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 128MiB 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 32KiB and define numjobs to 4 to
154 fork 4 identical jobs. The result is 4 processes each randomly writing
155 to their own 64MiB 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 fio also supports environment variable expansion in job files. Any
162 substring of the form "${VARNAME}" as part of an option value (in other
163 words, on the right of the `='), will be expanded to the value of the
164 environment variable called VARNAME. If no such environment variable
165 is defined, or VARNAME is the empty string, the empty string will be
168 As an example, let's look at a sample fio invocation and job file:
170 $ SIZE=64m NUMJOBS=4 fio jobfile.fio
172 ; -- start job file --
179 This will expand to the following equivalent job file at runtime:
181 ; -- start job file --
188 fio ships with a few example job files, you can also look there for
192 5.0 Detailed list of parameters
193 -------------------------------
195 This section describes in details each parameter associated with a job.
196 Some parameters take an option of a given type, such as an integer or
197 a string. The following types are used:
199 str String. This is a sequence of alpha characters.
200 int Integer. A whole number value, can be negative. If prefixed with
201 0x, the integer is assumed to be of base 16 (hexadecimal).
202 time Integer with possible time postfix. In seconds unless otherwise
203 specified, use eg 10m for 10 minutes. Accepts s/m/h for seconds,
205 siint SI integer. A whole number value, which may contain a postfix
206 describing the base of the number. Accepted postfixes are k/m/g,
207 meaning kilo, mega, and giga. So if you want to specify 4096,
208 you could either write out '4096' or just give 4k. The postfixes
209 signify base 2 values, so 1024 is 1k and 1024k is 1m and so on.
210 If the option accepts an upper and lower range, use a colon ':'
211 or minus '-' to separate such values. See irange.
212 bool Boolean. Usually parsed as an integer, however only defined for
213 true and false (1 and 0).
214 irange Integer range with postfix. Allows value range to be given, such
215 as 1024-4096. A colon may also be used as the separator, eg
216 1k:4k. If the option allows two sets of ranges, they can be
217 specified with a ',' or '/' delimiter: 1k-4k/8k-32k. Also see
220 With the above in mind, here follows the complete list of fio job
223 name=str ASCII name of the job. This may be used to override the
224 name printed by fio for this job. Otherwise the job
225 name is used. On the command line this parameter has the
226 special purpose of also signaling the start of a new
229 description=str Text description of the job. Doesn't do anything except
230 dump this text description when this job is run. It's
233 directory=str Prefix filenames with this directory. Used to places files
234 in a different location than "./".
236 filename=str Fio normally makes up a filename based on the job name,
237 thread number, and file number. If you want to share
238 files between threads in a job or several jobs, specify
239 a filename for each of them to override the default. If
240 the ioengine used is 'net', the filename is the host, port,
241 and protocol to use in the format of =host/port/protocol.
242 See ioengine=net for more. If the ioengine is file based, you
243 can specify a number of files by separating the names with a
244 ':' colon. So if you wanted a job to open /dev/sda and /dev/sdb
245 as the two working files, you would use
246 filename=/dev/sda:/dev/sdb. '-' is a reserved name, meaning
247 stdin or stdout. Which of the two depends on the read/write
250 opendir=str Tell fio to recursively add any file it can find in this
251 directory and down the file system tree.
253 lockfile=str Fio defaults to not doing any locking files before it does
254 IO to them. If a file or file descriptor is shared, fio
255 can serialize IO to that file to make the end result
256 consistent. This is usual for emulating real workloads that
257 share files. The lock modes are:
259 none No locking. The default.
260 exclusive Only one thread/process may do IO,
261 excluding all others.
262 readwrite Read-write locking on the file. Many
263 readers may access the file at the
264 same time, but writes get exclusive
267 The option may be post-fixed with a lock batch number. If
268 set, then each thread/process may do that amount of IOs to
269 the file before giving up the lock. Since lock acquisition is
270 expensive, batching the lock/unlocks will speed up IO.
273 rw=str Type of io pattern. Accepted values are:
275 read Sequential reads
276 write Sequential writes
277 randwrite Random writes
278 randread Random reads
279 rw Sequential mixed reads and writes
280 randrw Random mixed reads and writes
282 For the mixed io types, the default is to split them 50/50.
283 For certain types of io the result may still be skewed a bit,
284 since the speed may be different. It is possible to specify
285 a number of IO's to do before getting a new offset - this
286 is only useful for random IO, where fio would normally
287 generate a new random offset for every IO. If you append
288 eg 8 to randread, you would get a new random offset for
289 every 8 IO's. The result would be a seek for only every 8
290 IO's, instead of for every IO. Use rw=randread:8 to specify
293 randrepeat=bool For random IO workloads, seed the generator in a predictable
294 way so that results are repeatable across repetitions.
296 fadvise_hint=bool By default, fio will use fadvise() to advise the kernel
297 on what IO patterns it is likely to issue. Sometimes you
298 want to test specific IO patterns without telling the
299 kernel about it, in which case you can disable this option.
300 If set, fio will use POSIX_FADV_SEQUENTIAL for sequential
301 IO and POSIX_FADV_RANDOM for random IO.
303 size=siint The total size of file io for this job. Fio will run until
304 this many bytes has been transferred, unless runtime is
305 limited by other options (such as 'runtime', for instance).
306 Unless specific nr_files and filesize options are given,
307 fio will divide this size between the available files
308 specified by the job.
310 filesize=siint Individual file sizes. May be a range, in which case fio
311 will select sizes for files at random within the given range
312 and limited to 'size' in total (if that is given). If not
313 given, each created file is the same size.
315 fill_device=bool Sets size to something really large and waits for ENOSPC (no
316 space left on device) as the terminating condition. Only makes
317 sense with sequential write.
320 bs=siint The block size used for the io units. Defaults to 4k. Values
321 can be given for both read and writes. If a single siint is
322 given, it will apply to both. If a second siint is specified
323 after a comma, it will apply to writes only. In other words,
324 the format is either bs=read_and_write or bs=read,write.
325 bs=4k,8k will thus use 4k blocks for reads, and 8k blocks
326 for writes. If you only wish to set the write size, you
327 can do so by passing an empty read size - bs=,8k will set
328 8k for writes and leave the read default value.
330 blocksize_range=irange
331 bsrange=irange Instead of giving a single block size, specify a range
332 and fio will mix the issued io block sizes. The issued
333 io unit will always be a multiple of the minimum value
334 given (also see bs_unaligned). Applies to both reads and
335 writes, however a second range can be given after a comma.
338 bssplit=str Sometimes you want even finer grained control of the
339 block sizes issued, not just an even split between them.
340 This option allows you to weight various block sizes,
341 so that you are able to define a specific amount of
342 block sizes issued. The format for this option is:
344 bssplit=blocksize/percentage:blocksize/percentage
346 for as many block sizes as needed. So if you want to define
347 a workload that has 50% 64k blocks, 10% 4k blocks, and
348 40% 32k blocks, you would write:
350 bssplit=4k/10:64k/50:32k/40
352 Ordering does not matter. If the percentage is left blank,
353 fio will fill in the remaining values evenly. So a bssplit
354 option like this one:
356 bssplit=4k/50:1k/:32k/
358 would have 50% 4k ios, and 25% 1k and 32k ios. The percentages
359 always add up to 100, if bssplit is given a range that adds
360 up to more, it will error out.
363 bs_unaligned If this option is given, any byte size value within bsrange
364 may be used as a block range. This typically wont work with
365 direct IO, as that normally requires sector alignment.
367 zero_buffers If this option is given, fio will init the IO buffers to
368 all zeroes. The default is to fill them with random data.
370 refill_buffers If this option is given, fio will refill the IO buffers
371 on every submit. The default is to only fill it at init
372 time and reuse that data. Only makes sense if zero_buffers
373 isn't specified, naturally. If data verification is enabled,
374 refill_buffers is also automatically enabled.
376 nrfiles=int Number of files to use for this job. Defaults to 1.
378 openfiles=int Number of files to keep open at the same time. Defaults to
379 the same as nrfiles, can be set smaller to limit the number
382 file_service_type=str Defines how fio decides which file from a job to
383 service next. The following types are defined:
385 random Just choose a file at random.
387 roundrobin Round robin over open files. This
390 sequential Finish one file before moving on to
391 the next. Multiple files can still be
392 open depending on 'openfiles'.
394 The string can have a number appended, indicating how
395 often to switch to a new file. So if option random:4 is
396 given, fio will switch to a new random file after 4 ios
399 ioengine=str Defines how the job issues io to the file. The following
402 sync Basic read(2) or write(2) io. lseek(2) is
403 used to position the io location.
405 psync Basic pread(2) or pwrite(2) io.
407 vsync Basic readv(2) or writev(2) IO.
409 libaio Linux native asynchronous io. Note that Linux
410 may only support queued behaviour with
411 non-buffered IO (set direct=1 or buffered=0).
413 posixaio glibc posix asynchronous io.
415 solarisaio Solaris native asynchronous io.
417 mmap File is memory mapped and data copied
418 to/from using memcpy(3).
420 splice splice(2) is used to transfer the data and
421 vmsplice(2) to transfer data from user
424 syslet-rw Use the syslet system calls to make
425 regular read/write async.
427 sg SCSI generic sg v3 io. May either be
428 synchronous using the SG_IO ioctl, or if
429 the target is an sg character device
430 we use read(2) and write(2) for asynchronous
433 null Doesn't transfer any data, just pretends
434 to. This is mainly used to exercise fio
435 itself and for debugging/testing purposes.
437 net Transfer over the network to given host:port.
438 'filename' must be set appropriately to
439 filename=host/port/protocol regardless of send
440 or receive, if the latter only the port
441 argument is used. 'host' may be an IP address
442 or hostname, port is the port number to be used,
443 and protocol may be 'udp' or 'tcp'. If no
444 protocol is given, TCP is used.
446 netsplice Like net, but uses splice/vmsplice to
447 map data and send/receive.
449 cpuio Doesn't transfer any data, but burns CPU
450 cycles according to the cpuload= and
451 cpucycle= options. Setting cpuload=85
452 will cause that job to do nothing but burn
453 85% of the CPU. In case of SMP machines,
454 use numjobs=<no_of_cpu> to get desired CPU
455 usage, as the cpuload only loads a single
456 CPU at the desired rate.
458 guasi The GUASI IO engine is the Generic Userspace
459 Asyncronous Syscall Interface approach
462 http://www.xmailserver.org/guasi-lib.html
464 for more info on GUASI.
466 external Prefix to specify loading an external
467 IO engine object file. Append the engine
468 filename, eg ioengine=external:/tmp/foo.o
469 to load ioengine foo.o in /tmp.
471 iodepth=int This defines how many io units to keep in flight against
472 the file. The default is 1 for each file defined in this
473 job, can be overridden with a larger value for higher
476 iodepth_batch_submit=int
477 iodepth_batch=int This defines how many pieces of IO to submit at once.
478 It defaults to 1 which means that we submit each IO
479 as soon as it is available, but can be raised to submit
480 bigger batches of IO at the time.
482 iodepth_batch_complete=int This defines how many pieces of IO to retrieve
483 at once. It defaults to 1 which means that we'll ask
484 for a minimum of 1 IO in the retrieval process from
485 the kernel. The IO retrieval will go on until we
486 hit the limit set by iodepth_low. If this variable is
487 set to 0, then fio will always check for completed
488 events before queuing more IO. This helps reduce
489 IO latency, at the cost of more retrieval system calls.
491 iodepth_low=int The low water mark indicating when to start filling
492 the queue again. Defaults to the same as iodepth, meaning
493 that fio will attempt to keep the queue full at all times.
494 If iodepth is set to eg 16 and iodepth_low is set to 4, then
495 after fio has filled the queue of 16 requests, it will let
496 the depth drain down to 4 before starting to fill it again.
498 direct=bool If value is true, use non-buffered io. This is usually
501 buffered=bool If value is true, use buffered io. This is the opposite
502 of the 'direct' option. Defaults to true.
504 offset=siint Start io at the given offset in the file. The data before
505 the given offset will not be touched. This effectively
506 caps the file size at real_size - offset.
508 fsync=int If writing to a file, issue a sync of the dirty data
509 for every number of blocks given. For example, if you give
510 32 as a parameter, fio will sync the file for every 32
511 writes issued. If fio is using non-buffered io, we may
512 not sync the file. The exception is the sg io engine, which
513 synchronizes the disk cache anyway.
515 overwrite=bool If true, writes to a file will always overwrite existing
516 data. If the file doesn't already exist, it will be
517 created before the write phase begins. If the file exists
518 and is large enough for the specified write phase, nothing
521 end_fsync=bool If true, fsync file contents when the job exits.
523 fsync_on_close=bool If true, fio will fsync() a dirty file on close.
524 This differs from end_fsync in that it will happen on every
525 file close, not just at the end of the job.
527 rwmixread=int How large a percentage of the mix should be reads.
529 rwmixwrite=int How large a percentage of the mix should be writes. If both
530 rwmixread and rwmixwrite is given and the values do not add
531 up to 100%, the latter of the two will be used to override
534 norandommap Normally fio will cover every block of the file when doing
535 random IO. If this option is given, fio will just get a
536 new random offset without looking at past io history. This
537 means that some blocks may not be read or written, and that
538 some blocks may be read/written more than once. This option
539 is mutually exclusive with verify= if and only if multiple
540 blocksizes (via bsrange=) are used, since fio only tracks
541 complete rewrites of blocks.
543 softrandommap See norandommap. If fio runs with the random block map enabled
544 and it fails to allocate the map, if this option is set it
545 will continue without a random block map. As coverage will
546 not be as complete as with random maps, this option is
549 nice=int Run the job with the given nice value. See man nice(2).
551 prio=int Set the io priority value of this job. Linux limits us to
552 a positive value between 0 and 7, with 0 being the highest.
555 prioclass=int Set the io priority class. See man ionice(1).
557 thinktime=int Stall the job x microseconds after an io has completed before
558 issuing the next. May be used to simulate processing being
559 done by an application. See thinktime_blocks and
563 Only valid if thinktime is set - pretend to spend CPU time
564 doing something with the data received, before falling back
565 to sleeping for the rest of the period specified by
569 Only valid if thinktime is set - control how many blocks
570 to issue, before waiting 'thinktime' usecs. If not set,
571 defaults to 1 which will make fio wait 'thinktime' usecs
574 rate=int Cap the bandwidth used by this job to this number of KiB/sec.
576 ratemin=int Tell fio to do whatever it can to maintain at least this
577 bandwidth. Failing to meet this requirement, will cause
580 rate_iops=int Cap the bandwidth to this number of IOPS. Basically the same
581 as rate, just specified independently of bandwidth. If the
582 job is given a block size range instead of a fixed value,
583 the smallest block size is used as the metric.
585 rate_iops_min=int If fio doesn't meet this rate of IO, it will cause
588 ratecycle=int Average bandwidth for 'rate' and 'ratemin' over this number
591 cpumask=int Set the CPU affinity of this job. The parameter given is a
592 bitmask of allowed CPU's the job may run on. So if you want
593 the allowed CPUs to be 1 and 5, you would pass the decimal
594 value of (1 << 1 | 1 << 5), or 34. See man
595 sched_setaffinity(2). This may not work on all supported
596 operating systems or kernel versions. This option doesn't
597 work well for a higher CPU count than what you can store in
598 an integer mask, so it can only control cpus 1-32. For
599 boxes with larger CPU counts, use cpus_allowed.
601 cpus_allowed=str Controls the same options as cpumask, but it allows a text
602 setting of the permitted CPUs instead. So to use CPUs 1 and
603 5, you would specify cpus_allowed=1,5. This options also
604 allows a range of CPUs. Say you wanted a binding to CPUs
605 1, 5, and 8-15, you would set cpus_allowed=1,5,8-15.
607 startdelay=time Start this job the specified number of seconds after fio
608 has started. Only useful if the job file contains several
609 jobs, and you want to delay starting some jobs to a certain
612 runtime=time Tell fio to terminate processing after the specified number
613 of seconds. It can be quite hard to determine for how long
614 a specified job will run, so this parameter is handy to
615 cap the total runtime to a given time.
617 time_based If set, fio will run for the duration of the runtime
618 specified even if the file(s) are completely read or
619 written. It will simply loop over the same workload
620 as many times as the runtime allows.
622 ramp_time=time If set, fio will run the specified workload for this amount
623 of time before logging any performance numbers. Useful for
624 letting performance settle before logging results, thus
625 minimizing the runtime required for stable results. Note
626 that the ramp_time is considered lead in time for a job,
627 thus it will increase the total runtime if a special timeout
628 or runtime is specified.
630 invalidate=bool Invalidate the buffer/page cache parts for this file prior
631 to starting io. Defaults to true.
633 sync=bool Use sync io for buffered writes. For the majority of the
634 io engines, this means using O_SYNC.
637 mem=str Fio can use various types of memory as the io unit buffer.
638 The allowed values are:
640 malloc Use memory from malloc(3) as the buffers.
642 shm Use shared memory as the buffers. Allocated
645 shmhuge Same as shm, but use huge pages as backing.
647 mmap Use mmap to allocate buffers. May either be
648 anonymous memory, or can be file backed if
649 a filename is given after the option. The
650 format is mem=mmap:/path/to/file.
652 mmaphuge Use a memory mapped huge file as the buffer
653 backing. Append filename after mmaphuge, ala
654 mem=mmaphuge:/hugetlbfs/file
656 The area allocated is a function of the maximum allowed
657 bs size for the job, multiplied by the io depth given. Note
658 that for shmhuge and mmaphuge to work, the system must have
659 free huge pages allocated. This can normally be checked
660 and set by reading/writing /proc/sys/vm/nr_hugepages on a
661 Linux system. Fio assumes a huge page is 4MiB in size. So
662 to calculate the number of huge pages you need for a given
663 job file, add up the io depth of all jobs (normally one unless
664 iodepth= is used) and multiply by the maximum bs set. Then
665 divide that number by the huge page size. You can see the
666 size of the huge pages in /proc/meminfo. If no huge pages
667 are allocated by having a non-zero number in nr_hugepages,
668 using mmaphuge or shmhuge will fail. Also see hugepage-size.
670 mmaphuge also needs to have hugetlbfs mounted and the file
671 location should point there. So if it's mounted in /huge,
672 you would use mem=mmaphuge:/huge/somefile.
675 Defines the size of a huge page. Must at least be equal
676 to the system setting, see /proc/meminfo. Defaults to 4MiB.
677 Should probably always be a multiple of megabytes, so using
678 hugepage-size=Xm is the preferred way to set this to avoid
679 setting a non-pow-2 bad value.
681 exitall When one job finishes, terminate the rest. The default is
682 to wait for each job to finish, sometimes that is not the
685 bwavgtime=int Average the calculated bandwidth over the given time. Value
686 is specified in milliseconds.
688 create_serialize=bool If true, serialize the file creating for the jobs.
689 This may be handy to avoid interleaving of data
690 files, which may greatly depend on the filesystem
691 used and even the number of processors in the system.
693 create_fsync=bool fsync the data file after creation. This is the
696 unlink=bool Unlink the job files when done. Not the default, as repeated
697 runs of that job would then waste time recreating the file
700 loops=int Run the specified number of iterations of this job. Used
701 to repeat the same workload a given number of times. Defaults
704 do_verify=bool Run the verify phase after a write phase. Only makes sense if
705 verify is set. Defaults to 1.
707 verify=str If writing to a file, fio can verify the file contents
708 after each iteration of the job. The allowed values are:
710 md5 Use an md5 sum of the data area and store
711 it in the header of each block.
713 crc64 Use an experimental crc64 sum of the data
714 area and store it in the header of each
717 crc32c Use a crc32c sum of the data area and store
718 it in the header of each block.
720 crc32c-intel Use hardware assisted crc32c calcuation
721 provided on SSE4.2 enabled processors.
723 crc32 Use a crc32 sum of the data area and store
724 it in the header of each block.
726 crc16 Use a crc16 sum of the data area and store
727 it in the header of each block.
729 crc7 Use a crc7 sum of the data area and store
730 it in the header of each block.
732 sha512 Use sha512 as the checksum function.
734 sha256 Use sha256 as the checksum function.
736 meta Write extra information about each io
737 (timestamp, block number etc.). The block
740 null Only pretend to verify. Useful for testing
741 internals with ioengine=null, not for much
744 This option can be used for repeated burn-in tests of a
745 system to make sure that the written data is also
748 verifysort=bool If set, fio will sort written verify blocks when it deems
749 it faster to read them back in a sorted manner. This is
750 often the case when overwriting an existing file, since
751 the blocks are already laid out in the file system. You
752 can ignore this option unless doing huge amounts of really
753 fast IO where the red-black tree sorting CPU time becomes
756 verify_offset=siint Swap the verification header with data somewhere else
757 in the block before writing. Its swapped back before
760 verify_interval=siint Write the verification header at a finer granularity
761 than the blocksize. It will be written for chunks the
762 size of header_interval. blocksize should divide this
765 verify_pattern=int If set, fio will fill the io buffers with this
766 pattern. Fio defaults to filling with totally random
767 bytes, but sometimes it's interesting to fill with a known
768 pattern for io verification purposes. Depending on the
769 width of the pattern, fio will fill 1/2/3/4 bytes of the
770 buffer at the time. The verify_pattern cannot be larger than
773 verify_fatal=bool Normally fio will keep checking the entire contents
774 before quitting on a block verification failure. If this
775 option is set, fio will exit the job on the first observed
778 stonewall Wait for preceeding jobs in the job file to exit, before
779 starting this one. Can be used to insert serialization
780 points in the job file. A stone wall also implies starting
781 a new reporting group.
783 new_group Start a new reporting group. If this option isn't given,
784 jobs in a file will be part of the same reporting group
785 unless separated by a stone wall (or if it's a group
786 by itself, with the numjobs option).
788 numjobs=int Create the specified number of clones of this job. May be
789 used to setup a larger number of threads/processes doing
790 the same thing. We regard that grouping of jobs as a
793 group_reporting If 'numjobs' is set, it may be interesting to display
794 statistics for the group as a whole instead of for each
795 individual job. This is especially true of 'numjobs' is
796 large, looking at individual thread/process output quickly
797 becomes unwieldy. If 'group_reporting' is specified, fio
798 will show the final report per-group instead of per-job.
800 thread fio defaults to forking jobs, however if this option is
801 given, fio will use pthread_create(3) to create threads
804 zonesize=siint Divide a file into zones of the specified size. See zoneskip.
806 zoneskip=siint Skip the specified number of bytes when zonesize data has
807 been read. The two zone options can be used to only do
808 io on zones of a file.
810 write_iolog=str Write the issued io patterns to the specified file. See
813 read_iolog=str Open an iolog with the specified file name and replay the
814 io patterns it contains. This can be used to store a
815 workload and replay it sometime later. The iolog given
816 may also be a blktrace binary file, which allows fio
817 to replay a workload captured by blktrace. See blktrace
818 for how to capture such logging data. For blktrace replay,
819 the file needs to be turned into a blkparse binary data
820 file first (blktrace <device> -d file_for_fio.bin).
822 write_bw_log=str If given, write a bandwidth log of the jobs in this job
823 file. Can be used to store data of the bandwidth of the
824 jobs in their lifetime. The included fio_generate_plots
825 script uses gnuplot to turn these text files into nice
826 graphs. See write_log_log for behaviour of given
827 filename. For this option, the postfix is _bw.log.
829 write_lat_log=str Same as write_bw_log, except that this option stores io
830 completion latencies instead. If no filename is given
831 with this option, the default filename of "jobname_type.log"
832 is used. Even if the filename is given, fio will still
833 append the type of log. So if one specifies
837 The actual log names will be foo_clat.log and foo_slat.log.
838 This helps fio_generate_plot fine the logs automatically.
840 lockmem=siint Pin down the specified amount of memory with mlock(2). Can
841 potentially be used instead of removing memory or booting
842 with less memory to simulate a smaller amount of memory.
844 exec_prerun=str Before running this job, issue the command specified
847 exec_postrun=str After the job completes, issue the command specified
850 ioscheduler=str Attempt to switch the device hosting the file to the specified
851 io scheduler before running.
853 cpuload=int If the job is a CPU cycle eater, attempt to use the specified
854 percentage of CPU cycles.
856 cpuchunks=int If the job is a CPU cycle eater, split the load into
857 cycles of the given time. In milliseconds.
859 disk_util=bool Generate disk utilization statistics, if the platform
860 supports it. Defaults to on.
862 disable_clat=bool Disable measurements of completion latency numbers. Useful
863 only for cutting back the number of calls to gettimeofday,
864 as that does impact performance at really high IOPS rates.
865 Note that to really get rid of a large amount of these
866 calls, this option must be used with disable_slat and
869 disable_slat=bool Disable measurements of submission latency numbers. See
872 disable_bw=bool Disable measurements of throughput/bandwidth numbers. See
875 gtod_reduce=bool Enable all of the gettimeofday() reducing options
876 (disable_clat, disable_slat, disable_bw) plus reduce
877 precision of the timeout somewhat to really shrink
878 the gettimeofday() call count. With this option enabled,
879 we only do about 0.4% of the gtod() calls we would have
880 done if all time keeping was enabled.
882 gtod_cpu=int Sometimes it's cheaper to dedicate a single thread of
883 execution to just getting the current time. Fio (and
884 databases, for instance) are very intensive on gettimeofday()
885 calls. With this option, you can set one CPU aside for
886 doing nothing but logging current time to a shared memory
887 location. Then the other threads/processes that run IO
888 workloads need only copy that segment, instead of entering
889 the kernel with a gettimeofday() call. The CPU set aside
890 for doing these time calls will be excluded from other
891 uses. Fio will manually clear it from the CPU mask of other
895 6.0 Interpreting the output
896 ---------------------------
898 fio spits out a lot of output. While running, fio will display the
899 status of the jobs created. An example of that would be:
901 Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
903 The characters inside the square brackets denote the current status of
904 each thread. The possible values (in typical life cycle order) are:
908 P Thread setup, but not started.
910 I Thread initialized, waiting.
911 R Running, doing sequential reads.
912 r Running, doing random reads.
913 W Running, doing sequential writes.
914 w Running, doing random writes.
915 M Running, doing mixed sequential reads/writes.
916 m Running, doing mixed random reads/writes.
917 F Running, currently waiting for fsync()
918 V Running, doing verification of written data.
919 E Thread exited, not reaped by main thread yet.
922 The other values are fairly self explanatory - number of threads
923 currently running and doing io, rate of io since last check (read speed
924 listed first, then write speed), and the estimated completion percentage
925 and time for the running group. It's impossible to estimate runtime of
926 the following groups (if any).
928 When fio is done (or interrupted by ctrl-c), it will show the data for
929 each thread, group of threads, and disks in that order. For each data
930 direction, the output looks like:
932 Client1 (g=0): err= 0:
933 write: io= 32MiB, bw= 666KiB/s, runt= 50320msec
934 slat (msec): min= 0, max= 136, avg= 0.03, stdev= 1.92
935 clat (msec): min= 0, max= 631, avg=48.50, stdev=86.82
936 bw (KiB/s) : min= 0, max= 1196, per=51.00%, avg=664.02, stdev=681.68
937 cpu : usr=1.49%, sys=0.25%, ctx=7969, majf=0, minf=17
938 IO depths : 1=0.1%, 2=0.3%, 4=0.5%, 8=99.0%, 16=0.0%, 32=0.0%, >32=0.0%
939 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
940 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
941 issued r/w: total=0/32768, short=0/0
942 lat (msec): 2=1.6%, 4=0.0%, 10=3.2%, 20=12.8%, 50=38.4%, 100=24.8%,
943 lat (msec): 250=15.2%, 500=0.0%, 750=0.0%, 1000=0.0%, >=2048=0.0%
945 The client number is printed, along with the group id and error of that
946 thread. Below is the io statistics, here for writes. In the order listed,
949 io= Number of megabytes io performed
950 bw= Average bandwidth rate
951 runt= The runtime of that thread
952 slat= Submission latency (avg being the average, stdev being the
953 standard deviation). This is the time it took to submit
954 the io. For sync io, the slat is really the completion
955 latency, since queue/complete is one operation there. This
956 value can be in milliseconds or microseconds, fio will choose
957 the most appropriate base and print that. In the example
958 above, milliseconds is the best scale.
959 clat= Completion latency. Same names as slat, this denotes the
960 time from submission to completion of the io pieces. For
961 sync io, clat will usually be equal (or very close) to 0,
962 as the time from submit to complete is basically just
963 CPU time (io has already been done, see slat explanation).
964 bw= Bandwidth. Same names as the xlat stats, but also includes
965 an approximate percentage of total aggregate bandwidth
966 this thread received in this group. This last value is
967 only really useful if the threads in this group are on the
968 same disk, since they are then competing for disk access.
969 cpu= CPU usage. User and system time, along with the number
970 of context switches this thread went through, usage of
971 system and user time, and finally the number of major
972 and minor page faults.
973 IO depths= The distribution of io depths over the job life time. The
974 numbers are divided into powers of 2, so for example the
975 16= entries includes depths up to that value but higher
976 than the previous entry. In other words, it covers the
978 IO submit= How many pieces of IO were submitting in a single submit
979 call. Each entry denotes that amount and below, until
980 the previous entry - eg, 8=100% mean that we submitted
981 anywhere in between 5-8 ios per submit call.
982 IO complete= Like the above submit number, but for completions instead.
983 IO issued= The number of read/write requests issued, and how many
985 IO latencies= The distribution of IO completion latencies. This is the
986 time from when IO leaves fio and when it gets completed.
987 The numbers follow the same pattern as the IO depths,
988 meaning that 2=1.6% means that 1.6% of the IO completed
989 within 2 msecs, 20=12.8% means that 12.8% of the IO
990 took more than 10 msecs, but less than (or equal to) 20 msecs.
992 After each client has been listed, the group statistics are printed. They
995 Run status group 0 (all jobs):
996 READ: io=64MiB, aggrb=22178, minb=11355, maxb=11814, mint=2840msec, maxt=2955msec
997 WRITE: io=64MiB, aggrb=1302, minb=666, maxb=669, mint=50093msec, maxt=50320msec
999 For each data direction, it prints:
1001 io= Number of megabytes io performed.
1002 aggrb= Aggregate bandwidth of threads in this group.
1003 minb= The minimum average bandwidth a thread saw.
1004 maxb= The maximum average bandwidth a thread saw.
1005 mint= The smallest runtime of the threads in that group.
1006 maxt= The longest runtime of the threads in that group.
1008 And finally, the disk statistics are printed. They will look like this:
1010 Disk stats (read/write):
1011 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
1013 Each value is printed for both reads and writes, with reads first. The
1016 ios= Number of ios performed by all groups.
1017 merge= Number of merges io the io scheduler.
1018 ticks= Number of ticks we kept the disk busy.
1019 io_queue= Total time spent in the disk queue.
1020 util= The disk utilization. A value of 100% means we kept the disk
1021 busy constantly, 50% would be a disk idling half of the time.
1027 For scripted usage where you typically want to generate tables or graphs
1028 of the results, fio can output the results in a semicolon separated format.
1029 The format is one long line of values, such as:
1031 client1;0;0;1906777;1090804;1790;0;0;0.000000;0.000000;0;0;0.000000;0.000000;929380;1152890;25.510151%;1078276.333333;128948.113404;0;0;0;0;0;0.000000;0.000000;0;0;0.000000;0.000000;0;0;0.000000%;0.000000;0.000000;100.000000%;0.000000%;324;100.0%;0.0%;0.0%;0.0%;0.0%;0.0%;0.0%;100.0%;0.0%;0.0%;0.0%;0.0%;0.0%
1032 ;0.0%;0.0%;0.0%;0.0%;0.0%
1034 To enable terse output, use the --minimal command line option.
1036 Split up, the format is as follows:
1038 jobname, groupid, error
1040 KiB IO, bandwidth (KiB/sec), runtime (msec)
1041 Submission latency: min, max, mean, deviation
1042 Completion latency: min, max, mean, deviation
1043 Bw: min, max, aggregate percentage of total, mean, deviation
1045 KiB IO, bandwidth (KiB/sec), runtime (msec)
1046 Submission latency: min, max, mean, deviation
1047 Completion latency: min, max, mean, deviation
1048 Bw: min, max, aggregate percentage of total, mean, deviation
1049 CPU usage: user, system, context switches, major faults, minor faults
1050 IO depths: <=1, 2, 4, 8, 16, 32, >=64
1051 IO latencies: <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, >=2000