Ansible role BIND

An Ansible role for setting up ISC BIND as an authoritative-only DNS server for multiple domains. Specifically, the responsibilities of this role are to:

• install BIND
• set up the main configuration file (primary/secondary/forwarder server)
• set up forward and reverse lookup zone files

This role supports multiple forward and reverse zones, including for IPv6. Although enabling recursion is supported (albeit strongly discouraged), consider using another role if you want to set up a caching or forwarding name server.

If you like/use this role, please consider giving it a star and rating it on the role's Ansible Galaxy page. Thanks!

See the change log for notable changes between versions.

WARNING: If you've been using this role since before v5.0.0, please check the change log for important information on breaking changes. Old playbooks will fail if you upgrade to v5.0.0.

Supported platforms

This role can be used on several platforms, see meta/main.yml for an updated list. We strive to set up automated tests for each supported platform (see .ci.yml), but this is not always possible.

A few remarks on supported roles that are not included in automated tests

• Arch Linux and FreeBSD should work, but at this time, it's not possible to test the role on these distros, since no suitable Docker images are available.
• CentOS 6 should work, but idempotence tests fail even if BIND is installed successfully and acceptance tests succeed.

Requirements

The packages python-netaddr (required for the ipaddr filter) and dnspython should be installed on the management node

Role Variables

† Best practice for an authoritative name server is to leave recursion turned off. However, for some cases it may be necessary to have recursion turned on.

Minimal variables for a working zone

In order to set up an authoritative name server that is available to clients, you should at least define the following variables:

Domain definitions

bind_zones:
  # Example of a primary zone (hosts: and name_servers: ares defined)
  - name: mydomain.com           # Domain name
    create_reverse_zones: false  # Skip creation of reverse zones
    primaries:
      - 192.0.2.1                # Primary server(s) for this zone
    name_servers:
      - pub01.mydomain.com.
      - pub02.mydomain.com.
    hosts:
      - name: pub01
        ip: 192.0.2.1
        ipv6: 2001:db8::1
        aliases:
          - ns1
      - name: pub02
        ip: 192.0.2.2
        ipv6: 2001:db8::2
        aliases:
          - ns2
      - name: '@'                # Enables "http://mydomain.com/"
        ip:
          - 192.0.2.3            # Multiple IP addresses for a single host
          - 192.0.2.4            #   results in DNS round robin
        sshfp:                   # Secure shell fingerprint
          - "3 1 1262006f9a45bb36b1aa14f45f354b694b77d7c3"
          - "3 2 e5921564252fe10d2dbafeb243733ed8b1d165b8fa6d5a0e29198e5793f0623b"
        ipv6:
          - 2001:db8::2
          - 2001:db8::3
        aliases:
          - www
      - name: priv01             # This IP is in another subnet, will result in
        ip: 10.0.0.1             #   multiple reverse zones
      - name: mydomain.net.
        aliases:
          - name: sub01
            type: DNAME          # Example of a DNAME alias record
    networks:
      - '192.0.2'
      - '10'
      - '172.16'
    delegate:
      - zone: foo
        dns: 192.0.2.1
    services:
      - name: _ldap._tcp
        weight: 100
        port: 88
        target: dc001
    naptr:                       # Name Authority Pointer record, used for IP
      - name: "sip"              #   telephony
        order: 100
        pref: 10
        flags: "S"
        service: "SIP+D2T"
        regex: "!^.*$!sip:[email protected]!"
        replacement: "_sip._tcp.example.com."
  # Minimal example of a secondary zone
  - name: acme.com
    primaries:
      - 172.17.0.2
    networks:
      - "172.17"
  # Minimal example of a forward zone
  - name: acme.com
    forwarders:
      - 172.17.0.2
    networks:
      - "172.17"

Hosts

Host names that this DNS server should resolve can be specified in bind_zones.hosts as a list of mappings with keys name:, ip:, aliases: and sshfp:. Aliases can be CNAME (default) or DNAME records.

To allow to surf to http://example.com/, set the host name of your web server to '@' (must be quoted!). In BIND syntax, @ indicates the domain name itself.

If you want to specify multiple IP addresses for a host, add entries to bind_zones.hosts with the same name (e.g. priv01 in the code snippet). This results in multiple A/AAAA records for that host and allows DNS round robin, a simple load balancing technique. The order in which the IP addresses are returned can be configured with role variable bind_rrset_order.

Networks

As you can see, not all hosts are in the same subnet. This role will generate suitable reverse lookup zones for each subnet. All subnets should be specified in bind_zones.networks, though, or the host will not get a PTR record for reverse lookup.

Remark that only the network part should be specified here! When specifying a class B IP address (e.g. "172.16") in a variable file, it must be quoted. Otherwise, the Yaml parser will interpret it as a float.

Based on the idea and examples detailed at https://linuxmonk.ch/wordpress/index.php/2016/managing-dns-zones-with-ansible/ for the gdnsd package, the zone files are fully idempotent, and thus only get updated if "real" content changes.

Zone types and Zone type auto-detection

Zone type is an optional zone parameter that defines if the zone type should be of primary, secondary or forward type. When type parameter is omitted, zone type will be autodetected based on the intersection of host IP addresses and primaries record when configuring primary or secondary zone. When primaries is not defined and forwarders is defined, the zone type will be set to forward.

Zone auto-detection functionality is especially useful when deploying multi-site DNS infrastructure. It is convenient to have a "shared" bind_zones definitions in a single group inventory file for all dns servers ( ex. group_vars\dns.yml). Such an approach allows to switch between primary and secondary server(s) roles by updating primaries record only and rerunning the playbook. Zone type auto-detection can be tested with "shared_inventory" molecule scenario by running: molecule test --scenario-name shared_inventory

NOTE

• bind doesn't support automated multi-master configuration and primaries list should have a single entry only.
• When primaries record is updated to switch primary to secondary server roles, zones will be wiped out and recreated from template as we yet to support dynamic updates for existing zones.

Zone types can be also defined explicitly in per host inventory to skip autodetection:

# Primary Server
bind_zones:
  - name: mydomain.com
    type: primary
    primaries:
      - 192.0.2.1
...
# Secondary Server
bind_zones:
  - name: mydomain.com
      type: secondary
      primaries:
        - 192.0.2.1
...
# Forwarder Server
bind_zones:
  - name: anotherdomain.com
      type: forward
      forwarders:
        - 192.0.3.1

Zone delegation

To delegate a zone to a DNS server, it is sufficient to create a NS record (under delegate) which is the equivalent of:

foo IN NS 192.0.2.1

Service records

Service (SRV) records can be added with the services. This should be a list of mappings with mandatory keys name: (service name), target: (host providing the service), port: (TCP/UDP port of the service) and optional keys priority: (default = 0) and weight: (default = 0).

ACLs

ACLs can be defined like this:

bind_acls:
  - name: acl1
    match_list:
      - 192.0.2.0/24
      - 10.0.0.0/8

The names of the ACLs will be added to the allow-transfer clause in global options.

Binding Keys

Binding keys can be defined like this:

bind_dns_keys:
  - name: primary_key
    algorithm: hmac-sha256
    secret: "azertyAZERTY123456"
bind_extra_include_files:
  - "{{ bind_auth_file }}"

tip: Extra include file must be set as an ansible variable because file is OS dependant

This will be set in a file "{{ bind_auth_file }} (e.g. /etc/bind/auth_transfer.conf for Debian) which have to be added in the list variable bind_extra_include_files

Using TSIG for zone transfer (XFR) authorization

To authorize the transfer of zone between primary & secondary servers based on a TSIG key, set the mapping in the variable bind_key_mapping:

bind_key_mapping:
  primary_ip: TSIG-keyname

Each primary can only have one key (per view).

A check will be performed to ensure the key is actually present in the bind_dns_keys mapping. This will add a server statement for the a in bind_auth_file on a secondary server containing the specified key.

Dependencies

No dependencies.

Example Playbooks

See the test playbooks and inventory for an elaborate example that showcases most features.

Standard Inventory

• Variables common between all servers defined in all.yml
• bind_zone variable defined on per host basis (primary, secondary and forwarder)

❯ tree --dirsfirst molecule/default
molecule/default
├── group_vars
│   └── all.yml
├── host_vars
│   ├── ns1.yml    # Primary
│   ├── ns2.yml    # Secondary
│   └── ns3.yml    # Forwarder
├── converge.yml
...

Shared Inventory

Variables common between primary and secondary servers defined in all.yml

❯ tree --dirsfirst molecule/shared_inventory
molecule/shared_inventory
├── group_vars
│   └── all.yml
├── converge.yml
...

Testing

This role is tested using Ansible Molecule. Tests are launched automatically on Github Actions after each commit and PR.

This Molecule configuration will:

• Run Yamllint and Ansible Lint
• Create three Docker containers, one primary (ns1), one secondary (ns2) DNS server and forwarder(ns3) - default molecule scenario
• Run a syntax check
• Apply the role with a test playbook and check idempotence
• Run acceptance tests with verify playbook
• Create two additional Docker containers, one primary(ns4) and one secondary (ns5) and run shared_inventory scenario

This process is repeated for all the supported Linux distributions.

Local test environment

In order to run the acceptance tests on this role locally, you can install the necessary tools on your machine, or use this reproducible setup in a VirtualBox VM (set up with Vagrant): https://github.com/bertvv/ansible-testenv.

Steps to install the tools manually:

1. Docker should be installed on your machine
2. As recommended by Molecule, create a python virtual environment
3. Install the software tools python3 -m pip install molecule molecule-docker docker netaddr dnspython yamllint ansible-lint
4. Navigate to the root of the role directory and run molecule test

Molecule automatically deletes the containers after a test. If you would like to check out the containers yourself, run molecule converge followed by molecule login --host HOSTNAME.

The Docker containers are based on images created by Jeff Geerling, specifically for Ansible testing (look for images named geerlingguy/docker-DISTRO-ansible). You can use any of his images, but only the distributions mentioned in meta/main.yml are supported.

The default config will start three Centos 8 containers (the primary supported platform at this time). Choose another distro by setting the MOLECULE_DISTRO variable with the command, e.g.:

MOLECULE_DISTRO=debian9 molecule test

or

MOLECULE_DISTRO=debian9 molecule converge

You can run the acceptance tests on all servers with molecule verify.

Verification tests are done using "dig" lookup module by quering dns records and validating responses. This requires direct network communication between Ansible controller node (your machine running Ansible) and the target docker container.

NOTE

Molecule verify tests will fail if docker is running on MacOS, as MacOS cannot access container IP directly. This is a known issue. See #2670.

Workaround:

1. Run molecule linter: molecule lint
2. Provision containers: molecule converge
3. Connect to container: molecule login --host ns1
4. Go to role directory: cd /etc/ansible/roles/bertvv.bind
5. Run verify playbook:

ansible-playbook -c local -i "`hostname`," -i molecule/default/inventory.ini molecule/default/verify.yml

6. Repeat steps 2-4 for ns2 and ns3

License

BSD

Contributors

This role could only have been realized thanks to the contributions of many. If you have an idea to improve it even further, don't hesitate to pitch in!

Issues, feature requests, ideas, suggestions, etc. can be posted in the Issues section.

Pull requests are also very welcome. Please create a topic branch for your proposed changes. If you don't, this will create conflicts in your fork after the merge. Don't hesitate to add yourself to the contributor list below in your PR!

Maintainers:

• Bert Van Vreckem
• Gregory Shulov
• Stuart Knight

Contributors:

• Aido
• Angel Barrera
• B. Verschueren
• Boris Momčilović
• Brad Durrow
• Christopher Hicks
• David J. Haines
• Fabio Rocha
• Fazle Arefin
• flora-five
• Greg Cockburn
• Guillaume Darmont
• itbane
• jadjay
• Jascha Sticher
• Joanna Delaporte
• Jörg Eichhorn
• Jose Taas
• Lennart Weller
• Loic Dachary
• Mario Ciccarelli
• Miroslav Hudec
• Otto Sabart
• Paulius Mazeika
• Paulo E. Castro
• Peter Janes
• psa
• Rafael Bodill
• Rayford Johnson
• Robin Ophalvens
• Romuald
• roumano
• Shawn Wilsher
• Tom Meinlschmidt
• Jascha Sticher
• Zephyr82