Have a VM connected to a vss with its own uplink.
Another VM on same hypervisor also connected to a vss with its own uplink also.
These two uplinks are directly connected.
pings work fine between the two interfaces on the VMs.
However, when traffic is sent from one VM to the other with 802.1p (priority bits in the 802.1q header) the bits are non-zero when sending but the receiving VM sees all zeroes in the priority bits. This is verifired in wireshark on the two VMs.
Tried pktcap-uw on the sending and receiving vmnics (and tried capture points as close to the wire as possible) but they don't show any 802.1q at all. I.e. there's no vlan tagging at all. Not sure how the receiving VM got the 802.1q vlan tagging to show up then.
But the main problem is the clearing of the Priority Bits. Is there any kind of setting anywhere that might affect this?
For the diagram:
[disregard eno5,eno6, ens2f0, ens2f1 - they were when I was testing all of this on Ubuntu. Currently testing on esxi 6.7]
Traffic leaves ens224 on the VM "u2" which is connected to a vss with vmnic7 as the uplink. It has the value of 6 for the Priority Bits in the 802.1q header.
Traffic is received on ens224 on the VM "u1" which is connected to a vss with vmnic5. The frame is identical to what was sent except that the Priority Bits in the 802.1q header are all 0.
Concurrent wiresharks were run on the two VMs to verify the traffic.
