BGP Routes Not Appearing in ESXi Host T0 Distributed Router Forwarding Table
search cancel

BGP Routes Not Appearing in ESXi Host T0 Distributed Router Forwarding Table

book

Article ID: 407003

calendar_today

Updated On:

Products

VMware NSX

Issue/Introduction

GP routes learned by the T0 Service Router appear in the Edge node forwarding table using the get forwarding command. However, the same routes do not appear in NSX Manager logical router tables or ESXi host T0 Distributed Router forwarding tables. North/South traffic for specific prefix routes is directed through the default route (0.0.0.0/0) instead of the specific learned routes when viewed from the host perspective. Traffic flows through Intra-Tier Transit links between Distributed Router and Service Router components.

Error symptoms include:

  • BGP learned routes are visible in the Edge node T0 DR forwarding table
  • Same routes are missing from the NSX Manager logical router tables
  • Same routes are missing from ESXi host T0 DR forwarding tables
  • Unexpected routing behavior for specific prefixes

Example forwarding table output on Edge node:

Logical Router
UUID                                                  VRF    LR-ID              Name                              Type
#####-####-####-####-#####             #      #                      DR-#####                         DISTRIBUTED_ROUTER_TIER0

IPv4 Forwarding Table
IP Prefix                                    Gateway IP               Type        UUID                                                   Gateway MAC
0.0.0.0/0                                    ###.###.###.###      route       #####-####-####-####-#####            ##:##:##:##:##:##
###.###.###.###/                      # ###.###.###.##     route       #####-####-####-####-#####            ##:##:##:##:##:##

Environment

VMware NSX

Cause

This behavior is by design in NSX-T architecture. BGP routes learned by the T0 Service Router component on Edge nodes are not programmed into the ESXi host Distributed Router components. The NSX-T design requires all North/South traffic to traverse through Edge nodes. Dynamically learned BGP routes are only maintained in the Edge node forwarding tables.

Resolution

This is expected behavior and not a defect.

Understanding NSX-T Route Distribution

NSX-T handles routing distribution as follows:

  1. BGP routes are learned by T0 SR components running on Edge nodes
  2. Routes are stored locally in the Edge node's forwarding table
  3. ESXi host T0 DR components maintain only static routes and directly connected networks
  4. All North/South traffic from hosts is forwarded to Edge nodes via Intra-Tier Transit links
  5. Edge nodes make forwarding decisions based on their complete routing table, including BGP learned routes

Workaround Limitations

The following approaches do NOT achieve host-level route awareness:

Route Maps with BGP Attributes:

  • Route maps can set BGP local preference and other attributes
  • These attributes do not propagate to host transport nodes
  • Host T0 DR will still send traffic to any available T0 SR/Edge node

T0 Static Routes:

  • Static routes can steer traffic toward the preferred next hop with administrative distance
  • Host T0 DR has no awareness of next hop preferences
  • Limited effectiveness for location-specific routing requirements

T1 Static Routes:

  • T1 static routes are intended only for southbound traffic routing to VM segments
  • Cannot specify T0 uplinks as next hops
  • Do not influence North/South routing decisions

Verification Commands

On Edge Node:

##### > get forwarding

On ESXi Host Transport Node:

# Using nsxcli commands to check routing
nsxcli -c "get logical-routers"

# Get forwarding table for specific logical router
nsxcli -c "get logical-router <router-uuid>"
nsxcli -c "vrf <vrf-id>"
nsxcli -c "get forwarding"

# Check transport node status and routing
nsxcli -c "get transport-node status"

For environments requiring location-specific Edge node selection:

  • Consult with VMware Professional Services for design alternatives that work within NSX-T architecture constraints
  • Review Edge node placement and T0 gateway design for optimal traffic flow
  • Evaluate application-level solutions that might address latency or routing requirements
  • Consider network design patterns that leverage NSX-T capabilities appropriately

Additional Technical Details

  • This behavior applies to all dynamically learned routes (BGP, OSPF, etc.)
  • Static routes configured on T0 gateways will appear in both Edge and host forwarding tables
  • Troubleshooting North/South routing should focus on Edge node forwarding tables for complete route visibility
  • The distributed nature of NSX-T separates the control plane (Edge) from the data plane (host) forwarding decisions

If the error persists after following these steps, contact Broadcom Support for further assistance.

When opening a support request with Broadcom for routing-related issues, provide:

  • Complete network topology diagram
  • T0 and T1 gateway configurations
  • BGP peering configurations and neighbor status
  • Edge node forwarding table outputs using get forwarding
  • NSX Manager routing table exports
  • Specific traffic flow requirements and expected behavior

Additional Information

Powered by Wolken Software