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 Trim and write mixed workload. Blocks will be trimmed first, then the same
292 blocks will be written to.
295 For mixed I/O, the default split is 50/50. For certain types of io the result
296 may still be skewed a bit, since the speed may be different. It is possible to
297 specify a number of IO's to do before getting a new offset, this is done by
298 appending a `:\fI<nr>\fR to the end of the string given. For a random read, it
299 would look like \fBrw=randread:8\fR for passing in an offset modifier with a
300 value of 8. If the postfix is used with a sequential IO pattern, then the value
301 specified will be added to the generated offset for each IO. For instance,
302 using \fBrw=write:4k\fR will skip 4k for every write. It turns sequential IO
303 into sequential IO with holes. See the \fBrw_sequencer\fR option.
306 .BI rw_sequencer \fR=\fPstr
307 If an offset modifier is given by appending a number to the \fBrw=<str>\fR line,
308 then this option controls how that number modifies the IO offset being
309 generated. Accepted values are:
314 Generate sequential offset
317 Generate the same offset
320 \fBsequential\fR is only useful for random IO, where fio would normally
321 generate a new random offset for every IO. If you append eg 8 to randread, you
322 would get a new random offset for every 8 IO's. The result would be a seek for
323 only every 8 IO's, instead of for every IO. Use \fBrw=randread:8\fR to specify
324 that. As sequential IO is already sequential, setting \fBsequential\fR for that
325 would not result in any differences. \fBidentical\fR behaves in a similar
326 fashion, except it sends the same offset 8 number of times before generating a
331 .BI kb_base \fR=\fPint
332 The base unit for a kilobyte. The defacto base is 2^10, 1024. Storage
333 manufacturers like to use 10^3 or 1000 as a base ten unit instead, for obvious
334 reasons. Allowed values are 1024 or 1000, with 1024 being the default.
336 .BI unified_rw_reporting \fR=\fPbool
337 Fio normally reports statistics on a per data direction basis, meaning that
338 read, write, and trim are accounted and reported separately. If this option is
339 set fio sums the results and reports them as "mixed" instead.
341 .BI randrepeat \fR=\fPbool
342 Seed the random number generator used for random I/O patterns in a predictable
343 way so the pattern is repeatable across runs. Default: true.
345 .BI allrandrepeat \fR=\fPbool
346 Seed all random number generators in a predictable way so results are
347 repeatable across runs. Default: false.
349 .BI randseed \fR=\fPint
350 Seed the random number generators based on this seed value, to be able to
351 control what sequence of output is being generated. If not set, the random
352 sequence depends on the \fBrandrepeat\fR setting.
354 .BI fallocate \fR=\fPstr
355 Whether pre-allocation is performed when laying down files. Accepted values
361 Do not pre-allocate space.
364 Pre-allocate via \fBposix_fallocate\fR\|(3).
367 Pre-allocate via \fBfallocate\fR\|(2) with FALLOC_FL_KEEP_SIZE set.
370 Backward-compatible alias for 'none'.
373 Backward-compatible alias for 'posix'.
376 May not be available on all supported platforms. 'keep' is only
377 available on Linux. If using ZFS on Solaris this must be set to 'none'
378 because ZFS doesn't support it. Default: 'posix'.
381 .BI fadvise_hint \fR=\fPbool
382 Use \fBposix_fadvise\fR\|(2) to advise the kernel what I/O patterns
383 are likely to be issued. Default: true.
385 .BI fadvise_stream \fR=\fPint
386 Use \fBposix_fadvise\fR\|(2) to advise the kernel what stream ID the
387 writes issued belong to. Only supported on Linux. Note, this option
388 may change going forward.
391 Total size of I/O for this job. \fBfio\fR will run until this many bytes have
392 been transferred, unless limited by other options (\fBruntime\fR, for instance,
393 or increased/descreased by \fBio_size\fR). Unless \fBnrfiles\fR and
394 \fBfilesize\fR options are given, this amount will be divided between the
395 available files for the job. If not set, fio will use the full size of the
396 given files or devices. If the files do not exist, size must be given. It is
397 also possible to give size as a percentage between 1 and 100. If size=20% is
398 given, fio will use 20% of the full size of the given files or devices.
400 .BI io_size \fR=\fPint "\fR,\fB io_limit \fR=\fPint
401 Normally fio operates within the region set by \fBsize\fR, which means that
402 the \fBsize\fR option sets both the region and size of IO to be performed.
403 Sometimes that is not what you want. With this option, it is possible to
404 define just the amount of IO that fio should do. For instance, if \fBsize\fR
405 is set to 20G and \fBio_limit\fR is set to 5G, fio will perform IO within
406 the first 20G but exit when 5G have been done. The opposite is also
407 possible - if \fBsize\fR is set to 20G, and \fBio_size\fR is set to 40G, then
408 fio will do 40G of IO within the 0..20G region.
410 .BI fill_device \fR=\fPbool "\fR,\fB fill_fs" \fR=\fPbool
411 Sets size to something really large and waits for ENOSPC (no space left on
412 device) as the terminating condition. Only makes sense with sequential write.
413 For a read workload, the mount point will be filled first then IO started on
414 the result. This option doesn't make sense if operating on a raw device node,
415 since the size of that is already known by the file system. Additionally,
416 writing beyond end-of-device will not return ENOSPC there.
418 .BI filesize \fR=\fPirange
419 Individual file sizes. May be a range, in which case \fBfio\fR will select sizes
420 for files at random within the given range, limited to \fBsize\fR in total (if
421 that is given). If \fBfilesize\fR is not specified, each created file is the
424 .BI file_append \fR=\fPbool
425 Perform IO after the end of the file. Normally fio will operate within the
426 size of a file. If this option is set, then fio will append to the file
427 instead. This has identical behavior to setting \fRoffset\fP to the size
428 of a file. This option is ignored on non-regular files.
430 .BI blocksize \fR=\fPint[,int] "\fR,\fB bs" \fR=\fPint[,int]
431 Block size for I/O units. Default: 4k. Values for reads, writes, and trims
432 can be specified separately in the format \fIread\fR,\fIwrite\fR,\fItrim\fR
433 either of which may be empty to leave that value at its default. If a trailing
434 comma isn't given, the remainder will inherit the last value set.
436 .BI blocksize_range \fR=\fPirange[,irange] "\fR,\fB bsrange" \fR=\fPirange[,irange]
437 Specify a range of I/O block sizes. The issued I/O unit will always be a
438 multiple of the minimum size, unless \fBblocksize_unaligned\fR is set. Applies
439 to both reads and writes if only one range is given, but can be specified
440 separately with a comma separating the values. Example: bsrange=1k-4k,2k-8k.
441 Also (see \fBblocksize\fR).
443 .BI bssplit \fR=\fPstr
444 This option allows even finer grained control of the block sizes issued,
445 not just even splits between them. With this option, you can weight various
446 block sizes for exact control of the issued IO for a job that has mixed
447 block sizes. The format of the option is bssplit=blocksize/percentage,
448 optionally adding as many definitions as needed separated by a colon.
449 Example: bssplit=4k/10:64k/50:32k/40 would issue 50% 64k blocks, 10% 4k
450 blocks and 40% 32k blocks. \fBbssplit\fR also supports giving separate
451 splits to reads and writes. The format is identical to what the
452 \fBbs\fR option accepts, the read and write parts are separated with a
455 .B blocksize_unaligned\fR,\fP bs_unaligned
456 If set, any size in \fBblocksize_range\fR may be used. This typically won't
457 work with direct I/O, as that normally requires sector alignment.
459 .BI blockalign \fR=\fPint[,int] "\fR,\fB ba" \fR=\fPint[,int]
460 At what boundary to align random IO offsets. Defaults to the same as 'blocksize'
461 the minimum blocksize given. Minimum alignment is typically 512b
462 for using direct IO, though it usually depends on the hardware block size.
463 This option is mutually exclusive with using a random map for files, so it
464 will turn off that option.
466 .BI bs_is_seq_rand \fR=\fPbool
467 If this option is set, fio will use the normal read,write blocksize settings as
468 sequential,random instead. Any random read or write will use the WRITE
469 blocksize settings, and any sequential read or write will use the READ
473 Initialize buffers with all zeros. Default: fill buffers with random data.
476 If this option is given, fio will refill the IO buffers on every submit. The
477 default is to only fill it at init time and reuse that data. Only makes sense
478 if zero_buffers isn't specified, naturally. If data verification is enabled,
479 refill_buffers is also automatically enabled.
481 .BI scramble_buffers \fR=\fPbool
482 If \fBrefill_buffers\fR is too costly and the target is using data
483 deduplication, then setting this option will slightly modify the IO buffer
484 contents to defeat normal de-dupe attempts. This is not enough to defeat
485 more clever block compression attempts, but it will stop naive dedupe
486 of blocks. Default: true.
488 .BI buffer_compress_percentage \fR=\fPint
489 If this is set, then fio will attempt to provide IO buffer content (on WRITEs)
490 that compress to the specified level. Fio does this by providing a mix of
491 random data and a fixed pattern. The fixed pattern is either zeroes, or the
492 pattern specified by \fBbuffer_pattern\fR. If the pattern option is used, it
493 might skew the compression ratio slightly. Note that this is per block size
494 unit, for file/disk wide compression level that matches this setting. Note
495 that this is per block size unit, for file/disk wide compression level that
496 matches this setting, you'll also want to set refill_buffers.
498 .BI buffer_compress_chunk \fR=\fPint
499 See \fBbuffer_compress_percentage\fR. This setting allows fio to manage how
500 big the ranges of random data and zeroed data is. Without this set, fio will
501 provide \fBbuffer_compress_percentage\fR of blocksize random data, followed by
502 the remaining zeroed. With this set to some chunk size smaller than the block
503 size, fio can alternate random and zeroed data throughout the IO buffer.
505 .BI buffer_pattern \fR=\fPstr
506 If set, fio will fill the IO buffers with this pattern. If not set, the contents
507 of IO buffers is defined by the other options related to buffer contents. The
508 setting can be any pattern of bytes, and can be prefixed with 0x for hex
509 values. It may also be a string, where the string must then be wrapped with
512 .BI dedupe_percentage \fR=\fPint
513 If set, fio will generate this percentage of identical buffers when writing.
514 These buffers will be naturally dedupable. The contents of the buffers depend
515 on what other buffer compression settings have been set. It's possible to have
516 the individual buffers either fully compressible, or not at all. This option
517 only controls the distribution of unique buffers.
519 .BI nrfiles \fR=\fPint
520 Number of files to use for this job. Default: 1.
522 .BI openfiles \fR=\fPint
523 Number of files to keep open at the same time. Default: \fBnrfiles\fR.
525 .BI file_service_type \fR=\fPstr
526 Defines how files to service are selected. The following types are defined:
531 Choose a file at random.
534 Round robin over opened files (default).
537 Do each file in the set sequentially.
540 The number of I/Os to issue before switching to a new file can be specified by
541 appending `:\fIint\fR' to the service type.
544 .BI ioengine \fR=\fPstr
545 Defines how the job issues I/O. The following types are defined:
550 Basic \fBread\fR\|(2) or \fBwrite\fR\|(2) I/O. \fBfseek\fR\|(2) is used to
551 position the I/O location.
554 Basic \fBpread\fR\|(2) or \fBpwrite\fR\|(2) I/O.
557 Basic \fBreadv\fR\|(2) or \fBwritev\fR\|(2) I/O. Will emulate queuing by
558 coalescing adjacent IOs into a single submission.
561 Basic \fBpreadv\fR\|(2) or \fBpwritev\fR\|(2) I/O.
564 Linux native asynchronous I/O. This ioengine defines engine specific options.
567 POSIX asynchronous I/O using \fBaio_read\fR\|(3) and \fBaio_write\fR\|(3).
570 Solaris native asynchronous I/O.
573 Windows native asynchronous I/O.
576 File is memory mapped with \fBmmap\fR\|(2) and data copied using
580 \fBsplice\fR\|(2) is used to transfer the data and \fBvmsplice\fR\|(2) to
581 transfer data from user-space to the kernel.
584 Use the syslet system calls to make regular read/write asynchronous.
587 SCSI generic sg v3 I/O. May be either synchronous using the SG_IO ioctl, or if
588 the target is an sg character device, we use \fBread\fR\|(2) and
589 \fBwrite\fR\|(2) for asynchronous I/O.
592 Doesn't transfer any data, just pretends to. Mainly used to exercise \fBfio\fR
593 itself and for debugging and testing purposes.
596 Transfer over the network. The protocol to be used can be defined with the
597 \fBprotocol\fR parameter. Depending on the protocol, \fBfilename\fR,
598 \fBhostname\fR, \fBport\fR, or \fBlisten\fR must be specified.
599 This ioengine defines engine specific options.
602 Like \fBnet\fR, but uses \fBsplice\fR\|(2) and \fBvmsplice\fR\|(2) to map data
603 and send/receive. This ioengine defines engine specific options.
606 Doesn't transfer any data, but burns CPU cycles according to \fBcpuload\fR and
607 \fBcpucycles\fR parameters.
610 The GUASI I/O engine is the Generic Userspace Asynchronous Syscall Interface
611 approach to asynchronous I/O.
613 See <http://www.xmailserver.org/guasi\-lib.html>.
616 The RDMA I/O engine supports both RDMA memory semantics (RDMA_WRITE/RDMA_READ)
617 and channel semantics (Send/Recv) for the InfiniBand, RoCE and iWARP protocols.
620 Loads an external I/O engine object file. Append the engine filename as
624 IO engine that does regular linux native fallocate call to simulate data
625 transfer as fio ioengine
627 DDIR_READ does fallocate(,mode = FALLOC_FL_KEEP_SIZE,)
629 DIR_WRITE does fallocate(,mode = 0)
631 DDIR_TRIM does fallocate(,mode = FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE)
634 IO engine that does regular EXT4_IOC_MOVE_EXT ioctls to simulate defragment activity
635 request to DDIR_WRITE event
638 IO engine supporting direct access to Ceph Rados Block Devices (RBD) via librbd
639 without the need to use the kernel rbd driver. This ioengine defines engine specific
643 Using Glusterfs libgfapi sync interface to direct access to Glusterfs volumes without
644 having to go through FUSE. This ioengine defines engine specific
648 Using Glusterfs libgfapi async interface to direct access to Glusterfs volumes without
649 having to go through FUSE. This ioengine defines engine specific
653 Read and write through Hadoop (HDFS). The \fBfilename\fR option is used to
654 specify host,port of the hdfs name-node to connect. This engine interprets
655 offsets a little differently. In HDFS, files once created cannot be modified.
656 So random writes are not possible. To imitate this, libhdfs engine expects
657 bunch of small files to be created over HDFS, and engine will randomly pick a
658 file out of those files based on the offset generated by fio backend. (see the
659 example job file to create such files, use rw=write option). Please note, you
660 might want to set necessary environment variables to work with hdfs/libhdfs
664 Read, write and erase an MTD character device (e.g., /dev/mtd0). Discards are
665 treated as erases. Depending on the underlying device type, the I/O may have
666 to go in a certain pattern, e.g., on NAND, writing sequentially to erase blocks
667 and discarding before overwriting. The writetrim mode works well for this
673 .BI iodepth \fR=\fPint
674 Number of I/O units to keep in flight against the file. Note that increasing
675 iodepth beyond 1 will not affect synchronous ioengines (except for small
676 degress when verify_async is in use). Even async engines may impose OS
677 restrictions causing the desired depth not to be achieved. This may happen on
678 Linux when using libaio and not setting \fBdirect\fR=1, since buffered IO is
679 not async on that OS. Keep an eye on the IO depth distribution in the
680 fio output to verify that the achieved depth is as expected. Default: 1.
682 .BI iodepth_batch \fR=\fPint
683 Number of I/Os to submit at once. Default: \fBiodepth\fR.
685 .BI iodepth_batch_complete \fR=\fPint
686 This defines how many pieces of IO to retrieve at once. It defaults to 1 which
687 means that we'll ask for a minimum of 1 IO in the retrieval process from the
688 kernel. The IO retrieval will go on until we hit the limit set by
689 \fBiodepth_low\fR. If this variable is set to 0, then fio will always check for
690 completed events before queuing more IO. This helps reduce IO latency, at the
691 cost of more retrieval system calls.
693 .BI iodepth_low \fR=\fPint
694 Low watermark indicating when to start filling the queue again. Default:
697 .BI io_submit_mode \fR=\fPstr
698 This option controls how fio submits the IO to the IO engine. The default is
699 \fBinline\fR, which means that the fio job threads submit and reap IO directly.
700 If set to \fBoffload\fR, the job threads will offload IO submission to a
701 dedicated pool of IO threads. This requires some coordination and thus has a
702 bit of extra overhead, especially for lower queue depth IO where it can
703 increase latencies. The benefit is that fio can manage submission rates
704 independently of the device completion rates. This avoids skewed latency
705 reporting if IO gets back up on the device side (the coordinated omission
708 .BI direct \fR=\fPbool
709 If true, use non-buffered I/O (usually O_DIRECT). Default: false.
711 .BI atomic \fR=\fPbool
712 If value is true, attempt to use atomic direct IO. Atomic writes are guaranteed
713 to be stable once acknowledged by the operating system. Only Linux supports
716 .BI buffered \fR=\fPbool
717 If true, use buffered I/O. This is the opposite of the \fBdirect\fR parameter.
720 .BI offset \fR=\fPint
721 Offset in the file to start I/O. Data before the offset will not be touched.
723 .BI offset_increment \fR=\fPint
724 If this is provided, then the real offset becomes the
725 offset + offset_increment * thread_number, where the thread number is a
726 counter that starts at 0 and is incremented for each sub-job (i.e. when
727 numjobs option is specified). This option is useful if there are several jobs
728 which are intended to operate on a file in parallel disjoint segments, with
729 even spacing between the starting points.
731 .BI number_ios \fR=\fPint
732 Fio will normally perform IOs until it has exhausted the size of the region
733 set by \fBsize\fR, or if it exhaust the allocated time (or hits an error
734 condition). With this setting, the range/size can be set independently of
735 the number of IOs to perform. When fio reaches this number, it will exit
736 normally and report status. Note that this does not extend the amount
737 of IO that will be done, it will only stop fio if this condition is met
738 before other end-of-job criteria.
741 How many I/Os to perform before issuing an \fBfsync\fR\|(2) of dirty data. If
742 0, don't sync. Default: 0.
744 .BI fdatasync \fR=\fPint
745 Like \fBfsync\fR, but uses \fBfdatasync\fR\|(2) instead to only sync the
746 data parts of the file. Default: 0.
748 .BI write_barrier \fR=\fPint
749 Make every Nth write a barrier write.
751 .BI sync_file_range \fR=\fPstr:int
752 Use \fBsync_file_range\fR\|(2) for every \fRval\fP number of write operations. Fio will
753 track range of writes that have happened since the last \fBsync_file_range\fR\|(2) call.
754 \fRstr\fP can currently be one or more of:
758 SYNC_FILE_RANGE_WAIT_BEFORE
761 SYNC_FILE_RANGE_WRITE
764 SYNC_FILE_RANGE_WRITE
768 So if you do sync_file_range=wait_before,write:8, fio would use
769 \fBSYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE\fP for every 8 writes.
770 Also see the \fBsync_file_range\fR\|(2) man page. This option is Linux specific.
772 .BI overwrite \fR=\fPbool
773 If writing, setup the file first and do overwrites. Default: false.
775 .BI end_fsync \fR=\fPbool
776 Sync file contents when a write stage has completed. Default: false.
778 .BI fsync_on_close \fR=\fPbool
779 If true, sync file contents on close. This differs from \fBend_fsync\fR in that
780 it will happen on every close, not just at the end of the job. Default: false.
782 .BI rwmixread \fR=\fPint
783 Percentage of a mixed workload that should be reads. Default: 50.
785 .BI rwmixwrite \fR=\fPint
786 Percentage of a mixed workload that should be writes. If \fBrwmixread\fR and
787 \fBrwmixwrite\fR are given and do not sum to 100%, the latter of the two
788 overrides the first. This may interfere with a given rate setting, if fio is
789 asked to limit reads or writes to a certain rate. If that is the case, then
790 the distribution may be skewed. Default: 50.
792 .BI random_distribution \fR=\fPstr:float
793 By default, fio will use a completely uniform random distribution when asked
794 to perform random IO. Sometimes it is useful to skew the distribution in
795 specific ways, ensuring that some parts of the data is more hot than others.
796 Fio includes the following distribution models:
800 Uniform random distribution
810 When using a zipf or pareto distribution, an input value is also needed to
811 define the access pattern. For zipf, this is the zipf theta. For pareto,
812 it's the pareto power. Fio includes a test program, genzipf, that can be
813 used visualize what the given input values will yield in terms of hit rates.
814 If you wanted to use zipf with a theta of 1.2, you would use
815 random_distribution=zipf:1.2 as the option. If a non-uniform model is used,
816 fio will disable use of the random map.
818 .BI percentage_random \fR=\fPint
819 For a random workload, set how big a percentage should be random. This defaults
820 to 100%, in which case the workload is fully random. It can be set from
821 anywhere from 0 to 100. Setting it to 0 would make the workload fully
822 sequential. It is possible to set different values for reads, writes, and
823 trim. To do so, simply use a comma separated list. See \fBblocksize\fR.
826 Normally \fBfio\fR will cover every block of the file when doing random I/O. If
827 this parameter is given, a new offset will be chosen without looking at past
828 I/O history. This parameter is mutually exclusive with \fBverify\fR.
830 .BI softrandommap \fR=\fPbool
831 See \fBnorandommap\fR. If fio runs with the random block map enabled and it
832 fails to allocate the map, if this option is set it will continue without a
833 random block map. As coverage will not be as complete as with random maps, this
834 option is disabled by default.
836 .BI random_generator \fR=\fPstr
837 Fio supports the following engines for generating IO offsets for random IO:
841 Strong 2^88 cycle random number generator
844 Linear feedback shift register generator
848 Tausworthe is a strong random number generator, but it requires tracking on the
849 side if we want to ensure that blocks are only read or written once. LFSR
850 guarantees that we never generate the same offset twice, and it's also less
851 computationally expensive. It's not a true random generator, however, though
852 for IO purposes it's typically good enough. LFSR only works with single block
853 sizes, not with workloads that use multiple block sizes. If used with such a
854 workload, fio may read or write some blocks multiple times.
857 Run job with given nice value. See \fBnice\fR\|(2).
860 Set I/O priority value of this job between 0 (highest) and 7 (lowest). See
863 .BI prioclass \fR=\fPint
864 Set I/O priority class. See \fBionice\fR\|(1).
866 .BI thinktime \fR=\fPint
867 Stall job for given number of microseconds between issuing I/Os.
869 .BI thinktime_spin \fR=\fPint
870 Pretend to spend CPU time for given number of microseconds, sleeping the rest
871 of the time specified by \fBthinktime\fR. Only valid if \fBthinktime\fR is set.
873 .BI thinktime_blocks \fR=\fPint
874 Only valid if thinktime is set - control how many blocks to issue, before
875 waiting \fBthinktime\fR microseconds. If not set, defaults to 1 which will
876 make fio wait \fBthinktime\fR microseconds after every block. This
877 effectively makes any queue depth setting redundant, since no more than 1 IO
878 will be queued before we have to complete it and do our thinktime. In other
879 words, this setting effectively caps the queue depth if the latter is larger.
883 Cap bandwidth used by this job. The number is in bytes/sec, the normal postfix
884 rules apply. You can use \fBrate\fR=500k to limit reads and writes to 500k each,
885 or you can specify read and writes separately. Using \fBrate\fR=1m,500k would
886 limit reads to 1MB/sec and writes to 500KB/sec. Capping only reads or writes
887 can be done with \fBrate\fR=,500k or \fBrate\fR=500k,. The former will only
888 limit writes (to 500KB/sec), the latter will only limit reads.
890 .BI ratemin \fR=\fPint
891 Tell \fBfio\fR to do whatever it can to maintain at least the given bandwidth.
892 Failing to meet this requirement will cause the job to exit. The same format
893 as \fBrate\fR is used for read vs write separation.
895 .BI rate_iops \fR=\fPint
896 Cap the bandwidth to this number of IOPS. Basically the same as rate, just
897 specified independently of bandwidth. The same format as \fBrate\fR is used for
898 read vs write separation. If \fBblocksize\fR is a range, the smallest block
899 size is used as the metric.
901 .BI rate_iops_min \fR=\fPint
902 If this rate of I/O is not met, the job will exit. The same format as \fBrate\fR
903 is used for read vs write separation.
905 .BI ratecycle \fR=\fPint
906 Average bandwidth for \fBrate\fR and \fBratemin\fR over this number of
907 milliseconds. Default: 1000ms.
909 .BI latency_target \fR=\fPint
910 If set, fio will attempt to find the max performance point that the given
911 workload will run at while maintaining a latency below this target. The
912 values is given in microseconds. See \fBlatency_window\fR and
913 \fBlatency_percentile\fR.
915 .BI latency_window \fR=\fPint
916 Used with \fBlatency_target\fR to specify the sample window that the job
917 is run at varying queue depths to test the performance. The value is given
920 .BI latency_percentile \fR=\fPfloat
921 The percentage of IOs that must fall within the criteria specified by
922 \fBlatency_target\fR and \fBlatency_window\fR. If not set, this defaults
923 to 100.0, meaning that all IOs must be equal or below to the value set
924 by \fBlatency_target\fR.
926 .BI max_latency \fR=\fPint
927 If set, fio will exit the job if it exceeds this maximum latency. It will exit
930 .BI cpumask \fR=\fPint
931 Set CPU affinity for this job. \fIint\fR is a bitmask of allowed CPUs the job
932 may run on. See \fBsched_setaffinity\fR\|(2).
934 .BI cpus_allowed \fR=\fPstr
935 Same as \fBcpumask\fR, but allows a comma-delimited list of CPU numbers.
937 .BI cpus_allowed_policy \fR=\fPstr
938 Set the policy of how fio distributes the CPUs specified by \fBcpus_allowed\fR
939 or \fBcpumask\fR. Two policies are supported:
944 All jobs will share the CPU set specified.
947 Each job will get a unique CPU from the CPU set.
950 \fBshared\fR is the default behaviour, if the option isn't specified. If
951 \fBsplit\fR is specified, then fio will assign one cpu per job. If not enough
952 CPUs are given for the jobs listed, then fio will roundrobin the CPUs in
957 .BI numa_cpu_nodes \fR=\fPstr
958 Set this job running on specified NUMA nodes' CPUs. The arguments allow
959 comma delimited list of cpu numbers, A-B ranges, or 'all'.
961 .BI numa_mem_policy \fR=\fPstr
962 Set this job's memory policy and corresponding NUMA nodes. Format of
966 .B <mode>[:<nodelist>]
969 is one of the following memory policy:
971 .B default, prefer, bind, interleave, local
974 For \fBdefault\fR and \fBlocal\fR memory policy, no \fBnodelist\fR is
975 needed to be specified. For \fBprefer\fR, only one node is
976 allowed. For \fBbind\fR and \fBinterleave\fR, \fBnodelist\fR allows
977 comma delimited list of numbers, A-B ranges, or 'all'.
979 .BI startdelay \fR=\fPirange
980 Delay start of job for the specified number of seconds. Supports all time
981 suffixes to allow specification of hours, minutes, seconds and
982 milliseconds - seconds are the default if a unit is omitted.
983 Can be given as a range which causes each thread to choose randomly out of the
986 .BI runtime \fR=\fPint
987 Terminate processing after the specified number of seconds.
990 If given, run for the specified \fBruntime\fR duration even if the files are
991 completely read or written. The same workload will be repeated as many times
992 as \fBruntime\fR allows.
994 .BI ramp_time \fR=\fPint
995 If set, fio will run the specified workload for this amount of time before
996 logging any performance numbers. Useful for letting performance settle before
997 logging results, thus minimizing the runtime required for stable results. Note
998 that the \fBramp_time\fR is considered lead in time for a job, thus it will
999 increase the total runtime if a special timeout or runtime is specified.
1001 .BI invalidate \fR=\fPbool
1002 Invalidate buffer-cache for the file prior to starting I/O. Default: true.
1004 .BI sync \fR=\fPbool
1005 Use synchronous I/O for buffered writes. For the majority of I/O engines,
1006 this means using O_SYNC. Default: false.
1008 .BI iomem \fR=\fPstr "\fR,\fP mem" \fR=\fPstr
1009 Allocation method for I/O unit buffer. Allowed values are:
1014 Allocate memory with \fBmalloc\fR\|(3).
1017 Use shared memory buffers allocated through \fBshmget\fR\|(2).
1020 Same as \fBshm\fR, but use huge pages as backing.
1023 Use \fBmmap\fR\|(2) for allocation. Uses anonymous memory unless a filename
1024 is given after the option in the format `:\fIfile\fR'.
1027 Same as \fBmmap\fR, but use huge files as backing.
1030 The amount of memory allocated is the maximum allowed \fBblocksize\fR for the
1031 job multiplied by \fBiodepth\fR. For \fBshmhuge\fR or \fBmmaphuge\fR to work,
1032 the system must have free huge pages allocated. \fBmmaphuge\fR also needs to
1033 have hugetlbfs mounted, and \fIfile\fR must point there. At least on Linux,
1034 huge pages must be manually allocated. See \fB/proc/sys/vm/nr_hugehages\fR
1035 and the documentation for that. Normally you just need to echo an appropriate
1036 number, eg echoing 8 will ensure that the OS has 8 huge pages ready for
1040 .BI iomem_align \fR=\fPint "\fR,\fP mem_align" \fR=\fPint
1041 This indicates the memory alignment of the IO memory buffers. Note that the
1042 given alignment is applied to the first IO unit buffer, if using \fBiodepth\fR
1043 the alignment of the following buffers are given by the \fBbs\fR used. In
1044 other words, if using a \fBbs\fR that is a multiple of the page sized in the
1045 system, all buffers will be aligned to this value. If using a \fBbs\fR that
1046 is not page aligned, the alignment of subsequent IO memory buffers is the
1047 sum of the \fBiomem_align\fR and \fBbs\fR used.
1049 .BI hugepage\-size \fR=\fPint
1050 Defines the size of a huge page. Must be at least equal to the system setting.
1051 Should be a multiple of 1MB. Default: 4MB.
1054 Terminate all jobs when one finishes. Default: wait for each job to finish.
1056 .BI bwavgtime \fR=\fPint
1057 Average bandwidth calculations over the given time in milliseconds. Default:
1060 .BI iopsavgtime \fR=\fPint
1061 Average IOPS calculations over the given time in milliseconds. Default:
1064 .BI create_serialize \fR=\fPbool
1065 If true, serialize file creation for the jobs. Default: true.
1067 .BI create_fsync \fR=\fPbool
1068 \fBfsync\fR\|(2) data file after creation. Default: true.
1070 .BI create_on_open \fR=\fPbool
1071 If true, the files are not created until they are opened for IO by the job.
1073 .BI create_only \fR=\fPbool
1074 If true, fio will only run the setup phase of the job. If files need to be
1075 laid out or updated on disk, only that will be done. The actual job contents
1078 .BI pre_read \fR=\fPbool
1079 If this is given, files will be pre-read into memory before starting the given
1080 IO operation. This will also clear the \fR \fBinvalidate\fR flag, since it is
1081 pointless to pre-read and then drop the cache. This will only work for IO
1082 engines that are seekable, since they allow you to read the same data
1083 multiple times. Thus it will not work on eg network or splice IO.
1085 .BI unlink \fR=\fPbool
1086 Unlink job files when done. Default: false.
1088 .BI loops \fR=\fPint
1089 Specifies the number of iterations (runs of the same workload) of this job.
1092 .BI verify_only \fR=\fPbool
1093 Do not perform the specified workload, only verify data still matches previous
1094 invocation of this workload. This option allows one to check data multiple
1095 times at a later date without overwriting it. This option makes sense only for
1096 workloads that write data, and does not support workloads with the
1097 \fBtime_based\fR option set.
1099 .BI do_verify \fR=\fPbool
1100 Run the verify phase after a write phase. Only valid if \fBverify\fR is set.
1103 .BI verify \fR=\fPstr
1104 Method of verifying file contents after each iteration of the job. Allowed
1109 .B md5 crc16 crc32 crc32c crc32c-intel crc64 crc7 sha256 sha512 sha1 xxhash
1110 Store appropriate checksum in the header of each block. crc32c-intel is
1111 hardware accelerated SSE4.2 driven, falls back to regular crc32c if
1112 not supported by the system.
1115 Write extra information about each I/O (timestamp, block number, etc.). The
1116 block number is verified. See \fBverify_pattern\fR as well.
1119 Pretend to verify. Used for testing internals.
1122 This option can be used for repeated burn-in tests of a system to make sure
1123 that the written data is also correctly read back. If the data direction given
1124 is a read or random read, fio will assume that it should verify a previously
1125 written file. If the data direction includes any form of write, the verify will
1126 be of the newly written data.
1129 .BI verifysort \fR=\fPbool
1130 If true, written verify blocks are sorted if \fBfio\fR deems it to be faster to
1131 read them back in a sorted manner. Default: true.
1133 .BI verifysort_nr \fR=\fPint
1134 Pre-load and sort verify blocks for a read workload.
1136 .BI verify_offset \fR=\fPint
1137 Swap the verification header with data somewhere else in the block before
1138 writing. It is swapped back before verifying.
1140 .BI verify_interval \fR=\fPint
1141 Write the verification header for this number of bytes, which should divide
1142 \fBblocksize\fR. Default: \fBblocksize\fR.
1144 .BI verify_pattern \fR=\fPstr
1145 If set, fio will fill the io buffers with this pattern. Fio defaults to filling
1146 with totally random bytes, but sometimes it's interesting to fill with a known
1147 pattern for io verification purposes. Depending on the width of the pattern,
1148 fio will fill 1/2/3/4 bytes of the buffer at the time(it can be either a
1149 decimal or a hex number). The verify_pattern if larger than a 32-bit quantity
1150 has to be a hex number that starts with either "0x" or "0X". Use with
1153 .BI verify_fatal \fR=\fPbool
1154 If true, exit the job on the first observed verification failure. Default:
1157 .BI verify_dump \fR=\fPbool
1158 If set, dump the contents of both the original data block and the data block we
1159 read off disk to files. This allows later analysis to inspect just what kind of
1160 data corruption occurred. Off by default.
1162 .BI verify_async \fR=\fPint
1163 Fio will normally verify IO inline from the submitting thread. This option
1164 takes an integer describing how many async offload threads to create for IO
1165 verification instead, causing fio to offload the duty of verifying IO contents
1166 to one or more separate threads. If using this offload option, even sync IO
1167 engines can benefit from using an \fBiodepth\fR setting higher than 1, as it
1168 allows them to have IO in flight while verifies are running.
1170 .BI verify_async_cpus \fR=\fPstr
1171 Tell fio to set the given CPU affinity on the async IO verification threads.
1172 See \fBcpus_allowed\fP for the format used.
1174 .BI verify_backlog \fR=\fPint
1175 Fio will normally verify the written contents of a job that utilizes verify
1176 once that job has completed. In other words, everything is written then
1177 everything is read back and verified. You may want to verify continually
1178 instead for a variety of reasons. Fio stores the meta data associated with an
1179 IO block in memory, so for large verify workloads, quite a bit of memory would
1180 be used up holding this meta data. If this option is enabled, fio will write
1181 only N blocks before verifying these blocks.
1183 .BI verify_backlog_batch \fR=\fPint
1184 Control how many blocks fio will verify if verify_backlog is set. If not set,
1185 will default to the value of \fBverify_backlog\fR (meaning the entire queue is
1186 read back and verified). If \fBverify_backlog_batch\fR is less than
1187 \fBverify_backlog\fR then not all blocks will be verified, if
1188 \fBverify_backlog_batch\fR is larger than \fBverify_backlog\fR, some blocks
1189 will be verified more than once.
1191 .BI trim_percentage \fR=\fPint
1192 Number of verify blocks to discard/trim.
1194 .BI trim_verify_zero \fR=\fPbool
1195 Verify that trim/discarded blocks are returned as zeroes.
1197 .BI trim_backlog \fR=\fPint
1198 Trim after this number of blocks are written.
1200 .BI trim_backlog_batch \fR=\fPint
1201 Trim this number of IO blocks.
1203 .BI experimental_verify \fR=\fPbool
1204 Enable experimental verification.
1206 .BI verify_state_save \fR=\fPbool
1207 When a job exits during the write phase of a verify workload, save its
1208 current state. This allows fio to replay up until that point, if the
1209 verify state is loaded for the verify read phase.
1211 .BI verify_state_load \fR=\fPbool
1212 If a verify termination trigger was used, fio stores the current write
1213 state of each thread. This can be used at verification time so that fio
1214 knows how far it should verify. Without this information, fio will run
1215 a full verification pass, according to the settings in the job file used.
1217 .B stonewall "\fR,\fP wait_for_previous"
1218 Wait for preceding jobs in the job file to exit before starting this one.
1219 \fBstonewall\fR implies \fBnew_group\fR.
1222 Start a new reporting group. If not given, all jobs in a file will be part
1223 of the same reporting group, unless separated by a stonewall.
1225 .BI numjobs \fR=\fPint
1226 Number of clones (processes/threads performing the same workload) of this job.
1230 If set, display per-group reports instead of per-job when \fBnumjobs\fR is
1234 Use threads created with \fBpthread_create\fR\|(3) instead of processes created
1235 with \fBfork\fR\|(2).
1237 .BI zonesize \fR=\fPint
1238 Divide file into zones of the specified size in bytes. See \fBzoneskip\fR.
1240 .BI zonerange \fR=\fPint
1241 Give size of an IO zone. See \fBzoneskip\fR.
1243 .BI zoneskip \fR=\fPint
1244 Skip the specified number of bytes when \fBzonesize\fR bytes of data have been
1247 .BI write_iolog \fR=\fPstr
1248 Write the issued I/O patterns to the specified file. Specify a separate file
1249 for each job, otherwise the iologs will be interspersed and the file may be
1252 .BI read_iolog \fR=\fPstr
1253 Replay the I/O patterns contained in the specified file generated by
1254 \fBwrite_iolog\fR, or may be a \fBblktrace\fR binary file.
1256 .BI replay_no_stall \fR=\fPint
1257 While replaying I/O patterns using \fBread_iolog\fR the default behavior
1258 attempts to respect timing information between I/Os. Enabling
1259 \fBreplay_no_stall\fR causes I/Os to be replayed as fast as possible while
1260 still respecting ordering.
1262 .BI replay_redirect \fR=\fPstr
1263 While replaying I/O patterns using \fBread_iolog\fR the default behavior
1264 is to replay the IOPS onto the major/minor device that each IOP was recorded
1265 from. Setting \fBreplay_redirect\fR causes all IOPS to be replayed onto the
1266 single specified device regardless of the device it was recorded from.
1268 .BI write_bw_log \fR=\fPstr
1269 If given, write a bandwidth log of the jobs in this job file. Can be used to
1270 store data of the bandwidth of the jobs in their lifetime. The included
1271 fio_generate_plots script uses gnuplot to turn these text files into nice
1272 graphs. See \fBwrite_lat_log\fR for behaviour of given filename. For this
1273 option, the postfix is _bw.x.log, where x is the index of the job (1..N,
1274 where N is the number of jobs)
1276 .BI write_lat_log \fR=\fPstr
1277 Same as \fBwrite_bw_log\fR, but writes I/O completion latencies. If no
1278 filename is given with this option, the default filename of
1279 "jobname_type.x.log" is used, where x is the index of the job (1..N, where
1280 N is the number of jobs). Even if the filename is given, fio will still
1281 append the type of log.
1283 .BI write_iops_log \fR=\fPstr
1284 Same as \fBwrite_bw_log\fR, but writes IOPS. If no filename is given with this
1285 option, the default filename of "jobname_type.x.log" is used, where x is the
1286 index of the job (1..N, where N is the number of jobs). Even if the filename
1287 is given, fio will still append the type of log.
1289 .BI log_avg_msec \fR=\fPint
1290 By default, fio will log an entry in the iops, latency, or bw log for every
1291 IO that completes. When writing to the disk log, that can quickly grow to a
1292 very large size. Setting this option makes fio average the each log entry
1293 over the specified period of time, reducing the resolution of the log.
1296 .BI log_offset \fR=\fPbool
1297 If this is set, the iolog options will include the byte offset for the IO
1298 entry as well as the other data values.
1300 .BI log_compression \fR=\fPint
1301 If this is set, fio will compress the IO logs as it goes, to keep the memory
1302 footprint lower. When a log reaches the specified size, that chunk is removed
1303 and compressed in the background. Given that IO logs are fairly highly
1304 compressible, this yields a nice memory savings for longer runs. The downside
1305 is that the compression will consume some background CPU cycles, so it may
1306 impact the run. This, however, is also true if the logging ends up consuming
1307 most of the system memory. So pick your poison. The IO logs are saved
1308 normally at the end of a run, by decompressing the chunks and storing them
1309 in the specified log file. This feature depends on the availability of zlib.
1311 .BI log_store_compressed \fR=\fPbool
1312 If set, and \fBlog\fR_compression is also set, fio will store the log files in
1313 a compressed format. They can be decompressed with fio, using the
1314 \fB\-\-inflate-log\fR command line parameter. The files will be stored with a
1317 .BI block_error_percentiles \fR=\fPbool
1318 If set, record errors in trim block-sized units from writes and trims and output
1319 a histogram of how many trims it took to get to errors, and what kind of error
1322 .BI disable_lat \fR=\fPbool
1323 Disable measurements of total latency numbers. Useful only for cutting
1324 back the number of calls to \fBgettimeofday\fR\|(2), as that does impact performance at
1325 really high IOPS rates. Note that to really get rid of a large amount of these
1326 calls, this option must be used with disable_slat and disable_bw as well.
1328 .BI disable_clat \fR=\fPbool
1329 Disable measurements of completion latency numbers. See \fBdisable_lat\fR.
1331 .BI disable_slat \fR=\fPbool
1332 Disable measurements of submission latency numbers. See \fBdisable_lat\fR.
1334 .BI disable_bw_measurement \fR=\fPbool
1335 Disable measurements of throughput/bandwidth numbers. See \fBdisable_lat\fR.
1337 .BI lockmem \fR=\fPint
1338 Pin the specified amount of memory with \fBmlock\fR\|(2). Can be used to
1339 simulate a smaller amount of memory. The amount specified is per worker.
1341 .BI exec_prerun \fR=\fPstr
1342 Before running the job, execute the specified command with \fBsystem\fR\|(3).
1344 Output is redirected in a file called \fBjobname.prerun.txt\fR
1347 .BI exec_postrun \fR=\fPstr
1348 Same as \fBexec_prerun\fR, but the command is executed after the job completes.
1350 Output is redirected in a file called \fBjobname.postrun.txt\fR
1353 .BI ioscheduler \fR=\fPstr
1354 Attempt to switch the device hosting the file to the specified I/O scheduler.
1356 .BI disk_util \fR=\fPbool
1357 Generate disk utilization statistics if the platform supports it. Default: true.
1359 .BI clocksource \fR=\fPstr
1360 Use the given clocksource as the base of timing. The supported options are:
1364 \fBgettimeofday\fR\|(2)
1367 \fBclock_gettime\fR\|(2)
1370 Internal CPU clock source
1374 \fBcpu\fR is the preferred clocksource if it is reliable, as it is very fast
1375 (and fio is heavy on time calls). Fio will automatically use this clocksource
1376 if it's supported and considered reliable on the system it is running on,
1377 unless another clocksource is specifically set. For x86/x86-64 CPUs, this
1378 means supporting TSC Invariant.
1380 .BI gtod_reduce \fR=\fPbool
1381 Enable all of the \fBgettimeofday\fR\|(2) reducing options (disable_clat, disable_slat,
1382 disable_bw) plus reduce precision of the timeout somewhat to really shrink the
1383 \fBgettimeofday\fR\|(2) call count. With this option enabled, we only do about 0.4% of
1384 the gtod() calls we would have done if all time keeping was enabled.
1386 .BI gtod_cpu \fR=\fPint
1387 Sometimes it's cheaper to dedicate a single thread of execution to just getting
1388 the current time. Fio (and databases, for instance) are very intensive on
1389 \fBgettimeofday\fR\|(2) calls. With this option, you can set one CPU aside for doing
1390 nothing but logging current time to a shared memory location. Then the other
1391 threads/processes that run IO workloads need only copy that segment, instead of
1392 entering the kernel with a \fBgettimeofday\fR\|(2) call. The CPU set aside for doing
1393 these time calls will be excluded from other uses. Fio will manually clear it
1394 from the CPU mask of other jobs.
1396 .BI ignore_error \fR=\fPstr
1397 Sometimes you want to ignore some errors during test in that case you can specify
1398 error list for each error type.
1400 ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST
1402 errors for given error type is separated with ':'.
1403 Error may be symbol ('ENOSPC', 'ENOMEM') or an integer.
1405 Example: ignore_error=EAGAIN,ENOSPC:122 .
1407 This option will ignore EAGAIN from READ, and ENOSPC and 122(EDQUOT) from WRITE.
1409 .BI error_dump \fR=\fPbool
1410 If set dump every error even if it is non fatal, true by default. If disabled
1411 only fatal error will be dumped
1413 .BI profile \fR=\fPstr
1414 Select a specific builtin performance test.
1416 .BI cgroup \fR=\fPstr
1417 Add job to this control group. If it doesn't exist, it will be created.
1418 The system must have a mounted cgroup blkio mount point for this to work. If
1419 your system doesn't have it mounted, you can do so with:
1421 # mount \-t cgroup \-o blkio none /cgroup
1423 .BI cgroup_weight \fR=\fPint
1424 Set the weight of the cgroup to this value. See the documentation that comes
1425 with the kernel, allowed values are in the range of 100..1000.
1427 .BI cgroup_nodelete \fR=\fPbool
1428 Normally fio will delete the cgroups it has created after the job completion.
1429 To override this behavior and to leave cgroups around after the job completion,
1430 set cgroup_nodelete=1. This can be useful if one wants to inspect various
1431 cgroup files after job completion. Default: false
1434 Instead of running as the invoking user, set the user ID to this value before
1435 the thread/process does any work.
1438 Set group ID, see \fBuid\fR.
1440 .BI unit_base \fR=\fPint
1441 Base unit for reporting. Allowed values are:
1445 Use auto-detection (default).
1455 .BI flow_id \fR=\fPint
1456 The ID of the flow. If not specified, it defaults to being a global flow. See
1460 Weight in token-based flow control. If this value is used, then there is a
1461 \fBflow counter\fR which is used to regulate the proportion of activity between
1462 two or more jobs. fio attempts to keep this flow counter near zero. The
1463 \fBflow\fR parameter stands for how much should be added or subtracted to the
1464 flow counter on each iteration of the main I/O loop. That is, if one job has
1465 \fBflow=8\fR and another job has \fBflow=-1\fR, then there will be a roughly
1466 1:8 ratio in how much one runs vs the other.
1468 .BI flow_watermark \fR=\fPint
1469 The maximum value that the absolute value of the flow counter is allowed to
1470 reach before the job must wait for a lower value of the counter.
1472 .BI flow_sleep \fR=\fPint
1473 The period of time, in microseconds, to wait after the flow watermark has been
1474 exceeded before retrying operations
1476 .BI clat_percentiles \fR=\fPbool
1477 Enable the reporting of percentiles of completion latencies.
1479 .BI percentile_list \fR=\fPfloat_list
1480 Overwrite the default list of percentiles for completion latencies and the
1481 block error histogram. Each number is a floating number in the range (0,100],
1482 and the maximum length of the list is 20. Use ':' to separate the
1483 numbers. For example, \-\-percentile_list=99.5:99.9 will cause fio to
1484 report the values of completion latency below which 99.5% and 99.9% of
1485 the observed latencies fell, respectively.
1486 .SS "Ioengine Parameters List"
1487 Some parameters are only valid when a specific ioengine is in use. These are
1488 used identically to normal parameters, with the caveat that when used on the
1489 command line, they must come after the ioengine.
1491 .BI (cpu)cpuload \fR=\fPint
1492 Attempt to use the specified percentage of CPU cycles.
1494 .BI (cpu)cpuchunks \fR=\fPint
1495 Split the load into cycles of the given time. In microseconds.
1497 .BI (cpu)exit_on_io_done \fR=\fPbool
1498 Detect when IO threads are done, then exit.
1500 .BI (libaio)userspace_reap
1501 Normally, with the libaio engine in use, fio will use
1502 the io_getevents system call to reap newly returned events.
1503 With this flag turned on, the AIO ring will be read directly
1504 from user-space to reap events. The reaping mode is only
1505 enabled when polling for a minimum of 0 events (eg when
1506 iodepth_batch_complete=0).
1508 .BI (net,netsplice)hostname \fR=\fPstr
1509 The host name or IP address to use for TCP or UDP based IO.
1510 If the job is a TCP listener or UDP reader, the hostname is not
1511 used and must be omitted unless it is a valid UDP multicast address.
1513 .BI (net,netsplice)port \fR=\fPint
1514 The TCP or UDP port to bind to or connect to. If this is used with
1515 \fBnumjobs\fR to spawn multiple instances of the same job type, then
1516 this will be the starting port number since fio will use a range of ports.
1518 .BI (net,netsplice)interface \fR=\fPstr
1519 The IP address of the network interface used to send or receive UDP multicast
1522 .BI (net,netsplice)ttl \fR=\fPint
1523 Time-to-live value for outgoing UDP multicast packets. Default: 1
1525 .BI (net,netsplice)nodelay \fR=\fPbool
1526 Set TCP_NODELAY on TCP connections.
1528 .BI (net,netsplice)protocol \fR=\fPstr "\fR,\fP proto" \fR=\fPstr
1529 The network protocol to use. Accepted values are:
1534 Transmission control protocol
1537 Transmission control protocol V6
1540 User datagram protocol
1543 User datagram protocol V6
1549 When the protocol is TCP or UDP, the port must also be given,
1550 as well as the hostname if the job is a TCP listener or UDP
1551 reader. For unix sockets, the normal filename option should be
1552 used and the port is invalid.
1555 .BI (net,netsplice)listen
1556 For TCP network connections, tell fio to listen for incoming
1557 connections rather than initiating an outgoing connection. The
1558 hostname must be omitted if this option is used.
1560 .BI (net, pingpong) \fR=\fPbool
1561 Normally a network writer will just continue writing data, and a network reader
1562 will just consume packets. If pingpong=1 is set, a writer will send its normal
1563 payload to the reader, then wait for the reader to send the same payload back.
1564 This allows fio to measure network latencies. The submission and completion
1565 latencies then measure local time spent sending or receiving, and the
1566 completion latency measures how long it took for the other end to receive and
1567 send back. For UDP multicast traffic pingpong=1 should only be set for a single
1568 reader when multiple readers are listening to the same address.
1570 .BI (net, window_size) \fR=\fPint
1571 Set the desired socket buffer size for the connection.
1573 .BI (net, mss) \fR=\fPint
1574 Set the TCP maximum segment size (TCP_MAXSEG).
1576 .BI (e4defrag,donorname) \fR=\fPstr
1577 File will be used as a block donor (swap extents between files)
1579 .BI (e4defrag,inplace) \fR=\fPint
1580 Configure donor file block allocation strategy
1583 Preallocate donor's file on init
1586 allocate space immediately inside defragment event, and free right after event
1589 .BI (rbd)rbdname \fR=\fPstr
1590 Specifies the name of the RBD.
1592 .BI (rbd)pool \fR=\fPstr
1593 Specifies the name of the Ceph pool containing the RBD.
1595 .BI (rbd)clientname \fR=\fPstr
1596 Specifies the username (without the 'client.' prefix) used to access the Ceph cluster.
1598 .BI (mtd)skipbad \fR=\fPbool
1599 Skip operations against known bad blocks.
1601 While running, \fBfio\fR will display the status of the created jobs. For
1605 Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
1608 The characters in the first set of brackets denote the current status of each
1609 threads. The possible values are:
1615 Setup but not started.
1621 Initialized, waiting.
1624 Running, doing sequential reads.
1627 Running, doing random reads.
1630 Running, doing sequential writes.
1633 Running, doing random writes.
1636 Running, doing mixed sequential reads/writes.
1639 Running, doing mixed random reads/writes.
1642 Running, currently waiting for \fBfsync\fR\|(2).
1645 Running, verifying written data.
1648 Exited, not reaped by main thread.
1651 Exited, thread reaped.
1655 The second set of brackets shows the estimated completion percentage of
1656 the current group. The third set shows the read and write I/O rate,
1657 respectively. Finally, the estimated run time of the job is displayed.
1659 When \fBfio\fR completes (or is interrupted by Ctrl-C), it will show data
1660 for each thread, each group of threads, and each disk, in that order.
1662 Per-thread statistics first show the threads client number, group-id, and
1663 error code. The remaining figures are as follows:
1667 Number of megabytes of I/O performed.
1670 Average data rate (bandwidth).
1676 Submission latency minimum, maximum, average and standard deviation. This is
1677 the time it took to submit the I/O.
1680 Completion latency minimum, maximum, average and standard deviation. This
1681 is the time between submission and completion.
1684 Bandwidth minimum, maximum, percentage of aggregate bandwidth received, average
1685 and standard deviation.
1688 CPU usage statistics. Includes user and system time, number of context switches
1689 this thread went through and number of major and minor page faults.
1692 Distribution of I/O depths. Each depth includes everything less than (or equal)
1693 to it, but greater than the previous depth.
1696 Number of read/write requests issued, and number of short read/write requests.
1699 Distribution of I/O completion latencies. The numbers follow the same pattern
1703 The group statistics show:
1708 Number of megabytes I/O performed.
1711 Aggregate bandwidth of threads in the group.
1714 Minimum average bandwidth a thread saw.
1717 Maximum average bandwidth a thread saw.
1720 Shortest runtime of threads in the group.
1723 Longest runtime of threads in the group.
1727 Finally, disk statistics are printed with reads first:
1732 Number of I/Os performed by all groups.
1735 Number of merges in the I/O scheduler.
1738 Number of ticks we kept the disk busy.
1741 Total time spent in the disk queue.
1748 It is also possible to get fio to dump the current output while it is
1749 running, without terminating the job. To do that, send fio the \fBUSR1\fR
1752 If the \fB\-\-minimal\fR / \fB\-\-append-terse\fR options are given, the
1753 results will be printed/appended in a semicolon-delimited format suitable for
1755 A job description (if provided) follows on a new line. Note that the first
1756 number in the line is the version number. If the output has to be changed
1757 for some reason, this number will be incremented by 1 to signify that
1758 change. The fields are:
1761 .B terse version, fio version, jobname, groupid, error
1765 .B Total I/O \fR(KB)\fP, bandwidth \fR(KB/s)\fP, IOPS, runtime \fR(ms)\fP
1769 .B min, max, mean, standard deviation
1773 .B min, max, mean, standard deviation
1775 Completion latency percentiles (20 fields):
1777 .B Xth percentile=usec
1781 .B min, max, mean, standard deviation
1785 .B min, max, aggregate percentage of total, mean, standard deviation
1791 .B Total I/O \fR(KB)\fP, bandwidth \fR(KB/s)\fP, IOPS, runtime \fR(ms)\fP
1795 .B min, max, mean, standard deviation
1799 .B min, max, mean, standard deviation
1801 Completion latency percentiles (20 fields):
1803 .B Xth percentile=usec
1807 .B min, max, mean, standard deviation
1811 .B min, max, aggregate percentage of total, mean, standard deviation
1817 .B user, system, context switches, major page faults, minor page faults
1820 IO depth distribution:
1822 .B <=1, 2, 4, 8, 16, 32, >=64
1825 IO latency distribution:
1829 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
1833 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
1837 Disk utilization (1 for each disk used):
1839 .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
1842 Error Info (dependent on continue_on_error, default off):
1844 .B total # errors, first error code
1847 .B text description (if provided in config - appears on newline)
1850 Normally you would run fio as a stand-alone application on the machine
1851 where the IO workload should be generated. However, it is also possible to
1852 run the frontend and backend of fio separately. This makes it possible to
1853 have a fio server running on the machine(s) where the IO workload should
1854 be running, while controlling it from another machine.
1856 To start the server, you would do:
1858 \fBfio \-\-server=args\fR
1860 on that machine, where args defines what fio listens to. The arguments
1861 are of the form 'type:hostname or IP:port'. 'type' is either 'ip' (or ip4)
1862 for TCP/IP v4, 'ip6' for TCP/IP v6, or 'sock' for a local unix domain
1863 socket. 'hostname' is either a hostname or IP address, and 'port' is the port to
1864 listen to (only valid for TCP/IP, not a local socket). Some examples:
1868 Start a fio server, listening on all interfaces on the default port (8765).
1870 2) fio \-\-server=ip:hostname,4444
1872 Start a fio server, listening on IP belonging to hostname and on port 4444.
1874 3) fio \-\-server=ip6:::1,4444
1876 Start a fio server, listening on IPv6 localhost ::1 and on port 4444.
1878 4) fio \-\-server=,4444
1880 Start a fio server, listening on all interfaces on port 4444.
1882 5) fio \-\-server=1.2.3.4
1884 Start a fio server, listening on IP 1.2.3.4 on the default port.
1886 6) fio \-\-server=sock:/tmp/fio.sock
1888 Start a fio server, listening on the local socket /tmp/fio.sock.
1890 When a server is running, you can connect to it from a client. The client
1893 fio \-\-local-args \-\-client=server \-\-remote-args <job file(s)>
1895 where \-\-local-args are arguments that are local to the client where it is
1896 running, 'server' is the connect string, and \-\-remote-args and <job file(s)>
1897 are sent to the server. The 'server' string follows the same format as it
1898 does on the server side, to allow IP/hostname/socket and port strings.
1899 You can connect to multiple clients as well, to do that you could run:
1901 fio \-\-client=server2 \-\-client=server2 <job file(s)>
1903 If the job file is located on the fio server, then you can tell the server
1904 to load a local file as well. This is done by using \-\-remote-config:
1906 fio \-\-client=server \-\-remote-config /path/to/file.fio
1908 Then the fio serer will open this local (to the server) job file instead
1909 of being passed one from the client.
1913 was written by Jens Axboe <jens.axboe@oracle.com>,
1914 now Jens Axboe <axboe@fb.com>.
1916 This man page was written by Aaron Carroll <aaronc@cse.unsw.edu.au> based
1917 on documentation by Jens Axboe.
1918 .SH "REPORTING BUGS"
1919 Report bugs to the \fBfio\fR mailing list <fio@vger.kernel.org>.
1922 For further documentation see \fBHOWTO\fR and \fBREADME\fR.
1924 Sample jobfiles are available in the \fBexamples\fR directory.
projects / fio.git / blob