Per port polling, is it still "experimental"?


In our network, he have routers called ISG/BNG which are responsible from termination PPPoE sessions from customers. On the router these sessions are treated as virtual interfaces, so when i add these routers to librenms, they are also being processed. Im talking about 20k-30k interfaces, so when i add them up, CPU of the routers are getting %90 values.

When i checked the forum, i see that the only sensible solution is excluding them via “per port polling” feature. But i see that this feature is still “experimental” so i just wanted to if there is any other solution to overcome my problem. Or if im going to proceed with “per port polling” might i face with another issue since its experimental.

Thanks for your answers.

Yes, it is experimental because it doesn’t work better in every case. If it did, we could just make it the default. Improvements welcome.