Commit | Line | Data |
---|---|---|
f80dba8d MT |
1 | How fio works |
2 | ------------- | |
3 | ||
4 | The first step in getting fio to simulate a desired I/O workload, is writing a | |
5 | job file describing that specific setup. A job file may contain any number of | |
6 | threads and/or files -- the typical contents of the job file is a *global* | |
7 | section defining shared parameters, and one or more job sections describing the | |
8 | jobs involved. When run, fio parses this file and sets everything up as | |
9 | described. If we break down a job from top to bottom, it contains the following | |
10 | basic parameters: | |
11 | ||
12 | `I/O type`_ | |
13 | ||
14 | Defines the I/O pattern issued to the file(s). We may only be reading | |
15 | sequentially from this file(s), or we may be writing randomly. Or even | |
16 | mixing reads and writes, sequentially or randomly. | |
17 | Should we be doing buffered I/O, or direct/raw I/O? | |
18 | ||
19 | `Block size`_ | |
20 | ||
21 | In how large chunks are we issuing I/O? This may be a single value, | |
22 | or it may describe a range of block sizes. | |
23 | ||
24 | `I/O size`_ | |
25 | ||
26 | How much data are we going to be reading/writing. | |
27 | ||
28 | `I/O engine`_ | |
29 | ||
30 | How do we issue I/O? We could be memory mapping the file, we could be | |
31 | using regular read/write, we could be using splice, async I/O, or even | |
32 | SG (SCSI generic sg). | |
33 | ||
34 | `I/O depth`_ | |
35 | ||
36 | If the I/O engine is async, how large a queuing depth do we want to | |
37 | maintain? | |
38 | ||
39 | ||
40 | `Target file/device`_ | |
41 | ||
42 | How many files are we spreading the workload over. | |
43 | ||
44 | `Threads, processes and job synchronization`_ | |
45 | ||
46 | How many threads or processes should we spread this workload over. | |
47 | ||
48 | The above are the basic parameters defined for a workload, in addition there's a | |
49 | multitude of parameters that modify other aspects of how this job behaves. | |
50 | ||
51 | ||
52 | Command line options | |
53 | -------------------- | |
54 | ||
55 | .. option:: --debug=type | |
56 | ||
f50fbdda | 57 | Enable verbose tracing `type` of various fio actions. May be ``all`` for all types |
b034c0dd TK |
58 | or individual types separated by a comma (e.g. ``--debug=file,mem`` will |
59 | enable file and memory debugging). Currently, additional logging is | |
60 | available for: | |
f80dba8d | 61 | |
b034c0dd | 62 | *process* |
f80dba8d | 63 | Dump info related to processes. |
b034c0dd | 64 | *file* |
f80dba8d | 65 | Dump info related to file actions. |
b034c0dd | 66 | *io* |
f80dba8d | 67 | Dump info related to I/O queuing. |
b034c0dd | 68 | *mem* |
f80dba8d | 69 | Dump info related to memory allocations. |
b034c0dd | 70 | *blktrace* |
f80dba8d | 71 | Dump info related to blktrace setup. |
b034c0dd | 72 | *verify* |
f80dba8d | 73 | Dump info related to I/O verification. |
b034c0dd | 74 | *all* |
f80dba8d | 75 | Enable all debug options. |
b034c0dd | 76 | *random* |
f80dba8d | 77 | Dump info related to random offset generation. |
b034c0dd | 78 | *parse* |
f80dba8d | 79 | Dump info related to option matching and parsing. |
b034c0dd | 80 | *diskutil* |
f80dba8d | 81 | Dump info related to disk utilization updates. |
b034c0dd | 82 | *job:x* |
f80dba8d | 83 | Dump info only related to job number x. |
b034c0dd | 84 | *mutex* |
f80dba8d | 85 | Dump info only related to mutex up/down ops. |
b034c0dd | 86 | *profile* |
f80dba8d | 87 | Dump info related to profile extensions. |
b034c0dd | 88 | *time* |
f80dba8d | 89 | Dump info related to internal time keeping. |
b034c0dd | 90 | *net* |
f80dba8d | 91 | Dump info related to networking connections. |
b034c0dd | 92 | *rate* |
f80dba8d | 93 | Dump info related to I/O rate switching. |
b034c0dd | 94 | *compress* |
f80dba8d | 95 | Dump info related to log compress/decompress. |
a02ec45a VF |
96 | *steadystate* |
97 | Dump info related to steadystate detection. | |
98 | *helperthread* | |
99 | Dump info related to the helper thread. | |
100 | *zbd* | |
101 | Dump info related to support for zoned block devices. | |
b034c0dd | 102 | *?* or *help* |
f80dba8d MT |
103 | Show available debug options. |
104 | ||
105 | .. option:: --parse-only | |
106 | ||
25cd4b95 | 107 | Parse options only, don't start any I/O. |
f80dba8d | 108 | |
b9921d1a DZ |
109 | .. option:: --merge-blktrace-only |
110 | ||
111 | Merge blktraces only, don't start any I/O. | |
112 | ||
f80dba8d MT |
113 | .. option:: --output=filename |
114 | ||
115 | Write output to file `filename`. | |
116 | ||
f50fbdda | 117 | .. option:: --output-format=format |
b8f7e412 | 118 | |
f50fbdda | 119 | Set the reporting `format` to `normal`, `terse`, `json`, or `json+`. Multiple |
b8f7e412 TK |
120 | formats can be selected, separated by a comma. `terse` is a CSV based |
121 | format. `json+` is like `json`, except it adds a full dump of the latency | |
122 | buckets. | |
123 | ||
f80dba8d MT |
124 | .. option:: --bandwidth-log |
125 | ||
126 | Generate aggregate bandwidth logs. | |
127 | ||
128 | .. option:: --minimal | |
129 | ||
130 | Print statistics in a terse, semicolon-delimited format. | |
131 | ||
132 | .. option:: --append-terse | |
133 | ||
b034c0dd TK |
134 | Print statistics in selected mode AND terse, semicolon-delimited format. |
135 | **Deprecated**, use :option:`--output-format` instead to select multiple | |
136 | formats. | |
f80dba8d | 137 | |
f50fbdda | 138 | .. option:: --terse-version=version |
f80dba8d | 139 | |
f50fbdda | 140 | Set terse `version` output format (default 3, or 2 or 4 or 5). |
f80dba8d MT |
141 | |
142 | .. option:: --version | |
143 | ||
b8f7e412 | 144 | Print version information and exit. |
f80dba8d MT |
145 | |
146 | .. option:: --help | |
147 | ||
113f0e7c | 148 | Print a summary of the command line options and exit. |
f80dba8d MT |
149 | |
150 | .. option:: --cpuclock-test | |
151 | ||
152 | Perform test and validation of internal CPU clock. | |
153 | ||
113f0e7c | 154 | .. option:: --crctest=[test] |
f80dba8d | 155 | |
b034c0dd TK |
156 | Test the speed of the built-in checksumming functions. If no argument is |
157 | given, all of them are tested. Alternatively, a comma separated list can | |
158 | be passed, in which case the given ones are tested. | |
f80dba8d MT |
159 | |
160 | .. option:: --cmdhelp=command | |
161 | ||
162 | Print help information for `command`. May be ``all`` for all commands. | |
163 | ||
164 | .. option:: --enghelp=[ioengine[,command]] | |
165 | ||
f50fbdda TK |
166 | List all commands defined by `ioengine`, or print help for `command` |
167 | defined by `ioengine`. If no `ioengine` is given, list all | |
b034c0dd | 168 | available ioengines. |
f80dba8d | 169 | |
57fd9225 | 170 | .. option:: --showcmd |
f80dba8d | 171 | |
57fd9225 | 172 | Convert given job files to a set of command-line options. |
f80dba8d MT |
173 | |
174 | .. option:: --readonly | |
175 | ||
4027b2a1 VF |
176 | Turn on safety read-only checks, preventing writes and trims. The |
177 | ``--readonly`` option is an extra safety guard to prevent users from | |
178 | accidentally starting a write or trim workload when that is not desired. | |
179 | Fio will only modify the device under test if | |
180 | `rw=write/randwrite/rw/randrw/trim/randtrim/trimwrite` is given. This | |
181 | safety net can be used as an extra precaution. | |
f80dba8d MT |
182 | |
183 | .. option:: --eta=when | |
184 | ||
b8f7e412 | 185 | Specifies when real-time ETA estimate should be printed. `when` may be |
db37d890 JA |
186 | `always`, `never` or `auto`. `auto` is the default, it prints ETA |
187 | when requested if the output is a TTY. `always` disregards the output | |
188 | type, and prints ETA when requested. `never` never prints ETA. | |
189 | ||
190 | .. option:: --eta-interval=time | |
191 | ||
192 | By default, fio requests client ETA status roughly every second. With | |
193 | this option, the interval is configurable. Fio imposes a minimum | |
194 | allowed time to avoid flooding the console, less than 250 msec is | |
195 | not supported. | |
f80dba8d MT |
196 | |
197 | .. option:: --eta-newline=time | |
198 | ||
947e0fe0 SW |
199 | Force a new line for every `time` period passed. When the unit is omitted, |
200 | the value is interpreted in seconds. | |
f80dba8d MT |
201 | |
202 | .. option:: --status-interval=time | |
203 | ||
aa6cb459 VF |
204 | Force a full status dump of cumulative (from job start) values at `time` |
205 | intervals. This option does *not* provide per-period measurements. So | |
206 | values such as bandwidth are running averages. When the time unit is omitted, | |
c1f4de8a JA |
207 | `time` is interpreted in seconds. Note that using this option with |
208 | ``--output-format=json`` will yield output that technically isn't valid | |
209 | json, since the output will be collated sets of valid json. It will need | |
210 | to be split into valid sets of json after the run. | |
f80dba8d MT |
211 | |
212 | .. option:: --section=name | |
213 | ||
b034c0dd TK |
214 | Only run specified section `name` in job file. Multiple sections can be specified. |
215 | The ``--section`` option allows one to combine related jobs into one file. | |
216 | E.g. one job file could define light, moderate, and heavy sections. Tell | |
217 | fio to run only the "heavy" section by giving ``--section=heavy`` | |
218 | command line option. One can also specify the "write" operations in one | |
219 | section and "verify" operation in another section. The ``--section`` option | |
220 | only applies to job sections. The reserved *global* section is always | |
221 | parsed and used. | |
f80dba8d MT |
222 | |
223 | .. option:: --alloc-size=kb | |
224 | ||
4a419903 VF |
225 | Allocate additional internal smalloc pools of size `kb` in KiB. The |
226 | ``--alloc-size`` option increases shared memory set aside for use by fio. | |
b034c0dd TK |
227 | If running large jobs with randommap enabled, fio can run out of memory. |
228 | Smalloc is an internal allocator for shared structures from a fixed size | |
229 | memory pool and can grow to 16 pools. The pool size defaults to 16MiB. | |
f80dba8d | 230 | |
b034c0dd TK |
231 | NOTE: While running :file:`.fio_smalloc.*` backing store files are visible |
232 | in :file:`/tmp`. | |
f80dba8d MT |
233 | |
234 | .. option:: --warnings-fatal | |
235 | ||
b034c0dd TK |
236 | All fio parser warnings are fatal, causing fio to exit with an |
237 | error. | |
f80dba8d MT |
238 | |
239 | .. option:: --max-jobs=nr | |
240 | ||
f50fbdda | 241 | Set the maximum number of threads/processes to support to `nr`. |
818322cc | 242 | NOTE: On Linux, it may be necessary to increase the shared-memory |
71aa48eb | 243 | limit (:file:`/proc/sys/kernel/shmmax`) if fio runs into errors while |
818322cc | 244 | creating jobs. |
f80dba8d MT |
245 | |
246 | .. option:: --server=args | |
247 | ||
b034c0dd TK |
248 | Start a backend server, with `args` specifying what to listen to. |
249 | See `Client/Server`_ section. | |
f80dba8d MT |
250 | |
251 | .. option:: --daemonize=pidfile | |
252 | ||
b034c0dd | 253 | Background a fio server, writing the pid to the given `pidfile` file. |
f80dba8d MT |
254 | |
255 | .. option:: --client=hostname | |
256 | ||
f50fbdda | 257 | Instead of running the jobs locally, send and run them on the given `hostname` |
71aa48eb | 258 | or set of `hostname`\s. See `Client/Server`_ section. |
f80dba8d MT |
259 | |
260 | .. option:: --remote-config=file | |
261 | ||
f50fbdda | 262 | Tell fio server to load this local `file`. |
f80dba8d MT |
263 | |
264 | .. option:: --idle-prof=option | |
265 | ||
b8f7e412 | 266 | Report CPU idleness. `option` is one of the following: |
113f0e7c SW |
267 | |
268 | **calibrate** | |
269 | Run unit work calibration only and exit. | |
270 | ||
271 | **system** | |
272 | Show aggregate system idleness and unit work. | |
273 | ||
274 | **percpu** | |
275 | As **system** but also show per CPU idleness. | |
f80dba8d MT |
276 | |
277 | .. option:: --inflate-log=log | |
278 | ||
f50fbdda | 279 | Inflate and output compressed `log`. |
f80dba8d MT |
280 | |
281 | .. option:: --trigger-file=file | |
282 | ||
f50fbdda | 283 | Execute trigger command when `file` exists. |
f80dba8d | 284 | |
f50fbdda | 285 | .. option:: --trigger-timeout=time |
f80dba8d | 286 | |
f50fbdda | 287 | Execute trigger at this `time`. |
f80dba8d | 288 | |
f50fbdda | 289 | .. option:: --trigger=command |
f80dba8d | 290 | |
f50fbdda | 291 | Set this `command` as local trigger. |
f80dba8d | 292 | |
f50fbdda | 293 | .. option:: --trigger-remote=command |
f80dba8d | 294 | |
f50fbdda | 295 | Set this `command` as remote trigger. |
f80dba8d MT |
296 | |
297 | .. option:: --aux-path=path | |
298 | ||
f4401bf8 SW |
299 | Use the directory specified by `path` for generated state files instead |
300 | of the current working directory. | |
f80dba8d MT |
301 | |
302 | Any parameters following the options will be assumed to be job files, unless | |
303 | they match a job file parameter. Multiple job files can be listed and each job | |
304 | file will be regarded as a separate group. Fio will :option:`stonewall` | |
305 | execution between each group. | |
306 | ||
307 | ||
308 | Job file format | |
309 | --------------- | |
310 | ||
311 | As previously described, fio accepts one or more job files describing what it is | |
312 | supposed to do. The job file format is the classic ini file, where the names | |
c60ebc45 | 313 | enclosed in [] brackets define the job name. You are free to use any ASCII name |
f80dba8d MT |
314 | you want, except *global* which has special meaning. Following the job name is |
315 | a sequence of zero or more parameters, one per line, that define the behavior of | |
316 | the job. If the first character in a line is a ';' or a '#', the entire line is | |
317 | discarded as a comment. | |
318 | ||
319 | A *global* section sets defaults for the jobs described in that file. A job may | |
320 | override a *global* section parameter, and a job file may even have several | |
321 | *global* sections if so desired. A job is only affected by a *global* section | |
322 | residing above it. | |
323 | ||
f50fbdda TK |
324 | The :option:`--cmdhelp` option also lists all options. If used with a `command` |
325 | argument, :option:`--cmdhelp` will detail the given `command`. | |
f80dba8d MT |
326 | |
327 | See the `examples/` directory for inspiration on how to write job files. Note | |
328 | the copyright and license requirements currently apply to `examples/` files. | |
329 | ||
330 | So let's look at a really simple job file that defines two processes, each | |
331 | randomly reading from a 128MiB file: | |
332 | ||
333 | .. code-block:: ini | |
334 | ||
335 | ; -- start job file -- | |
336 | [global] | |
337 | rw=randread | |
338 | size=128m | |
339 | ||
340 | [job1] | |
341 | ||
342 | [job2] | |
343 | ||
344 | ; -- end job file -- | |
345 | ||
346 | As you can see, the job file sections themselves are empty as all the described | |
347 | parameters are shared. As no :option:`filename` option is given, fio makes up a | |
348 | `filename` for each of the jobs as it sees fit. On the command line, this job | |
349 | would look as follows:: | |
350 | ||
351 | $ fio --name=global --rw=randread --size=128m --name=job1 --name=job2 | |
352 | ||
353 | ||
354 | Let's look at an example that has a number of processes writing randomly to | |
355 | files: | |
356 | ||
357 | .. code-block:: ini | |
358 | ||
359 | ; -- start job file -- | |
360 | [random-writers] | |
361 | ioengine=libaio | |
362 | iodepth=4 | |
363 | rw=randwrite | |
364 | bs=32k | |
365 | direct=0 | |
366 | size=64m | |
367 | numjobs=4 | |
368 | ; -- end job file -- | |
369 | ||
370 | Here we have no *global* section, as we only have one job defined anyway. We | |
371 | want to use async I/O here, with a depth of 4 for each file. We also increased | |
372 | the buffer size used to 32KiB and define numjobs to 4 to fork 4 identical | |
373 | jobs. The result is 4 processes each randomly writing to their own 64MiB | |
374 | file. Instead of using the above job file, you could have given the parameters | |
375 | on the command line. For this case, you would specify:: | |
376 | ||
377 | $ fio --name=random-writers --ioengine=libaio --iodepth=4 --rw=randwrite --bs=32k --direct=0 --size=64m --numjobs=4 | |
378 | ||
379 | When fio is utilized as a basis of any reasonably large test suite, it might be | |
380 | desirable to share a set of standardized settings across multiple job files. | |
381 | Instead of copy/pasting such settings, any section may pull in an external | |
382 | :file:`filename.fio` file with *include filename* directive, as in the following | |
383 | example:: | |
384 | ||
385 | ; -- start job file including.fio -- | |
386 | [global] | |
387 | filename=/tmp/test | |
388 | filesize=1m | |
389 | include glob-include.fio | |
390 | ||
391 | [test] | |
392 | rw=randread | |
393 | bs=4k | |
394 | time_based=1 | |
395 | runtime=10 | |
396 | include test-include.fio | |
397 | ; -- end job file including.fio -- | |
398 | ||
399 | .. code-block:: ini | |
400 | ||
401 | ; -- start job file glob-include.fio -- | |
402 | thread=1 | |
403 | group_reporting=1 | |
404 | ; -- end job file glob-include.fio -- | |
405 | ||
406 | .. code-block:: ini | |
407 | ||
408 | ; -- start job file test-include.fio -- | |
409 | ioengine=libaio | |
410 | iodepth=4 | |
411 | ; -- end job file test-include.fio -- | |
412 | ||
413 | Settings pulled into a section apply to that section only (except *global* | |
414 | section). Include directives may be nested in that any included file may contain | |
415 | further include directive(s). Include files may not contain [] sections. | |
416 | ||
417 | ||
418 | Environment variables | |
419 | ~~~~~~~~~~~~~~~~~~~~~ | |
420 | ||
421 | Fio also supports environment variable expansion in job files. Any sub-string of | |
422 | the form ``${VARNAME}`` as part of an option value (in other words, on the right | |
423 | of the '='), will be expanded to the value of the environment variable called | |
424 | `VARNAME`. If no such environment variable is defined, or `VARNAME` is the | |
425 | empty string, the empty string will be substituted. | |
426 | ||
427 | As an example, let's look at a sample fio invocation and job file:: | |
428 | ||
429 | $ SIZE=64m NUMJOBS=4 fio jobfile.fio | |
430 | ||
431 | .. code-block:: ini | |
432 | ||
433 | ; -- start job file -- | |
434 | [random-writers] | |
435 | rw=randwrite | |
436 | size=${SIZE} | |
437 | numjobs=${NUMJOBS} | |
438 | ; -- end job file -- | |
439 | ||
440 | This will expand to the following equivalent job file at runtime: | |
441 | ||
442 | .. code-block:: ini | |
443 | ||
444 | ; -- start job file -- | |
445 | [random-writers] | |
446 | rw=randwrite | |
447 | size=64m | |
448 | numjobs=4 | |
449 | ; -- end job file -- | |
450 | ||
451 | Fio ships with a few example job files, you can also look there for inspiration. | |
452 | ||
453 | Reserved keywords | |
454 | ~~~~~~~~~~~~~~~~~ | |
455 | ||
456 | Additionally, fio has a set of reserved keywords that will be replaced | |
457 | internally with the appropriate value. Those keywords are: | |
458 | ||
459 | **$pagesize** | |
460 | ||
461 | The architecture page size of the running system. | |
462 | ||
463 | **$mb_memory** | |
464 | ||
465 | Megabytes of total memory in the system. | |
466 | ||
467 | **$ncpus** | |
468 | ||
469 | Number of online available CPUs. | |
470 | ||
471 | These can be used on the command line or in the job file, and will be | |
472 | automatically substituted with the current system values when the job is | |
473 | run. Simple math is also supported on these keywords, so you can perform actions | |
474 | like:: | |
475 | ||
b034c0dd | 476 | size=8*$mb_memory |
f80dba8d MT |
477 | |
478 | and get that properly expanded to 8 times the size of memory in the machine. | |
479 | ||
480 | ||
481 | Job file parameters | |
482 | ------------------- | |
483 | ||
484 | This section describes in details each parameter associated with a job. Some | |
485 | parameters take an option of a given type, such as an integer or a | |
486 | string. Anywhere a numeric value is required, an arithmetic expression may be | |
487 | used, provided it is surrounded by parentheses. Supported operators are: | |
488 | ||
489 | - addition (+) | |
490 | - subtraction (-) | |
491 | - multiplication (*) | |
492 | - division (/) | |
493 | - modulus (%) | |
494 | - exponentiation (^) | |
495 | ||
496 | For time values in expressions, units are microseconds by default. This is | |
497 | different than for time values not in expressions (not enclosed in | |
498 | parentheses). The following types are used: | |
499 | ||
500 | ||
501 | Parameter types | |
502 | ~~~~~~~~~~~~~~~ | |
503 | ||
504 | **str** | |
b034c0dd | 505 | String: A sequence of alphanumeric characters. |
f80dba8d MT |
506 | |
507 | **time** | |
008d0feb SW |
508 | Integer with possible time suffix. Without a unit value is interpreted as |
509 | seconds unless otherwise specified. Accepts a suffix of 'd' for days, 'h' for | |
510 | hours, 'm' for minutes, 's' for seconds, 'ms' (or 'msec') for milliseconds and | |
511 | 'us' (or 'usec') for microseconds. For example, use 10m for 10 minutes. | |
f80dba8d MT |
512 | |
513 | .. _int: | |
514 | ||
515 | **int** | |
516 | Integer. A whole number value, which may contain an integer prefix | |
517 | and an integer suffix: | |
518 | ||
b034c0dd | 519 | [*integer prefix*] **number** [*integer suffix*] |
f80dba8d MT |
520 | |
521 | The optional *integer prefix* specifies the number's base. The default | |
522 | is decimal. *0x* specifies hexadecimal. | |
523 | ||
524 | The optional *integer suffix* specifies the number's units, and includes an | |
525 | optional unit prefix and an optional unit. For quantities of data, the | |
947e0fe0 SW |
526 | default unit is bytes. For quantities of time, the default unit is seconds |
527 | unless otherwise specified. | |
f80dba8d | 528 | |
9207a0cb | 529 | With :option:`kb_base`\=1000, fio follows international standards for unit |
f80dba8d MT |
530 | prefixes. To specify power-of-10 decimal values defined in the |
531 | International System of Units (SI): | |
532 | ||
eccce61a TK |
533 | * *K* -- means kilo (K) or 1000 |
534 | * *M* -- means mega (M) or 1000**2 | |
535 | * *G* -- means giga (G) or 1000**3 | |
536 | * *T* -- means tera (T) or 1000**4 | |
537 | * *P* -- means peta (P) or 1000**5 | |
f80dba8d MT |
538 | |
539 | To specify power-of-2 binary values defined in IEC 80000-13: | |
540 | ||
eccce61a TK |
541 | * *Ki* -- means kibi (Ki) or 1024 |
542 | * *Mi* -- means mebi (Mi) or 1024**2 | |
543 | * *Gi* -- means gibi (Gi) or 1024**3 | |
544 | * *Ti* -- means tebi (Ti) or 1024**4 | |
545 | * *Pi* -- means pebi (Pi) or 1024**5 | |
f80dba8d | 546 | |
193aaf6a G |
547 | For Zone Block Device Mode: |
548 | * *z* -- means Zone | |
549 | ||
9207a0cb | 550 | With :option:`kb_base`\=1024 (the default), the unit prefixes are opposite |
f80dba8d MT |
551 | from those specified in the SI and IEC 80000-13 standards to provide |
552 | compatibility with old scripts. For example, 4k means 4096. | |
553 | ||
554 | For quantities of data, an optional unit of 'B' may be included | |
b8f7e412 | 555 | (e.g., 'kB' is the same as 'k'). |
f80dba8d MT |
556 | |
557 | The *integer suffix* is not case sensitive (e.g., m/mi mean mebi/mega, | |
558 | not milli). 'b' and 'B' both mean byte, not bit. | |
559 | ||
9207a0cb | 560 | Examples with :option:`kb_base`\=1000: |
f80dba8d MT |
561 | |
562 | * *4 KiB*: 4096, 4096b, 4096B, 4ki, 4kib, 4kiB, 4Ki, 4KiB | |
563 | * *1 MiB*: 1048576, 1mi, 1024ki | |
564 | * *1 MB*: 1000000, 1m, 1000k | |
565 | * *1 TiB*: 1099511627776, 1ti, 1024gi, 1048576mi | |
566 | * *1 TB*: 1000000000, 1t, 1000m, 1000000k | |
567 | ||
9207a0cb | 568 | Examples with :option:`kb_base`\=1024 (default): |
f80dba8d MT |
569 | |
570 | * *4 KiB*: 4096, 4096b, 4096B, 4k, 4kb, 4kB, 4K, 4KB | |
571 | * *1 MiB*: 1048576, 1m, 1024k | |
572 | * *1 MB*: 1000000, 1mi, 1000ki | |
573 | * *1 TiB*: 1099511627776, 1t, 1024g, 1048576m | |
574 | * *1 TB*: 1000000000, 1ti, 1000mi, 1000000ki | |
575 | ||
576 | To specify times (units are not case sensitive): | |
577 | ||
578 | * *D* -- means days | |
579 | * *H* -- means hours | |
4502cb42 | 580 | * *M* -- means minutes |
f80dba8d MT |
581 | * *s* -- or sec means seconds (default) |
582 | * *ms* -- or *msec* means milliseconds | |
583 | * *us* -- or *usec* means microseconds | |
584 | ||
585 | If the option accepts an upper and lower range, use a colon ':' or | |
586 | minus '-' to separate such values. See :ref:`irange <irange>`. | |
4502cb42 SW |
587 | If the lower value specified happens to be larger than the upper value |
588 | the two values are swapped. | |
f80dba8d MT |
589 | |
590 | .. _bool: | |
591 | ||
592 | **bool** | |
593 | Boolean. Usually parsed as an integer, however only defined for | |
594 | true and false (1 and 0). | |
595 | ||
596 | .. _irange: | |
597 | ||
598 | **irange** | |
599 | Integer range with suffix. Allows value range to be given, such as | |
c60ebc45 | 600 | 1024-4096. A colon may also be used as the separator, e.g. 1k:4k. If the |
f80dba8d MT |
601 | option allows two sets of ranges, they can be specified with a ',' or '/' |
602 | delimiter: 1k-4k/8k-32k. Also see :ref:`int <int>`. | |
603 | ||
604 | **float_list** | |
605 | A list of floating point numbers, separated by a ':' character. | |
606 | ||
f5c3bcf2 TK |
607 | With the above in mind, here follows the complete list of fio job parameters. |
608 | ||
f80dba8d MT |
609 | |
610 | Units | |
611 | ~~~~~ | |
612 | ||
613 | .. option:: kb_base=int | |
614 | ||
615 | Select the interpretation of unit prefixes in input parameters. | |
616 | ||
617 | **1000** | |
618 | Inputs comply with IEC 80000-13 and the International | |
619 | System of Units (SI). Use: | |
620 | ||
621 | - power-of-2 values with IEC prefixes (e.g., KiB) | |
622 | - power-of-10 values with SI prefixes (e.g., kB) | |
623 | ||
624 | **1024** | |
625 | Compatibility mode (default). To avoid breaking old scripts: | |
626 | ||
627 | - power-of-2 values with SI prefixes | |
628 | - power-of-10 values with IEC prefixes | |
629 | ||
630 | See :option:`bs` for more details on input parameters. | |
631 | ||
632 | Outputs always use correct prefixes. Most outputs include both | |
633 | side-by-side, like:: | |
634 | ||
635 | bw=2383.3kB/s (2327.4KiB/s) | |
636 | ||
637 | If only one value is reported, then kb_base selects the one to use: | |
638 | ||
639 | **1000** -- SI prefixes | |
640 | ||
641 | **1024** -- IEC prefixes | |
642 | ||
643 | .. option:: unit_base=int | |
644 | ||
645 | Base unit for reporting. Allowed values are: | |
646 | ||
647 | **0** | |
648 | Use auto-detection (default). | |
649 | **8** | |
650 | Byte based. | |
651 | **1** | |
652 | Bit based. | |
653 | ||
654 | ||
f80dba8d MT |
655 | Job description |
656 | ~~~~~~~~~~~~~~~ | |
657 | ||
658 | .. option:: name=str | |
659 | ||
660 | ASCII name of the job. This may be used to override the name printed by fio | |
661 | for this job. Otherwise the job name is used. On the command line this | |
662 | parameter has the special purpose of also signaling the start of a new job. | |
663 | ||
664 | .. option:: description=str | |
665 | ||
666 | Text description of the job. Doesn't do anything except dump this text | |
667 | description when this job is run. It's not parsed. | |
668 | ||
669 | .. option:: loops=int | |
670 | ||
671 | Run the specified number of iterations of this job. Used to repeat the same | |
672 | workload a given number of times. Defaults to 1. | |
673 | ||
674 | .. option:: numjobs=int | |
675 | ||
79591fa9 TK |
676 | Create the specified number of clones of this job. Each clone of job |
677 | is spawned as an independent thread or process. May be used to setup a | |
f80dba8d MT |
678 | larger number of threads/processes doing the same thing. Each thread is |
679 | reported separately; to see statistics for all clones as a whole, use | |
680 | :option:`group_reporting` in conjunction with :option:`new_group`. | |
a47b697c | 681 | See :option:`--max-jobs`. Default: 1. |
f80dba8d MT |
682 | |
683 | ||
684 | Time related parameters | |
685 | ~~~~~~~~~~~~~~~~~~~~~~~ | |
686 | ||
687 | .. option:: runtime=time | |
688 | ||
c7927863 H |
689 | Limit runtime. The test will run until it completes the configured I/O |
690 | workload or until it has run for this specified amount of time, whichever | |
691 | occurs first. It can be quite hard to determine for how long a specified | |
692 | job will run, so this parameter is handy to cap the total runtime to a | |
693 | given time. When the unit is omitted, the value is interpreted in | |
694 | seconds. | |
f80dba8d MT |
695 | |
696 | .. option:: time_based | |
697 | ||
698 | If set, fio will run for the duration of the :option:`runtime` specified | |
699 | even if the file(s) are completely read or written. It will simply loop over | |
700 | the same workload as many times as the :option:`runtime` allows. | |
701 | ||
a881438b | 702 | .. option:: startdelay=irange(time) |
f80dba8d | 703 | |
947e0fe0 SW |
704 | Delay the start of job for the specified amount of time. Can be a single |
705 | value or a range. When given as a range, each thread will choose a value | |
706 | randomly from within the range. Value is in seconds if a unit is omitted. | |
f80dba8d MT |
707 | |
708 | .. option:: ramp_time=time | |
709 | ||
710 | If set, fio will run the specified workload for this amount of time before | |
711 | logging any performance numbers. Useful for letting performance settle | |
712 | before logging results, thus minimizing the runtime required for stable | |
713 | results. Note that the ``ramp_time`` is considered lead in time for a job, | |
714 | thus it will increase the total runtime if a special timeout or | |
f75ede1d SW |
715 | :option:`runtime` is specified. When the unit is omitted, the value is |
716 | given in seconds. | |
f80dba8d MT |
717 | |
718 | .. option:: clocksource=str | |
719 | ||
720 | Use the given clocksource as the base of timing. The supported options are: | |
721 | ||
722 | **gettimeofday** | |
723 | :manpage:`gettimeofday(2)` | |
724 | ||
725 | **clock_gettime** | |
726 | :manpage:`clock_gettime(2)` | |
727 | ||
728 | **cpu** | |
729 | Internal CPU clock source | |
730 | ||
731 | cpu is the preferred clocksource if it is reliable, as it is very fast (and | |
732 | fio is heavy on time calls). Fio will automatically use this clocksource if | |
733 | it's supported and considered reliable on the system it is running on, | |
734 | unless another clocksource is specifically set. For x86/x86-64 CPUs, this | |
735 | means supporting TSC Invariant. | |
736 | ||
737 | .. option:: gtod_reduce=bool | |
738 | ||
739 | Enable all of the :manpage:`gettimeofday(2)` reducing options | |
f75ede1d | 740 | (:option:`disable_clat`, :option:`disable_slat`, :option:`disable_bw_measurement`) plus |
f80dba8d MT |
741 | reduce precision of the timeout somewhat to really shrink the |
742 | :manpage:`gettimeofday(2)` call count. With this option enabled, we only do | |
743 | about 0.4% of the :manpage:`gettimeofday(2)` calls we would have done if all | |
744 | time keeping was enabled. | |
745 | ||
746 | .. option:: gtod_cpu=int | |
747 | ||
748 | Sometimes it's cheaper to dedicate a single thread of execution to just | |
749 | getting the current time. Fio (and databases, for instance) are very | |
750 | intensive on :manpage:`gettimeofday(2)` calls. With this option, you can set | |
751 | one CPU aside for doing nothing but logging current time to a shared memory | |
752 | location. Then the other threads/processes that run I/O workloads need only | |
753 | copy that segment, instead of entering the kernel with a | |
754 | :manpage:`gettimeofday(2)` call. The CPU set aside for doing these time | |
755 | calls will be excluded from other uses. Fio will manually clear it from the | |
756 | CPU mask of other jobs. | |
757 | ||
758 | ||
759 | Target file/device | |
760 | ~~~~~~~~~~~~~~~~~~ | |
761 | ||
762 | .. option:: directory=str | |
763 | ||
764 | Prefix filenames with this directory. Used to place files in a different | |
765 | location than :file:`./`. You can specify a number of directories by | |
766 | separating the names with a ':' character. These directories will be | |
02dd2689 | 767 | assigned equally distributed to job clones created by :option:`numjobs` as |
f80dba8d MT |
768 | long as they are using generated filenames. If specific `filename(s)` are |
769 | set fio will use the first listed directory, and thereby matching the | |
f4401bf8 SW |
770 | `filename` semantic (which generates a file for each clone if not |
771 | specified, but lets all clones use the same file if set). | |
f80dba8d | 772 | |
3b803fe1 SW |
773 | See the :option:`filename` option for information on how to escape "``:``" |
774 | characters within the directory path itself. | |
f80dba8d | 775 | |
f4401bf8 SW |
776 | Note: To control the directory fio will use for internal state files |
777 | use :option:`--aux-path`. | |
778 | ||
f80dba8d MT |
779 | .. option:: filename=str |
780 | ||
781 | Fio normally makes up a `filename` based on the job name, thread number, and | |
02dd2689 SW |
782 | file number (see :option:`filename_format`). If you want to share files |
783 | between threads in a job or several | |
79591fa9 TK |
784 | jobs with fixed file paths, specify a `filename` for each of them to override |
785 | the default. If the ioengine is file based, you can specify a number of files | |
786 | by separating the names with a ':' colon. So if you wanted a job to open | |
787 | :file:`/dev/sda` and :file:`/dev/sdb` as the two working files, you would use | |
788 | ``filename=/dev/sda:/dev/sdb``. This also means that whenever this option is | |
789 | specified, :option:`nrfiles` is ignored. The size of regular files specified | |
02dd2689 | 790 | by this option will be :option:`size` divided by number of files unless an |
79591fa9 TK |
791 | explicit size is specified by :option:`filesize`. |
792 | ||
3b803fe1 | 793 | Each colon in the wanted path must be escaped with a ``\`` |
02dd2689 SW |
794 | character. For instance, if the path is :file:`/dev/dsk/foo@3,0:c` then you |
795 | would use ``filename=/dev/dsk/foo@3,0\:c`` and if the path is | |
3b803fe1 | 796 | :file:`F:\\filename` then you would use ``filename=F\:\filename``. |
02dd2689 | 797 | |
f80dba8d MT |
798 | On Windows, disk devices are accessed as :file:`\\\\.\\PhysicalDrive0` for |
799 | the first device, :file:`\\\\.\\PhysicalDrive1` for the second etc. | |
800 | Note: Windows and FreeBSD prevent write access to areas | |
02dd2689 SW |
801 | of the disk containing in-use data (e.g. filesystems). |
802 | ||
803 | The filename "`-`" is a reserved name, meaning *stdin* or *stdout*. Which | |
804 | of the two depends on the read/write direction set. | |
f80dba8d MT |
805 | |
806 | .. option:: filename_format=str | |
807 | ||
808 | If sharing multiple files between jobs, it is usually necessary to have fio | |
809 | generate the exact names that you want. By default, fio will name a file | |
810 | based on the default file format specification of | |
811 | :file:`jobname.jobnumber.filenumber`. With this option, that can be | |
812 | customized. Fio will recognize and replace the following keywords in this | |
813 | string: | |
814 | ||
815 | **$jobname** | |
816 | The name of the worker thread or process. | |
8d53c5f8 TG |
817 | **$clientuid** |
818 | IP of the fio process when using client/server mode. | |
f80dba8d MT |
819 | **$jobnum** |
820 | The incremental number of the worker thread or process. | |
821 | **$filenum** | |
822 | The incremental number of the file for that worker thread or | |
823 | process. | |
824 | ||
825 | To have dependent jobs share a set of files, this option can be set to have | |
826 | fio generate filenames that are shared between the two. For instance, if | |
827 | :file:`testfiles.$filenum` is specified, file number 4 for any job will be | |
828 | named :file:`testfiles.4`. The default of :file:`$jobname.$jobnum.$filenum` | |
829 | will be used if no other format specifier is given. | |
830 | ||
645943c0 JB |
831 | If you specify a path then the directories will be created up to the |
832 | main directory for the file. So for example if you specify | |
833 | ``filename_format=a/b/c/$jobnum`` then the directories a/b/c will be | |
834 | created before the file setup part of the job. If you specify | |
835 | :option:`directory` then the path will be relative that directory, | |
836 | otherwise it is treated as the absolute path. | |
837 | ||
f80dba8d MT |
838 | .. option:: unique_filename=bool |
839 | ||
840 | To avoid collisions between networked clients, fio defaults to prefixing any | |
841 | generated filenames (with a directory specified) with the source of the | |
842 | client connecting. To disable this behavior, set this option to 0. | |
843 | ||
844 | .. option:: opendir=str | |
845 | ||
846 | Recursively open any files below directory `str`. | |
847 | ||
848 | .. option:: lockfile=str | |
849 | ||
850 | Fio defaults to not locking any files before it does I/O to them. If a file | |
851 | or file descriptor is shared, fio can serialize I/O to that file to make the | |
852 | end result consistent. This is usual for emulating real workloads that share | |
853 | files. The lock modes are: | |
854 | ||
855 | **none** | |
856 | No locking. The default. | |
857 | **exclusive** | |
858 | Only one thread or process may do I/O at a time, excluding all | |
859 | others. | |
860 | **readwrite** | |
861 | Read-write locking on the file. Many readers may | |
862 | access the file at the same time, but writes get exclusive access. | |
863 | ||
864 | .. option:: nrfiles=int | |
865 | ||
79591fa9 TK |
866 | Number of files to use for this job. Defaults to 1. The size of files |
867 | will be :option:`size` divided by this unless explicit size is specified by | |
868 | :option:`filesize`. Files are created for each thread separately, and each | |
869 | file will have a file number within its name by default, as explained in | |
870 | :option:`filename` section. | |
871 | ||
f80dba8d MT |
872 | |
873 | .. option:: openfiles=int | |
874 | ||
875 | Number of files to keep open at the same time. Defaults to the same as | |
876 | :option:`nrfiles`, can be set smaller to limit the number simultaneous | |
877 | opens. | |
878 | ||
879 | .. option:: file_service_type=str | |
880 | ||
881 | Defines how fio decides which file from a job to service next. The following | |
882 | types are defined: | |
883 | ||
884 | **random** | |
885 | Choose a file at random. | |
886 | ||
887 | **roundrobin** | |
888 | Round robin over opened files. This is the default. | |
889 | ||
890 | **sequential** | |
891 | Finish one file before moving on to the next. Multiple files can | |
f50fbdda | 892 | still be open depending on :option:`openfiles`. |
f80dba8d MT |
893 | |
894 | **zipf** | |
c60ebc45 | 895 | Use a *Zipf* distribution to decide what file to access. |
f80dba8d MT |
896 | |
897 | **pareto** | |
c60ebc45 | 898 | Use a *Pareto* distribution to decide what file to access. |
f80dba8d | 899 | |
dd3503d3 | 900 | **normal** |
c60ebc45 | 901 | Use a *Gaussian* (normal) distribution to decide what file to |
f80dba8d MT |
902 | access. |
903 | ||
dd3503d3 SW |
904 | **gauss** |
905 | Alias for normal. | |
906 | ||
f80dba8d MT |
907 | For *random*, *roundrobin*, and *sequential*, a postfix can be appended to |
908 | tell fio how many I/Os to issue before switching to a new file. For example, | |
909 | specifying ``file_service_type=random:8`` would cause fio to issue | |
910 | 8 I/Os before selecting a new file at random. For the non-uniform | |
911 | distributions, a floating point postfix can be given to influence how the | |
912 | distribution is skewed. See :option:`random_distribution` for a description | |
913 | of how that would work. | |
914 | ||
915 | .. option:: ioscheduler=str | |
916 | ||
917 | Attempt to switch the device hosting the file to the specified I/O scheduler | |
918 | before running. | |
919 | ||
920 | .. option:: create_serialize=bool | |
921 | ||
922 | If true, serialize the file creation for the jobs. This may be handy to | |
923 | avoid interleaving of data files, which may greatly depend on the filesystem | |
a47b697c | 924 | used and even the number of processors in the system. Default: true. |
f80dba8d MT |
925 | |
926 | .. option:: create_fsync=bool | |
927 | ||
22413915 | 928 | :manpage:`fsync(2)` the data file after creation. This is the default. |
f80dba8d MT |
929 | |
930 | .. option:: create_on_open=bool | |
931 | ||
730bd7d9 SW |
932 | If true, don't pre-create files but allow the job's open() to create a file |
933 | when it's time to do I/O. Default: false -- pre-create all necessary files | |
934 | when the job starts. | |
f80dba8d MT |
935 | |
936 | .. option:: create_only=bool | |
937 | ||
938 | If true, fio will only run the setup phase of the job. If files need to be | |
4502cb42 | 939 | laid out or updated on disk, only that will be done -- the actual job contents |
a47b697c | 940 | are not executed. Default: false. |
f80dba8d MT |
941 | |
942 | .. option:: allow_file_create=bool | |
943 | ||
730bd7d9 SW |
944 | If true, fio is permitted to create files as part of its workload. If this |
945 | option is false, then fio will error out if | |
f80dba8d MT |
946 | the files it needs to use don't already exist. Default: true. |
947 | ||
948 | .. option:: allow_mounted_write=bool | |
949 | ||
c60ebc45 | 950 | If this isn't set, fio will abort jobs that are destructive (e.g. that write) |
f80dba8d MT |
951 | to what appears to be a mounted device or partition. This should help catch |
952 | creating inadvertently destructive tests, not realizing that the test will | |
b1db0375 TK |
953 | destroy data on the mounted file system. Note that some platforms don't allow |
954 | writing against a mounted device regardless of this option. Default: false. | |
f80dba8d MT |
955 | |
956 | .. option:: pre_read=bool | |
957 | ||
958 | If this is given, files will be pre-read into memory before starting the | |
959 | given I/O operation. This will also clear the :option:`invalidate` flag, | |
960 | since it is pointless to pre-read and then drop the cache. This will only | |
961 | work for I/O engines that are seek-able, since they allow you to read the | |
a47b697c SW |
962 | same data multiple times. Thus it will not work on non-seekable I/O engines |
963 | (e.g. network, splice). Default: false. | |
f80dba8d MT |
964 | |
965 | .. option:: unlink=bool | |
966 | ||
967 | Unlink the job files when done. Not the default, as repeated runs of that | |
a47b697c SW |
968 | job would then waste time recreating the file set again and again. Default: |
969 | false. | |
f80dba8d MT |
970 | |
971 | .. option:: unlink_each_loop=bool | |
972 | ||
a47b697c | 973 | Unlink job files after each iteration or loop. Default: false. |
f80dba8d | 974 | |
7b865a2f BVA |
975 | .. option:: zonemode=str |
976 | ||
977 | Accepted values are: | |
978 | ||
979 | **none** | |
b8dd9750 HH |
980 | The :option:`zonerange`, :option:`zonesize`, |
981 | :option `zonecapacity` and option:`zoneskip` | |
982 | parameters are ignored. | |
7b865a2f BVA |
983 | **strided** |
984 | I/O happens in a single zone until | |
985 | :option:`zonesize` bytes have been transferred. | |
986 | After that number of bytes has been | |
987 | transferred processing of the next zone | |
b8dd9750 | 988 | starts. :option `zonecapacity` is ignored. |
7b865a2f BVA |
989 | **zbd** |
990 | Zoned block device mode. I/O happens | |
991 | sequentially in each zone, even if random I/O | |
992 | has been selected. Random I/O happens across | |
993 | all zones instead of being restricted to a | |
994 | single zone. The :option:`zoneskip` parameter | |
995 | is ignored. :option:`zonerange` and | |
996 | :option:`zonesize` must be identical. | |
2455851d SK |
997 | Trim is handled using a zone reset operation. |
998 | Trim only considers non-empty sequential write | |
999 | required and sequential write preferred zones. | |
7b865a2f | 1000 | |
5faddc64 | 1001 | .. option:: zonerange=int |
f80dba8d | 1002 | |
7b865a2f BVA |
1003 | Size of a single zone. See also :option:`zonesize` and |
1004 | :option:`zoneskip`. | |
f80dba8d | 1005 | |
5faddc64 | 1006 | .. option:: zonesize=int |
f80dba8d | 1007 | |
7b865a2f BVA |
1008 | For :option:`zonemode` =strided, this is the number of bytes to |
1009 | transfer before skipping :option:`zoneskip` bytes. If this parameter | |
1010 | is smaller than :option:`zonerange` then only a fraction of each zone | |
1011 | with :option:`zonerange` bytes will be accessed. If this parameter is | |
1012 | larger than :option:`zonerange` then each zone will be accessed | |
1013 | multiple times before skipping to the next zone. | |
1014 | ||
1015 | For :option:`zonemode` =zbd, this is the size of a single zone. The | |
1016 | :option:`zonerange` parameter is ignored in this mode. | |
f80dba8d | 1017 | |
b8dd9750 HH |
1018 | |
1019 | .. option:: zonecapacity=int | |
1020 | ||
1021 | For :option:`zonemode` =zbd, this defines the capacity of a single zone, | |
1022 | which is the accessible area starting from the zone start address. | |
1023 | This parameter only applies when using :option:`zonemode` =zbd in | |
1024 | combination with regular block devices. If not specified it defaults to | |
1025 | the zone size. If the target device is a zoned block device, the zone | |
1026 | capacity is obtained from the device information and this option is | |
1027 | ignored. | |
1028 | ||
f80dba8d MT |
1029 | .. option:: zoneskip=int |
1030 | ||
7b865a2f BVA |
1031 | For :option:`zonemode` =strided, the number of bytes to skip after |
1032 | :option:`zonesize` bytes of data have been transferred. This parameter | |
1033 | must be zero for :option:`zonemode` =zbd. | |
f80dba8d | 1034 | |
bfbdd35b BVA |
1035 | .. option:: read_beyond_wp=bool |
1036 | ||
1037 | This parameter applies to :option:`zonemode` =zbd only. | |
1038 | ||
1039 | Zoned block devices are block devices that consist of multiple zones. | |
1040 | Each zone has a type, e.g. conventional or sequential. A conventional | |
1041 | zone can be written at any offset that is a multiple of the block | |
1042 | size. Sequential zones must be written sequentially. The position at | |
1043 | which a write must occur is called the write pointer. A zoned block | |
1044 | device can be either drive managed, host managed or host aware. For | |
1045 | host managed devices the host must ensure that writes happen | |
1046 | sequentially. Fio recognizes host managed devices and serializes | |
1047 | writes to sequential zones for these devices. | |
1048 | ||
1049 | If a read occurs in a sequential zone beyond the write pointer then | |
1050 | the zoned block device will complete the read without reading any data | |
1051 | from the storage medium. Since such reads lead to unrealistically high | |
1052 | bandwidth and IOPS numbers fio only reads beyond the write pointer if | |
1053 | explicitly told to do so. Default: false. | |
1054 | ||
59b07544 BVA |
1055 | .. option:: max_open_zones=int |
1056 | ||
1c97d909 SK |
1057 | A zone of a zoned block device is in the open state when it is partially |
1058 | written (i.e. not all sectors of the zone have been written). Zoned | |
1059 | block devices may have a limit on the total number of zones that can | |
1060 | be simultaneously in the open state, that is, the number of zones that | |
1061 | can be written to simultaneously. The :option:`max_open_zones` parameter | |
1062 | limits the number of zones to which write commands are issued by all fio | |
1063 | jobs, that is, limits the number of zones that will be in the open | |
1064 | state. This parameter is relevant only if the :option:`zonemode` =zbd is | |
1065 | used. The default value is always equal to maximum number of open zones | |
1066 | of the target zoned block device and a value higher than this limit | |
1067 | cannot be specified by users unless the option | |
1068 | :option:`ignore_zone_limits` is specified. When | |
1069 | :option:`ignore_zone_limits` is specified or the target device has no | |
1070 | limit on the number of zones that can be in an open state, | |
1071 | :option:`max_open_zones` can specify 0 to disable any limit on the | |
1072 | number of zones that can be simultaneously written to by all jobs. | |
59b07544 | 1073 | |
3b78a972 AK |
1074 | .. option:: job_max_open_zones=int |
1075 | ||
1c97d909 SK |
1076 | In the same manner as :option:`max_open_zones`, limit the number of open |
1077 | zones per fio job, that is, the number of zones that a single job can | |
1078 | simultaneously write to. A value of zero indicates no limit. | |
1079 | Default: zero. | |
3b78a972 | 1080 | |
12324d56 | 1081 | .. option:: ignore_zone_limits=bool |
a3a6f105 | 1082 | |
12324d56 DLM |
1083 | If this option is used, fio will ignore the maximum number of open |
1084 | zones limit of the zoned block device in use, thus allowing the | |
1085 | option :option:`max_open_zones` value to be larger than the device | |
1086 | reported limit. Default: false. | |
1087 | ||
a7c2b6fc BVA |
1088 | .. option:: zone_reset_threshold=float |
1089 | ||
d65625eb SK |
1090 | A number between zero and one that indicates the ratio of written bytes |
1091 | in the zones with write pointers in the IO range to the size of the IO | |
1092 | range. When current ratio is above this ratio, zones are reset | |
b3e9bd03 SK |
1093 | periodically as :option:`zone_reset_frequency` specifies. If there are |
1094 | multiple jobs when using this option, the IO range for all write jobs | |
1095 | has to be the same. | |
a7c2b6fc BVA |
1096 | |
1097 | .. option:: zone_reset_frequency=float | |
1098 | ||
1099 | A number between zero and one that indicates how often a zone reset | |
1100 | should be issued if the zone reset threshold has been exceeded. A zone | |
1101 | reset is submitted after each (1 / zone_reset_frequency) write | |
1102 | requests. This and the previous parameter can be used to simulate | |
1103 | garbage collection activity. | |
1104 | ||
f80dba8d MT |
1105 | |
1106 | I/O type | |
1107 | ~~~~~~~~ | |
1108 | ||
1109 | .. option:: direct=bool | |
1110 | ||
1111 | If value is true, use non-buffered I/O. This is usually O_DIRECT. Note that | |
8e889110 | 1112 | OpenBSD and ZFS on Solaris don't support direct I/O. On Windows the synchronous |
f80dba8d MT |
1113 | ioengines don't support direct I/O. Default: false. |
1114 | ||
f80dba8d MT |
1115 | .. option:: buffered=bool |
1116 | ||
1117 | If value is true, use buffered I/O. This is the opposite of the | |
1118 | :option:`direct` option. Defaults to true. | |
1119 | ||
1120 | .. option:: readwrite=str, rw=str | |
1121 | ||
1122 | Type of I/O pattern. Accepted values are: | |
1123 | ||
1124 | **read** | |
1125 | Sequential reads. | |
1126 | **write** | |
1127 | Sequential writes. | |
1128 | **trim** | |
3740cfc8 VF |
1129 | Sequential trims (Linux block devices and SCSI |
1130 | character devices only). | |
f80dba8d MT |
1131 | **randread** |
1132 | Random reads. | |
2831be97 SW |
1133 | **randwrite** |
1134 | Random writes. | |
f80dba8d | 1135 | **randtrim** |
3740cfc8 VF |
1136 | Random trims (Linux block devices and SCSI |
1137 | character devices only). | |
f80dba8d MT |
1138 | **rw,readwrite** |
1139 | Sequential mixed reads and writes. | |
1140 | **randrw** | |
1141 | Random mixed reads and writes. | |
1142 | **trimwrite** | |
1143 | Sequential trim+write sequences. Blocks will be trimmed first, | |
08996af4 AK |
1144 | then the same blocks will be written to. So if ``io_size=64K`` |
1145 | is specified, Fio will trim a total of 64K bytes and also | |
1146 | write 64K bytes on the same trimmed blocks. This behaviour | |
1147 | will be consistent with ``number_ios`` or other Fio options | |
1148 | limiting the total bytes or number of I/O's. | |
c16dc793 JA |
1149 | **randtrimwrite** |
1150 | Like trimwrite, but uses random offsets rather | |
1151 | than sequential writes. | |
f80dba8d MT |
1152 | |
1153 | Fio defaults to read if the option is not specified. For the mixed I/O | |
1154 | types, the default is to split them 50/50. For certain types of I/O the | |
730bd7d9 SW |
1155 | result may still be skewed a bit, since the speed may be different. |
1156 | ||
1157 | It is possible to specify the number of I/Os to do before getting a new | |
1158 | offset by appending ``:<nr>`` to the end of the string given. For a | |
f80dba8d MT |
1159 | random read, it would look like ``rw=randread:8`` for passing in an offset |
1160 | modifier with a value of 8. If the suffix is used with a sequential I/O | |
730bd7d9 SW |
1161 | pattern, then the *<nr>* value specified will be **added** to the generated |
1162 | offset for each I/O turning sequential I/O into sequential I/O with holes. | |
1163 | For instance, using ``rw=write:4k`` will skip 4k for every write. Also see | |
1164 | the :option:`rw_sequencer` option. | |
f80dba8d MT |
1165 | |
1166 | .. option:: rw_sequencer=str | |
1167 | ||
1168 | If an offset modifier is given by appending a number to the ``rw=<str>`` | |
1169 | line, then this option controls how that number modifies the I/O offset | |
1170 | being generated. Accepted values are: | |
1171 | ||
1172 | **sequential** | |
1173 | Generate sequential offset. | |
1174 | **identical** | |
1175 | Generate the same offset. | |
1176 | ||
1177 | ``sequential`` is only useful for random I/O, where fio would normally | |
537e0d23 AK |
1178 | generate a new random offset for every I/O. If you append e.g. 8 to |
1179 | randread, i.e. ``rw=randread:8`` you would get a new random offset for | |
1180 | every 8 I/Os. The result would be a sequence of 8 sequential offsets | |
1181 | with a random starting point. However this behavior may change if a | |
1182 | sequential I/O reaches end of the file. As sequential I/O is already | |
1183 | sequential, setting ``sequential`` for that would not result in any | |
1184 | difference. ``identical`` behaves in a similar fashion, except it sends | |
1185 | the same offset 8 number of times before generating a new offset. | |
1186 | ||
1187 | Example #1:: | |
1188 | ||
1189 | rw=randread:8 | |
1190 | rw_sequencer=sequential | |
1191 | bs=4k | |
1192 | ||
1193 | The generated sequence of offsets will look like this: | |
1194 | 4k, 8k, 12k, 16k, 20k, 24k, 28k, 32k, 92k, 96k, 100k, 104k, 108k, | |
1195 | 112k, 116k, 120k, 48k, 52k ... | |
1196 | ||
1197 | Example #2:: | |
1198 | ||
1199 | rw=randread:8 | |
1200 | rw_sequencer=identical | |
1201 | bs=4k | |
1202 | ||
1203 | The generated sequence of offsets will look like this: | |
1204 | 4k, 4k, 4k, 4k, 4k, 4k, 4k, 4k, 92k, 92k, 92k, 92k, 92k, 92k, 92k, 92k, | |
1205 | 48k, 48k, 48k ... | |
f80dba8d | 1206 | |
5cb8a8cd | 1207 | .. option:: unified_rw_reporting=str |
f80dba8d MT |
1208 | |
1209 | Fio normally reports statistics on a per data direction basis, meaning that | |
5cb8a8cd BP |
1210 | reads, writes, and trims are accounted and reported separately. This option |
1211 | determines whether fio reports the results normally, summed together, or as | |
1212 | both options. | |
1213 | Accepted values are: | |
1214 | ||
1215 | **none** | |
1216 | Normal statistics reporting. | |
1217 | ||
1218 | **mixed** | |
1219 | Statistics are summed per data direction and reported together. | |
1220 | ||
1221 | **both** | |
1222 | Statistics are reported normally, followed by the mixed statistics. | |
1223 | ||
1224 | **0** | |
1225 | Backward-compatible alias for **none**. | |
1226 | ||
1227 | **1** | |
1228 | Backward-compatible alias for **mixed**. | |
9326926b | 1229 | |
5cb8a8cd BP |
1230 | **2** |
1231 | Alias for **both**. | |
f80dba8d MT |
1232 | |
1233 | .. option:: randrepeat=bool | |
1234 | ||
1235 | Seed the random number generator used for random I/O patterns in a | |
1236 | predictable way so the pattern is repeatable across runs. Default: true. | |
1237 | ||
1238 | .. option:: allrandrepeat=bool | |
1239 | ||
1240 | Seed all random number generators in a predictable way so results are | |
1241 | repeatable across runs. Default: false. | |
1242 | ||
1243 | .. option:: randseed=int | |
1244 | ||
1245 | Seed the random number generators based on this seed value, to be able to | |
1246 | control what sequence of output is being generated. If not set, the random | |
1247 | sequence depends on the :option:`randrepeat` setting. | |
1248 | ||
1249 | .. option:: fallocate=str | |
1250 | ||
1251 | Whether pre-allocation is performed when laying down files. | |
1252 | Accepted values are: | |
1253 | ||
1254 | **none** | |
1255 | Do not pre-allocate space. | |
1256 | ||
2c3e17be SW |
1257 | **native** |
1258 | Use a platform's native pre-allocation call but fall back to | |
1259 | **none** behavior if it fails/is not implemented. | |
1260 | ||
f80dba8d MT |
1261 | **posix** |
1262 | Pre-allocate via :manpage:`posix_fallocate(3)`. | |
1263 | ||
1264 | **keep** | |
1265 | Pre-allocate via :manpage:`fallocate(2)` with | |
1266 | FALLOC_FL_KEEP_SIZE set. | |
1267 | ||
38ca5f03 TV |
1268 | **truncate** |
1269 | Extend file to final size via :manpage:`ftruncate(2)` | |
1270 | instead of allocating. | |
1271 | ||
f80dba8d MT |
1272 | **0** |
1273 | Backward-compatible alias for **none**. | |
1274 | ||
1275 | **1** | |
1276 | Backward-compatible alias for **posix**. | |
1277 | ||
1278 | May not be available on all supported platforms. **keep** is only available | |
2c3e17be SW |
1279 | on Linux. If using ZFS on Solaris this cannot be set to **posix** |
1280 | because ZFS doesn't support pre-allocation. Default: **native** if any | |
38ca5f03 TV |
1281 | pre-allocation methods except **truncate** are available, **none** if not. |
1282 | ||
1283 | Note that using **truncate** on Windows will interact surprisingly | |
1284 | with non-sequential write patterns. When writing to a file that has | |
1285 | been extended by setting the end-of-file information, Windows will | |
1286 | backfill the unwritten portion of the file up to that offset with | |
1287 | zeroes before issuing the new write. This means that a single small | |
1288 | write to the end of an extended file will stall until the entire | |
1289 | file has been filled with zeroes. | |
f80dba8d MT |
1290 | |
1291 | .. option:: fadvise_hint=str | |
1292 | ||
c712c97a JA |
1293 | Use :manpage:`posix_fadvise(2)` or :manpage:`posix_fadvise(2)` to |
1294 | advise the kernel on what I/O patterns are likely to be issued. | |
1295 | Accepted values are: | |
f80dba8d MT |
1296 | |
1297 | **0** | |
1298 | Backwards-compatible hint for "no hint". | |
1299 | ||
1300 | **1** | |
1301 | Backwards compatible hint for "advise with fio workload type". This | |
1302 | uses **FADV_RANDOM** for a random workload, and **FADV_SEQUENTIAL** | |
1303 | for a sequential workload. | |
1304 | ||
1305 | **sequential** | |
1306 | Advise using **FADV_SEQUENTIAL**. | |
1307 | ||
1308 | **random** | |
1309 | Advise using **FADV_RANDOM**. | |
1310 | ||
8f4b9f24 | 1311 | .. option:: write_hint=str |
f80dba8d | 1312 | |
8f4b9f24 JA |
1313 | Use :manpage:`fcntl(2)` to advise the kernel what life time to expect |
1314 | from a write. Only supported on Linux, as of version 4.13. Accepted | |
1315 | values are: | |
1316 | ||
1317 | **none** | |
1318 | No particular life time associated with this file. | |
1319 | ||
1320 | **short** | |
1321 | Data written to this file has a short life time. | |
1322 | ||
1323 | **medium** | |
1324 | Data written to this file has a medium life time. | |
1325 | ||
1326 | **long** | |
1327 | Data written to this file has a long life time. | |
1328 | ||
1329 | **extreme** | |
1330 | Data written to this file has a very long life time. | |
1331 | ||
1332 | The values are all relative to each other, and no absolute meaning | |
1333 | should be associated with them. | |
f80dba8d MT |
1334 | |
1335 | .. option:: offset=int | |
1336 | ||
82dbb8cb | 1337 | Start I/O at the provided offset in the file, given as either a fixed size in |
193aaf6a | 1338 | bytes, zones or a percentage. If a percentage is given, the generated offset will be |
83c8b093 JF |
1339 | aligned to the minimum ``blocksize`` or to the value of ``offset_align`` if |
1340 | provided. Data before the given offset will not be touched. This | |
89978a6b BW |
1341 | effectively caps the file size at `real_size - offset`. Can be combined with |
1342 | :option:`size` to constrain the start and end range of the I/O workload. | |
44bb1142 | 1343 | A percentage can be specified by a number between 1 and 100 followed by '%', |
adcc0730 | 1344 | for example, ``offset=20%`` to specify 20%. In ZBD mode, value can be set as |
193aaf6a | 1345 | number of zones using 'z'. |
f80dba8d | 1346 | |
83c8b093 JF |
1347 | .. option:: offset_align=int |
1348 | ||
1349 | If set to non-zero value, the byte offset generated by a percentage ``offset`` | |
1350 | is aligned upwards to this value. Defaults to 0 meaning that a percentage | |
1351 | offset is aligned to the minimum block size. | |
1352 | ||
f80dba8d MT |
1353 | .. option:: offset_increment=int |
1354 | ||
1355 | If this is provided, then the real offset becomes `offset + offset_increment | |
1356 | * thread_number`, where the thread number is a counter that starts at 0 and | |
1357 | is incremented for each sub-job (i.e. when :option:`numjobs` option is | |
1358 | specified). This option is useful if there are several jobs which are | |
1359 | intended to operate on a file in parallel disjoint segments, with even | |
0b288ba1 VF |
1360 | spacing between the starting points. Percentages can be used for this option. |
1361 | If a percentage is given, the generated offset will be aligned to the minimum | |
193aaf6a G |
1362 | ``blocksize`` or to the value of ``offset_align`` if provided. In ZBD mode, value can |
1363 | also be set as number of zones using 'z'. | |
f80dba8d MT |
1364 | |
1365 | .. option:: number_ios=int | |
1366 | ||
c60ebc45 | 1367 | Fio will normally perform I/Os until it has exhausted the size of the region |
f80dba8d MT |
1368 | set by :option:`size`, or if it exhaust the allocated time (or hits an error |
1369 | condition). With this setting, the range/size can be set independently of | |
c60ebc45 | 1370 | the number of I/Os to perform. When fio reaches this number, it will exit |
f80dba8d MT |
1371 | normally and report status. Note that this does not extend the amount of I/O |
1372 | that will be done, it will only stop fio if this condition is met before | |
1373 | other end-of-job criteria. | |
1374 | ||
1375 | .. option:: fsync=int | |
1376 | ||
730bd7d9 SW |
1377 | If writing to a file, issue an :manpage:`fsync(2)` (or its equivalent) of |
1378 | the dirty data for every number of blocks given. For example, if you give 32 | |
1379 | as a parameter, fio will sync the file after every 32 writes issued. If fio is | |
1380 | using non-buffered I/O, we may not sync the file. The exception is the sg | |
1381 | I/O engine, which synchronizes the disk cache anyway. Defaults to 0, which | |
1382 | means fio does not periodically issue and wait for a sync to complete. Also | |
1383 | see :option:`end_fsync` and :option:`fsync_on_close`. | |
f80dba8d MT |
1384 | |
1385 | .. option:: fdatasync=int | |
1386 | ||
1387 | Like :option:`fsync` but uses :manpage:`fdatasync(2)` to only sync data and | |
2550c71f | 1388 | not metadata blocks. In Windows, DragonFlyBSD or OSX there is no |
730bd7d9 SW |
1389 | :manpage:`fdatasync(2)` so this falls back to using :manpage:`fsync(2)`. |
1390 | Defaults to 0, which means fio does not periodically issue and wait for a | |
1391 | data-only sync to complete. | |
f80dba8d MT |
1392 | |
1393 | .. option:: write_barrier=int | |
1394 | ||
2831be97 | 1395 | Make every `N-th` write a barrier write. |
f80dba8d | 1396 | |
f50fbdda | 1397 | .. option:: sync_file_range=str:int |
f80dba8d | 1398 | |
f50fbdda | 1399 | Use :manpage:`sync_file_range(2)` for every `int` number of write |
f80dba8d MT |
1400 | operations. Fio will track range of writes that have happened since the last |
1401 | :manpage:`sync_file_range(2)` call. `str` can currently be one or more of: | |
1402 | ||
1403 | **wait_before** | |
1404 | SYNC_FILE_RANGE_WAIT_BEFORE | |
1405 | **write** | |
1406 | SYNC_FILE_RANGE_WRITE | |
1407 | **wait_after** | |
1408 | SYNC_FILE_RANGE_WAIT_AFTER | |
1409 | ||
1410 | So if you do ``sync_file_range=wait_before,write:8``, fio would use | |
1411 | ``SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE`` for every 8 | |
1412 | writes. Also see the :manpage:`sync_file_range(2)` man page. This option is | |
1413 | Linux specific. | |
1414 | ||
1415 | .. option:: overwrite=bool | |
1416 | ||
1417 | If true, writes to a file will always overwrite existing data. If the file | |
1418 | doesn't already exist, it will be created before the write phase begins. If | |
1419 | the file exists and is large enough for the specified write phase, nothing | |
a47b697c | 1420 | will be done. Default: false. |
f80dba8d MT |
1421 | |
1422 | .. option:: end_fsync=bool | |
1423 | ||
a47b697c SW |
1424 | If true, :manpage:`fsync(2)` file contents when a write stage has completed. |
1425 | Default: false. | |
f80dba8d MT |
1426 | |
1427 | .. option:: fsync_on_close=bool | |
1428 | ||
1429 | If true, fio will :manpage:`fsync(2)` a dirty file on close. This differs | |
a47b697c SW |
1430 | from :option:`end_fsync` in that it will happen on every file close, not |
1431 | just at the end of the job. Default: false. | |
f80dba8d MT |
1432 | |
1433 | .. option:: rwmixread=int | |
1434 | ||
1435 | Percentage of a mixed workload that should be reads. Default: 50. | |
1436 | ||
1437 | .. option:: rwmixwrite=int | |
1438 | ||
1439 | Percentage of a mixed workload that should be writes. If both | |
1440 | :option:`rwmixread` and :option:`rwmixwrite` is given and the values do not | |
1441 | add up to 100%, the latter of the two will be used to override the | |
1442 | first. This may interfere with a given rate setting, if fio is asked to | |
1443 | limit reads or writes to a certain rate. If that is the case, then the | |
1444 | distribution may be skewed. Default: 50. | |
1445 | ||
a87c90fd | 1446 | .. option:: random_distribution=str:float[:float][,str:float][,str:float] |
f80dba8d MT |
1447 | |
1448 | By default, fio will use a completely uniform random distribution when asked | |
1449 | to perform random I/O. Sometimes it is useful to skew the distribution in | |
1450 | specific ways, ensuring that some parts of the data is more hot than others. | |
1451 | fio includes the following distribution models: | |
1452 | ||
1453 | **random** | |
1454 | Uniform random distribution | |
1455 | ||
1456 | **zipf** | |
1457 | Zipf distribution | |
1458 | ||
1459 | **pareto** | |
1460 | Pareto distribution | |
1461 | ||
b2f4b559 | 1462 | **normal** |
c60ebc45 | 1463 | Normal (Gaussian) distribution |
f80dba8d MT |
1464 | |
1465 | **zoned** | |
1466 | Zoned random distribution | |
1467 | ||
59466396 JA |
1468 | **zoned_abs** |
1469 | Zone absolute random distribution | |
1470 | ||
f80dba8d | 1471 | When using a **zipf** or **pareto** distribution, an input value is also |
f50fbdda | 1472 | needed to define the access pattern. For **zipf**, this is the `Zipf |
c60ebc45 | 1473 | theta`. For **pareto**, it's the `Pareto power`. Fio includes a test |
f50fbdda | 1474 | program, :command:`fio-genzipf`, that can be used visualize what the given input |
f80dba8d MT |
1475 | values will yield in terms of hit rates. If you wanted to use **zipf** with |
1476 | a `theta` of 1.2, you would use ``random_distribution=zipf:1.2`` as the | |
1477 | option. If a non-uniform model is used, fio will disable use of the random | |
b2f4b559 SW |
1478 | map. For the **normal** distribution, a normal (Gaussian) deviation is |
1479 | supplied as a value between 0 and 100. | |
f80dba8d | 1480 | |
a87c90fd | 1481 | The second, optional float is allowed for **pareto**, **zipf** and **normal** distributions. |
12efafa3 | 1482 | It allows one to set base of distribution in non-default place, giving more control |
a87c90fd AK |
1483 | over most probable outcome. This value is in range [0-1] which maps linearly to |
1484 | range of possible random values. | |
1485 | Defaults are: random for **pareto** and **zipf**, and 0.5 for **normal**. | |
1486 | If you wanted to use **zipf** with a `theta` of 1.2 centered on 1/4 of allowed value range, | |
fc002f14 | 1487 | you would use ``random_distribution=zipf:1.2:0.25``. |
a87c90fd | 1488 | |
f80dba8d MT |
1489 | For a **zoned** distribution, fio supports specifying percentages of I/O |
1490 | access that should fall within what range of the file or device. For | |
1491 | example, given a criteria of: | |
1492 | ||
f50fbdda TK |
1493 | * 60% of accesses should be to the first 10% |
1494 | * 30% of accesses should be to the next 20% | |
1495 | * 8% of accesses should be to the next 30% | |
1496 | * 2% of accesses should be to the next 40% | |
f80dba8d MT |
1497 | |
1498 | we can define that through zoning of the random accesses. For the above | |
1499 | example, the user would do:: | |
1500 | ||
1501 | random_distribution=zoned:60/10:30/20:8/30:2/40 | |
1502 | ||
59466396 JA |
1503 | A **zoned_abs** distribution works exactly like the **zoned**, except |
1504 | that it takes absolute sizes. For example, let's say you wanted to | |
1505 | define access according to the following criteria: | |
1506 | ||
1507 | * 60% of accesses should be to the first 20G | |
1508 | * 30% of accesses should be to the next 100G | |
1509 | * 10% of accesses should be to the next 500G | |
1510 | ||
1511 | we can define an absolute zoning distribution with: | |
1512 | ||
1513 | random_distribution=zoned_abs=60/20G:30/100G:10/500g | |
1514 | ||
6a16ece8 JA |
1515 | For both **zoned** and **zoned_abs**, fio supports defining up to |
1516 | 256 separate zones. | |
1517 | ||
59466396 JA |
1518 | Similarly to how :option:`bssplit` works for setting ranges and |
1519 | percentages of block sizes. Like :option:`bssplit`, it's possible to | |
1520 | specify separate zones for reads, writes, and trims. If just one set | |
1521 | is given, it'll apply to all of them. This goes for both **zoned** | |
1522 | **zoned_abs** distributions. | |
f80dba8d MT |
1523 | |
1524 | .. option:: percentage_random=int[,int][,int] | |
1525 | ||
1526 | For a random workload, set how big a percentage should be random. This | |
1527 | defaults to 100%, in which case the workload is fully random. It can be set | |
1528 | from anywhere from 0 to 100. Setting it to 0 would make the workload fully | |
1529 | sequential. Any setting in between will result in a random mix of sequential | |
1530 | and random I/O, at the given percentages. Comma-separated values may be | |
1531 | specified for reads, writes, and trims as described in :option:`blocksize`. | |
1532 | ||
1533 | .. option:: norandommap | |
1534 | ||
1535 | Normally fio will cover every block of the file when doing random I/O. If | |
1536 | this option is given, fio will just get a new random offset without looking | |
1537 | at past I/O history. This means that some blocks may not be read or written, | |
1538 | and that some blocks may be read/written more than once. If this option is | |
1539 | used with :option:`verify` and multiple blocksizes (via :option:`bsrange`), | |
1540 | only intact blocks are verified, i.e., partially-overwritten blocks are | |
47e6a6e5 SW |
1541 | ignored. With an async I/O engine and an I/O depth > 1, it is possible for |
1542 | the same block to be overwritten, which can cause verification errors. Either | |
1543 | do not use norandommap in this case, or also use the lfsr random generator. | |
f80dba8d MT |
1544 | |
1545 | .. option:: softrandommap=bool | |
1546 | ||
1547 | See :option:`norandommap`. If fio runs with the random block map enabled and | |
1548 | it fails to allocate the map, if this option is set it will continue without | |
1549 | a random block map. As coverage will not be as complete as with random maps, | |
1550 | this option is disabled by default. | |
1551 | ||
1552 | .. option:: random_generator=str | |
1553 | ||
f50fbdda | 1554 | Fio supports the following engines for generating I/O offsets for random I/O: |
f80dba8d MT |
1555 | |
1556 | **tausworthe** | |
f50fbdda | 1557 | Strong 2^88 cycle random number generator. |
f80dba8d | 1558 | **lfsr** |
f50fbdda | 1559 | Linear feedback shift register generator. |
f80dba8d | 1560 | **tausworthe64** |
f50fbdda | 1561 | Strong 64-bit 2^258 cycle random number generator. |
f80dba8d MT |
1562 | |
1563 | **tausworthe** is a strong random number generator, but it requires tracking | |
1564 | on the side if we want to ensure that blocks are only read or written | |
f50fbdda | 1565 | once. **lfsr** guarantees that we never generate the same offset twice, and |
f80dba8d | 1566 | it's also less computationally expensive. It's not a true random generator, |
f50fbdda | 1567 | however, though for I/O purposes it's typically good enough. **lfsr** only |
f80dba8d MT |
1568 | works with single block sizes, not with workloads that use multiple block |
1569 | sizes. If used with such a workload, fio may read or write some blocks | |
1570 | multiple times. The default value is **tausworthe**, unless the required | |
1571 | space exceeds 2^32 blocks. If it does, then **tausworthe64** is | |
1572 | selected automatically. | |
1573 | ||
1574 | ||
1575 | Block size | |
1576 | ~~~~~~~~~~ | |
1577 | ||
1578 | .. option:: blocksize=int[,int][,int], bs=int[,int][,int] | |
1579 | ||
1580 | The block size in bytes used for I/O units. Default: 4096. A single value | |
1581 | applies to reads, writes, and trims. Comma-separated values may be | |
1582 | specified for reads, writes, and trims. A value not terminated in a comma | |
1583 | applies to subsequent types. | |
1584 | ||
1585 | Examples: | |
1586 | ||
1587 | **bs=256k** | |
1588 | means 256k for reads, writes and trims. | |
1589 | ||
1590 | **bs=8k,32k** | |
1591 | means 8k for reads, 32k for writes and trims. | |
1592 | ||
1593 | **bs=8k,32k,** | |
1594 | means 8k for reads, 32k for writes, and default for trims. | |
1595 | ||
1596 | **bs=,8k** | |
1597 | means default for reads, 8k for writes and trims. | |
1598 | ||
1599 | **bs=,8k,** | |
b443ae44 | 1600 | means default for reads, 8k for writes, and default for trims. |
f80dba8d MT |
1601 | |
1602 | .. option:: blocksize_range=irange[,irange][,irange], bsrange=irange[,irange][,irange] | |
1603 | ||
1604 | A range of block sizes in bytes for I/O units. The issued I/O unit will | |
1605 | always be a multiple of the minimum size, unless | |
1606 | :option:`blocksize_unaligned` is set. | |
1607 | ||
1608 | Comma-separated ranges may be specified for reads, writes, and trims as | |
1609 | described in :option:`blocksize`. | |
1610 | ||
1611 | Example: ``bsrange=1k-4k,2k-8k``. | |
1612 | ||
1613 | .. option:: bssplit=str[,str][,str] | |
1614 | ||
6a16ece8 JA |
1615 | Sometimes you want even finer grained control of the block sizes |
1616 | issued, not just an even split between them. This option allows you to | |
1617 | weight various block sizes, so that you are able to define a specific | |
1618 | amount of block sizes issued. The format for this option is:: | |
f80dba8d MT |
1619 | |
1620 | bssplit=blocksize/percentage:blocksize/percentage | |
1621 | ||
6a16ece8 JA |
1622 | for as many block sizes as needed. So if you want to define a workload |
1623 | that has 50% 64k blocks, 10% 4k blocks, and 40% 32k blocks, you would | |
1624 | write:: | |
f80dba8d MT |
1625 | |
1626 | bssplit=4k/10:64k/50:32k/40 | |
1627 | ||
6a16ece8 JA |
1628 | Ordering does not matter. If the percentage is left blank, fio will |
1629 | fill in the remaining values evenly. So a bssplit option like this one:: | |
f80dba8d MT |
1630 | |
1631 | bssplit=4k/50:1k/:32k/ | |
1632 | ||
6a16ece8 JA |
1633 | would have 50% 4k ios, and 25% 1k and 32k ios. The percentages always |
1634 | add up to 100, if bssplit is given a range that adds up to more, it | |
1635 | will error out. | |
f80dba8d MT |
1636 | |
1637 | Comma-separated values may be specified for reads, writes, and trims as | |
1638 | described in :option:`blocksize`. | |
1639 | ||
6a16ece8 JA |
1640 | If you want a workload that has 50% 2k reads and 50% 4k reads, while |
1641 | having 90% 4k writes and 10% 8k writes, you would specify:: | |
f80dba8d | 1642 | |
cf04b906 | 1643 | bssplit=2k/50:4k/50,4k/90:8k/10 |
f80dba8d | 1644 | |
6a16ece8 JA |
1645 | Fio supports defining up to 64 different weights for each data |
1646 | direction. | |
1647 | ||
f80dba8d MT |
1648 | .. option:: blocksize_unaligned, bs_unaligned |
1649 | ||
1650 | If set, fio will issue I/O units with any size within | |
1651 | :option:`blocksize_range`, not just multiples of the minimum size. This | |
1652 | typically won't work with direct I/O, as that normally requires sector | |
1653 | alignment. | |
1654 | ||
589e88b7 | 1655 | .. option:: bs_is_seq_rand=bool |
f80dba8d MT |
1656 | |
1657 | If this option is set, fio will use the normal read,write blocksize settings | |
1658 | as sequential,random blocksize settings instead. Any random read or write | |
1659 | will use the WRITE blocksize settings, and any sequential read or write will | |
1660 | use the READ blocksize settings. | |
1661 | ||
1662 | .. option:: blockalign=int[,int][,int], ba=int[,int][,int] | |
1663 | ||
1664 | Boundary to which fio will align random I/O units. Default: | |
1665 | :option:`blocksize`. Minimum alignment is typically 512b for using direct | |
1666 | I/O, though it usually depends on the hardware block size. This option is | |
1667 | mutually exclusive with using a random map for files, so it will turn off | |
1668 | that option. Comma-separated values may be specified for reads, writes, and | |
1669 | trims as described in :option:`blocksize`. | |
1670 | ||
1671 | ||
1672 | Buffers and memory | |
1673 | ~~~~~~~~~~~~~~~~~~ | |
1674 | ||
1675 | .. option:: zero_buffers | |
1676 | ||
1677 | Initialize buffers with all zeros. Default: fill buffers with random data. | |
1678 | ||
1679 | .. option:: refill_buffers | |
1680 | ||
1681 | If this option is given, fio will refill the I/O buffers on every | |
72592780 SW |
1682 | submit. Only makes sense if :option:`zero_buffers` isn't specified, |
1683 | naturally. Defaults to being unset i.e., the buffer is only filled at | |
1684 | init time and the data in it is reused when possible but if any of | |
1685 | :option:`verify`, :option:`buffer_compress_percentage` or | |
1686 | :option:`dedupe_percentage` are enabled then `refill_buffers` is also | |
1687 | automatically enabled. | |
f80dba8d MT |
1688 | |
1689 | .. option:: scramble_buffers=bool | |
1690 | ||
1691 | If :option:`refill_buffers` is too costly and the target is using data | |
1692 | deduplication, then setting this option will slightly modify the I/O buffer | |
1693 | contents to defeat normal de-dupe attempts. This is not enough to defeat | |
1694 | more clever block compression attempts, but it will stop naive dedupe of | |
1695 | blocks. Default: true. | |
1696 | ||
1697 | .. option:: buffer_compress_percentage=int | |
1698 | ||
72592780 SW |
1699 | If this is set, then fio will attempt to provide I/O buffer content |
1700 | (on WRITEs) that compresses to the specified level. Fio does this by | |
1701 | providing a mix of random data followed by fixed pattern data. The | |
1702 | fixed pattern is either zeros, or the pattern specified by | |
1703 | :option:`buffer_pattern`. If the `buffer_pattern` option is used, it | |
1704 | might skew the compression ratio slightly. Setting | |
1705 | `buffer_compress_percentage` to a value other than 100 will also | |
1706 | enable :option:`refill_buffers` in order to reduce the likelihood that | |
1707 | adjacent blocks are so similar that they over compress when seen | |
1708 | together. See :option:`buffer_compress_chunk` for how to set a finer or | |
1709 | coarser granularity for the random/fixed data region. Defaults to unset | |
1710 | i.e., buffer data will not adhere to any compression level. | |
f80dba8d MT |
1711 | |
1712 | .. option:: buffer_compress_chunk=int | |
1713 | ||
72592780 SW |
1714 | This setting allows fio to manage how big the random/fixed data region |
1715 | is when using :option:`buffer_compress_percentage`. When | |
1716 | `buffer_compress_chunk` is set to some non-zero value smaller than the | |
1717 | block size, fio can repeat the random/fixed region throughout the I/O | |
1718 | buffer at the specified interval (which particularly useful when | |
1719 | bigger block sizes are used for a job). When set to 0, fio will use a | |
1720 | chunk size that matches the block size resulting in a single | |
1721 | random/fixed region within the I/O buffer. Defaults to 512. When the | |
1722 | unit is omitted, the value is interpreted in bytes. | |
f80dba8d MT |
1723 | |
1724 | .. option:: buffer_pattern=str | |
1725 | ||
a1554f65 SB |
1726 | If set, fio will fill the I/O buffers with this pattern or with the contents |
1727 | of a file. If not set, the contents of I/O buffers are defined by the other | |
1728 | options related to buffer contents. The setting can be any pattern of bytes, | |
1729 | and can be prefixed with 0x for hex values. It may also be a string, where | |
1730 | the string must then be wrapped with ``""``. Or it may also be a filename, | |
1731 | where the filename must be wrapped with ``''`` in which case the file is | |
1732 | opened and read. Note that not all the file contents will be read if that | |
1733 | would cause the buffers to overflow. So, for example:: | |
1734 | ||
1735 | buffer_pattern='filename' | |
1736 | ||
1737 | or:: | |
f80dba8d MT |
1738 | |
1739 | buffer_pattern="abcd" | |
1740 | ||
1741 | or:: | |
1742 | ||
1743 | buffer_pattern=-12 | |
1744 | ||
1745 | or:: | |
1746 | ||
1747 | buffer_pattern=0xdeadface | |
1748 | ||
1749 | Also you can combine everything together in any order:: | |
1750 | ||
a1554f65 | 1751 | buffer_pattern=0xdeadface"abcd"-12'filename' |
f80dba8d MT |
1752 | |
1753 | .. option:: dedupe_percentage=int | |
1754 | ||
1755 | If set, fio will generate this percentage of identical buffers when | |
1756 | writing. These buffers will be naturally dedupable. The contents of the | |
1757 | buffers depend on what other buffer compression settings have been set. It's | |
1758 | possible to have the individual buffers either fully compressible, or not at | |
72592780 SW |
1759 | all -- this option only controls the distribution of unique buffers. Setting |
1760 | this option will also enable :option:`refill_buffers` to prevent every buffer | |
1761 | being identical. | |
f80dba8d | 1762 | |
0d71aa98 BD |
1763 | .. option:: dedupe_mode=str |
1764 | ||
1765 | If ``dedupe_percentage=<int>`` is given, then this option controls how fio | |
1766 | generates the dedupe buffers. | |
1767 | ||
1768 | **repeat** | |
1769 | Generate dedupe buffers by repeating previous writes | |
1770 | **working_set** | |
1771 | Generate dedupe buffers from working set | |
1772 | ||
1773 | ``repeat`` is the default option for fio. Dedupe buffers are generated | |
1774 | by repeating previous unique write. | |
1775 | ||
1776 | ``working_set`` is a more realistic workload. | |
1777 | With ``working_set``, ``dedupe_working_set_percentage=<int>`` should be provided. | |
1778 | Given that, fio will use the initial unique write buffers as its working set. | |
1779 | Upon deciding to dedupe, fio will randomly choose a buffer from the working set. | |
1780 | Note that by using ``working_set`` the dedupe percentage will converge | |
1781 | to the desired over time while ``repeat`` maintains the desired percentage | |
1782 | throughout the job. | |
1783 | ||
1784 | .. option:: dedupe_working_set_percentage=int | |
1785 | ||
1786 | If ``dedupe_mode=<str>`` is set to ``working_set``, then this controls | |
1787 | the percentage of size of the file or device used as the buffers | |
1788 | fio will choose to generate the dedupe buffers from | |
1789 | ||
1790 | Note that size needs to be explicitly provided and only 1 file per | |
1791 | job is supported | |
1792 | ||
c49cfc76 BD |
1793 | .. option:: dedupe_global=bool |
1794 | ||
1795 | This controls whether the deduplication buffers will be shared amongst | |
1796 | all jobs that have this option set. The buffers are spread evenly between | |
1797 | participating jobs. | |
1798 | ||
f80dba8d MT |
1799 | .. option:: invalidate=bool |
1800 | ||
730bd7d9 SW |
1801 | Invalidate the buffer/page cache parts of the files to be used prior to |
1802 | starting I/O if the platform and file type support it. Defaults to true. | |
21c1b29e TK |
1803 | This will be ignored if :option:`pre_read` is also specified for the |
1804 | same job. | |
f80dba8d | 1805 | |
eb9f8d7f AF |
1806 | .. option:: sync=str |
1807 | ||
1808 | Whether, and what type, of synchronous I/O to use for writes. The allowed | |
1809 | values are: | |
1810 | ||
1811 | **none** | |
1812 | Do not use synchronous IO, the default. | |
1813 | ||
1814 | **0** | |
1815 | Same as **none**. | |
1816 | ||
1817 | **sync** | |
1818 | Use synchronous file IO. For the majority of I/O engines, | |
1819 | this means using O_SYNC. | |
1820 | ||
1821 | **1** | |
1822 | Same as **sync**. | |
1823 | ||
1824 | **dsync** | |
1825 | Use synchronous data IO. For the majority of I/O engines, | |
1826 | this means using O_DSYNC. | |
f80dba8d | 1827 | |
f80dba8d MT |
1828 | |
1829 | .. option:: iomem=str, mem=str | |
1830 | ||
1831 | Fio can use various types of memory as the I/O unit buffer. The allowed | |
1832 | values are: | |
1833 | ||
1834 | **malloc** | |
1835 | Use memory from :manpage:`malloc(3)` as the buffers. Default memory | |
1836 | type. | |
1837 | ||
1838 | **shm** | |
1839 | Use shared memory as the buffers. Allocated through | |
1840 | :manpage:`shmget(2)`. | |
1841 | ||
1842 | **shmhuge** | |
1843 | Same as shm, but use huge pages as backing. | |
1844 | ||
1845 | **mmap** | |
22413915 | 1846 | Use :manpage:`mmap(2)` to allocate buffers. May either be anonymous memory, or can |
f80dba8d MT |
1847 | be file backed if a filename is given after the option. The format |
1848 | is `mem=mmap:/path/to/file`. | |
1849 | ||
1850 | **mmaphuge** | |
1851 | Use a memory mapped huge file as the buffer backing. Append filename | |
1852 | after mmaphuge, ala `mem=mmaphuge:/hugetlbfs/file`. | |
1853 | ||
1854 | **mmapshared** | |
1855 | Same as mmap, but use a MMAP_SHARED mapping. | |
1856 | ||
03553853 YR |
1857 | **cudamalloc** |
1858 | Use GPU memory as the buffers for GPUDirect RDMA benchmark. | |
f50fbdda | 1859 | The :option:`ioengine` must be `rdma`. |
03553853 | 1860 | |
f80dba8d MT |
1861 | The area allocated is a function of the maximum allowed bs size for the job, |
1862 | multiplied by the I/O depth given. Note that for **shmhuge** and | |
1863 | **mmaphuge** to work, the system must have free huge pages allocated. This | |
1864 | can normally be checked and set by reading/writing | |
1865 | :file:`/proc/sys/vm/nr_hugepages` on a Linux system. Fio assumes a huge page | |
cb8dcafa VF |
1866 | is 2 or 4MiB in size depending on the platform. So to calculate the |
1867 | number of huge pages you need for a given job file, add up the I/O | |
1868 | depth of all jobs (normally one unless :option:`iodepth` is used) and | |
1869 | multiply by the maximum bs set. Then divide that number by the huge | |
1870 | page size. You can see the size of the huge pages in | |
1871 | :file:`/proc/meminfo`. If no huge pages are allocated by having a | |
1872 | non-zero number in `nr_hugepages`, using **mmaphuge** or **shmhuge** | |
1873 | will fail. Also see :option:`hugepage-size`. | |
f80dba8d MT |
1874 | |
1875 | **mmaphuge** also needs to have hugetlbfs mounted and the file location | |
1876 | should point there. So if it's mounted in :file:`/huge`, you would use | |
1877 | `mem=mmaphuge:/huge/somefile`. | |
1878 | ||
f50fbdda | 1879 | .. option:: iomem_align=int, mem_align=int |
f80dba8d MT |
1880 | |
1881 | This indicates the memory alignment of the I/O memory buffers. Note that | |
1882 | the given alignment is applied to the first I/O unit buffer, if using | |
1883 | :option:`iodepth` the alignment of the following buffers are given by the | |
1884 | :option:`bs` used. In other words, if using a :option:`bs` that is a | |
1885 | multiple of the page sized in the system, all buffers will be aligned to | |
1886 | this value. If using a :option:`bs` that is not page aligned, the alignment | |
1887 | of subsequent I/O memory buffers is the sum of the :option:`iomem_align` and | |
1888 | :option:`bs` used. | |
1889 | ||
1890 | .. option:: hugepage-size=int | |
1891 | ||
cb8dcafa VF |
1892 | Defines the size of a huge page. Must at least be equal to the system |
1893 | setting, see :file:`/proc/meminfo` and | |
1894 | :file:`/sys/kernel/mm/hugepages/`. Defaults to 2 or 4MiB depending on | |
1895 | the platform. Should probably always be a multiple of megabytes, so | |
1896 | using ``hugepage-size=Xm`` is the preferred way to set this to avoid | |
1897 | setting a non-pow-2 bad value. | |
f80dba8d MT |
1898 | |
1899 | .. option:: lockmem=int | |
1900 | ||
1901 | Pin the specified amount of memory with :manpage:`mlock(2)`. Can be used to | |
1902 | simulate a smaller amount of memory. The amount specified is per worker. | |
1903 | ||
1904 | ||
1905 | I/O size | |
1906 | ~~~~~~~~ | |
1907 | ||
1908 | .. option:: size=int | |
1909 | ||
79591fa9 | 1910 | The total size of file I/O for each thread of this job. Fio will run until |
942d66c8 AK |
1911 | this many bytes has been transferred, unless runtime is altered by other means |
1912 | such as (1) :option:`runtime`, (2) :option:`io_size` (3) :option:`number_ios`, | |
1913 | (4) gaps/holes while doing I/O's such as ``rw=read:16K``, or (5) sequential | |
1914 | I/O reaching end of the file which is possible when :option:`percentage_random` | |
1915 | is less than 100. | |
79591fa9 TK |
1916 | Fio will divide this size between the available files determined by options |
1917 | such as :option:`nrfiles`, :option:`filename`, unless :option:`filesize` is | |
1918 | specified by the job. If the result of division happens to be 0, the size is | |
c4aa2d08 | 1919 | set to the physical size of the given files or devices if they exist. |
79591fa9 | 1920 | If this option is not specified, fio will use the full size of the given |
f80dba8d MT |
1921 | files or devices. If the files do not exist, size must be given. It is also |
1922 | possible to give size as a percentage between 1 and 100. If ``size=20%`` is | |
adcc0730 | 1923 | given, fio will use 20% of the full size of the given files or devices. |
193aaf6a | 1924 | In ZBD mode, value can also be set as number of zones using 'z'. |
9d25d068 SW |
1925 | Can be combined with :option:`offset` to constrain the start and end range |
1926 | that I/O will be done within. | |
f80dba8d MT |
1927 | |
1928 | .. option:: io_size=int, io_limit=int | |
1929 | ||
1930 | Normally fio operates within the region set by :option:`size`, which means | |
1931 | that the :option:`size` option sets both the region and size of I/O to be | |
1932 | performed. Sometimes that is not what you want. With this option, it is | |
1933 | possible to define just the amount of I/O that fio should do. For instance, | |
1934 | if :option:`size` is set to 20GiB and :option:`io_size` is set to 5GiB, fio | |
1935 | will perform I/O within the first 20GiB but exit when 5GiB have been | |
1936 | done. The opposite is also possible -- if :option:`size` is set to 20GiB, | |
1937 | and :option:`io_size` is set to 40GiB, then fio will do 40GiB of I/O within | |
1938 | the 0..20GiB region. | |
1939 | ||
7fdd97ca | 1940 | .. option:: filesize=irange(int) |
f80dba8d | 1941 | |
2a929257 NR |
1942 | Individual file sizes. May be a range, in which case fio will select sizes for |
1943 | files at random within the given range. If not given, each created file is the | |
1944 | same size. This option overrides :option:`size` in terms of file size, i.e. if | |
1945 | :option:`filesize` is specified then :option:`size` becomes merely the default | |
1946 | for :option:`io_size` and has no effect at all if :option:`io_size` is set | |
1947 | explicitly. | |
f80dba8d MT |
1948 | |
1949 | .. option:: file_append=bool | |
1950 | ||
1951 | Perform I/O after the end of the file. Normally fio will operate within the | |
1952 | size of a file. If this option is set, then fio will append to the file | |
1953 | instead. This has identical behavior to setting :option:`offset` to the size | |
1954 | of a file. This option is ignored on non-regular files. | |
1955 | ||
1956 | .. option:: fill_device=bool, fill_fs=bool | |
1957 | ||
1958 | Sets size to something really large and waits for ENOSPC (no space left on | |
418f5399 MB |
1959 | device) or EDQUOT (disk quota exceeded) |
1960 | as the terminating condition. Only makes sense with sequential | |
f80dba8d MT |
1961 | write. For a read workload, the mount point will be filled first then I/O |
1962 | started on the result. This option doesn't make sense if operating on a raw | |
1963 | device node, since the size of that is already known by the file system. | |
1964 | Additionally, writing beyond end-of-device will not return ENOSPC there. | |
1965 | ||
1966 | ||
1967 | I/O engine | |
1968 | ~~~~~~~~~~ | |
1969 | ||
1970 | .. option:: ioengine=str | |
1971 | ||
1972 | Defines how the job issues I/O to the file. The following types are defined: | |
1973 | ||
1974 | **sync** | |
1975 | Basic :manpage:`read(2)` or :manpage:`write(2)` | |
1976 | I/O. :manpage:`lseek(2)` is used to position the I/O location. | |
54227e6b | 1977 | See :option:`fsync` and :option:`fdatasync` for syncing write I/Os. |
f80dba8d MT |
1978 | |
1979 | **psync** | |
1980 | Basic :manpage:`pread(2)` or :manpage:`pwrite(2)` I/O. Default on | |
1981 | all supported operating systems except for Windows. | |
1982 | ||
1983 | **vsync** | |
1984 | Basic :manpage:`readv(2)` or :manpage:`writev(2)` I/O. Will emulate | |
c60ebc45 | 1985 | queuing by coalescing adjacent I/Os into a single submission. |
f80dba8d MT |
1986 | |
1987 | **pvsync** | |
1988 | Basic :manpage:`preadv(2)` or :manpage:`pwritev(2)` I/O. | |
1989 | ||
1990 | **pvsync2** | |
1991 | Basic :manpage:`preadv2(2)` or :manpage:`pwritev2(2)` I/O. | |
1992 | ||
029b42ac JA |
1993 | **io_uring** |
1994 | Fast Linux native asynchronous I/O. Supports async IO | |
1995 | for both direct and buffered IO. | |
1996 | This engine defines engine specific options. | |
1997 | ||
3716f9f1 AK |
1998 | **io_uring_cmd** |
1999 | Fast Linux native asynchronous I/O for pass through commands. | |
2000 | This engine defines engine specific options. | |
2001 | ||
f80dba8d MT |
2002 | **libaio** |
2003 | Linux native asynchronous I/O. Note that Linux may only support | |
22413915 | 2004 | queued behavior with non-buffered I/O (set ``direct=1`` or |
f80dba8d MT |
2005 | ``buffered=0``). |
2006 | This engine defines engine specific options. | |
2007 | ||
2008 | **posixaio** | |
2009 | POSIX asynchronous I/O using :manpage:`aio_read(3)` and | |
2010 | :manpage:`aio_write(3)`. | |
2011 | ||
2012 | **solarisaio** | |
2013 | Solaris native asynchronous I/O. | |
2014 | ||
2015 | **windowsaio** | |
2016 | Windows native asynchronous I/O. Default on Windows. | |
2017 | ||
2018 | **mmap** | |
2019 | File is memory mapped with :manpage:`mmap(2)` and data copied | |
2020 | to/from using :manpage:`memcpy(3)`. | |
2021 | ||
2022 | **splice** | |
2023 | :manpage:`splice(2)` is used to transfer the data and | |
2024 | :manpage:`vmsplice(2)` to transfer data from user space to the | |
2025 | kernel. | |
2026 | ||
2027 | **sg** | |
2028 | SCSI generic sg v3 I/O. May either be synchronous using the SG_IO | |
2029 | ioctl, or if the target is an sg character device we use | |
2030 | :manpage:`read(2)` and :manpage:`write(2)` for asynchronous | |
f50fbdda | 2031 | I/O. Requires :option:`filename` option to specify either block or |
3740cfc8 | 2032 | character devices. This engine supports trim operations. |
52b81b7c | 2033 | The sg engine includes engine specific options. |
f80dba8d | 2034 | |
2455851d SK |
2035 | **libzbc** |
2036 | Read, write, trim and ZBC/ZAC operations to a zoned | |
2037 | block device using libzbc library. The target can be | |
2038 | either an SG character device or a block device file. | |
2039 | ||
f80dba8d MT |
2040 | **null** |
2041 | Doesn't transfer any data, just pretends to. This is mainly used to | |
2042 | exercise fio itself and for debugging/testing purposes. | |
2043 | ||
2044 | **net** | |
2045 | Transfer over the network to given ``host:port``. Depending on the | |
2046 | :option:`protocol` used, the :option:`hostname`, :option:`port`, | |
2047 | :option:`listen` and :option:`filename` options are used to specify | |
2048 | what sort of connection to make, while the :option:`protocol` option | |
2049 | determines which protocol will be used. This engine defines engine | |
2050 | specific options. | |
2051 | ||
2052 | **netsplice** | |
2053 | Like **net**, but uses :manpage:`splice(2)` and | |
2054 | :manpage:`vmsplice(2)` to map data and send/receive. | |
2055 | This engine defines engine specific options. | |
2056 | ||
2057 | **cpuio** | |
2058 | Doesn't transfer any data, but burns CPU cycles according to the | |
9de473a8 EV |
2059 | :option:`cpuload`, :option:`cpuchunks` and :option:`cpumode` options. |
2060 | Setting :option:`cpuload`\=85 will cause that job to do nothing but burn 85% | |
71aa48eb | 2061 | of the CPU. In case of SMP machines, use :option:`numjobs`\=<nr_of_cpu> |
f50fbdda | 2062 | to get desired CPU usage, as the cpuload only loads a |
f80dba8d MT |
2063 | single CPU at the desired rate. A job never finishes unless there is |
2064 | at least one non-cpuio job. | |
9de473a8 EV |
2065 | Setting :option:`cpumode`\=qsort replace the default noop instructions loop |
2066 | by a qsort algorithm to consume more energy. | |
f80dba8d | 2067 | |
f80dba8d MT |
2068 | **rdma** |
2069 | The RDMA I/O engine supports both RDMA memory semantics | |
2070 | (RDMA_WRITE/RDMA_READ) and channel semantics (Send/Recv) for the | |
609ac152 SB |
2071 | InfiniBand, RoCE and iWARP protocols. This engine defines engine |
2072 | specific options. | |
f80dba8d MT |
2073 | |
2074 | **falloc** | |
2075 | I/O engine that does regular fallocate to simulate data transfer as | |
2076 | fio ioengine. | |
2077 | ||
2078 | DDIR_READ | |
2079 | does fallocate(,mode = FALLOC_FL_KEEP_SIZE,). | |
2080 | ||
2081 | DDIR_WRITE | |
2082 | does fallocate(,mode = 0). | |
2083 | ||
2084 | DDIR_TRIM | |
2085 | does fallocate(,mode = FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE). | |
2086 | ||
761cd093 SW |
2087 | **ftruncate** |
2088 | I/O engine that sends :manpage:`ftruncate(2)` operations in response | |
2089 | to write (DDIR_WRITE) events. Each ftruncate issued sets the file's | |
f50fbdda | 2090 | size to the current block offset. :option:`blocksize` is ignored. |
761cd093 | 2091 | |
f80dba8d MT |
2092 | **e4defrag** |
2093 | I/O engine that does regular EXT4_IOC_MOVE_EXT ioctls to simulate | |
2094 | defragment activity in request to DDIR_WRITE event. | |
2095 | ||
f3f96717 IF |
2096 | **rados** |
2097 | I/O engine supporting direct access to Ceph Reliable Autonomic | |
2098 | Distributed Object Store (RADOS) via librados. This ioengine | |
2099 | defines engine specific options. | |
2100 | ||
f80dba8d MT |
2101 | **rbd** |
2102 | I/O engine supporting direct access to Ceph Rados Block Devices | |
2103 | (RBD) via librbd without the need to use the kernel rbd driver. This | |
2104 | ioengine defines engine specific options. | |
2105 | ||
c2f6a13d LMB |
2106 | **http** |
2107 | I/O engine supporting GET/PUT requests over HTTP(S) with libcurl to | |
2108 | a WebDAV or S3 endpoint. This ioengine defines engine specific options. | |
2109 | ||
2110 | This engine only supports direct IO of iodepth=1; you need to scale this | |
2111 | via numjobs. blocksize defines the size of the objects to be created. | |
2112 | ||
2113 | TRIM is translated to object deletion. | |
2114 | ||
f80dba8d | 2115 | **gfapi** |
ac8ca2af SW |
2116 | Using GlusterFS libgfapi sync interface to direct access to |
2117 | GlusterFS volumes without having to go through FUSE. This ioengine | |
f80dba8d MT |
2118 | defines engine specific options. |
2119 | ||
2120 | **gfapi_async** | |
ac8ca2af SW |
2121 | Using GlusterFS libgfapi async interface to direct access to |
2122 | GlusterFS volumes without having to go through FUSE. This ioengine | |
f80dba8d MT |
2123 | defines engine specific options. |
2124 | ||
2125 | **libhdfs** | |
f50fbdda | 2126 | Read and write through Hadoop (HDFS). The :option:`filename` option |
f80dba8d MT |
2127 | is used to specify host,port of the hdfs name-node to connect. This |
2128 | engine interprets offsets a little differently. In HDFS, files once | |
e25c0c91 SW |
2129 | created cannot be modified so random writes are not possible. To |
2130 | imitate this the libhdfs engine expects a bunch of small files to be | |
2131 | created over HDFS and will randomly pick a file from them | |
2132 | based on the offset generated by fio backend (see the example | |
f80dba8d | 2133 | job file to create such files, use ``rw=write`` option). Please |
e25c0c91 SW |
2134 | note, it may be necessary to set environment variables to work |
2135 | with HDFS/libhdfs properly. Each job uses its own connection to | |
f80dba8d MT |
2136 | HDFS. |
2137 | ||
2138 | **mtd** | |
2139 | Read, write and erase an MTD character device (e.g., | |
2140 | :file:`/dev/mtd0`). Discards are treated as erases. Depending on the | |
2141 | underlying device type, the I/O may have to go in a certain pattern, | |
2142 | e.g., on NAND, writing sequentially to erase blocks and discarding | |
c298ee71 | 2143 | before overwriting. The `trimwrite` mode works well for this |
f80dba8d MT |
2144 | constraint. |
2145 | ||
f80dba8d MT |
2146 | **dev-dax** |
2147 | Read and write using device DAX to a persistent memory device (e.g., | |
363a5f65 | 2148 | /dev/dax0.0) through the PMDK libpmem library. |
f80dba8d MT |
2149 | |
2150 | **external** | |
2151 | Prefix to specify loading an external I/O engine object file. Append | |
c60ebc45 | 2152 | the engine filename, e.g. ``ioengine=external:/tmp/foo.o`` to load |
d243fd6d TK |
2153 | ioengine :file:`foo.o` in :file:`/tmp`. The path can be either |
2154 | absolute or relative. See :file:`engines/skeleton_external.c` for | |
2155 | details of writing an external I/O engine. | |
f80dba8d | 2156 | |
1216cc5a | 2157 | **filecreate** |
b71968b1 | 2158 | Simply create the files and do no I/O to them. You still need to |
1216cc5a | 2159 | set `filesize` so that all the accounting still occurs, but no |
b71968b1 | 2160 | actual I/O will be done other than creating the file. |
f80dba8d | 2161 | |
73ccd14e SF |
2162 | **filestat** |
2163 | Simply do stat() and do no I/O to the file. You need to set 'filesize' | |
2164 | and 'nrfiles', so that files will be created. | |
2165 | This engine is to measure file lookup and meta data access. | |
2166 | ||
5561e9dd FS |
2167 | **filedelete** |
2168 | Simply delete the files by unlink() and do no I/O to them. You need to set 'filesize' | |
2169 | and 'nrfiles', so that the files will be created. | |
2170 | This engine is to measure file delete. | |
2171 | ||
ae0db592 TI |
2172 | **libpmem** |
2173 | Read and write using mmap I/O to a file on a filesystem | |
363a5f65 | 2174 | mounted with DAX on a persistent memory device through the PMDK |
ae0db592 TI |
2175 | libpmem library. |
2176 | ||
a40e7a59 GB |
2177 | **ime_psync** |
2178 | Synchronous read and write using DDN's Infinite Memory Engine (IME). | |
2179 | This engine is very basic and issues calls to IME whenever an IO is | |
2180 | queued. | |
2181 | ||
2182 | **ime_psyncv** | |
2183 | Synchronous read and write using DDN's Infinite Memory Engine (IME). | |
2184 | This engine uses iovecs and will try to stack as much IOs as possible | |
2185 | (if the IOs are "contiguous" and the IO depth is not exceeded) | |
2186 | before issuing a call to IME. | |
2187 | ||
2188 | **ime_aio** | |
2189 | Asynchronous read and write using DDN's Infinite Memory Engine (IME). | |
2190 | This engine will try to stack as much IOs as possible by creating | |
2191 | requests for IME. FIO will then decide when to commit these requests. | |
68522f38 | 2192 | |
247ef2aa KZ |
2193 | **libiscsi** |
2194 | Read and write iscsi lun with libiscsi. | |
68522f38 | 2195 | |
d643a1e2 | 2196 | **nbd** |
f2d6de5d | 2197 | Read and write a Network Block Device (NBD). |
a40e7a59 | 2198 | |
10756b2c BS |
2199 | **libcufile** |
2200 | I/O engine supporting libcufile synchronous access to nvidia-fs and a | |
2201 | GPUDirect Storage-supported filesystem. This engine performs | |
2202 | I/O without transferring buffers between user-space and the kernel, | |
2203 | unless :option:`verify` is set or :option:`cuda_io` is `posix`. | |
2204 | :option:`iomem` must not be `cudamalloc`. This ioengine defines | |
2205 | engine specific options. | |
68522f38 | 2206 | |
c363fdd7 JL |
2207 | **dfs** |
2208 | I/O engine supporting asynchronous read and write operations to the | |
2209 | DAOS File System (DFS) via libdfs. | |
10756b2c | 2210 | |
9326926b TG |
2211 | **nfs** |
2212 | I/O engine supporting asynchronous read and write operations to | |
2213 | NFS filesystems from userspace via libnfs. This is useful for | |
2214 | achieving higher concurrency and thus throughput than is possible | |
2215 | via kernel NFS. | |
2216 | ||
b50590bc EV |
2217 | **exec** |
2218 | Execute 3rd party tools. Could be used to perform monitoring during jobs runtime. | |
2219 | ||
454154e6 AK |
2220 | **xnvme** |
2221 | I/O engine using the xNVMe C API, for NVMe devices. The xnvme engine provides | |
2222 | flexibility to access GNU/Linux Kernel NVMe driver via libaio, IOCTLs, io_uring, | |
2223 | the SPDK NVMe driver, or your own custom NVMe driver. The xnvme engine includes | |
2224 | engine specific options. (See https://xnvme.io). | |
2225 | ||
a601337a AF |
2226 | **libblkio** |
2227 | Use the libblkio library | |
2228 | (https://gitlab.com/libblkio/libblkio). The specific | |
2229 | *driver* to use must be set using | |
ef9b6f2f AF |
2230 | :option:`libblkio_driver`. If |
2231 | :option:`mem`/:option:`iomem` is not specified, memory | |
2232 | allocation is delegated to libblkio (and so is | |
3afc2d8a AF |
2233 | guaranteed to work with the selected *driver*). One |
2234 | libblkio instance is used per process, so all jobs | |
2235 | setting option :option:`thread` will share a single | |
2236 | instance (with one queue per thread) and must specify | |
2237 | compatible options. Note that some drivers don't allow | |
2238 | several instances to access the same device or file | |
2239 | simultaneously, but allow it for threads. | |
a601337a | 2240 | |
f80dba8d MT |
2241 | I/O engine specific parameters |
2242 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
2243 | ||
2244 | In addition, there are some parameters which are only valid when a specific | |
f50fbdda TK |
2245 | :option:`ioengine` is in use. These are used identically to normal parameters, |
2246 | with the caveat that when used on the command line, they must come after the | |
f80dba8d MT |
2247 | :option:`ioengine` that defines them is selected. |
2248 | ||
e9f6567a DLM |
2249 | .. option:: cmdprio_percentage=int[,int] : [io_uring] [libaio] |
2250 | ||
2251 | Set the percentage of I/O that will be issued with the highest priority. | |
2252 | Default: 0. A single value applies to reads and writes. Comma-separated | |
acf2e2d9 | 2253 | values may be specified for reads and writes. For this option to be |
68522f38 VF |
2254 | effective, NCQ priority must be supported and enabled, and the :option:`direct` |
2255 | option must be set. fio must also be run as the root user. Unlike | |
bebf1407 NC |
2256 | slat/clat/lat stats, which can be tracked and reported independently, per |
2257 | priority stats only track and report a single type of latency. By default, | |
2258 | completion latency (clat) will be reported, if :option:`lat_percentiles` is | |
2259 | set, total latency (lat) will be reported. | |
029b42ac | 2260 | |
12f9d54a DLM |
2261 | .. option:: cmdprio_class=int[,int] : [io_uring] [libaio] |
2262 | ||
2263 | Set the I/O priority class to use for I/Os that must be issued with | |
a48f0cc7 DLM |
2264 | a priority when :option:`cmdprio_percentage` or |
2265 | :option:`cmdprio_bssplit` is set. If not specified when | |
2266 | :option:`cmdprio_percentage` or :option:`cmdprio_bssplit` is set, | |
2267 | this defaults to the highest priority class. A single value applies | |
2268 | to reads and writes. Comma-separated values may be specified for | |
2269 | reads and writes. See :manpage:`ionice(1)`. See also the | |
2270 | :option:`prioclass` option. | |
12f9d54a DLM |
2271 | |
2272 | .. option:: cmdprio=int[,int] : [io_uring] [libaio] | |
2273 | ||
2274 | Set the I/O priority value to use for I/Os that must be issued with | |
a48f0cc7 DLM |
2275 | a priority when :option:`cmdprio_percentage` or |
2276 | :option:`cmdprio_bssplit` is set. If not specified when | |
2277 | :option:`cmdprio_percentage` or :option:`cmdprio_bssplit` is set, | |
2278 | this defaults to 0. | |
12f9d54a DLM |
2279 | Linux limits us to a positive value between 0 and 7, with 0 being the |
2280 | highest. A single value applies to reads and writes. Comma-separated | |
2281 | values may be specified for reads and writes. See :manpage:`ionice(1)`. | |
2282 | Refer to an appropriate manpage for other operating systems since | |
2283 | meaning of priority may differ. See also the :option:`prio` option. | |
2284 | ||
a48f0cc7 | 2285 | .. option:: cmdprio_bssplit=str[,str] : [io_uring] [libaio] |
68522f38 | 2286 | |
a48f0cc7 DLM |
2287 | To get a finer control over I/O priority, this option allows |
2288 | specifying the percentage of IOs that must have a priority set | |
2289 | depending on the block size of the IO. This option is useful only | |
2290 | when used together with the :option:`bssplit` option, that is, | |
2291 | multiple different block sizes are used for reads and writes. | |
f0547200 NC |
2292 | |
2293 | The first accepted format for this option is the same as the format of | |
2294 | the :option:`bssplit` option: | |
2295 | ||
2296 | cmdprio_bssplit=blocksize/percentage:blocksize/percentage | |
2297 | ||
2298 | In this case, each entry will use the priority class and priority | |
2299 | level defined by the options :option:`cmdprio_class` and | |
2300 | :option:`cmdprio` respectively. | |
2301 | ||
2302 | The second accepted format for this option is: | |
2303 | ||
2304 | cmdprio_bssplit=blocksize/percentage/class/level:blocksize/percentage/class/level | |
2305 | ||
2306 | In this case, the priority class and priority level is defined inside | |
2307 | each entry. In comparison with the first accepted format, the second | |
2308 | accepted format does not restrict all entries to have the same priority | |
2309 | class and priority level. | |
2310 | ||
2311 | For both formats, only the read and write data directions are supported, | |
2312 | values for trim IOs are ignored. This option is mutually exclusive with | |
2313 | the :option:`cmdprio_percentage` option. | |
a48f0cc7 | 2314 | |
3716f9f1 AK |
2315 | .. option:: fixedbufs : [io_uring] [io_uring_cmd] |
2316 | ||
2317 | If fio is asked to do direct IO, then Linux will map pages for each | |
2318 | IO call, and release them when IO is done. If this option is set, the | |
2319 | pages are pre-mapped before IO is started. This eliminates the need to | |
2320 | map and release for each IO. This is more efficient, and reduces the | |
2321 | IO latency as well. | |
2322 | ||
d6f936d1 | 2323 | .. option:: nonvectored=int : [io_uring] [io_uring_cmd] |
029b42ac | 2324 | |
3716f9f1 AK |
2325 | With this option, fio will use non-vectored read/write commands, where |
2326 | address must contain the address directly. Default is -1. | |
b2a432bf | 2327 | |
3716f9f1 AK |
2328 | .. option:: force_async=int : [io_uring] [io_uring_cmd] |
2329 | ||
2330 | Normal operation for io_uring is to try and issue an sqe as | |
2331 | non-blocking first, and if that fails, execute it in an async manner. | |
2332 | With this option set to N, then every N request fio will ask sqe to | |
2333 | be issued in an async manner. Default is 0. | |
2334 | ||
2335 | .. option:: registerfiles : [io_uring] [io_uring_cmd] | |
2c870598 | 2336 | |
5ffd5626 JA |
2337 | With this option, fio registers the set of files being used with the |
2338 | kernel. This avoids the overhead of managing file counts in the kernel, | |
2339 | making the submission and completion part more lightweight. Required | |
2340 | for the below :option:`sqthread_poll` option. | |
2341 | ||
3716f9f1 | 2342 | .. option:: sqthread_poll : [io_uring] [io_uring_cmd] [xnvme] |
029b42ac JA |
2343 | |
2344 | Normally fio will submit IO by issuing a system call to notify the | |
2345 | kernel of available items in the SQ ring. If this option is set, the | |
2346 | act of submitting IO will be done by a polling thread in the kernel. | |
2347 | This frees up cycles for fio, at the cost of using more CPU in the | |
72044c66 AK |
2348 | system. As submission is just the time it takes to fill in the sqe |
2349 | entries and any syscall required to wake up the idle kernel thread, | |
2350 | fio will not report submission latencies. | |
029b42ac | 2351 | |
d6f936d1 | 2352 | .. option:: sqthread_poll_cpu=int : [io_uring] [io_uring_cmd] |
029b42ac JA |
2353 | |
2354 | When :option:`sqthread_poll` is set, this option provides a way to | |
2355 | define which CPU should be used for the polling thread. | |
2356 | ||
3716f9f1 AK |
2357 | .. option:: cmd_type=str : [io_uring_cmd] |
2358 | ||
2359 | Specifies the type of uring passthrough command to be used. Supported | |
2360 | value is nvme. Default is nvme. | |
2361 | ||
8253a66b VF |
2362 | .. option:: hipri |
2363 | ||
3716f9f1 | 2364 | [io_uring] [io_uring_cmd] [xnvme] |
8253a66b VF |
2365 | |
2366 | If this option is set, fio will attempt to use polled IO completions. | |
2367 | Normal IO completions generate interrupts to signal the completion of | |
2368 | IO, polled completions do not. Hence they are require active reaping | |
2369 | by the application. The benefits are more efficient IO for high IOPS | |
2370 | scenarios, and lower latencies for low queue depth IO. | |
2371 | ||
a870d6ff AF |
2372 | [libblkio] |
2373 | ||
b158577d | 2374 | Use poll queues. This is incompatible with |
b1bd09b5 AF |
2375 | :option:`libblkio_wait_mode=eventfd <libblkio_wait_mode>` and |
2376 | :option:`libblkio_force_enable_completion_eventfd`. | |
a870d6ff | 2377 | |
8253a66b VF |
2378 | [pvsync2] |
2379 | ||
2380 | Set RWF_HIPRI on I/O, indicating to the kernel that it's of higher priority | |
2381 | than normal. | |
2382 | ||
8253a66b VF |
2383 | [sg] |
2384 | ||
2385 | If this option is set, fio will attempt to use polled IO completions. | |
2386 | This will have a similar effect as (io_uring)hipri. Only SCSI READ and | |
2387 | WRITE commands will have the SGV4_FLAG_HIPRI set (not UNMAP (trim) nor | |
2388 | VERIFY). Older versions of the Linux sg driver that do not support | |
2389 | hipri will simply ignore this flag and do normal IO. The Linux SCSI | |
2390 | Low Level Driver (LLD) that "owns" the device also needs to support | |
2391 | hipri (also known as iopoll and mq_poll). The MegaRAID driver is an | |
2392 | example of a SCSI LLD. Default: clear (0) which does normal | |
2393 | (interrupted based) IO. | |
2394 | ||
f80dba8d MT |
2395 | .. option:: userspace_reap : [libaio] |
2396 | ||
2397 | Normally, with the libaio engine in use, fio will use the | |
2398 | :manpage:`io_getevents(2)` system call to reap newly returned events. With | |
2399 | this flag turned on, the AIO ring will be read directly from user-space to | |
2400 | reap events. The reaping mode is only enabled when polling for a minimum of | |
c60ebc45 | 2401 | 0 events (e.g. when :option:`iodepth_batch_complete` `=0`). |
f80dba8d | 2402 | |
a0679ce5 SB |
2403 | .. option:: hipri_percentage : [pvsync2] |
2404 | ||
f50fbdda | 2405 | When hipri is set this determines the probability of a pvsync2 I/O being high |
a0679ce5 SB |
2406 | priority. The default is 100%. |
2407 | ||
d6f936d1 | 2408 | .. option:: nowait=bool : [pvsync2] [libaio] [io_uring] [io_uring_cmd] |
7d42e66e KK |
2409 | |
2410 | By default if a request cannot be executed immediately (e.g. resource starvation, | |
2411 | waiting on locks) it is queued and the initiating process will be blocked until | |
2412 | the required resource becomes free. | |
2413 | ||
2414 | This option sets the RWF_NOWAIT flag (supported from the 4.14 Linux kernel) and | |
2415 | the call will return instantly with EAGAIN or a partial result rather than waiting. | |
2416 | ||
2417 | It is useful to also use ignore_error=EAGAIN when using this option. | |
2418 | ||
2419 | Note: glibc 2.27, 2.28 have a bug in syscall wrappers preadv2, pwritev2. | |
2420 | They return EOPNOTSUP instead of EAGAIN. | |
2421 | ||
2422 | For cached I/O, using this option usually means a request operates only with | |
2423 | cached data. Currently the RWF_NOWAIT flag does not supported for cached write. | |
2424 | ||
2425 | For direct I/O, requests will only succeed if cache invalidation isn't required, | |
2426 | file blocks are fully allocated and the disk request could be issued immediately. | |
2427 | ||
a7e8aae0 KB |
2428 | .. option:: fdp=bool : [io_uring_cmd] |
2429 | ||
2430 | Enable Flexible Data Placement mode for write commands. | |
2431 | ||
2432 | .. option:: fdp_pli=str : [io_uring_cmd] | |
2433 | ||
2434 | Select which Placement ID Index/Indicies this job is allowed to use for | |
2435 | writes. By default, the job will cycle through all available Placement | |
2436 | IDs, so use this to isolate these identifiers to specific jobs. If you | |
2437 | want fio to use placement identifier only at indices 0, 2 and 5 specify | |
2438 | ``fdp_pli=0,2,5``. | |
2439 | ||
f80dba8d MT |
2440 | .. option:: cpuload=int : [cpuio] |
2441 | ||
da19cdb4 TK |
2442 | Attempt to use the specified percentage of CPU cycles. This is a mandatory |
2443 | option when using cpuio I/O engine. | |
f80dba8d MT |
2444 | |
2445 | .. option:: cpuchunks=int : [cpuio] | |
2446 | ||
2447 | Split the load into cycles of the given time. In microseconds. | |
2448 | ||
8a7bf04c VF |
2449 | .. option:: cpumode=str : [cpuio] |
2450 | ||
2451 | Specify how to stress the CPU. It can take these two values: | |
2452 | ||
2453 | **noop** | |
2454 | This is the default where the CPU executes noop instructions. | |
2455 | **qsort** | |
2456 | Replace the default noop instructions loop with a qsort algorithm to | |
2457 | consume more energy. | |
2458 | ||
f80dba8d MT |
2459 | .. option:: exit_on_io_done=bool : [cpuio] |
2460 | ||
2461 | Detect when I/O threads are done, then exit. | |
2462 | ||
f80dba8d MT |
2463 | .. option:: namenode=str : [libhdfs] |
2464 | ||
22413915 | 2465 | The hostname or IP address of a HDFS cluster namenode to contact. |
f80dba8d MT |
2466 | |
2467 | .. option:: port=int | |
2468 | ||
f50fbdda TK |
2469 | [libhdfs] |
2470 | ||
2471 | The listening port of the HFDS cluster namenode. | |
2472 | ||
f80dba8d MT |
2473 | [netsplice], [net] |
2474 | ||
2475 | The TCP or UDP port to bind to or connect to. If this is used with | |
2476 | :option:`numjobs` to spawn multiple instances of the same job type, then | |
2477 | this will be the starting port number since fio will use a range of | |
2478 | ports. | |
2479 | ||
e4c4625f | 2480 | [rdma], [librpma_*] |
609ac152 SB |
2481 | |
2482 | The port to use for RDMA-CM communication. This should be the same value | |
2483 | on the client and the server side. | |
2484 | ||
2485 | .. option:: hostname=str : [netsplice] [net] [rdma] | |
f80dba8d | 2486 | |
609ac152 SB |
2487 | The hostname or IP address to use for TCP, UDP or RDMA-CM based I/O. If the job |
2488 | is a TCP listener or UDP reader, the hostname is not used and must be omitted | |
f50fbdda | 2489 | unless it is a valid UDP multicast address. |
f80dba8d | 2490 | |
e4c4625f JM |
2491 | .. option:: serverip=str : [librpma_*] |
2492 | ||
2493 | The IP address to be used for RDMA-CM based I/O. | |
2494 | ||
2495 | .. option:: direct_write_to_pmem=bool : [librpma_*] | |
2496 | ||
2497 | Set to 1 only when Direct Write to PMem from the remote host is possible. | |
2498 | Otherwise, set to 0. | |
2499 | ||
6a229978 OS |
2500 | .. option:: busy_wait_polling=bool : [librpma_*_server] |
2501 | ||
2502 | Set to 0 to wait for completion instead of busy-wait polling completion. | |
2503 | Default: 1. | |
2504 | ||
f80dba8d MT |
2505 | .. option:: interface=str : [netsplice] [net] |
2506 | ||
2507 | The IP address of the network interface used to send or receive UDP | |
2508 | multicast. | |
2509 | ||
2510 | .. option:: ttl=int : [netsplice] [net] | |
2511 | ||
2512 | Time-to-live value for outgoing UDP multicast packets. Default: 1. | |
2513 | ||
2514 | .. option:: nodelay=bool : [netsplice] [net] | |
2515 | ||
2516 | Set TCP_NODELAY on TCP connections. | |
2517 | ||
f50fbdda | 2518 | .. option:: protocol=str, proto=str : [netsplice] [net] |
f80dba8d MT |
2519 | |
2520 | The network protocol to use. Accepted values are: | |
2521 | ||
2522 | **tcp** | |
2523 | Transmission control protocol. | |
2524 | **tcpv6** | |
2525 | Transmission control protocol V6. | |
2526 | **udp** | |
2527 | User datagram protocol. | |
2528 | **udpv6** | |
2529 | User datagram protocol V6. | |
2530 | **unix** | |
2531 | UNIX domain socket. | |
2532 | ||
2533 | When the protocol is TCP or UDP, the port must also be given, as well as the | |
2534 | hostname if the job is a TCP listener or UDP reader. For unix sockets, the | |
f50fbdda | 2535 | normal :option:`filename` option should be used and the port is invalid. |
f80dba8d | 2536 | |
e9184ec1 | 2537 | .. option:: listen : [netsplice] [net] |
f80dba8d MT |
2538 | |
2539 | For TCP network connections, tell fio to listen for incoming connections | |
2540 | rather than initiating an outgoing connection. The :option:`hostname` must | |
2541 | be omitted if this option is used. | |
2542 | ||
e9184ec1 | 2543 | .. option:: pingpong : [netsplice] [net] |
f80dba8d MT |
2544 | |
2545 | Normally a network writer will just continue writing data, and a network | |
2546 | reader will just consume packages. If ``pingpong=1`` is set, a writer will | |
2547 | send its normal payload to the reader, then wait for the reader to send the | |
2548 | same payload back. This allows fio to measure network latencies. The | |
2549 | submission and completion latencies then measure local time spent sending or | |
2550 | receiving, and the completion latency measures how long it took for the | |
2551 | other end to receive and send back. For UDP multicast traffic | |
2552 | ``pingpong=1`` should only be set for a single reader when multiple readers | |
2553 | are listening to the same address. | |
2554 | ||
e9184ec1 | 2555 | .. option:: window_size : [netsplice] [net] |
f80dba8d MT |
2556 | |
2557 | Set the desired socket buffer size for the connection. | |
2558 | ||
e9184ec1 | 2559 | .. option:: mss : [netsplice] [net] |
f80dba8d MT |
2560 | |
2561 | Set the TCP maximum segment size (TCP_MAXSEG). | |
2562 | ||
2563 | .. option:: donorname=str : [e4defrag] | |
2564 | ||
730bd7d9 | 2565 | File will be used as a block donor (swap extents between files). |
f80dba8d MT |
2566 | |
2567 | .. option:: inplace=int : [e4defrag] | |
2568 | ||
2569 | Configure donor file blocks allocation strategy: | |
2570 | ||
2571 | **0** | |
2572 | Default. Preallocate donor's file on init. | |
2573 | **1** | |
2b455dbf | 2574 | Allocate space immediately inside defragment event, and free right |
f80dba8d MT |
2575 | after event. |
2576 | ||
f3f96717 | 2577 | .. option:: clustername=str : [rbd,rados] |
f80dba8d MT |
2578 | |
2579 | Specifies the name of the Ceph cluster. | |
2580 | ||
2581 | .. option:: rbdname=str : [rbd] | |
2582 | ||
2583 | Specifies the name of the RBD. | |
2584 | ||
f3f96717 | 2585 | .. option:: clientname=str : [rbd,rados] |
f80dba8d MT |
2586 | |
2587 | Specifies the username (without the 'client.' prefix) used to access the | |
2588 | Ceph cluster. If the *clustername* is specified, the *clientname* shall be | |
2589 | the full *type.id* string. If no type. prefix is given, fio will add | |
2590 | 'client.' by default. | |
2591 | ||
873db854 | 2592 | .. option:: conf=str : [rados] |
2593 | ||
2594 | Specifies the configuration path of ceph cluster, so conf file does not | |
2595 | have to be /etc/ceph/ceph.conf. | |
2596 | ||
f3f96717 IF |
2597 | .. option:: busy_poll=bool : [rbd,rados] |
2598 | ||
2599 | Poll store instead of waiting for completion. Usually this provides better | |
2600 | throughput at cost of higher(up to 100%) CPU utilization. | |
2601 | ||
2b728756 AK |
2602 | .. option:: touch_objects=bool : [rados] |
2603 | ||
2604 | During initialization, touch (create if do not exist) all objects (files). | |
2605 | Touching all objects affects ceph caches and likely impacts test results. | |
2606 | Enabled by default. | |
2607 | ||
68522f38 VF |
2608 | .. option:: pool=str : |
2609 | ||
2610 | [rbd,rados] | |
2611 | ||
2612 | Specifies the name of the Ceph pool containing RBD or RADOS data. | |
2613 | ||
2614 | [dfs] | |
2615 | ||
2616 | Specify the label or UUID of the DAOS pool to connect to. | |
2617 | ||
2618 | .. option:: cont=str : [dfs] | |
2619 | ||
2620 | Specify the label or UUID of the DAOS container to open. | |
2621 | ||
19d8e50a VF |
2622 | .. option:: chunk_size=int |
2623 | ||
2624 | [dfs] | |
68522f38 | 2625 | |
ffe1d11f | 2626 | Specify a different chunk size (in bytes) for the dfs file. |
68522f38 VF |
2627 | Use DAOS container's chunk size by default. |
2628 | ||
19d8e50a VF |
2629 | [libhdfs] |
2630 | ||
2631 | The size of the chunk to use for each file. | |
2632 | ||
68522f38 VF |
2633 | .. option:: object_class=str : [dfs] |
2634 | ||
ffe1d11f | 2635 | Specify a different object class for the dfs file. |
68522f38 VF |
2636 | Use DAOS container's object class by default. |
2637 | ||
f80dba8d MT |
2638 | .. option:: skip_bad=bool : [mtd] |
2639 | ||
2640 | Skip operations against known bad blocks. | |
2641 | ||
2642 | .. option:: hdfsdirectory : [libhdfs] | |
2643 | ||
2644 | libhdfs will create chunk in this HDFS directory. | |
2645 | ||
609ac152 SB |
2646 | .. option:: verb=str : [rdma] |
2647 | ||
2648 | The RDMA verb to use on this side of the RDMA ioengine connection. Valid | |
2649 | values are write, read, send and recv. These correspond to the equivalent | |
2650 | RDMA verbs (e.g. write = rdma_write etc.). Note that this only needs to be | |
2651 | specified on the client side of the connection. See the examples folder. | |
2652 | ||
2653 | .. option:: bindname=str : [rdma] | |
2654 | ||
2655 | The name to use to bind the local RDMA-CM connection to a local RDMA device. | |
2656 | This could be a hostname or an IPv4 or IPv6 address. On the server side this | |
2657 | will be passed into the rdma_bind_addr() function and on the client site it | |
2658 | will be used in the rdma_resolve_add() function. This can be useful when | |
2659 | multiple paths exist between the client and the server or in certain loopback | |
2660 | configurations. | |
f80dba8d | 2661 | |
93a13ba5 | 2662 | .. option:: stat_type=str : [filestat] |
c446eff0 | 2663 | |
93a13ba5 TK |
2664 | Specify stat system call type to measure lookup/getattr performance. |
2665 | Default is **stat** for :manpage:`stat(2)`. | |
c446eff0 | 2666 | |
52b81b7c KD |
2667 | .. option:: readfua=bool : [sg] |
2668 | ||
2669 | With readfua option set to 1, read operations include | |
2670 | the force unit access (fua) flag. Default is 0. | |
2671 | ||
2672 | .. option:: writefua=bool : [sg] | |
2673 | ||
2674 | With writefua option set to 1, write operations include | |
2675 | the force unit access (fua) flag. Default is 0. | |
2676 | ||
2c3a9150 | 2677 | .. option:: sg_write_mode=str : [sg] |
3740cfc8 | 2678 | |
2c3a9150 VF |
2679 | Specify the type of write commands to issue. This option can take three values: |
2680 | ||
2681 | **write** | |
2682 | This is the default where write opcodes are issued as usual. | |
eadf3260 | 2683 | **write_and_verify** |
2c3a9150 VF |
2684 | Issue WRITE AND VERIFY commands. The BYTCHK bit is set to 0. This |
2685 | directs the device to carry out a medium verification with no data | |
2686 | comparison. The writefua option is ignored with this selection. | |
eadf3260 VF |
2687 | **verify** |
2688 | This option is deprecated. Use write_and_verify instead. | |
2689 | **write_same** | |
2c3a9150 VF |
2690 | Issue WRITE SAME commands. This transfers a single block to the device |
2691 | and writes this same block of data to a contiguous sequence of LBAs | |
2692 | beginning at the specified offset. fio's block size parameter specifies | |
2693 | the amount of data written with each command. However, the amount of data | |
2694 | actually transferred to the device is equal to the device's block | |
2695 | (sector) size. For a device with 512 byte sectors, blocksize=8k will | |
2696 | write 16 sectors with each command. fio will still generate 8k of data | |
2697 | for each command but only the first 512 bytes will be used and | |
2698 | transferred to the device. The writefua option is ignored with this | |
2699 | selection. | |
eadf3260 VF |
2700 | **same** |
2701 | This option is deprecated. Use write_same instead. | |
91e13ff5 VF |
2702 | **write_same_ndob** |
2703 | Issue WRITE SAME(16) commands as above but with the No Data Output | |
2704 | Buffer (NDOB) bit set. No data will be transferred to the device with | |
2705 | this bit set. Data written will be a pre-determined pattern such as | |
2706 | all zeroes. | |
71efbed6 VF |
2707 | **write_stream** |
2708 | Issue WRITE STREAM(16) commands. Use the **stream_id** option to specify | |
2709 | the stream identifier. | |
e8ab121c VF |
2710 | **verify_bytchk_00** |
2711 | Issue VERIFY commands with BYTCHK set to 00. This directs the | |
2712 | device to carry out a medium verification with no data comparison. | |
2713 | **verify_bytchk_01** | |
2714 | Issue VERIFY commands with BYTCHK set to 01. This directs the device to | |
2715 | compare the data on the device with the data transferred to the device. | |
2716 | **verify_bytchk_11** | |
2717 | Issue VERIFY commands with BYTCHK set to 11. This transfers a | |
2718 | single block to the device and compares the contents of this block with the | |
2719 | data on the device beginning at the specified offset. fio's block size | |
2720 | parameter specifies the total amount of data compared with this command. | |
2721 | However, only one block (sector) worth of data is transferred to the device. | |
2722 | This is similar to the WRITE SAME command except that data is compared instead | |
2723 | of written. | |
52b81b7c | 2724 | |
71efbed6 VF |
2725 | .. option:: stream_id=int : [sg] |
2726 | ||
2727 | Set the stream identifier for WRITE STREAM commands. If this is set to 0 (which is not | |
2728 | a valid stream identifier) fio will open a stream and then close it when done. Default | |
2729 | is 0. | |
2730 | ||
c2f6a13d LMB |
2731 | .. option:: http_host=str : [http] |
2732 | ||
2733 | Hostname to connect to. For S3, this could be the bucket hostname. | |
2734 | Default is **localhost** | |
2735 | ||
2736 | .. option:: http_user=str : [http] | |
2737 | ||
2738 | Username for HTTP authentication. | |
2739 | ||
2740 | .. option:: http_pass=str : [http] | |
2741 | ||
2742 | Password for HTTP authentication. | |
2743 | ||
09fd2966 | 2744 | .. option:: https=str : [http] |
c2f6a13d | 2745 | |
09fd2966 LMB |
2746 | Enable HTTPS instead of http. *on* enables HTTPS; *insecure* |
2747 | will enable HTTPS, but disable SSL peer verification (use with | |
2748 | caution!). Default is **off** | |
c2f6a13d | 2749 | |
09fd2966 | 2750 | .. option:: http_mode=str : [http] |
c2f6a13d | 2751 | |
09fd2966 LMB |
2752 | Which HTTP access mode to use: *webdav*, *swift*, or *s3*. |
2753 | Default is **webdav** | |
c2f6a13d LMB |
2754 | |
2755 | .. option:: http_s3_region=str : [http] | |
2756 | ||
2757 | The S3 region/zone string. | |
2758 | Default is **us-east-1** | |
2759 | ||
2760 | .. option:: http_s3_key=str : [http] | |
2761 | ||
2762 | The S3 secret key. | |
2763 | ||
2764 | .. option:: http_s3_keyid=str : [http] | |
2765 | ||
2766 | The S3 key/access id. | |
2767 | ||
a2084df0 FH |
2768 | .. option:: http_s3_sse_customer_key=str : [http] |
2769 | ||
2770 | The encryption customer key in SSE server side. | |
2771 | ||
2772 | .. option:: http_s3_sse_customer_algorithm=str : [http] | |
2773 | ||
2774 | The encryption customer algorithm in SSE server side. | |
2775 | Default is **AES256** | |
2776 | ||
2777 | .. option:: http_s3_storage_class=str : [http] | |
2778 | ||
2779 | Which storage class to access. User-customizable settings. | |
2780 | Default is **STANDARD** | |
2781 | ||
09fd2966 LMB |
2782 | .. option:: http_swift_auth_token=str : [http] |
2783 | ||
2784 | The Swift auth token. See the example configuration file on how | |
2785 | to retrieve this. | |
2786 | ||
c2f6a13d LMB |
2787 | .. option:: http_verbose=int : [http] |
2788 | ||
2789 | Enable verbose requests from libcurl. Useful for debugging. 1 | |
2790 | turns on verbose logging from libcurl, 2 additionally enables | |
2791 | HTTP IO tracing. Default is **0** | |
2792 | ||
f2d6de5d RJ |
2793 | .. option:: uri=str : [nbd] |
2794 | ||
2795 | Specify the NBD URI of the server to test. The string | |
2796 | is a standard NBD URI | |
2797 | (see https://github.com/NetworkBlockDevice/nbd/tree/master/doc). | |
2798 | Example URIs: nbd://localhost:10809 | |
2799 | nbd+unix:///?socket=/tmp/socket | |
2800 | nbds://tlshost/exportname | |
2801 | ||
10756b2c BS |
2802 | .. option:: gpu_dev_ids=str : [libcufile] |
2803 | ||
2804 | Specify the GPU IDs to use with CUDA. This is a colon-separated list of | |
2805 | int. GPUs are assigned to workers roundrobin. Default is 0. | |
2806 | ||
2807 | .. option:: cuda_io=str : [libcufile] | |
2808 | ||
2809 | Specify the type of I/O to use with CUDA. Default is **cufile**. | |
2810 | ||
2811 | **cufile** | |
2812 | Use libcufile and nvidia-fs. This option performs I/O directly | |
2813 | between a GPUDirect Storage filesystem and GPU buffers, | |
2814 | avoiding use of a bounce buffer. If :option:`verify` is set, | |
2815 | cudaMemcpy is used to copy verificaton data between RAM and GPU. | |
2816 | Verification data is copied from RAM to GPU before a write | |
2817 | and from GPU to RAM after a read. :option:`direct` must be 1. | |
2818 | **posix** | |
2819 | Use POSIX to perform I/O with a RAM buffer, and use cudaMemcpy | |
2820 | to transfer data between RAM and the GPUs. Data is copied from | |
2821 | GPU to RAM before a write and copied from RAM to GPU after a | |
2822 | read. :option:`verify` does not affect use of cudaMemcpy. | |
2823 | ||
9326926b TG |
2824 | .. option:: nfs_url=str : [nfs] |
2825 | ||
2826 | URL in libnfs format, eg nfs://<server|ipv4|ipv6>/path[?arg=val[&arg=val]*] | |
2827 | Refer to the libnfs README for more details. | |
2828 | ||
b50590bc EV |
2829 | .. option:: program=str : [exec] |
2830 | ||
2831 | Specify the program to execute. | |
2832 | ||
2833 | .. option:: arguments=str : [exec] | |
2834 | ||
2835 | Specify arguments to pass to program. | |
2836 | Some special variables can be expanded to pass fio's job details to the program. | |
2837 | ||
2838 | **%r** | |
2839 | Replaced by the duration of the job in seconds. | |
2840 | **%n** | |
2841 | Replaced by the name of the job. | |
2842 | ||
2843 | .. option:: grace_time=int : [exec] | |
2844 | ||
2845 | Specify the time between the SIGTERM and SIGKILL signals. Default is 1 second. | |
2846 | ||
81c7079c | 2847 | .. option:: std_redirect=bool : [exec] |
b50590bc EV |
2848 | |
2849 | If set, stdout and stderr streams are redirected to files named from the job name. Default is true. | |
2850 | ||
454154e6 AK |
2851 | .. option:: xnvme_async=str : [xnvme] |
2852 | ||
2853 | Select the xnvme async command interface. This can take these values. | |
2854 | ||
2855 | **emu** | |
4deb92f9 AK |
2856 | This is default and use to emulate asynchronous I/O by using a |
2857 | single thread to create a queue pair on top of a synchronous | |
2858 | I/O interface using the NVMe driver IOCTL. | |
454154e6 | 2859 | **thrpool** |
4deb92f9 AK |
2860 | Emulate an asynchronous I/O interface with a pool of userspace |
2861 | threads on top of a synchronous I/O interface using the NVMe | |
2862 | driver IOCTL. By default four threads are used. | |
454154e6 | 2863 | **io_uring** |
4deb92f9 AK |
2864 | Linux native asynchronous I/O interface which supports both |
2865 | direct and buffered I/O. | |
2866 | **io_uring_cmd** | |
2867 | Fast Linux native asynchronous I/O interface for NVMe pass | |
2868 | through commands. This only works with NVMe character device | |
2869 | (/dev/ngXnY). | |
454154e6 AK |
2870 | **libaio** |
2871 | Use Linux aio for Asynchronous I/O. | |
2872 | **posix** | |
4deb92f9 AK |
2873 | Use the posix asynchronous I/O interface to perform one or |
2874 | more I/O operations asynchronously. | |
203a4c7c AK |
2875 | **vfio** |
2876 | Use the user-space VFIO-based backend, implemented using | |
2877 | libvfn instead of SPDK. | |
454154e6 | 2878 | **nil** |
4deb92f9 AK |
2879 | Do not transfer any data; just pretend to. This is mainly used |
2880 | for introspective performance evaluation. | |
454154e6 AK |
2881 | |
2882 | .. option:: xnvme_sync=str : [xnvme] | |
2883 | ||
2884 | Select the xnvme synchronous command interface. This can take these values. | |
2885 | ||
2886 | **nvme** | |
4deb92f9 AK |
2887 | This is default and uses Linux NVMe Driver ioctl() for |
2888 | synchronous I/O. | |
454154e6 | 2889 | **psync** |
4deb92f9 AK |
2890 | This supports regular as well as vectored pread() and pwrite() |
2891 | commands. | |
2892 | **block** | |
2893 | This is the same as psync except that it also supports zone | |
2894 | management commands using Linux block layer IOCTLs. | |
454154e6 AK |
2895 | |
2896 | .. option:: xnvme_admin=str : [xnvme] | |
2897 | ||
2898 | Select the xnvme admin command interface. This can take these values. | |
2899 | ||
2900 | **nvme** | |
4deb92f9 AK |
2901 | This is default and uses linux NVMe Driver ioctl() for admin |
2902 | commands. | |
454154e6 AK |
2903 | **block** |
2904 | Use Linux Block Layer ioctl() and sysfs for admin commands. | |
454154e6 AK |
2905 | |
2906 | .. option:: xnvme_dev_nsid=int : [xnvme] | |
2907 | ||
203a4c7c | 2908 | xnvme namespace identifier for userspace NVMe driver, SPDK or vfio. |
454154e6 | 2909 | |
efbafe2a AK |
2910 | .. option:: xnvme_dev_subnqn=str : [xnvme] |
2911 | ||
2912 | Sets the subsystem NQN for fabrics. This is for xNVMe to utilize a | |
2913 | fabrics target with multiple systems. | |
2914 | ||
c945074c AK |
2915 | .. option:: xnvme_mem=str : [xnvme] |
2916 | ||
2917 | Select the xnvme memory backend. This can take these values. | |
2918 | ||
2919 | **posix** | |
2920 | This is the default posix memory backend for linux NVMe driver. | |
2921 | **hugepage** | |
2922 | Use hugepages, instead of existing posix memory backend. The | |
2923 | memory backend uses hugetlbfs. This require users to allocate | |
2924 | hugepages, mount hugetlbfs and set an enviornment variable for | |
2925 | XNVME_HUGETLB_PATH. | |
2926 | **spdk** | |
2927 | Uses SPDK's memory allocator. | |
2928 | **vfio** | |
2929 | Uses libvfn's memory allocator. This also specifies the use | |
2930 | of libvfn backend instead of SPDK. | |
2931 | ||
454154e6 AK |
2932 | .. option:: xnvme_iovec=int : [xnvme] |
2933 | ||
2934 | If this option is set. xnvme will use vectored read/write commands. | |
2935 | ||
a601337a AF |
2936 | .. option:: libblkio_driver=str : [libblkio] |
2937 | ||
2938 | The libblkio *driver* to use. Different drivers access devices through | |
2939 | different underlying interfaces. Available drivers depend on the | |
2940 | libblkio version in use and are listed at | |
2941 | https://libblkio.gitlab.io/libblkio/blkio.html#drivers | |
2942 | ||
13fffdfb AF |
2943 | .. option:: libblkio_path=str : [libblkio] |
2944 | ||
2945 | Sets the value of the driver-specific "path" property before connecting | |
2946 | the libblkio instance, which identifies the target device or file on | |
2947 | which to perform I/O. Its exact semantics are driver-dependent and not | |
2948 | all drivers may support it; see | |
2949 | https://libblkio.gitlab.io/libblkio/blkio.html#drivers | |
2950 | ||
a601337a AF |
2951 | .. option:: libblkio_pre_connect_props=str : [libblkio] |
2952 | ||
13fffdfb AF |
2953 | A colon-separated list of additional libblkio properties to be set after |
2954 | creating but before connecting the libblkio instance. Each property must | |
2955 | have the format ``<name>=<value>``. Colons can be escaped as ``\:``. | |
2956 | These are set after the engine sets any other properties, so those can | |
2957 | be overriden. Available properties depend on the libblkio version in use | |
a601337a AF |
2958 | and are listed at |
2959 | https://libblkio.gitlab.io/libblkio/blkio.html#properties | |
2960 | ||
13fffdfb AF |
2961 | .. option:: libblkio_num_entries=int : [libblkio] |
2962 | ||
2963 | Sets the value of the driver-specific "num-entries" property before | |
2964 | starting the libblkio instance. Its exact semantics are driver-dependent | |
2965 | and not all drivers may support it; see | |
2966 | https://libblkio.gitlab.io/libblkio/blkio.html#drivers | |
2967 | ||
2968 | .. option:: libblkio_queue_size=int : [libblkio] | |
2969 | ||
2970 | Sets the value of the driver-specific "queue-size" property before | |
2971 | starting the libblkio instance. Its exact semantics are driver-dependent | |
2972 | and not all drivers may support it; see | |
2973 | https://libblkio.gitlab.io/libblkio/blkio.html#drivers | |
2974 | ||
a601337a AF |
2975 | .. option:: libblkio_pre_start_props=str : [libblkio] |
2976 | ||
13fffdfb AF |
2977 | A colon-separated list of additional libblkio properties to be set after |
2978 | connecting but before starting the libblkio instance. Each property must | |
2979 | have the format ``<name>=<value>``. Colons can be escaped as ``\:``. | |
2980 | These are set after the engine sets any other properties, so those can | |
2981 | be overriden. Available properties depend on the libblkio version in use | |
a601337a AF |
2982 | and are listed at |
2983 | https://libblkio.gitlab.io/libblkio/blkio.html#properties | |
2984 | ||
6dd4291c AF |
2985 | .. option:: libblkio_vectored : [libblkio] |
2986 | ||
2987 | Submit vectored read and write requests. | |
2988 | ||
464981ff AF |
2989 | .. option:: libblkio_write_zeroes_on_trim : [libblkio] |
2990 | ||
2991 | Submit trims as "write zeroes" requests instead of discard requests. | |
2992 | ||
b158577d AF |
2993 | .. option:: libblkio_wait_mode=str : [libblkio] |
2994 | ||
2995 | How to wait for completions: | |
2996 | ||
2997 | **block** (default) | |
2998 | Use a blocking call to ``blkioq_do_io()``. | |
2999 | **eventfd** | |
3000 | Use a blocking call to ``read()`` on the completion eventfd. | |
3001 | **loop** | |
3002 | Use a busy loop with a non-blocking call to ``blkioq_do_io()``. | |
3003 | ||
b1bd09b5 AF |
3004 | .. option:: libblkio_force_enable_completion_eventfd : [libblkio] |
3005 | ||
3006 | Enable the queue's completion eventfd even when unused. This may impact | |
3007 | performance. The default is to enable it only if | |
3008 | :option:`libblkio_wait_mode=eventfd <libblkio_wait_mode>`. | |
3009 | ||
f80dba8d MT |
3010 | I/O depth |
3011 | ~~~~~~~~~ | |
3012 | ||
3013 | .. option:: iodepth=int | |
3014 | ||
3015 | Number of I/O units to keep in flight against the file. Note that | |
3016 | increasing *iodepth* beyond 1 will not affect synchronous ioengines (except | |
c60ebc45 | 3017 | for small degrees when :option:`verify_async` is in use). Even async |
f80dba8d MT |
3018 | engines may impose OS restrictions causing the desired depth not to be |
3019 | achieved. This may happen on Linux when using libaio and not setting | |
9207a0cb | 3020 | :option:`direct`\=1, since buffered I/O is not async on that OS. Keep an |
f80dba8d MT |
3021 | eye on the I/O depth distribution in the fio output to verify that the |
3022 | achieved depth is as expected. Default: 1. | |
3023 | ||
3024 | .. option:: iodepth_batch_submit=int, iodepth_batch=int | |
3025 | ||
3026 | This defines how many pieces of I/O to submit at once. It defaults to 1 | |
3027 | which means that we submit each I/O as soon as it is available, but can be | |
3028 | raised to submit bigger batches of I/O at the time. If it is set to 0 the | |
3029 | :option:`iodepth` value will be used. | |
3030 | ||
3031 | .. option:: iodepth_batch_complete_min=int, iodepth_batch_complete=int | |
3032 | ||
3033 | This defines how many pieces of I/O to retrieve at once. It defaults to 1 | |
3034 | which means that we'll ask for a minimum of 1 I/O in the retrieval process | |
3035 | from the kernel. The I/O retrieval will go on until we hit the limit set by | |
3036 | :option:`iodepth_low`. If this variable is set to 0, then fio will always | |
3037 | check for completed events before queuing more I/O. This helps reduce I/O | |
3038 | latency, at the cost of more retrieval system calls. | |
3039 | ||
3040 | .. option:: iodepth_batch_complete_max=int | |
3041 | ||
3042 | This defines maximum pieces of I/O to retrieve at once. This variable should | |
9207a0cb | 3043 | be used along with :option:`iodepth_batch_complete_min`\=int variable, |
f80dba8d | 3044 | specifying the range of min and max amount of I/O which should be |
730bd7d9 | 3045 | retrieved. By default it is equal to the :option:`iodepth_batch_complete_min` |
f80dba8d MT |
3046 | value. |
3047 | ||
3048 | Example #1:: | |
3049 | ||
3050 | iodepth_batch_complete_min=1 | |
3051 | iodepth_batch_complete_max=<iodepth> | |
3052 | ||
3053 | which means that we will retrieve at least 1 I/O and up to the whole | |
3054 | submitted queue depth. If none of I/O has been completed yet, we will wait. | |
3055 | ||
3056 | Example #2:: | |
3057 | ||
3058 | iodepth_batch_complete_min=0 | |
3059 | iodepth_batch_complete_max=<iodepth> | |
3060 | ||
3061 | which means that we can retrieve up to the whole submitted queue depth, but | |
3062 | if none of I/O has been completed yet, we will NOT wait and immediately exit | |
3063 | the system call. In this example we simply do polling. | |
3064 | ||
3065 | .. option:: iodepth_low=int | |
3066 | ||
3067 | The low water mark indicating when to start filling the queue | |
3068 | again. Defaults to the same as :option:`iodepth`, meaning that fio will | |
3069 | attempt to keep the queue full at all times. If :option:`iodepth` is set to | |
c60ebc45 | 3070 | e.g. 16 and *iodepth_low* is set to 4, then after fio has filled the queue of |
f80dba8d MT |
3071 | 16 requests, it will let the depth drain down to 4 before starting to fill |
3072 | it again. | |
3073 | ||
997b5680 SW |
3074 | .. option:: serialize_overlap=bool |
3075 | ||
3076 | Serialize in-flight I/Os that might otherwise cause or suffer from data races. | |
3077 | When two or more I/Os are submitted simultaneously, there is no guarantee that | |
3078 | the I/Os will be processed or completed in the submitted order. Further, if | |
3079 | two or more of those I/Os are writes, any overlapping region between them can | |
3080 | become indeterminate/undefined on certain storage. These issues can cause | |
3081 | verification to fail erratically when at least one of the racing I/Os is | |
3082 | changing data and the overlapping region has a non-zero size. Setting | |
3083 | ``serialize_overlap`` tells fio to avoid provoking this behavior by explicitly | |
3084 | serializing in-flight I/Os that have a non-zero overlap. Note that setting | |
ee21ebee | 3085 | this option can reduce both performance and the :option:`iodepth` achieved. |
3d6a6f04 VF |
3086 | |
3087 | This option only applies to I/Os issued for a single job except when it is | |
a02ec45a | 3088 | enabled along with :option:`io_submit_mode`\=offload. In offload mode, fio |
3d6a6f04 | 3089 | will check for overlap among all I/Os submitted by offload jobs with :option:`serialize_overlap` |
307f2246 | 3090 | enabled. |
3d6a6f04 VF |
3091 | |
3092 | Default: false. | |
997b5680 | 3093 | |
f80dba8d MT |
3094 | .. option:: io_submit_mode=str |
3095 | ||
3096 | This option controls how fio submits the I/O to the I/O engine. The default | |
3097 | is `inline`, which means that the fio job threads submit and reap I/O | |
3098 | directly. If set to `offload`, the job threads will offload I/O submission | |
3099 | to a dedicated pool of I/O threads. This requires some coordination and thus | |
3100 | has a bit of extra overhead, especially for lower queue depth I/O where it | |
3101 | can increase latencies. The benefit is that fio can manage submission rates | |
3102 | independently of the device completion rates. This avoids skewed latency | |
730bd7d9 | 3103 | reporting if I/O gets backed up on the device side (the coordinated omission |
abfd235a JA |
3104 | problem). Note that this option cannot reliably be used with async IO |
3105 | engines. | |
f80dba8d MT |
3106 | |
3107 | ||
3108 | I/O rate | |
3109 | ~~~~~~~~ | |
3110 | ||
a881438b | 3111 | .. option:: thinktime=time |
f80dba8d | 3112 | |
f75ede1d SW |
3113 | Stall the job for the specified period of time after an I/O has completed before issuing the |
3114 | next. May be used to simulate processing being done by an application. | |
947e0fe0 | 3115 | When the unit is omitted, the value is interpreted in microseconds. See |
f7942acd | 3116 | :option:`thinktime_blocks`, :option:`thinktime_iotime` and :option:`thinktime_spin`. |
f80dba8d | 3117 | |
a881438b | 3118 | .. option:: thinktime_spin=time |
f80dba8d MT |
3119 | |
3120 | Only valid if :option:`thinktime` is set - pretend to spend CPU time doing | |
3121 | something with the data received, before falling back to sleeping for the | |
f75ede1d | 3122 | rest of the period specified by :option:`thinktime`. When the unit is |
947e0fe0 | 3123 | omitted, the value is interpreted in microseconds. |
f80dba8d MT |
3124 | |
3125 | .. option:: thinktime_blocks=int | |
3126 | ||
3127 | Only valid if :option:`thinktime` is set - control how many blocks to issue, | |
f50fbdda TK |
3128 | before waiting :option:`thinktime` usecs. If not set, defaults to 1 which will make |
3129 | fio wait :option:`thinktime` usecs after every block. This effectively makes any | |
f80dba8d | 3130 | queue depth setting redundant, since no more than 1 I/O will be queued |
f50fbdda | 3131 | before we have to complete it and do our :option:`thinktime`. In other words, this |
f80dba8d | 3132 | setting effectively caps the queue depth if the latter is larger. |
71bfa161 | 3133 | |
33f42c20 HQ |
3134 | .. option:: thinktime_blocks_type=str |
3135 | ||
3136 | Only valid if :option:`thinktime` is set - control how :option:`thinktime_blocks` | |
3137 | triggers. The default is `complete`, which triggers thinktime when fio completes | |
3138 | :option:`thinktime_blocks` blocks. If this is set to `issue`, then the trigger happens | |
3139 | at the issue side. | |
3140 | ||
f7942acd SK |
3141 | .. option:: thinktime_iotime=time |
3142 | ||
3143 | Only valid if :option:`thinktime` is set - control :option:`thinktime` | |
3144 | interval by time. The :option:`thinktime` stall is repeated after IOs | |
3145 | are executed for :option:`thinktime_iotime`. For example, | |
3146 | ``--thinktime_iotime=9s --thinktime=1s`` repeat 10-second cycle with IOs | |
3147 | for 9 seconds and stall for 1 second. When the unit is omitted, | |
3148 | :option:`thinktime_iotime` is interpreted as a number of seconds. If | |
3149 | this option is used together with :option:`thinktime_blocks`, the | |
3150 | :option:`thinktime` stall is repeated after :option:`thinktime_iotime` | |
3151 | or after :option:`thinktime_blocks` IOs, whichever happens first. | |
3152 | ||
f80dba8d | 3153 | .. option:: rate=int[,int][,int] |
71bfa161 | 3154 | |
f80dba8d MT |
3155 | Cap the bandwidth used by this job. The number is in bytes/sec, the normal |
3156 | suffix rules apply. Comma-separated values may be specified for reads, | |
3157 | writes, and trims as described in :option:`blocksize`. | |
71bfa161 | 3158 | |
b25b3464 SW |
3159 | For example, using `rate=1m,500k` would limit reads to 1MiB/sec and writes to |
3160 | 500KiB/sec. Capping only reads or writes can be done with `rate=,500k` or | |
3161 | `rate=500k,` where the former will only limit writes (to 500KiB/sec) and the | |
3162 | latter will only limit reads. | |
3163 | ||
f80dba8d | 3164 | .. option:: rate_min=int[,int][,int] |
71bfa161 | 3165 | |
f80dba8d MT |
3166 | Tell fio to do whatever it can to maintain at least this bandwidth. Failing |
3167 | to meet this requirement will cause the job to exit. Comma-separated values | |
3168 | may be specified for reads, writes, and trims as described in | |
3169 | :option:`blocksize`. | |
71bfa161 | 3170 | |
f80dba8d | 3171 | .. option:: rate_iops=int[,int][,int] |
71bfa161 | 3172 | |
f80dba8d MT |
3173 | Cap the bandwidth to this number of IOPS. Basically the same as |
3174 | :option:`rate`, just specified independently of bandwidth. If the job is | |
3175 | given a block size range instead of a fixed value, the smallest block size | |
3176 | is used as the metric. Comma-separated values may be specified for reads, | |
3177 | writes, and trims as described in :option:`blocksize`. | |
71bfa161 | 3178 | |
f80dba8d | 3179 | .. option:: rate_iops_min=int[,int][,int] |
71bfa161 | 3180 | |
f80dba8d MT |
3181 | If fio doesn't meet this rate of I/O, it will cause the job to exit. |
3182 | Comma-separated values may be specified for reads, writes, and trims as | |
3183 | described in :option:`blocksize`. | |
71bfa161 | 3184 | |
f80dba8d | 3185 | .. option:: rate_process=str |
66c098b8 | 3186 | |
f80dba8d MT |
3187 | This option controls how fio manages rated I/O submissions. The default is |
3188 | `linear`, which submits I/O in a linear fashion with fixed delays between | |
c60ebc45 | 3189 | I/Os that gets adjusted based on I/O completion rates. If this is set to |
f80dba8d MT |
3190 | `poisson`, fio will submit I/O based on a more real world random request |
3191 | flow, known as the Poisson process | |
3192 | (https://en.wikipedia.org/wiki/Poisson_point_process). The lambda will be | |
3193 | 10^6 / IOPS for the given workload. | |
71bfa161 | 3194 | |
1a9bf814 JA |
3195 | .. option:: rate_ignore_thinktime=bool |
3196 | ||
3197 | By default, fio will attempt to catch up to the specified rate setting, | |
3198 | if any kind of thinktime setting was used. If this option is set, then | |
3199 | fio will ignore the thinktime and continue doing IO at the specified | |
3200 | rate, instead of entering a catch-up mode after thinktime is done. | |
3201 | ||
71bfa161 | 3202 | |
f80dba8d MT |
3203 | I/O latency |
3204 | ~~~~~~~~~~~ | |
71bfa161 | 3205 | |
a881438b | 3206 | .. option:: latency_target=time |
71bfa161 | 3207 | |
f80dba8d | 3208 | If set, fio will attempt to find the max performance point that the given |
f75ede1d | 3209 | workload will run at while maintaining a latency below this target. When |
947e0fe0 | 3210 | the unit is omitted, the value is interpreted in microseconds. See |
f75ede1d | 3211 | :option:`latency_window` and :option:`latency_percentile`. |
71bfa161 | 3212 | |
a881438b | 3213 | .. option:: latency_window=time |
71bfa161 | 3214 | |
f80dba8d | 3215 | Used with :option:`latency_target` to specify the sample window that the job |
f75ede1d | 3216 | is run at varying queue depths to test the performance. When the unit is |
947e0fe0 | 3217 | omitted, the value is interpreted in microseconds. |
b4692828 | 3218 | |
f80dba8d | 3219 | .. option:: latency_percentile=float |
71bfa161 | 3220 | |
c60ebc45 | 3221 | The percentage of I/Os that must fall within the criteria specified by |
f80dba8d | 3222 | :option:`latency_target` and :option:`latency_window`. If not set, this |
c60ebc45 | 3223 | defaults to 100.0, meaning that all I/Os must be equal or below to the value |
f80dba8d | 3224 | set by :option:`latency_target`. |
71bfa161 | 3225 | |
e1bcd541 SL |
3226 | .. option:: latency_run=bool |
3227 | ||
3228 | Used with :option:`latency_target`. If false (default), fio will find | |
3229 | the highest queue depth that meets :option:`latency_target` and exit. If | |
3230 | true, fio will continue running and try to meet :option:`latency_target` | |
3231 | by adjusting queue depth. | |
3232 | ||
f7cf63bf | 3233 | .. option:: max_latency=time[,time][,time] |
71bfa161 | 3234 | |
f75ede1d | 3235 | If set, fio will exit the job with an ETIMEDOUT error if it exceeds this |
947e0fe0 | 3236 | maximum latency. When the unit is omitted, the value is interpreted in |
f7cf63bf VR |
3237 | microseconds. Comma-separated values may be specified for reads, writes, |
3238 | and trims as described in :option:`blocksize`. | |
71bfa161 | 3239 | |
f80dba8d | 3240 | .. option:: rate_cycle=int |
71bfa161 | 3241 | |
f80dba8d | 3242 | Average bandwidth for :option:`rate` and :option:`rate_min` over this number |
a47b697c | 3243 | of milliseconds. Defaults to 1000. |
71bfa161 | 3244 | |
71bfa161 | 3245 | |
f80dba8d MT |
3246 | I/O replay |
3247 | ~~~~~~~~~~ | |
71bfa161 | 3248 | |
f80dba8d | 3249 | .. option:: write_iolog=str |
c2b1e753 | 3250 | |
f80dba8d MT |
3251 | Write the issued I/O patterns to the specified file. See |
3252 | :option:`read_iolog`. Specify a separate file for each job, otherwise the | |
02a36caa VF |
3253 | iologs will be interspersed and the file may be corrupt. This file will |
3254 | be opened in append mode. | |
c2b1e753 | 3255 | |
f80dba8d | 3256 | .. option:: read_iolog=str |
71bfa161 | 3257 | |
22413915 | 3258 | Open an iolog with the specified filename and replay the I/O patterns it |
f80dba8d MT |
3259 | contains. This can be used to store a workload and replay it sometime |
3260 | later. The iolog given may also be a blktrace binary file, which allows fio | |
3261 | to replay a workload captured by :command:`blktrace`. See | |
3262 | :manpage:`blktrace(8)` for how to capture such logging data. For blktrace | |
3263 | replay, the file needs to be turned into a blkparse binary data file first | |
3264 | (``blkparse <device> -o /dev/null -d file_for_fio.bin``). | |
78439a18 JA |
3265 | You can specify a number of files by separating the names with a ':' |
3266 | character. See the :option:`filename` option for information on how to | |
3b803fe1 | 3267 | escape ':' characters within the file names. These files will |
78439a18 | 3268 | be sequentially assigned to job clones created by :option:`numjobs`. |
d19c04d1 | 3269 | '-' is a reserved name, meaning read from stdin, notably if |
3270 | :option:`filename` is set to '-' which means stdin as well, then | |
3271 | this flag can't be set to '-'. | |
71bfa161 | 3272 | |
77be374d AK |
3273 | .. option:: read_iolog_chunked=bool |
3274 | ||
3275 | Determines how iolog is read. If false(default) entire :option:`read_iolog` | |
3276 | will be read at once. If selected true, input from iolog will be read | |
3277 | gradually. Useful when iolog is very large, or it is generated. | |
3278 | ||
b9921d1a DZ |
3279 | .. option:: merge_blktrace_file=str |
3280 | ||
3281 | When specified, rather than replaying the logs passed to :option:`read_iolog`, | |
3282 | the logs go through a merge phase which aggregates them into a single | |
3283 | blktrace. The resulting file is then passed on as the :option:`read_iolog` | |
3284 | parameter. The intention here is to make the order of events consistent. | |
3285 | This limits the influence of the scheduler compared to replaying multiple | |
3286 | blktraces via concurrent jobs. | |
3287 | ||
87a48ada DZ |
3288 | .. option:: merge_blktrace_scalars=float_list |
3289 | ||
3290 | This is a percentage based option that is index paired with the list of | |
3291 | files passed to :option:`read_iolog`. When merging is performed, scale | |
3292 | the time of each event by the corresponding amount. For example, | |
3293 | ``--merge_blktrace_scalars="50:100"`` runs the first trace in halftime | |
3294 | and the second trace in realtime. This knob is separately tunable from | |
3295 | :option:`replay_time_scale` which scales the trace during runtime and | |
3296 | does not change the output of the merge unlike this option. | |
3297 | ||
55bfd8c8 DZ |
3298 | .. option:: merge_blktrace_iters=float_list |
3299 | ||
3300 | This is a whole number option that is index paired with the list of files | |
3301 | passed to :option:`read_iolog`. When merging is performed, run each trace | |
3302 | for the specified number of iterations. For example, | |
3303 | ``--merge_blktrace_iters="2:1"`` runs the first trace for two iterations | |
3304 | and the second trace for one iteration. | |
3305 | ||
589e88b7 | 3306 | .. option:: replay_no_stall=bool |
71bfa161 | 3307 | |
f80dba8d | 3308 | When replaying I/O with :option:`read_iolog` the default behavior is to |
22413915 | 3309 | attempt to respect the timestamps within the log and replay them with the |
f80dba8d MT |
3310 | appropriate delay between IOPS. By setting this variable fio will not |
3311 | respect the timestamps and attempt to replay them as fast as possible while | |
3312 | still respecting ordering. The result is the same I/O pattern to a given | |
3313 | device, but different timings. | |
71bfa161 | 3314 | |
6dd7fa77 JA |
3315 | .. option:: replay_time_scale=int |
3316 | ||
3317 | When replaying I/O with :option:`read_iolog`, fio will honor the | |
3318 | original timing in the trace. With this option, it's possible to scale | |
3319 | the time. It's a percentage option, if set to 50 it means run at 50% | |
3320 | the original IO rate in the trace. If set to 200, run at twice the | |
3321 | original IO rate. Defaults to 100. | |
3322 | ||
f80dba8d | 3323 | .. option:: replay_redirect=str |
b4692828 | 3324 | |
f80dba8d MT |
3325 | While replaying I/O patterns using :option:`read_iolog` the default behavior |
3326 | is to replay the IOPS onto the major/minor device that each IOP was recorded | |
3327 | from. This is sometimes undesirable because on a different machine those | |
3328 | major/minor numbers can map to a different device. Changing hardware on the | |
3329 | same system can also result in a different major/minor mapping. | |
730bd7d9 | 3330 | ``replay_redirect`` causes all I/Os to be replayed onto the single specified |
f80dba8d | 3331 | device regardless of the device it was recorded |
9207a0cb | 3332 | from. i.e. :option:`replay_redirect`\= :file:`/dev/sdc` would cause all I/O |
f80dba8d MT |
3333 | in the blktrace or iolog to be replayed onto :file:`/dev/sdc`. This means |
3334 | multiple devices will be replayed onto a single device, if the trace | |
3335 | contains multiple devices. If you want multiple devices to be replayed | |
3336 | concurrently to multiple redirected devices you must blkparse your trace | |
3337 | into separate traces and replay them with independent fio invocations. | |
3338 | Unfortunately this also breaks the strict time ordering between multiple | |
3339 | device accesses. | |
71bfa161 | 3340 | |
f80dba8d | 3341 | .. option:: replay_align=int |
74929ac2 | 3342 | |
350a535d DZ |
3343 | Force alignment of the byte offsets in a trace to this value. The value |
3344 | must be a power of 2. | |
3c54bc46 | 3345 | |
f80dba8d | 3346 | .. option:: replay_scale=int |
3c54bc46 | 3347 | |
350a535d DZ |
3348 | Scale byte offsets down by this factor when replaying traces. Should most |
3349 | likely use :option:`replay_align` as well. | |
3c54bc46 | 3350 | |
38f68906 JA |
3351 | .. option:: replay_skip=str |
3352 | ||
3353 | Sometimes it's useful to skip certain IO types in a replay trace. | |
3354 | This could be, for instance, eliminating the writes in the trace. | |
3355 | Or not replaying the trims/discards, if you are redirecting to | |
3356 | a device that doesn't support them. This option takes a comma | |
3357 | separated list of read, write, trim, sync. | |
3358 | ||
3c54bc46 | 3359 | |
f80dba8d MT |
3360 | Threads, processes and job synchronization |
3361 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
3c54bc46 | 3362 | |
f80dba8d | 3363 | .. option:: thread |
3c54bc46 | 3364 | |
730bd7d9 SW |
3365 | Fio defaults to creating jobs by using fork, however if this option is |
3366 | given, fio will create jobs by using POSIX Threads' function | |
3367 | :manpage:`pthread_create(3)` to create threads instead. | |
71bfa161 | 3368 | |
f80dba8d | 3369 | .. option:: wait_for=str |
74929ac2 | 3370 | |
730bd7d9 SW |
3371 | If set, the current job won't be started until all workers of the specified |
3372 | waitee job are done. | |
74929ac2 | 3373 | |
f80dba8d MT |
3374 | ``wait_for`` operates on the job name basis, so there are a few |
3375 | limitations. First, the waitee must be defined prior to the waiter job | |
3376 | (meaning no forward references). Second, if a job is being referenced as a | |
3377 | waitee, it must have a unique name (no duplicate waitees). | |
74929ac2 | 3378 | |
f80dba8d | 3379 | .. option:: nice=int |
892a6ffc | 3380 | |
f80dba8d | 3381 | Run the job with the given nice value. See man :manpage:`nice(2)`. |
892a6ffc | 3382 | |
f80dba8d MT |
3383 | On Windows, values less than -15 set the process class to "High"; -1 through |
3384 | -15 set "Above Normal"; 1 through 15 "Below Normal"; and above 15 "Idle" | |
3385 | priority class. | |
74929ac2 | 3386 | |
f80dba8d | 3387 | .. option:: prio=int |
71bfa161 | 3388 | |
f80dba8d MT |
3389 | Set the I/O priority value of this job. Linux limits us to a positive value |
3390 | between 0 and 7, with 0 being the highest. See man | |
3391 | :manpage:`ionice(1)`. Refer to an appropriate manpage for other operating | |
b2a432bf | 3392 | systems since meaning of priority may differ. For per-command priority |
12f9d54a DLM |
3393 | setting, see I/O engine specific :option:`cmdprio_percentage` and |
3394 | :option:`cmdprio` options. | |
71bfa161 | 3395 | |
f80dba8d | 3396 | .. option:: prioclass=int |
d59aa780 | 3397 | |
b2a432bf | 3398 | Set the I/O priority class. See man :manpage:`ionice(1)`. For per-command |
12f9d54a DLM |
3399 | priority setting, see I/O engine specific :option:`cmdprio_percentage` |
3400 | and :option:`cmdprio_class` options. | |
d59aa780 | 3401 | |
f80dba8d | 3402 | .. option:: cpus_allowed=str |
6d500c2e | 3403 | |
730bd7d9 | 3404 | Controls the same options as :option:`cpumask`, but accepts a textual |
b570e037 SW |
3405 | specification of the permitted CPUs instead and CPUs are indexed from 0. So |
3406 | to use CPUs 0 and 5 you would specify ``cpus_allowed=0,5``. This option also | |
3407 | allows a range of CPUs to be specified -- say you wanted a binding to CPUs | |
3408 | 0, 5, and 8 to 15, you would set ``cpus_allowed=0,5,8-15``. | |
3409 | ||
3410 | On Windows, when ``cpus_allowed`` is unset only CPUs from fio's current | |
3411 | processor group will be used and affinity settings are inherited from the | |
3412 | system. An fio build configured to target Windows 7 makes options that set | |
3413 | CPUs processor group aware and values will set both the processor group | |
3414 | and a CPU from within that group. For example, on a system where processor | |
3415 | group 0 has 40 CPUs and processor group 1 has 32 CPUs, ``cpus_allowed`` | |
3416 | values between 0 and 39 will bind CPUs from processor group 0 and | |
3417 | ``cpus_allowed`` values between 40 and 71 will bind CPUs from processor | |
3418 | group 1. When using ``cpus_allowed_policy=shared`` all CPUs specified by a | |
3419 | single ``cpus_allowed`` option must be from the same processor group. For | |
3420 | Windows fio builds not built for Windows 7, CPUs will only be selected from | |
3421 | (and be relative to) whatever processor group fio happens to be running in | |
3422 | and CPUs from other processor groups cannot be used. | |
6d500c2e | 3423 | |
f80dba8d | 3424 | .. option:: cpus_allowed_policy=str |
6d500c2e | 3425 | |
f80dba8d | 3426 | Set the policy of how fio distributes the CPUs specified by |
730bd7d9 | 3427 | :option:`cpus_allowed` or :option:`cpumask`. Two policies are supported: |
6d500c2e | 3428 | |
f80dba8d MT |
3429 | **shared** |
3430 | All jobs will share the CPU set specified. | |
3431 | **split** | |
3432 | Each job will get a unique CPU from the CPU set. | |
6d500c2e | 3433 | |
22413915 | 3434 | **shared** is the default behavior, if the option isn't specified. If |
b21fc93f | 3435 | **split** is specified, then fio will assign one cpu per job. If not |
f80dba8d MT |
3436 | enough CPUs are given for the jobs listed, then fio will roundrobin the CPUs |
3437 | in the set. | |
6d500c2e | 3438 | |
b570e037 SW |
3439 | .. option:: cpumask=int |
3440 | ||
3441 | Set the CPU affinity of this job. The parameter given is a bit mask of | |
3442 | allowed CPUs the job may run on. So if you want the allowed CPUs to be 1 | |
3443 | and 5, you would pass the decimal value of (1 << 1 | 1 << 5), or 34. See man | |
3444 | :manpage:`sched_setaffinity(2)`. This may not work on all supported | |
3445 | operating systems or kernel versions. This option doesn't work well for a | |
3446 | higher CPU count than what you can store in an integer mask, so it can only | |
3447 | control cpus 1-32. For boxes with larger CPU counts, use | |
3448 | :option:`cpus_allowed`. | |
3449 | ||
f80dba8d | 3450 | .. option:: numa_cpu_nodes=str |
6d500c2e | 3451 | |
f80dba8d MT |
3452 | Set this job running on specified NUMA nodes' CPUs. The arguments allow |
3453 | comma delimited list of cpu numbers, A-B ranges, or `all`. Note, to enable | |
ac8ca2af | 3454 | NUMA options support, fio must be built on a system with libnuma-dev(el) |
f80dba8d | 3455 | installed. |
61b9861d | 3456 | |
f80dba8d | 3457 | .. option:: numa_mem_policy=str |
61b9861d | 3458 | |
f80dba8d MT |
3459 | Set this job's memory policy and corresponding NUMA nodes. Format of the |
3460 | arguments:: | |
5c94b008 | 3461 | |
f80dba8d | 3462 | <mode>[:<nodelist>] |
ce35b1ec | 3463 | |
804c0839 | 3464 | ``mode`` is one of the following memory policies: ``default``, ``prefer``, |
730bd7d9 SW |
3465 | ``bind``, ``interleave`` or ``local``. For ``default`` and ``local`` memory |
3466 | policies, no node needs to be specified. For ``prefer``, only one node is | |
3467 | allowed. For ``bind`` and ``interleave`` the ``nodelist`` may be as | |
3468 | follows: a comma delimited list of numbers, A-B ranges, or `all`. | |
71bfa161 | 3469 | |
f80dba8d | 3470 | .. option:: cgroup=str |
390b1537 | 3471 | |
f80dba8d MT |
3472 | Add job to this control group. If it doesn't exist, it will be created. The |
3473 | system must have a mounted cgroup blkio mount point for this to work. If | |
3474 | your system doesn't have it mounted, you can do so with:: | |
5af1c6f3 | 3475 | |
f80dba8d | 3476 | # mount -t cgroup -o blkio none /cgroup |
5af1c6f3 | 3477 | |
f80dba8d | 3478 | .. option:: cgroup_weight=int |
5af1c6f3 | 3479 | |
f80dba8d MT |
3480 | Set the weight of the cgroup to this value. See the documentation that comes |
3481 | with the kernel, allowed values are in the range of 100..1000. | |
a086c257 | 3482 | |
f80dba8d | 3483 | .. option:: cgroup_nodelete=bool |
8c07860d | 3484 | |
f80dba8d MT |
3485 | Normally fio will delete the cgroups it has created after the job |
3486 | completion. To override this behavior and to leave cgroups around after the | |
3487 | job completion, set ``cgroup_nodelete=1``. This can be useful if one wants | |
3488 | to inspect various cgroup files after job completion. Default: false. | |
8c07860d | 3489 | |
f80dba8d | 3490 | .. option:: flow_id=int |
8c07860d | 3491 | |
f80dba8d MT |
3492 | The ID of the flow. If not specified, it defaults to being a global |
3493 | flow. See :option:`flow`. | |
1907dbc6 | 3494 | |
f80dba8d | 3495 | .. option:: flow=int |
71bfa161 | 3496 | |
73f168ea VF |
3497 | Weight in token-based flow control. If this value is used, then fio |
3498 | regulates the activity between two or more jobs sharing the same | |
3499 | flow_id. Fio attempts to keep each job activity proportional to other | |
3500 | jobs' activities in the same flow_id group, with respect to requested | |
3501 | weight per job. That is, if one job has `flow=3', another job has | |
3502 | `flow=2' and another with `flow=1`, then there will be a roughly 3:2:1 | |
3503 | ratio in how much one runs vs the others. | |
71bfa161 | 3504 | |
f80dba8d | 3505 | .. option:: flow_sleep=int |
82407585 | 3506 | |
d4e74fda DB |
3507 | The period of time, in microseconds, to wait after the flow counter |
3508 | has exceeded its proportion before retrying operations. | |
82407585 | 3509 | |
f80dba8d | 3510 | .. option:: stonewall, wait_for_previous |
82407585 | 3511 | |
f80dba8d MT |
3512 | Wait for preceding jobs in the job file to exit, before starting this |
3513 | one. Can be used to insert serialization points in the job file. A stone | |
3514 | wall also implies starting a new reporting group, see | |
3515 | :option:`group_reporting`. | |
3516 | ||
3517 | .. option:: exitall | |
3518 | ||
64402a8a HW |
3519 | By default, fio will continue running all other jobs when one job finishes. |
3520 | Sometimes this is not the desired action. Setting ``exitall`` will instead | |
3521 | make fio terminate all jobs in the same group, as soon as one job of that | |
3522 | group finishes. | |
3523 | ||
7fc3a553 | 3524 | .. option:: exit_what=str |
64402a8a HW |
3525 | |
3526 | By default, fio will continue running all other jobs when one job finishes. | |
7fc3a553 | 3527 | Sometimes this is not the desired action. Setting ``exitall`` will |
64402a8a HW |
3528 | instead make fio terminate all jobs in the same group. The option |
3529 | ``exit_what`` allows to control which jobs get terminated when ``exitall`` is | |
3530 | enabled. The default is ``group`` and does not change the behaviour of | |
3531 | ``exitall``. The setting ``all`` terminates all jobs. The setting ``stonewall`` | |
3532 | terminates all currently running jobs across all groups and continues execution | |
3533 | with the next stonewalled group. | |
f80dba8d MT |
3534 | |
3535 | .. option:: exec_prerun=str | |
3536 | ||
3537 | Before running this job, issue the command specified through | |
3538 | :manpage:`system(3)`. Output is redirected in a file called | |
3539 | :file:`jobname.prerun.txt`. | |
3540 | ||
3541 | .. option:: exec_postrun=str | |
3542 | ||
3543 | After the job completes, issue the command specified though | |
3544 | :manpage:`system(3)`. Output is redirected in a file called | |
3545 | :file:`jobname.postrun.txt`. | |
3546 | ||
3547 | .. option:: uid=int | |
3548 | ||
3549 | Instead of running as the invoking user, set the user ID to this value | |
3550 | before the thread/process does any work. | |
3551 | ||
3552 | .. option:: gid=int | |
3553 | ||
3554 | Set group ID, see :option:`uid`. | |
3555 | ||
3556 | ||
3557 | Verification | |
3558 | ~~~~~~~~~~~~ | |
3559 | ||
3560 | .. option:: verify_only | |
3561 | ||
3562 | Do not perform specified workload, only verify data still matches previous | |
3563 | invocation of this workload. This option allows one to check data multiple | |
3564 | times at a later date without overwriting it. This option makes sense only | |
3565 | for workloads that write data, and does not support workloads with the | |
3566 | :option:`time_based` option set. | |
3567 | ||
3568 | .. option:: do_verify=bool | |
3569 | ||
3570 | Run the verify phase after a write phase. Only valid if :option:`verify` is | |
3571 | set. Default: true. | |
3572 | ||
3573 | .. option:: verify=str | |
3574 | ||
3575 | If writing to a file, fio can verify the file contents after each iteration | |
3576 | of the job. Each verification method also implies verification of special | |
3577 | header, which is written to the beginning of each block. This header also | |
3578 | includes meta information, like offset of the block, block number, timestamp | |
3579 | when block was written, etc. :option:`verify` can be combined with | |
3580 | :option:`verify_pattern` option. The allowed values are: | |
3581 | ||
3582 | **md5** | |
3583 | Use an md5 sum of the data area and store it in the header of | |
3584 | each block. | |
3585 | ||
3586 | **crc64** | |
3587 | Use an experimental crc64 sum of the data area and store it in the | |
3588 | header of each block. | |
3589 | ||
3590 | **crc32c** | |
a5896300 SW |
3591 | Use a crc32c sum of the data area and store it in the header of |
3592 | each block. This will automatically use hardware acceleration | |
3593 | (e.g. SSE4.2 on an x86 or CRC crypto extensions on ARM64) but will | |
3594 | fall back to software crc32c if none is found. Generally the | |
804c0839 | 3595 | fastest checksum fio supports when hardware accelerated. |
f80dba8d MT |
3596 | |
3597 | **crc32c-intel** | |
a5896300 | 3598 | Synonym for crc32c. |
f80dba8d MT |
3599 | |
3600 | **crc32** | |
3601 | Use a crc32 sum of the data area and store it in the header of each | |
3602 | block. | |
3603 | ||
3604 | **crc16** | |
3605 | Use a crc16 sum of the data area and store it in the header of each | |
3606 | block. | |
3607 | ||
3608 | **crc7** | |
3609 | Use a crc7 sum of the data area and store it in the header of each | |
3610 | block. | |
3611 | ||
3612 | **xxhash** | |
3613 | Use xxhash as the checksum function. Generally the fastest software | |
3614 | checksum that fio supports. | |
3615 | ||
3616 | **sha512** | |
3617 | Use sha512 as the checksum function. | |
3618 | ||
3619 | **sha256** | |
3620 | Use sha256 as the checksum function. | |
3621 | ||
3622 | **sha1** | |
3623 | Use optimized sha1 as the checksum function. | |
82407585 | 3624 | |
ae3a5acc JA |
3625 | **sha3-224** |
3626 | Use optimized sha3-224 as the checksum function. | |
3627 | ||
3628 | **sha3-256** | |
3629 | Use optimized sha3-256 as the checksum function. | |
3630 | ||
3631 | **sha3-384** | |
3632 | Use optimized sha3-384 as the checksum function. | |
3633 | ||
3634 | **sha3-512** | |
3635 | Use optimized sha3-512 as the checksum function. | |
3636 | ||
f80dba8d MT |
3637 | **meta** |
3638 | This option is deprecated, since now meta information is included in | |
3639 | generic verification header and meta verification happens by | |
3640 | default. For detailed information see the description of the | |
3641 | :option:`verify` setting. This option is kept because of | |
3642 | compatibility's sake with old configurations. Do not use it. | |
3643 | ||
3644 | **pattern** | |
3645 | Verify a strict pattern. Normally fio includes a header with some | |
3646 | basic information and checksumming, but if this option is set, only | |
3647 | the specific pattern set with :option:`verify_pattern` is verified. | |
3648 | ||
3649 | **null** | |
3650 | Only pretend to verify. Useful for testing internals with | |
9207a0cb | 3651 | :option:`ioengine`\=null, not for much else. |
f80dba8d MT |
3652 | |
3653 | This option can be used for repeated burn-in tests of a system to make sure | |
3654 | that the written data is also correctly read back. If the data direction | |
3655 | given is a read or random read, fio will assume that it should verify a | |
3656 | previously written file. If the data direction includes any form of write, | |
3657 | the verify will be of the newly written data. | |
3658 | ||
47e6a6e5 SW |
3659 | To avoid false verification errors, do not use the norandommap option when |
3660 | verifying data with async I/O engines and I/O depths > 1. Or use the | |
3661 | norandommap and the lfsr random generator together to avoid writing to the | |
fc002f14 | 3662 | same offset with multiple outstanding I/Os. |
47e6a6e5 | 3663 | |
f80dba8d MT |
3664 | .. option:: verify_offset=int |
3665 | ||
3666 | Swap the verification header with data somewhere else in the block before | |
3667 | writing. It is swapped back before verifying. | |
3668 | ||
3669 | .. option:: verify_interval=int | |
3670 | ||
3671 | Write the verification header at a finer granularity than the | |
3672 | :option:`blocksize`. It will be written for chunks the size of | |
3673 | ``verify_interval``. :option:`blocksize` should divide this evenly. | |
3674 | ||
3675 | .. option:: verify_pattern=str | |
3676 | ||
3677 | If set, fio will fill the I/O buffers with this pattern. Fio defaults to | |
3678 | filling with totally random bytes, but sometimes it's interesting to fill | |
3679 | with a known pattern for I/O verification purposes. Depending on the width | |
730bd7d9 | 3680 | of the pattern, fio will fill 1/2/3/4 bytes of the buffer at the time (it can |
f80dba8d MT |
3681 | be either a decimal or a hex number). The ``verify_pattern`` if larger than |
3682 | a 32-bit quantity has to be a hex number that starts with either "0x" or | |
3683 | "0X". Use with :option:`verify`. Also, ``verify_pattern`` supports %o | |
3684 | format, which means that for each block offset will be written and then | |
3685 | verified back, e.g.:: | |
61b9861d RP |
3686 | |
3687 | verify_pattern=%o | |
3688 | ||
f80dba8d MT |
3689 | Or use combination of everything:: |
3690 | ||
61b9861d | 3691 | verify_pattern=0xff%o"abcd"-12 |
e28218f3 | 3692 | |
f80dba8d MT |
3693 | .. option:: verify_fatal=bool |
3694 | ||
3695 | Normally fio will keep checking the entire contents before quitting on a | |
3696 | block verification failure. If this option is set, fio will exit the job on | |
3697 | the first observed failure. Default: false. | |
3698 | ||
3699 | .. option:: verify_dump=bool | |
3700 | ||
3701 | If set, dump the contents of both the original data block and the data block | |
3702 | we read off disk to files. This allows later analysis to inspect just what | |
3703 | kind of data corruption occurred. Off by default. | |
3704 | ||
3705 | .. option:: verify_async=int | |
3706 | ||
3707 | Fio will normally verify I/O inline from the submitting thread. This option | |
3708 | takes an integer describing how many async offload threads to create for I/O | |
3709 | verification instead, causing fio to offload the duty of verifying I/O | |
3710 | contents to one or more separate threads. If using this offload option, even | |
3711 | sync I/O engines can benefit from using an :option:`iodepth` setting higher | |
3712 | than 1, as it allows them to have I/O in flight while verifies are running. | |
d7e6ea1c | 3713 | Defaults to 0 async threads, i.e. verification is not asynchronous. |
f80dba8d MT |
3714 | |
3715 | .. option:: verify_async_cpus=str | |
3716 | ||
3717 | Tell fio to set the given CPU affinity on the async I/O verification | |
3718 | threads. See :option:`cpus_allowed` for the format used. | |
3719 | ||
3720 | .. option:: verify_backlog=int | |
3721 | ||
3722 | Fio will normally verify the written contents of a job that utilizes verify | |
3723 | once that job has completed. In other words, everything is written then | |
3724 | everything is read back and verified. You may want to verify continually | |
3725 | instead for a variety of reasons. Fio stores the meta data associated with | |
3726 | an I/O block in memory, so for large verify workloads, quite a bit of memory | |
3727 | would be used up holding this meta data. If this option is enabled, fio will | |
3728 | write only N blocks before verifying these blocks. | |
3729 | ||
3730 | .. option:: verify_backlog_batch=int | |
3731 | ||
3732 | Control how many blocks fio will verify if :option:`verify_backlog` is | |
3733 | set. If not set, will default to the value of :option:`verify_backlog` | |
3734 | (meaning the entire queue is read back and verified). If | |
3735 | ``verify_backlog_batch`` is less than :option:`verify_backlog` then not all | |
3736 | blocks will be verified, if ``verify_backlog_batch`` is larger than | |
3737 | :option:`verify_backlog`, some blocks will be verified more than once. | |
3738 | ||
3739 | .. option:: verify_state_save=bool | |
3740 | ||
3741 | When a job exits during the write phase of a verify workload, save its | |
3742 | current state. This allows fio to replay up until that point, if the verify | |
3743 | state is loaded for the verify read phase. The format of the filename is, | |
3744 | roughly:: | |
3745 | ||
f50fbdda | 3746 | <type>-<jobname>-<jobindex>-verify.state. |
f80dba8d MT |
3747 | |
3748 | <type> is "local" for a local run, "sock" for a client/server socket | |
3749 | connection, and "ip" (192.168.0.1, for instance) for a networked | |
d7e6ea1c | 3750 | client/server connection. Defaults to true. |
f80dba8d MT |
3751 | |
3752 | .. option:: verify_state_load=bool | |
3753 | ||
3754 | If a verify termination trigger was used, fio stores the current write state | |
3755 | of each thread. This can be used at verification time so that fio knows how | |
3756 | far it should verify. Without this information, fio will run a full | |
a47b697c SW |
3757 | verification pass, according to the settings in the job file used. Default |
3758 | false. | |
f80dba8d MT |
3759 | |
3760 | .. option:: trim_percentage=int | |
3761 | ||
3762 | Number of verify blocks to discard/trim. | |
3763 | ||
3764 | .. option:: trim_verify_zero=bool | |
3765 | ||
22413915 | 3766 | Verify that trim/discarded blocks are returned as zeros. |
f80dba8d MT |
3767 | |
3768 | .. option:: trim_backlog=int | |
3769 | ||
5cfd1e9a | 3770 | Trim after this number of blocks are written. |
f80dba8d MT |
3771 | |
3772 | .. option:: trim_backlog_batch=int | |
3773 | ||
3774 | Trim this number of I/O blocks. | |
3775 | ||
3776 | .. option:: experimental_verify=bool | |
3777 | ||
967c5441 VF |
3778 | Enable experimental verification. Standard verify records I/O metadata |
3779 | for later use during the verification phase. Experimental verify | |
3780 | instead resets the file after the write phase and then replays I/Os for | |
3781 | the verification phase. | |
f80dba8d | 3782 | |
f80dba8d MT |
3783 | Steady state |
3784 | ~~~~~~~~~~~~ | |
3785 | ||
3786 | .. option:: steadystate=str:float, ss=str:float | |
3787 | ||
3788 | Define the criterion and limit for assessing steady state performance. The | |
3789 | first parameter designates the criterion whereas the second parameter sets | |
3790 | the threshold. When the criterion falls below the threshold for the | |
3791 | specified duration, the job will stop. For example, `iops_slope:0.1%` will | |
3792 | direct fio to terminate the job when the least squares regression slope | |
3793 | falls below 0.1% of the mean IOPS. If :option:`group_reporting` is enabled | |
3794 | this will apply to all jobs in the group. Below is the list of available | |
3795 | steady state assessment criteria. All assessments are carried out using only | |
3796 | data from the rolling collection window. Threshold limits can be expressed | |
3797 | as a fixed value or as a percentage of the mean in the collection window. | |
3798 | ||
1cb049d9 VF |
3799 | When using this feature, most jobs should include the :option:`time_based` |
3800 | and :option:`runtime` options or the :option:`loops` option so that fio does not | |
3801 | stop running after it has covered the full size of the specified file(s) or device(s). | |
3802 | ||
f80dba8d MT |
3803 | **iops** |
3804 | Collect IOPS data. Stop the job if all individual IOPS measurements | |
3805 | are within the specified limit of the mean IOPS (e.g., ``iops:2`` | |
3806 | means that all individual IOPS values must be within 2 of the mean, | |
3807 | whereas ``iops:0.2%`` means that all individual IOPS values must be | |
3808 | within 0.2% of the mean IOPS to terminate the job). | |
3809 | ||
3810 | **iops_slope** | |
3811 | Collect IOPS data and calculate the least squares regression | |
3812 | slope. Stop the job if the slope falls below the specified limit. | |
3813 | ||
3814 | **bw** | |
3815 | Collect bandwidth data. Stop the job if all individual bandwidth | |
3816 | measurements are within the specified limit of the mean bandwidth. | |
3817 | ||
3818 | **bw_slope** | |
3819 | Collect bandwidth data and calculate the least squares regression | |
3820 | slope. Stop the job if the slope falls below the specified limit. | |
3821 | ||
3822 | .. option:: steadystate_duration=time, ss_dur=time | |
3823 | ||
3824 | A rolling window of this duration will be used to judge whether steady state | |
3825 | has been reached. Data will be collected once per second. The default is 0 | |
f75ede1d | 3826 | which disables steady state detection. When the unit is omitted, the |
947e0fe0 | 3827 | value is interpreted in seconds. |
f80dba8d MT |
3828 | |
3829 | .. option:: steadystate_ramp_time=time, ss_ramp=time | |
3830 | ||
3831 | Allow the job to run for the specified duration before beginning data | |
3832 | collection for checking the steady state job termination criterion. The | |
947e0fe0 | 3833 | default is 0. When the unit is omitted, the value is interpreted in seconds. |
f80dba8d MT |
3834 | |
3835 | ||
3836 | Measurements and reporting | |
3837 | ~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
3838 | ||
3839 | .. option:: per_job_logs=bool | |
3840 | ||
3841 | If set, this generates bw/clat/iops log with per file private filenames. If | |
3842 | not set, jobs with identical names will share the log filename. Default: | |
3843 | true. | |
3844 | ||
3845 | .. option:: group_reporting | |
3846 | ||
3847 | It may sometimes be interesting to display statistics for groups of jobs as | |
3848 | a whole instead of for each individual job. This is especially true if | |
3849 | :option:`numjobs` is used; looking at individual thread/process output | |
3850 | quickly becomes unwieldy. To see the final report per-group instead of | |
3851 | per-job, use :option:`group_reporting`. Jobs in a file will be part of the | |
3852 | same reporting group, unless if separated by a :option:`stonewall`, or by | |
3853 | using :option:`new_group`. | |
3854 | ||
3855 | .. option:: new_group | |
3856 | ||
3857 | Start a new reporting group. See: :option:`group_reporting`. If not given, | |
3858 | all jobs in a file will be part of the same reporting group, unless | |
3859 | separated by a :option:`stonewall`. | |
3860 | ||
589e88b7 | 3861 | .. option:: stats=bool |
8243be59 JA |
3862 | |
3863 | By default, fio collects and shows final output results for all jobs | |
3864 | that run. If this option is set to 0, then fio will ignore it in | |
3865 | the final stat output. | |
3866 | ||
f80dba8d MT |
3867 | .. option:: write_bw_log=str |
3868 | ||
3869 | If given, write a bandwidth log for this job. Can be used to store data of | |
074f0817 | 3870 | the bandwidth of the jobs in their lifetime. |
f80dba8d | 3871 | |
074f0817 SW |
3872 | If no str argument is given, the default filename of |
3873 | :file:`jobname_type.x.log` is used. Even when the argument is given, fio | |
3874 | will still append the type of log. So if one specifies:: | |
3875 | ||
3876 | write_bw_log=foo | |
f80dba8d | 3877 | |
074f0817 SW |
3878 | The actual log name will be :file:`foo_bw.x.log` where `x` is the index |
3879 | of the job (`1..N`, where `N` is the number of jobs). If | |
3880 | :option:`per_job_logs` is false, then the filename will not include the | |
3881 | `.x` job index. | |
e3cedca7 | 3882 | |
074f0817 SW |
3883 | The included :command:`fio_generate_plots` script uses :command:`gnuplot` to turn these |
3884 | text files into nice graphs. See `Log File Formats`_ for how data is | |
3885 | structured within the file. | |
3886 | ||
3887 | .. option:: write_lat_log=str | |
e3cedca7 | 3888 | |
074f0817 | 3889 | Same as :option:`write_bw_log`, except this option creates I/O |
77b7e675 SW |
3890 | submission (e.g., :file:`name_slat.x.log`), completion (e.g., |
3891 | :file:`name_clat.x.log`), and total (e.g., :file:`name_lat.x.log`) | |
074f0817 SW |
3892 | latency files instead. See :option:`write_bw_log` for details about |
3893 | the filename format and `Log File Formats`_ for how data is structured | |
3894 | within the files. | |
be4ecfdf | 3895 | |
f80dba8d | 3896 | .. option:: write_hist_log=str |
06842027 | 3897 | |
074f0817 | 3898 | Same as :option:`write_bw_log` but writes an I/O completion latency |
77b7e675 | 3899 | histogram file (e.g., :file:`name_hist.x.log`) instead. Note that this |
074f0817 SW |
3900 | file will be empty unless :option:`log_hist_msec` has also been set. |
3901 | See :option:`write_bw_log` for details about the filename format and | |
3902 | `Log File Formats`_ for how data is structured within the file. | |
06842027 | 3903 | |
f80dba8d | 3904 | .. option:: write_iops_log=str |
06842027 | 3905 | |
074f0817 | 3906 | Same as :option:`write_bw_log`, but writes an IOPS file (e.g. |
15417073 SW |
3907 | :file:`name_iops.x.log`) instead. Because fio defaults to individual |
3908 | I/O logging, the value entry in the IOPS log will be 1 unless windowed | |
3909 | logging (see :option:`log_avg_msec`) has been enabled. See | |
3910 | :option:`write_bw_log` for details about the filename format and `Log | |
3911 | File Formats`_ for how data is structured within the file. | |
06842027 | 3912 | |
0a852a50 DLM |
3913 | .. option:: log_entries=int |
3914 | ||
3915 | By default, fio will log an entry in the iops, latency, or bw log for | |
3916 | every I/O that completes. The initial number of I/O log entries is 1024. | |
3917 | When the log entries are all used, new log entries are dynamically | |
3918 | allocated. This dynamic log entry allocation may negatively impact | |
3919 | time-related statistics such as I/O tail latencies (e.g. 99.9th percentile | |
3920 | completion latency). This option allows specifying a larger initial | |
3921 | number of log entries to avoid run-time allocations of new log entries, | |
3922 | resulting in more precise time-related I/O statistics. | |
3923 | Also see :option:`log_avg_msec`. Defaults to 1024. | |
3924 | ||
f80dba8d | 3925 | .. option:: log_avg_msec=int |
06842027 | 3926 | |
f80dba8d MT |
3927 | By default, fio will log an entry in the iops, latency, or bw log for every |
3928 | I/O that completes. When writing to the disk log, that can quickly grow to a | |
3929 | very large size. Setting this option makes fio average the each log entry | |
3930 | over the specified period of time, reducing the resolution of the log. See | |
3931 | :option:`log_max_value` as well. Defaults to 0, logging all entries. | |
6fc82095 | 3932 | Also see `Log File Formats`_. |
06842027 | 3933 | |
f80dba8d | 3934 | .. option:: log_hist_msec=int |
06842027 | 3935 | |
f80dba8d MT |
3936 | Same as :option:`log_avg_msec`, but logs entries for completion latency |
3937 | histograms. Computing latency percentiles from averages of intervals using | |
c60ebc45 | 3938 | :option:`log_avg_msec` is inaccurate. Setting this option makes fio log |
f80dba8d MT |
3939 | histogram entries over the specified period of time, reducing log sizes for |
3940 | high IOPS devices while retaining percentile accuracy. See | |
074f0817 SW |
3941 | :option:`log_hist_coarseness` and :option:`write_hist_log` as well. |
3942 | Defaults to 0, meaning histogram logging is disabled. | |
06842027 | 3943 | |
f80dba8d | 3944 | .. option:: log_hist_coarseness=int |
06842027 | 3945 | |
f80dba8d MT |
3946 | Integer ranging from 0 to 6, defining the coarseness of the resolution of |
3947 | the histogram logs enabled with :option:`log_hist_msec`. For each increment | |
3948 | in coarseness, fio outputs half as many bins. Defaults to 0, for which | |
074f0817 SW |
3949 | histogram logs contain 1216 latency bins. See :option:`write_hist_log` |
3950 | and `Log File Formats`_. | |
8b28bd41 | 3951 | |
f80dba8d | 3952 | .. option:: log_max_value=bool |
66c098b8 | 3953 | |
f80dba8d MT |
3954 | If :option:`log_avg_msec` is set, fio logs the average over that window. If |
3955 | you instead want to log the maximum value, set this option to 1. Defaults to | |
3956 | 0, meaning that averaged values are logged. | |
a696fa2a | 3957 | |
589e88b7 | 3958 | .. option:: log_offset=bool |
a696fa2a | 3959 | |
f80dba8d | 3960 | If this is set, the iolog options will include the byte offset for the I/O |
5a83478f SW |
3961 | entry as well as the other data values. Defaults to 0 meaning that |
3962 | offsets are not present in logs. Also see `Log File Formats`_. | |
71bfa161 | 3963 | |
f80dba8d | 3964 | .. option:: log_compression=int |
7de87099 | 3965 | |
f80dba8d MT |
3966 | If this is set, fio will compress the I/O logs as it goes, to keep the |
3967 | memory footprint lower. When a log reaches the specified size, that chunk is | |
3968 | removed and compressed in the background. Given that I/O logs are fairly | |
3969 | highly compressible, this yields a nice memory savings for longer runs. The | |
3970 | downside is that the compression will consume some background CPU cycles, so | |
3971 | it may impact the run. This, however, is also true if the logging ends up | |
3972 | consuming most of the system memory. So pick your poison. The I/O logs are | |
3973 | saved normally at the end of a run, by decompressing the chunks and storing | |
3974 | them in the specified log file. This feature depends on the availability of | |
3975 | zlib. | |
e0b0d892 | 3976 | |
f80dba8d | 3977 | .. option:: log_compression_cpus=str |
e0b0d892 | 3978 | |
f80dba8d MT |
3979 | Define the set of CPUs that are allowed to handle online log compression for |
3980 | the I/O jobs. This can provide better isolation between performance | |
0cf90a62 SW |
3981 | sensitive jobs, and background compression work. See |
3982 | :option:`cpus_allowed` for the format used. | |
9e684a49 | 3983 | |
f80dba8d | 3984 | .. option:: log_store_compressed=bool |
9e684a49 | 3985 | |
f80dba8d MT |
3986 | If set, fio will store the log files in a compressed format. They can be |
3987 | decompressed with fio, using the :option:`--inflate-log` command line | |
3988 | parameter. The files will be stored with a :file:`.fz` suffix. | |
9e684a49 | 3989 | |
f80dba8d | 3990 | .. option:: log_unix_epoch=bool |
9e684a49 | 3991 | |
f80dba8d MT |
3992 | If set, fio will log Unix timestamps to the log files produced by enabling |
3993 | write_type_log for each log type, instead of the default zero-based | |
3994 | timestamps. | |
3995 | ||
d5b3cfd4 | 3996 | .. option:: log_alternate_epoch=bool |
3997 | ||
3998 | If set, fio will log timestamps based on the epoch used by the clock specified | |
3999 | in the log_alternate_epoch_clock_id option, to the log files produced by | |
4000 | enabling write_type_log for each log type, instead of the default zero-based | |
4001 | timestamps. | |
4002 | ||
4003 | .. option:: log_alternate_epoch_clock_id=int | |
4004 | ||
4005 | Specifies the clock_id to be used by clock_gettime to obtain the alternate epoch | |
4006 | if either log_unix_epoch or log_alternate_epoch are true. Otherwise has no | |
4007 | effect. Default value is 0, or CLOCK_REALTIME. | |
4008 | ||
f80dba8d MT |
4009 | .. option:: block_error_percentiles=bool |
4010 | ||
4011 | If set, record errors in trim block-sized units from writes and trims and | |
4012 | output a histogram of how many trims it took to get to errors, and what kind | |
4013 | of error was encountered. | |
4014 | ||
4015 | .. option:: bwavgtime=int | |
4016 | ||
4017 | Average the calculated bandwidth over the given time. Value is specified in | |
4018 | milliseconds. If the job also does bandwidth logging through | |
4019 | :option:`write_bw_log`, then the minimum of this option and | |
4020 | :option:`log_avg_msec` will be used. Default: 500ms. | |
4021 | ||
4022 | .. option:: iopsavgtime=int | |
4023 | ||
4024 | Average the calculated IOPS over the given time. Value is specified in | |
4025 | milliseconds. If the job also does IOPS logging through | |
4026 | :option:`write_iops_log`, then the minimum of this option and | |
4027 | :option:`log_avg_msec` will be used. Default: 500ms. | |
4028 | ||
4029 | .. option:: disk_util=bool | |
4030 | ||
4031 | Generate disk utilization statistics, if the platform supports it. | |
4032 | Default: true. | |
4033 | ||
4034 | .. option:: disable_lat=bool | |
4035 | ||
4036 | Disable measurements of total latency numbers. Useful only for cutting back | |
4037 | the number of calls to :manpage:`gettimeofday(2)`, as that does impact | |
4038 | performance at really high IOPS rates. Note that to really get rid of a | |
4039 | large amount of these calls, this option must be used with | |
f75ede1d | 4040 | :option:`disable_slat` and :option:`disable_bw_measurement` as well. |
f80dba8d MT |
4041 | |
4042 | .. option:: disable_clat=bool | |
4043 | ||
4044 | Disable measurements of completion latency numbers. See | |
4045 | :option:`disable_lat`. | |
4046 | ||
4047 | .. option:: disable_slat=bool | |
4048 | ||
4049 | Disable measurements of submission latency numbers. See | |
f50fbdda | 4050 | :option:`disable_lat`. |
f80dba8d | 4051 | |
f75ede1d | 4052 | .. option:: disable_bw_measurement=bool, disable_bw=bool |
f80dba8d MT |
4053 | |
4054 | Disable measurements of throughput/bandwidth numbers. See | |
4055 | :option:`disable_lat`. | |
4056 | ||
dd39b9ce VF |
4057 | .. option:: slat_percentiles=bool |
4058 | ||
4059 | Report submission latency percentiles. Submission latency is not recorded | |
4060 | for synchronous ioengines. | |
4061 | ||
f80dba8d MT |
4062 | .. option:: clat_percentiles=bool |
4063 | ||
dd39b9ce | 4064 | Report completion latency percentiles. |
b599759b JA |
4065 | |
4066 | .. option:: lat_percentiles=bool | |
4067 | ||
dd39b9ce VF |
4068 | Report total latency percentiles. Total latency is the sum of submission |
4069 | latency and completion latency. | |
f80dba8d MT |
4070 | |
4071 | .. option:: percentile_list=float_list | |
4072 | ||
dd39b9ce VF |
4073 | Overwrite the default list of percentiles for latencies and the block error |
4074 | histogram. Each number is a floating point number in the range (0,100], and | |
4075 | the maximum length of the list is 20. Use ``:`` to separate the numbers. For | |
c32ba107 | 4076 | example, ``--percentile_list=99.5:99.9`` will cause fio to report the |
dd39b9ce VF |
4077 | latency durations below which 99.5% and 99.9% of the observed latencies fell, |
4078 | respectively. | |
f80dba8d | 4079 | |
e883cb35 JF |
4080 | .. option:: significant_figures=int |
4081 | ||
c32ba107 JA |
4082 | If using :option:`--output-format` of `normal`, set the significant |
4083 | figures to this value. Higher values will yield more precise IOPS and | |
4084 | throughput units, while lower values will round. Requires a minimum | |
4085 | value of 1 and a maximum value of 10. Defaults to 4. | |
e883cb35 | 4086 | |
f80dba8d MT |
4087 | |
4088 | Error handling | |
4089 | ~~~~~~~~~~~~~~ | |
4090 | ||
4091 | .. option:: exitall_on_error | |
4092 | ||
4093 | When one job finishes in error, terminate the rest. The default is to wait | |
4094 | for each job to finish. | |
4095 | ||
4096 | .. option:: continue_on_error=str | |
4097 | ||
4098 | Normally fio will exit the job on the first observed failure. If this option | |
4099 | is set, fio will continue the job when there is a 'non-fatal error' (EIO or | |
4100 | EILSEQ) until the runtime is exceeded or the I/O size specified is | |
4101 | completed. If this option is used, there are two more stats that are | |
4102 | appended, the total error count and the first error. The error field given | |
4103 | in the stats is the first error that was hit during the run. | |
4104 | ||
dc305989 KK |
4105 | Note: a write error from the device may go unnoticed by fio when using |
4106 | buffered IO, as the write() (or similar) system call merely dirties the | |
4107 | kernel pages, unless :option:`sync` or :option:`direct` is used. Device IO | |
4108 | errors occur when the dirty data is actually written out to disk. If fully | |
4109 | sync writes aren't desirable, :option:`fsync` or :option:`fdatasync` can be | |
4110 | used as well. This is specific to writes, as reads are always synchronous. | |
4111 | ||
f80dba8d MT |
4112 | The allowed values are: |
4113 | ||
4114 | **none** | |
4115 | Exit on any I/O or verify errors. | |
4116 | ||
4117 | **read** | |
4118 | Continue on read errors, exit on all others. | |
4119 | ||
4120 | **write** | |
4121 | Continue on write errors, exit on all others. | |
4122 | ||
4123 | **io** | |
4124 | Continue on any I/O error, exit on all others. | |
4125 | ||
4126 | **verify** | |
4127 | Continue on verify errors, exit on all others. | |
4128 | ||
4129 | **all** | |
4130 | Continue on all errors. | |
4131 | ||
4132 | **0** | |
4133 | Backward-compatible alias for 'none'. | |
4134 | ||
4135 | **1** | |
4136 | Backward-compatible alias for 'all'. | |
4137 | ||
4138 | .. option:: ignore_error=str | |
4139 | ||
4140 | Sometimes you want to ignore some errors during test in that case you can | |
a35ef7cb TK |
4141 | specify error list for each error type, instead of only being able to |
4142 | ignore the default 'non-fatal error' using :option:`continue_on_error`. | |
f80dba8d MT |
4143 | ``ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST`` errors for |
4144 | given error type is separated with ':'. Error may be symbol ('ENOSPC', | |
4145 | 'ENOMEM') or integer. Example:: | |
4146 | ||
4147 | ignore_error=EAGAIN,ENOSPC:122 | |
4148 | ||
4149 | This option will ignore EAGAIN from READ, and ENOSPC and 122(EDQUOT) from | |
a35ef7cb TK |
4150 | WRITE. This option works by overriding :option:`continue_on_error` with |
4151 | the list of errors for each error type if any. | |
f80dba8d MT |
4152 | |
4153 | .. option:: error_dump=bool | |
4154 | ||
4155 | If set dump every error even if it is non fatal, true by default. If | |
4156 | disabled only fatal error will be dumped. | |
4157 | ||
f75ede1d SW |
4158 | Running predefined workloads |
4159 | ---------------------------- | |
4160 | ||
4161 | Fio includes predefined profiles that mimic the I/O workloads generated by | |
4162 | other tools. | |
4163 | ||
4164 | .. option:: profile=str | |
4165 | ||
4166 | The predefined workload to run. Current profiles are: | |
4167 | ||
4168 | **tiobench** | |
4169 | Threaded I/O bench (tiotest/tiobench) like workload. | |
4170 | ||
4171 | **act** | |
4172 | Aerospike Certification Tool (ACT) like workload. | |
4173 | ||
4174 | To view a profile's additional options use :option:`--cmdhelp` after specifying | |
4175 | the profile. For example:: | |
4176 | ||
f50fbdda | 4177 | $ fio --profile=act --cmdhelp |
f75ede1d SW |
4178 | |
4179 | Act profile options | |
4180 | ~~~~~~~~~~~~~~~~~~~ | |
4181 | ||
4182 | .. option:: device-names=str | |
4183 | :noindex: | |
4184 | ||
4185 | Devices to use. | |
4186 | ||
4187 | .. option:: load=int | |
4188 | :noindex: | |
4189 | ||
4190 | ACT load multiplier. Default: 1. | |
4191 | ||
4192 | .. option:: test-duration=time | |
4193 | :noindex: | |
4194 | ||
947e0fe0 SW |
4195 | How long the entire test takes to run. When the unit is omitted, the value |
4196 | is given in seconds. Default: 24h. | |
f75ede1d SW |
4197 | |
4198 | .. option:: threads-per-queue=int | |
4199 | :noindex: | |
4200 | ||
f50fbdda | 4201 | Number of read I/O threads per device. Default: 8. |
f75ede1d SW |
4202 | |
4203 | .. option:: read-req-num-512-blocks=int | |
4204 | :noindex: | |
4205 | ||
4206 | Number of 512B blocks to read at the time. Default: 3. | |
4207 | ||
4208 | .. option:: large-block-op-kbytes=int | |
4209 | :noindex: | |
4210 | ||
4211 | Size of large block ops in KiB (writes). Default: 131072. | |
4212 | ||
4213 | .. option:: prep | |
4214 | :noindex: | |
4215 | ||
4216 | Set to run ACT prep phase. | |
4217 | ||
4218 | Tiobench profile options | |
4219 | ~~~~~~~~~~~~~~~~~~~~~~~~ | |
4220 | ||
4221 | .. option:: size=str | |
4222 | :noindex: | |
4223 | ||
f50fbdda | 4224 | Size in MiB. |
f75ede1d SW |
4225 | |
4226 | .. option:: block=int | |
4227 | :noindex: | |
4228 | ||
4229 | Block size in bytes. Default: 4096. | |
4230 | ||
4231 | .. option:: numruns=int | |
4232 | :noindex: | |
4233 | ||
4234 | Number of runs. | |
4235 | ||
4236 | .. option:: dir=str | |
4237 | :noindex: | |
4238 | ||
4239 | Test directory. | |
4240 | ||
4241 | .. option:: threads=int | |
4242 | :noindex: | |
4243 | ||
4244 | Number of threads. | |
f80dba8d MT |
4245 | |
4246 | Interpreting the output | |
4247 | ----------------------- | |
4248 | ||
36214730 SW |
4249 | .. |
4250 | Example output was based on the following: | |
4251 | TZ=UTC fio --iodepth=8 --ioengine=null --size=100M --time_based \ | |
4252 | --rate=1256k --bs=14K --name=quick --runtime=1s --name=mixed \ | |
4253 | --runtime=2m --rw=rw | |
4254 | ||
f80dba8d MT |
4255 | Fio spits out a lot of output. While running, fio will display the status of the |
4256 | jobs created. An example of that would be:: | |
4257 | ||
9d25d068 | 4258 | 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] |
f80dba8d | 4259 | |
36214730 SW |
4260 | The characters inside the first set of square brackets denote the current status of |
4261 | each thread. The first character is the first job defined in the job file, and so | |
4262 | forth. The possible values (in typical life cycle order) are: | |
f80dba8d MT |
4263 | |
4264 | +------+-----+-----------------------------------------------------------+ | |
4265 | | Idle | Run | | | |
4266 | +======+=====+===========================================================+ | |
4267 | | P | | Thread setup, but not started. | | |
4268 | +------+-----+-----------------------------------------------------------+ | |
4269 | | C | | Thread created. | | |
4270 | +------+-----+-----------------------------------------------------------+ | |
4271 | | I | | Thread initialized, waiting or generating necessary data. | | |
4272 | +------+-----+-----------------------------------------------------------+ | |
4273 | | | p | Thread running pre-reading file(s). | | |
4274 | +------+-----+-----------------------------------------------------------+ | |
36214730 SW |
4275 | | | / | Thread is in ramp period. | |
4276 | +------+-----+-----------------------------------------------------------+ | |
f80dba8d MT |
4277 | | | R | Running, doing sequential reads. | |
4278 | +------+-----+-----------------------------------------------------------+ | |
4279 | | | r | Running, doing random reads. | | |
4280 | +------+-----+-----------------------------------------------------------+ | |
4281 | | | W | Running, doing sequential writes. | | |
4282 | +------+-----+-----------------------------------------------------------+ | |
4283 | | | w | Running, doing random writes. | | |
4284 | +------+-----+-----------------------------------------------------------+ | |
4285 | | | M | Running, doing mixed sequential reads/writes. | | |
4286 | +------+-----+-----------------------------------------------------------+ | |
4287 | | | m | Running, doing mixed random reads/writes. | | |
4288 | +------+-----+-----------------------------------------------------------+ | |
36214730 SW |
4289 | | | D | Running, doing sequential trims. | |
4290 | +------+-----+-----------------------------------------------------------+ | |
4291 | | | d | Running, doing random trims. | | |
4292 | +------+-----+-----------------------------------------------------------+ | |
4293 | | | F | Running, currently waiting for :manpage:`fsync(2)`. | | |
f80dba8d MT |
4294 | +------+-----+-----------------------------------------------------------+ |
4295 | | | V | Running, doing verification of written data. | | |
4296 | +------+-----+-----------------------------------------------------------+ | |
36214730 SW |
4297 | | f | | Thread finishing. | |
4298 | +------+-----+-----------------------------------------------------------+ | |
f80dba8d MT |
4299 | | E | | Thread exited, not reaped by main thread yet. | |
4300 | +------+-----+-----------------------------------------------------------+ | |
36214730 | 4301 | | _ | | Thread reaped. | |
f80dba8d MT |
4302 | +------+-----+-----------------------------------------------------------+ |
4303 | | X | | Thread reaped, exited with an error. | | |
4304 | +------+-----+-----------------------------------------------------------+ | |
4305 | | K | | Thread reaped, exited due to signal. | | |
4306 | +------+-----+-----------------------------------------------------------+ | |
4307 | ||
36214730 SW |
4308 | .. |
4309 | Example output was based on the following: | |
4310 | TZ=UTC fio --iodepth=8 --ioengine=null --size=100M --runtime=58m \ | |
4311 | --time_based --rate=2512k --bs=256K --numjobs=10 \ | |
4312 | --name=readers --rw=read --name=writers --rw=write | |
4313 | ||
f80dba8d | 4314 | Fio will condense the thread string as not to take up more space on the command |
36214730 | 4315 | line than needed. For instance, if you have 10 readers and 10 writers running, |
f80dba8d MT |
4316 | the output would look like this:: |
4317 | ||
9d25d068 | 4318 | 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] |
f80dba8d | 4319 | |
36214730 SW |
4320 | Note that the status string is displayed in order, so it's possible to tell which of |
4321 | the jobs are currently doing what. In the example above this means that jobs 1--10 | |
4322 | are readers and 11--20 are writers. | |
f80dba8d MT |
4323 | |
4324 | The other values are fairly self explanatory -- number of threads currently | |
36214730 SW |
4325 | running and doing I/O, the number of currently open files (f=), the estimated |
4326 | completion percentage, the rate of I/O since last check (read speed listed first, | |
f50fbdda TK |
4327 | then write speed and optionally trim speed) in terms of bandwidth and IOPS, |
4328 | and time to completion for the current running group. It's impossible to estimate | |
4329 | runtime of the following groups (if any). | |
36214730 SW |
4330 | |
4331 | .. | |
4332 | Example output was based on the following: | |
4333 | TZ=UTC fio --iodepth=16 --ioengine=posixaio --filename=/tmp/fiofile \ | |
4334 | --direct=1 --size=100M --time_based --runtime=50s --rate_iops=89 \ | |
4335 | --bs=7K --name=Client1 --rw=write | |
4336 | ||
4337 | When fio is done (or interrupted by :kbd:`Ctrl-C`), it will show the data for | |
4338 | each thread, group of threads, and disks in that order. For each overall thread (or | |
4339 | group) the output looks like:: | |
4340 | ||
4341 | Client1: (groupid=0, jobs=1): err= 0: pid=16109: Sat Jun 24 12:07:54 2017 | |
4342 | write: IOPS=88, BW=623KiB/s (638kB/s)(30.4MiB/50032msec) | |
4343 | slat (nsec): min=500, max=145500, avg=8318.00, stdev=4781.50 | |
4344 | clat (usec): min=170, max=78367, avg=4019.02, stdev=8293.31 | |
4345 | lat (usec): min=174, max=78375, avg=4027.34, stdev=8291.79 | |
4346 | clat percentiles (usec): | |
4347 | | 1.00th=[ 302], 5.00th=[ 326], 10.00th=[ 343], 20.00th=[ 363], | |
4348 | | 30.00th=[ 392], 40.00th=[ 404], 50.00th=[ 416], 60.00th=[ 445], | |
4349 | | 70.00th=[ 816], 80.00th=[ 6718], 90.00th=[12911], 95.00th=[21627], | |
4350 | | 99.00th=[43779], 99.50th=[51643], 99.90th=[68682], 99.95th=[72877], | |
4351 | | 99.99th=[78119] | |
4352 | bw ( KiB/s): min= 532, max= 686, per=0.10%, avg=622.87, stdev=24.82, samples= 100 | |
4353 | iops : min= 76, max= 98, avg=88.98, stdev= 3.54, samples= 100 | |
29092211 VF |
4354 | lat (usec) : 250=0.04%, 500=64.11%, 750=4.81%, 1000=2.79% |
4355 | lat (msec) : 2=4.16%, 4=1.84%, 10=4.90%, 20=11.33%, 50=5.37% | |
4356 | lat (msec) : 100=0.65% | |
36214730 SW |
4357 | cpu : usr=0.27%, sys=0.18%, ctx=12072, majf=0, minf=21 |
4358 | IO depths : 1=85.0%, 2=13.1%, 4=1.8%, 8=0.1%, 16=0.0%, 32=0.0%, >=64=0.0% | |
4359 | submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0% | |
4360 | complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0% | |
4361 | issued rwt: total=0,4450,0, short=0,0,0, dropped=0,0,0 | |
4362 | latency : target=0, window=0, percentile=100.00%, depth=8 | |
4363 | ||
4364 | The job name (or first job's name when using :option:`group_reporting`) is printed, | |
4365 | along with the group id, count of jobs being aggregated, last error id seen (which | |
4366 | is 0 when there are no errors), pid/tid of that thread and the time the job/group | |
4367 | completed. Below are the I/O statistics for each data direction performed (showing | |
4368 | writes in the example above). In the order listed, they denote: | |
4369 | ||
4370 | **read/write/trim** | |
4371 | The string before the colon shows the I/O direction the statistics | |
4372 | are for. **IOPS** is the average I/Os performed per second. **BW** | |
4373 | is the average bandwidth rate shown as: value in power of 2 format | |
4374 | (value in power of 10 format). The last two values show: (**total | |
4375 | I/O performed** in power of 2 format / **runtime** of that thread). | |
f80dba8d MT |
4376 | |
4377 | **slat** | |
36214730 SW |
4378 | Submission latency (**min** being the minimum, **max** being the |
4379 | maximum, **avg** being the average, **stdev** being the standard | |
13ddd98b VF |
4380 | deviation). This is the time from when fio initialized the I/O |
4381 | to submission. For synchronous ioengines this includes the time | |
4382 | up until just before the ioengine's queue function is called. | |
4383 | For asynchronous ioengines this includes the time up through the | |
4384 | completion of the ioengine's queue function (and commit function | |
4385 | if it is defined). For sync I/O this row is not displayed as the | |
4386 | slat is negligible. This value can be in nanoseconds, | |
4387 | microseconds or milliseconds --- fio will choose the most | |
4388 | appropriate base and print that (in the example above | |
4389 | nanoseconds was the best scale). Note: in :option:`--minimal` | |
4390 | mode latencies are always expressed in microseconds. | |
f80dba8d MT |
4391 | |
4392 | **clat** | |
4393 | Completion latency. Same names as slat, this denotes the time from | |
13ddd98b VF |
4394 | submission to completion of the I/O pieces. For sync I/O, this |
4395 | represents the time from when the I/O was submitted to the | |
4396 | operating system to when it was completed. For asynchronous | |
4397 | ioengines this is the time from when the ioengine's queue (and | |
4398 | commit if available) functions were completed to when the I/O's | |
4399 | completion was reaped by fio. | |
f80dba8d | 4400 | |
29092211 VF |
4401 | **lat** |
4402 | Total latency. Same names as slat and clat, this denotes the time from | |
4403 | when fio created the I/O unit to completion of the I/O operation. | |
13ddd98b | 4404 | It is the sum of submission and completion latency. |
29092211 | 4405 | |
f80dba8d | 4406 | **bw** |
36214730 SW |
4407 | Bandwidth statistics based on samples. Same names as the xlat stats, |
4408 | but also includes the number of samples taken (**samples**) and an | |
4409 | approximate percentage of total aggregate bandwidth this thread | |
4410 | received in its group (**per**). This last value is only really | |
4411 | useful if the threads in this group are on the same disk, since they | |
4412 | are then competing for disk access. | |
4413 | ||
4414 | **iops** | |
4415 | IOPS statistics based on samples. Same names as bw. | |
f80dba8d | 4416 | |
29092211 VF |
4417 | **lat (nsec/usec/msec)** |
4418 | The distribution of I/O completion latencies. This is the time from when | |
4419 | I/O leaves fio and when it gets completed. Unlike the separate | |
4420 | read/write/trim sections above, the data here and in the remaining | |
4421 | sections apply to all I/Os for the reporting group. 250=0.04% means that | |
4422 | 0.04% of the I/Os completed in under 250us. 500=64.11% means that 64.11% | |
4423 | of the I/Os required 250 to 499us for completion. | |
4424 | ||
f80dba8d MT |
4425 | **cpu** |
4426 | CPU usage. User and system time, along with the number of context | |
4427 | switches this thread went through, usage of system and user time, and | |
4428 | finally the number of major and minor page faults. The CPU utilization | |
4429 | numbers are averages for the jobs in that reporting group, while the | |
23a8e176 | 4430 | context and fault counters are summed. |
f80dba8d MT |
4431 | |
4432 | **IO depths** | |
a2140525 SW |
4433 | The distribution of I/O depths over the job lifetime. The numbers are |
4434 | divided into powers of 2 and each entry covers depths from that value | |
4435 | up to those that are lower than the next entry -- e.g., 16= covers | |
4436 | depths from 16 to 31. Note that the range covered by a depth | |
4437 | distribution entry can be different to the range covered by the | |
4438 | equivalent submit/complete distribution entry. | |
f80dba8d MT |
4439 | |
4440 | **IO submit** | |
4441 | How many pieces of I/O were submitting in a single submit call. Each | |
c60ebc45 | 4442 | entry denotes that amount and below, until the previous entry -- e.g., |
a2140525 SW |
4443 | 16=100% means that we submitted anywhere between 9 to 16 I/Os per submit |
4444 | call. Note that the range covered by a submit distribution entry can | |
4445 | be different to the range covered by the equivalent depth distribution | |
4446 | entry. | |
f80dba8d MT |
4447 | |
4448 | **IO complete** | |
4449 | Like the above submit number, but for completions instead. | |
4450 | ||
36214730 SW |
4451 | **IO issued rwt** |
4452 | The number of read/write/trim requests issued, and how many of them were | |
4453 | short or dropped. | |
f80dba8d | 4454 | |
29092211 | 4455 | **IO latency** |
ee21ebee | 4456 | These values are for :option:`latency_target` and related options. When |
29092211 VF |
4457 | these options are engaged, this section describes the I/O depth required |
4458 | to meet the specified latency target. | |
71bfa161 | 4459 | |
36214730 SW |
4460 | .. |
4461 | Example output was based on the following: | |
4462 | TZ=UTC fio --ioengine=null --iodepth=2 --size=100M --numjobs=2 \ | |
4463 | --rate_process=poisson --io_limit=32M --name=read --bs=128k \ | |
4464 | --rate=11M --name=write --rw=write --bs=2k --rate=700k | |
4465 | ||
71bfa161 | 4466 | After each client has been listed, the group statistics are printed. They |
f80dba8d | 4467 | will look like this:: |
71bfa161 | 4468 | |
f80dba8d | 4469 | Run status group 0 (all jobs): |
36214730 SW |
4470 | 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 |
4471 | WRITE: bw=1231KiB/s (1261kB/s), 616KiB/s-621KiB/s (630kB/s-636kB/s), io=64.0MiB (67.1MB), run=52747-53223msec | |
71bfa161 | 4472 | |
36214730 | 4473 | For each data direction it prints: |
71bfa161 | 4474 | |
36214730 SW |
4475 | **bw** |
4476 | Aggregate bandwidth of threads in this group followed by the | |
4477 | minimum and maximum bandwidth of all the threads in this group. | |
4478 | Values outside of brackets are power-of-2 format and those | |
4479 | within are the equivalent value in a power-of-10 format. | |
f80dba8d | 4480 | **io** |
36214730 SW |
4481 | Aggregate I/O performed of all threads in this group. The |
4482 | format is the same as bw. | |
4483 | **run** | |
4484 | The smallest and longest runtimes of the threads in this group. | |
71bfa161 | 4485 | |
f50fbdda | 4486 | And finally, the disk statistics are printed. This is Linux specific. They will look like this:: |
71bfa161 | 4487 | |
f80dba8d MT |
4488 | Disk stats (read/write): |
4489 | sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00% | |
71bfa161 JA |
4490 | |
4491 | Each value is printed for both reads and writes, with reads first. The | |
4492 | numbers denote: | |
4493 | ||
f80dba8d | 4494 | **ios** |
c60ebc45 | 4495 | Number of I/Os performed by all groups. |
f80dba8d | 4496 | **merge** |
007c7be9 | 4497 | Number of merges performed by the I/O scheduler. |
f80dba8d MT |
4498 | **ticks** |
4499 | Number of ticks we kept the disk busy. | |
36214730 | 4500 | **in_queue** |
f80dba8d MT |
4501 | Total time spent in the disk queue. |
4502 | **util** | |
4503 | The disk utilization. A value of 100% means we kept the disk | |
71bfa161 JA |
4504 | busy constantly, 50% would be a disk idling half of the time. |
4505 | ||
f80dba8d MT |
4506 | It is also possible to get fio to dump the current output while it is running, |
4507 | without terminating the job. To do that, send fio the **USR1** signal. You can | |
4508 | also get regularly timed dumps by using the :option:`--status-interval` | |
4509 | parameter, or by creating a file in :file:`/tmp` named | |
4510 | :file:`fio-dump-status`. If fio sees this file, it will unlink it and dump the | |
4511 | current output status. | |
8423bd11 | 4512 | |
71bfa161 | 4513 | |
f80dba8d MT |
4514 | Terse output |
4515 | ------------ | |
71bfa161 | 4516 | |
f80dba8d MT |
4517 | For scripted usage where you typically want to generate tables or graphs of the |
4518 | results, fio can output the results in a semicolon separated format. The format | |
4519 | is one long line of values, such as:: | |
71bfa161 | 4520 | |
f80dba8d MT |
4521 | 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% |
4522 | A description of this job goes here. | |
562c2d2f | 4523 | |
4e757af1 VF |
4524 | The job description (if provided) follows on a second line for terse v2. |
4525 | It appears on the same line for other terse versions. | |
71bfa161 | 4526 | |
a7f77fa6 SW |
4527 | To enable terse output, use the :option:`--minimal` or |
4528 | :option:`--output-format`\=terse command line options. The | |
f80dba8d MT |
4529 | first value is the version of the terse output format. If the output has to be |
4530 | changed for some reason, this number will be incremented by 1 to signify that | |
4531 | change. | |
6820cb3b | 4532 | |
a2c95580 | 4533 | Split up, the format is as follows (comments in brackets denote when a |
007c7be9 | 4534 | field was introduced or whether it's specific to some terse version): |
71bfa161 | 4535 | |
f80dba8d MT |
4536 | :: |
4537 | ||
f50fbdda | 4538 | terse version, fio version [v3], jobname, groupid, error |
f80dba8d MT |
4539 | |
4540 | READ status:: | |
4541 | ||
4542 | Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec) | |
4543 | Submission latency: min, max, mean, stdev (usec) | |
4544 | Completion latency: min, max, mean, stdev (usec) | |
4545 | Completion latency percentiles: 20 fields (see below) | |
4546 | Total latency: min, max, mean, stdev (usec) | |
f50fbdda TK |
4547 | Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev, number of samples [v5] |
4548 | IOPS [v5]: min, max, mean, stdev, number of samples | |
f80dba8d MT |
4549 | |
4550 | WRITE status: | |
4551 | ||
4552 | :: | |
4553 | ||
4554 | Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec) | |
4555 | Submission latency: min, max, mean, stdev (usec) | |
247823cc | 4556 | Completion latency: min, max, mean, stdev (usec) |
f80dba8d MT |
4557 | Completion latency percentiles: 20 fields (see below) |
4558 | Total latency: min, max, mean, stdev (usec) | |
f50fbdda TK |
4559 | Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev, number of samples [v5] |
4560 | IOPS [v5]: min, max, mean, stdev, number of samples | |
a2c95580 AH |
4561 | |
4562 | TRIM status [all but version 3]: | |
4563 | ||
f50fbdda | 4564 | Fields are similar to READ/WRITE status. |
f80dba8d MT |
4565 | |
4566 | CPU usage:: | |
4567 | ||
4568 | user, system, context switches, major faults, minor faults | |
4569 | ||
4570 | I/O depths:: | |
4571 | ||
4572 | <=1, 2, 4, 8, 16, 32, >=64 | |
4573 | ||
4574 | I/O latencies microseconds:: | |
4575 | ||
4576 | <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000 | |
4577 | ||
4578 | I/O latencies milliseconds:: | |
4579 | ||
4580 | <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000 | |
4581 | ||
a2c95580 | 4582 | Disk utilization [v3]:: |
f80dba8d | 4583 | |
f50fbdda TK |
4584 | disk name, read ios, write ios, read merges, write merges, read ticks, write ticks, |
4585 | time spent in queue, disk utilization percentage | |
f80dba8d MT |
4586 | |
4587 | Additional Info (dependent on continue_on_error, default off):: | |
4588 | ||
4589 | total # errors, first error code | |
4590 | ||
4591 | Additional Info (dependent on description being set):: | |
4592 | ||
4593 | Text description | |
4594 | ||
4595 | Completion latency percentiles can be a grouping of up to 20 sets, so for the | |
4596 | terse output fio writes all of them. Each field will look like this:: | |
1db92cb6 | 4597 | |
f50fbdda | 4598 | 1.00%=6112 |
1db92cb6 | 4599 | |
f80dba8d | 4600 | which is the Xth percentile, and the `usec` latency associated with it. |
1db92cb6 | 4601 | |
f50fbdda | 4602 | For `Disk utilization`, all disks used by fio are shown. So for each disk there |
f80dba8d | 4603 | will be a disk utilization section. |
f2f788dd | 4604 | |
2fc26c3d | 4605 | Below is a single line containing short names for each of the fields in the |
2831be97 | 4606 | minimal output v3, separated by semicolons:: |
2fc26c3d | 4607 | |
f95689d3 | 4608 | terse_version_3;fio_version;jobname;groupid;error;read_kb;read_bandwidth_kb;read_iops;read_runtime_ms;read_slat_min_us;read_slat_max_us;read_slat_mean_us;read_slat_dev_us;read_clat_min_us;read_clat_max_us;read_clat_mean_us;read_clat_dev_us;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_us;read_lat_max_us;read_lat_mean_us;read_lat_dev_us;read_bw_min_kb;read_bw_max_kb;read_bw_agg_pct;read_bw_mean_kb;read_bw_dev_kb;write_kb;write_bandwidth_kb;write_iops;write_runtime_ms;write_slat_min_us;write_slat_max_us;write_slat_mean_us;write_slat_dev_us;write_clat_min_us;write_clat_max_us;write_clat_mean_us;write_clat_dev_us;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_us;write_lat_max_us;write_lat_mean_us;write_lat_dev_us;write_bw_min_kb;write_bw_max_kb;write_bw_agg_pct;write_bw_mean_kb;write_bw_dev_kb;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 | 4609 | |
4e757af1 VF |
4610 | In client/server mode terse output differs from what appears when jobs are run |
4611 | locally. Disk utilization data is omitted from the standard terse output and | |
4612 | for v3 and later appears on its own separate line at the end of each terse | |
4613 | reporting cycle. | |
4614 | ||
25c8b9d7 | 4615 | |
44c82dba VF |
4616 | JSON output |
4617 | ------------ | |
4618 | ||
4619 | The `json` output format is intended to be both human readable and convenient | |
4620 | for automated parsing. For the most part its sections mirror those of the | |
4621 | `normal` output. The `runtime` value is reported in msec and the `bw` value is | |
4622 | reported in 1024 bytes per second units. | |
4623 | ||
4624 | ||
d29c4a91 VF |
4625 | JSON+ output |
4626 | ------------ | |
4627 | ||
4628 | The `json+` output format is identical to the `json` output format except that it | |
4629 | adds a full dump of the completion latency bins. Each `bins` object contains a | |
4630 | set of (key, value) pairs where keys are latency durations and values count how | |
4631 | many I/Os had completion latencies of the corresponding duration. For example, | |
4632 | consider: | |
4633 | ||
4634 | "bins" : { "87552" : 1, "89600" : 1, "94720" : 1, "96768" : 1, "97792" : 1, "99840" : 1, "100864" : 2, "103936" : 6, "104960" : 534, "105984" : 5995, "107008" : 7529, ... } | |
4635 | ||
4636 | This data indicates that one I/O required 87,552ns to complete, two I/Os required | |
4637 | 100,864ns to complete, and 7529 I/Os required 107,008ns to complete. | |
4638 | ||
4639 | Also included with fio is a Python script `fio_jsonplus_clat2csv` that takes | |
4640 | json+ output and generates CSV-formatted latency data suitable for plotting. | |
4641 | ||
4642 | The latency durations actually represent the midpoints of latency intervals. | |
f50fbdda | 4643 | For details refer to :file:`stat.h`. |
d29c4a91 VF |
4644 | |
4645 | ||
f80dba8d MT |
4646 | Trace file format |
4647 | ----------------- | |
4648 | ||
4649 | There are two trace file format that you can encounter. The older (v1) format is | |
4650 | unsupported since version 1.20-rc3 (March 2008). It will still be described | |
25c8b9d7 PD |
4651 | below in case that you get an old trace and want to understand it. |
4652 | ||
4653 | In any case the trace is a simple text file with a single action per line. | |
4654 | ||
4655 | ||
f80dba8d MT |
4656 | Trace file format v1 |
4657 | ~~~~~~~~~~~~~~~~~~~~ | |
4658 | ||
4659 | Each line represents a single I/O action in the following format:: | |
4660 | ||
4661 | rw, offset, length | |
25c8b9d7 | 4662 | |
f50fbdda | 4663 | where `rw=0/1` for read/write, and the `offset` and `length` entries being in bytes. |
25c8b9d7 | 4664 | |
22413915 | 4665 | This format is not supported in fio versions >= 1.20-rc3. |
25c8b9d7 | 4666 | |
25c8b9d7 | 4667 | |
f80dba8d MT |
4668 | Trace file format v2 |
4669 | ~~~~~~~~~~~~~~~~~~~~ | |
25c8b9d7 | 4670 | |
f80dba8d | 4671 | The second version of the trace file format was added in fio version 1.17. It |
12efafa3 | 4672 | allows one to access more than one file per trace and has a bigger set of possible |
f80dba8d | 4673 | file actions. |
25c8b9d7 | 4674 | |
f80dba8d | 4675 | The first line of the trace file has to be:: |
25c8b9d7 | 4676 | |
f80dba8d | 4677 | fio version 2 iolog |
25c8b9d7 PD |
4678 | |
4679 | Following this can be lines in two different formats, which are described below. | |
4680 | ||
f80dba8d | 4681 | The file management format:: |
25c8b9d7 | 4682 | |
f80dba8d | 4683 | filename action |
25c8b9d7 | 4684 | |
f50fbdda | 4685 | The `filename` is given as an absolute path. The `action` can be one of these: |
25c8b9d7 | 4686 | |
f80dba8d | 4687 | **add** |
f50fbdda | 4688 | Add the given `filename` to the trace. |
f80dba8d | 4689 | **open** |
f50fbdda | 4690 | Open the file with the given `filename`. The `filename` has to have |
f80dba8d MT |
4691 | been added with the **add** action before. |
4692 | **close** | |
f50fbdda | 4693 | Close the file with the given `filename`. The file has to have been |
f80dba8d MT |
4694 | opened before. |
4695 | ||
4696 | ||
4697 | The file I/O action format:: | |
4698 | ||
4699 | filename action offset length | |
4700 | ||
4701 | The `filename` is given as an absolute path, and has to have been added and | |
4702 | opened before it can be used with this format. The `offset` and `length` are | |
4703 | given in bytes. The `action` can be one of these: | |
4704 | ||
4705 | **wait** | |
4706 | Wait for `offset` microseconds. Everything below 100 is discarded. | |
5c2c0db4 MG |
4707 | The time is relative to the previous `wait` statement. Note that |
4708 | action `wait` is not allowed as of version 3, as the same behavior | |
4709 | can be achieved using timestamps. | |
f80dba8d MT |
4710 | **read** |
4711 | Read `length` bytes beginning from `offset`. | |
4712 | **write** | |
4713 | Write `length` bytes beginning from `offset`. | |
4714 | **sync** | |
4715 | :manpage:`fsync(2)` the file. | |
4716 | **datasync** | |
4717 | :manpage:`fdatasync(2)` the file. | |
4718 | **trim** | |
4719 | Trim the given file from the given `offset` for `length` bytes. | |
4720 | ||
b9921d1a | 4721 | |
5c2c0db4 MG |
4722 | Trace file format v3 |
4723 | ~~~~~~~~~~~~~~~~~~~~ | |
4724 | ||
4725 | The third version of the trace file format was added in fio version 3.31. It | |
4726 | forces each action to have a timestamp associated with it. | |
4727 | ||
4728 | The first line of the trace file has to be:: | |
4729 | ||
4730 | fio version 3 iolog | |
4731 | ||
4732 | Following this can be lines in two different formats, which are described below. | |
4733 | ||
4734 | The file management format:: | |
4735 | ||
4736 | timestamp filename action | |
4737 | ||
4738 | The file I/O action format:: | |
4739 | ||
4740 | timestamp filename action offset length | |
4741 | ||
4742 | The `timestamp` is relative to the beginning of the run (ie starts at 0). The | |
4743 | `filename`, `action`, `offset` and `length` are identical to version 2, except | |
4744 | that version 3 does not allow the `wait` action. | |
4745 | ||
4746 | ||
b9921d1a DZ |
4747 | I/O Replay - Merging Traces |
4748 | --------------------------- | |
4749 | ||
4750 | Colocation is a common practice used to get the most out of a machine. | |
4751 | Knowing which workloads play nicely with each other and which ones don't is | |
4752 | a much harder task. While fio can replay workloads concurrently via multiple | |
4753 | jobs, it leaves some variability up to the scheduler making results harder to | |
4754 | reproduce. Merging is a way to make the order of events consistent. | |
4755 | ||
4756 | Merging is integrated into I/O replay and done when a | |
4757 | :option:`merge_blktrace_file` is specified. The list of files passed to | |
4758 | :option:`read_iolog` go through the merge process and output a single file | |
4759 | stored to the specified file. The output file is passed on as if it were the | |
4760 | only file passed to :option:`read_iolog`. An example would look like:: | |
4761 | ||
4762 | $ fio --read_iolog="<file1>:<file2>" --merge_blktrace_file="<output_file>" | |
4763 | ||
4764 | Creating only the merged file can be done by passing the command line argument | |
d443e3af | 4765 | :option:`--merge-blktrace-only`. |
b9921d1a | 4766 | |
87a48ada DZ |
4767 | Scaling traces can be done to see the relative impact of any particular trace |
4768 | being slowed down or sped up. :option:`merge_blktrace_scalars` takes in a colon | |
4769 | separated list of percentage scalars. It is index paired with the files passed | |
4770 | to :option:`read_iolog`. | |
4771 | ||
55bfd8c8 DZ |
4772 | With scaling, it may be desirable to match the running time of all traces. |
4773 | This can be done with :option:`merge_blktrace_iters`. It is index paired with | |
4774 | :option:`read_iolog` just like :option:`merge_blktrace_scalars`. | |
4775 | ||
4776 | In an example, given two traces, A and B, each 60s long. If we want to see | |
4777 | the impact of trace A issuing IOs twice as fast and repeat trace A over the | |
4778 | runtime of trace B, the following can be done:: | |
4779 | ||
4780 | $ fio --read_iolog="<trace_a>:"<trace_b>" --merge_blktrace_file"<output_file>" --merge_blktrace_scalars="50:100" --merge_blktrace_iters="2:1" | |
4781 | ||
4782 | This runs trace A at 2x the speed twice for approximately the same runtime as | |
4783 | a single run of trace B. | |
4784 | ||
b9921d1a | 4785 | |
f80dba8d MT |
4786 | CPU idleness profiling |
4787 | ---------------------- | |
4788 | ||
4789 | In some cases, we want to understand CPU overhead in a test. For example, we | |
4790 | test patches for the specific goodness of whether they reduce CPU usage. | |
4791 | Fio implements a balloon approach to create a thread per CPU that runs at idle | |
4792 | priority, meaning that it only runs when nobody else needs the cpu. | |
4793 | By measuring the amount of work completed by the thread, idleness of each CPU | |
4794 | can be derived accordingly. | |
4795 | ||
4796 | An unit work is defined as touching a full page of unsigned characters. Mean and | |
4797 | standard deviation of time to complete an unit work is reported in "unit work" | |
4798 | section. Options can be chosen to report detailed percpu idleness or overall | |
4799 | system idleness by aggregating percpu stats. | |
4800 | ||
4801 | ||
4802 | Verification and triggers | |
4803 | ------------------------- | |
4804 | ||
4805 | Fio is usually run in one of two ways, when data verification is done. The first | |
4806 | is a normal write job of some sort with verify enabled. When the write phase has | |
4807 | completed, fio switches to reads and verifies everything it wrote. The second | |
4808 | model is running just the write phase, and then later on running the same job | |
4809 | (but with reads instead of writes) to repeat the same I/O patterns and verify | |
4810 | the contents. Both of these methods depend on the write phase being completed, | |
4811 | as fio otherwise has no idea how much data was written. | |
4812 | ||
4813 | With verification triggers, fio supports dumping the current write state to | |
4814 | local files. Then a subsequent read verify workload can load this state and know | |
4815 | exactly where to stop. This is useful for testing cases where power is cut to a | |
4816 | server in a managed fashion, for instance. | |
99b9a85a JA |
4817 | |
4818 | A verification trigger consists of two things: | |
4819 | ||
f80dba8d MT |
4820 | 1) Storing the write state of each job. |
4821 | 2) Executing a trigger command. | |
99b9a85a | 4822 | |
f80dba8d MT |
4823 | The write state is relatively small, on the order of hundreds of bytes to single |
4824 | kilobytes. It contains information on the number of completions done, the last X | |
4825 | completions, etc. | |
99b9a85a | 4826 | |
f80dba8d MT |
4827 | A trigger is invoked either through creation ('touch') of a specified file in |
4828 | the system, or through a timeout setting. If fio is run with | |
9207a0cb | 4829 | :option:`--trigger-file`\= :file:`/tmp/trigger-file`, then it will continually |
f80dba8d MT |
4830 | check for the existence of :file:`/tmp/trigger-file`. When it sees this file, it |
4831 | will fire off the trigger (thus saving state, and executing the trigger | |
99b9a85a JA |
4832 | command). |
4833 | ||
f80dba8d MT |
4834 | For client/server runs, there's both a local and remote trigger. If fio is |
4835 | running as a server backend, it will send the job states back to the client for | |
4836 | safe storage, then execute the remote trigger, if specified. If a local trigger | |
4837 | is specified, the server will still send back the write state, but the client | |
4838 | will then execute the trigger. | |
99b9a85a | 4839 | |
f80dba8d MT |
4840 | Verification trigger example |
4841 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
99b9a85a | 4842 | |
f50fbdda TK |
4843 | Let's say we want to run a powercut test on the remote Linux machine 'server'. |
4844 | Our write workload is in :file:`write-test.fio`. We want to cut power to 'server' at | |
f80dba8d MT |
4845 | some point during the run, and we'll run this test from the safety or our local |
4846 | machine, 'localbox'. On the server, we'll start the fio backend normally:: | |
99b9a85a | 4847 | |
f80dba8d | 4848 | server# fio --server |
99b9a85a | 4849 | |
f80dba8d | 4850 | and on the client, we'll fire off the workload:: |
99b9a85a | 4851 | |
f80dba8d | 4852 | localbox$ fio --client=server --trigger-file=/tmp/my-trigger --trigger-remote="bash -c \"echo b > /proc/sysrq-triger\"" |
99b9a85a | 4853 | |
f80dba8d | 4854 | We set :file:`/tmp/my-trigger` as the trigger file, and we tell fio to execute:: |
99b9a85a | 4855 | |
f80dba8d | 4856 | echo b > /proc/sysrq-trigger |
99b9a85a | 4857 | |
f80dba8d MT |
4858 | on the server once it has received the trigger and sent us the write state. This |
4859 | will work, but it's not **really** cutting power to the server, it's merely | |
4860 | abruptly rebooting it. If we have a remote way of cutting power to the server | |
4861 | through IPMI or similar, we could do that through a local trigger command | |
4502cb42 | 4862 | instead. Let's assume we have a script that does IPMI reboot of a given hostname, |
f80dba8d MT |
4863 | ipmi-reboot. On localbox, we could then have run fio with a local trigger |
4864 | instead:: | |
99b9a85a | 4865 | |
f80dba8d | 4866 | localbox$ fio --client=server --trigger-file=/tmp/my-trigger --trigger="ipmi-reboot server" |
99b9a85a | 4867 | |
f80dba8d MT |
4868 | For this case, fio would wait for the server to send us the write state, then |
4869 | execute ``ipmi-reboot server`` when that happened. | |
4870 | ||
4871 | Loading verify state | |
4872 | ~~~~~~~~~~~~~~~~~~~~ | |
4873 | ||
4502cb42 | 4874 | To load stored write state, a read verification job file must contain the |
f80dba8d | 4875 | :option:`verify_state_load` option. If that is set, fio will load the previously |
99b9a85a | 4876 | stored state. For a local fio run this is done by loading the files directly, |
f80dba8d MT |
4877 | and on a client/server run, the server backend will ask the client to send the |
4878 | files over and load them from there. | |
a3ae5b05 JA |
4879 | |
4880 | ||
f80dba8d MT |
4881 | Log File Formats |
4882 | ---------------- | |
a3ae5b05 JA |
4883 | |
4884 | Fio supports a variety of log file formats, for logging latencies, bandwidth, | |
4885 | and IOPS. The logs share a common format, which looks like this: | |
4886 | ||
5a83478f | 4887 | *time* (`msec`), *value*, *data direction*, *block size* (`bytes`), |
1a953d97 | 4888 | *offset* (`bytes`), *command priority* |
a3ae5b05 | 4889 | |
5a83478f | 4890 | *Time* for the log entry is always in milliseconds. The *value* logged depends |
a3ae5b05 JA |
4891 | on the type of log, it will be one of the following: |
4892 | ||
f80dba8d | 4893 | **Latency log** |
168bb587 | 4894 | Value is latency in nsecs |
f80dba8d MT |
4895 | **Bandwidth log** |
4896 | Value is in KiB/sec | |
4897 | **IOPS log** | |
4898 | Value is IOPS | |
4899 | ||
4900 | *Data direction* is one of the following: | |
4901 | ||
4902 | **0** | |
4903 | I/O is a READ | |
4904 | **1** | |
4905 | I/O is a WRITE | |
4906 | **2** | |
4907 | I/O is a TRIM | |
4908 | ||
15417073 SW |
4909 | The entry's *block size* is always in bytes. The *offset* is the position in bytes |
4910 | from the start of the file for that particular I/O. The logging of the offset can be | |
5a83478f | 4911 | toggled with :option:`log_offset`. |
f80dba8d | 4912 | |
1a953d97 PC |
4913 | *Command priority* is 0 for normal priority and 1 for high priority. This is controlled |
4914 | by the ioengine specific :option:`cmdprio_percentage`. | |
4915 | ||
15417073 SW |
4916 | Fio defaults to logging every individual I/O but when windowed logging is set |
4917 | through :option:`log_avg_msec`, either the average (by default) or the maximum | |
4918 | (:option:`log_max_value` is set) *value* seen over the specified period of time | |
4919 | is recorded. Each *data direction* seen within the window period will aggregate | |
4920 | its values in a separate row. Further, when using windowed logging the *block | |
4921 | size* and *offset* entries will always contain 0. | |
f80dba8d | 4922 | |
4e757af1 | 4923 | |
b8f7e412 | 4924 | Client/Server |
f80dba8d MT |
4925 | ------------- |
4926 | ||
4927 | Normally fio is invoked as a stand-alone application on the machine where the | |
6cf30ac0 SW |
4928 | I/O workload should be generated. However, the backend and frontend of fio can |
4929 | be run separately i.e., the fio server can generate an I/O workload on the "Device | |
4930 | Under Test" while being controlled by a client on another machine. | |
f80dba8d MT |
4931 | |
4932 | Start the server on the machine which has access to the storage DUT:: | |
4933 | ||
f50fbdda | 4934 | $ fio --server=args |
f80dba8d | 4935 | |
dbb257bb | 4936 | where `args` defines what fio listens to. The arguments are of the form |
f80dba8d MT |
4937 | ``type,hostname`` or ``IP,port``. *type* is either ``ip`` (or ip4) for TCP/IP |
4938 | v4, ``ip6`` for TCP/IP v6, or ``sock`` for a local unix domain socket. | |
4939 | *hostname* is either a hostname or IP address, and *port* is the port to listen | |
4940 | to (only valid for TCP/IP, not a local socket). Some examples: | |
4941 | ||
4942 | 1) ``fio --server`` | |
4943 | ||
4944 | Start a fio server, listening on all interfaces on the default port (8765). | |
4945 | ||
4946 | 2) ``fio --server=ip:hostname,4444`` | |
4947 | ||
4948 | Start a fio server, listening on IP belonging to hostname and on port 4444. | |
4949 | ||
4950 | 3) ``fio --server=ip6:::1,4444`` | |
4951 | ||
4952 | Start a fio server, listening on IPv6 localhost ::1 and on port 4444. | |
4953 | ||
4954 | 4) ``fio --server=,4444`` | |
4955 | ||
4956 | Start a fio server, listening on all interfaces on port 4444. | |
4957 | ||
4958 | 5) ``fio --server=1.2.3.4`` | |
4959 | ||
4960 | Start a fio server, listening on IP 1.2.3.4 on the default port. | |
4961 | ||
4962 | 6) ``fio --server=sock:/tmp/fio.sock`` | |
4963 | ||
dbb257bb | 4964 | Start a fio server, listening on the local socket :file:`/tmp/fio.sock`. |
f80dba8d MT |
4965 | |
4966 | Once a server is running, a "client" can connect to the fio server with:: | |
4967 | ||
4968 | fio <local-args> --client=<server> <remote-args> <job file(s)> | |
4969 | ||
4970 | where `local-args` are arguments for the client where it is running, `server` | |
4971 | is the connect string, and `remote-args` and `job file(s)` are sent to the | |
4972 | server. The `server` string follows the same format as it does on the server | |
4973 | side, to allow IP/hostname/socket and port strings. | |
4974 | ||
4975 | Fio can connect to multiple servers this way:: | |
4976 | ||
4977 | fio --client=<server1> <job file(s)> --client=<server2> <job file(s)> | |
4978 | ||
4979 | If the job file is located on the fio server, then you can tell the server to | |
4980 | load a local file as well. This is done by using :option:`--remote-config` :: | |
4981 | ||
4982 | fio --client=server --remote-config /path/to/file.fio | |
4983 | ||
4984 | Then fio will open this local (to the server) job file instead of being passed | |
4985 | one from the client. | |
4986 | ||
4987 | If you have many servers (example: 100 VMs/containers), you can input a pathname | |
4988 | of a file containing host IPs/names as the parameter value for the | |
4989 | :option:`--client` option. For example, here is an example :file:`host.list` | |
4990 | file containing 2 hostnames:: | |
4991 | ||
4992 | host1.your.dns.domain | |
4993 | host2.your.dns.domain | |
4994 | ||
4995 | The fio command would then be:: | |
a3ae5b05 | 4996 | |
f80dba8d | 4997 | fio --client=host.list <job file(s)> |
a3ae5b05 | 4998 | |
f80dba8d MT |
4999 | In this mode, you cannot input server-specific parameters or job files -- all |
5000 | servers receive the same job file. | |
a3ae5b05 | 5001 | |
f80dba8d MT |
5002 | In order to let ``fio --client`` runs use a shared filesystem from multiple |
5003 | hosts, ``fio --client`` now prepends the IP address of the server to the | |
4502cb42 | 5004 | filename. For example, if fio is using the directory :file:`/mnt/nfs/fio` and is |
f80dba8d MT |
5005 | writing filename :file:`fileio.tmp`, with a :option:`--client` `hostfile` |
5006 | containing two hostnames ``h1`` and ``h2`` with IP addresses 192.168.10.120 and | |
5007 | 192.168.10.121, then fio will create two files:: | |
a3ae5b05 | 5008 | |
f80dba8d MT |
5009 | /mnt/nfs/fio/192.168.10.120.fileio.tmp |
5010 | /mnt/nfs/fio/192.168.10.121.fileio.tmp | |
4e757af1 VF |
5011 | |
5012 | Terse output in client/server mode will differ slightly from what is produced | |
5013 | when fio is run in stand-alone mode. See the terse output section for details. |