Release 26.02

[ipspace]

I'm pretty happy with how far we got in the last month(ish), so it might be time to push out release 26.02. There's no issue that would scream to be fixed, and I'll merge the remaining two PRs in the next few days (having anyone actually look at them would be appreciated, though ;). I might also try to do #2991 (it should be pretty easy).

Any thoughts @DanPartelly @sdargoeuves @jbemmel @ssasso ?

[sdargoeuves]

All good for me, I've tested the dev branch on our main lab, I didn't see anything wrong.
The only thing I had to do was to update, in my playbooks running post netlab up, the extension of the file host_vars/xxx/topology.yml file to json.

[ipspace]

All good for me, I've tested the dev branch on our main lab, I didn't see anything wrong.

Thanks a million.

The only thing I had to do was to update, in my playbooks running post netlab up, the extension of the file host_vars/xxx/topology.yml file to json.

I would love to know how you use those files. I thought Ansible should take care of that change. Anyway, let's add another system default ;))

[ipspace]

Release 26.02 is out

[sdargoeuves]

Sorry @ipspace I haven't been very responsive... just got back from OneNetworkVendor Live 2026

Thanks for offering this solution, but in the end, I've updated the extension. Ansible will read the yaml or the json the same way, I just had to tell it which file to look for, no other changes.

Now to give a bit more context, after doing netlab up we want to create different scenarios, where the network will change. For some playbook, we have started to use the topology.yml (now .json) to avoid hard-coding the scenario, or to reduce the hard-coding part (IP address of p2p link, interface...)

In the example below, we were fixing the interface we wanted to shut. If someone changes the topology, and a new connection is added, then the playbook could shut a different interface. We have scenarios, where the playbook uses hard-coded p2p IP address. It caused quite a mess when I updated the topology, as I had to update a number of those playbooks with the new addressing.

I could also hard-code the p2p IP in the topology, but I prefer not.

---
- name: Shutdown links towards MPLS
  hosts: [ d1xr01, d1xr02, d2xr01, d3xr01, d3xr02 ]
  gather_facts: false

  tasks:
    - name: Load topology data
      include_vars:
        file: "../../host_vars/{{ inventory_hostname }}/topology.json"
        name: topology

    - name: Find interfaces connected to MPLS core (c0xr devices)
      set_fact:
        # Assumes point-to-point links, so we only check the first neighbor (index 0)
        mpls_interfaces: "{{ topology.interfaces | 
                            selectattr('neighbors', 'defined') | 
                            selectattr('neighbors.0.node', 'match', '^c0xr0\\d$') | 
                            map(attribute='ifname') | 
                            list }}"

    - name: Shutdown MPLS interfaces
      cisco.ios.ios_config:
        lines:
          - shutdown
        parents: interface {{ item }}
      loop: "{{ mpls_interfaces }}"

    - name: Display shutdown interfaces
      debug:
        msg: "Shutdown interfaces {{ mpls_interfaces }} on {{ inventory_hostname }}"

At some point, I will go through the other playbooks and make them more dynamic.

I hope that makes sense, let me know if you want more info on how we use it.

[ipspace]

Sorry @ipspace I haven't been very responsive... just got back from OneNetworkVendor Live 2026

No worries.

Thanks for offering this solution, but in the end, I've updated the extension. Ansible will read the yaml or the json the same way, I just had to tell it which file to look for, no other changes.

👍

Now to give a bit more context, after doing netlab up we want to create different scenarios, where the network will change.

Assuming the hosts.yml file is the same as the one generated by netlab, you could just point Ansible to the "correct" inventory source in local ansible.cfg, but yeah, nothing wrong with reading host variables from somewhere else (that's one of the reasons I went with this particular directory structure).

In the example below, we were fixing the interface we wanted to shut.

FWIW, in that particular case, you could use link role attribute to identify the interfaces. It's copied into the interface role attribute.

Thanks for the feedback, Ivan

[sdargoeuves]

Assuming the hosts.yml file is the same as the one generated by netlab

oh yes, I am using the hosts.yml generated by netlab

you could just point Ansible to the "correct" inventory source in local ansible.cfg, but yeah, nothing wrong with reading host variables from somewhere else (that's one of the reasons I went with this particular directory structure).

You're right — I was overcomplicating it by explicitly loading the topology file with include_vars. I can just reference interfaces directly:

- name: Shutdown links towards MPLS
  hosts: [ d1xr01, d1xr02, d2xr01, d3xr01, d3xr02 ]
  gather_facts: false
  tasks:
    - name: Find interfaces connected to MPLS core (c0xr devices)
      set_fact:
        # Assumes point-to-point links, so we only check the first neighbor (index 0)
        mpls_interfaces: "{{ interfaces | 
                            selectattr('neighbors', 'defined') | 
                            selectattr('neighbors.0.node', 'match', '^c0xr0\\d$') | 
                            map(attribute='ifname') | 
                            list }}"

Thanks for pointing it out!