Add vsync io engine
[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
d0ff85df 52 could be using splice, async io, syslet, or even
71bfa161
JA
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
65db0851
JA
111section residing above it. If the first character in a line is a ';' or a
112'#', the entire line is discarded as a comment.
71bfa161
JA
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.
6d16ecb6
JA
173int Integer. A whole number value, can be negative. If prefixed with
174 0x, the integer is assumed to be of base 16 (hexidecimal).
71bfa161
JA
175siint SI integer. A whole number value, which may contain a postfix
176 describing the base of the number. Accepted postfixes are k/m/g,
6c219763 177 meaning kilo, mega, and giga. So if you want to specify 4096,
71bfa161
JA
178 you could either write out '4096' or just give 4k. The postfixes
179 signify base 2 values, so 1024 is 1k and 1024k is 1m and so on.
43159d18
JA
180 If the option accepts an upper and lower range, use a colon ':'
181 or minus '-' to seperate such values. See irange.
71bfa161
JA
182bool Boolean. Usually parsed as an integer, however only defined for
183 true and false (1 and 0).
184irange Integer range with postfix. Allows value range to be given, such
0c9baf91
JA
185 as 1024-4096. A colon may also be used as the seperator, eg
186 1k:4k. If the option allows two sets of ranges, they can be
187 specified with a ',' or '/' delimiter: 1k-4k/8k-32k. Also see
188 siint.
71bfa161
JA
189
190With the above in mind, here follows the complete list of fio job
191parameters.
192
193name=str ASCII name of the job. This may be used to override the
194 name printed by fio for this job. Otherwise the job
c2b1e753 195 name is used. On the command line this parameter has the
6c219763 196 special purpose of also signaling the start of a new
c2b1e753 197 job.
71bfa161 198
61697c37
JA
199description=str Text description of the job. Doesn't do anything except
200 dump this text description when this job is run. It's
201 not parsed.
202
71bfa161
JA
203directory=str Prefix filenames with this directory. Used to places files
204 in a different location than "./".
205
206filename=str Fio normally makes up a filename based on the job name,
207 thread number, and file number. If you want to share
208 files between threads in a job or several jobs, specify
ed92ac0c
JA
209 a filename for each of them to override the default. If
210 the ioengine used is 'net', the filename is the host and
9f9214f2 211 port to connect to in the format of =host/port. If the
af52b345
JA
212 ioengine is file based, you can specify a number of files
213 by seperating the names with a ':' colon. So if you wanted
214 a job to open /dev/sda and /dev/sdb as the two working files,
66159828
JA
215 you would use filename=/dev/sda:/dev/sdb. '-' is a reserved
216 name, meaning stdin or stdout. Which of the two depends
217 on the read/write direction set.
71bfa161 218
bbf6b540
JA
219opendir=str Tell fio to recursively add any file it can find in this
220 directory and down the file system tree.
221
d3aad8f2 222readwrite=str
71bfa161
JA
223rw=str Type of io pattern. Accepted values are:
224
225 read Sequential reads
226 write Sequential writes
227 randwrite Random writes
228 randread Random reads
229 rw Sequential mixed reads and writes
230 randrw Random mixed reads and writes
231
232 For the mixed io types, the default is to split them 50/50.
233 For certain types of io the result may still be skewed a bit,
211097b2
JA
234 since the speed may be different. It is possible to specify
235 a number of IO's to do before getting a new offset - this
236 is only useful for random IO, where fio would normally
237 generate a new random offset for every IO. If you append
238 eg 8 to randread, you would get a new random offset for
239 every 8 IO's. The result would be a seek for only every 8
240 IO's, instead of for every IO. Use rw=randread:8 to specify
241 that.
71bfa161 242
ee738499
JA
243randrepeat=bool For random IO workloads, seed the generator in a predictable
244 way so that results are repeatable across repetitions.
245
d2f3ac35
JA
246fadvise_hint=bool By default, fio will use fadvise() to advise the kernel
247 on what IO patterns it is likely to issue. Sometimes you
248 want to test specific IO patterns without telling the
249 kernel about it, in which case you can disable this option.
250 If set, fio will use POSIX_FADV_SEQUENTIAL for sequential
251 IO and POSIX_FADV_RANDOM for random IO.
252
7616cafe
JA
253size=siint The total size of file io for this job. Fio will run until
254 this many bytes has been transferred, unless runtime is
255 limited by other options (such as 'runtime', for instance).
256 Unless specific nr_files and filesize options are given,
257 fio will divide this size between the available files
258 specified by the job.
71bfa161 259
9c60ce64
JA
260filesize=siint Individual file sizes. May be a range, in which case fio
261 will select sizes for files at random within the given range
262 and limited to 'size' in total (if that is given). If not
263 given, each created file is the same size.
264
aa31f1f1
SL
265fill_device=bool Sets size to something really large and waits for ENOSPC (no
266 space left on device) as the terminating condition. Only makes
267 sense with sequential write.
268
d3aad8f2 269blocksize=siint
f90eff5a
JA
270bs=siint The block size used for the io units. Defaults to 4k. Values
271 can be given for both read and writes. If a single siint is
272 given, it will apply to both. If a second siint is specified
273 after a comma, it will apply to writes only. In other words,
274 the format is either bs=read_and_write or bs=read,write.
275 bs=4k,8k will thus use 4k blocks for reads, and 8k blocks
787f7e95
JA
276 for writes. If you only wish to set the write size, you
277 can do so by passing an empty read size - bs=,8k will set
278 8k for writes and leave the read default value.
a00735e6 279
d3aad8f2 280blocksize_range=irange
71bfa161
JA
281bsrange=irange Instead of giving a single block size, specify a range
282 and fio will mix the issued io block sizes. The issued
283 io unit will always be a multiple of the minimum value
f90eff5a
JA
284 given (also see bs_unaligned). Applies to both reads and
285 writes, however a second range can be given after a comma.
286 See bs=.
a00735e6 287
564ca972
JA
288bssplit=str Sometimes you want even finer grained control of the
289 block sizes issued, not just an even split between them.
290 This option allows you to weight various block sizes,
291 so that you are able to define a specific amount of
292 block sizes issued. The format for this option is:
293
294 bssplit=blocksize/percentage:blocksize/percentage
295
296 for as many block sizes as needed. So if you want to define
297 a workload that has 50% 64k blocks, 10% 4k blocks, and
298 40% 32k blocks, you would write:
299
300 bssplit=4k/10:64k/50:32k/40
301
302 Ordering does not matter. If the percentage is left blank,
303 fio will fill in the remaining values evenly. So a bssplit
304 option like this one:
305
306 bssplit=4k/50:1k/:32k/
307
308 would have 50% 4k ios, and 25% 1k and 32k ios. The percentages
309 always add up to 100, if bssplit is given a range that adds
310 up to more, it will error out.
311
d3aad8f2 312blocksize_unaligned
690adba3
JA
313bs_unaligned If this option is given, any byte size value within bsrange
314 may be used as a block range. This typically wont work with
315 direct IO, as that normally requires sector alignment.
71bfa161 316
e9459e5a
JA
317zero_buffers If this option is given, fio will init the IO buffers to
318 all zeroes. The default is to fill them with random data.
319
71bfa161
JA
320nrfiles=int Number of files to use for this job. Defaults to 1.
321
390b1537
JA
322openfiles=int Number of files to keep open at the same time. Defaults to
323 the same as nrfiles, can be set smaller to limit the number
324 simultaneous opens.
325
5af1c6f3
JA
326file_service_type=str Defines how fio decides which file from a job to
327 service next. The following types are defined:
328
329 random Just choose a file at random.
330
331 roundrobin Round robin over open files. This
332 is the default.
333
1907dbc6
JA
334 The string can have a number appended, indicating how
335 often to switch to a new file. So if option random:4 is
336 given, fio will switch to a new random file after 4 ios
337 have been issued.
338
71bfa161
JA
339ioengine=str Defines how the job issues io to the file. The following
340 types are defined:
341
342 sync Basic read(2) or write(2) io. lseek(2) is
343 used to position the io location.
344
a31041ea 345 psync Basic pread(2) or pwrite(2) io.
346
1d2af02a
JA
347 vsync Basic readv(2) or writev(3) IO.
348
71bfa161
JA
349 libaio Linux native asynchronous io.
350
351 posixaio glibc posix asynchronous io.
352
353 mmap File is memory mapped and data copied
354 to/from using memcpy(3).
355
356 splice splice(2) is used to transfer the data and
357 vmsplice(2) to transfer data from user
358 space to the kernel.
359
d0ff85df
JA
360 syslet-rw Use the syslet system calls to make
361 regular read/write async.
362
71bfa161 363 sg SCSI generic sg v3 io. May either be
6c219763 364 synchronous using the SG_IO ioctl, or if
71bfa161
JA
365 the target is an sg character device
366 we use read(2) and write(2) for asynchronous
367 io.
368
a94ea28b
JA
369 null Doesn't transfer any data, just pretends
370 to. This is mainly used to exercise fio
371 itself and for debugging/testing purposes.
372
ed92ac0c
JA
373 net Transfer over the network to given host:port.
374 'filename' must be set appropriately to
9f9214f2 375 filename=host/port regardless of send
ed92ac0c
JA
376 or receive, if the latter only the port
377 argument is used.
378
9cce02e8
JA
379 netsplice Like net, but uses splice/vmsplice to
380 map data and send/receive.
381
53aec0a4 382 cpuio Doesn't transfer any data, but burns CPU
ba0fbe10
JA
383 cycles according to the cpuload= and
384 cpucycle= options. Setting cpuload=85
385 will cause that job to do nothing but burn
386 85% of the CPU.
387
e9a1806f
JA
388 guasi The GUASI IO engine is the Generic Userspace
389 Asyncronous Syscall Interface approach
390 to async IO. See
391
392 http://www.xmailserver.org/guasi-lib.html
393
394 for more info on GUASI.
395
8a7bd877
JA
396 external Prefix to specify loading an external
397 IO engine object file. Append the engine
398 filename, eg ioengine=external:/tmp/foo.o
399 to load ioengine foo.o in /tmp.
400
71bfa161
JA
401iodepth=int This defines how many io units to keep in flight against
402 the file. The default is 1 for each file defined in this
403 job, can be overridden with a larger value for higher
404 concurrency.
405
cb5ab512 406iodepth_batch=int This defines how many pieces of IO to submit at once.
89e820f6
JA
407 It defaults to 1 which means that we submit each IO
408 as soon as it is available, but can be raised to submit
409 bigger batches of IO at the time.
cb5ab512 410
e916b390
JA
411iodepth_low=int The low water mark indicating when to start filling
412 the queue again. Defaults to the same as iodepth, meaning
413 that fio will attempt to keep the queue full at all times.
414 If iodepth is set to eg 16 and iodepth_low is set to 4, then
415 after fio has filled the queue of 16 requests, it will let
416 the depth drain down to 4 before starting to fill it again.
417
71bfa161 418direct=bool If value is true, use non-buffered io. This is usually
76a43db4
JA
419 O_DIRECT.
420
421buffered=bool If value is true, use buffered io. This is the opposite
422 of the 'direct' option. Defaults to true.
71bfa161
JA
423
424offset=siint Start io at the given offset in the file. The data before
425 the given offset will not be touched. This effectively
426 caps the file size at real_size - offset.
427
428fsync=int If writing to a file, issue a sync of the dirty data
429 for every number of blocks given. For example, if you give
430 32 as a parameter, fio will sync the file for every 32
431 writes issued. If fio is using non-buffered io, we may
432 not sync the file. The exception is the sg io engine, which
6c219763 433 synchronizes the disk cache anyway.
71bfa161
JA
434
435overwrite=bool If writing to a file, setup the file first and do overwrites.
436
437end_fsync=bool If true, fsync file contents when the job exits.
438
ebb1415f
JA
439fsync_on_close=bool If true, fio will fsync() a dirty file on close.
440 This differs from end_fsync in that it will happen on every
441 file close, not just at the end of the job.
442
6c219763 443rwmixcycle=int Value in milliseconds describing how often to switch between
71bfa161
JA
444 reads and writes for a mixed workload. The default is
445 500 msecs.
446
447rwmixread=int How large a percentage of the mix should be reads.
448
449rwmixwrite=int How large a percentage of the mix should be writes. If both
450 rwmixread and rwmixwrite is given and the values do not add
451 up to 100%, the latter of the two will be used to override
452 the first.
453
bb8895e0
JA
454norandommap Normally fio will cover every block of the file when doing
455 random IO. If this option is given, fio will just get a
456 new random offset without looking at past io history. This
457 means that some blocks may not be read or written, and that
458 some blocks may be read/written more than once. This option
7616cafe
JA
459 is mutually exclusive with verify= for that reason, since
460 fio doesn't track potential block rewrites which may alter
461 the calculated checksum for that block.
bb8895e0 462
71bfa161
JA
463nice=int Run the job with the given nice value. See man nice(2).
464
465prio=int Set the io priority value of this job. Linux limits us to
466 a positive value between 0 and 7, with 0 being the highest.
467 See man ionice(1).
468
469prioclass=int Set the io priority class. See man ionice(1).
470
471thinktime=int Stall the job x microseconds after an io has completed before
472 issuing the next. May be used to simulate processing being
48097d5c
JA
473 done by an application. See thinktime_blocks and
474 thinktime_spin.
475
476thinktime_spin=int
477 Only valid if thinktime is set - pretend to spend CPU time
478 doing something with the data received, before falling back
479 to sleeping for the rest of the period specified by
480 thinktime.
9c1f7434
JA
481
482thinktime_blocks
483 Only valid if thinktime is set - control how many blocks
484 to issue, before waiting 'thinktime' usecs. If not set,
485 defaults to 1 which will make fio wait 'thinktime' usecs
486 after every block.
71bfa161
JA
487
488rate=int Cap the bandwidth used by this job to this number of KiB/sec.
489
490ratemin=int Tell fio to do whatever it can to maintain at least this
4e991c23
JA
491 bandwidth. Failing to meet this requirement, will cause
492 the job to exit.
493
494rate_iops=int Cap the bandwidth to this number of IOPS. Basically the same
495 as rate, just specified independently of bandwidth. If the
496 job is given a block size range instead of a fixed value,
497 the smallest block size is used as the metric.
498
499rate_iops_min=int If fio doesn't meet this rate of IO, it will cause
500 the job to exit.
71bfa161
JA
501
502ratecycle=int Average bandwidth for 'rate' and 'ratemin' over this number
6c219763 503 of milliseconds.
71bfa161
JA
504
505cpumask=int Set the CPU affinity of this job. The parameter given is a
a08bc17f
JA
506 bitmask of allowed CPU's the job may run on. So if you want
507 the allowed CPUs to be 1 and 5, you would pass the decimal
508 value of (1 << 1 | 1 << 5), or 34. See man
7dbb6eba
JA
509 sched_setaffinity(2). This may not work on all supported
510 operating systems or kernel versions.
71bfa161 511
d2e268b0
JA
512cpus_allowed=str Controls the same options as cpumask, but it allows a text
513 setting of the permitted CPUs instead. So to use CPUs 1 and
514 5, you would specify cpus_allowed=1,5.
515
71bfa161
JA
516startdelay=int Start this job the specified number of seconds after fio
517 has started. Only useful if the job file contains several
518 jobs, and you want to delay starting some jobs to a certain
519 time.
520
03b74b3e 521runtime=int Tell fio to terminate processing after the specified number
71bfa161
JA
522 of seconds. It can be quite hard to determine for how long
523 a specified job will run, so this parameter is handy to
524 cap the total runtime to a given time.
525
cf4464ca
JA
526time_based If set, fio will run for the duration of the runtime
527 specified even if the file(s) are completey read or
528 written. It will simply loop over the same workload
529 as many times as the runtime allows.
530
71bfa161
JA
531invalidate=bool Invalidate the buffer/page cache parts for this file prior
532 to starting io. Defaults to true.
533
534sync=bool Use sync io for buffered writes. For the majority of the
535 io engines, this means using O_SYNC.
536
d3aad8f2 537iomem=str
71bfa161
JA
538mem=str Fio can use various types of memory as the io unit buffer.
539 The allowed values are:
540
541 malloc Use memory from malloc(3) as the buffers.
542
543 shm Use shared memory as the buffers. Allocated
544 through shmget(2).
545
74b025b0
JA
546 shmhuge Same as shm, but use huge pages as backing.
547
313cb206
JA
548 mmap Use mmap to allocate buffers. May either be
549 anonymous memory, or can be file backed if
550 a filename is given after the option. The
551 format is mem=mmap:/path/to/file.
71bfa161 552
d0bdaf49
JA
553 mmaphuge Use a memory mapped huge file as the buffer
554 backing. Append filename after mmaphuge, ala
555 mem=mmaphuge:/hugetlbfs/file
556
71bfa161 557 The area allocated is a function of the maximum allowed
5394ae5f
JA
558 bs size for the job, multiplied by the io depth given. Note
559 that for shmhuge and mmaphuge to work, the system must have
560 free huge pages allocated. This can normally be checked
561 and set by reading/writing /proc/sys/vm/nr_hugepages on a
562 Linux system. Fio assumes a huge page is 4MiB in size. So
563 to calculate the number of huge pages you need for a given
564 job file, add up the io depth of all jobs (normally one unless
565 iodepth= is used) and multiply by the maximum bs set. Then
566 divide that number by the huge page size. You can see the
567 size of the huge pages in /proc/meminfo. If no huge pages
568 are allocated by having a non-zero number in nr_hugepages,
56bb17f2 569 using mmaphuge or shmhuge will fail. Also see hugepage-size.
5394ae5f
JA
570
571 mmaphuge also needs to have hugetlbfs mounted and the file
572 location should point there. So if it's mounted in /huge,
573 you would use mem=mmaphuge:/huge/somefile.
71bfa161 574
56bb17f2
JA
575hugepage-size=siint
576 Defines the size of a huge page. Must at least be equal
577 to the system setting, see /proc/meminfo. Defaults to 4MiB.
c51074e7
JA
578 Should probably always be a multiple of megabytes, so using
579 hugepage-size=Xm is the preferred way to set this to avoid
580 setting a non-pow-2 bad value.
56bb17f2 581
71bfa161
JA
582exitall When one job finishes, terminate the rest. The default is
583 to wait for each job to finish, sometimes that is not the
584 desired action.
585
586bwavgtime=int Average the calculated bandwidth over the given time. Value
6c219763 587 is specified in milliseconds.
71bfa161
JA
588
589create_serialize=bool If true, serialize the file creating for the jobs.
590 This may be handy to avoid interleaving of data
591 files, which may greatly depend on the filesystem
592 used and even the number of processors in the system.
593
594create_fsync=bool fsync the data file after creation. This is the
595 default.
596
e545a6ce
JA
597unlink=bool Unlink the job files when done. Not the default, as repeated
598 runs of that job would then waste time recreating the fileset
599 again and again.
71bfa161
JA
600
601loops=int Run the specified number of iterations of this job. Used
602 to repeat the same workload a given number of times. Defaults
603 to 1.
604
68e1f29a 605do_verify=bool Run the verify phase after a write phase. Only makes sense if
e84c73a8
SL
606 verify is set. Defaults to 1.
607
71bfa161
JA
608verify=str If writing to a file, fio can verify the file contents
609 after each iteration of the job. The allowed values are:
610
611 md5 Use an md5 sum of the data area and store
612 it in the header of each block.
613
17dc34df
JA
614 crc64 Use an experimental crc64 sum of the data
615 area and store it in the header of each
616 block.
617
71bfa161
JA
618 crc32 Use a crc32 sum of the data area and store
619 it in the header of each block.
620
969f7ed3
JA
621 crc16 Use a crc16 sum of the data area and store
622 it in the header of each block.
623
17dc34df
JA
624 crc7 Use a crc7 sum of the data area and store
625 it in the header of each block.
626
cd14cc10
JA
627 sha512 Use sha512 as the checksum function.
628
629 sha256 Use sha256 as the checksum function.
630
7437ee87
SL
631 meta Write extra information about each io
632 (timestamp, block number etc.). The block
633 number is verified.
634
36690c9b
JA
635 null Only pretend to verify. Useful for testing
636 internals with ioengine=null, not for much
637 else.
638
6c219763 639 This option can be used for repeated burn-in tests of a
71bfa161
JA
640 system to make sure that the written data is also
641 correctly read back.
642
160b966d
JA
643verifysort=bool If set, fio will sort written verify blocks when it deems
644 it faster to read them back in a sorted manner. This is
645 often the case when overwriting an existing file, since
646 the blocks are already laid out in the file system. You
647 can ignore this option unless doing huge amounts of really
648 fast IO where the red-black tree sorting CPU time becomes
649 significant.
3f9f4e26 650
a59e170d 651verify_offset=siint Swap the verification header with data somewhere else
546a9142
SL
652 in the block before writing. Its swapped back before
653 verifying.
654
a59e170d 655verify_interval=siint Write the verification header at a finer granularity
3f9f4e26
SL
656 than the blocksize. It will be written for chunks the
657 size of header_interval. blocksize should divide this
658 evenly.
90059d65 659
e28218f3
SL
660verify_pattern=int If set, fio will fill the io buffers with this
661 pattern. Fio defaults to filling with totally random
662 bytes, but sometimes it's interesting to fill with a known
663 pattern for io verification purposes. Depending on the
664 width of the pattern, fio will fill 1/2/3/4 bytes of the
665 buffer at the time. The verify_pattern cannot be larger than
666 a 32-bit quantity.
667
68e1f29a 668verify_fatal=bool Normally fio will keep checking the entire contents
a12a3b4d
JA
669 before quitting on a block verification failure. If this
670 option is set, fio will exit the job on the first observed
671 failure.
160b966d 672
71bfa161
JA
673stonewall Wait for preceeding jobs in the job file to exit, before
674 starting this one. Can be used to insert serialization
b3d62a75
JA
675 points in the job file. A stone wall also implies starting
676 a new reporting group.
677
678new_group Start a new reporting group. If this option isn't given,
679 jobs in a file will be part of the same reporting group
680 unless seperated by a stone wall (or if it's a group
681 by itself, with the numjobs option).
71bfa161
JA
682
683numjobs=int Create the specified number of clones of this job. May be
684 used to setup a larger number of threads/processes doing
fa28c85a
JA
685 the same thing. We regard that grouping of jobs as a
686 specific group.
687
688group_reporting If 'numjobs' is set, it may be interesting to display
689 statistics for the group as a whole instead of for each
690 individual job. This is especially true of 'numjobs' is
691 large, looking at individual thread/process output quickly
692 becomes unwieldy. If 'group_reporting' is specified, fio
693 will show the final report per-group instead of per-job.
71bfa161
JA
694
695thread fio defaults to forking jobs, however if this option is
696 given, fio will use pthread_create(3) to create threads
697 instead.
698
699zonesize=siint Divide a file into zones of the specified size. See zoneskip.
700
701zoneskip=siint Skip the specified number of bytes when zonesize data has
702 been read. The two zone options can be used to only do
703 io on zones of a file.
704
076efc7c
JA
705write_iolog=str Write the issued io patterns to the specified file. See
706 read_iolog.
71bfa161 707
076efc7c 708read_iolog=str Open an iolog with the specified file name and replay the
71bfa161 709 io patterns it contains. This can be used to store a
6df8adaa
JA
710 workload and replay it sometime later. The iolog given
711 may also be a blktrace binary file, which allows fio
712 to replay a workload captured by blktrace. See blktrace
713 for how to capture such logging data. For blktrace replay,
714 the file needs to be turned into a blkparse binary data
715 file first (blktrace <device> -d file_for_fio.bin).
71bfa161
JA
716
717write_bw_log If given, write a bandwidth log of the jobs in this job
718 file. Can be used to store data of the bandwidth of the
e0da9bc2
JA
719 jobs in their lifetime. The included fio_generate_plots
720 script uses gnuplot to turn these text files into nice
721 graphs.
71bfa161
JA
722
723write_lat_log Same as write_bw_log, except that this option stores io
724 completion latencies instead.
725
726lockmem=siint Pin down the specified amount of memory with mlock(2). Can
727 potentially be used instead of removing memory or booting
728 with less memory to simulate a smaller amount of memory.
729
730exec_prerun=str Before running this job, issue the command specified
731 through system(3).
732
733exec_postrun=str After the job completes, issue the command specified
734 though system(3).
735
736ioscheduler=str Attempt to switch the device hosting the file to the specified
737 io scheduler before running.
738
739cpuload=int If the job is a CPU cycle eater, attempt to use the specified
740 percentage of CPU cycles.
741
742cpuchunks=int If the job is a CPU cycle eater, split the load into
6c219763 743 cycles of the given time. In milliseconds.
71bfa161 744
0a839f30
JA
745disk_util=bool Generate disk utilization statistics, if the platform
746 supports it. Defaults to on.
747
71bfa161
JA
748
7496.0 Interpreting the output
750---------------------------
751
752fio spits out a lot of output. While running, fio will display the
753status of the jobs created. An example of that would be:
754
73c8b082 755Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
71bfa161
JA
756
757The characters inside the square brackets denote the current status of
758each thread. The possible values (in typical life cycle order) are:
759
760Idle Run
761---- ---
762P Thread setup, but not started.
763C Thread created.
764I Thread initialized, waiting.
765 R Running, doing sequential reads.
766 r Running, doing random reads.
767 W Running, doing sequential writes.
768 w Running, doing random writes.
769 M Running, doing mixed sequential reads/writes.
770 m Running, doing mixed random reads/writes.
771 F Running, currently waiting for fsync()
772V Running, doing verification of written data.
773E Thread exited, not reaped by main thread yet.
774_ Thread reaped.
775
776The other values are fairly self explanatory - number of threads
c9f60304
JA
777currently running and doing io, rate of io since last check (read speed
778listed first, then write speed), and the estimated completion percentage
779and time for the running group. It's impossible to estimate runtime of
780the following groups (if any).
71bfa161
JA
781
782When fio is done (or interrupted by ctrl-c), it will show the data for
783each thread, group of threads, and disks in that order. For each data
784direction, the output looks like:
785
786Client1 (g=0): err= 0:
787 write: io= 32MiB, bw= 666KiB/s, runt= 50320msec
6104ddb6
JA
788 slat (msec): min= 0, max= 136, avg= 0.03, stdev= 1.92
789 clat (msec): min= 0, max= 631, avg=48.50, stdev=86.82
790 bw (KiB/s) : min= 0, max= 1196, per=51.00%, avg=664.02, stdev=681.68
e7823a94 791 cpu : usr=1.49%, sys=0.25%, ctx=7969, majf=0, minf=17
71619dc2 792 IO depths : 1=0.1%, 2=0.3%, 4=0.5%, 8=99.0%, 16=0.0%, 32=0.0%, >32=0.0%
30061b97 793 issued r/w: total=0/32768, short=0/0
8abdce66
JA
794 lat (msec): 2=1.6%, 4=0.0%, 10=3.2%, 20=12.8%, 50=38.4%, 100=24.8%,
795 lat (msec): 250=15.2%, 500=0.0%, 750=0.0%, 1000=0.0%, >=2048=0.0%
71bfa161
JA
796
797The client number is printed, along with the group id and error of that
798thread. Below is the io statistics, here for writes. In the order listed,
799they denote:
800
801io= Number of megabytes io performed
802bw= Average bandwidth rate
803runt= The runtime of that thread
72fbda2a 804 slat= Submission latency (avg being the average, stdev being the
71bfa161
JA
805 standard deviation). This is the time it took to submit
806 the io. For sync io, the slat is really the completion
8a35c71e
JA
807 latency, since queue/complete is one operation there. This
808 value can be in miliseconds or microseconds, fio will choose
809 the most appropriate base and print that. In the example
810 above, miliseconds is the best scale.
71bfa161
JA
811 clat= Completion latency. Same names as slat, this denotes the
812 time from submission to completion of the io pieces. For
813 sync io, clat will usually be equal (or very close) to 0,
814 as the time from submit to complete is basically just
815 CPU time (io has already been done, see slat explanation).
816 bw= Bandwidth. Same names as the xlat stats, but also includes
817 an approximate percentage of total aggregate bandwidth
818 this thread received in this group. This last value is
819 only really useful if the threads in this group are on the
820 same disk, since they are then competing for disk access.
821cpu= CPU usage. User and system time, along with the number
e7823a94
JA
822 of context switches this thread went through, usage of
823 system and user time, and finally the number of major
824 and minor page faults.
71619dc2
JA
825IO depths= The distribution of io depths over the job life time. The
826 numbers are divided into powers of 2, so for example the
827 16= entries includes depths up to that value but higher
828 than the previous entry. In other words, it covers the
829 range from 16 to 31.
30061b97
JA
830IO issued= The number of read/write requests issued, and how many
831 of them were short.
ec118304
JA
832IO latencies= The distribution of IO completion latencies. This is the
833 time from when IO leaves fio and when it gets completed.
834 The numbers follow the same pattern as the IO depths,
835 meaning that 2=1.6% means that 1.6% of the IO completed
8abdce66
JA
836 within 2 msecs, 20=12.8% means that 12.8% of the IO
837 took more than 10 msecs, but less than (or equal to) 20 msecs.
71bfa161
JA
838
839After each client has been listed, the group statistics are printed. They
840will look like this:
841
842Run status group 0 (all jobs):
843 READ: io=64MiB, aggrb=22178, minb=11355, maxb=11814, mint=2840msec, maxt=2955msec
844 WRITE: io=64MiB, aggrb=1302, minb=666, maxb=669, mint=50093msec, maxt=50320msec
845
846For each data direction, it prints:
847
848io= Number of megabytes io performed.
849aggrb= Aggregate bandwidth of threads in this group.
850minb= The minimum average bandwidth a thread saw.
851maxb= The maximum average bandwidth a thread saw.
852mint= The smallest runtime of the threads in that group.
853maxt= The longest runtime of the threads in that group.
854
855And finally, the disk statistics are printed. They will look like this:
856
857Disk stats (read/write):
858 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
859
860Each value is printed for both reads and writes, with reads first. The
861numbers denote:
862
863ios= Number of ios performed by all groups.
864merge= Number of merges io the io scheduler.
865ticks= Number of ticks we kept the disk busy.
866io_queue= Total time spent in the disk queue.
867util= The disk utilization. A value of 100% means we kept the disk
868 busy constantly, 50% would be a disk idling half of the time.
869
870
8717.0 Terse output
872----------------
873
874For scripted usage where you typically want to generate tables or graphs
6af019c9 875of the results, fio can output the results in a semicolon separated format.
71bfa161
JA
876The format is one long line of values, such as:
877
6af019c9
JA
878client1;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%
879;0.0%;0.0%;0.0%;0.0%;0.0%
71bfa161
JA
880
881Split up, the format is as follows:
882
883 jobname, groupid, error
884 READ status:
885 KiB IO, bandwidth (KiB/sec), runtime (msec)
886 Submission latency: min, max, mean, deviation
887 Completion latency: min, max, mean, deviation
6c219763 888 Bw: min, max, aggregate percentage of total, mean, deviation
71bfa161
JA
889 WRITE status:
890 KiB IO, bandwidth (KiB/sec), runtime (msec)
891 Submission latency: min, max, mean, deviation
892 Completion latency: min, max, mean, deviation
6c219763 893 Bw: min, max, aggregate percentage of total, mean, deviation
046ee302 894 CPU usage: user, system, context switches, major faults, minor faults
2270890c
JA
895 IO depths: <=1, 2, 4, 8, 16, 32, >=64
896 IO latencies: <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, >=2000
897 Text description
71bfa161 898