As we can see in the picture, I have a RB450 (Switch1) which act as a switch.
At Ether1, I have unmanageable switch (Switch2) connected which located on different floor of my office building.

One day, a staff, by a mistake plug a UTP cable on switch2 at port 2 and plug the other end of the cable at port 3.
Then the network goes down since the looping occured.
I can see at Switch1 ether1 RX/TX goes to 75Mbps/0Mbps and cpu resource become 100%.
I also try to torch to examine what packet goes to ether1 and bridge1, but I can see no information on torch.

Is there any way to prevent this situation by disabling automatically ether1 at Switch1 when it happen again?

I already try enable STP/RSTP on bridge1 at Switch1. And set higher path cost (20) at bridge port ether1.
And it does not works. I understand that STP works base on BPDU, so it will not works since BPDU not received at other port on Switch1.

Any solution?

BTW, I'm sorry for my bad english and a picture above It is easier for me to use my hand to draw picture rather than use any software

The only (partial) solution I can think of is to use "/tool traffic-monitor" on switch 1 to trigger a script when a utilization threshold is crossed, shutting down the port. Of course that will also prevent legitimate traffic from heavily utilizing the port and could potentially aggravate the users with false positives, particularly since you'd have to only use an 'above threshold' script and not automatically re-enable the port when traffic goes down again since you'd be in a loop at that point, too - threshold is crossed, port shuts down, traffic goes down, threshold is crossed, port is opened, traffic goes up etc. - only this time you're potentially using even more resources on switch 1.

The better solution is to swap out switch 2 for a managed one that can participate in spanning tree and prevent loops altogether.

Hi fewi,

Yes i already think about that solution too. But the problem, just like you said, this setup will prevent legitimate traffic from heavily utilizing the port. Anyway I'll try it and see what will happen. Thank you for your solution

Aldhy

Search the forum. Someone suggested a solution based on pinging a host known to be good and taking action based on the type and number of packets which return to detect a loop.