is the empty string, the empty string will be
169substituted.
170
171As an example, let's look at a sample fio invocation and job file:
172
173$ SIZE=64m NUMJOBS=4 fio jobfile.fio
174
175; -- start job file --
176[random-writers]
177rw=randwrite
178size=${SIZE}
179numjobs=${NUMJOBS}
180; -- end job file --
181
182This will expand to the following equivalent job file at runtime:
183
184; -- start job file --
185[random-writers]
186rw=randwrite
187size=64m
188numjobs=4
189; -- end job file --
190
71bfa161
JA
191fio ships with a few example job files, you can also look there for
192inspiration.
193
74929ac2
JA
1944.2 Reserved keywords
195---------------------
196
197Additionally, fio has a set of reserved keywords that will be replaced
198internally with the appropriate value. Those keywords are:
199
200$pagesize The architecture page size of the running system
201$mb_memory Megabytes of total memory in the system
202$ncpus Number of online available CPUs
203
204These can be used on the command line or in the job file, and will be
205automatically substituted with the current system values when the job
892a6ffc
JA
206is run. Simple math is also supported on these keywords, so you can
207perform actions like:
208
209size=8*$mb_memory
210
211and get that properly expanded to 8 times the size of memory in the
212machine.
74929ac2 213
71bfa161
JA
214
2155.0 Detailed list of parameters
216-------------------------------
217
218This section describes in details each parameter associated with a job.
219Some parameters take an option of a given type, such as an integer or
220a string. The following types are used:
221
222str String. This is a sequence of alpha characters.
b09da8fa 223time Integer with possible time suffix. In seconds unless otherwise
e417fd66
JA
224 specified, use eg 10m for 10 minutes. Accepts s/m/h for seconds,
225 minutes, and hours.
b09da8fa
JA
226int SI integer. A whole number value, which may contain a suffix
227 describing the base of the number. Accepted suffixes are k/m/g/t/p,
228 meaning kilo, mega, giga, tera, and peta. The suffix is not case
57fc29fa
JA
229 sensitive, and you may also include trailing 'b' (eg 'kb' is the same
230 as 'k'). So if you want to specify 4096, you could either write
b09da8fa 231 out '4096' or just give 4k. The suffixes signify base 2 values, so
57fc29fa
JA
232 1024 is 1k and 1024k is 1m and so on, unless the suffix is explicitly
233 set to a base 10 value using 'kib', 'mib', 'gib', etc. If that is the
234 case, then 1000 is used as the multiplier. This can be handy for
235 disks, since manufacturers generally use base 10 values when listing
236 the capacity of a drive. If the option accepts an upper and lower
237 range, use a colon ':' or minus '-' to separate such values. May also
238 include a prefix to indicate numbers base. If 0x is used, the number
239 is assumed to be hexadecimal. See irange.
71bfa161
JA
240bool Boolean. Usually parsed as an integer, however only defined for
241 true and false (1 and 0).
b09da8fa 242irange Integer range with suffix. Allows value range to be given, such
bf9a3edb 243 as 1024-4096. A colon may also be used as the separator, eg
0c9baf91
JA
244 1k:4k. If the option allows two sets of ranges, they can be
245 specified with a ',' or '/' delimiter: 1k-4k/8k-32k. Also see
f7fa2653 246 int.
83349190 247float_list A list of floating numbers, separated by a ':' character.
71bfa161
JA
248
249With the above in mind, here follows the complete list of fio job
250parameters.
251
252name=str ASCII name of the job. This may be used to override the
253 name printed by fio for this job. Otherwise the job
c2b1e753 254 name is used. On the command line this parameter has the
6c219763 255 special purpose of also signaling the start of a new
c2b1e753 256 job.
71bfa161 257
61697c37
JA
258description=str Text description of the job. Doesn't do anything except
259 dump this text description when this job is run. It's
260 not parsed.
261
3776041e 262directory=str Prefix filenames with this directory. Used to place files
71bfa161
JA
263 in a different location than "./".
264
265filename=str Fio normally makes up a filename based on the job name,
266 thread number, and file number. If you want to share
267 files between threads in a job or several jobs, specify
ed92ac0c 268 a filename for each of them to override the default. If
414c2a3e 269 the ioengine used is 'net', the filename is the host, port,
0fd666bf 270 and protocol to use in the format of =host,port,protocol.
414c2a3e
JA
271 See ioengine=net for more. If the ioengine is file based, you
272 can specify a number of files by separating the names with a
273 ':' colon. So if you wanted a job to open /dev/sda and /dev/sdb
274 as the two working files, you would use
03e20d68 275 filename=/dev/sda:/dev/sdb. On Windows, disk devices are accessed
ecc314ba
BC
276 as \\.\PhysicalDrive0 for the first device, \\.\PhysicalDrive1
277 for the second etc. If the wanted filename does need to
278 include a colon, then escape that with a '\' character.
279 For instance, if the filename is "/dev/dsk/foo@3,0:c",
280 then you would use filename="/dev/dsk/foo@3,0\:c".
03e20d68
BC
281 '-' is a reserved name, meaning stdin or stdout. Which of the
282 two depends on the read/write direction set.
71bfa161 283
bbf6b540
JA
284opendir=str Tell fio to recursively add any file it can find in this
285 directory and down the file system tree.
286
3776041e 287lockfile=str Fio defaults to not locking any files before it does
4d4e80f2
JA
288 IO to them. If a file or file descriptor is shared, fio
289 can serialize IO to that file to make the end result
290 consistent. This is usual for emulating real workloads that
291 share files. The lock modes are:
292
293 none No locking. The default.
294 exclusive Only one thread/process may do IO,
295 excluding all others.
296 readwrite Read-write locking on the file. Many
297 readers may access the file at the
298 same time, but writes get exclusive
299 access.
300
301 The option may be post-fixed with a lock batch number. If
302 set, then each thread/process may do that amount of IOs to
bf9a3edb 303 the file before giving up the lock. Since lock acquisition is
4d4e80f2 304 expensive, batching the lock/unlocks will speed up IO.
29c1349f 305
d3aad8f2 306readwrite=str
71bfa161
JA
307rw=str Type of io pattern. Accepted values are:
308
309 read Sequential reads
310 write Sequential writes
311 randwrite Random writes
312 randread Random reads
313 rw Sequential mixed reads and writes
314 randrw Random mixed reads and writes
315
316 For the mixed io types, the default is to split them 50/50.
317 For certain types of io the result may still be skewed a bit,
211097b2 318 since the speed may be different. It is possible to specify
38dad62d
JA
319 a number of IO's to do before getting a new offset, this is
320 one by appending a ':<nr>' to the end of the string given.
321 For a random read, it would look like 'rw=randread:8' for
059b0802
JA
322 passing in an offset modifier with a value of 8. If the
323 postfix is used with a sequential IO pattern, then the value
324 specified will be added to the generated offset for each IO.
325 For instance, using rw=write:4k will skip 4k for every
326 write. It turns sequential IO into sequential IO with holes.
327 See the 'rw_sequencer' option.
38dad62d
JA
328
329rw_sequencer=str If an offset modifier is given by appending a number to
330 the rw=<str> line, then this option controls how that
331 number modifies the IO offset being generated. Accepted
332 values are:
333
334 sequential Generate sequential offset
335 identical Generate the same offset
336
337 'sequential' is only useful for random IO, where fio would
338 normally generate a new random offset for every IO. If you
339 append eg 8 to randread, you would get a new random offset for
211097b2
JA
340 every 8 IO's. The result would be a seek for only every 8
341 IO's, instead of for every IO. Use rw=randread:8 to specify
38dad62d
JA
342 that. As sequential IO is already sequential, setting
343 'sequential' for that would not result in any differences.
344 'identical' behaves in a similar fashion, except it sends
345 the same offset 8 number of times before generating a new
346 offset.
71bfa161 347
90fef2d1
JA
348kb_base=int The base unit for a kilobyte. The defacto base is 2^10, 1024.
349 Storage manufacturers like to use 10^3 or 1000 as a base
350 ten unit instead, for obvious reasons. Allow values are
351 1024 or 1000, with 1024 being the default.
352
ee738499
JA
353randrepeat=bool For random IO workloads, seed the generator in a predictable
354 way so that results are repeatable across repetitions.
355
2615cc4b
JA
356use_os_rand=bool Fio can either use the random generator supplied by the OS
357 to generator random offsets, or it can use it's own internal
358 generator (based on Tausworthe). Default is to use the
359 internal generator, which is often of better quality and
360 faster.
361
a596f047
EG
362fallocate=str Whether pre-allocation is performed when laying down files.
363 Accepted values are:
364
365 none Do not pre-allocate space
366 posix Pre-allocate via posix_fallocate()
367 keep Pre-allocate via fallocate() with
368 FALLOC_FL_KEEP_SIZE set
369 0 Backward-compatible alias for 'none'
370 1 Backward-compatible alias for 'posix'
371
372 May not be available on all supported platforms. 'keep' is only
373 available on Linux.If using ZFS on Solaris this must be set to
374 'none' because ZFS doesn't support it. Default: 'posix'.
7bc8c2cf 375
d2f3ac35
JA
376fadvise_hint=bool By default, fio will use fadvise() to advise the kernel
377 on what IO patterns it is likely to issue. Sometimes you
378 want to test specific IO patterns without telling the
379 kernel about it, in which case you can disable this option.
380 If set, fio will use POSIX_FADV_SEQUENTIAL for sequential
381 IO and POSIX_FADV_RANDOM for random IO.
382
f7fa2653 383size=int The total size of file io for this job. Fio will run until
7616cafe
JA
384 this many bytes has been transferred, unless runtime is
385 limited by other options (such as 'runtime', for instance).
3776041e 386 Unless specific nrfiles and filesize options are given,
7616cafe 387 fio will divide this size between the available files
d6667268
JA
388 specified by the job. If not set, fio will use the full
389 size of the given files or devices. If the the files
7bb59102
JA
390 do not exist, size must be given. It is also possible to
391 give size as a percentage between 1 and 100. If size=20%
392 is given, fio will use 20% of the full size of the given
393 files or devices.
71bfa161 394
f7fa2653 395filesize=int Individual file sizes. May be a range, in which case fio
9c60ce64
JA
396 will select sizes for files at random within the given range
397 and limited to 'size' in total (if that is given). If not
398 given, each created file is the same size.
399
74586c1e
JA
400fill_device=bool
401fill_fs=bool Sets size to something really large and waits for ENOSPC (no
aa31f1f1 402 space left on device) as the terminating condition. Only makes
3ce9dcaf 403 sense with sequential write. For a read workload, the mount
4f12432e
JA
404 point will be filled first then IO started on the result. This
405 option doesn't make sense if operating on a raw device node,
406 since the size of that is already known by the file system.
407 Additionally, writing beyond end-of-device will not return
408 ENOSPC there.
aa31f1f1 409
f7fa2653
JA
410blocksize=int
411bs=int The block size used for the io units. Defaults to 4k. Values
412 can be given for both read and writes. If a single int is
413 given, it will apply to both. If a second int is specified
f90eff5a
JA
414 after a comma, it will apply to writes only. In other words,
415 the format is either bs=read_and_write or bs=read,write.
416 bs=4k,8k will thus use 4k blocks for reads, and 8k blocks
787f7e95
JA
417 for writes. If you only wish to set the write size, you
418 can do so by passing an empty read size - bs=,8k will set
419 8k for writes and leave the read default value.
a00735e6 420
2b7a01d0
JA
421blockalign=int
422ba=int At what boundary to align random IO offsets. Defaults to
423 the same as 'blocksize' the minimum blocksize given.
424 Minimum alignment is typically 512b for using direct IO,
425 though it usually depends on the hardware block size. This
426 option is mutually exclusive with using a random map for
427 files, so it will turn off that option.
428
d3aad8f2 429blocksize_range=irange
71bfa161
JA
430bsrange=irange Instead of giving a single block size, specify a range
431 and fio will mix the issued io block sizes. The issued
432 io unit will always be a multiple of the minimum value
f90eff5a
JA
433 given (also see bs_unaligned). Applies to both reads and
434 writes, however a second range can be given after a comma.
435 See bs=.
a00735e6 436
564ca972
JA
437bssplit=str Sometimes you want even finer grained control of the
438 block sizes issued, not just an even split between them.
439 This option allows you to weight various block sizes,
440 so that you are able to define a specific amount of
441 block sizes issued. The format for this option is:
442
443 bssplit=blocksize/percentage:blocksize/percentage
444
445 for as many block sizes as needed. So if you want to define
446 a workload that has 50% 64k blocks, 10% 4k blocks, and
447 40% 32k blocks, you would write:
448
449 bssplit=4k/10:64k/50:32k/40
450
451 Ordering does not matter. If the percentage is left blank,
452 fio will fill in the remaining values evenly. So a bssplit
453 option like this one:
454
455 bssplit=4k/50:1k/:32k/
456
457 would have 50% 4k ios, and 25% 1k and 32k ios. The percentages
458 always add up to 100, if bssplit is given a range that adds
459 up to more, it will error out.
460
720e84ad
JA
461 bssplit also supports giving separate splits to reads and
462 writes. The format is identical to what bs= accepts. You
463 have to separate the read and write parts with a comma. So
464 if you want a workload that has 50% 2k reads and 50% 4k reads,
465 while having 90% 4k writes and 10% 8k writes, you would
466 specify:
467
468 bssplit=2k/50:4k/50,4k/90,8k/10
469
d3aad8f2 470blocksize_unaligned
690adba3
JA
471bs_unaligned If this option is given, any byte size value within bsrange
472 may be used as a block range. This typically wont work with
473 direct IO, as that normally requires sector alignment.
71bfa161 474
e9459e5a
JA
475zero_buffers If this option is given, fio will init the IO buffers to
476 all zeroes. The default is to fill them with random data.
477
5973cafb
JA
478refill_buffers If this option is given, fio will refill the IO buffers
479 on every submit. The default is to only fill it at init
480 time and reuse that data. Only makes sense if zero_buffers
41ccd845
JA
481 isn't specified, naturally. If data verification is enabled,
482 refill_buffers is also automatically enabled.
5973cafb 483
fd68418e
JA
484scramble_buffers=bool If refill_buffers is too costly and the target is
485 using data deduplication, then setting this option will
486 slightly modify the IO buffer contents to defeat normal
487 de-dupe attempts. This is not enough to defeat more clever
488 block compression attempts, but it will stop naive dedupe of
489 blocks. Default: true.
490
71bfa161
JA
491nrfiles=int Number of files to use for this job. Defaults to 1.
492
390b1537
JA
493openfiles=int Number of files to keep open at the same time. Defaults to
494 the same as nrfiles, can be set smaller to limit the number
495 simultaneous opens.
496
5af1c6f3
JA
497file_service_type=str Defines how fio decides which file from a job to
498 service next. The following types are defined:
499
500 random Just choose a file at random.
501
502 roundrobin Round robin over open files. This
503 is the default.
504
a086c257
JA
505 sequential Finish one file before moving on to
506 the next. Multiple files can still be
507 open depending on 'openfiles'.
508
1907dbc6
JA
509 The string can have a number appended, indicating how
510 often to switch to a new file. So if option random:4 is
511 given, fio will switch to a new random file after 4 ios
512 have been issued.
513
71bfa161
JA
514ioengine=str Defines how the job issues io to the file. The following
515 types are defined:
516
517 sync Basic read(2) or write(2) io. lseek(2) is
518 used to position the io location.
519
a31041ea 520 psync Basic pread(2) or pwrite(2) io.
521
e05af9e5 522 vsync Basic readv(2) or writev(2) IO.
1d2af02a 523
15d182aa
JA
524 libaio Linux native asynchronous io. Note that Linux
525 may only support queued behaviour with
526 non-buffered IO (set direct=1 or buffered=0).
de890a1e 527 This engine defines engine specific options.
71bfa161
JA
528
529 posixaio glibc posix asynchronous io.
530
417f0068
JA
531 solarisaio Solaris native asynchronous io.
532
03e20d68
BC
533 windowsaio Windows native asynchronous io.
534
71bfa161
JA
535 mmap File is memory mapped and data copied
536 to/from using memcpy(3).
537
538 splice splice(2) is used to transfer the data and
539 vmsplice(2) to transfer data from user
540 space to the kernel.
541
d0ff85df
JA
542 syslet-rw Use the syslet system calls to make
543 regular read/write async.
544
71bfa161 545 sg SCSI generic sg v3 io. May either be
6c219763 546 synchronous using the SG_IO ioctl, or if
71bfa161
JA
547 the target is an sg character device
548 we use read(2) and write(2) for asynchronous
549 io.
550
a94ea28b
JA
551 null Doesn't transfer any data, just pretends
552 to. This is mainly used to exercise fio
553 itself and for debugging/testing purposes.
554
ed92ac0c 555 net Transfer over the network to given host:port.
de890a1e
SL
556 Depending on the protocol used, the hostname,
557 port, listen and filename options are used to
558 specify what sort of connection to make, while
559 the protocol option determines which protocol
560 will be used.
561 This engine defines engine specific options.
ed92ac0c 562
9cce02e8
JA
563 netsplice Like net, but uses splice/vmsplice to
564 map data and send/receive.
de890a1e 565 This engine defines engine specific options.
9cce02e8 566
53aec0a4 567 cpuio Doesn't transfer any data, but burns CPU
ba0fbe10
JA
568 cycles according to the cpuload= and
569 cpucycle= options. Setting cpuload=85
570 will cause that job to do nothing but burn
36ecec83
GP
571 85% of the CPU. In case of SMP machines,
572 use numjobs=<no_of_cpu> to get desired CPU
573 usage, as the cpuload only loads a single
574 CPU at the desired rate.
ba0fbe10 575
e9a1806f
JA
576 guasi The GUASI IO engine is the Generic Userspace
577 Asyncronous Syscall Interface approach
578 to async IO. See
579
580 http://www.xmailserver.org/guasi-lib.html
581
582 for more info on GUASI.
583
21b8aee8 584 rdma The RDMA I/O engine supports both RDMA
eb52fa3f
BVA
585 memory semantics (RDMA_WRITE/RDMA_READ) and
586 channel semantics (Send/Recv) for the
587 InfiniBand, RoCE and iWARP protocols.
21b8aee8 588
8a7bd877
JA
589 external Prefix to specify loading an external
590 IO engine object file. Append the engine
591 filename, eg ioengine=external:/tmp/foo.o
592 to load ioengine foo.o in /tmp.
593
71bfa161
JA
594iodepth=int This defines how many io units to keep in flight against
595 the file. The default is 1 for each file defined in this
596 job, can be overridden with a larger value for higher
ee72ca09
JA
597 concurrency. Note that increasing iodepth beyond 1 will not
598 affect synchronous ioengines (except for small degress when
9b836561 599 verify_async is in use). Even async engines may impose OS
ee72ca09
JA
600 restrictions causing the desired depth not to be achieved.
601 This may happen on Linux when using libaio and not setting
602 direct=1, since buffered IO is not async on that OS. Keep an
603 eye on the IO depth distribution in the fio output to verify
604 that the achieved depth is as expected. Default: 1.
71bfa161 605
4950421a 606iodepth_batch_submit=int
cb5ab512 607iodepth_batch=int This defines how many pieces of IO to submit at once.
89e820f6
JA
608 It defaults to 1 which means that we submit each IO
609 as soon as it is available, but can be raised to submit
610 bigger batches of IO at the time.
cb5ab512 611
4950421a
JA
612iodepth_batch_complete=int This defines how many pieces of IO to retrieve
613 at once. It defaults to 1 which means that we'll ask
614 for a minimum of 1 IO in the retrieval process from
615 the kernel. The IO retrieval will go on until we
616 hit the limit set by iodepth_low. If this variable is
617 set to 0, then fio will always check for completed
618 events before queuing more IO. This helps reduce
619 IO latency, at the cost of more retrieval system calls.
620
e916b390
JA
621iodepth_low=int The low water mark indicating when to start filling
622 the queue again. Defaults to the same as iodepth, meaning
623 that fio will attempt to keep the queue full at all times.
624 If iodepth is set to eg 16 and iodepth_low is set to 4, then
625 after fio has filled the queue of 16 requests, it will let
626 the depth drain down to 4 before starting to fill it again.
627
71bfa161 628direct=bool If value is true, use non-buffered io. This is usually
9b836561 629 O_DIRECT. Note that ZFS on Solaris doesn't support direct io.
93bcfd20 630 On Windows the synchronous ioengines don't support direct io.
76a43db4
JA
631
632buffered=bool If value is true, use buffered io. This is the opposite
633 of the 'direct' option. Defaults to true.
71bfa161 634
f7fa2653 635offset=int Start io at the given offset in the file. The data before
71bfa161
JA
636 the given offset will not be touched. This effectively
637 caps the file size at real_size - offset.
638
639fsync=int If writing to a file, issue a sync of the dirty data
640 for every number of blocks given. For example, if you give
641 32 as a parameter, fio will sync the file for every 32
642 writes issued. If fio is using non-buffered io, we may
643 not sync the file. The exception is the sg io engine, which
6c219763 644 synchronizes the disk cache anyway.
71bfa161 645
e76b1da4 646fdatasync=int Like fsync= but uses fdatasync() to only sync data and not
5f9099ea 647 metadata blocks.
93bcfd20 648 In FreeBSD and Windows there is no fdatasync(), this falls back to
e72fa4d4 649 using fsync()
5f9099ea 650
e76b1da4
JA
651sync_file_range=str:val Use sync_file_range() for every 'val' number of
652 write operations. Fio will track range of writes that
653 have happened since the last sync_file_range() call. 'str'
654 can currently be one or more of:
655
656 wait_before SYNC_FILE_RANGE_WAIT_BEFORE
657 write SYNC_FILE_RANGE_WRITE
658 wait_after SYNC_FILE_RANGE_WAIT_AFTER
659
660 So if you do sync_file_range=wait_before,write:8, fio would
661 use SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE for
662 every 8 writes. Also see the sync_file_range(2) man page.
663 This option is Linux specific.
664
5036fc1e
JA
665overwrite=bool If true, writes to a file will always overwrite existing
666 data. If the file doesn't already exist, it will be
667 created before the write phase begins. If the file exists
668 and is large enough for the specified write phase, nothing
669 will be done.
71bfa161
JA
670
671end_fsync=bool If true, fsync file contents when the job exits.
672
ebb1415f
JA
673fsync_on_close=bool If true, fio will fsync() a dirty file on close.
674 This differs from end_fsync in that it will happen on every
675 file close, not just at the end of the job.
676
71bfa161
JA
677rwmixread=int How large a percentage of the mix should be reads.
678
679rwmixwrite=int How large a percentage of the mix should be writes. If both
680 rwmixread and rwmixwrite is given and the values do not add
681 up to 100%, the latter of the two will be used to override
c35dd7a6
JA
682 the first. This may interfere with a given rate setting,
683 if fio is asked to limit reads or writes to a certain rate.
684 If that is the case, then the distribution may be skewed.
71bfa161 685
bb8895e0
JA
686norandommap Normally fio will cover every block of the file when doing
687 random IO. If this option is given, fio will just get a
688 new random offset without looking at past io history. This
689 means that some blocks may not be read or written, and that
690 some blocks may be read/written more than once. This option
8347239a
JA
691 is mutually exclusive with verify= if and only if multiple
692 blocksizes (via bsrange=) are used, since fio only tracks
693 complete rewrites of blocks.
bb8895e0 694
0408c206
JA
695softrandommap=bool See norandommap. If fio runs with the random block map
696 enabled and it fails to allocate the map, if this option is
697 set it will continue without a random block map. As coverage
698 will not be as complete as with random maps, this option is
2b386d25
JA
699 disabled by default.
700
71bfa161
JA
701nice=int Run the job with the given nice value. See man nice(2).
702
703prio=int Set the io priority value of this job. Linux limits us to
704 a positive value between 0 and 7, with 0 being the highest.
705 See man ionice(1).
706
707prioclass=int Set the io priority class. See man ionice(1).
708
709thinktime=int Stall the job x microseconds after an io has completed before
710 issuing the next. May be used to simulate processing being
48097d5c
JA
711 done by an application. See thinktime_blocks and
712 thinktime_spin.
713
714thinktime_spin=int
715 Only valid if thinktime is set - pretend to spend CPU time
716 doing something with the data received, before falling back
717 to sleeping for the rest of the period specified by
718 thinktime.
9c1f7434
JA
719
720thinktime_blocks
721 Only valid if thinktime is set - control how many blocks
722 to issue, before waiting 'thinktime' usecs. If not set,
723 defaults to 1 which will make fio wait 'thinktime' usecs
724 after every block.
71bfa161 725
581e7141 726rate=int Cap the bandwidth used by this job. The number is in bytes/sec,
b09da8fa 727 the normal suffix rules apply. You can use rate=500k to limit
581e7141
JA
728 reads and writes to 500k each, or you can specify read and
729 writes separately. Using rate=1m,500k would limit reads to
730 1MB/sec and writes to 500KB/sec. Capping only reads or
731 writes can be done with rate=,500k or rate=500k,. The former
732 will only limit writes (to 500KB/sec), the latter will only
733 limit reads.
71bfa161
JA
734
735ratemin=int Tell fio to do whatever it can to maintain at least this
4e991c23 736 bandwidth. Failing to meet this requirement, will cause
581e7141
JA
737 the job to exit. The same format as rate is used for
738 read vs write separation.
4e991c23
JA
739
740rate_iops=int Cap the bandwidth to this number of IOPS. Basically the same
741 as rate, just specified independently of bandwidth. If the
742 job is given a block size range instead of a fixed value,
581e7141
JA
743 the smallest block size is used as the metric. The same format
744 as rate is used for read vs write seperation.
4e991c23
JA
745
746rate_iops_min=int If fio doesn't meet this rate of IO, it will cause
581e7141
JA
747 the job to exit. The same format as rate is used for read vs
748 write seperation.
71bfa161
JA
749
750ratecycle=int Average bandwidth for 'rate' and 'ratemin' over this number
6c219763 751 of milliseconds.
71bfa161
JA
752
753cpumask=int Set the CPU affinity of this job. The parameter given is a
a08bc17f
JA
754 bitmask of allowed CPU's the job may run on. So if you want
755 the allowed CPUs to be 1 and 5, you would pass the decimal
756 value of (1 << 1 | 1 << 5), or 34. See man
7dbb6eba 757 sched_setaffinity(2). This may not work on all supported
b0ea08ce
JA
758 operating systems or kernel versions. This option doesn't
759 work well for a higher CPU count than what you can store in
760 an integer mask, so it can only control cpus 1-32. For
761 boxes with larger CPU counts, use cpus_allowed.
71bfa161 762
d2e268b0
JA
763cpus_allowed=str Controls the same options as cpumask, but it allows a text
764 setting of the permitted CPUs instead. So to use CPUs 1 and
62a7273d
JA
765 5, you would specify cpus_allowed=1,5. This options also
766 allows a range of CPUs. Say you wanted a binding to CPUs
767 1, 5, and 8-15, you would set cpus_allowed=1,5,8-15.
d2e268b0 768
e417fd66 769startdelay=time Start this job the specified number of seconds after fio
71bfa161
JA
770 has started. Only useful if the job file contains several
771 jobs, and you want to delay starting some jobs to a certain
772 time.
773
e417fd66 774runtime=time Tell fio to terminate processing after the specified number
71bfa161
JA
775 of seconds. It can be quite hard to determine for how long
776 a specified job will run, so this parameter is handy to
777 cap the total runtime to a given time.
778
cf4464ca 779time_based If set, fio will run for the duration of the runtime
bf9a3edb 780 specified even if the file(s) are completely read or
cf4464ca
JA
781 written. It will simply loop over the same workload
782 as many times as the runtime allows.
783
e417fd66 784ramp_time=time If set, fio will run the specified workload for this amount
721938ae
JA
785 of time before logging any performance numbers. Useful for
786 letting performance settle before logging results, thus
b29ee5b3
JA
787 minimizing the runtime required for stable results. Note
788 that the ramp_time is considered lead in time for a job,
789 thus it will increase the total runtime if a special timeout
790 or runtime is specified.
721938ae 791
71bfa161
JA
792invalidate=bool Invalidate the buffer/page cache parts for this file prior
793 to starting io. Defaults to true.
794
795sync=bool Use sync io for buffered writes. For the majority of the
796 io engines, this means using O_SYNC.
797
d3aad8f2 798iomem=str
71bfa161
JA
799mem=str Fio can use various types of memory as the io unit buffer.
800 The allowed values are:
801
802 malloc Use memory from malloc(3) as the buffers.
803
804 shm Use shared memory as the buffers. Allocated
805 through shmget(2).
806
74b025b0
JA
807 shmhuge Same as shm, but use huge pages as backing.
808
313cb206
JA
809 mmap Use mmap to allocate buffers. May either be
810 anonymous memory, or can be file backed if
811 a filename is given after the option. The
812 format is mem=mmap:/path/to/file.
71bfa161 813
d0bdaf49
JA
814 mmaphuge Use a memory mapped huge file as the buffer
815 backing. Append filename after mmaphuge, ala
816 mem=mmaphuge:/hugetlbfs/file
817
71bfa161 818 The area allocated is a function of the maximum allowed
5394ae5f
JA
819 bs size for the job, multiplied by the io depth given. Note
820 that for shmhuge and mmaphuge to work, the system must have
821 free huge pages allocated. This can normally be checked
822 and set by reading/writing /proc/sys/vm/nr_hugepages on a
b22989b9 823 Linux system. Fio assumes a huge page is 4MB in size. So
5394ae5f
JA
824 to calculate the number of huge pages you need for a given
825 job file, add up the io depth of all jobs (normally one unless
826 iodepth= is used) and multiply by the maximum bs set. Then
827 divide that number by the huge page size. You can see the
828 size of the huge pages in /proc/meminfo. If no huge pages
829 are allocated by having a non-zero number in nr_hugepages,
56bb17f2 830 using mmaphuge or shmhuge will fail. Also see hugepage-size.
5394ae5f
JA
831
832 mmaphuge also needs to have hugetlbfs mounted and the file
833 location should point there. So if it's mounted in /huge,
834 you would use mem=mmaphuge:/huge/somefile.
71bfa161 835
d529ee19
JA
836iomem_align=int This indiciates the memory alignment of the IO memory buffers.
837 Note that the given alignment is applied to the first IO unit
838 buffer, if using iodepth the alignment of the following buffers
839 are given by the bs used. In other words, if using a bs that is
840 a multiple of the page sized in the system, all buffers will
841 be aligned to this value. If using a bs that is not page
842 aligned, the alignment of subsequent IO memory buffers is the
843 sum of the iomem_align and bs used.
844
f7fa2653 845hugepage-size=int
56bb17f2 846 Defines the size of a huge page. Must at least be equal
b22989b9 847 to the system setting, see /proc/meminfo. Defaults to 4MB.
c51074e7
JA
848 Should probably always be a multiple of megabytes, so using
849 hugepage-size=Xm is the preferred way to set this to avoid
850 setting a non-pow-2 bad value.
56bb17f2 851
71bfa161
JA
852exitall When one job finishes, terminate the rest. The default is
853 to wait for each job to finish, sometimes that is not the
854 desired action.
855
856bwavgtime=int Average the calculated bandwidth over the given time. Value
6c219763 857 is specified in milliseconds.
71bfa161 858
c8eeb9df
JA
859iopsavgtime=int Average the calculated IOPS over the given time. Value
860 is specified in milliseconds.
861
71bfa161
JA
862create_serialize=bool If true, serialize the file creating for the jobs.
863 This may be handy to avoid interleaving of data
864 files, which may greatly depend on the filesystem
865 used and even the number of processors in the system.
866
867create_fsync=bool fsync the data file after creation. This is the
868 default.
869
814452bd
JA
870create_on_open=bool Don't pre-setup the files for IO, just create open()
871 when it's time to do IO to that file.
872
afad68f7 873pre_read=bool If this is given, files will be pre-read into memory before
34f1c044
JA
874 starting the given IO operation. This will also clear
875 the 'invalidate' flag, since it is pointless to pre-read
9c0d2241
JA
876 and then drop the cache. This will only work for IO engines
877 that are seekable, since they allow you to read the same data
878 multiple times. Thus it will not work on eg network or splice
879 IO.
afad68f7 880
e545a6ce 881unlink=bool Unlink the job files when done. Not the default, as repeated
bf9a3edb
JA
882 runs of that job would then waste time recreating the file
883 set again and again.
71bfa161
JA
884
885loops=int Run the specified number of iterations of this job. Used
886 to repeat the same workload a given number of times. Defaults
887 to 1.
888
68e1f29a 889do_verify=bool Run the verify phase after a write phase. Only makes sense if
e84c73a8
SL
890 verify is set. Defaults to 1.
891
71bfa161
JA
892verify=str If writing to a file, fio can verify the file contents
893 after each iteration of the job. The allowed values are:
894
895 md5 Use an md5 sum of the data area and store
896 it in the header of each block.
897
17dc34df
JA
898 crc64 Use an experimental crc64 sum of the data
899 area and store it in the header of each
900 block.
901
bac39e0e
JA
902 crc32c Use a crc32c sum of the data area and store
903 it in the header of each block.
904
3845591f 905 crc32c-intel Use hardware assisted crc32c calcuation
0539d758
JA
906 provided on SSE4.2 enabled processors. Falls
907 back to regular software crc32c, if not
908 supported by the system.
3845591f 909
71bfa161
JA
910 crc32 Use a crc32 sum of the data area and store
911 it in the header of each block.
912
969f7ed3
JA
913 crc16 Use a crc16 sum of the data area and store
914 it in the header of each block.
915
17dc34df
JA
916 crc7 Use a crc7 sum of the data area and store
917 it in the header of each block.
918
cd14cc10
JA
919 sha512 Use sha512 as the checksum function.
920
921 sha256 Use sha256 as the checksum function.
922
7c353ceb
JA
923 sha1 Use optimized sha1 as the checksum function.
924
7437ee87
SL
925 meta Write extra information about each io
926 (timestamp, block number etc.). The block
996093bb 927 number is verified. See also verify_pattern.
7437ee87 928
36690c9b
JA
929 null Only pretend to verify. Useful for testing
930 internals with ioengine=null, not for much
931 else.
932
6c219763 933 This option can be used for repeated burn-in tests of a
71bfa161 934 system to make sure that the written data is also
b892dc08
JA
935 correctly read back. If the data direction given is
936 a read or random read, fio will assume that it should
937 verify a previously written file. If the data direction
938 includes any form of write, the verify will be of the
939 newly written data.
71bfa161 940
160b966d
JA
941verifysort=bool If set, fio will sort written verify blocks when it deems
942 it faster to read them back in a sorted manner. This is
943 often the case when overwriting an existing file, since
944 the blocks are already laid out in the file system. You
945 can ignore this option unless doing huge amounts of really
946 fast IO where the red-black tree sorting CPU time becomes
947 significant.
3f9f4e26 948
f7fa2653 949verify_offset=int Swap the verification header with data somewhere else
546a9142
SL
950 in the block before writing. Its swapped back before
951 verifying.
952
f7fa2653 953verify_interval=int Write the verification header at a finer granularity
3f9f4e26
SL
954 than the blocksize. It will be written for chunks the
955 size of header_interval. blocksize should divide this
956 evenly.
90059d65 957
0e92f873 958verify_pattern=str If set, fio will fill the io buffers with this
e28218f3
SL
959 pattern. Fio defaults to filling with totally random
960 bytes, but sometimes it's interesting to fill with a known
961 pattern for io verification purposes. Depending on the
962 width of the pattern, fio will fill 1/2/3/4 bytes of the
0e92f873
RR
963 buffer at the time(it can be either a decimal or a hex number).
964 The verify_pattern if larger than a 32-bit quantity has to
996093bb
JA
965 be a hex number that starts with either "0x" or "0X". Use
966 with verify=meta.
e28218f3 967
68e1f29a 968verify_fatal=bool Normally fio will keep checking the entire contents
a12a3b4d
JA
969 before quitting on a block verification failure. If this
970 option is set, fio will exit the job on the first observed
971 failure.
e8462bd8 972
b463e936
JA
973verify_dump=bool If set, dump the contents of both the original data
974 block and the data block we read off disk to files. This
975 allows later analysis to inspect just what kind of data
ef71e317 976 corruption occurred. Off by default.
b463e936 977
e8462bd8
JA
978verify_async=int Fio will normally verify IO inline from the submitting
979 thread. This option takes an integer describing how many
980 async offload threads to create for IO verification instead,
981 causing fio to offload the duty of verifying IO contents
c85c324c
JA
982 to one or more separate threads. If using this offload
983 option, even sync IO engines can benefit from using an
984 iodepth setting higher than 1, as it allows them to have
985 IO in flight while verifies are running.
e8462bd8
JA
986
987verify_async_cpus=str Tell fio to set the given CPU affinity on the
988 async IO verification threads. See cpus_allowed for the
989 format used.
6f87418f
JA
990
991verify_backlog=int Fio will normally verify the written contents of a
992 job that utilizes verify once that job has completed. In
993 other words, everything is written then everything is read
994 back and verified. You may want to verify continually
995 instead for a variety of reasons. Fio stores the meta data
996 associated with an IO block in memory, so for large
997 verify workloads, quite a bit of memory would be used up
998 holding this meta data. If this option is enabled, fio
f42195a3
JA
999 will write only N blocks before verifying these blocks.
1000
6f87418f
JA
1001 will verify the previously written blocks before continuing
1002 to write new ones.
1003
1004verify_backlog_batch=int Control how many blocks fio will verify
1005 if verify_backlog is set. If not set, will default to
1006 the value of verify_backlog (meaning the entire queue
f42195a3
JA
1007 is read back and verified). If verify_backlog_batch is
1008 less than verify_backlog then not all blocks will be verified,
1009 if verify_backlog_batch is larger than verify_backlog, some
1010 blocks will be verified more than once.
160b966d 1011
d392365e
JA
1012stonewall
1013wait_for_previous Wait for preceeding jobs in the job file to exit, before
71bfa161 1014 starting this one. Can be used to insert serialization
b3d62a75
JA
1015 points in the job file. A stone wall also implies starting
1016 a new reporting group.
1017
1018new_group Start a new reporting group. If this option isn't given,
1019 jobs in a file will be part of the same reporting group
bf9a3edb 1020 unless separated by a stone wall (or if it's a group
b3d62a75 1021 by itself, with the numjobs option).
71bfa161
JA
1022
1023numjobs=int Create the specified number of clones of this job. May be
1024 used to setup a larger number of threads/processes doing
fa28c85a
JA
1025 the same thing. We regard that grouping of jobs as a
1026 specific group.
1027
1028group_reporting If 'numjobs' is set, it may be interesting to display
1029 statistics for the group as a whole instead of for each
1030 individual job. This is especially true of 'numjobs' is
1031 large, looking at individual thread/process output quickly
1032 becomes unwieldy. If 'group_reporting' is specified, fio
1033 will show the final report per-group instead of per-job.
71bfa161
JA
1034
1035thread fio defaults to forking jobs, however if this option is
1036 given, fio will use pthread_create(3) to create threads
1037 instead.
1038
f7fa2653 1039zonesize=int Divide a file into zones of the specified size. See zoneskip.
71bfa161 1040
f7fa2653 1041zoneskip=int Skip the specified number of bytes when zonesize data has
71bfa161
JA
1042 been read. The two zone options can be used to only do
1043 io on zones of a file.
1044
076efc7c 1045write_iolog=str Write the issued io patterns to the specified file. See
5b42a488
SH
1046 read_iolog. Specify a separate file for each job, otherwise
1047 the iologs will be interspersed and the file may be corrupt.
71bfa161 1048
076efc7c 1049read_iolog=str Open an iolog with the specified file name and replay the
71bfa161 1050 io patterns it contains. This can be used to store a
6df8adaa
JA
1051 workload and replay it sometime later. The iolog given
1052 may also be a blktrace binary file, which allows fio
1053 to replay a workload captured by blktrace. See blktrace
1054 for how to capture such logging data. For blktrace replay,
1055 the file needs to be turned into a blkparse binary data
ea3e51c3 1056 file first (blkparse <device> -o /dev/null -d file_for_fio.bin).
64bbb865
DN
1057
1058replay_no_stall=int When replaying I/O with read_iolog the default behavior
62776229
JA
1059 is to attempt to respect the time stamps within the log and
1060 replay them with the appropriate delay between IOPS. By
1061 setting this variable fio will not respect the timestamps and
1062 attempt to replay them as fast as possible while still
1063 respecting ordering. The result is the same I/O pattern to a
1064 given device, but different timings.
71bfa161 1065
d1c46c04
DN
1066replay_redirect=str While replaying I/O patterns using read_iolog the
1067 default behavior is to replay the IOPS onto the major/minor
1068 device that each IOP was recorded from. This is sometimes
1069 undesireable because on a different machine those major/minor
1070 numbers can map to a different device. Changing hardware on
1071 the same system can also result in a different major/minor
1072 mapping. Replay_redirect causes all IOPS to be replayed onto
1073 the single specified device regardless of the device it was
1074 recorded from. i.e. replay_redirect=/dev/sdc would cause all
1075 IO in the blktrace to be replayed onto /dev/sdc. This means
1076 multiple devices will be replayed onto a single, if the trace
1077 contains multiple devices. If you want multiple devices to be
1078 replayed concurrently to multiple redirected devices you must
1079 blkparse your trace into separate traces and replay them with
1080 independent fio invocations. Unfortuantely this also breaks
1081 the strict time ordering between multiple device accesses.
1082
e3cedca7 1083write_bw_log=str If given, write a bandwidth log of the jobs in this job
71bfa161 1084 file. Can be used to store data of the bandwidth of the
e0da9bc2
JA
1085 jobs in their lifetime. The included fio_generate_plots
1086 script uses gnuplot to turn these text files into nice
e3cedca7
JA
1087 graphs. See write_log_log for behaviour of given
1088 filename. For this option, the postfix is _bw.log.
71bfa161 1089
e3cedca7 1090write_lat_log=str Same as write_bw_log, except that this option stores io
02af0988
JA
1091 submission, completion, and total latencies instead. If no
1092 filename is given with this option, the default filename of
1093 "jobname_type.log" is used. Even if the filename is given,
1094 fio will still append the type of log. So if one specifies
e3cedca7
JA
1095
1096 write_lat_log=foo
1097
02af0988
JA
1098 The actual log names will be foo_slat.log, foo_slat.log,
1099 and foo_lat.log. This helps fio_generate_plot fine the logs
1100 automatically.
71bfa161 1101
c8eeb9df
JA
1102write_bw_log=str If given, write an IOPS log of the jobs in this job
1103 file. See write_bw_log.
1104
b8bc8cba
JA
1105write_iops_log=str Same as write_bw_log, but writes IOPS. If no filename is
1106 given with this option, the default filename of
1107 "jobname_type.log" is used. Even if the filename is given,
1108 fio will still append the type of log.
1109
1110log_avg_msec=int By default, fio will log an entry in the iops, latency,
1111 or bw log for every IO that completes. When writing to the
1112 disk log, that can quickly grow to a very large size. Setting
1113 this option makes fio average the each log entry over the
1114 specified period of time, reducing the resolution of the log.
1115 Defaults to 0.
1116
f7fa2653 1117lockmem=int Pin down the specified amount of memory with mlock(2). Can
71bfa161
JA
1118 potentially be used instead of removing memory or booting
1119 with less memory to simulate a smaller amount of memory.
1120
1121exec_prerun=str Before running this job, issue the command specified
1122 through system(3).
1123
1124exec_postrun=str After the job completes, issue the command specified
1125 though system(3).
1126
1127ioscheduler=str Attempt to switch the device hosting the file to the specified
1128 io scheduler before running.
1129
1130cpuload=int If the job is a CPU cycle eater, attempt to use the specified
1131 percentage of CPU cycles.
1132
1133cpuchunks=int If the job is a CPU cycle eater, split the load into
26eca2db 1134 cycles of the given time. In microseconds.
71bfa161 1135
0a839f30
JA
1136disk_util=bool Generate disk utilization statistics, if the platform
1137 supports it. Defaults to on.
1138
02af0988 1139disable_lat=bool Disable measurements of total latency numbers. Useful
9520ebb9
JA
1140 only for cutting back the number of calls to gettimeofday,
1141 as that does impact performance at really high IOPS rates.
1142 Note that to really get rid of a large amount of these
1143 calls, this option must be used with disable_slat and
1144 disable_bw as well.
1145
02af0988
JA
1146disable_clat=bool Disable measurements of completion latency numbers. See
1147 disable_lat.
1148
9520ebb9 1149disable_slat=bool Disable measurements of submission latency numbers. See
02af0988 1150 disable_slat.
9520ebb9
JA
1151
1152disable_bw=bool Disable measurements of throughput/bandwidth numbers. See
02af0988 1153 disable_lat.
9520ebb9 1154
83349190
YH
1155clat_percentiles=bool Enable the reporting of percentiles of
1156 completion latencies.
1157
1158percentile_list=float_list Overwrite the default list of percentiles
1159 for completion latencies. Each number is a floating
1160 number in the range (0,100], and the maximum length of
1161 the list is 20. Use ':' to separate the numbers, and
1162 list the numbers in ascending order. For example,
1163 --percentile_list=99.5:99.9 will cause fio to report
1164 the values of completion latency below which 99.5% and
1165 99.9% of the observed latencies fell, respectively.
1166
993bf48b
JA
1167gtod_reduce=bool Enable all of the gettimeofday() reducing options
1168 (disable_clat, disable_slat, disable_bw) plus reduce
1169 precision of the timeout somewhat to really shrink
1170 the gettimeofday() call count. With this option enabled,
1171 we only do about 0.4% of the gtod() calls we would have
1172 done if all time keeping was enabled.
1173
be4ecfdf
JA
1174gtod_cpu=int Sometimes it's cheaper to dedicate a single thread of
1175 execution to just getting the current time. Fio (and
1176 databases, for instance) are very intensive on gettimeofday()
1177 calls. With this option, you can set one CPU aside for
1178 doing nothing but logging current time to a shared memory
1179 location. Then the other threads/processes that run IO
1180 workloads need only copy that segment, instead of entering
1181 the kernel with a gettimeofday() call. The CPU set aside
1182 for doing these time calls will be excluded from other
1183 uses. Fio will manually clear it from the CPU mask of other
1184 jobs.
a696fa2a 1185
06842027 1186continue_on_error=str Normally fio will exit the job on the first observed
f2bba182
RR
1187 failure. If this option is set, fio will continue the job when
1188 there is a 'non-fatal error' (EIO or EILSEQ) until the runtime
1189 is exceeded or the I/O size specified is completed. If this
1190 option is used, there are two more stats that are appended,
1191 the total error count and the first error. The error field
1192 given in the stats is the first error that was hit during the
1193 run.
be4ecfdf 1194
06842027
SL
1195 The allowed values are:
1196
1197 none Exit on any IO or verify errors.
1198
1199 read Continue on read errors, exit on all others.
1200
1201 write Continue on write errors, exit on all others.
1202
1203 io Continue on any IO error, exit on all others.
1204
1205 verify Continue on verify errors, exit on all others.
1206
1207 all Continue on all errors.
1208
1209 0 Backward-compatible alias for 'none'.
1210
1211 1 Backward-compatible alias for 'all'.
1212
6adb38a1
JA
1213cgroup=str Add job to this control group. If it doesn't exist, it will
1214 be created. The system must have a mounted cgroup blkio
1215 mount point for this to work. If your system doesn't have it
1216 mounted, you can do so with:
a696fa2a
JA
1217
1218 # mount -t cgroup -o blkio none /cgroup
1219
a696fa2a
JA
1220cgroup_weight=int Set the weight of the cgroup to this value. See
1221 the documentation that comes with the kernel, allowed values
1222 are in the range of 100..1000.
71bfa161 1223
7de87099
VG
1224cgroup_nodelete=bool Normally fio will delete the cgroups it has created after
1225 the job completion. To override this behavior and to leave
1226 cgroups around after the job completion, set cgroup_nodelete=1.
1227 This can be useful if one wants to inspect various cgroup
1228 files after job completion. Default: false
1229
e0b0d892
JA
1230uid=int Instead of running as the invoking user, set the user ID to
1231 this value before the thread/process does any work.
1232
1233gid=int Set group ID, see uid.
1234
9e684a49
DE
1235flow_id=int The ID of the flow. If not specified, it defaults to being a
1236 global flow. See flow.
1237
1238flow=int Weight in token-based flow control. If this value is used, then
1239 there is a 'flow counter' which is used to regulate the
1240 proportion of activity between two or more jobs. fio attempts
1241 to keep this flow counter near zero. The 'flow' parameter
1242 stands for how much should be added or subtracted to the flow
1243 counter on each iteration of the main I/O loop. That is, if
1244 one job has flow=8 and another job has flow=-1, then there
1245 will be a roughly 1:8 ratio in how much one runs vs the other.
1246
1247flow_watermark=int The maximum value that the absolute value of the flow
1248 counter is allowed to reach before the job must wait for a
1249 lower value of the counter.
1250
1251flow_sleep=int The period of time, in microseconds, to wait after the flow
1252 watermark has been exceeded before retrying operations
1253
de890a1e
SL
1254In addition, there are some parameters which are only valid when a specific
1255ioengine is in use. These are used identically to normal parameters, with the
1256caveat that when used on the command line, they must come after the ioengine
1257that defines them is selected.
1258
1259[libaio] userspace_reap Normally, with the libaio engine in use, fio will use
1260 the io_getevents system call to reap newly returned events.
1261 With this flag turned on, the AIO ring will be read directly
1262 from user-space to reap events. The reaping mode is only
1263 enabled when polling for a minimum of 0 events (eg when
1264 iodepth_batch_complete=0).
1265
1266[netsplice] hostname=str
1267[net] hostname=str The host name or IP address to use for TCP or UDP based IO.
1268 If the job is a TCP listener or UDP reader, the hostname is not
1269 used and must be omitted.
1270
1271[netsplice] port=int
1272[net] port=int The TCP or UDP port to bind to or connect to.
1273
1274[netsplice] protocol=str
1275[netsplice] proto=str
1276[net] protocol=str
1277[net] proto=str The network protocol to use. Accepted values are:
1278
1279 tcp Transmission control protocol
1280 udp Unreliable datagram protocol
1281 unix UNIX domain socket
1282
1283 When the protocol is TCP or UDP, the port must also be given,
1284 as well as the hostname if the job is a TCP listener or UDP
1285 reader. For unix sockets, the normal filename option should be
1286 used and the port is invalid.
1287
1288[net] listen For TCP network connections, tell fio to listen for incoming
1289 connections rather than initiating an outgoing connection. The
1290 hostname must be omitted if this option is used.
1291
1292
71bfa161
JA
12936.0 Interpreting the output
1294---------------------------
1295
1296fio spits out a lot of output. While running, fio will display the
1297status of the jobs created. An example of that would be:
1298
73c8b082 1299Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
71bfa161
JA
1300
1301The characters inside the square brackets denote the current status of
1302each thread. The possible values (in typical life cycle order) are:
1303
1304Idle Run
1305---- ---
1306P Thread setup, but not started.
1307C Thread created.
1308I Thread initialized, waiting.
b0f65863 1309 p Thread running pre-reading file(s).
71bfa161
JA
1310 R Running, doing sequential reads.
1311 r Running, doing random reads.
1312 W Running, doing sequential writes.
1313 w Running, doing random writes.
1314 M Running, doing mixed sequential reads/writes.
1315 m Running, doing mixed random reads/writes.
1316 F Running, currently waiting for fsync()
fc6bd43c 1317 V Running, doing verification of written data.
71bfa161
JA
1318E Thread exited, not reaped by main thread yet.
1319_ Thread reaped.
1320
1321The other values are fairly self explanatory - number of threads
c9f60304
JA
1322currently running and doing io, rate of io since last check (read speed
1323listed first, then write speed), and the estimated completion percentage
1324and time for the running group. It's impossible to estimate runtime of
1325the following groups (if any).
71bfa161
JA
1326
1327When fio is done (or interrupted by ctrl-c), it will show the data for
1328each thread, group of threads, and disks in that order. For each data
1329direction, the output looks like:
1330
1331Client1 (g=0): err= 0:
35649e58 1332 write: io= 32MB, bw= 666KB/s, iops=89 , runt= 50320msec
6104ddb6
JA
1333 slat (msec): min= 0, max= 136, avg= 0.03, stdev= 1.92
1334 clat (msec): min= 0, max= 631, avg=48.50, stdev=86.82
b22989b9 1335 bw (KB/s) : min= 0, max= 1196, per=51.00%, avg=664.02, stdev=681.68
e7823a94 1336 cpu : usr=1.49%, sys=0.25%, ctx=7969, majf=0, minf=17
71619dc2 1337 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
1338 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
1339 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
30061b97 1340 issued r/w: total=0/32768, short=0/0
8abdce66
JA
1341 lat (msec): 2=1.6%, 4=0.0%, 10=3.2%, 20=12.8%, 50=38.4%, 100=24.8%,
1342 lat (msec): 250=15.2%, 500=0.0%, 750=0.0%, 1000=0.0%, >=2048=0.0%
71bfa161
JA
1343
1344The client number is printed, along with the group id and error of that
1345thread. Below is the io statistics, here for writes. In the order listed,
1346they denote:
1347
1348io= Number of megabytes io performed
1349bw= Average bandwidth rate
35649e58 1350iops= Average IOs performed per second
71bfa161 1351runt= The runtime of that thread
72fbda2a 1352 slat= Submission latency (avg being the average, stdev being the
71bfa161
JA
1353 standard deviation). This is the time it took to submit
1354 the io. For sync io, the slat is really the completion
8a35c71e 1355 latency, since queue/complete is one operation there. This
bf9a3edb 1356 value can be in milliseconds or microseconds, fio will choose
8a35c71e 1357 the most appropriate base and print that. In the example
bf9a3edb 1358 above, milliseconds is the best scale.
71bfa161
JA
1359 clat= Completion latency. Same names as slat, this denotes the
1360 time from submission to completion of the io pieces. For
1361 sync io, clat will usually be equal (or very close) to 0,
1362 as the time from submit to complete is basically just
1363 CPU time (io has already been done, see slat explanation).
1364 bw= Bandwidth. Same names as the xlat stats, but also includes
1365 an approximate percentage of total aggregate bandwidth
1366 this thread received in this group. This last value is
1367 only really useful if the threads in this group are on the
1368 same disk, since they are then competing for disk access.
1369cpu= CPU usage. User and system time, along with the number
e7823a94
JA
1370 of context switches this thread went through, usage of
1371 system and user time, and finally the number of major
1372 and minor page faults.
71619dc2
JA
1373IO depths= The distribution of io depths over the job life time. The
1374 numbers are divided into powers of 2, so for example the
1375 16= entries includes depths up to that value but higher
1376 than the previous entry. In other words, it covers the
1377 range from 16 to 31.
838bc709
JA
1378IO submit= How many pieces of IO were submitting in a single submit
1379 call. Each entry denotes that amount and below, until
1380 the previous entry - eg, 8=100% mean that we submitted
1381 anywhere in between 5-8 ios per submit call.
1382IO complete= Like the above submit number, but for completions instead.
30061b97
JA
1383IO issued= The number of read/write requests issued, and how many
1384 of them were short.
ec118304
JA
1385IO latencies= The distribution of IO completion latencies. This is the
1386 time from when IO leaves fio and when it gets completed.
1387 The numbers follow the same pattern as the IO depths,
1388 meaning that 2=1.6% means that 1.6% of the IO completed
8abdce66
JA
1389 within 2 msecs, 20=12.8% means that 12.8% of the IO
1390 took more than 10 msecs, but less than (or equal to) 20 msecs.
71bfa161
JA
1391
1392After each client has been listed, the group statistics are printed. They
1393will look like this:
1394
1395Run status group 0 (all jobs):
b22989b9
JA
1396 READ: io=64MB, aggrb=22178, minb=11355, maxb=11814, mint=2840msec, maxt=2955msec
1397 WRITE: io=64MB, aggrb=1302, minb=666, maxb=669, mint=50093msec, maxt=50320msec
71bfa161
JA
1398
1399For each data direction, it prints:
1400
1401io= Number of megabytes io performed.
1402aggrb= Aggregate bandwidth of threads in this group.
1403minb= The minimum average bandwidth a thread saw.
1404maxb= The maximum average bandwidth a thread saw.
1405mint= The smallest runtime of the threads in that group.
1406maxt= The longest runtime of the threads in that group.
1407
1408And finally, the disk statistics are printed. They will look like this:
1409
1410Disk stats (read/write):
1411 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
1412
1413Each value is printed for both reads and writes, with reads first. The
1414numbers denote:
1415
1416ios= Number of ios performed by all groups.
1417merge= Number of merges io the io scheduler.
1418ticks= Number of ticks we kept the disk busy.
1419io_queue= Total time spent in the disk queue.
1420util= The disk utilization. A value of 100% means we kept the disk
1421 busy constantly, 50% would be a disk idling half of the time.
1422
1423
14247.0 Terse output
1425----------------
1426
1427For scripted usage where you typically want to generate tables or graphs
6af019c9 1428of the results, fio can output the results in a semicolon separated format.
71bfa161
JA
1429The format is one long line of values, such as:
1430
562c2d2f
DN
14312;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%
1432A description of this job goes here.
1433
1434The job description (if provided) follows on a second line.
71bfa161 1435
525c2bfa
JA
1436To enable terse output, use the --minimal command line option. The first
1437value is the version of the terse output format. If the output has to
1438be changed for some reason, this number will be incremented by 1 to
1439signify that change.
6820cb3b 1440
71bfa161
JA
1441Split up, the format is as follows:
1442
5e726d0a 1443 terse version, fio version, jobname, groupid, error
71bfa161 1444 READ status:
312b4af2 1445 Total IO (KB), bandwidth (KB/sec), IOPS, runtime (msec)
71bfa161
JA
1446 Submission latency: min, max, mean, deviation
1447 Completion latency: min, max, mean, deviation
1db92cb6 1448 Completion latency percentiles: 20 fields (see below)
525c2bfa 1449 Total latency: min, max, mean, deviation
6c219763 1450 Bw: min, max, aggregate percentage of total, mean, deviation
71bfa161 1451 WRITE status:
312b4af2 1452 Total IO (KB), bandwidth (KB/sec), IOPS, runtime (msec)
71bfa161
JA
1453 Submission latency: min, max, mean, deviation
1454 Completion latency: min, max, mean, deviation
1db92cb6 1455 Completion latency percentiles: 20 fields (see below)
525c2bfa 1456 Total latency: min, max, mean, deviation
6c219763 1457 Bw: min, max, aggregate percentage of total, mean, deviation
046ee302 1458 CPU usage: user, system, context switches, major faults, minor faults
2270890c 1459 IO depths: <=1, 2, 4, 8, 16, 32, >=64
562c2d2f
DN
1460 IO latencies microseconds: <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
1461 IO latencies milliseconds: <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
f2f788dd
JA
1462 Disk utilization: Disk name, Read ios, write ios,
1463 Read merges, write merges,
1464 Read ticks, write ticks,
3d7cd9b4 1465 Time spent in queue, disk utilization percentage
562c2d2f
DN
1466 Additional Info (dependant on continue_on_error, default off): total # errors, first error code
1467
f42195a3 1468 Additional Info (dependant on description being set): Text description
25c8b9d7 1469
1db92cb6
JA
1470Completion latency percentiles can be a grouping of up to 20 sets, so
1471for the terse output fio writes all of them. Each field will look like this:
1472
1473 1.00%=6112
1474
1475which is the Xth percentile, and the usec latency associated with it.
1476
f2f788dd
JA
1477For disk utilization, all disks used by fio are shown. So for each disk
1478there will be a disk utilization section.
1479
25c8b9d7
PD
1480
14818.0 Trace file format
1482---------------------
1483There are two trace file format that you can encounter. The older (v1) format
1484is unsupported since version 1.20-rc3 (March 2008). It will still be described
1485below in case that you get an old trace and want to understand it.
1486
1487In any case the trace is a simple text file with a single action per line.
1488
1489
14908.1 Trace file format v1
1491------------------------
1492Each line represents a single io action in the following format:
1493
1494rw, offset, length
1495
1496where rw=0/1 for read/write, and the offset and length entries being in bytes.
1497
1498This format is not supported in Fio versions => 1.20-rc3.
1499
1500
15018.2 Trace file format v2
1502------------------------
1503The second version of the trace file format was added in Fio version 1.17.
1504It allows to access more then one file per trace and has a bigger set of
1505possible file actions.
1506
1507The first line of the trace file has to be:
1508
1509fio version 2 iolog
1510
1511Following this can be lines in two different formats, which are described below.
1512
1513The file management format:
1514
1515filename action
1516
1517The filename is given as an absolute path. The action can be one of these:
1518
1519add Add the given filename to the trace
1520open Open the file with the given filename. The filename has to have
1521 been added with the add action before.
1522close Close the file with the given filename. The file has to have been
1523 opened before.
1524
1525
1526The file io action format:
1527
1528filename action offset length
1529
1530The filename is given as an absolute path, and has to have been added and opened
1531before it can be used with this format. The offset and length are given in
1532bytes. The action can be one of these:
1533
1534wait Wait for 'offset' microseconds. Everything below 100 is discarded.
1535read Read 'length' bytes beginning from 'offset'
1536write Write 'length' bytes beginning from 'offset'
1537sync fsync() the file
1538datasync fdatasync() the file
1539trim trim the given file from the given 'offset' for 'length' bytes