False “BGP Down” alarms (APIPA neighbors) after NSX upgrade

Products

• VMware NSX

Issue/Introduction

After upgrading a federated NSX‑T deployment from 3.x to 4.x the NSX Manager UI continuously raises BGP Down alarms that reference APIPA (169.254.x.x) addresses.

• The alarms re‑appear a few minutes after being manually cleared.
• All user‑defined BGP sessions are Up/Established and traffic is unaffected.
• The environment consists of two data‑centers, each with Edge VMs serving a stretched Tier‑0 gateway.
• LM has overridden the BGP configuration from GM

Environment

VMware NSX

Cause

he alarms are false positives caused by a synchronization mismatch in the NSX‑T control plane:
Cause: "BGP" was disabled and later re-enabled by overriding the config on Local Manager.

When BGP is re‑enabled, the manager creates internal “inter‑SR” BGP sessions that use APIPA addresses.
The data‑plane establishes these sessions successfully, but the monitoring service does not receive the “BGP Established” event for the inter‑SR peers.
Consequently, the manager continuously raises BGP Down alarms for those remote inter‑SR neighbors even though the sessions are up.

Resolution

The issue is resolved in NSX 4.2.4 or later, and in VCF 9.1 or later.

Workarounds:
Perform the steps during a scheduled maintenance window because Edge VM dataplane traffic may be briefly interrupted.

Option 1 (For single site deployments): Toggle Inter‑SR iBGP (clears local‑site alarms only)

• Log in to NSX Manager UI.
• Navigate to Networking → Tier‑0 Gateways.
• Select the affected Tier‑0 and click Edit → expand BGP.
• Disable Inter SR iBGP → Save.
• Wait ~30 seconds, then Enable Inter SR iBGP again → Save.
• Result: Alarms for the local‑site inter‑SR neighbors are cleared. Remote (cross‑site) alarms may persist.

Option 2 (For federation deployments with multiple sites): Maintenance‑mode bounce (clears all inter‑SR alarms)

• In the NSX Manager UI go to System → Fabric → Nodes → Edge.
• For each Edge VM reporting the alarms, perform the following:
  
  • Select the Edge VM.
  • Click **Enter Maintenance Mode** and confirm.
  • Wait until the node reports “in maintenance”.
  • Click **Exit Maintenance Mode** and confirm.
  • Verify that the BGP Down alarms have disappeared from the UI.
• Result: The alarm state is refreshed for both local and remote inter‑SR peers, fully clearing the false alarms.

Verification:

• Open Monitoring → Alarms and confirm that no “BGP Down” entries remain.
• Run show bgp summary on any Edge VM – all APIPA peers should show ESTABLISHED.

Note that in federation deployments, issue will only occur if BGP config is overridden from Local Manager (by performing actions such as disable and enable BGP on LM).