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' and 'G', denoting kilo (1024),
82 mega (1024*1024) and giga (1024*1024*1024) respectively. If prefixed with '0x',
83 the value is assumed to be base 16 (hexadecimal).
86 Boolean: a true or false value. `0' denotes false, `1' denotes true.
89 Integer range: a range of integers specified in the format
90 \fIlower\fR:\fIupper\fR or \fIlower\fR\-\fIupper\fR. \fIlower\fR and
91 \fIupper\fR may contain a suffix as described above. If an option allows two
92 sets of ranges, they are separated with a `,' or `/' character. For example:
97 May be used to override the job name. On the command line, this parameter
98 has the special purpose of signalling the start of a new job.
100 .BI description \fR=\fPstr
101 Human-readable description of the job. It is printed when the job is run, but
102 otherwise has no special purpose.
104 .BI directory \fR=\fPstr
105 Prefix filenames with this directory. Used to place files in a location other
108 .BI filename \fR=\fPstr
110 normally makes up a file name based on the job name, thread number, and file
111 number. If you want to share files between threads in a job or several jobs,
112 specify a \fIfilename\fR for each of them to override the default. If the I/O
113 engine used is `net', \fIfilename\fR is the host and port to connect to in the
114 format \fIhost\fR/\fIport\fR. If the I/O engine is file-based, you can specify
115 a number of files by separating the names with a `:' character. `\-' is a
116 reserved name, meaning stdin or stdout, depending on the read/write direction
119 .BI lockfile \fR=\fPstr
120 Fio defaults to not locking any files before it does IO to them. If a file or
121 file descriptor is shared, fio can serialize IO to that file to make the end
122 result consistent. This is usual for emulating real workloads that share files.
128 No locking. This is the default.
131 Only one thread or process may do IO at the time, excluding all others.
134 Read-write locking on the file. Many readers may access the file at the same
135 time, but writes get exclusive access.
138 The option may be post-fixed with a lock batch number. If set, then each
139 thread/process may do that amount of IOs to the file before giving up the lock.
140 Since lock acquisition is expensive, batching the lock/unlocks will speed up IO.
143 .BI opendir \fR=\fPstr
144 Recursively open any files below directory \fIstr\fR.
146 .BI readwrite \fR=\fPstr "\fR,\fP rw" \fR=\fPstr
147 Type of I/O pattern. Accepted values are:
164 Mixed sequential reads and writes.
167 Mixed random reads and writes.
170 For mixed I/O, the default split is 50/50. For random I/O, the number of I/Os
171 to perform before getting a new offset can be specified by appending
172 `:\fIint\fR' to the pattern type. The default is 1.
175 .BI kb_base \fR=\fPint
176 The base unit for a kilobyte. The defacto base is 2^10, 1024. Storage
177 manufacturers like to use 10^3 or 1000 as a base ten unit instead, for obvious
178 reasons. Allow values are 1024 or 1000, with 1024 being the default.
180 .BI randrepeat \fR=\fPbool
181 Seed the random number generator in a predictable way so results are repeatable
182 across runs. Default: true.
184 .BI fadvise_hint \fR=\fPbool
185 Disable use of \fIposix_fadvise\fR\|(2) to advise the kernel what I/O patterns
186 are likely to be issued. Default: true.
189 Total size of I/O for this job. \fBfio\fR will run until this many bytes have
190 been transfered, unless limited by other options (\fBruntime\fR, for instance).
191 Unless \fBnr_files\fR and \fBfilesize\fR options are given, this amount will be
192 divided between the available files for the job.
194 .BI fill_device \fR=\fPbool
195 Sets size to something really large and waits for ENOSPC (no space left on
196 device) as the terminating condition. Only makes sense with sequential write.
197 For a read workload, the mount point will be filled first then IO started on
200 .BI filesize \fR=\fPirange
201 Individual file sizes. May be a range, in which case \fBfio\fR will select sizes
202 for files at random within the given range, limited to \fBsize\fR in total (if
203 that is given). If \fBfilesize\fR is not specified, each created file is the
206 .BI blocksize \fR=\fPint[,int] "\fR,\fB bs" \fR=\fPint[,int]
207 Block size for I/O units. Default: 4k. Values for reads and writes can be
208 specified seperately in the format \fIread\fR,\fIwrite\fR, either of
209 which may be empty to leave that value at its default.
211 .BI blocksize_range \fR=\fPirange[,irange] "\fR,\fB bsrange" \fR=\fPirange[,irange]
212 Specify a range of I/O block sizes. The issued I/O unit will always be a
213 multiple of the minimum size, unless \fBblocksize_unaligned\fR is set. Applies
214 to both reads and writes if only one range is given, but can be specified
215 seperately with a comma seperating the values. Example: bsrange=1k-4k,2k-8k.
216 Also (see \fBblocksize\fR).
218 .BI bssplit \fR=\fPstr
219 This option allows even finer grained control of the block sizes issued,
220 not just even splits between them. With this option, you can weight various
221 block sizes for exact control of the issued IO for a job that has mixed
222 block sizes. The format of the option is bssplit=blocksize/percentage,
223 optionally adding as many definitions as needed seperated by a colon.
224 Example: bssplit=4k/10:64k/50:32k/40 would issue 50% 64k blocks, 10% 4k
225 blocks and 40% 32k blocks. \fBbssplit\fR also supports giving separate
226 splits to reads and writes. The format is identical to what the
227 \fBbs\fR option accepts, the read and write parts are separated with a
230 .B blocksize_unaligned\fR,\fP bs_unaligned
231 If set, any size in \fBblocksize_range\fR may be used. This typically won't
232 work with direct I/O, as that normally requires sector alignment.
234 .BI blockalign \fR=\fPint[,int] "\fR,\fB ba" \fR=\fPint[,int]
235 At what boundary to align random IO offsets. Defaults to the same as 'blocksize'
236 the minimum blocksize given. Minimum alignment is typically 512b
237 for using direct IO, though it usually depends on the hardware block size.
238 This option is mutually exclusive with using a random map for files, so it
239 will turn off that option.
242 Initialise buffers with all zeros. Default: fill buffers with random data.
245 If this option is given, fio will refill the IO buffers on every submit. The
246 default is to only fill it at init time and reuse that data. Only makes sense
247 if zero_buffers isn't specified, naturally. If data verification is enabled,
248 refill_buffers is also automatically enabled.
250 .BI nrfiles \fR=\fPint
251 Number of files to use for this job. Default: 1.
253 .BI openfiles \fR=\fPint
254 Number of files to keep open at the same time. Default: \fBnrfiles\fR.
256 .BI file_service_type \fR=\fPstr
257 Defines how files to service are selected. The following types are defined:
262 Choose a file at random
265 Round robin over open files (default).
267 Do each file in the set sequentially.
270 The number of I/Os to issue before switching a new file can be specified by
271 appending `:\fIint\fR' to the service type.
274 .BI ioengine \fR=\fPstr
275 Defines how the job issues I/O. The following types are defined:
280 Basic \fIread\fR\|(2) or \fIwrite\fR\|(2) I/O. \fIfseek\fR\|(2) is used to
281 position the I/O location.
284 Basic \fIpread\fR\|(2) or \fIpwrite\fR\|(2) I/O.
287 Basic \fIreadv\fR\|(2) or \fIwritev\fR\|(2) I/O. Will emulate queuing by
288 coalescing adjacents IOs into a single submission.
291 Linux native asynchronous I/O.
294 glibc POSIX asynchronous I/O using \fIaio_read\fR\|(3) and \fIaio_write\fR\|(3).
297 File is memory mapped with \fImmap\fR\|(2) and data copied using
301 \fIsplice\fR\|(2) is used to transfer the data and \fIvmsplice\fR\|(2) to
302 transfer data from user-space to the kernel.
305 Use the syslet system calls to make regular read/write asynchronous.
308 SCSI generic sg v3 I/O. May be either synchronous using the SG_IO ioctl, or if
309 the target is an sg character device, we use \fIread\fR\|(2) and
310 \fIwrite\fR\|(2) for asynchronous I/O.
313 Doesn't transfer any data, just pretends to. Mainly used to exercise \fBfio\fR
314 itself and for debugging and testing purposes.
317 Transfer over the network. \fBfilename\fR must be set appropriately to
318 `\fIhost\fR/\fIport\fR' regardless of data direction. If receiving, only the
319 \fIport\fR argument is used.
322 Like \fBnet\fR, but uses \fIsplice\fR\|(2) and \fIvmsplice\fR\|(2) to map data
326 Doesn't transfer any data, but burns CPU cycles according to \fBcpuload\fR and
327 \fBcpucycles\fR parameters.
330 The GUASI I/O engine is the Generic Userspace Asynchronous Syscall Interface
331 approach to asycnronous I/O.
333 See <http://www.xmailserver.org/guasi\-lib.html>.
336 Loads an external I/O engine object file. Append the engine filename as
341 .BI iodepth \fR=\fPint
342 Number of I/O units to keep in flight against the file. Default: 1.
344 .BI iodepth_batch \fR=\fPint
345 Number of I/Os to submit at once. Default: \fBiodepth\fR.
347 .BI iodepth_batch_complete \fR=\fPint
348 This defines how many pieces of IO to retrieve at once. It defaults to 1 which
349 means that we'll ask for a minimum of 1 IO in the retrieval process from the
350 kernel. The IO retrieval will go on until we hit the limit set by
351 \fBiodepth_low\fR. If this variable is set to 0, then fio will always check for
352 completed events before queuing more IO. This helps reduce IO latency, at the
353 cost of more retrieval system calls.
355 .BI iodepth_low \fR=\fPint
356 Low watermark indicating when to start filling the queue again. Default:
359 .BI direct \fR=\fPbool
360 If true, use non-buffered I/O (usually O_DIRECT). Default: false.
362 .BI buffered \fR=\fPbool
363 If true, use buffered I/O. This is the opposite of the \fBdirect\fR parameter.
366 .BI offset \fR=\fPint
367 Offset in the file to start I/O. Data before the offset will not be touched.
370 How many I/Os to perform before issuing an \fBfsync\fR\|(2) of dirty data. If
371 0, don't sync. Default: 0.
373 .BI fdatasync \fR=\fPint
374 Like \fBfsync\fR, but uses \fBfdatasync\fR\|(2) instead to only sync the
375 data parts of the file. Default: 0.
377 .BI overwrite \fR=\fPbool
378 If writing, setup the file first and do overwrites. Default: false.
380 .BI end_fsync \fR=\fPbool
381 Sync file contents when job exits. Default: false.
383 .BI fsync_on_close \fR=\fPbool
384 If true, sync file contents on close. This differs from \fBend_fsync\fR in that
385 it will happen on every close, not just at the end of the job. Default: false.
387 .BI rwmixcycle \fR=\fPint
388 How many milliseconds before switching between reads and writes for a mixed
389 workload. Default: 500ms.
391 .BI rwmixread \fR=\fPint
392 Percentage of a mixed workload that should be reads. Default: 50.
394 .BI rwmixwrite \fR=\fPint
395 Percentage of a mixed workload that should be writes. If \fBrwmixread\fR and
396 \fBrwmixwrite\fR are given and do not sum to 100%, the latter of the two
397 overrides the first. This may interfere with a given rate setting, if fio is
398 asked to limit reads or writes to a certain rate. If that is the case, then
399 the distribution may be skewed. Default: 50.
402 Normally \fBfio\fR will cover every block of the file when doing random I/O. If
403 this parameter is given, a new offset will be chosen without looking at past
404 I/O history. This parameter is mutually exclusive with \fBverify\fR.
407 See \fBnorandommap\fR. If fio runs with the random block map enabled and it
408 fails to allocate the map, if this option is set it will continue without a
409 random block map. As coverage will not be as complete as with random maps, this
410 option is disabled by default.
413 Run job with given nice value. See \fInice\fR\|(2).
416 Set I/O priority value of this job between 0 (highest) and 7 (lowest). See
419 .BI prioclass \fR=\fPint
420 Set I/O priority class. See \fIionice\fR\|(1).
422 .BI thinktime \fR=\fPint
423 Stall job for given number of microseconds between issuing I/Os.
425 .BI thinktime_spin \fR=\fPint
426 Pretend to spend CPU time for given number of microseconds, sleeping the rest
427 of the time specified by \fBthinktime\fR. Only valid if \fBthinktime\fR is set.
429 .BI thinktime_blocks \fR=\fPint
430 Number of blocks to issue before waiting \fBthinktime\fR microseconds.
434 Cap bandwidth used by this job. The number is in bytes/sec, the normal postfix
435 rules apply. You can use \fBrate\fR=500k to limit reads and writes to 500k each,
436 or you can specify read and writes separately. Using \fBrate\fR=1m,500k would
437 limit reads to 1MB/sec and writes to 500KB/sec. Capping only reads or writes
438 can be done with \fBrate\fR=,500k or \fBrate\fR=500k,. The former will only
439 limit writes (to 500KB/sec), the latter will only limit reads.
441 .BI ratemin \fR=\fPint
442 Tell \fBfio\fR to do whatever it can to maintain at least the given bandwidth.
443 Failing to meet this requirement will cause the job to exit. The same format
444 as \fBrate\fR is used for read vs write separation.
446 .BI rate_iops \fR=\fPint
447 Cap the bandwidth to this number of IOPS. Basically the same as rate, just
448 specified independently of bandwidth. The same format as \fBrate\fR is used for
449 read vs write seperation. If \fBblocksize\fR is a range, the smallest block
450 size is used as the metric.
452 .BI rate_iops_min \fR=\fPint
453 If this rate of I/O is not met, the job will exit. The same format as \fBrate\fR
454 is used for read vs write seperation.
456 .BI ratecycle \fR=\fPint
457 Average bandwidth for \fBrate\fR and \fBratemin\fR over this number of
458 milliseconds. Default: 1000ms.
460 .BI cpumask \fR=\fPint
461 Set CPU affinity for this job. \fIint\fR is a bitmask of allowed CPUs the job
462 may run on. See \fBsched_setaffinity\fR\|(2).
464 .BI cpus_allowed \fR=\fPstr
465 Same as \fBcpumask\fR, but allows a comma-delimited list of CPU numbers.
467 .BI startdelay \fR=\fPint
468 Delay start of job for the specified number of seconds.
470 .BI runtime \fR=\fPint
471 Terminate processing after the specified number of seconds.
474 If given, run for the specified \fBruntime\fR duration even if the files are
475 completely read or written. The same workload will be repeated as many times
476 as \fBruntime\fR allows.
478 .BI ramp_time \fR=\fPint
479 If set, fio will run the specified workload for this amount of time before
480 logging any performance numbers. Useful for letting performance settle before
481 logging results, thus minimizing the runtime required for stable results. Note
482 that the \fBramp_time\fR is considered lead in time for a job, thus it will
483 increase the total runtime if a special timeout or runtime is specified.
485 .BI invalidate \fR=\fPbool
486 Invalidate buffer-cache for the file prior to starting I/O. Default: true.
489 Use synchronous I/O for buffered writes. For the majority of I/O engines,
490 this means using O_SYNC. Default: false.
492 .BI iomem \fR=\fPstr "\fR,\fP mem" \fR=\fPstr
493 Allocation method for I/O unit buffer. Allowed values are:
498 Allocate memory with \fImalloc\fR\|(3).
501 Use shared memory buffers allocated through \fIshmget\fR\|(2).
504 Same as \fBshm\fR, but use huge pages as backing.
507 Use \fImmap\fR\|(2) for allocation. Uses anonymous memory unless a filename
508 is given after the option in the format `:\fIfile\fR'.
511 Same as \fBmmap\fR, but use huge files as backing.
514 The amount of memory allocated is the maximum allowed \fBblocksize\fR for the
515 job multiplied by \fBiodepth\fR. For \fBshmhuge\fR or \fBmmaphuge\fR to work,
516 the system must have free huge pages allocated. \fBmmaphuge\fR also needs to
517 have hugetlbfs mounted, and \fIfile\fR must point there.
520 .BI iomem_align \fR=\fPint
521 This indiciates the memory alignment of the IO memory buffers. Note that the
522 given alignment is applied to the first IO unit buffer, if using \fBiodepth\fR
523 the alignment of the following buffers are given by the \fBbs\fR used. In
524 other words, if using a \fBbs\fR that is a multiple of the page sized in the
525 system, all buffers will be aligned to this value. If using a \fBbs\fR that
526 is not page aligned, the alignment of subsequent IO memory buffers is the
527 sum of the \fBiomem_align\fR and \fBbs\fR used.
529 .BI hugepage\-size \fR=\fPint
530 Defines the size of a huge page. Must be at least equal to the system setting.
531 Should be a multiple of 1MB. Default: 4MB.
534 Terminate all jobs when one finishes. Default: wait for each job to finish.
536 .BI bwavgtime \fR=\fPint
537 Average bandwidth calculations over the given time in milliseconds. Default:
540 .BI create_serialize \fR=\fPbool
541 If true, serialize file creation for the jobs. Default: true.
543 .BI create_fsync \fR=\fPbool
544 \fIfsync\fR\|(2) data file after creation. Default: true.
546 .BI create_on_open \fR=\fPbool
547 If true, the files are not created until they are opened for IO by the job.
549 .BI pre_read \fR=\fPbool
550 If this is given, files will be pre-read into memory before starting the given
551 IO operation. This will also clear the \fR \fBinvalidate\fR flag, since it is
552 pointless to pre-read and then drop the cache. This will only work for IO
553 engines that are seekable, since they allow you to read the same data
554 multiple times. Thus it will not work on eg network or splice IO.
556 .BI unlink \fR=\fPbool
557 Unlink job files when done. Default: false.
560 Specifies the number of iterations (runs of the same workload) of this job.
563 .BI do_verify \fR=\fPbool
564 Run the verify phase after a write phase. Only valid if \fBverify\fR is set.
567 .BI verify \fR=\fPstr
568 Method of verifying file contents after each iteration of the job. Allowed
573 .B md5 crc16 crc32 crc64 crc7 sha256 sha512
574 Store appropriate checksum in the header of each block.
577 Write extra information about each I/O (timestamp, block number, etc.). The
578 block number is verified.
581 Fill I/O buffers with a specific pattern that is used to verify. The pattern is
582 specified by appending `:\fIint\fR' to the parameter. \fIint\fR cannot be larger
586 Pretend to verify. Used for testing internals.
590 .BI verify_sort \fR=\fPbool
591 If true, written verify blocks are sorted if \fBfio\fR deems it to be faster to
592 read them back in a sorted manner. Default: true.
594 .BI verify_offset \fR=\fPint
595 Swap the verification header with data somewhere else in the block before
596 writing. It is swapped back before verifying.
598 .BI verify_interval \fR=\fPint
599 Write the verification header for this number of bytes, which should divide
600 \fBblocksize\fR. Default: \fBblocksize\fR.
602 .BI verify_fatal \fR=\fPbool
603 If true, exit the job on the first observed verification failure. Default:
606 .BI verify_async \fR=\fPint
607 Fio will normally verify IO inline from the submitting thread. This option
608 takes an integer describing how many async offload threads to create for IO
609 verification instead, causing fio to offload the duty of verifying IO contents
610 to one or more separate threads. If using this offload option, even sync IO
611 engines can benefit from using an \fBiodepth\fR setting higher than 1, as it
612 allows them to have IO in flight while verifies are running.
614 .BI verify_async_cpus \fR=\fPstr
615 Tell fio to set the given CPU affinity on the async IO verification threads.
616 See \fBcpus_allowed\fP for the format used.
619 Wait for preceeding jobs in the job file to exit before starting this one.
620 \fBstonewall\fR implies \fBnew_group\fR.
623 Start a new reporting group. If not given, all jobs in a file will be part
624 of the same reporting group, unless separated by a stonewall.
626 .BI numjobs \fR=\fPint
627 Number of clones (processes/threads performing the same workload) of this job.
631 If set, display per-group reports instead of per-job when \fBnumjobs\fR is
635 Use threads created with \fBpthread_create\fR\|(3) instead of processes created
636 with \fBfork\fR\|(2).
638 .BI zonesize \fR=\fPint
639 Divide file into zones of the specified size in bytes. See \fBzoneskip\fR.
641 .BI zoneskip \fR=\fPint
642 Skip the specified number of bytes when \fBzonesize\fR bytes of data have been
645 .BI write_iolog \fR=\fPstr
646 Write the issued I/O patterns to the specified file.
648 .BI read_iolog \fR=\fPstr
649 Replay the I/O patterns contained in the specified file generated by
650 \fBwrite_iolog\fR, or may be a \fBblktrace\fR binary file.
652 .B write_bw_log \fR=\fPstr
653 If given, write a bandwidth log of the jobs in this job file. Can be used to
654 store data of the bandwidth of the jobs in their lifetime. The included
655 fio_generate_plots script uses gnuplot to turn these text files into nice
656 graphs. See \fBwrite_log_log\fR for behaviour of given filename. For this
657 option, the postfix is _bw.log.
660 Same as \fBwrite_bw_log\fR, but writes I/O completion latencies. If no
661 filename is given with this option, the default filename of "jobname_type.log"
662 is used. Even if the filename is given, fio will still append the type of log.
664 .B disable_clat \fR=\fPbool
665 Disable measurements of completion latency numbers. Useful only for cutting
666 back the number of calls to gettimeofday, as that does impact performance at
667 really high IOPS rates. Note that to really get rid of a large amount of these
668 calls, this option must be used with disable_slat and disable_bw as well.
670 .B disable_slat \fR=\fPbool
671 Disable measurements of submission latency numbers. See \fBdisable_clat\fR.
673 .B disable_bw_measurement \fR=\fPbool
674 Disable measurements of throughput/bandwidth numbers. See \fBdisable_clat\fR.
676 .BI lockmem \fR=\fPint
677 Pin the specified amount of memory with \fBmlock\fR\|(2). Can be used to
678 simulate a smaller amount of memory.
680 .BI exec_prerun \fR=\fPstr
681 Before running the job, execute the specified command with \fBsystem\fR\|(3).
683 .BI exec_postrun \fR=\fPstr
684 Same as \fBexec_prerun\fR, but the command is executed after the job completes.
686 .BI ioscheduler \fR=\fPstr
687 Attempt to switch the device hosting the file to the specified I/O scheduler.
689 .BI cpuload \fR=\fPint
690 If the job is a CPU cycle-eater, attempt to use the specified percentage of
693 .BI cpuchunks \fR=\fPint
694 If the job is a CPU cycle-eater, split the load into cycles of the
695 given time in milliseconds.
697 .BI disk_util \fR=\fPbool
698 Generate disk utilization statistics if the platform supports it. Default: true.
700 .BI gtod_reduce \fR=\fPbool
701 Enable all of the gettimeofday() reducing options (disable_clat, disable_slat,
702 disable_bw) plus reduce precision of the timeout somewhat to really shrink the
703 gettimeofday() call count. With this option enabled, we only do about 0.4% of
704 the gtod() calls we would have done if all time keeping was enabled.
706 .BI gtod_cpu \fR=\fPint
707 Sometimes it's cheaper to dedicate a single thread of execution to just getting
708 the current time. Fio (and databases, for instance) are very intensive on
709 gettimeofday() calls. With this option, you can set one CPU aside for doing
710 nothing but logging current time to a shared memory location. Then the other
711 threads/processes that run IO workloads need only copy that segment, instead of
712 entering the kernel with a gettimeofday() call. The CPU set aside for doing
713 these time calls will be excluded from other uses. Fio will manually clear it
714 from the CPU mask of other jobs.
716 .BI continue_on_error \fR=\fPbool
717 Normally fio will exit the job on the first observed failure. If this option is
718 set, fio will continue the job when there is a 'non-fatal error'
719 (\fBEIO\fR or \fBEILSEQ\fR) until the runtime is exceeded or the I/O size
720 specified is completed. If this option is used, there are two more stats that
721 are appended, the total error count and the first error. The error field given
722 in the stats is the first error that was hit during the run.
724 While running, \fBfio\fR will display the status of the created jobs. For
728 Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
731 The characters in the first set of brackets denote the current status of each
732 threads. The possible values are:
738 Setup but not started.
744 Initialized, waiting.
747 Running, doing sequential reads.
750 Running, doing random reads.
753 Running, doing sequential writes.
756 Running, doing random writes.
759 Running, doing mixed sequential reads/writes.
762 Running, doing mixed random reads/writes.
765 Running, currently waiting for \fBfsync\fR\|(2).
768 Running, verifying written data.
771 Exited, not reaped by main thread.
774 Exited, thread reaped.
778 The second set of brackets shows the estimated completion percentage of
779 the current group. The third set shows the read and write I/O rate,
780 respectively. Finally, the estimated run time of the job is displayed.
782 When \fBfio\fR completes (or is interrupted by Ctrl-C), it will show data
783 for each thread, each group of threads, and each disk, in that order.
785 Per-thread statistics first show the threads client number, group-id, and
786 error code. The remaining figures are as follows:
790 Number of megabytes of I/O performed.
793 Average data rate (bandwidth).
799 Submission latency minimum, maximum, average and standard deviation. This is
800 the time it took to submit the I/O.
803 Completion latency minimum, maximum, average and standard deviation. This
804 is the time between submission and completion.
807 Bandwidth minimum, maximum, percentage of aggregate bandwidth received, average
808 and standard deviation.
811 CPU usage statistics. Includes user and system time, number of context switches
812 this thread went through and number of major and minor page faults.
815 Distribution of I/O depths. Each depth includes everything less than (or equal)
816 to it, but greater than the previous depth.
819 Number of read/write requests issued, and number of short read/write requests.
822 Distribution of I/O completion latencies. The numbers follow the same pattern
826 The group statistics show:
831 Number of megabytes I/O performed.
834 Aggregate bandwidth of threads in the group.
837 Minimum average bandwidth a thread saw.
840 Maximum average bandwidth a thread saw.
843 Shortest runtime of threads in the group.
846 Longest runtime of threads in the group.
850 Finally, disk statistics are printed with reads first:
855 Number of I/Os performed by all groups.
858 Number of merges in the I/O scheduler.
861 Number of ticks we kept the disk busy.
864 Total time spent in the disk queue.
871 If the \fB\-\-minimal\fR option is given, the results will be printed in a
872 semicolon-delimited format suitable for scripted use. The fields are:
875 .B jobname, groupid, error
879 .B KB I/O, bandwidth \fR(KB/s)\fP, runtime \fR(ms)\fP
883 .B min, max, mean, standard deviation
887 .B min, max, mean, standard deviation
891 .B min, max, aggregate percentage of total, mean, standard deviation
897 .B KB I/O, bandwidth \fR(KB/s)\fP, runtime \fR(ms)\fP
901 .B min, max, mean, standard deviation
905 .B min, max, mean, standard deviation
909 .B min, max, aggregate percentage of total, mean, standard deviation
915 .B user, system, context switches, major page faults, minor page faults
918 IO depth distribution:
920 .B <=1, 2, 4, 8, 16, 32, >=64
923 IO latency distribution (ms):
925 .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, >=2000
932 was written by Jens Axboe <jens.axboe@oracle.com>.
934 This man page was written by Aaron Carroll <aaronc@cse.unsw.edu.au> based
935 on documentation by Jens Axboe.
937 Report bugs to the \fBfio\fR mailing list <fio@vger.kernel.org>.
940 For further documentation see \fBHOWTO\fR and \fBREADME\fR.
942 Sample jobfiles are available in the \fBexamples\fR directory.
projects / fio.git / blob