I have 3 routers for my home network, all running RouterOS 6.46.4
RB3011 - used for work PC, Linux stuff PC, Linux dev test “server”. 192.168.1.0/24
RB941 - used for Android TV (wifi), XBOX (wired) and phones/tablets (wifi) 192.168.2.0/24
RB931 - used for Android TV (wifi) and a gaming PC (wired) and phones/tablets (wifi) 192.168.3.0/24
RB3011 eth1 connects to fiber CPE
RB941 eth1 (192.168.1.2) connects to RB3011 eth6
RB931 eth1 (192.168.1.3) connects to RB3011 eth3
All 3 routers are configured as “Router” in Quick Set.
All ports except eth1 is bridged on all routers.
I also have hairpin NAT configured on the RB3011 due to work I do on the Linux dev test "server’.
The first problem I have is that my Linux stuff PC on 192.168.1.0/24 cannot see the gaming PC on 192.169.3.0/24 - I can’t figure out the cause.
Presumably this is a symptom of the above:
ping 192.168.3.235
PING 192.168.3.235 (192.168.3.235) 56(84) bytes of data.
From 192.168.1.1: icmp_seq=2 Redirect Host(New nexthop: 192.168.1.3)
From 192.168.1.1: icmp_seq=3 Redirect Host(New nexthop: 192.168.1.3)
The second issue is more a question regarding performance. When I use the BTest Server and Bandwidth Test tools I find the throughput is not that great when doing a TCP test - around 30Mbps. This seems much too low? I hope it has something to do with my configuration on the routers.
Any help will be much apppreciated, even just real basic pointers on where I am getting it wrong.
I guess the problem is that with default settings RB931 and RB941 both perform nat between their “LAN” and “WAN” sides (and RB3011 is on the WAN side of both).
And both also perform firewalling which might be what you intended or it might not be. In any case it affects throughput as both devices are relatively weak devices.
I’m sure you could simplify your LAN setup so that both “small” devices would only act as switches. So all 3 LAN subnets would be directly routed by RB3011 (it woul have 3 LAN IP addresses and would route/firewall traffic between them).
If you don’t want to firewall traffic between parts of home LAN, then the setup could be simplified even further to single IP subnet.
As you didn’t describe what kind if network you actually want to have (you just mentioned what you do have) I won’t go into details of each possible solutions.
I want to keep things as simple as possible. All PCs and “servers” must be reachable from within the network, irrespective of subnet.
The “small” routers don’t firewall, it’s not required. And yes, I see NAT is enabled on both.
So putting the little ones in bridge mode should do the trick? In that event, do I exclude their ports (the ports they are connected to on the RB3011) from the bridge on the RB3011?
Because I did a System → Reset Configuration (with no default config after reset) on each router I also had to
configure the WLAN again
set the System Identity and Password
check that the MAC address from each router still matched the expected MAC
All good. The result is a nice flat network with everything on 192.168.1.0/24
It seems the BT Test results (send or receive, not both) have also improved a lot, which is great. I guess the RB941 and RB931 are just too low spec to provide a respectable “both” throughput result.
The limitation of bandwidth-test is that the number of server processes it starts often saturate the CPU well before the communication channel. It is best performed to a powerful router beyond the router you want to test. Using the “both” mode doubles the number of server processes, which degrades comm performance way more than half.
