In specific NSX environments, enabling the "Static Routes" advertisement/redistribution setting on a Tier-1 (T1) Gateway causes immediate traffic loss to segments attached to that T1. This occurs even if no static routes are currently defined, or if a previously entered static route was invalid. The issue is often isolated to a single T1 while others in the same cluster function normally.
Symptoms:
Example log:
Affected networks from route graph are : [{ Network : <IP_ADDRESS>/32, Direction : UNKNOWN, nhToRealExitNextHopMap : ... }]
VMware NSX-T / NSX 3.x, 4.x, 9.0
This is caused by stale entry in NSX
This issue is resolved in NSX 9.1.
If an upgrade is not immediately possible, invoke the Reprocess API for the affected Tier-1 Gateway. This forces the Manager to recompute the route graph from scratch and clear stale in-memory nodes.
API Call:
POST https://<nsx-manager-ip>/policy/api/v1/infra/tier-1s/<tier-1-id>?action=reprocess
Note: This action can be performed without a maintenance window as it triggers a background re-calculation of the routing state.