IMPORTANT UPGRADE INFORMATION: In version 6.0.0 of this module there are significant changes to the handling of default zones that may require preparations prior to upgrading. See DEFAULT_ZONES.md for details.

Control BIND name servers and zones

The BIND module provides an interface for managing a BIND name server, including installation of software, configuration of the server, creation of keys, and definitions for zones.

• package installation and service control for BIND
• configuration of the server, zones, acls, keys, and views
• creation of TSIG and DNSSEC keys

The BIND module depends on ripienaar/module_data for hiera module data support and puppetlabs/stdlib for certain parser functions. When using the BIND module, please install the module using puppet module install or an equivalent substitute in order to ensure that its dependencies are met.

To begin using the BIND module with default parameters, declare the class

class { 'bind': }

Puppet code that uses anything from the BIND module requires that the core bind classes be declared.

bind provides a few parameters that control server-level configuration parameters in the named.conf file, and also defines the overall structure of DNS service on the node.

class { 'bind':
    forwarders => [
        '8.8.8.8',
        '8.8.4.4',
    ],
    dnssec     => true,
    version    => 'Controlled by Puppet',
}

Puppet will manage the entire named.conf file and its includes. Most parameters are set to a fixed value, but the server's upstream resolvers are controlled using forwarders, enabling of DNSSec signature validation is controlled using dnssec, and the reported version is controlled using version. It is unlikely that you will need to define an alternate value for confdir or cachedir.

You can setup bind with chroot using the $chroot parameter: Example for CentOS7:

class { 'bind':
    chroot                => true,
    # Note: this file MUST be into the /etc/named directory so the
    # RHEL7 specific setup-named-chroot.sh script will make it available into
    # the chroot.
    default_zones_include => '/etc/named/default-zones.conf',
    forwarders            => [
        '8.8.8.8',
        '8.8.4.4',
    ],
    dnssec                => true,
    version               => 'Controlled by Puppet',
}

Note: chroot functionality is not suitable for moving an existing BIND installation and configuration into a chroot. Please use only for provisioning of new servers.

The bind::updater class is an alternate entrypoint into the module. This class installs the BIND client tools but not a name server. The installed tools are sufficient to allow the use of the resource_record custom resource.

Creates a TSIG key file. Only the secret parameter is required, but it is recommended to explicitly supply the algorithm as well. The key file will be stored in ${::bind::confdir}/keys with a filename derived from the title of the bind::key declaration.

bind::key { 'local-update':
    algorithm => 'hmac-sha256', # default: 'hmac-sha256'
    secret    => '012345678901345678901234567890123456789=',
    owner     => 'root',
    group     => 'bind',
}

If no secret is specified, the bind::key define will generate one. The secret_bits parameter controls the size of the secret.

bind::key { 'local-update':
    secret_bits => 512, # default: 256
}

Declares an acl in the server's configuration. The acl's name is the title of the bind::acl declaration.

bind::acl { 'rfc1918':
    addresses => [
        '10.0.0.0/8',
        '172.16.0.0/12',
        '192.168.0.0/16',
    ]
}

bind::acl { 'secondary-dns':
    addresses => '192.0.2.4/32',
}

Declares a zone. Each zone must be included in at least one view in order to be included in the server's configuration, and may be included in multiple views. The corresponding zone file will be created if it is absent, but any existing file will not be overwritten. Only the zone_type is required. If domain is unspecified, the title of the bind::zone declaration will be used as the domain.

A master zone with a zone file managed directly by Puppet:

bind::zone { 'example.org':
    zone_type       => 'master',
    dynamic         => false,
    source          => 'puppet:https:///dns/db.example.org',
    allow_transfers => [ 'secondary-dns', ],
}

A master zone with DNSSec disabled which allows updates using a TSIG key and zone transfers to servers matching an acl:

bind::zone { 'example.com-internal':
    zone_type       => 'master',
    domain          => 'example.com',
    allow_updates   => [ 'key local-update', ],
    allow_transfers => [ 'secondary-dns', ],
    ns_notify       => true,
    dnssec          => false,
}

A master zone with DNSSec enabled which allows updates using a TSIG key and zone transfers to servers matching an acl:

bind::zone { 'example.com-external':
    zone_type       => 'master',
    domain          => 'example.com',
    allow_updates   => [ 'key local-update', ],
    allow_transfers => [ 'secondary-dns', ],
    ns_notify       => true,
    dnssec          => true,
    key_directory   => '/var/cache/bind/example.com',
}

A master zone which is initialized with a pre-existing zone file (for example, to migrate an existing zone to a bind-module controlled server or to recover from a backup):

bind::zone { 'example.com':
    zone_type => 'master',
    source    => 'puppet:https:///backups/dns/example.com',
}

A slave zone which allows notifications from servers matched by IP:

bind::zone { 'example.net':
    zone_type    => 'slave',
    masters      => [ '198.0.2.2' ],
    allow_notify => [ '192.0.2.2' ],
    ns_notify    => false,
}

A forward zone:

bind::zone { 'example.org':
    zone_type  => 'forward',
    forwarders => [ '10.0.2.4', ],
    forward    => 'only',
}

Declares a view in the BIND configuration. In order to declare zones in a server configuration there must be at least one view declaration which includes the zones.

A common use for views is to use a single dual-homed nameserver as a resolver on a private network and an authoritative non-resolving nameserver on the Internet. Furthermore, the Internet-facing and private network-facing views may present different authoritative results for a domain. Given a BIND server connected to the internet with the address 198.0.2.2 and connected to a private network with the address 10.0.2.2, here are the bind::view declarations that would create this configuration:

bind::view { 'internet':
    recursion          => false,
    match_destinations => [ '198.0.2.2', ],
    zones              => [ 'example.net', 'example.com-external', ],
}

bind::view { 'private':
    recursion          => true,
    match_destinations => [ '10.0.2.2', ],
    zones              => [ 'example.net', 'example.com-internal', ],
}

In this scenario, the example.com domain has two separate zones that are presented in each of the internet and private views. These two zones are independent, and TSIG-signed updates to example.com must be made to either 198.0.2.2 or 10.0.2.2, to change the internet or private views of this domain. Updates to example.net may be made via either address, since the zone is included in both views.

Another use for views is to control access to the DNS server's services. In this example, service is restricted to a specific set of client address ranges, and queries for the example.org domain are handled using a declared zone (see bind::zone declaration for example.org above):

bind::view { 'clients':
    recursion     => true,
    match_clients => [
        '10.10.0.0/24',
        '10.100.0.0/24',
    ],
    zones         => [
        'example.org',
    ],
}

View declarations can also include server clause configuration. The servers property of bind::view accepts an array value which specifies each server clause in the view as a hash. The hash must contain an ip_addr key which specifies the IP address (optionally, a CIDR address range), and may contain a keys key with a string value. The value of keys will be used as the name of a key in the server clause. In this example, the ns view will contain a server clause that configures BIND to use the key internal-ns to TSIG-sign transactions with 192.168.24.2 and the key hurricane-electric to TSIG-sign transactions with 216.218.130.2:

bind::view { 'ns':
    servers => [
        {
            'ip_addr' => '192.168.24.2',
            'keys'    => 'internal-ns',
        },
        {
            'ip_addr' => '216.218.130.2',
            'keys'    => 'hurricane-electric',
        }
    ],
    ...
}

Declares a resource record. For example:

resource_record { 'www.example.com address':
    ensure  => present,
    record  => 'www.example.com',
    type    => 'A',
    data    => [ '172.16.32.10', '172.16.32.11' ],
    ttl     => 86400,
    zone    => 'example.com',
    server  => 'ns.example.com',
    keyname => 'local',
    hmac    => 'hmac-sha1',
    secret  => 'aLE5LA=='
}

This resource declaration will result in address records with the addresses 172.16.32.10 and 172.16.32.11 (data), a TTL of 86400 (ttl) in the zone example.com (zone). Any updates necessary to create, update, or destroy these records are authenticated using a TSIG key named 'local' (keyname) of the given type (hmac) with the given secret.

No semantic information is communicated in the resource title. It is strictly for disambiguation of resources within Puppet.

record is required, and is the fully qualified record to be managed.

type is required, and is the record type. It must be one of: A AAAA CNAME NS MX SPF SRV NAPTR PTR or TXT. Other DNS record types are not currently supported.

rrclass is the class of the record. The default value is IN and allowed values are IN, CH, and HS.

data is required, and may be a scalar value or an array of scalar values whose format conform to the type of DNS resource record being created. data is an ensurable property and changes will be reflected in DNS. Note: for record types that have a DNS name as either the whole value or a component of the value (e.g. NS, 'MX', CNAME, PTR, NAPTR, or SRV) you must specify the name as a fully-qualified name with a trailing dot in order to satisfy both BIND, which will otherwise consider it a name relative, and Puppet, which will not consider the dot-qualified output of dig equal to a non-dot-qualified value in the manifest.

ttl defaults to 43200 and need not be specified. ttl is an ensurable property and changes will be reflected in DNS.

zone is not required, and generally not needed. It is only necessary to specify the zone to be updated if the target nameserver has the record in multiple zones, e.g. the NS records of a zone whose parent zone is served by the same nameserver.

server defaults to "localhost" and need not be specified. The value may be either a hostname or IP address.

query_section indicates the section of the DNS response to check for existing record values. It must be one of answer, authority, or additional. Defaults to: answer

keyname defaults to "update" and need not be specified. This parameter specifies the name of a TSIG key to be used to authenticate the update. The resource only uses a TSIG key if a secret is specified.

keyfile specifies the name of a key file to use to sign requests. This parameter has no effect if a secret is specified.

hmac defaults to "hmac-sha1" and need not be specified. This parameter specifies the algorithm of the TSIG key to be used to authenticate the update. The resource only uses a TSIG key if a secret is specified.

secret is optional. This parameter specifies the encoded cryptographic secret of the TSIG key to be used to authenticate the update. If no secret is specified, then the update will not use TSIG authentication.

Mail exchangers for a domain. Declares three mail exchangers for the domain example.com, which are mx.example.com, mx2.example.com, and mx.mail-host.ex with priorities 10, 20, and 30, respectively (note the trailing dots in the values to denote fully-qualified names):

resource_record { 'example.com mail exchangers':
    record => 'example.com',
    type   => 'MX',
    data   => [ '10 mx.example.com.', '20 mx2.example.com.', '20 mx.mail-host.ex.', ],
}

Nameserver records for a zone. Declares three nameserver records for the zone example.com, which are ns1.example.com, ns2.example.com, and ns.dns-host.ex:

resource_record { 'example.com name servers':
    record => 'example.com',
    type   => 'NS',
    data   => [ 'ns1.example.com.', 'ns2.example.com.', 'ns.dns-host.ex.' ],
}

Delegating nameserver records in a parent zone. Declares a nameserver record in the parent zone in order to delegate authority for a subdomain:

resource_record { 'sub.example.com delegation':
    record        => 'sub.example.com'
    type          => 'NS',
    zone          => 'example.com',
    query_section => 'authority',
    data          => 'sub-ns.example.com.',
}

Service locators records for a domain. Declares a service locator for SIP over UDP to the domain example.com, in which the service located at port 5060 of inbound.sip-host.ex is given priority 5 and weight 100.

resource_record { 'example.com SIP service locator':
    record => '_sip._udp.example.com',
    type   => 'SRV',
    data   => [ '5 100 5060 inbound.sip-host.ex.', ],
}

This error usually has one of two causes.

First of all, the bind module has dependencies. These are declared in the module manifest, and they are properly installed via the puppet module install command and when using librarian-puppet.

Secondly, after installing the module and its dependencies it is necessary to restart the Puppet master if you are using master mode. The ripienaar/module_data module contains functionality which is only loaded at startup.

Support for specific platforms is explicit in this module. If you see this message and would like to try to make the module work on your platform, you can do this by setting bind::defaults::supported to true in hiera, and then overriding other parameters of the bind::defaults class as necessary to make the module work. If you succeed or if you require help, please open an issue. I would be happy to include your changes in order to support additional platforms.