I used TACACS+ using image form here that work well with both following methods, until I see the accounting logs:
By docker run, it can logs the actual switches address for the NAS field.
By Rancher Deployment, it logs the Rancher Nodes address instead the switches address for the NAS field, like this with yellow marked.
Is it expected behaviour for Rancher/K8s? Or it need some tweak on my deployment? Here is my Port mapping setup.
Honestly, I'd prefer to use method no. 2 due to its beneficial (scaling, self-healing, and so on).
I think I found the culprit by manage hostNetwork from false to true, it works now.
...
spec:
containers:
- image: lfkeitel/tacacs_plus
name: tacacs
...
hostNetwork: true
...
...