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 \-\-cmdhelp \fR=\fPcommand
41 Print help information for \fIcommand\fR. May be `all' for all commands.
43 .BI \-\-debug \fR=\fPtype
44 Enable verbose tracing of various fio actions. May be `all' for all types
45 or individual types seperated by a comma (eg --debug=io,file). `help' will
46 list all available tracing options.
48 Print help information for \fIcommand\fR. May be `all' for all commands.
51 Display usage information and exit.
54 Display version information and exit.
56 Job files are in `ini' format. They consist of one or more
57 job definitions, which begin with a job name in square brackets and
58 extend to the next job name. The job name can be any ASCII string
59 except `global', which has a special meaning. Following the job name is
60 a sequence of zero or more parameters, one per line, that define the
61 behavior of the job. Any line starting with a `;' or `#' character is
62 considered a comment and ignored.
64 If \fIjobfile\fR is specified as `-', the job file will be read from
67 The global section contains default parameters for jobs specified in the
68 job file. A job is only affected by global sections residing above it,
69 and there may be any number of global sections. Specific job definitions
70 may override any parameter set in global sections.
73 Some parameters may take arguments of a specific type. The types used are:
76 String: a sequence of alphanumeric characters.
79 Integer: a whole number, possibly negative. If prefixed with `0x', the value
80 is assumed to be base 16 (hexadecimal).
83 SI integer: a whole number, possibly containing a suffix denoting the base unit
84 of the value. Accepted suffixes are `k', 'M' and 'G', denoting kilo (1024),
85 mega (1024*1024) and giga (1024*1024*1024) respectively.
88 Boolean: a true or false value. `0' denotes false, `1' denotes true.
91 Integer range: a range of integers specified in the format
92 \fIlower\fR:\fIupper\fR or \fIlower\fR\-\fIupper\fR. \fIlower\fR and
93 \fIupper\fR may contain a suffix as described above. If an option allows two
94 sets of ranges, they are separated with a `,' or `/' character. For example:
99 May be used to override the job name. On the command line, this parameter
100 has the special purpose of signalling the start of a new job.
102 .BI description \fR=\fPstr
103 Human-readable description of the job. It is printed when the job is run, but
104 otherwise has no special purpose.
106 .BI directory \fR=\fPstr
107 Prefix filenames with this directory. Used to place files in a location other
110 .BI filename \fR=\fPstr
112 normally makes up a file name based on the job name, thread number, and file
113 number. If you want to share files between threads in a job or several jobs,
114 specify a \fIfilename\fR for each of them to override the default. If the I/O
115 engine used is `net', \fIfilename\fR is the host and port to connect to in the
116 format \fIhost\fR/\fIport\fR. If the I/O engine is file-based, you can specify
117 a number of files by separating the names with a `:' character. `\-' is a
118 reserved name, meaning stdin or stdout, depending on the read/write direction
121 .BI opendir \fR=\fPstr
122 Recursively open any files below directory \fIstr\fR.
124 .BI readwrite \fR=\fPstr "\fR,\fP rw" \fR=\fPstr
125 Type of I/O pattern. Accepted values are:
142 Mixed sequential reads and writes.
145 Mixed random reads and writes.
148 For mixed I/O, the default split is 50/50. For random I/O, the number of I/Os
149 to perform before getting a new offset can be specified by appending
150 `:\fIint\fR' to the pattern type. The default is 1.
153 .BI randrepeat \fR=\fPbool
154 Seed the random number generator in a predictable way so results are repeatable
155 across runs. Default: true.
157 .BI fadvise_hint \fR=\fPbool
158 Disable use of \fIposix_fadvise\fR\|(2) to advise the kernel what I/O patterns
159 are likely to be issued. Default: true.
161 .BI size \fR=\fPsiint
162 Total size of I/O for this job. \fBfio\fR will run until this many bytes have
163 been transfered, unless limited by other options (\fBruntime\fR, for instance).
164 Unless \fBnr_files\fR and \fBfilesize\fR options are given, this amount will be
165 divided between the available files for the job.
167 .BI filesize \fR=\fPirange
168 Individual file sizes. May be a range, in which case \fBfio\fR will select sizes
169 for files at random within the given range, limited to \fBsize\fR in total (if
170 that is given). If \fBfilesize\fR is not specified, each created file is the
173 .BI blocksize \fR=\fPsiint[,siint] "\fR,\fB bs" \fR=\fPsiint[,siint]
174 Block size for I/O units. Default: 4k. Values for reads and writes can be
175 specified seperately in the format \fIread\fR,\fIwrite\fR, either of
176 which may be empty to leave that value at its default.
178 .BI blocksize_range \fR=\fPirange[,irange] "\fR,\fB bsrange" \fR=\fPirange[,irange]
179 Specify a range of I/O block sizes. The issued I/O unit will always be a
180 multiple of the minimum size, unless \fBblocksize_unaligned\fR is set. Applies
181 to both reads and writes if only one range is given, but can be specified
182 seperately with a comma seperating the values. Example: bsrange=1k-4k,2k-8k.
183 Also (see \fBblocksize\fR).
185 .BI bssplit \fR=\fPstr
186 This option allows even finer grained control of the block sizes issued,
187 not just even splits between them. With this option, you can weight various
188 block sizes for exact control of the issued IO for a job that has mixed
189 block sizes. The format of the option is bssplit=blocksize/percentage,
190 optionally adding as many definitions as needed seperated by a colon.
191 Example: bssplit=4k/10:64k/50:32k/40 would issue 50% 64k blocks, 10% 4k
192 blocks and 40% 32k blocks.
194 .B blocksize_unaligned\fR,\fP bs_unaligned
195 If set, any size in \fBblocksize_range\fR may be used. This typically won't
196 work with direct I/O, as that normally requires sector alignment.
199 Initialise buffers with all zeros. Default: fill buffers with random data.
201 .BI nrfiles \fR=\fPint
202 Number of files to use for this job. Default: 1.
204 .BI openfiles \fR=\fPint
205 Number of files to keep open at the same time. Default: \fBnrfiles\fR.
207 .BI file_service_type \fR=\fPstr
208 Defines how files to service are selected. The following types are defined:
213 Choose a file at random
216 Round robin over open files (default).
219 The number of I/Os to issue before switching a new file can be specified by
220 appending `:\fIint\fR' to the service type.
223 .BI ioengine \fR=\fPstr
224 Defines how the job issues I/O. The following types are defined:
229 Basic \fIread\fR\|(2) or \fIwrite\fR\|(2) I/O. \fIfseek\fR\|(2) is used to
230 position the I/O location.
233 Basic \fIpread\fR\|(2) or \fIpwrite\fR\|(2) I/O.
236 Basic \fIreadv\fR\|(2) or \fIwritev\fR\|(2) I/O. Will emulate queuing by
237 coalescing adjacents IOs into a single submission.
240 Linux native asynchronous I/O.
243 glibc POSIX asynchronous I/O using \fIaio_read\fR\|(3) and \fIaio_write\fR\|(3).
246 File is memory mapped with \fImmap\fR\|(2) and data copied using
250 \fIsplice\fR\|(2) is used to transfer the data and \fIvmsplice\fR\|(2) to
251 transfer data from user-space to the kernel.
254 Use the syslet system calls to make regular read/write asynchronous.
257 SCSI generic sg v3 I/O. May be either synchronous using the SG_IO ioctl, or if
258 the target is an sg character device, we use \fIread\fR\|(2) and
259 \fIwrite\fR\|(2) for asynchronous I/O.
262 Doesn't transfer any data, just pretends to. Mainly used to exercise \fBfio\fR
263 itself and for debugging and testing purposes.
266 Transfer over the network. \fBfilename\fR must be set appropriately to
267 `\fIhost\fR/\fIport\fR' regardless of data direction. If receiving, only the
268 \fIport\fR argument is used.
271 Like \fBnet\fR, but uses \fIsplice\fR\|(2) and \fIvmsplice\fR\|(2) to map data
275 Doesn't transfer any data, but burns CPU cycles according to \fBcpuload\fR and
276 \fBcpucycles\fR parameters.
279 The GUASI I/O engine is the Generic Userspace Asynchronous Syscall Interface
280 approach to asycnronous I/O.
282 See <http://www.xmailserver.org/guasi\-lib.html>.
285 Loads an external I/O engine object file. Append the engine filename as
290 .BI iodepth \fR=\fPint
291 Number of I/O units to keep in flight against the file. Default: 1.
293 .BI iodepth_batch \fR=\fPint
294 Number of I/Os to submit at once. Default: \fBiodepth\fR.
296 .BI iodepth_low \fR=\fPint
297 Low watermark indicating when to start filling the queue again. Default:
300 .BI direct \fR=\fPbool
301 If true, use non-buffered I/O (usually O_DIRECT). Default: false.
303 .BI buffered \fR=\fPbool
304 If true, use buffered I/O. This is the opposite of the \fBdirect\fR parameter.
307 .BI offset \fR=\fPsiint
308 Offset in the file to start I/O. Data before the offset will not be touched.
311 How many I/Os to perform before issuing an \fBfsync\fR\|(2) of dirty data. If
312 0, don't sync. Default: 0.
314 .BI overwrite \fR=\fPbool
315 If writing, setup the file first and do overwrites. Default: false.
317 .BI end_fsync \fR=\fPbool
318 Sync file contents when job exits. Default: false.
320 .BI fsync_on_close \fR=\fPbool
321 If true, sync file contents on close. This differs from \fBend_fsync\fR in that
322 it will happen on every close, not just at the end of the job. Default: false.
324 .BI rwmixcycle \fR=\fPint
325 How many milliseconds before switching between reads and writes for a mixed
326 workload. Default: 500ms.
328 .BI rwmixread \fR=\fPint
329 Percentage of a mixed workload that should be reads. Default: 50.
331 .BI rwmixwrite \fR=\fPint
332 Percentage of a mixed workload that should be writes. If \fBrwmixread\fR and
333 \fBwrmixwrite\fR are given and do not sum to 100%, the latter of the two
334 overrides the first. Default: 50.
337 Normally \fBfio\fR will cover every block of the file when doing random I/O. If
338 this parameter is given, a new offset will be chosen without looking at past
339 I/O history. This parameter is mutually exclusive with \fBverify\fR.
342 Run job with given nice value. See \fInice\fR\|(2).
345 Set I/O priority value of this job between 0 (highest) and 7 (lowest). See
348 .BI prioclass \fR=\fPint
349 Set I/O priority class. See \fIionice\fR\|(1).
351 .BI thinktime \fR=\fPint
352 Stall job for given number of microseconds between issuing I/Os.
354 .BI thinktime_spin \fR=\fPint
355 Pretend to spend CPU time for given number of microseconds, sleeping the rest
356 of the time specified by \fBthinktime\fR. Only valid if \fBthinktime\fR is set.
358 .BI thinktime_blocks \fR=\fPint
359 Number of blocks to issue before waiting \fBthinktime\fR microseconds.
363 Cap bandwidth used by this job to this number of KiB/s.
365 .BI ratemin \fR=\fPint
366 Tell \fBfio\fR to do whatever it can to maintain at least the given bandwidth.
367 Failing to meet this requirement will cause the job to exit.
369 .BI rate_iops \fR=\fPint
370 Cap the bandwidth to this number of IOPS. If \fBblocksize\fR is a range, the
371 smallest block size is used as the metric.
373 .BI rate_iops_min \fR=\fPint
374 If this rate of I/O is not met, the job will exit.
376 .BI ratecycle \fR=\fPint
377 Average bandwidth for \fBrate\fR and \fBratemin\fR over this number of
378 milliseconds. Default: 1000ms.
380 .BI cpumask \fR=\fPint
381 Set CPU affinity for this job. \fIint\fR is a bitmask of allowed CPUs the job
382 may run on. See \fBsched_setaffinity\fR\|(2).
384 .BI cpus_allowed \fR=\fPstr
385 Same as \fBcpumask\fR, but allows a comma-delimited list of CPU numbers.
387 .BI startdelay \fR=\fPint
388 Delay start of job for the specified number of seconds.
390 .BI runtime \fR=\fPint
391 Terminate processing after the specified number of seconds.
394 If given, run for the specified \fBruntime\fR duration even if the files are
395 completely read or written. The same workload will be repeated as many times
396 as \fBruntime\fR allows.
398 .BI invalidate \fR=\fPbool
399 Invalidate buffer-cache for the file prior to starting I/O. Default: true.
402 Use synchronous I/O for buffered writes. For the majority of I/O engines,
403 this means using O_SYNC. Default: false.
405 .BI iomem \fR=\fPstr "\fR,\fP mem" \fR=\fPstr
406 Allocation method for I/O unit buffer. Allowed values are:
411 Allocate memory with \fImalloc\fR\|(3).
414 Use shared memory buffers allocated through \fIshmget\fR\|(2).
417 Same as \fBshm\fR, but use huge pages as backing.
420 Use \fImmap\fR\|(2) for allocation. Uses anonymous memory unless a filename
421 is given after the option in the format `:\fIfile\fR'.
424 Same as \fBmmap\fR, but use huge files as backing.
427 The amount of memory allocated is the maximum allowed \fBblocksize\fR for the
428 job multiplied by \fBiodepth\fR. For \fBshmhuge\fR or \fBmmaphuge\fR to work,
429 the system must have free huge pages allocated. \fBmmaphuge\fR also needs to
430 have hugetlbfs mounted, and \fIfile\fR must point there.
433 .BI hugepage\-size \fR=\fPsiint
434 Defines the size of a huge page. Must be at least equal to the system setting.
435 Should be a multiple of 1MiB. Default: 4MiB.
438 Terminate all jobs when one finishes. Default: wait for each job to finish.
440 .BI bwavgtime \fR=\fPint
441 Average bandwidth calculations over the given time in milliseconds. Default:
444 .BI create_serialize \fR=\fPbool
445 If true, serialize file creation for the jobs. Default: true.
447 .BI create_fsync \fR=\fPbool
448 \fIfsync\fR\|(2) data file after creation. Default: true.
450 .BI unlink \fR=\fPbool
451 Unlink job files when done. Default: false.
454 Specifies the number of iterations (runs of the same workload) of this job.
457 .BI do_verify \fR=\fPbool
458 Run the verify phase after a write phase. Only valid if \fBverify\fR is set.
461 .BI verify \fR=\fPstr
462 Method of verifying file contents after each iteration of the job. Allowed
467 .B md5 crc16 crc32 crc64 crc7 sha256 sha512
468 Store appropriate checksum in the header of each block.
471 Write extra information about each I/O (timestamp, block number, etc.). The
472 block number is verified.
475 Fill I/O buffers with a specific pattern that is used to verify. The pattern is
476 specified by appending `:\fIint\fR' to the parameter. \fIint\fR cannot be larger
480 Pretend to verify. Used for testing internals.
484 .BI verify_sort \fR=\fPbool
485 If true, written verify blocks are sorted if \fBfio\fR deems it to be faster to
486 read them back in a sorted manner. Default: true.
488 .BI verify_offset \fR=\fPsiint
489 Swap the verification header with data somewhere else in the block before
490 writing. It is swapped back before verifying.
492 .BI verify_interval \fR=\fPsiint
493 Write the verification header for this number of bytes, which should divide
494 \fBblocksize\fR. Default: \fBblocksize\fR.
496 .BI verify_fatal \fR=\fPbool
497 If true, exit the job on the first observed verification failure. Default:
501 Wait for preceeding jobs in the job file to exit before starting this one.
502 \fBstonewall\fR implies \fBnew_group\fR.
505 Start a new reporting group. If not given, all jobs in a file will be part
506 of the same reporting group, unless separated by a stonewall.
508 .BI numjobs \fR=\fPint
509 Number of clones (processes/threads performing the same workload) of this job.
513 If set, display per-group reports instead of per-job when \fBnumjobs\fR is
517 Use threads created with \fBpthread_create\fR\|(3) instead of processes created
518 with \fBfork\fR\|(2).
520 .BI zonesize \fR=\fPsiint
521 Divide file into zones of the specified size in bytes. See \fBzoneskip\fR.
523 .BI zoneskip \fR=\fPsiint
524 Skip the specified number of bytes when \fBzonesize\fR bytes of data have been
527 .BI write_iolog \fR=\fPstr
528 Write the issued I/O patterns to the specified file.
530 .BI read_iolog \fR=\fPstr
531 Replay the I/O patterns contained in the specified file generated by
532 \fBwrite_iolog\fR, or may be a \fBblktrace\fR binary file.
535 If given, write bandwidth logs of the jobs in this file.
538 Same as \fBwrite_bw_log\fR, but writes I/O completion latencies.
540 .BI lockmem \fR=\fPsiint
541 Pin the specified amount of memory with \fBmlock\fR\|(2). Can be used to
542 simulate a smaller amount of memory.
544 .BI exec_prerun \fR=\fPstr
545 Before running the job, execute the specified command with \fBsystem\fR\|(3).
547 .BI exec_postrun \fR=\fPstr
548 Same as \fBexec_prerun\fR, but the command is executed after the job completes.
550 .BI ioscheduler \fR=\fPstr
551 Attempt to switch the device hosting the file to the specified I/O scheduler.
553 .BI cpuload \fR=\fPint
554 If the job is a CPU cycle-eater, attempt to use the specified percentage of
557 .BI cpuchunks \fR=\fPint
558 If the job is a CPU cycle-eater, split the load into cycles of the
559 given time in milliseconds.
561 .BI disk_util \fR=\fPbool
562 Generate disk utilization statistics if the platform supports it. Default: true.
564 While running, \fBfio\fR will display the status of the created jobs. For
568 Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
571 The characters in the first set of brackets denote the current status of each
572 threads. The possible values are:
578 Setup but not started.
584 Initialized, waiting.
587 Running, doing sequential reads.
590 Running, doing random reads.
593 Running, doing sequential writes.
596 Running, doing random writes.
599 Running, doing mixed sequential reads/writes.
602 Running, doing mixed random reads/writes.
605 Running, currently waiting for \fBfsync\fR\|(2).
608 Running, verifying written data.
611 Exited, not reaped by main thread.
614 Exited, thread reaped.
618 The second set of brackets shows the estimated completion percentage of
619 the current group. The third set shows the read and write I/O rate,
620 respectively. Finally, the estimated run time of the job is displayed.
622 When \fBfio\fR completes (or is interrupted by Ctrl-C), it will show data
623 for each thread, each group of threads, and each disk, in that order.
625 Per-thread statistics first show the threads client number, group-id, and
626 error code. The remaining figures are as follows:
630 Number of megabytes of I/O performed.
633 Average data rate (bandwidth).
639 Submission latency minimum, maximum, average and standard deviation. This is
640 the time it took to submit the I/O.
643 Completion latency minimum, maximum, average and standard deviation. This
644 is the time between submission and completion.
647 Bandwidth minimum, maximum, percentage of aggregate bandwidth received, average
648 and standard deviation.
651 CPU usage statistics. Includes user and system time, number of context switches
652 this thread went through and number of major and minor page faults.
655 Distribution of I/O depths. Each depth includes everything less than (or equal)
656 to it, but greater than the previous depth.
659 Number of read/write requests issued, and number of short read/write requests.
662 Distribution of I/O completion latencies. The numbers follow the same pattern
666 The group statistics show:
671 Number of megabytes I/O performed.
674 Aggregate bandwidth of threads in the group.
677 Minimum average bandwidth a thread saw.
680 Maximum average bandwidth a thread saw.
683 Shortest runtime of threads in the group.
686 Longest runtime of threads in the group.
690 Finally, disk statistics are printed with reads first:
695 Number of I/Os performed by all groups.
698 Number of merges in the I/O scheduler.
701 Number of ticks we kept the disk busy.
704 Total time spent in the disk queue.
711 If the \fB\-\-minimal\fR option is given, the results will be printed in a
712 semicolon-delimited format suitable for scripted use. The fields are:
715 .B jobname, groupid, error
719 .B KiB I/O, bandwidth \fR(KiB/s)\fP, runtime \fR(ms)\fP
723 .B min, max, mean, standard deviation
727 .B min, max, mean, standard deviation
731 .B min, max, aggregate percentage of total, mean, standard deviation
737 .B KiB I/O, bandwidth \fR(KiB/s)\fP, runtime \fR(ms)\fP
741 .B min, max, mean, standard deviation
745 .B min, max, mean, standard deviation
749 .B min, max, aggregate percentage of total, mean, standard deviation
755 .B user, system, context switches
758 IO depth distribution:
760 .B <=1, 2, 4, 8, 16, 32, >=64
763 IO latency distribution (ms):
765 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, >=2000
772 was written by Jens Axboe <jens.axboe@oracle.com>.
774 This man page was written by Aaron Carroll <aaronc@cse.unsw.edu.au> based
775 on documentation by Jens Axboe.
777 Report bugs to the \fBfio\fR mailing list <fio-devel@kernel.dk>.
780 For further documentation see \fBHOWTO\fR and \fBREADME\fR.
782 Sample jobfiles are available in the \fBexamples\fR directory.
projects / fio.git / blob