BGP Sessions down on Edge node during LB Config change

Products

• VMware NSX

Issue/Introduction

• BGP sessions go down on edge node during LB config change:
  
  2025-01-13T19:16:24.602Z edge NSX 1 ROUTING [nsx@6876 comp="nsx-edge" subcomp="nsxa" s2comp="routing" level="ERROR" eventId="vmwNSXRoutingStatus"] {"event_state":0,"event_external_reason":"All BGP sessions DOWN","event_src_comp_id":"xxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxx"}

• High memory alarms are observed for the Edge node during the same time on NSX UI. 

Environment

VMware NSX

Cause

When a Load Balancer is reconfigured, the nginx worker process quits and a new process is forked, however, the old worker process only quits after all live connections handled by this process are closed.
The worker processes in a shutting down state makes the edge run out of memory, thus impacting the datapath and BGP sessions.

2026-01-13T19:16:24.013Z edge kernel - - - [6674015.199700] Out of memory: Killed process 14179 (nginx) total-vm:3066896kB, anon-rss:21232kB, file-rss:0kB, shmem-rss:2105672kB, UID:134 pgtables:4800kB oom_score_adj:0

Resolution

Follow the workaround mentioned in KB https://knowledge.broadcom.com/external/article/371702/edge-memory-usage-high-alarm-might-appea.html