globaliolimits.cfg(5)

1GLOBALIOLIMITS.CFG(5)                                    GLOBALIOLIMITS.CFG(5)
2
3
4

NAME

6       globaliolimits.cfg - global I/O limiting configuration
7

DESCRIPTION

9       The file globaliolimits.cfg contains configuration of the global I/O
10       limiter.
11

SYNTAX

13       Syntax is:
14
15       option value
16
17       Lines starting with # character are ignored.
18

OPTIONS

20       Configuration options:
21
22       subsystem <subsystem>
23           The cgroups subsystem by which clients are classified. If left
24           unspecified, all clients are considered unclassified (see below).
25
26       limit unclassified <throughput in KiB/s>
27           This is a special entry for clients that don’t match any group
28           specified in configuration file or for all clients if subsystem is
29           unspecified. If this entry is unspecified and subsystem is
30           unspecified as well, I/O limiting is disabled entirely. If this
31           entry is unspecified but subsystem is specified, unclassified
32           clients are not allowed to perform I/O.
33
34       limit <group> <throughput in KiB/s>
35           Set limit for clients belonging to the cgroups group <group>. In
36           LizardFS, subgroups of <group> constitute independent groups; they
37           are not allowed to use <group>'s bandwidth reservation and they
38           don’t count against <group>'s usage.
39

EXAMPLES

41           # empty file
42
43       I/O limiting is disabled, no limits are enforced.
44
45           limit unclassified 1024
46
47       All clients are unclassified and share 1MiB/s of bandwidth.
48
49           subsystem blkio
50           limit /a 1024
51
52       Clients in the blkio /a group are limited to 1MiB/s, no other clients
53       can perform any I/O.
54
55           subsystem blkio
56           limit unclassified 256
57           limit /a   1024
58           limit /b/a 2048
59
60       The blkio group /a is allowed to transfer 1MiB/s, while /b/a gets
61       2MiB/s. Clients from other groups (e.g. /b, /z, /a/a, /b/z) are
62       considered unclassified and share 256KiB/s of bandwidth.
63

TUNING NOTES

65       Global I/O limiting is managed by the master server. Mount instances
66       reserve bandwidth allocations from master when they want to perform I/O
67       to chunkservers.
68
69       To avoid overloading the master under heavy traffic, mounts try to
70       predict their future usage and reserve at once all the bandwidth they
71       will for the next renegotiation period (see mfsmaster.cfg(5)).
72
73       Such reservation are wasted if the traffic at given mount instance
74       suddenly drops.
75
76       The ratio of bandwidth being wasted due to this phenomenon shouldn’t
77       exceed fsp/b, where:
78
79           f is the frequency of sudden traffic drops in the whole installation (in 1/s)
80           s is the average size of such drop (in KiB/s)
81           p is the renegotiation period (in s)
82           b is the bandwidth limit (in KiB/s)
83
84       This applies to each group separately, because groups reserve their
85       bandwidth independently from each other.
86

COPYRIGHT

88       Copyright 2008-2009 Gemius SA, 2013-2015 Skytechnology sp. z o.o.
89
90       LizardFS is free software: you can redistribute it and/or modify it
91       under the terms of the GNU General Public License as published by the
92       Free Software Foundation, version 3.
93
94       LizardFS is distributed in the hope that it will be useful, but WITHOUT
95       ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
96       FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
97       for more details.
98
99       You should have received a copy of the GNU General Public License along
100       with LizardFS. If not, see http://www.gnu.org/licenses/.
101