NSX-T upgrade fails with "NSX Manager upgrade dry run failed, Do not proceed with the upgrade."

Products

• VMware NSX

Issue/Introduction

• You are upgrading to NSX-T 3.2.0 or higher.
• Your NSX-T managers are small form factor appliances.
• In the NSX-T UI the following message is displayed: 

"NSX Manager upgrade dry run failed. Do not proceed with the upgrade. Please collect the support bundle and contact VMWare GS. Failed migrations: Extracting the required files from MP vmdk Starting parallel Corfu Migrating data 2023-03-23 07:31:42.560093 Error running logical data migration tool. return value 137, log file /var/log/proton/logical-migration.log Ending Manager dry-run script"

• On the NSX-T manager the following log file is not present: /var/log/proton/logical-migration.log
• In log file /var/log/upgrade-coordinator/upgrade-coordinator.log we see the same message:

2023-03-23 07:31:42.560093 Error running logical data migration tool. return value 137, log file /var/log/proton/logical-migration.log

Environment

VMware NSX-T Data Center
VMware NSX

Cause

There are two issues here:

• The message display with "return value 137" indicates that the Appliance is running low on memory. NSX-T manager in a production environment should be medium or large form factor, please refer to the following guide for further details NSX-T Installation Guide.

The dry run, referred to in the error is part of the pre checks carried out before an upgrade starts.

• The path for logical-migration.log file displayed in the error message is incorrect, the correct location of the logical-migration.log file is /var/log/upgrade-coordinator/logical-migration.log

Resolution

• This is a pre check and as such has warned of this issue before your upgrade has begun, that is you have not upgraded any components yet.
  Please increase the NSX-T manager appliance form factor to a supported size (medium or large) and then complete the upgrade.
• The issue with the wrong log location displayed in the error is resolved in NSX 3.2.4 and 4.x.
  
  Workaround:
  If this is a lab or testing environment, then you can try restarting the NSX-T Manager appliance which is running the pre checks. 
  Post reboot, the NSX-T manager should get enough free memory for the pre checks to complete.
  If this is not a small appliance and you encounter this issue, please open a support request and reference this KB.