5 collectd-perl - Documentation of collectd's C<perl plugin>
12 IncludeDir "/path/to/perl/plugins"
13 BaseName "Collectd::Plugins"
24 The C<perl plugin> embeds a Perl-interpreter into collectd and provides an
25 interface to collectd's plugin system. This makes it possible to write plugins
26 for collectd in Perl. This is a lot more efficient than executing a
27 Perl-script every time you want to read a value with the C<exec plugin> (see
28 L<collectd-exec(5)>) and provides a lot more functionality, too.
34 =item B<LoadPlugin> I<Plugin>
36 Loads the Perl plugin I<Plugin>. This does basically the same as B<use> would
37 do in a Perl program. As a side effect, the first occurrence of this option
38 causes the Perl-interpreter to be initialized.
40 =item B<BaseName> I<Name>
42 Prepends I<Name>B<::> to all plugin names loaded after this option. This is
43 provided for convenience to keep plugin names short. All Perl-based plugins
44 provided with the I<collectd> distributions reside in the C<Collectd::Plugins>
47 =item E<lt>B<Plugin> I<Name>E<gt> block
49 This block may be used to pass on configuration settings to a Perl plugin. The
50 configuration is converted into a config-item data type which is passed to the
51 registered configuration callback. See below for details about the config-item
52 data type and how to register callbacks.
54 The I<name> identifies the callback. It is used literally and independent of
55 the B<BaseName> setting.
57 =item B<EnableDebugger> I<Package>[=I<option>,...]
59 Run collectd under the control of the Perl source debugger. If I<Package> is
60 not the empty string, control is passed to the debugging, profiling, or
61 tracing module installed as Devel::I<Package>. A comma-separated list of
62 options may be specified after the "=" character. Please note that you may not
63 leave out the I<Package> option even if you specify B<"">. This is the same as
64 using the B<-d:Package> command line option.
66 See L<perldebug> for detailed documentation about debugging Perl.
68 This option does not prevent collectd from daemonizing, so you should start
69 collectd with the B<-f> command line option. Else you will not be able to use
70 the command line driven interface of the debugger.
72 =item B<IncludeDir> I<Dir>
74 Adds I<Dir> to the B<@INC> array. This is the same as using the B<-IDir>
75 command line option or B<use lib Dir> in the source code. Please note that it
76 only has effect on plugins loaded after this option.
78 =item B<RegisterLegacyFlush> I<true|false>
80 The C<Perl plugin> used to register one flush callback (called B<"perl">) and
81 call all Perl-based flush handlers when this callback was called. Newer versions
82 of the plugin wrap the Perl flush handlers and register them directly with the
83 daemon I<in addition> to the legacy B<"perl"> callback. This allows to call
84 specific Perl flush handlers, but has the downside that flushing I<all> plugins
85 now calls the Perl flush handlers twice (once directly and once via the legacy
86 callback). Unfortunately, removing the B<"perl"> callback would break backwards
89 This option allows you to disable the legacy B<"perl"> flush callback if you care
90 about the double call and don't call the B<"perl"> callback in your setup.
94 =head1 WRITING YOUR OWN PLUGINS
96 Writing your own plugins is quite simple. collectd manages plugins by means of
97 B<dispatch functions> which call the appropriate B<callback functions>
98 registered by the plugins. Any plugin basically consists of the implementation
99 of these callback functions and initializing code which registers the
100 functions with collectd. See the section "EXAMPLES" below for a really basic
101 example. The following types of B<callback functions> are known to collectd
102 (all of them are optional):
106 =item configuration functions
108 This type of functions is called during configuration if an appropriate
109 B<Plugin> block has been encountered. It is called once for each B<Plugin>
110 block which matches the name of the callback as provided with the
111 B<plugin_register> method - see below.
115 This type of functions is called once after loading the module and before any
116 calls to the read and write functions. It should be used to initialize the
117 internal state of the plugin (e.E<nbsp>g. open sockets, ...). If the return
118 value evaluates to B<false>, the plugin will be disabled.
122 This type of function is used to collect the actual data. It is called once
123 per interval (see the B<Interval> configuration option of collectd). Usually
124 it will call B<plugin_dispatch_values> to dispatch the values to collectd
125 which will pass them on to all registered B<write functions>. If the return
126 value evaluates to B<false> the plugin will be skipped for an increasing
127 amount of time until it returns B<true> again.
129 =item write functions
131 This type of function is used to write the dispatched values. It is called
132 once for each call to B<plugin_dispatch_values>.
134 =item flush functions
136 This type of function is used to flush internal caches of plugins. It is
137 usually triggered by the user only. Any plugin which caches data before
138 writing it to disk should provide this kind of callback function.
142 This type of function is used to pass messages of plugins or the daemon itself
145 =item notification function
147 This type of function is used to act upon notifications. In general, a
148 notification is a status message that may be associated with a data instance.
149 Usually, a notification is generated by the daemon if a configured threshold
150 has been exceeded (see the section "THRESHOLD CONFIGURATION" in
151 L<collectd.conf(5)> for more details), but any plugin may dispatch
152 notifications as well.
154 =item shutdown functions
156 This type of function is called once before the daemon shuts down. It should
157 be used to clean up the plugin (e.g. close sockets, ...).
161 Any function (except log functions) may set the B<$@> variable to describe
162 errors in more detail. The message will be passed on to the user using
163 collectd's logging mechanism.
165 See the documentation of the B<plugin_register> method in the section
166 "METHODS" below for the number and types of arguments passed to each
167 B<callback function>. This section also explains how to register B<callback
168 functions> with collectd.
170 To enable a plugin, copy it to a place where Perl can find it (i.E<nbsp>e. a
171 directory listed in the B<@INC> array) just as any other Perl plugin and add
172 an appropriate B<LoadPlugin> option to the configuration file. After
173 restarting collectd you're done.
177 The following complex types are used to pass values between the Perl plugin
184 A config-item is one structure which keeps the information provided in the
185 configuration file. The array of children keeps one entry for each
186 configuration option. Each such entry is another config-item structure, which
187 may nest further if nested blocks are used.
191 values => [ val1, val2, ... ],
192 children => [ { ... }, { ... }, ... ]
197 A data-set is a list of one or more data-sources. Each data-source defines a
198 name, type, min- and max-value and the data-set wraps them up into one
199 structure. The general layout looks like this:
202 name => 'data_source_name',
203 type => DS_TYPE_COUNTER || DS_TYPE_GAUGE || DS_TYPE_DERIVE || DS_TYPE_ABSOLUTE,
204 min => value || undef,
205 max => value || undef
210 A value-list is one structure which features an array of values and fields to
211 identify the values, i.E<nbsp>e. time and host, plugin name and
212 plugin-instance as well as a type and type-instance. Since the "type" is not
213 included in the value-list but is passed as an extra argument, the general
214 layout looks like this:
217 values => [123, 0.5],
219 interval => plugin_get_interval (),
221 plugin => 'myplugin',
223 plugin_instance => '',
229 A notification is one structure defining the severity, time and message of the
230 status message as well as an identification of a data instance. Also, it
231 includes an optional list of user-defined meta information represented as
235 severity => NOTIF_FAILURE || NOTIF_WARNING || NOTIF_OKAY,
237 message => 'status message',
239 plugin => 'myplugin',
241 plugin_instance => '',
243 meta => [ { name => <name>, value => <value> }, ... ]
248 A match-proc is one structure storing the callbacks of a "match" of the filter
249 chain infrastructure. The general layout looks like this:
252 create => 'my_create',
253 destroy => 'my_destroy',
259 A target-proc is one structure storing the callbacks of a "target" of the
260 filter chain infrastructure. The general layout looks like this:
263 create => 'my_create',
264 destroy => 'my_destroy',
265 invoke => 'my_invoke'
272 The following functions provide the C-interface to Perl-modules. They are
273 exported by the ":plugin" export tag (see the section "EXPORTS" below).
277 =item B<plugin_register> (I<type>, I<name>, I<data>)
279 Registers a callback-function or data-set.
281 I<type> can be one of:
305 I<name> is the name of the callback-function or the type of the data-set,
306 depending on the value of I<type>. (Please note that the type of the data-set
307 is the value passed as I<name> here and has nothing to do with the I<type>
308 argument which simply tells B<plugin_register> what is being registered.)
310 The last argument, I<data>, is either a function name or an array-reference.
311 If I<type> is B<TYPE_DATASET>, then the I<data> argument must be an
312 array-reference which points to an array of hashes. Each hash describes one
313 data-set. For the exact layout see B<Data-Set> above. Please note that
314 there is a large number of predefined data-sets available in the B<types.db>
315 file which are automatically registered with collectd - see L<types.db(5)> for
316 a description of the format of this file.
318 B<Note>: Using B<plugin_register> to register a data-set is deprecated. Add
319 the new type to a custom L<types.db(5)> file instead. This functionality might
320 be removed in a future version of collectd.
322 If the I<type> argument is any of the other types (B<TYPE_INIT>, B<TYPE_READ>,
323 ...) then I<data> is expected to be a function name. If the name is not
324 prefixed with the plugin's package name collectd will add it automatically.
325 The interface slightly differs from the C interface (which expects a function
326 pointer instead) because Perl does not support to share references to
327 subroutines between threads.
329 These functions are called in the various stages of the daemon (see the
330 section "WRITING YOUR OWN PLUGINS" above) and are passed the following
337 The only argument passed is I<config-item>. See above for the layout of this
346 No arguments are passed.
350 The arguments passed are I<type>, I<data-set>, and I<value-list>. I<type> is a
351 string. For the layout of I<data-set> and I<value-list> see above.
355 The arguments passed are I<timeout> and I<identifier>. I<timeout> indicates
356 that only data older than I<timeout> seconds is to be flushed. I<identifier>
357 specifies which values are to be flushed.
361 The arguments are I<log-level> and I<message>. The log level is small for
362 important messages and high for less important messages. The least important
363 level is B<LOG_DEBUG>, the most important level is B<LOG_ERR>. In between there
364 are (from least to most important): B<LOG_INFO>, B<LOG_NOTICE>, and
365 B<LOG_WARNING>. I<message> is simply a string B<without> a newline at the end.
369 The only argument passed is I<notification>. See above for the layout of this
374 =item B<plugin_unregister> (I<type>, I<plugin>)
376 Removes a callback or data-set from collectd's internal list of
377 functionsE<nbsp>/ datasets.
379 =item B<plugin_dispatch_values> (I<value-list>)
381 Submits a I<value-list> to the daemon. If the data-set identified by
382 I<value-list>->{I<type>}
383 is found (and the number of values matches the number of data-sources) then the
384 type, data-set and value-list is passed to all write-callbacks that are
385 registered with the daemon.
387 =item B<plugin_write> ([B<plugins> => I<...>][, B<datasets> => I<...>],
388 B<valuelists> => I<...>)
390 Calls the write function of the given I<plugins> with the provided I<data
391 sets> and I<value lists>. In contrast to B<plugin_dispatch_values>, it does
392 not update collectd's internal cache and bypasses the filter mechanism (see
393 L<collectd.conf(5)> for details). If the B<plugins> argument has been omitted,
394 the values will be dispatched to all registered write plugins. If the
395 B<datasets> argument has been omitted, the required data sets are looked up
396 according to the C<type> member in the appropriate value list. The value of
397 all three arguments may either be a single scalar or a reference to an array.
398 If the B<datasets> argument has been specified, the number of data sets has to
399 equal the number of specified value lists.
401 =item B<plugin_flush> ([B<timeout> => I<timeout>][, B<plugins> => I<...>][,
402 B<identifiers> => I<...>])
404 Flush one or more plugins. I<timeout> and the specified I<identifiers> are
405 passed on to the registered flush-callbacks. If omitted, the timeout defaults
406 to C<-1>. The identifier defaults to the undefined value. If the B<plugins>
407 argument has been specified, only named plugins will be flushed. The value of
408 the B<plugins> and B<identifiers> arguments may either be a string or a
409 reference to an array of strings.
411 =item B<plugin_dispatch_notification> (I<notification>)
413 Submits a I<notification> to the daemon which will then pass it to all
414 notification-callbacks that are registered.
416 =item B<plugin_log> (I<log-level>, I<message>)
418 Submits a I<message> of level I<log-level> to collectd's logging mechanism.
419 The message is passed to all log-callbacks that are registered with collectd.
421 =item B<ERROR>, B<WARNING>, B<NOTICE>, B<INFO>, B<DEBUG> (I<message>)
423 Wrappers around B<plugin_log>, using B<LOG_ERR>, B<LOG_WARNING>,
424 B<LOG_NOTICE>, B<LOG_INFO> and B<LOG_DEBUG> respectively as I<log-level>.
426 =item B<plugin_get_interval> ()
428 Returns the interval of the current plugin as a floating point number in
429 seconds. This value depends on the interval configured within the
430 C<LoadPlugin perl> block or the global interval (see L<collectd.conf(5)> for
435 The following function provides the filter chain C-interface to Perl-modules.
436 It is exported by the ":filter_chain" export tag (see the section "EXPORTS"
441 =item B<fc_register> (I<type>, I<name>, I<proc>)
443 Registers filter chain callbacks with collectd.
445 I<type> may be any of:
455 I<name> is the name of the match or target. By this name, the callbacks are
456 identified in the configuration file when specifying a B<Match> or B<Target>
457 block (see L<collectd.conf(5)> for details).
459 I<proc> is a hash reference. The hash includes up to three callbacks: an
460 optional constructor (B<create>) and destructor (B<destroy>) and a mandatory
461 B<match> or B<invoke> callback. B<match> is called whenever processing an
462 appropriate match, while B<invoke> is called whenever processing an
463 appropriate target (see the section "FILTER CONFIGURATION" in
464 L<collectd.conf(5)> for details). Just like any other callbacks, filter chain
465 callbacks are identified by the function name rather than a function pointer
466 because Perl does not support to share references to subroutines between
467 threads. The following arguments are passed to the callbacks:
473 The arguments passed are I<config-item> and I<user-data>. See above for the
474 layout of the config-item data-type. I<user-data> is a reference to a scalar
475 value that may be used to store any information specific to this particular
476 instance. The daemon does not care about this information at all. It's for the
481 The only argument passed is I<user-data> which is a reference to the user data
482 initialized in the B<create> callback. This callback may be used to cleanup
483 instance-specific information and settings.
487 The arguments passed are I<data-set>, I<value-list>, I<meta> and I<user-data>.
488 See above for the layout of the data-set and value-list data-types. I<meta> is
489 a pointer to an array of meta information, just like the B<meta> member of the
490 notification data-type (see above). I<user-data> is a reference to the user
491 data initialized in the B<create> callback.
497 =head1 GLOBAL VARIABLES
503 As the name suggests this variable keeps the hostname of the system collectd
504 is running on. The value might be influenced by the B<Hostname> or
505 B<FQDNLookup> configuration options (see L<collectd.conf(5)> for details).
509 This variable keeps the interval in seconds in which the read functions are
510 queried (see the B<Interval> configuration option).
512 B<Note:> This variable should no longer be used in favor of
513 C<plugin_get_interval()> (see above). This function takes any plugin-specific
514 interval settings into account (see the C<Interval> option of C<LoadPlugin> in
515 L<collectd.conf(5)> for details).
519 Any changes to these variables will be globally visible in collectd.
523 By default no symbols are exported. However, the following export tags are
524 available (B<:all> will export all of them):
532 =item B<plugin_register> ()
534 =item B<plugin_unregister> ()
536 =item B<plugin_dispatch_values> ()
538 =item B<plugin_flush> ()
540 =item B<plugin_flush_one> ()
542 =item B<plugin_flush_all> ()
544 =item B<plugin_dispatch_notification> ()
546 =item B<plugin_log> ()
564 =item B<TYPE_SHUTDOWN>
568 =item B<TYPE_DATASET>
576 =item B<DS_TYPE_COUNTER>
578 =item B<DS_TYPE_GAUGE>
580 =item B<DS_TYPE_DERIVE>
582 =item B<DS_TYPE_ABSOLUTE>
612 =item B<:filter_chain>
618 =item B<FC_MATCH_NO_MATCH>
620 =item B<FC_MATCH_MATCHES>
622 =item B<FC_TARGET_CONTINUE>
624 =item B<FC_TARGET_STOP>
626 =item B<FC_TARGET_RETURN>
644 =item B<NOTIF_FAILURE>
646 =item B<NOTIF_WARNING>
666 Any Perl plugin will start similar to:
668 package Collectd::Plugins::FooBar;
673 use Collectd qw( :all );
675 A very simple read function might look like:
679 my $vl = { plugin => 'foobar', type => 'gauge' };
680 $vl->{'values'} = [ rand(42) ];
681 plugin_dispatch_values ($vl);
685 A very simple write function might look like:
689 my ($type, $ds, $vl) = @_;
690 for (my $i = 0; $i < scalar (@$ds); ++$i) {
691 print "$vl->{'plugin'} ($vl->{'type'}): $vl->{'values'}->[$i]\n";
696 A very simple match callback might look like:
700 my ($ds, $vl, $meta, $user_data) = @_;
701 if (matches($ds, $vl)) {
702 return FC_MATCH_MATCHES;
704 return FC_MATCH_NO_MATCH;
708 To register those functions with collectd:
710 plugin_register (TYPE_READ, "foobar", "foobar_read");
711 plugin_register (TYPE_WRITE, "foobar", "foobar_write");
713 fc_register (FC_MATCH, "foobar", "foobar_match");
715 See the section "DATA TYPES" above for a complete documentation of the data
716 types used by the read, write and match functions.
724 Please feel free to send in new plugins to collectd's mailing list at
725 E<lt>collectdE<nbsp>atE<nbsp>collectd.orgE<gt> for review and, possibly,
726 inclusion in the main distribution. In the latter case, we will take care of
727 keeping the plugin up to date and adapting it to new versions of collectd.
729 Before submitting your plugin, please take a look at
730 L<http://collectd.org/dev-info.shtml>.
740 collectd is heavily multi-threaded. Each collectd thread accessing the perl
741 plugin will be mapped to a Perl interpreter thread (see L<threads(3perl)>).
742 Any such thread will be created and destroyed transparently and on-the-fly.
744 Hence, any plugin has to be thread-safe if it provides several entry points
745 from collectd (i.E<nbsp>e. if it registers more than one callback or if a
746 registered callback may be called more than once in parallel). Please note
747 that no data is shared between threads by default. You have to use the
748 B<threads::shared> module to do so.
752 Each function name registered with collectd has to be available before the
753 first thread has been created (i.E<nbsp>e. basically at compile time). This
754 basically means that hacks (yes, I really consider this to be a hack) like
755 C<*foo = \&bar; plugin_register (TYPE_READ, "plugin", "foo");> most likely
756 will not work. This is due to the fact that the symbol table is not shared
757 across different threads.
761 Each plugin is usually only loaded once and kept in memory for performance
762 reasons. Therefore, END blocks are only executed once when collectd shuts
763 down. You should not rely on END blocks anyway - use B<shutdown functions>
768 The perl plugin exports the internal API of collectd which is considered
769 unstable and subject to change at any time. We try hard to not break backwards
770 compatibility in the Perl API during the life cycle of one major release.
771 However, this cannot be guaranteed at all times. Watch out for warnings
772 dispatched by the perl plugin after upgrades.
784 L<threads::shared(3perl)>,
789 The C<perl plugin> has been written by Sebastian Harl
790 E<lt>shE<nbsp>atE<nbsp>tokkee.orgE<gt>.
792 This manpage has been written by Florian Forster
793 E<lt>octoE<nbsp>atE<nbsp>collectd.orgE<gt> and Sebastian Harl
794 E<lt>shE<nbsp>atE<nbsp>tokkee.orgE<gt>.