1How fio works 2------------- 3 4The first step in getting fio to simulate a desired I/O workload, is writing a 5job file describing that specific setup. A job file may contain any number of 6threads and/or files -- the typical contents of the job file is a *global* 7section defining shared parameters, and one or more job sections describing the 8jobs involved. When run, fio parses this file and sets everything up as 9described. If we break down a job from top to bottom, it contains the following 10basic 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 48The above are the basic parameters defined for a workload, in addition there's a 49multitude of parameters that modify other aspects of how this job behaves. 50 51 52Command line options 53-------------------- 54 55.. option:: --debug=type 56 57 Enable verbose tracing of various fio actions. May be ``all`` for all types 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: 61 62 *process* 63 Dump info related to processes. 64 *file* 65 Dump info related to file actions. 66 *io* 67 Dump info related to I/O queuing. 68 *mem* 69 Dump info related to memory allocations. 70 *blktrace* 71 Dump info related to blktrace setup. 72 *verify* 73 Dump info related to I/O verification. 74 *all* 75 Enable all debug options. 76 *random* 77 Dump info related to random offset generation. 78 *parse* 79 Dump info related to option matching and parsing. 80 *diskutil* 81 Dump info related to disk utilization updates. 82 *job:x* 83 Dump info only related to job number x. 84 *mutex* 85 Dump info only related to mutex up/down ops. 86 *profile* 87 Dump info related to profile extensions. 88 *time* 89 Dump info related to internal time keeping. 90 *net* 91 Dump info related to networking connections. 92 *rate* 93 Dump info related to I/O rate switching. 94 *compress* 95 Dump info related to log compress/decompress. 96 *?* or *help* 97 Show available debug options. 98 99.. option:: --parse-only 100 101 Parse options only, don\'t start any I/O. 102 103.. option:: --output=filename 104 105 Write output to file `filename`. 106 107.. option:: --bandwidth-log 108 109 Generate aggregate bandwidth logs. 110 111.. option:: --minimal 112 113 Print statistics in a terse, semicolon-delimited format. 114 115.. option:: --append-terse 116 117 Print statistics in selected mode AND terse, semicolon-delimited format. 118 **deprecated**, use :option:`--output-format` instead to select multiple 119 formats. 120 121.. option:: --output-format=type 122 123 Set the reporting format to `normal`, `terse`, `json`, or `json+`. Multiple 124 formats can be selected, separate by a comma. `terse` is a CSV based 125 format. `json+` is like `json`, except it adds a full dump of the latency 126 buckets. 127 128.. option:: --terse-version=type 129 130 Set terse version output format (default 3, or 2 or 4). 131 132.. option:: --version 133 134 Print version info and exit. 135 136.. option:: --help 137 138 Print this page. 139 140.. option:: --cpuclock-test 141 142 Perform test and validation of internal CPU clock. 143 144.. option:: --crctest=test 145 146 Test the speed of the builtin checksumming functions. If no argument is 147 given, all of them are tested. Or a comma separated list can be passed, in 148 which case the given ones are tested. 149 150.. option:: --cmdhelp=command 151 152 Print help information for `command`. May be ``all`` for all commands. 153 154.. option:: --enghelp=[ioengine[,command]] 155 156 List all commands defined by :option:`ioengine`, or print help for `command` 157 defined by :option:`ioengine`. If no :option:`ioengine` is given, list all 158 available ioengines. 159 160.. option:: --showcmd=jobfile 161 162 Turn a job file into command line options. 163 164.. option:: --readonly 165 166 Turn on safety read-only checks, preventing writes. The ``--readonly`` 167 option is an extra safety guard to prevent users from accidentally starting 168 a write workload when that is not desired. Fio will only write if 169 `rw=write/randwrite/rw/randrw` is given. This extra safety net can be used 170 as an extra precaution as ``--readonly`` will also enable a write check in 171 the I/O engine core to prevent writes due to unknown user space bug(s). 172 173.. option:: --eta=when 174 175 When real-time ETA estimate should be printed. May be `always`, `never` or 176 `auto`. 177 178.. option:: --eta-newline=time 179 180 Force a new line for every `time` period passed. 181 182.. option:: --status-interval=time 183 184 Force full status dump every `time` period passed. 185 186.. option:: --section=name 187 188 Only run specified section in job file. Multiple sections can be specified. 189 The ``--section`` option allows one to combine related jobs into one file. 190 E.g. one job file could define light, moderate, and heavy sections. Tell 191 fio to run only the "heavy" section by giving ``--section=heavy`` 192 command line option. One can also specify the "write" operations in one 193 section and "verify" operation in another section. The ``--section`` option 194 only applies to job sections. The reserved *global* section is always 195 parsed and used. 196 197.. option:: --alloc-size=kb 198 199 Set the internal smalloc pool to this size in kb (def 1024). The 200 ``--alloc-size`` switch allows one to use a larger pool size for smalloc. 201 If running large jobs with randommap enabled, fio can run out of memory. 202 Smalloc is an internal allocator for shared structures from a fixed size 203 memory pool. The pool size defaults to 16M and can grow to 8 pools. 204 205 NOTE: While running :file:`.fio_smalloc.*` backing store files are visible 206 in :file:`/tmp`. 207 208.. option:: --warnings-fatal 209 210 All fio parser warnings are fatal, causing fio to exit with an 211 error. 212 213.. option:: --max-jobs=nr 214 215 Maximum number of threads/processes to support. 216 217.. option:: --server=args 218 219 Start a backend server, with `args` specifying what to listen to. 220 See `Client/Server`_ section. 221 222.. option:: --daemonize=pidfile 223 224 Background a fio server, writing the pid to the given `pidfile` file. 225 226.. option:: --client=hostname 227 228 Instead of running the jobs locally, send and run them on the given host or 229 set of hosts. See `Client/Server`_ section. 230 231.. option:: --remote-config=file 232 233 Tell fio server to load this local file. 234 235.. option:: --idle-prof=option 236 237 Report cpu idleness on a system or percpu basis 238 ``--idle-prof=system,percpu`` or 239 run unit work calibration only ``--idle-prof=calibrate``. 240 241.. option:: --inflate-log=log 242 243 Inflate and output compressed log. 244 245.. option:: --trigger-file=file 246 247 Execute trigger cmd when file exists. 248 249.. option:: --trigger-timeout=t 250 251 Execute trigger at this time. 252 253.. option:: --trigger=cmd 254 255 Set this command as local trigger. 256 257.. option:: --trigger-remote=cmd 258 259 Set this command as remote trigger. 260 261.. option:: --aux-path=path 262 263 Use this path for fio state generated files. 264 265Any parameters following the options will be assumed to be job files, unless 266they match a job file parameter. Multiple job files can be listed and each job 267file will be regarded as a separate group. Fio will :option:`stonewall` 268execution between each group. 269 270 271Job file format 272--------------- 273 274As previously described, fio accepts one or more job files describing what it is 275supposed to do. The job file format is the classic ini file, where the names 276enclosed in [] brackets define the job name. You are free to use any ASCII name 277you want, except *global* which has special meaning. Following the job name is 278a sequence of zero or more parameters, one per line, that define the behavior of 279the job. If the first character in a line is a ';' or a '#', the entire line is 280discarded as a comment. 281 282A *global* section sets defaults for the jobs described in that file. A job may 283override a *global* section parameter, and a job file may even have several 284*global* sections if so desired. A job is only affected by a *global* section 285residing above it. 286 287The :option:`--cmdhelp` option also lists all options. If used with an `option` 288argument, :option:`--cmdhelp` will detail the given `option`. 289 290See the `examples/` directory for inspiration on how to write job files. Note 291the copyright and license requirements currently apply to `examples/` files. 292 293So let's look at a really simple job file that defines two processes, each 294randomly reading from a 128MiB file: 295 296.. code-block:: ini 297 298 ; -- start job file -- 299 [global] 300 rw=randread 301 size=128m 302 303 [job1] 304 305 [job2] 306 307 ; -- end job file -- 308 309As you can see, the job file sections themselves are empty as all the described 310parameters are shared. As no :option:`filename` option is given, fio makes up a 311`filename` for each of the jobs as it sees fit. On the command line, this job 312would look as follows:: 313 314$ fio --name=global --rw=randread --size=128m --name=job1 --name=job2 315 316 317Let's look at an example that has a number of processes writing randomly to 318files: 319 320.. code-block:: ini 321 322 ; -- start job file -- 323 [random-writers] 324 ioengine=libaio 325 iodepth=4 326 rw=randwrite 327 bs=32k 328 direct=0 329 size=64m 330 numjobs=4 331 ; -- end job file -- 332 333Here we have no *global* section, as we only have one job defined anyway. We 334want to use async I/O here, with a depth of 4 for each file. We also increased 335the buffer size used to 32KiB and define numjobs to 4 to fork 4 identical 336jobs. The result is 4 processes each randomly writing to their own 64MiB 337file. Instead of using the above job file, you could have given the parameters 338on the command line. For this case, you would specify:: 339 340$ fio --name=random-writers --ioengine=libaio --iodepth=4 --rw=randwrite --bs=32k --direct=0 --size=64m --numjobs=4 341 342When fio is utilized as a basis of any reasonably large test suite, it might be 343desirable to share a set of standardized settings across multiple job files. 344Instead of copy/pasting such settings, any section may pull in an external 345:file:`filename.fio` file with *include filename* directive, as in the following 346example:: 347 348 ; -- start job file including.fio -- 349 [global] 350 filename=/tmp/test 351 filesize=1m 352 include glob-include.fio 353 354 [test] 355 rw=randread 356 bs=4k 357 time_based=1 358 runtime=10 359 include test-include.fio 360 ; -- end job file including.fio -- 361 362.. code-block:: ini 363 364 ; -- start job file glob-include.fio -- 365 thread=1 366 group_reporting=1 367 ; -- end job file glob-include.fio -- 368 369.. code-block:: ini 370 371 ; -- start job file test-include.fio -- 372 ioengine=libaio 373 iodepth=4 374 ; -- end job file test-include.fio -- 375 376Settings pulled into a section apply to that section only (except *global* 377section). Include directives may be nested in that any included file may contain 378further include directive(s). Include files may not contain [] sections. 379 380 381Environment variables 382~~~~~~~~~~~~~~~~~~~~~ 383 384Fio also supports environment variable expansion in job files. Any sub-string of 385the form ``${VARNAME}`` as part of an option value (in other words, on the right 386of the '='), will be expanded to the value of the environment variable called 387`VARNAME`. If no such environment variable is defined, or `VARNAME` is the 388empty string, the empty string will be substituted. 389 390As an example, let's look at a sample fio invocation and job file:: 391 392$ SIZE=64m NUMJOBS=4 fio jobfile.fio 393 394.. code-block:: ini 395 396 ; -- start job file -- 397 [random-writers] 398 rw=randwrite 399 size=${SIZE} 400 numjobs=${NUMJOBS} 401 ; -- end job file -- 402 403This will expand to the following equivalent job file at runtime: 404 405.. code-block:: ini 406 407 ; -- start job file -- 408 [random-writers] 409 rw=randwrite 410 size=64m 411 numjobs=4 412 ; -- end job file -- 413 414Fio ships with a few example job files, you can also look there for inspiration. 415 416Reserved keywords 417~~~~~~~~~~~~~~~~~ 418 419Additionally, fio has a set of reserved keywords that will be replaced 420internally with the appropriate value. Those keywords are: 421 422**$pagesize** 423 424 The architecture page size of the running system. 425 426**$mb_memory** 427 428 Megabytes of total memory in the system. 429 430**$ncpus** 431 432 Number of online available CPUs. 433 434These can be used on the command line or in the job file, and will be 435automatically substituted with the current system values when the job is 436run. Simple math is also supported on these keywords, so you can perform actions 437like:: 438 439 size=8*$mb_memory 440 441and get that properly expanded to 8 times the size of memory in the machine. 442 443 444Job file parameters 445------------------- 446 447This section describes in details each parameter associated with a job. Some 448parameters take an option of a given type, such as an integer or a 449string. Anywhere a numeric value is required, an arithmetic expression may be 450used, provided it is surrounded by parentheses. Supported operators are: 451 452 - addition (+) 453 - subtraction (-) 454 - multiplication (*) 455 - division (/) 456 - modulus (%) 457 - exponentiation (^) 458 459For time values in expressions, units are microseconds by default. This is 460different than for time values not in expressions (not enclosed in 461parentheses). The following types are used: 462 463 464Parameter types 465~~~~~~~~~~~~~~~ 466 467**str** 468 String. This is a sequence of alpha characters. 469 470**time** 471 Integer with possible time suffix. In seconds unless otherwise 472 specified, use e.g. 10m for 10 minutes. Accepts s/m/h for seconds, minutes, 473 and hours, and accepts 'ms' (or 'msec') for milliseconds, and 'us' (or 474 'usec') for microseconds. 475 476.. _int: 477 478**int** 479 Integer. A whole number value, which may contain an integer prefix 480 and an integer suffix: 481 482 [*integer prefix*] **number** [*integer suffix*] 483 484 The optional *integer prefix* specifies the number's base. The default 485 is decimal. *0x* specifies hexadecimal. 486 487 The optional *integer suffix* specifies the number's units, and includes an 488 optional unit prefix and an optional unit. For quantities of data, the 489 default unit is bytes. For quantities of time, the default unit is seconds. 490 491 With :option:`kb_base` =1000, fio follows international standards for unit 492 prefixes. To specify power-of-10 decimal values defined in the 493 International System of Units (SI): 494 495 * *Ki* -- means kilo (K) or 1000 496 * *Mi* -- means mega (M) or 1000**2 497 * *Gi* -- means giga (G) or 1000**3 498 * *Ti* -- means tera (T) or 1000**4 499 * *Pi* -- means peta (P) or 1000**5 500 501 To specify power-of-2 binary values defined in IEC 80000-13: 502 503 * *k* -- means kibi (Ki) or 1024 504 * *M* -- means mebi (Mi) or 1024**2 505 * *G* -- means gibi (Gi) or 1024**3 506 * *T* -- means tebi (Ti) or 1024**4 507 * *P* -- means pebi (Pi) or 1024**5 508 509 With :option:`kb_base` =1024 (the default), the unit prefixes are opposite 510 from those specified in the SI and IEC 80000-13 standards to provide 511 compatibility with old scripts. For example, 4k means 4096. 512 513 For quantities of data, an optional unit of 'B' may be included 514 (e.g., 'kB' is the same as 'k'). 515 516 The *integer suffix* is not case sensitive (e.g., m/mi mean mebi/mega, 517 not milli). 'b' and 'B' both mean byte, not bit. 518 519 Examples with :option:`kb_base` =1000: 520 521 * *4 KiB*: 4096, 4096b, 4096B, 4ki, 4kib, 4kiB, 4Ki, 4KiB 522 * *1 MiB*: 1048576, 1mi, 1024ki 523 * *1 MB*: 1000000, 1m, 1000k 524 * *1 TiB*: 1099511627776, 1ti, 1024gi, 1048576mi 525 * *1 TB*: 1000000000, 1t, 1000m, 1000000k 526 527 Examples with :option:`kb_base` =1024 (default): 528 529 * *4 KiB*: 4096, 4096b, 4096B, 4k, 4kb, 4kB, 4K, 4KB 530 * *1 MiB*: 1048576, 1m, 1024k 531 * *1 MB*: 1000000, 1mi, 1000ki 532 * *1 TiB*: 1099511627776, 1t, 1024g, 1048576m 533 * *1 TB*: 1000000000, 1ti, 1000mi, 1000000ki 534 535 To specify times (units are not case sensitive): 536 537 * *D* -- means days 538 * *H* -- means hours 539 * *M* -- mean minutes 540 * *s* -- or sec means seconds (default) 541 * *ms* -- or *msec* means milliseconds 542 * *us* -- or *usec* means microseconds 543 544 If the option accepts an upper and lower range, use a colon ':' or 545 minus '-' to separate such values. See :ref:`irange <irange>`. 546 If the lower value specified happens to be larger than the upper value, 547 two values are swapped. 548 549.. _bool: 550 551**bool** 552 Boolean. Usually parsed as an integer, however only defined for 553 true and false (1 and 0). 554 555.. _irange: 556 557**irange** 558 Integer range with suffix. Allows value range to be given, such as 559 1024-4096. A colon may also be used as the separator, e.g. 1k:4k. If the 560 option allows two sets of ranges, they can be specified with a ',' or '/' 561 delimiter: 1k-4k/8k-32k. Also see :ref:`int <int>`. 562 563**float_list** 564 A list of floating point numbers, separated by a ':' character. 565 566 567Units 568~~~~~ 569 570.. option:: kb_base=int 571 572 Select the interpretation of unit prefixes in input parameters. 573 574 **1000** 575 Inputs comply with IEC 80000-13 and the International 576 System of Units (SI). Use: 577 578 - power-of-2 values with IEC prefixes (e.g., KiB) 579 - power-of-10 values with SI prefixes (e.g., kB) 580 581 **1024** 582 Compatibility mode (default). To avoid breaking old scripts: 583 584 - power-of-2 values with SI prefixes 585 - power-of-10 values with IEC prefixes 586 587 See :option:`bs` for more details on input parameters. 588 589 Outputs always use correct prefixes. Most outputs include both 590 side-by-side, like:: 591 592 bw=2383.3kB/s (2327.4KiB/s) 593 594 If only one value is reported, then kb_base selects the one to use: 595 596 **1000** -- SI prefixes 597 598 **1024** -- IEC prefixes 599 600.. option:: unit_base=int 601 602 Base unit for reporting. Allowed values are: 603 604 **0** 605 Use auto-detection (default). 606 **8** 607 Byte based. 608 **1** 609 Bit based. 610 611 612With the above in mind, here follows the complete list of fio job parameters. 613 614 615Job description 616~~~~~~~~~~~~~~~ 617 618.. option:: name=str 619 620 ASCII name of the job. This may be used to override the name printed by fio 621 for this job. Otherwise the job name is used. On the command line this 622 parameter has the special purpose of also signaling the start of a new job. 623 624.. option:: description=str 625 626 Text description of the job. Doesn't do anything except dump this text 627 description when this job is run. It's not parsed. 628 629.. option:: loops=int 630 631 Run the specified number of iterations of this job. Used to repeat the same 632 workload a given number of times. Defaults to 1. 633 634.. option:: numjobs=int 635 636 Create the specified number of clones of this job. Each clone of job 637 is spawned as an independent thread or process. May be used to setup a 638 larger number of threads/processes doing the same thing. Each thread is 639 reported separately; to see statistics for all clones as a whole, use 640 :option:`group_reporting` in conjunction with :option:`new_group`. 641 See :option:`--max-jobs`. 642 643 644Time related parameters 645~~~~~~~~~~~~~~~~~~~~~~~ 646 647.. option:: runtime=time 648 649 Tell fio to terminate processing after the specified period of time. It 650 can be quite hard to determine for how long a specified job will run, so 651 this parameter is handy to cap the total runtime to a given time. When 652 the unit is omitted, the value is given in seconds. 653 654.. option:: time_based 655 656 If set, fio will run for the duration of the :option:`runtime` specified 657 even if the file(s) are completely read or written. It will simply loop over 658 the same workload as many times as the :option:`runtime` allows. 659 660.. option:: startdelay=irange(time) 661 662 Delay start of job for the specified number of seconds. Supports all time 663 suffixes to allow specification of hours, minutes, seconds and milliseconds 664 -- seconds are the default if a unit is omitted. Can be given as a range 665 which causes each thread to choose randomly out of the range. 666 667.. option:: ramp_time=time 668 669 If set, fio will run the specified workload for this amount of time before 670 logging any performance numbers. Useful for letting performance settle 671 before logging results, thus minimizing the runtime required for stable 672 results. Note that the ``ramp_time`` is considered lead in time for a job, 673 thus it will increase the total runtime if a special timeout or 674 :option:`runtime` is specified. When the unit is omitted, the value is 675 given in seconds. 676 677.. option:: clocksource=str 678 679 Use the given clocksource as the base of timing. The supported options are: 680 681 **gettimeofday** 682 :manpage:`gettimeofday(2)` 683 684 **clock_gettime** 685 :manpage:`clock_gettime(2)` 686 687 **cpu** 688 Internal CPU clock source 689 690 cpu is the preferred clocksource if it is reliable, as it is very fast (and 691 fio is heavy on time calls). Fio will automatically use this clocksource if 692 it's supported and considered reliable on the system it is running on, 693 unless another clocksource is specifically set. For x86/x86-64 CPUs, this 694 means supporting TSC Invariant. 695 696.. option:: gtod_reduce=bool 697 698 Enable all of the :manpage:`gettimeofday(2)` reducing options 699 (:option:`disable_clat`, :option:`disable_slat`, :option:`disable_bw_measurement`) plus 700 reduce precision of the timeout somewhat to really shrink the 701 :manpage:`gettimeofday(2)` call count. With this option enabled, we only do 702 about 0.4% of the :manpage:`gettimeofday(2)` calls we would have done if all 703 time keeping was enabled. 704 705.. option:: gtod_cpu=int 706 707 Sometimes it's cheaper to dedicate a single thread of execution to just 708 getting the current time. Fio (and databases, for instance) are very 709 intensive on :manpage:`gettimeofday(2)` calls. With this option, you can set 710 one CPU aside for doing nothing but logging current time to a shared memory 711 location. Then the other threads/processes that run I/O workloads need only 712 copy that segment, instead of entering the kernel with a 713 :manpage:`gettimeofday(2)` call. The CPU set aside for doing these time 714 calls will be excluded from other uses. Fio will manually clear it from the 715 CPU mask of other jobs. 716 717 718Target file/device 719~~~~~~~~~~~~~~~~~~ 720 721.. option:: directory=str 722 723 Prefix filenames with this directory. Used to place files in a different 724 location than :file:`./`. You can specify a number of directories by 725 separating the names with a ':' character. These directories will be 726 assigned equally distributed to job clones creates with :option:`numjobs` as 727 long as they are using generated filenames. If specific `filename(s)` are 728 set fio will use the first listed directory, and thereby matching the 729 `filename` semantic which generates a file each clone if not specified, but 730 let all clones use the same if set. 731 732 See the :option:`filename` option for escaping certain characters. 733 734.. option:: filename=str 735 736 Fio normally makes up a `filename` based on the job name, thread number, and 737 file number. If you want to share files between threads in a job or several 738 jobs with fixed file paths, specify a `filename` for each of them to override 739 the default. If the ioengine is file based, you can specify a number of files 740 by separating the names with a ':' colon. So if you wanted a job to open 741 :file:`/dev/sda` and :file:`/dev/sdb` as the two working files, you would use 742 ``filename=/dev/sda:/dev/sdb``. This also means that whenever this option is 743 specified, :option:`nrfiles` is ignored. The size of regular files specified 744 by this option will be :option:`size` divided by number of files unless 745 explicit size is specified by :option:`filesize`. 746 747 On Windows, disk devices are accessed as :file:`\\\\.\\PhysicalDrive0` for 748 the first device, :file:`\\\\.\\PhysicalDrive1` for the second etc. 749 Note: Windows and FreeBSD prevent write access to areas 750 of the disk containing in-use data (e.g. filesystems). If the wanted 751 `filename` does need to include a colon, then escape that with a ``\`` 752 character. For instance, if the `filename` is :file:`/dev/dsk/foo@3,0:c`, 753 then you would use ``filename="/dev/dsk/foo@3,0\:c"``. The 754 :file:`-` is a reserved name, meaning stdin or stdout. Which of the two 755 depends on the read/write direction set. 756 757.. option:: filename_format=str 758 759 If sharing multiple files between jobs, it is usually necessary to have fio 760 generate the exact names that you want. By default, fio will name a file 761 based on the default file format specification of 762 :file:`jobname.jobnumber.filenumber`. With this option, that can be 763 customized. Fio will recognize and replace the following keywords in this 764 string: 765 766 **$jobname** 767 The name of the worker thread or process. 768 **$jobnum** 769 The incremental number of the worker thread or process. 770 **$filenum** 771 The incremental number of the file for that worker thread or 772 process. 773 774 To have dependent jobs share a set of files, this option can be set to have 775 fio generate filenames that are shared between the two. For instance, if 776 :file:`testfiles.$filenum` is specified, file number 4 for any job will be 777 named :file:`testfiles.4`. The default of :file:`$jobname.$jobnum.$filenum` 778 will be used if no other format specifier is given. 779 780.. option:: unique_filename=bool 781 782 To avoid collisions between networked clients, fio defaults to prefixing any 783 generated filenames (with a directory specified) with the source of the 784 client connecting. To disable this behavior, set this option to 0. 785 786.. option:: opendir=str 787 788 Recursively open any files below directory `str`. 789 790.. option:: lockfile=str 791 792 Fio defaults to not locking any files before it does I/O to them. If a file 793 or file descriptor is shared, fio can serialize I/O to that file to make the 794 end result consistent. This is usual for emulating real workloads that share 795 files. The lock modes are: 796 797 **none** 798 No locking. The default. 799 **exclusive** 800 Only one thread or process may do I/O at a time, excluding all 801 others. 802 **readwrite** 803 Read-write locking on the file. Many readers may 804 access the file at the same time, but writes get exclusive access. 805 806.. option:: nrfiles=int 807 808 Number of files to use for this job. Defaults to 1. The size of files 809 will be :option:`size` divided by this unless explicit size is specified by 810 :option:`filesize`. Files are created for each thread separately, and each 811 file will have a file number within its name by default, as explained in 812 :option:`filename` section. 813 814 815.. option:: openfiles=int 816 817 Number of files to keep open at the same time. Defaults to the same as 818 :option:`nrfiles`, can be set smaller to limit the number simultaneous 819 opens. 820 821.. option:: file_service_type=str 822 823 Defines how fio decides which file from a job to service next. The following 824 types are defined: 825 826 **random** 827 Choose a file at random. 828 829 **roundrobin** 830 Round robin over opened files. This is the default. 831 832 **sequential** 833 Finish one file before moving on to the next. Multiple files can 834 still be open depending on 'openfiles'. 835 836 **zipf** 837 Use a *Zipf* distribution to decide what file to access. 838 839 **pareto** 840 Use a *Pareto* distribution to decide what file to access. 841 842 **gauss** 843 Use a *Gaussian* (normal) distribution to decide what file to 844 access. 845 846 For *random*, *roundrobin*, and *sequential*, a postfix can be appended to 847 tell fio how many I/Os to issue before switching to a new file. For example, 848 specifying ``file_service_type=random:8`` would cause fio to issue 849 8 I/Os before selecting a new file at random. For the non-uniform 850 distributions, a floating point postfix can be given to influence how the 851 distribution is skewed. See :option:`random_distribution` for a description 852 of how that would work. 853 854.. option:: ioscheduler=str 855 856 Attempt to switch the device hosting the file to the specified I/O scheduler 857 before running. 858 859.. option:: create_serialize=bool 860 861 If true, serialize the file creation for the jobs. This may be handy to 862 avoid interleaving of data files, which may greatly depend on the filesystem 863 used and even the number of processors in the system. 864 865.. option:: create_fsync=bool 866 867 fsync the data file after creation. This is the default. 868 869.. option:: create_on_open=bool 870 871 Don't pre-setup the files for I/O, just create open() when it's time to do 872 I/O to that file. 873 874.. option:: create_only=bool 875 876 If true, fio will only run the setup phase of the job. If files need to be 877 laid out or updated on disk, only that will be done. The actual job contents 878 are not executed. 879 880.. option:: allow_file_create=bool 881 882 If true, fio is permitted to create files as part of its workload. This is 883 the default behavior. If this option is false, then fio will error out if 884 the files it needs to use don't already exist. Default: true. 885 886.. option:: allow_mounted_write=bool 887 888 If this isn't set, fio will abort jobs that are destructive (e.g. that write) 889 to what appears to be a mounted device or partition. This should help catch 890 creating inadvertently destructive tests, not realizing that the test will 891 destroy data on the mounted file system. Note that some platforms don't allow 892 writing against a mounted device regardless of this option. Default: false. 893 894.. option:: pre_read=bool 895 896 If this is given, files will be pre-read into memory before starting the 897 given I/O operation. This will also clear the :option:`invalidate` flag, 898 since it is pointless to pre-read and then drop the cache. This will only 899 work for I/O engines that are seek-able, since they allow you to read the 900 same data multiple times. Thus it will not work on e.g. network or splice I/O. 901 902.. option:: unlink=bool 903 904 Unlink the job files when done. Not the default, as repeated runs of that 905 job would then waste time recreating the file set again and again. 906 907.. option:: unlink_each_loop=bool 908 909 Unlink job files after each iteration or loop. 910 911.. option:: zonesize=int 912 913 Divide a file into zones of the specified size. See :option:`zoneskip`. 914 915.. option:: zonerange=int 916 917 Give size of an I/O zone. See :option:`zoneskip`. 918 919.. option:: zoneskip=int 920 921 Skip the specified number of bytes when :option:`zonesize` data has been 922 read. The two zone options can be used to only do I/O on zones of a file. 923 924 925I/O type 926~~~~~~~~ 927 928.. option:: direct=bool 929 930 If value is true, use non-buffered I/O. This is usually O_DIRECT. Note that 931 ZFS on Solaris doesn't support direct I/O. On Windows the synchronous 932 ioengines don't support direct I/O. Default: false. 933 934.. option:: atomic=bool 935 936 If value is true, attempt to use atomic direct I/O. Atomic writes are 937 guaranteed to be stable once acknowledged by the operating system. Only 938 Linux supports O_ATOMIC right now. 939 940.. option:: buffered=bool 941 942 If value is true, use buffered I/O. This is the opposite of the 943 :option:`direct` option. Defaults to true. 944 945.. option:: readwrite=str, rw=str 946 947 Type of I/O pattern. Accepted values are: 948 949 **read** 950 Sequential reads. 951 **write** 952 Sequential writes. 953 **trim** 954 Sequential trims (Linux block devices only). 955 **randwrite** 956 Random writes. 957 **randread** 958 Random reads. 959 **randtrim** 960 Random trims (Linux block devices only). 961 **rw,readwrite** 962 Sequential mixed reads and writes. 963 **randrw** 964 Random mixed reads and writes. 965 **trimwrite** 966 Sequential trim+write sequences. Blocks will be trimmed first, 967 then the same blocks will be written to. 968 969 Fio defaults to read if the option is not specified. For the mixed I/O 970 types, the default is to split them 50/50. For certain types of I/O the 971 result may still be skewed a bit, since the speed may be different. It is 972 possible to specify a number of I/O's to do before getting a new offset, 973 this is done by appending a ``:<nr>`` to the end of the string given. For a 974 random read, it would look like ``rw=randread:8`` for passing in an offset 975 modifier with a value of 8. If the suffix is used with a sequential I/O 976 pattern, then the value specified will be added to the generated offset for 977 each I/O. For instance, using ``rw=write:4k`` will skip 4k for every 978 write. It turns sequential I/O into sequential I/O with holes. See the 979 :option:`rw_sequencer` option. 980 981.. option:: rw_sequencer=str 982 983 If an offset modifier is given by appending a number to the ``rw=<str>`` 984 line, then this option controls how that number modifies the I/O offset 985 being generated. Accepted values are: 986 987 **sequential** 988 Generate sequential offset. 989 **identical** 990 Generate the same offset. 991 992 ``sequential`` is only useful for random I/O, where fio would normally 993 generate a new random offset for every I/O. If you append e.g. 8 to randread, 994 you would get a new random offset for every 8 I/O's. The result would be a 995 seek for only every 8 I/O's, instead of for every I/O. Use ``rw=randread:8`` 996 to specify that. As sequential I/O is already sequential, setting 997 ``sequential`` for that would not result in any differences. ``identical`` 998 behaves in a similar fashion, except it sends the same offset 8 number of 999 times before generating a new offset. 1000 1001.. option:: unified_rw_reporting=bool 1002 1003 Fio normally reports statistics on a per data direction basis, meaning that 1004 reads, writes, and trims are accounted and reported separately. If this 1005 option is set fio sums the results and report them as "mixed" instead. 1006 1007.. option:: randrepeat=bool 1008 1009 Seed the random number generator used for random I/O patterns in a 1010 predictable way so the pattern is repeatable across runs. Default: true. 1011 1012.. option:: allrandrepeat=bool 1013 1014 Seed all random number generators in a predictable way so results are 1015 repeatable across runs. Default: false. 1016 1017.. option:: randseed=int 1018 1019 Seed the random number generators based on this seed value, to be able to 1020 control what sequence of output is being generated. If not set, the random 1021 sequence depends on the :option:`randrepeat` setting. 1022 1023.. option:: fallocate=str 1024 1025 Whether pre-allocation is performed when laying down files. 1026 Accepted values are: 1027 1028 **none** 1029 Do not pre-allocate space. 1030 1031 **posix** 1032 Pre-allocate via :manpage:`posix_fallocate(3)`. 1033 1034 **keep** 1035 Pre-allocate via :manpage:`fallocate(2)` with 1036 FALLOC_FL_KEEP_SIZE set. 1037 1038 **0** 1039 Backward-compatible alias for **none**. 1040 1041 **1** 1042 Backward-compatible alias for **posix**. 1043 1044 May not be available on all supported platforms. **keep** is only available 1045 on Linux. If using ZFS on Solaris this must be set to **none** because ZFS 1046 doesn't support it. Default: **posix**. 1047 1048.. option:: fadvise_hint=str 1049 1050 Use :manpage:`posix_fadvise(2)` to advise the kernel on what I/O patterns 1051 are likely to be issued. Accepted values are: 1052 1053 **0** 1054 Backwards-compatible hint for "no hint". 1055 1056 **1** 1057 Backwards compatible hint for "advise with fio workload type". This 1058 uses **FADV_RANDOM** for a random workload, and **FADV_SEQUENTIAL** 1059 for a sequential workload. 1060 1061 **sequential** 1062 Advise using **FADV_SEQUENTIAL**. 1063 1064 **random** 1065 Advise using **FADV_RANDOM**. 1066 1067.. option:: fadvise_stream=int 1068 1069 Use :manpage:`posix_fadvise(2)` to advise the kernel what stream ID the 1070 writes issued belong to. Only supported on Linux. Note, this option may 1071 change going forward. 1072 1073.. option:: offset=int 1074 1075 Start I/O at the given offset in the file. The data before the given offset 1076 will not be touched. This effectively caps the file size at `real_size - 1077 offset`. Can be combined with :option:`size` to constrain the start and 1078 end range that I/O will be done within. 1079 1080.. option:: offset_increment=int 1081 1082 If this is provided, then the real offset becomes `offset + offset_increment 1083 * thread_number`, where the thread number is a counter that starts at 0 and 1084 is incremented for each sub-job (i.e. when :option:`numjobs` option is 1085 specified). This option is useful if there are several jobs which are 1086 intended to operate on a file in parallel disjoint segments, with even 1087 spacing between the starting points. 1088 1089.. option:: number_ios=int 1090 1091 Fio will normally perform I/Os until it has exhausted the size of the region 1092 set by :option:`size`, or if it exhaust the allocated time (or hits an error 1093 condition). With this setting, the range/size can be set independently of 1094 the number of I/Os to perform. When fio reaches this number, it will exit 1095 normally and report status. Note that this does not extend the amount of I/O 1096 that will be done, it will only stop fio if this condition is met before 1097 other end-of-job criteria. 1098 1099.. option:: fsync=int 1100 1101 If writing to a file, issue a sync of the dirty data for every number of 1102 blocks given. For example, if you give 32 as a parameter, fio will sync the 1103 file for every 32 writes issued. If fio is using non-buffered I/O, we may 1104 not sync the file. The exception is the sg I/O engine, which synchronizes 1105 the disk cache anyway. Defaults to 0, which means no sync every certain 1106 number of writes. 1107 1108.. option:: fdatasync=int 1109 1110 Like :option:`fsync` but uses :manpage:`fdatasync(2)` to only sync data and 1111 not metadata blocks. In Windows, FreeBSD, and DragonFlyBSD there is no 1112 :manpage:`fdatasync(2)`, this falls back to using :manpage:`fsync(2)`. 1113 Defaults to 0, which means no sync data every certain number of writes. 1114 1115.. option:: write_barrier=int 1116 1117 Make every `N-th` write a barrier write. 1118 1119.. option:: sync_file_range=str:val 1120 1121 Use :manpage:`sync_file_range(2)` for every `val` number of write 1122 operations. Fio will track range of writes that have happened since the last 1123 :manpage:`sync_file_range(2)` call. `str` can currently be one or more of: 1124 1125 **wait_before** 1126 SYNC_FILE_RANGE_WAIT_BEFORE 1127 **write** 1128 SYNC_FILE_RANGE_WRITE 1129 **wait_after** 1130 SYNC_FILE_RANGE_WAIT_AFTER 1131 1132 So if you do ``sync_file_range=wait_before,write:8``, fio would use 1133 ``SYNC_FILE_RANGE_WAIT_BEFORE | SYNC_FILE_RANGE_WRITE`` for every 8 1134 writes. Also see the :manpage:`sync_file_range(2)` man page. This option is 1135 Linux specific. 1136 1137.. option:: overwrite=bool 1138 1139 If true, writes to a file will always overwrite existing data. If the file 1140 doesn't already exist, it will be created before the write phase begins. If 1141 the file exists and is large enough for the specified write phase, nothing 1142 will be done. 1143 1144.. option:: end_fsync=bool 1145 1146 If true, fsync file contents when a write stage has completed. 1147 1148.. option:: fsync_on_close=bool 1149 1150 If true, fio will :manpage:`fsync(2)` a dirty file on close. This differs 1151 from end_fsync in that it will happen on every file close, not just at the 1152 end of the job. 1153 1154.. option:: rwmixread=int 1155 1156 Percentage of a mixed workload that should be reads. Default: 50. 1157 1158.. option:: rwmixwrite=int 1159 1160 Percentage of a mixed workload that should be writes. If both 1161 :option:`rwmixread` and :option:`rwmixwrite` is given and the values do not 1162 add up to 100%, the latter of the two will be used to override the 1163 first. This may interfere with a given rate setting, if fio is asked to 1164 limit reads or writes to a certain rate. If that is the case, then the 1165 distribution may be skewed. Default: 50. 1166 1167.. option:: random_distribution=str:float[,str:float][,str:float] 1168 1169 By default, fio will use a completely uniform random distribution when asked 1170 to perform random I/O. Sometimes it is useful to skew the distribution in 1171 specific ways, ensuring that some parts of the data is more hot than others. 1172 fio includes the following distribution models: 1173 1174 **random** 1175 Uniform random distribution 1176 1177 **zipf** 1178 Zipf distribution 1179 1180 **pareto** 1181 Pareto distribution 1182 1183 **gauss** 1184 Normal (Gaussian) distribution 1185 1186 **zoned** 1187 Zoned random distribution 1188 1189 When using a **zipf** or **pareto** distribution, an input value is also 1190 needed to define the access pattern. For **zipf**, this is the `zipf 1191 theta`. For **pareto**, it's the `Pareto power`. Fio includes a test 1192 program, :command:`genzipf`, that can be used visualize what the given input 1193 values will yield in terms of hit rates. If you wanted to use **zipf** with 1194 a `theta` of 1.2, you would use ``random_distribution=zipf:1.2`` as the 1195 option. If a non-uniform model is used, fio will disable use of the random 1196 map. For the **gauss** distribution, a normal deviation is supplied as a 1197 value between 0 and 100. 1198 1199 For a **zoned** distribution, fio supports specifying percentages of I/O 1200 access that should fall within what range of the file or device. For 1201 example, given a criteria of: 1202 1203 * 60% of accesses should be to the first 10% 1204 * 30% of accesses should be to the next 20% 1205 * 8% of accesses should be to to the next 30% 1206 * 2% of accesses should be to the next 40% 1207 1208 we can define that through zoning of the random accesses. For the above 1209 example, the user would do:: 1210 1211 random_distribution=zoned:60/10:30/20:8/30:2/40 1212 1213 similarly to how :option:`bssplit` works for setting ranges and percentages 1214 of block sizes. Like :option:`bssplit`, it's possible to specify separate 1215 zones for reads, writes, and trims. If just one set is given, it'll apply to 1216 all of them. 1217 1218.. option:: percentage_random=int[,int][,int] 1219 1220 For a random workload, set how big a percentage should be random. This 1221 defaults to 100%, in which case the workload is fully random. It can be set 1222 from anywhere from 0 to 100. Setting it to 0 would make the workload fully 1223 sequential. Any setting in between will result in a random mix of sequential 1224 and random I/O, at the given percentages. Comma-separated values may be 1225 specified for reads, writes, and trims as described in :option:`blocksize`. 1226 1227.. option:: norandommap 1228 1229 Normally fio will cover every block of the file when doing random I/O. If 1230 this option is given, fio will just get a new random offset without looking 1231 at past I/O history. This means that some blocks may not be read or written, 1232 and that some blocks may be read/written more than once. If this option is 1233 used with :option:`verify` and multiple blocksizes (via :option:`bsrange`), 1234 only intact blocks are verified, i.e., partially-overwritten blocks are 1235 ignored. 1236 1237.. option:: softrandommap=bool 1238 1239 See :option:`norandommap`. If fio runs with the random block map enabled and 1240 it fails to allocate the map, if this option is set it will continue without 1241 a random block map. As coverage will not be as complete as with random maps, 1242 this option is disabled by default. 1243 1244.. option:: random_generator=str 1245 1246 Fio supports the following engines for generating 1247 I/O offsets for random I/O: 1248 1249 **tausworthe** 1250 Strong 2^88 cycle random number generator 1251 **lfsr** 1252 Linear feedback shift register generator 1253 **tausworthe64** 1254 Strong 64-bit 2^258 cycle random number generator 1255 1256 **tausworthe** is a strong random number generator, but it requires tracking 1257 on the side if we want to ensure that blocks are only read or written 1258 once. **LFSR** guarantees that we never generate the same offset twice, and 1259 it's also less computationally expensive. It's not a true random generator, 1260 however, though for I/O purposes it's typically good enough. **LFSR** only 1261 works with single block sizes, not with workloads that use multiple block 1262 sizes. If used with such a workload, fio may read or write some blocks 1263 multiple times. The default value is **tausworthe**, unless the required 1264 space exceeds 2^32 blocks. If it does, then **tausworthe64** is 1265 selected automatically. 1266 1267 1268Block size 1269~~~~~~~~~~ 1270 1271.. option:: blocksize=int[,int][,int], bs=int[,int][,int] 1272 1273 The block size in bytes used for I/O units. Default: 4096. A single value 1274 applies to reads, writes, and trims. Comma-separated values may be 1275 specified for reads, writes, and trims. A value not terminated in a comma 1276 applies to subsequent types. 1277 1278 Examples: 1279 1280 **bs=256k** 1281 means 256k for reads, writes and trims. 1282 1283 **bs=8k,32k** 1284 means 8k for reads, 32k for writes and trims. 1285 1286 **bs=8k,32k,** 1287 means 8k for reads, 32k for writes, and default for trims. 1288 1289 **bs=,8k** 1290 means default for reads, 8k for writes and trims. 1291 1292 **bs=,8k,** 1293 means default for reads, 8k for writes, and default for writes. 1294 1295.. option:: blocksize_range=irange[,irange][,irange], bsrange=irange[,irange][,irange] 1296 1297 A range of block sizes in bytes for I/O units. The issued I/O unit will 1298 always be a multiple of the minimum size, unless 1299 :option:`blocksize_unaligned` is set. 1300 1301 Comma-separated ranges may be specified for reads, writes, and trims as 1302 described in :option:`blocksize`. 1303 1304 Example: ``bsrange=1k-4k,2k-8k``. 1305 1306.. option:: bssplit=str[,str][,str] 1307 1308 Sometimes you want even finer grained control of the block sizes issued, not 1309 just an even split between them. This option allows you to weight various 1310 block sizes, so that you are able to define a specific amount of block sizes 1311 issued. The format for this option is:: 1312 1313 bssplit=blocksize/percentage:blocksize/percentage 1314 1315 for as many block sizes as needed. So if you want to define a workload that 1316 has 50% 64k blocks, 10% 4k blocks, and 40% 32k blocks, you would write:: 1317 1318 bssplit=4k/10:64k/50:32k/40 1319 1320 Ordering does not matter. If the percentage is left blank, fio will fill in 1321 the remaining values evenly. So a bssplit option like this one:: 1322 1323 bssplit=4k/50:1k/:32k/ 1324 1325 would have 50% 4k ios, and 25% 1k and 32k ios. The percentages always add up 1326 to 100, if bssplit is given a range that adds up to more, it will error out. 1327 1328 Comma-separated values may be specified for reads, writes, and trims as 1329 described in :option:`blocksize`. 1330 1331 If you want a workload that has 50% 2k reads and 50% 4k reads, while having 1332 90% 4k writes and 10% 8k writes, you would specify:: 1333 1334 bssplit=2k/50:4k/50,4k/90,8k/10 1335 1336.. option:: blocksize_unaligned, bs_unaligned 1337 1338 If set, fio will issue I/O units with any size within 1339 :option:`blocksize_range`, not just multiples of the minimum size. This 1340 typically won't work with direct I/O, as that normally requires sector 1341 alignment. 1342 1343.. option:: bs_is_seq_rand 1344 1345 If this option is set, fio will use the normal read,write blocksize settings 1346 as sequential,random blocksize settings instead. Any random read or write 1347 will use the WRITE blocksize settings, and any sequential read or write will 1348 use the READ blocksize settings. 1349 1350.. option:: blockalign=int[,int][,int], ba=int[,int][,int] 1351 1352 Boundary to which fio will align random I/O units. Default: 1353 :option:`blocksize`. Minimum alignment is typically 512b for using direct 1354 I/O, though it usually depends on the hardware block size. This option is 1355 mutually exclusive with using a random map for files, so it will turn off 1356 that option. Comma-separated values may be specified for reads, writes, and 1357 trims as described in :option:`blocksize`. 1358 1359 1360Buffers and memory 1361~~~~~~~~~~~~~~~~~~ 1362 1363.. option:: zero_buffers 1364 1365 Initialize buffers with all zeros. Default: fill buffers with random data. 1366 1367.. option:: refill_buffers 1368 1369 If this option is given, fio will refill the I/O buffers on every 1370 submit. The default is to only fill it at init time and reuse that 1371 data. Only makes sense if zero_buffers isn't specified, naturally. If data 1372 verification is enabled, `refill_buffers` is also automatically enabled. 1373 1374.. option:: scramble_buffers=bool 1375 1376 If :option:`refill_buffers` is too costly and the target is using data 1377 deduplication, then setting this option will slightly modify the I/O buffer 1378 contents to defeat normal de-dupe attempts. This is not enough to defeat 1379 more clever block compression attempts, but it will stop naive dedupe of 1380 blocks. Default: true. 1381 1382.. option:: buffer_compress_percentage=int 1383 1384 If this is set, then fio will attempt to provide I/O buffer content (on 1385 WRITEs) that compress to the specified level. Fio does this by providing a 1386 mix of random data and a fixed pattern. The fixed pattern is either zeroes, 1387 or the pattern specified by :option:`buffer_pattern`. If the pattern option 1388 is used, it might skew the compression ratio slightly. Note that this is per 1389 block size unit, for file/disk wide compression level that matches this 1390 setting, you'll also want to set :option:`refill_buffers`. 1391 1392.. option:: buffer_compress_chunk=int 1393 1394 See :option:`buffer_compress_percentage`. This setting allows fio to manage 1395 how big the ranges of random data and zeroed data is. Without this set, fio 1396 will provide :option:`buffer_compress_percentage` of blocksize random data, 1397 followed by the remaining zeroed. With this set to some chunk size smaller 1398 than the block size, fio can alternate random and zeroed data throughout the 1399 I/O buffer. 1400 1401.. option:: buffer_pattern=str 1402 1403 If set, fio will fill the I/O buffers with this pattern. If not set, the 1404 contents of I/O buffers is defined by the other options related to buffer 1405 contents. The setting can be any pattern of bytes, and can be prefixed with 1406 0x for hex values. It may also be a string, where the string must then be 1407 wrapped with ``""``, e.g.:: 1408 1409 buffer_pattern="abcd" 1410 1411 or:: 1412 1413 buffer_pattern=-12 1414 1415 or:: 1416 1417 buffer_pattern=0xdeadface 1418 1419 Also you can combine everything together in any order:: 1420 1421 buffer_pattern=0xdeadface"abcd"-12 1422 1423.. option:: dedupe_percentage=int 1424 1425 If set, fio will generate this percentage of identical buffers when 1426 writing. These buffers will be naturally dedupable. The contents of the 1427 buffers depend on what other buffer compression settings have been set. It's 1428 possible to have the individual buffers either fully compressible, or not at 1429 all. This option only controls the distribution of unique buffers. 1430 1431.. option:: invalidate=bool 1432 1433 Invalidate the buffer/page cache parts for this file prior to starting 1434 I/O if the platform and file type support it. Defaults to true. 1435 This will be ignored if :option:`pre_read` is also specified for the 1436 same job. 1437 1438.. option:: sync=bool 1439 1440 Use synchronous I/O for buffered writes. For the majority of I/O engines, 1441 this means using O_SYNC. Default: false. 1442 1443.. option:: iomem=str, mem=str 1444 1445 Fio can use various types of memory as the I/O unit buffer. The allowed 1446 values are: 1447 1448 **malloc** 1449 Use memory from :manpage:`malloc(3)` as the buffers. Default memory 1450 type. 1451 1452 **shm** 1453 Use shared memory as the buffers. Allocated through 1454 :manpage:`shmget(2)`. 1455 1456 **shmhuge** 1457 Same as shm, but use huge pages as backing. 1458 1459 **mmap** 1460 Use mmap to allocate buffers. May either be anonymous memory, or can 1461 be file backed if a filename is given after the option. The format 1462 is `mem=mmap:/path/to/file`. 1463 1464 **mmaphuge** 1465 Use a memory mapped huge file as the buffer backing. Append filename 1466 after mmaphuge, ala `mem=mmaphuge:/hugetlbfs/file`. 1467 1468 **mmapshared** 1469 Same as mmap, but use a MMAP_SHARED mapping. 1470 1471 **cudamalloc** 1472 Use GPU memory as the buffers for GPUDirect RDMA benchmark. 1473 1474 The area allocated is a function of the maximum allowed bs size for the job, 1475 multiplied by the I/O depth given. Note that for **shmhuge** and 1476 **mmaphuge** to work, the system must have free huge pages allocated. This 1477 can normally be checked and set by reading/writing 1478 :file:`/proc/sys/vm/nr_hugepages` on a Linux system. Fio assumes a huge page 1479 is 4MiB in size. So to calculate the number of huge pages you need for a 1480 given job file, add up the I/O depth of all jobs (normally one unless 1481 :option:`iodepth` is used) and multiply by the maximum bs set. Then divide 1482 that number by the huge page size. You can see the size of the huge pages in 1483 :file:`/proc/meminfo`. If no huge pages are allocated by having a non-zero 1484 number in `nr_hugepages`, using **mmaphuge** or **shmhuge** will fail. Also 1485 see :option:`hugepage-size`. 1486 1487 **mmaphuge** also needs to have hugetlbfs mounted and the file location 1488 should point there. So if it's mounted in :file:`/huge`, you would use 1489 `mem=mmaphuge:/huge/somefile`. 1490 1491.. option:: iomem_align=int 1492 1493 This indicates the memory alignment of the I/O memory buffers. Note that 1494 the given alignment is applied to the first I/O unit buffer, if using 1495 :option:`iodepth` the alignment of the following buffers are given by the 1496 :option:`bs` used. In other words, if using a :option:`bs` that is a 1497 multiple of the page sized in the system, all buffers will be aligned to 1498 this value. If using a :option:`bs` that is not page aligned, the alignment 1499 of subsequent I/O memory buffers is the sum of the :option:`iomem_align` and 1500 :option:`bs` used. 1501 1502.. option:: hugepage-size=int 1503 1504 Defines the size of a huge page. Must at least be equal to the system 1505 setting, see :file:`/proc/meminfo`. Defaults to 4MiB. Should probably 1506 always be a multiple of megabytes, so using ``hugepage-size=Xm`` is the 1507 preferred way to set this to avoid setting a non-pow-2 bad value. 1508 1509.. option:: lockmem=int 1510 1511 Pin the specified amount of memory with :manpage:`mlock(2)`. Can be used to 1512 simulate a smaller amount of memory. The amount specified is per worker. 1513 1514 1515I/O size 1516~~~~~~~~ 1517 1518.. option:: size=int 1519 1520 The total size of file I/O for each thread of this job. Fio will run until 1521 this many bytes has been transferred, unless runtime is limited by other options 1522 (such as :option:`runtime`, for instance, or increased/decreased by :option:`io_size`). 1523 Fio will divide this size between the available files determined by options 1524 such as :option:`nrfiles`, :option:`filename`, unless :option:`filesize` is 1525 specified by the job. If the result of division happens to be 0, the size is 1526 set to the physical size of the given files or devices if they exist. 1527 If this option is not specified, fio will use the full size of the given 1528 files or devices. If the files do not exist, size must be given. It is also 1529 possible to give size as a percentage between 1 and 100. If ``size=20%`` is 1530 given, fio will use 20% of the full size of the given files or devices. 1531 Can be combined with :option:`offset` to constrain the start and end range 1532 that I/O will be done within. 1533 1534.. option:: io_size=int, io_limit=int 1535 1536 Normally fio operates within the region set by :option:`size`, which means 1537 that the :option:`size` option sets both the region and size of I/O to be 1538 performed. Sometimes that is not what you want. With this option, it is 1539 possible to define just the amount of I/O that fio should do. For instance, 1540 if :option:`size` is set to 20GiB and :option:`io_size` is set to 5GiB, fio 1541 will perform I/O within the first 20GiB but exit when 5GiB have been 1542 done. The opposite is also possible -- if :option:`size` is set to 20GiB, 1543 and :option:`io_size` is set to 40GiB, then fio will do 40GiB of I/O within 1544 the 0..20GiB region. 1545 1546.. option:: filesize=int 1547 1548 Individual file sizes. May be a range, in which case fio will select sizes 1549 for files at random within the given range and limited to :option:`size` in 1550 total (if that is given). If not given, each created file is the same size. 1551 This option overrides :option:`size` in terms of file size, which means 1552 this value is used as a fixed size or possible range of each file. 1553 1554.. option:: file_append=bool 1555 1556 Perform I/O after the end of the file. Normally fio will operate within the 1557 size of a file. If this option is set, then fio will append to the file 1558 instead. This has identical behavior to setting :option:`offset` to the size 1559 of a file. This option is ignored on non-regular files. 1560 1561.. option:: fill_device=bool, fill_fs=bool 1562 1563 Sets size to something really large and waits for ENOSPC (no space left on 1564 device) as the terminating condition. Only makes sense with sequential 1565 write. For a read workload, the mount point will be filled first then I/O 1566 started on the result. This option doesn't make sense if operating on a raw 1567 device node, since the size of that is already known by the file system. 1568 Additionally, writing beyond end-of-device will not return ENOSPC there. 1569 1570 1571I/O engine 1572~~~~~~~~~~ 1573 1574.. option:: ioengine=str 1575 1576 Defines how the job issues I/O to the file. The following types are defined: 1577 1578 **sync** 1579 Basic :manpage:`read(2)` or :manpage:`write(2)` 1580 I/O. :manpage:`lseek(2)` is used to position the I/O location. 1581 See :option:`fsync` and :option:`fdatasync` for syncing write I/Os. 1582 1583 **psync** 1584 Basic :manpage:`pread(2)` or :manpage:`pwrite(2)` I/O. Default on 1585 all supported operating systems except for Windows. 1586 1587 **vsync** 1588 Basic :manpage:`readv(2)` or :manpage:`writev(2)` I/O. Will emulate 1589 queuing by coalescing adjacent I/Os into a single submission. 1590 1591 **pvsync** 1592 Basic :manpage:`preadv(2)` or :manpage:`pwritev(2)` I/O. 1593 1594 **pvsync2** 1595 Basic :manpage:`preadv2(2)` or :manpage:`pwritev2(2)` I/O. 1596 1597 **libaio** 1598 Linux native asynchronous I/O. Note that Linux may only support 1599 queued behaviour with non-buffered I/O (set ``direct=1`` or 1600 ``buffered=0``). 1601 This engine defines engine specific options. 1602 1603 **posixaio** 1604 POSIX asynchronous I/O using :manpage:`aio_read(3)` and 1605 :manpage:`aio_write(3)`. 1606 1607 **solarisaio** 1608 Solaris native asynchronous I/O. 1609 1610 **windowsaio** 1611 Windows native asynchronous I/O. Default on Windows. 1612 1613 **mmap** 1614 File is memory mapped with :manpage:`mmap(2)` and data copied 1615 to/from using :manpage:`memcpy(3)`. 1616 1617 **splice** 1618 :manpage:`splice(2)` is used to transfer the data and 1619 :manpage:`vmsplice(2)` to transfer data from user space to the 1620 kernel. 1621 1622 **sg** 1623 SCSI generic sg v3 I/O. May either be synchronous using the SG_IO 1624 ioctl, or if the target is an sg character device we use 1625 :manpage:`read(2)` and :manpage:`write(2)` for asynchronous 1626 I/O. Requires filename option to specify either block or character 1627 devices. 1628 1629 **null** 1630 Doesn't transfer any data, just pretends to. This is mainly used to 1631 exercise fio itself and for debugging/testing purposes. 1632 1633 **net** 1634 Transfer over the network to given ``host:port``. Depending on the 1635 :option:`protocol` used, the :option:`hostname`, :option:`port`, 1636 :option:`listen` and :option:`filename` options are used to specify 1637 what sort of connection to make, while the :option:`protocol` option 1638 determines which protocol will be used. This engine defines engine 1639 specific options. 1640 1641 **netsplice** 1642 Like **net**, but uses :manpage:`splice(2)` and 1643 :manpage:`vmsplice(2)` to map data and send/receive. 1644 This engine defines engine specific options. 1645 1646 **cpuio** 1647 Doesn't transfer any data, but burns CPU cycles according to the 1648 :option:`cpuload` and :option:`cpuchunks` options. Setting 1649 :option:`cpuload` =85 will cause that job to do nothing but burn 85% 1650 of the CPU. In case of SMP machines, use :option:`numjobs` 1651 =<no_of_cpu> to get desired CPU usage, as the cpuload only loads a 1652 single CPU at the desired rate. A job never finishes unless there is 1653 at least one non-cpuio job. 1654 1655 **guasi** 1656 The GUASI I/O engine is the Generic Userspace Asyncronous Syscall 1657 Interface approach to async I/O. See 1658 1659 http://www.xmailserver.org/guasi-lib.html 1660 1661 for more info on GUASI. 1662 1663 **rdma** 1664 The RDMA I/O engine supports both RDMA memory semantics 1665 (RDMA_WRITE/RDMA_READ) and channel semantics (Send/Recv) for the 1666 InfiniBand, RoCE and iWARP protocols. 1667 1668 **falloc** 1669 I/O engine that does regular fallocate to simulate data transfer as 1670 fio ioengine. 1671 1672 DDIR_READ 1673 does fallocate(,mode = FALLOC_FL_KEEP_SIZE,). 1674 1675 DDIR_WRITE 1676 does fallocate(,mode = 0). 1677 1678 DDIR_TRIM 1679 does fallocate(,mode = FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE). 1680 1681 **ftruncate** 1682 I/O engine that sends :manpage:`ftruncate(2)` operations in response 1683 to write (DDIR_WRITE) events. Each ftruncate issued sets the file's 1684 size to the current block offset. Block size is ignored. 1685 1686 **e4defrag** 1687 I/O engine that does regular EXT4_IOC_MOVE_EXT ioctls to simulate 1688 defragment activity in request to DDIR_WRITE event. 1689 1690 **rbd** 1691 I/O engine supporting direct access to Ceph Rados Block Devices 1692 (RBD) via librbd without the need to use the kernel rbd driver. This 1693 ioengine defines engine specific options. 1694 1695 **gfapi** 1696 Using Glusterfs libgfapi sync interface to direct access to 1697 Glusterfs volumes without having to go through FUSE. This ioengine 1698 defines engine specific options. 1699 1700 **gfapi_async** 1701 Using Glusterfs libgfapi async interface to direct access to 1702 Glusterfs volumes without having to go through FUSE. This ioengine 1703 defines engine specific options. 1704 1705 **libhdfs** 1706 Read and write through Hadoop (HDFS). The :file:`filename` option 1707 is used to specify host,port of the hdfs name-node to connect. This 1708 engine interprets offsets a little differently. In HDFS, files once 1709 created cannot be modified. So random writes are not possible. To 1710 imitate this, libhdfs engine expects bunch of small files to be 1711 created over HDFS, and engine will randomly pick a file out of those 1712 files based on the offset generated by fio backend. (see the example 1713 job file to create such files, use ``rw=write`` option). Please 1714 note, you might want to set necessary environment variables to work 1715 with hdfs/libhdfs properly. Each job uses its own connection to 1716 HDFS. 1717 1718 **mtd** 1719 Read, write and erase an MTD character device (e.g., 1720 :file:`/dev/mtd0`). Discards are treated as erases. Depending on the 1721 underlying device type, the I/O may have to go in a certain pattern, 1722 e.g., on NAND, writing sequentially to erase blocks and discarding 1723 before overwriting. The writetrim mode works well for this 1724 constraint. 1725 1726 **pmemblk** 1727 Read and write using filesystem DAX to a file on a filesystem 1728 mounted with DAX on a persistent memory device through the NVML 1729 libpmemblk library. 1730 1731 **dev-dax** 1732 Read and write using device DAX to a persistent memory device (e.g., 1733 /dev/dax0.0) through the NVML libpmem library. 1734 1735 **external** 1736 Prefix to specify loading an external I/O engine object file. Append 1737 the engine filename, e.g. ``ioengine=external:/tmp/foo.o`` to load 1738 ioengine :file:`foo.o` in :file:`/tmp`. 1739 1740 1741I/O engine specific parameters 1742~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 1743 1744In addition, there are some parameters which are only valid when a specific 1745ioengine is in use. These are used identically to normal parameters, with the 1746caveat that when used on the command line, they must come after the 1747:option:`ioengine` that defines them is selected. 1748 1749.. option:: userspace_reap : [libaio] 1750 1751 Normally, with the libaio engine in use, fio will use the 1752 :manpage:`io_getevents(2)` system call to reap newly returned events. With 1753 this flag turned on, the AIO ring will be read directly from user-space to 1754 reap events. The reaping mode is only enabled when polling for a minimum of 1755 0 events (e.g. when :option:`iodepth_batch_complete` `=0`). 1756 1757.. option:: hipri : [pvsync2] 1758 1759 Set RWF_HIPRI on I/O, indicating to the kernel that it's of higher priority 1760 than normal. 1761 1762.. option:: cpuload=int : [cpuio] 1763 1764 Attempt to use the specified percentage of CPU cycles. This is a mandatory 1765 option when using cpuio I/O engine. 1766 1767.. option:: cpuchunks=int : [cpuio] 1768 1769 Split the load into cycles of the given time. In microseconds. 1770 1771.. option:: exit_on_io_done=bool : [cpuio] 1772 1773 Detect when I/O threads are done, then exit. 1774 1775.. option:: hostname=str : [netsplice] [net] 1776 1777 The host name or IP address to use for TCP or UDP based I/O. If the job is 1778 a TCP listener or UDP reader, the host name is not used and must be omitted 1779 unless it is a valid UDP multicast address. 1780 1781.. option:: namenode=str : [libhdfs] 1782 1783 The host name or IP address of a HDFS cluster namenode to contact. 1784 1785.. option:: port=int 1786 1787 [netsplice], [net] 1788 1789 The TCP or UDP port to bind to or connect to. If this is used with 1790 :option:`numjobs` to spawn multiple instances of the same job type, then 1791 this will be the starting port number since fio will use a range of 1792 ports. 1793 1794 [libhdfs] 1795 1796 the listening port of the HFDS cluster namenode. 1797 1798.. option:: interface=str : [netsplice] [net] 1799 1800 The IP address of the network interface used to send or receive UDP 1801 multicast. 1802 1803.. option:: ttl=int : [netsplice] [net] 1804 1805 Time-to-live value for outgoing UDP multicast packets. Default: 1. 1806 1807.. option:: nodelay=bool : [netsplice] [net] 1808 1809 Set TCP_NODELAY on TCP connections. 1810 1811.. option:: protocol=str : [netsplice] [net] 1812 1813.. option:: proto=str : [netsplice] [net] 1814 1815 The network protocol to use. Accepted values are: 1816 1817 **tcp** 1818 Transmission control protocol. 1819 **tcpv6** 1820 Transmission control protocol V6. 1821 **udp** 1822 User datagram protocol. 1823 **udpv6** 1824 User datagram protocol V6. 1825 **unix** 1826 UNIX domain socket. 1827 1828 When the protocol is TCP or UDP, the port must also be given, as well as the 1829 hostname if the job is a TCP listener or UDP reader. For unix sockets, the 1830 normal filename option should be used and the port is invalid. 1831 1832.. option:: listen : [net] 1833 1834 For TCP network connections, tell fio to listen for incoming connections 1835 rather than initiating an outgoing connection. The :option:`hostname` must 1836 be omitted if this option is used. 1837 1838.. option:: pingpong : [net] 1839 1840 Normally a network writer will just continue writing data, and a network 1841 reader will just consume packages. If ``pingpong=1`` is set, a writer will 1842 send its normal payload to the reader, then wait for the reader to send the 1843 same payload back. This allows fio to measure network latencies. The 1844 submission and completion latencies then measure local time spent sending or 1845 receiving, and the completion latency measures how long it took for the 1846 other end to receive and send back. For UDP multicast traffic 1847 ``pingpong=1`` should only be set for a single reader when multiple readers 1848 are listening to the same address. 1849 1850.. option:: window_size : [net] 1851 1852 Set the desired socket buffer size for the connection. 1853 1854.. option:: mss : [net] 1855 1856 Set the TCP maximum segment size (TCP_MAXSEG). 1857 1858.. option:: donorname=str : [e4defrag] 1859 1860 File will be used as a block donor(swap extents between files). 1861 1862.. option:: inplace=int : [e4defrag] 1863 1864 Configure donor file blocks allocation strategy: 1865 1866 **0** 1867 Default. Preallocate donor's file on init. 1868 **1** 1869 Allocate space immediately inside defragment event, and free right 1870 after event. 1871 1872.. option:: clustername=str : [rbd] 1873 1874 Specifies the name of the Ceph cluster. 1875 1876.. option:: rbdname=str : [rbd] 1877 1878 Specifies the name of the RBD. 1879 1880.. option:: pool=str : [rbd] 1881 1882 Specifies the name of the Ceph pool containing RBD. 1883 1884.. option:: clientname=str : [rbd] 1885 1886 Specifies the username (without the 'client.' prefix) used to access the 1887 Ceph cluster. If the *clustername* is specified, the *clientname* shall be 1888 the full *type.id* string. If no type. prefix is given, fio will add 1889 'client.' by default. 1890 1891.. option:: skip_bad=bool : [mtd] 1892 1893 Skip operations against known bad blocks. 1894 1895.. option:: hdfsdirectory : [libhdfs] 1896 1897 libhdfs will create chunk in this HDFS directory. 1898 1899.. option:: chunk_size : [libhdfs] 1900 1901 the size of the chunk to use for each file. 1902 1903 1904I/O depth 1905~~~~~~~~~ 1906 1907.. option:: iodepth=int 1908 1909 Number of I/O units to keep in flight against the file. Note that 1910 increasing *iodepth* beyond 1 will not affect synchronous ioengines (except 1911 for small degrees when :option:`verify_async` is in use). Even async 1912 engines may impose OS restrictions causing the desired depth not to be 1913 achieved. This may happen on Linux when using libaio and not setting 1914 :option:`direct` =1, since buffered I/O is not async on that OS. Keep an 1915 eye on the I/O depth distribution in the fio output to verify that the 1916 achieved depth is as expected. Default: 1. 1917 1918.. option:: iodepth_batch_submit=int, iodepth_batch=int 1919 1920 This defines how many pieces of I/O to submit at once. It defaults to 1 1921 which means that we submit each I/O as soon as it is available, but can be 1922 raised to submit bigger batches of I/O at the time. If it is set to 0 the 1923 :option:`iodepth` value will be used. 1924 1925.. option:: iodepth_batch_complete_min=int, iodepth_batch_complete=int 1926 1927 This defines how many pieces of I/O to retrieve at once. It defaults to 1 1928 which means that we'll ask for a minimum of 1 I/O in the retrieval process 1929 from the kernel. The I/O retrieval will go on until we hit the limit set by 1930 :option:`iodepth_low`. If this variable is set to 0, then fio will always 1931 check for completed events before queuing more I/O. This helps reduce I/O 1932 latency, at the cost of more retrieval system calls. 1933 1934.. option:: iodepth_batch_complete_max=int 1935 1936 This defines maximum pieces of I/O to retrieve at once. This variable should 1937 be used along with :option:`iodepth_batch_complete_min` =int variable, 1938 specifying the range of min and max amount of I/O which should be 1939 retrieved. By default it is equal to :option:`iodepth_batch_complete_min` 1940 value. 1941 1942 Example #1:: 1943 1944 iodepth_batch_complete_min=1 1945 iodepth_batch_complete_max=<iodepth> 1946 1947 which means that we will retrieve at least 1 I/O and up to the whole 1948 submitted queue depth. If none of I/O has been completed yet, we will wait. 1949 1950 Example #2:: 1951 1952 iodepth_batch_complete_min=0 1953 iodepth_batch_complete_max=<iodepth> 1954 1955 which means that we can retrieve up to the whole submitted queue depth, but 1956 if none of I/O has been completed yet, we will NOT wait and immediately exit 1957 the system call. In this example we simply do polling. 1958 1959.. option:: iodepth_low=int 1960 1961 The low water mark indicating when to start filling the queue 1962 again. Defaults to the same as :option:`iodepth`, meaning that fio will 1963 attempt to keep the queue full at all times. If :option:`iodepth` is set to 1964 e.g. 16 and *iodepth_low* is set to 4, then after fio has filled the queue of 1965 16 requests, it will let the depth drain down to 4 before starting to fill 1966 it again. 1967 1968.. option:: io_submit_mode=str 1969 1970 This option controls how fio submits the I/O to the I/O engine. The default 1971 is `inline`, which means that the fio job threads submit and reap I/O 1972 directly. If set to `offload`, the job threads will offload I/O submission 1973 to a dedicated pool of I/O threads. This requires some coordination and thus 1974 has a bit of extra overhead, especially for lower queue depth I/O where it 1975 can increase latencies. The benefit is that fio can manage submission rates 1976 independently of the device completion rates. This avoids skewed latency 1977 reporting if I/O gets back up on the device side (the coordinated omission 1978 problem). 1979 1980 1981I/O rate 1982~~~~~~~~ 1983 1984.. option:: thinktime=time 1985 1986 Stall the job for the specified period of time after an I/O has completed before issuing the 1987 next. May be used to simulate processing being done by an application. 1988 When the unit is omitted, the value is given in microseconds. See 1989 :option:`thinktime_blocks` and :option:`thinktime_spin`. 1990 1991.. option:: thinktime_spin=time 1992 1993 Only valid if :option:`thinktime` is set - pretend to spend CPU time doing 1994 something with the data received, before falling back to sleeping for the 1995 rest of the period specified by :option:`thinktime`. When the unit is 1996 omitted, the value is given in microseconds. 1997 1998.. option:: thinktime_blocks=int 1999 2000 Only valid if :option:`thinktime` is set - control how many blocks to issue, 2001 before waiting `thinktime` usecs. If not set, defaults to 1 which will make 2002 fio wait `thinktime` usecs after every block. This effectively makes any 2003 queue depth setting redundant, since no more than 1 I/O will be queued 2004 before we have to complete it and do our thinktime. In other words, this 2005 setting effectively caps the queue depth if the latter is larger. 2006 2007.. option:: rate=int[,int][,int] 2008 2009 Cap the bandwidth used by this job. The number is in bytes/sec, the normal 2010 suffix rules apply. Comma-separated values may be specified for reads, 2011 writes, and trims as described in :option:`blocksize`. 2012 2013.. option:: rate_min=int[,int][,int] 2014 2015 Tell fio to do whatever it can to maintain at least this bandwidth. Failing 2016 to meet this requirement will cause the job to exit. Comma-separated values 2017 may be specified for reads, writes, and trims as described in 2018 :option:`blocksize`. 2019 2020.. option:: rate_iops=int[,int][,int] 2021 2022 Cap the bandwidth to this number of IOPS. Basically the same as 2023 :option:`rate`, just specified independently of bandwidth. If the job is 2024 given a block size range instead of a fixed value, the smallest block size 2025 is used as the metric. Comma-separated values may be specified for reads, 2026 writes, and trims as described in :option:`blocksize`. 2027 2028.. option:: rate_iops_min=int[,int][,int] 2029 2030 If fio doesn't meet this rate of I/O, it will cause the job to exit. 2031 Comma-separated values may be specified for reads, writes, and trims as 2032 described in :option:`blocksize`. 2033 2034.. option:: rate_process=str 2035 2036 This option controls how fio manages rated I/O submissions. The default is 2037 `linear`, which submits I/O in a linear fashion with fixed delays between 2038 I/Os that gets adjusted based on I/O completion rates. If this is set to 2039 `poisson`, fio will submit I/O based on a more real world random request 2040 flow, known as the Poisson process 2041 (https://en.wikipedia.org/wiki/Poisson_point_process). The lambda will be 2042 10^6 / IOPS for the given workload. 2043 2044 2045I/O latency 2046~~~~~~~~~~~ 2047 2048.. option:: latency_target=time 2049 2050 If set, fio will attempt to find the max performance point that the given 2051 workload will run at while maintaining a latency below this target. When 2052 the unit is omitted, the value is given in microseconds. See 2053 :option:`latency_window` and :option:`latency_percentile`. 2054 2055.. option:: latency_window=time 2056 2057 Used with :option:`latency_target` to specify the sample window that the job 2058 is run at varying queue depths to test the performance. When the unit is 2059 omitted, the value is given in microseconds. 2060 2061.. option:: latency_percentile=float 2062 2063 The percentage of I/Os that must fall within the criteria specified by 2064 :option:`latency_target` and :option:`latency_window`. If not set, this 2065 defaults to 100.0, meaning that all I/Os must be equal or below to the value 2066 set by :option:`latency_target`. 2067 2068.. option:: max_latency=time 2069 2070 If set, fio will exit the job with an ETIMEDOUT error if it exceeds this 2071 maximum latency. When the unit is omitted, the value is given in 2072 microseconds. 2073 2074.. option:: rate_cycle=int 2075 2076 Average bandwidth for :option:`rate` and :option:`rate_min` over this number 2077 of milliseconds. 2078 2079 2080I/O replay 2081~~~~~~~~~~ 2082 2083.. option:: write_iolog=str 2084 2085 Write the issued I/O patterns to the specified file. See 2086 :option:`read_iolog`. Specify a separate file for each job, otherwise the 2087 iologs will be interspersed and the file may be corrupt. 2088 2089.. option:: read_iolog=str 2090 2091 Open an iolog with the specified file name and replay the I/O patterns it 2092 contains. This can be used to store a workload and replay it sometime 2093 later. The iolog given may also be a blktrace binary file, which allows fio 2094 to replay a workload captured by :command:`blktrace`. See 2095 :manpage:`blktrace(8)` for how to capture such logging data. For blktrace 2096 replay, the file needs to be turned into a blkparse binary data file first 2097 (``blkparse <device> -o /dev/null -d file_for_fio.bin``). 2098 2099.. option:: replay_no_stall=int 2100 2101 When replaying I/O with :option:`read_iolog` the default behavior is to 2102 attempt to respect the time stamps within the log and replay them with the 2103 appropriate delay between IOPS. By setting this variable fio will not 2104 respect the timestamps and attempt to replay them as fast as possible while 2105 still respecting ordering. The result is the same I/O pattern to a given 2106 device, but different timings. 2107 2108.. option:: replay_redirect=str 2109 2110 While replaying I/O patterns using :option:`read_iolog` the default behavior 2111 is to replay the IOPS onto the major/minor device that each IOP was recorded 2112 from. This is sometimes undesirable because on a different machine those 2113 major/minor numbers can map to a different device. Changing hardware on the 2114 same system can also result in a different major/minor mapping. 2115 ``replay_redirect`` causes all IOPS to be replayed onto the single specified 2116 device regardless of the device it was recorded 2117 from. i.e. :option:`replay_redirect` = :file:`/dev/sdc` would cause all I/O 2118 in the blktrace or iolog to be replayed onto :file:`/dev/sdc`. This means 2119 multiple devices will be replayed onto a single device, if the trace 2120 contains multiple devices. If you want multiple devices to be replayed 2121 concurrently to multiple redirected devices you must blkparse your trace 2122 into separate traces and replay them with independent fio invocations. 2123 Unfortunately this also breaks the strict time ordering between multiple 2124 device accesses. 2125 2126.. option:: replay_align=int 2127 2128 Force alignment of I/O offsets and lengths in a trace to this power of 2 2129 value. 2130 2131.. option:: replay_scale=int 2132 2133 Scale sector offsets down by this factor when replaying traces. 2134 2135 2136Threads, processes and job synchronization 2137~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 2138 2139.. option:: thread 2140 2141 Fio defaults to forking jobs, however if this option is given, fio will use 2142 POSIX Threads function :manpage:`pthread_create(3)` to create threads instead 2143 of forking processes. 2144 2145.. option:: wait_for=str 2146 2147 Specifies the name of the already defined job to wait for. Single waitee 2148 name only may be specified. If set, the job won't be started until all 2149 workers of the waitee job are done. 2150 2151 ``wait_for`` operates on the job name basis, so there are a few 2152 limitations. First, the waitee must be defined prior to the waiter job 2153 (meaning no forward references). Second, if a job is being referenced as a 2154 waitee, it must have a unique name (no duplicate waitees). 2155 2156.. option:: nice=int 2157 2158 Run the job with the given nice value. See man :manpage:`nice(2)`. 2159 2160 On Windows, values less than -15 set the process class to "High"; -1 through 2161 -15 set "Above Normal"; 1 through 15 "Below Normal"; and above 15 "Idle" 2162 priority class. 2163 2164.. option:: prio=int 2165 2166 Set the I/O priority value of this job. Linux limits us to a positive value 2167 between 0 and 7, with 0 being the highest. See man 2168 :manpage:`ionice(1)`. Refer to an appropriate manpage for other operating 2169 systems since meaning of priority may differ. 2170 2171.. option:: prioclass=int 2172 2173 Set the I/O priority class. See man :manpage:`ionice(1)`. 2174 2175.. option:: cpumask=int 2176 2177 Set the CPU affinity of this job. The parameter given is a bitmask of 2178 allowed CPU's the job may run on. So if you want the allowed CPUs to be 1 2179 and 5, you would pass the decimal value of (1 << 1 | 1 << 5), or 34. See man 2180 :manpage:`sched_setaffinity(2)`. This may not work on all supported 2181 operating systems or kernel versions. This option doesn't work well for a 2182 higher CPU count than what you can store in an integer mask, so it can only 2183 control cpus 1-32. For boxes with larger CPU counts, use 2184 :option:`cpus_allowed`. 2185 2186.. option:: cpus_allowed=str 2187 2188 Controls the same options as :option:`cpumask`, but it allows a text setting 2189 of the permitted CPUs instead. So to use CPUs 1 and 5, you would specify 2190 ``cpus_allowed=1,5``. This options also allows a range of CPUs. Say you 2191 wanted a binding to CPUs 1, 5, and 8-15, you would set 2192 ``cpus_allowed=1,5,8-15``. 2193 2194.. option:: cpus_allowed_policy=str 2195 2196 Set the policy of how fio distributes the CPUs specified by 2197 :option:`cpus_allowed` or cpumask. Two policies are supported: 2198 2199 **shared** 2200 All jobs will share the CPU set specified. 2201 **split** 2202 Each job will get a unique CPU from the CPU set. 2203 2204 **shared** is the default behaviour, if the option isn't specified. If 2205 **split** is specified, then fio will will assign one cpu per job. If not 2206 enough CPUs are given for the jobs listed, then fio will roundrobin the CPUs 2207 in the set. 2208 2209.. option:: numa_cpu_nodes=str 2210 2211 Set this job running on specified NUMA nodes' CPUs. The arguments allow 2212 comma delimited list of cpu numbers, A-B ranges, or `all`. Note, to enable 2213 numa options support, fio must be built on a system with libnuma-dev(el) 2214 installed. 2215 2216.. option:: numa_mem_policy=str 2217 2218 Set this job's memory policy and corresponding NUMA nodes. Format of the 2219 arguments:: 2220 2221 <mode>[:<nodelist>] 2222 2223 ``mode`` is one of the following memory policy: ``default``, ``prefer``, 2224 ``bind``, ``interleave``, ``local`` For ``default`` and ``local`` memory 2225 policy, no node is needed to be specified. For ``prefer``, only one node is 2226 allowed. For ``bind`` and ``interleave``, it allow comma delimited list of 2227 numbers, A-B ranges, or `all`. 2228 2229.. option:: cgroup=str 2230 2231 Add job to this control group. If it doesn't exist, it will be created. The 2232 system must have a mounted cgroup blkio mount point for this to work. If 2233 your system doesn't have it mounted, you can do so with:: 2234 2235 # mount -t cgroup -o blkio none /cgroup 2236 2237.. option:: cgroup_weight=int 2238 2239 Set the weight of the cgroup to this value. See the documentation that comes 2240 with the kernel, allowed values are in the range of 100..1000. 2241 2242.. option:: cgroup_nodelete=bool 2243 2244 Normally fio will delete the cgroups it has created after the job 2245 completion. To override this behavior and to leave cgroups around after the 2246 job completion, set ``cgroup_nodelete=1``. This can be useful if one wants 2247 to inspect various cgroup files after job completion. Default: false. 2248 2249.. option:: flow_id=int 2250 2251 The ID of the flow. If not specified, it defaults to being a global 2252 flow. See :option:`flow`. 2253 2254.. option:: flow=int 2255 2256 Weight in token-based flow control. If this value is used, then there is a 2257 'flow counter' which is used to regulate the proportion of activity between 2258 two or more jobs. Fio attempts to keep this flow counter near zero. The 2259 ``flow`` parameter stands for how much should be added or subtracted to the 2260 flow counter on each iteration of the main I/O loop. That is, if one job has 2261 ``flow=8`` and another job has ``flow=-1``, then there will be a roughly 1:8 2262 ratio in how much one runs vs the other. 2263 2264.. option:: flow_watermark=int 2265 2266 The maximum value that the absolute value of the flow counter is allowed to 2267 reach before the job must wait for a lower value of the counter. 2268 2269.. option:: flow_sleep=int 2270 2271 The period of time, in microseconds, to wait after the flow watermark has 2272 been exceeded before retrying operations. 2273 2274.. option:: stonewall, wait_for_previous 2275 2276 Wait for preceding jobs in the job file to exit, before starting this 2277 one. Can be used to insert serialization points in the job file. A stone 2278 wall also implies starting a new reporting group, see 2279 :option:`group_reporting`. 2280 2281.. option:: exitall 2282 2283 When one job finishes, terminate the rest. The default is to wait for each 2284 job to finish, sometimes that is not the desired action. 2285 2286.. option:: exec_prerun=str 2287 2288 Before running this job, issue the command specified through 2289 :manpage:`system(3)`. Output is redirected in a file called 2290 :file:`jobname.prerun.txt`. 2291 2292.. option:: exec_postrun=str 2293 2294 After the job completes, issue the command specified though 2295 :manpage:`system(3)`. Output is redirected in a file called 2296 :file:`jobname.postrun.txt`. 2297 2298.. option:: uid=int 2299 2300 Instead of running as the invoking user, set the user ID to this value 2301 before the thread/process does any work. 2302 2303.. option:: gid=int 2304 2305 Set group ID, see :option:`uid`. 2306 2307 2308Verification 2309~~~~~~~~~~~~ 2310 2311.. option:: verify_only 2312 2313 Do not perform specified workload, only verify data still matches previous 2314 invocation of this workload. This option allows one to check data multiple 2315 times at a later date without overwriting it. This option makes sense only 2316 for workloads that write data, and does not support workloads with the 2317 :option:`time_based` option set. 2318 2319.. option:: do_verify=bool 2320 2321 Run the verify phase after a write phase. Only valid if :option:`verify` is 2322 set. Default: true. 2323 2324.. option:: verify=str 2325 2326 If writing to a file, fio can verify the file contents after each iteration 2327 of the job. Each verification method also implies verification of special 2328 header, which is written to the beginning of each block. This header also 2329 includes meta information, like offset of the block, block number, timestamp 2330 when block was written, etc. :option:`verify` can be combined with 2331 :option:`verify_pattern` option. The allowed values are: 2332 2333 **md5** 2334 Use an md5 sum of the data area and store it in the header of 2335 each block. 2336 2337 **crc64** 2338 Use an experimental crc64 sum of the data area and store it in the 2339 header of each block. 2340 2341 **crc32c** 2342 Use a crc32c sum of the data area and store it in the header of each 2343 block. 2344 2345 **crc32c-intel** 2346 Use hardware assisted crc32c calculation provided on SSE4.2 enabled 2347 processors. Falls back to regular software crc32c, if not supported 2348 by the system. 2349 2350 **crc32** 2351 Use a crc32 sum of the data area and store it in the header of each 2352 block. 2353 2354 **crc16** 2355 Use a crc16 sum of the data area and store it in the header of each 2356 block. 2357 2358 **crc7** 2359 Use a crc7 sum of the data area and store it in the header of each 2360 block. 2361 2362 **xxhash** 2363 Use xxhash as the checksum function. Generally the fastest software 2364 checksum that fio supports. 2365 2366 **sha512** 2367 Use sha512 as the checksum function. 2368 2369 **sha256** 2370 Use sha256 as the checksum function. 2371 2372 **sha1** 2373 Use optimized sha1 as the checksum function. 2374 2375 **sha3-224** 2376 Use optimized sha3-224 as the checksum function. 2377 2378 **sha3-256** 2379 Use optimized sha3-256 as the checksum function. 2380 2381 **sha3-384** 2382 Use optimized sha3-384 as the checksum function. 2383 2384 **sha3-512** 2385 Use optimized sha3-512 as the checksum function. 2386 2387 **meta** 2388 This option is deprecated, since now meta information is included in 2389 generic verification header and meta verification happens by 2390 default. For detailed information see the description of the 2391 :option:`verify` setting. This option is kept because of 2392 compatibility's sake with old configurations. Do not use it. 2393 2394 **pattern** 2395 Verify a strict pattern. Normally fio includes a header with some 2396 basic information and checksumming, but if this option is set, only 2397 the specific pattern set with :option:`verify_pattern` is verified. 2398 2399 **null** 2400 Only pretend to verify. Useful for testing internals with 2401 :option:`ioengine` `=null`, not for much else. 2402 2403 This option can be used for repeated burn-in tests of a system to make sure 2404 that the written data is also correctly read back. If the data direction 2405 given is a read or random read, fio will assume that it should verify a 2406 previously written file. If the data direction includes any form of write, 2407 the verify will be of the newly written data. 2408 2409.. option:: verifysort=bool 2410 2411 If true, fio will sort written verify blocks when it deems it faster to read 2412 them back in a sorted manner. This is often the case when overwriting an 2413 existing file, since the blocks are already laid out in the file system. You 2414 can ignore this option unless doing huge amounts of really fast I/O where 2415 the red-black tree sorting CPU time becomes significant. Default: true. 2416 2417.. option:: verifysort_nr=int 2418 2419 Pre-load and sort verify blocks for a read workload. 2420 2421.. option:: verify_offset=int 2422 2423 Swap the verification header with data somewhere else in the block before 2424 writing. It is swapped back before verifying. 2425 2426.. option:: verify_interval=int 2427 2428 Write the verification header at a finer granularity than the 2429 :option:`blocksize`. It will be written for chunks the size of 2430 ``verify_interval``. :option:`blocksize` should divide this evenly. 2431 2432.. option:: verify_pattern=str 2433 2434 If set, fio will fill the I/O buffers with this pattern. Fio defaults to 2435 filling with totally random bytes, but sometimes it's interesting to fill 2436 with a known pattern for I/O verification purposes. Depending on the width 2437 of the pattern, fio will fill 1/2/3/4 bytes of the buffer at the time(it can 2438 be either a decimal or a hex number). The ``verify_pattern`` if larger than 2439 a 32-bit quantity has to be a hex number that starts with either "0x" or 2440 "0X". Use with :option:`verify`. Also, ``verify_pattern`` supports %o 2441 format, which means that for each block offset will be written and then 2442 verified back, e.g.:: 2443 2444 verify_pattern=%o 2445 2446 Or use combination of everything:: 2447 2448 verify_pattern=0xff%o"abcd"-12 2449 2450.. option:: verify_fatal=bool 2451 2452 Normally fio will keep checking the entire contents before quitting on a 2453 block verification failure. If this option is set, fio will exit the job on 2454 the first observed failure. Default: false. 2455 2456.. option:: verify_dump=bool 2457 2458 If set, dump the contents of both the original data block and the data block 2459 we read off disk to files. This allows later analysis to inspect just what 2460 kind of data corruption occurred. Off by default. 2461 2462.. option:: verify_async=int 2463 2464 Fio will normally verify I/O inline from the submitting thread. This option 2465 takes an integer describing how many async offload threads to create for I/O 2466 verification instead, causing fio to offload the duty of verifying I/O 2467 contents to one or more separate threads. If using this offload option, even 2468 sync I/O engines can benefit from using an :option:`iodepth` setting higher 2469 than 1, as it allows them to have I/O in flight while verifies are running. 2470 2471.. option:: verify_async_cpus=str 2472 2473 Tell fio to set the given CPU affinity on the async I/O verification 2474 threads. See :option:`cpus_allowed` for the format used. 2475 2476.. option:: verify_backlog=int 2477 2478 Fio will normally verify the written contents of a job that utilizes verify 2479 once that job has completed. In other words, everything is written then 2480 everything is read back and verified. You may want to verify continually 2481 instead for a variety of reasons. Fio stores the meta data associated with 2482 an I/O block in memory, so for large verify workloads, quite a bit of memory 2483 would be used up holding this meta data. If this option is enabled, fio will 2484 write only N blocks before verifying these blocks. 2485 2486.. option:: verify_backlog_batch=int 2487 2488 Control how many blocks fio will verify if :option:`verify_backlog` is 2489 set. If not set, will default to the value of :option:`verify_backlog` 2490 (meaning the entire queue is read back and verified). If 2491 ``verify_backlog_batch`` is less than :option:`verify_backlog` then not all 2492 blocks will be verified, if ``verify_backlog_batch`` is larger than 2493 :option:`verify_backlog`, some blocks will be verified more than once. 2494 2495.. option:: verify_state_save=bool 2496 2497 When a job exits during the write phase of a verify workload, save its 2498 current state. This allows fio to replay up until that point, if the verify 2499 state is loaded for the verify read phase. The format of the filename is, 2500 roughly:: 2501 2502 <type>-<jobname>-<jobindex>-verify.state. 2503 2504 <type> is "local" for a local run, "sock" for a client/server socket 2505 connection, and "ip" (192.168.0.1, for instance) for a networked 2506 client/server connection. 2507 2508.. option:: verify_state_load=bool 2509 2510 If a verify termination trigger was used, fio stores the current write state 2511 of each thread. This can be used at verification time so that fio knows how 2512 far it should verify. Without this information, fio will run a full 2513 verification pass, according to the settings in the job file used. 2514 2515.. option:: trim_percentage=int 2516 2517 Number of verify blocks to discard/trim. 2518 2519.. option:: trim_verify_zero=bool 2520 2521 Verify that trim/discarded blocks are returned as zeroes. 2522 2523.. option:: trim_backlog=int 2524 2525 Verify that trim/discarded blocks are returned as zeroes. 2526 2527.. option:: trim_backlog_batch=int 2528 2529 Trim this number of I/O blocks. 2530 2531.. option:: experimental_verify=bool 2532 2533 Enable experimental verification. 2534 2535 2536Steady state 2537~~~~~~~~~~~~ 2538 2539.. option:: steadystate=str:float, ss=str:float 2540 2541 Define the criterion and limit for assessing steady state performance. The 2542 first parameter designates the criterion whereas the second parameter sets 2543 the threshold. When the criterion falls below the threshold for the 2544 specified duration, the job will stop. For example, `iops_slope:0.1%` will 2545 direct fio to terminate the job when the least squares regression slope 2546 falls below 0.1% of the mean IOPS. If :option:`group_reporting` is enabled 2547 this will apply to all jobs in the group. Below is the list of available 2548 steady state assessment criteria. All assessments are carried out using only 2549 data from the rolling collection window. Threshold limits can be expressed 2550 as a fixed value or as a percentage of the mean in the collection window. 2551 2552 **iops** 2553 Collect IOPS data. Stop the job if all individual IOPS measurements 2554 are within the specified limit of the mean IOPS (e.g., ``iops:2`` 2555 means that all individual IOPS values must be within 2 of the mean, 2556 whereas ``iops:0.2%`` means that all individual IOPS values must be 2557 within 0.2% of the mean IOPS to terminate the job). 2558 2559 **iops_slope** 2560 Collect IOPS data and calculate the least squares regression 2561 slope. Stop the job if the slope falls below the specified limit. 2562 2563 **bw** 2564 Collect bandwidth data. Stop the job if all individual bandwidth 2565 measurements are within the specified limit of the mean bandwidth. 2566 2567 **bw_slope** 2568 Collect bandwidth data and calculate the least squares regression 2569 slope. Stop the job if the slope falls below the specified limit. 2570 2571.. option:: steadystate_duration=time, ss_dur=time 2572 2573 A rolling window of this duration will be used to judge whether steady state 2574 has been reached. Data will be collected once per second. The default is 0 2575 which disables steady state detection. When the unit is omitted, the 2576 value is given in seconds. 2577 2578.. option:: steadystate_ramp_time=time, ss_ramp=time 2579 2580 Allow the job to run for the specified duration before beginning data 2581 collection for checking the steady state job termination criterion. The 2582 default is 0. When the unit is omitted, the value is given in seconds. 2583 2584 2585Measurements and reporting 2586~~~~~~~~~~~~~~~~~~~~~~~~~~ 2587 2588.. option:: per_job_logs=bool 2589 2590 If set, this generates bw/clat/iops log with per file private filenames. If 2591 not set, jobs with identical names will share the log filename. Default: 2592 true. 2593 2594.. option:: group_reporting 2595 2596 It may sometimes be interesting to display statistics for groups of jobs as 2597 a whole instead of for each individual job. This is especially true if 2598 :option:`numjobs` is used; looking at individual thread/process output 2599 quickly becomes unwieldy. To see the final report per-group instead of 2600 per-job, use :option:`group_reporting`. Jobs in a file will be part of the 2601 same reporting group, unless if separated by a :option:`stonewall`, or by 2602 using :option:`new_group`. 2603 2604.. option:: new_group 2605 2606 Start a new reporting group. See: :option:`group_reporting`. If not given, 2607 all jobs in a file will be part of the same reporting group, unless 2608 separated by a :option:`stonewall`. 2609 2610.. option:: stats 2611 2612 By default, fio collects and shows final output results for all jobs 2613 that run. If this option is set to 0, then fio will ignore it in 2614 the final stat output. 2615 2616.. option:: write_bw_log=str 2617 2618 If given, write a bandwidth log for this job. Can be used to store data of 2619 the bandwidth of the jobs in their lifetime. The included 2620 :command:`fio_generate_plots` script uses :command:`gnuplot` to turn these 2621 text files into nice graphs. See :option:`write_lat_log` for behaviour of 2622 given filename. For this option, the postfix is :file:`_bw.x.log`, where `x` 2623 is the index of the job (`1..N`, where `N` is the number of jobs). If 2624 :option:`per_job_logs` is false, then the filename will not include the job 2625 index. See `Log File Formats`_. 2626 2627.. option:: write_lat_log=str 2628 2629 Same as :option:`write_bw_log`, except that this option stores I/O 2630 submission, completion, and total latencies instead. If no filename is given 2631 with this option, the default filename of :file:`jobname_type.log` is 2632 used. Even if the filename is given, fio will still append the type of 2633 log. So if one specifies:: 2634 2635 write_lat_log=foo 2636 2637 The actual log names will be :file:`foo_slat.x.log`, :file:`foo_clat.x.log`, 2638 and :file:`foo_lat.x.log`, where `x` is the index of the job (1..N, where N 2639 is the number of jobs). This helps :command:`fio_generate_plot` find the 2640 logs automatically. If :option:`per_job_logs` is false, then the filename 2641 will not include the job index. See `Log File Formats`_. 2642 2643.. option:: write_hist_log=str 2644 2645 Same as :option:`write_lat_log`, but writes I/O completion latency 2646 histograms. If no filename is given with this option, the default filename 2647 of :file:`jobname_clat_hist.x.log` is used, where `x` is the index of the 2648 job (1..N, where `N` is the number of jobs). Even if the filename is given, 2649 fio will still append the type of log. If :option:`per_job_logs` is false, 2650 then the filename will not include the job index. See `Log File Formats`_. 2651 2652.. option:: write_iops_log=str 2653 2654 Same as :option:`write_bw_log`, but writes IOPS. If no filename is given 2655 with this option, the default filename of :file:`jobname_type.x.log` is 2656 used,where `x` is the index of the job (1..N, where `N` is the number of 2657 jobs). Even if the filename is given, fio will still append the type of 2658 log. If :option:`per_job_logs` is false, then the filename will not include 2659 the job index. See `Log File Formats`_. 2660 2661.. option:: log_avg_msec=int 2662 2663 By default, fio will log an entry in the iops, latency, or bw log for every 2664 I/O that completes. When writing to the disk log, that can quickly grow to a 2665 very large size. Setting this option makes fio average the each log entry 2666 over the specified period of time, reducing the resolution of the log. See 2667 :option:`log_max_value` as well. Defaults to 0, logging all entries. 2668 2669.. option:: log_hist_msec=int 2670 2671 Same as :option:`log_avg_msec`, but logs entries for completion latency 2672 histograms. Computing latency percentiles from averages of intervals using 2673 :option:`log_avg_msec` is inaccurate. Setting this option makes fio log 2674 histogram entries over the specified period of time, reducing log sizes for 2675 high IOPS devices while retaining percentile accuracy. See 2676 :option:`log_hist_coarseness` as well. Defaults to 0, meaning histogram 2677 logging is disabled. 2678 2679.. option:: log_hist_coarseness=int 2680 2681 Integer ranging from 0 to 6, defining the coarseness of the resolution of 2682 the histogram logs enabled with :option:`log_hist_msec`. For each increment 2683 in coarseness, fio outputs half as many bins. Defaults to 0, for which 2684 histogram logs contain 1216 latency bins. See `Log File Formats`_. 2685 2686.. option:: log_max_value=bool 2687 2688 If :option:`log_avg_msec` is set, fio logs the average over that window. If 2689 you instead want to log the maximum value, set this option to 1. Defaults to 2690 0, meaning that averaged values are logged. 2691 2692.. option:: log_offset=int 2693 2694 If this is set, the iolog options will include the byte offset for the I/O 2695 entry as well as the other data values. 2696 2697.. option:: log_compression=int 2698 2699 If this is set, fio will compress the I/O logs as it goes, to keep the 2700 memory footprint lower. When a log reaches the specified size, that chunk is 2701 removed and compressed in the background. Given that I/O logs are fairly 2702 highly compressible, this yields a nice memory savings for longer runs. The 2703 downside is that the compression will consume some background CPU cycles, so 2704 it may impact the run. This, however, is also true if the logging ends up 2705 consuming most of the system memory. So pick your poison. The I/O logs are 2706 saved normally at the end of a run, by decompressing the chunks and storing 2707 them in the specified log file. This feature depends on the availability of 2708 zlib. 2709 2710.. option:: log_compression_cpus=str 2711 2712 Define the set of CPUs that are allowed to handle online log compression for 2713 the I/O jobs. This can provide better isolation between performance 2714 sensitive jobs, and background compression work. 2715 2716.. option:: log_store_compressed=bool 2717 2718 If set, fio will store the log files in a compressed format. They can be 2719 decompressed with fio, using the :option:`--inflate-log` command line 2720 parameter. The files will be stored with a :file:`.fz` suffix. 2721 2722.. option:: log_unix_epoch=bool 2723 2724 If set, fio will log Unix timestamps to the log files produced by enabling 2725 write_type_log for each log type, instead of the default zero-based 2726 timestamps. 2727 2728.. option:: block_error_percentiles=bool 2729 2730 If set, record errors in trim block-sized units from writes and trims and 2731 output a histogram of how many trims it took to get to errors, and what kind 2732 of error was encountered. 2733 2734.. option:: bwavgtime=int 2735 2736 Average the calculated bandwidth over the given time. Value is specified in 2737 milliseconds. If the job also does bandwidth logging through 2738 :option:`write_bw_log`, then the minimum of this option and 2739 :option:`log_avg_msec` will be used. Default: 500ms. 2740 2741.. option:: iopsavgtime=int 2742 2743 Average the calculated IOPS over the given time. Value is specified in 2744 milliseconds. If the job also does IOPS logging through 2745 :option:`write_iops_log`, then the minimum of this option and 2746 :option:`log_avg_msec` will be used. Default: 500ms. 2747 2748.. option:: disk_util=bool 2749 2750 Generate disk utilization statistics, if the platform supports it. 2751 Default: true. 2752 2753.. option:: disable_lat=bool 2754 2755 Disable measurements of total latency numbers. Useful only for cutting back 2756 the number of calls to :manpage:`gettimeofday(2)`, as that does impact 2757 performance at really high IOPS rates. Note that to really get rid of a 2758 large amount of these calls, this option must be used with 2759 :option:`disable_slat` and :option:`disable_bw_measurement` as well. 2760 2761.. option:: disable_clat=bool 2762 2763 Disable measurements of completion latency numbers. See 2764 :option:`disable_lat`. 2765 2766.. option:: disable_slat=bool 2767 2768 Disable measurements of submission latency numbers. See 2769 :option:`disable_slat`. 2770 2771.. option:: disable_bw_measurement=bool, disable_bw=bool 2772 2773 Disable measurements of throughput/bandwidth numbers. See 2774 :option:`disable_lat`. 2775 2776.. option:: clat_percentiles=bool 2777 2778 Enable the reporting of percentiles of completion latencies. 2779 2780.. option:: percentile_list=float_list 2781 2782 Overwrite the default list of percentiles for completion latencies and the 2783 block error histogram. Each number is a floating number in the range 2784 (0,100], and the maximum length of the list is 20. Use ``:`` to separate the 2785 numbers, and list the numbers in ascending order. For example, 2786 ``--percentile_list=99.5:99.9`` will cause fio to report the values of 2787 completion latency below which 99.5% and 99.9% of the observed latencies 2788 fell, respectively. 2789 2790 2791Error handling 2792~~~~~~~~~~~~~~ 2793 2794.. option:: exitall_on_error 2795 2796 When one job finishes in error, terminate the rest. The default is to wait 2797 for each job to finish. 2798 2799.. option:: continue_on_error=str 2800 2801 Normally fio will exit the job on the first observed failure. If this option 2802 is set, fio will continue the job when there is a 'non-fatal error' (EIO or 2803 EILSEQ) until the runtime is exceeded or the I/O size specified is 2804 completed. If this option is used, there are two more stats that are 2805 appended, the total error count and the first error. The error field given 2806 in the stats is the first error that was hit during the run. 2807 2808 The allowed values are: 2809 2810 **none** 2811 Exit on any I/O or verify errors. 2812 2813 **read** 2814 Continue on read errors, exit on all others. 2815 2816 **write** 2817 Continue on write errors, exit on all others. 2818 2819 **io** 2820 Continue on any I/O error, exit on all others. 2821 2822 **verify** 2823 Continue on verify errors, exit on all others. 2824 2825 **all** 2826 Continue on all errors. 2827 2828 **0** 2829 Backward-compatible alias for 'none'. 2830 2831 **1** 2832 Backward-compatible alias for 'all'. 2833 2834.. option:: ignore_error=str 2835 2836 Sometimes you want to ignore some errors during test in that case you can 2837 specify error list for each error type. 2838 ``ignore_error=READ_ERR_LIST,WRITE_ERR_LIST,VERIFY_ERR_LIST`` errors for 2839 given error type is separated with ':'. Error may be symbol ('ENOSPC', 2840 'ENOMEM') or integer. Example:: 2841 2842 ignore_error=EAGAIN,ENOSPC:122 2843 2844 This option will ignore EAGAIN from READ, and ENOSPC and 122(EDQUOT) from 2845 WRITE. 2846 2847.. option:: error_dump=bool 2848 2849 If set dump every error even if it is non fatal, true by default. If 2850 disabled only fatal error will be dumped. 2851 2852Running predefined workloads 2853---------------------------- 2854 2855Fio includes predefined profiles that mimic the I/O workloads generated by 2856other tools. 2857 2858.. option:: profile=str 2859 2860 The predefined workload to run. Current profiles are: 2861 2862 **tiobench** 2863 Threaded I/O bench (tiotest/tiobench) like workload. 2864 2865 **act** 2866 Aerospike Certification Tool (ACT) like workload. 2867 2868To view a profile's additional options use :option:`--cmdhelp` after specifying 2869the profile. For example:: 2870 2871$ fio --profile=act --cmdhelp 2872 2873Act profile options 2874~~~~~~~~~~~~~~~~~~~ 2875 2876.. option:: device-names=str 2877 :noindex: 2878 2879 Devices to use. 2880 2881.. option:: load=int 2882 :noindex: 2883 2884 ACT load multiplier. Default: 1. 2885 2886.. option:: test-duration=time 2887 :noindex: 2888 2889 How long the entire test takes to run. Default: 24h. 2890 2891.. option:: threads-per-queue=int 2892 :noindex: 2893 2894 Number of read IO threads per device. Default: 8. 2895 2896.. option:: read-req-num-512-blocks=int 2897 :noindex: 2898 2899 Number of 512B blocks to read at the time. Default: 3. 2900 2901.. option:: large-block-op-kbytes=int 2902 :noindex: 2903 2904 Size of large block ops in KiB (writes). Default: 131072. 2905 2906.. option:: prep 2907 :noindex: 2908 2909 Set to run ACT prep phase. 2910 2911Tiobench profile options 2912~~~~~~~~~~~~~~~~~~~~~~~~ 2913 2914.. option:: size=str 2915 :noindex: 2916 2917 Size in MiB 2918 2919.. option:: block=int 2920 :noindex: 2921 2922 Block size in bytes. Default: 4096. 2923 2924.. option:: numruns=int 2925 :noindex: 2926 2927 Number of runs. 2928 2929.. option:: dir=str 2930 :noindex: 2931 2932 Test directory. 2933 2934.. option:: threads=int 2935 :noindex: 2936 2937 Number of threads. 2938 2939Interpreting the output 2940----------------------- 2941 2942Fio spits out a lot of output. While running, fio will display the status of the 2943jobs created. An example of that would be:: 2944 2945 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] 2946 2947The characters inside the square brackets denote the current status of each 2948thread. The possible values (in typical life cycle order) are: 2949 2950+------+-----+-----------------------------------------------------------+ 2951| Idle | Run | | 2952+======+=====+===========================================================+ 2953| P | | Thread setup, but not started. | 2954+------+-----+-----------------------------------------------------------+ 2955| C | | Thread created. | 2956+------+-----+-----------------------------------------------------------+ 2957| I | | Thread initialized, waiting or generating necessary data. | 2958+------+-----+-----------------------------------------------------------+ 2959| | p | Thread running pre-reading file(s). | 2960+------+-----+-----------------------------------------------------------+ 2961| | R | Running, doing sequential reads. | 2962+------+-----+-----------------------------------------------------------+ 2963| | r | Running, doing random reads. | 2964+------+-----+-----------------------------------------------------------+ 2965| | W | Running, doing sequential writes. | 2966+------+-----+-----------------------------------------------------------+ 2967| | w | Running, doing random writes. | 2968+------+-----+-----------------------------------------------------------+ 2969| | M | Running, doing mixed sequential reads/writes. | 2970+------+-----+-----------------------------------------------------------+ 2971| | m | Running, doing mixed random reads/writes. | 2972+------+-----+-----------------------------------------------------------+ 2973| | F | Running, currently waiting for :manpage:`fsync(2)` | 2974+------+-----+-----------------------------------------------------------+ 2975| | V | Running, doing verification of written data. | 2976+------+-----+-----------------------------------------------------------+ 2977| E | | Thread exited, not reaped by main thread yet. | 2978+------+-----+-----------------------------------------------------------+ 2979| _ | | Thread reaped, or | 2980+------+-----+-----------------------------------------------------------+ 2981| X | | Thread reaped, exited with an error. | 2982+------+-----+-----------------------------------------------------------+ 2983| K | | Thread reaped, exited due to signal. | 2984+------+-----+-----------------------------------------------------------+ 2985 2986Fio will condense the thread string as not to take up more space on the command 2987line as is needed. For instance, if you have 10 readers and 10 writers running, 2988the output would look like this:: 2989 2990 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] 2991 2992Fio will still maintain the ordering, though. So the above means that jobs 1..10 2993are readers, and 11..20 are writers. 2994 2995The other values are fairly self explanatory -- number of threads currently 2996running and doing I/O, the number of currently open files (f=), the rate of I/O 2997since last check (read speed listed first, then write speed and optionally trim 2998speed), and the estimated completion percentage and time for the current 2999running group. It's impossible to estimate runtime of the following groups (if 3000any). Note that the string is displayed in order, so it's possible to tell which 3001of the jobs are currently doing what. The first character is the first job 3002defined in the job file, and so forth. 3003 3004When fio is done (or interrupted by :kbd:`ctrl-c`), it will show the data for 3005each thread, group of threads, and disks in that order. For each data direction, 3006the output looks like:: 3007 3008 Client1 (g=0): err= 0: 3009 write: io= 32MiB, bw= 666KiB/s, iops=89 , runt= 50320msec 3010 slat (msec): min= 0, max= 136, avg= 0.03, stdev= 1.92 3011 clat (msec): min= 0, max= 631, avg=48.50, stdev=86.82 3012 bw (KiB/s) : min= 0, max= 1196, per=51.00%, avg=664.02, stdev=681.68 3013 cpu : usr=1.49%, sys=0.25%, ctx=7969, majf=0, minf=17 3014 IO depths : 1=0.1%, 2=0.3%, 4=0.5%, 8=99.0%, 16=0.0%, 32=0.0%, >32=0.0% 3015 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0% 3016 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0% 3017 issued r/w: total=0/32768, short=0/0 3018 lat (msec): 2=1.6%, 4=0.0%, 10=3.2%, 20=12.8%, 50=38.4%, 100=24.8%, 3019 lat (msec): 250=15.2%, 500=0.0%, 750=0.0%, 1000=0.0%, >=2048=0.0% 3020 3021The client number is printed, along with the group id and error of that 3022thread. Below is the I/O statistics, here for writes. In the order listed, they 3023denote: 3024 3025**io** 3026 Number of megabytes I/O performed. 3027 3028**bw** 3029 Average bandwidth rate. 3030 3031**iops** 3032 Average I/Os performed per second. 3033 3034**runt** 3035 The runtime of that thread. 3036 3037**slat** 3038 Submission latency (avg being the average, stdev being the standard 3039 deviation). This is the time it took to submit the I/O. For sync I/O, 3040 the slat is really the completion latency, since queue/complete is one 3041 operation there. This value can be in milliseconds or microseconds, fio 3042 will choose the most appropriate base and print that. In the example 3043 above, milliseconds is the best scale. Note: in :option:`--minimal` mode 3044 latencies are always expressed in microseconds. 3045 3046**clat** 3047 Completion latency. Same names as slat, this denotes the time from 3048 submission to completion of the I/O pieces. For sync I/O, clat will 3049 usually be equal (or very close) to 0, as the time from submit to 3050 complete is basically just CPU time (I/O has already been done, see slat 3051 explanation). 3052 3053**bw** 3054 Bandwidth. Same names as the xlat stats, but also includes an 3055 approximate percentage of total aggregate bandwidth this thread received 3056 in this group. This last value is only really useful if the threads in 3057 this group are on the same disk, since they are then competing for disk 3058 access. 3059 3060**cpu** 3061 CPU usage. User and system time, along with the number of context 3062 switches this thread went through, usage of system and user time, and 3063 finally the number of major and minor page faults. The CPU utilization 3064 numbers are averages for the jobs in that reporting group, while the 3065 context and fault counters are summed. 3066 3067**IO depths** 3068 The distribution of I/O depths over the job life time. The numbers are 3069 divided into powers of 2, so for example the 16= entries includes depths 3070 up to that value but higher than the previous entry. In other words, it 3071 covers the range from 16 to 31. 3072 3073**IO submit** 3074 How many pieces of I/O were submitting in a single submit call. Each 3075 entry denotes that amount and below, until the previous entry -- e.g., 3076 8=100% mean that we submitted anywhere in between 5-8 I/Os per submit 3077 call. 3078 3079**IO complete** 3080 Like the above submit number, but for completions instead. 3081 3082**IO issued** 3083 The number of read/write requests issued, and how many of them were 3084 short. 3085 3086**IO latencies** 3087 The distribution of I/O completion latencies. This is the time from when 3088 I/O leaves fio and when it gets completed. The numbers follow the same 3089 pattern as the I/O depths, meaning that 2=1.6% means that 1.6% of the 3090 I/O completed within 2 msecs, 20=12.8% means that 12.8% of the I/O took 3091 more than 10 msecs, but less than (or equal to) 20 msecs. 3092 3093After each client has been listed, the group statistics are printed. They 3094will look like this:: 3095 3096 Run status group 0 (all jobs): 3097 READ: io=64MB, aggrb=22178, minb=11355, maxb=11814, mint=2840msec, maxt=2955msec 3098 WRITE: io=64MB, aggrb=1302, minb=666, maxb=669, mint=50093msec, maxt=50320msec 3099 3100For each data direction, it prints: 3101 3102**io** 3103 Number of megabytes I/O performed. 3104**aggrb** 3105 Aggregate bandwidth of threads in this group. 3106**minb** 3107 The minimum average bandwidth a thread saw. 3108**maxb** 3109 The maximum average bandwidth a thread saw. 3110**mint** 3111 The smallest runtime of the threads in that group. 3112**maxt** 3113 The longest runtime of the threads in that group. 3114 3115And finally, the disk statistics are printed. They will look like this:: 3116 3117 Disk stats (read/write): 3118 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00% 3119 3120Each value is printed for both reads and writes, with reads first. The 3121numbers denote: 3122 3123**ios** 3124 Number of I/Os performed by all groups. 3125**merge** 3126 Number of merges I/O the I/O scheduler. 3127**ticks** 3128 Number of ticks we kept the disk busy. 3129**io_queue** 3130 Total time spent in the disk queue. 3131**util** 3132 The disk utilization. A value of 100% means we kept the disk 3133 busy constantly, 50% would be a disk idling half of the time. 3134 3135It is also possible to get fio to dump the current output while it is running, 3136without terminating the job. To do that, send fio the **USR1** signal. You can 3137also get regularly timed dumps by using the :option:`--status-interval` 3138parameter, or by creating a file in :file:`/tmp` named 3139:file:`fio-dump-status`. If fio sees this file, it will unlink it and dump the 3140current output status. 3141 3142 3143Terse output 3144------------ 3145 3146For scripted usage where you typically want to generate tables or graphs of the 3147results, fio can output the results in a semicolon separated format. The format 3148is one long line of values, such as:: 3149 3150 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% 3151 A description of this job goes here. 3152 3153The job description (if provided) follows on a second line. 3154 3155To enable terse output, use the :option:`--minimal` command line option. The 3156first value is the version of the terse output format. If the output has to be 3157changed for some reason, this number will be incremented by 1 to signify that 3158change. 3159 3160Split up, the format is as follows: 3161 3162 :: 3163 3164 terse version, fio version, jobname, groupid, error 3165 3166 READ status:: 3167 3168 Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec) 3169 Submission latency: min, max, mean, stdev (usec) 3170 Completion latency: min, max, mean, stdev (usec) 3171 Completion latency percentiles: 20 fields (see below) 3172 Total latency: min, max, mean, stdev (usec) 3173 Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev 3174 3175 WRITE status: 3176 3177 :: 3178 3179 Total IO (KiB), bandwidth (KiB/sec), IOPS, runtime (msec) 3180 Submission latency: min, max, mean, stdev (usec) 3181 Completion latency: min, max, mean, stdev(usec) 3182 Completion latency percentiles: 20 fields (see below) 3183 Total latency: min, max, mean, stdev (usec) 3184 Bw (KiB/s): min, max, aggregate percentage of total, mean, stdev 3185 3186 CPU usage:: 3187 3188 user, system, context switches, major faults, minor faults 3189 3190 I/O depths:: 3191 3192 <=1, 2, 4, 8, 16, 32, >=64 3193 3194 I/O latencies microseconds:: 3195 3196 <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000 3197 3198 I/O latencies milliseconds:: 3199 3200 <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, 2000, >=2000 3201 3202 Disk utilization:: 3203 3204 Disk name, Read ios, write ios, 3205 Read merges, write merges, 3206 Read ticks, write ticks, 3207 Time spent in queue, disk utilization percentage 3208 3209 Additional Info (dependent on continue_on_error, default off):: 3210 3211 total # errors, first error code 3212 3213 Additional Info (dependent on description being set):: 3214 3215 Text description 3216 3217Completion latency percentiles can be a grouping of up to 20 sets, so for the 3218terse output fio writes all of them. Each field will look like this:: 3219 3220 1.00%=6112 3221 3222which is the Xth percentile, and the `usec` latency associated with it. 3223 3224For disk utilization, all disks used by fio are shown. So for each disk there 3225will be a disk utilization section. 3226 3227 3228Trace file format 3229----------------- 3230 3231There are two trace file format that you can encounter. The older (v1) format is 3232unsupported since version 1.20-rc3 (March 2008). It will still be described 3233below in case that you get an old trace and want to understand it. 3234 3235In any case the trace is a simple text file with a single action per line. 3236 3237 3238Trace file format v1 3239~~~~~~~~~~~~~~~~~~~~ 3240 3241Each line represents a single I/O action in the following format:: 3242 3243 rw, offset, length 3244 3245where `rw=0/1` for read/write, and the offset and length entries being in bytes. 3246 3247This format is not supported in fio versions => 1.20-rc3. 3248 3249 3250Trace file format v2 3251~~~~~~~~~~~~~~~~~~~~ 3252 3253The second version of the trace file format was added in fio version 1.17. It 3254allows to access more then one file per trace and has a bigger set of possible 3255file actions. 3256 3257The first line of the trace file has to be:: 3258 3259 fio version 2 iolog 3260 3261Following this can be lines in two different formats, which are described below. 3262 3263The file management format:: 3264 3265 filename action 3266 3267The filename is given as an absolute path. The action can be one of these: 3268 3269**add** 3270 Add the given filename to the trace. 3271**open** 3272 Open the file with the given filename. The filename has to have 3273 been added with the **add** action before. 3274**close** 3275 Close the file with the given filename. The file has to have been 3276 opened before. 3277 3278 3279The file I/O action format:: 3280 3281 filename action offset length 3282 3283The `filename` is given as an absolute path, and has to have been added and 3284opened before it can be used with this format. The `offset` and `length` are 3285given in bytes. The `action` can be one of these: 3286 3287**wait** 3288 Wait for `offset` microseconds. Everything below 100 is discarded. 3289 The time is relative to the previous `wait` statement. 3290**read** 3291 Read `length` bytes beginning from `offset`. 3292**write** 3293 Write `length` bytes beginning from `offset`. 3294**sync** 3295 :manpage:`fsync(2)` the file. 3296**datasync** 3297 :manpage:`fdatasync(2)` the file. 3298**trim** 3299 Trim the given file from the given `offset` for `length` bytes. 3300 3301CPU idleness profiling 3302---------------------- 3303 3304In some cases, we want to understand CPU overhead in a test. For example, we 3305test patches for the specific goodness of whether they reduce CPU usage. 3306Fio implements a balloon approach to create a thread per CPU that runs at idle 3307priority, meaning that it only runs when nobody else needs the cpu. 3308By measuring the amount of work completed by the thread, idleness of each CPU 3309can be derived accordingly. 3310 3311An unit work is defined as touching a full page of unsigned characters. Mean and 3312standard deviation of time to complete an unit work is reported in "unit work" 3313section. Options can be chosen to report detailed percpu idleness or overall 3314system idleness by aggregating percpu stats. 3315 3316 3317Verification and triggers 3318------------------------- 3319 3320Fio is usually run in one of two ways, when data verification is done. The first 3321is a normal write job of some sort with verify enabled. When the write phase has 3322completed, fio switches to reads and verifies everything it wrote. The second 3323model is running just the write phase, and then later on running the same job 3324(but with reads instead of writes) to repeat the same I/O patterns and verify 3325the contents. Both of these methods depend on the write phase being completed, 3326as fio otherwise has no idea how much data was written. 3327 3328With verification triggers, fio supports dumping the current write state to 3329local files. Then a subsequent read verify workload can load this state and know 3330exactly where to stop. This is useful for testing cases where power is cut to a 3331server in a managed fashion, for instance. 3332 3333A verification trigger consists of two things: 3334 33351) Storing the write state of each job. 33362) Executing a trigger command. 3337 3338The write state is relatively small, on the order of hundreds of bytes to single 3339kilobytes. It contains information on the number of completions done, the last X 3340completions, etc. 3341 3342A trigger is invoked either through creation ('touch') of a specified file in 3343the system, or through a timeout setting. If fio is run with 3344:option:`--trigger-file` = :file:`/tmp/trigger-file`, then it will continually 3345check for the existence of :file:`/tmp/trigger-file`. When it sees this file, it 3346will fire off the trigger (thus saving state, and executing the trigger 3347command). 3348 3349For client/server runs, there's both a local and remote trigger. If fio is 3350running as a server backend, it will send the job states back to the client for 3351safe storage, then execute the remote trigger, if specified. If a local trigger 3352is specified, the server will still send back the write state, but the client 3353will then execute the trigger. 3354 3355Verification trigger example 3356~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 3357 3358Lets say we want to run a powercut test on the remote machine 'server'. Our 3359write workload is in :file:`write-test.fio`. We want to cut power to 'server' at 3360some point during the run, and we'll run this test from the safety or our local 3361machine, 'localbox'. On the server, we'll start the fio backend normally:: 3362 3363 server# fio --server 3364 3365and on the client, we'll fire off the workload:: 3366 3367 localbox$ fio --client=server --trigger-file=/tmp/my-trigger --trigger-remote="bash -c \"echo b > /proc/sysrq-triger\"" 3368 3369We set :file:`/tmp/my-trigger` as the trigger file, and we tell fio to execute:: 3370 3371 echo b > /proc/sysrq-trigger 3372 3373on the server once it has received the trigger and sent us the write state. This 3374will work, but it's not **really** cutting power to the server, it's merely 3375abruptly rebooting it. If we have a remote way of cutting power to the server 3376through IPMI or similar, we could do that through a local trigger command 3377instead. Lets assume we have a script that does IPMI reboot of a given hostname, 3378ipmi-reboot. On localbox, we could then have run fio with a local trigger 3379instead:: 3380 3381 localbox$ fio --client=server --trigger-file=/tmp/my-trigger --trigger="ipmi-reboot server" 3382 3383For this case, fio would wait for the server to send us the write state, then 3384execute ``ipmi-reboot server`` when that happened. 3385 3386Loading verify state 3387~~~~~~~~~~~~~~~~~~~~ 3388 3389To load store write state, read verification job file must contain the 3390:option:`verify_state_load` option. If that is set, fio will load the previously 3391stored state. For a local fio run this is done by loading the files directly, 3392and on a client/server run, the server backend will ask the client to send the 3393files over and load them from there. 3394 3395 3396Log File Formats 3397---------------- 3398 3399Fio supports a variety of log file formats, for logging latencies, bandwidth, 3400and IOPS. The logs share a common format, which looks like this: 3401 3402 *time* (`msec`), *value*, *data direction*, *offset* 3403 3404Time for the log entry is always in milliseconds. The *value* logged depends 3405on the type of log, it will be one of the following: 3406 3407 **Latency log** 3408 Value is latency in usecs 3409 **Bandwidth log** 3410 Value is in KiB/sec 3411 **IOPS log** 3412 Value is IOPS 3413 3414*Data direction* is one of the following: 3415 3416 **0** 3417 I/O is a READ 3418 **1** 3419 I/O is a WRITE 3420 **2** 3421 I/O is a TRIM 3422 3423The *offset* is the offset, in bytes, from the start of the file, for that 3424particular I/O. The logging of the offset can be toggled with 3425:option:`log_offset`. 3426 3427If windowed logging is enabled through :option:`log_avg_msec` then fio doesn't 3428log individual I/Os. Instead of logs the average values over the specified period 3429of time. Since 'data direction' and 'offset' are per-I/O values, they aren't 3430applicable if windowed logging is enabled. If windowed logging is enabled and 3431:option:`log_max_value` is set, then fio logs maximum values in that window 3432instead of averages. 3433 3434 3435Client/server 3436------------- 3437 3438Normally fio is invoked as a stand-alone application on the machine where the 3439I/O workload should be generated. However, the frontend and backend of fio can 3440be run separately. Ie the fio server can generate an I/O workload on the "Device 3441Under Test" while being controlled from another machine. 3442 3443Start the server on the machine which has access to the storage DUT:: 3444 3445 fio --server=args 3446 3447where args defines what fio listens to. The arguments are of the form 3448``type,hostname`` or ``IP,port``. *type* is either ``ip`` (or ip4) for TCP/IP 3449v4, ``ip6`` for TCP/IP v6, or ``sock`` for a local unix domain socket. 3450*hostname* is either a hostname or IP address, and *port* is the port to listen 3451to (only valid for TCP/IP, not a local socket). Some examples: 3452 34531) ``fio --server`` 3454 3455 Start a fio server, listening on all interfaces on the default port (8765). 3456 34572) ``fio --server=ip:hostname,4444`` 3458 3459 Start a fio server, listening on IP belonging to hostname and on port 4444. 3460 34613) ``fio --server=ip6:::1,4444`` 3462 3463 Start a fio server, listening on IPv6 localhost ::1 and on port 4444. 3464 34654) ``fio --server=,4444`` 3466 3467 Start a fio server, listening on all interfaces on port 4444. 3468 34695) ``fio --server=1.2.3.4`` 3470 3471 Start a fio server, listening on IP 1.2.3.4 on the default port. 3472 34736) ``fio --server=sock:/tmp/fio.sock`` 3474 3475 Start a fio server, listening on the local socket /tmp/fio.sock. 3476 3477Once a server is running, a "client" can connect to the fio server with:: 3478 3479 fio <local-args> --client=<server> <remote-args> <job file(s)> 3480 3481where `local-args` are arguments for the client where it is running, `server` 3482is the connect string, and `remote-args` and `job file(s)` are sent to the 3483server. The `server` string follows the same format as it does on the server 3484side, to allow IP/hostname/socket and port strings. 3485 3486Fio can connect to multiple servers this way:: 3487 3488 fio --client=<server1> <job file(s)> --client=<server2> <job file(s)> 3489 3490If the job file is located on the fio server, then you can tell the server to 3491load a local file as well. This is done by using :option:`--remote-config` :: 3492 3493 fio --client=server --remote-config /path/to/file.fio 3494 3495Then fio will open this local (to the server) job file instead of being passed 3496one from the client. 3497 3498If you have many servers (example: 100 VMs/containers), you can input a pathname 3499of a file containing host IPs/names as the parameter value for the 3500:option:`--client` option. For example, here is an example :file:`host.list` 3501file containing 2 hostnames:: 3502 3503 host1.your.dns.domain 3504 host2.your.dns.domain 3505 3506The fio command would then be:: 3507 3508 fio --client=host.list <job file(s)> 3509 3510In this mode, you cannot input server-specific parameters or job files -- all 3511servers receive the same job file. 3512 3513In order to let ``fio --client`` runs use a shared filesystem from multiple 3514hosts, ``fio --client`` now prepends the IP address of the server to the 3515filename. For example, if fio is using directory :file:`/mnt/nfs/fio` and is 3516writing filename :file:`fileio.tmp`, with a :option:`--client` `hostfile` 3517containing two hostnames ``h1`` and ``h2`` with IP addresses 192.168.10.120 and 3518192.168.10.121, then fio will create two files:: 3519 3520 /mnt/nfs/fio/192.168.10.120.fileio.tmp 3521 /mnt/nfs/fio/192.168.10.121.fileio.tmp 3522