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 queueing
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.
91 fio does not need to run as root, except if the files or devices specified
92 in the job section requires that. Some other options may also be restricted,
93 such as memory locking, io scheduler switching, and descreasing the nice value.
98 As previously described, fio accepts one or more job files describing
99 what it is supposed to do. The job file format is the classic ini file,
100 where the names enclosed in [] brackets define the job name. You are free
101 to use any ascii name you want, except 'global' which has special meaning.
102 A global section sets defaults for the jobs described in that file. A job
103 may override a global section parameter, and a job file may even have
104 several global sections if so desired. A job is only affected by a global
105 section residing above it. If the first character in a line is a ';', the
106 entire line is discarded as a comment.
108 So lets look at a really simple job file that define to threads, each
109 randomly reading from a 128MiB file.
111 ; -- start job file --
122 As you can see, the job file sections themselves are empty as all the
123 described parameters are shared. As no filename= option is given, fio
124 makes up a filename for each of the jobs as it sees fit.
126 Lets look at an example that have a number of processes writing randomly
129 ; -- start job file --
141 Here we have no global section, as we only have one job defined anyway.
142 We want to use async io here, with a depth of 4 for each file. We also
143 increased the buffer size used to 32KiB and define numjobs to 4 to
144 fork 4 identical jobs. The result is 4 processes each randomly writing
145 to their own 64MiB file. Instead of using the above job file, you could
146 have given the parameters on the command line. For this case, you would
149 $ fio --name=random-writers --ioengine=libaio --iodepth=4 --rw=randwrite --bs=32k --direct=0 --size=64m --numjobs=4
151 fio ships with a few example job files, you can also look there for
155 5.0 Detailed list of parameters
156 -------------------------------
158 This section describes in details each parameter associated with a job.
159 Some parameters take an option of a given type, such as an integer or
160 a string. The following types are used:
162 str String. This is a sequence of alpha characters.
163 int Integer. A whole number value, may be negative.
164 siint SI integer. A whole number value, which may contain a postfix
165 describing the base of the number. Accepted postfixes are k/m/g,
166 meaning kilo, mega, and giga. So if you want to specifiy 4096,
167 you could either write out '4096' or just give 4k. The postfixes
168 signify base 2 values, so 1024 is 1k and 1024k is 1m and so on.
169 bool Boolean. Usually parsed as an integer, however only defined for
170 true and false (1 and 0).
171 irange Integer range with postfix. Allows value range to be given, such
172 as 1024-4096. Also see siint.
174 With the above in mind, here follows the complete list of fio job
177 name=str ASCII name of the job. This may be used to override the
178 name printed by fio for this job. Otherwise the job
181 directory=str Prefix filenames with this directory. Used to places files
182 in a different location than "./".
184 filename=str Fio normally makes up a filename based on the job name,
185 thread number, and file number. If you want to share
186 files between threads in a job or several jobs, specify
187 a filename for each of them to override the default.
189 rw=str Type of io pattern. Accepted values are:
191 read Sequential reads
192 write Sequential writes
193 randwrite Random writes
194 randread Random reads
195 rw Sequential mixed reads and writes
196 randrw Random mixed reads and writes
198 For the mixed io types, the default is to split them 50/50.
199 For certain types of io the result may still be skewed a bit,
200 since the speed may be different.
202 size=siint The total size of file io for this job. This may describe
203 the size of the single file the job uses, or it may be
204 divided between the number of files in the job. If the
205 file already exists, the file size will be adjusted to this
206 size if larger than the current file size. If this parameter
207 is not given and the file exists, the file size will be used.
209 bs=siint The block size used for the io units. Defaults to 4k.
211 bsrange=irange Instead of giving a single block size, specify a range
212 and fio will mix the issued io block sizes. The issued
213 io unit will always be a multiple of the minimum value
216 nrfiles=int Number of files to use for this job. Defaults to 1.
218 ioengine=str Defines how the job issues io to the file. The following
221 sync Basic read(2) or write(2) io. lseek(2) is
222 used to position the io location.
224 libaio Linux native asynchronous io.
226 posixaio glibc posix asynchronous io.
228 mmap File is memory mapped and data copied
229 to/from using memcpy(3).
231 splice splice(2) is used to transfer the data and
232 vmsplice(2) to transfer data from user
235 sg SCSI generic sg v3 io. May either be
236 syncrhonous using the SG_IO ioctl, or if
237 the target is an sg character device
238 we use read(2) and write(2) for asynchronous
241 iodepth=int This defines how many io units to keep in flight against
242 the file. The default is 1 for each file defined in this
243 job, can be overridden with a larger value for higher
246 direct=bool If value is true, use non-buffered io. This is usually
247 O_DIRECT. Defaults to true.
249 offset=siint Start io at the given offset in the file. The data before
250 the given offset will not be touched. This effectively
251 caps the file size at real_size - offset.
253 fsync=int If writing to a file, issue a sync of the dirty data
254 for every number of blocks given. For example, if you give
255 32 as a parameter, fio will sync the file for every 32
256 writes issued. If fio is using non-buffered io, we may
257 not sync the file. The exception is the sg io engine, which
258 syncronizes the disk cache anyway.
260 overwrite=bool If writing to a file, setup the file first and do overwrites.
262 end_fsync=bool If true, fsync file contents when the job exits.
264 rwmixcycle=int Value in miliseconds describing how often to switch between
265 reads and writes for a mixed workload. The default is
268 rwmixread=int How large a percentage of the mix should be reads.
270 rwmixwrite=int How large a percentage of the mix should be writes. If both
271 rwmixread and rwmixwrite is given and the values do not add
272 up to 100%, the latter of the two will be used to override
275 nice=int Run the job with the given nice value. See man nice(2).
277 prio=int Set the io priority value of this job. Linux limits us to
278 a positive value between 0 and 7, with 0 being the highest.
281 prioclass=int Set the io priority class. See man ionice(1).
283 thinktime=int Stall the job x microseconds after an io has completed before
284 issuing the next. May be used to simulate processing being
285 done by an application.
287 rate=int Cap the bandwidth used by this job to this number of KiB/sec.
289 ratemin=int Tell fio to do whatever it can to maintain at least this
292 ratecycle=int Average bandwidth for 'rate' and 'ratemin' over this number
295 cpumask=int Set the CPU affinity of this job. The parameter given is a
296 bitmask of allowed CPU's the job may run on. See man
297 sched_setaffinity(2).
299 startdelay=int Start this job the specified number of seconds after fio
300 has started. Only useful if the job file contains several
301 jobs, and you want to delay starting some jobs to a certain
304 timeout=int Tell fio to terminate processing after the specified number
305 of seconds. It can be quite hard to determine for how long
306 a specified job will run, so this parameter is handy to
307 cap the total runtime to a given time.
309 invalidate=bool Invalidate the buffer/page cache parts for this file prior
310 to starting io. Defaults to true.
312 sync=bool Use sync io for buffered writes. For the majority of the
313 io engines, this means using O_SYNC.
315 mem=str Fio can use various types of memory as the io unit buffer.
316 The allowed values are:
318 malloc Use memory from malloc(3) as the buffers.
320 shm Use shared memory as the buffers. Allocated
323 mmap Use anonymous memory maps as the buffers.
324 Allocated through mmap(2).
326 The area allocated is a function of the maximum allowed
327 bs size for the job, multiplied by the io depth given.
329 exitall When one job finishes, terminate the rest. The default is
330 to wait for each job to finish, sometimes that is not the
333 bwavgtime=int Average the calculated bandwidth over the given time. Value
334 is specified in miliseconds.
336 create_serialize=bool If true, serialize the file creating for the jobs.
337 This may be handy to avoid interleaving of data
338 files, which may greatly depend on the filesystem
339 used and even the number of processors in the system.
341 create_fsync=bool fsync the data file after creation. This is the
344 unlink Unlink the job files when done. fio defaults to doing this,
345 if it created the file itself.
347 loops=int Run the specified number of iterations of this job. Used
348 to repeat the same workload a given number of times. Defaults
351 verify=str If writing to a file, fio can verify the file contents
352 after each iteration of the job. The allowed values are:
354 md5 Use an md5 sum of the data area and store
355 it in the header of each block.
357 crc32 Use a crc32 sum of the data area and store
358 it in the header of each block.
360 This option can be used for repeated burnin tests of a
361 system to make sure that the written data is also
364 stonewall Wait for preceeding jobs in the job file to exit, before
365 starting this one. Can be used to insert serialization
366 points in the job file.
368 numjobs=int Create the specified number of clones of this job. May be
369 used to setup a larger number of threads/processes doing
372 thread fio defaults to forking jobs, however if this option is
373 given, fio will use pthread_create(3) to create threads
376 zonesize=siint Divide a file into zones of the specified size. See zoneskip.
378 zoneskip=siint Skip the specified number of bytes when zonesize data has
379 been read. The two zone options can be used to only do
380 io on zones of a file.
382 write_iolog=str Write the issued io patterns to the specified file. See
385 read_iolog=str Open an iolog with the specified file name and replay the
386 io patterns it contains. This can be used to store a
387 workload and replay it sometime later.
389 write_bw_log If given, write a bandwidth log of the jobs in this job
390 file. Can be used to store data of the bandwidth of the
391 jobs in their lifetime. The included fio_generate_plots
392 script uses gnuplot to turn these text files into nice
395 write_lat_log Same as write_bw_log, except that this option stores io
396 completion latencies instead.
398 lockmem=siint Pin down the specified amount of memory with mlock(2). Can
399 potentially be used instead of removing memory or booting
400 with less memory to simulate a smaller amount of memory.
402 exec_prerun=str Before running this job, issue the command specified
405 exec_postrun=str After the job completes, issue the command specified
408 ioscheduler=str Attempt to switch the device hosting the file to the specified
409 io scheduler before running.
411 cpuload=int If the job is a CPU cycle eater, attempt to use the specified
412 percentage of CPU cycles.
414 cpuchunks=int If the job is a CPU cycle eater, split the load into
415 cycles of the given time. In miliseconds.
418 6.0 Interpreting the output
419 ---------------------------
421 fio spits out a lot of output. While running, fio will display the
422 status of the jobs created. An example of that would be:
424 Threads running: 1: [_r] [24.79% done] [eta 00h:01m:31s]
426 The characters inside the square brackets denote the current status of
427 each thread. The possible values (in typical life cycle order) are:
431 P Thread setup, but not started.
433 I Thread initialized, waiting.
434 R Running, doing sequential reads.
435 r Running, doing random reads.
436 W Running, doing sequential writes.
437 w Running, doing random writes.
438 M Running, doing mixed sequential reads/writes.
439 m Running, doing mixed random reads/writes.
440 F Running, currently waiting for fsync()
441 V Running, doing verification of written data.
442 E Thread exited, not reaped by main thread yet.
445 The other values are fairly self explanatory - number of threads
446 currently running and doing io, and the estimated completion percentage
447 and time for the running group. It's impossible to estimate runtime
448 of the following groups (if any).
450 When fio is done (or interrupted by ctrl-c), it will show the data for
451 each thread, group of threads, and disks in that order. For each data
452 direction, the output looks like:
454 Client1 (g=0): err= 0:
455 write: io= 32MiB, bw= 666KiB/s, runt= 50320msec
456 slat (msec): min= 0, max= 136, avg= 0.03, dev= 1.92
457 clat (msec): min= 0, max= 631, avg=48.50, dev=86.82
458 bw (KiB/s) : min= 0, max= 1196, per=51.00%, avg=664.02, dev=681.68
459 cpu : usr=1.49%, sys=0.25%, ctx=7969
461 The client number is printed, along with the group id and error of that
462 thread. Below is the io statistics, here for writes. In the order listed,
465 io= Number of megabytes io performed
466 bw= Average bandwidth rate
467 runt= The runtime of that thread
468 slat= Submission latency (avg being the average, dev being the
469 standard deviation). This is the time it took to submit
470 the io. For sync io, the slat is really the completion
471 latency, since queue/complete is one operation there.
472 clat= Completion latency. Same names as slat, this denotes the
473 time from submission to completion of the io pieces. For
474 sync io, clat will usually be equal (or very close) to 0,
475 as the time from submit to complete is basically just
476 CPU time (io has already been done, see slat explanation).
477 bw= Bandwidth. Same names as the xlat stats, but also includes
478 an approximate percentage of total aggregate bandwidth
479 this thread received in this group. This last value is
480 only really useful if the threads in this group are on the
481 same disk, since they are then competing for disk access.
482 cpu= CPU usage. User and system time, along with the number
483 of context switches this thread went through.
485 After each client has been listed, the group statistics are printed. They
488 Run status group 0 (all jobs):
489 READ: io=64MiB, aggrb=22178, minb=11355, maxb=11814, mint=2840msec, maxt=2955msec
490 WRITE: io=64MiB, aggrb=1302, minb=666, maxb=669, mint=50093msec, maxt=50320msec
492 For each data direction, it prints:
494 io= Number of megabytes io performed.
495 aggrb= Aggregate bandwidth of threads in this group.
496 minb= The minimum average bandwidth a thread saw.
497 maxb= The maximum average bandwidth a thread saw.
498 mint= The smallest runtime of the threads in that group.
499 maxt= The longest runtime of the threads in that group.
501 And finally, the disk statistics are printed. They will look like this:
503 Disk stats (read/write):
504 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
506 Each value is printed for both reads and writes, with reads first. The
509 ios= Number of ios performed by all groups.
510 merge= Number of merges io the io scheduler.
511 ticks= Number of ticks we kept the disk busy.
512 io_queue= Total time spent in the disk queue.
513 util= The disk utilization. A value of 100% means we kept the disk
514 busy constantly, 50% would be a disk idling half of the time.
520 For scripted usage where you typically want to generate tables or graphs
521 of the results, fio can output the results in a comma seperated format.
522 The format is one long line of values, such as:
524 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
526 Split up, the format is as follows:
528 jobname, groupid, error
530 KiB IO, bandwidth (KiB/sec), runtime (msec)
531 Submission latency: min, max, mean, deviation
532 Completion latency: min, max, mean, deviation
533 Bw: min, max, aggreate percentage of total, mean, deviation
535 KiB IO, bandwidth (KiB/sec), runtime (msec)
536 Submission latency: min, max, mean, deviation
537 Completion latency: min, max, mean, deviation
538 Bw: min, max, aggreate percentage of total, mean, deviation
539 CPU usage: user, system, context switches
projects / fio.git / blob