Skip to content

Designate with the BIND9 back-end in Kolla-Ansible. Name servers, NS records, Neutron DNS integration, and how to check a record end to end

Context

Designate gives projects DNS zones and records through an API, and with Neutron's DNS integration it creates records for instances and floating IPs automatically. With enable_designate: "yes" and the default designate_backend: "bind9", Kolla-Ansible deploys the Designate services and a BIND9 server that serves the zones.

Three things decide whether the names are actually resolvable, and they sit partly outside Kolla-Ansible:

  1. the name servers' names (designate_ns_record), which become the NS records of every zone;
  2. where BIND9 listens, and who can reach it;
  3. the delegation of your zones from the parent DNS, which Kolla-Ansible does not do.

This page covers those, Neutron's integration, the optional designate-sink, and how to check a record from the API down to a DNS answer.

Topology

From the 2026.1 sample multinode inventory:

Container Group Role
designate_api, designate_central, designate_producer, designate_worker designate (control) API, zone logic, periodic tasks, pushing zones to the back-end
designate_backend_bind9 designate-backend-bind9 (designate) The BIND9 server: authoritative for the zones, port 53
designate_mdns designate-mdns (network) Serves zones to BIND9 by zone transfer, port 5354
designate_sink designate-sink (designate) Optional: records from Nova and Neutron notifications, off by default

With more than one designate_worker, the guide requires a coordination back-end, and only Valkey is supported: designate_coordination_backend defaults to valkey when enable_valkey is true, and is empty otherwise. With three controllers, enable Valkey.

Network design

In 2026.1, named.conf.j2 makes BIND9 listen on port 53 on the host's API address, and also on the dns_interface address when that interface differs from the API one. dns_interface defaults to network_interface. The guide adds that the mDNS node "requires the dns_interface to be reachable from management network".

So decide who must query BIND9:

  • only clients inside the platform: the API network is enough;
  • resolvers or clients outside: give the BIND9 hosts a dns_interface on a network those resolvers can reach, and open port 53 (UDP and TCP) to it.

BIND9 here is authoritative only: designate_recursion defaults to "no". It does not resolve other names for instances; their resolver stays the one their subnet's DHCP gives them.

Storage design

Zones and records live in the Designate database (MariaDB); BIND9 keeps its copies under /var/lib/named in its container's volume. Nothing to size beyond that.

Inventory and configuration layout

# globals.yml
enable_designate: "yes"
designate_backend: "bind9"                  # the default
designate_ns_record:                        # names, not addresses; they become the NS records of every zone
  - "ns1.cloud.example.org"
  - "ns2.cloud.example.org"
dns_interface: "<interface-reachable-by-resolvers>"   # default: network_interface
neutron_dns_domain: "cloud.example.org."    # must end with a dot and differ from "openstacklocal"
enable_valkey: "yes"                        # coordination for several designate_worker

designate_ns_record defaults to ns1.example.org: left as is, every zone announces a name server that is not yours. The names must resolve, in the parent DNS, to the addresses BIND9 listens on, and the parent zone must delegate your zones to them; Kolla-Ansible configures neither.

pools.yaml is generated from the inventory (one target per designate-backend-bind9 host, mDNS hosts as masters) unless you place your own at /etc/kolla/config/designate/pools.yaml. The deploy and reconfigure run designate-manage pool update in designate_worker, so a change to the NS records reaches existing zones.

Neutron DNS integration

With Designate enabled, neutron_dns_integration defaults to true. Neutron then sets external_dns_driver = designate and dns_domain = <neutron_dns_domain>, and loads the subnet_dns_publish_fixed_ip extension driver (2026.1 roles). Records are created in Designate when a port has a dns_name and its network has a dns_domain that matches an existing zone; floating IPs get records the same way. See Neutron's DNS integration with an external service guide for the exact rules per network type.

The guide warns: neutron_dns_domain "has to be different to openstacklocal (its default value) and has to end with a period".

Applying it

1. Create the delegation first (outside OpenStack): ns1/ns2 names in the parent zone, pointing at the BIND9 hosts' addresses on dns_interface, and NS records delegating your zone to them.

2. Deploy.

⚠️ State-changing. Creates the Designate database, service user and endpoints, the Designate and BIND9 containers, and reconfigures Neutron (external_dns_driver, the extension driver), whose containers restart where their configuration changed, and Nova: enabling Designate turns on Nova's notifications in nova.conf (2026.1 templates), so the Nova containers restart, on every compute too. Running instances keep running. The new HAProxy frontend for the Designate API restarts the haproxy containers on every load balancer host, backups first, then the VIP holder: a brief interruption of every API behind the VIPs. Plan it in a window. Undo: enable_designate: "no" and reconfigure Neutron and Nova; the Designate containers and HAProxy fragment stay until removed by hand, as described for the monitoring page.

kolla-ansible deploy -i <inventory> --tags common,loadbalancer,valkey,designate,neutron,nova

3. Create a zone and use it (the guide's sequence, with your names):

⚠️ State-changing. Creates a zone and records; assigning a dns_domain to a network makes Neutron create records for new ports on it.

openstack zone create --email dnsadmin@example.org cloud.example.org.
openstack zone show cloud.example.org. -c status -c serial      # ACTIVE once BIND9 has it
openstack network set <network> --dns-domain cloud.example.org.
openstack server create --image <image> --flavor <flavor> --network <network> test-dns
openstack recordset list cloud.example.org.                      # an A record for test-dns

On a self-service (tenant) network, Neutron publishes fixed-IP records only for subnets with dns_publish_fixed_ips set, the attribute the subnet_dns_publish_fixed_ip extension adds (default False; Neutron's guide, use case 3a). Without it, only provider networks and floating IPs get records, and the A record above does not appear. For the test on a tenant network:

openstack subnet set --dns-publish-fixed-ip <subnet>    # before creating the instance

4. Check each hop (read-only):

dig +short -p 5354 @<mdns-host-dns-address> test-dns.cloud.example.org. A   # mDNS, what Designate serves
dig +short @<bind9-host-address> test-dns.cloud.example.org. A              # BIND9, the authoritative answer
dig +short test-dns.cloud.example.org. A                                    # through your resolvers: needs the delegation
dig +short NS cloud.example.org.                                            # the names from designate_ns_record

When the zone stays PENDING, Designate could not push it to BIND9: the worker, mDNS and BIND9 logs below say which hop failed.

designate-sink (optional)

The sink creates records from Nova and Neutron notifications, the older alternative to Neutron's integration. It is off by default (designate_enable_notifications_sink: true enables it) and needs the zone ID in /etc/kolla/config/designate/designate-sink.conf under [handler:nova_fixed] and [handler:neutron_floatingip], then reconfigure --tags designate,neutron,nova (guide). Use one mechanism, not both, so records are not created twice: the guide's own example output shows two A records for the same instance.

Where to look when it fails

Symptom First log to read
Zone stays PENDING or goes to ERROR /var/log/kolla/designate/designate-worker.log: rndc to BIND9 on port 953 (authentication uses designate_rndc_key from passwords.yml, rendered into rndc.key for the worker and BIND9), or the pool targets
BIND9 has no copy of the zone /var/log/kolla/designate/designate-mdns.log (zone transfers) and /var/log/kolla/named/ on the BIND9 host
API errors /var/log/kolla/designate/designate-api.log, designate-central.log
No record for a new instance /var/log/kolla/neutron/neutron-server.log: the external DNS driver's calls to Designate; the network's dns_domain and the port's dns_name
Name resolves on BIND9 but not from outside the delegation and the designate_ns_record names in the parent zone; port 53 reachability on dns_interface
A container does not start docker logs <container> (or podman logs)

Key choices and their consequences

Choice Alternative(s) Why this one here Consequence in operation
BIND9 deployed by Kolla-Ansible An external BIND9 (designate_backend_external) or Infoblox Nothing else to operate Your zones are served from the controllers; their reachability on port 53 is yours to provide
Neutron DNS integration designate-sink The default; records follow ports and floating IPs Records only for networks with a dns_domain and ports with a dns_name
Authoritative only (designate_recursion: "no") Recursion with forwarders Avoids an open resolver Instances keep their own resolvers
Valkey enabled One designate_worker Required by the guide with several workers One more service to run

What went wrong during deployment

Nothing observed yet: written from the Kolla-Ansible 2026.1 Designate guide and the designate and neutron roles.

Version notes

  • 2026.1: coordination for several workers is Valkey only (Redis was replaced by Valkey in Kolla-Ansible).
  • The Infoblox back-end needs the mDNS container to listen on port 53 and therefore to be rebuilt to run as root (guide); not covered here.

Lessons learned

No first-hand lessons yet. What the sources make clear:

  • Set designate_ns_record and arrange the delegation before the first zone: the NS records of every zone come from it.
  • Check each hop with dig: mDNS on 5354, BIND9 on 53, then your resolvers.
  • Choose Neutron's integration or the sink, not both.

Validity and limits

Written from Kolla-Ansible 2026.1 (Designate guide, ansible/group_vars/all/designate.yml, the designate role and its named.conf.j2 and pools.yaml.j2 templates, the neutron role defaults and templates, the sample multinode inventory) and the Neutron DNS integration guide. Not run on a lab: no output on this page was observed. Not covered: Infoblox and external BIND9 back-ends, DNSSEC signing of zones, reverse zones for floating IPs, several pools, and Designate quotas and tenancy.


Provenance: written by agent:ozarland-author, from the Kolla-Ansible 2026.1 documentation and code, and the Neutron documentation. Reviewed in its pull request (see the page history). Checked by a human operator: not yet.

Worked for you, or not? Tell us in one line: your release, your topology, and what happened.