8 5. Detailed list of parameters
13 1.0 Overview and history
14 ------------------------
15 fio was originally written to save me the hassle of writing special test
16 case programs when I wanted to test a specific workload, either for
17 performance reasons or to find/reproduce a bug. The process of writing
18 such a test app can be tiresome, especially if you have to do it often.
19 Hence I needed a tool that would be able to simulate a given io workload
20 without resorting to writing a tailored test case again and again.
22 A test work load is difficult to define, though. There can be any number
23 of processes or threads involved, and they can each be using their own
24 way of generating io. You could have someone dirtying large amounts of
25 memory in an memory mapped file, or maybe several threads issuing
26 reads using asynchronous io. fio needed to be flexible enough to
27 simulate both of these cases, and many more.
31 The first step in getting fio to simulate a desired io workload, is
32 writing a job file describing that specific setup. A job file may contain
33 any number of threads and/or files - the typical contents of the job file
34 is a global section defining shared parameters, and one or more job
35 sections describing the jobs involved. When run, fio parses this file
36 and sets everything up as described. If we break down a job from top to
37 bottom, it contains the following basic parameters:
39 IO type Defines the io pattern issued to the file(s).
40 We may only be reading sequentially from this
41 file(s), or we may be writing randomly. Or even
42 mixing reads and writes, sequentially or randomly.
44 Block size In how large chunks are we issuing io? This may be
45 a single value, or it may describe a range of
48 IO size How much data are we going to be reading/writing.
50 IO engine How do we issue io? We could be memory mapping the
51 file, we could be using regular read/write, we
52 could be using splice, async io, syslet, or even
55 IO depth If the io engine is async, how large a queuing
56 depth do we want to maintain?
58 IO type Should we be doing buffered io, or direct/raw io?
60 Num files How many files are we spreading the workload over.
62 Num threads How many threads or processes should we spread
65 The above are the basic parameters defined for a workload, in addition
66 there's a multitude of parameters that modify other aspects of how this
72 See the README file for command line parameters, there are only a few
75 Running fio is normally the easiest part - you just give it the job file
76 (or job files) as parameters:
80 and it will start doing what the job_file tells it to do. You can give
81 more than one job file on the command line, fio will serialize the running
82 of those files. Internally that is the same as using the 'stonewall'
83 parameter described the the parameter section.
85 If the job file contains only one job, you may as well just give the
86 parameters on the command line. The command line parameters are identical
87 to the job parameters, with a few extra that control global parameters
88 (see README). For example, for the job file parameter iodepth=2, the
89 mirror command line option would be --iodepth 2 or --iodepth=2. You can
90 also use the command line for giving more than one job entry. For each
91 --name option that fio sees, it will start a new job with that name.
92 Command line entries following a --name entry will apply to that job,
93 until there are no more entries or a new --name entry is seen. This is
94 similar to the job file options, where each option applies to the current
95 job until a new [] job entry is seen.
97 fio does not need to run as root, except if the files or devices specified
98 in the job section requires that. Some other options may also be restricted,
99 such as memory locking, io scheduler switching, and decreasing the nice value.
104 As previously described, fio accepts one or more job files describing
105 what it is supposed to do. The job file format is the classic ini file,
106 where the names enclosed in [] brackets define the job name. You are free
107 to use any ascii name you want, except 'global' which has special meaning.
108 A global section sets defaults for the jobs described in that file. A job
109 may override a global section parameter, and a job file may even have
110 several global sections if so desired. A job is only affected by a global
111 section residing above it. If the first character in a line is a ';' or a
112 '#', the entire line is discarded as a comment.
114 So let's look at a really simple job file that defines two processes, each
115 randomly reading from a 128MB file.
117 ; -- start job file --
128 As you can see, the job file sections themselves are empty as all the
129 described parameters are shared. As no filename= option is given, fio
130 makes up a filename for each of the jobs as it sees fit. On the command
131 line, this job would look as follows:
133 $ fio --name=global --rw=randread --size=128m --name=job1 --name=job2
136 Let's look at an example that has a number of processes writing randomly
139 ; -- start job file --
151 Here we have no global section, as we only have one job defined anyway.
152 We want to use async io here, with a depth of 4 for each file. We also
153 increased the buffer size used to 32KB and define numjobs to 4 to
154 fork 4 identical jobs. The result is 4 processes each randomly writing
155 to their own 64MB file. Instead of using the above job file, you could
156 have given the parameters on the command line. For this case, you would
159 $ fio --name=random-writers --ioengine=libaio --iodepth=4 --rw=randwrite --bs=32k --direct=0 --size=64m --numjobs=4
161 4.1 Environment variables
162 -------------------------
164 fio also supports environment variable expansion in job files. Any
165 substring of the form "${VARNAME}" as part of an option value (in other
166 words, on the right of the `='), will be expanded to the value of the
167 environment variable called VARNAME. If no such environment variable
168 is defined, or VARNAME is the empty string, the empty string will be
171 As an example, let's look at a sample fio invocation and job file:
173 $ SIZE=64m NUMJOBS=4 fio jobfile.fio
175 ; -- start job file --
182 This will expand to the following equivalent job file at runtime:
184 ; -- start job file --
191 fio ships with a few example job files, you can also look there for
194 4.2 Reserved keywords
195 ---------------------
197 Additionally, fio has a set of reserved keywords that will be replaced
198 internally with the appropriate value. Those keywords are:
200 $pagesize The architecture page size of the running system
201 $mb_memory Megabytes of total memory in the system
202 $ncpus Number of online available CPUs
204 These can be used on the command line or in the job file, and will be
205 automatically substituted with the current system values when the job
206 is run. Simple math is also supported on these keywords, so you can
207 perform actions like:
211 and get that properly expanded to 8 times the size of memory in the
215 5.0 Detailed list of parameters
216 -------------------------------
218 This section describes in details each parameter associated with a job.
219 Some parameters take an option of a given type, such as an integer or
220 a string. The following types are used:
222 str String. This is a sequence of alpha characters.
223 time Integer with possible time suffix. In seconds unless otherwise
224 specified, use eg 10m for 10 minutes. Accepts s/m/h for seconds,
226 int SI integer. A whole number value, which may contain a suffix
227 describing the base of the number. Accepted suffixes are k/m/g/t/p,
228 meaning kilo, mega, giga, tera, and peta. The suffix is not case
229 sensitive, and you may also include trailing 'b' (eg 'kb' is the same
230 as 'k'). So if you want to specify 4096, you could either write
231 out '4096' or just give 4k. The suffixes signify base 2 values, so
232 1024 is 1k and 1024k is 1m and so on, unless the suffix is explicitly
233 set to a base 10 value using 'kib', 'mib', 'gib', etc. If that is the
234 case, then 1000 is used as the multiplier. This can be handy for
235 disks, since manufacturers generally use base 10 values when listing
236 the capacity of a drive. If the option accepts an upper and lower
237 range, use a colon ':' or minus '-' to separate such values. May also
238 include a prefix to indicate numbers base. If 0x is used, the number
239 is assumed to be hexadecimal. See irange.
240 bool Boolean. Usually parsed as an integer, however only defined for
241 true and false (1 and 0).
242 irange Integer range with suffix. Allows value range to be given, such
243 as 1024-4096. A colon may also be used as the separator, eg
244 1k:4k. If the option allows two sets of ranges, they can be
245 specified with a ',' or '/' delimiter: 1k-4k/8k-32k. Also see
247 float_list A list of floating numbers, separated by a ':' character.
249 With the above in mind, here follows the complete list of fio job
252 name=str ASCII name of the job. This may be used to override the
253 name printed by fio for this job. Otherwise the job
254 name is used. On the command line this parameter has the
255 special purpose of also signaling the start of a new
258 description=str Text description of the job. Doesn't do anything except
259 dump this text description when this job is run. It's
262 directory=str Prefix filenames with this directory. Used to place files
263 in a different location than "./".
265 filename=str Fio normally makes up a filename based on the job name,
266 thread number, and file number. If you want to share
267 files between threads in a job or several jobs, specify
268 a filename for each of them to override the default. If
269 the ioengine used is 'net', the filename is the host, port,
270 and protocol to use in the format of =host,port,protocol.
271 See ioengine=net for more. If the ioengine is file based, you
272 can specify a number of files by separating the names with a
273 ':' colon. So if you wanted a job to open /dev/sda and /dev/sdb
274 as the two working files, you would use
275 filename=/dev/sda:/dev/sdb. On Windows, disk devices are accessed
276 as \\.\PhysicalDrive0 for the first device, \\.\PhysicalDrive1
277 for the second etc. If the wanted filename does need to
278 include a colon, then escape that with a '\' character.
279 For instance, if the filename is "/dev/dsk/foo@3,0:c",
280 then you would use filename="/dev/dsk/foo@3,0\:c".
281 '-' is a reserved name, meaning stdin or stdout. Which of the
282 two depends on the read/write direction set.
284 opendir=str Tell fio to recursively add any file it can find in this
285 directory and down the file system tree.
287 lockfile=str Fio defaults to not locking any files before it does
288 IO to them. If a file or file descriptor is shared, fio
289 can serialize IO to that file to make the end result
290 consistent. This is usual for emulating real workloads that
291 share files. The lock modes are:
293 none No locking. The default.
294 exclusive Only one thread/process may do IO,
295 excluding all others.
296 readwrite Read-write locking on the file. Many
297 readers may access the file at the
298 same time, but writes get exclusive
301 The option may be post-fixed with a lock batch number. If
302 set, then each thread/process may do that amount of IOs to
303 the file before giving up the lock. Since lock acquisition is
304 expensive, batching the lock/unlocks will speed up IO.
307 rw=str Type of io pattern. Accepted values are:
309 read Sequential reads
310 write Sequential writes
311 randwrite Random writes
312 randread Random reads
313 rw,readwrite Sequential mixed reads and writes
314 randrw Random mixed reads and writes
316 For the mixed io types, the default is to split them 50/50.
317 For certain types of io the result may still be skewed a bit,
318 since the speed may be different. It is possible to specify
319 a number of IO's to do before getting a new offset, this is
320 one by appending a ':<nr>' to the end of the string given.
321 For a random read, it would look like 'rw=randread:8' for
322 passing in an offset modifier with a value of 8. If the
323 suffix is used with a sequential IO pattern, then the value
324 specified will be added to the generated offset for each IO.
325 For instance, using rw=write:4k will skip 4k for every
326 write. It turns sequential IO into sequential IO with holes.
327 See the 'rw_sequencer' option.
329 rw_sequencer=str If an offset modifier is given by appending a number to
330 the rw=<str> line, then this option controls how that
331 number modifies the IO offset being generated. Accepted
334 sequential Generate sequential offset
335 identical Generate the same offset
337 'sequential' is only useful for random IO, where fio would
338 normally generate a new random offset for every IO. If you
339 append eg 8 to randread, you would get a new random offset for
340 every 8 IO's. The result would be a seek for only every 8
341 IO's, instead of for every IO. Use rw=randread:8 to specify
342 that. As sequential IO is already sequential, setting
343 'sequential' for that would not result in any differences.
344 'identical' behaves in a similar fashion, except it sends
345 the same offset 8 number of times before generating a new
348 kb_base=int The base unit for a kilobyte. The defacto base is 2^10, 1024.
349 Storage manufacturers like to use 10^3 or 1000 as a base
350 ten unit instead, for obvious reasons. Allow values are
351 1024 or 1000, with 1024 being the default.
353 randrepeat=bool For random IO workloads, seed the generator in a predictable
354 way so that results are repeatable across repetitions.
356 use_os_rand=bool Fio can either use the random generator supplied by the OS
357 to generator random offsets, or it can use it's own internal
358 generator (based on Tausworthe). Default is to use the
359 internal generator, which is often of better quality and
362 fallocate=str Whether pre-allocation is performed when laying down files.
365 none Do not pre-allocate space
366 posix Pre-allocate via posix_fallocate()
367 keep Pre-allocate via fallocate() with
368 FALLOC_FL_KEEP_SIZE set
369 0 Backward-compatible alias for 'none'
370 1 Backward-compatible alias for 'posix'
372 May not be available on all supported platforms. 'keep' is only
373 available on Linux.If using ZFS on Solaris this must be set to
374 'none' because ZFS doesn't support it. Default: 'posix'.
376 fadvise_hint=bool By default, fio will use fadvise() to advise the kernel
377 on what IO patterns it is likely to issue. Sometimes you
378 want to test specific IO patterns without telling the
379 kernel about it, in which case you can disable this option.
380 If set, fio will use POSIX_FADV_SEQUENTIAL for sequential
381 IO and POSIX_FADV_RANDOM for random IO.
383 size=int The total size of file io for this job. Fio will run until
384 this many bytes has been transferred, unless runtime is
385 limited by other options (such as 'runtime', for instance).
386 Unless specific nrfiles and filesize options are given,
387 fio will divide this size between the available files
388 specified by the job. If not set, fio will use the full
389 size of the given files or devices. If the the files
390 do not exist, size must be given. It is also possible to
391 give size as a percentage between 1 and 100. If size=20%
392 is given, fio will use 20% of the full size of the given
395 filesize=int Individual file sizes. May be a range, in which case fio
396 will select sizes for files at random within the given range
397 and limited to 'size' in total (if that is given). If not
398 given, each created file is the same size.
401 fill_fs=bool Sets size to something really large and waits for ENOSPC (no
402 space left on device) as the terminating condition. Only makes
403 sense with sequential write. For a read workload, the mount
404 point will be filled first then IO started on the result. This
405 option doesn't make sense if operating on a raw device node,
406 since the size of that is already known by the file system.
407 Additionally, writing beyond end-of-device will not return
411 bs=int The block size used for the io units. Defaults to 4k. Values
412 can be given for both read and writes. If a single int is
413 given, it will apply to both. If a second int is specified
414 after a comma, it will apply to writes only. In other words,
415 the format is either bs=read_and_write or bs=read,write.
416 bs=4k,8k will thus use 4k blocks for reads, and 8k blocks
417 for writes. If you only wish to set the write size, you
418 can do so by passing an empty read size - bs=,8k will set
419 8k for writes and leave the read default value.
422 ba=int At what boundary to align random IO offsets. Defaults to
423 the same as 'blocksize' the minimum blocksize given.
424 Minimum alignment is typically 512b for using direct IO,
425 though it usually depends on the hardware block size. This
426 option is mutually exclusive with using a random map for
427 files, so it will turn off that option.
429 blocksize_range=irange
430 bsrange=irange Instead of giving a single block size, specify a range
431 and fio will mix the issued io block sizes. The issued
432 io unit will always be a multiple of the minimum value
433 given (also see bs_unaligned). Applies to both reads and
434 writes, however a second range can be given after a comma.
437 bssplit=str Sometimes you want even finer grained control of the
438 block sizes issued, not just an even split between them.
439 This option allows you to weight various block sizes,
440 so that you are able to define a specific amount of
441 block sizes issued. The format for this option is:
443 bssplit=blocksize/percentage:blocksize/percentage
445 for as many block sizes as needed. So if you want to define
446 a workload that has 50% 64k blocks, 10% 4k blocks, and
447 40% 32k blocks, you would write:
449 bssplit=4k/10:64k/50:32k/40
451 Ordering does not matter. If the percentage is left blank,
452 fio will fill in the remaining values evenly. So a bssplit
453 option like this one:
455 bssplit=4k/50:1k/:32k/
457 would have 50% 4k ios, and 25% 1k and 32k ios. The percentages
458 always add up to 100, if bssplit is given a range that adds
459 up to more, it will error out.
461 bssplit also supports giving separate splits to reads and
462 writes. The format is identical to what bs= accepts. You
463 have to separate the read and write parts with a comma. So
464 if you want a workload that has 50% 2k reads and 50% 4k reads,
465 while having 90% 4k writes and 10% 8k writes, you would
468 bssplit=2k/50:4k/50,4k/90,8k/10
471 bs_unaligned If this option is given, any byte size value within bsrange
472 may be used as a block range. This typically wont work with
473 direct IO, as that normally requires sector alignment.
475 zero_buffers If this option is given, fio will init the IO buffers to
476 all zeroes. The default is to fill them with random data.
478 refill_buffers If this option is given, fio will refill the IO buffers
479 on every submit. The default is to only fill it at init
480 time and reuse that data. Only makes sense if zero_buffers
481 isn't specified, naturally. If data verification is enabled,
482 refill_buffers is also automatically enabled.
484 scramble_buffers=bool If refill_buffers is too costly and the target is
485 using data deduplication, then setting this option will
486 slightly modify the IO buffer contents to defeat normal
487 de-dupe attempts. This is not enough to defeat more clever
488 block compression attempts, but it will stop naive dedupe of
489 blocks. Default: true.
491 buffer_compress_percentage=int If this is set, then fio will attempt to
492 provide IO buffer content (on WRITEs) that compress to
493 the specified level. Fio does this by providing a mix of
494 random data and zeroes. Note that this is per block size
495 unit, for file/disk wide compression level that matches
496 this setting, you'll also want to set refill_buffers.
498 buffer_compress_chunk=int See buffer_compress_percentage. This
499 setting allows fio to manage how big the ranges of random
500 data and zeroed data is. Without this set, fio will
501 provide buffer_compress_percentage of blocksize random
502 data, followed by the remaining zeroed. With this set
503 to some chunk size smaller than the block size, fio can
504 alternate random and zeroed data throughout the IO
507 nrfiles=int Number of files to use for this job. Defaults to 1.
509 openfiles=int Number of files to keep open at the same time. Defaults to
510 the same as nrfiles, can be set smaller to limit the number
513 file_service_type=str Defines how fio decides which file from a job to
514 service next. The following types are defined:
516 random Just choose a file at random.
518 roundrobin Round robin over open files. This
521 sequential Finish one file before moving on to
522 the next. Multiple files can still be
523 open depending on 'openfiles'.
525 The string can have a number appended, indicating how
526 often to switch to a new file. So if option random:4 is
527 given, fio will switch to a new random file after 4 ios
530 ioengine=str Defines how the job issues io to the file. The following
533 sync Basic read(2) or write(2) io. lseek(2) is
534 used to position the io location.
536 psync Basic pread(2) or pwrite(2) io.
538 vsync Basic readv(2) or writev(2) IO.
540 libaio Linux native asynchronous io. Note that Linux
541 may only support queued behaviour with
542 non-buffered IO (set direct=1 or buffered=0).
543 This engine defines engine specific options.
545 posixaio glibc posix asynchronous io.
547 solarisaio Solaris native asynchronous io.
549 windowsaio Windows native asynchronous io.
551 mmap File is memory mapped and data copied
552 to/from using memcpy(3).
554 splice splice(2) is used to transfer the data and
555 vmsplice(2) to transfer data from user
558 syslet-rw Use the syslet system calls to make
559 regular read/write async.
561 sg SCSI generic sg v3 io. May either be
562 synchronous using the SG_IO ioctl, or if
563 the target is an sg character device
564 we use read(2) and write(2) for asynchronous
567 null Doesn't transfer any data, just pretends
568 to. This is mainly used to exercise fio
569 itself and for debugging/testing purposes.
571 net Transfer over the network to given host:port.
572 Depending on the protocol used, the hostname,
573 port, listen and filename options are used to
574 specify what sort of connection to make, while
575 the protocol option determines which protocol
577 This engine defines engine specific options.
579 netsplice Like net, but uses splice/vmsplice to
580 map data and send/receive.
581 This engine defines engine specific options.
583 cpuio Doesn't transfer any data, but burns CPU
584 cycles according to the cpuload= and
585 cpucycle= options. Setting cpuload=85
586 will cause that job to do nothing but burn
587 85% of the CPU. In case of SMP machines,
588 use numjobs=<no_of_cpu> to get desired CPU
589 usage, as the cpuload only loads a single
590 CPU at the desired rate.
592 guasi The GUASI IO engine is the Generic Userspace
593 Asyncronous Syscall Interface approach
596 http://www.xmailserver.org/guasi-lib.html
598 for more info on GUASI.
600 rdma The RDMA I/O engine supports both RDMA
601 memory semantics (RDMA_WRITE/RDMA_READ) and
602 channel semantics (Send/Recv) for the
603 InfiniBand, RoCE and iWARP protocols.
605 falloc Perform fallocate/punch_hole to files
607 external Prefix to specify loading an external
608 IO engine object file. Append the engine
609 filename, eg ioengine=external:/tmp/foo.o
610 to load ioengine foo.o in /tmp.
612 iodepth=int This defines how many io units to keep in flight against
613 the file. The default is 1 for each file defined in this
614 job, can be overridden with a larger value for higher
615 concurrency. Note that increasing iodepth beyond 1 will not
616 affect synchronous ioengines (except for small degress when
617 verify_async is in use). Even async engines may impose OS
618 restrictions causing the desired depth not to be achieved.
619 This may happen on Linux when using libaio and not setting
620 direct=1, since buffered IO is not async on that OS. Keep an
621 eye on the IO depth distribution in the fio output to verify
622 that the achieved depth is as expected. Default: 1.
624 iodepth_batch_submit=int
625 iodepth_batch=int This defines how many pieces of IO to submit at once.
626 It defaults to 1 which means that we submit each IO
627 as soon as it is available, but can be raised to submit
628 bigger batches of IO at the time.
630 iodepth_batch_complete=int This defines how many pieces of IO to retrieve
631 at once. It defaults to 1 which means that we'll ask
632 for a minimum of 1 IO in the retrieval process from
633 the kernel. The IO retrieval will go on until we
634 hit the limit set by iodepth_low. If this variable is
635 set to 0, then fio will always check for completed
636 events before queuing more IO. This helps reduce
637 IO latency, at the cost of more retrieval system calls.
639 iodepth_low=int The low water mark indicating when to start filling
640 the queue again. Defaults to the same as iodepth, meaning
641 that fio will attempt to keep the queue full at all times.
642 If iodepth is set to eg 16 and iodepth_low is set to 4, then
643 after fio has filled the queue of 16 requests, it will let
644 the depth drain down to 4 before starting to fill it again.
646 direct=bool If value is true, use non-buffered io. This is usually
647 O_DIRECT. Note that ZFS on Solaris doesn't support direct io.
648 On Windows the synchronous ioengines don't support direct io.
650 buffered=bool If value is true, use buffered io. This is the opposite
651 of the 'direct' option. Defaults to true.
653 offset=int Start io at the given offset in the file. The data before
654 the given offset will not be touched. This effectively
655 caps the file size at real_size - offset.
657 offset_increment=int If this is provided, then the real offset becomes
658 the offset + offset_increment * thread_number, where the
659 thread number is a counter that starts at 0 and is incremented
660 for each job. This option is useful if there are several jobs
661 which are intended to operate on a file in parallel in disjoint
662 segments, with even spacing between the starting points.
664 fsync=int If writing to a file, issue a sync of the dirty data
665 for every number of blocks given. For example, if you give
666 32 as a parameter, fio will sync the file for every 32
667 writes issued. If fio is using non-buffered io, we may
668 not sync the file. The exception is the sg io engine, which
669 synchronizes the disk cache anyway.
671 fdatasync=int Like fsync= but uses fdatasync() to only sync data and not
673 In FreeBSD and Windows there is no fdatasync(), this falls back to
676 sync_file_range=str:val Use sync_file_range() for every 'val' number of
677 write operations. Fio will track range of writes that
678 have happened since the last sync_file_range() call. 'str'
679 can currently be one or more of:
681 wait_before SYNC_FILE_RANGE_WAIT_BEFORE
682 write SYNC_FILE_RANGE_WRITE
683 wait_after SYNC_FILE_RANGE_WAIT_AFTER
685 So if you do sync_file_range=wait_before,write:8, fio would
686 use SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE for
687 every 8 writes. Also see the sync_file_range(2) man page.
688 This option is Linux specific.
690 overwrite=bool If true, writes to a file will always overwrite existing
691 data. If the file doesn't already exist, it will be
692 created before the write phase begins. If the file exists
693 and is large enough for the specified write phase, nothing
696 end_fsync=bool If true, fsync file contents when the job exits.
698 fsync_on_close=bool If true, fio will fsync() a dirty file on close.
699 This differs from end_fsync in that it will happen on every
700 file close, not just at the end of the job.
702 rwmixread=int How large a percentage of the mix should be reads.
704 rwmixwrite=int How large a percentage of the mix should be writes. If both
705 rwmixread and rwmixwrite is given and the values do not add
706 up to 100%, the latter of the two will be used to override
707 the first. This may interfere with a given rate setting,
708 if fio is asked to limit reads or writes to a certain rate.
709 If that is the case, then the distribution may be skewed.
711 norandommap Normally fio will cover every block of the file when doing
712 random IO. If this option is given, fio will just get a
713 new random offset without looking at past io history. This
714 means that some blocks may not be read or written, and that
715 some blocks may be read/written more than once. This option
716 is mutually exclusive with verify= if and only if multiple
717 blocksizes (via bsrange=) are used, since fio only tracks
718 complete rewrites of blocks.
720 softrandommap=bool See norandommap. If fio runs with the random block map
721 enabled and it fails to allocate the map, if this option is
722 set it will continue without a random block map. As coverage
723 will not be as complete as with random maps, this option is
726 nice=int Run the job with the given nice value. See man nice(2).
728 prio=int Set the io priority value of this job. Linux limits us to
729 a positive value between 0 and 7, with 0 being the highest.
732 prioclass=int Set the io priority class. See man ionice(1).
734 thinktime=int Stall the job x microseconds after an io has completed before
735 issuing the next. May be used to simulate processing being
736 done by an application. See thinktime_blocks and
740 Only valid if thinktime is set - pretend to spend CPU time
741 doing something with the data received, before falling back
742 to sleeping for the rest of the period specified by
746 Only valid if thinktime is set - control how many blocks
747 to issue, before waiting 'thinktime' usecs. If not set,
748 defaults to 1 which will make fio wait 'thinktime' usecs
751 rate=int Cap the bandwidth used by this job. The number is in bytes/sec,
752 the normal suffix rules apply. You can use rate=500k to limit
753 reads and writes to 500k each, or you can specify read and
754 writes separately. Using rate=1m,500k would limit reads to
755 1MB/sec and writes to 500KB/sec. Capping only reads or
756 writes can be done with rate=,500k or rate=500k,. The former
757 will only limit writes (to 500KB/sec), the latter will only
760 ratemin=int Tell fio to do whatever it can to maintain at least this
761 bandwidth. Failing to meet this requirement, will cause
762 the job to exit. The same format as rate is used for
763 read vs write separation.
765 rate_iops=int Cap the bandwidth to this number of IOPS. Basically the same
766 as rate, just specified independently of bandwidth. If the
767 job is given a block size range instead of a fixed value,
768 the smallest block size is used as the metric. The same format
769 as rate is used for read vs write seperation.
771 rate_iops_min=int If fio doesn't meet this rate of IO, it will cause
772 the job to exit. The same format as rate is used for read vs
775 ratecycle=int Average bandwidth for 'rate' and 'ratemin' over this number
778 cpumask=int Set the CPU affinity of this job. The parameter given is a
779 bitmask of allowed CPU's the job may run on. So if you want
780 the allowed CPUs to be 1 and 5, you would pass the decimal
781 value of (1 << 1 | 1 << 5), or 34. See man
782 sched_setaffinity(2). This may not work on all supported
783 operating systems or kernel versions. This option doesn't
784 work well for a higher CPU count than what you can store in
785 an integer mask, so it can only control cpus 1-32. For
786 boxes with larger CPU counts, use cpus_allowed.
788 cpus_allowed=str Controls the same options as cpumask, but it allows a text
789 setting of the permitted CPUs instead. So to use CPUs 1 and
790 5, you would specify cpus_allowed=1,5. This options also
791 allows a range of CPUs. Say you wanted a binding to CPUs
792 1, 5, and 8-15, you would set cpus_allowed=1,5,8-15.
794 startdelay=time Start this job the specified number of seconds after fio
795 has started. Only useful if the job file contains several
796 jobs, and you want to delay starting some jobs to a certain
799 runtime=time Tell fio to terminate processing after the specified number
800 of seconds. It can be quite hard to determine for how long
801 a specified job will run, so this parameter is handy to
802 cap the total runtime to a given time.
804 time_based If set, fio will run for the duration of the runtime
805 specified even if the file(s) are completely read or
806 written. It will simply loop over the same workload
807 as many times as the runtime allows.
809 ramp_time=time If set, fio will run the specified workload for this amount
810 of time before logging any performance numbers. Useful for
811 letting performance settle before logging results, thus
812 minimizing the runtime required for stable results. Note
813 that the ramp_time is considered lead in time for a job,
814 thus it will increase the total runtime if a special timeout
815 or runtime is specified.
817 invalidate=bool Invalidate the buffer/page cache parts for this file prior
818 to starting io. Defaults to true.
820 sync=bool Use sync io for buffered writes. For the majority of the
821 io engines, this means using O_SYNC.
824 mem=str Fio can use various types of memory as the io unit buffer.
825 The allowed values are:
827 malloc Use memory from malloc(3) as the buffers.
829 shm Use shared memory as the buffers. Allocated
832 shmhuge Same as shm, but use huge pages as backing.
834 mmap Use mmap to allocate buffers. May either be
835 anonymous memory, or can be file backed if
836 a filename is given after the option. The
837 format is mem=mmap:/path/to/file.
839 mmaphuge Use a memory mapped huge file as the buffer
840 backing. Append filename after mmaphuge, ala
841 mem=mmaphuge:/hugetlbfs/file
843 The area allocated is a function of the maximum allowed
844 bs size for the job, multiplied by the io depth given. Note
845 that for shmhuge and mmaphuge to work, the system must have
846 free huge pages allocated. This can normally be checked
847 and set by reading/writing /proc/sys/vm/nr_hugepages on a
848 Linux system. Fio assumes a huge page is 4MB in size. So
849 to calculate the number of huge pages you need for a given
850 job file, add up the io depth of all jobs (normally one unless
851 iodepth= is used) and multiply by the maximum bs set. Then
852 divide that number by the huge page size. You can see the
853 size of the huge pages in /proc/meminfo. If no huge pages
854 are allocated by having a non-zero number in nr_hugepages,
855 using mmaphuge or shmhuge will fail. Also see hugepage-size.
857 mmaphuge also needs to have hugetlbfs mounted and the file
858 location should point there. So if it's mounted in /huge,
859 you would use mem=mmaphuge:/huge/somefile.
861 iomem_align=int This indiciates the memory alignment of the IO memory buffers.
862 Note that the given alignment is applied to the first IO unit
863 buffer, if using iodepth the alignment of the following buffers
864 are given by the bs used. In other words, if using a bs that is
865 a multiple of the page sized in the system, all buffers will
866 be aligned to this value. If using a bs that is not page
867 aligned, the alignment of subsequent IO memory buffers is the
868 sum of the iomem_align and bs used.
871 Defines the size of a huge page. Must at least be equal
872 to the system setting, see /proc/meminfo. Defaults to 4MB.
873 Should probably always be a multiple of megabytes, so using
874 hugepage-size=Xm is the preferred way to set this to avoid
875 setting a non-pow-2 bad value.
877 exitall When one job finishes, terminate the rest. The default is
878 to wait for each job to finish, sometimes that is not the
881 bwavgtime=int Average the calculated bandwidth over the given time. Value
882 is specified in milliseconds.
884 iopsavgtime=int Average the calculated IOPS over the given time. Value
885 is specified in milliseconds.
887 create_serialize=bool If true, serialize the file creating for the jobs.
888 This may be handy to avoid interleaving of data
889 files, which may greatly depend on the filesystem
890 used and even the number of processors in the system.
892 create_fsync=bool fsync the data file after creation. This is the
895 create_on_open=bool Don't pre-setup the files for IO, just create open()
896 when it's time to do IO to that file.
898 create_only=bool If true, fio will only run the setup phase of the job.
899 If files need to be laid out or updated on disk, only
900 that will be done. The actual job contents are not
903 pre_read=bool If this is given, files will be pre-read into memory before
904 starting the given IO operation. This will also clear
905 the 'invalidate' flag, since it is pointless to pre-read
906 and then drop the cache. This will only work for IO engines
907 that are seekable, since they allow you to read the same data
908 multiple times. Thus it will not work on eg network or splice
911 unlink=bool Unlink the job files when done. Not the default, as repeated
912 runs of that job would then waste time recreating the file
915 loops=int Run the specified number of iterations of this job. Used
916 to repeat the same workload a given number of times. Defaults
919 do_verify=bool Run the verify phase after a write phase. Only makes sense if
920 verify is set. Defaults to 1.
922 verify=str If writing to a file, fio can verify the file contents
923 after each iteration of the job. The allowed values are:
925 md5 Use an md5 sum of the data area and store
926 it in the header of each block.
928 crc64 Use an experimental crc64 sum of the data
929 area and store it in the header of each
932 crc32c Use a crc32c sum of the data area and store
933 it in the header of each block.
935 crc32c-intel Use hardware assisted crc32c calcuation
936 provided on SSE4.2 enabled processors. Falls
937 back to regular software crc32c, if not
938 supported by the system.
940 crc32 Use a crc32 sum of the data area and store
941 it in the header of each block.
943 crc16 Use a crc16 sum of the data area and store
944 it in the header of each block.
946 crc7 Use a crc7 sum of the data area and store
947 it in the header of each block.
949 sha512 Use sha512 as the checksum function.
951 sha256 Use sha256 as the checksum function.
953 sha1 Use optimized sha1 as the checksum function.
955 meta Write extra information about each io
956 (timestamp, block number etc.). The block
957 number is verified. See also verify_pattern.
959 null Only pretend to verify. Useful for testing
960 internals with ioengine=null, not for much
963 This option can be used for repeated burn-in tests of a
964 system to make sure that the written data is also
965 correctly read back. If the data direction given is
966 a read or random read, fio will assume that it should
967 verify a previously written file. If the data direction
968 includes any form of write, the verify will be of the
971 verifysort=bool If set, fio will sort written verify blocks when it deems
972 it faster to read them back in a sorted manner. This is
973 often the case when overwriting an existing file, since
974 the blocks are already laid out in the file system. You
975 can ignore this option unless doing huge amounts of really
976 fast IO where the red-black tree sorting CPU time becomes
979 verify_offset=int Swap the verification header with data somewhere else
980 in the block before writing. Its swapped back before
983 verify_interval=int Write the verification header at a finer granularity
984 than the blocksize. It will be written for chunks the
985 size of header_interval. blocksize should divide this
988 verify_pattern=str If set, fio will fill the io buffers with this
989 pattern. Fio defaults to filling with totally random
990 bytes, but sometimes it's interesting to fill with a known
991 pattern for io verification purposes. Depending on the
992 width of the pattern, fio will fill 1/2/3/4 bytes of the
993 buffer at the time(it can be either a decimal or a hex number).
994 The verify_pattern if larger than a 32-bit quantity has to
995 be a hex number that starts with either "0x" or "0X". Use
998 verify_fatal=bool Normally fio will keep checking the entire contents
999 before quitting on a block verification failure. If this
1000 option is set, fio will exit the job on the first observed
1003 verify_dump=bool If set, dump the contents of both the original data
1004 block and the data block we read off disk to files. This
1005 allows later analysis to inspect just what kind of data
1006 corruption occurred. Off by default.
1008 verify_async=int Fio will normally verify IO inline from the submitting
1009 thread. This option takes an integer describing how many
1010 async offload threads to create for IO verification instead,
1011 causing fio to offload the duty of verifying IO contents
1012 to one or more separate threads. If using this offload
1013 option, even sync IO engines can benefit from using an
1014 iodepth setting higher than 1, as it allows them to have
1015 IO in flight while verifies are running.
1017 verify_async_cpus=str Tell fio to set the given CPU affinity on the
1018 async IO verification threads. See cpus_allowed for the
1021 verify_backlog=int Fio will normally verify the written contents of a
1022 job that utilizes verify once that job has completed. In
1023 other words, everything is written then everything is read
1024 back and verified. You may want to verify continually
1025 instead for a variety of reasons. Fio stores the meta data
1026 associated with an IO block in memory, so for large
1027 verify workloads, quite a bit of memory would be used up
1028 holding this meta data. If this option is enabled, fio
1029 will write only N blocks before verifying these blocks.
1031 will verify the previously written blocks before continuing
1034 verify_backlog_batch=int Control how many blocks fio will verify
1035 if verify_backlog is set. If not set, will default to
1036 the value of verify_backlog (meaning the entire queue
1037 is read back and verified). If verify_backlog_batch is
1038 less than verify_backlog then not all blocks will be verified,
1039 if verify_backlog_batch is larger than verify_backlog, some
1040 blocks will be verified more than once.
1043 wait_for_previous Wait for preceeding jobs in the job file to exit, before
1044 starting this one. Can be used to insert serialization
1045 points in the job file. A stone wall also implies starting
1046 a new reporting group.
1048 new_group Start a new reporting group. If this option isn't given,
1049 jobs in a file will be part of the same reporting group
1050 unless separated by a stone wall (or if it's a group
1051 by itself, with the numjobs option).
1053 numjobs=int Create the specified number of clones of this job. May be
1054 used to setup a larger number of threads/processes doing
1055 the same thing. We regard that grouping of jobs as a
1058 group_reporting If 'numjobs' is set, it may be interesting to display
1059 statistics for the group as a whole instead of for each
1060 individual job. This is especially true of 'numjobs' is
1061 large, looking at individual thread/process output quickly
1062 becomes unwieldy. If 'group_reporting' is specified, fio
1063 will show the final report per-group instead of per-job.
1065 thread fio defaults to forking jobs, however if this option is
1066 given, fio will use pthread_create(3) to create threads
1069 zonesize=int Divide a file into zones of the specified size. See zoneskip.
1071 zoneskip=int Skip the specified number of bytes when zonesize data has
1072 been read. The two zone options can be used to only do
1073 io on zones of a file.
1075 write_iolog=str Write the issued io patterns to the specified file. See
1076 read_iolog. Specify a separate file for each job, otherwise
1077 the iologs will be interspersed and the file may be corrupt.
1079 read_iolog=str Open an iolog with the specified file name and replay the
1080 io patterns it contains. This can be used to store a
1081 workload and replay it sometime later. The iolog given
1082 may also be a blktrace binary file, which allows fio
1083 to replay a workload captured by blktrace. See blktrace
1084 for how to capture such logging data. For blktrace replay,
1085 the file needs to be turned into a blkparse binary data
1086 file first (blkparse <device> -o /dev/null -d file_for_fio.bin).
1088 replay_no_stall=int When replaying I/O with read_iolog the default behavior
1089 is to attempt to respect the time stamps within the log and
1090 replay them with the appropriate delay between IOPS. By
1091 setting this variable fio will not respect the timestamps and
1092 attempt to replay them as fast as possible while still
1093 respecting ordering. The result is the same I/O pattern to a
1094 given device, but different timings.
1096 replay_redirect=str While replaying I/O patterns using read_iolog the
1097 default behavior is to replay the IOPS onto the major/minor
1098 device that each IOP was recorded from. This is sometimes
1099 undesireable because on a different machine those major/minor
1100 numbers can map to a different device. Changing hardware on
1101 the same system can also result in a different major/minor
1102 mapping. Replay_redirect causes all IOPS to be replayed onto
1103 the single specified device regardless of the device it was
1104 recorded from. i.e. replay_redirect=/dev/sdc would cause all
1105 IO in the blktrace to be replayed onto /dev/sdc. This means
1106 multiple devices will be replayed onto a single, if the trace
1107 contains multiple devices. If you want multiple devices to be
1108 replayed concurrently to multiple redirected devices you must
1109 blkparse your trace into separate traces and replay them with
1110 independent fio invocations. Unfortuantely this also breaks
1111 the strict time ordering between multiple device accesses.
1113 write_bw_log=str If given, write a bandwidth log of the jobs in this job
1114 file. Can be used to store data of the bandwidth of the
1115 jobs in their lifetime. The included fio_generate_plots
1116 script uses gnuplot to turn these text files into nice
1117 graphs. See write_lat_log for behaviour of given
1118 filename. For this option, the suffix is _bw.log.
1120 write_lat_log=str Same as write_bw_log, except that this option stores io
1121 submission, completion, and total latencies instead. If no
1122 filename is given with this option, the default filename of
1123 "jobname_type.log" is used. Even if the filename is given,
1124 fio will still append the type of log. So if one specifies
1128 The actual log names will be foo_slat.log, foo_slat.log,
1129 and foo_lat.log. This helps fio_generate_plot fine the logs
1132 write_bw_log=str If given, write an IOPS log of the jobs in this job
1133 file. See write_bw_log.
1135 write_iops_log=str Same as write_bw_log, but writes IOPS. If no filename is
1136 given with this option, the default filename of
1137 "jobname_type.log" is used. Even if the filename is given,
1138 fio will still append the type of log.
1140 log_avg_msec=int By default, fio will log an entry in the iops, latency,
1141 or bw log for every IO that completes. When writing to the
1142 disk log, that can quickly grow to a very large size. Setting
1143 this option makes fio average the each log entry over the
1144 specified period of time, reducing the resolution of the log.
1147 lockmem=int Pin down the specified amount of memory with mlock(2). Can
1148 potentially be used instead of removing memory or booting
1149 with less memory to simulate a smaller amount of memory.
1151 exec_prerun=str Before running this job, issue the command specified
1154 exec_postrun=str After the job completes, issue the command specified
1157 ioscheduler=str Attempt to switch the device hosting the file to the specified
1158 io scheduler before running.
1160 cpuload=int If the job is a CPU cycle eater, attempt to use the specified
1161 percentage of CPU cycles.
1163 cpuchunks=int If the job is a CPU cycle eater, split the load into
1164 cycles of the given time. In microseconds.
1166 disk_util=bool Generate disk utilization statistics, if the platform
1167 supports it. Defaults to on.
1169 disable_lat=bool Disable measurements of total latency numbers. Useful
1170 only for cutting back the number of calls to gettimeofday,
1171 as that does impact performance at really high IOPS rates.
1172 Note that to really get rid of a large amount of these
1173 calls, this option must be used with disable_slat and
1176 disable_clat=bool Disable measurements of completion latency numbers. See
1179 disable_slat=bool Disable measurements of submission latency numbers. See
1182 disable_bw=bool Disable measurements of throughput/bandwidth numbers. See
1185 clat_percentiles=bool Enable the reporting of percentiles of
1186 completion latencies.
1188 percentile_list=float_list Overwrite the default list of percentiles
1189 for completion latencies. Each number is a floating
1190 number in the range (0,100], and the maximum length of
1191 the list is 20. Use ':' to separate the numbers, and
1192 list the numbers in ascending order. For example,
1193 --percentile_list=99.5:99.9 will cause fio to report
1194 the values of completion latency below which 99.5% and
1195 99.9% of the observed latencies fell, respectively.
1197 gtod_reduce=bool Enable all of the gettimeofday() reducing options
1198 (disable_clat, disable_slat, disable_bw) plus reduce
1199 precision of the timeout somewhat to really shrink
1200 the gettimeofday() call count. With this option enabled,
1201 we only do about 0.4% of the gtod() calls we would have
1202 done if all time keeping was enabled.
1204 gtod_cpu=int Sometimes it's cheaper to dedicate a single thread of
1205 execution to just getting the current time. Fio (and
1206 databases, for instance) are very intensive on gettimeofday()
1207 calls. With this option, you can set one CPU aside for
1208 doing nothing but logging current time to a shared memory
1209 location. Then the other threads/processes that run IO
1210 workloads need only copy that segment, instead of entering
1211 the kernel with a gettimeofday() call. The CPU set aside
1212 for doing these time calls will be excluded from other
1213 uses. Fio will manually clear it from the CPU mask of other
1216 continue_on_error=str Normally fio will exit the job on the first observed
1217 failure. If this option is set, fio will continue the job when
1218 there is a 'non-fatal error' (EIO or EILSEQ) until the runtime
1219 is exceeded or the I/O size specified is completed. If this
1220 option is used, there are two more stats that are appended,
1221 the total error count and the first error. The error field
1222 given in the stats is the first error that was hit during the
1225 The allowed values are:
1227 none Exit on any IO or verify errors.
1229 read Continue on read errors, exit on all others.
1231 write Continue on write errors, exit on all others.
1233 io Continue on any IO error, exit on all others.
1235 verify Continue on verify errors, exit on all others.
1237 all Continue on all errors.
1239 0 Backward-compatible alias for 'none'.
1241 1 Backward-compatible alias for 'all'.
1243 cgroup=str Add job to this control group. If it doesn't exist, it will
1244 be created. The system must have a mounted cgroup blkio
1245 mount point for this to work. If your system doesn't have it
1246 mounted, you can do so with:
1248 # mount -t cgroup -o blkio none /cgroup
1250 cgroup_weight=int Set the weight of the cgroup to this value. See
1251 the documentation that comes with the kernel, allowed values
1252 are in the range of 100..1000.
1254 cgroup_nodelete=bool Normally fio will delete the cgroups it has created after
1255 the job completion. To override this behavior and to leave
1256 cgroups around after the job completion, set cgroup_nodelete=1.
1257 This can be useful if one wants to inspect various cgroup
1258 files after job completion. Default: false
1260 uid=int Instead of running as the invoking user, set the user ID to
1261 this value before the thread/process does any work.
1263 gid=int Set group ID, see uid.
1265 flow_id=int The ID of the flow. If not specified, it defaults to being a
1266 global flow. See flow.
1268 flow=int Weight in token-based flow control. If this value is used, then
1269 there is a 'flow counter' which is used to regulate the
1270 proportion of activity between two or more jobs. fio attempts
1271 to keep this flow counter near zero. The 'flow' parameter
1272 stands for how much should be added or subtracted to the flow
1273 counter on each iteration of the main I/O loop. That is, if
1274 one job has flow=8 and another job has flow=-1, then there
1275 will be a roughly 1:8 ratio in how much one runs vs the other.
1277 flow_watermark=int The maximum value that the absolute value of the flow
1278 counter is allowed to reach before the job must wait for a
1279 lower value of the counter.
1281 flow_sleep=int The period of time, in microseconds, to wait after the flow
1282 watermark has been exceeded before retrying operations
1284 In addition, there are some parameters which are only valid when a specific
1285 ioengine is in use. These are used identically to normal parameters, with the
1286 caveat that when used on the command line, they must come after the ioengine
1287 that defines them is selected.
1289 [libaio] userspace_reap Normally, with the libaio engine in use, fio will use
1290 the io_getevents system call to reap newly returned events.
1291 With this flag turned on, the AIO ring will be read directly
1292 from user-space to reap events. The reaping mode is only
1293 enabled when polling for a minimum of 0 events (eg when
1294 iodepth_batch_complete=0).
1296 [netsplice] hostname=str
1297 [net] hostname=str The host name or IP address to use for TCP or UDP based IO.
1298 If the job is a TCP listener or UDP reader, the hostname is not
1299 used and must be omitted.
1301 [netsplice] port=int
1302 [net] port=int The TCP or UDP port to bind to or connect to.
1304 [netsplice] protocol=str
1305 [netsplice] proto=str
1307 [net] proto=str The network protocol to use. Accepted values are:
1309 tcp Transmission control protocol
1310 udp Unreliable datagram protocol
1311 unix UNIX domain socket
1313 When the protocol is TCP or UDP, the port must also be given,
1314 as well as the hostname if the job is a TCP listener or UDP
1315 reader. For unix sockets, the normal filename option should be
1316 used and the port is invalid.
1318 [net] listen For TCP network connections, tell fio to listen for incoming
1319 connections rather than initiating an outgoing connection. The
1320 hostname must be omitted if this option is used.
1323 6.0 Interpreting the output
1324 ---------------------------
1326 fio spits out a lot of output. While running, fio will display the
1327 status of the jobs created. An example of that would be:
1329 Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
1331 The characters inside the square brackets denote the current status of
1332 each thread. The possible values (in typical life cycle order) are:
1336 P Thread setup, but not started.
1338 I Thread initialized, waiting.
1339 p Thread running pre-reading file(s).
1340 R Running, doing sequential reads.
1341 r Running, doing random reads.
1342 W Running, doing sequential writes.
1343 w Running, doing random writes.
1344 M Running, doing mixed sequential reads/writes.
1345 m Running, doing mixed random reads/writes.
1346 F Running, currently waiting for fsync()
1347 V Running, doing verification of written data.
1348 E Thread exited, not reaped by main thread yet.
1350 X Thread reaped, exited with an error.
1351 K Thread reaped, exited due to signal.
1353 The other values are fairly self explanatory - number of threads
1354 currently running and doing io, rate of io since last check (read speed
1355 listed first, then write speed), and the estimated completion percentage
1356 and time for the running group. It's impossible to estimate runtime of
1357 the following groups (if any). Note that the string is displayed in order,
1358 so it's possible to tell which of the jobs are currently doing what. The
1359 first character is the first job defined in the job file, and so forth.
1361 When fio is done (or interrupted by ctrl-c), it will show the data for
1362 each thread, group of threads, and disks in that order. For each data
1363 direction, the output looks like:
1365 Client1 (g=0): err= 0:
1366 write: io= 32MB, bw= 666KB/s, iops=89 , runt= 50320msec
1367 slat (msec): min= 0, max= 136, avg= 0.03, stdev= 1.92
1368 clat (msec): min= 0, max= 631, avg=48.50, stdev=86.82
1369 bw (KB/s) : min= 0, max= 1196, per=51.00%, avg=664.02, stdev=681.68
1370 cpu : usr=1.49%, sys=0.25%, ctx=7969, majf=0, minf=17
1371 IO depths : 1=0.1%, 2=0.3%, 4=0.5%, 8=99.0%, 16=0.0%, 32=0.0%, >32=0.0%
1372 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
1373 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
1374 issued r/w: total=0/32768, short=0/0
1375 lat (msec): 2=1.6%, 4=0.0%, 10=3.2%, 20=12.8%, 50=38.4%, 100=24.8%,
1376 lat (msec): 250=15.2%, 500=0.0%, 750=0.0%, 1000=0.0%, >=2048=0.0%
1378 The client number is printed, along with the group id and error of that
1379 thread. Below is the io statistics, here for writes. In the order listed,
1382 io= Number of megabytes io performed
1383 bw= Average bandwidth rate
1384 iops= Average IOs performed per second
1385 runt= The runtime of that thread
1386 slat= Submission latency (avg being the average, stdev being the
1387 standard deviation). This is the time it took to submit
1388 the io. For sync io, the slat is really the completion
1389 latency, since queue/complete is one operation there. This
1390 value can be in milliseconds or microseconds, fio will choose
1391 the most appropriate base and print that. In the example
1392 above, milliseconds is the best scale. Note: in --minimal mode
1393 latencies are always expressed in microseconds.
1394 clat= Completion latency. Same names as slat, this denotes the
1395 time from submission to completion of the io pieces. For
1396 sync io, clat will usually be equal (or very close) to 0,
1397 as the time from submit to complete is basically just
1398 CPU time (io has already been done, see slat explanation).
1399 bw= Bandwidth. Same names as the xlat stats, but also includes
1400 an approximate percentage of total aggregate bandwidth
1401 this thread received in this group. This last value is
1402 only really useful if the threads in this group are on the
1403 same disk, since they are then competing for disk access.
1404 cpu= CPU usage. User and system time, along with the number
1405 of context switches this thread went through, usage of
1406 system and user time, and finally the number of major
1407 and minor page faults.
1408 IO depths= The distribution of io depths over the job life time. The
1409 numbers are divided into powers of 2, so for example the
1410 16= entries includes depths up to that value but higher
1411 than the previous entry. In other words, it covers the
1412 range from 16 to 31.
1413 IO submit= How many pieces of IO were submitting in a single submit
1414 call. Each entry denotes that amount and below, until
1415 the previous entry - eg, 8=100% mean that we submitted
1416 anywhere in between 5-8 ios per submit call.
1417 IO complete= Like the above submit number, but for completions instead.
1418 IO issued= The number of read/write requests issued, and how many
1420 IO latencies= The distribution of IO completion latencies. This is the
1421 time from when IO leaves fio and when it gets completed.
1422 The numbers follow the same pattern as the IO depths,
1423 meaning that 2=1.6% means that 1.6% of the IO completed
1424 within 2 msecs, 20=12.8% means that 12.8% of the IO
1425 took more than 10 msecs, but less than (or equal to) 20 msecs.
1427 After each client has been listed, the group statistics are printed. They
1428 will look like this:
1430 Run status group 0 (all jobs):
1431 READ: io=64MB, aggrb=22178, minb=11355, maxb=11814, mint=2840msec, maxt=2955msec
1432 WRITE: io=64MB, aggrb=1302, minb=666, maxb=669, mint=50093msec, maxt=50320msec
1434 For each data direction, it prints:
1436 io= Number of megabytes io performed.
1437 aggrb= Aggregate bandwidth of threads in this group.
1438 minb= The minimum average bandwidth a thread saw.
1439 maxb= The maximum average bandwidth a thread saw.
1440 mint= The smallest runtime of the threads in that group.
1441 maxt= The longest runtime of the threads in that group.
1443 And finally, the disk statistics are printed. They will look like this:
1445 Disk stats (read/write):
1446 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
1448 Each value is printed for both reads and writes, with reads first. The
1451 ios= Number of ios performed by all groups.
1452 merge= Number of merges io the io scheduler.
1453 ticks= Number of ticks we kept the disk busy.
1454 io_queue= Total time spent in the disk queue.
1455 util= The disk utilization. A value of 100% means we kept the disk
1456 busy constantly, 50% would be a disk idling half of the time.
1458 It is also possible to get fio to dump the current output while it is
1459 running, without terminating the job. To do that, send fio the USR1 signal.
1465 For scripted usage where you typically want to generate tables or graphs
1466 of the results, fio can output the results in a semicolon separated format.
1467 The format is one long line of values, such as:
1469 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%
1470 A description of this job goes here.
1472 The job description (if provided) follows on a second line.
1474 To enable terse output, use the --minimal command line option. The first
1475 value is the version of the terse output format. If the output has to
1476 be changed for some reason, this number will be incremented by 1 to
1477 signify that change.
1479 Split up, the format is as follows:
1481 terse version, fio version, jobname, groupid, error
1483 Total IO (KB), bandwidth (KB/sec), IOPS, runtime (msec)
1484 Submission latency: min, max, mean, deviation (usec)
1485 Completion latency: min, max, mean, deviation (usec)
1486 Completion latency percentiles: 20 fields (see below)
1487 Total latency: min, max, mean, deviation (usec)
1488 Bw (KB/s): min, max, aggregate percentage of total, mean, deviation
1490 Total IO (KB), bandwidth (KB/sec), IOPS, runtime (msec)
1491 Submission latency: min, max, mean, deviation (usec)
1492 Completion latency: min, max, mean, deviation (usec)
1493 Completion latency percentiles: 20 fields (see below)
1494 Total latency: min, max, mean, deviation (usec)
1495 Bw (KB/s): min, max, aggregate percentage of total, mean, deviation
1496 CPU usage: user, system, context switches, major faults, minor faults
1497 IO depths: <=1, 2, 4, 8, 16, 32, >=64
1498 IO latencies microseconds: <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
1499 IO latencies milliseconds: <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
1500 Disk utilization: Disk name, Read ios, write ios,
1501 Read merges, write merges,
1502 Read ticks, write ticks,
1503 Time spent in queue, disk utilization percentage
1504 Additional Info (dependant on continue_on_error, default off): total # errors, first error code
1506 Additional Info (dependant on description being set): Text description
1508 Completion latency percentiles can be a grouping of up to 20 sets, so
1509 for the terse output fio writes all of them. Each field will look like this:
1513 which is the Xth percentile, and the usec latency associated with it.
1515 For disk utilization, all disks used by fio are shown. So for each disk
1516 there will be a disk utilization section.
1519 8.0 Trace file format
1520 ---------------------
1521 There are two trace file format that you can encounter. The older (v1) format
1522 is unsupported since version 1.20-rc3 (March 2008). It will still be described
1523 below in case that you get an old trace and want to understand it.
1525 In any case the trace is a simple text file with a single action per line.
1528 8.1 Trace file format v1
1529 ------------------------
1530 Each line represents a single io action in the following format:
1534 where rw=0/1 for read/write, and the offset and length entries being in bytes.
1536 This format is not supported in Fio versions => 1.20-rc3.
1539 8.2 Trace file format v2
1540 ------------------------
1541 The second version of the trace file format was added in Fio version 1.17.
1542 It allows to access more then one file per trace and has a bigger set of
1543 possible file actions.
1545 The first line of the trace file has to be:
1549 Following this can be lines in two different formats, which are described below.
1551 The file management format:
1555 The filename is given as an absolute path. The action can be one of these:
1557 add Add the given filename to the trace
1558 open Open the file with the given filename. The filename has to have
1559 been added with the add action before.
1560 close Close the file with the given filename. The file has to have been
1564 The file io action format:
1566 filename action offset length
1568 The filename is given as an absolute path, and has to have been added and opened
1569 before it can be used with this format. The offset and length are given in
1570 bytes. The action can be one of these:
1572 wait Wait for 'offset' microseconds. Everything below 100 is discarded.
1573 read Read 'length' bytes beginning from 'offset'
1574 write Write 'length' bytes beginning from 'offset'
1575 sync fsync() the file
1576 datasync fdatasync() the file
1577 trim trim the given file from the given 'offset' for 'length' bytes
projects / fio.git / blob