Commit | Line | Data |
---|---|---|
805e021f CE |
1 | =head1 NAME |
2 | ||
3 | fs_bypassthreshold - get/set Cache Bypass file size threshold | |
4 | ||
5 | =head1 SYNOPSIS | |
6 | ||
7 | =for html | |
8 | <div class="synopsis"> | |
9 | ||
10 | B<fs bypassthreshold> S<<< [B<-size> <I<file size>>] >>> [B<-help>] | |
11 | ||
12 | B<fs bypassthresh> S<<< [B<-s> <I<file size>>] >>> [B<-h>] | |
13 | ||
14 | =for html | |
15 | </div> | |
16 | ||
17 | =head1 DESCRIPTION | |
18 | ||
19 | The B<fs bypassthreshold> command either changes the AFS client Cache | |
20 | Bypass file size threshold, or reports the current threshold. Normally, | |
21 | when an application tries to read data via the OpenAFS client, the client | |
22 | will fetch the required data from the fileserver, store the results in | |
23 | the local cache, and then serve the application request from the local | |
24 | cache. When the Cache Bypass feature is enabled, certain files will be | |
25 | read directly from the network, and the data will not be saved to the | |
26 | local cache. In some scenarios, this can provide a significant | |
27 | performance improvement, especially if the data is only read once. | |
28 | ||
29 | Currently, the only way to specify which files should bypass the cache | |
30 | and which files should not, is by file size. In the future, different | |
31 | mechanisms may exist for specifying which files should bypass the local | |
32 | cache and which should not. | |
33 | ||
34 | If B<-size> is not specified, B<fs bypassthreshold> just reports the | |
35 | current threshold and exits. | |
36 | ||
37 | =head1 CAUTIONS | |
38 | ||
39 | Cache Bypass is still considered an experimental feature. Even in stable | |
40 | releases of OpenAFS, it is not considered as stable as the rest of OpenAFS. | |
41 | ||
42 | =head1 OPTIONS | |
43 | ||
44 | =over 4 | |
45 | ||
46 | =item B<-size> <I<file size>> | |
47 | ||
48 | Specifies how large a file must be in order to enable Cache Bypass mode. | |
49 | If a file is larger than the specified file size (in bytes), reading from | |
50 | that file will bypass the local cache and be read directly from the | |
51 | network. A file size of -1 indicates that Cache Bypass should be disabled | |
52 | complete, so no files will enable Cache Bypass mode, no matter how large | |
53 | they are. | |
54 | ||
55 | By default, Cache Bypass is disabled. | |
56 | ||
57 | =item B<-help> | |
58 | ||
59 | Prints the online help for this command. All other valid options are | |
60 | ignored. | |
61 | ||
62 | =back | |
63 | ||
64 | =head1 OUTPUT | |
65 | ||
66 | The output format is identical whether or not B<-size> was specified. If | |
67 | B<-size> is specified, the bypass threshold is changed, and the output | |
68 | represents the new, changed, threshold. Otherwise the output represents | |
69 | the current threshold. | |
70 | ||
71 | =head1 EXAMPLES | |
72 | ||
73 | The following example turns on Cache Bypass with a threshold of 4KiB: | |
74 | ||
75 | % fs bypassthreshold 4096 | |
76 | Cache bypass threshold 4096 | |
77 | ||
78 | The following example disables Cache Bypass: | |
79 | ||
80 | % fs bypassthreshold -1 | |
81 | Cache bypass threshold -1 (disabled) | |
82 | ||
83 | The following example examines the current Cache Bypass threshold: | |
84 | ||
85 | % fs bypassthreshold | |
86 | Cache bypass threshold 4096 | |
87 | ||
88 | =head1 PRIVILEGE REQUIRED | |
89 | ||
90 | To set the bypass threshold, the issuer must be logged in as the local | |
91 | superuser C<root>. Otherwise, no privilege is required. | |
92 | ||
93 | =head1 COPYRIGHT | |
94 | ||
95 | Copyright 2012 Sine Nomine Associates | |
96 | ||
97 | This documentation is covered by the BSD License as written in the | |
98 | doc/LICENSE file. This man page was written by Andrew Deason for | |
99 | OpenAFS. |