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. Suffixes currently do not
121 work with arithmetic expressions. Supported operators are
122 addition, subtraction, multiplication and division.
126 String: a sequence of alphanumeric characters.
129 SI integer: a whole number, possibly containing a suffix denoting the base unit
130 of the value. Accepted suffixes are `k', 'M', 'G', 'T', and 'P', denoting
131 kilo (1024), mega (1024^2), giga (1024^3), tera (1024^4), and peta (1024^5)
132 respectively. If prefixed with '0x', the value is assumed to be base 16
133 (hexadecimal). A suffix may include a trailing 'b', for instance 'kb' is
134 identical to 'k'. You can specify a base 10 value by using 'KiB', 'MiB','GiB',
135 etc. This is useful for disk drives where values are often given in base 10
136 values. Specifying '30GiB' will get you 30*1000^3 bytes.
137 When specifying times the default suffix meaning changes, still denoting the
138 base unit of the value, but accepted suffixes are 'D' (days), 'H' (hours), 'M'
139 (minutes), 'S' Seconds, 'ms' (or msec) milli seconds, 'us' (or 'usec') micro
140 seconds. Time values without a unit specify seconds.
141 The suffixes are not case sensitive.
144 Boolean: a true or false value. `0' denotes false, `1' denotes true.
147 Integer range: a range of integers specified in the format
148 \fIlower\fR:\fIupper\fR or \fIlower\fR\-\fIupper\fR. \fIlower\fR and
149 \fIupper\fR may contain a suffix as described above. If an option allows two
150 sets of ranges, they are separated with a `,' or `/' character. For example:
154 List of floating numbers: A list of floating numbers, separated by
159 May be used to override the job name. On the command line, this parameter
160 has the special purpose of signalling the start of a new job.
162 .BI description \fR=\fPstr
163 Human-readable description of the job. It is printed when the job is run, but
164 otherwise has no special purpose.
166 .BI directory \fR=\fPstr
167 Prefix filenames with this directory. Used to place files in a location other
169 You can specify a number of directories by separating the names with a ':'
170 character. These directories will be assigned equally distributed to job clones
171 creates with \fInumjobs\fR as long as they are using generated filenames.
172 If specific \fIfilename(s)\fR are set fio will use the first listed directory,
173 and thereby matching the \fIfilename\fR semantic which generates a file each
174 clone if not specified, but let all clones use the same if set. See
175 \fIfilename\fR for considerations regarding escaping certain characters on
178 .BI filename \fR=\fPstr
180 normally makes up a file name based on the job name, thread number, and file
181 number. If you want to share files between threads in a job or several jobs,
182 specify a \fIfilename\fR for each of them to override the default.
183 If the I/O engine is file-based, you can specify
184 a number of files by separating the names with a `:' character. `\-' is a
185 reserved name, meaning stdin or stdout, depending on the read/write direction
186 set. On Windows, disk devices are accessed as \\.\PhysicalDrive0 for the first
187 device, \\.\PhysicalDrive1 for the second etc. Note: Windows and FreeBSD
188 prevent write access to areas of the disk containing in-use data
189 (e.g. filesystems). If the wanted filename does need to include a colon, then
190 escape that with a '\\' character. For instance, if the filename is
191 "/dev/dsk/foo@3,0:c", then you would use filename="/dev/dsk/foo@3,0\\:c".
193 .BI filename_format \fR=\fPstr
194 If sharing multiple files between jobs, it is usually necessary to have
195 fio generate the exact names that you want. By default, fio will name a file
196 based on the default file format specification of
197 \fBjobname.jobnumber.filenumber\fP. With this option, that can be
198 customized. Fio will recognize and replace the following keywords in this
204 The name of the worker thread or process.
207 The incremental number of the worker thread or process.
210 The incremental number of the file for that worker thread or process.
213 To have dependent jobs share a set of files, this option can be set to
214 have fio generate filenames that are shared between the two. For instance,
215 if \fBtestfiles.$filenum\fR is specified, file number 4 for any job will
216 be named \fBtestfiles.4\fR. The default of \fB$jobname.$jobnum.$filenum\fR
217 will be used if no other format specifier is given.
221 .BI lockfile \fR=\fPstr
222 Fio defaults to not locking any files before it does IO to them. If a file or
223 file descriptor is shared, fio can serialize IO to that file to make the end
224 result consistent. This is usual for emulating real workloads that share files.
230 No locking. This is the default.
233 Only one thread or process may do IO at a time, excluding all others.
236 Read-write locking on the file. Many readers may access the file at the same
237 time, but writes get exclusive access.
241 .BI opendir \fR=\fPstr
242 Recursively open any files below directory \fIstr\fR.
244 .BI readwrite \fR=\fPstr "\fR,\fP rw" \fR=\fPstr
245 Type of I/O pattern. Accepted values are:
256 Sequential trim (Linux block devices only).
265 Random trim (Linux block devices only).
268 Mixed sequential reads and writes.
271 Mixed random reads and writes.
274 For mixed I/O, the default split is 50/50. For certain types of io the result
275 may still be skewed a bit, since the speed may be different. It is possible to
276 specify a number of IO's to do before getting a new offset, this is done by
277 appending a `:\fI<nr>\fR to the end of the string given. For a random read, it
278 would look like \fBrw=randread:8\fR for passing in an offset modifier with a
279 value of 8. If the postfix is used with a sequential IO pattern, then the value
280 specified will be added to the generated offset for each IO. For instance,
281 using \fBrw=write:4k\fR will skip 4k for every write. It turns sequential IO
282 into sequential IO with holes. See the \fBrw_sequencer\fR option.
285 .BI rw_sequencer \fR=\fPstr
286 If an offset modifier is given by appending a number to the \fBrw=<str>\fR line,
287 then this option controls how that number modifies the IO offset being
288 generated. Accepted values are:
293 Generate sequential offset
296 Generate the same offset
299 \fBsequential\fR is only useful for random IO, where fio would normally
300 generate a new random offset for every IO. If you append eg 8 to randread, you
301 would get a new random offset for every 8 IO's. The result would be a seek for
302 only every 8 IO's, instead of for every IO. Use \fBrw=randread:8\fR to specify
303 that. As sequential IO is already sequential, setting \fBsequential\fR for that
304 would not result in any differences. \fBidentical\fR behaves in a similar
305 fashion, except it sends the same offset 8 number of times before generating a
310 .BI kb_base \fR=\fPint
311 The base unit for a kilobyte. The defacto base is 2^10, 1024. Storage
312 manufacturers like to use 10^3 or 1000 as a base ten unit instead, for obvious
313 reasons. Allowed values are 1024 or 1000, with 1024 being the default.
315 .BI unified_rw_reporting \fR=\fPbool
316 Fio normally reports statistics on a per data direction basis, meaning that
317 read, write, and trim are accounted and reported separately. If this option is
318 set fio sums the results and reports them as "mixed" instead.
320 .BI randrepeat \fR=\fPbool
321 Seed the random number generator used for random I/O patterns in a predictable
322 way so the pattern is repeatable across runs. Default: true.
324 .BI allrandrepeat \fR=\fPbool
325 Seed all random number generators in a predictable way so results are
326 repeatable across runs. Default: false.
328 .BI randseed \fR=\fPint
329 Seed the random number generators based on this seed value, to be able to
330 control what sequence of output is being generated. If not set, the random
331 sequence depends on the \fBrandrepeat\fR setting.
333 .BI use_os_rand \fR=\fPbool
334 Fio can either use the random generator supplied by the OS to generate random
335 offsets, or it can use its own internal generator (based on Tausworthe).
336 Default is to use the internal generator, which is often of better quality and
337 faster. Default: false.
339 .BI fallocate \fR=\fPstr
340 Whether pre-allocation is performed when laying down files. Accepted values
346 Do not pre-allocate space.
349 Pre-allocate via \fBposix_fallocate\fR\|(3).
352 Pre-allocate via \fBfallocate\fR\|(2) with FALLOC_FL_KEEP_SIZE set.
355 Backward-compatible alias for 'none'.
358 Backward-compatible alias for 'posix'.
361 May not be available on all supported platforms. 'keep' is only
362 available on Linux. If using ZFS on Solaris this must be set to 'none'
363 because ZFS doesn't support it. Default: 'posix'.
366 .BI fadvise_hint \fR=\fPbool
367 Use \fBposix_fadvise\fR\|(2) to advise the kernel what I/O patterns
368 are likely to be issued. Default: true.
371 Total size of I/O for this job. \fBfio\fR will run until this many bytes have
372 been transferred, unless limited by other options (\fBruntime\fR, for instance).
373 Unless \fBnrfiles\fR and \fBfilesize\fR options are given, this amount will be
374 divided between the available files for the job. If not set, fio will use the
375 full size of the given files or devices. If the files do not exist, size
376 must be given. It is also possible to give size as a percentage between 1 and
377 100. If size=20% is given, fio will use 20% of the full size of the given
380 .BI io_limit \fR=\fPint
381 Normally fio operates within the region set by \fBsize\fR, which means that
382 the \fBsize\fR option sets both the region and size of IO to be performed.
383 Sometimes that is not what you want. With this option, it is possible to
384 define just the amount of IO that fio should do. For instance, if \fBsize\fR
385 is set to 20G and \fBio_limit\fR is set to 5G, fio will perform IO within
386 the first 20G but exit when 5G have been done.
388 .BI fill_device \fR=\fPbool "\fR,\fB fill_fs" \fR=\fPbool
389 Sets size to something really large and waits for ENOSPC (no space left on
390 device) as the terminating condition. Only makes sense with sequential write.
391 For a read workload, the mount point will be filled first then IO started on
392 the result. This option doesn't make sense if operating on a raw device node,
393 since the size of that is already known by the file system. Additionally,
394 writing beyond end-of-device will not return ENOSPC there.
396 .BI filesize \fR=\fPirange
397 Individual file sizes. May be a range, in which case \fBfio\fR will select sizes
398 for files at random within the given range, limited to \fBsize\fR in total (if
399 that is given). If \fBfilesize\fR is not specified, each created file is the
402 .BI file_append \fR=\fPbool
403 Perform IO after the end of the file. Normally fio will operate within the
404 size of a file. If this option is set, then fio will append to the file
405 instead. This has identical behavior to setting \fRoffset\fP to the size
406 of a file. This option is ignored on non-regular files.
408 .BI blocksize \fR=\fPint[,int] "\fR,\fB bs" \fR=\fPint[,int]
409 Block size for I/O units. Default: 4k. Values for reads, writes, and trims
410 can be specified separately in the format \fIread\fR,\fIwrite\fR,\fItrim\fR
411 either of which may be empty to leave that value at its default. If a trailing
412 comma isn't given, the remainder will inherit the last value set.
414 .BI blocksize_range \fR=\fPirange[,irange] "\fR,\fB bsrange" \fR=\fPirange[,irange]
415 Specify a range of I/O block sizes. The issued I/O unit will always be a
416 multiple of the minimum size, unless \fBblocksize_unaligned\fR is set. Applies
417 to both reads and writes if only one range is given, but can be specified
418 separately with a comma separating the values. Example: bsrange=1k-4k,2k-8k.
419 Also (see \fBblocksize\fR).
421 .BI bssplit \fR=\fPstr
422 This option allows even finer grained control of the block sizes issued,
423 not just even splits between them. With this option, you can weight various
424 block sizes for exact control of the issued IO for a job that has mixed
425 block sizes. The format of the option is bssplit=blocksize/percentage,
426 optionally adding as many definitions as needed separated by a colon.
427 Example: bssplit=4k/10:64k/50:32k/40 would issue 50% 64k blocks, 10% 4k
428 blocks and 40% 32k blocks. \fBbssplit\fR also supports giving separate
429 splits to reads and writes. The format is identical to what the
430 \fBbs\fR option accepts, the read and write parts are separated with a
433 .B blocksize_unaligned\fR,\fP bs_unaligned
434 If set, any size in \fBblocksize_range\fR may be used. This typically won't
435 work with direct I/O, as that normally requires sector alignment.
437 .BI blockalign \fR=\fPint[,int] "\fR,\fB ba" \fR=\fPint[,int]
438 At what boundary to align random IO offsets. Defaults to the same as 'blocksize'
439 the minimum blocksize given. Minimum alignment is typically 512b
440 for using direct IO, though it usually depends on the hardware block size.
441 This option is mutually exclusive with using a random map for files, so it
442 will turn off that option.
444 .BI bs_is_seq_rand \fR=\fPbool
445 If this option is set, fio will use the normal read,write blocksize settings as
446 sequential,random instead. Any random read or write will use the WRITE
447 blocksize settings, and any sequential read or write will use the READ
451 Initialize buffers with all zeros. Default: fill buffers with random data.
452 The resulting IO buffers will not be completely zeroed, unless
453 \fPscramble_buffers\fR is also turned off.
456 If this option is given, fio will refill the IO buffers on every submit. The
457 default is to only fill it at init time and reuse that data. Only makes sense
458 if zero_buffers isn't specified, naturally. If data verification is enabled,
459 refill_buffers is also automatically enabled.
461 .BI scramble_buffers \fR=\fPbool
462 If \fBrefill_buffers\fR is too costly and the target is using data
463 deduplication, then setting this option will slightly modify the IO buffer
464 contents to defeat normal de-dupe attempts. This is not enough to defeat
465 more clever block compression attempts, but it will stop naive dedupe
466 of blocks. Default: true.
468 .BI buffer_compress_percentage \fR=\fPint
469 If this is set, then fio will attempt to provide IO buffer content (on WRITEs)
470 that compress to the specified level. Fio does this by providing a mix of
471 random data and zeroes. Note that this is per block size unit, for file/disk
472 wide compression level that matches this setting, you'll also want to set
473 \fBrefill_buffers\fR.
475 .BI buffer_compress_chunk \fR=\fPint
476 See \fBbuffer_compress_percentage\fR. This setting allows fio to manage how
477 big the ranges of random data and zeroed data is. Without this set, fio will
478 provide \fBbuffer_compress_percentage\fR of blocksize random data, followed by
479 the remaining zeroed. With this set to some chunk size smaller than the block
480 size, fio can alternate random and zeroed data throughout the IO buffer.
482 .BI buffer_pattern \fR=\fPstr
483 If set, fio will fill the IO buffers with this pattern. If not set, the contents
484 of IO buffers is defined by the other options related to buffer contents. The
485 setting can be any pattern of bytes, and can be prefixed with 0x for hex
486 values. It may also be a string, where the string must then be wrapped with
489 .BI dedupe_percentage \fR=\fPint
490 If set, fio will generate this percentage of identical buffers when writing.
491 These buffers will be naturally dedupable. The contents of the buffers depend
492 on what other buffer compression settings have been set. It's possible to have
493 the individual buffers either fully compressible, or not at all. This option
494 only controls the distribution of unique buffers.
496 .BI nrfiles \fR=\fPint
497 Number of files to use for this job. Default: 1.
499 .BI openfiles \fR=\fPint
500 Number of files to keep open at the same time. Default: \fBnrfiles\fR.
502 .BI file_service_type \fR=\fPstr
503 Defines how files to service are selected. The following types are defined:
508 Choose a file at random.
511 Round robin over opened files (default).
514 Do each file in the set sequentially.
517 The number of I/Os to issue before switching to a new file can be specified by
518 appending `:\fIint\fR' to the service type.
521 .BI ioengine \fR=\fPstr
522 Defines how the job issues I/O. The following types are defined:
527 Basic \fBread\fR\|(2) or \fBwrite\fR\|(2) I/O. \fBfseek\fR\|(2) is used to
528 position the I/O location.
531 Basic \fBpread\fR\|(2) or \fBpwrite\fR\|(2) I/O.
534 Basic \fBreadv\fR\|(2) or \fBwritev\fR\|(2) I/O. Will emulate queuing by
535 coalescing adjacent IOs into a single submission.
538 Basic \fBpreadv\fR\|(2) or \fBpwritev\fR\|(2) I/O.
541 Linux native asynchronous I/O. This ioengine defines engine specific options.
544 POSIX asynchronous I/O using \fBaio_read\fR\|(3) and \fBaio_write\fR\|(3).
547 Solaris native asynchronous I/O.
550 Windows native asynchronous I/O.
553 File is memory mapped with \fBmmap\fR\|(2) and data copied using
557 \fBsplice\fR\|(2) is used to transfer the data and \fBvmsplice\fR\|(2) to
558 transfer data from user-space to the kernel.
561 Use the syslet system calls to make regular read/write asynchronous.
564 SCSI generic sg v3 I/O. May be either synchronous using the SG_IO ioctl, or if
565 the target is an sg character device, we use \fBread\fR\|(2) and
566 \fBwrite\fR\|(2) for asynchronous I/O.
569 Doesn't transfer any data, just pretends to. Mainly used to exercise \fBfio\fR
570 itself and for debugging and testing purposes.
573 Transfer over the network. The protocol to be used can be defined with the
574 \fBprotocol\fR parameter. Depending on the protocol, \fBfilename\fR,
575 \fBhostname\fR, \fBport\fR, or \fBlisten\fR must be specified.
576 This ioengine defines engine specific options.
579 Like \fBnet\fR, but uses \fBsplice\fR\|(2) and \fBvmsplice\fR\|(2) to map data
580 and send/receive. This ioengine defines engine specific options.
583 Doesn't transfer any data, but burns CPU cycles according to \fBcpuload\fR and
584 \fBcpucycles\fR parameters.
587 The GUASI I/O engine is the Generic Userspace Asynchronous Syscall Interface
588 approach to asynchronous I/O.
590 See <http://www.xmailserver.org/guasi\-lib.html>.
593 The RDMA I/O engine supports both RDMA memory semantics (RDMA_WRITE/RDMA_READ)
594 and channel semantics (Send/Recv) for the InfiniBand, RoCE and iWARP protocols.
597 Loads an external I/O engine object file. Append the engine filename as
601 IO engine that does regular linux native fallocate call to simulate data
602 transfer as fio ioengine
604 DDIR_READ does fallocate(,mode = FALLOC_FL_KEEP_SIZE,)
606 DIR_WRITE does fallocate(,mode = 0)
608 DDIR_TRIM does fallocate(,mode = FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE)
611 IO engine that does regular EXT4_IOC_MOVE_EXT ioctls to simulate defragment activity
612 request to DDIR_WRITE event
615 IO engine supporting direct access to Ceph Rados Block Devices (RBD) via librbd
616 without the need to use the kernel rbd driver. This ioengine defines engine specific
620 Using Glusterfs libgfapi sync interface to direct access to Glusterfs volumes without
621 having to go through FUSE. This ioengine defines engine specific
625 Using Glusterfs libgfapi async interface to direct access to Glusterfs volumes without
626 having to go through FUSE. This ioengine defines engine specific
630 Read and write through Hadoop (HDFS). The \fBfilename\fR option is used to
631 specify host,port of the hdfs name-node to connect. This engine interprets
632 offsets a little differently. In HDFS, files once created cannot be modified.
633 So random writes are not possible. To imitate this, libhdfs engine expects
634 bunch of small files to be created over HDFS, and engine will randomly pick a
635 file out of those files based on the offset generated by fio backend. (see the
636 example job file to create such files, use rw=write option). Please note, you
637 might want to set necessary environment variables to work with hdfs/libhdfs
643 .BI iodepth \fR=\fPint
644 Number of I/O units to keep in flight against the file. Note that increasing
645 iodepth beyond 1 will not affect synchronous ioengines (except for small
646 degress when verify_async is in use). Even async engines may impose OS
647 restrictions causing the desired depth not to be achieved. This may happen on
648 Linux when using libaio and not setting \fBdirect\fR=1, since buffered IO is
649 not async on that OS. Keep an eye on the IO depth distribution in the
650 fio output to verify that the achieved depth is as expected. Default: 1.
652 .BI iodepth_batch \fR=\fPint
653 Number of I/Os to submit at once. Default: \fBiodepth\fR.
655 .BI iodepth_batch_complete \fR=\fPint
656 This defines how many pieces of IO to retrieve at once. It defaults to 1 which
657 means that we'll ask for a minimum of 1 IO in the retrieval process from the
658 kernel. The IO retrieval will go on until we hit the limit set by
659 \fBiodepth_low\fR. If this variable is set to 0, then fio will always check for
660 completed events before queuing more IO. This helps reduce IO latency, at the
661 cost of more retrieval system calls.
663 .BI iodepth_low \fR=\fPint
664 Low watermark indicating when to start filling the queue again. Default:
667 .BI direct \fR=\fPbool
668 If true, use non-buffered I/O (usually O_DIRECT). Default: false.
670 .BI atomic \fR=\fPbool
671 If value is true, attempt to use atomic direct IO. Atomic writes are guaranteed
672 to be stable once acknowledged by the operating system. Only Linux supports
675 .BI buffered \fR=\fPbool
676 If true, use buffered I/O. This is the opposite of the \fBdirect\fR parameter.
679 .BI offset \fR=\fPint
680 Offset in the file to start I/O. Data before the offset will not be touched.
682 .BI offset_increment \fR=\fPint
683 If this is provided, then the real offset becomes the
684 offset + offset_increment * thread_number, where the thread number is a
685 counter that starts at 0 and is incremented for each sub-job (i.e. when
686 numjobs option is specified). This option is useful if there are several jobs
687 which are intended to operate on a file in parallel disjoint segments, with
688 even spacing between the starting points.
690 .BI number_ios \fR=\fPint
691 Fio will normally perform IOs until it has exhausted the size of the region
692 set by \fBsize\fR, or if it exhaust the allocated time (or hits an error
693 condition). With this setting, the range/size can be set independently of
694 the number of IOs to perform. When fio reaches this number, it will exit
695 normally and report status. Note that this does not extend the amount
696 of IO that will be done, it will only stop fio if this condition is met
697 before other end-of-job criteria.
700 How many I/Os to perform before issuing an \fBfsync\fR\|(2) of dirty data. If
701 0, don't sync. Default: 0.
703 .BI fdatasync \fR=\fPint
704 Like \fBfsync\fR, but uses \fBfdatasync\fR\|(2) instead to only sync the
705 data parts of the file. Default: 0.
707 .BI write_barrier \fR=\fPint
708 Make every Nth write a barrier write.
710 .BI sync_file_range \fR=\fPstr:int
711 Use \fBsync_file_range\fR\|(2) for every \fRval\fP number of write operations. Fio will
712 track range of writes that have happened since the last \fBsync_file_range\fR\|(2) call.
713 \fRstr\fP can currently be one or more of:
717 SYNC_FILE_RANGE_WAIT_BEFORE
720 SYNC_FILE_RANGE_WRITE
723 SYNC_FILE_RANGE_WRITE
727 So if you do sync_file_range=wait_before,write:8, fio would use
728 \fBSYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE\fP for every 8 writes.
729 Also see the \fBsync_file_range\fR\|(2) man page. This option is Linux specific.
731 .BI overwrite \fR=\fPbool
732 If writing, setup the file first and do overwrites. Default: false.
734 .BI end_fsync \fR=\fPbool
735 Sync file contents when a write stage has completed. Default: false.
737 .BI fsync_on_close \fR=\fPbool
738 If true, sync file contents on close. This differs from \fBend_fsync\fR in that
739 it will happen on every close, not just at the end of the job. Default: false.
741 .BI rwmixread \fR=\fPint
742 Percentage of a mixed workload that should be reads. Default: 50.
744 .BI rwmixwrite \fR=\fPint
745 Percentage of a mixed workload that should be writes. If \fBrwmixread\fR and
746 \fBrwmixwrite\fR are given and do not sum to 100%, the latter of the two
747 overrides the first. This may interfere with a given rate setting, if fio is
748 asked to limit reads or writes to a certain rate. If that is the case, then
749 the distribution may be skewed. Default: 50.
751 .BI random_distribution \fR=\fPstr:float
752 By default, fio will use a completely uniform random distribution when asked
753 to perform random IO. Sometimes it is useful to skew the distribution in
754 specific ways, ensuring that some parts of the data is more hot than others.
755 Fio includes the following distribution models:
759 Uniform random distribution
769 When using a zipf or pareto distribution, an input value is also needed to
770 define the access pattern. For zipf, this is the zipf theta. For pareto,
771 it's the pareto power. Fio includes a test program, genzipf, that can be
772 used visualize what the given input values will yield in terms of hit rates.
773 If you wanted to use zipf with a theta of 1.2, you would use
774 random_distribution=zipf:1.2 as the option. If a non-uniform model is used,
775 fio will disable use of the random map.
777 .BI percentage_random \fR=\fPint
778 For a random workload, set how big a percentage should be random. This defaults
779 to 100%, in which case the workload is fully random. It can be set from
780 anywhere from 0 to 100. Setting it to 0 would make the workload fully
781 sequential. It is possible to set different values for reads, writes, and
782 trim. To do so, simply use a comma separated list. See \fBblocksize\fR.
785 Normally \fBfio\fR will cover every block of the file when doing random I/O. If
786 this parameter is given, a new offset will be chosen without looking at past
787 I/O history. This parameter is mutually exclusive with \fBverify\fR.
789 .BI softrandommap \fR=\fPbool
790 See \fBnorandommap\fR. If fio runs with the random block map enabled and it
791 fails to allocate the map, if this option is set it will continue without a
792 random block map. As coverage will not be as complete as with random maps, this
793 option is disabled by default.
795 .BI random_generator \fR=\fPstr
796 Fio supports the following engines for generating IO offsets for random IO:
800 Strong 2^88 cycle random number generator
803 Linear feedback shift register generator
807 Tausworthe is a strong random number generator, but it requires tracking on the
808 side if we want to ensure that blocks are only read or written once. LFSR
809 guarantees that we never generate the same offset twice, and it's also less
810 computationally expensive. It's not a true random generator, however, though
811 for IO purposes it's typically good enough. LFSR only works with single block
812 sizes, not with workloads that use multiple block sizes. If used with such a
813 workload, fio may read or write some blocks multiple times.
816 Run job with given nice value. See \fBnice\fR\|(2).
819 Set I/O priority value of this job between 0 (highest) and 7 (lowest). See
822 .BI prioclass \fR=\fPint
823 Set I/O priority class. See \fBionice\fR\|(1).
825 .BI thinktime \fR=\fPint
826 Stall job for given number of microseconds between issuing I/Os.
828 .BI thinktime_spin \fR=\fPint
829 Pretend to spend CPU time for given number of microseconds, sleeping the rest
830 of the time specified by \fBthinktime\fR. Only valid if \fBthinktime\fR is set.
832 .BI thinktime_blocks \fR=\fPint
833 Only valid if thinktime is set - control how many blocks to issue, before
834 waiting \fBthinktime\fR microseconds. If not set, defaults to 1 which will
835 make fio wait \fBthinktime\fR microseconds after every block. This
836 effectively makes any queue depth setting redundant, since no more than 1 IO
837 will be queued before we have to complete it and do our thinktime. In other
838 words, this setting effectively caps the queue depth if the latter is larger.
842 Cap bandwidth used by this job. The number is in bytes/sec, the normal postfix
843 rules apply. You can use \fBrate\fR=500k to limit reads and writes to 500k each,
844 or you can specify read and writes separately. Using \fBrate\fR=1m,500k would
845 limit reads to 1MB/sec and writes to 500KB/sec. Capping only reads or writes
846 can be done with \fBrate\fR=,500k or \fBrate\fR=500k,. The former will only
847 limit writes (to 500KB/sec), the latter will only limit reads.
849 .BI ratemin \fR=\fPint
850 Tell \fBfio\fR to do whatever it can to maintain at least the given bandwidth.
851 Failing to meet this requirement will cause the job to exit. The same format
852 as \fBrate\fR is used for read vs write separation.
854 .BI rate_iops \fR=\fPint
855 Cap the bandwidth to this number of IOPS. Basically the same as rate, just
856 specified independently of bandwidth. The same format as \fBrate\fR is used for
857 read vs write separation. If \fBblocksize\fR is a range, the smallest block
858 size is used as the metric.
860 .BI rate_iops_min \fR=\fPint
861 If this rate of I/O is not met, the job will exit. The same format as \fBrate\fR
862 is used for read vs write separation.
864 .BI ratecycle \fR=\fPint
865 Average bandwidth for \fBrate\fR and \fBratemin\fR over this number of
866 milliseconds. Default: 1000ms.
868 .BI latency_target \fR=\fPint
869 If set, fio will attempt to find the max performance point that the given
870 workload will run at while maintaining a latency below this target. The
871 values is given in microseconds. See \fBlatency_window\fR and
872 \fBlatency_percentile\fR.
874 .BI latency_window \fR=\fPint
875 Used with \fBlatency_target\fR to specify the sample window that the job
876 is run at varying queue depths to test the performance. The value is given
879 .BI latency_percentile \fR=\fPfloat
880 The percentage of IOs that must fall within the criteria specified by
881 \fBlatency_target\fR and \fBlatency_window\fR. If not set, this defaults
882 to 100.0, meaning that all IOs must be equal or below to the value set
883 by \fBlatency_target\fR.
885 .BI max_latency \fR=\fPint
886 If set, fio will exit the job if it exceeds this maximum latency. It will exit
889 .BI cpumask \fR=\fPint
890 Set CPU affinity for this job. \fIint\fR is a bitmask of allowed CPUs the job
891 may run on. See \fBsched_setaffinity\fR\|(2).
893 .BI cpus_allowed \fR=\fPstr
894 Same as \fBcpumask\fR, but allows a comma-delimited list of CPU numbers.
896 .BI cpus_allowed_policy \fR=\fPstr
897 Set the policy of how fio distributes the CPUs specified by \fBcpus_allowed\fR
898 or \fBcpumask\fR. Two policies are supported:
903 All jobs will share the CPU set specified.
906 Each job will get a unique CPU from the CPU set.
909 \fBshared\fR is the default behaviour, if the option isn't specified. If
910 \fBsplit\fR is specified, then fio will assign one cpu per job. If not enough
911 CPUs are given for the jobs listed, then fio will roundrobin the CPUs in
916 .BI numa_cpu_nodes \fR=\fPstr
917 Set this job running on specified NUMA nodes' CPUs. The arguments allow
918 comma delimited list of cpu numbers, A-B ranges, or 'all'.
920 .BI numa_mem_policy \fR=\fPstr
921 Set this job's memory policy and corresponding NUMA nodes. Format of
925 .B <mode>[:<nodelist>]
928 is one of the following memory policy:
930 .B default, prefer, bind, interleave, local
933 For \fBdefault\fR and \fBlocal\fR memory policy, no \fBnodelist\fR is
934 needed to be specified. For \fBprefer\fR, only one node is
935 allowed. For \fBbind\fR and \fBinterleave\fR, \fBnodelist\fR allows
936 comma delimited list of numbers, A-B ranges, or 'all'.
938 .BI startdelay \fR=\fPirange
939 Delay start of job for the specified number of seconds. Supports all time
940 suffixes to allow specification of hours, minutes, seconds and
941 milliseconds - seconds are the default if a unit is ommited.
942 Can be given as a range which causes each thread to choose randomly out of the
945 .BI runtime \fR=\fPint
946 Terminate processing after the specified number of seconds.
949 If given, run for the specified \fBruntime\fR duration even if the files are
950 completely read or written. The same workload will be repeated as many times
951 as \fBruntime\fR allows.
953 .BI ramp_time \fR=\fPint
954 If set, fio will run the specified workload for this amount of time before
955 logging any performance numbers. Useful for letting performance settle before
956 logging results, thus minimizing the runtime required for stable results. Note
957 that the \fBramp_time\fR is considered lead in time for a job, thus it will
958 increase the total runtime if a special timeout or runtime is specified.
960 .BI invalidate \fR=\fPbool
961 Invalidate buffer-cache for the file prior to starting I/O. Default: true.
964 Use synchronous I/O for buffered writes. For the majority of I/O engines,
965 this means using O_SYNC. Default: false.
967 .BI iomem \fR=\fPstr "\fR,\fP mem" \fR=\fPstr
968 Allocation method for I/O unit buffer. Allowed values are:
973 Allocate memory with \fBmalloc\fR\|(3).
976 Use shared memory buffers allocated through \fBshmget\fR\|(2).
979 Same as \fBshm\fR, but use huge pages as backing.
982 Use \fBmmap\fR\|(2) for allocation. Uses anonymous memory unless a filename
983 is given after the option in the format `:\fIfile\fR'.
986 Same as \fBmmap\fR, but use huge files as backing.
989 The amount of memory allocated is the maximum allowed \fBblocksize\fR for the
990 job multiplied by \fBiodepth\fR. For \fBshmhuge\fR or \fBmmaphuge\fR to work,
991 the system must have free huge pages allocated. \fBmmaphuge\fR also needs to
992 have hugetlbfs mounted, and \fIfile\fR must point there. At least on Linux,
993 huge pages must be manually allocated. See \fB/proc/sys/vm/nr_hugehages\fR
994 and the documentation for that. Normally you just need to echo an appropriate
995 number, eg echoing 8 will ensure that the OS has 8 huge pages ready for
999 .BI iomem_align \fR=\fPint "\fR,\fP mem_align" \fR=\fPint
1000 This indicates the memory alignment of the IO memory buffers. Note that the
1001 given alignment is applied to the first IO unit buffer, if using \fBiodepth\fR
1002 the alignment of the following buffers are given by the \fBbs\fR used. In
1003 other words, if using a \fBbs\fR that is a multiple of the page sized in the
1004 system, all buffers will be aligned to this value. If using a \fBbs\fR that
1005 is not page aligned, the alignment of subsequent IO memory buffers is the
1006 sum of the \fBiomem_align\fR and \fBbs\fR used.
1008 .BI hugepage\-size \fR=\fPint
1009 Defines the size of a huge page. Must be at least equal to the system setting.
1010 Should be a multiple of 1MB. Default: 4MB.
1013 Terminate all jobs when one finishes. Default: wait for each job to finish.
1015 .BI bwavgtime \fR=\fPint
1016 Average bandwidth calculations over the given time in milliseconds. Default:
1019 .BI iopsavgtime \fR=\fPint
1020 Average IOPS calculations over the given time in milliseconds. Default:
1023 .BI create_serialize \fR=\fPbool
1024 If true, serialize file creation for the jobs. Default: true.
1026 .BI create_fsync \fR=\fPbool
1027 \fBfsync\fR\|(2) data file after creation. Default: true.
1029 .BI create_on_open \fR=\fPbool
1030 If true, the files are not created until they are opened for IO by the job.
1032 .BI create_only \fR=\fPbool
1033 If true, fio will only run the setup phase of the job. If files need to be
1034 laid out or updated on disk, only that will be done. The actual job contents
1037 .BI pre_read \fR=\fPbool
1038 If this is given, files will be pre-read into memory before starting the given
1039 IO operation. This will also clear the \fR \fBinvalidate\fR flag, since it is
1040 pointless to pre-read and then drop the cache. This will only work for IO
1041 engines that are seekable, since they allow you to read the same data
1042 multiple times. Thus it will not work on eg network or splice IO.
1044 .BI unlink \fR=\fPbool
1045 Unlink job files when done. Default: false.
1047 .BI loops \fR=\fPint
1048 Specifies the number of iterations (runs of the same workload) of this job.
1051 .BI verify_only \fR=\fPbool
1052 Do not perform the specified workload, only verify data still matches previous
1053 invocation of this workload. This option allows one to check data multiple
1054 times at a later date without overwriting it. This option makes sense only for
1055 workloads that write data, and does not support workloads with the
1056 \fBtime_based\fR option set.
1058 .BI do_verify \fR=\fPbool
1059 Run the verify phase after a write phase. Only valid if \fBverify\fR is set.
1062 .BI verify \fR=\fPstr
1063 Method of verifying file contents after each iteration of the job. Allowed
1068 .B md5 crc16 crc32 crc32c crc32c-intel crc64 crc7 sha256 sha512 sha1 xxhash
1069 Store appropriate checksum in the header of each block. crc32c-intel is
1070 hardware accelerated SSE4.2 driven, falls back to regular crc32c if
1071 not supported by the system.
1074 Write extra information about each I/O (timestamp, block number, etc.). The
1075 block number is verified. See \fBverify_pattern\fR as well.
1078 Pretend to verify. Used for testing internals.
1081 This option can be used for repeated burn-in tests of a system to make sure
1082 that the written data is also correctly read back. If the data direction given
1083 is a read or random read, fio will assume that it should verify a previously
1084 written file. If the data direction includes any form of write, the verify will
1085 be of the newly written data.
1088 .BI verifysort \fR=\fPbool
1089 If true, written verify blocks are sorted if \fBfio\fR deems it to be faster to
1090 read them back in a sorted manner. Default: true.
1092 .BI verifysort_nr \fR=\fPint
1093 Pre-load and sort verify blocks for a read workload.
1095 .BI verify_offset \fR=\fPint
1096 Swap the verification header with data somewhere else in the block before
1097 writing. It is swapped back before verifying.
1099 .BI verify_interval \fR=\fPint
1100 Write the verification header for this number of bytes, which should divide
1101 \fBblocksize\fR. Default: \fBblocksize\fR.
1103 .BI verify_pattern \fR=\fPstr
1104 If set, fio will fill the io buffers with this pattern. Fio defaults to filling
1105 with totally random bytes, but sometimes it's interesting to fill with a known
1106 pattern for io verification purposes. Depending on the width of the pattern,
1107 fio will fill 1/2/3/4 bytes of the buffer at the time(it can be either a
1108 decimal or a hex number). The verify_pattern if larger than a 32-bit quantity
1109 has to be a hex number that starts with either "0x" or "0X". Use with
1112 .BI verify_fatal \fR=\fPbool
1113 If true, exit the job on the first observed verification failure. Default:
1116 .BI verify_dump \fR=\fPbool
1117 If set, dump the contents of both the original data block and the data block we
1118 read off disk to files. This allows later analysis to inspect just what kind of
1119 data corruption occurred. Off by default.
1121 .BI verify_async \fR=\fPint
1122 Fio will normally verify IO inline from the submitting thread. This option
1123 takes an integer describing how many async offload threads to create for IO
1124 verification instead, causing fio to offload the duty of verifying IO contents
1125 to one or more separate threads. If using this offload option, even sync IO
1126 engines can benefit from using an \fBiodepth\fR setting higher than 1, as it
1127 allows them to have IO in flight while verifies are running.
1129 .BI verify_async_cpus \fR=\fPstr
1130 Tell fio to set the given CPU affinity on the async IO verification threads.
1131 See \fBcpus_allowed\fP for the format used.
1133 .BI verify_backlog \fR=\fPint
1134 Fio will normally verify the written contents of a job that utilizes verify
1135 once that job has completed. In other words, everything is written then
1136 everything is read back and verified. You may want to verify continually
1137 instead for a variety of reasons. Fio stores the meta data associated with an
1138 IO block in memory, so for large verify workloads, quite a bit of memory would
1139 be used up holding this meta data. If this option is enabled, fio will write
1140 only N blocks before verifying these blocks.
1142 .BI verify_backlog_batch \fR=\fPint
1143 Control how many blocks fio will verify if verify_backlog is set. If not set,
1144 will default to the value of \fBverify_backlog\fR (meaning the entire queue is
1145 read back and verified). If \fBverify_backlog_batch\fR is less than
1146 \fBverify_backlog\fR then not all blocks will be verified, if
1147 \fBverify_backlog_batch\fR is larger than \fBverify_backlog\fR, some blocks
1148 will be verified more than once.
1150 .BI trim_percentage \fR=\fPint
1151 Number of verify blocks to discard/trim.
1153 .BI trim_verify_zero \fR=\fPbool
1154 Verify that trim/discarded blocks are returned as zeroes.
1156 .BI trim_backlog \fR=\fPint
1157 Trim after this number of blocks are written.
1159 .BI trim_backlog_batch \fR=\fPint
1160 Trim this number of IO blocks.
1162 .BI experimental_verify \fR=\fPbool
1163 Enable experimental verification.
1165 .B stonewall "\fR,\fP wait_for_previous"
1166 Wait for preceding jobs in the job file to exit before starting this one.
1167 \fBstonewall\fR implies \fBnew_group\fR.
1170 Start a new reporting group. If not given, all jobs in a file will be part
1171 of the same reporting group, unless separated by a stonewall.
1173 .BI numjobs \fR=\fPint
1174 Number of clones (processes/threads performing the same workload) of this job.
1178 If set, display per-group reports instead of per-job when \fBnumjobs\fR is
1182 Use threads created with \fBpthread_create\fR\|(3) instead of processes created
1183 with \fBfork\fR\|(2).
1185 .BI zonesize \fR=\fPint
1186 Divide file into zones of the specified size in bytes. See \fBzoneskip\fR.
1188 .BI zonerange \fR=\fPint
1189 Give size of an IO zone. See \fBzoneskip\fR.
1191 .BI zoneskip \fR=\fPint
1192 Skip the specified number of bytes when \fBzonesize\fR bytes of data have been
1195 .BI write_iolog \fR=\fPstr
1196 Write the issued I/O patterns to the specified file. Specify a separate file
1197 for each job, otherwise the iologs will be interspersed and the file may be
1200 .BI read_iolog \fR=\fPstr
1201 Replay the I/O patterns contained in the specified file generated by
1202 \fBwrite_iolog\fR, or may be a \fBblktrace\fR binary file.
1204 .BI replay_no_stall \fR=\fPint
1205 While replaying I/O patterns using \fBread_iolog\fR the default behavior
1206 attempts to respect timing information between I/Os. Enabling
1207 \fBreplay_no_stall\fR causes I/Os to be replayed as fast as possible while
1208 still respecting ordering.
1210 .BI replay_redirect \fR=\fPstr
1211 While replaying I/O patterns using \fBread_iolog\fR the default behavior
1212 is to replay the IOPS onto the major/minor device that each IOP was recorded
1213 from. Setting \fBreplay_redirect\fR causes all IOPS to be replayed onto the
1214 single specified device regardless of the device it was recorded from.
1216 .BI write_bw_log \fR=\fPstr
1217 If given, write a bandwidth log of the jobs in this job file. Can be used to
1218 store data of the bandwidth of the jobs in their lifetime. The included
1219 fio_generate_plots script uses gnuplot to turn these text files into nice
1220 graphs. See \fBwrite_lat_log\fR for behaviour of given filename. For this
1221 option, the postfix is _bw.x.log, where x is the index of the job (1..N,
1222 where N is the number of jobs)
1224 .BI write_lat_log \fR=\fPstr
1225 Same as \fBwrite_bw_log\fR, but writes I/O completion latencies. If no
1226 filename is given with this option, the default filename of
1227 "jobname_type.x.log" is used, where x is the index of the job (1..N, where
1228 N is the number of jobs). Even if the filename is given, fio will still
1229 append the type of log.
1231 .BI write_iops_log \fR=\fPstr
1232 Same as \fBwrite_bw_log\fR, but writes IOPS. If no filename is given with this
1233 option, the default filename of "jobname_type.x.log" is used, where x is the
1234 index of the job (1..N, where N is the number of jobs). Even if the filename
1235 is given, fio will still append the type of log.
1237 .BI log_avg_msec \fR=\fPint
1238 By default, fio will log an entry in the iops, latency, or bw log for every
1239 IO that completes. When writing to the disk log, that can quickly grow to a
1240 very large size. Setting this option makes fio average the each log entry
1241 over the specified period of time, reducing the resolution of the log.
1244 .BI log_offset \fR=\fPbool
1245 If this is set, the iolog options will include the byte offset for the IO
1246 entry as well as the other data values.
1248 .BI log_compression \fR=\fPint
1249 If this is set, fio will compress the IO logs as it goes, to keep the memory
1250 footprint lower. When a log reaches the specified size, that chunk is removed
1251 and compressed in the background. Given that IO logs are fairly highly
1252 compressible, this yields a nice memory savings for longer runs. The downside
1253 is that the compression will consume some background CPU cycles, so it may
1254 impact the run. This, however, is also true if the logging ends up consuming
1255 most of the system memory. So pick your poison. The IO logs are saved
1256 normally at the end of a run, by decompressing the chunks and storing them
1257 in the specified log file. This feature depends on the availability of zlib.
1259 .BI log_store_compressed \fR=\fPbool
1260 If set, and \fBlog\fR_compression is also set, fio will store the log files in
1261 a compressed format. They can be decompressed with fio, using the
1262 \fB\-\-inflate-log\fR command line parameter. The files will be stored with a
1265 .BI disable_lat \fR=\fPbool
1266 Disable measurements of total latency numbers. Useful only for cutting
1267 back the number of calls to \fBgettimeofday\fR\|(2), as that does impact performance at
1268 really high IOPS rates. Note that to really get rid of a large amount of these
1269 calls, this option must be used with disable_slat and disable_bw as well.
1271 .BI disable_clat \fR=\fPbool
1272 Disable measurements of completion latency numbers. See \fBdisable_lat\fR.
1274 .BI disable_slat \fR=\fPbool
1275 Disable measurements of submission latency numbers. See \fBdisable_lat\fR.
1277 .BI disable_bw_measurement \fR=\fPbool
1278 Disable measurements of throughput/bandwidth numbers. See \fBdisable_lat\fR.
1280 .BI lockmem \fR=\fPint
1281 Pin the specified amount of memory with \fBmlock\fR\|(2). Can be used to
1282 simulate a smaller amount of memory. The amount specified is per worker.
1284 .BI exec_prerun \fR=\fPstr
1285 Before running the job, execute the specified command with \fBsystem\fR\|(3).
1287 Output is redirected in a file called \fBjobname.prerun.txt\fR
1290 .BI exec_postrun \fR=\fPstr
1291 Same as \fBexec_prerun\fR, but the command is executed after the job completes.
1293 Output is redirected in a file called \fBjobname.postrun.txt\fR
1296 .BI ioscheduler \fR=\fPstr
1297 Attempt to switch the device hosting the file to the specified I/O scheduler.
1299 .BI disk_util \fR=\fPbool
1300 Generate disk utilization statistics if the platform supports it. Default: true.
1302 .BI clocksource \fR=\fPstr
1303 Use the given clocksource as the base of timing. The supported options are:
1307 \fBgettimeofday\fR\|(2)
1310 \fBclock_gettime\fR\|(2)
1313 Internal CPU clock source
1317 \fBcpu\fR is the preferred clocksource if it is reliable, as it is very fast
1318 (and fio is heavy on time calls). Fio will automatically use this clocksource
1319 if it's supported and considered reliable on the system it is running on,
1320 unless another clocksource is specifically set. For x86/x86-64 CPUs, this
1321 means supporting TSC Invariant.
1323 .BI gtod_reduce \fR=\fPbool
1324 Enable all of the \fBgettimeofday\fR\|(2) reducing options (disable_clat, disable_slat,
1325 disable_bw) plus reduce precision of the timeout somewhat to really shrink the
1326 \fBgettimeofday\fR\|(2) call count. With this option enabled, we only do about 0.4% of
1327 the gtod() calls we would have done if all time keeping was enabled.
1329 .BI gtod_cpu \fR=\fPint
1330 Sometimes it's cheaper to dedicate a single thread of execution to just getting
1331 the current time. Fio (and databases, for instance) are very intensive on
1332 \fBgettimeofday\fR\|(2) calls. With this option, you can set one CPU aside for doing
1333 nothing but logging current time to a shared memory location. Then the other
1334 threads/processes that run IO workloads need only copy that segment, instead of
1335 entering the kernel with a \fBgettimeofday\fR\|(2) call. The CPU set aside for doing
1336 these time calls will be excluded from other uses. Fio will manually clear it
1337 from the CPU mask of other jobs.
1339 .BI ignore_error \fR=\fPstr
1340 Sometimes you want to ignore some errors during test in that case you can specify
1341 error list for each error type.
1343 ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST
1345 errors for given error type is separated with ':'.
1346 Error may be symbol ('ENOSPC', 'ENOMEM') or an integer.
1348 Example: ignore_error=EAGAIN,ENOSPC:122 .
1350 This option will ignore EAGAIN from READ, and ENOSPC and 122(EDQUOT) from WRITE.
1352 .BI error_dump \fR=\fPbool
1353 If set dump every error even if it is non fatal, true by default. If disabled
1354 only fatal error will be dumped
1356 .BI profile \fR=\fPstr
1357 Select a specific builtin performance test.
1359 .BI cgroup \fR=\fPstr
1360 Add job to this control group. If it doesn't exist, it will be created.
1361 The system must have a mounted cgroup blkio mount point for this to work. If
1362 your system doesn't have it mounted, you can do so with:
1364 # mount \-t cgroup \-o blkio none /cgroup
1366 .BI cgroup_weight \fR=\fPint
1367 Set the weight of the cgroup to this value. See the documentation that comes
1368 with the kernel, allowed values are in the range of 100..1000.
1370 .BI cgroup_nodelete \fR=\fPbool
1371 Normally fio will delete the cgroups it has created after the job completion.
1372 To override this behavior and to leave cgroups around after the job completion,
1373 set cgroup_nodelete=1. This can be useful if one wants to inspect various
1374 cgroup files after job completion. Default: false
1377 Instead of running as the invoking user, set the user ID to this value before
1378 the thread/process does any work.
1381 Set group ID, see \fBuid\fR.
1383 .BI unit_base \fR=\fPint
1384 Base unit for reporting. Allowed values are:
1388 Use auto-detection (default).
1398 .BI flow_id \fR=\fPint
1399 The ID of the flow. If not specified, it defaults to being a global flow. See
1403 Weight in token-based flow control. If this value is used, then there is a
1404 \fBflow counter\fR which is used to regulate the proportion of activity between
1405 two or more jobs. fio attempts to keep this flow counter near zero. The
1406 \fBflow\fR parameter stands for how much should be added or subtracted to the
1407 flow counter on each iteration of the main I/O loop. That is, if one job has
1408 \fBflow=8\fR and another job has \fBflow=-1\fR, then there will be a roughly
1409 1:8 ratio in how much one runs vs the other.
1411 .BI flow_watermark \fR=\fPint
1412 The maximum value that the absolute value of the flow counter is allowed to
1413 reach before the job must wait for a lower value of the counter.
1415 .BI flow_sleep \fR=\fPint
1416 The period of time, in microseconds, to wait after the flow watermark has been
1417 exceeded before retrying operations
1419 .BI clat_percentiles \fR=\fPbool
1420 Enable the reporting of percentiles of completion latencies.
1422 .BI percentile_list \fR=\fPfloat_list
1423 Overwrite the default list of percentiles for completion
1424 latencies. Each number is a floating number in the range (0,100], and
1425 the maximum length of the list is 20. Use ':' to separate the
1426 numbers. For example, \-\-percentile_list=99.5:99.9 will cause fio to
1427 report the values of completion latency below which 99.5% and 99.9% of
1428 the observed latencies fell, respectively.
1429 .SS "Ioengine Parameters List"
1430 Some parameters are only valid when a specific ioengine is in use. These are
1431 used identically to normal parameters, with the caveat that when used on the
1432 command line, they must come after the ioengine.
1434 .BI (cpu)cpuload \fR=\fPint
1435 Attempt to use the specified percentage of CPU cycles.
1437 .BI (cpu)cpuchunks \fR=\fPint
1438 Split the load into cycles of the given time. In microseconds.
1440 .BI (cpu)exit_on_io_done \fR=\fPbool
1441 Detect when IO threads are done, then exit.
1443 .BI (libaio)userspace_reap
1444 Normally, with the libaio engine in use, fio will use
1445 the io_getevents system call to reap newly returned events.
1446 With this flag turned on, the AIO ring will be read directly
1447 from user-space to reap events. The reaping mode is only
1448 enabled when polling for a minimum of 0 events (eg when
1449 iodepth_batch_complete=0).
1451 .BI (net,netsplice)hostname \fR=\fPstr
1452 The host name or IP address to use for TCP or UDP based IO.
1453 If the job is a TCP listener or UDP reader, the hostname is not
1454 used and must be omitted unless it is a valid UDP multicast address.
1456 .BI (net,netsplice)port \fR=\fPint
1457 The TCP or UDP port to bind to or connect to.
1459 .BI (net,netsplice)interface \fR=\fPstr
1460 The IP address of the network interface used to send or receive UDP multicast
1463 .BI (net,netsplice)ttl \fR=\fPint
1464 Time-to-live value for outgoing UDP multicast packets. Default: 1
1466 .BI (net,netsplice)nodelay \fR=\fPbool
1467 Set TCP_NODELAY on TCP connections.
1469 .BI (net,netsplice)protocol \fR=\fPstr "\fR,\fP proto" \fR=\fPstr
1470 The network protocol to use. Accepted values are:
1475 Transmission control protocol
1478 Transmission control protocol V6
1481 User datagram protocol
1484 User datagram protocol V6
1490 When the protocol is TCP or UDP, the port must also be given,
1491 as well as the hostname if the job is a TCP listener or UDP
1492 reader. For unix sockets, the normal filename option should be
1493 used and the port is invalid.
1496 .BI (net,netsplice)listen
1497 For TCP network connections, tell fio to listen for incoming
1498 connections rather than initiating an outgoing connection. The
1499 hostname must be omitted if this option is used.
1501 .BI (net, pingpong) \fR=\fPbool
1502 Normally a network writer will just continue writing data, and a network reader
1503 will just consume packets. If pingpong=1 is set, a writer will send its normal
1504 payload to the reader, then wait for the reader to send the same payload back.
1505 This allows fio to measure network latencies. The submission and completion
1506 latencies then measure local time spent sending or receiving, and the
1507 completion latency measures how long it took for the other end to receive and
1508 send back. For UDP multicast traffic pingpong=1 should only be set for a single
1509 reader when multiple readers are listening to the same address.
1511 .BI (e4defrag,donorname) \fR=\fPstr
1512 File will be used as a block donor (swap extents between files)
1514 .BI (e4defrag,inplace) \fR=\fPint
1515 Configure donor file block allocation strategy
1518 Preallocate donor's file on init
1521 allocate space immediately inside defragment event, and free right after event
1524 .BI (rbd)rbdname \fR=\fPstr
1525 Specifies the name of the RBD.
1527 .BI (rbd)pool \fR=\fPstr
1528 Specifies the name of the Ceph pool containing the RBD.
1530 .BI (rbd)clientname \fR=\fPstr
1531 Specifies the username (without the 'client.' prefix) used to access the Ceph cluster.
1533 While running, \fBfio\fR will display the status of the created jobs. For
1537 Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
1540 The characters in the first set of brackets denote the current status of each
1541 threads. The possible values are:
1547 Setup but not started.
1553 Initialized, waiting.
1556 Running, doing sequential reads.
1559 Running, doing random reads.
1562 Running, doing sequential writes.
1565 Running, doing random writes.
1568 Running, doing mixed sequential reads/writes.
1571 Running, doing mixed random reads/writes.
1574 Running, currently waiting for \fBfsync\fR\|(2).
1577 Running, verifying written data.
1580 Exited, not reaped by main thread.
1583 Exited, thread reaped.
1587 The second set of brackets shows the estimated completion percentage of
1588 the current group. The third set shows the read and write I/O rate,
1589 respectively. Finally, the estimated run time of the job is displayed.
1591 When \fBfio\fR completes (or is interrupted by Ctrl-C), it will show data
1592 for each thread, each group of threads, and each disk, in that order.
1594 Per-thread statistics first show the threads client number, group-id, and
1595 error code. The remaining figures are as follows:
1599 Number of megabytes of I/O performed.
1602 Average data rate (bandwidth).
1608 Submission latency minimum, maximum, average and standard deviation. This is
1609 the time it took to submit the I/O.
1612 Completion latency minimum, maximum, average and standard deviation. This
1613 is the time between submission and completion.
1616 Bandwidth minimum, maximum, percentage of aggregate bandwidth received, average
1617 and standard deviation.
1620 CPU usage statistics. Includes user and system time, number of context switches
1621 this thread went through and number of major and minor page faults.
1624 Distribution of I/O depths. Each depth includes everything less than (or equal)
1625 to it, but greater than the previous depth.
1628 Number of read/write requests issued, and number of short read/write requests.
1631 Distribution of I/O completion latencies. The numbers follow the same pattern
1635 The group statistics show:
1640 Number of megabytes I/O performed.
1643 Aggregate bandwidth of threads in the group.
1646 Minimum average bandwidth a thread saw.
1649 Maximum average bandwidth a thread saw.
1652 Shortest runtime of threads in the group.
1655 Longest runtime of threads in the group.
1659 Finally, disk statistics are printed with reads first:
1664 Number of I/Os performed by all groups.
1667 Number of merges in the I/O scheduler.
1670 Number of ticks we kept the disk busy.
1673 Total time spent in the disk queue.
1680 It is also possible to get fio to dump the current output while it is
1681 running, without terminating the job. To do that, send fio the \fBUSR1\fR
1684 If the \fB\-\-minimal\fR / \fB\-\-append-terse\fR options are given, the
1685 results will be printed/appended in a semicolon-delimited format suitable for
1687 A job description (if provided) follows on a new line. Note that the first
1688 number in the line is the version number. If the output has to be changed
1689 for some reason, this number will be incremented by 1 to signify that
1690 change. The fields are:
1693 .B terse version, fio version, jobname, groupid, error
1697 .B Total I/O \fR(KB)\fP, bandwidth \fR(KB/s)\fP, IOPS, runtime \fR(ms)\fP
1701 .B min, max, mean, standard deviation
1705 .B min, max, mean, standard deviation
1707 Completion latency percentiles (20 fields):
1709 .B Xth percentile=usec
1713 .B min, max, mean, standard deviation
1717 .B min, max, aggregate percentage of total, mean, standard deviation
1723 .B Total I/O \fR(KB)\fP, bandwidth \fR(KB/s)\fP, IOPS, runtime \fR(ms)\fP
1727 .B min, max, mean, standard deviation
1731 .B min, max, mean, standard deviation
1733 Completion latency percentiles (20 fields):
1735 .B Xth percentile=usec
1739 .B min, max, mean, standard deviation
1743 .B min, max, aggregate percentage of total, mean, standard deviation
1749 .B user, system, context switches, major page faults, minor page faults
1752 IO depth distribution:
1754 .B <=1, 2, 4, 8, 16, 32, >=64
1757 IO latency distribution:
1761 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
1765 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
1769 Disk utilization (1 for each disk used):
1771 .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
1774 Error Info (dependent on continue_on_error, default off):
1776 .B total # errors, first error code
1779 .B text description (if provided in config - appears on newline)
1782 Normally you would run fio as a stand-alone application on the machine
1783 where the IO workload should be generated. However, it is also possible to
1784 run the frontend and backend of fio separately. This makes it possible to
1785 have a fio server running on the machine(s) where the IO workload should
1786 be running, while controlling it from another machine.
1788 To start the server, you would do:
1790 \fBfio \-\-server=args\fR
1792 on that machine, where args defines what fio listens to. The arguments
1793 are of the form 'type:hostname or IP:port'. 'type' is either 'ip' (or ip4)
1794 for TCP/IP v4, 'ip6' for TCP/IP v6, or 'sock' for a local unix domain
1795 socket. 'hostname' is either a hostname or IP address, and 'port' is the port to
1796 listen to (only valid for TCP/IP, not a local socket). Some examples:
1800 Start a fio server, listening on all interfaces on the default port (8765).
1802 2) fio \-\-server=ip:hostname,4444
1804 Start a fio server, listening on IP belonging to hostname and on port 4444.
1806 3) fio \-\-server=ip6:::1,4444
1808 Start a fio server, listening on IPv6 localhost ::1 and on port 4444.
1810 4) fio \-\-server=,4444
1812 Start a fio server, listening on all interfaces on port 4444.
1814 5) fio \-\-server=1.2.3.4
1816 Start a fio server, listening on IP 1.2.3.4 on the default port.
1818 6) fio \-\-server=sock:/tmp/fio.sock
1820 Start a fio server, listening on the local socket /tmp/fio.sock.
1822 When a server is running, you can connect to it from a client. The client
1825 fio \-\-local-args \-\-client=server \-\-remote-args <job file(s)>
1827 where \-\-local-args are arguments that are local to the client where it is
1828 running, 'server' is the connect string, and \-\-remote-args and <job file(s)>
1829 are sent to the server. The 'server' string follows the same format as it
1830 does on the server side, to allow IP/hostname/socket and port strings.
1831 You can connect to multiple clients as well, to do that you could run:
1833 fio \-\-client=server2 \-\-client=server2 <job file(s)>
1837 was written by Jens Axboe <jens.axboe@oracle.com>,
1838 now Jens Axboe <jaxboe@fusionio.com>.
1840 This man page was written by Aaron Carroll <aaronc@cse.unsw.edu.au> based
1841 on documentation by Jens Axboe.
1842 .SH "REPORTING BUGS"
1843 Report bugs to the \fBfio\fR mailing list <fio@vger.kernel.org>.
1846 For further documentation see \fBHOWTO\fR and \fBREADME\fR.
1848 Sample jobfiles are available in the \fBexamples\fR directory.
projects / fio.git / blob