diff options
author | Niko Tyni <ntyni@iki.fi> | 2005-02-13 20:23:04 +0100 |
---|---|---|
committer | Niko Tyni <ntyni@iki.fi> | 2005-02-13 20:23:04 +0100 |
commit | 6d76521656e91daa160bc8019828f1b68d7aa5dc (patch) | |
tree | aaa27615a0702942fa1606d9a5c89f0a3547467c /doc | |
parent | 6dba1afbe4b475a7d34f5ef867b7b37291cd1484 (diff) | |
download | smokeping-6d76521656e91daa160bc8019828f1b68d7aa5dc.tar.gz smokeping-6d76521656e91daa160bc8019828f1b68d7aa5dc.tar.xz |
Moved probes, matchers and ciscoRttMonMIB modules to lib/Smokeping.
Diffstat (limited to 'doc')
-rw-r--r-- | doc/smokeping_extend.pod | 24 |
1 files changed, 12 insertions, 12 deletions
diff --git a/doc/smokeping_extend.pod b/doc/smokeping_extend.pod index 536ac77..4d3f24d 100644 --- a/doc/smokeping_extend.pod +++ b/doc/smokeping_extend.pod @@ -17,17 +17,17 @@ source of this document are most appreciated. The first thing you should decide is which base class you should use for your probe. For most (if not all) uses it's a choice between -C<probes::base> and C<probes::basefork>. The former is intended for probes +C<Smokeping::probes::base> and C<Smokeping::probes::basefork>. The former is intended for probes that can measure their targets all in one go, while the latter is for probing them one at a time, possibly in several concurrent subprocesses. -At the moment, the only probes that use C<probes::base> are the FPing -derivatives. All the others use C<probes::basefork>, and chances are +At the moment, the only probes that use C<Smokeping::probes::base> are the FPing +derivatives. All the others use C<Smokeping::probes::basefork>, and chances are you should too. This document will thus concentrate on the latter case. =head1 SKELETON FILE -The C<probes::skel> module is a non-functional probe that is intended +The C<Smokeping::probes::skel> module is a non-functional probe that is intended to make a good basis for a new probe module. Copy the file, C<lib/probes/skel.pm>, to a new name and just fill out the blanks :) Note that real probe modules must have at least one capital letter @@ -63,7 +63,7 @@ C<probevars> method for probe-specific variables and a C<targetvars> method for target-specific variables. If you don't know the difference between these yet, see the L<smokeping_examples> document. -(The probes that are derived from C<probes::base> don't support +(The probes that are derived from C<Smokeping::probes::base> don't support target-specific variables, so they only use the C<probevars> method.) The base classes offer these methods too to provide the variables that @@ -88,7 +88,7 @@ convenience method called C<_makevars> that does this, and the common idiom is The variables are declared in a syntax that comes from the module used for parsing the configuration file, C<ISG::ParseConfig>. Each variable should be a hash that uses the "special variable keys" documented in -the C<ISG::ParseConfig> manual. See the C<probes::skel> and the other +the C<ISG::ParseConfig> manual. See the C<Smokeping::probes::skel> and the other probes for examples. For reference, here are the keys the hash should have. Much of this @@ -154,13 +154,13 @@ If you must do something at probe initialization time, like check that the external program you're going to use behaves as you expect, you should do it in the C<new> method. You should probably also take care that you don't run the tests needlessly while in CGI mode. The usual way to -do this is to test for $ENV{SERVER_SOFTWARE}. See the C<probes::skel> +do this is to test for $ENV{SERVER_SOFTWARE}. See the C<Smokeping::probes::skel> module for an example. =head1 PINGING All the real action happens in the C<pingone> method (or, for -C<probes::base>-derived probes, in the C<ping> method.) The arguments +C<Smokeping::probes::base>-derived probes, in the C<ping> method.) The arguments for C<pingone> are C<$self>, the module instance (since this is a method) and C<$target>, the target to be probed. @@ -191,7 +191,7 @@ speak up if you think you'd use it. If you deal with timeouts (for example because your program offers a parameter for specifying the timeout for the pings), you should know a few things. -First, there's timeout logic in C<probes::basefork> that kills the probe +First, there's timeout logic in C<Smokeping::probes::basefork> that kills the probe when the timeout is reached. By default the timeout is (# of pings * 5 seconds) + 1 second. If you expect that your pings can take longer, you should modify the default value of the probe-specific variable C<timeout>. @@ -208,14 +208,14 @@ This would be done like this: If you want to provide a target-specific C<timeout> setting, you should delete the probe-specific variable and be sure to provide a default for -your target-specific one. See eg. C<probes::AnotherDNS> for an example of +your target-specific one. See eg. C<Smokeping::probes::AnotherDNS> for an example of how this is done. Providing a target-specific C<timeout> will make the timeout in -C<probes::basefork> be (# of pings * the maximum timeout of all targets) +C<Smokeping::probes::basefork> be (# of pings * the maximum timeout of all targets) + 1 second. The 1 second is added so that the own timeout logic of the probe has time to kick in even in the worst case (ie. all pings are lost) -before C<probes::basefork> starts killing the processes. +before C<Smokeping::probes::basefork> starts killing the processes. =head1 COPYRIGHT |