I have 3 radios that are all part of the same network. These all are powered/routed by a single RB750UP. My goal is to expand this setup to each radio having its own RB750UP in the event that one of the RB750 switches fails. I have migrated one other RB750UP and it keeps dropping 3-4 pings at a time.
What routing protocols should be in place for the following configuration shown in the photo? OSPF? RIP? Nothing?
How does the configuration of the ports being either all in a bridge or master/slave config affect routing?
Thanks guys. 
The simpliest way is to clean all configuration, put all ether2..5 as slave on each device…
If one RB750UL is broken, simply move cable to another working, without do nohing more…
But the most secure way is to use only one device, and the other disconnected from anything, as spare parts…
Only ports from the same switch group may be master/slave ports. So I could set ports 3,4,5 as a slave to port 2. The purpose of this setup is to create a added redundancy to this system. At some point or another, one of the RB750UP’s will die. When this happens I want to ensure that the network will stay up. The other two radios serve as redundant links for my clients.
Also, why would you consider it more secure to have one device?
Only ports from the same switch group may be master/slave ports
I know, but you not catch the point, ether2…5 on same switch group,
ether 2 linked on main switch with all other ether2, and the acces point can be put on any ether3..5.
If one of the RB750UP die, you must phisical move ethernet to other rb750…
When this happens I want to ensure that the network will stay up.
The network can not stay up, but if your 3 ap cover the same area, and the other two are used for redundancy,
you can script one 750 for start another ap…
Also, why would you consider it more secure to have one device?
Because if power from main line fail with overvoltage/discharge, you broken all the device at the same time…
Is more logical to reduce the fail point putting only two rb750UP,
and only two AP phisiscally connected,
leaving third ap and third rb750 as spare part, unplugged and unpowered.
About AP you must use one board like RB433AH, it have 3 ethernet, you must plug ether1 to first rb750up, ether2 to second rb750up and on rb433ah box power the board from both rb750up, not modifyng the board, but using dedicated connector
one script inside the two rb750up decide what device are powering the rb, or if hte connector are well made, both can power the rb, and if one rb750up fail, you not touch anything and all still working without the minimal interruption…
Can you tell us more about the L2 and L3 relation ship the 750UPs have with each other and the radios? If you need to maintain L2 connectivity between the radios your currently doing the best setup aside from maybe adding a second switch. If the 750UPs are routing between the radios and the switch you could remove the switch and create a full mesh connection between the three 750UPs and use a dynamic routing protocol like OSPF to manage failover.
The 750UP’s are all on the same L3 subnet. They are in between the radios and the main switch as shown in the diagram. I would like to have a dynamic routing protocol in place like OSPF but I wanted to ensure that was ideal for this scenario. I was not aware that you can create a mesh with only switches, could you elaborate on that thought? Thanks
You can not really have a non-blocking mesh of switches. If you connect switches together STP will block all but one of the ports. You mentioned the 750s doing routing so I thought it was different subnets. The next best thing you could do is plug in two 750’s setup VRRP for your L3 gateway and split the radios across the two 750s. If you have only one uplink from the site all of this really accomplishes nothing and it would be a better use of energy to fix grounding, ESD, and lightning protection issues.