8 5. Detailed list of parameters
12 9. CPU idleness profiling
13 10. Verification and triggers
17 1.0 Overview and history
18 ------------------------
19 fio was originally written to save me the hassle of writing special test
20 case programs when I wanted to test a specific workload, either for
21 performance reasons or to find/reproduce a bug. The process of writing
22 such a test app can be tiresome, especially if you have to do it often.
23 Hence I needed a tool that would be able to simulate a given io workload
24 without resorting to writing a tailored test case again and again.
26 A test work load is difficult to define, though. There can be any number
27 of processes or threads involved, and they can each be using their own
28 way of generating io. You could have someone dirtying large amounts of
29 memory in an memory mapped file, or maybe several threads issuing
30 reads using asynchronous io. fio needed to be flexible enough to
31 simulate both of these cases, and many more.
35 The first step in getting fio to simulate a desired io workload, is
36 writing a job file describing that specific setup. A job file may contain
37 any number of threads and/or files - the typical contents of the job file
38 is a global section defining shared parameters, and one or more job
39 sections describing the jobs involved. When run, fio parses this file
40 and sets everything up as described. If we break down a job from top to
41 bottom, it contains the following basic parameters:
43 IO type Defines the io pattern issued to the file(s).
44 We may only be reading sequentially from this
45 file(s), or we may be writing randomly. Or even
46 mixing reads and writes, sequentially or randomly.
48 Block size In how large chunks are we issuing io? This may be
49 a single value, or it may describe a range of
52 IO size How much data are we going to be reading/writing.
54 IO engine How do we issue io? We could be memory mapping the
55 file, we could be using regular read/write, we
56 could be using splice, async io, or even SG
59 IO depth If the io engine is async, how large a queuing
60 depth do we want to maintain?
62 IO type Should we be doing buffered io, or direct/raw io?
64 Num files How many files are we spreading the workload over.
66 Num threads How many threads or processes should we spread
69 The above are the basic parameters defined for a workload, in addition
70 there's a multitude of parameters that modify other aspects of how this
76 See the README file for command line parameters, there are only a few
79 Running fio is normally the easiest part - you just give it the job file
80 (or job files) as parameters:
84 and it will start doing what the job_file tells it to do. You can give
85 more than one job file on the command line, fio will serialize the running
86 of those files. Internally that is the same as using the 'stonewall'
87 parameter described in the parameter section.
89 If the job file contains only one job, you may as well just give the
90 parameters on the command line. The command line parameters are identical
91 to the job parameters, with a few extra that control global parameters
92 (see README). For example, for the job file parameter iodepth=2, the
93 mirror command line option would be --iodepth 2 or --iodepth=2. You can
94 also use the command line for giving more than one job entry. For each
95 --name option that fio sees, it will start a new job with that name.
96 Command line entries following a --name entry will apply to that job,
97 until there are no more entries or a new --name entry is seen. This is
98 similar to the job file options, where each option applies to the current
99 job until a new [] job entry is seen.
101 fio does not need to run as root, except if the files or devices specified
102 in the job section requires that. Some other options may also be restricted,
103 such as memory locking, io scheduler switching, and decreasing the nice value.
108 As previously described, fio accepts one or more job files describing
109 what it is supposed to do. The job file format is the classic ini file,
110 where the names enclosed in [] brackets define the job name. You are free
111 to use any ascii name you want, except 'global' which has special meaning.
112 A global section sets defaults for the jobs described in that file. A job
113 may override a global section parameter, and a job file may even have
114 several global sections if so desired. A job is only affected by a global
115 section residing above it. If the first character in a line is a ';' or a
116 '#', the entire line is discarded as a comment.
118 So let's look at a really simple job file that defines two processes, each
119 randomly reading from a 128MiB file.
121 ; -- start job file --
132 As you can see, the job file sections themselves are empty as all the
133 described parameters are shared. As no filename= option is given, fio
134 makes up a filename for each of the jobs as it sees fit. On the command
135 line, this job would look as follows:
137 $ fio --name=global --rw=randread --size=128m --name=job1 --name=job2
140 Let's look at an example that has a number of processes writing randomly
143 ; -- start job file --
155 Here we have no global section, as we only have one job defined anyway.
156 We want to use async io here, with a depth of 4 for each file. We also
157 increased the buffer size used to 32KiB and define numjobs to 4 to
158 fork 4 identical jobs. The result is 4 processes each randomly writing
159 to their own 64MiB file. Instead of using the above job file, you could
160 have given the parameters on the command line. For this case, you would
163 $ fio --name=random-writers --ioengine=libaio --iodepth=4 --rw=randwrite --bs=32k --direct=0 --size=64m --numjobs=4
165 When fio is utilized as a basis of any reasonably large test suite, it might be
166 desirable to share a set of standardized settings across multiple job files.
167 Instead of copy/pasting such settings, any section may pull in an external
168 .fio file with 'include filename' directive, as in the following example:
170 ; -- start job file including.fio --
174 include glob-include.fio
181 include test-include.fio
182 ; -- end job file including.fio --
184 ; -- start job file glob-include.fio --
187 ; -- end job file glob-include.fio --
189 ; -- start job file test-include.fio --
192 ; -- end job file test-include.fio --
194 Settings pulled into a section apply to that section only (except global
195 section). Include directives may be nested in that any included file may
196 contain further include directive(s). Include files may not contain []
200 4.1 Environment variables
201 -------------------------
203 fio also supports environment variable expansion in job files. Any
204 sub-string of the form "${VARNAME}" as part of an option value (in other
205 words, on the right of the `='), will be expanded to the value of the
206 environment variable called VARNAME. If no such environment variable
207 is defined, or VARNAME is the empty string, the empty string will be
210 As an example, let's look at a sample fio invocation and job file:
212 $ SIZE=64m NUMJOBS=4 fio jobfile.fio
214 ; -- start job file --
221 This will expand to the following equivalent job file at runtime:
223 ; -- start job file --
230 fio ships with a few example job files, you can also look there for
233 4.2 Reserved keywords
234 ---------------------
236 Additionally, fio has a set of reserved keywords that will be replaced
237 internally with the appropriate value. Those keywords are:
239 $pagesize The architecture page size of the running system
240 $mb_memory Megabytes of total memory in the system
241 $ncpus Number of online available CPUs
243 These can be used on the command line or in the job file, and will be
244 automatically substituted with the current system values when the job
245 is run. Simple math is also supported on these keywords, so you can
246 perform actions like:
250 and get that properly expanded to 8 times the size of memory in the
254 5.0 Detailed list of parameters
255 -------------------------------
257 This section describes in details each parameter associated with a job.
258 Some parameters take an option of a given type, such as an integer or
259 a string. Anywhere a numeric value is required, an arithmetic expression
260 may be used, provided it is surrounded by parentheses. Supported operators
270 For time values in expressions, units are microseconds by default. This is
271 different than for time values not in expressions (not enclosed in
272 parentheses). The following types are used:
274 str String. This is a sequence of alpha characters.
275 time Integer with possible time suffix. In seconds unless otherwise
276 specified, use eg 10m for 10 minutes. Accepts s/m/h for seconds,
277 minutes, and hours, and accepts 'ms' (or 'msec') for milliseconds,
278 and 'us' (or 'usec') for microseconds.
280 int Integer. A whole number value, which may contain an integer prefix
281 and an integer suffix.
282 [integer prefix]number[integer suffix]
284 The optional integer prefix specifies the number's base. The default
285 is decimal. 0x specifies hexadecimal.
287 The optional integer suffix specifies the number's units, and includes
288 an optional unit prefix and an optional unit. For quantities of data,
289 the default unit is bytes. For quantities of time, the default unit
292 With kb_base=1000, fio follows international standards for unit prefixes.
293 To specify power-of-10 decimal values defined in the International
294 System of Units (SI):
295 Ki means kilo (K) or 1000
296 Mi means mega (M) or 1000**2
297 Gi means giga (G) or 1000**3
298 Ti means tera (T) or 1000**4
299 Pi means peta (P) or 1000**5
301 To specify power-of-2 binary values defined in IEC 80000-13:
302 k means kibi (Ki) or 1024
303 M means mebi (Mi) or 1024**2
304 G means gibi (Gi) or 1024**3
305 T means tebi (Ti) or 1024**4
306 P means pebi (Pi) or 1024**5
308 With kb_base=1024 (the default), the unit prefixes are opposite from
309 those specified in the SI and IEC 80000-13 standards to provide
310 compatibility with old scripts. For example, 4k means 4096.
312 For quantities of data, an optional unit of 'B' may be included
313 (e.g., 'kB' is the same as 'k').
315 The integer suffix is not case sensitive (e.g., m/mi mean mebi/mega,
316 not milli). 'b' and 'B' both mean byte, not bit.
318 Examples with kb_base=1000:
319 4 KiB: 4096, 4096b, 4096B, 4ki, 4kib, 4kiB, 4Ki, 4KiB
320 1 MiB: 1048576, 1mi, 1024ki
321 1 MB: 1000000, 1m, 1000k
322 1 TiB: 1073741824, 1ti, 1024mi, 1048576ki
323 1 TB: 1000000000, 1t, 1000m, 1000000k
325 Examples with kb_base=1024 (default):
326 4 KiB: 4096, 4096b, 4096B, 4k, 4kb, 4kB, 4K, 4KB
327 1 MiB: 1048576, 1m, 1024k
328 1 MB: 1000000, 1mi, 1000ki
329 1 TiB: 1073741824, 1t, 1024m, 1048576k
330 1 TB: 1000000000, 1ti, 1000mi, 1000000ki
332 To specify times (units are not case sensitive):
336 s or sec means seconds (default)
337 ms or msec means milliseconds
338 us or usec means microseconds
340 If the option accepts an upper and lower range, use a colon ':' or
341 minus '-' to separate such values. See irange.
343 bool Boolean. Usually parsed as an integer, however only defined for
344 true and false (1 and 0).
345 irange Integer range with suffix. Allows value range to be given, such
346 as 1024-4096. A colon may also be used as the separator, eg
347 1k:4k. If the option allows two sets of ranges, they can be
348 specified with a ',' or '/' delimiter: 1k-4k/8k-32k. Also see
350 float_list A list of floating point numbers, separated by a ':' character.
352 With the above in mind, here follows the complete list of fio job
355 name=str ASCII name of the job. This may be used to override the
356 name printed by fio for this job. Otherwise the job
357 name is used. On the command line this parameter has the
358 special purpose of also signaling the start of a new
361 wait_for=str Specifies the name of the already defined job to wait
362 for. Single waitee name only may be specified. If set, the job
363 won't be started until all workers of the waitee job are done.
365 Wait_for operates on the job name basis, so there are a few
366 limitations. First, the waitee must be defined prior to the
367 waiter job (meaning no forward references). Second, if a job
368 is being referenced as a waitee, it must have a unique name
369 (no duplicate waitees).
371 description=str Text description of the job. Doesn't do anything except
372 dump this text description when this job is run. It's
375 directory=str Prefix filenames with this directory. Used to place files
376 in a different location than "./". See the 'filename' option
377 for escaping certain characters.
379 filename=str Fio normally makes up a filename based on the job name,
380 thread number, and file number. If you want to share
381 files between threads in a job or several jobs, specify
382 a filename for each of them to override the default.
383 If the ioengine is file based, you can specify a number of
384 files by separating the names with a ':' colon. So if you
385 wanted a job to open /dev/sda and /dev/sdb as the two working
386 files, you would use filename=/dev/sda:/dev/sdb. On Windows,
387 disk devices are accessed as \\.\PhysicalDrive0 for the first
388 device, \\.\PhysicalDrive1 for the second etc. Note: Windows
389 and FreeBSD prevent write access to areas of the disk
390 containing in-use data (e.g. filesystems).
391 If the wanted filename does need to include a colon, then
392 escape that with a '\' character. For instance, if the filename
393 is "/dev/dsk/foo@3,0:c", then you would use
394 filename="/dev/dsk/foo@3,0\:c". '-' is a reserved name, meaning
395 stdin or stdout. Which of the two depends on the read/write
399 If sharing multiple files between jobs, it is usually necessary
400 to have fio generate the exact names that you want. By default,
401 fio will name a file based on the default file format
402 specification of jobname.jobnumber.filenumber. With this
403 option, that can be customized. Fio will recognize and replace
404 the following keywords in this string:
407 The name of the worker thread or process.
410 The incremental number of the worker thread or
414 The incremental number of the file for that worker
417 To have dependent jobs share a set of files, this option can
418 be set to have fio generate filenames that are shared between
419 the two. For instance, if testfiles.$filenum is specified,
420 file number 4 for any job will be named testfiles.4. The
421 default of $jobname.$jobnum.$filenum will be used if
422 no other format specifier is given.
424 unique_filename=bool To avoid collisions between networked clients, fio
425 defaults to prefixing any generated filenames (with a directory
426 specified) with the source of the client connecting. To disable
427 this behavior, set this option to 0.
429 opendir=str Tell fio to recursively add any file it can find in this
430 directory and down the file system tree.
432 lockfile=str Fio defaults to not locking any files before it does
433 IO to them. If a file or file descriptor is shared, fio
434 can serialize IO to that file to make the end result
435 consistent. This is usual for emulating real workloads that
436 share files. The lock modes are:
438 none No locking. The default.
439 exclusive Only one thread/process may do IO,
440 excluding all others.
441 readwrite Read-write locking on the file. Many
442 readers may access the file at the
443 same time, but writes get exclusive
447 rw=str Type of io pattern. Accepted values are:
449 read Sequential reads
450 write Sequential writes
451 trim Sequential trims
452 randwrite Random writes
453 randread Random reads
454 randtrim Random trims
455 rw,readwrite Sequential mixed reads and writes
456 randrw Random mixed reads and writes
457 trimwrite Sequential trim+write sequences
459 Fio defaults to read if the option is not specified.
460 For the mixed io types, the default is to split them 50/50.
461 For certain types of io the result may still be skewed a bit,
462 since the speed may be different. It is possible to specify
463 a number of IO's to do before getting a new offset, this is
464 done by appending a ':<nr>' to the end of the string given.
465 For a random read, it would look like 'rw=randread:8' for
466 passing in an offset modifier with a value of 8. If the
467 suffix is used with a sequential IO pattern, then the value
468 specified will be added to the generated offset for each IO.
469 For instance, using rw=write:4k will skip 4k for every
470 write. It turns sequential IO into sequential IO with holes.
471 See the 'rw_sequencer' option.
473 rw_sequencer=str If an offset modifier is given by appending a number to
474 the rw=<str> line, then this option controls how that
475 number modifies the IO offset being generated. Accepted
478 sequential Generate sequential offset
479 identical Generate the same offset
481 'sequential' is only useful for random IO, where fio would
482 normally generate a new random offset for every IO. If you
483 append eg 8 to randread, you would get a new random offset for
484 every 8 IO's. The result would be a seek for only every 8
485 IO's, instead of for every IO. Use rw=randread:8 to specify
486 that. As sequential IO is already sequential, setting
487 'sequential' for that would not result in any differences.
488 'identical' behaves in a similar fashion, except it sends
489 the same offset 8 number of times before generating a new
492 kb_base=int Select the interpretation of unit prefixes in input parameters.
493 1000 = Inputs comply with IEC 80000-13 and the International
494 System of Units (SI). Use:
495 - power-of-2 values with IEC prefixes (e.g., KiB)
496 - power-of-10 values with SI prefixes (e.g., kB)
497 1024 = Compatibility mode (default). To avoid breaking
499 - power-of-2 values with SI prefixes
500 - power-of-10 values with IEC prefixes
501 See bs= for more details on input parameters.
503 Outputs always use correct prefixes. Most outputs include
504 both side-by-side, like:
505 bw=2383.3kB/s (2327.4KiB/s)
506 If only one value is reported, then kb_base selects the
511 unified_rw_reporting=bool Fio normally reports statistics on a per
512 data direction basis, meaning that reads, writes, and trims are
513 accounted and reported separately. If this option is set,
514 the fio will sum the results and report them as "mixed"
517 randrepeat=bool For random IO workloads, seed the generator in a predictable
518 way so that results are repeatable across repetitions.
521 randseed=int Seed the random number generators based on this seed value, to
522 be able to control what sequence of output is being generated.
523 If not set, the random sequence depends on the randrepeat
526 fallocate=str Whether pre-allocation is performed when laying down files.
529 none Do not pre-allocate space
530 posix Pre-allocate via posix_fallocate()
531 keep Pre-allocate via fallocate() with
532 FALLOC_FL_KEEP_SIZE set
533 0 Backward-compatible alias for 'none'
534 1 Backward-compatible alias for 'posix'
536 May not be available on all supported platforms. 'keep' is only
537 available on Linux.If using ZFS on Solaris this must be set to
538 'none' because ZFS doesn't support it. Default: 'posix'.
540 fadvise_hint=bool By default, fio will use fadvise() to advise the kernel
541 on what IO patterns it is likely to issue. Sometimes you
542 want to test specific IO patterns without telling the
543 kernel about it, in which case you can disable this option.
544 The following options are supported:
546 sequential Use FADV_SEQUENTIAL
547 random Use FADV_RANDOM
548 1 Backwards-compatible hint for basing
549 the hint on the fio workload. Will use
550 FADV_SEQUENTIAL for a sequential
551 workload, and FADV_RANDOM for a random
553 0 Backwards-compatible setting for not
554 issing a fadvise hint.
556 fadvise_stream=int Notify the kernel what write stream ID to place these
557 writes under. Only supported on Linux. Note, this option
558 may change going forward.
560 size=int The total size of file io for this job. Fio will run until
561 this many bytes has been transferred, unless runtime is
562 limited by other options (such as 'runtime', for instance,
563 or increased/decreased by 'io_size'). Unless specific nrfiles
564 and filesize options are given, fio will divide this size
565 between the available files specified by the job. If not set,
566 fio will use the full size of the given files or devices.
567 If the files do not exist, size must be given. It is also
568 possible to give size as a percentage between 1 and 100. If
569 size=20% is given, fio will use 20% of the full size of the
570 given files or devices.
573 io_limit=int Normally fio operates within the region set by 'size', which
574 means that the 'size' option sets both the region and size of
575 IO to be performed. Sometimes that is not what you want. With
576 this option, it is possible to define just the amount of IO
577 that fio should do. For instance, if 'size' is set to 20GiB and
578 'io_size' is set to 5GiB, fio will perform IO within the first
579 20GiB but exit when 5GiB have been done. The opposite is also
580 possible - if 'size' is set to 20GiB, and 'io_size' is set to
581 40GiB, then fio will do 40GiB of IO within the 0..20GiB region.
583 filesize=int Individual file sizes. May be a range, in which case fio
584 will select sizes for files at random within the given range
585 and limited to 'size' in total (if that is given). If not
586 given, each created file is the same size.
588 file_append=bool Perform IO after the end of the file. Normally fio will
589 operate within the size of a file. If this option is set, then
590 fio will append to the file instead. This has identical
591 behavior to setting offset to the size of a file. This option
592 is ignored on non-regular files.
595 fill_fs=bool Sets size to something really large and waits for ENOSPC (no
596 space left on device) as the terminating condition. Only makes
597 sense with sequential write. For a read workload, the mount
598 point will be filled first then IO started on the result. This
599 option doesn't make sense if operating on a raw device node,
600 since the size of that is already known by the file system.
601 Additionally, writing beyond end-of-device will not return
604 blocksize=int[,int][,int]
606 The block size in bytes used for I/O units. Default: 4096.
607 A single value applies to reads, writes, and trims.
608 Comma-separated values may be specified for reads, writes,
609 and trims. A value not terminated in a comma applies to
613 bs=256k means 256k for reads, writes and trims
614 bs=8k,32k means 8k for reads, 32k for writes and trims
615 bs=8k,32k, means 8k for reads, 32k for writes, and
617 bs=,8k means default for reads, 8k for writes and trims
618 bs=,8k, means default for reads, 8k for writes, and
621 blocksize_range=irange[,irange][,irange]
622 bsrange=irange[,irange][,irange]
623 A range of block sizes in bytes for I/O units.
624 The issued I/O unit will always be a multiple of the minimum
625 size, unless blocksize_unaligned is set.
627 Comma-separated ranges may be specified for reads, writes,
628 and trims as described in 'blocksize'.
630 Example: bsrange=1k-4k,2k-8k
632 bssplit=str[,str][,str]
633 Sometimes you want even finer grained control of the
634 block sizes issued, not just an even split between them.
635 This option allows you to weight various block sizes,
636 so that you are able to define a specific amount of
637 block sizes issued. The format for this option is:
639 bssplit=blocksize/percentage:blocksize/percentage
641 for as many block sizes as needed. So if you want to define
642 a workload that has 50% 64k blocks, 10% 4k blocks, and
643 40% 32k blocks, you would write:
645 bssplit=4k/10:64k/50:32k/40
647 Ordering does not matter. If the percentage is left blank,
648 fio will fill in the remaining values evenly. So a bssplit
649 option like this one:
651 bssplit=4k/50:1k/:32k/
653 would have 50% 4k ios, and 25% 1k and 32k ios. The percentages
654 always add up to 100, if bssplit is given a range that adds
655 up to more, it will error out.
657 Comma-separated values may be specified for reads, writes,
658 and trims as described in 'blocksize'.
660 If you want a workload that has 50% 2k reads and 50% 4k reads,
661 while having 90% 4k writes and 10% 8k writes, you would
664 bssplit=2k/50:4k/50,4k/90:8k/10
667 bs_unaligned If set, fio will issue I/O units with any size within
668 blocksize_range, not just multiples of the minimum size.
669 This typically won't work with direct I/O, as that normally
670 requires sector alignment.
672 bs_is_seq_rand If this option is set, fio will use the normal read,write
673 blocksize settings as sequential,random blocksize settings
674 instead. Any random read or write will use the WRITE blocksize
675 settings, and any sequential read or write will use the READ
678 blockalign=int[,int][,int]
680 Boundary to which fio will align random I/O units.
681 Default: 'blocksize'.
682 Minimum alignment is typically 512b for using direct IO,
683 though it usually depends on the hardware block size. This
684 option is mutually exclusive with using a random map for
685 files, so it will turn off that option.
686 Comma-separated values may be specified for reads, writes,
687 and trims as described in 'blocksize'.
689 zero_buffers If this option is given, fio will init the IO buffers to
690 all zeroes. The default is to fill them with random data.
692 refill_buffers If this option is given, fio will refill the IO buffers
693 on every submit. The default is to only fill it at init
694 time and reuse that data. Only makes sense if zero_buffers
695 isn't specified, naturally. If data verification is enabled,
696 refill_buffers is also automatically enabled.
698 scramble_buffers=bool If refill_buffers is too costly and the target is
699 using data deduplication, then setting this option will
700 slightly modify the IO buffer contents to defeat normal
701 de-dupe attempts. This is not enough to defeat more clever
702 block compression attempts, but it will stop naive dedupe of
703 blocks. Default: true.
705 buffer_compress_percentage=int If this is set, then fio will attempt to
706 provide IO buffer content (on WRITEs) that compress to
707 the specified level. Fio does this by providing a mix of
708 random data and a fixed pattern. The fixed pattern is either
709 zeroes, or the pattern specified by buffer_pattern. If the
710 pattern option is used, it might skew the compression ratio
711 slightly. Note that this is per block size unit, for file/disk
712 wide compression level that matches this setting, you'll also
713 want to set refill_buffers.
715 buffer_compress_chunk=int See buffer_compress_percentage. This
716 setting allows fio to manage how big the ranges of random
717 data and zeroed data is. Without this set, fio will
718 provide buffer_compress_percentage of blocksize random
719 data, followed by the remaining zeroed. With this set
720 to some chunk size smaller than the block size, fio can
721 alternate random and zeroed data throughout the IO
724 buffer_pattern=str If set, fio will fill the io buffers with this
725 pattern. If not set, the contents of io buffers is defined by
726 the other options related to buffer contents. The setting can
727 be any pattern of bytes, and can be prefixed with 0x for hex
728 values. It may also be a string, where the string must then
729 be wrapped with "", e.g.:
731 buffer_pattern="abcd"
735 buffer_pattern=0xdeadface
737 Also you can combine everything together in any order:
738 buffer_pattern=0xdeadface"abcd"-12
740 dedupe_percentage=int If set, fio will generate this percentage of
741 identical buffers when writing. These buffers will be
742 naturally dedupable. The contents of the buffers depend on
743 what other buffer compression settings have been set. It's
744 possible to have the individual buffers either fully
745 compressible, or not at all. This option only controls the
746 distribution of unique buffers.
748 nrfiles=int Number of files to use for this job. Defaults to 1.
750 openfiles=int Number of files to keep open at the same time. Defaults to
751 the same as nrfiles, can be set smaller to limit the number
754 file_service_type=str Defines how fio decides which file from a job to
755 service next. The following types are defined:
757 random Just choose a file at random.
759 roundrobin Round robin over open files. This
762 sequential Finish one file before moving on to
763 the next. Multiple files can still be
764 open depending on 'openfiles'.
766 zipf Use a zipfian distribution to decide what file
769 pareto Use a pareto distribution to decide what file
772 gauss Use a gaussian (normal) distribution to decide
775 For random, roundrobin, and sequential, a postfix can be
776 appended to tell fio how many I/Os to issue before switching
777 to a new file. For example, specifying
778 'file_service_type=random:8' would cause fio to issue 8 I/Os
779 before selecting a new file at random. For the non-uniform
780 distributions, a floating point postfix can be given to
781 influence how the distribution is skewed. See
782 'random_distribution' for a description of how that would work.
784 ioengine=str Defines how the job issues io to the file. The following
787 sync Basic read(2) or write(2) io. lseek(2) is
788 used to position the io location.
790 psync Basic pread(2) or pwrite(2) io. Default on all
791 supported operating systems except for Windows.
793 vsync Basic readv(2) or writev(2) IO.
795 pvsync Basic preadv(2) or pwritev(2) IO.
797 pvsync2 Basic preadv2(2) or pwritev2(2) IO.
799 libaio Linux native asynchronous io. Note that Linux
800 may only support queued behaviour with
801 non-buffered IO (set direct=1 or buffered=0).
802 This engine defines engine specific options.
804 posixaio glibc posix asynchronous io.
806 solarisaio Solaris native asynchronous io.
808 windowsaio Windows native asynchronous io.
811 mmap File is memory mapped and data copied
812 to/from using memcpy(3).
814 splice splice(2) is used to transfer the data and
815 vmsplice(2) to transfer data from user
818 sg SCSI generic sg v3 io. May either be
819 synchronous using the SG_IO ioctl, or if
820 the target is an sg character device
821 we use read(2) and write(2) for asynchronous
824 null Doesn't transfer any data, just pretends
825 to. This is mainly used to exercise fio
826 itself and for debugging/testing purposes.
828 net Transfer over the network to given host:port.
829 Depending on the protocol used, the hostname,
830 port, listen and filename options are used to
831 specify what sort of connection to make, while
832 the protocol option determines which protocol
834 This engine defines engine specific options.
836 netsplice Like net, but uses splice/vmsplice to
837 map data and send/receive.
838 This engine defines engine specific options.
840 cpuio Doesn't transfer any data, but burns CPU
841 cycles according to the cpuload= and
842 cpuchunks= options. Setting cpuload=85
843 will cause that job to do nothing but burn
844 85% of the CPU. In case of SMP machines,
845 use numjobs=<no_of_cpu> to get desired CPU
846 usage, as the cpuload only loads a single
847 CPU at the desired rate. A job never finishes
848 unless there is at least one non-cpuio job.
850 guasi The GUASI IO engine is the Generic Userspace
851 Asyncronous Syscall Interface approach
854 http://www.xmailserver.org/guasi-lib.html
856 for more info on GUASI.
858 rdma The RDMA I/O engine supports both RDMA
859 memory semantics (RDMA_WRITE/RDMA_READ) and
860 channel semantics (Send/Recv) for the
861 InfiniBand, RoCE and iWARP protocols.
863 falloc IO engine that does regular fallocate to
864 simulate data transfer as fio ioengine.
865 DDIR_READ does fallocate(,mode = keep_size,)
866 DDIR_WRITE does fallocate(,mode = 0)
867 DDIR_TRIM does fallocate(,mode = punch_hole)
869 e4defrag IO engine that does regular EXT4_IOC_MOVE_EXT
870 ioctls to simulate defragment activity in
871 request to DDIR_WRITE event
873 rbd IO engine supporting direct access to Ceph
874 Rados Block Devices (RBD) via librbd without
875 the need to use the kernel rbd driver. This
876 ioengine defines engine specific options.
878 gfapi Using Glusterfs libgfapi sync interface to
879 direct access to Glusterfs volumes without
882 gfapi_async Using Glusterfs libgfapi async interface
883 to direct access to Glusterfs volumes without
884 having to go through FUSE. This ioengine
885 defines engine specific options.
887 libhdfs Read and write through Hadoop (HDFS).
888 This engine interprets offsets a little
889 differently. In HDFS, files once created
890 cannot be modified. So random writes are not
891 possible. To imitate this, libhdfs engine
892 creates bunch of small files, and engine will
893 pick a file out of those files based on the
894 offset generated by fio backend. Each jobs uses
895 it's own connection to HDFS.
897 mtd Read, write and erase an MTD character device
898 (e.g., /dev/mtd0). Discards are treated as
899 erases. Depending on the underlying device
900 type, the I/O may have to go in a certain
901 pattern, e.g., on NAND, writing sequentially
902 to erase blocks and discarding before
903 overwriting. The writetrim mode works well
906 pmemblk Read and write through the NVML libpmemblk
909 dev-dax Read and write through a DAX device exposed
910 from persistent memory.
912 external Prefix to specify loading an external
913 IO engine object file. Append the engine
914 filename, eg ioengine=external:/tmp/foo.o
915 to load ioengine foo.o in /tmp.
917 iodepth=int This defines how many I/O units to keep in flight against
918 the file. The default is 1 for each file defined in this
919 job, can be overridden with a larger value for higher
920 concurrency. Note that increasing iodepth beyond 1 will not
921 affect synchronous ioengines (except for small degress when
922 verify_async is in use). Even async engines may impose OS
923 restrictions causing the desired depth not to be achieved.
924 This may happen on Linux when using libaio and not setting
925 direct=1, since buffered IO is not async on that OS. Keep an
926 eye on the IO depth distribution in the fio output to verify
927 that the achieved depth is as expected. Default: 1.
929 iodepth_batch_submit=int
930 iodepth_batch=int This defines how many pieces of IO to submit at once.
931 It defaults to 1 which means that we submit each IO
932 as soon as it is available, but can be raised to submit
933 bigger batches of IO at the time. If it is set to 0 the iodepth
936 iodepth_batch_complete_min=int
937 iodepth_batch_complete=int This defines how many pieces of IO to retrieve
938 at once. It defaults to 1 which means that we'll ask
939 for a minimum of 1 IO in the retrieval process from
940 the kernel. The IO retrieval will go on until we
941 hit the limit set by iodepth_low. If this variable is
942 set to 0, then fio will always check for completed
943 events before queuing more IO. This helps reduce
944 IO latency, at the cost of more retrieval system calls.
946 iodepth_batch_complete_max=int This defines maximum pieces of IO to
947 retrieve at once. This variable should be used along with
948 iodepth_batch_complete_min=int variable, specifying the range
949 of min and max amount of IO which should be retrieved. By default
950 it is equal to iodepth_batch_complete_min value.
954 iodepth_batch_complete_min=1
955 iodepth_batch_complete_max=<iodepth>
957 which means that we will retrieve at least 1 IO and up to the
958 whole submitted queue depth. If none of IO has been completed
963 iodepth_batch_complete_min=0
964 iodepth_batch_complete_max=<iodepth>
966 which means that we can retrieve up to the whole submitted
967 queue depth, but if none of IO has been completed yet, we will
968 NOT wait and immediately exit the system call. In this example
969 we simply do polling.
971 iodepth_low=int The low water mark indicating when to start filling
972 the queue again. Defaults to the same as iodepth, meaning
973 that fio will attempt to keep the queue full at all times.
974 If iodepth is set to eg 16 and iodepth_low is set to 4, then
975 after fio has filled the queue of 16 requests, it will let
976 the depth drain down to 4 before starting to fill it again.
978 io_submit_mode=str This option controls how fio submits the IO to
979 the IO engine. The default is 'inline', which means that the
980 fio job threads submit and reap IO directly. If set to
981 'offload', the job threads will offload IO submission to a
982 dedicated pool of IO threads. This requires some coordination
983 and thus has a bit of extra overhead, especially for lower
984 queue depth IO where it can increase latencies. The benefit
985 is that fio can manage submission rates independently of
986 the device completion rates. This avoids skewed latency
987 reporting if IO gets back up on the device side (the
988 coordinated omission problem).
990 direct=bool If value is true, use non-buffered io. This is usually
991 O_DIRECT. Note that ZFS on Solaris doesn't support direct io.
992 On Windows the synchronous ioengines don't support direct io.
994 atomic=bool If value is true, attempt to use atomic direct IO. Atomic
995 writes are guaranteed to be stable once acknowledged by
996 the operating system. Only Linux supports O_ATOMIC right
999 buffered=bool If value is true, use buffered io. This is the opposite
1000 of the 'direct' option. Defaults to true.
1002 offset=int Start io at the given offset in the file. The data before
1003 the given offset will not be touched. This effectively
1004 caps the file size at real_size - offset.
1006 offset_increment=int If this is provided, then the real offset becomes
1007 offset + offset_increment * thread_number, where the thread
1008 number is a counter that starts at 0 and is incremented for
1009 each sub-job (i.e. when numjobs option is specified). This
1010 option is useful if there are several jobs which are intended
1011 to operate on a file in parallel disjoint segments, with
1012 even spacing between the starting points.
1014 number_ios=int Fio will normally perform IOs until it has exhausted the size
1015 of the region set by size=, or if it exhaust the allocated
1016 time (or hits an error condition). With this setting, the
1017 range/size can be set independently of the number of IOs to
1018 perform. When fio reaches this number, it will exit normally
1019 and report status. Note that this does not extend the amount
1020 of IO that will be done, it will only stop fio if this
1021 condition is met before other end-of-job criteria.
1023 fsync=int If writing to a file, issue a sync of the dirty data
1024 for every number of blocks given. For example, if you give
1025 32 as a parameter, fio will sync the file for every 32
1026 writes issued. If fio is using non-buffered io, we may
1027 not sync the file. The exception is the sg io engine, which
1028 synchronizes the disk cache anyway.
1030 fdatasync=int Like fsync= but uses fdatasync() to only sync data and not
1032 In FreeBSD and Windows there is no fdatasync(), this falls back
1035 sync_file_range=str:val Use sync_file_range() for every 'val' number of
1036 write operations. Fio will track range of writes that
1037 have happened since the last sync_file_range() call. 'str'
1038 can currently be one or more of:
1040 wait_before SYNC_FILE_RANGE_WAIT_BEFORE
1041 write SYNC_FILE_RANGE_WRITE
1042 wait_after SYNC_FILE_RANGE_WAIT_AFTER
1044 So if you do sync_file_range=wait_before,write:8, fio would
1045 use SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE for
1046 every 8 writes. Also see the sync_file_range(2) man page.
1047 This option is Linux specific.
1049 overwrite=bool If true, writes to a file will always overwrite existing
1050 data. If the file doesn't already exist, it will be
1051 created before the write phase begins. If the file exists
1052 and is large enough for the specified write phase, nothing
1055 end_fsync=bool If true, fsync file contents when a write stage has completed.
1057 fsync_on_close=bool If true, fio will fsync() a dirty file on close.
1058 This differs from end_fsync in that it will happen on every
1059 file close, not just at the end of the job.
1061 rwmixread=int How large a percentage of the mix should be reads.
1063 rwmixwrite=int How large a percentage of the mix should be writes. If both
1064 rwmixread and rwmixwrite is given and the values do not add
1065 up to 100%, the latter of the two will be used to override
1066 the first. This may interfere with a given rate setting,
1067 if fio is asked to limit reads or writes to a certain rate.
1068 If that is the case, then the distribution may be skewed.
1070 random_distribution=str:float[,str:float][,str:float]
1071 By default, fio will use a completely uniform
1072 random distribution when asked to perform random IO. Sometimes
1073 it is useful to skew the distribution in specific ways,
1074 ensuring that some parts of the data is more hot than others.
1075 fio includes the following distribution models:
1077 random Uniform random distribution
1078 zipf Zipf distribution
1079 pareto Pareto distribution
1080 gauss Normal (gaussian) distribution
1081 zoned Zoned random distribution
1083 When using a zipf or pareto distribution, an input value
1084 is also needed to define the access pattern. For zipf, this
1085 is the zipf theta. For pareto, it's the pareto power. Fio
1086 includes a test program, genzipf, that can be used visualize
1087 what the given input values will yield in terms of hit rates.
1088 If you wanted to use zipf with a theta of 1.2, you would use
1089 random_distribution=zipf:1.2 as the option. If a non-uniform
1090 model is used, fio will disable use of the random map. For
1091 the gauss distribution, a normal deviation is supplied as
1092 a value between 0 and 100.
1094 For a zoned distribution, fio supports specifying percentages
1095 of IO access that should fall within what range of the file or
1096 device. For example, given a criteria of:
1098 60% of accesses should be to the first 10%
1099 30% of accesses should be to the next 20%
1100 8% of accesses should be to to the next 30%
1101 2% of accesses should be to the next 40%
1103 we can define that through zoning of the random accesses. For
1104 the above example, the user would do:
1106 random_distribution=zoned:60/10:30/20:8/30:2/40
1108 similarly to how bssplit works for setting ranges and
1109 percentages of block sizes. Like bssplit, it's possible to
1110 specify separate zones for reads, writes, and trims. If just
1111 one set is given, it'll apply to all of them.
1113 percentage_random=int[,int][,int]
1114 For a random workload, set how big a percentage should
1115 be random. This defaults to 100%, in which case the workload
1116 is fully random. It can be set from anywhere from 0 to 100.
1117 Setting it to 0 would make the workload fully sequential. Any
1118 setting in between will result in a random mix of sequential
1119 and random IO, at the given percentages.
1120 Comma-separated values may be specified for reads, writes,
1121 and trims as described in 'blocksize'.
1123 norandommap Normally fio will cover every block of the file when doing
1124 random IO. If this option is given, fio will just get a
1125 new random offset without looking at past io history. This
1126 means that some blocks may not be read or written, and that
1127 some blocks may be read/written more than once. If this option
1128 is used with verify= and multiple blocksizes (via bsrange=),
1129 only intact blocks are verified, i.e., partially-overwritten
1132 softrandommap=bool See norandommap. If fio runs with the random block map
1133 enabled and it fails to allocate the map, if this option is
1134 set it will continue without a random block map. As coverage
1135 will not be as complete as with random maps, this option is
1136 disabled by default.
1138 random_generator=str Fio supports the following engines for generating
1139 IO offsets for random IO:
1141 tausworthe Strong 2^88 cycle random number generator
1142 lfsr Linear feedback shift register generator
1143 tausworthe64 Strong 64-bit 2^258 cycle random number
1146 Tausworthe is a strong random number generator, but it
1147 requires tracking on the side if we want to ensure that
1148 blocks are only read or written once. LFSR guarantees
1149 that we never generate the same offset twice, and it's
1150 also less computationally expensive. It's not a true
1151 random generator, however, though for IO purposes it's
1152 typically good enough. LFSR only works with single
1153 block sizes, not with workloads that use multiple block
1154 sizes. If used with such a workload, fio may read or write
1155 some blocks multiple times. The default value is tausworthe,
1156 unless the required space exceeds 2^32 blocks. If it does,
1157 then tausworthe64 is selected automatically.
1159 nice=int Run the job with the given nice value. See man nice(2).
1161 On Windows, values less than -15 set the process class to "High";
1162 -1 through -15 set "Above Normal"; 1 through 15 "Below Normal";
1163 and above 15 "Idle" priority class.
1165 prio=int Set the io priority value of this job. Linux limits us to
1166 a positive value between 0 and 7, with 0 being the highest.
1167 See man ionice(1). Refer to an appropriate manpage for
1168 other operating systems since meaning of priority may differ.
1170 prioclass=int Set the io priority class. See man ionice(1).
1172 thinktime=int Stall the job x microseconds after an io has completed before
1173 issuing the next. May be used to simulate processing being
1174 done by an application. See thinktime_blocks and
1178 Only valid if thinktime is set - pretend to spend CPU time
1179 doing something with the data received, before falling back
1180 to sleeping for the rest of the period specified by
1183 thinktime_blocks=int
1184 Only valid if thinktime is set - control how many blocks
1185 to issue, before waiting 'thinktime' usecs. If not set,
1186 defaults to 1 which will make fio wait 'thinktime' usecs
1187 after every block. This effectively makes any queue depth
1188 setting redundant, since no more than 1 IO will be queued
1189 before we have to complete it and do our thinktime. In
1190 other words, this setting effectively caps the queue depth
1191 if the latter is larger.
1193 rate=int[,int][,int]
1194 Cap the bandwidth used by this job. The number is in bytes/sec,
1195 the normal suffix rules apply.
1196 Comma-separated values may be specified for reads, writes,
1197 and trims as described in 'blocksize'.
1199 rate_min=int[,int][,int]
1200 Tell fio to do whatever it can to maintain at least this
1201 bandwidth. Failing to meet this requirement will cause
1203 Comma-separated values may be specified for reads, writes,
1204 and trims as described in 'blocksize'.
1206 rate_iops=int[,int][,int]
1207 Cap the bandwidth to this number of IOPS. Basically the same
1208 as rate, just specified independently of bandwidth. If the
1209 job is given a block size range instead of a fixed value,
1210 the smallest block size is used as the metric.
1211 Comma-separated values may be specified for reads, writes,
1212 and trims as described in 'blocksize'.
1214 rate_iops_min=int[,int][,int]
1215 If fio doesn't meet this rate of IO, it will cause
1217 Comma-separated values may be specified for reads, writes,
1218 and trims as described in 'blocksize'.
1220 rate_process=str This option controls how fio manages rated IO
1221 submissions. The default is 'linear', which submits IO in a
1222 linear fashion with fixed delays between IOs that gets
1223 adjusted based on IO completion rates. If this is set to
1224 'poisson', fio will submit IO based on a more real world
1225 random request flow, known as the Poisson process
1226 (https://en.wikipedia.org/wiki/Poisson_process). The lambda
1227 will be 10^6 / IOPS for the given workload.
1229 latency_target=int If set, fio will attempt to find the max performance
1230 point that the given workload will run at while maintaining a
1231 latency below this target. The values is given in microseconds.
1232 See latency_window and latency_percentile
1234 latency_window=int Used with latency_target to specify the sample window
1235 that the job is run at varying queue depths to test the
1236 performance. The value is given in microseconds.
1238 latency_percentile=float The percentage of IOs that must fall within the
1239 criteria specified by latency_target and latency_window. If not
1240 set, this defaults to 100.0, meaning that all IOs must be equal
1241 or below to the value set by latency_target.
1243 max_latency=int If set, fio will exit the job if it exceeds this maximum
1244 latency. It will exit with an ETIME error.
1246 rate_cycle=int Average bandwidth for 'rate' and 'rate_min' over this number
1249 cpumask=int Set the CPU affinity of this job. The parameter given is a
1250 bitmask of allowed CPU's the job may run on. So if you want
1251 the allowed CPUs to be 1 and 5, you would pass the decimal
1252 value of (1 << 1 | 1 << 5), or 34. See man
1253 sched_setaffinity(2). This may not work on all supported
1254 operating systems or kernel versions. This option doesn't
1255 work well for a higher CPU count than what you can store in
1256 an integer mask, so it can only control cpus 1-32. For
1257 boxes with larger CPU counts, use cpus_allowed.
1259 cpus_allowed=str Controls the same options as cpumask, but it allows a text
1260 setting of the permitted CPUs instead. So to use CPUs 1 and
1261 5, you would specify cpus_allowed=1,5. This options also
1262 allows a range of CPUs. Say you wanted a binding to CPUs
1263 1, 5, and 8-15, you would set cpus_allowed=1,5,8-15.
1265 cpus_allowed_policy=str Set the policy of how fio distributes the CPUs
1266 specified by cpus_allowed or cpumask. Two policies are
1269 shared All jobs will share the CPU set specified.
1270 split Each job will get a unique CPU from the CPU set.
1272 'shared' is the default behaviour, if the option isn't
1273 specified. If split is specified, then fio will will assign
1274 one cpu per job. If not enough CPUs are given for the jobs
1275 listed, then fio will roundrobin the CPUs in the set.
1277 numa_cpu_nodes=str Set this job running on specified NUMA nodes' CPUs. The
1278 arguments allow comma delimited list of cpu numbers,
1279 A-B ranges, or 'all'. Note, to enable numa options support,
1280 fio must be built on a system with libnuma-dev(el) installed.
1282 numa_mem_policy=str Set this job's memory policy and corresponding NUMA
1283 nodes. Format of the arguments:
1285 `mode' is one of the following memory policy:
1286 default, prefer, bind, interleave, local
1287 For `default' and `local' memory policy, no node is
1288 needed to be specified.
1289 For `prefer', only one node is allowed.
1290 For `bind' and `interleave', it allow comma delimited
1291 list of numbers, A-B ranges, or 'all'.
1293 startdelay=time Start this job the specified number of seconds after fio
1294 has started. Only useful if the job file contains several
1295 jobs, and you want to delay starting some jobs to a certain
1298 runtime=time Tell fio to terminate processing after the specified number
1299 of seconds. It can be quite hard to determine for how long
1300 a specified job will run, so this parameter is handy to
1301 cap the total runtime to a given time.
1303 time_based If set, fio will run for the duration of the runtime
1304 specified even if the file(s) are completely read or
1305 written. It will simply loop over the same workload
1306 as many times as the runtime allows.
1308 ramp_time=time If set, fio will run the specified workload for this amount
1309 of time before logging any performance numbers. Useful for
1310 letting performance settle before logging results, thus
1311 minimizing the runtime required for stable results. Note
1312 that the ramp_time is considered lead in time for a job,
1313 thus it will increase the total runtime if a special timeout
1314 or runtime is specified.
1316 steadystate=str:float
1317 ss=str:float Define the criterion and limit for assessing steady state
1318 performance. The first parameter designates the criterion
1319 whereas the second parameter sets the threshold. When the
1320 criterion falls below the threshold for the specified duration,
1321 the job will stop. For example, iops_slope:0.1% will direct fio
1322 to terminate the job when the least squares regression slope
1323 falls below 0.1% of the mean IOPS. If group_reporting is
1324 enabled this will apply to all jobs in the group. Below is the
1325 list of available steady state assessment criteria. All
1326 assessments are carried out using only data from the rolling
1327 collection window. Threshold limits can be expressed as a fixed
1328 value or as a percentage of the mean in the collection window.
1329 iops Collect IOPS data. Stop the job if all
1330 individual IOPS measurements are within the
1331 specified limit of the mean IOPS (e.g., iops:2
1332 means that all individual IOPS values must be
1333 within 2 of the mean, whereas iops:0.2% means
1334 that all individual IOPS values must be within
1335 0.2% of the mean IOPS to terminate the job).
1337 Collect IOPS data and calculate the least
1338 squares regression slope. Stop the job if the
1339 slope falls below the specified limit.
1340 bw Collect bandwidth data. Stop the job if all
1341 individual bandwidth measurements are within
1342 the specified limit of the mean bandwidth.
1344 Collect bandwidth data and calculate the least
1345 squares regression slope. Stop the job if the
1346 slope falls below the specified limit.
1348 steadystate_duration=time
1349 ss_dur=time A rolling window of this duration will be used to judge whether
1350 steady state has been reached. Data will be collected once per
1351 second. The default is 0 which disables steady state detection.
1353 steadystate_ramp_time=time
1354 ss_ramp=time Allow the job to run for the specified duration before
1355 beginning data collection for checking the steady state job
1356 termination criterion. The default is 0.
1358 invalidate=bool Invalidate the buffer/page cache parts for this file prior
1359 to starting io. Defaults to true.
1361 sync=bool Use sync io for buffered writes. For the majority of the
1362 io engines, this means using O_SYNC.
1365 mem=str Fio can use various types of memory as the I/O unit buffer.
1366 The allowed values are:
1368 malloc Use memory from malloc(3) as the buffers.
1369 Default memory type.
1371 shm Use shared memory as the buffers. Allocated
1374 shmhuge Same as shm, but use huge pages as backing.
1376 mmap Use mmap to allocate buffers. May either be
1377 anonymous memory, or can be file backed if
1378 a filename is given after the option. The
1379 format is mem=mmap:/path/to/file.
1381 mmaphuge Use a memory mapped huge file as the buffer
1382 backing. Append filename after mmaphuge, ala
1383 mem=mmaphuge:/hugetlbfs/file
1385 mmapshared Same as mmap, but use a MMAP_SHARED
1388 The area allocated is a function of the maximum allowed
1389 bs size for the job, multiplied by the io depth given. Note
1390 that for shmhuge and mmaphuge to work, the system must have
1391 free huge pages allocated. This can normally be checked
1392 and set by reading/writing /proc/sys/vm/nr_hugepages on a
1393 Linux system. Fio assumes a huge page is 4MiB in size. So
1394 to calculate the number of huge pages you need for a given
1395 job file, add up the io depth of all jobs (normally one unless
1396 iodepth= is used) and multiply by the maximum bs set. Then
1397 divide that number by the huge page size. You can see the
1398 size of the huge pages in /proc/meminfo. If no huge pages
1399 are allocated by having a non-zero number in nr_hugepages,
1400 using mmaphuge or shmhuge will fail. Also see hugepage-size.
1402 mmaphuge also needs to have hugetlbfs mounted and the file
1403 location should point there. So if it's mounted in /huge,
1404 you would use mem=mmaphuge:/huge/somefile.
1406 iomem_align=int This indicates the memory alignment of the IO memory buffers.
1407 Note that the given alignment is applied to the first I/O unit
1408 buffer, if using iodepth the alignment of the following buffers
1409 are given by the bs used. In other words, if using a bs that is
1410 a multiple of the page sized in the system, all buffers will
1411 be aligned to this value. If using a bs that is not page
1412 aligned, the alignment of subsequent IO memory buffers is the
1413 sum of the iomem_align and bs used.
1416 Defines the size of a huge page. Must at least be equal
1417 to the system setting, see /proc/meminfo. Defaults to 4MiB.
1418 Should probably always be a multiple of megabytes, so using
1419 hugepage-size=Xm is the preferred way to set this to avoid
1420 setting a non-pow-2 bad value.
1422 exitall When one job finishes, terminate the rest. The default is
1423 to wait for each job to finish, sometimes that is not the
1426 exitall_on_error When one job finishes in error, terminate the rest. The
1427 default is to wait for each job to finish.
1429 bwavgtime=int Average the calculated bandwidth over the given time. Value
1430 is specified in milliseconds. If the job also does bandwidth
1431 logging through 'write_bw_log', then the minimum of this option
1432 and 'log_avg_msec' will be used. Default: 500ms.
1434 iopsavgtime=int Average the calculated IOPS over the given time. Value
1435 is specified in milliseconds. If the job also does IOPS logging
1436 through 'write_iops_log', then the minimum of this option and
1437 'log_avg_msec' will be used. Default: 500ms.
1439 create_serialize=bool If true, serialize the file creation for the jobs.
1440 This may be handy to avoid interleaving of data
1441 files, which may greatly depend on the filesystem
1442 used and even the number of processors in the system.
1444 create_fsync=bool fsync the data file after creation. This is the
1447 create_on_open=bool Don't pre-setup the files for IO, just create open()
1448 when it's time to do IO to that file.
1450 create_only=bool If true, fio will only run the setup phase of the job.
1451 If files need to be laid out or updated on disk, only
1452 that will be done. The actual job contents are not
1455 allow_file_create=bool If true, fio is permitted to create files as part
1456 of its workload. This is the default behavior. If this
1457 option is false, then fio will error out if the files it
1458 needs to use don't already exist. Default: true.
1460 allow_mounted_write=bool If this isn't set, fio will abort jobs that
1461 are destructive (eg that write) to what appears to be a
1462 mounted device or partition. This should help catch creating
1463 inadvertently destructive tests, not realizing that the test
1464 will destroy data on the mounted file system. Default: false.
1466 pre_read=bool If this is given, files will be pre-read into memory before
1467 starting the given IO operation. This will also clear
1468 the 'invalidate' flag, since it is pointless to pre-read
1469 and then drop the cache. This will only work for IO engines
1470 that are seek-able, since they allow you to read the same data
1471 multiple times. Thus it will not work on eg network or splice
1474 unlink=bool Unlink the job files when done. Not the default, as repeated
1475 runs of that job would then waste time recreating the file
1476 set again and again.
1478 unlink_each_loop=bool Unlink job files after each iteration or loop.
1480 loops=int Run the specified number of iterations of this job. Used
1481 to repeat the same workload a given number of times. Defaults
1484 verify_only Do not perform specified workload---only verify data still
1485 matches previous invocation of this workload. This option
1486 allows one to check data multiple times at a later date
1487 without overwriting it. This option makes sense only for
1488 workloads that write data, and does not support workloads
1489 with the time_based option set.
1491 do_verify=bool Run the verify phase after a write phase. Only makes sense if
1492 verify is set. Defaults to 1.
1494 verify=str If writing to a file, fio can verify the file contents
1495 after each iteration of the job. Each verification method also implies
1496 verification of special header, which is written to the beginning of
1497 each block. This header also includes meta information, like offset
1498 of the block, block number, timestamp when block was written, etc.
1499 verify=str can be combined with verify_pattern=str option.
1500 The allowed values are:
1502 md5 Use an md5 sum of the data area and store
1503 it in the header of each block.
1505 crc64 Use an experimental crc64 sum of the data
1506 area and store it in the header of each
1509 crc32c Use a crc32c sum of the data area and store
1510 it in the header of each block.
1512 crc32c-intel Use hardware assisted crc32c calculation
1513 provided on SSE4.2 enabled processors. Falls
1514 back to regular software crc32c, if not
1515 supported by the system.
1517 crc32 Use a crc32 sum of the data area and store
1518 it in the header of each block.
1520 crc16 Use a crc16 sum of the data area and store
1521 it in the header of each block.
1523 crc7 Use a crc7 sum of the data area and store
1524 it in the header of each block.
1526 xxhash Use xxhash as the checksum function. Generally
1527 the fastest software checksum that fio
1530 sha512 Use sha512 as the checksum function.
1532 sha256 Use sha256 as the checksum function.
1534 sha1 Use optimized sha1 as the checksum function.
1536 meta This option is deprecated, since now meta information is
1537 included in generic verification header and meta verification
1538 happens by default. For detailed information see the description
1539 of the verify=str setting. This option is kept because of
1540 compatibility's sake with old configurations. Do not use it.
1542 pattern Verify a strict pattern. Normally fio includes
1543 a header with some basic information and
1544 checksumming, but if this option is set, only
1545 the specific pattern set with 'verify_pattern'
1548 null Only pretend to verify. Useful for testing
1549 internals with ioengine=null, not for much
1552 This option can be used for repeated burn-in tests of a
1553 system to make sure that the written data is also
1554 correctly read back. If the data direction given is
1555 a read or random read, fio will assume that it should
1556 verify a previously written file. If the data direction
1557 includes any form of write, the verify will be of the
1560 verifysort=bool If set, fio will sort written verify blocks when it deems
1561 it faster to read them back in a sorted manner. This is
1562 often the case when overwriting an existing file, since
1563 the blocks are already laid out in the file system. You
1564 can ignore this option unless doing huge amounts of really
1565 fast IO where the red-black tree sorting CPU time becomes
1568 verify_offset=int Swap the verification header with data somewhere else
1569 in the block before writing. Its swapped back before
1572 verify_interval=int Write the verification header at a finer granularity
1573 than the blocksize. It will be written for chunks the
1574 size of header_interval. blocksize should divide this
1577 verify_pattern=str If set, fio will fill the io buffers with this
1578 pattern. Fio defaults to filling with totally random
1579 bytes, but sometimes it's interesting to fill with a known
1580 pattern for io verification purposes. Depending on the
1581 width of the pattern, fio will fill 1/2/3/4 bytes of the
1582 buffer at the time(it can be either a decimal or a hex number).
1583 The verify_pattern if larger than a 32-bit quantity has to
1584 be a hex number that starts with either "0x" or "0X". Use
1585 with verify=str. Also, verify_pattern supports %o format,
1586 which means that for each block offset will be written and
1587 then verified back, e.g.:
1591 Or use combination of everything:
1592 verify_pattern=0xff%o"abcd"-12
1594 verify_fatal=bool Normally fio will keep checking the entire contents
1595 before quitting on a block verification failure. If this
1596 option is set, fio will exit the job on the first observed
1599 verify_dump=bool If set, dump the contents of both the original data
1600 block and the data block we read off disk to files. This
1601 allows later analysis to inspect just what kind of data
1602 corruption occurred. Off by default.
1604 verify_async=int Fio will normally verify IO inline from the submitting
1605 thread. This option takes an integer describing how many
1606 async offload threads to create for IO verification instead,
1607 causing fio to offload the duty of verifying IO contents
1608 to one or more separate threads. If using this offload
1609 option, even sync IO engines can benefit from using an
1610 iodepth setting higher than 1, as it allows them to have
1611 IO in flight while verifies are running.
1613 verify_async_cpus=str Tell fio to set the given CPU affinity on the
1614 async IO verification threads. See cpus_allowed for the
1617 verify_backlog=int Fio will normally verify the written contents of a
1618 job that utilizes verify once that job has completed. In
1619 other words, everything is written then everything is read
1620 back and verified. You may want to verify continually
1621 instead for a variety of reasons. Fio stores the meta data
1622 associated with an IO block in memory, so for large
1623 verify workloads, quite a bit of memory would be used up
1624 holding this meta data. If this option is enabled, fio
1625 will write only N blocks before verifying these blocks.
1627 verify_backlog_batch=int Control how many blocks fio will verify
1628 if verify_backlog is set. If not set, will default to
1629 the value of verify_backlog (meaning the entire queue
1630 is read back and verified). If verify_backlog_batch is
1631 less than verify_backlog then not all blocks will be verified,
1632 if verify_backlog_batch is larger than verify_backlog, some
1633 blocks will be verified more than once.
1635 verify_state_save=bool When a job exits during the write phase of a verify
1636 workload, save its current state. This allows fio to replay
1637 up until that point, if the verify state is loaded for the
1638 verify read phase. The format of the filename is, roughly,
1639 <type>-<jobname>-<jobindex>-verify.state. <type> is "local"
1640 for a local run, "sock" for a client/server socket connection,
1641 and "ip" (192.168.0.1, for instance) for a networked
1642 client/server connection.
1644 verify_state_load=bool If a verify termination trigger was used, fio stores
1645 the current write state of each thread. This can be used at
1646 verification time so that fio knows how far it should verify.
1647 Without this information, fio will run a full verification
1648 pass, according to the settings in the job file used.
1651 wait_for_previous Wait for preceding jobs in the job file to exit, before
1652 starting this one. Can be used to insert serialization
1653 points in the job file. A stone wall also implies starting
1654 a new reporting group.
1656 new_group Start a new reporting group. See: group_reporting.
1658 numjobs=int Create the specified number of clones of this job. May be
1659 used to setup a larger number of threads/processes doing
1660 the same thing. Each thread is reported separately; to see
1661 statistics for all clones as a whole, use group_reporting in
1662 conjunction with new_group.
1664 group_reporting It may sometimes be interesting to display statistics for
1665 groups of jobs as a whole instead of for each individual job.
1666 This is especially true if 'numjobs' is used; looking at
1667 individual thread/process output quickly becomes unwieldy.
1668 To see the final report per-group instead of per-job, use
1669 'group_reporting'. Jobs in a file will be part of the same
1670 reporting group, unless if separated by a stonewall, or by
1673 thread fio defaults to forking jobs, however if this option is
1674 given, fio will use pthread_create(3) to create threads
1677 zonesize=int Divide a file into zones of the specified size. See zoneskip.
1679 zoneskip=int Skip the specified number of bytes when zonesize data has
1680 been read. The two zone options can be used to only do
1681 io on zones of a file.
1683 write_iolog=str Write the issued io patterns to the specified file. See
1684 read_iolog. Specify a separate file for each job, otherwise
1685 the iologs will be interspersed and the file may be corrupt.
1687 read_iolog=str Open an iolog with the specified file name and replay the
1688 io patterns it contains. This can be used to store a
1689 workload and replay it sometime later. The iolog given
1690 may also be a blktrace binary file, which allows fio
1691 to replay a workload captured by blktrace. See blktrace
1692 for how to capture such logging data. For blktrace replay,
1693 the file needs to be turned into a blkparse binary data
1694 file first (blkparse <device> -o /dev/null -d file_for_fio.bin).
1696 replay_no_stall=int When replaying I/O with read_iolog the default behavior
1697 is to attempt to respect the time stamps within the log and
1698 replay them with the appropriate delay between IOPS. By
1699 setting this variable fio will not respect the timestamps and
1700 attempt to replay them as fast as possible while still
1701 respecting ordering. The result is the same I/O pattern to a
1702 given device, but different timings.
1704 replay_redirect=str While replaying I/O patterns using read_iolog the
1705 default behavior is to replay the IOPS onto the major/minor
1706 device that each IOP was recorded from. This is sometimes
1707 undesirable because on a different machine those major/minor
1708 numbers can map to a different device. Changing hardware on
1709 the same system can also result in a different major/minor
1710 mapping. Replay_redirect causes all IOPS to be replayed onto
1711 the single specified device regardless of the device it was
1712 recorded from. i.e. replay_redirect=/dev/sdc would cause all
1713 IO in the blktrace or iolog to be replayed onto /dev/sdc.
1714 This means multiple devices will be replayed onto a single
1715 device, if the trace contains multiple devices. If you want
1716 multiple devices to be replayed concurrently to multiple
1717 redirected devices you must blkparse your trace into separate
1718 traces and replay them with independent fio invocations.
1719 Unfortunately this also breaks the strict time ordering
1720 between multiple device accesses.
1722 replay_align=int Force alignment of IO offsets and lengths in a trace
1723 to this power of 2 value.
1725 replay_scale=int Scale sector offsets down by this factor when
1728 per_job_logs=bool If set, this generates bw/clat/iops log with per
1729 file private filenames. If not set, jobs with identical names
1730 will share the log filename. Default: true.
1732 write_bw_log=str If given, write a bandwidth log of the jobs in this job
1733 file. Can be used to store data of the bandwidth of the
1734 jobs in their lifetime. The included fio_generate_plots
1735 script uses gnuplot to turn these text files into nice
1736 graphs. See write_lat_log for behaviour of given
1737 filename. For this option, the suffix is _bw.x.log, where
1738 x is the index of the job (1..N, where N is the number of
1739 jobs). If 'per_job_logs' is false, then the filename will not
1740 include the job index. See 'Log File Formats'.
1742 write_lat_log=str Same as write_bw_log, except that this option stores io
1743 submission, completion, and total latencies instead. If no
1744 filename is given with this option, the default filename of
1745 "jobname_type.log" is used. Even if the filename is given,
1746 fio will still append the type of log. So if one specifies
1750 The actual log names will be foo_slat.x.log, foo_clat.x.log,
1751 and foo_lat.x.log, where x is the index of the job (1..N,
1752 where N is the number of jobs). This helps fio_generate_plot
1753 find the logs automatically. If 'per_job_logs' is false, then
1754 the filename will not include the job index. See 'Log File
1757 write_hist_log=str Same as write_lat_log, but writes I/O completion
1758 latency histograms. If no filename is given with this option, the
1759 default filename of "jobname_clat_hist.x.log" is used, where x is
1760 the index of the job (1..N, where N is the number of jobs). Even
1761 if the filename is given, fio will still append the type of log.
1762 If per_job_logs is false, then the filename will not include the
1763 job index. See 'Log File Formats'.
1765 write_iops_log=str Same as write_bw_log, but writes IOPS. If no filename is
1766 given with this option, the default filename of
1767 "jobname_type.x.log" is used,where x is the index of the job
1768 (1..N, where N is the number of jobs). Even if the filename
1769 is given, fio will still append the type of log. If
1770 'per_job_logs' is false, then the filename will not include
1771 the job index. See 'Log File Formats'.
1773 log_avg_msec=int By default, fio will log an entry in the iops, latency,
1774 or bw log for every IO that completes. When writing to the
1775 disk log, that can quickly grow to a very large size. Setting
1776 this option makes fio average the each log entry over the
1777 specified period of time, reducing the resolution of the log.
1778 See log_max_value as well. Defaults to 0, logging all entries.
1780 log_hist_msec=int Same as log_avg_msec, but logs entries for completion
1781 latency histograms. Computing latency percentiles from averages of
1782 intervals using log_avg_msec is innacurate. Setting this option makes
1783 fio log histogram entries over the specified period of time, reducing
1784 log sizes for high IOPS devices while retaining percentile accuracy.
1785 See log_hist_coarseness as well. Defaults to 0, meaning histogram
1786 logging is disabled.
1788 log_hist_coarseness=int Integer ranging from 0 to 6, defining the coarseness
1789 of the resolution of the histogram logs enabled with log_hist_msec. For
1790 each increment in coarseness, fio outputs half as many bins. Defaults to
1791 0, for which histogram logs contain 1216 latency bins. See
1794 log_max_value=bool If log_avg_msec is set, fio logs the average over that
1795 window. If you instead want to log the maximum value, set this
1796 option to 1. Defaults to 0, meaning that averaged values are
1799 log_offset=int If this is set, the iolog options will include the byte
1800 offset for the IO entry as well as the other data values.
1802 log_compression=int If this is set, fio will compress the IO logs as
1803 it goes, to keep the memory footprint lower. When a log
1804 reaches the specified size, that chunk is removed and
1805 compressed in the background. Given that IO logs are
1806 fairly highly compressible, this yields a nice memory
1807 savings for longer runs. The downside is that the
1808 compression will consume some background CPU cycles, so
1809 it may impact the run. This, however, is also true if
1810 the logging ends up consuming most of the system memory.
1811 So pick your poison. The IO logs are saved normally at the
1812 end of a run, by decompressing the chunks and storing them
1813 in the specified log file. This feature depends on the
1814 availability of zlib.
1816 log_compression_cpus=str Define the set of CPUs that are allowed to
1817 handle online log compression for the IO jobs. This can
1818 provide better isolation between performance sensitive jobs,
1819 and background compression work.
1821 log_store_compressed=bool If set, fio will store the log files in a
1822 compressed format. They can be decompressed with fio, using
1823 the --inflate-log command line parameter. The files will be
1824 stored with a .fz suffix.
1826 log_unix_epoch=bool If set, fio will log Unix timestamps to the log
1827 files produced by enabling write_type_log for each log type, instead
1828 of the default zero-based timestamps.
1830 block_error_percentiles=bool If set, record errors in trim block-sized
1831 units from writes and trims and output a histogram of
1832 how many trims it took to get to errors, and what kind
1833 of error was encountered.
1835 lockmem=int Pin down the specified amount of memory with mlock(2). Can
1836 potentially be used instead of removing memory or booting
1837 with less memory to simulate a smaller amount of memory.
1838 The amount specified is per worker.
1840 exec_prerun=str Before running this job, issue the command specified
1841 through system(3). Output is redirected in a file called
1844 exec_postrun=str After the job completes, issue the command specified
1845 though system(3). Output is redirected in a file called
1846 jobname.postrun.txt.
1848 ioscheduler=str Attempt to switch the device hosting the file to the specified
1849 io scheduler before running.
1851 disk_util=bool Generate disk utilization statistics, if the platform
1852 supports it. Defaults to on.
1854 disable_lat=bool Disable measurements of total latency numbers. Useful
1855 only for cutting back the number of calls to gettimeofday,
1856 as that does impact performance at really high IOPS rates.
1857 Note that to really get rid of a large amount of these
1858 calls, this option must be used with disable_slat and
1861 disable_clat=bool Disable measurements of completion latency numbers. See
1864 disable_slat=bool Disable measurements of submission latency numbers. See
1867 disable_bw=bool Disable measurements of throughput/bandwidth numbers. See
1870 clat_percentiles=bool Enable the reporting of percentiles of
1871 completion latencies.
1873 percentile_list=float_list Overwrite the default list of percentiles
1874 for completion latencies and the block error histogram.
1875 Each number is a floating number in the range (0,100],
1876 and the maximum length of the list is 20. Use ':'
1877 to separate the numbers, and list the numbers in ascending
1878 order. For example, --percentile_list=99.5:99.9 will cause
1879 fio to report the values of completion latency below which
1880 99.5% and 99.9% of the observed latencies fell, respectively.
1882 clocksource=str Use the given clocksource as the base of timing. The
1883 supported options are:
1885 gettimeofday gettimeofday(2)
1887 clock_gettime clock_gettime(2)
1889 cpu Internal CPU clock source
1891 cpu is the preferred clocksource if it is reliable, as it
1892 is very fast (and fio is heavy on time calls). Fio will
1893 automatically use this clocksource if it's supported and
1894 considered reliable on the system it is running on, unless
1895 another clocksource is specifically set. For x86/x86-64 CPUs,
1896 this means supporting TSC Invariant.
1898 gtod_reduce=bool Enable all of the gettimeofday() reducing options
1899 (disable_clat, disable_slat, disable_bw) plus reduce
1900 precision of the timeout somewhat to really shrink
1901 the gettimeofday() call count. With this option enabled,
1902 we only do about 0.4% of the gtod() calls we would have
1903 done if all time keeping was enabled.
1905 gtod_cpu=int Sometimes it's cheaper to dedicate a single thread of
1906 execution to just getting the current time. Fio (and
1907 databases, for instance) are very intensive on gettimeofday()
1908 calls. With this option, you can set one CPU aside for
1909 doing nothing but logging current time to a shared memory
1910 location. Then the other threads/processes that run IO
1911 workloads need only copy that segment, instead of entering
1912 the kernel with a gettimeofday() call. The CPU set aside
1913 for doing these time calls will be excluded from other
1914 uses. Fio will manually clear it from the CPU mask of other
1917 continue_on_error=str Normally fio will exit the job on the first observed
1918 failure. If this option is set, fio will continue the job when
1919 there is a 'non-fatal error' (EIO or EILSEQ) until the runtime
1920 is exceeded or the I/O size specified is completed. If this
1921 option is used, there are two more stats that are appended,
1922 the total error count and the first error. The error field
1923 given in the stats is the first error that was hit during the
1926 The allowed values are:
1928 none Exit on any IO or verify errors.
1930 read Continue on read errors, exit on all others.
1932 write Continue on write errors, exit on all others.
1934 io Continue on any IO error, exit on all others.
1936 verify Continue on verify errors, exit on all others.
1938 all Continue on all errors.
1940 0 Backward-compatible alias for 'none'.
1942 1 Backward-compatible alias for 'all'.
1944 ignore_error=str Sometimes you want to ignore some errors during test
1945 in that case you can specify error list for each error type.
1946 ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST
1947 errors for given error type is separated with ':'. Error
1948 may be symbol ('ENOSPC', 'ENOMEM') or integer.
1950 ignore_error=EAGAIN,ENOSPC:122
1951 This option will ignore EAGAIN from READ, and ENOSPC and
1952 122(EDQUOT) from WRITE.
1954 error_dump=bool If set dump every error even if it is non fatal, true
1955 by default. If disabled only fatal error will be dumped
1957 cgroup=str Add job to this control group. If it doesn't exist, it will
1958 be created. The system must have a mounted cgroup blkio
1959 mount point for this to work. If your system doesn't have it
1960 mounted, you can do so with:
1962 # mount -t cgroup -o blkio none /cgroup
1964 cgroup_weight=int Set the weight of the cgroup to this value. See
1965 the documentation that comes with the kernel, allowed values
1966 are in the range of 100..1000.
1968 cgroup_nodelete=bool Normally fio will delete the cgroups it has created after
1969 the job completion. To override this behavior and to leave
1970 cgroups around after the job completion, set cgroup_nodelete=1.
1971 This can be useful if one wants to inspect various cgroup
1972 files after job completion. Default: false
1974 uid=int Instead of running as the invoking user, set the user ID to
1975 this value before the thread/process does any work.
1977 gid=int Set group ID, see uid.
1979 flow_id=int The ID of the flow. If not specified, it defaults to being a
1980 global flow. See flow.
1982 flow=int Weight in token-based flow control. If this value is used, then
1983 there is a 'flow counter' which is used to regulate the
1984 proportion of activity between two or more jobs. fio attempts
1985 to keep this flow counter near zero. The 'flow' parameter
1986 stands for how much should be added or subtracted to the flow
1987 counter on each iteration of the main I/O loop. That is, if
1988 one job has flow=8 and another job has flow=-1, then there
1989 will be a roughly 1:8 ratio in how much one runs vs the other.
1991 flow_watermark=int The maximum value that the absolute value of the flow
1992 counter is allowed to reach before the job must wait for a
1993 lower value of the counter.
1995 flow_sleep=int The period of time, in microseconds, to wait after the flow
1996 watermark has been exceeded before retrying operations
1998 In addition, there are some parameters which are only valid when a specific
1999 ioengine is in use. These are used identically to normal parameters, with the
2000 caveat that when used on the command line, they must come after the ioengine
2001 that defines them is selected.
2003 [libaio] userspace_reap Normally, with the libaio engine in use, fio will use
2004 the io_getevents system call to reap newly returned events.
2005 With this flag turned on, the AIO ring will be read directly
2006 from user-space to reap events. The reaping mode is only
2007 enabled when polling for a minimum of 0 events (eg when
2008 iodepth_batch_complete=0).
2010 [psyncv2] hipri Set RWF_HIPRI on IO, indicating to the kernel that
2011 it's of higher priority than normal.
2013 [cpuio] cpuload=int Attempt to use the specified percentage of CPU cycles.
2015 [cpuio] cpuchunks=int Split the load into cycles of the given time. In
2018 [cpuio] exit_on_io_done=bool Detect when IO threads are done, then exit.
2020 [netsplice] hostname=str
2021 [net] hostname=str The host name or IP address to use for TCP or UDP based IO.
2022 If the job is a TCP listener or UDP reader, the hostname is not
2023 used and must be omitted unless it is a valid UDP multicast
2025 [libhdfs] namenode=str The host name or IP address of a HDFS cluster namenode to contact.
2027 [netsplice] port=int
2028 [net] port=int The TCP or UDP port to bind to or connect to. If this is used
2029 with numjobs to spawn multiple instances of the same job type, then this will
2030 be the starting port number since fio will use a range of ports.
2031 [libhdfs] port=int the listening port of the HFDS cluster namenode.
2033 [netsplice] interface=str
2034 [net] interface=str The IP address of the network interface used to send or
2035 receive UDP multicast
2038 [net] ttl=int Time-to-live value for outgoing UDP multicast packets.
2041 [netsplice] nodelay=bool
2042 [net] nodelay=bool Set TCP_NODELAY on TCP connections.
2044 [netsplice] protocol=str
2045 [netsplice] proto=str
2047 [net] proto=str The network protocol to use. Accepted values are:
2049 tcp Transmission control protocol
2050 tcpv6 Transmission control protocol V6
2051 udp User datagram protocol
2052 udpv6 User datagram protocol V6
2053 unix UNIX domain socket
2055 When the protocol is TCP or UDP, the port must also be given,
2056 as well as the hostname if the job is a TCP listener or UDP
2057 reader. For unix sockets, the normal filename option should be
2058 used and the port is invalid.
2060 [net] listen For TCP network connections, tell fio to listen for incoming
2061 connections rather than initiating an outgoing connection. The
2062 hostname must be omitted if this option is used.
2064 [net] pingpong Normally a network writer will just continue writing data, and
2065 a network reader will just consume packages. If pingpong=1
2066 is set, a writer will send its normal payload to the reader,
2067 then wait for the reader to send the same payload back. This
2068 allows fio to measure network latencies. The submission
2069 and completion latencies then measure local time spent
2070 sending or receiving, and the completion latency measures
2071 how long it took for the other end to receive and send back.
2072 For UDP multicast traffic pingpong=1 should only be set for a
2073 single reader when multiple readers are listening to the same
2076 [net] window_size Set the desired socket buffer size for the connection.
2078 [net] mss Set the TCP maximum segment size (TCP_MAXSEG).
2080 [e4defrag] donorname=str
2081 File will be used as a block donor(swap extents between files)
2082 [e4defrag] inplace=int
2083 Configure donor file blocks allocation strategy
2084 0(default): Preallocate donor's file on init
2085 1 : allocate space immediately inside defragment event,
2086 and free right after event
2088 [rbd] clustername=str Specifies the name of the Ceph cluster.
2089 [rbd] rbdname=str Specifies the name of the RBD.
2090 [rbd] pool=str Specifies the name of the Ceph pool containing RBD.
2091 [rbd] clientname=str Specifies the username (without the 'client.' prefix)
2092 used to access the Ceph cluster. If the clustername is
2093 specified, the clientname shall be the full type.id
2094 string. If no type. prefix is given, fio will add
2095 'client.' by default.
2097 [mtd] skip_bad=bool Skip operations against known bad blocks.
2099 [libhdfs] hdfsdirectory libhdfs will create chunk in this HDFS directory
2100 [libhdfs] chunk_size the size of the chunk to use for each file.
2103 6.0 Interpreting the output
2104 ---------------------------
2106 fio spits out a lot of output. While running, fio will display the
2107 status of the jobs created. An example of that would be:
2109 Jobs: 1: [_r] [24.8% done] [r=20992KiB/s,w=24064KiB/s,t=0KiB/s] [r=82,w=94,t=0 iops] [eta 00h:01m:31s]
2111 The characters inside the square brackets denote the current status of
2112 each thread. The possible values (in typical life cycle order) are:
2116 P Thread setup, but not started.
2118 I Thread initialized, waiting or generating necessary data.
2119 p Thread running pre-reading file(s).
2120 R Running, doing sequential reads.
2121 r Running, doing random reads.
2122 W Running, doing sequential writes.
2123 w Running, doing random writes.
2124 M Running, doing mixed sequential reads/writes.
2125 m Running, doing mixed random reads/writes.
2126 F Running, currently waiting for fsync()
2127 f Running, finishing up (writing IO logs, etc)
2128 V Running, doing verification of written data.
2129 E Thread exited, not reaped by main thread yet.
2131 X Thread reaped, exited with an error.
2132 K Thread reaped, exited due to signal.
2134 Fio will condense the thread string as not to take up more space on the
2135 command line as is needed. For instance, if you have 10 readers and 10
2136 writers running, the output would look like this:
2138 Jobs: 20 (f=20): [R(10),W(10)] [4.0% done] [r=20992KiB/s,w=24064KiB/s,t=0KiB/s] [r=82,w=94,t=0 iops] [eta 57m:36s]
2140 Fio will still maintain the ordering, though. So the above means that jobs
2141 1..10 are readers, and 11..20 are writers.
2143 The other values are fairly self explanatory - number of threads
2144 currently running and doing io, rate of io since last check (read speed
2145 listed first, then write speed), and the estimated completion percentage
2146 and time for the running group. It's impossible to estimate runtime of
2147 the following groups (if any). Note that the string is displayed in order,
2148 so it's possible to tell which of the jobs are currently doing what. The
2149 first character is the first job defined in the job file, and so forth.
2151 When fio is done (or interrupted by ctrl-c), it will show the data for
2152 each thread, group of threads, and disks in that order. For each data
2153 direction, the output looks like:
2155 Client1 (g=0): err= 0:
2156 write: io= 32MiB, bw= 666KiB/s, iops=89 , runt= 50320msec
2157 slat (msec): min= 0, max= 136, avg= 0.03, stdev= 1.92
2158 clat (msec): min= 0, max= 631, avg=48.50, stdev=86.82
2159 bw (KiB/s) : min= 0, max= 1196, per=51.00%, avg=664.02, stdev=681.68
2160 cpu : usr=1.49%, sys=0.25%, ctx=7969, majf=0, minf=17
2161 IO depths : 1=0.1%, 2=0.3%, 4=0.5%, 8=99.0%, 16=0.0%, 32=0.0%, >32=0.0%
2162 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
2163 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
2164 issued r/w: total=0/32768, short=0/0
2165 lat (msec): 2=1.6%, 4=0.0%, 10=3.2%, 20=12.8%, 50=38.4%, 100=24.8%,
2166 lat (msec): 250=15.2%, 500=0.0%, 750=0.0%, 1000=0.0%, >=2048=0.0%
2168 The client number is printed, along with the group id and error of that
2169 thread. Below is the io statistics, here for writes. In the order listed,
2172 io= Number of megabytes io performed
2173 bw= Average bandwidth rate
2174 iops= Average IOs performed per second
2175 runt= The runtime of that thread
2176 slat= Submission latency (avg being the average, stdev being the
2177 standard deviation). This is the time it took to submit
2178 the io. For sync io, the slat is really the completion
2179 latency, since queue/complete is one operation there. This
2180 value can be in milliseconds or microseconds, fio will choose
2181 the most appropriate base and print that. In the example
2182 above, milliseconds is the best scale. Note: in --minimal mode
2183 latencies are always expressed in microseconds.
2184 clat= Completion latency. Same names as slat, this denotes the
2185 time from submission to completion of the io pieces. For
2186 sync io, clat will usually be equal (or very close) to 0,
2187 as the time from submit to complete is basically just
2188 CPU time (io has already been done, see slat explanation).
2189 bw= Bandwidth. Same names as the xlat stats, but also includes
2190 an approximate percentage of total aggregate bandwidth
2191 this thread received in this group. This last value is
2192 only really useful if the threads in this group are on the
2193 same disk, since they are then competing for disk access.
2194 cpu= CPU usage. User and system time, along with the number
2195 of context switches this thread went through, usage of
2196 system and user time, and finally the number of major
2197 and minor page faults. The CPU utilization numbers are
2198 averages for the jobs in that reporting group, while the
2199 context and fault counters are summed.
2200 IO depths= The distribution of io depths over the job life time. The
2201 numbers are divided into powers of 2, so for example the
2202 16= entries includes depths up to that value but higher
2203 than the previous entry. In other words, it covers the
2204 range from 16 to 31.
2205 IO submit= How many pieces of IO were submitting in a single submit
2206 call. Each entry denotes that amount and below, until
2207 the previous entry - eg, 8=100% mean that we submitted
2208 anywhere in between 5-8 ios per submit call.
2209 IO complete= Like the above submit number, but for completions instead.
2210 IO issued= The number of read/write requests issued, and how many
2212 IO latencies= The distribution of IO completion latencies. This is the
2213 time from when IO leaves fio and when it gets completed.
2214 The numbers follow the same pattern as the IO depths,
2215 meaning that 2=1.6% means that 1.6% of the IO completed
2216 within 2 msecs, 20=12.8% means that 12.8% of the IO
2217 took more than 10 msecs, but less than (or equal to) 20 msecs.
2219 After each client has been listed, the group statistics are printed. They
2220 will look like this:
2222 Run status group 0 (all jobs):
2223 READ: io=64MB, aggrb=22178, minb=11355, maxb=11814, mint=2840msec, maxt=2955msec
2224 WRITE: io=64MB, aggrb=1302, minb=666, maxb=669, mint=50093msec, maxt=50320msec
2226 For each data direction, it prints:
2228 io= Number of megabytes io performed.
2229 aggrb= Aggregate bandwidth of threads in this group.
2230 minb= The minimum average bandwidth a thread saw.
2231 maxb= The maximum average bandwidth a thread saw.
2232 mint= The smallest runtime of the threads in that group.
2233 maxt= The longest runtime of the threads in that group.
2235 And finally, the disk statistics are printed. They will look like this:
2237 Disk stats (read/write):
2238 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
2240 Each value is printed for both reads and writes, with reads first. The
2243 ios= Number of ios performed by all groups.
2244 merge= Number of merges io the io scheduler.
2245 ticks= Number of ticks we kept the disk busy.
2246 io_queue= Total time spent in the disk queue.
2247 util= The disk utilization. A value of 100% means we kept the disk
2248 busy constantly, 50% would be a disk idling half of the time.
2250 It is also possible to get fio to dump the current output while it is
2251 running, without terminating the job. To do that, send fio the USR1 signal.
2252 You can also get regularly timed dumps by using the --status-interval
2253 parameter, or by creating a file in /tmp named fio-dump-status. If fio
2254 sees this file, it will unlink it and dump the current output status.
2260 For scripted usage where you typically want to generate tables or graphs
2261 of the results, fio can output the results in a semicolon separated format.
2262 The format is one long line of values, such as:
2264 2;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%
2265 A description of this job goes here.
2267 The job description (if provided) follows on a second line.
2269 To enable terse output, use the --minimal command line option. The first
2270 value is the version of the terse output format. If the output has to
2271 be changed for some reason, this number will be incremented by 1 to
2272 signify that change.
2274 Split up, the format is as follows:
2276 terse version, fio version, jobname, groupid, error
2278 Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec)
2279 Submission latency: min, max, mean, stdev (usec)
2280 Completion latency: min, max, mean, stdev (usec)
2281 Completion latency percentiles: 20 fields (see below)
2282 Total latency: min, max, mean, stdev (usec)
2283 Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev
2285 Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec)
2286 Submission latency: min, max, mean, stdev (usec)
2287 Completion latency: min, max, mean, stdev(usec)
2288 Completion latency percentiles: 20 fields (see below)
2289 Total latency: min, max, mean, stdev (usec)
2290 Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev
2291 CPU usage: user, system, context switches, major faults, minor faults
2292 IO depths: <=1, 2, 4, 8, 16, 32, >=64
2293 IO latencies microseconds: <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
2294 IO latencies milliseconds: <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
2295 Disk utilization: Disk name, Read ios, write ios,
2296 Read merges, write merges,
2297 Read ticks, write ticks,
2298 Time spent in queue, disk utilization percentage
2299 Additional Info (dependent on continue_on_error, default off): total # errors, first error code
2301 Additional Info (dependent on description being set): Text description
2303 Completion latency percentiles can be a grouping of up to 20 sets, so
2304 for the terse output fio writes all of them. Each field will look like this:
2308 which is the Xth percentile, and the usec latency associated with it.
2310 For disk utilization, all disks used by fio are shown. So for each disk
2311 there will be a disk utilization section.
2314 8.0 Trace file format
2315 ---------------------
2316 There are two trace file format that you can encounter. The older (v1) format
2317 is unsupported since version 1.20-rc3 (March 2008). It will still be described
2318 below in case that you get an old trace and want to understand it.
2320 In any case the trace is a simple text file with a single action per line.
2323 8.1 Trace file format v1
2324 ------------------------
2325 Each line represents a single io action in the following format:
2329 where rw=0/1 for read/write, and the offset and length entries being in bytes.
2331 This format is not supported in Fio versions => 1.20-rc3.
2334 8.2 Trace file format v2
2335 ------------------------
2336 The second version of the trace file format was added in Fio version 1.17.
2337 It allows to access more then one file per trace and has a bigger set of
2338 possible file actions.
2340 The first line of the trace file has to be:
2344 Following this can be lines in two different formats, which are described below.
2346 The file management format:
2350 The filename is given as an absolute path. The action can be one of these:
2352 add Add the given filename to the trace
2353 open Open the file with the given filename. The filename has to have
2354 been added with the add action before.
2355 close Close the file with the given filename. The file has to have been
2359 The file io action format:
2361 filename action offset length
2363 The filename is given as an absolute path, and has to have been added and opened
2364 before it can be used with this format. The offset and length are given in
2365 bytes. The action can be one of these:
2367 wait Wait for 'offset' microseconds. Everything below 100 is discarded.
2368 The time is relative to the previous wait statement.
2369 read Read 'length' bytes beginning from 'offset'
2370 write Write 'length' bytes beginning from 'offset'
2371 sync fsync() the file
2372 datasync fdatasync() the file
2373 trim trim the given file from the given 'offset' for 'length' bytes
2376 9.0 CPU idleness profiling
2377 --------------------------
2378 In some cases, we want to understand CPU overhead in a test. For example,
2379 we test patches for the specific goodness of whether they reduce CPU usage.
2380 fio implements a balloon approach to create a thread per CPU that runs at
2381 idle priority, meaning that it only runs when nobody else needs the cpu.
2382 By measuring the amount of work completed by the thread, idleness of each
2383 CPU can be derived accordingly.
2385 An unit work is defined as touching a full page of unsigned characters. Mean
2386 and standard deviation of time to complete an unit work is reported in "unit
2387 work" section. Options can be chosen to report detailed percpu idleness or
2388 overall system idleness by aggregating percpu stats.
2391 10.0 Verification and triggers
2392 ------------------------------
2393 Fio is usually run in one of two ways, when data verification is done. The
2394 first is a normal write job of some sort with verify enabled. When the
2395 write phase has completed, fio switches to reads and verifies everything
2396 it wrote. The second model is running just the write phase, and then later
2397 on running the same job (but with reads instead of writes) to repeat the
2398 same IO patterns and verify the contents. Both of these methods depend
2399 on the write phase being completed, as fio otherwise has no idea how much
2402 With verification triggers, fio supports dumping the current write state
2403 to local files. Then a subsequent read verify workload can load this state
2404 and know exactly where to stop. This is useful for testing cases where
2405 power is cut to a server in a managed fashion, for instance.
2407 A verification trigger consists of two things:
2409 1) Storing the write state of each job
2410 2) Executing a trigger command
2412 The write state is relatively small, on the order of hundreds of bytes
2413 to single kilobytes. It contains information on the number of completions
2414 done, the last X completions, etc.
2416 A trigger is invoked either through creation ('touch') of a specified
2417 file in the system, or through a timeout setting. If fio is run with
2418 --trigger-file=/tmp/trigger-file, then it will continually check for
2419 the existence of /tmp/trigger-file. When it sees this file, it will
2420 fire off the trigger (thus saving state, and executing the trigger
2423 For client/server runs, there's both a local and remote trigger. If
2424 fio is running as a server backend, it will send the job states back
2425 to the client for safe storage, then execute the remote trigger, if
2426 specified. If a local trigger is specified, the server will still send
2427 back the write state, but the client will then execute the trigger.
2429 10.1 Verification trigger example
2430 ---------------------------------
2431 Lets say we want to run a powercut test on the remote machine 'server'.
2432 Our write workload is in write-test.fio. We want to cut power to 'server'
2433 at some point during the run, and we'll run this test from the safety
2434 or our local machine, 'localbox'. On the server, we'll start the fio
2437 server# fio --server
2439 and on the client, we'll fire off the workload:
2441 localbox$ fio --client=server --trigger-file=/tmp/my-trigger --trigger-remote="bash -c \"echo b > /proc/sysrq-triger\""
2443 We set /tmp/my-trigger as the trigger file, and we tell fio to execute
2445 echo b > /proc/sysrq-trigger
2447 on the server once it has received the trigger and sent us the write
2448 state. This will work, but it's not _really_ cutting power to the server,
2449 it's merely abruptly rebooting it. If we have a remote way of cutting
2450 power to the server through IPMI or similar, we could do that through
2451 a local trigger command instead. Lets assume we have a script that does
2452 IPMI reboot of a given hostname, ipmi-reboot. On localbox, we could
2453 then have run fio with a local trigger instead:
2455 localbox$ fio --client=server --trigger-file=/tmp/my-trigger --trigger="ipmi-reboot server"
2457 For this case, fio would wait for the server to send us the write state,
2458 then execute 'ipmi-reboot server' when that happened.
2460 10.2 Loading verify state
2461 -------------------------
2462 To load store write state, read verification job file must contain
2463 the verify_state_load option. If that is set, fio will load the previously
2464 stored state. For a local fio run this is done by loading the files directly,
2465 and on a client/server run, the server backend will ask the client to send
2466 the files over and load them from there.
2469 11.0 Log File Formats
2470 ---------------------
2472 Fio supports a variety of log file formats, for logging latencies, bandwidth,
2473 and IOPS. The logs share a common format, which looks like this:
2475 time (msec), value, data direction, offset
2477 Time for the log entry is always in milliseconds. The value logged depends
2478 on the type of log, it will be one of the following:
2480 Latency log Value is latency in usecs
2481 Bandwidth log Value is in KiB/sec
2482 IOPS log Value is IOPS
2484 Data direction is one of the following:
2490 The offset is the offset, in bytes, from the start of the file, for that
2491 particular IO. The logging of the offset can be toggled with 'log_offset'.
2493 If windowed logging is enabled through 'log_avg_msec', then fio doesn't log
2494 individual IOs. Instead of logs the average values over the specified
2495 period of time. Since 'data direction' and 'offset' are per-IO values,
2496 they aren't applicable if windowed logging is enabled. If windowed logging
2497 is enabled and 'log_max_value' is set, then fio logs maximum values in
2498 that window instead of averages.
projects / fio.git / blob