net: support for unix domain sockets
[fio.git] / fio.1
CommitLineData
d60e92d1
AC
1.TH fio 1 "September 2007" "User Manual"
2.SH NAME
3fio \- flexible I/O tester
4.SH SYNOPSIS
5.B fio
6[\fIoptions\fR] [\fIjobfile\fR]...
7.SH DESCRIPTION
8.B fio
9is a tool that will spawn a number of threads or processes doing a
10particular type of I/O action as specified by the user.
11The typical use of fio is to write a job file matching the I/O load
12one wants to simulate.
13.SH OPTIONS
14.TP
15.BI \-\-output \fR=\fPfilename
16Write output to \fIfilename\fR.
17.TP
18.BI \-\-timeout \fR=\fPtimeout
19Limit run time to \fItimeout\fR seconds.
20.TP
21.B \-\-latency\-log
22Generate per-job latency logs.
23.TP
24.B \-\-bandwidth\-log
25Generate per-job bandwidth logs.
26.TP
27.B \-\-minimal
d1429b5c 28Print statistics in a terse, semicolon-delimited format.
d60e92d1
AC
29.TP
30.BI \-\-showcmd \fR=\fPjobfile
31Convert \fIjobfile\fR to a set of command-line options.
32.TP
33.B \-\-readonly
34Enable read-only safety checks.
35.TP
36.BI \-\-eta \fR=\fPwhen
37Specifies when real-time ETA estimate should be printed. \fIwhen\fR may
38be one of `always', `never' or `auto'.
39.TP
c0a5d35e
AC
40.BI \-\-section \fR=\fPsec
41Only run section \fIsec\fR from job file.
42.TP
d60e92d1
AC
43.BI \-\-cmdhelp \fR=\fPcommand
44Print help information for \fIcommand\fR. May be `all' for all commands.
45.TP
9183788d
JA
46.BI \-\-debug \fR=\fPtype
47Enable verbose tracing of various fio actions. May be `all' for all types
5982a925 48or individual types separated by a comma (eg \-\-debug=io,file). `help' will
9183788d
JA
49list all available tracing options.
50.TP
d60e92d1
AC
51.B \-\-help
52Display usage information and exit.
53.TP
54.B \-\-version
55Display version information and exit.
f57a9c59
JA
56.TP
57.B \-\-terse\-version\fR=\fPtype
58Terse version output format
d60e92d1
AC
59.SH "JOB FILE FORMAT"
60Job files are in `ini' format. They consist of one or more
61job definitions, which begin with a job name in square brackets and
62extend to the next job name. The job name can be any ASCII string
63except `global', which has a special meaning. Following the job name is
64a sequence of zero or more parameters, one per line, that define the
65behavior of the job. Any line starting with a `;' or `#' character is
d1429b5c 66considered a comment and ignored.
d9956b64
AC
67.P
68If \fIjobfile\fR is specified as `-', the job file will be read from
69standard input.
d60e92d1
AC
70.SS "Global Section"
71The global section contains default parameters for jobs specified in the
72job file. A job is only affected by global sections residing above it,
73and there may be any number of global sections. Specific job definitions
74may override any parameter set in global sections.
75.SH "JOB PARAMETERS"
76.SS Types
77Some parameters may take arguments of a specific type. The types used are:
78.TP
79.I str
80String: a sequence of alphanumeric characters.
81.TP
82.I int
d60e92d1 83SI integer: a whole number, possibly containing a suffix denoting the base unit
b09da8fa
JA
84of the value. Accepted suffixes are `k', 'M', 'G', 'T', and 'P', denoting
85kilo (1024), mega (1024^2), giga (1024^3), tera (1024^4), and peta (1024^5)
86respectively. The suffix is not case sensitive. If prefixed with '0x', the
5982a925
MS
87value is assumed to be base 16 (hexadecimal). A suffix may include a trailing 'b',
88for instance 'kb' is identical to 'k'. You can specify a base 10 value
57fc29fa
JA
89by using 'KiB', 'MiB', 'GiB', etc. This is useful for disk drives where
90values are often given in base 10 values. Specifying '30GiB' will get you
9130*1000^3 bytes.
d60e92d1
AC
92.TP
93.I bool
94Boolean: a true or false value. `0' denotes false, `1' denotes true.
95.TP
96.I irange
97Integer range: a range of integers specified in the format
d1429b5c
AC
98\fIlower\fR:\fIupper\fR or \fIlower\fR\-\fIupper\fR. \fIlower\fR and
99\fIupper\fR may contain a suffix as described above. If an option allows two
100sets of ranges, they are separated with a `,' or `/' character. For example:
101`8\-8k/8M\-4G'.
83349190
YH
102.TP
103.I float_list
104List of floating numbers: A list of floating numbers, separated by
105a ':' charcater.
d60e92d1
AC
106.SS "Parameter List"
107.TP
108.BI name \fR=\fPstr
d9956b64 109May be used to override the job name. On the command line, this parameter
d60e92d1
AC
110has the special purpose of signalling the start of a new job.
111.TP
112.BI description \fR=\fPstr
113Human-readable description of the job. It is printed when the job is run, but
114otherwise has no special purpose.
115.TP
116.BI directory \fR=\fPstr
117Prefix filenames with this directory. Used to place files in a location other
118than `./'.
119.TP
120.BI filename \fR=\fPstr
121.B fio
122normally makes up a file name based on the job name, thread number, and file
d1429b5c
AC
123number. If you want to share files between threads in a job or several jobs,
124specify a \fIfilename\fR for each of them to override the default. If the I/O
125engine used is `net', \fIfilename\fR is the host and port to connect to in the
126format \fIhost\fR/\fIport\fR. If the I/O engine is file-based, you can specify
127a number of files by separating the names with a `:' character. `\-' is a
128reserved name, meaning stdin or stdout, depending on the read/write direction
129set.
d60e92d1 130.TP
3ce9dcaf
JA
131.BI lockfile \fR=\fPstr
132Fio defaults to not locking any files before it does IO to them. If a file or
133file descriptor is shared, fio can serialize IO to that file to make the end
134result consistent. This is usual for emulating real workloads that share files.
135The lock modes are:
136.RS
137.RS
138.TP
139.B none
140No locking. This is the default.
141.TP
142.B exclusive
143Only one thread or process may do IO at the time, excluding all others.
144.TP
145.B readwrite
146Read-write locking on the file. Many readers may access the file at the same
147time, but writes get exclusive access.
148.RE
149.P
150The option may be post-fixed with a lock batch number. If set, then each
151thread/process may do that amount of IOs to the file before giving up the lock.
152Since lock acquisition is expensive, batching the lock/unlocks will speed up IO.
153.RE
154.P
d60e92d1
AC
155.BI opendir \fR=\fPstr
156Recursively open any files below directory \fIstr\fR.
157.TP
158.BI readwrite \fR=\fPstr "\fR,\fP rw" \fR=\fPstr
159Type of I/O pattern. Accepted values are:
160.RS
161.RS
162.TP
163.B read
d1429b5c 164Sequential reads.
d60e92d1
AC
165.TP
166.B write
d1429b5c 167Sequential writes.
d60e92d1
AC
168.TP
169.B randread
d1429b5c 170Random reads.
d60e92d1
AC
171.TP
172.B randwrite
d1429b5c 173Random writes.
d60e92d1
AC
174.TP
175.B rw
d1429b5c 176Mixed sequential reads and writes.
d60e92d1
AC
177.TP
178.B randrw
d1429b5c 179Mixed random reads and writes.
d60e92d1
AC
180.RE
181.P
38dad62d
JA
182For mixed I/O, the default split is 50/50. For certain types of io the result
183may still be skewed a bit, since the speed may be different. It is possible to
184specify a number of IO's to do before getting a new offset, this is one by
185appending a `:\fI<nr>\fR to the end of the string given. For a random read, it
186would look like \fBrw=randread:8\fR for passing in an offset modifier with a
059b0802
JA
187value of 8. If the postfix is used with a sequential IO pattern, then the value
188specified will be added to the generated offset for each IO. For instance,
189using \fBrw=write:4k\fR will skip 4k for every write. It turns sequential IO
190into sequential IO with holes. See the \fBrw_sequencer\fR option.
d60e92d1
AC
191.RE
192.TP
38dad62d
JA
193.BI rw_sequencer \fR=\fPstr
194If an offset modifier is given by appending a number to the \fBrw=<str>\fR line,
195then this option controls how that number modifies the IO offset being
196generated. Accepted values are:
197.RS
198.RS
199.TP
200.B sequential
201Generate sequential offset
202.TP
203.B identical
204Generate the same offset
205.RE
206.P
207\fBsequential\fR is only useful for random IO, where fio would normally
208generate a new random offset for every IO. If you append eg 8 to randread, you
209would get a new random offset for every 8 IO's. The result would be a seek for
210only every 8 IO's, instead of for every IO. Use \fBrw=randread:8\fR to specify
211that. As sequential IO is already sequential, setting \fBsequential\fR for that
212would not result in any differences. \fBidentical\fR behaves in a similar
213fashion, except it sends the same offset 8 number of times before generating a
214new offset.
215.RE
216.P
217.TP
90fef2d1
JA
218.BI kb_base \fR=\fPint
219The base unit for a kilobyte. The defacto base is 2^10, 1024. Storage
220manufacturers like to use 10^3 or 1000 as a base ten unit instead, for obvious
221reasons. Allow values are 1024 or 1000, with 1024 being the default.
222.TP
d60e92d1
AC
223.BI randrepeat \fR=\fPbool
224Seed the random number generator in a predictable way so results are repeatable
d1429b5c 225across runs. Default: true.
d60e92d1 226.TP
2615cc4b
JA
227.BI use_os_rand \fR=\fPbool
228Fio can either use the random generator supplied by the OS to generator random
229offsets, or it can use it's own internal generator (based on Tausworthe).
230Default is to use the internal generator, which is often of better quality and
231faster. Default: false.
232.TP
a596f047
EG
233.BI fallocate \fR=\fPstr
234Whether pre-allocation is performed when laying down files. Accepted values
235are:
236.RS
237.RS
238.TP
239.B none
240Do not pre-allocate space.
241.TP
242.B posix
243Pre-allocate via posix_fallocate().
244.TP
245.B keep
246Pre-allocate via fallocate() with FALLOC_FL_KEEP_SIZE set.
247.TP
248.B 0
249Backward-compatible alias for 'none'.
250.TP
251.B 1
252Backward-compatible alias for 'posix'.
253.RE
254.P
255May not be available on all supported platforms. 'keep' is only
256available on Linux. If using ZFS on Solaris this must be set to 'none'
257because ZFS doesn't support it. Default: 'posix'.
258.RE
7bc8c2cf 259.TP
d60e92d1 260.BI fadvise_hint \fR=\fPbool
d1429b5c
AC
261Disable use of \fIposix_fadvise\fR\|(2) to advise the kernel what I/O patterns
262are likely to be issued. Default: true.
d60e92d1 263.TP
f7fa2653 264.BI size \fR=\fPint
d60e92d1
AC
265Total size of I/O for this job. \fBfio\fR will run until this many bytes have
266been transfered, unless limited by other options (\fBruntime\fR, for instance).
d7c8be03 267Unless \fBnrfiles\fR and \fBfilesize\fR options are given, this amount will be
d6667268
JA
268divided between the available files for the job. If not set, fio will use the
269full size of the given files or devices. If the the files do not exist, size
7bb59102
JA
270must be given. It is also possible to give size as a percentage between 1 and
271100. If size=20% is given, fio will use 20% of the full size of the given files
272or devices.
d60e92d1 273.TP
74586c1e 274.BI fill_device \fR=\fPbool "\fR,\fB fill_fs" \fR=\fPbool
3ce9dcaf
JA
275Sets size to something really large and waits for ENOSPC (no space left on
276device) as the terminating condition. Only makes sense with sequential write.
277For a read workload, the mount point will be filled first then IO started on
4f12432e
JA
278the result. This option doesn't make sense if operating on a raw device node,
279since the size of that is already known by the file system. Additionally,
280writing beyond end-of-device will not return ENOSPC there.
3ce9dcaf 281.TP
d60e92d1
AC
282.BI filesize \fR=\fPirange
283Individual file sizes. May be a range, in which case \fBfio\fR will select sizes
d1429b5c
AC
284for files at random within the given range, limited to \fBsize\fR in total (if
285that is given). If \fBfilesize\fR is not specified, each created file is the
286same size.
d60e92d1 287.TP
f7fa2653 288.BI blocksize \fR=\fPint[,int] "\fR,\fB bs" \fR=\fPint[,int]
d60e92d1 289Block size for I/O units. Default: 4k. Values for reads and writes can be
656ebab7 290specified separately in the format \fIread\fR,\fIwrite\fR, either of
d60e92d1
AC
291which may be empty to leave that value at its default.
292.TP
9183788d 293.BI blocksize_range \fR=\fPirange[,irange] "\fR,\fB bsrange" \fR=\fPirange[,irange]
d1429b5c
AC
294Specify a range of I/O block sizes. The issued I/O unit will always be a
295multiple of the minimum size, unless \fBblocksize_unaligned\fR is set. Applies
9183788d 296to both reads and writes if only one range is given, but can be specified
656ebab7 297separately with a comma seperating the values. Example: bsrange=1k-4k,2k-8k.
9183788d
JA
298Also (see \fBblocksize\fR).
299.TP
300.BI bssplit \fR=\fPstr
301This option allows even finer grained control of the block sizes issued,
302not just even splits between them. With this option, you can weight various
303block sizes for exact control of the issued IO for a job that has mixed
304block sizes. The format of the option is bssplit=blocksize/percentage,
5982a925 305optionally adding as many definitions as needed separated by a colon.
9183788d 306Example: bssplit=4k/10:64k/50:32k/40 would issue 50% 64k blocks, 10% 4k
c83cdd3e
JA
307blocks and 40% 32k blocks. \fBbssplit\fR also supports giving separate
308splits to reads and writes. The format is identical to what the
309\fBbs\fR option accepts, the read and write parts are separated with a
310comma.
d60e92d1
AC
311.TP
312.B blocksize_unaligned\fR,\fP bs_unaligned
d1429b5c
AC
313If set, any size in \fBblocksize_range\fR may be used. This typically won't
314work with direct I/O, as that normally requires sector alignment.
d60e92d1 315.TP
2b7a01d0 316.BI blockalign \fR=\fPint[,int] "\fR,\fB ba" \fR=\fPint[,int]
639ce0f3
MS
317At what boundary to align random IO offsets. Defaults to the same as 'blocksize'
318the minimum blocksize given. Minimum alignment is typically 512b
2b7a01d0
JA
319for using direct IO, though it usually depends on the hardware block size.
320This option is mutually exclusive with using a random map for files, so it
321will turn off that option.
43602667 322.TP
d60e92d1
AC
323.B zero_buffers
324Initialise buffers with all zeros. Default: fill buffers with random data.
325.TP
901bb994
JA
326.B refill_buffers
327If this option is given, fio will refill the IO buffers on every submit. The
328default is to only fill it at init time and reuse that data. Only makes sense
329if zero_buffers isn't specified, naturally. If data verification is enabled,
330refill_buffers is also automatically enabled.
331.TP
fd68418e
JA
332.BI scramble_buffers \fR=\fPbool
333If \fBrefill_buffers\fR is too costly and the target is using data
334deduplication, then setting this option will slightly modify the IO buffer
335contents to defeat normal de-dupe attempts. This is not enough to defeat
336more clever block compression attempts, but it will stop naive dedupe
337of blocks. Default: true.
338.TP
d60e92d1
AC
339.BI nrfiles \fR=\fPint
340Number of files to use for this job. Default: 1.
341.TP
342.BI openfiles \fR=\fPint
343Number of files to keep open at the same time. Default: \fBnrfiles\fR.
344.TP
345.BI file_service_type \fR=\fPstr
346Defines how files to service are selected. The following types are defined:
347.RS
348.RS
349.TP
350.B random
351Choose a file at random
352.TP
353.B roundrobin
354Round robin over open files (default).
6b7f6851
JA
355.B sequential
356Do each file in the set sequentially.
d60e92d1
AC
357.RE
358.P
359The number of I/Os to issue before switching a new file can be specified by
360appending `:\fIint\fR' to the service type.
361.RE
362.TP
363.BI ioengine \fR=\fPstr
364Defines how the job issues I/O. The following types are defined:
365.RS
366.RS
367.TP
368.B sync
369Basic \fIread\fR\|(2) or \fIwrite\fR\|(2) I/O. \fIfseek\fR\|(2) is used to
370position the I/O location.
371.TP
a31041ea 372.B psync
373Basic \fIpread\fR\|(2) or \fIpwrite\fR\|(2) I/O.
374.TP
9183788d
JA
375.B vsync
376Basic \fIreadv\fR\|(2) or \fIwritev\fR\|(2) I/O. Will emulate queuing by
377coalescing adjacents IOs into a single submission.
378.TP
d60e92d1 379.B libaio
c44b1ff5
JA
380Linux native asynchronous I/O. This engine also has a sub-option,
381\fBuserspace_reap\fR. To set it, use \fBioengine=libaio:userspace_reap\fR.
382Normally, with the libaio engine in use, fio will use the
383\fIio_getevents\fR\|(3) system call to reap newly returned events. With this
384flag turned on, the AIO ring will be read directly from user-space to reap
385events. The reaping mode is only enabled when polling for a minimum of \fB0\fR
386events (eg when \fBiodepth_batch_complete=0\fR).
d60e92d1
AC
387.TP
388.B posixaio
03e20d68
BC
389POSIX asynchronous I/O using \fIaio_read\fR\|(3) and \fIaio_write\fR\|(3).
390.TP
391.B solarisaio
392Solaris native asynchronous I/O.
393.TP
394.B windowsaio
395Windows native asynchronous I/O.
d60e92d1
AC
396.TP
397.B mmap
d1429b5c
AC
398File is memory mapped with \fImmap\fR\|(2) and data copied using
399\fImemcpy\fR\|(3).
d60e92d1
AC
400.TP
401.B splice
d1429b5c
AC
402\fIsplice\fR\|(2) is used to transfer the data and \fIvmsplice\fR\|(2) to
403transfer data from user-space to the kernel.
d60e92d1
AC
404.TP
405.B syslet-rw
406Use the syslet system calls to make regular read/write asynchronous.
407.TP
408.B sg
409SCSI generic sg v3 I/O. May be either synchronous using the SG_IO ioctl, or if
d1429b5c
AC
410the target is an sg character device, we use \fIread\fR\|(2) and
411\fIwrite\fR\|(2) for asynchronous I/O.
d60e92d1
AC
412.TP
413.B null
414Doesn't transfer any data, just pretends to. Mainly used to exercise \fBfio\fR
415itself and for debugging and testing purposes.
416.TP
417.B net
418Transfer over the network. \fBfilename\fR must be set appropriately to
0fd666bf
JA
419`\fIhost\fR,\fIport\fR,\fItype\fR' regardless of data direction. \fItype\fR
420is one of \fBtcp\fR, \fBudp\fR, or \fBunix\fR. For UNIX domain sockets,
421the \fIhost\fR parameter is a file system path.
d60e92d1
AC
422.TP
423.B netsplice
424Like \fBnet\fR, but uses \fIsplice\fR\|(2) and \fIvmsplice\fR\|(2) to map data
425and send/receive.
426.TP
53aec0a4 427.B cpuio
d60e92d1
AC
428Doesn't transfer any data, but burns CPU cycles according to \fBcpuload\fR and
429\fBcpucycles\fR parameters.
430.TP
431.B guasi
432The GUASI I/O engine is the Generic Userspace Asynchronous Syscall Interface
433approach to asycnronous I/O.
d1429b5c
AC
434.br
435See <http://www.xmailserver.org/guasi\-lib.html>.
d60e92d1 436.TP
21b8aee8 437.B rdma
85286c5c
BVA
438The RDMA I/O engine supports both RDMA memory semantics (RDMA_WRITE/RDMA_READ)
439and channel semantics (Send/Recv) for the InfiniBand, RoCE and iWARP protocols.
21b8aee8 440.TP
d60e92d1
AC
441.B external
442Loads an external I/O engine object file. Append the engine filename as
443`:\fIenginepath\fR'.
444.RE
445.RE
446.TP
447.BI iodepth \fR=\fPint
8489dae4
SK
448Number of I/O units to keep in flight against the file. Note that increasing
449iodepth beyond 1 will not affect synchronous ioengines (except for small
ee72ca09
JA
450degress when verify_async is in use). Even async engines my impose OS
451restrictions causing the desired depth not to be achieved. This may happen on
452Linux when using libaio and not setting \fBdirect\fR=1, since buffered IO is
453not async on that OS. Keep an eye on the IO depth distribution in the
454fio output to verify that the achieved depth is as expected. Default: 1.
d60e92d1
AC
455.TP
456.BI iodepth_batch \fR=\fPint
457Number of I/Os to submit at once. Default: \fBiodepth\fR.
458.TP
3ce9dcaf
JA
459.BI iodepth_batch_complete \fR=\fPint
460This defines how many pieces of IO to retrieve at once. It defaults to 1 which
461 means that we'll ask for a minimum of 1 IO in the retrieval process from the
462kernel. The IO retrieval will go on until we hit the limit set by
463\fBiodepth_low\fR. If this variable is set to 0, then fio will always check for
464completed events before queuing more IO. This helps reduce IO latency, at the
465cost of more retrieval system calls.
466.TP
d60e92d1
AC
467.BI iodepth_low \fR=\fPint
468Low watermark indicating when to start filling the queue again. Default:
469\fBiodepth\fR.
470.TP
471.BI direct \fR=\fPbool
472If true, use non-buffered I/O (usually O_DIRECT). Default: false.
473.TP
474.BI buffered \fR=\fPbool
475If true, use buffered I/O. This is the opposite of the \fBdirect\fR parameter.
476Default: true.
477.TP
f7fa2653 478.BI offset \fR=\fPint
d60e92d1
AC
479Offset in the file to start I/O. Data before the offset will not be touched.
480.TP
481.BI fsync \fR=\fPint
d1429b5c
AC
482How many I/Os to perform before issuing an \fBfsync\fR\|(2) of dirty data. If
4830, don't sync. Default: 0.
d60e92d1 484.TP
5f9099ea
JA
485.BI fdatasync \fR=\fPint
486Like \fBfsync\fR, but uses \fBfdatasync\fR\|(2) instead to only sync the
487data parts of the file. Default: 0.
488.TP
e76b1da4
JA
489.BI sync_file_range \fR=\fPstr:int
490Use sync_file_range() for every \fRval\fP number of write operations. Fio will
491track range of writes that have happened since the last sync_file_range() call.
492\fRstr\fP can currently be one or more of:
493.RS
494.TP
495.B wait_before
496SYNC_FILE_RANGE_WAIT_BEFORE
497.TP
498.B write
499SYNC_FILE_RANGE_WRITE
500.TP
501.B wait_after
502SYNC_FILE_RANGE_WRITE
503.TP
504.RE
505.P
506So if you do sync_file_range=wait_before,write:8, fio would use
507\fBSYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE\fP for every 8 writes.
508Also see the sync_file_range(2) man page. This option is Linux specific.
509.TP
d60e92d1 510.BI overwrite \fR=\fPbool
d1429b5c 511If writing, setup the file first and do overwrites. Default: false.
d60e92d1
AC
512.TP
513.BI end_fsync \fR=\fPbool
d1429b5c 514Sync file contents when job exits. Default: false.
d60e92d1
AC
515.TP
516.BI fsync_on_close \fR=\fPbool
517If true, sync file contents on close. This differs from \fBend_fsync\fR in that
d1429b5c 518it will happen on every close, not just at the end of the job. Default: false.
d60e92d1
AC
519.TP
520.BI rwmixcycle \fR=\fPint
521How many milliseconds before switching between reads and writes for a mixed
522workload. Default: 500ms.
523.TP
524.BI rwmixread \fR=\fPint
525Percentage of a mixed workload that should be reads. Default: 50.
526.TP
527.BI rwmixwrite \fR=\fPint
d1429b5c 528Percentage of a mixed workload that should be writes. If \fBrwmixread\fR and
c35dd7a6
JA
529\fBrwmixwrite\fR are given and do not sum to 100%, the latter of the two
530overrides the first. This may interfere with a given rate setting, if fio is
531asked to limit reads or writes to a certain rate. If that is the case, then
532the distribution may be skewed. Default: 50.
d60e92d1
AC
533.TP
534.B norandommap
535Normally \fBfio\fR will cover every block of the file when doing random I/O. If
536this parameter is given, a new offset will be chosen without looking at past
537I/O history. This parameter is mutually exclusive with \fBverify\fR.
538.TP
744492c9 539.BI softrandommap \fR=\fPbool
3ce9dcaf
JA
540See \fBnorandommap\fR. If fio runs with the random block map enabled and it
541fails to allocate the map, if this option is set it will continue without a
542random block map. As coverage will not be as complete as with random maps, this
543option is disabled by default.
544.TP
d60e92d1
AC
545.BI nice \fR=\fPint
546Run job with given nice value. See \fInice\fR\|(2).
547.TP
548.BI prio \fR=\fPint
549Set I/O priority value of this job between 0 (highest) and 7 (lowest). See
550\fIionice\fR\|(1).
551.TP
552.BI prioclass \fR=\fPint
553Set I/O priority class. See \fIionice\fR\|(1).
554.TP
555.BI thinktime \fR=\fPint
556Stall job for given number of microseconds between issuing I/Os.
557.TP
558.BI thinktime_spin \fR=\fPint
559Pretend to spend CPU time for given number of microseconds, sleeping the rest
560of the time specified by \fBthinktime\fR. Only valid if \fBthinktime\fR is set.
561.TP
562.BI thinktime_blocks \fR=\fPint
563Number of blocks to issue before waiting \fBthinktime\fR microseconds.
564Default: 1.
565.TP
566.BI rate \fR=\fPint
c35dd7a6
JA
567Cap bandwidth used by this job. The number is in bytes/sec, the normal postfix
568rules apply. You can use \fBrate\fR=500k to limit reads and writes to 500k each,
569or you can specify read and writes separately. Using \fBrate\fR=1m,500k would
570limit reads to 1MB/sec and writes to 500KB/sec. Capping only reads or writes
571can be done with \fBrate\fR=,500k or \fBrate\fR=500k,. The former will only
572limit writes (to 500KB/sec), the latter will only limit reads.
d60e92d1
AC
573.TP
574.BI ratemin \fR=\fPint
575Tell \fBfio\fR to do whatever it can to maintain at least the given bandwidth.
c35dd7a6
JA
576Failing to meet this requirement will cause the job to exit. The same format
577as \fBrate\fR is used for read vs write separation.
d60e92d1
AC
578.TP
579.BI rate_iops \fR=\fPint
c35dd7a6
JA
580Cap the bandwidth to this number of IOPS. Basically the same as rate, just
581specified independently of bandwidth. The same format as \fBrate\fR is used for
582read vs write seperation. If \fBblocksize\fR is a range, the smallest block
583size is used as the metric.
d60e92d1
AC
584.TP
585.BI rate_iops_min \fR=\fPint
c35dd7a6
JA
586If this rate of I/O is not met, the job will exit. The same format as \fBrate\fR
587is used for read vs write seperation.
d60e92d1
AC
588.TP
589.BI ratecycle \fR=\fPint
590Average bandwidth for \fBrate\fR and \fBratemin\fR over this number of
591milliseconds. Default: 1000ms.
592.TP
593.BI cpumask \fR=\fPint
594Set CPU affinity for this job. \fIint\fR is a bitmask of allowed CPUs the job
595may run on. See \fBsched_setaffinity\fR\|(2).
596.TP
597.BI cpus_allowed \fR=\fPstr
598Same as \fBcpumask\fR, but allows a comma-delimited list of CPU numbers.
599.TP
600.BI startdelay \fR=\fPint
601Delay start of job for the specified number of seconds.
602.TP
603.BI runtime \fR=\fPint
604Terminate processing after the specified number of seconds.
605.TP
606.B time_based
607If given, run for the specified \fBruntime\fR duration even if the files are
608completely read or written. The same workload will be repeated as many times
609as \fBruntime\fR allows.
610.TP
901bb994
JA
611.BI ramp_time \fR=\fPint
612If set, fio will run the specified workload for this amount of time before
613logging any performance numbers. Useful for letting performance settle before
614logging results, thus minimizing the runtime required for stable results. Note
c35dd7a6
JA
615that the \fBramp_time\fR is considered lead in time for a job, thus it will
616increase the total runtime if a special timeout or runtime is specified.
901bb994 617.TP
d60e92d1
AC
618.BI invalidate \fR=\fPbool
619Invalidate buffer-cache for the file prior to starting I/O. Default: true.
620.TP
621.BI sync \fR=\fPbool
622Use synchronous I/O for buffered writes. For the majority of I/O engines,
d1429b5c 623this means using O_SYNC. Default: false.
d60e92d1
AC
624.TP
625.BI iomem \fR=\fPstr "\fR,\fP mem" \fR=\fPstr
626Allocation method for I/O unit buffer. Allowed values are:
627.RS
628.RS
629.TP
630.B malloc
631Allocate memory with \fImalloc\fR\|(3).
632.TP
633.B shm
634Use shared memory buffers allocated through \fIshmget\fR\|(2).
635.TP
636.B shmhuge
637Same as \fBshm\fR, but use huge pages as backing.
638.TP
639.B mmap
640Use \fImmap\fR\|(2) for allocation. Uses anonymous memory unless a filename
641is given after the option in the format `:\fIfile\fR'.
642.TP
643.B mmaphuge
644Same as \fBmmap\fR, but use huge files as backing.
645.RE
646.P
647The amount of memory allocated is the maximum allowed \fBblocksize\fR for the
648job multiplied by \fBiodepth\fR. For \fBshmhuge\fR or \fBmmaphuge\fR to work,
649the system must have free huge pages allocated. \fBmmaphuge\fR also needs to
2e266ba6
JA
650have hugetlbfs mounted, and \fIfile\fR must point there. At least on Linux,
651huge pages must be manually allocated. See \fB/proc/sys/vm/nr_hugehages\fR
652and the documentation for that. Normally you just need to echo an appropriate
653number, eg echoing 8 will ensure that the OS has 8 huge pages ready for
654use.
d60e92d1
AC
655.RE
656.TP
d392365e 657.BI iomem_align \fR=\fPint "\fR,\fP mem_align" \fR=\fPint
d529ee19
JA
658This indiciates the memory alignment of the IO memory buffers. Note that the
659given alignment is applied to the first IO unit buffer, if using \fBiodepth\fR
660the alignment of the following buffers are given by the \fBbs\fR used. In
661other words, if using a \fBbs\fR that is a multiple of the page sized in the
662system, all buffers will be aligned to this value. If using a \fBbs\fR that
663is not page aligned, the alignment of subsequent IO memory buffers is the
664sum of the \fBiomem_align\fR and \fBbs\fR used.
665.TP
f7fa2653 666.BI hugepage\-size \fR=\fPint
d60e92d1 667Defines the size of a huge page. Must be at least equal to the system setting.
b22989b9 668Should be a multiple of 1MB. Default: 4MB.
d60e92d1
AC
669.TP
670.B exitall
671Terminate all jobs when one finishes. Default: wait for each job to finish.
672.TP
673.BI bwavgtime \fR=\fPint
674Average bandwidth calculations over the given time in milliseconds. Default:
675500ms.
676.TP
c8eeb9df
JA
677.BI iopsavgtime \fR=\fPint
678Average IOPS calculations over the given time in milliseconds. Default:
679500ms.
680.TP
d60e92d1 681.BI create_serialize \fR=\fPbool
d1429b5c 682If true, serialize file creation for the jobs. Default: true.
d60e92d1
AC
683.TP
684.BI create_fsync \fR=\fPbool
685\fIfsync\fR\|(2) data file after creation. Default: true.
686.TP
6b7f6851
JA
687.BI create_on_open \fR=\fPbool
688If true, the files are not created until they are opened for IO by the job.
689.TP
e9f48479
JA
690.BI pre_read \fR=\fPbool
691If this is given, files will be pre-read into memory before starting the given
692IO operation. This will also clear the \fR \fBinvalidate\fR flag, since it is
9c0d2241
JA
693pointless to pre-read and then drop the cache. This will only work for IO
694engines that are seekable, since they allow you to read the same data
695multiple times. Thus it will not work on eg network or splice IO.
e9f48479 696.TP
d60e92d1
AC
697.BI unlink \fR=\fPbool
698Unlink job files when done. Default: false.
699.TP
700.BI loops \fR=\fPint
701Specifies the number of iterations (runs of the same workload) of this job.
702Default: 1.
703.TP
704.BI do_verify \fR=\fPbool
705Run the verify phase after a write phase. Only valid if \fBverify\fR is set.
706Default: true.
707.TP
708.BI verify \fR=\fPstr
709Method of verifying file contents after each iteration of the job. Allowed
710values are:
711.RS
712.RS
713.TP
b892dc08 714.B md5 crc16 crc32 crc32c crc32c-intel crc64 crc7 sha256 sha512 sha1
0539d758
JA
715Store appropriate checksum in the header of each block. crc32c-intel is
716hardware accelerated SSE4.2 driven, falls back to regular crc32c if
717not supported by the system.
d60e92d1
AC
718.TP
719.B meta
720Write extra information about each I/O (timestamp, block number, etc.). The
996093bb 721block number is verified. See \fBverify_pattern\fR as well.
d60e92d1
AC
722.TP
723.B null
724Pretend to verify. Used for testing internals.
725.RE
b892dc08
JA
726
727This option can be used for repeated burn-in tests of a system to make sure
728that the written data is also correctly read back. If the data direction given
729is a read or random read, fio will assume that it should verify a previously
730written file. If the data direction includes any form of write, the verify will
731be of the newly written data.
d60e92d1
AC
732.RE
733.TP
734.BI verify_sort \fR=\fPbool
735If true, written verify blocks are sorted if \fBfio\fR deems it to be faster to
736read them back in a sorted manner. Default: true.
737.TP
f7fa2653 738.BI verify_offset \fR=\fPint
d60e92d1 739Swap the verification header with data somewhere else in the block before
d1429b5c 740writing. It is swapped back before verifying.
d60e92d1 741.TP
f7fa2653 742.BI verify_interval \fR=\fPint
d60e92d1
AC
743Write the verification header for this number of bytes, which should divide
744\fBblocksize\fR. Default: \fBblocksize\fR.
745.TP
996093bb
JA
746.BI verify_pattern \fR=\fPstr
747If set, fio will fill the io buffers with this pattern. Fio defaults to filling
748with totally random bytes, but sometimes it's interesting to fill with a known
749pattern for io verification purposes. Depending on the width of the pattern,
750fio will fill 1/2/3/4 bytes of the buffer at the time(it can be either a
751decimal or a hex number). The verify_pattern if larger than a 32-bit quantity
752has to be a hex number that starts with either "0x" or "0X". Use with
753\fBverify\fP=meta.
754.TP
d60e92d1
AC
755.BI verify_fatal \fR=\fPbool
756If true, exit the job on the first observed verification failure. Default:
757false.
758.TP
b463e936
JA
759.BI verify_dump \fR=\fPbool
760If set, dump the contents of both the original data block and the data block we
761read off disk to files. This allows later analysis to inspect just what kind of
762data corruption occurred. On by default.
763.TP
e8462bd8
JA
764.BI verify_async \fR=\fPint
765Fio will normally verify IO inline from the submitting thread. This option
766takes an integer describing how many async offload threads to create for IO
767verification instead, causing fio to offload the duty of verifying IO contents
c85c324c
JA
768to one or more separate threads. If using this offload option, even sync IO
769engines can benefit from using an \fBiodepth\fR setting higher than 1, as it
770allows them to have IO in flight while verifies are running.
e8462bd8
JA
771.TP
772.BI verify_async_cpus \fR=\fPstr
773Tell fio to set the given CPU affinity on the async IO verification threads.
774See \fBcpus_allowed\fP for the format used.
775.TP
6f87418f
JA
776.BI verify_backlog \fR=\fPint
777Fio will normally verify the written contents of a job that utilizes verify
778once that job has completed. In other words, everything is written then
779everything is read back and verified. You may want to verify continually
780instead for a variety of reasons. Fio stores the meta data associated with an
781IO block in memory, so for large verify workloads, quite a bit of memory would
092f707f
DN
782be used up holding this meta data. If this option is enabled, fio will write
783only N blocks before verifying these blocks.
6f87418f
JA
784.TP
785.BI verify_backlog_batch \fR=\fPint
786Control how many blocks fio will verify if verify_backlog is set. If not set,
787will default to the value of \fBverify_backlog\fR (meaning the entire queue is
092f707f
DN
788read back and verified). If \fBverify_backlog_batch\fR is less than
789\fBverify_backlog\fR then not all blocks will be verified, if
790\fBverify_backlog_batch\fR is larger than \fBverify_backlog\fR, some blocks
791will be verified more than once.
6f87418f 792.TP
d392365e 793.B stonewall "\fR,\fP wait_for_previous"
5982a925 794Wait for preceding jobs in the job file to exit before starting this one.
d60e92d1
AC
795\fBstonewall\fR implies \fBnew_group\fR.
796.TP
797.B new_group
798Start a new reporting group. If not given, all jobs in a file will be part
799of the same reporting group, unless separated by a stonewall.
800.TP
801.BI numjobs \fR=\fPint
802Number of clones (processes/threads performing the same workload) of this job.
803Default: 1.
804.TP
805.B group_reporting
806If set, display per-group reports instead of per-job when \fBnumjobs\fR is
807specified.
808.TP
809.B thread
810Use threads created with \fBpthread_create\fR\|(3) instead of processes created
811with \fBfork\fR\|(2).
812.TP
f7fa2653 813.BI zonesize \fR=\fPint
d60e92d1
AC
814Divide file into zones of the specified size in bytes. See \fBzoneskip\fR.
815.TP
f7fa2653 816.BI zoneskip \fR=\fPint
d1429b5c 817Skip the specified number of bytes when \fBzonesize\fR bytes of data have been
d60e92d1
AC
818read.
819.TP
820.BI write_iolog \fR=\fPstr
5b42a488
SH
821Write the issued I/O patterns to the specified file. Specify a separate file
822for each job, otherwise the iologs will be interspersed and the file may be
823corrupt.
d60e92d1
AC
824.TP
825.BI read_iolog \fR=\fPstr
826Replay the I/O patterns contained in the specified file generated by
827\fBwrite_iolog\fR, or may be a \fBblktrace\fR binary file.
828.TP
64bbb865
DN
829.BI replay_no_stall \fR=\fPint
830While replaying I/O patterns using \fBread_iolog\fR the default behavior
831attempts to respect timing information between I/Os. Enabling
832\fBreplay_no_stall\fR causes I/Os to be replayed as fast as possible while
833still respecting ordering.
834.TP
d1c46c04
DN
835.BI replay_redirect \fR=\fPstr
836While replaying I/O patterns using \fBread_iolog\fR the default behavior
837is to replay the IOPS onto the major/minor device that each IOP was recorded
838from. Setting \fBreplay_redirect\fR causes all IOPS to be replayed onto the
839single specified device regardless of the device it was recorded from.
840.TP
836bad52 841.BI write_bw_log \fR=\fPstr
901bb994
JA
842If given, write a bandwidth log of the jobs in this job file. Can be used to
843store data of the bandwidth of the jobs in their lifetime. The included
844fio_generate_plots script uses gnuplot to turn these text files into nice
845graphs. See \fBwrite_log_log\fR for behaviour of given filename. For this
846option, the postfix is _bw.log.
d60e92d1 847.TP
836bad52 848.BI write_lat_log \fR=\fPstr
901bb994
JA
849Same as \fBwrite_bw_log\fR, but writes I/O completion latencies. If no
850filename is given with this option, the default filename of "jobname_type.log"
851is used. Even if the filename is given, fio will still append the type of log.
852.TP
c8eeb9df
JA
853.BI write_iops_log \fR=\fPstr
854Same as \fBwrite_bw_log\fR, but writes IOPS. If no filename is given with this
855option, the default filename of "jobname_type.log" is used. Even if the
856filename is given, fio will still append the type of log.
857.TP
836bad52 858.BI disable_lat \fR=\fPbool
02af0988 859Disable measurements of total latency numbers. Useful only for cutting
901bb994
JA
860back the number of calls to gettimeofday, as that does impact performance at
861really high IOPS rates. Note that to really get rid of a large amount of these
862calls, this option must be used with disable_slat and disable_bw as well.
863.TP
836bad52 864.BI disable_clat \fR=\fPbool
c95f9daf 865Disable measurements of completion latency numbers. See \fBdisable_lat\fR.
02af0988 866.TP
836bad52 867.BI disable_slat \fR=\fPbool
02af0988 868Disable measurements of submission latency numbers. See \fBdisable_lat\fR.
901bb994 869.TP
836bad52 870.BI disable_bw_measurement \fR=\fPbool
02af0988 871Disable measurements of throughput/bandwidth numbers. See \fBdisable_lat\fR.
d60e92d1 872.TP
f7fa2653 873.BI lockmem \fR=\fPint
d60e92d1
AC
874Pin the specified amount of memory with \fBmlock\fR\|(2). Can be used to
875simulate a smaller amount of memory.
876.TP
877.BI exec_prerun \fR=\fPstr
878Before running the job, execute the specified command with \fBsystem\fR\|(3).
879.TP
880.BI exec_postrun \fR=\fPstr
881Same as \fBexec_prerun\fR, but the command is executed after the job completes.
882.TP
883.BI ioscheduler \fR=\fPstr
884Attempt to switch the device hosting the file to the specified I/O scheduler.
885.TP
886.BI cpuload \fR=\fPint
887If the job is a CPU cycle-eater, attempt to use the specified percentage of
888CPU cycles.
889.TP
890.BI cpuchunks \fR=\fPint
891If the job is a CPU cycle-eater, split the load into cycles of the
892given time in milliseconds.
893.TP
894.BI disk_util \fR=\fPbool
d1429b5c 895Generate disk utilization statistics if the platform supports it. Default: true.
901bb994
JA
896.TP
897.BI gtod_reduce \fR=\fPbool
898Enable all of the gettimeofday() reducing options (disable_clat, disable_slat,
899disable_bw) plus reduce precision of the timeout somewhat to really shrink the
900gettimeofday() call count. With this option enabled, we only do about 0.4% of
901the gtod() calls we would have done if all time keeping was enabled.
902.TP
903.BI gtod_cpu \fR=\fPint
904Sometimes it's cheaper to dedicate a single thread of execution to just getting
905the current time. Fio (and databases, for instance) are very intensive on
906gettimeofday() calls. With this option, you can set one CPU aside for doing
907nothing but logging current time to a shared memory location. Then the other
908threads/processes that run IO workloads need only copy that segment, instead of
909entering the kernel with a gettimeofday() call. The CPU set aside for doing
910these time calls will be excluded from other uses. Fio will manually clear it
911from the CPU mask of other jobs.
f2bba182 912.TP
a696fa2a
JA
913.BI cgroup \fR=\fPstr
914Add job to this control group. If it doesn't exist, it will be created.
6adb38a1
JA
915The system must have a mounted cgroup blkio mount point for this to work. If
916your system doesn't have it mounted, you can do so with:
917
5982a925 918# mount \-t cgroup \-o blkio none /cgroup
a696fa2a
JA
919.TP
920.BI cgroup_weight \fR=\fPint
921Set the weight of the cgroup to this value. See the documentation that comes
922with the kernel, allowed values are in the range of 100..1000.
e0b0d892 923.TP
7de87099
VG
924.BI cgroup_nodelete \fR=\fPbool
925Normally fio will delete the cgroups it has created after the job completion.
926To override this behavior and to leave cgroups around after the job completion,
927set cgroup_nodelete=1. This can be useful if one wants to inspect various
928cgroup files after job completion. Default: false
929.TP
e0b0d892
JA
930.BI uid \fR=\fPint
931Instead of running as the invoking user, set the user ID to this value before
932the thread/process does any work.
933.TP
934.BI gid \fR=\fPint
935Set group ID, see \fBuid\fR.
83349190
YH
936.TP
937.BI clat_percentiles \fR=\fPbool
938Enable the reporting of percentiles of completion latencies.
939.TP
940.BI percentile_list \fR=\fPfloat_list
941Overwrite the default list of percentiles for completion
942latencies. Each number is a floating number in the range (0,100], and
943the maximum length of the list is 20. Use ':' to separate the
3eb07285 944numbers. For example, \-\-percentile_list=99.5:99.9 will cause fio to
83349190
YH
945report the values of completion latency below which 99.5% and 99.9% of
946the observed latencies fell, respectively.
d60e92d1 947.SH OUTPUT
d1429b5c
AC
948While running, \fBfio\fR will display the status of the created jobs. For
949example:
d60e92d1 950.RS
d1429b5c 951.P
d60e92d1
AC
952Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s]
953.RE
954.P
d1429b5c
AC
955The characters in the first set of brackets denote the current status of each
956threads. The possible values are:
957.P
958.PD 0
d60e92d1
AC
959.RS
960.TP
961.B P
962Setup but not started.
963.TP
964.B C
965Thread created.
966.TP
967.B I
968Initialized, waiting.
969.TP
970.B R
971Running, doing sequential reads.
972.TP
973.B r
974Running, doing random reads.
975.TP
976.B W
977Running, doing sequential writes.
978.TP
979.B w
980Running, doing random writes.
981.TP
982.B M
983Running, doing mixed sequential reads/writes.
984.TP
985.B m
986Running, doing mixed random reads/writes.
987.TP
988.B F
989Running, currently waiting for \fBfsync\fR\|(2).
990.TP
991.B V
992Running, verifying written data.
993.TP
994.B E
995Exited, not reaped by main thread.
996.TP
997.B \-
998Exited, thread reaped.
999.RE
d1429b5c 1000.PD
d60e92d1
AC
1001.P
1002The second set of brackets shows the estimated completion percentage of
1003the current group. The third set shows the read and write I/O rate,
1004respectively. Finally, the estimated run time of the job is displayed.
1005.P
1006When \fBfio\fR completes (or is interrupted by Ctrl-C), it will show data
1007for each thread, each group of threads, and each disk, in that order.
1008.P
1009Per-thread statistics first show the threads client number, group-id, and
1010error code. The remaining figures are as follows:
1011.RS
d60e92d1
AC
1012.TP
1013.B io
1014Number of megabytes of I/O performed.
1015.TP
1016.B bw
1017Average data rate (bandwidth).
1018.TP
1019.B runt
1020Threads run time.
1021.TP
1022.B slat
1023Submission latency minimum, maximum, average and standard deviation. This is
1024the time it took to submit the I/O.
1025.TP
1026.B clat
1027Completion latency minimum, maximum, average and standard deviation. This
1028is the time between submission and completion.
1029.TP
1030.B bw
1031Bandwidth minimum, maximum, percentage of aggregate bandwidth received, average
1032and standard deviation.
1033.TP
1034.B cpu
1035CPU usage statistics. Includes user and system time, number of context switches
1036this thread went through and number of major and minor page faults.
1037.TP
1038.B IO depths
1039Distribution of I/O depths. Each depth includes everything less than (or equal)
1040to it, but greater than the previous depth.
1041.TP
1042.B IO issued
1043Number of read/write requests issued, and number of short read/write requests.
1044.TP
1045.B IO latencies
1046Distribution of I/O completion latencies. The numbers follow the same pattern
1047as \fBIO depths\fR.
1048.RE
d60e92d1
AC
1049.P
1050The group statistics show:
d1429b5c 1051.PD 0
d60e92d1
AC
1052.RS
1053.TP
1054.B io
1055Number of megabytes I/O performed.
1056.TP
1057.B aggrb
1058Aggregate bandwidth of threads in the group.
1059.TP
1060.B minb
1061Minimum average bandwidth a thread saw.
1062.TP
1063.B maxb
1064Maximum average bandwidth a thread saw.
1065.TP
1066.B mint
d1429b5c 1067Shortest runtime of threads in the group.
d60e92d1
AC
1068.TP
1069.B maxt
1070Longest runtime of threads in the group.
1071.RE
d1429b5c 1072.PD
d60e92d1
AC
1073.P
1074Finally, disk statistics are printed with reads first:
d1429b5c 1075.PD 0
d60e92d1
AC
1076.RS
1077.TP
1078.B ios
1079Number of I/Os performed by all groups.
1080.TP
1081.B merge
1082Number of merges in the I/O scheduler.
1083.TP
1084.B ticks
1085Number of ticks we kept the disk busy.
1086.TP
1087.B io_queue
1088Total time spent in the disk queue.
1089.TP
1090.B util
1091Disk utilization.
1092.RE
d1429b5c 1093.PD
d60e92d1
AC
1094.SH TERSE OUTPUT
1095If the \fB\-\-minimal\fR option is given, the results will be printed in a
562c2d2f
DN
1096semicolon-delimited format suitable for scripted use - a job description
1097(if provided) follows on a new line. Note that the first
525c2bfa
JA
1098number in the line is the version number. If the output has to be changed
1099for some reason, this number will be incremented by 1 to signify that
1100change. The fields are:
d60e92d1
AC
1101.P
1102.RS
525c2bfa 1103.B version, jobname, groupid, error
d60e92d1
AC
1104.P
1105Read status:
1106.RS
85549ba0 1107.B Total I/O \fR(KB)\fP, bandwidth \fR(KB/s)\fP, runtime \fR(ms)\fP
d60e92d1
AC
1108.P
1109Submission latency:
1110.RS
1111.B min, max, mean, standard deviation
1112.RE
1113Completion latency:
1114.RS
1115.B min, max, mean, standard deviation
1116.RE
525c2bfa
JA
1117Total latency:
1118.RS
1119.B min, max, mean, standard deviation
1120.RE
d60e92d1
AC
1121Bandwidth:
1122.RS
1123.B min, max, aggregate percentage of total, mean, standard deviation
1124.RE
1125.RE
1126.P
1127Write status:
1128.RS
85549ba0 1129.B Total I/O \fR(KB)\fP, bandwidth \fR(KB/s)\fP, runtime \fR(ms)\fP
d60e92d1
AC
1130.P
1131Submission latency:
1132.RS
1133.B min, max, mean, standard deviation
1134.RE
1135Completion latency:
1136.RS
1137.B min, max, mean, standard deviation
1138.RE
525c2bfa
JA
1139Total latency:
1140.RS
1141.B min, max, mean, standard deviation
1142.RE
d60e92d1
AC
1143Bandwidth:
1144.RS
1145.B min, max, aggregate percentage of total, mean, standard deviation
1146.RE
1147.RE
1148.P
d1429b5c 1149CPU usage:
d60e92d1 1150.RS
bd2626f0 1151.B user, system, context switches, major page faults, minor page faults
d60e92d1
AC
1152.RE
1153.P
1154IO depth distribution:
1155.RS
1156.B <=1, 2, 4, 8, 16, 32, >=64
1157.RE
1158.P
562c2d2f 1159IO latency distribution:
d60e92d1 1160.RS
562c2d2f
DN
1161Microseconds:
1162.RS
1163.B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000
1164.RE
1165Milliseconds:
1166.RS
1167.B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000
1168.RE
1169.RE
1170.P
5982a925 1171Error Info (dependent on continue_on_error, default off):
562c2d2f
DN
1172.RS
1173.B total # errors, first error code
d60e92d1
AC
1174.RE
1175.P
562c2d2f 1176.B text description (if provided in config - appears on newline)
d60e92d1
AC
1177.RE
1178.SH AUTHORS
1179.B fio
aa58d252
JA
1180was written by Jens Axboe <jens.axboe@oracle.com>,
1181now Jens Axboe <jaxboe@fusionio.com>.
d1429b5c
AC
1182.br
1183This man page was written by Aaron Carroll <aaronc@cse.unsw.edu.au> based
d60e92d1
AC
1184on documentation by Jens Axboe.
1185.SH "REPORTING BUGS"
482900c9 1186Report bugs to the \fBfio\fR mailing list <fio@vger.kernel.org>.
d1429b5c 1187See \fBREADME\fR.
d60e92d1 1188.SH "SEE ALSO"
d1429b5c
AC
1189For further documentation see \fBHOWTO\fR and \fBREADME\fR.
1190.br
1191Sample jobfiles are available in the \fBexamples\fR directory.
d60e92d1 1192