During NSX-T upgrade, ESXi host may experience a PSOD

Products

• VMware NSX
• VMware HCX

Issue/Introduction

 

• You are upgrading NSX-T from an NSX-T version less than 3.2.0 to version 3.2.0 or higher.
• You are using the In-Place option to upgrade the ESXi hosts.
• You have configured policy based routing with HCX.
• On the ESXi host, running net-vdr --policyTable -l will show results such as, which indicate HCX policy routing is enabled:

#net-vdr --policyTable -l

Policy Tables
Id Name Ref
--- ------------------------------ ---
5 8926fa40-xxxx-xxxx-xxxx
Policy Table ID 5 Information

#net-vdr --policy -l -B 5
 

Policy Table 5
Destination GenMask Flags Ref Action HitCount
----------- ------- ----- --- ------ --------
0.0.0.0 0.0.0.0 1 allow 0
10.#.#.# 255.0.0.0 1 allow 0
172.#.#.# 255.240.0.0 1 allow 0
192.168.0.0 255.255.0.0 1 allow 0
Policy count = 4
============================================================================================
Dumping DR FL tables
============================================================================================
IPv4 FLE entries
FL Table Dump.

Environment

VMware NSX-T

Cause

When policy routing is configured with HCX and you are doing an In-Place upgrade, this can trigger a PSOD on the ESXi host.

Note: For further details on HCX policy routing Mobility Optimized Networking Policy Route

Resolution

This is a known issue impacting NSX-T data center.

Workaround:

When doing the NSX-T upgrade, instead of an in place upgrade, use maintenance mode for the ESXi hosts upgrade.
or
In HCX if possible, you can disable the policy routing before the upgrade.