1 .TH fio 1 "September 2007" "User Manual"
3 fio \- flexible I/O tester
6 [\fIoptions\fR] [\fIjobfile\fR]...
9 is a tool that will spawn a number of threads or processes doing a
10 particular type of I/O action as specified by the user.
11 The typical use of fio is to write a job file matching the I/O load
12 one wants to simulate.
15 .BI \-\-output \fR=\fPfilename
16 Write output to \fIfilename\fR.
18 .BI \-\-timeout \fR=\fPtimeout
19 Limit run time to \fItimeout\fR seconds.
22 Generate per-job latency logs.
25 Generate per-job bandwidth logs.
28 Print statistics in a terse, semicolon-delimited format.
30 .BI \-\-showcmd \fR=\fPjobfile
31 Convert \fIjobfile\fR to a set of command-line options.
34 Enable read-only safety checks.
36 .BI \-\-eta \fR=\fPwhen
37 Specifies when real-time ETA estimate should be printed. \fIwhen\fR may
38 be one of `always', `never' or `auto'.
40 .BI \-\-section \fR=\fPsec
41 Only run section \fIsec\fR from job file.
43 .BI \-\-cmdhelp \fR=\fPcommand
44 Print help information for \fIcommand\fR. May be `all' for all commands.
46 .BI \-\-debug \fR=\fPtype
47 Enable verbose tracing of various fio actions. May be `all' for all types
48 or individual types separated by a comma (eg \-\-debug=io,file). `help' will
49 list all available tracing options.
52 Display usage information and exit.
55 Display version information and exit.
57 Job files are in `ini' format. They consist of one or more
58 job definitions, which begin with a job name in square brackets and
59 extend to the next job name. The job name can be any ASCII string
60 except `global', which has a special meaning. Following the job name is
61 a sequence of zero or more parameters, one per line, that define the
62 behavior of the job. Any line starting with a `;' or `#' character is
63 considered a comment and ignored.
65 If \fIjobfile\fR is specified as `-', the job file will be read from
68 The global section contains default parameters for jobs specified in the
69 job file. A job is only affected by global sections residing above it,
70 and there may be any number of global sections. Specific job definitions
71 may override any parameter set in global sections.
74 Some parameters may take arguments of a specific type. The types used are:
77 String: a sequence of alphanumeric characters.
80 SI integer: a whole number, possibly containing a suffix denoting the base unit
81 of the value. Accepted suffixes are `k', 'M', 'G', 'T', and 'P', denoting
82 kilo (1024), mega (1024^2), giga (1024^3), tera (1024^4), and peta (1024^5)
83 respectively. The suffix is not case sensitive. If prefixed with '0x', the
84 value is assumed to be base 16 (hexadecimal). A suffix may include a trailing 'b',
85 for instance 'kb' is identical to 'k'. You can specify a base 10 value
86 by using 'KiB', 'MiB', 'GiB', etc. This is useful for disk drives where
87 values are often given in base 10 values. Specifying '30GiB' will get you
91 Boolean: a true or false value. `0' denotes false, `1' denotes true.
94 Integer range: a range of integers specified in the format
95 \fIlower\fR:\fIupper\fR or \fIlower\fR\-\fIupper\fR. \fIlower\fR and
96 \fIupper\fR may contain a suffix as described above. If an option allows two
97 sets of ranges, they are separated with a `,' or `/' character. For example:
101 List of floating numbers: A list of floating numbers, separated by
106 May be used to override the job name. On the command line, this parameter
107 has the special purpose of signalling the start of a new job.
109 .BI description \fR=\fPstr
110 Human-readable description of the job. It is printed when the job is run, but
111 otherwise has no special purpose.
113 .BI directory \fR=\fPstr
114 Prefix filenames with this directory. Used to place files in a location other
117 .BI filename \fR=\fPstr
119 normally makes up a file name based on the job name, thread number, and file
120 number. If you want to share files between threads in a job or several jobs,
121 specify a \fIfilename\fR for each of them to override the default. If the I/O
122 engine used is `net', \fIfilename\fR is the host and port to connect to in the
123 format \fIhost\fR/\fIport\fR. If the I/O engine is file-based, you can specify
124 a number of files by separating the names with a `:' character. `\-' is a
125 reserved name, meaning stdin or stdout, depending on the read/write direction
128 .BI lockfile \fR=\fPstr
129 Fio defaults to not locking any files before it does IO to them. If a file or
130 file descriptor is shared, fio can serialize IO to that file to make the end
131 result consistent. This is usual for emulating real workloads that share files.
137 No locking. This is the default.
140 Only one thread or process may do IO at the time, excluding all others.
143 Read-write locking on the file. Many readers may access the file at the same
144 time, but writes get exclusive access.
147 The option may be post-fixed with a lock batch number. If set, then each
148 thread/process may do that amount of IOs to the file before giving up the lock.
149 Since lock acquisition is expensive, batching the lock/unlocks will speed up IO.
152 .BI opendir \fR=\fPstr
153 Recursively open any files below directory \fIstr\fR.
155 .BI readwrite \fR=\fPstr "\fR,\fP rw" \fR=\fPstr
156 Type of I/O pattern. Accepted values are:
173 Mixed sequential reads and writes.
176 Mixed random reads and writes.
179 For mixed I/O, the default split is 50/50. For certain types of io the result
180 may still be skewed a bit, since the speed may be different. It is possible to
181 specify a number of IO's to do before getting a new offset, this is one by
182 appending a `:\fI<nr>\fR to the end of the string given. For a random read, it
183 would look like \fBrw=randread:8\fR for passing in an offset modifier with a
184 value of 8. If the postfix is used with a sequential IO pattern, then the value
185 specified will be added to the generated offset for each IO. For instance,
186 using \fBrw=write:4k\fR will skip 4k for every write. It turns sequential IO
187 into sequential IO with holes. See the \fBrw_sequencer\fR option.
190 .BI rw_sequencer \fR=\fPstr
191 If an offset modifier is given by appending a number to the \fBrw=<str>\fR line,
192 then this option controls how that number modifies the IO offset being
193 generated. Accepted values are:
198 Generate sequential offset
201 Generate the same offset
204 \fBsequential\fR is only useful for random IO, where fio would normally
205 generate a new random offset for every IO. If you append eg 8 to randread, you
206 would get a new random offset for every 8 IO's. The result would be a seek for
207 only every 8 IO's, instead of for every IO. Use \fBrw=randread:8\fR to specify
208 that. As sequential IO is already sequential, setting \fBsequential\fR for that
209 would not result in any differences. \fBidentical\fR behaves in a similar
210 fashion, except it sends the same offset 8 number of times before generating a
215 .BI kb_base \fR=\fPint
216 The base unit for a kilobyte. The defacto base is 2^10, 1024. Storage
217 manufacturers like to use 10^3 or 1000 as a base ten unit instead, for obvious
218 reasons. Allow values are 1024 or 1000, with 1024 being the default.
220 .BI randrepeat \fR=\fPbool
221 Seed the random number generator in a predictable way so results are repeatable
222 across runs. Default: true.
224 .BI use_os_rand \fR=\fPbool
225 Fio can either use the random generator supplied by the OS to generator random
226 offsets, or it can use it's own internal generator (based on Tausworthe).
227 Default is to use the internal generator, which is often of better quality and
228 faster. Default: false.
230 .BI fallocate \fR=\fPstr
231 Whether pre-allocation is performed when laying down files. Accepted values
237 Do not pre-allocate space.
240 Pre-allocate via posix_fallocate().
243 Pre-allocate via fallocate() with FALLOC_FL_KEEP_SIZE set.
246 Backward-compatible alias for 'none'.
249 Backward-compatible alias for 'posix'.
252 May not be available on all supported platforms. 'keep' is only
253 available on Linux. If using ZFS on Solaris this must be set to 'none'
254 because ZFS doesn't support it. Default: 'posix'.
257 .BI fadvise_hint \fR=\fPbool
258 Disable use of \fIposix_fadvise\fR\|(2) to advise the kernel what I/O patterns
259 are likely to be issued. Default: true.
262 Total size of I/O for this job. \fBfio\fR will run until this many bytes have
263 been transfered, unless limited by other options (\fBruntime\fR, for instance).
264 Unless \fBnrfiles\fR and \fBfilesize\fR options are given, this amount will be
265 divided between the available files for the job. If not set, fio will use the
266 full size of the given files or devices. If the the files do not exist, size
267 must be given. It is also possible to give size as a percentage between 1 and
268 100. If size=20% is given, fio will use 20% of the full size of the given files
271 .BI fill_device \fR=\fPbool "\fR,\fB fill_fs" \fR=\fPbool
272 Sets size to something really large and waits for ENOSPC (no space left on
273 device) as the terminating condition. Only makes sense with sequential write.
274 For a read workload, the mount point will be filled first then IO started on
275 the result. This option doesn't make sense if operating on a raw device node,
276 since the size of that is already known by the file system. Additionally,
277 writing beyond end-of-device will not return ENOSPC there.
279 .BI filesize \fR=\fPirange
280 Individual file sizes. May be a range, in which case \fBfio\fR will select sizes
281 for files at random within the given range, limited to \fBsize\fR in total (if
282 that is given). If \fBfilesize\fR is not specified, each created file is the
285 .BI blocksize \fR=\fPint[,int] "\fR,\fB bs" \fR=\fPint[,int]
286 Block size for I/O units. Default: 4k. Values for reads and writes can be
287 specified separately in the format \fIread\fR,\fIwrite\fR, either of
288 which may be empty to leave that value at its default.
290 .BI blocksize_range \fR=\fPirange[,irange] "\fR,\fB bsrange" \fR=\fPirange[,irange]
291 Specify a range of I/O block sizes. The issued I/O unit will always be a
292 multiple of the minimum size, unless \fBblocksize_unaligned\fR is set. Applies
293 to both reads and writes if only one range is given, but can be specified
294 separately with a comma seperating the values. Example: bsrange=1k-4k,2k-8k.
295 Also (see \fBblocksize\fR).
297 .BI bssplit \fR=\fPstr
298 This option allows even finer grained control of the block sizes issued,
299 not just even splits between them. With this option, you can weight various
300 block sizes for exact control of the issued IO for a job that has mixed
301 block sizes. The format of the option is bssplit=blocksize/percentage,
302 optionally adding as many definitions as needed separated by a colon.
303 Example: bssplit=4k/10:64k/50:32k/40 would issue 50% 64k blocks, 10% 4k
304 blocks and 40% 32k blocks. \fBbssplit\fR also supports giving separate
305 splits to reads and writes. The format is identical to what the
306 \fBbs\fR option accepts, the read and write parts are separated with a
309 .B blocksize_unaligned\fR,\fP bs_unaligned
310 If set, any size in \fBblocksize_range\fR may be used. This typically won't
311 work with direct I/O, as that normally requires sector alignment.
313 .BI blockalign \fR=\fPint[,int] "\fR,\fB ba" \fR=\fPint[,int]
314 At what boundary to align random IO offsets. Defaults to the same as 'blocksize'
315 the minimum blocksize given. Minimum alignment is typically 512b
316 for using direct IO, though it usually depends on the hardware block size.
317 This option is mutually exclusive with using a random map for files, so it
318 will turn off that option.
321 Initialise buffers with all zeros. Default: fill buffers with random data.
324 If this option is given, fio will refill the IO buffers on every submit. The
325 default is to only fill it at init time and reuse that data. Only makes sense
326 if zero_buffers isn't specified, naturally. If data verification is enabled,
327 refill_buffers is also automatically enabled.
329 .BI nrfiles \fR=\fPint
330 Number of files to use for this job. Default: 1.
332 .BI openfiles \fR=\fPint
333 Number of files to keep open at the same time. Default: \fBnrfiles\fR.
335 .BI file_service_type \fR=\fPstr
336 Defines how files to service are selected. The following types are defined:
341 Choose a file at random
344 Round robin over open files (default).
346 Do each file in the set sequentially.
349 The number of I/Os to issue before switching a new file can be specified by
350 appending `:\fIint\fR' to the service type.
353 .BI ioengine \fR=\fPstr
354 Defines how the job issues I/O. The following types are defined:
359 Basic \fIread\fR\|(2) or \fIwrite\fR\|(2) I/O. \fIfseek\fR\|(2) is used to
360 position the I/O location.
363 Basic \fIpread\fR\|(2) or \fIpwrite\fR\|(2) I/O.
366 Basic \fIreadv\fR\|(2) or \fIwritev\fR\|(2) I/O. Will emulate queuing by
367 coalescing adjacents IOs into a single submission.
370 Linux native asynchronous I/O.
373 POSIX asynchronous I/O using \fIaio_read\fR\|(3) and \fIaio_write\fR\|(3).
376 Solaris native asynchronous I/O.
379 Windows native asynchronous I/O.
382 File is memory mapped with \fImmap\fR\|(2) and data copied using
386 \fIsplice\fR\|(2) is used to transfer the data and \fIvmsplice\fR\|(2) to
387 transfer data from user-space to the kernel.
390 Use the syslet system calls to make regular read/write asynchronous.
393 SCSI generic sg v3 I/O. May be either synchronous using the SG_IO ioctl, or if
394 the target is an sg character device, we use \fIread\fR\|(2) and
395 \fIwrite\fR\|(2) for asynchronous I/O.
398 Doesn't transfer any data, just pretends to. Mainly used to exercise \fBfio\fR
399 itself and for debugging and testing purposes.
402 Transfer over the network. \fBfilename\fR must be set appropriately to
403 `\fIhost\fR/\fIport\fR' regardless of data direction. If receiving, only the
404 \fIport\fR argument is used.
407 Like \fBnet\fR, but uses \fIsplice\fR\|(2) and \fIvmsplice\fR\|(2) to map data
411 Doesn't transfer any data, but burns CPU cycles according to \fBcpuload\fR and
412 \fBcpucycles\fR parameters.
415 The GUASI I/O engine is the Generic Userspace Asynchronous Syscall Interface
416 approach to asycnronous I/O.
418 See <http://www.xmailserver.org/guasi\-lib.html>.
421 The RDMA I/O engine supports both RDMA memory semantics (RDMA_WRITE/RDMA_READ)
422 and channel semantics (Send/Recv) for the InfiniBand, RoCE and iWARP protocols.
425 Loads an external I/O engine object file. Append the engine filename as
430 .BI iodepth \fR=\fPint
431 Number of I/O units to keep in flight against the file. Note that increasing
432 iodepth beyond 1 will not affect synchronous ioengines (except for small
433 degress when verify_async is in use). Even async engines my impose OS
434 restrictions causing the desired depth not to be achieved. This may happen on
435 Linux when using libaio and not setting \fBdirect\fR=1, since buffered IO is
436 not async on that OS. Keep an eye on the IO depth distribution in the
437 fio output to verify that the achieved depth is as expected. Default: 1.
439 .BI iodepth_batch \fR=\fPint
440 Number of I/Os to submit at once. Default: \fBiodepth\fR.
442 .BI iodepth_batch_complete \fR=\fPint
443 This defines how many pieces of IO to retrieve at once. It defaults to 1 which
444 means that we'll ask for a minimum of 1 IO in the retrieval process from the
445 kernel. The IO retrieval will go on until we hit the limit set by
446 \fBiodepth_low\fR. If this variable is set to 0, then fio will always check for
447 completed events before queuing more IO. This helps reduce IO latency, at the
448 cost of more retrieval system calls.
450 .BI iodepth_low \fR=\fPint
451 Low watermark indicating when to start filling the queue again. Default:
454 .BI direct \fR=\fPbool
455 If true, use non-buffered I/O (usually O_DIRECT). Default: false.
457 .BI buffered \fR=\fPbool
458 If true, use buffered I/O. This is the opposite of the \fBdirect\fR parameter.
461 .BI offset \fR=\fPint
462 Offset in the file to start I/O. Data before the offset will not be touched.
465 How many I/Os to perform before issuing an \fBfsync\fR\|(2) of dirty data. If
466 0, don't sync. Default: 0.
468 .BI fdatasync \fR=\fPint
469 Like \fBfsync\fR, but uses \fBfdatasync\fR\|(2) instead to only sync the
470 data parts of the file. Default: 0.
472 .BI sync_file_range \fR=\fPstr:int
473 Use sync_file_range() for every \fRval\fP number of write operations. Fio will
474 track range of writes that have happened since the last sync_file_range() call.
475 \fRstr\fP can currently be one or more of:
479 SYNC_FILE_RANGE_WAIT_BEFORE
482 SYNC_FILE_RANGE_WRITE
485 SYNC_FILE_RANGE_WRITE
489 So if you do sync_file_range=wait_before,write:8, fio would use
490 \fBSYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE\fP for every 8 writes.
491 Also see the sync_file_range(2) man page. This option is Linux specific.
493 .BI overwrite \fR=\fPbool
494 If writing, setup the file first and do overwrites. Default: false.
496 .BI end_fsync \fR=\fPbool
497 Sync file contents when job exits. Default: false.
499 .BI fsync_on_close \fR=\fPbool
500 If true, sync file contents on close. This differs from \fBend_fsync\fR in that
501 it will happen on every close, not just at the end of the job. Default: false.
503 .BI rwmixcycle \fR=\fPint
504 How many milliseconds before switching between reads and writes for a mixed
505 workload. Default: 500ms.
507 .BI rwmixread \fR=\fPint
508 Percentage of a mixed workload that should be reads. Default: 50.
510 .BI rwmixwrite \fR=\fPint
511 Percentage of a mixed workload that should be writes. If \fBrwmixread\fR and
512 \fBrwmixwrite\fR are given and do not sum to 100%, the latter of the two
513 overrides the first. This may interfere with a given rate setting, if fio is
514 asked to limit reads or writes to a certain rate. If that is the case, then
515 the distribution may be skewed. Default: 50.
518 Normally \fBfio\fR will cover every block of the file when doing random I/O. If
519 this parameter is given, a new offset will be chosen without looking at past
520 I/O history. This parameter is mutually exclusive with \fBverify\fR.
522 .BI softrandommap \fR=\fPbool
523 See \fBnorandommap\fR. If fio runs with the random block map enabled and it
524 fails to allocate the map, if this option is set it will continue without a
525 random block map. As coverage will not be as complete as with random maps, this
526 option is disabled by default.
529 Run job with given nice value. See \fInice\fR\|(2).
532 Set I/O priority value of this job between 0 (highest) and 7 (lowest). See
535 .BI prioclass \fR=\fPint
536 Set I/O priority class. See \fIionice\fR\|(1).
538 .BI thinktime \fR=\fPint
539 Stall job for given number of microseconds between issuing I/Os.
541 .BI thinktime_spin \fR=\fPint
542 Pretend to spend CPU time for given number of microseconds, sleeping the rest
543 of the time specified by \fBthinktime\fR. Only valid if \fBthinktime\fR is set.
545 .BI thinktime_blocks \fR=\fPint
546 Number of blocks to issue before waiting \fBthinktime\fR microseconds.
550 Cap bandwidth used by this job. The number is in bytes/sec, the normal postfix
551 rules apply. You can use \fBrate\fR=500k to limit reads and writes to 500k each,
552 or you can specify read and writes separately. Using \fBrate\fR=1m,500k would
553 limit reads to 1MB/sec and writes to 500KB/sec. Capping only reads or writes
554 can be done with \fBrate\fR=,500k or \fBrate\fR=500k,. The former will only
555 limit writes (to 500KB/sec), the latter will only limit reads.
557 .BI ratemin \fR=\fPint
558 Tell \fBfio\fR to do whatever it can to maintain at least the given bandwidth.
559 Failing to meet this requirement will cause the job to exit. The same format
560 as \fBrate\fR is used for read vs write separation.
562 .BI rate_iops \fR=\fPint
563 Cap the bandwidth to this number of IOPS. Basically the same as rate, just
564 specified independently of bandwidth. The same format as \fBrate\fR is used for
565 read vs write seperation. If \fBblocksize\fR is a range, the smallest block
566 size is used as the metric.
568 .BI rate_iops_min \fR=\fPint
569 If this rate of I/O is not met, the job will exit. The same format as \fBrate\fR
570 is used for read vs write seperation.
572 .BI ratecycle \fR=\fPint
573 Average bandwidth for \fBrate\fR and \fBratemin\fR over this number of
574 milliseconds. Default: 1000ms.
576 .BI cpumask \fR=\fPint
577 Set CPU affinity for this job. \fIint\fR is a bitmask of allowed CPUs the job
578 may run on. See \fBsched_setaffinity\fR\|(2).
580 .BI cpus_allowed \fR=\fPstr
581 Same as \fBcpumask\fR, but allows a comma-delimited list of CPU numbers.
583 .BI startdelay \fR=\fPint
584 Delay start of job for the specified number of seconds.
586 .BI runtime \fR=\fPint
587 Terminate processing after the specified number of seconds.
590 If given, run for the specified \fBruntime\fR duration even if the files are
591 completely read or written. The same workload will be repeated as many times
592 as \fBruntime\fR allows.
594 .BI ramp_time \fR=\fPint
595 If set, fio will run the specified workload for this amount of time before
596 logging any performance numbers. Useful for letting performance settle before
597 logging results, thus minimizing the runtime required for stable results. Note
598 that the \fBramp_time\fR is considered lead in time for a job, thus it will
599 increase the total runtime if a special timeout or runtime is specified.
601 .BI invalidate \fR=\fPbool
602 Invalidate buffer-cache for the file prior to starting I/O. Default: true.
605 Use synchronous I/O for buffered writes. For the majority of I/O engines,
606 this means using O_SYNC. Default: false.
608 .BI iomem \fR=\fPstr "\fR,\fP mem" \fR=\fPstr
609 Allocation method for I/O unit buffer. Allowed values are:
614 Allocate memory with \fImalloc\fR\|(3).
617 Use shared memory buffers allocated through \fIshmget\fR\|(2).
620 Same as \fBshm\fR, but use huge pages as backing.
623 Use \fImmap\fR\|(2) for allocation. Uses anonymous memory unless a filename
624 is given after the option in the format `:\fIfile\fR'.
627 Same as \fBmmap\fR, but use huge files as backing.
630 The amount of memory allocated is the maximum allowed \fBblocksize\fR for the
631 job multiplied by \fBiodepth\fR. For \fBshmhuge\fR or \fBmmaphuge\fR to work,
632 the system must have free huge pages allocated. \fBmmaphuge\fR also needs to
633 have hugetlbfs mounted, and \fIfile\fR must point there. At least on Linux,
634 huge pages must be manually allocated. See \fB/proc/sys/vm/nr_hugehages\fR
635 and the documentation for that. Normally you just need to echo an appropriate
636 number, eg echoing 8 will ensure that the OS has 8 huge pages ready for
640 .BI iomem_align \fR=\fPint "\fR,\fP mem_align" \fR=\fPint
641 This indiciates the memory alignment of the IO memory buffers. Note that the
642 given alignment is applied to the first IO unit buffer, if using \fBiodepth\fR
643 the alignment of the following buffers are given by the \fBbs\fR used. In
644 other words, if using a \fBbs\fR that is a multiple of the page sized in the
645 system, all buffers will be aligned to this value. If using a \fBbs\fR that
646 is not page aligned, the alignment of subsequent IO memory buffers is the
647 sum of the \fBiomem_align\fR and \fBbs\fR used.
649 .BI hugepage\-size \fR=\fPint
650 Defines the size of a huge page. Must be at least equal to the system setting.
651 Should be a multiple of 1MB. Default: 4MB.
654 Terminate all jobs when one finishes. Default: wait for each job to finish.
656 .BI bwavgtime \fR=\fPint
657 Average bandwidth calculations over the given time in milliseconds. Default:
660 .BI create_serialize \fR=\fPbool
661 If true, serialize file creation for the jobs. Default: true.
663 .BI create_fsync \fR=\fPbool
664 \fIfsync\fR\|(2) data file after creation. Default: true.
666 .BI create_on_open \fR=\fPbool
667 If true, the files are not created until they are opened for IO by the job.
669 .BI pre_read \fR=\fPbool
670 If this is given, files will be pre-read into memory before starting the given
671 IO operation. This will also clear the \fR \fBinvalidate\fR flag, since it is
672 pointless to pre-read and then drop the cache. This will only work for IO
673 engines that are seekable, since they allow you to read the same data
674 multiple times. Thus it will not work on eg network or splice IO.
676 .BI unlink \fR=\fPbool
677 Unlink job files when done. Default: false.
680 Specifies the number of iterations (runs of the same workload) of this job.
683 .BI do_verify \fR=\fPbool
684 Run the verify phase after a write phase. Only valid if \fBverify\fR is set.
687 .BI verify \fR=\fPstr
688 Method of verifying file contents after each iteration of the job. Allowed
693 .B md5 crc16 crc32 crc32c crc32c-intel crc64 crc7 sha256 sha512 sha1
694 Store appropriate checksum in the header of each block. crc32c-intel is
695 hardware accelerated SSE4.2 driven, falls back to regular crc32c if
696 not supported by the system.
699 Write extra information about each I/O (timestamp, block number, etc.). The
700 block number is verified. See \fBverify_pattern\fR as well.
703 Pretend to verify. Used for testing internals.
706 This option can be used for repeated burn-in tests of a system to make sure
707 that the written data is also correctly read back. If the data direction given
708 is a read or random read, fio will assume that it should verify a previously
709 written file. If the data direction includes any form of write, the verify will
710 be of the newly written data.
713 .BI verify_sort \fR=\fPbool
714 If true, written verify blocks are sorted if \fBfio\fR deems it to be faster to
715 read them back in a sorted manner. Default: true.
717 .BI verify_offset \fR=\fPint
718 Swap the verification header with data somewhere else in the block before
719 writing. It is swapped back before verifying.
721 .BI verify_interval \fR=\fPint
722 Write the verification header for this number of bytes, which should divide
723 \fBblocksize\fR. Default: \fBblocksize\fR.
725 .BI verify_pattern \fR=\fPstr
726 If set, fio will fill the io buffers with this pattern. Fio defaults to filling
727 with totally random bytes, but sometimes it's interesting to fill with a known
728 pattern for io verification purposes. Depending on the width of the pattern,
729 fio will fill 1/2/3/4 bytes of the buffer at the time(it can be either a
730 decimal or a hex number). The verify_pattern if larger than a 32-bit quantity
731 has to be a hex number that starts with either "0x" or "0X". Use with
734 .BI verify_fatal \fR=\fPbool
735 If true, exit the job on the first observed verification failure. Default:
738 .BI verify_dump \fR=\fPbool
739 If set, dump the contents of both the original data block and the data block we
740 read off disk to files. This allows later analysis to inspect just what kind of
741 data corruption occurred. On by default.
743 .BI verify_async \fR=\fPint
744 Fio will normally verify IO inline from the submitting thread. This option
745 takes an integer describing how many async offload threads to create for IO
746 verification instead, causing fio to offload the duty of verifying IO contents
747 to one or more separate threads. If using this offload option, even sync IO
748 engines can benefit from using an \fBiodepth\fR setting higher than 1, as it
749 allows them to have IO in flight while verifies are running.
751 .BI verify_async_cpus \fR=\fPstr
752 Tell fio to set the given CPU affinity on the async IO verification threads.
753 See \fBcpus_allowed\fP for the format used.
755 .BI verify_backlog \fR=\fPint
756 Fio will normally verify the written contents of a job that utilizes verify
757 once that job has completed. In other words, everything is written then
758 everything is read back and verified. You may want to verify continually
759 instead for a variety of reasons. Fio stores the meta data associated with an
760 IO block in memory, so for large verify workloads, quite a bit of memory would
761 be used up holding this meta data. If this option is enabled, fio will write
762 only N blocks before verifying these blocks.
764 .BI verify_backlog_batch \fR=\fPint
765 Control how many blocks fio will verify if verify_backlog is set. If not set,
766 will default to the value of \fBverify_backlog\fR (meaning the entire queue is
767 read back and verified). If \fBverify_backlog_batch\fR is less than
768 \fBverify_backlog\fR then not all blocks will be verified, if
769 \fBverify_backlog_batch\fR is larger than \fBverify_backlog\fR, some blocks
770 will be verified more than once.
772 .B stonewall "\fR,\fP wait_for_previous"
773 Wait for preceding jobs in the job file to exit before starting this one.
774 \fBstonewall\fR implies \fBnew_group\fR.
777 Start a new reporting group. If not given, all jobs in a file will be part
778 of the same reporting group, unless separated by a stonewall.
780 .BI numjobs \fR=\fPint
781 Number of clones (processes/threads performing the same workload) of this job.
785 If set, display per-group reports instead of per-job when \fBnumjobs\fR is
789 Use threads created with \fBpthread_create\fR\|(3) instead of processes created
790 with \fBfork\fR\|(2).
792 .BI zonesize \fR=\fPint
793 Divide file into zones of the specified size in bytes. See \fBzoneskip\fR.
795 .BI zoneskip \fR=\fPint
796 Skip the specified number of bytes when \fBzonesize\fR bytes of data have been
799 .BI write_iolog \fR=\fPstr
800 Write the issued I/O patterns to the specified file. Specify a separate file
801 for each job, otherwise the iologs will be interspersed and the file may be
804 .BI read_iolog \fR=\fPstr
805 Replay the I/O patterns contained in the specified file generated by
806 \fBwrite_iolog\fR, or may be a \fBblktrace\fR binary file.
808 .BI replay_no_stall \fR=\fPint
809 While replaying I/O patterns using \fBread_iolog\fR the default behavior
810 attempts to respect timing information between I/Os. Enabling
811 \fBreplay_no_stall\fR causes I/Os to be replayed as fast as possible while
812 still respecting ordering.
814 .BI replay_redirect \fR=\fPstr
815 While replaying I/O patterns using \fBread_iolog\fR the default behavior
816 is to replay the IOPS onto the major/minor device that each IOP was recorded
817 from. Setting \fBreplay_redirect\fR causes all IOPS to be replayed onto the
818 single specified device regardless of the device it was recorded from.
820 .B write_bw_log \fR=\fPstr
821 If given, write a bandwidth log of the jobs in this job file. Can be used to
822 store data of the bandwidth of the jobs in their lifetime. The included
823 fio_generate_plots script uses gnuplot to turn these text files into nice
824 graphs. See \fBwrite_log_log\fR for behaviour of given filename. For this
825 option, the postfix is _bw.log.
827 .B write_lat_log \fR=\fPstr
828 Same as \fBwrite_bw_log\fR, but writes I/O completion latencies. If no
829 filename is given with this option, the default filename of "jobname_type.log"
830 is used. Even if the filename is given, fio will still append the type of log.
832 .B disable_lat \fR=\fPbool
833 Disable measurements of total latency numbers. Useful only for cutting
834 back the number of calls to gettimeofday, as that does impact performance at
835 really high IOPS rates. Note that to really get rid of a large amount of these
836 calls, this option must be used with disable_slat and disable_bw as well.
838 .B disable_clat \fR=\fPbool
839 Disable measurements of completion latency numbers. See \fBdisable_lat\fR.
841 .B disable_slat \fR=\fPbool
842 Disable measurements of submission latency numbers. See \fBdisable_lat\fR.
844 .B disable_bw_measurement \fR=\fPbool
845 Disable measurements of throughput/bandwidth numbers. See \fBdisable_lat\fR.
847 .BI lockmem \fR=\fPint
848 Pin the specified amount of memory with \fBmlock\fR\|(2). Can be used to
849 simulate a smaller amount of memory.
851 .BI exec_prerun \fR=\fPstr
852 Before running the job, execute the specified command with \fBsystem\fR\|(3).
854 .BI exec_postrun \fR=\fPstr
855 Same as \fBexec_prerun\fR, but the command is executed after the job completes.
857 .BI ioscheduler \fR=\fPstr
858 Attempt to switch the device hosting the file to the specified I/O scheduler.
860 .BI cpuload \fR=\fPint
861 If the job is a CPU cycle-eater, attempt to use the specified percentage of
864 .BI cpuchunks \fR=\fPint
865 If the job is a CPU cycle-eater, split the load into cycles of the
866 given time in milliseconds.
868 .BI disk_util \fR=\fPbool
869 Generate disk utilization statistics if the platform supports it. Default: true.
871 .BI gtod_reduce \fR=\fPbool
872 Enable all of the gettimeofday() reducing options (disable_clat, disable_slat,
873 disable_bw) plus reduce precision of the timeout somewhat to really shrink the
874 gettimeofday() call count. With this option enabled, we only do about 0.4% of
875 the gtod() calls we would have done if all time keeping was enabled.
877 .BI gtod_cpu \fR=\fPint
878 Sometimes it's cheaper to dedicate a single thread of execution to just getting
879 the current time. Fio (and databases, for instance) are very intensive on
880 gettimeofday() calls. With this option, you can set one CPU aside for doing
881 nothing but logging current time to a shared memory location. Then the other
882 threads/processes that run IO workloads need only copy that segment, instead of
883 entering the kernel with a gettimeofday() call. The CPU set aside for doing
884 these time calls will be excluded from other uses. Fio will manually clear it
885 from the CPU mask of other jobs.
887 .BI cgroup \fR=\fPstr
888 Add job to this control group. If it doesn't exist, it will be created.
889 The system must have a mounted cgroup blkio mount point for this to work. If
890 your system doesn't have it mounted, you can do so with:
892 # mount \-t cgroup \-o blkio none /cgroup
894 .BI cgroup_weight \fR=\fPint
895 Set the weight of the cgroup to this value. See the documentation that comes
896 with the kernel, allowed values are in the range of 100..1000.
898 .BI cgroup_nodelete \fR=\fPbool
899 Normally fio will delete the cgroups it has created after the job completion.
900 To override this behavior and to leave cgroups around after the job completion,
901 set cgroup_nodelete=1. This can be useful if one wants to inspect various
902 cgroup files after job completion. Default: false
905 Instead of running as the invoking user, set the user ID to this value before
906 the thread/process does any work.
909 Set group ID, see \fBuid\fR.
911 .BI clat_percentiles \fR=\fPbool
912 Enable the reporting of percentiles of completion latencies.
914 .BI percentile_list \fR=\fPfloat_list
915 Overwrite the default list of percentiles for completion
916 latencies. Each number is a floating number in the range (0,100], and
917 the maximum length of the list is 20. Use ':' to separate the
918 numbers. For example, --percentile_list=99.5:99.9 will cause fio to
919 report the values of completion latency below which 99.5% and 99.9% of
920 the observed latencies fell, respectively.
922 While running, \fBfio\fR will display the status of the created jobs. For
926 Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
929 The characters in the first set of brackets denote the current status of each
930 threads. The possible values are:
936 Setup but not started.
942 Initialized, waiting.
945 Running, doing sequential reads.
948 Running, doing random reads.
951 Running, doing sequential writes.
954 Running, doing random writes.
957 Running, doing mixed sequential reads/writes.
960 Running, doing mixed random reads/writes.
963 Running, currently waiting for \fBfsync\fR\|(2).
966 Running, verifying written data.
969 Exited, not reaped by main thread.
972 Exited, thread reaped.
976 The second set of brackets shows the estimated completion percentage of
977 the current group. The third set shows the read and write I/O rate,
978 respectively. Finally, the estimated run time of the job is displayed.
980 When \fBfio\fR completes (or is interrupted by Ctrl-C), it will show data
981 for each thread, each group of threads, and each disk, in that order.
983 Per-thread statistics first show the threads client number, group-id, and
984 error code. The remaining figures are as follows:
988 Number of megabytes of I/O performed.
991 Average data rate (bandwidth).
997 Submission latency minimum, maximum, average and standard deviation. This is
998 the time it took to submit the I/O.
1001 Completion latency minimum, maximum, average and standard deviation. This
1002 is the time between submission and completion.
1005 Bandwidth minimum, maximum, percentage of aggregate bandwidth received, average
1006 and standard deviation.
1009 CPU usage statistics. Includes user and system time, number of context switches
1010 this thread went through and number of major and minor page faults.
1013 Distribution of I/O depths. Each depth includes everything less than (or equal)
1014 to it, but greater than the previous depth.
1017 Number of read/write requests issued, and number of short read/write requests.
1020 Distribution of I/O completion latencies. The numbers follow the same pattern
1024 The group statistics show:
1029 Number of megabytes I/O performed.
1032 Aggregate bandwidth of threads in the group.
1035 Minimum average bandwidth a thread saw.
1038 Maximum average bandwidth a thread saw.
1041 Shortest runtime of threads in the group.
1044 Longest runtime of threads in the group.
1048 Finally, disk statistics are printed with reads first:
1053 Number of I/Os performed by all groups.
1056 Number of merges in the I/O scheduler.
1059 Number of ticks we kept the disk busy.
1062 Total time spent in the disk queue.
1069 If the \fB\-\-minimal\fR option is given, the results will be printed in a
1070 semicolon-delimited format suitable for scripted use - a job description
1071 (if provided) follows on a new line. Note that the first
1072 number in the line is the version number. If the output has to be changed
1073 for some reason, this number will be incremented by 1 to signify that
1074 change. The fields are:
1077 .B version, jobname, groupid, error
1081 .B KB I/O, bandwidth \fR(KB/s)\fP, runtime \fR(ms)\fP
1085 .B min, max, mean, standard deviation
1089 .B min, max, mean, standard deviation
1093 .B min, max, mean, standard deviation
1097 .B min, max, aggregate percentage of total, mean, standard deviation
1103 .B KB I/O, bandwidth \fR(KB/s)\fP, runtime \fR(ms)\fP
1107 .B min, max, mean, standard deviation
1111 .B min, max, mean, standard deviation
1115 .B min, max, mean, standard deviation
1119 .B min, max, aggregate percentage of total, mean, standard deviation
1125 .B user, system, context switches, major page faults, minor page faults
1128 IO depth distribution:
1130 .B <=1, 2, 4, 8, 16, 32, >=64
1133 IO latency distribution:
1137 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
1141 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
1145 Error Info (dependent on continue_on_error, default off):
1147 .B total # errors, first error code
1150 .B text description (if provided in config - appears on newline)
1154 was written by Jens Axboe <jens.axboe@oracle.com>,
1155 now Jens Axboe <jaxboe@fusionio.com>.
1157 This man page was written by Aaron Carroll <aaronc@cse.unsw.edu.au> based
1158 on documentation by Jens Axboe.
1159 .SH "REPORTING BUGS"
1160 Report bugs to the \fBfio\fR mailing list <fio@vger.kernel.org>.
1163 For further documentation see \fBHOWTO\fR and \fBREADME\fR.
1165 Sample jobfiles are available in the \fBexamples\fR directory.
projects / fio.git / blob