[PATCH] Update irange documentation
[fio.git] / HOWTO
CommitLineData
71bfa161
JA
1Table of contents
2-----------------
3
41. Overview
52. How fio works
63. Running fio
74. Job file format
85. Detailed list of parameters
96. Normal output
107. Terse output
11
12
131.0 Overview and history
14------------------------
15fio was originally written to save me the hassle of writing special test
16case programs when I wanted to test a specific workload, either for
17performance reasons or to find/reproduce a bug. The process of writing
18such a test app can be tiresome, especially if you have to do it often.
19Hence I needed a tool that would be able to simulate a given io workload
20without resorting to writing a tailored test case again and again.
21
22A test work load is difficult to define, though. There can be any number
23of processes or threads involved, and they can each be using their own
24way of generating io. You could have someone dirtying large amounts of
25memory in an memory mapped file, or maybe several threads issuing
26reads using asynchronous io. fio needed to be flexible enough to
27simulate both of these cases, and many more.
28
292.0 How fio works
30-----------------
31The first step in getting fio to simulate a desired io workload, is
32writing a job file describing that specific setup. A job file may contain
33any number of threads and/or files - the typical contents of the job file
34is a global section defining shared parameters, and one or more job
35sections describing the jobs involved. When run, fio parses this file
36and sets everything up as described. If we break down a job from top to
37bottom, it contains the following basic parameters:
38
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.
43
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
46 block sizes.
47
48 IO size How much data are we going to be reading/writing.
49
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
53 SG (SCSI generic sg).
54
6c219763 55 IO depth If the io engine is async, how large a queuing
71bfa161
JA
56 depth do we want to maintain?
57
58 IO type Should we be doing buffered io, or direct/raw io?
59
60 Num files How many files are we spreading the workload over.
61
62 Num threads How many threads or processes should we spread
63 this workload over.
64
65The above are the basic parameters defined for a workload, in addition
66there's a multitude of parameters that modify other aspects of how this
67job behaves.
68
69
703.0 Running fio
71---------------
72See the README file for command line parameters, there are only a few
73of them.
74
75Running fio is normally the easiest part - you just give it the job file
76(or job files) as parameters:
77
78$ fio job_file
79
80and it will start doing what the job_file tells it to do. You can give
81more than one job file on the command line, fio will serialize the running
82of those files. Internally that is the same as using the 'stonewall'
83parameter described the the parameter section.
84
b4692828
JA
85If the job file contains only one job, you may as well just give the
86parameters on the command line. The command line parameters are identical
87to the job parameters, with a few extra that control global parameters
88(see README). For example, for the job file parameter iodepth=2, the
c2b1e753
JA
89mirror command line option would be --iodepth 2 or --iodepth=2. You can
90also 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.
92Command line entries following a --name entry will apply to that job,
93until there are no more entries or a new --name entry is seen. This is
94similar to the job file options, where each option applies to the current
95job until a new [] job entry is seen.
b4692828 96
71bfa161
JA
97fio does not need to run as root, except if the files or devices specified
98in the job section requires that. Some other options may also be restricted,
6c219763 99such as memory locking, io scheduler switching, and decreasing the nice value.
71bfa161
JA
100
101
1024.0 Job file format
103-------------------
104As previously described, fio accepts one or more job files describing
105what it is supposed to do. The job file format is the classic ini file,
106where the names enclosed in [] brackets define the job name. You are free
107to use any ascii name you want, except 'global' which has special meaning.
108A global section sets defaults for the jobs described in that file. A job
109may override a global section parameter, and a job file may even have
110several global sections if so desired. A job is only affected by a global
111section residing above it. If the first character in a line is a ';', the
112entire line is discarded as a comment.
113
114So lets look at a really simple job file that define to threads, each
115randomly reading from a 128MiB file.
116
117; -- start job file --
118[global]
119rw=randread
120size=128m
121
122[job1]
123
124[job2]
125
126; -- end job file --
127
128As you can see, the job file sections themselves are empty as all the
129described parameters are shared. As no filename= option is given, fio
c2b1e753
JA
130makes up a filename for each of the jobs as it sees fit. On the command
131line, this job would look as follows:
132
133$ fio --name=global --rw=randread --size=128m --name=job1 --name=job2
134
71bfa161
JA
135
136Lets look at an example that have a number of processes writing randomly
137to files.
138
139; -- start job file --
140[random-writers]
141ioengine=libaio
142iodepth=4
143rw=randwrite
144bs=32k
145direct=0
146size=64m
147numjobs=4
148
149; -- end job file --
150
151Here we have no global section, as we only have one job defined anyway.
152We want to use async io here, with a depth of 4 for each file. We also
153increased the buffer size used to 32KiB and define numjobs to 4 to
154fork 4 identical jobs. The result is 4 processes each randomly writing
b4692828
JA
155to their own 64MiB file. Instead of using the above job file, you could
156have given the parameters on the command line. For this case, you would
157specify:
158
159$ fio --name=random-writers --ioengine=libaio --iodepth=4 --rw=randwrite --bs=32k --direct=0 --size=64m --numjobs=4
71bfa161
JA
160
161fio ships with a few example job files, you can also look there for
162inspiration.
163
164
1655.0 Detailed list of parameters
166-------------------------------
167
168This section describes in details each parameter associated with a job.
169Some parameters take an option of a given type, such as an integer or
170a string. The following types are used:
171
172str String. This is a sequence of alpha characters.
173int Integer. A whole number value, may be negative.
174siint SI integer. A whole number value, which may contain a postfix
175 describing the base of the number. Accepted postfixes are k/m/g,
6c219763 176 meaning kilo, mega, and giga. So if you want to specify 4096,
71bfa161
JA
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.
179bool Boolean. Usually parsed as an integer, however only defined for
180 true and false (1 and 0).
181irange Integer range with postfix. Allows value range to be given, such
0c9baf91
JA
182 as 1024-4096. A colon may also be used as the seperator, eg
183 1k:4k. If the option allows two sets of ranges, they can be
184 specified with a ',' or '/' delimiter: 1k-4k/8k-32k. Also see
185 siint.
71bfa161
JA
186
187With the above in mind, here follows the complete list of fio job
188parameters.
189
190name=str ASCII name of the job. This may be used to override the
191 name printed by fio for this job. Otherwise the job
c2b1e753 192 name is used. On the command line this parameter has the
6c219763 193 special purpose of also signaling the start of a new
c2b1e753 194 job.
71bfa161
JA
195
196directory=str Prefix filenames with this directory. Used to places files
197 in a different location than "./".
198
199filename=str Fio normally makes up a filename based on the job name,
200 thread number, and file number. If you want to share
201 files between threads in a job or several jobs, specify
202 a filename for each of them to override the default.
203
204rw=str Type of io pattern. Accepted values are:
205
206 read Sequential reads
207 write Sequential writes
208 randwrite Random writes
209 randread Random reads
210 rw Sequential mixed reads and writes
211 randrw Random mixed reads and writes
212
213 For the mixed io types, the default is to split them 50/50.
214 For certain types of io the result may still be skewed a bit,
215 since the speed may be different.
216
ee738499
JA
217randrepeat=bool For random IO workloads, seed the generator in a predictable
218 way so that results are repeatable across repetitions.
219
71bfa161
JA
220size=siint The total size of file io for this job. This may describe
221 the size of the single file the job uses, or it may be
222 divided between the number of files in the job. If the
223 file already exists, the file size will be adjusted to this
224 size if larger than the current file size. If this parameter
225 is not given and the file exists, the file size will be used.
226
f90eff5a
JA
227bs=siint The block size used for the io units. Defaults to 4k. Values
228 can be given for both read and writes. If a single siint is
229 given, it will apply to both. If a second siint is specified
230 after a comma, it will apply to writes only. In other words,
231 the format is either bs=read_and_write or bs=read,write.
232 bs=4k,8k will thus use 4k blocks for reads, and 8k blocks
787f7e95
JA
233 for writes. If you only wish to set the write size, you
234 can do so by passing an empty read size - bs=,8k will set
235 8k for writes and leave the read default value.
a00735e6 236
71bfa161
JA
237bsrange=irange Instead of giving a single block size, specify a range
238 and fio will mix the issued io block sizes. The issued
239 io unit will always be a multiple of the minimum value
f90eff5a
JA
240 given (also see bs_unaligned). Applies to both reads and
241 writes, however a second range can be given after a comma.
242 See bs=.
a00735e6 243
690adba3
JA
244bs_unaligned If this option is given, any byte size value within bsrange
245 may be used as a block range. This typically wont work with
246 direct IO, as that normally requires sector alignment.
71bfa161
JA
247
248nrfiles=int Number of files to use for this job. Defaults to 1.
249
250ioengine=str Defines how the job issues io to the file. The following
251 types are defined:
252
253 sync Basic read(2) or write(2) io. lseek(2) is
254 used to position the io location.
255
256 libaio Linux native asynchronous io.
257
258 posixaio glibc posix asynchronous io.
259
260 mmap File is memory mapped and data copied
261 to/from using memcpy(3).
262
263 splice splice(2) is used to transfer the data and
264 vmsplice(2) to transfer data from user
265 space to the kernel.
266
267 sg SCSI generic sg v3 io. May either be
6c219763 268 synchronous using the SG_IO ioctl, or if
71bfa161
JA
269 the target is an sg character device
270 we use read(2) and write(2) for asynchronous
271 io.
272
a94ea28b
JA
273 null Doesn't transfer any data, just pretends
274 to. This is mainly used to exercise fio
275 itself and for debugging/testing purposes.
276
71bfa161
JA
277iodepth=int This defines how many io units to keep in flight against
278 the file. The default is 1 for each file defined in this
279 job, can be overridden with a larger value for higher
280 concurrency.
281
282direct=bool If value is true, use non-buffered io. This is usually
76a43db4
JA
283 O_DIRECT.
284
285buffered=bool If value is true, use buffered io. This is the opposite
286 of the 'direct' option. Defaults to true.
71bfa161
JA
287
288offset=siint Start io at the given offset in the file. The data before
289 the given offset will not be touched. This effectively
290 caps the file size at real_size - offset.
291
292fsync=int If writing to a file, issue a sync of the dirty data
293 for every number of blocks given. For example, if you give
294 32 as a parameter, fio will sync the file for every 32
295 writes issued. If fio is using non-buffered io, we may
296 not sync the file. The exception is the sg io engine, which
6c219763 297 synchronizes the disk cache anyway.
71bfa161
JA
298
299overwrite=bool If writing to a file, setup the file first and do overwrites.
300
301end_fsync=bool If true, fsync file contents when the job exits.
302
6c219763 303rwmixcycle=int Value in milliseconds describing how often to switch between
71bfa161
JA
304 reads and writes for a mixed workload. The default is
305 500 msecs.
306
307rwmixread=int How large a percentage of the mix should be reads.
308
309rwmixwrite=int How large a percentage of the mix should be writes. If both
310 rwmixread and rwmixwrite is given and the values do not add
311 up to 100%, the latter of the two will be used to override
312 the first.
313
bb8895e0
JA
314norandommap Normally fio will cover every block of the file when doing
315 random IO. If this option is given, fio will just get a
316 new random offset without looking at past io history. This
317 means that some blocks may not be read or written, and that
318 some blocks may be read/written more than once. This option
319 is mutually exclusive with verify= for that reason.
320
71bfa161
JA
321nice=int Run the job with the given nice value. See man nice(2).
322
323prio=int Set the io priority value of this job. Linux limits us to
324 a positive value between 0 and 7, with 0 being the highest.
325 See man ionice(1).
326
327prioclass=int Set the io priority class. See man ionice(1).
328
329thinktime=int Stall the job x microseconds after an io has completed before
330 issuing the next. May be used to simulate processing being
9c1f7434
JA
331 done by an application. See thinktime_blocks.
332
333thinktime_blocks
334 Only valid if thinktime is set - control how many blocks
335 to issue, before waiting 'thinktime' usecs. If not set,
336 defaults to 1 which will make fio wait 'thinktime' usecs
337 after every block.
71bfa161
JA
338
339rate=int Cap the bandwidth used by this job to this number of KiB/sec.
340
341ratemin=int Tell fio to do whatever it can to maintain at least this
342 bandwidth.
343
344ratecycle=int Average bandwidth for 'rate' and 'ratemin' over this number
6c219763 345 of milliseconds.
71bfa161
JA
346
347cpumask=int Set the CPU affinity of this job. The parameter given is a
348 bitmask of allowed CPU's the job may run on. See man
349 sched_setaffinity(2).
350
351startdelay=int Start this job the specified number of seconds after fio
352 has started. Only useful if the job file contains several
353 jobs, and you want to delay starting some jobs to a certain
354 time.
355
03b74b3e 356runtime=int Tell fio to terminate processing after the specified number
71bfa161
JA
357 of seconds. It can be quite hard to determine for how long
358 a specified job will run, so this parameter is handy to
359 cap the total runtime to a given time.
360
361invalidate=bool Invalidate the buffer/page cache parts for this file prior
362 to starting io. Defaults to true.
363
364sync=bool Use sync io for buffered writes. For the majority of the
365 io engines, this means using O_SYNC.
366
367mem=str Fio can use various types of memory as the io unit buffer.
368 The allowed values are:
369
370 malloc Use memory from malloc(3) as the buffers.
371
372 shm Use shared memory as the buffers. Allocated
373 through shmget(2).
374
74b025b0
JA
375 shmhuge Same as shm, but use huge pages as backing.
376
313cb206
JA
377 mmap Use mmap to allocate buffers. May either be
378 anonymous memory, or can be file backed if
379 a filename is given after the option. The
380 format is mem=mmap:/path/to/file.
71bfa161 381
d0bdaf49
JA
382 mmaphuge Use a memory mapped huge file as the buffer
383 backing. Append filename after mmaphuge, ala
384 mem=mmaphuge:/hugetlbfs/file
385
71bfa161 386 The area allocated is a function of the maximum allowed
5394ae5f
JA
387 bs size for the job, multiplied by the io depth given. Note
388 that for shmhuge and mmaphuge to work, the system must have
389 free huge pages allocated. This can normally be checked
390 and set by reading/writing /proc/sys/vm/nr_hugepages on a
391 Linux system. Fio assumes a huge page is 4MiB in size. So
392 to calculate the number of huge pages you need for a given
393 job file, add up the io depth of all jobs (normally one unless
394 iodepth= is used) and multiply by the maximum bs set. Then
395 divide that number by the huge page size. You can see the
396 size of the huge pages in /proc/meminfo. If no huge pages
397 are allocated by having a non-zero number in nr_hugepages,
56bb17f2 398 using mmaphuge or shmhuge will fail. Also see hugepage-size.
5394ae5f
JA
399
400 mmaphuge also needs to have hugetlbfs mounted and the file
401 location should point there. So if it's mounted in /huge,
402 you would use mem=mmaphuge:/huge/somefile.
71bfa161 403
56bb17f2
JA
404hugepage-size=siint
405 Defines the size of a huge page. Must at least be equal
406 to the system setting, see /proc/meminfo. Defaults to 4MiB.
c51074e7
JA
407 Should probably always be a multiple of megabytes, so using
408 hugepage-size=Xm is the preferred way to set this to avoid
409 setting a non-pow-2 bad value.
56bb17f2 410
71bfa161
JA
411exitall When one job finishes, terminate the rest. The default is
412 to wait for each job to finish, sometimes that is not the
413 desired action.
414
415bwavgtime=int Average the calculated bandwidth over the given time. Value
6c219763 416 is specified in milliseconds.
71bfa161
JA
417
418create_serialize=bool If true, serialize the file creating for the jobs.
419 This may be handy to avoid interleaving of data
420 files, which may greatly depend on the filesystem
421 used and even the number of processors in the system.
422
423create_fsync=bool fsync the data file after creation. This is the
424 default.
425
8aeebd55 426unlink=bool Unlink the job files when done. fio defaults to doing this,
71bfa161
JA
427 if it created the file itself.
428
429loops=int Run the specified number of iterations of this job. Used
430 to repeat the same workload a given number of times. Defaults
431 to 1.
432
433verify=str If writing to a file, fio can verify the file contents
434 after each iteration of the job. The allowed values are:
435
436 md5 Use an md5 sum of the data area and store
437 it in the header of each block.
438
439 crc32 Use a crc32 sum of the data area and store
440 it in the header of each block.
441
6c219763 442 This option can be used for repeated burn-in tests of a
71bfa161
JA
443 system to make sure that the written data is also
444 correctly read back.
445
446stonewall Wait for preceeding jobs in the job file to exit, before
447 starting this one. Can be used to insert serialization
448 points in the job file.
449
450numjobs=int Create the specified number of clones of this job. May be
451 used to setup a larger number of threads/processes doing
452 the same thing.
453
454thread fio defaults to forking jobs, however if this option is
455 given, fio will use pthread_create(3) to create threads
456 instead.
457
458zonesize=siint Divide a file into zones of the specified size. See zoneskip.
459
460zoneskip=siint Skip the specified number of bytes when zonesize data has
461 been read. The two zone options can be used to only do
462 io on zones of a file.
463
076efc7c
JA
464write_iolog=str Write the issued io patterns to the specified file. See
465 read_iolog.
71bfa161 466
076efc7c 467read_iolog=str Open an iolog with the specified file name and replay the
71bfa161
JA
468 io patterns it contains. This can be used to store a
469 workload and replay it sometime later.
470
471write_bw_log If given, write a bandwidth log of the jobs in this job
472 file. Can be used to store data of the bandwidth of the
e0da9bc2
JA
473 jobs in their lifetime. The included fio_generate_plots
474 script uses gnuplot to turn these text files into nice
475 graphs.
71bfa161
JA
476
477write_lat_log Same as write_bw_log, except that this option stores io
478 completion latencies instead.
479
480lockmem=siint Pin down the specified amount of memory with mlock(2). Can
481 potentially be used instead of removing memory or booting
482 with less memory to simulate a smaller amount of memory.
483
484exec_prerun=str Before running this job, issue the command specified
485 through system(3).
486
487exec_postrun=str After the job completes, issue the command specified
488 though system(3).
489
490ioscheduler=str Attempt to switch the device hosting the file to the specified
491 io scheduler before running.
492
493cpuload=int If the job is a CPU cycle eater, attempt to use the specified
494 percentage of CPU cycles.
495
496cpuchunks=int If the job is a CPU cycle eater, split the load into
6c219763 497 cycles of the given time. In milliseconds.
71bfa161
JA
498
499
5006.0 Interpreting the output
501---------------------------
502
503fio spits out a lot of output. While running, fio will display the
504status of the jobs created. An example of that would be:
505
6043c579 506Threads running: 1: [_r] [24.79% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
71bfa161
JA
507
508The characters inside the square brackets denote the current status of
509each thread. The possible values (in typical life cycle order) are:
510
511Idle Run
512---- ---
513P Thread setup, but not started.
514C Thread created.
515I Thread initialized, waiting.
516 R Running, doing sequential reads.
517 r Running, doing random reads.
518 W Running, doing sequential writes.
519 w Running, doing random writes.
520 M Running, doing mixed sequential reads/writes.
521 m Running, doing mixed random reads/writes.
522 F Running, currently waiting for fsync()
523V Running, doing verification of written data.
524E Thread exited, not reaped by main thread yet.
525_ Thread reaped.
526
527The other values are fairly self explanatory - number of threads
6043c579
JA
528currently running and doing io, rate of io since last check, and the estimated
529completion percentage and time for the running group. It's impossible to
530estimate runtime of the following groups (if any).
71bfa161
JA
531
532When fio is done (or interrupted by ctrl-c), it will show the data for
533each thread, group of threads, and disks in that order. For each data
534direction, the output looks like:
535
536Client1 (g=0): err= 0:
537 write: io= 32MiB, bw= 666KiB/s, runt= 50320msec
6104ddb6
JA
538 slat (msec): min= 0, max= 136, avg= 0.03, stdev= 1.92
539 clat (msec): min= 0, max= 631, avg=48.50, stdev=86.82
540 bw (KiB/s) : min= 0, max= 1196, per=51.00%, avg=664.02, stdev=681.68
71bfa161
JA
541 cpu : usr=1.49%, sys=0.25%, ctx=7969
542
543The client number is printed, along with the group id and error of that
544thread. Below is the io statistics, here for writes. In the order listed,
545they denote:
546
547io= Number of megabytes io performed
548bw= Average bandwidth rate
549runt= The runtime of that thread
550 slat= Submission latency (avg being the average, dev being the
551 standard deviation). This is the time it took to submit
552 the io. For sync io, the slat is really the completion
553 latency, since queue/complete is one operation there.
554 clat= Completion latency. Same names as slat, this denotes the
555 time from submission to completion of the io pieces. For
556 sync io, clat will usually be equal (or very close) to 0,
557 as the time from submit to complete is basically just
558 CPU time (io has already been done, see slat explanation).
559 bw= Bandwidth. Same names as the xlat stats, but also includes
560 an approximate percentage of total aggregate bandwidth
561 this thread received in this group. This last value is
562 only really useful if the threads in this group are on the
563 same disk, since they are then competing for disk access.
564cpu= CPU usage. User and system time, along with the number
565 of context switches this thread went through.
566
567After each client has been listed, the group statistics are printed. They
568will look like this:
569
570Run status group 0 (all jobs):
571 READ: io=64MiB, aggrb=22178, minb=11355, maxb=11814, mint=2840msec, maxt=2955msec
572 WRITE: io=64MiB, aggrb=1302, minb=666, maxb=669, mint=50093msec, maxt=50320msec
573
574For each data direction, it prints:
575
576io= Number of megabytes io performed.
577aggrb= Aggregate bandwidth of threads in this group.
578minb= The minimum average bandwidth a thread saw.
579maxb= The maximum average bandwidth a thread saw.
580mint= The smallest runtime of the threads in that group.
581maxt= The longest runtime of the threads in that group.
582
583And finally, the disk statistics are printed. They will look like this:
584
585Disk stats (read/write):
586 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
587
588Each value is printed for both reads and writes, with reads first. The
589numbers denote:
590
591ios= Number of ios performed by all groups.
592merge= Number of merges io the io scheduler.
593ticks= Number of ticks we kept the disk busy.
594io_queue= Total time spent in the disk queue.
595util= The disk utilization. A value of 100% means we kept the disk
596 busy constantly, 50% would be a disk idling half of the time.
597
598
5997.0 Terse output
600----------------
601
602For scripted usage where you typically want to generate tables or graphs
6c219763 603of the results, fio can output the results in a comma separated format.
71bfa161
JA
604The format is one long line of values, such as:
605
606client1,0,0,936,331,2894,0,0,0.000000,0.000000,1,170,22.115385,34.290410,16,714,84.252874%,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
607
608Split up, the format is as follows:
609
610 jobname, groupid, error
611 READ status:
612 KiB IO, bandwidth (KiB/sec), runtime (msec)
613 Submission latency: min, max, mean, deviation
614 Completion latency: min, max, mean, deviation
6c219763 615 Bw: min, max, aggregate percentage of total, mean, deviation
71bfa161
JA
616 WRITE status:
617 KiB IO, bandwidth (KiB/sec), runtime (msec)
618 Submission latency: min, max, mean, deviation
619 Completion latency: min, max, mean, deviation
6c219763 620 Bw: min, max, aggregate percentage of total, mean, deviation
71bfa161
JA
621 CPU usage: user, system, context switches
622