[PATCH] Support for splice IO engine
[fio.git] / README
CommitLineData
ebac4655
JA
1fio
2---
3
4fio is a tool that will spawn a number of thread doing a particular
5type of io action as specified by the user. fio takes a number of
6global parameters, each inherited by the thread unless otherwise
7parameters given to them overriding that setting is given.
8
2b02b546
JA
9
10Source
11------
12
13fio resides in a git repo, the canonical place is:
14
15git://brick.kernel.dk/data/git/fio.git
16
17Snapshots are frequently generated as well and they include the git
18meta data as well. You can download them here:
19
20http://brick.kernel.dk/snaps/
21
22
ebac4655
JA
23Options
24-------
25
26$ fio
27 -s IO is sequential
28 -b block size in KiB for each io
29 -t <sec> Runtime in seconds
30 -r For random io, sequence must be repeatable
31 -R <on> If one thread fails to meet rate, quit all
32 -o <on> Use direct IO is 1, buffered if 0
33 -l Generate per-job latency logs
34 -w Generate per-job bandwidth logs
35 -f <file> Read <file> for job descriptions
4785f995 36 -h Print help info
ebac4655
JA
37 -v Print version information and exit
38
39The <jobs> format is as follows:
40
41 directory=x Use 'x' as the top level directory for storing files
3d60d1ed
JA
42 rw=x 'x' may be: read, randread, write, randwrite,
43 rw (read-write mix), randrw (read-write random mix)
ebac4655
JA
44 size=x Set file size to x bytes (x string can include k/m/g)
45 ioengine=x 'x' may be: aio/libaio/linuxaio for Linux aio,
46 posixaio for POSIX aio, sync for regular read/write io,
8756e4d4
JA
47 mmap for mmap'ed io, splice for using splice/vmsplice,
48 or sgio for direct SG_IO io. The latter only works on
49 Linux on SCSI (or SCSI-like devices, such as
50 usb-storage or sata/libata driven) devices.
ebac4655
JA
51 iodepth=x For async io, allow 'x' ios in flight
52 overwrite=x If 'x', layout a write file first.
53 prio=x Run io at prio X, 0-7 is the kernel allowed range
54 prioclass=x Run io at prio class X
55 bs=x Use 'x' for thread blocksize. May include k/m postfix.
56 bsrange=x-y Mix thread block sizes randomly between x and y. May
57 also include k/m postfix.
58 direct=x 1 for direct IO, 0 for buffered IO
59 thinktime=x "Think" x usec after each io
60 rate=x Throttle rate to x KiB/sec
61 ratemin=x Quit if rate of x KiB/sec can't be met
62 ratecycle=x ratemin averaged over x msecs
63 cpumask=x Only allow job to run on CPUs defined by mask.
64 fsync=x If writing, fsync after every x blocks have been written
65 startdelay=x Start this thread x seconds after startup
66 timeout=x Terminate x seconds after startup
67 offset=x Start io at offset x (x string can include k/m/g)
68 invalidate=x Invalidate page cache for file prior to doing io
69 sync=x Use sync writes if x and writing
70 mem=x If x == malloc, use malloc for buffers. If x == shm,
71 use shm for buffers. If x == mmap, use anon mmap.
72 exitall When one thread quits, terminate the others
73 bwavgtime=x Average bandwidth stats over an x msec window.
74 create_serialize=x If 'x', serialize file creation.
75 create_fsync=x If 'x', run fsync() after file creation.
76 loops=x Run the job 'x' number of times.
77 verify=x If 'x' == md5, use md5 for verifies. If 'x' == crc32,
78 use crc32 for verifies. md5 is 'safer', but crc32 is
79 a lot faster. Only makes sense for writing to a file.
80 stonewall Wait for preceeding jobs to end before running.
81 numjobs=x Create 'x' similar entries for this job
82 thread Use pthreads instead of forked jobs
20dc95c4
JA
83 zonesize=x
84 zoneskip=y Zone options must be paired. If given, the job
85 will skip y bytes for every x read/written. This
86 can be used to gauge hard drive speed over the entire
87 platter, without reading everything. Both x/y can
88 include k/m/g suffix.
aea47d44
JA
89 iolog=x Open and read io pattern from file 'x'. The file must
90 contain one io action per line in the following format:
91 rw, offset, length
92 where with rw=0/1 for read/write, and the offset
93 and length entries being in bytes.
ebac4655
JA
94
95
96Examples using a job file
97-------------------------
98
99A sample job file doing the same as above would look like this:
100
101[read_file]
102rw=0
103bs=4096
104
105[write_file]
106rw=1
107bs=16384
108
109And fio would be invoked as:
110
111$ fio -o1 -s -f file_with_above
112
113The second example would look like this:
114
115[rf1]
116rw=0
117prio=6
118
119[rf2]
120rw=0
121prio=3
122
123[rf3]
124rw=0
125prio=0
126direct=1
127
128And fio would be invoked as:
129
130$ fio -o0 -s -b4096 -f file_with_above
131
132'global' is a reserved keyword. When used as the filename, it sets the
133default options for the threads following that section. It is possible
134to have more than one global section in the file, as it only affects
135subsequent jobs.
136
137Also see the examples/ dir for sample job files.
138
139
140Interpreting the output
141-----------------------
142
143fio spits out a lot of output. While running, fio will display the
144status of the jobs created. An example of that would be:
145
146Threads now running: 2 : [ww] [5.73% done]
147
148The characters inside the square brackets denote the current status of
149each thread. The possible values (in typical life cycle order) are:
150
151Idle Run
152---- ---
153P Thread setup, but not started.
154C Thread created and running, but not doing anything yet
155 R Running, doing sequential reads.
156 r Running, doing random reads.
157 W Running, doing sequential writes.
158 w Running, doing random writes.
159V Running, doing verification of written data.
160E Thread exited, not reaped by main thread yet.
161_ Thread reaped.
162
163The other values are fairly self explanatory - number of thread currently
164running and doing io, and the estimated completion percentage.
165
166When fio is done (or interrupted by ctrl-c), it will show the data for
167each thread, group of threads, and disks in that order. For each data
168direction, the output looks like:
169
170Client1 (g=0): err= 0:
171 write: io= 32MiB, bw= 666KiB/s, runt= 50320msec
172 slat (msec): min= 0, max= 136, avg= 0.03, dev= 1.92
173 clat (msec): min= 0, max= 631, avg=48.50, dev=86.82
174 bw (KiB/s) : min= 0, max= 1196, per=51.00%, avg=664.02, dev=681.68
175 cpu : usr=1.49%, sys=0.25%, ctx=7969
176
177The client number is printed, along with the group id and error of that
178thread. Below is the io statistics, here for writes. In the order listed,
179they denote:
180
181io= Number of megabytes io performed
182bw= Average bandwidth rate
183runt= The runtime of that thread
184 slat= Submission latency (avg being the average, dev being the
185 standard deviation). This is the time it took to submit
186 the io. For sync io, the slat is really the completion
187 latency, since queue/complete is one operation there.
188 clat= Completion latency. Same names as slat, this denotes the
189 time from submission to completion of the io pieces. For
190 sync io, clat will usually be equal (or very close) to 0,
191 as the time from submit to complete is basically just
192 CPU time (io has already been done, see slat explanation).
193 bw= Bandwidth. Same names as the xlat stats, but also includes
194 an approximate percentage of total aggregate bandwidth
195 this thread received in this group. This last value is
196 only really useful if the threads in this group are on the
197 same disk, since they are then competing for disk access.
198cpu= CPU usage. User and system time, along with the number
199 of context switches this thread went through.
200
201After each client has been listed, the group statistics are printed. They
202will look like this:
203
204Run status group 0 (all jobs):
205 READ: io=64MiB, aggrb=22178, minb=11355, maxb=11814, mint=2840msec, maxt=2955msec
206 WRITE: io=64MiB, aggrb=1302, minb=666, maxb=669, mint=50093msec, maxt=50320msec
207
208For each data direction, it prints:
209
210io= Number of megabytes io performed.
211aggrb= Aggregate bandwidth of threads in this group.
212minb= The minimum average bandwidth a thread saw.
213maxb= The maximum average bandwidth a thread saw.
214mint= The minimum runtime of a thread.
215maxt= The maximum runtime of a thread.
216
217And finally, the disk statistics are printed. They will look like this:
218
219Disk stats (read/write):
220 sda: ios=16398/16511, merge=30/162, ticks=6853/819634, in_queue=826487, util=100.00%
221
222Each value is printed for both reads and writes, with reads first. The
223numbers denote:
224
225ios= Number of ios performed by all groups.
226merge= Number of merges io the io scheduler.
227ticks= Number of ticks we kept the disk busy.
228io_queue= Total time spent in the disk queue.
229util= The disk utilization. A value of 100% means we kept the disk
230 busy constantly, 50% would be a disk idling half of the time.