NSX Edge HA Failover and BGP Adjacency Drops Due to Host Resource Contention

Products

• VMware NSX

Issue/Introduction

• NSX Edge experiences intermittent tunnel connectivity loss, as seen in the Edge /var/log/syslog:
  2026-03-11T12:29:08.892Z edge.example.com NSX 1 FABRIC [nsx@6876 comp="nsx-edge" subcomp="nsxa" s2comp="ha-cluster" level="INFO"] HA tunnel <local-vtep-ip>:<remote-vtep-ip> state changed from Up to Unreachable
2026-03-11T12:29:08.896Z edge.example.com NSX 1 FABRIC [nsx@6876 comp="nsx-edge" subcomp="nsxa" s2comp="ha-cluster" level="INFO"] HA tunnel <local-vtep-ip>:<remote-vtep-ip> state changed from Up to Unreachable
  
  
• This loss of connectivity triggers High Availability (HA) failovers, as seen in the Edge /var/log/syslog:
  2026-03-11T12:29:08.926Z edge.example.com NSX 1 FABRIC [nsx@6876 comp="nsx-edge" subcomp="nsxa" s2comp="ha-cluster" level="WARN" eventId="vmwNSXClusterNodeStatus"] {"event_state":4,"event_external_reason":"Edge node status changed: Up -> Down , reason: VTEP tunnels down","event_src_comp_id":"1990####-####-####-####-#######575ac","event_sources":{"id":"1990####-####-####-####-#######575ac"}}
2026-03-11T12:29:09.051Z edge.example.com NSX 1 FABRIC [nsx@6876 comp="nsx-edge" subcomp="nsxa" s2comp="ha-cluster" level="WARN" eventId="vmwNSXClusterFailoverStatus"] {"event_state":1,"event_external_reason":"Service router switches over from Active to Down. edge node down","event_src_comp_id":"1990####-####-####-####-#######575ac","event_sources":{"id":"2171####-####-####-####-########0ebe","router_id":"69db####-####-####-####-#######baff"}}
  
  
• This causes BGP adjacencies on the edge to drop, as seen in the Edge /var/log/frr/frr.log:
  2026/03/11 12:29:10.298195 BGP: %NOTIFICATION: sent to neighbor <neighbor-ip-address> 6/2 (Cease/Administratively Shutdown) 0 bytes
  
  
• When tunnel connectivity is restored, HA failover is triggered again as the Edge reassumes BGP adjacencies.
  
  
• Symptoms are resolved by migrating (vMotion) the Edge VM to a different host.

Environment

VMware NSX
VMware ESXi

Cause

The NSX Edge VM is experiencing compute or network resource starvation due to severe resource contention from co-located workload VMs on the same ESXi host.

Resolution

• Relocate the affected NSX Edge VMs to a dedicated Edge cluster or separate ESXi hosts to prevent co-location with general workload VMs.

• If temporary co-location is unavoidable, configure strict CPU and Memory reservations for the Edge VMs to guarantee resource availability and prevent datapath starvation.

• Validate host placement and Distributed Resource Scheduler (DRS) anti-affinity rules to ensure Edge VMs are not inadvertently migrated back to congested workload hosts.

 

Additional Information

NSX Edge VM CPU and Memory Requirements