| 1 | fio |
| 2 | --- |
| 3 | |
| 4 | fio is a tool that will spawn a number of threads or processes doing a |
| 5 | particular type of io action as specified by the user. fio takes a |
| 6 | number of global parameters, each inherited by the thread unless |
| 7 | otherwise parameters given to them overriding that setting is given. |
| 8 | The typical use of fio is to write a job file matching the io load |
| 9 | one wants to simulate. |
| 10 | |
| 11 | |
| 12 | Source |
| 13 | ------ |
| 14 | |
| 15 | fio resides in a git repo, the canonical place is: |
| 16 | |
| 17 | git://git.kernel.dk/fio.git |
| 18 | |
| 19 | The http protocol also works, path is the same. |
| 20 | |
| 21 | Snapshots are frequently generated and they include the git meta data as |
| 22 | well. You can download them here: |
| 23 | |
| 24 | http://brick.kernel.dk/snaps/ |
| 25 | |
| 26 | Pascal Bleser <guru@unixtech.be> has fio RPMs in his repository for |
| 27 | SUSE variants, you can find them here: |
| 28 | |
| 29 | http://linux01.gwdg.de/~pbleser/rpm-navigation.php?cat=System/fio |
| 30 | |
| 31 | Dag Wieƫrs has RPMs for Red Hat related distros, find them here: |
| 32 | |
| 33 | http://dag.wieers.com/rpm/packages/fio/ |
| 34 | |
| 35 | Mandriva has integrated fio into their package repository, so installing |
| 36 | on that distro should be as easy as typing 'urpmi fio'. |
| 37 | |
| 38 | |
| 39 | Mailing list |
| 40 | ------------ |
| 41 | |
| 42 | There's a mailing list associated with fio. It's meant for general |
| 43 | discussion, bug reporting, questions, and development - basically anything |
| 44 | that has to do with fio. An automated mail detailing recent commits is |
| 45 | automatically sent to the list at most daily. The list address is |
| 46 | fio@vger.kernel.org, subscribe by sending an email to |
| 47 | majordomo@vger.kernel.org with |
| 48 | |
| 49 | subscribe fio |
| 50 | |
| 51 | in the body of the email. There is no archive for the new list yet, |
| 52 | archives for the old list can be found here: |
| 53 | |
| 54 | http://maillist.kernel.dk/fio-devel/ |
| 55 | |
| 56 | |
| 57 | Building |
| 58 | -------- |
| 59 | |
| 60 | Just type 'make' and 'make install'. If on FreeBSD, for now you have to |
| 61 | specify the FreeBSD Makefile with -f and use gmake (not make), eg: |
| 62 | |
| 63 | $ gmake -f Makefile.Freebsd && gmake -f Makefile.FreeBSD install |
| 64 | |
| 65 | Likewise with OpenSolaris, use the Makefile.solaris to compile there. |
| 66 | The OpenSolaris make should work fine. This might change in the |
| 67 | future if I opt for an autoconf type setup. |
| 68 | |
| 69 | If your compile fails with an error like this: |
| 70 | |
| 71 | CC gettime.o |
| 72 | In file included from fio.h:23, |
| 73 | from gettime.c:8: |
| 74 | os/os.h:15:20: error: libaio.h: No such file or directory |
| 75 | In file included from gettime.c:8: |
| 76 | fio.h:119: error: field 'iocb' has incomplete type |
| 77 | make: *** [gettime.o] Error 1 |
| 78 | |
| 79 | Check that you have the libaio development package installed. On RPM |
| 80 | based distros, it's typically called libaio-devel. |
| 81 | |
| 82 | |
| 83 | Command line |
| 84 | ------------ |
| 85 | |
| 86 | $ fio |
| 87 | --debug Enable some debugging options (see below) |
| 88 | --output Write output to file |
| 89 | --timeout Runtime in seconds |
| 90 | --latency-log Generate per-job latency logs |
| 91 | --bandwidth-log Generate per-job bandwidth logs |
| 92 | --minimal Minimal (terse) output |
| 93 | --version Print version info and exit |
| 94 | --help Print this page |
| 95 | --cmdhelp=cmd Print command help, "all" for all of them |
| 96 | --showcmd Turn a job file into command line options |
| 97 | --readonly Turn on safety read-only checks, preventing writes |
| 98 | --eta=when When ETA estimate should be printed |
| 99 | May be "always", "never" or "auto" |
| 100 | --section=name Only run specified section in job file |
| 101 | --alloc-size=kb Set smalloc pool to this size in kb (def 1024) |
| 102 | |
| 103 | |
| 104 | Any parameters following the options will be assumed to be job files, |
| 105 | unless they match a job file parameter. You can add as many as you want, |
| 106 | each job file will be regarded as a separate group and fio will stonewall |
| 107 | its execution. |
| 108 | |
| 109 | The --readonly switch is an extra safety guard to prevent accidentically |
| 110 | turning on a write setting when that is not desired. Fio will only write |
| 111 | if rw=write/randwrite/rw/randrw is given, but this extra safety net can |
| 112 | be used as an extra precaution. It will also enable a write check in the |
| 113 | io engine core to prevent an accidental write due to a fio bug. |
| 114 | |
| 115 | The debug switch allows adding options that trigger certain logging |
| 116 | options in fio. Currently the options are: |
| 117 | |
| 118 | process Dump info related to processes |
| 119 | file Dump info related to file actions |
| 120 | io Dump info related to IO queuing |
| 121 | mem Dump info related to memory allocations |
| 122 | blktrace Dump info related to blktrace setup |
| 123 | verify Dump info related to IO verification |
| 124 | all Enable all debug options |
| 125 | random Dump info related to random offset generation |
| 126 | parse Dump info related to option matching and parsing |
| 127 | diskutil Dump info related to disk utilization updates |
| 128 | job:x Dump info only related to job number x |
| 129 | mutex Dump info only related to mutex up/down ops |
| 130 | ? or help Show available debug options. |
| 131 | |
| 132 | You can specify as many as you want, eg --debug=file,mem will enable |
| 133 | file and memory debugging. |
| 134 | |
| 135 | The section switch is meant to make it easier to ship a bigger job file |
| 136 | instead of several smaller ones. Say you define a job file with light, |
| 137 | moderate, and heavy parts. Then you can ask fio to run the given part |
| 138 | only by giving it a --section=heavy command line option. The section |
| 139 | option only applies to job sections, the reserved 'global' section is |
| 140 | always parsed and taken into account. |
| 141 | |
| 142 | Fio has an internal allocator for shared memory called smalloc. It |
| 143 | allocates shared structures from this pool. The pool defaults to 1024k |
| 144 | in size, and can grow to 32 pools. If running large jobs with randommap |
| 145 | enabled it can run out of memory, in which case the --alloc-size switch |
| 146 | is handy for starting with a larger pool size. |
| 147 | |
| 148 | |
| 149 | Job file |
| 150 | -------- |
| 151 | |
| 152 | See the HOWTO file for a more detailed description of parameters and what |
| 153 | they mean. This file contains the terse version. You can describe big and |
| 154 | complex setups with the command line, but generally it's a lot easier to |
| 155 | just write a simple job file to describe the workload. The job file format |
| 156 | is in the ini style format, as that is easy to read and write for the user. |
| 157 | |
| 158 | The job file parameters are: |
| 159 | |
| 160 | name=x Use 'x' as the identifier for this job. |
| 161 | description=x 'x' is a text description of the job. |
| 162 | directory=x Use 'x' as the top level directory for storing files |
| 163 | filename=x Force the use of 'x' as the filename for all files |
| 164 | in this thread. If not given, fio will make up |
| 165 | a suitable filename based on the thread and file |
| 166 | number. |
| 167 | rw=x 'x' may be: read, randread, write, randwrite, |
| 168 | rw (read-write mix), randrw (read-write random mix) |
| 169 | rwmixcycle=x Base cycle for switching between read and write |
| 170 | in msecs. |
| 171 | rwmixread=x 'x' percentage of rw mix ios will be reads. If |
| 172 | rwmixwrite is also given, the last of the two will |
| 173 | be used if they don't add up to 100%. |
| 174 | rwmixwrite=x 'x' percentage of rw mix ios will be writes. See |
| 175 | rwmixread. |
| 176 | rand_repeatable=x The sequence of random io blocks can be repeatable |
| 177 | across runs, if 'x' is 1. |
| 178 | size=x Set file size to x bytes (x string can include k/m/g) |
| 179 | ioengine=x 'x' may be: aio/libaio/linuxaio for Linux aio, |
| 180 | posixaio for POSIX aio, solarisaio for Solaris |
| 181 | native async IO, sync for regular read/write io, |
| 182 | psync for regular pread/pwrite io, vsync for regular |
| 183 | readv/writev (with queuing emulation) mmap for mmap'ed |
| 184 | io, syslet-rw for syslet driven read/write, splice for |
| 185 | using splice/vmsplice, sg for direct SG_IO io, net |
| 186 | for network io, or cpuio for a cycler burner load. sg |
| 187 | only works on Linux on SCSI (or SCSI-like devices, such |
| 188 | as usb-storage or sata/libata driven) devices. Fio also |
| 189 | has a null io engine, which is mainly used for testing |
| 190 | fio itself. |
| 191 | |
| 192 | iodepth=x For async io, allow 'x' ios in flight |
| 193 | overwrite=x If 'x', layout a write file first. |
| 194 | nrfiles=x Spread io load over 'x' number of files per job, |
| 195 | if possible. |
| 196 | prio=x Run io at prio X, 0-7 is the kernel allowed range |
| 197 | prioclass=x Run io at prio class X |
| 198 | bs=x Use 'x' for thread blocksize. May include k/m postfix. |
| 199 | bsrange=x-y Mix thread block sizes randomly between x and y. May |
| 200 | also include k/m postfix. |
| 201 | direct=x 1 for direct IO, 0 for buffered IO |
| 202 | thinktime=x "Think" x usec after each io |
| 203 | rate=x Throttle rate to x KiB/sec |
| 204 | ratemin=x Quit if rate of x KiB/sec can't be met |
| 205 | ratecycle=x ratemin averaged over x msecs |
| 206 | cpumask=x Only allow job to run on CPUs defined by mask. |
| 207 | cpus_allowed=x Like 'cpumask', but allow text setting of CPU affinity. |
| 208 | fsync=x If writing with buffered IO, fsync after every |
| 209 | 'x' blocks have been written. |
| 210 | end_fsync=x If 'x', run fsync() after end-of-job. |
| 211 | startdelay=x Start this thread x seconds after startup |
| 212 | runtime=x Terminate x seconds after startup. Can include a |
| 213 | normal time suffix if not given in seconds, such as |
| 214 | 'm' for minutes, 'h' for hours, and 'd' for days. |
| 215 | offset=x Start io at offset x (x string can include k/m/g) |
| 216 | invalidate=x Invalidate page cache for file prior to doing io |
| 217 | sync=x Use sync writes if x and writing buffered IO. |
| 218 | mem=x If x == malloc, use malloc for buffers. If x == shm, |
| 219 | use shared memory for buffers. If x == mmap, use |
| 220 | anonymous mmap. |
| 221 | exitall When one thread quits, terminate the others |
| 222 | bwavgtime=x Average bandwidth stats over an x msec window. |
| 223 | create_serialize=x If 'x', serialize file creation. |
| 224 | create_fsync=x If 'x', run fsync() after file creation. |
| 225 | unlink If set, unlink files when done. |
| 226 | loops=x Run the job 'x' number of times. |
| 227 | verify=x If 'x' == md5, use md5 for verifies. If 'x' == crc32, |
| 228 | use crc32 for verifies. md5 is 'safer', but crc32 is |
| 229 | a lot faster. Only makes sense for writing to a file. |
| 230 | For other types of checksumming, see HOWTO. |
| 231 | stonewall Wait for preceeding jobs to end before running. |
| 232 | numjobs=x Create 'x' similar entries for this job |
| 233 | thread Use pthreads instead of forked jobs |
| 234 | zonesize=x |
| 235 | zoneskip=y Zone options must be paired. If given, the job |
| 236 | will skip y bytes for every x read/written. This |
| 237 | can be used to gauge hard drive speed over the entire |
| 238 | platter, without reading everything. Both x/y can |
| 239 | include k/m/g suffix. |
| 240 | iolog=x Open and read io pattern from file 'x'. The file must |
| 241 | contain one io action per line in the following format: |
| 242 | rw, offset, length |
| 243 | where with rw=0/1 for read/write, and the offset |
| 244 | and length entries being in bytes. |
| 245 | write_iolog=x Write an iolog to file 'x' in the same format as iolog. |
| 246 | The iolog options are exclusive, if both given the |
| 247 | read iolog will be performed. |
| 248 | write_bw_log Write a bandwidth log. |
| 249 | write_lat_log Write a latency log. |
| 250 | lockmem=x Lock down x amount of memory on the machine, to |
| 251 | simulate a machine with less memory available. x can |
| 252 | include k/m/g suffix. |
| 253 | nice=x Run job at given nice value. |
| 254 | exec_prerun=x Run 'x' before job io is begun. |
| 255 | exec_postrun=x Run 'x' after job io has finished. |
| 256 | ioscheduler=x Use ioscheduler 'x' for this job. |
| 257 | cpuload=x For a CPU io thread, percentage of CPU time to attempt |
| 258 | to burn. |
| 259 | cpuchunks=x Split burn cycles into pieces of x usecs. |
| 260 | |
| 261 | |
| 262 | |
| 263 | Platforms |
| 264 | --------- |
| 265 | |
| 266 | Fio works on (at least) Linux, Solaris, and FreeBSD. Some features and/or |
| 267 | options may only be available on some of the platforms, typically because |
| 268 | those features only apply to that platform (like the solarisaio engine, or |
| 269 | the splice engine on Linux). |
| 270 | |
| 271 | Some features are not available on FreeBSD/Solaris even if they could be |
| 272 | implemented, I'd be happy to take patches for that. An example of that is |
| 273 | disk utility statistics and (I think) huge page support, support for that |
| 274 | does exist in FreeBSD/Solaris. |
| 275 | |
| 276 | Fio uses pthread mutexes for signalling and locking and FreeBSD does not |
| 277 | support process shared pthread mutexes. As a result, only threads are |
| 278 | supported on FreeBSD. This could be fixed with sysv ipc locking or |
| 279 | other locking alternatives. |
| 280 | |
| 281 | Other *BSD platforms are untested, but fio should work there almost out |
| 282 | of the box. Since I don't do test runs or even compiles on those platforms, |
| 283 | your mileage may vary. Sending me patches for other platforms is greatly |
| 284 | appreciated. There's a lot of value in having the same test/benchmark tool |
| 285 | available on all platforms. |
| 286 | |
| 287 | |
| 288 | |
| 289 | Author |
| 290 | ------ |
| 291 | |
| 292 | Fio was written by Jens Axboe <axboe@kernel.dk> to enable flexible testing |
| 293 | of the Linux IO subsystem and schedulers. He got tired of writing |
| 294 | specific test applications to simulate a given workload, and found that |
| 295 | the existing io benchmark/test tools out there weren't flexible enough |
| 296 | to do what he wanted. |
| 297 | |
| 298 | Jens Axboe <axboe@kernel.dk> 20060905 |
| 299 | |