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, 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 ';', the
112 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, may 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 bool Boolean. Usually parsed as an integer, however only defined for
180 true and false (1 and 0).
181 irange Integer range with postfix. Allows value range to be given, such
182 as 1024-4096. Also see siint.
184 With the above in mind, here follows the complete list of fio job
187 name=str ASCII name of the job. This may be used to override the
188 name printed by fio for this job. Otherwise the job
189 name is used. On the command line this parameter has the
190 special purpose of also signaling the start of a new
193 directory=str Prefix filenames with this directory. Used to places files
194 in a different location than "./".
196 filename=str Fio normally makes up a filename based on the job name,
197 thread number, and file number. If you want to share
198 files between threads in a job or several jobs, specify
199 a filename for each of them to override the default.
201 rw=str Type of io pattern. Accepted values are:
203 read Sequential reads
204 write Sequential writes
205 randwrite Random writes
206 randread Random reads
207 rw Sequential mixed reads and writes
208 randrw Random mixed reads and writes
210 For the mixed io types, the default is to split them 50/50.
211 For certain types of io the result may still be skewed a bit,
212 since the speed may be different.
214 size=siint The total size of file io for this job. This may describe
215 the size of the single file the job uses, or it may be
216 divided between the number of files in the job. If the
217 file already exists, the file size will be adjusted to this
218 size if larger than the current file size. If this parameter
219 is not given and the file exists, the file size will be used.
221 bs=siint The block size used for the io units. Defaults to 4k. Values
222 can be given for both read and writes. If a single siint is
223 given, it will apply to both. If a second siint is specified
224 after a comma, it will apply to writes only. In other words,
225 the format is either bs=read_and_write or bs=read,write.
226 bs=4k,8k will thus use 4k blocks for reads, and 8k blocks
227 for writes. If you only wish to set the write size, you
228 can do so by passing an empty read size - bs=,8k will set
229 8k for writes and leave the read default value.
231 bsrange=irange Instead of giving a single block size, specify a range
232 and fio will mix the issued io block sizes. The issued
233 io unit will always be a multiple of the minimum value
234 given (also see bs_unaligned). Applies to both reads and
235 writes, however a second range can be given after a comma.
238 bs_unaligned If this option is given, any byte size value within bsrange
239 may be used as a block range. This typically wont work with
240 direct IO, as that normally requires sector alignment.
242 nrfiles=int Number of files to use for this job. Defaults to 1.
244 ioengine=str Defines how the job issues io to the file. The following
247 sync Basic read(2) or write(2) io. lseek(2) is
248 used to position the io location.
250 libaio Linux native asynchronous io.
252 posixaio glibc posix asynchronous io.
254 mmap File is memory mapped and data copied
255 to/from using memcpy(3).
257 splice splice(2) is used to transfer the data and
258 vmsplice(2) to transfer data from user
261 sg SCSI generic sg v3 io. May either be
262 synchronous using the SG_IO ioctl, or if
263 the target is an sg character device
264 we use read(2) and write(2) for asynchronous
267 null Doesn't transfer any data, just pretends
268 to. This is mainly used to exercise fio
269 itself and for debugging/testing purposes.
271 iodepth=int This defines how many io units to keep in flight against
272 the file. The default is 1 for each file defined in this
273 job, can be overridden with a larger value for higher
276 direct=bool If value is true, use non-buffered io. This is usually
277 O_DIRECT. Defaults to true.
279 offset=siint Start io at the given offset in the file. The data before
280 the given offset will not be touched. This effectively
281 caps the file size at real_size - offset.
283 fsync=int If writing to a file, issue a sync of the dirty data
284 for every number of blocks given. For example, if you give
285 32 as a parameter, fio will sync the file for every 32
286 writes issued. If fio is using non-buffered io, we may
287 not sync the file. The exception is the sg io engine, which
288 synchronizes the disk cache anyway.
290 overwrite=bool If writing to a file, setup the file first and do overwrites.
292 end_fsync=bool If true, fsync file contents when the job exits.
294 rwmixcycle=int Value in milliseconds describing how often to switch between
295 reads and writes for a mixed workload. The default is
298 rwmixread=int How large a percentage of the mix should be reads.
300 rwmixwrite=int How large a percentage of the mix should be writes. If both
301 rwmixread and rwmixwrite is given and the values do not add
302 up to 100%, the latter of the two will be used to override
305 norandommap Normally fio will cover every block of the file when doing
306 random IO. If this option is given, fio will just get a
307 new random offset without looking at past io history. This
308 means that some blocks may not be read or written, and that
309 some blocks may be read/written more than once. This option
310 is mutually exclusive with verify= for that reason.
312 nice=int Run the job with the given nice value. See man nice(2).
314 prio=int Set the io priority value of this job. Linux limits us to
315 a positive value between 0 and 7, with 0 being the highest.
318 prioclass=int Set the io priority class. See man ionice(1).
320 thinktime=int Stall the job x microseconds after an io has completed before
321 issuing the next. May be used to simulate processing being
322 done by an application. See thinktime_blocks.
325 Only valid if thinktime is set - control how many blocks
326 to issue, before waiting 'thinktime' usecs. If not set,
327 defaults to 1 which will make fio wait 'thinktime' usecs
330 rate=int Cap the bandwidth used by this job to this number of KiB/sec.
332 ratemin=int Tell fio to do whatever it can to maintain at least this
335 ratecycle=int Average bandwidth for 'rate' and 'ratemin' over this number
338 cpumask=int Set the CPU affinity of this job. The parameter given is a
339 bitmask of allowed CPU's the job may run on. See man
340 sched_setaffinity(2).
342 startdelay=int Start this job the specified number of seconds after fio
343 has started. Only useful if the job file contains several
344 jobs, and you want to delay starting some jobs to a certain
347 timeout=int Tell fio to terminate processing after the specified number
348 of seconds. It can be quite hard to determine for how long
349 a specified job will run, so this parameter is handy to
350 cap the total runtime to a given time.
352 invalidate=bool Invalidate the buffer/page cache parts for this file prior
353 to starting io. Defaults to true.
355 sync=bool Use sync io for buffered writes. For the majority of the
356 io engines, this means using O_SYNC.
358 mem=str Fio can use various types of memory as the io unit buffer.
359 The allowed values are:
361 malloc Use memory from malloc(3) as the buffers.
363 shm Use shared memory as the buffers. Allocated
366 shmhuge Same as shm, but use huge pages as backing.
368 mmap Use mmap to allocate buffers. May either be
369 anonymous memory, or can be file backed if
370 a filename is given after the option. The
371 format is mem=mmap:/path/to/file.
373 mmaphuge Use a memory mapped huge file as the buffer
374 backing. Append filename after mmaphuge, ala
375 mem=mmaphuge:/hugetlbfs/file
377 The area allocated is a function of the maximum allowed
378 bs size for the job, multiplied by the io depth given. Note
379 that for shmhuge and mmaphuge to work, the system must have
380 free huge pages allocated. This can normally be checked
381 and set by reading/writing /proc/sys/vm/nr_hugepages on a
382 Linux system. Fio assumes a huge page is 4MiB in size. So
383 to calculate the number of huge pages you need for a given
384 job file, add up the io depth of all jobs (normally one unless
385 iodepth= is used) and multiply by the maximum bs set. Then
386 divide that number by the huge page size. You can see the
387 size of the huge pages in /proc/meminfo. If no huge pages
388 are allocated by having a non-zero number in nr_hugepages,
389 using mmaphuge or shmhuge will fail. Also see hugepage-size.
391 mmaphuge also needs to have hugetlbfs mounted and the file
392 location should point there. So if it's mounted in /huge,
393 you would use mem=mmaphuge:/huge/somefile.
396 Defines the size of a huge page. Must at least be equal
397 to the system setting, see /proc/meminfo. Defaults to 4MiB.
398 Should probably always be a multiple of megabytes, so using
399 hugepage-size=Xm is the preferred way to set this to avoid
400 setting a non-pow-2 bad value.
402 exitall When one job finishes, terminate the rest. The default is
403 to wait for each job to finish, sometimes that is not the
406 bwavgtime=int Average the calculated bandwidth over the given time. Value
407 is specified in milliseconds.
409 create_serialize=bool If true, serialize the file creating for the jobs.
410 This may be handy to avoid interleaving of data
411 files, which may greatly depend on the filesystem
412 used and even the number of processors in the system.
414 create_fsync=bool fsync the data file after creation. This is the
417 unlink=bool Unlink the job files when done. fio defaults to doing this,
418 if it created the file itself.
420 loops=int Run the specified number of iterations of this job. Used
421 to repeat the same workload a given number of times. Defaults
424 verify=str If writing to a file, fio can verify the file contents
425 after each iteration of the job. The allowed values are:
427 md5 Use an md5 sum of the data area and store
428 it in the header of each block.
430 crc32 Use a crc32 sum of the data area and store
431 it in the header of each block.
433 This option can be used for repeated burn-in tests of a
434 system to make sure that the written data is also
437 stonewall Wait for preceeding jobs in the job file to exit, before
438 starting this one. Can be used to insert serialization
439 points in the job file.
441 numjobs=int Create the specified number of clones of this job. May be
442 used to setup a larger number of threads/processes doing
445 thread fio defaults to forking jobs, however if this option is
446 given, fio will use pthread_create(3) to create threads
449 zonesize=siint Divide a file into zones of the specified size. See zoneskip.
451 zoneskip=siint Skip the specified number of bytes when zonesize data has
452 been read. The two zone options can be used to only do
453 io on zones of a file.
455 write_iolog=str Write the issued io patterns to the specified file. See
458 read_iolog=str Open an iolog with the specified file name and replay the
459 io patterns it contains. This can be used to store a
460 workload and replay it sometime later.
462 write_bw_log If given, write a bandwidth log of the jobs in this job
463 file. Can be used to store data of the bandwidth of the
464 jobs in their lifetime. The included fio_generate_plots
465 script uses gnuplot to turn these text files into nice
468 write_lat_log Same as write_bw_log, except that this option stores io
469 completion latencies instead.
471 lockmem=siint Pin down the specified amount of memory with mlock(2). Can
472 potentially be used instead of removing memory or booting
473 with less memory to simulate a smaller amount of memory.
475 exec_prerun=str Before running this job, issue the command specified
478 exec_postrun=str After the job completes, issue the command specified
481 ioscheduler=str Attempt to switch the device hosting the file to the specified
482 io scheduler before running.
484 cpuload=int If the job is a CPU cycle eater, attempt to use the specified
485 percentage of CPU cycles.
487 cpuchunks=int If the job is a CPU cycle eater, split the load into
488 cycles of the given time. In milliseconds.
491 6.0 Interpreting the output
492 ---------------------------
494 fio spits out a lot of output. While running, fio will display the
495 status of the jobs created. An example of that would be:
497 Threads running: 1: [_r] [24.79% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
499 The characters inside the square brackets denote the current status of
500 each thread. The possible values (in typical life cycle order) are:
504 P Thread setup, but not started.
506 I Thread initialized, waiting.
507 R Running, doing sequential reads.
508 r Running, doing random reads.
509 W Running, doing sequential writes.
510 w Running, doing random writes.
511 M Running, doing mixed sequential reads/writes.
512 m Running, doing mixed random reads/writes.
513 F Running, currently waiting for fsync()
514 V Running, doing verification of written data.
515 E Thread exited, not reaped by main thread yet.
518 The other values are fairly self explanatory - number of threads
519 currently running and doing io, rate of io since last check, and the estimated
520 completion percentage and time for the running group. It's impossible to
521 estimate runtime of the following groups (if any).
523 When fio is done (or interrupted by ctrl-c), it will show the data for
524 each thread, group of threads, and disks in that order. For each data
525 direction, the output looks like:
527 Client1 (g=0): err= 0:
528 write: io= 32MiB, bw= 666KiB/s, runt= 50320msec
529 slat (msec): min= 0, max= 136, avg= 0.03, dev= 1.92
530 clat (msec): min= 0, max= 631, avg=48.50, dev=86.82
531 bw (KiB/s) : min= 0, max= 1196, per=51.00%, avg=664.02, dev=681.68
532 cpu : usr=1.49%, sys=0.25%, ctx=7969
534 The client number is printed, along with the group id and error of that
535 thread. Below is the io statistics, here for writes. In the order listed,
538 io= Number of megabytes io performed
539 bw= Average bandwidth rate
540 runt= The runtime of that thread
541 slat= Submission latency (avg being the average, dev being the
542 standard deviation). This is the time it took to submit
543 the io. For sync io, the slat is really the completion
544 latency, since queue/complete is one operation there.
545 clat= Completion latency. Same names as slat, this denotes the
546 time from submission to completion of the io pieces. For
547 sync io, clat will usually be equal (or very close) to 0,
548 as the time from submit to complete is basically just
549 CPU time (io has already been done, see slat explanation).
550 bw= Bandwidth. Same names as the xlat stats, but also includes
551 an approximate percentage of total aggregate bandwidth
552 this thread received in this group. This last value is
553 only really useful if the threads in this group are on the
554 same disk, since they are then competing for disk access.
555 cpu= CPU usage. User and system time, along with the number
556 of context switches this thread went through.
558 After each client has been listed, the group statistics are printed. They
561 Run status group 0 (all jobs):
562 READ: io=64MiB, aggrb=22178, minb=11355, maxb=11814, mint=2840msec, maxt=2955msec
563 WRITE: io=64MiB, aggrb=1302, minb=666, maxb=669, mint=50093msec, maxt=50320msec
565 For each data direction, it prints:
567 io= Number of megabytes io performed.
568 aggrb= Aggregate bandwidth of threads in this group.
569 minb= The minimum average bandwidth a thread saw.
570 maxb= The maximum average bandwidth a thread saw.
571 mint= The smallest runtime of the threads in that group.
572 maxt= The longest runtime of the threads in that group.
574 And finally, the disk statistics are printed. They will look like this:
576 Disk stats (read/write):
577 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
579 Each value is printed for both reads and writes, with reads first. The
582 ios= Number of ios performed by all groups.
583 merge= Number of merges io the io scheduler.
584 ticks= Number of ticks we kept the disk busy.
585 io_queue= Total time spent in the disk queue.
586 util= The disk utilization. A value of 100% means we kept the disk
587 busy constantly, 50% would be a disk idling half of the time.
593 For scripted usage where you typically want to generate tables or graphs
594 of the results, fio can output the results in a comma separated format.
595 The format is one long line of values, such as:
597 client1,0,0,936,331,2894,0,0,0.000000,0.000000,1,170,22.115385,34.290410,16,714,84.252874%,366.500000,566.417819,3496,1237,2894,0,0,0.000000,0.000000,0,246,6.671625,21.436952,0,2534,55.465300%,1406.600000,2008.044216,0.000000%,0.431928%,1109
599 Split up, the format is as follows:
601 jobname, groupid, error
603 KiB IO, bandwidth (KiB/sec), runtime (msec)
604 Submission latency: min, max, mean, deviation
605 Completion latency: min, max, mean, deviation
606 Bw: min, max, aggregate percentage of total, mean, deviation
608 KiB IO, bandwidth (KiB/sec), runtime (msec)
609 Submission latency: min, max, mean, deviation
610 Completion latency: min, max, mean, deviation
611 Bw: min, max, aggregate percentage of total, mean, deviation
612 CPU usage: user, system, context switches