Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 1 | .TH fio 1 "September 2007" "User Manual" |
| 2 | .SH NAME |
| 3 | fio \- flexible I/O tester |
| 4 | .SH SYNOPSIS |
| 5 | .B fio |
| 6 | [\fIoptions\fR] [\fIjobfile\fR]... |
| 7 | .SH DESCRIPTION |
| 8 | .B fio |
| 9 | is a tool that will spawn a number of threads or processes doing a |
| 10 | particular type of I/O action as specified by the user. |
| 11 | The typical use of fio is to write a job file matching the I/O load |
| 12 | one wants to simulate. |
| 13 | .SH OPTIONS |
| 14 | .TP |
| 15 | .BI \-\-output \fR=\fPfilename |
| 16 | Write output to \fIfilename\fR. |
| 17 | .TP |
| 18 | .BI \-\-timeout \fR=\fPtimeout |
| 19 | Limit run time to \fItimeout\fR seconds. |
| 20 | .TP |
| 21 | .B \-\-latency\-log |
| 22 | Generate per-job latency logs. |
| 23 | .TP |
| 24 | .B \-\-bandwidth\-log |
| 25 | Generate per-job bandwidth logs. |
| 26 | .TP |
| 27 | .B \-\-minimal |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 28 | Print statistics in a terse, semicolon-delimited format. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 29 | .TP |
| 30 | .BI \-\-showcmd \fR=\fPjobfile |
| 31 | Convert \fIjobfile\fR to a set of command-line options. |
| 32 | .TP |
| 33 | .B \-\-readonly |
| 34 | Enable read-only safety checks. |
| 35 | .TP |
| 36 | .BI \-\-eta \fR=\fPwhen |
| 37 | Specifies when real-time ETA estimate should be printed. \fIwhen\fR may |
| 38 | be one of `always', `never' or `auto'. |
| 39 | .TP |
Aaron Carroll | c0a5d35 | 2008-02-26 23:10:39 +0100 | [diff] [blame] | 40 | .BI \-\-section \fR=\fPsec |
| 41 | Only run section \fIsec\fR from job file. |
| 42 | .TP |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 43 | .BI \-\-cmdhelp \fR=\fPcommand |
| 44 | Print help information for \fIcommand\fR. May be `all' for all commands. |
| 45 | .TP |
Jens Axboe | 9183788 | 2008-02-05 12:02:07 +0100 | [diff] [blame] | 46 | .BI \-\-debug \fR=\fPtype |
| 47 | Enable verbose tracing of various fio actions. May be `all' for all types |
Michael Prokop | c6e13ea | 2009-08-12 17:24:41 +0200 | [diff] [blame] | 48 | or individual types seperated by a comma (eg \-\-debug=io,file). `help' will |
Jens Axboe | 9183788 | 2008-02-05 12:02:07 +0100 | [diff] [blame] | 49 | list all available tracing options. |
| 50 | .TP |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 51 | .B \-\-help |
| 52 | Display usage information and exit. |
| 53 | .TP |
| 54 | .B \-\-version |
| 55 | Display version information and exit. |
| 56 | .SH "JOB FILE FORMAT" |
| 57 | Job files are in `ini' format. They consist of one or more |
| 58 | job definitions, which begin with a job name in square brackets and |
| 59 | extend to the next job name. The job name can be any ASCII string |
| 60 | except `global', which has a special meaning. Following the job name is |
| 61 | a sequence of zero or more parameters, one per line, that define the |
| 62 | behavior of the job. Any line starting with a `;' or `#' character is |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 63 | considered a comment and ignored. |
Aaron Carroll | d9956b6 | 2007-11-16 12:12:45 +0100 | [diff] [blame] | 64 | .P |
| 65 | If \fIjobfile\fR is specified as `-', the job file will be read from |
| 66 | standard input. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 67 | .SS "Global Section" |
| 68 | The global section contains default parameters for jobs specified in the |
| 69 | job file. A job is only affected by global sections residing above it, |
| 70 | and there may be any number of global sections. Specific job definitions |
| 71 | may override any parameter set in global sections. |
| 72 | .SH "JOB PARAMETERS" |
| 73 | .SS Types |
| 74 | Some parameters may take arguments of a specific type. The types used are: |
| 75 | .TP |
| 76 | .I str |
| 77 | String: a sequence of alphanumeric characters. |
| 78 | .TP |
| 79 | .I int |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 80 | SI integer: a whole number, possibly containing a suffix denoting the base unit |
Jens Axboe | b09da8f | 2009-07-17 23:16:17 +0200 | [diff] [blame] | 81 | of the value. Accepted suffixes are `k', 'M', 'G', 'T', and 'P', denoting |
| 82 | kilo (1024), mega (1024^2), giga (1024^3), tera (1024^4), and peta (1024^5) |
| 83 | respectively. The suffix is not case sensitive. If prefixed with '0x', the |
| 84 | value is assumed to be base 16 (hexadecimal). |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 85 | .TP |
| 86 | .I bool |
| 87 | Boolean: a true or false value. `0' denotes false, `1' denotes true. |
| 88 | .TP |
| 89 | .I irange |
| 90 | Integer range: a range of integers specified in the format |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 91 | \fIlower\fR:\fIupper\fR or \fIlower\fR\-\fIupper\fR. \fIlower\fR and |
| 92 | \fIupper\fR may contain a suffix as described above. If an option allows two |
| 93 | sets of ranges, they are separated with a `,' or `/' character. For example: |
| 94 | `8\-8k/8M\-4G'. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 95 | .SS "Parameter List" |
| 96 | .TP |
| 97 | .BI name \fR=\fPstr |
Aaron Carroll | d9956b6 | 2007-11-16 12:12:45 +0100 | [diff] [blame] | 98 | May be used to override the job name. On the command line, this parameter |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 99 | has the special purpose of signalling the start of a new job. |
| 100 | .TP |
| 101 | .BI description \fR=\fPstr |
| 102 | Human-readable description of the job. It is printed when the job is run, but |
| 103 | otherwise has no special purpose. |
| 104 | .TP |
| 105 | .BI directory \fR=\fPstr |
| 106 | Prefix filenames with this directory. Used to place files in a location other |
| 107 | than `./'. |
| 108 | .TP |
| 109 | .BI filename \fR=\fPstr |
| 110 | .B fio |
| 111 | normally makes up a file name based on the job name, thread number, and file |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 112 | number. If you want to share files between threads in a job or several jobs, |
| 113 | specify a \fIfilename\fR for each of them to override the default. If the I/O |
| 114 | engine used is `net', \fIfilename\fR is the host and port to connect to in the |
| 115 | format \fIhost\fR/\fIport\fR. If the I/O engine is file-based, you can specify |
| 116 | a number of files by separating the names with a `:' character. `\-' is a |
| 117 | reserved name, meaning stdin or stdout, depending on the read/write direction |
| 118 | set. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 119 | .TP |
Jens Axboe | 3ce9dca | 2009-06-10 08:55:21 +0200 | [diff] [blame] | 120 | .BI lockfile \fR=\fPstr |
| 121 | Fio defaults to not locking any files before it does IO to them. If a file or |
| 122 | file descriptor is shared, fio can serialize IO to that file to make the end |
| 123 | result consistent. This is usual for emulating real workloads that share files. |
| 124 | The lock modes are: |
| 125 | .RS |
| 126 | .RS |
| 127 | .TP |
| 128 | .B none |
| 129 | No locking. This is the default. |
| 130 | .TP |
| 131 | .B exclusive |
| 132 | Only one thread or process may do IO at the time, excluding all others. |
| 133 | .TP |
| 134 | .B readwrite |
| 135 | Read-write locking on the file. Many readers may access the file at the same |
| 136 | time, but writes get exclusive access. |
| 137 | .RE |
| 138 | .P |
| 139 | The option may be post-fixed with a lock batch number. If set, then each |
| 140 | thread/process may do that amount of IOs to the file before giving up the lock. |
| 141 | Since lock acquisition is expensive, batching the lock/unlocks will speed up IO. |
| 142 | .RE |
| 143 | .P |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 144 | .BI opendir \fR=\fPstr |
| 145 | Recursively open any files below directory \fIstr\fR. |
| 146 | .TP |
| 147 | .BI readwrite \fR=\fPstr "\fR,\fP rw" \fR=\fPstr |
| 148 | Type of I/O pattern. Accepted values are: |
| 149 | .RS |
| 150 | .RS |
| 151 | .TP |
| 152 | .B read |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 153 | Sequential reads. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 154 | .TP |
| 155 | .B write |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 156 | Sequential writes. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 157 | .TP |
| 158 | .B randread |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 159 | Random reads. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 160 | .TP |
| 161 | .B randwrite |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 162 | Random writes. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 163 | .TP |
| 164 | .B rw |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 165 | Mixed sequential reads and writes. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 166 | .TP |
| 167 | .B randrw |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 168 | Mixed random reads and writes. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 169 | .RE |
| 170 | .P |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 171 | For mixed I/O, the default split is 50/50. For random I/O, the number of I/Os |
| 172 | to perform before getting a new offset can be specified by appending |
| 173 | `:\fIint\fR' to the pattern type. The default is 1. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 174 | .RE |
| 175 | .TP |
Jens Axboe | 90fef2d | 2009-07-17 22:33:32 +0200 | [diff] [blame] | 176 | .BI kb_base \fR=\fPint |
| 177 | The base unit for a kilobyte. The defacto base is 2^10, 1024. Storage |
| 178 | manufacturers like to use 10^3 or 1000 as a base ten unit instead, for obvious |
| 179 | reasons. Allow values are 1024 or 1000, with 1024 being the default. |
| 180 | .TP |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 181 | .BI randrepeat \fR=\fPbool |
| 182 | Seed the random number generator in a predictable way so results are repeatable |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 183 | across runs. Default: true. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 184 | .TP |
| 185 | .BI fadvise_hint \fR=\fPbool |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 186 | Disable use of \fIposix_fadvise\fR\|(2) to advise the kernel what I/O patterns |
| 187 | are likely to be issued. Default: true. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 188 | .TP |
Jens Axboe | f7fa265 | 2009-03-09 14:20:20 +0100 | [diff] [blame] | 189 | .BI size \fR=\fPint |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 190 | Total size of I/O for this job. \fBfio\fR will run until this many bytes have |
| 191 | been transfered, unless limited by other options (\fBruntime\fR, for instance). |
| 192 | Unless \fBnr_files\fR and \fBfilesize\fR options are given, this amount will be |
| 193 | divided between the available files for the job. |
| 194 | .TP |
Jens Axboe | 3ce9dca | 2009-06-10 08:55:21 +0200 | [diff] [blame] | 195 | .BI fill_device \fR=\fPbool |
| 196 | Sets size to something really large and waits for ENOSPC (no space left on |
| 197 | device) as the terminating condition. Only makes sense with sequential write. |
| 198 | For a read workload, the mount point will be filled first then IO started on |
| 199 | the result. |
| 200 | .TP |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 201 | .BI filesize \fR=\fPirange |
| 202 | Individual file sizes. May be a range, in which case \fBfio\fR will select sizes |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 203 | for files at random within the given range, limited to \fBsize\fR in total (if |
| 204 | that is given). If \fBfilesize\fR is not specified, each created file is the |
| 205 | same size. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 206 | .TP |
Jens Axboe | f7fa265 | 2009-03-09 14:20:20 +0100 | [diff] [blame] | 207 | .BI blocksize \fR=\fPint[,int] "\fR,\fB bs" \fR=\fPint[,int] |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 208 | Block size for I/O units. Default: 4k. Values for reads and writes can be |
| 209 | specified seperately in the format \fIread\fR,\fIwrite\fR, either of |
| 210 | which may be empty to leave that value at its default. |
| 211 | .TP |
Jens Axboe | 9183788 | 2008-02-05 12:02:07 +0100 | [diff] [blame] | 212 | .BI blocksize_range \fR=\fPirange[,irange] "\fR,\fB bsrange" \fR=\fPirange[,irange] |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 213 | Specify a range of I/O block sizes. The issued I/O unit will always be a |
| 214 | multiple of the minimum size, unless \fBblocksize_unaligned\fR is set. Applies |
Jens Axboe | 9183788 | 2008-02-05 12:02:07 +0100 | [diff] [blame] | 215 | to both reads and writes if only one range is given, but can be specified |
| 216 | seperately with a comma seperating the values. Example: bsrange=1k-4k,2k-8k. |
| 217 | Also (see \fBblocksize\fR). |
| 218 | .TP |
| 219 | .BI bssplit \fR=\fPstr |
| 220 | This option allows even finer grained control of the block sizes issued, |
| 221 | not just even splits between them. With this option, you can weight various |
| 222 | block sizes for exact control of the issued IO for a job that has mixed |
| 223 | block sizes. The format of the option is bssplit=blocksize/percentage, |
| 224 | optionally adding as many definitions as needed seperated by a colon. |
| 225 | Example: bssplit=4k/10:64k/50:32k/40 would issue 50% 64k blocks, 10% 4k |
Jens Axboe | c83cdd3 | 2009-04-24 14:23:59 +0200 | [diff] [blame] | 226 | blocks and 40% 32k blocks. \fBbssplit\fR also supports giving separate |
| 227 | splits to reads and writes. The format is identical to what the |
| 228 | \fBbs\fR option accepts, the read and write parts are separated with a |
| 229 | comma. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 230 | .TP |
| 231 | .B blocksize_unaligned\fR,\fP bs_unaligned |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 232 | If set, any size in \fBblocksize_range\fR may be used. This typically won't |
| 233 | work with direct I/O, as that normally requires sector alignment. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 234 | .TP |
Jens Axboe | 2b7a01d | 2009-03-11 11:00:13 +0100 | [diff] [blame] | 235 | .BI blockalign \fR=\fPint[,int] "\fR,\fB ba" \fR=\fPint[,int] |
Martin Steigerwald | 639ce0f | 2009-05-20 11:33:49 +0200 | [diff] [blame] | 236 | At what boundary to align random IO offsets. Defaults to the same as 'blocksize' |
| 237 | the minimum blocksize given. Minimum alignment is typically 512b |
Jens Axboe | 2b7a01d | 2009-03-11 11:00:13 +0100 | [diff] [blame] | 238 | for using direct IO, though it usually depends on the hardware block size. |
| 239 | This option is mutually exclusive with using a random map for files, so it |
| 240 | will turn off that option. |
Jens Axboe | 4360266 | 2009-03-14 20:08:47 +0100 | [diff] [blame] | 241 | .TP |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 242 | .B zero_buffers |
| 243 | Initialise buffers with all zeros. Default: fill buffers with random data. |
| 244 | .TP |
Jens Axboe | 901bb99 | 2009-03-14 20:17:36 +0100 | [diff] [blame] | 245 | .B refill_buffers |
| 246 | If this option is given, fio will refill the IO buffers on every submit. The |
| 247 | default is to only fill it at init time and reuse that data. Only makes sense |
| 248 | if zero_buffers isn't specified, naturally. If data verification is enabled, |
| 249 | refill_buffers is also automatically enabled. |
| 250 | .TP |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 251 | .BI nrfiles \fR=\fPint |
| 252 | Number of files to use for this job. Default: 1. |
| 253 | .TP |
| 254 | .BI openfiles \fR=\fPint |
| 255 | Number of files to keep open at the same time. Default: \fBnrfiles\fR. |
| 256 | .TP |
| 257 | .BI file_service_type \fR=\fPstr |
| 258 | Defines how files to service are selected. The following types are defined: |
| 259 | .RS |
| 260 | .RS |
| 261 | .TP |
| 262 | .B random |
| 263 | Choose a file at random |
| 264 | .TP |
| 265 | .B roundrobin |
| 266 | Round robin over open files (default). |
Jens Axboe | 6b7f685 | 2009-03-09 14:22:56 +0100 | [diff] [blame] | 267 | .B sequential |
| 268 | Do each file in the set sequentially. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 269 | .RE |
| 270 | .P |
| 271 | The number of I/Os to issue before switching a new file can be specified by |
| 272 | appending `:\fIint\fR' to the service type. |
| 273 | .RE |
| 274 | .TP |
| 275 | .BI ioengine \fR=\fPstr |
| 276 | Defines how the job issues I/O. The following types are defined: |
| 277 | .RS |
| 278 | .RS |
| 279 | .TP |
| 280 | .B sync |
| 281 | Basic \fIread\fR\|(2) or \fIwrite\fR\|(2) I/O. \fIfseek\fR\|(2) is used to |
| 282 | position the I/O location. |
| 283 | .TP |
gurudas pai | a31041e | 2007-10-23 15:12:30 +0200 | [diff] [blame] | 284 | .B psync |
| 285 | Basic \fIpread\fR\|(2) or \fIpwrite\fR\|(2) I/O. |
| 286 | .TP |
Jens Axboe | 9183788 | 2008-02-05 12:02:07 +0100 | [diff] [blame] | 287 | .B vsync |
| 288 | Basic \fIreadv\fR\|(2) or \fIwritev\fR\|(2) I/O. Will emulate queuing by |
| 289 | coalescing adjacents IOs into a single submission. |
| 290 | .TP |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 291 | .B libaio |
| 292 | Linux native asynchronous I/O. |
| 293 | .TP |
| 294 | .B posixaio |
| 295 | glibc POSIX asynchronous I/O using \fIaio_read\fR\|(3) and \fIaio_write\fR\|(3). |
| 296 | .TP |
| 297 | .B mmap |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 298 | File is memory mapped with \fImmap\fR\|(2) and data copied using |
| 299 | \fImemcpy\fR\|(3). |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 300 | .TP |
| 301 | .B splice |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 302 | \fIsplice\fR\|(2) is used to transfer the data and \fIvmsplice\fR\|(2) to |
| 303 | transfer data from user-space to the kernel. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 304 | .TP |
| 305 | .B syslet-rw |
| 306 | Use the syslet system calls to make regular read/write asynchronous. |
| 307 | .TP |
| 308 | .B sg |
| 309 | SCSI generic sg v3 I/O. May be either synchronous using the SG_IO ioctl, or if |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 310 | the target is an sg character device, we use \fIread\fR\|(2) and |
| 311 | \fIwrite\fR\|(2) for asynchronous I/O. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 312 | .TP |
| 313 | .B null |
| 314 | Doesn't transfer any data, just pretends to. Mainly used to exercise \fBfio\fR |
| 315 | itself and for debugging and testing purposes. |
| 316 | .TP |
| 317 | .B net |
| 318 | Transfer over the network. \fBfilename\fR must be set appropriately to |
| 319 | `\fIhost\fR/\fIport\fR' regardless of data direction. If receiving, only the |
| 320 | \fIport\fR argument is used. |
| 321 | .TP |
| 322 | .B netsplice |
| 323 | Like \fBnet\fR, but uses \fIsplice\fR\|(2) and \fIvmsplice\fR\|(2) to map data |
| 324 | and send/receive. |
| 325 | .TP |
gurudas pai | 53aec0a | 2007-10-05 13:20:18 +0200 | [diff] [blame] | 326 | .B cpuio |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 327 | Doesn't transfer any data, but burns CPU cycles according to \fBcpuload\fR and |
| 328 | \fBcpucycles\fR parameters. |
| 329 | .TP |
| 330 | .B guasi |
| 331 | The GUASI I/O engine is the Generic Userspace Asynchronous Syscall Interface |
| 332 | approach to asycnronous I/O. |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 333 | .br |
| 334 | See <http://www.xmailserver.org/guasi\-lib.html>. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 335 | .TP |
| 336 | .B external |
| 337 | Loads an external I/O engine object file. Append the engine filename as |
| 338 | `:\fIenginepath\fR'. |
| 339 | .RE |
| 340 | .RE |
| 341 | .TP |
| 342 | .BI iodepth \fR=\fPint |
| 343 | Number of I/O units to keep in flight against the file. Default: 1. |
| 344 | .TP |
| 345 | .BI iodepth_batch \fR=\fPint |
| 346 | Number of I/Os to submit at once. Default: \fBiodepth\fR. |
| 347 | .TP |
Jens Axboe | 3ce9dca | 2009-06-10 08:55:21 +0200 | [diff] [blame] | 348 | .BI iodepth_batch_complete \fR=\fPint |
| 349 | This defines how many pieces of IO to retrieve at once. It defaults to 1 which |
| 350 | means that we'll ask for a minimum of 1 IO in the retrieval process from the |
| 351 | kernel. The IO retrieval will go on until we hit the limit set by |
| 352 | \fBiodepth_low\fR. If this variable is set to 0, then fio will always check for |
| 353 | completed events before queuing more IO. This helps reduce IO latency, at the |
| 354 | cost of more retrieval system calls. |
| 355 | .TP |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 356 | .BI iodepth_low \fR=\fPint |
| 357 | Low watermark indicating when to start filling the queue again. Default: |
| 358 | \fBiodepth\fR. |
| 359 | .TP |
| 360 | .BI direct \fR=\fPbool |
| 361 | If true, use non-buffered I/O (usually O_DIRECT). Default: false. |
| 362 | .TP |
| 363 | .BI buffered \fR=\fPbool |
| 364 | If true, use buffered I/O. This is the opposite of the \fBdirect\fR parameter. |
| 365 | Default: true. |
| 366 | .TP |
Jens Axboe | f7fa265 | 2009-03-09 14:20:20 +0100 | [diff] [blame] | 367 | .BI offset \fR=\fPint |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 368 | Offset in the file to start I/O. Data before the offset will not be touched. |
| 369 | .TP |
| 370 | .BI fsync \fR=\fPint |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 371 | How many I/Os to perform before issuing an \fBfsync\fR\|(2) of dirty data. If |
| 372 | 0, don't sync. Default: 0. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 373 | .TP |
Jens Axboe | 5f9099e | 2009-06-16 22:40:26 +0200 | [diff] [blame] | 374 | .BI fdatasync \fR=\fPint |
| 375 | Like \fBfsync\fR, but uses \fBfdatasync\fR\|(2) instead to only sync the |
| 376 | data parts of the file. Default: 0. |
| 377 | .TP |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 378 | .BI overwrite \fR=\fPbool |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 379 | If writing, setup the file first and do overwrites. Default: false. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 380 | .TP |
| 381 | .BI end_fsync \fR=\fPbool |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 382 | Sync file contents when job exits. Default: false. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 383 | .TP |
| 384 | .BI fsync_on_close \fR=\fPbool |
| 385 | If true, sync file contents on close. This differs from \fBend_fsync\fR in that |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 386 | it will happen on every close, not just at the end of the job. Default: false. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 387 | .TP |
| 388 | .BI rwmixcycle \fR=\fPint |
| 389 | How many milliseconds before switching between reads and writes for a mixed |
| 390 | workload. Default: 500ms. |
| 391 | .TP |
| 392 | .BI rwmixread \fR=\fPint |
| 393 | Percentage of a mixed workload that should be reads. Default: 50. |
| 394 | .TP |
| 395 | .BI rwmixwrite \fR=\fPint |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 396 | Percentage of a mixed workload that should be writes. If \fBrwmixread\fR and |
Jens Axboe | c35dd7a | 2009-06-10 08:39:16 +0200 | [diff] [blame] | 397 | \fBrwmixwrite\fR are given and do not sum to 100%, the latter of the two |
| 398 | overrides the first. This may interfere with a given rate setting, if fio is |
| 399 | asked to limit reads or writes to a certain rate. If that is the case, then |
| 400 | the distribution may be skewed. Default: 50. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 401 | .TP |
| 402 | .B norandommap |
| 403 | Normally \fBfio\fR will cover every block of the file when doing random I/O. If |
| 404 | this parameter is given, a new offset will be chosen without looking at past |
| 405 | I/O history. This parameter is mutually exclusive with \fBverify\fR. |
| 406 | .TP |
Jens Axboe | 3ce9dca | 2009-06-10 08:55:21 +0200 | [diff] [blame] | 407 | .B softrandommap |
| 408 | See \fBnorandommap\fR. If fio runs with the random block map enabled and it |
| 409 | fails to allocate the map, if this option is set it will continue without a |
| 410 | random block map. As coverage will not be as complete as with random maps, this |
| 411 | option is disabled by default. |
| 412 | .TP |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 413 | .BI nice \fR=\fPint |
| 414 | Run job with given nice value. See \fInice\fR\|(2). |
| 415 | .TP |
| 416 | .BI prio \fR=\fPint |
| 417 | Set I/O priority value of this job between 0 (highest) and 7 (lowest). See |
| 418 | \fIionice\fR\|(1). |
| 419 | .TP |
| 420 | .BI prioclass \fR=\fPint |
| 421 | Set I/O priority class. See \fIionice\fR\|(1). |
| 422 | .TP |
| 423 | .BI thinktime \fR=\fPint |
| 424 | Stall job for given number of microseconds between issuing I/Os. |
| 425 | .TP |
| 426 | .BI thinktime_spin \fR=\fPint |
| 427 | Pretend to spend CPU time for given number of microseconds, sleeping the rest |
| 428 | of the time specified by \fBthinktime\fR. Only valid if \fBthinktime\fR is set. |
| 429 | .TP |
| 430 | .BI thinktime_blocks \fR=\fPint |
| 431 | Number of blocks to issue before waiting \fBthinktime\fR microseconds. |
| 432 | Default: 1. |
| 433 | .TP |
| 434 | .BI rate \fR=\fPint |
Jens Axboe | c35dd7a | 2009-06-10 08:39:16 +0200 | [diff] [blame] | 435 | Cap bandwidth used by this job. The number is in bytes/sec, the normal postfix |
| 436 | rules apply. You can use \fBrate\fR=500k to limit reads and writes to 500k each, |
| 437 | or you can specify read and writes separately. Using \fBrate\fR=1m,500k would |
| 438 | limit reads to 1MB/sec and writes to 500KB/sec. Capping only reads or writes |
| 439 | can be done with \fBrate\fR=,500k or \fBrate\fR=500k,. The former will only |
| 440 | limit writes (to 500KB/sec), the latter will only limit reads. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 441 | .TP |
| 442 | .BI ratemin \fR=\fPint |
| 443 | Tell \fBfio\fR to do whatever it can to maintain at least the given bandwidth. |
Jens Axboe | c35dd7a | 2009-06-10 08:39:16 +0200 | [diff] [blame] | 444 | Failing to meet this requirement will cause the job to exit. The same format |
| 445 | as \fBrate\fR is used for read vs write separation. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 446 | .TP |
| 447 | .BI rate_iops \fR=\fPint |
Jens Axboe | c35dd7a | 2009-06-10 08:39:16 +0200 | [diff] [blame] | 448 | Cap the bandwidth to this number of IOPS. Basically the same as rate, just |
| 449 | specified independently of bandwidth. The same format as \fBrate\fR is used for |
| 450 | read vs write seperation. If \fBblocksize\fR is a range, the smallest block |
| 451 | size is used as the metric. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 452 | .TP |
| 453 | .BI rate_iops_min \fR=\fPint |
Jens Axboe | c35dd7a | 2009-06-10 08:39:16 +0200 | [diff] [blame] | 454 | If this rate of I/O is not met, the job will exit. The same format as \fBrate\fR |
| 455 | is used for read vs write seperation. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 456 | .TP |
| 457 | .BI ratecycle \fR=\fPint |
| 458 | Average bandwidth for \fBrate\fR and \fBratemin\fR over this number of |
| 459 | milliseconds. Default: 1000ms. |
| 460 | .TP |
| 461 | .BI cpumask \fR=\fPint |
| 462 | Set CPU affinity for this job. \fIint\fR is a bitmask of allowed CPUs the job |
| 463 | may run on. See \fBsched_setaffinity\fR\|(2). |
| 464 | .TP |
| 465 | .BI cpus_allowed \fR=\fPstr |
| 466 | Same as \fBcpumask\fR, but allows a comma-delimited list of CPU numbers. |
| 467 | .TP |
| 468 | .BI startdelay \fR=\fPint |
| 469 | Delay start of job for the specified number of seconds. |
| 470 | .TP |
| 471 | .BI runtime \fR=\fPint |
| 472 | Terminate processing after the specified number of seconds. |
| 473 | .TP |
| 474 | .B time_based |
| 475 | If given, run for the specified \fBruntime\fR duration even if the files are |
| 476 | completely read or written. The same workload will be repeated as many times |
| 477 | as \fBruntime\fR allows. |
| 478 | .TP |
Jens Axboe | 901bb99 | 2009-03-14 20:17:36 +0100 | [diff] [blame] | 479 | .BI ramp_time \fR=\fPint |
| 480 | If set, fio will run the specified workload for this amount of time before |
| 481 | logging any performance numbers. Useful for letting performance settle before |
| 482 | logging results, thus minimizing the runtime required for stable results. Note |
Jens Axboe | c35dd7a | 2009-06-10 08:39:16 +0200 | [diff] [blame] | 483 | that the \fBramp_time\fR is considered lead in time for a job, thus it will |
| 484 | increase the total runtime if a special timeout or runtime is specified. |
Jens Axboe | 901bb99 | 2009-03-14 20:17:36 +0100 | [diff] [blame] | 485 | .TP |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 486 | .BI invalidate \fR=\fPbool |
| 487 | Invalidate buffer-cache for the file prior to starting I/O. Default: true. |
| 488 | .TP |
| 489 | .BI sync \fR=\fPbool |
| 490 | Use synchronous I/O for buffered writes. For the majority of I/O engines, |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 491 | this means using O_SYNC. Default: false. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 492 | .TP |
| 493 | .BI iomem \fR=\fPstr "\fR,\fP mem" \fR=\fPstr |
| 494 | Allocation method for I/O unit buffer. Allowed values are: |
| 495 | .RS |
| 496 | .RS |
| 497 | .TP |
| 498 | .B malloc |
| 499 | Allocate memory with \fImalloc\fR\|(3). |
| 500 | .TP |
| 501 | .B shm |
| 502 | Use shared memory buffers allocated through \fIshmget\fR\|(2). |
| 503 | .TP |
| 504 | .B shmhuge |
| 505 | Same as \fBshm\fR, but use huge pages as backing. |
| 506 | .TP |
| 507 | .B mmap |
| 508 | Use \fImmap\fR\|(2) for allocation. Uses anonymous memory unless a filename |
| 509 | is given after the option in the format `:\fIfile\fR'. |
| 510 | .TP |
| 511 | .B mmaphuge |
| 512 | Same as \fBmmap\fR, but use huge files as backing. |
| 513 | .RE |
| 514 | .P |
| 515 | The amount of memory allocated is the maximum allowed \fBblocksize\fR for the |
| 516 | job multiplied by \fBiodepth\fR. For \fBshmhuge\fR or \fBmmaphuge\fR to work, |
| 517 | the system must have free huge pages allocated. \fBmmaphuge\fR also needs to |
Jens Axboe | 2e266ba | 2009-09-14 08:56:53 +0200 | [diff] [blame] | 518 | have hugetlbfs mounted, and \fIfile\fR must point there. At least on Linux, |
| 519 | huge pages must be manually allocated. See \fB/proc/sys/vm/nr_hugehages\fR |
| 520 | and the documentation for that. Normally you just need to echo an appropriate |
| 521 | number, eg echoing 8 will ensure that the OS has 8 huge pages ready for |
| 522 | use. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 523 | .RE |
| 524 | .TP |
Jens Axboe | d529ee1 | 2009-07-01 10:33:03 +0200 | [diff] [blame] | 525 | .BI iomem_align \fR=\fPint |
| 526 | This indiciates the memory alignment of the IO memory buffers. Note that the |
| 527 | given alignment is applied to the first IO unit buffer, if using \fBiodepth\fR |
| 528 | the alignment of the following buffers are given by the \fBbs\fR used. In |
| 529 | other words, if using a \fBbs\fR that is a multiple of the page sized in the |
| 530 | system, all buffers will be aligned to this value. If using a \fBbs\fR that |
| 531 | is not page aligned, the alignment of subsequent IO memory buffers is the |
| 532 | sum of the \fBiomem_align\fR and \fBbs\fR used. |
| 533 | .TP |
Jens Axboe | f7fa265 | 2009-03-09 14:20:20 +0100 | [diff] [blame] | 534 | .BI hugepage\-size \fR=\fPint |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 535 | Defines the size of a huge page. Must be at least equal to the system setting. |
Jens Axboe | b22989b | 2009-07-17 22:29:23 +0200 | [diff] [blame] | 536 | Should be a multiple of 1MB. Default: 4MB. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 537 | .TP |
| 538 | .B exitall |
| 539 | Terminate all jobs when one finishes. Default: wait for each job to finish. |
| 540 | .TP |
| 541 | .BI bwavgtime \fR=\fPint |
| 542 | Average bandwidth calculations over the given time in milliseconds. Default: |
| 543 | 500ms. |
| 544 | .TP |
| 545 | .BI create_serialize \fR=\fPbool |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 546 | If true, serialize file creation for the jobs. Default: true. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 547 | .TP |
| 548 | .BI create_fsync \fR=\fPbool |
| 549 | \fIfsync\fR\|(2) data file after creation. Default: true. |
| 550 | .TP |
Jens Axboe | 6b7f685 | 2009-03-09 14:22:56 +0100 | [diff] [blame] | 551 | .BI create_on_open \fR=\fPbool |
| 552 | If true, the files are not created until they are opened for IO by the job. |
| 553 | .TP |
Jens Axboe | e9f4847 | 2009-06-03 12:14:08 +0200 | [diff] [blame] | 554 | .BI pre_read \fR=\fPbool |
| 555 | If this is given, files will be pre-read into memory before starting the given |
| 556 | IO operation. This will also clear the \fR \fBinvalidate\fR flag, since it is |
Jens Axboe | 9c0d224 | 2009-07-01 12:26:28 +0200 | [diff] [blame] | 557 | pointless to pre-read and then drop the cache. This will only work for IO |
| 558 | engines that are seekable, since they allow you to read the same data |
| 559 | multiple times. Thus it will not work on eg network or splice IO. |
Jens Axboe | e9f4847 | 2009-06-03 12:14:08 +0200 | [diff] [blame] | 560 | .TP |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 561 | .BI unlink \fR=\fPbool |
| 562 | Unlink job files when done. Default: false. |
| 563 | .TP |
| 564 | .BI loops \fR=\fPint |
| 565 | Specifies the number of iterations (runs of the same workload) of this job. |
| 566 | Default: 1. |
| 567 | .TP |
| 568 | .BI do_verify \fR=\fPbool |
| 569 | Run the verify phase after a write phase. Only valid if \fBverify\fR is set. |
| 570 | Default: true. |
| 571 | .TP |
| 572 | .BI verify \fR=\fPstr |
| 573 | Method of verifying file contents after each iteration of the job. Allowed |
| 574 | values are: |
| 575 | .RS |
| 576 | .RS |
| 577 | .TP |
Jens Axboe | b892dc0 | 2009-09-05 20:37:35 +0200 | [diff] [blame] | 578 | .B md5 crc16 crc32 crc32c crc32c-intel crc64 crc7 sha256 sha512 sha1 |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 579 | Store appropriate checksum in the header of each block. |
| 580 | .TP |
| 581 | .B meta |
| 582 | Write extra information about each I/O (timestamp, block number, etc.). The |
| 583 | block number is verified. |
| 584 | .TP |
| 585 | .B pattern |
Radha Ramachandran | 0e92f87 | 2009-10-27 20:14:27 +0100 | [diff] [blame] | 586 | Fill I/O buffers with a specific pattern that is used to verify. If the pattern |
| 587 | is < 4bytes, it can either be a decimal or a hexadecimal number. If the pattern |
| 588 | is > 4bytes, currently, it can only be a hexadecimal pattern starting with |
| 589 | either "0x" or "0X". |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 590 | .TP |
| 591 | .B null |
| 592 | Pretend to verify. Used for testing internals. |
| 593 | .RE |
Jens Axboe | b892dc0 | 2009-09-05 20:37:35 +0200 | [diff] [blame] | 594 | |
| 595 | This option can be used for repeated burn-in tests of a system to make sure |
| 596 | that the written data is also correctly read back. If the data direction given |
| 597 | is a read or random read, fio will assume that it should verify a previously |
| 598 | written file. If the data direction includes any form of write, the verify will |
| 599 | be of the newly written data. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 600 | .RE |
| 601 | .TP |
| 602 | .BI verify_sort \fR=\fPbool |
| 603 | If true, written verify blocks are sorted if \fBfio\fR deems it to be faster to |
| 604 | read them back in a sorted manner. Default: true. |
| 605 | .TP |
Jens Axboe | f7fa265 | 2009-03-09 14:20:20 +0100 | [diff] [blame] | 606 | .BI verify_offset \fR=\fPint |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 607 | Swap the verification header with data somewhere else in the block before |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 608 | writing. It is swapped back before verifying. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 609 | .TP |
Jens Axboe | f7fa265 | 2009-03-09 14:20:20 +0100 | [diff] [blame] | 610 | .BI verify_interval \fR=\fPint |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 611 | Write the verification header for this number of bytes, which should divide |
| 612 | \fBblocksize\fR. Default: \fBblocksize\fR. |
| 613 | .TP |
| 614 | .BI verify_fatal \fR=\fPbool |
| 615 | If true, exit the job on the first observed verification failure. Default: |
| 616 | false. |
| 617 | .TP |
Jens Axboe | e8462bd | 2009-07-06 12:59:04 +0200 | [diff] [blame] | 618 | .BI verify_async \fR=\fPint |
| 619 | Fio will normally verify IO inline from the submitting thread. This option |
| 620 | takes an integer describing how many async offload threads to create for IO |
| 621 | verification instead, causing fio to offload the duty of verifying IO contents |
Jens Axboe | c85c324 | 2009-07-06 14:12:57 +0200 | [diff] [blame] | 622 | to one or more separate threads. If using this offload option, even sync IO |
| 623 | engines can benefit from using an \fBiodepth\fR setting higher than 1, as it |
| 624 | allows them to have IO in flight while verifies are running. |
Jens Axboe | e8462bd | 2009-07-06 12:59:04 +0200 | [diff] [blame] | 625 | .TP |
| 626 | .BI verify_async_cpus \fR=\fPstr |
| 627 | Tell fio to set the given CPU affinity on the async IO verification threads. |
| 628 | See \fBcpus_allowed\fP for the format used. |
| 629 | .TP |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 630 | .B stonewall |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 631 | Wait for preceeding jobs in the job file to exit before starting this one. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 632 | \fBstonewall\fR implies \fBnew_group\fR. |
| 633 | .TP |
| 634 | .B new_group |
| 635 | Start a new reporting group. If not given, all jobs in a file will be part |
| 636 | of the same reporting group, unless separated by a stonewall. |
| 637 | .TP |
| 638 | .BI numjobs \fR=\fPint |
| 639 | Number of clones (processes/threads performing the same workload) of this job. |
| 640 | Default: 1. |
| 641 | .TP |
| 642 | .B group_reporting |
| 643 | If set, display per-group reports instead of per-job when \fBnumjobs\fR is |
| 644 | specified. |
| 645 | .TP |
| 646 | .B thread |
| 647 | Use threads created with \fBpthread_create\fR\|(3) instead of processes created |
| 648 | with \fBfork\fR\|(2). |
| 649 | .TP |
Jens Axboe | f7fa265 | 2009-03-09 14:20:20 +0100 | [diff] [blame] | 650 | .BI zonesize \fR=\fPint |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 651 | Divide file into zones of the specified size in bytes. See \fBzoneskip\fR. |
| 652 | .TP |
Jens Axboe | f7fa265 | 2009-03-09 14:20:20 +0100 | [diff] [blame] | 653 | .BI zoneskip \fR=\fPint |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 654 | Skip the specified number of bytes when \fBzonesize\fR bytes of data have been |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 655 | read. |
| 656 | .TP |
| 657 | .BI write_iolog \fR=\fPstr |
| 658 | Write the issued I/O patterns to the specified file. |
| 659 | .TP |
| 660 | .BI read_iolog \fR=\fPstr |
| 661 | Replay the I/O patterns contained in the specified file generated by |
| 662 | \fBwrite_iolog\fR, or may be a \fBblktrace\fR binary file. |
| 663 | .TP |
Jens Axboe | 901bb99 | 2009-03-14 20:17:36 +0100 | [diff] [blame] | 664 | .B write_bw_log \fR=\fPstr |
| 665 | If given, write a bandwidth log of the jobs in this job file. Can be used to |
| 666 | store data of the bandwidth of the jobs in their lifetime. The included |
| 667 | fio_generate_plots script uses gnuplot to turn these text files into nice |
| 668 | graphs. See \fBwrite_log_log\fR for behaviour of given filename. For this |
| 669 | option, the postfix is _bw.log. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 670 | .TP |
| 671 | .B write_lat_log |
Jens Axboe | 901bb99 | 2009-03-14 20:17:36 +0100 | [diff] [blame] | 672 | Same as \fBwrite_bw_log\fR, but writes I/O completion latencies. If no |
| 673 | filename is given with this option, the default filename of "jobname_type.log" |
| 674 | is used. Even if the filename is given, fio will still append the type of log. |
| 675 | .TP |
| 676 | .B disable_clat \fR=\fPbool |
| 677 | Disable measurements of completion latency numbers. Useful only for cutting |
| 678 | back the number of calls to gettimeofday, as that does impact performance at |
| 679 | really high IOPS rates. Note that to really get rid of a large amount of these |
| 680 | calls, this option must be used with disable_slat and disable_bw as well. |
| 681 | .TP |
| 682 | .B disable_slat \fR=\fPbool |
| 683 | Disable measurements of submission latency numbers. See \fBdisable_clat\fR. |
| 684 | .TP |
| 685 | .B disable_bw_measurement \fR=\fPbool |
| 686 | Disable measurements of throughput/bandwidth numbers. See \fBdisable_clat\fR. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 687 | .TP |
Jens Axboe | f7fa265 | 2009-03-09 14:20:20 +0100 | [diff] [blame] | 688 | .BI lockmem \fR=\fPint |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 689 | Pin the specified amount of memory with \fBmlock\fR\|(2). Can be used to |
| 690 | simulate a smaller amount of memory. |
| 691 | .TP |
| 692 | .BI exec_prerun \fR=\fPstr |
| 693 | Before running the job, execute the specified command with \fBsystem\fR\|(3). |
| 694 | .TP |
| 695 | .BI exec_postrun \fR=\fPstr |
| 696 | Same as \fBexec_prerun\fR, but the command is executed after the job completes. |
| 697 | .TP |
| 698 | .BI ioscheduler \fR=\fPstr |
| 699 | Attempt to switch the device hosting the file to the specified I/O scheduler. |
| 700 | .TP |
| 701 | .BI cpuload \fR=\fPint |
| 702 | If the job is a CPU cycle-eater, attempt to use the specified percentage of |
| 703 | CPU cycles. |
| 704 | .TP |
| 705 | .BI cpuchunks \fR=\fPint |
| 706 | If the job is a CPU cycle-eater, split the load into cycles of the |
| 707 | given time in milliseconds. |
| 708 | .TP |
| 709 | .BI disk_util \fR=\fPbool |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 710 | Generate disk utilization statistics if the platform supports it. Default: true. |
Jens Axboe | 901bb99 | 2009-03-14 20:17:36 +0100 | [diff] [blame] | 711 | .TP |
| 712 | .BI gtod_reduce \fR=\fPbool |
| 713 | Enable all of the gettimeofday() reducing options (disable_clat, disable_slat, |
| 714 | disable_bw) plus reduce precision of the timeout somewhat to really shrink the |
| 715 | gettimeofday() call count. With this option enabled, we only do about 0.4% of |
| 716 | the gtod() calls we would have done if all time keeping was enabled. |
| 717 | .TP |
| 718 | .BI gtod_cpu \fR=\fPint |
| 719 | Sometimes it's cheaper to dedicate a single thread of execution to just getting |
| 720 | the current time. Fio (and databases, for instance) are very intensive on |
| 721 | gettimeofday() calls. With this option, you can set one CPU aside for doing |
| 722 | nothing but logging current time to a shared memory location. Then the other |
| 723 | threads/processes that run IO workloads need only copy that segment, instead of |
| 724 | entering the kernel with a gettimeofday() call. The CPU set aside for doing |
| 725 | these time calls will be excluded from other uses. Fio will manually clear it |
| 726 | from the CPU mask of other jobs. |
Radha Ramachandran | f2bba18 | 2009-06-15 08:40:16 +0200 | [diff] [blame] | 727 | .TP |
Jens Axboe | a696fa2 | 2009-12-04 10:05:02 +0100 | [diff] [blame] | 728 | .BI cgroup_root \fR=\fPstr |
| 729 | Root of the mounted blkio cgroup file systems. This is a Linux |
| 730 | specific IO controller. If your system doesn't have it mounted, |
| 731 | you can do so with: |
| 732 | |
| 733 | # mount -t cgroup -o blkio none /cgroup |
| 734 | |
| 735 | The cgroup_root defaults to /cgroup, if mounted elsewhere please specify this |
| 736 | option. |
| 737 | .TP |
| 738 | .BI cgroup \fR=\fPstr |
| 739 | Add job to this control group. If it doesn't exist, it will be created. |
| 740 | .TP |
| 741 | .BI cgroup_weight \fR=\fPint |
| 742 | Set the weight of the cgroup to this value. See the documentation that comes |
| 743 | with the kernel, allowed values are in the range of 100..1000. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 744 | .SH OUTPUT |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 745 | While running, \fBfio\fR will display the status of the created jobs. For |
| 746 | example: |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 747 | .RS |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 748 | .P |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 749 | Threads: 1: [_r] [24.8% done] [ 13509/ 8334 kb/s] [eta 00h:01m:31s] |
| 750 | .RE |
| 751 | .P |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 752 | The characters in the first set of brackets denote the current status of each |
| 753 | threads. The possible values are: |
| 754 | .P |
| 755 | .PD 0 |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 756 | .RS |
| 757 | .TP |
| 758 | .B P |
| 759 | Setup but not started. |
| 760 | .TP |
| 761 | .B C |
| 762 | Thread created. |
| 763 | .TP |
| 764 | .B I |
| 765 | Initialized, waiting. |
| 766 | .TP |
| 767 | .B R |
| 768 | Running, doing sequential reads. |
| 769 | .TP |
| 770 | .B r |
| 771 | Running, doing random reads. |
| 772 | .TP |
| 773 | .B W |
| 774 | Running, doing sequential writes. |
| 775 | .TP |
| 776 | .B w |
| 777 | Running, doing random writes. |
| 778 | .TP |
| 779 | .B M |
| 780 | Running, doing mixed sequential reads/writes. |
| 781 | .TP |
| 782 | .B m |
| 783 | Running, doing mixed random reads/writes. |
| 784 | .TP |
| 785 | .B F |
| 786 | Running, currently waiting for \fBfsync\fR\|(2). |
| 787 | .TP |
| 788 | .B V |
| 789 | Running, verifying written data. |
| 790 | .TP |
| 791 | .B E |
| 792 | Exited, not reaped by main thread. |
| 793 | .TP |
| 794 | .B \- |
| 795 | Exited, thread reaped. |
| 796 | .RE |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 797 | .PD |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 798 | .P |
| 799 | The second set of brackets shows the estimated completion percentage of |
| 800 | the current group. The third set shows the read and write I/O rate, |
| 801 | respectively. Finally, the estimated run time of the job is displayed. |
| 802 | .P |
| 803 | When \fBfio\fR completes (or is interrupted by Ctrl-C), it will show data |
| 804 | for each thread, each group of threads, and each disk, in that order. |
| 805 | .P |
| 806 | Per-thread statistics first show the threads client number, group-id, and |
| 807 | error code. The remaining figures are as follows: |
| 808 | .RS |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 809 | .TP |
| 810 | .B io |
| 811 | Number of megabytes of I/O performed. |
| 812 | .TP |
| 813 | .B bw |
| 814 | Average data rate (bandwidth). |
| 815 | .TP |
| 816 | .B runt |
| 817 | Threads run time. |
| 818 | .TP |
| 819 | .B slat |
| 820 | Submission latency minimum, maximum, average and standard deviation. This is |
| 821 | the time it took to submit the I/O. |
| 822 | .TP |
| 823 | .B clat |
| 824 | Completion latency minimum, maximum, average and standard deviation. This |
| 825 | is the time between submission and completion. |
| 826 | .TP |
| 827 | .B bw |
| 828 | Bandwidth minimum, maximum, percentage of aggregate bandwidth received, average |
| 829 | and standard deviation. |
| 830 | .TP |
| 831 | .B cpu |
| 832 | CPU usage statistics. Includes user and system time, number of context switches |
| 833 | this thread went through and number of major and minor page faults. |
| 834 | .TP |
| 835 | .B IO depths |
| 836 | Distribution of I/O depths. Each depth includes everything less than (or equal) |
| 837 | to it, but greater than the previous depth. |
| 838 | .TP |
| 839 | .B IO issued |
| 840 | Number of read/write requests issued, and number of short read/write requests. |
| 841 | .TP |
| 842 | .B IO latencies |
| 843 | Distribution of I/O completion latencies. The numbers follow the same pattern |
| 844 | as \fBIO depths\fR. |
| 845 | .RE |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 846 | .P |
| 847 | The group statistics show: |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 848 | .PD 0 |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 849 | .RS |
| 850 | .TP |
| 851 | .B io |
| 852 | Number of megabytes I/O performed. |
| 853 | .TP |
| 854 | .B aggrb |
| 855 | Aggregate bandwidth of threads in the group. |
| 856 | .TP |
| 857 | .B minb |
| 858 | Minimum average bandwidth a thread saw. |
| 859 | .TP |
| 860 | .B maxb |
| 861 | Maximum average bandwidth a thread saw. |
| 862 | .TP |
| 863 | .B mint |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 864 | Shortest runtime of threads in the group. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 865 | .TP |
| 866 | .B maxt |
| 867 | Longest runtime of threads in the group. |
| 868 | .RE |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 869 | .PD |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 870 | .P |
| 871 | Finally, disk statistics are printed with reads first: |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 872 | .PD 0 |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 873 | .RS |
| 874 | .TP |
| 875 | .B ios |
| 876 | Number of I/Os performed by all groups. |
| 877 | .TP |
| 878 | .B merge |
| 879 | Number of merges in the I/O scheduler. |
| 880 | .TP |
| 881 | .B ticks |
| 882 | Number of ticks we kept the disk busy. |
| 883 | .TP |
| 884 | .B io_queue |
| 885 | Total time spent in the disk queue. |
| 886 | .TP |
| 887 | .B util |
| 888 | Disk utilization. |
| 889 | .RE |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 890 | .PD |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 891 | .SH TERSE OUTPUT |
| 892 | If the \fB\-\-minimal\fR option is given, the results will be printed in a |
| 893 | semicolon-delimited format suitable for scripted use. The fields are: |
| 894 | .P |
| 895 | .RS |
| 896 | .B jobname, groupid, error |
| 897 | .P |
| 898 | Read status: |
| 899 | .RS |
Jens Axboe | b22989b | 2009-07-17 22:29:23 +0200 | [diff] [blame] | 900 | .B KB I/O, bandwidth \fR(KB/s)\fP, runtime \fR(ms)\fP |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 901 | .P |
| 902 | Submission latency: |
| 903 | .RS |
| 904 | .B min, max, mean, standard deviation |
| 905 | .RE |
| 906 | Completion latency: |
| 907 | .RS |
| 908 | .B min, max, mean, standard deviation |
| 909 | .RE |
| 910 | Bandwidth: |
| 911 | .RS |
| 912 | .B min, max, aggregate percentage of total, mean, standard deviation |
| 913 | .RE |
| 914 | .RE |
| 915 | .P |
| 916 | Write status: |
| 917 | .RS |
Jens Axboe | b22989b | 2009-07-17 22:29:23 +0200 | [diff] [blame] | 918 | .B KB I/O, bandwidth \fR(KB/s)\fP, runtime \fR(ms)\fP |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 919 | .P |
| 920 | Submission latency: |
| 921 | .RS |
| 922 | .B min, max, mean, standard deviation |
| 923 | .RE |
| 924 | Completion latency: |
| 925 | .RS |
| 926 | .B min, max, mean, standard deviation |
| 927 | .RE |
| 928 | Bandwidth: |
| 929 | .RS |
| 930 | .B min, max, aggregate percentage of total, mean, standard deviation |
| 931 | .RE |
| 932 | .RE |
| 933 | .P |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 934 | CPU usage: |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 935 | .RS |
Carl Henrik Lunde | bd2626f | 2008-06-12 09:17:46 +0200 | [diff] [blame] | 936 | .B user, system, context switches, major page faults, minor page faults |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 937 | .RE |
| 938 | .P |
| 939 | IO depth distribution: |
| 940 | .RS |
| 941 | .B <=1, 2, 4, 8, 16, 32, >=64 |
| 942 | .RE |
| 943 | .P |
| 944 | IO latency distribution (ms): |
| 945 | .RS |
| 946 | .B <=2, 4, 10, 20, 50, 100, 250, 500, 750, 1000, >=2000 |
| 947 | .RE |
| 948 | .P |
| 949 | .B text description |
| 950 | .RE |
| 951 | .SH AUTHORS |
| 952 | .B fio |
| 953 | was written by Jens Axboe <jens.axboe@oracle.com>. |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 954 | .br |
| 955 | This man page was written by Aaron Carroll <aaronc@cse.unsw.edu.au> based |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 956 | on documentation by Jens Axboe. |
| 957 | .SH "REPORTING BUGS" |
Jens Axboe | 482900c | 2009-06-02 12:15:51 +0200 | [diff] [blame] | 958 | Report bugs to the \fBfio\fR mailing list <fio@vger.kernel.org>. |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 959 | See \fBREADME\fR. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 960 | .SH "SEE ALSO" |
Aaron Carroll | d1429b5 | 2007-09-18 08:10:57 +0200 | [diff] [blame] | 961 | For further documentation see \fBHOWTO\fR and \fBREADME\fR. |
| 962 | .br |
| 963 | Sample jobfiles are available in the \fBexamples\fR directory. |
Aaron Carroll | d60e92d | 2007-09-17 10:32:59 +0200 | [diff] [blame] | 964 | |