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 lets look at a really simple job file that define to threads, 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 Lets look at an example that have 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 ships with a few example job files, you can also look there for
165 5.0 Detailed list of parameters
166 -------------------------------
168 This section describes in details each parameter associated with a job.
169 Some parameters take an option of a given type, such as an integer or
170 a string. The following types are used:
172 str String. This is a sequence of alpha characters.
173 int Integer. A whole number value, can be negative. If prefixed with
174 0x, the integer is assumed to be of base 16 (hexidecimal).
175 siint SI integer. A whole number value, which may contain a postfix
176 describing the base of the number. Accepted postfixes are k/m/g,
177 meaning kilo, mega, and giga. So if you want to specify 4096,
178 you could either write out '4096' or just give 4k. The postfixes
179 signify base 2 values, so 1024 is 1k and 1024k is 1m and so on.
180 If the option accepts an upper and lower range, use a colon ':'
181 or minus '-' to seperate such values. See irange.
182 bool Boolean. Usually parsed as an integer, however only defined for
183 true and false (1 and 0).
184 irange Integer range with postfix. Allows value range to be given, such
185 as 1024-4096. A colon may also be used as the seperator, eg
186 1k:4k. If the option allows two sets of ranges, they can be
187 specified with a ',' or '/' delimiter: 1k-4k/8k-32k. Also see
190 With the above in mind, here follows the complete list of fio job
193 name=str ASCII name of the job. This may be used to override the
194 name printed by fio for this job. Otherwise the job
195 name is used. On the command line this parameter has the
196 special purpose of also signaling the start of a new
199 description=str Text description of the job. Doesn't do anything except
200 dump this text description when this job is run. It's
203 directory=str Prefix filenames with this directory. Used to places files
204 in a different location than "./".
206 filename=str Fio normally makes up a filename based on the job name,
207 thread number, and file number. If you want to share
208 files between threads in a job or several jobs, specify
209 a filename for each of them to override the default. If
210 the ioengine used is 'net', the filename is the host and
211 port to connect to in the format of =host/port. If the
212 ioengine is file based, you can specify a number of files
213 by seperating the names with a ':' colon. So if you wanted
214 a job to open /dev/sda and /dev/sdb as the two working files,
215 you would use filename=/dev/sda:/dev/sdb. '-' is a reserved
216 name, meaning stdin or stdout. Which of the two depends
217 on the read/write direction set.
219 opendir=str Tell fio to recursively add any file it can find in this
220 directory and down the file system tree.
223 rw=str Type of io pattern. Accepted values are:
225 read Sequential reads
226 write Sequential writes
227 randwrite Random writes
228 randread Random reads
229 rw Sequential mixed reads and writes
230 randrw Random mixed reads and writes
232 For the mixed io types, the default is to split them 50/50.
233 For certain types of io the result may still be skewed a bit,
234 since the speed may be different. It is possible to specify
235 a number of IO's to do before getting a new offset - this
236 is only useful for random IO, where fio would normally
237 generate a new random offset for every IO. If you append
238 eg 8 to randread, you would get a new random offset for
239 every 8 IO's. The result would be a seek for only every 8
240 IO's, instead of for every IO. Use rw=randread:8 to specify
243 randrepeat=bool For random IO workloads, seed the generator in a predictable
244 way so that results are repeatable across repetitions.
246 fadvise_hint=bool By default, fio will use fadvise() to advise the kernel
247 on what IO patterns it is likely to issue. Sometimes you
248 want to test specific IO patterns without telling the
249 kernel about it, in which case you can disable this option.
250 If set, fio will use POSIX_FADV_SEQUENTIAL for sequential
251 IO and POSIX_FADV_RANDOM for random IO.
253 size=siint The total size of file io for this job. Fio will run until
254 this many bytes has been transferred, unless runtime is
255 limited by other options (such as 'runtime', for instance).
256 Unless specific nr_files and filesize options are given,
257 fio will divide this size between the available files
258 specified by the job.
260 filesize=siint Individual file sizes. May be a range, in which case fio
261 will select sizes for files at random within the given range
262 and limited to 'size' in total (if that is given). If not
263 given, each created file is the same size.
266 bs=siint The block size used for the io units. Defaults to 4k. Values
267 can be given for both read and writes. If a single siint is
268 given, it will apply to both. If a second siint is specified
269 after a comma, it will apply to writes only. In other words,
270 the format is either bs=read_and_write or bs=read,write.
271 bs=4k,8k will thus use 4k blocks for reads, and 8k blocks
272 for writes. If you only wish to set the write size, you
273 can do so by passing an empty read size - bs=,8k will set
274 8k for writes and leave the read default value.
276 blocksize_range=irange
277 bsrange=irange Instead of giving a single block size, specify a range
278 and fio will mix the issued io block sizes. The issued
279 io unit will always be a multiple of the minimum value
280 given (also see bs_unaligned). Applies to both reads and
281 writes, however a second range can be given after a comma.
285 bs_unaligned If this option is given, any byte size value within bsrange
286 may be used as a block range. This typically wont work with
287 direct IO, as that normally requires sector alignment.
289 zero_buffers If this option is given, fio will init the IO buffers to
290 all zeroes. The default is to fill them with random data.
292 nrfiles=int Number of files to use for this job. Defaults to 1.
294 openfiles=int Number of files to keep open at the same time. Defaults to
295 the same as nrfiles, can be set smaller to limit the number
298 file_service_type=str Defines how fio decides which file from a job to
299 service next. The following types are defined:
301 random Just choose a file at random.
303 roundrobin Round robin over open files. This
306 The string can have a number appended, indicating how
307 often to switch to a new file. So if option random:4 is
308 given, fio will switch to a new random file after 4 ios
311 ioengine=str Defines how the job issues io to the file. The following
314 sync Basic read(2) or write(2) io. lseek(2) is
315 used to position the io location.
317 libaio Linux native asynchronous io.
319 posixaio glibc posix asynchronous io.
321 mmap File is memory mapped and data copied
322 to/from using memcpy(3).
324 splice splice(2) is used to transfer the data and
325 vmsplice(2) to transfer data from user
328 syslet-rw Use the syslet system calls to make
329 regular read/write async.
331 sg SCSI generic sg v3 io. May either be
332 synchronous using the SG_IO ioctl, or if
333 the target is an sg character device
334 we use read(2) and write(2) for asynchronous
337 null Doesn't transfer any data, just pretends
338 to. This is mainly used to exercise fio
339 itself and for debugging/testing purposes.
341 net Transfer over the network to given host:port.
342 'filename' must be set appropriately to
343 filename=host/port regardless of send
344 or receive, if the latter only the port
347 netsplice Like net, but uses splice/vmsplice to
348 map data and send/receive.
350 cpu Doesn't transfer any data, but burns CPU
351 cycles according to the cpuload= and
352 cpucycle= options. Setting cpuload=85
353 will cause that job to do nothing but burn
356 guasi The GUASI IO engine is the Generic Userspace
357 Asyncronous Syscall Interface approach
360 http://www.xmailserver.org/guasi-lib.html
362 for more info on GUASI.
364 external Prefix to specify loading an external
365 IO engine object file. Append the engine
366 filename, eg ioengine=external:/tmp/foo.o
367 to load ioengine foo.o in /tmp.
369 iodepth=int This defines how many io units to keep in flight against
370 the file. The default is 1 for each file defined in this
371 job, can be overridden with a larger value for higher
374 iodepth_batch=int This defines how many pieces of IO to submit at once.
375 It defaults to the same as iodepth, but can be set lower
378 iodepth_low=int The low water mark indicating when to start filling
379 the queue again. Defaults to the same as iodepth, meaning
380 that fio will attempt to keep the queue full at all times.
381 If iodepth is set to eg 16 and iodepth_low is set to 4, then
382 after fio has filled the queue of 16 requests, it will let
383 the depth drain down to 4 before starting to fill it again.
385 direct=bool If value is true, use non-buffered io. This is usually
388 buffered=bool If value is true, use buffered io. This is the opposite
389 of the 'direct' option. Defaults to true.
391 offset=siint Start io at the given offset in the file. The data before
392 the given offset will not be touched. This effectively
393 caps the file size at real_size - offset.
395 fsync=int If writing to a file, issue a sync of the dirty data
396 for every number of blocks given. For example, if you give
397 32 as a parameter, fio will sync the file for every 32
398 writes issued. If fio is using non-buffered io, we may
399 not sync the file. The exception is the sg io engine, which
400 synchronizes the disk cache anyway.
402 overwrite=bool If writing to a file, setup the file first and do overwrites.
404 end_fsync=bool If true, fsync file contents when the job exits.
406 fsync_on_close=bool If true, fio will fsync() a dirty file on close.
407 This differs from end_fsync in that it will happen on every
408 file close, not just at the end of the job.
410 rwmixcycle=int Value in milliseconds describing how often to switch between
411 reads and writes for a mixed workload. The default is
414 rwmixread=int How large a percentage of the mix should be reads.
416 rwmixwrite=int How large a percentage of the mix should be writes. If both
417 rwmixread and rwmixwrite is given and the values do not add
418 up to 100%, the latter of the two will be used to override
421 norandommap Normally fio will cover every block of the file when doing
422 random IO. If this option is given, fio will just get a
423 new random offset without looking at past io history. This
424 means that some blocks may not be read or written, and that
425 some blocks may be read/written more than once. This option
426 is mutually exclusive with verify= for that reason, since
427 fio doesn't track potential block rewrites which may alter
428 the calculated checksum for that block.
430 nice=int Run the job with the given nice value. See man nice(2).
432 prio=int Set the io priority value of this job. Linux limits us to
433 a positive value between 0 and 7, with 0 being the highest.
436 prioclass=int Set the io priority class. See man ionice(1).
438 thinktime=int Stall the job x microseconds after an io has completed before
439 issuing the next. May be used to simulate processing being
440 done by an application. See thinktime_blocks and
444 Only valid if thinktime is set - pretend to spend CPU time
445 doing something with the data received, before falling back
446 to sleeping for the rest of the period specified by
450 Only valid if thinktime is set - control how many blocks
451 to issue, before waiting 'thinktime' usecs. If not set,
452 defaults to 1 which will make fio wait 'thinktime' usecs
455 rate=int Cap the bandwidth used by this job to this number of KiB/sec.
457 ratemin=int Tell fio to do whatever it can to maintain at least this
458 bandwidth. Failing to meet this requirement, will cause
461 rate_iops=int Cap the bandwidth to this number of IOPS. Basically the same
462 as rate, just specified independently of bandwidth. If the
463 job is given a block size range instead of a fixed value,
464 the smallest block size is used as the metric.
466 rate_iops_min=int If fio doesn't meet this rate of IO, it will cause
469 ratecycle=int Average bandwidth for 'rate' and 'ratemin' over this number
472 cpumask=int Set the CPU affinity of this job. The parameter given is a
473 bitmask of allowed CPU's the job may run on. So if you want
474 the allowed CPUs to be 1 and 5, you would pass the decimal
475 value of (1 << 1 | 1 << 5), or 34. See man
476 sched_setaffinity(2). This may not work on all supported
477 operating systems or kernel versions.
479 cpus_allowed=str Controls the same options as cpumask, but it allows a text
480 setting of the permitted CPUs instead. So to use CPUs 1 and
481 5, you would specify cpus_allowed=1,5.
483 startdelay=int Start this job the specified number of seconds after fio
484 has started. Only useful if the job file contains several
485 jobs, and you want to delay starting some jobs to a certain
488 runtime=int Tell fio to terminate processing after the specified number
489 of seconds. It can be quite hard to determine for how long
490 a specified job will run, so this parameter is handy to
491 cap the total runtime to a given time.
493 time_based If set, fio will run for the duration of the runtime
494 specified even if the file(s) are completey read or
495 written. It will simply loop over the same workload
496 as many times as the runtime allows.
498 invalidate=bool Invalidate the buffer/page cache parts for this file prior
499 to starting io. Defaults to true.
501 sync=bool Use sync io for buffered writes. For the majority of the
502 io engines, this means using O_SYNC.
505 mem=str Fio can use various types of memory as the io unit buffer.
506 The allowed values are:
508 malloc Use memory from malloc(3) as the buffers.
510 shm Use shared memory as the buffers. Allocated
513 shmhuge Same as shm, but use huge pages as backing.
515 mmap Use mmap to allocate buffers. May either be
516 anonymous memory, or can be file backed if
517 a filename is given after the option. The
518 format is mem=mmap:/path/to/file.
520 mmaphuge Use a memory mapped huge file as the buffer
521 backing. Append filename after mmaphuge, ala
522 mem=mmaphuge:/hugetlbfs/file
524 The area allocated is a function of the maximum allowed
525 bs size for the job, multiplied by the io depth given. Note
526 that for shmhuge and mmaphuge to work, the system must have
527 free huge pages allocated. This can normally be checked
528 and set by reading/writing /proc/sys/vm/nr_hugepages on a
529 Linux system. Fio assumes a huge page is 4MiB in size. So
530 to calculate the number of huge pages you need for a given
531 job file, add up the io depth of all jobs (normally one unless
532 iodepth= is used) and multiply by the maximum bs set. Then
533 divide that number by the huge page size. You can see the
534 size of the huge pages in /proc/meminfo. If no huge pages
535 are allocated by having a non-zero number in nr_hugepages,
536 using mmaphuge or shmhuge will fail. Also see hugepage-size.
538 mmaphuge also needs to have hugetlbfs mounted and the file
539 location should point there. So if it's mounted in /huge,
540 you would use mem=mmaphuge:/huge/somefile.
543 Defines the size of a huge page. Must at least be equal
544 to the system setting, see /proc/meminfo. Defaults to 4MiB.
545 Should probably always be a multiple of megabytes, so using
546 hugepage-size=Xm is the preferred way to set this to avoid
547 setting a non-pow-2 bad value.
549 exitall When one job finishes, terminate the rest. The default is
550 to wait for each job to finish, sometimes that is not the
553 bwavgtime=int Average the calculated bandwidth over the given time. Value
554 is specified in milliseconds.
556 create_serialize=bool If true, serialize the file creating for the jobs.
557 This may be handy to avoid interleaving of data
558 files, which may greatly depend on the filesystem
559 used and even the number of processors in the system.
561 create_fsync=bool fsync the data file after creation. This is the
564 unlink=bool Unlink the job files when done. Not the default, as repeated
565 runs of that job would then waste time recreating the fileset
568 loops=int Run the specified number of iterations of this job. Used
569 to repeat the same workload a given number of times. Defaults
572 do_verify=bool Run the verify phase after a write phase. Only makes sense if
573 verify is set. Defaults to 1.
575 verify=str If writing to a file, fio can verify the file contents
576 after each iteration of the job. The allowed values are:
578 md5 Use an md5 sum of the data area and store
579 it in the header of each block.
581 crc64 Use an experimental crc64 sum of the data
582 area and store it in the header of each
585 crc32 Use a crc32 sum of the data area and store
586 it in the header of each block.
588 crc16 Use a crc16 sum of the data area and store
589 it in the header of each block.
591 crc7 Use a crc7 sum of the data area and store
592 it in the header of each block.
594 sha512 Use sha512 as the checksum function.
596 sha256 Use sha256 as the checksum function.
598 meta Write extra information about each io
599 (timestamp, block number etc.). The block
602 pattern Fill the IO buffers with a specific pattern,
603 that we can use to verify. Depending on the
604 width of the pattern, fio will fill 1/2/3/4
605 bytes of the buffer at the time. The pattern
606 cannot be larger than a 32-bit quantity. The
607 given pattern is given as a postfix to this
608 option, ala: verify=pattern:0x5a. It accepts
609 both hex and dec values.
611 null Only pretend to verify. Useful for testing
612 internals with ioengine=null, not for much
615 This option can be used for repeated burn-in tests of a
616 system to make sure that the written data is also
619 verifysort=bool If set, fio will sort written verify blocks when it deems
620 it faster to read them back in a sorted manner. This is
621 often the case when overwriting an existing file, since
622 the blocks are already laid out in the file system. You
623 can ignore this option unless doing huge amounts of really
624 fast IO where the red-black tree sorting CPU time becomes
627 verify_offset=siint Swap the verification header with data somewhere else
628 in the block before writing. Its swapped back before
631 verify_interval=siint Write the verification header at a finer granularity
632 than the blocksize. It will be written for chunks the
633 size of header_interval. blocksize should divide this
636 verify_fatal=bool Normally fio will keep checking the entire contents
637 before quitting on a block verification failure. If this
638 option is set, fio will exit the job on the first observed
641 stonewall Wait for preceeding jobs in the job file to exit, before
642 starting this one. Can be used to insert serialization
643 points in the job file. A stone wall also implies starting
644 a new reporting group.
646 new_group Start a new reporting group. If this option isn't given,
647 jobs in a file will be part of the same reporting group
648 unless seperated by a stone wall (or if it's a group
649 by itself, with the numjobs option).
651 numjobs=int Create the specified number of clones of this job. May be
652 used to setup a larger number of threads/processes doing
653 the same thing. We regard that grouping of jobs as a
656 group_reporting If 'numjobs' is set, it may be interesting to display
657 statistics for the group as a whole instead of for each
658 individual job. This is especially true of 'numjobs' is
659 large, looking at individual thread/process output quickly
660 becomes unwieldy. If 'group_reporting' is specified, fio
661 will show the final report per-group instead of per-job.
663 thread fio defaults to forking jobs, however if this option is
664 given, fio will use pthread_create(3) to create threads
667 zonesize=siint Divide a file into zones of the specified size. See zoneskip.
669 zoneskip=siint Skip the specified number of bytes when zonesize data has
670 been read. The two zone options can be used to only do
671 io on zones of a file.
673 write_iolog=str Write the issued io patterns to the specified file. See
676 read_iolog=str Open an iolog with the specified file name and replay the
677 io patterns it contains. This can be used to store a
678 workload and replay it sometime later. The iolog given
679 may also be a blktrace binary file, which allows fio
680 to replay a workload captured by blktrace. See blktrace
681 for how to capture such logging data. For blktrace replay,
682 the file needs to be turned into a blkparse binary data
683 file first (blktrace <device> -d file_for_fio.bin).
685 write_bw_log If given, write a bandwidth log of the jobs in this job
686 file. Can be used to store data of the bandwidth of the
687 jobs in their lifetime. The included fio_generate_plots
688 script uses gnuplot to turn these text files into nice
691 write_lat_log Same as write_bw_log, except that this option stores io
692 completion latencies instead.
694 lockmem=siint Pin down the specified amount of memory with mlock(2). Can
695 potentially be used instead of removing memory or booting
696 with less memory to simulate a smaller amount of memory.
698 exec_prerun=str Before running this job, issue the command specified
701 exec_postrun=str After the job completes, issue the command specified
704 ioscheduler=str Attempt to switch the device hosting the file to the specified
705 io scheduler before running.
707 cpuload=int If the job is a CPU cycle eater, attempt to use the specified
708 percentage of CPU cycles.
710 cpuchunks=int If the job is a CPU cycle eater, split the load into
711 cycles of the given time. In milliseconds.
713 disk_util=bool Generate disk utilization statistics, if the platform
714 supports it. Defaults to on.
717 6.0 Interpreting the output
718 ---------------------------
720 fio spits out a lot of output. While running, fio will display the
721 status of the jobs created. An example of that would be:
723 Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
725 The characters inside the square brackets denote the current status of
726 each thread. The possible values (in typical life cycle order) are:
730 P Thread setup, but not started.
732 I Thread initialized, waiting.
733 R Running, doing sequential reads.
734 r Running, doing random reads.
735 W Running, doing sequential writes.
736 w Running, doing random writes.
737 M Running, doing mixed sequential reads/writes.
738 m Running, doing mixed random reads/writes.
739 F Running, currently waiting for fsync()
740 V Running, doing verification of written data.
741 E Thread exited, not reaped by main thread yet.
744 The other values are fairly self explanatory - number of threads
745 currently running and doing io, rate of io since last check (read speed
746 listed first, then write speed), and the estimated completion percentage
747 and time for the running group. It's impossible to estimate runtime of
748 the following groups (if any).
750 When fio is done (or interrupted by ctrl-c), it will show the data for
751 each thread, group of threads, and disks in that order. For each data
752 direction, the output looks like:
754 Client1 (g=0): err= 0:
755 write: io= 32MiB, bw= 666KiB/s, runt= 50320msec
756 slat (msec): min= 0, max= 136, avg= 0.03, stdev= 1.92
757 clat (msec): min= 0, max= 631, avg=48.50, stdev=86.82
758 bw (KiB/s) : min= 0, max= 1196, per=51.00%, avg=664.02, stdev=681.68
759 cpu : usr=1.49%, sys=0.25%, ctx=7969
760 IO depths : 1=0.1%, 2=0.3%, 4=0.5%, 8=99.0%, 16=0.0%, 32=0.0%, >32=0.0%
761 issued r/w: total=0/32768, short=0/0
762 lat (msec): 2=1.6%, 4=0.0%, 10=3.2%, 20=12.8%, 50=38.4%, 100=24.8%,
763 lat (msec): 250=15.2%, 500=0.0%, 750=0.0%, 1000=0.0%, >=2048=0.0%
765 The client number is printed, along with the group id and error of that
766 thread. Below is the io statistics, here for writes. In the order listed,
769 io= Number of megabytes io performed
770 bw= Average bandwidth rate
771 runt= The runtime of that thread
772 slat= Submission latency (avg being the average, stdev being the
773 standard deviation). This is the time it took to submit
774 the io. For sync io, the slat is really the completion
775 latency, since queue/complete is one operation there. This
776 value can be in miliseconds or microseconds, fio will choose
777 the most appropriate base and print that. In the example
778 above, miliseconds is the best scale.
779 clat= Completion latency. Same names as slat, this denotes the
780 time from submission to completion of the io pieces. For
781 sync io, clat will usually be equal (or very close) to 0,
782 as the time from submit to complete is basically just
783 CPU time (io has already been done, see slat explanation).
784 bw= Bandwidth. Same names as the xlat stats, but also includes
785 an approximate percentage of total aggregate bandwidth
786 this thread received in this group. This last value is
787 only really useful if the threads in this group are on the
788 same disk, since they are then competing for disk access.
789 cpu= CPU usage. User and system time, along with the number
790 of context switches this thread went through.
791 IO depths= The distribution of io depths over the job life time. The
792 numbers are divided into powers of 2, so for example the
793 16= entries includes depths up to that value but higher
794 than the previous entry. In other words, it covers the
796 IO issued= The number of read/write requests issued, and how many
798 IO latencies= The distribution of IO completion latencies. This is the
799 time from when IO leaves fio and when it gets completed.
800 The numbers follow the same pattern as the IO depths,
801 meaning that 2=1.6% means that 1.6% of the IO completed
802 within 2 msecs, 20=12.8% means that 12.8% of the IO
803 took more than 10 msecs, but less than (or equal to) 20 msecs.
805 After each client has been listed, the group statistics are printed. They
808 Run status group 0 (all jobs):
809 READ: io=64MiB, aggrb=22178, minb=11355, maxb=11814, mint=2840msec, maxt=2955msec
810 WRITE: io=64MiB, aggrb=1302, minb=666, maxb=669, mint=50093msec, maxt=50320msec
812 For each data direction, it prints:
814 io= Number of megabytes io performed.
815 aggrb= Aggregate bandwidth of threads in this group.
816 minb= The minimum average bandwidth a thread saw.
817 maxb= The maximum average bandwidth a thread saw.
818 mint= The smallest runtime of the threads in that group.
819 maxt= The longest runtime of the threads in that group.
821 And finally, the disk statistics are printed. They will look like this:
823 Disk stats (read/write):
824 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
826 Each value is printed for both reads and writes, with reads first. The
829 ios= Number of ios performed by all groups.
830 merge= Number of merges io the io scheduler.
831 ticks= Number of ticks we kept the disk busy.
832 io_queue= Total time spent in the disk queue.
833 util= The disk utilization. A value of 100% means we kept the disk
834 busy constantly, 50% would be a disk idling half of the time.
840 For scripted usage where you typically want to generate tables or graphs
841 of the results, fio can output the results in a semicolon separated format.
842 The format is one long line of values, such as:
844 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%
845 ;0.0%;0.0%;0.0%;0.0%;0.0%
847 Split up, the format is as follows:
849 jobname, groupid, error
851 KiB IO, bandwidth (KiB/sec), runtime (msec)
852 Submission latency: min, max, mean, deviation
853 Completion latency: min, max, mean, deviation
854 Bw: min, max, aggregate percentage of total, mean, deviation
856 KiB IO, bandwidth (KiB/sec), runtime (msec)
857 Submission latency: min, max, mean, deviation
858 Completion latency: min, max, mean, deviation
859 Bw: min, max, aggregate percentage of total, mean, deviation
860 CPU usage: user, system, context switches
861 IO depths: <=1, 2, 4, 8, 16, 32, >=64
862 IO latencies: <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, >=2000