1bonnie++(8) System Manager's Manual bonnie++(8)
2
6 bonnie++ - program to test hard drive performance.
7
10 bonnie++ [-d dir] [-c concurrency] [-s size(MiB)[:chunk-size(b)]] [-n
11 number-to-stat(*1024)[:max-size[:min-size][:num-directo‐
12 ries[:chunk-size]]]] [-m machine-name] [-r ram-size-in-MiB] [-x num‐
13 ber-of-tests] [-u uid-to-use:gid-to-use] [-g gid-to-use] [-q] [-f
14 size-for-char-io] [-b] [-D] [-p processes | -y p|s ] [-z seed-num|-Z
15 random-file]
16
19 This manual page documents briefly the bonnie++, program.
20 Bonnie++ is a program to test hard drives and file systems for perfor‐
22 mance or the lack therof. There are a many different types of file sys‐
23 tem operations which different applications use to different degrees.
24 Bonnie++ tests some of them and for each test gives a result of the
25 amount of work done per second and the percentage of CPU time this
26 took. For performance results higher numbers are better, for CPU usage
27 and latency lower are better (NB a configuration scoring a performance
28 result of 2000 and a CPU result of 90% is better in terms of CPU use
29 than a configuration delivering performance of 1000 and CPU usage of
30 60%).
31 There are two sections to the program's operations. The first is to
33 test the IO throughput in a fashion that is designed to simulate some
34 types of database applications. The second is to test creation, read‐
35 ing, and deleting many small files in a fashion similar to the usage
36 patterns of programs such as Squid or INN.
37 All the details of the tests performed by Bonnie++ are contained in the
39 file /usr/share/doc/bonnie++/readme.html
40
43 For Bonnie++ every option is of the form of a hyphen followed by a let‐
44 ter and then the next parameter contains the value.
45 -d the directory to use for the tests.
47 -c the level of concurrency (default 1). The number of copies of
49 the test to be performed at the same time.
50 -s the size of the file(s) for IO performance measures in
52 megabytes. If the size is greater than 1G then multiple files
53 will be used to store the data, and each file will be up to 1G
54 in size. This parameter may include the chunk size seperated
55 from the size by a colon. The chunk-size is measured in bytes
56 and must be a power of two from 256 to 1048576, the default is
57 8192. NB You can specify the size in giga-bytes or the
58 chunk-size in kilo-bytes if you add g or k to the end of the
59 number respectively.
60 If the specified size is 0 then this test will be skipped.
62 -n the number of files for the file creation test. This is measured
64 in multiples of 1024 files. This is because no-one will want to
65 test less than 1024 files, and we need the extra space on
66 braille displays.
67 If the specified number is 0 then this test will be skipped.
69 The default for this test is to test with 0 byte files. To use
71 files of other sizes you can specify number:max:min:num-directo‐
72 ries:chunk-size where max is the maximum size and min is the
73 minimum size (both default to 0 if not specified). If minimum
74 and maximum sizes are specified then every file will have a ran‐
75 dom size from the range min..max inclusive. If you specify a
76 number of directories then the files will be evenly distributed
77 amoung that many sub-directories.
78 If max is -1 then hard links will be created instead of files.
80 If max is -2 then soft links will be created instead of files.
81 -m name of the machine - for display purposes only.
85 -r RAM size in megabytes. If you specify this the other parameters
88 will be checked to ensure they make sense for a machine of that
89 much RAM. You should not need to do this in general use as it
90 should be able to discover the RAM size. NB If you specify a
91 size of 0 then all checks will be disabled...
92 -x number of test runs. This is useful if you want to perform more
95 than one test. It will dump output continuously in CSV format
96 until either the number of tests have been completed, or it is
97 killed.
98 -u user-id to use. When running as root specify the UID to use for
101 the tests. It is not recommended to use root (since the occa‐
102 sion when a Bonnie++ bug wiped out someone's system), so if you
103 really want to run as root then use -u root. Also if you want
104 to specify the group to run as then use the user:group format.
105 If you specify a user by name but no group then the primary
106 group of that user will be chosen. If you specify a user by
107 number and no group then the group will be nogroup.
108 -g group-id to use. Same as using :group for the -u parameter,
111 just a different way to specify it for compatibility with other
112 programs.
113 -q quiet mode. If specified then some of the extra informational
116 messages will be suppressed. Also the csv data will be the only
117 output on standard out and the plain text data will be on stan‐
118 dard error. This means you can run bonnie++ -q >> file.csv to
119 record your csv data.
120 -f size-for-char-io
123 fast mode control, skips per-char IO tests if no parameter, oth‐
124 erwise specifies the size of the tests for per-char IO tests
125 (default 20M).
126 -b no write buffering. fsync() after every write.
129 -p number of processes to serve semaphores for. This is used to
132 create the semaphores for synchronising multiple Bonnie++ pro‐
133 cesses. All the processes which are told to use the semaphore
134 with -ys will start each test with synchronization. Use "-p -1"
135 to delete the semaphore.
136 -y s|p perform synchronization before each test. Option s for sema‐
139 phores and option p for prompting.
140 -D use direct IO (O_DIRECT) for the bulk IO tests
143 -z seed
146 random number seed to repeat the same test.
147 -Z random-file
150 file containing random data in network byte order.
151
153 Run the following commands to run three copies of Bonnie++ simultane‐
154 ously:
155 bonnie++ -p3
157 bonnie++ -y > out1 &
159 bonnie++ -y > out2 &
161 bonnie++ -y > out3 &
163
166 The primary output is plain-text in 80 columns which is designed to fit
167 well when pasted into email and which will work well with Braille dis‐
168 plays.
169 The second type of output is CSV (Comma Seperated Values). This can
171 easily be imported into any spread-sheet or database program. Also I
172 have included the programs bon_csv2html and bon_csv2txt to convert CSV
173 data to HTML and plain-ascii respectively.
174 For every test two numbers are reported, the amount of work done
176 (higher numbers are better) and the percentage of CPU time taken to
177 perform the work (lower numbers are better). If a test completes in
178 less than 500ms then the output will be displayed as "++++". This is
179 because such a test result can't be calculated accurately due to round‐
180 ing errors and I would rather display no result than a wrong result.
181 Data volumes for the 80 column text display use "K" for KiB (1024
183 bytes), "M" for MiB (1048576 bytes), and "G" for GiB (1073741824
184 bytes). So K/sec means a multiple of 1024 bytes per second.
185
188 This program, its manual page, and the Debian package were written by
189 Russell Coker <russell@coker.com.au>, parts of the program are based on
190 the work of Tim Bray <tbray@textuality.com>.
191 The documentation, the Perl scripts, and all the code for testing the
193 creation of thousands of files was written by Russell Coker, but the
194 entire package is under joint copyright with Tim Bray.
195
198 Handles SIGINT and does a cleanup (which may take some time), a second
199 SIGINT or a SIGQUIT will cause it to immediately die.
200 SIGXCPU and SIGXFSZ act like SIGINT.
202 Ignores SIGHUP.
204
208 The random file sizes will add up to different values for different
209 test runs. I plan to add some code that checks the sum and ensures
210 that the sum of the values will be the same on seperate runs.
211
214 The source is available from http://www.coker.com.au/bonnie++ .
215 See http://etbe.coker.com.au/category/benchmark for further informa‐
217 tion.
218
221 zcav(8), getc_putc(8), bon_csv2html(1), bon_csv2txt(1)
222 bonnie++(8)