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 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.
45 Perform test and validation of internal CPU clock
47 .BI \-\-crctest[\fR=\fPtest]
48 Test the speed of the builtin checksumming functions. If no argument is given,
49 all of them are tested. Or a comma separated list can be passed, in which
50 case the given ones are tested.
52 .BI \-\-cmdhelp \fR=\fPcommand
53 Print help information for \fIcommand\fR. May be `all' for all commands.
55 .BI \-\-enghelp \fR=\fPioengine[,command]
56 List all commands defined by \fIioengine\fR, or print help for \fIcommand\fR defined by \fIioengine\fR.
58 .BI \-\-showcmd \fR=\fPjobfile
59 Convert \fIjobfile\fR to a set of command-line options.
61 .BI \-\-eta \fR=\fPwhen
62 Specifies when real-time ETA estimate should be printed. \fIwhen\fR may
63 be one of `always', `never' or `auto'.
65 .BI \-\-eta\-newline \fR=\fPtime
66 Force an ETA newline for every `time` period passed.
68 .BI \-\-status\-interval \fR=\fPtime
69 Report full output status every `time` period passed.
72 Turn on safety read-only checks, preventing any attempted write.
74 .BI \-\-section \fR=\fPsec
75 Only run section \fIsec\fR from job file. Multiple of these options can be given, adding more sections to run.
77 .BI \-\-alloc\-size \fR=\fPkb
78 Set the internal smalloc pool size to \fIkb\fP kilobytes.
80 .BI \-\-warnings\-fatal
81 All fio parser warnings are fatal, causing fio to exit with an error.
83 .BI \-\-max\-jobs \fR=\fPnr
84 Set the maximum allowed number of jobs (threads/processes) to support.
86 .BI \-\-server \fR=\fPargs
87 Start a backend server, with \fIargs\fP specifying what to listen to. See client/server section.
89 .BI \-\-daemonize \fR=\fPpidfile
90 Background a fio server, writing the pid to the given pid file.
92 .BI \-\-client \fR=\fPhost
93 Instead of running the jobs locally, send and run them on the given host.
95 .BI \-\-idle\-prof \fR=\fPoption
96 Report cpu idleness on a system or percpu basis (\fIoption\fP=system,percpu) or run unit work calibration only (\fIoption\fP=calibrate).
98 Job files are in `ini' format. They consist of one or more
99 job definitions, which begin with a job name in square brackets and
100 extend to the next job name. The job name can be any ASCII string
101 except `global', which has a special meaning. Following the job name is
102 a sequence of zero or more parameters, one per line, that define the
103 behavior of the job. Any line starting with a `;' or `#' character is
104 considered a comment and ignored.
106 If \fIjobfile\fR is specified as `-', the job file will be read from
109 The global section contains default parameters for jobs specified in the
110 job file. A job is only affected by global sections residing above it,
111 and there may be any number of global sections. Specific job definitions
112 may override any parameter set in global sections.
115 Some parameters may take arguments of a specific type. The types used are:
118 String: a sequence of alphanumeric characters.
121 SI integer: a whole number, possibly containing a suffix denoting the base unit
122 of the value. Accepted suffixes are `k', 'M', 'G', 'T', and 'P', denoting
123 kilo (1024), mega (1024^2), giga (1024^3), tera (1024^4), and peta (1024^5)
124 respectively. If prefixed with '0x', the value is assumed to be base 16
125 (hexadecimal). A suffix may include a trailing 'b', for instance 'kb' is
126 identical to 'k'. You can specify a base 10 value by using 'KiB', 'MiB','GiB',
127 etc. This is useful for disk drives where values are often given in base 10
128 values. Specifying '30GiB' will get you 30*1000^3 bytes.
129 When specifying times the default suffix meaning changes, still denoting the
130 base unit of the value, but accepted suffixes are 'D' (days), 'H' (hours), 'M'
131 (minutes), 'S' Seconds, 'ms' milli seconds. Time values without a unit specify
133 The suffixes are not case sensitive.
136 Boolean: a true or false value. `0' denotes false, `1' denotes true.
139 Integer range: a range of integers specified in the format
140 \fIlower\fR:\fIupper\fR or \fIlower\fR\-\fIupper\fR. \fIlower\fR and
141 \fIupper\fR may contain a suffix as described above. If an option allows two
142 sets of ranges, they are separated with a `,' or `/' character. For example:
146 List of floating numbers: A list of floating numbers, separated by
151 May be used to override the job name. On the command line, this parameter
152 has the special purpose of signalling the start of a new job.
154 .BI description \fR=\fPstr
155 Human-readable description of the job. It is printed when the job is run, but
156 otherwise has no special purpose.
158 .BI directory \fR=\fPstr
159 Prefix filenames with this directory. Used to place files in a location other
162 .BI filename \fR=\fPstr
164 normally makes up a file name based on the job name, thread number, and file
165 number. If you want to share files between threads in a job or several jobs,
166 specify a \fIfilename\fR for each of them to override the default.
167 If the I/O engine is file-based, you can specify
168 a number of files by separating the names with a `:' character. `\-' is a
169 reserved name, meaning stdin or stdout, depending on the read/write direction
172 .BI filename_format \fR=\fPstr
173 If sharing multiple files between jobs, it is usually necessary to have
174 fio generate the exact names that you want. By default, fio will name a file
175 based on the default file format specification of
176 \fBjobname.jobnumber.filenumber\fP. With this option, that can be
177 customized. Fio will recognize and replace the following keywords in this
183 The name of the worker thread or process.
186 The incremental number of the worker thread or process.
189 The incremental number of the file for that worker thread or process.
192 To have dependent jobs share a set of files, this option can be set to
193 have fio generate filenames that are shared between the two. For instance,
194 if \fBtestfiles.$filenum\fR is specified, file number 4 for any job will
195 be named \fBtestfiles.4\fR. The default of \fB$jobname.$jobnum.$filenum\fR
196 will be used if no other format specifier is given.
200 .BI lockfile \fR=\fPstr
201 Fio defaults to not locking any files before it does IO to them. If a file or
202 file descriptor is shared, fio can serialize IO to that file to make the end
203 result consistent. This is usual for emulating real workloads that share files.
209 No locking. This is the default.
212 Only one thread or process may do IO at the time, excluding all others.
215 Read-write locking on the file. Many readers may access the file at the same
216 time, but writes get exclusive access.
220 .BI opendir \fR=\fPstr
221 Recursively open any files below directory \fIstr\fR.
223 .BI readwrite \fR=\fPstr "\fR,\fP rw" \fR=\fPstr
224 Type of I/O pattern. Accepted values are:
235 Sequential trim (Linux block devices only).
244 Random trim (Linux block devices only).
247 Mixed sequential reads and writes.
250 Mixed random reads and writes.
253 For mixed I/O, the default split is 50/50. For certain types of io the result
254 may still be skewed a bit, since the speed may be different. It is possible to
255 specify a number of IO's to do before getting a new offset, this is done by
256 appending a `:\fI<nr>\fR to the end of the string given. For a random read, it
257 would look like \fBrw=randread:8\fR for passing in an offset modifier with a
258 value of 8. If the postfix is used with a sequential IO pattern, then the value
259 specified will be added to the generated offset for each IO. For instance,
260 using \fBrw=write:4k\fR will skip 4k for every write. It turns sequential IO
261 into sequential IO with holes. See the \fBrw_sequencer\fR option.
264 .BI rw_sequencer \fR=\fPstr
265 If an offset modifier is given by appending a number to the \fBrw=<str>\fR line,
266 then this option controls how that number modifies the IO offset being
267 generated. Accepted values are:
272 Generate sequential offset
275 Generate the same offset
278 \fBsequential\fR is only useful for random IO, where fio would normally
279 generate a new random offset for every IO. If you append eg 8 to randread, you
280 would get a new random offset for every 8 IO's. The result would be a seek for
281 only every 8 IO's, instead of for every IO. Use \fBrw=randread:8\fR to specify
282 that. As sequential IO is already sequential, setting \fBsequential\fR for that
283 would not result in any differences. \fBidentical\fR behaves in a similar
284 fashion, except it sends the same offset 8 number of times before generating a
289 .BI kb_base \fR=\fPint
290 The base unit for a kilobyte. The defacto base is 2^10, 1024. Storage
291 manufacturers like to use 10^3 or 1000 as a base ten unit instead, for obvious
292 reasons. Allowed values are 1024 or 1000, with 1024 being the default.
294 .BI unified_rw_reporting \fR=\fPbool
295 Fio normally reports statistics on a per data direction basis, meaning that
296 read, write, and trim are accounted and reported separately. If this option is
297 set, the fio will sum the results and report them as "mixed" instead.
299 .BI randrepeat \fR=\fPbool
300 Seed the random number generator used for random I/O patterns in a predictable
301 way so the pattern is repeatable across runs. Default: true.
303 .BI allrandrepeat \fR=\fPbool
304 Seed all random number generators in a predictable way so results are
305 repeatable across runs. Default: false.
307 .BI randseed \fR=\fPint
308 Seed the random number generators based on this seed value, to be able to
309 control what sequence of output is being generated. If not set, the random
310 sequence depends on the \fBrandrepeat\fR setting.
312 .BI use_os_rand \fR=\fPbool
313 Fio can either use the random generator supplied by the OS to generator random
314 offsets, or it can use it's own internal generator (based on Tausworthe).
315 Default is to use the internal generator, which is often of better quality and
316 faster. Default: false.
318 .BI fallocate \fR=\fPstr
319 Whether pre-allocation is performed when laying down files. Accepted values
325 Do not pre-allocate space.
328 Pre-allocate via \fBposix_fallocate\fR\|(3).
331 Pre-allocate via \fBfallocate\fR\|(2) with FALLOC_FL_KEEP_SIZE set.
334 Backward-compatible alias for 'none'.
337 Backward-compatible alias for 'posix'.
340 May not be available on all supported platforms. 'keep' is only
341 available on Linux. If using ZFS on Solaris this must be set to 'none'
342 because ZFS doesn't support it. Default: 'posix'.
345 .BI fadvise_hint \fR=\fPbool
346 Use of \fBposix_fadvise\fR\|(2) to advise the kernel what I/O patterns
347 are likely to be issued. Default: true.
350 Total size of I/O for this job. \fBfio\fR will run until this many bytes have
351 been transferred, unless limited by other options (\fBruntime\fR, for instance).
352 Unless \fBnrfiles\fR and \fBfilesize\fR options are given, this amount will be
353 divided between the available files for the job. If not set, fio will use the
354 full size of the given files or devices. If the files do not exist, size
355 must be given. It is also possible to give size as a percentage between 1 and
356 100. If size=20% is given, fio will use 20% of the full size of the given files
359 .BI fill_device \fR=\fPbool "\fR,\fB fill_fs" \fR=\fPbool
360 Sets size to something really large and waits for ENOSPC (no space left on
361 device) as the terminating condition. Only makes sense with sequential write.
362 For a read workload, the mount point will be filled first then IO started on
363 the result. This option doesn't make sense if operating on a raw device node,
364 since the size of that is already known by the file system. Additionally,
365 writing beyond end-of-device will not return ENOSPC there.
367 .BI filesize \fR=\fPirange
368 Individual file sizes. May be a range, in which case \fBfio\fR will select sizes
369 for files at random within the given range, limited to \fBsize\fR in total (if
370 that is given). If \fBfilesize\fR is not specified, each created file is the
373 .BI blocksize \fR=\fPint[,int] "\fR,\fB bs" \fR=\fPint[,int]
374 Block size for I/O units. Default: 4k. Values for reads, writes, and trims
375 can be specified separately in the format \fIread\fR,\fIwrite\fR,\fItrim\fR
376 either of which may be empty to leave that value at its default. If a trailing
377 comma isn't given, the remainder will inherit the last value set.
379 .BI blocksize_range \fR=\fPirange[,irange] "\fR,\fB bsrange" \fR=\fPirange[,irange]
380 Specify a range of I/O block sizes. The issued I/O unit will always be a
381 multiple of the minimum size, unless \fBblocksize_unaligned\fR is set. Applies
382 to both reads and writes if only one range is given, but can be specified
383 separately with a comma separating the values. Example: bsrange=1k-4k,2k-8k.
384 Also (see \fBblocksize\fR).
386 .BI bssplit \fR=\fPstr
387 This option allows even finer grained control of the block sizes issued,
388 not just even splits between them. With this option, you can weight various
389 block sizes for exact control of the issued IO for a job that has mixed
390 block sizes. The format of the option is bssplit=blocksize/percentage,
391 optionally adding as many definitions as needed separated by a colon.
392 Example: bssplit=4k/10:64k/50:32k/40 would issue 50% 64k blocks, 10% 4k
393 blocks and 40% 32k blocks. \fBbssplit\fR also supports giving separate
394 splits to reads and writes. The format is identical to what the
395 \fBbs\fR option accepts, the read and write parts are separated with a
398 .B blocksize_unaligned\fR,\fP bs_unaligned
399 If set, any size in \fBblocksize_range\fR may be used. This typically won't
400 work with direct I/O, as that normally requires sector alignment.
402 .BI blockalign \fR=\fPint[,int] "\fR,\fB ba" \fR=\fPint[,int]
403 At what boundary to align random IO offsets. Defaults to the same as 'blocksize'
404 the minimum blocksize given. Minimum alignment is typically 512b
405 for using direct IO, though it usually depends on the hardware block size.
406 This option is mutually exclusive with using a random map for files, so it
407 will turn off that option.
409 .BI bs_is_seq_rand \fR=\fPbool
410 If this option is set, fio will use the normal read,write blocksize settings as
411 sequential,random instead. Any random read or write will use the WRITE
412 blocksize settings, and any sequential read or write will use the READ
416 Initialise buffers with all zeros. Default: fill buffers with random data.
419 If this option is given, fio will refill the IO buffers on every submit. The
420 default is to only fill it at init time and reuse that data. Only makes sense
421 if zero_buffers isn't specified, naturally. If data verification is enabled,
422 refill_buffers is also automatically enabled.
424 .BI scramble_buffers \fR=\fPbool
425 If \fBrefill_buffers\fR is too costly and the target is using data
426 deduplication, then setting this option will slightly modify the IO buffer
427 contents to defeat normal de-dupe attempts. This is not enough to defeat
428 more clever block compression attempts, but it will stop naive dedupe
429 of blocks. Default: true.
431 .BI buffer_compress_percentage \fR=\fPint
432 If this is set, then fio will attempt to provide IO buffer content (on WRITEs)
433 that compress to the specified level. Fio does this by providing a mix of
434 random data and zeroes. Note that this is per block size unit, for file/disk
435 wide compression level that matches this setting, you'll also want to set
436 \fBrefill_buffers\fR.
438 .BI buffer_compress_chunk \fR=\fPint
439 See \fBbuffer_compress_percentage\fR. This setting allows fio to manage how
440 big the ranges of random data and zeroed data is. Without this set, fio will
441 provide \fBbuffer_compress_percentage\fR of blocksize random data, followed by
442 the remaining zeroed. With this set to some chunk size smaller than the block
443 size, fio can alternate random and zeroed data throughout the IO buffer.
445 .BI buffer_pattern \fR=\fPstr
446 If set, fio will fill the io buffers with this pattern. If not set, the contents
447 of io buffers is defined by the other options related to buffer contents. The
448 setting can be any pattern of bytes, and can be prefixed with 0x for hex
451 .BI nrfiles \fR=\fPint
452 Number of files to use for this job. Default: 1.
454 .BI openfiles \fR=\fPint
455 Number of files to keep open at the same time. Default: \fBnrfiles\fR.
457 .BI file_service_type \fR=\fPstr
458 Defines how files to service are selected. The following types are defined:
463 Choose a file at random.
466 Round robin over open files (default).
469 Do each file in the set sequentially.
472 The number of I/Os to issue before switching a new file can be specified by
473 appending `:\fIint\fR' to the service type.
476 .BI ioengine \fR=\fPstr
477 Defines how the job issues I/O. The following types are defined:
482 Basic \fBread\fR\|(2) or \fBwrite\fR\|(2) I/O. \fBfseek\fR\|(2) is used to
483 position the I/O location.
486 Basic \fBpread\fR\|(2) or \fBpwrite\fR\|(2) I/O.
489 Basic \fBreadv\fR\|(2) or \fBwritev\fR\|(2) I/O. Will emulate queuing by
490 coalescing adjacent IOs into a single submission.
493 Basic \fBpreadv\fR\|(2) or \fBpwritev\fR\|(2) I/O.
496 Linux native asynchronous I/O. This ioengine defines engine specific options.
499 POSIX asynchronous I/O using \fBaio_read\fR\|(3) and \fBaio_write\fR\|(3).
502 Solaris native asynchronous I/O.
505 Windows native asynchronous I/O.
508 File is memory mapped with \fBmmap\fR\|(2) and data copied using
512 \fBsplice\fR\|(2) is used to transfer the data and \fBvmsplice\fR\|(2) to
513 transfer data from user-space to the kernel.
516 Use the syslet system calls to make regular read/write asynchronous.
519 SCSI generic sg v3 I/O. May be either synchronous using the SG_IO ioctl, or if
520 the target is an sg character device, we use \fBread\fR\|(2) and
521 \fBwrite\fR\|(2) for asynchronous I/O.
524 Doesn't transfer any data, just pretends to. Mainly used to exercise \fBfio\fR
525 itself and for debugging and testing purposes.
528 Transfer over the network. The protocol to be used can be defined with the
529 \fBprotocol\fR parameter. Depending on the protocol, \fBfilename\fR,
530 \fBhostname\fR, \fBport\fR, or \fBlisten\fR must be specified.
531 This ioengine defines engine specific options.
534 Like \fBnet\fR, but uses \fBsplice\fR\|(2) and \fBvmsplice\fR\|(2) to map data
535 and send/receive. This ioengine defines engine specific options.
538 Doesn't transfer any data, but burns CPU cycles according to \fBcpuload\fR and
539 \fBcpucycles\fR parameters.
542 The GUASI I/O engine is the Generic Userspace Asynchronous Syscall Interface
543 approach to asynchronous I/O.
545 See <http://www.xmailserver.org/guasi\-lib.html>.
548 The RDMA I/O engine supports both RDMA memory semantics (RDMA_WRITE/RDMA_READ)
549 and channel semantics (Send/Recv) for the InfiniBand, RoCE and iWARP protocols.
552 Loads an external I/O engine object file. Append the engine filename as
556 IO engine that does regular linux native fallocate call to simulate data
557 transfer as fio ioengine
559 DDIR_READ does fallocate(,mode = FALLOC_FL_KEEP_SIZE,)
561 DIR_WRITE does fallocate(,mode = 0)
563 DDIR_TRIM does fallocate(,mode = FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE)
566 IO engine that does regular EXT4_IOC_MOVE_EXT ioctls to simulate defragment activity
567 request to DDIR_WRITE event
570 IO engine supporting direct access to Ceph Rados Block Devices (RBD) via librbd
571 without the need to use the kernel rbd driver. This ioengine defines engine specific
577 .BI iodepth \fR=\fPint
578 Number of I/O units to keep in flight against the file. Note that increasing
579 iodepth beyond 1 will not affect synchronous ioengines (except for small
580 degress when verify_async is in use). Even async engines my impose OS
581 restrictions causing the desired depth not to be achieved. This may happen on
582 Linux when using libaio and not setting \fBdirect\fR=1, since buffered IO is
583 not async on that OS. Keep an eye on the IO depth distribution in the
584 fio output to verify that the achieved depth is as expected. Default: 1.
586 .BI iodepth_batch \fR=\fPint
587 Number of I/Os to submit at once. Default: \fBiodepth\fR.
589 .BI iodepth_batch_complete \fR=\fPint
590 This defines how many pieces of IO to retrieve at once. It defaults to 1 which
591 means that we'll ask for a minimum of 1 IO in the retrieval process from the
592 kernel. The IO retrieval will go on until we hit the limit set by
593 \fBiodepth_low\fR. If this variable is set to 0, then fio will always check for
594 completed events before queuing more IO. This helps reduce IO latency, at the
595 cost of more retrieval system calls.
597 .BI iodepth_low \fR=\fPint
598 Low watermark indicating when to start filling the queue again. Default:
601 .BI direct \fR=\fPbool
602 If true, use non-buffered I/O (usually O_DIRECT). Default: false.
604 .BI atomic \fR=\fPbool
605 If value is true, attempt to use atomic direct IO. Atomic writes are guaranteed
606 to be stable once acknowledged by the operating system. Only Linux supports
609 .BI buffered \fR=\fPbool
610 If true, use buffered I/O. This is the opposite of the \fBdirect\fR parameter.
613 .BI offset \fR=\fPint
614 Offset in the file to start I/O. Data before the offset will not be touched.
616 .BI offset_increment \fR=\fPint
617 If this is provided, then the real offset becomes the
618 offset + offset_increment * thread_number, where the thread number is a counter
619 that starts at 0 and is incremented for each job. This option is useful if
620 there are several jobs which are intended to operate on a file in parallel in
621 disjoint segments, with even spacing between the starting points.
623 .BI number_ios \fR=\fPint
624 Fio will normally perform IOs until it has exhausted the size of the region
625 set by \fBsize\fR, or if it exhaust the allocated time (or hits an error
626 condition). With this setting, the range/size can be set independently of
627 the number of IOs to perform. When fio reaches this number, it will exit
628 normally and report status.
631 How many I/Os to perform before issuing an \fBfsync\fR\|(2) of dirty data. If
632 0, don't sync. Default: 0.
634 .BI fdatasync \fR=\fPint
635 Like \fBfsync\fR, but uses \fBfdatasync\fR\|(2) instead to only sync the
636 data parts of the file. Default: 0.
638 .BI write_barrier \fR=\fPint
639 Make every Nth write a barrier write.
641 .BI sync_file_range \fR=\fPstr:int
642 Use \fBsync_file_range\fR\|(2) for every \fRval\fP number of write operations. Fio will
643 track range of writes that have happened since the last \fBsync_file_range\fR\|(2) call.
644 \fRstr\fP can currently be one or more of:
648 SYNC_FILE_RANGE_WAIT_BEFORE
651 SYNC_FILE_RANGE_WRITE
654 SYNC_FILE_RANGE_WRITE
658 So if you do sync_file_range=wait_before,write:8, fio would use
659 \fBSYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE\fP for every 8 writes.
660 Also see the \fBsync_file_range\fR\|(2) man page. This option is Linux specific.
662 .BI overwrite \fR=\fPbool
663 If writing, setup the file first and do overwrites. Default: false.
665 .BI end_fsync \fR=\fPbool
666 Sync file contents when a write stage has completed. Default: false.
668 .BI fsync_on_close \fR=\fPbool
669 If true, sync file contents on close. This differs from \fBend_fsync\fR in that
670 it will happen on every close, not just at the end of the job. Default: false.
672 .BI rwmixread \fR=\fPint
673 Percentage of a mixed workload that should be reads. Default: 50.
675 .BI rwmixwrite \fR=\fPint
676 Percentage of a mixed workload that should be writes. If \fBrwmixread\fR and
677 \fBrwmixwrite\fR are given and do not sum to 100%, the latter of the two
678 overrides the first. This may interfere with a given rate setting, if fio is
679 asked to limit reads or writes to a certain rate. If that is the case, then
680 the distribution may be skewed. Default: 50.
682 .BI random_distribution \fR=\fPstr:float
683 By default, fio will use a completely uniform random distribution when asked
684 to perform random IO. Sometimes it is useful to skew the distribution in
685 specific ways, ensuring that some parts of the data is more hot than others.
686 Fio includes the following distribution models:
690 Uniform random distribution
700 When using a zipf or pareto distribution, an input value is also needed to
701 define the access pattern. For zipf, this is the zipf theta. For pareto,
702 it's the pareto power. Fio includes a test program, genzipf, that can be
703 used visualize what the given input values will yield in terms of hit rates.
704 If you wanted to use zipf with a theta of 1.2, you would use
705 random_distribution=zipf:1.2 as the option. If a non-uniform model is used,
706 fio will disable use of the random map.
708 .BI percentage_random \fR=\fPint
709 For a random workload, set how big a percentage should be random. This defaults
710 to 100%, in which case the workload is fully random. It can be set from
711 anywhere from 0 to 100. Setting it to 0 would make the workload fully
712 sequential. It is possible to set different values for reads, writes, and
713 trim. To do so, simply use a comma separated list. See \fBblocksize\fR.
716 Normally \fBfio\fR will cover every block of the file when doing random I/O. If
717 this parameter is given, a new offset will be chosen without looking at past
718 I/O history. This parameter is mutually exclusive with \fBverify\fR.
720 .BI softrandommap \fR=\fPbool
721 See \fBnorandommap\fR. If fio runs with the random block map enabled and it
722 fails to allocate the map, if this option is set it will continue without a
723 random block map. As coverage will not be as complete as with random maps, this
724 option is disabled by default.
726 .BI random_generator \fR=\fPstr
727 Fio supports the following engines for generating IO offsets for random IO:
731 Strong 2^88 cycle random number generator
734 Linear feedback shift register generator
738 Tausworthe is a strong random number generator, but it requires tracking on the
739 side if we want to ensure that blocks are only read or written once. LFSR
740 guarantees that we never generate the same offset twice, and it's also less
741 computationally expensive. It's not a true random generator, however, though
742 for IO purposes it's typically good enough. LFSR only works with single block
743 sizes, not with workloads that use multiple block sizes. If used with such a
744 workload, fio may read or write some blocks multiple times.
747 Run job with given nice value. See \fBnice\fR\|(2).
750 Set I/O priority value of this job between 0 (highest) and 7 (lowest). See
753 .BI prioclass \fR=\fPint
754 Set I/O priority class. See \fBionice\fR\|(1).
756 .BI thinktime \fR=\fPint
757 Stall job for given number of microseconds between issuing I/Os.
759 .BI thinktime_spin \fR=\fPint
760 Pretend to spend CPU time for given number of microseconds, sleeping the rest
761 of the time specified by \fBthinktime\fR. Only valid if \fBthinktime\fR is set.
763 .BI thinktime_blocks \fR=\fPint
764 Only valid if thinktime is set - control how many blocks to issue, before
765 waiting \fBthinktime\fR microseconds. If not set, defaults to 1 which will
766 make fio wait \fBthinktime\fR microseconds after every block. This
767 effectively makes any queue depth setting redundant, since no more than 1 IO
768 will be queued before we have to complete it and do our thinktime. In other
769 words, this setting effectively caps the queue depth if the latter is larger.
773 Cap bandwidth used by this job. The number is in bytes/sec, the normal postfix
774 rules apply. You can use \fBrate\fR=500k to limit reads and writes to 500k each,
775 or you can specify read and writes separately. Using \fBrate\fR=1m,500k would
776 limit reads to 1MB/sec and writes to 500KB/sec. Capping only reads or writes
777 can be done with \fBrate\fR=,500k or \fBrate\fR=500k,. The former will only
778 limit writes (to 500KB/sec), the latter will only limit reads.
780 .BI ratemin \fR=\fPint
781 Tell \fBfio\fR to do whatever it can to maintain at least the given bandwidth.
782 Failing to meet this requirement will cause the job to exit. The same format
783 as \fBrate\fR is used for read vs write separation.
785 .BI rate_iops \fR=\fPint
786 Cap the bandwidth to this number of IOPS. Basically the same as rate, just
787 specified independently of bandwidth. The same format as \fBrate\fR is used for
788 read vs write separation. If \fBblocksize\fR is a range, the smallest block
789 size is used as the metric.
791 .BI rate_iops_min \fR=\fPint
792 If this rate of I/O is not met, the job will exit. The same format as \fBrate\fR
793 is used for read vs write separation.
795 .BI ratecycle \fR=\fPint
796 Average bandwidth for \fBrate\fR and \fBratemin\fR over this number of
797 milliseconds. Default: 1000ms.
799 .BI latency_target \fR=\fPint
800 If set, fio will attempt to find the max performance point that the given
801 workload will run at while maintaining a latency below this target. The
802 values is given in microseconds. See \fBlatency_window\fR and
803 \fBlatency_percentile\fR.
805 .BI latency_window \fR=\fPint
806 Used with \fBlatency_target\fR to specify the sample window that the job
807 is run at varying queue depths to test the performance. The value is given
810 .BI latency_percentile \fR=\fPfloat
811 The percentage of IOs that must fall within the criteria specified by
812 \fBlatency_target\fR and \fBlatency_window\fR. If not set, this defaults
813 to 100.0, meaning that all IOs must be equal or below to the value set
814 by \fBlatency_target\fR.
816 .BI max_latency \fR=\fPint
817 If set, fio will exit the job if it exceeds this maximum latency. It will exit
820 .BI cpumask \fR=\fPint
821 Set CPU affinity for this job. \fIint\fR is a bitmask of allowed CPUs the job
822 may run on. See \fBsched_setaffinity\fR\|(2).
824 .BI cpus_allowed \fR=\fPstr
825 Same as \fBcpumask\fR, but allows a comma-delimited list of CPU numbers.
827 .BI numa_cpu_nodes \fR=\fPstr
828 Set this job running on specified NUMA nodes' CPUs. The arguments allow
829 comma delimited list of cpu numbers, A-B ranges, or 'all'.
831 .BI numa_mem_policy \fR=\fPstr
832 Set this job's memory policy and corresponding NUMA nodes. Format of
836 .B <mode>[:<nodelist>]
839 is one of the following memory policy:
841 .B default, prefer, bind, interleave, local
844 For \fBdefault\fR and \fBlocal\fR memory policy, no \fBnodelist\fR is
845 needed to be specified. For \fBprefer\fR, only one node is
846 allowed. For \fBbind\fR and \fBinterleave\fR, \fBnodelist\fR allows
847 comma delimited list of numbers, A-B ranges, or 'all'.
849 .BI startdelay \fR=\fPirange
850 Delay start of job for the specified number of seconds. Supports all time
851 suffixes to allow specification of hours, minutes, seconds and
852 milliseconds - seconds are the default if a unit is ommited.
853 Can be given as a range which causes each thread to choose randomly out of the
856 .BI runtime \fR=\fPint
857 Terminate processing after the specified number of seconds.
860 If given, run for the specified \fBruntime\fR duration even if the files are
861 completely read or written. The same workload will be repeated as many times
862 as \fBruntime\fR allows.
864 .BI ramp_time \fR=\fPint
865 If set, fio will run the specified workload for this amount of time before
866 logging any performance numbers. Useful for letting performance settle before
867 logging results, thus minimizing the runtime required for stable results. Note
868 that the \fBramp_time\fR is considered lead in time for a job, thus it will
869 increase the total runtime if a special timeout or runtime is specified.
871 .BI invalidate \fR=\fPbool
872 Invalidate buffer-cache for the file prior to starting I/O. Default: true.
875 Use synchronous I/O for buffered writes. For the majority of I/O engines,
876 this means using O_SYNC. Default: false.
878 .BI iomem \fR=\fPstr "\fR,\fP mem" \fR=\fPstr
879 Allocation method for I/O unit buffer. Allowed values are:
884 Allocate memory with \fBmalloc\fR\|(3).
887 Use shared memory buffers allocated through \fBshmget\fR\|(2).
890 Same as \fBshm\fR, but use huge pages as backing.
893 Use \fBmmap\fR\|(2) for allocation. Uses anonymous memory unless a filename
894 is given after the option in the format `:\fIfile\fR'.
897 Same as \fBmmap\fR, but use huge files as backing.
900 The amount of memory allocated is the maximum allowed \fBblocksize\fR for the
901 job multiplied by \fBiodepth\fR. For \fBshmhuge\fR or \fBmmaphuge\fR to work,
902 the system must have free huge pages allocated. \fBmmaphuge\fR also needs to
903 have hugetlbfs mounted, and \fIfile\fR must point there. At least on Linux,
904 huge pages must be manually allocated. See \fB/proc/sys/vm/nr_hugehages\fR
905 and the documentation for that. Normally you just need to echo an appropriate
906 number, eg echoing 8 will ensure that the OS has 8 huge pages ready for
910 .BI iomem_align \fR=\fPint "\fR,\fP mem_align" \fR=\fPint
911 This indicates the memory alignment of the IO memory buffers. Note that the
912 given alignment is applied to the first IO unit buffer, if using \fBiodepth\fR
913 the alignment of the following buffers are given by the \fBbs\fR used. In
914 other words, if using a \fBbs\fR that is a multiple of the page sized in the
915 system, all buffers will be aligned to this value. If using a \fBbs\fR that
916 is not page aligned, the alignment of subsequent IO memory buffers is the
917 sum of the \fBiomem_align\fR and \fBbs\fR used.
919 .BI hugepage\-size \fR=\fPint
920 Defines the size of a huge page. Must be at least equal to the system setting.
921 Should be a multiple of 1MB. Default: 4MB.
924 Terminate all jobs when one finishes. Default: wait for each job to finish.
926 .BI bwavgtime \fR=\fPint
927 Average bandwidth calculations over the given time in milliseconds. Default:
930 .BI iopsavgtime \fR=\fPint
931 Average IOPS calculations over the given time in milliseconds. Default:
934 .BI create_serialize \fR=\fPbool
935 If true, serialize file creation for the jobs. Default: true.
937 .BI create_fsync \fR=\fPbool
938 \fBfsync\fR\|(2) data file after creation. Default: true.
940 .BI create_on_open \fR=\fPbool
941 If true, the files are not created until they are opened for IO by the job.
943 .BI create_only \fR=\fPbool
944 If true, fio will only run the setup phase of the job. If files need to be
945 laid out or updated on disk, only that will be done. The actual job contents
948 .BI pre_read \fR=\fPbool
949 If this is given, files will be pre-read into memory before starting the given
950 IO operation. This will also clear the \fR \fBinvalidate\fR flag, since it is
951 pointless to pre-read and then drop the cache. This will only work for IO
952 engines that are seekable, since they allow you to read the same data
953 multiple times. Thus it will not work on eg network or splice IO.
955 .BI unlink \fR=\fPbool
956 Unlink job files when done. Default: false.
959 Specifies the number of iterations (runs of the same workload) of this job.
962 .BI verify_only \fR=\fPbool
963 Do not perform the specified workload, only verify data still matches previous
964 invocation of this workload. This option allows one to check data multiple
965 times at a later date without overwriting it. This option makes sense only for
966 workloads that write data, and does not support workloads with the
967 \fBtime_based\fR option set.
969 .BI do_verify \fR=\fPbool
970 Run the verify phase after a write phase. Only valid if \fBverify\fR is set.
973 .BI verify \fR=\fPstr
974 Method of verifying file contents after each iteration of the job. Allowed
979 .B md5 crc16 crc32 crc32c crc32c-intel crc64 crc7 sha256 sha512 sha1
980 Store appropriate checksum in the header of each block. crc32c-intel is
981 hardware accelerated SSE4.2 driven, falls back to regular crc32c if
982 not supported by the system.
985 Write extra information about each I/O (timestamp, block number, etc.). The
986 block number is verified. See \fBverify_pattern\fR as well.
989 Pretend to verify. Used for testing internals.
992 This option can be used for repeated burn-in tests of a system to make sure
993 that the written data is also correctly read back. If the data direction given
994 is a read or random read, fio will assume that it should verify a previously
995 written file. If the data direction includes any form of write, the verify will
996 be of the newly written data.
999 .BI verifysort \fR=\fPbool
1000 If true, written verify blocks are sorted if \fBfio\fR deems it to be faster to
1001 read them back in a sorted manner. Default: true.
1003 .BI verifysort_nr \fR=\fPint
1004 Pre-load and sort verify blocks for a read workload.
1006 .BI verify_offset \fR=\fPint
1007 Swap the verification header with data somewhere else in the block before
1008 writing. It is swapped back before verifying.
1010 .BI verify_interval \fR=\fPint
1011 Write the verification header for this number of bytes, which should divide
1012 \fBblocksize\fR. Default: \fBblocksize\fR.
1014 .BI verify_pattern \fR=\fPstr
1015 If set, fio will fill the io buffers with this pattern. Fio defaults to filling
1016 with totally random bytes, but sometimes it's interesting to fill with a known
1017 pattern for io verification purposes. Depending on the width of the pattern,
1018 fio will fill 1/2/3/4 bytes of the buffer at the time(it can be either a
1019 decimal or a hex number). The verify_pattern if larger than a 32-bit quantity
1020 has to be a hex number that starts with either "0x" or "0X". Use with
1023 .BI verify_fatal \fR=\fPbool
1024 If true, exit the job on the first observed verification failure. Default:
1027 .BI verify_dump \fR=\fPbool
1028 If set, dump the contents of both the original data block and the data block we
1029 read off disk to files. This allows later analysis to inspect just what kind of
1030 data corruption occurred. Off by default.
1032 .BI verify_async \fR=\fPint
1033 Fio will normally verify IO inline from the submitting thread. This option
1034 takes an integer describing how many async offload threads to create for IO
1035 verification instead, causing fio to offload the duty of verifying IO contents
1036 to one or more separate threads. If using this offload option, even sync IO
1037 engines can benefit from using an \fBiodepth\fR setting higher than 1, as it
1038 allows them to have IO in flight while verifies are running.
1040 .BI verify_async_cpus \fR=\fPstr
1041 Tell fio to set the given CPU affinity on the async IO verification threads.
1042 See \fBcpus_allowed\fP for the format used.
1044 .BI verify_backlog \fR=\fPint
1045 Fio will normally verify the written contents of a job that utilizes verify
1046 once that job has completed. In other words, everything is written then
1047 everything is read back and verified. You may want to verify continually
1048 instead for a variety of reasons. Fio stores the meta data associated with an
1049 IO block in memory, so for large verify workloads, quite a bit of memory would
1050 be used up holding this meta data. If this option is enabled, fio will write
1051 only N blocks before verifying these blocks.
1053 .BI verify_backlog_batch \fR=\fPint
1054 Control how many blocks fio will verify if verify_backlog is set. If not set,
1055 will default to the value of \fBverify_backlog\fR (meaning the entire queue is
1056 read back and verified). If \fBverify_backlog_batch\fR is less than
1057 \fBverify_backlog\fR then not all blocks will be verified, if
1058 \fBverify_backlog_batch\fR is larger than \fBverify_backlog\fR, some blocks
1059 will be verified more than once.
1061 .BI trim_percentage \fR=\fPint
1062 Number of verify blocks to discard/trim.
1064 .BI trim_verify_zero \fR=\fPbool
1065 Verify that trim/discarded blocks are returned as zeroes.
1067 .BI trim_backlog \fR=\fPint
1068 Trim after this number of blocks are written.
1070 .BI trim_backlog_batch \fR=\fPint
1071 Trim this number of IO blocks.
1073 .BI experimental_verify \fR=\fPbool
1074 Enable experimental verification.
1076 .B stonewall "\fR,\fP wait_for_previous"
1077 Wait for preceding jobs in the job file to exit before starting this one.
1078 \fBstonewall\fR implies \fBnew_group\fR.
1081 Start a new reporting group. If not given, all jobs in a file will be part
1082 of the same reporting group, unless separated by a stonewall.
1084 .BI numjobs \fR=\fPint
1085 Number of clones (processes/threads performing the same workload) of this job.
1089 If set, display per-group reports instead of per-job when \fBnumjobs\fR is
1093 Use threads created with \fBpthread_create\fR\|(3) instead of processes created
1094 with \fBfork\fR\|(2).
1096 .BI zonesize \fR=\fPint
1097 Divide file into zones of the specified size in bytes. See \fBzoneskip\fR.
1099 .BI zonerange \fR=\fPint
1100 Give size of an IO zone. See \fBzoneskip\fR.
1102 .BI zoneskip \fR=\fPint
1103 Skip the specified number of bytes when \fBzonesize\fR bytes of data have been
1106 .BI write_iolog \fR=\fPstr
1107 Write the issued I/O patterns to the specified file. Specify a separate file
1108 for each job, otherwise the iologs will be interspersed and the file may be
1111 .BI read_iolog \fR=\fPstr
1112 Replay the I/O patterns contained in the specified file generated by
1113 \fBwrite_iolog\fR, or may be a \fBblktrace\fR binary file.
1115 .BI replay_no_stall \fR=\fPint
1116 While replaying I/O patterns using \fBread_iolog\fR the default behavior
1117 attempts to respect timing information between I/Os. Enabling
1118 \fBreplay_no_stall\fR causes I/Os to be replayed as fast as possible while
1119 still respecting ordering.
1121 .BI replay_redirect \fR=\fPstr
1122 While replaying I/O patterns using \fBread_iolog\fR the default behavior
1123 is to replay the IOPS onto the major/minor device that each IOP was recorded
1124 from. Setting \fBreplay_redirect\fR causes all IOPS to be replayed onto the
1125 single specified device regardless of the device it was recorded from.
1127 .BI write_bw_log \fR=\fPstr
1128 If given, write a bandwidth log of the jobs in this job file. Can be used to
1129 store data of the bandwidth of the jobs in their lifetime. The included
1130 fio_generate_plots script uses gnuplot to turn these text files into nice
1131 graphs. See \fBwrite_lat_log\fR for behaviour of given filename. For this
1132 option, the postfix is _bw.log.
1134 .BI write_lat_log \fR=\fPstr
1135 Same as \fBwrite_bw_log\fR, but writes I/O completion latencies. If no
1136 filename is given with this option, the default filename of "jobname_type.log"
1137 is used. Even if the filename is given, fio will still append the type of log.
1139 .BI write_iops_log \fR=\fPstr
1140 Same as \fBwrite_bw_log\fR, but writes IOPS. If no filename is given with this
1141 option, the default filename of "jobname_type.log" is used. Even if the
1142 filename is given, fio will still append the type of log.
1144 .BI log_avg_msec \fR=\fPint
1145 By default, fio will log an entry in the iops, latency, or bw log for every
1146 IO that completes. When writing to the disk log, that can quickly grow to a
1147 very large size. Setting this option makes fio average the each log entry
1148 over the specified period of time, reducing the resolution of the log.
1151 .BI disable_lat \fR=\fPbool
1152 Disable measurements of total latency numbers. Useful only for cutting
1153 back the number of calls to \fBgettimeofday\fR\|(2), as that does impact performance at
1154 really high IOPS rates. Note that to really get rid of a large amount of these
1155 calls, this option must be used with disable_slat and disable_bw as well.
1157 .BI disable_clat \fR=\fPbool
1158 Disable measurements of completion latency numbers. See \fBdisable_lat\fR.
1160 .BI disable_slat \fR=\fPbool
1161 Disable measurements of submission latency numbers. See \fBdisable_lat\fR.
1163 .BI disable_bw_measurement \fR=\fPbool
1164 Disable measurements of throughput/bandwidth numbers. See \fBdisable_lat\fR.
1166 .BI lockmem \fR=\fPint
1167 Pin the specified amount of memory with \fBmlock\fR\|(2). Can be used to
1168 simulate a smaller amount of memory. The amount specified is per worker.
1170 .BI exec_prerun \fR=\fPstr
1171 Before running the job, execute the specified command with \fBsystem\fR\|(3).
1173 Output is redirected in a file called \fBjobname.prerun.txt\fR
1176 .BI exec_postrun \fR=\fPstr
1177 Same as \fBexec_prerun\fR, but the command is executed after the job completes.
1179 Output is redirected in a file called \fBjobname.postrun.txt\fR
1182 .BI ioscheduler \fR=\fPstr
1183 Attempt to switch the device hosting the file to the specified I/O scheduler.
1185 .BI cpuload \fR=\fPint
1186 If the job is a CPU cycle-eater, attempt to use the specified percentage of
1189 .BI cpuchunks \fR=\fPint
1190 If the job is a CPU cycle-eater, split the load into cycles of the
1191 given time in milliseconds.
1193 .BI disk_util \fR=\fPbool
1194 Generate disk utilization statistics if the platform supports it. Default: true.
1196 .BI clocksource \fR=\fPstr
1197 Use the given clocksource as the base of timing. The supported options are:
1201 \fBgettimeofday\fR\|(2)
1204 \fBclock_gettime\fR\|(2)
1207 Internal CPU clock source
1211 \fBcpu\fR is the preferred clocksource if it is reliable, as it is very fast
1212 (and fio is heavy on time calls). Fio will automatically use this clocksource
1213 if it's supported and considered reliable on the system it is running on,
1214 unless another clocksource is specifically set. For x86/x86-64 CPUs, this
1215 means supporting TSC Invariant.
1217 .BI gtod_reduce \fR=\fPbool
1218 Enable all of the \fBgettimeofday\fR\|(2) reducing options (disable_clat, disable_slat,
1219 disable_bw) plus reduce precision of the timeout somewhat to really shrink the
1220 \fBgettimeofday\fR\|(2) call count. With this option enabled, we only do about 0.4% of
1221 the gtod() calls we would have done if all time keeping was enabled.
1223 .BI gtod_cpu \fR=\fPint
1224 Sometimes it's cheaper to dedicate a single thread of execution to just getting
1225 the current time. Fio (and databases, for instance) are very intensive on
1226 \fBgettimeofday\fR\|(2) calls. With this option, you can set one CPU aside for doing
1227 nothing but logging current time to a shared memory location. Then the other
1228 threads/processes that run IO workloads need only copy that segment, instead of
1229 entering the kernel with a \fBgettimeofday\fR\|(2) call. The CPU set aside for doing
1230 these time calls will be excluded from other uses. Fio will manually clear it
1231 from the CPU mask of other jobs.
1233 .BI ignore_error \fR=\fPstr
1234 Sometimes you want to ignore some errors during test in that case you can specify
1235 error list for each error type.
1237 ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST
1239 errors for given error type is separated with ':'.
1240 Error may be symbol ('ENOSPC', 'ENOMEM') or an integer.
1242 Example: ignore_error=EAGAIN,ENOSPC:122 .
1244 This option will ignore EAGAIN from READ, and ENOSPC and 122(EDQUOT) from WRITE.
1246 .BI error_dump \fR=\fPbool
1247 If set dump every error even if it is non fatal, true by default. If disabled
1248 only fatal error will be dumped
1250 .BI profile \fR=\fPstr
1251 Select a specific builtin performance test.
1253 .BI cgroup \fR=\fPstr
1254 Add job to this control group. If it doesn't exist, it will be created.
1255 The system must have a mounted cgroup blkio mount point for this to work. If
1256 your system doesn't have it mounted, you can do so with:
1258 # mount \-t cgroup \-o blkio none /cgroup
1260 .BI cgroup_weight \fR=\fPint
1261 Set the weight of the cgroup to this value. See the documentation that comes
1262 with the kernel, allowed values are in the range of 100..1000.
1264 .BI cgroup_nodelete \fR=\fPbool
1265 Normally fio will delete the cgroups it has created after the job completion.
1266 To override this behavior and to leave cgroups around after the job completion,
1267 set cgroup_nodelete=1. This can be useful if one wants to inspect various
1268 cgroup files after job completion. Default: false
1271 Instead of running as the invoking user, set the user ID to this value before
1272 the thread/process does any work.
1275 Set group ID, see \fBuid\fR.
1277 .BI unit_base \fR=\fPint
1278 Base unit for reporting. Allowed values are:
1282 Use auto-detection (default).
1292 .BI flow_id \fR=\fPint
1293 The ID of the flow. If not specified, it defaults to being a global flow. See
1297 Weight in token-based flow control. If this value is used, then there is a
1298 \fBflow counter\fR which is used to regulate the proportion of activity between
1299 two or more jobs. fio attempts to keep this flow counter near zero. The
1300 \fBflow\fR parameter stands for how much should be added or subtracted to the
1301 flow counter on each iteration of the main I/O loop. That is, if one job has
1302 \fBflow=8\fR and another job has \fBflow=-1\fR, then there will be a roughly
1303 1:8 ratio in how much one runs vs the other.
1305 .BI flow_watermark \fR=\fPint
1306 The maximum value that the absolute value of the flow counter is allowed to
1307 reach before the job must wait for a lower value of the counter.
1309 .BI flow_sleep \fR=\fPint
1310 The period of time, in microseconds, to wait after the flow watermark has been
1311 exceeded before retrying operations
1313 .BI clat_percentiles \fR=\fPbool
1314 Enable the reporting of percentiles of completion latencies.
1316 .BI percentile_list \fR=\fPfloat_list
1317 Overwrite the default list of percentiles for completion
1318 latencies. Each number is a floating number in the range (0,100], and
1319 the maximum length of the list is 20. Use ':' to separate the
1320 numbers. For example, \-\-percentile_list=99.5:99.9 will cause fio to
1321 report the values of completion latency below which 99.5% and 99.9% of
1322 the observed latencies fell, respectively.
1323 .SS "Ioengine Parameters List"
1324 Some parameters are only valid when a specific ioengine is in use. These are
1325 used identically to normal parameters, with the caveat that when used on the
1326 command line, the must come after the ioengine that defines them is selected.
1328 .BI (cpu)cpuload \fR=\fPint
1329 Attempt to use the specified percentage of CPU cycles.
1331 .BI (cpu)cpuchunks \fR=\fPint
1332 Split the load into cycles of the given time. In microseconds.
1334 .BI (libaio)userspace_reap
1335 Normally, with the libaio engine in use, fio will use
1336 the io_getevents system call to reap newly returned events.
1337 With this flag turned on, the AIO ring will be read directly
1338 from user-space to reap events. The reaping mode is only
1339 enabled when polling for a minimum of 0 events (eg when
1340 iodepth_batch_complete=0).
1342 .BI (net,netsplice)hostname \fR=\fPstr
1343 The host name or IP address to use for TCP or UDP based IO.
1344 If the job is a TCP listener or UDP reader, the hostname is not
1345 used and must be omitted unless it is a valid UDP multicast address.
1347 .BI (net,netsplice)port \fR=\fPint
1348 The TCP or UDP port to bind to or connect to.
1350 .BI (net,netsplice)interface \fR=\fPstr
1351 The IP address of the network interface used to send or receive UDP multicast
1354 .BI (net,netsplice)ttl \fR=\fPint
1355 Time-to-live value for outgoing UDP multicast packets. Default: 1
1357 .BI (net,netsplice)nodelay \fR=\fPbool
1358 Set TCP_NODELAY on TCP connections.
1360 .BI (net,netsplice)protocol \fR=\fPstr "\fR,\fP proto" \fR=\fPstr
1361 The network protocol to use. Accepted values are:
1366 Transmission control protocol
1369 Transmission control protocol V6
1372 User datagram protocol
1375 User datagram protocol V6
1381 When the protocol is TCP or UDP, the port must also be given,
1382 as well as the hostname if the job is a TCP listener or UDP
1383 reader. For unix sockets, the normal filename option should be
1384 used and the port is invalid.
1387 .BI (net,netsplice)listen
1388 For TCP network connections, tell fio to listen for incoming
1389 connections rather than initiating an outgoing connection. The
1390 hostname must be omitted if this option is used.
1392 .BI (net, pingpong) \fR=\fPbool
1393 Normally a network writer will just continue writing data, and a network reader
1394 will just consume packages. If pingpong=1 is set, a writer will send its normal
1395 payload to the reader, then wait for the reader to send the same payload back.
1396 This allows fio to measure network latencies. The submission and completion
1397 latencies then measure local time spent sending or receiving, and the
1398 completion latency measures how long it took for the other end to receive and
1399 send back. For UDP multicast traffic pingpong=1 should only be set for a single
1400 reader when multiple readers are listening to the same address.
1402 .BI (e4defrag,donorname) \fR=\fPstr
1403 File will be used as a block donor (swap extents between files)
1405 .BI (e4defrag,inplace) \fR=\fPint
1406 Configure donor file block allocation strategy
1409 Preallocate donor's file on init
1412 allocate space immediately inside defragment event, and free right after event
1415 .BI (rbd)rbdname \fR=\fPstr
1416 Specifies the name of the RBD.
1418 .BI (rbd)pool \fR=\fPstr
1419 Specifies the name of the Ceph pool containing the RBD.
1421 .BI (rbd)clientname \fR=\fPstr
1422 Specifies the username (without the 'client.' prefix) used to access the Ceph cluster.
1424 While running, \fBfio\fR will display the status of the created jobs. For
1428 Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
1431 The characters in the first set of brackets denote the current status of each
1432 threads. The possible values are:
1438 Setup but not started.
1444 Initialized, waiting.
1447 Running, doing sequential reads.
1450 Running, doing random reads.
1453 Running, doing sequential writes.
1456 Running, doing random writes.
1459 Running, doing mixed sequential reads/writes.
1462 Running, doing mixed random reads/writes.
1465 Running, currently waiting for \fBfsync\fR\|(2).
1468 Running, verifying written data.
1471 Exited, not reaped by main thread.
1474 Exited, thread reaped.
1478 The second set of brackets shows the estimated completion percentage of
1479 the current group. The third set shows the read and write I/O rate,
1480 respectively. Finally, the estimated run time of the job is displayed.
1482 When \fBfio\fR completes (or is interrupted by Ctrl-C), it will show data
1483 for each thread, each group of threads, and each disk, in that order.
1485 Per-thread statistics first show the threads client number, group-id, and
1486 error code. The remaining figures are as follows:
1490 Number of megabytes of I/O performed.
1493 Average data rate (bandwidth).
1499 Submission latency minimum, maximum, average and standard deviation. This is
1500 the time it took to submit the I/O.
1503 Completion latency minimum, maximum, average and standard deviation. This
1504 is the time between submission and completion.
1507 Bandwidth minimum, maximum, percentage of aggregate bandwidth received, average
1508 and standard deviation.
1511 CPU usage statistics. Includes user and system time, number of context switches
1512 this thread went through and number of major and minor page faults.
1515 Distribution of I/O depths. Each depth includes everything less than (or equal)
1516 to it, but greater than the previous depth.
1519 Number of read/write requests issued, and number of short read/write requests.
1522 Distribution of I/O completion latencies. The numbers follow the same pattern
1526 The group statistics show:
1531 Number of megabytes I/O performed.
1534 Aggregate bandwidth of threads in the group.
1537 Minimum average bandwidth a thread saw.
1540 Maximum average bandwidth a thread saw.
1543 Shortest runtime of threads in the group.
1546 Longest runtime of threads in the group.
1550 Finally, disk statistics are printed with reads first:
1555 Number of I/Os performed by all groups.
1558 Number of merges in the I/O scheduler.
1561 Number of ticks we kept the disk busy.
1564 Total time spent in the disk queue.
1571 It is also possible to get fio to dump the current output while it is
1572 running, without terminating the job. To do that, send fio the \fBUSR1\fR
1575 If the \fB\-\-minimal\fR option is given, the results will be printed in a
1576 semicolon-delimited format suitable for scripted use - a job description
1577 (if provided) follows on a new line. Note that the first
1578 number in the line is the version number. If the output has to be changed
1579 for some reason, this number will be incremented by 1 to signify that
1580 change. The fields are:
1583 .B terse version, fio version, jobname, groupid, error
1587 .B Total I/O \fR(KB)\fP, bandwidth \fR(KB/s)\fP, IOPS, runtime \fR(ms)\fP
1591 .B min, max, mean, standard deviation
1595 .B min, max, mean, standard deviation
1597 Completion latency percentiles (20 fields):
1599 .B Xth percentile=usec
1603 .B min, max, mean, standard deviation
1607 .B min, max, aggregate percentage of total, mean, standard deviation
1613 .B Total I/O \fR(KB)\fP, bandwidth \fR(KB/s)\fP, IOPS, runtime \fR(ms)\fP
1617 .B min, max, mean, standard deviation
1621 .B min, max, mean, standard deviation
1623 Completion latency percentiles (20 fields):
1625 .B Xth percentile=usec
1629 .B min, max, mean, standard deviation
1633 .B min, max, aggregate percentage of total, mean, standard deviation
1639 .B user, system, context switches, major page faults, minor page faults
1642 IO depth distribution:
1644 .B <=1, 2, 4, 8, 16, 32, >=64
1647 IO latency distribution:
1651 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
1655 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
1659 Disk utilization (1 for each disk used):
1661 .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
1664 Error Info (dependent on continue_on_error, default off):
1666 .B total # errors, first error code
1669 .B text description (if provided in config - appears on newline)
1672 Normally you would run fio as a stand-alone application on the machine
1673 where the IO workload should be generated. However, it is also possible to
1674 run the frontend and backend of fio separately. This makes it possible to
1675 have a fio server running on the machine(s) where the IO workload should
1676 be running, while controlling it from another machine.
1678 To start the server, you would do:
1680 \fBfio \-\-server=args\fR
1682 on that machine, where args defines what fio listens to. The arguments
1683 are of the form 'type:hostname or IP:port'. 'type' is either 'ip' (or ip4)
1684 for TCP/IP v4, 'ip6' for TCP/IP v6, or 'sock' for a local unix domain
1685 socket. 'hostname' is either a hostname or IP address, and 'port' is the port to
1686 listen to (only valid for TCP/IP, not a local socket). Some examples:
1690 Start a fio server, listening on all interfaces on the default port (8765).
1692 2) fio \-\-server=ip:hostname,4444
1694 Start a fio server, listening on IP belonging to hostname and on port 4444.
1696 3) fio \-\-server=ip6:::1,4444
1698 Start a fio server, listening on IPv6 localhost ::1 and on port 4444.
1700 4) fio \-\-server=,4444
1702 Start a fio server, listening on all interfaces on port 4444.
1704 5) fio \-\-server=1.2.3.4
1706 Start a fio server, listening on IP 1.2.3.4 on the default port.
1708 6) fio \-\-server=sock:/tmp/fio.sock
1710 Start a fio server, listening on the local socket /tmp/fio.sock.
1712 When a server is running, you can connect to it from a client. The client
1715 fio \-\-local-args \-\-client=server \-\-remote-args <job file(s)>
1717 where \-\-local-args are arguments that are local to the client where it is
1718 running, 'server' is the connect string, and \-\-remote-args and <job file(s)>
1719 are sent to the server. The 'server' string follows the same format as it
1720 does on the server side, to allow IP/hostname/socket and port strings.
1721 You can connect to multiple clients as well, to do that you could run:
1723 fio \-\-client=server2 \-\-client=server2 <job file(s)>
1727 was written by Jens Axboe <jens.axboe@oracle.com>,
1728 now Jens Axboe <jaxboe@fusionio.com>.
1730 This man page was written by Aaron Carroll <aaronc@cse.unsw.edu.au> based
1731 on documentation by Jens Axboe.
1732 .SH "REPORTING BUGS"
1733 Report bugs to the \fBfio\fR mailing list <fio@vger.kernel.org>.
1736 For further documentation see \fBHOWTO\fR and \fBREADME\fR.
1738 Sample jobfiles are available in the \fBexamples\fR directory.