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 128MB 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 32KB and define numjobs to 4 to
158 fork 4 identical jobs. The result is 4 processes each randomly writing
159 to their own 64MB 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.
279 int SI integer. A whole number value, which may contain a suffix
280 describing the base of the number. Accepted suffixes are k/m/g/t/p,
281 meaning kilo, mega, giga, tera, and peta. The suffix is not case
282 sensitive, and you may also include trailing 'b' (eg 'kb' is the same
283 as 'k'). So if you want to specify 4096, you could either write
284 out '4096' or just give 4k. The suffixes signify base 2 values, so
285 1024 is 1k and 1024k is 1m and so on, unless the suffix is explicitly
286 set to a base 10 value using 'kib', 'mib', 'gib', etc. If that is the
287 case, then 1000 is used as the multiplier. This can be handy for
288 disks, since manufacturers generally use base 10 values when listing
289 the capacity of a drive. If the option accepts an upper and lower
290 range, use a colon ':' or minus '-' to separate such values. May also
291 include a prefix to indicate numbers base. If 0x is used, the number
292 is assumed to be hexadecimal. See irange.
293 bool Boolean. Usually parsed as an integer, however only defined for
294 true and false (1 and 0).
295 irange Integer range with suffix. Allows value range to be given, such
296 as 1024-4096. A colon may also be used as the separator, eg
297 1k:4k. If the option allows two sets of ranges, they can be
298 specified with a ',' or '/' delimiter: 1k-4k/8k-32k. Also see
300 float_list A list of floating numbers, separated by a ':' character.
302 With the above in mind, here follows the complete list of fio job
305 name=str ASCII name of the job. This may be used to override the
306 name printed by fio for this job. Otherwise the job
307 name is used. On the command line this parameter has the
308 special purpose of also signaling the start of a new
311 wait_for=str Specifies the name of the already defined job to wait
312 for. Single waitee name only may be specified. If set, the job
313 won't be started until all workers of the waitee job are done.
315 Wait_for operates on the job name basis, so there are a few
316 limitations. First, the waitee must be defined prior to the
317 waiter job (meaning no forward references). Second, if a job
318 is being referenced as a waitee, it must have a unique name
319 (no duplicate waitees).
321 description=str Text description of the job. Doesn't do anything except
322 dump this text description when this job is run. It's
325 directory=str Prefix filenames with this directory. Used to place files
326 in a different location than "./". See the 'filename' option
327 for escaping certain characters.
329 filename=str Fio normally makes up a filename based on the job name,
330 thread number, and file number. If you want to share
331 files between threads in a job or several jobs, specify
332 a filename for each of them to override the default.
333 If the ioengine is file based, you can specify a number of
334 files by separating the names with a ':' colon. So if you
335 wanted a job to open /dev/sda and /dev/sdb as the two working
336 files, you would use filename=/dev/sda:/dev/sdb. On Windows,
337 disk devices are accessed as \\.\PhysicalDrive0 for the first
338 device, \\.\PhysicalDrive1 for the second etc. Note: Windows
339 and FreeBSD prevent write access to areas of the disk
340 containing in-use data (e.g. filesystems).
341 If the wanted filename does need to include a colon, then
342 escape that with a '\' character. For instance, if the filename
343 is "/dev/dsk/foo@3,0:c", then you would use
344 filename="/dev/dsk/foo@3,0\:c". '-' is a reserved name, meaning
345 stdin or stdout. Which of the two depends on the read/write
349 If sharing multiple files between jobs, it is usually necessary
350 to have fio generate the exact names that you want. By default,
351 fio will name a file based on the default file format
352 specification of jobname.jobnumber.filenumber. With this
353 option, that can be customized. Fio will recognize and replace
354 the following keywords in this string:
357 The name of the worker thread or process.
360 The incremental number of the worker thread or
364 The incremental number of the file for that worker
367 To have dependent jobs share a set of files, this option can
368 be set to have fio generate filenames that are shared between
369 the two. For instance, if testfiles.$filenum is specified,
370 file number 4 for any job will be named testfiles.4. The
371 default of $jobname.$jobnum.$filenum will be used if
372 no other format specifier is given.
374 unique_filename=bool To avoid collisions between networked clients, fio
375 defaults to prefixing any generated filenames (with a directory
376 specified) with the source of the client connecting. To disable
377 this behavior, set this option to 0.
379 opendir=str Tell fio to recursively add any file it can find in this
380 directory and down the file system tree.
382 lockfile=str Fio defaults to not locking any files before it does
383 IO to them. If a file or file descriptor is shared, fio
384 can serialize IO to that file to make the end result
385 consistent. This is usual for emulating real workloads that
386 share files. The lock modes are:
388 none No locking. The default.
389 exclusive Only one thread/process may do IO,
390 excluding all others.
391 readwrite Read-write locking on the file. Many
392 readers may access the file at the
393 same time, but writes get exclusive
397 rw=str Type of io pattern. Accepted values are:
399 read Sequential reads
400 write Sequential writes
401 randwrite Random writes
402 randread Random reads
403 rw,readwrite Sequential mixed reads and writes
404 randrw Random mixed reads and writes
405 trimwrite Mixed trims and writes. Blocks will be
406 trimmed first, then written to.
408 Fio defaults to read if the option is not specified.
409 For the mixed io types, the default is to split them 50/50.
410 For certain types of io the result may still be skewed a bit,
411 since the speed may be different. It is possible to specify
412 a number of IO's to do before getting a new offset, this is
413 done by appending a ':<nr>' to the end of the string given.
414 For a random read, it would look like 'rw=randread:8' for
415 passing in an offset modifier with a value of 8. If the
416 suffix is used with a sequential IO pattern, then the value
417 specified will be added to the generated offset for each IO.
418 For instance, using rw=write:4k will skip 4k for every
419 write. It turns sequential IO into sequential IO with holes.
420 See the 'rw_sequencer' option.
422 rw_sequencer=str If an offset modifier is given by appending a number to
423 the rw=<str> line, then this option controls how that
424 number modifies the IO offset being generated. Accepted
427 sequential Generate sequential offset
428 identical Generate the same offset
430 'sequential' is only useful for random IO, where fio would
431 normally generate a new random offset for every IO. If you
432 append eg 8 to randread, you would get a new random offset for
433 every 8 IO's. The result would be a seek for only every 8
434 IO's, instead of for every IO. Use rw=randread:8 to specify
435 that. As sequential IO is already sequential, setting
436 'sequential' for that would not result in any differences.
437 'identical' behaves in a similar fashion, except it sends
438 the same offset 8 number of times before generating a new
441 kb_base=int The base unit for a kilobyte. The defacto base is 2^10, 1024.
442 Storage manufacturers like to use 10^3 or 1000 as a base
443 ten unit instead, for obvious reasons. Allow values are
444 1024 or 1000, with 1024 being the default.
446 unified_rw_reporting=bool Fio normally reports statistics on a per
447 data direction basis, meaning that read, write, and trim are
448 accounted and reported separately. If this option is set,
449 the fio will sum the results and report them as "mixed"
452 randrepeat=bool For random IO workloads, seed the generator in a predictable
453 way so that results are repeatable across repetitions.
456 randseed=int Seed the random number generators based on this seed value, to
457 be able to control what sequence of output is being generated.
458 If not set, the random sequence depends on the randrepeat
461 fallocate=str Whether pre-allocation is performed when laying down files.
464 none Do not pre-allocate space
465 posix Pre-allocate via posix_fallocate()
466 keep Pre-allocate via fallocate() with
467 FALLOC_FL_KEEP_SIZE set
468 0 Backward-compatible alias for 'none'
469 1 Backward-compatible alias for 'posix'
471 May not be available on all supported platforms. 'keep' is only
472 available on Linux.If using ZFS on Solaris this must be set to
473 'none' because ZFS doesn't support it. Default: 'posix'.
475 fadvise_hint=bool By default, fio will use fadvise() to advise the kernel
476 on what IO patterns it is likely to issue. Sometimes you
477 want to test specific IO patterns without telling the
478 kernel about it, in which case you can disable this option.
479 If set, fio will use POSIX_FADV_SEQUENTIAL for sequential
480 IO and POSIX_FADV_RANDOM for random IO.
482 fadvise_stream=int Notify the kernel what write stream ID to place these
483 writes under. Only supported on Linux. Note, this option
484 may change going forward.
486 size=int The total size of file io for this job. Fio will run until
487 this many bytes has been transferred, unless runtime is
488 limited by other options (such as 'runtime', for instance,
489 or increased/decreased by 'io_size'). Unless specific nrfiles
490 and filesize options are given, fio will divide this size
491 between the available files specified by the job. If not set,
492 fio will use the full size of the given files or devices.
493 If the files do not exist, size must be given. It is also
494 possible to give size as a percentage between 1 and 100. If
495 size=20% is given, fio will use 20% of the full size of the
496 given files or devices.
499 io_limit=int Normally fio operates within the region set by 'size', which
500 means that the 'size' option sets both the region and size of
501 IO to be performed. Sometimes that is not what you want. With
502 this option, it is possible to define just the amount of IO
503 that fio should do. For instance, if 'size' is set to 20G and
504 'io_size' is set to 5G, fio will perform IO within the first
505 20G but exit when 5G have been done. The opposite is also
506 possible - if 'size' is set to 20G, and 'io_size' is set to
507 40G, then fio will do 40G of IO within the 0..20G region.
509 filesize=int Individual file sizes. May be a range, in which case fio
510 will select sizes for files at random within the given range
511 and limited to 'size' in total (if that is given). If not
512 given, each created file is the same size.
514 file_append=bool Perform IO after the end of the file. Normally fio will
515 operate within the size of a file. If this option is set, then
516 fio will append to the file instead. This has identical
517 behavior to setting offset to the size of a file. This option
518 is ignored on non-regular files.
521 fill_fs=bool Sets size to something really large and waits for ENOSPC (no
522 space left on device) as the terminating condition. Only makes
523 sense with sequential write. For a read workload, the mount
524 point will be filled first then IO started on the result. This
525 option doesn't make sense if operating on a raw device node,
526 since the size of that is already known by the file system.
527 Additionally, writing beyond end-of-device will not return
531 bs=int The block size used for the io units. Defaults to 4k. Values
532 can be given for both read and writes. If a single int is
533 given, it will apply to both. If a second int is specified
534 after a comma, it will apply to writes only. In other words,
535 the format is either bs=read_and_write or bs=read,write,trim.
536 bs=4k,8k will thus use 4k blocks for reads, 8k blocks for
537 writes, and 8k for trims. You can terminate the list with
538 a trailing comma. bs=4k,8k, would use the default value for
539 trims.. If you only wish to set the write size, you
540 can do so by passing an empty read size - bs=,8k will set
541 8k for writes and leave the read default value.
544 ba=int At what boundary to align random IO offsets. Defaults to
545 the same as 'blocksize' the minimum blocksize given.
546 Minimum alignment is typically 512b for using direct IO,
547 though it usually depends on the hardware block size. This
548 option is mutually exclusive with using a random map for
549 files, so it will turn off that option.
551 blocksize_range=irange
552 bsrange=irange Instead of giving a single block size, specify a range
553 and fio will mix the issued io block sizes. The issued
554 io unit will always be a multiple of the minimum value
555 given (also see bs_unaligned). Applies to both reads and
556 writes, however a second range can be given after a comma.
559 bssplit=str Sometimes you want even finer grained control of the
560 block sizes issued, not just an even split between them.
561 This option allows you to weight various block sizes,
562 so that you are able to define a specific amount of
563 block sizes issued. The format for this option is:
565 bssplit=blocksize/percentage:blocksize/percentage
567 for as many block sizes as needed. So if you want to define
568 a workload that has 50% 64k blocks, 10% 4k blocks, and
569 40% 32k blocks, you would write:
571 bssplit=4k/10:64k/50:32k/40
573 Ordering does not matter. If the percentage is left blank,
574 fio will fill in the remaining values evenly. So a bssplit
575 option like this one:
577 bssplit=4k/50:1k/:32k/
579 would have 50% 4k ios, and 25% 1k and 32k ios. The percentages
580 always add up to 100, if bssplit is given a range that adds
581 up to more, it will error out.
583 bssplit also supports giving separate splits to reads and
584 writes. The format is identical to what bs= accepts. You
585 have to separate the read and write parts with a comma. So
586 if you want a workload that has 50% 2k reads and 50% 4k reads,
587 while having 90% 4k writes and 10% 8k writes, you would
590 bssplit=2k/50:4k/50,4k/90:8k/10
593 bs_unaligned If this option is given, any byte size value within bsrange
594 may be used as a block range. This typically wont work with
595 direct IO, as that normally requires sector alignment.
597 bs_is_seq_rand If this option is set, fio will use the normal read,write
598 blocksize settings as sequential,random instead. Any random
599 read or write will use the WRITE blocksize settings, and any
600 sequential read or write will use the READ blocksize setting.
602 zero_buffers If this option is given, fio will init the IO buffers to
603 all zeroes. The default is to fill them with random data.
605 refill_buffers If this option is given, fio will refill the IO buffers
606 on every submit. The default is to only fill it at init
607 time and reuse that data. Only makes sense if zero_buffers
608 isn't specified, naturally. If data verification is enabled,
609 refill_buffers is also automatically enabled.
611 scramble_buffers=bool If refill_buffers is too costly and the target is
612 using data deduplication, then setting this option will
613 slightly modify the IO buffer contents to defeat normal
614 de-dupe attempts. This is not enough to defeat more clever
615 block compression attempts, but it will stop naive dedupe of
616 blocks. Default: true.
618 buffer_compress_percentage=int If this is set, then fio will attempt to
619 provide IO buffer content (on WRITEs) that compress to
620 the specified level. Fio does this by providing a mix of
621 random data and a fixed pattern. The fixed pattern is either
622 zeroes, or the pattern specified by buffer_pattern. If the
623 pattern option is used, it might skew the compression ratio
624 slightly. Note that this is per block size unit, for file/disk
625 wide compression level that matches this setting, you'll also
626 want to set refill_buffers.
628 buffer_compress_chunk=int See buffer_compress_percentage. This
629 setting allows fio to manage how big the ranges of random
630 data and zeroed data is. Without this set, fio will
631 provide buffer_compress_percentage of blocksize random
632 data, followed by the remaining zeroed. With this set
633 to some chunk size smaller than the block size, fio can
634 alternate random and zeroed data throughout the IO
637 buffer_pattern=str If set, fio will fill the io buffers with this
638 pattern. If not set, the contents of io buffers is defined by
639 the other options related to buffer contents. The setting can
640 be any pattern of bytes, and can be prefixed with 0x for hex
641 values. It may also be a string, where the string must then
642 be wrapped with "", e.g.:
644 buffer_pattern="abcd"
648 buffer_pattern=0xdeadface
650 Also you can combine everything together in any order:
651 buffer_pattern=0xdeadface"abcd"-12
653 dedupe_percentage=int If set, fio will generate this percentage of
654 identical buffers when writing. These buffers will be
655 naturally dedupable. The contents of the buffers depend on
656 what other buffer compression settings have been set. It's
657 possible to have the individual buffers either fully
658 compressible, or not at all. This option only controls the
659 distribution of unique buffers.
661 nrfiles=int Number of files to use for this job. Defaults to 1.
663 openfiles=int Number of files to keep open at the same time. Defaults to
664 the same as nrfiles, can be set smaller to limit the number
667 file_service_type=str Defines how fio decides which file from a job to
668 service next. The following types are defined:
670 random Just choose a file at random.
672 roundrobin Round robin over open files. This
675 sequential Finish one file before moving on to
676 the next. Multiple files can still be
677 open depending on 'openfiles'.
679 zipf Use a zipfian distribution to decide what file
682 pareto Use a pareto distribution to decide what file
685 gauss Use a gaussian (normal) distribution to decide
688 For random, roundrobin, and sequential, a postfix can be
689 appended to tell fio how many I/Os to issue before switching
690 to a new file. For example, specifying
691 'file_service_type=random:8' would cause fio to issue 8 I/Os
692 before selecting a new file at random. For the non-uniform
693 distributions, a floating point postfix can be given to
694 influence how the distribution is skewed. See
695 'random_distribution' for a description of how that would work.
697 ioengine=str Defines how the job issues io to the file. The following
700 sync Basic read(2) or write(2) io. lseek(2) is
701 used to position the io location.
703 psync Basic pread(2) or pwrite(2) io. Default on all
704 supported operating systems except for Windows.
706 vsync Basic readv(2) or writev(2) IO.
708 pvsync Basic preadv(2) or pwritev(2) IO.
710 pvsync2 Basic preadv2(2) or pwritev2(2) IO.
712 libaio Linux native asynchronous io. Note that Linux
713 may only support queued behaviour with
714 non-buffered IO (set direct=1 or buffered=0).
715 This engine defines engine specific options.
717 posixaio glibc posix asynchronous io.
719 solarisaio Solaris native asynchronous io.
721 windowsaio Windows native asynchronous io.
724 mmap File is memory mapped and data copied
725 to/from using memcpy(3).
727 splice splice(2) is used to transfer the data and
728 vmsplice(2) to transfer data from user
731 sg SCSI generic sg v3 io. May either be
732 synchronous using the SG_IO ioctl, or if
733 the target is an sg character device
734 we use read(2) and write(2) for asynchronous
737 null Doesn't transfer any data, just pretends
738 to. This is mainly used to exercise fio
739 itself and for debugging/testing purposes.
741 net Transfer over the network to given host:port.
742 Depending on the protocol used, the hostname,
743 port, listen and filename options are used to
744 specify what sort of connection to make, while
745 the protocol option determines which protocol
747 This engine defines engine specific options.
749 netsplice Like net, but uses splice/vmsplice to
750 map data and send/receive.
751 This engine defines engine specific options.
753 cpuio Doesn't transfer any data, but burns CPU
754 cycles according to the cpuload= and
755 cpuchunks= options. Setting cpuload=85
756 will cause that job to do nothing but burn
757 85% of the CPU. In case of SMP machines,
758 use numjobs=<no_of_cpu> to get desired CPU
759 usage, as the cpuload only loads a single
760 CPU at the desired rate.
762 guasi The GUASI IO engine is the Generic Userspace
763 Asyncronous Syscall Interface approach
766 http://www.xmailserver.org/guasi-lib.html
768 for more info on GUASI.
770 rdma The RDMA I/O engine supports both RDMA
771 memory semantics (RDMA_WRITE/RDMA_READ) and
772 channel semantics (Send/Recv) for the
773 InfiniBand, RoCE and iWARP protocols.
775 falloc IO engine that does regular fallocate to
776 simulate data transfer as fio ioengine.
777 DDIR_READ does fallocate(,mode = keep_size,)
778 DDIR_WRITE does fallocate(,mode = 0)
779 DDIR_TRIM does fallocate(,mode = punch_hole)
781 e4defrag IO engine that does regular EXT4_IOC_MOVE_EXT
782 ioctls to simulate defragment activity in
783 request to DDIR_WRITE event
785 rbd IO engine supporting direct access to Ceph
786 Rados Block Devices (RBD) via librbd without
787 the need to use the kernel rbd driver. This
788 ioengine defines engine specific options.
790 gfapi Using Glusterfs libgfapi sync interface to
791 direct access to Glusterfs volumes without
794 gfapi_async Using Glusterfs libgfapi async interface
795 to direct access to Glusterfs volumes without
796 having to go through FUSE. This ioengine
797 defines engine specific options.
799 libhdfs Read and write through Hadoop (HDFS).
800 This engine interprets offsets a little
801 differently. In HDFS, files once created
802 cannot be modified. So random writes are not
803 possible. To imitate this, libhdfs engine
804 creates bunch of small files, and engine will
805 pick a file out of those files based on the
806 offset enerated by fio backend. Each jobs uses
807 it's own connection to HDFS.
809 mtd Read, write and erase an MTD character device
810 (e.g., /dev/mtd0). Discards are treated as
811 erases. Depending on the underlying device
812 type, the I/O may have to go in a certain
813 pattern, e.g., on NAND, writing sequentially
814 to erase blocks and discarding before
815 overwriting. The writetrim mode works well
818 pmemblk Read and write through the NVML libpmemblk
821 external Prefix to specify loading an external
822 IO engine object file. Append the engine
823 filename, eg ioengine=external:/tmp/foo.o
824 to load ioengine foo.o in /tmp.
826 iodepth=int This defines how many io units to keep in flight against
827 the file. The default is 1 for each file defined in this
828 job, can be overridden with a larger value for higher
829 concurrency. Note that increasing iodepth beyond 1 will not
830 affect synchronous ioengines (except for small degress when
831 verify_async is in use). Even async engines may impose OS
832 restrictions causing the desired depth not to be achieved.
833 This may happen on Linux when using libaio and not setting
834 direct=1, since buffered IO is not async on that OS. Keep an
835 eye on the IO depth distribution in the fio output to verify
836 that the achieved depth is as expected. Default: 1.
838 iodepth_batch_submit=int
839 iodepth_batch=int This defines how many pieces of IO to submit at once.
840 It defaults to 1 which means that we submit each IO
841 as soon as it is available, but can be raised to submit
842 bigger batches of IO at the time. If it is set to 0 the iodepth
845 iodepth_batch_complete_min=int
846 iodepth_batch_complete=int This defines how many pieces of IO to retrieve
847 at once. It defaults to 1 which means that we'll ask
848 for a minimum of 1 IO in the retrieval process from
849 the kernel. The IO retrieval will go on until we
850 hit the limit set by iodepth_low. If this variable is
851 set to 0, then fio will always check for completed
852 events before queuing more IO. This helps reduce
853 IO latency, at the cost of more retrieval system calls.
855 iodepth_batch_complete_max=int This defines maximum pieces of IO to
856 retrieve at once. This variable should be used along with
857 iodepth_batch_complete_min=int variable, specifying the range
858 of min and max amount of IO which should be retrieved. By default
859 it is equal to iodepth_batch_complete_min value.
863 iodepth_batch_complete_min=1
864 iodepth_batch_complete_max=<iodepth>
866 which means that we will retrieve at leat 1 IO and up to the
867 whole submitted queue depth. If none of IO has been completed
872 iodepth_batch_complete_min=0
873 iodepth_batch_complete_max=<iodepth>
875 which means that we can retrieve up to the whole submitted
876 queue depth, but if none of IO has been completed yet, we will
877 NOT wait and immediately exit the system call. In this example
878 we simply do polling.
880 iodepth_low=int The low water mark indicating when to start filling
881 the queue again. Defaults to the same as iodepth, meaning
882 that fio will attempt to keep the queue full at all times.
883 If iodepth is set to eg 16 and iodepth_low is set to 4, then
884 after fio has filled the queue of 16 requests, it will let
885 the depth drain down to 4 before starting to fill it again.
887 io_submit_mode=str This option controls how fio submits the IO to
888 the IO engine. The default is 'inline', which means that the
889 fio job threads submit and reap IO directly. If set to
890 'offload', the job threads will offload IO submission to a
891 dedicated pool of IO threads. This requires some coordination
892 and thus has a bit of extra overhead, especially for lower
893 queue depth IO where it can increase latencies. The benefit
894 is that fio can manage submission rates independently of
895 the device completion rates. This avoids skewed latency
896 reporting if IO gets back up on the device side (the
897 coordinated omission problem).
899 direct=bool If value is true, use non-buffered io. This is usually
900 O_DIRECT. Note that ZFS on Solaris doesn't support direct io.
901 On Windows the synchronous ioengines don't support direct io.
903 atomic=bool If value is true, attempt to use atomic direct IO. Atomic
904 writes are guaranteed to be stable once acknowledged by
905 the operating system. Only Linux supports O_ATOMIC right
908 buffered=bool If value is true, use buffered io. This is the opposite
909 of the 'direct' option. Defaults to true.
911 offset=int Start io at the given offset in the file. The data before
912 the given offset will not be touched. This effectively
913 caps the file size at real_size - offset.
915 offset_increment=int If this is provided, then the real offset becomes
916 offset + offset_increment * thread_number, where the thread
917 number is a counter that starts at 0 and is incremented for
918 each sub-job (i.e. when numjobs option is specified). This
919 option is useful if there are several jobs which are intended
920 to operate on a file in parallel disjoint segments, with
921 even spacing between the starting points.
923 number_ios=int Fio will normally perform IOs until it has exhausted the size
924 of the region set by size=, or if it exhaust the allocated
925 time (or hits an error condition). With this setting, the
926 range/size can be set independently of the number of IOs to
927 perform. When fio reaches this number, it will exit normally
928 and report status. Note that this does not extend the amount
929 of IO that will be done, it will only stop fio if this
930 condition is met before other end-of-job criteria.
932 fsync=int If writing to a file, issue a sync of the dirty data
933 for every number of blocks given. For example, if you give
934 32 as a parameter, fio will sync the file for every 32
935 writes issued. If fio is using non-buffered io, we may
936 not sync the file. The exception is the sg io engine, which
937 synchronizes the disk cache anyway.
939 fdatasync=int Like fsync= but uses fdatasync() to only sync data and not
941 In FreeBSD and Windows there is no fdatasync(), this falls back
944 sync_file_range=str:val Use sync_file_range() for every 'val' number of
945 write operations. Fio will track range of writes that
946 have happened since the last sync_file_range() call. 'str'
947 can currently be one or more of:
949 wait_before SYNC_FILE_RANGE_WAIT_BEFORE
950 write SYNC_FILE_RANGE_WRITE
951 wait_after SYNC_FILE_RANGE_WAIT_AFTER
953 So if you do sync_file_range=wait_before,write:8, fio would
954 use SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE for
955 every 8 writes. Also see the sync_file_range(2) man page.
956 This option is Linux specific.
958 overwrite=bool If true, writes to a file will always overwrite existing
959 data. If the file doesn't already exist, it will be
960 created before the write phase begins. If the file exists
961 and is large enough for the specified write phase, nothing
964 end_fsync=bool If true, fsync file contents when a write stage has completed.
966 fsync_on_close=bool If true, fio will fsync() a dirty file on close.
967 This differs from end_fsync in that it will happen on every
968 file close, not just at the end of the job.
970 rwmixread=int How large a percentage of the mix should be reads.
972 rwmixwrite=int How large a percentage of the mix should be writes. If both
973 rwmixread and rwmixwrite is given and the values do not add
974 up to 100%, the latter of the two will be used to override
975 the first. This may interfere with a given rate setting,
976 if fio is asked to limit reads or writes to a certain rate.
977 If that is the case, then the distribution may be skewed.
979 random_distribution=str:float By default, fio will use a completely uniform
980 random distribution when asked to perform random IO. Sometimes
981 it is useful to skew the distribution in specific ways,
982 ensuring that some parts of the data is more hot than others.
983 fio includes the following distribution models:
985 random Uniform random distribution
986 zipf Zipf distribution
987 pareto Pareto distribution
988 gauss Normal (guassian) distribution
989 zoned Zoned random distribution
991 When using a zipf or pareto distribution, an input value
992 is also needed to define the access pattern. For zipf, this
993 is the zipf theta. For pareto, it's the pareto power. Fio
994 includes a test program, genzipf, that can be used visualize
995 what the given input values will yield in terms of hit rates.
996 If you wanted to use zipf with a theta of 1.2, you would use
997 random_distribution=zipf:1.2 as the option. If a non-uniform
998 model is used, fio will disable use of the random map. For
999 the gauss distribution, a normal deviation is supplied as
1000 a value between 0 and 100.
1002 For a zoned distribution, fio supports specifying percentages
1003 of IO access that should fall within what range of the file or
1004 device. For example, given a criteria of:
1006 60% of accesses should be to the first 10%
1007 30% of accesses should be to the next 20%
1008 8% of accesses should be to to the next 30%
1009 2% of accesses should be to the next 40%
1011 we can define that through zoning of the random accesses. For
1012 the above example, the user would do:
1014 random_distribution=zoned:60/10:30/20:8/30:2/40
1016 similarly to how bssplit works for setting ranges and
1017 percentages of block sizes. Like bssplit, it's possible to
1018 specify separate zones for reads, writes, and trims. If just
1019 one set is given, it'll apply to all of them.
1021 percentage_random=int For a random workload, set how big a percentage should
1022 be random. This defaults to 100%, in which case the workload
1023 is fully random. It can be set from anywhere from 0 to 100.
1024 Setting it to 0 would make the workload fully sequential. Any
1025 setting in between will result in a random mix of sequential
1026 and random IO, at the given percentages. It is possible to
1027 set different values for reads, writes, and trim. To do so,
1028 simply use a comma separated list. See blocksize.
1030 norandommap Normally fio will cover every block of the file when doing
1031 random IO. If this option is given, fio will just get a
1032 new random offset without looking at past io history. This
1033 means that some blocks may not be read or written, and that
1034 some blocks may be read/written more than once. If this option
1035 is used with verify= and multiple blocksizes (via bsrange=),
1036 only intact blocks are verified, i.e., partially-overwritten
1039 softrandommap=bool See norandommap. If fio runs with the random block map
1040 enabled and it fails to allocate the map, if this option is
1041 set it will continue without a random block map. As coverage
1042 will not be as complete as with random maps, this option is
1043 disabled by default.
1045 random_generator=str Fio supports the following engines for generating
1046 IO offsets for random IO:
1048 tausworthe Strong 2^88 cycle random number generator
1049 lfsr Linear feedback shift register generator
1050 tausworthe64 Strong 64-bit 2^258 cycle random number
1053 Tausworthe is a strong random number generator, but it
1054 requires tracking on the side if we want to ensure that
1055 blocks are only read or written once. LFSR guarantees
1056 that we never generate the same offset twice, and it's
1057 also less computationally expensive. It's not a true
1058 random generator, however, though for IO purposes it's
1059 typically good enough. LFSR only works with single
1060 block sizes, not with workloads that use multiple block
1061 sizes. If used with such a workload, fio may read or write
1062 some blocks multiple times. The default value is tausworthe,
1063 unless the required space exceeds 2^32 blocks. If it does,
1064 then tausworthe64 is selected automatically.
1066 nice=int Run the job with the given nice value. See man nice(2).
1068 prio=int Set the io priority value of this job. Linux limits us to
1069 a positive value between 0 and 7, with 0 being the highest.
1070 See man ionice(1). Refer to an appropriate manpage for
1071 other operating systems since meaning of priority may differ.
1073 prioclass=int Set the io priority class. See man ionice(1).
1075 thinktime=int Stall the job x microseconds after an io has completed before
1076 issuing the next. May be used to simulate processing being
1077 done by an application. See thinktime_blocks and
1081 Only valid if thinktime is set - pretend to spend CPU time
1082 doing something with the data received, before falling back
1083 to sleeping for the rest of the period specified by
1086 thinktime_blocks=int
1087 Only valid if thinktime is set - control how many blocks
1088 to issue, before waiting 'thinktime' usecs. If not set,
1089 defaults to 1 which will make fio wait 'thinktime' usecs
1090 after every block. This effectively makes any queue depth
1091 setting redundant, since no more than 1 IO will be queued
1092 before we have to complete it and do our thinktime. In
1093 other words, this setting effectively caps the queue depth
1094 if the latter is larger.
1096 rate=int Cap the bandwidth used by this job. The number is in bytes/sec,
1097 the normal suffix rules apply. You can use rate=500k to limit
1098 reads and writes to 500k each, or you can specify read and
1099 writes separately. Using rate=1m,500k would limit reads to
1100 1MB/sec and writes to 500KB/sec. Capping only reads or
1101 writes can be done with rate=,500k or rate=500k,. The former
1102 will only limit writes (to 500KB/sec), the latter will only
1105 rate_min=int Tell fio to do whatever it can to maintain at least this
1106 bandwidth. Failing to meet this requirement, will cause
1107 the job to exit. The same format as rate is used for
1108 read vs write separation.
1110 rate_iops=int Cap the bandwidth to this number of IOPS. Basically the same
1111 as rate, just specified independently of bandwidth. If the
1112 job is given a block size range instead of a fixed value,
1113 the smallest block size is used as the metric. The same format
1114 as rate is used for read vs write separation.
1116 rate_iops_min=int If fio doesn't meet this rate of IO, it will cause
1117 the job to exit. The same format as rate is used for read vs
1120 rate_process=str This option controls how fio manages rated IO
1121 submissions. The default is 'linear', which submits IO in a
1122 linear fashion with fixed delays between IOs that gets
1123 adjusted based on IO completion rates. If this is set to
1124 'poisson', fio will submit IO based on a more real world
1125 random request flow, known as the Poisson process
1126 (https://en.wikipedia.org/wiki/Poisson_process). The lambda
1127 will be 10^6 / IOPS for the given workload.
1129 latency_target=int If set, fio will attempt to find the max performance
1130 point that the given workload will run at while maintaining a
1131 latency below this target. The values is given in microseconds.
1132 See latency_window and latency_percentile
1134 latency_window=int Used with latency_target to specify the sample window
1135 that the job is run at varying queue depths to test the
1136 performance. The value is given in microseconds.
1138 latency_percentile=float The percentage of IOs that must fall within the
1139 criteria specified by latency_target and latency_window. If not
1140 set, this defaults to 100.0, meaning that all IOs must be equal
1141 or below to the value set by latency_target.
1143 max_latency=int If set, fio will exit the job if it exceeds this maximum
1144 latency. It will exit with an ETIME error.
1146 rate_cycle=int Average bandwidth for 'rate' and 'rate_min' over this number
1149 cpumask=int Set the CPU affinity of this job. The parameter given is a
1150 bitmask of allowed CPU's the job may run on. So if you want
1151 the allowed CPUs to be 1 and 5, you would pass the decimal
1152 value of (1 << 1 | 1 << 5), or 34. See man
1153 sched_setaffinity(2). This may not work on all supported
1154 operating systems or kernel versions. This option doesn't
1155 work well for a higher CPU count than what you can store in
1156 an integer mask, so it can only control cpus 1-32. For
1157 boxes with larger CPU counts, use cpus_allowed.
1159 cpus_allowed=str Controls the same options as cpumask, but it allows a text
1160 setting of the permitted CPUs instead. So to use CPUs 1 and
1161 5, you would specify cpus_allowed=1,5. This options also
1162 allows a range of CPUs. Say you wanted a binding to CPUs
1163 1, 5, and 8-15, you would set cpus_allowed=1,5,8-15.
1165 cpus_allowed_policy=str Set the policy of how fio distributes the CPUs
1166 specified by cpus_allowed or cpumask. Two policies are
1169 shared All jobs will share the CPU set specified.
1170 split Each job will get a unique CPU from the CPU set.
1172 'shared' is the default behaviour, if the option isn't
1173 specified. If split is specified, then fio will will assign
1174 one cpu per job. If not enough CPUs are given for the jobs
1175 listed, then fio will roundrobin the CPUs in the set.
1177 numa_cpu_nodes=str Set this job running on spcified NUMA nodes' CPUs. The
1178 arguments allow comma delimited list of cpu numbers,
1179 A-B ranges, or 'all'. Note, to enable numa options support,
1180 fio must be built on a system with libnuma-dev(el) installed.
1182 numa_mem_policy=str Set this job's memory policy and corresponding NUMA
1183 nodes. Format of the argements:
1185 `mode' is one of the following memory policy:
1186 default, prefer, bind, interleave, local
1187 For `default' and `local' memory policy, no node is
1188 needed to be specified.
1189 For `prefer', only one node is allowed.
1190 For `bind' and `interleave', it allow comma delimited
1191 list of numbers, A-B ranges, or 'all'.
1193 startdelay=time Start this job the specified number of seconds after fio
1194 has started. Only useful if the job file contains several
1195 jobs, and you want to delay starting some jobs to a certain
1198 runtime=time Tell fio to terminate processing after the specified number
1199 of seconds. It can be quite hard to determine for how long
1200 a specified job will run, so this parameter is handy to
1201 cap the total runtime to a given time.
1203 time_based If set, fio will run for the duration of the runtime
1204 specified even if the file(s) are completely read or
1205 written. It will simply loop over the same workload
1206 as many times as the runtime allows.
1208 ramp_time=time If set, fio will run the specified workload for this amount
1209 of time before logging any performance numbers. Useful for
1210 letting performance settle before logging results, thus
1211 minimizing the runtime required for stable results. Note
1212 that the ramp_time is considered lead in time for a job,
1213 thus it will increase the total runtime if a special timeout
1214 or runtime is specified.
1216 invalidate=bool Invalidate the buffer/page cache parts for this file prior
1217 to starting io. Defaults to true.
1219 sync=bool Use sync io for buffered writes. For the majority of the
1220 io engines, this means using O_SYNC.
1223 mem=str Fio can use various types of memory as the io unit buffer.
1224 The allowed values are:
1226 malloc Use memory from malloc(3) as the buffers.
1227 Default memory type.
1229 shm Use shared memory as the buffers. Allocated
1232 shmhuge Same as shm, but use huge pages as backing.
1234 mmap Use mmap to allocate buffers. May either be
1235 anonymous memory, or can be file backed if
1236 a filename is given after the option. The
1237 format is mem=mmap:/path/to/file.
1239 mmaphuge Use a memory mapped huge file as the buffer
1240 backing. Append filename after mmaphuge, ala
1241 mem=mmaphuge:/hugetlbfs/file
1243 mmapshared Same as mmap, but use a MMAP_SHARED
1246 The area allocated is a function of the maximum allowed
1247 bs size for the job, multiplied by the io depth given. Note
1248 that for shmhuge and mmaphuge to work, the system must have
1249 free huge pages allocated. This can normally be checked
1250 and set by reading/writing /proc/sys/vm/nr_hugepages on a
1251 Linux system. Fio assumes a huge page is 4MB in size. So
1252 to calculate the number of huge pages you need for a given
1253 job file, add up the io depth of all jobs (normally one unless
1254 iodepth= is used) and multiply by the maximum bs set. Then
1255 divide that number by the huge page size. You can see the
1256 size of the huge pages in /proc/meminfo. If no huge pages
1257 are allocated by having a non-zero number in nr_hugepages,
1258 using mmaphuge or shmhuge will fail. Also see hugepage-size.
1260 mmaphuge also needs to have hugetlbfs mounted and the file
1261 location should point there. So if it's mounted in /huge,
1262 you would use mem=mmaphuge:/huge/somefile.
1264 iomem_align=int This indiciates the memory alignment of the IO memory buffers.
1265 Note that the given alignment is applied to the first IO unit
1266 buffer, if using iodepth the alignment of the following buffers
1267 are given by the bs used. In other words, if using a bs that is
1268 a multiple of the page sized in the system, all buffers will
1269 be aligned to this value. If using a bs that is not page
1270 aligned, the alignment of subsequent IO memory buffers is the
1271 sum of the iomem_align and bs used.
1274 Defines the size of a huge page. Must at least be equal
1275 to the system setting, see /proc/meminfo. Defaults to 4MB.
1276 Should probably always be a multiple of megabytes, so using
1277 hugepage-size=Xm is the preferred way to set this to avoid
1278 setting a non-pow-2 bad value.
1280 exitall When one job finishes, terminate the rest. The default is
1281 to wait for each job to finish, sometimes that is not the
1284 exitall_on_error When one job finishes in error, terminate the rest. The
1285 default is to wait for each job to finish.
1287 bwavgtime=int Average the calculated bandwidth over the given time. Value
1288 is specified in milliseconds. If the job also does bandwidth
1289 logging through 'write_bw_log', then the minimum of this option
1290 and 'log_avg_msec' will be used. Default: 500ms.
1292 iopsavgtime=int Average the calculated IOPS over the given time. Value
1293 is specified in milliseconds. If the job also does IOPS logging
1294 through 'write_iops_log', then the minimum of this option and
1295 'log_avg_msec' will be used. Default: 500ms.
1297 create_serialize=bool If true, serialize the file creating for the jobs.
1298 This may be handy to avoid interleaving of data
1299 files, which may greatly depend on the filesystem
1300 used and even the number of processors in the system.
1302 create_fsync=bool fsync the data file after creation. This is the
1305 create_on_open=bool Don't pre-setup the files for IO, just create open()
1306 when it's time to do IO to that file.
1308 create_only=bool If true, fio will only run the setup phase of the job.
1309 If files need to be laid out or updated on disk, only
1310 that will be done. The actual job contents are not
1313 allow_file_create=bool If true, fio is permitted to create files as part
1314 of its workload. This is the default behavior. If this
1315 option is false, then fio will error out if the files it
1316 needs to use don't already exist. Default: true.
1318 allow_mounted_write=bool If this isn't set, fio will abort jobs that
1319 are destructive (eg that write) to what appears to be a
1320 mounted device or partition. This should help catch creating
1321 inadvertently destructive tests, not realizing that the test
1322 will destroy data on the mounted file system. Default: false.
1324 pre_read=bool If this is given, files will be pre-read into memory before
1325 starting the given IO operation. This will also clear
1326 the 'invalidate' flag, since it is pointless to pre-read
1327 and then drop the cache. This will only work for IO engines
1328 that are seekable, since they allow you to read the same data
1329 multiple times. Thus it will not work on eg network or splice
1332 unlink=bool Unlink the job files when done. Not the default, as repeated
1333 runs of that job would then waste time recreating the file
1334 set again and again.
1336 loops=int Run the specified number of iterations of this job. Used
1337 to repeat the same workload a given number of times. Defaults
1340 verify_only Do not perform specified workload---only verify data still
1341 matches previous invocation of this workload. This option
1342 allows one to check data multiple times at a later date
1343 without overwriting it. This option makes sense only for
1344 workloads that write data, and does not support workloads
1345 with the time_based option set.
1347 do_verify=bool Run the verify phase after a write phase. Only makes sense if
1348 verify is set. Defaults to 1.
1350 verify=str If writing to a file, fio can verify the file contents
1351 after each iteration of the job. Each verification method also implies
1352 verification of special header, which is written to the beginning of
1353 each block. This header also includes meta information, like offset
1354 of the block, block number, timestamp when block was written, etc.
1355 verify=str can be combined with verify_pattern=str option.
1356 The allowed values are:
1358 md5 Use an md5 sum of the data area and store
1359 it in the header of each block.
1361 crc64 Use an experimental crc64 sum of the data
1362 area and store it in the header of each
1365 crc32c Use a crc32c sum of the data area and store
1366 it in the header of each block.
1368 crc32c-intel Use hardware assisted crc32c calcuation
1369 provided on SSE4.2 enabled processors. Falls
1370 back to regular software crc32c, if not
1371 supported by the system.
1373 crc32 Use a crc32 sum of the data area and store
1374 it in the header of each block.
1376 crc16 Use a crc16 sum of the data area and store
1377 it in the header of each block.
1379 crc7 Use a crc7 sum of the data area and store
1380 it in the header of each block.
1382 xxhash Use xxhash as the checksum function. Generally
1383 the fastest software checksum that fio
1386 sha512 Use sha512 as the checksum function.
1388 sha256 Use sha256 as the checksum function.
1390 sha1 Use optimized sha1 as the checksum function.
1392 meta This option is deprecated, since now meta information is
1393 included in generic verification header and meta verification
1394 happens by default. For detailed information see the description
1395 of the verify=str setting. This option is kept because of
1396 compatibility's sake with old configurations. Do not use it.
1398 pattern Verify a strict pattern. Normally fio includes
1399 a header with some basic information and
1400 checksumming, but if this option is set, only
1401 the specific pattern set with 'verify_pattern'
1404 null Only pretend to verify. Useful for testing
1405 internals with ioengine=null, not for much
1408 This option can be used for repeated burn-in tests of a
1409 system to make sure that the written data is also
1410 correctly read back. If the data direction given is
1411 a read or random read, fio will assume that it should
1412 verify a previously written file. If the data direction
1413 includes any form of write, the verify will be of the
1416 verifysort=bool If set, fio will sort written verify blocks when it deems
1417 it faster to read them back in a sorted manner. This is
1418 often the case when overwriting an existing file, since
1419 the blocks are already laid out in the file system. You
1420 can ignore this option unless doing huge amounts of really
1421 fast IO where the red-black tree sorting CPU time becomes
1424 verify_offset=int Swap the verification header with data somewhere else
1425 in the block before writing. Its swapped back before
1428 verify_interval=int Write the verification header at a finer granularity
1429 than the blocksize. It will be written for chunks the
1430 size of header_interval. blocksize should divide this
1433 verify_pattern=str If set, fio will fill the io buffers with this
1434 pattern. Fio defaults to filling with totally random
1435 bytes, but sometimes it's interesting to fill with a known
1436 pattern for io verification purposes. Depending on the
1437 width of the pattern, fio will fill 1/2/3/4 bytes of the
1438 buffer at the time(it can be either a decimal or a hex number).
1439 The verify_pattern if larger than a 32-bit quantity has to
1440 be a hex number that starts with either "0x" or "0X". Use
1441 with verify=str. Also, verify_pattern supports %o format,
1442 which means that for each block offset will be written and
1443 then verifyied back, e.g.:
1447 Or use combination of everything:
1448 verify_pattern=0xff%o"abcd"-12
1450 verify_fatal=bool Normally fio will keep checking the entire contents
1451 before quitting on a block verification failure. If this
1452 option is set, fio will exit the job on the first observed
1455 verify_dump=bool If set, dump the contents of both the original data
1456 block and the data block we read off disk to files. This
1457 allows later analysis to inspect just what kind of data
1458 corruption occurred. Off by default.
1460 verify_async=int Fio will normally verify IO inline from the submitting
1461 thread. This option takes an integer describing how many
1462 async offload threads to create for IO verification instead,
1463 causing fio to offload the duty of verifying IO contents
1464 to one or more separate threads. If using this offload
1465 option, even sync IO engines can benefit from using an
1466 iodepth setting higher than 1, as it allows them to have
1467 IO in flight while verifies are running.
1469 verify_async_cpus=str Tell fio to set the given CPU affinity on the
1470 async IO verification threads. See cpus_allowed for the
1473 verify_backlog=int Fio will normally verify the written contents of a
1474 job that utilizes verify once that job has completed. In
1475 other words, everything is written then everything is read
1476 back and verified. You may want to verify continually
1477 instead for a variety of reasons. Fio stores the meta data
1478 associated with an IO block in memory, so for large
1479 verify workloads, quite a bit of memory would be used up
1480 holding this meta data. If this option is enabled, fio
1481 will write only N blocks before verifying these blocks.
1483 verify_backlog_batch=int Control how many blocks fio will verify
1484 if verify_backlog is set. If not set, will default to
1485 the value of verify_backlog (meaning the entire queue
1486 is read back and verified). If verify_backlog_batch is
1487 less than verify_backlog then not all blocks will be verified,
1488 if verify_backlog_batch is larger than verify_backlog, some
1489 blocks will be verified more than once.
1491 verify_state_save=bool When a job exits during the write phase of a verify
1492 workload, save its current state. This allows fio to replay
1493 up until that point, if the verify state is loaded for the
1494 verify read phase. The format of the filename is, roughly,
1495 <type>-<jobname>-<jobindex>-verify.state. <type> is "local"
1496 for a local run, "sock" for a client/server socket connection,
1497 and "ip" (192.168.0.1, for instance) for a networked
1498 client/server connection.
1500 verify_state_load=bool If a verify termination trigger was used, fio stores
1501 the current write state of each thread. This can be used at
1502 verification time so that fio knows how far it should verify.
1503 Without this information, fio will run a full verification
1504 pass, according to the settings in the job file used.
1507 wait_for_previous Wait for preceding jobs in the job file to exit, before
1508 starting this one. Can be used to insert serialization
1509 points in the job file. A stone wall also implies starting
1510 a new reporting group.
1512 new_group Start a new reporting group. See: group_reporting.
1514 numjobs=int Create the specified number of clones of this job. May be
1515 used to setup a larger number of threads/processes doing
1516 the same thing. Each thread is reported separately; to see
1517 statistics for all clones as a whole, use group_reporting in
1518 conjunction with new_group.
1520 group_reporting It may sometimes be interesting to display statistics for
1521 groups of jobs as a whole instead of for each individual job.
1522 This is especially true if 'numjobs' is used; looking at
1523 individual thread/process output quickly becomes unwieldy.
1524 To see the final report per-group instead of per-job, use
1525 'group_reporting'. Jobs in a file will be part of the same
1526 reporting group, unless if separated by a stonewall, or by
1529 thread fio defaults to forking jobs, however if this option is
1530 given, fio will use pthread_create(3) to create threads
1533 zonesize=int Divide a file into zones of the specified size. See zoneskip.
1535 zoneskip=int Skip the specified number of bytes when zonesize data has
1536 been read. The two zone options can be used to only do
1537 io on zones of a file.
1539 write_iolog=str Write the issued io patterns to the specified file. See
1540 read_iolog. Specify a separate file for each job, otherwise
1541 the iologs will be interspersed and the file may be corrupt.
1543 read_iolog=str Open an iolog with the specified file name and replay the
1544 io patterns it contains. This can be used to store a
1545 workload and replay it sometime later. The iolog given
1546 may also be a blktrace binary file, which allows fio
1547 to replay a workload captured by blktrace. See blktrace
1548 for how to capture such logging data. For blktrace replay,
1549 the file needs to be turned into a blkparse binary data
1550 file first (blkparse <device> -o /dev/null -d file_for_fio.bin).
1552 replay_no_stall=int When replaying I/O with read_iolog the default behavior
1553 is to attempt to respect the time stamps within the log and
1554 replay them with the appropriate delay between IOPS. By
1555 setting this variable fio will not respect the timestamps and
1556 attempt to replay them as fast as possible while still
1557 respecting ordering. The result is the same I/O pattern to a
1558 given device, but different timings.
1560 replay_redirect=str While replaying I/O patterns using read_iolog the
1561 default behavior is to replay the IOPS onto the major/minor
1562 device that each IOP was recorded from. This is sometimes
1563 undesirable because on a different machine those major/minor
1564 numbers can map to a different device. Changing hardware on
1565 the same system can also result in a different major/minor
1566 mapping. Replay_redirect causes all IOPS to be replayed onto
1567 the single specified device regardless of the device it was
1568 recorded from. i.e. replay_redirect=/dev/sdc would cause all
1569 IO in the blktrace to be replayed onto /dev/sdc. This means
1570 multiple devices will be replayed onto a single, if the trace
1571 contains multiple devices. If you want multiple devices to be
1572 replayed concurrently to multiple redirected devices you must
1573 blkparse your trace into separate traces and replay them with
1574 independent fio invocations. Unfortuantely this also breaks
1575 the strict time ordering between multiple device accesses.
1577 replay_align=int Force alignment of IO offsets and lengths in a trace
1578 to this power of 2 value.
1580 replay_scale=int Scale sector offsets down by this factor when
1583 per_job_logs=bool If set, this generates bw/clat/iops log with per
1584 file private filenames. If not set, jobs with identical names
1585 will share the log filename. Default: true.
1587 write_bw_log=str If given, write a bandwidth log of the jobs in this job
1588 file. Can be used to store data of the bandwidth of the
1589 jobs in their lifetime. The included fio_generate_plots
1590 script uses gnuplot to turn these text files into nice
1591 graphs. See write_lat_log for behaviour of given
1592 filename. For this option, the suffix is _bw.x.log, where
1593 x is the index of the job (1..N, where N is the number of
1594 jobs). If 'per_job_logs' is false, then the filename will not
1595 include the job index. See 'Log File Formats'.
1597 write_lat_log=str Same as write_bw_log, except that this option stores io
1598 submission, completion, and total latencies instead. If no
1599 filename is given with this option, the default filename of
1600 "jobname_type.log" is used. Even if the filename is given,
1601 fio will still append the type of log. So if one specifies
1605 The actual log names will be foo_slat.x.log, foo_clat.x.log,
1606 and foo_lat.x.log, where x is the index of the job (1..N,
1607 where N is the number of jobs). This helps fio_generate_plot
1608 fine the logs automatically. If 'per_job_logs' is false, then
1609 the filename will not include the job index. See 'Log File
1612 write_iops_log=str Same as write_bw_log, but writes IOPS. If no filename is
1613 given with this option, the default filename of
1614 "jobname_type.x.log" is used,where x is the index of the job
1615 (1..N, where N is the number of jobs). Even if the filename
1616 is given, fio will still append the type of log. If
1617 'per_job_logs' is false, then the filename will not include
1618 the job index. See 'Log File Formats'.
1620 log_avg_msec=int By default, fio will log an entry in the iops, latency,
1621 or bw log for every IO that completes. When writing to the
1622 disk log, that can quickly grow to a very large size. Setting
1623 this option makes fio average the each log entry over the
1624 specified period of time, reducing the resolution of the log.
1625 See log_max_value as well. Defaults to 0, logging all entries.
1627 log_max_value=bool If log_avg_msec is set, fio logs the average over that
1628 window. If you instead want to log the maximum value, set this
1629 option to 1. Defaults to 0, meaning that averaged values are
1632 log_offset=int If this is set, the iolog options will include the byte
1633 offset for the IO entry as well as the other data values.
1635 log_compression=int If this is set, fio will compress the IO logs as
1636 it goes, to keep the memory footprint lower. When a log
1637 reaches the specified size, that chunk is removed and
1638 compressed in the background. Given that IO logs are
1639 fairly highly compressible, this yields a nice memory
1640 savings for longer runs. The downside is that the
1641 compression will consume some background CPU cycles, so
1642 it may impact the run. This, however, is also true if
1643 the logging ends up consuming most of the system memory.
1644 So pick your poison. The IO logs are saved normally at the
1645 end of a run, by decompressing the chunks and storing them
1646 in the specified log file. This feature depends on the
1647 availability of zlib.
1649 log_compression_cpus=str Define the set of CPUs that are allowed to
1650 handle online log compression for the IO jobs. This can
1651 provide better isolation between performance sensitive jobs,
1652 and background compression work.
1654 log_store_compressed=bool If set, fio will store the log files in a
1655 compressed format. They can be decompressed with fio, using
1656 the --inflate-log command line parameter. The files will be
1657 stored with a .fz suffix.
1659 block_error_percentiles=bool If set, record errors in trim block-sized
1660 units from writes and trims and output a histogram of
1661 how many trims it took to get to errors, and what kind
1662 of error was encountered.
1664 lockmem=int Pin down the specified amount of memory with mlock(2). Can
1665 potentially be used instead of removing memory or booting
1666 with less memory to simulate a smaller amount of memory.
1667 The amount specified is per worker.
1669 exec_prerun=str Before running this job, issue the command specified
1670 through system(3). Output is redirected in a file called
1673 exec_postrun=str After the job completes, issue the command specified
1674 though system(3). Output is redirected in a file called
1675 jobname.postrun.txt.
1677 ioscheduler=str Attempt to switch the device hosting the file to the specified
1678 io scheduler before running.
1680 disk_util=bool Generate disk utilization statistics, if the platform
1681 supports it. Defaults to on.
1683 disable_lat=bool Disable measurements of total latency numbers. Useful
1684 only for cutting back the number of calls to gettimeofday,
1685 as that does impact performance at really high IOPS rates.
1686 Note that to really get rid of a large amount of these
1687 calls, this option must be used with disable_slat and
1690 disable_clat=bool Disable measurements of completion latency numbers. See
1693 disable_slat=bool Disable measurements of submission latency numbers. See
1696 disable_bw=bool Disable measurements of throughput/bandwidth numbers. See
1699 clat_percentiles=bool Enable the reporting of percentiles of
1700 completion latencies.
1702 percentile_list=float_list Overwrite the default list of percentiles
1703 for completion latencies and the block error histogram.
1704 Each number is a floating number in the range (0,100],
1705 and the maximum length of the list is 20. Use ':'
1706 to separate the numbers, and list the numbers in ascending
1707 order. For example, --percentile_list=99.5:99.9 will cause
1708 fio to report the values of completion latency below which
1709 99.5% and 99.9% of the observed latencies fell, respectively.
1711 clocksource=str Use the given clocksource as the base of timing. The
1712 supported options are:
1714 gettimeofday gettimeofday(2)
1716 clock_gettime clock_gettime(2)
1718 cpu Internal CPU clock source
1720 cpu is the preferred clocksource if it is reliable, as it
1721 is very fast (and fio is heavy on time calls). Fio will
1722 automatically use this clocksource if it's supported and
1723 considered reliable on the system it is running on, unless
1724 another clocksource is specifically set. For x86/x86-64 CPUs,
1725 this means supporting TSC Invariant.
1727 gtod_reduce=bool Enable all of the gettimeofday() reducing options
1728 (disable_clat, disable_slat, disable_bw) plus reduce
1729 precision of the timeout somewhat to really shrink
1730 the gettimeofday() call count. With this option enabled,
1731 we only do about 0.4% of the gtod() calls we would have
1732 done if all time keeping was enabled.
1734 gtod_cpu=int Sometimes it's cheaper to dedicate a single thread of
1735 execution to just getting the current time. Fio (and
1736 databases, for instance) are very intensive on gettimeofday()
1737 calls. With this option, you can set one CPU aside for
1738 doing nothing but logging current time to a shared memory
1739 location. Then the other threads/processes that run IO
1740 workloads need only copy that segment, instead of entering
1741 the kernel with a gettimeofday() call. The CPU set aside
1742 for doing these time calls will be excluded from other
1743 uses. Fio will manually clear it from the CPU mask of other
1746 continue_on_error=str Normally fio will exit the job on the first observed
1747 failure. If this option is set, fio will continue the job when
1748 there is a 'non-fatal error' (EIO or EILSEQ) until the runtime
1749 is exceeded or the I/O size specified is completed. If this
1750 option is used, there are two more stats that are appended,
1751 the total error count and the first error. The error field
1752 given in the stats is the first error that was hit during the
1755 The allowed values are:
1757 none Exit on any IO or verify errors.
1759 read Continue on read errors, exit on all others.
1761 write Continue on write errors, exit on all others.
1763 io Continue on any IO error, exit on all others.
1765 verify Continue on verify errors, exit on all others.
1767 all Continue on all errors.
1769 0 Backward-compatible alias for 'none'.
1771 1 Backward-compatible alias for 'all'.
1773 ignore_error=str Sometimes you want to ignore some errors during test
1774 in that case you can specify error list for each error type.
1775 ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST
1776 errors for given error type is separated with ':'. Error
1777 may be symbol ('ENOSPC', 'ENOMEM') or integer.
1779 ignore_error=EAGAIN,ENOSPC:122
1780 This option will ignore EAGAIN from READ, and ENOSPC and
1781 122(EDQUOT) from WRITE.
1783 error_dump=bool If set dump every error even if it is non fatal, true
1784 by default. If disabled only fatal error will be dumped
1786 cgroup=str Add job to this control group. If it doesn't exist, it will
1787 be created. The system must have a mounted cgroup blkio
1788 mount point for this to work. If your system doesn't have it
1789 mounted, you can do so with:
1791 # mount -t cgroup -o blkio none /cgroup
1793 cgroup_weight=int Set the weight of the cgroup to this value. See
1794 the documentation that comes with the kernel, allowed values
1795 are in the range of 100..1000.
1797 cgroup_nodelete=bool Normally fio will delete the cgroups it has created after
1798 the job completion. To override this behavior and to leave
1799 cgroups around after the job completion, set cgroup_nodelete=1.
1800 This can be useful if one wants to inspect various cgroup
1801 files after job completion. Default: false
1803 uid=int Instead of running as the invoking user, set the user ID to
1804 this value before the thread/process does any work.
1806 gid=int Set group ID, see uid.
1808 flow_id=int The ID of the flow. If not specified, it defaults to being a
1809 global flow. See flow.
1811 flow=int Weight in token-based flow control. If this value is used, then
1812 there is a 'flow counter' which is used to regulate the
1813 proportion of activity between two or more jobs. fio attempts
1814 to keep this flow counter near zero. The 'flow' parameter
1815 stands for how much should be added or subtracted to the flow
1816 counter on each iteration of the main I/O loop. That is, if
1817 one job has flow=8 and another job has flow=-1, then there
1818 will be a roughly 1:8 ratio in how much one runs vs the other.
1820 flow_watermark=int The maximum value that the absolute value of the flow
1821 counter is allowed to reach before the job must wait for a
1822 lower value of the counter.
1824 flow_sleep=int The period of time, in microseconds, to wait after the flow
1825 watermark has been exceeded before retrying operations
1827 In addition, there are some parameters which are only valid when a specific
1828 ioengine is in use. These are used identically to normal parameters, with the
1829 caveat that when used on the command line, they must come after the ioengine
1830 that defines them is selected.
1832 [libaio] userspace_reap Normally, with the libaio engine in use, fio will use
1833 the io_getevents system call to reap newly returned events.
1834 With this flag turned on, the AIO ring will be read directly
1835 from user-space to reap events. The reaping mode is only
1836 enabled when polling for a minimum of 0 events (eg when
1837 iodepth_batch_complete=0).
1839 [psyncv2] hipri Set RWF_HIPRI on IO, indicating to the kernel that
1840 it's of higher priority than normal.
1842 [cpu] cpuload=int Attempt to use the specified percentage of CPU cycles.
1844 [cpu] cpuchunks=int Split the load into cycles of the given time. In
1847 [cpu] exit_on_io_done=bool Detect when IO threads are done, then exit.
1849 [netsplice] hostname=str
1850 [net] hostname=str The host name or IP address to use for TCP or UDP based IO.
1851 If the job is a TCP listener or UDP reader, the hostname is not
1852 used and must be omitted unless it is a valid UDP multicast
1854 [libhdfs] namenode=str The host name or IP address of a HDFS cluster namenode to contact.
1856 [netsplice] port=int
1857 [net] port=int The TCP or UDP port to bind to or connect to. If this is used
1858 with numjobs to spawn multiple instances of the same job type, then this will
1859 be the starting port number since fio will use a range of ports.
1860 [libhdfs] port=int the listening port of the HFDS cluster namenode.
1862 [netsplice] interface=str
1863 [net] interface=str The IP address of the network interface used to send or
1864 receive UDP multicast
1867 [net] ttl=int Time-to-live value for outgoing UDP multicast packets.
1870 [netsplice] nodelay=bool
1871 [net] nodelay=bool Set TCP_NODELAY on TCP connections.
1873 [netsplice] protocol=str
1874 [netsplice] proto=str
1876 [net] proto=str The network protocol to use. Accepted values are:
1878 tcp Transmission control protocol
1879 tcpv6 Transmission control protocol V6
1880 udp User datagram protocol
1881 udpv6 User datagram protocol V6
1882 unix UNIX domain socket
1884 When the protocol is TCP or UDP, the port must also be given,
1885 as well as the hostname if the job is a TCP listener or UDP
1886 reader. For unix sockets, the normal filename option should be
1887 used and the port is invalid.
1889 [net] listen For TCP network connections, tell fio to listen for incoming
1890 connections rather than initiating an outgoing connection. The
1891 hostname must be omitted if this option is used.
1893 [net] pingpong Normaly a network writer will just continue writing data, and
1894 a network reader will just consume packages. If pingpong=1
1895 is set, a writer will send its normal payload to the reader,
1896 then wait for the reader to send the same payload back. This
1897 allows fio to measure network latencies. The submission
1898 and completion latencies then measure local time spent
1899 sending or receiving, and the completion latency measures
1900 how long it took for the other end to receive and send back.
1901 For UDP multicast traffic pingpong=1 should only be set for a
1902 single reader when multiple readers are listening to the same
1905 [net] window_size Set the desired socket buffer size for the connection.
1907 [net] mss Set the TCP maximum segment size (TCP_MAXSEG).
1909 [e4defrag] donorname=str
1910 File will be used as a block donor(swap extents between files)
1911 [e4defrag] inplace=int
1912 Configure donor file blocks allocation strategy
1913 0(default): Preallocate donor's file on init
1914 1 : allocate space immidietly inside defragment event,
1915 and free right after event
1917 [rbd] clustername=str Specifies the name of the Ceph cluster.
1918 [rbd] rbdname=str Specifies the name of the RBD.
1919 [rbd] pool=str Specifies the naem of the Ceph pool containing RBD.
1920 [rbd] clientname=str Specifies the username (without the 'client.' prefix)
1921 used to access the Ceph cluster. If the clustername is
1922 specified, the clientmae shall be the full type.id
1923 string. If no type. prefix is given, fio will add
1924 'client.' by default.
1926 [mtd] skip_bad=bool Skip operations against known bad blocks.
1928 [libhdfs] hdfsdirectory libhdfs will create chunk in this HDFS directory
1929 [libhdfs] chunck_size the size of the chunck to use for each file.
1932 6.0 Interpreting the output
1933 ---------------------------
1935 fio spits out a lot of output. While running, fio will display the
1936 status of the jobs created. An example of that would be:
1938 Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
1940 The characters inside the square brackets denote the current status of
1941 each thread. The possible values (in typical life cycle order) are:
1945 P Thread setup, but not started.
1947 I Thread initialized, waiting or generating necessary data.
1948 p Thread running pre-reading file(s).
1949 R Running, doing sequential reads.
1950 r Running, doing random reads.
1951 W Running, doing sequential writes.
1952 w Running, doing random writes.
1953 M Running, doing mixed sequential reads/writes.
1954 m Running, doing mixed random reads/writes.
1955 F Running, currently waiting for fsync()
1956 f Running, finishing up (writing IO logs, etc)
1957 V Running, doing verification of written data.
1958 E Thread exited, not reaped by main thread yet.
1960 X Thread reaped, exited with an error.
1961 K Thread reaped, exited due to signal.
1963 Fio will condense the thread string as not to take up more space on the
1964 command line as is needed. For instance, if you have 10 readers and 10
1965 writers running, the output would look like this:
1967 Jobs: 20 (f=20): [R(10),W(10)] [4.0% done] [2103MB/0KB/0KB /s] [538K/0/0 iops] [eta 57m:36s]
1969 Fio will still maintain the ordering, though. So the above means that jobs
1970 1..10 are readers, and 11..20 are writers.
1972 The other values are fairly self explanatory - number of threads
1973 currently running and doing io, rate of io since last check (read speed
1974 listed first, then write speed), and the estimated completion percentage
1975 and time for the running group. It's impossible to estimate runtime of
1976 the following groups (if any). Note that the string is displayed in order,
1977 so it's possible to tell which of the jobs are currently doing what. The
1978 first character is the first job defined in the job file, and so forth.
1980 When fio is done (or interrupted by ctrl-c), it will show the data for
1981 each thread, group of threads, and disks in that order. For each data
1982 direction, the output looks like:
1984 Client1 (g=0): err= 0:
1985 write: io= 32MB, bw= 666KB/s, iops=89 , runt= 50320msec
1986 slat (msec): min= 0, max= 136, avg= 0.03, stdev= 1.92
1987 clat (msec): min= 0, max= 631, avg=48.50, stdev=86.82
1988 bw (KB/s) : min= 0, max= 1196, per=51.00%, avg=664.02, stdev=681.68
1989 cpu : usr=1.49%, sys=0.25%, ctx=7969, majf=0, minf=17
1990 IO depths : 1=0.1%, 2=0.3%, 4=0.5%, 8=99.0%, 16=0.0%, 32=0.0%, >32=0.0%
1991 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
1992 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
1993 issued r/w: total=0/32768, short=0/0
1994 lat (msec): 2=1.6%, 4=0.0%, 10=3.2%, 20=12.8%, 50=38.4%, 100=24.8%,
1995 lat (msec): 250=15.2%, 500=0.0%, 750=0.0%, 1000=0.0%, >=2048=0.0%
1997 The client number is printed, along with the group id and error of that
1998 thread. Below is the io statistics, here for writes. In the order listed,
2001 io= Number of megabytes io performed
2002 bw= Average bandwidth rate
2003 iops= Average IOs performed per second
2004 runt= The runtime of that thread
2005 slat= Submission latency (avg being the average, stdev being the
2006 standard deviation). This is the time it took to submit
2007 the io. For sync io, the slat is really the completion
2008 latency, since queue/complete is one operation there. This
2009 value can be in milliseconds or microseconds, fio will choose
2010 the most appropriate base and print that. In the example
2011 above, milliseconds is the best scale. Note: in --minimal mode
2012 latencies are always expressed in microseconds.
2013 clat= Completion latency. Same names as slat, this denotes the
2014 time from submission to completion of the io pieces. For
2015 sync io, clat will usually be equal (or very close) to 0,
2016 as the time from submit to complete is basically just
2017 CPU time (io has already been done, see slat explanation).
2018 bw= Bandwidth. Same names as the xlat stats, but also includes
2019 an approximate percentage of total aggregate bandwidth
2020 this thread received in this group. This last value is
2021 only really useful if the threads in this group are on the
2022 same disk, since they are then competing for disk access.
2023 cpu= CPU usage. User and system time, along with the number
2024 of context switches this thread went through, usage of
2025 system and user time, and finally the number of major
2026 and minor page faults. The CPU utilization numbers are
2027 averages for the jobs in that reporting group, while the
2028 context and fault counters are summed.
2029 IO depths= The distribution of io depths over the job life time. The
2030 numbers are divided into powers of 2, so for example the
2031 16= entries includes depths up to that value but higher
2032 than the previous entry. In other words, it covers the
2033 range from 16 to 31.
2034 IO submit= How many pieces of IO were submitting in a single submit
2035 call. Each entry denotes that amount and below, until
2036 the previous entry - eg, 8=100% mean that we submitted
2037 anywhere in between 5-8 ios per submit call.
2038 IO complete= Like the above submit number, but for completions instead.
2039 IO issued= The number of read/write requests issued, and how many
2041 IO latencies= The distribution of IO completion latencies. This is the
2042 time from when IO leaves fio and when it gets completed.
2043 The numbers follow the same pattern as the IO depths,
2044 meaning that 2=1.6% means that 1.6% of the IO completed
2045 within 2 msecs, 20=12.8% means that 12.8% of the IO
2046 took more than 10 msecs, but less than (or equal to) 20 msecs.
2048 After each client has been listed, the group statistics are printed. They
2049 will look like this:
2051 Run status group 0 (all jobs):
2052 READ: io=64MB, aggrb=22178, minb=11355, maxb=11814, mint=2840msec, maxt=2955msec
2053 WRITE: io=64MB, aggrb=1302, minb=666, maxb=669, mint=50093msec, maxt=50320msec
2055 For each data direction, it prints:
2057 io= Number of megabytes io performed.
2058 aggrb= Aggregate bandwidth of threads in this group.
2059 minb= The minimum average bandwidth a thread saw.
2060 maxb= The maximum average bandwidth a thread saw.
2061 mint= The smallest runtime of the threads in that group.
2062 maxt= The longest runtime of the threads in that group.
2064 And finally, the disk statistics are printed. They will look like this:
2066 Disk stats (read/write):
2067 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
2069 Each value is printed for both reads and writes, with reads first. The
2072 ios= Number of ios performed by all groups.
2073 merge= Number of merges io the io scheduler.
2074 ticks= Number of ticks we kept the disk busy.
2075 io_queue= Total time spent in the disk queue.
2076 util= The disk utilization. A value of 100% means we kept the disk
2077 busy constantly, 50% would be a disk idling half of the time.
2079 It is also possible to get fio to dump the current output while it is
2080 running, without terminating the job. To do that, send fio the USR1 signal.
2081 You can also get regularly timed dumps by using the --status-interval
2082 parameter, or by creating a file in /tmp named fio-dump-status. If fio
2083 sees this file, it will unlink it and dump the current output status.
2089 For scripted usage where you typically want to generate tables or graphs
2090 of the results, fio can output the results in a semicolon separated format.
2091 The format is one long line of values, such as:
2093 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%
2094 A description of this job goes here.
2096 The job description (if provided) follows on a second line.
2098 To enable terse output, use the --minimal command line option. The first
2099 value is the version of the terse output format. If the output has to
2100 be changed for some reason, this number will be incremented by 1 to
2101 signify that change.
2103 Split up, the format is as follows:
2105 terse version, fio version, jobname, groupid, error
2107 Total IO (KB), bandwidth (KB/sec), IOPS, runtime (msec)
2108 Submission latency: min, max, mean, stdev (usec)
2109 Completion latency: min, max, mean, stdev (usec)
2110 Completion latency percentiles: 20 fields (see below)
2111 Total latency: min, max, mean, stdev (usec)
2112 Bw (KB/s): min, max, aggregate percentage of total, mean, stdev
2114 Total IO (KB), bandwidth (KB/sec), IOPS, runtime (msec)
2115 Submission latency: min, max, mean, stdev (usec)
2116 Completion latency: min, max, mean, stdev(usec)
2117 Completion latency percentiles: 20 fields (see below)
2118 Total latency: min, max, mean, stdev (usec)
2119 Bw (KB/s): min, max, aggregate percentage of total, mean, stdev
2120 CPU usage: user, system, context switches, major faults, minor faults
2121 IO depths: <=1, 2, 4, 8, 16, 32, >=64
2122 IO latencies microseconds: <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
2123 IO latencies milliseconds: <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
2124 Disk utilization: Disk name, Read ios, write ios,
2125 Read merges, write merges,
2126 Read ticks, write ticks,
2127 Time spent in queue, disk utilization percentage
2128 Additional Info (dependent on continue_on_error, default off): total # errors, first error code
2130 Additional Info (dependent on description being set): Text description
2132 Completion latency percentiles can be a grouping of up to 20 sets, so
2133 for the terse output fio writes all of them. Each field will look like this:
2137 which is the Xth percentile, and the usec latency associated with it.
2139 For disk utilization, all disks used by fio are shown. So for each disk
2140 there will be a disk utilization section.
2143 8.0 Trace file format
2144 ---------------------
2145 There are two trace file format that you can encounter. The older (v1) format
2146 is unsupported since version 1.20-rc3 (March 2008). It will still be described
2147 below in case that you get an old trace and want to understand it.
2149 In any case the trace is a simple text file with a single action per line.
2152 8.1 Trace file format v1
2153 ------------------------
2154 Each line represents a single io action in the following format:
2158 where rw=0/1 for read/write, and the offset and length entries being in bytes.
2160 This format is not supported in Fio versions => 1.20-rc3.
2163 8.2 Trace file format v2
2164 ------------------------
2165 The second version of the trace file format was added in Fio version 1.17.
2166 It allows to access more then one file per trace and has a bigger set of
2167 possible file actions.
2169 The first line of the trace file has to be:
2173 Following this can be lines in two different formats, which are described below.
2175 The file management format:
2179 The filename is given as an absolute path. The action can be one of these:
2181 add Add the given filename to the trace
2182 open Open the file with the given filename. The filename has to have
2183 been added with the add action before.
2184 close Close the file with the given filename. The file has to have been
2188 The file io action format:
2190 filename action offset length
2192 The filename is given as an absolute path, and has to have been added and opened
2193 before it can be used with this format. The offset and length are given in
2194 bytes. The action can be one of these:
2196 wait Wait for 'offset' microseconds. Everything below 100 is discarded.
2197 The time is relative to the previous wait statement.
2198 read Read 'length' bytes beginning from 'offset'
2199 write Write 'length' bytes beginning from 'offset'
2200 sync fsync() the file
2201 datasync fdatasync() the file
2202 trim trim the given file from the given 'offset' for 'length' bytes
2205 9.0 CPU idleness profiling
2206 --------------------------
2207 In some cases, we want to understand CPU overhead in a test. For example,
2208 we test patches for the specific goodness of whether they reduce CPU usage.
2209 fio implements a balloon approach to create a thread per CPU that runs at
2210 idle priority, meaning that it only runs when nobody else needs the cpu.
2211 By measuring the amount of work completed by the thread, idleness of each
2212 CPU can be derived accordingly.
2214 An unit work is defined as touching a full page of unsigned characters. Mean
2215 and standard deviation of time to complete an unit work is reported in "unit
2216 work" section. Options can be chosen to report detailed percpu idleness or
2217 overall system idleness by aggregating percpu stats.
2220 10.0 Verification and triggers
2221 ------------------------------
2222 Fio is usually run in one of two ways, when data verification is done. The
2223 first is a normal write job of some sort with verify enabled. When the
2224 write phase has completed, fio switches to reads and verifies everything
2225 it wrote. The second model is running just the write phase, and then later
2226 on running the same job (but with reads instead of writes) to repeat the
2227 same IO patterns and verify the contents. Both of these methods depend
2228 on the write phase being completed, as fio otherwise has no idea how much
2231 With verification triggers, fio supports dumping the current write state
2232 to local files. Then a subsequent read verify workload can load this state
2233 and know exactly where to stop. This is useful for testing cases where
2234 power is cut to a server in a managed fashion, for instance.
2236 A verification trigger consists of two things:
2238 1) Storing the write state of each job
2239 2) Executing a trigger command
2241 The write state is relatively small, on the order of hundreds of bytes
2242 to single kilobytes. It contains information on the number of completions
2243 done, the last X completions, etc.
2245 A trigger is invoked either through creation ('touch') of a specified
2246 file in the system, or through a timeout setting. If fio is run with
2247 --trigger-file=/tmp/trigger-file, then it will continually check for
2248 the existence of /tmp/trigger-file. When it sees this file, it will
2249 fire off the trigger (thus saving state, and executing the trigger
2252 For client/server runs, there's both a local and remote trigger. If
2253 fio is running as a server backend, it will send the job states back
2254 to the client for safe storage, then execute the remote trigger, if
2255 specified. If a local trigger is specified, the server will still send
2256 back the write state, but the client will then execute the trigger.
2258 10.1 Verification trigger example
2259 ---------------------------------
2260 Lets say we want to run a powercut test on the remote machine 'server'.
2261 Our write workload is in write-test.fio. We want to cut power to 'server'
2262 at some point during the run, and we'll run this test from the safety
2263 or our local machine, 'localbox'. On the server, we'll start the fio
2266 server# fio --server
2268 and on the client, we'll fire off the workload:
2270 localbox$ fio --client=server --trigger-file=/tmp/my-trigger --trigger-remote="bash -c \"echo b > /proc/sysrq-triger\""
2272 We set /tmp/my-trigger as the trigger file, and we tell fio to execute
2274 echo b > /proc/sysrq-trigger
2276 on the server once it has received the trigger and sent us the write
2277 state. This will work, but it's not _really_ cutting power to the server,
2278 it's merely abruptly rebooting it. If we have a remote way of cutting
2279 power to the server through IPMI or similar, we could do that through
2280 a local trigger command instead. Lets assume we have a script that does
2281 IPMI reboot of a given hostname, ipmi-reboot. On localbox, we could
2282 then have run fio with a local trigger instead:
2284 localbox$ fio --client=server --trigger-file=/tmp/my-trigger --trigger="ipmi-reboot server"
2286 For this case, fio would wait for the server to send us the write state,
2287 then execute 'ipmi-reboot server' when that happened.
2289 10.2 Loading verify state
2290 -------------------------
2291 To load store write state, read verification job file must contain
2292 the verify_state_load option. If that is set, fio will load the previously
2293 stored state. For a local fio run this is done by loading the files directly,
2294 and on a client/server run, the server backend will ask the client to send
2295 the files over and load them from there.
2298 11.0 Log File Formats
2299 ---------------------
2301 Fio supports a variety of log file formats, for logging latencies, bandwidth,
2302 and IOPS. The logs share a common format, which looks like this:
2304 time (msec), value, data direction, offset
2306 Time for the log entry is always in milliseconds. The value logged depends
2307 on the type of log, it will be one of the following:
2309 Latency log Value is latency in usecs
2310 Bandwidth log Value is in KB/sec
2311 IOPS log Value is IOPS
2313 Data direction is one of the following:
2319 The offset is the offset, in bytes, from the start of the file, for that
2320 particular IO. The logging of the offset can be toggled with 'log_offset'.
2322 If windowed logging is enabled though 'log_avg_msec', then fio doesn't log
2323 individual IOs. Instead of logs the average values over the specified
2324 period of time. Since 'data direction' and 'offset' are per-IO values,
2325 they aren't applicable if windowed logging is enabled. If windowed logging
2326 is enabled and 'log_max_value' is set, then fio logs maximum values in
2327 that window instead of averages.