1 .TH fio 1 "September 2007" "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 \-\-output \fR=\fPfilename
16 Write output to \fIfilename\fR.
18 .BI \-\-timeout \fR=\fPtimeout
19 Limit run time to \fItimeout\fR seconds.
22 Generate per-job latency logs.
25 Generate per-job bandwidth logs.
28 Print statistics in a terse, semicolon-delimited format.
30 .BI \-\-showcmd \fR=\fPjobfile
31 Convert \fIjobfile\fR to a set of command-line options.
34 Enable read-only safety checks.
36 .BI \-\-eta \fR=\fPwhen
37 Specifies when real-time ETA estimate should be printed. \fIwhen\fR may
38 be one of `always', `never' or `auto'.
40 .BI \-\-section \fR=\fPsec
41 Only run section \fIsec\fR from job file.
43 .BI \-\-cmdhelp \fR=\fPcommand
44 Print help information for \fIcommand\fR. May be `all' for all commands.
46 .BI \-\-debug \fR=\fPtype
47 Enable verbose tracing of various fio actions. May be `all' for all types
48 or individual types seperated by a comma (eg \-\-debug=io,file). `help' will
49 list all available tracing options.
52 Display usage information and exit.
55 Display version information and exit.
57 Job files are in `ini' format. They consist of one or more
58 job definitions, which begin with a job name in square brackets and
59 extend to the next job name. The job name can be any ASCII string
60 except `global', which has a special meaning. Following the job name is
61 a sequence of zero or more parameters, one per line, that define the
62 behavior of the job. Any line starting with a `;' or `#' character is
63 considered a comment and ignored.
65 If \fIjobfile\fR is specified as `-', the job file will be read from
68 The global section contains default parameters for jobs specified in the
69 job file. A job is only affected by global sections residing above it,
70 and there may be any number of global sections. Specific job definitions
71 may override any parameter set in global sections.
74 Some parameters may take arguments of a specific type. The types used are:
77 String: a sequence of alphanumeric characters.
80 SI integer: a whole number, possibly containing a suffix denoting the base unit
81 of the value. Accepted suffixes are `k', 'M', 'G', 'T', and 'P', denoting
82 kilo (1024), mega (1024^2), giga (1024^3), tera (1024^4), and peta (1024^5)
83 respectively. The suffix is not case sensitive. If prefixed with '0x', the
84 value is assumed to be base 16 (hexadecimal).
87 Boolean: a true or false value. `0' denotes false, `1' denotes true.
90 Integer range: a range of integers specified in the format
91 \fIlower\fR:\fIupper\fR or \fIlower\fR\-\fIupper\fR. \fIlower\fR and
92 \fIupper\fR may contain a suffix as described above. If an option allows two
93 sets of ranges, they are separated with a `,' or `/' character. For example:
98 May be used to override the job name. On the command line, this parameter
99 has the special purpose of signalling the start of a new job.
101 .BI description \fR=\fPstr
102 Human-readable description of the job. It is printed when the job is run, but
103 otherwise has no special purpose.
105 .BI directory \fR=\fPstr
106 Prefix filenames with this directory. Used to place files in a location other
109 .BI filename \fR=\fPstr
111 normally makes up a file name based on the job name, thread number, and file
112 number. If you want to share files between threads in a job or several jobs,
113 specify a \fIfilename\fR for each of them to override the default. If the I/O
114 engine used is `net', \fIfilename\fR is the host and port to connect to in the
115 format \fIhost\fR/\fIport\fR. If the I/O engine is file-based, you can specify
116 a number of files by separating the names with a `:' character. `\-' is a
117 reserved name, meaning stdin or stdout, depending on the read/write direction
120 .BI lockfile \fR=\fPstr
121 Fio defaults to not locking any files before it does IO to them. If a file or
122 file descriptor is shared, fio can serialize IO to that file to make the end
123 result consistent. This is usual for emulating real workloads that share files.
129 No locking. This is the default.
132 Only one thread or process may do IO at the time, excluding all others.
135 Read-write locking on the file. Many readers may access the file at the same
136 time, but writes get exclusive access.
139 The option may be post-fixed with a lock batch number. If set, then each
140 thread/process may do that amount of IOs to the file before giving up the lock.
141 Since lock acquisition is expensive, batching the lock/unlocks will speed up IO.
144 .BI opendir \fR=\fPstr
145 Recursively open any files below directory \fIstr\fR.
147 .BI readwrite \fR=\fPstr "\fR,\fP rw" \fR=\fPstr
148 Type of I/O pattern. Accepted values are:
165 Mixed sequential reads and writes.
168 Mixed random reads and writes.
171 For mixed I/O, the default split is 50/50. For random I/O, the number of I/Os
172 to perform before getting a new offset can be specified by appending
173 `:\fIint\fR' to the pattern type. The default is 1.
176 .BI kb_base \fR=\fPint
177 The base unit for a kilobyte. The defacto base is 2^10, 1024. Storage
178 manufacturers like to use 10^3 or 1000 as a base ten unit instead, for obvious
179 reasons. Allow values are 1024 or 1000, with 1024 being the default.
181 .BI randrepeat \fR=\fPbool
182 Seed the random number generator in a predictable way so results are repeatable
183 across runs. Default: true.
185 .BI fallocate \fR=\fPbool
186 By default, fio will use fallocate() to advise the system of the size of the
187 file we are going to write. This can be turned off with fallocate=0. May not
188 be available on all supported platforms.
190 .BI fadvise_hint \fR=\fPbool
191 Disable use of \fIposix_fadvise\fR\|(2) to advise the kernel what I/O patterns
192 are likely to be issued. Default: true.
195 Total size of I/O for this job. \fBfio\fR will run until this many bytes have
196 been transfered, unless limited by other options (\fBruntime\fR, for instance).
197 Unless \fBnr_files\fR and \fBfilesize\fR options are given, this amount will be
198 divided between the available files for the job.
200 .BI fill_device \fR=\fPbool
201 Sets size to something really large and waits for ENOSPC (no space left on
202 device) as the terminating condition. Only makes sense with sequential write.
203 For a read workload, the mount point will be filled first then IO started on
206 .BI filesize \fR=\fPirange
207 Individual file sizes. May be a range, in which case \fBfio\fR will select sizes
208 for files at random within the given range, limited to \fBsize\fR in total (if
209 that is given). If \fBfilesize\fR is not specified, each created file is the
212 .BI blocksize \fR=\fPint[,int] "\fR,\fB bs" \fR=\fPint[,int]
213 Block size for I/O units. Default: 4k. Values for reads and writes can be
214 specified seperately in the format \fIread\fR,\fIwrite\fR, either of
215 which may be empty to leave that value at its default.
217 .BI blocksize_range \fR=\fPirange[,irange] "\fR,\fB bsrange" \fR=\fPirange[,irange]
218 Specify a range of I/O block sizes. The issued I/O unit will always be a
219 multiple of the minimum size, unless \fBblocksize_unaligned\fR is set. Applies
220 to both reads and writes if only one range is given, but can be specified
221 seperately with a comma seperating the values. Example: bsrange=1k-4k,2k-8k.
222 Also (see \fBblocksize\fR).
224 .BI bssplit \fR=\fPstr
225 This option allows even finer grained control of the block sizes issued,
226 not just even splits between them. With this option, you can weight various
227 block sizes for exact control of the issued IO for a job that has mixed
228 block sizes. The format of the option is bssplit=blocksize/percentage,
229 optionally adding as many definitions as needed seperated by a colon.
230 Example: bssplit=4k/10:64k/50:32k/40 would issue 50% 64k blocks, 10% 4k
231 blocks and 40% 32k blocks. \fBbssplit\fR also supports giving separate
232 splits to reads and writes. The format is identical to what the
233 \fBbs\fR option accepts, the read and write parts are separated with a
236 .B blocksize_unaligned\fR,\fP bs_unaligned
237 If set, any size in \fBblocksize_range\fR may be used. This typically won't
238 work with direct I/O, as that normally requires sector alignment.
240 .BI blockalign \fR=\fPint[,int] "\fR,\fB ba" \fR=\fPint[,int]
241 At what boundary to align random IO offsets. Defaults to the same as 'blocksize'
242 the minimum blocksize given. Minimum alignment is typically 512b
243 for using direct IO, though it usually depends on the hardware block size.
244 This option is mutually exclusive with using a random map for files, so it
245 will turn off that option.
248 Initialise buffers with all zeros. Default: fill buffers with random data.
251 If this option is given, fio will refill the IO buffers on every submit. The
252 default is to only fill it at init time and reuse that data. Only makes sense
253 if zero_buffers isn't specified, naturally. If data verification is enabled,
254 refill_buffers is also automatically enabled.
256 .BI nrfiles \fR=\fPint
257 Number of files to use for this job. Default: 1.
259 .BI openfiles \fR=\fPint
260 Number of files to keep open at the same time. Default: \fBnrfiles\fR.
262 .BI file_service_type \fR=\fPstr
263 Defines how files to service are selected. The following types are defined:
268 Choose a file at random
271 Round robin over open files (default).
273 Do each file in the set sequentially.
276 The number of I/Os to issue before switching a new file can be specified by
277 appending `:\fIint\fR' to the service type.
280 .BI ioengine \fR=\fPstr
281 Defines how the job issues I/O. The following types are defined:
286 Basic \fIread\fR\|(2) or \fIwrite\fR\|(2) I/O. \fIfseek\fR\|(2) is used to
287 position the I/O location.
290 Basic \fIpread\fR\|(2) or \fIpwrite\fR\|(2) I/O.
293 Basic \fIreadv\fR\|(2) or \fIwritev\fR\|(2) I/O. Will emulate queuing by
294 coalescing adjacents IOs into a single submission.
297 Linux native asynchronous I/O.
300 glibc POSIX asynchronous I/O using \fIaio_read\fR\|(3) and \fIaio_write\fR\|(3).
303 File is memory mapped with \fImmap\fR\|(2) and data copied using
307 \fIsplice\fR\|(2) is used to transfer the data and \fIvmsplice\fR\|(2) to
308 transfer data from user-space to the kernel.
311 Use the syslet system calls to make regular read/write asynchronous.
314 SCSI generic sg v3 I/O. May be either synchronous using the SG_IO ioctl, or if
315 the target is an sg character device, we use \fIread\fR\|(2) and
316 \fIwrite\fR\|(2) for asynchronous I/O.
319 Doesn't transfer any data, just pretends to. Mainly used to exercise \fBfio\fR
320 itself and for debugging and testing purposes.
323 Transfer over the network. \fBfilename\fR must be set appropriately to
324 `\fIhost\fR/\fIport\fR' regardless of data direction. If receiving, only the
325 \fIport\fR argument is used.
328 Like \fBnet\fR, but uses \fIsplice\fR\|(2) and \fIvmsplice\fR\|(2) to map data
332 Doesn't transfer any data, but burns CPU cycles according to \fBcpuload\fR and
333 \fBcpucycles\fR parameters.
336 The GUASI I/O engine is the Generic Userspace Asynchronous Syscall Interface
337 approach to asycnronous I/O.
339 See <http://www.xmailserver.org/guasi\-lib.html>.
342 Loads an external I/O engine object file. Append the engine filename as
347 .BI iodepth \fR=\fPint
348 Number of I/O units to keep in flight against the file. Default: 1.
350 .BI iodepth_batch \fR=\fPint
351 Number of I/Os to submit at once. Default: \fBiodepth\fR.
353 .BI iodepth_batch_complete \fR=\fPint
354 This defines how many pieces of IO to retrieve at once. It defaults to 1 which
355 means that we'll ask for a minimum of 1 IO in the retrieval process from the
356 kernel. The IO retrieval will go on until we hit the limit set by
357 \fBiodepth_low\fR. If this variable is set to 0, then fio will always check for
358 completed events before queuing more IO. This helps reduce IO latency, at the
359 cost of more retrieval system calls.
361 .BI iodepth_low \fR=\fPint
362 Low watermark indicating when to start filling the queue again. Default:
365 .BI direct \fR=\fPbool
366 If true, use non-buffered I/O (usually O_DIRECT). Default: false.
368 .BI buffered \fR=\fPbool
369 If true, use buffered I/O. This is the opposite of the \fBdirect\fR parameter.
372 .BI offset \fR=\fPint
373 Offset in the file to start I/O. Data before the offset will not be touched.
376 How many I/Os to perform before issuing an \fBfsync\fR\|(2) of dirty data. If
377 0, don't sync. Default: 0.
379 .BI fdatasync \fR=\fPint
380 Like \fBfsync\fR, but uses \fBfdatasync\fR\|(2) instead to only sync the
381 data parts of the file. Default: 0.
383 .BI overwrite \fR=\fPbool
384 If writing, setup the file first and do overwrites. Default: false.
386 .BI end_fsync \fR=\fPbool
387 Sync file contents when job exits. Default: false.
389 .BI fsync_on_close \fR=\fPbool
390 If true, sync file contents on close. This differs from \fBend_fsync\fR in that
391 it will happen on every close, not just at the end of the job. Default: false.
393 .BI rwmixcycle \fR=\fPint
394 How many milliseconds before switching between reads and writes for a mixed
395 workload. Default: 500ms.
397 .BI rwmixread \fR=\fPint
398 Percentage of a mixed workload that should be reads. Default: 50.
400 .BI rwmixwrite \fR=\fPint
401 Percentage of a mixed workload that should be writes. If \fBrwmixread\fR and
402 \fBrwmixwrite\fR are given and do not sum to 100%, the latter of the two
403 overrides the first. This may interfere with a given rate setting, if fio is
404 asked to limit reads or writes to a certain rate. If that is the case, then
405 the distribution may be skewed. Default: 50.
408 Normally \fBfio\fR will cover every block of the file when doing random I/O. If
409 this parameter is given, a new offset will be chosen without looking at past
410 I/O history. This parameter is mutually exclusive with \fBverify\fR.
413 See \fBnorandommap\fR. If fio runs with the random block map enabled and it
414 fails to allocate the map, if this option is set it will continue without a
415 random block map. As coverage will not be as complete as with random maps, this
416 option is disabled by default.
419 Run job with given nice value. See \fInice\fR\|(2).
422 Set I/O priority value of this job between 0 (highest) and 7 (lowest). See
425 .BI prioclass \fR=\fPint
426 Set I/O priority class. See \fIionice\fR\|(1).
428 .BI thinktime \fR=\fPint
429 Stall job for given number of microseconds between issuing I/Os.
431 .BI thinktime_spin \fR=\fPint
432 Pretend to spend CPU time for given number of microseconds, sleeping the rest
433 of the time specified by \fBthinktime\fR. Only valid if \fBthinktime\fR is set.
435 .BI thinktime_blocks \fR=\fPint
436 Number of blocks to issue before waiting \fBthinktime\fR microseconds.
440 Cap bandwidth used by this job. The number is in bytes/sec, the normal postfix
441 rules apply. You can use \fBrate\fR=500k to limit reads and writes to 500k each,
442 or you can specify read and writes separately. Using \fBrate\fR=1m,500k would
443 limit reads to 1MB/sec and writes to 500KB/sec. Capping only reads or writes
444 can be done with \fBrate\fR=,500k or \fBrate\fR=500k,. The former will only
445 limit writes (to 500KB/sec), the latter will only limit reads.
447 .BI ratemin \fR=\fPint
448 Tell \fBfio\fR to do whatever it can to maintain at least the given bandwidth.
449 Failing to meet this requirement will cause the job to exit. The same format
450 as \fBrate\fR is used for read vs write separation.
452 .BI rate_iops \fR=\fPint
453 Cap the bandwidth to this number of IOPS. Basically the same as rate, just
454 specified independently of bandwidth. The same format as \fBrate\fR is used for
455 read vs write seperation. If \fBblocksize\fR is a range, the smallest block
456 size is used as the metric.
458 .BI rate_iops_min \fR=\fPint
459 If this rate of I/O is not met, the job will exit. The same format as \fBrate\fR
460 is used for read vs write seperation.
462 .BI ratecycle \fR=\fPint
463 Average bandwidth for \fBrate\fR and \fBratemin\fR over this number of
464 milliseconds. Default: 1000ms.
466 .BI cpumask \fR=\fPint
467 Set CPU affinity for this job. \fIint\fR is a bitmask of allowed CPUs the job
468 may run on. See \fBsched_setaffinity\fR\|(2).
470 .BI cpus_allowed \fR=\fPstr
471 Same as \fBcpumask\fR, but allows a comma-delimited list of CPU numbers.
473 .BI startdelay \fR=\fPint
474 Delay start of job for the specified number of seconds.
476 .BI runtime \fR=\fPint
477 Terminate processing after the specified number of seconds.
480 If given, run for the specified \fBruntime\fR duration even if the files are
481 completely read or written. The same workload will be repeated as many times
482 as \fBruntime\fR allows.
484 .BI ramp_time \fR=\fPint
485 If set, fio will run the specified workload for this amount of time before
486 logging any performance numbers. Useful for letting performance settle before
487 logging results, thus minimizing the runtime required for stable results. Note
488 that the \fBramp_time\fR is considered lead in time for a job, thus it will
489 increase the total runtime if a special timeout or runtime is specified.
491 .BI invalidate \fR=\fPbool
492 Invalidate buffer-cache for the file prior to starting I/O. Default: true.
495 Use synchronous I/O for buffered writes. For the majority of I/O engines,
496 this means using O_SYNC. Default: false.
498 .BI iomem \fR=\fPstr "\fR,\fP mem" \fR=\fPstr
499 Allocation method for I/O unit buffer. Allowed values are:
504 Allocate memory with \fImalloc\fR\|(3).
507 Use shared memory buffers allocated through \fIshmget\fR\|(2).
510 Same as \fBshm\fR, but use huge pages as backing.
513 Use \fImmap\fR\|(2) for allocation. Uses anonymous memory unless a filename
514 is given after the option in the format `:\fIfile\fR'.
517 Same as \fBmmap\fR, but use huge files as backing.
520 The amount of memory allocated is the maximum allowed \fBblocksize\fR for the
521 job multiplied by \fBiodepth\fR. For \fBshmhuge\fR or \fBmmaphuge\fR to work,
522 the system must have free huge pages allocated. \fBmmaphuge\fR also needs to
523 have hugetlbfs mounted, and \fIfile\fR must point there. At least on Linux,
524 huge pages must be manually allocated. See \fB/proc/sys/vm/nr_hugehages\fR
525 and the documentation for that. Normally you just need to echo an appropriate
526 number, eg echoing 8 will ensure that the OS has 8 huge pages ready for
530 .BI iomem_align \fR=\fPint
531 This indiciates the memory alignment of the IO memory buffers. Note that the
532 given alignment is applied to the first IO unit buffer, if using \fBiodepth\fR
533 the alignment of the following buffers are given by the \fBbs\fR used. In
534 other words, if using a \fBbs\fR that is a multiple of the page sized in the
535 system, all buffers will be aligned to this value. If using a \fBbs\fR that
536 is not page aligned, the alignment of subsequent IO memory buffers is the
537 sum of the \fBiomem_align\fR and \fBbs\fR used.
539 .BI hugepage\-size \fR=\fPint
540 Defines the size of a huge page. Must be at least equal to the system setting.
541 Should be a multiple of 1MB. Default: 4MB.
544 Terminate all jobs when one finishes. Default: wait for each job to finish.
546 .BI bwavgtime \fR=\fPint
547 Average bandwidth calculations over the given time in milliseconds. Default:
550 .BI create_serialize \fR=\fPbool
551 If true, serialize file creation for the jobs. Default: true.
553 .BI create_fsync \fR=\fPbool
554 \fIfsync\fR\|(2) data file after creation. Default: true.
556 .BI create_on_open \fR=\fPbool
557 If true, the files are not created until they are opened for IO by the job.
559 .BI pre_read \fR=\fPbool
560 If this is given, files will be pre-read into memory before starting the given
561 IO operation. This will also clear the \fR \fBinvalidate\fR flag, since it is
562 pointless to pre-read and then drop the cache. This will only work for IO
563 engines that are seekable, since they allow you to read the same data
564 multiple times. Thus it will not work on eg network or splice IO.
566 .BI unlink \fR=\fPbool
567 Unlink job files when done. Default: false.
570 Specifies the number of iterations (runs of the same workload) of this job.
573 .BI do_verify \fR=\fPbool
574 Run the verify phase after a write phase. Only valid if \fBverify\fR is set.
577 .BI verify \fR=\fPstr
578 Method of verifying file contents after each iteration of the job. Allowed
583 .B md5 crc16 crc32 crc32c crc32c-intel crc64 crc7 sha256 sha512 sha1
584 Store appropriate checksum in the header of each block.
587 Write extra information about each I/O (timestamp, block number, etc.). The
588 block number is verified.
591 Fill I/O buffers with a specific pattern that is used to verify. If the pattern
592 is < 4bytes, it can either be a decimal or a hexadecimal number. If the pattern
593 is > 4bytes, currently, it can only be a hexadecimal pattern starting with
597 Pretend to verify. Used for testing internals.
600 This option can be used for repeated burn-in tests of a system to make sure
601 that the written data is also correctly read back. If the data direction given
602 is a read or random read, fio will assume that it should verify a previously
603 written file. If the data direction includes any form of write, the verify will
604 be of the newly written data.
607 .BI verify_sort \fR=\fPbool
608 If true, written verify blocks are sorted if \fBfio\fR deems it to be faster to
609 read them back in a sorted manner. Default: true.
611 .BI verify_offset \fR=\fPint
612 Swap the verification header with data somewhere else in the block before
613 writing. It is swapped back before verifying.
615 .BI verify_interval \fR=\fPint
616 Write the verification header for this number of bytes, which should divide
617 \fBblocksize\fR. Default: \fBblocksize\fR.
619 .BI verify_fatal \fR=\fPbool
620 If true, exit the job on the first observed verification failure. Default:
623 .BI verify_async \fR=\fPint
624 Fio will normally verify IO inline from the submitting thread. This option
625 takes an integer describing how many async offload threads to create for IO
626 verification instead, causing fio to offload the duty of verifying IO contents
627 to one or more separate threads. If using this offload option, even sync IO
628 engines can benefit from using an \fBiodepth\fR setting higher than 1, as it
629 allows them to have IO in flight while verifies are running.
631 .BI verify_async_cpus \fR=\fPstr
632 Tell fio to set the given CPU affinity on the async IO verification threads.
633 See \fBcpus_allowed\fP for the format used.
636 Wait for preceeding jobs in the job file to exit before starting this one.
637 \fBstonewall\fR implies \fBnew_group\fR.
640 Start a new reporting group. If not given, all jobs in a file will be part
641 of the same reporting group, unless separated by a stonewall.
643 .BI numjobs \fR=\fPint
644 Number of clones (processes/threads performing the same workload) of this job.
648 If set, display per-group reports instead of per-job when \fBnumjobs\fR is
652 Use threads created with \fBpthread_create\fR\|(3) instead of processes created
653 with \fBfork\fR\|(2).
655 .BI zonesize \fR=\fPint
656 Divide file into zones of the specified size in bytes. See \fBzoneskip\fR.
658 .BI zoneskip \fR=\fPint
659 Skip the specified number of bytes when \fBzonesize\fR bytes of data have been
662 .BI write_iolog \fR=\fPstr
663 Write the issued I/O patterns to the specified file.
665 .BI read_iolog \fR=\fPstr
666 Replay the I/O patterns contained in the specified file generated by
667 \fBwrite_iolog\fR, or may be a \fBblktrace\fR binary file.
669 .B write_bw_log \fR=\fPstr
670 If given, write a bandwidth log of the jobs in this job file. Can be used to
671 store data of the bandwidth of the jobs in their lifetime. The included
672 fio_generate_plots script uses gnuplot to turn these text files into nice
673 graphs. See \fBwrite_log_log\fR for behaviour of given filename. For this
674 option, the postfix is _bw.log.
677 Same as \fBwrite_bw_log\fR, but writes I/O completion latencies. If no
678 filename is given with this option, the default filename of "jobname_type.log"
679 is used. Even if the filename is given, fio will still append the type of log.
681 .B disable_clat \fR=\fPbool
682 Disable measurements of completion latency numbers. Useful only for cutting
683 back the number of calls to gettimeofday, as that does impact performance at
684 really high IOPS rates. Note that to really get rid of a large amount of these
685 calls, this option must be used with disable_slat and disable_bw as well.
687 .B disable_slat \fR=\fPbool
688 Disable measurements of submission latency numbers. See \fBdisable_clat\fR.
690 .B disable_bw_measurement \fR=\fPbool
691 Disable measurements of throughput/bandwidth numbers. See \fBdisable_clat\fR.
693 .BI lockmem \fR=\fPint
694 Pin the specified amount of memory with \fBmlock\fR\|(2). Can be used to
695 simulate a smaller amount of memory.
697 .BI exec_prerun \fR=\fPstr
698 Before running the job, execute the specified command with \fBsystem\fR\|(3).
700 .BI exec_postrun \fR=\fPstr
701 Same as \fBexec_prerun\fR, but the command is executed after the job completes.
703 .BI ioscheduler \fR=\fPstr
704 Attempt to switch the device hosting the file to the specified I/O scheduler.
706 .BI cpuload \fR=\fPint
707 If the job is a CPU cycle-eater, attempt to use the specified percentage of
710 .BI cpuchunks \fR=\fPint
711 If the job is a CPU cycle-eater, split the load into cycles of the
712 given time in milliseconds.
714 .BI disk_util \fR=\fPbool
715 Generate disk utilization statistics if the platform supports it. Default: true.
717 .BI gtod_reduce \fR=\fPbool
718 Enable all of the gettimeofday() reducing options (disable_clat, disable_slat,
719 disable_bw) plus reduce precision of the timeout somewhat to really shrink the
720 gettimeofday() call count. With this option enabled, we only do about 0.4% of
721 the gtod() calls we would have done if all time keeping was enabled.
723 .BI gtod_cpu \fR=\fPint
724 Sometimes it's cheaper to dedicate a single thread of execution to just getting
725 the current time. Fio (and databases, for instance) are very intensive on
726 gettimeofday() calls. With this option, you can set one CPU aside for doing
727 nothing but logging current time to a shared memory location. Then the other
728 threads/processes that run IO workloads need only copy that segment, instead of
729 entering the kernel with a gettimeofday() call. The CPU set aside for doing
730 these time calls will be excluded from other uses. Fio will manually clear it
731 from the CPU mask of other jobs.
733 .BI cgroup \fR=\fPstr
734 Add job to this control group. If it doesn't exist, it will be created.
735 The system must have a mounted cgroup blkio mount point for this to work. If
736 your system doesn't have it mounted, you can do so with:
738 # mount -t cgroup -o blkio none /cgroup
740 .BI cgroup_weight \fR=\fPint
741 Set the weight of the cgroup to this value. See the documentation that comes
742 with the kernel, allowed values are in the range of 100..1000.
745 Instead of running as the invoking user, set the user ID to this value before
746 the thread/process does any work.
749 Set group ID, see \fBuid\fR.
751 While running, \fBfio\fR will display the status of the created jobs. For
755 Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
758 The characters in the first set of brackets denote the current status of each
759 threads. The possible values are:
765 Setup but not started.
771 Initialized, waiting.
774 Running, doing sequential reads.
777 Running, doing random reads.
780 Running, doing sequential writes.
783 Running, doing random writes.
786 Running, doing mixed sequential reads/writes.
789 Running, doing mixed random reads/writes.
792 Running, currently waiting for \fBfsync\fR\|(2).
795 Running, verifying written data.
798 Exited, not reaped by main thread.
801 Exited, thread reaped.
805 The second set of brackets shows the estimated completion percentage of
806 the current group. The third set shows the read and write I/O rate,
807 respectively. Finally, the estimated run time of the job is displayed.
809 When \fBfio\fR completes (or is interrupted by Ctrl-C), it will show data
810 for each thread, each group of threads, and each disk, in that order.
812 Per-thread statistics first show the threads client number, group-id, and
813 error code. The remaining figures are as follows:
817 Number of megabytes of I/O performed.
820 Average data rate (bandwidth).
826 Submission latency minimum, maximum, average and standard deviation. This is
827 the time it took to submit the I/O.
830 Completion latency minimum, maximum, average and standard deviation. This
831 is the time between submission and completion.
834 Bandwidth minimum, maximum, percentage of aggregate bandwidth received, average
835 and standard deviation.
838 CPU usage statistics. Includes user and system time, number of context switches
839 this thread went through and number of major and minor page faults.
842 Distribution of I/O depths. Each depth includes everything less than (or equal)
843 to it, but greater than the previous depth.
846 Number of read/write requests issued, and number of short read/write requests.
849 Distribution of I/O completion latencies. The numbers follow the same pattern
853 The group statistics show:
858 Number of megabytes I/O performed.
861 Aggregate bandwidth of threads in the group.
864 Minimum average bandwidth a thread saw.
867 Maximum average bandwidth a thread saw.
870 Shortest runtime of threads in the group.
873 Longest runtime of threads in the group.
877 Finally, disk statistics are printed with reads first:
882 Number of I/Os performed by all groups.
885 Number of merges in the I/O scheduler.
888 Number of ticks we kept the disk busy.
891 Total time spent in the disk queue.
898 If the \fB\-\-minimal\fR option is given, the results will be printed in a
899 semicolon-delimited format suitable for scripted use. The fields are:
902 .B jobname, groupid, error
906 .B KB I/O, bandwidth \fR(KB/s)\fP, runtime \fR(ms)\fP
910 .B min, max, mean, standard deviation
914 .B min, max, mean, standard deviation
918 .B min, max, aggregate percentage of total, mean, standard deviation
924 .B KB I/O, bandwidth \fR(KB/s)\fP, runtime \fR(ms)\fP
928 .B min, max, mean, standard deviation
932 .B min, max, mean, standard deviation
936 .B min, max, aggregate percentage of total, mean, standard deviation
942 .B user, system, context switches, major page faults, minor page faults
945 IO depth distribution:
947 .B <=1, 2, 4, 8, 16, 32, >=64
950 IO latency distribution (ms):
952 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, >=2000
959 was written by Jens Axboe <jens.axboe@oracle.com>.
961 This man page was written by Aaron Carroll <aaronc@cse.unsw.edu.au> based
962 on documentation by Jens Axboe.
964 Report bugs to the \fBfio\fR mailing list <fio@vger.kernel.org>.
967 For further documentation see \fBHOWTO\fR and \fBREADME\fR.
969 Sample jobfiles are available in the \fBexamples\fR directory.