Hi!
I would like to add redundancy to a compute cluster network. Currently a two-port LACP trunk is accepting the traffic (VLANs) for the stack and adds the QinQ tag. This VLAN is then tagged to the compute nodes.
This works well with one input / output interface.
Whats happens if I add another LACP-trunk to another switch in the stack?
a) RSTP is detected on link (untagged) and the whole trunk is blocked - would work
b) RSTP is ignored and the network floods
c) RSTP is ignored on link level but not on VLANs as they get bridged inside the metro vpn (like directly connecting them) - switch will probably block one trunk while the carrier switch does not know, packetloss
The customer switch (Cisco SG300) is managed by me, so RSTP could be guaranteed.
What needs to be done, to get redundancy in a QinQ deployment?