Monitor-MetricsAPI/lib/Monitor/MetricsAPI/Server/Routes.pm
use strict;
use warnings;
package Monitor::MetricsAPI::Server::Routes;
$Monitor::MetricsAPI::Server::Routes::VERSION = '0.900';
use Dancer2;
set serializer => 'JSON';
=head1 NAME
Monitor::MetricsAPI::Server::Routes
=head1 SYNOPSIS
You should not interact with this module directly in your code. Please refer to
L<Monitor::MetricsAPI> for how to integrate this service with your application.
=head1 DESCRIPTION
The base Dancer2 application which provides HTTP API routes for MetricsAPI.
=cut
=head1 ROUTES
The following routes are provided by this server for viewing the collected
metrics.
=cut
=head2 /
Returns a small JSON structure with metadata about the metrics collector. May
be used as a heartbeat for the service, as it performs no computations and does
not invoke any metrics callbacks.
=cut
get '/' => sub {
return {
status => 'ok',
message => 'Metrics service alive.',
service => _service_info(),
};
};
=head2 /all
Returns a full export of all metrics that have been configured in the
containing application. This will include the invoking of all callback metrics.
Depending on your metrics collection, this can be a computationally expensive
process.
If you have any non-trivial callback metrics collected by your application, you
should exercise care in your calls to this route.
=cut
get '/all' => sub {
my $coll = Monitor::MetricsAPI->collector;
return {
status => 'collector_fail',
message => 'Could not access metrics collector.',
service => _service_info(),
} unless defined $coll;
return {
status => 'no_metrics',
message => 'The collector does not have any metrics defined.',
service => _service_info(),
} unless scalar(keys(%{$coll->metrics})) > 0;
return {
status => 'ok',
message => 'Request processed successfully.',
service => _service_info(),
metrics => _expand_metrics(values %{$coll->metrics}),
};
};
=head2 /metric/**
Returns a single named metric, instead of the entire collection of metrics. The
use of this route is substantially preferred for monitoring systems which use
checks against individual metrics, as it limits the invocations of potentially
expensive callback metrics (assuming your application configures any).
A complete metric name must be provided. For returning groups of metrics under
a common namespace, use the /metrics/ route instead.
=cut
get '/metric/**' => sub {
my ($mparts) = splat;
my $coll = Monitor::MetricsAPI->collector;
my $metric_name = join('/', @{$mparts});
return {
status => 'collector_fail',
message => 'Could not access metrics collector.',
service => _service_info(),
} unless defined $coll;
return {
status => 'not_found',
message => 'Invalid metric name provided.',
service => _service_info(),
} unless exists $coll->metrics->{$metric_name};
return {
status => 'ok',
message => 'Request processed successfully.',
service => _service_info(),
metrics => _expand_metrics($coll->metric($metric_name)),
};
};
=head2 /metrics/**
Returns a collection of metrics under the given namespace. Each component of
the namespace prefix you wish to return must be fully specified, but you can
select as deep or as shallow a namespace as you like. In other words, if you
have the following metrics:
messages/incoming/total
messages/incoming/rejected
messages/outgoing/total
messages/outgoing/supressed
users/total
And your monitoring application calls this service with the following path:
/metrics/messages/outgoing
Then this route will return two metrics:
messages/outgoing/total
messages/outgoing/supressed
But if you try to call this service with:
/metrics/messages/out
You will receive an error response.
If you specify the full name of a single metric, you will receive only that
metric back in the response output. Given that, why would you ever use the
"/metric/**" route over this one? Using the single-metric route when you truly
only want a single metric will return an error if the name you give is a
namespace instead of a single metric. It is possible this may be useful in
some circumstances.
=cut
get '/metrics/**' => sub {
my ($nsparts) = splat;
my $coll = Monitor::MetricsAPI->collector;
my $prefix = join('/', @{$nsparts});
return {
status => 'collector_fail',
message => 'Could not access metrics collector.',
service => _service_info(),
} unless defined $coll;
return {
status => 'no_group',
message => 'Must provide metric group prefix.',
service => _service_info(),
} unless defined $prefix && $prefix =~ m{\w+};
my @metrics =
map { $coll->metric($_) }
grep { $_ =~ m{^$prefix(/|$)} }
keys %{$coll->metrics};
return {
status => 'not_found',
message => 'Invalid metric group name provided.',
service => _service_info(),
} unless @metrics > 0;
return {
status => 'ok',
message => 'Request processed successfully.',
service => _service_info(),
metrics => _expand_metrics(@metrics),
};
};
sub _expand_metrics {
my (@metrics) = @_;
my %m;
foreach my $metric (@metrics) {
my @path = split(m|/|, $metric->name);
my $name = pop @path;
my $val = $metric->value;
my $key = \%m;
foreach my $part (@path) {
$key->{$part} = {} unless exists $key->{$part};
$key = $key->{$part};
}
$key->{$name} = $val;
}
return \%m;
}
sub _service_info {
return {
name => 'Monitor::MetricsAPI',
version => $Monitor::MetricsAPI::VERSION,
};
}
=head1 OUTPUT STRUCTURE
Unless otherwise noted, all routes return a similar data structure to the one
described in this section.
=head2 Complete Example
{ "status": "ok",
"message: "Request processed successfully.",
"service": {
"name": "Monitor::MetricsAPI",
"version": "0.001",
},
"metrics": {
"messages": {
"incoming": {
"total": 5019,
"rejected": 104
},
"outgoing": {
"total": 1627,
"suppressed": 5
},
},
"users": {
"total": 1928
}
}
}
=head2 Status
The status attribute is present in every response, and anything other than the
string "ok" indicates an error condition. Responses with a non-"ok" status may
not contain any additional attributes beyond a message which will contain an
error message.
Note that the API server will still return an HTTP status code of 200 even when
there is an error displaying your metrics (e.g. you have requested a metric
which does not exist). The HTTP status codes are used only for indicating
whether the HTTP server itself is functioning properly and is able to process
the incoming HTTP request. HTTP status codes are not overloaded to also serve
as indicators of metrics "health."
=head2 Message
Contains a human-readable message, useful for discerning the reason for failed
requests. Otherwise easily ignored.
=head2 Service
The value of the service attribute is an object containing metadata about the
metrics reporting service.
=head2 Metrics
The metrics attribute contains an object of all metrics which matched the
request, nested according to the categories you used when defining your
metrics. In the case of requests to the "/all" route, this will be every metric
collected by the service for your application. For both the "/metric" and
"/metrics" routes, only those metrics which matched your request will appear in
the object, and all others will be omitted from the output. Callback metrics
which are not present in the API output are not invoked when constructing the
response.
The data type of each metric's value will depend on the type of metric that is
being collected. Counters and gauges will return numbers, boolean metrics will
return 1 (true), 0 (false), or null, and string metrics will return, well,
strings. Callback metrics can return any type, entirely dependent upon what the
subroutine you provided for the callback does. List metrics will return an
array containing values of whatever type was push()'ed onto the list within
your application.
=head1 AUTHORS
Jon Sime <jonsime@gmail.com>
=head1 LICENSE AND COPYRIGHT
This software is copyright (c) 2015 by OmniTI Computer Consulting, Inc.
This module is free software; you can redistribute it and/or
modify it under the same terms as Perl itself. See L<perlartistic>.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
=cut
1;