We've been struggling with an NSX-T deployment on hosts with only 2 pnics. So far we've managed to deploy the managers and migrate all vmkernels over to a single N-VDS, and we now have all VM's running on the platform too.
We're currently trying to deploy an Edge node on top of it to take over all the routing for the VM's using a single Tier0 router. Tier 1 isn't required as it is not multi-tenanted. We then create a LS linked to the edge node and have moved a test VM onto it. It can ping the interface on the the T0 router on the same subnet, but is not able to ping the interface IP assigned to the uplink, and therefore can't get out to the internet.
In terms of the setup of the Edge TN, we have the following:
N-VDS - nic1 assigned to existing management segment
From the Edge node NSX config screen:
N-VDS1 - nic2/fp-eth0 - linked to tz-overlay with transport vlan 20 for Edge TEP IP(different from the esxi host transport node TEP VLAN).
N-VDS2 - nic3/fp-eth1 - Segment on VLAN 10 - uplink to physical network / router / additional transport zone (tz-edge-vlan)
nic4/fp-eth2 - Disabled (disconnected from the Edge VM in vSphere)
One thing that bugs me is that the if I look at the transport zone page, the tz-edge-vlan has a status of unknown.
In terms of the model, we're aiming for the last model mentioned in this blog: https://blogs.vmware.com/networkvirtualization/2018/10/flexible-deployment-options-for-nsx-t-edge-vm.html/
Any help or ideas greatly appreciated, and apologies for the confusing overview, but NSX-T is new to me.