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'.
58 Turn on safety read-only checks, preventing any attempted write.
60 .BI \-\-section \fR=\fPsec
61 Only run section \fIsec\fR from job file. Multiple of these options can be given, adding more sections to run.
63 .BI \-\-alloc\-size \fR=\fPkb
64 Set the internal smalloc pool size to \fIkb\fP kilobytes.
66 .BI \-\-warnings\-fatal
67 All fio parser warnings are fatal, causing fio to exit with an error.
69 .BI \-\-max\-jobs \fR=\fPnr
70 Set the maximum allowed number of jobs (threads/processes) to support.
72 .BI \-\-server \fR=\fPargs
73 Start a backend server, with \fIargs\fP specifying what to listen to. See client/server section.
75 .BI \-\-daemonize \fR=\fPpidfile
76 Background a fio server, writing the pid to the given pid file.
78 .BI \-\-client \fR=\fPhost
79 Instead of running the jobs locally, send and run them on the given host.
81 .BI \-\-idle\-prof \fR=\fPoption
82 Report cpu idleness on a system or percpu basis (\fIoption\fP=system,percpu) or run unit work calibration only (\fIoption\fP=calibrate).
84 Job files are in `ini' format. They consist of one or more
85 job definitions, which begin with a job name in square brackets and
86 extend to the next job name. The job name can be any ASCII string
87 except `global', which has a special meaning. Following the job name is
88 a sequence of zero or more parameters, one per line, that define the
89 behavior of the job. Any line starting with a `;' or `#' character is
90 considered a comment and ignored.
92 If \fIjobfile\fR is specified as `-', the job file will be read from
95 The global section contains default parameters for jobs specified in the
96 job file. A job is only affected by global sections residing above it,
97 and there may be any number of global sections. Specific job definitions
98 may override any parameter set in global sections.
101 Some parameters may take arguments of a specific type. The types used are:
104 String: a sequence of alphanumeric characters.
107 SI integer: a whole number, possibly containing a suffix denoting the base unit
108 of the value. Accepted suffixes are `k', 'M', 'G', 'T', and 'P', denoting
109 kilo (1024), mega (1024^2), giga (1024^3), tera (1024^4), and peta (1024^5)
110 respectively. The suffix is not case sensitive. If prefixed with '0x', the
111 value is assumed to be base 16 (hexadecimal). A suffix may include a trailing 'b',
112 for instance 'kb' is identical to 'k'. You can specify a base 10 value
113 by using 'KiB', 'MiB', 'GiB', etc. This is useful for disk drives where
114 values are often given in base 10 values. Specifying '30GiB' will get you
118 Boolean: a true or false value. `0' denotes false, `1' denotes true.
121 Integer range: a range of integers specified in the format
122 \fIlower\fR:\fIupper\fR or \fIlower\fR\-\fIupper\fR. \fIlower\fR and
123 \fIupper\fR may contain a suffix as described above. If an option allows two
124 sets of ranges, they are separated with a `,' or `/' character. For example:
128 List of floating numbers: A list of floating numbers, separated by
133 May be used to override the job name. On the command line, this parameter
134 has the special purpose of signalling the start of a new job.
136 .BI description \fR=\fPstr
137 Human-readable description of the job. It is printed when the job is run, but
138 otherwise has no special purpose.
140 .BI directory \fR=\fPstr
141 Prefix filenames with this directory. Used to place files in a location other
144 .BI filename \fR=\fPstr
146 normally makes up a file name based on the job name, thread number, and file
147 number. If you want to share files between threads in a job or several jobs,
148 specify a \fIfilename\fR for each of them to override the default.
149 If the I/O engine is file-based, you can specify
150 a number of files by separating the names with a `:' character. `\-' is a
151 reserved name, meaning stdin or stdout, depending on the read/write direction
154 .BI filename_format \fR=\fPstr
155 If sharing multiple files between jobs, it is usually necessary to have
156 fio generate the exact names that you want. By default, fio will name a file
157 based on the default file format specification of
158 \fBjobname.jobnumber.filenumber\fP. With this option, that can be
159 customized. Fio will recognize and replace the following keywords in this
165 The name of the worker thread or process.
168 The incremental number of the worker thread or process.
171 The incremental number of the file for that worker thread or process.
174 To have dependent jobs share a set of files, this option can be set to
175 have fio generate filenames that are shared between the two. For instance,
176 if \fBtestfiles.$filenum\fR is specified, file number 4 for any job will
177 be named \fBtestfiles.4\fR. The default of \fB$jobname.$jobnum.$filenum\fR
178 will be used if no other format specifier is given.
182 .BI lockfile \fR=\fPstr
183 Fio defaults to not locking any files before it does IO to them. If a file or
184 file descriptor is shared, fio can serialize IO to that file to make the end
185 result consistent. This is usual for emulating real workloads that share files.
191 No locking. This is the default.
194 Only one thread or process may do IO at the time, excluding all others.
197 Read-write locking on the file. Many readers may access the file at the same
198 time, but writes get exclusive access.
202 .BI opendir \fR=\fPstr
203 Recursively open any files below directory \fIstr\fR.
205 .BI readwrite \fR=\fPstr "\fR,\fP rw" \fR=\fPstr
206 Type of I/O pattern. Accepted values are:
223 Mixed sequential reads and writes.
226 Mixed random reads and writes.
229 For mixed I/O, the default split is 50/50. For certain types of io the result
230 may still be skewed a bit, since the speed may be different. It is possible to
231 specify a number of IO's to do before getting a new offset, this is done by
232 appending a `:\fI<nr>\fR to the end of the string given. For a random read, it
233 would look like \fBrw=randread:8\fR for passing in an offset modifier with a
234 value of 8. If the postfix is used with a sequential IO pattern, then the value
235 specified will be added to the generated offset for each IO. For instance,
236 using \fBrw=write:4k\fR will skip 4k for every write. It turns sequential IO
237 into sequential IO with holes. See the \fBrw_sequencer\fR option.
240 .BI rw_sequencer \fR=\fPstr
241 If an offset modifier is given by appending a number to the \fBrw=<str>\fR line,
242 then this option controls how that number modifies the IO offset being
243 generated. Accepted values are:
248 Generate sequential offset
251 Generate the same offset
254 \fBsequential\fR is only useful for random IO, where fio would normally
255 generate a new random offset for every IO. If you append eg 8 to randread, you
256 would get a new random offset for every 8 IO's. The result would be a seek for
257 only every 8 IO's, instead of for every IO. Use \fBrw=randread:8\fR to specify
258 that. As sequential IO is already sequential, setting \fBsequential\fR for that
259 would not result in any differences. \fBidentical\fR behaves in a similar
260 fashion, except it sends the same offset 8 number of times before generating a
265 .BI kb_base \fR=\fPint
266 The base unit for a kilobyte. The defacto base is 2^10, 1024. Storage
267 manufacturers like to use 10^3 or 1000 as a base ten unit instead, for obvious
268 reasons. Allow values are 1024 or 1000, with 1024 being the default.
270 .BI unified_rw_reporting \fR=\fPbool
271 Fio normally reports statistics on a per data direction basis, meaning that
272 read, write, and trim are accounted and reported separately. If this option is
273 set, the fio will sum the results and report them as "mixed" instead.
275 .BI randrepeat \fR=\fPbool
276 Seed the random number generator in a predictable way so results are repeatable
277 across runs. Default: true.
279 .BI use_os_rand \fR=\fPbool
280 Fio can either use the random generator supplied by the OS to generator random
281 offsets, or it can use it's own internal generator (based on Tausworthe).
282 Default is to use the internal generator, which is often of better quality and
283 faster. Default: false.
285 .BI fallocate \fR=\fPstr
286 Whether pre-allocation is performed when laying down files. Accepted values
292 Do not pre-allocate space.
295 Pre-allocate via posix_fallocate().
298 Pre-allocate via fallocate() with FALLOC_FL_KEEP_SIZE set.
301 Backward-compatible alias for 'none'.
304 Backward-compatible alias for 'posix'.
307 May not be available on all supported platforms. 'keep' is only
308 available on Linux. If using ZFS on Solaris this must be set to 'none'
309 because ZFS doesn't support it. Default: 'posix'.
312 .BI fadvise_hint \fR=\fPbool
313 Use of \fIposix_fadvise\fR\|(2) to advise the kernel what I/O patterns
314 are likely to be issued. Default: true.
317 Total size of I/O for this job. \fBfio\fR will run until this many bytes have
318 been transfered, unless limited by other options (\fBruntime\fR, for instance).
319 Unless \fBnrfiles\fR and \fBfilesize\fR options are given, this amount will be
320 divided between the available files for the job. If not set, fio will use the
321 full size of the given files or devices. If the the files do not exist, size
322 must be given. It is also possible to give size as a percentage between 1 and
323 100. If size=20% is given, fio will use 20% of the full size of the given files
326 .BI fill_device \fR=\fPbool "\fR,\fB fill_fs" \fR=\fPbool
327 Sets size to something really large and waits for ENOSPC (no space left on
328 device) as the terminating condition. Only makes sense with sequential write.
329 For a read workload, the mount point will be filled first then IO started on
330 the result. This option doesn't make sense if operating on a raw device node,
331 since the size of that is already known by the file system. Additionally,
332 writing beyond end-of-device will not return ENOSPC there.
334 .BI filesize \fR=\fPirange
335 Individual file sizes. May be a range, in which case \fBfio\fR will select sizes
336 for files at random within the given range, limited to \fBsize\fR in total (if
337 that is given). If \fBfilesize\fR is not specified, each created file is the
340 .BI blocksize \fR=\fPint[,int] "\fR,\fB bs" \fR=\fPint[,int]
341 Block size for I/O units. Default: 4k. Values for reads and writes can be
342 specified separately in the format \fIread\fR,\fIwrite\fR, either of
343 which may be empty to leave that value at its default.
345 .BI blocksize_range \fR=\fPirange[,irange] "\fR,\fB bsrange" \fR=\fPirange[,irange]
346 Specify a range of I/O block sizes. The issued I/O unit will always be a
347 multiple of the minimum size, unless \fBblocksize_unaligned\fR is set. Applies
348 to both reads and writes if only one range is given, but can be specified
349 separately with a comma seperating the values. Example: bsrange=1k-4k,2k-8k.
350 Also (see \fBblocksize\fR).
352 .BI bssplit \fR=\fPstr
353 This option allows even finer grained control of the block sizes issued,
354 not just even splits between them. With this option, you can weight various
355 block sizes for exact control of the issued IO for a job that has mixed
356 block sizes. The format of the option is bssplit=blocksize/percentage,
357 optionally adding as many definitions as needed separated by a colon.
358 Example: bssplit=4k/10:64k/50:32k/40 would issue 50% 64k blocks, 10% 4k
359 blocks and 40% 32k blocks. \fBbssplit\fR also supports giving separate
360 splits to reads and writes. The format is identical to what the
361 \fBbs\fR option accepts, the read and write parts are separated with a
364 .B blocksize_unaligned\fR,\fP bs_unaligned
365 If set, any size in \fBblocksize_range\fR may be used. This typically won't
366 work with direct I/O, as that normally requires sector alignment.
368 .BI blockalign \fR=\fPint[,int] "\fR,\fB ba" \fR=\fPint[,int]
369 At what boundary to align random IO offsets. Defaults to the same as 'blocksize'
370 the minimum blocksize given. Minimum alignment is typically 512b
371 for using direct IO, though it usually depends on the hardware block size.
372 This option is mutually exclusive with using a random map for files, so it
373 will turn off that option.
376 Initialise buffers with all zeros. Default: fill buffers with random data.
379 If this option is given, fio will refill the IO buffers on every submit. The
380 default is to only fill it at init time and reuse that data. Only makes sense
381 if zero_buffers isn't specified, naturally. If data verification is enabled,
382 refill_buffers is also automatically enabled.
384 .BI scramble_buffers \fR=\fPbool
385 If \fBrefill_buffers\fR is too costly and the target is using data
386 deduplication, then setting this option will slightly modify the IO buffer
387 contents to defeat normal de-dupe attempts. This is not enough to defeat
388 more clever block compression attempts, but it will stop naive dedupe
389 of blocks. Default: true.
391 .BI buffer_compress_percentage \fR=\fPint
392 If this is set, then fio will attempt to provide IO buffer content (on WRITEs)
393 that compress to the specified level. Fio does this by providing a mix of
394 random data and zeroes. Note that this is per block size unit, for file/disk
395 wide compression level that matches this setting, you'll also want to set
396 \fBrefill_buffers\fR.
398 .BI buffer_compress_chunk \fR=\fPint
399 See \fBbuffer_compress_percentage\fR. This setting allows fio to manage how
400 big the ranges of random data and zeroed data is. Without this set, fio will
401 provide \fBbuffer_compress_percentage\fR of blocksize random data, followed by
402 the remaining zeroed. With this set to some chunk size smaller than the block
403 size, fio can alternate random and zeroed data throughout the IO buffer.
405 .BI nrfiles \fR=\fPint
406 Number of files to use for this job. Default: 1.
408 .BI openfiles \fR=\fPint
409 Number of files to keep open at the same time. Default: \fBnrfiles\fR.
411 .BI file_service_type \fR=\fPstr
412 Defines how files to service are selected. The following types are defined:
417 Choose a file at random
420 Round robin over open files (default).
422 Do each file in the set sequentially.
425 The number of I/Os to issue before switching a new file can be specified by
426 appending `:\fIint\fR' to the service type.
429 .BI ioengine \fR=\fPstr
430 Defines how the job issues I/O. The following types are defined:
435 Basic \fIread\fR\|(2) or \fIwrite\fR\|(2) I/O. \fIfseek\fR\|(2) is used to
436 position the I/O location.
439 Basic \fIpread\fR\|(2) or \fIpwrite\fR\|(2) I/O.
442 Basic \fIreadv\fR\|(2) or \fIwritev\fR\|(2) I/O. Will emulate queuing by
443 coalescing adjacents IOs into a single submission.
446 Linux native asynchronous I/O. This ioengine defines engine specific options.
449 POSIX asynchronous I/O using \fIaio_read\fR\|(3) and \fIaio_write\fR\|(3).
452 Solaris native asynchronous I/O.
455 Windows native asynchronous I/O.
458 File is memory mapped with \fImmap\fR\|(2) and data copied using
462 \fIsplice\fR\|(2) is used to transfer the data and \fIvmsplice\fR\|(2) to
463 transfer data from user-space to the kernel.
466 Use the syslet system calls to make regular read/write asynchronous.
469 SCSI generic sg v3 I/O. May be either synchronous using the SG_IO ioctl, or if
470 the target is an sg character device, we use \fIread\fR\|(2) and
471 \fIwrite\fR\|(2) for asynchronous I/O.
474 Doesn't transfer any data, just pretends to. Mainly used to exercise \fBfio\fR
475 itself and for debugging and testing purposes.
478 Transfer over the network. The protocol to be used can be defined with the
479 \fBprotocol\fR parameter. Depending on the protocol, \fBfilename\fR,
480 \fBhostname\fR, \fBport\fR, or \fBlisten\fR must be specified.
481 This ioengine defines engine specific options.
484 Like \fBnet\fR, but uses \fIsplice\fR\|(2) and \fIvmsplice\fR\|(2) to map data
485 and send/receive. This ioengine defines engine specific options.
488 Doesn't transfer any data, but burns CPU cycles according to \fBcpuload\fR and
489 \fBcpucycles\fR parameters.
492 The GUASI I/O engine is the Generic Userspace Asynchronous Syscall Interface
493 approach to asycnronous I/O.
495 See <http://www.xmailserver.org/guasi\-lib.html>.
498 The RDMA I/O engine supports both RDMA memory semantics (RDMA_WRITE/RDMA_READ)
499 and channel semantics (Send/Recv) for the InfiniBand, RoCE and iWARP protocols.
502 Loads an external I/O engine object file. Append the engine filename as
506 IO engine that does regular linux native fallocate callt to simulate data
507 transfer as fio ioengine
509 DDIR_READ does fallocate(,mode = FALLOC_FL_KEEP_SIZE,)
511 DIR_WRITE does fallocate(,mode = 0)
513 DDIR_TRIM does fallocate(,mode = FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE)
516 IO engine that does regular EXT4_IOC_MOVE_EXT ioctls to simulate defragment activity
517 request to DDIR_WRITE event
522 .BI iodepth \fR=\fPint
523 Number of I/O units to keep in flight against the file. Note that increasing
524 iodepth beyond 1 will not affect synchronous ioengines (except for small
525 degress when verify_async is in use). Even async engines my impose OS
526 restrictions causing the desired depth not to be achieved. This may happen on
527 Linux when using libaio and not setting \fBdirect\fR=1, since buffered IO is
528 not async on that OS. Keep an eye on the IO depth distribution in the
529 fio output to verify that the achieved depth is as expected. Default: 1.
531 .BI iodepth_batch \fR=\fPint
532 Number of I/Os to submit at once. Default: \fBiodepth\fR.
534 .BI iodepth_batch_complete \fR=\fPint
535 This defines how many pieces of IO to retrieve at once. It defaults to 1 which
536 means that we'll ask for a minimum of 1 IO in the retrieval process from the
537 kernel. The IO retrieval will go on until we hit the limit set by
538 \fBiodepth_low\fR. If this variable is set to 0, then fio will always check for
539 completed events before queuing more IO. This helps reduce IO latency, at the
540 cost of more retrieval system calls.
542 .BI iodepth_low \fR=\fPint
543 Low watermark indicating when to start filling the queue again. Default:
546 .BI direct \fR=\fPbool
547 If true, use non-buffered I/O (usually O_DIRECT). Default: false.
549 .BI buffered \fR=\fPbool
550 If true, use buffered I/O. This is the opposite of the \fBdirect\fR parameter.
553 .BI offset \fR=\fPint
554 Offset in the file to start I/O. Data before the offset will not be touched.
556 .BI offset_increment \fR=\fPint
557 If this is provided, then the real offset becomes the
558 offset + offset_increment * thread_number, where the thread number is a counter
559 that starts at 0 and is incremented for each job. This option is useful if
560 there are several jobs which are intended to operate on a file in parallel in
561 disjoint segments, with even spacing between the starting points.
564 How many I/Os to perform before issuing an \fBfsync\fR\|(2) of dirty data. If
565 0, don't sync. Default: 0.
567 .BI fdatasync \fR=\fPint
568 Like \fBfsync\fR, but uses \fBfdatasync\fR\|(2) instead to only sync the
569 data parts of the file. Default: 0.
571 .BI sync_file_range \fR=\fPstr:int
572 Use sync_file_range() for every \fRval\fP number of write operations. Fio will
573 track range of writes that have happened since the last sync_file_range() call.
574 \fRstr\fP can currently be one or more of:
578 SYNC_FILE_RANGE_WAIT_BEFORE
581 SYNC_FILE_RANGE_WRITE
584 SYNC_FILE_RANGE_WRITE
588 So if you do sync_file_range=wait_before,write:8, fio would use
589 \fBSYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE\fP for every 8 writes.
590 Also see the sync_file_range(2) man page. This option is Linux specific.
592 .BI overwrite \fR=\fPbool
593 If writing, setup the file first and do overwrites. Default: false.
595 .BI end_fsync \fR=\fPbool
596 Sync file contents when a write stage has completed. Default: false.
598 .BI fsync_on_close \fR=\fPbool
599 If true, sync file contents on close. This differs from \fBend_fsync\fR in that
600 it will happen on every close, not just at the end of the job. Default: false.
602 .BI rwmixread \fR=\fPint
603 Percentage of a mixed workload that should be reads. Default: 50.
605 .BI rwmixwrite \fR=\fPint
606 Percentage of a mixed workload that should be writes. If \fBrwmixread\fR and
607 \fBrwmixwrite\fR are given and do not sum to 100%, the latter of the two
608 overrides the first. This may interfere with a given rate setting, if fio is
609 asked to limit reads or writes to a certain rate. If that is the case, then
610 the distribution may be skewed. Default: 50.
612 .BI random_distribution \fR=\fPstr:float
613 By default, fio will use a completely uniform random distribution when asked
614 to perform random IO. Sometimes it is useful to skew the distribution in
615 specific ways, ensuring that some parts of the data is more hot than others.
616 Fio includes the following distribution models:
620 Uniform random distribution
630 When using a zipf or pareto distribution, an input value is also needed to
631 define the access pattern. For zipf, this is the zipf theta. For pareto,
632 it's the pareto power. Fio includes a test program, genzipf, that can be
633 used visualize what the given input values will yield in terms of hit rates.
634 If you wanted to use zipf with a theta of 1.2, you would use
635 random_distribution=zipf:1.2 as the option. If a non-uniform model is used,
636 fio will disable use of the random map.
639 Normally \fBfio\fR will cover every block of the file when doing random I/O. If
640 this parameter is given, a new offset will be chosen without looking at past
641 I/O history. This parameter is mutually exclusive with \fBverify\fR.
643 .BI softrandommap \fR=\fPbool
644 See \fBnorandommap\fR. If fio runs with the random block map enabled and it
645 fails to allocate the map, if this option is set it will continue without a
646 random block map. As coverage will not be as complete as with random maps, this
647 option is disabled by default.
649 .BI random_generator \fR=\fPstr
650 Fio supports the following engines for generating IO offsets for random IO:
654 Strong 2^88 cycle random number generator
657 Linear feedback shift register generator
661 Tausworthe is a strong random number generator, but it requires tracking on the
662 side if we want to ensure that blocks are only read or written once. LFSR
663 guarantees that we never generate the same offset twice, and it's also less
664 computationally expensive. It's not a true random generator, however, though
665 for IO purposes it's typically good enough. LFSR only works with single block
666 sizes, not with workloads that use multiple block sizes. If used with such a
667 workload, fio may read or write some blocks multiple times.
670 Run job with given nice value. See \fInice\fR\|(2).
673 Set I/O priority value of this job between 0 (highest) and 7 (lowest). See
676 .BI prioclass \fR=\fPint
677 Set I/O priority class. See \fIionice\fR\|(1).
679 .BI thinktime \fR=\fPint
680 Stall job for given number of microseconds between issuing I/Os.
682 .BI thinktime_spin \fR=\fPint
683 Pretend to spend CPU time for given number of microseconds, sleeping the rest
684 of the time specified by \fBthinktime\fR. Only valid if \fBthinktime\fR is set.
686 .BI thinktime_blocks \fR=\fPint
687 Number of blocks to issue before waiting \fBthinktime\fR microseconds.
691 Cap bandwidth used by this job. The number is in bytes/sec, the normal postfix
692 rules apply. You can use \fBrate\fR=500k to limit reads and writes to 500k each,
693 or you can specify read and writes separately. Using \fBrate\fR=1m,500k would
694 limit reads to 1MB/sec and writes to 500KB/sec. Capping only reads or writes
695 can be done with \fBrate\fR=,500k or \fBrate\fR=500k,. The former will only
696 limit writes (to 500KB/sec), the latter will only limit reads.
698 .BI ratemin \fR=\fPint
699 Tell \fBfio\fR to do whatever it can to maintain at least the given bandwidth.
700 Failing to meet this requirement will cause the job to exit. The same format
701 as \fBrate\fR is used for read vs write separation.
703 .BI rate_iops \fR=\fPint
704 Cap the bandwidth to this number of IOPS. Basically the same as rate, just
705 specified independently of bandwidth. The same format as \fBrate\fR is used for
706 read vs write seperation. If \fBblocksize\fR is a range, the smallest block
707 size is used as the metric.
709 .BI rate_iops_min \fR=\fPint
710 If this rate of I/O is not met, the job will exit. The same format as \fBrate\fR
711 is used for read vs write seperation.
713 .BI ratecycle \fR=\fPint
714 Average bandwidth for \fBrate\fR and \fBratemin\fR over this number of
715 milliseconds. Default: 1000ms.
717 .BI max_latency \fR=\fPint
718 If set, fio will exit the job if it exceeds this maximum latency. It will exit
721 .BI cpumask \fR=\fPint
722 Set CPU affinity for this job. \fIint\fR is a bitmask of allowed CPUs the job
723 may run on. See \fBsched_setaffinity\fR\|(2).
725 .BI cpus_allowed \fR=\fPstr
726 Same as \fBcpumask\fR, but allows a comma-delimited list of CPU numbers.
728 .BI numa_cpu_nodes \fR=\fPstr
729 Set this job running on spcified NUMA nodes' CPUs. The arguments allow
730 comma delimited list of cpu numbers, A-B ranges, or 'all'.
732 .BI numa_mem_policy \fR=\fPstr
733 Set this job's memory policy and corresponding NUMA nodes. Format of
737 .B <mode>[:<nodelist>]
740 is one of the following memory policy:
742 .B default, prefer, bind, interleave, local
745 For \fBdefault\fR and \fBlocal\fR memory policy, no \fBnodelist\fR is
746 needed to be specified. For \fBprefer\fR, only one node is
747 allowed. For \fBbind\fR and \fBinterleave\fR, \fBnodelist\fR allows
748 comma delimited list of numbers, A-B ranges, or 'all'.
750 .BI startdelay \fR=\fPint
751 Delay start of job for the specified number of seconds.
753 .BI runtime \fR=\fPint
754 Terminate processing after the specified number of seconds.
757 If given, run for the specified \fBruntime\fR duration even if the files are
758 completely read or written. The same workload will be repeated as many times
759 as \fBruntime\fR allows.
761 .BI ramp_time \fR=\fPint
762 If set, fio will run the specified workload for this amount of time before
763 logging any performance numbers. Useful for letting performance settle before
764 logging results, thus minimizing the runtime required for stable results. Note
765 that the \fBramp_time\fR is considered lead in time for a job, thus it will
766 increase the total runtime if a special timeout or runtime is specified.
768 .BI invalidate \fR=\fPbool
769 Invalidate buffer-cache for the file prior to starting I/O. Default: true.
772 Use synchronous I/O for buffered writes. For the majority of I/O engines,
773 this means using O_SYNC. Default: false.
775 .BI iomem \fR=\fPstr "\fR,\fP mem" \fR=\fPstr
776 Allocation method for I/O unit buffer. Allowed values are:
781 Allocate memory with \fImalloc\fR\|(3).
784 Use shared memory buffers allocated through \fIshmget\fR\|(2).
787 Same as \fBshm\fR, but use huge pages as backing.
790 Use \fImmap\fR\|(2) for allocation. Uses anonymous memory unless a filename
791 is given after the option in the format `:\fIfile\fR'.
794 Same as \fBmmap\fR, but use huge files as backing.
797 The amount of memory allocated is the maximum allowed \fBblocksize\fR for the
798 job multiplied by \fBiodepth\fR. For \fBshmhuge\fR or \fBmmaphuge\fR to work,
799 the system must have free huge pages allocated. \fBmmaphuge\fR also needs to
800 have hugetlbfs mounted, and \fIfile\fR must point there. At least on Linux,
801 huge pages must be manually allocated. See \fB/proc/sys/vm/nr_hugehages\fR
802 and the documentation for that. Normally you just need to echo an appropriate
803 number, eg echoing 8 will ensure that the OS has 8 huge pages ready for
807 .BI iomem_align \fR=\fPint "\fR,\fP mem_align" \fR=\fPint
808 This indiciates the memory alignment of the IO memory buffers. Note that the
809 given alignment is applied to the first IO unit buffer, if using \fBiodepth\fR
810 the alignment of the following buffers are given by the \fBbs\fR used. In
811 other words, if using a \fBbs\fR that is a multiple of the page sized in the
812 system, all buffers will be aligned to this value. If using a \fBbs\fR that
813 is not page aligned, the alignment of subsequent IO memory buffers is the
814 sum of the \fBiomem_align\fR and \fBbs\fR used.
816 .BI hugepage\-size \fR=\fPint
817 Defines the size of a huge page. Must be at least equal to the system setting.
818 Should be a multiple of 1MB. Default: 4MB.
821 Terminate all jobs when one finishes. Default: wait for each job to finish.
823 .BI bwavgtime \fR=\fPint
824 Average bandwidth calculations over the given time in milliseconds. Default:
827 .BI iopsavgtime \fR=\fPint
828 Average IOPS calculations over the given time in milliseconds. Default:
831 .BI create_serialize \fR=\fPbool
832 If true, serialize file creation for the jobs. Default: true.
834 .BI create_fsync \fR=\fPbool
835 \fIfsync\fR\|(2) data file after creation. Default: true.
837 .BI create_on_open \fR=\fPbool
838 If true, the files are not created until they are opened for IO by the job.
840 .BI create_only \fR=\fPbool
841 If true, fio will only run the setup phase of the job. If files need to be
842 laid out or updated on disk, only that will be done. The actual job contents
845 .BI pre_read \fR=\fPbool
846 If this is given, files will be pre-read into memory before starting the given
847 IO operation. This will also clear the \fR \fBinvalidate\fR flag, since it is
848 pointless to pre-read and then drop the cache. This will only work for IO
849 engines that are seekable, since they allow you to read the same data
850 multiple times. Thus it will not work on eg network or splice IO.
852 .BI unlink \fR=\fPbool
853 Unlink job files when done. Default: false.
856 Specifies the number of iterations (runs of the same workload) of this job.
859 .BI do_verify \fR=\fPbool
860 Run the verify phase after a write phase. Only valid if \fBverify\fR is set.
863 .BI verify \fR=\fPstr
864 Method of verifying file contents after each iteration of the job. Allowed
869 .B md5 crc16 crc32 crc32c crc32c-intel crc64 crc7 sha256 sha512 sha1
870 Store appropriate checksum in the header of each block. crc32c-intel is
871 hardware accelerated SSE4.2 driven, falls back to regular crc32c if
872 not supported by the system.
875 Write extra information about each I/O (timestamp, block number, etc.). The
876 block number is verified. See \fBverify_pattern\fR as well.
879 Pretend to verify. Used for testing internals.
882 This option can be used for repeated burn-in tests of a system to make sure
883 that the written data is also correctly read back. If the data direction given
884 is a read or random read, fio will assume that it should verify a previously
885 written file. If the data direction includes any form of write, the verify will
886 be of the newly written data.
889 .BI verify_sort \fR=\fPbool
890 If true, written verify blocks are sorted if \fBfio\fR deems it to be faster to
891 read them back in a sorted manner. Default: true.
893 .BI verify_offset \fR=\fPint
894 Swap the verification header with data somewhere else in the block before
895 writing. It is swapped back before verifying.
897 .BI verify_interval \fR=\fPint
898 Write the verification header for this number of bytes, which should divide
899 \fBblocksize\fR. Default: \fBblocksize\fR.
901 .BI verify_pattern \fR=\fPstr
902 If set, fio will fill the io buffers with this pattern. Fio defaults to filling
903 with totally random bytes, but sometimes it's interesting to fill with a known
904 pattern for io verification purposes. Depending on the width of the pattern,
905 fio will fill 1/2/3/4 bytes of the buffer at the time(it can be either a
906 decimal or a hex number). The verify_pattern if larger than a 32-bit quantity
907 has to be a hex number that starts with either "0x" or "0X". Use with
910 .BI verify_fatal \fR=\fPbool
911 If true, exit the job on the first observed verification failure. Default:
914 .BI verify_dump \fR=\fPbool
915 If set, dump the contents of both the original data block and the data block we
916 read off disk to files. This allows later analysis to inspect just what kind of
917 data corruption occurred. Off by default.
919 .BI verify_async \fR=\fPint
920 Fio will normally verify IO inline from the submitting thread. This option
921 takes an integer describing how many async offload threads to create for IO
922 verification instead, causing fio to offload the duty of verifying IO contents
923 to one or more separate threads. If using this offload option, even sync IO
924 engines can benefit from using an \fBiodepth\fR setting higher than 1, as it
925 allows them to have IO in flight while verifies are running.
927 .BI verify_async_cpus \fR=\fPstr
928 Tell fio to set the given CPU affinity on the async IO verification threads.
929 See \fBcpus_allowed\fP for the format used.
931 .BI verify_backlog \fR=\fPint
932 Fio will normally verify the written contents of a job that utilizes verify
933 once that job has completed. In other words, everything is written then
934 everything is read back and verified. You may want to verify continually
935 instead for a variety of reasons. Fio stores the meta data associated with an
936 IO block in memory, so for large verify workloads, quite a bit of memory would
937 be used up holding this meta data. If this option is enabled, fio will write
938 only N blocks before verifying these blocks.
940 .BI verify_backlog_batch \fR=\fPint
941 Control how many blocks fio will verify if verify_backlog is set. If not set,
942 will default to the value of \fBverify_backlog\fR (meaning the entire queue is
943 read back and verified). If \fBverify_backlog_batch\fR is less than
944 \fBverify_backlog\fR then not all blocks will be verified, if
945 \fBverify_backlog_batch\fR is larger than \fBverify_backlog\fR, some blocks
946 will be verified more than once.
948 .B stonewall "\fR,\fP wait_for_previous"
949 Wait for preceding jobs in the job file to exit before starting this one.
950 \fBstonewall\fR implies \fBnew_group\fR.
953 Start a new reporting group. If not given, all jobs in a file will be part
954 of the same reporting group, unless separated by a stonewall.
956 .BI numjobs \fR=\fPint
957 Number of clones (processes/threads performing the same workload) of this job.
961 If set, display per-group reports instead of per-job when \fBnumjobs\fR is
965 Use threads created with \fBpthread_create\fR\|(3) instead of processes created
966 with \fBfork\fR\|(2).
968 .BI zonesize \fR=\fPint
969 Divide file into zones of the specified size in bytes. See \fBzoneskip\fR.
971 .BI zoneskip \fR=\fPint
972 Skip the specified number of bytes when \fBzonesize\fR bytes of data have been
975 .BI write_iolog \fR=\fPstr
976 Write the issued I/O patterns to the specified file. Specify a separate file
977 for each job, otherwise the iologs will be interspersed and the file may be
980 .BI read_iolog \fR=\fPstr
981 Replay the I/O patterns contained in the specified file generated by
982 \fBwrite_iolog\fR, or may be a \fBblktrace\fR binary file.
984 .BI replay_no_stall \fR=\fPint
985 While replaying I/O patterns using \fBread_iolog\fR the default behavior
986 attempts to respect timing information between I/Os. Enabling
987 \fBreplay_no_stall\fR causes I/Os to be replayed as fast as possible while
988 still respecting ordering.
990 .BI replay_redirect \fR=\fPstr
991 While replaying I/O patterns using \fBread_iolog\fR the default behavior
992 is to replay the IOPS onto the major/minor device that each IOP was recorded
993 from. Setting \fBreplay_redirect\fR causes all IOPS to be replayed onto the
994 single specified device regardless of the device it was recorded from.
996 .BI write_bw_log \fR=\fPstr
997 If given, write a bandwidth log of the jobs in this job file. Can be used to
998 store data of the bandwidth of the jobs in their lifetime. The included
999 fio_generate_plots script uses gnuplot to turn these text files into nice
1000 graphs. See \fBwrite_log_log\fR for behaviour of given filename. For this
1001 option, the postfix is _bw.log.
1003 .BI write_lat_log \fR=\fPstr
1004 Same as \fBwrite_bw_log\fR, but writes I/O completion latencies. If no
1005 filename is given with this option, the default filename of "jobname_type.log"
1006 is used. Even if the filename is given, fio will still append the type of log.
1008 .BI write_iops_log \fR=\fPstr
1009 Same as \fBwrite_bw_log\fR, but writes IOPS. If no filename is given with this
1010 option, the default filename of "jobname_type.log" is used. Even if the
1011 filename is given, fio will still append the type of log.
1013 .BI log_avg_msec \fR=\fPint
1014 By default, fio will log an entry in the iops, latency, or bw log for every
1015 IO that completes. When writing to the disk log, that can quickly grow to a
1016 very large size. Setting this option makes fio average the each log entry
1017 over the specified period of time, reducing the resolution of the log.
1020 .BI disable_lat \fR=\fPbool
1021 Disable measurements of total latency numbers. Useful only for cutting
1022 back the number of calls to gettimeofday, as that does impact performance at
1023 really high IOPS rates. Note that to really get rid of a large amount of these
1024 calls, this option must be used with disable_slat and disable_bw as well.
1026 .BI disable_clat \fR=\fPbool
1027 Disable measurements of completion latency numbers. See \fBdisable_lat\fR.
1029 .BI disable_slat \fR=\fPbool
1030 Disable measurements of submission latency numbers. See \fBdisable_lat\fR.
1032 .BI disable_bw_measurement \fR=\fPbool
1033 Disable measurements of throughput/bandwidth numbers. See \fBdisable_lat\fR.
1035 .BI lockmem \fR=\fPint
1036 Pin the specified amount of memory with \fBmlock\fR\|(2). Can be used to
1037 simulate a smaller amount of memory.
1039 .BI exec_prerun \fR=\fPstr
1040 Before running the job, execute the specified command with \fBsystem\fR\|(3).
1042 .BI exec_postrun \fR=\fPstr
1043 Same as \fBexec_prerun\fR, but the command is executed after the job completes.
1045 .BI ioscheduler \fR=\fPstr
1046 Attempt to switch the device hosting the file to the specified I/O scheduler.
1048 .BI cpuload \fR=\fPint
1049 If the job is a CPU cycle-eater, attempt to use the specified percentage of
1052 .BI cpuchunks \fR=\fPint
1053 If the job is a CPU cycle-eater, split the load into cycles of the
1054 given time in milliseconds.
1056 .BI disk_util \fR=\fPbool
1057 Generate disk utilization statistics if the platform supports it. Default: true.
1059 .BI clocksource \fR=\fPstr
1060 Use the given clocksource as the base of timing. The supported options are:
1070 Internal CPU clock source
1074 \fBcpu\fR is the preferred clocksource if it is reliable, as it is very fast
1075 (and fio is heavy on time calls). Fio will automatically use this clocksource
1076 if it's supported and considered reliable on the system it is running on,
1077 unless another clocksource is specifically set. For x86/x86-64 CPUs, this
1078 means supporting TSC Invariant.
1080 .BI gtod_reduce \fR=\fPbool
1081 Enable all of the gettimeofday() reducing options (disable_clat, disable_slat,
1082 disable_bw) plus reduce precision of the timeout somewhat to really shrink the
1083 gettimeofday() call count. With this option enabled, we only do about 0.4% of
1084 the gtod() calls we would have done if all time keeping was enabled.
1086 .BI gtod_cpu \fR=\fPint
1087 Sometimes it's cheaper to dedicate a single thread of execution to just getting
1088 the current time. Fio (and databases, for instance) are very intensive on
1089 gettimeofday() calls. With this option, you can set one CPU aside for doing
1090 nothing but logging current time to a shared memory location. Then the other
1091 threads/processes that run IO workloads need only copy that segment, instead of
1092 entering the kernel with a gettimeofday() call. The CPU set aside for doing
1093 these time calls will be excluded from other uses. Fio will manually clear it
1094 from the CPU mask of other jobs.
1096 .BI ignore_error \fR=\fPstr
1097 Sometimes you want to ignore some errors during test in that case you can specify
1098 error list for each error type.
1100 ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST
1102 errors for given error type is separated with ':'.
1103 Error may be symbol ('ENOSPC', 'ENOMEM') or an integer.
1105 Example: ignore_error=EAGAIN,ENOSPC:122 .
1107 This option will ignore EAGAIN from READ, and ENOSPC and 122(EDQUOT) from WRITE.
1109 .BI error_dump \fR=\fPbool
1110 If set dump every error even if it is non fatal, true by default. If disabled
1111 only fatal error will be dumped
1113 .BI cgroup \fR=\fPstr
1114 Add job to this control group. If it doesn't exist, it will be created.
1115 The system must have a mounted cgroup blkio mount point for this to work. If
1116 your system doesn't have it mounted, you can do so with:
1118 # mount \-t cgroup \-o blkio none /cgroup
1120 .BI cgroup_weight \fR=\fPint
1121 Set the weight of the cgroup to this value. See the documentation that comes
1122 with the kernel, allowed values are in the range of 100..1000.
1124 .BI cgroup_nodelete \fR=\fPbool
1125 Normally fio will delete the cgroups it has created after the job completion.
1126 To override this behavior and to leave cgroups around after the job completion,
1127 set cgroup_nodelete=1. This can be useful if one wants to inspect various
1128 cgroup files after job completion. Default: false
1131 Instead of running as the invoking user, set the user ID to this value before
1132 the thread/process does any work.
1135 Set group ID, see \fBuid\fR.
1137 .BI flow_id \fR=\fPint
1138 The ID of the flow. If not specified, it defaults to being a global flow. See
1142 Weight in token-based flow control. If this value is used, then there is a
1143 \fBflow counter\fR which is used to regulate the proportion of activity between
1144 two or more jobs. fio attempts to keep this flow counter near zero. The
1145 \fBflow\fR parameter stands for how much should be added or subtracted to the
1146 flow counter on each iteration of the main I/O loop. That is, if one job has
1147 \fBflow=8\fR and another job has \fBflow=-1\fR, then there will be a roughly
1148 1:8 ratio in how much one runs vs the other.
1150 .BI flow_watermark \fR=\fPint
1151 The maximum value that the absolute value of the flow counter is allowed to
1152 reach before the job must wait for a lower value of the counter.
1154 .BI flow_sleep \fR=\fPint
1155 The period of time, in microseconds, to wait after the flow watermark has been
1156 exceeded before retrying operations
1158 .BI clat_percentiles \fR=\fPbool
1159 Enable the reporting of percentiles of completion latencies.
1161 .BI percentile_list \fR=\fPfloat_list
1162 Overwrite the default list of percentiles for completion
1163 latencies. Each number is a floating number in the range (0,100], and
1164 the maximum length of the list is 20. Use ':' to separate the
1165 numbers. For example, \-\-percentile_list=99.5:99.9 will cause fio to
1166 report the values of completion latency below which 99.5% and 99.9% of
1167 the observed latencies fell, respectively.
1168 .SS "Ioengine Parameters List"
1169 Some parameters are only valid when a specific ioengine is in use. These are
1170 used identically to normal parameters, with the caveat that when used on the
1171 command line, the must come after the ioengine that defines them is selected.
1173 .BI (cpu)cpuload \fR=\fPint
1174 Attempt to use the specified percentage of CPU cycles.
1176 .BI (cpu)cpuchunks \fR=\fPint
1177 Split the load into cycles of the given time. In microseconds.
1179 .BI (libaio)userspace_reap
1180 Normally, with the libaio engine in use, fio will use
1181 the io_getevents system call to reap newly returned events.
1182 With this flag turned on, the AIO ring will be read directly
1183 from user-space to reap events. The reaping mode is only
1184 enabled when polling for a minimum of 0 events (eg when
1185 iodepth_batch_complete=0).
1187 .BI (net,netsplice)hostname \fR=\fPstr
1188 The host name or IP address to use for TCP or UDP based IO.
1189 If the job is a TCP listener or UDP reader, the hostname is not
1190 used and must be omitted.
1192 .BI (net,netsplice)port \fR=\fPint
1193 The TCP or UDP port to bind to or connect to.
1195 .BI (net,netsplice)nodelay \fR=\fPbool
1196 Set TCP_NODELAY on TCP connections.
1198 .BI (net,netsplice)protocol \fR=\fPstr "\fR,\fP proto" \fR=\fPstr
1199 The network protocol to use. Accepted values are:
1204 Transmission control protocol
1207 User datagram protocol
1213 When the protocol is TCP or UDP, the port must also be given,
1214 as well as the hostname if the job is a TCP listener or UDP
1215 reader. For unix sockets, the normal filename option should be
1216 used and the port is invalid.
1219 .BI (net,netsplice)listen
1220 For TCP network connections, tell fio to listen for incoming
1221 connections rather than initiating an outgoing connection. The
1222 hostname must be omitted if this option is used.
1224 .BI (net, pingpong) \fR=\fPbool
1225 Normal a network writer will just continue writing data, and a network reader
1226 will just consume packages. If pingpong=1 is set, a writer will send its normal
1227 payload to the reader, then wait for the reader to send the same payload back.
1228 This allows fio to measure network latencies. The submission and completion
1229 latencies then measure local time spent sending or receiving, and the
1230 completion latency measures how long it took for the other end to receive and
1233 .BI (e4defrag,donorname) \fR=\fPstr
1234 File will be used as a block donor (swap extents between files)
1236 .BI (e4defrag,inplace) \fR=\fPint
1237 Configure donor file block allocation strategy
1240 Preallocate donor's file on init
1243 allocate space immidietly inside defragment event, and free right after event
1247 While running, \fBfio\fR will display the status of the created jobs. For
1251 Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
1254 The characters in the first set of brackets denote the current status of each
1255 threads. The possible values are:
1261 Setup but not started.
1267 Initialized, waiting.
1270 Running, doing sequential reads.
1273 Running, doing random reads.
1276 Running, doing sequential writes.
1279 Running, doing random writes.
1282 Running, doing mixed sequential reads/writes.
1285 Running, doing mixed random reads/writes.
1288 Running, currently waiting for \fBfsync\fR\|(2).
1291 Running, verifying written data.
1294 Exited, not reaped by main thread.
1297 Exited, thread reaped.
1301 The second set of brackets shows the estimated completion percentage of
1302 the current group. The third set shows the read and write I/O rate,
1303 respectively. Finally, the estimated run time of the job is displayed.
1305 When \fBfio\fR completes (or is interrupted by Ctrl-C), it will show data
1306 for each thread, each group of threads, and each disk, in that order.
1308 Per-thread statistics first show the threads client number, group-id, and
1309 error code. The remaining figures are as follows:
1313 Number of megabytes of I/O performed.
1316 Average data rate (bandwidth).
1322 Submission latency minimum, maximum, average and standard deviation. This is
1323 the time it took to submit the I/O.
1326 Completion latency minimum, maximum, average and standard deviation. This
1327 is the time between submission and completion.
1330 Bandwidth minimum, maximum, percentage of aggregate bandwidth received, average
1331 and standard deviation.
1334 CPU usage statistics. Includes user and system time, number of context switches
1335 this thread went through and number of major and minor page faults.
1338 Distribution of I/O depths. Each depth includes everything less than (or equal)
1339 to it, but greater than the previous depth.
1342 Number of read/write requests issued, and number of short read/write requests.
1345 Distribution of I/O completion latencies. The numbers follow the same pattern
1349 The group statistics show:
1354 Number of megabytes I/O performed.
1357 Aggregate bandwidth of threads in the group.
1360 Minimum average bandwidth a thread saw.
1363 Maximum average bandwidth a thread saw.
1366 Shortest runtime of threads in the group.
1369 Longest runtime of threads in the group.
1373 Finally, disk statistics are printed with reads first:
1378 Number of I/Os performed by all groups.
1381 Number of merges in the I/O scheduler.
1384 Number of ticks we kept the disk busy.
1387 Total time spent in the disk queue.
1394 It is also possible to get fio to dump the current output while it is
1395 running, without terminating the job. To do that, send fio the \fBUSR1\fR
1398 If the \fB\-\-minimal\fR option is given, the results will be printed in a
1399 semicolon-delimited format suitable for scripted use - a job description
1400 (if provided) follows on a new line. Note that the first
1401 number in the line is the version number. If the output has to be changed
1402 for some reason, this number will be incremented by 1 to signify that
1403 change. The fields are:
1406 .B terse version, fio version, jobname, groupid, error
1410 .B Total I/O \fR(KB)\fP, bandwidth \fR(KB/s)\fP, IOPS, runtime \fR(ms)\fP
1414 .B min, max, mean, standard deviation
1418 .B min, max, mean, standard deviation
1420 Completion latency percentiles (20 fields):
1422 .B Xth percentile=usec
1426 .B min, max, mean, standard deviation
1430 .B min, max, aggregate percentage of total, mean, standard deviation
1436 .B Total I/O \fR(KB)\fP, bandwidth \fR(KB/s)\fP, IOPS, runtime \fR(ms)\fP
1440 .B min, max, mean, standard deviation
1444 .B min, max, mean, standard deviation
1446 Completion latency percentiles (20 fields):
1448 .B Xth percentile=usec
1452 .B min, max, mean, standard deviation
1456 .B min, max, aggregate percentage of total, mean, standard deviation
1462 .B user, system, context switches, major page faults, minor page faults
1465 IO depth distribution:
1467 .B <=1, 2, 4, 8, 16, 32, >=64
1470 IO latency distribution:
1474 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
1478 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
1482 Disk utilization (1 for each disk used):
1484 .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
1487 Error Info (dependent on continue_on_error, default off):
1489 .B total # errors, first error code
1492 .B text description (if provided in config - appears on newline)
1495 Normally you would run fio as a stand-alone application on the machine
1496 where the IO workload should be generated. However, it is also possible to
1497 run the frontend and backend of fio separately. This makes it possible to
1498 have a fio server running on the machine(s) where the IO workload should
1499 be running, while controlling it from another machine.
1501 To start the server, you would do:
1503 \fBfio \-\-server=args\fR
1505 on that machine, where args defines what fio listens to. The arguments
1506 are of the form 'type:hostname or IP:port'. 'type' is either 'ip' (or ip4)
1507 for TCP/IP v4, 'ip6' for TCP/IP v6, or 'sock' for a local unix domain
1508 socket. 'hostname' is either a hostname or IP address, and 'port' is the port to
1509 listen to (only valid for TCP/IP, not a local socket). Some examples:
1513 Start a fio server, listening on all interfaces on the default port (8765).
1515 2) fio \-\-server=ip:hostname,4444
1517 Start a fio server, listening on IP belonging to hostname and on port 4444.
1519 3) fio \-\-server=ip6:::1,4444
1521 Start a fio server, listening on IPv6 localhost ::1 and on port 4444.
1523 4) fio \-\-server=,4444
1525 Start a fio server, listening on all interfaces on port 4444.
1527 5) fio \-\-server=1.2.3.4
1529 Start a fio server, listening on IP 1.2.3.4 on the default port.
1531 6) fio \-\-server=sock:/tmp/fio.sock
1533 Start a fio server, listening on the local socket /tmp/fio.sock.
1535 When a server is running, you can connect to it from a client. The client
1538 fio \-\-local-args \-\-client=server \-\-remote-args <job file(s)>
1540 where \-\-local-args are arguments that are local to the client where it is
1541 running, 'server' is the connect string, and \-\-remote-args and <job file(s)>
1542 are sent to the server. The 'server' string follows the same format as it
1543 does on the server side, to allow IP/hostname/socket and port strings.
1544 You can connect to multiple clients as well, to do that you could run:
1546 fio \-\-client=server2 \-\-client=server2 <job file(s)>
1550 was written by Jens Axboe <jens.axboe@oracle.com>,
1551 now Jens Axboe <jaxboe@fusionio.com>.
1553 This man page was written by Aaron Carroll <aaronc@cse.unsw.edu.au> based
1554 on documentation by Jens Axboe.
1555 .SH "REPORTING BUGS"
1556 Report bugs to the \fBfio\fR mailing list <fio@vger.kernel.org>.
1559 For further documentation see \fBHOWTO\fR and \fBREADME\fR.
1561 Sample jobfiles are available in the \fBexamples\fR directory.
projects / fio.git / blob