vault: keep systemd-tmpfiles from reaping openbao plugin sockets

go-plugin gives each secrets plugin a control socket at /tmp/pluginNNN. The daily systemd-tmpfiles-clean deletes /tmp files older than the OS policy, which severs the socket of a long-lived plugin while its process keeps running. Every request then fails with rpc Unavailable / dial unix /tmp/pluginNNN: no such file, until a plugin reload respawns it. This hit the litellm engine (terraform-vault#112) but the risk is shared by every OpenBao plugin (gpg, rancher, gitea, ...). Ship a tmpfiles.d exclude (x /tmp/plugin*) on all vault servers so plugin sockets survive the cleaner. Takes effect on the next clean, no restart. Claude-Session: https://claude.ai/code/session_01JUoARVdmhxKQHyyyp1pxeT
chore: update dns nameserver in dhcp (#507 )
2026-08-06 22:31:58 +10:00 · 2026-08-06 22:23:51 +10:00 · 2026-08-02 17:25:18 +10:00 · 2026-07-30 20:50:30 +10:00 · 2026-07-28 21:42:58 +10:00 · 2026-07-28 21:42:32 +10:00
80 changed files with 1380 additions and 123 deletions
@@ -0,0 +1,18 @@
+when:
+  - event: pull_request
+
+steps:
+  - name: bolt-validate
+    image: git.unkin.net/unkin/almalinux9-puppet-pr-validator:20260317
+    commands:
+      - uvx pre-commit run --all-files --config ci/bolt-validate.yaml
+    backend_options:
+      kubernetes:
+        serviceAccountName: default
+        resources:
+          requests:
+            memory: 512Mi
+            cpu: 250m
+          limits:
+            memory: 2Gi
+            cpu: 1
@@ -0,0 +1,18 @@
+when:
+  - event: pull_request
+
+steps:
+  - name: epp-validate
+    image: git.unkin.net/unkin/almalinux9-puppet-pr-validator:20260317
+    commands:
+      - uvx pre-commit run --all-files --config ci/epp-validate.yaml
+    backend_options:
+      kubernetes:
+        serviceAccountName: default
+        resources:
+          requests:
+            memory: 512Mi
+            cpu: 250m
+          limits:
+            memory: 2Gi
+            cpu: 1
@@ -0,0 +1,18 @@
+when:
+  - event: pull_request
+
+steps:
+  - name: erb-validate
+    image: git.unkin.net/unkin/almalinux9-puppet-pr-validator:20260317
+    commands:
+      - uvx pre-commit run --all-files --config ci/erb-validate.yaml
+    backend_options:
+      kubernetes:
+        serviceAccountName: default
+        resources:
+          requests:
+            memory: 512Mi
+            cpu: 250m
+          limits:
+            memory: 2Gi
+            cpu: 1
@@ -1,10 +0,0 @@
-when:
-  - event: pull_request
-
-steps:
-  - name: pre-commit
-    image: git.unkin.net/unkin/almalinux9-base:latest
-    commands:
-      - dnf groupinstall -y "Development Tools" -y
-      - dnf install uv rubygems ruby-devel gcc make redhat-rpm-config glibc-headers glibc-devel libffi libffi-devel -y
-      - uvx pre-commit run --all-files
@@ -0,0 +1,18 @@
+when:
+  - event: pull_request
+
+steps:
+  - name: puppet-lint
+    image: git.unkin.net/unkin/almalinux9-puppet-pr-validator:20260317
+    commands:
+      - uvx pre-commit run --all-files --config ci/puppet-lint.yaml
+    backend_options:
+      kubernetes:
+        serviceAccountName: default
+        resources:
+          requests:
+            memory: 512Mi
+            cpu: 250m
+          limits:
+            memory: 2Gi
+            cpu: 1
@@ -0,0 +1,18 @@
+when:
+  - event: pull_request
+
+steps:
+  - name: puppet-validate
+    image: git.unkin.net/unkin/almalinux9-puppet-pr-validator:20260317
+    commands:
+      - uvx pre-commit run --all-files --config ci/puppet-validate.yaml
+    backend_options:
+      kubernetes:
+        serviceAccountName: default
+        resources:
+          requests:
+            memory: 512Mi
+            cpu: 1
+          limits:
+            memory: 2Gi
+            cpu: 2
@@ -0,0 +1,18 @@
+when:
+  - event: pull_request
+
+steps:
+  - name: ruby-check
+    image: git.unkin.net/unkin/almalinux9-puppet-pr-validator:20260317
+    commands:
+      - uvx pre-commit run --all-files --config ci/ruby-check.yaml
+    backend_options:
+      kubernetes:
+        serviceAccountName: default
+        resources:
+          requests:
+            memory: 512Mi
+            cpu: 250m
+          limits:
+            memory: 2Gi
+            cpu: 1
@@ -0,0 +1,18 @@
+when:
+  - event: pull_request
+
+steps:
+  - name: ruby-validate
+    image: git.unkin.net/unkin/almalinux9-puppet-pr-validator:20260317
+    commands:
+      - uvx pre-commit run --all-files --config ci/ruby-validate.yaml
+    backend_options:
+      kubernetes:
+        serviceAccountName: default
+        resources:
+          requests:
+            memory: 512Mi
+            cpu: 250m
+          limits:
+            memory: 2Gi
+            cpu: 1
@@ -0,0 +1,18 @@
+when:
+  - event: pull_request
+
+steps:
+  - name: yamllint
+    image: git.unkin.net/unkin/almalinux9-base:20260606
+    commands:
+      - uvx pre-commit run --all-files --config ci/yamllint.yaml
+    backend_options:
+      kubernetes:
+        serviceAccountName: default
+        resources:
+          requests:
+            memory: 512Mi
+            cpu: 250m
+          limits:
+            memory: 2Gi
+            cpu: 1
@@ -53,7 +53,6 @@ mod 'saz-ssh', '13.1.0'
 mod 'saz-limits', '5.0.0'
 mod 'ghoneycutt-timezone', '4.0.0'
 mod 'ghoneycutt-puppet', '3.3.0'
-mod 'dalen-puppetdbquery', '3.0.1'
 mod 'markt-galera', '3.1.0'
 mod 'kogitoapp-minio', '1.1.4'
 mod 'broadinstitute-certs', '3.0.1'
@@ -0,0 +1,5 @@
+repos:
+  - repo: 'https://github.com/chriskuehl/puppet-pre-commit-hooks.git'
+    rev: v2.2.0
+    hooks:
+      - id: bolt-validate
@@ -0,0 +1,5 @@
+repos:
+  - repo: 'https://github.com/chriskuehl/puppet-pre-commit-hooks.git'
+    rev: v2.2.0
+    hooks:
+      - id: epp-validate
@@ -0,0 +1,5 @@
+repos:
+  - repo: 'https://github.com/chriskuehl/puppet-pre-commit-hooks.git'
+    rev: v2.2.0
+    hooks:
+      - id: erb-validate
@@ -0,0 +1,10 @@
+repos:
+  - repo: 'https://github.com/chriskuehl/puppet-pre-commit-hooks.git'
+    rev: v2.2.0
+    hooks:
+      - id: puppet-lint
+        args:
+          - --no-80chars-check
+          - --no-documentation-check
+          - --no-puppet_url_without_modules-check
+          - --fail-on-warnings
@@ -0,0 +1,5 @@
+repos:
+  - repo: 'https://github.com/chriskuehl/puppet-pre-commit-hooks.git'
+    rev: v2.2.0
+    hooks:
+      - id: puppet-validate
@@ -0,0 +1,10 @@
+repos:
+  - repo: 'https://github.com/chriskuehl/puppet-pre-commit-hooks.git'
+    rev: v2.2.0
+    hooks:
+      - id: ruby-validate
+  - repo: 'https://github.com/jumanjihouse/pre-commit-hooks'
+    rev: 3.0.0
+    hooks:
+      - id: reek
+      - id: rubocop
@@ -0,0 +1,5 @@
+repos:
+  - repo: 'https://github.com/chriskuehl/puppet-pre-commit-hooks.git'
+    rev: v2.2.0
+    hooks:
+      - id: ruby-validate
@@ -0,0 +1,10 @@
+repos:
+  - repo: 'https://github.com/adrienverge/yamllint'
+    rev: v1.32.0
+    hooks:
+      - id: 'yamllint'
+        args:
+          [
+            "-d {extends: relaxed, rules: {line-length: disable}, ignore: chart}",
+            "-s",
+          ]
@@ -30,6 +30,7 @@ hierarchy:
      - "roles/%{::enc_role_tier1}.eyaml"
      - "roles/%{::enc_role_tier1}.yaml"
      - "virtual/%{facts.virtual}.yaml"
+      - "os/%{facts.os.name}/%{facts.os.name}%{facts.os.release.major}.%{facts.os.release.minor}.yaml"
      - "os/%{facts.os.name}/%{facts.os.name}%{facts.os.release.major}.yaml"
      - "os/%{facts.os.name}/all_releases.yaml"
      - "common.eyaml"
@@ -6,3 +6,4 @@ profiles::consul::token::node_editor::secret_id: ENC[PKCS7,MIIBmQYJKoZIhvcNAQcDo
 profiles::consul::server::acl_tokens_initial_management: ENC[PKCS7,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]
 profiles::consul::server::acl_tokens_default: ENC[PKCS7,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]
 profiles::consul::server::acl_tokens_replication: ENC[PKCS7,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]
+profiles::dns::updater::key_secret: ENC[PKCS7,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]
@@ -146,6 +146,8 @@ lookup_options:
      strategy: deep
  profiles::etcd::node::initial_cluster_token:
    convert_to: Sensitive
+  profiles::dns::updater::key_secret:
+    convert_to: Sensitive
  sysctl::base::values:
    merge:
      strategy: deep
@@ -206,8 +208,28 @@ vault::manage_config_dir: true
 vault::disable_mlock: false

 profiles::dns::base::nameservers:
-  - 198.18.19.16
+  - 198.18.200.7
 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)
+# k8s is now authoritative: clients resolve via the k8s bind-resolvers
+# (profiles::dns::base::nameservers 198.18.200.7, since #490), which forward the
+# unkin.net/main.unkin.net/*.18.198.in-addr.arpa zones to the k8s
+# bind-authoritative cluster fed by the nsupdate path below. The legacy
+# exported-resources -> VM-master /var/named/sources files are no longer in any
+# client's resolution path, so stop exporting them. Rollback: set back to true.
+profiles::dns::updater::manage_export: false
+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
@@ -2,7 +2,7 @@
 timezone::timezone: 'Australia/Sydney'
 certbot::client::webserver: ausyd1nxvm2057.main.unkin.net
 profiles_dns_upstream_forwarder_unkin:
-  - 198.18.19.15
+  - 198.18.200.7
 profiles_dns_upstream_forwarder_consul:
  - 198.18.19.14
 profiles_dns_upstream_forwarder_k8s:
@@ -1,4 +1,7 @@
 ---
+haproxy_server_k8s_syd1_traefik_internal: 'k8s-traefik-internal 198.18.200.4:443 ssl verify none check inter 2s rise 3 fall 2'
+haproxy_server_k8s_syd1_traefik_external: 'k8s-traefik-external 198.18.199.0:443 ssl verify none check inter 2s rise 3 fall 2'
+
 profiles::haproxy::dns::ipaddr: "%{hiera('anycast_ip')}"
 profiles::haproxy::dns::vrrp_cnames:
  - sonarr.main.unkin.net
@@ -16,6 +19,7 @@ profiles::haproxy::dns::vrrp_cnames:
  - mail.main.unkin.net
  - autoconfig.main.unkin.net
  - autodiscover.main.unkin.net
+  - auth.unkin.net

 profiles::haproxy::mappings:
  fe_http:
@@ -37,6 +41,7 @@ profiles::haproxy::mappings:
      - 'mail-webadmin.main.unkin.net be_stalwart_webadmin'
      - 'autoconfig.main.unkin.net be_stalwart_webadmin'
      - 'autodiscovery.main.unkin.net be_stalwart_webadmin'
+      - 'auth.unkin.net be_k8s_kanidm'
  fe_https:
    ensure: present
    mappings:
@@ -56,6 +61,7 @@ profiles::haproxy::mappings:
      - 'mail-webadmin.main.unkin.net be_stalwart_webadmin'
      - 'autoconfig.main.unkin.net be_stalwart_webadmin'
      - 'autodiscovery.main.unkin.net be_stalwart_webadmin'
+      - 'auth.unkin.net be_k8s_kanidm'

 profiles::haproxy::frontends:
  fe_http:
@@ -80,6 +86,7 @@ profiles::haproxy::frontends:
        - 'acl_stalwart_webadmin req.hdr(host) -i mail-webadmin.main.unkin.net'
        - 'acl_stalwart_webadmin req.hdr(host) -i autoconfig.main.unkin.net'
        - 'acl_stalwart_webadmin req.hdr(host) -i autodiscovery.main.unkin.net'
+        - 'acl_kanidm req.hdr(host) -i auth.unkin.net'
        - 'acl_internalsubnets src 198.18.0.0/16 10.10.12.0/24'
      use_backend:
        - "%[req.hdr(host),lower,map(/etc/haproxy/fe_https.map,be_default)]"
@@ -99,6 +106,7 @@ profiles::haproxy::frontends:
        - 'set-header X-Frame-Options DENY if acl_grafana'
        - 'set-header X-Frame-Options DENY if acl_ceph_dashboard'
        - 'set-header X-Frame-Options DENY if acl_stalwart_webadmin'
+        - 'set-header X-Frame-Options DENY if acl_kanidm'
        - 'set-header X-Content-Type-Options nosniff'
        - 'set-header X-XSS-Protection 1;mode=block'

@@ -320,6 +328,26 @@ profiles::haproxy::backends:
        - add-header X-Forwarded-Proto https if { dst_port 9443 }
      redirect: 'scheme https if !{ ssl_fc }'
      stick-table: 'type ip size 200k expire 30m'
+  be_k8s_kanidm:
+    description: Backend for Kanidm (auth.unkin.net via Kubernetes internal Traefik)
+    collect_exported: false
+    options:
+      balance: roundrobin
+      option:
+        - httpchk
+        - forwardfor
+        - http-keep-alive
+        - prefer-last-server
+      http-check:
+        - 'connect ssl sni auth.unkin.net'
+        - 'send meth GET uri /status ver HTTP/1.1 hdr Host auth.unkin.net'
+        - 'expect status 200'
+      http-reuse: always
+      http-request:
+        - set-header X-Forwarded-Port %[dst_port]
+        - add-header X-Forwarded-Proto https if { dst_port 443 }
+      redirect: 'scheme https if !{ ssl_fc }'
+      server: "%{lookup('haproxy_server_k8s_syd1_traefik_internal')} sni str(auth.unkin.net)"
  be_stalwart_imap:
    description: Backend for Stalwart IMAP (STARTTLS)
    collect_exported: false
@@ -393,6 +421,7 @@ profiles::haproxy::certlist::certificates:
  - /etc/pki/tls/letsencrypt/git.unkin.net/fullchain_combined.pem
  - /etc/pki/tls/letsencrypt/grafana.unkin.net/fullchain_combined.pem
  - /etc/pki/tls/letsencrypt/dashboard.ceph.unkin.net/fullchain_combined.pem
+  - /etc/pki/tls/letsencrypt/auth.unkin.net/fullchain_combined.pem
  - /etc/pki/tls/vault/certificate.pem

 # additional altnames
@@ -422,3 +451,4 @@ certbot::client::domains:
  - git.unkin.net
  - grafana.unkin.net
  - dashboard.ceph.unkin.net
+  - auth.unkin.net
@@ -0,0 +1,3 @@
+---
+# Canary for the puppet-on-k8s migration (roles::base test host).
+profiles::puppet::migrate::enabled: true
@@ -1,7 +1,7 @@
 # hieradata/os/AlmaLinux/AlmaLinux8.yaml
 ---
 crypto_policies::policy: 'DEFAULT'
-
+almalinux-base-repo: almalinux
 profiles::packages::include:
  network-scripts: {}

@@ -0,0 +1,2 @@
+---
+almalinux-base-repo: almalinux-vault
@@ -1,7 +1,7 @@
 # hieradata/os/AlmaLinux/AlmaLinux9.yaml
 ---
 crypto_policies::policy: 'DEFAULT:SHA1'
-
+almalinux-base-repo: almalinux
 profiles::yum::global::repos:
  crb:
    ensure: present
@@ -23,45 +23,45 @@ profiles::yum::global::repos:
    name: baseos
    descr: baseos repository
    target: /etc/yum.repos.d/baseos.repo
-    baseurl: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/almalinux/%{facts.os.release.full}/BaseOS/%{facts.os.architecture}/os/
-    gpgkey: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/almalinux/%{facts.os.release.full}/BaseOS/%{facts.os.architecture}/os/RPM-GPG-KEY-AlmaLinux-%{facts.os.release.major}
+    baseurl: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/%{lookup('almalinux-base-repo')}/%{facts.os.release.full}/BaseOS/%{facts.os.architecture}/os/
+    gpgkey: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/%{lookup('almalinux-base-repo')}/%{facts.os.release.full}/BaseOS/%{facts.os.architecture}/os/RPM-GPG-KEY-AlmaLinux-%{facts.os.release.major}
    mirrorlist: absent
  extras:
    name: extras
    descr: extras repository
    target: /etc/yum.repos.d/extras.repo
-    baseurl: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/almalinux/%{facts.os.release.full}/extras/%{facts.os.architecture}/os/
-    gpgkey: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/almalinux/%{facts.os.release.full}/extras/%{facts.os.architecture}/os/RPM-GPG-KEY-AlmaLinux-%{facts.os.release.major}
+    baseurl: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/%{lookup('almalinux-base-repo')}/%{facts.os.release.full}/extras/%{facts.os.architecture}/os/
+    gpgkey: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/%{lookup('almalinux-base-repo')}/%{facts.os.release.full}/extras/%{facts.os.architecture}/os/RPM-GPG-KEY-AlmaLinux-%{facts.os.release.major}
    mirrorlist: absent
  appstream:
    name: appstream
    descr: appstream repository
    target: /etc/yum.repos.d/appstream.repo
-    baseurl: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/almalinux/%{facts.os.release.full}/AppStream/%{facts.os.architecture}/os/
-    gpgkey: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/almalinux/%{facts.os.release.full}/AppStream/%{facts.os.architecture}/os/RPM-GPG-KEY-AlmaLinux-%{facts.os.release.major}
+    baseurl: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/%{lookup('almalinux-base-repo')}/%{facts.os.release.full}/AppStream/%{facts.os.architecture}/os/
+    gpgkey: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/%{lookup('almalinux-base-repo')}/%{facts.os.release.full}/AppStream/%{facts.os.architecture}/os/RPM-GPG-KEY-AlmaLinux-%{facts.os.release.major}
    mirrorlist: absent
  highavailability:
    name: highavailability
    descr: highavailability repository
    target: /etc/yum.repos.d/highavailability.repo
-    baseurl: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/almalinux/%{facts.os.release.full}/HighAvailability/%{facts.os.architecture}/os/
-    gpgkey: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/almalinux/%{facts.os.release.full}/HighAvailability/%{facts.os.architecture}/os/RPM-GPG-KEY-AlmaLinux-%{facts.os.release.major}
+    baseurl: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/%{lookup('almalinux-base-repo')}/%{facts.os.release.full}/HighAvailability/%{facts.os.architecture}/os/
+    gpgkey: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/%{lookup('almalinux-base-repo')}/%{facts.os.release.full}/HighAvailability/%{facts.os.architecture}/os/RPM-GPG-KEY-AlmaLinux-%{facts.os.release.major}
    mirrorlist: absent
  crb:
    ensure: absent
    name: crb
    descr: crb repository
    target: /etc/yum.repos.d/crb.repo
-    baseurl: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/almalinux/%{facts.os.release.full}/CRB/%{facts.os.architecture}/os/
-    gpgkey: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/almalinux/%{facts.os.release.full}/CRB/%{facts.os.architecture}/os/RPM-GPG-KEY-AlmaLinux-%{facts.os.release.major}
+    baseurl: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/%{lookup('almalinux-base-repo')}/%{facts.os.release.full}/CRB/%{facts.os.architecture}/os/
+    gpgkey: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/%{lookup('almalinux-base-repo')}/%{facts.os.release.full}/CRB/%{facts.os.architecture}/os/RPM-GPG-KEY-AlmaLinux-%{facts.os.release.major}
    mirrorlist: absent
  powertools:
    ensure: absent
    name: powertools
    descr: powertools repository
    target: /etc/yum.repos.d/powertools.repo
-    baseurl: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/almalinux/%{facts.os.release.full}/PowerTools/%{facts.os.architecture}/os/
-    gpgkey: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/almalinux/%{facts.os.release.full}/PowerTools/%{facts.os.architecture}/os/RPM-GPG-KEY-AlmaLinux-%{facts.os.release.major}
+    baseurl: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/%{lookup('almalinux-base-repo')}/%{facts.os.release.full}/PowerTools/%{facts.os.architecture}/os/
+    gpgkey: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/%{lookup('almalinux-base-repo')}/%{facts.os.release.full}/PowerTools/%{facts.os.architecture}/os/RPM-GPG-KEY-AlmaLinux-%{facts.os.release.major}
    mirrorlist: absent
  epel:
    name: epel
@@ -70,12 +70,34 @@ profiles::yum::global::repos:
    baseurl: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/epel/%{facts.os.release.major}/Everything/%{facts.os.architecture}
    gpgkey: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/epel/RPM-GPG-KEY-EPEL-%{facts.os.release.major}
    mirrorlist: absent
-  unkinben:
-    name: unkinben
-    descr: unkinben repository
-    target: /etc/yum.repos.d/unkin.repo
-    baseurl: https://git.query.consul/api/packages/unkinben/rpm/el%{facts.os.release.major}
-    gpgkey: https://git.query.consul/api/packages/unkinben/rpm/repository.key
+  rpm-internal:
+    name: rpm-internal
+    descr: rpm-internal repository
+    target: /etc/yum.repos.d/rpm-internal.repo
+    baseurl: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/local/rpm-internal/
+    gpgcheck: false
+    mirrorlist: absent
+  rpm-vendor:
+    name: rpm-vendor
+    descr: rpm-vendor repository
+    target: /etc/yum.repos.d/rpm-vendor.repo
+    baseurl: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/local/rpm-vendor/
+    gpgcheck: false
+    mirrorlist: absent
+  # Per-release variants, resolved from the host's EL major version so el8
+  # hosts pull rpm-internal-el8/rpm-vendor-el8, el9 hosts el9, etc.
+  rpm-internal-release:
+    name: rpm-internal-el%{facts.os.release.major}
+    descr: rpm-internal-el%{facts.os.release.major} repository
+    target: /etc/yum.repos.d/rpm-internal-release.repo
+    baseurl: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/local/rpm-internal-el%{facts.os.release.major}/
+    gpgcheck: false
+    mirrorlist: absent
+  rpm-vendor-release:
+    name: rpm-vendor-el%{facts.os.release.major}
+    descr: rpm-vendor-el%{facts.os.release.major} repository
+    target: /etc/yum.repos.d/rpm-vendor-release.repo
+    baseurl: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/local/rpm-vendor-el%{facts.os.release.major}/
    gpgcheck: false
    mirrorlist: absent

@@ -247,8 +269,8 @@ profiles::yum::global::repos:
    name: rpmfusion-free
    descr: rpmfusion-free repository
    target: /etc/yum.repos.d/rpmfusion-free.repo
-    baseurl: https://packagerepo.service.consul/rpmfusion-free-el%{facts.os.release.major}-%{facts.os.architecture}/
-    gpgkey: https://packagerepo.service.consul/rpmfusion-free-el%{facts.os.release.major}-%{facts.os.architecture}/repodata/repomd.xml.key
+    baseurl: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/rpmfusion-free/el/updates/%{facts.os.release.major}/%{facts.os.architecture}/
+    gpgkey: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/rpmfusion-free/el/RPM-GPG-KEY-rpmfusion-free-el-%{facts.os.release.major}
    gpgcheck: 1
    mirrorlist: absent
  rpmfusion-nonfree:
@@ -256,15 +278,7 @@ profiles::yum::global::repos:
    name: rpmfusion-nonfree
    descr: rpmfusion-nonfree repository
    target: /etc/yum.repos.d/rpmfusion-nonfree.repo
-    baseurl: https://packagerepo.service.consul/rpmfusion-nonfree-el%{facts.os.release.major}-%{facts.os.architecture}/
-    gpgkey: https://packagerepo.service.consul/rpmfusion-nonfree-el%{facts.os.release.major}-%{facts.os.architecture}/repodata/repomd.xml.key
+    baseurl: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/rpmfusion-nonfree/el/updates/%{facts.os.release.major}/%{facts.os.architecture}/
+    gpgkey: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/rpmfusion-nonfree/el/RPM-GPG-KEY-rpmfusion-nonfree-el-%{facts.os.release.major}
    gpgcheck: 1
    mirrorlist: absent
-  unkin:
-    name: unkin
-    descr: unkin repository
-    target: /etc/yum.repos.d/unkin.repo
-    baseurl: https://git.query.consul/api/packages/unkin/rpm/almalinux/el%{facts.os.release.major}
-    gpgkey: https://git.query.consul/api/packages/unkin/rpm/repository.key
-    gpgcheck: false
-    mirrorlist: absent
@@ -0,0 +1,69 @@
+# hieradata/os/Fedora/all_releases.yaml
+---
+# Fedora reuses the RedHat osfamily code paths (dnf/yum, crypto-policies,
+# firewalld, openvox). Everything here keys off facts.os.release.major so a
+# future Fedora release only needs its artifactapi local repos created, not new
+# hieradata.
+
+# crypto-policies: use the distro default. Kept here (not per-release) so newer
+# Fedora releases inherit it for free.
+crypto_policies::policy: 'DEFAULT'
+
+# Puppet agent via OpenVox. RedHat-family builds for Fedora live under
+# openvox7/fedora/<major> (see profiles::puppet::agent).
+profiles::puppet::agent::openvox_enable: true
+
+lm-sensors::package: lm_sensors
+
+# Fedora-specific base setup (NetworkManager, future workstation/laptop hooks).
+hiera_include:
+  - profiles::fedora::base
+
+# Base package set adjustments for Fedora:
+#   - p7zip was dropped from Fedora; the real 7-Zip ships as "7zip"
+#   - dstat was removed from Fedora with no drop-in successor in the base set
+#   - iotop is provided by the C rewrite package "iotop-c"
+profiles::packages::exclude:
+  - p7zip
+  - dstat
+  - iotop
+profiles::packages::include:
+  7zip: {}
+  iotop-c: {}
+  crypto-policies-scripts: {}
+  lzo: {}
+  policycoreutils: {}
+  unar: {}
+  xz: {}
+
+profiles::yum::global::repos:
+  fedora:
+    name: fedora
+    descr: Fedora %{facts.os.release.major} - %{facts.os.architecture}
+    target: /etc/yum.repos.d/fedora.repo
+    baseurl: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/fedora/releases/%{facts.os.release.major}/Everything/%{facts.os.architecture}/os/
+    gpgkey: file:///etc/pki/rpm-gpg/RPM-GPG-KEY-fedora-%{facts.os.release.major}-%{facts.os.architecture}
+    metalink: absent
+    mirrorlist: absent
+  updates:
+    name: updates
+    descr: Fedora %{facts.os.release.major} - %{facts.os.architecture} - Updates
+    target: /etc/yum.repos.d/fedora-updates.repo
+    baseurl: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/fedora/updates/%{facts.os.release.major}/Everything/%{facts.os.architecture}/
+    gpgkey: file:///etc/pki/rpm-gpg/RPM-GPG-KEY-fedora-%{facts.os.release.major}-%{facts.os.architecture}
+    metalink: absent
+    mirrorlist: absent
+  rpm-internal:
+    name: rpm-internal-f%{facts.os.release.major}
+    descr: rpm-internal-f%{facts.os.release.major} repository
+    target: /etc/yum.repos.d/rpm-internal.repo
+    baseurl: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/local/rpm-internal-f%{facts.os.release.major}/
+    gpgcheck: false
+    mirrorlist: absent
+  rpm-vendor:
+    name: rpm-vendor-f%{facts.os.release.major}
+    descr: rpm-vendor-f%{facts.os.release.major} repository
+    target: /etc/yum.repos.d/rpm-vendor.repo
+    baseurl: https://artifactapi.k8s.syd1.au.unkin.net/api/v1/local/rpm-vendor-f%{facts.os.release.major}/
+    gpgcheck: false
+    mirrorlist: absent
@@ -57,5 +57,3 @@ profiles::yum::global::repos:
    ensure: present
  rpmfusion-nonfree:
    ensure: present
-  unkinben:
-    ensure: present
@@ -14,6 +14,9 @@ profiles::pki::vault::alt_names:
  - radosgw.service.consul
  - radosgw.query.consul
  - "radosgw.service.%{facts.country}-%{facts.region}.consul"
+  # S3 endpoint name (s3.ceph.unkin.net CNAME -> radosgw.service.consul); needed
+  # on the cert so TLS validates for that hostname.
+  - s3.ceph.unkin.net

 # additional repos
 profiles::yum::global::repos:
@@ -26,6 +29,8 @@ profiles::yum::global::repos:
 profiles::nginx::simpleproxy::nginx_vhost: 'radosgw.service.consul'
 profiles::nginx::simpleproxy::nginx_aliases:
  - radosgw.service.au-syd1.consul
+  # nginx (443 -> radosgw:7480) must answer for the S3 endpoint name too.
+  - s3.ceph.unkin.net
 profiles::nginx::simpleproxy::proxy_port: 7480
 profiles::nginx::simpleproxy::proxy_path: '/'
 nginx::client_max_body_size: 5000M
@@ -6,8 +6,10 @@ hiera_include:
 profiles::dns::resolver::acls:
  acl-main.unkin.net:
    addresses:
-      - 10.10.8.1/32
+      - 198.18.1.10/32
+      - 198.18.2.160/27
      - 198.18.21.160/27
+      - 198.18.2.192/27
      - 198.18.21.192/27
      - 198.18.13.0/24
      - 198.18.14.0/24
@@ -82,8 +82,14 @@ profiles::sql::postgresdb::dbname: gitea
 profiles::sql::postgresdb::dbuser: gitea

 # deploy gitea
-gitea::ensure: '1.22.4'
-gitea::checksum: 'd549104f55067e6fb156e7ba060c9af488f36e12d5e747db7563fcc99eaf8532'
+gitea::base_url: 'https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/gitea-dl/gitea'
+gitea::install::checksums:
+  1.26.2:
+    linux:
+      amd64: 5b37937b625de69196748f7293eee1a7363f8637ae6e3da3c359bb380bd61a6a
+
+gitea::ensure: '1.26.2'
+gitea::checksum: '5b37937b625de69196748f7293eee1a7363f8637ae6e3da3c359bb380bd61a6a'
 gitea::manage_user: false
 gitea::manage_group: false
 gitea::manage_home: false
@@ -5,6 +5,10 @@ hiera_include:
  - incus
  - zfs
  - profiles::ceph::node
+  - profiles::ceph::mon
+  - profiles::ceph::mgr
+  - profiles::ceph::mds
+  - profiles::ceph::osd
  - profiles::ceph::client
  - profiles::ceph::dashboard
  - profiles::storage::cephfsvols
@@ -99,7 +103,7 @@ profiles::yum::global::repos:
 profiles::dns::base::primary_interface: loopback0

 # dashboard/haproxy
-profiles::ceph::dashboard::ipaddress: "%{hiera('networking_loopback0_ip')}"
+profiles::ceph::dashboard::ipaddress: "%{hiera('networking_loopback2_ip')}"

 # networking
 systemd::manage_networkd: true
@@ -2,6 +2,7 @@
 hiera_include:
  - profiles::selinux::setenforce
  - profiles::ceph::node
+  - profiles::ceph::osd
  - profiles::ceph::client
  - exporters::frr_exporter
  - frrouting
@@ -10,6 +11,62 @@ hiera_include:
 # manage rke2
 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')}"
@@ -125,6 +182,17 @@ frrouting::ospf_exclude_k8s_enable: true
 frrouting::k8s_cluster_cidr: '10.42.0.0/16'  # RKE2 cluster-cidr (pods)
 frrouting::k8s_service_cidr: '10.43.0.0/16'  # RKE2 service-cidr

+# sysctl recommendations
+sysctl::base::values:
+  net.ipv4.conf.default.rp_filter:
+    value: '0'
+  net.ipv4.conf.all.rp_filter:
+    value: '0'
+  fs.inotify.max_user_watches:
+    value: '524288'
+  fs.inotify.max_user_instances:
+    value: '512'
+
 # add loopback interfaces to ssh list
 ssh::server::options:
  ListenAddress:
@@ -11,6 +11,7 @@ profiles::metrics::grafana::db_name: "%{hiera('profiles::sql::postgresdb::dbname
 profiles::metrics::grafana::db_user: "%{hiera('profiles::sql::postgresdb::dbuser')}"
 profiles::metrics::grafana::db_pass: "%{hiera('profiles::sql::postgresdb::dbpass')}"
 profiles::metrics::grafana::pgsql_backend: true
+profiles::metrics::grafana::version: '13.0.2'
 profiles::metrics::grafana::plugins:
  victoriametrics-logs-datasource:
    ensure: present
@@ -16,3 +16,4 @@ certbot::domains:
  - git.unkin.net
  - grafana.unkin.net
  - dashboard.ceph.unkin.net
+  - auth.unkin.net
@@ -26,7 +26,7 @@ profiles::puppet::cobbler_enc::packages:
  - 'requests'
  - 'PyYAML'
 profiles::puppet::enc::repo: https://git.service.au-syd1.consul/unkinben/puppet-enc.git
-profiles::puppet::r10k::r10k_repo: https://git.service.au-syd1.consul/unkin/puppet-r10k.git
+profiles::puppet::r10k::r10k_repo: https://git.unkin.net/unkin/puppet-r10k.git
 profiles::puppet::g10k::bin_path: '/usr/bin/g10k'
 profiles::puppet::g10k::cfg_path: '/etc/puppetlabs/r10k/r10k.yaml'
 profiles::puppet::g10k::environments_path: '/etc/puppetlabs/code/environments'
@@ -27,4 +27,16 @@ profiles::nginx::simpleproxy::proxy_port: 8200
 profiles::nginx::simpleproxy::proxy_path: '/'

 profiles::packages::include:
+  # openbao-plugins (base bundle) left unpinned; it tracks the openbao package.
  openbao-plugins: {}
+  # Secrets plugins pinned to the exact version whose binary matches the sha256
+  # registered in terraform-vault (config/plugins/*.yaml). Bump both in lockstep
+  # on upgrade, or OpenBao refuses to launch the plugin after a restart.
+  openbao-plugin-secrets-litellm:
+    ensure: '0.1.1'
+  openbao-plugin-secrets-gpg:
+    ensure: '0.1.0'
+  openbao-plugin-secrets-rancher:
+    ensure: '0.1.1'
+  openbao-plugin-secrets-gitea:
+    ensure: '0.1.0'
@@ -28,8 +28,8 @@ class externaldns::master inherits externaldns {
      dynamic         => true,
      allow_updates   => ['key externaldns-key'],
      allow_transfers => empty($slave_ips) ? {
-        true  => [],
-        false => ['dns-slaves'],
+        true  => ['key externaldns-key'],
+        false => ['key externaldns-key','dns-slaves'],
      },
      ns_notify       => !empty($slave_ips),
      also_notify     => $slave_ips,
@@ -42,4 +42,4 @@ class externaldns::master inherits externaldns {
    recursion => false,
    zones     => $externaldns::k8s_zones,
  }
-}
+}
@@ -22,7 +22,12 @@ class incus::cluster (
    }

    # if it is, find hosts, sort them so they dont cause changes every run
-    $servers_array = sort(query_nodes("enc_role='${members_role}' and region='${facts['region']}'", 'networking.fqdn'))
+    $servers_array = sort(puppetdb_query(
+      "facts[certname] {
+        name = 'enc_role' and value = '${members_role}' and
+        certname in facts[certname] { name = 'region' and value = '${facts['region']}' }
+      }"
+    ).map |$fact| { $fact['certname'] })

  # else use provided array from params
  }else{
@@ -0,0 +1,23 @@
+# frozen_string_literal: true
+
+require 'facter'
+
+Facter.add(:ceph_osd_devices) do
+  confine kernel: 'Linux'
+  confine { Facter.value(:is_virtual) == false }
+  confine { Facter::Core::Execution.which('pvs') }
+  setcode do
+    devices = []
+    output = Facter::Core::Execution.execute(
+      'pvs --noheadings -o pv_name,vg_name 2>/dev/null',
+      on_fail: nil
+    )
+    output&.each_line do |line|
+      pv_name, vg_name = line.split
+      next unless pv_name && vg_name
+
+      devices << pv_name if vg_name.start_with?('ceph-')
+    end
+    devices.sort
+  end
+end
@@ -0,0 +1,56 @@
+# frozen_string_literal: true
+
+require 'facter'
+
+# Detects active ceph service instances via systemctl and exposes facts
+# for use in ceph service management profiles.
+# rubocop:disable Style/ClassAndModuleChildren
+module Unkin
+  module Ceph
+    # Detects active ceph service instances via systemctl and exposes Facter facts.
+    module Utils
+      TYPES = %w[mon mgr mds osd].freeze
+
+      def self.services
+        output = Facter::Core::Execution.execute(
+          'systemctl list-units "ceph*" --no-legend --plain --all 2>/dev/null',
+          on_fail: ''
+        )
+        parse_units(output)
+      end
+
+      def self.parse_units(output)
+        result = TYPES.each_with_object({}) { |type, hash| hash[type] = [] }
+        output.each_line do |line|
+          unit = line.split.first
+          next unless unit
+
+          match_unit(result, unit)
+        end
+        result
+      end
+
+      def self.match_unit(result, unit)
+        TYPES.each do |type|
+          match = unit.match(/\Aceph-#{type}@(.+)\.service\z/)
+          result[type] << "ceph-#{type}@#{match[1]}" if match
+        end
+      end
+
+      TYPES.each do |type|
+        define_singleton_method(:"#{type}?") { !services[type].empty? }
+      end
+    end
+  end
+end
+# rubocop:enable Style/ClassAndModuleChildren
+
+Facter.add('ceph_services') do
+  setcode { Unkin::Ceph::Utils.services }
+end
+
+Unkin::Ceph::Utils::TYPES.each do |type|
+  Facter.add("is_ceph_#{type}") do
+    setcode { Unkin::Ceph::Utils.public_send(:"#{type}?") }
+  end
+end
@@ -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
@@ -20,7 +20,12 @@ class redisha::redis (
    }

    # if it is, find hosts, sort them so they dont cause changes every run
-    $servers_array = sort(query_nodes("enc_role='${redisha_members_role}' and region='${facts['region']}'", 'networking.fqdn'))
+    $servers_array = sort(puppetdb_query(
+      "facts[certname] {
+        name = 'enc_role' and value = '${redisha_members_role}' and
+        certname in facts[certname] { name = 'region' and value = '${facts['region']}' }
+      }"
+    ).map |$fact| { $fact['certname'] })

  # else use provided array from params
  }else{
@@ -23,7 +23,12 @@ class redisha::sentinel (
    }

    # if it is, find hosts, sort them so they dont cause changes every run
-    $servers_array = sort(query_nodes("enc_role='${redisha_members_role}' and region='${facts['region']}'", 'networking.fqdn'))
+    $servers_array = sort(puppetdb_query(
+      "facts[certname] {
+        name = 'enc_role' and value = '${redisha_members_role}' and
+        certname in facts[certname] { name = 'region' and value = '${facts['region']}' }
+      }"
+    ).map |$fact| { $fact['certname'] })

  # else use provided array from params
  }else{
@@ -7,6 +7,8 @@ class rke2::config (
  Stdlib::Fqdn            $bootstrap_node = $rke2::bootstrap_node,
  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
@@ -28,6 +30,24 @@ class rke2::config (
    $config = $config_hash
  }

+  if $manage_registries {
+    file { '/etc/rancher/rke2/registries.yaml':
+      ensure  => file,
+      owner   => 'root',
+      group   => 'root',
+      mode    => '0644',
+      content => epp('rke2/registries.yaml.epp', { registries => $registries }),
+      require => Package["rke2-${node_type}"],
+      notify  => Service["rke2-${node_type}"],
+    }
+  }else{
+    file { '/etc/rancher/rke2/registries.yaml':
+      ensure  => absent,
+      require => Package["rke2-${node_type}"],
+      notify  => Service["rke2-${node_type}"],
+    }
+  }
+
  # create the config file
  file { $config_file:
    ensure  => file,
@@ -12,6 +12,8 @@ class rke2 (
  Hash                    $helm_repos = $rke2::params::helm_repos,
  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
@@ -12,4 +12,6 @@ class rke2::params (
  Hash                    $helm_repos = {},
  Array[String[1]]        $extra_config_files = [],
  Stdlib::HTTPUrl         $container_archive_source = 'https://github.com/rancher/rke2/releases/download',
+  Boolean                 $manage_registries = false,
+  Hash                    $registries = {},
 ) {}
@@ -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
+<%- } -%>
+<%- } -%>
@@ -167,7 +167,13 @@ class stalwart (

  # Query cluster members for validation
  $cluster_query = "enc_role='${cluster_role}' and country='${facts['country']}' and region='${facts['region']}'"
-  $cluster_members_raw = query_nodes($cluster_query, 'networking.fqdn')
+  $cluster_members_raw = puppetdb_query(
+    "facts[certname] {
+      name = 'enc_role' and value = '${cluster_role}' and
+      certname in facts[certname] { name = 'country' and value = '${facts['country']}' } and
+      certname in facts[certname] { name = 'region' and value = '${facts['region']}' }
+    }"
+  ).map |$fact| { $fact['certname'] }
  $cluster_members = $cluster_members_raw ? {
    undef   => [],
    default => $cluster_members_raw,
@@ -180,7 +186,20 @@ class stalwart (

  # Query HAProxy nodes for proxy trusted networks
  $haproxy_query = "enc_role='${haproxy_role}' and country='${facts['country']}' and region='${facts['region']}'"
-  $haproxy_members_raw = query_nodes($haproxy_query, 'networking.ip')
+  $haproxy_members_raw = puppetdb_query(
+    "facts[certname,value] {
+      name = 'networking' and
+      certname in facts[certname] {
+        name = 'enc_role' and value = '${haproxy_role}'
+      } and
+      certname in facts[certname] {
+        name = 'country' and value = '${facts['country']}'
+      } and
+      certname in facts[certname] {
+        name = 'region' and value = '${facts['region']}'
+      }
+    }"
+  ).map |$fact| { $fact['value']['ip'] }
  $haproxy_ips = $haproxy_members_raw ? {
    undef   => [],
    default => sort($haproxy_members_raw),
@@ -12,6 +12,9 @@ class profiles::base () {
    # manage the puppet agent
    include profiles::puppet::agent
    include profiles::puppet::client
+    # k8s migration switch; ships disabled, flip its ::enabled boolean in
+    # hiera (node/role/common) to repoint a target onto puppet-on-k8s.
+    include profiles::puppet::migrate

    # include the base profiles
    include profiles::base::repos
@@ -0,0 +1,13 @@
+class profiles::ceph::mds (
+  Boolean $ensure_running = true,
+) {
+
+  if $ensure_running and $facts['is_ceph_mds'] {
+    $facts['ceph_services']['mds'].each |String $svc| {
+      service { $svc:
+        ensure => running,
+        enable => true,
+      }
+    }
+  }
+}
@@ -0,0 +1,13 @@
+class profiles::ceph::mgr (
+  Boolean $ensure_running = true,
+) {
+
+  if $ensure_running and $facts['is_ceph_mgr'] {
+    $facts['ceph_services']['mgr'].each |String $svc| {
+      service { $svc:
+        ensure => running,
+        enable => true,
+      }
+    }
+  }
+}
@@ -0,0 +1,13 @@
+class profiles::ceph::mon (
+  Boolean $ensure_running = true,
+) {
+
+  if $ensure_running and $facts['is_ceph_mon'] {
+    $facts['ceph_services']['mon'].each |String $svc| {
+      service { $svc:
+        ensure => running,
+        enable => true,
+      }
+    }
+  }
+}
@@ -0,0 +1,13 @@
+class profiles::ceph::osd (
+  Boolean $ensure_running = true,
+) {
+
+  if $ensure_running and $facts['is_ceph_osd'] {
+    $facts['ceph_services']['osd'].each |String $svc| {
+      service { $svc:
+        ensure => running,
+        enable => true,
+      }
+    }
+  }
+}
@@ -68,7 +68,10 @@ class profiles::cobbler::config {
  # export cnames for cobbler
  if $is_cobbler_master {
    profiles::dns::record { "${::facts['networking']['fqdn']}_${service_cname}_CNAME":
-      value  => $::facts['networking']['hostname'],
+      # CNAME target must be a fully-qualified name (trailing dot): dns-updater
+      # parses the value with no $ORIGIN, so a bare label becomes root-absolute
+      # ("ausyd1nxvm2098.") and dead-ends in NXDOMAIN.
+      value  => "${::facts['networking']['fqdn']}.",
      type   => 'CNAME',
      record => "${service_cname}.",
      zone   => $::facts['networking']['domain'],
@@ -28,7 +28,12 @@ class profiles::consul::client (
      }

      # if it is, find hosts, sort them so they dont cause changes every run
-      $servers_array = sort(query_nodes("enc_role='${members_role}' and region='${::facts['region']}'", 'networking.fqdn'))
+      $servers_array = sort(puppetdb_query(
+        "facts[certname] {
+          name = 'enc_role' and value = '${members_role}' and
+          certname in facts[certname] { name = 'region' and value = '${::facts['region']}' }
+        }"
+      ).map |$fact| { $fact['certname'] })

    # else use provided array from params
    }else{
@@ -65,12 +65,22 @@ class profiles::consul::server (
      }

      # if it is, find hosts, sort them so they dont cause changes every run
-      $servers_array = sort(query_nodes("enc_role='${members_role}' and region='${::facts['region']}'", 'networking.fqdn'))
+      $servers_array = sort(puppetdb_query(
+        "facts[certname] {
+          name = 'enc_role' and value = '${members_role}' and
+          certname in facts[certname] { name = 'region' and value = '${::facts['region']}' }
+        }"
+      ).map |$fact| { $fact['certname'] })

      if $join_remote_regions {
        # get all nodes in the members_role for each other region
        $region_to_servers = $remote_regions.reduce({}) |$memo, $region| {
-          $servers = sort(query_nodes("enc_role='${members_role}' and region='${region}'", 'networking.fqdn'))
+          $servers = sort(puppetdb_query(
+            "facts[certname] {
+              name = 'enc_role' and value = '${members_role}' and
+              certname in facts[certname] { name = 'region' and value = '${region}' }
+            }"
+          ).map |$fact| { $fact['certname'] })
          $memo + { $region => $servers }
        }

@@ -11,16 +11,38 @@ 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,
    default => $use_ns ? {
-      'all'     => query_nodes("enc_role='${ns_role}'", 'networking.ip'),
-      'region'  => query_nodes("enc_role='${ns_role}' and region=${facts['region']}", 'networking.ip'),
-      'country' => query_nodes("enc_role='${ns_role}' and country=${facts['country']}", 'networking.ip'),
+      'all'     => puppetdb_query(
+        "facts[certname,value] {
+          name = 'networking' and
+          certname in nodes[certname] { facts.enc_role = '${ns_role}' }
+        }"
+      ).map |$fact| { $fact['value']['ip'] },
+      'region'  => puppetdb_query(
+        "facts[certname,value] {
+          name = 'networking' and
+          certname in nodes[certname] {
+            facts.enc_role = '${ns_role}' and facts.region = '${facts['region']}'
+          }
+        }"
+      ).map |$fact| { $fact['value']['ip'] },
+      'country' => puppetdb_query(
+        "facts[certname,value] {
+          name = 'networking' and
+          certname in nodes[certname] {
+            facts.enc_role = '${ns_role}' and facts.country = '${facts['country']}'
+          }
+        }"
+      ).map |$fact| { $fact['value']['ip'] },
    }
  }

@@ -20,9 +20,21 @@ class profiles::dns::master (
  $nameservers_array = $ns_role ? {
    undef   => [$facts['networking']['fqdn']],
    default => $use_ns ? {
-      'all'     => sort(query_nodes("enc_role='${ns_role}'", 'networking.fqdn')),
-      'region'  => sort(query_nodes("enc_role='${ns_role}' and region=${facts['region']}", 'networking.fqdn')),
-      'country' => sort(query_nodes("enc_role='${ns_role}' and country=${facts['country']}", 'networking.fqdn')),
+      'all'     => sort(puppetdb_query(
+        "facts[certname] { name = 'enc_role' and value = '${ns_role}' }"
+      ).map |$fact| { $fact['certname'] }),
+      'region'  => sort(puppetdb_query(
+        "facts[certname] {
+          name = 'enc_role' and value = '${ns_role}' and
+          certname in facts[certname] { name = 'region' and value = '${facts['region']}' }
+        }"
+      ).map |$fact| { $fact['certname'] }),
+      'country' => sort(puppetdb_query(
+        "facts[certname] {
+          name = 'enc_role' and value = '${ns_role}' and
+          certname in facts[certname] { name = 'country' and value = '${facts['country']}' }
+        }"
+      ).map |$fact| { $fact['certname'] }),
    }
  }

@@ -32,7 +44,9 @@ class profiles::dns::master (
      $facts['networking']['fqdn'] => $facts['networking']['ip']
    },
    default => $nameservers_array.reduce({}) |$acc, $fqdn| {
-      $result = query_nodes("networking.fqdn='${fqdn}'", 'networking.ip')
+      $result = puppetdb_query(
+        "facts[certname,value] { name = 'networking' and certname = '${fqdn}' }"
+      ).map |$fact| { $fact['value']['ip'] }
      $ip = $result[0]
      $acc + { "${fqdn}." => $ip }
    }
@@ -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 $ttl = 300,
 ) {
-  @@concat::fragment { "${zone}_${name}":
-    target  => "${basedir}/${zone}.conf",
-    content => "${record} IN ${type} ${value}\n",
-    order   => $order,
+  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}_${name}":
+      target  => "${profiles::dns::updater::master_basedir}/${zone}.conf",
+      content => "${record} IN ${type} ${value}\n",
+      order   => $order,
+    }
  }
 }
@@ -0,0 +1,173 @@
+# profiles::dns::updater
+#
+# Publishes this host's DNS records. Two methods, independently toggled so both
+# can run during the k8s cutover (profiles::dns::record honours the same flags):
+#
+#   - nsupdate ($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'),
+  # dns-updater daemon (replaces the dns-update shell script). 'latest' so hosts
+  # pick up new releases (e.g. the record filter); rpm-internal metadata_expire
+  # is 1h so this does not thrash.
+  String                      $package_ensure   = 'latest',
+  Stdlib::AbsolutePath        $api_socket       = '/run/dns-updater/api.sock',
+  String                      $resync           = '10m',
+  Enum['debug', 'info', 'warn', 'error'] $log_level = 'info',
+  Boolean                     $watch_interfaces = true,
+  # Never publish records whose address falls in these ranges: the k8s pod CIDR
+  # (10.42.0.0/16) and service CIDR (10.43.0.0/16), the LB VIP pool
+  # (198.18.200.0/24) and 198.18.199.0/24. Keeps kube-lb0/flannel/etc. out of
+  # the authoritative zones and stops NOTAUTH updates for zones we do not host.
+  Array[String]               $deny_ranges     = ['198.18.199.0/24', '198.18.200.0/24', '10.42.0.0/16', '10.43.0.0/16'],
+  Array[String]               $allow_ranges    = [],
+  Array[String]               $deny_domains    = [],
+  Array[String]               $allow_domains   = [],
+) {
+
+  $state_file  = "${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,
+        })),
+      }
+
+      # dns-updater daemon: watches the records file (inotify) and network
+      # interfaces, pushes TSIG-signed RFC2136 updates to $server natively.
+      package { 'dns-updater':
+        ensure => $package_ensure,
+      }
+
+      $deny_ranges_str   = join($deny_ranges, ',')
+      $allow_ranges_str  = join($allow_ranges, ',')
+      $deny_domains_str  = join($deny_domains, ',')
+      $allow_domains_str = join($allow_domains, ',')
+
+      $env_content = @("ENV")
+        # Managed by puppet (profiles::dns::updater).
+        DNS_UPDATER_SERVER=${server}
+        DNS_UPDATER_KEY_FILE=${key_file}
+        DNS_UPDATER_RECORDS_FILE=${records_file}
+        DNS_UPDATER_STATE_FILE=${state_file}
+        DNS_UPDATER_API=${api_socket}
+        DNS_UPDATER_RESYNC=${resync}
+        DNS_UPDATER_WATCH_INTERFACES=${watch_interfaces}
+        DNS_UPDATER_LOG_LEVEL=${log_level}
+        DNS_UPDATER_DENY_RANGES=${deny_ranges_str}
+        DNS_UPDATER_ALLOW_RANGES=${allow_ranges_str}
+        DNS_UPDATER_DENY_DOMAINS=${deny_domains_str}
+        DNS_UPDATER_ALLOW_DOMAINS=${allow_domains_str}
+        | ENV
+
+      file { "${config_dir}/env":
+        ensure  => file,
+        owner   => 'root',
+        group   => 'root',
+        mode    => '0644',
+        content => $env_content,
+        require => [File[$config_dir], Package['dns-updater']],
+      }
+
+      # Restart only on config/key change; records-file changes are picked up by
+      # the daemon's own inotify watch, so no service churn on record edits.
+      service { 'dns-updater':
+        ensure    => running,
+        enable    => true,
+        subscribe => [File["${config_dir}/env"], File[$key_file]],
+        require   => [Package['dns-updater'], Concat[$records_file], File[$key_file]],
+      }
+
+      # Retire the previous shell-based mechanism.
+      file { '/usr/local/bin/dns-update':
+        ensure => absent,
+      }
+
+      systemd::unit_file { 'dns-update.service':
+        ensure => absent,
+        active => false,
+        enable => false,
+      }
+
+      systemd::unit_file { 'dns-update.path':
+        ensure => absent,
+        active => false,
+        enable => false,
+      }
+    }
+  }
+}
@@ -18,7 +18,12 @@ class profiles::etcd::node (
    }

    # if it is, find hosts, sort them so they dont cause changes every run
-    $servers_array = sort(query_nodes("enc_role='${members_role}' and region='${facts['region']}'", 'networking.fqdn'))
+    $servers_array = sort(puppetdb_query(
+      "facts[certname] {
+        name = 'enc_role' and value = '${members_role}' and
+        certname in facts[certname] { name = 'region' and value = '${facts['region']}' }
+      }"
+    ).map |$fact| { $fact['certname'] })

  # else use provided array from params
  }else{
@@ -31,7 +36,7 @@ class profiles::etcd::node (
    $initial_cluster = $servers_array.map |$fqdn| {

      # lookup the ip address for the current fqdn
-      $ip = query_nodes("networking.fqdn='${fqdn}'", 'networking.ip')[0]
+      $ip = puppetdb_query("facts[certname,value] { name = 'networking' and certname = '${fqdn}' }").map |$fact| { $fact['value']['ip'] }[0]

      # construct the string for this server
      "${fqdn}=https://${ip}:${peer_port}"
@@ -0,0 +1,14 @@
+# base fedora settings
+#
+# Fedora hosts share the RedHat-family base profiles (yum/dnf, crypto-policies,
+# firewalld); this class carries the handful of Fedora-specific bits. Kept
+# deliberately small so it can grow into workstation/laptop (e.g. sway) support
+# later without disturbing servers.
+class profiles::fedora::base {
+  # Fedora manages interfaces through NetworkManager. Ensure it is enabled and
+  # running (some minimal and cloud images ship it disabled).
+  service { 'NetworkManager':
+    ensure => running,
+    enable => true,
+  }
+}
@@ -21,7 +21,10 @@ class profiles::haproxy::dns (
    $parts = split($cname, '\.')
    $domain = join($parts[1, $parts.length], '.')
    profiles::dns::record { "${::facts['networking']['fqdn']}_${cname}_CNAME":
-      value  => "${location_environment}-halb",
+      # CNAME target must be a fully-qualified name (trailing dot): dns-updater
+      # parses the value with no $ORIGIN, so a bare label becomes root-absolute
+      # ("au-syd1-prod-halb.") and dead-ends in NXDOMAIN.
+      value  => "${location_environment}-halb.${facts['networking']['domain']}.",
      type   => 'CNAME',
      record => "${cname}.",
      zone   => $domain,
@@ -30,13 +33,14 @@ class profiles::haproxy::dns (
  }

  # if it is, find hosts, sort them so they dont cause changes every run
-  $servers_array = sort(query_nodes(
-    "enc_role='${facts['enc_role']}' and
-      country='${facts['country']}' and
-      region='${facts['region']}' and
-      environment='${facts['environment']}'",
-    'networking.fqdn'
-  ))
+  $servers_array = sort(puppetdb_query(
+    "facts[certname] {
+      name = 'enc_role' and value = '${facts['enc_role']}' and
+      certname in facts[certname] { name = 'country' and value = '${facts['country']}' } and
+      certname in facts[certname] { name = 'region' and value = '${facts['region']}' } and
+      certname in facts[certname] { name = 'environment' and value = '${facts['environment']}' }
+    }"
+  ).map |$fact| { $fact['certname'] })

  # give enough time for a few hosts to be provisioned
  if length($servers_array) >= 3 {
@@ -57,7 +61,10 @@ class profiles::haproxy::dns (
        $parts = split($cname, '\.')
        $domain = join($parts[1, $parts.length], '.')
        profiles::dns::record { "${::facts['networking']['fqdn']}_${cname}_CNAME":
-          value  => "${location_environment}-halb-vrrp",
+          # Fully-qualified target (trailing dot); see the note on the non-vrrp
+          # cnames above. The matching A record is published in main.unkin.net
+          # and unkin.net just above.
+          value  => "${location_environment}-halb-vrrp.${facts['networking']['domain']}.",
          type   => 'CNAME',
          record => "${cname}.",
          zone   => $domain,
@@ -1,6 +1,7 @@
 # profiles::metrics::grafana
 class profiles::metrics::grafana (
  String    $ldap_bind_pass,
+  String    $version        = 'installed',
  Stdlib::Port $http_port   = 8080,
  String    $app_mode       = 'production',
  Boolean   $allow_sign_up  = false,
@@ -107,6 +108,7 @@ class profiles::metrics::grafana (

  # deploy grafana
  class { 'grafana':
+    version  => $version,
    cfg      => $cfg,
    ldap_cfg => $ldap_cfg,
    plugins  => $plugins,
@@ -98,8 +98,15 @@ class profiles::minio::server (
    }

    # if it is, find hosts, sort them so they dont cause changes every run
-    #$servers_array = sort(query_nodes("enc_role='${minio_members_role}'", 'networking.fqdn'))
-    $servers_array = sort(query_nodes("enc_role='${minio_members_role}' and minio_region='${minio_region}'", 'networking.fqdn'))
+    #$servers_array = sort(puppetdb_query(
+    #  "facts[certname] { name = 'enc_role' and value = '${minio_members_role}' }"
+    #).map |$fact| { $fact['certname'] })
+    $servers_array = sort(puppetdb_query(
+      "facts[certname] {
+        name = 'enc_role' and value = '${minio_members_role}' and
+        certname in facts[certname] { name = 'minio_region' and value = '${minio_region}' }
+      }"
+    ).map |$fact| { $fact['certname'] })

  # else use provided array from params
  }else{
@@ -26,9 +26,21 @@ class profiles::ntp::client (
    $ntpserver_array = $ntp_role ? {
      undef   => $peers,
      default => $use_ntp ? {
-        'all'     => query_nodes("enc_role='${ntp_role}'", 'networking.fqdn'),
-        'region'  => query_nodes("enc_role='${ntp_role}' and region=${facts['region']}", 'networking.fqdn'),
-        'country' => query_nodes("enc_role='${ntp_role}' and country=${facts['country']}", 'networking.fqdn'),
+        'all'     => puppetdb_query(
+          "facts[certname] { name = 'enc_role' and value = '${ntp_role}' }"
+        ).map |$fact| { $fact['certname'] },
+        'region'  => puppetdb_query(
+          "facts[certname] {
+            name = 'enc_role' and value = '${ntp_role}' and
+            certname in facts[certname] { name = 'region' and value = '${facts['region']}' }
+          }"
+        ).map |$fact| { $fact['certname'] },
+        'country' => puppetdb_query(
+          "facts[certname] {
+            name = 'enc_role' and value = '${ntp_role}' and
+            certname in facts[certname] { name = 'country' and value = '${facts['country']}' }
+          }"
+        ).map |$fact| { $fact['certname'] },
      }
    }

@@ -24,10 +24,13 @@ class profiles::proxmox::clusterinit {
    }
  }

-  $servers_array = sort(query_nodes(
-    "enc_role='${membersrole}' and country='${facts['country']}' and region='${facts['region']}'",
-    'networking.fqdn'
-  ))
+  $servers_array = sort(puppetdb_query(
+    "facts[certname] {
+      name = 'enc_role' and value = '${membersrole}' and
+      certname in facts[certname] { name = 'country' and value = '${facts['country']}' } and
+      certname in facts[certname] { name = 'region' and value = '${facts['region']}' }
+    }"
+  ).map |$fact| { $fact['certname'] })

  if ! $profiles::proxmox::params::pve_clusterinit_master {
    if !empty($servers_array) {
@@ -11,13 +11,14 @@ class profiles::proxmox::clusterjoin {
  $root_password      = $profiles::proxmox::params::root_password

  # query puppetdb for list of cluster members
-  $members_array = sort(query_nodes(
-    "enc_role='${membersrole}' and \
-    country='${facts['country']}' and \
-    region='${facts['region']}' and \
-    pve_cluster.cluster_name='${clustername}'",
-    'networking.fqdn'
-  ))
+  $members_array = sort(puppetdb_query(
+    "facts[certname] {
+      name = 'enc_role' and value = '${membersrole}' and
+      certname in facts[certname] { name = 'country' and value = '${facts['country']}' } and
+      certname in facts[certname] { name = 'region' and value = '${facts['region']}' } and
+      certname in facts[certname] { name = 'pve_cluster' and value.cluster_name = '${clustername}' }
+    }"
+  ).map |$fact| { $fact['certname'] })

  # check if the pve kernerl is running
  if $facts['kernelrelease'] == $profiles::proxmox::params::pve_kernel_release {
@@ -16,11 +16,18 @@ class profiles::puppet::agent (
    $use_service = 'puppet'
  }

+  # OpenVox publishes RedHat-family builds per distribution: enterprise-linux
+  # under el/<major>, Fedora under fedora/<major>.
+  $openvox_dist = $facts['os']['name'] ? {
+    'Fedora' => 'fedora',
+    default  => 'el',
+  }
+
  # manage the yumrepo for the given package
  if $openvox_enable and $facts['os']['family'] == 'RedHat' {
    yumrepo { 'openvox':
      ensure  => 'present',
-      baseurl => "https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/openvox/openvox7/el/${facts['os']['release']['major']}/${facts['os']['architecture']}/",
+      baseurl => "https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/openvox/openvox7/${openvox_dist}/${facts['os']['release']['major']}/${facts['os']['architecture']}/",
      descr   => 'openvox repository',
      gpgkey  => 'https://artifactapi.k8s.syd1.au.unkin.net/api/v1/remote/openvox/GPG-KEY-openvox.pub',
      notify  => Exec['dnf_makecache'],
@@ -4,19 +4,64 @@
 #
 # site/profile/manifests/puppet/client.pp
 class profiles::puppet::client (
-  Array  $dns_alt_names       = [$trusted['certname']],
-  String $server              = 'puppetmaster',
-  String $ca_server           = 'puppetca',
-  String $environment         = 'develop',
-  Integer $runinterval        = 1800,
-  Integer $runtimeout         = 3600,
-  Boolean $show_diff          = true,
-  Boolean $usecacheonfailure  = false,
-  Integer $facts_soft_limit   = 4096,
-  Boolean $splay              = true,
-  Integer $splaylimit         = 600,
+  Array  $dns_alt_names                  = [$trusted['certname']],
+  String $server                         = 'puppetmaster',
+  String $ca_server                      = 'puppetca',
+  Optional[String] $report_server        = undef,
+  String $environment                    = 'develop',
+  Integer $runinterval                   = 1800,
+  Integer $runtimeout                    = 3600,
+  Boolean $show_diff                     = true,
+  Boolean $usecacheonfailure             = false,
+  Integer $facts_soft_limit              = 4096,
+  Boolean $splay                         = true,
+  Integer $splaylimit                    = 600,
+  Optional[Stdlib::Absolutepath] $ssldir = undef,
 ) {

+  # Pull in the k8s-migration switch. When
+  # profiles::puppet::migrate::enabled is true at ANY hiera layer
+  # (node/role/common) the effective endpoints below flip to the k8s
+  # servers and a fresh ssldir -- no other hiera keys required. This class
+  # never includes client.pp, so there is no include cycle.
+  include profiles::puppet::migrate
+
+  # Effective values fed to the template. Precedence differs by param type:
+  #
+  #   ssldir / report_server (Optional, default undef): an explicit client
+  #     param wins; else the migrate-derived value when enabled; else the
+  #     legacy default (undef -> template omits ssldir; report_server -> server).
+  #
+  #   server / ca_server (String, no undef sentinel): the migrate value wins
+  #     when migrate::enabled, otherwise the client param (which is the legacy
+  #     default unless an operator set it explicitly in hiera). To pin a
+  #     bespoke server while enabled, leave enabled => false for that target
+  #     and set profiles::puppet::client::server directly.
+  $migrate_on = $profiles::puppet::migrate::enabled
+
+  $effective_server = $migrate_on ? {
+    true    => $profiles::puppet::migrate::new_server,
+    default => $server,
+  }
+  $effective_ca_server = $migrate_on ? {
+    true    => $profiles::puppet::migrate::new_ca_server,
+    default => $ca_server,
+  }
+  # report_server: explicit param wins; else follow the effective server.
+  $effective_report_server = $report_server ? {
+    undef   => $effective_server,
+    default => $report_server,
+  }
+  # ssldir: explicit param wins; else the fresh k8s ssldir when migrating;
+  # else undef (template omits the key, i.e. puppet's built-in default).
+  $effective_ssldir = $ssldir ? {
+    undef   => $migrate_on ? {
+      true    => $profiles::puppet::migrate::new_ssldir,
+      default => undef,
+    },
+    default => $ssldir,
+  }
+
  # dont manage puppet.conf if this is a puppetmaster
  if $facts['enc_role'] != 'roles::infra::puppet::master' {

@@ -0,0 +1,87 @@
+# Class: profiles::puppet::migrate
+#
+# Repoints a VM puppet agent from the legacy VM puppetmasters onto the new
+# puppet-on-kubernetes servers.
+#
+# The agent's existing ssldir holds a client certificate signed by the OLD CA;
+# the new k8s CA will neither trust that cert nor recognise the agent. Rather
+# than revoke/clean the old creds (which would burn the rollback path), this
+# class switches the agent to a FRESH ssldir. On the next run the agent finds
+# an empty ssldir, generates a new key + CSR, submits it to the new CA
+# (autosign = true on the k8s side signs it immediately), and then pulls its
+# catalog from the new compilers. The OLD ssldir is left untouched on disk, so
+# reverting the hiera below restores the node to the legacy masters with no
+# re-enrolment.
+#
+# The catalog that performs the switch is compiled by the OLD server (the node
+# still points at it when it applies this change). profiles::puppet::client
+# rewrites puppet.conf with the new server / ca_server / report_server / ssldir
+# and notifies Service['puppet'], so the very next agent run lands on k8s.
+#
+# NOTE: this is deliberately hiera-driven per node/role so waves can be rolled
+# out and rolled back one target at a time. The class only manages the fresh
+# ssldir; the actual puppet.conf changes ride on profiles::puppet::client's
+# existing template. client.pp `include`s this class and, when enabled, folds
+# new_server / new_ca_server / new_ssldir into the effective server /
+# ca_server / report_server / ssldir it renders. Do NOT add a second File
+# resource for puppet.conf here -- client.pp already owns it.
+#
+# == Wiring (single boolean)
+#
+# This class ships wired into profiles::base with enabled => false, so every
+# node evaluates it but nothing changes until the boolean is flipped. Because
+# client.pp includes it and reads its params, setting the ONE key below at any
+# hiera layer repoints a target -- no client.pp param overrides needed.
+#
+# The endpoints (new_server / new_ca_server / new_ssldir) are param defaults
+# on this class; override them in hiera too if the k8s names ever change.
+#
+# --- Canary: one node -> hieradata/nodes/<certname>.yaml -------------------
+#   profiles::puppet::migrate::enabled: true
+# ---------------------------------------------------------------------------
+#
+# --- Wave: a whole role -> hieradata/roles/<t1>/<t2>[/<t3>].yaml -----------
+#   profiles::puppet::migrate::enabled: true
+# ---------------------------------------------------------------------------
+#
+# --- Estate flip -> hieradata/common.yaml (retires the legacy masters) -----
+#   profiles::puppet::migrate::enabled: true
+# ---------------------------------------------------------------------------
+#
+# Manual override: profiles::puppet::client::ssldir and ::report_server are
+# Optional and, when set explicitly in hiera, win over the migrate-derived
+# value. server / ca_server have no undef sentinel, so while enabled => true
+# the migrate endpoints win; to point a target at a bespoke server leave
+# enabled => false for it and set profiles::puppet::client::server directly.
+#
+# == Rollback (per target)
+#
+# Remove the profiles::puppet::migrate::enabled key (or set it back to false)
+# at the node/role/common layer. The next run rewrites puppet.conf back to the
+# legacy server + legacy ssldir, which still contains the original CA-signed
+# cert, and the node is back on the VM masters.
+#
+# site/profiles/manifests/puppet/migrate.pp
+class profiles::puppet::migrate (
+  Boolean $enabled                          = false,
+  String $new_server                        = 'puppet.k8s.syd1.au.unkin.net',
+  String $new_ca_server                     = 'puppetca.k8s.syd1.au.unkin.net',
+  Stdlib::Absolutepath $new_ssldir          = '/etc/puppetlabs/puppet/ssl-k8s',
+) {
+
+  # never touch the puppet masters themselves (same guard as client.pp)
+  if $enabled and $facts['enc_role'] != 'roles::infra::puppet::master' {
+
+    # Ensure the fresh ssldir exists with agent-correct ownership/perms. The
+    # puppet agent will create it on demand, but pre-creating it keeps the very
+    # first post-switch run from racing directory creation and makes the
+    # migration state auditable. Contents (keys, certs, crl) are populated by
+    # the agent against the new CA -- we only own the directory itself.
+    file { $new_ssldir:
+      ensure => 'directory',
+      owner  => 'root',
+      group  => 'root',
+      mode   => '0771',
+    }
+  }
+}
@@ -134,4 +134,64 @@ class profiles::puppet::server (
    enable          => true,
    require         => File['/usr/local/bin/puppet_generate_types.sh'],
  }
+
+  # Guard against an OpenJDK (or other JVM library) package upgrade landing
+  # underneath a long-running puppetserver. When the java package is replaced
+  # out-of-band (e.g. a yum repo/version bump), the old versioned JAVA_HOME is
+  # removed while the JVM keeps running against the now-deleted files. The JVM
+  # still execs jspawnhelper from its original (deleted) JAVA_HOME on every
+  # subprocess spawn, so ProcessBuilder fails with "error=2, No such file or
+  # directory" -- which breaks the exec ENC (/opt/cobbler-enc/cobbler-enc) and
+  # 500s every catalog compile fleet-wide. A restart re-binds to the new JVM.
+  # This timer detects the running JVM executing from deleted binaries and
+  # restarts the service to recover automatically.
+  file { '/usr/local/bin/puppetserver_jvm_guard.sh':
+    ensure  => file,
+    mode    => '0755',
+    content => @(EOF),
+      #!/bin/bash
+      # Restart puppetserver if its running JVM is executing deleted binaries
+      # (e.g. an OpenJDK package upgrade replaced the versioned JAVA_HOME),
+      # which breaks subprocess spawning and thus ENC / catalog compilation.
+      for pid in $(pgrep -f puppet-server-release.jar); do
+        exe=$(readlink "/proc/${pid}/exe" 2>/dev/null)
+        case "${exe}" in
+          *'(deleted)'*)
+            logger -t puppetserver-jvm-guard "puppetserver JVM (pid ${pid}) running on deleted binaries; restarting"
+            systemctl restart puppetserver
+            exit 0
+            ;;
+        esac
+      done
+      exit 0
+      | EOF
+  }
+
+  $_guard_timer = @(EOT)
+  [Unit]
+  Description=puppetserver JVM guard timer
+  [Timer]
+  OnCalendar=*:0/5
+  RandomizedDelaySec=30s
+  [Install]
+  WantedBy=timers.target
+  EOT
+
+  $_guard_service = @(EOT)
+  [Unit]
+  Description=puppetserver JVM guard service
+  [Service]
+  Type=oneshot
+  ExecStart=/usr/local/bin/puppetserver_jvm_guard.sh
+  User=root
+  Group=root
+  EOT
+
+  systemd::timer { 'puppetserver-jvm-guard.timer':
+    timer_content   => $_guard_timer,
+    service_content => $_guard_service,
+    active          => true,
+    enable          => true,
+    require         => File['/usr/local/bin/puppetserver_jvm_guard.sh'],
+  }
 }
@@ -48,7 +48,12 @@ class profiles::sql::galera_member (
    }

    # if it is, find hosts, sort them so they dont cause changes every run
-    $servers_array = sort(query_nodes("enc_role='${galera_members_role}' and region='${facts['region']}'", 'networking.fqdn'))
+    $servers_array = sort(puppetdb_query(
+      "facts[certname] {
+        name = 'enc_role' and value = '${galera_members_role}' and
+        certname in facts[certname] { name = 'region' and value = '${facts['region']}' }
+      }"
+    ).map |$fact| { $fact['certname'] })

  # else use provided array from params
  }else{
@@ -18,7 +18,12 @@ class profiles::sql::postgresdb (
    }

    # if it is, find hosts, sort them so they dont cause changes every run
-    $servers_array = sort(query_nodes("enc_role='${members_role}' and region='${facts['region']}'", 'networking.fqdn'))
+    $servers_array = sort(puppetdb_query(
+      "facts[certname] {
+        name = 'enc_role' and value = '${members_role}' and
+        certname in facts[certname] { name = 'region' and value = '${facts['region']}' }
+      }"
+    ).map |$fact| { $fact['certname'] })

  # else use provided array from params
  }else{
@@ -29,7 +29,12 @@ class profiles::vault::server (
  if $members_lookup and $members_role != undef {

    # if it is, find hosts, sort them so they dont cause changes every run
-    $servers_array = sort(query_nodes("enc_role='${members_role}' and region='${::facts['region']}'", 'networking.fqdn'))
+    $servers_array = sort(puppetdb_query(
+      "facts[certname] {
+        name = 'enc_role' and value = '${members_role}' and
+        certname in facts[certname] { name = 'region' and value = '${::facts['region']}' }
+      }"
+    ).map |$fact| { $fact['certname'] })

  # else use provided array from params
  }else{
@@ -117,6 +122,15 @@ class profiles::vault::server (
      ]
    }

+    # go-plugin creates each secrets plugin's control socket as /tmp/pluginNNN;
+    # systemd-tmpfiles-clean reaps files aged past the /tmp policy, severing the
+    # socket of a long-lived plugin and orphaning the still-running process
+    # (every request then fails: rpc Unavailable, dial unix /tmp/pluginNNN: no
+    # such file). Exclude the sockets from cleaning for all plugins.
+    systemd::tmpfile { 'openbao-plugin-sockets.conf':
+      content => "x /tmp/plugin*\n",
+    }
+
    # ensure the vault audit log exists
    file { $audit_log:
      ensure => 'file',
@@ -0,0 +1,5 @@
+<%- | String $name, String $algorithm, String $secret | -%>
+key "<%= $name %>" {
+    algorithm <%= $algorithm %>;
+    secret "<%= $secret %>";
+};
@@ -1,12 +1,15 @@
 [main]
 dns_alt_names = <%= @dns_alt_names_string %>
+<% unless @effective_ssldir.nil? -%>
+ssldir = <%= @effective_ssldir %>
+<% end -%>

 [agent]
-server = <%= @server %>
-ca_server = <%= @ca_server %>
+server = <%= @effective_server %>
+ca_server = <%= @effective_ca_server %>
 environment = <%= @environment %>
 report = true
-report_server = <%= @server %>
+report_server = <%= @effective_report_server.nil? ? @effective_server : @effective_report_server %>
 runinterval = <%= @runinterval %>
 runtimeout = <%= @runtimeout %>
 show_diff = <%= @show_diff %>