I am in the process of upgrading a network I bought from all dumb switches to RB493AHs, to later convert to a routed network.
As I’ve started putting out RB493AHs, I’ve noticed that traffic will stop passing, and generally a reboot of one of the RB493AHs will remedy the problem across the whole network. The watchdog will catch this lack of communication and reboot a router.
Is this some sort of broadcast issue? Ideas to fix it?
Could you post a diagram of how your network is constructed? What ROS version are you running, and is spanning tree enabled on these bridges? STP or RSTP?
10.255.1.1 is my first RB. There are RBs at ICS, Mill Rd., and Mill Rd. just after the other side of the UBNT links.
ICS is 3.31, all others are 4.10.
There is no spanning tree of any kind because there are no loops.
bump
I could’ve sworn I had posted a reply here.
So, all of the orange devices are RouterBOARDS? I don’t imagine your bridge configuration is anything complex; that setup should work. You may want to enable spannning tree on the RB’s just so they have a correct ‘view’ of the network topology and avoid blocking any upstream ports.
Also, what is the CPU load percentage when the network locks up? High load would likely indicate a broadcast storm or packet loop.
There are RB493AHs at ICS, Mill Rd. Far South, and then the 10.255.1.1 in the center.
Graphing wasn’t enabled on all of them, so I just turned that on. Didn’t want to trust the Dude, since it would lose connectivity to the box.
So just turn on STP for each of the bridges?
Sure. That would be a first step. Assuming you only have a single bridge on each RB you’ll want to run ‘/interface bridge set 0 protocol-mode=rstp priority=4096’ on 10.255.1.1, then ‘/interface bridge set 0 protocol-mode=rstp’ on the others.
You may want to go into Safe Mode before you run those commands as you may get disconnected, but not likely. You can monitor the status of each bridge and whether or not it sees the root bridge with ‘/interface bridge monitor 0’.