Summary

When building a large-ish topology (34 devices), there was an IP clash on the ansible node and device node. Both the device and the ansible node were allocated 192.168.0.6.

Expected Behaviour

Expected the ansible node to have the last ip, especially since it is one of the last devices in the .yml file.

Workaround

Removed the ansible node, as it is not being used in this lab anyway.

If using port-channels in the underlay with core_interfaces data model, the peer-interface will show up as Port-ChannelX in the generated topology file which doesn't work.

Not sure how to work around this at the moment...

Summary

At the moment, the logic just uses the management interfaces from structured config.

ACT only supports the management interfaces being in a single /24 network, however the AVD configuration may have generated config for devices with mgmt interfaces in different subnets.

Desired behaviour

• As it is now, the generated configuration will not work. So we should at least error out and not generate a topology that cannot be used.

New format, instead of defining neighbors per node and having to define every link in 2 places, this can be used:

links:
  - connection:
    - <host1>:<interface>
    - <host2>:<interface>