Programming

Getting Parallels Plesk Panel to do something without admin's interaction is not tricky. My favorite method of remote-controlling Plesk is via its RPC API. I am a co-author of Perl-implementation API::Plesk, which is available in CPAN.

All XML RPC -requests should be directed towards your Plesk-server at URL
https://-your-plesk-box-here-:8443/enterprise/control/agent.php

Raw XML

First we'll need to get the internal site ID of a domain. A request to get all the subscriptions looks like this:

<?xml version="1.0" encoding="UTF-8"?>
<packet version="1.6.3.5">
    <webspace>
        <get>
            <filter/>
            <dataset>
                <gen_info/>
            </dataset>
        </get>
    </webspace>
</packet>

Note: It would have been possible to filter a specific subscription by domain name, but in this case we just wanted a list of all.

A response to it will contain domain names and their Ids:

<?xml version="1.0" encoding="UTF-8"?>
<packet version="1.6.3.5">
    <webspace>
        <get>
            <result>
                <status>ok</status>
                <filter-id>1</filter-id>
                <id>1</id>
                <data>
                    <gen_info>
                        <name>www.testdomain.org</name>
                    </gen_info>
                </data>
            </result>
        </get>
    </webspace>
</packet>

The response packet contains internal ID and name. We'll be using the internal ID of 1 to get all the DNS-records of the zone:

<?xml version="1.0" encoding="UTF-8"?>
<packet version="1.6.3.5">
    <dns>
        <get_rec>
            <filter>
                <site-id>1</site-id>
            </filter>
        </get_rec>
    </dns>
</packet>

A response packet will look like this:

<?xml version="1.0" encoding="UTF-8"?>
<packet version="1.6.3.5">
    <dns>
        <get_rec>
            <result>
                <status>ok</status>
                <id>111</id>
                <data>
                    <site-id>1</site-id>
                    <type>CNAME</type>
                    <host>www.testdomain.org.</host>
                    <value>testdomain.org.</value>
                    <opt/>
                </data>
            </result>
        </get_rec>
    </dns>
</packet>

There seems not to be any other way of picking a specific record. A filter with type/name would be welcome. Any further operations would be done with the domain record's ID. In this case it is 111.

Perl-code

With a software library, the access is much easier. The same requests would be something like this in Perl:

my $plesk_client = API::Plesk->new('api_version' => '1.6.3.5',
  'secret_key' => $plesk_api_key,
  'url'=>'https://-your-plesk-box-here-:8443/enterprise/control/agent.php',
  'debug'        => 0);

$res = $plesk_client->webspace->get();
die "Subscriptions->get() failed!\n" . $res->error . "\n" if (!$res->is_success);

my @domains = @{$res->results()};
my $cnt = $#domains + 1;
for (my $idx = 0; $idx < $cnt; ++$idx) {
    my $domainId = $domains[$idx]{"id"};
    $domainId += 0;    # toInt

    my $res = $plesk_client->dns->get('site-id' => $domainId);
    die "DNS->get() failed!\n" . $res->error . "\n" if (!$res->is_success);
    my %dns = %{@{$res->results()}[0]};
    print Dump::Dumper(%dns);
}

That is pretty much it.

Update (2nd Nov 2013)

To get all of the domains will require a two-step process (order does not matter): 1) get all the subscriptions (kind of main domains) and 2) get the other domains under subscriptions.

In my Perl-code I do it like this:

# NOTE: This is from the above code
# 1st round:
# Get all the subscriptions.
# There we have the "main" domains
$res = $plesk_client->webspace->get();
die "Subscriptions->get() failed!\n" . $res->error . "\n" if (!$res->is_success);

# NOTE: New one:
# 2nd round:
# Get all the sites.
# There we have the "non-main" domains
$res = $plesk_client->site->get();
die "Sites->get() failed!\n" . $res->error . "\n" if (!$res->is_success);
@domains = @{$res->results()};

In my case, the $res-hash is fed into a ExtractDomains()-function to get the details I need from them. If only the name is required, then no further processing is necessary.