1 .TH fio 1 "December 2014" "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 \-\-output-format \fR=\fPformat
24 Set the reporting format to \fInormal\fR, \fIterse\fR, or \fIjson\fR.
26 .BI \-\-runtime \fR=\fPruntime
27 Limit run time to \fIruntime\fR seconds.
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. Supported operators are:
128 .B multiplication (*)
134 .B exponentiation (^)
138 For time values in expressions, units are microseconds by default. This is
139 different than for time values not in expressions (not enclosed in
140 parentheses). The types used are:
143 String: a sequence of alphanumeric characters.
146 SI integer: a whole number, possibly containing a suffix denoting the base unit
147 of the value. Accepted suffixes are `k', 'M', 'G', 'T', and 'P', denoting
148 kilo (1024), mega (1024^2), giga (1024^3), tera (1024^4), and peta (1024^5)
149 respectively. If prefixed with '0x', the value is assumed to be base 16
150 (hexadecimal). A suffix may include a trailing 'b', for instance 'kb' is
151 identical to 'k'. You can specify a base 10 value by using 'KiB', 'MiB','GiB',
152 etc. This is useful for disk drives where values are often given in base 10
153 values. Specifying '30GiB' will get you 30*1000^3 bytes.
154 When specifying times the default suffix meaning changes, still denoting the
155 base unit of the value, but accepted suffixes are 'D' (days), 'H' (hours), 'M'
156 (minutes), 'S' Seconds, 'ms' (or msec) milli seconds, 'us' (or 'usec') micro
157 seconds. Time values without a unit specify seconds.
158 The suffixes are not case sensitive.
161 Boolean: a true or false value. `0' denotes false, `1' denotes true.
164 Integer range: a range of integers specified in the format
165 \fIlower\fR:\fIupper\fR or \fIlower\fR\-\fIupper\fR. \fIlower\fR and
166 \fIupper\fR may contain a suffix as described above. If an option allows two
167 sets of ranges, they are separated with a `,' or `/' character. For example:
171 List of floating numbers: A list of floating numbers, separated by
176 May be used to override the job name. On the command line, this parameter
177 has the special purpose of signalling the start of a new job.
179 .BI description \fR=\fPstr
180 Human-readable description of the job. It is printed when the job is run, but
181 otherwise has no special purpose.
183 .BI directory \fR=\fPstr
184 Prefix filenames with this directory. Used to place files in a location other
186 You can specify a number of directories by separating the names with a ':'
187 character. These directories will be assigned equally distributed to job clones
188 creates with \fInumjobs\fR as long as they are using generated filenames.
189 If specific \fIfilename(s)\fR are set fio will use the first listed directory,
190 and thereby matching the \fIfilename\fR semantic which generates a file each
191 clone if not specified, but let all clones use the same if set. See
192 \fIfilename\fR for considerations regarding escaping certain characters on
195 .BI filename \fR=\fPstr
197 normally makes up a file name based on the job name, thread number, and file
198 number. If you want to share files between threads in a job or several jobs,
199 specify a \fIfilename\fR for each of them to override the default.
200 If the I/O engine is file-based, you can specify
201 a number of files by separating the names with a `:' character. `\-' is a
202 reserved name, meaning stdin or stdout, depending on the read/write direction
203 set. On Windows, disk devices are accessed as \\.\PhysicalDrive0 for the first
204 device, \\.\PhysicalDrive1 for the second etc. Note: Windows and FreeBSD
205 prevent write access to areas of the disk containing in-use data
206 (e.g. filesystems). If the wanted filename does need to include a colon, then
207 escape that with a '\\' character. For instance, if the filename is
208 "/dev/dsk/foo@3,0:c", then you would use filename="/dev/dsk/foo@3,0\\:c".
210 .BI filename_format \fR=\fPstr
211 If sharing multiple files between jobs, it is usually necessary to have
212 fio generate the exact names that you want. By default, fio will name a file
213 based on the default file format specification of
214 \fBjobname.jobnumber.filenumber\fP. With this option, that can be
215 customized. Fio will recognize and replace the following keywords in this
221 The name of the worker thread or process.
224 The incremental number of the worker thread or process.
227 The incremental number of the file for that worker thread or process.
230 To have dependent jobs share a set of files, this option can be set to
231 have fio generate filenames that are shared between the two. For instance,
232 if \fBtestfiles.$filenum\fR is specified, file number 4 for any job will
233 be named \fBtestfiles.4\fR. The default of \fB$jobname.$jobnum.$filenum\fR
234 will be used if no other format specifier is given.
238 .BI lockfile \fR=\fPstr
239 Fio defaults to not locking any files before it does IO to them. If a file or
240 file descriptor is shared, fio can serialize IO to that file to make the end
241 result consistent. This is usual for emulating real workloads that share files.
247 No locking. This is the default.
250 Only one thread or process may do IO at a time, excluding all others.
253 Read-write locking on the file. Many readers may access the file at the same
254 time, but writes get exclusive access.
258 .BI opendir \fR=\fPstr
259 Recursively open any files below directory \fIstr\fR.
261 .BI readwrite \fR=\fPstr "\fR,\fP rw" \fR=\fPstr
262 Type of I/O pattern. Accepted values are:
273 Sequential trim (Linux block devices only).
282 Random trim (Linux block devices only).
285 Mixed sequential reads and writes.
288 Mixed random reads and writes.
291 For mixed I/O, the default split is 50/50. For certain types of io the result
292 may still be skewed a bit, since the speed may be different. It is possible to
293 specify a number of IO's to do before getting a new offset, this is done by
294 appending a `:\fI<nr>\fR to the end of the string given. For a random read, it
295 would look like \fBrw=randread:8\fR for passing in an offset modifier with a
296 value of 8. If the postfix is used with a sequential IO pattern, then the value
297 specified will be added to the generated offset for each IO. For instance,
298 using \fBrw=write:4k\fR will skip 4k for every write. It turns sequential IO
299 into sequential IO with holes. See the \fBrw_sequencer\fR option.
302 .BI rw_sequencer \fR=\fPstr
303 If an offset modifier is given by appending a number to the \fBrw=<str>\fR line,
304 then this option controls how that number modifies the IO offset being
305 generated. Accepted values are:
310 Generate sequential offset
313 Generate the same offset
316 \fBsequential\fR is only useful for random IO, where fio would normally
317 generate a new random offset for every IO. If you append eg 8 to randread, you
318 would get a new random offset for every 8 IO's. The result would be a seek for
319 only every 8 IO's, instead of for every IO. Use \fBrw=randread:8\fR to specify
320 that. As sequential IO is already sequential, setting \fBsequential\fR for that
321 would not result in any differences. \fBidentical\fR behaves in a similar
322 fashion, except it sends the same offset 8 number of times before generating a
327 .BI kb_base \fR=\fPint
328 The base unit for a kilobyte. The defacto base is 2^10, 1024. Storage
329 manufacturers like to use 10^3 or 1000 as a base ten unit instead, for obvious
330 reasons. Allowed values are 1024 or 1000, with 1024 being the default.
332 .BI unified_rw_reporting \fR=\fPbool
333 Fio normally reports statistics on a per data direction basis, meaning that
334 read, write, and trim are accounted and reported separately. If this option is
335 set fio sums the results and reports them as "mixed" instead.
337 .BI randrepeat \fR=\fPbool
338 Seed the random number generator used for random I/O patterns in a predictable
339 way so the pattern is repeatable across runs. Default: true.
341 .BI allrandrepeat \fR=\fPbool
342 Seed all random number generators in a predictable way so results are
343 repeatable across runs. Default: false.
345 .BI randseed \fR=\fPint
346 Seed the random number generators based on this seed value, to be able to
347 control what sequence of output is being generated. If not set, the random
348 sequence depends on the \fBrandrepeat\fR setting.
350 .BI fallocate \fR=\fPstr
351 Whether pre-allocation is performed when laying down files. Accepted values
357 Do not pre-allocate space.
360 Pre-allocate via \fBposix_fallocate\fR\|(3).
363 Pre-allocate via \fBfallocate\fR\|(2) with FALLOC_FL_KEEP_SIZE set.
366 Backward-compatible alias for 'none'.
369 Backward-compatible alias for 'posix'.
372 May not be available on all supported platforms. 'keep' is only
373 available on Linux. If using ZFS on Solaris this must be set to 'none'
374 because ZFS doesn't support it. Default: 'posix'.
377 .BI fadvise_hint \fR=\fPbool
378 Use \fBposix_fadvise\fR\|(2) to advise the kernel what I/O patterns
379 are likely to be issued. Default: true.
381 .BI fadvise_stream \fR=\fPint
382 Use \fBposix_fadvise\fR\|(2) to advise the kernel what stream ID the
383 writes issued belong to. Only supported on Linux. Note, this option
384 may change going forward.
387 Total size of I/O for this job. \fBfio\fR will run until this many bytes have
388 been transferred, unless limited by other options (\fBruntime\fR, for instance,
389 or increased/descreased by \fBio_size\fR). Unless \fBnrfiles\fR and
390 \fBfilesize\fR options are given, this amount will be divided between the
391 available files for the job. If not set, fio will use the full size of the
392 given files or devices. If the files do not exist, size must be given. It is
393 also possible to give size as a percentage between 1 and 100. If size=20% is
394 given, fio will use 20% of the full size of the given files or devices.
396 .BI io_size \fR=\fPint "\fR,\fB io_limit \fR=\fPint
397 Normally fio operates within the region set by \fBsize\fR, which means that
398 the \fBsize\fR option sets both the region and size of IO to be performed.
399 Sometimes that is not what you want. With this option, it is possible to
400 define just the amount of IO that fio should do. For instance, if \fBsize\fR
401 is set to 20G and \fBio_limit\fR is set to 5G, fio will perform IO within
402 the first 20G but exit when 5G have been done. The opposite is also
403 possible - if \fBsize\fR is set to 20G, and \fBio_size\fR is set to 40G, then
404 fio will do 40G of IO within the 0..20G region.
406 .BI fill_device \fR=\fPbool "\fR,\fB fill_fs" \fR=\fPbool
407 Sets size to something really large and waits for ENOSPC (no space left on
408 device) as the terminating condition. Only makes sense with sequential write.
409 For a read workload, the mount point will be filled first then IO started on
410 the result. This option doesn't make sense if operating on a raw device node,
411 since the size of that is already known by the file system. Additionally,
412 writing beyond end-of-device will not return ENOSPC there.
414 .BI filesize \fR=\fPirange
415 Individual file sizes. May be a range, in which case \fBfio\fR will select sizes
416 for files at random within the given range, limited to \fBsize\fR in total (if
417 that is given). If \fBfilesize\fR is not specified, each created file is the
420 .BI file_append \fR=\fPbool
421 Perform IO after the end of the file. Normally fio will operate within the
422 size of a file. If this option is set, then fio will append to the file
423 instead. This has identical behavior to setting \fRoffset\fP to the size
424 of a file. This option is ignored on non-regular files.
426 .BI blocksize \fR=\fPint[,int] "\fR,\fB bs" \fR=\fPint[,int]
427 Block size for I/O units. Default: 4k. Values for reads, writes, and trims
428 can be specified separately in the format \fIread\fR,\fIwrite\fR,\fItrim\fR
429 either of which may be empty to leave that value at its default. If a trailing
430 comma isn't given, the remainder will inherit the last value set.
432 .BI blocksize_range \fR=\fPirange[,irange] "\fR,\fB bsrange" \fR=\fPirange[,irange]
433 Specify a range of I/O block sizes. The issued I/O unit will always be a
434 multiple of the minimum size, unless \fBblocksize_unaligned\fR is set. Applies
435 to both reads and writes if only one range is given, but can be specified
436 separately with a comma separating the values. Example: bsrange=1k-4k,2k-8k.
437 Also (see \fBblocksize\fR).
439 .BI bssplit \fR=\fPstr
440 This option allows even finer grained control of the block sizes issued,
441 not just even splits between them. With this option, you can weight various
442 block sizes for exact control of the issued IO for a job that has mixed
443 block sizes. The format of the option is bssplit=blocksize/percentage,
444 optionally adding as many definitions as needed separated by a colon.
445 Example: bssplit=4k/10:64k/50:32k/40 would issue 50% 64k blocks, 10% 4k
446 blocks and 40% 32k blocks. \fBbssplit\fR also supports giving separate
447 splits to reads and writes. The format is identical to what the
448 \fBbs\fR option accepts, the read and write parts are separated with a
451 .B blocksize_unaligned\fR,\fP bs_unaligned
452 If set, any size in \fBblocksize_range\fR may be used. This typically won't
453 work with direct I/O, as that normally requires sector alignment.
455 .BI blockalign \fR=\fPint[,int] "\fR,\fB ba" \fR=\fPint[,int]
456 At what boundary to align random IO offsets. Defaults to the same as 'blocksize'
457 the minimum blocksize given. Minimum alignment is typically 512b
458 for using direct IO, though it usually depends on the hardware block size.
459 This option is mutually exclusive with using a random map for files, so it
460 will turn off that option.
462 .BI bs_is_seq_rand \fR=\fPbool
463 If this option is set, fio will use the normal read,write blocksize settings as
464 sequential,random instead. Any random read or write will use the WRITE
465 blocksize settings, and any sequential read or write will use the READ
469 Initialize buffers with all zeros. Default: fill buffers with random data.
472 If this option is given, fio will refill the IO buffers on every submit. The
473 default is to only fill it at init time and reuse that data. Only makes sense
474 if zero_buffers isn't specified, naturally. If data verification is enabled,
475 refill_buffers is also automatically enabled.
477 .BI scramble_buffers \fR=\fPbool
478 If \fBrefill_buffers\fR is too costly and the target is using data
479 deduplication, then setting this option will slightly modify the IO buffer
480 contents to defeat normal de-dupe attempts. This is not enough to defeat
481 more clever block compression attempts, but it will stop naive dedupe
482 of blocks. Default: true.
484 .BI buffer_compress_percentage \fR=\fPint
485 If this is set, then fio will attempt to provide IO buffer content (on WRITEs)
486 that compress to the specified level. Fio does this by providing a mix of
487 random data and a fixed pattern. The fixed pattern is either zeroes, or the
488 pattern specified by \fBbuffer_pattern\fR. If the pattern option is used, it
489 might skew the compression ratio slightly. Note that this is per block size
490 unit, for file/disk wide compression level that matches this setting. Note
491 that this is per block size unit, for file/disk wide compression level that
492 matches this setting, you'll also want to set refill_buffers.
494 .BI buffer_compress_chunk \fR=\fPint
495 See \fBbuffer_compress_percentage\fR. This setting allows fio to manage how
496 big the ranges of random data and zeroed data is. Without this set, fio will
497 provide \fBbuffer_compress_percentage\fR of blocksize random data, followed by
498 the remaining zeroed. With this set to some chunk size smaller than the block
499 size, fio can alternate random and zeroed data throughout the IO buffer.
501 .BI buffer_pattern \fR=\fPstr
502 If set, fio will fill the IO buffers with this pattern. If not set, the contents
503 of IO buffers is defined by the other options related to buffer contents. The
504 setting can be any pattern of bytes, and can be prefixed with 0x for hex
505 values. It may also be a string, where the string must then be wrapped with
508 .BI dedupe_percentage \fR=\fPint
509 If set, fio will generate this percentage of identical buffers when writing.
510 These buffers will be naturally dedupable. The contents of the buffers depend
511 on what other buffer compression settings have been set. It's possible to have
512 the individual buffers either fully compressible, or not at all. This option
513 only controls the distribution of unique buffers.
515 .BI nrfiles \fR=\fPint
516 Number of files to use for this job. Default: 1.
518 .BI openfiles \fR=\fPint
519 Number of files to keep open at the same time. Default: \fBnrfiles\fR.
521 .BI file_service_type \fR=\fPstr
522 Defines how files to service are selected. The following types are defined:
527 Choose a file at random.
530 Round robin over opened files (default).
533 Do each file in the set sequentially.
536 The number of I/Os to issue before switching to a new file can be specified by
537 appending `:\fIint\fR' to the service type.
540 .BI ioengine \fR=\fPstr
541 Defines how the job issues I/O. The following types are defined:
546 Basic \fBread\fR\|(2) or \fBwrite\fR\|(2) I/O. \fBfseek\fR\|(2) is used to
547 position the I/O location.
550 Basic \fBpread\fR\|(2) or \fBpwrite\fR\|(2) I/O.
553 Basic \fBreadv\fR\|(2) or \fBwritev\fR\|(2) I/O. Will emulate queuing by
554 coalescing adjacent IOs into a single submission.
557 Basic \fBpreadv\fR\|(2) or \fBpwritev\fR\|(2) I/O.
560 Linux native asynchronous I/O. This ioengine defines engine specific options.
563 POSIX asynchronous I/O using \fBaio_read\fR\|(3) and \fBaio_write\fR\|(3).
566 Solaris native asynchronous I/O.
569 Windows native asynchronous I/O.
572 File is memory mapped with \fBmmap\fR\|(2) and data copied using
576 \fBsplice\fR\|(2) is used to transfer the data and \fBvmsplice\fR\|(2) to
577 transfer data from user-space to the kernel.
580 Use the syslet system calls to make regular read/write asynchronous.
583 SCSI generic sg v3 I/O. May be either synchronous using the SG_IO ioctl, or if
584 the target is an sg character device, we use \fBread\fR\|(2) and
585 \fBwrite\fR\|(2) for asynchronous I/O.
588 Doesn't transfer any data, just pretends to. Mainly used to exercise \fBfio\fR
589 itself and for debugging and testing purposes.
592 Transfer over the network. The protocol to be used can be defined with the
593 \fBprotocol\fR parameter. Depending on the protocol, \fBfilename\fR,
594 \fBhostname\fR, \fBport\fR, or \fBlisten\fR must be specified.
595 This ioengine defines engine specific options.
598 Like \fBnet\fR, but uses \fBsplice\fR\|(2) and \fBvmsplice\fR\|(2) to map data
599 and send/receive. This ioengine defines engine specific options.
602 Doesn't transfer any data, but burns CPU cycles according to \fBcpuload\fR and
603 \fBcpucycles\fR parameters.
606 The GUASI I/O engine is the Generic Userspace Asynchronous Syscall Interface
607 approach to asynchronous I/O.
609 See <http://www.xmailserver.org/guasi\-lib.html>.
612 The RDMA I/O engine supports both RDMA memory semantics (RDMA_WRITE/RDMA_READ)
613 and channel semantics (Send/Recv) for the InfiniBand, RoCE and iWARP protocols.
616 Loads an external I/O engine object file. Append the engine filename as
620 IO engine that does regular linux native fallocate call to simulate data
621 transfer as fio ioengine
623 DDIR_READ does fallocate(,mode = FALLOC_FL_KEEP_SIZE,)
625 DIR_WRITE does fallocate(,mode = 0)
627 DDIR_TRIM does fallocate(,mode = FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE)
630 IO engine that does regular EXT4_IOC_MOVE_EXT ioctls to simulate defragment activity
631 request to DDIR_WRITE event
634 IO engine supporting direct access to Ceph Rados Block Devices (RBD) via librbd
635 without the need to use the kernel rbd driver. This ioengine defines engine specific
639 Using Glusterfs libgfapi sync interface to direct access to Glusterfs volumes without
640 having to go through FUSE. This ioengine defines engine specific
644 Using Glusterfs libgfapi async interface to direct access to Glusterfs volumes without
645 having to go through FUSE. This ioengine defines engine specific
649 Read and write through Hadoop (HDFS). The \fBfilename\fR option is used to
650 specify host,port of the hdfs name-node to connect. This engine interprets
651 offsets a little differently. In HDFS, files once created cannot be modified.
652 So random writes are not possible. To imitate this, libhdfs engine expects
653 bunch of small files to be created over HDFS, and engine will randomly pick a
654 file out of those files based on the offset generated by fio backend. (see the
655 example job file to create such files, use rw=write option). Please note, you
656 might want to set necessary environment variables to work with hdfs/libhdfs
660 Read, write and erase an MTD character device (e.g., /dev/mtd0). Discards are
661 treated as erases. Depending on the underlying device type, the I/O may have
662 to go in a certain pattern, e.g., on NAND, writing sequentially to erase blocks
663 and discarding before overwriting. The writetrim mode works well for this
669 .BI iodepth \fR=\fPint
670 Number of I/O units to keep in flight against the file. Note that increasing
671 iodepth beyond 1 will not affect synchronous ioengines (except for small
672 degress when verify_async is in use). Even async engines may impose OS
673 restrictions causing the desired depth not to be achieved. This may happen on
674 Linux when using libaio and not setting \fBdirect\fR=1, since buffered IO is
675 not async on that OS. Keep an eye on the IO depth distribution in the
676 fio output to verify that the achieved depth is as expected. Default: 1.
678 .BI iodepth_batch \fR=\fPint
679 Number of I/Os to submit at once. Default: \fBiodepth\fR.
681 .BI iodepth_batch_complete \fR=\fPint
682 This defines how many pieces of IO to retrieve at once. It defaults to 1 which
683 means that we'll ask for a minimum of 1 IO in the retrieval process from the
684 kernel. The IO retrieval will go on until we hit the limit set by
685 \fBiodepth_low\fR. If this variable is set to 0, then fio will always check for
686 completed events before queuing more IO. This helps reduce IO latency, at the
687 cost of more retrieval system calls.
689 .BI iodepth_low \fR=\fPint
690 Low watermark indicating when to start filling the queue again. Default:
693 .BI direct \fR=\fPbool
694 If true, use non-buffered I/O (usually O_DIRECT). Default: false.
696 .BI atomic \fR=\fPbool
697 If value is true, attempt to use atomic direct IO. Atomic writes are guaranteed
698 to be stable once acknowledged by the operating system. Only Linux supports
701 .BI buffered \fR=\fPbool
702 If true, use buffered I/O. This is the opposite of the \fBdirect\fR parameter.
705 .BI offset \fR=\fPint
706 Offset in the file to start I/O. Data before the offset will not be touched.
708 .BI offset_increment \fR=\fPint
709 If this is provided, then the real offset becomes the
710 offset + offset_increment * thread_number, where the thread number is a
711 counter that starts at 0 and is incremented for each sub-job (i.e. when
712 numjobs option is specified). This option is useful if there are several jobs
713 which are intended to operate on a file in parallel disjoint segments, with
714 even spacing between the starting points.
716 .BI number_ios \fR=\fPint
717 Fio will normally perform IOs until it has exhausted the size of the region
718 set by \fBsize\fR, or if it exhaust the allocated time (or hits an error
719 condition). With this setting, the range/size can be set independently of
720 the number of IOs to perform. When fio reaches this number, it will exit
721 normally and report status. Note that this does not extend the amount
722 of IO that will be done, it will only stop fio if this condition is met
723 before other end-of-job criteria.
726 How many I/Os to perform before issuing an \fBfsync\fR\|(2) of dirty data. If
727 0, don't sync. Default: 0.
729 .BI fdatasync \fR=\fPint
730 Like \fBfsync\fR, but uses \fBfdatasync\fR\|(2) instead to only sync the
731 data parts of the file. Default: 0.
733 .BI write_barrier \fR=\fPint
734 Make every Nth write a barrier write.
736 .BI sync_file_range \fR=\fPstr:int
737 Use \fBsync_file_range\fR\|(2) for every \fRval\fP number of write operations. Fio will
738 track range of writes that have happened since the last \fBsync_file_range\fR\|(2) call.
739 \fRstr\fP can currently be one or more of:
743 SYNC_FILE_RANGE_WAIT_BEFORE
746 SYNC_FILE_RANGE_WRITE
749 SYNC_FILE_RANGE_WRITE
753 So if you do sync_file_range=wait_before,write:8, fio would use
754 \fBSYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE\fP for every 8 writes.
755 Also see the \fBsync_file_range\fR\|(2) man page. This option is Linux specific.
757 .BI overwrite \fR=\fPbool
758 If writing, setup the file first and do overwrites. Default: false.
760 .BI end_fsync \fR=\fPbool
761 Sync file contents when a write stage has completed. Default: false.
763 .BI fsync_on_close \fR=\fPbool
764 If true, sync file contents on close. This differs from \fBend_fsync\fR in that
765 it will happen on every close, not just at the end of the job. Default: false.
767 .BI rwmixread \fR=\fPint
768 Percentage of a mixed workload that should be reads. Default: 50.
770 .BI rwmixwrite \fR=\fPint
771 Percentage of a mixed workload that should be writes. If \fBrwmixread\fR and
772 \fBrwmixwrite\fR are given and do not sum to 100%, the latter of the two
773 overrides the first. This may interfere with a given rate setting, if fio is
774 asked to limit reads or writes to a certain rate. If that is the case, then
775 the distribution may be skewed. Default: 50.
777 .BI random_distribution \fR=\fPstr:float
778 By default, fio will use a completely uniform random distribution when asked
779 to perform random IO. Sometimes it is useful to skew the distribution in
780 specific ways, ensuring that some parts of the data is more hot than others.
781 Fio includes the following distribution models:
785 Uniform random distribution
795 When using a zipf or pareto distribution, an input value is also needed to
796 define the access pattern. For zipf, this is the zipf theta. For pareto,
797 it's the pareto power. Fio includes a test program, genzipf, that can be
798 used visualize what the given input values will yield in terms of hit rates.
799 If you wanted to use zipf with a theta of 1.2, you would use
800 random_distribution=zipf:1.2 as the option. If a non-uniform model is used,
801 fio will disable use of the random map.
803 .BI percentage_random \fR=\fPint
804 For a random workload, set how big a percentage should be random. This defaults
805 to 100%, in which case the workload is fully random. It can be set from
806 anywhere from 0 to 100. Setting it to 0 would make the workload fully
807 sequential. It is possible to set different values for reads, writes, and
808 trim. To do so, simply use a comma separated list. See \fBblocksize\fR.
811 Normally \fBfio\fR will cover every block of the file when doing random I/O. If
812 this parameter is given, a new offset will be chosen without looking at past
813 I/O history. This parameter is mutually exclusive with \fBverify\fR.
815 .BI softrandommap \fR=\fPbool
816 See \fBnorandommap\fR. If fio runs with the random block map enabled and it
817 fails to allocate the map, if this option is set it will continue without a
818 random block map. As coverage will not be as complete as with random maps, this
819 option is disabled by default.
821 .BI random_generator \fR=\fPstr
822 Fio supports the following engines for generating IO offsets for random IO:
826 Strong 2^88 cycle random number generator
829 Linear feedback shift register generator
833 Tausworthe is a strong random number generator, but it requires tracking on the
834 side if we want to ensure that blocks are only read or written once. LFSR
835 guarantees that we never generate the same offset twice, and it's also less
836 computationally expensive. It's not a true random generator, however, though
837 for IO purposes it's typically good enough. LFSR only works with single block
838 sizes, not with workloads that use multiple block sizes. If used with such a
839 workload, fio may read or write some blocks multiple times.
842 Run job with given nice value. See \fBnice\fR\|(2).
845 Set I/O priority value of this job between 0 (highest) and 7 (lowest). See
848 .BI prioclass \fR=\fPint
849 Set I/O priority class. See \fBionice\fR\|(1).
851 .BI thinktime \fR=\fPint
852 Stall job for given number of microseconds between issuing I/Os.
854 .BI thinktime_spin \fR=\fPint
855 Pretend to spend CPU time for given number of microseconds, sleeping the rest
856 of the time specified by \fBthinktime\fR. Only valid if \fBthinktime\fR is set.
858 .BI thinktime_blocks \fR=\fPint
859 Only valid if thinktime is set - control how many blocks to issue, before
860 waiting \fBthinktime\fR microseconds. If not set, defaults to 1 which will
861 make fio wait \fBthinktime\fR microseconds after every block. This
862 effectively makes any queue depth setting redundant, since no more than 1 IO
863 will be queued before we have to complete it and do our thinktime. In other
864 words, this setting effectively caps the queue depth if the latter is larger.
868 Cap bandwidth used by this job. The number is in bytes/sec, the normal postfix
869 rules apply. You can use \fBrate\fR=500k to limit reads and writes to 500k each,
870 or you can specify read and writes separately. Using \fBrate\fR=1m,500k would
871 limit reads to 1MB/sec and writes to 500KB/sec. Capping only reads or writes
872 can be done with \fBrate\fR=,500k or \fBrate\fR=500k,. The former will only
873 limit writes (to 500KB/sec), the latter will only limit reads.
875 .BI ratemin \fR=\fPint
876 Tell \fBfio\fR to do whatever it can to maintain at least the given bandwidth.
877 Failing to meet this requirement will cause the job to exit. The same format
878 as \fBrate\fR is used for read vs write separation.
880 .BI rate_iops \fR=\fPint
881 Cap the bandwidth to this number of IOPS. Basically the same as rate, just
882 specified independently of bandwidth. The same format as \fBrate\fR is used for
883 read vs write separation. If \fBblocksize\fR is a range, the smallest block
884 size is used as the metric.
886 .BI rate_iops_min \fR=\fPint
887 If this rate of I/O is not met, the job will exit. The same format as \fBrate\fR
888 is used for read vs write separation.
890 .BI ratecycle \fR=\fPint
891 Average bandwidth for \fBrate\fR and \fBratemin\fR over this number of
892 milliseconds. Default: 1000ms.
894 .BI latency_target \fR=\fPint
895 If set, fio will attempt to find the max performance point that the given
896 workload will run at while maintaining a latency below this target. The
897 values is given in microseconds. See \fBlatency_window\fR and
898 \fBlatency_percentile\fR.
900 .BI latency_window \fR=\fPint
901 Used with \fBlatency_target\fR to specify the sample window that the job
902 is run at varying queue depths to test the performance. The value is given
905 .BI latency_percentile \fR=\fPfloat
906 The percentage of IOs that must fall within the criteria specified by
907 \fBlatency_target\fR and \fBlatency_window\fR. If not set, this defaults
908 to 100.0, meaning that all IOs must be equal or below to the value set
909 by \fBlatency_target\fR.
911 .BI max_latency \fR=\fPint
912 If set, fio will exit the job if it exceeds this maximum latency. It will exit
915 .BI cpumask \fR=\fPint
916 Set CPU affinity for this job. \fIint\fR is a bitmask of allowed CPUs the job
917 may run on. See \fBsched_setaffinity\fR\|(2).
919 .BI cpus_allowed \fR=\fPstr
920 Same as \fBcpumask\fR, but allows a comma-delimited list of CPU numbers.
922 .BI cpus_allowed_policy \fR=\fPstr
923 Set the policy of how fio distributes the CPUs specified by \fBcpus_allowed\fR
924 or \fBcpumask\fR. Two policies are supported:
929 All jobs will share the CPU set specified.
932 Each job will get a unique CPU from the CPU set.
935 \fBshared\fR is the default behaviour, if the option isn't specified. If
936 \fBsplit\fR is specified, then fio will assign one cpu per job. If not enough
937 CPUs are given for the jobs listed, then fio will roundrobin the CPUs in
942 .BI numa_cpu_nodes \fR=\fPstr
943 Set this job running on specified NUMA nodes' CPUs. The arguments allow
944 comma delimited list of cpu numbers, A-B ranges, or 'all'.
946 .BI numa_mem_policy \fR=\fPstr
947 Set this job's memory policy and corresponding NUMA nodes. Format of
951 .B <mode>[:<nodelist>]
954 is one of the following memory policy:
956 .B default, prefer, bind, interleave, local
959 For \fBdefault\fR and \fBlocal\fR memory policy, no \fBnodelist\fR is
960 needed to be specified. For \fBprefer\fR, only one node is
961 allowed. For \fBbind\fR and \fBinterleave\fR, \fBnodelist\fR allows
962 comma delimited list of numbers, A-B ranges, or 'all'.
964 .BI startdelay \fR=\fPirange
965 Delay start of job for the specified number of seconds. Supports all time
966 suffixes to allow specification of hours, minutes, seconds and
967 milliseconds - seconds are the default if a unit is omitted.
968 Can be given as a range which causes each thread to choose randomly out of the
971 .BI runtime \fR=\fPint
972 Terminate processing after the specified number of seconds.
975 If given, run for the specified \fBruntime\fR duration even if the files are
976 completely read or written. The same workload will be repeated as many times
977 as \fBruntime\fR allows.
979 .BI ramp_time \fR=\fPint
980 If set, fio will run the specified workload for this amount of time before
981 logging any performance numbers. Useful for letting performance settle before
982 logging results, thus minimizing the runtime required for stable results. Note
983 that the \fBramp_time\fR is considered lead in time for a job, thus it will
984 increase the total runtime if a special timeout or runtime is specified.
986 .BI invalidate \fR=\fPbool
987 Invalidate buffer-cache for the file prior to starting I/O. Default: true.
990 Use synchronous I/O for buffered writes. For the majority of I/O engines,
991 this means using O_SYNC. Default: false.
993 .BI iomem \fR=\fPstr "\fR,\fP mem" \fR=\fPstr
994 Allocation method for I/O unit buffer. Allowed values are:
999 Allocate memory with \fBmalloc\fR\|(3).
1002 Use shared memory buffers allocated through \fBshmget\fR\|(2).
1005 Same as \fBshm\fR, but use huge pages as backing.
1008 Use \fBmmap\fR\|(2) for allocation. Uses anonymous memory unless a filename
1009 is given after the option in the format `:\fIfile\fR'.
1012 Same as \fBmmap\fR, but use huge files as backing.
1015 The amount of memory allocated is the maximum allowed \fBblocksize\fR for the
1016 job multiplied by \fBiodepth\fR. For \fBshmhuge\fR or \fBmmaphuge\fR to work,
1017 the system must have free huge pages allocated. \fBmmaphuge\fR also needs to
1018 have hugetlbfs mounted, and \fIfile\fR must point there. At least on Linux,
1019 huge pages must be manually allocated. See \fB/proc/sys/vm/nr_hugehages\fR
1020 and the documentation for that. Normally you just need to echo an appropriate
1021 number, eg echoing 8 will ensure that the OS has 8 huge pages ready for
1025 .BI iomem_align \fR=\fPint "\fR,\fP mem_align" \fR=\fPint
1026 This indicates the memory alignment of the IO memory buffers. Note that the
1027 given alignment is applied to the first IO unit buffer, if using \fBiodepth\fR
1028 the alignment of the following buffers are given by the \fBbs\fR used. In
1029 other words, if using a \fBbs\fR that is a multiple of the page sized in the
1030 system, all buffers will be aligned to this value. If using a \fBbs\fR that
1031 is not page aligned, the alignment of subsequent IO memory buffers is the
1032 sum of the \fBiomem_align\fR and \fBbs\fR used.
1034 .BI hugepage\-size \fR=\fPint
1035 Defines the size of a huge page. Must be at least equal to the system setting.
1036 Should be a multiple of 1MB. Default: 4MB.
1039 Terminate all jobs when one finishes. Default: wait for each job to finish.
1041 .BI bwavgtime \fR=\fPint
1042 Average bandwidth calculations over the given time in milliseconds. Default:
1045 .BI iopsavgtime \fR=\fPint
1046 Average IOPS calculations over the given time in milliseconds. Default:
1049 .BI create_serialize \fR=\fPbool
1050 If true, serialize file creation for the jobs. Default: true.
1052 .BI create_fsync \fR=\fPbool
1053 \fBfsync\fR\|(2) data file after creation. Default: true.
1055 .BI create_on_open \fR=\fPbool
1056 If true, the files are not created until they are opened for IO by the job.
1058 .BI create_only \fR=\fPbool
1059 If true, fio will only run the setup phase of the job. If files need to be
1060 laid out or updated on disk, only that will be done. The actual job contents
1063 .BI pre_read \fR=\fPbool
1064 If this is given, files will be pre-read into memory before starting the given
1065 IO operation. This will also clear the \fR \fBinvalidate\fR flag, since it is
1066 pointless to pre-read and then drop the cache. This will only work for IO
1067 engines that are seekable, since they allow you to read the same data
1068 multiple times. Thus it will not work on eg network or splice IO.
1070 .BI unlink \fR=\fPbool
1071 Unlink job files when done. Default: false.
1073 .BI loops \fR=\fPint
1074 Specifies the number of iterations (runs of the same workload) of this job.
1077 .BI verify_only \fR=\fPbool
1078 Do not perform the specified workload, only verify data still matches previous
1079 invocation of this workload. This option allows one to check data multiple
1080 times at a later date without overwriting it. This option makes sense only for
1081 workloads that write data, and does not support workloads with the
1082 \fBtime_based\fR option set.
1084 .BI do_verify \fR=\fPbool
1085 Run the verify phase after a write phase. Only valid if \fBverify\fR is set.
1088 .BI verify \fR=\fPstr
1089 Method of verifying file contents after each iteration of the job. Allowed
1094 .B md5 crc16 crc32 crc32c crc32c-intel crc64 crc7 sha256 sha512 sha1 xxhash
1095 Store appropriate checksum in the header of each block. crc32c-intel is
1096 hardware accelerated SSE4.2 driven, falls back to regular crc32c if
1097 not supported by the system.
1100 Write extra information about each I/O (timestamp, block number, etc.). The
1101 block number is verified. See \fBverify_pattern\fR as well.
1104 Pretend to verify. Used for testing internals.
1107 This option can be used for repeated burn-in tests of a system to make sure
1108 that the written data is also correctly read back. If the data direction given
1109 is a read or random read, fio will assume that it should verify a previously
1110 written file. If the data direction includes any form of write, the verify will
1111 be of the newly written data.
1114 .BI verifysort \fR=\fPbool
1115 If true, written verify blocks are sorted if \fBfio\fR deems it to be faster to
1116 read them back in a sorted manner. Default: true.
1118 .BI verifysort_nr \fR=\fPint
1119 Pre-load and sort verify blocks for a read workload.
1121 .BI verify_offset \fR=\fPint
1122 Swap the verification header with data somewhere else in the block before
1123 writing. It is swapped back before verifying.
1125 .BI verify_interval \fR=\fPint
1126 Write the verification header for this number of bytes, which should divide
1127 \fBblocksize\fR. Default: \fBblocksize\fR.
1129 .BI verify_pattern \fR=\fPstr
1130 If set, fio will fill the io buffers with this pattern. Fio defaults to filling
1131 with totally random bytes, but sometimes it's interesting to fill with a known
1132 pattern for io verification purposes. Depending on the width of the pattern,
1133 fio will fill 1/2/3/4 bytes of the buffer at the time(it can be either a
1134 decimal or a hex number). The verify_pattern if larger than a 32-bit quantity
1135 has to be a hex number that starts with either "0x" or "0X". Use with
1138 .BI verify_fatal \fR=\fPbool
1139 If true, exit the job on the first observed verification failure. Default:
1142 .BI verify_dump \fR=\fPbool
1143 If set, dump the contents of both the original data block and the data block we
1144 read off disk to files. This allows later analysis to inspect just what kind of
1145 data corruption occurred. Off by default.
1147 .BI verify_async \fR=\fPint
1148 Fio will normally verify IO inline from the submitting thread. This option
1149 takes an integer describing how many async offload threads to create for IO
1150 verification instead, causing fio to offload the duty of verifying IO contents
1151 to one or more separate threads. If using this offload option, even sync IO
1152 engines can benefit from using an \fBiodepth\fR setting higher than 1, as it
1153 allows them to have IO in flight while verifies are running.
1155 .BI verify_async_cpus \fR=\fPstr
1156 Tell fio to set the given CPU affinity on the async IO verification threads.
1157 See \fBcpus_allowed\fP for the format used.
1159 .BI verify_backlog \fR=\fPint
1160 Fio will normally verify the written contents of a job that utilizes verify
1161 once that job has completed. In other words, everything is written then
1162 everything is read back and verified. You may want to verify continually
1163 instead for a variety of reasons. Fio stores the meta data associated with an
1164 IO block in memory, so for large verify workloads, quite a bit of memory would
1165 be used up holding this meta data. If this option is enabled, fio will write
1166 only N blocks before verifying these blocks.
1168 .BI verify_backlog_batch \fR=\fPint
1169 Control how many blocks fio will verify if verify_backlog is set. If not set,
1170 will default to the value of \fBverify_backlog\fR (meaning the entire queue is
1171 read back and verified). If \fBverify_backlog_batch\fR is less than
1172 \fBverify_backlog\fR then not all blocks will be verified, if
1173 \fBverify_backlog_batch\fR is larger than \fBverify_backlog\fR, some blocks
1174 will be verified more than once.
1176 .BI trim_percentage \fR=\fPint
1177 Number of verify blocks to discard/trim.
1179 .BI trim_verify_zero \fR=\fPbool
1180 Verify that trim/discarded blocks are returned as zeroes.
1182 .BI trim_backlog \fR=\fPint
1183 Trim after this number of blocks are written.
1185 .BI trim_backlog_batch \fR=\fPint
1186 Trim this number of IO blocks.
1188 .BI experimental_verify \fR=\fPbool
1189 Enable experimental verification.
1191 .BI verify_state_save \fR=\fPbool
1192 When a job exits during the write phase of a verify workload, save its
1193 current state. This allows fio to replay up until that point, if the
1194 verify state is loaded for the verify read phase.
1196 .BI verify_state_load \fR=\fPbool
1197 If a verify termination trigger was used, fio stores the current write
1198 state of each thread. This can be used at verification time so that fio
1199 knows how far it should verify. Without this information, fio will run
1200 a full verification pass, according to the settings in the job file used.
1202 .B stonewall "\fR,\fP wait_for_previous"
1203 Wait for preceding jobs in the job file to exit before starting this one.
1204 \fBstonewall\fR implies \fBnew_group\fR.
1207 Start a new reporting group. If not given, all jobs in a file will be part
1208 of the same reporting group, unless separated by a stonewall.
1210 .BI numjobs \fR=\fPint
1211 Number of clones (processes/threads performing the same workload) of this job.
1215 If set, display per-group reports instead of per-job when \fBnumjobs\fR is
1219 Use threads created with \fBpthread_create\fR\|(3) instead of processes created
1220 with \fBfork\fR\|(2).
1222 .BI zonesize \fR=\fPint
1223 Divide file into zones of the specified size in bytes. See \fBzoneskip\fR.
1225 .BI zonerange \fR=\fPint
1226 Give size of an IO zone. See \fBzoneskip\fR.
1228 .BI zoneskip \fR=\fPint
1229 Skip the specified number of bytes when \fBzonesize\fR bytes of data have been
1232 .BI write_iolog \fR=\fPstr
1233 Write the issued I/O patterns to the specified file. Specify a separate file
1234 for each job, otherwise the iologs will be interspersed and the file may be
1237 .BI read_iolog \fR=\fPstr
1238 Replay the I/O patterns contained in the specified file generated by
1239 \fBwrite_iolog\fR, or may be a \fBblktrace\fR binary file.
1241 .BI replay_no_stall \fR=\fPint
1242 While replaying I/O patterns using \fBread_iolog\fR the default behavior
1243 attempts to respect timing information between I/Os. Enabling
1244 \fBreplay_no_stall\fR causes I/Os to be replayed as fast as possible while
1245 still respecting ordering.
1247 .BI replay_redirect \fR=\fPstr
1248 While replaying I/O patterns using \fBread_iolog\fR the default behavior
1249 is to replay the IOPS onto the major/minor device that each IOP was recorded
1250 from. Setting \fBreplay_redirect\fR causes all IOPS to be replayed onto the
1251 single specified device regardless of the device it was recorded from.
1253 .BI write_bw_log \fR=\fPstr
1254 If given, write a bandwidth log of the jobs in this job file. Can be used to
1255 store data of the bandwidth of the jobs in their lifetime. The included
1256 fio_generate_plots script uses gnuplot to turn these text files into nice
1257 graphs. See \fBwrite_lat_log\fR for behaviour of given filename. For this
1258 option, the postfix is _bw.x.log, where x is the index of the job (1..N,
1259 where N is the number of jobs)
1261 .BI write_lat_log \fR=\fPstr
1262 Same as \fBwrite_bw_log\fR, but writes I/O completion latencies. If no
1263 filename is given with this option, the default filename of
1264 "jobname_type.x.log" is used, where x is the index of the job (1..N, where
1265 N is the number of jobs). Even if the filename is given, fio will still
1266 append the type of log.
1268 .BI write_iops_log \fR=\fPstr
1269 Same as \fBwrite_bw_log\fR, but writes IOPS. If no filename is given with this
1270 option, the default filename of "jobname_type.x.log" is used, where x is the
1271 index of the job (1..N, where N is the number of jobs). Even if the filename
1272 is given, fio will still append the type of log.
1274 .BI log_avg_msec \fR=\fPint
1275 By default, fio will log an entry in the iops, latency, or bw log for every
1276 IO that completes. When writing to the disk log, that can quickly grow to a
1277 very large size. Setting this option makes fio average the each log entry
1278 over the specified period of time, reducing the resolution of the log.
1281 .BI log_offset \fR=\fPbool
1282 If this is set, the iolog options will include the byte offset for the IO
1283 entry as well as the other data values.
1285 .BI log_compression \fR=\fPint
1286 If this is set, fio will compress the IO logs as it goes, to keep the memory
1287 footprint lower. When a log reaches the specified size, that chunk is removed
1288 and compressed in the background. Given that IO logs are fairly highly
1289 compressible, this yields a nice memory savings for longer runs. The downside
1290 is that the compression will consume some background CPU cycles, so it may
1291 impact the run. This, however, is also true if the logging ends up consuming
1292 most of the system memory. So pick your poison. The IO logs are saved
1293 normally at the end of a run, by decompressing the chunks and storing them
1294 in the specified log file. This feature depends on the availability of zlib.
1296 .BI log_store_compressed \fR=\fPbool
1297 If set, and \fBlog\fR_compression is also set, fio will store the log files in
1298 a compressed format. They can be decompressed with fio, using the
1299 \fB\-\-inflate-log\fR command line parameter. The files will be stored with a
1302 .BI block_error_percentiles \fR=\fPbool
1303 If set, record errors in trim block-sized units from writes and trims and output
1304 a histogram of how many trims it took to get to errors, and what kind of error
1307 .BI disable_lat \fR=\fPbool
1308 Disable measurements of total latency numbers. Useful only for cutting
1309 back the number of calls to \fBgettimeofday\fR\|(2), as that does impact performance at
1310 really high IOPS rates. Note that to really get rid of a large amount of these
1311 calls, this option must be used with disable_slat and disable_bw as well.
1313 .BI disable_clat \fR=\fPbool
1314 Disable measurements of completion latency numbers. See \fBdisable_lat\fR.
1316 .BI disable_slat \fR=\fPbool
1317 Disable measurements of submission latency numbers. See \fBdisable_lat\fR.
1319 .BI disable_bw_measurement \fR=\fPbool
1320 Disable measurements of throughput/bandwidth numbers. See \fBdisable_lat\fR.
1322 .BI lockmem \fR=\fPint
1323 Pin the specified amount of memory with \fBmlock\fR\|(2). Can be used to
1324 simulate a smaller amount of memory. The amount specified is per worker.
1326 .BI exec_prerun \fR=\fPstr
1327 Before running the job, execute the specified command with \fBsystem\fR\|(3).
1329 Output is redirected in a file called \fBjobname.prerun.txt\fR
1332 .BI exec_postrun \fR=\fPstr
1333 Same as \fBexec_prerun\fR, but the command is executed after the job completes.
1335 Output is redirected in a file called \fBjobname.postrun.txt\fR
1338 .BI ioscheduler \fR=\fPstr
1339 Attempt to switch the device hosting the file to the specified I/O scheduler.
1341 .BI disk_util \fR=\fPbool
1342 Generate disk utilization statistics if the platform supports it. Default: true.
1344 .BI clocksource \fR=\fPstr
1345 Use the given clocksource as the base of timing. The supported options are:
1349 \fBgettimeofday\fR\|(2)
1352 \fBclock_gettime\fR\|(2)
1355 Internal CPU clock source
1359 \fBcpu\fR is the preferred clocksource if it is reliable, as it is very fast
1360 (and fio is heavy on time calls). Fio will automatically use this clocksource
1361 if it's supported and considered reliable on the system it is running on,
1362 unless another clocksource is specifically set. For x86/x86-64 CPUs, this
1363 means supporting TSC Invariant.
1365 .BI gtod_reduce \fR=\fPbool
1366 Enable all of the \fBgettimeofday\fR\|(2) reducing options (disable_clat, disable_slat,
1367 disable_bw) plus reduce precision of the timeout somewhat to really shrink the
1368 \fBgettimeofday\fR\|(2) call count. With this option enabled, we only do about 0.4% of
1369 the gtod() calls we would have done if all time keeping was enabled.
1371 .BI gtod_cpu \fR=\fPint
1372 Sometimes it's cheaper to dedicate a single thread of execution to just getting
1373 the current time. Fio (and databases, for instance) are very intensive on
1374 \fBgettimeofday\fR\|(2) calls. With this option, you can set one CPU aside for doing
1375 nothing but logging current time to a shared memory location. Then the other
1376 threads/processes that run IO workloads need only copy that segment, instead of
1377 entering the kernel with a \fBgettimeofday\fR\|(2) call. The CPU set aside for doing
1378 these time calls will be excluded from other uses. Fio will manually clear it
1379 from the CPU mask of other jobs.
1381 .BI ignore_error \fR=\fPstr
1382 Sometimes you want to ignore some errors during test in that case you can specify
1383 error list for each error type.
1385 ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST
1387 errors for given error type is separated with ':'.
1388 Error may be symbol ('ENOSPC', 'ENOMEM') or an integer.
1390 Example: ignore_error=EAGAIN,ENOSPC:122 .
1392 This option will ignore EAGAIN from READ, and ENOSPC and 122(EDQUOT) from WRITE.
1394 .BI error_dump \fR=\fPbool
1395 If set dump every error even if it is non fatal, true by default. If disabled
1396 only fatal error will be dumped
1398 .BI profile \fR=\fPstr
1399 Select a specific builtin performance test.
1401 .BI cgroup \fR=\fPstr
1402 Add job to this control group. If it doesn't exist, it will be created.
1403 The system must have a mounted cgroup blkio mount point for this to work. If
1404 your system doesn't have it mounted, you can do so with:
1406 # mount \-t cgroup \-o blkio none /cgroup
1408 .BI cgroup_weight \fR=\fPint
1409 Set the weight of the cgroup to this value. See the documentation that comes
1410 with the kernel, allowed values are in the range of 100..1000.
1412 .BI cgroup_nodelete \fR=\fPbool
1413 Normally fio will delete the cgroups it has created after the job completion.
1414 To override this behavior and to leave cgroups around after the job completion,
1415 set cgroup_nodelete=1. This can be useful if one wants to inspect various
1416 cgroup files after job completion. Default: false
1419 Instead of running as the invoking user, set the user ID to this value before
1420 the thread/process does any work.
1423 Set group ID, see \fBuid\fR.
1425 .BI unit_base \fR=\fPint
1426 Base unit for reporting. Allowed values are:
1430 Use auto-detection (default).
1440 .BI flow_id \fR=\fPint
1441 The ID of the flow. If not specified, it defaults to being a global flow. See
1445 Weight in token-based flow control. If this value is used, then there is a
1446 \fBflow counter\fR which is used to regulate the proportion of activity between
1447 two or more jobs. fio attempts to keep this flow counter near zero. The
1448 \fBflow\fR parameter stands for how much should be added or subtracted to the
1449 flow counter on each iteration of the main I/O loop. That is, if one job has
1450 \fBflow=8\fR and another job has \fBflow=-1\fR, then there will be a roughly
1451 1:8 ratio in how much one runs vs the other.
1453 .BI flow_watermark \fR=\fPint
1454 The maximum value that the absolute value of the flow counter is allowed to
1455 reach before the job must wait for a lower value of the counter.
1457 .BI flow_sleep \fR=\fPint
1458 The period of time, in microseconds, to wait after the flow watermark has been
1459 exceeded before retrying operations
1461 .BI clat_percentiles \fR=\fPbool
1462 Enable the reporting of percentiles of completion latencies.
1464 .BI percentile_list \fR=\fPfloat_list
1465 Overwrite the default list of percentiles for completion latencies and the
1466 block error histogram. Each number is a floating number in the range (0,100],
1467 and the maximum length of the list is 20. Use ':' to separate the
1468 numbers. For example, \-\-percentile_list=99.5:99.9 will cause fio to
1469 report the values of completion latency below which 99.5% and 99.9% of
1470 the observed latencies fell, respectively.
1471 .SS "Ioengine Parameters List"
1472 Some parameters are only valid when a specific ioengine is in use. These are
1473 used identically to normal parameters, with the caveat that when used on the
1474 command line, they must come after the ioengine.
1476 .BI (cpu)cpuload \fR=\fPint
1477 Attempt to use the specified percentage of CPU cycles.
1479 .BI (cpu)cpuchunks \fR=\fPint
1480 Split the load into cycles of the given time. In microseconds.
1482 .BI (cpu)exit_on_io_done \fR=\fPbool
1483 Detect when IO threads are done, then exit.
1485 .BI (libaio)userspace_reap
1486 Normally, with the libaio engine in use, fio will use
1487 the io_getevents system call to reap newly returned events.
1488 With this flag turned on, the AIO ring will be read directly
1489 from user-space to reap events. The reaping mode is only
1490 enabled when polling for a minimum of 0 events (eg when
1491 iodepth_batch_complete=0).
1493 .BI (net,netsplice)hostname \fR=\fPstr
1494 The host name or IP address to use for TCP or UDP based IO.
1495 If the job is a TCP listener or UDP reader, the hostname is not
1496 used and must be omitted unless it is a valid UDP multicast address.
1498 .BI (net,netsplice)port \fR=\fPint
1499 The TCP or UDP port to bind to or connect to. If this is used with
1500 \fBnumjobs\fR to spawn multiple instances of the same job type, then
1501 this will be the starting port number since fio will use a range of ports.
1503 .BI (net,netsplice)interface \fR=\fPstr
1504 The IP address of the network interface used to send or receive UDP multicast
1507 .BI (net,netsplice)ttl \fR=\fPint
1508 Time-to-live value for outgoing UDP multicast packets. Default: 1
1510 .BI (net,netsplice)nodelay \fR=\fPbool
1511 Set TCP_NODELAY on TCP connections.
1513 .BI (net,netsplice)protocol \fR=\fPstr "\fR,\fP proto" \fR=\fPstr
1514 The network protocol to use. Accepted values are:
1519 Transmission control protocol
1522 Transmission control protocol V6
1525 User datagram protocol
1528 User datagram protocol V6
1534 When the protocol is TCP or UDP, the port must also be given,
1535 as well as the hostname if the job is a TCP listener or UDP
1536 reader. For unix sockets, the normal filename option should be
1537 used and the port is invalid.
1540 .BI (net,netsplice)listen
1541 For TCP network connections, tell fio to listen for incoming
1542 connections rather than initiating an outgoing connection. The
1543 hostname must be omitted if this option is used.
1545 .BI (net, pingpong) \fR=\fPbool
1546 Normally a network writer will just continue writing data, and a network reader
1547 will just consume packets. If pingpong=1 is set, a writer will send its normal
1548 payload to the reader, then wait for the reader to send the same payload back.
1549 This allows fio to measure network latencies. The submission and completion
1550 latencies then measure local time spent sending or receiving, and the
1551 completion latency measures how long it took for the other end to receive and
1552 send back. For UDP multicast traffic pingpong=1 should only be set for a single
1553 reader when multiple readers are listening to the same address.
1555 .BI (net, window_size) \fR=\fPint
1556 Set the desired socket buffer size for the connection.
1558 .BI (net, mss) \fR=\fPint
1559 Set the TCP maximum segment size (TCP_MAXSEG).
1561 .BI (e4defrag,donorname) \fR=\fPstr
1562 File will be used as a block donor (swap extents between files)
1564 .BI (e4defrag,inplace) \fR=\fPint
1565 Configure donor file block allocation strategy
1568 Preallocate donor's file on init
1571 allocate space immediately inside defragment event, and free right after event
1574 .BI (rbd)rbdname \fR=\fPstr
1575 Specifies the name of the RBD.
1577 .BI (rbd)pool \fR=\fPstr
1578 Specifies the name of the Ceph pool containing the RBD.
1580 .BI (rbd)clientname \fR=\fPstr
1581 Specifies the username (without the 'client.' prefix) used to access the Ceph cluster.
1583 .BI (mtd)skipbad \fR=\fPbool
1584 Skip operations against known bad blocks.
1586 While running, \fBfio\fR will display the status of the created jobs. For
1590 Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
1593 The characters in the first set of brackets denote the current status of each
1594 threads. The possible values are:
1600 Setup but not started.
1606 Initialized, waiting.
1609 Running, doing sequential reads.
1612 Running, doing random reads.
1615 Running, doing sequential writes.
1618 Running, doing random writes.
1621 Running, doing mixed sequential reads/writes.
1624 Running, doing mixed random reads/writes.
1627 Running, currently waiting for \fBfsync\fR\|(2).
1630 Running, verifying written data.
1633 Exited, not reaped by main thread.
1636 Exited, thread reaped.
1640 The second set of brackets shows the estimated completion percentage of
1641 the current group. The third set shows the read and write I/O rate,
1642 respectively. Finally, the estimated run time of the job is displayed.
1644 When \fBfio\fR completes (or is interrupted by Ctrl-C), it will show data
1645 for each thread, each group of threads, and each disk, in that order.
1647 Per-thread statistics first show the threads client number, group-id, and
1648 error code. The remaining figures are as follows:
1652 Number of megabytes of I/O performed.
1655 Average data rate (bandwidth).
1661 Submission latency minimum, maximum, average and standard deviation. This is
1662 the time it took to submit the I/O.
1665 Completion latency minimum, maximum, average and standard deviation. This
1666 is the time between submission and completion.
1669 Bandwidth minimum, maximum, percentage of aggregate bandwidth received, average
1670 and standard deviation.
1673 CPU usage statistics. Includes user and system time, number of context switches
1674 this thread went through and number of major and minor page faults.
1677 Distribution of I/O depths. Each depth includes everything less than (or equal)
1678 to it, but greater than the previous depth.
1681 Number of read/write requests issued, and number of short read/write requests.
1684 Distribution of I/O completion latencies. The numbers follow the same pattern
1688 The group statistics show:
1693 Number of megabytes I/O performed.
1696 Aggregate bandwidth of threads in the group.
1699 Minimum average bandwidth a thread saw.
1702 Maximum average bandwidth a thread saw.
1705 Shortest runtime of threads in the group.
1708 Longest runtime of threads in the group.
1712 Finally, disk statistics are printed with reads first:
1717 Number of I/Os performed by all groups.
1720 Number of merges in the I/O scheduler.
1723 Number of ticks we kept the disk busy.
1726 Total time spent in the disk queue.
1733 It is also possible to get fio to dump the current output while it is
1734 running, without terminating the job. To do that, send fio the \fBUSR1\fR
1737 If the \fB\-\-minimal\fR / \fB\-\-append-terse\fR options are given, the
1738 results will be printed/appended in a semicolon-delimited format suitable for
1740 A job description (if provided) follows on a new line. Note that the first
1741 number in the line is the version number. If the output has to be changed
1742 for some reason, this number will be incremented by 1 to signify that
1743 change. The fields are:
1746 .B terse version, fio version, jobname, groupid, error
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 Total I/O \fR(KB)\fP, bandwidth \fR(KB/s)\fP, IOPS, runtime \fR(ms)\fP
1780 .B min, max, mean, standard deviation
1784 .B min, max, mean, standard deviation
1786 Completion latency percentiles (20 fields):
1788 .B Xth percentile=usec
1792 .B min, max, mean, standard deviation
1796 .B min, max, aggregate percentage of total, mean, standard deviation
1802 .B user, system, context switches, major page faults, minor page faults
1805 IO depth distribution:
1807 .B <=1, 2, 4, 8, 16, 32, >=64
1810 IO latency distribution:
1814 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
1818 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
1822 Disk utilization (1 for each disk used):
1824 .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
1827 Error Info (dependent on continue_on_error, default off):
1829 .B total # errors, first error code
1832 .B text description (if provided in config - appears on newline)
1835 Normally you would run fio as a stand-alone application on the machine
1836 where the IO workload should be generated. However, it is also possible to
1837 run the frontend and backend of fio separately. This makes it possible to
1838 have a fio server running on the machine(s) where the IO workload should
1839 be running, while controlling it from another machine.
1841 To start the server, you would do:
1843 \fBfio \-\-server=args\fR
1845 on that machine, where args defines what fio listens to. The arguments
1846 are of the form 'type:hostname or IP:port'. 'type' is either 'ip' (or ip4)
1847 for TCP/IP v4, 'ip6' for TCP/IP v6, or 'sock' for a local unix domain
1848 socket. 'hostname' is either a hostname or IP address, and 'port' is the port to
1849 listen to (only valid for TCP/IP, not a local socket). Some examples:
1853 Start a fio server, listening on all interfaces on the default port (8765).
1855 2) fio \-\-server=ip:hostname,4444
1857 Start a fio server, listening on IP belonging to hostname and on port 4444.
1859 3) fio \-\-server=ip6:::1,4444
1861 Start a fio server, listening on IPv6 localhost ::1 and on port 4444.
1863 4) fio \-\-server=,4444
1865 Start a fio server, listening on all interfaces on port 4444.
1867 5) fio \-\-server=1.2.3.4
1869 Start a fio server, listening on IP 1.2.3.4 on the default port.
1871 6) fio \-\-server=sock:/tmp/fio.sock
1873 Start a fio server, listening on the local socket /tmp/fio.sock.
1875 When a server is running, you can connect to it from a client. The client
1878 fio \-\-local-args \-\-client=server \-\-remote-args <job file(s)>
1880 where \-\-local-args are arguments that are local to the client where it is
1881 running, 'server' is the connect string, and \-\-remote-args and <job file(s)>
1882 are sent to the server. The 'server' string follows the same format as it
1883 does on the server side, to allow IP/hostname/socket and port strings.
1884 You can connect to multiple clients as well, to do that you could run:
1886 fio \-\-client=server2 \-\-client=server2 <job file(s)>
1888 If the job file is located on the fio server, then you can tell the server
1889 to load a local file as well. This is done by using \-\-remote-config:
1891 fio \-\-client=server \-\-remote-config /path/to/file.fio
1893 Then the fio serer will open this local (to the server) job file instead
1894 of being passed one from the client.
1898 was written by Jens Axboe <jens.axboe@oracle.com>,
1899 now Jens Axboe <axboe@fb.com>.
1901 This man page was written by Aaron Carroll <aaronc@cse.unsw.edu.au> based
1902 on documentation by Jens Axboe.
1903 .SH "REPORTING BUGS"
1904 Report bugs to the \fBfio\fR mailing list <fio@vger.kernel.org>.
1907 For further documentation see \fBHOWTO\fR and \fBREADME\fR.
1909 Sample jobfiles are available in the \fBexamples\fR directory.
projects / fio.git / blob