Is there a way to streamline this use case?

[lwhitworth]

Apologies if I'm missing something obvious, or trying to ram a square peg into a round hole......but.....I'm trying to use netlab to validate our Broadcom Enterprise SONiC switch configs which are entirely built in Ansible already.

My topology (at the moment) looks like:

---
defaults.device: sonic
nodes:
  exit-01:
  core-01:
  dist-01:
  vlan-003-client1:
    device: linux
    provider: clab

links:
  - exit-01:
      ifindex: 1
    core-01:
      ifindex: 4
    pool: l2only
  - core-01:
      ifindex: 1
    dist-01:
      ifindex: 1
    pool: l2only
  - dist-01:
      ifindex: 3
    vlan-003-client1:
      ifindex: 1
    pool: l2only

tools:
  graphite:

If I run netlab up --no-config and then run our ansible playbooks against them everything all comes up perfectly and (once I configure it with the right IP) vlan-003-client1 can ping all the way up to exit-01 and get a response. So that's brilliant and exactly what I need.

However, once I netlab down it everything's gone and I'm back to a blank slate. This I expect, but no level of reading docs and netlab collect exploration seems to be able to get me to a point where I can just netlab up the lab and continue where I left off.

Am I just missing something obvious here?

[lwhitworth]

Just to further my rambling chain of thought on this, after a bit more playing I suppose I could just do something akin to:

#!/bin/bash
netlab up --no-config
ansible-playbook playbooks/deploy_simulation.yml --limit dist-01,core-01,exit-01
netlab exec vlan-003-client1 /sbin/ip addr add 192.168.29.10/24 dev eth1
netlab exec vlan-003-client1 /sbin/ip route replace default via 192.168.29.200
netlab exec vlan-003-client1 ping 192.168.23.151 # test visibilty from a "user" device all the way up to the exit node which is using 192.168.23.151

That'll work for some of my use cases.....but if I could bring the SONiC nodes up with a pre-determined minimum config then I wouldn't have to run the whole deploy_simulation.yml every time. Still, the tool has allowed me to get this far so massive thanks for that 😄

[ipspace]

There's no standard way (yet) to do what you'd like to achieve, so you'll have to work with a custom bash script at the moment. However:

• Someone already implemented configuration gathering for SONiC, so you could download the configurations with netlab collect --limit sonic before shutting down the lab
• You can configure groups of devices (in your case, the Linux hosts) with netlab initial --limit group. Device type is always an Ansible group, so you could add netlab initial --limit linux to get the addressing and IP routing on Linux hosts configured.
• You could also use the collected SONiC configurations and deploy them with a custom Ansible playbook.

If you find a generic solution to the "deploy collected SONiC configurations", submit a PR and I'll integrate it to make "netlab up --reload-config" work for you.

[lwhitworth]

Thanks for the prompt response, really appreciated. If I end up with anything useful I'll certainly submit a PR.