ESX PSODed due to panic related to nsxt-vdl2

Products

• VMware vSphere ESXi
• VMware NSX

Issue/Introduction

• It is a scaled / large environment managed by a single NSX Manager cluster.
• NSX is configured on ESXi.
• ESXi goes down with a PSOD after support bundle collection. 
• PSOD screen includes the line-
         
         "Module(s) involved in panic: [nsxt-vdl2-####]"

Environment

• VMware NSX-T 3.2.x
• VMware NSX 4.x
• VCF 9.0.1

Cause

When number of unique vtep entries across unions of all vteps in LS/RD are more than 2048, there is a buffer overrrun during support-bundle collection causing the PSOD. 

Resolution

This is a known issue impacting VMware NSX.

Workaround:

1. Don't collect support bundle on esx host if number of global vtep entries are more than 2048.
2. Second workaround would be to reduce the scale of LS/RD vtep spans so that global vtep count is less than 2048.
3. There is a workaround script developed by Broadcom. If this script is run, it will prevent a PSOD even after support bundle collection. If you need this script, please open a support-request to Broadcom and refer to this KB.