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
248 With the above in mind, here follows the complete list of fio job
251 name=str ASCII name of the job. This may be used to override the
252 name printed by fio for this job. Otherwise the job
253 name is used. On the command line this parameter has the
254 special purpose of also signaling the start of a new
257 description=str Text description of the job. Doesn't do anything except
258 dump this text description when this job is run. It's
261 directory=str Prefix filenames with this directory. Used to place files
262 in a different location than "./".
264 filename=str Fio normally makes up a filename based on the job name,
265 thread number, and file number. If you want to share
266 files between threads in a job or several jobs, specify
267 a filename for each of them to override the default. If
268 the ioengine used is 'net', the filename is the host, port,
269 and protocol to use in the format of =host/port/protocol.
270 See ioengine=net for more. If the ioengine is file based, you
271 can specify a number of files by separating the names with a
272 ':' colon. So if you wanted a job to open /dev/sda and /dev/sdb
273 as the two working files, you would use
274 filename=/dev/sda:/dev/sdb. On Windows, disk devices are accessed
275 as \\.\PhysicalDrive0 for the first device, \\.\PhysicalDrive1
276 for the second etc. If the wanted filename does need to
277 include a colon, then escape that with a '\' character.
278 For instance, if the filename is "/dev/dsk/foo@3,0:c",
279 then you would use filename="/dev/dsk/foo@3,0\:c".
280 '-' is a reserved name, meaning stdin or stdout. Which of the
281 two depends on the read/write direction set.
283 opendir=str Tell fio to recursively add any file it can find in this
284 directory and down the file system tree.
286 lockfile=str Fio defaults to not locking any files before it does
287 IO to them. If a file or file descriptor is shared, fio
288 can serialize IO to that file to make the end result
289 consistent. This is usual for emulating real workloads that
290 share files. The lock modes are:
292 none No locking. The default.
293 exclusive Only one thread/process may do IO,
294 excluding all others.
295 readwrite Read-write locking on the file. Many
296 readers may access the file at the
297 same time, but writes get exclusive
300 The option may be post-fixed with a lock batch number. If
301 set, then each thread/process may do that amount of IOs to
302 the file before giving up the lock. Since lock acquisition is
303 expensive, batching the lock/unlocks will speed up IO.
306 rw=str Type of io pattern. Accepted values are:
308 read Sequential reads
309 write Sequential writes
310 randwrite Random writes
311 randread Random reads
312 rw Sequential mixed reads and writes
313 randrw Random mixed reads and writes
315 For the mixed io types, the default is to split them 50/50.
316 For certain types of io the result may still be skewed a bit,
317 since the speed may be different. It is possible to specify
318 a number of IO's to do before getting a new offset, this is
319 one by appending a ':<nr>' to the end of the string given.
320 For a random read, it would look like 'rw=randread:8' for
321 passing in an offset modifier with a value of 8. See the
322 'rw_sequencer' option.
324 rw_sequencer=str If an offset modifier is given by appending a number to
325 the rw=<str> line, then this option controls how that
326 number modifies the IO offset being generated. Accepted
329 sequential Generate sequential offset
330 identical Generate the same offset
332 'sequential' is only useful for random IO, where fio would
333 normally generate a new random offset for every IO. If you
334 append eg 8 to randread, you would get a new random offset for
335 every 8 IO's. The result would be a seek for only every 8
336 IO's, instead of for every IO. Use rw=randread:8 to specify
337 that. As sequential IO is already sequential, setting
338 'sequential' for that would not result in any differences.
339 'identical' behaves in a similar fashion, except it sends
340 the same offset 8 number of times before generating a new
343 kb_base=int The base unit for a kilobyte. The defacto base is 2^10, 1024.
344 Storage manufacturers like to use 10^3 or 1000 as a base
345 ten unit instead, for obvious reasons. Allow values are
346 1024 or 1000, with 1024 being the default.
348 randrepeat=bool For random IO workloads, seed the generator in a predictable
349 way so that results are repeatable across repetitions.
351 fallocate=bool By default, fio will use fallocate() to advise the system
352 of the size of the file we are going to write. This can be
353 turned off with fallocate=0. May not be available on all
354 supported platforms. If using ZFS on Solaris this must be
355 set to 0 because ZFS doesn't support it.
357 fadvise_hint=bool By default, fio will use fadvise() to advise the kernel
358 on what IO patterns it is likely to issue. Sometimes you
359 want to test specific IO patterns without telling the
360 kernel about it, in which case you can disable this option.
361 If set, fio will use POSIX_FADV_SEQUENTIAL for sequential
362 IO and POSIX_FADV_RANDOM for random IO.
364 size=int The total size of file io for this job. Fio will run until
365 this many bytes has been transferred, unless runtime is
366 limited by other options (such as 'runtime', for instance).
367 Unless specific nrfiles and filesize options are given,
368 fio will divide this size between the available files
369 specified by the job. If not set, fio will use the full
370 size of the given files or devices. If the the files
371 do not exist, size must be given.
373 filesize=int Individual file sizes. May be a range, in which case fio
374 will select sizes for files at random within the given range
375 and limited to 'size' in total (if that is given). If not
376 given, each created file is the same size.
378 fill_device=bool Sets size to something really large and waits for ENOSPC (no
379 space left on device) as the terminating condition. Only makes
380 sense with sequential write. For a read workload, the mount
381 point will be filled first then IO started on the result. This
382 option doesn't make sense if operating on a raw device node,
383 since the size of that is already known by the file system.
384 Additionally, writing beyond end-of-device will not return
388 bs=int The block size used for the io units. Defaults to 4k. Values
389 can be given for both read and writes. If a single int is
390 given, it will apply to both. If a second int is specified
391 after a comma, it will apply to writes only. In other words,
392 the format is either bs=read_and_write or bs=read,write.
393 bs=4k,8k will thus use 4k blocks for reads, and 8k blocks
394 for writes. If you only wish to set the write size, you
395 can do so by passing an empty read size - bs=,8k will set
396 8k for writes and leave the read default value.
399 ba=int At what boundary to align random IO offsets. Defaults to
400 the same as 'blocksize' the minimum blocksize given.
401 Minimum alignment is typically 512b for using direct IO,
402 though it usually depends on the hardware block size. This
403 option is mutually exclusive with using a random map for
404 files, so it will turn off that option.
406 blocksize_range=irange
407 bsrange=irange Instead of giving a single block size, specify a range
408 and fio will mix the issued io block sizes. The issued
409 io unit will always be a multiple of the minimum value
410 given (also see bs_unaligned). Applies to both reads and
411 writes, however a second range can be given after a comma.
414 bssplit=str Sometimes you want even finer grained control of the
415 block sizes issued, not just an even split between them.
416 This option allows you to weight various block sizes,
417 so that you are able to define a specific amount of
418 block sizes issued. The format for this option is:
420 bssplit=blocksize/percentage:blocksize/percentage
422 for as many block sizes as needed. So if you want to define
423 a workload that has 50% 64k blocks, 10% 4k blocks, and
424 40% 32k blocks, you would write:
426 bssplit=4k/10:64k/50:32k/40
428 Ordering does not matter. If the percentage is left blank,
429 fio will fill in the remaining values evenly. So a bssplit
430 option like this one:
432 bssplit=4k/50:1k/:32k/
434 would have 50% 4k ios, and 25% 1k and 32k ios. The percentages
435 always add up to 100, if bssplit is given a range that adds
436 up to more, it will error out.
438 bssplit also supports giving separate splits to reads and
439 writes. The format is identical to what bs= accepts. You
440 have to separate the read and write parts with a comma. So
441 if you want a workload that has 50% 2k reads and 50% 4k reads,
442 while having 90% 4k writes and 10% 8k writes, you would
445 bssplit=2k/50:4k/50,4k/90,8k/10
448 bs_unaligned If this option is given, any byte size value within bsrange
449 may be used as a block range. This typically wont work with
450 direct IO, as that normally requires sector alignment.
452 zero_buffers If this option is given, fio will init the IO buffers to
453 all zeroes. The default is to fill them with random data.
455 refill_buffers If this option is given, fio will refill the IO buffers
456 on every submit. The default is to only fill it at init
457 time and reuse that data. Only makes sense if zero_buffers
458 isn't specified, naturally. If data verification is enabled,
459 refill_buffers is also automatically enabled.
461 nrfiles=int Number of files to use for this job. Defaults to 1.
463 openfiles=int Number of files to keep open at the same time. Defaults to
464 the same as nrfiles, can be set smaller to limit the number
467 file_service_type=str Defines how fio decides which file from a job to
468 service next. The following types are defined:
470 random Just choose a file at random.
472 roundrobin Round robin over open files. This
475 sequential Finish one file before moving on to
476 the next. Multiple files can still be
477 open depending on 'openfiles'.
479 The string can have a number appended, indicating how
480 often to switch to a new file. So if option random:4 is
481 given, fio will switch to a new random file after 4 ios
484 ioengine=str Defines how the job issues io to the file. The following
487 sync Basic read(2) or write(2) io. lseek(2) is
488 used to position the io location.
490 psync Basic pread(2) or pwrite(2) io.
492 vsync Basic readv(2) or writev(2) IO.
494 libaio Linux native asynchronous io. Note that Linux
495 may only support queued behaviour with
496 non-buffered IO (set direct=1 or buffered=0).
498 posixaio glibc posix asynchronous io.
500 solarisaio Solaris native asynchronous io.
502 windowsaio Windows native asynchronous io.
504 mmap File is memory mapped and data copied
505 to/from using memcpy(3).
507 splice splice(2) is used to transfer the data and
508 vmsplice(2) to transfer data from user
511 syslet-rw Use the syslet system calls to make
512 regular read/write async.
514 sg SCSI generic sg v3 io. May either be
515 synchronous using the SG_IO ioctl, or if
516 the target is an sg character device
517 we use read(2) and write(2) for asynchronous
520 null Doesn't transfer any data, just pretends
521 to. This is mainly used to exercise fio
522 itself and for debugging/testing purposes.
524 net Transfer over the network to given host:port.
525 'filename' must be set appropriately to
526 filename=host/port/protocol regardless of send
527 or receive, if the latter only the port
528 argument is used. 'host' may be an IP address
529 or hostname, port is the port number to be used,
530 and protocol may be 'udp' or 'tcp'. If no
531 protocol is given, TCP is used.
533 netsplice Like net, but uses splice/vmsplice to
534 map data and send/receive.
536 cpuio Doesn't transfer any data, but burns CPU
537 cycles according to the cpuload= and
538 cpucycle= options. Setting cpuload=85
539 will cause that job to do nothing but burn
540 85% of the CPU. In case of SMP machines,
541 use numjobs=<no_of_cpu> to get desired CPU
542 usage, as the cpuload only loads a single
543 CPU at the desired rate.
545 guasi The GUASI IO engine is the Generic Userspace
546 Asyncronous Syscall Interface approach
549 http://www.xmailserver.org/guasi-lib.html
551 for more info on GUASI.
553 external Prefix to specify loading an external
554 IO engine object file. Append the engine
555 filename, eg ioengine=external:/tmp/foo.o
556 to load ioengine foo.o in /tmp.
558 iodepth=int This defines how many io units to keep in flight against
559 the file. The default is 1 for each file defined in this
560 job, can be overridden with a larger value for higher
561 concurrency. Note that increasing iodepth beyond 1 will not
562 affect synchronous ioengines (except for small degress when
563 verify_async is in use). Even async engines may impose OS
564 restrictions causing the desired depth not to be achieved.
565 This may happen on Linux when using libaio and not setting
566 direct=1, since buffered IO is not async on that OS. Keep an
567 eye on the IO depth distribution in the fio output to verify
568 that the achieved depth is as expected. Default: 1.
570 iodepth_batch_submit=int
571 iodepth_batch=int This defines how many pieces of IO to submit at once.
572 It defaults to 1 which means that we submit each IO
573 as soon as it is available, but can be raised to submit
574 bigger batches of IO at the time.
576 iodepth_batch_complete=int This defines how many pieces of IO to retrieve
577 at once. It defaults to 1 which means that we'll ask
578 for a minimum of 1 IO in the retrieval process from
579 the kernel. The IO retrieval will go on until we
580 hit the limit set by iodepth_low. If this variable is
581 set to 0, then fio will always check for completed
582 events before queuing more IO. This helps reduce
583 IO latency, at the cost of more retrieval system calls.
585 iodepth_low=int The low water mark indicating when to start filling
586 the queue again. Defaults to the same as iodepth, meaning
587 that fio will attempt to keep the queue full at all times.
588 If iodepth is set to eg 16 and iodepth_low is set to 4, then
589 after fio has filled the queue of 16 requests, it will let
590 the depth drain down to 4 before starting to fill it again.
592 direct=bool If value is true, use non-buffered io. This is usually
593 O_DIRECT. Note that ZFS on Solaris doesn't support direct io.
595 buffered=bool If value is true, use buffered io. This is the opposite
596 of the 'direct' option. Defaults to true.
598 offset=int Start io at the given offset in the file. The data before
599 the given offset will not be touched. This effectively
600 caps the file size at real_size - offset.
602 fsync=int If writing to a file, issue a sync of the dirty data
603 for every number of blocks given. For example, if you give
604 32 as a parameter, fio will sync the file for every 32
605 writes issued. If fio is using non-buffered io, we may
606 not sync the file. The exception is the sg io engine, which
607 synchronizes the disk cache anyway.
609 fdatasync=int Like fsync= but uses fdatasync() to only sync data and not
611 In FreeBSD there is no fdatasync(), this falls back to
614 sync_file_range=str:val Use sync_file_range() for every 'val' number of
615 write operations. Fio will track range of writes that
616 have happened since the last sync_file_range() call. 'str'
617 can currently be one or more of:
619 wait_before SYNC_FILE_RANGE_WAIT_BEFORE
620 write SYNC_FILE_RANGE_WRITE
621 wait_after SYNC_FILE_RANGE_WAIT_AFTER
623 So if you do sync_file_range=wait_before,write:8, fio would
624 use SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE for
625 every 8 writes. Also see the sync_file_range(2) man page.
626 This option is Linux specific.
628 overwrite=bool If true, writes to a file will always overwrite existing
629 data. If the file doesn't already exist, it will be
630 created before the write phase begins. If the file exists
631 and is large enough for the specified write phase, nothing
634 end_fsync=bool If true, fsync file contents when the job exits.
636 fsync_on_close=bool If true, fio will fsync() a dirty file on close.
637 This differs from end_fsync in that it will happen on every
638 file close, not just at the end of the job.
640 rwmixread=int How large a percentage of the mix should be reads.
642 rwmixwrite=int How large a percentage of the mix should be writes. If both
643 rwmixread and rwmixwrite is given and the values do not add
644 up to 100%, the latter of the two will be used to override
645 the first. This may interfere with a given rate setting,
646 if fio is asked to limit reads or writes to a certain rate.
647 If that is the case, then the distribution may be skewed.
649 norandommap Normally fio will cover every block of the file when doing
650 random IO. If this option is given, fio will just get a
651 new random offset without looking at past io history. This
652 means that some blocks may not be read or written, and that
653 some blocks may be read/written more than once. This option
654 is mutually exclusive with verify= if and only if multiple
655 blocksizes (via bsrange=) are used, since fio only tracks
656 complete rewrites of blocks.
658 softrandommap See norandommap. If fio runs with the random block map enabled
659 and it fails to allocate the map, if this option is set it
660 will continue without a random block map. As coverage will
661 not be as complete as with random maps, this option is
664 nice=int Run the job with the given nice value. See man nice(2).
666 prio=int Set the io priority value of this job. Linux limits us to
667 a positive value between 0 and 7, with 0 being the highest.
670 prioclass=int Set the io priority class. See man ionice(1).
672 thinktime=int Stall the job x microseconds after an io has completed before
673 issuing the next. May be used to simulate processing being
674 done by an application. See thinktime_blocks and
678 Only valid if thinktime is set - pretend to spend CPU time
679 doing something with the data received, before falling back
680 to sleeping for the rest of the period specified by
684 Only valid if thinktime is set - control how many blocks
685 to issue, before waiting 'thinktime' usecs. If not set,
686 defaults to 1 which will make fio wait 'thinktime' usecs
689 rate=int Cap the bandwidth used by this job. The number is in bytes/sec,
690 the normal suffix rules apply. You can use rate=500k to limit
691 reads and writes to 500k each, or you can specify read and
692 writes separately. Using rate=1m,500k would limit reads to
693 1MB/sec and writes to 500KB/sec. Capping only reads or
694 writes can be done with rate=,500k or rate=500k,. The former
695 will only limit writes (to 500KB/sec), the latter will only
698 ratemin=int Tell fio to do whatever it can to maintain at least this
699 bandwidth. Failing to meet this requirement, will cause
700 the job to exit. The same format as rate is used for
701 read vs write separation.
703 rate_iops=int Cap the bandwidth to this number of IOPS. Basically the same
704 as rate, just specified independently of bandwidth. If the
705 job is given a block size range instead of a fixed value,
706 the smallest block size is used as the metric. The same format
707 as rate is used for read vs write seperation.
709 rate_iops_min=int If fio doesn't meet this rate of IO, it will cause
710 the job to exit. The same format as rate is used for read vs
713 ratecycle=int Average bandwidth for 'rate' and 'ratemin' over this number
716 cpumask=int Set the CPU affinity of this job. The parameter given is a
717 bitmask of allowed CPU's the job may run on. So if you want
718 the allowed CPUs to be 1 and 5, you would pass the decimal
719 value of (1 << 1 | 1 << 5), or 34. See man
720 sched_setaffinity(2). This may not work on all supported
721 operating systems or kernel versions. This option doesn't
722 work well for a higher CPU count than what you can store in
723 an integer mask, so it can only control cpus 1-32. For
724 boxes with larger CPU counts, use cpus_allowed.
726 cpus_allowed=str Controls the same options as cpumask, but it allows a text
727 setting of the permitted CPUs instead. So to use CPUs 1 and
728 5, you would specify cpus_allowed=1,5. This options also
729 allows a range of CPUs. Say you wanted a binding to CPUs
730 1, 5, and 8-15, you would set cpus_allowed=1,5,8-15.
732 startdelay=time Start this job the specified number of seconds after fio
733 has started. Only useful if the job file contains several
734 jobs, and you want to delay starting some jobs to a certain
737 runtime=time Tell fio to terminate processing after the specified number
738 of seconds. It can be quite hard to determine for how long
739 a specified job will run, so this parameter is handy to
740 cap the total runtime to a given time.
742 time_based If set, fio will run for the duration of the runtime
743 specified even if the file(s) are completely read or
744 written. It will simply loop over the same workload
745 as many times as the runtime allows.
747 ramp_time=time If set, fio will run the specified workload for this amount
748 of time before logging any performance numbers. Useful for
749 letting performance settle before logging results, thus
750 minimizing the runtime required for stable results. Note
751 that the ramp_time is considered lead in time for a job,
752 thus it will increase the total runtime if a special timeout
753 or runtime is specified.
755 invalidate=bool Invalidate the buffer/page cache parts for this file prior
756 to starting io. Defaults to true.
758 sync=bool Use sync io for buffered writes. For the majority of the
759 io engines, this means using O_SYNC.
762 mem=str Fio can use various types of memory as the io unit buffer.
763 The allowed values are:
765 malloc Use memory from malloc(3) as the buffers.
767 shm Use shared memory as the buffers. Allocated
770 shmhuge Same as shm, but use huge pages as backing.
772 mmap Use mmap to allocate buffers. May either be
773 anonymous memory, or can be file backed if
774 a filename is given after the option. The
775 format is mem=mmap:/path/to/file.
777 mmaphuge Use a memory mapped huge file as the buffer
778 backing. Append filename after mmaphuge, ala
779 mem=mmaphuge:/hugetlbfs/file
781 The area allocated is a function of the maximum allowed
782 bs size for the job, multiplied by the io depth given. Note
783 that for shmhuge and mmaphuge to work, the system must have
784 free huge pages allocated. This can normally be checked
785 and set by reading/writing /proc/sys/vm/nr_hugepages on a
786 Linux system. Fio assumes a huge page is 4MB in size. So
787 to calculate the number of huge pages you need for a given
788 job file, add up the io depth of all jobs (normally one unless
789 iodepth= is used) and multiply by the maximum bs set. Then
790 divide that number by the huge page size. You can see the
791 size of the huge pages in /proc/meminfo. If no huge pages
792 are allocated by having a non-zero number in nr_hugepages,
793 using mmaphuge or shmhuge will fail. Also see hugepage-size.
795 mmaphuge also needs to have hugetlbfs mounted and the file
796 location should point there. So if it's mounted in /huge,
797 you would use mem=mmaphuge:/huge/somefile.
799 iomem_align=int This indiciates the memory alignment of the IO memory buffers.
800 Note that the given alignment is applied to the first IO unit
801 buffer, if using iodepth the alignment of the following buffers
802 are given by the bs used. In other words, if using a bs that is
803 a multiple of the page sized in the system, all buffers will
804 be aligned to this value. If using a bs that is not page
805 aligned, the alignment of subsequent IO memory buffers is the
806 sum of the iomem_align and bs used.
809 Defines the size of a huge page. Must at least be equal
810 to the system setting, see /proc/meminfo. Defaults to 4MB.
811 Should probably always be a multiple of megabytes, so using
812 hugepage-size=Xm is the preferred way to set this to avoid
813 setting a non-pow-2 bad value.
815 exitall When one job finishes, terminate the rest. The default is
816 to wait for each job to finish, sometimes that is not the
819 bwavgtime=int Average the calculated bandwidth over the given time. Value
820 is specified in milliseconds.
822 create_serialize=bool If true, serialize the file creating for the jobs.
823 This may be handy to avoid interleaving of data
824 files, which may greatly depend on the filesystem
825 used and even the number of processors in the system.
827 create_fsync=bool fsync the data file after creation. This is the
830 create_on_open=bool Don't pre-setup the files for IO, just create open()
831 when it's time to do IO to that file.
833 pre_read=bool If this is given, files will be pre-read into memory before
834 starting the given IO operation. This will also clear
835 the 'invalidate' flag, since it is pointless to pre-read
836 and then drop the cache. This will only work for IO engines
837 that are seekable, since they allow you to read the same data
838 multiple times. Thus it will not work on eg network or splice
841 unlink=bool Unlink the job files when done. Not the default, as repeated
842 runs of that job would then waste time recreating the file
845 loops=int Run the specified number of iterations of this job. Used
846 to repeat the same workload a given number of times. Defaults
849 do_verify=bool Run the verify phase after a write phase. Only makes sense if
850 verify is set. Defaults to 1.
852 verify=str If writing to a file, fio can verify the file contents
853 after each iteration of the job. The allowed values are:
855 md5 Use an md5 sum of the data area and store
856 it in the header of each block.
858 crc64 Use an experimental crc64 sum of the data
859 area and store it in the header of each
862 crc32c Use a crc32c sum of the data area and store
863 it in the header of each block.
865 crc32c-intel Use hardware assisted crc32c calcuation
866 provided on SSE4.2 enabled processors. Falls
867 back to regular software crc32c, if not
868 supported by the system.
870 crc32 Use a crc32 sum of the data area and store
871 it in the header of each block.
873 crc16 Use a crc16 sum of the data area and store
874 it in the header of each block.
876 crc7 Use a crc7 sum of the data area and store
877 it in the header of each block.
879 sha512 Use sha512 as the checksum function.
881 sha256 Use sha256 as the checksum function.
883 sha1 Use optimized sha1 as the checksum function.
885 meta Write extra information about each io
886 (timestamp, block number etc.). The block
887 number is verified. See also verify_pattern.
889 null Only pretend to verify. Useful for testing
890 internals with ioengine=null, not for much
893 This option can be used for repeated burn-in tests of a
894 system to make sure that the written data is also
895 correctly read back. If the data direction given is
896 a read or random read, fio will assume that it should
897 verify a previously written file. If the data direction
898 includes any form of write, the verify will be of the
901 verifysort=bool If set, fio will sort written verify blocks when it deems
902 it faster to read them back in a sorted manner. This is
903 often the case when overwriting an existing file, since
904 the blocks are already laid out in the file system. You
905 can ignore this option unless doing huge amounts of really
906 fast IO where the red-black tree sorting CPU time becomes
909 verify_offset=int Swap the verification header with data somewhere else
910 in the block before writing. Its swapped back before
913 verify_interval=int Write the verification header at a finer granularity
914 than the blocksize. It will be written for chunks the
915 size of header_interval. blocksize should divide this
918 verify_pattern=str If set, fio will fill the io buffers with this
919 pattern. Fio defaults to filling with totally random
920 bytes, but sometimes it's interesting to fill with a known
921 pattern for io verification purposes. Depending on the
922 width of the pattern, fio will fill 1/2/3/4 bytes of the
923 buffer at the time(it can be either a decimal or a hex number).
924 The verify_pattern if larger than a 32-bit quantity has to
925 be a hex number that starts with either "0x" or "0X". Use
928 verify_fatal=bool Normally fio will keep checking the entire contents
929 before quitting on a block verification failure. If this
930 option is set, fio will exit the job on the first observed
933 verify_dump=bool If set, dump the contents of both the original data
934 block and the data block we read off disk to files. This
935 allows later analysis to inspect just what kind of data
936 corruption occurred. On by default.
938 verify_async=int Fio will normally verify IO inline from the submitting
939 thread. This option takes an integer describing how many
940 async offload threads to create for IO verification instead,
941 causing fio to offload the duty of verifying IO contents
942 to one or more separate threads. If using this offload
943 option, even sync IO engines can benefit from using an
944 iodepth setting higher than 1, as it allows them to have
945 IO in flight while verifies are running.
947 verify_async_cpus=str Tell fio to set the given CPU affinity on the
948 async IO verification threads. See cpus_allowed for the
951 verify_backlog=int Fio will normally verify the written contents of a
952 job that utilizes verify once that job has completed. In
953 other words, everything is written then everything is read
954 back and verified. You may want to verify continually
955 instead for a variety of reasons. Fio stores the meta data
956 associated with an IO block in memory, so for large
957 verify workloads, quite a bit of memory would be used up
958 holding this meta data. If this option is enabled, fio
959 will write only N blocks before verifying these blocks.
961 will verify the previously written blocks before continuing
964 verify_backlog_batch=int Control how many blocks fio will verify
965 if verify_backlog is set. If not set, will default to
966 the value of verify_backlog (meaning the entire queue
967 is read back and verified). If verify_backlog_batch is
968 less than verify_backlog then not all blocks will be verified,
969 if verify_backlog_batch is larger than verify_backlog, some
970 blocks will be verified more than once.
972 stonewall Wait for preceeding jobs in the job file to exit, before
973 starting this one. Can be used to insert serialization
974 points in the job file. A stone wall also implies starting
975 a new reporting group.
977 new_group Start a new reporting group. If this option isn't given,
978 jobs in a file will be part of the same reporting group
979 unless separated by a stone wall (or if it's a group
980 by itself, with the numjobs option).
982 numjobs=int Create the specified number of clones of this job. May be
983 used to setup a larger number of threads/processes doing
984 the same thing. We regard that grouping of jobs as a
987 group_reporting If 'numjobs' is set, it may be interesting to display
988 statistics for the group as a whole instead of for each
989 individual job. This is especially true of 'numjobs' is
990 large, looking at individual thread/process output quickly
991 becomes unwieldy. If 'group_reporting' is specified, fio
992 will show the final report per-group instead of per-job.
994 thread fio defaults to forking jobs, however if this option is
995 given, fio will use pthread_create(3) to create threads
998 zonesize=int Divide a file into zones of the specified size. See zoneskip.
1000 zoneskip=int Skip the specified number of bytes when zonesize data has
1001 been read. The two zone options can be used to only do
1002 io on zones of a file.
1004 write_iolog=str Write the issued io patterns to the specified file. See
1005 read_iolog. Specify a separate file for each job, otherwise
1006 the iologs will be interspersed and the file may be corrupt.
1008 read_iolog=str Open an iolog with the specified file name and replay the
1009 io patterns it contains. This can be used to store a
1010 workload and replay it sometime later. The iolog given
1011 may also be a blktrace binary file, which allows fio
1012 to replay a workload captured by blktrace. See blktrace
1013 for how to capture such logging data. For blktrace replay,
1014 the file needs to be turned into a blkparse binary data
1015 file first (blkparse <device> -o /dev/null -d file_for_fio.bin).
1017 replay_no_stall=int When replaying I/O with read_iolog the default behavior
1018 is to attempt to respect the time stamps within the log and
1019 replay them with the appropriate delay between IOPS. By
1020 setting this variable fio will not respect the timestamps and
1021 attempt to replay them as fast as possible while still
1022 respecting ordering. The result is the same I/O pattern to a
1023 given device, but different timings.
1025 replay_redirect=str While replaying I/O patterns using read_iolog the
1026 default behavior is to replay the IOPS onto the major/minor
1027 device that each IOP was recorded from. This is sometimes
1028 undesireable because on a different machine those major/minor
1029 numbers can map to a different device. Changing hardware on
1030 the same system can also result in a different major/minor
1031 mapping. Replay_redirect causes all IOPS to be replayed onto
1032 the single specified device regardless of the device it was
1033 recorded from. i.e. replay_redirect=/dev/sdc would cause all
1034 IO in the blktrace to be replayed onto /dev/sdc. This means
1035 multiple devices will be replayed onto a single, if the trace
1036 contains multiple devices. If you want multiple devices to be
1037 replayed concurrently to multiple redirected devices you must
1038 blkparse your trace into separate traces and replay them with
1039 independent fio invocations. Unfortuantely this also breaks
1040 the strict time ordering between multiple device accesses.
1042 write_bw_log=str If given, write a bandwidth log of the jobs in this job
1043 file. Can be used to store data of the bandwidth of the
1044 jobs in their lifetime. The included fio_generate_plots
1045 script uses gnuplot to turn these text files into nice
1046 graphs. See write_log_log for behaviour of given
1047 filename. For this option, the postfix is _bw.log.
1049 write_lat_log=str Same as write_bw_log, except that this option stores io
1050 submission, completion, and total latencies instead. If no
1051 filename is given with this option, the default filename of
1052 "jobname_type.log" is used. Even if the filename is given,
1053 fio will still append the type of log. So if one specifies
1057 The actual log names will be foo_slat.log, foo_slat.log,
1058 and foo_lat.log. This helps fio_generate_plot fine the logs
1061 lockmem=int Pin down the specified amount of memory with mlock(2). Can
1062 potentially be used instead of removing memory or booting
1063 with less memory to simulate a smaller amount of memory.
1065 exec_prerun=str Before running this job, issue the command specified
1068 exec_postrun=str After the job completes, issue the command specified
1071 ioscheduler=str Attempt to switch the device hosting the file to the specified
1072 io scheduler before running.
1074 cpuload=int If the job is a CPU cycle eater, attempt to use the specified
1075 percentage of CPU cycles.
1077 cpuchunks=int If the job is a CPU cycle eater, split the load into
1078 cycles of the given time. In microseconds.
1080 disk_util=bool Generate disk utilization statistics, if the platform
1081 supports it. Defaults to on.
1083 disable_lat=bool Disable measurements of total latency numbers. Useful
1084 only for cutting back the number of calls to gettimeofday,
1085 as that does impact performance at really high IOPS rates.
1086 Note that to really get rid of a large amount of these
1087 calls, this option must be used with disable_slat and
1090 disable_clat=bool Disable measurements of completion latency numbers. See
1093 disable_slat=bool Disable measurements of submission latency numbers. See
1096 disable_bw=bool Disable measurements of throughput/bandwidth numbers. See
1099 gtod_reduce=bool Enable all of the gettimeofday() reducing options
1100 (disable_clat, disable_slat, disable_bw) plus reduce
1101 precision of the timeout somewhat to really shrink
1102 the gettimeofday() call count. With this option enabled,
1103 we only do about 0.4% of the gtod() calls we would have
1104 done if all time keeping was enabled.
1106 gtod_cpu=int Sometimes it's cheaper to dedicate a single thread of
1107 execution to just getting the current time. Fio (and
1108 databases, for instance) are very intensive on gettimeofday()
1109 calls. With this option, you can set one CPU aside for
1110 doing nothing but logging current time to a shared memory
1111 location. Then the other threads/processes that run IO
1112 workloads need only copy that segment, instead of entering
1113 the kernel with a gettimeofday() call. The CPU set aside
1114 for doing these time calls will be excluded from other
1115 uses. Fio will manually clear it from the CPU mask of other
1118 continue_on_error=bool Normally fio will exit the job on the first observed
1119 failure. If this option is set, fio will continue the job when
1120 there is a 'non-fatal error' (EIO or EILSEQ) until the runtime
1121 is exceeded or the I/O size specified is completed. If this
1122 option is used, there are two more stats that are appended,
1123 the total error count and the first error. The error field
1124 given in the stats is the first error that was hit during the
1127 cgroup=str Add job to this control group. If it doesn't exist, it will
1128 be created. The system must have a mounted cgroup blkio
1129 mount point for this to work. If your system doesn't have it
1130 mounted, you can do so with:
1132 # mount -t cgroup -o blkio none /cgroup
1134 cgroup_weight=int Set the weight of the cgroup to this value. See
1135 the documentation that comes with the kernel, allowed values
1136 are in the range of 100..1000.
1138 cgroup_nodelete=bool Normally fio will delete the cgroups it has created after
1139 the job completion. To override this behavior and to leave
1140 cgroups around after the job completion, set cgroup_nodelete=1.
1141 This can be useful if one wants to inspect various cgroup
1142 files after job completion. Default: false
1144 uid=int Instead of running as the invoking user, set the user ID to
1145 this value before the thread/process does any work.
1147 gid=int Set group ID, see uid.
1149 6.0 Interpreting the output
1150 ---------------------------
1152 fio spits out a lot of output. While running, fio will display the
1153 status of the jobs created. An example of that would be:
1155 Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
1157 The characters inside the square brackets denote the current status of
1158 each thread. The possible values (in typical life cycle order) are:
1162 P Thread setup, but not started.
1164 I Thread initialized, waiting.
1165 p Thread running pre-reading file(s).
1166 R Running, doing sequential reads.
1167 r Running, doing random reads.
1168 W Running, doing sequential writes.
1169 w Running, doing random writes.
1170 M Running, doing mixed sequential reads/writes.
1171 m Running, doing mixed random reads/writes.
1172 F Running, currently waiting for fsync()
1173 V Running, doing verification of written data.
1174 E Thread exited, not reaped by main thread yet.
1177 The other values are fairly self explanatory - number of threads
1178 currently running and doing io, rate of io since last check (read speed
1179 listed first, then write speed), and the estimated completion percentage
1180 and time for the running group. It's impossible to estimate runtime of
1181 the following groups (if any).
1183 When fio is done (or interrupted by ctrl-c), it will show the data for
1184 each thread, group of threads, and disks in that order. For each data
1185 direction, the output looks like:
1187 Client1 (g=0): err= 0:
1188 write: io= 32MB, bw= 666KB/s, runt= 50320msec
1189 slat (msec): min= 0, max= 136, avg= 0.03, stdev= 1.92
1190 clat (msec): min= 0, max= 631, avg=48.50, stdev=86.82
1191 bw (KB/s) : min= 0, max= 1196, per=51.00%, avg=664.02, stdev=681.68
1192 cpu : usr=1.49%, sys=0.25%, ctx=7969, majf=0, minf=17
1193 IO depths : 1=0.1%, 2=0.3%, 4=0.5%, 8=99.0%, 16=0.0%, 32=0.0%, >32=0.0%
1194 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
1195 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
1196 issued r/w: total=0/32768, short=0/0
1197 lat (msec): 2=1.6%, 4=0.0%, 10=3.2%, 20=12.8%, 50=38.4%, 100=24.8%,
1198 lat (msec): 250=15.2%, 500=0.0%, 750=0.0%, 1000=0.0%, >=2048=0.0%
1200 The client number is printed, along with the group id and error of that
1201 thread. Below is the io statistics, here for writes. In the order listed,
1204 io= Number of megabytes io performed
1205 bw= Average bandwidth rate
1206 runt= The runtime of that thread
1207 slat= Submission latency (avg being the average, stdev being the
1208 standard deviation). This is the time it took to submit
1209 the io. For sync io, the slat is really the completion
1210 latency, since queue/complete is one operation there. This
1211 value can be in milliseconds or microseconds, fio will choose
1212 the most appropriate base and print that. In the example
1213 above, milliseconds is the best scale.
1214 clat= Completion latency. Same names as slat, this denotes the
1215 time from submission to completion of the io pieces. For
1216 sync io, clat will usually be equal (or very close) to 0,
1217 as the time from submit to complete is basically just
1218 CPU time (io has already been done, see slat explanation).
1219 bw= Bandwidth. Same names as the xlat stats, but also includes
1220 an approximate percentage of total aggregate bandwidth
1221 this thread received in this group. This last value is
1222 only really useful if the threads in this group are on the
1223 same disk, since they are then competing for disk access.
1224 cpu= CPU usage. User and system time, along with the number
1225 of context switches this thread went through, usage of
1226 system and user time, and finally the number of major
1227 and minor page faults.
1228 IO depths= The distribution of io depths over the job life time. The
1229 numbers are divided into powers of 2, so for example the
1230 16= entries includes depths up to that value but higher
1231 than the previous entry. In other words, it covers the
1232 range from 16 to 31.
1233 IO submit= How many pieces of IO were submitting in a single submit
1234 call. Each entry denotes that amount and below, until
1235 the previous entry - eg, 8=100% mean that we submitted
1236 anywhere in between 5-8 ios per submit call.
1237 IO complete= Like the above submit number, but for completions instead.
1238 IO issued= The number of read/write requests issued, and how many
1240 IO latencies= The distribution of IO completion latencies. This is the
1241 time from when IO leaves fio and when it gets completed.
1242 The numbers follow the same pattern as the IO depths,
1243 meaning that 2=1.6% means that 1.6% of the IO completed
1244 within 2 msecs, 20=12.8% means that 12.8% of the IO
1245 took more than 10 msecs, but less than (or equal to) 20 msecs.
1247 After each client has been listed, the group statistics are printed. They
1248 will look like this:
1250 Run status group 0 (all jobs):
1251 READ: io=64MB, aggrb=22178, minb=11355, maxb=11814, mint=2840msec, maxt=2955msec
1252 WRITE: io=64MB, aggrb=1302, minb=666, maxb=669, mint=50093msec, maxt=50320msec
1254 For each data direction, it prints:
1256 io= Number of megabytes io performed.
1257 aggrb= Aggregate bandwidth of threads in this group.
1258 minb= The minimum average bandwidth a thread saw.
1259 maxb= The maximum average bandwidth a thread saw.
1260 mint= The smallest runtime of the threads in that group.
1261 maxt= The longest runtime of the threads in that group.
1263 And finally, the disk statistics are printed. They will look like this:
1265 Disk stats (read/write):
1266 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
1268 Each value is printed for both reads and writes, with reads first. The
1271 ios= Number of ios performed by all groups.
1272 merge= Number of merges io the io scheduler.
1273 ticks= Number of ticks we kept the disk busy.
1274 io_queue= Total time spent in the disk queue.
1275 util= The disk utilization. A value of 100% means we kept the disk
1276 busy constantly, 50% would be a disk idling half of the time.
1282 For scripted usage where you typically want to generate tables or graphs
1283 of the results, fio can output the results in a semicolon separated format.
1284 The format is one long line of values, such as:
1286 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%
1287 A description of this job goes here.
1289 The job description (if provided) follows on a second line.
1291 To enable terse output, use the --minimal command line option. The first
1292 value is the version of the terse output format. If the output has to
1293 be changed for some reason, this number will be incremented by 1 to
1294 signify that change.
1296 Split up, the format is as follows:
1298 version, jobname, groupid, error
1300 KB IO, bandwidth (KB/sec), runtime (msec)
1301 Submission latency: min, max, mean, deviation
1302 Completion latency: min, max, mean, deviation
1303 Total latency: min, max, mean, deviation
1304 Bw: min, max, aggregate percentage of total, mean, deviation
1306 KB IO, bandwidth (KB/sec), runtime (msec)
1307 Submission latency: min, max, mean, deviation
1308 Completion latency: min, max, mean, deviation
1309 Total latency: min, max, mean, deviation
1310 Bw: min, max, aggregate percentage of total, mean, deviation
1311 CPU usage: user, system, context switches, major faults, minor faults
1312 IO depths: <=1, 2, 4, 8, 16, 32, >=64
1313 IO latencies microseconds: <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
1314 IO latencies milliseconds: <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
1315 Additional Info (dependant on continue_on_error, default off): total # errors, first error code
1317 Additional Info (dependant on description being set): Text description