Ansible and the Labgrid playbook (openwrt-tests)¶

playbook_labgrid.yml and Ansible inventory live in aparcar/openwrt-tests (ansible/playbook_labgrid.yml, ansible/inventory*). fcefyn_testbed_utils adds roles, templates, and ansible/playbook_testbed.yml for FCEFyN-specific extras (Arduino relay, PoE, ZeroTier, observability). Host deployment (tags, files, ZeroTier, etc.): host-config - Ansible.

1. Summary¶

The Ansible playbook (playbook_labgrid.yml in aparcar/openwrt-tests) automates full setup of a lab host:

User labgrid-dev, SSH keys, groups
Labgrid (pipx), PDUDaemon, dnsmasq
labgrid-switch-abstraction (pipx, provides the switch-vlan CLI)
Netplan (VLANs), exporter, coordinator
TFTP, udev, labgrid-bound-connect

For the full contribution flow (openwrt-tests-only lab vs libremesh-capable lab), see Contributing a new lab.

Everything host-config describes as manual configuration can be deployed with this playbook, provided ansible/files/exporter/<lab>/ exists with lab-specific files (netplan, dnsmasq, pdudaemon, exporter).

2. Concepts: control node vs managed node¶

Role	Description
Control node	Machine where `ansible-playbook` runs. Must have Ansible installed and access (SSH or local) to managed nodes.
Managed node	Machine Ansible configures. Receives playbook tasks (packages, files, services).

Example: If ansible-playbook runs from the Lenovo to labgrid-fcefyn, the Lenovo is the control node and host labgrid-fcefyn is the managed node. When that host is the same Lenovo, use ansible_connection=local to avoid SSH to itself.

3. Use cases and flows¶

3.1 Maintainer configuring their own lab (openwrt-tests)¶

The Lenovo is both control and managed node. Run the playbook locally against the same host.

flowchart LR
    subgraph lenovo ["Lenovo (labgrid-fcefyn)"]
        CN["Control node - ansible"]
        MN["Managed node - local lab"]
    end
    CN -->|local| MN

Hold "Alt" / "Option" to enable pan & zoom

Inventory: labgrid-fcefyn ansible_connection=local
Command: ansible-playbook playbook_labgrid.yml -l labgrid-fcefyn -K

From the Lenovo (control node), with the lab in inventory and ansible_host set:

3.2 Maintainer configuring another contributor's lab (openwrt-tests)¶

When a developer contributes their lab to openwrt-tests, the maintainer runs the playbook from the Lenovo toward the contributor's host. The contributor provides SSH access (public key, VPN, etc.).

flowchart LR
    subgraph lenovo_ctrl ["Lenovo (control node)"]
        AP["ansible-playbook -l labgrid-dev-x"]
    end
    subgraph lab_dev ["Dev lab (managed node)"]
        LG["labgrid-dev - dnsmasq, exporter"]
    end
    AP -->|SSH| LG

Hold "Alt" / "Option" to enable pan & zoom

Inventory: labgrid-dev-x ansible_host=192.168.x.x (or resolvable hostname)
Command: ansible-playbook playbook_labgrid.yml -l labgrid-dev-x -K

ansible-playbook playbook_labgrid.yml -l labgrid-fcefyn -K --check

3.3 openwrt-tests (Aparcar) configuring the lab¶

The Lenovo is only a managed node. The control node lives in Aparcar/upstream infrastructure.

flowchart LR
    subgraph aparcar ["Aparcar (control node)"]
        Inv["Remote inventory"]
        Play["Remote playbook"]
    end
    subgraph lenovo_mn ["Lenovo (managed node)"]
        Lab["labgrid-fcefyn"]
    end
    Play -->|SSH| Lab

Hold "Alt" / "Option" to enable pan & zoom

Do not use the local inventory or playbook. Aparcar has theirs and they point at the lab host. The Lenovo only needs to be reachable by SSH from their control node.

3.4 Tagged tasks only¶

If the playbook uses tags (e.g. export for exporter/coordinator):

ansible-playbook playbook_labgrid.yml -l labgrid-fcefyn -K --tags export

3.5 Summary of Lenovo roles¶

Scenario	Lenovo is
Self-setup of own lab (3.1)	Control + managed (local)
Setting up another contributor's lab (3.2)	Control node
Aparcar managing the host upstream (3.3)	Managed node only

4. Files and purpose¶

All of this lives in the aparcar/openwrt-tests repo, under ansible/:

File	Purpose
`inventory.ini`	Host list (labs, coordinator). Defines connection (local or SSH) and `ansible_host` if needed.
`ansible.cfg`	Ansible settings (default inventory, etc.).
`playbook_labgrid.yml`	Main playbook. Host setup tasks.
`files/exporter/<lab>/netplan.yaml`	Network config (VLANs) for that lab. Copied to `/etc/netplan/labnet.yaml`.
`files/exporter/<lab>/dnsmasq.conf`	dnsmasq config. Copied to `/etc/dnsmasq.conf`.
`files/exporter/<lab>/pdudaemon.conf`	PDUDaemon config (power control).
`files/exporter/<lab>/exporter.yaml`	Labgrid exporter config.
`files/coordinator/places.yaml.j2`	Template to generate coordinator `places.yaml`. Uses `labnet.yaml`.
`files/exporter/labgrid-exporter.service`	Exporter systemd unit.
`files/exporter/pdudaemon.service`	PDUDaemon systemd unit.
`files/coordinator/labgrid-coordinator.service`	Coordinator systemd unit.
`files/exporter/labgrid-bound-connect`	Script for SSH connections bound to a VLAN interface.
`files/exporter/usbsdmux.rules`	udev rules for USB-SD-Mux (if applicable).

labnet.yaml is at the repo root (not under ansible/) and is loaded as variables for the playbook.

5. Inventory¶

5.1 Structure¶

[labs]
labgrid-fcefyn ansible_connection=local
# labgrid-other-dev ansible_host=192.168.1.50   # When someone else contributes

[coordinator]
labgrid-fcefyn ansible_connection=local

[labs]: Hosts the playbook configures (exporter, dnsmasq, netplan, etc.).
[coordinator]: Host running labgrid-coordinator. Can be the same as a lab.

5.2 `ansible_connection=local`¶

When the managed node is the same machine running Ansible, use ansible_connection=local. That avoids SSH and runs tasks locally. Required for case 3.1 (self-setup).

5.3 Adding another contributor's lab¶

The contributor provides SSH (key or network access).

Add a line to inventory.ini:

labgrid-new-lab ansible_host=IP_OR_HOSTNAME

Create ansible/files/exporter/labgrid-new-lab/ with netplan, dnsmasq, pdudaemon, exporter.yaml.
Add the lab and its devices to labnet.yaml.
Run: ansible-playbook playbook_labgrid.yml -l labgrid-new-lab -K

6. Playbook¶

6.1 Variable loading¶

The playbook needs labs from labnet.yaml to:

Fetch developers' SSH keys from GitHub (using maintainers + access lists) and add to labgrid-dev
Generate the TFTP device/instance list
Generate coordinator places.yaml

Options:

vars_files: - ../labnet.yaml in the playbook (recommended): automatic load.
-e @../labnet.yaml on the command line: manual load.

6.2 What the playbook does¶

Approximate task order:

Task	Result
Set hostname	Managed node hostname
Create labgrid-dev	User for Labgrid
Add SSH keys	Fetch from GitHub for `maintainers` + `access` users, add to labgrid-dev
Install packages	pipx, microcom, ser2net, socat, iptables, etc.
Install Labgrid	via pipx
Install PDUDaemon	via pipx
Install labgrid-switch-abstraction	via pipx (`switch-vlan` CLI for optional VLAN automation)
Configure netplan	Copy host-specific `netplan.yaml` to `/etc/netplan/labnet.yaml`
Configure dnsmasq	Copy host-specific `dnsmasq.conf`
Configure PDUDaemon	Copy `pdudaemon.conf`, deploy unit
Copy exporter	Copy `exporter.yaml` to `/etc/labgrid/`
Add places.yaml	Generate from `labnet.yaml`
labgrid-bound-connect	Script + sudoers for bound SSH
Udev rules	USB-SD-Mux (if applicable)
TFTP folders	`/srv/tftp/` structure per device instance
Enable IP forwarding	For routing between VLANs
Restart services	dnsmasq, coordinator, exporter, pdudaemon

6.3 Host-specific file priority¶

For netplan, dnsmasq, and pdudaemon, the playbook checks files/exporter/<inventory_hostname>/. If present, it uses that file; otherwise the default template under files/exporter/. For labgrid-fcefyn the host-specific directory should always exist.

7. labnet.yaml¶

File at the openwrt-tests repo root (../labnet.yaml relative to ansible/); see aparcar/openwrt-tests/labnet.yaml. Not part of Ansible directly; loaded as variables.

7.1 Structure¶

devices:
  openwrt_one: { name, target, firmware }
  bananapi_bpi-r4: ...
  linksys_e8450: ...
  librerouter_librerouter-v1: ...

labs:
  labgrid-fcefyn:
    devices: [linksys_e8450, openwrt_one, ...]
    device_instances:
      linksys_e8450: [belkin_rt3200_1, belkin_rt3200_2, belkin_rt3200_3]
      librerouter_librerouter-v1: [librerouter_1]
    maintainers: [francoriba, ccasanueva7, javierbrk]
    access: [francoriba, aparcar]

No top-level developers: block with sshkey: fields - since commit 2f1aed1 SSH keys are fetched from GitHub.

7.2 Use in the playbook¶

labnet.labs[inventory_hostname] → current lab config (maintainers, access, devices).
lookup('url', 'https://github.com/' + item + '.keys') → SSH keys fetched at deploy time.
lookup('file', '../labnet.yaml') → generating places.yaml.

8. Commands by scenario¶

Sudo requirement¶

The playbook uses become (sudo) to copy files under /etc/, restart services, etc. If the user has no passwordless sudo:

Use -K (or --ask-become-pass) so Ansible prompts for the password.
Or configure passwordless sudo (typical on the lab host): echo "USER ALL=(ALL) NOPASSWD: ALL" | sudo tee /etc/sudoers.d/labgrid

Example: ansible-playbook playbook_labgrid.yml -l labgrid-fcefyn --ask-become-pass.

8.1 Self-setup (local lab on the Lenovo)¶

From the control node (the Lenovo):

cd ~/testbed_fcefyn/openwrt-tests/ansible
ansible-playbook playbook_labgrid.yml -l labgrid-fcefyn -K

-K asks for the sudo password. Run from ansible/ so ansible.cfg picks up the right inventory. If the playbook has vars_files: - ../labnet.yaml, you do not need -e @../labnet.yaml.

8.2 Setup of another contributor's lab¶

cd ~/testbed_fcefyn/openwrt-tests/ansible
ansible-playbook playbook_labgrid.yml -l labgrid-new-lab -K

8.3 Dry run (check mode)¶

ansible-playbook playbook_labgrid.yml -l labgrid-fcefyn -K --check

Ansible and the Labgrid playbook (openwrt-tests)¶

1. Summary¶

2. Concepts: control node vs managed node¶

3. Use cases and flows¶

3.1 Maintainer configuring their own lab (openwrt-tests)¶

3.2 Maintainer configuring another contributor's lab (openwrt-tests)¶

3.3 openwrt-tests (Aparcar) configuring the lab¶

3.4 Tagged tasks only¶

3.5 Summary of Lenovo roles¶

4. Files and purpose¶

5. Inventory¶

5.1 Structure¶

5.2 ansible_connection=local¶

5.3 Adding another contributor's lab¶

6. Playbook¶

6.1 Variable loading¶

6.2 What the playbook does¶

6.3 Host-specific file priority¶

7. labnet.yaml¶

7.1 Structure¶

7.2 Use in the playbook¶

8. Commands by scenario¶

Sudo requirement¶

8.1 Self-setup (local lab on the Lenovo)¶

8.2 Setup of another contributor's lab¶

8.3 Dry run (check mode)¶

5.2 `ansible_connection=local`¶