File: blkreplay.pod

package info (click to toggle)
blkreplay 1.0-3.1
  • links: PTS
  • area: main
  • in suites: bookworm, bullseye, sid
  • size: 25,672 kB
  • sloc: sh: 3,175; ansic: 2,057; makefile: 23
file content (203 lines) | stat: -rw-r--r-- 3,434 bytes parent folder | download | duplicates (2)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
=encoding utf8

=head1 NAME

blkreplay - block device testing and benchmarking tool

=head1 SYNOPSIS

B<blkreplay> [I<options>] I<device>

=head1 DESCRIPTION

B<blkreplay> is a utility driving the block layer of the operating
system while measuring latency and throughput of I/O operations
for later visualisation.

B<blkreplay> can create artificial loads (random read-write sweeps,
various kinds of overload tests) or replay natural loads which
have been recorded by blktrace or a similar utility run at
production servers.

B<blkreplay> can be used to test physical hardware, to compare
different brands of hard disks or RAID controllers, to
evaluate the effect of SSD caching, to compare different
block level transports like iSCSI vs Fibrechannel and so on.

B<blkreplay> parses the load data from stdin. To create load data from
B<blktrace> output, use B<conv_blktrace_to_load.sh> script.

=head1 OPTIONS

=over

=item Influence replay duration:

=over

=item B<--replay-start=>I<val>

start offset (in seconds, 0=from_start)

=item B<--replay-end>=I<val>

end offset (in seconds, 0=unlimited)

=item B<--replay-duration>=I<val>

alternatively specify the end offset as delta

=item B<--replay-out>=I<val>

start offset, used for output (in seconds)

=item B<--start-grace>=I<val>

start after grace period for filling the pipes (in seconds)

=back

=item Handling of conflicting I/O requests:

=over

=item B<--with-conflicts>

conflicting writes are ALLOWED (damaged I/O)

=item B<--with-drop>

conflicting writes are simply dropped

=item B<--with-partial>

partial ordering by pushing back conflicts (default)

=item B<--with-ordering>

enforce total order in case of conflicts

=item B<--strong>=I<val>

mode between 0 and 2, see docs (default=1)

=back

=item Replay parameters:

=over

=item B<--threads>=I<val>

parallelism (default=1024)

=item B<--fill-random>=I<val>

fill data blocks with random bytes (%, default=0)

=back

=item Verification modes:

=over

=item B<--no-overhead>

verify is OFF (default)

=item B<--with-verify>

verify on reads

=item B<--with-final-verify>

additional verify pass at the end

=item B<--with-paranoia>

re-read after each write (destroys performance)

=back

=item Convenience:

=over

=item B<--verbose>=I<val>

increase verbosity, show additional INFO: output

=back

=item Expert options (DANGEROUS):

=over

=item B<--o-direct>

use O_DIRECT (default)

=item B<--no-o-direct>

don't use O_DIRECT, deliver FAKE results

=item B<--o-sync>

use O_SYNC

=item B<--no-o-sync>

don't use O_SYNC (default)

=item B<--dry-run>

don't actually do I/O, measure internal overhead

=item B<--fake-io>

omit lseek() and tags, even less internal overhead

=item B<--simulate-io>=I<val>

delay value for I/O simulation (timespec I<sec.nsec>)

=item B<--ahead-limit>=I<val>

limit pipe fillahead (realtime I<sec.nsec>)

=item B<--fan-out>=I<val>

only for kernel hackers (default=4)

=item B<--no-dispatcher>

only for kernel hackers

=item B<--bottleneck>=I<val>

max #requests on dispatch

=item B<--speedup>=I<val>

speedup / slowdown by REAL factor (default=1.0)

=item B<--mmap-mode>

use mmap() instead of read() / write() [NYI]

=back

=back

=head1 AUTHORS

B<blkreplay> was written by Thomas Schoebel-Theuer. This manual page was created from B<blkreplay> documentation by Andrew Shadura.

=head1 COPYRIGHT

Copyright 2009-2012 Thomas Schoebel-Theuer

=head1 SEE ALSO

blktrace(8)