Commit | Line | Data |
---|---|---|
523bad63 | 1 | .TH fio 1 "August 2017" "User Manual" |
d60e92d1 AC |
2 | .SH NAME |
3 | fio \- flexible I/O tester | |
4 | .SH SYNOPSIS | |
5 | .B fio | |
6 | [\fIoptions\fR] [\fIjobfile\fR]... | |
7 | .SH DESCRIPTION | |
8 | .B fio | |
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. | |
13 | .SH OPTIONS | |
14 | .TP | |
49da1240 | 15 | .BI \-\-debug \fR=\fPtype |
7db7a5a0 TK |
16 | Enable verbose tracing \fItype\fR of various fio actions. May be `all' for all \fItype\fRs |
17 | or individual types separated by a comma (e.g. `\-\-debug=file,mem' will enable | |
bdd88be3 TK |
18 | file and memory debugging). `help' will list all available tracing options. |
19 | .TP | |
7db7a5a0 | 20 | .BI \-\-parse\-only |
bdd88be3 | 21 | Parse options only, don't start any I/O. |
49da1240 | 22 | .TP |
b9921d1a DZ |
23 | .BI \-\-merge\-blktrace\-only |
24 | Merge blktraces only, don't start any I/O. | |
25 | .TP | |
d60e92d1 AC |
26 | .BI \-\-output \fR=\fPfilename |
27 | Write output to \fIfilename\fR. | |
28 | .TP | |
7db7a5a0 TK |
29 | .BI \-\-output\-format \fR=\fPformat |
30 | Set the reporting \fIformat\fR to `normal', `terse', `json', or | |
31 | `json+'. Multiple formats can be selected, separate by a comma. `terse' | |
32 | is a CSV based format. `json+' is like `json', except it adds a full | |
513e37ee | 33 | dump of the latency buckets. |
e28ee21d | 34 | .TP |
7db7a5a0 | 35 | .BI \-\-bandwidth\-log |
d23ae827 | 36 | Generate aggregate bandwidth logs. |
d60e92d1 | 37 | .TP |
7db7a5a0 TK |
38 | .BI \-\-minimal |
39 | Print statistics in a terse, semicolon\-delimited format. | |
d60e92d1 | 40 | .TP |
7db7a5a0 TK |
41 | .BI \-\-append\-terse |
42 | Print statistics in selected mode AND terse, semicolon\-delimited format. | |
43 | \fBDeprecated\fR, use \fB\-\-output\-format\fR instead to select multiple formats. | |
f6a7df53 | 44 | .TP |
065248bf | 45 | .BI \-\-terse\-version \fR=\fPversion |
7db7a5a0 | 46 | Set terse \fIversion\fR output format (default `3', or `2', `4', `5'). |
49da1240 | 47 | .TP |
7db7a5a0 | 48 | .BI \-\-version |
bdd88be3 TK |
49 | Print version information and exit. |
50 | .TP | |
7db7a5a0 | 51 | .BI \-\-help |
bdd88be3 | 52 | Print a summary of the command line options and exit. |
49da1240 | 53 | .TP |
7db7a5a0 | 54 | .BI \-\-cpuclock\-test |
bdd88be3 | 55 | Perform test and validation of internal CPU clock. |
fec0f21c | 56 | .TP |
bdd88be3 | 57 | .BI \-\-crctest \fR=\fP[test] |
7db7a5a0 | 58 | Test the speed of the built\-in checksumming functions. If no argument is given, |
bdd88be3 | 59 | all of them are tested. Alternatively, a comma separated list can be passed, in which |
fec0f21c JA |
60 | case the given ones are tested. |
61 | .TP | |
49da1240 | 62 | .BI \-\-cmdhelp \fR=\fPcommand |
bdd88be3 | 63 | Print help information for \fIcommand\fR. May be `all' for all commands. |
49da1240 | 64 | .TP |
7db7a5a0 TK |
65 | .BI \-\-enghelp \fR=\fP[ioengine[,command]] |
66 | List all commands defined by \fIioengine\fR, or print help for \fIcommand\fR | |
67 | defined by \fIioengine\fR. If no \fIioengine\fR is given, list all | |
68 | available ioengines. | |
de890a1e | 69 | .TP |
d60e92d1 | 70 | .BI \-\-showcmd \fR=\fPjobfile |
7db7a5a0 | 71 | Convert \fIjobfile\fR to a set of command\-line options. |
d60e92d1 | 72 | .TP |
bdd88be3 | 73 | .BI \-\-readonly |
4027b2a1 | 74 | Turn on safety read\-only checks, preventing writes and trims. The \fB\-\-readonly\fR |
bdd88be3 | 75 | option is an extra safety guard to prevent users from accidentally starting |
4027b2a1 VF |
76 | a write or trim workload when that is not desired. Fio will only modify the |
77 | device under test if `rw=write/randwrite/rw/randrw/trim/randtrim/trimwrite' | |
78 | is given. This safety net can be used as an extra precaution. | |
bdd88be3 | 79 | .TP |
d60e92d1 | 80 | .BI \-\-eta \fR=\fPwhen |
7db7a5a0 | 81 | Specifies when real\-time ETA estimate should be printed. \fIwhen\fR may |
db37d890 JA |
82 | be `always', `never' or `auto'. `auto' is the default, it prints ETA when |
83 | requested if the output is a TTY. `always' disregards the output type, and | |
84 | prints ETA when requested. `never' never prints ETA. | |
85 | .TP | |
86 | .BI \-\-eta\-interval \fR=\fPtime | |
87 | By default, fio requests client ETA status roughly every second. With this | |
88 | option, the interval is configurable. Fio imposes a minimum allowed time to | |
89 | avoid flooding the console, less than 250 msec is not supported. | |
d60e92d1 | 90 | .TP |
30b5d57f | 91 | .BI \-\-eta\-newline \fR=\fPtime |
bdd88be3 TK |
92 | Force a new line for every \fItime\fR period passed. When the unit is omitted, |
93 | the value is interpreted in seconds. | |
30b5d57f JA |
94 | .TP |
95 | .BI \-\-status\-interval \fR=\fPtime | |
aa6cb459 VF |
96 | Force a full status dump of cumulative (from job start) values at \fItime\fR |
97 | intervals. This option does *not* provide per-period measurements. So | |
98 | values such as bandwidth are running averages. When the time unit is omitted, | |
c1f4de8a JA |
99 | \fItime\fR is interpreted in seconds. Note that using this option with |
100 | `\-\-output-format=json' will yield output that technically isn't valid json, | |
101 | since the output will be collated sets of valid json. It will need to be split | |
102 | into valid sets of json after the run. | |
bdd88be3 TK |
103 | .TP |
104 | .BI \-\-section \fR=\fPname | |
105 | Only run specified section \fIname\fR in job file. Multiple sections can be specified. | |
7db7a5a0 | 106 | The \fB\-\-section\fR option allows one to combine related jobs into one file. |
bdd88be3 | 107 | E.g. one job file could define light, moderate, and heavy sections. Tell |
7db7a5a0 | 108 | fio to run only the "heavy" section by giving `\-\-section=heavy' |
bdd88be3 | 109 | command line option. One can also specify the "write" operations in one |
7db7a5a0 | 110 | section and "verify" operation in another section. The \fB\-\-section\fR option |
bdd88be3 TK |
111 | only applies to job sections. The reserved *global* section is always |
112 | parsed and used. | |
c0a5d35e | 113 | .TP |
49da1240 | 114 | .BI \-\-alloc\-size \fR=\fPkb |
4a419903 VF |
115 | Allocate additional internal smalloc pools of size \fIkb\fR in KiB. The |
116 | \fB\-\-alloc\-size\fR option increases shared memory set aside for use by fio. | |
bdd88be3 TK |
117 | If running large jobs with randommap enabled, fio can run out of memory. |
118 | Smalloc is an internal allocator for shared structures from a fixed size | |
119 | memory pool and can grow to 16 pools. The pool size defaults to 16MiB. | |
7db7a5a0 TK |
120 | NOTE: While running `.fio_smalloc.*' backing store files are visible |
121 | in `/tmp'. | |
d60e92d1 | 122 | .TP |
49da1240 JA |
123 | .BI \-\-warnings\-fatal |
124 | All fio parser warnings are fatal, causing fio to exit with an error. | |
9183788d | 125 | .TP |
49da1240 | 126 | .BI \-\-max\-jobs \fR=\fPnr |
7db7a5a0 | 127 | Set the maximum number of threads/processes to support to \fInr\fR. |
7f4811bb RNS |
128 | NOTE: On Linux, it may be necessary to increase the shared-memory limit |
129 | (`/proc/sys/kernel/shmmax') if fio runs into errors while creating jobs. | |
d60e92d1 | 130 | .TP |
49da1240 | 131 | .BI \-\-server \fR=\fPargs |
7db7a5a0 TK |
132 | Start a backend server, with \fIargs\fR specifying what to listen to. |
133 | See \fBCLIENT/SERVER\fR section. | |
f57a9c59 | 134 | .TP |
49da1240 | 135 | .BI \-\-daemonize \fR=\fPpidfile |
7db7a5a0 | 136 | Background a fio server, writing the pid to the given \fIpidfile\fR file. |
49da1240 | 137 | .TP |
bdd88be3 | 138 | .BI \-\-client \fR=\fPhostname |
7db7a5a0 TK |
139 | Instead of running the jobs locally, send and run them on the given \fIhostname\fR |
140 | or set of \fIhostname\fRs. See \fBCLIENT/SERVER\fR section. | |
bdd88be3 | 141 | .TP |
7db7a5a0 TK |
142 | .BI \-\-remote\-config \fR=\fPfile |
143 | Tell fio server to load this local \fIfile\fR. | |
f2a2ce0e HL |
144 | .TP |
145 | .BI \-\-idle\-prof \fR=\fPoption | |
7db7a5a0 | 146 | Report CPU idleness. \fIoption\fR is one of the following: |
bdd88be3 TK |
147 | .RS |
148 | .RS | |
149 | .TP | |
150 | .B calibrate | |
151 | Run unit work calibration only and exit. | |
152 | .TP | |
153 | .B system | |
154 | Show aggregate system idleness and unit work. | |
155 | .TP | |
156 | .B percpu | |
7db7a5a0 | 157 | As \fBsystem\fR but also show per CPU idleness. |
bdd88be3 TK |
158 | .RE |
159 | .RE | |
160 | .TP | |
7db7a5a0 TK |
161 | .BI \-\-inflate\-log \fR=\fPlog |
162 | Inflate and output compressed \fIlog\fR. | |
bdd88be3 | 163 | .TP |
7db7a5a0 TK |
164 | .BI \-\-trigger\-file \fR=\fPfile |
165 | Execute trigger command when \fIfile\fR exists. | |
bdd88be3 | 166 | .TP |
7db7a5a0 TK |
167 | .BI \-\-trigger\-timeout \fR=\fPtime |
168 | Execute trigger at this \fItime\fR. | |
bdd88be3 | 169 | .TP |
7db7a5a0 TK |
170 | .BI \-\-trigger \fR=\fPcommand |
171 | Set this \fIcommand\fR as local trigger. | |
bdd88be3 | 172 | .TP |
7db7a5a0 TK |
173 | .BI \-\-trigger\-remote \fR=\fPcommand |
174 | Set this \fIcommand\fR as remote trigger. | |
bdd88be3 | 175 | .TP |
7db7a5a0 | 176 | .BI \-\-aux\-path \fR=\fPpath |
f4401bf8 SW |
177 | Use the directory specified by \fIpath\fP for generated state files instead |
178 | of the current working directory. | |
d60e92d1 | 179 | .SH "JOB FILE FORMAT" |
7a14cf18 TK |
180 | Any parameters following the options will be assumed to be job files, unless |
181 | they match a job file parameter. Multiple job files can be listed and each job | |
7db7a5a0 | 182 | file will be regarded as a separate group. Fio will \fBstonewall\fR execution |
7a14cf18 TK |
183 | between each group. |
184 | ||
185 | Fio accepts one or more job files describing what it is | |
186 | supposed to do. The job file format is the classic ini file, where the names | |
187 | enclosed in [] brackets define the job name. You are free to use any ASCII name | |
188 | you want, except *global* which has special meaning. Following the job name is | |
189 | a sequence of zero or more parameters, one per line, that define the behavior of | |
190 | the job. If the first character in a line is a ';' or a '#', the entire line is | |
191 | discarded as a comment. | |
192 | ||
193 | A *global* section sets defaults for the jobs described in that file. A job may | |
194 | override a *global* section parameter, and a job file may even have several | |
195 | *global* sections if so desired. A job is only affected by a *global* section | |
196 | residing above it. | |
197 | ||
7db7a5a0 TK |
198 | The \fB\-\-cmdhelp\fR option also lists all options. If used with an \fIcommand\fR |
199 | argument, \fB\-\-cmdhelp\fR will detail the given \fIcommand\fR. | |
7a14cf18 | 200 | |
7db7a5a0 TK |
201 | See the `examples/' directory for inspiration on how to write job files. Note |
202 | the copyright and license requirements currently apply to | |
203 | `examples/' files. | |
07a2919d NP |
204 | |
205 | Note that the maximum length of a line in the job file is 8192 bytes. | |
54eb4569 TK |
206 | .SH "JOB FILE PARAMETERS" |
207 | Some parameters take an option of a given type, such as an integer or a | |
208 | string. Anywhere a numeric value is required, an arithmetic expression may be | |
209 | used, provided it is surrounded by parentheses. Supported operators are: | |
d59aa780 | 210 | .RS |
7db7a5a0 | 211 | .P |
d59aa780 | 212 | .B addition (+) |
7db7a5a0 TK |
213 | .P |
214 | .B subtraction (\-) | |
215 | .P | |
d59aa780 | 216 | .B multiplication (*) |
7db7a5a0 | 217 | .P |
d59aa780 | 218 | .B division (/) |
7db7a5a0 | 219 | .P |
d59aa780 | 220 | .B modulus (%) |
7db7a5a0 | 221 | .P |
d59aa780 JA |
222 | .B exponentiation (^) |
223 | .RE | |
d59aa780 JA |
224 | .P |
225 | For time values in expressions, units are microseconds by default. This is | |
226 | different than for time values not in expressions (not enclosed in | |
54eb4569 TK |
227 | parentheses). |
228 | .SH "PARAMETER TYPES" | |
229 | The following parameter types are used. | |
d60e92d1 AC |
230 | .TP |
231 | .I str | |
6b86fc18 TK |
232 | String. A sequence of alphanumeric characters. |
233 | .TP | |
234 | .I time | |
235 | Integer with possible time suffix. Without a unit value is interpreted as | |
236 | seconds unless otherwise specified. Accepts a suffix of 'd' for days, 'h' for | |
237 | hours, 'm' for minutes, 's' for seconds, 'ms' (or 'msec') for milliseconds and 'us' | |
238 | (or 'usec') for microseconds. For example, use 10m for 10 minutes. | |
d60e92d1 AC |
239 | .TP |
240 | .I int | |
6d500c2e RE |
241 | Integer. A whole number value, which may contain an integer prefix |
242 | and an integer suffix. | |
0b43a833 TK |
243 | .RS |
244 | .RS | |
245 | .P | |
6b86fc18 | 246 | [*integer prefix*] **number** [*integer suffix*] |
0b43a833 TK |
247 | .RE |
248 | .P | |
6b86fc18 TK |
249 | The optional *integer prefix* specifies the number's base. The default |
250 | is decimal. *0x* specifies hexadecimal. | |
0b43a833 | 251 | .P |
6b86fc18 TK |
252 | The optional *integer suffix* specifies the number's units, and includes an |
253 | optional unit prefix and an optional unit. For quantities of data, the | |
254 | default unit is bytes. For quantities of time, the default unit is seconds | |
255 | unless otherwise specified. | |
0b43a833 TK |
256 | .P |
257 | With `kb_base=1000', fio follows international standards for unit | |
338f2db5 | 258 | prefixes. To specify power-of-10 decimal values defined in the |
6b86fc18 | 259 | International System of Units (SI): |
0b43a833 TK |
260 | .RS |
261 | .P | |
7db7a5a0 | 262 | .PD 0 |
eccce61a | 263 | K means kilo (K) or 1000 |
7db7a5a0 | 264 | .P |
eccce61a | 265 | M means mega (M) or 1000**2 |
7db7a5a0 | 266 | .P |
eccce61a | 267 | G means giga (G) or 1000**3 |
7db7a5a0 | 268 | .P |
eccce61a | 269 | T means tera (T) or 1000**4 |
7db7a5a0 | 270 | .P |
eccce61a | 271 | P means peta (P) or 1000**5 |
7db7a5a0 | 272 | .PD |
0b43a833 TK |
273 | .RE |
274 | .P | |
338f2db5 | 275 | To specify power-of-2 binary values defined in IEC 80000-13: |
0b43a833 TK |
276 | .RS |
277 | .P | |
7db7a5a0 | 278 | .PD 0 |
eccce61a | 279 | Ki means kibi (Ki) or 1024 |
7db7a5a0 | 280 | .P |
eccce61a | 281 | Mi means mebi (Mi) or 1024**2 |
7db7a5a0 | 282 | .P |
eccce61a | 283 | Gi means gibi (Gi) or 1024**3 |
7db7a5a0 | 284 | .P |
eccce61a | 285 | Ti means tebi (Ti) or 1024**4 |
7db7a5a0 | 286 | .P |
eccce61a | 287 | Pi means pebi (Pi) or 1024**5 |
7db7a5a0 | 288 | .PD |
0b43a833 TK |
289 | .RE |
290 | .P | |
291 | With `kb_base=1024' (the default), the unit prefixes are opposite | |
338f2db5 | 292 | from those specified in the SI and IEC 80000-13 standards to provide |
6b86fc18 | 293 | compatibility with old scripts. For example, 4k means 4096. |
0b43a833 | 294 | .P |
6b86fc18 TK |
295 | For quantities of data, an optional unit of 'B' may be included |
296 | (e.g., 'kB' is the same as 'k'). | |
0b43a833 | 297 | .P |
6b86fc18 TK |
298 | The *integer suffix* is not case sensitive (e.g., m/mi mean mebi/mega, |
299 | not milli). 'b' and 'B' both mean byte, not bit. | |
0b43a833 TK |
300 | .P |
301 | Examples with `kb_base=1000': | |
302 | .RS | |
303 | .P | |
7db7a5a0 | 304 | .PD 0 |
6d500c2e | 305 | 4 KiB: 4096, 4096b, 4096B, 4k, 4kb, 4kB, 4K, 4KB |
7db7a5a0 | 306 | .P |
6d500c2e | 307 | 1 MiB: 1048576, 1m, 1024k |
7db7a5a0 | 308 | .P |
6d500c2e | 309 | 1 MB: 1000000, 1mi, 1000ki |
7db7a5a0 | 310 | .P |
6d500c2e | 311 | 1 TiB: 1073741824, 1t, 1024m, 1048576k |
7db7a5a0 | 312 | .P |
6d500c2e | 313 | 1 TB: 1000000000, 1ti, 1000mi, 1000000ki |
7db7a5a0 | 314 | .PD |
0b43a833 TK |
315 | .RE |
316 | .P | |
317 | Examples with `kb_base=1024' (default): | |
318 | .RS | |
319 | .P | |
7db7a5a0 | 320 | .PD 0 |
6d500c2e | 321 | 4 KiB: 4096, 4096b, 4096B, 4k, 4kb, 4kB, 4K, 4KB |
7db7a5a0 | 322 | .P |
6d500c2e | 323 | 1 MiB: 1048576, 1m, 1024k |
7db7a5a0 | 324 | .P |
6d500c2e | 325 | 1 MB: 1000000, 1mi, 1000ki |
7db7a5a0 | 326 | .P |
6d500c2e | 327 | 1 TiB: 1073741824, 1t, 1024m, 1048576k |
7db7a5a0 | 328 | .P |
6d500c2e | 329 | 1 TB: 1000000000, 1ti, 1000mi, 1000000ki |
7db7a5a0 | 330 | .PD |
0b43a833 TK |
331 | .RE |
332 | .P | |
6d500c2e | 333 | To specify times (units are not case sensitive): |
0b43a833 TK |
334 | .RS |
335 | .P | |
7db7a5a0 | 336 | .PD 0 |
6d500c2e | 337 | D means days |
7db7a5a0 | 338 | .P |
6d500c2e | 339 | H means hours |
7db7a5a0 | 340 | .P |
6d500c2e | 341 | M mean minutes |
7db7a5a0 | 342 | .P |
6d500c2e | 343 | s or sec means seconds (default) |
7db7a5a0 | 344 | .P |
6d500c2e | 345 | ms or msec means milliseconds |
7db7a5a0 | 346 | .P |
6d500c2e | 347 | us or usec means microseconds |
7db7a5a0 | 348 | .PD |
0b43a833 TK |
349 | .RE |
350 | .P | |
6b86fc18 | 351 | If the option accepts an upper and lower range, use a colon ':' or |
7db7a5a0 | 352 | minus '\-' to separate such values. See \fIirange\fR parameter type. |
6b86fc18 TK |
353 | If the lower value specified happens to be larger than the upper value |
354 | the two values are swapped. | |
0b43a833 | 355 | .RE |
d60e92d1 AC |
356 | .TP |
357 | .I bool | |
6b86fc18 TK |
358 | Boolean. Usually parsed as an integer, however only defined for |
359 | true and false (1 and 0). | |
d60e92d1 AC |
360 | .TP |
361 | .I irange | |
6b86fc18 | 362 | Integer range with suffix. Allows value range to be given, such as |
7db7a5a0 | 363 | 1024\-4096. A colon may also be used as the separator, e.g. 1k:4k. If the |
6b86fc18 | 364 | option allows two sets of ranges, they can be specified with a ',' or '/' |
7db7a5a0 | 365 | delimiter: 1k\-4k/8k\-32k. Also see \fIint\fR parameter type. |
83349190 YH |
366 | .TP |
367 | .I float_list | |
6b86fc18 | 368 | A list of floating point numbers, separated by a ':' character. |
523bad63 | 369 | .SH "JOB PARAMETERS" |
54eb4569 | 370 | With the above in mind, here follows the complete list of fio job parameters. |
523bad63 | 371 | .SS "Units" |
d60e92d1 | 372 | .TP |
523bad63 TK |
373 | .BI kb_base \fR=\fPint |
374 | Select the interpretation of unit prefixes in input parameters. | |
375 | .RS | |
376 | .RS | |
d60e92d1 | 377 | .TP |
523bad63 | 378 | .B 1000 |
338f2db5 | 379 | Inputs comply with IEC 80000-13 and the International |
523bad63 TK |
380 | System of Units (SI). Use: |
381 | .RS | |
382 | .P | |
383 | .PD 0 | |
338f2db5 | 384 | \- power-of-2 values with IEC prefixes (e.g., KiB) |
523bad63 | 385 | .P |
338f2db5 | 386 | \- power-of-10 values with SI prefixes (e.g., kB) |
523bad63 TK |
387 | .PD |
388 | .RE | |
389 | .TP | |
390 | .B 1024 | |
391 | Compatibility mode (default). To avoid breaking old scripts: | |
392 | .P | |
393 | .RS | |
394 | .PD 0 | |
338f2db5 | 395 | \- power-of-2 values with SI prefixes |
523bad63 | 396 | .P |
338f2db5 | 397 | \- power-of-10 values with IEC prefixes |
523bad63 TK |
398 | .PD |
399 | .RE | |
400 | .RE | |
401 | .P | |
402 | See \fBbs\fR for more details on input parameters. | |
403 | .P | |
404 | Outputs always use correct prefixes. Most outputs include both | |
338f2db5 | 405 | side-by-side, like: |
523bad63 TK |
406 | .P |
407 | .RS | |
408 | bw=2383.3kB/s (2327.4KiB/s) | |
409 | .RE | |
410 | .P | |
411 | If only one value is reported, then kb_base selects the one to use: | |
412 | .P | |
413 | .RS | |
414 | .PD 0 | |
415 | 1000 \-\- SI prefixes | |
416 | .P | |
417 | 1024 \-\- IEC prefixes | |
418 | .PD | |
419 | .RE | |
420 | .RE | |
421 | .TP | |
422 | .BI unit_base \fR=\fPint | |
423 | Base unit for reporting. Allowed values are: | |
424 | .RS | |
425 | .RS | |
426 | .TP | |
427 | .B 0 | |
338f2db5 | 428 | Use auto-detection (default). |
523bad63 TK |
429 | .TP |
430 | .B 8 | |
431 | Byte based. | |
432 | .TP | |
433 | .B 1 | |
434 | Bit based. | |
435 | .RE | |
436 | .RE | |
437 | .SS "Job description" | |
438 | .TP | |
439 | .BI name \fR=\fPstr | |
440 | ASCII name of the job. This may be used to override the name printed by fio | |
441 | for this job. Otherwise the job name is used. On the command line this | |
442 | parameter has the special purpose of also signaling the start of a new job. | |
9cc8cb91 | 443 | .TP |
d60e92d1 | 444 | .BI description \fR=\fPstr |
523bad63 TK |
445 | Text description of the job. Doesn't do anything except dump this text |
446 | description when this job is run. It's not parsed. | |
447 | .TP | |
448 | .BI loops \fR=\fPint | |
449 | Run the specified number of iterations of this job. Used to repeat the same | |
450 | workload a given number of times. Defaults to 1. | |
451 | .TP | |
452 | .BI numjobs \fR=\fPint | |
453 | Create the specified number of clones of this job. Each clone of job | |
454 | is spawned as an independent thread or process. May be used to setup a | |
455 | larger number of threads/processes doing the same thing. Each thread is | |
456 | reported separately; to see statistics for all clones as a whole, use | |
457 | \fBgroup_reporting\fR in conjunction with \fBnew_group\fR. | |
458 | See \fB\-\-max\-jobs\fR. Default: 1. | |
459 | .SS "Time related parameters" | |
460 | .TP | |
461 | .BI runtime \fR=\fPtime | |
462 | Tell fio to terminate processing after the specified period of time. It | |
463 | can be quite hard to determine for how long a specified job will run, so | |
464 | this parameter is handy to cap the total runtime to a given time. When | |
f1dd3fb1 | 465 | the unit is omitted, the value is interpreted in seconds. |
523bad63 TK |
466 | .TP |
467 | .BI time_based | |
468 | If set, fio will run for the duration of the \fBruntime\fR specified | |
469 | even if the file(s) are completely read or written. It will simply loop over | |
470 | the same workload as many times as the \fBruntime\fR allows. | |
471 | .TP | |
472 | .BI startdelay \fR=\fPirange(int) | |
473 | Delay the start of job for the specified amount of time. Can be a single | |
474 | value or a range. When given as a range, each thread will choose a value | |
475 | randomly from within the range. Value is in seconds if a unit is omitted. | |
476 | .TP | |
477 | .BI ramp_time \fR=\fPtime | |
478 | If set, fio will run the specified workload for this amount of time before | |
479 | logging any performance numbers. Useful for letting performance settle | |
480 | before logging results, thus minimizing the runtime required for stable | |
481 | results. Note that the \fBramp_time\fR is considered lead in time for a job, | |
482 | thus it will increase the total runtime if a special timeout or | |
483 | \fBruntime\fR is specified. When the unit is omitted, the value is | |
484 | given in seconds. | |
485 | .TP | |
486 | .BI clocksource \fR=\fPstr | |
487 | Use the given clocksource as the base of timing. The supported options are: | |
488 | .RS | |
489 | .RS | |
490 | .TP | |
491 | .B gettimeofday | |
492 | \fBgettimeofday\fR\|(2) | |
493 | .TP | |
494 | .B clock_gettime | |
495 | \fBclock_gettime\fR\|(2) | |
496 | .TP | |
497 | .B cpu | |
498 | Internal CPU clock source | |
499 | .RE | |
500 | .P | |
501 | \fBcpu\fR is the preferred clocksource if it is reliable, as it is very fast (and | |
502 | fio is heavy on time calls). Fio will automatically use this clocksource if | |
503 | it's supported and considered reliable on the system it is running on, | |
504 | unless another clocksource is specifically set. For x86/x86\-64 CPUs, this | |
505 | means supporting TSC Invariant. | |
506 | .RE | |
507 | .TP | |
508 | .BI gtod_reduce \fR=\fPbool | |
509 | Enable all of the \fBgettimeofday\fR\|(2) reducing options | |
510 | (\fBdisable_clat\fR, \fBdisable_slat\fR, \fBdisable_bw_measurement\fR) plus | |
511 | reduce precision of the timeout somewhat to really shrink the | |
512 | \fBgettimeofday\fR\|(2) call count. With this option enabled, we only do | |
513 | about 0.4% of the \fBgettimeofday\fR\|(2) calls we would have done if all | |
514 | time keeping was enabled. | |
515 | .TP | |
516 | .BI gtod_cpu \fR=\fPint | |
517 | Sometimes it's cheaper to dedicate a single thread of execution to just | |
518 | getting the current time. Fio (and databases, for instance) are very | |
519 | intensive on \fBgettimeofday\fR\|(2) calls. With this option, you can set | |
520 | one CPU aside for doing nothing but logging current time to a shared memory | |
521 | location. Then the other threads/processes that run I/O workloads need only | |
522 | copy that segment, instead of entering the kernel with a | |
523 | \fBgettimeofday\fR\|(2) call. The CPU set aside for doing these time | |
524 | calls will be excluded from other uses. Fio will manually clear it from the | |
525 | CPU mask of other jobs. | |
526 | .SS "Target file/device" | |
d60e92d1 AC |
527 | .TP |
528 | .BI directory \fR=\fPstr | |
523bad63 TK |
529 | Prefix \fBfilename\fRs with this directory. Used to place files in a different |
530 | location than `./'. You can specify a number of directories by | |
531 | separating the names with a ':' character. These directories will be | |
532 | assigned equally distributed to job clones created by \fBnumjobs\fR as | |
533 | long as they are using generated filenames. If specific \fBfilename\fR(s) are | |
534 | set fio will use the first listed directory, and thereby matching the | |
f4401bf8 SW |
535 | \fBfilename\fR semantic (which generates a file for each clone if not |
536 | specified, but lets all clones use the same file if set). | |
523bad63 TK |
537 | .RS |
538 | .P | |
3b803fe1 | 539 | See the \fBfilename\fR option for information on how to escape ':' |
523bad63 | 540 | characters within the directory path itself. |
f4401bf8 SW |
541 | .P |
542 | Note: To control the directory fio will use for internal state files | |
543 | use \fB\-\-aux\-path\fR. | |
523bad63 | 544 | .RE |
d60e92d1 AC |
545 | .TP |
546 | .BI filename \fR=\fPstr | |
523bad63 TK |
547 | Fio normally makes up a \fBfilename\fR based on the job name, thread number, and |
548 | file number (see \fBfilename_format\fR). If you want to share files | |
549 | between threads in a job or several | |
550 | jobs with fixed file paths, specify a \fBfilename\fR for each of them to override | |
551 | the default. If the ioengine is file based, you can specify a number of files | |
552 | by separating the names with a ':' colon. So if you wanted a job to open | |
553 | `/dev/sda' and `/dev/sdb' as the two working files, you would use | |
554 | `filename=/dev/sda:/dev/sdb'. This also means that whenever this option is | |
555 | specified, \fBnrfiles\fR is ignored. The size of regular files specified | |
556 | by this option will be \fBsize\fR divided by number of files unless an | |
557 | explicit size is specified by \fBfilesize\fR. | |
558 | .RS | |
559 | .P | |
3b803fe1 | 560 | Each colon in the wanted path must be escaped with a '\\' |
523bad63 TK |
561 | character. For instance, if the path is `/dev/dsk/foo@3,0:c' then you |
562 | would use `filename=/dev/dsk/foo@3,0\\:c' and if the path is | |
3b803fe1 | 563 | `F:\\filename' then you would use `filename=F\\:\\filename'. |
523bad63 | 564 | .P |
ffc90a44 SW |
565 | On Windows, disk devices are accessed as `\\\\.\\PhysicalDrive0' for |
566 | the first device, `\\\\.\\PhysicalDrive1' for the second etc. | |
523bad63 | 567 | Note: Windows and FreeBSD prevent write access to areas |
338f2db5 | 568 | of the disk containing in-use data (e.g. filesystems). |
523bad63 TK |
569 | .P |
570 | The filename `\-' is a reserved name, meaning *stdin* or *stdout*. Which | |
571 | of the two depends on the read/write direction set. | |
572 | .RE | |
d60e92d1 | 573 | .TP |
de98bd30 | 574 | .BI filename_format \fR=\fPstr |
523bad63 TK |
575 | If sharing multiple files between jobs, it is usually necessary to have fio |
576 | generate the exact names that you want. By default, fio will name a file | |
de98bd30 | 577 | based on the default file format specification of |
523bad63 | 578 | `jobname.jobnumber.filenumber'. With this option, that can be |
de98bd30 JA |
579 | customized. Fio will recognize and replace the following keywords in this |
580 | string: | |
581 | .RS | |
582 | .RS | |
583 | .TP | |
584 | .B $jobname | |
585 | The name of the worker thread or process. | |
586 | .TP | |
587 | .B $jobnum | |
588 | The incremental number of the worker thread or process. | |
589 | .TP | |
590 | .B $filenum | |
591 | The incremental number of the file for that worker thread or process. | |
592 | .RE | |
593 | .P | |
523bad63 TK |
594 | To have dependent jobs share a set of files, this option can be set to have |
595 | fio generate filenames that are shared between the two. For instance, if | |
596 | `testfiles.$filenum' is specified, file number 4 for any job will be | |
597 | named `testfiles.4'. The default of `$jobname.$jobnum.$filenum' | |
de98bd30 | 598 | will be used if no other format specifier is given. |
645943c0 JB |
599 | .P |
600 | If you specify a path then the directories will be created up to the main | |
601 | directory for the file. So for example if you specify `a/b/c/$jobnum` then the | |
602 | directories a/b/c will be created before the file setup part of the job. If you | |
603 | specify \fBdirectory\fR then the path will be relative that directory, otherwise | |
604 | it is treated as the absolute path. | |
de98bd30 | 605 | .RE |
de98bd30 | 606 | .TP |
922a5be8 | 607 | .BI unique_filename \fR=\fPbool |
523bad63 TK |
608 | To avoid collisions between networked clients, fio defaults to prefixing any |
609 | generated filenames (with a directory specified) with the source of the | |
610 | client connecting. To disable this behavior, set this option to 0. | |
611 | .TP | |
612 | .BI opendir \fR=\fPstr | |
613 | Recursively open any files below directory \fIstr\fR. | |
922a5be8 | 614 | .TP |
3ce9dcaf | 615 | .BI lockfile \fR=\fPstr |
523bad63 TK |
616 | Fio defaults to not locking any files before it does I/O to them. If a file |
617 | or file descriptor is shared, fio can serialize I/O to that file to make the | |
618 | end result consistent. This is usual for emulating real workloads that share | |
619 | files. The lock modes are: | |
3ce9dcaf JA |
620 | .RS |
621 | .RS | |
622 | .TP | |
623 | .B none | |
523bad63 | 624 | No locking. The default. |
3ce9dcaf JA |
625 | .TP |
626 | .B exclusive | |
523bad63 | 627 | Only one thread or process may do I/O at a time, excluding all others. |
3ce9dcaf JA |
628 | .TP |
629 | .B readwrite | |
523bad63 TK |
630 | Read\-write locking on the file. Many readers may |
631 | access the file at the same time, but writes get exclusive access. | |
3ce9dcaf | 632 | .RE |
ce594fbe | 633 | .RE |
523bad63 TK |
634 | .TP |
635 | .BI nrfiles \fR=\fPint | |
636 | Number of files to use for this job. Defaults to 1. The size of files | |
637 | will be \fBsize\fR divided by this unless explicit size is specified by | |
638 | \fBfilesize\fR. Files are created for each thread separately, and each | |
639 | file will have a file number within its name by default, as explained in | |
640 | \fBfilename\fR section. | |
641 | .TP | |
642 | .BI openfiles \fR=\fPint | |
643 | Number of files to keep open at the same time. Defaults to the same as | |
644 | \fBnrfiles\fR, can be set smaller to limit the number simultaneous | |
645 | opens. | |
646 | .TP | |
647 | .BI file_service_type \fR=\fPstr | |
648 | Defines how fio decides which file from a job to service next. The following | |
649 | types are defined: | |
650 | .RS | |
651 | .RS | |
652 | .TP | |
653 | .B random | |
654 | Choose a file at random. | |
655 | .TP | |
656 | .B roundrobin | |
657 | Round robin over opened files. This is the default. | |
658 | .TP | |
659 | .B sequential | |
660 | Finish one file before moving on to the next. Multiple files can | |
661 | still be open depending on \fBopenfiles\fR. | |
662 | .TP | |
663 | .B zipf | |
664 | Use a Zipf distribution to decide what file to access. | |
665 | .TP | |
666 | .B pareto | |
667 | Use a Pareto distribution to decide what file to access. | |
668 | .TP | |
669 | .B normal | |
670 | Use a Gaussian (normal) distribution to decide what file to access. | |
671 | .TP | |
672 | .B gauss | |
673 | Alias for normal. | |
674 | .RE | |
3ce9dcaf | 675 | .P |
523bad63 TK |
676 | For \fBrandom\fR, \fBroundrobin\fR, and \fBsequential\fR, a postfix can be appended to |
677 | tell fio how many I/Os to issue before switching to a new file. For example, | |
678 | specifying `file_service_type=random:8' would cause fio to issue | |
338f2db5 | 679 | 8 I/Os before selecting a new file at random. For the non-uniform |
523bad63 TK |
680 | distributions, a floating point postfix can be given to influence how the |
681 | distribution is skewed. See \fBrandom_distribution\fR for a description | |
682 | of how that would work. | |
683 | .RE | |
684 | .TP | |
685 | .BI ioscheduler \fR=\fPstr | |
686 | Attempt to switch the device hosting the file to the specified I/O scheduler | |
687 | before running. | |
688 | .TP | |
689 | .BI create_serialize \fR=\fPbool | |
690 | If true, serialize the file creation for the jobs. This may be handy to | |
691 | avoid interleaving of data files, which may greatly depend on the filesystem | |
692 | used and even the number of processors in the system. Default: true. | |
693 | .TP | |
694 | .BI create_fsync \fR=\fPbool | |
695 | \fBfsync\fR\|(2) the data file after creation. This is the default. | |
696 | .TP | |
697 | .BI create_on_open \fR=\fPbool | |
338f2db5 SW |
698 | If true, don't pre-create files but allow the job's open() to create a file |
699 | when it's time to do I/O. Default: false \-\- pre-create all necessary files | |
523bad63 TK |
700 | when the job starts. |
701 | .TP | |
702 | .BI create_only \fR=\fPbool | |
703 | If true, fio will only run the setup phase of the job. If files need to be | |
704 | laid out or updated on disk, only that will be done \-\- the actual job contents | |
705 | are not executed. Default: false. | |
706 | .TP | |
707 | .BI allow_file_create \fR=\fPbool | |
708 | If true, fio is permitted to create files as part of its workload. If this | |
709 | option is false, then fio will error out if | |
710 | the files it needs to use don't already exist. Default: true. | |
711 | .TP | |
712 | .BI allow_mounted_write \fR=\fPbool | |
713 | If this isn't set, fio will abort jobs that are destructive (e.g. that write) | |
714 | to what appears to be a mounted device or partition. This should help catch | |
715 | creating inadvertently destructive tests, not realizing that the test will | |
716 | destroy data on the mounted file system. Note that some platforms don't allow | |
717 | writing against a mounted device regardless of this option. Default: false. | |
718 | .TP | |
719 | .BI pre_read \fR=\fPbool | |
338f2db5 | 720 | If this is given, files will be pre-read into memory before starting the |
523bad63 | 721 | given I/O operation. This will also clear the \fBinvalidate\fR flag, |
338f2db5 SW |
722 | since it is pointless to pre-read and then drop the cache. This will only |
723 | work for I/O engines that are seek-able, since they allow you to read the | |
724 | same data multiple times. Thus it will not work on non-seekable I/O engines | |
523bad63 TK |
725 | (e.g. network, splice). Default: false. |
726 | .TP | |
727 | .BI unlink \fR=\fPbool | |
728 | Unlink the job files when done. Not the default, as repeated runs of that | |
729 | job would then waste time recreating the file set again and again. Default: | |
730 | false. | |
731 | .TP | |
732 | .BI unlink_each_loop \fR=\fPbool | |
733 | Unlink job files after each iteration or loop. Default: false. | |
734 | .TP | |
7b865a2f BVA |
735 | .BI zonemode \fR=\fPstr |
736 | Accepted values are: | |
737 | .RS | |
738 | .RS | |
739 | .TP | |
740 | .B none | |
b8dd9750 HH |
741 | The \fBzonerange\fR, \fBzonesize\fR \fBzonecapacity\fR and \fBzoneskip\fR |
742 | parameters are ignored. | |
7b865a2f BVA |
743 | .TP |
744 | .B strided | |
745 | I/O happens in a single zone until \fBzonesize\fR bytes have been transferred. | |
746 | After that number of bytes has been transferred processing of the next zone | |
b8dd9750 | 747 | starts. The \fBzonecapacity\fR parameter is ignored. |
7b865a2f BVA |
748 | .TP |
749 | .B zbd | |
750 | Zoned block device mode. I/O happens sequentially in each zone, even if random | |
751 | I/O has been selected. Random I/O happens across all zones instead of being | |
752 | restricted to a single zone. | |
753 | .RE | |
754 | .RE | |
523bad63 TK |
755 | .TP |
756 | .BI zonerange \fR=\fPint | |
d4e058cd DLM |
757 | For \fBzonemode\fR=strided, this is the size of a single zone. See also |
758 | \fBzonesize\fR and \fBzoneskip\fR. | |
759 | ||
760 | For \fBzonemode\fR=zbd, this parameter is ignored. | |
5faddc64 BVA |
761 | .TP |
762 | .BI zonesize \fR=\fPint | |
7b865a2f BVA |
763 | For \fBzonemode\fR=strided, this is the number of bytes to transfer before |
764 | skipping \fBzoneskip\fR bytes. If this parameter is smaller than | |
765 | \fBzonerange\fR then only a fraction of each zone with \fBzonerange\fR bytes | |
766 | will be accessed. If this parameter is larger than \fBzonerange\fR then each | |
767 | zone will be accessed multiple times before skipping to the next zone. | |
768 | ||
d4e058cd DLM |
769 | For \fBzonemode\fR=zbd, this is the size of a single zone. The |
770 | \fBzonerange\fR parameter is ignored in this mode. For a job accessing a | |
771 | zoned block device, the specified \fBzonesize\fR must be 0 or equal to the | |
772 | device zone size. For a regular block device or file, the specified | |
773 | \fBzonesize\fR must be at least 512B. | |
523bad63 | 774 | .TP |
b8dd9750 HH |
775 | .BI zonecapacity \fR=\fPint |
776 | For \fBzonemode\fR=zbd, this defines the capacity of a single zone, which is | |
777 | the accessible area starting from the zone start address. This parameter only | |
778 | applies when using \fBzonemode\fR=zbd in combination with regular block devices. | |
779 | If not specified it defaults to the zone size. If the target device is a zoned | |
780 | block device, the zone capacity is obtained from the device information and this | |
781 | option is ignored. | |
782 | .TP | |
523bad63 | 783 | .BI zoneskip \fR=\fPint |
7b865a2f | 784 | For \fBzonemode\fR=strided, the number of bytes to skip after \fBzonesize\fR |
4d37720a DLM |
785 | bytes of data have been transferred. |
786 | ||
787 | For \fBzonemode\fR=zbd, the \fBzonesize\fR aligned number of bytes to skip | |
788 | once a zone is fully written (write workloads) or all written data in the | |
789 | zone have been read (read workloads). This parameter is valid only for | |
790 | sequential workloads and ignored for random workloads. For read workloads, | |
791 | see also \fBread_beyond_wp\fR. | |
5faddc64 | 792 | |
bfbdd35b BVA |
793 | .TP |
794 | .BI read_beyond_wp \fR=\fPbool | |
795 | This parameter applies to \fBzonemode=zbd\fR only. | |
796 | ||
797 | Zoned block devices are block devices that consist of multiple zones. Each | |
798 | zone has a type, e.g. conventional or sequential. A conventional zone can be | |
799 | written at any offset that is a multiple of the block size. Sequential zones | |
800 | must be written sequentially. The position at which a write must occur is | |
402f0887 DLM |
801 | called the write pointer. A zoned block device can be either host managed or |
802 | host aware. For host managed devices the host must ensure that writes happen | |
803 | sequentially. Fio recognizes host managed devices and serializes writes to | |
804 | sequential zones for these devices. | |
bfbdd35b BVA |
805 | |
806 | If a read occurs in a sequential zone beyond the write pointer then the zoned | |
807 | block device will complete the read without reading any data from the storage | |
808 | medium. Since such reads lead to unrealistically high bandwidth and IOPS | |
809 | numbers fio only reads beyond the write pointer if explicitly told to do | |
810 | so. Default: false. | |
59b07544 BVA |
811 | .TP |
812 | .BI max_open_zones \fR=\fPint | |
813 | When running a random write test across an entire drive many more zones will be | |
814 | open than in a typical application workload. Hence this command line option | |
815 | that allows to limit the number of open zones. The number of open zones is | |
219c662d AD |
816 | defined as the number of zones to which write commands are issued by all |
817 | threads/processes. | |
818 | .TP | |
819 | .BI job_max_open_zones \fR=\fPint | |
820 | Limit on the number of simultaneously opened zones per single thread/process. | |
a7c2b6fc BVA |
821 | .TP |
822 | .BI zone_reset_threshold \fR=\fPfloat | |
823 | A number between zero and one that indicates the ratio of logical blocks with | |
824 | data to the total number of logical blocks in the test above which zones | |
825 | should be reset periodically. | |
826 | .TP | |
827 | .BI zone_reset_frequency \fR=\fPfloat | |
828 | A number between zero and one that indicates how often a zone reset should be | |
829 | issued if the zone reset threshold has been exceeded. A zone reset is | |
830 | submitted after each (1 / zone_reset_frequency) write requests. This and the | |
831 | previous parameter can be used to simulate garbage collection activity. | |
bfbdd35b | 832 | |
523bad63 TK |
833 | .SS "I/O type" |
834 | .TP | |
835 | .BI direct \fR=\fPbool | |
338f2db5 | 836 | If value is true, use non-buffered I/O. This is usually O_DIRECT. Note that |
8e889110 | 837 | OpenBSD and ZFS on Solaris don't support direct I/O. On Windows the synchronous |
523bad63 TK |
838 | ioengines don't support direct I/O. Default: false. |
839 | .TP | |
840 | .BI atomic \fR=\fPbool | |
841 | If value is true, attempt to use atomic direct I/O. Atomic writes are | |
842 | guaranteed to be stable once acknowledged by the operating system. Only | |
843 | Linux supports O_ATOMIC right now. | |
844 | .TP | |
845 | .BI buffered \fR=\fPbool | |
846 | If value is true, use buffered I/O. This is the opposite of the | |
847 | \fBdirect\fR option. Defaults to true. | |
d60e92d1 AC |
848 | .TP |
849 | .BI readwrite \fR=\fPstr "\fR,\fP rw" \fR=\fPstr | |
523bad63 | 850 | Type of I/O pattern. Accepted values are: |
d60e92d1 AC |
851 | .RS |
852 | .RS | |
853 | .TP | |
854 | .B read | |
d1429b5c | 855 | Sequential reads. |
d60e92d1 AC |
856 | .TP |
857 | .B write | |
d1429b5c | 858 | Sequential writes. |
d60e92d1 | 859 | .TP |
fa769d44 | 860 | .B trim |
3740cfc8 | 861 | Sequential trims (Linux block devices and SCSI character devices only). |
fa769d44 | 862 | .TP |
d60e92d1 | 863 | .B randread |
d1429b5c | 864 | Random reads. |
d60e92d1 AC |
865 | .TP |
866 | .B randwrite | |
d1429b5c | 867 | Random writes. |
d60e92d1 | 868 | .TP |
fa769d44 | 869 | .B randtrim |
3740cfc8 | 870 | Random trims (Linux block devices and SCSI character devices only). |
fa769d44 | 871 | .TP |
523bad63 TK |
872 | .B rw,readwrite |
873 | Sequential mixed reads and writes. | |
d60e92d1 | 874 | .TP |
ff6bb260 | 875 | .B randrw |
523bad63 | 876 | Random mixed reads and writes. |
82a90686 JA |
877 | .TP |
878 | .B trimwrite | |
523bad63 TK |
879 | Sequential trim+write sequences. Blocks will be trimmed first, |
880 | then the same blocks will be written to. | |
d60e92d1 AC |
881 | .RE |
882 | .P | |
523bad63 TK |
883 | Fio defaults to read if the option is not specified. For the mixed I/O |
884 | types, the default is to split them 50/50. For certain types of I/O the | |
885 | result may still be skewed a bit, since the speed may be different. | |
886 | .P | |
887 | It is possible to specify the number of I/Os to do before getting a new | |
888 | offset by appending `:<nr>' to the end of the string given. For a | |
889 | random read, it would look like `rw=randread:8' for passing in an offset | |
890 | modifier with a value of 8. If the suffix is used with a sequential I/O | |
891 | pattern, then the `<nr>' value specified will be added to the generated | |
892 | offset for each I/O turning sequential I/O into sequential I/O with holes. | |
893 | For instance, using `rw=write:4k' will skip 4k for every write. Also see | |
894 | the \fBrw_sequencer\fR option. | |
d60e92d1 AC |
895 | .RE |
896 | .TP | |
38dad62d | 897 | .BI rw_sequencer \fR=\fPstr |
523bad63 TK |
898 | If an offset modifier is given by appending a number to the `rw=\fIstr\fR' |
899 | line, then this option controls how that number modifies the I/O offset | |
900 | being generated. Accepted values are: | |
38dad62d JA |
901 | .RS |
902 | .RS | |
903 | .TP | |
904 | .B sequential | |
523bad63 | 905 | Generate sequential offset. |
38dad62d JA |
906 | .TP |
907 | .B identical | |
523bad63 | 908 | Generate the same offset. |
38dad62d JA |
909 | .RE |
910 | .P | |
523bad63 TK |
911 | \fBsequential\fR is only useful for random I/O, where fio would normally |
912 | generate a new random offset for every I/O. If you append e.g. 8 to randread, | |
913 | you would get a new random offset for every 8 I/Os. The result would be a | |
914 | seek for only every 8 I/Os, instead of for every I/O. Use `rw=randread:8' | |
915 | to specify that. As sequential I/O is already sequential, setting | |
916 | \fBsequential\fR for that would not result in any differences. \fBidentical\fR | |
917 | behaves in a similar fashion, except it sends the same offset 8 number of | |
918 | times before generating a new offset. | |
38dad62d | 919 | .RE |
90fef2d1 | 920 | .TP |
771e58be JA |
921 | .BI unified_rw_reporting \fR=\fPbool |
922 | Fio normally reports statistics on a per data direction basis, meaning that | |
523bad63 TK |
923 | reads, writes, and trims are accounted and reported separately. If this |
924 | option is set fio sums the results and report them as "mixed" instead. | |
771e58be | 925 | .TP |
d60e92d1 | 926 | .BI randrepeat \fR=\fPbool |
523bad63 TK |
927 | Seed the random number generator used for random I/O patterns in a |
928 | predictable way so the pattern is repeatable across runs. Default: true. | |
56e2a5fc CE |
929 | .TP |
930 | .BI allrandrepeat \fR=\fPbool | |
931 | Seed all random number generators in a predictable way so results are | |
523bad63 | 932 | repeatable across runs. Default: false. |
d60e92d1 | 933 | .TP |
04778baf JA |
934 | .BI randseed \fR=\fPint |
935 | Seed the random number generators based on this seed value, to be able to | |
936 | control what sequence of output is being generated. If not set, the random | |
937 | sequence depends on the \fBrandrepeat\fR setting. | |
938 | .TP | |
a596f047 | 939 | .BI fallocate \fR=\fPstr |
338f2db5 | 940 | Whether pre-allocation is performed when laying down files. |
523bad63 | 941 | Accepted values are: |
a596f047 EG |
942 | .RS |
943 | .RS | |
944 | .TP | |
945 | .B none | |
338f2db5 | 946 | Do not pre-allocate space. |
a596f047 | 947 | .TP |
2c3e17be | 948 | .B native |
338f2db5 | 949 | Use a platform's native pre-allocation call but fall back to |
523bad63 | 950 | \fBnone\fR behavior if it fails/is not implemented. |
2c3e17be | 951 | .TP |
a596f047 | 952 | .B posix |
338f2db5 | 953 | Pre-allocate via \fBposix_fallocate\fR\|(3). |
a596f047 EG |
954 | .TP |
955 | .B keep | |
338f2db5 | 956 | Pre-allocate via \fBfallocate\fR\|(2) with |
523bad63 | 957 | FALLOC_FL_KEEP_SIZE set. |
a596f047 | 958 | .TP |
38ca5f03 TV |
959 | .B truncate |
960 | Extend file to final size using \fBftruncate\fR|(2) | |
961 | instead of allocating. | |
962 | .TP | |
a596f047 | 963 | .B 0 |
338f2db5 | 964 | Backward-compatible alias for \fBnone\fR. |
a596f047 EG |
965 | .TP |
966 | .B 1 | |
338f2db5 | 967 | Backward-compatible alias for \fBposix\fR. |
a596f047 EG |
968 | .RE |
969 | .P | |
523bad63 TK |
970 | May not be available on all supported platforms. \fBkeep\fR is only available |
971 | on Linux. If using ZFS on Solaris this cannot be set to \fBposix\fR | |
338f2db5 | 972 | because ZFS doesn't support pre-allocation. Default: \fBnative\fR if any |
38ca5f03 TV |
973 | pre-allocation methods except \fBtruncate\fR are available, \fBnone\fR if not. |
974 | .P | |
975 | Note that using \fBtruncate\fR on Windows will interact surprisingly | |
976 | with non-sequential write patterns. When writing to a file that has | |
977 | been extended by setting the end-of-file information, Windows will | |
978 | backfill the unwritten portion of the file up to that offset with | |
979 | zeroes before issuing the new write. This means that a single small | |
980 | write to the end of an extended file will stall until the entire | |
981 | file has been filled with zeroes. | |
a596f047 | 982 | .RE |
7bc8c2cf | 983 | .TP |
ecb2083d | 984 | .BI fadvise_hint \fR=\fPstr |
c712c97a JA |
985 | Use \fBposix_fadvise\fR\|(2) or \fBposix_madvise\fR\|(2) to advise the kernel |
986 | what I/O patterns are likely to be issued. Accepted values are: | |
ecb2083d JA |
987 | .RS |
988 | .RS | |
989 | .TP | |
990 | .B 0 | |
991 | Backwards compatible hint for "no hint". | |
992 | .TP | |
993 | .B 1 | |
994 | Backwards compatible hint for "advise with fio workload type". This | |
523bad63 | 995 | uses FADV_RANDOM for a random workload, and FADV_SEQUENTIAL |
ecb2083d JA |
996 | for a sequential workload. |
997 | .TP | |
998 | .B sequential | |
523bad63 | 999 | Advise using FADV_SEQUENTIAL. |
ecb2083d JA |
1000 | .TP |
1001 | .B random | |
523bad63 | 1002 | Advise using FADV_RANDOM. |
ecb2083d JA |
1003 | .RE |
1004 | .RE | |
d60e92d1 | 1005 | .TP |
8f4b9f24 | 1006 | .BI write_hint \fR=\fPstr |
523bad63 TK |
1007 | Use \fBfcntl\fR\|(2) to advise the kernel what life time to expect |
1008 | from a write. Only supported on Linux, as of version 4.13. Accepted | |
8f4b9f24 JA |
1009 | values are: |
1010 | .RS | |
1011 | .RS | |
1012 | .TP | |
1013 | .B none | |
1014 | No particular life time associated with this file. | |
1015 | .TP | |
1016 | .B short | |
1017 | Data written to this file has a short life time. | |
1018 | .TP | |
1019 | .B medium | |
1020 | Data written to this file has a medium life time. | |
1021 | .TP | |
1022 | .B long | |
1023 | Data written to this file has a long life time. | |
1024 | .TP | |
1025 | .B extreme | |
1026 | Data written to this file has a very long life time. | |
1027 | .RE | |
523bad63 TK |
1028 | .P |
1029 | The values are all relative to each other, and no absolute meaning | |
1030 | should be associated with them. | |
8f4b9f24 | 1031 | .RE |
37659335 | 1032 | .TP |
523bad63 TK |
1033 | .BI offset \fR=\fPint |
1034 | Start I/O at the provided offset in the file, given as either a fixed size in | |
83c8b093 JF |
1035 | bytes or a percentage. If a percentage is given, the generated offset will be |
1036 | aligned to the minimum \fBblocksize\fR or to the value of \fBoffset_align\fR if | |
1037 | provided. Data before the given offset will not be touched. This | |
523bad63 TK |
1038 | effectively caps the file size at `real_size \- offset'. Can be combined with |
1039 | \fBsize\fR to constrain the start and end range of the I/O workload. | |
1040 | A percentage can be specified by a number between 1 and 100 followed by '%', | |
1041 | for example, `offset=20%' to specify 20%. | |
6d500c2e | 1042 | .TP |
83c8b093 JF |
1043 | .BI offset_align \fR=\fPint |
1044 | If set to non-zero value, the byte offset generated by a percentage \fBoffset\fR | |
1045 | is aligned upwards to this value. Defaults to 0 meaning that a percentage | |
1046 | offset is aligned to the minimum block size. | |
1047 | .TP | |
523bad63 TK |
1048 | .BI offset_increment \fR=\fPint |
1049 | If this is provided, then the real offset becomes `\fBoffset\fR + \fBoffset_increment\fR | |
1050 | * thread_number', where the thread number is a counter that starts at 0 and | |
338f2db5 | 1051 | is incremented for each sub-job (i.e. when \fBnumjobs\fR option is |
523bad63 TK |
1052 | specified). This option is useful if there are several jobs which are |
1053 | intended to operate on a file in parallel disjoint segments, with even | |
0b288ba1 VF |
1054 | spacing between the starting points. Percentages can be used for this option. |
1055 | If a percentage is given, the generated offset will be aligned to the minimum | |
1056 | \fBblocksize\fR or to the value of \fBoffset_align\fR if provided. | |
6d500c2e | 1057 | .TP |
523bad63 TK |
1058 | .BI number_ios \fR=\fPint |
1059 | Fio will normally perform I/Os until it has exhausted the size of the region | |
1060 | set by \fBsize\fR, or if it exhaust the allocated time (or hits an error | |
1061 | condition). With this setting, the range/size can be set independently of | |
1062 | the number of I/Os to perform. When fio reaches this number, it will exit | |
1063 | normally and report status. Note that this does not extend the amount of I/O | |
1064 | that will be done, it will only stop fio if this condition is met before | |
338f2db5 | 1065 | other end-of-job criteria. |
d60e92d1 | 1066 | .TP |
523bad63 TK |
1067 | .BI fsync \fR=\fPint |
1068 | If writing to a file, issue an \fBfsync\fR\|(2) (or its equivalent) of | |
1069 | the dirty data for every number of blocks given. For example, if you give 32 | |
1070 | as a parameter, fio will sync the file after every 32 writes issued. If fio is | |
338f2db5 | 1071 | using non-buffered I/O, we may not sync the file. The exception is the sg |
523bad63 TK |
1072 | I/O engine, which synchronizes the disk cache anyway. Defaults to 0, which |
1073 | means fio does not periodically issue and wait for a sync to complete. Also | |
1074 | see \fBend_fsync\fR and \fBfsync_on_close\fR. | |
6d500c2e | 1075 | .TP |
523bad63 TK |
1076 | .BI fdatasync \fR=\fPint |
1077 | Like \fBfsync\fR but uses \fBfdatasync\fR\|(2) to only sync data and | |
44f668d7 | 1078 | not metadata blocks. In Windows, FreeBSD, DragonFlyBSD or OSX there is no |
523bad63 TK |
1079 | \fBfdatasync\fR\|(2) so this falls back to using \fBfsync\fR\|(2). |
1080 | Defaults to 0, which means fio does not periodically issue and wait for a | |
338f2db5 | 1081 | data-only sync to complete. |
d60e92d1 | 1082 | .TP |
523bad63 TK |
1083 | .BI write_barrier \fR=\fPint |
1084 | Make every N\-th write a barrier write. | |
901bb994 | 1085 | .TP |
523bad63 TK |
1086 | .BI sync_file_range \fR=\fPstr:int |
1087 | Use \fBsync_file_range\fR\|(2) for every \fIint\fR number of write | |
1088 | operations. Fio will track range of writes that have happened since the last | |
1089 | \fBsync_file_range\fR\|(2) call. \fIstr\fR can currently be one or more of: | |
1090 | .RS | |
1091 | .RS | |
fd68418e | 1092 | .TP |
523bad63 TK |
1093 | .B wait_before |
1094 | SYNC_FILE_RANGE_WAIT_BEFORE | |
c5751c62 | 1095 | .TP |
523bad63 TK |
1096 | .B write |
1097 | SYNC_FILE_RANGE_WRITE | |
c5751c62 | 1098 | .TP |
523bad63 TK |
1099 | .B wait_after |
1100 | SYNC_FILE_RANGE_WRITE_AFTER | |
2fa5a241 | 1101 | .RE |
523bad63 TK |
1102 | .P |
1103 | So if you do `sync_file_range=wait_before,write:8', fio would use | |
1104 | `SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE' for every 8 | |
1105 | writes. Also see the \fBsync_file_range\fR\|(2) man page. This option is | |
1106 | Linux specific. | |
2fa5a241 | 1107 | .RE |
ce35b1ec | 1108 | .TP |
523bad63 TK |
1109 | .BI overwrite \fR=\fPbool |
1110 | If true, writes to a file will always overwrite existing data. If the file | |
1111 | doesn't already exist, it will be created before the write phase begins. If | |
1112 | the file exists and is large enough for the specified write phase, nothing | |
1113 | will be done. Default: false. | |
5c94b008 | 1114 | .TP |
523bad63 TK |
1115 | .BI end_fsync \fR=\fPbool |
1116 | If true, \fBfsync\fR\|(2) file contents when a write stage has completed. | |
1117 | Default: false. | |
d60e92d1 | 1118 | .TP |
523bad63 TK |
1119 | .BI fsync_on_close \fR=\fPbool |
1120 | If true, fio will \fBfsync\fR\|(2) a dirty file on close. This differs | |
1121 | from \fBend_fsync\fR in that it will happen on every file close, not | |
1122 | just at the end of the job. Default: false. | |
d60e92d1 | 1123 | .TP |
523bad63 TK |
1124 | .BI rwmixread \fR=\fPint |
1125 | Percentage of a mixed workload that should be reads. Default: 50. | |
1126 | .TP | |
1127 | .BI rwmixwrite \fR=\fPint | |
1128 | Percentage of a mixed workload that should be writes. If both | |
1129 | \fBrwmixread\fR and \fBrwmixwrite\fR is given and the values do not | |
1130 | add up to 100%, the latter of the two will be used to override the | |
1131 | first. This may interfere with a given rate setting, if fio is asked to | |
1132 | limit reads or writes to a certain rate. If that is the case, then the | |
1133 | distribution may be skewed. Default: 50. | |
1134 | .TP | |
1135 | .BI random_distribution \fR=\fPstr:float[,str:float][,str:float] | |
1136 | By default, fio will use a completely uniform random distribution when asked | |
1137 | to perform random I/O. Sometimes it is useful to skew the distribution in | |
1138 | specific ways, ensuring that some parts of the data is more hot than others. | |
1139 | fio includes the following distribution models: | |
d60e92d1 AC |
1140 | .RS |
1141 | .RS | |
1142 | .TP | |
1143 | .B random | |
523bad63 | 1144 | Uniform random distribution |
8c07860d JA |
1145 | .TP |
1146 | .B zipf | |
523bad63 | 1147 | Zipf distribution |
8c07860d JA |
1148 | .TP |
1149 | .B pareto | |
523bad63 | 1150 | Pareto distribution |
8c07860d | 1151 | .TP |
dd3503d3 | 1152 | .B normal |
523bad63 | 1153 | Normal (Gaussian) distribution |
dd3503d3 | 1154 | .TP |
523bad63 TK |
1155 | .B zoned |
1156 | Zoned random distribution | |
59466396 JA |
1157 | .B zoned_abs |
1158 | Zoned absolute random distribution | |
d60e92d1 AC |
1159 | .RE |
1160 | .P | |
523bad63 TK |
1161 | When using a \fBzipf\fR or \fBpareto\fR distribution, an input value is also |
1162 | needed to define the access pattern. For \fBzipf\fR, this is the `Zipf theta'. | |
1163 | For \fBpareto\fR, it's the `Pareto power'. Fio includes a test | |
1164 | program, \fBfio\-genzipf\fR, that can be used visualize what the given input | |
1165 | values will yield in terms of hit rates. If you wanted to use \fBzipf\fR with | |
1166 | a `theta' of 1.2, you would use `random_distribution=zipf:1.2' as the | |
1167 | option. If a non\-uniform model is used, fio will disable use of the random | |
1168 | map. For the \fBnormal\fR distribution, a normal (Gaussian) deviation is | |
1169 | supplied as a value between 0 and 100. | |
1170 | .P | |
1171 | For a \fBzoned\fR distribution, fio supports specifying percentages of I/O | |
1172 | access that should fall within what range of the file or device. For | |
1173 | example, given a criteria of: | |
d60e92d1 | 1174 | .RS |
523bad63 TK |
1175 | .P |
1176 | .PD 0 | |
1177 | 60% of accesses should be to the first 10% | |
1178 | .P | |
1179 | 30% of accesses should be to the next 20% | |
1180 | .P | |
1181 | 8% of accesses should be to the next 30% | |
1182 | .P | |
1183 | 2% of accesses should be to the next 40% | |
1184 | .PD | |
1185 | .RE | |
1186 | .P | |
1187 | we can define that through zoning of the random accesses. For the above | |
1188 | example, the user would do: | |
1189 | .RS | |
1190 | .P | |
1191 | random_distribution=zoned:60/10:30/20:8/30:2/40 | |
1192 | .RE | |
1193 | .P | |
59466396 JA |
1194 | A \fBzoned_abs\fR distribution works exactly like the\fBzoned\fR, except that |
1195 | it takes absolute sizes. For example, let's say you wanted to define access | |
1196 | according to the following criteria: | |
1197 | .RS | |
1198 | .P | |
1199 | .PD 0 | |
1200 | 60% of accesses should be to the first 20G | |
1201 | .P | |
1202 | 30% of accesses should be to the next 100G | |
1203 | .P | |
1204 | 10% of accesses should be to the next 500G | |
1205 | .PD | |
1206 | .RE | |
1207 | .P | |
1208 | we can define an absolute zoning distribution with: | |
1209 | .RS | |
1210 | .P | |
1211 | random_distribution=zoned:60/10:30/20:8/30:2/40 | |
1212 | .RE | |
1213 | .P | |
6a16ece8 JA |
1214 | For both \fBzoned\fR and \fBzoned_abs\fR, fio supports defining up to 256 |
1215 | separate zones. | |
1216 | .P | |
59466396 | 1217 | Similarly to how \fBbssplit\fR works for setting ranges and percentages |
523bad63 TK |
1218 | of block sizes. Like \fBbssplit\fR, it's possible to specify separate |
1219 | zones for reads, writes, and trims. If just one set is given, it'll apply to | |
1220 | all of them. | |
1221 | .RE | |
1222 | .TP | |
1223 | .BI percentage_random \fR=\fPint[,int][,int] | |
1224 | For a random workload, set how big a percentage should be random. This | |
1225 | defaults to 100%, in which case the workload is fully random. It can be set | |
1226 | from anywhere from 0 to 100. Setting it to 0 would make the workload fully | |
1227 | sequential. Any setting in between will result in a random mix of sequential | |
338f2db5 | 1228 | and random I/O, at the given percentages. Comma-separated values may be |
523bad63 TK |
1229 | specified for reads, writes, and trims as described in \fBblocksize\fR. |
1230 | .TP | |
1231 | .BI norandommap | |
1232 | Normally fio will cover every block of the file when doing random I/O. If | |
1233 | this option is given, fio will just get a new random offset without looking | |
1234 | at past I/O history. This means that some blocks may not be read or written, | |
1235 | and that some blocks may be read/written more than once. If this option is | |
1236 | used with \fBverify\fR and multiple blocksizes (via \fBbsrange\fR), | |
338f2db5 | 1237 | only intact blocks are verified, i.e., partially-overwritten blocks are |
47e6a6e5 SW |
1238 | ignored. With an async I/O engine and an I/O depth > 1, it is possible for |
1239 | the same block to be overwritten, which can cause verification errors. Either | |
1240 | do not use norandommap in this case, or also use the lfsr random generator. | |
523bad63 TK |
1241 | .TP |
1242 | .BI softrandommap \fR=\fPbool | |
1243 | See \fBnorandommap\fR. If fio runs with the random block map enabled and | |
1244 | it fails to allocate the map, if this option is set it will continue without | |
1245 | a random block map. As coverage will not be as complete as with random maps, | |
1246 | this option is disabled by default. | |
1247 | .TP | |
1248 | .BI random_generator \fR=\fPstr | |
1249 | Fio supports the following engines for generating I/O offsets for random I/O: | |
1250 | .RS | |
1251 | .RS | |
1252 | .TP | |
1253 | .B tausworthe | |
1254 | Strong 2^88 cycle random number generator. | |
1255 | .TP | |
1256 | .B lfsr | |
1257 | Linear feedback shift register generator. | |
1258 | .TP | |
1259 | .B tausworthe64 | |
1260 | Strong 64\-bit 2^258 cycle random number generator. | |
1261 | .RE | |
1262 | .P | |
1263 | \fBtausworthe\fR is a strong random number generator, but it requires tracking | |
1264 | on the side if we want to ensure that blocks are only read or written | |
1265 | once. \fBlfsr\fR guarantees that we never generate the same offset twice, and | |
1266 | it's also less computationally expensive. It's not a true random generator, | |
1267 | however, though for I/O purposes it's typically good enough. \fBlfsr\fR only | |
1268 | works with single block sizes, not with workloads that use multiple block | |
1269 | sizes. If used with such a workload, fio may read or write some blocks | |
1270 | multiple times. The default value is \fBtausworthe\fR, unless the required | |
1271 | space exceeds 2^32 blocks. If it does, then \fBtausworthe64\fR is | |
1272 | selected automatically. | |
1273 | .RE | |
1274 | .SS "Block size" | |
1275 | .TP | |
1276 | .BI blocksize \fR=\fPint[,int][,int] "\fR,\fB bs" \fR=\fPint[,int][,int] | |
1277 | The block size in bytes used for I/O units. Default: 4096. A single value | |
338f2db5 | 1278 | applies to reads, writes, and trims. Comma-separated values may be |
523bad63 TK |
1279 | specified for reads, writes, and trims. A value not terminated in a comma |
1280 | applies to subsequent types. Examples: | |
1281 | .RS | |
1282 | .RS | |
1283 | .P | |
1284 | .PD 0 | |
1285 | bs=256k means 256k for reads, writes and trims. | |
1286 | .P | |
1287 | bs=8k,32k means 8k for reads, 32k for writes and trims. | |
1288 | .P | |
1289 | bs=8k,32k, means 8k for reads, 32k for writes, and default for trims. | |
1290 | .P | |
1291 | bs=,8k means default for reads, 8k for writes and trims. | |
1292 | .P | |
1293 | bs=,8k, means default for reads, 8k for writes, and default for trims. | |
1294 | .PD | |
1295 | .RE | |
1296 | .RE | |
1297 | .TP | |
1298 | .BI blocksize_range \fR=\fPirange[,irange][,irange] "\fR,\fB bsrange" \fR=\fPirange[,irange][,irange] | |
1299 | A range of block sizes in bytes for I/O units. The issued I/O unit will | |
1300 | always be a multiple of the minimum size, unless | |
1301 | \fBblocksize_unaligned\fR is set. | |
338f2db5 | 1302 | Comma-separated ranges may be specified for reads, writes, and trims as |
523bad63 TK |
1303 | described in \fBblocksize\fR. Example: |
1304 | .RS | |
1305 | .RS | |
1306 | .P | |
1307 | bsrange=1k\-4k,2k\-8k | |
1308 | .RE | |
1309 | .RE | |
1310 | .TP | |
1311 | .BI bssplit \fR=\fPstr[,str][,str] | |
1312 | Sometimes you want even finer grained control of the block sizes issued, not | |
1313 | just an even split between them. This option allows you to weight various | |
1314 | block sizes, so that you are able to define a specific amount of block sizes | |
1315 | issued. The format for this option is: | |
1316 | .RS | |
1317 | .RS | |
1318 | .P | |
1319 | bssplit=blocksize/percentage:blocksize/percentage | |
1320 | .RE | |
1321 | .P | |
1322 | for as many block sizes as needed. So if you want to define a workload that | |
1323 | has 50% 64k blocks, 10% 4k blocks, and 40% 32k blocks, you would write: | |
1324 | .RS | |
1325 | .P | |
1326 | bssplit=4k/10:64k/50:32k/40 | |
1327 | .RE | |
1328 | .P | |
1329 | Ordering does not matter. If the percentage is left blank, fio will fill in | |
1330 | the remaining values evenly. So a bssplit option like this one: | |
1331 | .RS | |
1332 | .P | |
1333 | bssplit=4k/50:1k/:32k/ | |
1334 | .RE | |
1335 | .P | |
1336 | would have 50% 4k ios, and 25% 1k and 32k ios. The percentages always add up | |
1337 | to 100, if bssplit is given a range that adds up to more, it will error out. | |
1338 | .P | |
338f2db5 | 1339 | Comma-separated values may be specified for reads, writes, and trims as |
523bad63 TK |
1340 | described in \fBblocksize\fR. |
1341 | .P | |
1342 | If you want a workload that has 50% 2k reads and 50% 4k reads, while having | |
1343 | 90% 4k writes and 10% 8k writes, you would specify: | |
1344 | .RS | |
1345 | .P | |
cf04b906 | 1346 | bssplit=2k/50:4k/50,4k/90:8k/10 |
523bad63 | 1347 | .RE |
6a16ece8 JA |
1348 | .P |
1349 | Fio supports defining up to 64 different weights for each data direction. | |
523bad63 TK |
1350 | .RE |
1351 | .TP | |
1352 | .BI blocksize_unaligned "\fR,\fB bs_unaligned" | |
1353 | If set, fio will issue I/O units with any size within | |
1354 | \fBblocksize_range\fR, not just multiples of the minimum size. This | |
1355 | typically won't work with direct I/O, as that normally requires sector | |
1356 | alignment. | |
1357 | .TP | |
1358 | .BI bs_is_seq_rand \fR=\fPbool | |
1359 | If this option is set, fio will use the normal read,write blocksize settings | |
1360 | as sequential,random blocksize settings instead. Any random read or write | |
1361 | will use the WRITE blocksize settings, and any sequential read or write will | |
1362 | use the READ blocksize settings. | |
1363 | .TP | |
1364 | .BI blockalign \fR=\fPint[,int][,int] "\fR,\fB ba" \fR=\fPint[,int][,int] | |
1365 | Boundary to which fio will align random I/O units. Default: | |
1366 | \fBblocksize\fR. Minimum alignment is typically 512b for using direct | |
1367 | I/O, though it usually depends on the hardware block size. This option is | |
1368 | mutually exclusive with using a random map for files, so it will turn off | |
338f2db5 | 1369 | that option. Comma-separated values may be specified for reads, writes, and |
523bad63 TK |
1370 | trims as described in \fBblocksize\fR. |
1371 | .SS "Buffers and memory" | |
1372 | .TP | |
1373 | .BI zero_buffers | |
1374 | Initialize buffers with all zeros. Default: fill buffers with random data. | |
1375 | .TP | |
1376 | .BI refill_buffers | |
1377 | If this option is given, fio will refill the I/O buffers on every | |
1378 | submit. The default is to only fill it at init time and reuse that | |
1379 | data. Only makes sense if zero_buffers isn't specified, naturally. If data | |
1380 | verification is enabled, \fBrefill_buffers\fR is also automatically enabled. | |
1381 | .TP | |
1382 | .BI scramble_buffers \fR=\fPbool | |
1383 | If \fBrefill_buffers\fR is too costly and the target is using data | |
1384 | deduplication, then setting this option will slightly modify the I/O buffer | |
338f2db5 | 1385 | contents to defeat normal de-dupe attempts. This is not enough to defeat |
523bad63 TK |
1386 | more clever block compression attempts, but it will stop naive dedupe of |
1387 | blocks. Default: true. | |
1388 | .TP | |
1389 | .BI buffer_compress_percentage \fR=\fPint | |
72592780 SW |
1390 | If this is set, then fio will attempt to provide I/O buffer content |
1391 | (on WRITEs) that compresses to the specified level. Fio does this by | |
1392 | providing a mix of random data followed by fixed pattern data. The | |
1393 | fixed pattern is either zeros, or the pattern specified by | |
1394 | \fBbuffer_pattern\fR. If the \fBbuffer_pattern\fR option is used, it | |
1395 | might skew the compression ratio slightly. Setting | |
1396 | \fBbuffer_compress_percentage\fR to a value other than 100 will also | |
1397 | enable \fBrefill_buffers\fR in order to reduce the likelihood that | |
1398 | adjacent blocks are so similar that they over compress when seen | |
1399 | together. See \fBbuffer_compress_chunk\fR for how to set a finer or | |
1400 | coarser granularity of the random/fixed data regions. Defaults to unset | |
1401 | i.e., buffer data will not adhere to any compression level. | |
523bad63 TK |
1402 | .TP |
1403 | .BI buffer_compress_chunk \fR=\fPint | |
72592780 SW |
1404 | This setting allows fio to manage how big the random/fixed data region |
1405 | is when using \fBbuffer_compress_percentage\fR. When | |
1406 | \fBbuffer_compress_chunk\fR is set to some non-zero value smaller than the | |
1407 | block size, fio can repeat the random/fixed region throughout the I/O | |
1408 | buffer at the specified interval (which particularly useful when | |
1409 | bigger block sizes are used for a job). When set to 0, fio will use a | |
1410 | chunk size that matches the block size resulting in a single | |
1411 | random/fixed region within the I/O buffer. Defaults to 512. When the | |
1412 | unit is omitted, the value is interpreted in bytes. | |
523bad63 TK |
1413 | .TP |
1414 | .BI buffer_pattern \fR=\fPstr | |
1415 | If set, fio will fill the I/O buffers with this pattern or with the contents | |
1416 | of a file. If not set, the contents of I/O buffers are defined by the other | |
1417 | options related to buffer contents. The setting can be any pattern of bytes, | |
1418 | and can be prefixed with 0x for hex values. It may also be a string, where | |
1419 | the string must then be wrapped with "". Or it may also be a filename, | |
1420 | where the filename must be wrapped with '' in which case the file is | |
1421 | opened and read. Note that not all the file contents will be read if that | |
1422 | would cause the buffers to overflow. So, for example: | |
1423 | .RS | |
1424 | .RS | |
1425 | .P | |
1426 | .PD 0 | |
1427 | buffer_pattern='filename' | |
1428 | .P | |
1429 | or: | |
1430 | .P | |
1431 | buffer_pattern="abcd" | |
1432 | .P | |
1433 | or: | |
1434 | .P | |
1435 | buffer_pattern=\-12 | |
1436 | .P | |
1437 | or: | |
1438 | .P | |
1439 | buffer_pattern=0xdeadface | |
1440 | .PD | |
1441 | .RE | |
1442 | .P | |
1443 | Also you can combine everything together in any order: | |
1444 | .RS | |
1445 | .P | |
1446 | buffer_pattern=0xdeadface"abcd"\-12'filename' | |
1447 | .RE | |
1448 | .RE | |
1449 | .TP | |
1450 | .BI dedupe_percentage \fR=\fPint | |
1451 | If set, fio will generate this percentage of identical buffers when | |
1452 | writing. These buffers will be naturally dedupable. The contents of the | |
1453 | buffers depend on what other buffer compression settings have been set. It's | |
1454 | possible to have the individual buffers either fully compressible, or not at | |
72592780 SW |
1455 | all \-\- this option only controls the distribution of unique buffers. Setting |
1456 | this option will also enable \fBrefill_buffers\fR to prevent every buffer | |
1457 | being identical. | |
523bad63 TK |
1458 | .TP |
1459 | .BI invalidate \fR=\fPbool | |
1460 | Invalidate the buffer/page cache parts of the files to be used prior to | |
1461 | starting I/O if the platform and file type support it. Defaults to true. | |
1462 | This will be ignored if \fBpre_read\fR is also specified for the | |
1463 | same job. | |
1464 | .TP | |
1465 | .BI sync \fR=\fPbool | |
1466 | Use synchronous I/O for buffered writes. For the majority of I/O engines, | |
1467 | this means using O_SYNC. Default: false. | |
1468 | .TP | |
1469 | .BI iomem \fR=\fPstr "\fR,\fP mem" \fR=\fPstr | |
1470 | Fio can use various types of memory as the I/O unit buffer. The allowed | |
1471 | values are: | |
1472 | .RS | |
1473 | .RS | |
1474 | .TP | |
1475 | .B malloc | |
1476 | Use memory from \fBmalloc\fR\|(3) as the buffers. Default memory type. | |
1477 | .TP | |
1478 | .B shm | |
1479 | Use shared memory as the buffers. Allocated through \fBshmget\fR\|(2). | |
1480 | .TP | |
1481 | .B shmhuge | |
1482 | Same as \fBshm\fR, but use huge pages as backing. | |
1483 | .TP | |
1484 | .B mmap | |
1485 | Use \fBmmap\fR\|(2) to allocate buffers. May either be anonymous memory, or can | |
1486 | be file backed if a filename is given after the option. The format | |
1487 | is `mem=mmap:/path/to/file'. | |
1488 | .TP | |
1489 | .B mmaphuge | |
1490 | Use a memory mapped huge file as the buffer backing. Append filename | |
1491 | after mmaphuge, ala `mem=mmaphuge:/hugetlbfs/file'. | |
1492 | .TP | |
1493 | .B mmapshared | |
1494 | Same as \fBmmap\fR, but use a MMAP_SHARED mapping. | |
1495 | .TP | |
1496 | .B cudamalloc | |
1497 | Use GPU memory as the buffers for GPUDirect RDMA benchmark. | |
1498 | The \fBioengine\fR must be \fBrdma\fR. | |
1499 | .RE | |
1500 | .P | |
1501 | The area allocated is a function of the maximum allowed bs size for the job, | |
1502 | multiplied by the I/O depth given. Note that for \fBshmhuge\fR and | |
1503 | \fBmmaphuge\fR to work, the system must have free huge pages allocated. This | |
1504 | can normally be checked and set by reading/writing | |
1505 | `/proc/sys/vm/nr_hugepages' on a Linux system. Fio assumes a huge page | |
1506 | is 4MiB in size. So to calculate the number of huge pages you need for a | |
1507 | given job file, add up the I/O depth of all jobs (normally one unless | |
1508 | \fBiodepth\fR is used) and multiply by the maximum bs set. Then divide | |
1509 | that number by the huge page size. You can see the size of the huge pages in | |
338f2db5 | 1510 | `/proc/meminfo'. If no huge pages are allocated by having a non-zero |
523bad63 TK |
1511 | number in `nr_hugepages', using \fBmmaphuge\fR or \fBshmhuge\fR will fail. Also |
1512 | see \fBhugepage\-size\fR. | |
1513 | .P | |
1514 | \fBmmaphuge\fR also needs to have hugetlbfs mounted and the file location | |
1515 | should point there. So if it's mounted in `/huge', you would use | |
1516 | `mem=mmaphuge:/huge/somefile'. | |
1517 | .RE | |
1518 | .TP | |
1519 | .BI iomem_align \fR=\fPint "\fR,\fP mem_align" \fR=\fPint | |
1520 | This indicates the memory alignment of the I/O memory buffers. Note that | |
1521 | the given alignment is applied to the first I/O unit buffer, if using | |
1522 | \fBiodepth\fR the alignment of the following buffers are given by the | |
1523 | \fBbs\fR used. In other words, if using a \fBbs\fR that is a | |
1524 | multiple of the page sized in the system, all buffers will be aligned to | |
1525 | this value. If using a \fBbs\fR that is not page aligned, the alignment | |
1526 | of subsequent I/O memory buffers is the sum of the \fBiomem_align\fR and | |
1527 | \fBbs\fR used. | |
1528 | .TP | |
1529 | .BI hugepage\-size \fR=\fPint | |
1530 | Defines the size of a huge page. Must at least be equal to the system | |
1531 | setting, see `/proc/meminfo'. Defaults to 4MiB. Should probably | |
1532 | always be a multiple of megabytes, so using `hugepage\-size=Xm' is the | |
338f2db5 | 1533 | preferred way to set this to avoid setting a non-pow-2 bad value. |
523bad63 TK |
1534 | .TP |
1535 | .BI lockmem \fR=\fPint | |
1536 | Pin the specified amount of memory with \fBmlock\fR\|(2). Can be used to | |
1537 | simulate a smaller amount of memory. The amount specified is per worker. | |
1538 | .SS "I/O size" | |
1539 | .TP | |
1540 | .BI size \fR=\fPint | |
1541 | The total size of file I/O for each thread of this job. Fio will run until | |
1542 | this many bytes has been transferred, unless runtime is limited by other options | |
1543 | (such as \fBruntime\fR, for instance, or increased/decreased by \fBio_size\fR). | |
1544 | Fio will divide this size between the available files determined by options | |
1545 | such as \fBnrfiles\fR, \fBfilename\fR, unless \fBfilesize\fR is | |
1546 | specified by the job. If the result of division happens to be 0, the size is | |
1547 | set to the physical size of the given files or devices if they exist. | |
1548 | If this option is not specified, fio will use the full size of the given | |
1549 | files or devices. If the files do not exist, size must be given. It is also | |
1550 | possible to give size as a percentage between 1 and 100. If `size=20%' is | |
1551 | given, fio will use 20% of the full size of the given files or devices. | |
1552 | Can be combined with \fBoffset\fR to constrain the start and end range | |
1553 | that I/O will be done within. | |
1554 | .TP | |
1555 | .BI io_size \fR=\fPint "\fR,\fB io_limit" \fR=\fPint | |
1556 | Normally fio operates within the region set by \fBsize\fR, which means | |
1557 | that the \fBsize\fR option sets both the region and size of I/O to be | |
1558 | performed. Sometimes that is not what you want. With this option, it is | |
1559 | possible to define just the amount of I/O that fio should do. For instance, | |
1560 | if \fBsize\fR is set to 20GiB and \fBio_size\fR is set to 5GiB, fio | |
1561 | will perform I/O within the first 20GiB but exit when 5GiB have been | |
1562 | done. The opposite is also possible \-\- if \fBsize\fR is set to 20GiB, | |
1563 | and \fBio_size\fR is set to 40GiB, then fio will do 40GiB of I/O within | |
f248a525 AD |
1564 | the 0..20GiB region. Value can be set as percentage: \fBio_size\fR=N%. |
1565 | In this case \fBio_size\fR multiplies \fBsize\fR= value. | |
523bad63 TK |
1566 | .TP |
1567 | .BI filesize \fR=\fPirange(int) | |
1568 | Individual file sizes. May be a range, in which case fio will select sizes | |
1569 | for files at random within the given range and limited to \fBsize\fR in | |
1570 | total (if that is given). If not given, each created file is the same size. | |
1571 | This option overrides \fBsize\fR in terms of file size, which means | |
1572 | this value is used as a fixed size or possible range of each file. | |
1573 | .TP | |
1574 | .BI file_append \fR=\fPbool | |
1575 | Perform I/O after the end of the file. Normally fio will operate within the | |
1576 | size of a file. If this option is set, then fio will append to the file | |
1577 | instead. This has identical behavior to setting \fBoffset\fR to the size | |
338f2db5 | 1578 | of a file. This option is ignored on non-regular files. |
523bad63 TK |
1579 | .TP |
1580 | .BI fill_device \fR=\fPbool "\fR,\fB fill_fs" \fR=\fPbool | |
1581 | Sets size to something really large and waits for ENOSPC (no space left on | |
1582 | device) as the terminating condition. Only makes sense with sequential | |
1583 | write. For a read workload, the mount point will be filled first then I/O | |
1584 | started on the result. This option doesn't make sense if operating on a raw | |
1585 | device node, since the size of that is already known by the file system. | |
338f2db5 | 1586 | Additionally, writing beyond end-of-device will not return ENOSPC there. |
523bad63 TK |
1587 | .SS "I/O engine" |
1588 | .TP | |
1589 | .BI ioengine \fR=\fPstr | |
1590 | Defines how the job issues I/O to the file. The following types are defined: | |
1591 | .RS | |
1592 | .RS | |
1593 | .TP | |
1594 | .B sync | |
1595 | Basic \fBread\fR\|(2) or \fBwrite\fR\|(2) | |
1596 | I/O. \fBlseek\fR\|(2) is used to position the I/O location. | |
1597 | See \fBfsync\fR and \fBfdatasync\fR for syncing write I/Os. | |
1598 | .TP | |
1599 | .B psync | |
1600 | Basic \fBpread\fR\|(2) or \fBpwrite\fR\|(2) I/O. Default on | |
1601 | all supported operating systems except for Windows. | |
1602 | .TP | |
1603 | .B vsync | |
1604 | Basic \fBreadv\fR\|(2) or \fBwritev\fR\|(2) I/O. Will emulate | |
1605 | queuing by coalescing adjacent I/Os into a single submission. | |
1606 | .TP | |
1607 | .B pvsync | |
1608 | Basic \fBpreadv\fR\|(2) or \fBpwritev\fR\|(2) I/O. | |
a46c5e01 | 1609 | .TP |
2cafffbe JA |
1610 | .B pvsync2 |
1611 | Basic \fBpreadv2\fR\|(2) or \fBpwritev2\fR\|(2) I/O. | |
1612 | .TP | |
d60e92d1 | 1613 | .B libaio |
523bad63 | 1614 | Linux native asynchronous I/O. Note that Linux may only support |
338f2db5 | 1615 | queued behavior with non-buffered I/O (set `direct=1' or |
523bad63 TK |
1616 | `buffered=0'). |
1617 | This engine defines engine specific options. | |
d60e92d1 AC |
1618 | .TP |
1619 | .B posixaio | |
523bad63 TK |
1620 | POSIX asynchronous I/O using \fBaio_read\fR\|(3) and |
1621 | \fBaio_write\fR\|(3). | |
03e20d68 BC |
1622 | .TP |
1623 | .B solarisaio | |
1624 | Solaris native asynchronous I/O. | |
1625 | .TP | |
1626 | .B windowsaio | |
38f8c318 | 1627 | Windows native asynchronous I/O. Default on Windows. |
d60e92d1 AC |
1628 | .TP |
1629 | .B mmap | |
523bad63 TK |
1630 | File is memory mapped with \fBmmap\fR\|(2) and data copied |
1631 | to/from using \fBmemcpy\fR\|(3). | |
d60e92d1 AC |
1632 | .TP |
1633 | .B splice | |
523bad63 TK |
1634 | \fBsplice\fR\|(2) is used to transfer the data and |
1635 | \fBvmsplice\fR\|(2) to transfer data from user space to the | |
1636 | kernel. | |
d60e92d1 | 1637 | .TP |
d60e92d1 | 1638 | .B sg |
523bad63 TK |
1639 | SCSI generic sg v3 I/O. May either be synchronous using the SG_IO |
1640 | ioctl, or if the target is an sg character device we use | |
1641 | \fBread\fR\|(2) and \fBwrite\fR\|(2) for asynchronous | |
1642 | I/O. Requires \fBfilename\fR option to specify either block or | |
3740cfc8 VF |
1643 | character devices. This engine supports trim operations. The |
1644 | sg engine includes engine specific options. | |
d60e92d1 | 1645 | .TP |
56a19325 DF |
1646 | .B libzbc |
1647 | Synchronous I/O engine for SMR hard-disks using the \fBlibzbc\fR | |
1648 | library. The target can be either an sg character device or | |
1649 | a block device file. This engine supports the zonemode=zbd zone | |
1650 | operations. | |
1651 | .TP | |
d60e92d1 | 1652 | .B null |
523bad63 TK |
1653 | Doesn't transfer any data, just pretends to. This is mainly used to |
1654 | exercise fio itself and for debugging/testing purposes. | |
d60e92d1 AC |
1655 | .TP |
1656 | .B net | |
523bad63 TK |
1657 | Transfer over the network to given `host:port'. Depending on the |
1658 | \fBprotocol\fR used, the \fBhostname\fR, \fBport\fR, | |
1659 | \fBlisten\fR and \fBfilename\fR options are used to specify | |
1660 | what sort of connection to make, while the \fBprotocol\fR option | |
1661 | determines which protocol will be used. This engine defines engine | |
1662 | specific options. | |
d60e92d1 AC |
1663 | .TP |
1664 | .B netsplice | |
523bad63 TK |
1665 | Like \fBnet\fR, but uses \fBsplice\fR\|(2) and |
1666 | \fBvmsplice\fR\|(2) to map data and send/receive. | |
1667 | This engine defines engine specific options. | |
d60e92d1 | 1668 | .TP |
53aec0a4 | 1669 | .B cpuio |
523bad63 TK |
1670 | Doesn't transfer any data, but burns CPU cycles according to the |
1671 | \fBcpuload\fR and \fBcpuchunks\fR options. Setting | |
1672 | \fBcpuload\fR\=85 will cause that job to do nothing but burn 85% | |
1673 | of the CPU. In case of SMP machines, use `numjobs=<nr_of_cpu>' | |
1674 | to get desired CPU usage, as the cpuload only loads a | |
1675 | single CPU at the desired rate. A job never finishes unless there is | |
338f2db5 | 1676 | at least one non-cpuio job. |
d60e92d1 | 1677 | .TP |
21b8aee8 | 1678 | .B rdma |
523bad63 TK |
1679 | The RDMA I/O engine supports both RDMA memory semantics |
1680 | (RDMA_WRITE/RDMA_READ) and channel semantics (Send/Recv) for the | |
609ac152 SB |
1681 | InfiniBand, RoCE and iWARP protocols. This engine defines engine |
1682 | specific options. | |
d54fce84 DM |
1683 | .TP |
1684 | .B falloc | |
523bad63 TK |
1685 | I/O engine that does regular fallocate to simulate data transfer as |
1686 | fio ioengine. | |
1687 | .RS | |
1688 | .P | |
1689 | .PD 0 | |
1690 | DDIR_READ does fallocate(,mode = FALLOC_FL_KEEP_SIZE,). | |
1691 | .P | |
1692 | DIR_WRITE does fallocate(,mode = 0). | |
1693 | .P | |
1694 | DDIR_TRIM does fallocate(,mode = FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE). | |
1695 | .PD | |
1696 | .RE | |
1697 | .TP | |
1698 | .B ftruncate | |
1699 | I/O engine that sends \fBftruncate\fR\|(2) operations in response | |
1700 | to write (DDIR_WRITE) events. Each ftruncate issued sets the file's | |
1701 | size to the current block offset. \fBblocksize\fR is ignored. | |
d54fce84 DM |
1702 | .TP |
1703 | .B e4defrag | |
523bad63 TK |
1704 | I/O engine that does regular EXT4_IOC_MOVE_EXT ioctls to simulate |
1705 | defragment activity in request to DDIR_WRITE event. | |
0d978694 | 1706 | .TP |
d5f9b0ea IF |
1707 | .B rados |
1708 | I/O engine supporting direct access to Ceph Reliable Autonomic Distributed | |
1709 | Object Store (RADOS) via librados. This ioengine defines engine specific | |
1710 | options. | |
1711 | .TP | |
0d978694 | 1712 | .B rbd |
523bad63 TK |
1713 | I/O engine supporting direct access to Ceph Rados Block Devices |
1714 | (RBD) via librbd without the need to use the kernel rbd driver. This | |
1715 | ioengine defines engine specific options. | |
a7c386f4 | 1716 | .TP |
c2f6a13d LMB |
1717 | .B http |
1718 | I/O engine supporting GET/PUT requests over HTTP(S) with libcurl to | |
1719 | a WebDAV or S3 endpoint. This ioengine defines engine specific options. | |
1720 | ||
1721 | This engine only supports direct IO of iodepth=1; you need to scale this | |
1722 | via numjobs. blocksize defines the size of the objects to be created. | |
1723 | ||
1724 | TRIM is translated to object deletion. | |
1725 | .TP | |
a7c386f4 | 1726 | .B gfapi |
523bad63 TK |
1727 | Using GlusterFS libgfapi sync interface to direct access to |
1728 | GlusterFS volumes without having to go through FUSE. This ioengine | |
1729 | defines engine specific options. | |
cc47f094 | 1730 | .TP |
1731 | .B gfapi_async | |
523bad63 TK |
1732 | Using GlusterFS libgfapi async interface to direct access to |
1733 | GlusterFS volumes without having to go through FUSE. This ioengine | |
1734 | defines engine specific options. | |
1b10477b | 1735 | .TP |
b74e419e | 1736 | .B libhdfs |
523bad63 TK |
1737 | Read and write through Hadoop (HDFS). The \fBfilename\fR option |
1738 | is used to specify host,port of the hdfs name\-node to connect. This | |
1739 | engine interprets offsets a little differently. In HDFS, files once | |
1740 | created cannot be modified so random writes are not possible. To | |
1741 | imitate this the libhdfs engine expects a bunch of small files to be | |
1742 | created over HDFS and will randomly pick a file from them | |
1743 | based on the offset generated by fio backend (see the example | |
1744 | job file to create such files, use `rw=write' option). Please | |
1745 | note, it may be necessary to set environment variables to work | |
1746 | with HDFS/libhdfs properly. Each job uses its own connection to | |
1747 | HDFS. | |
65fa28ca DE |
1748 | .TP |
1749 | .B mtd | |
523bad63 TK |
1750 | Read, write and erase an MTD character device (e.g., |
1751 | `/dev/mtd0'). Discards are treated as erases. Depending on the | |
1752 | underlying device type, the I/O may have to go in a certain pattern, | |
1753 | e.g., on NAND, writing sequentially to erase blocks and discarding | |
1754 | before overwriting. The \fBtrimwrite\fR mode works well for this | |
65fa28ca | 1755 | constraint. |
5c4ef02e JA |
1756 | .TP |
1757 | .B pmemblk | |
523bad63 | 1758 | Read and write using filesystem DAX to a file on a filesystem |
363a5f65 | 1759 | mounted with DAX on a persistent memory device through the PMDK |
523bad63 | 1760 | libpmemblk library. |
104ee4de | 1761 | .TP |
523bad63 TK |
1762 | .B dev\-dax |
1763 | Read and write using device DAX to a persistent memory device (e.g., | |
363a5f65 | 1764 | /dev/dax0.0) through the PMDK libpmem library. |
d60e92d1 | 1765 | .TP |
523bad63 TK |
1766 | .B external |
1767 | Prefix to specify loading an external I/O engine object file. Append | |
1768 | the engine filename, e.g. `ioengine=external:/tmp/foo.o' to load | |
d243fd6d TK |
1769 | ioengine `foo.o' in `/tmp'. The path can be either |
1770 | absolute or relative. See `engines/skeleton_external.c' in the fio source for | |
1771 | details of writing an external I/O engine. | |
1216cc5a JB |
1772 | .TP |
1773 | .B filecreate | |
b71968b1 SW |
1774 | Simply create the files and do no I/O to them. You still need to set |
1775 | \fBfilesize\fR so that all the accounting still occurs, but no actual I/O will be | |
1776 | done other than creating the file. | |
ae0db592 | 1777 | .TP |
73ccd14e SF |
1778 | .B filestat |
1779 | Simply do stat() and do no I/O to the file. You need to set 'filesize' | |
1780 | and 'nrfiles', so that files will be created. | |
1781 | This engine is to measure file lookup and meta data access. | |
1782 | .TP | |
ae0db592 TI |
1783 | .B libpmem |
1784 | Read and write using mmap I/O to a file on a filesystem | |
363a5f65 | 1785 | mounted with DAX on a persistent memory device through the PMDK |
ae0db592 | 1786 | libpmem library. |
07751e10 JA |
1787 | .TP |
1788 | .B ime_psync | |
1789 | Synchronous read and write using DDN's Infinite Memory Engine (IME). This | |
1790 | engine is very basic and issues calls to IME whenever an IO is queued. | |
1791 | .TP | |
1792 | .B ime_psyncv | |
1793 | Synchronous read and write using DDN's Infinite Memory Engine (IME). This | |
1794 | engine uses iovecs and will try to stack as much IOs as possible (if the IOs | |
1795 | are "contiguous" and the IO depth is not exceeded) before issuing a call to IME. | |
1796 | .TP | |
1797 | .B ime_aio | |
1798 | Asynchronous read and write using DDN's Infinite Memory Engine (IME). This | |
1799 | engine will try to stack as much IOs as possible by creating requests for IME. | |
1800 | FIO will then decide when to commit these requests. | |
247ef2aa KZ |
1801 | .TP |
1802 | .B libiscsi | |
1803 | Read and write iscsi lun with libiscsi. | |
d643a1e2 RJ |
1804 | .TP |
1805 | .B nbd | |
1806 | Synchronous read and write a Network Block Device (NBD). | |
523bad63 TK |
1807 | .SS "I/O engine specific parameters" |
1808 | In addition, there are some parameters which are only valid when a specific | |
1809 | \fBioengine\fR is in use. These are used identically to normal parameters, | |
1810 | with the caveat that when used on the command line, they must come after the | |
1811 | \fBioengine\fR that defines them is selected. | |
d60e92d1 | 1812 | .TP |
b2a432bf PC |
1813 | .BI (io_uring, libaio)cmdprio_percentage \fR=\fPint |
1814 | Set the percentage of I/O that will be issued with higher priority by setting | |
1815 | the priority bit. Non-read I/O is likely unaffected by ``cmdprio_percentage``. | |
1816 | This option cannot be used with the `prio` or `prioclass` options. For this | |
1817 | option to set the priority bit properly, NCQ priority must be supported and | |
7896180a VF |
1818 | enabled and `direct=1' option must be used. fio must also be run as the root |
1819 | user. | |
029b42ac JA |
1820 | .TP |
1821 | .BI (io_uring)fixedbufs | |
1822 | If fio is asked to do direct IO, then Linux will map pages for each IO call, and | |
1823 | release them when IO is done. If this option is set, the pages are pre-mapped | |
1824 | before IO is started. This eliminates the need to map and release for each IO. | |
1825 | This is more efficient, and reduces the IO latency as well. | |
1826 | .TP | |
b2a432bf PC |
1827 | .BI (io_uring)hipri |
1828 | If this option is set, fio will attempt to use polled IO completions. Normal IO | |
1829 | completions generate interrupts to signal the completion of IO, polled | |
1830 | completions do not. Hence they are require active reaping by the application. | |
1831 | The benefits are more efficient IO for high IOPS scenarios, and lower latencies | |
1832 | for low queue depth IO. | |
1833 | .TP | |
5ffd5626 JA |
1834 | .BI (io_uring)registerfiles |
1835 | With this option, fio registers the set of files being used with the kernel. | |
1836 | This avoids the overhead of managing file counts in the kernel, making the | |
1837 | submission and completion part more lightweight. Required for the below | |
1838 | sqthread_poll option. | |
1839 | .TP | |
029b42ac JA |
1840 | .BI (io_uring)sqthread_poll |
1841 | Normally fio will submit IO by issuing a system call to notify the kernel of | |
1842 | available items in the SQ ring. If this option is set, the act of submitting IO | |
1843 | will be done by a polling thread in the kernel. This frees up cycles for fio, at | |
1844 | the cost of using more CPU in the system. | |
1845 | .TP | |
1846 | .BI (io_uring)sqthread_poll_cpu | |
1847 | When `sqthread_poll` is set, this option provides a way to define which CPU | |
1848 | should be used for the polling thread. | |
1849 | .TP | |
523bad63 TK |
1850 | .BI (libaio)userspace_reap |
1851 | Normally, with the libaio engine in use, fio will use the | |
1852 | \fBio_getevents\fR\|(3) system call to reap newly returned events. With | |
338f2db5 | 1853 | this flag turned on, the AIO ring will be read directly from user-space to |
523bad63 TK |
1854 | reap events. The reaping mode is only enabled when polling for a minimum of |
1855 | 0 events (e.g. when `iodepth_batch_complete=0'). | |
3ce9dcaf | 1856 | .TP |
523bad63 TK |
1857 | .BI (pvsync2)hipri |
1858 | Set RWF_HIPRI on I/O, indicating to the kernel that it's of higher priority | |
1859 | than normal. | |
82407585 | 1860 | .TP |
523bad63 TK |
1861 | .BI (pvsync2)hipri_percentage |
1862 | When hipri is set this determines the probability of a pvsync2 I/O being high | |
1863 | priority. The default is 100%. | |
d60e92d1 | 1864 | .TP |
7d42e66e KK |
1865 | .BI (pvsync2,libaio,io_uring)nowait |
1866 | By default if a request cannot be executed immediately (e.g. resource starvation, | |
1867 | waiting on locks) it is queued and the initiating process will be blocked until | |
1868 | the required resource becomes free. | |
1869 | This option sets the RWF_NOWAIT flag (supported from the 4.14 Linux kernel) and | |
1870 | the call will return instantly with EAGAIN or a partial result rather than waiting. | |
1871 | ||
1872 | It is useful to also use \fBignore_error\fR=EAGAIN when using this option. | |
1873 | Note: glibc 2.27, 2.28 have a bug in syscall wrappers preadv2, pwritev2. | |
1874 | They return EOPNOTSUP instead of EAGAIN. | |
1875 | ||
1876 | For cached I/O, using this option usually means a request operates only with | |
1877 | cached data. Currently the RWF_NOWAIT flag does not supported for cached write. | |
1878 | For direct I/O, requests will only succeed if cache invalidation isn't required, | |
1879 | file blocks are fully allocated and the disk request could be issued immediately. | |
1880 | .TP | |
523bad63 TK |
1881 | .BI (cpuio)cpuload \fR=\fPint |
1882 | Attempt to use the specified percentage of CPU cycles. This is a mandatory | |
1883 | option when using cpuio I/O engine. | |
997b5680 | 1884 | .TP |
523bad63 TK |
1885 | .BI (cpuio)cpuchunks \fR=\fPint |
1886 | Split the load into cycles of the given time. In microseconds. | |
1ad01bd1 | 1887 | .TP |
523bad63 TK |
1888 | .BI (cpuio)exit_on_io_done \fR=\fPbool |
1889 | Detect when I/O threads are done, then exit. | |
d60e92d1 | 1890 | .TP |
523bad63 TK |
1891 | .BI (libhdfs)namenode \fR=\fPstr |
1892 | The hostname or IP address of a HDFS cluster namenode to contact. | |
d01612f3 | 1893 | .TP |
523bad63 TK |
1894 | .BI (libhdfs)port |
1895 | The listening port of the HFDS cluster namenode. | |
d60e92d1 | 1896 | .TP |
523bad63 TK |
1897 | .BI (netsplice,net)port |
1898 | The TCP or UDP port to bind to or connect to. If this is used with | |
1899 | \fBnumjobs\fR to spawn multiple instances of the same job type, then | |
1900 | this will be the starting port number since fio will use a range of | |
1901 | ports. | |
d60e92d1 | 1902 | .TP |
609ac152 SB |
1903 | .BI (rdma)port |
1904 | The port to use for RDMA-CM communication. This should be the same | |
1905 | value on the client and the server side. | |
1906 | .TP | |
1907 | .BI (netsplice,net, rdma)hostname \fR=\fPstr | |
1908 | The hostname or IP address to use for TCP, UDP or RDMA-CM based I/O. | |
1909 | If the job is a TCP listener or UDP reader, the hostname is not used | |
1910 | and must be omitted unless it is a valid UDP multicast address. | |
591e9e06 | 1911 | .TP |
523bad63 TK |
1912 | .BI (netsplice,net)interface \fR=\fPstr |
1913 | The IP address of the network interface used to send or receive UDP | |
1914 | multicast. | |
ddf24e42 | 1915 | .TP |
523bad63 TK |
1916 | .BI (netsplice,net)ttl \fR=\fPint |
1917 | Time\-to\-live value for outgoing UDP multicast packets. Default: 1. | |
d60e92d1 | 1918 | .TP |
523bad63 TK |
1919 | .BI (netsplice,net)nodelay \fR=\fPbool |
1920 | Set TCP_NODELAY on TCP connections. | |
fa769d44 | 1921 | .TP |
523bad63 TK |
1922 | .BI (netsplice,net)protocol \fR=\fPstr "\fR,\fP proto" \fR=\fPstr |
1923 | The network protocol to use. Accepted values are: | |
1924 | .RS | |
e76b1da4 JA |
1925 | .RS |
1926 | .TP | |
523bad63 TK |
1927 | .B tcp |
1928 | Transmission control protocol. | |
e76b1da4 | 1929 | .TP |
523bad63 TK |
1930 | .B tcpv6 |
1931 | Transmission control protocol V6. | |
e76b1da4 | 1932 | .TP |
523bad63 TK |
1933 | .B udp |
1934 | User datagram protocol. | |
1935 | .TP | |
1936 | .B udpv6 | |
1937 | User datagram protocol V6. | |
e76b1da4 | 1938 | .TP |
523bad63 TK |
1939 | .B unix |
1940 | UNIX domain socket. | |
e76b1da4 JA |
1941 | .RE |
1942 | .P | |
523bad63 TK |
1943 | When the protocol is TCP or UDP, the port must also be given, as well as the |
1944 | hostname if the job is a TCP listener or UDP reader. For unix sockets, the | |
1945 | normal \fBfilename\fR option should be used and the port is invalid. | |
1946 | .RE | |
1947 | .TP | |
1948 | .BI (netsplice,net)listen | |
1949 | For TCP network connections, tell fio to listen for incoming connections | |
1950 | rather than initiating an outgoing connection. The \fBhostname\fR must | |
1951 | be omitted if this option is used. | |
1952 | .TP | |
1953 | .BI (netsplice,net)pingpong | |
1954 | Normally a network writer will just continue writing data, and a network | |
1955 | reader will just consume packages. If `pingpong=1' is set, a writer will | |
1956 | send its normal payload to the reader, then wait for the reader to send the | |
1957 | same payload back. This allows fio to measure network latencies. The | |
1958 | submission and completion latencies then measure local time spent sending or | |
1959 | receiving, and the completion latency measures how long it took for the | |
1960 | other end to receive and send back. For UDP multicast traffic | |
1961 | `pingpong=1' should only be set for a single reader when multiple readers | |
1962 | are listening to the same address. | |
1963 | .TP | |
1964 | .BI (netsplice,net)window_size \fR=\fPint | |
1965 | Set the desired socket buffer size for the connection. | |
e76b1da4 | 1966 | .TP |
523bad63 TK |
1967 | .BI (netsplice,net)mss \fR=\fPint |
1968 | Set the TCP maximum segment size (TCP_MAXSEG). | |
d60e92d1 | 1969 | .TP |
523bad63 TK |
1970 | .BI (e4defrag)donorname \fR=\fPstr |
1971 | File will be used as a block donor (swap extents between files). | |
d60e92d1 | 1972 | .TP |
523bad63 TK |
1973 | .BI (e4defrag)inplace \fR=\fPint |
1974 | Configure donor file blocks allocation strategy: | |
1975 | .RS | |
1976 | .RS | |
d60e92d1 | 1977 | .TP |
523bad63 TK |
1978 | .B 0 |
1979 | Default. Preallocate donor's file on init. | |
d60e92d1 | 1980 | .TP |
523bad63 TK |
1981 | .B 1 |
1982 | Allocate space immediately inside defragment event, and free right | |
1983 | after event. | |
1984 | .RE | |
1985 | .RE | |
d60e92d1 | 1986 | .TP |
d5f9b0ea | 1987 | .BI (rbd,rados)clustername \fR=\fPstr |
523bad63 | 1988 | Specifies the name of the Ceph cluster. |
92d42d69 | 1989 | .TP |
523bad63 TK |
1990 | .BI (rbd)rbdname \fR=\fPstr |
1991 | Specifies the name of the RBD. | |
92d42d69 | 1992 | .TP |
d5f9b0ea IF |
1993 | .BI (rbd,rados)pool \fR=\fPstr |
1994 | Specifies the name of the Ceph pool containing RBD or RADOS data. | |
92d42d69 | 1995 | .TP |
d5f9b0ea | 1996 | .BI (rbd,rados)clientname \fR=\fPstr |
523bad63 TK |
1997 | Specifies the username (without the 'client.' prefix) used to access the |
1998 | Ceph cluster. If the \fBclustername\fR is specified, the \fBclientname\fR shall be | |
1999 | the full *type.id* string. If no type. prefix is given, fio will add 'client.' | |
2000 | by default. | |
92d42d69 | 2001 | .TP |
d5f9b0ea IF |
2002 | .BI (rbd,rados)busy_poll \fR=\fPbool |
2003 | Poll store instead of waiting for completion. Usually this provides better | |
2004 | throughput at cost of higher(up to 100%) CPU utilization. | |
2005 | .TP | |
c2f6a13d LMB |
2006 | .BI (http)http_host \fR=\fPstr |
2007 | Hostname to connect to. For S3, this could be the bucket name. Default | |
2008 | is \fBlocalhost\fR | |
2009 | .TP | |
2010 | .BI (http)http_user \fR=\fPstr | |
2011 | Username for HTTP authentication. | |
2012 | .TP | |
2013 | .BI (http)http_pass \fR=\fPstr | |
2014 | Password for HTTP authentication. | |
2015 | .TP | |
09fd2966 LMB |
2016 | .BI (http)https \fR=\fPstr |
2017 | Whether to use HTTPS instead of plain HTTP. \fRon\fP enables HTTPS; | |
2018 | \fRinsecure\fP will enable HTTPS, but disable SSL peer verification (use | |
2019 | with caution!). Default is \fBoff\fR. | |
c2f6a13d | 2020 | .TP |
09fd2966 LMB |
2021 | .BI (http)http_mode \fR=\fPstr |
2022 | Which HTTP access mode to use: webdav, swift, or s3. Default is | |
2023 | \fBwebdav\fR. | |
c2f6a13d LMB |
2024 | .TP |
2025 | .BI (http)http_s3_region \fR=\fPstr | |
2026 | The S3 region/zone to include in the request. Default is \fBus-east-1\fR. | |
2027 | .TP | |
2028 | .BI (http)http_s3_key \fR=\fPstr | |
2029 | The S3 secret key. | |
2030 | .TP | |
2031 | .BI (http)http_s3_keyid \fR=\fPstr | |
2032 | The S3 key/access id. | |
2033 | .TP | |
09fd2966 LMB |
2034 | .BI (http)http_swift_auth_token \fR=\fPstr |
2035 | The Swift auth token. See the example configuration file on how to | |
2036 | retrieve this. | |
2037 | .TP | |
c2f6a13d LMB |
2038 | .BI (http)http_verbose \fR=\fPint |
2039 | Enable verbose requests from libcurl. Useful for debugging. 1 turns on | |
2040 | verbose logging from libcurl, 2 additionally enables HTTP IO tracing. | |
2041 | Default is \fB0\fR | |
2042 | .TP | |
523bad63 TK |
2043 | .BI (mtd)skip_bad \fR=\fPbool |
2044 | Skip operations against known bad blocks. | |
8116fd24 | 2045 | .TP |
523bad63 TK |
2046 | .BI (libhdfs)hdfsdirectory |
2047 | libhdfs will create chunk in this HDFS directory. | |
e0a04ac1 | 2048 | .TP |
523bad63 TK |
2049 | .BI (libhdfs)chunk_size |
2050 | The size of the chunk to use for each file. | |
609ac152 SB |
2051 | .TP |
2052 | .BI (rdma)verb \fR=\fPstr | |
2053 | The RDMA verb to use on this side of the RDMA ioengine | |
2054 | connection. Valid values are write, read, send and recv. These | |
2055 | correspond to the equivalent RDMA verbs (e.g. write = rdma_write | |
2056 | etc.). Note that this only needs to be specified on the client side of | |
2057 | the connection. See the examples folder. | |
2058 | .TP | |
2059 | .BI (rdma)bindname \fR=\fPstr | |
2060 | The name to use to bind the local RDMA-CM connection to a local RDMA | |
2061 | device. This could be a hostname or an IPv4 or IPv6 address. On the | |
2062 | server side this will be passed into the rdma_bind_addr() function and | |
2063 | on the client site it will be used in the rdma_resolve_add() | |
2064 | function. This can be useful when multiple paths exist between the | |
2065 | client and the server or in certain loopback configurations. | |
52b81b7c | 2066 | .TP |
93a13ba5 TK |
2067 | .BI (filestat)stat_type \fR=\fPstr |
2068 | Specify stat system call type to measure lookup/getattr performance. | |
2069 | Default is \fBstat\fR for \fBstat\fR\|(2). | |
c446eff0 | 2070 | .TP |
52b81b7c KD |
2071 | .BI (sg)readfua \fR=\fPbool |
2072 | With readfua option set to 1, read operations include the force | |
2073 | unit access (fua) flag. Default: 0. | |
2074 | .TP | |
2075 | .BI (sg)writefua \fR=\fPbool | |
2076 | With writefua option set to 1, write operations include the force | |
2077 | unit access (fua) flag. Default: 0. | |
2c3a9150 VF |
2078 | .TP |
2079 | .BI (sg)sg_write_mode \fR=\fPstr | |
2080 | Specify the type of write commands to issue. This option can take three | |
2081 | values: | |
2082 | .RS | |
2083 | .RS | |
2084 | .TP | |
2085 | .B write (default) | |
2086 | Write opcodes are issued as usual | |
2087 | .TP | |
2088 | .B verify | |
2089 | Issue WRITE AND VERIFY commands. The BYTCHK bit is set to 0. This | |
2090 | directs the device to carry out a medium verification with no data | |
2091 | comparison. The writefua option is ignored with this selection. | |
2092 | .TP | |
2093 | .B same | |
2094 | Issue WRITE SAME commands. This transfers a single block to the device | |
2095 | and writes this same block of data to a contiguous sequence of LBAs | |
2096 | beginning at the specified offset. fio's block size parameter | |
2097 | specifies the amount of data written with each command. However, the | |
2098 | amount of data actually transferred to the device is equal to the | |
2099 | device's block (sector) size. For a device with 512 byte sectors, | |
2100 | blocksize=8k will write 16 sectors with each command. fio will still | |
2101 | generate 8k of data for each command butonly the first 512 bytes will | |
2102 | be used and transferred to the device. The writefua option is ignored | |
2103 | with this selection. | |
f2d6de5d RJ |
2104 | .RE |
2105 | .RE | |
2106 | .TP | |
2107 | .BI (nbd)uri \fR=\fPstr | |
2108 | Specify the NBD URI of the server to test. | |
2109 | The string is a standard NBD URI (see | |
2110 | \fIhttps://github.com/NetworkBlockDevice/nbd/tree/master/doc\fR). | |
2111 | Example URIs: | |
2112 | .RS | |
2113 | .RS | |
2114 | .TP | |
2115 | \fInbd://localhost:10809\fR | |
2116 | .TP | |
2117 | \fInbd+unix:///?socket=/tmp/socket\fR | |
2118 | .TP | |
2119 | \fInbds://tlshost/exportname\fR | |
2c3a9150 | 2120 | |
523bad63 TK |
2121 | .SS "I/O depth" |
2122 | .TP | |
2123 | .BI iodepth \fR=\fPint | |
2124 | Number of I/O units to keep in flight against the file. Note that | |
2125 | increasing \fBiodepth\fR beyond 1 will not affect synchronous ioengines (except | |
2126 | for small degrees when \fBverify_async\fR is in use). Even async | |
2127 | engines may impose OS restrictions causing the desired depth not to be | |
2128 | achieved. This may happen on Linux when using libaio and not setting | |
2129 | `direct=1', since buffered I/O is not async on that OS. Keep an | |
2130 | eye on the I/O depth distribution in the fio output to verify that the | |
2131 | achieved depth is as expected. Default: 1. | |
2132 | .TP | |
2133 | .BI iodepth_batch_submit \fR=\fPint "\fR,\fP iodepth_batch" \fR=\fPint | |
2134 | This defines how many pieces of I/O to submit at once. It defaults to 1 | |
2135 | which means that we submit each I/O as soon as it is available, but can be | |
2136 | raised to submit bigger batches of I/O at the time. If it is set to 0 the | |
2137 | \fBiodepth\fR value will be used. | |
2138 | .TP | |
2139 | .BI iodepth_batch_complete_min \fR=\fPint "\fR,\fP iodepth_batch_complete" \fR=\fPint | |
2140 | This defines how many pieces of I/O to retrieve at once. It defaults to 1 | |
2141 | which means that we'll ask for a minimum of 1 I/O in the retrieval process | |
2142 | from the kernel. The I/O retrieval will go on until we hit the limit set by | |
2143 | \fBiodepth_low\fR. If this variable is set to 0, then fio will always | |
2144 | check for completed events before queuing more I/O. This helps reduce I/O | |
2145 | latency, at the cost of more retrieval system calls. | |
2146 | .TP | |
2147 | .BI iodepth_batch_complete_max \fR=\fPint | |
2148 | This defines maximum pieces of I/O to retrieve at once. This variable should | |
2149 | be used along with \fBiodepth_batch_complete_min\fR=\fIint\fR variable, | |
2150 | specifying the range of min and max amount of I/O which should be | |
2151 | retrieved. By default it is equal to \fBiodepth_batch_complete_min\fR | |
2152 | value. Example #1: | |
e0a04ac1 | 2153 | .RS |
e0a04ac1 | 2154 | .RS |
e0a04ac1 | 2155 | .P |
523bad63 TK |
2156 | .PD 0 |
2157 | iodepth_batch_complete_min=1 | |
e0a04ac1 | 2158 | .P |
523bad63 TK |
2159 | iodepth_batch_complete_max=<iodepth> |
2160 | .PD | |
e0a04ac1 JA |
2161 | .RE |
2162 | .P | |
523bad63 TK |
2163 | which means that we will retrieve at least 1 I/O and up to the whole |
2164 | submitted queue depth. If none of I/O has been completed yet, we will wait. | |
2165 | Example #2: | |
e8b1961d | 2166 | .RS |
523bad63 TK |
2167 | .P |
2168 | .PD 0 | |
2169 | iodepth_batch_complete_min=0 | |
2170 | .P | |
2171 | iodepth_batch_complete_max=<iodepth> | |
2172 | .PD | |
e8b1961d JA |
2173 | .RE |
2174 | .P | |
523bad63 TK |
2175 | which means that we can retrieve up to the whole submitted queue depth, but |
2176 | if none of I/O has been completed yet, we will NOT wait and immediately exit | |
2177 | the system call. In this example we simply do polling. | |
2178 | .RE | |
e8b1961d | 2179 | .TP |
523bad63 TK |
2180 | .BI iodepth_low \fR=\fPint |
2181 | The low water mark indicating when to start filling the queue | |
2182 | again. Defaults to the same as \fBiodepth\fR, meaning that fio will | |
2183 | attempt to keep the queue full at all times. If \fBiodepth\fR is set to | |
2184 | e.g. 16 and \fBiodepth_low\fR is set to 4, then after fio has filled the queue of | |
2185 | 16 requests, it will let the depth drain down to 4 before starting to fill | |
2186 | it again. | |
d60e92d1 | 2187 | .TP |
523bad63 TK |
2188 | .BI serialize_overlap \fR=\fPbool |
2189 | Serialize in-flight I/Os that might otherwise cause or suffer from data races. | |
2190 | When two or more I/Os are submitted simultaneously, there is no guarantee that | |
2191 | the I/Os will be processed or completed in the submitted order. Further, if | |
2192 | two or more of those I/Os are writes, any overlapping region between them can | |
2193 | become indeterminate/undefined on certain storage. These issues can cause | |
2194 | verification to fail erratically when at least one of the racing I/Os is | |
2195 | changing data and the overlapping region has a non-zero size. Setting | |
2196 | \fBserialize_overlap\fR tells fio to avoid provoking this behavior by explicitly | |
2197 | serializing in-flight I/Os that have a non-zero overlap. Note that setting | |
2198 | this option can reduce both performance and the \fBiodepth\fR achieved. | |
3d6a6f04 VF |
2199 | .RS |
2200 | .P | |
2201 | This option only applies to I/Os issued for a single job except when it is | |
2202 | enabled along with \fBio_submit_mode\fR=offload. In offload mode, fio | |
2203 | will check for overlap among all I/Os submitted by offload jobs with \fBserialize_overlap\fR | |
307f2246 | 2204 | enabled. |
3d6a6f04 VF |
2205 | .P |
2206 | Default: false. | |
2207 | .RE | |
d60e92d1 | 2208 | .TP |
523bad63 TK |
2209 | .BI io_submit_mode \fR=\fPstr |
2210 | This option controls how fio submits the I/O to the I/O engine. The default | |
2211 | is `inline', which means that the fio job threads submit and reap I/O | |
2212 | directly. If set to `offload', the job threads will offload I/O submission | |
2213 | to a dedicated pool of I/O threads. This requires some coordination and thus | |
2214 | has a bit of extra overhead, especially for lower queue depth I/O where it | |
2215 | can increase latencies. The benefit is that fio can manage submission rates | |
2216 | independently of the device completion rates. This avoids skewed latency | |
2217 | reporting if I/O gets backed up on the device side (the coordinated omission | |
abfd235a | 2218 | problem). Note that this option cannot reliably be used with async IO engines. |
523bad63 | 2219 | .SS "I/O rate" |
d60e92d1 | 2220 | .TP |
523bad63 TK |
2221 | .BI thinktime \fR=\fPtime |
2222 | Stall the job for the specified period of time after an I/O has completed before issuing the | |
2223 | next. May be used to simulate processing being done by an application. | |
2224 | When the unit is omitted, the value is interpreted in microseconds. See | |
2225 | \fBthinktime_blocks\fR and \fBthinktime_spin\fR. | |
d60e92d1 | 2226 | .TP |
523bad63 | 2227 | .BI thinktime_spin \fR=\fPtime |
338f2db5 | 2228 | Only valid if \fBthinktime\fR is set - pretend to spend CPU time doing |
523bad63 TK |
2229 | something with the data received, before falling back to sleeping for the |
2230 | rest of the period specified by \fBthinktime\fR. When the unit is | |
2231 | omitted, the value is interpreted in microseconds. | |
d60e92d1 AC |
2232 | .TP |
2233 | .BI thinktime_blocks \fR=\fPint | |
338f2db5 | 2234 | Only valid if \fBthinktime\fR is set - control how many blocks to issue, |
523bad63 TK |
2235 | before waiting \fBthinktime\fR usecs. If not set, defaults to 1 which will make |
2236 | fio wait \fBthinktime\fR usecs after every block. This effectively makes any | |
2237 | queue depth setting redundant, since no more than 1 I/O will be queued | |
2238 | before we have to complete it and do our \fBthinktime\fR. In other words, this | |
2239 | setting effectively caps the queue depth if the latter is larger. | |
d60e92d1 | 2240 | .TP |
6d500c2e | 2241 | .BI rate \fR=\fPint[,int][,int] |
523bad63 | 2242 | Cap the bandwidth used by this job. The number is in bytes/sec, the normal |
338f2db5 | 2243 | suffix rules apply. Comma-separated values may be specified for reads, |
523bad63 TK |
2244 | writes, and trims as described in \fBblocksize\fR. |
2245 | .RS | |
2246 | .P | |
2247 | For example, using `rate=1m,500k' would limit reads to 1MiB/sec and writes to | |
2248 | 500KiB/sec. Capping only reads or writes can be done with `rate=,500k' or | |
2249 | `rate=500k,' where the former will only limit writes (to 500KiB/sec) and the | |
2250 | latter will only limit reads. | |
2251 | .RE | |
d60e92d1 | 2252 | .TP |
6d500c2e | 2253 | .BI rate_min \fR=\fPint[,int][,int] |
523bad63 | 2254 | Tell fio to do whatever it can to maintain at least this bandwidth. Failing |
338f2db5 | 2255 | to meet this requirement will cause the job to exit. Comma-separated values |
523bad63 TK |
2256 | may be specified for reads, writes, and trims as described in |
2257 | \fBblocksize\fR. | |
d60e92d1 | 2258 | .TP |
6d500c2e | 2259 | .BI rate_iops \fR=\fPint[,int][,int] |
523bad63 TK |
2260 | Cap the bandwidth to this number of IOPS. Basically the same as |
2261 | \fBrate\fR, just specified independently of bandwidth. If the job is | |
2262 | given a block size range instead of a fixed value, the smallest block size | |
338f2db5 | 2263 | is used as the metric. Comma-separated values may be specified for reads, |
523bad63 | 2264 | writes, and trims as described in \fBblocksize\fR. |
d60e92d1 | 2265 | .TP |
6d500c2e | 2266 | .BI rate_iops_min \fR=\fPint[,int][,int] |
523bad63 | 2267 | If fio doesn't meet this rate of I/O, it will cause the job to exit. |
338f2db5 | 2268 | Comma-separated values may be specified for reads, writes, and trims as |
523bad63 | 2269 | described in \fBblocksize\fR. |
d60e92d1 | 2270 | .TP |
6de65959 | 2271 | .BI rate_process \fR=\fPstr |
523bad63 TK |
2272 | This option controls how fio manages rated I/O submissions. The default is |
2273 | `linear', which submits I/O in a linear fashion with fixed delays between | |
2274 | I/Os that gets adjusted based on I/O completion rates. If this is set to | |
2275 | `poisson', fio will submit I/O based on a more real world random request | |
6de65959 | 2276 | flow, known as the Poisson process |
523bad63 | 2277 | (\fIhttps://en.wikipedia.org/wiki/Poisson_point_process\fR). The lambda will be |
5d02b083 | 2278 | 10^6 / IOPS for the given workload. |
1a9bf814 JA |
2279 | .TP |
2280 | .BI rate_ignore_thinktime \fR=\fPbool | |
2281 | By default, fio will attempt to catch up to the specified rate setting, if any | |
2282 | kind of thinktime setting was used. If this option is set, then fio will | |
2283 | ignore the thinktime and continue doing IO at the specified rate, instead of | |
2284 | entering a catch-up mode after thinktime is done. | |
523bad63 | 2285 | .SS "I/O latency" |
ff6bb260 | 2286 | .TP |
523bad63 | 2287 | .BI latency_target \fR=\fPtime |
3e260a46 | 2288 | If set, fio will attempt to find the max performance point that the given |
523bad63 TK |
2289 | workload will run at while maintaining a latency below this target. When |
2290 | the unit is omitted, the value is interpreted in microseconds. See | |
2291 | \fBlatency_window\fR and \fBlatency_percentile\fR. | |
3e260a46 | 2292 | .TP |
523bad63 | 2293 | .BI latency_window \fR=\fPtime |
3e260a46 | 2294 | Used with \fBlatency_target\fR to specify the sample window that the job |
523bad63 TK |
2295 | is run at varying queue depths to test the performance. When the unit is |
2296 | omitted, the value is interpreted in microseconds. | |
3e260a46 JA |
2297 | .TP |
2298 | .BI latency_percentile \fR=\fPfloat | |
523bad63 TK |
2299 | The percentage of I/Os that must fall within the criteria specified by |
2300 | \fBlatency_target\fR and \fBlatency_window\fR. If not set, this | |
2301 | defaults to 100.0, meaning that all I/Os must be equal or below to the value | |
2302 | set by \fBlatency_target\fR. | |
2303 | .TP | |
e1bcd541 SL |
2304 | .BI latency_run \fR=\fPbool |
2305 | Used with \fBlatency_target\fR. If false (default), fio will find the highest | |
2306 | queue depth that meets \fBlatency_target\fR and exit. If true, fio will continue | |
2307 | running and try to meet \fBlatency_target\fR by adjusting queue depth. | |
2308 | .TP | |
523bad63 TK |
2309 | .BI max_latency \fR=\fPtime |
2310 | If set, fio will exit the job with an ETIMEDOUT error if it exceeds this | |
2311 | maximum latency. When the unit is omitted, the value is interpreted in | |
2312 | microseconds. | |
2313 | .TP | |
2314 | .BI rate_cycle \fR=\fPint | |
2315 | Average bandwidth for \fBrate\fR and \fBrate_min\fR over this number | |
2316 | of milliseconds. Defaults to 1000. | |
2317 | .SS "I/O replay" | |
2318 | .TP | |
2319 | .BI write_iolog \fR=\fPstr | |
2320 | Write the issued I/O patterns to the specified file. See | |
2321 | \fBread_iolog\fR. Specify a separate file for each job, otherwise the | |
2322 | iologs will be interspersed and the file may be corrupt. | |
2323 | .TP | |
2324 | .BI read_iolog \fR=\fPstr | |
2325 | Open an iolog with the specified filename and replay the I/O patterns it | |
2326 | contains. This can be used to store a workload and replay it sometime | |
2327 | later. The iolog given may also be a blktrace binary file, which allows fio | |
2328 | to replay a workload captured by blktrace. See | |
2329 | \fBblktrace\fR\|(8) for how to capture such logging data. For blktrace | |
2330 | replay, the file needs to be turned into a blkparse binary data file first | |
2331 | (`blkparse <device> \-o /dev/null \-d file_for_fio.bin'). | |
c70c7f58 | 2332 | You can specify a number of files by separating the names with a ':' character. |
3b803fe1 | 2333 | See the \fBfilename\fR option for information on how to escape ':' |
c70c7f58 | 2334 | characters within the file names. These files will be sequentially assigned to |
d19c04d1 | 2335 | job clones created by \fBnumjobs\fR. '-' is a reserved name, meaning read from |
2336 | stdin, notably if \fBfilename\fR is set to '-' which means stdin as well, | |
2337 | then this flag can't be set to '-'. | |
3e260a46 | 2338 | .TP |
98e7161c AK |
2339 | .BI read_iolog_chunked \fR=\fPbool |
2340 | Determines how iolog is read. If false (default) entire \fBread_iolog\fR will | |
2341 | be read at once. If selected true, input from iolog will be read gradually. | |
2342 | Useful when iolog is very large, or it is generated. | |
2343 | .TP | |
b9921d1a DZ |
2344 | .BI merge_blktrace_file \fR=\fPstr |
2345 | When specified, rather than replaying the logs passed to \fBread_iolog\fR, | |
2346 | the logs go through a merge phase which aggregates them into a single blktrace. | |
2347 | The resulting file is then passed on as the \fBread_iolog\fR parameter. The | |
2348 | intention here is to make the order of events consistent. This limits the | |
2349 | influence of the scheduler compared to replaying multiple blktraces via | |
2350 | concurrent jobs. | |
2351 | .TP | |
87a48ada DZ |
2352 | .BI merge_blktrace_scalars \fR=\fPfloat_list |
2353 | This is a percentage based option that is index paired with the list of files | |
2354 | passed to \fBread_iolog\fR. When merging is performed, scale the time of each | |
2355 | event by the corresponding amount. For example, | |
2356 | `\-\-merge_blktrace_scalars="50:100"' runs the first trace in halftime and the | |
2357 | second trace in realtime. This knob is separately tunable from | |
2358 | \fBreplay_time_scale\fR which scales the trace during runtime and will not | |
2359 | change the output of the merge unlike this option. | |
2360 | .TP | |
55bfd8c8 DZ |
2361 | .BI merge_blktrace_iters \fR=\fPfloat_list |
2362 | This is a whole number option that is index paired with the list of files | |
2363 | passed to \fBread_iolog\fR. When merging is performed, run each trace for | |
2364 | the specified number of iterations. For example, | |
2365 | `\-\-merge_blktrace_iters="2:1"' runs the first trace for two iterations | |
2366 | and the second trace for one iteration. | |
2367 | .TP | |
523bad63 TK |
2368 | .BI replay_no_stall \fR=\fPbool |
2369 | When replaying I/O with \fBread_iolog\fR the default behavior is to | |
2370 | attempt to respect the timestamps within the log and replay them with the | |
2371 | appropriate delay between IOPS. By setting this variable fio will not | |
2372 | respect the timestamps and attempt to replay them as fast as possible while | |
2373 | still respecting ordering. The result is the same I/O pattern to a given | |
2374 | device, but different timings. | |
2375 | .TP | |
6dd7fa77 JA |
2376 | .BI replay_time_scale \fR=\fPint |
2377 | When replaying I/O with \fBread_iolog\fR, fio will honor the original timing | |
2378 | in the trace. With this option, it's possible to scale the time. It's a | |
2379 | percentage option, if set to 50 it means run at 50% the original IO rate in | |
2380 | the trace. If set to 200, run at twice the original IO rate. Defaults to 100. | |
2381 | .TP | |
523bad63 TK |
2382 | .BI replay_redirect \fR=\fPstr |
2383 | While replaying I/O patterns using \fBread_iolog\fR the default behavior | |
2384 | is to replay the IOPS onto the major/minor device that each IOP was recorded | |
2385 | from. This is sometimes undesirable because on a different machine those | |
2386 | major/minor numbers can map to a different device. Changing hardware on the | |
2387 | same system can also result in a different major/minor mapping. | |
2388 | \fBreplay_redirect\fR causes all I/Os to be replayed onto the single specified | |
2389 | device regardless of the device it was recorded | |
2390 | from. i.e. `replay_redirect=/dev/sdc' would cause all I/O | |
2391 | in the blktrace or iolog to be replayed onto `/dev/sdc'. This means | |
2392 | multiple devices will be replayed onto a single device, if the trace | |
2393 | contains multiple devices. If you want multiple devices to be replayed | |
2394 | concurrently to multiple redirected devices you must blkparse your trace | |
2395 | into separate traces and replay them with independent fio invocations. | |
2396 | Unfortunately this also breaks the strict time ordering between multiple | |
2397 | device accesses. | |
2398 | .TP | |
2399 | .BI replay_align \fR=\fPint | |
350a535d DZ |
2400 | Force alignment of the byte offsets in a trace to this value. The value |
2401 | must be a power of 2. | |
523bad63 TK |
2402 | .TP |
2403 | .BI replay_scale \fR=\fPint | |
350a535d DZ |
2404 | Scale bye offsets down by this factor when replaying traces. Should most |
2405 | likely use \fBreplay_align\fR as well. | |
523bad63 TK |
2406 | .SS "Threads, processes and job synchronization" |
2407 | .TP | |
38f68906 JA |
2408 | .BI replay_skip \fR=\fPstr |
2409 | Sometimes it's useful to skip certain IO types in a replay trace. This could | |
2410 | be, for instance, eliminating the writes in the trace. Or not replaying the | |
2411 | trims/discards, if you are redirecting to a device that doesn't support them. | |
2412 | This option takes a comma separated list of read, write, trim, sync. | |
2413 | .TP | |
523bad63 TK |
2414 | .BI thread |
2415 | Fio defaults to creating jobs by using fork, however if this option is | |
2416 | given, fio will create jobs by using POSIX Threads' function | |
2417 | \fBpthread_create\fR\|(3) to create threads instead. | |
2418 | .TP | |
2419 | .BI wait_for \fR=\fPstr | |
2420 | If set, the current job won't be started until all workers of the specified | |
2421 | waitee job are done. | |
2422 | .\" ignore blank line here from HOWTO as it looks normal without it | |
2423 | \fBwait_for\fR operates on the job name basis, so there are a few | |
2424 | limitations. First, the waitee must be defined prior to the waiter job | |
2425 | (meaning no forward references). Second, if a job is being referenced as a | |
2426 | waitee, it must have a unique name (no duplicate waitees). | |
2427 | .TP | |
2428 | .BI nice \fR=\fPint | |
2429 | Run the job with the given nice value. See man \fBnice\fR\|(2). | |
2430 | .\" ignore blank line here from HOWTO as it looks normal without it | |
2431 | On Windows, values less than \-15 set the process class to "High"; \-1 through | |
2432 | \-15 set "Above Normal"; 1 through 15 "Below Normal"; and above 15 "Idle" | |
2433 | priority class. | |
2434 | .TP | |
2435 | .BI prio \fR=\fPint | |
2436 | Set the I/O priority value of this job. Linux limits us to a positive value | |
2437 | between 0 and 7, with 0 being the highest. See man | |
2438 | \fBionice\fR\|(1). Refer to an appropriate manpage for other operating | |
b2a432bf PC |
2439 | systems since meaning of priority may differ. For per-command priority |
2440 | setting, see I/O engine specific `cmdprio_percentage` and `hipri_percentage` | |
2441 | options. | |
523bad63 TK |
2442 | .TP |
2443 | .BI prioclass \fR=\fPint | |
b2a432bf PC |
2444 | Set the I/O priority class. See man \fBionice\fR\|(1). For per-command |
2445 | priority setting, see I/O engine specific `cmdprio_percentage` and `hipri_percent` | |
2446 | options. | |
15501535 | 2447 | .TP |
d60e92d1 | 2448 | .BI cpus_allowed \fR=\fPstr |
523bad63 | 2449 | Controls the same options as \fBcpumask\fR, but accepts a textual |
b570e037 SW |
2450 | specification of the permitted CPUs instead and CPUs are indexed from 0. So |
2451 | to use CPUs 0 and 5 you would specify `cpus_allowed=0,5'. This option also | |
2452 | allows a range of CPUs to be specified \-\- say you wanted a binding to CPUs | |
2453 | 0, 5, and 8 to 15, you would set `cpus_allowed=0,5,8\-15'. | |
2454 | .RS | |
2455 | .P | |
2456 | On Windows, when `cpus_allowed' is unset only CPUs from fio's current | |
2457 | processor group will be used and affinity settings are inherited from the | |
2458 | system. An fio build configured to target Windows 7 makes options that set | |
2459 | CPUs processor group aware and values will set both the processor group | |
2460 | and a CPU from within that group. For example, on a system where processor | |
2461 | group 0 has 40 CPUs and processor group 1 has 32 CPUs, `cpus_allowed' | |
2462 | values between 0 and 39 will bind CPUs from processor group 0 and | |
2463 | `cpus_allowed' values between 40 and 71 will bind CPUs from processor | |
2464 | group 1. When using `cpus_allowed_policy=shared' all CPUs specified by a | |
2465 | single `cpus_allowed' option must be from the same processor group. For | |
2466 | Windows fio builds not built for Windows 7, CPUs will only be selected from | |
2467 | (and be relative to) whatever processor group fio happens to be running in | |
2468 | and CPUs from other processor groups cannot be used. | |
2469 | .RE | |
d60e92d1 | 2470 | .TP |
c2acfbac | 2471 | .BI cpus_allowed_policy \fR=\fPstr |
523bad63 TK |
2472 | Set the policy of how fio distributes the CPUs specified by |
2473 | \fBcpus_allowed\fR or \fBcpumask\fR. Two policies are supported: | |
c2acfbac JA |
2474 | .RS |
2475 | .RS | |
2476 | .TP | |
2477 | .B shared | |
2478 | All jobs will share the CPU set specified. | |
2479 | .TP | |
2480 | .B split | |
2481 | Each job will get a unique CPU from the CPU set. | |
2482 | .RE | |
2483 | .P | |
523bad63 | 2484 | \fBshared\fR is the default behavior, if the option isn't specified. If |
b21fc93f | 2485 | \fBsplit\fR is specified, then fio will assign one cpu per job. If not |
523bad63 TK |
2486 | enough CPUs are given for the jobs listed, then fio will roundrobin the CPUs |
2487 | in the set. | |
c2acfbac | 2488 | .RE |
c2acfbac | 2489 | .TP |
b570e037 SW |
2490 | .BI cpumask \fR=\fPint |
2491 | Set the CPU affinity of this job. The parameter given is a bit mask of | |
2492 | allowed CPUs the job may run on. So if you want the allowed CPUs to be 1 | |
2493 | and 5, you would pass the decimal value of (1 << 1 | 1 << 5), or 34. See man | |
2494 | \fBsched_setaffinity\fR\|(2). This may not work on all supported | |
2495 | operating systems or kernel versions. This option doesn't work well for a | |
2496 | higher CPU count than what you can store in an integer mask, so it can only | |
2497 | control cpus 1\-32. For boxes with larger CPU counts, use | |
2498 | \fBcpus_allowed\fR. | |
2499 | .TP | |
d0b937ed | 2500 | .BI numa_cpu_nodes \fR=\fPstr |
cecbfd47 | 2501 | Set this job running on specified NUMA nodes' CPUs. The arguments allow |
523bad63 TK |
2502 | comma delimited list of cpu numbers, A\-B ranges, or `all'. Note, to enable |
2503 | NUMA options support, fio must be built on a system with libnuma\-dev(el) | |
2504 | installed. | |
d0b937ed YR |
2505 | .TP |
2506 | .BI numa_mem_policy \fR=\fPstr | |
523bad63 TK |
2507 | Set this job's memory policy and corresponding NUMA nodes. Format of the |
2508 | arguments: | |
39c7a2ca VF |
2509 | .RS |
2510 | .RS | |
523bad63 TK |
2511 | .P |
2512 | <mode>[:<nodelist>] | |
39c7a2ca | 2513 | .RE |
523bad63 | 2514 | .P |
f1dd3fb1 | 2515 | `mode' is one of the following memory policies: `default', `prefer', |
523bad63 TK |
2516 | `bind', `interleave' or `local'. For `default' and `local' memory |
2517 | policies, no node needs to be specified. For `prefer', only one node is | |
2518 | allowed. For `bind' and `interleave' the `nodelist' may be as | |
2519 | follows: a comma delimited list of numbers, A\-B ranges, or `all'. | |
39c7a2ca VF |
2520 | .RE |
2521 | .TP | |
523bad63 TK |
2522 | .BI cgroup \fR=\fPstr |
2523 | Add job to this control group. If it doesn't exist, it will be created. The | |
2524 | system must have a mounted cgroup blkio mount point for this to work. If | |
2525 | your system doesn't have it mounted, you can do so with: | |
d60e92d1 AC |
2526 | .RS |
2527 | .RS | |
d60e92d1 | 2528 | .P |
523bad63 TK |
2529 | # mount \-t cgroup \-o blkio none /cgroup |
2530 | .RE | |
d60e92d1 AC |
2531 | .RE |
2532 | .TP | |
523bad63 TK |
2533 | .BI cgroup_weight \fR=\fPint |
2534 | Set the weight of the cgroup to this value. See the documentation that comes | |
2535 | with the kernel, allowed values are in the range of 100..1000. | |
d60e92d1 | 2536 | .TP |
523bad63 TK |
2537 | .BI cgroup_nodelete \fR=\fPbool |
2538 | Normally fio will delete the cgroups it has created after the job | |
2539 | completion. To override this behavior and to leave cgroups around after the | |
2540 | job completion, set `cgroup_nodelete=1'. This can be useful if one wants | |
2541 | to inspect various cgroup files after job completion. Default: false. | |
c8eeb9df | 2542 | .TP |
523bad63 TK |
2543 | .BI flow_id \fR=\fPint |
2544 | The ID of the flow. If not specified, it defaults to being a global | |
2545 | flow. See \fBflow\fR. | |
d60e92d1 | 2546 | .TP |
523bad63 | 2547 | .BI flow \fR=\fPint |
d4e74fda DB |
2548 | Weight in token-based flow control. If this value is used, |
2549 | then fio regulates the activity between two or more jobs | |
2550 | sharing the same flow_id. | |
2551 | Fio attempts to keep each job activity proportional to other jobs' activities | |
2552 | in the same flow_id group, with respect to requested weight per job. | |
2553 | That is, if one job has `flow=3', another job has `flow=2' | |
2554 | and another with `flow=1`, then there will be a roughly 3:2:1 ratio | |
2555 | in how much one runs vs the others. | |
6b7f6851 | 2556 | .TP |
523bad63 | 2557 | .BI flow_sleep \fR=\fPint |
d4e74fda DB |
2558 | The period of time, in microseconds, to wait after the flow counter |
2559 | has exceeded its proportion before retrying operations. | |
25460cf6 | 2560 | .TP |
523bad63 TK |
2561 | .BI stonewall "\fR,\fB wait_for_previous" |
2562 | Wait for preceding jobs in the job file to exit, before starting this | |
2563 | one. Can be used to insert serialization points in the job file. A stone | |
2564 | wall also implies starting a new reporting group, see | |
fd56c235 AW |
2565 | \fBgroup_reporting\fR. Optionally you can use `stonewall=0` to disable or |
2566 | `stonewall=1` to enable it. | |
2378826d | 2567 | .TP |
523bad63 | 2568 | .BI exitall |
64402a8a HW |
2569 | By default, fio will continue running all other jobs when one job finishes. |
2570 | Sometimes this is not the desired action. Setting \fBexitall\fR will instead | |
2571 | make fio terminate all jobs in the same group, as soon as one job of that | |
2572 | group finishes. | |
2573 | .TP | |
fd56c235 | 2574 | .BI exit_what \fR=\fPstr |
64402a8a | 2575 | By default, fio will continue running all other jobs when one job finishes. |
fd56c235 | 2576 | Sometimes this is not the desired action. Setting \fBexitall\fR will instead |
64402a8a | 2577 | make fio terminate all jobs in the same group. The option \fBexit_what\fR |
fd56c235 AW |
2578 | allows you to control which jobs get terminated when \fBexitall\fR is enabled. |
2579 | The default value is \fBgroup\fR. | |
2580 | The allowed values are: | |
2581 | .RS | |
2582 | .RS | |
2583 | .TP | |
2584 | .B all | |
2585 | terminates all jobs. | |
2586 | .TP | |
2587 | .B group | |
2588 | is the default and does not change the behaviour of \fBexitall\fR. | |
2589 | .TP | |
2590 | .B stonewall | |
2591 | terminates all currently running jobs across all groups and continues | |
2592 | execution with the next stonewalled group. | |
2593 | .RE | |
2594 | .RE | |
e81ecca3 | 2595 | .TP |
523bad63 TK |
2596 | .BI exec_prerun \fR=\fPstr |
2597 | Before running this job, issue the command specified through | |
2598 | \fBsystem\fR\|(3). Output is redirected in a file called `jobname.prerun.txt'. | |
e9f48479 | 2599 | .TP |
523bad63 TK |
2600 | .BI exec_postrun \fR=\fPstr |
2601 | After the job completes, issue the command specified though | |
2602 | \fBsystem\fR\|(3). Output is redirected in a file called `jobname.postrun.txt'. | |
d60e92d1 | 2603 | .TP |
523bad63 TK |
2604 | .BI uid \fR=\fPint |
2605 | Instead of running as the invoking user, set the user ID to this value | |
2606 | before the thread/process does any work. | |
39c1c323 | 2607 | .TP |
523bad63 TK |
2608 | .BI gid \fR=\fPint |
2609 | Set group ID, see \fBuid\fR. | |
2610 | .SS "Verification" | |
d60e92d1 | 2611 | .TP |
589e88b7 | 2612 | .BI verify_only |
523bad63 | 2613 | Do not perform specified workload, only verify data still matches previous |
5e4c7118 | 2614 | invocation of this workload. This option allows one to check data multiple |
523bad63 TK |
2615 | times at a later date without overwriting it. This option makes sense only |
2616 | for workloads that write data, and does not support workloads with the | |
5e4c7118 JA |
2617 | \fBtime_based\fR option set. |
2618 | .TP | |
d60e92d1 | 2619 | .BI do_verify \fR=\fPbool |
523bad63 TK |
2620 | Run the verify phase after a write phase. Only valid if \fBverify\fR is |
2621 | set. Default: true. | |
d60e92d1 AC |
2622 | .TP |
2623 | .BI verify \fR=\fPstr | |
523bad63 TK |
2624 | If writing to a file, fio can verify the file contents after each iteration |
2625 | of the job. Each verification method also implies verification of special | |
2626 | header, which is written to the beginning of each block. This header also | |
2627 | includes meta information, like offset of the block, block number, timestamp | |
2628 | when block was written, etc. \fBverify\fR can be combined with | |
2629 | \fBverify_pattern\fR option. The allowed values are: | |
d60e92d1 AC |
2630 | .RS |
2631 | .RS | |
2632 | .TP | |
523bad63 TK |
2633 | .B md5 |
2634 | Use an md5 sum of the data area and store it in the header of | |
2635 | each block. | |
2636 | .TP | |
2637 | .B crc64 | |
2638 | Use an experimental crc64 sum of the data area and store it in the | |
2639 | header of each block. | |
2640 | .TP | |
2641 | .B crc32c | |
2642 | Use a crc32c sum of the data area and store it in the header of | |
2643 | each block. This will automatically use hardware acceleration | |
2644 | (e.g. SSE4.2 on an x86 or CRC crypto extensions on ARM64) but will | |
2645 | fall back to software crc32c if none is found. Generally the | |
f1dd3fb1 | 2646 | fastest checksum fio supports when hardware accelerated. |
523bad63 TK |
2647 | .TP |
2648 | .B crc32c\-intel | |
2649 | Synonym for crc32c. | |
2650 | .TP | |
2651 | .B crc32 | |
2652 | Use a crc32 sum of the data area and store it in the header of each | |
2653 | block. | |
2654 | .TP | |
2655 | .B crc16 | |
2656 | Use a crc16 sum of the data area and store it in the header of each | |
2657 | block. | |
2658 | .TP | |
2659 | .B crc7 | |
2660 | Use a crc7 sum of the data area and store it in the header of each | |
2661 | block. | |
2662 | .TP | |
2663 | .B xxhash | |
2664 | Use xxhash as the checksum function. Generally the fastest software | |
2665 | checksum that fio supports. | |
2666 | .TP | |
2667 | .B sha512 | |
2668 | Use sha512 as the checksum function. | |
2669 | .TP | |
2670 | .B sha256 | |
2671 | Use sha256 as the checksum function. | |
2672 | .TP | |
2673 | .B sha1 | |
2674 | Use optimized sha1 as the checksum function. | |
2675 | .TP | |
2676 | .B sha3\-224 | |
2677 | Use optimized sha3\-224 as the checksum function. | |
2678 | .TP | |
2679 | .B sha3\-256 | |
2680 | Use optimized sha3\-256 as the checksum function. | |
2681 | .TP | |
2682 | .B sha3\-384 | |
2683 | Use optimized sha3\-384 as the checksum function. | |
2684 | .TP | |
2685 | .B sha3\-512 | |
2686 | Use optimized sha3\-512 as the checksum function. | |
d60e92d1 AC |
2687 | .TP |
2688 | .B meta | |
523bad63 TK |
2689 | This option is deprecated, since now meta information is included in |
2690 | generic verification header and meta verification happens by | |
2691 | default. For detailed information see the description of the | |
2692 | \fBverify\fR setting. This option is kept because of | |
2693 | compatibility's sake with old configurations. Do not use it. | |
d60e92d1 | 2694 | .TP |
59245381 | 2695 | .B pattern |
523bad63 TK |
2696 | Verify a strict pattern. Normally fio includes a header with some |
2697 | basic information and checksumming, but if this option is set, only | |
2698 | the specific pattern set with \fBverify_pattern\fR is verified. | |
59245381 | 2699 | .TP |
d60e92d1 | 2700 | .B null |
523bad63 TK |
2701 | Only pretend to verify. Useful for testing internals with |
2702 | `ioengine=null', not for much else. | |
d60e92d1 | 2703 | .RE |
523bad63 TK |
2704 | .P |
2705 | This option can be used for repeated burn\-in tests of a system to make sure | |
2706 | that the written data is also correctly read back. If the data direction | |
2707 | given is a read or random read, fio will assume that it should verify a | |
2708 | previously written file. If the data direction includes any form of write, | |
2709 | the verify will be of the newly written data. | |
47e6a6e5 SW |
2710 | .P |
2711 | To avoid false verification errors, do not use the norandommap option when | |
2712 | verifying data with async I/O engines and I/O depths > 1. Or use the | |
2713 | norandommap and the lfsr random generator together to avoid writing to the | |
2714 | same offset with muliple outstanding I/Os. | |
d60e92d1 AC |
2715 | .RE |
2716 | .TP | |
f7fa2653 | 2717 | .BI verify_offset \fR=\fPint |
d60e92d1 | 2718 | Swap the verification header with data somewhere else in the block before |
523bad63 | 2719 | writing. It is swapped back before verifying. |
d60e92d1 | 2720 | .TP |
f7fa2653 | 2721 | .BI verify_interval \fR=\fPint |
523bad63 TK |
2722 | Write the verification header at a finer granularity than the |
2723 | \fBblocksize\fR. It will be written for chunks the size of | |
2724 | \fBverify_interval\fR. \fBblocksize\fR should divide this evenly. | |
d60e92d1 | 2725 | .TP |
996093bb | 2726 | .BI verify_pattern \fR=\fPstr |
523bad63 TK |
2727 | If set, fio will fill the I/O buffers with this pattern. Fio defaults to |
2728 | filling with totally random bytes, but sometimes it's interesting to fill | |
2729 | with a known pattern for I/O verification purposes. Depending on the width | |
2730 | of the pattern, fio will fill 1/2/3/4 bytes of the buffer at the time (it can | |
2731 | be either a decimal or a hex number). The \fBverify_pattern\fR if larger than | |
2732 | a 32\-bit quantity has to be a hex number that starts with either "0x" or | |
2733 | "0X". Use with \fBverify\fR. Also, \fBverify_pattern\fR supports %o | |
2734 | format, which means that for each block offset will be written and then | |
2735 | verified back, e.g.: | |
2fa5a241 RP |
2736 | .RS |
2737 | .RS | |
523bad63 TK |
2738 | .P |
2739 | verify_pattern=%o | |
2fa5a241 | 2740 | .RE |
523bad63 | 2741 | .P |
2fa5a241 | 2742 | Or use combination of everything: |
2fa5a241 | 2743 | .RS |
523bad63 TK |
2744 | .P |
2745 | verify_pattern=0xff%o"abcd"\-12 | |
2fa5a241 RP |
2746 | .RE |
2747 | .RE | |
996093bb | 2748 | .TP |
d60e92d1 | 2749 | .BI verify_fatal \fR=\fPbool |
523bad63 TK |
2750 | Normally fio will keep checking the entire contents before quitting on a |
2751 | block verification failure. If this option is set, fio will exit the job on | |
2752 | the first observed failure. Default: false. | |
d60e92d1 | 2753 | .TP |
b463e936 | 2754 | .BI verify_dump \fR=\fPbool |
523bad63 TK |
2755 | If set, dump the contents of both the original data block and the data block |
2756 | we read off disk to files. This allows later analysis to inspect just what | |
2757 | kind of data corruption occurred. Off by default. | |
b463e936 | 2758 | .TP |
e8462bd8 | 2759 | .BI verify_async \fR=\fPint |
523bad63 TK |
2760 | Fio will normally verify I/O inline from the submitting thread. This option |
2761 | takes an integer describing how many async offload threads to create for I/O | |
2762 | verification instead, causing fio to offload the duty of verifying I/O | |
2763 | contents to one or more separate threads. If using this offload option, even | |
2764 | sync I/O engines can benefit from using an \fBiodepth\fR setting higher | |
2765 | than 1, as it allows them to have I/O in flight while verifies are running. | |
2766 | Defaults to 0 async threads, i.e. verification is not asynchronous. | |
e8462bd8 JA |
2767 | .TP |
2768 | .BI verify_async_cpus \fR=\fPstr | |
523bad63 TK |
2769 | Tell fio to set the given CPU affinity on the async I/O verification |
2770 | threads. See \fBcpus_allowed\fR for the format used. | |
e8462bd8 | 2771 | .TP |
6f87418f JA |
2772 | .BI verify_backlog \fR=\fPint |
2773 | Fio will normally verify the written contents of a job that utilizes verify | |
2774 | once that job has completed. In other words, everything is written then | |
2775 | everything is read back and verified. You may want to verify continually | |
523bad63 TK |
2776 | instead for a variety of reasons. Fio stores the meta data associated with |
2777 | an I/O block in memory, so for large verify workloads, quite a bit of memory | |
2778 | would be used up holding this meta data. If this option is enabled, fio will | |
2779 | write only N blocks before verifying these blocks. | |
6f87418f JA |
2780 | .TP |
2781 | .BI verify_backlog_batch \fR=\fPint | |
523bad63 TK |
2782 | Control how many blocks fio will verify if \fBverify_backlog\fR is |
2783 | set. If not set, will default to the value of \fBverify_backlog\fR | |
2784 | (meaning the entire queue is read back and verified). If | |
2785 | \fBverify_backlog_batch\fR is less than \fBverify_backlog\fR then not all | |
2786 | blocks will be verified, if \fBverify_backlog_batch\fR is larger than | |
2787 | \fBverify_backlog\fR, some blocks will be verified more than once. | |
2788 | .TP | |
2789 | .BI verify_state_save \fR=\fPbool | |
2790 | When a job exits during the write phase of a verify workload, save its | |
2791 | current state. This allows fio to replay up until that point, if the verify | |
2792 | state is loaded for the verify read phase. The format of the filename is, | |
2793 | roughly: | |
2794 | .RS | |
2795 | .RS | |
2796 | .P | |
2797 | <type>\-<jobname>\-<jobindex>\-verify.state. | |
2798 | .RE | |
2799 | .P | |
2800 | <type> is "local" for a local run, "sock" for a client/server socket | |
2801 | connection, and "ip" (192.168.0.1, for instance) for a networked | |
2802 | client/server connection. Defaults to true. | |
2803 | .RE | |
2804 | .TP | |
2805 | .BI verify_state_load \fR=\fPbool | |
2806 | If a verify termination trigger was used, fio stores the current write state | |
2807 | of each thread. This can be used at verification time so that fio knows how | |
2808 | far it should verify. Without this information, fio will run a full | |
2809 | verification pass, according to the settings in the job file used. Default | |
2810 | false. | |
6f87418f | 2811 | .TP |
fa769d44 SW |
2812 | .BI trim_percentage \fR=\fPint |
2813 | Number of verify blocks to discard/trim. | |
2814 | .TP | |
2815 | .BI trim_verify_zero \fR=\fPbool | |
523bad63 | 2816 | Verify that trim/discarded blocks are returned as zeros. |
fa769d44 SW |
2817 | .TP |
2818 | .BI trim_backlog \fR=\fPint | |
523bad63 | 2819 | Verify that trim/discarded blocks are returned as zeros. |
fa769d44 SW |
2820 | .TP |
2821 | .BI trim_backlog_batch \fR=\fPint | |
523bad63 | 2822 | Trim this number of I/O blocks. |
fa769d44 SW |
2823 | .TP |
2824 | .BI experimental_verify \fR=\fPbool | |
2825 | Enable experimental verification. | |
523bad63 | 2826 | .SS "Steady state" |
fa769d44 | 2827 | .TP |
523bad63 TK |
2828 | .BI steadystate \fR=\fPstr:float "\fR,\fP ss" \fR=\fPstr:float |
2829 | Define the criterion and limit for assessing steady state performance. The | |
2830 | first parameter designates the criterion whereas the second parameter sets | |
2831 | the threshold. When the criterion falls below the threshold for the | |
2832 | specified duration, the job will stop. For example, `iops_slope:0.1%' will | |
2833 | direct fio to terminate the job when the least squares regression slope | |
2834 | falls below 0.1% of the mean IOPS. If \fBgroup_reporting\fR is enabled | |
2835 | this will apply to all jobs in the group. Below is the list of available | |
2836 | steady state assessment criteria. All assessments are carried out using only | |
2837 | data from the rolling collection window. Threshold limits can be expressed | |
2838 | as a fixed value or as a percentage of the mean in the collection window. | |
2839 | .RS | |
1cb049d9 VF |
2840 | .P |
2841 | When using this feature, most jobs should include the \fBtime_based\fR | |
2842 | and \fBruntime\fR options or the \fBloops\fR option so that fio does not | |
2843 | stop running after it has covered the full size of the specified file(s) | |
2844 | or device(s). | |
2845 | .RS | |
523bad63 | 2846 | .RS |
d60e92d1 | 2847 | .TP |
523bad63 TK |
2848 | .B iops |
2849 | Collect IOPS data. Stop the job if all individual IOPS measurements | |
2850 | are within the specified limit of the mean IOPS (e.g., `iops:2' | |
2851 | means that all individual IOPS values must be within 2 of the mean, | |
2852 | whereas `iops:0.2%' means that all individual IOPS values must be | |
2853 | within 0.2% of the mean IOPS to terminate the job). | |
d60e92d1 | 2854 | .TP |
523bad63 TK |
2855 | .B iops_slope |
2856 | Collect IOPS data and calculate the least squares regression | |
2857 | slope. Stop the job if the slope falls below the specified limit. | |
d60e92d1 | 2858 | .TP |
523bad63 TK |
2859 | .B bw |
2860 | Collect bandwidth data. Stop the job if all individual bandwidth | |
2861 | measurements are within the specified limit of the mean bandwidth. | |
64bbb865 | 2862 | .TP |
523bad63 TK |
2863 | .B bw_slope |
2864 | Collect bandwidth data and calculate the least squares regression | |
2865 | slope. Stop the job if the slope falls below the specified limit. | |
2866 | .RE | |
2867 | .RE | |
d1c46c04 | 2868 | .TP |
523bad63 TK |
2869 | .BI steadystate_duration \fR=\fPtime "\fR,\fP ss_dur" \fR=\fPtime |
2870 | A rolling window of this duration will be used to judge whether steady state | |
2871 | has been reached. Data will be collected once per second. The default is 0 | |
2872 | which disables steady state detection. When the unit is omitted, the | |
2873 | value is interpreted in seconds. | |
0c63576e | 2874 | .TP |
523bad63 TK |
2875 | .BI steadystate_ramp_time \fR=\fPtime "\fR,\fP ss_ramp" \fR=\fPtime |
2876 | Allow the job to run for the specified duration before beginning data | |
2877 | collection for checking the steady state job termination criterion. The | |
2878 | default is 0. When the unit is omitted, the value is interpreted in seconds. | |
2879 | .SS "Measurements and reporting" | |
0c63576e | 2880 | .TP |
3a5db920 JA |
2881 | .BI per_job_logs \fR=\fPbool |
2882 | If set, this generates bw/clat/iops log with per file private filenames. If | |
523bad63 TK |
2883 | not set, jobs with identical names will share the log filename. Default: |
2884 | true. | |
2885 | .TP | |
2886 | .BI group_reporting | |
2887 | It may sometimes be interesting to display statistics for groups of jobs as | |
2888 | a whole instead of for each individual job. This is especially true if | |
2889 | \fBnumjobs\fR is used; looking at individual thread/process output | |
338f2db5 SW |
2890 | quickly becomes unwieldy. To see the final report per-group instead of |
2891 | per-job, use \fBgroup_reporting\fR. Jobs in a file will be part of the | |
523bad63 TK |
2892 | same reporting group, unless if separated by a \fBstonewall\fR, or by |
2893 | using \fBnew_group\fR. | |
2894 | .TP | |
2895 | .BI new_group | |
2896 | Start a new reporting group. See: \fBgroup_reporting\fR. If not given, | |
2897 | all jobs in a file will be part of the same reporting group, unless | |
2898 | separated by a \fBstonewall\fR. | |
2899 | .TP | |
2900 | .BI stats \fR=\fPbool | |
2901 | By default, fio collects and shows final output results for all jobs | |
2902 | that run. If this option is set to 0, then fio will ignore it in | |
2903 | the final stat output. | |
3a5db920 | 2904 | .TP |
836bad52 | 2905 | .BI write_bw_log \fR=\fPstr |
523bad63 | 2906 | If given, write a bandwidth log for this job. Can be used to store data of |
074f0817 | 2907 | the bandwidth of the jobs in their lifetime. |
523bad63 | 2908 | .RS |
074f0817 SW |
2909 | .P |
2910 | If no str argument is given, the default filename of | |
2911 | `jobname_type.x.log' is used. Even when the argument is given, fio | |
2912 | will still append the type of log. So if one specifies: | |
523bad63 TK |
2913 | .RS |
2914 | .P | |
074f0817 | 2915 | write_bw_log=foo |
523bad63 TK |
2916 | .RE |
2917 | .P | |
074f0817 SW |
2918 | The actual log name will be `foo_bw.x.log' where `x' is the index |
2919 | of the job (1..N, where N is the number of jobs). If | |
2920 | \fBper_job_logs\fR is false, then the filename will not include the | |
2921 | `.x` job index. | |
2922 | .P | |
2923 | The included \fBfio_generate_plots\fR script uses gnuplot to turn these | |
2924 | text files into nice graphs. See the \fBLOG FILE FORMATS\fR section for how data is | |
2925 | structured within the file. | |
523bad63 | 2926 | .RE |
901bb994 | 2927 | .TP |
074f0817 SW |
2928 | .BI write_lat_log \fR=\fPstr |
2929 | Same as \fBwrite_bw_log\fR, except this option creates I/O | |
2930 | submission (e.g., `name_slat.x.log'), completion (e.g., | |
2931 | `name_clat.x.log'), and total (e.g., `name_lat.x.log') latency | |
2932 | files instead. See \fBwrite_bw_log\fR for details about the | |
2933 | filename format and the \fBLOG FILE FORMATS\fR section for how data is structured | |
2934 | within the files. | |
2935 | .TP | |
1e613c9c | 2936 | .BI write_hist_log \fR=\fPstr |
074f0817 SW |
2937 | Same as \fBwrite_bw_log\fR but writes an I/O completion latency |
2938 | histogram file (e.g., `name_hist.x.log') instead. Note that this | |
2939 | file will be empty unless \fBlog_hist_msec\fR has also been set. | |
2940 | See \fBwrite_bw_log\fR for details about the filename format and | |
2941 | the \fBLOG FILE FORMATS\fR section for how data is structured | |
2942 | within the file. | |
1e613c9c | 2943 | .TP |
c8eeb9df | 2944 | .BI write_iops_log \fR=\fPstr |
074f0817 | 2945 | Same as \fBwrite_bw_log\fR, but writes an IOPS file (e.g. |
15417073 SW |
2946 | `name_iops.x.log`) instead. Because fio defaults to individual |
2947 | I/O logging, the value entry in the IOPS log will be 1 unless windowed | |
2948 | logging (see \fBlog_avg_msec\fR) has been enabled. See | |
2949 | \fBwrite_bw_log\fR for details about the filename format and \fBLOG | |
2950 | FILE FORMATS\fR for how data is structured within the file. | |
c8eeb9df | 2951 | .TP |
b8bc8cba JA |
2952 | .BI log_avg_msec \fR=\fPint |
2953 | By default, fio will log an entry in the iops, latency, or bw log for every | |
523bad63 | 2954 | I/O that completes. When writing to the disk log, that can quickly grow to a |
b8bc8cba | 2955 | very large size. Setting this option makes fio average the each log entry |
e6989e10 | 2956 | over the specified period of time, reducing the resolution of the log. See |
523bad63 TK |
2957 | \fBlog_max_value\fR as well. Defaults to 0, logging all entries. |
2958 | Also see \fBLOG FILE FORMATS\fR section. | |
b8bc8cba | 2959 | .TP |
1e613c9c | 2960 | .BI log_hist_msec \fR=\fPint |
523bad63 TK |
2961 | Same as \fBlog_avg_msec\fR, but logs entries for completion latency |
2962 | histograms. Computing latency percentiles from averages of intervals using | |
2963 | \fBlog_avg_msec\fR is inaccurate. Setting this option makes fio log | |
2964 | histogram entries over the specified period of time, reducing log sizes for | |
2965 | high IOPS devices while retaining percentile accuracy. See | |
074f0817 SW |
2966 | \fBlog_hist_coarseness\fR and \fBwrite_hist_log\fR as well. |
2967 | Defaults to 0, meaning histogram logging is disabled. | |
1e613c9c KC |
2968 | .TP |
2969 | .BI log_hist_coarseness \fR=\fPint | |
523bad63 TK |
2970 | Integer ranging from 0 to 6, defining the coarseness of the resolution of |
2971 | the histogram logs enabled with \fBlog_hist_msec\fR. For each increment | |
2972 | in coarseness, fio outputs half as many bins. Defaults to 0, for which | |
2973 | histogram logs contain 1216 latency bins. See \fBLOG FILE FORMATS\fR section. | |
2974 | .TP | |
2975 | .BI log_max_value \fR=\fPbool | |
2976 | If \fBlog_avg_msec\fR is set, fio logs the average over that window. If | |
2977 | you instead want to log the maximum value, set this option to 1. Defaults to | |
2978 | 0, meaning that averaged values are logged. | |
1e613c9c | 2979 | .TP |
ae588852 | 2980 | .BI log_offset \fR=\fPbool |
523bad63 TK |
2981 | If this is set, the iolog options will include the byte offset for the I/O |
2982 | entry as well as the other data values. Defaults to 0 meaning that | |
2983 | offsets are not present in logs. Also see \fBLOG FILE FORMATS\fR section. | |
ae588852 | 2984 | .TP |
aee2ab67 | 2985 | .BI log_compression \fR=\fPint |
523bad63 TK |
2986 | If this is set, fio will compress the I/O logs as it goes, to keep the |
2987 | memory footprint lower. When a log reaches the specified size, that chunk is | |
2988 | removed and compressed in the background. Given that I/O logs are fairly | |
2989 | highly compressible, this yields a nice memory savings for longer runs. The | |
2990 | downside is that the compression will consume some background CPU cycles, so | |
2991 | it may impact the run. This, however, is also true if the logging ends up | |
2992 | consuming most of the system memory. So pick your poison. The I/O logs are | |
2993 | saved normally at the end of a run, by decompressing the chunks and storing | |
2994 | them in the specified log file. This feature depends on the availability of | |
2995 | zlib. | |
aee2ab67 | 2996 | .TP |
c08f9fe2 | 2997 | .BI log_compression_cpus \fR=\fPstr |
523bad63 TK |
2998 | Define the set of CPUs that are allowed to handle online log compression for |
2999 | the I/O jobs. This can provide better isolation between performance | |
0cf90a62 SW |
3000 | sensitive jobs, and background compression work. See \fBcpus_allowed\fR for |
3001 | the format used. | |
c08f9fe2 | 3002 | .TP |
b26317c9 | 3003 | .BI log_store_compressed \fR=\fPbool |
c08f9fe2 | 3004 | If set, fio will store the log files in a compressed format. They can be |
523bad63 TK |
3005 | decompressed with fio, using the \fB\-\-inflate\-log\fR command line |
3006 | parameter. The files will be stored with a `.fz' suffix. | |
b26317c9 | 3007 | .TP |
3aea75b1 KC |
3008 | .BI log_unix_epoch \fR=\fPbool |
3009 | If set, fio will log Unix timestamps to the log files produced by enabling | |
338f2db5 | 3010 | write_type_log for each log type, instead of the default zero-based |
3aea75b1 KC |
3011 | timestamps. |
3012 | .TP | |
66347cfa | 3013 | .BI block_error_percentiles \fR=\fPbool |
338f2db5 | 3014 | If set, record errors in trim block-sized units from writes and trims and |
523bad63 TK |
3015 | output a histogram of how many trims it took to get to errors, and what kind |
3016 | of error was encountered. | |
d60e92d1 | 3017 | .TP |
523bad63 TK |
3018 | .BI bwavgtime \fR=\fPint |
3019 | Average the calculated bandwidth over the given time. Value is specified in | |
3020 | milliseconds. If the job also does bandwidth logging through | |
3021 | \fBwrite_bw_log\fR, then the minimum of this option and | |
3022 | \fBlog_avg_msec\fR will be used. Default: 500ms. | |
d60e92d1 | 3023 | .TP |
523bad63 TK |
3024 | .BI iopsavgtime \fR=\fPint |
3025 | Average the calculated IOPS over the given time. Value is specified in | |
3026 | milliseconds. If the job also does IOPS logging through | |
3027 | \fBwrite_iops_log\fR, then the minimum of this option and | |
3028 | \fBlog_avg_msec\fR will be used. Default: 500ms. | |
d60e92d1 | 3029 | .TP |
d60e92d1 | 3030 | .BI disk_util \fR=\fPbool |
523bad63 TK |
3031 | Generate disk utilization statistics, if the platform supports it. |
3032 | Default: true. | |
fa769d44 | 3033 | .TP |
523bad63 TK |
3034 | .BI disable_lat \fR=\fPbool |
3035 | Disable measurements of total latency numbers. Useful only for cutting back | |
3036 | the number of calls to \fBgettimeofday\fR\|(2), as that does impact | |
3037 | performance at really high IOPS rates. Note that to really get rid of a | |
3038 | large amount of these calls, this option must be used with | |
3039 | \fBdisable_slat\fR and \fBdisable_bw_measurement\fR as well. | |
9e684a49 | 3040 | .TP |
523bad63 TK |
3041 | .BI disable_clat \fR=\fPbool |
3042 | Disable measurements of completion latency numbers. See | |
3043 | \fBdisable_lat\fR. | |
9e684a49 | 3044 | .TP |
523bad63 TK |
3045 | .BI disable_slat \fR=\fPbool |
3046 | Disable measurements of submission latency numbers. See | |
3047 | \fBdisable_lat\fR. | |
9e684a49 | 3048 | .TP |
523bad63 TK |
3049 | .BI disable_bw_measurement \fR=\fPbool "\fR,\fP disable_bw" \fR=\fPbool |
3050 | Disable measurements of throughput/bandwidth numbers. See | |
3051 | \fBdisable_lat\fR. | |
9e684a49 | 3052 | .TP |
dd39b9ce VF |
3053 | .BI slat_percentiles \fR=\fPbool |
3054 | Report submission latency percentiles. Submission latency is not recorded | |
3055 | for synchronous ioengines. | |
3056 | .TP | |
83349190 | 3057 | .BI clat_percentiles \fR=\fPbool |
dd39b9ce | 3058 | Report completion latency percentiles. |
b599759b JA |
3059 | .TP |
3060 | .BI lat_percentiles \fR=\fPbool | |
dd39b9ce VF |
3061 | Report total latency percentiles. Total latency is the sum of submission |
3062 | latency and completion latency. | |
83349190 YH |
3063 | .TP |
3064 | .BI percentile_list \fR=\fPfloat_list | |
dd39b9ce VF |
3065 | Overwrite the default list of percentiles for latencies and the |
3066 | block error histogram. Each number is a floating point number in the range | |
523bad63 | 3067 | (0,100], and the maximum length of the list is 20. Use ':' to separate the |
dd39b9ce VF |
3068 | numbers. For example, `\-\-percentile_list=99.5:99.9' will cause fio to |
3069 | report the latency durations below which 99.5% and 99.9% of the observed | |
3070 | latencies fell, respectively. | |
e883cb35 JF |
3071 | .TP |
3072 | .BI significant_figures \fR=\fPint | |
c32ba107 JA |
3073 | If using \fB\-\-output\-format\fR of `normal', set the significant figures |
3074 | to this value. Higher values will yield more precise IOPS and throughput | |
3075 | units, while lower values will round. Requires a minimum value of 1 and a | |
e883cb35 | 3076 | maximum value of 10. Defaults to 4. |
523bad63 | 3077 | .SS "Error handling" |
e4585935 | 3078 | .TP |
523bad63 TK |
3079 | .BI exitall_on_error |
3080 | When one job finishes in error, terminate the rest. The default is to wait | |
3081 | for each job to finish. | |
e4585935 | 3082 | .TP |
523bad63 TK |
3083 | .BI continue_on_error \fR=\fPstr |
3084 | Normally fio will exit the job on the first observed failure. If this option | |
338f2db5 | 3085 | is set, fio will continue the job when there is a 'non-fatal error' (EIO or |
523bad63 TK |
3086 | EILSEQ) until the runtime is exceeded or the I/O size specified is |
3087 | completed. If this option is used, there are two more stats that are | |
3088 | appended, the total error count and the first error. The error field given | |
3089 | in the stats is the first error that was hit during the run. | |
3090 | The allowed values are: | |
3091 | .RS | |
3092 | .RS | |
046395d7 | 3093 | .TP |
523bad63 TK |
3094 | .B none |
3095 | Exit on any I/O or verify errors. | |
de890a1e | 3096 | .TP |
523bad63 TK |
3097 | .B read |
3098 | Continue on read errors, exit on all others. | |
2cafffbe | 3099 | .TP |
523bad63 TK |
3100 | .B write |
3101 | Continue on write errors, exit on all others. | |
a0679ce5 | 3102 | .TP |
523bad63 TK |
3103 | .B io |
3104 | Continue on any I/O error, exit on all others. | |
de890a1e | 3105 | .TP |
523bad63 TK |
3106 | .B verify |
3107 | Continue on verify errors, exit on all others. | |
de890a1e | 3108 | .TP |
523bad63 TK |
3109 | .B all |
3110 | Continue on all errors. | |
b93b6a2e | 3111 | .TP |
523bad63 | 3112 | .B 0 |
338f2db5 | 3113 | Backward-compatible alias for 'none'. |
d3a623de | 3114 | .TP |
523bad63 | 3115 | .B 1 |
338f2db5 | 3116 | Backward-compatible alias for 'all'. |
523bad63 TK |
3117 | .RE |
3118 | .RE | |
1d360ffb | 3119 | .TP |
523bad63 TK |
3120 | .BI ignore_error \fR=\fPstr |
3121 | Sometimes you want to ignore some errors during test in that case you can | |
3122 | specify error list for each error type, instead of only being able to | |
338f2db5 | 3123 | ignore the default 'non-fatal error' using \fBcontinue_on_error\fR. |
523bad63 TK |
3124 | `ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST' errors for |
3125 | given error type is separated with ':'. Error may be symbol ('ENOSPC', 'ENOMEM') | |
3126 | or integer. Example: | |
de890a1e SL |
3127 | .RS |
3128 | .RS | |
523bad63 TK |
3129 | .P |
3130 | ignore_error=EAGAIN,ENOSPC:122 | |
3131 | .RE | |
3132 | .P | |
3133 | This option will ignore EAGAIN from READ, and ENOSPC and 122(EDQUOT) from | |
3134 | WRITE. This option works by overriding \fBcontinue_on_error\fR with | |
3135 | the list of errors for each error type if any. | |
3136 | .RE | |
de890a1e | 3137 | .TP |
523bad63 TK |
3138 | .BI error_dump \fR=\fPbool |
3139 | If set dump every error even if it is non fatal, true by default. If | |
3140 | disabled only fatal error will be dumped. | |
3141 | .SS "Running predefined workloads" | |
3142 | Fio includes predefined profiles that mimic the I/O workloads generated by | |
3143 | other tools. | |
49ccb8c1 | 3144 | .TP |
523bad63 TK |
3145 | .BI profile \fR=\fPstr |
3146 | The predefined workload to run. Current profiles are: | |
3147 | .RS | |
3148 | .RS | |
de890a1e | 3149 | .TP |
523bad63 TK |
3150 | .B tiobench |
3151 | Threaded I/O bench (tiotest/tiobench) like workload. | |
49ccb8c1 | 3152 | .TP |
523bad63 TK |
3153 | .B act |
3154 | Aerospike Certification Tool (ACT) like workload. | |
3155 | .RE | |
de890a1e SL |
3156 | .RE |
3157 | .P | |
523bad63 TK |
3158 | To view a profile's additional options use \fB\-\-cmdhelp\fR after specifying |
3159 | the profile. For example: | |
3160 | .RS | |
3161 | .TP | |
3162 | $ fio \-\-profile=act \-\-cmdhelp | |
de890a1e | 3163 | .RE |
523bad63 | 3164 | .SS "Act profile options" |
de890a1e | 3165 | .TP |
523bad63 TK |
3166 | .BI device\-names \fR=\fPstr |
3167 | Devices to use. | |
d54fce84 | 3168 | .TP |
523bad63 TK |
3169 | .BI load \fR=\fPint |
3170 | ACT load multiplier. Default: 1. | |
7aeb1e94 | 3171 | .TP |
523bad63 TK |
3172 | .BI test\-duration\fR=\fPtime |
3173 | How long the entire test takes to run. When the unit is omitted, the value | |
3174 | is given in seconds. Default: 24h. | |
1008602c | 3175 | .TP |
523bad63 TK |
3176 | .BI threads\-per\-queue\fR=\fPint |
3177 | Number of read I/O threads per device. Default: 8. | |
e5f34d95 | 3178 | .TP |
523bad63 TK |
3179 | .BI read\-req\-num\-512\-blocks\fR=\fPint |
3180 | Number of 512B blocks to read at the time. Default: 3. | |
d54fce84 | 3181 | .TP |
523bad63 TK |
3182 | .BI large\-block\-op\-kbytes\fR=\fPint |
3183 | Size of large block ops in KiB (writes). Default: 131072. | |
d54fce84 | 3184 | .TP |
523bad63 TK |
3185 | .BI prep |
3186 | Set to run ACT prep phase. | |
3187 | .SS "Tiobench profile options" | |
6d500c2e | 3188 | .TP |
523bad63 TK |
3189 | .BI size\fR=\fPstr |
3190 | Size in MiB. | |
0d978694 | 3191 | .TP |
523bad63 TK |
3192 | .BI block\fR=\fPint |
3193 | Block size in bytes. Default: 4096. | |
0d978694 | 3194 | .TP |
523bad63 TK |
3195 | .BI numruns\fR=\fPint |
3196 | Number of runs. | |
0d978694 | 3197 | .TP |
523bad63 TK |
3198 | .BI dir\fR=\fPstr |
3199 | Test directory. | |
65fa28ca | 3200 | .TP |
523bad63 TK |
3201 | .BI threads\fR=\fPint |
3202 | Number of threads. | |
d60e92d1 | 3203 | .SH OUTPUT |
40943b9a TK |
3204 | Fio spits out a lot of output. While running, fio will display the status of the |
3205 | jobs created. An example of that would be: | |
d60e92d1 | 3206 | .P |
40943b9a TK |
3207 | .nf |
3208 | Jobs: 1 (f=1): [_(1),M(1)][24.8%][r=20.5MiB/s,w=23.5MiB/s][r=82,w=94 IOPS][eta 01m:31s] | |
3209 | .fi | |
d1429b5c | 3210 | .P |
40943b9a TK |
3211 | The characters inside the first set of square brackets denote the current status of |
3212 | each thread. The first character is the first job defined in the job file, and so | |
3213 | forth. The possible values (in typical life cycle order) are: | |
d60e92d1 AC |
3214 | .RS |
3215 | .TP | |
40943b9a | 3216 | .PD 0 |
d60e92d1 | 3217 | .B P |
40943b9a | 3218 | Thread setup, but not started. |
d60e92d1 AC |
3219 | .TP |
3220 | .B C | |
3221 | Thread created. | |
3222 | .TP | |
3223 | .B I | |
40943b9a TK |
3224 | Thread initialized, waiting or generating necessary data. |
3225 | .TP | |
522c29f6 | 3226 | .B p |
338f2db5 | 3227 | Thread running pre-reading file(s). |
40943b9a TK |
3228 | .TP |
3229 | .B / | |
3230 | Thread is in ramp period. | |
d60e92d1 AC |
3231 | .TP |
3232 | .B R | |
3233 | Running, doing sequential reads. | |
3234 | .TP | |
3235 | .B r | |
3236 | Running, doing random reads. | |
3237 | .TP | |
3238 | .B W | |
3239 | Running, doing sequential writes. | |
3240 | .TP | |
3241 | .B w | |
3242 | Running, doing random writes. | |
3243 | .TP | |
3244 | .B M | |
3245 | Running, doing mixed sequential reads/writes. | |
3246 | .TP | |
3247 | .B m | |
3248 | Running, doing mixed random reads/writes. | |
3249 | .TP | |
40943b9a TK |
3250 | .B D |
3251 | Running, doing sequential trims. | |
3252 | .TP | |
3253 | .B d | |
3254 | Running, doing random trims. | |
3255 | .TP | |
d60e92d1 AC |
3256 | .B F |
3257 | Running, currently waiting for \fBfsync\fR\|(2). | |
3258 | .TP | |
3259 | .B V | |
40943b9a TK |
3260 | Running, doing verification of written data. |
3261 | .TP | |
3262 | .B f | |
3263 | Thread finishing. | |
d60e92d1 AC |
3264 | .TP |
3265 | .B E | |
40943b9a | 3266 | Thread exited, not reaped by main thread yet. |
d60e92d1 AC |
3267 | .TP |
3268 | .B \- | |
40943b9a TK |
3269 | Thread reaped. |
3270 | .TP | |
3271 | .B X | |
3272 | Thread reaped, exited with an error. | |
3273 | .TP | |
3274 | .B K | |
3275 | Thread reaped, exited due to signal. | |
d1429b5c | 3276 | .PD |
40943b9a TK |
3277 | .RE |
3278 | .P | |
3279 | Fio will condense the thread string as not to take up more space on the command | |
3280 | line than needed. For instance, if you have 10 readers and 10 writers running, | |
3281 | the output would look like this: | |
3282 | .P | |
3283 | .nf | |
3284 | Jobs: 20 (f=20): [R(10),W(10)][4.0%][r=20.5MiB/s,w=23.5MiB/s][r=82,w=94 IOPS][eta 57m:36s] | |
3285 | .fi | |
d60e92d1 | 3286 | .P |
40943b9a TK |
3287 | Note that the status string is displayed in order, so it's possible to tell which of |
3288 | the jobs are currently doing what. In the example above this means that jobs 1\-\-10 | |
3289 | are readers and 11\-\-20 are writers. | |
d60e92d1 | 3290 | .P |
40943b9a TK |
3291 | The other values are fairly self explanatory \-\- number of threads currently |
3292 | running and doing I/O, the number of currently open files (f=), the estimated | |
3293 | completion percentage, the rate of I/O since last check (read speed listed first, | |
3294 | then write speed and optionally trim speed) in terms of bandwidth and IOPS, | |
3295 | and time to completion for the current running group. It's impossible to estimate | |
3296 | runtime of the following groups (if any). | |
d60e92d1 | 3297 | .P |
40943b9a TK |
3298 | When fio is done (or interrupted by Ctrl\-C), it will show the data for |
3299 | each thread, group of threads, and disks in that order. For each overall thread (or | |
3300 | group) the output looks like: | |
3301 | .P | |
3302 | .nf | |
3303 | Client1: (groupid=0, jobs=1): err= 0: pid=16109: Sat Jun 24 12:07:54 2017 | |
3304 | write: IOPS=88, BW=623KiB/s (638kB/s)(30.4MiB/50032msec) | |
3305 | slat (nsec): min=500, max=145500, avg=8318.00, stdev=4781.50 | |
3306 | clat (usec): min=170, max=78367, avg=4019.02, stdev=8293.31 | |
3307 | lat (usec): min=174, max=78375, avg=4027.34, stdev=8291.79 | |
3308 | clat percentiles (usec): | |
3309 | | 1.00th=[ 302], 5.00th=[ 326], 10.00th=[ 343], 20.00th=[ 363], | |
3310 | | 30.00th=[ 392], 40.00th=[ 404], 50.00th=[ 416], 60.00th=[ 445], | |
3311 | | 70.00th=[ 816], 80.00th=[ 6718], 90.00th=[12911], 95.00th=[21627], | |
3312 | | 99.00th=[43779], 99.50th=[51643], 99.90th=[68682], 99.95th=[72877], | |
3313 | | 99.99th=[78119] | |
3314 | bw ( KiB/s): min= 532, max= 686, per=0.10%, avg=622.87, stdev=24.82, samples= 100 | |
3315 | iops : min= 76, max= 98, avg=88.98, stdev= 3.54, samples= 100 | |
d3b9694d VF |
3316 | lat (usec) : 250=0.04%, 500=64.11%, 750=4.81%, 1000=2.79% |
3317 | lat (msec) : 2=4.16%, 4=1.84%, 10=4.90%, 20=11.33%, 50=5.37% | |
3318 | lat (msec) : 100=0.65% | |
40943b9a TK |
3319 | cpu : usr=0.27%, sys=0.18%, ctx=12072, majf=0, minf=21 |
3320 | IO depths : 1=85.0%, 2=13.1%, 4=1.8%, 8=0.1%, 16=0.0%, 32=0.0%, >=64=0.0% | |
3321 | submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0% | |
3322 | complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0% | |
3323 | issued rwt: total=0,4450,0, short=0,0,0, dropped=0,0,0 | |
3324 | latency : target=0, window=0, percentile=100.00%, depth=8 | |
3325 | .fi | |
3326 | .P | |
3327 | The job name (or first job's name when using \fBgroup_reporting\fR) is printed, | |
3328 | along with the group id, count of jobs being aggregated, last error id seen (which | |
3329 | is 0 when there are no errors), pid/tid of that thread and the time the job/group | |
3330 | completed. Below are the I/O statistics for each data direction performed (showing | |
3331 | writes in the example above). In the order listed, they denote: | |
d60e92d1 | 3332 | .RS |
d60e92d1 | 3333 | .TP |
40943b9a TK |
3334 | .B read/write/trim |
3335 | The string before the colon shows the I/O direction the statistics | |
3336 | are for. \fIIOPS\fR is the average I/Os performed per second. \fIBW\fR | |
3337 | is the average bandwidth rate shown as: value in power of 2 format | |
3338 | (value in power of 10 format). The last two values show: (total | |
3339 | I/O performed in power of 2 format / \fIruntime\fR of that thread). | |
d60e92d1 AC |
3340 | .TP |
3341 | .B slat | |
40943b9a TK |
3342 | Submission latency (\fImin\fR being the minimum, \fImax\fR being the |
3343 | maximum, \fIavg\fR being the average, \fIstdev\fR being the standard | |
3344 | deviation). This is the time it took to submit the I/O. For | |
3345 | sync I/O this row is not displayed as the slat is really the | |
3346 | completion latency (since queue/complete is one operation there). | |
3347 | This value can be in nanoseconds, microseconds or milliseconds \-\-\- | |
3348 | fio will choose the most appropriate base and print that (in the | |
3349 | example above nanoseconds was the best scale). Note: in \fB\-\-minimal\fR mode | |
3350 | latencies are always expressed in microseconds. | |
d60e92d1 AC |
3351 | .TP |
3352 | .B clat | |
40943b9a TK |
3353 | Completion latency. Same names as slat, this denotes the time from |
3354 | submission to completion of the I/O pieces. For sync I/O, clat will | |
3355 | usually be equal (or very close) to 0, as the time from submit to | |
3356 | complete is basically just CPU time (I/O has already been done, see slat | |
3357 | explanation). | |
d60e92d1 | 3358 | .TP |
d3b9694d VF |
3359 | .B lat |
3360 | Total latency. Same names as slat and clat, this denotes the time from | |
3361 | when fio created the I/O unit to completion of the I/O operation. | |
3362 | .TP | |
d60e92d1 | 3363 | .B bw |
40943b9a TK |
3364 | Bandwidth statistics based on samples. Same names as the xlat stats, |
3365 | but also includes the number of samples taken (\fIsamples\fR) and an | |
3366 | approximate percentage of total aggregate bandwidth this thread | |
3367 | received in its group (\fIper\fR). This last value is only really | |
3368 | useful if the threads in this group are on the same disk, since they | |
3369 | are then competing for disk access. | |
3370 | .TP | |
3371 | .B iops | |
3372 | IOPS statistics based on samples. Same names as \fBbw\fR. | |
d60e92d1 | 3373 | .TP |
d3b9694d VF |
3374 | .B lat (nsec/usec/msec) |
3375 | The distribution of I/O completion latencies. This is the time from when | |
3376 | I/O leaves fio and when it gets completed. Unlike the separate | |
3377 | read/write/trim sections above, the data here and in the remaining | |
3378 | sections apply to all I/Os for the reporting group. 250=0.04% means that | |
3379 | 0.04% of the I/Os completed in under 250us. 500=64.11% means that 64.11% | |
3380 | of the I/Os required 250 to 499us for completion. | |
3381 | .TP | |
d60e92d1 | 3382 | .B cpu |
40943b9a TK |
3383 | CPU usage. User and system time, along with the number of context |
3384 | switches this thread went through, usage of system and user time, and | |
3385 | finally the number of major and minor page faults. The CPU utilization | |
3386 | numbers are averages for the jobs in that reporting group, while the | |
3387 | context and fault counters are summed. | |
d60e92d1 AC |
3388 | .TP |
3389 | .B IO depths | |
40943b9a TK |
3390 | The distribution of I/O depths over the job lifetime. The numbers are |
3391 | divided into powers of 2 and each entry covers depths from that value | |
3392 | up to those that are lower than the next entry \-\- e.g., 16= covers | |
3393 | depths from 16 to 31. Note that the range covered by a depth | |
3394 | distribution entry can be different to the range covered by the | |
3395 | equivalent \fBsubmit\fR/\fBcomplete\fR distribution entry. | |
3396 | .TP | |
3397 | .B IO submit | |
3398 | How many pieces of I/O were submitting in a single submit call. Each | |
3399 | entry denotes that amount and below, until the previous entry \-\- e.g., | |
3400 | 16=100% means that we submitted anywhere between 9 to 16 I/Os per submit | |
3401 | call. Note that the range covered by a \fBsubmit\fR distribution entry can | |
3402 | be different to the range covered by the equivalent depth distribution | |
3403 | entry. | |
3404 | .TP | |
3405 | .B IO complete | |
3406 | Like the above \fBsubmit\fR number, but for completions instead. | |
3407 | .TP | |
3408 | .B IO issued rwt | |
3409 | The number of \fBread/write/trim\fR requests issued, and how many of them were | |
3410 | short or dropped. | |
d60e92d1 | 3411 | .TP |
d3b9694d | 3412 | .B IO latency |
ee21ebee | 3413 | These values are for \fBlatency_target\fR and related options. When |
d3b9694d VF |
3414 | these options are engaged, this section describes the I/O depth required |
3415 | to meet the specified latency target. | |
d60e92d1 | 3416 | .RE |
d60e92d1 | 3417 | .P |
40943b9a TK |
3418 | After each client has been listed, the group statistics are printed. They |
3419 | will look like this: | |
3420 | .P | |
3421 | .nf | |
3422 | Run status group 0 (all jobs): | |
3423 | READ: bw=20.9MiB/s (21.9MB/s), 10.4MiB/s\-10.8MiB/s (10.9MB/s\-11.3MB/s), io=64.0MiB (67.1MB), run=2973\-3069msec | |
3424 | WRITE: bw=1231KiB/s (1261kB/s), 616KiB/s\-621KiB/s (630kB/s\-636kB/s), io=64.0MiB (67.1MB), run=52747\-53223msec | |
3425 | .fi | |
3426 | .P | |
3427 | For each data direction it prints: | |
d60e92d1 AC |
3428 | .RS |
3429 | .TP | |
40943b9a TK |
3430 | .B bw |
3431 | Aggregate bandwidth of threads in this group followed by the | |
3432 | minimum and maximum bandwidth of all the threads in this group. | |
338f2db5 SW |
3433 | Values outside of brackets are power-of-2 format and those |
3434 | within are the equivalent value in a power-of-10 format. | |
d60e92d1 | 3435 | .TP |
40943b9a TK |
3436 | .B io |
3437 | Aggregate I/O performed of all threads in this group. The | |
3438 | format is the same as \fBbw\fR. | |
d60e92d1 | 3439 | .TP |
40943b9a TK |
3440 | .B run |
3441 | The smallest and longest runtimes of the threads in this group. | |
d60e92d1 | 3442 | .RE |
d60e92d1 | 3443 | .P |
40943b9a TK |
3444 | And finally, the disk statistics are printed. This is Linux specific. |
3445 | They will look like this: | |
3446 | .P | |
3447 | .nf | |
3448 | Disk stats (read/write): | |
3449 | sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00% | |
3450 | .fi | |
3451 | .P | |
3452 | Each value is printed for both reads and writes, with reads first. The | |
3453 | numbers denote: | |
d60e92d1 AC |
3454 | .RS |
3455 | .TP | |
3456 | .B ios | |
3457 | Number of I/Os performed by all groups. | |
3458 | .TP | |
3459 | .B merge | |
007c7be9 | 3460 | Number of merges performed by the I/O scheduler. |
d60e92d1 AC |
3461 | .TP |
3462 | .B ticks | |
3463 | Number of ticks we kept the disk busy. | |
3464 | .TP | |
40943b9a | 3465 | .B in_queue |
d60e92d1 AC |
3466 | Total time spent in the disk queue. |
3467 | .TP | |
3468 | .B util | |
40943b9a TK |
3469 | The disk utilization. A value of 100% means we kept the disk |
3470 | busy constantly, 50% would be a disk idling half of the time. | |
d60e92d1 | 3471 | .RE |
8423bd11 | 3472 | .P |
40943b9a TK |
3473 | It is also possible to get fio to dump the current output while it is running, |
3474 | without terminating the job. To do that, send fio the USR1 signal. You can | |
3475 | also get regularly timed dumps by using the \fB\-\-status\-interval\fR | |
3476 | parameter, or by creating a file in `/tmp' named | |
3477 | `fio\-dump\-status'. If fio sees this file, it will unlink it and dump the | |
3478 | current output status. | |
d60e92d1 | 3479 | .SH TERSE OUTPUT |
40943b9a TK |
3480 | For scripted usage where you typically want to generate tables or graphs of the |
3481 | results, fio can output the results in a semicolon separated format. The format | |
3482 | is one long line of values, such as: | |
d60e92d1 | 3483 | .P |
40943b9a TK |
3484 | .nf |
3485 | 2;card0;0;0;7139336;121836;60004;1;10109;27.932460;116.933948;220;126861;3495.446807;1085.368601;226;126864;3523.635629;1089.012448;24063;99944;50.275485%;59818.274627;5540.657370;7155060;122104;60004;1;8338;29.086342;117.839068;388;128077;5032.488518;1234.785715;391;128085;5061.839412;1236.909129;23436;100928;50.287926%;59964.832030;5644.844189;14.595833%;19.394167%;123706;0;7313;0.1%;0.1%;0.1%;0.1%;0.1%;0.1%;100.0%;0.00%;0.00%;0.00%;0.00%;0.00%;0.00%;0.01%;0.02%;0.05%;0.16%;6.04%;40.40%;52.68%;0.64%;0.01%;0.00%;0.01%;0.00%;0.00%;0.00%;0.00%;0.00% | |
3486 | A description of this job goes here. | |
3487 | .fi | |
d60e92d1 | 3488 | .P |
4e757af1 VF |
3489 | The job description (if provided) follows on a second line for terse v2. |
3490 | It appears on the same line for other terse versions. | |
d60e92d1 | 3491 | .P |
40943b9a TK |
3492 | To enable terse output, use the \fB\-\-minimal\fR or |
3493 | `\-\-output\-format=terse' command line options. The | |
3494 | first value is the version of the terse output format. If the output has to be | |
3495 | changed for some reason, this number will be incremented by 1 to signify that | |
3496 | change. | |
d60e92d1 | 3497 | .P |
40943b9a TK |
3498 | Split up, the format is as follows (comments in brackets denote when a |
3499 | field was introduced or whether it's specific to some terse version): | |
d60e92d1 | 3500 | .P |
40943b9a TK |
3501 | .nf |
3502 | terse version, fio version [v3], jobname, groupid, error | |
3503 | .fi | |
525c2bfa | 3504 | .RS |
40943b9a TK |
3505 | .P |
3506 | .B | |
3507 | READ status: | |
525c2bfa | 3508 | .RE |
40943b9a TK |
3509 | .P |
3510 | .nf | |
3511 | Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec) | |
3512 | Submission latency: min, max, mean, stdev (usec) | |
3513 | Completion latency: min, max, mean, stdev (usec) | |
3514 | Completion latency percentiles: 20 fields (see below) | |
3515 | Total latency: min, max, mean, stdev (usec) | |
3516 | Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev, number of samples [v5] | |
3517 | IOPS [v5]: min, max, mean, stdev, number of samples | |
3518 | .fi | |
d60e92d1 | 3519 | .RS |
40943b9a TK |
3520 | .P |
3521 | .B | |
3522 | WRITE status: | |
a2c95580 | 3523 | .RE |
40943b9a TK |
3524 | .P |
3525 | .nf | |
3526 | Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec) | |
3527 | Submission latency: min, max, mean, stdev (usec) | |
3528 | Completion latency: min, max, mean, stdev (usec) | |
3529 | Completion latency percentiles: 20 fields (see below) | |
3530 | Total latency: min, max, mean, stdev (usec) | |
3531 | Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev, number of samples [v5] | |
3532 | IOPS [v5]: min, max, mean, stdev, number of samples | |
3533 | .fi | |
a2c95580 | 3534 | .RS |
40943b9a TK |
3535 | .P |
3536 | .B | |
3537 | TRIM status [all but version 3]: | |
d60e92d1 AC |
3538 | .RE |
3539 | .P | |
40943b9a TK |
3540 | .nf |
3541 | Fields are similar to \fBREAD/WRITE\fR status. | |
3542 | .fi | |
a2c95580 | 3543 | .RS |
a2c95580 | 3544 | .P |
40943b9a | 3545 | .B |
d1429b5c | 3546 | CPU usage: |
d60e92d1 AC |
3547 | .RE |
3548 | .P | |
40943b9a TK |
3549 | .nf |
3550 | user, system, context switches, major faults, minor faults | |
3551 | .fi | |
d60e92d1 | 3552 | .RS |
40943b9a TK |
3553 | .P |
3554 | .B | |
3555 | I/O depths: | |
d60e92d1 AC |
3556 | .RE |
3557 | .P | |
40943b9a TK |
3558 | .nf |
3559 | <=1, 2, 4, 8, 16, 32, >=64 | |
3560 | .fi | |
562c2d2f | 3561 | .RS |
40943b9a TK |
3562 | .P |
3563 | .B | |
3564 | I/O latencies microseconds: | |
562c2d2f | 3565 | .RE |
40943b9a TK |
3566 | .P |
3567 | .nf | |
3568 | <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000 | |
3569 | .fi | |
562c2d2f | 3570 | .RS |
40943b9a TK |
3571 | .P |
3572 | .B | |
3573 | I/O latencies milliseconds: | |
562c2d2f DN |
3574 | .RE |
3575 | .P | |
40943b9a TK |
3576 | .nf |
3577 | <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000 | |
3578 | .fi | |
f2f788dd | 3579 | .RS |
40943b9a TK |
3580 | .P |
3581 | .B | |
3582 | Disk utilization [v3]: | |
f2f788dd JA |
3583 | .RE |
3584 | .P | |
40943b9a TK |
3585 | .nf |
3586 | disk name, read ios, write ios, read merges, write merges, read ticks, write ticks, time spent in queue, disk utilization percentage | |
3587 | .fi | |
562c2d2f | 3588 | .RS |
d60e92d1 | 3589 | .P |
40943b9a TK |
3590 | .B |
3591 | Additional Info (dependent on continue_on_error, default off): | |
d60e92d1 | 3592 | .RE |
2fc26c3d | 3593 | .P |
40943b9a TK |
3594 | .nf |
3595 | total # errors, first error code | |
3596 | .fi | |
2fc26c3d IC |
3597 | .RS |
3598 | .P | |
40943b9a TK |
3599 | .B |
3600 | Additional Info (dependent on description being set): | |
3601 | .RE | |
3602 | .P | |
2fc26c3d | 3603 | .nf |
40943b9a TK |
3604 | Text description |
3605 | .fi | |
3606 | .P | |
3607 | Completion latency percentiles can be a grouping of up to 20 sets, so for the | |
3608 | terse output fio writes all of them. Each field will look like this: | |
3609 | .P | |
3610 | .nf | |
3611 | 1.00%=6112 | |
3612 | .fi | |
3613 | .P | |
3614 | which is the Xth percentile, and the `usec' latency associated with it. | |
3615 | .P | |
3616 | For \fBDisk utilization\fR, all disks used by fio are shown. So for each disk there | |
3617 | will be a disk utilization section. | |
3618 | .P | |
3619 | Below is a single line containing short names for each of the fields in the | |
3620 | minimal output v3, separated by semicolons: | |
3621 | .P | |
3622 | .nf | |
3623 | terse_version_3;fio_version;jobname;groupid;error;read_kb;read_bandwidth;read_iops;read_runtime_ms;read_slat_min;read_slat_max;read_slat_mean;read_slat_dev;read_clat_min;read_clat_max;read_clat_mean;read_clat_dev;read_clat_pct01;read_clat_pct02;read_clat_pct03;read_clat_pct04;read_clat_pct05;read_clat_pct06;read_clat_pct07;read_clat_pct08;read_clat_pct09;read_clat_pct10;read_clat_pct11;read_clat_pct12;read_clat_pct13;read_clat_pct14;read_clat_pct15;read_clat_pct16;read_clat_pct17;read_clat_pct18;read_clat_pct19;read_clat_pct20;read_tlat_min;read_lat_max;read_lat_mean;read_lat_dev;read_bw_min;read_bw_max;read_bw_agg_pct;read_bw_mean;read_bw_dev;write_kb;write_bandwidth;write_iops;write_runtime_ms;write_slat_min;write_slat_max;write_slat_mean;write_slat_dev;write_clat_min;write_clat_max;write_clat_mean;write_clat_dev;write_clat_pct01;write_clat_pct02;write_clat_pct03;write_clat_pct04;write_clat_pct05;write_clat_pct06;write_clat_pct07;write_clat_pct08;write_clat_pct09;write_clat_pct10;write_clat_pct11;write_clat_pct12;write_clat_pct13;write_clat_pct14;write_clat_pct15;write_clat_pct16;write_clat_pct17;write_clat_pct18;write_clat_pct19;write_clat_pct20;write_tlat_min;write_lat_max;write_lat_mean;write_lat_dev;write_bw_min;write_bw_max;write_bw_agg_pct;write_bw_mean;write_bw_dev;cpu_user;cpu_sys;cpu_csw;cpu_mjf;cpu_minf;iodepth_1;iodepth_2;iodepth_4;iodepth_8;iodepth_16;iodepth_32;iodepth_64;lat_2us;lat_4us;lat_10us;lat_20us;lat_50us;lat_100us;lat_250us;lat_500us;lat_750us;lat_1000us;lat_2ms;lat_4ms;lat_10ms;lat_20ms;lat_50ms;lat_100ms;lat_250ms;lat_500ms;lat_750ms;lat_1000ms;lat_2000ms;lat_over_2000ms;disk_name;disk_read_iops;disk_write_iops;disk_read_merges;disk_write_merges;disk_read_ticks;write_ticks;disk_queue_time;disk_util | |
2fc26c3d | 3624 | .fi |
4e757af1 VF |
3625 | .P |
3626 | In client/server mode terse output differs from what appears when jobs are run | |
3627 | locally. Disk utilization data is omitted from the standard terse output and | |
3628 | for v3 and later appears on its own separate line at the end of each terse | |
3629 | reporting cycle. | |
44c82dba VF |
3630 | .SH JSON OUTPUT |
3631 | The \fBjson\fR output format is intended to be both human readable and convenient | |
3632 | for automated parsing. For the most part its sections mirror those of the | |
3633 | \fBnormal\fR output. The \fBruntime\fR value is reported in msec and the \fBbw\fR value is | |
3634 | reported in 1024 bytes per second units. | |
3635 | .fi | |
d9e557ab VF |
3636 | .SH JSON+ OUTPUT |
3637 | The \fBjson+\fR output format is identical to the \fBjson\fR output format except that it | |
3638 | adds a full dump of the completion latency bins. Each \fBbins\fR object contains a | |
3639 | set of (key, value) pairs where keys are latency durations and values count how | |
3640 | many I/Os had completion latencies of the corresponding duration. For example, | |
3641 | consider: | |
d9e557ab | 3642 | .RS |
40943b9a | 3643 | .P |
d9e557ab VF |
3644 | "bins" : { "87552" : 1, "89600" : 1, "94720" : 1, "96768" : 1, "97792" : 1, "99840" : 1, "100864" : 2, "103936" : 6, "104960" : 534, "105984" : 5995, "107008" : 7529, ... } |
3645 | .RE | |
40943b9a | 3646 | .P |
d9e557ab VF |
3647 | This data indicates that one I/O required 87,552ns to complete, two I/Os required |
3648 | 100,864ns to complete, and 7529 I/Os required 107,008ns to complete. | |
40943b9a | 3649 | .P |
d9e557ab | 3650 | Also included with fio is a Python script \fBfio_jsonplus_clat2csv\fR that takes |
338f2db5 | 3651 | json+ output and generates CSV-formatted latency data suitable for plotting. |
40943b9a | 3652 | .P |
d9e557ab | 3653 | The latency durations actually represent the midpoints of latency intervals. |
40943b9a | 3654 | For details refer to `stat.h' in the fio source. |
29dbd1e5 | 3655 | .SH TRACE FILE FORMAT |
40943b9a TK |
3656 | There are two trace file format that you can encounter. The older (v1) format is |
3657 | unsupported since version 1.20\-rc3 (March 2008). It will still be described | |
29dbd1e5 | 3658 | below in case that you get an old trace and want to understand it. |
29dbd1e5 | 3659 | .P |
40943b9a TK |
3660 | In any case the trace is a simple text file with a single action per line. |
3661 | .TP | |
29dbd1e5 | 3662 | .B Trace file format v1 |
40943b9a | 3663 | Each line represents a single I/O action in the following format: |
29dbd1e5 | 3664 | .RS |
40943b9a TK |
3665 | .RS |
3666 | .P | |
29dbd1e5 | 3667 | rw, offset, length |
29dbd1e5 JA |
3668 | .RE |
3669 | .P | |
40943b9a TK |
3670 | where `rw=0/1' for read/write, and the `offset' and `length' entries being in bytes. |
3671 | .P | |
3672 | This format is not supported in fio versions >= 1.20\-rc3. | |
3673 | .RE | |
3674 | .TP | |
29dbd1e5 | 3675 | .B Trace file format v2 |
40943b9a TK |
3676 | The second version of the trace file format was added in fio version 1.17. It |
3677 | allows to access more then one file per trace and has a bigger set of possible | |
3678 | file actions. | |
29dbd1e5 | 3679 | .RS |
40943b9a | 3680 | .P |
29dbd1e5 | 3681 | The first line of the trace file has to be: |
40943b9a TK |
3682 | .RS |
3683 | .P | |
3684 | "fio version 2 iolog" | |
3685 | .RE | |
3686 | .P | |
29dbd1e5 | 3687 | Following this can be lines in two different formats, which are described below. |
40943b9a TK |
3688 | .P |
3689 | .B | |
29dbd1e5 | 3690 | The file management format: |
40943b9a TK |
3691 | .RS |
3692 | filename action | |
29dbd1e5 | 3693 | .P |
40943b9a | 3694 | The `filename' is given as an absolute path. The `action' can be one of these: |
29dbd1e5 JA |
3695 | .RS |
3696 | .TP | |
3697 | .B add | |
40943b9a | 3698 | Add the given `filename' to the trace. |
29dbd1e5 JA |
3699 | .TP |
3700 | .B open | |
40943b9a TK |
3701 | Open the file with the given `filename'. The `filename' has to have |
3702 | been added with the \fBadd\fR action before. | |
29dbd1e5 JA |
3703 | .TP |
3704 | .B close | |
40943b9a TK |
3705 | Close the file with the given `filename'. The file has to have been |
3706 | \fBopen\fRed before. | |
3707 | .RE | |
29dbd1e5 | 3708 | .RE |
29dbd1e5 | 3709 | .P |
40943b9a TK |
3710 | .B |
3711 | The file I/O action format: | |
3712 | .RS | |
3713 | filename action offset length | |
29dbd1e5 | 3714 | .P |
40943b9a TK |
3715 | The `filename' is given as an absolute path, and has to have been \fBadd\fRed and |
3716 | \fBopen\fRed before it can be used with this format. The `offset' and `length' are | |
3717 | given in bytes. The `action' can be one of these: | |
29dbd1e5 JA |
3718 | .RS |
3719 | .TP | |
3720 | .B wait | |
40943b9a TK |
3721 | Wait for `offset' microseconds. Everything below 100 is discarded. |
3722 | The time is relative to the previous `wait' statement. | |
29dbd1e5 JA |
3723 | .TP |
3724 | .B read | |
40943b9a | 3725 | Read `length' bytes beginning from `offset'. |
29dbd1e5 JA |
3726 | .TP |
3727 | .B write | |
40943b9a | 3728 | Write `length' bytes beginning from `offset'. |
29dbd1e5 JA |
3729 | .TP |
3730 | .B sync | |
40943b9a | 3731 | \fBfsync\fR\|(2) the file. |
29dbd1e5 JA |
3732 | .TP |
3733 | .B datasync | |
40943b9a | 3734 | \fBfdatasync\fR\|(2) the file. |
29dbd1e5 JA |
3735 | .TP |
3736 | .B trim | |
40943b9a TK |
3737 | Trim the given file from the given `offset' for `length' bytes. |
3738 | .RE | |
29dbd1e5 | 3739 | .RE |
b9921d1a DZ |
3740 | .SH I/O REPLAY \- MERGING TRACES |
3741 | Colocation is a common practice used to get the most out of a machine. | |
3742 | Knowing which workloads play nicely with each other and which ones don't is | |
3743 | a much harder task. While fio can replay workloads concurrently via multiple | |
3744 | jobs, it leaves some variability up to the scheduler making results harder to | |
3745 | reproduce. Merging is a way to make the order of events consistent. | |
3746 | .P | |
3747 | Merging is integrated into I/O replay and done when a \fBmerge_blktrace_file\fR | |
3748 | is specified. The list of files passed to \fBread_iolog\fR go through the merge | |
3749 | process and output a single file stored to the specified file. The output file is | |
3750 | passed on as if it were the only file passed to \fBread_iolog\fR. An example would | |
3751 | look like: | |
3752 | .RS | |
3753 | .P | |
3754 | $ fio \-\-read_iolog="<file1>:<file2>" \-\-merge_blktrace_file="<output_file>" | |
3755 | .RE | |
3756 | .P | |
3757 | Creating only the merged file can be done by passing the command line argument | |
3758 | \fBmerge-blktrace-only\fR. | |
87a48ada DZ |
3759 | .P |
3760 | Scaling traces can be done to see the relative impact of any particular trace | |
3761 | being slowed down or sped up. \fBmerge_blktrace_scalars\fR takes in a colon | |
3762 | separated list of percentage scalars. It is index paired with the files passed | |
3763 | to \fBread_iolog\fR. | |
55bfd8c8 DZ |
3764 | .P |
3765 | With scaling, it may be desirable to match the running time of all traces. | |
3766 | This can be done with \fBmerge_blktrace_iters\fR. It is index paired with | |
3767 | \fBread_iolog\fR just like \fBmerge_blktrace_scalars\fR. | |
3768 | .P | |
3769 | In an example, given two traces, A and B, each 60s long. If we want to see | |
3770 | the impact of trace A issuing IOs twice as fast and repeat trace A over the | |
3771 | runtime of trace B, the following can be done: | |
3772 | .RS | |
3773 | .P | |
3774 | $ fio \-\-read_iolog="<trace_a>:"<trace_b>" \-\-merge_blktrace_file"<output_file>" \-\-merge_blktrace_scalars="50:100" \-\-merge_blktrace_iters="2:1" | |
3775 | .RE | |
3776 | .P | |
3777 | This runs trace A at 2x the speed twice for approximately the same runtime as | |
3778 | a single run of trace B. | |
29dbd1e5 | 3779 | .SH CPU IDLENESS PROFILING |
40943b9a TK |
3780 | In some cases, we want to understand CPU overhead in a test. For example, we |
3781 | test patches for the specific goodness of whether they reduce CPU usage. | |
3782 | Fio implements a balloon approach to create a thread per CPU that runs at idle | |
3783 | priority, meaning that it only runs when nobody else needs the cpu. | |
3784 | By measuring the amount of work completed by the thread, idleness of each CPU | |
3785 | can be derived accordingly. | |
3786 | .P | |
3787 | An unit work is defined as touching a full page of unsigned characters. Mean and | |
3788 | standard deviation of time to complete an unit work is reported in "unit work" | |
3789 | section. Options can be chosen to report detailed percpu idleness or overall | |
3790 | system idleness by aggregating percpu stats. | |
29dbd1e5 | 3791 | .SH VERIFICATION AND TRIGGERS |
40943b9a TK |
3792 | Fio is usually run in one of two ways, when data verification is done. The first |
3793 | is a normal write job of some sort with verify enabled. When the write phase has | |
3794 | completed, fio switches to reads and verifies everything it wrote. The second | |
3795 | model is running just the write phase, and then later on running the same job | |
3796 | (but with reads instead of writes) to repeat the same I/O patterns and verify | |
3797 | the contents. Both of these methods depend on the write phase being completed, | |
3798 | as fio otherwise has no idea how much data was written. | |
3799 | .P | |
3800 | With verification triggers, fio supports dumping the current write state to | |
3801 | local files. Then a subsequent read verify workload can load this state and know | |
3802 | exactly where to stop. This is useful for testing cases where power is cut to a | |
3803 | server in a managed fashion, for instance. | |
3804 | .P | |
29dbd1e5 | 3805 | A verification trigger consists of two things: |
29dbd1e5 | 3806 | .RS |
40943b9a TK |
3807 | .P |
3808 | 1) Storing the write state of each job. | |
3809 | .P | |
3810 | 2) Executing a trigger command. | |
29dbd1e5 | 3811 | .RE |
40943b9a TK |
3812 | .P |
3813 | The write state is relatively small, on the order of hundreds of bytes to single | |
3814 | kilobytes. It contains information on the number of completions done, the last X | |
3815 | completions, etc. | |
3816 | .P | |
3817 | A trigger is invoked either through creation ('touch') of a specified file in | |
3818 | the system, or through a timeout setting. If fio is run with | |
3819 | `\-\-trigger\-file=/tmp/trigger\-file', then it will continually | |
3820 | check for the existence of `/tmp/trigger\-file'. When it sees this file, it | |
3821 | will fire off the trigger (thus saving state, and executing the trigger | |
29dbd1e5 | 3822 | command). |
40943b9a TK |
3823 | .P |
3824 | For client/server runs, there's both a local and remote trigger. If fio is | |
3825 | running as a server backend, it will send the job states back to the client for | |
3826 | safe storage, then execute the remote trigger, if specified. If a local trigger | |
3827 | is specified, the server will still send back the write state, but the client | |
3828 | will then execute the trigger. | |
29dbd1e5 JA |
3829 | .RE |
3830 | .P | |
3831 | .B Verification trigger example | |
3832 | .RS | |
40943b9a TK |
3833 | Let's say we want to run a powercut test on the remote Linux machine 'server'. |
3834 | Our write workload is in `write\-test.fio'. We want to cut power to 'server' at | |
3835 | some point during the run, and we'll run this test from the safety or our local | |
3836 | machine, 'localbox'. On the server, we'll start the fio backend normally: | |
3837 | .RS | |
3838 | .P | |
3839 | server# fio \-\-server | |
3840 | .RE | |
3841 | .P | |
29dbd1e5 | 3842 | and on the client, we'll fire off the workload: |
40943b9a TK |
3843 | .RS |
3844 | .P | |
3845 | localbox$ fio \-\-client=server \-\-trigger\-file=/tmp/my\-trigger \-\-trigger\-remote="bash \-c "echo b > /proc/sysrq\-triger"" | |
3846 | .RE | |
3847 | .P | |
3848 | We set `/tmp/my\-trigger' as the trigger file, and we tell fio to execute: | |
3849 | .RS | |
3850 | .P | |
3851 | echo b > /proc/sysrq\-trigger | |
3852 | .RE | |
3853 | .P | |
3854 | on the server once it has received the trigger and sent us the write state. This | |
3855 | will work, but it's not really cutting power to the server, it's merely | |
3856 | abruptly rebooting it. If we have a remote way of cutting power to the server | |
3857 | through IPMI or similar, we could do that through a local trigger command | |
3858 | instead. Let's assume we have a script that does IPMI reboot of a given hostname, | |
3859 | ipmi\-reboot. On localbox, we could then have run fio with a local trigger | |
3860 | instead: | |
3861 | .RS | |
3862 | .P | |
3863 | localbox$ fio \-\-client=server \-\-trigger\-file=/tmp/my\-trigger \-\-trigger="ipmi\-reboot server" | |
3864 | .RE | |
3865 | .P | |
3866 | For this case, fio would wait for the server to send us the write state, then | |
3867 | execute `ipmi\-reboot server' when that happened. | |
29dbd1e5 JA |
3868 | .RE |
3869 | .P | |
3870 | .B Loading verify state | |
3871 | .RS | |
40943b9a TK |
3872 | To load stored write state, a read verification job file must contain the |
3873 | \fBverify_state_load\fR option. If that is set, fio will load the previously | |
29dbd1e5 | 3874 | stored state. For a local fio run this is done by loading the files directly, |
40943b9a TK |
3875 | and on a client/server run, the server backend will ask the client to send the |
3876 | files over and load them from there. | |
29dbd1e5 | 3877 | .RE |
a3ae5b05 | 3878 | .SH LOG FILE FORMATS |
a3ae5b05 JA |
3879 | Fio supports a variety of log file formats, for logging latencies, bandwidth, |
3880 | and IOPS. The logs share a common format, which looks like this: | |
40943b9a | 3881 | .RS |
a3ae5b05 | 3882 | .P |
1a953d97 PC |
3883 | time (msec), value, data direction, block size (bytes), offset (bytes), |
3884 | command priority | |
40943b9a TK |
3885 | .RE |
3886 | .P | |
3887 | `Time' for the log entry is always in milliseconds. The `value' logged depends | |
3888 | on the type of log, it will be one of the following: | |
3889 | .RS | |
a3ae5b05 JA |
3890 | .TP |
3891 | .B Latency log | |
168bb587 | 3892 | Value is latency in nsecs |
a3ae5b05 JA |
3893 | .TP |
3894 | .B Bandwidth log | |
6d500c2e | 3895 | Value is in KiB/sec |
a3ae5b05 JA |
3896 | .TP |
3897 | .B IOPS log | |
40943b9a TK |
3898 | Value is IOPS |
3899 | .RE | |
a3ae5b05 | 3900 | .P |
40943b9a TK |
3901 | `Data direction' is one of the following: |
3902 | .RS | |
a3ae5b05 JA |
3903 | .TP |
3904 | .B 0 | |
40943b9a | 3905 | I/O is a READ |
a3ae5b05 JA |
3906 | .TP |
3907 | .B 1 | |
40943b9a | 3908 | I/O is a WRITE |
a3ae5b05 JA |
3909 | .TP |
3910 | .B 2 | |
40943b9a | 3911 | I/O is a TRIM |
a3ae5b05 | 3912 | .RE |
40943b9a | 3913 | .P |
15417073 SW |
3914 | The entry's `block size' is always in bytes. The `offset' is the position in bytes |
3915 | from the start of the file for that particular I/O. The logging of the offset can be | |
40943b9a TK |
3916 | toggled with \fBlog_offset\fR. |
3917 | .P | |
1a953d97 PC |
3918 | `Command priority` is 0 for normal priority and 1 for high priority. This is controlled |
3919 | by the ioengine specific \fBcmdprio_percentage\fR. | |
3920 | .P | |
15417073 SW |
3921 | Fio defaults to logging every individual I/O but when windowed logging is set |
3922 | through \fBlog_avg_msec\fR, either the average (by default) or the maximum | |
3923 | (\fBlog_max_value\fR is set) `value' seen over the specified period of time | |
3924 | is recorded. Each `data direction' seen within the window period will aggregate | |
3925 | its values in a separate row. Further, when using windowed logging the `block | |
3926 | size' and `offset' entries will always contain 0. | |
49da1240 | 3927 | .SH CLIENT / SERVER |
338f2db5 | 3928 | Normally fio is invoked as a stand-alone application on the machine where the |
40943b9a TK |
3929 | I/O workload should be generated. However, the backend and frontend of fio can |
3930 | be run separately i.e., the fio server can generate an I/O workload on the "Device | |
3931 | Under Test" while being controlled by a client on another machine. | |
3932 | .P | |
3933 | Start the server on the machine which has access to the storage DUT: | |
3934 | .RS | |
3935 | .P | |
3936 | $ fio \-\-server=args | |
3937 | .RE | |
3938 | .P | |
3939 | where `args' defines what fio listens to. The arguments are of the form | |
3940 | `type,hostname' or `IP,port'. `type' is either `ip' (or ip4) for TCP/IP | |
3941 | v4, `ip6' for TCP/IP v6, or `sock' for a local unix domain socket. | |
3942 | `hostname' is either a hostname or IP address, and `port' is the port to listen | |
3943 | to (only valid for TCP/IP, not a local socket). Some examples: | |
3944 | .RS | |
3945 | .TP | |
e0ee7a8b | 3946 | 1) \fBfio \-\-server\fR |
40943b9a TK |
3947 | Start a fio server, listening on all interfaces on the default port (8765). |
3948 | .TP | |
e0ee7a8b | 3949 | 2) \fBfio \-\-server=ip:hostname,4444\fR |
40943b9a TK |
3950 | Start a fio server, listening on IP belonging to hostname and on port 4444. |
3951 | .TP | |
e0ee7a8b | 3952 | 3) \fBfio \-\-server=ip6:::1,4444\fR |
40943b9a TK |
3953 | Start a fio server, listening on IPv6 localhost ::1 and on port 4444. |
3954 | .TP | |
e0ee7a8b | 3955 | 4) \fBfio \-\-server=,4444\fR |
40943b9a TK |
3956 | Start a fio server, listening on all interfaces on port 4444. |
3957 | .TP | |
e0ee7a8b | 3958 | 5) \fBfio \-\-server=1.2.3.4\fR |
40943b9a TK |
3959 | Start a fio server, listening on IP 1.2.3.4 on the default port. |
3960 | .TP | |
e0ee7a8b | 3961 | 6) \fBfio \-\-server=sock:/tmp/fio.sock\fR |
40943b9a TK |
3962 | Start a fio server, listening on the local socket `/tmp/fio.sock'. |
3963 | .RE | |
3964 | .P | |
3965 | Once a server is running, a "client" can connect to the fio server with: | |
3966 | .RS | |
3967 | .P | |
3968 | $ fio <local\-args> \-\-client=<server> <remote\-args> <job file(s)> | |
3969 | .RE | |
3970 | .P | |
3971 | where `local\-args' are arguments for the client where it is running, `server' | |
3972 | is the connect string, and `remote\-args' and `job file(s)' are sent to the | |
3973 | server. The `server' string follows the same format as it does on the server | |
3974 | side, to allow IP/hostname/socket and port strings. | |
3975 | .P | |
3976 | Fio can connect to multiple servers this way: | |
3977 | .RS | |
3978 | .P | |
3979 | $ fio \-\-client=<server1> <job file(s)> \-\-client=<server2> <job file(s)> | |
3980 | .RE | |
3981 | .P | |
3982 | If the job file is located on the fio server, then you can tell the server to | |
3983 | load a local file as well. This is done by using \fB\-\-remote\-config\fR: | |
3984 | .RS | |
3985 | .P | |
3986 | $ fio \-\-client=server \-\-remote\-config /path/to/file.fio | |
3987 | .RE | |
3988 | .P | |
3989 | Then fio will open this local (to the server) job file instead of being passed | |
3990 | one from the client. | |
3991 | .P | |
ff6bb260 | 3992 | If you have many servers (example: 100 VMs/containers), you can input a pathname |
40943b9a TK |
3993 | of a file containing host IPs/names as the parameter value for the |
3994 | \fB\-\-client\fR option. For example, here is an example `host.list' | |
3995 | file containing 2 hostnames: | |
3996 | .RS | |
3997 | .P | |
3998 | .PD 0 | |
39b5f61e | 3999 | host1.your.dns.domain |
40943b9a | 4000 | .P |
39b5f61e | 4001 | host2.your.dns.domain |
40943b9a TK |
4002 | .PD |
4003 | .RE | |
4004 | .P | |
39b5f61e | 4005 | The fio command would then be: |
40943b9a TK |
4006 | .RS |
4007 | .P | |
4008 | $ fio \-\-client=host.list <job file(s)> | |
4009 | .RE | |
4010 | .P | |
338f2db5 | 4011 | In this mode, you cannot input server-specific parameters or job files \-\- all |
39b5f61e | 4012 | servers receive the same job file. |
40943b9a TK |
4013 | .P |
4014 | In order to let `fio \-\-client' runs use a shared filesystem from multiple | |
4015 | hosts, `fio \-\-client' now prepends the IP address of the server to the | |
4016 | filename. For example, if fio is using the directory `/mnt/nfs/fio' and is | |
4017 | writing filename `fileio.tmp', with a \fB\-\-client\fR `hostfile' | |
4018 | containing two hostnames `h1' and `h2' with IP addresses 192.168.10.120 and | |
4019 | 192.168.10.121, then fio will create two files: | |
4020 | .RS | |
4021 | .P | |
4022 | .PD 0 | |
39b5f61e | 4023 | /mnt/nfs/fio/192.168.10.120.fileio.tmp |
40943b9a | 4024 | .P |
39b5f61e | 4025 | /mnt/nfs/fio/192.168.10.121.fileio.tmp |
40943b9a TK |
4026 | .PD |
4027 | .RE | |
4e757af1 VF |
4028 | .P |
4029 | Terse output in client/server mode will differ slightly from what is produced | |
4030 | when fio is run in stand-alone mode. See the terse output section for details. | |
d60e92d1 AC |
4031 | .SH AUTHORS |
4032 | .B fio | |
d292596c | 4033 | was written by Jens Axboe <axboe@kernel.dk>. |
d1429b5c AC |
4034 | .br |
4035 | This man page was written by Aaron Carroll <aaronc@cse.unsw.edu.au> based | |
d60e92d1 | 4036 | on documentation by Jens Axboe. |
40943b9a TK |
4037 | .br |
4038 | This man page was rewritten by Tomohiro Kusumi <tkusumi@tuxera.com> based | |
4039 | on documentation by Jens Axboe. | |
d60e92d1 | 4040 | .SH "REPORTING BUGS" |
482900c9 | 4041 | Report bugs to the \fBfio\fR mailing list <fio@vger.kernel.org>. |
6468020d | 4042 | .br |
40943b9a TK |
4043 | See \fBREPORTING\-BUGS\fR. |
4044 | .P | |
4045 | \fBREPORTING\-BUGS\fR: \fIhttp://git.kernel.dk/cgit/fio/plain/REPORTING\-BUGS\fR | |
d60e92d1 | 4046 | .SH "SEE ALSO" |
d1429b5c AC |
4047 | For further documentation see \fBHOWTO\fR and \fBREADME\fR. |
4048 | .br | |
40943b9a | 4049 | Sample jobfiles are available in the `examples/' directory. |
9040e236 | 4050 | .br |
40943b9a TK |
4051 | These are typically located under `/usr/share/doc/fio'. |
4052 | .P | |
4053 | \fBHOWTO\fR: \fIhttp://git.kernel.dk/cgit/fio/plain/HOWTO\fR | |
9040e236 | 4054 | .br |
40943b9a | 4055 | \fBREADME\fR: \fIhttp://git.kernel.dk/cgit/fio/plain/README\fR |