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. The suffix is not case sensitive. If prefixed with '0x', the
125 value is assumed to be base 16 (hexadecimal). A suffix may include a trailing 'b',
126 for instance 'kb' is identical to 'k'. You can specify a base 10 value
127 by using 'KiB', 'MiB', 'GiB', etc. This is useful for disk drives where
128 values are often given in base 10 values. Specifying '30GiB' will get you
132 Boolean: a true or false value. `0' denotes false, `1' denotes true.
135 Integer range: a range of integers specified in the format
136 \fIlower\fR:\fIupper\fR or \fIlower\fR\-\fIupper\fR. \fIlower\fR and
137 \fIupper\fR may contain a suffix as described above. If an option allows two
138 sets of ranges, they are separated with a `,' or `/' character. For example:
142 List of floating numbers: A list of floating numbers, separated by
147 May be used to override the job name. On the command line, this parameter
148 has the special purpose of signalling the start of a new job.
150 .BI description \fR=\fPstr
151 Human-readable description of the job. It is printed when the job is run, but
152 otherwise has no special purpose.
154 .BI directory \fR=\fPstr
155 Prefix filenames with this directory. Used to place files in a location other
158 .BI filename \fR=\fPstr
160 normally makes up a file name based on the job name, thread number, and file
161 number. If you want to share files between threads in a job or several jobs,
162 specify a \fIfilename\fR for each of them to override the default.
163 If the I/O engine is file-based, you can specify
164 a number of files by separating the names with a `:' character. `\-' is a
165 reserved name, meaning stdin or stdout, depending on the read/write direction
168 .BI filename_format \fR=\fPstr
169 If sharing multiple files between jobs, it is usually necessary to have
170 fio generate the exact names that you want. By default, fio will name a file
171 based on the default file format specification of
172 \fBjobname.jobnumber.filenumber\fP. With this option, that can be
173 customized. Fio will recognize and replace the following keywords in this
179 The name of the worker thread or process.
182 The incremental number of the worker thread or process.
185 The incremental number of the file for that worker thread or process.
188 To have dependent jobs share a set of files, this option can be set to
189 have fio generate filenames that are shared between the two. For instance,
190 if \fBtestfiles.$filenum\fR is specified, file number 4 for any job will
191 be named \fBtestfiles.4\fR. The default of \fB$jobname.$jobnum.$filenum\fR
192 will be used if no other format specifier is given.
196 .BI lockfile \fR=\fPstr
197 Fio defaults to not locking any files before it does IO to them. If a file or
198 file descriptor is shared, fio can serialize IO to that file to make the end
199 result consistent. This is usual for emulating real workloads that share files.
205 No locking. This is the default.
208 Only one thread or process may do IO at the time, excluding all others.
211 Read-write locking on the file. Many readers may access the file at the same
212 time, but writes get exclusive access.
216 .BI opendir \fR=\fPstr
217 Recursively open any files below directory \fIstr\fR.
219 .BI readwrite \fR=\fPstr "\fR,\fP rw" \fR=\fPstr
220 Type of I/O pattern. Accepted values are:
231 Sequential trim (Linux block devices only).
240 Random trim (Linux block devices only).
243 Mixed sequential reads and writes.
246 Mixed random reads and writes.
249 For mixed I/O, the default split is 50/50. For certain types of io the result
250 may still be skewed a bit, since the speed may be different. It is possible to
251 specify a number of IO's to do before getting a new offset, this is done by
252 appending a `:\fI<nr>\fR to the end of the string given. For a random read, it
253 would look like \fBrw=randread:8\fR for passing in an offset modifier with a
254 value of 8. If the postfix is used with a sequential IO pattern, then the value
255 specified will be added to the generated offset for each IO. For instance,
256 using \fBrw=write:4k\fR will skip 4k for every write. It turns sequential IO
257 into sequential IO with holes. See the \fBrw_sequencer\fR option.
260 .BI rw_sequencer \fR=\fPstr
261 If an offset modifier is given by appending a number to the \fBrw=<str>\fR line,
262 then this option controls how that number modifies the IO offset being
263 generated. Accepted values are:
268 Generate sequential offset
271 Generate the same offset
274 \fBsequential\fR is only useful for random IO, where fio would normally
275 generate a new random offset for every IO. If you append eg 8 to randread, you
276 would get a new random offset for every 8 IO's. The result would be a seek for
277 only every 8 IO's, instead of for every IO. Use \fBrw=randread:8\fR to specify
278 that. As sequential IO is already sequential, setting \fBsequential\fR for that
279 would not result in any differences. \fBidentical\fR behaves in a similar
280 fashion, except it sends the same offset 8 number of times before generating a
285 .BI kb_base \fR=\fPint
286 The base unit for a kilobyte. The defacto base is 2^10, 1024. Storage
287 manufacturers like to use 10^3 or 1000 as a base ten unit instead, for obvious
288 reasons. Allowed values are 1024 or 1000, with 1024 being the default.
290 .BI unified_rw_reporting \fR=\fPbool
291 Fio normally reports statistics on a per data direction basis, meaning that
292 read, write, and trim are accounted and reported separately. If this option is
293 set, the fio will sum the results and report them as "mixed" instead.
295 .BI randrepeat \fR=\fPbool
296 Seed the random number generator in a predictable way so results are repeatable
297 across runs. Default: true.
299 .BI randseed \fR=\fPint
300 Seed the random number generators based on this seed value, to be able to
301 control what sequence of output is being generated. If not set, the random
302 sequence depends on the \fBrandrepeat\fR setting.
304 .BI use_os_rand \fR=\fPbool
305 Fio can either use the random generator supplied by the OS to generator random
306 offsets, or it can use it's own internal generator (based on Tausworthe).
307 Default is to use the internal generator, which is often of better quality and
308 faster. Default: false.
310 .BI fallocate \fR=\fPstr
311 Whether pre-allocation is performed when laying down files. Accepted values
317 Do not pre-allocate space.
320 Pre-allocate via \fBposix_fallocate\fR\|(3).
323 Pre-allocate via \fBfallocate\fR\|(2) with FALLOC_FL_KEEP_SIZE set.
326 Backward-compatible alias for 'none'.
329 Backward-compatible alias for 'posix'.
332 May not be available on all supported platforms. 'keep' is only
333 available on Linux. If using ZFS on Solaris this must be set to 'none'
334 because ZFS doesn't support it. Default: 'posix'.
337 .BI fadvise_hint \fR=\fPbool
338 Use of \fBposix_fadvise\fR\|(2) to advise the kernel what I/O patterns
339 are likely to be issued. Default: true.
342 Total size of I/O for this job. \fBfio\fR will run until this many bytes have
343 been transferred, unless limited by other options (\fBruntime\fR, for instance).
344 Unless \fBnrfiles\fR and \fBfilesize\fR options are given, this amount will be
345 divided between the available files for the job. If not set, fio will use the
346 full size of the given files or devices. If the files do not exist, size
347 must be given. It is also possible to give size as a percentage between 1 and
348 100. If size=20% is given, fio will use 20% of the full size of the given files
351 .BI fill_device \fR=\fPbool "\fR,\fB fill_fs" \fR=\fPbool
352 Sets size to something really large and waits for ENOSPC (no space left on
353 device) as the terminating condition. Only makes sense with sequential write.
354 For a read workload, the mount point will be filled first then IO started on
355 the result. This option doesn't make sense if operating on a raw device node,
356 since the size of that is already known by the file system. Additionally,
357 writing beyond end-of-device will not return ENOSPC there.
359 .BI filesize \fR=\fPirange
360 Individual file sizes. May be a range, in which case \fBfio\fR will select sizes
361 for files at random within the given range, limited to \fBsize\fR in total (if
362 that is given). If \fBfilesize\fR is not specified, each created file is the
365 .BI blocksize \fR=\fPint[,int] "\fR,\fB bs" \fR=\fPint[,int]
366 Block size for I/O units. Default: 4k. Values for reads, writes, and trims
367 can be specified separately in the format \fIread\fR,\fIwrite\fR,\fItrim\fR
368 either of which may be empty to leave that value at its default. If a trailing
369 comma isn't given, the remainder will inherit the last value set.
371 .BI blocksize_range \fR=\fPirange[,irange] "\fR,\fB bsrange" \fR=\fPirange[,irange]
372 Specify a range of I/O block sizes. The issued I/O unit will always be a
373 multiple of the minimum size, unless \fBblocksize_unaligned\fR is set. Applies
374 to both reads and writes if only one range is given, but can be specified
375 separately with a comma separating the values. Example: bsrange=1k-4k,2k-8k.
376 Also (see \fBblocksize\fR).
378 .BI bssplit \fR=\fPstr
379 This option allows even finer grained control of the block sizes issued,
380 not just even splits between them. With this option, you can weight various
381 block sizes for exact control of the issued IO for a job that has mixed
382 block sizes. The format of the option is bssplit=blocksize/percentage,
383 optionally adding as many definitions as needed separated by a colon.
384 Example: bssplit=4k/10:64k/50:32k/40 would issue 50% 64k blocks, 10% 4k
385 blocks and 40% 32k blocks. \fBbssplit\fR also supports giving separate
386 splits to reads and writes. The format is identical to what the
387 \fBbs\fR option accepts, the read and write parts are separated with a
390 .B blocksize_unaligned\fR,\fP bs_unaligned
391 If set, any size in \fBblocksize_range\fR may be used. This typically won't
392 work with direct I/O, as that normally requires sector alignment.
394 .BI blockalign \fR=\fPint[,int] "\fR,\fB ba" \fR=\fPint[,int]
395 At what boundary to align random IO offsets. Defaults to the same as 'blocksize'
396 the minimum blocksize given. Minimum alignment is typically 512b
397 for using direct IO, though it usually depends on the hardware block size.
398 This option is mutually exclusive with using a random map for files, so it
399 will turn off that option.
401 .BI bs_is_seq_rand \fR=\fPbool
402 If this option is set, fio will use the normal read,write blocksize settings as
403 sequential,random instead. Any random read or write will use the WRITE
404 blocksize settings, and any sequential read or write will use the READ
408 Initialise buffers with all zeros. Default: fill buffers with random data.
411 If this option is given, fio will refill the IO buffers on every submit. The
412 default is to only fill it at init time and reuse that data. Only makes sense
413 if zero_buffers isn't specified, naturally. If data verification is enabled,
414 refill_buffers is also automatically enabled.
416 .BI scramble_buffers \fR=\fPbool
417 If \fBrefill_buffers\fR is too costly and the target is using data
418 deduplication, then setting this option will slightly modify the IO buffer
419 contents to defeat normal de-dupe attempts. This is not enough to defeat
420 more clever block compression attempts, but it will stop naive dedupe
421 of blocks. Default: true.
423 .BI buffer_compress_percentage \fR=\fPint
424 If this is set, then fio will attempt to provide IO buffer content (on WRITEs)
425 that compress to the specified level. Fio does this by providing a mix of
426 random data and zeroes. Note that this is per block size unit, for file/disk
427 wide compression level that matches this setting, you'll also want to set
428 \fBrefill_buffers\fR.
430 .BI buffer_compress_chunk \fR=\fPint
431 See \fBbuffer_compress_percentage\fR. This setting allows fio to manage how
432 big the ranges of random data and zeroed data is. Without this set, fio will
433 provide \fBbuffer_compress_percentage\fR of blocksize random data, followed by
434 the remaining zeroed. With this set to some chunk size smaller than the block
435 size, fio can alternate random and zeroed data throughout the IO buffer.
437 .BI buffer_pattern \fR=\fPstr
438 If set, fio will fill the io buffers with this pattern. If not set, the contents
439 of io buffers is defined by the other options related to buffer contents. The
440 setting can be any pattern of bytes, and can be prefixed with 0x for hex
443 .BI nrfiles \fR=\fPint
444 Number of files to use for this job. Default: 1.
446 .BI openfiles \fR=\fPint
447 Number of files to keep open at the same time. Default: \fBnrfiles\fR.
449 .BI file_service_type \fR=\fPstr
450 Defines how files to service are selected. The following types are defined:
455 Choose a file at random.
458 Round robin over open files (default).
461 Do each file in the set sequentially.
464 The number of I/Os to issue before switching a new file can be specified by
465 appending `:\fIint\fR' to the service type.
468 .BI ioengine \fR=\fPstr
469 Defines how the job issues I/O. The following types are defined:
474 Basic \fBread\fR\|(2) or \fBwrite\fR\|(2) I/O. \fBfseek\fR\|(2) is used to
475 position the I/O location.
478 Basic \fBpread\fR\|(2) or \fBpwrite\fR\|(2) I/O.
481 Basic \fBreadv\fR\|(2) or \fBwritev\fR\|(2) I/O. Will emulate queuing by
482 coalescing adjacent IOs into a single submission.
485 Basic \fBpreadv\fR\|(2) or \fBpwritev\fR\|(2) I/O.
488 Linux native asynchronous I/O. This ioengine defines engine specific options.
491 POSIX asynchronous I/O using \fBaio_read\fR\|(3) and \fBaio_write\fR\|(3).
494 Solaris native asynchronous I/O.
497 Windows native asynchronous I/O.
500 File is memory mapped with \fBmmap\fR\|(2) and data copied using
504 \fBsplice\fR\|(2) is used to transfer the data and \fBvmsplice\fR\|(2) to
505 transfer data from user-space to the kernel.
508 Use the syslet system calls to make regular read/write asynchronous.
511 SCSI generic sg v3 I/O. May be either synchronous using the SG_IO ioctl, or if
512 the target is an sg character device, we use \fBread\fR\|(2) and
513 \fBwrite\fR\|(2) for asynchronous I/O.
516 Doesn't transfer any data, just pretends to. Mainly used to exercise \fBfio\fR
517 itself and for debugging and testing purposes.
520 Transfer over the network. The protocol to be used can be defined with the
521 \fBprotocol\fR parameter. Depending on the protocol, \fBfilename\fR,
522 \fBhostname\fR, \fBport\fR, or \fBlisten\fR must be specified.
523 This ioengine defines engine specific options.
526 Like \fBnet\fR, but uses \fBsplice\fR\|(2) and \fBvmsplice\fR\|(2) to map data
527 and send/receive. This ioengine defines engine specific options.
530 Doesn't transfer any data, but burns CPU cycles according to \fBcpuload\fR and
531 \fBcpucycles\fR parameters.
534 The GUASI I/O engine is the Generic Userspace Asynchronous Syscall Interface
535 approach to asynchronous I/O.
537 See <http://www.xmailserver.org/guasi\-lib.html>.
540 The RDMA I/O engine supports both RDMA memory semantics (RDMA_WRITE/RDMA_READ)
541 and channel semantics (Send/Recv) for the InfiniBand, RoCE and iWARP protocols.
544 Loads an external I/O engine object file. Append the engine filename as
548 IO engine that does regular linux native fallocate call to simulate data
549 transfer as fio ioengine
551 DDIR_READ does fallocate(,mode = FALLOC_FL_KEEP_SIZE,)
553 DIR_WRITE does fallocate(,mode = 0)
555 DDIR_TRIM does fallocate(,mode = FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE)
558 IO engine that does regular EXT4_IOC_MOVE_EXT ioctls to simulate defragment activity
559 request to DDIR_WRITE event
562 IO engine supporting direct access to Ceph Rados Block Devices (RBD) via librbd
563 without the need to use the kernel rbd driver. This ioengine defines engine specific
569 .BI iodepth \fR=\fPint
570 Number of I/O units to keep in flight against the file. Note that increasing
571 iodepth beyond 1 will not affect synchronous ioengines (except for small
572 degress when verify_async is in use). Even async engines my impose OS
573 restrictions causing the desired depth not to be achieved. This may happen on
574 Linux when using libaio and not setting \fBdirect\fR=1, since buffered IO is
575 not async on that OS. Keep an eye on the IO depth distribution in the
576 fio output to verify that the achieved depth is as expected. Default: 1.
578 .BI iodepth_batch \fR=\fPint
579 Number of I/Os to submit at once. Default: \fBiodepth\fR.
581 .BI iodepth_batch_complete \fR=\fPint
582 This defines how many pieces of IO to retrieve at once. It defaults to 1 which
583 means that we'll ask for a minimum of 1 IO in the retrieval process from the
584 kernel. The IO retrieval will go on until we hit the limit set by
585 \fBiodepth_low\fR. If this variable is set to 0, then fio will always check for
586 completed events before queuing more IO. This helps reduce IO latency, at the
587 cost of more retrieval system calls.
589 .BI iodepth_low \fR=\fPint
590 Low watermark indicating when to start filling the queue again. Default:
593 .BI direct \fR=\fPbool
594 If true, use non-buffered I/O (usually O_DIRECT). Default: false.
596 .BI atomic \fR=\fPbool
597 If value is true, attempt to use atomic direct IO. Atomic writes are guaranteed
598 to be stable once acknowledged by the operating system. Only Linux supports
601 .BI buffered \fR=\fPbool
602 If true, use buffered I/O. This is the opposite of the \fBdirect\fR parameter.
605 .BI offset \fR=\fPint
606 Offset in the file to start I/O. Data before the offset will not be touched.
608 .BI offset_increment \fR=\fPint
609 If this is provided, then the real offset becomes the
610 offset + offset_increment * thread_number, where the thread number is a counter
611 that starts at 0 and is incremented for each job. This option is useful if
612 there are several jobs which are intended to operate on a file in parallel in
613 disjoint segments, with even spacing between the starting points.
615 .BI number_ios \fR=\fPint
616 Fio will normally perform IOs until it has exhausted the size of the region
617 set by \fBsize\fR, or if it exhaust the allocated time (or hits an error
618 condition). With this setting, the range/size can be set independently of
619 the number of IOs to perform. When fio reaches this number, it will exit
620 normally and report status.
623 How many I/Os to perform before issuing an \fBfsync\fR\|(2) of dirty data. If
624 0, don't sync. Default: 0.
626 .BI fdatasync \fR=\fPint
627 Like \fBfsync\fR, but uses \fBfdatasync\fR\|(2) instead to only sync the
628 data parts of the file. Default: 0.
630 .BI write_barrier \fR=\fPint
631 Make every Nth write a barrier write.
633 .BI sync_file_range \fR=\fPstr:int
634 Use \fBsync_file_range\fR\|(2) for every \fRval\fP number of write operations. Fio will
635 track range of writes that have happened since the last \fBsync_file_range\fR\|(2) call.
636 \fRstr\fP can currently be one or more of:
640 SYNC_FILE_RANGE_WAIT_BEFORE
643 SYNC_FILE_RANGE_WRITE
646 SYNC_FILE_RANGE_WRITE
650 So if you do sync_file_range=wait_before,write:8, fio would use
651 \fBSYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE\fP for every 8 writes.
652 Also see the \fBsync_file_range\fR\|(2) man page. This option is Linux specific.
654 .BI overwrite \fR=\fPbool
655 If writing, setup the file first and do overwrites. Default: false.
657 .BI end_fsync \fR=\fPbool
658 Sync file contents when a write stage has completed. Default: false.
660 .BI fsync_on_close \fR=\fPbool
661 If true, sync file contents on close. This differs from \fBend_fsync\fR in that
662 it will happen on every close, not just at the end of the job. Default: false.
664 .BI rwmixread \fR=\fPint
665 Percentage of a mixed workload that should be reads. Default: 50.
667 .BI rwmixwrite \fR=\fPint
668 Percentage of a mixed workload that should be writes. If \fBrwmixread\fR and
669 \fBrwmixwrite\fR are given and do not sum to 100%, the latter of the two
670 overrides the first. This may interfere with a given rate setting, if fio is
671 asked to limit reads or writes to a certain rate. If that is the case, then
672 the distribution may be skewed. Default: 50.
674 .BI random_distribution \fR=\fPstr:float
675 By default, fio will use a completely uniform random distribution when asked
676 to perform random IO. Sometimes it is useful to skew the distribution in
677 specific ways, ensuring that some parts of the data is more hot than others.
678 Fio includes the following distribution models:
682 Uniform random distribution
692 When using a zipf or pareto distribution, an input value is also needed to
693 define the access pattern. For zipf, this is the zipf theta. For pareto,
694 it's the pareto power. Fio includes a test program, genzipf, that can be
695 used visualize what the given input values will yield in terms of hit rates.
696 If you wanted to use zipf with a theta of 1.2, you would use
697 random_distribution=zipf:1.2 as the option. If a non-uniform model is used,
698 fio will disable use of the random map.
700 .BI percentage_random \fR=\fPint
701 For a random workload, set how big a percentage should be random. This defaults
702 to 100%, in which case the workload is fully random. It can be set from
703 anywhere from 0 to 100. Setting it to 0 would make the workload fully
704 sequential. It is possible to set different values for reads, writes, and
705 trim. To do so, simply use a comma separated list. See \fBblocksize\fR.
708 Normally \fBfio\fR will cover every block of the file when doing random I/O. If
709 this parameter is given, a new offset will be chosen without looking at past
710 I/O history. This parameter is mutually exclusive with \fBverify\fR.
712 .BI softrandommap \fR=\fPbool
713 See \fBnorandommap\fR. If fio runs with the random block map enabled and it
714 fails to allocate the map, if this option is set it will continue without a
715 random block map. As coverage will not be as complete as with random maps, this
716 option is disabled by default.
718 .BI random_generator \fR=\fPstr
719 Fio supports the following engines for generating IO offsets for random IO:
723 Strong 2^88 cycle random number generator
726 Linear feedback shift register generator
730 Tausworthe is a strong random number generator, but it requires tracking on the
731 side if we want to ensure that blocks are only read or written once. LFSR
732 guarantees that we never generate the same offset twice, and it's also less
733 computationally expensive. It's not a true random generator, however, though
734 for IO purposes it's typically good enough. LFSR only works with single block
735 sizes, not with workloads that use multiple block sizes. If used with such a
736 workload, fio may read or write some blocks multiple times.
739 Run job with given nice value. See \fBnice\fR\|(2).
742 Set I/O priority value of this job between 0 (highest) and 7 (lowest). See
745 .BI prioclass \fR=\fPint
746 Set I/O priority class. See \fBionice\fR\|(1).
748 .BI thinktime \fR=\fPint
749 Stall job for given number of microseconds between issuing I/Os.
751 .BI thinktime_spin \fR=\fPint
752 Pretend to spend CPU time for given number of microseconds, sleeping the rest
753 of the time specified by \fBthinktime\fR. Only valid if \fBthinktime\fR is set.
755 .BI thinktime_blocks \fR=\fPint
756 Only valid if thinktime is set - control how many blocks to issue, before
757 waiting \fBthinktime\fR microseconds. If not set, defaults to 1 which will
758 make fio wait \fBthinktime\fR microseconds after every block. This
759 effectively makes any queue depth setting redundant, since no more than 1 IO
760 will be queued before we have to complete it and do our thinktime. In other
761 words, this setting effectively caps the queue depth if the latter is larger.
765 Cap bandwidth used by this job. The number is in bytes/sec, the normal postfix
766 rules apply. You can use \fBrate\fR=500k to limit reads and writes to 500k each,
767 or you can specify read and writes separately. Using \fBrate\fR=1m,500k would
768 limit reads to 1MB/sec and writes to 500KB/sec. Capping only reads or writes
769 can be done with \fBrate\fR=,500k or \fBrate\fR=500k,. The former will only
770 limit writes (to 500KB/sec), the latter will only limit reads.
772 .BI ratemin \fR=\fPint
773 Tell \fBfio\fR to do whatever it can to maintain at least the given bandwidth.
774 Failing to meet this requirement will cause the job to exit. The same format
775 as \fBrate\fR is used for read vs write separation.
777 .BI rate_iops \fR=\fPint
778 Cap the bandwidth to this number of IOPS. Basically the same as rate, just
779 specified independently of bandwidth. The same format as \fBrate\fR is used for
780 read vs write separation. If \fBblocksize\fR is a range, the smallest block
781 size is used as the metric.
783 .BI rate_iops_min \fR=\fPint
784 If this rate of I/O is not met, the job will exit. The same format as \fBrate\fR
785 is used for read vs write separation.
787 .BI ratecycle \fR=\fPint
788 Average bandwidth for \fBrate\fR and \fBratemin\fR over this number of
789 milliseconds. Default: 1000ms.
791 .BI latency_target \fR=\fPint
792 If set, fio will attempt to find the max performance point that the given
793 workload will run at while maintaining a latency below this target. The
794 values is given in microseconds. See \fBlatency_window\fR and
795 \fBlatency_percentile\fR.
797 .BI latency_window \fR=\fPint
798 Used with \fBlatency_target\fR to specify the sample window that the job
799 is run at varying queue depths to test the performance. The value is given
802 .BI latency_percentile \fR=\fPfloat
803 The percentage of IOs that must fall within the criteria specified by
804 \fBlatency_target\fR and \fBlatency_window\fR. If not set, this defaults
805 to 100.0, meaning that all IOs must be equal or below to the value set
806 by \fBlatency_target\fR.
808 .BI max_latency \fR=\fPint
809 If set, fio will exit the job if it exceeds this maximum latency. It will exit
812 .BI cpumask \fR=\fPint
813 Set CPU affinity for this job. \fIint\fR is a bitmask of allowed CPUs the job
814 may run on. See \fBsched_setaffinity\fR\|(2).
816 .BI cpus_allowed \fR=\fPstr
817 Same as \fBcpumask\fR, but allows a comma-delimited list of CPU numbers.
819 .BI numa_cpu_nodes \fR=\fPstr
820 Set this job running on specified NUMA nodes' CPUs. The arguments allow
821 comma delimited list of cpu numbers, A-B ranges, or 'all'.
823 .BI numa_mem_policy \fR=\fPstr
824 Set this job's memory policy and corresponding NUMA nodes. Format of
828 .B <mode>[:<nodelist>]
831 is one of the following memory policy:
833 .B default, prefer, bind, interleave, local
836 For \fBdefault\fR and \fBlocal\fR memory policy, no \fBnodelist\fR is
837 needed to be specified. For \fBprefer\fR, only one node is
838 allowed. For \fBbind\fR and \fBinterleave\fR, \fBnodelist\fR allows
839 comma delimited list of numbers, A-B ranges, or 'all'.
841 .BI startdelay \fR=\fPint
842 Delay start of job for the specified number of seconds.
844 .BI runtime \fR=\fPint
845 Terminate processing after the specified number of seconds.
848 If given, run for the specified \fBruntime\fR duration even if the files are
849 completely read or written. The same workload will be repeated as many times
850 as \fBruntime\fR allows.
852 .BI ramp_time \fR=\fPint
853 If set, fio will run the specified workload for this amount of time before
854 logging any performance numbers. Useful for letting performance settle before
855 logging results, thus minimizing the runtime required for stable results. Note
856 that the \fBramp_time\fR is considered lead in time for a job, thus it will
857 increase the total runtime if a special timeout or runtime is specified.
859 .BI invalidate \fR=\fPbool
860 Invalidate buffer-cache for the file prior to starting I/O. Default: true.
863 Use synchronous I/O for buffered writes. For the majority of I/O engines,
864 this means using O_SYNC. Default: false.
866 .BI iomem \fR=\fPstr "\fR,\fP mem" \fR=\fPstr
867 Allocation method for I/O unit buffer. Allowed values are:
872 Allocate memory with \fBmalloc\fR\|(3).
875 Use shared memory buffers allocated through \fBshmget\fR\|(2).
878 Same as \fBshm\fR, but use huge pages as backing.
881 Use \fBmmap\fR\|(2) for allocation. Uses anonymous memory unless a filename
882 is given after the option in the format `:\fIfile\fR'.
885 Same as \fBmmap\fR, but use huge files as backing.
888 The amount of memory allocated is the maximum allowed \fBblocksize\fR for the
889 job multiplied by \fBiodepth\fR. For \fBshmhuge\fR or \fBmmaphuge\fR to work,
890 the system must have free huge pages allocated. \fBmmaphuge\fR also needs to
891 have hugetlbfs mounted, and \fIfile\fR must point there. At least on Linux,
892 huge pages must be manually allocated. See \fB/proc/sys/vm/nr_hugehages\fR
893 and the documentation for that. Normally you just need to echo an appropriate
894 number, eg echoing 8 will ensure that the OS has 8 huge pages ready for
898 .BI iomem_align \fR=\fPint "\fR,\fP mem_align" \fR=\fPint
899 This indicates the memory alignment of the IO memory buffers. Note that the
900 given alignment is applied to the first IO unit buffer, if using \fBiodepth\fR
901 the alignment of the following buffers are given by the \fBbs\fR used. In
902 other words, if using a \fBbs\fR that is a multiple of the page sized in the
903 system, all buffers will be aligned to this value. If using a \fBbs\fR that
904 is not page aligned, the alignment of subsequent IO memory buffers is the
905 sum of the \fBiomem_align\fR and \fBbs\fR used.
907 .BI hugepage\-size \fR=\fPint
908 Defines the size of a huge page. Must be at least equal to the system setting.
909 Should be a multiple of 1MB. Default: 4MB.
912 Terminate all jobs when one finishes. Default: wait for each job to finish.
914 .BI bwavgtime \fR=\fPint
915 Average bandwidth calculations over the given time in milliseconds. Default:
918 .BI iopsavgtime \fR=\fPint
919 Average IOPS calculations over the given time in milliseconds. Default:
922 .BI create_serialize \fR=\fPbool
923 If true, serialize file creation for the jobs. Default: true.
925 .BI create_fsync \fR=\fPbool
926 \fBfsync\fR\|(2) data file after creation. Default: true.
928 .BI create_on_open \fR=\fPbool
929 If true, the files are not created until they are opened for IO by the job.
931 .BI create_only \fR=\fPbool
932 If true, fio will only run the setup phase of the job. If files need to be
933 laid out or updated on disk, only that will be done. The actual job contents
936 .BI pre_read \fR=\fPbool
937 If this is given, files will be pre-read into memory before starting the given
938 IO operation. This will also clear the \fR \fBinvalidate\fR flag, since it is
939 pointless to pre-read and then drop the cache. This will only work for IO
940 engines that are seekable, since they allow you to read the same data
941 multiple times. Thus it will not work on eg network or splice IO.
943 .BI unlink \fR=\fPbool
944 Unlink job files when done. Default: false.
947 Specifies the number of iterations (runs of the same workload) of this job.
950 .BI verify_only \fR=\fPbool
951 Do not perform the specified workload, only verify data still matches previous
952 invocation of this workload. This option allows one to check data multiple
953 times at a later date without overwriting it. This option makes sense only for
954 workloads that write data, and does not support workloads with the
955 \fBtime_based\fR option set.
957 .BI do_verify \fR=\fPbool
958 Run the verify phase after a write phase. Only valid if \fBverify\fR is set.
961 .BI verify \fR=\fPstr
962 Method of verifying file contents after each iteration of the job. Allowed
967 .B md5 crc16 crc32 crc32c crc32c-intel crc64 crc7 sha256 sha512 sha1
968 Store appropriate checksum in the header of each block. crc32c-intel is
969 hardware accelerated SSE4.2 driven, falls back to regular crc32c if
970 not supported by the system.
973 Write extra information about each I/O (timestamp, block number, etc.). The
974 block number is verified. See \fBverify_pattern\fR as well.
977 Pretend to verify. Used for testing internals.
980 This option can be used for repeated burn-in tests of a system to make sure
981 that the written data is also correctly read back. If the data direction given
982 is a read or random read, fio will assume that it should verify a previously
983 written file. If the data direction includes any form of write, the verify will
984 be of the newly written data.
987 .BI verifysort \fR=\fPbool
988 If true, written verify blocks are sorted if \fBfio\fR deems it to be faster to
989 read them back in a sorted manner. Default: true.
991 .BI verifysort_nr \fR=\fPint
992 Pre-load and sort verify blocks for a read workload.
994 .BI verify_offset \fR=\fPint
995 Swap the verification header with data somewhere else in the block before
996 writing. It is swapped back before verifying.
998 .BI verify_interval \fR=\fPint
999 Write the verification header for this number of bytes, which should divide
1000 \fBblocksize\fR. Default: \fBblocksize\fR.
1002 .BI verify_pattern \fR=\fPstr
1003 If set, fio will fill the io buffers with this pattern. Fio defaults to filling
1004 with totally random bytes, but sometimes it's interesting to fill with a known
1005 pattern for io verification purposes. Depending on the width of the pattern,
1006 fio will fill 1/2/3/4 bytes of the buffer at the time(it can be either a
1007 decimal or a hex number). The verify_pattern if larger than a 32-bit quantity
1008 has to be a hex number that starts with either "0x" or "0X". Use with
1011 .BI verify_fatal \fR=\fPbool
1012 If true, exit the job on the first observed verification failure. Default:
1015 .BI verify_dump \fR=\fPbool
1016 If set, dump the contents of both the original data block and the data block we
1017 read off disk to files. This allows later analysis to inspect just what kind of
1018 data corruption occurred. Off by default.
1020 .BI verify_async \fR=\fPint
1021 Fio will normally verify IO inline from the submitting thread. This option
1022 takes an integer describing how many async offload threads to create for IO
1023 verification instead, causing fio to offload the duty of verifying IO contents
1024 to one or more separate threads. If using this offload option, even sync IO
1025 engines can benefit from using an \fBiodepth\fR setting higher than 1, as it
1026 allows them to have IO in flight while verifies are running.
1028 .BI verify_async_cpus \fR=\fPstr
1029 Tell fio to set the given CPU affinity on the async IO verification threads.
1030 See \fBcpus_allowed\fP for the format used.
1032 .BI verify_backlog \fR=\fPint
1033 Fio will normally verify the written contents of a job that utilizes verify
1034 once that job has completed. In other words, everything is written then
1035 everything is read back and verified. You may want to verify continually
1036 instead for a variety of reasons. Fio stores the meta data associated with an
1037 IO block in memory, so for large verify workloads, quite a bit of memory would
1038 be used up holding this meta data. If this option is enabled, fio will write
1039 only N blocks before verifying these blocks.
1041 .BI verify_backlog_batch \fR=\fPint
1042 Control how many blocks fio will verify if verify_backlog is set. If not set,
1043 will default to the value of \fBverify_backlog\fR (meaning the entire queue is
1044 read back and verified). If \fBverify_backlog_batch\fR is less than
1045 \fBverify_backlog\fR then not all blocks will be verified, if
1046 \fBverify_backlog_batch\fR is larger than \fBverify_backlog\fR, some blocks
1047 will be verified more than once.
1049 .BI trim_percentage \fR=\fPint
1050 Number of verify blocks to discard/trim.
1052 .BI trim_verify_zero \fR=\fPbool
1053 Verify that trim/discarded blocks are returned as zeroes.
1055 .BI trim_backlog \fR=\fPint
1056 Trim after this number of blocks are written.
1058 .BI trim_backlog_batch \fR=\fPint
1059 Trim this number of IO blocks.
1061 .BI experimental_verify \fR=\fPbool
1062 Enable experimental verification.
1064 .B stonewall "\fR,\fP wait_for_previous"
1065 Wait for preceding jobs in the job file to exit before starting this one.
1066 \fBstonewall\fR implies \fBnew_group\fR.
1069 Start a new reporting group. If not given, all jobs in a file will be part
1070 of the same reporting group, unless separated by a stonewall.
1072 .BI numjobs \fR=\fPint
1073 Number of clones (processes/threads performing the same workload) of this job.
1077 If set, display per-group reports instead of per-job when \fBnumjobs\fR is
1081 Use threads created with \fBpthread_create\fR\|(3) instead of processes created
1082 with \fBfork\fR\|(2).
1084 .BI zonesize \fR=\fPint
1085 Divide file into zones of the specified size in bytes. See \fBzoneskip\fR.
1087 .BI zonerange \fR=\fPint
1088 Give size of an IO zone. See \fBzoneskip\fR.
1090 .BI zoneskip \fR=\fPint
1091 Skip the specified number of bytes when \fBzonesize\fR bytes of data have been
1094 .BI write_iolog \fR=\fPstr
1095 Write the issued I/O patterns to the specified file. Specify a separate file
1096 for each job, otherwise the iologs will be interspersed and the file may be
1099 .BI read_iolog \fR=\fPstr
1100 Replay the I/O patterns contained in the specified file generated by
1101 \fBwrite_iolog\fR, or may be a \fBblktrace\fR binary file.
1103 .BI replay_no_stall \fR=\fPint
1104 While replaying I/O patterns using \fBread_iolog\fR the default behavior
1105 attempts to respect timing information between I/Os. Enabling
1106 \fBreplay_no_stall\fR causes I/Os to be replayed as fast as possible while
1107 still respecting ordering.
1109 .BI replay_redirect \fR=\fPstr
1110 While replaying I/O patterns using \fBread_iolog\fR the default behavior
1111 is to replay the IOPS onto the major/minor device that each IOP was recorded
1112 from. Setting \fBreplay_redirect\fR causes all IOPS to be replayed onto the
1113 single specified device regardless of the device it was recorded from.
1115 .BI write_bw_log \fR=\fPstr
1116 If given, write a bandwidth log of the jobs in this job file. Can be used to
1117 store data of the bandwidth of the jobs in their lifetime. The included
1118 fio_generate_plots script uses gnuplot to turn these text files into nice
1119 graphs. See \fBwrite_lat_log\fR for behaviour of given filename. For this
1120 option, the postfix is _bw.log.
1122 .BI write_lat_log \fR=\fPstr
1123 Same as \fBwrite_bw_log\fR, but writes I/O completion latencies. If no
1124 filename is given with this option, the default filename of "jobname_type.log"
1125 is used. Even if the filename is given, fio will still append the type of log.
1127 .BI write_iops_log \fR=\fPstr
1128 Same as \fBwrite_bw_log\fR, but writes IOPS. If no filename is given with this
1129 option, the default filename of "jobname_type.log" is used. Even if the
1130 filename is given, fio will still append the type of log.
1132 .BI log_avg_msec \fR=\fPint
1133 By default, fio will log an entry in the iops, latency, or bw log for every
1134 IO that completes. When writing to the disk log, that can quickly grow to a
1135 very large size. Setting this option makes fio average the each log entry
1136 over the specified period of time, reducing the resolution of the log.
1139 .BI disable_lat \fR=\fPbool
1140 Disable measurements of total latency numbers. Useful only for cutting
1141 back the number of calls to \fBgettimeofday\fR\|(2), as that does impact performance at
1142 really high IOPS rates. Note that to really get rid of a large amount of these
1143 calls, this option must be used with disable_slat and disable_bw as well.
1145 .BI disable_clat \fR=\fPbool
1146 Disable measurements of completion latency numbers. See \fBdisable_lat\fR.
1148 .BI disable_slat \fR=\fPbool
1149 Disable measurements of submission latency numbers. See \fBdisable_lat\fR.
1151 .BI disable_bw_measurement \fR=\fPbool
1152 Disable measurements of throughput/bandwidth numbers. See \fBdisable_lat\fR.
1154 .BI lockmem \fR=\fPint
1155 Pin the specified amount of memory with \fBmlock\fR\|(2). Can be used to
1156 simulate a smaller amount of memory. The amount specified is per worker.
1158 .BI exec_prerun \fR=\fPstr
1159 Before running the job, execute the specified command with \fBsystem\fR\|(3).
1161 Output is redirected in a file called \fBjobname.prerun.txt\fR
1164 .BI exec_postrun \fR=\fPstr
1165 Same as \fBexec_prerun\fR, but the command is executed after the job completes.
1167 Output is redirected in a file called \fBjobname.postrun.txt\fR
1170 .BI ioscheduler \fR=\fPstr
1171 Attempt to switch the device hosting the file to the specified I/O scheduler.
1173 .BI cpuload \fR=\fPint
1174 If the job is a CPU cycle-eater, attempt to use the specified percentage of
1177 .BI cpuchunks \fR=\fPint
1178 If the job is a CPU cycle-eater, split the load into cycles of the
1179 given time in milliseconds.
1181 .BI disk_util \fR=\fPbool
1182 Generate disk utilization statistics if the platform supports it. Default: true.
1184 .BI clocksource \fR=\fPstr
1185 Use the given clocksource as the base of timing. The supported options are:
1189 \fBgettimeofday\fR\|(2)
1192 \fBclock_gettime\fR\|(2)
1195 Internal CPU clock source
1199 \fBcpu\fR is the preferred clocksource if it is reliable, as it is very fast
1200 (and fio is heavy on time calls). Fio will automatically use this clocksource
1201 if it's supported and considered reliable on the system it is running on,
1202 unless another clocksource is specifically set. For x86/x86-64 CPUs, this
1203 means supporting TSC Invariant.
1205 .BI gtod_reduce \fR=\fPbool
1206 Enable all of the \fBgettimeofday\fR\|(2) reducing options (disable_clat, disable_slat,
1207 disable_bw) plus reduce precision of the timeout somewhat to really shrink the
1208 \fBgettimeofday\fR\|(2) call count. With this option enabled, we only do about 0.4% of
1209 the gtod() calls we would have done if all time keeping was enabled.
1211 .BI gtod_cpu \fR=\fPint
1212 Sometimes it's cheaper to dedicate a single thread of execution to just getting
1213 the current time. Fio (and databases, for instance) are very intensive on
1214 \fBgettimeofday\fR\|(2) calls. With this option, you can set one CPU aside for doing
1215 nothing but logging current time to a shared memory location. Then the other
1216 threads/processes that run IO workloads need only copy that segment, instead of
1217 entering the kernel with a \fBgettimeofday\fR\|(2) call. The CPU set aside for doing
1218 these time calls will be excluded from other uses. Fio will manually clear it
1219 from the CPU mask of other jobs.
1221 .BI ignore_error \fR=\fPstr
1222 Sometimes you want to ignore some errors during test in that case you can specify
1223 error list for each error type.
1225 ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST
1227 errors for given error type is separated with ':'.
1228 Error may be symbol ('ENOSPC', 'ENOMEM') or an integer.
1230 Example: ignore_error=EAGAIN,ENOSPC:122 .
1232 This option will ignore EAGAIN from READ, and ENOSPC and 122(EDQUOT) from WRITE.
1234 .BI error_dump \fR=\fPbool
1235 If set dump every error even if it is non fatal, true by default. If disabled
1236 only fatal error will be dumped
1238 .BI profile \fR=\fPstr
1239 Select a specific builtin performance test.
1241 .BI cgroup \fR=\fPstr
1242 Add job to this control group. If it doesn't exist, it will be created.
1243 The system must have a mounted cgroup blkio mount point for this to work. If
1244 your system doesn't have it mounted, you can do so with:
1246 # mount \-t cgroup \-o blkio none /cgroup
1248 .BI cgroup_weight \fR=\fPint
1249 Set the weight of the cgroup to this value. See the documentation that comes
1250 with the kernel, allowed values are in the range of 100..1000.
1252 .BI cgroup_nodelete \fR=\fPbool
1253 Normally fio will delete the cgroups it has created after the job completion.
1254 To override this behavior and to leave cgroups around after the job completion,
1255 set cgroup_nodelete=1. This can be useful if one wants to inspect various
1256 cgroup files after job completion. Default: false
1259 Instead of running as the invoking user, set the user ID to this value before
1260 the thread/process does any work.
1263 Set group ID, see \fBuid\fR.
1265 .BI unit_base \fR=\fPint
1266 Base unit for reporting. Allowed values are:
1270 Use auto-detection (default).
1280 .BI flow_id \fR=\fPint
1281 The ID of the flow. If not specified, it defaults to being a global flow. See
1285 Weight in token-based flow control. If this value is used, then there is a
1286 \fBflow counter\fR which is used to regulate the proportion of activity between
1287 two or more jobs. fio attempts to keep this flow counter near zero. The
1288 \fBflow\fR parameter stands for how much should be added or subtracted to the
1289 flow counter on each iteration of the main I/O loop. That is, if one job has
1290 \fBflow=8\fR and another job has \fBflow=-1\fR, then there will be a roughly
1291 1:8 ratio in how much one runs vs the other.
1293 .BI flow_watermark \fR=\fPint
1294 The maximum value that the absolute value of the flow counter is allowed to
1295 reach before the job must wait for a lower value of the counter.
1297 .BI flow_sleep \fR=\fPint
1298 The period of time, in microseconds, to wait after the flow watermark has been
1299 exceeded before retrying operations
1301 .BI clat_percentiles \fR=\fPbool
1302 Enable the reporting of percentiles of completion latencies.
1304 .BI percentile_list \fR=\fPfloat_list
1305 Overwrite the default list of percentiles for completion
1306 latencies. Each number is a floating number in the range (0,100], and
1307 the maximum length of the list is 20. Use ':' to separate the
1308 numbers. For example, \-\-percentile_list=99.5:99.9 will cause fio to
1309 report the values of completion latency below which 99.5% and 99.9% of
1310 the observed latencies fell, respectively.
1311 .SS "Ioengine Parameters List"
1312 Some parameters are only valid when a specific ioengine is in use. These are
1313 used identically to normal parameters, with the caveat that when used on the
1314 command line, the must come after the ioengine that defines them is selected.
1316 .BI (cpu)cpuload \fR=\fPint
1317 Attempt to use the specified percentage of CPU cycles.
1319 .BI (cpu)cpuchunks \fR=\fPint
1320 Split the load into cycles of the given time. In microseconds.
1322 .BI (libaio)userspace_reap
1323 Normally, with the libaio engine in use, fio will use
1324 the io_getevents system call to reap newly returned events.
1325 With this flag turned on, the AIO ring will be read directly
1326 from user-space to reap events. The reaping mode is only
1327 enabled when polling for a minimum of 0 events (eg when
1328 iodepth_batch_complete=0).
1330 .BI (net,netsplice)hostname \fR=\fPstr
1331 The host name or IP address to use for TCP or UDP based IO.
1332 If the job is a TCP listener or UDP reader, the hostname is not
1333 used and must be omitted unless it is a valid UDP multicast address.
1335 .BI (net,netsplice)port \fR=\fPint
1336 The TCP or UDP port to bind to or connect to.
1338 .BI (net,netsplice)interface \fR=\fPstr
1339 The IP address of the network interface used to send or receive UDP multicast
1342 .BI (net,netsplice)ttl \fR=\fPint
1343 Time-to-live value for outgoing UDP multicast packets. Default: 1
1345 .BI (net,netsplice)nodelay \fR=\fPbool
1346 Set TCP_NODELAY on TCP connections.
1348 .BI (net,netsplice)protocol \fR=\fPstr "\fR,\fP proto" \fR=\fPstr
1349 The network protocol to use. Accepted values are:
1354 Transmission control protocol
1357 Transmission control protocol V6
1360 User datagram protocol
1363 User datagram protocol V6
1369 When the protocol is TCP or UDP, the port must also be given,
1370 as well as the hostname if the job is a TCP listener or UDP
1371 reader. For unix sockets, the normal filename option should be
1372 used and the port is invalid.
1375 .BI (net,netsplice)listen
1376 For TCP network connections, tell fio to listen for incoming
1377 connections rather than initiating an outgoing connection. The
1378 hostname must be omitted if this option is used.
1380 .BI (net, pingpong) \fR=\fPbool
1381 Normally a network writer will just continue writing data, and a network reader
1382 will just consume packages. If pingpong=1 is set, a writer will send its normal
1383 payload to the reader, then wait for the reader to send the same payload back.
1384 This allows fio to measure network latencies. The submission and completion
1385 latencies then measure local time spent sending or receiving, and the
1386 completion latency measures how long it took for the other end to receive and
1387 send back. For UDP multicast traffic pingpong=1 should only be set for a single
1388 reader when multiple readers are listening to the same address.
1390 .BI (e4defrag,donorname) \fR=\fPstr
1391 File will be used as a block donor (swap extents between files)
1393 .BI (e4defrag,inplace) \fR=\fPint
1394 Configure donor file block allocation strategy
1397 Preallocate donor's file on init
1400 allocate space immediately inside defragment event, and free right after event
1403 .BI (rbd)rbdname \fR=\fPstr
1404 Specifies the name of the RBD.
1406 .BI (rbd)pool \fR=\fPstr
1407 Specifies the name of the Ceph pool containing the RBD.
1409 .BI (rbd)clientname \fR=\fPstr
1410 Specifies the username (without the 'client.' prefix) used to access the Ceph cluster.
1412 While running, \fBfio\fR will display the status of the created jobs. For
1416 Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
1419 The characters in the first set of brackets denote the current status of each
1420 threads. The possible values are:
1426 Setup but not started.
1432 Initialized, waiting.
1435 Running, doing sequential reads.
1438 Running, doing random reads.
1441 Running, doing sequential writes.
1444 Running, doing random writes.
1447 Running, doing mixed sequential reads/writes.
1450 Running, doing mixed random reads/writes.
1453 Running, currently waiting for \fBfsync\fR\|(2).
1456 Running, verifying written data.
1459 Exited, not reaped by main thread.
1462 Exited, thread reaped.
1466 The second set of brackets shows the estimated completion percentage of
1467 the current group. The third set shows the read and write I/O rate,
1468 respectively. Finally, the estimated run time of the job is displayed.
1470 When \fBfio\fR completes (or is interrupted by Ctrl-C), it will show data
1471 for each thread, each group of threads, and each disk, in that order.
1473 Per-thread statistics first show the threads client number, group-id, and
1474 error code. The remaining figures are as follows:
1478 Number of megabytes of I/O performed.
1481 Average data rate (bandwidth).
1487 Submission latency minimum, maximum, average and standard deviation. This is
1488 the time it took to submit the I/O.
1491 Completion latency minimum, maximum, average and standard deviation. This
1492 is the time between submission and completion.
1495 Bandwidth minimum, maximum, percentage of aggregate bandwidth received, average
1496 and standard deviation.
1499 CPU usage statistics. Includes user and system time, number of context switches
1500 this thread went through and number of major and minor page faults.
1503 Distribution of I/O depths. Each depth includes everything less than (or equal)
1504 to it, but greater than the previous depth.
1507 Number of read/write requests issued, and number of short read/write requests.
1510 Distribution of I/O completion latencies. The numbers follow the same pattern
1514 The group statistics show:
1519 Number of megabytes I/O performed.
1522 Aggregate bandwidth of threads in the group.
1525 Minimum average bandwidth a thread saw.
1528 Maximum average bandwidth a thread saw.
1531 Shortest runtime of threads in the group.
1534 Longest runtime of threads in the group.
1538 Finally, disk statistics are printed with reads first:
1543 Number of I/Os performed by all groups.
1546 Number of merges in the I/O scheduler.
1549 Number of ticks we kept the disk busy.
1552 Total time spent in the disk queue.
1559 It is also possible to get fio to dump the current output while it is
1560 running, without terminating the job. To do that, send fio the \fBUSR1\fR
1563 If the \fB\-\-minimal\fR option is given, the results will be printed in a
1564 semicolon-delimited format suitable for scripted use - a job description
1565 (if provided) follows on a new line. Note that the first
1566 number in the line is the version number. If the output has to be changed
1567 for some reason, this number will be incremented by 1 to signify that
1568 change. The fields are:
1571 .B terse version, fio version, jobname, groupid, error
1575 .B Total I/O \fR(KB)\fP, bandwidth \fR(KB/s)\fP, IOPS, runtime \fR(ms)\fP
1579 .B min, max, mean, standard deviation
1583 .B min, max, mean, standard deviation
1585 Completion latency percentiles (20 fields):
1587 .B Xth percentile=usec
1591 .B min, max, mean, standard deviation
1595 .B min, max, aggregate percentage of total, mean, standard deviation
1601 .B Total I/O \fR(KB)\fP, bandwidth \fR(KB/s)\fP, IOPS, runtime \fR(ms)\fP
1605 .B min, max, mean, standard deviation
1609 .B min, max, mean, standard deviation
1611 Completion latency percentiles (20 fields):
1613 .B Xth percentile=usec
1617 .B min, max, mean, standard deviation
1621 .B min, max, aggregate percentage of total, mean, standard deviation
1627 .B user, system, context switches, major page faults, minor page faults
1630 IO depth distribution:
1632 .B <=1, 2, 4, 8, 16, 32, >=64
1635 IO latency distribution:
1639 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
1643 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
1647 Disk utilization (1 for each disk used):
1649 .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
1652 Error Info (dependent on continue_on_error, default off):
1654 .B total # errors, first error code
1657 .B text description (if provided in config - appears on newline)
1660 Normally you would run fio as a stand-alone application on the machine
1661 where the IO workload should be generated. However, it is also possible to
1662 run the frontend and backend of fio separately. This makes it possible to
1663 have a fio server running on the machine(s) where the IO workload should
1664 be running, while controlling it from another machine.
1666 To start the server, you would do:
1668 \fBfio \-\-server=args\fR
1670 on that machine, where args defines what fio listens to. The arguments
1671 are of the form 'type:hostname or IP:port'. 'type' is either 'ip' (or ip4)
1672 for TCP/IP v4, 'ip6' for TCP/IP v6, or 'sock' for a local unix domain
1673 socket. 'hostname' is either a hostname or IP address, and 'port' is the port to
1674 listen to (only valid for TCP/IP, not a local socket). Some examples:
1678 Start a fio server, listening on all interfaces on the default port (8765).
1680 2) fio \-\-server=ip:hostname,4444
1682 Start a fio server, listening on IP belonging to hostname and on port 4444.
1684 3) fio \-\-server=ip6:::1,4444
1686 Start a fio server, listening on IPv6 localhost ::1 and on port 4444.
1688 4) fio \-\-server=,4444
1690 Start a fio server, listening on all interfaces on port 4444.
1692 5) fio \-\-server=1.2.3.4
1694 Start a fio server, listening on IP 1.2.3.4 on the default port.
1696 6) fio \-\-server=sock:/tmp/fio.sock
1698 Start a fio server, listening on the local socket /tmp/fio.sock.
1700 When a server is running, you can connect to it from a client. The client
1703 fio \-\-local-args \-\-client=server \-\-remote-args <job file(s)>
1705 where \-\-local-args are arguments that are local to the client where it is
1706 running, 'server' is the connect string, and \-\-remote-args and <job file(s)>
1707 are sent to the server. The 'server' string follows the same format as it
1708 does on the server side, to allow IP/hostname/socket and port strings.
1709 You can connect to multiple clients as well, to do that you could run:
1711 fio \-\-client=server2 \-\-client=server2 <job file(s)>
1715 was written by Jens Axboe <jens.axboe@oracle.com>,
1716 now Jens Axboe <jaxboe@fusionio.com>.
1718 This man page was written by Aaron Carroll <aaronc@cse.unsw.edu.au> based
1719 on documentation by Jens Axboe.
1720 .SH "REPORTING BUGS"
1721 Report bugs to the \fBfio\fR mailing list <fio@vger.kernel.org>.
1724 For further documentation see \fBHOWTO\fR and \fBREADME\fR.
1726 Sample jobfiles are available in the \fBexamples\fR directory.
projects / fio.git / blob