1 .TH fio 1 "October 2013" "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 Print statistics in selected mode AND terse, semicolon-delimited format.
39 Display version information and exit.
41 .BI \-\-terse\-version \fR=\fPversion
42 Set terse version output format (Current version 3, or older version 2).
45 Display usage information and exit.
48 Perform test and validation of internal CPU clock
50 .BI \-\-crctest[\fR=\fPtest]
51 Test the speed of the builtin checksumming functions. If no argument is given,
52 all of them are tested. Or a comma separated list can be passed, in which
53 case the given ones are tested.
55 .BI \-\-cmdhelp \fR=\fPcommand
56 Print help information for \fIcommand\fR. May be `all' for all commands.
58 .BI \-\-enghelp \fR=\fPioengine[,command]
59 List all commands defined by \fIioengine\fR, or print help for \fIcommand\fR defined by \fIioengine\fR.
61 .BI \-\-showcmd \fR=\fPjobfile
62 Convert \fIjobfile\fR to a set of command-line options.
64 .BI \-\-eta \fR=\fPwhen
65 Specifies when real-time ETA estimate should be printed. \fIwhen\fR may
66 be one of `always', `never' or `auto'.
68 .BI \-\-eta\-newline \fR=\fPtime
69 Force an ETA newline for every `time` period passed.
71 .BI \-\-status\-interval \fR=\fPtime
72 Report full output status every `time` period passed.
75 Turn on safety read-only checks, preventing any attempted write.
77 .BI \-\-section \fR=\fPsec
78 Only run section \fIsec\fR from job file. Multiple of these options can be given, adding more sections to run.
80 .BI \-\-alloc\-size \fR=\fPkb
81 Set the internal smalloc pool size to \fIkb\fP kilobytes.
83 .BI \-\-warnings\-fatal
84 All fio parser warnings are fatal, causing fio to exit with an error.
86 .BI \-\-max\-jobs \fR=\fPnr
87 Set the maximum allowed number of jobs (threads/processes) to support.
89 .BI \-\-server \fR=\fPargs
90 Start a backend server, with \fIargs\fP specifying what to listen to. See client/server section.
92 .BI \-\-daemonize \fR=\fPpidfile
93 Background a fio server, writing the pid to the given pid file.
95 .BI \-\-client \fR=\fPhost
96 Instead of running the jobs locally, send and run them on the given host.
98 .BI \-\-idle\-prof \fR=\fPoption
99 Report cpu idleness on a system or percpu basis (\fIoption\fP=system,percpu) or run unit work calibration only (\fIoption\fP=calibrate).
100 .SH "JOB FILE FORMAT"
101 Job files are in `ini' format. They consist of one or more
102 job definitions, which begin with a job name in square brackets and
103 extend to the next job name. The job name can be any ASCII string
104 except `global', which has a special meaning. Following the job name is
105 a sequence of zero or more parameters, one per line, that define the
106 behavior of the job. Any line starting with a `;' or `#' character is
107 considered a comment and ignored.
109 If \fIjobfile\fR is specified as `-', the job file will be read from
112 The global section contains default parameters for jobs specified in the
113 job file. A job is only affected by global sections residing above it,
114 and there may be any number of global sections. Specific job definitions
115 may override any parameter set in global sections.
118 Some parameters may take arguments of a specific type. The types used are:
121 String: a sequence of alphanumeric characters.
124 SI integer: a whole number, possibly containing a suffix denoting the base unit
125 of the value. Accepted suffixes are `k', 'M', 'G', 'T', and 'P', denoting
126 kilo (1024), mega (1024^2), giga (1024^3), tera (1024^4), and peta (1024^5)
127 respectively. If prefixed with '0x', the value is assumed to be base 16
128 (hexadecimal). A suffix may include a trailing 'b', for instance 'kb' is
129 identical to 'k'. You can specify a base 10 value by using 'KiB', 'MiB','GiB',
130 etc. This is useful for disk drives where values are often given in base 10
131 values. Specifying '30GiB' will get you 30*1000^3 bytes.
132 When specifying times the default suffix meaning changes, still denoting the
133 base unit of the value, but accepted suffixes are 'D' (days), 'H' (hours), 'M'
134 (minutes), 'S' Seconds, 'ms' (or msec) milli seconds, 'us' (or 'usec') micro
135 seconds. Time values without a unit specify seconds.
136 The suffixes are not case sensitive.
139 Boolean: a true or false value. `0' denotes false, `1' denotes true.
142 Integer range: a range of integers specified in the format
143 \fIlower\fR:\fIupper\fR or \fIlower\fR\-\fIupper\fR. \fIlower\fR and
144 \fIupper\fR may contain a suffix as described above. If an option allows two
145 sets of ranges, they are separated with a `,' or `/' character. For example:
149 List of floating numbers: A list of floating numbers, separated by
154 May be used to override the job name. On the command line, this parameter
155 has the special purpose of signalling the start of a new job.
157 .BI description \fR=\fPstr
158 Human-readable description of the job. It is printed when the job is run, but
159 otherwise has no special purpose.
161 .BI directory \fR=\fPstr
162 Prefix filenames with this directory. Used to place files in a location other
164 You can specify a number of directories by separating the names with a ':'
165 character. These directories will be assigned equally distributed to job clones
166 creates with \fInumjobs\fR as long as they are using generated filenames.
167 If specific \fIfilename(s)\fR are set fio will use the first listed directory,
168 and thereby matching the \fIfilename\fR semantic which generates a file each
169 clone if not specified, but let all clones use the same if set.
171 .BI filename \fR=\fPstr
173 normally makes up a file name based on the job name, thread number, and file
174 number. If you want to share files between threads in a job or several jobs,
175 specify a \fIfilename\fR for each of them to override the default.
176 If the I/O engine is file-based, you can specify
177 a number of files by separating the names with a `:' character. `\-' is a
178 reserved name, meaning stdin or stdout, depending on the read/write direction
181 .BI filename_format \fR=\fPstr
182 If sharing multiple files between jobs, it is usually necessary to have
183 fio generate the exact names that you want. By default, fio will name a file
184 based on the default file format specification of
185 \fBjobname.jobnumber.filenumber\fP. With this option, that can be
186 customized. Fio will recognize and replace the following keywords in this
192 The name of the worker thread or process.
195 The incremental number of the worker thread or process.
198 The incremental number of the file for that worker thread or process.
201 To have dependent jobs share a set of files, this option can be set to
202 have fio generate filenames that are shared between the two. For instance,
203 if \fBtestfiles.$filenum\fR is specified, file number 4 for any job will
204 be named \fBtestfiles.4\fR. The default of \fB$jobname.$jobnum.$filenum\fR
205 will be used if no other format specifier is given.
209 .BI lockfile \fR=\fPstr
210 Fio defaults to not locking any files before it does IO to them. If a file or
211 file descriptor is shared, fio can serialize IO to that file to make the end
212 result consistent. This is usual for emulating real workloads that share files.
218 No locking. This is the default.
221 Only one thread or process may do IO at the time, excluding all others.
224 Read-write locking on the file. Many readers may access the file at the same
225 time, but writes get exclusive access.
229 .BI opendir \fR=\fPstr
230 Recursively open any files below directory \fIstr\fR.
232 .BI readwrite \fR=\fPstr "\fR,\fP rw" \fR=\fPstr
233 Type of I/O pattern. Accepted values are:
244 Sequential trim (Linux block devices only).
253 Random trim (Linux block devices only).
256 Mixed sequential reads and writes.
259 Mixed random reads and writes.
262 For mixed I/O, the default split is 50/50. For certain types of io the result
263 may still be skewed a bit, since the speed may be different. It is possible to
264 specify a number of IO's to do before getting a new offset, this is done by
265 appending a `:\fI<nr>\fR to the end of the string given. For a random read, it
266 would look like \fBrw=randread:8\fR for passing in an offset modifier with a
267 value of 8. If the postfix is used with a sequential IO pattern, then the value
268 specified will be added to the generated offset for each IO. For instance,
269 using \fBrw=write:4k\fR will skip 4k for every write. It turns sequential IO
270 into sequential IO with holes. See the \fBrw_sequencer\fR option.
273 .BI rw_sequencer \fR=\fPstr
274 If an offset modifier is given by appending a number to the \fBrw=<str>\fR line,
275 then this option controls how that number modifies the IO offset being
276 generated. Accepted values are:
281 Generate sequential offset
284 Generate the same offset
287 \fBsequential\fR is only useful for random IO, where fio would normally
288 generate a new random offset for every IO. If you append eg 8 to randread, you
289 would get a new random offset for every 8 IO's. The result would be a seek for
290 only every 8 IO's, instead of for every IO. Use \fBrw=randread:8\fR to specify
291 that. As sequential IO is already sequential, setting \fBsequential\fR for that
292 would not result in any differences. \fBidentical\fR behaves in a similar
293 fashion, except it sends the same offset 8 number of times before generating a
298 .BI kb_base \fR=\fPint
299 The base unit for a kilobyte. The defacto base is 2^10, 1024. Storage
300 manufacturers like to use 10^3 or 1000 as a base ten unit instead, for obvious
301 reasons. Allowed values are 1024 or 1000, with 1024 being the default.
303 .BI unified_rw_reporting \fR=\fPbool
304 Fio normally reports statistics on a per data direction basis, meaning that
305 read, write, and trim are accounted and reported separately. If this option is
306 set, the fio will sum the results and report them as "mixed" instead.
308 .BI randrepeat \fR=\fPbool
309 Seed the random number generator used for random I/O patterns in a predictable
310 way so the pattern is repeatable across runs. Default: true.
312 .BI allrandrepeat \fR=\fPbool
313 Seed all random number generators in a predictable way so results are
314 repeatable across runs. Default: false.
316 .BI randseed \fR=\fPint
317 Seed the random number generators based on this seed value, to be able to
318 control what sequence of output is being generated. If not set, the random
319 sequence depends on the \fBrandrepeat\fR setting.
321 .BI use_os_rand \fR=\fPbool
322 Fio can either use the random generator supplied by the OS to generator random
323 offsets, or it can use it's own internal generator (based on Tausworthe).
324 Default is to use the internal generator, which is often of better quality and
325 faster. Default: false.
327 .BI fallocate \fR=\fPstr
328 Whether pre-allocation is performed when laying down files. Accepted values
334 Do not pre-allocate space.
337 Pre-allocate via \fBposix_fallocate\fR\|(3).
340 Pre-allocate via \fBfallocate\fR\|(2) with FALLOC_FL_KEEP_SIZE set.
343 Backward-compatible alias for 'none'.
346 Backward-compatible alias for 'posix'.
349 May not be available on all supported platforms. 'keep' is only
350 available on Linux. If using ZFS on Solaris this must be set to 'none'
351 because ZFS doesn't support it. Default: 'posix'.
354 .BI fadvise_hint \fR=\fPbool
355 Use of \fBposix_fadvise\fR\|(2) to advise the kernel what I/O patterns
356 are likely to be issued. Default: true.
359 Total size of I/O for this job. \fBfio\fR will run until this many bytes have
360 been transferred, unless limited by other options (\fBruntime\fR, for instance).
361 Unless \fBnrfiles\fR and \fBfilesize\fR options are given, this amount will be
362 divided between the available files for the job. If not set, fio will use the
363 full size of the given files or devices. If the files do not exist, size
364 must be given. It is also possible to give size as a percentage between 1 and
365 100. If size=20% is given, fio will use 20% of the full size of the given files
368 .BI fill_device \fR=\fPbool "\fR,\fB fill_fs" \fR=\fPbool
369 Sets size to something really large and waits for ENOSPC (no space left on
370 device) as the terminating condition. Only makes sense with sequential write.
371 For a read workload, the mount point will be filled first then IO started on
372 the result. This option doesn't make sense if operating on a raw device node,
373 since the size of that is already known by the file system. Additionally,
374 writing beyond end-of-device will not return ENOSPC there.
376 .BI filesize \fR=\fPirange
377 Individual file sizes. May be a range, in which case \fBfio\fR will select sizes
378 for files at random within the given range, limited to \fBsize\fR in total (if
379 that is given). If \fBfilesize\fR is not specified, each created file is the
382 .BI blocksize \fR=\fPint[,int] "\fR,\fB bs" \fR=\fPint[,int]
383 Block size for I/O units. Default: 4k. Values for reads, writes, and trims
384 can be specified separately in the format \fIread\fR,\fIwrite\fR,\fItrim\fR
385 either of which may be empty to leave that value at its default. If a trailing
386 comma isn't given, the remainder will inherit the last value set.
388 .BI blocksize_range \fR=\fPirange[,irange] "\fR,\fB bsrange" \fR=\fPirange[,irange]
389 Specify a range of I/O block sizes. The issued I/O unit will always be a
390 multiple of the minimum size, unless \fBblocksize_unaligned\fR is set. Applies
391 to both reads and writes if only one range is given, but can be specified
392 separately with a comma separating the values. Example: bsrange=1k-4k,2k-8k.
393 Also (see \fBblocksize\fR).
395 .BI bssplit \fR=\fPstr
396 This option allows even finer grained control of the block sizes issued,
397 not just even splits between them. With this option, you can weight various
398 block sizes for exact control of the issued IO for a job that has mixed
399 block sizes. The format of the option is bssplit=blocksize/percentage,
400 optionally adding as many definitions as needed separated by a colon.
401 Example: bssplit=4k/10:64k/50:32k/40 would issue 50% 64k blocks, 10% 4k
402 blocks and 40% 32k blocks. \fBbssplit\fR also supports giving separate
403 splits to reads and writes. The format is identical to what the
404 \fBbs\fR option accepts, the read and write parts are separated with a
407 .B blocksize_unaligned\fR,\fP bs_unaligned
408 If set, any size in \fBblocksize_range\fR may be used. This typically won't
409 work with direct I/O, as that normally requires sector alignment.
411 .BI blockalign \fR=\fPint[,int] "\fR,\fB ba" \fR=\fPint[,int]
412 At what boundary to align random IO offsets. Defaults to the same as 'blocksize'
413 the minimum blocksize given. Minimum alignment is typically 512b
414 for using direct IO, though it usually depends on the hardware block size.
415 This option is mutually exclusive with using a random map for files, so it
416 will turn off that option.
418 .BI bs_is_seq_rand \fR=\fPbool
419 If this option is set, fio will use the normal read,write blocksize settings as
420 sequential,random instead. Any random read or write will use the WRITE
421 blocksize settings, and any sequential read or write will use the READ
425 Initialise buffers with all zeros. Default: fill buffers with random data.
428 If this option is given, fio will refill the IO buffers on every submit. The
429 default is to only fill it at init time and reuse that data. Only makes sense
430 if zero_buffers isn't specified, naturally. If data verification is enabled,
431 refill_buffers is also automatically enabled.
433 .BI scramble_buffers \fR=\fPbool
434 If \fBrefill_buffers\fR is too costly and the target is using data
435 deduplication, then setting this option will slightly modify the IO buffer
436 contents to defeat normal de-dupe attempts. This is not enough to defeat
437 more clever block compression attempts, but it will stop naive dedupe
438 of blocks. Default: true.
440 .BI buffer_compress_percentage \fR=\fPint
441 If this is set, then fio will attempt to provide IO buffer content (on WRITEs)
442 that compress to the specified level. Fio does this by providing a mix of
443 random data and zeroes. Note that this is per block size unit, for file/disk
444 wide compression level that matches this setting, you'll also want to set
445 \fBrefill_buffers\fR.
447 .BI buffer_compress_chunk \fR=\fPint
448 See \fBbuffer_compress_percentage\fR. This setting allows fio to manage how
449 big the ranges of random data and zeroed data is. Without this set, fio will
450 provide \fBbuffer_compress_percentage\fR of blocksize random data, followed by
451 the remaining zeroed. With this set to some chunk size smaller than the block
452 size, fio can alternate random and zeroed data throughout the IO buffer.
454 .BI buffer_pattern \fR=\fPstr
455 If set, fio will fill the io buffers with this pattern. If not set, the contents
456 of io buffers is defined by the other options related to buffer contents. The
457 setting can be any pattern of bytes, and can be prefixed with 0x for hex
460 .BI nrfiles \fR=\fPint
461 Number of files to use for this job. Default: 1.
463 .BI openfiles \fR=\fPint
464 Number of files to keep open at the same time. Default: \fBnrfiles\fR.
466 .BI file_service_type \fR=\fPstr
467 Defines how files to service are selected. The following types are defined:
472 Choose a file at random.
475 Round robin over open files (default).
478 Do each file in the set sequentially.
481 The number of I/Os to issue before switching a new file can be specified by
482 appending `:\fIint\fR' to the service type.
485 .BI ioengine \fR=\fPstr
486 Defines how the job issues I/O. The following types are defined:
491 Basic \fBread\fR\|(2) or \fBwrite\fR\|(2) I/O. \fBfseek\fR\|(2) is used to
492 position the I/O location.
495 Basic \fBpread\fR\|(2) or \fBpwrite\fR\|(2) I/O.
498 Basic \fBreadv\fR\|(2) or \fBwritev\fR\|(2) I/O. Will emulate queuing by
499 coalescing adjacent IOs into a single submission.
502 Basic \fBpreadv\fR\|(2) or \fBpwritev\fR\|(2) I/O.
505 Linux native asynchronous I/O. This ioengine defines engine specific options.
508 POSIX asynchronous I/O using \fBaio_read\fR\|(3) and \fBaio_write\fR\|(3).
511 Solaris native asynchronous I/O.
514 Windows native asynchronous I/O.
517 File is memory mapped with \fBmmap\fR\|(2) and data copied using
521 \fBsplice\fR\|(2) is used to transfer the data and \fBvmsplice\fR\|(2) to
522 transfer data from user-space to the kernel.
525 Use the syslet system calls to make regular read/write asynchronous.
528 SCSI generic sg v3 I/O. May be either synchronous using the SG_IO ioctl, or if
529 the target is an sg character device, we use \fBread\fR\|(2) and
530 \fBwrite\fR\|(2) for asynchronous I/O.
533 Doesn't transfer any data, just pretends to. Mainly used to exercise \fBfio\fR
534 itself and for debugging and testing purposes.
537 Transfer over the network. The protocol to be used can be defined with the
538 \fBprotocol\fR parameter. Depending on the protocol, \fBfilename\fR,
539 \fBhostname\fR, \fBport\fR, or \fBlisten\fR must be specified.
540 This ioengine defines engine specific options.
543 Like \fBnet\fR, but uses \fBsplice\fR\|(2) and \fBvmsplice\fR\|(2) to map data
544 and send/receive. This ioengine defines engine specific options.
547 Doesn't transfer any data, but burns CPU cycles according to \fBcpuload\fR and
548 \fBcpucycles\fR parameters.
551 The GUASI I/O engine is the Generic Userspace Asynchronous Syscall Interface
552 approach to asynchronous I/O.
554 See <http://www.xmailserver.org/guasi\-lib.html>.
557 The RDMA I/O engine supports both RDMA memory semantics (RDMA_WRITE/RDMA_READ)
558 and channel semantics (Send/Recv) for the InfiniBand, RoCE and iWARP protocols.
561 Loads an external I/O engine object file. Append the engine filename as
565 IO engine that does regular linux native fallocate call to simulate data
566 transfer as fio ioengine
568 DDIR_READ does fallocate(,mode = FALLOC_FL_KEEP_SIZE,)
570 DIR_WRITE does fallocate(,mode = 0)
572 DDIR_TRIM does fallocate(,mode = FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE)
575 IO engine that does regular EXT4_IOC_MOVE_EXT ioctls to simulate defragment activity
576 request to DDIR_WRITE event
579 IO engine supporting direct access to Ceph Rados Block Devices (RBD) via librbd
580 without the need to use the kernel rbd driver. This ioengine defines engine specific
586 .BI iodepth \fR=\fPint
587 Number of I/O units to keep in flight against the file. Note that increasing
588 iodepth beyond 1 will not affect synchronous ioengines (except for small
589 degress when verify_async is in use). Even async engines my impose OS
590 restrictions causing the desired depth not to be achieved. This may happen on
591 Linux when using libaio and not setting \fBdirect\fR=1, since buffered IO is
592 not async on that OS. Keep an eye on the IO depth distribution in the
593 fio output to verify that the achieved depth is as expected. Default: 1.
595 .BI iodepth_batch \fR=\fPint
596 Number of I/Os to submit at once. Default: \fBiodepth\fR.
598 .BI iodepth_batch_complete \fR=\fPint
599 This defines how many pieces of IO to retrieve at once. It defaults to 1 which
600 means that we'll ask for a minimum of 1 IO in the retrieval process from the
601 kernel. The IO retrieval will go on until we hit the limit set by
602 \fBiodepth_low\fR. If this variable is set to 0, then fio will always check for
603 completed events before queuing more IO. This helps reduce IO latency, at the
604 cost of more retrieval system calls.
606 .BI iodepth_low \fR=\fPint
607 Low watermark indicating when to start filling the queue again. Default:
610 .BI direct \fR=\fPbool
611 If true, use non-buffered I/O (usually O_DIRECT). Default: false.
613 .BI atomic \fR=\fPbool
614 If value is true, attempt to use atomic direct IO. Atomic writes are guaranteed
615 to be stable once acknowledged by the operating system. Only Linux supports
618 .BI buffered \fR=\fPbool
619 If true, use buffered I/O. This is the opposite of the \fBdirect\fR parameter.
622 .BI offset \fR=\fPint
623 Offset in the file to start I/O. Data before the offset will not be touched.
625 .BI offset_increment \fR=\fPint
626 If this is provided, then the real offset becomes the
627 offset + offset_increment * thread_number, where the thread number is a counter
628 that starts at 0 and is incremented for each job. This option is useful if
629 there are several jobs which are intended to operate on a file in parallel in
630 disjoint segments, with even spacing between the starting points.
632 .BI number_ios \fR=\fPint
633 Fio will normally perform IOs until it has exhausted the size of the region
634 set by \fBsize\fR, or if it exhaust the allocated time (or hits an error
635 condition). With this setting, the range/size can be set independently of
636 the number of IOs to perform. When fio reaches this number, it will exit
637 normally and report status.
640 How many I/Os to perform before issuing an \fBfsync\fR\|(2) of dirty data. If
641 0, don't sync. Default: 0.
643 .BI fdatasync \fR=\fPint
644 Like \fBfsync\fR, but uses \fBfdatasync\fR\|(2) instead to only sync the
645 data parts of the file. Default: 0.
647 .BI write_barrier \fR=\fPint
648 Make every Nth write a barrier write.
650 .BI sync_file_range \fR=\fPstr:int
651 Use \fBsync_file_range\fR\|(2) for every \fRval\fP number of write operations. Fio will
652 track range of writes that have happened since the last \fBsync_file_range\fR\|(2) call.
653 \fRstr\fP can currently be one or more of:
657 SYNC_FILE_RANGE_WAIT_BEFORE
660 SYNC_FILE_RANGE_WRITE
663 SYNC_FILE_RANGE_WRITE
667 So if you do sync_file_range=wait_before,write:8, fio would use
668 \fBSYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE\fP for every 8 writes.
669 Also see the \fBsync_file_range\fR\|(2) man page. This option is Linux specific.
671 .BI overwrite \fR=\fPbool
672 If writing, setup the file first and do overwrites. Default: false.
674 .BI end_fsync \fR=\fPbool
675 Sync file contents when a write stage has completed. Default: false.
677 .BI fsync_on_close \fR=\fPbool
678 If true, sync file contents on close. This differs from \fBend_fsync\fR in that
679 it will happen on every close, not just at the end of the job. Default: false.
681 .BI rwmixread \fR=\fPint
682 Percentage of a mixed workload that should be reads. Default: 50.
684 .BI rwmixwrite \fR=\fPint
685 Percentage of a mixed workload that should be writes. If \fBrwmixread\fR and
686 \fBrwmixwrite\fR are given and do not sum to 100%, the latter of the two
687 overrides the first. This may interfere with a given rate setting, if fio is
688 asked to limit reads or writes to a certain rate. If that is the case, then
689 the distribution may be skewed. Default: 50.
691 .BI random_distribution \fR=\fPstr:float
692 By default, fio will use a completely uniform random distribution when asked
693 to perform random IO. Sometimes it is useful to skew the distribution in
694 specific ways, ensuring that some parts of the data is more hot than others.
695 Fio includes the following distribution models:
699 Uniform random distribution
709 When using a zipf or pareto distribution, an input value is also needed to
710 define the access pattern. For zipf, this is the zipf theta. For pareto,
711 it's the pareto power. Fio includes a test program, genzipf, that can be
712 used visualize what the given input values will yield in terms of hit rates.
713 If you wanted to use zipf with a theta of 1.2, you would use
714 random_distribution=zipf:1.2 as the option. If a non-uniform model is used,
715 fio will disable use of the random map.
717 .BI percentage_random \fR=\fPint
718 For a random workload, set how big a percentage should be random. This defaults
719 to 100%, in which case the workload is fully random. It can be set from
720 anywhere from 0 to 100. Setting it to 0 would make the workload fully
721 sequential. It is possible to set different values for reads, writes, and
722 trim. To do so, simply use a comma separated list. See \fBblocksize\fR.
725 Normally \fBfio\fR will cover every block of the file when doing random I/O. If
726 this parameter is given, a new offset will be chosen without looking at past
727 I/O history. This parameter is mutually exclusive with \fBverify\fR.
729 .BI softrandommap \fR=\fPbool
730 See \fBnorandommap\fR. If fio runs with the random block map enabled and it
731 fails to allocate the map, if this option is set it will continue without a
732 random block map. As coverage will not be as complete as with random maps, this
733 option is disabled by default.
735 .BI random_generator \fR=\fPstr
736 Fio supports the following engines for generating IO offsets for random IO:
740 Strong 2^88 cycle random number generator
743 Linear feedback shift register generator
747 Tausworthe is a strong random number generator, but it requires tracking on the
748 side if we want to ensure that blocks are only read or written once. LFSR
749 guarantees that we never generate the same offset twice, and it's also less
750 computationally expensive. It's not a true random generator, however, though
751 for IO purposes it's typically good enough. LFSR only works with single block
752 sizes, not with workloads that use multiple block sizes. If used with such a
753 workload, fio may read or write some blocks multiple times.
756 Run job with given nice value. See \fBnice\fR\|(2).
759 Set I/O priority value of this job between 0 (highest) and 7 (lowest). See
762 .BI prioclass \fR=\fPint
763 Set I/O priority class. See \fBionice\fR\|(1).
765 .BI thinktime \fR=\fPint
766 Stall job for given number of microseconds between issuing I/Os.
768 .BI thinktime_spin \fR=\fPint
769 Pretend to spend CPU time for given number of microseconds, sleeping the rest
770 of the time specified by \fBthinktime\fR. Only valid if \fBthinktime\fR is set.
772 .BI thinktime_blocks \fR=\fPint
773 Only valid if thinktime is set - control how many blocks to issue, before
774 waiting \fBthinktime\fR microseconds. If not set, defaults to 1 which will
775 make fio wait \fBthinktime\fR microseconds after every block. This
776 effectively makes any queue depth setting redundant, since no more than 1 IO
777 will be queued before we have to complete it and do our thinktime. In other
778 words, this setting effectively caps the queue depth if the latter is larger.
782 Cap bandwidth used by this job. The number is in bytes/sec, the normal postfix
783 rules apply. You can use \fBrate\fR=500k to limit reads and writes to 500k each,
784 or you can specify read and writes separately. Using \fBrate\fR=1m,500k would
785 limit reads to 1MB/sec and writes to 500KB/sec. Capping only reads or writes
786 can be done with \fBrate\fR=,500k or \fBrate\fR=500k,. The former will only
787 limit writes (to 500KB/sec), the latter will only limit reads.
789 .BI ratemin \fR=\fPint
790 Tell \fBfio\fR to do whatever it can to maintain at least the given bandwidth.
791 Failing to meet this requirement will cause the job to exit. The same format
792 as \fBrate\fR is used for read vs write separation.
794 .BI rate_iops \fR=\fPint
795 Cap the bandwidth to this number of IOPS. Basically the same as rate, just
796 specified independently of bandwidth. The same format as \fBrate\fR is used for
797 read vs write separation. If \fBblocksize\fR is a range, the smallest block
798 size is used as the metric.
800 .BI rate_iops_min \fR=\fPint
801 If this rate of I/O is not met, the job will exit. The same format as \fBrate\fR
802 is used for read vs write separation.
804 .BI ratecycle \fR=\fPint
805 Average bandwidth for \fBrate\fR and \fBratemin\fR over this number of
806 milliseconds. Default: 1000ms.
808 .BI latency_target \fR=\fPint
809 If set, fio will attempt to find the max performance point that the given
810 workload will run at while maintaining a latency below this target. The
811 values is given in microseconds. See \fBlatency_window\fR and
812 \fBlatency_percentile\fR.
814 .BI latency_window \fR=\fPint
815 Used with \fBlatency_target\fR to specify the sample window that the job
816 is run at varying queue depths to test the performance. The value is given
819 .BI latency_percentile \fR=\fPfloat
820 The percentage of IOs that must fall within the criteria specified by
821 \fBlatency_target\fR and \fBlatency_window\fR. If not set, this defaults
822 to 100.0, meaning that all IOs must be equal or below to the value set
823 by \fBlatency_target\fR.
825 .BI max_latency \fR=\fPint
826 If set, fio will exit the job if it exceeds this maximum latency. It will exit
829 .BI cpumask \fR=\fPint
830 Set CPU affinity for this job. \fIint\fR is a bitmask of allowed CPUs the job
831 may run on. See \fBsched_setaffinity\fR\|(2).
833 .BI cpus_allowed \fR=\fPstr
834 Same as \fBcpumask\fR, but allows a comma-delimited list of CPU numbers.
836 .BI cpus_allowed_policy \fR=\fPstr
837 Set the policy of how fio distributes the CPUs specified by \fBcpus_allowed\fR
838 or \fBcpumask\fR. Two policies are supported:
843 All jobs will share the CPU set specified.
846 Each job will get a unique CPU from the CPU set.
849 \fBshared\fR is the default behaviour, if the option isn't specified. If
850 \fBsplit\fR is specified, then fio will error out if there are more jobs
851 defined than CPUs given in the set.
855 .BI numa_cpu_nodes \fR=\fPstr
856 Set this job running on specified NUMA nodes' CPUs. The arguments allow
857 comma delimited list of cpu numbers, A-B ranges, or 'all'.
859 .BI numa_mem_policy \fR=\fPstr
860 Set this job's memory policy and corresponding NUMA nodes. Format of
864 .B <mode>[:<nodelist>]
867 is one of the following memory policy:
869 .B default, prefer, bind, interleave, local
872 For \fBdefault\fR and \fBlocal\fR memory policy, no \fBnodelist\fR is
873 needed to be specified. For \fBprefer\fR, only one node is
874 allowed. For \fBbind\fR and \fBinterleave\fR, \fBnodelist\fR allows
875 comma delimited list of numbers, A-B ranges, or 'all'.
877 .BI startdelay \fR=\fPirange
878 Delay start of job for the specified number of seconds. Supports all time
879 suffixes to allow specification of hours, minutes, seconds and
880 milliseconds - seconds are the default if a unit is ommited.
881 Can be given as a range which causes each thread to choose randomly out of the
884 .BI runtime \fR=\fPint
885 Terminate processing after the specified number of seconds.
888 If given, run for the specified \fBruntime\fR duration even if the files are
889 completely read or written. The same workload will be repeated as many times
890 as \fBruntime\fR allows.
892 .BI ramp_time \fR=\fPint
893 If set, fio will run the specified workload for this amount of time before
894 logging any performance numbers. Useful for letting performance settle before
895 logging results, thus minimizing the runtime required for stable results. Note
896 that the \fBramp_time\fR is considered lead in time for a job, thus it will
897 increase the total runtime if a special timeout or runtime is specified.
899 .BI invalidate \fR=\fPbool
900 Invalidate buffer-cache for the file prior to starting I/O. Default: true.
903 Use synchronous I/O for buffered writes. For the majority of I/O engines,
904 this means using O_SYNC. Default: false.
906 .BI iomem \fR=\fPstr "\fR,\fP mem" \fR=\fPstr
907 Allocation method for I/O unit buffer. Allowed values are:
912 Allocate memory with \fBmalloc\fR\|(3).
915 Use shared memory buffers allocated through \fBshmget\fR\|(2).
918 Same as \fBshm\fR, but use huge pages as backing.
921 Use \fBmmap\fR\|(2) for allocation. Uses anonymous memory unless a filename
922 is given after the option in the format `:\fIfile\fR'.
925 Same as \fBmmap\fR, but use huge files as backing.
928 The amount of memory allocated is the maximum allowed \fBblocksize\fR for the
929 job multiplied by \fBiodepth\fR. For \fBshmhuge\fR or \fBmmaphuge\fR to work,
930 the system must have free huge pages allocated. \fBmmaphuge\fR also needs to
931 have hugetlbfs mounted, and \fIfile\fR must point there. At least on Linux,
932 huge pages must be manually allocated. See \fB/proc/sys/vm/nr_hugehages\fR
933 and the documentation for that. Normally you just need to echo an appropriate
934 number, eg echoing 8 will ensure that the OS has 8 huge pages ready for
938 .BI iomem_align \fR=\fPint "\fR,\fP mem_align" \fR=\fPint
939 This indicates the memory alignment of the IO memory buffers. Note that the
940 given alignment is applied to the first IO unit buffer, if using \fBiodepth\fR
941 the alignment of the following buffers are given by the \fBbs\fR used. In
942 other words, if using a \fBbs\fR that is a multiple of the page sized in the
943 system, all buffers will be aligned to this value. If using a \fBbs\fR that
944 is not page aligned, the alignment of subsequent IO memory buffers is the
945 sum of the \fBiomem_align\fR and \fBbs\fR used.
947 .BI hugepage\-size \fR=\fPint
948 Defines the size of a huge page. Must be at least equal to the system setting.
949 Should be a multiple of 1MB. Default: 4MB.
952 Terminate all jobs when one finishes. Default: wait for each job to finish.
954 .BI bwavgtime \fR=\fPint
955 Average bandwidth calculations over the given time in milliseconds. Default:
958 .BI iopsavgtime \fR=\fPint
959 Average IOPS calculations over the given time in milliseconds. Default:
962 .BI create_serialize \fR=\fPbool
963 If true, serialize file creation for the jobs. Default: true.
965 .BI create_fsync \fR=\fPbool
966 \fBfsync\fR\|(2) data file after creation. Default: true.
968 .BI create_on_open \fR=\fPbool
969 If true, the files are not created until they are opened for IO by the job.
971 .BI create_only \fR=\fPbool
972 If true, fio will only run the setup phase of the job. If files need to be
973 laid out or updated on disk, only that will be done. The actual job contents
976 .BI pre_read \fR=\fPbool
977 If this is given, files will be pre-read into memory before starting the given
978 IO operation. This will also clear the \fR \fBinvalidate\fR flag, since it is
979 pointless to pre-read and then drop the cache. This will only work for IO
980 engines that are seekable, since they allow you to read the same data
981 multiple times. Thus it will not work on eg network or splice IO.
983 .BI unlink \fR=\fPbool
984 Unlink job files when done. Default: false.
987 Specifies the number of iterations (runs of the same workload) of this job.
990 .BI verify_only \fR=\fPbool
991 Do not perform the specified workload, only verify data still matches previous
992 invocation of this workload. This option allows one to check data multiple
993 times at a later date without overwriting it. This option makes sense only for
994 workloads that write data, and does not support workloads with the
995 \fBtime_based\fR option set.
997 .BI do_verify \fR=\fPbool
998 Run the verify phase after a write phase. Only valid if \fBverify\fR is set.
1001 .BI verify \fR=\fPstr
1002 Method of verifying file contents after each iteration of the job. Allowed
1007 .B md5 crc16 crc32 crc32c crc32c-intel crc64 crc7 sha256 sha512 sha1 xxhash
1008 Store appropriate checksum in the header of each block. crc32c-intel is
1009 hardware accelerated SSE4.2 driven, falls back to regular crc32c if
1010 not supported by the system.
1013 Write extra information about each I/O (timestamp, block number, etc.). The
1014 block number is verified. See \fBverify_pattern\fR as well.
1017 Pretend to verify. Used for testing internals.
1020 This option can be used for repeated burn-in tests of a system to make sure
1021 that the written data is also correctly read back. If the data direction given
1022 is a read or random read, fio will assume that it should verify a previously
1023 written file. If the data direction includes any form of write, the verify will
1024 be of the newly written data.
1027 .BI verifysort \fR=\fPbool
1028 If true, written verify blocks are sorted if \fBfio\fR deems it to be faster to
1029 read them back in a sorted manner. Default: true.
1031 .BI verifysort_nr \fR=\fPint
1032 Pre-load and sort verify blocks for a read workload.
1034 .BI verify_offset \fR=\fPint
1035 Swap the verification header with data somewhere else in the block before
1036 writing. It is swapped back before verifying.
1038 .BI verify_interval \fR=\fPint
1039 Write the verification header for this number of bytes, which should divide
1040 \fBblocksize\fR. Default: \fBblocksize\fR.
1042 .BI verify_pattern \fR=\fPstr
1043 If set, fio will fill the io buffers with this pattern. Fio defaults to filling
1044 with totally random bytes, but sometimes it's interesting to fill with a known
1045 pattern for io verification purposes. Depending on the width of the pattern,
1046 fio will fill 1/2/3/4 bytes of the buffer at the time(it can be either a
1047 decimal or a hex number). The verify_pattern if larger than a 32-bit quantity
1048 has to be a hex number that starts with either "0x" or "0X". Use with
1051 .BI verify_fatal \fR=\fPbool
1052 If true, exit the job on the first observed verification failure. Default:
1055 .BI verify_dump \fR=\fPbool
1056 If set, dump the contents of both the original data block and the data block we
1057 read off disk to files. This allows later analysis to inspect just what kind of
1058 data corruption occurred. Off by default.
1060 .BI verify_async \fR=\fPint
1061 Fio will normally verify IO inline from the submitting thread. This option
1062 takes an integer describing how many async offload threads to create for IO
1063 verification instead, causing fio to offload the duty of verifying IO contents
1064 to one or more separate threads. If using this offload option, even sync IO
1065 engines can benefit from using an \fBiodepth\fR setting higher than 1, as it
1066 allows them to have IO in flight while verifies are running.
1068 .BI verify_async_cpus \fR=\fPstr
1069 Tell fio to set the given CPU affinity on the async IO verification threads.
1070 See \fBcpus_allowed\fP for the format used.
1072 .BI verify_backlog \fR=\fPint
1073 Fio will normally verify the written contents of a job that utilizes verify
1074 once that job has completed. In other words, everything is written then
1075 everything is read back and verified. You may want to verify continually
1076 instead for a variety of reasons. Fio stores the meta data associated with an
1077 IO block in memory, so for large verify workloads, quite a bit of memory would
1078 be used up holding this meta data. If this option is enabled, fio will write
1079 only N blocks before verifying these blocks.
1081 .BI verify_backlog_batch \fR=\fPint
1082 Control how many blocks fio will verify if verify_backlog is set. If not set,
1083 will default to the value of \fBverify_backlog\fR (meaning the entire queue is
1084 read back and verified). If \fBverify_backlog_batch\fR is less than
1085 \fBverify_backlog\fR then not all blocks will be verified, if
1086 \fBverify_backlog_batch\fR is larger than \fBverify_backlog\fR, some blocks
1087 will be verified more than once.
1089 .BI trim_percentage \fR=\fPint
1090 Number of verify blocks to discard/trim.
1092 .BI trim_verify_zero \fR=\fPbool
1093 Verify that trim/discarded blocks are returned as zeroes.
1095 .BI trim_backlog \fR=\fPint
1096 Trim after this number of blocks are written.
1098 .BI trim_backlog_batch \fR=\fPint
1099 Trim this number of IO blocks.
1101 .BI experimental_verify \fR=\fPbool
1102 Enable experimental verification.
1104 .B stonewall "\fR,\fP wait_for_previous"
1105 Wait for preceding jobs in the job file to exit before starting this one.
1106 \fBstonewall\fR implies \fBnew_group\fR.
1109 Start a new reporting group. If not given, all jobs in a file will be part
1110 of the same reporting group, unless separated by a stonewall.
1112 .BI numjobs \fR=\fPint
1113 Number of clones (processes/threads performing the same workload) of this job.
1117 If set, display per-group reports instead of per-job when \fBnumjobs\fR is
1121 Use threads created with \fBpthread_create\fR\|(3) instead of processes created
1122 with \fBfork\fR\|(2).
1124 .BI zonesize \fR=\fPint
1125 Divide file into zones of the specified size in bytes. See \fBzoneskip\fR.
1127 .BI zonerange \fR=\fPint
1128 Give size of an IO zone. See \fBzoneskip\fR.
1130 .BI zoneskip \fR=\fPint
1131 Skip the specified number of bytes when \fBzonesize\fR bytes of data have been
1134 .BI write_iolog \fR=\fPstr
1135 Write the issued I/O patterns to the specified file. Specify a separate file
1136 for each job, otherwise the iologs will be interspersed and the file may be
1139 .BI read_iolog \fR=\fPstr
1140 Replay the I/O patterns contained in the specified file generated by
1141 \fBwrite_iolog\fR, or may be a \fBblktrace\fR binary file.
1143 .BI replay_no_stall \fR=\fPint
1144 While replaying I/O patterns using \fBread_iolog\fR the default behavior
1145 attempts to respect timing information between I/Os. Enabling
1146 \fBreplay_no_stall\fR causes I/Os to be replayed as fast as possible while
1147 still respecting ordering.
1149 .BI replay_redirect \fR=\fPstr
1150 While replaying I/O patterns using \fBread_iolog\fR the default behavior
1151 is to replay the IOPS onto the major/minor device that each IOP was recorded
1152 from. Setting \fBreplay_redirect\fR causes all IOPS to be replayed onto the
1153 single specified device regardless of the device it was recorded from.
1155 .BI write_bw_log \fR=\fPstr
1156 If given, write a bandwidth log of the jobs in this job file. Can be used to
1157 store data of the bandwidth of the jobs in their lifetime. The included
1158 fio_generate_plots script uses gnuplot to turn these text files into nice
1159 graphs. See \fBwrite_lat_log\fR for behaviour of given filename. For this
1160 option, the postfix is _bw.log.
1162 .BI write_lat_log \fR=\fPstr
1163 Same as \fBwrite_bw_log\fR, but writes I/O completion latencies. If no
1164 filename is given with this option, the default filename of "jobname_type.log"
1165 is used. Even if the filename is given, fio will still append the type of log.
1167 .BI write_iops_log \fR=\fPstr
1168 Same as \fBwrite_bw_log\fR, but writes IOPS. If no filename is given with this
1169 option, the default filename of "jobname_type.log" is used. Even if the
1170 filename is given, fio will still append the type of log.
1172 .BI log_avg_msec \fR=\fPint
1173 By default, fio will log an entry in the iops, latency, or bw log for every
1174 IO that completes. When writing to the disk log, that can quickly grow to a
1175 very large size. Setting this option makes fio average the each log entry
1176 over the specified period of time, reducing the resolution of the log.
1179 .BI disable_lat \fR=\fPbool
1180 Disable measurements of total latency numbers. Useful only for cutting
1181 back the number of calls to \fBgettimeofday\fR\|(2), as that does impact performance at
1182 really high IOPS rates. Note that to really get rid of a large amount of these
1183 calls, this option must be used with disable_slat and disable_bw as well.
1185 .BI disable_clat \fR=\fPbool
1186 Disable measurements of completion latency numbers. See \fBdisable_lat\fR.
1188 .BI disable_slat \fR=\fPbool
1189 Disable measurements of submission latency numbers. See \fBdisable_lat\fR.
1191 .BI disable_bw_measurement \fR=\fPbool
1192 Disable measurements of throughput/bandwidth numbers. See \fBdisable_lat\fR.
1194 .BI lockmem \fR=\fPint
1195 Pin the specified amount of memory with \fBmlock\fR\|(2). Can be used to
1196 simulate a smaller amount of memory. The amount specified is per worker.
1198 .BI exec_prerun \fR=\fPstr
1199 Before running the job, execute the specified command with \fBsystem\fR\|(3).
1201 Output is redirected in a file called \fBjobname.prerun.txt\fR
1204 .BI exec_postrun \fR=\fPstr
1205 Same as \fBexec_prerun\fR, but the command is executed after the job completes.
1207 Output is redirected in a file called \fBjobname.postrun.txt\fR
1210 .BI ioscheduler \fR=\fPstr
1211 Attempt to switch the device hosting the file to the specified I/O scheduler.
1213 .BI cpuload \fR=\fPint
1214 If the job is a CPU cycle-eater, attempt to use the specified percentage of
1217 .BI cpuchunks \fR=\fPint
1218 If the job is a CPU cycle-eater, split the load into cycles of the
1219 given time in milliseconds.
1221 .BI disk_util \fR=\fPbool
1222 Generate disk utilization statistics if the platform supports it. Default: true.
1224 .BI clocksource \fR=\fPstr
1225 Use the given clocksource as the base of timing. The supported options are:
1229 \fBgettimeofday\fR\|(2)
1232 \fBclock_gettime\fR\|(2)
1235 Internal CPU clock source
1239 \fBcpu\fR is the preferred clocksource if it is reliable, as it is very fast
1240 (and fio is heavy on time calls). Fio will automatically use this clocksource
1241 if it's supported and considered reliable on the system it is running on,
1242 unless another clocksource is specifically set. For x86/x86-64 CPUs, this
1243 means supporting TSC Invariant.
1245 .BI gtod_reduce \fR=\fPbool
1246 Enable all of the \fBgettimeofday\fR\|(2) reducing options (disable_clat, disable_slat,
1247 disable_bw) plus reduce precision of the timeout somewhat to really shrink the
1248 \fBgettimeofday\fR\|(2) call count. With this option enabled, we only do about 0.4% of
1249 the gtod() calls we would have done if all time keeping was enabled.
1251 .BI gtod_cpu \fR=\fPint
1252 Sometimes it's cheaper to dedicate a single thread of execution to just getting
1253 the current time. Fio (and databases, for instance) are very intensive on
1254 \fBgettimeofday\fR\|(2) calls. With this option, you can set one CPU aside for doing
1255 nothing but logging current time to a shared memory location. Then the other
1256 threads/processes that run IO workloads need only copy that segment, instead of
1257 entering the kernel with a \fBgettimeofday\fR\|(2) call. The CPU set aside for doing
1258 these time calls will be excluded from other uses. Fio will manually clear it
1259 from the CPU mask of other jobs.
1261 .BI ignore_error \fR=\fPstr
1262 Sometimes you want to ignore some errors during test in that case you can specify
1263 error list for each error type.
1265 ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST
1267 errors for given error type is separated with ':'.
1268 Error may be symbol ('ENOSPC', 'ENOMEM') or an integer.
1270 Example: ignore_error=EAGAIN,ENOSPC:122 .
1272 This option will ignore EAGAIN from READ, and ENOSPC and 122(EDQUOT) from WRITE.
1274 .BI error_dump \fR=\fPbool
1275 If set dump every error even if it is non fatal, true by default. If disabled
1276 only fatal error will be dumped
1278 .BI profile \fR=\fPstr
1279 Select a specific builtin performance test.
1281 .BI cgroup \fR=\fPstr
1282 Add job to this control group. If it doesn't exist, it will be created.
1283 The system must have a mounted cgroup blkio mount point for this to work. If
1284 your system doesn't have it mounted, you can do so with:
1286 # mount \-t cgroup \-o blkio none /cgroup
1288 .BI cgroup_weight \fR=\fPint
1289 Set the weight of the cgroup to this value. See the documentation that comes
1290 with the kernel, allowed values are in the range of 100..1000.
1292 .BI cgroup_nodelete \fR=\fPbool
1293 Normally fio will delete the cgroups it has created after the job completion.
1294 To override this behavior and to leave cgroups around after the job completion,
1295 set cgroup_nodelete=1. This can be useful if one wants to inspect various
1296 cgroup files after job completion. Default: false
1299 Instead of running as the invoking user, set the user ID to this value before
1300 the thread/process does any work.
1303 Set group ID, see \fBuid\fR.
1305 .BI unit_base \fR=\fPint
1306 Base unit for reporting. Allowed values are:
1310 Use auto-detection (default).
1320 .BI flow_id \fR=\fPint
1321 The ID of the flow. If not specified, it defaults to being a global flow. See
1325 Weight in token-based flow control. If this value is used, then there is a
1326 \fBflow counter\fR which is used to regulate the proportion of activity between
1327 two or more jobs. fio attempts to keep this flow counter near zero. The
1328 \fBflow\fR parameter stands for how much should be added or subtracted to the
1329 flow counter on each iteration of the main I/O loop. That is, if one job has
1330 \fBflow=8\fR and another job has \fBflow=-1\fR, then there will be a roughly
1331 1:8 ratio in how much one runs vs the other.
1333 .BI flow_watermark \fR=\fPint
1334 The maximum value that the absolute value of the flow counter is allowed to
1335 reach before the job must wait for a lower value of the counter.
1337 .BI flow_sleep \fR=\fPint
1338 The period of time, in microseconds, to wait after the flow watermark has been
1339 exceeded before retrying operations
1341 .BI clat_percentiles \fR=\fPbool
1342 Enable the reporting of percentiles of completion latencies.
1344 .BI percentile_list \fR=\fPfloat_list
1345 Overwrite the default list of percentiles for completion
1346 latencies. Each number is a floating number in the range (0,100], and
1347 the maximum length of the list is 20. Use ':' to separate the
1348 numbers. For example, \-\-percentile_list=99.5:99.9 will cause fio to
1349 report the values of completion latency below which 99.5% and 99.9% of
1350 the observed latencies fell, respectively.
1351 .SS "Ioengine Parameters List"
1352 Some parameters are only valid when a specific ioengine is in use. These are
1353 used identically to normal parameters, with the caveat that when used on the
1354 command line, the must come after the ioengine that defines them is selected.
1356 .BI (cpu)cpuload \fR=\fPint
1357 Attempt to use the specified percentage of CPU cycles.
1359 .BI (cpu)cpuchunks \fR=\fPint
1360 Split the load into cycles of the given time. In microseconds.
1362 .BI (libaio)userspace_reap
1363 Normally, with the libaio engine in use, fio will use
1364 the io_getevents system call to reap newly returned events.
1365 With this flag turned on, the AIO ring will be read directly
1366 from user-space to reap events. The reaping mode is only
1367 enabled when polling for a minimum of 0 events (eg when
1368 iodepth_batch_complete=0).
1370 .BI (net,netsplice)hostname \fR=\fPstr
1371 The host name or IP address to use for TCP or UDP based IO.
1372 If the job is a TCP listener or UDP reader, the hostname is not
1373 used and must be omitted unless it is a valid UDP multicast address.
1375 .BI (net,netsplice)port \fR=\fPint
1376 The TCP or UDP port to bind to or connect to.
1378 .BI (net,netsplice)interface \fR=\fPstr
1379 The IP address of the network interface used to send or receive UDP multicast
1382 .BI (net,netsplice)ttl \fR=\fPint
1383 Time-to-live value for outgoing UDP multicast packets. Default: 1
1385 .BI (net,netsplice)nodelay \fR=\fPbool
1386 Set TCP_NODELAY on TCP connections.
1388 .BI (net,netsplice)protocol \fR=\fPstr "\fR,\fP proto" \fR=\fPstr
1389 The network protocol to use. Accepted values are:
1394 Transmission control protocol
1397 Transmission control protocol V6
1400 User datagram protocol
1403 User datagram protocol V6
1409 When the protocol is TCP or UDP, the port must also be given,
1410 as well as the hostname if the job is a TCP listener or UDP
1411 reader. For unix sockets, the normal filename option should be
1412 used and the port is invalid.
1415 .BI (net,netsplice)listen
1416 For TCP network connections, tell fio to listen for incoming
1417 connections rather than initiating an outgoing connection. The
1418 hostname must be omitted if this option is used.
1420 .BI (net, pingpong) \fR=\fPbool
1421 Normally a network writer will just continue writing data, and a network reader
1422 will just consume packages. If pingpong=1 is set, a writer will send its normal
1423 payload to the reader, then wait for the reader to send the same payload back.
1424 This allows fio to measure network latencies. The submission and completion
1425 latencies then measure local time spent sending or receiving, and the
1426 completion latency measures how long it took for the other end to receive and
1427 send back. For UDP multicast traffic pingpong=1 should only be set for a single
1428 reader when multiple readers are listening to the same address.
1430 .BI (e4defrag,donorname) \fR=\fPstr
1431 File will be used as a block donor (swap extents between files)
1433 .BI (e4defrag,inplace) \fR=\fPint
1434 Configure donor file block allocation strategy
1437 Preallocate donor's file on init
1440 allocate space immediately inside defragment event, and free right after event
1443 .BI (rbd)rbdname \fR=\fPstr
1444 Specifies the name of the RBD.
1446 .BI (rbd)pool \fR=\fPstr
1447 Specifies the name of the Ceph pool containing the RBD.
1449 .BI (rbd)clientname \fR=\fPstr
1450 Specifies the username (without the 'client.' prefix) used to access the Ceph cluster.
1452 While running, \fBfio\fR will display the status of the created jobs. For
1456 Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
1459 The characters in the first set of brackets denote the current status of each
1460 threads. The possible values are:
1466 Setup but not started.
1472 Initialized, waiting.
1475 Running, doing sequential reads.
1478 Running, doing random reads.
1481 Running, doing sequential writes.
1484 Running, doing random writes.
1487 Running, doing mixed sequential reads/writes.
1490 Running, doing mixed random reads/writes.
1493 Running, currently waiting for \fBfsync\fR\|(2).
1496 Running, verifying written data.
1499 Exited, not reaped by main thread.
1502 Exited, thread reaped.
1506 The second set of brackets shows the estimated completion percentage of
1507 the current group. The third set shows the read and write I/O rate,
1508 respectively. Finally, the estimated run time of the job is displayed.
1510 When \fBfio\fR completes (or is interrupted by Ctrl-C), it will show data
1511 for each thread, each group of threads, and each disk, in that order.
1513 Per-thread statistics first show the threads client number, group-id, and
1514 error code. The remaining figures are as follows:
1518 Number of megabytes of I/O performed.
1521 Average data rate (bandwidth).
1527 Submission latency minimum, maximum, average and standard deviation. This is
1528 the time it took to submit the I/O.
1531 Completion latency minimum, maximum, average and standard deviation. This
1532 is the time between submission and completion.
1535 Bandwidth minimum, maximum, percentage of aggregate bandwidth received, average
1536 and standard deviation.
1539 CPU usage statistics. Includes user and system time, number of context switches
1540 this thread went through and number of major and minor page faults.
1543 Distribution of I/O depths. Each depth includes everything less than (or equal)
1544 to it, but greater than the previous depth.
1547 Number of read/write requests issued, and number of short read/write requests.
1550 Distribution of I/O completion latencies. The numbers follow the same pattern
1554 The group statistics show:
1559 Number of megabytes I/O performed.
1562 Aggregate bandwidth of threads in the group.
1565 Minimum average bandwidth a thread saw.
1568 Maximum average bandwidth a thread saw.
1571 Shortest runtime of threads in the group.
1574 Longest runtime of threads in the group.
1578 Finally, disk statistics are printed with reads first:
1583 Number of I/Os performed by all groups.
1586 Number of merges in the I/O scheduler.
1589 Number of ticks we kept the disk busy.
1592 Total time spent in the disk queue.
1599 It is also possible to get fio to dump the current output while it is
1600 running, without terminating the job. To do that, send fio the \fBUSR1\fR
1603 If the \fB\-\-minimal\fR / \fB\-\-append-terse\fR options are given, the
1604 results will be printed/appended in a semicolon-delimited format suitable for
1606 A job description (if provided) follows on a new line. Note that the first
1607 number in the line is the version number. If the output has to be changed
1608 for some reason, this number will be incremented by 1 to signify that
1609 change. The fields are:
1612 .B terse version, fio version, jobname, groupid, error
1616 .B Total I/O \fR(KB)\fP, bandwidth \fR(KB/s)\fP, IOPS, runtime \fR(ms)\fP
1620 .B min, max, mean, standard deviation
1624 .B min, max, mean, standard deviation
1626 Completion latency percentiles (20 fields):
1628 .B Xth percentile=usec
1632 .B min, max, mean, standard deviation
1636 .B min, max, aggregate percentage of total, mean, standard deviation
1642 .B Total I/O \fR(KB)\fP, bandwidth \fR(KB/s)\fP, IOPS, runtime \fR(ms)\fP
1646 .B min, max, mean, standard deviation
1650 .B min, max, mean, standard deviation
1652 Completion latency percentiles (20 fields):
1654 .B Xth percentile=usec
1658 .B min, max, mean, standard deviation
1662 .B min, max, aggregate percentage of total, mean, standard deviation
1668 .B user, system, context switches, major page faults, minor page faults
1671 IO depth distribution:
1673 .B <=1, 2, 4, 8, 16, 32, >=64
1676 IO latency distribution:
1680 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
1684 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
1688 Disk utilization (1 for each disk used):
1690 .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
1693 Error Info (dependent on continue_on_error, default off):
1695 .B total # errors, first error code
1698 .B text description (if provided in config - appears on newline)
1701 Normally you would run fio as a stand-alone application on the machine
1702 where the IO workload should be generated. However, it is also possible to
1703 run the frontend and backend of fio separately. This makes it possible to
1704 have a fio server running on the machine(s) where the IO workload should
1705 be running, while controlling it from another machine.
1707 To start the server, you would do:
1709 \fBfio \-\-server=args\fR
1711 on that machine, where args defines what fio listens to. The arguments
1712 are of the form 'type:hostname or IP:port'. 'type' is either 'ip' (or ip4)
1713 for TCP/IP v4, 'ip6' for TCP/IP v6, or 'sock' for a local unix domain
1714 socket. 'hostname' is either a hostname or IP address, and 'port' is the port to
1715 listen to (only valid for TCP/IP, not a local socket). Some examples:
1719 Start a fio server, listening on all interfaces on the default port (8765).
1721 2) fio \-\-server=ip:hostname,4444
1723 Start a fio server, listening on IP belonging to hostname and on port 4444.
1725 3) fio \-\-server=ip6:::1,4444
1727 Start a fio server, listening on IPv6 localhost ::1 and on port 4444.
1729 4) fio \-\-server=,4444
1731 Start a fio server, listening on all interfaces on port 4444.
1733 5) fio \-\-server=1.2.3.4
1735 Start a fio server, listening on IP 1.2.3.4 on the default port.
1737 6) fio \-\-server=sock:/tmp/fio.sock
1739 Start a fio server, listening on the local socket /tmp/fio.sock.
1741 When a server is running, you can connect to it from a client. The client
1744 fio \-\-local-args \-\-client=server \-\-remote-args <job file(s)>
1746 where \-\-local-args are arguments that are local to the client where it is
1747 running, 'server' is the connect string, and \-\-remote-args and <job file(s)>
1748 are sent to the server. The 'server' string follows the same format as it
1749 does on the server side, to allow IP/hostname/socket and port strings.
1750 You can connect to multiple clients as well, to do that you could run:
1752 fio \-\-client=server2 \-\-client=server2 <job file(s)>
1756 was written by Jens Axboe <jens.axboe@oracle.com>,
1757 now Jens Axboe <jaxboe@fusionio.com>.
1759 This man page was written by Aaron Carroll <aaronc@cse.unsw.edu.au> based
1760 on documentation by Jens Axboe.
1761 .SH "REPORTING BUGS"
1762 Report bugs to the \fBfio\fR mailing list <fio@vger.kernel.org>.
1765 For further documentation see \fBHOWTO\fR and \fBREADME\fR.
1767 Sample jobfiles are available in the \fBexamples\fR directory.
projects / fio.git / blob