We have a Routerboard at a client site with the following config -
Ether 1 - Uplink to us.
Ether 3 - Slave of 1 (no ip address) - Connected to a Cisco Switch - behind which is our Sip Proxy device (Alix)
Ether 4 - Slave of 1 (no ip address) - Connected to ATA device.
Devices connected to Ether3/4 share the addresses from the Ether1 Wan subnet.
What we want to do is identify Voip traffic coming to us from Ether1, mangle the traffic, and then apply QoS on Ether1 to prioritize Voip over data. The phones are behind the Alix (sip proxy) connected to Ether3.
The issue I’m having is the RouterOS does not seem to be recognizing any Voip based connections on Ether1. And thus the mangle rules do not pick up any packets to mark. On interfaces tab, I do see traffic generated during phone calls on Ether3, which in turn then goes to Ether1 and I see bytes racking up there as well - but I see nothing on Firewall->Connections. I also tried Tool->Sniffer on Ether1 and don’t see any UDP or SIP traffic. I tried to Sniff on Ether3 but that does not work, I’m assuming it is because it is a slave/switched port.
Upon some more reading, it might be because the Master-Slave ports are using the switching hardware and not making it to RouterOS? if anyone could shed some light on this that would be great.
What kind of RouterBoard are you using? Typically, you will need traffic to go through the CPU which means bridging. In the CRS series, you can use switching and only bridge the CPU for routing out of the subnet.
in the short term, you can bridge all the needed ports together which will allow the mangle to see and mark the traffic into a queue.
I’m a bit new to this, but I’ve successfully gotten similar tasks done. It might be a good idea to post at least the relevant parts of your queue and mangle configs.
I had a lot of confusion when I failed to see traffic going a certain direction in my queues, and I found out that I needed to fix my queue target and destinations. If you’re seeing traffic go both ways in the queue graph this probably isn’t your problem.
My other thought is to make sure you’re not doing anything with fasttrack that might go around your mangles.
I would start with the queue graphs and make sure you’re seeing the appropriate traffic going in/out of your queue.
Are you marking the connection and then marking the packets?