Cross-site route sync for stretched Tier-0 gateway may fail

Products

• VMware NSX

Issue/Introduction

• You're running VMware NSX 3.2.0, 3.2.0.1 or 3.2.1 and its a Federated environment.
• You have recently deleted and recreated all the interfaces on edge nodes on one site.
• You observe data path impact on stretched networks.  
• Preforming a packet capture on RTEP doesn't show any traffic.
• Checking BGP status for the remote site shows connect state and local site show established state.

nsx-edge-1>set debug
nsx-edge-1> vrf internal
nsx-edge-1> get bgp neighbor summary
...
Neighbor AS State Up/DownTime BFD InMsgs OutMsgs InPfx OutPfx

169.254.32.4 64601 Estab 01:52:04 NC 6729 6729 51 51 -> local site
169.254.32.6 64601 Conne 07:35:24 NC 5447990 5448118 0 0 -> remote site

• If you attempt to list the L2 forwarders on the NSX edge CLI, it comes back empty.

site01-edge-1> get logical-switch <internal-routing-switch-uuid-for-tier0> l2forwarders

Environment

VMware NSX-T Data Center

Cause

This issue may happen if the L2 forwarder port on internal routing switch has not been cleared.

Resolution

This issue is resolved in VMware NSX 3.2.2, available at  Broadcom downloads.

Workaround:

1. Detach any workload attached to segment.
2. Shrink the span of Tier0 gateway and reduce it to a single site.
3. Expand the Tier0 gateway.