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 bandwidth logs.
30 Print statistics in a terse, semicolon-delimited format.
33 Print statistics in selected mode AND 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. This option can be used multiple times to add 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.
116 Anywhere a numeric value is required, an arithmetic expression may be used,
117 provided it is surrounded by parentheses. Supported operators are:
125 .B multiplication (*)
131 .B exponentiation (^)
135 For time values in expressions, units are microseconds by default. This is
136 different than for time values not in expressions (not enclosed in
137 parentheses). The types used are:
140 String: a sequence of alphanumeric characters.
143 SI integer: a whole number, possibly containing a suffix denoting the base unit
144 of the value. Accepted suffixes are `k', 'M', 'G', 'T', and 'P', denoting
145 kilo (1024), mega (1024^2), giga (1024^3), tera (1024^4), and peta (1024^5)
146 respectively. If prefixed with '0x', the value is assumed to be base 16
147 (hexadecimal). A suffix may include a trailing 'b', for instance 'kb' is
148 identical to 'k'. You can specify a base 10 value by using 'KiB', 'MiB','GiB',
149 etc. This is useful for disk drives where values are often given in base 10
150 values. Specifying '30GiB' will get you 30*1000^3 bytes.
151 When specifying times the default suffix meaning changes, still denoting the
152 base unit of the value, but accepted suffixes are 'D' (days), 'H' (hours), 'M'
153 (minutes), 'S' Seconds, 'ms' (or msec) milli seconds, 'us' (or 'usec') micro
154 seconds. Time values without a unit specify seconds.
155 The suffixes are not case sensitive.
158 Boolean: a true or false value. `0' denotes false, `1' denotes true.
161 Integer range: a range of integers specified in the format
162 \fIlower\fR:\fIupper\fR or \fIlower\fR\-\fIupper\fR. \fIlower\fR and
163 \fIupper\fR may contain a suffix as described above. If an option allows two
164 sets of ranges, they are separated with a `,' or `/' character. For example:
168 List of floating numbers: A list of floating numbers, separated by
173 May be used to override the job name. On the command line, this parameter
174 has the special purpose of signalling the start of a new job.
176 .BI description \fR=\fPstr
177 Human-readable description of the job. It is printed when the job is run, but
178 otherwise has no special purpose.
180 .BI directory \fR=\fPstr
181 Prefix filenames with this directory. Used to place files in a location other
183 You can specify a number of directories by separating the names with a ':'
184 character. These directories will be assigned equally distributed to job clones
185 creates with \fInumjobs\fR as long as they are using generated filenames.
186 If specific \fIfilename(s)\fR are set fio will use the first listed directory,
187 and thereby matching the \fIfilename\fR semantic which generates a file each
188 clone if not specified, but let all clones use the same if set. See
189 \fIfilename\fR for considerations regarding escaping certain characters on
192 .BI filename \fR=\fPstr
194 normally makes up a file name based on the job name, thread number, and file
195 number. If you want to share files between threads in a job or several jobs,
196 specify a \fIfilename\fR for each of them to override the default.
197 If the I/O engine is file-based, you can specify
198 a number of files by separating the names with a `:' character. `\-' is a
199 reserved name, meaning stdin or stdout, depending on the read/write direction
200 set. On Windows, disk devices are accessed as \\.\PhysicalDrive0 for the first
201 device, \\.\PhysicalDrive1 for the second etc. Note: Windows and FreeBSD
202 prevent write access to areas of the disk containing in-use data
203 (e.g. filesystems). If the wanted filename does need to include a colon, then
204 escape that with a '\\' character. For instance, if the filename is
205 "/dev/dsk/foo@3,0:c", then you would use filename="/dev/dsk/foo@3,0\\:c".
207 .BI filename_format \fR=\fPstr
208 If sharing multiple files between jobs, it is usually necessary to have
209 fio generate the exact names that you want. By default, fio will name a file
210 based on the default file format specification of
211 \fBjobname.jobnumber.filenumber\fP. With this option, that can be
212 customized. Fio will recognize and replace the following keywords in this
218 The name of the worker thread or process.
221 The incremental number of the worker thread or process.
224 The incremental number of the file for that worker thread or process.
227 To have dependent jobs share a set of files, this option can be set to
228 have fio generate filenames that are shared between the two. For instance,
229 if \fBtestfiles.$filenum\fR is specified, file number 4 for any job will
230 be named \fBtestfiles.4\fR. The default of \fB$jobname.$jobnum.$filenum\fR
231 will be used if no other format specifier is given.
235 .BI lockfile \fR=\fPstr
236 Fio defaults to not locking any files before it does IO to them. If a file or
237 file descriptor is shared, fio can serialize IO to that file to make the end
238 result consistent. This is usual for emulating real workloads that share files.
244 No locking. This is the default.
247 Only one thread or process may do IO at a time, excluding all others.
250 Read-write locking on the file. Many readers may access the file at the same
251 time, but writes get exclusive access.
255 .BI opendir \fR=\fPstr
256 Recursively open any files below directory \fIstr\fR.
258 .BI readwrite \fR=\fPstr "\fR,\fP rw" \fR=\fPstr
259 Type of I/O pattern. Accepted values are:
270 Sequential trim (Linux block devices only).
279 Random trim (Linux block devices only).
282 Mixed sequential reads and writes.
285 Mixed random reads and writes.
288 For mixed I/O, the default split is 50/50. For certain types of io the result
289 may still be skewed a bit, since the speed may be different. It is possible to
290 specify a number of IO's to do before getting a new offset, this is done by
291 appending a `:\fI<nr>\fR to the end of the string given. For a random read, it
292 would look like \fBrw=randread:8\fR for passing in an offset modifier with a
293 value of 8. If the postfix is used with a sequential IO pattern, then the value
294 specified will be added to the generated offset for each IO. For instance,
295 using \fBrw=write:4k\fR will skip 4k for every write. It turns sequential IO
296 into sequential IO with holes. See the \fBrw_sequencer\fR option.
299 .BI rw_sequencer \fR=\fPstr
300 If an offset modifier is given by appending a number to the \fBrw=<str>\fR line,
301 then this option controls how that number modifies the IO offset being
302 generated. Accepted values are:
307 Generate sequential offset
310 Generate the same offset
313 \fBsequential\fR is only useful for random IO, where fio would normally
314 generate a new random offset for every IO. If you append eg 8 to randread, you
315 would get a new random offset for every 8 IO's. The result would be a seek for
316 only every 8 IO's, instead of for every IO. Use \fBrw=randread:8\fR to specify
317 that. As sequential IO is already sequential, setting \fBsequential\fR for that
318 would not result in any differences. \fBidentical\fR behaves in a similar
319 fashion, except it sends the same offset 8 number of times before generating a
324 .BI kb_base \fR=\fPint
325 The base unit for a kilobyte. The defacto base is 2^10, 1024. Storage
326 manufacturers like to use 10^3 or 1000 as a base ten unit instead, for obvious
327 reasons. Allowed values are 1024 or 1000, with 1024 being the default.
329 .BI unified_rw_reporting \fR=\fPbool
330 Fio normally reports statistics on a per data direction basis, meaning that
331 read, write, and trim are accounted and reported separately. If this option is
332 set fio sums the results and reports them as "mixed" instead.
334 .BI randrepeat \fR=\fPbool
335 Seed the random number generator used for random I/O patterns in a predictable
336 way so the pattern is repeatable across runs. Default: true.
338 .BI allrandrepeat \fR=\fPbool
339 Seed all random number generators in a predictable way so results are
340 repeatable across runs. Default: false.
342 .BI randseed \fR=\fPint
343 Seed the random number generators based on this seed value, to be able to
344 control what sequence of output is being generated. If not set, the random
345 sequence depends on the \fBrandrepeat\fR setting.
347 .BI fallocate \fR=\fPstr
348 Whether pre-allocation is performed when laying down files. Accepted values
354 Do not pre-allocate space.
357 Pre-allocate via \fBposix_fallocate\fR\|(3).
360 Pre-allocate via \fBfallocate\fR\|(2) with FALLOC_FL_KEEP_SIZE set.
363 Backward-compatible alias for 'none'.
366 Backward-compatible alias for 'posix'.
369 May not be available on all supported platforms. 'keep' is only
370 available on Linux. If using ZFS on Solaris this must be set to 'none'
371 because ZFS doesn't support it. Default: 'posix'.
374 .BI fadvise_hint \fR=\fPbool
375 Use \fBposix_fadvise\fR\|(2) to advise the kernel what I/O patterns
376 are likely to be issued. Default: true.
379 Total size of I/O for this job. \fBfio\fR will run until this many bytes have
380 been transferred, unless limited by other options (\fBruntime\fR, for instance).
381 Unless \fBnrfiles\fR and \fBfilesize\fR options are given, this amount will be
382 divided between the available files for the job. If not set, fio will use the
383 full size of the given files or devices. If the files do not exist, size
384 must be given. It is also possible to give size as a percentage between 1 and
385 100. If size=20% is given, fio will use 20% of the full size of the given
388 .BI io_limit \fR=\fPint
389 Normally fio operates within the region set by \fBsize\fR, which means that
390 the \fBsize\fR option sets both the region and size of IO to be performed.
391 Sometimes that is not what you want. With this option, it is possible to
392 define just the amount of IO that fio should do. For instance, if \fBsize\fR
393 is set to 20G and \fBio_limit\fR is set to 5G, fio will perform IO within
394 the first 20G but exit when 5G have been done.
396 .BI fill_device \fR=\fPbool "\fR,\fB fill_fs" \fR=\fPbool
397 Sets size to something really large and waits for ENOSPC (no space left on
398 device) as the terminating condition. Only makes sense with sequential write.
399 For a read workload, the mount point will be filled first then IO started on
400 the result. This option doesn't make sense if operating on a raw device node,
401 since the size of that is already known by the file system. Additionally,
402 writing beyond end-of-device will not return ENOSPC there.
404 .BI filesize \fR=\fPirange
405 Individual file sizes. May be a range, in which case \fBfio\fR will select sizes
406 for files at random within the given range, limited to \fBsize\fR in total (if
407 that is given). If \fBfilesize\fR is not specified, each created file is the
410 .BI file_append \fR=\fPbool
411 Perform IO after the end of the file. Normally fio will operate within the
412 size of a file. If this option is set, then fio will append to the file
413 instead. This has identical behavior to setting \fRoffset\fP to the size
414 of a file. This option is ignored on non-regular files.
416 .BI blocksize \fR=\fPint[,int] "\fR,\fB bs" \fR=\fPint[,int]
417 Block size for I/O units. Default: 4k. Values for reads, writes, and trims
418 can be specified separately in the format \fIread\fR,\fIwrite\fR,\fItrim\fR
419 either of which may be empty to leave that value at its default. If a trailing
420 comma isn't given, the remainder will inherit the last value set.
422 .BI blocksize_range \fR=\fPirange[,irange] "\fR,\fB bsrange" \fR=\fPirange[,irange]
423 Specify a range of I/O block sizes. The issued I/O unit will always be a
424 multiple of the minimum size, unless \fBblocksize_unaligned\fR is set. Applies
425 to both reads and writes if only one range is given, but can be specified
426 separately with a comma separating the values. Example: bsrange=1k-4k,2k-8k.
427 Also (see \fBblocksize\fR).
429 .BI bssplit \fR=\fPstr
430 This option allows even finer grained control of the block sizes issued,
431 not just even splits between them. With this option, you can weight various
432 block sizes for exact control of the issued IO for a job that has mixed
433 block sizes. The format of the option is bssplit=blocksize/percentage,
434 optionally adding as many definitions as needed separated by a colon.
435 Example: bssplit=4k/10:64k/50:32k/40 would issue 50% 64k blocks, 10% 4k
436 blocks and 40% 32k blocks. \fBbssplit\fR also supports giving separate
437 splits to reads and writes. The format is identical to what the
438 \fBbs\fR option accepts, the read and write parts are separated with a
441 .B blocksize_unaligned\fR,\fP bs_unaligned
442 If set, any size in \fBblocksize_range\fR may be used. This typically won't
443 work with direct I/O, as that normally requires sector alignment.
445 .BI blockalign \fR=\fPint[,int] "\fR,\fB ba" \fR=\fPint[,int]
446 At what boundary to align random IO offsets. Defaults to the same as 'blocksize'
447 the minimum blocksize given. Minimum alignment is typically 512b
448 for using direct IO, though it usually depends on the hardware block size.
449 This option is mutually exclusive with using a random map for files, so it
450 will turn off that option.
452 .BI bs_is_seq_rand \fR=\fPbool
453 If this option is set, fio will use the normal read,write blocksize settings as
454 sequential,random instead. Any random read or write will use the WRITE
455 blocksize settings, and any sequential read or write will use the READ
459 Initialize buffers with all zeros. Default: fill buffers with random data.
460 The resulting IO buffers will not be completely zeroed, unless
461 \fPscramble_buffers\fR is also turned off.
464 If this option is given, fio will refill the IO buffers on every submit. The
465 default is to only fill it at init time and reuse that data. Only makes sense
466 if zero_buffers isn't specified, naturally. If data verification is enabled,
467 refill_buffers is also automatically enabled.
469 .BI scramble_buffers \fR=\fPbool
470 If \fBrefill_buffers\fR is too costly and the target is using data
471 deduplication, then setting this option will slightly modify the IO buffer
472 contents to defeat normal de-dupe attempts. This is not enough to defeat
473 more clever block compression attempts, but it will stop naive dedupe
474 of blocks. Default: true.
476 .BI buffer_compress_percentage \fR=\fPint
477 If this is set, then fio will attempt to provide IO buffer content (on WRITEs)
478 that compress to the specified level. Fio does this by providing a mix of
479 random data and zeroes. Note that this is per block size unit, for file/disk
480 wide compression level that matches this setting, you'll also want to set
481 \fBrefill_buffers\fR.
483 .BI buffer_compress_chunk \fR=\fPint
484 See \fBbuffer_compress_percentage\fR. This setting allows fio to manage how
485 big the ranges of random data and zeroed data is. Without this set, fio will
486 provide \fBbuffer_compress_percentage\fR of blocksize random data, followed by
487 the remaining zeroed. With this set to some chunk size smaller than the block
488 size, fio can alternate random and zeroed data throughout the IO buffer.
490 .BI buffer_pattern \fR=\fPstr
491 If set, fio will fill the IO buffers with this pattern. If not set, the contents
492 of IO buffers is defined by the other options related to buffer contents. The
493 setting can be any pattern of bytes, and can be prefixed with 0x for hex
494 values. It may also be a string, where the string must then be wrapped with
497 .BI dedupe_percentage \fR=\fPint
498 If set, fio will generate this percentage of identical buffers when writing.
499 These buffers will be naturally dedupable. The contents of the buffers depend
500 on what other buffer compression settings have been set. It's possible to have
501 the individual buffers either fully compressible, or not at all. This option
502 only controls the distribution of unique buffers.
504 .BI nrfiles \fR=\fPint
505 Number of files to use for this job. Default: 1.
507 .BI openfiles \fR=\fPint
508 Number of files to keep open at the same time. Default: \fBnrfiles\fR.
510 .BI file_service_type \fR=\fPstr
511 Defines how files to service are selected. The following types are defined:
516 Choose a file at random.
519 Round robin over opened files (default).
522 Do each file in the set sequentially.
525 The number of I/Os to issue before switching to a new file can be specified by
526 appending `:\fIint\fR' to the service type.
529 .BI ioengine \fR=\fPstr
530 Defines how the job issues I/O. The following types are defined:
535 Basic \fBread\fR\|(2) or \fBwrite\fR\|(2) I/O. \fBfseek\fR\|(2) is used to
536 position the I/O location.
539 Basic \fBpread\fR\|(2) or \fBpwrite\fR\|(2) I/O.
542 Basic \fBreadv\fR\|(2) or \fBwritev\fR\|(2) I/O. Will emulate queuing by
543 coalescing adjacent IOs into a single submission.
546 Basic \fBpreadv\fR\|(2) or \fBpwritev\fR\|(2) I/O.
549 Linux native asynchronous I/O. This ioengine defines engine specific options.
552 POSIX asynchronous I/O using \fBaio_read\fR\|(3) and \fBaio_write\fR\|(3).
555 Solaris native asynchronous I/O.
558 Windows native asynchronous I/O.
561 File is memory mapped with \fBmmap\fR\|(2) and data copied using
565 \fBsplice\fR\|(2) is used to transfer the data and \fBvmsplice\fR\|(2) to
566 transfer data from user-space to the kernel.
569 Use the syslet system calls to make regular read/write asynchronous.
572 SCSI generic sg v3 I/O. May be either synchronous using the SG_IO ioctl, or if
573 the target is an sg character device, we use \fBread\fR\|(2) and
574 \fBwrite\fR\|(2) for asynchronous I/O.
577 Doesn't transfer any data, just pretends to. Mainly used to exercise \fBfio\fR
578 itself and for debugging and testing purposes.
581 Transfer over the network. The protocol to be used can be defined with the
582 \fBprotocol\fR parameter. Depending on the protocol, \fBfilename\fR,
583 \fBhostname\fR, \fBport\fR, or \fBlisten\fR must be specified.
584 This ioengine defines engine specific options.
587 Like \fBnet\fR, but uses \fBsplice\fR\|(2) and \fBvmsplice\fR\|(2) to map data
588 and send/receive. This ioengine defines engine specific options.
591 Doesn't transfer any data, but burns CPU cycles according to \fBcpuload\fR and
592 \fBcpucycles\fR parameters.
595 The GUASI I/O engine is the Generic Userspace Asynchronous Syscall Interface
596 approach to asynchronous I/O.
598 See <http://www.xmailserver.org/guasi\-lib.html>.
601 The RDMA I/O engine supports both RDMA memory semantics (RDMA_WRITE/RDMA_READ)
602 and channel semantics (Send/Recv) for the InfiniBand, RoCE and iWARP protocols.
605 Loads an external I/O engine object file. Append the engine filename as
609 IO engine that does regular linux native fallocate call to simulate data
610 transfer as fio ioengine
612 DDIR_READ does fallocate(,mode = FALLOC_FL_KEEP_SIZE,)
614 DIR_WRITE does fallocate(,mode = 0)
616 DDIR_TRIM does fallocate(,mode = FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE)
619 IO engine that does regular EXT4_IOC_MOVE_EXT ioctls to simulate defragment activity
620 request to DDIR_WRITE event
623 IO engine supporting direct access to Ceph Rados Block Devices (RBD) via librbd
624 without the need to use the kernel rbd driver. This ioengine defines engine specific
628 Using Glusterfs libgfapi sync interface to direct access to Glusterfs volumes without
629 having to go through FUSE. This ioengine defines engine specific
633 Using Glusterfs libgfapi async interface to direct access to Glusterfs volumes without
634 having to go through FUSE. This ioengine defines engine specific
638 Read and write through Hadoop (HDFS). The \fBfilename\fR option is used to
639 specify host,port of the hdfs name-node to connect. This engine interprets
640 offsets a little differently. In HDFS, files once created cannot be modified.
641 So random writes are not possible. To imitate this, libhdfs engine expects
642 bunch of small files to be created over HDFS, and engine will randomly pick a
643 file out of those files based on the offset generated by fio backend. (see the
644 example job file to create such files, use rw=write option). Please note, you
645 might want to set necessary environment variables to work with hdfs/libhdfs
651 .BI iodepth \fR=\fPint
652 Number of I/O units to keep in flight against the file. Note that increasing
653 iodepth beyond 1 will not affect synchronous ioengines (except for small
654 degress when verify_async is in use). Even async engines may impose OS
655 restrictions causing the desired depth not to be achieved. This may happen on
656 Linux when using libaio and not setting \fBdirect\fR=1, since buffered IO is
657 not async on that OS. Keep an eye on the IO depth distribution in the
658 fio output to verify that the achieved depth is as expected. Default: 1.
660 .BI iodepth_batch \fR=\fPint
661 Number of I/Os to submit at once. Default: \fBiodepth\fR.
663 .BI iodepth_batch_complete \fR=\fPint
664 This defines how many pieces of IO to retrieve at once. It defaults to 1 which
665 means that we'll ask for a minimum of 1 IO in the retrieval process from the
666 kernel. The IO retrieval will go on until we hit the limit set by
667 \fBiodepth_low\fR. If this variable is set to 0, then fio will always check for
668 completed events before queuing more IO. This helps reduce IO latency, at the
669 cost of more retrieval system calls.
671 .BI iodepth_low \fR=\fPint
672 Low watermark indicating when to start filling the queue again. Default:
675 .BI direct \fR=\fPbool
676 If true, use non-buffered I/O (usually O_DIRECT). Default: false.
678 .BI atomic \fR=\fPbool
679 If value is true, attempt to use atomic direct IO. Atomic writes are guaranteed
680 to be stable once acknowledged by the operating system. Only Linux supports
683 .BI buffered \fR=\fPbool
684 If true, use buffered I/O. This is the opposite of the \fBdirect\fR parameter.
687 .BI offset \fR=\fPint
688 Offset in the file to start I/O. Data before the offset will not be touched.
690 .BI offset_increment \fR=\fPint
691 If this is provided, then the real offset becomes the
692 offset + offset_increment * thread_number, where the thread number is a
693 counter that starts at 0 and is incremented for each sub-job (i.e. when
694 numjobs option is specified). This option is useful if there are several jobs
695 which are intended to operate on a file in parallel disjoint segments, with
696 even spacing between the starting points.
698 .BI number_ios \fR=\fPint
699 Fio will normally perform IOs until it has exhausted the size of the region
700 set by \fBsize\fR, or if it exhaust the allocated time (or hits an error
701 condition). With this setting, the range/size can be set independently of
702 the number of IOs to perform. When fio reaches this number, it will exit
703 normally and report status. Note that this does not extend the amount
704 of IO that will be done, it will only stop fio if this condition is met
705 before other end-of-job criteria.
708 How many I/Os to perform before issuing an \fBfsync\fR\|(2) of dirty data. If
709 0, don't sync. Default: 0.
711 .BI fdatasync \fR=\fPint
712 Like \fBfsync\fR, but uses \fBfdatasync\fR\|(2) instead to only sync the
713 data parts of the file. Default: 0.
715 .BI write_barrier \fR=\fPint
716 Make every Nth write a barrier write.
718 .BI sync_file_range \fR=\fPstr:int
719 Use \fBsync_file_range\fR\|(2) for every \fRval\fP number of write operations. Fio will
720 track range of writes that have happened since the last \fBsync_file_range\fR\|(2) call.
721 \fRstr\fP can currently be one or more of:
725 SYNC_FILE_RANGE_WAIT_BEFORE
728 SYNC_FILE_RANGE_WRITE
731 SYNC_FILE_RANGE_WRITE
735 So if you do sync_file_range=wait_before,write:8, fio would use
736 \fBSYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE\fP for every 8 writes.
737 Also see the \fBsync_file_range\fR\|(2) man page. This option is Linux specific.
739 .BI overwrite \fR=\fPbool
740 If writing, setup the file first and do overwrites. Default: false.
742 .BI end_fsync \fR=\fPbool
743 Sync file contents when a write stage has completed. Default: false.
745 .BI fsync_on_close \fR=\fPbool
746 If true, sync file contents on close. This differs from \fBend_fsync\fR in that
747 it will happen on every close, not just at the end of the job. Default: false.
749 .BI rwmixread \fR=\fPint
750 Percentage of a mixed workload that should be reads. Default: 50.
752 .BI rwmixwrite \fR=\fPint
753 Percentage of a mixed workload that should be writes. If \fBrwmixread\fR and
754 \fBrwmixwrite\fR are given and do not sum to 100%, the latter of the two
755 overrides the first. This may interfere with a given rate setting, if fio is
756 asked to limit reads or writes to a certain rate. If that is the case, then
757 the distribution may be skewed. Default: 50.
759 .BI random_distribution \fR=\fPstr:float
760 By default, fio will use a completely uniform random distribution when asked
761 to perform random IO. Sometimes it is useful to skew the distribution in
762 specific ways, ensuring that some parts of the data is more hot than others.
763 Fio includes the following distribution models:
767 Uniform random distribution
777 When using a zipf or pareto distribution, an input value is also needed to
778 define the access pattern. For zipf, this is the zipf theta. For pareto,
779 it's the pareto power. Fio includes a test program, genzipf, that can be
780 used visualize what the given input values will yield in terms of hit rates.
781 If you wanted to use zipf with a theta of 1.2, you would use
782 random_distribution=zipf:1.2 as the option. If a non-uniform model is used,
783 fio will disable use of the random map.
785 .BI percentage_random \fR=\fPint
786 For a random workload, set how big a percentage should be random. This defaults
787 to 100%, in which case the workload is fully random. It can be set from
788 anywhere from 0 to 100. Setting it to 0 would make the workload fully
789 sequential. It is possible to set different values for reads, writes, and
790 trim. To do so, simply use a comma separated list. See \fBblocksize\fR.
793 Normally \fBfio\fR will cover every block of the file when doing random I/O. If
794 this parameter is given, a new offset will be chosen without looking at past
795 I/O history. This parameter is mutually exclusive with \fBverify\fR.
797 .BI softrandommap \fR=\fPbool
798 See \fBnorandommap\fR. If fio runs with the random block map enabled and it
799 fails to allocate the map, if this option is set it will continue without a
800 random block map. As coverage will not be as complete as with random maps, this
801 option is disabled by default.
803 .BI random_generator \fR=\fPstr
804 Fio supports the following engines for generating IO offsets for random IO:
808 Strong 2^88 cycle random number generator
811 Linear feedback shift register generator
815 Tausworthe is a strong random number generator, but it requires tracking on the
816 side if we want to ensure that blocks are only read or written once. LFSR
817 guarantees that we never generate the same offset twice, and it's also less
818 computationally expensive. It's not a true random generator, however, though
819 for IO purposes it's typically good enough. LFSR only works with single block
820 sizes, not with workloads that use multiple block sizes. If used with such a
821 workload, fio may read or write some blocks multiple times.
824 Run job with given nice value. See \fBnice\fR\|(2).
827 Set I/O priority value of this job between 0 (highest) and 7 (lowest). See
830 .BI prioclass \fR=\fPint
831 Set I/O priority class. See \fBionice\fR\|(1).
833 .BI thinktime \fR=\fPint
834 Stall job for given number of microseconds between issuing I/Os.
836 .BI thinktime_spin \fR=\fPint
837 Pretend to spend CPU time for given number of microseconds, sleeping the rest
838 of the time specified by \fBthinktime\fR. Only valid if \fBthinktime\fR is set.
840 .BI thinktime_blocks \fR=\fPint
841 Only valid if thinktime is set - control how many blocks to issue, before
842 waiting \fBthinktime\fR microseconds. If not set, defaults to 1 which will
843 make fio wait \fBthinktime\fR microseconds after every block. This
844 effectively makes any queue depth setting redundant, since no more than 1 IO
845 will be queued before we have to complete it and do our thinktime. In other
846 words, this setting effectively caps the queue depth if the latter is larger.
850 Cap bandwidth used by this job. The number is in bytes/sec, the normal postfix
851 rules apply. You can use \fBrate\fR=500k to limit reads and writes to 500k each,
852 or you can specify read and writes separately. Using \fBrate\fR=1m,500k would
853 limit reads to 1MB/sec and writes to 500KB/sec. Capping only reads or writes
854 can be done with \fBrate\fR=,500k or \fBrate\fR=500k,. The former will only
855 limit writes (to 500KB/sec), the latter will only limit reads.
857 .BI ratemin \fR=\fPint
858 Tell \fBfio\fR to do whatever it can to maintain at least the given bandwidth.
859 Failing to meet this requirement will cause the job to exit. The same format
860 as \fBrate\fR is used for read vs write separation.
862 .BI rate_iops \fR=\fPint
863 Cap the bandwidth to this number of IOPS. Basically the same as rate, just
864 specified independently of bandwidth. The same format as \fBrate\fR is used for
865 read vs write separation. If \fBblocksize\fR is a range, the smallest block
866 size is used as the metric.
868 .BI rate_iops_min \fR=\fPint
869 If this rate of I/O is not met, the job will exit. The same format as \fBrate\fR
870 is used for read vs write separation.
872 .BI ratecycle \fR=\fPint
873 Average bandwidth for \fBrate\fR and \fBratemin\fR over this number of
874 milliseconds. Default: 1000ms.
876 .BI latency_target \fR=\fPint
877 If set, fio will attempt to find the max performance point that the given
878 workload will run at while maintaining a latency below this target. The
879 values is given in microseconds. See \fBlatency_window\fR and
880 \fBlatency_percentile\fR.
882 .BI latency_window \fR=\fPint
883 Used with \fBlatency_target\fR to specify the sample window that the job
884 is run at varying queue depths to test the performance. The value is given
887 .BI latency_percentile \fR=\fPfloat
888 The percentage of IOs that must fall within the criteria specified by
889 \fBlatency_target\fR and \fBlatency_window\fR. If not set, this defaults
890 to 100.0, meaning that all IOs must be equal or below to the value set
891 by \fBlatency_target\fR.
893 .BI max_latency \fR=\fPint
894 If set, fio will exit the job if it exceeds this maximum latency. It will exit
897 .BI cpumask \fR=\fPint
898 Set CPU affinity for this job. \fIint\fR is a bitmask of allowed CPUs the job
899 may run on. See \fBsched_setaffinity\fR\|(2).
901 .BI cpus_allowed \fR=\fPstr
902 Same as \fBcpumask\fR, but allows a comma-delimited list of CPU numbers.
904 .BI cpus_allowed_policy \fR=\fPstr
905 Set the policy of how fio distributes the CPUs specified by \fBcpus_allowed\fR
906 or \fBcpumask\fR. Two policies are supported:
911 All jobs will share the CPU set specified.
914 Each job will get a unique CPU from the CPU set.
917 \fBshared\fR is the default behaviour, if the option isn't specified. If
918 \fBsplit\fR is specified, then fio will assign one cpu per job. If not enough
919 CPUs are given for the jobs listed, then fio will roundrobin the CPUs in
924 .BI numa_cpu_nodes \fR=\fPstr
925 Set this job running on specified NUMA nodes' CPUs. The arguments allow
926 comma delimited list of cpu numbers, A-B ranges, or 'all'.
928 .BI numa_mem_policy \fR=\fPstr
929 Set this job's memory policy and corresponding NUMA nodes. Format of
933 .B <mode>[:<nodelist>]
936 is one of the following memory policy:
938 .B default, prefer, bind, interleave, local
941 For \fBdefault\fR and \fBlocal\fR memory policy, no \fBnodelist\fR is
942 needed to be specified. For \fBprefer\fR, only one node is
943 allowed. For \fBbind\fR and \fBinterleave\fR, \fBnodelist\fR allows
944 comma delimited list of numbers, A-B ranges, or 'all'.
946 .BI startdelay \fR=\fPirange
947 Delay start of job for the specified number of seconds. Supports all time
948 suffixes to allow specification of hours, minutes, seconds and
949 milliseconds - seconds are the default if a unit is omitted.
950 Can be given as a range which causes each thread to choose randomly out of the
953 .BI runtime \fR=\fPint
954 Terminate processing after the specified number of seconds.
957 If given, run for the specified \fBruntime\fR duration even if the files are
958 completely read or written. The same workload will be repeated as many times
959 as \fBruntime\fR allows.
961 .BI ramp_time \fR=\fPint
962 If set, fio will run the specified workload for this amount of time before
963 logging any performance numbers. Useful for letting performance settle before
964 logging results, thus minimizing the runtime required for stable results. Note
965 that the \fBramp_time\fR is considered lead in time for a job, thus it will
966 increase the total runtime if a special timeout or runtime is specified.
968 .BI invalidate \fR=\fPbool
969 Invalidate buffer-cache for the file prior to starting I/O. Default: true.
972 Use synchronous I/O for buffered writes. For the majority of I/O engines,
973 this means using O_SYNC. Default: false.
975 .BI iomem \fR=\fPstr "\fR,\fP mem" \fR=\fPstr
976 Allocation method for I/O unit buffer. Allowed values are:
981 Allocate memory with \fBmalloc\fR\|(3).
984 Use shared memory buffers allocated through \fBshmget\fR\|(2).
987 Same as \fBshm\fR, but use huge pages as backing.
990 Use \fBmmap\fR\|(2) for allocation. Uses anonymous memory unless a filename
991 is given after the option in the format `:\fIfile\fR'.
994 Same as \fBmmap\fR, but use huge files as backing.
997 The amount of memory allocated is the maximum allowed \fBblocksize\fR for the
998 job multiplied by \fBiodepth\fR. For \fBshmhuge\fR or \fBmmaphuge\fR to work,
999 the system must have free huge pages allocated. \fBmmaphuge\fR also needs to
1000 have hugetlbfs mounted, and \fIfile\fR must point there. At least on Linux,
1001 huge pages must be manually allocated. See \fB/proc/sys/vm/nr_hugehages\fR
1002 and the documentation for that. Normally you just need to echo an appropriate
1003 number, eg echoing 8 will ensure that the OS has 8 huge pages ready for
1007 .BI iomem_align \fR=\fPint "\fR,\fP mem_align" \fR=\fPint
1008 This indicates the memory alignment of the IO memory buffers. Note that the
1009 given alignment is applied to the first IO unit buffer, if using \fBiodepth\fR
1010 the alignment of the following buffers are given by the \fBbs\fR used. In
1011 other words, if using a \fBbs\fR that is a multiple of the page sized in the
1012 system, all buffers will be aligned to this value. If using a \fBbs\fR that
1013 is not page aligned, the alignment of subsequent IO memory buffers is the
1014 sum of the \fBiomem_align\fR and \fBbs\fR used.
1016 .BI hugepage\-size \fR=\fPint
1017 Defines the size of a huge page. Must be at least equal to the system setting.
1018 Should be a multiple of 1MB. Default: 4MB.
1021 Terminate all jobs when one finishes. Default: wait for each job to finish.
1023 .BI bwavgtime \fR=\fPint
1024 Average bandwidth calculations over the given time in milliseconds. Default:
1027 .BI iopsavgtime \fR=\fPint
1028 Average IOPS calculations over the given time in milliseconds. Default:
1031 .BI create_serialize \fR=\fPbool
1032 If true, serialize file creation for the jobs. Default: true.
1034 .BI create_fsync \fR=\fPbool
1035 \fBfsync\fR\|(2) data file after creation. Default: true.
1037 .BI create_on_open \fR=\fPbool
1038 If true, the files are not created until they are opened for IO by the job.
1040 .BI create_only \fR=\fPbool
1041 If true, fio will only run the setup phase of the job. If files need to be
1042 laid out or updated on disk, only that will be done. The actual job contents
1045 .BI pre_read \fR=\fPbool
1046 If this is given, files will be pre-read into memory before starting the given
1047 IO operation. This will also clear the \fR \fBinvalidate\fR flag, since it is
1048 pointless to pre-read and then drop the cache. This will only work for IO
1049 engines that are seekable, since they allow you to read the same data
1050 multiple times. Thus it will not work on eg network or splice IO.
1052 .BI unlink \fR=\fPbool
1053 Unlink job files when done. Default: false.
1055 .BI loops \fR=\fPint
1056 Specifies the number of iterations (runs of the same workload) of this job.
1059 .BI verify_only \fR=\fPbool
1060 Do not perform the specified workload, only verify data still matches previous
1061 invocation of this workload. This option allows one to check data multiple
1062 times at a later date without overwriting it. This option makes sense only for
1063 workloads that write data, and does not support workloads with the
1064 \fBtime_based\fR option set.
1066 .BI do_verify \fR=\fPbool
1067 Run the verify phase after a write phase. Only valid if \fBverify\fR is set.
1070 .BI verify \fR=\fPstr
1071 Method of verifying file contents after each iteration of the job. Allowed
1076 .B md5 crc16 crc32 crc32c crc32c-intel crc64 crc7 sha256 sha512 sha1 xxhash
1077 Store appropriate checksum in the header of each block. crc32c-intel is
1078 hardware accelerated SSE4.2 driven, falls back to regular crc32c if
1079 not supported by the system.
1082 Write extra information about each I/O (timestamp, block number, etc.). The
1083 block number is verified. See \fBverify_pattern\fR as well.
1086 Pretend to verify. Used for testing internals.
1089 This option can be used for repeated burn-in tests of a system to make sure
1090 that the written data is also correctly read back. If the data direction given
1091 is a read or random read, fio will assume that it should verify a previously
1092 written file. If the data direction includes any form of write, the verify will
1093 be of the newly written data.
1096 .BI verifysort \fR=\fPbool
1097 If true, written verify blocks are sorted if \fBfio\fR deems it to be faster to
1098 read them back in a sorted manner. Default: true.
1100 .BI verifysort_nr \fR=\fPint
1101 Pre-load and sort verify blocks for a read workload.
1103 .BI verify_offset \fR=\fPint
1104 Swap the verification header with data somewhere else in the block before
1105 writing. It is swapped back before verifying.
1107 .BI verify_interval \fR=\fPint
1108 Write the verification header for this number of bytes, which should divide
1109 \fBblocksize\fR. Default: \fBblocksize\fR.
1111 .BI verify_pattern \fR=\fPstr
1112 If set, fio will fill the io buffers with this pattern. Fio defaults to filling
1113 with totally random bytes, but sometimes it's interesting to fill with a known
1114 pattern for io verification purposes. Depending on the width of the pattern,
1115 fio will fill 1/2/3/4 bytes of the buffer at the time(it can be either a
1116 decimal or a hex number). The verify_pattern if larger than a 32-bit quantity
1117 has to be a hex number that starts with either "0x" or "0X". Use with
1120 .BI verify_fatal \fR=\fPbool
1121 If true, exit the job on the first observed verification failure. Default:
1124 .BI verify_dump \fR=\fPbool
1125 If set, dump the contents of both the original data block and the data block we
1126 read off disk to files. This allows later analysis to inspect just what kind of
1127 data corruption occurred. Off by default.
1129 .BI verify_async \fR=\fPint
1130 Fio will normally verify IO inline from the submitting thread. This option
1131 takes an integer describing how many async offload threads to create for IO
1132 verification instead, causing fio to offload the duty of verifying IO contents
1133 to one or more separate threads. If using this offload option, even sync IO
1134 engines can benefit from using an \fBiodepth\fR setting higher than 1, as it
1135 allows them to have IO in flight while verifies are running.
1137 .BI verify_async_cpus \fR=\fPstr
1138 Tell fio to set the given CPU affinity on the async IO verification threads.
1139 See \fBcpus_allowed\fP for the format used.
1141 .BI verify_backlog \fR=\fPint
1142 Fio will normally verify the written contents of a job that utilizes verify
1143 once that job has completed. In other words, everything is written then
1144 everything is read back and verified. You may want to verify continually
1145 instead for a variety of reasons. Fio stores the meta data associated with an
1146 IO block in memory, so for large verify workloads, quite a bit of memory would
1147 be used up holding this meta data. If this option is enabled, fio will write
1148 only N blocks before verifying these blocks.
1150 .BI verify_backlog_batch \fR=\fPint
1151 Control how many blocks fio will verify if verify_backlog is set. If not set,
1152 will default to the value of \fBverify_backlog\fR (meaning the entire queue is
1153 read back and verified). If \fBverify_backlog_batch\fR is less than
1154 \fBverify_backlog\fR then not all blocks will be verified, if
1155 \fBverify_backlog_batch\fR is larger than \fBverify_backlog\fR, some blocks
1156 will be verified more than once.
1158 .BI trim_percentage \fR=\fPint
1159 Number of verify blocks to discard/trim.
1161 .BI trim_verify_zero \fR=\fPbool
1162 Verify that trim/discarded blocks are returned as zeroes.
1164 .BI trim_backlog \fR=\fPint
1165 Trim after this number of blocks are written.
1167 .BI trim_backlog_batch \fR=\fPint
1168 Trim this number of IO blocks.
1170 .BI experimental_verify \fR=\fPbool
1171 Enable experimental verification.
1173 .BI verify_state_save \fR=\fPbool
1174 When a job exits during the write phase of a verify workload, save its
1175 current state. This allows fio to replay up until that point, if the
1176 verify state is loaded for the verify read phase.
1178 .BI verify_state_load \fR=\fPbool
1179 If a verify termination trigger was used, fio stores the current write
1180 state of each thread. This can be used at verification time so that fio
1181 knows how far it should verify. Without this information, fio will run
1182 a full verification pass, according to the settings in the job file used.
1184 .B stonewall "\fR,\fP wait_for_previous"
1185 Wait for preceding jobs in the job file to exit before starting this one.
1186 \fBstonewall\fR implies \fBnew_group\fR.
1189 Start a new reporting group. If not given, all jobs in a file will be part
1190 of the same reporting group, unless separated by a stonewall.
1192 .BI numjobs \fR=\fPint
1193 Number of clones (processes/threads performing the same workload) of this job.
1197 If set, display per-group reports instead of per-job when \fBnumjobs\fR is
1201 Use threads created with \fBpthread_create\fR\|(3) instead of processes created
1202 with \fBfork\fR\|(2).
1204 .BI zonesize \fR=\fPint
1205 Divide file into zones of the specified size in bytes. See \fBzoneskip\fR.
1207 .BI zonerange \fR=\fPint
1208 Give size of an IO zone. See \fBzoneskip\fR.
1210 .BI zoneskip \fR=\fPint
1211 Skip the specified number of bytes when \fBzonesize\fR bytes of data have been
1214 .BI write_iolog \fR=\fPstr
1215 Write the issued I/O patterns to the specified file. Specify a separate file
1216 for each job, otherwise the iologs will be interspersed and the file may be
1219 .BI read_iolog \fR=\fPstr
1220 Replay the I/O patterns contained in the specified file generated by
1221 \fBwrite_iolog\fR, or may be a \fBblktrace\fR binary file.
1223 .BI replay_no_stall \fR=\fPint
1224 While replaying I/O patterns using \fBread_iolog\fR the default behavior
1225 attempts to respect timing information between I/Os. Enabling
1226 \fBreplay_no_stall\fR causes I/Os to be replayed as fast as possible while
1227 still respecting ordering.
1229 .BI replay_redirect \fR=\fPstr
1230 While replaying I/O patterns using \fBread_iolog\fR the default behavior
1231 is to replay the IOPS onto the major/minor device that each IOP was recorded
1232 from. Setting \fBreplay_redirect\fR causes all IOPS to be replayed onto the
1233 single specified device regardless of the device it was recorded from.
1235 .BI write_bw_log \fR=\fPstr
1236 If given, write a bandwidth log of the jobs in this job file. Can be used to
1237 store data of the bandwidth of the jobs in their lifetime. The included
1238 fio_generate_plots script uses gnuplot to turn these text files into nice
1239 graphs. See \fBwrite_lat_log\fR for behaviour of given filename. For this
1240 option, the postfix is _bw.x.log, where x is the index of the job (1..N,
1241 where N is the number of jobs)
1243 .BI write_lat_log \fR=\fPstr
1244 Same as \fBwrite_bw_log\fR, but writes I/O completion latencies. If no
1245 filename is given with this option, the default filename of
1246 "jobname_type.x.log" is used, where x is the index of the job (1..N, where
1247 N is the number of jobs). Even if the filename is given, fio will still
1248 append the type of log.
1250 .BI write_iops_log \fR=\fPstr
1251 Same as \fBwrite_bw_log\fR, but writes IOPS. If no filename is given with this
1252 option, the default filename of "jobname_type.x.log" is used, where x is the
1253 index of the job (1..N, where N is the number of jobs). Even if the filename
1254 is given, fio will still append the type of log.
1256 .BI log_avg_msec \fR=\fPint
1257 By default, fio will log an entry in the iops, latency, or bw log for every
1258 IO that completes. When writing to the disk log, that can quickly grow to a
1259 very large size. Setting this option makes fio average the each log entry
1260 over the specified period of time, reducing the resolution of the log.
1263 .BI log_offset \fR=\fPbool
1264 If this is set, the iolog options will include the byte offset for the IO
1265 entry as well as the other data values.
1267 .BI log_compression \fR=\fPint
1268 If this is set, fio will compress the IO logs as it goes, to keep the memory
1269 footprint lower. When a log reaches the specified size, that chunk is removed
1270 and compressed in the background. Given that IO logs are fairly highly
1271 compressible, this yields a nice memory savings for longer runs. The downside
1272 is that the compression will consume some background CPU cycles, so it may
1273 impact the run. This, however, is also true if the logging ends up consuming
1274 most of the system memory. So pick your poison. The IO logs are saved
1275 normally at the end of a run, by decompressing the chunks and storing them
1276 in the specified log file. This feature depends on the availability of zlib.
1278 .BI log_store_compressed \fR=\fPbool
1279 If set, and \fBlog\fR_compression is also set, fio will store the log files in
1280 a compressed format. They can be decompressed with fio, using the
1281 \fB\-\-inflate-log\fR command line parameter. The files will be stored with a
1284 .BI disable_lat \fR=\fPbool
1285 Disable measurements of total latency numbers. Useful only for cutting
1286 back the number of calls to \fBgettimeofday\fR\|(2), as that does impact performance at
1287 really high IOPS rates. Note that to really get rid of a large amount of these
1288 calls, this option must be used with disable_slat and disable_bw as well.
1290 .BI disable_clat \fR=\fPbool
1291 Disable measurements of completion latency numbers. See \fBdisable_lat\fR.
1293 .BI disable_slat \fR=\fPbool
1294 Disable measurements of submission latency numbers. See \fBdisable_lat\fR.
1296 .BI disable_bw_measurement \fR=\fPbool
1297 Disable measurements of throughput/bandwidth numbers. See \fBdisable_lat\fR.
1299 .BI lockmem \fR=\fPint
1300 Pin the specified amount of memory with \fBmlock\fR\|(2). Can be used to
1301 simulate a smaller amount of memory. The amount specified is per worker.
1303 .BI exec_prerun \fR=\fPstr
1304 Before running the job, execute the specified command with \fBsystem\fR\|(3).
1306 Output is redirected in a file called \fBjobname.prerun.txt\fR
1309 .BI exec_postrun \fR=\fPstr
1310 Same as \fBexec_prerun\fR, but the command is executed after the job completes.
1312 Output is redirected in a file called \fBjobname.postrun.txt\fR
1315 .BI ioscheduler \fR=\fPstr
1316 Attempt to switch the device hosting the file to the specified I/O scheduler.
1318 .BI disk_util \fR=\fPbool
1319 Generate disk utilization statistics if the platform supports it. Default: true.
1321 .BI clocksource \fR=\fPstr
1322 Use the given clocksource as the base of timing. The supported options are:
1326 \fBgettimeofday\fR\|(2)
1329 \fBclock_gettime\fR\|(2)
1332 Internal CPU clock source
1336 \fBcpu\fR is the preferred clocksource if it is reliable, as it is very fast
1337 (and fio is heavy on time calls). Fio will automatically use this clocksource
1338 if it's supported and considered reliable on the system it is running on,
1339 unless another clocksource is specifically set. For x86/x86-64 CPUs, this
1340 means supporting TSC Invariant.
1342 .BI gtod_reduce \fR=\fPbool
1343 Enable all of the \fBgettimeofday\fR\|(2) reducing options (disable_clat, disable_slat,
1344 disable_bw) plus reduce precision of the timeout somewhat to really shrink the
1345 \fBgettimeofday\fR\|(2) call count. With this option enabled, we only do about 0.4% of
1346 the gtod() calls we would have done if all time keeping was enabled.
1348 .BI gtod_cpu \fR=\fPint
1349 Sometimes it's cheaper to dedicate a single thread of execution to just getting
1350 the current time. Fio (and databases, for instance) are very intensive on
1351 \fBgettimeofday\fR\|(2) calls. With this option, you can set one CPU aside for doing
1352 nothing but logging current time to a shared memory location. Then the other
1353 threads/processes that run IO workloads need only copy that segment, instead of
1354 entering the kernel with a \fBgettimeofday\fR\|(2) call. The CPU set aside for doing
1355 these time calls will be excluded from other uses. Fio will manually clear it
1356 from the CPU mask of other jobs.
1358 .BI ignore_error \fR=\fPstr
1359 Sometimes you want to ignore some errors during test in that case you can specify
1360 error list for each error type.
1362 ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST
1364 errors for given error type is separated with ':'.
1365 Error may be symbol ('ENOSPC', 'ENOMEM') or an integer.
1367 Example: ignore_error=EAGAIN,ENOSPC:122 .
1369 This option will ignore EAGAIN from READ, and ENOSPC and 122(EDQUOT) from WRITE.
1371 .BI error_dump \fR=\fPbool
1372 If set dump every error even if it is non fatal, true by default. If disabled
1373 only fatal error will be dumped
1375 .BI profile \fR=\fPstr
1376 Select a specific builtin performance test.
1378 .BI cgroup \fR=\fPstr
1379 Add job to this control group. If it doesn't exist, it will be created.
1380 The system must have a mounted cgroup blkio mount point for this to work. If
1381 your system doesn't have it mounted, you can do so with:
1383 # mount \-t cgroup \-o blkio none /cgroup
1385 .BI cgroup_weight \fR=\fPint
1386 Set the weight of the cgroup to this value. See the documentation that comes
1387 with the kernel, allowed values are in the range of 100..1000.
1389 .BI cgroup_nodelete \fR=\fPbool
1390 Normally fio will delete the cgroups it has created after the job completion.
1391 To override this behavior and to leave cgroups around after the job completion,
1392 set cgroup_nodelete=1. This can be useful if one wants to inspect various
1393 cgroup files after job completion. Default: false
1396 Instead of running as the invoking user, set the user ID to this value before
1397 the thread/process does any work.
1400 Set group ID, see \fBuid\fR.
1402 .BI unit_base \fR=\fPint
1403 Base unit for reporting. Allowed values are:
1407 Use auto-detection (default).
1417 .BI flow_id \fR=\fPint
1418 The ID of the flow. If not specified, it defaults to being a global flow. See
1422 Weight in token-based flow control. If this value is used, then there is a
1423 \fBflow counter\fR which is used to regulate the proportion of activity between
1424 two or more jobs. fio attempts to keep this flow counter near zero. The
1425 \fBflow\fR parameter stands for how much should be added or subtracted to the
1426 flow counter on each iteration of the main I/O loop. That is, if one job has
1427 \fBflow=8\fR and another job has \fBflow=-1\fR, then there will be a roughly
1428 1:8 ratio in how much one runs vs the other.
1430 .BI flow_watermark \fR=\fPint
1431 The maximum value that the absolute value of the flow counter is allowed to
1432 reach before the job must wait for a lower value of the counter.
1434 .BI flow_sleep \fR=\fPint
1435 The period of time, in microseconds, to wait after the flow watermark has been
1436 exceeded before retrying operations
1438 .BI clat_percentiles \fR=\fPbool
1439 Enable the reporting of percentiles of completion latencies.
1441 .BI percentile_list \fR=\fPfloat_list
1442 Overwrite the default list of percentiles for completion
1443 latencies. Each number is a floating number in the range (0,100], and
1444 the maximum length of the list is 20. Use ':' to separate the
1445 numbers. For example, \-\-percentile_list=99.5:99.9 will cause fio to
1446 report the values of completion latency below which 99.5% and 99.9% of
1447 the observed latencies fell, respectively.
1448 .SS "Ioengine Parameters List"
1449 Some parameters are only valid when a specific ioengine is in use. These are
1450 used identically to normal parameters, with the caveat that when used on the
1451 command line, they must come after the ioengine.
1453 .BI (cpu)cpuload \fR=\fPint
1454 Attempt to use the specified percentage of CPU cycles.
1456 .BI (cpu)cpuchunks \fR=\fPint
1457 Split the load into cycles of the given time. In microseconds.
1459 .BI (cpu)exit_on_io_done \fR=\fPbool
1460 Detect when IO threads are done, then exit.
1462 .BI (libaio)userspace_reap
1463 Normally, with the libaio engine in use, fio will use
1464 the io_getevents system call to reap newly returned events.
1465 With this flag turned on, the AIO ring will be read directly
1466 from user-space to reap events. The reaping mode is only
1467 enabled when polling for a minimum of 0 events (eg when
1468 iodepth_batch_complete=0).
1470 .BI (net,netsplice)hostname \fR=\fPstr
1471 The host name or IP address to use for TCP or UDP based IO.
1472 If the job is a TCP listener or UDP reader, the hostname is not
1473 used and must be omitted unless it is a valid UDP multicast address.
1475 .BI (net,netsplice)port \fR=\fPint
1476 The TCP or UDP port to bind to or connect to. If this is used with
1477 \fBnumjobs\fR to spawn multiple instances of the same job type, then
1478 this will be the starting port number since fio will use a range of ports.
1480 .BI (net,netsplice)interface \fR=\fPstr
1481 The IP address of the network interface used to send or receive UDP multicast
1484 .BI (net,netsplice)ttl \fR=\fPint
1485 Time-to-live value for outgoing UDP multicast packets. Default: 1
1487 .BI (net,netsplice)nodelay \fR=\fPbool
1488 Set TCP_NODELAY on TCP connections.
1490 .BI (net,netsplice)protocol \fR=\fPstr "\fR,\fP proto" \fR=\fPstr
1491 The network protocol to use. Accepted values are:
1496 Transmission control protocol
1499 Transmission control protocol V6
1502 User datagram protocol
1505 User datagram protocol V6
1511 When the protocol is TCP or UDP, the port must also be given,
1512 as well as the hostname if the job is a TCP listener or UDP
1513 reader. For unix sockets, the normal filename option should be
1514 used and the port is invalid.
1517 .BI (net,netsplice)listen
1518 For TCP network connections, tell fio to listen for incoming
1519 connections rather than initiating an outgoing connection. The
1520 hostname must be omitted if this option is used.
1522 .BI (net, pingpong) \fR=\fPbool
1523 Normally a network writer will just continue writing data, and a network reader
1524 will just consume packets. If pingpong=1 is set, a writer will send its normal
1525 payload to the reader, then wait for the reader to send the same payload back.
1526 This allows fio to measure network latencies. The submission and completion
1527 latencies then measure local time spent sending or receiving, and the
1528 completion latency measures how long it took for the other end to receive and
1529 send back. For UDP multicast traffic pingpong=1 should only be set for a single
1530 reader when multiple readers are listening to the same address.
1532 .BI (net, window_size) \fR=\fPint
1533 Set the desired socket buffer size for the connection.
1535 .BI (net, mss) \fR=\fPint
1536 Set the TCP maximum segment size (TCP_MAXSEG).
1538 .BI (e4defrag,donorname) \fR=\fPstr
1539 File will be used as a block donor (swap extents between files)
1541 .BI (e4defrag,inplace) \fR=\fPint
1542 Configure donor file block allocation strategy
1545 Preallocate donor's file on init
1548 allocate space immediately inside defragment event, and free right after event
1551 .BI (rbd)rbdname \fR=\fPstr
1552 Specifies the name of the RBD.
1554 .BI (rbd)pool \fR=\fPstr
1555 Specifies the name of the Ceph pool containing the RBD.
1557 .BI (rbd)clientname \fR=\fPstr
1558 Specifies the username (without the 'client.' prefix) used to access the Ceph cluster.
1560 While running, \fBfio\fR will display the status of the created jobs. For
1564 Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
1567 The characters in the first set of brackets denote the current status of each
1568 threads. The possible values are:
1574 Setup but not started.
1580 Initialized, waiting.
1583 Running, doing sequential reads.
1586 Running, doing random reads.
1589 Running, doing sequential writes.
1592 Running, doing random writes.
1595 Running, doing mixed sequential reads/writes.
1598 Running, doing mixed random reads/writes.
1601 Running, currently waiting for \fBfsync\fR\|(2).
1604 Running, verifying written data.
1607 Exited, not reaped by main thread.
1610 Exited, thread reaped.
1614 The second set of brackets shows the estimated completion percentage of
1615 the current group. The third set shows the read and write I/O rate,
1616 respectively. Finally, the estimated run time of the job is displayed.
1618 When \fBfio\fR completes (or is interrupted by Ctrl-C), it will show data
1619 for each thread, each group of threads, and each disk, in that order.
1621 Per-thread statistics first show the threads client number, group-id, and
1622 error code. The remaining figures are as follows:
1626 Number of megabytes of I/O performed.
1629 Average data rate (bandwidth).
1635 Submission latency minimum, maximum, average and standard deviation. This is
1636 the time it took to submit the I/O.
1639 Completion latency minimum, maximum, average and standard deviation. This
1640 is the time between submission and completion.
1643 Bandwidth minimum, maximum, percentage of aggregate bandwidth received, average
1644 and standard deviation.
1647 CPU usage statistics. Includes user and system time, number of context switches
1648 this thread went through and number of major and minor page faults.
1651 Distribution of I/O depths. Each depth includes everything less than (or equal)
1652 to it, but greater than the previous depth.
1655 Number of read/write requests issued, and number of short read/write requests.
1658 Distribution of I/O completion latencies. The numbers follow the same pattern
1662 The group statistics show:
1667 Number of megabytes I/O performed.
1670 Aggregate bandwidth of threads in the group.
1673 Minimum average bandwidth a thread saw.
1676 Maximum average bandwidth a thread saw.
1679 Shortest runtime of threads in the group.
1682 Longest runtime of threads in the group.
1686 Finally, disk statistics are printed with reads first:
1691 Number of I/Os performed by all groups.
1694 Number of merges in the I/O scheduler.
1697 Number of ticks we kept the disk busy.
1700 Total time spent in the disk queue.
1707 It is also possible to get fio to dump the current output while it is
1708 running, without terminating the job. To do that, send fio the \fBUSR1\fR
1711 If the \fB\-\-minimal\fR / \fB\-\-append-terse\fR options are given, the
1712 results will be printed/appended in a semicolon-delimited format suitable for
1714 A job description (if provided) follows on a new line. Note that the first
1715 number in the line is the version number. If the output has to be changed
1716 for some reason, this number will be incremented by 1 to signify that
1717 change. The fields are:
1720 .B terse version, fio version, jobname, groupid, error
1724 .B Total I/O \fR(KB)\fP, bandwidth \fR(KB/s)\fP, IOPS, runtime \fR(ms)\fP
1728 .B min, max, mean, standard deviation
1732 .B min, max, mean, standard deviation
1734 Completion latency percentiles (20 fields):
1736 .B Xth percentile=usec
1740 .B min, max, mean, standard deviation
1744 .B min, max, aggregate percentage of total, mean, standard deviation
1750 .B Total I/O \fR(KB)\fP, bandwidth \fR(KB/s)\fP, IOPS, runtime \fR(ms)\fP
1754 .B min, max, mean, standard deviation
1758 .B min, max, mean, standard deviation
1760 Completion latency percentiles (20 fields):
1762 .B Xth percentile=usec
1766 .B min, max, mean, standard deviation
1770 .B min, max, aggregate percentage of total, mean, standard deviation
1776 .B user, system, context switches, major page faults, minor page faults
1779 IO depth distribution:
1781 .B <=1, 2, 4, 8, 16, 32, >=64
1784 IO latency distribution:
1788 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
1792 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
1796 Disk utilization (1 for each disk used):
1798 .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
1801 Error Info (dependent on continue_on_error, default off):
1803 .B total # errors, first error code
1806 .B text description (if provided in config - appears on newline)
1809 Normally you would run fio as a stand-alone application on the machine
1810 where the IO workload should be generated. However, it is also possible to
1811 run the frontend and backend of fio separately. This makes it possible to
1812 have a fio server running on the machine(s) where the IO workload should
1813 be running, while controlling it from another machine.
1815 To start the server, you would do:
1817 \fBfio \-\-server=args\fR
1819 on that machine, where args defines what fio listens to. The arguments
1820 are of the form 'type:hostname or IP:port'. 'type' is either 'ip' (or ip4)
1821 for TCP/IP v4, 'ip6' for TCP/IP v6, or 'sock' for a local unix domain
1822 socket. 'hostname' is either a hostname or IP address, and 'port' is the port to
1823 listen to (only valid for TCP/IP, not a local socket). Some examples:
1827 Start a fio server, listening on all interfaces on the default port (8765).
1829 2) fio \-\-server=ip:hostname,4444
1831 Start a fio server, listening on IP belonging to hostname and on port 4444.
1833 3) fio \-\-server=ip6:::1,4444
1835 Start a fio server, listening on IPv6 localhost ::1 and on port 4444.
1837 4) fio \-\-server=,4444
1839 Start a fio server, listening on all interfaces on port 4444.
1841 5) fio \-\-server=1.2.3.4
1843 Start a fio server, listening on IP 1.2.3.4 on the default port.
1845 6) fio \-\-server=sock:/tmp/fio.sock
1847 Start a fio server, listening on the local socket /tmp/fio.sock.
1849 When a server is running, you can connect to it from a client. The client
1852 fio \-\-local-args \-\-client=server \-\-remote-args <job file(s)>
1854 where \-\-local-args are arguments that are local to the client where it is
1855 running, 'server' is the connect string, and \-\-remote-args and <job file(s)>
1856 are sent to the server. The 'server' string follows the same format as it
1857 does on the server side, to allow IP/hostname/socket and port strings.
1858 You can connect to multiple clients as well, to do that you could run:
1860 fio \-\-client=server2 \-\-client=server2 <job file(s)>
1862 If the job file is located on the fio server, then you can tell the server
1863 to load a local file as well. This is done by using \-\-remote-config:
1865 fio \-\-client=server \-\-remote-config /path/to/file.fio
1867 Then the fio serer will open this local (to the server) job file instead
1868 of being passed one from the client.
1872 was written by Jens Axboe <jens.axboe@oracle.com>,
1873 now Jens Axboe <jaxboe@fusionio.com>.
1875 This man page was written by Aaron Carroll <aaronc@cse.unsw.edu.au> based
1876 on documentation by Jens Axboe.
1877 .SH "REPORTING BUGS"
1878 Report bugs to the \fBfio\fR mailing list <fio@vger.kernel.org>.
1881 For further documentation see \fBHOWTO\fR and \fBREADME\fR.
1883 Sample jobfiles are available in the \fBexamples\fR directory.
projects / fio.git / blob