8 5. Detailed list of parameters
12 9. CPU idleness profiling
14 1.0 Overview and history
15 ------------------------
16 fio was originally written to save me the hassle of writing special test
17 case programs when I wanted to test a specific workload, either for
18 performance reasons or to find/reproduce a bug. The process of writing
19 such a test app can be tiresome, especially if you have to do it often.
20 Hence I needed a tool that would be able to simulate a given io workload
21 without resorting to writing a tailored test case again and again.
23 A test work load is difficult to define, though. There can be any number
24 of processes or threads involved, and they can each be using their own
25 way of generating io. You could have someone dirtying large amounts of
26 memory in an memory mapped file, or maybe several threads issuing
27 reads using asynchronous io. fio needed to be flexible enough to
28 simulate both of these cases, and many more.
32 The first step in getting fio to simulate a desired io workload, is
33 writing a job file describing that specific setup. A job file may contain
34 any number of threads and/or files - the typical contents of the job file
35 is a global section defining shared parameters, and one or more job
36 sections describing the jobs involved. When run, fio parses this file
37 and sets everything up as described. If we break down a job from top to
38 bottom, it contains the following basic parameters:
40 IO type Defines the io pattern issued to the file(s).
41 We may only be reading sequentially from this
42 file(s), or we may be writing randomly. Or even
43 mixing reads and writes, sequentially or randomly.
45 Block size In how large chunks are we issuing io? This may be
46 a single value, or it may describe a range of
49 IO size How much data are we going to be reading/writing.
51 IO engine How do we issue io? We could be memory mapping the
52 file, we could be using regular read/write, we
53 could be using splice, async io, syslet, or even
56 IO depth If the io engine is async, how large a queuing
57 depth do we want to maintain?
59 IO type Should we be doing buffered io, or direct/raw io?
61 Num files How many files are we spreading the workload over.
63 Num threads How many threads or processes should we spread
66 The above are the basic parameters defined for a workload, in addition
67 there's a multitude of parameters that modify other aspects of how this
73 See the README file for command line parameters, there are only a few
76 Running fio is normally the easiest part - you just give it the job file
77 (or job files) as parameters:
81 and it will start doing what the job_file tells it to do. You can give
82 more than one job file on the command line, fio will serialize the running
83 of those files. Internally that is the same as using the 'stonewall'
84 parameter described the the parameter section.
86 If the job file contains only one job, you may as well just give the
87 parameters on the command line. The command line parameters are identical
88 to the job parameters, with a few extra that control global parameters
89 (see README). For example, for the job file parameter iodepth=2, the
90 mirror command line option would be --iodepth 2 or --iodepth=2. You can
91 also use the command line for giving more than one job entry. For each
92 --name option that fio sees, it will start a new job with that name.
93 Command line entries following a --name entry will apply to that job,
94 until there are no more entries or a new --name entry is seen. This is
95 similar to the job file options, where each option applies to the current
96 job until a new [] job entry is seen.
98 fio does not need to run as root, except if the files or devices specified
99 in the job section requires that. Some other options may also be restricted,
100 such as memory locking, io scheduler switching, and decreasing the nice value.
105 As previously described, fio accepts one or more job files describing
106 what it is supposed to do. The job file format is the classic ini file,
107 where the names enclosed in [] brackets define the job name. You are free
108 to use any ascii name you want, except 'global' which has special meaning.
109 A global section sets defaults for the jobs described in that file. A job
110 may override a global section parameter, and a job file may even have
111 several global sections if so desired. A job is only affected by a global
112 section residing above it. If the first character in a line is a ';' or a
113 '#', the entire line is discarded as a comment.
115 So let's look at a really simple job file that defines two processes, each
116 randomly reading from a 128MB file.
118 ; -- start job file --
129 As you can see, the job file sections themselves are empty as all the
130 described parameters are shared. As no filename= option is given, fio
131 makes up a filename for each of the jobs as it sees fit. On the command
132 line, this job would look as follows:
134 $ fio --name=global --rw=randread --size=128m --name=job1 --name=job2
137 Let's look at an example that has a number of processes writing randomly
140 ; -- start job file --
152 Here we have no global section, as we only have one job defined anyway.
153 We want to use async io here, with a depth of 4 for each file. We also
154 increased the buffer size used to 32KB and define numjobs to 4 to
155 fork 4 identical jobs. The result is 4 processes each randomly writing
156 to their own 64MB file. Instead of using the above job file, you could
157 have given the parameters on the command line. For this case, you would
160 $ fio --name=random-writers --ioengine=libaio --iodepth=4 --rw=randwrite --bs=32k --direct=0 --size=64m --numjobs=4
162 4.1 Environment variables
163 -------------------------
165 fio also supports environment variable expansion in job files. Any
166 substring of the form "${VARNAME}" as part of an option value (in other
167 words, on the right of the `='), will be expanded to the value of the
168 environment variable called VARNAME. If no such environment variable
169 is defined, or VARNAME is the empty string, the empty string will be
172 As an example, let's look at a sample fio invocation and job file:
174 $ SIZE=64m NUMJOBS=4 fio jobfile.fio
176 ; -- start job file --
183 This will expand to the following equivalent job file at runtime:
185 ; -- start job file --
192 fio ships with a few example job files, you can also look there for
195 4.2 Reserved keywords
196 ---------------------
198 Additionally, fio has a set of reserved keywords that will be replaced
199 internally with the appropriate value. Those keywords are:
201 $pagesize The architecture page size of the running system
202 $mb_memory Megabytes of total memory in the system
203 $ncpus Number of online available CPUs
205 These can be used on the command line or in the job file, and will be
206 automatically substituted with the current system values when the job
207 is run. Simple math is also supported on these keywords, so you can
208 perform actions like:
212 and get that properly expanded to 8 times the size of memory in the
216 5.0 Detailed list of parameters
217 -------------------------------
219 This section describes in details each parameter associated with a job.
220 Some parameters take an option of a given type, such as an integer or
221 a string. The following types are used:
223 str String. This is a sequence of alpha characters.
224 time Integer with possible time suffix. In seconds unless otherwise
225 specified, use eg 10m for 10 minutes. Accepts s/m/h for seconds,
227 int SI integer. A whole number value, which may contain a suffix
228 describing the base of the number. Accepted suffixes are k/m/g/t/p,
229 meaning kilo, mega, giga, tera, and peta. The suffix is not case
230 sensitive, and you may also include trailing 'b' (eg 'kb' is the same
231 as 'k'). So if you want to specify 4096, you could either write
232 out '4096' or just give 4k. The suffixes signify base 2 values, so
233 1024 is 1k and 1024k is 1m and so on, unless the suffix is explicitly
234 set to a base 10 value using 'kib', 'mib', 'gib', etc. If that is the
235 case, then 1000 is used as the multiplier. This can be handy for
236 disks, since manufacturers generally use base 10 values when listing
237 the capacity of a drive. If the option accepts an upper and lower
238 range, use a colon ':' or minus '-' to separate such values. May also
239 include a prefix to indicate numbers base. If 0x is used, the number
240 is assumed to be hexadecimal. See irange.
241 bool Boolean. Usually parsed as an integer, however only defined for
242 true and false (1 and 0).
243 irange Integer range with suffix. Allows value range to be given, such
244 as 1024-4096. A colon may also be used as the separator, eg
245 1k:4k. If the option allows two sets of ranges, they can be
246 specified with a ',' or '/' delimiter: 1k-4k/8k-32k. Also see
248 float_list A list of floating numbers, separated by a ':' character.
250 With the above in mind, here follows the complete list of fio job
253 name=str ASCII name of the job. This may be used to override the
254 name printed by fio for this job. Otherwise the job
255 name is used. On the command line this parameter has the
256 special purpose of also signaling the start of a new
259 description=str Text description of the job. Doesn't do anything except
260 dump this text description when this job is run. It's
263 directory=str Prefix filenames with this directory. Used to place files
264 in a different location than "./".
266 filename=str Fio normally makes up a filename based on the job name,
267 thread number, and file number. If you want to share
268 files between threads in a job or several jobs, specify
269 a filename for each of them to override the default. If
270 the ioengine used is 'net', the filename is the host, port,
271 and protocol to use in the format of =host,port,protocol.
272 See ioengine=net for more. If the ioengine is file based, you
273 can specify a number of files by separating the names with a
274 ':' colon. So if you wanted a job to open /dev/sda and /dev/sdb
275 as the two working files, you would use
276 filename=/dev/sda:/dev/sdb. On Windows, disk devices are
277 accessed as \\.\PhysicalDrive0 for the first device,
278 \\.\PhysicalDrive1 for the second etc. Note: Windows and
279 FreeBSD prevent write access to areas of the disk containing
280 in-use data (e.g. filesystems).
281 If the wanted filename does need to include a colon, then
282 escape that with a '\' character. For instance, if the filename
283 is "/dev/dsk/foo@3,0:c", then you would use
284 filename="/dev/dsk/foo@3,0\:c". '-' is a reserved name, meaning
285 stdin or stdout. Which of the two depends on the read/write
289 If sharing multiple files between jobs, it is usually necessary
290 to have fio generate the exact names that you want. By default,
291 fio will name a file based on the default file format
292 specification of jobname.jobnumber.filenumber. With this
293 option, that can be customized. Fio will recognize and replace
294 the following keywords in this string:
297 The name of the worker thread or process.
300 The incremental number of the worker thread or
304 The incremental number of the file for that worker
307 To have dependent jobs share a set of files, this option can
308 be set to have fio generate filenames that are shared between
309 the two. For instance, if testfiles.$filenum is specified,
310 file number 4 for any job will be named testfiles.4. The
311 default of $jobname.$jobnum.$filenum will be used if
312 no other format specifier is given.
314 opendir=str Tell fio to recursively add any file it can find in this
315 directory and down the file system tree.
317 lockfile=str Fio defaults to not locking any files before it does
318 IO to them. If a file or file descriptor is shared, fio
319 can serialize IO to that file to make the end result
320 consistent. This is usual for emulating real workloads that
321 share files. The lock modes are:
323 none No locking. The default.
324 exclusive Only one thread/process may do IO,
325 excluding all others.
326 readwrite Read-write locking on the file. Many
327 readers may access the file at the
328 same time, but writes get exclusive
332 rw=str Type of io pattern. Accepted values are:
334 read Sequential reads
335 write Sequential writes
336 randwrite Random writes
337 randread Random reads
338 rw,readwrite Sequential mixed reads and writes
339 randrw Random mixed reads and writes
341 For the mixed io types, the default is to split them 50/50.
342 For certain types of io the result may still be skewed a bit,
343 since the speed may be different. It is possible to specify
344 a number of IO's to do before getting a new offset, this is
345 one by appending a ':<nr>' to the end of the string given.
346 For a random read, it would look like 'rw=randread:8' for
347 passing in an offset modifier with a value of 8. If the
348 suffix is used with a sequential IO pattern, then the value
349 specified will be added to the generated offset for each IO.
350 For instance, using rw=write:4k will skip 4k for every
351 write. It turns sequential IO into sequential IO with holes.
352 See the 'rw_sequencer' option.
354 rw_sequencer=str If an offset modifier is given by appending a number to
355 the rw=<str> line, then this option controls how that
356 number modifies the IO offset being generated. Accepted
359 sequential Generate sequential offset
360 identical Generate the same offset
362 'sequential' is only useful for random IO, where fio would
363 normally generate a new random offset for every IO. If you
364 append eg 8 to randread, you would get a new random offset for
365 every 8 IO's. The result would be a seek for only every 8
366 IO's, instead of for every IO. Use rw=randread:8 to specify
367 that. As sequential IO is already sequential, setting
368 'sequential' for that would not result in any differences.
369 'identical' behaves in a similar fashion, except it sends
370 the same offset 8 number of times before generating a new
373 kb_base=int The base unit for a kilobyte. The defacto base is 2^10, 1024.
374 Storage manufacturers like to use 10^3 or 1000 as a base
375 ten unit instead, for obvious reasons. Allow values are
376 1024 or 1000, with 1024 being the default.
378 unified_rw_reporting=bool Fio normally reports statistics on a per
379 data direction basis, meaning that read, write, and trim are
380 accounted and reported separately. If this option is set,
381 the fio will sum the results and report them as "mixed"
384 randrepeat=bool For random IO workloads, seed the generator in a predictable
385 way so that results are repeatable across repetitions.
387 randseed=int Seed the random number generators based on this seed value, to
388 be able to control what sequence of output is being generated.
389 If not set, the random sequence depends on the randrepeat
392 use_os_rand=bool Fio can either use the random generator supplied by the OS
393 to generator random offsets, or it can use it's own internal
394 generator (based on Tausworthe). Default is to use the
395 internal generator, which is often of better quality and
398 fallocate=str Whether pre-allocation is performed when laying down files.
401 none Do not pre-allocate space
402 posix Pre-allocate via posix_fallocate()
403 keep Pre-allocate via fallocate() with
404 FALLOC_FL_KEEP_SIZE set
405 0 Backward-compatible alias for 'none'
406 1 Backward-compatible alias for 'posix'
408 May not be available on all supported platforms. 'keep' is only
409 available on Linux.If using ZFS on Solaris this must be set to
410 'none' because ZFS doesn't support it. Default: 'posix'.
412 fadvise_hint=bool By default, fio will use fadvise() to advise the kernel
413 on what IO patterns it is likely to issue. Sometimes you
414 want to test specific IO patterns without telling the
415 kernel about it, in which case you can disable this option.
416 If set, fio will use POSIX_FADV_SEQUENTIAL for sequential
417 IO and POSIX_FADV_RANDOM for random IO.
419 size=int The total size of file io for this job. Fio will run until
420 this many bytes has been transferred, unless runtime is
421 limited by other options (such as 'runtime', for instance).
422 Unless specific nrfiles and filesize options are given,
423 fio will divide this size between the available files
424 specified by the job. If not set, fio will use the full
425 size of the given files or devices. If the the files
426 do not exist, size must be given. It is also possible to
427 give size as a percentage between 1 and 100. If size=20%
428 is given, fio will use 20% of the full size of the given
431 filesize=int Individual file sizes. May be a range, in which case fio
432 will select sizes for files at random within the given range
433 and limited to 'size' in total (if that is given). If not
434 given, each created file is the same size.
437 fill_fs=bool Sets size to something really large and waits for ENOSPC (no
438 space left on device) as the terminating condition. Only makes
439 sense with sequential write. For a read workload, the mount
440 point will be filled first then IO started on the result. This
441 option doesn't make sense if operating on a raw device node,
442 since the size of that is already known by the file system.
443 Additionally, writing beyond end-of-device will not return
447 bs=int The block size used for the io units. Defaults to 4k. Values
448 can be given for both read and writes. If a single int is
449 given, it will apply to both. If a second int is specified
450 after a comma, it will apply to writes only. In other words,
451 the format is either bs=read_and_write or bs=read,write,trim.
452 bs=4k,8k will thus use 4k blocks for reads, 8k blocks for
453 writes, and 8k for trims. You can terminate the list with
454 a trailing comma. bs=4k,8k, would use the default value for
455 trims.. If you only wish to set the write size, you
456 can do so by passing an empty read size - bs=,8k will set
457 8k for writes and leave the read default value.
460 ba=int At what boundary to align random IO offsets. Defaults to
461 the same as 'blocksize' the minimum blocksize given.
462 Minimum alignment is typically 512b for using direct IO,
463 though it usually depends on the hardware block size. This
464 option is mutually exclusive with using a random map for
465 files, so it will turn off that option.
467 blocksize_range=irange
468 bsrange=irange Instead of giving a single block size, specify a range
469 and fio will mix the issued io block sizes. The issued
470 io unit will always be a multiple of the minimum value
471 given (also see bs_unaligned). Applies to both reads and
472 writes, however a second range can be given after a comma.
475 bssplit=str Sometimes you want even finer grained control of the
476 block sizes issued, not just an even split between them.
477 This option allows you to weight various block sizes,
478 so that you are able to define a specific amount of
479 block sizes issued. The format for this option is:
481 bssplit=blocksize/percentage:blocksize/percentage
483 for as many block sizes as needed. So if you want to define
484 a workload that has 50% 64k blocks, 10% 4k blocks, and
485 40% 32k blocks, you would write:
487 bssplit=4k/10:64k/50:32k/40
489 Ordering does not matter. If the percentage is left blank,
490 fio will fill in the remaining values evenly. So a bssplit
491 option like this one:
493 bssplit=4k/50:1k/:32k/
495 would have 50% 4k ios, and 25% 1k and 32k ios. The percentages
496 always add up to 100, if bssplit is given a range that adds
497 up to more, it will error out.
499 bssplit also supports giving separate splits to reads and
500 writes. The format is identical to what bs= accepts. You
501 have to separate the read and write parts with a comma. So
502 if you want a workload that has 50% 2k reads and 50% 4k reads,
503 while having 90% 4k writes and 10% 8k writes, you would
506 bssplit=2k/50:4k/50,4k/90,8k/10
509 bs_unaligned If this option is given, any byte size value within bsrange
510 may be used as a block range. This typically wont work with
511 direct IO, as that normally requires sector alignment.
513 bs_is_seq_rand If this option is set, fio will use the normal read,write
514 blocksize settings as sequential,random instead. Any random
515 read or write will use the WRITE blocksize settings, and any
516 sequential read or write will use the READ blocksize setting.
518 zero_buffers If this option is given, fio will init the IO buffers to
519 all zeroes. The default is to fill them with random data.
521 refill_buffers If this option is given, fio will refill the IO buffers
522 on every submit. The default is to only fill it at init
523 time and reuse that data. Only makes sense if zero_buffers
524 isn't specified, naturally. If data verification is enabled,
525 refill_buffers is also automatically enabled.
527 scramble_buffers=bool If refill_buffers is too costly and the target is
528 using data deduplication, then setting this option will
529 slightly modify the IO buffer contents to defeat normal
530 de-dupe attempts. This is not enough to defeat more clever
531 block compression attempts, but it will stop naive dedupe of
532 blocks. Default: true.
534 buffer_compress_percentage=int If this is set, then fio will attempt to
535 provide IO buffer content (on WRITEs) that compress to
536 the specified level. Fio does this by providing a mix of
537 random data and zeroes. Note that this is per block size
538 unit, for file/disk wide compression level that matches
539 this setting, you'll also want to set refill_buffers.
541 buffer_compress_chunk=int See buffer_compress_percentage. This
542 setting allows fio to manage how big the ranges of random
543 data and zeroed data is. Without this set, fio will
544 provide buffer_compress_percentage of blocksize random
545 data, followed by the remaining zeroed. With this set
546 to some chunk size smaller than the block size, fio can
547 alternate random and zeroed data throughout the IO
550 nrfiles=int Number of files to use for this job. Defaults to 1.
552 openfiles=int Number of files to keep open at the same time. Defaults to
553 the same as nrfiles, can be set smaller to limit the number
556 file_service_type=str Defines how fio decides which file from a job to
557 service next. The following types are defined:
559 random Just choose a file at random.
561 roundrobin Round robin over open files. This
564 sequential Finish one file before moving on to
565 the next. Multiple files can still be
566 open depending on 'openfiles'.
568 The string can have a number appended, indicating how
569 often to switch to a new file. So if option random:4 is
570 given, fio will switch to a new random file after 4 ios
573 ioengine=str Defines how the job issues io to the file. The following
576 sync Basic read(2) or write(2) io. lseek(2) is
577 used to position the io location.
579 psync Basic pread(2) or pwrite(2) io.
581 vsync Basic readv(2) or writev(2) IO.
583 psyncv Basic preadv(2) or pwritev(2) IO.
585 libaio Linux native asynchronous io. Note that Linux
586 may only support queued behaviour with
587 non-buffered IO (set direct=1 or buffered=0).
588 This engine defines engine specific options.
590 posixaio glibc posix asynchronous io.
592 solarisaio Solaris native asynchronous io.
594 windowsaio Windows native asynchronous io.
596 mmap File is memory mapped and data copied
597 to/from using memcpy(3).
599 splice splice(2) is used to transfer the data and
600 vmsplice(2) to transfer data from user
603 syslet-rw Use the syslet system calls to make
604 regular read/write async.
606 sg SCSI generic sg v3 io. May either be
607 synchronous using the SG_IO ioctl, or if
608 the target is an sg character device
609 we use read(2) and write(2) for asynchronous
612 null Doesn't transfer any data, just pretends
613 to. This is mainly used to exercise fio
614 itself and for debugging/testing purposes.
616 net Transfer over the network to given host:port.
617 Depending on the protocol used, the hostname,
618 port, listen and filename options are used to
619 specify what sort of connection to make, while
620 the protocol option determines which protocol
622 This engine defines engine specific options.
624 netsplice Like net, but uses splice/vmsplice to
625 map data and send/receive.
626 This engine defines engine specific options.
628 cpuio Doesn't transfer any data, but burns CPU
629 cycles according to the cpuload= and
630 cpucycle= options. Setting cpuload=85
631 will cause that job to do nothing but burn
632 85% of the CPU. In case of SMP machines,
633 use numjobs=<no_of_cpu> to get desired CPU
634 usage, as the cpuload only loads a single
635 CPU at the desired rate.
637 guasi The GUASI IO engine is the Generic Userspace
638 Asyncronous Syscall Interface approach
641 http://www.xmailserver.org/guasi-lib.html
643 for more info on GUASI.
645 rdma The RDMA I/O engine supports both RDMA
646 memory semantics (RDMA_WRITE/RDMA_READ) and
647 channel semantics (Send/Recv) for the
648 InfiniBand, RoCE and iWARP protocols.
650 falloc IO engine that does regular fallocate to
651 simulate data transfer as fio ioengine.
652 DDIR_READ does fallocate(,mode = keep_size,)
653 DDIR_WRITE does fallocate(,mode = 0)
654 DDIR_TRIM does fallocate(,mode = punch_hole)
656 e4defrag IO engine that does regular EXT4_IOC_MOVE_EXT
657 ioctls to simulate defragment activity in
658 request to DDIR_WRITE event
660 external Prefix to specify loading an external
661 IO engine object file. Append the engine
662 filename, eg ioengine=external:/tmp/foo.o
663 to load ioengine foo.o in /tmp.
665 iodepth=int This defines how many io units to keep in flight against
666 the file. The default is 1 for each file defined in this
667 job, can be overridden with a larger value for higher
668 concurrency. Note that increasing iodepth beyond 1 will not
669 affect synchronous ioengines (except for small degress when
670 verify_async is in use). Even async engines may impose OS
671 restrictions causing the desired depth not to be achieved.
672 This may happen on Linux when using libaio and not setting
673 direct=1, since buffered IO is not async on that OS. Keep an
674 eye on the IO depth distribution in the fio output to verify
675 that the achieved depth is as expected. Default: 1.
677 iodepth_batch_submit=int
678 iodepth_batch=int This defines how many pieces of IO to submit at once.
679 It defaults to 1 which means that we submit each IO
680 as soon as it is available, but can be raised to submit
681 bigger batches of IO at the time.
683 iodepth_batch_complete=int This defines how many pieces of IO to retrieve
684 at once. It defaults to 1 which means that we'll ask
685 for a minimum of 1 IO in the retrieval process from
686 the kernel. The IO retrieval will go on until we
687 hit the limit set by iodepth_low. If this variable is
688 set to 0, then fio will always check for completed
689 events before queuing more IO. This helps reduce
690 IO latency, at the cost of more retrieval system calls.
692 iodepth_low=int The low water mark indicating when to start filling
693 the queue again. Defaults to the same as iodepth, meaning
694 that fio will attempt to keep the queue full at all times.
695 If iodepth is set to eg 16 and iodepth_low is set to 4, then
696 after fio has filled the queue of 16 requests, it will let
697 the depth drain down to 4 before starting to fill it again.
699 direct=bool If value is true, use non-buffered io. This is usually
700 O_DIRECT. Note that ZFS on Solaris doesn't support direct io.
701 On Windows the synchronous ioengines don't support direct io.
703 atomic=bool If value is true, attempt to use atomic direct IO. Atomic
704 writes are guaranteed to be stable once acknowledged by
705 the operating system. Only Linux supports O_ATOMIC right
708 buffered=bool If value is true, use buffered io. This is the opposite
709 of the 'direct' option. Defaults to true.
711 offset=int Start io at the given offset in the file. The data before
712 the given offset will not be touched. This effectively
713 caps the file size at real_size - offset.
715 offset_increment=int If this is provided, then the real offset becomes
716 the offset + offset_increment * thread_number, where the
717 thread number is a counter that starts at 0 and is incremented
718 for each job. This option is useful if there are several jobs
719 which are intended to operate on a file in parallel in disjoint
720 segments, with even spacing between the starting points.
722 number_ios=int Fio will normally perform IOs until it has exhausted the size
723 of the region set by size=, or if it exhaust the allocated
724 time (or hits an error condition). With this setting, the
725 range/size can be set independently of the number of IOs to
726 perform. When fio reaches this number, it will exit normally
729 fsync=int If writing to a file, issue a sync of the dirty data
730 for every number of blocks given. For example, if you give
731 32 as a parameter, fio will sync the file for every 32
732 writes issued. If fio is using non-buffered io, we may
733 not sync the file. The exception is the sg io engine, which
734 synchronizes the disk cache anyway.
736 fdatasync=int Like fsync= but uses fdatasync() to only sync data and not
738 In FreeBSD and Windows there is no fdatasync(), this falls back to
741 sync_file_range=str:val Use sync_file_range() for every 'val' number of
742 write operations. Fio will track range of writes that
743 have happened since the last sync_file_range() call. 'str'
744 can currently be one or more of:
746 wait_before SYNC_FILE_RANGE_WAIT_BEFORE
747 write SYNC_FILE_RANGE_WRITE
748 wait_after SYNC_FILE_RANGE_WAIT_AFTER
750 So if you do sync_file_range=wait_before,write:8, fio would
751 use SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE for
752 every 8 writes. Also see the sync_file_range(2) man page.
753 This option is Linux specific.
755 overwrite=bool If true, writes to a file will always overwrite existing
756 data. If the file doesn't already exist, it will be
757 created before the write phase begins. If the file exists
758 and is large enough for the specified write phase, nothing
761 end_fsync=bool If true, fsync file contents when a write stage has completed.
763 fsync_on_close=bool If true, fio will fsync() a dirty file on close.
764 This differs from end_fsync in that it will happen on every
765 file close, not just at the end of the job.
767 rwmixread=int How large a percentage of the mix should be reads.
769 rwmixwrite=int How large a percentage of the mix should be writes. If both
770 rwmixread and rwmixwrite is given and the values do not add
771 up to 100%, the latter of the two will be used to override
772 the first. This may interfere with a given rate setting,
773 if fio is asked to limit reads or writes to a certain rate.
774 If that is the case, then the distribution may be skewed.
776 random_distribution=str:float By default, fio will use a completely uniform
777 random distribution when asked to perform random IO. Sometimes
778 it is useful to skew the distribution in specific ways,
779 ensuring that some parts of the data is more hot than others.
780 fio includes the following distribution models:
782 random Uniform random distribution
783 zipf Zipf distribution
784 pareto Pareto distribution
786 When using a zipf or pareto distribution, an input value
787 is also needed to define the access pattern. For zipf, this
788 is the zipf theta. For pareto, it's the pareto power. Fio
789 includes a test program, genzipf, that can be used visualize
790 what the given input values will yield in terms of hit rates.
791 If you wanted to use zipf with a theta of 1.2, you would use
792 random_distribution=zipf:1.2 as the option. If a non-uniform
793 model is used, fio will disable use of the random map.
795 percentage_random=int For a random workload, set how big a percentage should
796 be random. This defaults to 100%, in which case the workload
797 is fully random. It can be set from anywhere from 0 to 100.
798 Setting it to 0 would make the workload fully sequential. Any
799 setting in between will result in a random mix of sequential
800 and random IO, at the given percentages. It is possible to
801 set different values for reads, writes, and trim. To do so,
802 simply use a comma separated list. See blocksize.
804 norandommap Normally fio will cover every block of the file when doing
805 random IO. If this option is given, fio will just get a
806 new random offset without looking at past io history. This
807 means that some blocks may not be read or written, and that
808 some blocks may be read/written more than once. This option
809 is mutually exclusive with verify= if and only if multiple
810 blocksizes (via bsrange=) are used, since fio only tracks
811 complete rewrites of blocks.
813 softrandommap=bool See norandommap. If fio runs with the random block map
814 enabled and it fails to allocate the map, if this option is
815 set it will continue without a random block map. As coverage
816 will not be as complete as with random maps, this option is
819 random_generator=str Fio supports the following engines for generating
820 IO offsets for random IO:
822 tausworthe Strong 2^88 cycle random number generator
823 lfsr Linear feedback shift register generator
825 Tausworthe is a strong random number generator, but it
826 requires tracking on the side if we want to ensure that
827 blocks are only read or written once. LFSR guarantees
828 that we never generate the same offset twice, and it's
829 also less computationally expensive. It's not a true
830 random generator, however, though for IO purposes it's
831 typically good enough. LFSR only works with single
832 block sizes, not with workloads that use multiple block
833 sizes. If used with such a workload, fio may read or write
834 some blocks multiple times.
836 nice=int Run the job with the given nice value. See man nice(2).
838 prio=int Set the io priority value of this job. Linux limits us to
839 a positive value between 0 and 7, with 0 being the highest.
842 prioclass=int Set the io priority class. See man ionice(1).
844 thinktime=int Stall the job x microseconds after an io has completed before
845 issuing the next. May be used to simulate processing being
846 done by an application. See thinktime_blocks and
850 Only valid if thinktime is set - pretend to spend CPU time
851 doing something with the data received, before falling back
852 to sleeping for the rest of the period specified by
856 Only valid if thinktime is set - control how many blocks
857 to issue, before waiting 'thinktime' usecs. If not set,
858 defaults to 1 which will make fio wait 'thinktime' usecs
859 after every block. This effectively makes any queue depth
860 setting redundant, since no more than 1 IO will be queued
861 before we have to complete it and do our thinktime. In
862 other words, this setting effectively caps the queue depth
863 if the latter is larger.
865 rate=int Cap the bandwidth used by this job. The number is in bytes/sec,
866 the normal suffix rules apply. You can use rate=500k to limit
867 reads and writes to 500k each, or you can specify read and
868 writes separately. Using rate=1m,500k would limit reads to
869 1MB/sec and writes to 500KB/sec. Capping only reads or
870 writes can be done with rate=,500k or rate=500k,. The former
871 will only limit writes (to 500KB/sec), the latter will only
874 ratemin=int Tell fio to do whatever it can to maintain at least this
875 bandwidth. Failing to meet this requirement, will cause
876 the job to exit. The same format as rate is used for
877 read vs write separation.
879 rate_iops=int Cap the bandwidth to this number of IOPS. Basically the same
880 as rate, just specified independently of bandwidth. If the
881 job is given a block size range instead of a fixed value,
882 the smallest block size is used as the metric. The same format
883 as rate is used for read vs write separation.
885 rate_iops_min=int If fio doesn't meet this rate of IO, it will cause
886 the job to exit. The same format as rate is used for read vs
889 latency_target=int If set, fio will attempt to find the max performance
890 point that the given workload will run at while maintaining a
891 latency below this target. The values is given in microseconds.
892 See latency_window and latency_percentile
894 latency_window=int Used with latency_target to specify the sample window
895 that the job is run at varying queue depths to test the
896 performance. The value is given in microseconds.
898 latency_percentile=float The percentage of IOs that must fall within the
899 criteria specified by latency_target and latency_window. If not
900 set, this defaults to 100.0, meaning that all IOs must be equal
901 or below to the value set by latency_target.
903 max_latency=int If set, fio will exit the job if it exceeds this maximum
904 latency. It will exit with an ETIME error.
906 ratecycle=int Average bandwidth for 'rate' and 'ratemin' over this number
909 cpumask=int Set the CPU affinity of this job. The parameter given is a
910 bitmask of allowed CPU's the job may run on. So if you want
911 the allowed CPUs to be 1 and 5, you would pass the decimal
912 value of (1 << 1 | 1 << 5), or 34. See man
913 sched_setaffinity(2). This may not work on all supported
914 operating systems or kernel versions. This option doesn't
915 work well for a higher CPU count than what you can store in
916 an integer mask, so it can only control cpus 1-32. For
917 boxes with larger CPU counts, use cpus_allowed.
919 cpus_allowed=str Controls the same options as cpumask, but it allows a text
920 setting of the permitted CPUs instead. So to use CPUs 1 and
921 5, you would specify cpus_allowed=1,5. This options also
922 allows a range of CPUs. Say you wanted a binding to CPUs
923 1, 5, and 8-15, you would set cpus_allowed=1,5,8-15.
925 numa_cpu_nodes=str Set this job running on spcified NUMA nodes' CPUs. The
926 arguments allow comma delimited list of cpu numbers,
927 A-B ranges, or 'all'. Note, to enable numa options support,
928 fio must be built on a system with libnuma-dev(el) installed.
930 numa_mem_policy=str Set this job's memory policy and corresponding NUMA
931 nodes. Format of the argements:
933 `mode' is one of the following memory policy:
934 default, prefer, bind, interleave, local
935 For `default' and `local' memory policy, no node is
936 needed to be specified.
937 For `prefer', only one node is allowed.
938 For `bind' and `interleave', it allow comma delimited
939 list of numbers, A-B ranges, or 'all'.
941 startdelay=time Start this job the specified number of seconds after fio
942 has started. Only useful if the job file contains several
943 jobs, and you want to delay starting some jobs to a certain
946 runtime=time Tell fio to terminate processing after the specified number
947 of seconds. It can be quite hard to determine for how long
948 a specified job will run, so this parameter is handy to
949 cap the total runtime to a given time.
951 time_based If set, fio will run for the duration of the runtime
952 specified even if the file(s) are completely read or
953 written. It will simply loop over the same workload
954 as many times as the runtime allows.
956 ramp_time=time If set, fio will run the specified workload for this amount
957 of time before logging any performance numbers. Useful for
958 letting performance settle before logging results, thus
959 minimizing the runtime required for stable results. Note
960 that the ramp_time is considered lead in time for a job,
961 thus it will increase the total runtime if a special timeout
962 or runtime is specified.
964 invalidate=bool Invalidate the buffer/page cache parts for this file prior
965 to starting io. Defaults to true.
967 sync=bool Use sync io for buffered writes. For the majority of the
968 io engines, this means using O_SYNC.
971 mem=str Fio can use various types of memory as the io unit buffer.
972 The allowed values are:
974 malloc Use memory from malloc(3) as the buffers.
976 shm Use shared memory as the buffers. Allocated
979 shmhuge Same as shm, but use huge pages as backing.
981 mmap Use mmap to allocate buffers. May either be
982 anonymous memory, or can be file backed if
983 a filename is given after the option. The
984 format is mem=mmap:/path/to/file.
986 mmaphuge Use a memory mapped huge file as the buffer
987 backing. Append filename after mmaphuge, ala
988 mem=mmaphuge:/hugetlbfs/file
990 The area allocated is a function of the maximum allowed
991 bs size for the job, multiplied by the io depth given. Note
992 that for shmhuge and mmaphuge to work, the system must have
993 free huge pages allocated. This can normally be checked
994 and set by reading/writing /proc/sys/vm/nr_hugepages on a
995 Linux system. Fio assumes a huge page is 4MB in size. So
996 to calculate the number of huge pages you need for a given
997 job file, add up the io depth of all jobs (normally one unless
998 iodepth= is used) and multiply by the maximum bs set. Then
999 divide that number by the huge page size. You can see the
1000 size of the huge pages in /proc/meminfo. If no huge pages
1001 are allocated by having a non-zero number in nr_hugepages,
1002 using mmaphuge or shmhuge will fail. Also see hugepage-size.
1004 mmaphuge also needs to have hugetlbfs mounted and the file
1005 location should point there. So if it's mounted in /huge,
1006 you would use mem=mmaphuge:/huge/somefile.
1008 iomem_align=int This indiciates the memory alignment of the IO memory buffers.
1009 Note that the given alignment is applied to the first IO unit
1010 buffer, if using iodepth the alignment of the following buffers
1011 are given by the bs used. In other words, if using a bs that is
1012 a multiple of the page sized in the system, all buffers will
1013 be aligned to this value. If using a bs that is not page
1014 aligned, the alignment of subsequent IO memory buffers is the
1015 sum of the iomem_align and bs used.
1018 Defines the size of a huge page. Must at least be equal
1019 to the system setting, see /proc/meminfo. Defaults to 4MB.
1020 Should probably always be a multiple of megabytes, so using
1021 hugepage-size=Xm is the preferred way to set this to avoid
1022 setting a non-pow-2 bad value.
1024 exitall When one job finishes, terminate the rest. The default is
1025 to wait for each job to finish, sometimes that is not the
1028 bwavgtime=int Average the calculated bandwidth over the given time. Value
1029 is specified in milliseconds.
1031 iopsavgtime=int Average the calculated IOPS over the given time. Value
1032 is specified in milliseconds.
1034 create_serialize=bool If true, serialize the file creating for the jobs.
1035 This may be handy to avoid interleaving of data
1036 files, which may greatly depend on the filesystem
1037 used and even the number of processors in the system.
1039 create_fsync=bool fsync the data file after creation. This is the
1042 create_on_open=bool Don't pre-setup the files for IO, just create open()
1043 when it's time to do IO to that file.
1045 create_only=bool If true, fio will only run the setup phase of the job.
1046 If files need to be laid out or updated on disk, only
1047 that will be done. The actual job contents are not
1050 pre_read=bool If this is given, files will be pre-read into memory before
1051 starting the given IO operation. This will also clear
1052 the 'invalidate' flag, since it is pointless to pre-read
1053 and then drop the cache. This will only work for IO engines
1054 that are seekable, since they allow you to read the same data
1055 multiple times. Thus it will not work on eg network or splice
1058 unlink=bool Unlink the job files when done. Not the default, as repeated
1059 runs of that job would then waste time recreating the file
1060 set again and again.
1062 loops=int Run the specified number of iterations of this job. Used
1063 to repeat the same workload a given number of times. Defaults
1066 do_verify=bool Run the verify phase after a write phase. Only makes sense if
1067 verify is set. Defaults to 1.
1069 verify=str If writing to a file, fio can verify the file contents
1070 after each iteration of the job. The allowed values are:
1072 md5 Use an md5 sum of the data area and store
1073 it in the header of each block.
1075 crc64 Use an experimental crc64 sum of the data
1076 area and store it in the header of each
1079 crc32c Use a crc32c sum of the data area and store
1080 it in the header of each block.
1082 crc32c-intel Use hardware assisted crc32c calcuation
1083 provided on SSE4.2 enabled processors. Falls
1084 back to regular software crc32c, if not
1085 supported by the system.
1087 crc32 Use a crc32 sum of the data area and store
1088 it in the header of each block.
1090 crc16 Use a crc16 sum of the data area and store
1091 it in the header of each block.
1093 crc7 Use a crc7 sum of the data area and store
1094 it in the header of each block.
1096 sha512 Use sha512 as the checksum function.
1098 sha256 Use sha256 as the checksum function.
1100 sha1 Use optimized sha1 as the checksum function.
1102 meta Write extra information about each io
1103 (timestamp, block number etc.). The block
1104 number is verified. See also verify_pattern.
1106 null Only pretend to verify. Useful for testing
1107 internals with ioengine=null, not for much
1110 This option can be used for repeated burn-in tests of a
1111 system to make sure that the written data is also
1112 correctly read back. If the data direction given is
1113 a read or random read, fio will assume that it should
1114 verify a previously written file. If the data direction
1115 includes any form of write, the verify will be of the
1118 verifysort=bool If set, fio will sort written verify blocks when it deems
1119 it faster to read them back in a sorted manner. This is
1120 often the case when overwriting an existing file, since
1121 the blocks are already laid out in the file system. You
1122 can ignore this option unless doing huge amounts of really
1123 fast IO where the red-black tree sorting CPU time becomes
1126 verify_offset=int Swap the verification header with data somewhere else
1127 in the block before writing. Its swapped back before
1130 verify_interval=int Write the verification header at a finer granularity
1131 than the blocksize. It will be written for chunks the
1132 size of header_interval. blocksize should divide this
1135 verify_pattern=str If set, fio will fill the io buffers with this
1136 pattern. Fio defaults to filling with totally random
1137 bytes, but sometimes it's interesting to fill with a known
1138 pattern for io verification purposes. Depending on the
1139 width of the pattern, fio will fill 1/2/3/4 bytes of the
1140 buffer at the time(it can be either a decimal or a hex number).
1141 The verify_pattern if larger than a 32-bit quantity has to
1142 be a hex number that starts with either "0x" or "0X". Use
1145 verify_fatal=bool Normally fio will keep checking the entire contents
1146 before quitting on a block verification failure. If this
1147 option is set, fio will exit the job on the first observed
1150 verify_dump=bool If set, dump the contents of both the original data
1151 block and the data block we read off disk to files. This
1152 allows later analysis to inspect just what kind of data
1153 corruption occurred. Off by default.
1155 verify_async=int Fio will normally verify IO inline from the submitting
1156 thread. This option takes an integer describing how many
1157 async offload threads to create for IO verification instead,
1158 causing fio to offload the duty of verifying IO contents
1159 to one or more separate threads. If using this offload
1160 option, even sync IO engines can benefit from using an
1161 iodepth setting higher than 1, as it allows them to have
1162 IO in flight while verifies are running.
1164 verify_async_cpus=str Tell fio to set the given CPU affinity on the
1165 async IO verification threads. See cpus_allowed for the
1168 verify_backlog=int Fio will normally verify the written contents of a
1169 job that utilizes verify once that job has completed. In
1170 other words, everything is written then everything is read
1171 back and verified. You may want to verify continually
1172 instead for a variety of reasons. Fio stores the meta data
1173 associated with an IO block in memory, so for large
1174 verify workloads, quite a bit of memory would be used up
1175 holding this meta data. If this option is enabled, fio
1176 will write only N blocks before verifying these blocks.
1178 verify_backlog_batch=int Control how many blocks fio will verify
1179 if verify_backlog is set. If not set, will default to
1180 the value of verify_backlog (meaning the entire queue
1181 is read back and verified). If verify_backlog_batch is
1182 less than verify_backlog then not all blocks will be verified,
1183 if verify_backlog_batch is larger than verify_backlog, some
1184 blocks will be verified more than once.
1187 wait_for_previous Wait for preceding jobs in the job file to exit, before
1188 starting this one. Can be used to insert serialization
1189 points in the job file. A stone wall also implies starting
1190 a new reporting group.
1192 new_group Start a new reporting group. See: group_reporting.
1194 numjobs=int Create the specified number of clones of this job. May be
1195 used to setup a larger number of threads/processes doing
1196 the same thing. Each thread is reported separately; to see
1197 statistics for all clones as a whole, use group_reporting in
1198 conjunction with new_group.
1200 group_reporting It may sometimes be interesting to display statistics for
1201 groups of jobs as a whole instead of for each individual job.
1202 This is especially true if 'numjobs' is used; looking at
1203 individual thread/process output quickly becomes unwieldy.
1204 To see the final report per-group instead of per-job, use
1205 'group_reporting'. Jobs in a file will be part of the same
1206 reporting group, unless if separated by a stonewall, or by
1209 thread fio defaults to forking jobs, however if this option is
1210 given, fio will use pthread_create(3) to create threads
1213 zonesize=int Divide a file into zones of the specified size. See zoneskip.
1215 zoneskip=int Skip the specified number of bytes when zonesize data has
1216 been read. The two zone options can be used to only do
1217 io on zones of a file.
1219 write_iolog=str Write the issued io patterns to the specified file. See
1220 read_iolog. Specify a separate file for each job, otherwise
1221 the iologs will be interspersed and the file may be corrupt.
1223 read_iolog=str Open an iolog with the specified file name and replay the
1224 io patterns it contains. This can be used to store a
1225 workload and replay it sometime later. The iolog given
1226 may also be a blktrace binary file, which allows fio
1227 to replay a workload captured by blktrace. See blktrace
1228 for how to capture such logging data. For blktrace replay,
1229 the file needs to be turned into a blkparse binary data
1230 file first (blkparse <device> -o /dev/null -d file_for_fio.bin).
1232 replay_no_stall=int When replaying I/O with read_iolog the default behavior
1233 is to attempt to respect the time stamps within the log and
1234 replay them with the appropriate delay between IOPS. By
1235 setting this variable fio will not respect the timestamps and
1236 attempt to replay them as fast as possible while still
1237 respecting ordering. The result is the same I/O pattern to a
1238 given device, but different timings.
1240 replay_redirect=str While replaying I/O patterns using read_iolog the
1241 default behavior is to replay the IOPS onto the major/minor
1242 device that each IOP was recorded from. This is sometimes
1243 undesirable because on a different machine those major/minor
1244 numbers can map to a different device. Changing hardware on
1245 the same system can also result in a different major/minor
1246 mapping. Replay_redirect causes all IOPS to be replayed onto
1247 the single specified device regardless of the device it was
1248 recorded from. i.e. replay_redirect=/dev/sdc would cause all
1249 IO in the blktrace to be replayed onto /dev/sdc. This means
1250 multiple devices will be replayed onto a single, if the trace
1251 contains multiple devices. If you want multiple devices to be
1252 replayed concurrently to multiple redirected devices you must
1253 blkparse your trace into separate traces and replay them with
1254 independent fio invocations. Unfortuantely this also breaks
1255 the strict time ordering between multiple device accesses.
1257 write_bw_log=str If given, write a bandwidth log of the jobs in this job
1258 file. Can be used to store data of the bandwidth of the
1259 jobs in their lifetime. The included fio_generate_plots
1260 script uses gnuplot to turn these text files into nice
1261 graphs. See write_lat_log for behaviour of given
1262 filename. For this option, the suffix is _bw.log.
1264 write_lat_log=str Same as write_bw_log, except that this option stores io
1265 submission, completion, and total latencies instead. If no
1266 filename is given with this option, the default filename of
1267 "jobname_type.log" is used. Even if the filename is given,
1268 fio will still append the type of log. So if one specifies
1272 The actual log names will be foo_slat.log, foo_clat.log,
1273 and foo_lat.log. This helps fio_generate_plot fine the logs
1276 write_iops_log=str Same as write_bw_log, but writes IOPS. If no filename is
1277 given with this option, the default filename of
1278 "jobname_type.log" is used. Even if the filename is given,
1279 fio will still append the type of log.
1281 log_avg_msec=int By default, fio will log an entry in the iops, latency,
1282 or bw log for every IO that completes. When writing to the
1283 disk log, that can quickly grow to a very large size. Setting
1284 this option makes fio average the each log entry over the
1285 specified period of time, reducing the resolution of the log.
1288 lockmem=int Pin down the specified amount of memory with mlock(2). Can
1289 potentially be used instead of removing memory or booting
1290 with less memory to simulate a smaller amount of memory.
1291 The amount specified is per worker.
1293 exec_prerun=str Before running this job, issue the command specified
1294 through system(3). Output is redirected in a file called
1297 exec_postrun=str After the job completes, issue the command specified
1298 though system(3). Output is redirected in a file called
1299 jobname.postrun.txt.
1301 ioscheduler=str Attempt to switch the device hosting the file to the specified
1302 io scheduler before running.
1304 disk_util=bool Generate disk utilization statistics, if the platform
1305 supports it. Defaults to on.
1307 disable_lat=bool Disable measurements of total latency numbers. Useful
1308 only for cutting back the number of calls to gettimeofday,
1309 as that does impact performance at really high IOPS rates.
1310 Note that to really get rid of a large amount of these
1311 calls, this option must be used with disable_slat and
1314 disable_clat=bool Disable measurements of completion latency numbers. See
1317 disable_slat=bool Disable measurements of submission latency numbers. See
1320 disable_bw=bool Disable measurements of throughput/bandwidth numbers. See
1323 clat_percentiles=bool Enable the reporting of percentiles of
1324 completion latencies.
1326 percentile_list=float_list Overwrite the default list of percentiles
1327 for completion latencies. Each number is a floating
1328 number in the range (0,100], and the maximum length of
1329 the list is 20. Use ':' to separate the numbers, and
1330 list the numbers in ascending order. For example,
1331 --percentile_list=99.5:99.9 will cause fio to report
1332 the values of completion latency below which 99.5% and
1333 99.9% of the observed latencies fell, respectively.
1335 clocksource=str Use the given clocksource as the base of timing. The
1336 supported options are:
1338 gettimeofday gettimeofday(2)
1340 clock_gettime clock_gettime(2)
1342 cpu Internal CPU clock source
1344 cpu is the preferred clocksource if it is reliable, as it
1345 is very fast (and fio is heavy on time calls). Fio will
1346 automatically use this clocksource if it's supported and
1347 considered reliable on the system it is running on, unless
1348 another clocksource is specifically set. For x86/x86-64 CPUs,
1349 this means supporting TSC Invariant.
1351 gtod_reduce=bool Enable all of the gettimeofday() reducing options
1352 (disable_clat, disable_slat, disable_bw) plus reduce
1353 precision of the timeout somewhat to really shrink
1354 the gettimeofday() call count. With this option enabled,
1355 we only do about 0.4% of the gtod() calls we would have
1356 done if all time keeping was enabled.
1358 gtod_cpu=int Sometimes it's cheaper to dedicate a single thread of
1359 execution to just getting the current time. Fio (and
1360 databases, for instance) are very intensive on gettimeofday()
1361 calls. With this option, you can set one CPU aside for
1362 doing nothing but logging current time to a shared memory
1363 location. Then the other threads/processes that run IO
1364 workloads need only copy that segment, instead of entering
1365 the kernel with a gettimeofday() call. The CPU set aside
1366 for doing these time calls will be excluded from other
1367 uses. Fio will manually clear it from the CPU mask of other
1370 continue_on_error=str Normally fio will exit the job on the first observed
1371 failure. If this option is set, fio will continue the job when
1372 there is a 'non-fatal error' (EIO or EILSEQ) until the runtime
1373 is exceeded or the I/O size specified is completed. If this
1374 option is used, there are two more stats that are appended,
1375 the total error count and the first error. The error field
1376 given in the stats is the first error that was hit during the
1379 The allowed values are:
1381 none Exit on any IO or verify errors.
1383 read Continue on read errors, exit on all others.
1385 write Continue on write errors, exit on all others.
1387 io Continue on any IO error, exit on all others.
1389 verify Continue on verify errors, exit on all others.
1391 all Continue on all errors.
1393 0 Backward-compatible alias for 'none'.
1395 1 Backward-compatible alias for 'all'.
1397 ignore_error=str Sometimes you want to ignore some errors during test
1398 in that case you can specify error list for each error type.
1399 ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST
1400 errors for given error type is separated with ':'. Error
1401 may be symbol ('ENOSPC', 'ENOMEM') or integer.
1403 ignore_error=EAGAIN,ENOSPC:122
1404 This option will ignore EAGAIN from READ, and ENOSPC and
1405 122(EDQUOT) from WRITE.
1407 error_dump=bool If set dump every error even if it is non fatal, true
1408 by default. If disabled only fatal error will be dumped
1410 cgroup=str Add job to this control group. If it doesn't exist, it will
1411 be created. The system must have a mounted cgroup blkio
1412 mount point for this to work. If your system doesn't have it
1413 mounted, you can do so with:
1415 # mount -t cgroup -o blkio none /cgroup
1417 cgroup_weight=int Set the weight of the cgroup to this value. See
1418 the documentation that comes with the kernel, allowed values
1419 are in the range of 100..1000.
1421 cgroup_nodelete=bool Normally fio will delete the cgroups it has created after
1422 the job completion. To override this behavior and to leave
1423 cgroups around after the job completion, set cgroup_nodelete=1.
1424 This can be useful if one wants to inspect various cgroup
1425 files after job completion. Default: false
1427 uid=int Instead of running as the invoking user, set the user ID to
1428 this value before the thread/process does any work.
1430 gid=int Set group ID, see uid.
1432 flow_id=int The ID of the flow. If not specified, it defaults to being a
1433 global flow. See flow.
1435 flow=int Weight in token-based flow control. If this value is used, then
1436 there is a 'flow counter' which is used to regulate the
1437 proportion of activity between two or more jobs. fio attempts
1438 to keep this flow counter near zero. The 'flow' parameter
1439 stands for how much should be added or subtracted to the flow
1440 counter on each iteration of the main I/O loop. That is, if
1441 one job has flow=8 and another job has flow=-1, then there
1442 will be a roughly 1:8 ratio in how much one runs vs the other.
1444 flow_watermark=int The maximum value that the absolute value of the flow
1445 counter is allowed to reach before the job must wait for a
1446 lower value of the counter.
1448 flow_sleep=int The period of time, in microseconds, to wait after the flow
1449 watermark has been exceeded before retrying operations
1451 In addition, there are some parameters which are only valid when a specific
1452 ioengine is in use. These are used identically to normal parameters, with the
1453 caveat that when used on the command line, they must come after the ioengine
1454 that defines them is selected.
1456 [libaio] userspace_reap Normally, with the libaio engine in use, fio will use
1457 the io_getevents system call to reap newly returned events.
1458 With this flag turned on, the AIO ring will be read directly
1459 from user-space to reap events. The reaping mode is only
1460 enabled when polling for a minimum of 0 events (eg when
1461 iodepth_batch_complete=0).
1463 [cpu] cpuload=int Attempt to use the specified percentage of CPU cycles.
1465 [cpu] cpuchunks=int Split the load into cycles of the given time. In
1468 [netsplice] hostname=str
1469 [net] hostname=str The host name or IP address to use for TCP or UDP based IO.
1470 If the job is a TCP listener or UDP reader, the hostname is not
1471 used and must be omitted unless it is a valid UDP multicast
1474 [netsplice] port=int
1475 [net] port=int The TCP or UDP port to bind to or connect to.
1477 [netsplice] interface=str
1478 [net] interface=str The IP address of the network interface used to send or
1479 receive UDP multicast
1482 [net] ttl=int Time-to-live value for outgoing UDP multicast packets.
1485 [netsplice] nodelay=bool
1486 [net] nodelay=bool Set TCP_NODELAY on TCP connections.
1488 [netsplice] protocol=str
1489 [netsplice] proto=str
1491 [net] proto=str The network protocol to use. Accepted values are:
1493 tcp Transmission control protocol
1494 udp User datagram protocol
1495 unix UNIX domain socket
1497 When the protocol is TCP or UDP, the port must also be given,
1498 as well as the hostname if the job is a TCP listener or UDP
1499 reader. For unix sockets, the normal filename option should be
1500 used and the port is invalid.
1502 [net] listen For TCP network connections, tell fio to listen for incoming
1503 connections rather than initiating an outgoing connection. The
1504 hostname must be omitted if this option is used.
1505 [net] pingpong Normaly a network writer will just continue writing data, and
1506 a network reader will just consume packages. If pingpong=1
1507 is set, a writer will send its normal payload to the reader,
1508 then wait for the reader to send the same payload back. This
1509 allows fio to measure network latencies. The submission
1510 and completion latencies then measure local time spent
1511 sending or receiving, and the completion latency measures
1512 how long it took for the other end to receive and send back.
1513 For UDP multicast traffic pingpong=1 should only be set for a
1514 single reader when multiple readers are listening to the same
1517 [e4defrag] donorname=str
1518 File will be used as a block donor(swap extents between files)
1519 [e4defrag] inplace=int
1520 Configure donor file blocks allocation strategy
1521 0(default): Preallocate donor's file on init
1522 1 : allocate space immidietly inside defragment event,
1523 and free right after event
1527 6.0 Interpreting the output
1528 ---------------------------
1530 fio spits out a lot of output. While running, fio will display the
1531 status of the jobs created. An example of that would be:
1533 Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
1535 The characters inside the square brackets denote the current status of
1536 each thread. The possible values (in typical life cycle order) are:
1540 P Thread setup, but not started.
1542 I Thread initialized, waiting or generating necessary data.
1543 p Thread running pre-reading file(s).
1544 R Running, doing sequential reads.
1545 r Running, doing random reads.
1546 W Running, doing sequential writes.
1547 w Running, doing random writes.
1548 M Running, doing mixed sequential reads/writes.
1549 m Running, doing mixed random reads/writes.
1550 F Running, currently waiting for fsync()
1551 V Running, doing verification of written data.
1552 E Thread exited, not reaped by main thread yet.
1554 X Thread reaped, exited with an error.
1555 K Thread reaped, exited due to signal.
1557 The other values are fairly self explanatory - number of threads
1558 currently running and doing io, rate of io since last check (read speed
1559 listed first, then write speed), and the estimated completion percentage
1560 and time for the running group. It's impossible to estimate runtime of
1561 the following groups (if any). Note that the string is displayed in order,
1562 so it's possible to tell which of the jobs are currently doing what. The
1563 first character is the first job defined in the job file, and so forth.
1565 When fio is done (or interrupted by ctrl-c), it will show the data for
1566 each thread, group of threads, and disks in that order. For each data
1567 direction, the output looks like:
1569 Client1 (g=0): err= 0:
1570 write: io= 32MB, bw= 666KB/s, iops=89 , runt= 50320msec
1571 slat (msec): min= 0, max= 136, avg= 0.03, stdev= 1.92
1572 clat (msec): min= 0, max= 631, avg=48.50, stdev=86.82
1573 bw (KB/s) : min= 0, max= 1196, per=51.00%, avg=664.02, stdev=681.68
1574 cpu : usr=1.49%, sys=0.25%, ctx=7969, majf=0, minf=17
1575 IO depths : 1=0.1%, 2=0.3%, 4=0.5%, 8=99.0%, 16=0.0%, 32=0.0%, >32=0.0%
1576 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
1577 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
1578 issued r/w: total=0/32768, short=0/0
1579 lat (msec): 2=1.6%, 4=0.0%, 10=3.2%, 20=12.8%, 50=38.4%, 100=24.8%,
1580 lat (msec): 250=15.2%, 500=0.0%, 750=0.0%, 1000=0.0%, >=2048=0.0%
1582 The client number is printed, along with the group id and error of that
1583 thread. Below is the io statistics, here for writes. In the order listed,
1586 io= Number of megabytes io performed
1587 bw= Average bandwidth rate
1588 iops= Average IOs performed per second
1589 runt= The runtime of that thread
1590 slat= Submission latency (avg being the average, stdev being the
1591 standard deviation). This is the time it took to submit
1592 the io. For sync io, the slat is really the completion
1593 latency, since queue/complete is one operation there. This
1594 value can be in milliseconds or microseconds, fio will choose
1595 the most appropriate base and print that. In the example
1596 above, milliseconds is the best scale. Note: in --minimal mode
1597 latencies are always expressed in microseconds.
1598 clat= Completion latency. Same names as slat, this denotes the
1599 time from submission to completion of the io pieces. For
1600 sync io, clat will usually be equal (or very close) to 0,
1601 as the time from submit to complete is basically just
1602 CPU time (io has already been done, see slat explanation).
1603 bw= Bandwidth. Same names as the xlat stats, but also includes
1604 an approximate percentage of total aggregate bandwidth
1605 this thread received in this group. This last value is
1606 only really useful if the threads in this group are on the
1607 same disk, since they are then competing for disk access.
1608 cpu= CPU usage. User and system time, along with the number
1609 of context switches this thread went through, usage of
1610 system and user time, and finally the number of major
1611 and minor page faults.
1612 IO depths= The distribution of io depths over the job life time. The
1613 numbers are divided into powers of 2, so for example the
1614 16= entries includes depths up to that value but higher
1615 than the previous entry. In other words, it covers the
1616 range from 16 to 31.
1617 IO submit= How many pieces of IO were submitting in a single submit
1618 call. Each entry denotes that amount and below, until
1619 the previous entry - eg, 8=100% mean that we submitted
1620 anywhere in between 5-8 ios per submit call.
1621 IO complete= Like the above submit number, but for completions instead.
1622 IO issued= The number of read/write requests issued, and how many
1624 IO latencies= The distribution of IO completion latencies. This is the
1625 time from when IO leaves fio and when it gets completed.
1626 The numbers follow the same pattern as the IO depths,
1627 meaning that 2=1.6% means that 1.6% of the IO completed
1628 within 2 msecs, 20=12.8% means that 12.8% of the IO
1629 took more than 10 msecs, but less than (or equal to) 20 msecs.
1631 After each client has been listed, the group statistics are printed. They
1632 will look like this:
1634 Run status group 0 (all jobs):
1635 READ: io=64MB, aggrb=22178, minb=11355, maxb=11814, mint=2840msec, maxt=2955msec
1636 WRITE: io=64MB, aggrb=1302, minb=666, maxb=669, mint=50093msec, maxt=50320msec
1638 For each data direction, it prints:
1640 io= Number of megabytes io performed.
1641 aggrb= Aggregate bandwidth of threads in this group.
1642 minb= The minimum average bandwidth a thread saw.
1643 maxb= The maximum average bandwidth a thread saw.
1644 mint= The smallest runtime of the threads in that group.
1645 maxt= The longest runtime of the threads in that group.
1647 And finally, the disk statistics are printed. They will look like this:
1649 Disk stats (read/write):
1650 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
1652 Each value is printed for both reads and writes, with reads first. The
1655 ios= Number of ios performed by all groups.
1656 merge= Number of merges io the io scheduler.
1657 ticks= Number of ticks we kept the disk busy.
1658 io_queue= Total time spent in the disk queue.
1659 util= The disk utilization. A value of 100% means we kept the disk
1660 busy constantly, 50% would be a disk idling half of the time.
1662 It is also possible to get fio to dump the current output while it is
1663 running, without terminating the job. To do that, send fio the USR1 signal.
1664 You can also get regularly timed dumps by using the --status-interval
1665 parameter, or by creating a file in /tmp named fio-dump-status. If fio
1666 sees this file, it will unlink it and dump the current output status.
1672 For scripted usage where you typically want to generate tables or graphs
1673 of the results, fio can output the results in a semicolon separated format.
1674 The format is one long line of values, such as:
1676 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%
1677 A description of this job goes here.
1679 The job description (if provided) follows on a second line.
1681 To enable terse output, use the --minimal command line option. The first
1682 value is the version of the terse output format. If the output has to
1683 be changed for some reason, this number will be incremented by 1 to
1684 signify that change.
1686 Split up, the format is as follows:
1688 terse version, fio version, jobname, groupid, error
1690 Total IO (KB), bandwidth (KB/sec), IOPS, runtime (msec)
1691 Submission latency: min, max, mean, deviation (usec)
1692 Completion latency: min, max, mean, deviation (usec)
1693 Completion latency percentiles: 20 fields (see below)
1694 Total latency: min, max, mean, deviation (usec)
1695 Bw (KB/s): min, max, aggregate percentage of total, mean, deviation
1697 Total IO (KB), bandwidth (KB/sec), IOPS, runtime (msec)
1698 Submission latency: min, max, mean, deviation (usec)
1699 Completion latency: min, max, mean, deviation (usec)
1700 Completion latency percentiles: 20 fields (see below)
1701 Total latency: min, max, mean, deviation (usec)
1702 Bw (KB/s): min, max, aggregate percentage of total, mean, deviation
1703 CPU usage: user, system, context switches, major faults, minor faults
1704 IO depths: <=1, 2, 4, 8, 16, 32, >=64
1705 IO latencies microseconds: <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
1706 IO latencies milliseconds: <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
1707 Disk utilization: Disk name, Read ios, write ios,
1708 Read merges, write merges,
1709 Read ticks, write ticks,
1710 Time spent in queue, disk utilization percentage
1711 Additional Info (dependent on continue_on_error, default off): total # errors, first error code
1713 Additional Info (dependent on description being set): Text description
1715 Completion latency percentiles can be a grouping of up to 20 sets, so
1716 for the terse output fio writes all of them. Each field will look like this:
1720 which is the Xth percentile, and the usec latency associated with it.
1722 For disk utilization, all disks used by fio are shown. So for each disk
1723 there will be a disk utilization section.
1726 8.0 Trace file format
1727 ---------------------
1728 There are two trace file format that you can encounter. The older (v1) format
1729 is unsupported since version 1.20-rc3 (March 2008). It will still be described
1730 below in case that you get an old trace and want to understand it.
1732 In any case the trace is a simple text file with a single action per line.
1735 8.1 Trace file format v1
1736 ------------------------
1737 Each line represents a single io action in the following format:
1741 where rw=0/1 for read/write, and the offset and length entries being in bytes.
1743 This format is not supported in Fio versions => 1.20-rc3.
1746 8.2 Trace file format v2
1747 ------------------------
1748 The second version of the trace file format was added in Fio version 1.17.
1749 It allows to access more then one file per trace and has a bigger set of
1750 possible file actions.
1752 The first line of the trace file has to be:
1756 Following this can be lines in two different formats, which are described below.
1758 The file management format:
1762 The filename is given as an absolute path. The action can be one of these:
1764 add Add the given filename to the trace
1765 open Open the file with the given filename. The filename has to have
1766 been added with the add action before.
1767 close Close the file with the given filename. The file has to have been
1771 The file io action format:
1773 filename action offset length
1775 The filename is given as an absolute path, and has to have been added and opened
1776 before it can be used with this format. The offset and length are given in
1777 bytes. The action can be one of these:
1779 wait Wait for 'offset' microseconds. Everything below 100 is discarded.
1780 read Read 'length' bytes beginning from 'offset'
1781 write Write 'length' bytes beginning from 'offset'
1782 sync fsync() the file
1783 datasync fdatasync() the file
1784 trim trim the given file from the given 'offset' for 'length' bytes
1787 9.0 CPU idleness profiling
1788 --------------------------
1789 In some cases, we want to understand CPU overhead in a test. For example,
1790 we test patches for the specific goodness of whether they reduce CPU usage.
1791 fio implements a balloon approach to create a thread per CPU that runs at
1792 idle priority, meaning that it only runs when nobody else needs the cpu.
1793 By measuring the amount of work completed by the thread, idleness of each
1794 CPU can be derived accordingly.
1796 An unit work is defined as touching a full page of unsigned characters. Mean
1797 and standard deviation of time to complete an unit work is reported in "unit
1798 work" section. Options can be chosen to report detailed percpu idleness or
1799 overall system idleness by aggregating percpu stats.
projects / fio.git / blob