feat: stabilize tailscale observability exposure with declarative proxy class

.gitea/workflows/deploy.yml

@@ -243,6 +243,9 @@ jobs:
           ansible -i inventory.ini 'control_plane[0]' -m command -a "kubectl get storageclass"
           ansible -i inventory.ini 'control_plane[0]' -m command -a "kubectl -n observability get pods -o wide"
           ansible -i inventory.ini 'control_plane[0]' -m command -a "kubectl -n observability get pvc"
+          ansible -i inventory.ini 'control_plane[0]' -m command -a "kubectl -n tailscale-system get pods -o wide"
+          ansible -i inventory.ini 'control_plane[0]' -m command -a "kubectl -n observability get svc kube-prometheus-stack-grafana kube-prometheus-stack-prometheus"
+          ansible -i inventory.ini 'control_plane[0]' -m command -a "kubectl -n observability describe svc kube-prometheus-stack-grafana"
         env:
           ANSIBLE_HOST_KEY_CHECKING: "False"
 

README.md

@@ -166,6 +166,8 @@ Set these in your Gitea repository settings (**Settings** → **Secrets** → **
 | `S3_BUCKET` | S3 bucket name (e.g., `k8s-terraform-state`) |
 | `TAILSCALE_AUTH_KEY` | Tailscale auth key for node bootstrap |
 | `TAILSCALE_TAILNET` | Tailnet domain (e.g., `yourtailnet.ts.net`) |
+| `TAILSCALE_OAUTH_CLIENT_ID` | Tailscale OAuth client ID for Kubernetes Operator |
+| `TAILSCALE_OAUTH_CLIENT_SECRET` | Tailscale OAuth client secret for Kubernetes Operator |
 | `GRAFANA_ADMIN_PASSWORD` | Optional admin password for Grafana (auto-generated if unset) |
 | `RUNNER_ALLOWED_CIDRS` | Optional CIDR list for CI runner access if you choose to pass it via tfvars/secrets |
 | `SSH_PUBLIC_KEY` | SSH public key content |
@@ -179,10 +181,17 @@ The Ansible playbook deploys a lightweight observability stack in the `observabi
 - `loki`
 - `promtail`
 
-Services are kept internal for tailnet-first access.
+Services are kept internal by default, with optional declarative Tailscale exposure when the Tailscale Kubernetes Operator is healthy.
 
 ### Access Grafana and Prometheus
 
+Preferred (when Tailscale Operator is healthy):
+
+- Grafana: `http://grafana` (or `http://grafana.<your-tailnet>`)
+- Prometheus: `http://prometheus` (or `http://prometheus.<your-tailnet>`)
+
+Fallback (port-forward from a tailnet-connected machine):
+
 Run from a tailnet-connected machine:
 
 ```bash
@@ -200,6 +209,25 @@ Then open:
 Grafana user: `admin`
 Grafana password: value of `GRAFANA_ADMIN_PASSWORD` secret (or the generated value shown by Ansible output)
 
+### Verify Tailscale exposure
+
+```bash
+export KUBECONFIG=$(pwd)/outputs/kubeconfig
+
+kubectl -n tailscale-system get pods
+kubectl -n observability get svc kube-prometheus-stack-grafana kube-prometheus-stack-prometheus
+kubectl -n observability describe svc kube-prometheus-stack-grafana | grep TailscaleProxyReady
+kubectl -n observability describe svc kube-prometheus-stack-prometheus | grep TailscaleProxyReady
+```
+
+If `TailscaleProxyReady=False`, check:
+
+```bash
+kubectl -n tailscale-system logs deployment/operator --tail=100
+```
+
+Common cause: OAuth client missing tag/scopes permissions.
+
 ## File Structure
 
 ```
@@ -223,6 +251,7 @@ Grafana password: value of `GRAFANA_ADMIN_PASSWORD` secret (or the generated val
 │   │   ├── k3s-agent/
 │   │   ├── ccm/
 │   │   ├── csi/
+│   │   ├── tailscale-operator/
 │   │   └── observability/
 │   └── ansible.cfg
 ├── .gitea/

ansible/roles/observability/defaults/main.yml

@@ -20,3 +20,8 @@ loki_enabled: true
 tailscale_oauth_client_id: ""
 tailscale_oauth_client_secret: ""
 tailscale_tailnet: ""
+
+observability_tailscale_expose: true
+grafana_tailscale_hostname: "grafana"
+prometheus_tailscale_hostname: "prometheus"
+tailscale_proxyclass_name: "infra-stable"

ansible/roles/observability/tasks/main.yml

@@ -156,64 +156,66 @@
   changed_when: true
   when: loki_enabled
 
-- name: Configure Grafana for Tailscale access
+- name: Check Tailscale service readiness for Grafana
-  block:
+  command: kubectl -n {{ observability_namespace }} get svc kube-prometheus-stack-grafana -o jsonpath='{.status.conditions[?(@.type=="TailscaleProxyReady")].status}'
-    - name: Patch Grafana service for Tailscale
+  register: grafana_tailscale_ready
-      command: >-
+  changed_when: false
-        kubectl -n {{ observability_namespace }} patch svc kube-prometheus-stack-grafana
+  failed_when: false
-        -p '{"metadata":{"annotations":{"tailscale.com/hostname":"grafana"}},"spec":{"type":"LoadBalancer","loadBalancerClass":"tailscale"}}'
-      register: grafana_patch
-      changed_when: true
-
-    - name: Patch Prometheus service for Tailscale
-      command: >-
-        kubectl -n {{ observability_namespace }} patch svc kube-prometheus-stack-prometheus
-        -p '{"metadata":{"annotations":{"tailscale.com/hostname":"prometheus"}},"spec":{"type":"LoadBalancer","loadBalancerClass":"tailscale"}}'
-      register: prometheus_patch
-      changed_when: true
-
-    - name: Check Tailscale endpoint (IP/hostname) for Grafana
-      shell: >-
-        kubectl -n {{ observability_namespace }} get svc kube-prometheus-stack-grafana
-        -o go-template='{{"{{"}}range .status.loadBalancer.ingress{{"}}"}}{{"{{"}}if .ip{{"}}"}}{{"{{"}}.ip{{"}}"}}{{"{{"}}else{{"}}"}}{{"{{"}}.hostname{{"}}"}}{{"{{"}}end{{"}}"}}{{"{{"}}end{{"}}"}}'
-      register: grafana_lb_ip
-      changed_when: false
-      failed_when: false
-
-    - name: Check Tailscale endpoint (IP/hostname) for Prometheus
-      shell: >-
-        kubectl -n {{ observability_namespace }} get svc kube-prometheus-stack-prometheus
-        -o go-template='{{"{{"}}range .status.loadBalancer.ingress{{"}}"}}{{"{{"}}if .ip{{"}}"}}{{"{{"}}.ip{{"}}"}}{{"{{"}}else{{"}}"}}{{"{{"}}.hostname{{"}}"}}{{"{{"}}end{{"}}"}}{{"{{"}}end{{"}}"}}'
-      register: prometheus_lb_ip
-      changed_when: false
-      failed_when: false
-
-    - name: Warn if Tailscale endpoint assignment is still pending
-      debug:
-        msg: |
-          Tailscale service endpoint assignment is still pending.
-          Grafana endpoint: {{ grafana_lb_ip.stdout | default('') }}
-          Prometheus endpoint: {{ prometheus_lb_ip.stdout | default('') }}
-          Deployment continues; services may become reachable shortly.
-      when: (grafana_lb_ip.stdout | default('') | length == 0) or (prometheus_lb_ip.stdout | default('') | length == 0)
-
-    - name: Show Tailscale access details
-      debug:
-        msg: |
-          Observability stack deployed with Tailscale access!
-
-          Grafana: http://grafana{% if grafana_lb_ip.stdout | default('') | length > 0 %} (or http://{{ grafana_lb_ip.stdout }}){% endif %}
-          Prometheus: http://prometheus{% if prometheus_lb_ip.stdout | default('') | length > 0 %} (or http://{{ prometheus_lb_ip.stdout }}){% endif %}
-
-          Login: admin / {{ grafana_password_effective }}
-
-          Access via:
-          - MagicDNS: http://grafana or http://prometheus (if enabled)
-          - Direct endpoint: {% if grafana_lb_ip.stdout | default('') | length > 0 %}http://{{ grafana_lb_ip.stdout }}{% else %}(pending){% endif %} / {% if prometheus_lb_ip.stdout | default('') | length > 0 %}http://{{ prometheus_lb_ip.stdout }}{% else %}(pending){% endif %}
-          - Tailnet FQDN: http://grafana.{{ tailscale_tailnet | default('tailnet.ts.net') }}
-
-          Note: Ensure Tailscale Kubernetes Operator is installed first
   when:
+    - observability_tailscale_expose | bool
+    - tailscale_operator_ready | default(false) | bool
+
+- name: Check Tailscale service readiness for Prometheus
+  command: kubectl -n {{ observability_namespace }} get svc kube-prometheus-stack-prometheus -o jsonpath='{.status.conditions[?(@.type=="TailscaleProxyReady")].status}'
+  register: prometheus_tailscale_ready
+  changed_when: false
+  failed_when: false
+  when:
+    - observability_tailscale_expose | bool
+    - tailscale_operator_ready | default(false) | bool
+
+- name: Check Tailscale endpoint (IP/hostname) for Grafana
+  shell: >-
+    kubectl -n {{ observability_namespace }} get svc kube-prometheus-stack-grafana
+    -o go-template='{{"{{"}}range .status.loadBalancer.ingress{{"}}"}}{{"{{"}}if .ip{{"}}"}}{{"{{"}}.ip{{"}}"}}{{"{{"}}else{{"}}"}}{{"{{"}}.hostname{{"}}"}}{{"{{"}}end{{"}}"}}{{"{{"}}end{{"}}"}}'
+  register: grafana_lb_ip
+  changed_when: false
+  failed_when: false
+  when:
+    - observability_tailscale_expose | bool
+    - tailscale_operator_ready | default(false) | bool
+
+- name: Check Tailscale endpoint (IP/hostname) for Prometheus
+  shell: >-
+    kubectl -n {{ observability_namespace }} get svc kube-prometheus-stack-prometheus
+    -o go-template='{{"{{"}}range .status.loadBalancer.ingress{{"}}"}}{{"{{"}}if .ip{{"}}"}}{{"{{"}}.ip{{"}}"}}{{"{{"}}else{{"}}"}}{{"{{"}}.hostname{{"}}"}}{{"{{"}}end{{"}}"}}{{"{{"}}end{{"}}"}}'
+  register: prometheus_lb_ip
+  changed_when: false
+  failed_when: false
+  when:
+    - observability_tailscale_expose | bool
+    - tailscale_operator_ready | default(false) | bool
+
+- name: Show Tailscale access details
+  debug:
+    msg: |
+      Observability stack deployed with Tailscale access!
+
+      Grafana: http://{{ grafana_tailscale_hostname }}{% if grafana_lb_ip.stdout | default('') | length > 0 %} (or http://{{ grafana_lb_ip.stdout }}){% endif %}
+      Prometheus: http://{{ prometheus_tailscale_hostname }}{% if prometheus_lb_ip.stdout | default('') | length > 0 %} (or http://{{ prometheus_lb_ip.stdout }}){% endif %}
+
+      Login: admin / {{ grafana_password_effective }}
+
+      Tailscale readiness:
+      - Grafana proxy ready: {{ grafana_tailscale_ready.stdout | default('pending') }}
+      - Prometheus proxy ready: {{ prometheus_tailscale_ready.stdout | default('pending') }}
+
+      Access via:
+      - MagicDNS: http://{{ grafana_tailscale_hostname }} and http://{{ prometheus_tailscale_hostname }}
+      - Tailnet FQDN: http://{{ grafana_tailscale_hostname }}.{{ tailscale_tailnet | default('tailnet.ts.net') }}
+      - Direct endpoint: {% if grafana_lb_ip.stdout | default('') | length > 0 %}http://{{ grafana_lb_ip.stdout }}{% else %}(pending){% endif %} / {% if prometheus_lb_ip.stdout | default('') | length > 0 %}http://{{ prometheus_lb_ip.stdout }}{% else %}(pending){% endif %}
+  when:
+    - observability_tailscale_expose | bool
     - tailscale_operator_ready | default(false) | bool
 
 - name: Show observability access details (fallback)
@@ -230,4 +232,4 @@
       Loki: Disabled
       {% endif %}
   when:
-    - not (tailscale_operator_ready | default(false) | bool)
+    - not (observability_tailscale_expose | bool and (tailscale_operator_ready | default(false) | bool))

ansible/roles/observability/templates/kube-prometheus-stack-values.yaml.j2

@@ -6,8 +6,26 @@ grafana:
     storageClassName: {{ grafana_storage_class }}
     size: {{ grafana_storage_size }}
   service:
+{% if observability_tailscale_expose and (tailscale_operator_ready | default(false)) %}
+    type: LoadBalancer
+    loadBalancerClass: tailscale
+    annotations:
+      tailscale.com/hostname: {{ grafana_tailscale_hostname }}
+      tailscale.com/proxy-class: {{ tailscale_proxyclass_name }}
+{% else %}
     type: ClusterIP
+{% endif %}
 prometheus:
+  service:
+{% if observability_tailscale_expose and (tailscale_operator_ready | default(false)) %}
+    type: LoadBalancer
+    loadBalancerClass: tailscale
+    annotations:
+      tailscale.com/hostname: {{ prometheus_tailscale_hostname }}
+      tailscale.com/proxy-class: {{ tailscale_proxyclass_name }}
+{% else %}
+    type: ClusterIP
+{% endif %}
   prometheusSpec:
     retention: 7d
     storageSpec:

ansible/roles/tailscale-operator/defaults/main.yml

@@ -8,6 +8,8 @@ tailscale_oauth_client_secret: ""
 tailscale_operator_default_tags:
   - "tag:k8s"
 
+tailscale_proxyclass_name: "infra-stable"
+
 tailscale_operator_required: false
 
 tailscale_operator_node_selector:

ansible/roles/tailscale-operator/tasks/main.yml

@@ -157,3 +157,15 @@
       {{ tailscale_rollout.stdout | default('') }}
       {{ tailscale_deploy.stdout | default('') }}
   when: tailscale_rollout.rc != 0
+
+- name: Write Tailscale default ProxyClass manifest
+  template:
+    src: proxyclass.yaml.j2
+    dest: /tmp/tailscale-proxyclass.yaml
+    mode: "0644"
+  when: tailscale_operator_ready | default(false) | bool
+
+- name: Apply Tailscale default ProxyClass
+  command: kubectl apply -f /tmp/tailscale-proxyclass.yaml
+  changed_when: true
+  when: tailscale_operator_ready | default(false) | bool

ansible/roles/tailscale-operator/templates/operator-values.yaml.j2

@@ -21,3 +21,4 @@ installCRDs: true
 
 proxyConfig:
   defaultTags: "{{ tailscale_operator_default_tags | join(',') }}"
+  defaultProxyClass: "{{ tailscale_proxyclass_name }}"

ansible/roles/tailscale-operator/templates/proxyclass.yaml.j2

@@ -0,0 +1,17 @@
+apiVersion: tailscale.com/v1alpha1
+kind: ProxyClass
+metadata:
+  name: {{ tailscale_proxyclass_name }}
+spec:
+  statefulSet:
+    pod:
+      nodeSelector:
+{% for key, value in tailscale_operator_node_selector.items() %}
+        {{ key }}: "{{ value }}"
+{% endfor %}
+      tolerations:
+{% for tol in tailscale_operator_tolerations %}
+      - key: "{{ tol.key }}"
+        operator: "{{ tol.operator }}"
+        effect: "{{ tol.effect }}"
+{% endfor %}