dns: dual-write toggles + drift fact

Publish records both ways during the k8s cutover, and expose expected vs
deployed records for drift detection.

- profiles::dns::updater + ::record: manage_nsupdate and manage_export
  booleans (both default on); export keeps the legacy master flow, so
  disable it once k8s is authoritative
- dns_records fact: parses the expected records file and digs the
  authoritative server for each, reporting expected / in_sync / drift
  (plus dns_records_insync boolean); updater writes the server address
  to /var/lib/dns-updater/server for the fact
- hiera: manage_export/manage_nsupdate = true (cutover)

hieradata/common.yaml

@@ -208,6 +208,20 @@ vault::disable_mlock: false
 profiles::dns::base::nameservers:
   - 198.18.19.16
 profiles::dns::master::basedir: '/var/named/sources'
+
+# dns record publishing. During the k8s cutover both methods run; set
+# manage_export false once k8s is authoritative.
+#   - export:   legacy exported-resources -> puppet DNS master
+#   - nsupdate: RFC2136 to the k8s bind-authoritative write endpoint (.9),
+#               inert until the TSIG key is set in eyaml:
+#                 profiles::dns::updater::key_secret: ENC[...]
+#               (must match the key the bind-authoritative zones allow-update
+#               with; algorithm hmac-sha256)
+profiles::dns::updater::manage_export: true
+profiles::dns::updater::manage_nsupdate: true
+profiles::dns::updater::server: '198.18.200.9'
+profiles::dns::updater::key_name: 'client-update'
+profiles::dns::updater::key_algorithm: 'hmac-sha256'
 #profiles::dns::base::ns_role: 'roles::infra::dns::resolver'
 #profiles::dns::base::use_ns: 'region'
 profiles::consul::server::members_role: roles::infra::storage::consul

hieradata/roles/infra/k8s.yaml

@@ -12,6 +12,61 @@ hiera_include:
 rke2::bootstrap_node: prodnxsr0001.main.unkin.net
 rke2::join_url: https://join-k8s.service.consul:9345
 rke2::manage_registries: true
+rke2::registries:
+  docker.io:
+    endpoint:
+      - "https://artifactapi.k8s.syd1.au.unkin.net"
+    rewrite:
+      "^(.*)$": "dockerhub/$1"
+    disable-default-registry-endpoint: true
+  ghcr.io:
+    endpoint:
+      - "https://artifactapi.k8s.syd1.au.unkin.net"
+    rewrite:
+      "^(.*)$": "ghcr/$1"
+    disable-default-registry-endpoint: true
+  quay.io:
+    endpoint:
+      - "https://artifactapi.k8s.syd1.au.unkin.net"
+    rewrite:
+      "^(.*)$": "quay/$1"
+    disable-default-registry-endpoint: true
+  registry.k8s.io:
+    endpoint:
+      - "https://artifactapi.k8s.syd1.au.unkin.net"
+    rewrite:
+      "^(.*)$": "k8s-registry/$1"
+    disable-default-registry-endpoint: true
+  registry.gitlab.com:
+    endpoint:
+      - "https://artifactapi.k8s.syd1.au.unkin.net"
+    rewrite:
+      "^(.*)$": "gitlab/$1"
+    disable-default-registry-endpoint: true
+  docker.elastic.co:
+    endpoint:
+      - "https://artifactapi.k8s.syd1.au.unkin.net"
+    rewrite:
+      "^(.*)$": "elastic/$1"
+    disable-default-registry-endpoint: true
+  gcr.io:
+    endpoint:
+      - "https://artifactapi.k8s.syd1.au.unkin.net"
+    rewrite:
+      "^(.*)$": "gcr/$1"
+    disable-default-registry-endpoint: true
+  docker.litellm.ai:
+    endpoint:
+      - "https://artifactapi.k8s.syd1.au.unkin.net"
+    rewrite:
+      "^(.*)$": "litellm/$1"
+    disable-default-registry-endpoint: true
+  public.ecr.aws:
+    endpoint:
+      - "https://artifactapi.k8s.syd1.au.unkin.net"
+    rewrite:
+      "^(.*)$": "ecr-public/$1"
+    disable-default-registry-endpoint: true
 rke2::config_hash:
   bind-address: "%{hiera('networking_loopback0_ip')}"
   node-ip: "%{hiera('networking_loopback0_ip')}"

modules/libs/lib/facter/dns_records.rb

@@ -0,0 +1,88 @@
+# frozen_string_literal: true
+
+# lib/facter/dns_records.rb
+#
+# Reports this host's expected DNS records (assembled by profiles::dns::updater
+# into its records file) versus what is currently deployed on the authoritative
+# server, so puppet can detect drift and re-apply.
+#
+# Structured value:
+#   { server, count, expected => [{zone,fqdn,type,ttl,value}], in_sync,
+#     drift => [{...,deployed => [...]}] }
+
+# Helpers for the dns_records fact.
+module DnsRecordsFact
+  RECORDS_FILE = '/var/lib/dns-updater/records'
+  SERVER_FILE  = '/var/lib/dns-updater/server'
+
+  module_function
+
+  # normalise a value for comparison: strip, drop trailing dot, downcase
+  def norm(value)
+    value.to_s.strip.chomp('.').downcase
+  end
+
+  def server
+    File.exist?(SERVER_FILE) ? File.read(SERVER_FILE).strip : nil
+  end
+
+  # a name relative to a zone (or @) as a fully-qualified name
+  def to_fqdn(name, zone)
+    return "#{zone}." if name.to_s.empty? || name == '@'
+
+    "#{name}.#{zone}."
+  end
+
+  # parse one "zone|name|type|ttl|value" line into a record hash (nil to skip)
+  def parse_line(line)
+    line = line.strip
+    return nil if line.empty? || line.start_with?('#')
+
+    zone, name, type, ttl, value = line.split('|', 5)
+    return nil unless zone && type && value
+
+    { 'zone' => zone, 'fqdn' => to_fqdn(name, zone), 'type' => type, 'ttl' => ttl, 'value' => value }
+  end
+
+  # parse the records file into record hashes
+  def expected
+    return [] unless File.exist?(RECORDS_FILE)
+
+    File.readlines(RECORDS_FILE).filter_map { |line| parse_line(line) }
+  end
+
+  # the values currently deployed for a record, per the authoritative server
+  def deployed(record, srv)
+    cmd = ['dig', '+short', '+time=2', '+tries=1']
+    cmd << "@#{srv}" if srv && !srv.empty?
+    cmd += [record['fqdn'], record['type']]
+    out = Facter::Core::Execution.execute(cmd.join(' '), on_fail: '')
+    out.to_s.split("\n").map { |line| norm(line) }.reject(&:empty?)
+  end
+
+  def report
+    srv = server
+    exp = expected
+    drift = exp.filter_map do |record|
+      dep = deployed(record, srv)
+      record.merge('deployed' => dep) unless dep.include?(norm(record['value']))
+    end
+    { 'server' => srv, 'count' => exp.length, 'expected' => exp, 'in_sync' => drift.empty?, 'drift' => drift }
+  end
+end
+
+Facter.add(:dns_records) do
+  confine kernel: 'Linux'
+  setcode do
+    File.exist?(DnsRecordsFact::RECORDS_FILE) ? DnsRecordsFact.report : nil
+  end
+end
+
+# Convenience boolean for `if $facts['dns_records_insync']` guards.
+Facter.add(:dns_records_insync) do
+  confine kernel: 'Linux'
+  setcode do
+    v = Facter.value(:dns_records)
+    v.nil? ? nil : v['in_sync']
+  end
+end

modules/rke2/files/registries.yaml

@@ -1,38 +0,0 @@
----
-# DO NOT MODIFY - MANAGED BY PUPPET
-mirrors:
-  docker.io:
-    endpoint:
-      - "https://artifactapi.k8s.syd1.au.unkin.net"
-    rewrite:
-      "^(.*)$": "dockerhub/$1"
-  ghcr.io:
-    endpoint:
-      - "https://artifactapi.k8s.syd1.au.unkin.net"
-    rewrite:
-      "^(.*)$": "ghcr/$1"
-  quay.io:
-    endpoint:
-      - "https://artifactapi.k8s.syd1.au.unkin.net"
-    rewrite:
-      "^(.*)$": "quay/$1"
-  registry.k8s.io:
-    endpoint:
-      - "https://artifactapi.k8s.syd1.au.unkin.net"
-    rewrite:
-      "^(.*)$": "k8s-registry/$1"
-  registry.gitlab.com:
-    endpoint:
-      - "https://artifactapi.k8s.syd1.au.unkin.net"
-    rewrite:
-      "^(.*)$": "gitlab/$1"
-  docker.elastic.co:
-    endpoint:
-      - "https://artifactapi.k8s.syd1.au.unkin.net"
-    rewrite:
-      "^(.*)$": "elastic/$1"
-  gcr.io:
-    endpoint:
-      - "https://artifactapi.k8s.syd1.au.unkin.net"
-    rewrite:
-      "^(.*)$": "gcr/$1"

modules/rke2/manifests/config.pp

@@ -8,6 +8,7 @@ class rke2::config (
   String                  $node_token = $rke2::node_token,
   Array[String[1]]        $extra_config_files = $rke2::extra_config_files,
   Boolean                 $manage_registries = $rke2::manage_registries,
+  Hash                    $registries = $rke2::registries,
 ){
 
   # if its not the bootstrap node, add join path to config
@@ -35,7 +36,7 @@ class rke2::config (
       owner   => 'root',
       group   => 'root',
       mode    => '0644',
-      source  => 'puppet:///modules/rke2/registries.yaml',
+      content => epp('rke2/registries.yaml.epp', { registries => $registries }),
       require => Package["rke2-${node_type}"],
       notify  => Service["rke2-${node_type}"],
     }

modules/rke2/manifests/init.pp

@@ -13,6 +13,7 @@ class rke2 (
   Array[String[1]]        $extra_config_files = $rke2::params::extra_config_files,
   Stdlib::HTTPUrl         $container_archive_source = $rke2::params::container_archive_source,
   Boolean                 $manage_registries = $rke2::params::manage_registries,
+  Hash                    $registries = $rke2::params::registries,
 ) inherits rke2::params {
 
   include rke2::install

modules/rke2/manifests/params.pp

@@ -13,4 +13,5 @@ class rke2::params (
   Array[String[1]]        $extra_config_files = [],
   Stdlib::HTTPUrl         $container_archive_source = 'https://github.com/rancher/rke2/releases/download',
   Boolean                 $manage_registries = false,
+  Hash                    $registries = {},
 ) {}

modules/rke2/templates/registries.yaml.epp

@@ -0,0 +1,20 @@
+<%- | Hash $registries | -%>
+---
+# DO NOT MODIFY - MANAGED BY PUPPET
+mirrors:
+<%- $registries.each |$registry, $config| { -%>
+  <%= $registry %>:
+    endpoint:
+<%- $config['endpoint'].each |$ep| { -%>
+      - "<%= $ep %>"
+<%- } -%>
+<%- if $config['rewrite'] { -%>
+    rewrite:
+<%- $config['rewrite'].each |$pattern, $replacement| { -%>
+      "<%= $pattern %>": "<%= $replacement %>"
+<%- } -%>
+<%- } -%>
+<%- if $config['disable-default-registry-endpoint'] { -%>
+    disable-default-registry-endpoint: true
+<%- } -%>
+<%- } -%>

site/profiles/manifests/dns/base.pp

@@ -11,9 +11,12 @@ class profiles::dns::base (
   Optional[String] $ns_role = undef,
 ){
 
-  # install bind_utils
+  # install bind_utils (provides nsupdate)
   include bind::updater
 
+  # assemble the host's DNS records and nsupdate them to the authoritative server
+  include profiles::dns::updater
+
   # if ns_role is set, find all hosts matching that enc_role
   $nameserver_array = $ns_role ? {
     undef   => $nameservers,

site/profiles/manifests/dns/record.pp

@@ -1,4 +1,10 @@
-# defines the base record that will be exported
+# profiles::dns::record
+#
+# Declares a DNS record for this host. Publishes it via either or both methods,
+# controlled by profiles::dns::updater's toggles (both on during cutover):
+#   - nsupdate: a local concat fragment consumed by profiles::dns::updater,
+#     which nsupdates it to the authoritative server.
+#   - export: the legacy @@concat::fragment exported to the puppet DNS master.
 define profiles::dns::record (
   String  $record,
   Enum[
@@ -13,11 +19,26 @@ define profiles::dns::record (
   String  $value,
   String  $zone,
   Integer $order,
-  Stdlib::AbsolutePath $basedir = lookup('profiles::dns::master::basedir'),
+  Integer $ttl = 300,
 ) {
+  include profiles::dns::updater
+
+  # new: local records file consumed by the nsupdate service
+  if $profiles::dns::updater::manage_nsupdate {
+    # zone|name|type|ttl|value  (parsed by the dns-update script)
+    concat::fragment { "dns-record-${name}":
+      target  => $profiles::dns::updater::records_file,
+      content => "${zone}|${record}|${type}|${ttl}|${value}\n",
+      order   => sprintf('%03d', $order),
+    }
+  }
+
+  # legacy: export the fragment to the puppet DNS master
+  if $profiles::dns::updater::manage_export {
     @@concat::fragment { "${zone}_${name}":
-    target  => "${basedir}/${zone}.conf",
+      target  => "${profiles::dns::updater::master_basedir}/${zone}.conf",
       content => "${record} IN ${type} ${value}\n",
       order   => $order,
     }
+  }
 }

site/profiles/manifests/dns/updater.pp

@@ -0,0 +1,127 @@
+# profiles::dns::updater
+#
+# Publishes this host's DNS records. Two methods, independently toggled so both
+# can run during the k8s cutover (profiles::dns::record honours the same flags):
+#
+#   - nsupdate ($manage_nsupdate): assemble the records into a local file and
+#     nsupdate them to the k8s authoritative write endpoint via a systemd .path
+#     unit that watches the file. Inert until $key_secret (TSIG) is set.
+#   - export ($manage_export): the legacy exported-resources flow to the puppet
+#     DNS master. Kept during cutover; disable once k8s is authoritative.
+#
+# nsupdate comes from bind-utils (installed via bind::updater in
+# profiles::dns::base).
+class profiles::dns::updater (
+  Boolean                     $manage_nsupdate = true,
+  Boolean                     $manage_export   = true,
+  String                      $server          = '198.18.200.9',
+  String                      $key_name        = 'client-update',
+  String                      $key_algorithm   = 'hmac-sha256',
+  Optional[Sensitive[String]] $key_secret      = undef,
+  Integer                     $default_ttl     = 300,
+  Stdlib::AbsolutePath        $records_file    = '/var/lib/dns-updater/records',
+  Stdlib::AbsolutePath        $state_dir       = '/var/lib/dns-updater',
+  Stdlib::AbsolutePath        $config_dir      = '/etc/dns-updater',
+  Stdlib::AbsolutePath        $master_basedir  = lookup('profiles::dns::master::basedir'),
+) {
+
+  $state_file  = "${state_dir}/applied"
+  $server_file = "${state_dir}/server"
+  $key_file    = "${config_dir}/key"
+
+  if $manage_nsupdate {
+
+    file { $state_dir:
+      ensure => directory,
+      owner  => 'root',
+      group  => 'root',
+      mode   => '0755',
+    }
+
+    # Server address, read by the dns_records fact for drift detection.
+    file { $server_file:
+      ensure  => file,
+      owner   => 'root',
+      group   => 'root',
+      mode    => '0644',
+      content => "${server}\n",
+      require => File[$state_dir],
+    }
+
+    # Records file, assembled from profiles::dns::record fragments.
+    concat { $records_file:
+      ensure         => present,
+      owner          => 'root',
+      group          => 'root',
+      mode           => '0644',
+      ensure_newline => true,
+      warn           => false,
+      require        => File[$state_dir],
+    }
+
+    concat::fragment { 'dns-update-header':
+      target  => $records_file,
+      content => "# Managed by puppet (profiles::dns::record): zone|name|type|ttl|value\n",
+      order   => '00',
+    }
+
+    if $key_secret =~ Undef {
+      notify { 'dns-updater-inert':
+        message  => 'profiles::dns::updater: key_secret unset; records assembled but not applied.',
+        loglevel => 'info',
+      }
+    } else {
+      file { $config_dir:
+        ensure => directory,
+        owner  => 'root',
+        group  => 'root',
+        mode   => '0700',
+      }
+
+      file { $key_file:
+        ensure    => file,
+        owner     => 'root',
+        group     => 'root',
+        mode      => '0600',
+        show_diff => false,
+        content   => Sensitive(epp('profiles/dns/tsig-key.epp', {
+            'name'      => $key_name,
+            'algorithm' => $key_algorithm,
+            'secret'    => $key_secret.unwrap,
+        })),
+      }
+
+      file { '/usr/local/bin/dns-update':
+        ensure  => file,
+        owner   => 'root',
+        group   => 'root',
+        mode    => '0755',
+        content => epp('profiles/dns/dns-update.sh.epp', {
+            'server'       => $server,
+            'key_file'     => $key_file,
+            'records_file' => $records_file,
+            'state_file'   => $state_file,
+        }),
+      }
+
+      systemd::unit_file { 'dns-update.service':
+        content => epp('profiles/dns/dns-update.service.epp', { 'script' => '/usr/local/bin/dns-update' }),
+      }
+
+      # The .path unit watches the records file and triggers the service.
+      systemd::unit_file { 'dns-update.path':
+        content => epp('profiles/dns/dns-update.path.epp', { 'records_file' => $records_file }),
+        active  => true,
+        enable  => true,
+      }
+
+      # Also apply within the puppet run whenever the records change.
+      exec { 'dns-update-apply':
+        command     => '/usr/local/bin/dns-update',
+        refreshonly => true,
+        subscribe   => Concat[$records_file],
+        require     => [File['/usr/local/bin/dns-update'], File[$key_file]],
+      }
+    }
+  }
+}

site/profiles/templates/dns/dns-update.path.epp

@@ -0,0 +1,10 @@
+<%- | String $records_file | -%>
+[Unit]
+Description=Watch the DNS records file and apply changes
+
+[Path]
+PathModified=<%= $records_file %>
+Unit=dns-update.service
+
+[Install]
+WantedBy=multi-user.target

site/profiles/templates/dns/dns-update.service.epp

@@ -0,0 +1,9 @@
+<%- | String $script | -%>
+[Unit]
+Description=Apply host DNS records via nsupdate
+After=network-online.target
+Wants=network-online.target
+
+[Service]
+Type=oneshot
+ExecStart=<%= $script %>

site/profiles/templates/dns/dns-update.sh.epp

@@ -0,0 +1,56 @@
+<%- | String $server, String $key_file, String $records_file, String $state_file | -%>
+#!/bin/bash
+# Managed by puppet (profiles::dns::updater). Applies this host's records to the
+# authoritative DNS server via TSIG nsupdate. Only the delta since the last
+# successful run is sent; removed records are deleted.
+set -euo pipefail
+
+SERVER="<%= $server %>"
+KEYFILE="<%= $key_file %>"
+RECORDS="<%= $records_file %>"
+STATE="<%= $state_file %>"
+
+[ -f "$RECORDS" ] || exit 0
+touch "$STATE"
+
+# Format per line: zone|name|type|ttl|value  (name is relative to zone, or @).
+desired="$(grep -vE '^[[:space:]]*(#|$)' "$RECORDS" | sort -u || true)"
+applied="$(grep -vE '^[[:space:]]*(#|$)' "$STATE" 2>/dev/null | sort -u || true)"
+
+[ "$desired" = "$applied" ] && exit 0
+
+fqdn() { # name zone
+  if [ -z "$1" ] || [ "$1" = "@" ]; then printf '%s.' "$2"; else printf '%s.%s.' "$1" "$2"; fi
+}
+
+msg="$(mktemp)"
+trap 'rm -f "$msg"' EXIT
+printf 'server %s\n' "$SERVER" >> "$msg"
+
+# Process per zone so each UPDATE message targets a single zone.
+zones="$(printf '%s\n%s\n' "$desired" "$applied" | cut -d'|' -f1 | sort -u | grep -v '^$' || true)"
+for zone in $zones; do
+  printf 'zone %s.\n' "$zone" >> "$msg"
+  # Additions/updates: replace the RRset for every desired record in this zone.
+  printf '%s\n' "$desired" | awk -F'|' -v z="$zone" 'NF>=5 && $1==z' | \
+    while IFS='|' read -r z name type ttl value; do
+      f="$(fqdn "$name" "$z")"
+      printf 'update delete %s %s\n' "$f" "$type" >> "$msg"
+      printf 'update add %s %s %s %s\n' "$f" "$ttl" "$type" "$value" >> "$msg"
+    done
+  # Deletions: records present last run but gone now.
+  comm -23 <(printf '%s\n' "$applied") <(printf '%s\n' "$desired") | \
+    awk -F'|' -v z="$zone" 'NF>=5 && $1==z' | \
+    while IFS='|' read -r z name type ttl value; do
+      f="$(fqdn "$name" "$z")"
+      printf 'update delete %s %s %s\n' "$f" "$type" "$value" >> "$msg"
+    done
+  printf 'send\n' >> "$msg"
+done
+
+if nsupdate -k "$KEYFILE" "$msg"; then
+  printf '%s\n' "$desired" > "$STATE"
+else
+  echo "dns-update: nsupdate to ${SERVER} failed" >&2
+  exit 1
+fi

site/profiles/templates/dns/tsig-key.epp

@@ -0,0 +1,5 @@
+<%- | String $name, String $algorithm, String $secret | -%>
+key "<%= $name %>" {
+    algorithm <%= $algorithm %>;
+    secret "<%= $secret %>";
+};