Traffic Flow is interrupted between nodes on different subnets/vlans from node but succeed from pods

Issue

• Host -> pod (or svc) is broken between nodes, but pod->pod (overlay->overlay) is working fine between pair of nodes.
• Nodes are bridged via vlan management upstream from node layer (managed by vmware)
• Nodes are aware of their peers on secondary vlan and all nodes respond to ICMP ping when queried
• Nodes are unable to route traffic (curl to services/pods) hosted on secondary VLAN nodes; but PODS hosted on the same node CAN make the curl successfully.
• This may affect nodes on VLAN1 hosting the router-default haproxy pods, which rely on HostNetwork to connect to peers; and may intermittently be unable to query peer services and pods on VLAN2.
• Scaling up a test pod on host node on VLAN1 can consistently curl services/pods on VLAN2, even while host node is unable to complete the same.