LACP Bonding in ROS on a CRS328 switch

RouterOS General
1

Hi all,

I tried to setup an aggregated link on two SFP+ ports between a CRS328 switch and a Ubiquiti US-48.

The Ubiquiti AFAIK only does aggregation via LACP, so the first attempt has been with SwOS and it kinda worked (nothing to do, just left both ports on the SwOS as passive): both ports recognized the trunk on the two ports as the same (1) with the same partner mac id from the US-48. However a couple of days later, looking at my router logs I noticed a storm of:

ether2: bridge RX looped packet - MAC 18:fd:74:XX:XX:XX -> ff:ff:ff:ff:ff:ff VID 1 ETHERTYPE 0x0800 IP UDP 0.0.0.0:68 -> 255.255.255.255:67

The router is a HeX S router (recently updated to 7.9.2) which is connected to the Ubiquiti US-48 on the ether2 port.
As soon as I disconnect one of the two SFP+ connectors, the storm of looped packet stops immediately.

I tried to reboot the CRS328 in ROS and set up the bond but immediately hit the wall as I cannot remove the 2 sfp+ ports from the bridge (couldn’t remove - ether devices are not removable), and apparently cannot add them to the bond either if they’re part of the bridge. How should I proceed?

I’m a bit puzzled on the cause of this as I got the feeling (before noticing the log) that the aggregation worked.

2

Some more insights after upgrading HeX S to 7.10 and inspecting the log thoroughly with the help of packet sniffer. There’s an infinite flood of warning regarding all the DHCP servers on VLANs offering a lease for a MAC which belongs to the HeX S itself and is not being accepted:

dhcp-vlan-44 offering lease 10.44.3.254 for 18:FD:74:XX:XX:XX without success

Why is the HeX S offering leases to itself and how is this affected by the LAG? I’m a bit confused. On the packet sniffers I see a tx of DHCP lease offer immediately followed by a rx of the same packet, repeated for all VLANs and the bridge interface where I keep the DHCP for the untagged traffic.
I obviously thought of an accidental DHCP client created during the update, but that’s not the cause of course (it was a long shot).

Anyone can help shed some light?
Thanks

3

Addendum: I have the feeling this didn’t happen with 7.8 and this LAG setup has been pretty much up for 5 weeks.
Shall I try downgrading to 7.8 and see what happens? I cannot think of any other changes.