1 .TH fio 1 "October 2013" "User Manual"
3 fio \- flexible I/O tester
6 [\fIoptions\fR] [\fIjobfile\fR]...
9 is a tool that will spawn a number of threads or processes doing a
10 particular type of I/O action as specified by the user.
11 The typical use of fio is to write a job file matching the I/O load
12 one wants to simulate.
15 .BI \-\-debug \fR=\fPtype
16 Enable verbose tracing of various fio actions. May be `all' for all types
17 or individual types separated by a comma (eg \-\-debug=io,file). `help' will
18 list all available tracing options.
20 .BI \-\-output \fR=\fPfilename
21 Write output to \fIfilename\fR.
23 .BI \-\-runtime \fR=\fPruntime
24 Limit run time to \fIruntime\fR seconds.
27 Generate per-job latency logs.
30 Generate per-job bandwidth logs.
33 Print statistics in a terse, semicolon-delimited format.
36 Print statistics in selected mode AND terse, semicolon-delimited format.
39 Display version information and exit.
41 .BI \-\-terse\-version \fR=\fPversion
42 Set terse version output format (Current version 3, or older version 2).
45 Display usage information and exit.
48 Perform test and validation of internal CPU clock
50 .BI \-\-crctest[\fR=\fPtest]
51 Test the speed of the builtin checksumming functions. If no argument is given,
52 all of them are tested. Or a comma separated list can be passed, in which
53 case the given ones are tested.
55 .BI \-\-cmdhelp \fR=\fPcommand
56 Print help information for \fIcommand\fR. May be `all' for all commands.
58 .BI \-\-enghelp \fR=\fPioengine[,command]
59 List all commands defined by \fIioengine\fR, or print help for \fIcommand\fR defined by \fIioengine\fR.
61 .BI \-\-showcmd \fR=\fPjobfile
62 Convert \fIjobfile\fR to a set of command-line options.
64 .BI \-\-eta \fR=\fPwhen
65 Specifies when real-time ETA estimate should be printed. \fIwhen\fR may
66 be one of `always', `never' or `auto'.
68 .BI \-\-eta\-newline \fR=\fPtime
69 Force an ETA newline for every `time` period passed.
71 .BI \-\-status\-interval \fR=\fPtime
72 Report full output status every `time` period passed.
75 Turn on safety read-only checks, preventing any attempted write.
77 .BI \-\-section \fR=\fPsec
78 Only run section \fIsec\fR from job file. This option can be used multiple times to add more sections to run.
80 .BI \-\-alloc\-size \fR=\fPkb
81 Set the internal smalloc pool size to \fIkb\fP kilobytes.
83 .BI \-\-warnings\-fatal
84 All fio parser warnings are fatal, causing fio to exit with an error.
86 .BI \-\-max\-jobs \fR=\fPnr
87 Set the maximum allowed number of jobs (threads/processes) to support.
89 .BI \-\-server \fR=\fPargs
90 Start a backend server, with \fIargs\fP specifying what to listen to. See client/server section.
92 .BI \-\-daemonize \fR=\fPpidfile
93 Background a fio server, writing the pid to the given pid file.
95 .BI \-\-client \fR=\fPhost
96 Instead of running the jobs locally, send and run them on the given host.
98 .BI \-\-idle\-prof \fR=\fPoption
99 Report cpu idleness on a system or percpu basis (\fIoption\fP=system,percpu) or run unit work calibration only (\fIoption\fP=calibrate).
100 .SH "JOB FILE FORMAT"
101 Job files are in `ini' format. They consist of one or more
102 job definitions, which begin with a job name in square brackets and
103 extend to the next job name. The job name can be any ASCII string
104 except `global', which has a special meaning. Following the job name is
105 a sequence of zero or more parameters, one per line, that define the
106 behavior of the job. Any line starting with a `;' or `#' character is
107 considered a comment and ignored.
109 If \fIjobfile\fR is specified as `-', the job file will be read from
112 The global section contains default parameters for jobs specified in the
113 job file. A job is only affected by global sections residing above it,
114 and there may be any number of global sections. Specific job definitions
115 may override any parameter set in global sections.
118 Some parameters may take arguments of a specific type.
119 Anywhere a numeric value is required, an arithmetic expression may be used,
120 provided it is surrounded by parentheses.
121 Supported operators are
122 addition, subtraction, multiplication and division.
123 For time values in expressions
124 units are microseconds by default. This is different than for time
125 values not in expressions (not enclosed in parentheses).
129 String: a sequence of alphanumeric characters.
132 SI integer: a whole number, possibly containing a suffix denoting the base unit
133 of the value. Accepted suffixes are `k', 'M', 'G', 'T', and 'P', denoting
134 kilo (1024), mega (1024^2), giga (1024^3), tera (1024^4), and peta (1024^5)
135 respectively. If prefixed with '0x', the value is assumed to be base 16
136 (hexadecimal). A suffix may include a trailing 'b', for instance 'kb' is
137 identical to 'k'. You can specify a base 10 value by using 'KiB', 'MiB','GiB',
138 etc. This is useful for disk drives where values are often given in base 10
139 values. Specifying '30GiB' will get you 30*1000^3 bytes.
140 When specifying times the default suffix meaning changes, still denoting the
141 base unit of the value, but accepted suffixes are 'D' (days), 'H' (hours), 'M'
142 (minutes), 'S' Seconds, 'ms' (or msec) milli seconds, 'us' (or 'usec') micro
143 seconds. Time values without a unit specify seconds.
144 The suffixes are not case sensitive.
147 Boolean: a true or false value. `0' denotes false, `1' denotes true.
150 Integer range: a range of integers specified in the format
151 \fIlower\fR:\fIupper\fR or \fIlower\fR\-\fIupper\fR. \fIlower\fR and
152 \fIupper\fR may contain a suffix as described above. If an option allows two
153 sets of ranges, they are separated with a `,' or `/' character. For example:
157 List of floating numbers: A list of floating numbers, separated by
162 May be used to override the job name. On the command line, this parameter
163 has the special purpose of signalling the start of a new job.
165 .BI description \fR=\fPstr
166 Human-readable description of the job. It is printed when the job is run, but
167 otherwise has no special purpose.
169 .BI directory \fR=\fPstr
170 Prefix filenames with this directory. Used to place files in a location other
172 You can specify a number of directories by separating the names with a ':'
173 character. These directories will be assigned equally distributed to job clones
174 creates with \fInumjobs\fR as long as they are using generated filenames.
175 If specific \fIfilename(s)\fR are set fio will use the first listed directory,
176 and thereby matching the \fIfilename\fR semantic which generates a file each
177 clone if not specified, but let all clones use the same if set. See
178 \fIfilename\fR for considerations regarding escaping certain characters on
181 .BI filename \fR=\fPstr
183 normally makes up a file name based on the job name, thread number, and file
184 number. If you want to share files between threads in a job or several jobs,
185 specify a \fIfilename\fR for each of them to override the default.
186 If the I/O engine is file-based, you can specify
187 a number of files by separating the names with a `:' character. `\-' is a
188 reserved name, meaning stdin or stdout, depending on the read/write direction
189 set. On Windows, disk devices are accessed as \\.\PhysicalDrive0 for the first
190 device, \\.\PhysicalDrive1 for the second etc. Note: Windows and FreeBSD
191 prevent write access to areas of the disk containing in-use data
192 (e.g. filesystems). If the wanted filename does need to include a colon, then
193 escape that with a '\\' character. For instance, if the filename is
194 "/dev/dsk/foo@3,0:c", then you would use filename="/dev/dsk/foo@3,0\\:c".
196 .BI filename_format \fR=\fPstr
197 If sharing multiple files between jobs, it is usually necessary to have
198 fio generate the exact names that you want. By default, fio will name a file
199 based on the default file format specification of
200 \fBjobname.jobnumber.filenumber\fP. With this option, that can be
201 customized. Fio will recognize and replace the following keywords in this
207 The name of the worker thread or process.
210 The incremental number of the worker thread or process.
213 The incremental number of the file for that worker thread or process.
216 To have dependent jobs share a set of files, this option can be set to
217 have fio generate filenames that are shared between the two. For instance,
218 if \fBtestfiles.$filenum\fR is specified, file number 4 for any job will
219 be named \fBtestfiles.4\fR. The default of \fB$jobname.$jobnum.$filenum\fR
220 will be used if no other format specifier is given.
224 .BI lockfile \fR=\fPstr
225 Fio defaults to not locking any files before it does IO to them. If a file or
226 file descriptor is shared, fio can serialize IO to that file to make the end
227 result consistent. This is usual for emulating real workloads that share files.
233 No locking. This is the default.
236 Only one thread or process may do IO at a time, excluding all others.
239 Read-write locking on the file. Many readers may access the file at the same
240 time, but writes get exclusive access.
244 .BI opendir \fR=\fPstr
245 Recursively open any files below directory \fIstr\fR.
247 .BI readwrite \fR=\fPstr "\fR,\fP rw" \fR=\fPstr
248 Type of I/O pattern. Accepted values are:
259 Sequential trim (Linux block devices only).
268 Random trim (Linux block devices only).
271 Mixed sequential reads and writes.
274 Mixed random reads and writes.
277 For mixed I/O, the default split is 50/50. For certain types of io the result
278 may still be skewed a bit, since the speed may be different. It is possible to
279 specify a number of IO's to do before getting a new offset, this is done by
280 appending a `:\fI<nr>\fR to the end of the string given. For a random read, it
281 would look like \fBrw=randread:8\fR for passing in an offset modifier with a
282 value of 8. If the postfix is used with a sequential IO pattern, then the value
283 specified will be added to the generated offset for each IO. For instance,
284 using \fBrw=write:4k\fR will skip 4k for every write. It turns sequential IO
285 into sequential IO with holes. See the \fBrw_sequencer\fR option.
288 .BI rw_sequencer \fR=\fPstr
289 If an offset modifier is given by appending a number to the \fBrw=<str>\fR line,
290 then this option controls how that number modifies the IO offset being
291 generated. Accepted values are:
296 Generate sequential offset
299 Generate the same offset
302 \fBsequential\fR is only useful for random IO, where fio would normally
303 generate a new random offset for every IO. If you append eg 8 to randread, you
304 would get a new random offset for every 8 IO's. The result would be a seek for
305 only every 8 IO's, instead of for every IO. Use \fBrw=randread:8\fR to specify
306 that. As sequential IO is already sequential, setting \fBsequential\fR for that
307 would not result in any differences. \fBidentical\fR behaves in a similar
308 fashion, except it sends the same offset 8 number of times before generating a
313 .BI kb_base \fR=\fPint
314 The base unit for a kilobyte. The defacto base is 2^10, 1024. Storage
315 manufacturers like to use 10^3 or 1000 as a base ten unit instead, for obvious
316 reasons. Allowed values are 1024 or 1000, with 1024 being the default.
318 .BI unified_rw_reporting \fR=\fPbool
319 Fio normally reports statistics on a per data direction basis, meaning that
320 read, write, and trim are accounted and reported separately. If this option is
321 set fio sums the results and reports them as "mixed" instead.
323 .BI randrepeat \fR=\fPbool
324 Seed the random number generator used for random I/O patterns in a predictable
325 way so the pattern is repeatable across runs. Default: true.
327 .BI allrandrepeat \fR=\fPbool
328 Seed all random number generators in a predictable way so results are
329 repeatable across runs. Default: false.
331 .BI randseed \fR=\fPint
332 Seed the random number generators based on this seed value, to be able to
333 control what sequence of output is being generated. If not set, the random
334 sequence depends on the \fBrandrepeat\fR setting.
336 .BI use_os_rand \fR=\fPbool
337 Fio can either use the random generator supplied by the OS to generate random
338 offsets, or it can use its own internal generator (based on Tausworthe).
339 Default is to use the internal generator, which is often of better quality and
340 faster. Default: false.
342 .BI fallocate \fR=\fPstr
343 Whether pre-allocation is performed when laying down files. Accepted values
349 Do not pre-allocate space.
352 Pre-allocate via \fBposix_fallocate\fR\|(3).
355 Pre-allocate via \fBfallocate\fR\|(2) with FALLOC_FL_KEEP_SIZE set.
358 Backward-compatible alias for 'none'.
361 Backward-compatible alias for 'posix'.
364 May not be available on all supported platforms. 'keep' is only
365 available on Linux. If using ZFS on Solaris this must be set to 'none'
366 because ZFS doesn't support it. Default: 'posix'.
369 .BI fadvise_hint \fR=\fPbool
370 Use \fBposix_fadvise\fR\|(2) to advise the kernel what I/O patterns
371 are likely to be issued. Default: true.
374 Total size of I/O for this job. \fBfio\fR will run until this many bytes have
375 been transferred, unless limited by other options (\fBruntime\fR, for instance).
376 Unless \fBnrfiles\fR and \fBfilesize\fR options are given, this amount will be
377 divided between the available files for the job. If not set, fio will use the
378 full size of the given files or devices. If the files do not exist, size
379 must be given. It is also possible to give size as a percentage between 1 and
380 100. If size=20% is given, fio will use 20% of the full size of the given
383 .BI io_limit \fR=\fPint
384 Normally fio operates within the region set by \fBsize\fR, which means that
385 the \fBsize\fR option sets both the region and size of IO to be performed.
386 Sometimes that is not what you want. With this option, it is possible to
387 define just the amount of IO that fio should do. For instance, if \fBsize\fR
388 is set to 20G and \fBio_limit\fR is set to 5G, fio will perform IO within
389 the first 20G but exit when 5G have been done.
391 .BI fill_device \fR=\fPbool "\fR,\fB fill_fs" \fR=\fPbool
392 Sets size to something really large and waits for ENOSPC (no space left on
393 device) as the terminating condition. Only makes sense with sequential write.
394 For a read workload, the mount point will be filled first then IO started on
395 the result. This option doesn't make sense if operating on a raw device node,
396 since the size of that is already known by the file system. Additionally,
397 writing beyond end-of-device will not return ENOSPC there.
399 .BI filesize \fR=\fPirange
400 Individual file sizes. May be a range, in which case \fBfio\fR will select sizes
401 for files at random within the given range, limited to \fBsize\fR in total (if
402 that is given). If \fBfilesize\fR is not specified, each created file is the
405 .BI file_append \fR=\fPbool
406 Perform IO after the end of the file. Normally fio will operate within the
407 size of a file. If this option is set, then fio will append to the file
408 instead. This has identical behavior to setting \fRoffset\fP to the size
409 of a file. This option is ignored on non-regular files.
411 .BI blocksize \fR=\fPint[,int] "\fR,\fB bs" \fR=\fPint[,int]
412 Block size for I/O units. Default: 4k. Values for reads, writes, and trims
413 can be specified separately in the format \fIread\fR,\fIwrite\fR,\fItrim\fR
414 either of which may be empty to leave that value at its default. If a trailing
415 comma isn't given, the remainder will inherit the last value set.
417 .BI blocksize_range \fR=\fPirange[,irange] "\fR,\fB bsrange" \fR=\fPirange[,irange]
418 Specify a range of I/O block sizes. The issued I/O unit will always be a
419 multiple of the minimum size, unless \fBblocksize_unaligned\fR is set. Applies
420 to both reads and writes if only one range is given, but can be specified
421 separately with a comma separating the values. Example: bsrange=1k-4k,2k-8k.
422 Also (see \fBblocksize\fR).
424 .BI bssplit \fR=\fPstr
425 This option allows even finer grained control of the block sizes issued,
426 not just even splits between them. With this option, you can weight various
427 block sizes for exact control of the issued IO for a job that has mixed
428 block sizes. The format of the option is bssplit=blocksize/percentage,
429 optionally adding as many definitions as needed separated by a colon.
430 Example: bssplit=4k/10:64k/50:32k/40 would issue 50% 64k blocks, 10% 4k
431 blocks and 40% 32k blocks. \fBbssplit\fR also supports giving separate
432 splits to reads and writes. The format is identical to what the
433 \fBbs\fR option accepts, the read and write parts are separated with a
436 .B blocksize_unaligned\fR,\fP bs_unaligned
437 If set, any size in \fBblocksize_range\fR may be used. This typically won't
438 work with direct I/O, as that normally requires sector alignment.
440 .BI blockalign \fR=\fPint[,int] "\fR,\fB ba" \fR=\fPint[,int]
441 At what boundary to align random IO offsets. Defaults to the same as 'blocksize'
442 the minimum blocksize given. Minimum alignment is typically 512b
443 for using direct IO, though it usually depends on the hardware block size.
444 This option is mutually exclusive with using a random map for files, so it
445 will turn off that option.
447 .BI bs_is_seq_rand \fR=\fPbool
448 If this option is set, fio will use the normal read,write blocksize settings as
449 sequential,random instead. Any random read or write will use the WRITE
450 blocksize settings, and any sequential read or write will use the READ
454 Initialize buffers with all zeros. Default: fill buffers with random data.
455 The resulting IO buffers will not be completely zeroed, unless
456 \fPscramble_buffers\fR is also turned off.
459 If this option is given, fio will refill the IO buffers on every submit. The
460 default is to only fill it at init time and reuse that data. Only makes sense
461 if zero_buffers isn't specified, naturally. If data verification is enabled,
462 refill_buffers is also automatically enabled.
464 .BI scramble_buffers \fR=\fPbool
465 If \fBrefill_buffers\fR is too costly and the target is using data
466 deduplication, then setting this option will slightly modify the IO buffer
467 contents to defeat normal de-dupe attempts. This is not enough to defeat
468 more clever block compression attempts, but it will stop naive dedupe
469 of blocks. Default: true.
471 .BI buffer_compress_percentage \fR=\fPint
472 If this is set, then fio will attempt to provide IO buffer content (on WRITEs)
473 that compress to the specified level. Fio does this by providing a mix of
474 random data and zeroes. Note that this is per block size unit, for file/disk
475 wide compression level that matches this setting, you'll also want to set
476 \fBrefill_buffers\fR.
478 .BI buffer_compress_chunk \fR=\fPint
479 See \fBbuffer_compress_percentage\fR. This setting allows fio to manage how
480 big the ranges of random data and zeroed data is. Without this set, fio will
481 provide \fBbuffer_compress_percentage\fR of blocksize random data, followed by
482 the remaining zeroed. With this set to some chunk size smaller than the block
483 size, fio can alternate random and zeroed data throughout the IO buffer.
485 .BI buffer_pattern \fR=\fPstr
486 If set, fio will fill the IO buffers with this pattern. If not set, the contents
487 of IO buffers is defined by the other options related to buffer contents. The
488 setting can be any pattern of bytes, and can be prefixed with 0x for hex
489 values. It may also be a string, where the string must then be wrapped with
492 .BI dedupe_percentage \fR=\fPint
493 If set, fio will generate this percentage of identical buffers when writing.
494 These buffers will be naturally dedupable. The contents of the buffers depend
495 on what other buffer compression settings have been set. It's possible to have
496 the individual buffers either fully compressible, or not at all. This option
497 only controls the distribution of unique buffers.
499 .BI nrfiles \fR=\fPint
500 Number of files to use for this job. Default: 1.
502 .BI openfiles \fR=\fPint
503 Number of files to keep open at the same time. Default: \fBnrfiles\fR.
505 .BI file_service_type \fR=\fPstr
506 Defines how files to service are selected. The following types are defined:
511 Choose a file at random.
514 Round robin over opened files (default).
517 Do each file in the set sequentially.
520 The number of I/Os to issue before switching to a new file can be specified by
521 appending `:\fIint\fR' to the service type.
524 .BI ioengine \fR=\fPstr
525 Defines how the job issues I/O. The following types are defined:
530 Basic \fBread\fR\|(2) or \fBwrite\fR\|(2) I/O. \fBfseek\fR\|(2) is used to
531 position the I/O location.
534 Basic \fBpread\fR\|(2) or \fBpwrite\fR\|(2) I/O.
537 Basic \fBreadv\fR\|(2) or \fBwritev\fR\|(2) I/O. Will emulate queuing by
538 coalescing adjacent IOs into a single submission.
541 Basic \fBpreadv\fR\|(2) or \fBpwritev\fR\|(2) I/O.
544 Linux native asynchronous I/O. This ioengine defines engine specific options.
547 POSIX asynchronous I/O using \fBaio_read\fR\|(3) and \fBaio_write\fR\|(3).
550 Solaris native asynchronous I/O.
553 Windows native asynchronous I/O.
556 File is memory mapped with \fBmmap\fR\|(2) and data copied using
560 \fBsplice\fR\|(2) is used to transfer the data and \fBvmsplice\fR\|(2) to
561 transfer data from user-space to the kernel.
564 Use the syslet system calls to make regular read/write asynchronous.
567 SCSI generic sg v3 I/O. May be either synchronous using the SG_IO ioctl, or if
568 the target is an sg character device, we use \fBread\fR\|(2) and
569 \fBwrite\fR\|(2) for asynchronous I/O.
572 Doesn't transfer any data, just pretends to. Mainly used to exercise \fBfio\fR
573 itself and for debugging and testing purposes.
576 Transfer over the network. The protocol to be used can be defined with the
577 \fBprotocol\fR parameter. Depending on the protocol, \fBfilename\fR,
578 \fBhostname\fR, \fBport\fR, or \fBlisten\fR must be specified.
579 This ioengine defines engine specific options.
582 Like \fBnet\fR, but uses \fBsplice\fR\|(2) and \fBvmsplice\fR\|(2) to map data
583 and send/receive. This ioengine defines engine specific options.
586 Doesn't transfer any data, but burns CPU cycles according to \fBcpuload\fR and
587 \fBcpucycles\fR parameters.
590 The GUASI I/O engine is the Generic Userspace Asynchronous Syscall Interface
591 approach to asynchronous I/O.
593 See <http://www.xmailserver.org/guasi\-lib.html>.
596 The RDMA I/O engine supports both RDMA memory semantics (RDMA_WRITE/RDMA_READ)
597 and channel semantics (Send/Recv) for the InfiniBand, RoCE and iWARP protocols.
600 Loads an external I/O engine object file. Append the engine filename as
604 IO engine that does regular linux native fallocate call to simulate data
605 transfer as fio ioengine
607 DDIR_READ does fallocate(,mode = FALLOC_FL_KEEP_SIZE,)
609 DIR_WRITE does fallocate(,mode = 0)
611 DDIR_TRIM does fallocate(,mode = FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE)
614 IO engine that does regular EXT4_IOC_MOVE_EXT ioctls to simulate defragment activity
615 request to DDIR_WRITE event
618 IO engine supporting direct access to Ceph Rados Block Devices (RBD) via librbd
619 without the need to use the kernel rbd driver. This ioengine defines engine specific
623 Using Glusterfs libgfapi sync interface to direct access to Glusterfs volumes without
624 having to go through FUSE. This ioengine defines engine specific
628 Using Glusterfs libgfapi async interface to direct access to Glusterfs volumes without
629 having to go through FUSE. This ioengine defines engine specific
633 Read and write through Hadoop (HDFS). The \fBfilename\fR option is used to
634 specify host,port of the hdfs name-node to connect. This engine interprets
635 offsets a little differently. In HDFS, files once created cannot be modified.
636 So random writes are not possible. To imitate this, libhdfs engine expects
637 bunch of small files to be created over HDFS, and engine will randomly pick a
638 file out of those files based on the offset generated by fio backend. (see the
639 example job file to create such files, use rw=write option). Please note, you
640 might want to set necessary environment variables to work with hdfs/libhdfs
646 .BI iodepth \fR=\fPint
647 Number of I/O units to keep in flight against the file. Note that increasing
648 iodepth beyond 1 will not affect synchronous ioengines (except for small
649 degress when verify_async is in use). Even async engines may impose OS
650 restrictions causing the desired depth not to be achieved. This may happen on
651 Linux when using libaio and not setting \fBdirect\fR=1, since buffered IO is
652 not async on that OS. Keep an eye on the IO depth distribution in the
653 fio output to verify that the achieved depth is as expected. Default: 1.
655 .BI iodepth_batch \fR=\fPint
656 Number of I/Os to submit at once. Default: \fBiodepth\fR.
658 .BI iodepth_batch_complete \fR=\fPint
659 This defines how many pieces of IO to retrieve at once. It defaults to 1 which
660 means that we'll ask for a minimum of 1 IO in the retrieval process from the
661 kernel. The IO retrieval will go on until we hit the limit set by
662 \fBiodepth_low\fR. If this variable is set to 0, then fio will always check for
663 completed events before queuing more IO. This helps reduce IO latency, at the
664 cost of more retrieval system calls.
666 .BI iodepth_low \fR=\fPint
667 Low watermark indicating when to start filling the queue again. Default:
670 .BI direct \fR=\fPbool
671 If true, use non-buffered I/O (usually O_DIRECT). Default: false.
673 .BI atomic \fR=\fPbool
674 If value is true, attempt to use atomic direct IO. Atomic writes are guaranteed
675 to be stable once acknowledged by the operating system. Only Linux supports
678 .BI buffered \fR=\fPbool
679 If true, use buffered I/O. This is the opposite of the \fBdirect\fR parameter.
682 .BI offset \fR=\fPint
683 Offset in the file to start I/O. Data before the offset will not be touched.
685 .BI offset_increment \fR=\fPint
686 If this is provided, then the real offset becomes the
687 offset + offset_increment * thread_number, where the thread number is a
688 counter that starts at 0 and is incremented for each sub-job (i.e. when
689 numjobs option is specified). This option is useful if there are several jobs
690 which are intended to operate on a file in parallel disjoint segments, with
691 even spacing between the starting points.
693 .BI number_ios \fR=\fPint
694 Fio will normally perform IOs until it has exhausted the size of the region
695 set by \fBsize\fR, or if it exhaust the allocated time (or hits an error
696 condition). With this setting, the range/size can be set independently of
697 the number of IOs to perform. When fio reaches this number, it will exit
698 normally and report status. Note that this does not extend the amount
699 of IO that will be done, it will only stop fio if this condition is met
700 before other end-of-job criteria.
703 How many I/Os to perform before issuing an \fBfsync\fR\|(2) of dirty data. If
704 0, don't sync. Default: 0.
706 .BI fdatasync \fR=\fPint
707 Like \fBfsync\fR, but uses \fBfdatasync\fR\|(2) instead to only sync the
708 data parts of the file. Default: 0.
710 .BI write_barrier \fR=\fPint
711 Make every Nth write a barrier write.
713 .BI sync_file_range \fR=\fPstr:int
714 Use \fBsync_file_range\fR\|(2) for every \fRval\fP number of write operations. Fio will
715 track range of writes that have happened since the last \fBsync_file_range\fR\|(2) call.
716 \fRstr\fP can currently be one or more of:
720 SYNC_FILE_RANGE_WAIT_BEFORE
723 SYNC_FILE_RANGE_WRITE
726 SYNC_FILE_RANGE_WRITE
730 So if you do sync_file_range=wait_before,write:8, fio would use
731 \fBSYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE\fP for every 8 writes.
732 Also see the \fBsync_file_range\fR\|(2) man page. This option is Linux specific.
734 .BI overwrite \fR=\fPbool
735 If writing, setup the file first and do overwrites. Default: false.
737 .BI end_fsync \fR=\fPbool
738 Sync file contents when a write stage has completed. Default: false.
740 .BI fsync_on_close \fR=\fPbool
741 If true, sync file contents on close. This differs from \fBend_fsync\fR in that
742 it will happen on every close, not just at the end of the job. Default: false.
744 .BI rwmixread \fR=\fPint
745 Percentage of a mixed workload that should be reads. Default: 50.
747 .BI rwmixwrite \fR=\fPint
748 Percentage of a mixed workload that should be writes. If \fBrwmixread\fR and
749 \fBrwmixwrite\fR are given and do not sum to 100%, the latter of the two
750 overrides the first. This may interfere with a given rate setting, if fio is
751 asked to limit reads or writes to a certain rate. If that is the case, then
752 the distribution may be skewed. Default: 50.
754 .BI random_distribution \fR=\fPstr:float
755 By default, fio will use a completely uniform random distribution when asked
756 to perform random IO. Sometimes it is useful to skew the distribution in
757 specific ways, ensuring that some parts of the data is more hot than others.
758 Fio includes the following distribution models:
762 Uniform random distribution
772 When using a zipf or pareto distribution, an input value is also needed to
773 define the access pattern. For zipf, this is the zipf theta. For pareto,
774 it's the pareto power. Fio includes a test program, genzipf, that can be
775 used visualize what the given input values will yield in terms of hit rates.
776 If you wanted to use zipf with a theta of 1.2, you would use
777 random_distribution=zipf:1.2 as the option. If a non-uniform model is used,
778 fio will disable use of the random map.
780 .BI percentage_random \fR=\fPint
781 For a random workload, set how big a percentage should be random. This defaults
782 to 100%, in which case the workload is fully random. It can be set from
783 anywhere from 0 to 100. Setting it to 0 would make the workload fully
784 sequential. It is possible to set different values for reads, writes, and
785 trim. To do so, simply use a comma separated list. See \fBblocksize\fR.
788 Normally \fBfio\fR will cover every block of the file when doing random I/O. If
789 this parameter is given, a new offset will be chosen without looking at past
790 I/O history. This parameter is mutually exclusive with \fBverify\fR.
792 .BI softrandommap \fR=\fPbool
793 See \fBnorandommap\fR. If fio runs with the random block map enabled and it
794 fails to allocate the map, if this option is set it will continue without a
795 random block map. As coverage will not be as complete as with random maps, this
796 option is disabled by default.
798 .BI random_generator \fR=\fPstr
799 Fio supports the following engines for generating IO offsets for random IO:
803 Strong 2^88 cycle random number generator
806 Linear feedback shift register generator
810 Tausworthe is a strong random number generator, but it requires tracking on the
811 side if we want to ensure that blocks are only read or written once. LFSR
812 guarantees that we never generate the same offset twice, and it's also less
813 computationally expensive. It's not a true random generator, however, though
814 for IO purposes it's typically good enough. LFSR only works with single block
815 sizes, not with workloads that use multiple block sizes. If used with such a
816 workload, fio may read or write some blocks multiple times.
819 Run job with given nice value. See \fBnice\fR\|(2).
822 Set I/O priority value of this job between 0 (highest) and 7 (lowest). See
825 .BI prioclass \fR=\fPint
826 Set I/O priority class. See \fBionice\fR\|(1).
828 .BI thinktime \fR=\fPint
829 Stall job for given number of microseconds between issuing I/Os.
831 .BI thinktime_spin \fR=\fPint
832 Pretend to spend CPU time for given number of microseconds, sleeping the rest
833 of the time specified by \fBthinktime\fR. Only valid if \fBthinktime\fR is set.
835 .BI thinktime_blocks \fR=\fPint
836 Only valid if thinktime is set - control how many blocks to issue, before
837 waiting \fBthinktime\fR microseconds. If not set, defaults to 1 which will
838 make fio wait \fBthinktime\fR microseconds after every block. This
839 effectively makes any queue depth setting redundant, since no more than 1 IO
840 will be queued before we have to complete it and do our thinktime. In other
841 words, this setting effectively caps the queue depth if the latter is larger.
845 Cap bandwidth used by this job. The number is in bytes/sec, the normal postfix
846 rules apply. You can use \fBrate\fR=500k to limit reads and writes to 500k each,
847 or you can specify read and writes separately. Using \fBrate\fR=1m,500k would
848 limit reads to 1MB/sec and writes to 500KB/sec. Capping only reads or writes
849 can be done with \fBrate\fR=,500k or \fBrate\fR=500k,. The former will only
850 limit writes (to 500KB/sec), the latter will only limit reads.
852 .BI ratemin \fR=\fPint
853 Tell \fBfio\fR to do whatever it can to maintain at least the given bandwidth.
854 Failing to meet this requirement will cause the job to exit. The same format
855 as \fBrate\fR is used for read vs write separation.
857 .BI rate_iops \fR=\fPint
858 Cap the bandwidth to this number of IOPS. Basically the same as rate, just
859 specified independently of bandwidth. The same format as \fBrate\fR is used for
860 read vs write separation. If \fBblocksize\fR is a range, the smallest block
861 size is used as the metric.
863 .BI rate_iops_min \fR=\fPint
864 If this rate of I/O is not met, the job will exit. The same format as \fBrate\fR
865 is used for read vs write separation.
867 .BI ratecycle \fR=\fPint
868 Average bandwidth for \fBrate\fR and \fBratemin\fR over this number of
869 milliseconds. Default: 1000ms.
871 .BI latency_target \fR=\fPint
872 If set, fio will attempt to find the max performance point that the given
873 workload will run at while maintaining a latency below this target. The
874 values is given in microseconds. See \fBlatency_window\fR and
875 \fBlatency_percentile\fR.
877 .BI latency_window \fR=\fPint
878 Used with \fBlatency_target\fR to specify the sample window that the job
879 is run at varying queue depths to test the performance. The value is given
882 .BI latency_percentile \fR=\fPfloat
883 The percentage of IOs that must fall within the criteria specified by
884 \fBlatency_target\fR and \fBlatency_window\fR. If not set, this defaults
885 to 100.0, meaning that all IOs must be equal or below to the value set
886 by \fBlatency_target\fR.
888 .BI max_latency \fR=\fPint
889 If set, fio will exit the job if it exceeds this maximum latency. It will exit
892 .BI cpumask \fR=\fPint
893 Set CPU affinity for this job. \fIint\fR is a bitmask of allowed CPUs the job
894 may run on. See \fBsched_setaffinity\fR\|(2).
896 .BI cpus_allowed \fR=\fPstr
897 Same as \fBcpumask\fR, but allows a comma-delimited list of CPU numbers.
899 .BI cpus_allowed_policy \fR=\fPstr
900 Set the policy of how fio distributes the CPUs specified by \fBcpus_allowed\fR
901 or \fBcpumask\fR. Two policies are supported:
906 All jobs will share the CPU set specified.
909 Each job will get a unique CPU from the CPU set.
912 \fBshared\fR is the default behaviour, if the option isn't specified. If
913 \fBsplit\fR is specified, then fio will assign one cpu per job. If not enough
914 CPUs are given for the jobs listed, then fio will roundrobin the CPUs in
919 .BI numa_cpu_nodes \fR=\fPstr
920 Set this job running on specified NUMA nodes' CPUs. The arguments allow
921 comma delimited list of cpu numbers, A-B ranges, or 'all'.
923 .BI numa_mem_policy \fR=\fPstr
924 Set this job's memory policy and corresponding NUMA nodes. Format of
928 .B <mode>[:<nodelist>]
931 is one of the following memory policy:
933 .B default, prefer, bind, interleave, local
936 For \fBdefault\fR and \fBlocal\fR memory policy, no \fBnodelist\fR is
937 needed to be specified. For \fBprefer\fR, only one node is
938 allowed. For \fBbind\fR and \fBinterleave\fR, \fBnodelist\fR allows
939 comma delimited list of numbers, A-B ranges, or 'all'.
941 .BI startdelay \fR=\fPirange
942 Delay start of job for the specified number of seconds. Supports all time
943 suffixes to allow specification of hours, minutes, seconds and
944 milliseconds - seconds are the default if a unit is ommited.
945 Can be given as a range which causes each thread to choose randomly out of the
948 .BI runtime \fR=\fPint
949 Terminate processing after the specified number of seconds.
952 If given, run for the specified \fBruntime\fR duration even if the files are
953 completely read or written. The same workload will be repeated as many times
954 as \fBruntime\fR allows.
956 .BI ramp_time \fR=\fPint
957 If set, fio will run the specified workload for this amount of time before
958 logging any performance numbers. Useful for letting performance settle before
959 logging results, thus minimizing the runtime required for stable results. Note
960 that the \fBramp_time\fR is considered lead in time for a job, thus it will
961 increase the total runtime if a special timeout or runtime is specified.
963 .BI invalidate \fR=\fPbool
964 Invalidate buffer-cache for the file prior to starting I/O. Default: true.
967 Use synchronous I/O for buffered writes. For the majority of I/O engines,
968 this means using O_SYNC. Default: false.
970 .BI iomem \fR=\fPstr "\fR,\fP mem" \fR=\fPstr
971 Allocation method for I/O unit buffer. Allowed values are:
976 Allocate memory with \fBmalloc\fR\|(3).
979 Use shared memory buffers allocated through \fBshmget\fR\|(2).
982 Same as \fBshm\fR, but use huge pages as backing.
985 Use \fBmmap\fR\|(2) for allocation. Uses anonymous memory unless a filename
986 is given after the option in the format `:\fIfile\fR'.
989 Same as \fBmmap\fR, but use huge files as backing.
992 The amount of memory allocated is the maximum allowed \fBblocksize\fR for the
993 job multiplied by \fBiodepth\fR. For \fBshmhuge\fR or \fBmmaphuge\fR to work,
994 the system must have free huge pages allocated. \fBmmaphuge\fR also needs to
995 have hugetlbfs mounted, and \fIfile\fR must point there. At least on Linux,
996 huge pages must be manually allocated. See \fB/proc/sys/vm/nr_hugehages\fR
997 and the documentation for that. Normally you just need to echo an appropriate
998 number, eg echoing 8 will ensure that the OS has 8 huge pages ready for
1002 .BI iomem_align \fR=\fPint "\fR,\fP mem_align" \fR=\fPint
1003 This indicates the memory alignment of the IO memory buffers. Note that the
1004 given alignment is applied to the first IO unit buffer, if using \fBiodepth\fR
1005 the alignment of the following buffers are given by the \fBbs\fR used. In
1006 other words, if using a \fBbs\fR that is a multiple of the page sized in the
1007 system, all buffers will be aligned to this value. If using a \fBbs\fR that
1008 is not page aligned, the alignment of subsequent IO memory buffers is the
1009 sum of the \fBiomem_align\fR and \fBbs\fR used.
1011 .BI hugepage\-size \fR=\fPint
1012 Defines the size of a huge page. Must be at least equal to the system setting.
1013 Should be a multiple of 1MB. Default: 4MB.
1016 Terminate all jobs when one finishes. Default: wait for each job to finish.
1018 .BI bwavgtime \fR=\fPint
1019 Average bandwidth calculations over the given time in milliseconds. Default:
1022 .BI iopsavgtime \fR=\fPint
1023 Average IOPS calculations over the given time in milliseconds. Default:
1026 .BI create_serialize \fR=\fPbool
1027 If true, serialize file creation for the jobs. Default: true.
1029 .BI create_fsync \fR=\fPbool
1030 \fBfsync\fR\|(2) data file after creation. Default: true.
1032 .BI create_on_open \fR=\fPbool
1033 If true, the files are not created until they are opened for IO by the job.
1035 .BI create_only \fR=\fPbool
1036 If true, fio will only run the setup phase of the job. If files need to be
1037 laid out or updated on disk, only that will be done. The actual job contents
1040 .BI pre_read \fR=\fPbool
1041 If this is given, files will be pre-read into memory before starting the given
1042 IO operation. This will also clear the \fR \fBinvalidate\fR flag, since it is
1043 pointless to pre-read and then drop the cache. This will only work for IO
1044 engines that are seekable, since they allow you to read the same data
1045 multiple times. Thus it will not work on eg network or splice IO.
1047 .BI unlink \fR=\fPbool
1048 Unlink job files when done. Default: false.
1050 .BI loops \fR=\fPint
1051 Specifies the number of iterations (runs of the same workload) of this job.
1054 .BI verify_only \fR=\fPbool
1055 Do not perform the specified workload, only verify data still matches previous
1056 invocation of this workload. This option allows one to check data multiple
1057 times at a later date without overwriting it. This option makes sense only for
1058 workloads that write data, and does not support workloads with the
1059 \fBtime_based\fR option set.
1061 .BI do_verify \fR=\fPbool
1062 Run the verify phase after a write phase. Only valid if \fBverify\fR is set.
1065 .BI verify \fR=\fPstr
1066 Method of verifying file contents after each iteration of the job. Allowed
1071 .B md5 crc16 crc32 crc32c crc32c-intel crc64 crc7 sha256 sha512 sha1 xxhash
1072 Store appropriate checksum in the header of each block. crc32c-intel is
1073 hardware accelerated SSE4.2 driven, falls back to regular crc32c if
1074 not supported by the system.
1077 Write extra information about each I/O (timestamp, block number, etc.). The
1078 block number is verified. See \fBverify_pattern\fR as well.
1081 Pretend to verify. Used for testing internals.
1084 This option can be used for repeated burn-in tests of a system to make sure
1085 that the written data is also correctly read back. If the data direction given
1086 is a read or random read, fio will assume that it should verify a previously
1087 written file. If the data direction includes any form of write, the verify will
1088 be of the newly written data.
1091 .BI verifysort \fR=\fPbool
1092 If true, written verify blocks are sorted if \fBfio\fR deems it to be faster to
1093 read them back in a sorted manner. Default: true.
1095 .BI verifysort_nr \fR=\fPint
1096 Pre-load and sort verify blocks for a read workload.
1098 .BI verify_offset \fR=\fPint
1099 Swap the verification header with data somewhere else in the block before
1100 writing. It is swapped back before verifying.
1102 .BI verify_interval \fR=\fPint
1103 Write the verification header for this number of bytes, which should divide
1104 \fBblocksize\fR. Default: \fBblocksize\fR.
1106 .BI verify_pattern \fR=\fPstr
1107 If set, fio will fill the io buffers with this pattern. Fio defaults to filling
1108 with totally random bytes, but sometimes it's interesting to fill with a known
1109 pattern for io verification purposes. Depending on the width of the pattern,
1110 fio will fill 1/2/3/4 bytes of the buffer at the time(it can be either a
1111 decimal or a hex number). The verify_pattern if larger than a 32-bit quantity
1112 has to be a hex number that starts with either "0x" or "0X". Use with
1115 .BI verify_fatal \fR=\fPbool
1116 If true, exit the job on the first observed verification failure. Default:
1119 .BI verify_dump \fR=\fPbool
1120 If set, dump the contents of both the original data block and the data block we
1121 read off disk to files. This allows later analysis to inspect just what kind of
1122 data corruption occurred. Off by default.
1124 .BI verify_async \fR=\fPint
1125 Fio will normally verify IO inline from the submitting thread. This option
1126 takes an integer describing how many async offload threads to create for IO
1127 verification instead, causing fio to offload the duty of verifying IO contents
1128 to one or more separate threads. If using this offload option, even sync IO
1129 engines can benefit from using an \fBiodepth\fR setting higher than 1, as it
1130 allows them to have IO in flight while verifies are running.
1132 .BI verify_async_cpus \fR=\fPstr
1133 Tell fio to set the given CPU affinity on the async IO verification threads.
1134 See \fBcpus_allowed\fP for the format used.
1136 .BI verify_backlog \fR=\fPint
1137 Fio will normally verify the written contents of a job that utilizes verify
1138 once that job has completed. In other words, everything is written then
1139 everything is read back and verified. You may want to verify continually
1140 instead for a variety of reasons. Fio stores the meta data associated with an
1141 IO block in memory, so for large verify workloads, quite a bit of memory would
1142 be used up holding this meta data. If this option is enabled, fio will write
1143 only N blocks before verifying these blocks.
1145 .BI verify_backlog_batch \fR=\fPint
1146 Control how many blocks fio will verify if verify_backlog is set. If not set,
1147 will default to the value of \fBverify_backlog\fR (meaning the entire queue is
1148 read back and verified). If \fBverify_backlog_batch\fR is less than
1149 \fBverify_backlog\fR then not all blocks will be verified, if
1150 \fBverify_backlog_batch\fR is larger than \fBverify_backlog\fR, some blocks
1151 will be verified more than once.
1153 .BI trim_percentage \fR=\fPint
1154 Number of verify blocks to discard/trim.
1156 .BI trim_verify_zero \fR=\fPbool
1157 Verify that trim/discarded blocks are returned as zeroes.
1159 .BI trim_backlog \fR=\fPint
1160 Trim after this number of blocks are written.
1162 .BI trim_backlog_batch \fR=\fPint
1163 Trim this number of IO blocks.
1165 .BI experimental_verify \fR=\fPbool
1166 Enable experimental verification.
1168 .B stonewall "\fR,\fP wait_for_previous"
1169 Wait for preceding jobs in the job file to exit before starting this one.
1170 \fBstonewall\fR implies \fBnew_group\fR.
1173 Start a new reporting group. If not given, all jobs in a file will be part
1174 of the same reporting group, unless separated by a stonewall.
1176 .BI numjobs \fR=\fPint
1177 Number of clones (processes/threads performing the same workload) of this job.
1181 If set, display per-group reports instead of per-job when \fBnumjobs\fR is
1185 Use threads created with \fBpthread_create\fR\|(3) instead of processes created
1186 with \fBfork\fR\|(2).
1188 .BI zonesize \fR=\fPint
1189 Divide file into zones of the specified size in bytes. See \fBzoneskip\fR.
1191 .BI zonerange \fR=\fPint
1192 Give size of an IO zone. See \fBzoneskip\fR.
1194 .BI zoneskip \fR=\fPint
1195 Skip the specified number of bytes when \fBzonesize\fR bytes of data have been
1198 .BI write_iolog \fR=\fPstr
1199 Write the issued I/O patterns to the specified file. Specify a separate file
1200 for each job, otherwise the iologs will be interspersed and the file may be
1203 .BI read_iolog \fR=\fPstr
1204 Replay the I/O patterns contained in the specified file generated by
1205 \fBwrite_iolog\fR, or may be a \fBblktrace\fR binary file.
1207 .BI replay_no_stall \fR=\fPint
1208 While replaying I/O patterns using \fBread_iolog\fR the default behavior
1209 attempts to respect timing information between I/Os. Enabling
1210 \fBreplay_no_stall\fR causes I/Os to be replayed as fast as possible while
1211 still respecting ordering.
1213 .BI replay_redirect \fR=\fPstr
1214 While replaying I/O patterns using \fBread_iolog\fR the default behavior
1215 is to replay the IOPS onto the major/minor device that each IOP was recorded
1216 from. Setting \fBreplay_redirect\fR causes all IOPS to be replayed onto the
1217 single specified device regardless of the device it was recorded from.
1219 .BI write_bw_log \fR=\fPstr
1220 If given, write a bandwidth log of the jobs in this job file. Can be used to
1221 store data of the bandwidth of the jobs in their lifetime. The included
1222 fio_generate_plots script uses gnuplot to turn these text files into nice
1223 graphs. See \fBwrite_lat_log\fR for behaviour of given filename. For this
1224 option, the postfix is _bw.x.log, where x is the index of the job (1..N,
1225 where N is the number of jobs)
1227 .BI write_lat_log \fR=\fPstr
1228 Same as \fBwrite_bw_log\fR, but writes I/O completion latencies. If no
1229 filename is given with this option, the default filename of
1230 "jobname_type.x.log" is used, where x is the index of the job (1..N, where
1231 N is the number of jobs). Even if the filename is given, fio will still
1232 append the type of log.
1234 .BI write_iops_log \fR=\fPstr
1235 Same as \fBwrite_bw_log\fR, but writes IOPS. If no filename is given with this
1236 option, the default filename of "jobname_type.x.log" is used, where x is the
1237 index of the job (1..N, where N is the number of jobs). Even if the filename
1238 is given, fio will still append the type of log.
1240 .BI log_avg_msec \fR=\fPint
1241 By default, fio will log an entry in the iops, latency, or bw log for every
1242 IO that completes. When writing to the disk log, that can quickly grow to a
1243 very large size. Setting this option makes fio average the each log entry
1244 over the specified period of time, reducing the resolution of the log.
1247 .BI log_offset \fR=\fPbool
1248 If this is set, the iolog options will include the byte offset for the IO
1249 entry as well as the other data values.
1251 .BI log_compression \fR=\fPint
1252 If this is set, fio will compress the IO logs as it goes, to keep the memory
1253 footprint lower. When a log reaches the specified size, that chunk is removed
1254 and compressed in the background. Given that IO logs are fairly highly
1255 compressible, this yields a nice memory savings for longer runs. The downside
1256 is that the compression will consume some background CPU cycles, so it may
1257 impact the run. This, however, is also true if the logging ends up consuming
1258 most of the system memory. So pick your poison. The IO logs are saved
1259 normally at the end of a run, by decompressing the chunks and storing them
1260 in the specified log file. This feature depends on the availability of zlib.
1262 .BI log_store_compressed \fR=\fPbool
1263 If set, and \fBlog\fR_compression is also set, fio will store the log files in
1264 a compressed format. They can be decompressed with fio, using the
1265 \fB\-\-inflate-log\fR command line parameter. The files will be stored with a
1268 .BI disable_lat \fR=\fPbool
1269 Disable measurements of total latency numbers. Useful only for cutting
1270 back the number of calls to \fBgettimeofday\fR\|(2), as that does impact performance at
1271 really high IOPS rates. Note that to really get rid of a large amount of these
1272 calls, this option must be used with disable_slat and disable_bw as well.
1274 .BI disable_clat \fR=\fPbool
1275 Disable measurements of completion latency numbers. See \fBdisable_lat\fR.
1277 .BI disable_slat \fR=\fPbool
1278 Disable measurements of submission latency numbers. See \fBdisable_lat\fR.
1280 .BI disable_bw_measurement \fR=\fPbool
1281 Disable measurements of throughput/bandwidth numbers. See \fBdisable_lat\fR.
1283 .BI lockmem \fR=\fPint
1284 Pin the specified amount of memory with \fBmlock\fR\|(2). Can be used to
1285 simulate a smaller amount of memory. The amount specified is per worker.
1287 .BI exec_prerun \fR=\fPstr
1288 Before running the job, execute the specified command with \fBsystem\fR\|(3).
1290 Output is redirected in a file called \fBjobname.prerun.txt\fR
1293 .BI exec_postrun \fR=\fPstr
1294 Same as \fBexec_prerun\fR, but the command is executed after the job completes.
1296 Output is redirected in a file called \fBjobname.postrun.txt\fR
1299 .BI ioscheduler \fR=\fPstr
1300 Attempt to switch the device hosting the file to the specified I/O scheduler.
1302 .BI disk_util \fR=\fPbool
1303 Generate disk utilization statistics if the platform supports it. Default: true.
1305 .BI clocksource \fR=\fPstr
1306 Use the given clocksource as the base of timing. The supported options are:
1310 \fBgettimeofday\fR\|(2)
1313 \fBclock_gettime\fR\|(2)
1316 Internal CPU clock source
1320 \fBcpu\fR is the preferred clocksource if it is reliable, as it is very fast
1321 (and fio is heavy on time calls). Fio will automatically use this clocksource
1322 if it's supported and considered reliable on the system it is running on,
1323 unless another clocksource is specifically set. For x86/x86-64 CPUs, this
1324 means supporting TSC Invariant.
1326 .BI gtod_reduce \fR=\fPbool
1327 Enable all of the \fBgettimeofday\fR\|(2) reducing options (disable_clat, disable_slat,
1328 disable_bw) plus reduce precision of the timeout somewhat to really shrink the
1329 \fBgettimeofday\fR\|(2) call count. With this option enabled, we only do about 0.4% of
1330 the gtod() calls we would have done if all time keeping was enabled.
1332 .BI gtod_cpu \fR=\fPint
1333 Sometimes it's cheaper to dedicate a single thread of execution to just getting
1334 the current time. Fio (and databases, for instance) are very intensive on
1335 \fBgettimeofday\fR\|(2) calls. With this option, you can set one CPU aside for doing
1336 nothing but logging current time to a shared memory location. Then the other
1337 threads/processes that run IO workloads need only copy that segment, instead of
1338 entering the kernel with a \fBgettimeofday\fR\|(2) call. The CPU set aside for doing
1339 these time calls will be excluded from other uses. Fio will manually clear it
1340 from the CPU mask of other jobs.
1342 .BI ignore_error \fR=\fPstr
1343 Sometimes you want to ignore some errors during test in that case you can specify
1344 error list for each error type.
1346 ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST
1348 errors for given error type is separated with ':'.
1349 Error may be symbol ('ENOSPC', 'ENOMEM') or an integer.
1351 Example: ignore_error=EAGAIN,ENOSPC:122 .
1353 This option will ignore EAGAIN from READ, and ENOSPC and 122(EDQUOT) from WRITE.
1355 .BI error_dump \fR=\fPbool
1356 If set dump every error even if it is non fatal, true by default. If disabled
1357 only fatal error will be dumped
1359 .BI profile \fR=\fPstr
1360 Select a specific builtin performance test.
1362 .BI cgroup \fR=\fPstr
1363 Add job to this control group. If it doesn't exist, it will be created.
1364 The system must have a mounted cgroup blkio mount point for this to work. If
1365 your system doesn't have it mounted, you can do so with:
1367 # mount \-t cgroup \-o blkio none /cgroup
1369 .BI cgroup_weight \fR=\fPint
1370 Set the weight of the cgroup to this value. See the documentation that comes
1371 with the kernel, allowed values are in the range of 100..1000.
1373 .BI cgroup_nodelete \fR=\fPbool
1374 Normally fio will delete the cgroups it has created after the job completion.
1375 To override this behavior and to leave cgroups around after the job completion,
1376 set cgroup_nodelete=1. This can be useful if one wants to inspect various
1377 cgroup files after job completion. Default: false
1380 Instead of running as the invoking user, set the user ID to this value before
1381 the thread/process does any work.
1384 Set group ID, see \fBuid\fR.
1386 .BI unit_base \fR=\fPint
1387 Base unit for reporting. Allowed values are:
1391 Use auto-detection (default).
1401 .BI flow_id \fR=\fPint
1402 The ID of the flow. If not specified, it defaults to being a global flow. See
1406 Weight in token-based flow control. If this value is used, then there is a
1407 \fBflow counter\fR which is used to regulate the proportion of activity between
1408 two or more jobs. fio attempts to keep this flow counter near zero. The
1409 \fBflow\fR parameter stands for how much should be added or subtracted to the
1410 flow counter on each iteration of the main I/O loop. That is, if one job has
1411 \fBflow=8\fR and another job has \fBflow=-1\fR, then there will be a roughly
1412 1:8 ratio in how much one runs vs the other.
1414 .BI flow_watermark \fR=\fPint
1415 The maximum value that the absolute value of the flow counter is allowed to
1416 reach before the job must wait for a lower value of the counter.
1418 .BI flow_sleep \fR=\fPint
1419 The period of time, in microseconds, to wait after the flow watermark has been
1420 exceeded before retrying operations
1422 .BI clat_percentiles \fR=\fPbool
1423 Enable the reporting of percentiles of completion latencies.
1425 .BI percentile_list \fR=\fPfloat_list
1426 Overwrite the default list of percentiles for completion
1427 latencies. Each number is a floating number in the range (0,100], and
1428 the maximum length of the list is 20. Use ':' to separate the
1429 numbers. For example, \-\-percentile_list=99.5:99.9 will cause fio to
1430 report the values of completion latency below which 99.5% and 99.9% of
1431 the observed latencies fell, respectively.
1432 .SS "Ioengine Parameters List"
1433 Some parameters are only valid when a specific ioengine is in use. These are
1434 used identically to normal parameters, with the caveat that when used on the
1435 command line, they must come after the ioengine.
1437 .BI (cpu)cpuload \fR=\fPint
1438 Attempt to use the specified percentage of CPU cycles.
1440 .BI (cpu)cpuchunks \fR=\fPint
1441 Split the load into cycles of the given time. In microseconds.
1443 .BI (cpu)exit_on_io_done \fR=\fPbool
1444 Detect when IO threads are done, then exit.
1446 .BI (libaio)userspace_reap
1447 Normally, with the libaio engine in use, fio will use
1448 the io_getevents system call to reap newly returned events.
1449 With this flag turned on, the AIO ring will be read directly
1450 from user-space to reap events. The reaping mode is only
1451 enabled when polling for a minimum of 0 events (eg when
1452 iodepth_batch_complete=0).
1454 .BI (net,netsplice)hostname \fR=\fPstr
1455 The host name or IP address to use for TCP or UDP based IO.
1456 If the job is a TCP listener or UDP reader, the hostname is not
1457 used and must be omitted unless it is a valid UDP multicast address.
1459 .BI (net,netsplice)port \fR=\fPint
1460 The TCP or UDP port to bind to or connect to.
1462 .BI (net,netsplice)interface \fR=\fPstr
1463 The IP address of the network interface used to send or receive UDP multicast
1466 .BI (net,netsplice)ttl \fR=\fPint
1467 Time-to-live value for outgoing UDP multicast packets. Default: 1
1469 .BI (net,netsplice)nodelay \fR=\fPbool
1470 Set TCP_NODELAY on TCP connections.
1472 .BI (net,netsplice)protocol \fR=\fPstr "\fR,\fP proto" \fR=\fPstr
1473 The network protocol to use. Accepted values are:
1478 Transmission control protocol
1481 Transmission control protocol V6
1484 User datagram protocol
1487 User datagram protocol V6
1493 When the protocol is TCP or UDP, the port must also be given,
1494 as well as the hostname if the job is a TCP listener or UDP
1495 reader. For unix sockets, the normal filename option should be
1496 used and the port is invalid.
1499 .BI (net,netsplice)listen
1500 For TCP network connections, tell fio to listen for incoming
1501 connections rather than initiating an outgoing connection. The
1502 hostname must be omitted if this option is used.
1504 .BI (net, pingpong) \fR=\fPbool
1505 Normally a network writer will just continue writing data, and a network reader
1506 will just consume packets. If pingpong=1 is set, a writer will send its normal
1507 payload to the reader, then wait for the reader to send the same payload back.
1508 This allows fio to measure network latencies. The submission and completion
1509 latencies then measure local time spent sending or receiving, and the
1510 completion latency measures how long it took for the other end to receive and
1511 send back. For UDP multicast traffic pingpong=1 should only be set for a single
1512 reader when multiple readers are listening to the same address.
1514 .BI (e4defrag,donorname) \fR=\fPstr
1515 File will be used as a block donor (swap extents between files)
1517 .BI (e4defrag,inplace) \fR=\fPint
1518 Configure donor file block allocation strategy
1521 Preallocate donor's file on init
1524 allocate space immediately inside defragment event, and free right after event
1527 .BI (rbd)rbdname \fR=\fPstr
1528 Specifies the name of the RBD.
1530 .BI (rbd)pool \fR=\fPstr
1531 Specifies the name of the Ceph pool containing the RBD.
1533 .BI (rbd)clientname \fR=\fPstr
1534 Specifies the username (without the 'client.' prefix) used to access the Ceph cluster.
1536 While running, \fBfio\fR will display the status of the created jobs. For
1540 Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
1543 The characters in the first set of brackets denote the current status of each
1544 threads. The possible values are:
1550 Setup but not started.
1556 Initialized, waiting.
1559 Running, doing sequential reads.
1562 Running, doing random reads.
1565 Running, doing sequential writes.
1568 Running, doing random writes.
1571 Running, doing mixed sequential reads/writes.
1574 Running, doing mixed random reads/writes.
1577 Running, currently waiting for \fBfsync\fR\|(2).
1580 Running, verifying written data.
1583 Exited, not reaped by main thread.
1586 Exited, thread reaped.
1590 The second set of brackets shows the estimated completion percentage of
1591 the current group. The third set shows the read and write I/O rate,
1592 respectively. Finally, the estimated run time of the job is displayed.
1594 When \fBfio\fR completes (or is interrupted by Ctrl-C), it will show data
1595 for each thread, each group of threads, and each disk, in that order.
1597 Per-thread statistics first show the threads client number, group-id, and
1598 error code. The remaining figures are as follows:
1602 Number of megabytes of I/O performed.
1605 Average data rate (bandwidth).
1611 Submission latency minimum, maximum, average and standard deviation. This is
1612 the time it took to submit the I/O.
1615 Completion latency minimum, maximum, average and standard deviation. This
1616 is the time between submission and completion.
1619 Bandwidth minimum, maximum, percentage of aggregate bandwidth received, average
1620 and standard deviation.
1623 CPU usage statistics. Includes user and system time, number of context switches
1624 this thread went through and number of major and minor page faults.
1627 Distribution of I/O depths. Each depth includes everything less than (or equal)
1628 to it, but greater than the previous depth.
1631 Number of read/write requests issued, and number of short read/write requests.
1634 Distribution of I/O completion latencies. The numbers follow the same pattern
1638 The group statistics show:
1643 Number of megabytes I/O performed.
1646 Aggregate bandwidth of threads in the group.
1649 Minimum average bandwidth a thread saw.
1652 Maximum average bandwidth a thread saw.
1655 Shortest runtime of threads in the group.
1658 Longest runtime of threads in the group.
1662 Finally, disk statistics are printed with reads first:
1667 Number of I/Os performed by all groups.
1670 Number of merges in the I/O scheduler.
1673 Number of ticks we kept the disk busy.
1676 Total time spent in the disk queue.
1683 It is also possible to get fio to dump the current output while it is
1684 running, without terminating the job. To do that, send fio the \fBUSR1\fR
1687 If the \fB\-\-minimal\fR / \fB\-\-append-terse\fR options are given, the
1688 results will be printed/appended in a semicolon-delimited format suitable for
1690 A job description (if provided) follows on a new line. Note that the first
1691 number in the line is the version number. If the output has to be changed
1692 for some reason, this number will be incremented by 1 to signify that
1693 change. The fields are:
1696 .B terse version, fio version, jobname, groupid, error
1700 .B Total I/O \fR(KB)\fP, bandwidth \fR(KB/s)\fP, IOPS, runtime \fR(ms)\fP
1704 .B min, max, mean, standard deviation
1708 .B min, max, mean, standard deviation
1710 Completion latency percentiles (20 fields):
1712 .B Xth percentile=usec
1716 .B min, max, mean, standard deviation
1720 .B min, max, aggregate percentage of total, mean, standard deviation
1726 .B Total I/O \fR(KB)\fP, bandwidth \fR(KB/s)\fP, IOPS, runtime \fR(ms)\fP
1730 .B min, max, mean, standard deviation
1734 .B min, max, mean, standard deviation
1736 Completion latency percentiles (20 fields):
1738 .B Xth percentile=usec
1742 .B min, max, mean, standard deviation
1746 .B min, max, aggregate percentage of total, mean, standard deviation
1752 .B user, system, context switches, major page faults, minor page faults
1755 IO depth distribution:
1757 .B <=1, 2, 4, 8, 16, 32, >=64
1760 IO latency distribution:
1764 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
1768 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
1772 Disk utilization (1 for each disk used):
1774 .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
1777 Error Info (dependent on continue_on_error, default off):
1779 .B total # errors, first error code
1782 .B text description (if provided in config - appears on newline)
1785 Normally you would run fio as a stand-alone application on the machine
1786 where the IO workload should be generated. However, it is also possible to
1787 run the frontend and backend of fio separately. This makes it possible to
1788 have a fio server running on the machine(s) where the IO workload should
1789 be running, while controlling it from another machine.
1791 To start the server, you would do:
1793 \fBfio \-\-server=args\fR
1795 on that machine, where args defines what fio listens to. The arguments
1796 are of the form 'type:hostname or IP:port'. 'type' is either 'ip' (or ip4)
1797 for TCP/IP v4, 'ip6' for TCP/IP v6, or 'sock' for a local unix domain
1798 socket. 'hostname' is either a hostname or IP address, and 'port' is the port to
1799 listen to (only valid for TCP/IP, not a local socket). Some examples:
1803 Start a fio server, listening on all interfaces on the default port (8765).
1805 2) fio \-\-server=ip:hostname,4444
1807 Start a fio server, listening on IP belonging to hostname and on port 4444.
1809 3) fio \-\-server=ip6:::1,4444
1811 Start a fio server, listening on IPv6 localhost ::1 and on port 4444.
1813 4) fio \-\-server=,4444
1815 Start a fio server, listening on all interfaces on port 4444.
1817 5) fio \-\-server=1.2.3.4
1819 Start a fio server, listening on IP 1.2.3.4 on the default port.
1821 6) fio \-\-server=sock:/tmp/fio.sock
1823 Start a fio server, listening on the local socket /tmp/fio.sock.
1825 When a server is running, you can connect to it from a client. The client
1828 fio \-\-local-args \-\-client=server \-\-remote-args <job file(s)>
1830 where \-\-local-args are arguments that are local to the client where it is
1831 running, 'server' is the connect string, and \-\-remote-args and <job file(s)>
1832 are sent to the server. The 'server' string follows the same format as it
1833 does on the server side, to allow IP/hostname/socket and port strings.
1834 You can connect to multiple clients as well, to do that you could run:
1836 fio \-\-client=server2 \-\-client=server2 <job file(s)>
1840 was written by Jens Axboe <jens.axboe@oracle.com>,
1841 now Jens Axboe <jaxboe@fusionio.com>.
1843 This man page was written by Aaron Carroll <aaronc@cse.unsw.edu.au> based
1844 on documentation by Jens Axboe.
1845 .SH "REPORTING BUGS"
1846 Report bugs to the \fBfio\fR mailing list <fio@vger.kernel.org>.
1849 For further documentation see \fBHOWTO\fR and \fBREADME\fR.
1851 Sample jobfiles are available in the \fBexamples\fR directory.
projects / fio.git / blob