workqueue: move private exit code to caller
[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'
550b1db6 84parameter described in the parameter section.
71bfa161 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
df5ad464
AK
162When fio is utilized as a basis of any reasonably large test suite, it might be
163desirable to share a set of standardized settings across multiple job files.
164Instead of copy/pasting such settings, any section may pull in an external
165.fio file with 'include filename' directive, as in the following example:
166
167; -- start job file including.fio --
168[global]
169filename=/tmp/test
170filesize=1m
171include glob-include.fio
172
173[test]
174rw=randread
175bs=4k
176time_based=1
177runtime=10
178include test-include.fio
179; -- end job file including.fio --
180
181; -- start job file glob-include.fio --
182thread=1
183group_reporting=1
184; -- end job file glob-include.fio --
185
186; -- start job file test-include.fio --
187ioengine=libaio
188iodepth=4
189; -- end job file test-include.fio --
190
191Settings pulled into a section apply to that section only (except global
192section). Include directives may be nested in that any included file may
73568e1b
JA
193contain further include directive(s). Include files may not contain []
194sections.
df5ad464
AK
195
196
74929ac2
JA
1974.1 Environment variables
198-------------------------
199
3c54bc46 200fio also supports environment variable expansion in job files. Any
4fbe1860 201sub-string of the form "${VARNAME}" as part of an option value (in other
3c54bc46
AC
202words, on the right of the `='), will be expanded to the value of the
203environment variable called VARNAME. If no such environment variable
204is defined, or VARNAME is the empty string, the empty string will be
205substituted.
206
207As an example, let's look at a sample fio invocation and job file:
208
209$ SIZE=64m NUMJOBS=4 fio jobfile.fio
210
211; -- start job file --
212[random-writers]
213rw=randwrite
214size=${SIZE}
215numjobs=${NUMJOBS}
216; -- end job file --
217
218This will expand to the following equivalent job file at runtime:
219
220; -- start job file --
221[random-writers]
222rw=randwrite
223size=64m
224numjobs=4
225; -- end job file --
226
71bfa161
JA
227fio ships with a few example job files, you can also look there for
228inspiration.
229
74929ac2
JA
2304.2 Reserved keywords
231---------------------
232
233Additionally, fio has a set of reserved keywords that will be replaced
234internally with the appropriate value. Those keywords are:
235
236$pagesize The architecture page size of the running system
237$mb_memory Megabytes of total memory in the system
238$ncpus Number of online available CPUs
239
240These can be used on the command line or in the job file, and will be
241automatically substituted with the current system values when the job
892a6ffc
JA
242is run. Simple math is also supported on these keywords, so you can
243perform actions like:
244
245size=8*$mb_memory
246
247and get that properly expanded to 8 times the size of memory in the
248machine.
74929ac2 249
71bfa161
JA
250
2515.0 Detailed list of parameters
252-------------------------------
253
254This section describes in details each parameter associated with a job.
255Some parameters take an option of a given type, such as an integer or
d59aa780
JA
256a string. Anywhere a numeric value is required, an arithmetic expression
257may be used, provided it is surrounded by parentheses. Supported operators
258are:
259
260 addition (+)
261 subtraction (-)
262 multiplication (*)
263 division (/)
264 modulus (%)
265 exponentiation (^)
266
267For time values in expressions, units are microseconds by default. This is
268different than for time values not in expressions (not enclosed in
269parentheses). The following types are used:
71bfa161
JA
270
271str String. This is a sequence of alpha characters.
b09da8fa 272time Integer with possible time suffix. In seconds unless otherwise
e417fd66 273 specified, use eg 10m for 10 minutes. Accepts s/m/h for seconds,
0de5b26f
JA
274 minutes, and hours, and accepts 'ms' (or 'msec') for milliseconds,
275 and 'us' (or 'usec') for microseconds.
b09da8fa
JA
276int SI integer. A whole number value, which may contain a suffix
277 describing the base of the number. Accepted suffixes are k/m/g/t/p,
278 meaning kilo, mega, giga, tera, and peta. The suffix is not case
57fc29fa
JA
279 sensitive, and you may also include trailing 'b' (eg 'kb' is the same
280 as 'k'). So if you want to specify 4096, you could either write
b09da8fa 281 out '4096' or just give 4k. The suffixes signify base 2 values, so
57fc29fa
JA
282 1024 is 1k and 1024k is 1m and so on, unless the suffix is explicitly
283 set to a base 10 value using 'kib', 'mib', 'gib', etc. If that is the
284 case, then 1000 is used as the multiplier. This can be handy for
285 disks, since manufacturers generally use base 10 values when listing
286 the capacity of a drive. If the option accepts an upper and lower
287 range, use a colon ':' or minus '-' to separate such values. May also
288 include a prefix to indicate numbers base. If 0x is used, the number
289 is assumed to be hexadecimal. See irange.
71bfa161
JA
290bool Boolean. Usually parsed as an integer, however only defined for
291 true and false (1 and 0).
b09da8fa 292irange Integer range with suffix. Allows value range to be given, such
bf9a3edb 293 as 1024-4096. A colon may also be used as the separator, eg
0c9baf91
JA
294 1k:4k. If the option allows two sets of ranges, they can be
295 specified with a ',' or '/' delimiter: 1k-4k/8k-32k. Also see
f7fa2653 296 int.
83349190 297float_list A list of floating numbers, separated by a ':' character.
71bfa161
JA
298
299With the above in mind, here follows the complete list of fio job
300parameters.
301
302name=str ASCII name of the job. This may be used to override the
303 name printed by fio for this job. Otherwise the job
c2b1e753 304 name is used. On the command line this parameter has the
6c219763 305 special purpose of also signaling the start of a new
c2b1e753 306 job.
71bfa161 307
61697c37
JA
308description=str Text description of the job. Doesn't do anything except
309 dump this text description when this job is run. It's
310 not parsed.
311
3776041e 312directory=str Prefix filenames with this directory. Used to place files
67445b63
JA
313 in a different location than "./". See the 'filename' option
314 for escaping certain characters.
71bfa161
JA
315
316filename=str Fio normally makes up a filename based on the job name,
317 thread number, and file number. If you want to share
318 files between threads in a job or several jobs, specify
ed92ac0c 319 a filename for each of them to override the default. If
414c2a3e 320 the ioengine used is 'net', the filename is the host, port,
0fd666bf 321 and protocol to use in the format of =host,port,protocol.
414c2a3e
JA
322 See ioengine=net for more. If the ioengine is file based, you
323 can specify a number of files by separating the names with a
324 ':' colon. So if you wanted a job to open /dev/sda and /dev/sdb
325 as the two working files, you would use
30a4588a
JA
326 filename=/dev/sda:/dev/sdb. On Windows, disk devices are
327 accessed as \\.\PhysicalDrive0 for the first device,
328 \\.\PhysicalDrive1 for the second etc. Note: Windows and
329 FreeBSD prevent write access to areas of the disk containing
330 in-use data (e.g. filesystems).
331 If the wanted filename does need to include a colon, then
332 escape that with a '\' character. For instance, if the filename
333 is "/dev/dsk/foo@3,0:c", then you would use
334 filename="/dev/dsk/foo@3,0\:c". '-' is a reserved name, meaning
335 stdin or stdout. Which of the two depends on the read/write
336 direction set.
71bfa161 337
de98bd30
JA
338filename_format=str
339 If sharing multiple files between jobs, it is usually necessary
340 to have fio generate the exact names that you want. By default,
341 fio will name a file based on the default file format
342 specification of jobname.jobnumber.filenumber. With this
343 option, that can be customized. Fio will recognize and replace
344 the following keywords in this string:
345
346 $jobname
347 The name of the worker thread or process.
348
349 $jobnum
350 The incremental number of the worker thread or
351 process.
352
353 $filenum
354 The incremental number of the file for that worker
355 thread or process.
356
357 To have dependent jobs share a set of files, this option can
358 be set to have fio generate filenames that are shared between
359 the two. For instance, if testfiles.$filenum is specified,
360 file number 4 for any job will be named testfiles.4. The
361 default of $jobname.$jobnum.$filenum will be used if
362 no other format specifier is given.
363
bbf6b540
JA
364opendir=str Tell fio to recursively add any file it can find in this
365 directory and down the file system tree.
366
3776041e 367lockfile=str Fio defaults to not locking any files before it does
4d4e80f2
JA
368 IO to them. If a file or file descriptor is shared, fio
369 can serialize IO to that file to make the end result
370 consistent. This is usual for emulating real workloads that
371 share files. The lock modes are:
372
373 none No locking. The default.
374 exclusive Only one thread/process may do IO,
375 excluding all others.
376 readwrite Read-write locking on the file. Many
377 readers may access the file at the
378 same time, but writes get exclusive
379 access.
380
d3aad8f2 381readwrite=str
71bfa161
JA
382rw=str Type of io pattern. Accepted values are:
383
384 read Sequential reads
385 write Sequential writes
386 randwrite Random writes
387 randread Random reads
10b023db 388 rw,readwrite Sequential mixed reads and writes
71bfa161 389 randrw Random mixed reads and writes
82a90686
JA
390 trimwrite Mixed trims and writes. Blocks will be
391 trimmed first, then written to.
71bfa161
JA
392
393 For the mixed io types, the default is to split them 50/50.
394 For certain types of io the result may still be skewed a bit,
211097b2 395 since the speed may be different. It is possible to specify
38dad62d 396 a number of IO's to do before getting a new offset, this is
892ea9bd 397 done by appending a ':<nr>' to the end of the string given.
38dad62d 398 For a random read, it would look like 'rw=randread:8' for
059b0802 399 passing in an offset modifier with a value of 8. If the
ddb754db 400 suffix is used with a sequential IO pattern, then the value
059b0802
JA
401 specified will be added to the generated offset for each IO.
402 For instance, using rw=write:4k will skip 4k for every
403 write. It turns sequential IO into sequential IO with holes.
404 See the 'rw_sequencer' option.
38dad62d
JA
405
406rw_sequencer=str If an offset modifier is given by appending a number to
407 the rw=<str> line, then this option controls how that
408 number modifies the IO offset being generated. Accepted
409 values are:
410
411 sequential Generate sequential offset
412 identical Generate the same offset
413
414 'sequential' is only useful for random IO, where fio would
415 normally generate a new random offset for every IO. If you
416 append eg 8 to randread, you would get a new random offset for
211097b2
JA
417 every 8 IO's. The result would be a seek for only every 8
418 IO's, instead of for every IO. Use rw=randread:8 to specify
38dad62d
JA
419 that. As sequential IO is already sequential, setting
420 'sequential' for that would not result in any differences.
421 'identical' behaves in a similar fashion, except it sends
422 the same offset 8 number of times before generating a new
423 offset.
71bfa161 424
90fef2d1
JA
425kb_base=int The base unit for a kilobyte. The defacto base is 2^10, 1024.
426 Storage manufacturers like to use 10^3 or 1000 as a base
427 ten unit instead, for obvious reasons. Allow values are
428 1024 or 1000, with 1024 being the default.
429
771e58be
JA
430unified_rw_reporting=bool Fio normally reports statistics on a per
431 data direction basis, meaning that read, write, and trim are
432 accounted and reported separately. If this option is set,
433 the fio will sum the results and report them as "mixed"
434 instead.
435
ee738499
JA
436randrepeat=bool For random IO workloads, seed the generator in a predictable
437 way so that results are repeatable across repetitions.
40fe5e7b 438 Defaults to true.
ee738499 439
04778baf
JA
440randseed=int Seed the random number generators based on this seed value, to
441 be able to control what sequence of output is being generated.
442 If not set, the random sequence depends on the randrepeat
443 setting.
444
a596f047
EG
445fallocate=str Whether pre-allocation is performed when laying down files.
446 Accepted values are:
447
448 none Do not pre-allocate space
449 posix Pre-allocate via posix_fallocate()
450 keep Pre-allocate via fallocate() with
451 FALLOC_FL_KEEP_SIZE set
452 0 Backward-compatible alias for 'none'
453 1 Backward-compatible alias for 'posix'
454
455 May not be available on all supported platforms. 'keep' is only
456 available on Linux.If using ZFS on Solaris this must be set to
457 'none' because ZFS doesn't support it. Default: 'posix'.
7bc8c2cf 458
d2f3ac35
JA
459fadvise_hint=bool By default, fio will use fadvise() to advise the kernel
460 on what IO patterns it is likely to issue. Sometimes you
461 want to test specific IO patterns without telling the
462 kernel about it, in which case you can disable this option.
463 If set, fio will use POSIX_FADV_SEQUENTIAL for sequential
464 IO and POSIX_FADV_RANDOM for random IO.
465
37659335
JA
466fadvise_stream=int Notify the kernel what write stream ID to place these
467 writes under. Only supported on Linux. Note, this option
468 may change going forward.
469
f7fa2653 470size=int The total size of file io for this job. Fio will run until
7616cafe 471 this many bytes has been transferred, unless runtime is
a4d3b4db
JA
472 limited by other options (such as 'runtime', for instance,
473 or increased/decreased by 'io_size'). Unless specific nrfiles
474 and filesize options are given, fio will divide this size
475 between the available files specified by the job. If not set,
476 fio will use the full size of the given files or devices.
477 If the files do not exist, size must be given. It is also
478 possible to give size as a percentage between 1 and 100. If
479 size=20% is given, fio will use 20% of the full size of the
480 given files or devices.
481
482io_size=int
77731b29
JA
483io_limit=int Normally fio operates within the region set by 'size', which
484 means that the 'size' option sets both the region and size of
485 IO to be performed. Sometimes that is not what you want. With
486 this option, it is possible to define just the amount of IO
487 that fio should do. For instance, if 'size' is set to 20G and
a4d3b4db
JA
488 'io_size' is set to 5G, fio will perform IO within the first
489 20G but exit when 5G have been done. The opposite is also
490 possible - if 'size' is set to 20G, and 'io_size' is set to
491 40G, then fio will do 40G of IO within the 0..20G region.
77731b29 492
f7fa2653 493filesize=int Individual file sizes. May be a range, in which case fio
9c60ce64
JA
494 will select sizes for files at random within the given range
495 and limited to 'size' in total (if that is given). If not
496 given, each created file is the same size.
497
bedc9dc2
JA
498file_append=bool Perform IO after the end of the file. Normally fio will
499 operate within the size of a file. If this option is set, then
500 fio will append to the file instead. This has identical
0aae4ce7
JA
501 behavior to setting offset to the size of a file. This option
502 is ignored on non-regular files.
bedc9dc2 503
74586c1e
JA
504fill_device=bool
505fill_fs=bool Sets size to something really large and waits for ENOSPC (no
aa31f1f1 506 space left on device) as the terminating condition. Only makes
de98bd30 507 sense with sequential write. For a read workload, the mount
4f12432e
JA
508 point will be filled first then IO started on the result. This
509 option doesn't make sense if operating on a raw device node,
510 since the size of that is already known by the file system.
511 Additionally, writing beyond end-of-device will not return
512 ENOSPC there.
aa31f1f1 513
f7fa2653
JA
514blocksize=int
515bs=int The block size used for the io units. Defaults to 4k. Values
516 can be given for both read and writes. If a single int is
517 given, it will apply to both. If a second int is specified
f90eff5a 518 after a comma, it will apply to writes only. In other words,
d9472271
JA
519 the format is either bs=read_and_write or bs=read,write,trim.
520 bs=4k,8k will thus use 4k blocks for reads, 8k blocks for
521 writes, and 8k for trims. You can terminate the list with
522 a trailing comma. bs=4k,8k, would use the default value for
523 trims.. If you only wish to set the write size, you
787f7e95
JA
524 can do so by passing an empty read size - bs=,8k will set
525 8k for writes and leave the read default value.
a00735e6 526
2b7a01d0
JA
527blockalign=int
528ba=int At what boundary to align random IO offsets. Defaults to
529 the same as 'blocksize' the minimum blocksize given.
530 Minimum alignment is typically 512b for using direct IO,
531 though it usually depends on the hardware block size. This
532 option is mutually exclusive with using a random map for
533 files, so it will turn off that option.
534
d3aad8f2 535blocksize_range=irange
71bfa161
JA
536bsrange=irange Instead of giving a single block size, specify a range
537 and fio will mix the issued io block sizes. The issued
538 io unit will always be a multiple of the minimum value
f90eff5a
JA
539 given (also see bs_unaligned). Applies to both reads and
540 writes, however a second range can be given after a comma.
541 See bs=.
a00735e6 542
564ca972
JA
543bssplit=str Sometimes you want even finer grained control of the
544 block sizes issued, not just an even split between them.
545 This option allows you to weight various block sizes,
546 so that you are able to define a specific amount of
547 block sizes issued. The format for this option is:
548
549 bssplit=blocksize/percentage:blocksize/percentage
550
551 for as many block sizes as needed. So if you want to define
552 a workload that has 50% 64k blocks, 10% 4k blocks, and
553 40% 32k blocks, you would write:
554
555 bssplit=4k/10:64k/50:32k/40
556
557 Ordering does not matter. If the percentage is left blank,
558 fio will fill in the remaining values evenly. So a bssplit
559 option like this one:
560
561 bssplit=4k/50:1k/:32k/
562
563 would have 50% 4k ios, and 25% 1k and 32k ios. The percentages
564 always add up to 100, if bssplit is given a range that adds
565 up to more, it will error out.
566
720e84ad
JA
567 bssplit also supports giving separate splits to reads and
568 writes. The format is identical to what bs= accepts. You
569 have to separate the read and write parts with a comma. So
570 if you want a workload that has 50% 2k reads and 50% 4k reads,
571 while having 90% 4k writes and 10% 8k writes, you would
572 specify:
573
892ea9bd 574 bssplit=2k/50:4k/50,4k/90:8k/10
720e84ad 575
d3aad8f2 576blocksize_unaligned
690adba3
JA
577bs_unaligned If this option is given, any byte size value within bsrange
578 may be used as a block range. This typically wont work with
579 direct IO, as that normally requires sector alignment.
71bfa161 580
6aca9b3d
JA
581bs_is_seq_rand If this option is set, fio will use the normal read,write
582 blocksize settings as sequential,random instead. Any random
583 read or write will use the WRITE blocksize settings, and any
584 sequential read or write will use the READ blocksize setting.
585
e9459e5a
JA
586zero_buffers If this option is given, fio will init the IO buffers to
587 all zeroes. The default is to fill them with random data.
588
5973cafb
JA
589refill_buffers If this option is given, fio will refill the IO buffers
590 on every submit. The default is to only fill it at init
591 time and reuse that data. Only makes sense if zero_buffers
41ccd845
JA
592 isn't specified, naturally. If data verification is enabled,
593 refill_buffers is also automatically enabled.
5973cafb 594
fd68418e
JA
595scramble_buffers=bool If refill_buffers is too costly and the target is
596 using data deduplication, then setting this option will
597 slightly modify the IO buffer contents to defeat normal
598 de-dupe attempts. This is not enough to defeat more clever
599 block compression attempts, but it will stop naive dedupe of
600 blocks. Default: true.
601
c5751c62
JA
602buffer_compress_percentage=int If this is set, then fio will attempt to
603 provide IO buffer content (on WRITEs) that compress to
604 the specified level. Fio does this by providing a mix of
d1af2894
JA
605 random data and a fixed pattern. The fixed pattern is either
606 zeroes, or the pattern specified by buffer_pattern. If the
607 pattern option is used, it might skew the compression ratio
608 slightly. Note that this is per block size unit, for file/disk
609 wide compression level that matches this setting, you'll also
610 want to set refill_buffers.
c5751c62
JA
611
612buffer_compress_chunk=int See buffer_compress_percentage. This
613 setting allows fio to manage how big the ranges of random
614 data and zeroed data is. Without this set, fio will
615 provide buffer_compress_percentage of blocksize random
616 data, followed by the remaining zeroed. With this set
617 to some chunk size smaller than the block size, fio can
618 alternate random and zeroed data throughout the IO
619 buffer.
620
5c94b008
JA
621buffer_pattern=str If set, fio will fill the io buffers with this
622 pattern. If not set, the contents of io buffers is defined by
623 the other options related to buffer contents. The setting can
624 be any pattern of bytes, and can be prefixed with 0x for hex
625 values. It may also be a string, where the string must then
61b9861d
RP
626 be wrapped with "", e.g.:
627
628 buffer_pattern="abcd"
629 or
630 buffer_pattern=-12
631 or
632 buffer_pattern=0xdeadface
633
634 Also you can combine everything together in any order:
635 buffer_pattern=0xdeadface"abcd"-12
5c94b008
JA
636
637dedupe_percentage=int If set, fio will generate this percentage of
638 identical buffers when writing. These buffers will be
639 naturally dedupable. The contents of the buffers depend on
640 what other buffer compression settings have been set. It's
641 possible to have the individual buffers either fully
642 compressible, or not at all. This option only controls the
643 distribution of unique buffers.
ce35b1ec 644
71bfa161
JA
645nrfiles=int Number of files to use for this job. Defaults to 1.
646
390b1537
JA
647openfiles=int Number of files to keep open at the same time. Defaults to
648 the same as nrfiles, can be set smaller to limit the number
649 simultaneous opens.
650
5af1c6f3
JA
651file_service_type=str Defines how fio decides which file from a job to
652 service next. The following types are defined:
653
654 random Just choose a file at random.
655
656 roundrobin Round robin over open files. This
657 is the default.
658
a086c257
JA
659 sequential Finish one file before moving on to
660 the next. Multiple files can still be
661 open depending on 'openfiles'.
662
1907dbc6
JA
663 The string can have a number appended, indicating how
664 often to switch to a new file. So if option random:4 is
665 given, fio will switch to a new random file after 4 ios
666 have been issued.
667
71bfa161
JA
668ioengine=str Defines how the job issues io to the file. The following
669 types are defined:
670
671 sync Basic read(2) or write(2) io. lseek(2) is
672 used to position the io location.
673
a31041ea 674 psync Basic pread(2) or pwrite(2) io.
675
e05af9e5 676 vsync Basic readv(2) or writev(2) IO.
1d2af02a 677
a46c5e01
JA
678 psyncv Basic preadv(2) or pwritev(2) IO.
679
15d182aa
JA
680 libaio Linux native asynchronous io. Note that Linux
681 may only support queued behaviour with
682 non-buffered IO (set direct=1 or buffered=0).
de890a1e 683 This engine defines engine specific options.
71bfa161
JA
684
685 posixaio glibc posix asynchronous io.
686
417f0068
JA
687 solarisaio Solaris native asynchronous io.
688
03e20d68
BC
689 windowsaio Windows native asynchronous io.
690
71bfa161
JA
691 mmap File is memory mapped and data copied
692 to/from using memcpy(3).
693
694 splice splice(2) is used to transfer the data and
695 vmsplice(2) to transfer data from user
696 space to the kernel.
697
d0ff85df
JA
698 syslet-rw Use the syslet system calls to make
699 regular read/write async.
700
71bfa161 701 sg SCSI generic sg v3 io. May either be
6c219763 702 synchronous using the SG_IO ioctl, or if
71bfa161
JA
703 the target is an sg character device
704 we use read(2) and write(2) for asynchronous
705 io.
706
a94ea28b
JA
707 null Doesn't transfer any data, just pretends
708 to. This is mainly used to exercise fio
709 itself and for debugging/testing purposes.
710
ed92ac0c 711 net Transfer over the network to given host:port.
de890a1e
SL
712 Depending on the protocol used, the hostname,
713 port, listen and filename options are used to
714 specify what sort of connection to make, while
715 the protocol option determines which protocol
716 will be used.
717 This engine defines engine specific options.
ed92ac0c 718
9cce02e8
JA
719 netsplice Like net, but uses splice/vmsplice to
720 map data and send/receive.
de890a1e 721 This engine defines engine specific options.
9cce02e8 722
53aec0a4 723 cpuio Doesn't transfer any data, but burns CPU
ba0fbe10
JA
724 cycles according to the cpuload= and
725 cpucycle= options. Setting cpuload=85
726 will cause that job to do nothing but burn
36ecec83
GP
727 85% of the CPU. In case of SMP machines,
728 use numjobs=<no_of_cpu> to get desired CPU
729 usage, as the cpuload only loads a single
730 CPU at the desired rate.
ba0fbe10 731
e9a1806f
JA
732 guasi The GUASI IO engine is the Generic Userspace
733 Asyncronous Syscall Interface approach
734 to async IO. See
735
736 http://www.xmailserver.org/guasi-lib.html
737
738 for more info on GUASI.
739
21b8aee8 740 rdma The RDMA I/O engine supports both RDMA
eb52fa3f
BVA
741 memory semantics (RDMA_WRITE/RDMA_READ) and
742 channel semantics (Send/Recv) for the
743 InfiniBand, RoCE and iWARP protocols.
21b8aee8 744
b861be9f
JA
745 falloc IO engine that does regular fallocate to
746 simulate data transfer as fio ioengine.
747 DDIR_READ does fallocate(,mode = keep_size,)
748 DDIR_WRITE does fallocate(,mode = 0)
749 DDIR_TRIM does fallocate(,mode = punch_hole)
d54fce84
DM
750
751 e4defrag IO engine that does regular EXT4_IOC_MOVE_EXT
b861be9f
JA
752 ioctls to simulate defragment activity in
753 request to DDIR_WRITE event
754
755 rbd IO engine supporting direct access to Ceph
756 Rados Block Devices (RBD) via librbd without
757 the need to use the kernel rbd driver. This
758 ioengine defines engine specific options.
759
760 gfapi Using Glusterfs libgfapi sync interface to
761 direct access to Glusterfs volumes without
762 options.
763
764 gfapi_async Using Glusterfs libgfapi async interface
765 to direct access to Glusterfs volumes without
766 having to go through FUSE. This ioengine
767 defines engine specific options.
0981fd71 768
b74e419e
MM
769 libhdfs Read and write through Hadoop (HDFS).
770 The 'filename' option is used to specify host,
771 port of the hdfs name-node to connect. This
772 engine interprets offsets a little
773 differently. In HDFS, files once created
774 cannot be modified. So random writes are not
775 possible. To imitate this, libhdfs engine
776 expects bunch of small files to be created
777 over HDFS, and engine will randomly pick a
778 file out of those files based on the offset
779 generated by fio backend. (see the example
780 job file to create such files, use rw=write
781 option). Please note, you might want to set
782 necessary environment variables to work with
783 hdfs/libhdfs properly.
1b10477b 784
65fa28ca
DE
785 mtd Read, write and erase an MTD character device
786 (e.g., /dev/mtd0). Discards are treated as
787 erases. Depending on the underlying device
788 type, the I/O may have to go in a certain
789 pattern, e.g., on NAND, writing sequentially
790 to erase blocks and discarding before
791 overwriting. The writetrim mode works well
792 for this constraint.
793
8a7bd877
JA
794 external Prefix to specify loading an external
795 IO engine object file. Append the engine
796 filename, eg ioengine=external:/tmp/foo.o
797 to load ioengine foo.o in /tmp.
798
71bfa161
JA
799iodepth=int This defines how many io units to keep in flight against
800 the file. The default is 1 for each file defined in this
801 job, can be overridden with a larger value for higher
ee72ca09
JA
802 concurrency. Note that increasing iodepth beyond 1 will not
803 affect synchronous ioengines (except for small degress when
9b836561 804 verify_async is in use). Even async engines may impose OS
ee72ca09
JA
805 restrictions causing the desired depth not to be achieved.
806 This may happen on Linux when using libaio and not setting
807 direct=1, since buffered IO is not async on that OS. Keep an
808 eye on the IO depth distribution in the fio output to verify
809 that the achieved depth is as expected. Default: 1.
71bfa161 810
4950421a 811iodepth_batch_submit=int
cb5ab512 812iodepth_batch=int This defines how many pieces of IO to submit at once.
89e820f6
JA
813 It defaults to 1 which means that we submit each IO
814 as soon as it is available, but can be raised to submit
e63a0b2f
RP
815 bigger batches of IO at the time. If it is set to 0 the iodepth
816 value will be used.
cb5ab512 817
82407585 818iodepth_batch_complete_min=int
4950421a
JA
819iodepth_batch_complete=int This defines how many pieces of IO to retrieve
820 at once. It defaults to 1 which means that we'll ask
821 for a minimum of 1 IO in the retrieval process from
822 the kernel. The IO retrieval will go on until we
823 hit the limit set by iodepth_low. If this variable is
824 set to 0, then fio will always check for completed
825 events before queuing more IO. This helps reduce
826 IO latency, at the cost of more retrieval system calls.
827
82407585
RP
828iodepth_batch_complete_max=int This defines maximum pieces of IO to
829 retrieve at once. This variable should be used along with
830 iodepth_batch_complete_min=int variable, specifying the range
831 of min and max amount of IO which should be retrieved. By default
832 it is equal to iodepth_batch_complete_min value.
833
834 Example #1:
835
836 iodepth_batch_complete_min=1
837 iodepth_batch_complete_max=<iodepth>
838
839 which means that we will retrieve at leat 1 IO and up to the
840 whole submitted queue depth. If none of IO has been completed
841 yet, we will wait.
842
843 Example #2:
844
845 iodepth_batch_complete_min=0
846 iodepth_batch_complete_max=<iodepth>
847
848 which means that we can retrieve up to the whole submitted
849 queue depth, but if none of IO has been completed yet, we will
850 NOT wait and immediately exit the system call. In this example
851 we simply do polling.
852
e916b390
JA
853iodepth_low=int The low water mark indicating when to start filling
854 the queue again. Defaults to the same as iodepth, meaning
855 that fio will attempt to keep the queue full at all times.
856 If iodepth is set to eg 16 and iodepth_low is set to 4, then
857 after fio has filled the queue of 16 requests, it will let
858 the depth drain down to 4 before starting to fill it again.
859
1ad01bd1
JA
860io_submit_mode=str This option controls how fio submits the IO to
861 the IO engine. The default is 'inline', which means that the
862 fio job threads submit and reap IO directly. If set to
863 'offload', the job threads will offload IO submission to a
864 dedicated pool of IO threads. This requires some coordination
865 and thus has a bit of extra overhead, especially for lower
866 queue depth IO where it can increase latencies. The benefit
867 is that fio can manage submission rates independently of
868 the device completion rates. This avoids skewed latency
869 reporting if IO gets back up on the device side (the
870 coordinated omission problem).
871
71bfa161 872direct=bool If value is true, use non-buffered io. This is usually
9b836561 873 O_DIRECT. Note that ZFS on Solaris doesn't support direct io.
93bcfd20 874 On Windows the synchronous ioengines don't support direct io.
76a43db4 875
d01612f3
CM
876atomic=bool If value is true, attempt to use atomic direct IO. Atomic
877 writes are guaranteed to be stable once acknowledged by
878 the operating system. Only Linux supports O_ATOMIC right
879 now.
880
76a43db4
JA
881buffered=bool If value is true, use buffered io. This is the opposite
882 of the 'direct' option. Defaults to true.
71bfa161 883
f7fa2653 884offset=int Start io at the given offset in the file. The data before
71bfa161
JA
885 the given offset will not be touched. This effectively
886 caps the file size at real_size - offset.
887
214ac7e0 888offset_increment=int If this is provided, then the real offset becomes
69bdd6ba
JH
889 offset + offset_increment * thread_number, where the thread
890 number is a counter that starts at 0 and is incremented for
891 each sub-job (i.e. when numjobs option is specified). This
892 option is useful if there are several jobs which are intended
893 to operate on a file in parallel disjoint segments, with
894 even spacing between the starting points.
214ac7e0 895
ddf24e42
JA
896number_ios=int Fio will normally perform IOs until it has exhausted the size
897 of the region set by size=, or if it exhaust the allocated
898 time (or hits an error condition). With this setting, the
899 range/size can be set independently of the number of IOs to
900 perform. When fio reaches this number, it will exit normally
be3fec7d
JA
901 and report status. Note that this does not extend the amount
902 of IO that will be done, it will only stop fio if this
903 condition is met before other end-of-job criteria.
ddf24e42 904
71bfa161
JA
905fsync=int If writing to a file, issue a sync of the dirty data
906 for every number of blocks given. For example, if you give
907 32 as a parameter, fio will sync the file for every 32
908 writes issued. If fio is using non-buffered io, we may
909 not sync the file. The exception is the sg io engine, which
6c219763 910 synchronizes the disk cache anyway.
71bfa161 911
e76b1da4 912fdatasync=int Like fsync= but uses fdatasync() to only sync data and not
5f9099ea 913 metadata blocks.
37db59d6
JA
914 In FreeBSD and Windows there is no fdatasync(), this falls back
915 to using fsync()
5f9099ea 916
e76b1da4
JA
917sync_file_range=str:val Use sync_file_range() for every 'val' number of
918 write operations. Fio will track range of writes that
919 have happened since the last sync_file_range() call. 'str'
920 can currently be one or more of:
921
922 wait_before SYNC_FILE_RANGE_WAIT_BEFORE
923 write SYNC_FILE_RANGE_WRITE
924 wait_after SYNC_FILE_RANGE_WAIT_AFTER
925
926 So if you do sync_file_range=wait_before,write:8, fio would
927 use SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE for
928 every 8 writes. Also see the sync_file_range(2) man page.
929 This option is Linux specific.
930
5036fc1e
JA
931overwrite=bool If true, writes to a file will always overwrite existing
932 data. If the file doesn't already exist, it will be
933 created before the write phase begins. If the file exists
934 and is large enough for the specified write phase, nothing
935 will be done.
71bfa161 936
dbd11ead 937end_fsync=bool If true, fsync file contents when a write stage has completed.
71bfa161 938
ebb1415f
JA
939fsync_on_close=bool If true, fio will fsync() a dirty file on close.
940 This differs from end_fsync in that it will happen on every
941 file close, not just at the end of the job.
942
71bfa161
JA
943rwmixread=int How large a percentage of the mix should be reads.
944
945rwmixwrite=int How large a percentage of the mix should be writes. If both
946 rwmixread and rwmixwrite is given and the values do not add
947 up to 100%, the latter of the two will be used to override
c35dd7a6
JA
948 the first. This may interfere with a given rate setting,
949 if fio is asked to limit reads or writes to a certain rate.
950 If that is the case, then the distribution may be skewed.
71bfa161 951
92d42d69
JA
952random_distribution=str:float By default, fio will use a completely uniform
953 random distribution when asked to perform random IO. Sometimes
954 it is useful to skew the distribution in specific ways,
955 ensuring that some parts of the data is more hot than others.
956 fio includes the following distribution models:
957
958 random Uniform random distribution
959 zipf Zipf distribution
960 pareto Pareto distribution
961
962 When using a zipf or pareto distribution, an input value
963 is also needed to define the access pattern. For zipf, this
964 is the zipf theta. For pareto, it's the pareto power. Fio
965 includes a test program, genzipf, that can be used visualize
966 what the given input values will yield in terms of hit rates.
967 If you wanted to use zipf with a theta of 1.2, you would use
968 random_distribution=zipf:1.2 as the option. If a non-uniform
969 model is used, fio will disable use of the random map.
970
211c9b89
JA
971percentage_random=int For a random workload, set how big a percentage should
972 be random. This defaults to 100%, in which case the workload
973 is fully random. It can be set from anywhere from 0 to 100.
974 Setting it to 0 would make the workload fully sequential. Any
975 setting in between will result in a random mix of sequential
d9472271
JA
976 and random IO, at the given percentages. It is possible to
977 set different values for reads, writes, and trim. To do so,
978 simply use a comma separated list. See blocksize.
211c9b89 979
bb8895e0
JA
980norandommap Normally fio will cover every block of the file when doing
981 random IO. If this option is given, fio will just get a
982 new random offset without looking at past io history. This
983 means that some blocks may not be read or written, and that
83da8fbf
JE
984 some blocks may be read/written more than once. If this option
985 is used with verify= and multiple blocksizes (via bsrange=),
986 only intact blocks are verified, i.e., partially-overwritten
987 blocks are ignored.
bb8895e0 988
0408c206
JA
989softrandommap=bool See norandommap. If fio runs with the random block map
990 enabled and it fails to allocate the map, if this option is
991 set it will continue without a random block map. As coverage
992 will not be as complete as with random maps, this option is
2b386d25
JA
993 disabled by default.
994
e8b1961d
JA
995random_generator=str Fio supports the following engines for generating
996 IO offsets for random IO:
997
998 tausworthe Strong 2^88 cycle random number generator
999 lfsr Linear feedback shift register generator
c3546b53
JA
1000 tausworthe64 Strong 64-bit 2^258 cycle random number
1001 generator
e8b1961d
JA
1002
1003 Tausworthe is a strong random number generator, but it
1004 requires tracking on the side if we want to ensure that
1005 blocks are only read or written once. LFSR guarantees
1006 that we never generate the same offset twice, and it's
1007 also less computationally expensive. It's not a true
1008 random generator, however, though for IO purposes it's
1009 typically good enough. LFSR only works with single
1010 block sizes, not with workloads that use multiple block
1011 sizes. If used with such a workload, fio may read or write
1012 some blocks multiple times.
43f09da1 1013
71bfa161
JA
1014nice=int Run the job with the given nice value. See man nice(2).
1015
1016prio=int Set the io priority value of this job. Linux limits us to
1017 a positive value between 0 and 7, with 0 being the highest.
1018 See man ionice(1).
1019
1020prioclass=int Set the io priority class. See man ionice(1).
1021
1022thinktime=int Stall the job x microseconds after an io has completed before
1023 issuing the next. May be used to simulate processing being
48097d5c
JA
1024 done by an application. See thinktime_blocks and
1025 thinktime_spin.
1026
1027thinktime_spin=int
1028 Only valid if thinktime is set - pretend to spend CPU time
1029 doing something with the data received, before falling back
1030 to sleeping for the rest of the period specified by
1031 thinktime.
9c1f7434 1032
4d01ece6 1033thinktime_blocks=int
9c1f7434
JA
1034 Only valid if thinktime is set - control how many blocks
1035 to issue, before waiting 'thinktime' usecs. If not set,
1036 defaults to 1 which will make fio wait 'thinktime' usecs
4d01ece6
JA
1037 after every block. This effectively makes any queue depth
1038 setting redundant, since no more than 1 IO will be queued
1039 before we have to complete it and do our thinktime. In
1040 other words, this setting effectively caps the queue depth
1041 if the latter is larger.
71bfa161 1042
581e7141 1043rate=int Cap the bandwidth used by this job. The number is in bytes/sec,
b09da8fa 1044 the normal suffix rules apply. You can use rate=500k to limit
581e7141
JA
1045 reads and writes to 500k each, or you can specify read and
1046 writes separately. Using rate=1m,500k would limit reads to
1047 1MB/sec and writes to 500KB/sec. Capping only reads or
1048 writes can be done with rate=,500k or rate=500k,. The former
1049 will only limit writes (to 500KB/sec), the latter will only
1050 limit reads.
71bfa161 1051
6d428bcd 1052rate_min=int Tell fio to do whatever it can to maintain at least this
4e991c23 1053 bandwidth. Failing to meet this requirement, will cause
581e7141
JA
1054 the job to exit. The same format as rate is used for
1055 read vs write separation.
4e991c23
JA
1056
1057rate_iops=int Cap the bandwidth to this number of IOPS. Basically the same
1058 as rate, just specified independently of bandwidth. If the
1059 job is given a block size range instead of a fixed value,
581e7141 1060 the smallest block size is used as the metric. The same format
de8f6de9 1061 as rate is used for read vs write separation.
4e991c23
JA
1062
1063rate_iops_min=int If fio doesn't meet this rate of IO, it will cause
581e7141 1064 the job to exit. The same format as rate is used for read vs
de8f6de9 1065 write separation.
71bfa161 1066
6de65959
JA
1067rate_process=str This option controls how fio manages rated IO
1068 submissions. The default is 'linear', which submits IO in a
1069 linear fashion with fixed delays between IOs that gets
1070 adjusted based on IO completion rates. If this is set to
1071 'poisson', fio will submit IO based on a more real world
1072 random request flow, known as the Poisson process
5d02b083
JA
1073 (https://en.wikipedia.org/wiki/Poisson_process). The lambda
1074 will be 10^6 / IOPS for the given workload.
e7b24047 1075
3e260a46
JA
1076latency_target=int If set, fio will attempt to find the max performance
1077 point that the given workload will run at while maintaining a
1078 latency below this target. The values is given in microseconds.
1079 See latency_window and latency_percentile
1080
1081latency_window=int Used with latency_target to specify the sample window
1082 that the job is run at varying queue depths to test the
1083 performance. The value is given in microseconds.
1084
1085latency_percentile=float The percentage of IOs that must fall within the
1086 criteria specified by latency_target and latency_window. If not
1087 set, this defaults to 100.0, meaning that all IOs must be equal
1088 or below to the value set by latency_target.
1089
15501535
JA
1090max_latency=int If set, fio will exit the job if it exceeds this maximum
1091 latency. It will exit with an ETIME error.
1092
6d428bcd 1093rate_cycle=int Average bandwidth for 'rate' and 'rate_min' over this number
6c219763 1094 of milliseconds.
71bfa161
JA
1095
1096cpumask=int Set the CPU affinity of this job. The parameter given is a
a08bc17f
JA
1097 bitmask of allowed CPU's the job may run on. So if you want
1098 the allowed CPUs to be 1 and 5, you would pass the decimal
1099 value of (1 << 1 | 1 << 5), or 34. See man
7dbb6eba 1100 sched_setaffinity(2). This may not work on all supported
b0ea08ce
JA
1101 operating systems or kernel versions. This option doesn't
1102 work well for a higher CPU count than what you can store in
1103 an integer mask, so it can only control cpus 1-32. For
1104 boxes with larger CPU counts, use cpus_allowed.
71bfa161 1105
d2e268b0
JA
1106cpus_allowed=str Controls the same options as cpumask, but it allows a text
1107 setting of the permitted CPUs instead. So to use CPUs 1 and
62a7273d
JA
1108 5, you would specify cpus_allowed=1,5. This options also
1109 allows a range of CPUs. Say you wanted a binding to CPUs
1110 1, 5, and 8-15, you would set cpus_allowed=1,5,8-15.
d2e268b0 1111
c2acfbac
JA
1112cpus_allowed_policy=str Set the policy of how fio distributes the CPUs
1113 specified by cpus_allowed or cpumask. Two policies are
1114 supported:
1115
1116 shared All jobs will share the CPU set specified.
1117 split Each job will get a unique CPU from the CPU set.
1118
1119 'shared' is the default behaviour, if the option isn't
ada083cd
JA
1120 specified. If split is specified, then fio will will assign
1121 one cpu per job. If not enough CPUs are given for the jobs
1122 listed, then fio will roundrobin the CPUs in the set.
c2acfbac 1123
d0b937ed
YR
1124numa_cpu_nodes=str Set this job running on spcified NUMA nodes' CPUs. The
1125 arguments allow comma delimited list of cpu numbers,
1126 A-B ranges, or 'all'. Note, to enable numa options support,
67bf9823 1127 fio must be built on a system with libnuma-dev(el) installed.
d0b937ed
YR
1128
1129numa_mem_policy=str Set this job's memory policy and corresponding NUMA
1130 nodes. Format of the argements:
1131 <mode>[:<nodelist>]
1132 `mode' is one of the following memory policy:
1133 default, prefer, bind, interleave, local
1134 For `default' and `local' memory policy, no node is
1135 needed to be specified.
1136 For `prefer', only one node is allowed.
1137 For `bind' and `interleave', it allow comma delimited
1138 list of numbers, A-B ranges, or 'all'.
1139
e417fd66 1140startdelay=time Start this job the specified number of seconds after fio
71bfa161
JA
1141 has started. Only useful if the job file contains several
1142 jobs, and you want to delay starting some jobs to a certain
1143 time.
1144
e417fd66 1145runtime=time Tell fio to terminate processing after the specified number
71bfa161
JA
1146 of seconds. It can be quite hard to determine for how long
1147 a specified job will run, so this parameter is handy to
1148 cap the total runtime to a given time.
1149
cf4464ca 1150time_based If set, fio will run for the duration of the runtime
bf9a3edb 1151 specified even if the file(s) are completely read or
cf4464ca
JA
1152 written. It will simply loop over the same workload
1153 as many times as the runtime allows.
1154
e417fd66 1155ramp_time=time If set, fio will run the specified workload for this amount
721938ae
JA
1156 of time before logging any performance numbers. Useful for
1157 letting performance settle before logging results, thus
b29ee5b3
JA
1158 minimizing the runtime required for stable results. Note
1159 that the ramp_time is considered lead in time for a job,
1160 thus it will increase the total runtime if a special timeout
1161 or runtime is specified.
721938ae 1162
71bfa161
JA
1163invalidate=bool Invalidate the buffer/page cache parts for this file prior
1164 to starting io. Defaults to true.
1165
1166sync=bool Use sync io for buffered writes. For the majority of the
1167 io engines, this means using O_SYNC.
1168
d3aad8f2 1169iomem=str
71bfa161
JA
1170mem=str Fio can use various types of memory as the io unit buffer.
1171 The allowed values are:
1172
1173 malloc Use memory from malloc(3) as the buffers.
1174
1175 shm Use shared memory as the buffers. Allocated
1176 through shmget(2).
1177
74b025b0
JA
1178 shmhuge Same as shm, but use huge pages as backing.
1179
313cb206
JA
1180 mmap Use mmap to allocate buffers. May either be
1181 anonymous memory, or can be file backed if
1182 a filename is given after the option. The
1183 format is mem=mmap:/path/to/file.
71bfa161 1184
d0bdaf49
JA
1185 mmaphuge Use a memory mapped huge file as the buffer
1186 backing. Append filename after mmaphuge, ala
1187 mem=mmaphuge:/hugetlbfs/file
1188
09c782bb
JA
1189 mmapshared Same as mmap, but use a MMAP_SHARED
1190 mapping.
1191
71bfa161 1192 The area allocated is a function of the maximum allowed
5394ae5f
JA
1193 bs size for the job, multiplied by the io depth given. Note
1194 that for shmhuge and mmaphuge to work, the system must have
1195 free huge pages allocated. This can normally be checked
1196 and set by reading/writing /proc/sys/vm/nr_hugepages on a
b22989b9 1197 Linux system. Fio assumes a huge page is 4MB in size. So
5394ae5f
JA
1198 to calculate the number of huge pages you need for a given
1199 job file, add up the io depth of all jobs (normally one unless
1200 iodepth= is used) and multiply by the maximum bs set. Then
1201 divide that number by the huge page size. You can see the
1202 size of the huge pages in /proc/meminfo. If no huge pages
1203 are allocated by having a non-zero number in nr_hugepages,
56bb17f2 1204 using mmaphuge or shmhuge will fail. Also see hugepage-size.
5394ae5f
JA
1205
1206 mmaphuge also needs to have hugetlbfs mounted and the file
1207 location should point there. So if it's mounted in /huge,
1208 you would use mem=mmaphuge:/huge/somefile.
71bfa161 1209
d529ee19
JA
1210iomem_align=int This indiciates the memory alignment of the IO memory buffers.
1211 Note that the given alignment is applied to the first IO unit
1212 buffer, if using iodepth the alignment of the following buffers
1213 are given by the bs used. In other words, if using a bs that is
1214 a multiple of the page sized in the system, all buffers will
1215 be aligned to this value. If using a bs that is not page
1216 aligned, the alignment of subsequent IO memory buffers is the
1217 sum of the iomem_align and bs used.
1218
f7fa2653 1219hugepage-size=int
56bb17f2 1220 Defines the size of a huge page. Must at least be equal
b22989b9 1221 to the system setting, see /proc/meminfo. Defaults to 4MB.
c51074e7
JA
1222 Should probably always be a multiple of megabytes, so using
1223 hugepage-size=Xm is the preferred way to set this to avoid
1224 setting a non-pow-2 bad value.
56bb17f2 1225
71bfa161
JA
1226exitall When one job finishes, terminate the rest. The default is
1227 to wait for each job to finish, sometimes that is not the
1228 desired action.
1229
1230bwavgtime=int Average the calculated bandwidth over the given time. Value
6c219763 1231 is specified in milliseconds.
71bfa161 1232
c8eeb9df
JA
1233iopsavgtime=int Average the calculated IOPS over the given time. Value
1234 is specified in milliseconds.
1235
71bfa161
JA
1236create_serialize=bool If true, serialize the file creating for the jobs.
1237 This may be handy to avoid interleaving of data
1238 files, which may greatly depend on the filesystem
1239 used and even the number of processors in the system.
1240
1241create_fsync=bool fsync the data file after creation. This is the
1242 default.
1243
814452bd
JA
1244create_on_open=bool Don't pre-setup the files for IO, just create open()
1245 when it's time to do IO to that file.
1246
25460cf6
JA
1247create_only=bool If true, fio will only run the setup phase of the job.
1248 If files need to be laid out or updated on disk, only
1249 that will be done. The actual job contents are not
1250 executed.
1251
2378826d
JA
1252allow_file_create=bool If true, fio is permitted to create files as part
1253 of its workload. This is the default behavior. If this
1254 option is false, then fio will error out if the files it
1255 needs to use don't already exist. Default: true.
1256
e81ecca3
JA
1257allow_mounted_write=bool If this isn't set, fio will abort jobs that
1258 are destructive (eg that write) to what appears to be a
1259 mounted device or partition. This should help catch creating
1260 inadvertently destructive tests, not realizing that the test
1261 will destroy data on the mounted file system. Default: false.
1262
afad68f7 1263pre_read=bool If this is given, files will be pre-read into memory before
34f1c044
JA
1264 starting the given IO operation. This will also clear
1265 the 'invalidate' flag, since it is pointless to pre-read
9c0d2241
JA
1266 and then drop the cache. This will only work for IO engines
1267 that are seekable, since they allow you to read the same data
1268 multiple times. Thus it will not work on eg network or splice
1269 IO.
afad68f7 1270
e545a6ce 1271unlink=bool Unlink the job files when done. Not the default, as repeated
bf9a3edb
JA
1272 runs of that job would then waste time recreating the file
1273 set again and again.
71bfa161
JA
1274
1275loops=int Run the specified number of iterations of this job. Used
1276 to repeat the same workload a given number of times. Defaults
1277 to 1.
1278
62167762
JC
1279verify_only Do not perform specified workload---only verify data still
1280 matches previous invocation of this workload. This option
1281 allows one to check data multiple times at a later date
1282 without overwriting it. This option makes sense only for
1283 workloads that write data, and does not support workloads
1284 with the time_based option set.
1285
68e1f29a 1286do_verify=bool Run the verify phase after a write phase. Only makes sense if
e84c73a8
SL
1287 verify is set. Defaults to 1.
1288
71bfa161 1289verify=str If writing to a file, fio can verify the file contents
b638d82f
RP
1290 after each iteration of the job. Each verification method also implies
1291 verification of special header, which is written to the beginning of
1292 each block. This header also includes meta information, like offset
1293 of the block, block number, timestamp when block was written, etc.
1294 verify=str can be combined with verify_pattern=str option.
1295 The allowed values are:
71bfa161
JA
1296
1297 md5 Use an md5 sum of the data area and store
1298 it in the header of each block.
1299
17dc34df
JA
1300 crc64 Use an experimental crc64 sum of the data
1301 area and store it in the header of each
1302 block.
1303
bac39e0e
JA
1304 crc32c Use a crc32c sum of the data area and store
1305 it in the header of each block.
1306
3845591f 1307 crc32c-intel Use hardware assisted crc32c calcuation
0539d758
JA
1308 provided on SSE4.2 enabled processors. Falls
1309 back to regular software crc32c, if not
1310 supported by the system.
3845591f 1311
71bfa161
JA
1312 crc32 Use a crc32 sum of the data area and store
1313 it in the header of each block.
1314
969f7ed3
JA
1315 crc16 Use a crc16 sum of the data area and store
1316 it in the header of each block.
1317
17dc34df
JA
1318 crc7 Use a crc7 sum of the data area and store
1319 it in the header of each block.
1320
844ea602
JA
1321 xxhash Use xxhash as the checksum function. Generally
1322 the fastest software checksum that fio
1323 supports.
1324
cd14cc10
JA
1325 sha512 Use sha512 as the checksum function.
1326
1327 sha256 Use sha256 as the checksum function.
1328
7c353ceb
JA
1329 sha1 Use optimized sha1 as the checksum function.
1330
b638d82f
RP
1331 meta This option is deprecated, since now meta information is
1332 included in generic verification header and meta verification
1333 happens by default. For detailed information see the description
1334 of the verify=str setting. This option is kept because of
1335 compatibility's sake with old configurations. Do not use it.
7437ee87 1336
59245381
JA
1337 pattern Verify a strict pattern. Normally fio includes
1338 a header with some basic information and
1339 checksumming, but if this option is set, only
1340 the specific pattern set with 'verify_pattern'
1341 is verified.
1342
36690c9b
JA
1343 null Only pretend to verify. Useful for testing
1344 internals with ioengine=null, not for much
1345 else.
1346
6c219763 1347 This option can be used for repeated burn-in tests of a
71bfa161 1348 system to make sure that the written data is also
b892dc08
JA
1349 correctly read back. If the data direction given is
1350 a read or random read, fio will assume that it should
1351 verify a previously written file. If the data direction
1352 includes any form of write, the verify will be of the
1353 newly written data.
71bfa161 1354
160b966d
JA
1355verifysort=bool If set, fio will sort written verify blocks when it deems
1356 it faster to read them back in a sorted manner. This is
1357 often the case when overwriting an existing file, since
1358 the blocks are already laid out in the file system. You
1359 can ignore this option unless doing huge amounts of really
1360 fast IO where the red-black tree sorting CPU time becomes
1361 significant.
3f9f4e26 1362
f7fa2653 1363verify_offset=int Swap the verification header with data somewhere else
546a9142
SL
1364 in the block before writing. Its swapped back before
1365 verifying.
1366
f7fa2653 1367verify_interval=int Write the verification header at a finer granularity
3f9f4e26
SL
1368 than the blocksize. It will be written for chunks the
1369 size of header_interval. blocksize should divide this
1370 evenly.
90059d65 1371
0e92f873 1372verify_pattern=str If set, fio will fill the io buffers with this
e28218f3
SL
1373 pattern. Fio defaults to filling with totally random
1374 bytes, but sometimes it's interesting to fill with a known
1375 pattern for io verification purposes. Depending on the
1376 width of the pattern, fio will fill 1/2/3/4 bytes of the
0e92f873
RR
1377 buffer at the time(it can be either a decimal or a hex number).
1378 The verify_pattern if larger than a 32-bit quantity has to
996093bb 1379 be a hex number that starts with either "0x" or "0X". Use
b638d82f 1380 with verify=str. Also, verify_pattern supports %o format,
61b9861d
RP
1381 which means that for each block offset will be written and
1382 then verifyied back, e.g.:
1383
1384 verify_pattern=%o
1385
1386 Or use combination of everything:
1387 verify_pattern=0xff%o"abcd"-12
e28218f3 1388
68e1f29a 1389verify_fatal=bool Normally fio will keep checking the entire contents
a12a3b4d
JA
1390 before quitting on a block verification failure. If this
1391 option is set, fio will exit the job on the first observed
1392 failure.
e8462bd8 1393
b463e936
JA
1394verify_dump=bool If set, dump the contents of both the original data
1395 block and the data block we read off disk to files. This
1396 allows later analysis to inspect just what kind of data
ef71e317 1397 corruption occurred. Off by default.
b463e936 1398
e8462bd8
JA
1399verify_async=int Fio will normally verify IO inline from the submitting
1400 thread. This option takes an integer describing how many
1401 async offload threads to create for IO verification instead,
1402 causing fio to offload the duty of verifying IO contents
c85c324c
JA
1403 to one or more separate threads. If using this offload
1404 option, even sync IO engines can benefit from using an
1405 iodepth setting higher than 1, as it allows them to have
1406 IO in flight while verifies are running.
e8462bd8
JA
1407
1408verify_async_cpus=str Tell fio to set the given CPU affinity on the
1409 async IO verification threads. See cpus_allowed for the
1410 format used.
6f87418f
JA
1411
1412verify_backlog=int Fio will normally verify the written contents of a
1413 job that utilizes verify once that job has completed. In
1414 other words, everything is written then everything is read
1415 back and verified. You may want to verify continually
1416 instead for a variety of reasons. Fio stores the meta data
1417 associated with an IO block in memory, so for large
1418 verify workloads, quite a bit of memory would be used up
1419 holding this meta data. If this option is enabled, fio
f42195a3
JA
1420 will write only N blocks before verifying these blocks.
1421
6f87418f
JA
1422verify_backlog_batch=int Control how many blocks fio will verify
1423 if verify_backlog is set. If not set, will default to
1424 the value of verify_backlog (meaning the entire queue
f42195a3
JA
1425 is read back and verified). If verify_backlog_batch is
1426 less than verify_backlog then not all blocks will be verified,
1427 if verify_backlog_batch is larger than verify_backlog, some
1428 blocks will be verified more than once.
66c098b8 1429
ca09be4b
JA
1430verify_state_save=bool When a job exits during the write phase of a verify
1431 workload, save its current state. This allows fio to replay
1432 up until that point, if the verify state is loaded for the
1433 verify read phase. The format of the filename is, roughly,
1434 <type>-<jobname>-<jobindex>-verify.state. <type> is "local"
1435 for a local run, "sock" for a client/server socket connection,
1436 and "ip" (192.168.0.1, for instance) for a networked
1437 client/server connection.
1438
1439verify_state_load=bool If a verify termination trigger was used, fio stores
1440 the current write state of each thread. This can be used at
1441 verification time so that fio knows how far it should verify.
1442 Without this information, fio will run a full verification
1443 pass, according to the settings in the job file used.
1444
d392365e 1445stonewall
de8f6de9 1446wait_for_previous Wait for preceding jobs in the job file to exit, before
71bfa161 1447 starting this one. Can be used to insert serialization
b3d62a75
JA
1448 points in the job file. A stone wall also implies starting
1449 a new reporting group.
1450
abcab6af 1451new_group Start a new reporting group. See: group_reporting.
71bfa161
JA
1452
1453numjobs=int Create the specified number of clones of this job. May be
1454 used to setup a larger number of threads/processes doing
abcab6af
AV
1455 the same thing. Each thread is reported separately; to see
1456 statistics for all clones as a whole, use group_reporting in
1457 conjunction with new_group.
1458
1459group_reporting It may sometimes be interesting to display statistics for
04b2f799
JA
1460 groups of jobs as a whole instead of for each individual job.
1461 This is especially true if 'numjobs' is used; looking at
1462 individual thread/process output quickly becomes unwieldy.
1463 To see the final report per-group instead of per-job, use
1464 'group_reporting'. Jobs in a file will be part of the same
1465 reporting group, unless if separated by a stonewall, or by
1466 using 'new_group'.
71bfa161
JA
1467
1468thread fio defaults to forking jobs, however if this option is
1469 given, fio will use pthread_create(3) to create threads
1470 instead.
1471
f7fa2653 1472zonesize=int Divide a file into zones of the specified size. See zoneskip.
71bfa161 1473
f7fa2653 1474zoneskip=int Skip the specified number of bytes when zonesize data has
71bfa161
JA
1475 been read. The two zone options can be used to only do
1476 io on zones of a file.
1477
076efc7c 1478write_iolog=str Write the issued io patterns to the specified file. See
5b42a488
SH
1479 read_iolog. Specify a separate file for each job, otherwise
1480 the iologs will be interspersed and the file may be corrupt.
71bfa161 1481
076efc7c 1482read_iolog=str Open an iolog with the specified file name and replay the
71bfa161 1483 io patterns it contains. This can be used to store a
6df8adaa
JA
1484 workload and replay it sometime later. The iolog given
1485 may also be a blktrace binary file, which allows fio
1486 to replay a workload captured by blktrace. See blktrace
1487 for how to capture such logging data. For blktrace replay,
1488 the file needs to be turned into a blkparse binary data
ea3e51c3 1489 file first (blkparse <device> -o /dev/null -d file_for_fio.bin).
66c098b8 1490
64bbb865 1491replay_no_stall=int When replaying I/O with read_iolog the default behavior
62776229
JA
1492 is to attempt to respect the time stamps within the log and
1493 replay them with the appropriate delay between IOPS. By
1494 setting this variable fio will not respect the timestamps and
1495 attempt to replay them as fast as possible while still
1496 respecting ordering. The result is the same I/O pattern to a
1497 given device, but different timings.
71bfa161 1498
d1c46c04
DN
1499replay_redirect=str While replaying I/O patterns using read_iolog the
1500 default behavior is to replay the IOPS onto the major/minor
1501 device that each IOP was recorded from. This is sometimes
de8f6de9 1502 undesirable because on a different machine those major/minor
d1c46c04
DN
1503 numbers can map to a different device. Changing hardware on
1504 the same system can also result in a different major/minor
1505 mapping. Replay_redirect causes all IOPS to be replayed onto
1506 the single specified device regardless of the device it was
1507 recorded from. i.e. replay_redirect=/dev/sdc would cause all
1508 IO in the blktrace to be replayed onto /dev/sdc. This means
1509 multiple devices will be replayed onto a single, if the trace
1510 contains multiple devices. If you want multiple devices to be
1511 replayed concurrently to multiple redirected devices you must
1512 blkparse your trace into separate traces and replay them with
1513 independent fio invocations. Unfortuantely this also breaks
1514 the strict time ordering between multiple device accesses.
1515
0c63576e
JA
1516replay_align=int Force alignment of IO offsets and lengths in a trace
1517 to this power of 2 value.
1518
1519replay_scale=int Scale sector offsets down by this factor when
1520 replaying traces.
1521
3a5db920
JA
1522per_job_logs=bool If set, this generates bw/clat/iops log with per
1523 file private filenames. If not set, jobs with identical names
1524 will share the log filename. Default: true.
1525
e3cedca7 1526write_bw_log=str If given, write a bandwidth log of the jobs in this job
71bfa161 1527 file. Can be used to store data of the bandwidth of the
e0da9bc2
JA
1528 jobs in their lifetime. The included fio_generate_plots
1529 script uses gnuplot to turn these text files into nice
ddb754db 1530 graphs. See write_lat_log for behaviour of given
8ad3b3dd
JA
1531 filename. For this option, the suffix is _bw.x.log, where
1532 x is the index of the job (1..N, where N is the number of
3a5db920
JA
1533 jobs). If 'per_job_logs' is false, then the filename will not
1534 include the job index.
71bfa161 1535
e3cedca7 1536write_lat_log=str Same as write_bw_log, except that this option stores io
02af0988
JA
1537 submission, completion, and total latencies instead. If no
1538 filename is given with this option, the default filename of
1539 "jobname_type.log" is used. Even if the filename is given,
1540 fio will still append the type of log. So if one specifies
e3cedca7
JA
1541
1542 write_lat_log=foo
1543
8ad3b3dd
JA
1544 The actual log names will be foo_slat.x.log, foo_clat.x.log,
1545 and foo_lat.x.log, where x is the index of the job (1..N,
1546 where N is the number of jobs). This helps fio_generate_plot
3a5db920
JA
1547 fine the logs automatically. If 'per_job_logs' is false, then
1548 the filename will not include the job index.
1549
71bfa161 1550
b8bc8cba
JA
1551write_iops_log=str Same as write_bw_log, but writes IOPS. If no filename is
1552 given with this option, the default filename of
8ad3b3dd
JA
1553 "jobname_type.x.log" is used,where x is the index of the job
1554 (1..N, where N is the number of jobs). Even if the filename
3a5db920
JA
1555 is given, fio will still append the type of log. If
1556 'per_job_logs' is false, then the filename will not include
1557 the job index.
b8bc8cba
JA
1558
1559log_avg_msec=int By default, fio will log an entry in the iops, latency,
1560 or bw log for every IO that completes. When writing to the
1561 disk log, that can quickly grow to a very large size. Setting
1562 this option makes fio average the each log entry over the
1563 specified period of time, reducing the resolution of the log.
1564 Defaults to 0.
1565
ae588852
JA
1566log_offset=int If this is set, the iolog options will include the byte
1567 offset for the IO entry as well as the other data values.
1568
aee2ab67
JA
1569log_compression=int If this is set, fio will compress the IO logs as
1570 it goes, to keep the memory footprint lower. When a log
1571 reaches the specified size, that chunk is removed and
1572 compressed in the background. Given that IO logs are
1573 fairly highly compressible, this yields a nice memory
1574 savings for longer runs. The downside is that the
1575 compression will consume some background CPU cycles, so
1576 it may impact the run. This, however, is also true if
1577 the logging ends up consuming most of the system memory.
1578 So pick your poison. The IO logs are saved normally at the
1579 end of a run, by decompressing the chunks and storing them
1580 in the specified log file. This feature depends on the
1581 availability of zlib.
1582
b26317c9
JA
1583log_store_compressed=bool If set, and log_compression is also set,
1584 fio will store the log files in a compressed format. They
1585 can be decompressed with fio, using the --inflate-log
1586 command line parameter. The files will be stored with a
1587 .fz suffix.
1588
66347cfa
DE
1589block_error_percentiles=bool If set, record errors in trim block-sized
1590 units from writes and trims and output a histogram of
1591 how many trims it took to get to errors, and what kind
1592 of error was encountered.
1593
f7fa2653 1594lockmem=int Pin down the specified amount of memory with mlock(2). Can
71bfa161
JA
1595 potentially be used instead of removing memory or booting
1596 with less memory to simulate a smaller amount of memory.
81c6b6cd 1597 The amount specified is per worker.
71bfa161
JA
1598
1599exec_prerun=str Before running this job, issue the command specified
74c8c488
JA
1600 through system(3). Output is redirected in a file called
1601 jobname.prerun.txt.
71bfa161
JA
1602
1603exec_postrun=str After the job completes, issue the command specified
74c8c488
JA
1604 though system(3). Output is redirected in a file called
1605 jobname.postrun.txt.
71bfa161
JA
1606
1607ioscheduler=str Attempt to switch the device hosting the file to the specified
1608 io scheduler before running.
1609
0a839f30
JA
1610disk_util=bool Generate disk utilization statistics, if the platform
1611 supports it. Defaults to on.
1612
02af0988 1613disable_lat=bool Disable measurements of total latency numbers. Useful
9520ebb9
JA
1614 only for cutting back the number of calls to gettimeofday,
1615 as that does impact performance at really high IOPS rates.
1616 Note that to really get rid of a large amount of these
1617 calls, this option must be used with disable_slat and
1618 disable_bw as well.
1619
02af0988
JA
1620disable_clat=bool Disable measurements of completion latency numbers. See
1621 disable_lat.
1622
9520ebb9 1623disable_slat=bool Disable measurements of submission latency numbers. See
02af0988 1624 disable_slat.
9520ebb9
JA
1625
1626disable_bw=bool Disable measurements of throughput/bandwidth numbers. See
02af0988 1627 disable_lat.
9520ebb9 1628
83349190
YH
1629clat_percentiles=bool Enable the reporting of percentiles of
1630 completion latencies.
1631
1632percentile_list=float_list Overwrite the default list of percentiles
66347cfa
DE
1633 for completion latencies and the block error histogram.
1634 Each number is a floating number in the range (0,100],
1635 and the maximum length of the list is 20. Use ':'
1636 to separate the numbers, and list the numbers in ascending
1637 order. For example, --percentile_list=99.5:99.9 will cause
1638 fio to report the values of completion latency below which
1639 99.5% and 99.9% of the observed latencies fell, respectively.
83349190 1640
23893646
JA
1641clocksource=str Use the given clocksource as the base of timing. The
1642 supported options are:
1643
1644 gettimeofday gettimeofday(2)
1645
1646 clock_gettime clock_gettime(2)
1647
1648 cpu Internal CPU clock source
1649
1650 cpu is the preferred clocksource if it is reliable, as it
1651 is very fast (and fio is heavy on time calls). Fio will
1652 automatically use this clocksource if it's supported and
1653 considered reliable on the system it is running on, unless
1654 another clocksource is specifically set. For x86/x86-64 CPUs,
1655 this means supporting TSC Invariant.
1656
993bf48b
JA
1657gtod_reduce=bool Enable all of the gettimeofday() reducing options
1658 (disable_clat, disable_slat, disable_bw) plus reduce
1659 precision of the timeout somewhat to really shrink
1660 the gettimeofday() call count. With this option enabled,
1661 we only do about 0.4% of the gtod() calls we would have
1662 done if all time keeping was enabled.
1663
be4ecfdf
JA
1664gtod_cpu=int Sometimes it's cheaper to dedicate a single thread of
1665 execution to just getting the current time. Fio (and
1666 databases, for instance) are very intensive on gettimeofday()
1667 calls. With this option, you can set one CPU aside for
1668 doing nothing but logging current time to a shared memory
1669 location. Then the other threads/processes that run IO
1670 workloads need only copy that segment, instead of entering
1671 the kernel with a gettimeofday() call. The CPU set aside
1672 for doing these time calls will be excluded from other
1673 uses. Fio will manually clear it from the CPU mask of other
1674 jobs.
a696fa2a 1675
06842027 1676continue_on_error=str Normally fio will exit the job on the first observed
f2bba182
RR
1677 failure. If this option is set, fio will continue the job when
1678 there is a 'non-fatal error' (EIO or EILSEQ) until the runtime
1679 is exceeded or the I/O size specified is completed. If this
1680 option is used, there are two more stats that are appended,
1681 the total error count and the first error. The error field
1682 given in the stats is the first error that was hit during the
1683 run.
be4ecfdf 1684
06842027
SL
1685 The allowed values are:
1686
1687 none Exit on any IO or verify errors.
1688
1689 read Continue on read errors, exit on all others.
1690
1691 write Continue on write errors, exit on all others.
1692
1693 io Continue on any IO error, exit on all others.
1694
1695 verify Continue on verify errors, exit on all others.
1696
1697 all Continue on all errors.
1698
1699 0 Backward-compatible alias for 'none'.
1700
1701 1 Backward-compatible alias for 'all'.
1702
8b28bd41
DM
1703ignore_error=str Sometimes you want to ignore some errors during test
1704 in that case you can specify error list for each error type.
1705 ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST
1706 errors for given error type is separated with ':'. Error
1707 may be symbol ('ENOSPC', 'ENOMEM') or integer.
1708 Example:
1709 ignore_error=EAGAIN,ENOSPC:122
66c098b8
BC
1710 This option will ignore EAGAIN from READ, and ENOSPC and
1711 122(EDQUOT) from WRITE.
8b28bd41
DM
1712
1713error_dump=bool If set dump every error even if it is non fatal, true
1714 by default. If disabled only fatal error will be dumped
66c098b8 1715
6adb38a1
JA
1716cgroup=str Add job to this control group. If it doesn't exist, it will
1717 be created. The system must have a mounted cgroup blkio
1718 mount point for this to work. If your system doesn't have it
1719 mounted, you can do so with:
a696fa2a
JA
1720
1721 # mount -t cgroup -o blkio none /cgroup
1722
a696fa2a
JA
1723cgroup_weight=int Set the weight of the cgroup to this value. See
1724 the documentation that comes with the kernel, allowed values
1725 are in the range of 100..1000.
71bfa161 1726
7de87099
VG
1727cgroup_nodelete=bool Normally fio will delete the cgroups it has created after
1728 the job completion. To override this behavior and to leave
1729 cgroups around after the job completion, set cgroup_nodelete=1.
1730 This can be useful if one wants to inspect various cgroup
1731 files after job completion. Default: false
1732
e0b0d892
JA
1733uid=int Instead of running as the invoking user, set the user ID to
1734 this value before the thread/process does any work.
1735
1736gid=int Set group ID, see uid.
1737
9e684a49
DE
1738flow_id=int The ID of the flow. If not specified, it defaults to being a
1739 global flow. See flow.
1740
1741flow=int Weight in token-based flow control. If this value is used, then
1742 there is a 'flow counter' which is used to regulate the
1743 proportion of activity between two or more jobs. fio attempts
1744 to keep this flow counter near zero. The 'flow' parameter
1745 stands for how much should be added or subtracted to the flow
1746 counter on each iteration of the main I/O loop. That is, if
1747 one job has flow=8 and another job has flow=-1, then there
1748 will be a roughly 1:8 ratio in how much one runs vs the other.
1749
1750flow_watermark=int The maximum value that the absolute value of the flow
1751 counter is allowed to reach before the job must wait for a
1752 lower value of the counter.
1753
1754flow_sleep=int The period of time, in microseconds, to wait after the flow
1755 watermark has been exceeded before retrying operations
1756
de890a1e
SL
1757In addition, there are some parameters which are only valid when a specific
1758ioengine is in use. These are used identically to normal parameters, with the
1759caveat that when used on the command line, they must come after the ioengine
1760that defines them is selected.
1761
1762[libaio] userspace_reap Normally, with the libaio engine in use, fio will use
1763 the io_getevents system call to reap newly returned events.
1764 With this flag turned on, the AIO ring will be read directly
1765 from user-space to reap events. The reaping mode is only
1766 enabled when polling for a minimum of 0 events (eg when
1767 iodepth_batch_complete=0).
1768
0353050f
JA
1769[cpu] cpuload=int Attempt to use the specified percentage of CPU cycles.
1770
1771[cpu] cpuchunks=int Split the load into cycles of the given time. In
1772 microseconds.
1773
046395d7
JA
1774[cpu] exit_on_io_done=bool Detect when IO threads are done, then exit.
1775
de890a1e
SL
1776[netsplice] hostname=str
1777[net] hostname=str The host name or IP address to use for TCP or UDP based IO.
1778 If the job is a TCP listener or UDP reader, the hostname is not
b511c9aa
SB
1779 used and must be omitted unless it is a valid UDP multicast
1780 address.
de890a1e
SL
1781
1782[netsplice] port=int
6315af9d
JA
1783[net] port=int The TCP or UDP port to bind to or connect to. If this is used
1784with numjobs to spawn multiple instances of the same job type, then this will
1785be the starting port number since fio will use a range of ports.
de890a1e 1786
b93b6a2e
SB
1787[netsplice] interface=str
1788[net] interface=str The IP address of the network interface used to send or
1789 receive UDP multicast
1790
d3a623de
SB
1791[netsplice] ttl=int
1792[net] ttl=int Time-to-live value for outgoing UDP multicast packets.
1793 Default: 1
1794
1d360ffb
JA
1795[netsplice] nodelay=bool
1796[net] nodelay=bool Set TCP_NODELAY on TCP connections.
1797
de890a1e
SL
1798[netsplice] protocol=str
1799[netsplice] proto=str
1800[net] protocol=str
1801[net] proto=str The network protocol to use. Accepted values are:
1802
1803 tcp Transmission control protocol
49ccb8c1 1804 tcpv6 Transmission control protocol V6
f5cc3d0e 1805 udp User datagram protocol
49ccb8c1 1806 udpv6 User datagram protocol V6
de890a1e
SL
1807 unix UNIX domain socket
1808
1809 When the protocol is TCP or UDP, the port must also be given,
1810 as well as the hostname if the job is a TCP listener or UDP
1811 reader. For unix sockets, the normal filename option should be
1812 used and the port is invalid.
1813
1814[net] listen For TCP network connections, tell fio to listen for incoming
1815 connections rather than initiating an outgoing connection. The
1816 hostname must be omitted if this option is used.
1008602c 1817
b511c9aa 1818[net] pingpong Normaly a network writer will just continue writing data, and
7aeb1e94
JA
1819 a network reader will just consume packages. If pingpong=1
1820 is set, a writer will send its normal payload to the reader,
1821 then wait for the reader to send the same payload back. This
1822 allows fio to measure network latencies. The submission
1823 and completion latencies then measure local time spent
1824 sending or receiving, and the completion latency measures
1825 how long it took for the other end to receive and send back.
b511c9aa
SB
1826 For UDP multicast traffic pingpong=1 should only be set for a
1827 single reader when multiple readers are listening to the same
1828 address.
7aeb1e94 1829
1008602c
JA
1830[net] window_size Set the desired socket buffer size for the connection.
1831
e5f34d95
JA
1832[net] mss Set the TCP maximum segment size (TCP_MAXSEG).
1833
d54fce84
DM
1834[e4defrag] donorname=str
1835 File will be used as a block donor(swap extents between files)
1836[e4defrag] inplace=int
66c098b8 1837 Configure donor file blocks allocation strategy
d54fce84
DM
1838 0(default): Preallocate donor's file on init
1839 1 : allocate space immidietly inside defragment event,
1840 and free right after event
1841
65fa28ca 1842[mtd] skip_bad=bool Skip operations against known bad blocks.
de890a1e
SL
1843
1844
71bfa161
JA
18456.0 Interpreting the output
1846---------------------------
1847
1848fio spits out a lot of output. While running, fio will display the
1849status of the jobs created. An example of that would be:
1850
73c8b082 1851Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
71bfa161
JA
1852
1853The characters inside the square brackets denote the current status of
1854each thread. The possible values (in typical life cycle order) are:
1855
1856Idle Run
1857---- ---
1858P Thread setup, but not started.
1859C Thread created.
9c6f6316 1860I Thread initialized, waiting or generating necessary data.
b0f65863 1861 p Thread running pre-reading file(s).
71bfa161
JA
1862 R Running, doing sequential reads.
1863 r Running, doing random reads.
1864 W Running, doing sequential writes.
1865 w Running, doing random writes.
1866 M Running, doing mixed sequential reads/writes.
1867 m Running, doing mixed random reads/writes.
1868 F Running, currently waiting for fsync()
3d434057 1869 f Running, finishing up (writing IO logs, etc)
fc6bd43c 1870 V Running, doing verification of written data.
71bfa161 1871E Thread exited, not reaped by main thread yet.
4f7e57a4
JA
1872_ Thread reaped, or
1873X Thread reaped, exited with an error.
a5e371a6 1874K Thread reaped, exited due to signal.
71bfa161 1875
3e2e48a7
JA
1876Fio will condense the thread string as not to take up more space on the
1877command line as is needed. For instance, if you have 10 readers and 10
1878writers running, the output would look like this:
1879
1880Jobs: 20 (f=20): [R(10),W(10)] [4.0% done] [2103MB/0KB/0KB /s] [538K/0/0 iops] [eta 57m:36s]
1881
1882Fio will still maintain the ordering, though. So the above means that jobs
18831..10 are readers, and 11..20 are writers.
1884
71bfa161 1885The other values are fairly self explanatory - number of threads
c9f60304
JA
1886currently running and doing io, rate of io since last check (read speed
1887listed first, then write speed), and the estimated completion percentage
1888and time for the running group. It's impossible to estimate runtime of
4f7e57a4
JA
1889the following groups (if any). Note that the string is displayed in order,
1890so it's possible to tell which of the jobs are currently doing what. The
1891first character is the first job defined in the job file, and so forth.
71bfa161
JA
1892
1893When fio is done (or interrupted by ctrl-c), it will show the data for
1894each thread, group of threads, and disks in that order. For each data
1895direction, the output looks like:
1896
1897Client1 (g=0): err= 0:
35649e58 1898 write: io= 32MB, bw= 666KB/s, iops=89 , runt= 50320msec
6104ddb6
JA
1899 slat (msec): min= 0, max= 136, avg= 0.03, stdev= 1.92
1900 clat (msec): min= 0, max= 631, avg=48.50, stdev=86.82
b22989b9 1901 bw (KB/s) : min= 0, max= 1196, per=51.00%, avg=664.02, stdev=681.68
e7823a94 1902 cpu : usr=1.49%, sys=0.25%, ctx=7969, majf=0, minf=17
71619dc2 1903 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
1904 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
1905 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
30061b97 1906 issued r/w: total=0/32768, short=0/0
8abdce66
JA
1907 lat (msec): 2=1.6%, 4=0.0%, 10=3.2%, 20=12.8%, 50=38.4%, 100=24.8%,
1908 lat (msec): 250=15.2%, 500=0.0%, 750=0.0%, 1000=0.0%, >=2048=0.0%
71bfa161
JA
1909
1910The client number is printed, along with the group id and error of that
1911thread. Below is the io statistics, here for writes. In the order listed,
1912they denote:
1913
1914io= Number of megabytes io performed
1915bw= Average bandwidth rate
35649e58 1916iops= Average IOs performed per second
71bfa161 1917runt= The runtime of that thread
72fbda2a 1918 slat= Submission latency (avg being the average, stdev being the
71bfa161
JA
1919 standard deviation). This is the time it took to submit
1920 the io. For sync io, the slat is really the completion
8a35c71e 1921 latency, since queue/complete is one operation there. This
bf9a3edb 1922 value can be in milliseconds or microseconds, fio will choose
8a35c71e 1923 the most appropriate base and print that. In the example
0d237712
LAG
1924 above, milliseconds is the best scale. Note: in --minimal mode
1925 latencies are always expressed in microseconds.
71bfa161
JA
1926 clat= Completion latency. Same names as slat, this denotes the
1927 time from submission to completion of the io pieces. For
1928 sync io, clat will usually be equal (or very close) to 0,
1929 as the time from submit to complete is basically just
1930 CPU time (io has already been done, see slat explanation).
1931 bw= Bandwidth. Same names as the xlat stats, but also includes
1932 an approximate percentage of total aggregate bandwidth
1933 this thread received in this group. This last value is
1934 only really useful if the threads in this group are on the
1935 same disk, since they are then competing for disk access.
1936cpu= CPU usage. User and system time, along with the number
e7823a94
JA
1937 of context switches this thread went through, usage of
1938 system and user time, and finally the number of major
1939 and minor page faults.
71619dc2
JA
1940IO depths= The distribution of io depths over the job life time. The
1941 numbers are divided into powers of 2, so for example the
1942 16= entries includes depths up to that value but higher
1943 than the previous entry. In other words, it covers the
1944 range from 16 to 31.
838bc709
JA
1945IO submit= How many pieces of IO were submitting in a single submit
1946 call. Each entry denotes that amount and below, until
1947 the previous entry - eg, 8=100% mean that we submitted
1948 anywhere in between 5-8 ios per submit call.
1949IO complete= Like the above submit number, but for completions instead.
30061b97
JA
1950IO issued= The number of read/write requests issued, and how many
1951 of them were short.
ec118304
JA
1952IO latencies= The distribution of IO completion latencies. This is the
1953 time from when IO leaves fio and when it gets completed.
1954 The numbers follow the same pattern as the IO depths,
1955 meaning that 2=1.6% means that 1.6% of the IO completed
8abdce66
JA
1956 within 2 msecs, 20=12.8% means that 12.8% of the IO
1957 took more than 10 msecs, but less than (or equal to) 20 msecs.
71bfa161
JA
1958
1959After each client has been listed, the group statistics are printed. They
1960will look like this:
1961
1962Run status group 0 (all jobs):
b22989b9
JA
1963 READ: io=64MB, aggrb=22178, minb=11355, maxb=11814, mint=2840msec, maxt=2955msec
1964 WRITE: io=64MB, aggrb=1302, minb=666, maxb=669, mint=50093msec, maxt=50320msec
71bfa161
JA
1965
1966For each data direction, it prints:
1967
1968io= Number of megabytes io performed.
1969aggrb= Aggregate bandwidth of threads in this group.
1970minb= The minimum average bandwidth a thread saw.
1971maxb= The maximum average bandwidth a thread saw.
1972mint= The smallest runtime of the threads in that group.
1973maxt= The longest runtime of the threads in that group.
1974
1975And finally, the disk statistics are printed. They will look like this:
1976
1977Disk stats (read/write):
1978 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
1979
1980Each value is printed for both reads and writes, with reads first. The
1981numbers denote:
1982
1983ios= Number of ios performed by all groups.
1984merge= Number of merges io the io scheduler.
1985ticks= Number of ticks we kept the disk busy.
1986io_queue= Total time spent in the disk queue.
1987util= The disk utilization. A value of 100% means we kept the disk
1988 busy constantly, 50% would be a disk idling half of the time.
1989
8423bd11
JA
1990It is also possible to get fio to dump the current output while it is
1991running, without terminating the job. To do that, send fio the USR1 signal.
06464907
JA
1992You can also get regularly timed dumps by using the --status-interval
1993parameter, or by creating a file in /tmp named fio-dump-status. If fio
1994sees this file, it will unlink it and dump the current output status.
8423bd11 1995
71bfa161
JA
1996
19977.0 Terse output
1998----------------
1999
2000For scripted usage where you typically want to generate tables or graphs
6af019c9 2001of the results, fio can output the results in a semicolon separated format.
71bfa161
JA
2002The format is one long line of values, such as:
2003
562c2d2f
DN
20042;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%
2005A description of this job goes here.
2006
2007The job description (if provided) follows on a second line.
71bfa161 2008
525c2bfa
JA
2009To enable terse output, use the --minimal command line option. The first
2010value is the version of the terse output format. If the output has to
2011be changed for some reason, this number will be incremented by 1 to
2012signify that change.
6820cb3b 2013
71bfa161
JA
2014Split up, the format is as follows:
2015
5e726d0a 2016 terse version, fio version, jobname, groupid, error
71bfa161 2017 READ status:
312b4af2 2018 Total IO (KB), bandwidth (KB/sec), IOPS, runtime (msec)
d86ae56c
CW
2019 Submission latency: min, max, mean, stdev (usec)
2020 Completion latency: min, max, mean, stdev (usec)
1db92cb6 2021 Completion latency percentiles: 20 fields (see below)
d86ae56c
CW
2022 Total latency: min, max, mean, stdev (usec)
2023 Bw (KB/s): min, max, aggregate percentage of total, mean, stdev
71bfa161 2024 WRITE status:
312b4af2 2025 Total IO (KB), bandwidth (KB/sec), IOPS, runtime (msec)
d86ae56c
CW
2026 Submission latency: min, max, mean, stdev (usec)
2027 Completion latency: min, max, mean, stdev(usec)
1db92cb6 2028 Completion latency percentiles: 20 fields (see below)
d86ae56c
CW
2029 Total latency: min, max, mean, stdev (usec)
2030 Bw (KB/s): min, max, aggregate percentage of total, mean, stdev
046ee302 2031 CPU usage: user, system, context switches, major faults, minor faults
2270890c 2032 IO depths: <=1, 2, 4, 8, 16, 32, >=64
562c2d2f
DN
2033 IO latencies microseconds: <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
2034 IO latencies milliseconds: <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
f2f788dd
JA
2035 Disk utilization: Disk name, Read ios, write ios,
2036 Read merges, write merges,
2037 Read ticks, write ticks,
3d7cd9b4 2038 Time spent in queue, disk utilization percentage
de8f6de9 2039 Additional Info (dependent on continue_on_error, default off): total # errors, first error code
66c098b8 2040
de8f6de9 2041 Additional Info (dependent on description being set): Text description
25c8b9d7 2042
1db92cb6
JA
2043Completion latency percentiles can be a grouping of up to 20 sets, so
2044for the terse output fio writes all of them. Each field will look like this:
2045
2046 1.00%=6112
2047
2048which is the Xth percentile, and the usec latency associated with it.
2049
f2f788dd
JA
2050For disk utilization, all disks used by fio are shown. So for each disk
2051there will be a disk utilization section.
2052
25c8b9d7
PD
2053
20548.0 Trace file format
2055---------------------
66c098b8 2056There are two trace file format that you can encounter. The older (v1) format
25c8b9d7
PD
2057is unsupported since version 1.20-rc3 (March 2008). It will still be described
2058below in case that you get an old trace and want to understand it.
2059
2060In any case the trace is a simple text file with a single action per line.
2061
2062
20638.1 Trace file format v1
2064------------------------
2065Each line represents a single io action in the following format:
2066
2067rw, offset, length
2068
2069where rw=0/1 for read/write, and the offset and length entries being in bytes.
2070
2071This format is not supported in Fio versions => 1.20-rc3.
2072
2073
20748.2 Trace file format v2
2075------------------------
2076The second version of the trace file format was added in Fio version 1.17.
2077It allows to access more then one file per trace and has a bigger set of
2078possible file actions.
2079
2080The first line of the trace file has to be:
2081
2082fio version 2 iolog
2083
2084Following this can be lines in two different formats, which are described below.
2085
2086The file management format:
2087
2088filename action
2089
2090The filename is given as an absolute path. The action can be one of these:
2091
2092add Add the given filename to the trace
66c098b8 2093open Open the file with the given filename. The filename has to have
25c8b9d7
PD
2094 been added with the add action before.
2095close Close the file with the given filename. The file has to have been
2096 opened before.
2097
2098
2099The file io action format:
2100
2101filename action offset length
2102
2103The filename is given as an absolute path, and has to have been added and opened
66c098b8 2104before it can be used with this format. The offset and length are given in
25c8b9d7
PD
2105bytes. The action can be one of these:
2106
2107wait Wait for 'offset' microseconds. Everything below 100 is discarded.
5c7808fe 2108 The time is relative to the previous wait statement.
25c8b9d7
PD
2109read Read 'length' bytes beginning from 'offset'
2110write Write 'length' bytes beginning from 'offset'
2111sync fsync() the file
2112datasync fdatasync() the file
2113trim trim the given file from the given 'offset' for 'length' bytes
f2a2ce0e
HL
2114
2115
21169.0 CPU idleness profiling
06464907 2117--------------------------
f2a2ce0e
HL
2118In some cases, we want to understand CPU overhead in a test. For example,
2119we test patches for the specific goodness of whether they reduce CPU usage.
2120fio implements a balloon approach to create a thread per CPU that runs at
2121idle priority, meaning that it only runs when nobody else needs the cpu.
2122By measuring the amount of work completed by the thread, idleness of each
2123CPU can be derived accordingly.
2124
2125An unit work is defined as touching a full page of unsigned characters. Mean
2126and standard deviation of time to complete an unit work is reported in "unit
2127work" section. Options can be chosen to report detailed percpu idleness or
2128overall system idleness by aggregating percpu stats.
99b9a85a
JA
2129
2130
213110.0 Verification and triggers
2132------------------------------
2133Fio is usually run in one of two ways, when data verification is done. The
2134first is a normal write job of some sort with verify enabled. When the
2135write phase has completed, fio switches to reads and verifies everything
2136it wrote. The second model is running just the write phase, and then later
2137on running the same job (but with reads instead of writes) to repeat the
2138same IO patterns and verify the contents. Both of these methods depend
2139on the write phase being completed, as fio otherwise has no idea how much
2140data was written.
2141
2142With verification triggers, fio supports dumping the current write state
2143to local files. Then a subsequent read verify workload can load this state
2144and know exactly where to stop. This is useful for testing cases where
2145power is cut to a server in a managed fashion, for instance.
2146
2147A verification trigger consists of two things:
2148
21491) Storing the write state of each job
21502) Executing a trigger command
2151
2152The write state is relatively small, on the order of hundreds of bytes
2153to single kilobytes. It contains information on the number of completions
2154done, the last X completions, etc.
2155
2156A trigger is invoked either through creation ('touch') of a specified
2157file in the system, or through a timeout setting. If fio is run with
2158--trigger-file=/tmp/trigger-file, then it will continually check for
2159the existence of /tmp/trigger-file. When it sees this file, it will
2160fire off the trigger (thus saving state, and executing the trigger
2161command).
2162
2163For client/server runs, there's both a local and remote trigger. If
2164fio is running as a server backend, it will send the job states back
2165to the client for safe storage, then execute the remote trigger, if
2166specified. If a local trigger is specified, the server will still send
2167back the write state, but the client will then execute the trigger.
2168
216910.1 Verification trigger example
2170---------------------------------
2171Lets say we want to run a powercut test on the remote machine 'server'.
2172Our write workload is in write-test.fio. We want to cut power to 'server'
2173at some point during the run, and we'll run this test from the safety
2174or our local machine, 'localbox'. On the server, we'll start the fio
2175backend normally:
2176
2177server# fio --server
2178
2179and on the client, we'll fire off the workload:
2180
2181localbox$ fio --client=server --trigger-file=/tmp/my-trigger --trigger-remote="bash -c \"echo b > /proc/sysrq-triger\""
2182
2183We set /tmp/my-trigger as the trigger file, and we tell fio to execute
2184
2185echo b > /proc/sysrq-trigger
2186
2187on the server once it has received the trigger and sent us the write
2188state. This will work, but it's not _really_ cutting power to the server,
2189it's merely abruptly rebooting it. If we have a remote way of cutting
2190power to the server through IPMI or similar, we could do that through
2191a local trigger command instead. Lets assume we have a script that does
2192IPMI reboot of a given hostname, ipmi-reboot. On localbox, we could
2193then have run fio with a local trigger instead:
2194
2195localbox$ fio --client=server --trigger-file=/tmp/my-trigger --trigger="ipmi-reboot server"
2196
2197For this case, fio would wait for the server to send us the write state,
2198then execute 'ipmi-reboot server' when that happened.
2199
220010.1 Loading verify state
2201-------------------------
2202To load store write state, read verification job file must contain
2203the verify_state_load option. If that is set, fio will load the previously
2204stored state. For a local fio run this is done by loading the files directly,
2205and on a client/server run, the server backend will ask the client to send
2206the files over and load them from there.