# NetBird Configuration as Code Guide ## Overview Manage your NetBird logical configuration (groups, policies, networks, DNS, posture checks, setup keys, account settings) as YAML files stored in Git. Changes are reviewed via pull requests and applied via the `configure_netbird` playbook. ``` Edit YAML → PR → Review → Merge → Apply ``` ## Prerequisites Install the collection: ```bash ansible-galaxy collection install community.ansible_netbird ``` ## Quick Start ### 1. Export Current State Capture your current NetBird configuration as clean YAML files: ```bash ansible-playbook community.ansible_netbird.export_netbird_config \ -e "netbird_api_url=https://netbird.example.com" \ -e "netbird_api_token=your-token" ``` Output is written to `/tmp/netbird_config_export/` by default. ### 2. Set Up Your Config Directory Copy the exported files to your project (or start from the skeleton): ```bash # From export cp -r /tmp/netbird_config_export/ my_netbird_config/ # Or from skeleton (empty defaults with commented examples) cp -r ~/.ansible/collections/ansible_collections/community/ansible_netbird/config_skeleton/ my_netbird_config/ ``` ### 3. Preview Changes The playbook runs in **preview mode by default** — read-only, no modifications: ```bash ansible-playbook community.ansible_netbird.configure_netbird \ -e "config_dir=$(pwd)/my_netbird_config" \ -e "netbird_api_url=https://netbird.example.com" \ -e "netbird_api_token=your-token" ``` ### 4. Apply Changes ```bash ansible-playbook community.ansible_netbird.configure_netbird \ -e "config_dir=$(pwd)/my_netbird_config" \ -e "netbird_api_url=https://netbird.example.com" \ -e "netbird_api_token=your-token" \ -e "commit=true" ``` ## Config Directory Structure ``` my_netbird_config/ ├── settings.yml → netbird_settings ├── networks.yml → netbird_networks ├── setup_keys.yml → netbird_setup_keys (optional) ├── access_control/ │ ├── groups.yml → netbird_groups │ ├── posture_checks.yml → netbird_posture_checks │ └── policies.yml → netbird_policies └── dns/ ├── nameservers.yml → netbird_dns_nameserver_groups ├── zones.yml → netbird_dns_zones └── settings.yml → netbird_dns_disabled_management_groups ``` The directory layout mirrors the NetBird UI navigation: - `settings.yml` — Account-wide settings (UI: Settings) - `access_control/` — Groups, posture checks, policies (UI: Access Control) - `dns/` — Nameservers, zones, DNS settings (UI: DNS) - `networks.yml` — Networks with routers and resources (UI: Networks) - `setup_keys.yml` — Peer enrollment keys (optional, UI: Setup Keys) ## Resource Dependencies Resources are applied in dependency order (handled automatically): 1. **Account settings** — no dependencies 2. **Posture checks** — no dependencies (referenced by policies) 3. **Groups** — no dependencies (referenced by everything else) 4. **Setup keys** — depends on groups (for auto_groups) 5. **DNS** — depends on groups 6. **Networks** — depends on groups 7. **Policies** — depends on groups + posture checks ## Adding/Modifying Resources Edit the appropriate file in your config directory. ### Groups ```yaml # access_control/groups.yml netbird_groups: - name: "developers" state: present - name: "production-servers" state: present - name: "deprecated-group" state: absent # Will be deleted ``` ### Posture Checks ```yaml # access_control/posture_checks.yml netbird_posture_checks: - name: "minimum-version" description: "Require minimum NetBird version" checks: nb_version_check: min_version: "0.25.0" state: present ``` ### Policies ```yaml # access_control/policies.yml netbird_policies: - name: "developers-ssh" description: "Allow developers SSH to production" enabled: true source_posture_checks: - minimum-version rules: - name: "ssh-access" description: "SSH access rule" enabled: true sources: - developers destinations: - production-servers bidirectional: false protocol: "tcp" ports: ["22"] action: "accept" state: present ``` ### DNS Nameservers ```yaml # dns/nameservers.yml netbird_dns_nameserver_groups: - name: "corporate-dns" description: "Corporate DNS servers" nameservers: - ip: "10.0.0.53" ns_type: "udp" port: 53 groups: - developers domains: - "corp.example.com" enabled: true primary: false state: present ``` ### DNS Zones ```yaml # dns/zones.yml netbird_dns_zones: - name: "Office Zone" domain: "office.example.com" enabled: true enable_search_domain: false distribution_groups: - developers records: - name: "server1" type: "A" content: "10.0.1.1" ttl: 300 state: present ``` ### Networks ```yaml # networks.yml netbird_networks: - name: "internal-network" description: "Corporate internal network" routers: - peer: "gateway-peer-id" # Peer IDs stay as-is (peers are dynamic) metric: 100 masquerade: true enabled: true resources: - address: "172.16.0.0/16" name: "internal-range" enabled: true groups: - developers state: present ``` ### Setup Keys ```yaml # setup_keys.yml netbird_setup_keys: - name: "server-enrollment" key_type: "reusable" # one-off or reusable expires_in: 604800 # 7 days in seconds auto_groups: - servers # resolved to ID automatically usage_limit: 0 # 0 = unlimited (reusable only) ephemeral: false state: present ``` > **Important:** Setup key values are returned by the NetBird API **only at creation time**. On subsequent runs, the key exists and `changed=false` is reported — the key value is not retrievable. Store key values securely when they are first created. The role registers results as `netbird_setup_key_results`, making key values available to `post_tasks:` in wrapper playbooks for downstream storage (e.g., HashiCorp Vault, AWS Secrets Manager). ### Setup Key Rotation The NetBird API does not support regenerating a setup key — the Update endpoint can only change `revoked` and `auto_groups`. Rotation requires a two-phase approach: **Phase 1 — Create new key alongside old (both active):** ```yaml # setup_keys.yml netbird_setup_keys: - name: "server-enrollment" # existing key, still active key_type: "reusable" auto_groups: [servers] - name: "server-enrollment-rotated" # new key key_type: "reusable" auto_groups: [servers] ``` Apply with `commit=true`. The new key is created and its value is displayed (or stored via `post_tasks:`). Deploy the new key to clients while the old key still works. **Phase 2 — Revoke old key after clients migrated:** ```yaml # setup_keys.yml netbird_setup_keys: - name: "server-enrollment" revoked: true # revoked, no longer usable - name: "server-enrollment-rotated" key_type: "reusable" auto_groups: [servers] ``` The old key is revoked. You can later change it to `state: absent` to delete it entirely. ## Name-Based ID Resolution Config files use **plain names** for groups and posture checks — no IDs, no Jinja2 syntax. The playbook resolves names to API IDs automatically at runtime. How it works: 1. Groups and posture checks are created first (dependency order) 2. The playbook fetches all groups/posture checks from the API and builds lookup maps 3. When applying DNS, networks, and policies, names are resolved to IDs transparently ```yaml # Just use plain names — the playbook handles the rest sources: - developers distribution_groups: - production-servers source_posture_checks: - minimum-version ``` > **Note:** Router `peer` values in networks remain as peer IDs because peers are dynamic (they register via setup keys). The export playbook annotates peer IDs with hostnames in comments for reference. **Not managed by IaC (intentional):** - **Peers** — dynamic, register via setup keys - **Users** — managed via IdP/LDAP sync ## Previewing Changes (Dry-Run Diff) The playbook runs in **preview mode by default** — read-only, no modifications. To apply changes, pass `-e "commit=true"`. Example output: ``` ══════════════════════════════════════════════════════════════ NETBIRD CONFIGURATION PREVIEW — MY_NETBIRD_CONFIG ══════════════════════════════════════════════════════════════ ── Groups ──────────────────────────────────────────────────── + ADD: "monitoring-servers" - REMOVE: "deprecated-group" ~ EXISTS (re-apply): "developers" ── Policies ────────────────────────────────────────────────── ~ EXISTS (re-apply): "developers-ssh" ══════════════════════════════════════════════════════════════ SUMMARY ────────────────────────────────────────────────────────────── + Add: 1 resource(s) ~ Exists: 2 resource(s) - Remove: 1 resource(s) ══════════════════════════════════════════════════════════════ ``` ## Strict Mode (Full IaC Enforcement) By default, the playbook only manages resources defined in YAML. Resources created manually in the webUI are left untouched. With **strict mode**, any resource in the API that is NOT defined in your YAML config will be deleted — making the YAML files the single source of truth. ```bash # Preview what strict mode would do ansible-playbook community.ansible_netbird.configure_netbird \ -e "config_dir=$(pwd)/my_netbird_config" \ -e "netbird_api_url=https://netbird.example.com" \ -e "netbird_api_token=your-token" \ -e "strict=true" # Apply with strict enforcement ansible-playbook community.ansible_netbird.configure_netbird \ -e "config_dir=$(pwd)/my_netbird_config" \ -e "netbird_api_url=https://netbird.example.com" \ -e "netbird_api_token=your-token" \ -e "commit=true" \ -e "strict=true" ``` **Protected resources** (never deleted by strict mode): - "All" group (auto-managed by NetBird) - JWT-issued groups (synced from IdP) - Peers (dynamic, register via setup keys) - Users (managed via IdP/LDAP) The preview always shows orphaned resources so you can see what strict mode would remove, even without `-e "strict=true"`. ## Using with Inventory (target_hosts) By default, the playbooks run on `localhost`. If you use an Ansible inventory (e.g., with AAP or for multi-environment setups), set `target_hosts` to your inventory group: ```bash ansible-playbook community.ansible_netbird.configure_netbird \ -i inventory \ -e "target_hosts=netbird_control_nodes" \ -l netbird_control_nodes_preprod \ -e "config_dir=/path/to/netbird_config/preprod" \ -e "netbird_api_url=https://netbird.example.com" \ -e "netbird_api_token=your-token" ``` Or create your own playbooks that use the roles directly — this is the recommended approach for inventory-based workflows: ```yaml # configure_netbird.yml (using the role directly) - name: Configure NetBird hosts: netbird_control_nodes gather_facts: false roles: - role: community.ansible_netbird.configure run_once: true vars: config_dir: "{{ playbook_dir }}/../netbird_config/{{ netbird_env }}" ``` Then run with just a limit: `ansible-playbook configure_netbird.yml -i inventory -l preprod` Using roles directly gives you full control over `hosts`, `gather_facts`, and variable resolution — and avoids `import_playbook` path resolution issues in AAP. ## Remote / Public API Endpoint If your NetBird management plane runs on OpenShift or a cloud platform without SSH access, you can run the playbooks against the public API endpoint directly from `localhost`: ```yaml # configure_netbird.yml (remote API) - name: Configure NetBird via Public API hosts: localhost gather_facts: false roles: - role: community.ansible_netbird.configure run_once: true vars: netbird_api_url: "https://netbird.example.com" netbird_api_token: "{{ vault_netbird_token }}" config_dir: "{{ playbook_dir }}/../netbird_config/{{ netbird_env }}" ``` Or use the collection playbook directly: ```bash ansible-playbook community.ansible_netbird.configure_netbird \ -e "config_dir=$(pwd)/netbird_config/prod" \ -e "netbird_api_url=https://netbird.example.com" \ -e "netbird_api_token=your-token" \ -e "commit=true" ``` Both patterns work identically — the only difference is whether Ansible SSHes to a control node (and queries `localhost:33073`) or runs locally against a remote HTTPS endpoint. ## Multi-Environment Setup For managing multiple environments (e.g., production and staging), create separate config directories: ``` netbird_config/ ├── prod/ │ ├── settings.yml │ ├── setup_keys.yml │ ├── access_control/ │ ├── dns/ │ └── networks.yml └── staging/ ├── settings.yml ├── setup_keys.yml ├── access_control/ ├── dns/ └── networks.yml ``` Then target each environment with a different `config_dir`: ```bash # Preview staging ansible-playbook community.ansible_netbird.configure_netbird \ -e "config_dir=$(pwd)/netbird_config/staging" \ -e "netbird_api_url=https://staging-netbird.example.com" \ -e "netbird_api_token=staging-token" # Apply production ansible-playbook community.ansible_netbird.configure_netbird \ -e "config_dir=$(pwd)/netbird_config/prod" \ -e "netbird_api_url=https://netbird.example.com" \ -e "netbird_api_token=prod-token" \ -e "commit=true" ``` ## Disaster Recovery To rebuild NetBird configuration from scratch using your YAML files: ```bash ansible-playbook community.ansible_netbird.configure_netbird \ -e "config_dir=$(pwd)/my_netbird_config" \ -e "netbird_api_url=https://netbird.example.com" \ -e "netbird_api_token=your-token" \ -e "commit=true" ``` All resources defined in your config will be recreated. Add `-e "strict=true"` to also remove resources not in the config files. ## Idempotency All operations are idempotent: - Running the playbook twice produces the same result - Resources that already match the desired state are not modified - `changed=false` is reported when no changes are needed