Port goes down when added to MLAG PortChannel (SN2010M)

Port configured for 1GB and 10GB speed has a 1GB Transceiver.

When added to a MLAG PortChannel, the port goes down.

Is the SN2010M expecting 10GB links in an MLAG PortChannel?


SN2010M is an HPE part number, so the best place to ask for assistance is through HP Enterprise support. SN2010 is the NVIDIA product that is almost identical.

If this was a SN2010, the best way to start looking into this is to check if/why the interface is down as the physical port should be able to come up regardless of port-channel config when a supported SFP/cable is connected to a remote device configured to match. You can get details on the autonegotiation, speed, FEC, power, linkstate, and more with the following command (replace X with the front panel port number)

Onyx Interface State:
For SN2xxx switches:
X is the port number
(config)# fae mlxlink -d /dev/mst/mt52100_pciconf0 -p X -m -e -c

Cumulus Linux/Sonic:
sudo mlxlink -d 03:00.0 -p X -m -e -c

If the physical port does not report any issues, then there is probably a misconfiguration between the two devices related to LACP. All the ports in a bond are required to be running at the same speed, and all links on both sides of the bond need to use the same LACP system MAC and be part of the same bond. If for example you have 2 bonds on the same switch cabled to one bond or across 2 different switches cabled to one bond, then links in only a single bond would be allowed to be up due to the mismatched LACP bond ID or system ID respectively. In this case all the interfaces on each side of the bond would need to be placed in the same bond/port-channel or the same mlag respectively.

The best place to troubleshoot this would be a support case, so please contact the support for your switch after checking if this is a NVIDIA or HPE product.