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 \-\-debug \fR=\fPtype
16 Enable verbose tracing of various fio actions. May be `all' for all types
17 or individual types separated by a comma (eg \-\-debug=io,file). `help' will
18 list all available tracing options.
20 .BI \-\-output \fR=\fPfilename
21 Write output to \fIfilename\fR.
23 .BI \-\-runtime \fR=\fPruntime
24 Limit run time to \fIruntime\fR seconds.
27 Generate per-job latency logs.
30 Generate per-job bandwidth logs.
33 Print statistics in a terse, semicolon-delimited format.
36 Display version information and exit.
38 .BI \-\-terse\-version \fR=\fPversion
39 Set terse version output format (Current version 3, or older version 2).
42 Display usage information and exit.
44 .BI \-\-cmdhelp \fR=\fPcommand
45 Print help information for \fIcommand\fR. May be `all' for all commands.
47 .BI \-\-enghelp \fR=\fPioengine[,command]
48 List all commands defined by \fIioengine\fR, or print help for \fIcommand\fR defined by \fIioengine\fR.
50 .BI \-\-showcmd \fR=\fPjobfile
51 Convert \fIjobfile\fR to a set of command-line options.
53 .BI \-\-eta \fR=\fPwhen
54 Specifies when real-time ETA estimate should be printed. \fIwhen\fR may
55 be one of `always', `never' or `auto'.
57 .BI \-\-eta\-newline \fR=\fPtime
58 Force an ETA newline for every `time` period passed.
60 .BI \-\-status\-interval \fR=\fPtime
61 Report full output status every `time` period passed.
64 Turn on safety read-only checks, preventing any attempted write.
66 .BI \-\-section \fR=\fPsec
67 Only run section \fIsec\fR from job file. Multiple of these options can be given, adding more sections to run.
69 .BI \-\-alloc\-size \fR=\fPkb
70 Set the internal smalloc pool size to \fIkb\fP kilobytes.
72 .BI \-\-warnings\-fatal
73 All fio parser warnings are fatal, causing fio to exit with an error.
75 .BI \-\-max\-jobs \fR=\fPnr
76 Set the maximum allowed number of jobs (threads/processes) to support.
78 .BI \-\-server \fR=\fPargs
79 Start a backend server, with \fIargs\fP specifying what to listen to. See client/server section.
81 .BI \-\-daemonize \fR=\fPpidfile
82 Background a fio server, writing the pid to the given pid file.
84 .BI \-\-client \fR=\fPhost
85 Instead of running the jobs locally, send and run them on the given host.
87 .BI \-\-idle\-prof \fR=\fPoption
88 Report cpu idleness on a system or percpu basis (\fIoption\fP=system,percpu) or run unit work calibration only (\fIoption\fP=calibrate).
90 Job files are in `ini' format. They consist of one or more
91 job definitions, which begin with a job name in square brackets and
92 extend to the next job name. The job name can be any ASCII string
93 except `global', which has a special meaning. Following the job name is
94 a sequence of zero or more parameters, one per line, that define the
95 behavior of the job. Any line starting with a `;' or `#' character is
96 considered a comment and ignored.
98 If \fIjobfile\fR is specified as `-', the job file will be read from
101 The global section contains default parameters for jobs specified in the
102 job file. A job is only affected by global sections residing above it,
103 and there may be any number of global sections. Specific job definitions
104 may override any parameter set in global sections.
107 Some parameters may take arguments of a specific type. The types used are:
110 String: a sequence of alphanumeric characters.
113 SI integer: a whole number, possibly containing a suffix denoting the base unit
114 of the value. Accepted suffixes are `k', 'M', 'G', 'T', and 'P', denoting
115 kilo (1024), mega (1024^2), giga (1024^3), tera (1024^4), and peta (1024^5)
116 respectively. The suffix is not case sensitive. If prefixed with '0x', the
117 value is assumed to be base 16 (hexadecimal). A suffix may include a trailing 'b',
118 for instance 'kb' is identical to 'k'. You can specify a base 10 value
119 by using 'KiB', 'MiB', 'GiB', etc. This is useful for disk drives where
120 values are often given in base 10 values. Specifying '30GiB' will get you
124 Boolean: a true or false value. `0' denotes false, `1' denotes true.
127 Integer range: a range of integers specified in the format
128 \fIlower\fR:\fIupper\fR or \fIlower\fR\-\fIupper\fR. \fIlower\fR and
129 \fIupper\fR may contain a suffix as described above. If an option allows two
130 sets of ranges, they are separated with a `,' or `/' character. For example:
134 List of floating numbers: A list of floating numbers, separated by
139 May be used to override the job name. On the command line, this parameter
140 has the special purpose of signalling the start of a new job.
142 .BI description \fR=\fPstr
143 Human-readable description of the job. It is printed when the job is run, but
144 otherwise has no special purpose.
146 .BI directory \fR=\fPstr
147 Prefix filenames with this directory. Used to place files in a location other
150 .BI filename \fR=\fPstr
152 normally makes up a file name based on the job name, thread number, and file
153 number. If you want to share files between threads in a job or several jobs,
154 specify a \fIfilename\fR for each of them to override the default.
155 If the I/O engine is file-based, you can specify
156 a number of files by separating the names with a `:' character. `\-' is a
157 reserved name, meaning stdin or stdout, depending on the read/write direction
160 .BI filename_format \fR=\fPstr
161 If sharing multiple files between jobs, it is usually necessary to have
162 fio generate the exact names that you want. By default, fio will name a file
163 based on the default file format specification of
164 \fBjobname.jobnumber.filenumber\fP. With this option, that can be
165 customized. Fio will recognize and replace the following keywords in this
171 The name of the worker thread or process.
174 The incremental number of the worker thread or process.
177 The incremental number of the file for that worker thread or process.
180 To have dependent jobs share a set of files, this option can be set to
181 have fio generate filenames that are shared between the two. For instance,
182 if \fBtestfiles.$filenum\fR is specified, file number 4 for any job will
183 be named \fBtestfiles.4\fR. The default of \fB$jobname.$jobnum.$filenum\fR
184 will be used if no other format specifier is given.
188 .BI lockfile \fR=\fPstr
189 Fio defaults to not locking any files before it does IO to them. If a file or
190 file descriptor is shared, fio can serialize IO to that file to make the end
191 result consistent. This is usual for emulating real workloads that share files.
197 No locking. This is the default.
200 Only one thread or process may do IO at the time, excluding all others.
203 Read-write locking on the file. Many readers may access the file at the same
204 time, but writes get exclusive access.
208 .BI opendir \fR=\fPstr
209 Recursively open any files below directory \fIstr\fR.
211 .BI readwrite \fR=\fPstr "\fR,\fP rw" \fR=\fPstr
212 Type of I/O pattern. Accepted values are:
229 Mixed sequential reads and writes.
232 Mixed random reads and writes.
235 For mixed I/O, the default split is 50/50. For certain types of io the result
236 may still be skewed a bit, since the speed may be different. It is possible to
237 specify a number of IO's to do before getting a new offset, this is done by
238 appending a `:\fI<nr>\fR to the end of the string given. For a random read, it
239 would look like \fBrw=randread:8\fR for passing in an offset modifier with a
240 value of 8. If the postfix is used with a sequential IO pattern, then the value
241 specified will be added to the generated offset for each IO. For instance,
242 using \fBrw=write:4k\fR will skip 4k for every write. It turns sequential IO
243 into sequential IO with holes. See the \fBrw_sequencer\fR option.
246 .BI rw_sequencer \fR=\fPstr
247 If an offset modifier is given by appending a number to the \fBrw=<str>\fR line,
248 then this option controls how that number modifies the IO offset being
249 generated. Accepted values are:
254 Generate sequential offset
257 Generate the same offset
260 \fBsequential\fR is only useful for random IO, where fio would normally
261 generate a new random offset for every IO. If you append eg 8 to randread, you
262 would get a new random offset for every 8 IO's. The result would be a seek for
263 only every 8 IO's, instead of for every IO. Use \fBrw=randread:8\fR to specify
264 that. As sequential IO is already sequential, setting \fBsequential\fR for that
265 would not result in any differences. \fBidentical\fR behaves in a similar
266 fashion, except it sends the same offset 8 number of times before generating a
271 .BI kb_base \fR=\fPint
272 The base unit for a kilobyte. The defacto base is 2^10, 1024. Storage
273 manufacturers like to use 10^3 or 1000 as a base ten unit instead, for obvious
274 reasons. Allow values are 1024 or 1000, with 1024 being the default.
276 .BI unified_rw_reporting \fR=\fPbool
277 Fio normally reports statistics on a per data direction basis, meaning that
278 read, write, and trim are accounted and reported separately. If this option is
279 set, the fio will sum the results and report them as "mixed" instead.
281 .BI randrepeat \fR=\fPbool
282 Seed the random number generator in a predictable way so results are repeatable
283 across runs. Default: true.
285 .BI use_os_rand \fR=\fPbool
286 Fio can either use the random generator supplied by the OS to generator random
287 offsets, or it can use it's own internal generator (based on Tausworthe).
288 Default is to use the internal generator, which is often of better quality and
289 faster. Default: false.
291 .BI fallocate \fR=\fPstr
292 Whether pre-allocation is performed when laying down files. Accepted values
298 Do not pre-allocate space.
301 Pre-allocate via posix_fallocate().
304 Pre-allocate via fallocate() with FALLOC_FL_KEEP_SIZE set.
307 Backward-compatible alias for 'none'.
310 Backward-compatible alias for 'posix'.
313 May not be available on all supported platforms. 'keep' is only
314 available on Linux. If using ZFS on Solaris this must be set to 'none'
315 because ZFS doesn't support it. Default: 'posix'.
318 .BI fadvise_hint \fR=\fPbool
319 Use of \fIposix_fadvise\fR\|(2) to advise the kernel what I/O patterns
320 are likely to be issued. Default: true.
323 Total size of I/O for this job. \fBfio\fR will run until this many bytes have
324 been transfered, unless limited by other options (\fBruntime\fR, for instance).
325 Unless \fBnrfiles\fR and \fBfilesize\fR options are given, this amount will be
326 divided between the available files for the job. If not set, fio will use the
327 full size of the given files or devices. If the the files do not exist, size
328 must be given. It is also possible to give size as a percentage between 1 and
329 100. If size=20% is given, fio will use 20% of the full size of the given files
332 .BI fill_device \fR=\fPbool "\fR,\fB fill_fs" \fR=\fPbool
333 Sets size to something really large and waits for ENOSPC (no space left on
334 device) as the terminating condition. Only makes sense with sequential write.
335 For a read workload, the mount point will be filled first then IO started on
336 the result. This option doesn't make sense if operating on a raw device node,
337 since the size of that is already known by the file system. Additionally,
338 writing beyond end-of-device will not return ENOSPC there.
340 .BI filesize \fR=\fPirange
341 Individual file sizes. May be a range, in which case \fBfio\fR will select sizes
342 for files at random within the given range, limited to \fBsize\fR in total (if
343 that is given). If \fBfilesize\fR is not specified, each created file is the
346 .BI blocksize \fR=\fPint[,int] "\fR,\fB bs" \fR=\fPint[,int]
347 Block size for I/O units. Default: 4k. Values for reads, writes, and trims
348 can be specified separately in the format \fIread\fR,\fIwrite\fR,\fItrim\fR
349 either of which may be empty to leave that value at its default. If a trailing
350 comma isn't given, the remainder will inherit the last value set.
352 .BI blocksize_range \fR=\fPirange[,irange] "\fR,\fB bsrange" \fR=\fPirange[,irange]
353 Specify a range of I/O block sizes. The issued I/O unit will always be a
354 multiple of the minimum size, unless \fBblocksize_unaligned\fR is set. Applies
355 to both reads and writes if only one range is given, but can be specified
356 separately with a comma seperating the values. Example: bsrange=1k-4k,2k-8k.
357 Also (see \fBblocksize\fR).
359 .BI bssplit \fR=\fPstr
360 This option allows even finer grained control of the block sizes issued,
361 not just even splits between them. With this option, you can weight various
362 block sizes for exact control of the issued IO for a job that has mixed
363 block sizes. The format of the option is bssplit=blocksize/percentage,
364 optionally adding as many definitions as needed separated by a colon.
365 Example: bssplit=4k/10:64k/50:32k/40 would issue 50% 64k blocks, 10% 4k
366 blocks and 40% 32k blocks. \fBbssplit\fR also supports giving separate
367 splits to reads and writes. The format is identical to what the
368 \fBbs\fR option accepts, the read and write parts are separated with a
371 .B blocksize_unaligned\fR,\fP bs_unaligned
372 If set, any size in \fBblocksize_range\fR may be used. This typically won't
373 work with direct I/O, as that normally requires sector alignment.
375 .BI blockalign \fR=\fPint[,int] "\fR,\fB ba" \fR=\fPint[,int]
376 At what boundary to align random IO offsets. Defaults to the same as 'blocksize'
377 the minimum blocksize given. Minimum alignment is typically 512b
378 for using direct IO, though it usually depends on the hardware block size.
379 This option is mutually exclusive with using a random map for files, so it
380 will turn off that option.
383 Initialise buffers with all zeros. Default: fill buffers with random data.
386 If this option is given, fio will refill the IO buffers on every submit. The
387 default is to only fill it at init time and reuse that data. Only makes sense
388 if zero_buffers isn't specified, naturally. If data verification is enabled,
389 refill_buffers is also automatically enabled.
391 .BI scramble_buffers \fR=\fPbool
392 If \fBrefill_buffers\fR is too costly and the target is using data
393 deduplication, then setting this option will slightly modify the IO buffer
394 contents to defeat normal de-dupe attempts. This is not enough to defeat
395 more clever block compression attempts, but it will stop naive dedupe
396 of blocks. Default: true.
398 .BI buffer_compress_percentage \fR=\fPint
399 If this is set, then fio will attempt to provide IO buffer content (on WRITEs)
400 that compress to the specified level. Fio does this by providing a mix of
401 random data and zeroes. Note that this is per block size unit, for file/disk
402 wide compression level that matches this setting, you'll also want to set
403 \fBrefill_buffers\fR.
405 .BI buffer_compress_chunk \fR=\fPint
406 See \fBbuffer_compress_percentage\fR. This setting allows fio to manage how
407 big the ranges of random data and zeroed data is. Without this set, fio will
408 provide \fBbuffer_compress_percentage\fR of blocksize random data, followed by
409 the remaining zeroed. With this set to some chunk size smaller than the block
410 size, fio can alternate random and zeroed data throughout the IO buffer.
412 .BI nrfiles \fR=\fPint
413 Number of files to use for this job. Default: 1.
415 .BI openfiles \fR=\fPint
416 Number of files to keep open at the same time. Default: \fBnrfiles\fR.
418 .BI file_service_type \fR=\fPstr
419 Defines how files to service are selected. The following types are defined:
424 Choose a file at random
427 Round robin over open files (default).
429 Do each file in the set sequentially.
432 The number of I/Os to issue before switching a new file can be specified by
433 appending `:\fIint\fR' to the service type.
436 .BI ioengine \fR=\fPstr
437 Defines how the job issues I/O. The following types are defined:
442 Basic \fIread\fR\|(2) or \fIwrite\fR\|(2) I/O. \fIfseek\fR\|(2) is used to
443 position the I/O location.
446 Basic \fIpread\fR\|(2) or \fIpwrite\fR\|(2) I/O.
449 Basic \fIreadv\fR\|(2) or \fIwritev\fR\|(2) I/O. Will emulate queuing by
450 coalescing adjacents IOs into a single submission.
453 Basic \fIpreadv\fR\|(2) or \fIpwritev\fR\|(2) I/O.
456 Linux native asynchronous I/O. This ioengine defines engine specific options.
459 POSIX asynchronous I/O using \fIaio_read\fR\|(3) and \fIaio_write\fR\|(3).
462 Solaris native asynchronous I/O.
465 Windows native asynchronous I/O.
468 File is memory mapped with \fImmap\fR\|(2) and data copied using
472 \fIsplice\fR\|(2) is used to transfer the data and \fIvmsplice\fR\|(2) to
473 transfer data from user-space to the kernel.
476 Use the syslet system calls to make regular read/write asynchronous.
479 SCSI generic sg v3 I/O. May be either synchronous using the SG_IO ioctl, or if
480 the target is an sg character device, we use \fIread\fR\|(2) and
481 \fIwrite\fR\|(2) for asynchronous I/O.
484 Doesn't transfer any data, just pretends to. Mainly used to exercise \fBfio\fR
485 itself and for debugging and testing purposes.
488 Transfer over the network. The protocol to be used can be defined with the
489 \fBprotocol\fR parameter. Depending on the protocol, \fBfilename\fR,
490 \fBhostname\fR, \fBport\fR, or \fBlisten\fR must be specified.
491 This ioengine defines engine specific options.
494 Like \fBnet\fR, but uses \fIsplice\fR\|(2) and \fIvmsplice\fR\|(2) to map data
495 and send/receive. This ioengine defines engine specific options.
498 Doesn't transfer any data, but burns CPU cycles according to \fBcpuload\fR and
499 \fBcpucycles\fR parameters.
502 The GUASI I/O engine is the Generic Userspace Asynchronous Syscall Interface
503 approach to asycnronous I/O.
505 See <http://www.xmailserver.org/guasi\-lib.html>.
508 The RDMA I/O engine supports both RDMA memory semantics (RDMA_WRITE/RDMA_READ)
509 and channel semantics (Send/Recv) for the InfiniBand, RoCE and iWARP protocols.
512 Loads an external I/O engine object file. Append the engine filename as
516 IO engine that does regular linux native fallocate callt to simulate data
517 transfer as fio ioengine
519 DDIR_READ does fallocate(,mode = FALLOC_FL_KEEP_SIZE,)
521 DIR_WRITE does fallocate(,mode = 0)
523 DDIR_TRIM does fallocate(,mode = FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE)
526 IO engine that does regular EXT4_IOC_MOVE_EXT ioctls to simulate defragment activity
527 request to DDIR_WRITE event
532 .BI iodepth \fR=\fPint
533 Number of I/O units to keep in flight against the file. Note that increasing
534 iodepth beyond 1 will not affect synchronous ioengines (except for small
535 degress when verify_async is in use). Even async engines my impose OS
536 restrictions causing the desired depth not to be achieved. This may happen on
537 Linux when using libaio and not setting \fBdirect\fR=1, since buffered IO is
538 not async on that OS. Keep an eye on the IO depth distribution in the
539 fio output to verify that the achieved depth is as expected. Default: 1.
541 .BI iodepth_batch \fR=\fPint
542 Number of I/Os to submit at once. Default: \fBiodepth\fR.
544 .BI iodepth_batch_complete \fR=\fPint
545 This defines how many pieces of IO to retrieve at once. It defaults to 1 which
546 means that we'll ask for a minimum of 1 IO in the retrieval process from the
547 kernel. The IO retrieval will go on until we hit the limit set by
548 \fBiodepth_low\fR. If this variable is set to 0, then fio will always check for
549 completed events before queuing more IO. This helps reduce IO latency, at the
550 cost of more retrieval system calls.
552 .BI iodepth_low \fR=\fPint
553 Low watermark indicating when to start filling the queue again. Default:
556 .BI direct \fR=\fPbool
557 If true, use non-buffered I/O (usually O_DIRECT). Default: false.
559 .BI buffered \fR=\fPbool
560 If true, use buffered I/O. This is the opposite of the \fBdirect\fR parameter.
563 .BI offset \fR=\fPint
564 Offset in the file to start I/O. Data before the offset will not be touched.
566 .BI offset_increment \fR=\fPint
567 If this is provided, then the real offset becomes the
568 offset + offset_increment * thread_number, where the thread number is a counter
569 that starts at 0 and is incremented for each job. This option is useful if
570 there are several jobs which are intended to operate on a file in parallel in
571 disjoint segments, with even spacing between the starting points.
574 How many I/Os to perform before issuing an \fBfsync\fR\|(2) of dirty data. If
575 0, don't sync. Default: 0.
577 .BI fdatasync \fR=\fPint
578 Like \fBfsync\fR, but uses \fBfdatasync\fR\|(2) instead to only sync the
579 data parts of the file. Default: 0.
581 .BI sync_file_range \fR=\fPstr:int
582 Use sync_file_range() for every \fRval\fP number of write operations. Fio will
583 track range of writes that have happened since the last sync_file_range() call.
584 \fRstr\fP can currently be one or more of:
588 SYNC_FILE_RANGE_WAIT_BEFORE
591 SYNC_FILE_RANGE_WRITE
594 SYNC_FILE_RANGE_WRITE
598 So if you do sync_file_range=wait_before,write:8, fio would use
599 \fBSYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE\fP for every 8 writes.
600 Also see the sync_file_range(2) man page. This option is Linux specific.
602 .BI overwrite \fR=\fPbool
603 If writing, setup the file first and do overwrites. Default: false.
605 .BI end_fsync \fR=\fPbool
606 Sync file contents when a write stage has completed. Default: false.
608 .BI fsync_on_close \fR=\fPbool
609 If true, sync file contents on close. This differs from \fBend_fsync\fR in that
610 it will happen on every close, not just at the end of the job. Default: false.
612 .BI rwmixread \fR=\fPint
613 Percentage of a mixed workload that should be reads. Default: 50.
615 .BI rwmixwrite \fR=\fPint
616 Percentage of a mixed workload that should be writes. If \fBrwmixread\fR and
617 \fBrwmixwrite\fR are given and do not sum to 100%, the latter of the two
618 overrides the first. This may interfere with a given rate setting, if fio is
619 asked to limit reads or writes to a certain rate. If that is the case, then
620 the distribution may be skewed. Default: 50.
622 .BI random_distribution \fR=\fPstr:float
623 By default, fio will use a completely uniform random distribution when asked
624 to perform random IO. Sometimes it is useful to skew the distribution in
625 specific ways, ensuring that some parts of the data is more hot than others.
626 Fio includes the following distribution models:
630 Uniform random distribution
640 When using a zipf or pareto distribution, an input value is also needed to
641 define the access pattern. For zipf, this is the zipf theta. For pareto,
642 it's the pareto power. Fio includes a test program, genzipf, that can be
643 used visualize what the given input values will yield in terms of hit rates.
644 If you wanted to use zipf with a theta of 1.2, you would use
645 random_distribution=zipf:1.2 as the option. If a non-uniform model is used,
646 fio will disable use of the random map.
648 .BI percentage_random \fR=\fPint
649 For a random workload, set how big a percentage should be random. This defaults
650 to 100%, in which case the workload is fully random. It can be set from
651 anywhere from 0 to 100. Setting it to 0 would make the workload fully
652 sequential. It is possible to set different values for reads, writes, and
653 trim. To do so, simply use a comma separated list. See \fBblocksize\fR.
656 Normally \fBfio\fR will cover every block of the file when doing random I/O. If
657 this parameter is given, a new offset will be chosen without looking at past
658 I/O history. This parameter is mutually exclusive with \fBverify\fR.
660 .BI softrandommap \fR=\fPbool
661 See \fBnorandommap\fR. If fio runs with the random block map enabled and it
662 fails to allocate the map, if this option is set it will continue without a
663 random block map. As coverage will not be as complete as with random maps, this
664 option is disabled by default.
666 .BI random_generator \fR=\fPstr
667 Fio supports the following engines for generating IO offsets for random IO:
671 Strong 2^88 cycle random number generator
674 Linear feedback shift register generator
678 Tausworthe is a strong random number generator, but it requires tracking on the
679 side if we want to ensure that blocks are only read or written once. LFSR
680 guarantees that we never generate the same offset twice, and it's also less
681 computationally expensive. It's not a true random generator, however, though
682 for IO purposes it's typically good enough. LFSR only works with single block
683 sizes, not with workloads that use multiple block sizes. If used with such a
684 workload, fio may read or write some blocks multiple times.
687 Run job with given nice value. See \fInice\fR\|(2).
690 Set I/O priority value of this job between 0 (highest) and 7 (lowest). See
693 .BI prioclass \fR=\fPint
694 Set I/O priority class. See \fIionice\fR\|(1).
696 .BI thinktime \fR=\fPint
697 Stall job for given number of microseconds between issuing I/Os.
699 .BI thinktime_spin \fR=\fPint
700 Pretend to spend CPU time for given number of microseconds, sleeping the rest
701 of the time specified by \fBthinktime\fR. Only valid if \fBthinktime\fR is set.
703 .BI thinktime_blocks \fR=\fPint
704 Only valid if thinktime is set - control how many blocks to issue, before
705 waiting \fBthinktime\fR microseconds. If not set, defaults to 1 which will
706 make fio wait \fBthinktime\fR microseconds after every block. This
707 effectively makes any queue depth setting redundant, since no more than 1 IO
708 will be queued before we have to complete it and do our thinktime. In other
709 words, this setting effectively caps the queue depth if the latter is larger.
713 Cap bandwidth used by this job. The number is in bytes/sec, the normal postfix
714 rules apply. You can use \fBrate\fR=500k to limit reads and writes to 500k each,
715 or you can specify read and writes separately. Using \fBrate\fR=1m,500k would
716 limit reads to 1MB/sec and writes to 500KB/sec. Capping only reads or writes
717 can be done with \fBrate\fR=,500k or \fBrate\fR=500k,. The former will only
718 limit writes (to 500KB/sec), the latter will only limit reads.
720 .BI ratemin \fR=\fPint
721 Tell \fBfio\fR to do whatever it can to maintain at least the given bandwidth.
722 Failing to meet this requirement will cause the job to exit. The same format
723 as \fBrate\fR is used for read vs write separation.
725 .BI rate_iops \fR=\fPint
726 Cap the bandwidth to this number of IOPS. Basically the same as rate, just
727 specified independently of bandwidth. The same format as \fBrate\fR is used for
728 read vs write seperation. If \fBblocksize\fR is a range, the smallest block
729 size is used as the metric.
731 .BI rate_iops_min \fR=\fPint
732 If this rate of I/O is not met, the job will exit. The same format as \fBrate\fR
733 is used for read vs write seperation.
735 .BI ratecycle \fR=\fPint
736 Average bandwidth for \fBrate\fR and \fBratemin\fR over this number of
737 milliseconds. Default: 1000ms.
739 .BI max_latency \fR=\fPint
740 If set, fio will exit the job if it exceeds this maximum latency. It will exit
743 .BI cpumask \fR=\fPint
744 Set CPU affinity for this job. \fIint\fR is a bitmask of allowed CPUs the job
745 may run on. See \fBsched_setaffinity\fR\|(2).
747 .BI cpus_allowed \fR=\fPstr
748 Same as \fBcpumask\fR, but allows a comma-delimited list of CPU numbers.
750 .BI numa_cpu_nodes \fR=\fPstr
751 Set this job running on spcified NUMA nodes' CPUs. The arguments allow
752 comma delimited list of cpu numbers, A-B ranges, or 'all'.
754 .BI numa_mem_policy \fR=\fPstr
755 Set this job's memory policy and corresponding NUMA nodes. Format of
759 .B <mode>[:<nodelist>]
762 is one of the following memory policy:
764 .B default, prefer, bind, interleave, local
767 For \fBdefault\fR and \fBlocal\fR memory policy, no \fBnodelist\fR is
768 needed to be specified. For \fBprefer\fR, only one node is
769 allowed. For \fBbind\fR and \fBinterleave\fR, \fBnodelist\fR allows
770 comma delimited list of numbers, A-B ranges, or 'all'.
772 .BI startdelay \fR=\fPint
773 Delay start of job for the specified number of seconds.
775 .BI runtime \fR=\fPint
776 Terminate processing after the specified number of seconds.
779 If given, run for the specified \fBruntime\fR duration even if the files are
780 completely read or written. The same workload will be repeated as many times
781 as \fBruntime\fR allows.
783 .BI ramp_time \fR=\fPint
784 If set, fio will run the specified workload for this amount of time before
785 logging any performance numbers. Useful for letting performance settle before
786 logging results, thus minimizing the runtime required for stable results. Note
787 that the \fBramp_time\fR is considered lead in time for a job, thus it will
788 increase the total runtime if a special timeout or runtime is specified.
790 .BI invalidate \fR=\fPbool
791 Invalidate buffer-cache for the file prior to starting I/O. Default: true.
794 Use synchronous I/O for buffered writes. For the majority of I/O engines,
795 this means using O_SYNC. Default: false.
797 .BI iomem \fR=\fPstr "\fR,\fP mem" \fR=\fPstr
798 Allocation method for I/O unit buffer. Allowed values are:
803 Allocate memory with \fImalloc\fR\|(3).
806 Use shared memory buffers allocated through \fIshmget\fR\|(2).
809 Same as \fBshm\fR, but use huge pages as backing.
812 Use \fImmap\fR\|(2) for allocation. Uses anonymous memory unless a filename
813 is given after the option in the format `:\fIfile\fR'.
816 Same as \fBmmap\fR, but use huge files as backing.
819 The amount of memory allocated is the maximum allowed \fBblocksize\fR for the
820 job multiplied by \fBiodepth\fR. For \fBshmhuge\fR or \fBmmaphuge\fR to work,
821 the system must have free huge pages allocated. \fBmmaphuge\fR also needs to
822 have hugetlbfs mounted, and \fIfile\fR must point there. At least on Linux,
823 huge pages must be manually allocated. See \fB/proc/sys/vm/nr_hugehages\fR
824 and the documentation for that. Normally you just need to echo an appropriate
825 number, eg echoing 8 will ensure that the OS has 8 huge pages ready for
829 .BI iomem_align \fR=\fPint "\fR,\fP mem_align" \fR=\fPint
830 This indiciates the memory alignment of the IO memory buffers. Note that the
831 given alignment is applied to the first IO unit buffer, if using \fBiodepth\fR
832 the alignment of the following buffers are given by the \fBbs\fR used. In
833 other words, if using a \fBbs\fR that is a multiple of the page sized in the
834 system, all buffers will be aligned to this value. If using a \fBbs\fR that
835 is not page aligned, the alignment of subsequent IO memory buffers is the
836 sum of the \fBiomem_align\fR and \fBbs\fR used.
838 .BI hugepage\-size \fR=\fPint
839 Defines the size of a huge page. Must be at least equal to the system setting.
840 Should be a multiple of 1MB. Default: 4MB.
843 Terminate all jobs when one finishes. Default: wait for each job to finish.
845 .BI bwavgtime \fR=\fPint
846 Average bandwidth calculations over the given time in milliseconds. Default:
849 .BI iopsavgtime \fR=\fPint
850 Average IOPS calculations over the given time in milliseconds. Default:
853 .BI create_serialize \fR=\fPbool
854 If true, serialize file creation for the jobs. Default: true.
856 .BI create_fsync \fR=\fPbool
857 \fIfsync\fR\|(2) data file after creation. Default: true.
859 .BI create_on_open \fR=\fPbool
860 If true, the files are not created until they are opened for IO by the job.
862 .BI create_only \fR=\fPbool
863 If true, fio will only run the setup phase of the job. If files need to be
864 laid out or updated on disk, only that will be done. The actual job contents
867 .BI pre_read \fR=\fPbool
868 If this is given, files will be pre-read into memory before starting the given
869 IO operation. This will also clear the \fR \fBinvalidate\fR flag, since it is
870 pointless to pre-read and then drop the cache. This will only work for IO
871 engines that are seekable, since they allow you to read the same data
872 multiple times. Thus it will not work on eg network or splice IO.
874 .BI unlink \fR=\fPbool
875 Unlink job files when done. Default: false.
878 Specifies the number of iterations (runs of the same workload) of this job.
881 .BI do_verify \fR=\fPbool
882 Run the verify phase after a write phase. Only valid if \fBverify\fR is set.
885 .BI verify \fR=\fPstr
886 Method of verifying file contents after each iteration of the job. Allowed
891 .B md5 crc16 crc32 crc32c crc32c-intel crc64 crc7 sha256 sha512 sha1
892 Store appropriate checksum in the header of each block. crc32c-intel is
893 hardware accelerated SSE4.2 driven, falls back to regular crc32c if
894 not supported by the system.
897 Write extra information about each I/O (timestamp, block number, etc.). The
898 block number is verified. See \fBverify_pattern\fR as well.
901 Pretend to verify. Used for testing internals.
904 This option can be used for repeated burn-in tests of a system to make sure
905 that the written data is also correctly read back. If the data direction given
906 is a read or random read, fio will assume that it should verify a previously
907 written file. If the data direction includes any form of write, the verify will
908 be of the newly written data.
911 .BI verify_sort \fR=\fPbool
912 If true, written verify blocks are sorted if \fBfio\fR deems it to be faster to
913 read them back in a sorted manner. Default: true.
915 .BI verify_offset \fR=\fPint
916 Swap the verification header with data somewhere else in the block before
917 writing. It is swapped back before verifying.
919 .BI verify_interval \fR=\fPint
920 Write the verification header for this number of bytes, which should divide
921 \fBblocksize\fR. Default: \fBblocksize\fR.
923 .BI verify_pattern \fR=\fPstr
924 If set, fio will fill the io buffers with this pattern. Fio defaults to filling
925 with totally random bytes, but sometimes it's interesting to fill with a known
926 pattern for io verification purposes. Depending on the width of the pattern,
927 fio will fill 1/2/3/4 bytes of the buffer at the time(it can be either a
928 decimal or a hex number). The verify_pattern if larger than a 32-bit quantity
929 has to be a hex number that starts with either "0x" or "0X". Use with
932 .BI verify_fatal \fR=\fPbool
933 If true, exit the job on the first observed verification failure. Default:
936 .BI verify_dump \fR=\fPbool
937 If set, dump the contents of both the original data block and the data block we
938 read off disk to files. This allows later analysis to inspect just what kind of
939 data corruption occurred. Off by default.
941 .BI verify_async \fR=\fPint
942 Fio will normally verify IO inline from the submitting thread. This option
943 takes an integer describing how many async offload threads to create for IO
944 verification instead, causing fio to offload the duty of verifying IO contents
945 to one or more separate threads. If using this offload option, even sync IO
946 engines can benefit from using an \fBiodepth\fR setting higher than 1, as it
947 allows them to have IO in flight while verifies are running.
949 .BI verify_async_cpus \fR=\fPstr
950 Tell fio to set the given CPU affinity on the async IO verification threads.
951 See \fBcpus_allowed\fP for the format used.
953 .BI verify_backlog \fR=\fPint
954 Fio will normally verify the written contents of a job that utilizes verify
955 once that job has completed. In other words, everything is written then
956 everything is read back and verified. You may want to verify continually
957 instead for a variety of reasons. Fio stores the meta data associated with an
958 IO block in memory, so for large verify workloads, quite a bit of memory would
959 be used up holding this meta data. If this option is enabled, fio will write
960 only N blocks before verifying these blocks.
962 .BI verify_backlog_batch \fR=\fPint
963 Control how many blocks fio will verify if verify_backlog is set. If not set,
964 will default to the value of \fBverify_backlog\fR (meaning the entire queue is
965 read back and verified). If \fBverify_backlog_batch\fR is less than
966 \fBverify_backlog\fR then not all blocks will be verified, if
967 \fBverify_backlog_batch\fR is larger than \fBverify_backlog\fR, some blocks
968 will be verified more than once.
970 .B stonewall "\fR,\fP wait_for_previous"
971 Wait for preceding jobs in the job file to exit before starting this one.
972 \fBstonewall\fR implies \fBnew_group\fR.
975 Start a new reporting group. If not given, all jobs in a file will be part
976 of the same reporting group, unless separated by a stonewall.
978 .BI numjobs \fR=\fPint
979 Number of clones (processes/threads performing the same workload) of this job.
983 If set, display per-group reports instead of per-job when \fBnumjobs\fR is
987 Use threads created with \fBpthread_create\fR\|(3) instead of processes created
988 with \fBfork\fR\|(2).
990 .BI zonesize \fR=\fPint
991 Divide file into zones of the specified size in bytes. See \fBzoneskip\fR.
993 .BI zoneskip \fR=\fPint
994 Skip the specified number of bytes when \fBzonesize\fR bytes of data have been
997 .BI write_iolog \fR=\fPstr
998 Write the issued I/O patterns to the specified file. Specify a separate file
999 for each job, otherwise the iologs will be interspersed and the file may be
1002 .BI read_iolog \fR=\fPstr
1003 Replay the I/O patterns contained in the specified file generated by
1004 \fBwrite_iolog\fR, or may be a \fBblktrace\fR binary file.
1006 .BI replay_no_stall \fR=\fPint
1007 While replaying I/O patterns using \fBread_iolog\fR the default behavior
1008 attempts to respect timing information between I/Os. Enabling
1009 \fBreplay_no_stall\fR causes I/Os to be replayed as fast as possible while
1010 still respecting ordering.
1012 .BI replay_redirect \fR=\fPstr
1013 While replaying I/O patterns using \fBread_iolog\fR the default behavior
1014 is to replay the IOPS onto the major/minor device that each IOP was recorded
1015 from. Setting \fBreplay_redirect\fR causes all IOPS to be replayed onto the
1016 single specified device regardless of the device it was recorded from.
1018 .BI write_bw_log \fR=\fPstr
1019 If given, write a bandwidth log of the jobs in this job file. Can be used to
1020 store data of the bandwidth of the jobs in their lifetime. The included
1021 fio_generate_plots script uses gnuplot to turn these text files into nice
1022 graphs. See \fBwrite_log_log\fR for behaviour of given filename. For this
1023 option, the postfix is _bw.log.
1025 .BI write_lat_log \fR=\fPstr
1026 Same as \fBwrite_bw_log\fR, but writes I/O completion latencies. If no
1027 filename is given with this option, the default filename of "jobname_type.log"
1028 is used. Even if the filename is given, fio will still append the type of log.
1030 .BI write_iops_log \fR=\fPstr
1031 Same as \fBwrite_bw_log\fR, but writes IOPS. If no filename is given with this
1032 option, the default filename of "jobname_type.log" is used. Even if the
1033 filename is given, fio will still append the type of log.
1035 .BI log_avg_msec \fR=\fPint
1036 By default, fio will log an entry in the iops, latency, or bw log for every
1037 IO that completes. When writing to the disk log, that can quickly grow to a
1038 very large size. Setting this option makes fio average the each log entry
1039 over the specified period of time, reducing the resolution of the log.
1042 .BI disable_lat \fR=\fPbool
1043 Disable measurements of total latency numbers. Useful only for cutting
1044 back the number of calls to gettimeofday, as that does impact performance at
1045 really high IOPS rates. Note that to really get rid of a large amount of these
1046 calls, this option must be used with disable_slat and disable_bw as well.
1048 .BI disable_clat \fR=\fPbool
1049 Disable measurements of completion latency numbers. See \fBdisable_lat\fR.
1051 .BI disable_slat \fR=\fPbool
1052 Disable measurements of submission latency numbers. See \fBdisable_lat\fR.
1054 .BI disable_bw_measurement \fR=\fPbool
1055 Disable measurements of throughput/bandwidth numbers. See \fBdisable_lat\fR.
1057 .BI lockmem \fR=\fPint
1058 Pin the specified amount of memory with \fBmlock\fR\|(2). Can be used to
1059 simulate a smaller amount of memory. The amount specified is per worker.
1061 .BI exec_prerun \fR=\fPstr
1062 Before running the job, execute the specified command with \fBsystem\fR\|(3).
1064 Output is redirected in a file called \fBjobname.prerun.txt\fR
1067 .BI exec_postrun \fR=\fPstr
1068 Same as \fBexec_prerun\fR, but the command is executed after the job completes.
1070 Output is redirected in a file called \fBjobname.postrun.txt\fR
1073 .BI ioscheduler \fR=\fPstr
1074 Attempt to switch the device hosting the file to the specified I/O scheduler.
1076 .BI cpuload \fR=\fPint
1077 If the job is a CPU cycle-eater, attempt to use the specified percentage of
1080 .BI cpuchunks \fR=\fPint
1081 If the job is a CPU cycle-eater, split the load into cycles of the
1082 given time in milliseconds.
1084 .BI disk_util \fR=\fPbool
1085 Generate disk utilization statistics if the platform supports it. Default: true.
1087 .BI clocksource \fR=\fPstr
1088 Use the given clocksource as the base of timing. The supported options are:
1098 Internal CPU clock source
1102 \fBcpu\fR is the preferred clocksource if it is reliable, as it is very fast
1103 (and fio is heavy on time calls). Fio will automatically use this clocksource
1104 if it's supported and considered reliable on the system it is running on,
1105 unless another clocksource is specifically set. For x86/x86-64 CPUs, this
1106 means supporting TSC Invariant.
1108 .BI gtod_reduce \fR=\fPbool
1109 Enable all of the gettimeofday() reducing options (disable_clat, disable_slat,
1110 disable_bw) plus reduce precision of the timeout somewhat to really shrink the
1111 gettimeofday() call count. With this option enabled, we only do about 0.4% of
1112 the gtod() calls we would have done if all time keeping was enabled.
1114 .BI gtod_cpu \fR=\fPint
1115 Sometimes it's cheaper to dedicate a single thread of execution to just getting
1116 the current time. Fio (and databases, for instance) are very intensive on
1117 gettimeofday() calls. With this option, you can set one CPU aside for doing
1118 nothing but logging current time to a shared memory location. Then the other
1119 threads/processes that run IO workloads need only copy that segment, instead of
1120 entering the kernel with a gettimeofday() call. The CPU set aside for doing
1121 these time calls will be excluded from other uses. Fio will manually clear it
1122 from the CPU mask of other jobs.
1124 .BI ignore_error \fR=\fPstr
1125 Sometimes you want to ignore some errors during test in that case you can specify
1126 error list for each error type.
1128 ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST
1130 errors for given error type is separated with ':'.
1131 Error may be symbol ('ENOSPC', 'ENOMEM') or an integer.
1133 Example: ignore_error=EAGAIN,ENOSPC:122 .
1135 This option will ignore EAGAIN from READ, and ENOSPC and 122(EDQUOT) from WRITE.
1137 .BI error_dump \fR=\fPbool
1138 If set dump every error even if it is non fatal, true by default. If disabled
1139 only fatal error will be dumped
1141 .BI cgroup \fR=\fPstr
1142 Add job to this control group. If it doesn't exist, it will be created.
1143 The system must have a mounted cgroup blkio mount point for this to work. If
1144 your system doesn't have it mounted, you can do so with:
1146 # mount \-t cgroup \-o blkio none /cgroup
1148 .BI cgroup_weight \fR=\fPint
1149 Set the weight of the cgroup to this value. See the documentation that comes
1150 with the kernel, allowed values are in the range of 100..1000.
1152 .BI cgroup_nodelete \fR=\fPbool
1153 Normally fio will delete the cgroups it has created after the job completion.
1154 To override this behavior and to leave cgroups around after the job completion,
1155 set cgroup_nodelete=1. This can be useful if one wants to inspect various
1156 cgroup files after job completion. Default: false
1159 Instead of running as the invoking user, set the user ID to this value before
1160 the thread/process does any work.
1163 Set group ID, see \fBuid\fR.
1165 .BI flow_id \fR=\fPint
1166 The ID of the flow. If not specified, it defaults to being a global flow. See
1170 Weight in token-based flow control. If this value is used, then there is a
1171 \fBflow counter\fR which is used to regulate the proportion of activity between
1172 two or more jobs. fio attempts to keep this flow counter near zero. The
1173 \fBflow\fR parameter stands for how much should be added or subtracted to the
1174 flow counter on each iteration of the main I/O loop. That is, if one job has
1175 \fBflow=8\fR and another job has \fBflow=-1\fR, then there will be a roughly
1176 1:8 ratio in how much one runs vs the other.
1178 .BI flow_watermark \fR=\fPint
1179 The maximum value that the absolute value of the flow counter is allowed to
1180 reach before the job must wait for a lower value of the counter.
1182 .BI flow_sleep \fR=\fPint
1183 The period of time, in microseconds, to wait after the flow watermark has been
1184 exceeded before retrying operations
1186 .BI clat_percentiles \fR=\fPbool
1187 Enable the reporting of percentiles of completion latencies.
1189 .BI percentile_list \fR=\fPfloat_list
1190 Overwrite the default list of percentiles for completion
1191 latencies. Each number is a floating number in the range (0,100], and
1192 the maximum length of the list is 20. Use ':' to separate the
1193 numbers. For example, \-\-percentile_list=99.5:99.9 will cause fio to
1194 report the values of completion latency below which 99.5% and 99.9% of
1195 the observed latencies fell, respectively.
1196 .SS "Ioengine Parameters List"
1197 Some parameters are only valid when a specific ioengine is in use. These are
1198 used identically to normal parameters, with the caveat that when used on the
1199 command line, the must come after the ioengine that defines them is selected.
1201 .BI (cpu)cpuload \fR=\fPint
1202 Attempt to use the specified percentage of CPU cycles.
1204 .BI (cpu)cpuchunks \fR=\fPint
1205 Split the load into cycles of the given time. In microseconds.
1207 .BI (libaio)userspace_reap
1208 Normally, with the libaio engine in use, fio will use
1209 the io_getevents system call to reap newly returned events.
1210 With this flag turned on, the AIO ring will be read directly
1211 from user-space to reap events. The reaping mode is only
1212 enabled when polling for a minimum of 0 events (eg when
1213 iodepth_batch_complete=0).
1215 .BI (net,netsplice)hostname \fR=\fPstr
1216 The host name or IP address to use for TCP or UDP based IO.
1217 If the job is a TCP listener or UDP reader, the hostname is not
1218 used and must be omitted unless it is a valid UDP multicast address.
1220 .BI (net,netsplice)port \fR=\fPint
1221 The TCP or UDP port to bind to or connect to.
1223 .BI (net,netsplice)interface \fR=\fPstr
1224 The IP address of the network interface used to send or receive UDP multicast
1227 .BI (net,netsplice)ttl \fR=\fPint
1228 Time-to-live value for outgoing UDP multicast packets. Default: 1
1230 .BI (net,netsplice)nodelay \fR=\fPbool
1231 Set TCP_NODELAY on TCP connections.
1233 .BI (net,netsplice)protocol \fR=\fPstr "\fR,\fP proto" \fR=\fPstr
1234 The network protocol to use. Accepted values are:
1239 Transmission control protocol
1242 User datagram protocol
1248 When the protocol is TCP or UDP, the port must also be given,
1249 as well as the hostname if the job is a TCP listener or UDP
1250 reader. For unix sockets, the normal filename option should be
1251 used and the port is invalid.
1254 .BI (net,netsplice)listen
1255 For TCP network connections, tell fio to listen for incoming
1256 connections rather than initiating an outgoing connection. The
1257 hostname must be omitted if this option is used.
1259 .BI (net, pingpong) \fR=\fPbool
1260 Normaly a network writer will just continue writing data, and a network reader
1261 will just consume packages. If pingpong=1 is set, a writer will send its normal
1262 payload to the reader, then wait for the reader to send the same payload back.
1263 This allows fio to measure network latencies. The submission and completion
1264 latencies then measure local time spent sending or receiving, and the
1265 completion latency measures how long it took for the other end to receive and
1266 send back. For UDP multicast traffic pingpong=1 should only be set for a single
1267 reader when multiple readers are listening to the same address.
1269 .BI (e4defrag,donorname) \fR=\fPstr
1270 File will be used as a block donor (swap extents between files)
1272 .BI (e4defrag,inplace) \fR=\fPint
1273 Configure donor file block allocation strategy
1276 Preallocate donor's file on init
1279 allocate space immidietly inside defragment event, and free right after event
1283 While running, \fBfio\fR will display the status of the created jobs. For
1287 Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
1290 The characters in the first set of brackets denote the current status of each
1291 threads. The possible values are:
1297 Setup but not started.
1303 Initialized, waiting.
1306 Running, doing sequential reads.
1309 Running, doing random reads.
1312 Running, doing sequential writes.
1315 Running, doing random writes.
1318 Running, doing mixed sequential reads/writes.
1321 Running, doing mixed random reads/writes.
1324 Running, currently waiting for \fBfsync\fR\|(2).
1327 Running, verifying written data.
1330 Exited, not reaped by main thread.
1333 Exited, thread reaped.
1337 The second set of brackets shows the estimated completion percentage of
1338 the current group. The third set shows the read and write I/O rate,
1339 respectively. Finally, the estimated run time of the job is displayed.
1341 When \fBfio\fR completes (or is interrupted by Ctrl-C), it will show data
1342 for each thread, each group of threads, and each disk, in that order.
1344 Per-thread statistics first show the threads client number, group-id, and
1345 error code. The remaining figures are as follows:
1349 Number of megabytes of I/O performed.
1352 Average data rate (bandwidth).
1358 Submission latency minimum, maximum, average and standard deviation. This is
1359 the time it took to submit the I/O.
1362 Completion latency minimum, maximum, average and standard deviation. This
1363 is the time between submission and completion.
1366 Bandwidth minimum, maximum, percentage of aggregate bandwidth received, average
1367 and standard deviation.
1370 CPU usage statistics. Includes user and system time, number of context switches
1371 this thread went through and number of major and minor page faults.
1374 Distribution of I/O depths. Each depth includes everything less than (or equal)
1375 to it, but greater than the previous depth.
1378 Number of read/write requests issued, and number of short read/write requests.
1381 Distribution of I/O completion latencies. The numbers follow the same pattern
1385 The group statistics show:
1390 Number of megabytes I/O performed.
1393 Aggregate bandwidth of threads in the group.
1396 Minimum average bandwidth a thread saw.
1399 Maximum average bandwidth a thread saw.
1402 Shortest runtime of threads in the group.
1405 Longest runtime of threads in the group.
1409 Finally, disk statistics are printed with reads first:
1414 Number of I/Os performed by all groups.
1417 Number of merges in the I/O scheduler.
1420 Number of ticks we kept the disk busy.
1423 Total time spent in the disk queue.
1430 It is also possible to get fio to dump the current output while it is
1431 running, without terminating the job. To do that, send fio the \fBUSR1\fR
1434 If the \fB\-\-minimal\fR option is given, the results will be printed in a
1435 semicolon-delimited format suitable for scripted use - a job description
1436 (if provided) follows on a new line. Note that the first
1437 number in the line is the version number. If the output has to be changed
1438 for some reason, this number will be incremented by 1 to signify that
1439 change. The fields are:
1442 .B terse version, fio version, jobname, groupid, error
1446 .B Total I/O \fR(KB)\fP, bandwidth \fR(KB/s)\fP, IOPS, runtime \fR(ms)\fP
1450 .B min, max, mean, standard deviation
1454 .B min, max, mean, standard deviation
1456 Completion latency percentiles (20 fields):
1458 .B Xth percentile=usec
1462 .B min, max, mean, standard deviation
1466 .B min, max, aggregate percentage of total, mean, standard deviation
1472 .B Total I/O \fR(KB)\fP, bandwidth \fR(KB/s)\fP, IOPS, runtime \fR(ms)\fP
1476 .B min, max, mean, standard deviation
1480 .B min, max, mean, standard deviation
1482 Completion latency percentiles (20 fields):
1484 .B Xth percentile=usec
1488 .B min, max, mean, standard deviation
1492 .B min, max, aggregate percentage of total, mean, standard deviation
1498 .B user, system, context switches, major page faults, minor page faults
1501 IO depth distribution:
1503 .B <=1, 2, 4, 8, 16, 32, >=64
1506 IO latency distribution:
1510 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
1514 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
1518 Disk utilization (1 for each disk used):
1520 .B name, read ios, write ios, read merges, write merges, read ticks, write ticks, read in-queue time, write in-queue time, disk utilization percentage
1523 Error Info (dependent on continue_on_error, default off):
1525 .B total # errors, first error code
1528 .B text description (if provided in config - appears on newline)
1531 Normally you would run fio as a stand-alone application on the machine
1532 where the IO workload should be generated. However, it is also possible to
1533 run the frontend and backend of fio separately. This makes it possible to
1534 have a fio server running on the machine(s) where the IO workload should
1535 be running, while controlling it from another machine.
1537 To start the server, you would do:
1539 \fBfio \-\-server=args\fR
1541 on that machine, where args defines what fio listens to. The arguments
1542 are of the form 'type:hostname or IP:port'. 'type' is either 'ip' (or ip4)
1543 for TCP/IP v4, 'ip6' for TCP/IP v6, or 'sock' for a local unix domain
1544 socket. 'hostname' is either a hostname or IP address, and 'port' is the port to
1545 listen to (only valid for TCP/IP, not a local socket). Some examples:
1549 Start a fio server, listening on all interfaces on the default port (8765).
1551 2) fio \-\-server=ip:hostname,4444
1553 Start a fio server, listening on IP belonging to hostname and on port 4444.
1555 3) fio \-\-server=ip6:::1,4444
1557 Start a fio server, listening on IPv6 localhost ::1 and on port 4444.
1559 4) fio \-\-server=,4444
1561 Start a fio server, listening on all interfaces on port 4444.
1563 5) fio \-\-server=1.2.3.4
1565 Start a fio server, listening on IP 1.2.3.4 on the default port.
1567 6) fio \-\-server=sock:/tmp/fio.sock
1569 Start a fio server, listening on the local socket /tmp/fio.sock.
1571 When a server is running, you can connect to it from a client. The client
1574 fio \-\-local-args \-\-client=server \-\-remote-args <job file(s)>
1576 where \-\-local-args are arguments that are local to the client where it is
1577 running, 'server' is the connect string, and \-\-remote-args and <job file(s)>
1578 are sent to the server. The 'server' string follows the same format as it
1579 does on the server side, to allow IP/hostname/socket and port strings.
1580 You can connect to multiple clients as well, to do that you could run:
1582 fio \-\-client=server2 \-\-client=server2 <job file(s)>
1586 was written by Jens Axboe <jens.axboe@oracle.com>,
1587 now Jens Axboe <jaxboe@fusionio.com>.
1589 This man page was written by Aaron Carroll <aaronc@cse.unsw.edu.au> based
1590 on documentation by Jens Axboe.
1591 .SH "REPORTING BUGS"
1592 Report bugs to the \fBfio\fR mailing list <fio@vger.kernel.org>.
1595 For further documentation see \fBHOWTO\fR and \fBREADME\fR.
1597 Sample jobfiles are available in the \fBexamples\fR directory.
projects / fio.git / blob