PSOD with exception 13 and 14 for mtx_lock nsxt-vsip

Products

• VMware NSX

Issue/Introduction

This article provides information on how to overcome the PSOD issue with below symptoms.

Symptoms:

• After upgrading from NSX-T 3.0.2 to 3.1.2, host started logging and then crashed with PSOD exceptions 13 and 14.
• ESXI version is 6.7
• Below logs can be seen at Core logs 
  
  021-06-02T09:49:39.604Z cpu85:55315934)Backtrace for current CPU #85, worldID=55315934, fp=0x431ad12aaf48
  2021-06-02T09:49:39.604Z cpu85:55315934)0x451aed09be88:[0x41801179c8bc]mtx_lock@(nsxt-vsip-17883598)#<None>+0x0 stack: 0x431ad12aaf48, 0x3f5, 0x431ad12ab9b4, 0x1, 0x45cd88930400
  2021-06-02T09:49:39.629Z cpu85:55315934)ESC[45mESC[33;1mVMware ESXi 6.7.0 [Releasebuild-17167734 x86_64]ESC[0m
  #GP Exception 13 in world 55315934:VSIP PF Purg @ 0x41801179c8bc
  2021-06-02T09:48:57.943Z cpu78:55316435)pf_set_state_attributes_five_tuple: failed to set attribute (ret = 45)
  2021-06-02T09:48:57.943Z cpu78:55316435)VSIPConversionSetConnAttr: failed to set connection attributes
  
  

Environment

VMware NSX-T Data Center 3.x
VMware NSX-T Data Center

Cause

There is a small window where an attribute connection created, can be destroyed by the purge thread before being linked to a state.
This will result in a new state being linked to a deleted attribute connection and cause a PSOD during rule validation

Resolution

Upgrade to NSX-T 3.1.3.1

Workaround:
Currently there are no workarounds.

Additional Information

Impact/Risks:

Environment impacted due to host crashes.