Issue with Link Failover on Mellanox SN2010 MLAG setup

Infrastructure & Networking Switches and Gateways Ethernet Switches
1

Hi, I believe I’m facing a setup issue. Here is the topology:

Drawing1
Drawing11127×987 74.1 KB

As you can see, I have a pair of Mellanox switches configured with MLAG, which are connected to two standalone DMZ switches — DMZ-SW1 and DMZ-SW2 — using a cross-connection for redundancy.

The issue I’m facing happens during a link failover test. When I remove swp1 from both Mellanox switches, I get disconnected. I tried enabling BPDU Guard and PortAdminEdge, but then I started experiencing flapping (ping drops and recovery).

Could this be an STP issue?

More details:

  1. The DMZ switches are Cisco switches running RSTP.

  2. The Mellanox switches are running Cumulus Linux version 5.12 in pure Layer 2 mode.

  3. STP mode is RSTP.

  4. BPDU Guard is enabled only on swp1 and swp2. The rest of the ports do not have it enabled.

  5. There is only one bridge domain in use (br_default).

  6. The links are configured as trunk ports. (not bonded)

2

I would suggest to open a support case with the Nvidia support group and provide the cl-support from both switches.