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.
174 siint SI integer. A whole number value, which may contain a postfix
175 describing the base of the number. Accepted postfixes are k/m/g,
176 meaning kilo, mega, and giga. So if you want to specify 4096,
177 you could either write out '4096' or just give 4k. The postfixes
178 signify base 2 values, so 1024 is 1k and 1024k is 1m and so on.
179 If the option accepts an upper and lower range, use a colon ':'
180 or minus '-' to seperate such values. See irange.
181 bool Boolean. Usually parsed as an integer, however only defined for
182 true and false (1 and 0).
183 irange Integer range with postfix. Allows value range to be given, such
184 as 1024-4096. A colon may also be used as the seperator, eg
185 1k:4k. If the option allows two sets of ranges, they can be
186 specified with a ',' or '/' delimiter: 1k-4k/8k-32k. Also see
189 With the above in mind, here follows the complete list of fio job
192 name=str ASCII name of the job. This may be used to override the
193 name printed by fio for this job. Otherwise the job
194 name is used. On the command line this parameter has the
195 special purpose of also signaling the start of a new
198 description=str Text description of the job. Doesn't do anything except
199 dump this text description when this job is run. It's
202 directory=str Prefix filenames with this directory. Used to places files
203 in a different location than "./".
205 filename=str Fio normally makes up a filename based on the job name,
206 thread number, and file number. If you want to share
207 files between threads in a job or several jobs, specify
208 a filename for each of them to override the default. If
209 the ioengine used is 'net', the filename is the host and
210 port to connect to in the format of =host/port. If the
211 ioengine is file based, you can specify a number of files
212 by seperating the names with a ':' colon. So if you wanted
213 a job to open /dev/sda and /dev/sdb as the two working files,
214 you would use filename=/dev/sda:/dev/sdb. '-' is a reserved
215 name, meaning stdin or stdout. Which of the two depends
216 on the read/write direction set.
218 opendir=str Tell fio to recursively add any file it can find in this
219 directory and down the file system tree.
222 rw=str Type of io pattern. Accepted values are:
224 read Sequential reads
225 write Sequential writes
226 randwrite Random writes
227 randread Random reads
228 rw Sequential mixed reads and writes
229 randrw Random mixed reads and writes
231 For the mixed io types, the default is to split them 50/50.
232 For certain types of io the result may still be skewed a bit,
233 since the speed may be different. It is possible to specify
234 a number of IO's to do before getting a new offset - this
235 is only useful for random IO, where fio would normally
236 generate a new random offset for every IO. If you append
237 eg 8 to randread, you would get a new random offset for
238 every 8 IO's. The result would be a seek for only every 8
239 IO's, instead of for every IO. Use rw=randread:8 to specify
242 randrepeat=bool For random IO workloads, seed the generator in a predictable
243 way so that results are repeatable across repetitions.
245 fadvise_hint=bool By default, fio will use fadvise() to advise the kernel
246 on what IO patterns it is likely to issue. Sometimes you
247 want to test specific IO patterns without telling the
248 kernel about it, in which case you can disable this option.
249 If set, fio will use POSIX_FADV_SEQUENTIAL for sequential
250 IO and POSIX_FADV_RANDOM for random IO.
252 size=siint The total size of file io for this job. Fio will run until
253 this many bytes has been transferred, unless runtime is
254 limited by other options (such as 'runtime', for instance).
255 Unless specific nr_files and filesize options are given,
256 fio will divide this size between the available files
257 specified by the job.
259 filesize=siint Individual file sizes. May be a range, in which case fio
260 will select sizes for files at random within the given range
261 and limited to 'size' in total (if that is given). If not
262 given, each created file is the same size.
265 bs=siint The block size used for the io units. Defaults to 4k. Values
266 can be given for both read and writes. If a single siint is
267 given, it will apply to both. If a second siint is specified
268 after a comma, it will apply to writes only. In other words,
269 the format is either bs=read_and_write or bs=read,write.
270 bs=4k,8k will thus use 4k blocks for reads, and 8k blocks
271 for writes. If you only wish to set the write size, you
272 can do so by passing an empty read size - bs=,8k will set
273 8k for writes and leave the read default value.
275 blocksize_range=irange
276 bsrange=irange Instead of giving a single block size, specify a range
277 and fio will mix the issued io block sizes. The issued
278 io unit will always be a multiple of the minimum value
279 given (also see bs_unaligned). Applies to both reads and
280 writes, however a second range can be given after a comma.
284 bs_unaligned If this option is given, any byte size value within bsrange
285 may be used as a block range. This typically wont work with
286 direct IO, as that normally requires sector alignment.
288 zero_buffers If this option is given, fio will init the IO buffers to
289 all zeroes. The default is to fill them with random data.
291 nrfiles=int Number of files to use for this job. Defaults to 1.
293 openfiles=int Number of files to keep open at the same time. Defaults to
294 the same as nrfiles, can be set smaller to limit the number
297 file_service_type=str Defines how fio decides which file from a job to
298 service next. The following types are defined:
300 random Just choose a file at random.
302 roundrobin Round robin over open files. This
305 The string can have a number appended, indicating how
306 often to switch to a new file. So if option random:4 is
307 given, fio will switch to a new random file after 4 ios
310 ioengine=str Defines how the job issues io to the file. The following
313 sync Basic read(2) or write(2) io. lseek(2) is
314 used to position the io location.
316 libaio Linux native asynchronous io.
318 posixaio glibc posix asynchronous io.
320 mmap File is memory mapped and data copied
321 to/from using memcpy(3).
323 splice splice(2) is used to transfer the data and
324 vmsplice(2) to transfer data from user
327 syslet-rw Use the syslet system calls to make
328 regular read/write async.
330 sg SCSI generic sg v3 io. May either be
331 synchronous using the SG_IO ioctl, or if
332 the target is an sg character device
333 we use read(2) and write(2) for asynchronous
336 null Doesn't transfer any data, just pretends
337 to. This is mainly used to exercise fio
338 itself and for debugging/testing purposes.
340 net Transfer over the network to given host:port.
341 'filename' must be set appropriately to
342 filename=host/port regardless of send
343 or receive, if the latter only the port
346 netsplice Like net, but uses splice/vmsplice to
347 map data and send/receive.
349 cpu Doesn't transfer any data, but burns CPU
350 cycles according to the cpuload= and
351 cpucycle= options. Setting cpuload=85
352 will cause that job to do nothing but burn
355 guasi The GUASI IO engine is the Generic Userspace
356 Asyncronous Syscall Interface approach
359 http://www.xmailserver.org/guasi-lib.html
361 for more info on GUASI.
363 external Prefix to specify loading an external
364 IO engine object file. Append the engine
365 filename, eg ioengine=external:/tmp/foo.o
366 to load ioengine foo.o in /tmp.
368 iodepth=int This defines how many io units to keep in flight against
369 the file. The default is 1 for each file defined in this
370 job, can be overridden with a larger value for higher
373 iodepth_batch=int This defines how many pieces of IO to submit at once.
374 It defaults to the same as iodepth, but can be set lower
377 iodepth_low=int The low water mark indicating when to start filling
378 the queue again. Defaults to the same as iodepth, meaning
379 that fio will attempt to keep the queue full at all times.
380 If iodepth is set to eg 16 and iodepth_low is set to 4, then
381 after fio has filled the queue of 16 requests, it will let
382 the depth drain down to 4 before starting to fill it again.
384 direct=bool If value is true, use non-buffered io. This is usually
387 buffered=bool If value is true, use buffered io. This is the opposite
388 of the 'direct' option. Defaults to true.
390 offset=siint Start io at the given offset in the file. The data before
391 the given offset will not be touched. This effectively
392 caps the file size at real_size - offset.
394 fsync=int If writing to a file, issue a sync of the dirty data
395 for every number of blocks given. For example, if you give
396 32 as a parameter, fio will sync the file for every 32
397 writes issued. If fio is using non-buffered io, we may
398 not sync the file. The exception is the sg io engine, which
399 synchronizes the disk cache anyway.
401 overwrite=bool If writing to a file, setup the file first and do overwrites.
403 end_fsync=bool If true, fsync file contents when the job exits.
405 fsync_on_close=bool If true, fio will fsync() a dirty file on close.
406 This differs from end_fsync in that it will happen on every
407 file close, not just at the end of the job.
409 rwmixcycle=int Value in milliseconds describing how often to switch between
410 reads and writes for a mixed workload. The default is
413 rwmixread=int How large a percentage of the mix should be reads.
415 rwmixwrite=int How large a percentage of the mix should be writes. If both
416 rwmixread and rwmixwrite is given and the values do not add
417 up to 100%, the latter of the two will be used to override
420 norandommap Normally fio will cover every block of the file when doing
421 random IO. If this option is given, fio will just get a
422 new random offset without looking at past io history. This
423 means that some blocks may not be read or written, and that
424 some blocks may be read/written more than once. This option
425 is mutually exclusive with verify= for that reason, since
426 fio doesn't track potential block rewrites which may alter
427 the calculated checksum for that block.
429 nice=int Run the job with the given nice value. See man nice(2).
431 prio=int Set the io priority value of this job. Linux limits us to
432 a positive value between 0 and 7, with 0 being the highest.
435 prioclass=int Set the io priority class. See man ionice(1).
437 thinktime=int Stall the job x microseconds after an io has completed before
438 issuing the next. May be used to simulate processing being
439 done by an application. See thinktime_blocks and
443 Only valid if thinktime is set - pretend to spend CPU time
444 doing something with the data received, before falling back
445 to sleeping for the rest of the period specified by
449 Only valid if thinktime is set - control how many blocks
450 to issue, before waiting 'thinktime' usecs. If not set,
451 defaults to 1 which will make fio wait 'thinktime' usecs
454 rate=int Cap the bandwidth used by this job to this number of KiB/sec.
456 ratemin=int Tell fio to do whatever it can to maintain at least this
457 bandwidth. Failing to meet this requirement, will cause
460 rate_iops=int Cap the bandwidth to this number of IOPS. Basically the same
461 as rate, just specified independently of bandwidth. If the
462 job is given a block size range instead of a fixed value,
463 the smallest block size is used as the metric.
465 rate_iops_min=int If fio doesn't meet this rate of IO, it will cause
468 ratecycle=int Average bandwidth for 'rate' and 'ratemin' over this number
471 cpumask=int Set the CPU affinity of this job. The parameter given is a
472 bitmask of allowed CPU's the job may run on. So if you want
473 the allowed CPUs to be 1 and 5, you would pass the decimal
474 value of (1 << 1 | 1 << 5), or 34. See man
475 sched_setaffinity(2). This may not work on all supported
476 operating systems or kernel versions.
478 cpus_allowed=str Controls the same options as cpumask, but it allows a text
479 setting of the permitted CPUs instead. So to use CPUs 1 and
480 5, you would specify cpus_allowed=1,5.
482 startdelay=int Start this job the specified number of seconds after fio
483 has started. Only useful if the job file contains several
484 jobs, and you want to delay starting some jobs to a certain
487 runtime=int Tell fio to terminate processing after the specified number
488 of seconds. It can be quite hard to determine for how long
489 a specified job will run, so this parameter is handy to
490 cap the total runtime to a given time.
492 time_based If set, fio will run for the duration of the runtime
493 specified even if the file(s) are completey read or
494 written. It will simply loop over the same workload
495 as many times as the runtime allows.
497 invalidate=bool Invalidate the buffer/page cache parts for this file prior
498 to starting io. Defaults to true.
500 sync=bool Use sync io for buffered writes. For the majority of the
501 io engines, this means using O_SYNC.
504 mem=str Fio can use various types of memory as the io unit buffer.
505 The allowed values are:
507 malloc Use memory from malloc(3) as the buffers.
509 shm Use shared memory as the buffers. Allocated
512 shmhuge Same as shm, but use huge pages as backing.
514 mmap Use mmap to allocate buffers. May either be
515 anonymous memory, or can be file backed if
516 a filename is given after the option. The
517 format is mem=mmap:/path/to/file.
519 mmaphuge Use a memory mapped huge file as the buffer
520 backing. Append filename after mmaphuge, ala
521 mem=mmaphuge:/hugetlbfs/file
523 The area allocated is a function of the maximum allowed
524 bs size for the job, multiplied by the io depth given. Note
525 that for shmhuge and mmaphuge to work, the system must have
526 free huge pages allocated. This can normally be checked
527 and set by reading/writing /proc/sys/vm/nr_hugepages on a
528 Linux system. Fio assumes a huge page is 4MiB in size. So
529 to calculate the number of huge pages you need for a given
530 job file, add up the io depth of all jobs (normally one unless
531 iodepth= is used) and multiply by the maximum bs set. Then
532 divide that number by the huge page size. You can see the
533 size of the huge pages in /proc/meminfo. If no huge pages
534 are allocated by having a non-zero number in nr_hugepages,
535 using mmaphuge or shmhuge will fail. Also see hugepage-size.
537 mmaphuge also needs to have hugetlbfs mounted and the file
538 location should point there. So if it's mounted in /huge,
539 you would use mem=mmaphuge:/huge/somefile.
542 Defines the size of a huge page. Must at least be equal
543 to the system setting, see /proc/meminfo. Defaults to 4MiB.
544 Should probably always be a multiple of megabytes, so using
545 hugepage-size=Xm is the preferred way to set this to avoid
546 setting a non-pow-2 bad value.
548 exitall When one job finishes, terminate the rest. The default is
549 to wait for each job to finish, sometimes that is not the
552 bwavgtime=int Average the calculated bandwidth over the given time. Value
553 is specified in milliseconds.
555 create_serialize=bool If true, serialize the file creating for the jobs.
556 This may be handy to avoid interleaving of data
557 files, which may greatly depend on the filesystem
558 used and even the number of processors in the system.
560 create_fsync=bool fsync the data file after creation. This is the
563 unlink=bool Unlink the job files when done. Not the default, as repeated
564 runs of that job would then waste time recreating the fileset
567 loops=int Run the specified number of iterations of this job. Used
568 to repeat the same workload a given number of times. Defaults
571 verify=str If writing to a file, fio can verify the file contents
572 after each iteration of the job. The allowed values are:
574 md5 Use an md5 sum of the data area and store
575 it in the header of each block.
577 crc32 Use a crc32 sum of the data area and store
578 it in the header of each block.
580 null Only pretend to verify. Useful for testing
581 internals with ioengine=null, not for much
584 This option can be used for repeated burn-in tests of a
585 system to make sure that the written data is also
588 verifysort=bool If set, fio will sort written verify blocks when it deems
589 it faster to read them back in a sorted manner. This is
590 often the case when overwriting an existing file, since
591 the blocks are already laid out in the file system. You
592 can ignore this option unless doing huge amounts of really
593 fast IO where the red-black tree sorting CPU time becomes
596 stonewall Wait for preceeding jobs in the job file to exit, before
597 starting this one. Can be used to insert serialization
598 points in the job file. A stone wall also implies starting
599 a new reporting group.
601 new_group Start a new reporting group. If this option isn't given,
602 jobs in a file will be part of the same reporting group
603 unless seperated by a stone wall (or if it's a group
604 by itself, with the numjobs option).
606 numjobs=int Create the specified number of clones of this job. May be
607 used to setup a larger number of threads/processes doing
608 the same thing. We regard that grouping of jobs as a
611 group_reporting If 'numjobs' is set, it may be interesting to display
612 statistics for the group as a whole instead of for each
613 individual job. This is especially true of 'numjobs' is
614 large, looking at individual thread/process output quickly
615 becomes unwieldy. If 'group_reporting' is specified, fio
616 will show the final report per-group instead of per-job.
618 thread fio defaults to forking jobs, however if this option is
619 given, fio will use pthread_create(3) to create threads
622 zonesize=siint Divide a file into zones of the specified size. See zoneskip.
624 zoneskip=siint Skip the specified number of bytes when zonesize data has
625 been read. The two zone options can be used to only do
626 io on zones of a file.
628 write_iolog=str Write the issued io patterns to the specified file. See
631 read_iolog=str Open an iolog with the specified file name and replay the
632 io patterns it contains. This can be used to store a
633 workload and replay it sometime later. The iolog given
634 may also be a blktrace binary file, which allows fio
635 to replay a workload captured by blktrace. See blktrace
636 for how to capture such logging data. For blktrace replay,
637 the file needs to be turned into a blkparse binary data
638 file first (blktrace <device> -d file_for_fio.bin).
640 write_bw_log If given, write a bandwidth log of the jobs in this job
641 file. Can be used to store data of the bandwidth of the
642 jobs in their lifetime. The included fio_generate_plots
643 script uses gnuplot to turn these text files into nice
646 write_lat_log Same as write_bw_log, except that this option stores io
647 completion latencies instead.
649 lockmem=siint Pin down the specified amount of memory with mlock(2). Can
650 potentially be used instead of removing memory or booting
651 with less memory to simulate a smaller amount of memory.
653 exec_prerun=str Before running this job, issue the command specified
656 exec_postrun=str After the job completes, issue the command specified
659 ioscheduler=str Attempt to switch the device hosting the file to the specified
660 io scheduler before running.
662 cpuload=int If the job is a CPU cycle eater, attempt to use the specified
663 percentage of CPU cycles.
665 cpuchunks=int If the job is a CPU cycle eater, split the load into
666 cycles of the given time. In milliseconds.
668 disk_util=bool Generate disk utilization statistics, if the platform
669 supports it. Defaults to on.
672 6.0 Interpreting the output
673 ---------------------------
675 fio spits out a lot of output. While running, fio will display the
676 status of the jobs created. An example of that would be:
678 Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
680 The characters inside the square brackets denote the current status of
681 each thread. The possible values (in typical life cycle order) are:
685 P Thread setup, but not started.
687 I Thread initialized, waiting.
688 R Running, doing sequential reads.
689 r Running, doing random reads.
690 W Running, doing sequential writes.
691 w Running, doing random writes.
692 M Running, doing mixed sequential reads/writes.
693 m Running, doing mixed random reads/writes.
694 F Running, currently waiting for fsync()
695 V Running, doing verification of written data.
696 E Thread exited, not reaped by main thread yet.
699 The other values are fairly self explanatory - number of threads
700 currently running and doing io, rate of io since last check (read speed
701 listed first, then write speed), and the estimated completion percentage
702 and time for the running group. It's impossible to estimate runtime of
703 the following groups (if any).
705 When fio is done (or interrupted by ctrl-c), it will show the data for
706 each thread, group of threads, and disks in that order. For each data
707 direction, the output looks like:
709 Client1 (g=0): err= 0:
710 write: io= 32MiB, bw= 666KiB/s, runt= 50320msec
711 slat (msec): min= 0, max= 136, avg= 0.03, stdev= 1.92
712 clat (msec): min= 0, max= 631, avg=48.50, stdev=86.82
713 bw (KiB/s) : min= 0, max= 1196, per=51.00%, avg=664.02, stdev=681.68
714 cpu : usr=1.49%, sys=0.25%, ctx=7969
715 IO depths : 1=0.1%, 2=0.3%, 4=0.5%, 8=99.0%, 16=0.0%, 32=0.0%, >32=0.0%
716 issued r/w: total=0/32768, short=0/0
717 lat (msec): 2=1.6%, 4=0.0%, 10=3.2%, 20=12.8%, 50=38.4%, 100=24.8%,
718 lat (msec): 250=15.2%, 500=0.0%, 750=0.0%, 1000=0.0%, >=2048=0.0%
720 The client number is printed, along with the group id and error of that
721 thread. Below is the io statistics, here for writes. In the order listed,
724 io= Number of megabytes io performed
725 bw= Average bandwidth rate
726 runt= The runtime of that thread
727 slat= Submission latency (avg being the average, stdev being the
728 standard deviation). This is the time it took to submit
729 the io. For sync io, the slat is really the completion
730 latency, since queue/complete is one operation there. This
731 value can be in miliseconds or microseconds, fio will choose
732 the most appropriate base and print that. In the example
733 above, miliseconds is the best scale.
734 clat= Completion latency. Same names as slat, this denotes the
735 time from submission to completion of the io pieces. For
736 sync io, clat will usually be equal (or very close) to 0,
737 as the time from submit to complete is basically just
738 CPU time (io has already been done, see slat explanation).
739 bw= Bandwidth. Same names as the xlat stats, but also includes
740 an approximate percentage of total aggregate bandwidth
741 this thread received in this group. This last value is
742 only really useful if the threads in this group are on the
743 same disk, since they are then competing for disk access.
744 cpu= CPU usage. User and system time, along with the number
745 of context switches this thread went through.
746 IO depths= The distribution of io depths over the job life time. The
747 numbers are divided into powers of 2, so for example the
748 16= entries includes depths up to that value but higher
749 than the previous entry. In other words, it covers the
751 IO issued= The number of read/write requests issued, and how many
753 IO latencies= The distribution of IO completion latencies. This is the
754 time from when IO leaves fio and when it gets completed.
755 The numbers follow the same pattern as the IO depths,
756 meaning that 2=1.6% means that 1.6% of the IO completed
757 within 2 msecs, 20=12.8% means that 12.8% of the IO
758 took more than 10 msecs, but less than (or equal to) 20 msecs.
760 After each client has been listed, the group statistics are printed. They
763 Run status group 0 (all jobs):
764 READ: io=64MiB, aggrb=22178, minb=11355, maxb=11814, mint=2840msec, maxt=2955msec
765 WRITE: io=64MiB, aggrb=1302, minb=666, maxb=669, mint=50093msec, maxt=50320msec
767 For each data direction, it prints:
769 io= Number of megabytes io performed.
770 aggrb= Aggregate bandwidth of threads in this group.
771 minb= The minimum average bandwidth a thread saw.
772 maxb= The maximum average bandwidth a thread saw.
773 mint= The smallest runtime of the threads in that group.
774 maxt= The longest runtime of the threads in that group.
776 And finally, the disk statistics are printed. They will look like this:
778 Disk stats (read/write):
779 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
781 Each value is printed for both reads and writes, with reads first. The
784 ios= Number of ios performed by all groups.
785 merge= Number of merges io the io scheduler.
786 ticks= Number of ticks we kept the disk busy.
787 io_queue= Total time spent in the disk queue.
788 util= The disk utilization. A value of 100% means we kept the disk
789 busy constantly, 50% would be a disk idling half of the time.
795 For scripted usage where you typically want to generate tables or graphs
796 of the results, fio can output the results in a semicolon separated format.
797 The format is one long line of values, such as:
799 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%
800 ;0.0%;0.0%;0.0%;0.0%;0.0%
802 Split up, the format is as follows:
804 jobname, groupid, error
806 KiB IO, bandwidth (KiB/sec), runtime (msec)
807 Submission latency: min, max, mean, deviation
808 Completion latency: min, max, mean, deviation
809 Bw: min, max, aggregate percentage of total, mean, deviation
811 KiB IO, bandwidth (KiB/sec), runtime (msec)
812 Submission latency: min, max, mean, deviation
813 Completion latency: min, max, mean, deviation
814 Bw: min, max, aggregate percentage of total, mean, deviation
815 CPU usage: user, system, context switches
816 IO depths: <=1, 2, 4, 8, 16, 32, >=64
817 IO latencies: <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, >=2000